摘要:

The chemical basis of protein splicing Henry Paulus Protein splicing is a recently discovered mechanism for the post-translational processing of proteins. It involves the selfcatalyzed excision of an intervening polypeptide the intein from an inactive enzyme precursor and the formation of an active enzyme by joining the flanking regions by a peptide bond. Protein splicing occurs at a catalytic center that resides entirely within the intein. The catalyzed reactions include rearrangement of a peptide bond adjacent to cysteine or serine to yield a peptide ester intramolecular transesterification involving a second cysteine serine or threonine side chain to yield a branched protein and cyclization of an asparagine residue coupled to peptide bond cleavage to effect intein excision.This review discusses the mechanisms of these reactions and of similar reactions that underlie other types of protein rearrangements as well as the current state of knowledge on how these reactions are catalyzed. 1 Introduction Since the discovery of zymogens in the 1930s it has been clear that many proteins are synthesized as inactive precursors that are activated by the rearrangement of peptide bonds. Such posttranslational processing is usually catalyzed by other proteins as in the activation of the blood clotting factors by highly selective proteases. However in the last 10 years many examples of self-catalyzed peptide bond rearrangements have been described. These include protein splicing the autoprocessing of hedgehog proteins the autocleavage of amidohydrolase precursors and the formation of pyruvoyl enzymes.A common feature of these self-catalyzed reactions is that they are initiated by the N?S or N?O acyl rearrangement of a peptide bond involving the amino group of cysteine serine or threonine. Moreover protein splicing elements and the autoprocessing domains of hedgehog proteins have significant structural and sequence homology suggesting an evolutionary relationship.1 Protein splicing which occurs in many types of organisms and whose mechanism is now relatively well understood will thus serve as the paradigm for self-catalyzed protein rearrangements. As we discuss the mechanism of Henry Paulus was educated at the University of Chicago where he carried out research for the PhD (1959) under the guidance of Eugene P.Kennedy. He was an NSF postdoctoral fellow at Cambridge University with Alexander Todd and at Harvard University with Konrad Bloch. In 1961 he joined the Department of Biological Chemistry at the Harvard Medical School where he still holds a faculty appointment but in 1975 he moved his laboratory to the Boston Biomedical Research Institute where he is a Senior Scientist. Boston Biomedical Research Institute Boston MA 02114 and Department of Biological Chemistry and Molecular Pharmacology Harvard Medical School Boston MA 02115 USA protein splicing we will point out similarities with the reactions that underlie other types of protein rearrangements. Protein splicing was discovered in 1990 first in yeast a unicellular eukaryote but then also in bacteria and archaea.About 50 examples of protein splicing are known to date of which 36 are listed in the latest published compilation.2 Protein splicing involves the excision of an intervening polypeptide segment (the ‘intein’) which usually interrupts a functional domain of an enzyme and the joining of the flanking regions (the ‘exteins’) through a peptide bond thereby generating an active enzyme. This process is formally analogous to RNA splicing in which an intervening sequence (the ‘intron’) is excised and the flanking regions (the ‘exons’) are joined by a normal 3A,5A-phosphodiester bond to yield a functional messenger RNA molecule (see Fig. 1). RNA splicing and protein splicing may be viewed as alternative methods for the expression of interrupted genes which differ only in the level at which excision of the intervening sequences occurs.From this perspective both introns and inteins are manifestations of ‘selfish DNA’ whose excision is essential for the survival of the organism. Many introns and inteins also harbor homing endonucleases which can initiate the insertion of the DNA sequences encoding these elements into other genes either in the same or in different organisms. Such intervening sequences can be considered parasites that insert themselves into vital host genes yet avoid killing their hosts because they can excise themselves and thereby restore host gene function. RNA splicing and protein splicing may also have regulatory functions.Introns are generally positioned in regions of genes that correspond to domain boundaries in the encoded proteins and alternate RNA splicing provides a mechanism for the modular assembly of functionally different proteins from a single gene. In contrast inteins are usually embedded in highly conserved regions of functional protein domains and protein splicing is a mechanism for the activation of an inactive precursor protein. Both types of intervening sequences can thus serve to control gene expression albeit to different ends. Because there are introns and inteins that lack homing endonuclease domains one might speculate that the primary function of intervening sequences is indeed regulatory but that they are frequently harnessed by homing endonucleases as vehicles for horizontal transmission to other organisms and species.Perhaps the strongest argument for an origin of inteins as regulatory elements is the homology of the protein splicing domains to the autoprocessing domains of hedgehog proteins. The latter occur in essentially all metazoans ranging from nematodes and fruit flies to man and fulfill a purely regulatory function in early developmental pattern formation. This suggests that protein splicing elements known to occur only in unicellular organisms were modified early in the evolution of multicellular organisms to regulate cell fate in embryonic development. 2 Chemical strategy of protein splicing 2.1 General characteristics of the protein splicing process The investigation of the chemical mechanisms underlying protein splicing was made possible by the classic experiments 375 Chemical Society Reviews 1998 volume 27 Fig.1 Comparison of the excision of intervening sequences by (a) RNA splicing and (b) protein splicing of Xu and coworkers.3 The coding sequence for the intein embedded in the DNA polymerase of the hyperthermophilic archeon Pyrococcus sp. GB-D was inserted into a foreign context so as to allow synthesis of the unspliced precursor at low temperatures and the induction of splicing of the purified precursor by an increase in temperature. In these hybrid constructs the maltose binding protein of Escherichia coli was used as the N-extein to allow rapid affinity purification of the unspliced precursor on amylose resin.These experiments yielded two important insights into the general nature of protein splicing (i) protein splicing occurs with highly purified precursor proteins in which the intein is inserted between exteins that bear no relationship to protein splicing and is expressed in an organism (E. coli) in which protein splicing normally does not occur indicating that protein splicing requires no accessory enzymes and that all necessary catalytic groups are part of the intein itself; and (ii) protein splicing is induced merely by increasing the temperature in the absence of cofactors such as ATP indicating that protein splicing is a spontaneous process requiring neither organic cofactors nor sources of metabolic energy.Additional information on the nature of protein splicing came from the laboratory of Anraku,4 who developed an in vitro protein splicing system based on the intein from the VMA subunit of the vacuolar ATPase of Saccharomyces cerevisiae. Expression of recombinant proteins containing this intein in E. coli leads to the accumulation of unspliced precursor in the form of inclusion bodies that can be solubilized with 6 M guanidin- Fig. 2 Major sequence features of a typical protein splicing element inserted between an N-extein (shown partially on the left) and a C- extein (right). The bold rectangle represents the protein splicing element with its N-terminus at the left and the shaded boxes labeled A B F and G represent conserved sequence motifs common to all protein splicing elements.The size and positions of the rectangles are approximately to scale. Invariant amino acid residues that occur in all inteins are indicated below the rectangle. The portion of the intein enclosed in dotted lines represents the homing endonuclease domain which is found in most but not all inteins; its size exceeds that of the protein splicing element considerably and is not represented on scale. Chemical Society Reviews 1998 volume 27 376 ium chloride and undergo splicing upon removal of the denaturant. Using this experimental system Anraku and coworkers made two important observations:5 (i) protein splicing is refractory to inhibitors of cysteine- serine- aspartic- and metallo-proteases suggesting that the protein splicing catalytic center does not resemble the active sites of any of the four major classes of proteases and (ii) no swapping of exteins occurs when mixtures of protein splicing precursors in which the VMA intein was inserted between different pairs of exteins are allowed to undergo splicing indicating that protein splicing is exclusively an intramolecular reaction.The general picture of protein splicing that emerges from these in vitro observations is as follows 4 protein splicing is catalyzed solely by amino acid residues contained in the intein; 4 protein splicing requires no coenzymes or sources of metabolic energy and therefore involves bond rearrangements rather than bond cleavage followed by resynthesis; 4 protein splicing is an intramolecular process; 4 the amino acid residues that catalyze protein splicing are arranged in an active center which differs from that of any of the four major classes of proteases.2.2 Summary of the reaction steps that underlie protein splicing The general nature of these reactions was anticipated on the basis of the two types of amino acid side chains that are invariably found adjacent to the splicing junctions (see Fig. 2). (i) The N-termini of the intein and the C-extein are always amino acids that carry a nucleophilic side chain cysteine (Cys) serine (Ser) or threonine (Thr). These amino acid side chains often function in nucleophilic attacks on amide and ester bonds including—albeit usually under extreme conditions—attacks on amide bonds involving their own amino groups.(ii) At the C-terminus of the intein one always finds an asparagine (Asn) residue. In proteins with long biological half-lives or in synthetic peptides during the acidic conditions encountered in deprotection steps Asn residues—especially when adjacent to glycine—often undergo spontaneous cyclization reactions that can lead to deamidation or peptide bond cleavage. Detailed biochemical studies on protein splicing during the period 1993–96 succeeded in defining the chemical reactions that underlie this unusual catalytic process.6–11 These reactions which consist of the four steps summarized in Fig. 3 are described in the next four sections of this review. Briefly these steps are Step 1 Formation of a linear ester intermediate by N?S or N?O acyl rearrangement involving the nucleophilic amino acid residue at the upstream splice junction; Step 2 Formation of a branched ester intermediate by the attack of the nucleophilic residue at the downstream splice junction on the linear ester intermediate; Step 3 Cyclization of the asparagine residue adjacent to the downstream splice junction coupled to cleavage of the branched ester intermediate to yield an excised intein with a C-terminal aminosuccinimide residue and the two exteins joined by an ester bond; Step 4 Spontaneous hydrolysis of the aminosuccinimide residue and rearrangement of the ester linking the exteins to the more stable amide bond.3 N?O and N?S acyl rearrangements (Scheme 1) Scheme 1 Fig.3 Mechanism of protein splicing. The amino acid residues that participate directly in the chemical transformations are shown (X = O or S). The remainder of the intein and exteins are shown by boxes which are not to scale. 3.1 Acid-induced N?O acyl rearrangements When treated with strong acids at low temperatures peptide bonds involving the amino groups of serine or threonine rearrange to ester bonds involving the corresponding amino acid side chains (reviewed in ref. 12). Such N?O acyl rearrangements occur through the nucleophilic attack by the amino acid side chain on the peptide carbonyl carbon and are thought to proceed through a hydroxyoxazolidine intermediate (Fig. 4). At neutral pH the equilibrium position for N?O acyl shifts favors the amide but under strongly acidic conditions the ester is stabilized by protonation of the free amino group.Reversal of acid-induced N?O acyl shifts can be prevented by the acetylation formylation dinitrophenylation or deamination of amino groups at low pH. This allows subsequent alkaline hydrolysis of the ester bonds thereby achieving selective cleavage of proteins adjacent to Ser and Thr residues. Although N?S acyl rearrangements involving the analogous reaction of 377 Chemical Society Reviews 1998 volume 27 Fig. 4 N?O and N?S acyl rearrangement with the postulated hydroxythiozolidine or hydroxyoxazolidine intermediate Cys residues and hydroxythiazolidine intermediates can also occur these have not been studied. Acid-induced N?S acyl shifts are less frequent in proteins than N?O acyl shifts owing to the low levels of Cys in proteins relative to Ser and Thr and the equilibrium is even more unfavorable thioesters being about 50-fold less stable than oxygen esters.13 Moreover at pH 5.0 the lowest pH at which amino groups can be trapped by acylation following N?O acyl shifts,12 the rates of S?N acyl rearrangements are extremely rapid with half-lives of only 24 s,8 precluding chemical trapping of the thioester product for quantitation under neutral conditions.3.2 N?O and N?S acyl rearrangements in protein splicing The occurrence of N?O and N?S acyl rearrangements in protein splicing was demonstrated using an in vitro splicing system involving the intein from the DNA polymerase of the hyperthermophilic archeon Pyrococcus sp.GB-D.7 Because thioesters react much more rapidly than oxygen esters with nitrogen nucleophiles at neutral pH protein splicing precursors in which the Ser residue of interest was replaced by Cys were constructed and purified. In the presence of 0.25 M hydroxylamine or 0.1 M ethylenediamine at pH 6 or higher these constructs undergo rapid cleavage at the upstream splice junction consistent with the aminolysis of a thioester (Fig. 5). The site of hydroxylaminolysis was identified by analysis of the C-terminus of the polypeptide cleavage products. Comparison of the C-terminal peptide hydroxamate with synthetic peptide hydroxamates with respect to chromatographic mobility colorimetric assay amino acid composition and high resolution mass spectrometry showed that the hydroxylamine-sensitive site in the splicing precursor is the peptide bond adjacent to the serine residue at the upstream splice junction.These studies were subsequently extended to the VMA intein of S. cerevisiae in which Cys occurs naturally at the upstream splice junction by studying the hydroxylaminolysis of intein variants that were prevented from undergoing protein splicing by the replacement of essential amino acid residues at the intein C-terminus.11 Studies on the N?S acyl rearrangements of such mutant derivatives of the S. cerevisiae VMA intein showed that the thioester intermediate can also be cleaved by transesterification with a large excess of low molecular weight thiols such as cysteine or 1,4-dithiothreitol (Fig.5) an observation that made possible the development of a protein expression/purification system with self-cleavable affinity tags.14 The fact that the thioester product of N?S acyl shifts cannot be trapped chemically at neutral pH when generated by acidinduced rearrangements of proteins (see Section 3.1) but can be trapped by hydroxylamine or thiols when generated catalyt- Chemical Society Reviews 1998 volume 27 378 Fig. 5 Trapping of thioester intermediates by nucleophilic displacement by hydroxylamine or an excess of thiol ically during protein splicing suggests that the position of the NÔS equilibrium differs in the two cases. This thermodynamic difference obviously cannot be the result of catalysis.Rather its existence implies that the bond that undergoes the N?S or N?O acyl shift in protein splicing is not a typical peptide bond but is under strain that can be relieved by the rearrangement. Support for this idea was provided by the crystallographic analysis of the GyrA intein from Mycobacterium xenopi which showed the peptide bond in question to be in the uncommon cis configuration.15 The energy of cis-peptide bonds is about 5 kcal mol21 higher than that of trans-peptide bonds making the equilibrium constant of N?S acyl rearrangements in which a cis-peptide bond is rearranged to an ester bond about 4000 times larger. Furthermore a cis-peptide precursor resembles the cyclic hydroxythiazolidine intermediate more closely than a trans-peptide precursor with respect to atomic coordinates (Fig.6) suggesting that N?S acyl shifts involving precursors with a cis-peptide bond have lower activation energies and thus may have a kinetic as well as a thermodynamic advantage over the corresponding reactions in an all-trans peptide. It will be of great interest to learn whether the peptide bond linking the intein to the N-extein is always in the cis-configuration. Studies on the role of the C-terminal amino acid of the N-extein in protein splicing involving the S. cerevisiae VMA intein support the notion that the geometry of the peptide bond undergoing the N?S acyl shift is important.16 When Val Leu Glu Phe Tyr Trp Lys or Thr participate in this peptide bond substantial proportions of thioester intermediate can be trapped by denaturation of the unspliced precursor with 8 M urea indicating that the NÔS equilibrium is far displaced towards the thioester with equilibrium constants at pH 7.6 as high as 10 (for the Leu-Cys bond).The ester intermediate was identified by thiol-induced cleavage using a large excess of dithiothreitol and is not observed when the scissile bond involves Gly Ala Ile Ser Gln His Cys Asn and Pro. The observation that the NÔS equilibrium for a splicing junction comprising Leu-Cys favors thioester formation whereas an Ile-Cys junction exists mainly in the amide precursor form serves to illustrate the subtle structural factors that drive protein splicing. Little is known about the catalytic groups involved in the N?S acyl rearrangement beyond what can be inferred from the crystal structure of the GyrA intein from M.xenopi.15 This structure was obtained using an intein analog in which Cys-1 was replaced by Ser and which carried a single Ala residue as the N-extein but lacked a C-extein. (To avoid confusion we will refer to Cys-1 in the discussion that follows although the crystal structure was obtained with the Ser-1 analog of the GyrA intein.) A key element in the reaction appears to be the conserved intein Block B which contains the sequence Thr-x-x- His located about 75 residues from the upstream splice junction (see Fig. 2).2,17 A similar Thr-x-x-His motif is found in the hedgehog proteins (see Section 3.3.1). In the GyrA intein from M.xenopi this conserved sequence is Thr-Ala-Asn-His and comprises intein residues 72–74. An H-bond between Thr-72 and the amide N of Cys-1 as well as steric constraints force the Fig. 6 Mechanism of N?S acyl rearrangement inferred from the crystal structure of an analog of the GyrA intein from M. xenopi. The diagram shows the amino acid residues at the N-terminal splice junction with Cys replacing Ser at the intein N-terminus of the crystal structure and the side chains of the amino acids thought to be involved in catalysis. The positions of the C N and O atoms are based on the atomic coordinates of the precursor determined by Klabunde et al.15 and were modeled by the RasMol program. The positions of H atoms and the atoms of the oxythiazolidine anion are estimates.scissile peptide bond into the cis conformation (Fig. 6). The thiol group of Cys-1 is poised for nucleophilic attack on the carbonyl of the scissile peptide bond. (The nucleophilic attack probably involves the thiolate anion of Cys-1 which is replaced by a Ser hydroxy in the crystal structure.) The resulting tetrahedral intermediate an oxythiazolidine anion is stabilized by the side chain hydroxy of Thr-72 and the side chain amide of Asn-74 analogous to the stabilization of the tetrahedral intermediate in serine protease catalysis by the so-called oxyanion hole. The imidazolium group of His-75 is capable of donating a proton to the amino N of the scissile bond to promote the collapse of the oxythiazolidine anion to the thioester product (Fig.6). 3.3 Other self-catalyzed N?O and N?S acyl rearrangements Nucleophilic attack by a Ser or Cys side chain on the adjacent peptide bond leading to N?S or N?O acyl shifts is not only seen in protein splicing but also in many other self-catalyzed protein rearrangements (Fig. 7). The broad distribution and diverse biological functions of the proteins involved suggest that such self- catalyzed rearrangements arose relatively early in evolution. 3.3.1 Autoprocessing of hedgehog proteins The hedgehog proteins are signaling proteins that function in embryonic patterning of multicellular animals from nematodes to mammals. A typical example is the hedgehog protein from Drosophila melanogaster. A 45 kDa protein precursor is secreted and then undergoes self-catalyzed processing involving polypeptide chain cleavage adjacent to a Cys residue to yield a 25 kDa C-terminal fragment (Hh-C or autoprocessing domain) and a 20 kDa protein (Hh-N or signaling domain) with a C terminus that is esterified with cholesterol.The esterified Hh-N signaling domain which is lipophilic owing to its C-terminal cholesterol moiety interacts with specific cellsurface receptors and is responsible for developmental signaling. 18 The autocleavage in hedgehog proteins occurs adjacent to a highly conserved Cys residue. This Cys is followed by a 12-amino acid sequence motif that resembles the conserved Block A motif of self-splicing proteins (see Fig. 2),19 suggesting mechanistic similarities between protein splicing and hedgehog autoprocessing.20 The mechanism of hedgehog protein autoprocessing was studied using a hedgehog protein analog in which a hexahistidine sequence replaced most of the Hh-N domain thereby allowing for facile purification of the precursor protein by affinity chromatography on Ni2+ resin after expression in E.coli. Evidence for a thioester intermediate came from the observation that hydroxylamine and thiols promote cleavage of the precursor yielding the hydroxamate or thiol ester of Hh-N (see Fig. 5).20,21 Replacement of His-329 which resides in a conserved domain homologous to the Block B motif of selfsplicing proteins (see Fig. 2) blocks the autoprocessing reaction suggesting that Block B plays a role in the acyl rearrangements leading to the ester intermediates in protein splicing and hedgehog protein autoprocessing.Comparison of the amino acid sequences of hedgehog autoprocessing domains (Hh-C) and protein splicing elements shows considerable homology,22–24 which is preserved on the level of protein structure,24 suggesting a close evolutionary relationship between protein splicing elements and hedgehog autoprocessing domains. 3.3.2 Autocleavage of glycosylasparaginase precursors Glycosylasparaginase hydrolyzes N4-(b-N-acetylglucosaminyl)-l-aspartic acid and related glycans and its deficiency in humans leads to aspartylglycosaminuria a genetic disorder of glycoprotein degradation. The human and Flavobacterium meningosepticum glycosylasparaginases have a high degree of amino acid sequence and structural homology both being composed of a and b subunits which are encoded by a single 379 Chemical Society Reviews 1998 volume 27 Fig.7 The role of N?O and N?S acyl rearrangements in self-catalyzed protein rearrangements gene. The primary gene product an inactive polypeptide precursor is converted to the active enzyme by self-catalyzed cleavage adjacent to a Thr residue to yield the a and b subunits the latter with an N-terminal Thr.25 Replacement of this Thr residue in bacterial glycosylasparaginase with amino acids other than Ser or Cys completely prevents autocleavage whereas replacement with Cys or Ser greatly reduces its rate.26 The use of affinity-tagged recombinant proteins in which the critical Thr residue is replaced by Cys in conjunction with the discovery that glycine severely inhibits autocleavage made possible the purification of glycosylasparaginase precursors by affinity chromatography and the study of the autocleavage mechanism.Evidence for the involvement of an ester intermediate in autocleavage comes from the observations that the cleavage of the Thr?Cys mutant protein is promoted by the nucleophile hydroxylamine and inhibited by the thiol-blocking reagent iodoacetamide.26,27 These studies also provided important insights into the mechanism of the N?O acyl shift leading to ester formation in which the tripeptide His-Asp-Thr at the potential cleavage site plays a critical role. The imidazole side chain of His-150 acts as proton acceptor to facilitate the nucleophilic attack of the Thr-152 hydroxy group on the peptide carbonyl of the Asp-Thr peptide bond [Fig.8(a)]. The Asp-151 side chain is anchored in the partially formed substrate binding pocket of the enzyme precursor in a conformation that strains the Asp-Thr peptide bond. The relief of this strain upon N?O acyl rearrangement serves to compensate for the unfavorable equilibrium constant associated with N?O acyl shifts. Binding of the Asp-151 side chain in the incipient substrate binding site is stabilized by ionic interactions with the guanidino group of Arg-180. Owing to its small size and ionic interactions of its carboxy and amino groups with Arg-180 and Asp-183 respectively free glycine can also bind strongly to the same site.The important role of the binding of the Asp-151 b-carboxylate moiety in the incipient substrate binding pocket is supported by three lines of evidence (i) the strong inhibition of autocleavage by free glycine which competes with the Asp side chain for the incipient substrate binding site; (ii) the suppression of autocleavage by mutations in which the Asp-151 residue is replaced by other amino acids; and (iii) the activation of autocleavage in an Asp-151?Gly mutant by the addition of glycine which by Chemical Society Reviews 1998 volume 27 380 substituting for the missing Asp-151 side chain in the substrate binding pocket restores the original strained protein conformation. 27 An interesting aspect of the autocleavage of the glycosylasparaginase precursor is that the N-terminal Thr residue of the b subunit generated by this process is essential for enzyme activity.In this respect glycosylasparaginase resembles a group of other amidohydrolases termed N-terminal nucleophile amidohydrolases which include glutaminephosphoribosylpyrophosphate amidotransferase penicillin acylase and the proteasome. These enzymes have similar threedimensional structures and their activities depend on an N-teminal Ser Thr or Cys residue that is generated by the cleavage of a precursor protein probably through self-catalysis involving N?O or N?S acyl rearrangements.28 It is interesting that the nucleophilic amino acid in the N-terminal nucleophile amidohydrolases plays quite similar roles mechanistically in the autoprocessing reaction and in the catalysis of amide hydrolysis by the mature enzyme.This is illustrated in Fig. 8 for glycosylasparaginase.27,29 The possibility of a similar dual role for the N-terminal nucleophilic residue of inteins in the first and third steps of protein splicing is discussed in Section 5.2. In spite of these possible functional similarities the N-terminal nucleophile amidohydrolases exhibit no structural or sequence similarity to protein splicing elements. 3.3.3 Autocleavage leading to pyruvoyl enzyme formation The activity of pyridoxyl phosphate-independent bacterial amino acid decarboxylases and reductases depends on N-terminal pyruvoyl prosthetic groups. The pioneering work of E.E. Snell and coworkers (reviewed in ref.30) established that the enzyme-bound pyruvate is generated through the self-catalyzed cleavage of a protein precursor. In the case of the histidine decarboxylase from Lactobacillus 30a processing of the proenzyme involves cleavage of the peptide bond beween Ser- 81 and Ser-82 coupled to the conversion of Ser-82 to an N-terminal pyruvate residue by b-elimination. The first suggestion that this process involves an ester intermediate came from isotopic labeling experiments which showed that 18O from the side chain of Ser-82 is transferred to the Ser-81 carboxy group in the course of the cleavage reaction. The putative ester Fig. 8 Comparison of the role of Thr-152 in (a) autocleavage of the glycosylasparaginase precursor and (b) catalysis of amide hydrolysis by the mature enzyme based on the studies of Guan and coworkers.27,29 Only the residues at the catalytic site are shown with the a and b domains of the protein indicated by rectangular boxes.Abbreviation NAcGlc b-N-acetylgucosaminyl. intermediate could be trapped as the hydroxamate after replacing the relevant Ser residue with Cys,31 owing to the much greater susceptibility of thioesters than oxygen esters to attack by nitrogen nucleophiles. Direct evidence for an ester intermediate in the autoprocessing of pyruvoyl enzyme precursors came from X-ray crystallographic studies on a pyruvoyl enzyme aspartate decarboxylase from E. coli.32 Aspartate decarboxylase is synthesized as a single polypeptide chain p that undergoes self-cleavage to generate a and b subunits with the b subunit carrying an N-terminal pyruvate residue.The active enzyme crystallizes as a tetramer with the structure (ab)3(p) which has only 3 pyruvoyl groups presumably owing to negative cooperativity in the processing reaction. The 2.2 Å electron density map is inconsistent with a peptide bond at the potential cleavage site between Gly-24 and Ser-25 of the unprocessed p subunit but shows a satisfactory fit with a Gly- Ser ester bond suggesting that an N?O acyl shift had occurred in the p subunit to yield a stable ester. The stability of the peptide ester in the crystals of the aspartate decarboxylase tetramer grown at pH 7.5 is remarkable. Two factors may be responsible for the stabilization of the ester (i) the protonation of the free serine amino group by Tyr-58 and (ii) the relief of steric strain within the 5-amino acid loop containing Gly-24 and Ser-25 which is unusually short for a loop linking two b strands.32 The aspartate decarboxylase from E.coli shows neither sequence nor structural homology with other pyruvoyl enzymes nor with glycosylasparaginases hedgehog proteins or protein splicing elements. This suggests that except for protein splicing elements and hedgehog proteins which show clear signs of an evolutionary relationship self-processing proteins that take advantage of N?O or N?S acyl shifts to effect peptide bond cleavage have evolved independently. An independent origin of these self-processing events is also suggested by the diverse mechanisms for overcoming the unfavorable equilibrium of N?O and N?S acyl rearrangements.Although based on the same underlying principle— destabilization of the peptide bond that is to undergo the acyl rearrangement—the various mechanisms by which this is achieved include a cispeptide bond (GyrA intein) anchoring of an amino acid side chain in an incipient substrate binding pocket (glycosylasparaginase) and tertiary structure constraints (aspartate decarboxylase). 4 Transesterification (Scheme 2) Intramolecular transesterifications are relatively rapid and facile reactions involving both effective nucleophiles and favorable equilibria. In protein splicing transesterification occurs either between a thiol and a thioester an alcohol and an oxygen ester or an alcohol and a thioester but never between a thiol and an oxygen ester.2 In the first two cases the equilibrium constant at neutral pH is near unity; in the alcohol–thioester reaction it is about 50.13 Accordingly in 40% of the known inteins where an N?S acyl shift is followed by transesterification with a Ser or Thr side chain at the downstream splice junction the transesterification step will partially compensate for the unfavorable equilibrium of the N?S acyl shift.However even when the equilibrium constant for transesterification is close to one as in the alcohol–oxygen ester or thiol– thioester reactions kinetic stabilization occurs as discussed below. Evidence for the occurrence of transesterification in protein splicing came from the characterization of a transient interme- 381 Chemical Society Reviews 1998 volume 27 Scheme 2 diate that occurs in the in vitro splicing reaction involving the DNA polymerase intein from Pyrococcus sp.GB-D discussed earlier.3 The intermediate which appears at 37 °C with a halftime of about 15 min and migrates abnormally slowly upon polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate was found to have two N-termini corresponding to those of the N-extein and the intein. The unusual branched structure of this intermediate accounts for its abnormal electrophoretic mobility and implies that two polypeptides with free N-termini are linked to a single Ser residue one to its a-amino group and the other to its b-hydroxy group.In order to determine which of the two N-terminal polypeptides the N-extein or the intein is esterified with the Ser side chain hydroxy group protein splicing was interrupted by adding 6 M guanidinium chloride when an appreciable amount of the branched intermediate had accumulated and the mixture was incubated at pH 9.0 and 65 °C at which amide bonds are stable but esters are hydrolyzed. Under these conditions free N-extein is released from the branched intermediate indicating that it had been linked by an ester bond consistent with the reaction pathway outlined in Fig. 3.9 The intein from the DNA polymerase of Pyrococcus sp. GBD has Ser residues at both splice junctions and the transesterification reaction should therefore have an equilibrium constant of one and be freely reversible.Since the transesterification equilibrium is coupled to the pH-dependent N?O acyl rearrangement equilibrium the reversibility of transesterification can be experimentally verified by controlling the pH. Raising the pH to 10 causes the branched intermediate to revert to the linear precursor form with a t1/2 of about 10 min; readjusting the pH to 6 leads to regeneration of the branched intermediate and subsequent protein splicing indicating full reversibility.3 It is interesting to compare the t1/2 for the reversal of transesterification with the t1/2 for the O?N acyl shift which is 27 min at pH 6.0 and declines precipitously with increasing pH (to about 2 min at pH 7.0 and an unmeasurably small value at higher pH).8 The rate at which the branched intermediate reverts to the linear precursor is several orders of magnitude slower than the rate of the O?N shift per se,3 suggesting that the transesterification reaction provides significant kinetic stabilization of ester intermediates.Such kinetic stabilization may play an important role when protein splicing involves an Chemical Society Reviews 1998 volume 27 382 N?S acyl rearrangement because the estimated t1/2 for the S?N acyl shift at pH 7 is only about 0.1 s. The crystal structures of the S. cerevisiae VMA intein33 and the M. xenopi GyrA intein15 were determined with modified proteins lacking the C-terminal Cys or Thr residues that are involved in transesterification and little can therefore be said about the amino acid side chains that might assist the transesterification reaction.However even when the position of the C-terminal Thr is approximated by modeling no amino acid side chain that might provide catalytic assistance can be identified.15 It is possible that the thiol of the Cys that participates in transesterification in the yeast VMA intein and many other inteins has an unusually low pKa and is therefore an effective nucleophile between pH 6 and 7 where splicing is optimal. The observation that a disulfide bond between the Cys residues at the splice junctions of the M. tuberculosis RecA intein forms rapidly under very mildly oxidizing conditions during the reconstitution of a functional protein splicing element from N- and C-terminal fragments34,35 supports this notion and also indicates that the spatial arrangement of the Cys thiols at the two splice junctions is well suited for transesterification.Although esters and thioesters play an important role in protein metabolism ranging from protein degradation to the defensive functions of the complement system and a-macroglobulins and the biosynthesis of lipids and peptide antibiotics there are no documented examples of transesterification reactions that occur within a single polypeptide chain other than the reactions that participate in protein splicing. It seems therefore that the self-catalyzed intramolecular transesterifications observed in protein splicing represent a unique evolutionary adaptation for rearranging polypeptide backbones efficiently and without the necessity of an external energy source.Hedgehog proteins which bear a close evolutionary relationship to protein splicing elements may mediate an intramolecular transesterification similar to that seen in the second step of protein splicing.23 These proteins undergo an N?S acyl shift followed by intermolecular transesterification with the 3-OH group of cholesterol catalyzed by the autoprocessing domain.18 It has been suggested that the reaction with cholesterol occurs after prior intramolecular transesterification with a Cys residue found in a highly conserved region near the C-terminus of the autoprocessing domain thereby producing a branched ester intermediate similar to that seen in protein splicing.23 5 Peptide cleavage coupled to asparagine cyclization (Scheme 3) 5.1 Spontaneous cyclization of asparagine in peptides and proteins The cyclization of Asn residues is frequently observed in peptides and proteins (reviewed in ref.36). This reaction can follow either of two routes (i) attack by the peptide bond N on the carbonyl C of the Asn b-amide leading to deamidation [Fig. 9(a)] or (ii) attack by the Asn b-amide N on the carbonyl C of the peptide bond leading to peptide bond cleavage [Fig. 9(b)]. Asn cyclization can occur at extremes of pH and can become a significant side reaction during the deprotection of synthetic peptides under acidic conditions but it is also observed at neutral pH albeit at much lower rates.A critical determinant of the propensity for Asn cyclization is the conformation of the peptide bond that participates in the reaction particularly the dihedral angle y as well as the orientation of the Asn side chain defined by the dihedral angle c.36 The optimal y (+120° or 2120°) and c angles (+90° or 290°) for Asn cyclization are not those favored in typical polypeptides. As a result not all Asn residues in proteins are equally susceptible to cyclization. Asn residues adjacent to Gly which allows a much greater range of dihedral angles than amino acids with bulky side chains are particularly prone to the generation of Asp and iso-Asp residues (Fig. 9).37 Asn cyclization leading to peptide bond cleavage occurs at high temperatures in peptides in which Asn is followed by an amino acid with a bulky side chain,37 suggesting that steric constraints for attack by the side chain amide N are different than for attack by the peptide bond N.This mode of peptide cleavage has also been found in a-crystallin an unusually long-lived protein.38 5.2 Asparagine cyclization in protein splicing The presence of an Asn residue at the C-terminus of all inteins provides the opportunity for the excision of the intein by Asn cyclization. That such a reaction indeed occurs was demonstrated by analysis of the C-terminal residue of the excised DNA polymerase intein from Pyrococcus sp. GB-D. The excised intein was cleaved with CNBr at a Met residue that had been introduced near the C-terminus of the intein by a conservative amino acid replacement.The resulting C-terminal peptide was purified by reverse phase high performance liquid chromatography analyzed both by mass spectrometry and a specific color test for succinimides and compared with the corresponding chemically synthesized peptide.6,9 Similar experiments confirmed that the excised VMA intein of yeast also has a C-terminal aminosuccinimide residue suggesting that the same cleavage mechanism functions in the intein of hyperthermophiles and mesophiles regardless of whether protein splicing involves oxygen or thioesters.11 Scheme 3 The mechanism of Asn cyclization and the attendant cleavage of the peptide bond linking the intein to the C-extein is not yet clear. Evidence that the adjacent His residue plays an important role comes from the observation that its replacement with other amino acids leads to the accumulation of the branched intermediate.10 Examination of the crystal structure of the M.xenopi GyrA intein extended by modeling a C-terminal Thr shows that the imidazole moiety of His-197 can donate a proton cyclization. The most common mode of Asn cyclization in proteins is the one leading to deamidation. Even though its rate is slow under physiological conditions deamidation can be a significant reaction in long-lived proteins and can affect protein function owing to racemization and ring opening which leads to Fig. 9 Alternate modes of cyclization of Asn residues in peptides and proteins leading either to (a) deamidation or (b) peptide bond cleavage.(Reproduced from ref. 6. Copyright 1995 American Chemical Society). 383 Chemical Society Reviews 1998 volume 27 to the amido N of the scissile peptide bond to facilitate bond cleavage. Although His-197 can also form a H-bond with the carbonyl O of the Asn side chain no clear mechanism for enhancing the nucleophilicity of the Asn amide N is apparent. An alternative mechanism for Asn cyclization involving isomerization to an imide and the attack of the side chain carbonyl O on the peptide carbonyl C has been suggested.36 This mechanism would yield an isoimide rather than a succinimide but offers no advantages with respect to the catalytic groups available at the protein splicing active center. On the other hand an intriguing possibility that avoids the problem posed by the poor nucleophilicity of the Asn side chain involves the participation of the nucleophilic side chain of the N-terminal Cys or Ser residue of the intein which would function like the N-terminal nucleophile in amide hydrolysis by the mature N-terminal nucleophile amidohydrolases [see Fig.8(b)].28 This alternative mechanism for Asn cyclization which was suggested by Dr A.L. Nussbaum Harvard Medical School (personal communication) is illustrated in Fig. 10. The Cys or Fig. 10 Postulated mechanism for Asn cyclization in Step 3 of protein splicing with participation of the nucleophilic residue at the N-terminus of the intein and an Asn ester intermediate Ser at the N terminus of the intein released as an N-terminal nucleophile by the N?S or N?O acyl shift and transesterification reactions attacks the peptide bond linking the C-terminal Asn to the C-extein in a manner analogous to catalysis by N-terminal nucleophile amidohydrolases.This leads to the liberation of the C-extein and the formation of a macrocylic intermediate in which the N- and C-termini of the intein are linked by an ester bond. The esterified C-terminal Asn residue in the macrocyclic intermediate then cyclizes by attack of the Chemical Society Reviews 1998 volume 27 384 amide N on the ester carbonyl C leading to ester hydrolysis and a linear excised intein with a C-terminal aminosuccinimide residue. Experimental support for this mechanism comes from studies on the effect of amino acid substitutions on the splicing of chimeric proteins in which an intein has been inserted in a foreign context.When plasmids encoding such chimeric proteins are expressed in E. coli one observes not only protein splicing but also hydrolytic side reactions at either splice junction especially when normal protein splicing is attenuated by amino acid substitutions.3,9–11,39 These cleavage reactions have been shown to involve the same basic mechanisms as protein splicing11 and thus offer the opportunity to study the effect of amino acid substitutions on individual steps of the protein splicing mechanism.10,16 Using such experimental systems it was found that the replacement of the N-terminal nucleophilic residue of the intein with Ala causes at least a 25-fold reduction in the rate of Asn cyclization and attendant cleavage of the C-terminal splicing junction.10 Moreover when the transesterification step in precursors involving the yeast VMA intein is prevented by replacing the Cys residue at the downstream splice junction with Ala and the cyclization of the C-terminal Asn-454 is attenuated by replacing the adjacent His- 453 with Gln Asn cyclization and peptide bond cleavage at the downstream splice junction requires prior thiol-induced cleavage at the upstream splice junction.16 These results show that Asn cyclization coupled to peptide bond cleavage is much enhanced by a free intein N-terminus consisting of a nucleophilic amino acid residue.With the caveat that these observations were made in experimental systems where protein splicing was compromised by amino acid substitutions and may not be representative of normal inteins a plausible interpretation is that Asn cyclization is preceded by a nucleophilic displacement at the peptide bond carbonyl of the downstream splice junction involving an N-terminal nucleophile analogous to amide hydrolysis by N-terminal nucleophile amidohydrolases [compare Figs.8(b) and 10]. This would lead to a macrocyclic peptide ester involving the Asn carboxy group which would be resolved into a linear polypeptide by Asn cyclization (Fig. 10). Owing to the free rotation of ester bonds cyclization of Asn in the postulated ester intermediate would be subject to much less steric constraint than cyclization of the same Asn linked by a planar peptide bond to an amino acid whose side chain is esterified with the N-extein.On the other hand cleavage at the C-terminal splice junction proceeds slowly in the absence of N-terminal cleavage3,10,39 or when the N-terminal nucleophile is replaced by Ala,10 suggesting that the direct cyclization of Asn can also occur at a moderate rate. Nevertheless the notion that the major route to Asn cylization involves attack by an N-terminal nucleophile to form a macrocylic ester intermediate is attractive because it would assure that C-terminal cleavage can occur only after cleavage at the N-terminal splice junction and the formation of the branched intermediate. Such a mechanism would assure that the steps in protein splicing occur sequentially and thereby minimize side reactions leading to abortive cleavage.6 Uncatalyzed finishing reactions (Scheme 4) With the cyclization of the C-terminal Asn residue and the attendant cleavage of the peptide bond between the intein and the C-extein protein splicing has been achieved in the sense that the intein has been excised and the exteins have been linked together. However the excised intein contains an unnatural C-terminal residue aminosuccinimide and the exteins are linked by an unnatural ester bond. The completion of the splicing process requires the elimination of these unnatural features through some finishing reactions. Owing to the fact that the intein which catalyzed the earlier steps in protein splicing has now been excised and lacks specific affinity for the exteins the finishing reactions by necessity have to be spontaneous reactions that can proceed rapidly in the absence of catalysts.Scheme 4 6.1 Hydrolysis of C-terminal aminosuccinimides The rate of hydrolysis of C-terminal aminosuccinimides to Asn or iso-Asn [see Fig. 9(a)] was measured using synthetic tetrapeptides corresponding to the C-terminus of the DNA polymerase intein from Pyrococcus species GB-D.6 The rate of succinimide hydrolysis at 37 °C is strikingly pH-dependent with a half-life of 350 h at pH 5.5 which declines to 17 h at pH 7.4. These rates are considerably slower than the rates of hydrolysis of N-substituted cyclic imides produced at internal positions of polypeptide chains by nucleophilic attack of the peptide bond N on the carbonyl C of the Asn b-amide [Fig.9(a)] (summarized in ref. 6). In model peptides the hydrolysis of C-terminal aminosuccinimide proceeds with an activation energy (DH‡) of 23.2 kcal mol21 and with a relatively low entropy of activation (2TDS‡ = 2.1 kcal mol21).6 The relatively high stability of C-terminal aminosuccinimides implies that at least in mesophilic organisms a substantial fraction of the excised inteins carry C-terminal aminosuccinimide residues. This prediction is supported by the observation that about 50% of excised inteins isolated from E. coli transformed with an appropriate recombinant plasmid are terminated with aminosuccinimide.9 6.2 S?N and O?N acyl rearrangements in peptide esters In order to estimate the rates of S?N and O?N acyl shifts the depsipeptides N-(9-fluorenyl)methoxycarbonyl (Fmoc)-Gluo-Ser-Asp-Gly-Tyr-NH2 and Fmoc-Glu-s-Cys-Asp-Gly-Tyr- NH2 were synthesized and deprotected under acid conditions where the ester form is stable.(The designation Glu-o-Ser and Glu-s-Cys indicates that these amino acids are linked by a ester or thioester bond involving the side chains of Ser or Cys respectively rather than by a peptide bond.) The conversion of the deprotected depsipeptides to true peptides was estimated by reverse phase chromatographic separation of peptide esters and their amide rearrangement products in the presence of 0.1% trifluoroacetic acid which prevents further S?N and O?N acyl rearrangement.8 Both the rates and equilibrium constants for the S?N and O?N acyl rearrangements were found to increase rapidly with pH.At pH 6.0 the half-life of the oxygen ester Fmoc-Glu-o-Ser-Asp-Gly-Tyr-NH2 is 27 min; at pH 5.0 the half-life of the corresponding thioester Fmoc-Glu-s-Cys- Asp-Gly-Tyr-NH2 is only 24 s; at pH 7 the rates of acyl rearrangement of both esters are too fast for measurement by these techniques. The difference in acyl rearrangement rates for the oxygen esters and thioesters under identical conditions is about 1000. The activation energy (DH‡) for the O?N acyl shift at pH 6.0 is only 5.2 kcal mol21; that for the S?N acyl shift at pH 4.0 is even less (4.2 kcal mol21); and the entropies of activation (2TDS‡) are also very low (2700 and 1200 kcal mol21 respectively).The extremely rapid uncatalyzed rates and the virtual irreversibility of S?N and O?N acyl rearrangements at neutral pH makes these acyl shifts ideal finishing reactions that rapidly drive protein splicing to completion under physiological conditions even without catalysts. 7 Conclusions Protein splicing is a remarkable process in which nonfunctional polypeptides undergo complex and highly specific rearrangements to yield functional proteins. It may therefore seem surprising that a review of the chemical reactions that underlie protein splicing reveals only reaction types that are already well known to the protein chemist. However what is remarkable about protein splicing is not the novelty of the chemistry but the way in which protein splicing elements organize these chemical reactions into an effective pathway for remodeling proteins in a highly specific manner with a minimum of side reactions.Indeed it is amazing how protein splicing elements can catalyze this complex set of chemical transformations using less than 150 amino acid residues35,40 and no external cofactors. This contrasts with other types of post-translational modification which often require a complex enzymatic machinery proteases kinases phosphatases nucleotidyl transferases etc. Now that the chemical basis of protein splicing is well understood the major remaining challenge is to elucidate in detail how protein splicing elements can catalyze the reactions of the protein splicing pathway with such sparse means.As summarized in this review significant progress has already been made in this direction but many unanswered questions remain. Even though the crystal structures of two inteins have been solved,15,33 the proteins that were crystallized lack one or both exteins and therefore fail to reveal the most critical interactions needed for the catalytic process. Much information has been gained from studying the effect of amino acid substitutions on intein function,10,11 but observations based on systems perturbed by mutations must be interpreted with caution. As is evident from this review some mechanistic insights can be gained by comparing protein splicing with other self- catalyzed protein rearrangements such as hedgehog protein autoprocessing and glycosylasparaginase autocleavage but these serve primarily as starting points for further study.A promising experimental system for addressing structure–function relationships in protein splicing is the recently developed in vitro trans-splicing system based on highly truncated intein fragments.34,35 As shown in Fig. 11 N- and C-terminal segments of the RecA intein from Mycobacterium tuberculosis each about 100 amino acids long and fused to appropriate exteins can be reconstituted into a functional protein splicing element by renaturation from 6 M urea or guanidinium 385 Chemical Society Reviews 1998 volume 27 Fig. 11 Diagram illustrating the reconstitution and splicing of fusion proteins containing N- and C-terminal segments of the RecA intein linked to exteins A and B respectively chloride.34 Under mildly oxidizing conditions the intein segments are almost quantitatively reconstituted into a nonreactive covalent dimer in which the side chains of the Cys residues at the splice junctions are linked by a disulfide bond.Upon treatment with disulfide reducing agents the essential thiol groups are regenerated and protein splicing ensues. A semisynthetic protein splicing element can be generated by replacing the C-terminal intein segment with synthetic polypeptides corresponding to 35–50 of the C-terminal amino acids of the RecA intein.35 An especially valuable feature of this experimental system is that the reconstitution of the intein fragments to yield a functional protein splicing element can be studied independently of the protein splicing process per se thus allowing separate investigation of structural and catalytic aspects of the protein splicing element.Moreover the fact that a semisynthetic protein splicing element is now available in which the C-terminal segment of the intein consists entirely of a synthetic polypeptide35 offers the opportunity for probing the structure and function of the protein splicing active center by substituting other amino acids including unnatural amino acids and structural probes at specific positions of the polypeptide. Experimental approaches of this type should rapidly advance our understanding of protein splicing so that it may soon be possible to write a definitive review on the enzymatic basis of protein splicing.8 Acknowledgements I am deeply grateful to my collaborators at the Boston Biomedical Research Institute and New England Biolabs who are named on our joint publications cited in this review for their important contributions to the elucidation of the mechanism of protein splicing and for the pleasure of doing research together. I also thank Alex Nussbaum for stimulating discussions on the mechanism of asparagine cyclization to Ken Mills and Peter Coleman for their constructive comments on the manuscript to Thomas Klabunde for providing the atomic coordinates of the M. xenopi GyrA intein crystal structure prior to public release and to Don Comb for introducing me to protein splicing.The research in my laboratory was funded by fellowships from New England Biolabs and a research grant (R01 GM55875) from the Chemical Society Reviews 1998 volume 27 386 National Institutes of Health. Their research support is greatly appreciated. 9 References 1 F. B. Perler M.-Q. Xu and H. Paulus Curr. Opin. Chem. Biol. 1997 1 292. 2 F. B. Perler G. J. Olsen and E. Adam Nucl. Acids Res. 1997 25 1087. 3 M.-Q. Xu M. W. Southworth F. B. Mersha L. J. Hornstra and F. B. Perler Cell 1993 75 1371. 4 M. Kawasaki S.-i. Makino H. Matsuzawa Y. Satow Y. Ohya and Y. Anraku Biochem. Biophys. Res. Comm. 1996 222 827. 5 M. Kawasaki Y. Satow Y. Ohya and Y. Anraku FEBS Lett. 1997 412 518. 6 Y. Shao M.-Q. Xu and H. Paulus Biochemistry 1995 34 10844.7 Y. Shao M.-Q. Xu and H. Paulus Biochemistry 1996 35 3810. 8 Y. Shao and H. Paulus J. Peptide Res. 1997 50 193. 9 M.-Q. Xu D. G. Comb H. Paulus C. J. Noren Y. Shao and F. B. Perler EMBO J. 1994 13 5517. 10 M.-Q. Xu and F. B. Perler EMBO J. 1996 15 5146. 11 S. Chong Y. Shao H. Paulus J. Benner F. B. Perler and M. Q. Xu J. Biol. Chem. 1996 271 22159. 12 K. Iwai and T. Ando Methods Enzymol. 1967 11 262. 13 W. P. Jencks S. Cordes and J. Carriulo J. Biol. Chem. 1960 235 15 T. Klabunde S. Sharma A. Telenti W. R. Jacobs and J. C. Sacchettini 3608. 14 S. Chong F. B. Mersha D. G. Comb M. E. Scott D. Landry L. M. Vence F. B. Perler J. Benner R. Kucera C. A. Hirvonen J. J. Pelletier H. Paulus and M. Q. Xu Gene 1997 192 271. Nature Struct.Biol. 1998 5 31. 16 S. Chong K. S. Williams C. Wotkovicz and M.-Q. Xu J. Biol. Chem. 1998 273 10567. 17 S. Pietrokovski Protein Sci. 1994 3 2340. 18 J. A. Porter K. E. Young and P. A. Beachy Science 1996 274 255. 19 E. V. Koonin Trends Biochem. Sci. 1995 20 141. 20 J. A. Porter D. P. Kessler S. C. Ekker K. E. Young J. J. Lee K. Moses and P. A. Beachy Nature 1995 374 363. 21 J. A. Porter S. C. Ekker W. J. Park D. P. v. Kessler K. E. Young C. H. Chen Y. Ma A. S. Woods R. J. Cotter E. V. Koonin and P. A. Beachy Cell 1996 86 21. 22 S. Pietrokovski Protein Sci. 1998 7 64. 23 J. Z. Dalgaard M. J. Moser R. Hughey and I. S. Mian J. Comput. Biol. 1997 4 193. 24 T. M. T. Hall J. A. Porter K. E. Young E. V. Koonin P. E. Beachy and D. J. Leahy Cell 1997 91 85. 25 R. Tikkanen A. Riikonen C. Oinonen J. Rouvinen and L. Peltonen EMBO J. 1996 15 2954. 26 C. Guan T. Cui V. Rao W. Liao J. Benner C. L. Lin and D. Comb J. Biol. Chem. 1996 271 1732. 27 C. Guan Y. Liu Y. Shao T. Cui W. Liao A. Ewel R. Whitaker and H. Paulus J. Biol. Chem. 1998 273 9695. 28 J. A. Brannigan G. Dodson H. J. Duggleby P. C. E. Moody J. L. Smith D. R. Tomchick and A. G. Murzin Nature 1995 278 416. 29 Y. Liu C. Guan and N. N. Aronson J. Biol. Chem. 1998 273 9688. 30 P. A. Recsei Q. K. Huynh and E. E. Snell Proc. Natl. Acad. Sci. USA 1983 80 973. 31 P. Vanderslice W. C. Copeland and J. D. Robertus J. Biol. Chem. 1988 263 10583. 32 A. Albert V. Dhanaraj U. Genschel G. Khan M. K. Ranjee R. Pulido B. L. Sibanda F. v. Delft M. Witty T. L. Blundell A. G. Smith and C. Abell Nature Struct. Biol. 1998 5 289. 33 X. Duan F. S. Gimble and F. A. Quiocho Cell 1997 89 555. 34 K. V. Mills B. M. Lews S.-Q. Jiang and H. Paulus Proc. Natl. Acad. Sci. USA 1998 95 3453. 35 B. M. Lew K. V. Mills and H. Paulus J. Biol. Chem. 1998 273 15 887. 36 S. Clarke Int. J. Peptide Protein Res. 1987 30 808. 37 T. Geiger and S. Clarke J. Biol. Chem. 1987 262 785. 39 C. E. M. Voorter W. A. de Haard-Hoekman P. J. M. Van den Oetelaar H. Bloemendal and W. W. de Jong J. Biol. Chem. 1988 263 19020. 39 K. Shingledecker S. Q. Jiang and H. Paulus Gene 1998 207 187. 40 V. Derbyshire D. W. Wood W. Wu J. T. Dansereau L. Z. Dalgaard and M. Belfort Proc. Natl. Acad. Sci. USA 1997 94 11466. Received 2nd June 1998 Accepted 17th June 1998

ISSN:0306-0012    

DOI:10.1039/a827375z

出版商:RSC    

年代:1998

数据来源: RSC

摘要:

Mechanistic elucidation of small molecule – transition metal interactions by kinetic techniques Siegfried Schindler Colin D. Hubbard and Rudi van Eldik* Institute for Inorganic Chemistry University of Erlangen-N�urnberg Egerlandstr. 1 91058 Erlangen Germany This article focuses on the interaction of small molecules such as dioxygen carbon monoxide carbon dioxide sulfur dioxide and nitric oxide with transition metal centres in solution with the objective of establishing reaction mechanisms. The systems described are themselves biochemical reactions model reactions for them environmentally relevant reactions or are reactions that are important for industrial processes. The experimental approach is principally based on appropriately designed kinetics measurements using a variety of techniques with considerable emphasis on the application of high pressure methods.1 Introduction Several small molecules such as dioxygen carbon monoxide carbon dioxide sulfur dioxide and nitric oxide play an important role in biochemical systems in environmentally significant reactions and/or in technically important processes. The activation of these small molecules usually involves a direct or indirect interaction with the transition metal centre situated in a very specific coordination environment. It is a subject that has received considerable attention from numerous groups working on model or catalytically active systems and many examples are given in the quoted references.1–4 Siegfried Schindler studied Chemistry at the Technical University of Darmstadt (Germany) where he obtained his PhD under the supervision of Professor Horst Elias in 1989.He then became a research associate of Dr Carol Creutz at Brookhaven National Laboratory (Long Island New York). He returned to Germany in 1993 and joined the group of Professor Rudi van Eldik first at the University of Witten/Herdecke and later at the University of Erlangen-N�unberg. He completed his “Habilitation” in inorganic chemistry in 1997 and is now Privatdozent at the University of Erlangen-N�urnberg. Colin D. Hubbard a native of Norfolk England studied Chemistry at the University of Sheffield and obtained his PhD under the supervision of Professor Ralph G. Wilkins. C. D. Hubbard S. Schindler We have prepared a progress report that covers a divergent set of reaction types and as requested mostly represents recent investigations from our own laboratories with emphasis on the application of high pressure techniques.We will focus on the interactions of transition metal centres with small molecules with the purpose of developing an understanding of the detailed reaction mechanisms. In the majority of cases the transition metal centres cited will be from the first row and are present in metalloproteins themselves or are complexes designed to mimic the function of the metalloproteins or can be simple aquated ions. In order to endeavour to understand the mechanisms of the reactions involved detailed kinetic measurements are performed as a function of temperature and pressure.Since appropriate instrumentation is not widely available thermodynamic and kinetic measurements on reactions in solution as a function of pressure are relatively rare. The goal of equilibrium and kinetic studies at different elevated pressures is to obtain information on the overall volume change (DV) and volume of activation (DV‡) associated with the reaction under study. This enables an analysis to be performed of the chemical events in terms of volume changes along the reaction coordinate. In the particular case of inorganic reactions the potential of this approach was noted a few decades ago.5 The kinetic character- Following post-doctoral research a MIT and Cornell (Professor G. G. Hammes) and Berkeley (Professor J.F. Kirsch) he was appointed to the faculty of chemistry at the University of New Hampshire becoming a tenured Associate Professor in 1972 and Professor in 1979. He was a frequent research visitor to the University of Leicester (Dr J. Burgess) and to the University of Witten/Herdecke (Professor R. van Eldik). Since 1994 he has been on the staff of the Institute for Inorganic Chemistry at the University of Erlangen-N�urnberg. Rudi van Eldik was born in Amsterdam but grew up in South Africa where he obtained his PhD at the Potchefstroom University in 1971. He worked as a post-doctoral fellow with Professor Gordon M. Harris (Buffalo New York) and Professor Hartwig Kelm (Frankfurt am Main). He completed his “Habilitation” in physical chemistry at the University of Frankfurt am Main in 1982 and moved to the University of Witten/Herdecke as Professor of Inorganic Chemistry in 1987.In 1994 he accepted a call to the University of Erlangen-N�urnberg where he is presently Professor of Inorganic and Analytical Chemistry. In 1997 the Potchefstroom University awarded him an honorary doctorate. R. van Eldik 387 Chemical Society Reviews 1998 volume 27 isations (i.e. rate laws) together with the thermal and pressure activation parameters (i.e. DH‡ DS‡ and DV‡) together with other information lead to the postulation of reaction mechanisms. The basic principles at issue in high pressure kinetics have been the subject of several accounts.5–8 The equation developed for the pressure dependence of the equilibrium constant for a reaction system can be transformed into (dln k/dP)T = 2DV‡/ RT since within transition state theory an equilibrium is assumed between the activated complex and the reactants.Upon integration this equation can be applied to determine the volume of activation (DV‡) by plotting the natural logarithm of the rate constant k versus applied pressure P. If the plot is linear and it frequently is if the pressure is < 200 MPa then DV‡ can be obtained directly from the slope of the plot. Thus a reaction is accelerated by pressure when there is a volume reduction upon reaching the transition state i.e. DV‡ is negative and retarded by pressure when DV‡ is positive (expansion upon reaching the transition state).These in turn are indicative in the absence of any solvational changes of associative and dissociative mechanisms respectively. A reaction that is completely insensitive to pressure will yield DV‡ equal to zero. In many reactions DV‡ has a magnitude within the range of +30 to 230 cm3 mol21. These values correspond to rate retardations and accelerations of about four fold at 100 MPa. The diagnostic mechanistic value of the magnitude of these values has to be carefully considered with both intrinsic and solvational contributions taken into account. Further details on the interpretation of the volume of activation are provided in the references.5–8 A volume profile charts strictly partial molar volume changes along the reaction coordinate as a free energy profile illustrates Gibbs free energy changes along the reaction coordinate.When actual partial molar volumes of either products and reactants are known or when the partial molar volume of one of the latter together with the reaction volume are known then the volume profile is on an absolute rather than a relative basis. Mechanistic conclusions based on the interpretation of volume profiles will be demonstrated in this account. The extreme lability of many of the reactions reported herein requires the utilisation of rapid reaction techniques such as stopped-flow temperature-jump and flash photolysis as well as conventional UV–VIS spectrophotometry and other standard methods and instrumentation. Establishment of volume profiles for unsymmetrical reactions has been an integral part of our mechanistic characterisation in many cases an approach providing considerable insight.Other spectroscopic techniques for example NMR ESR as well as X-ray structural determinations often provide vital or supporting evidence for a particular mechanism and complementary information. Some examples of these investigations perfor laboratories are summarised in this report. The nature of this review precludes a full bibliographic survey; hence the context of such studies and additional literature are best obtained from the references cited. 2 Reactions with dioxygen and carbon monoxide The interaction of dioxygen and carbon monoxide with metal centres is usually investigated in parallel since the binding to the metal centre can involve a similar mechanism with the difference that the subsequent reactions involving the activation of the bound small molecule frequently only occur for dioxygen.Model complexes for dioxygen carrier proteins as well as for redox active metalloenzymes have been synthesised and investigated.2–4 In a very simplified case it was possible to construct a volume profile for the reversible binding of dioxygen to a Co(ii) macrocycle viz. L = hexamethylcyclam to produce (L)Co- O2 2+ which is a Co(iii)-superoxo species [eqn. (1)].9 Chemical Society Reviews 1998 volume 27 388 (1) CoII(L)(H2O)2 2+ + O2 " CoIII(L)(H2O)(O22)2+ + H2O The kinetics of the overall reaction could be studied by flashphotolysis since the dioxygen complex can be photo-dissociated and the subsequent reequilibration could be followed in the microsecond time range.A combination of the activation volumes for the binding and release of dioxygen results in a value of the reaction volume that is in very good agreement with that determined directly from equilibrium measurements as a function of pressure. The volume profile for the reaction is shown in Fig. 1. The small volume of activation associated with the forward reaction could be interpreted as evidence for a ratelimiting interchange of the ligands dioxygen for water which is followed by an intramolecular electron-transfer reaction between Co(ii) and O2 to form CoIII-O22 a superoxo species. It is the latter process that accounts for the large volume reduction en route to the reaction products.Thus during flash-photolysis electron transfer in the reverse direction occurs due to irradiation into the CT (charge-transfer) band. This is followed by the rapid release of dioxygen. The mechanism of the binding of small molecules such as O2 and CO to ferrous hemes and hemoproteins which are more complex systems has been the focus of many investigations in recent years.1–4 Model heme complexes were usually employed in an effort to improve the understanding of the reactions of the corresponding proteins. Two model heme systems (monochelated protoheme (MCPH) and protoheme dimethyl ester) and various neutral ligands were used to study the bimolecular addition to the five- coordinate ferrous model heme complexes using two different photolysis techniques.10 The reported DV‡ data correlate well with the addition rate constants.For the slower reactions bond formation is rate determining and results in negative DV‡ values while for the faster reactions the processes become diffusion controlled and are slowed down by increasing pressure due to the large increase in solvent (toluene) viscosity. In a subsequent study the reaction of CO with MCPH was studied as a function of pressure in a very viscous medium.11 The data showed that a changeover in ratedetermining step occurred from one involving bond formation to a diffusion-controlled process upon increasing the pressure i.e from a DV‡ value of 29.6 to a value of +7.1 cm3 mol21.These are examples of a changeover from activation control to diffusion control over reaction rates where the mechanistic diagnosis has emerged from kinetic measurements at high pressures. In another study the binding of CO to lacunar Fe(ii) complexes was studied in detail as a function of temperature and pressure in acetonitrile.12,13 The overall reaction for the binding of CO to [FeII(PhBzXy)](PF6)2 can be summarised as in Scheme 1. The volume profile for this reaction is presented in Fig. 2. The large volume collapse associated with the forward reaction suggests that in the transition state CO “disappears” L = hexamethyl cyclam at 298 K Fig. 1 Volume profile for the reaction of dioxygen with the Co(ii)L complex CoII(L)(H2O)2 2+ + O2 " CoIII(L)(H2O)(O22)2+ + H2O where completely into the ligand pocket during partial Fe–CO bond formation followed by a high spin to low spin transition on Fe(ii) during which the metal centre moves into the ligand plane and accounts for the subsequent volume decrease.13 Similar kinetic techniques were applied to study the effect of pressure on the bimolecular rate constant for the reaction of sperm whale myoglobin with a series of neutral ligands in water.10,14 It followed from the data that only the reaction with CO is characterized by a negative DV‡ value which is typical for a bond formation process.The positive DV‡ values found for the other ligands were ascribed to the entry of the ligand into the protein matrix which will be accompanied by pronounced desolvation and presumably conformational changes on the protein chain.The effect of pressure on the escape of the ligand Scheme 1 Fig. 2 Volume profile for the reaction of [FeII(PhBzXy)](PF6)2 with CO in acetonitrile at 298 K from the protein-separated pair resulted in distinctly positive DV‡ values.10 These values are consistent with the notion of a “gate” that operates in both directions of the process. 2 The large difference in DV‡ observed for the binding between O2 and CO to deoxymyoglobin stimulated a more comprehensive study to be carried out for both systems.14 The profiles are shown in Fig. 3. The volume profile for the binding of O2 is characterised by a substantial increase in volume in going from the reactant to the transition state followed by a significant volume reduction on going to the product state.The volume increase was ascribed to rate-determining movement of O through the protein to the heme pocket which may involve hydrogen bonding to the distal histidine as well as desolvation. This step is followed by rapid bond formation with the Fe(ii) centre during which a change in spin state from high to low the movement of the Fe(ii) centre into the porphyrin plane and the associated conformational changes account for the drastic volume reduction. The overall reaction volume of 218 cm3 mol21 demonstrates the large volume reduction caused by the binding of O2. The volume profile for the binding of CO shows a considerable volume decrease on going from the reactant to the transition state which has been ascribed to rate-determining bond formation.The reverse bond cleavage reaction is accompanied by a volume decrease which may be related to the different bonding mode of CO compared with O2. This difference in bonding mode must also account for the much smaller absolute reaction volume observed in this case. A volume profile was also generated (Fig. 4) for the binding of dioxygen to hemerythrin,15 for which the overall reaction is given in Scheme 2. The DV‡ values for the overall “on” and “off” reactions as well as the overall reaction volume are ca. twice the magnitudes of those for the corresponding myoglobin case. In the hemerythrin system two Fe(ii) centres are oxidized to Fe(iii) during which dioxygen is reduced and bound as hydroperoxide to one Fe(iii) center.The DV‡ on value can partly Fig. 3 Volume profile for the reactions of CO and O2 with myoglobin at 298 K Mb + O2 " MbO2 Mb + CO " MbCO Hr + O Fig. 4 Volume profile for the binding of dioxygen to hemerythrin at 298 K 2 " HrO2 389 Chemical Society Reviews 1998 volume 27 be accounted for in terms of desolvation of oxygen during its entrance into the protein. The value is however such that it suggests some form of dynamic “reathing”motion of the protein that momentarily causes an opening up of a cleft and enables oxygen to enter the protein. The significant volume decrease that occurs following the formation of the transition state can be ascribed to the oxidation of the Fe centres and the reduction of O2 to O2 22.The fact that the overall volume collapse is almost double that observed for the oxygenation of myoglobin may indicate similar structural features in oxyhemerythrin and oxymyoglobin. This would suggest that a description of the bonding mode as FeIII-O22 or FeIII-O2H (H from histidine E7) instead of FeII-O2 may be more appropriate for oxymyoglobin. A suitable model for the oxygen carrier protein hemerythrin is [Fe2(Et-HPTB)(OBz)](BF4)2 Et-HPTB = N,N,NA,N-tetrakis[( N-ethyl-2-benzimidazolyl)methyl]-2-hydroxy-1,3-diaminopropane OBz = benzoate; it can mimic the formation of a binuclear peroxo iron complex in the natural system (Scheme 3).16 In this case it was possible to follow the irreversible uptake of dioxygen.The measured value of 212.8 cm3 mol21 for the activation volume of the reaction together with the negative value of the activation entropy confirms the highly structured nature of the transition state. Oxidation reactions of chelated Fe(ii) complexes are all markedly accelerated by pressure and accompanied by negative volumes of activation.8 These can be ascribed to the binding of dioxygen that is accompanied by the oxidation of Fe(ii) to Fe(iii) and the reduction of dioxygen to superoxide and peroxide (eqns. (2) and (3)) processes that are all expected to lead to a decrease in partial molar volume. (2) FeII(L)(H2O) + O2 " FeIII(L)(O22) + H2O FeII(L)(H2O) + FeIII(L)(O22) ? (L)FeIII(O2 22)FeIII(L) + H2O (3) In a recent reinvestigation of the FeII(edta) oxidation reaction where L = edta in eqns.(2) and (3) it was possible to resolve the different reaction steps that form part of the oxidation process and to interpret the negative volumes of activation in a more detailed way.17 Tyrosinase a dinuclear copper protein is a monoxygenase which activates dioxygen for the ortho hydroxylation of monophenols. Efforts to model this protein functionally led to the development of a series of dinuclear copper model complexes.4,18,19 It has been established that during the reaction of a series of dinuclear copper(i) complexes with dioxygen intramolecular ligand hydroxylation occurs leading to pheno- Scheme 2 Scheme 3 Chemical Society Reviews 1998 volume 27 390 late bridged copper(ii) complexes.4,18,19 In a kinetic study of the reaction of dioxygen with [Cu2(H-BPB-H)(CH3CN)2]- (PF6)2 (H-BPB-H = 1,3-bis[N-(2-pyridylethyl)formimidoyl]- benzene) a dinuclear copper Schiff base complex in acetone it was not possible (even at low temperatures) to observe the postulated copper peroxo intermediate complex spectroscopically (Scheme 4).However kinetic findings provided indirect evidence for the transient existence of this complex.18 Even though kinetic investigations were complicated because of side reactions it was possible to extract appropriate kinetic data and the activation parameters DH‡ = 47 ± 9 kJ mol21 DS‡ = 253 ± 11 J K21 mol21 and DV‡ = 29.5 ± 0.5 cm3 mol21 were obtained. The results support a mechanism that is very similar to that proposed in our study of the reaction of [Cu2(mac)(CH3CN)2](PF6)2 (mac = 3,6,9,17,20,23-hexaazatricyclo [23.3.1.111 15]triaconta- 1(29),2,9,11(30),12,14,16,23,25,27-decaene) in methanol with dioxygen.19 In a rate determining step a peroxo complex is formed as an intermediate which then reacts in a very fast reaction to give the final product.The negative DS‡ and DV‡ values support the idea of a highly structured transition state that is formed as a result of the presence of the highly reactive and easily oxidizable cuprous species. The negative volume of activation is a strong indication of copper–oxygen bond formation that is accompanied by electron transfer to produce the Cu(ii)-O2-Cu(ii) peroxo intermediate. The formal oxidation of Cu(i) to Cu(ii) and reduction of O2 to O2 22 are expected to be accompanied by a significant volume collapse partly due to intrinsic and solvational volume changes.The activation parameters for the reaction of dioxygen with [Cu2(H-BPBH)( CH3CN)2](PF6)2 compare well with those obtained for the corresponding reaction of [Cu2(mac)(CH3CN)2](PF6)2 (DH‡ = 32 ± 2 kJ mol21 DS‡ = 2146 ± 8 J K21 mol21 and DV‡ = 221 ± 1 cm3 mol21). The higher values for the activation entropy and activation volume in the latter case are probably caused by larger geometrical rearrangements for the macrocyclic complex compared to those for the open complex [Cu2(H-BPB-H)(CH3CN)2](PF6)2. It was assumed that the rate determining step within a whole sequence of steps must be the attack of dioxygen on the first copper(i) ion accompanied by an electron transfer step (leading formally to a copper(ii) superoxo complex).This species then reacts very quickly to produce the peroxo complex and then again in a fast reaction sequence to give the product. The negative DV‡ values found for the oxidation of dinuclear copper Schiff base complexes are close to the average value of 222 ± 2 cm3 mol21 reported for the oxidation of Cu(i)(phen)2 by dioxygen in aqueous medium.20 In Scheme 4 2Cu(i)-O2 species. the latter study it was concluded that the significantly negative volume of activation mainly arises from the large volume reduction associated with the formation of the intermediate (phen) The mononuclear copper(i) complex of the tripodal amine ligand Me6tren (tris(2-dimethylaminoethyl)amine) reacts reversibly with oxygen at low temperatures to form superoxo and peroxo complexes according to eqns (4) and (5),21 where L = Me6tren and R = CH3 or C2H5.The reaction could be followed by employing a low temperature stopped-flow instrument and the spectral changes that occurred during the formation of the superoxo complex (lmax = 412 nm) at 290 °C are shown in Fig. 5 (absorbance vs. time trace is shown as an insertion). The formation of the superoxo species was predictably much faster at higher temperatures and therefore only the decomposition of the superoxo complex and the formation of the peroxo complex could be observed in the temperature range 290 to 230 °C.(4) [Cu(L)RCN]+ + O2 " [Cu(L)O2]+ + RCN (5) [Cu(L)O2]+ + [Cu(L)RCN]+ " [Cu2(L)2O2]2+ + RCN 3 Interaction with carbon dioxide (6) Early work on the binding of CO2 by inert transition metal hydroxo complexes showed22 that the transition state consists of a species with incipient oxygen–carbon bond formation (oxygen of the hydroxo ligand) and no breakage of the metal oxygen bond. The particular complexes employed to generate the bicarbonate species and upon ionisation the carbonato-complex were hydroxo complexes of M(NH3)5 3+ where M = Co Rh or Ir (eqn. (6)).22 From a study of the pressure dependence of the kinetics of both the forward and reverse reactions it could be concluded that on a volume basis the bond formation and breakage respectively are approximately 50% completed in the transition state.M(NH3)5OH2+ + CO2 ? M(NH3)5OCO2 + + H+ The active centre of the zinc containing metalloenzyme carbonic anhydrase (CA) is comprised of three histidine residues and one water molecule coordinated to zinc in a Fig. 5 Reaction of [Cu(Me6tren)(CH3CN)]PF6 with dioxygen at 290 °C in dry propionitrile. The insert shows the absorbance vs. time trace at 412 nm. slightly distorted tetrahedral geometry. The reaction catalysed by CA is given in eqn. (7). Catalytic activity is integrally related to the ionisation (pKa value ca. 7) of the coordinated water molecule and for human CA II the mechanism is known as the zinc hydroxide mechanism which has been described and modelled theoretically in considerable detail.23 CO2 + H2O " HCO32 + H+ (7) A simplistic expectation would be that model metal complexes designed to mimic the means of activation of CO2 for catalysis of its hydration and of bicarbonate for the reverse dehydration reaction would possess corresponding coordinating units and geometry.It prevails that the highest catalytic activity observed to date for the hydration reaction occurs when a five coordinate Zn(ii) complex ([12]aneN4–Zn(II)-OH2)(ClO4)2 ([12]aneN4 = 1,4,7,10-tetraazacyclododecane)) in its deprotonated form was employed in solution in the neutral pH region.24 Dehydration was also accelerated by the same complex (in its +2 form). Interestingly a model zinc complex more closely resembling the zinc centre of CA viz.[12]aneN3–Zn(II),25 is less effective catalytically. This is associated with the ability or lack thereof of the formation of a bicarbonate intermediate the five coordinate complex of [12]aneN4 does not allow bicarbonate chelation and it is this fact coupled with the apparently appropriate pKa value of the coordinated water which together appear to provide the catalytic supremacy. The CA catalysed reactions themselves have been studied in intricate detail by many investigators using a wide variety of techniques and have formed the subject of several theoretical calculations and computer simulations.23 The application of high pressure kinetic measurements provided further mechanistic distinction than was previously available.26 An encouraging close agreement was obtained between the reaction volume for the uncatalysed reaction obtained earlier27 and that derived from kinetic measurements on the catalysed reaction in both directions.The first complete detailed volume profile (see Fig. 6) for an enzyme catalysed reaction was generated. The Zn2+ bound hydroxy moiety subjects the carbon of CO2 to nucleophilic attack resulting in the formation of an oxygen– carbon bond and the results are consistent with a unidentate bonding of bicarbonate. For this process the transition state lies 391 Chemical Society Reviews 1998 volume 27 approximately halfway between the reactant and product states (see left part of volume profile). The substitution of coordinated water by bicarbonate tends more toward a limiting D mechanism (see right part of volume profile) than would be predicted on the basis of the coordination chemistry of aquated Zn2+ which may result from the influence of the environment of the active centre of the enzyme.26 Because of the complexity of this particular system readers are kindly advised to consult the original reference for more details.Fujita and van Eldik and others28 have used a wide range of methods to study cobalt complexes with tetraazamacrocyclic ligands as potential catalysts for the reduction of CO2. The species 2 in CH3CN leads to a five- 3CN)+ formed through addition of CH3CN (see eqn. (8) interaction of the low spin CoI(HMD)+ HMD = 5,7,7,12,14,14-hexamethyl-1,4,8,11-tetraazacyclotetradecane-4,11-diene with CO coordinate species CoI(HMD)(CO2)+ which is in equilibrium with a six-coordinate complex ion CoIII(HMD)(CO2 22)- (CH and (9)).(8) CoI(HMD)+ + CO2 " CoI(HMD)(CO2)+ CoI(HMD)(CO2)+ + CH3CN " CoIII(HMD) (CO2 22)(CH3CN)+ (9) Results from an XANES study together with other information provide a clear indication that in the six-coordinate complex cobalt is in the oxidation state plus three meaning that Scheme 5 Fig. 6 Volume profile for the carbonic anhydrase catalyzed hydration of CO2 and dehydration of HCO32 at 298 K CO2 + H2O " HCO32 + H+ Chemical Society Reviews 1998 volume 27 392 the complex ion is Co(iii)-CO2 22 (carboxylate). Hence the initial cobalt complex has reduced the bound CO2. Since the sixcoordinate species is yellow and the five-coordinate species is purple the change of coordination number equilibrium can be studied readily by UV–VIS spectrophotometry; the thermodynamic parameters are DH° = 229 kJ mol21 DS° = 2113 J mol21 K21 and DV° = 217.7 cm3 mol21.The latter two are mutually compatible and consistent with a highly ordered and compact six-coordinate complex ion. It has been proposed that a major part of the volume decrease arises from the intramolecular electron transfer process accompanied by a short- 3CN addition. ening of the Co–CO2 bond (as supported by XANES and EXAFS studies) and an increase in electrostriction. Only a relatively minor contribution to the large negative reaction volume results from the intrinsic effect of CH Even though it is well known that Ni(0) complexes can react with carbon dioxide our results showed that the complex [Ni(bpy)(COD)] COD = cyclooctadiene is unreactive towards carbon dioxide.29 For the reaction of [Ni(bpy)(COD)] with propionaldehyde and carbon dioxide according to Scheme 5 a thorough kinetic study clearly demonstrated that carbon dioxide does not bind to the Ni(0) center but instead reacts with the activated propionaldehyde of the product complex [Ni(bpy)- (propionaldehyde)].29 4 Interaction with sulfur dioxide and nitric oxide The catalytic role of metal ions and complexes in the autoxidation of sulfur and nitrogen oxides has been studied in detail because of the importance of such reactions in the treatment of gaseous effluents of coal-fired power plants and in atmospheric oxidation processes.30,31 The application of fast kinetic and spectroscopic techniques has assisted in defining the underlying reaction mechanism.It was possible to identify and characterise the redox cycling of the catalytically active metal ions and complexes during the autoxidation of SOx species.32,33 A general mechanism in terms of the participating reactions based on the detailed kinetic studies referred to above is presented in eqn. (10)–(19) and illustrates the important role of redox cycling of the metal ion in such catalytic processes. (10) Mn+ + HSO32 ? M(n21)+ + SO32 + H+ (11) (12) SO32 + O2 ? SO52 M(n21)+ + SO52 + H+ ? Mn+ + HSO52 (13) M(n21)+ + HSO52 ? Mn+ + SO42 + OH2 (14) M(n21)+ + SO42 ? Mn+ + SO4 22 SO52 + HSO32 ? HSO52 + SO32 SO52 + HSO32 ? SO4 22 + SO42 + H+ SO42 + HSO32 ? SO4 22 + SO32 + H+ SO52 + SO52 ? S2O8 22 + O2 (15) (16) (17) (18) (19) HSO52 + HSO32 + H+ ? 2SO4 22 + 3H+ Nitrogen oxides as well as mixed sulfur–nitrogen oxides produced during the spontaneous or metal catalysed interaction of NO/HONO/NO22 with SO2/HSO32/SO3 22 can interact with metal ions and complexes and affect their oxidation state.In general Fe(ii) complexes can bind NO rapidly to produce species of the type FeIII(L)NO2 which decompose to FeIII(L) and NO2 where the latter species hydrolyses to N2O (eqns. (20) and (21) where L = polyaminecarboxylate chelate). (20) FeII(L)(H2O) + NO " FeII(L)NO + H2O (21) FeII(L)NO * FeIII(L)(NO2) ? FeIII(L) + NO2 Coordinated NO can interact with bisulfite to produce hydroxylaminedisulfonate (HADS) N2O and iron(iii).34,35 Manganese( iii) and iron(iii) ions and complexes can affect the hydrolysis reactions of such mixed sulfur–nitrogen oxides.36,37 For instance HADS rapidly hydrolyses under such conditions to hydroxylaminemonosulfonate during which the metal ions are reduced to the plus two oxidation state.In one case evidence for the redox cycling of manganese(ii/iii) in the presence of a sulfur–nitrogen oxide was observed.38 The role of these reactions in atmospheric oxidation processes in water vapour or aerosols is presently unknown despite their potential importance especially in terms of controlling the oxidation state of the catalytic metal ions.5 Conclusions It was our intention to provide a succinct account of the means by which the interaction of small molecules with transition metal centres occurs. At the initial level of mechanistic understanding achieved so far the capability of monitoring UV–VIS spectral changes over a wide time range and under a variety of conditions results in an opportunity to obtain sufficient information for mechanistic proposals. Techniques other than kinetic are also vital in the establishment of mechanism. Even though the chemistry of the reported reactions is quite different the experimental approach and techniques employed in the mechanistic resolutions are very similar. We believe that the overall approach can be exploited for the study of the interaction of other small molecules with transition metal centres.The cited literature could be consulted for provision of a broader context and additional background for the investigations described herein. 6 Acknowledgements The authors gratefully acknowledge the contributions of many students and collaborators cited in the references as well as financial support for their work from the Deutsche Forschungsgemeinschaft Bundesministerium f�ur Bildung Wissenschaft Forschung und Technologie Volkswagen-Stiftung and Fonds der Chemischen Industrie. 7 References 1 A. M. Valentine and S. J. Lippard J. Chem. Soc. Dalton Trans. 1997 3925. 2 L. Que J. Chem. Soc. Dalton Trans. 1997 3933. 3 W. B. Tolman Acc. Chem.Res. 1997 30 227. 4 K. D. Karlin S. Kaderli and A. D. Zuberb�uhler Acc. Chem. Res. 1997 30 139. 5 D. R. Stranks Pure Appl. Chem. 1974 38 303. 6 S. F. Lincoln and A. E. Merbach Adv. Inorg. Chem. 1995 42 1. 7 C. D. Hubbard and R. van Eldik in Chemistry under Extreme or Nonclassical Conditions R. van Eldik and C. D. Hubbard (eds.) Wiley 8 A. Drljaca C. D. Hubbard R. van Eldik T. Asano M. V. Basilevsky 9 M. Zhang R. van Eldik J. H. Espenson and A. Bakac Inorg. Chem. New York 1997 ch. 2. and W. J. le Noble Chem. Rev. 1998 98 2167. 1994 33 130. J. Am. Chem. Soc. 1990 112 6880. 1992 114 4340. Chem. Soc. 1997 119 5867. 1116. 10 D. J. Taube H.-D. Projahn R. van Eldik D. Magde and T. G. Traylor 11 T. G. Traylor J. Luo J. A. Simon and P.C. Ford J. Am. Chem. Soc. 12 M. Buchalova P. R. Warburton R. van Eldik and D. H. Busch J. Am. 13 M. Buchalova D. H. Busch and R. van Eldik Inorg. Chem. 1998 37 14 H.-D. Projahn and R. van Eldik Inorg. Chem. 1992 30 3288. 15 H.-D. Projahn S. Schindler R. van Eldik D. G. Fortier C. R. Andrew and A. G. Sykes Inorg. Chem. 1995 34 5935. 16 A. L. Feig M. Becker S. Schindler R. van Eldik and S. J. Lippard Inorg. Chem. 1996 35 2590. 17 S. Seibig and R. van Eldik Inorg. Chem. 1997 36 4115. 18 S. Ryan H. Adams D. E. Fenton M. Becker and S. Schindler Inorg. Chem. 1998 37 2134. 19 M. Becker S. Schindler and R. van Eldik Inorg. Chem. 1994 33 5370. 20 S. Goldstein G. Czapski R. van Eldik H. Cohen and D. Meyerstein J. Phys. Chem. 1991 95 1282. 21 M. Becker F.W. Heinemann F. Knoch W. Donaubauer G. Liehr S. Schindler G. Golub H. Cohen and D. Meyerstein submitted for publication. 22 U. Spitzer R. van Eldik and H. Kelm Inorg. Chem. 1982 21 2821. 23 D. N. Silverman and S. Lindskog Acc. Chem. Res. 1988 21 30 and the literature survey in ref. 26. 24 X. Zhang and R. van Eldik Inorg. Chem. 1995 34 5606. 25 X. Zhang R. van Eldik T. Koike and E. Kimura Inorg. Chem. 1993 32 5749. 26 X. Zhang C. D. Hubbard and R. van Eldik J. Phys. Chem. 1996 100 9161. 27 R. van Eldik and D. A. Palmer J. Solution Chem. 1982 11 239. 28 E. Fujita and R. van Eldik Inorg. Chem. 1998 37 360; and literature cited therein. 29 C. Geyer E. Dinjus and S. Schindler Organometallics 1998 17 98. 30 C. Brandt and R. van Eldik Chem. Rev. 1995 95 119. 31 P. Warneck P. Mirabel G. A. Salmon R. van Eldik C. Vinkier K. J. Wannowius and C. Zetzsch in Heterogeneous and Liquid-Phase Processes P. Warneck (ed.) Springer Berlin 1996 ch. 2. 32 J. Berglund S. Fronaeus and L. I. Elding Inorg. Chem. 1993 32 4527. 33 C. Brandt I. Fabian and R. van Eldik Inorg. Chem. 1994 33 687. 34 V. Zang and R. van Eldik Inorg. Chem. 1990 29 4462. 35 V. Zang and R. van Eldik J. Chem. Soc. Dalton. Trans. 1993 111. 36 F. F. Prinsloo J. J. Pienaar R. van Eldik and H. Gutberlet J. Chem. 37 V. Lepentsiotis F. F. Prinsloo R. van Eldik and H. Gutberlet 38 F. F. Prinsloo J. J. Pienaar and R. van Eldik J. Chem. Soc. Dalton Soc. Dalton Trans. 1994 2373. J. Chem. Soc. Dalton Trans. 1996 2135. Trans. 1995 293. Received 3rd March 1998 Accepted 19th June 1998 393 Chemical Society Reviews 1998 volum

ISSN:0306-0012    

DOI:10.1039/a827387z

出版商:RSC    

年代:1998

数据来源: RSC

摘要:

Sodium borohydride in carboxylic acid media a phenomenal reduction system Gordon W. Gribble Department of Chemistry Dartmouth College Hanover New Hampshire 03755 USA The union of sodium borohydride and carboxylic acids has yielded an amazingly versatile and efficient set of reducing reagents. These acyloxyborohydride species reduce and N-alkylate indoles quinolines isoquinolines related heterocycles imines enamines oximes enamides and similar functional groups. They reduce amides and nitriles aryl alcohols and ketones aldehydes in the presence of ketones and b-hydroxyketones to 1,3-diols stereoselectively. This reagent is also extraordinarily useful for the N-alkylation of primary and secondary amines with aldehydes and ketones in a novel reductive amination process.1 Introduction Like an artist without paint the synthetic chemist is impotent without the necessary chemical reagents to synthesize molecules of interest. As synthetic targets increase in complexity so must the tools of the chemist increase in efficiency and selectivity. The present article summarizes the enormous range of chemical transformations available through the use of the relatively new reagent combination of sodium borohydride in carboxylic acids leading to the generation of sodium acyloxyborohydrides [eqns. (1) and (2)]. The less reactive sodium triacyloxyborohydrides 1 form with 3 equivalents of a carboxylic acid (or excess) and the more reactive sodium acyloxyborohydrides 2 form with one equivalent of a carboxylic acid.1,2 NaBH (1) 4 + 3RCO2H ––? NaBH(OCOR)3 + 3H2 (2) NaBH4 + RCO2H ––? NaBH3OCOR + H2 Gordon W.Gribble was born in San Francisco California and received his undergraduate training at the University of California Berkeley. He obtained his PhD at the University of Oregon with Lloyd J. Dolby in 1967 and did postdoctoral research at the University of California Los Angeles with Frank A. L. Anet. Professor Gribble has been at Dartmouth since 1968. In addition to his research program on heterocyclic chemistry and natural product synthesis he has a fascination with naturally occurring organohalogen compounds and he has published more than 160 papers in these areas. As an amateur winemaker he also has a strong interest in the chemistry of wine and winemaking.AcO OAc – B H O O Ph Following a report by Marshall and Johnson on the compatibility of NaBH4 and acetic acid (HOAc) in the reduction of enamines,3 we investigated this unlikely chemical combination as a possible new indole (3) reduction method. Surprisingly not only is the indole double bond rapidly and efficiently reduced presumably via indolenium ion 4 but the nitrogen is alkylated by the acetic acid solvent to afford N-ethylindoline (6) (Scheme 1).4 Further study of this novel transformation revealed that the N-alkylation can be circumvented using NaBH3CN–HOAc to afford indoline (5) in essentially quantitative yield.4,5 H H NaBH4 HOAc NH NH 86% 3 5 NaBH3CN 94% HOAc NaBH4 NaBH4 HOAc HOAc NH N + H 4 N CH2CH3 N CH2CH3 88% 86% 5 7 the reviews.1,2 NaBH4 53% HCO2H N CH3 NaBH4 69% N NH CH2CH2CH3 3 49% N CH2CH(CH3)2 6 Scheme 1 These fascinating results sparked our further studies with the NaBH4–RCO2H reagent system an odyssey that has continued for 25 years and has led to an extraordinarily versatile unique and efficient set of reducing agents.1,2 Throughout this presentation uncited reactions can be found in 2 Chemistry of acyloxyborohydrides 2.1 Reduction of indoles The reduction and N-alkylation of indoles with NaBH4–RCO2H to give N-alkylindoles (Scheme 2)4 and the reduction of indoles with NaBH3CN–HOAc to give indolines are quite general CH3CH2CO2H NaBH4 (CH3)2CHCO2H Scheme 2 Chemical Society Reviews 1998 volume 27 395 processes.The latter reaction is the preeminent method for reducing the indole double bond provided that electronwithdrawing groups are not present to retard the initial indole protonation (i.e. 3?4). Thus an indole double bond containing an ester group at C-2 or C-3 (8–10) is inert to the action of NaBH3CN–HOAc. Likewise 5-nitroindole is not reduced to 11 under these conditions. Such selectivity has been of great utility in the synthesis of the antitumor agent CC-1065 and analogues. However at higher temperatures one may encounter N-ethylation with NaBH3CN–HOAc. CO2Me MeO MeO NH 98% H O O HN NH H 98% MeO Me N Me MeO N Et N N Me BnO2C NH NH 75% A remarkable illustration of the selectivity of this indole reduction method is seen in 12 ? 13 [eqn.(3)],6 in which the protonated basic nitrogen presumably prevents a second protonation of the proximal indole double bond. Me N NH NH 12 Me N NH 13 NaBH4 N CF3CO2H NH 0 °C 90% 14 O HN N NH Ph Cl 16 Chemical Society Reviews 1998 volume 27 396 Generally compounds containing a basic nitrogen atom such as the ubiquitous indolo[2,3-a]quinolizidine alkaloids (e.g. 14) can only be reduced using NaBH4 in trifluoroacetic acid (TFA) [eqn. (4)].4,7 Thus only the imine and not the indole bond in 16 is reduced with NaBH3CN–HOAc [eqn. (5)]. Under the influence of NaBH4–TFA indole (3) is converted to a mixture of indoline (5) N-trifluoroethylindoline (17) and the Baeyer condensation product 18 [eqn.(6)].8 NaBH4 + EtO2C CF3CO2H CO2Me NH NH HN 5 (40%) NH CF3 8 (0%) NH 9 (96%) O2N TMS N CH2CF3 CO2Et NH 18 (13%) 11 (0%) NH N OBn NH OMe O OBn OMe 10 (61%) Et NHEt 88% NaBH3CN NaBH4 HOAc 20 °C NH2 HOAc CF3CO2H (3:1) 71% 4 69% NaBH4 2. NaBH4 80% Me3CCO2H HOAc 40–45 °C (3) rt ® D EtCO2H 50 °C 70% 2. NaBH4 HOAc 68% 1. acetone 2. NaBH HOAc 50-55 °C 79% NH N i–Pr NHCH2C(CH3)3 Scheme 3 H H (4) N H NH 15 NaBH3CN HOAc 10 °C 15 min O HN (5) HN NH Ph Cl 3 2.2 N-Alkylation of amines Our observation that NaBH4–HOAc gives N-ethylation of indoline (5) (Scheme 1) led us to explore the scope of this unprecedented amine N-alkylation reaction.Reactions of aniline with NaBH4–RCO2H are shown in Scheme 3.4 One can N n–Pr NaBH4 HN i–Pr 1. acetone NaBH4 HOAc 50–55 °C Et effect mono- or dialkylation depending on the temperature and thus achieve the conversion of a primary amine to an unsymmetrical tertiary amine in one pot. The reductive amination of added aldehydes or ketones increases the versatility of the method. Pivalic acid affords N-neopentylaniline in 80% yield. Similar chemistry is observed with benzylamine and other aliphatic amines.9 Control experiments revealed that the mechanism of this Nalkylation does not involve reduction of a precursor amide,4 and gas evolution measurements and isolation studies10 indicated that the borohydride species formed under these conditions of excess acetic acid is NaBH(OAc)3 (1 R = Me) [eqn.(1)]. Furthermore we were able to isolate the 2,4-DNP derivative of acetaldehyde from the evolved gases of the reaction of NaBH4 with glacial HOAc. Therefore we believe that the NaBH(OCOR) 3 species undergoes self-reduction to free aldehyde (or a synthetic equivalent) which then reacts with the amine in a typical reductive amination sequence (Scheme 4). + N CH2CF3 (6) 17 (5%) N CH2CF3 NEt2 NaBH4 HOAc 50–60 °C 74% NHCH2Ph 4 1. PhCHO 2. NaBH HOAc 20 °C 1. PhCHO 80% 50–60 °C Et N CH2Ph HO2CR :OH O O R—C—O R—C—O R—C—H amines reduction RCH2OH trapped as 2,4-DNP (R = CH3 CF3) NH2 NO2 HN NO2 HN 2 O NH O Ph HN CO2Me H B(OCOR)3 – N-alkylation ester formation Scheme 4 NaBH(OAc)3 HOAc ClCH2CH2Cl 23 h 66% NaBH(OAc)3 HOAc ClCH2CH2Cl 87% NaBH(OAc)3 CH2Cl2 80% Scheme 5 H—B(OCOR)2 – Using the isolated reducing reagent NaBH(OAc)3 Abdel- Magid has extended this amine N-alkylation method into a powerful general reductive amination protocol for aldehydes and ketones.11 Some recent examples are shown in Scheme 5.11–13 O + NH NH O + O O Ph + CHO H2NCH2CO2Me Obviously these amine N-alkylations (reductive aminations) succeed because NaBH(OAc)3 only slowly reduces aldehydes and ketones relative to the rapid reduction of iminium and immonium ions.2.3 Reduction of other heterocycles The facile reduction of indole (3) (Scheme 1) with NaBH4– RCO2H portended that other nitrogen-containing heterocycles that are susceptible to protonation would undergo a similar reduction/alkylation sequence. Indeed quinolines isoquinolines acridines quinazolines quinoxalines phthalazines pteridines benzoxazines adenines some pyrroles and pyrylium salts are reduced by NaBH4–RCO2H.1,2 More aromatic pyridines are normally unaffected by NaBH4–RCO2H. Illustrative of the versatility of this methodology is the chemistry shown in Scheme 6 involving quinoline.14 NH 71% NaBH3CN N Some additional examples are shown below [eqns.(7)–(9)]. N N N Ph Me Me O+ N SO2Ph 2.4 Reduction of imines enamines and related compounds A wide range of imines enamines enamides vinylogous amides and carbamates can be reduced using NaBH4–RCO2H (Scheme 7). 2.5 Reduction of oximes Oximes are reduced and reductively alkylated with NaBH3CN– HOAc and NaBH4–RCO2H respectively (Scheme 8).15 Whereas at room temperature cyclohexanone oxime is reduced to cyclohexylhydroxylamine with NaBH4–HOAc at higher temperatures the reaction proceeds to afford N,N-diethylcyclohexylamine. Similarly oxime ethers are reduced to O-alkylated hydroxylamines under these conditions. Such an example involving a hydroxy-directed reduction is illustrated in eqn. (10). OH Chemical Society Reviews 1998 volume 27 50 °C 1 h NaBH3CN NaBH4 HOAc N NaBH HOAc HCO2H acetone 52% 59% Scheme 6 HN NaBH4 N CF3CO2H NH THF rt 75% Ph NaBH4 HOAc O Me 40 °C 99% NaBH3CN CF3CO2H N 75% SO2Ph Me4NBH(OAc)3 N CH3CN HOAc OBn –35 °C 95% OH NH OBn anti/syn 95 5 N CH2CH3 HOAc 68% 4 N CH3 (7) (8) Me (9) (10) 397 F F N O Me CN N Br H N MeO2C Cl OMe N OMe O – NHBn S + NHBn O EtO O N Ph N (CH2)10CH3 Ts N NH MeO2C 2.6 Reduction of amides to amines Although NaBH(OCOR)3 (1) does not reduce amides Umino discovered that NaBH3OCOR does reduce amides and lactams to the corresponding amines.16 Some examples of this useful reaction are listed in Scheme 9.Note that carbamates and sultams are unaffected by these conditions. Interestingly the indole double bond in the last example is not reduced.17 2.7 Reduction of nitriles to primary amines Umino also discovered that NaBH3OCOR particularly NaBH3OCOCF3 reduces nitriles to primary amines.18 A selection of Chemical Society Reviews 1998 volume 27 398 BHNa CO2 N 3 O i-Bu O CH2Cl2 -40 °C ® -5 °C 92% NaBH3CN CF3CO2H 45% NaBH3CN Bn HOAc MeCN rt 56% NaBH4 HOAc THF 20 °C 62% NaBH(OAc)3 HOAc 10 °C 89% NaBH3CN HCO2H 90% NaBH3CN CH2Cl2 TFA -45 °C 76% NaBH4 O HOAc 10 °C 90% Scheme 7 F NH F O 95% ee N Br H Cl NH CO2Me (+ aziridine) OMe N Bn OMe (+ 3% trans) O – NHBn + S NHBn O EtO O N Ph N (CH2)10CH3 Ts (22% cis) N O NH H H MeO2C (+ trans 1 1) NaBH4 NaBH4 HOAc HOAc OH 20 °C 3.5 h N NHOH 63% NaBH4 CH3CH2CO2H NaBH3CN CH3CN HOAc 30% 20 °C 3 h 81% OMe MeO MeO NH H O NaBH3OCOCF3 HN BOC NH O HN BOC BOC NH O HN NH O O O N Ph O O O HN OH O HO O HN O O2S O2S Ph O N N Scheme 8 NaBH4 HOAc dioxane 97% HN BOC HN NaBH4 TFA THF 73% OH Me4NBH(OAc)3 HOAc MeCN 25 °C 63% NaBH4 TFA dioxane 100 °C 66% NaBH3OCOCF3 dioxane 82% NaBH4 CF3CO2H dioxane D 90% Scheme 9 examples is shown in Scheme 10.Noteworthy is that this reduction reaction occurs in the presence of nitro and 1,2-oxazine functionalities. Et N Et CH3CN 46% 45 °C Pr Pr N 50 °C OMe NH H THF 20 °C 65% HN NH OH OH O N Ph O O > 98% de HN OH OH HN Ph H N N Me Cl N CN N NO CN CH3 BnO NO2 O O O CN Ar O O Ar CN O N CO2Bn F F CN F F F 2.8 Reduction of alcohols to hydrocarbons Early in our studies we thought that NaBH4–CF3CO2H (TFA) might serve to reduce certain alcohols to hydrocarbons in view of the propensity of TFA to stabilize carbocations. Indeed this reagent combination provides an efficient and general method for reducing di- and triarylcarbinols to di- and triarylmethanes (Scheme 11).19 Other alcohols unless the derived carbocation is highly stabilized are not reduced cleanly under these conditions.19 OH COH 3 In the case of carbinol 19 the intermediate carbocation 20 is ambushed by the o-phenyl group prior to reduction affording only 9-phenylfluorene (21) [eqn. (11)].19 The generality and selectivity of this reduction method is illustrated by the examples in eqns. (12)–(16). The chemoselectivity exhibited by NaBH(OCOCF3)3 vis-`a-vis NaBH 3OCOCF3 in eqns. (14)–(15) is remarkable. 2.9 Reduction of ketones to hydrocarbons Diarylketones are smoothly reduced to diarylmethanes with NaBH4–TFA.20 As we have seen a wide range of functional Me N Cl N NaBH3OCOCF3 NH2 OH NO2 2 THF 10-15 °C 57% NH2 CH3 NaBH3OCOCF3 BnO NO2 THF rt 70% O O O CH2NH2 Ar NaBH3OCOCF3 O O THF rt 70% Ar = p-tolyl Ar CH2NH2 NaBH3OCOCF3 O N CO2Bn THF rt 48% F NH F 2 NaBH3OCOCF3 F F THF rt 67% F Scheme 10 Ph OH NaBH4 CF3CO2H N 15-20 °C 93% MeS NaBH4 CF3CO2H 15-20 °C 94% Scheme 11 19 O S NH N HO Chemical Society Reviews 1998 volume 27 21 NaBH4 CF3CO2H 90% NaBH3OCOCF3 + NaBH4 CF3CO2H 15–20 °C 93% (11) OH I NaBH4 O CF3CO2H CH2Cl2 rt I 86% NEt2 (12) O I NEt2 NaBH4 O OH CF3CO2H Et2O rt 68% (13) O S Ph (14) N major (87:13) MeS OH O (15) N N THF 4 h >75% NH N NaBH4 (16) 20 I O OH N HN N HN TFA CH2Cl2 20 °C 100% 399 groups will tolerate these reaction conditions (Scheme 12).The mechanism presumably involves reduction to the diarylmethanol solvolysis to the carbocation and reduction to the hydrocarbon. Only in the case of strong electron-withdrawing groups (e.g. p-NO2) is the reaction incomplete. The last reaction appears to be the first reduction of a formyl group to a methyl group using this methodology. This method provides for a very useful synthesis of 3-alkylindoles [eqn. (17)] and alkyl-substituted ferrocenes [eqn. (18)].21 Ac2O N SO2Ph AlCl3 CH2Cl2 25 °C 98% NaBH4 CF3CO2H 15 °C ® 25 °C 99% CO2H NaBH4 O Fe CF3CO2H CH2Cl2 90% rt The combination of NaBH4–TFA converts enones to alkenes [eqns.(19) and (20)],22,23 and isopropylidene acylmalonates 5-acylbarbituric acids and 3-acyl-4-hydroxycoumarins are all reduced to the corresponding methylene derivatives with NaBH3CN–HOAc [e.g. eqn. (21)]. O NaBH4 CF3CO2H O CH2Cl2 93% O NaBH4 CF3CO2H CH2Cl2 CH3CN 69% O O O NaBH3CN Bn O HOAc rt O 85% 2.10 Reductive cleavage of acetals ketals ethers and related compounds The action of NaBH4–TFA serves to reductively cleave a variety of acetals ketals ethers and ozonides as summarized in Scheme 13.24–26 The deoxygenation of 1,4-epoxy-1,4-dihy- Chemical Society Reviews 1998 volume 27 400 O CH3 N SO2Ph CH3 (17) N SO2Ph CO2H Fe (18) OH (19) (20) O O (21) Bn O O R S HO OHC O droarenes is particularly useful for the synthesis of polycyclic aromatic hydrocarbons.2.11 Alkylation of arenes (Baeyer condensation) As noted earlier indole (3) with NaBH4–TFA gives some of the Baeyer condensation product 18 [eqn. (6)]. Indeed this alkylation reaction of arenes involving the generation of trifluoroacetaldehyde is reminiscent of the synthesis of DDT from chlorobenzene trichloroacetaldehyde (chloral) and sulfuric acid. We have found that several arenes give analogous products under these conditions (Scheme 14).27 Congested arenes like durene and mesitylene stop at the carbinol stage.2.12 Selective reduction of aldehydes The reduction of aldehydes and ketones to alcohols is one of the most important reactions in organic chemistry. Although many reagents are available for this reaction few are chemoselective for aldehydes and such methods are in great demand. From the beginning of our work in this area it was clear that aldehydes and especially ketones were reduced relatively slowly by these acyloxyborohydrides. Indeed this is precisely why the N-alkylation of amines works! Thus although benzaldehyde is completely reduced to benzyl alcohol after 1 hour at 15 °C with a large excess of NaBH4 in glacial acetic acid acetophenone is only reduced to the extent of 60% at 25 °C after 40 hours! By comparison in alcoholic solution both reductions are complete in seconds.These and related observations paved the way for the chemoselective reduction of aldehydes in the presence of ketones.10,28 The isolated reagents NaBH(OAc)3 10 and n-Bu4NBH(OAc)3 28 work extremely well and some examples with the latter reagent are depicted in Scheme 15.28 Some additional examples with NaBH(OAc)3 are listed in eqns. (22)–(25). Notable is the selective reduction of the aldehyde in the presence of a trifluoromethyl ketone [eqn. (24)] O NaBH4 CF3CO2H R 15-20 °C R % yield R % yield 2 90 82 93 73 93 CN NMe NHPh CO2H CO2Me NO H Me OMe OH F Br 43 (+57% alcohol) 92 89 88 90 82 94 2 S S NaBH4 CF3CO2H S THF O 89% Cl Cl O Cl Cl NaBH4 CF3CO2H HO O O CH2Cl2 CO2Et CO2Et 66% Me CO2Me CO2Me NaBH3CN CO2Me CO N 2Me N CF3CO2H N N O O O 60% Scheme 12 O O NaBH4 TFA THF 20 °C 83% F CH3 NaBH4 O F CH3 TFA THF 15 °C 92% Br CH3 F O Mg N CH3 93% PhO2S 1.NaBH4–TFA 2. NaOH 77% MeO NaBH4 O MeO TFA 0-5 °C 38% Cl O NaBH3CN O Ar OBn O CF3CO2H DMF NHAc BnO 84% Ar = p-Methoxyphenyl n-C5H11 NaBH4 O O O CF3CO2H n-C5H11 CH2Cl2 33% Scheme 13 and the selective reduction of the less sterically encumbered aldehyde in a dialdehyde [eqn. (25)].29 2.13 Hydroxy-directed reduction of ketones During our study of the chemoselective reduction of aldehydes in the presence of ketones we observed the reduction of ketoaldehyde 22 to diol 23 presumably involving internal hydride delivery as shown in eqn.(26).28 Saksena independently discovered this same reaction and observed excellent stereoselectivities in the reduction of steroidal b-hydroxy ketones. Evans thoroughly explored the scope of this powerful methodology and he fully characterized several BH(OAc)3 species for the first time.30 In the intervening OH O CH F 3 F CH3 CH3 O N PhO CH3 2S CH3 NH CH3 MeO OH MeO Cl OHC O O Ph O O Ar OBn HO NHAc BnO n-C5H11 O n-C5H11 O ten years a very large number of hydroxy-directed carbonyl reductions using triacetoxyborohydride have been described.2a,30 A few examples are shown in Scheme 16. Several a-hydroxyketones have also been reported to undergo this intramolecular reduction. Arene + NaBH4–CF3CO2H 22 O NaBH(OAc)3 PhH rt 83% O O O O O O O NaBH4 CF3 N HOAc PhH O CH3 95% CHO CHO H B – H CHO Ph O O CF3 53% CF3 Me 65% D NaBH(OAc)3 PhH 67% AcO O Scheme 14 Chemical Society Reviews 1998 volume 27 CHO OH 23 78% CF Me 3 OH (22) O CHO (23) HO (24) O D CH3 (25) O HOAc PhH OH N H O OAc O (26) Ph NaBH(OAc)3 OH OMe CF3 76% CF3 OH OMe Me Me 48% 401 H O + O CH O 3 CHO + O CHO OH O OH O O OTIPS MeO O OH O O O N Bn O O OH O OBOM OH O OH N O EtO2C O BnO O OMe O OMe 2.14 Hydroboration of alkenes One of the first applications of the NaBH4–HOAc reagent system was the hydroboration of alkenes as described by Marshall and Johnson.3 The actual reagent is the more reactive NaBH3OAc which can also be generated from NaBH4 and Chemical Society Reviews 1998 volume 27 CH2OH + 95% O OH OH 98:2 anti/syn O OH OH 13:1 anti/syn O OH OH N Bn > 99:1 OH OH OBOM OH N EtO2C BnO Me4NBH(OAc)3 HOAc MeCN -20 °C 84% 402 excess n-Bu4NBH(OAc)3 recovered ketone PhH 94% 96% O CH3 excess n-Bu4NBH(OAc)3 PhH 96% excess n-Bu4NBH(OAc)3 OH PhH 77% Scheme 15 Me4NBH(OAc)3 HOAc MeCN -20 °C 92% Me4NBH(OAc)3 OTIPS MeO HOAc MeCN -40 °C 93% O Me4NBH(OAc)3 O HOAc MeCN rt 81% Me4NBH(OAc)3 O 75% O OH NaBH(OAc)3 OH + OH O O HOAc CH2Cl2 rt 82% OBn OH O OBn OH OH 97% ds Scheme 16 Hg(OAc)2.Some examples of alkene hydroboration are listed in Scheme 17. OH 1. NaBH4 HOAc 2. H2O2 OH– 89% OTMS OTMS 1. NaBH3OAc OH 2. H2O2 OH– 83% O 1. NaBH3OAc 2. NaOMe CHCl3 3. H2O2 OH– 67% O 1. NaBH3OAc 2. PCC 85% Scheme 17 2.15 Miscellaneous reductions Alkenes that yield stable carbocations upon protonation can be reduced to alkanes with NaBH4–TFA [eqn.(27)],19 but such examples are exceedingly rare. It seems likely that NaBH4– CF3SO3H may work in this regard with other alkenes. NaBH4 (27) CF3CO2H 0 °C 93% Enones and enals give primarily 1,2-reduction with NaBH3OAc [eqns. (28) and (29)]. Small amounts of the 1,4-reduction products were also found. NaBH3OAc Ph Ph Ph (28) + THF 25 °C O OH O 70% 4% 96% CHO CH2OH CHO NaBH3OAc (29) THF 25 °C 86% 1% 99% Twenty years ago we observed that both 1- and 2-naphthol yielded 10–20% naphthalene upon treatment with NaBH4–TFA (Scheme 18). In fact when the crude reaction products were OH NaBH4 CF3CO2H rt Scheme 18 allowed to stand undisturbed for six months pure naphthalene had sublimed onto the upper part of the flask.Presumably this transformation involves ring protonation carbonyl reduction and dehydration. More recently this novel deoxygenation of phenols has been reported for a hydroxyphenanthrene [eqn. (30)].31 OH NaBH4 CF3CO2H rt 20% These results are consistent with our earlier observations that 9,10-phenanthrenequinone and 2-methyl-1,4-naphthoquinone undergo reduction to the corresponding aromatic hydrocarbons albeit in low yield (Scheme 19).1 O NaBH4 CF3CO2H O NaBH O 4 CF3CO2H O O NaBH4 CF3CO2H O Scheme 19 Esters are not normally reduced to primary alcohols by NaBH4–CO2H but such a reduction is observed at higher temperatures in the case of amino acid and peptide esters [eqns.(31) and (32)].32 Importantly racemization is not seen. In view of the facile reduction of carboxylic acids to aldehydes with NaBH4 in the course of the reductive amination sequence (vide supra) it is not surprising that complete reduction to primary alcohols has been found (Scheme 20). Turnbull has recently described the reduction and subsequent N-trifluoroethylation of aroyl azides with NaBH4–TFA [eqn. (33)].33 The parent compound yielded the Baeyer condensation product 24 in 87% yield. Organomercurials can be reduced to alkanes with NaBH( OAc)3 [eqn. (34)],34 and the NaBH(OCOR)3 reagents can also acylate alcohols phenols and thiophenols [eqns. (35)–(38)] presumably by direct acylation of an acyloxyborohydride intermediate. OH + OH –H2O 10% CON3 Cl (30) O Me OMe 72% O CF3CO2H NaBH4 Me 23% 12% 57% O2N CO2Me NaBH4 BOC HN HOAc dioxane 80 °C 90% CO2Me NaBH4 FMOC HN HOAc dioxane 80 °C 98% NaBH4 CF3CO2H rt 91% CF3 (F3CH2C)2N HgOAc NaBH(OAc) CO2Me THF 0 °C 82% H + OMe 80% OH NaBH4 HOAc 95% SH D NaBH4 HOAc 95% CH2OH 4 24 3 CO2Me O2N D NaBH EtCO2H 90% NH2 EtCO2H D 95% D NaBH4 Chemical Society Reviews 1998 volume 27 OH (31) HN BOC OH (32) HN FMOC N(CH2CF3)2 (33) Cl N(CH2CF3)2 H (34) CO2Me OMe 20% OAc (35) SAc (36) CH2OCOEt (37) NHCOEt (38) 403 CO2H O MeO (CH2)8CO2H Amine alkylation b-Hydroxyketone reduction Oxime reduction alkylation Acetal ketal reductive cleavage 3 Concluding remarks As summarized in Scheme 21 the combination of NaBH4– RCO2H leading to acyloxyborohydrides is a remarkably versatile and unique chemical system.These chemical species have emerged as the preeminent reagents of choice for a wide spectrum of chemical transformations. The ability to control chemoselectivity regioselectivity and stereoselectivity by adjusting the carboxylic acid borohydride reagent stoichiometry and temperature has no parallel in the repertoire of the organic chemist. Nevertheless much work remains to be done with acyloxyborohydrides particularly with regard to understanding the mechanisms of some of the reactions in applying these reagents to asymmetric synthesis and in uncovering new applications.4 Acknowledgements I am especially grateful to my many co-workers on this project over the past 25 years. In particular graduate students Jerry Skotnicki at the beginning and Charlie Nutaitis in later years. But my deepest appreciation goes out to the many Dartmouth undergraduates who explored and developed much of the chemistry presented here. These include Steve Dietz Duncan Ferguson Peter Heald Bud Leese Myles Sheehan Joe Chemical Society Reviews 1998 volume 27 404 NaBH4 OH Hoffman Sandy Emery Joe Jasinski John Pellicone and Steve Wright. Thank you all. THF 90% O NaBH4 MeO (CH2)9OH TFA THF 25 °C 78% Scheme 20 Indole reduction Aryl carbinol reduction Selective aldehyde reduction NaBH4 RCO2H Quinoline Isoquinoline reduction Arene alkylation Amide and nitrile reduction Scheme 21 5 References 1 For a review and historical background on acyloxyborohydrides see G.W. Gribble and C. F. Nutaitis Org. Prep. Proc. Int. 1985 17 317. 2 For recent reviews of the synthetic utility of acyloxyborohydrides see (a) G. W. Gribble in Reductions in Organic Synthesis ed. Ahmed F. Abdel-Magid ACS Symposium Series 641 American Chemical Society Washington DC 1996 pp. 167–200; (b) G. W. Gribble Chemtech 1996 12 26. 3 J. A. Marshall and W. S. Johnson J. Org. Chem. 1963 28 421. 4 G. W. Gribble P. D. Lord J. Skotnicki S. E. Dietz J. T. Eaton and J.L. Johnson J. Am. Chem. Soc. 1974 96 7812. 5 G. W. Gribble and J. H. Hoffman Synthesis 1977 859. 6 M. Somei F. Yamada and H. Morikawa Heterocycles 1997 46 91. 7 G. W. Gribble J. L. Johnson and M. G. Saulnier Heterocycles 1981 8 G. W. Gribble C. F. Nutaitis and R. M. Leese Heterocycles 1984 22 9 G. W. Gribble J. M. Jasinski J. T. Pellicone and J. A. Panetta 16 2109. 379. Synthesis 1978 766. 1975 535. R. D. Shah J. Org. Chem. 1996 61 3849. 1998 63 968. 10 G. W. Gribble and D. C. Ferguson J. Chem. Soc. Chem. Commun. 11 A. F. Abdel-Magid K. G. Carson B. D. Harris C. A. Maryanoff and 12 J. Quirante C. Escolano A. Merino and J. Bonjoch J. Org. Chem. 13 J. Cossy and D. Belotti J. Org. Chem. 1997 62 7900. 14 G. W. Gribble and P. W. Heald Synthesis 1975 650. 15 G. W. Gribble R. W. Leiby and M. N. Sheehan Synthesis 1977 856. 16 N. Umino T. Iwakuma and M. Itoh Tetrahedron Lett. 1976 763. 17 A. Padwa S. R. Harring and M. A. Semones J. Org. Chem. 1998 63 44. 18 N. Umino T. Iwakuma and N. Itoh Tetrahedron Lett. 1976 2875. 19 G. W. Gribble R. M. Leese and B. E. Evans Synthesis 1977 172. 20 G. W. Gribble W. J. Kelly and S. E. Emery Synthesis 1978 763. 21 S. Bhattacharyya J. Chem. Soc. Perkin Trans. 1 1996 1381. 22 M. Beckmann T. Meyer F. Schulz and E. Winterfeldt Chem. Ber. 1994 127 2505. 23 V. H. Rawal A. Fabr�e and S. Iwasa Tetrahedron Lett. 1995 36 6851. 24 C. F. Nutaitis and G. W. Gribble Org. Prep. Proc. Int. 1985 17 11. 25 G. W. Gribble W. J. Kelly and M. P. Sibi Synthesis 1982 143. 26 G. W. Gribble M. G. Saulnier M. P. Sibi and J. A. Obaza-Nutaitis J. Org. Chem. 1984 49 4518. 27 C. F. Nutaitis and G. W. Gribble Synthesis 1985 756. 28 C. F. Nutaitis and G. W. Gribble Tetrahedron Lett. 1983 24 4287. 29 G. Stork F. West H. Y. Lee R. C. A. Isaacs and S. Manabe J. Am. Chem. Soc. 1996 118 10 660. 30 D. A. Evans K. T. Chapman and E. M. Carreira J. Am. Chem. Soc. 1988 110 3560. 31 A. K. Banerjee J. C. Acevedo R. Gonz�alez and A. Rojas Tetrahedron 1991 47 2081. 32 M. Soucek J. Urban and D. Saman Collect. Czech. Chem. Commun. 1990 55 761. 33 D. M. Krein P. J. Sullivan and K. Turnbull Tetrahedron Lett. 1996 37 7213. 34 F. H. Gouzoules and R. A. Whitney J. Org. Chem. 1986 51 2024. Received 16th June 1998 Accepted 26th June

ISSN:0306-0012    

DOI:10.1039/a827395z

出版商:RSC    

年代:1998

数据来源: RSC

摘要:

Chemistry with a sense of direction—the stereodynamics of bimolecular reactions Andrew J. Alexander Mark Brouard Konstantinos S. Kalogerakis and John P. Simons* The Physical and Theoretical Chemistry Laboratory South Parks Road The University of Oxford Oxford UK OX1 3QZ. This review outlines some of the exciting new developments in the experimental study of the dynamics of elementary bimolecular reactions. Emphasis is placed on the ‘new wave’ of stereodynamical studies of photon initiated bimolecular reactions using Doppler-resolved polarised laser pump and probe techniques. A few key studies which are discussed in some detail provide a taste of what has already been achieved as well as a hint of the new experiments that can be anticipated in the near future.1 Introduction The development of scientific innovation is often triggered by the advent of new experimental strategies which allow questions to be addressed by design rather than conjecture. At first the new thinking is confined to a small group of laboratories but if the innovation is addressing important questions the ‘word Dr Alexander gained his first degree in Chemical Physics from the University of Edinburgh in 1994 and his D.Phil. from the University of Oxford working in Professor Simons’ group in 1997. He currently holds a post-doctoral research fellowship with Professor R. N. Zare in the Department of Chemistry University of Stanford. Dr Brouard obtained his D.Phil. from the University of Oxford in 1986 where he worked in the group of Professor M.J. Pilling in the field of gas phase reaction kinetics. He subsequently moved to Nottingham University initially as a post-doctoral researcher in Professor Simons’ group and then from 1989 as a University Lecturer. In 1993 he returned to the University of Oxford where he is now a lecturer in the Department of Chemistry and a Tutorial Fellow at Jesus College. His primary research interests in recent years have been in the fields of photodissociation and bimolecular reaction dynamics. Dr Kalogerakis obtained his first degree in Chemistry from the John P. Simons Andrew J. Alexander soon gets around’ and before too long scientists in other laboratories or disciplines begin to get wind of the fact that something new and exciting is happening.This is the path that (may) lead one day to Nobel prizes. When that day has come and gone although a new sub-discipline has been created the buzz may begin to subside and the new sub-discipline can become an esoteric speciality with its own high priests language priorities biennial conferences—and even Review articles. A prime purpose of the present Review is to transmit some of the current buzz and passion which animates the (far from esoteric) world of molecular reaction dynamics; to show how some of the questions signalled by its Nobel Laureates can now be addressed using powerful new laser-based experimental strategies; and to report some of the extraordinarily detailed insights into the nature of chemically reactive collisions that are being revealed.The central challenge is easy to state far less easy to achieve it is to describe and understand the stereo- University of Athens Greece in 1989 and his PhD from the University of Stanford USA working in Professor R. N. Zare’s group in 1995. He currently holds an EPSRC post-doctoral research fellowship with Professor Simons and Dr Brouard in the Department of Chemistry University of Oxford a College lectureship at Keble College Oxford and is a Junior Research Fellow at Worcester College Oxford. Professor Simons is the Dr Lee’s Professor of Chemistry at the University of Oxford and a Professorial Fellow of Exeter College. He is a Fellow of the Royal Society a past-President of the Faraday Division of the RSC and a past-Chairman of the Molecular Beams and Dynamics and the Gas Kinetics Groups.He received the Tilden Lectureship in 1983 the Chemical Dynamics Award (sponsored by BP) in 1993 the Polanyi Medal and Lectureship in 1996 and presented the George C. Pimentel Memorial Lecture at Berkeley in 1998. He has been engaged in the study of molecular reaction dynamics and in particular the stereodynamics of photon initiated reactions for some thirty years. Konstantinos S. Kalogerakis Mark Brouard 405 Chemical Society Reviews 1998 volume 27 dynamics of a reactive molecular collision to present a threedimensional view of the passage through the transition state from reagents to products. The most direct way of probing the stereodynamics of the collision is through monitoring the angular distribution or better the angle-resolved velocity distribution of the scattered products best represented on a polar product scattering map.The most direct way of obtaining such a map is through monitoring the products scattered from two colliding reagent molecular beams using a (universal) mass spectrometer detector. Fig. 1 shows the scattering map for the abstraction reaction determined in this way by Y. T. Lee’s group;1 |n A jAÅ represent the vibrational and rotational levels of the nascent HF. It has become a famous picture since almost uniquely the kinematics of the reaction allow the angular distributions of each populated vibrational level n A in the scattered HF molecules to be separately resolved through their velocity spectrum.In the vast majority of cross-beam studies the angular distribution represents an average over all the populated product quantum states. For the F + H2 reaction there is a switch from ‘backward’ to ‘forward’ scattering when the HF is excited into its highest energetically accessible vibrational level n A = 3. This subtle change which would otherwise have been hidden has taken a decade to understand and in concert with ab initio theory and parallel studies of the photoelectron spectroscopy of Fig. 1 (A) A contour plot (upper panel) and a three-dimensional view (lower panel) of the scattering map for the F + H2 ? HF(nA) + H reaction at a collision energy of 11.5 kJ mol21 adapted from reference.1 The scattering into different product vibrational levels n A is indicated.0° corresponds to forward scattering with respect to the incoming F atom. (B) Schematic illustrations of the scattering behaviour shown quantitatively in part (A). Note in particular the contrasting scattering for HF molecules born in the n A = 1,2 and n A = 3 vibrational levels. Chemical Society Reviews 1998 volume 27 406 the FH22 anion it has stimulated some of the most profound insights into the dynamics of a benchmark chemical reaction. 2 (1) F + H2 ? HF|n A j AÅ + H Optical detection provides an alternative to mass spectrometric detection; techniques such as tunable laser induced fluorescence (LIF) or resonantly enhanced multiphoton ionisation (REMPI) spectroscopy are both sensitive and necessarily product quantum state selective.If there is no population in the level being excited then of course there is no signal! In principle localised angle- and time-resolved optical detection around the cross-beam scattering zone could provide a means of probing the distribution of the scattered products but the dilution in their concentrations as they move away from the collision zone places severe demands on the detection sensitivity. Fortunately there is an alternative approach which circumvents this difficulty—Doppler-resolved † optical detection of the scattered reaction products at the collision zone. Products moving towards or away from the detection laser experience a blue or red shift in their absorption spectra and their angle-resolved velocity distributions are ‘encoded’ in the Doppler contours of their spectral lines.This strategy originally pioneered by Kinsey in the late 1970s,3 provided the first quantum state-resolved product angular distributions for a bimolecular reaction. An elegant and visually appealing means to the same end couples selective laser ionisation at the scattering zone with angular imaging of the ionised product trajectories (which remain virtually undeflected by the loss of an electron) using a CCD camera. Transformation of the two-dimensional projections into three-dimensional angular distributions generates a family of product-state resolved angular scattering maps.4 These new experiments begin the process of resolving the swarm of successful (i.e.reactive) molecular collisions into their constituent sub-sets and exploring the dynamics of their individual classical trajectories; in quantum mechanical terms the experiments begin the process of identifying individual elements in the reactive scattering matrix. The process can be continued further by taking advantage of the polarisation of the probe laser beam to explore the spatial distribution of the rotational angular momentum jA of the scattered reaction products. In the classical limit of high jA Q branch molecular transitions with ½jA = 0 have their transition dipoles m aligned parallel to jA; P or R branch features with ½jA = ±1 have mdirected perpendicular to jA.5 In consequence the polarisation dependence of the Doppler-resolved laser induced excitation spectra reflects stereodynamic preferences in both the linear and the angular momentum spatial distributions e.g.preferences for forward or backward scattering or for ‘cartwheel’ ‘discus’ or ‘propeller-like’ rotational motion. The application of polarised Doppler-resolved laser probe strategies has more than any other provided an especially high-powered lens through which the stereodynamics of individual molecular collisions may be viewed. . . . How was it created? 2 The vectorial approach In a prescient paper6 on the ‘statistical theory of angular momentum polarisation in chemical reactions’ published more than 20 years ago Case and Herschbach wrote ‘directional or vector properties of chemical reaction dynamics contain much information not provided by energetic or scalar properties’ † We include under the heading of Doppler-resolved optical detection REMPI techniques coupled with time-of-flight mass spectrometric detection.The important feature is that the optical technique must be sensitive to the (state-resolved) velocity distribution of the reaction product. The paper was stimulated by a few pioneering experiments probing polarisation of rotational angular momentum from bimolecular scattering events. A few years later Case McClelland and Herschbach noted7 ‘the wedding of lasers and molecular beams. now makes possible a wide variety of new experiments particularly by means of laser induced fluorescence. . . . The method has great sensitivity and may allow the polarisation of individual vibration-rotation states to be measured as a function of the scattering angle’.How right they were will become apparent as this review progresses though it is only in the last year or so that the hope expressed in their second sentence has become an experimental reality. A landmark conference which was organised by Richard Bernstein Dudley Herschbach and Raphael Levine in Jerusalem in 1986,8 marked a turning point in the evolution of the study of reaction stereodynamics. One of its prime concerns was the measurement of the vector (spatial) correlations between k j kA and jA the relative velocities and rotational angular momenta of the colliding or dissociating reagents and products and their analysis in terms of the stereodynamics they reflect.The power of polarised Doppler-resolved probing in revealing the stereodynamics of molecular photodissociation5 was already strongly apparent (following a key paper by Zare and Herschbach in 19635 and the subsequent development of the necessary analytical theory during the 1980s particularly by Greene and Zare9 and Dixon10). Indeed so successful had the strategy been that it distracted attention away from bimolecular reactions throughout most of the 1980s; it was not until the current decade that the balance was reversed though the current ‘new wave’ of experiments leans heavily on the earlier studies of molecular photodissociation. The most powerful strategy uses polarised photodissociation of an appropriate molecular precursor to generate velocity aligned atomic or free radical reagent ‘beams in a bulb’; the products scattered from their subsequent secondary collisions are probed by a second tunable polarised laser after a short delay (short enough to ensure ‘single collision conditions’ and the avoidance of subsequent collisional relaxation).The philosophy is best illustrated by one of the first pioneering experiments to be conducted in this way in Richard Bersohn’s laboratory at Columbia University.11 They compared LIF Doppler-resolved profiles of (i) the H atoms generated by polarised photodissociation of the precursor molecule H2S [reaction (2)] and (ii) the D atoms recoiling from their subsequent collision with the target molecule SiD4 [reaction (3)].H S + h n (193 nm)Æ (2) 2 H + SiD (3) ææææ H + HS N2 reaction).11 4 ––? HSiD3 + D The H atoms are generated with very high translational energy in excess of 150 kJ mol21 and their velocity is aligned perpendicular to the polarisation vector e of the absorbed photons (since the electronic transition is polarised perpendicular to the molecular plane of the H2S). H atoms recoiling perpendicular to e present a double-peaked Doppler contour reflecting their motion towards or away from the ‘observer’; when the probe laser is directed parallel to e the Doppler profile narrows and only presents a single central peak. Remarkably the D atoms generated in the secondary reaction (3) present very similar behaviour—the D atoms tend to emerge with velocities directed parallel to those of the incident H atom i.e.with k || kA but moving more slowly to conserve momentum. The results have been interpreted in terms of a displacement mechanism proceeding through transition state structures approximating a trigonal bipyramid i.e. a ‘collinear’ inversion mechanism (reminiscent of an S This innovative experiment provided a flavour of the ‘shape’ of experiments to come but despite its elegance it was limited in its scope in two or three important respects. Since the centreof-mass of the colliding reagents lies very close to the heavy target molecule the velocity of the centre-of-mass of the reaction system is very small. In consequence the velocities (observed in the laboratory frame via the Doppler spectrum) of products scattered in the same direction as the reagent H atoms i.e.forwards are virtually the same as any that are scattered backwards. The interpretation in terms of forward scattering although highly plausible remains an interpretation. Secondly the experiment was restricted to monitoring the atomic product only. The full flowering of the new strategy has not been slow however and in the last three years it has been applied at steadily evolving levels of refinement to a wide range of bimolecular reactions. The measurement of both linear and angular momentum correlations in reactions involving both atomic and molecular reagents has provided a wealth of new dynamical information which either complements or supersedes that gained from crossed molecular beam studies.The developments foreseen by Case McClelland and Herschbach over twenty years ago7 are at last being realised; for illustrative examples read the rest of this Review and for alternative sources see the excellent earlier reviews written by Orr-Ewing and Zare12 and/or by (two of) the present authors13 (the reader can judge as to the excellence of the latter). 3 The stereodynamics of photon initiated bimolecular reactions concepts and machinery 3.1 The concept Consider the idealised photon initiated reaction sequence14,15 (4) and (5) in which monoenergetic velocity aligned atomic (pol) hn (4) t (5) Ê (6) æææÆ A + D mA � v A m + mBC ~ Ë Á AD At + BC|iÅ ? AB|fÅ + C reagents At generated in step (4) collide with a stationary target molecule BC in quantum state |iÅ to generate a product AB in a quantum state |fÅ which is subsequently probed (stateselectively) via Doppler (or time-of-flight) resolved laser excitation.The Doppler broadened spectrum of the scattered products reflects their speed distribution in the reference frame of the observer i.e. in the laboratory or LAB frame but this speed distribution is determined by the dynamics of the collision in the molecular frame referenced to the relative velocity of the reactants k with an origin at the centre-of-mass of the colliding reagents i.e. the CM frame. If the reagent atom is moving much faster than the target molecule the velocity of the centre-of-mass vCM will be given by eqn.(6). vCM (7) � � A Fig. 2 presents simple velocity vector diagrams (‘collapsed’ Newton diagrams) for reactive collisions in which the molecular reagent velocity vBC ~ 0. The LAB frame velocities of products AB (or C) scattered at an angle qt and velocities wAB (or wc) in the CM frame will be given by eqn. (7). If the vAB = vCM + wAB or vC = vCM + wC products are scattered forwards i.e. with qt ? 0° their LAB velocity will be enhanced by vCM; if they are scattered backwards their LAB velocity will be diminished by vCM —the simple vector sum in eqn. (7) provides a means of establishing the full product angular distribution. The sensitivity is maximised when nCM = wAB (or wC). In the example of the H + SiD4 reaction discussed above reactive scattering of a light atom (H) by a heavy target molecule (SiD4) nCM 8 wC occurred and it was not possible to distinguish between forward and backward scattering.Fortunately however this represents an extreme and rare situation. Chemical Society Reviews 1998 volume 27 407 The Law of Cosines is given in eqn. (8). If the kinetic energy (8) n2 AB = n2 CM + w2 AB + 2nCMwAB cos qt AB|fÅ of A the internal energy of BC and the exoergicity (i.e. the energy release) in the reactive collision (5) are known a priori (as they commonly are) and C carries no internal excitation then selection of the product quantum state |fÅ will fix the kinetic energy of the scattered products and therefore the speed w through energy conservation.Under these conditions each laboratory speed,nAB|fÅ maps uniquely onto a centre-of-mass scattering angle qt (see Fig. 2) and the LAB speed distribution determined from the Doppler spectrum provides the stateselected product scattering angular distribution in the CM frame P(qt) proportional to the state resolved differential cross-section. In a less than ideal world of course the method is not quite that simple but the central concept is sustained. 3.2 The unobserved products In the vast majority of reactions the constraint limiting the (usually) unobserved product C of the bimolecular reaction (5) to a structureless atom will not operate. Even when C is monatomic it is likely to be an open shell atom the energy of which may well be split through spin-orbit interaction e.g.I(2P3/2) and I(2P1/2). If the unobserved product is molecular as for example in a reaction of type (9) it will be generated in a At + BCD ? AB|fÅ + CD|fAÅ (9) range of ro-vibrational states and the monitored products AB |fÅ will be scattered with a corresponding spread of kinetic energies/velocities wAB|fÅ. Under these more general circumstances the AB product speed distribution in the LAB frame will reflect both the CM angular distribution (the AB|fÅ stateresolved differential cross-section) and the kinetic energy distribution (reflecting the internal energy distribution in the correlated partner). Were this the end of the story there would be less to relate but fortunately the increased complexity also offers increased opportunity since there are ways of extracting both the angular distributions and the correlated kinetic energy distributions from the Doppler-resolved spectra.16–18 The key lies in taking advantage of measurements using alternative detection geometries e.g.with the photolysis and detection beams aligned either parallel or perpendicular to each other and with the further option of a parallel or perpendicular alignment of their polariisation vectors. The alternative geometries allow the scattered products to be ‘viewed’ from different perspectives in the LAB frame. An example set of results derived in the above fashion is shown in Fig. 3. The data are for the OH products of the H + CO2 reaction,17 and are plotted as a scattering map representing the joint distribution of the CM speed (wAB|fÅ) and scattering Fig.2 ‘Collapsed’ Newton diagrams for the situation in which the velocity of the target molecule can be neglected i.e. nBC ~ 0. CM scattering into the forward (upper figure) and backward hemispheres (lower figure) leads to different product speeds nAB in the laboratory frame. nCM is the velocity of the centre-of-mass and wAB is the AB product velocity in the CM frame. Chemical Society Reviews 1998 volume 27 408 angle (cos qt) of the OH. Although the scattering map is reminiscent of that shown for F + H2 in Fig. 1 the new map shown in Fig. 3 corresponds to the scattering of a specific product OH quantum state (n A = 0 N = 5). Furthermore measurement of the product speed distribution together with the constraint of energy conservation allows the correlated kinetic energy (or product speed) distribution to be converted into the correlated internal energy distribution P(EfA) in the unobserved products which accompany AB|fÅ.3.3 Rotational polarisation product angular momentum distributions As well as reflecting the (vectorial) angle-resolved velocity distributions of the scattered reaction products and the (scalar) product internal quantum state distributions the Doppler (or REMPI time-of-flight) contours also reflect a third dynamical factor—the state-resolved product angular momentum distributions. 6,7,12–14,19–21 Their influence is imaged in the LAB frame by the dependence of the Doppler broadened spectrum on the rotational polarisation of the scattered products.If the stateresolved product angular momentum distributtion is anisotropic (i.e. polarised) in the LAB frame both the intensity and the shape of the Doppler spectrum will vary with the experimental configuration and the type of rotational transition selected by tuning the probe laser e.g. Q— or P R—. To appreciate the dynamics properly however they need to be viewed through the ‘molecule’s eye’ i.e. in the molecular or CM frame which has its z-axis parallel to the relative velocity of the reactants; in this frame the laboratory velocities are replaced by the relative velocities k and kA and the k kA jA distribution can be defined in terms of the angles shown in Fig. 4. qt represents the scattering angle between the reagent and product velocity vectors k and kA.A preference for qt?0° for example would indicate forward scattering in the CM frame; a preference for qt?180° would indicate backward or ‘rebound’ dynamics. The angles q angles of the rotational angular momentum jA referenced to the reagents velocity k and the scattering plane (k kA). A distribution peaking at (q example would reflect a preference for the products to spin away from the reactive collision like a frisbee or discus; a j and fj are the polar and azimuthal j fj) = (90° 90°) or (90° 270°) for distribution peaking at (qj fj) = (90° 0°) could indicate either Fig. 3 A contour plot (upper figure) and three-dimensional view (lower figure) of the product scattering map for the reaction H + CO2?OH(n A = 0 NA = 5 AA) + CO at a mean collision energy of 240 kJ mol21 adapted reference.17 The data reveal a forward-backward peaking product angular distribution with a bias in the backward direction and a broad OH(nA = 0 NA = 5 AA) CM speed distribution which peaks at high OH speeds.This speed distribution reflects the internal energy distribution in the CO co-products to OH(n A = 0 NA = 5 AA) which from energy conservation must be born internally cold. a preference for cartwheel motion or propeller-like motion depending on the preferred scattering angle qt. The full CM angular distribution of k kA and jA is expressed in terms of three angles i.e. P(qt qj fj); it describes the dependence of the angular momentum polarisation on the CM scattering angle.6,7 When integrated over scattering angle the resulting (scattering angle averaged) rotational angular momentum distribution P(qj fj) can be displayed in the form of a polar map—a typical example taken from classical trajectory data for the F + H2 reaction,22 is shown in Fig.5. The full CM angular distribution may be written (semiclassically‡) in terms of expansions either in bipolar or spherical harmonics.12,19,20 The ‘moments’ or coefficients defining these expansions are known either as the polarisation dependent differential cross-sections19 (PDDCS’s) or the related bipolar moments.20 Only the low order moments of the angular momentum polarisation distribution can be extracted from the dependence of the Doppler contours on pump-probe geometry and rotational transition.20,21 Low order even alignment moments are obtained using linearly polarised probe laser radiation whilst low order odd orientation moments may in principle be determined using circularly polarised light.3.4 Transformation from the LAB frame to the CM frame Transformation of data obtained in the experimental laboratory frame to the molecular centre-of-mass frame can be effected through a least squares fitting procedure based upon a set of basis functions,24 which depend parametrically on the angles of interest. The simplest example reflects the correlation between k and kA i.e. the conventional differential cross-section determined by the angular distribution in qt.An example basis Fig. 4 (a) The definition of the scattering angle qt and the polar angles qj fj which define the direction of jA with respect to the k–kA scattering plane. In part (b) these definitions are illustrated for AB products rotating and scattering in the particular directions shown. The reactant relative motion defines the CM z-axis (see the left figure) and the xz plane is defined by the relative motion of the product molecules (right figure). ‡ A comprehensive quantum mechanical treatment of angular momentum polarisation in elementary bimolecular reactions has recently been given by Miranda and Clary.23 This paper also contains an excellent review of angular momentum polarisation in elementary chemical reactions. set which was used to analyse the dynamics of reaction (10) is (10) H + CO2 ? OH + CO shown in Fig.6; the basis functions depend parametrically on qt and the energy EAt released into translation in the scattered Fig. 5 An example of a polar plot of the angular momentum polarisation for the F +H2(n = 0 j = 0) ?HF(nA = 3) + H based on the classical trajectory calculations presented in reference 22. The upper figure shows the full distribution employing an expansion in seven moments. The lower figure illustrates the oriented rotational motion of the vibrationally excited HF products. As shown in Fig. 4 the centre-of-mass frame xz plane corresponds to the scattering plane and thus contains the HF product relative velocity vector. Recall that the particular HF(n A = 3) products in question are scattered in the forward direction (see Fig.1) i.e. with velocity vectors lying nearly parallel to the z axis. Fig 6 Example basis functions for the H + CO2 reaction at a mean collision energy of 240 kJ mol21. Only the dependence on CM scattering angle q is shown for a fixed value of the fractional kinetic energy release ft = t 0.725. The basis functions were employed to fit the sum (a) and difference (b) of experimental Doppler profiles obtained in parallel and perpendicular pump-probe geometries. Those shown in (a) depend only on the OH speed distribution in the LAB frame whilst those in (b) depend on the OH LAB frame translational anisotropy. The derived CM polar scattering map has already been shown in Fig. 3. 409 Chemical Society Reviews 1998 volume 27 products (expressed as the fraction ft · EAt/Etotal) which itself is determined by the internal energy distribution in the unobserved CO fragment.The corresponding polar scattering map derived from fits to experimental data has already been shown in Fig. 3. The state-resolved differential cross-sections may and generally do depend on the selected product channel and thereby reflect the detailed anatomy of the reactive collisions. Crossbeam scattering experiments which employ conventional timeof-flight mass spectrometry rather than optical spectroscopy are generally far more limited in their scope. It is possible to extend the basis function strategy to determine additional parameters for example the collision energy dependence of the reaction probability measured by the reaction cross-section sR(Et),25 or the rotational polarisation of the scattered products reflected in the sensitivity to the choice of detection via Q— or P R— probe transitions.17 The most powerful formulations enable a set of polarisation dependent differential cross-sections to be determined from which the low order moments of the full spatial distribution of the products linear and angular momenta P(qt qj fj) can be determined17,21,26,27 —realising the dream expressed by Case McClelland and Herschbach more than 20 years ago8 3.5 A few caveats In general the contours of the Doppler broadened product spectra can be influenced by a range of factors some of which can be controlled by the experimenter and some of which reflect the dynamical behaviour of the reactive collisions i.e.what is actually sought. The former (may) include i The thermal spread of velocities in the reagent source and the molecular target;14,28 these can be accommodated by appropriate averaging or virtually eliminated by coexpanding the source and target molecular reagents in a supersonic nozzle expansion. ii The spread in the internal quantum states of the molecular reagent; these too can be constrained by nozzle beam expansion or (so far rarely) the reagent may be state selected through prior optical excitation. iii A spread in the collision energies associated with the dynamics of the photon initiation step e.g. when the source is a polyatomic molecule.This can be accommodated either by assuming the dynamics/cross-section of the subsequent bimolecular reaction to be insensitive to the spread of collision energies or better by making a virtue of necessity and using the (known) spread to explore the energy dependence25 or better still by choosing (if possible) an alternative mono-energetic but ‘tunable’ photolytic source. 4 Case histories 4.1 Overview The ability to determine the distributions of linear and rotational angular momenta among the quantum state-resolved products of reactive molecular collisions represents literally a quantumjump in our ability to penetrate the microscopic world of chemical reaction dynamics. Their interpretation in the light of accurate quantum scattering and/or quasi-classical trajectory (QCT) calculations on reliable ab initio potential energy surfaces provides a profoundly detailed view of the dynamics of reactive molecular collisions; witness the recent triumphs in understanding the dynamics of the reactions of H(D) and F with H2 and its isotopes.2 The development of polarised Dopplerresolved optical detection methods has provided a new and extremely powerful general strategy towards the dynamicists’ ultimate objective—a full three-dimensional perspective of the stereodynamics of individual reactive collisions.The remainder of this Review provides a flavour of some of the prizes that have already been won (hard-won) through their application. Chemical Society Reviews 1998 volume 27 410 Early intimations of what was to come arrived at the start of the decade with the studies of Bersohn and his co-workers11— the first to exploit the anisotropy of molecular photodissociation to probe the dynamics of a subsequent secondary reaction e.g.the displacement reaction (11) see Section 2. These studies (11) H + CD4(SiD4) ? CD3H(SiD3H) + D were followed rapidly by new experiments which utilised Doppler resolved polarised laser detection to probe the full range of state-resolved (k kA jA) correlations among the scattered molecular products of reactive collisions. Examples include reactions (12)–(21) many of which have either been mentioned already or were discussed in earlier reviews where further leading references can be found.12,13 Reaction (21) the first to employ both product and reagent state selection will be discussed in more detail in Section 4.2.O(1D) + N2O ? NO|fÅ + NO|fAÅ (ref. 13) O(1D) + H2 18O ? OH|fÅ + 18OH|fAÅ (ref. 13) H + O2 ? OH|fÅ + O (ref. 24) O(1D) + HCl ? OH|fÅ + Cl (ref. 25) O(3P) + CS ? CO|fÅ + S (ref. 29) O(3P) + H2S ? OH|fÅ + HS (ref. 29) H + CO2 ? OH|fÅ + CO (ref. 17) H + H2O ? OH|fÅ + H2 (ref. 18) O(1D) + CH4 ? OH|fÅ + CH3 (ref. 16) Cl + CH4|iÅ ? HCl|fÅ + CH3 (ref. 15) 2 (12) (13) (14) (15) (16) (17) (18) (19) (20) (21) The pioneering studies of reactions (14)24 and (20)16 provided the first examples of reactive collisions proceeding through (potentially) bound intermediates respectively HO and CH3OH formed through addition or insertion.In each case a detailed analysis of the polarisation-dependent Dopplerresolved laser induced fluorescence of the scattered OH products provided a rich foretaste of the delights to come. The OH radical (like NO) is wonderfully equipped to act as an eloquent reporter of the stereodynamics involved in its formation. Its ground electronic state carries both spin and electronic orbital angular momentum and its fluorescent electronic transition A2S / X2P presents both Q and P,R branches to allow a full determination of the vectorial distribution of k kA jA. Furthermore the coupling of its molecular rotation jA with its electronic orbital angular momentum L splits each rotational level into two components known as ‘lambda doublets’.They can be identified with radicals in which the odd electron occupies either the p-orbital lobe directed perpendicular to the rotation plane p (AB) or the lobe lying in the rotation plane p(AA) see Fig. 7. Their unequal population would reflect a propensity for electronic orbital alignment and perhaps an insight into the electronic motion within the collision complex. Initially it was thought that the vector correlations associated with the nuclear motions i.e. those involving k kA and jA would be independent of those involving the electronic motions but it was not to be. The first clue came in analysing the Doppler contours of OH scattered from the reaction of H with O2 [reaction (14)]. It simply was not possible to obtain consistent results unless the OH products scattered into the two alternative lambda-doublet states were endowed with different rotational polarisations ‘frisbee-like’ for the p(AA) states but isotropic for the p(AB) partners.24 4.2 Some case studies the state-of-the-art Some of the most recent optical studies of the stereodynamics of bimolecular reactions have been conducted in Oxford and in Stanford by Richard Zare’s group; they provide excellent illustrations of the present state-of-the-art.Key studies include the reaction of atomic chlorine with methane and ethane and some of their deuterated isotopomers,21,27,30 and of electronically excited oxygen atoms O(1D2) with hydrogen and methane.16,26,31There are fundamental differences in the way the two reagents interact in reactions (22) and (23) which are associated with the differing topography of the potential energy surfaces that the two attacking atoms encounter when they approach the C–H (or H–H) bond.In reaction (22) the chlorine atoms prefer to attack at the H atom end of the bond and need to surmount an energy barrier for the reaction to proceed; the net process is an endothermic ‘abstraction’ reaction. In contrast singlet oxygen atoms prefer to attack the H–H or C–H bond ‘side-on’ and ‘insert’; reactions (23) and (24) are strongly exothermic and proceed over a deeply attractive potential energy surface (correlating with H2O or CH3OH) which presents little or no barrier to the approach of the oxygen atom (see Fig. 8). The strikingly different mechanisms for the Cl and O(1D2) atom reactions are reflected in their own characteristic stereodynamics.Cl(2P3/2) + CH4 ? HCl|n A = 0 1 jAÅ + CH3 O(1D2) + CH4 ? OH|nA @ 4 jAÅ + CH3 O(1D2) + H2 ? OH|nA @ 4 jAÅ + H (22) (23) (24) That is not the whole story however since the O(1D2) atom may also have an opportunity to interact with the target molecule over one (or more) electronically excited potential energy surface(s) correlating with dissociative electronically excited states of H2O or CH3OH.31–33 On its own the O(1D2) atom is five-fold degenerate but under the lowered symmetry of the collision this degeneracy is lifted. For O(1D) + H2 the lowest of these in a linear configuration designated 1S would correlate with the ground state of water but the next two 1P and 1D correspond to electronically excited states.Ab initio computations of the 1P surface (which splits into 1AA and 1AB surfaces in a bent collision complex) indicate a small energy barrier for near collinear configurations and a surface topography closely resembling that for the interaction of H2 with the halogen atoms F or Cl (one up from O in the Periodic Table) see Fig. 9. If the collision energy is high enough to surmount the low entrance barrier abstraction could begin to compete with insertion. But that may not be the whole story either. . . ! There is a possibility that collisions initially governed by one potential energy surface may switch ‘their allegiance’ to another at shorter range. The surfaces may intersect allowing reactive collision trajectories to ‘hop’ from one surface to another if this Fig.7 The limiting high jA unpaired electron density in the AA and AB lambda-doublets in the ground 2Pstate of OH. Q—transitions probe the AB level while P/R— transitions probe the AA lambda-doublet level. occurs some collisions which started out on an ‘abstractive’ pathway over an excited energy surface might still end up on an ‘insertion’ pathway proceeding over the ground state surface. Molecular reaction dynamics can be a subtle business. 4.2.1 The reaction of Cl with CH4 The reaction of Cl(2P3/2) atoms with CH4 (in its vibrational ground state) reaction (22) is endothermic (DHÆ(0 K) = +7.9 Fig. 8 Sections through the potential energy surfaces for the reactions of O(lD) with CH4 33 (upper panel) and with H2 (lower panel).32 The reactant valley in both figures is on the upper left and is shown by the black arrow.Two alternative product channels are shown as red arrows. Fig. 9 The variation of the potential energy along the reaction coordinate for the 1AA 2AA and AB electronic states of the O(lD2) + H2 reaction adapted from Fig. 1 of Schatz et al. Faraday Discuss. Chem. Soc. 108 reference 32. The curves shown in (b) are for collinear O–H–H configurations while those in (a) are for bent O–H–H. 411 Chemical Society Reviews 1998 volume 27 3/2) kJ mol21); in addition there is also a considerable energetic barrier to reaction (ca. 15 kJ mol21) which limits the reaction probability measured by its reactive cross-section sR.The cross-section can be enhanced by increasing the collision energy or much more effectively through vibrational excitation of the CH4; the effect of reagent vibrational excitation on the reaction stereodynamics has been investigated by Zare and co-workers,15,30 using infra-red (IR) laser radiation to excite one quantum of the asymmetric stretching mode (n3 = 1). The two reagents Cl2 and CH4 were co-expanded as a ‘mixed’ molecular beam in helium (to minimise the effects of thermal motion on the spread of reagent velocities) and crossed with an IR laser beam tuned to excite the transition n3 = 0 ? 1 in the CH4. The abstraction reaction (22) was initiated by photolysis of the Cl2 using a second linearly polarised UV laser operating at 355 nm which produces a velocity-aligned pulse of Cl(2P reagent atoms.After a delay of a few tens of nanoseconds the HCl|n A jAÅ products were interrogated by a third tunable (REMPI) laser pulse which selectively ionised the products from the specifically populated rovibrational states |n A jAÅ. Their velocity distribution was measured by recording the timeof-flight of the HCl+ ions to a remote detector employing a strategy known as ‘core extraction’,15 which differs a little from the Doppler technique described earlier. However using analogous procedures to those outlined in Section 3 the LAB frame velocity distribution of the state-selected reaction products could be converted into a family of (state-resolved) angular distributions see Fig.10 referenced to the relative velocity of the reagents—the key step which provides the first direct insight into the dynamics of the reactive collisions and of the molecular interactions which govern them. The experimental results were intriguing HCl products generated in |n A = 0 high jAÅ were scattered backwards and sideways with respect to the velocity of the incoming Cl atom. In contrast the HCl molecules generated in |n A = 1 jAÅ were scattered predominantly forward for products with low rotation but shifted towards the backward hemisphere as the product rotation increased. Very similar results were obtained when CD3H was substituted for CH4 suggesting a ‘spectator’ role for the unobserved partner fragments CH3 or CD3.The results could be understood if HCl |n A = 1 high jAÅ products Fig. 10 Experimental HCl(n A = 1 jA = 0–3) product state-specific time-offlight spectra (left panel) together with the fits to the data (solid lines) and the derived CM angular distributions (right panel) for the reaction Cl + CH4(nA3 = 1) ?HCl(nA3 = 1 jA = 0–3) + CH3. The figure is adapted from reference 30. Chemical Society Reviews 1998 volume 27 412 were generated by near head-on ‘hard-sphere’ collisions of Cl with the H–C bond leading to backward or ‘rebound’ scattering while the HCl |n A = 0 high jAÅ was associated with more ‘glancing’ trajectories (see Fig. 11). Forward scattered products HCl|n A = 1 low jAÅ were generated through peripheral or tangential collisions in which the C–H bond was oriented perpendicular to the velocity of the incoming Cl atom.A similar dynamical mechanism described as ‘peripheral abstraction’ had been encountered earlier in a study of the O(1D2) + N2O reaction.13 This interpretation was confirmed by using a polarised IR laser to pre-orient the C–H bond axis through excitation of the symmetric C–H stretch mode n1 in the ‘designer’ target molecule CD3H.30 The photolysis laser polarisation could also be used to control the direction of the (velocity-aligned) Cl atoms and it was possible therefore to align the vibrating C–H bond either parallel or perpendicular to the reagent Cl atom velocity. As expected the switch from an ‘end-on’ to a ‘side-on’ collision geometry enhanced the level of forward scattering from CD3H(n1) in agreement with the mechanism proposed.These quantum state-dependent scattering angular distributions provide a uniquely resolved dynamical picture of the full ensemble of reactive molecular collisions detail that would have been washed out if the HCl products had been detected without distinguishing the individual rovibrational states c.f. a conventional crossed molecular beam experiment. In effect the spread of product state-resolved angular distributions reflects the range of initial collision conditions and the ‘accuracy’ of the initial collision trajectories i.e. how closely they approach the target molecular ‘bull’s eye’ measured by the impact parameter b and its direction or angular orientation.The HCl generated through reaction with vibrationally unexcited CH4 was generated exclusively in |n A = 0Å with low rotational excitation and predominantly backward scattered. This was attributed to a ‘tightening’ of the conditions for reactivity to collision trajectories oriented along the H–CH3 axis with low impact parameters (i.e. near direct hits). The angular spread of successful collision trajectories i.e. the target molecule’s ‘coneof-acceptance’ widens when the molecule is vibrationally activated. 4.2.2 The reaction of O(1D2) with H2 and CH4 The reaction of electronically excited singlet oxygen atoms with H2) reaction (24) is strongly exothermic [DHÆ (0 K) = 2182 kJ mol21] and at room temperature where the mean collision energies are low (ca.3.6 kJ mol21) reaction over the ground state potential energy surface will be dominant since the surface presents no barrier. Not surprisingly the reaction has a very Fig. 11 Schematic illustration of the contrasting reaction dynamics leading to HCl products in n A = 0 and n A = 1 in low and high jA levels for the Cl + CH4(n3 = 1) reaction. The figure is adapted from reference 30. Some of the experimental data on which the figure is based is shown in Fig. 10. high rate constant at room temperature. The OH fragments tend to spin away from the collision with high rotational angular momentum jA and all the energetically accessible vibrational levels OH(n A = 0 ? 4) are populated. The angular distribution of the scattered products was first measured at Berkeley by Yuan Lee and his co-workers34 using the method of crossed molecular beams coupled with mass spectrometric detection.The distribution necessarily averaged over all product quantum states was nearly symmetric with peaks in both the forward and backward hemispheres— referenced to the incoming reagent oxygen atom. Among other possibilities this could be interpreted to indicate the existence of a transiently bound rotating (H2O) collision ‘complex’ with a sufficiently long average lifetime åtdÅ with respect to its rotational period åtrÅ to result in a scrambling of the angular distribution i.e. with no preference for either of the forward or backward directions (symmetric about qt = 90°). The rotation derived from the orbital angular momentum of the collision partners would be expected to be fast with a period in the femtosecond regime in view of the low moment of inertia of the (H2O) complex.Subsequent calculations using classical mechanics to simulate the ensemble of reactive collision trajectories on a computed ab initio potential energy surface confirmed this time-scale but also indicated a somewhat more complicated story.13,31 Fig. 12 shows the predicted state-resolved angular distributions of the products OH |n A jAÅ for two different vibrational quantum states |n AÅ. Products generated in |n A = 4Å are predicted to be symmetrically distributed but those formed in |n A = 0Å are mainly backward scattered. Shortly after these calculations were published the first stateresolved experimental study of the reaction was reported using the newly-developed strategy of photon initiated reaction using N2O as the photolytic source of O(1D) and optical stateresolved detection of the OH|n A jAÅ products.31 Since the scattered reaction products are probed using polarised laser radiation and the measurements are Doppler selective they are sensitive both to the velocity and the rotational polarisation of Fig.12 The QCT derived angular scattering maps for the reaction O(lD) + H2 ? OH(n A jA) + H .13 Part (a) is specific to the OH(jA = 0–10) products in nA = 0 whilst part (b) is for the jA = 0–17 products in n A = 4. The data can be compared with the experimental OH state-specific scattering maps obtained for the O(lD) + CH4 shown in Fig.14. the individually state-selected products allowing the construction of a unique map of the correlated vectorial distribution between the reagent velocity k and the products’ velocity and rotational angular momentum kA and jA in the aftermath of the reaction. Fig. 13 shows the results for k kA the angular distributions for the OH products formed in |n A = 0 jA = 5 or 14Å show a preference for backward scattering in good agreement with the predictions of classical trajectory simulation. 26,31 Similar experiments have also been conducted for the analogous reaction of O(1D) atoms with CH4 reaction (23).16 This system is more ‘user-friendly’ since the scattered products OH and CH3 with almost the same relative molecular mass separate with equally fast velocities in marked contrast to the product pair OH and H where momentum conservation greatly constrains the velocity of the OH.The experimental scattering maps for OH|n A = 0 jA = 5Å and OH|n A = 4 jA = 8Å shown in Fig. 14 can be compared with the corresponding maps for the reaction of O(1D2) with H2 presented in Fig. 12. The similarities are striking. The scattering maps for OH|n A = 0Å both display a strong backward peak and a weak forward peak while those for OH|n A = 4Å both approach forward-backward symmetry. The correlations between the incoming and outgoing relative velocity vectors k kA i.e. the scattering distributions are remarkably similar for the two reactions not surprisingly perhaps in view of the similarity of their potential energy surfaces shown in Fig.8. The scalar energy correlations for the channels producing OH|n A = 0Å are very different for the two reactions however since the kinetic energy released to the OH|n A = 0Å from the reaction of O(1D2) with CH4 is remarkably low most of the exoergicity appears as internal excitation of the (unobserved) polyatomic CH3 fragment implying that a considerable redistribution of vibrational energy takes place during the lifetime of the transient (CH3OH) complex åtdÅ a lifetime that is shorter than its mean rotation period åtrÅ. The Fig. 13 Comparison between the experimental (bold line) and QCT calculated (dashed line) angular distribution for the reaction O(1D) + H2? OH(n A = 0 NA) + H.26 Part (a) is for the NA = 5 products whilst part (b) is for the NA = 14 products.413 Chemical Society Reviews 1998 volume 27 increased moment of inertia in (CH3OH) compared to (HOH) suggests a considerable increase in the relevant time-scales. The rotational period provides a ‘clock’ against which the lifetime of the intermediate collision complex can be measured. Channels with intermediate lifetimes longer than the mean rotational period will be expected to display symmetric product angular distributions. The problem is how to measure the absolute ‘clock rate’ and thereby ‘time’ the reactive events. One way is to use an ultrafast laser to probe the rate of appearance of the scattered products directly on a femtosecond time-scale. This has actually been done by Stephenson and van Zee,35 who probed the rate of appearance of OH|n A = 0 jAÅ from a (CH3OH) collision complex prepared by photolysis of O3 [the source of O(lD)] bound to CH4 in a van der Waals complex generated in the low temperature environment of a molecular beam expansion.The result indicates a transient complex lifetime åtdÅ ~ 3 ps. Unfortunately the corresponding timescales for the analogous reaction with H2 are too fast to allow a similar direct measurement. There is a way out however via simulations using classical trajectory calculations. They indicate rotational periods åt increasing from åtdÅ ~ 30 fs to 85 fs for decomposition into OH|n A = 0Å to |n A = 4Å fully consistent with the change in the angular distribution from backward scattering to near symmetric.For OH scattered into the higher vibrational levels |n A = 4Å the lifetimes åtdÅ ~ åtrÅ and the angular distributions display nearly equal forward and backward peaks. When the OH fragments are scattered into the lowest quantum states |n A = 0Å the complex lifetime åt predominantly into the backward hemisphere because the intermediates make on average less than a full rotation before they proceed to products. The real-time and rotational clocks used to analyse the results for the reaction of O(1D2) with CH4 indicated time-scales in the picosecond rather than the femtosecond range—time enough for extensive vibrational energy redistribution into the CH3 fragment and associated no doubt with the larger moment of inertia of the rotating (CH rÅ ~ 100 fs but (HOH) lifetimes dÅ @ åtrÅ and the fragments scatter 3OH) Fig.14 The experimentally derived angular scattering maps for the reaction O(1D) + CH4 ?OH(nA jA) + CH3. Part (a) is for OH(nA = 0 NA = 5 AA) and part (b) is for OH(nA = 4 NA = 8). The data on which the plots are based is taken from reference 16. Chemical Society Reviews 1998 volume 27 414 dÅ/åtrÅ in the channels generating OH |n A complex. Nevertheless the remarkable similarity between the two sets of scattering data (Fig. 12 and 14) suggests that the relative time-scales åt = 0Å and |n A = 4Å are little changed when CH4 is substituted for H2. We turn now to the collisional energy dependence of the reaction. If there is truly no barrier in the entrance channel the reaction probability measured by the cross-section sR should actually decrease with increasing collision energy.O(1D) atoms approaching the target molecule at long range (large impact parameters) may still be drawn into the ‘reaction field’ provided they are not travelling too rapidly. If they are then they may avoid reaction because their momentum can carry them out of harm’s way. To be captured they would need to follow a closer trajectory (with a smaller impact parameter). The ‘reactive’ cross-sectional area presented by the target will shrink at elevated collision energies and the ‘excitation function’ sR (Et) should fall as the collision energy Et increases. This is precisely the behaviour observed.31 Fig. 15 shows the excitation function for the scattering of OH|n A = 0Å from collisions of O(1D) with H2 determined through analysis of its Dopplerresolved spectral band contours; it compares very well with the simulation based upon classical trajectories over the ab initio potential energy surface for the ground electronic state of the collision complex.Suppose the reactive collisions had proceeded instead over the initially repulsive electronically excited potential energy surface(s). In this situation increasing collision energy should increase the probability of reaction; once the energy was sufficient to overcome the initial barrier the excitation function would be expected to increase monotonically from a threshold value. Trajectory simulations for this pathway32 also predict totally different patterns of energy and momentum disposal in the scattered products with abstraction favouring high vibrational excitation and low rotation (the inverse of the distribution generated via an insertion pathway over the ground electronic Fig.15 Comparison between the experimental (bold line) and QCT calculated (dashed line) state-resolved excitation functions sR(Et) for the reaction O(1D) + H2 ? OH(n A = 0 NA) + H leading to OH products in NA = 5 (a) and NA = 14 (b).26 potential energy surface) and a strongly backward scattered angular distribution. Although some experimental evidence has been presented in favour of this alternative abstraction pathway at elevated collision energies its actual (or even real) contribution remains to be established.31,32 5 Forward look The case histories presented above illustrate the power of the new stereodynamical methods in exploring the intimate details of individual reactive molecular collisions.In the past three or four years experimental procedures have been optimised and the new results obtained have revealed many new features of state-resolved reaction dynamics which were hitherto unforeseen. Diversity rather than conformity appears to be the rule; each reactive channel displays its own unique dynamical signature. Particularly exciting discoveries are being made in areas where experiment and theory interact most closely and the results from the new experiments are proving a demanding test of ab initio theory.It is perhaps sobering that many dynamical aspects of the ‘simple’ gas phase reaction O(lD) + H2 remain to be rationalised theoretical studies of this reaction highlight the current ‘state-of-the-art’ in ab initio theory. Following in the footsteps of Herschbach and co-workers,6,7 theorists and experimentalists are beginning to turn increasingly to angular momentum polarisation to provide additional insights into reaction mechanism. Such studies will help maintain (for some at least) the ‘buzz and passion’ which animates the world of molecular reaction dynamics. 6 Acknowledgements It is a pleasure to thank our past and present colleagues whose work has contributed to this Review Article and to acknowledge particularly Professor Javier Aoiz at the Complutense University Madrid with whom we have enjoyed a long and fruitful collaboration.7 References 1 D. M. Neumark A. M. Wodtke G. N. Robinson C. C. Hayden and Y. T. Lee J. Chem. Phys. 1985 82 3045. 2 D. E. Manolopoulos J. Chem. Soc. Faraday Trans. 1997 93 673 and references therein. 3 E. J. Murphy J. H. Brophy G. S. Arnold W. L. Dimpfl and J. L. Kinsey J. Chem. Phys. 1979 70 5910. 4 P. L. Houston Acc. Chem. Res. 1995 28 453. 5 R. N. Zare and D. R. Herschbach Proc. IEEE 1963 51 173; P. L. Houston J. Phys. Chem. 1987 91 5388 and J. P. Simons ibid 5378. 6 D. E. Case and D. R. Herschbach Mol. Phys. 1975 30 1537 and references therein. 7 D. E. Case G. M. McClelland and D. R. Herschbach Mol. Phys. 1978 35 541. 8 Published in J.Phys. Chem. 1987 91 5365 to 5515. See in particular the opening review by R. B. Bernstein D. R. Herschbach and R. D. Levine ibid 5365. 9 C. H. Greene and R. N. Zare J. Chem. Phys. 1983 78 6742. 10 R. N. Dixon J. Chem. Phys. 1986 85 1866. 11 B. Katz J. Park S. Satyapal S. Tasaki A Chattopadhyay W. Yi and R. Bersohn Faraday Discuss. Chem. Soc. 91 73. 12 A. J. Orr-Ewing and R. N. Zare in Chemical dynamics and kinetics of small free radicals eds. K. Liu and A. L. Wagner World Scientific Singapore 1995 Part II p. 936; A. J. Orr-Ewing and R. N. Zare Annu. Rev. Phys. Chem. 1994 45 315; A. J. Orr-Ewing J. Chem. Soc. Faraday Trans. 1996 92 881 and references therein. 13 M. Brouard and J. P. Simons in Chemical dynamics and kinetics of small free radicals eds.K. Liu and A. L. Wagner World Scientific Singapore 1995 Part II p. 936; J. P. Simons J. Chem. Soc. Faraday. Trans. 1997 93 4095 and references therein. 14 F. J. Aoiz M. Brouard P. A. Enriquez and R. Sayos J. Chem. Soc. Faraday Trans. 1993 89 1427. 15 N. E. Shafer A. J. Orr-Ewing W. R. Simpson H. Xu and R. N. Zare Chem. Phys. Lett. 1993 212 155; W. R. Simpson A. J. Orr-Ewing and R. N. Zare ibid. 163. 16 M. Brouard H. M. Lambert C. L. Russell J. Short and J. P. Simons Faraday Discuss. Chem. Soc. 1995 102 179. 17 M. Brouard H. M. Lambert S. P. Rayner and J. P. Simons Mol. Phys. 1996 89 403. 18 M. Brouard I. Burak G. A. J. Markillie K. McGrath and C. Vallance Chem. Phys. Lett. 281 97. 19 N. E. Shafer A. J. Orr-Ewing and R. N.Zare J. Phys. Chem. 1995 99 7591. 20 F. J. Aoiz M. Brouard and P. A. Enriquez J. Chem. Phys. 1996 105 4981. 21 T. P. Rakitzis S. A. Kandel and R. N. Zare J. Chem. Phys. 1997 107 9382; T. P. Rakitzis S. A. Kandel T. Lev-On and R. N. Zare ibid. 9392. 22 F. J. Aoiz M. Brouard V. J. Herrero V. Saez Rabanos and K. Stark Chem. Phys. Lett. 1997 264 487. 23 M. P. de Miranda and D. C. Clary J. Chem. Phys. 1997 106 4509 and references therein . 24 H. L. Kim M. A. Wickramaaratchi X. Zheng and G. E. Hall J. Chem. Phys. 1994 101 2033; R. Fei X. S. Zheng and G. E. Hall J. Phys. Chem. 1997 101 2541. 25 A. J. Alexander M. Brouard S. P. Rayner and J. P. Simons Chem. Phys. 1996 207 215. 26 A. J. Alexander F. J. Aoiz L. Ba�nares M. Brouard J. Short and J.P. Simons J. Phys. Chem. 1997 101 7544. 27 A. J. Orr-Ewing W. R. Simpson T. P. Rakitzis S. A. Kandel and R. N. Zare J. Chem. Phys. 1997 106 5961. 28 F. P. Gilbert G. Maitland A. Watson and K. G. McKendrick J. Chem. Soc. Faraday Trans. 1993 89 1527. 29 F. Green G. Hancock and A. J. Orr-Ewing Faraday Discuss. Chem. Soc. 1991 91 79; D. Summerfield M. L. Costen G. A. D. Ritchie G. Hancock T. W. R. Hancock and A. J. Orr-Ewing J. Chem. Phys. 1997 106 1391; M. L. Costen D.Phil. Thesis University of Oxford 1997. 30 W. R. Simpson T. P. Rakitzis S. A. Kandel A. J. Orr-Ewing and R. N. Zare J. Chem. Phys. 1995 103 7299 and 7313; W. R. Simpson T. P. Rakitzis S. A. Kandel T. Lev-On and R. N. Zare J. Phys. Chem. 1996 100 7938. 31 A. J. Alexander F. J. Aoiz M. Brouard and J. P. Simons Chem. Phys. Lett. 1996 256 561; A. J. Alexander F. J. Aoiz M. Brouard I. Burak Y. Fujimura J. Short and J. P. Simons ibid. 262 589; A. J. Alexander D. A. Blunt M. Brouard J. P. Simons F. J. Aoiz L. Banares Y. Fujimura and M. Tsubouchi Faraday Discuss. Chem. Soc. 1998 108 375. 32 T.-S. Ho T. Hollebeek H. Rabitz L. B. Harding and G. C. Schatz J. Chem. Phys. 1996 105 10472; G. C. Schatz A. Papaioannou L. A. Pederson L. B. Harding T.-S. Ho T. Hollebeek and H. Rabitz ibid. 1997 107 2340; G. C. Schatz L. A. Pederson and P. J. Kuntz Faraday Discuss. Chem. Soc. 1998 108 357. 33 H. Arai S. Kato and S. Koda J. Phys. Chem. 1994 98 12. 34 R. J. Buss P. Casavecchia T. Hirooka S. J. Sibener and Y. T. Lee Chem. Phys. Lett. 1981 82 386. 35 R. D. van Zee and J. C. Stephenson J. Chem. Phys. 1995 102 6946. Received 4th June 1998 Accepted 13th July 1998 415 Chemical Society Reviews 1998 volume

ISSN:0306-0012    

DOI:10.1039/a827405z

出版商:RSC    

年代:1998

数据来源: RSC

摘要:

Metal-directed self-assembly of two- and three-dimensional synthetic receptors Makoto Fujita Coordination Chemistry Laboratories Institute for Molecular Science Myodaiji Okazaki 444-8585 Japan This article reviews recent progress in the study of the transition-metal mediated self-assembly of two- and threedimensional synthetic receptors. Whereas macrocyclization under kinetic control is undoubtedly an unfavorable process the self-assembly strategy offers quite efficient methods for constructing macrocycles under thermodynamic control. In particular cis-protected Pd(II) and Pt(II) blocks are quite effective in obtaining the cyclic framework from simple molecules. Examples disclosed in this article are spontaneously assembled in quantitative yields by just mixing component molecules.This approach is successfully applied to the construction of cage compounds. The selfassembly of nanosized macrocycles and cages is also discussed. 1 Introduction Since the discovery of crown ethers numerous studies on cyclic and cage compounds that are capable of binding atoms (ions) or molecules in their cavities have developed new fields in current chemistry such as host–guest chemistry molecular recognition or supramolecular chemistry. However there seem little examples of obvious practical use for macrocycles and in particular cage compounds despite their remarkable potential utilities. This is partially due to synthetic difficulties encountered in their preparation. In fact the preparation of macrocycles and cages are in general quite low yield processes and require for example high-dilution techniques which are not suitable for their practical syntheses.Whereas macrocyclization under kinetic conditions is an unfavorable process macrocycles are sometimes obtained in high yields under thermodynamic conditions. Typical examples of high yield formation of macrocycles under thermodynamic conditions are provided by phenol-formaldehyde (calixarenes) or resorcinol-aldehyde cyclic oligomers which are formed under equilibrium conditions. The facile formation of such macrocycles suggests that Makoto Fujita is an associate professor of the Institute for Molecular Science (IMS) Okazaki Japan. He received his MS degree from Chiba University in 1982 and PhD from the Tokyo Institute of Technology in 1987.Then he joined the Faculty of Engineering Chiba University in 1988 and moved to IMS in 1997. He is currently a leader of the CREST (Core Research for Evolutional Science and Technology) project of Japan Science and Technology Corporation. NH N Cu2+ +CH2O NH thermodynamic control should be a good strategy for macrocycle (or cage) synthesis. This principle has been realized by recently developed noncovalent syntheses. That is spontaneous generation of welldefined structures has been achieved under thermodynamic conditions by supramolecular self-assembly. In particular coordination chemistry has been successfully used to construct such discrete structures as helices rods macrocycles and cages.1 These structures are spontaneously generated by simply mixing component ligands and metals in solution.Among defined structures macrocycles and cages are particularly interesting because of their potential as synthetic receptors. Exploring the function of self-assembled metal-complexes the present article focuses on recent progress in the metal-directed self-assembly of two- and three-dimensional synthetic receptors. 2 2 Macrocycles incorporating naked metals in their backbones a In the literature several metal–ligand macrocycles have been described. An earlier example was the macrocyclic dinuclear Rh(i) complex 1,3 which was however not designed as a receptor and no binding property was reported for this or related macrocycles.The first self-assembled macrocyclic host 2 containing two Cu(ii) ions was reported by Maverick and Klavetter.4 This inorganic macrocycle showed a strong binding affinity towards 1,4-diazabicyclo[2.2.2]octane (Ka = 220 L mol21) as evidenced by X-ray crystallography. Two point acid– base binding observed in the crystal structure is obviously important because little affinity was shown for monoamines such as pyridine (K = 0.5 L mol21). Self-assembled macrocycle 3 with a hydrophobic binding site was shown to be effective for pyrene transport through a liquid membrane.5 The rate of the transport depended on the metal ion employed and Ni(ii) and Co(ii) showed the highest efficiency for the cooperative binding of aromatic guests. The binding ability of hosts 2 3 and 5 is summarized in Table 1.Macrocyclic dinuclear Pd(ii) complex 4 also has a hydrophobic cavity.6 This complex was prepared by simply mixing PdCl2 and a bisphosphine bridging ligand and fully characterized by FABMS study. The dimerization of a zinc–porphyrin with a pyridine terminus resulted in the formation of macrocyclic complex 5 with inwardly directed hydrogen bonding sites.7 This macrocycle showed a binding affinity towards a terephthalic acid derivative (Ka = ca. 40 L mol21) through efficient hydrogen bond formation. Quite recently Lehn and co-workers reported the selfassembly of circular helices whose frameworks are templated by their counter anions.8 That is the self-assembly of trisbipyridine ligand 6 with FeCl2 salts yields pentanuclear complex 7 which incorporates a chloride ion in the cavity as evidenced by X-ray analysis whereas the same ligand gives hexanuclear complex 8 when treated with Fe(BF4)2 FeSO4 or FeSiF6.Most probably complex 8 carries one counter ion in the void at the center. 417 Chemical Society Reviews 1998 volume 27 P( n-Bu)2 ( n-Bu)2P OC Rh Cl Rh CO Cl P( n-Bu)2 ( n-Bu)2P 1 O O O Cu O N N O O O Cu O 2 O 2 O H N M O NH 2 O O P -O O 2 O N H M O NH 2 O 3 Ph Cl Ph Ph Ph Pd P P Cl Cl Pd P P Ph Ph Cl Ph Ph 4 O N M HN N N O NH N N HN N N N NH M N N O 5 A catalytic property has been reported for self-assembled macrocyclic Cu(i) complex 9 i.e.it catalyzed the hydrolysis of ß-amino acid derivatives by the cooperative effect of two Cu(i) Chemical Society Reviews 1998 volume 27 418 Table 1 Association constants between self-assembled macrocycles and various guests Guest Host 2 Pyridine Pyradine Quinuclidine DABCO 3 (M = Co) Pyrene 5 11a 19 O P O- 1-(Naphthyloxy)acetate Indole Tryptamine d or l-N-Acetyltryptophan N,NA-Dihexylterephthalamide Dimethyl terephthalate Dimethyl isophthalate N-(2-Naphthyl)acetamide 1,3,5-Tri(methoxy)benzene p-Dimethoxybenzene m-Dimethoxybenzene o-Dimethoxybenzene p-Bis(methoxymethyl)benzene 1,4-(Dimethoxy)cyclohexane N-(2-Naphthyl)acetamide 1,3,5-Tri(methoxy)benzene p-Dimethoxybenzene m-Dimethoxybenzene o-Dimethoxybenzene p-Bis(methoxymethyl)benzene p-Dinitrobenzene 1,4-(Dimethoxy)cyclohexane a n.c.not complexed. ions. Interestingly this macrocycle inhibited the hydrolysis of a-amino acid derivatives.9 3 Macrocycles involving cis-protected Pd(ii) or Pt(ii) blocks N N In the studies on inorganic macrocycles discussed above naked metal ions were employed to induce the self-assembly of macrocyclic frameworks. The most frequently employed metals are tetrahedral Cu(i) octahedral Fe(ii) Co(ii) Ni(ii) and square planar Pd(ii). However when such naked ions are employed it is often difficult to control the number and the direction of coordinating organic ligands. If a metal ion is appropriately protected coordination sites are limited and the number and the direction of ligands are easily controlled.Furthermore the protection of the metal gives rise to the design of simple selfassembly from non-sophisticated monodentate ligands (e.g. 4-pyridyl group). Based on such an idea novel cis-protected Pd(ii) and Pt(ii) building blocks 10 were designed. As discussed in the following sections these metal building blocks have shown a remarkable ability for inducing the self-assembly of a variety of macrocycles and catenanes.10 O 3.1 Self-assembly of Pd(ii)- or Pt(ii)-linked square complexes The first example of the self-assembled complex possessing cisprotected Pd(ii) blocks is the tetranuclear square compound 11a (eqn.(1)).11 In the design of complex 11a the combination of the 90 degree coordination angle of the metal with the 180 degree divergence of a linear ligand provides a square framework which has been hitherto unrealized despite its simplicity. The self-assembly takes place without losing the Pd(en) framework because the dissociation of the en ligand Association constant/L mol21 0.5 5 7 220 14 500 6 060 848 60 6 > 1 400 40 < 1 1 800 750 330 580 30 10 n.c.a 15 2 500 2 680 1 560 1 300 560 30 n.c. N N N 6 N N N Fe N N N N N N Fe N N N from Pd(ii) is negligible under ordinary conditions but monodentate ligands undergo rapid dissociation on Pd(ii) ion giving rise to the self-assembly of 11a under thermodynamic equilibration.The procedure for the preparation of 11a is very simple. By N N N N N N Fe N N N N N Cl – N N N Fe N N N 7 N N N Fe N N N N N N Fe N N N just mixing an equimolar amount of 10a and 4,4A-bipyridine in aqueous solution the quantitative self-assembly of 11a was observed which was precipitated in a very pure form by adding ethanol to the reaction solution. The solution structure of 11a 9+ N N Fe N N N N N Fe N N N N N Fe N N N N 8 Chemical Society Reviews 1998 volume 27 12+ N N N Fe N N N N N N Fe N N N 419 Me3N+CH2O was confirmed by NMR and electrospray ionization mass spectrometry (ESI-MS) studies whereas the solid structure was determined by X-ray crystallography (Fig.1).12 In the crystal structure of 11a is found an almost perfect square framework with facial conformation of all pyridine nuclei. The interplanar surface-to-surface separation inside the 9 NH2 M NH 2 ONO ONO2 2 10 + Fig. 1 The X-ray structure of 11a N N N 420 Chemical Society Reviews 1998 volume 27 a M = Pd; b M = Pt NHR' NH Cu2+ N Cu2+ NHR' NH2 M NH 2 NH ONO ONO2 2 10 8+ H2N NH2 a M = Pd; b M = Pt 2 NH N N H N 2 N M NH2 M M N N NH 2 (1) N H N 2 N M H2N (NO3 –)8 11 H2N NH2 Pd X N NH NH 2 2 N N X X 3 N N 2 Pd H N 2 X N N Pd NH Pd N 2 H2N •(NO3)8 12 4 H2N NH2 Pd N N X X (2) N N N H 2 N H2 N H Pd N N X NH Pd 2 H2N •(NO3)6 13 a X = -CºC-; b X = -CH=CH-; c X = -CºC-CºC-; d X = -C6H4- cavity is approximately 8 Å corresponding to the diameter of the cavity of b-cyclodextrin.The square structure was easily expanded by incorporating phenylene or acetylene spacers into Fig. 2 The X-ray structure of 27·(G)4 (G = adamantanecarboxylate ion) the bipyridine framework. However an equilibrium between square 12 and triangle 13 was observed depending on the concentration of the components (eqn. (2)).12 Hong and co- Ph2P PPh2 P PM M N N P M N Ph2 N N N N 2 M N Ph2 N M P Ph P PPh2 Ph2P •(NO 14 (M = Pd Pt) Cl N Pt Cl N N N M N N N N Pt Cl Cl H2N O H2N workers reported recently that this equilibration can be controlled by induced-fit molecular recognition.That is the equilibrium ratio is pushed toward triangle 13b in the presence PPh2 Ph2 Ph2 N N N Cl(OC)3Re 3)8 N N M N N N N M N N 16 H2N O NH2 N N Pt Pt N N OH N N OH HO OH Pt Pt N N NH2 H O 2N 17 N Re(CO)3Cl N N Ph2 P N M Ph2P •(NO3)8 15 (M = Pd Pt) Cl N Pt Cl N N N M N N N Pt Cl N Cl 2 NH O NH2 421 Chemical Society Reviews 1998 volume 27 of a small guest. In contrast a large guest induced the reorganization of square complex 12b.13 The self-assembly of Pt(ii) analog 11b was very slow due to the inactivity of the Pt(ii)–pyridine bond.Thus upon treatment of 10b with 4,4A-bipyridine a kinetically distributed oligomer mixture was initially formed. However the mixture gradually turned into the thermodynamically most stable 11b after the solution was heated for a few weeks at 100 °C as monitored by NMR measurement.14 3.2 Molecular recognition ability of the self-assembled square complexes Due to the cationic structure complex 11a is highly water soluble. In water however it provides a very efficient hydrophobic cavity in which neutral organic molecules are effectively recognized (Table 1).15 In addition to a hydrophobic interaction an electrostatic or charge transfer interaction should be important for the host–guest complexation because electrondeficient host 11a recognized electron-rich guests more efficiently.Organic carboxylates are also bound in the cavity and a significant upfield shift for guest signals (up to Dd = 22.8 for a (2-naphthyl)acetate ring proton) was observed. In the recognition of dimethoxybenzenes the association constants were 30 580 and 330 L mol21 for ortho- meta- and paraisomers respectively. The preferential recognition of meta- and para-isomers are probably due to two-point electrostatic interaction which can not be considered for the ortho-isomer. 3.3 Other molecular squares Following square complex 11 a variety of square compounds have appeared in which transition metals provide the 90° angle at every corner of the square (Fig.2). Stang and Cao modified the structure of 11 into organic soluble phosphine derivative 14.16 Using the phosphine protective group his group has been extending the chemistry of square compounds. For example chiral squares metal-hypervalent iodine hybrid squares nanosized squares functional squares with ferrocene crown ether or porphyrin units have been developed.17 A similar self-assembly strategy was employed by Hupp and co-workers for the preparation of Pd(ii)–Re(i) or Pt(ii)–Re(i) bimetallic square complex 15 that shows a luminescent property.18 Porphyrin square 16 was also reported by Drain and Lehn,19 though its structure was deduced by inference. An “inorganic calix[4]arene” 17 was also prepared by Lippert and co-workers.20 3.4 Di- and tri-nuclear macrocycles Rigid bridging ligands give macrocyclic tetranuclear complexes on complexation with cis-protected metals whereas flexible bridging ligands favor the self-assembly of fewer membered di- or tri-nuclear macrocycles.It has been shown that the selfassembly strategy employing the cis-protected Pd(ii) 10 unit provides a general and highly efficient synthesis of macrocycles. 10 When ring strain exists in the framework to some extent the dinuclear structure becomes in equilibrium with a trinuclear macrocycle (e.g. 18).21 Their structures were confirmed by X-ray crystallographic analyses and/or mass spectrometry. A macrocycle involving the 3-pyridyl group was also prepared.The binding property of these macrocycles is worthy of note. For example dinuclear complex 19 possessing two tetrafluorophenylene units showed a remarkable molecular recognition ability for an electron rich aromatic compound.22 Thus the association constant in aqueous media increased with increasing electron density of the guest compounds (e.g. Ka = 2680 L Chemical Society Reviews 1998 volume 27 422 6+ H2N H2C Pd N NH2 N 2 N N H Pd CH2 N NH 2 N NH2 Pd N H2N H2C (NO3 –)6 18 3)2 mol21 for p-dimethoxybenzene whereas 30 L mol21 for p-dinitrobenzene; Table 1). Quite recently Lippert and co-workers reported the highly efficient assembly of a molecular triangle from (en)Pt(NO and 2,2A-bipyradine.23 3.5 Self-assembly of nanometer-scale macrocycles The construction of nanometer-scale host compounds by selfassembly is quite attractive since they can be synthetic receptors for big molecules such as C60.Although there are few reports on the metal-directed self-assembly of nano-sized macrocycles their structures have been often determined by spectroscopy and inference.24 The first example of a nano-sized macrotricycle which was prepared by metal-directed self-assembly and fully characterized by X-ray analysis was hexanuclear Pd(ii) complex 20 assembling from ten small components the treatment of tripyridyl compound 21 with 10a gave nano-sized macrotricyclic complex 20 in which four ligand molecules were held together by six metal ions.25 The molecular size of 20 is roughly 30 3 23 3 22 Å and the furthest Pd–Pd distance is 19 Å.A proposed pathway leading to 20 involves dinuclear macrocycle 22 possessing two uncoordinated pyridyl groups. Complex 20 showed selective binding properties toward dicarboxylate dianions. For example a significant upfield shift (Dd 20.29) was observed for aromatic protons of benzenediacetate in D2O upon complexation with one equiv. of 20. This complexation is probably due to two-point electrostatic attraction between the negative charge around COO2 groups and the positive charge around Pd(ii) ions. 4+ F F 2 N F F N H2 N H N Pd Pd F F N N NH NH 2 2 (NO3 –)4 F F 19 N H2 N Pd N NH 2 N Another nano-sized macrocycle 23 whose topology is the same as that of 20 also self-assembled in a quantitative yield from 10a and tris(3-pyridyl)-1,3,5-triazine.The structure of 22 was unambiguously determined by X-ray analysis and spectroscopic studies. 4 Metal-directed self-assembly of cage compounds 4.1 Earlier examples Recent progress in the self-assembly of cage compounds shows that three dimensional receptors provide a chemically localized environment.26 The construction of such three-dimensional systems by metal-directed self-assembly requires a more precise design for metal–ligand recognition units. The first major advance in forming a three-dimensional cage complex may be ascribed to Saalfranks “damantanoid” assembly 24 prepared by metallation of ethyl malonate in the presence of MgI2 followed by the addition of oxalyl chloride.The structure NH2 H2N Pd N N NH2 H2N N N N N Pd N N N N N N N Pd N N N N N H2N NH2 N N Pd NH2 H2N 20 N N N N N N 21 NH2 H2N Pd N N NH2 H2N N N N Pd N N N 22 12+ H2 N N Pd N NH 2 •(NO3 –)12 N N N H2N Pd N AcO N N H2N H2 N Pd Pd H2N N N NH 2 AcO N Pd H2N 23 n– 24 EtO EtOCO COOEtOEt O EtO EtOCO O of 24 was evidenced by X-ray crystallography.27 The metal center can be replaced by transition metals such as manganese cobalt nickel or iron.28 The three-dimensional cavity can be expanded by inserting a phenylene space into the ligand framework.29 Analogous M4L6 cage structures have been also reported by Raymond and co-workers.30 A fascinating example of the self-assembled 3D cage is Lehn’s cylindrical complex 25 consisting of six Cu(i) ions and five ligands.31 The X-ray analysis of 25 showed that the quaterpyridine ligands are tilted by 66° with respect to the axis passing through the top and bottom of the cage.When Cu(i) is replaced by Ag(i) the quaterpyridine ligands became almost vertical. 4.2 Self-assembly of nanocages While the 3D cage complexes discussed above possess a small cavity in their framework and thus show no significant binding property (except the recognition of inorganic anion by 25) a nanosized roughly spherical 3D cage with an extraordinarily large cavity was recently constructed by metal-directed selfassembly.32 When coplanar exotridentate ligand 26 was complexed with cis-protected Pd(ii) complex 10a ten small Chemical Society Reviews 1998 volume 27 12+ NH2 N OAc N N H2 N NH2 Pd Pd NH2 N N NH 2 OAc N •(NO3 –)12 NH2 Mg2+ Mn2+ Co2+ Ni2+ ( n = 4) Fe3+ ( n = 0) – – O O O COOEtOEt – – O O O 423 N N N N N N N N N N N N N N N N N N N N N N N N 25 molecules (four ligand and six metals) were found to selfassemble into a highly symmetric M6L4 type adamantanoid complex 27 in a quantitative yield (eqn. (3)). The thermodynamic stability of complex 27 is remarkable. Even if 10a was combined with ligand 26 in a 4:2 molar ratio complex 27 with a 3:2 stoichiometry was formed quantitatively and excess 10b remained intact.The cage has a large spherical central void with a diameter of ca. 11 Å. Within this self-assembling cage four molecules of adamantane carboxylate were encapsulated and this clathrate complex was fully characterized by X-ray crystallography (Fig. 2). The inclusion geometry of the guest in the cage of 27 is very interesting. Hydrophobic adamantyl groups are located inside while hydrophilic carboxylate groups are outside. In addition the main axes of the four guest molecules point to the corners of a tetrahedron. A spectroscopic study based on NMR showed that the same host–guest aggregate was also organized even in solution.It is remarkable that the same 1 4 complexation was also observed with neutral non-substituted adamantane. More interestingly o-carborane a more bulky spherical guest was also encapsulated efficiently in a 1 4 stoichiometry.33 Shape selectivity in the recognition was also observed e.g. cis-stilbene was effectively bound but its trans isomer was not. Since nanocage 27a stands as a result of thermodynamic equilibration this structure is not stable enough to maintain its framework under forcing conditions (e.g. acidic basic or nucleophilic conditions). In contrast Pt(ii) counterpart 27b was shown to be very stable and it does not decompose even in the presence of an acid (HNO3) a base (K2CO3) or a nucleophile (NEt3) due to the inertness of a Pt(ii)–pyridine coordinate bond.The stability of 27b toward acid and base made it possible to design a pH controlled host–guest complexation. Thus N,NAdimethylaniline was effectively encapsulated in cage 27b under neutral conditions but liberated into the aqueous bulk phase by acidification (pH 1). After neutralization with K2CO3 the liberated guest recapsulated in 27b.34 5 Guest-induced organization of synthetic receptors An interesting and unique example of a cage-like complex is compound 29 assembled from three molecules of 10a and two Chemical Society Reviews 1998 volume 27 424 N N N + N N N 26 N N N N N N N N M N M N N N N N N N N M N a M = Pd b M = Pt M 6 Summary M 27 simple pyridine-based ligands 28 (eqn.(4)). The behavior of this complex is quite different from that of other selfassembling receptors because 29 assembles only in the presence of an appropriate guest.35 When 10a and 28 are simply combined an intractable mixture of oligomeric products is obtained. However the addition of an appropriate guest induces the assembly of cage-like complex 29. This guest-induced assembly process was monitored by a time-dependent 1H NMR measurement. NMR titration experiments showed 1 1 host– guest complexation between 29 and various organic carboxylates. For example the 1 1 complex with 4-(methoxy)- phenylacetate ion was isolated in 94% yield. Besides anionic hosts a neutral hydrophobic guest such as p-xylene was also effective for inducing the organization of 29.Significant upfield shifts of guest signals up to ca. 3 ppm were observed when the guests effectively induced the organization of 29. This guestinduced assembly can be regarded as a model for induced fit because a guest molecule induces the organization of its own receptor. This article has demonstrated the effectiveness of the metalmediated self-assembly strategy for constructing two- and 10 12+ N N N M (3) N N N •(NO3–)12 H N 2 M M = NH 2 N N + 10a N 28 6+ N N Pd Pd (4) N NH2 N Pd NH 2 N H2 N 2 NH NH N H2 N 2 •(NO3 –)12 29 three-dimensional receptor frameworks. This strategy provides the facile preparation of synthetic receptors and more significantly enables the construction of nanosized precise structures never before prepared by conventional covalent synthesis.Whereas self-assembly of helicate complexes has been studied extensively during the last decade the area discussed in this article has been less explored and will presumably attract a great deal of attention. A more important aspect in this area is that the self-assembled complexes may exhibit new and unexpected properties particularly owing to the binding abilities of the receptor frameworks and the redox or magnetic properties of the metals. In addition success in the construction of nano-sized receptors leads to multi guest inclusion systems which are essential to the design of an artificial micro space where a substrate and a reagent have an opportunity to be included together and to react with each other.In other words the metal directed self-assembly of synthetic receptors will bring a variety of applications to current chemistry which have never been achieved by covalent bond chemistry. 7 References 1 Templating Self-assembly and Self-organization as vol. 9 of Comprehensive Supramolecular Chemistry eds. J.-P. Sauvage and M. W. Hosseini in particular the following chapters P. N. W. Baxter ch. 5; E. Constable ch. 6; M. Fujita ch. 7. 2 J. M. Canary and B. C. Gibb Progress in Inorganic Chemistry Wiley New York 1997 p. 45 1. 3 A. J. Pryde B. L. Shaw and B. Weeks J. Chem. Soc. Chem. Commun.1973 947. 4 A. W. Maverick and F. E. Klavetter Inorg. Chem. 1984 23 4129. 5 A. W. Schwabacher J. Lee and H. Lei J. Am. Chem. Soc. 1992 114 6 M. Fujita J. Yazaki T. Kuramochi and K. Ogura Bull. Chem. Soc. Jpn. 7 C. A. Hunter and L. D. Sarson Angew. Chem. Int. Ed. Engl. 1994 33 8 B. Hasenknopf J.-M. Lehn N. Boumediene A. Dupont-Gervais 7597. 1993 66 1837. 2313. A. V. Dorsselaer B. Kneisel and D. Fenske J. Am. Chem. Soc. 1997 9 P. Scrimin P. Tecilla U. Tonellato and M. Vignana J. Chem. Soc. 119 10 956. Chem. Commun. 1991 449. M. Fujita Acc. Chem. Res. in press. 5645. and K. Ogura Chem. Commun. 1996 1535. Lett. 1998 39 873. 10 M. Fujita and K. Ogura Bull. Chem. Soc. Jpn. 1996 69 1471; 11 M. Fujita J. Yazaki and K. Ogura J. Am.Chem. Soc. 1990 112 12 M. Fujita O. Sasaki T. Mitsuhashi T. Fujita J. Yazaki K. Yamaguchi 13 S. B. Lee S. Hwang D. S. Chung H. Yun and J.-I. Hong Tetrahedron 14 M. Fujita J. Yazaki and K. Ogura Chem. Lett. 1991 1031. 15 M. Fujita J. Yazaki and K. Ogura Tetrahedron Lett. 1991 32 5589. 16 P. J. Stang and D. H. Cao J. Am. Chem. Soc. 1994 116 4981. 17 P. J. Stang and B. Olenyuk Acc. Chem. Res. 1997 30 502. 18 R. M. Nielson J. T. Hupp and E. I. Yoon J. Am. Chem. Soc. 1995 117 9085. 19 C. M. Drain and J.-M. Lehn J. Chem. Soc. Chem. Commun. 1994 2313. 20 H. Rauter E. C. Hillgeris A. Erxleben and B. Lippert J. Am. Chem. Soc. 1994 116 616. 21 M. Fujita M. Aoyagi and K. Ogura Inorg. Chim. Acta 1996 246 53. 22 M. Fujita S. Nagao M. Iida K. Ogata and K. Ogura J. Am. Chem. Soc. 1993 115 1574. 23 R.-D. Schnebeck L. Randaccio E. Zangrando and B. Lippert Angew. Chem. Int. Ed. Engl. 1998 37 119. 24 J. Manna J. A. Whiteford P.J. Stang D. C. Muddiman and R. D. J. Smith J. Am. Chem. Soc. 1996 118 8731. 25 M. Fujita S-Y. Yu T. Kusukawa H. Funaki K. Ogura and K. Yamaguchi Angew. Chem. Int. Ed. Engl. 1998 37 in the press. 26 D. J. Cram Nature 1992 356 29. 27 R. W. Saalfrank A. Stark K. Peters and H. G. von Schnering Angew. 28 R. W. Saalfrank A. Stark M. Bremer and H.-U. Hummel Angew. 29 R. W. Saalfrank B. H�orner D. Stalke and J. Salbeck Angew. Chem. 30 T. Baisel R. E. Powers and K. N. Raymond Angew. Chem. Int. Ed. 31 P. Baxter J.-M. Lehn and A. DeCian Angew. Chem. Int. Ed. Engl. 32 M. Fujita D. Oguro M. Miyazawa H. Oka K. Yamaguchi and 33 T. Kusukawa and M. Fujita Angew. Chem. Int. Ed. Engl. in the 34 F. Ibukuro T. Kusukawa and M. Fujita J. Am. Chem. Soc. 1998 120 35 M. Fujita S. Nagao and K. Ogura J. Am. Chem. Soc. 1995 117 Chem. Int. Ed. Engl. 1988 27 851. Chem. Int. Ed. Engl. 1990 29 311. Int. Ed. Engl. 1993 32 1179. Engl. 1996 35 1084. 1993 32 69. K. Ogura Nature 1995 378 469. press. 8561. 1649. Received 24th March 1998 Accepted 23rd June 1998 425 Chemical Society Reviews 1998 volume

ISSN:0306-0012    

DOI:10.1039/a827417z

出版商:RSC    

年代:1998

数据来源: RSC

摘要:

Variations on a theme—recent developments on the mechanism of the Heck reaction and their implications for Pd(0) R (1) base EWG synthesis Geoffrey T. Crisp Department of Chemistry The University of Adelaide Adelaide South Australia Australia 5005 The Heck reaction has been used extensively over the past 30 years for the elaboration of alkenes. Although many examples have been published in the literature it is only in the last few years that detailed mechanistic studies have been undertaken and a clearer understanding of the active species has emerged. This review will examine the implications of recent details on the mechanism of the Heck reaction using both traditional and non-traditional catalytic systems highlight mechanistic details which must be updated in textbooks and summarise the protocols which a practising chemist might use to perform a successful Heck coupling.1 Introduction The arylation and alkenylation of alkenes under the influence of a palladium catalyst commonly referred to as the Heck reaction has been extensively exploited by synthetic chemists since its debut in the late 1960’s.1 A traditional Heck coupling was based on an aryl iodide or bromide as the electrophilic partner and a terminal alkene as the nucleophilic partner [eqn. (1) R = aryl vinyl and X = I Br]. The initial promise of a convenient method for carbon–carbon bond formation was only realised satisfactorily for terminal alkenes possessing an electron withdrawing group (for example EWG = CO2R CN Ph).+ RX EWG The regiochemistry found for such transformations was consistent with carbon–carbon bond formation at the least Geoffrey Crisp was born in Sydney Australia in 1955 and obtained his BSc from the University of Queensland in 1978 and his PhD from the Australian National University (with Professor Martin Bennett) in 1981. He was an Alexander von Humboldt Fellow at the Max-Planck-Institute f�ur Kohlenforschung in Germany (with Dr P. W. Jolly 1981–83) a Postdoctoral Associate with the late Professor J. K. Stille at Colorado State University (1983–84) and returned to Australia as a Postdoctoral Associate with Dr S. B. Wild (1984–85). He joined the academic staff of the University of Melbourne as a lecturer from 1985–88 before moving to the University of Adelaide where he is currently a Senior Lecturer and Deputy Head.His research interests revolve around the use of metal catalysis for the synthesis of biologically active molecules and for the construction of nanoframeworks. PdL2 O O Pd NHR O R H eclipsed boatO O hindered terminus of the alkene much like in a Michael addition.2 The stereochemistry of addition (discussed in more detail below) usually delivered a trans disubstituted alkene. The formation of regioisomers arose as a major problem for electronically neutral alkenes or those substituted with an electron donating group (such as OR or NR2).2 After considerable experimentation with a variety of ligands palladium sources solvents and additives many of the disadvantages associated with the traditional Heck conditions have been alleviated and we now have a convenient methodology for the construction of complex multifunctional molecules.The key to using the Heck reaction as a crucial step in synthetic endeavours is to identify the class of Heck reaction in terms of both the type of alkene (whether electron donating or electron withdrawing) and the electrophile (whether a halide or a trifluoromethanesulfonate is the leaving group) and then select the most appropriate conditions in order to maximise the conversion. Various aspects of the Heck reaction have been reviewed in recent years including an extensive compilation of recent applications to the synthesis of complex natural products and cascade processes,2 the use of the intramolecular Heck reaction for the construction of quaternary centres,3,4 the intramolecular asymmetric Heck reaction5 and a detailed summary of mechanistic details for the Heck reaction under a variety of conditions.6 In view of this plethora of excellent reviews why attempt another catalogue of past achievements in the area? This review will not be encyclopaedic but concentrate on the implications of recently published details on the mechanism of the Heck reaction using both traditional and non-traditional catalytic systems highlight mechanistic details which must be updated in textbooks and summarise the protocols which a practising chemist might use to perform a successful Heck coupling.2 Traditional mechanism proposed for the Heck reaction We begin this excursion by reviewing the traditional mechanism for a Heck transformation namely an aryl or vinyl halide (RX where X = Br I) coupling with an acrylate (CH2NCHEWG) under the influence of either Pd(OAc)2 with added L or PdCl2L2 or PdL4 [where L is a tertiary phosphine ligand such as PPh3 or P(o-tolyl)3] as summarised in Scheme 1.Although the general features of this mechanism are still retained in recent proposals the exact nature of the catalytically active species and the influence of associated ligands have been considerably modified. The traditional mechanism proposes that irrespective of the nature of the palladium precursor the active catalytic unit is the coordinately unsaturated 14 electron species PdL2.2 This would seem reasonable since PdL2 is electron rich and nucleophilic in character and has vacant sites so that the organic electrophile RX can undergo oxidative addition to give the known trans- RPdXL2 (1) intermediate in which the R group (aryl or vinyl) is s-bonded to the Pd(ii).The trans isomer (1) is formed by an isomerization of the thermodynamically less stable cis isomer. A vacant site must now be created to accommodate and activate 427 Chemical Society Reviews 1998 volume 27 base•HX base PdL2 RX H L Pd L X EWG R L Pd R L X L X 1 Pd R L R L – L EWG Pd EWG H + L X H EWG H EWG EWG H H H H H H Pd R Pd R EWG H EWG H R H R H H Pd 2 Scheme 1 the alkene and for monodentate L this is usually assumed to be the phosphine so that a neutral RPdXL(CH2NCHEWG) (2) species is formed.The coordinated alkene then undergoes syn addition to form an unstable s-bonded complex which will rotate around the carbon–carbon bond so that the palladium and b-hydrogen are syn coplanar and b-hydride elimination takes place to generate the observed trans substituted alkene and the catalytically inactive HPdXL2 as outlined in Scheme 2. The Scheme 2 base which must be present to enact a successful catalytic cycle reduces the HPdXL2 to regenerate PdL2 and the whole cycle repeats (Scheme 1). This mechanism served adequately for many years and rationalised the regio- and stereochemistry of the substituted alkene product and the need for stabilising ligands around the palladium.This mechanism does not rationalise many of the ligand and solvent effects which have been reported over the past decade nor does it explain the marked accelerating effects observed for certain additives. It is worth detailing some of these recent insights into the various steps of the Heck reaction and how these have lead to new transformations some of which were impossible with traditional Heck conditions. 3 Recent mechanistic details for the Heck reaction 3)2(OAc)2.7 This Pd(ii) substrate spontaneously forms n(OAc)]2 (n = 2 3) as the reduced partner in an overall 3.1 Pd(OAc)2 and PPh3 A common catalyst combination for the traditional Heck reaction is Pd(OAc)2 and 2–4 equivalents of PPh3.2 In the presence of the alkene an oxidation–reduction sequence was proposed which eventually delivered the proposed catalytically active species Pd(PPh3)2.1 Recent examination of the stoichiometric reaction of polymeric Pd(OAc)2 with PPh3 has shown that an immediate transformation takes place in polar solvents such as dimethylformamide (DMF) with formation of Pd(PPh [PdL redox process.Complex [Pd(PPh3)2(OAc)]2 is formally coordinately unsaturated and highly nucleophilic two of the characteristi required for a rapid oxidative addition of RX. [Pd(PPh3)2(OAc)]2 has been shown to undergo a rapid reaction with iodobenzene to give the oxidative addition product Chemical Society Reviews 1998 volume 27 428 PhPd(PPh3)2(OAc) and not the expected (and often proposed) PhPd(PPh3)2I (Scheme 3).7 2PPh3 Pd(OAc)2(PPh3)2 Pd(OAc)2 –PPh3 PhI [Pd(OAc)(PPh3)]– 2PPh3 [Pd(OAc)(PPh [Pd(OAc)(PPh PhPd(OAc)(PPh 3)2]– 3)2 3)3]– Scheme 3 Electrochemical studies indicate that anionic palladium complexes are more readily oxidised than the corresponding neutrals but that the rate of oxidative addition of iodobenzene is the same for both.Pentacoordinate anionic Pd(ii) species have been proposed following the oxidative addition of aryl halides to [PdX(PPh3)2]2 (Scheme 4).8 The addition of excess acetate L L S I I + Ph Pd Ph Pd PhI [PdL2Cl] S Cl L L S = solvent L I Ph Pd L Scheme 4 anion (AcO2) to PhPdI(PPh3)2 also results in an equilibrium with formation of PhPd(PPh3)2(OAc) probably via a dissociative mechanism involving iodide displacement.The affinity of Pd(ii) for the anion X has been found to be Cl2 > Br2 > I2 > AcO2.7 Since many of the catalyst precursors for Heck reactions involve adding 4 or more equivalents of L to assist in the stabilisation of the Pd(0) it is of interest to note that the addition of excess L (over the 3 equivalents required to generate [Pd(PPh3)2(OAc)]2) retards the rate of oxidative addition of RX. These results would suggest that anionic complexes should favour the oxidative addition of RX to Pd(0). The carbon–carbon bond forming step of the Heck reaction was studied using styrene which reacted with PhPd(PPh3)2(OAc) at room temperature in DMF to afford stilbene whereas PhPdI(PPh3)2 did not react under these same conditions (Scheme 5).7 When acetate anion was added to a AgBF Ph 4 RPdI(PPh [RPd(PPh3)2] BF4 no reaction 3)2 rt AcO rt slow Ph RPd(OAc)(PPh3)2 R Ph + AcO [RPd(PPh3)2] rt fast Ph R Ph Scheme 5 mixture of PhPdI(PPh3)2 and styrene the formation of stilbene was observed at room temperature.These observations are consistent with the dissociation of AcO2 cationic complex [PhPd(PPh from PhPd(PPh3)2(OAc) to form an equilibrium mixture containing the cationic complex [PhPd(PPh3)2]+. In contrast the preformed 3)2]+ BF42 reacted only slowly with styrene at room temperature to afford stilbene. What can we infer from these observations? The counteranion would appear to play a crucial role in the electrophilicity of the Pd(ii) species and it is likely that the formation of tight ion pairs with the palladium retards the coordination and insertion of the alkene.These results do confirm that PdL2 is not a major player in the oxidative addition process when anionic ligands such as halide or acetate are present. These stoichiometric studies have all been performed on the very reactive electrophile iodobenzene and the electron deficient alkene styrene. Aryl bromides are much less reactive than aryl iodides towards oxidative addition and require higher temperatures for the Heck reaction. When the catalyst system Pd(OAc)2–PPh3 is used in high temperature Heck reactions (usually above 100 °C) then the phenyl group is cleaved from the phosphine ligand at the arylpalladium(ii) stage.9 This causes the palladium catalyst to decompose prematurely and the phenyl group to be coupled to the alkene resulting in product contamination.Even under rather mild conditions such as 60 °C in chloroform preformed ArPd(PPh3)2Cl complexes have been reported to undergo aryl exchange and thus product contamination is a significant obstacle to the use of PPh3 with less reactive electrophiles.10 These high temperature conditions are necessary for aryl bromides containing electron donating groups and for unreactive aryl chlorides because of the slow rate of oxidative addition at temperatures below 100 °C and so the necessity for more stable catalysts prompted investigations of alternative ligands.3.2 Pd(OAc)2 and P(o-tolyl)3 In order to prolong the life of the palladium complex it has long been known that P(o-tolyl)3 could be used the assumption being that the bulky phosphine formed a stable PdL2 species and that quaternisation of the phosphorus by the aryl halide (a problem because of the high temperatures) could be minimised. Recent studies of the reaction of Pd(OAc)2 and P(o-tolyl)3 have lead to a significant modification to the mechanism proposed in standard textbooks. The addition of slightly more than 1 equivalent of P(o-tolyl)3 to Pd(OAc)2 results in the formation of the dimeric palladacycle 3 which in solution is in equilibrium with solvent bound monomers such as 4 (Scheme 6).11 These palladacycles are extremely thermally stable showing no decomposition up to 250 °C.What is intriguing about these complexes is that they are Pd(ii) species and do not appear to undergo an obvious reduction to Pd(0) during the Heck reaction. A Pd(ii) to Pd(iv) sequence cannot be ruled out at this stage nor can the existence of a short lived highly reactive Pd(0). Palladium(iv) species have been isolated and shown to be involved in the oxidative addition of particularly reactive alkyl halides (such as CH3I) to Pd(ii) complexes containing nitrogen chelating ligands and alkyl or aryl groups (7 and 8).12 The Pd(ii) to Pd(iv) sequence is normally only invoked when a Pd(0) to Pd(ii) conversion is clearly not available or when the Pd(ii) species cannot readily undergo b-hydride elimination or reductive elimination.12 For the Heck reaction involving monomer 4 palladacycles no oxidative addition of the aryl halide occurs in the absence of the alkene and then only at elevated temperatures.Catalysts such as 4 are capable of very high turnovers ( > 100 000 at 0.001% catalyst) for the coupling of butyl acrylate with 4-bromoacetophenone and in the presence of n-Bu4NBr turnover numbers approaching 1 000 000 are possible.11 Excess halide replaces the acetate group of 3 or 4 and forms 5 in the reaction mixture with the bridging halide 6 being isolated at the completion of the coupling. These catalysts which appear to only operate at high temperatures suggest that a number of alternative pathways are available for the Heck coupling and that a detailed unified mechanism is probably not operating under all conditions.A general mechanistic description of oxidative addition alkene insertion and b-hydride elimination will then be the common features for all Heck couplings with the oxidation state and charge on the palladium being dependent on the exact reaction conditions and ligands. What are the advantages of using these heavy duty catalysts? Clearly they can be used for substrates which are reluctant to couple at room temperature because of a high activation energy for the oxidative addition such as electron rich aryl bromides or aryl chlorides. The catalysts or halide derivatives such as 6 can be isolated from the reaction mixtures and recycled with little + P Pd(OAc)2 H3C R R R P Pd 4 X H3C R = X = Cl Br N CH3 CH3 Pd N CH3 I 7 Scheme 6 Chemical Society Reviews 1998 volume 27 5 H3C 3 CH3 R O O P Pd Pd P O O R R CH3 3 O CH3 O R R X P Pd X R R X P Pd Pd P X R R 6 N Pd N I 8 429 loss in activity.Only one equivalent of phosphine is required to generate the catalyst so that the retarding effect of excess phosphine seen in many studies can be avoided. These catalysts will have applications in industrial based processes and for thermally stable substrates. Catalyst 3 can induce a coupling between 1,1-disubstituted alkenes and aryl bromides in the presence of amine bases to give the internal alkene 9 selectively whereas in the presence of bases such as NaOAc and Na2CO3 a mixture of regioisomers (9 and 10) will be formed with the terminal alkene (10) as the major product (Scheme 7).13 Interestingly these authors have CH3 + Br Y Z Y = 4-F 4-OCH3 4-Cl Z = CO2Bu Ph Y Y + Z Z H3C 10 9 Me Pri 2 N P Pri 2 N I Me Pd TFA Pd Me I N P Pri 2 N Me Pri 2 12 11 3 Scheme 7 proposed a reduction step whereby the palladacycle 3 is reduced to an unknown Pd(0) species and the rest of the coupling follows a traditional Heck mechanism (as in Scheme 1).The search for alternative ligands which would not be cleaved at high temperatures prompted the development of N-heterocyclic carbenes as ligands.14 The Pd(ii) catalyst precursor 11 represents an example of non-chelating cis ligands with an X-ray crystal structure of 11 confirming the cis stereochemistry and the fact that the carbene ligands are rotated out of the plane (about 70°) formed by the palladium and associated iodide ligands.Carbene ligands will be good s- and p-donors but poor p-acceptors in contrast to phosphines which are good s-donors and good p-acceptors. These Pd(ii) complexes show an initial induction period (presumably a reduction step) which can be eliminated by the addition of a reducing agent (such as hydrazine or formate). A particularly active catalyst which shows no induction period is formed from Pd(dba)2 and 2 equivalents of 1,3-dimethyldihydroimidazol- 2-ylidene.These carbene-containing catalysts are not active below 80 °C and so are most appropriate for non activated aryl bromides and aryl chlorides or substrates which are thermally stable.14 An alternative cyclometallated complex also displaying extraordinary stability at high temperature is derived from 12.15 This catalyst is not sensitive to air or moisture and Heck reactions can be conducted in air with no noticeable decomposition and the catalyst can even be recovered unchanged (apart from the trifluoroacetate TFA being replaced with the halide from RX) after 300 h at 140 °C. Since a cyclometallated palladium catalyst can be recovered from the reaction mixture it seems unlikely that Pd(0) intermediates are involved and a cycle Chemical Society Reviews 1998 volume 27 430 involving Pd(ii) to Pd(iv) interconversions appears to be more likely.Although the precise identities for the proposed Pd(ii) and Pd(iv) intermediates have not been experimentally established a mechanistic rationale involving them has been proposed by Shaw.16 Since electron rich Pd(ii) intermediates would be necessary in order to encourage oxidative addition the conversion of a coordinated alkene into a s-bonded alkyl species would increase the electron density around the Pd(ii) and also explain the observation that catalysts such as 3 are inactive in the absence of alkene.16 Alkenes are more susceptible to nucleophilic attack when coordinated to a Pd(ii) complex and there are a number of nucleophiles (Y = AcO2 X2 HO2 RNH2 or R2NH Scheme 8) available under the typical EWG X P C P EWG Pd Pd Pd C C P X X 3 or 6 + Y Y Y R P P RBr EWG EWG Pd Pd C C X X Br – Y R EWG P EWG 6 + R Pd C X Br Y = AcO– X– HO– RNH2 or R2NH Scheme 8 conditions of a Heck reaction.Since catalysts involving Pd(ii) to Pd(iv) interconversions operate at high temperatures the addition of the nucleophile to the coordinated alkene is likely to be reversible. Shaw has also proposed that under the mild phase transfer conditions of Jeffery (discussed in detail below)17 a reversible nucleophilic addition can also occur and so generate chelating dialkyl ligands which would facilitate oxidative addition to Pd(ii). 3.3 Mild reaction conditions 3.3.1 PPh3 ligands Much of the recent effort directed towards the Heck reaction has centred on developing protocols which will allow coupling at temperatures below 60 °C with non activated alkenes.In order to achieve this the first requirement is a rapid oxidative addition of RX and the obvious choice would be an aryl or vinyl iodide which are known to oxidatively add to Pd(0) complexes at room temperature or below.6 In the presence of halide anions whether from the electrophile RX or from added n-Bu4NX anionic Pd(0) complexes are usually involved.8 Such anionic species should be more nucleophilic than the corresponding neutral species and so react more rapidly with the electrophile RX. the active Pd(0) species is [PdL An investigation of the reduction of PdCl2L2 catalyst (L = PPh3) precursors has shown that in the presence of Cl2 2Cl]2 (13 in Scheme 9).18 This [PdL2Cl2]2 X 15 Pd L L – Cl X L Pd X Ar [PdL2Cl] + Ar L 13 Cl L Ar 17 X Pd Ar L EWG [PdL2Cl]2 2 EWG 18 14 16 Scheme 9 anion is in equilibrium with the dimeric anion [PdL2Cl]2 22 (14) and in the presence of excess Cl2 also with [PdL2Cl2]22 (15).Aryl iodides (ArI) then undergo an oxidative addition to [PdL2Cl]2 to generate a five coordinate anionic Pd(ii) complex (16) which can eventually lead to ArPdL2I (17) (the often proposed and sometimes observed product of stoichiometric reactions) this latter process being significantly accelerated by added Na+ ions which presumably form a tight Na+Cl2 ion pair and so assist in the removal of chloride ions.The formation of ArPdL2I is not essential for the overall Heck cycle since an alkene can displace Cl2 from 16 and form 18.17 An informative series of experiments was reported by Jeffery17 on the use of phase transfer conditions for performing Heck reactions at 50–60 °C with Pd(OAc)2 and PPh3. In particular the effects of added halide PPh3 and water showed that specific combinations of these additives can be beneficial for the coupling while other combinations can be highly detrimental depending on the particular base present. Using a standard Heck reaction involving reactive substrates namely iodobenzene and methyl acrylate Pd(OAc)2 and 2 equivalents of PPh3 at 60 °C anhydrous conditions were found to be most effective when n-Bu4NCl or n-Bu4NHSO4 was used in CH3CN or DMF if either NaHCO3 or KHCO3 were the base.Anhydrous conditions were also necessary when using NaOAc KOAc or 2CO3 and n-Bu4NX were present. Under these 3 was unnecessary. This dramatic 4NHSO4 and KCl. Jeffery proposed that 4NX resulted from an efficient n-Bu4NOAc as the base either with or without PPh3 but in this case n-Bu4NCl or n-Bu4NBr were equally effective in promoting the coupling but n-Bu4NHSO4 was not. In stark contrast the addition of water was clearly beneficial when the base was Na2CO3 or K2CO3. In fact reactions could be performed in water without any organic solvent at room temperature provided both K conditions the addition of PPh accelerating effect of the base and added n-Bu4NX depended on the presence of the tetraalkylammonium portion of the salt since addition of LiX or NaX had a negligible effect (probably because of the formation of tight ion pairs).For reactions performed in the absence of phosphine n-Bu4NCl was consistently an efficient promoter of the coupling as was a combination of n-Bu the accelerating effect of n-Bu base catalysed reduction of Pd(PPh3)2HX to regenerate the Pd(0) catalyst. It is also plausible that the added halide results in the formation of anionic Pd(0) complexes and five-coordinate Pd(ii) intermediates which would also be expected to increase the rate of the oxidative addition and alkene insertion respectively.18 The accelerating effect of water was rationalised in terms of its influence on the concentration of halide or base in solution and in changing the solid–liquid phase (for organic solvents under anhydrous conditions) to liquid–liquid phase reaction conditions.Water has been reported to be a beneficial co-solvent for the Heck couplings of purine and pyrimidine iodides with 2,3-dihydrofuran with NaOAc or NaHCO3 in combination with Et3N.19 These results highlight the complex interactions which occur in the coordination sphere of the palladium during Heck reactions. It is clear that halides can have an accelerating affect on the Heck coupling and that reactions are possible under mild conditions (even room temperature) for suitably activated substrates. So far we have discussed the oxidative addition of aryl or vinyl halides to Pd(0).Many of the recent examples of Heck reactions have involved the use of aryl or vinyl triflates (trifluoromethanesulfonates OTf). The major difference with this substrate is the greater lability of the Pd–OTf bond compared to the Pd–X bond (X = Br I) and its propensity to be readily displaced by other ligands including solvents. Oxidative addition of Ar–OTf to Pd(PPh3)4 occurs readily at room temperature.20 Electron withdrawing groups on the Ar–OTf substrate increase the rate of oxidative addition consistent with a nucleophilic attack of the palladium on the electrophilic triflate. In the absence of added halide it is assumed that Pd(PPh3)2 reacts with Ar–OTf to give cationic palladium(ii) species.In polar solvents separated ion pairs are formed (19) while in non polar solvents tight ion pairs are formed whereas in the presence of added n-Bu4NCl the square planar chloride complex 20 is formed (Scheme 10). The rate of oxidative L + TfO Pd Ar L 19 Ar L L Pd PdL2 + Ar OTf Ar OTf L L Pd Cl Cl 20 Scheme 10 addition of aryl triflates to Pd(0) increases in the presence of chloride ions (probably by formation of [PdL2Cl]2 as described previously). The order of the oxidative addition of aryl electrophiles RX to Pd(0) is I > OTf ~ Br > > Cl.20 3.3.2 Solvent ligands The combination of a vinyl triflate with phosphine and halide free phase transfer conditions proved to be an efficient procedure for the Heck couplings of non activated alkenes under mild conditions (50 °C) as depicted in Scheme 11.21 OTf OAc + NHCbz Pd(OAc)2 Bu4NOTf K2CO3 OAc NHCbz Scheme 11 Water was found to be necessary only when n-BuN4Cl was used and made no difference to the yield or rate of reaction when n-Bu4NOTf was used as the additive.In addition only a catalytic quantity of n-Bu4NOTf was required for an efficient coupling. Although water may induce a change from the solid– liquid phase to a liquid–liquid phase with a more rapid regeneration of the Pd(0) catalyst as proposed by Jeffery,17 water could also accelerate the reaction rate by the formation of hydroxide anions which could act as a ligand for the palladium 431 Chemical Society Reviews 1998 volume 27 (21 Scheme 12).In the absence of water but in the presence of Cl2 the catalytically active palladium species is ligated to chloride ions whereas in water chloride ligands could be displaced by hydroxide ions giving a weaker and kinetically more labile ligand and a faster Heck coupling. Halide ions appear to inhibit the Heck coupling of vinyl halides with alkenes lacking an activating electron withdrawing substituent whereas the coupling of acrylates proceed satisfactorily in the presence of halide ions. The coordination of a chloride ion to a Pd(ii) complex inhibits the alkene insertion of an electron-rich alkene because of the repulsive interactions between the filled molecular orbital of the alkene and the filled 5s-orbital of the neutral Pd(ii) complex.22 The higher reactivity observed when using vinyl triflates and tetrabutylammonium triflate is then attributed to the high rate of dissociation of the triflate anion from the neutral Pd(ii) intermediate and formation of a very reactive cationic palladium species (22).3.3.3 Chelating phosphine ligands Chelating phosphines are considered to be inappropriate for the Heck coupling of aryl halides with alkenes containing an electron withdrawing group but this situation is reversed for the couplings of electron rich alkenes with aryl and vinyl triflates [eqn. (2)].6 L L L R R R Y (2) Pd Pd Pd L L L OTf Y The key to the success of triflate electrophiles in Heck couplings rests with the formation of cationic Pd(ii) intermediates which will bind electron-rich alkenes (poor p-acceptors but good s-donors) strongly.On the other hand neutral Pd(ii) complexes will bind electron deficient alkenes preferentially as these alkenes are better p-acceptors but poorer s-donors. It is possible to switch from cationic to neutral Pd(ii) manifolds by the addition of a soluble halide source to reactions involving aryl or vinyl triflates and to switch from neutral to cationic Pd(ii) manifolds by the addition of a halide sequestering agent (such as Ag2CO3 or AgOTf) to reactions involving aryl or vinyl halides. The regioselectivity of the Heck reaction for both electron rich and electron deficient alkenes is now better rationalised with an understanding of both neutral and cationic Pd(ii) pathways.6 Under neutral conditions alkenes containing an electron donating group attached directly to the alkene carbon will give mixtures of regioisomers largely determined by OTf R'OTf Oxidative Addition R Pd(OAc) PdLX 2 L Pd R' Cl base induced reduction slow for electron-rich alkenes R b-Hydride Elimination Pd L R Pd Cl X H X = AcO– Cl– HO– TfO– L = solvent Scheme 12 Chemical Society Reviews 1998 volume 27 432 steric factors the key carbon–carbon bond forming step occurring at the least hindered carbon (Fig.1).6 Intermediates which are cationic will favour carbon–carbon bond formation at the least electron-rich terminus of the alkene or the terminus which can stabilise a developing positive charge (Fig.1).6 Much of the recent mechanistic work on the Heck reaction has been concerned with the intermolecular asymmetric Heck arylations of 2,3-dihydrofurans. Numerous reports have consistently shown that the Heck coupling of phenyl triflate with 2,3-dihydrofuran in the presence of (R)-BINAP (Fig. 2) and Pd(OAc)2 will give (R)-2-aryl-2,3-dihydrofuran (23) in > 95% enantiomeric excess (ee) whereas the use of aryl iodides under these conditions will give only racemic products (Scheme 13).23 The enantiomeric purity of (R)-23 can be enhanced by a kinetic resolution involving the selective elimination of (S)-23 through formation of the minor isomeric product (S)-24. Molecular models show that (R)-23 is derived from the cationic intermediate 25 while (S)-24 is derived from the cationic intermediate 26 (Scheme 14).These two diastereomers result from the 2,3-dihydrofuran presenting a different face for coordination to the palladium. The enantiomeric purity of (R)-23 and the yield of (S)-24 are both dependent on the strength of the base used to reduce the Pd(ii) hydride with the highly basic 1,8-bis(dimethylamino)naphthalene giving the highest ee for (R)-23 and the highest proportion of (S)-24. The rationalisation for these observations is that 25 undergoes a rapid b-hydride elimination to give an hydrido alkene 27 which experiences severe steric interactions between the phenyl ring of the dihydrofuran and the phenyl rings of the (R)-BINAP ligand. This unfavourable steric interaction can be alleviated by a rotation of the dihydrofuran ring to give 29 which can now undergo a series of rapid b-hydride additions and eliminations to give (R)-23.For 26 the initial hydrido alkene 28 will not experience these unfavourable steric interactions so that rotation of the dihydrofuran is less likely and 28 can form (S)-24 directly (Scheme 14). These studies have highlighted the ease with which reversible b-hydride additions and eliminations will take place during Heck couplings. In a mechanistic study of this reaction using (S)-BINAP extensive use of 31P NMR spectroscopy has revealed the ease with which alkene insertions can take place and the importance of the base in minimising the lifetime of the hydrido Pd(ii) intermediates.24 Isomer (S)-23 was formed in up to 91% enantiomeric excess along with the minor isomeric product (R)-24,24 consistent with previous reports based on (R)-BINAP.23 Intermediate 30 was characterised by 31P and two dimensional 1H NMR spectroscopy as well as mass spectrometry (Scheme 15).Carbon–carbon bond formation occurred at 230 °C and 30 was stable at this temperature for L R + Pd R' X L R R X L Pd R' L Pd R' OH 22 21 fast for electron-rich alkenes slow for electron-rich alkenes R R R' R' R' Pd L X Pd L OH L several hours.24 In the absence of base the palladium hydride adds to the 2,3-dihydrofuran to give a mixture of diastereomers 31 whereas in the presence of a strong organic base none of this product was observed.The ability of the chelating phosphine to maintain the bidentate coordination is essential to the proposed rationalisation of the enantioselectivity of these Heck reactions. The use of chiral phosphinooxazoline ligands (such as the tert-butyloxazoline in Fig. 2) has further enhanced the ee’s Ar L + Pd L Ar L Pd + L X ( R)-BINAP + PhOTf O Ph P Ph Pd Ph H O 25 O ( R)-23 Ph P Ph H Pd O Ph 26 R R Ar PPh2 PPh2 Pd(OAc)2 ( R)-BINAP Ph P Ph R Ph H Ph P Ph Scheme 13 ( R)-23 Ph Pd H S Scheme 14 28 OBu Ar Ar N OBu Ar Ar Ar Fig. 1 PPh2 phosphinooxazoline Fig. 2 H Ph O Ph Ph P P Ph Ph O N CO2CH3 O Ar CO2CH3 O Ar O CH3 N CH3 H3C + O ( S)-24 Ph Ph P P Ph Ph Pd O H 27 Ph Ph P P Ph Pd H Ph O 29 Ph Ph Ph H O ( S)-24 Ph L L Ph O Pd Pd + O L L solvent solvent 30 Ph O L H O L O ( S)-23 Pd Pd L solvent L solvent 31 Scheme 15 H available from these reactions and offers an alternative product profile compared to BINAP.25 The major product from the reaction of phenyl triflate with 2,3-dihydrofuran in the presence of (R)-BINAP was (R)-23 with (S)-24 being a minor yet persistent byproduct.23 The same reaction performed with (S)-phosphinooxazoline (Fig.2) gave (R)-24 as the sole product (97% ee) and none of the isomeric 2,3-dihydrofuran 23 therefore suggesting that reversible b-hydride elimination and addition is less favourable with this ligand.The catalysts formed from phosphinooxazoline ligands are less reactive than those formed from BINAP with Heck couplings typically performed at 50–80 °C over 3–7 days. The reactivity profile for this ligand suggested that cationic Pd(ii) intermediates were involved as did the observation that chloride ions dramatically reduced the rate of coupling. 2,3-Dihydropyrrole can also be used successfully for asymmetric Heck couplings as can nonactivated alkenes such as cyclopentene and cyclohexene (although with lower ee’s). Ph 3.3.4 Intramolecular Heck couplings The early reports of asymmetric Heck reactions concentrated on intramolecular examples which involved the formation of tertiary centres usually used (R)-BINAP as the chiral ligand and were performed in polar aprotic solvents such as N-methylpyrrolidin-2-one.3,5 When aryl or vinyl halides were used as the electrophile a halide sequestering agent (Ag+) was added to ensure a cationic Pd(ii) intermediate.The use of AgOAc was found to be detrimental to high ee’s since AcO2 would bind strongly to the Pd(ii) and discourage formation of the necessary cation. The most efficacious sequestering agent was Ag3PO4 which afforded high ee’s because of the low nucleophilicity of PO4 32 and consequential low affinity for Pd(ii). If triflates were used instead of halides then a Ag+ source was unnecessary and nonpolar solvents preferred.The selectivity of Ag+ salts in effecting a non-reductive cyclisation was elegantly demonstrated during the synthesis of Pancratistatin as outlined in Scheme 16.26 In the absence of Ag+ salts only the reductive cyclisation product was observed (the inversion of the OBn alpha to the carbonyl was inconsequential to the Heck coupling). The inherent reluctance of tri- and tetrasubstituted alkenes to participate in Heck couplings can be overcome by using an intramolecular reaction rather than a bimolecular one.4 The success of this approach has been demonstrated with the construction of sterically hindered quaternary centres.4 During an intermolecular Heck coupling the alkene can present two alternative faces to the palladium with steric hindrance between the substituents on the alkene and the ligands on the palladium determining which face is preferred.For an intramolecular 433 Chemical Society Reviews 1998 volume 27 OBn O OBn O O O Heck coupling restrictions in the conformation of the backbone linking the alkene to the electrophilic centre will usually determine a preferred orientation of the alkene. During the total synthesis of Amaryllidaceae alkaloids the stereochemistry of the intramolecular Heck coupling was shown to favour an eclipsed orientation of the Pd–C bond and the alkene (Scheme 17).4 Using the catalytic system Pd(OAc)2–PPh3–Ag2CO3 in THF at 66 0C a single cyclised product 32 was formed in 90% yield. The epimeric product 33 was not detected suggesting that the preferred orientation of the tethered alkene is eclipsed (assuming the conformations shown in Scheme 17).PdL2 O O Pd R eclipsed boatO O O O O O 32 It is also possible to form quaternary carbon centres with high enantioselectivity by an intramolecular Heck coupling using 2dba3.CHCl3.4 Using either (R)- or (S)-BINAP and Pd (S)-BINAP and Z alkene 34 an intramolecular Heck coupling Chemical Society Reviews 1998 volume 27 434 O I Pd(OAc)2 PPh3 O O Pd(OAc)2 PPh3–AgNO3 O O O Scheme 16 O O I O H O NHR O PdL2 H RHN NHR O O H O O O H O NHR O Scheme 17 OBn OBn O will form 3,3-disubstituted-2-oxindole 35 in 95% ee and eventually (2)-physostigmine whereas 34 and (R)-BINAP can be used to form ent-(+)-physostigmine (Scheme 18).Analogues of 34 with an E stereochemistry around the alkene also cyclised but with much lower ee’s (50%). The interesting feature of these intramolecular Heck transformations is the high ee’s observed even in the presence of halide. Earlier discussions in this review have highlighted the fact that halide free conditions were necessary to effect high ee’s in intermolecular couplings. MeO OBn OBn O These cyclisations result from an equatorial orientation of the large group and an eclipsed conformation of the Pd–C s-bond and the alkene. For the formation of large rings via an intramolecular Heck coupling an endo selectivity is usually observed because both faces of the alkene are accessible [eqn.3].3 For medium sized rings a mixture of endo and exo cyclisation is often observed [eqn. (4)] while for small rings exo cyclisation is predominant [eqn. (5)]. It is noted that it is possible to design substrates which cyclise contrary to these general observations.3 Pd Br O O R twisted chair O O Br O H NHR 33 A further example of this difference in the reaction pathways for inter- and intramolecular Heck reactions is seen in the greatly enhanced enantioselectivity resulting from the addition of n-Bu4NX to Pd–(R)-BINAP induced cyclisation of triflate 36 (ee of 90–95% Scheme 19).27 In the absence of n-Bu4NX an ee of 43% was observed. For the cyclisation of iodide 37 using this same catalyst an ee of 91–95% was observed in the presence or absence of added halide yet the ee dropped dramatically to 43% Me CHO OTIPS I MeO O O N Me Me N Me 34 35 Me MeNHCO2 N Me N Me (–)-physostigmine Scheme 18 I (3) n > 2 n > 2 endo product (4) endo product exo product (5) exo product OSiR3 Me X OSiR3 O O N N Me Me Me 38 36 X = OTf 37 X = I Scheme 19 for the iodide in the presence of AgOTf.In order to rationalise these observations a five- coordinate neutral Pd(ii) precursor to a cationic Pd(ii) intermediate was proposed and the effect of the added halide was to divert the palladium from a dissociative pathway involving triflate loss (Scheme 20).Ar P P Ar Pd Pd P P X X P Ar P Pd Ar Pd P P X X Scheme 20 By using the poorly coordinating ligand AsPh3 in conjunction with Pd(OAc)2 it is possible to encourage reversible b-hydride eliminations and additions and so form the thermodynamically more stable alkene (39) by an intramolecular cyclisation (Scheme 21).28 Alternatively by using a halide sequestering agent such as TlOAc and PPh3 the expected spiro tetrahydropyridine 40 can be formed with minimum double bond isomerisation. I Pd(OAc)2 AsPh3 N N CHO CHO 39 Pd(OAc)2–PPh3 TlOAc N CHO 40 Scheme 21 A stoichiometric investigation of the intermediates formed during the intramolecular Heck cyclisation of 41 with Pd(dppf) (dppf = diphenylphosphinoferrocene) resulted in the isolation of 42 (Scheme 22).29 The X-ray structure of 42 showed a distorted square planar structure with the aryl group perpendicular to the mean square plane of the palladium and associated ligands and no interaction of the tethered alkene with the metal centre.Complex 42 was stable at room temperature showing no signs of undergoing the Heck cyclisation. Reaction of 42 with AgOTf at 278 °C caused an immediate precipitation of AgI and the only species detected by NMR at 240 °C was 43 the product of insertion and reversible b-hydride elimination and addition.29 This implies that the insertion of the palladium–aryl O Pd(dppf)Cl2 I reducing agent 41 OTf L O Pd L L H O Pd L Scheme 22 bond into the alkene is rapid and that the b-hydride elimination and addition is also rapid at this low temperature.This sequence implies that if the stereochemistry of the intermediates is favourable then rapid Heck coupling can take place at low temperatures. 4 Recent modifications to traditional reaction conditions An effective method for controlling the position of the double bond from an asymmetric intramolecular Heck coupling is the use of a terminal trimethylsilyl group as in 44.30 A catalyst system consisting of Pd(OAc)2 PPh3 KOAc and n-Pr4NBr gave the vinyl product 45 whereas the catalyst system Pd2dba3– PPh3–Ag2O gave predominantly the vinyl silane 46 indicating that the presence of the silane has completely suppressed alkene isomerisation (Scheme 23).NCOCF3 I Me3Si 44 Scheme 23 With the increased interest in using Heck reactions for industrial processes and for the synthesis of pharmaceuticals the loss of palladium catalyst and product contamination with the catalyst have emerged as serious issues. The use of controlledpore glass beads coated with a palladium catalyst derived from Pd(OAc)2 or PdCl2 and a water soluble phosphine ligand (m-NaO3SC6H4)3P has been shown to catalyse the classical Heck reaction between aryl bromides or iodides and acrylates with less than 1 ppm leaching of the palladium.31 A silicasupported poly-g-mercaptopropylsiloxane Pd(0) catalyst prepared from the condensation of fused silica and g-mercaptopropyltriethoxysilane and PdCl2 followed by reduction with hydrazine can be used to prepare functionalised styrenes and acrylic acids in good yield.32 The catalyst can be recovered by simple filtration and reused with only a 3% reduction in activity for the second cycle.Only 0.4% catalyst need be employed for the couplings and the reduced catalyst can Chemical Society Reviews 1998 volume 27 I L O Pd L 42 AgOTf L O Pd L L OTf Pd O L H3C 43 NCOCF3 H R 45 R = H 46 R =SiMe3 435 be exposed to air for up to a week with no apparent reduction in activity. The disadvantage for this particular catalyst is that only aryl iodides and not aryl bromides will couple with the acrylates. 5 Summary of protocols Can we now propose a series of protocols which can be used to assist the practising chemist in performing a successful Heck coupling? There are no certainties in synthesis yet the large body of published data and detailed mechanistic information can be used to establish some general guidelines.For intermolecular couplings involving reactive electrophiles (aryl or vinyl iodides) and alkenes containing an electron withdrawing group a traditional catalyst systems such as Pd(OAc)2 and 2–4 equivalents of L or PdL2Cl2 or PdL4 [L = PPh3 or P(o-tolyl)3] with an organic or inorganic base will usually suffice. Such systems will usually require temperatures in the range 50–100 °C. In order to lower the temperature the most effective protocol is to add R4NX (X = Cl Br) and use an aqueous solvent with K2CO3 as the base.For electrophiles which undergo oxidative addition more slowly (aryl bromides with electron donating groups or aryl chlorides) high temperatures (above 120 °C) are usually required and a ligand which will not decompose is essential for a long lived catalyst (L = PPh3 will not be suitable). For aryl or vinyl triflates with alkenes containing an electron withdrawing group a traditional catalyst system can also be used. For alkenes which do not contain an electron withdrawing group then halide free conditions achieved by either using aryl or vinyl triflates as the electrophile or adding a halide sequestering agent (Ag+) for aryl or vinyl halides will be advantageous. For intramolecular Heck cyclisations the reaction conditions appear to vary depending on whether a tertiary or quaternary centre is being formed on the ring size and the stereochemistry of the alkene.The presence of halides does not appear to impede the cyclisation at elevated temperatures and can be beneficial for high ee’s in asymmetric Heck couplings. The use of halide free conditions can produce rapid Heck couplings but variable ee’s for asymmetric cyclisations. 6 References 1 R. F. Heck J. Am. Chem. Soc. 1968 90 5518. 2 A. de Meijere and F. E. Meyer Angew. Chem. Int. Ed. Engl. 1994 33 2379. Chemical Society Reviews 1998 volume 27 436 3 S. E. Gibson and R. J. Middleton Contemp. Org. Synth. 1996 3 447. 4 L. E. Overman Pure Appl. Chem. 1994 66 1423. 5 M. Shibasaki C. D. J. Boden and A. Kojima Tetrahedron 1997 22 7371. 6 W. Cabri and I. Candiani Acc. Chem. Res. 1995 28 2. 7 C. Amatore E. Carre A. Jutand M. A. M’Barki and G. Meyer Organometallics 1995 14 5605. 8 C. Amatore A. Jutand and A. Saurez J. Am. Chem. Soc. 1993 115 9531. 9 W. A. Herrmann C. Brossmer K. � Ofele C.-P. Reisinger T. H. Priermeier M. Beller and H. Fischer Angew. Chem. Int. Ed. Engl. 1995 34 1844. 10 W. A. Herrmann C. Brossmer T. H. Priermeier and K. � Ofele J. Organomet. Chem. 1994 481 97. 11 W. A. Herrmann C. Brossmer C.-P. Reisinger T. H. Riermeier K. � Ofele and M. Beller Chem. Eur. J. 1997 3 1357. 12 A. J. Canty Acc. Chem. Res. 1992 25 83. 13 M. Beller and T. H. Riermeier Eur. J. Inorg. Chem. 1998 29. 14 W. A. Herrmann M. Elison J. Fischer C. Kocher and G. R. Artus Angew. Chem. Int. Ed. Engl. 1995 34 2371. 15 M. Ohff A. Ohff M. van der Boom and D. Milstein J. Am. Chem. Soc. 1997 119 11 687. 16 B. L. Shaw New J. Chem. 1998 77. 17 T. Jeffery Tetrahedron 1996 52 10 113. 18 C. Amotore A. Jutand and A. Suarez J. Am. Chem. Soc. 1993 115 9531. 19 H.-C. Zhang and G. D. Daves Jr. Organometallics 1993 12 1499. 20 A. Jutand and A. Mosleh Organometallics 1995 14 1810. 21 G. T. Crisp and M. G. Gebauer Tetrahedron 1996 52 12 465. 22 P. E. M. Siegbahn J. Organomet. Chem. 1994 478 83. 23 F. Ozawa A. Kubo Y. Matsumoto T. Hayashi E. Nishioka K. Yanagi and K. Moriguchi Organometallics 1993 12 4188. 24 K. K. Hii T. D. W. Claridge and J. M. Brown Angew. Chem. Int. Ed. Engl. 1997 36 984. 25 O. Loiseleur M. Hayashi N. Schmees and A. Pfaltz Synthesis 1997 1338. 26 G. K. Friestad and B. P. Branchaud Tetrahedron Lett. 1997 38 5933. 27 L. E. Overman and D. J. Poon Angew. Chem. Int. Ed. Engl. 1997 36 518. 28 L. Ripa and A. Hallberg J. Org. Chem. 1997 62 595. 29 J. M. Brown J. J. Perez-Torrente N. W. Alcock and H. J. Clase Organometallics 1995 14 207. 30 L. F. Tietze and R. Schimpf Angew. Chem. Int. Ed. Engl. 1994 33 1089. 31 L. Tonks M. S. Anson K. Hellgardt A. R. Mirza D. F. Thompson and J. M. J. Williams Tetrahedron Lett. 1997 24 4319. 32 M.-Z. Cai C.-S. Song and X. Huang Synthesis 1997 521. Received 28th April 1998 Accepted 2nd Jun

ISSN:0306-0012    

DOI:10.1039/a827427z

出版商:RSC    

年代:1998

数据来源: RSC

摘要:

Synthesis chemistry and conformational properties of piperazic acids Marco A. Ciufolini*,a,b and Ning Xib a Laboratoire de Synth`ese et M�ethodologie Organiques Universit�e Claude Bernard Lyon 1 et � Ecole Superieure de Chimie Physique et Electronique de Lyon 43 Bd. du 11 Novembre 1918 69622 Villeurbanne cedex France. E-mail ciufi@cpe.fr b Department of Chemistry MS 60 Rice University 6100 Main Street Houston TX 77005–1892 USA We review methods for the preparation of piperazic acids (“piz”) and significant aspects of their chemistry. Special attention is devoted to issues relating to acylation of these non-proteinogenic amino acids given the importance of these operations for the synthesis of piz-containing oligopeptides. We also provide an account of the unique conformational properties of piperazic acids and discuss their behavior as rigid proline equivalents that should be quite useful for the creation and study of peptide turn mimics.1 Introduction Piperazic acids (“piz”) are non-proteinogenic cyclic a-hydrazino acids of general structure 1 2. Compound 1 the first 4 HO COOH 3 H 5 COOH NH 6 2 X N 1 1 X = Y = H 2 X Y = p bond NH 3 N Y member of the family to be described in the literature was discovered by Hassall and coworkers as a component of monamycins a group of cyclodepsipeptide natural products.1 In the intervening years piperazic acids and their C-4 oxygenated variants 3 have been detected in several other peptide natural substances,2,3 the number of which continues to grow.4 The majority of these compounds display remarkable biological properties.For instance antrimycins2b are tuberculostatic agents; GE3 and GE3B4a are antitumor antibiotics that inhibit the progression of cell cycle from the G1 to the S phase; L-156,3733 is a potent oxytocin antagonist; L-156,6023 is an antiinflammatory agent that inhibits binding of anaphilatoxin C5a to its receptor; luzopeptins5 and quinoxapeptins5 are strongly active against HIV.6 Appreciable biological activity is Marco Ciufolini (BS chemistry 1978 Spring Hill College; PhD 1981 University of Michigan M. Koreeda; postdoc 1982–1984 Yale University S. J. Danishefsky) started his independent career at Rice in 1984. He won an A. P. Sloan Fellowship in 1994. In January 1998 he assumed his current duties in Lyon France.Ning Xi Marco Ciufolini H N R1 O N H O O H N H 93 found even at the level of piz-containing structures much simpler than those of the above natural products. For instance the antihypertensive agent cilazapril® 4,7 features a unique O COOH HN N EtOOC 4 N bicyclic derivative of 1 as a key subunit while piperazic acid 1 itself appears to interfere with g-aminobutyric acid (“GABA”) uptake.8 Our own interest in 1–3 developed in connection with work directed toward the total synthesis of luzopeptins. These efforts have provided us with much opportunity to explore the preparation chemistry and conformational properties of piperazic acids especially the unsaturated variants 2 and 3 and their N-2 acyl derivatives.Our own work and that of other researchers before us has unveiled a number of interesting properties of these molecules. Aspects of their synthesis chemistry and unusual conformational properties are reviewed herein. 2 The chemistry of piperazic acids 2.1 Synthesis of the piz framework A number of methods for the synthesis of piperazic acids are currently available undoubtedly as a result of their growing importance in natural products chemistry and of their potential in pharmaceutical research. In general the piz framework is assembled by cyclization of a suitable a-hydrazino acid precursor which in turn may be obtained either by N-amination of an a-amino acid derivative or by delivery of a complete hydrazine unit to an appropriate substrate.Ning Xi obtained his BS (1984) and MS (1987) degrees from Peking University joined the Ciufolini group at Rice in 1992 and received his PhD in 1996. After a one-year postdoc also with M. A. Ciufolini he accepted a position at Albany Molecular Research Inc. Albany NY. 437 Chemical Society Reviews 1998 volume 27 A typical route to piperazic acids through de novo construction of the hydrazine unit is exemplified in the conversion of 5 to 7. Thus a classical N-nitrosation of compound 5 followed by nitroso group reduction and in situ acetylation were carried out as a prelude to cyclization to 7 (Scheme 1).9 Hydrazines may be COOiPr NHCbz 5 MsO assembled even more expeditiously thanks to contemporary methodology for N-amination.This is illustrated in a recent synthesis of piperazic acid analogs a central feature of which was the reaction of oxazolone 8 with O-(diphenylphosphinyl) hydroxylamine. The emerging 9 was then cyclized to 10 with aqueous acid (Scheme 2).10 The recently described Collet OMOM H MeO O HN O 8 COOR2 Ph COOR2 HN CONH2 13 MeO 3).11b R1 NH2 11 R1 438 oxaziridines11a seem to hold excellent potential for the creation of the hydrazine unit of an ultimate piperazic acid as apparent from the successful N-amination of various a-amino acids (cf. 11 ? 12). Similar transformations should also be possible through the so-called Shestakov reaction an analog of the Hoffman rearrangement that utilizes ureas as substrates.Improved protocols for the conduct of this reaction with hydantoic acids are now available (cf. 13 ? 14 Scheme Most known routes to piperazic acids however involve delivery of a complete hydrazine subunit to a suitable acceptor. The heterocycloaddition between dienes and azodicarboxylate Chemical Society Reviews 1998 volume 27 9 Aq. HCl H N COOR2 NH COOiPr 1. NaNO2 NCbz 2. Zn HOAc– Ac2O MsO NHAc 6 NaH COOiPr NCbz N Ac 7 Scheme 1 OMOM H O Ph2P-O-NH2 O MeO N NaH MeO H2N O MOMO O N O 10 Scheme 2 O R1 N–BOC BOC–NH 12 COOR2 R1 KOCl HN NH2 14 Scheme 3 esters or cyclic variants thereof constitutes one of the earliest synthetic methods for 1 and 2.12 As shown in Scheme 4 the reaction may be conducted in an enantioselective mode; e.g.by the use of dienes displaying a glucose-derived chiral auxiliary (Scheme 4).13 A different strategy rests on the regioselective OAc COOMe AcO O O AcO NCOOEt OAc H R* O HO MeO O H MeO COOH COOH MeO OMe NH 17 18 H+ COOR O O O 15 H N N COOEt H2N–NH2 H2N–NHBOC 3 noteworthy that the regioselectivity of nucleophilic opening of epoxides bearing an electron-withdrawing substituent at one of the oxiranyl carbons may be weak and it is often in favor of the b-carbon relative to the substituent in question. In the present case however complete a-selectivity was observed.This may be due to conversion of the substrate acid to the corresponding carboxylate salt prior to oxirane cleavage since the presence of a carboxylate anion on the epoxy ring is known to promote good a-selectivity.15 A related approach involves reduction of the imino subunit of a hydrazone obtained from an a-ketoester and an alkyl carbazate (Scheme 6).16 An interesting route to piz COOR NHBOC O O 19 1 + EtOOCN COOMe COOEt Scheme 4 Scheme 5 Scheme 6 1 hydrazinolysis of the oxirane ring of an a,b-epoxy acid prepared in turn from the corresponding allylic alcohol by enantioselective Sharpless methodology and oxidation. This approach was demonstrated in a landmark synthesis of a piperazic acid component of luzopeptins (Scheme 5).14 It is H2N N 20 1.NaBH3CN 2. H+ frameworks through Speckamp-type intramolecular amidoalkylation reaction of N- acylhydrazinium species is shown in Scheme 7. This technique also allows preparation of benzo analogs 24 of piperazic acids.17a COOMe OAc COOMe TMS Et2AlCl N N COO–All N Alloc N Bn Bn 21 COOMe OAc COOMe SnCl4 N N COO–All 22 Alloc N N Bn Bn 23 All = allyl Alloc = allyloxycarbonyl Scheme 7 O O LDA Xc Xc (tBuOOC–N=)2 N–BOC N Br BOC 25 24 Especially versatile avenues to 1–3 have emerged18 with the advent of methods for electrophilic hydrazination oates.19 This chemistry may well represent the most practical avenue to piperazic acids currently available especially when combined with protocols for asymmetric derivatization of enolates; e.g.through the use of Evans chiral auxiliaries. In many cases the azodicarboxylate component is the bis tert-butyl ester (“DBAD”) though other azodicarboxylates may also be employed. A variant of this reaction leads directly to the perhydropyridazine ring system through condensation of azodicarboxylates with enolates of 4-halobutyrate esters (Scheme 8).20 O O N Xc = Ph Scheme 8 O O O OH (text) OR2 R1 R1 27 OH O 26 Luzopeptins and quinoxapeptins incorporate piperazic acids that display a free OH or an acyloxy group at C-4. Very direct asymmetric syntheses of these materials may be accomplished by diastereoselective hydrazination of the dianions of scalemic b-hydroxyesters (Scheme 9).21 Diastereomeric excesses ob- OR2 OR2 R1 BOCHN Scheme 9 28 LDA (tBuOOC–N=)2 N–BOC 29 served in the course of these reactions are good to excellent (94% in our hands).In turn the hydroxyester substrates are readily available by enantioselective reduction of the corresponding b-ketoesters either by Noyori-type hydrogenation or by treatment with fermenting bakers’ yeast.22 It will be seen shortly that the preparation of piperazic acid for an ultimate incorporation into a peptide chain often requires access to terminally blocked a-hydrazino acid intermediates of the type 32. Compounds of this type may be obtained from DBAD adducts by release of both BOC units (TFA) and treatment of the intervening free hydrazino acid with BOC2O.The terminal NH2 of the hydrazine is considerably more reactive than the inner NH group so that product 32 is obtained with excellent regioselectivity. However this procedure is inefficient.23 It is much better to start with adducts 30 of dibenzyl azodicarboxylate. Hydrogenolysis of the benzyl carbamates in the presence of BOC2O leads to 32 in excellent yields.24 Similar procedures have been described as a “transprotection”. 17b It is not clear whether the free hydrazine 31 is an intermediate in these reactions but if so its reaction with BOC2O must be considerably faster than its hydrogenolysis. It should be noted that hydrazines such as 31 are quite sensitive to oxidation.Exposure to the atmosphere causes deazoniation probably through the series of events depicted in Scheme 10. Transprotection bypasses the need to handle sensitive intermediates during blocking group exchange. O O O R1 R1 R1 H2 Pd(C) X X X BOC2O NH N NH COOBn (?) BOCHN BnOOCHN H2N 30 31 32 X = OR2 or Evans auxiliary X X O O2 R1 R1 OH O 31 R1 (?) X N N N H N 34 33 Scheme 10 2.2 Reactivity of piperazic acids Saturated and unsaturated variants of piperazic acids may be readily interconverted when the N-2 position is acylated; therefore any method leading to one type of compound is readily adaptable to the creation of the other. Acidic NaBH3CN easily reduces the imino unit of structures of the type 2,23 while oxidation of molecules of the type 1 may be effected with tertbutyl hypochlorite (Scheme 11).25a It should be noted that controlled oxidations may be difficult to induce if the N-2 position is not acylated in which case the substrate tends to undergo aromatization to a pyridazine.This is shown with compound 37 a product of partial hydrolysis of luzopeptins. Reaction with Br2 furnished 38 (Scheme 12).26 COOR COOR tBuOCl base N N NaBH3CN acid N COR' COR' NH 35 36 Scheme 11 Oxidation of N-acylpiperazic acids with other mild agents may lead to useful derivatives. Bock and coworkers have described several noteworthy transformations of compound 39 a potent oxytocin antagonist.25b This material reacts regioselectively at the piperazic acid A unit under a variety of conditions.Treatment with hydrogen peroxide attacks the imino linkage in ring A resulting in the formation of cyclic hydrazide 40; while reaction with 2,3-dichloro-5,6-dicyano-1,4-benzoqui- 439 Chemical Society Reviews 1998 volume 27 Me O N COOH HN N NH O OH Br2 Me O N COOH N N NH O O H HN NH N O N O 37 O H Bn N N O N A Me DDQ O H N O N A 41 Et3SiO H COOiBu N N O N N H HN Ac N O O 43 O 42 38 Scheme 12 none (“DDQ”) produces 41 apparently as a relatively stable compound (Scheme 13). By contrast we find that compounds Bn O K2CO3 MeOH N + 39 Scheme 13 COOiBu of the type 43 which are regioisomers of 41 are fragile substances.27 These materials may be obtained upon b-elimination of C-4-oxygenated piperazic acids and their esters e.g.42. It will be seen later that this reaction is often facile and it constitutes a major stumbling block in sequences directed toward the incorporation of fragments 3 in growing peptide chains. Exposure of 43 to the atmosphere tends to promote aromatization with concomitant release of the N-2 acyl unit as well as other reactions (Scheme 14). On the other hand A COOiBu O H O air COOH H N HN O O 44 Scheme 14 43 may be intercepted in situ with nucleophilic amines e.g. pyrrolidine to furnish products of conjugate addition in Chemical Society Reviews 1998 volume 27 440 moderate yield (Scheme 15).24 The difference in stability between structures 41 and 43 is probably due to a stereoelectronic effect.Compound 44 favors a conformation 46 which places the sole ring hydrogen at a pseudoequatorial position. Improper alignment of the C–H s orbital with the ring p system retards removal of this H atom by radical species such as e.g. ground- state triplet O2. Compound 43 by contrast exists in a conformation 47 wherein a hydrogen atom is always pseudoaxial and properly aligned with the ring p system for facile abstraction (Scheme 16). N COOiBu O N NH N H 42 43 HN O 45 O Scheme 15 N B R O RO R' NH O O N O N H2O2 H R' H N N H O H 47 46 N Scheme 16 O NH O 40 The N–N linkage of piperazic acids is susceptible to hydrogenolysis especially if the N-2 position is free.Thus reaction of 37 with H2 in the presence of PtO2 leads to 48 (Scheme 17).26 It is likely that this transformation involves an O HO H COOH N NH O NH N 37 H2 PtO2 O HO H COOH N NH O NH2 H2N + other products 48 Scheme 17 initial reduction of the imino linkage followed by a faster cleavage of the N–N bond. However hydrogenolysis of benzyl groups may be effected without rupture of the N–N bond in monoacyl derivatives of 1 as well as in acyclic a-hydrazino ester precursors to piperazic acids. Even N-benzyl groups may be cleaved under appropriate conditions without harm to the N–N linkage (Scheme 18; see also Scheme 10).Piperazic acids are relatively stable to acidic reagents. Even some of the C-4 oxygenated analogs survive exposure to strong trifluoroacetic acid solutions for short periods of time (10–15 min). Tolerance to basic agents is likewise good; however either free acids or esters of general structure 3 may suffer b-elimination under basic conditions. This propensity complicates the task of incorporating fragments containing 3 into growing peptide chains since even the weakly basic agents required during these rections (e.g. N-methylmorpholine) may H H COO-iPr COO-iPr H2 [Ref. 9] Pd(C) N NH N N COOBn Ac Ac O OBn OAc O OH OAc H2 O-iBu O-iBu Pd(C) O N O N [Ref. 24] BOCHN BOCHN Ac O N N 51 O O H H H2 [Ref. 17] Pd(C) N N 50 O COOMe BOC 49 O COOMe BOC N Bn OH OAc COOH 54 Aq.NaOH O O N N complex mixture N N NH NH 56 55 O O O 52 NH Scheme 18 COOiBu O Scheme 19 53 cause unwanted side reactions. Occasionally b-elimination may be controlled by the judicious choice of conditions. To illustrate attempted hydrolysis of 55 with aqueous LiOH produced complex mixtures; however hydroxy acid 56 was obtained in essentially quantitative yield after 5 minutes of contact time with aqueous NaOH (Scheme 19).24 Of course Aq. LiOH longer exposure also resulted in decomposition. It is noteworthy that hydrolytic cleavage of N-2 acyl substituents in piperazic acids of the type 2 and 3 occurs considerably more readily than one would expect for an amide linkage.Thus prolonged reaction of luzopeptin A 57 with 0.1 M aqueous NaOH at 37 °C yields fragment 37 (Scheme 20).26 Identical treatment induced hydrolysis of the oxazolone ring in compound 10 a transformation that normally requires considerably more vigourous conditions (Scheme 21).10 The ease of N-2 deacylation in piperazic acid derivatives may be due to resonant dispersal of the negative charge accumulating on N- 2 during framentation of a tetrahedral intermediate such as 60 into the adjacent imino linkage (Scheme 22). It is also possible that the inductive effect of the imino nitrogen aids the fragmentation of 60 by diminishing the basicity of the departing nitrogen anion. Notice that formation of 37 occurs without b-elimination within the oxygenated piperazic acid.This may be due to the fact that the piz unit is now present as a secondary amide instead of a free acid or an ester. The considerably diminished ability of secondary amides to undergo carbonyl a-deprotonation is probably responsible for the survival of the otherwise sensitive C-4 oxygenated functionality. This is also apparent in the examples of Scheme 23. We were unable to suppress b-elimination of the piperazic acid unit and subsequent decomposition of the substrates during Kunieda-type cleavage of the oxazolone ring in piz esters 62. By contrast secondary amide 63 underwent the reaction cleanly and in high yield.24 Piperazic acids occur in nature only as components of peptide natural products and in all known cases they are found as N-2 OH O O AcO H O N N NH H O N O N H HN O N O 2 OH MeO 57 0.1 M NaOH O HO H COOH N NH O NH N 37 + O COOH H N NH OH OH MeO Scheme 20 MOMO 0.1 M 10 OH NaOH N Scheme 21 R2 O R2 O O– OH N N N– N R1 60 61a R2 O – N N 61b Scheme 22 OAc OH CO–R CO–R Cs2CO3 O N MeOH N N O N N–BOC HO NHBOC O O 64 62 R = OiBu or OMe 63 R = NH-nBu 90% yield Scheme 23 58 H NH 59 – R1–COOH b-elim.dec. Chemical Society Reviews 1998 volume 27 441 acyl derivatives. Interestingly N-2 acylation of a free piperazic acid is problematic. As early as 1979 it was observed that 1 reacts selectively at N-1 with various acylating agents and that N-2 acylation of the resultant 65 is possible only with acid chlorides (Scheme 24).12a In support of these observations we H H COOH COOH H COOH R2–COCl R1–CO–X NH N R2 N N NH O NH O O R1 R1 1 66 OH OAc OBOC COOEt COOEt COOEt BOC2O Ac2O 10 h Pyridine NH NH N N N 65 Scheme 24 found that 67 is inert toward carboxylic acids activated by a variety of common peptide coupling agents and even toward carboxylic anhydrides.23 Thus 68 is the exclusive product of reaction with Ac2O and pyridine while prolonged treatment with BOC2O yields 69 (Scheme 25).The weak nucleophilicity NH BOC BOC BOC 69 68 R1–O H 67 Scheme 25 of the N-2 site is reduced even further in structures of the type 2.Attempted acylation of these molecules under gentle conditions normally fails while more vigorous conditions severely damage the substrate. This is especially true of oxygenated variants of the type 69 which tend to undergo b-elimination when exposed to several basic agents necessary for the conduct of acylation reactions. The resulting 70 then decomposed by a variety of pathways as indicated above (Scheme 26). Notice that the decrease in N-2 nucleophilicity in 69 is consonant with the ease of N-2 deacylation discussed earlier. COOR2 COOR2 COOR2 + other products NH N NH N N N 69 71 R1 = H or R3Si R2 = Me Et 70 Scheme 26 The reasons for the unusually poor nucleophilicity of piperazic acids have been researched in some detail.10 The fact that the N atom to be acylated is now part of a 6-membered ring suggests a parallel between piz and pipecolinic acid derivatives.While the latter are notoriously troublesome substrates in peptide-forming reactions they are nonetheless considerably more reactive than piz. For example pipecolinic units may be N-derivatized with carboxylic acids activated by N-methyl- 2-chloropyridinium chloride,28 while piperazic acids are completely inert under these conditions. In pipecolinic acid derivatives the difficulty of N-acylation may be due primarily to a stereoelectronic effect.29 In piperazic acids stereoelectronic problems appear to be greatly exacerbated by inductive erosion of N-2 nucleophilicity by the carboxy group but not by steric or conformational effects.Indeed compound 59 which is sterically and conformationally very similar to 2 or 3 reacts at N-2 even with weak acylating agents; e.g. 4- nitrophenyl esters in Chemical Society Reviews 1998 volume 27 442 the presence of N-hydroxybenzotriazole (“HOBt”). It is also worthy of note that reaction of 59 with Ac2O in pyridine delivers only diacetyl compound 73 with no evidence of N-1 acylation and consequent enamide formation or of tautomerization to an enamine and C-6-acylation (Scheme 27).10 As an MOMO H OOCC7H15- n O2N OH N N COC7H15- n HOBt 72 OAc Ac Scheme 27 CSA O H O H N + Cl– Me2N=CH2 O N A O A Me2N 74 Scheme 28 59 Ac2O pyrid.MOMO H N N 73 interesting aside it should be mentioned that hydrazone tautomerization and consequent C-6 derivatization in N-2 acyl variants of 1 and 2 are possible under more vigorous conditions. In particular reaction of 39 with camphorsulfonic acid (“CSA”) and N,N-dimethylimmonium chloride (“Eschenmoser’s salt”) produces Mannich-type derivative 74.25b Presumably CSA serves to promote conversion to ene hydrazine 73 by protonation of the imino nitrogen and prototropic shift (Scheme 28). To our knowledge however no examples of C-6 acylation of 2 or 3 are known. 39 N NH 73 The foregoing difficulties greatly complicate the creation of peptides incorporating N-2 acyl piperazic acids. Early methods for N-2 acylation of intermediates of the type 65 seem now less attractive than more recent techniques involving acylation of a terminally monoprotected a-hydrazino acid derivative such as 32.The free NH group of these substances also displays abnormally low nucleophilicity necessitating the use of highly reactive acid chlorides. This is exemplified by the conversion 75 ?77 in Scheme 29.16 The sequence leading to 77 also reflects recent advances in the preparation and use of acyl chlorides of Fmoc-protected amino acids.30 Hydrazines of general type 32 have also been acylated by an alternative technique involving condensation of their N-TMS derivatives with acyl fluorides. This is exemplified by the reaction of compound 78 with 79 (Scheme 30).31 As of this writing however this method does not appear to have been used for the preparation of peptides incorporating piperazic acids.The creation of N-2 serinyl derivatives of piz is especially problematic due to the propensity of many serinyl chlorides to undergo b-elimination of the protected oxygen functionality under the conditions of the acylation reaction. Yet the N-2 serinyl piz is a crucial motif of synthetically appealing targets such as luzopeptins and quinoxapeptins. This difficulty may be remedied by the use of reagent 81 (d configuration shown) wherein b-elimination is retarded on stereoelectronic grounds.32 The availability of this serinylating agent was central O H COOMe O NH BOCNH 75 H NH-Fmoc ClOC N3 H O H H COOMe COOMe O O N O N N H+ BOCNH H H NH-Fmoc NH-Fmoc H H N3 N3 77 76 Scheme 29 to the success of the synthesis of compounds 62–64,24 heretofore elusive dipeptide components of many luzopeptins and of quinoxapeptins.Its use is detailed below in the context of a synthesis of 87 (Scheme 31),33 a dipeptide subunit of the E series of luzopeptins. It is likely that acyl chloride 82 may be likewise employed for the preparation of N-2 threoninyl piperazic acids. The advent of reliable technology for the preparation of N-2 acyl piz frameworks has revealed a unique property of these susbtances. Briefly peptides incorporating proline or pipecolinic acid exist as mixtures of rotamers of the tertiary amide. As shown in Table 1 conformation A is always favored over B typically to the extent of 1.5:1 to 5:1 (NMR) though in some cases this preference may be stronger (cf.90). In sharp contrast analogous piz amides and peptides are rigidly fixed as rotamer A around the N-2-acyl bond.33 This preference is calculated to be especially strong in peptides based on unsaturated piperazic acids 2 and 3. Indeed variable temperature NMR experiments with some derivatives of 2 and 3 failed to reveal significant motion about the N-2-acyl bond in a temperature range from 289 °C to +120 °C suggesting also that the conformational energy barrier is large. The rigidity of these molecules may be attributed to an electrostatic interaction between the N-1 nitrogen and the N-2 acyl oxygen atoms that greatly disfavors conformer B thereby strongly reinforcing the molecules innate preference for A (Fig.1). More recently we have confirmed the calculated preferences of a number of piz derivatives by single crystal X-ray diffractometry.34 O H H O COOBn F + N NH TMS BOC–NH O 79 78 COOBn O N H H BOC–NH O NH O 80 Scheme 30 O H H H O Me Cl Cl O N N O Ac Ac O O 81 82 O H 1. 81 O 2. H+ 83 H COOMe N O N NHBOC COOMe O H O N O 84 Z = Ac N2H4 85 Z = H R O H NH H N N R = Ac R = BOC 88 6) 1.5 1 O (DMSO- d6) H O Z A 87 CO–Z H R 1.2 1 2.2 1 (DMSO- d COOMe N–Ac O CO–Z H R BOC2O N A COOMe O NHBOC OH H O N H OH d – N 86 Z = BOC Scheme 31 Table 1 CO–Z H O 4 1 (CDCl3) CONH–C8H17- n 15 1 (CDCl3) NHBOC 90 CO–Z N H O d – R BOCNH Cs2CO3 MeOH 89 Z = OR NHR Fig.1 COOMe H N N OH N COOMe O N–R B O O H N O N N d – B 91 d – These observations may have significant ramifications in problems relating to peptide secondary structure. A number of important aspects of protein structure and function are currently perceived to be intimately related to the presence of turn motifs Chemical Society Reviews 1998 volume 27 443 in peptide chains; that is of regions where a peptide chain reverses its direction.35 A diagram representing a typical peptide turn appears in Fig.2. Such turns are believed to be key structural features of b-sheet architectures and of binding domains involved in ligand–receptor interactions including those between proteins and nucleic acids. Accordingly functional structural and mimicry aspects of turns motifs have commanded considerable attention and remain to this day exceptionally active areas of scientific inquiry.36 H H R1 R2 N H N R1 H N O O O H N H O O N H O H N H 92 93 H R1 N H N O O H O N H 94 Fig. 2 Strain is often present in proximity of the amino acids occupying the peripheral position of a turn. This strain may be relieved by replacement of a secondary amide with a conformationally more flexible teriary amide at a turn site e.g.by the introdution of proline (“pro” cf. 93). Indeed this amino acid appears often at turn sites probably because it can readily accommodate a conformation of the type A (Fig. 1) which is conducive to turn formation. Turns are even more greatly favored if the peripheral amino acids are of opposite a-configuration the resulting structure being often described as a “heterochiral” sequence. Not surprisingly a heterochiral sequence of the type e.g. N-(d-aminoacyl)-l-pro (cf. 94) constitutes an especially good turn promoter. Indeed similar heterochiral sequences have been utilized extensively in turn mimicry. However the introduction of a proline within a peptide chain even in a heterochiral mode is insufficient to force a turn because as detailed above N-acyl pro generally do not manifest a sufficiently great preference for turn-promoting conformer A.Thus artificial peptide turns must be further stabilized by a significant number of other intramolecular interactions (multiple hydrogen bonds metal coordination disulfide bridges rigid templates etc.).37 Heterochiral as well as homochiral sequences wherein pro is replaced by a piperazic acid especially an unsaturated one like 2 or 3 display vector properties of a peptide turn. Recall N-2 piz peptides are locked in a conformation of the type A; therefore it seems plausible that these entities may be utilized to force turns in peptide chains. Piz peptides may thus become useful perhaps exceedingly so as building blocks for turn mimetics consisting of just a handful of amino acids.The resulting small molecule (mass < 500 Dalton) mimics of turn motifs may well find significant applications in various aspects of life science and health care. These are currently areas of intense research in our group. Structural evidence in support of the turn-forming ability of piz may be identified in a seminal paper by Bock and collaborators at Merck & Co. USA.25b The X-ray crystal structure of oxytocin antagonist 39 reveals turn motifs at the level of the piz subunits which interestingly are each part of a heterochiral sequence. An additional turn motif occurs at the level of the proline. The Merck scientists also observed significant changes in the biological activity of 39 upon reduction of the imino groups.It is conceivable that removal of Chemical Society Reviews 1998 volume 27 444 the imino subunit renders the molecule more flexible and consequently less apt to assume a conformation suitable for binding to the oxytocin receptor. This would be consistent with the calculated greater rigidity of unsaturated piz peptides vs. their unsaturated analogs. A further potential benefit of piz-based turns is implicit in Fig. 3 which shows an overlay of the calculated (PM3) structures of an l-pro peptide and of the corresponding l-2 analog. Proline and piperazic acid seem sufficiently similar that turn mimics based on 2 should display only marginal differences relative to their naturally occurring originals increasing the likelihood that sufficiently strong interactions will sussist once the “counterfeit” turn docks with a natural receptor site.Fig. 3 3 Conclusions and outlook Our interest in piperazic acids developed as a result of observations made during a synthetic venture in the luzopeptin domain. Of course it is our expectation that the climax of this endeavor the total synthesis of at least some of the luzopeptins will soon be forthcoming. This accomplishment is likely to provide the scientific community with the means necessary to investigate details of the mechanism of action of the natural products and to conduct medicinal chemistry and SAR studies. An equally important payoff of our work perhaps one of even greater importance is the discovery of the unusual conformational properties of piz units and their potential as turn-inducing instruments.Further research in this area will probably reveal interesting and novel aspects of peptide secondary structure and indeed the exploration of these issues is well underway in our group as of this writing. The chronology of the piperazic acid story just narrated provides also compelling evidence that an area of organic chemistry too often suspected of having already seen its heydays Synthesis has once again proven to be as central as ever in delivering new and potentially valuable research leads to the scientific community. 4 References 1 K. Bevan J. S. Davies C. H. Hassall R. B. Morton and D. A. S. Phillips J. Chem.Soc. (C) 1971 514 and refs. cited therein. 2 e.g. (a) A83586C T. A. Smitka J. B. Deeter A. H. Hunt F. P. Mertz R. M. Ellis L. D. Boeck and R. C. Yao J. Antibiot. 1988 41 726; (b) Antrimycins K. Morimoto N. Shimada H. Nakagawa T. Takita and H. Umezawa J. Antibiot. 1982 35 378 and refs. cited therein; (c) Aurantimycins U. Gr�afe R. Schlegel R. Ritzau W. Ihn K. Dornberger C. Stengel N. F. Fleck W. Gutsche and AH�arti J. Antibiot. 1995 48 119; (d) Variapeptin M. Nakagawa Y. Hayakawa K. Furihata and H. Seto J. Antibiot. 1980 43 477. 3 An excellent bibliography of many important natural products containing piperazic acids (aurantimycins azinothricin citropeptin L-156,373 L-156,602 verucopeptin) appears in ref. 20. 4 Recently described natural products containing piperazic acids (a) GE3 and GE3B T.Agatsuma Y. Sakai T. Mizukami and Y. Saitoh J. Antibiot. 1997 50 704; (b) Polyoxypeptins K. Umezawa K. Nakazawa T. Uemura Y. Ikeda S. Kondo H. Naganawa N. Kinoshita H. Hashizume M. Hamada T. Takeuchi and S. Ohba Tetrahedron Lett. 1998 39 1389. 5 Luzopeptins M. Konishi H. Ohkuma F. Sakai T. Tsuno H. Koshiyama T. Naito and H. Kawaguchi J. Am. Chem. Soc. 1981 103 1241 and refs. cited therein; Quinoxapeptins R. B. Lingham A. Hsu J. A. O’Brien J. M. Sigmund M. Sanchez M. N. Gagliardi B. K. Heimbuch O. Genilloud I. Martin M. T. Diez C. F. Hirsch D. L. Zink J. M. Liesch G. E. Koch S. E. Gartner G. M. Garrity N. N. Tsou and G. M. Salituro J. Antibiot. 1996 49 253. 6 Anti-HIV activity of luzopeptins Y. Inouye Y.Take S. Nakamura H. Nakashima N. Yamamoto and H. Kawaguchi J. Antibiot. 1987 50 100. 7 M. R. Attwood C. H. Hassall A. Kr�ohn G. Lawton and S. Redshaw J. Chem. Soc. Perkin Trans. 1 1986 1011. 8 R. W. Horton J. F. Collins G. M. Anlezark and B. S. Meldrum Eur. J. Pharmacol. 1979 59 75. 9 U. Schmidt C. Braun and H. Sutoris Synthesis 1996 223. 10 M. A. Ciufolini T. Shimizu S. Swaminathan and N. Xi Tetrahedron Lett. 1997 38 4947. 11 (a) J. Vidal S. Damestoy L. Guy J.-C. Hannachi A. Aubry and A. Collet Chem. Eur. J. 1997 3 1691; (b) J. Viret J. Gabard and A. Collet Tetrahedron 1987 43 891. 12 (a) C. H. Hassall W. H. Johnson and C. J. Theobald J. Chem. Soc. Perkin Trans. 1 1979 1451; (b) C. R. Davies and J. S. Davies J. Chem. Soc. Perkin Trans.1 1976 2390. 13 I. H. Aspinall P. M. Cowley G. Mitchell and R. J. Stoodley J. Chem. Soc. Chem. Commun. 1993 1179. 14 P. Hughes and J. Clardy J. Org. Chem. 1989 54 3260; P. Hughes and J. Clardy J. Org. Chem. 1995 60 2950. 15 Discussion K. B. Sharpless C. H. Behrens T. Katsuki A. W. M. Lee V. S. Martin M. Takatani S. M. Viti F. J. Walker and S. S. Woodard Pure Appl. Chem. 1983 55 589. 16 U. Schmidt and B. Riedl Synthesis 1993 809. 17 (a) F. P. J. T. Rutjes N. M. Teerhuis H. Hiemstra and W. N. Speckamp Tetrahedron 1993 49 8605; (b) F. P. J. T. Rutjes M. M. Paz H. Hiemstra and W. N. Speckamp Tetrahedron Lett. 1991 32 6629. 18 Y. Nakamura and C.-G. Shin Chem. Lett. 1991 1953; U. Schmidt and B. Reidl J. Chem. Soc. Chem. Commun. 1992 1186; K. J.Hale V. M. Delisser and S. Manaviazar Tetrahedron Lett. 1992 33 7613. 19 C. Gennari L. Colombo and G. Bertolini J. Am. Chem. Soc. 1986 108 6394; D. A. Evans T. C. Britton R. L. Dorow and J. F. Dellaria J. Am. Chem. Soc. 1986 108 6395; L. A. Trimble and J. C. Vederas J. Am. Chem. Soc. 1986 108 6397. 20 K. J. Hale J. Cai V. Delisser S. Manaviazar S. A. Peak G. S. Bhatia T. C. Collins and N. Jogiya Tetrahedron 1996 52 1047. 21 (a) G. Guanti L. Banfi and E. Narisano Tetrahedron 1988 44 5553; (b) C. Greck L. Bischoff and J. P. Genet Tetrahedron Asymm. 1995 6 1989. 22 e.g. (a) M. Kitamura M. Tokunaga T. Ohkuma and R. Noyori Org. Synth. 1992 71 1 (hydrogenation); (b) D. Seebach M. A. Sutter R. H. Weber and M. F. Z�uger Org. Synth. 1984 63 1 (Bakers’ yeast).23 M. A. Ciufolini and N. Xi J. Chem. Soc. Chem. Commun. 1994 1867. 24 M. A. Ciufolini and N. Xi J. Org. Chem. 1997 62 2320. 25 (a) M. G. Bock R. M. DiPardo P. D. Williams R. D. Tung J. M. Erb N. P. Gould W. L. Whitter D. S. Perlow G. F. Lundell D. J. Pettibone B. V. Clineschmidt D. F. Weber and R. M. Freidinger Vasopressin Colloque INSERM eds. S. Jard and R. Jamison John Libbey Eurotext Ltd UK 1991 vol. 208 pp. 349–355; (b) M. G. Bock R. M. DiPardo P. D. Williams D. J. Pettibone B. V. Clineschmidt R. G. Ball D. F. Weber and R. M. Freidinger J. Med. Chem. 1990 33 2323. 26 M. Konishi H. Ohkuma F. Sakai T. Tsuno H. Koshiyama T. Naito and H. Kawaguchi J. Antibiot. 1981 34 148. 27 N. Xi Dissertation Rice University 1997. 28 e.g. T. K. Jones S. G. Mills R. A. Reamer D. Askin R. Desmond R. P. Volante and I. Shinkai J. Am. Chem. Soc. 1989 111 1157 29 cf. M. A. Ciufolini M. A. Rivera-Fortin V. Zuzukin and K. H. Whitmire J. Am. Chem. Soc. 1994 116 1272 footnote 26. 30 L. A. Carpino M. Beyermann H. Wenschuh and M. Bienert Acc. Chem. Res. 1996 29 268 31 M. D. Ferguson J. P. Meara H. Nakanishi M. S. Lee and M. Kahn Tetrahedron Lett. 1997 38 6961. 32 N. Xi and M. A. Ciufolini Tetrahedron Lett. 1995 36 6595. 33 N. Xi L. B. Alemany and M. A. Ciufolini J. Am. Chem. Soc. 1998 120 80. 34 Unpublished results from these laboratories. We thank our colleague Professor Monique Perrin and her collaborators for their invaluable assistance with these X-ray studies details of which will be published 35 cf. J. Rizo and L. M. Gierasch Ann. Rev. Biochem. 1992 61 387. See 36 cf. e.g. C. K. Smith and L. Regan Acc. Chem. Res. 1997 30 153 and 37 cf. e.g. M. D. Struthers R. P. Cheng and B. Imperiali Science 1996 shortly. also ref. 33. references cited therein. 271 342 and references cited therein. Received 21st May 1998 Accepted 8th June 1998 445 Chemical Society Reviews 1998 vol

ISSN:0306-0012    

DOI:10.1039/a827437z

出版商:RSC    

年代:1998

数据来源: RSC

摘要:

The structures and magnetic properties of complexes containing 3d- and 4f-metals Richard E. P. Winpenny The University of Edinburgh West Mains Road Edinburgh UK EH9 3JJ Heterometallic complexes containing 3d-/4f-metals have been made with a variety of ligands including Schiff-bases pyridonates and amino-alcohols. The majority of species with Schiff-base ligands are trinuclear with Cu2Ln cores and with other ligands larger oligomers are found ranging as large as Cu12La8 and Cu12Ln6 clusters (Ln = Y Nd Sm or Gd). The magnetic properties displayed by these polynuclear species are discussed and the magnetic coupling between CuII and GdIII is always found to be ferromagnetic. 1 Introduction Whenever two magnetic ions are brought into close proximity our natural inclination is to assume that like bar magnets they will arrange themselves so that they couple anti-ferromagnetically and thus reduce the overall magnetism of the dimer.This prejudice is probably reinforced by studies of dimeric copper complexes such as copper acetate or by the magnetic behaviour of the active sites of any metalloprotein which contains more than one magnetic metal centre. When the prejudice is confronted by contrary fact the observation gains a good deal of piquancy and the divergence from clich�e is in itself an important scientific reminder that the magnetic behaviour of atoms and ions is not as simple as that of bar magnets. In 1985 Italian scientists led by Gatteschi,1 observed that when copper(ii) and gadolinium(iii) were brought into close proximity the coupling between the ions was ferromagnetic.This was surprising as gadolinium(iii) has unpaired electrons in all seven 4f-orbitals and it might be expected that at least one of these orbitals would overlap with the semi-occupied orbital on copper(ii). Such an interaction between two half-occupied orbitals would be anti-ferromagnetic as it would create a molecular orbital which could contain both electrons. Such an interaction is observed between CuII and the d5-centre MnII where the coupling is always anti-ferromagnetic.2 The explanation cannot lie in the interaction between ground state configurations. The 4f-orbitals of the lanthanoids are small internal orbitals and this leads to the interaction between the CuII 3d- and GdIII 4f-orbitals being minimal and hence the overlap which would lead to anti-ferromagnetic coupling is of little importance.The important interaction is between the semi-occupied orbital say dx22y2 on CuII and an empty orbital on the GdIII centre.3,4 In such a charge transfer configuration there are two possible orientations for the electron transferred from CuII to GdIII. If it aligns antiparallel to the seven electrons in the 4f-orbitals this would lead to an S = 3 spin state and would be equivalent to Richard Winpenny completed his BSc degree at Imperial College in 1985. His PhD studies were carried out at Imperial with Professor D. M. L. Goodgame before a post-doctoral period with Professor J. P. Fackler Jr. at Texas A&M University.Dr Winpenny joined the staff at Edinburgh in January 1990. anti-ferromagnetic exchange. If the additional electron aligns parallel to the seven 4f-electrons this would give an S = 4 spin state and is equivalent to ferromagnetic exchange. Due to Hund’s rule the latter will be lower in energy than the former and hence this excited state of the system favours ferromagnetic exchange. Initially there was some debate about which gadolinium orbital was the electron acceptor but it now seems clear that the 5d-orbitals are best suited to this role.4 It is the interaction of this charge transfer configuration with the ground state which leads to the ferromagnetic coupling that is always observed. The interaction reported was weak1—with a ferromagnetic J value of around 5 cm21 and was found in a trinuclear Gd2Cu complex.The initial report posed two questions could the interaction ever be sufficiently strong to be of more than purely academic interest and if it were found in larger or even polymeric complexes could it be a way of producing high spin molecules or even molecular ferromagnets. At about the same time as this chemistry appeared the high Tc superconductors containing copper–lanthanoid mixtures were also reported and a second potential use of these heterometallic cages became apparent—perhaps they could be used as precursors for producing new mixed-metal perovskite phases. Therefore in addition to Cu–Gd complexes Cu–Ln complexes in general became the subject for study.2 Complexes using Schiff-base ligands The initial work in the area used polydentate Schiff-base ligands to bridge between metal centres; later others have used oxamido ligands and our own work has been based on pyridonates. All these ligands have in common a mixed-donor set where it was intended that hard O-donors would favour binding to the oxophilic lanthanoid centres while the softer N-donor would bind to the copper centres. More recent work has shown that this differentiation is unnecessary and ligands with exclusively O-donors will work equally well. Several Schiff-base ligands have been used (L1–L10).1,3–6 In all cases the procedure used to make mixed-metal complexes was first to complex a CuII centre into a cavity of the Schiffbase normally consisting of an N2O2 donor set.Having isolated a copper complex this was then used as a ligand for the lanthanoid metal with the hydroxy-oxygen atoms acting as m2-bridges. The high coordination number favoured by 4f-metal centres normally led to coordination of two copper–Schiff-base complexes to each centre and hence trinuclear Cu2Gd complexes (Fig. 1). In one such synthesis incorporation of hydroxide combined with hydrolysis of the Schiff-base gave a tetranuclear Cu2Gd2 cage.3 Use of the ligand L7 gave dimerisation of the trinuclear Cu2Ln units through the carboxylate function (Fig. 2).4 This oligomerisation presents problems for studying the magnetic properties of such cages as allowance needs to be made for anti-ferromagnetic exchange between like metals in addition to ferromagnetic exchange between Cu–Gd however results supported the initial observation that the Cu–Gd exchange was always ferromagnetic.The problem of oligomer- 447 Chemical Society Reviews 1998 volume 27 isation was recognised by Sakamoto et al. and the Schiff-base ligands L5 L8 and L9 were used to attempt to restrict the cages to discrete dinuclear species.6 Unfortunately no crystallographic studies were reported for these molecules however this paper has been influential in later work. 3 Complexes using non-Schiff-base ligands Work with pyridonate ligands led immediately to structures with higher nuclearity and with largely unpredictable structures. The first complexes reported used 2-pyridone in a reaction with copper hydroxide and an hydrated lanthanoid nitrate to give Cu4Ln2 cages (where Ln = Gd or Dy),7 where the six metal centres are found at the vertices of a distorted octahedron with the Ln metals trans to one another (Fig.3). A very similar reaction but starting with copper methoxide led to still larger Cu8Ln2 cages (Ln = Nd or Y).8 The ten metals at the core of the structure lie in two mutually-perpendicular intersecting planes (Fig. 4). The six metals two Ln and four Cu in N OH N OH N N N N OH O OH OH N L7 O N N NH N L9 L1 N NH2 L4 N Chemical Society Reviews 1998 volume 27 448 N OH OH OH OH N L3 L2 OH O N OH OH OH OH N O OH L6 L5 O N OH OH NH OH OH N O L8 N ( OH O N O )3 L10 the first plane lie on the vertices of two squares with the two Ln sites common to both.The second plane of four Cu atoms passes through the centre of the larger plane and again the Cu atoms are arranged at the vertices of a square. These reactions used a copper salt either hydroxide or methoxide to deprotonate the 2-pyridone in situ. Deprotonation of the ligand prior to reaction with metal salts resulted in a Cu4La4 cage where a central La4O2 ‘butterfly’eparates two dinuclear copper units.9 At this stage it became apparent that isolation of the homometallic copper complexes of the pyridonate ligands prior to reaction with a lanthanoid salt might lead to cleaner and higher yielding reactions.Pursuing this strategy and using analogues of 2-pyridone substituted in the 6-position we produced a series of predominantly tetranuclear Key for Figs. 1–9 Fig. 1 The structure of [Cu2Gd(L1)2(H2O)3]3+1 Fig. 2 The structure of [Cu4Pr2(L7)4(NO3)6]4 2O2 cages where the exact product depends on the lanthanoid solvent and pyridonate derivative involved.10–15 It is noticeable that the work with the substituted pyridonates leads in general to lower nuclearities than cages which feature the parent ligand. The majority of structures fall into two broad types. In Type I structures,13 there is a central Cu2O2 ring with pyridonate ligands attached to the copper centres through the ring N-donor and which bridge on to peripheral Ln atoms through the O-donor (Fig.5). The oxygen atoms of the central Cu2O2 ring are derived from either methoxide if methanol is used as a solvent or hydroxide if other solvents are used. This structure is favoured by the 6-methyl-2-pyridonate ligand (mhp),13,14 and is also found for complexes of the heavier 4f-elements with 6-chloro- and 6-bromo-2-pyridonate. Type II structures feature either Cu3Ln or Cu2Ln2 held together by eight pyridonate ligands (Fig. 6).14,15 These structures are found for the 6-haloderivatives of these ligands. For the Cu3Ln cores a central CuLnO2 ring is present with the oxygen atoms provided by two pyridonate ligands with the N-donors of these pyridonates attached to the remaining Cu centres. Two further pyridonates are attached to the Cu of the central ring through N-donors with the oxygen attached to the remaining Cu centres.The final four pyridonates bridge between the Ln and the external copper atoms. For the Cu2Ln2 cores a very similar arrangement of pyridonates is found but here the central core is an Ln ring. Fig. 3 The structure of [Cu4Gd2(hp)8(Hhp)4(OH)2(NO3)4(H2O)2] (where Hhp = 2-pyridone)7 Fig. 4 The structure of [Cu8Nd2(O)2(hp)12(Cl)2(OMe)4(H2O)4] (where Hhp = 2-pyridone)8 Initially we thought that Type I structures were favoured by use of MeOH as a solvent however it has become apparent that the controlling influences are more subtle. For the mhp ligand Type I structures are found even in the absence of MeOH with bridging hydroxides incorporated instead.14 With 6-fluoro- 2-pyridonate Type II structures are found even in the presence of MeOH.15 One control is probably the basicity of the deprotonated pyridonate ligand.For mhp the electron donating methyl group leads to a reasonable basicity for the ligand which allows deprotonation of either the solvent or any waters of crystallisation present and hence leads to bridging methoxide or hydroxide ligands and Type I cages. For 6-fluoro-2-pyridonate the strongly electronegative fluorine reduces the basicity and deprotonation of solvent is never seen and hence coordination of solvent has little structural importance. For the 6-chloro- and 6-bromo-2-pyridonates which have intermediate basicity early 4f-elements give Type I and late lanthanoids Type II cages.This observation is more difficult to rationalise. We also find that lanthanum itself has a quite different coordination chemistry with these ligands than the smaller 4f-elements probably because the larger radius of the La centre Fig. 5 The ‘Type I’ structure of [Cu2Gd2(OMe)2(mhp)4(NO3)4- (Hmhp)2(MeOH)2] (where Hmhp = 6-methyl-2-pyridone)13 Fig. 6 The ‘Type II’ structure of [Cu3Gd(chp)8(NO3)] (where Hchp = 6- chloro-2-pyridone)15 449 Chemical Society Reviews 1998 volume 27 2Ln2 complexes but here a 12La8(m3-OH)24 core results (Fig. 7). The leads to a higher coordination number. Both the Cu4La4 cage mentioned above,9 and an extremely asymmetric Cu3La cage seem unique to this metal,11 however the most unusual result involves the reaction of [Cu6Na(mhp)12][NO3] with hydrated lanthanum nitrate in dichloromethane.15 With later lanthanoids this reaction generates Type I Cu cage with a Cu structure has non-crystallographic Oh symmetry with eight La atoms at the corners of a cube and the twelve Cu centres at the mid-points of the edges of the cube (thus forming a cuboctahedron).The cage is stabilised by a shell of disordered pyridonate nitrate and water molecules attached to the lanthanum sites which are nine or ten coordinate. None of these peripheral molecules bridge between metal sites and the cage is exclusively held together by hydroxy bridges. There is also a central disordered nitrate anion captured within the cage. This last very high nuclearity cage appears to be an attractive by-way in the pyridonate chemistry.We have been unable to scale-up the reaction or make the cage with any lanthanoid except lanthanum. However work from China using betaine ligands—either pyridinioacetate16 or pyridiniopropionate17— has generated cages containing a Cu12Ln6(m3-OH)24 core (Ln = Y Nd Sm or Gd) (Fig. 8). The curiosity here is that the cuboctahedron of copper centres is largely unchanged from the core in the Cu12La8(m3-OH)24 cage but now rather than being contained in a cube of 4f-metals it is contained in an octahedron. The non-crystallographic Oh symmetry is also retained however the 24 m3-hydroxides which hold together the cage are supplemented by bridging betaine ligands which span Gd···Cu vectors. The structure includes a central disordered perchlorate anion which may be involved in templating the structure.These later reactions are moderately high-yielding and seem applicable to many of the 4f-metals. Unfortunately the magnetic properties of these cages are dominated by the antiferromagnetic exchange between CuII ions and hence only low spin species result.16 Two further classes of mixed-donor ligands have been investigated. Ligands based on oxamido ligands have been used to make both trinuclear18 and polymeric Cu–Gd cages.19,20 In both cases the Cu–Gd coupling is found to be ferromagnetic. The latter complexes contain 2D-sheets linked by oxamido and oxalate ligands,19,20 and in the initial reports there was no evidence such as phase transitions of long-range magnetic Fig.7 The [Cu12La8(OH)24] core of the polynuclear complex (where Hmhp [Cu12La8(OH)24(NO3)21.2(Hmhp)13(H2O)5.5][NO3]2.8 = 6-methyl-2-pyridone).15 The view taken is perpendicular to one square face of the Cu12 cuboctahedron and the equivalent view is also used for Fig. 8 Chemical Society Reviews 1998 volume 27 450 ordering even down to 1.3 K. Later studies suggest that twodimensional ordering is found at lower temperatures.21 The final class of mixed-donor ligands investigated are amino-alcohols. Wang and co-workers have reported the use of both 1,3-bis(dimethylamino)propan-2-ol22 and 2,6-bis- (dimethylamino)-4-methylphenol in this chemistry.23 The former ligand generates triangular Cu2Ln cages in good yield (Ln = La or Nd) and a dimeric Cu–Pr cage in very low yield.24 The latter amino-alcohol gave good yields of dimeric Cu–Pr cages by reaction of copper methoxide and praesodymium tris(hexafluoroacetylacetonate) with the ligand in THF.23 Two of these series of cages—Cu2Ln2 cages with 6-methyl- 2-pyridonate13 and Cu2Ln cages with 1,3-bis(dimethylamino)- propan-2-ol22—have been studied as potential precursors for synthesis of mixed-metal oxides.The observations appear to be similar; initial decomposition is to oxide or oxide-carbonate phases of the individual metals followed by reaction of these phases to give mixed-metal oxides at moderate temperatures e.g. X-ray powder diffraction analysis indicates that La2CuO4 begins to form at around 670 °C and Yb2Cu2O5 at around 520 °C.13 Unfortunately diffraction peaks due to CuO are found in all these samples at low temperatures and prolonged annealing is required to produce even moderately pure mixedmetal oxides.Preparation of thin films of superconducting oxides has been reported by Wang using salts of copper barium and lanthanoids mixed in the presence of amino-alcohols but without isolating molecular precursors.25 4 Magnetic studies and possible magneto-structural correlations Magnetic studies of these complexes have shown that in all cases the interaction between CuII and GdIII appears to be ferromagnetic confirming initial observations. The range of cages reported has allowed exploration of possible correlations between the structures of these heterometallic cages and the magnitude of the magnetic exchange interaction J between CuII and GdIII centres.However a difficulty is that the absolute range covered by these measurements is rather small with the largest J-value reported being some 7.4 cm21 (found for {Cu(L3)2Gd(H2O)}) and the smallest around 0.1 cm21. There is some correlation between J and the Cu···Gd distance and this can be fitted to an exponential function such that 2J = Aexp[BdGd···Cu] (where A = 6.5 3 104 and B = 22.833 with J in cm21 and dGd···Cu in Å).14 This correlation was made on purely empirical grounds and remains to be proven but Fig. 8 The [Cu12Gd8(OH)24] core of the polynuclear complex [Cu12Gd8- (OH)24(O2CCH2CH2NC5H5)12(H2O)16(ClO4)][ClO4]17 17 indicates that the nearer the Cu and Gd approach each other the larger the exchange.This might be expected given that the accepted mechanism for this ferromagnetic interaction involves an excited state in which an electron has been transferred from Cu to Gd. Perhaps more importantly it also sets an upper limit for this exchange interaction of around 10 cm21 as the Gd···Cu separation is never likely to be less than 3.1 Å. The best means of testing such a correlation is to synthesise discrete dinuclear CuGd complexes where no anti-ferromagnetic exchange between like metals is present to confuse the interpretation of the Gd···Cu exchange. This was recognised several years ago by Sakamoto et al.6 The Wang group managed to make a discrete Cu–Pr complex using 2,6-bis(dimethylamino)-4-methylphenol,23 but did not make the Gd analogue where the magnetic properties would have been amenable to modelling.Oligomerisation is prevented in this cage by using three hexafluoroacetylacetonate ligands to block coordination sites on the praesodymium. Similar approaches have since been reported by Costes et al.,26 using Schiff-base L6 and blocking six sites on Gd using nitrate Kahn’s group using Schiff-base L3 and blocking six sites using hexafluoroacetylacetonate,27 and a Polish group using 1,3-bis(dimethylamino)propan-2-ol as a bridging ligand and blocking six sites using triflate.28 2 These results suggest that a correlation between distance and the exchange integral is oversimplified and Kahn has suggested27 that because the magnitude of the exchange coupling is dependent on the exchange transfer integral between the Gd 5d-orbitals and the Cu 3d-orbitals then in systems with a GdCuO2 ring the exchange integral will be correlated with the angle between the GdO2 and CuO2 planes,27 with the maximum value found when the two planes are co-planar.The argument also supports the view that the Cu···Gd distance is important as the transfer integral is certainly dependent on the inter-nuclear distance. Results from Costes et al. feature a planar GdCuO ring,26 and the coupling in question is only 7.0 cm21 which again support the view that the ferromagnetic coupling between Cu and Gd is never likely to be greater than 10 cm21. A problem with this latter correlation is that it has limited immediate applicability—essentially it is only of use for compounds which contain the two metals bridged by two m-oxygen donors.Many Cu–Gd complexes are known which do not contain this structural feature. A different approach to synthesising discrete units is to design a ligand which provides all the donor groups necessary to encapsulate both Cu and Gd centres. The tripodal Schiff-base ligand L10 provides three N-donors three hydroxy O-donors and six acetal oxygen donors and therefore seemed suitable for such a role.29 Unfortunately while a Ni–La complex could be made with such a ligand copper could not be readily incorporated into the cavity ‘designed’ for a 3d-metal centre and there is a considerable problem of hydrolysis of the acetal groups.29 More robust versions of this tripodal ligand can be envisaged.In none of these complexes have particularly high spin ground states been observed. In trinuclear Cu2Gd species an S = 9/2 ground state is expected and found but for all higher oligomers (as opposed to oxamido-bridged polymers) antiferromagnetic exchange between copper centres predominates and leads to molecules with low spin or even diamagnetic ground states. The problem is in the topology of the larger cages where short Cu···Cu contacts are normally found. Other than the work with oxamido-ligands no means of segregating the metals into alternating Cu and Gd sites have yet been reported. 5 Conclusions The future of magnetic studies of heterometallic 3d–4f complexes is in the study of more complicated ions than CuII and GdIII.All other paramagnetic 4f-ions are anisotropic and this will introduce anisotropy into the magnetic ground state of any cage. As anisotropy appears vital in forming ‘singlemolecule magnets’ then studies of such cages are potentially exciting. The work to isolate discrete dinuclear Cu–Gd cages will be particularly useful as this will allow study of isostructural Cu–Ln complexes. Kahn has already studied a series of oxamido-bridged polymers of formula Ln2[Cu(opba)]3,30 where Ln = any 4f-element from Tb to Yb and opba = ortho-phenylenebis(oxamato). For Tb and Dy there is a divergence in cMT at low temperature (cM is the molar magnetic susceptibility) indicating magnetic ordering and possibly the presence of one-dimensional ferro- or ferrimagnets.For Ln = Ho Er Tm or Yb no significant Cu···Ln interaction could be measured. These results are probably the first concrete indication of the likely diversity of magnetic properties that can be expected when orbitally-degenerate 4f-metals are involved in mixed-metal complexes. Equally moving to other 3d-metals will be challenging and may lead to novel properties. We have shown that pyridonate ligands can be used to make Co–Ln Ni–Ln31 and Mn–Ln cages32 using similar procedures to those adopted for Cu–Ln cages. 6Sm cage has Perhaps most excitingly a very beautiful Ni been reported using l-prolinate as a ligand (Fig. 9).33 This last result is particularly intriguing because the procedure used should be general to many 3d-metals and it should be possible to replace the central Sm ion with a range of other 4f-metals.The structure contains a central icosahedrally coordinated Sm ion surrounded by an octahedron of Ni centres. The Ni···Ni contacts are much longer (5.23 Å) than the Sm···Ni vectors (3.70 Å) and this may lead for the correct combination of 3d- and 4f-metals to ferromagnetic exchange between dissimilar metals dominating over anti-ferromagnetic exchange between like. No magnetic properties of this cage or related cages have yet been reported. 6 Acknowledgements The synthetic work carried out in Edinburgh was performed by Paul Milne Craig Grant Steve Archibald and Euan Brechin with crystallographic characterisation due to Sandy Blake Bob Gould and Simon Parsons.The project has been supported by the EPSRC(UK) The Leverhulme Trust and NATO. I am also grateful to Professor Cristiano Benelli (Florence) for magnetic studies on the mixed-metal complexes we have made and to Dr Fig. 9 The structure of the heptanuclear cage [Sm{Ni(pro)2}6]3+ (where pro = l-proline)33 451 Chemical Society Reviews 1998 volume 27 Peter Thornton (QMWC London) for preliminary magnetic studies on hexanuclear Cu–Gd complexes. 7 References 1 A. Bencini C. Benelli A. Caneschi R. L. Carlin A. Dei and D. Gatteschi J. Am. Chem. Soc. 1985 107 8128. 2 For example Y. Pei Y. Journaux O. Kahn A. Dei and D. Gatteschi J. Chem. Soc. Chem. Commun. 1986 1300. 3 C. Benelli A. Caneschi D. Gatteschi O. Guillou and L.Pardi Inorg. Chem. 1990 29 1750. 4 M. Andruh I. Ramade E. Codjovi O. Guillou O. Kahn and J. C. Trombe J. Am. Chem. Soc. 1993 115 1822. 5 A. Bencini C. Benelli A. Caneschi A. Dei and D. Gatteschi Inorg. Chem. 1986 25 572. 6 M. Sakamoto M. Hashimura K. Matsuki N. Matsumato K. Inoue and H. Okawa Bull. Chem. Soc. Jpn. 1991 64 3639. 7 D. M. L. Goodgame D. J. Williams and R. E. P. Winpenny Polyhedron 1989 8 1531. 8 S. Wang Z. Pang and M. J. Wagner Inorg. Chem. 1992 31 5381. 9 A. J. Blake P. E. Y. Milne R. E. P. Winpenny and P. Thornton Angew. Chem. Int. Ed. Engl. 1991 30 1139. 10 A. J. Blake R. O. Gould P. E. Y. Milne and R. E. P. Winpenny J. Chem. Soc. Chem. Commun. 1991 1453. 11 A. J. Blake R. O. Gould P. E. Y. Milne and R. E. P. Winpenny J.Chem. Soc. Chem. Commun. 1992 522. 12 A. J. Blake P. E. Y. Milne and R. E. P. Winpenny J. Chem. Soc. Dalton Trans. 1993 3727. 13 A. J. Blake V. Cherepanov A. A. Dunlop C. M. Grant P. E. Y. Milne J. M. Rawson and R. E. P. Winpenny J. Chem. Soc. Dalton Trans. 1994 2719. 14 C. Benelli A. J. Blake P. E. Y. Milne J. M. Rawson and R. E. P. Winpenny Chem. Eur. J. 1995 1 614. 15 A. J. Blake R. O. Gould C. M. Grant P. E. Y. Milne S. Parsons and R. E. P. Winpenny Dalton Trans. 1997 485. Chemical Society Reviews 1998 volume 27 452 16 X.-M. Chen S. M. J. Aubin Y.-L. Wu Y.-S. Yang T. C. W. Mak and D. N. Hendrickson J. Am. Chem. Soc. 1995 117 9600. 17 X.-M. Chen Y.-L. Wu Y.-X. Tong and X.-Y. Huang J. Chem. Soc. Dalton Trans. 1996 2443. 18 C.Benelli A. C. Fabretti and A. Giusti J. Chem. Soc. Dalton Trans. 1993 409. 19 O. Guillou O. Kahn R. L. Oushoorn K. Boubekar and P. Batail Angew. Chem. Int. Ed. Engl. 1992 31 626. 20 O. Guillou P. Bergerat O. Kahn E. Bakalbassis K. Boubekar P. Batail and M. Guillot Inorg. Chem. 1992 31 110. 21 F. Bartolome J. Bartolome R. L. Oushoorn O. Guillou and O. Kahn J. Magn. Magn. Mater. 1995 140 1711. 22 S. Wang Z. Pang K. D. L. Smith and M. J. Wagner J. Chem. Soc. Dalton Trans. 1994 955. 23 L. Chen S. R. Breeze R. J. Rousseau S. Wang and L. K. Thompson Inorg. Chem. 1993 34 454. 24 S. Wang Z. Pang and K. D. L. Smith Inorg. Chem. 1993 32 4992. 25 S. Wang Polyhedron 1998 17 831. 26 J.-P. Costes F. Dahan A. Dupuis and J.-P. Laurent Inorg. Chem. 1996 35 2400. 27 I. R. Ramade O. Kahn Y. Jeannin and F. Robert Inorg. Chem. 1997 36 930. 28 S. Gao O. Borgmeier and H. Leuken Acta Phys. Pol. A 1996 90 393. 29 S. J. Archibald A. J. Blake S. Parsons M. Schr�oder and R. E. P. Winpenny Dalton Trans. 1997 173. 30 R. L. Oushoorn K. Boubekeur P. Batail O. Guillou and O. Kahn Bull. Soc. Chim. Fr. 1996 133 777. 31 E. K. Brechin S. G. Harris S. Parsons and R. E. P. Winpenny Dalton Trans. 1997 1665. 32 M. Murrie S. Parsons and R. E. P. Winpenny unpublished results. 33 Y. Yukawa S. Igarashi A. Yamano and S. Sato Chem. Commun. 1997 711. Received 28th April 1998 Accepted 17th June 19

ISSN:0306-0012    

DOI:10.1039/a827447z

出版商:RSC    

年代:1998

数据来源: RSC

摘要:

Aqueous chemistry of N-halo-compounds X. L. Armesto M. Canle L. M. V. García and J. A. Santaballa* Departamento de Química Fundamental e Industrial Facultade de Ciencias Universidade de A Coru�na A Zapateira s/n. E-15071 A Coru�na Galicia Spain. E-mail xlvarb@udc.es mcanle@udc.es vicky@udc.es arturo@udc.es Halogens in aqueous solution are still used world-wide as disinfectants. During the process of halogenation the substances present in water undergo several chemical processes yielding relatively unstable intermediate species; their life-times in the medium depend on their structure and on the physico-chemical conditions. Several low molecular weight hydrocarbons are formed during water halogenation some of them potent mutagens and/or carcinogens. Halogenation also takes place in vivo involving the system myeloperoxidase/ H2O2/halide which increases the relevance of such reactions and opens new research fields.1 Introduction. The general chemistry of aqueous halogen oxidants X2 (aq) has long received attention and has been summarized a number of times by different authors in general environmental chemistry textbooks.1 Of course there have also been many Xos�e Luis Armesto Barbeito born in A Coru�na received his BS degree in 1970 and his PhD degree in Physical Chemistry in 1975 from the Universidade de Santiago de Compostela (Spain). He joined the Facultade de Ciencias at the Universidade de A Coru�na as Physical Chemistry Professor in 1993. Dr X. L. Armesto has been Master in Environmental Science and Technology since 1992 and his research is related to Environmental Chemistry.Mois�es Canle L�opez was born in A Coru�na. He graduated in Chemistry from the Universidade de Santiago de Compostela in 1990 and he received his PhD degree in Environmental Chemistry from the Universidade de A Coru�na in 1994. After two short periods working with Professor Rory A. More O’Ferrall and with Professor Howard Maskill he has spend a long period working with Professor Steen Steenken at the MPI at M�ulheim an der Ruhr (Germany) and has become interested in fast reactions in solution. He is Assistant Professor at the Universidade de A Coru�na. Mar�ýa Victoria Garc�ýa Dopico was born in A Coru�na and graduated in Chemistry from the Universidade de Santiago de X.L. Armesto M. V. Garc�ýa J. A. Santaballa specialized symposia and publications related specifically to the environmental and health problems of water treatment.2 The chemistry of N-halo-amines has also been reviewed in an overall manner with emphasis on N-Cl- and N-Br-amines.3 Recently the IUPAC has devoted a ‘White Book on Chlorine’ to a broad review covering among other topics industrial environmental and health aspects of chlorine and organochlorine chemistry.4 Even though aqueous halogens and their derivatives are still used world-wide as disinfectants in the last few years halogenbased water treatment has become more and more unpopular. Most of the developed countries endeavour to apply new water disinfection techniques that avoid the use of halogens.However the situation is still far from an end and halogen-based water treatment will continue to be unavoidable in the near future. All these facts give only a vague idea of the enormous importance that aqueous halogen chemistry has reached both Compostela in 1990. She received her PhD degree in Environmental Chemistry from the Universidade de A Coru�na in 1995. After a short stay as Assistant Professor in the Universidade de Santiago de Compostela she moved to the Universidade de A Coru�na also as Assistant Professor where she continues her research mainly related to halogenation processes of environmental concern. Juan Arturo Santaballa L�opez obtained his Licenciate degree in Chemistry in 1980 and his PhD in Physical Organic Chemistry from the Universidade de Santiago de Compostela in 1985.He then worked as Assistant Professor in the Departamento de Qu�ýmica F�ýsica at the Universidade de Santiago de Compostela. After two short post-doc fellowships with Dr Jerry Kresge in the Chemistry Department at the University of Toronto (Canada) and with Dr Rory A. More O’Ferrall in the Chemistry Department of the University College of Dublin (Ireland) he joined the academic staff of the Sciences Faculty at the Universidade de A Coru�na in 1990. His current research interests include chemical reactivity in aqueous solution highly reactive intermediates and theoretical chemistry. M. Canle L. 453 Chemical Society Reviews 1998 volume 27 from the fundamental and industrial point of view.Here we attempt to present a broad view of the state-of-the-art of the chemistry of N-halo-compounds in aqueous solution their generation reactivity and of the renewed interest of these substances in biological systems. 2 Generation of N-halo-compounds. Aqueous solutions of halogens have a strong oxidizing character. Different species can be responsible for such oxidizing character depending on the acidity of the medium. Some of the possible equilibria between the oxidizing species are summarized in eqns. (1)–(8) the equilibrium constants are given for a 298 K and are in the order Cl Br I.2,5,6 X2 (g) " X2 (aq) K1 = 0.062 0.21 0.0013 X2 (aq) + 2 H2O (l) " HOX (aq) + H3O+ (aq) + X2 (aq) K2 = 4.2 3 1024 7.2 3 1029 2.0 3 10213 X2 (aq) + 2 OH2 (aq) " XO2 (aq) + X2 (aq) + H2O (l) K3 = 7.5 3 1015 2 3 108 30 X2 (aq) " X+ (aq) + X2 (aq) K4 ~ 10240 10230 10221 X2 (aq) + H2O (l) " H2OX+ (aq) + X2 (aq) K5 ~ 10230 10220 10210 HOX (aq) + H2O (l) " XO2 (aq) + H3O+ (aq) K6 = 3.4 3 1028 2 3 1029 2 3 10211 3 XO2 (aq) " 2 X2 (aq) + XO32 (aq) (1) (2) (3) (4) (5) (6) (7) K7 = 1027 1015 1020 X2 (aq) + X2 (aq) " X32 (aq) K8 = 0.18 15.9 698 (8) These oxidant species react readily with N-compounds to give the corresponding N-halo-derivatives.The oxidation products depend on the ratio [X2 (aq)]/[N-compounds] and on the acidity of the medium. Such dependence is usually known as ‘breakpoint’ and is illustrated in Fig. 1 for a general case in Fig.1 General scheme of the breakpoint chlorination of a sample of natural water which a sample of natural water is chlorinated.1 ‘Free chlorine’ is the content of Cl2 HOCl and ClO2; ‘combined chlorine’ is the total concentration of chloramines (NR2Cl + NRCl2 + NCl3) and their sum (free chlorine + combined chlorine) is called ‘total chlorine’. The dose of chlorine is the analytical concentration of chlorine dissolved in water and the difference between this and the residual chlorine reflects the demand of chlorine that is how much chlorine is consumed to oxidize the pollutants (organic and inorganic) contained in the sample. As a consequence of the world-wide use of chlorination in water and wastewater treatment studies regarding the halogenation of N-compounds have been mostly restricted to the use of chlorine and bromine derivatives.The relevance of bromine for water treatment comes mainly from the relative abundance of bromide in natural waters mainly in seawater; in the presence of Chemical Society Reviews 1998 volume 27 454 the bromide ion the following fast process given in eqn. (9) takes place.2 (9) HOCl(aq) + Br2 (aq) ? HOBr (aq) + Cl2 (aq) k = 2.95 3 103 mol21 dm3 s21 (T = 298 K) i.e. when [Br2] is relatively high chlorination is effectively equivalent to bromination. An analogous process is observed with I2 but at the same time fast disproportionation reactions occur,5 which seriously complicate kinetic studies. Interest in the application of I-derivatives to water disinfection comes from their use on board the Shuttle Orbiter and in the near future in the International Space Station Alpha.7 Fluorine is extremely reactive and oxidation products are usually obtained rather than the N-fluoroamines.2.1 Monohalogenated amino compounds When the ratio of the amino compound concentration to the halogen concentration is 1 1 or higher the monohalogena halogenation by molecular halogens [X2 (aq)]9,10 involves the fast bimolecular halogen transfer to the unprotonated nitrogen formally as X+ with release of halide ion [eqn. (10)]. R2NH (aq) + X2 (aq) ? R2NXH+ (aq) + X2 (aq) (10) Bimolecular rate constants for these reactions are collected in Fig. 2 the values being very high i.e.at or close to the diffusion control limit. When pH > 5 [X2 (aq)] is so small that the previously mentioned pathway is no longer significant HOX and/or XO2 becoming the active species. The generation of N-Cl-compounds follows a second order rate law first order in both the nitrogenated compound and the halogenating agent. A bell-shaped dependence of the rate constant with the acidity is observed (shown in Fig. 3 for the case of amino acids) according to eqn. (11) where a b and c are empirical parameters. k = a obs (11) [H3O+] (b + [H3O+])(c + [H3O+]) = k 3Ka a The observed behaviour is adequately explained by assuming a reaction between the halogenating agent and the nitrogenated compound. However identifying the nature of the rate determining step is not straightforward.Considering the major species present under the experimental conditions (HOX 1 XO2 12 R2NH 2 and R2NH2 + 2H+) four elementary steps are possible in the case of amines. If more than one acid–base equilibrium takes place for the nitrogenated compound the number of possible elementary steps increases. The experimental evidence is consistent with both HOX 1 + R2NH 2 (‘molecular pathway’) and XO2 12 + R2NH2 + 2H+ (‘ionic pathway’) two kinetically indistinguishable processes. A detailed study of the structure–activity effects and the comparison with analogous processes allowed identification of the ‘molecular pathway’ as the one taking place.11 An expression like eqn. (11) is deduced for the pH dependence of kobs with a 2 b = Ka2 and c = K5 where k is the second order rate constant for the chlorination reaction and Ka2 and K5 are respectively the equilibrium constants for deprotonation of the ammonium cation and of HOCl.Thus the maximum lies at the average of the pKa’s of HOCl 1 (pK6) and R2NH2 + 2H+ (pKa2). Fig. 2 broadly sketches the typical order of magnitude of k measured for the chlorination of different amino compounds 2 with HOCl. They are of the same order of magnitude (107–108 mol21 dm3 s21) for all the amines with conjugate acids of pK > 9. Such similarity together with the very low activation enthalpies points to a nearly diffusion controlled bimolecular process. However this is not in agreement with the observed k ~ 108 mol21 dm3 s21 for HOCl and k > 109 mol21 dm3 s21 for Cl2 (aq).12 A cyclic structure for the transition state 3 implying Fig.2 Bimolecular rate constants for halogenating agents a number of hydrogen-bonded solvent molecules (Scheme 1) could account for the activation parameters. For less basic N-compounds a curved dependence of the second order rate constant for chlorination with the basicity of the amino group has been found (Fig. 4). The analysis of the curvature in terms of Marcus’ theory13 (using the More O’Ferrall–Lewis approach) gives an intrinsic barrier D‡G0 0 Å 15 kJ mol21 for the chlorine transfer between the HOCl and the unprotonated nitrogen. The lack of measurements of the Gibbs free energy for the reaction does not allow estimation of the work terms.Chlorination of tertiary amines is a much slower reaction the rate equation being consistent with reaction of HOCl and the unprotonated amine (‘molecular’ pathway). Second order rate constants are some three orders of magnitude lower than those for primary or secondary amines of similar base strength. The fact that these substrates are incapable of hydrogen bonding to the solvent at the transition state may explain such rate differences and reinforces the suggestion of the involvement of water molecules at the transition state (TS) in the chlorination of primary and secondary aliphatic amines. A comparison of the substituent effects for the chlorination of primary/secondary and tertiary amines shows opposite slopes in the Taft plots.This is interpreted in terms of a development of positive charge (r < 0 ) on the nitrogen for tertiary amines and of negative charge (r > 0 ) for primary/secondary amines. Fig. 3 pH-dependence of the rate of chlorination of a-amino acids by HOCl; (:) glycine (5) isoleucine (-) amino isobutiric acid In the presence of acetic/acetate buffer CH3COOCl is formed according to eqn. (12).14 CH3COOH (aq) + HOCl (aq) " CH3COOCl (aq) + H2O (l) K (12) 9 = 2.5 3 1023 (298.0 K) This species is even more powerful as a chlorinating agent than Cl2 (aq) itself. In the chlorination of N-methylacetamide,15 the corresponding second order rate constant is 100 times higher 2 (aq). It is noticeable that chlorination by ClO2 than that of Cl also takes place which is not observable in the chlorination of amines or a-amino acids.The following sequence holds for the reactivity of the different aqueous chlorinating agents mentioned HOCl < ClO2 < Cl2 < AcOCl. ‡ H R2 Cl O N R1 R2 R2 HOCl + H N •• •• N H H R1 R1 H O H O H H 2 1 4 H O H + H2O 455 3 Scheme 1 Fig. 4 Rate-equilibrium plot in the chlorination of N-compounds by HOCl Chemical Society Reviews 1998 volume 27 Halogenation of nitrogenated compounds can also be achieved with other N-halo-compounds. In fact some of them have been used in water treatment in order to minimize the formation of trihalomethanes and other halogenated byproducts. 2 N-Cl- and N-Br-succinimide readily halogenate amines by direct halogen transfer between the N atoms.16 Several other N-halo-compounds e.g.N-Cl-toluenep-sulfonamide N-Cl-saccharine etc. have been used for this purpose but it is unclear if direct halogenation takes place or whether the N-halo-compound first undergoes hydrolysis to the corresponding HOX which is then responsible for the halogenation. Although bromination by HOBr/BrO2 would be expected to be very similar to chlorination by HOCl/ClO2 it has been proposed that bromination also takes place via a reaction between the unprotonated amine and BrO2.17 ClO2 is frequently used in water treatment and is sometimes preferred to aqueous chlorine since it leads to fewer chlorinated by-products and no CHCl3. The reaction is first order in both ClO2 and the N-compound.The observed rate constant shows a marked dependence on the pH of the medium.18 Both the free and the protonated amines yield the corresponding N-Clderivatives. The protonated form is ca. four orders of magnitude less reactive toward chlorine dioxide than the free form the chlorination of the latter by ClO2 being much slower than by HOCl/OCl2 (see Fig. 2). a units). 2.2 Dihalogenated amino compounds Dihalogenated compounds are formed when the concentration of the halogenating agent is higher than that of the N-compound. If this ratio is two or higher the dihalogenated compound is the main product.8 The reactions suffered by monohalogenated compounds following their formation complicate the study of the generation of the dihalogenated ones.It would be expected that dihalogenation taking place by reaction between the monohalogenated substrate and a halogenating agent follows a mechanism similar to that previously described for the initial halogenation. As shown in Fig. 2 the reaction is much slower due to the remarkably lower basicity of the N-halo-substituted amino group relative to amines (about 8 pK Mixed dihalo-amines,10 i.e. (N-Br N-Cl)-methylamine have been observed when N-Cl-amines react with HOBr. The mechanism is expected to be similar to that previously mentioned (vide supra). Often dihalo-compounds are generated by disproportionation of two monohalamines [eqn. (13)].19 The observed rate constants for such reactions vary with the acidity of the medium the maximum of the bell shaped pH-dependence giving an estimation of the pKa of the protonated N-halo-amine.The rate determining step for these processes seems to be the first step in eqn. (13) i.e. the direct halogen transfer between N-halo-amines followed by fast deprotonation of the dihaloamine. (13) RNXH2 + (aq) + RNXH(aq) ? RNX2H+(aq) + RNH2(aq) ?RNX2 (aq) + RNH3 + (aq) 3 Decomposition of N-halo-compounds The main efforts in this field have been devoted to the decomposition of N-Cl and N-Br-compounds. N-I-compounds have received much less attention due to the difficulty of their direct generation together with their expected higher instability. Once formed N-halo-compounds 5 can undergo several processes depending on the conditions of the medium (Scheme 2).Generally speaking the products of such reactions are carbonyl compounds 6 7 halide ions X2 and ammonia/amines 8 and/or cyano derivatives 9. Here we classify the possible processes in terms of types of reactions rather than considering their products. Chemical Society Reviews 1998 volume 27 456 R4 R2 R3 = CO2 O R1 R4 R4 R2 N B X R4 = H R3 O R2 5 B R1 = X R2 = R4 = H 7 R3 C N + 2X– + 2BH+ Scheme 2 3.1 Grob fragmentation Most studies20 have concentrated on the decomposition of N-halo-a-amino acids due to their environmental relevance in near-neutral conditions where the anionic form 10 is the main species. As shown in Scheme 3 the Grob fragmentation yields R4 R4 R2 R2 – CO2 + N N R1 X X R1 12 10 R4 R2 X– + 6 O O– O O– N+ R1 13 Scheme 3 an aldehyde or ketone 6 with one carbon fewer than the parent amino acid carbon dioxide ammonia or primary amines 8 and halide ions (Strecker degradation).The decomposition takes place in two consecutive steps. First a unimolecular ratedetermining fragmentation of the N-halo-amino acid 10 to halide CO2 and an imine 11 takes place followed by a fast hydrolysis of the latter to the corresponding amine 8 and carbonyl compound 6. The process is first order in the N-halo-amino acid 10 independent of the ionic strength and pH and faster as the solvent polarity decreases. Analysis of the effect of alkyl substituents both on the Ca and on the N and a comparison of the behaviour of N-Cl- and N-Br-a-amino acids allow the structure of the transition state to be described,21 using a More O’Ferrall diagram.13 As shown in Fig.5 the Grob fragmentation is a nearly synchronous concerted DEDN process with a product-like transition state (shaded zone). This is supported by the observation of a bigger inductive effect for the alkyl substituents on the Ca than for the nucleofuge in the N. In the transition state the electrofuge and the nucleofuge (CO2 and X2) adopt an antiperiplanar configuration and the Ca–CO22 bond breaking is slightly ahead of the N–X bond breaking. The simplicity of the mechanism involved in the decarboxylation of N-Cl-a-amino acids has been used to test the phenomenological theory of solvent effects on chemical reaction rates.22 At lower pH values N-halo-a-amino acids can also occur in three protonation states other than the anionic.Grob fragmentation is observed for such species if they are derived from secondary amino acids in which case the disproportionation (Section 2.2) becomes rather slow. If pH < 5 the carboxylate is protonated and the same is true for the amino group when [H+] > 1.0 mol dm23. The neutral form of N-Cl-a-amino acids is favoured relative to the zwitterionic form by a factor of ca. 105 contrary to the case of a-amino acids. + R1NH2 + R3 + X– 8 + R1NH2 + BH+ + X– 8 9 R2 R4 + CO2 + X– slow N R1 11 H2O fast R2 R4 + R1NH2 8 O 6 Fig. 5 More O’Ferrall plot for the Grob fragmentation of N-halo-a-amino acids Under acidic conditions all four protonation states undergo fragmentation.23 As expected the decomposition of the zwitterion is faster and for the neutral 14 and the protonated species 15 ‡ ‡ O O R2 R2 O R3 O R3 R1 N+ N Cl H Cl H R1 H 15 14 a cyclic transition state has been proposed.N-Halo-amino alcohols 16 also undergo fragmentation,24 the electrofuge and nucleofuge being respectively formaldehyde and halide as shown in Scheme 4. The 2OH group is only significantly O– R1 N R1 N CH2 + CH2O + X– X 16 H2O R1 NH2 + CH2O 8 17 fast Scheme 4 deprotonated in strongly alkaline medium. Under such conditions an imine 17 is formed which then quickly hydrolyses to the final products.The lack of precise pKa measurements for the hydroxylammonium ions prevents good estimations of the first order rate constants. Nevertheless considering the values available in the literature this fragmentation is some three orders of magnitude faster for N-Cl-alcoholamines than for similar N-Cl-a-amino acids. 3.2 Elimination In the presence of bases N-halo-amines 18 undergo an elimination process according to Scheme 2. The reaction products are similar to those obtained in the Grob fragmentation. Two different elimination processes have been identified in the decomposition of N-halo-amines an intermolecular elimination (well known) and an intramolecular one (recently proposed). 3.2.1 Intermolecular elimination in halo-amines The reaction follows a second order kinetic law first order in both the N-halo-compound 18 and the base.25 It shows general base catalysis and a marked effect of the ionic strength which in some cases has led to misinterpretations of the kinetic data.Scheme 5 displays the detailed mechanism a first step involves B R1 H R1 R2 R1 B – BH+ + N N X X R3 R3 R3 R2 R2 19 18 20 fast H2O H N + + R1NH2 R2 8 O 7 slow R1 X R2 R3 ‡ X– + B R1 N H R2 R3 •• N + BH+ + X– 457 21 Scheme 5 proton abstraction from the Ca to the N and expulsion of X2 an imine 19 being formed. In a subsequent step the imine 19 undergoes a fast hydrolysis to ammonia/amine 8 and to the corresponding carbonyl compound 7 (an a-keto acid in the case of N-halo-a-amino acids).The base-promoted generation of imines has been extensively studied26 for many substrates and leaving groups even in the case of N-Cl-amines.27 Most of the available studies have been carried out in mixed and non-aqueous media. Such studies stress the effect of steric hindrance on the reactivity and support an AxhDHDN concerted mechanism (E2 according to the Ingold nomenclature). The studies in aqueous solution are also in agreement with such behaviour. In the case of N-haloa-amino acids studies of the Brønsted and pseudo-Brønsted plots (b blg),13 kinetic isotope effects crossed-interaction parameters etc. allow the process to be identified as an asynchronous concerted AxhDHDN elimination.25 The N–X bond breaking is generally ahead of the C–H bond breaking.The process shows an important Thornton (perpendicular) effect,13 shown in Fig. 6 more important as the alkyl substituent on the Ca gets bulkier. Parallel to this effect the TS 22 changes its structure from slightly carbanion-like to nitrenium-like. Fig. 6 More O’Ferrall–Albery–Jencks plot for the base promoted elimination of N-halo-a-amino acids R3 22 The decomposition of N-halo-dipeptides follows the same general pattern already described. However some relevant differences are observed. Using NH3 (g) and Cl2 selective electrodes the formation of the imine 19 and its decomposition becomes evident as shown in Fig. 7. NH3 production does not start until the concentration of Cl2 is close to its maximum.As with other N-halo-amines the first step in the basepromoted decomposition of N-halo-dipeptides (Scheme 5) is an Chemical Society Reviews 1998 volume 27 Fig. 7 Spectrophotometric Cl2 NH3 kinetic profiles in the base promoted decomposition of N-Cl-Ala-Gly AxhDHDN process in this case with a reactant-like transition state,28 and the second step the hydrolysis of the imine 19 is slower than for N-halo-amines and N-halo-amino acids. No detailed mechanistic studies are available for the decomposition of (N,N)-dihalo-amines 23. In the simpler cases the experimental evidence suggests the existence of two consecutive dehydrohalogenation processes yielding eventually the corresponding cyano derivative 9 as shown in Scheme 6.The intermediacy of an N-halo-imine 24 has been clearly demonstrated. 8 Although detailed kinetic and mechanistic studies on this process would be desirable two concerted base-promoted eliminations are likely to be involved in the mechanism. If a second dehydrohalogenation is not possible the hydrolysis of the N-halo-imine 24 yields the corresponding carbonyl compound 7 and NH2X a very active halogenating species.2 R2 + X– + BH+ R2 B B X R3 R3 C N + X– + BH+ R3 N R2 = H N X X 23 9 O + NH2X quickly hydrolyzes. –O H N 24 R2 H /= R3 N R3 R2 R2 X 25 O OH + R2 R3 7 Scheme 6 3.2.2 Intramolecular elimination An intramolecular elimination process has been observed24 in the decomposition of N-Cl-amino alcohols 25 a process illustrated in Scheme 7.Following the deprotonation of the OH group an intramolecular proton transfer from the Ca to the oxygen takes place. In this process an imine 26 is formed that OH + X– H2N R2 R3 27 7 26 H2O fast Scheme 7 Comparison of the reaction rate for this process with that observed for similar compounds lacking the OH allows an Chemical Society Reviews 1998 volume 27 458 estimation of the effectiveness of the intramolecular pathway. A value of 106 mol dm23 has been measured for the effective molarity,13 an unexpectedly high value for such a process which may be due to the uncertainty in the pKa of the OH of the amino alcohol. 5 Photolysis and radiolysis Some work has been done on the applications of N-centered radicals aminyl radicals and the corresponding protonated species aminium radicals.The latter are much more electrophilic and reactive than the former.29 The implication of aminium radicals in the Hoffman–Löffler–Freytag pyrrolidine synthesis (cyclization of N-halogenated amines) as well as in the chlorination of hydrocarbons and other organic compounds by chloramines was proposed long ago. Aminyl and aminium radicals can be generated in several ways i.e. electrochemistry pulse radiolysis,30 and photolysis. In the laser flash photolysis (LFP) and pulse radiolysis (PR) studies of (N-Cl,N-phenyl)- glycine 28,31 photohomolysis and reductive homolysis (shown in Scheme 8) were observed.The corresponding aminyl radical (a g-distonic radical anion) 29 and aminium radical (a g-distonic radical zwitterion) 30 are generated. +• O O O Cl N •N N H O O O – – – +H+ hn -Cl• –H+ +e– aq -Cl– 28 30 hn -e– Products +• O Cl N O– 31 6 In vivo N-halogenation 29 Products Products Scheme 8 The photoionization process is monophotonic when the excitation is performed with 193 248 and 266 nm light and biphotonic if 308 nm radiation is used. The photoionization and photohomolysis quantum yields decrease with increasing excitation wavelengths. These results have shown that N-centered radicals are generated from N-halo-compounds during UV-based water treatment. Despite the increasing interest in the synthetic use of aminium and aminyl radicals and of their possible environmental and health implications there is a lack of detailed mechanistic and kinetic studies of their behaviour.There has been much controversy about the benefits and risks of halogen-based water treatment to human health. In fact several countries are looking for safer and cheap water treatment methods. For a long time there has been speculation about the carcinogenic and/or mutagenic activity of chloramines or HOCl itself. Studies on the mutagenic activity of chloramines suggest that chloramines are unlikely to penetrate the skin and mucous membranes.32 Furthermore the high SCN2 content in human saliva acts as a reducing agent toward ingested chloramines.HOCl and chloramines are also generated in vivo. When infection or tissue injury takes place phagocytic leukocytes release certain compounds with antimicrobial and antitumor activity. Although such substances are able to fight against infection and tumor growth at the same time they provoke inflammatory tissue destruction and (maybe) carcinogenesis in other tissues. The in vivo production of HOCl involves the enzyme myeloperoxidase (MPO) which is secreted by neutrophiles (neutrophilic polymorphonuclear leukocytes PMN). In the presence of H2O2 myeloperoxidase 32 catalyzes the oxidation of Cl2 to HOCl (Scheme 9) which rapidly oxidizes MPO (Fe+3) + H2O2 HOX A + H+ X– 32 Compound I (Fe+4 O) AH AH A + H+ 33 Compound II (Fe+4 O) 34 Scheme 9 other substances present in the medium.HOCl is membrane permeable but has lower or no mutagenic activity it is very reactive (see Section 2) and the carcinogenic/mutagenic activity could be related to the chloramines so generated. It has long been known that even partial chlorination of DNA bases interferes with vital biological functions of the DNA. The reaction of HOCl with DNA is slow and chlorination of the amino groups of DNA bases causes its inactivation.33 The hydrogen bonding is lost and consequently the double strand dissociates into single strands. Reaction with NH4 + rapidly produces NH2Cl which is highly toxic and also membrane permeable. In fact it modifies isolated DNA and it has been shown to be mutagenic for Bacillus Subtilus.The mutagenic activity shown by lipophilic chloramines like monochlorohistamine is rather higher than for hydrophilic chloramines such as monochlorotaurine. Histamine facilitates the entry of leukocytes into the inflammatory site. Hence as a result of its action HOCl is generated and consequently chlorohistamine formed which could be responsible for chronic inflammation and eventually carcinogenesis. The formation of chlorotaurine is enzyme-catalyzed by myeloperoxidase rather than being formed by direct reaction with HOCl.34 Presumably this essentially inert chloramine exerts a protective effect against HOCl/OCl2 damage. In vitro experiences show that dichloramines have higher mutagenic activity than chloramines.32 As stated in Section 2.2 these are formed from chloramines at low pH values and high Cl2 concentration or at high HOCl amine ratios.There is indirect evidence of small amounts of dichloramines being generated by stimulated neutrophiles but their role in vivo is still unclear. There are many other examples of in vivo oxidative damage by chloramines. Amongst them it is worth mentioning that collagen displays greatly increased proteolytic susceptibility following chloramine and HOCl/OCl2 treatment.35 Collagen degradation by collagenase is increased three-fold after exposure to chloramines with the exception of chlorotaurine which has no effect. Hypohalous acids and haloamines produced as a consequence of leukocyte action show unequivocally an inactivating effect on the antiproteolytic effect of human a2-macroglobulin.Oxidation of Br2 by myeloperoxidase has also been observed. In the presence of amines bromamines are generated which in contrast to chloramines react with the H2O2 present in the medium. As for Cl2 and Br2 peroxidases enzymatically oxidize I2 to I2 and for example tyrosine is iodinated but it remains unclear whether the iodinating species is HOI or I2 itself. The mutagenic and/or carcinogenic effect of haloamines could also be due to the effect of secondary mutagens i.e. to the by-products. For example the decomposition of N-haloa-amino acids yields aldehydes or ketones (Section 3) which show mutagenic and/or carcinogenic activity. Research in the field of in vivo halogenation is growing offering an extremely interesting field of investigation from the basic applied and theoretical chemical and biological points of view.7 Theoretical studies All the computational studies on the reactivity of N-halocompounds are quite recent all of them referring to the Grob fragmentation mechanism (Section 3.1). The unimolecular decomposition of the anionic form of N-Cl-glycine 10 (R2 = R4 = H) was studied36 using analytical gradients and AM1 and PM3 semiempirical procedures and the ab initio RHF and UHF methods at the 4-31G 6-31G* and 6-311+G* basis set levels. Correlation effects were also analyzed at the MP2/6-31G* basis set level. Although the results are for the gas phase the overall picture agrees with the experimentally observed behaviour in solution.The possible intermediates a carbanion 12 and a nitrenium ion 13 are not stationary points on the potential energy surface. In agreement with experiment calculations point to a concerted and slightly asynchronous process starting from the antiperiplanar form the N–Cl bond breaking being ahead of the Ca–CO22 bond breaking with a product-like transition state. An analysis of the substituent effects on the transition stucture has also been carried out concluding that the size of the substituent relates to Thornton effects while their type and number relate to Hammond effects.37 8 Final comments Rather than summarizing the material presented in previous sections we highlight here the aspects that merit further careful examination.A lot of analytical work has been done on the identification detection and quantitative determination of the chlorination by-products but the picture is still far from complete. For simple substances like proline chlorination yields more than 10 products some of which are still not unambiguously identified.2 Some of these substances come partly from the chlorination of the initial decomposition products and partly from subsequent reactions of the main proline chlorination products. Halogenation of more complex substrates like humic acids produces a broad spectrum of organochlorine derivatives including the nitrogenated ones which should be identified. The same is true for the chlorination by-products of other nitrogenbased common pollutants of water like pesticides herbicides etc.Reliable analytical identifications are also needed for the products generated after UV-irradiation of N-halo-compounds and to the chemicals involved in in vivo halogenation. These analytical studies are needed for a better understanding of the reaction mechanisms for which a lot of kinetic research is still needed. Simple processes like the decomposition of (N,N)- dihalo-amines the halogenation of ureas or carbamates and the fate of the products so-generated are not known in detail or at all. A challenge for those looking for more complex mechanistic puzzles is for instance a complete description of the chlorination of humic and fulvic acids. The chlorination of some model compounds has been analysed like kynurenine 35,38 a urinary tryptophan metabolite that yields chloroform 39 chloroacetonitriles 36 40 and chloroacetic acids 37 38 (Scheme 10) all of them potentially dangerous.Although an approximate explanation for each pathway could be given the detailed mechanism of some steps remains unclear. Generally speaking the pathways from haloamines to potentially toxic compounds like trihalomethanes haloacetonitriles or haloacetic acids have no detailed mechanistic explanation. Not much attention has been paid to the reactions involving heavy metals and haloamines. A wide variety of such processes is to be expected some of which could be of great environmental concern. Radiolysis and photolysis of haloamines offer an interesting and fruitful field.The study of the reactions taking place after UV-irradiation of haloamines has just started. The increasing availability of laser flash photolysis instrumentation for the study of transient species is an opportunity for those 459 Chemical Society Reviews 1998 volume 27 Cl Cl Cl O O Cl Cl 42 41 O NH2 O O Cl OH Cl Cl OH OH Cl O Cl 37 35 Cl CHCl 38 3 C N Cl C N Cl 40 36 Scheme 10 39 willing to enter the fascinating world of short-lived intermediates. Such species are extremely interesting both from the basic and applied chemical point of view. Apart from the understanding of the reactivity of N-centered radicals they have interesting applications for example in organic synthesis. The environmental aspect should be kept in mind considering the recent trend to combine UV-irradiation and chlorination in water treatment.The corresponding by-products must be identified their reaction mechanisms explained and especially toxicological activity determined. Moreover some of the processes involved in radiolysis and photolysis could be used to model the damage to DNA induced by irradiation or by the N-centered radicals themselves. Another highly attractive developing field is the study of in vivo halogenation and effects of halogenation by-products. Some of the enzymatic and nonenzymatic mechanisms involving haloamines have been related to carcinogenesis. As already stated the study of the iodination of N-compounds has been a tough problem.A good possibility for such study is the use of the system {peroxidase/H2O2/I2}. The reactivity of haloamines with sulfur compounds deserves careful study provided mutagenic N-halo-compounds are rapidly inactivated in the presence of certain sulfur compounds. On the chemical–biological side an explanation is lacking for the protection mechanism exhibited by microorganisms like Escherichia coli which shows negative chemotaxis in gradients of different substances like for example N-chlorotaurine.39 Computational studies are at their very beginning. Only the Strecker degradation of N-Cl-a-amino acids in the gas phase has been analyzed. More complex processes like the generation of N-halo-compounds their base-promoted elimination the structure and reactivity of aminium and aminyl radicals and of course the in vivo reactions would merit further theoretical study.9 Acknowledgements We are indebted to Dr J. M. Antelo (Universidade de Santiago de Compostela Galicia España) for his collaboration at the very early stages of our research in the field of halogenation mechanisms. We are also indebted to Dr H. Maskill (University of Newcastle upon Tyne UK) for his helpful comments on this review. 10 References 1 W. Stummn and J. J. Morgan Aquatic Chemistry. Chemical Equilibria and Rates in Natural Waters John Wiley New York 3rd edn. 1996. 2 R. L. Jolley L. W. Condie J. D. Johnson S. Katz R. A. Minear J. S. Maticce and V. A. Jacobs eds. Water Chlorination Chemistry Environmental and Health Effects Lewis Publishers Michigan 1990 vol.6. Previous volumes are properly referenced in this one. 3 P. Kovacic M. K. Lowery and K. W. Field Chem. Rev. 1970 70 639. Chemical Society Reviews 1998 volume 27 460 4 R. P. Martin G. J. Martens G. Porter T. E. Graedel W. C.Keene G. W. Gribble J. Fauvarque K. R. Solomon H. Galal-Gorchev J. Miyamoto M. J. Molina H. W. Sidebottom J. A. Franklin K. Ballschmiter Ch. Rappe A. Hanberg R. Papp G. Menges and A. E. Fischli Pure Appl. Chem. 1996 68 1683. 5 F. A. Cotton and G. Wilkinson Advanced Inorganic Chemistry Wiley New York 4th. edn. 1986. 6 J. Arotsky and M. C. R Symons Quart. Rev. Chem. Soc. 1962 16 282. 7 J. E. Atwater R. L. Sauer and J. R. Schultz J. Environ. Sci. Health (A) 1996 31 1965. 8 T.C. Fox D. J. Keefe F. E. Scully and A. Laikter Environ. Sci. Tech. 1997 31 1979. 9 D. W. Margerum E. T. Gray and R. P. Huffman in Organometals and Organometalloids Occurrence and Fate in the Environment, eds. F. E. Brickman and J. M. Bellama ACS Symposium Series No. 82 pp. 278–291 American Chemical Society Washington DC 1978. 10 M. Gazda and D. W. Margerum Inorg. Chem. 1994 33 118. 11 L. Abia X. L. Armesto M. Canle L. Mª. Victoria García and J. A. Santaballa Tetrahedron 1998 54 521 and references therein. 12 X. L. Armesto M. Canle L. and J. A. Santaballa Electronic Conference on Trends in Organic Chemistry (ECTOC-1) ISBN 0 85404 899 5 eds. H. S. Rzepa J. M. Goodman and C. Leach CD-ROM The Royal Society of Chemistry Publications 1996. 13 H. Maskill The Physical Basis of Organic Chemistry Oxford University Press 1993.14 M. Wayman and E. W. C. W. Thomm Can. J. Chem. 1969 47 2561. 15 J. M. Antelo F. Arce M. Parajó A. I. Pousa and J. C. Pérez-Moure Int. J. Chem. Kin. 1995 27 1021. 16 J. M. Antelo F. Arce J. Crugeiras and M. Parajó J. Phys. Org. Chem. 1997 10 631. 17 J. E. Wajon and J. C. Morris Inorg. Chem. 1982 21 4258. 18 J. Hoigné and H. Bader Wat. Res. 1994 28 45. 19 J. M. Antelo F. Arce M. C. Castro J. Crugeiras J. C. Pérez Moure and P. Rodríguez Int. J. Chem. Kinet. 1995 27 703. 20 V. C. Hand M. P. Synder and D. W. Margerum J. Am. Chem. Soc. 1983 105 4022. 21 X. L. Armesto M. Canle L. M. Losada and J. A. Santaballa J. Org. Chem. 1994 59 4659 and references therein. 22 J.M. LePree and K. A. Connors J. Pharm. Sci. 1996 85 560. 23 X. L. Armesto M. Canle A. M. Gamper M. Losada and J. A. Santaballa Tetrahedron 1994 50 10509. 24 X. L. Armesto M. Canle P. Carretero M. V. García and J. A. Santaballa Tetrahedron 1997 53 2565. 25 X. L. Armesto M. Canle M. V. García M. Losada and J. A. Santaballa J. Phys. Org. Chem. 1996 9 552. 26 R. V. Hoffman R. A. Bartsch and B. R. Cho Acc. Chem. Res. 1989 22 211. 27 Q. Meng and A. Thibblin J. Am. Chem. Soc. 1997 119 1224. 28 X. L. Armesto M. Canle M. V. García and J. A. Santaballa Tetrahedron 1997 53 12615. 29 B.D. Wagner G. Ruel and J. Lusztyk J. Am. Chem. Soc. 1996 118 13. 30 J. Lind M. Jonsson T. E. Eriksen G. Merényi and L. Eberson J. Phys. Chem. 1993 97 1610. 31 M. Canle L. J. A. Santaballa and S. Steenken submitted to Eur. J. Chem. See also Book of Abstracts Fast Reactions in Solution Meeting Group (FRIS’97) P11 Copenhagen 1–4 September 1997 and references therein. 32 E. L. Thomas M. M. Jefferson J. J. Bennett and D. L. Learn Mutat. Res. 1987 188 35 and references therein. 33 W. A. Prütz Arch. Biochem. Biophys. 1995 332 110. 34 L. A. Marquez and H. B. Dunford J. Biol. Chem. 1994 269 7950. 35 J. M. S. Davies D. A. Horwitz and K. J. A. Davies Free Radical Biol. 36 J. Andrés J. J. Queralt V. S. Safont M. Canle and J. A. Santaballa J. 37 J. Andrés J. J. Queralt V. S. Safont M. Canle and J. A. Santaballa J. 38 H. Ueno T. Moto T. Sayato and K. Nakamuro Chemosphere 1996 33 39 L. Benov and I. Fridovich Proc. Natl. Acad. Sci. USA 1996 93 Med. 1993 15 637. Phys. Chem. 1996 100 3561. Phys. Org. Chem. 1996 9 371. 1425. 4999. Received 20th April 1998 Accepted 7th July 1998

ISSN:0306-0012    

DOI:10.1039/a827453z

出版商:RSC    

年代:1998

数据来源: RSC