摘要:

J. Chem. Soc., Perkin Trans. 2, 1999, 389–391 389 The first electrochemical study of epidithiopiperazine-2,5-diones, a special class of ·,·9-disulfide bridged cyclic dipeptides Christina L. L. Chai,*a Graham A. Heath,b Paul B. Huleatt a and Gavin A. O’Shea a a Department of Chemistry, The Faculties, Australian National University, ACT 0200 b Research School of Chemistry, Institute of Advanced Studies, Australian National University, ACT 0200 Received (in Cambridge) 26th November 1998, Accepted 28th January 1999 Polarographic studies of 3,6-epidithiopiperazine-2,5- diones (ETP) including naturally occurring gliotoxin and simple synthetic analogues are pertinent to understanding their biological action; coulometric measurements on 1,4- diethyl ETP in acetonitrile establish the existence of a ready one-electron reduction overall, in sharp contrast to the familiar two-electron cleavage of acyclic disulfides.Epidithiopiperazine-2,5-diones (abbreviated as ETP) 1 are an important class of biologically active compounds.1 The fungal metabolite gliotoxin 2 displays antitumor, antiviral and immunosuppressive properties whereas sporidesmin 3 is the causative agent of facial eczema in ruminants.Comparable patterns of activity persist in certain simple synthetic analogues such as 1,4-dimethyl and 1,4-diethyl ETP.2,3 One proposal 4 to account for this behaviour envisages redox cycling of the disulfide to the dithiol piperazine-2,5-dione, in a 2-electron/2H1 process coupled with a cellular reductant such as glutathione (see Scheme 1).However little is known directly of the redox properties of this class of compounds so even an empirical correlation with biological activity remains to be established. As part of a broad investigation of the mode of action of epidithiopiperazine-2,5-diones,3 we now report the voltammetric behaviour of gliotoxin (2), selected synthetic analogues (4–7) and a structural variant (8).Epidithiopiperazine-2,5- diones 4–7 were prepared by modifications of known procedures 2,5 and 1,4-diacetylcystine anhydride (8) by acetylating cystine anhydride.6 These studies suggest that the bridgehead N N O O R2 S2 R3 R4 R1 4 R1 = R2 = Et 5 R1 = Bn, R2 = Me 6 R1 = R2 = Bn 7 R1 = R2 = p-MeOBn 2 3 8 1 R1 N N O O R2 S2 N N O O CH2OH OH S2 N N O O S2 N MeO MeO Cl OH OH H N N O O Ac Ac S2 disulfide moiety in ETP compounds is exceptionally easy to reduce, and that one-electron reduction is a special property of ETP compounds.To our knowledge, this is the first reported electrochemical study of this particular class of disulfides. After preliminary trials with other working electrodes, the voltammetric behaviour of compounds 2, 4–8 in acetonitrile was studied by dc and ac polarographic techniques. The halfwave potentials are summarized in Table 1. Reduction of the synthetic ETP compounds 4–7 in acetonitrile takes place readily on mercury, between ca. 20.4 to 20.6 V, although a complex wave with two components is observed in each case. Typical ac and dc polarograms are shown in Fig. 1. Rather than the two components representing successive oneelectron steps, detailed examination established one-electron reduction of the ETP compound overall (see below). The shape of the first feature suggests specific adsorption of the ETP compounds 4–7 on the negatively polarised electrode. For gliotoxin (2), the first two cathodic processes are similar to those of the synthetic ETP compounds, though the ac peaks are not fully resolved.Gliotoxin also displays a strong broad polarographic maximum at ca. 20.9 V, apparently unrelated to its primary reduction. In contrast to all the ETP compounds, no reduction of 1,4-diacetylcystine anhydride (8) is observed within the scan limit (22.0 V) under these conditions. The choice of acetonitrile enabled us to compare a range of simple organosoluble derivatives and to explore the partici- Scheme 1 S S SH SH cellular reductant e.g.AH2 A 2O2 2O2 – 1 Table 1 Redox properties of cyclic disulfides at the dropping mercury electrodea,b Compound Ferrocene (internal reference) 1,4-Diethyl ETP (4) 1-Benzyl-4-methyl ETP (5) 1,4-Dibenzyl ETP (6) 1,4-Di(p-methoxybenzyl) ETP (7) Gliotoxin (2) c 1,4-Diacetylcystine anhydride (8) Peak 1/V 10.55 20.38 20.39 20.43 20.50 20.45 not observed Peak 2/V 20.60 20.53 20.53 20.58 20.52 (shoulder) not observed a E1/2 values vs.Ag/AgCl/Cl2 in acetonitrile, recorded as ac peak potentials. b Polarographic conditions: scan rate 10 mV s21, drop time 0.5 s, ac frequency 205 Hz. c Gliotoxin has dc E1/2 values of 20.37 and 20.48 V, with a strong broad maximum at 20.9 V.390 J. Chem. Soc., Perkin Trans. 2, 1999 389–391 pation of water. These conditions may mimic the biological environment of ETP compounds as there is much evidence that ETP compounds are readily taken up by cells.7 Progressive addition of water (up to 2% by volume, i.e.ca. 100 × the concentration of ETP) did not greatly aVect the observations or undermine the pattern emerging in Table 1. Further work should be directed to the comparison of i–E curves of 4–7 in aqueous media and to developing water soluble analogues. The reduction of 1,4-diethyl ETP (4) on the dropping mercury electrode was examined in greater detail. The impression that wave 1 is due to adsorption was reinforced by concentration- dependence studies and by the observation that low levels of added Li1 completely suppress the first component in both ac and dc scans, leaving component 2 unchanged. Cyclic voltammetric experiments on ETP 4 at a stationary mercury electrode (hanging mercury drop electrode, HMDE) confirm that component 1 is tensammetric in nature, while component 2 represents a reversible Faradaic reduction (Fig. 2).This pattern of behaviour (including inhibition by cations) is in accord with specific adsorption of the ETP compound on the negatively polarised electrode surface (in the 20.3 to 20.5 V domain).8 This, along with observations on the shape of the curves,† suggests that the first wave is characteristic of an adsorption phenomenon, whereas the second wave is that resulting from a ‘true’ reduction process.In accord with this, attempted coulometry at the Hg pool polarised at 20.5 V leads to complete current decay within seconds with very obvious simultaneous development of a dark, apparently insulating patina on the mercury surface.The adsorption of disulfides and diselenides onto mercury electrodes is widely documented, though there is no consensus on the chemical formulation of the adsorbed species.9 Coulometry of 4 at a Hg-pool polarised at 21.1 V, i.e. well beyond the second step in the overall reduction, consumes only Fig. 1 Ac and dc polarograms of 3,6-epidithio-1,4-diethylpiperazine- 2,5-dione (4). Fig. 2 CV of 3,6-epidithio-1,4-diethylpiperazine-2,5-dione (4) (scan rate of 500 mV s21).one Faraday per mole of ETP compound. This unexpected outcome is consistent with initial formation of a radical anion disulfide species (as opposed to the more obvious RSSR/2 × RSH couple). Although no definitive evidence was obtained to support this formulation, a persistent orange–red colour of the homogeneous electrolysed solution was observed. This sensitive solution was immediately decolourised on exposure to air.Pulse radiolysis studies on a number of dialkyl disulfides have shown that [RSSR]2? species can be generated and characterised spectrophotometrically by broad absorption maxima at ca. 410 nm.10 In particular, the cyclic disulfide radical anion of lipoic acid has a lifetime of the order of 0.1 millisecond; 11 the present ETP structure is much more conducive to achieving the >100- fold increase in lifetime implicit in our reversible ac polarography and CV results.Polarographic analysis after bulk electrolysis revealed the disappearance of the original ETP cathodic processes, and the emergence of a new anodic wave of comparable height near 10.1 V. The oxidation of cis-1,4- diethyl-3,6-dithiolpiperazine-2,5-dione (the dithiol related to ETP 4) was measured separately and was found to occur at 10.65 V at the dropping mercury electrode. Accordingly, the ETP compounds described here are reduced more readily than simple dialkyl disulfides which generally display a two-electron polarographic process well beyond 21.0 V.12 This may reflect the atypical conformation and more strained S–S bond in ETP compounds13 compared to acyclic analogues (S–S bond length 2.08 Å vs. 2.04 Å; CSSC dihedral angle of ca. 108 vs. 908). The cyclic disulfide system of 8, by contrast, is relatively normal in conformation (for the parent cystine anhydride, S–S bond length 2.00 Å; CSSC dihedral angle 908).14 It appears that the inherent constraint on the ETP disulfide bridge has two (exquisitely balanced) consequences i.e.to weaken the bond by ~0.04 Å such that the s* level is lowered in energy, making formation of the hemi-bonded disulfide radical more accessible, and to protect the (albeit weak) hemi bond from heterolytic cleavage, or further reduction, by the enforced apposition of the neighbouring atoms. This suggests that the original 2e/2H1 redox cycling scheme (Scheme 1) deserves further consideration, with possible participation at a microscopic level of a one electron process.These results illustrate the practical utility of the Hg electrode in eliciting an electro-analytical response for ETP derivatives, despite the attendant complications which are often encountered in the polarography of organo-thio compounds.9,15 The present studies indicate that the reduction processes for these five ETP compounds are broadly similar to one another, notwithstanding the more marked biological activity of gliotoxin itself.3 There is room to believe that redox cycling is fundamental to their common mode of action, while attendant molecular features of 2 vs. 4–7 lend selectivity to gliotoxin’s role in the physiological cycle. Experimental Polarographic and coulometric experiments were performed using a PAR 170 Electrochemistry System coupled to Metrohm 505 and 663 electrode stands using a Pt-wire counter electrode, a Ag/AgCl reference electrode, and tetrabutylammonium tetrafluoroborate (recrystallized ex methanol) as the supporting electrolyte in acetonitrile.The purified solvent was distilled from calcium hydride under nitrogen, immediately before use. Typical experiments utilised 2 × 1023 M solution of the ETP derivative in 10 ml of CH3CN–Bu4NBF4 (0.1 M) electrolyte, stringently purged of oxygen. Polarograms were recorded in the range 0.0 V to 22.0 V (vs. Ag/AgCl) and also calibrated with ferrocene. Multiple coulometric measurements on 25 mg of ETP 4 yielded concordant values of 1.0 Faraday per mole.Acknowledgements The authors thank Dr P. Waring for the gift of gliotoxin and the Australian Research Council for financial support of this project.J. Chem. Soc., Perkin Trans. 2, 1999, 389–391 391 Notes and references † We thank a referee for encouraging us to explore this issue further. 1 A. Taylor, in Microbial Toxins, ed. S. Kadis, A. Ciegler and S. Ajl, Academic Press, New York, 1971, Part VII, p. 337–371; T.W. Jordan and S. J. Cordiner, Trends Pharmacol. Sci., 1987, 8, 144; P. G. Sammes, Fortschr. Chem. Org. Naturst., 1975, 32, 51; P. Waring, R. D. Eichner and A. Müllbacher, Med. Res. Rev., 1988, 8, 499; P. Waring and J. Beaver, Gen. Pharmacol., 1996, 27, 1311. 2 P. W. Trown, Biochem. Biophys. Res. Commun., 1968, 33, 402. 3 A. M. Hurne, J. Simpson, P. Waring and C. L. L. Chai, Bioorg. Med. Chem. Lett., 1997, 7, 2645. 4 R. Munday, J. Appl. Toxicol., 1987, 7, 17. 5 H. Poisel and U.Schmidt, Angew. Chem., Int. Ed. Engl., 1971, 10, 130. 6 B. Kamber, Helv. Chim. Acta, 1971, 54, 927. 7 P. Waring, N. Newcombe, M. Edel, Q. H. Lin, H. Jiang, A. Sjaarda, T. Piva and A. Müllbacher, Toxicon., 1994, 32, 491. 8 B. Breyer and H. H. Bauer, in Chemical Analysis, New York, eds. P. J. Elving and I. M. KolthoV, Interscience Publishers, 1963, vol. 8. 9 B. Nygård, Ark. Kemi, 1967, 28, 75; B. Nygård, Ark. Kemi, 1967, 28, 89; B. Nygård and L. Schotte, Acta Chem. Scand., 1956, 10, 469; J. Ludvík and B. Nygård, J. Electroanal. Chem., 1997, 423, 1. 10 R. L. Wilson, J. Chem. Soc., Chem. Commun., 1970, 1425. 11 K.-D. Asmus, in Methods in Enzymology, Academic Press, New York, eds. L. Packer and A. N. Glazer, 1990, vol. 186, p. 168; M. Z. HoVman and E. Hayon, J. Am. Chem. Soc., 1972, 94, 7950. 12 See for example M. E. Hall, Anal. Chem., 1953, 25, 556. 13 J. Fridrichsons and A. McL. Mathieson, Acta Crystallogr., 1965, 18, 1043; J. Fridrichsons and A. McL. Mathieson, Acta Crystallogr., 1967, 23, 439. 14 H. Mez, Cryst. Struct. Commun., 1974, 3, 657. 15 For early examples, see I. M. KolthoV and C. Barnum, J. Am. Chem. Soc., 1941, 63, 520; D. G. Davis and E. Bianco, J. Electroanal. Chem., 1966, 12, 254. Communication 8/09266D

ISSN:1472-779X    

DOI:10.1039/a809266d

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

J. Chem. Soc., Perkin Trans. 2, 1997 397 Site selectivity in self-catalysed functionalization of helical polypeptide structures Klas Broo,a Malin Allert,a Linda Andersson, Pernilla Erlandsson,a Gunnar Stenhagen,b Joakim Wigström,a Per Ahlberg *,a and Lars Baltzer *,a a Department of Organic Chemistry, Göteborg University, 412 96 Göteborg, Sweden b Department of Organic Chemistry, Chalmers University of Technology, 412 96 Göteborg, Sweden Histidine side chains in helical structures catalyse the acylation of flanking lysine, ornithine and 1,3-diaminobutyric acid residues provided they are in positions i 2 3 and i + 4, but not in positions i 2 4, i 2 1, i + 2, i + 3, relative to a histidine in position i, in a novel site-selective functionalization reaction that enhances the potential of polypeptide and protein design.We recently reported that a histidine side chain catalyses the site-selective acylation of a flanking ornithine residue in a helical structure [eqn.(1)] using mono-p-nitrophenyl fumarate.1 The mechanism was determined2 and it was found that an initial rate-limiting attack of the deprotonated form of the histidine side chain to form an acyl intermediate was followed by a fast intramolecular transfer of the acyl group to form an amide at the side chain of the flanking Orn. The reaction was first discovered in a designed polypeptide with 42 amino acid residues, RA-42, that folds into a helix-loop-helix dimer.1 To explore the generality of the reaction we have further investigated the selectivity and reactivity in other helix-loop-helix motifs with 42 amino acid residues and in model peptides with 20 residues that in solution form helical structures.All model peptides were derived from helix I in RA-42 (Fig. 1). We now wish to report that the site selectivity in this reaction is high, that only two sites in the vicinity of the histidine are acylated and that the flanking residue may be Lys, Orn or 1,3- diaminobutyric acid (DAB).His-11 catalyses the acylation of Lys-15, Orn-15, DAB-15 and Lys-8, but not Lys-7, Lys-10, Lys- 13 and Lys-14 (Fig. 1). Positions 9 and 12 were not investigated since they were part of the hydrophobic core of the amphiphilic helices. So far, site selectivity has been explored by us in helices, although b sheets can also be expected to function as templates in design and functionalization. The helical content is conveniently assessed by CD spectroscopy 3 and the mean residue ellipticity of the model peptides under reaction conditions was between 29 000 and 220 000 deg cm2 dmol21, which corresponds to 25–60% of helix.4,5 Typical reaction conditions for model peptides with 20 amino acids and for helix-loop-helix motifs 1 were 0.5–1 mM concentration of peptide in 10 vol% 2,2,2-trifluoroethanol at 290 K and pH 5.85.The trifluoroethanol solution was needed to ensure helical conformation for the shorter peptides.6 The mean residue ellipticity of the helixloop- helix dimers had larger negative values than 219 000 deg cm2 dmol21 corresponding to more than 60% helix.In all peptides His-11 was flanked by DAB, Lys or Orn residues in positions 7, 8, 10, 13, 14 or 15. The second-order rate constants were determined in helix-loop-helix motifs for some flanking residues. A histidine residue in a helix-loop-helix dimer that was not flanked by Arg, Lys, Orn or His was previously shown to catalyse the formation of p-nitrophenol with a second-order rate constant of 5.3 × 1023 M21 s21 in 10 vol% TFE at pH 5.85.The rate constant correlated very well with that of 4-methylimidazole 1 (3.4 × 1023 M21 s21) if the difference in pKa and a Brønsted coefficient b of 0.8 7 for nucleophilic catalysis of ester hydrolysis was taken into account.2 RA-42, a polypeptide with 42 amino acids that folds into a helix-loophelix motif in solution, where Orn-15 flanks His-11, reacts with a second-order rate constant of 5.1 × 1022 M21 s21 in aqueous solution at pH 5.8 and 290 K.LA-42 and MA-42 differed by only one amino acid residue from the sequence of RA-42. In the former, Lys-15 replaced Orn-15 and in the latter Arg-15 replaced Orn-15. The second-order rate constants were Fig. 1 Schematic representation of helical structure illustrating the geometric relationship between His-11 and positions that are acylated by mono-p-nitrophenyl fumarate. The amino acid sequence is that of helix I in RA-42 given in three-letter code.Aib is a-aminoisobutyric acid and Nle is norleucine. All residues that surround His-11 in space have been searched for possible acylation except residues that form part of the hydrophobic core. Only in positions 8 and 11 are DAB, Lys or Orn acylated. The residues shown are those from Ala-3 to Lys-19 and Ala-8 is underscored to indicate that a lysine in that position would have been acylated.398 J. Chem. Soc., Perkin Trans. 2, 1997 Fig. 2 Schematic representation of the result of trypsin cleavage of RA-42 after reaction with mono-p-nitrophenyl fumarate. Residues 1–19 and 24– 42 were designed to fold into helical structures and the sequence from 20–23 was designed to form the loop. Trypsin is known to cleave polypeptides on the C-terminal side of basic residues but will not cleave at that site if the basic residue has been modified. The vertical solid lines have been used to indicate cleaved peptide bonds.Vertical dashed lines indicate peptide bonds that would have been cleaved in the absence of acylation, but where no cleavage was observed. Segments Asn-1 to Lys-10, His-11 to Lys-33, His-11 to Lys-19 and F-35 to Arg-40 were found by LC ESMS showing that Lys- 10, Lys-33 and Arg-40 were not acylated. The observation of segment His-11 to Lys-33 as well as His-11 to Lys-19 shows that Lys-19 is acylated to some extent as indicated by using both a solid and a dashed vertical line.The fragments His-11 to Orn-15 or Nle-16 to Lys-33 were not found and Orn-15 is therefore acylated prior to Lys-19. 5.6 × 1022 M21 s21 for LA-42 and 3.1 × 1022 M21 s21 for MA-42. Arg, Lys or Orn in position i + 4, relative to a histidine in position i, thus increases the second-order rate constant by factors of 3–6 in aqueous solution at pH 5.85 and with factors as large as 38 in 30 vol% TFE, most likely by hydrogen bonding or other electrostatic stabilization in the transition state.The similarity in the rate increase between Arg, Lys and Orn substituted peptides was surprising since the side chains of the stabilizing residues differ in hydrogen bond donating ability (different pKa values 8) and length. In contrast, protonated flanking histidine side chains have recently been shown to provide rate enhancements of more than 103 over 4-MeIm catalysed reactions.9 The reaction rates of the shorter peptides were found to be concentration dependent and these peptides were only used for studies of reaction products.Identification of reaction products was accomplished by electrospray mass spectrometry (ESMS) or trypsin cleavage followed by LC ESMS. The acylation reaction was carried out by measuring the reaction rate for each peptide with 0.1 mM of mono-p-nitrophenyl fumarate 1 followed by the addition of 3–5 times excess of the same substrate. In RA-42 ESMS showed the predominant formation of monoacylated product.Trypsin treatment of the monoacylated reaction product followed by LC ESMS showed the formation of the fragments expected from cleavage of the peptide on the C-terminal side of the residues Lys-10, Lys-19, Lys-33, Orn-34 and Arg-40 (Fig. 2). No cleavage product was observed that corresponded to cleavage at Orn-15. In addition, one fragment (MW 1078) showed the molecular weight corresponding to the fragment His-11 to Lys- 19 (980) plus the weight of fumaric acid (116) less the weight of water (18).The formation of an amide at the side chain of Orn- 15 was thus established. Some reaction product from RA-42 also showed acylation at the side chain of Lys-19 (Fig. 2) as one fragment showed the molecular weight of the segment from His-11 to Lys-33 plus two fumaryl residues, less the molecular weight of two water molecules (Fig. 2). No other site of acylation was detected. Interestingly, 1H NMR spectroscopic investigations, a-H chemical shift indices 10 and medium-range NOEs,11 reveal that the structure of RA-42 is not helical at Lys- 19 which suggests that some sequence specific conformation is the reason for the observed acylation reaction.The difference in reactivity between Orn-15 and Lys-19 is such that for the addition of one equivalent of substrate only Orn-15 acylation was detected. The acylation of RA-42 and LA-42 also takes place in 30 vol% TFE where the helix-loop-helices do not form hairpin motifs and are monomeric.12 The reaction is therefore intramolecular.Several polypeptides were synthesised and reacted with excess substrate to explore the site selectivity of the reaction and the reaction products were analysed by ESMS or trypsin cleavage followed by LC ESMS. All peptides investigated had a histidine in position 11. LA-42 has the same sequence as RA-42 except that Orn-15 has been replaced by Lys-15. LA-42 is also acylated in position 15. The polypeptide JW-20Lys (Lys-7, Lys- 14) showed no acylation, the polypeptides KB-20 (Lys-8, Lys- 13), LB-20 (Lys-7, DAB-15) and JW-20Orn (Lys-7, Orn-15) were monoacylated and the polypeptide PB-20 (Lys-8, Orn-15) was monoacylated at Orn-15 showing that only positions 8 and 15 were functionalized by His-11 and that Orn-15 was preferentially acylated over Lys-8.Trypsin cleavage of the reaction product confirms preferential acylation of Orn-15 in the peptide PB-20. In the absence of a flanking residue that could be acylated selectivity was lost.In MA-42, a helix-loop-helix motif, excess substrate led to acylation at several sites. MA-42 has the same amino acid sequence as RA-42 except that Arg-15 replaces Orn-15. This reaction has thus been carried out in a number of different helical peptides and the reaction appears to be quite general with large potential for the incorporation of functionality in folded polypeptides and proteins. So far we have incorporated amino acid derivatives, fumaryl residues 1 and a nicotinyl residue (dehydrogenase mimic) and the incorporation of carbohydrates is currently being explored.The structural basis for the observed selectivity is probably that unstrained cyclic transitions states are possible for lysine, ornithine and DAB residues in position 15 and for lysine residues in position 8, but not for the other cases reported here. Possible areas of application are the immobilization of proteins and the introduction of nonnatural, sensitive or bulky functional groups into peptides and proteins.Surprisingly simple combinations of amino acid residues on the surface of helical structures can accomplish high selectivity and reactivity. The fact that the reaction takes place in aqueous solution as well as in TFE mixtures suggests that peptide catalysed transformations may have played a role in the prebiotic soup at the origin of life. References 1 L. Baltzer, A.-C. Lundh, K. Broo, S. Olofsson and P. Ahlberg, J. Chem. Soc., Perkin Trans. 2, 1996, 1671. 2 K. Broo, L. Brive, A.-C. Lundh, P. Ahlberg and L. Baltzer, J. Am. Chem. Soc., 1996, 118, 8172. 3 W. C. Johnson, Jr., Proteins, 1990, 7, 205. 4 Y.-H. Chen, J. T. Yang and K. H. Chau, Biochemistry, 1974, 13, 3350. 5 M. Engel, R. W. Williams and B. W. Erickson, Biochemistry, 1991, 30, 3161. 6 J. W. Nelson and N. R. Kallenbach, Proteins, 1986, 1, 211. 7 T. C. Bruice and R. Lapinski, J. Am. Chem. Soc., 1958, 80, 2265. 8 C. Tanford, Adv. Protein Chem., 1962, 17, 69. 9 K. Ottosson, L. Brive, P. Ahlberg and L. Baltzer, unpublished results. 10 D. S. Wishart, B. D. Sykes and F. M. Richards, J. Mol. Biol., 1991, 222, 311. 11 K. Wutrich, NMR of Proteins and Nucleic Acids, Wiley, New York, 1985. 12 S. Olofsson and L. Baltzer, Folding Des., 1996, 1, 347. Paper 6/05776D Received 19th August 1996 Accepted 23rd December 1996

ISSN:1472-779X    

DOI:10.1039/a605776d

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

J. Chem. Soc., Perkin Trans. 2, 1999, 399–410 399 A theoretical study of electronic factors aVecting hydroxylation by model ferryl complexes of cytochrome P-450 and horseradish peroxidase Michael Filatov, Nathan Harris and Sason Shaik* Department of Chemistry and The Lise Meitner-Minerva Center for Computational Quantum Chemistry, The Hebrew University, 91904 Jerusalem, Israel Received (in Cambridge) 1st December 1998, Accepted 15th January 1999 Density functional theory (DFT) is used to study model ferryl species of cytochrome P-450 and horseradish peroxidase (HRP), as well as of the product complex due to oxidation of H2 by the P-450 species (1–4 and 7).The ferryl species studied include neutral and cation radical states of the porphyrin, as well as high- and low-spin situations. A few issues are addressed concerning the mechanism of alkane hydroxylation, and theoretical support is provided for: (i) the contention that spin inversion occurs along the reaction path, (ii) that the cation radical state of the porphyrin is an essential feature required to accommodate an excess electron from the ferryl moiety and thereby stabilize the ground state of the hydroxylation product, and (iii) that the donor property of the proximal ligand has a significant influence on the energy of the ferryl-to-ring charge-transfer states which are essential to convert the reactant state to the hydroxylation product state.In this sense, our study sheds some light on the diVerence between the oxidized and reduced HRP forms, HRP(I) and HRP(II), and suggests that the combination of a cation radical porphyrin state and a good p-donor proximal ligand like thiolate, could be the underlying reason for the potent hydroxylation ability of the P-450 ferryl-complex.Introduction Cytochrome P-450 and horseradish peroxidase (HRP) represent two families of heme enzymes that are used in nature as a means of biological oxidation of toxic compounds.1–5 The active species in these enzyme families is considered 1,3–5 to be the iron-oxo compound, shown in Scheme 1, in which the ferryl group FeIVO is embedded into the protoporphyrin IX ring.While in P-450 oxidation there exists evidence 6,7 for an additional active species (e.g., peroxo-complex), attention here will be restricted to the ferryl-complex. An interesting feature of the ferryl moiety is its high-spin O2-type bonding with two triplet electrons occupying the p*-orbitals of FeO.1–5 The P-450 and HRP active species diVer in the identity of the sixth ligand, the so called ‘proximal ligand’, which is a nitrogen from the imidazole of a histidine residue for the HRP species and a negatively charged sulfur moiety of a cysteine residue in the P-450 species.Other members of these families are secondaryamine monooxygenase (SAMO) which is analogous to HRP, and chloroperoxidase (CPO) which is analogous to P-450, and so on. The ferryl species of HRP (Scheme 1) appears in two forms, HRP(I) and HRP(II), which diVer in the oxidation state of the porphyrin ring.Thus, in HRP(I) the porphyrin moiety is a cation radical porphyrin, while in the HRP(II) form, the porphyrin is closed-shell.4,5 The active form of P-450 has not been isolated, but model studies 1,3,4,8 suggest that it is analogous to HRP(I) with a cation radical state for the porphyrin. The species with the cation radical situation are typified by a highspin state in which the odd electron of the cation radical is ferromagnetically coupled to the triplet electrons of the ferryl moiety (S = 3/2, i.e., a quartet state).1–5 In a few studies 9 it was shown that the cation radical state is an essential ingredient, in the absence of which the oxygen transfer capability of the ferryl species diminishes.This finding appears intriguing, since bare ferryl is capable of carrying out oxygen transfer reactions in the gas phase,10 and one wonders what could be the precise diVerence between the nature of free ferryl and of the one embedded in a porphyrin ring. Thus, the ‘hole’ in the porphyrin ring appears to be a fundamental oxidative feature that requires elucidation.Among the important oxidation reactions is the hydroxyl- Scheme 1 N N N N –O2C –O2C Fe O N N H L N N N N –O2C –O2C Fe O L N N H [Por+•]FeO [Por]FeO his L= L= L=(cys)S – his HRP(I) Ferryl HRP(II) Ferryl P-450 Ferryl S=3/2 S=1 •• ••400 J.Chem. Soc., Perkin Trans. 2, 1999, 399–410 ation of alkanes to alcohols, in Scheme 2. The net eVect of the reaction is the insertion of the ferryl oxygen into the C–H bond to form the alcohol complex followed by regeneration of the resting state of the enzyme which is the iron–water complex. In this respect, P-450 is a very potent oxidant capable of hydroxylating even nonactivated C–H bonds of simple alkanes.1,3 On the other hand, HRP(I) hydroxylates only activated C–H bonds, while HRP(II) is apparently not a hydroxylating agent; again implying a key role for the ‘hole state’ of the porphyrin.9 It has been proposed that the diVerence between HRP(I) and P-450 may originate in diVerences of accessibility of the iron center due to shielding by respective proteins.1–3 Nevertheless, it is still important to ascertain whether or not any of the reactivity patterns are also associated with electronic factors due to the diVerent proximal ligands of the two species. Indeed, it was proposed in numerous studies 1,3,8b,11 that the cysteinato proximal ligand plays a significant role in the ‘O’ insertion capability of P-450 into nonactivated C–H bonds.Generally speaking, electronic eVects due to proximal ligands are well documented in epoxidation reactions by model compounds,12 and hence the role of the proximal ligand is another intriguing issue which merits attention. A third fundamental issue is concerned with the nature of the states which participate in the hydroxylation mechanism.In a Scheme 2 N N N N –O2C –O2C Fe O L N N N N –O2C –O2C Fe O L N N N N –O2C –O2C Fe O L H N N N N –O2C –O2C Fe O L H H + R-H (RH) R + • + • H2O R-OH series of studies on the reactivity of bare ferryl, it was shown that the high-spin hydroxylation surface is typified by a high barrier and generates an addition adduct which is not very stable.13 Consequently, the high-spin surface is crossed by a lowspin surface which is initially an excited state of the ferryl, and which is typified by low barriers and a stable hydroxylation adduct.The existing evidence in the area of P-450 hydroxylation suggests a similar mechanism in at least one case where spin state information is available for the hydroxylation product. Thus, whereas the ground state of the active species 1–5 is high-spin, the product-complex resulting from hydroxylation of camphor is low-spin.14 Indeed, a hydroxylation mechanism based on a two-state-reactivity (TSR) paradigm has been recently proposed for P-450 hydroxylation,15 in which a key role is played by the transitions between high- and low-spin potential energy surfaces of the reaction system, as shown in Fig. 1. All these pieces of evidence, taken together, indicate that even a minimalistic approach would require the consideration of two spin states of the ferryl species, for any hydroxylation mechanism.4,6,16 The above three fundamental issues form a broad topic which requires an equally extensive program before any satisfactory resolution of the problems can be achieved.As a first step we present in this paper density functional theoretical studies of various spin and state situations for model P-450 and HRP reactant species with closed-shell and radical-cationic porphyrin situations, as well as of a hydroxylation-product species (1–4, and 7 in Scheme 3 later). This paper provides supporting theoretical evidence for a TSR reactivity scenario in which the cation radical state of the porphyrin is required to stabilize the ground state of the hydroxylation product, and in which the proximal ligand aVects the energy of the intramolecular charge-transfer states which fill the hole and eventually correlate to the hydroxylation product state.Theoretical methods and strategy Methods Our method of choice is density functional theory (DFT), which has become a powerful tool for investigating transition metal compounds.17 The calculations presented later have been done with the unrestricted Kohn–Sham method and the BP86 density functional 18 as implemented in the CADPAC519 suite of programs.This functional which includes nonlocal exchange and correlation corrections has already proven successful in the study of the various state and spin situations of an HRP complex and other ferryl complexes.20 A basis set of double-zeta quality with polarization functions (DZVP2) especially opti- Fig. 1 Qualitative energy profiles following reference 15 showing a possible two-state-reactivity (TSR) situation for hydroxylation of alkane by P-450 ferryl complex. The TSR refers to the spin-state crossing along the reaction coordinate. NN N Fe N O L H H NN N Fe N O L low-spin high-spin Ferryl + RH Products / R-OH / R-H ROH complex TSR-MechanismJ. Chem. Soc., Perkin Trans. 2, 1999, 399–410 401 mized for DFT calculations 21 was used on the metal, while for the coordination sphere of iron we used the 6-31G* basis,22 and the STO-3G basis 22 on the C and H porphyrin atoms.Other studies performed by us on some of the species, with FT97 23 as well as BP86 with larger basis sets 24 gave virtually the same results, but were abandoned for reasons of computer time economy. All the structures were obtained by full geometry optimization, with the exception of a D4h constraint of the porphyrin ring. Below we describe the ground and excited state species which have been calculated, and provide necessary theoretical details for the calculation of the excited states using DFT.Strategy Model species. Scheme 3 shows, in 1–4, models of the active species of P-450, HRP(I) and HRP(II), where the protoporphyrin substituents and proximal ligands have been simplified by necessity to save computer time and enable thereby the geometry optimization. Thus, the N-ligation of imidazole is simpli- fied initially to NH3 in 2, and then to H2C]] NH in 3; the latter ligand has p-orbitals which can interact with the porphyrin in a manner at least analogous to imidazole.The cysteinato ligand of P-450 is simplified to HS2 in 4. In view of the fact that the cysteinato moiety bears two electron attracting substituents (NH3 1 and CO2H), it was felt that HS2 which possesses a higher electron aYnity can serve as a better model than CH3S2. The choice is not crucial for the present paper, though in general a judicious modeling of the ligand is important.Compound 1 serves as a reference in which the ferryl-complex is devoid of a proximal sixth ligand. All the species in 1–4 are studied in two oxidation states which will presumably give rise to closed-shell as well as cation radical porphyrin states. Furthermore, for all the species we also screened some excited states which may be relevant for the hydroxylation process. Also shown in Scheme 3 are O2 (5) and FeO21 (6) which serve as bonding analogs for the ferryl moiety in the ferrylcomplexes. 15 Note that FeO21 in 6 has a formal oxidation state FeIV as in the porphyrin complexes.1–5 Finally, in 7 we study the Scheme 3 N N N N Fe O N N N N Fe N O N N N N Fe N O N N N N Fe S O H2C H H H H H 1 (1+) 2 (2+) 3 (3+) 4 (4 –) 5 6 N N N N Fe S O H 7 H H O O Fe O2+ high- and low-spin states of the hydroxylation product of H2 by the model P-450 active species. Excited state calculations. Two types of excited states which are likely to have a role 15 along the hydroxylation path have been calculated.These are: the O2-like excited states and the internal charge-transfer states of the types ferryl-to-ring and ligand-to-ring. Our interest is primarily in the qualitative trends along the series of species 1–4, and hence the correct symmetry assignment for these excited states is important to us. Fig. 2 depicts the corresponding orbital energy diagrams for the high-spin states of O2 and FeO21. Both species have states which are also eigenstates of the angular momentum operator.Since the unrestricted Kohn–Sham (UKS) method is not capable of producing states with the correct angular momentum, except for high-spin states, we applied the vector coupling scheme,25 which enabled us to calculate the energy for the correct state symmetry from the single Kohn–Sham (KS) determinant energies. This scheme is approximate because it does not optimize orbitals for the respective states, and therefore its reliability is in a way after the fact.In this sense, it is pointed out that the scheme has been successfully applied to excited state calculations for various transition metal complexes and has demonstrated a good ability to predict the state energies as well as the corresponding structural parameters.26 Unlike methods of annihilation 27 or projection 28 within the UKS approach, this scheme enables one to calculate multiplets with spatial degeneracies.25 From the theoretical point of view, each KS determinant contributing into the energy of a multiplet state corresponds to a mixed-symmetry state which possesses a single determinant noninteracting reference.29 The equations for the vector coupling scheme are collected in the Appendix to this paper, while here we give a short description of these states.Thus, O2 possesses a 3Sg 2 ground state and 1Dg/1Sg 1 excited states, all nascent from a (p*)2 electronic con- figuration [Fig. 2(a)].Thus for example, as shown in Fig. 2(a) below the orbital diagram, the low lying 1Dg excited state contains the ‘classical’ doubly bonded O]] O species, whereas the ground state is a triplet p-diradical, 3Sg 2. The orbital diagram of FeO21 in Fig. 2(b) shows the valence configuration 1s22s2p4d2(p*)2 for the high-spin 5S1 state. The corresponding O2-like low-spin states are nascent from the same valence configuration 1s22s2p4d2(p*)2 by electron reshuZing in the p*- and d-orbitals.The 3D and the 3S2 state derive from the high-spin 5S1 state by realignment of the p* electrons, while keeping the d-electrons in a high-spin relation. Fig. 2 Valence orbital energy diagrams and ground state configurations for O2 (in a) and FeO21 (in b). Below each diagram are shown the excited states nascent by distributing electrons among the highest lying p* and/or d orbitals. The boxes depict the doubly bonded excited state D-type species. O O Fe O O2 FeO S S S S D D , , O O Fe O D D s 1s 2s p p*x y p p x y * * 3s 1s 2s d d x2– y2 xy p p xz yz p p*xz yz * z z 2+ 3 g – +5 3– 1 g +1 g 3 1 g: 2+ 3 : (a) (b) (p*)2 d2(p*)2402 J.Chem. Soc., Perkin Trans. 2, 1999, 399–410 Fig. 3 (a) The a1u and a2u-type orbitals, and the state symmetries for the closed-shell and cation radical situations of porphyrin. (b and c) The valence orbital diagram of the ferryl moiety in C4v and Cs point groups. In the boxes are shown the combined state symmetries of the ferryl-porphyrin complex.The cation radical porphyrin generates both high- and low-spin states due to diVerent spin coupling as shown in Scheme 4. N N N Fe N O N N N Fe• N O • L N N N N Fe D C C A A A A A A A A A A A A A A A A A A A A A A A , , A A A A , , X X X X X X z x 4h 4v s p *xz 2b 1 p *yz 2b 2 p xz 1b 1 p yz 1b 2 d x2– y2 a1 p *xz a' p *yz a" p xz a' p yz a" d x2– y2 a' a a 2 1u 2u 2 a a 1 1u 2u 2 a a 2 1u 2u 1 1 1g 2 1u 2 2u Por Por Por 1 1g 3 2 3 2 2 1u 3 2 4 1 2 2u 3 2 4 2 1 1g 3 2 1u 3 2 2u 3 (a) (b) (c) " " " 3 " 4 ' 4 " 3 2 3 " a1u a 2u 2 2 " ' 2 1 2 2 Thus, the 3D state is an analog of the 1Dg state of O2 with an Fe]] O double bond, while 3S2 is an analog of the 1Sg 1 state of O2.The orbital schemes for the ferryl complexes are shown in Fig. 3. The porphyrin frontier orbitals which belong to the a2u and a1u representations in the D4h point group20 are shown in (a) along with the local symmetry of the porphyrin state for the closed-shell and the two potential cation radical situations.The ferryl orbitals are shown in (b) and (c) for the C4v and Cs ligand fields pertaining to the model compounds 1–4 in the oxidized and reduced forms. The d-orbitals of the ferryl group split in the ligand field and only the lower lying dx22y2 is occupied with two electrons. The p*-antibonding orbitals remain degenerate in the C4v ligand field and their splitting in the Cs ligand field is small. As such, these orbitals are occupied in (b) and (c) in a high-spin triplet situation, giving rise to 3A2 and 3A0 state symmetry, which are analogs of the high-spin ground state of O2.Below the orbital diagrams in Fig. 3(b) and 3(c) we show the combined state symmetry for a closed-shell and cation radical porphyrin situations. Thus, the complexes with the closed-shell porphyrin remain 3A2 and 3A0 for the C4v and Cs ligand fields, respectively. In contrast, the states of the complexes with cation radical porphyrin become 2,4A1,2, 2,4A9 and 2,4A0 depending on the ligand field symmetry (C4v and Cs), the identity of the porphyrin’s singly occupied orbital (a2u or a1u), and the mode of spin-coupling between the triplet ferryl module and the cation radical situation in the porphyrin.There are two modes of spincoupling which are depicted in Scheme 4; mode (a) shows the high-spin quartet state which arises from the ferromagnetic coupling of the three electrons, while mode (b) shows the corresponding antiferromagnetic coupling.20a,30–32 The ferromagnetic states have been computed for the ferryl complexes 21, 31 and 4 (Scheme 1), while the antiferromagnetic states have been computed only for the representative cases, 31 and 4.Two types of excited low-spin states implicated in the TSR reactivity of P-450 15 were computed for the ferryl complexes: (i) excited states of the ferryl moiety that involve a reshuZe of the p*-electrons as the corresponding O2-like states in Fig. 2(a), and (ii) intramolecular electron transferred states which fill the ‘hole’ in the porphyrin, either from the p*-electrons of the ferryl group or from ligand orbitals. Fig. 4 shows the O2-like excitations along with their state symmetries in the C4v and Cs ligand fields. In the C4v ligand field, the 1D-type states split into two states of B1 and B2 symmetry, while in the Cs ligand field these states become A9 and A0. The 1S1-type states become either A1 or A9.All these low-spin states are either singlets, for cases with closed-shell porphyrin, or doublets whenever the porphyrin has a cation radical situation. Next, we turn to Figs. 5 and 6 to consider the intramolecular charge-transfer states which involve an electron transfer from either the ferryl moiety or the ligand into the singly occupied porphyrin orbital. These states are analyzed only for complexes with the cation radical porphyrin, since the compounds with a closed-shell ring have no appropriate low-lying orbital for accepting the transferred electron. Fig. 5 shows the charge-transfer states arising by electron transfer from the ferryl p* orbitals to the singly occupied a2u orbital, henceforth ‘ferryl-to-ring charge-transfer states’. In the C4v point group in (a) these excited states belong to the doubly degenerate representation (2E). In the Cs group, in (b), this doubly degenerate charge-transfer state splits into 2A9 and 2A0 states nascent by electron transfer from the 2a0 and 2a9 p* orbitals into the a2u type porphyrin orbital.In case (a) the 2E Scheme 4 p* xz p* yz a2u (a1u) p* xz p* yz a2u (a1u) (a) (b) S=3/2 S=1/2 4A2 (4A1) 4A' (4A") 2A2 (2A1) 2A' (2A") C4v CS (ferromagnetic coupling) (antiferromagnetic coupling)J. Chem. Soc., Perkin Trans. 2, 1999, 399–410 403 Table 1 Relative energetic, selected structural parameters, and spin densities in 1–4 and 6 Point Relative energy/ Spin densities Entry Species group State kcal mol21 FeNring/Å FeO/Å FeX/Å Fe O FeO[Por1?] complexes 1 2 3 4 11 21 31 4 C4v Cs Cs Cs 4A1 4A2 4A9 4A0 4A9 4A0 a 4A9 4A0 b 22.8 0.0 12.4 0.0 12.6 0.0 19.7 0.0 2.014 2.028 2.026 2.038 2.026 2.038 2.028 2.032 1.638 1.632 1.651 1.651 1.652 1.653 1.689 1.676 2.174 2.149 2.112 2.083 2.433 2.471 1.26 1.27 1.18 1.17 1.13 1.11 1.21 1.21 0.79 0.80 0.87 0.88 0.89 0.89 0.86 0.86 FeO[Por] complexes 5678 12342 C4v Cs Cs Cs 3A2 3A0 3A0 3A0 2.024 2.035 2.035 2.037 1.643 1.653 1.652 1.684 2.169 2.095 2.504 1.30 1.22 1.16 1.28 0.75 0.83 0.85 0.82 Bare ferryl 9 6 C•v 5S1 1.616 3.06 0.94 a The antiferromagnetic state for 31 lies 11 cm21 higher.b The antiferromagnetic state for 4 lies 31 cm21 higher. In both cases, the state energies were calculated as in ref. 20a. state does not have a symmetry match with the O2-like low-spin states (in Fig. 4), and the two state types cannot mix. In case (b), there is a symmetry match between the two state types, but the coupling matrix elements between the symmetry matched states are expected to be small since they involve overlap between the ferryl p*-orbitals and a2u or a1u orbitals of the cation radical porphyrin.Hence, the charge-transfer states have been assumed to be noninteracting with the O2-like states in the parent ferryl-complex geometry. As such, in both symmetry point groups, C4v and Cs, the charge-transfer states can be represented by single determinants. Due to what seem to be serious degeneracies in this energy range, the spin-unrestricted SCF-KS procedure does not converge to the desirable states.Hence, the energies of the ‘ferryl-to-ring charge-transfer states’ have been estimated nonself-consistently, computing the Fig. 4 O2-like low-spin excited states of the ferryl complex and their symmetry assignment in C4v and Cs point groups. S A 3 g – Dg S 1 g + 3 " B 1 2 1 1 1 A 3 2 a-A 1 ' 4 B A A 1 1 1 B 2 b-A 1 ' 1 B 2 2 A 2 1 A 4 " A 1 " a- A 2 ' A 2 " b-A 2 ' , , Fe L O N N N N , , -like -like [Por]FeO [Por ]FeO z x p xz * 2b1 a' , p yz * 2b2 a" , , D 1 g-like S 1 g + -like -like S 3 g –-like +• energies of single determinants constructed from the ground state eigenvectors. These charge-transfer excitation energies are no doubt overestimated, but are expected to yield qualitative trends on the eVect of diVerent proximal ligands. Fig. 6 shows the charge-transfer states arising by electron transfer from the ligand out-of-plane type orbitals to the singly occupied ‘a2u’ orbital, henceforth ‘ligand-to-ring chargetransfer states’.When the ligand is HS2, in (a), the electron is transferred from the p-lone pair orbital of sulfur, while when the ligand is HN]] CH2, in (b) the electron is transferred from the pN]] C orbital. Ferromagnetic excited states as well as an antiferromagnetic excited state are expected, depending on the spin coupling between the odd electron on the ligand and the triplet electrons of the ferryl group.The ‘ligandto- ring charge-transfer states’ are computed in a SCF-KS procedure. Results Ground state properties of oxidized and reduced ferryl complexes Table 1 summarizes the ground state properties of 1–4 and 6, while Fig. 7 provides additional structural data for the ground states of the ferryl-complexes. Electronic structure of the ground states. The data in Table 1 show that in all the oxidized forms in entries (1)–(4), the ground state is a high-spin quartet state typified by a cation radical porphyrin coupled ferromagnetically with a triplet p-diradical ferryl group.The antiferromagnetic states lie slightly higher (see comments in the Table), and indicate a very small coupling of the ferryl to the porphyrin odd electrons. In the absence of a proximal sixth ligand (entry 1), the ground state 4A1 involves the porphyrin cation radical in the 2A1u situation, while in the presence of an axial sixth ligand the ground state 4A0 involves the 2A2u porphyrin cation radical state.These results are in accord with experimental data,1–5 with DFT calculations 20 as well as with CASCCF30 and other calculations ranging from iterative extended Hückel 31 to X-a32 and UHF;33 all of which indicate that in the sixth coordinated ferryl-complex the porphyrin has a hole in the a2u orbital. Comparing the complexes 21 and 31 shows that better p-donation of the sixth ligand in the latter complex has a negligible eVect on the ground state properties of these complexes.404 J.Chem. Soc., Perkin Trans. 2, 1999, 399–410 In contrast, the thiolate ligand in 4 has a more significant eVect on the state separation. The ground states of the reduced forms, 1–42, of the ferrylcomplexes in entries (5)–(8), are all triplet states corresponding to O2-like p-diradicals in accord with CASSCF30 and experimental data.31,34 Thus, as analyzed in Fig. 3 above, it is apparent from Table 1 that the oxidized form of the analogs of HRP(I) and P-450 ferryls (11–4) possesses two closely lying highspin states with a ‘hole’ in the porphyrin ring and an O2-like ferryl moiety, while in the reduced form analogs of HRP(II) there is one high-spin state associated with an O2-like ferryl moiety.Geometric features of the ground states. The optimized geometric parameters in Table 1 show that in the ferryl complexes, the Fe–O bond distance is in the range of 1.64–1.69 Å, well within the range of experimentally measured bond lengths for ferryl in active compounds: 1.604–1.68 Å.1–5,20 For comparison we show in entry (9) that the calculated bond length for the 5S1 ground state of the FeO21 is 1.616 Å; also within the range of experimentally measured bond lengths for ferryls.The FeIV–S distance in 4 and 42 is a bit long, but since the calculations reproduce a reasonable FeIII–S bond length 11b for the water complex (see later Fig. 8), we may consider that the BP86 result for the FeIV–S bond length in the ferryl complexes is Fig. 5 Ferryl-to-ring charge-transfer states and their symmetry assignment in (a) C4v point group, and in (b) Cs point group. Fe L O N N N N pxz * p yz * 2b1 2b2 e a 2u 2 – a2u A 4 2 E 2 1 * 2u p z x (a) C4v (b) Cs p xz * p yz * 2a" e– a2u A" 4 2a' p xz * p yz * 2a" 2a' p xz * p yz * 2a" 2a' a2u a2u A' 2 A" 2 a 2u 2 (a') (a") 1 1a 2u 2 a reliable. In fact, all the optimized structural data for the ferryl complexes are in agreement with most models constructed from X-ray data and by restricted optimizations of the Fe–O bond.4,11b,20,31–33,35 The complementary structural data in Fig. 7 show that without the sixth ligand the iron lies significantly above the porphyrin ring; by as much as 0.305 Å. The sixth ligand ‘pulls’ the iron back into the plane, albeit not completely. This trend is also in line with known experimental data.1–5 There is no special ligand eVect on this structural feature.11b The only structural eVect exerted by the ligand is on the Fe–O bond, which as seen from Table 1 undergoes elongation for thiolate as a ligand.This latter eVect is common to the oxidized and reduced ferryl-complexes, and provides some support for the ‘push’ eVect discussed for thiolate ligands,1,3,11b–d and is generally expected from ligands which are good p-donors.12b Finally, Table 1 shows also the computed spin-density distribution on Fe and O within the ferryl complexes. The spin density is almost equally distributed over the two atoms, in accord with the expected triplet p-diradical character of the Fe–O bond.The values of the spin densities are in accord with the recent DFT calculations of Kuramochi et al.20a performed with a larger basis set, with (SCC)X-a results,32 as well as with the results of CASSCF calculations.30 Fig. 6 Ligand-to-ring excited charge-transfer states for L = SH2 in (a) and L = NH]] CH2 in (b). The electron in the ligand orbital is drawn with a doubled-direction spin, such that the low- and high-spin situations correspond to antiferromagnetic and ferromagnetic spin coupling modes with that electron.Fe S O N N N N H Fe N O N N N N CH2 H x z z x (a) (b) p xz * p yz e– a2u * p xz * p yz * e– a2u p xz * p yz * a2u p xz * p yz * a2u pNC pNC p S p S A' A' / 2 4 A' A' / 2 4 Fig. 7 Fe-ring displacement parameters for the ferryl complexes (1–4) in the ground states of the oxidized and reduced forms.L molecular charge d/Å none (1) NH3 (2) H2C=NH (3) SH – (4) + , 0 + , 0 + , 0 0 , – 0.268, 0.305 0.116, 0.149 0.100, 0.129 0.100, 0.075 Fe dJ. Chem. Soc., Perkin Trans. 2, 1999, 399–410 405 Electronic structure of a product-complex state: a TSR scenario Having ascertained the high-spin nature of the ground state, we turn to the product-complex state (7), which would correspond to the H2 hydroxylation reaction nascent from the high-spin P-450 model compound (4).The results are depicted in Fig. 8 which shows the reactants, and the most stable structures for the high- as well as low-spin product complexes (7HS, 7LS, 7LS9).36 The ground state of the complex is the low-spin doublet state of the product-complex (7LS); a result in accord with recent electron spin echo envelope modulation (ESEEM) spectroscopy of the water complex of P-450,37 as well as with other experimental data of low-spin six coordinated FeIII complexes which possess thiolate groups.38 The plane of the water molecule and the porphyrin ring in 7LS are parallel, and this provides some extra stabilization due to internal hydrogen bonding between the protons of the water molecule and the negatively charged nitrogen atoms of the porphyrin. Indeed the two hydrogen bonded nitrogen atoms are found to carry a significantly higher negative charge than the other two.According to the ESEEM spectroscopic data 37 the water molecule is in an upright position.Our calculations show that this conformation shown in 7LS9 is of higher energy. It was postulated 37 that the upright position is stabilized by hydrogen bonding with the protein which is in line with our interpretation that the parallel conformer 7LS is the most stable one due to the internal hydrogen bonding interactions. For both low-spin structures, the Fe–O bond lengths are well within the range of known experimental values for water molecules bonded to Fe.14,35a Other low-spin complexes of A9 symmetry were found to be significantly higher in energy and are not reported here.The corresponding highspin quartet state 4A0 (7LS) is 13.6 kcal mol21 less stable than the low-spin complex 7LS and has a virtually unbound water molecule. We conclude therefore that the ground state of the water-complex is 7LS/7LS9 with 2A0 symmetry. A rationale for the low-spin ground state of the water complex is provided in Scheme 5 using the upright water position for simplicity.The scheme shows the d5 electronic configuration in the low- and the high-spin situations in (a) and (b). For a strong ligand field a low-spin occupation would be expected to possess lower energy than the corresponding high-spin complex, because the latter requires excitation of an electron from the dxz orbital to the high lying s*(dz2) orbital. This electron occupation pattern results in a considerable Fe–O bond weakening and hence destabilization of the quartet high-spin product-complex, as evident from Fig. 8. The above low-spin ground state of the product complex models experimental findings that the product complex of 5-hydroxycamphor with P-450cam is low-spin with a short Fe–O bond.14,38a,39 This in turn shows that the hydroxylation process would generally be spin-nonconserving and involve crossover of high- and low-spin states along the reaction pathway, in accord with the recently proposed TSR paradigm15 which is schematized in Fig. 1. Furthermore, as evident from Scheme 5, in both high- and low-spin product states the a2u orbital is doubly occupied and the porphyrin ring is closed-shell; in contrast to the reactant ground state, where the a2u orbital is singly occupied. Thus, there must be an internal electron transfer which fills the porphyrin’s hole along the reaction pathway. The hole state then plays an essential role in the reaction. The above results provide an incentive to look for excited low-spin states which may, on the one hand, fit the TSR concept and which, on the other hand, take into account the changes in the electronic structure from the reactant to the product state.These excited states are described below. O2-Like excited states of FeO21 and ferryl complexes Table 2 collects the excitation energies for the species which serve as bonding models for the ferryl unit; the O2 molecule, as well as the O2-like excitations of FeO21. The calculations for FeO21 have been carried out for the equilibrium bond length as well as for a stretched bond, 1.65 Å which corresponds to an Table 2 O2-like excitation energies (eV) of gas phase FeO21 and O2 excitation energies a Entry Transition Excitation energy O2 excitation energies 12 1Dg–3Sg 2 1Sg 1–3Sg 2 1.025 2.042 FeO21 excitation energies 34 3D–5S1 3S2–5S1 rFeO = 1.616 Å 1.172 2.344 rFeO = 1.650 Å 1.113 2.225 a See Fig. 2 for state definitions, and eqns. (1)–(4) in the Appendix. Fig. 8 Ground state of the P-450 ferryl complex model and the low- and high-spin product complexes for H2 oxidation.N N N N N N N N O N N N N O S Fe Fe S S H H H H H H H H2 D D A A HS LS A S Fe+• O N N N N O Fe S H H H D A LS' 1.676 2.471 8.204 – 0.307 – 0.169 0.976 0.983 102.8 102.4 + E = 0.0 kcal mol –1 E = – 40.2 kcal mol –1 high-spin E = – 53.7 kcal mol –1 low-spin 2 " 4 " 7 7 4 " 1 g + 2.343 2.178 2.257 0.100 – 0.133 0.971 110.1 E = – 47.1 kcal mol –1 low-spin 2 " 7 2.183 2.159 2.296406 J.Chem. Soc., Perkin Trans. 2, 1999, 399–410 Table 3 Energies (eV) of the O2-like excitations for ferryl-complexes a Entry Complex Group Ground state Excited state/Excitation energy 1 234 5 678 11 21 31 4 1 234 2 C4v Cs Cs Cs C4v Cs Cs Cs 4A1 4A0 4A0 4A0 3A2 3A0 3A0 3A0 2B1 0.698 (0.651 b) 2A0 0.689 0.659 0.698 1B2 0.696 2A0 0.689 0.647 0.696 2B2 0.883 (1.000 b) a-2A9 0.916 0.866 0.873 1B1 0.879 a-2A9 0.904 0.837 0.866 2A1 1.581 (1.586 b) b-2A9 1.606 1.544 1.598 1A1 1.575 b-2A9 1.593 1.517 1.592 a See Fig. 4 for state definitions and eqns. (5)–(11) in the Appendix. b Datum obtained with the ROKS method.41 average value in the ferryl-complexes (see Table 1). It is seen that the excitation energies (entries 1 and 2) from the high-spin ground state of O2 to the corresponding low-spin excited states are in reasonable accord with experimental data.40 The same results were obtained recently using the restricted open-shell Scheme 5 N N N N Fe S O H H H N N N N Fe S O H H H x z y x z y (a) (b) 7LS' 7HS d x 2 – y 2 d yz a2u a2u d yz d x 2 – y 2 s*(d z2) s*(d z2) d xz d xz Kohn–Sham (ROKS) method which employs orbital optimization, 41 thereby indicating that the vector coupling scheme 25 is quite reliable.The corresponding excitations of FeO21 in entries 3 and 4 in Table 2 are of similar magnitudes, and are not very sensitive to the variations in the Fe–O bond length. The similarity between the states of the two species is indeed striking, albeit expected based on their common electronic origins (see Fig. 2). The oxygen-like excitations in the ferryl-complexes 1–4 in their oxidized and reduced forms are presented in Table 3. The parenthetical values in entry 1 obtained with the ROKS method41 show that the vector coupling scheme gives results within a few hundredth of eV from a result obtained by orbital optimization. The various states can be discussed with reference to Fig. 4. We recall that in the C4v ligand field, in 1, the D-type excited states split into two states of B1 and B2 symmetry, while in the Cs point group of 2–4, these states are A0 and A9.The computed data in Table 3 show that the excitation energies of this type are quite insensitive to the nature of the axial ligand and state of the porphyrin, cation radical or closed-shell. The lowest excitation energy is ca. 30% lower than in the FeO21 or oxygen molecules. These excitation energies are however in the same range as those calculated by CASPT2 for the bare FeO1 molecule, which is another possible model for embedded ferryl: 0.5–0.8 eV.13a Hence, by analogy with the FeO1 species, the low lying doublet state where the ferryl moiety is in a perfect pairing Fe]] O situation, is energetically accessible and can become involved in the hydroxylation of alkanes via spin-state crossing. Charge-transfer excited states of ferryl complexes To identify low-spin excited states which are responsible for diVerences between the ferryl-complexes and account for the requisite internal electron transfer, we turn to the chargetransfer states, described in Figs. 5 and 6. The results of these calculations are given in Table 4; entries 1–4 provide the ferrylto- ring excitations, while entries 5 and 6 list the ligand-to-ring excitations. Inspecting entries 1–4, it is seen that, in the absence of the sixth axial ligand (in 11) the ferryl-to-ring charge-transfer excitation is very high, while a nitrogen ligand lowers the 4A0Æ2A9 excitation, the value still remains large.The most noticeable eVect is produced by the thiolate ligand (4) which lowers the excitation energies to the two ferryl-to-ring charge-transfer states, with a larger eVect on the 2A9 state. A similar eVect of the proximal ligand is apparent from the ligand-to-ring chargetransfer excitations in entries 5 and 6 which show that the thiolate ligand gives rise to low lying states, in comparison with the p-donor ligand H2C]] NH (in 31).These qualitative trends are in general agreement with findings that the thiolate ligand produces low energy excited states 11d in comparison with imidazole type ligands.J. Chem. Soc., Perkin Trans. 2, 1999, 399–410 407 Table 4 Energies (eV) of the charge-transfer (CT) excitations Entry Complex Group Ground state Excited state/Excitation energy Ferryl-to-ring CT state 1 11 C4v 4A1 2E 4.839 a 234 21 31 4 Cs Cs Cs 4A0 4A0 4A0 2A9 4.804 b 3.505 b 2.182 b 2A0 4.804 c 4.873 c 4.447 c Ligand-to-ring CT state 56 31 4 Cs Cs 4A0 4A0 2A9 6.219 d 0.552 d 4A9 6.927 e 0.897 e a Calculated as single determinant energy (see the Appendix for specifications of the orbitals) E(2E) = E(exa2ua�2u).b Calculated as E(2A9) = E(a9a2ua�2u). c Calculated as E(2A0) = E(a0a2ua�2u). d Calculated as E(2A9) = 3– 2E(a9a0lp – a0) 2 1– 2E(a9a0lpa0). e Calculated as E(4A9) = E(a9a0lpa0). Discussion The computational results make a case for a TSR reactivity 15 mechanism of hydroxylation, where quartet and doublet spin states intersect along the reaction path, and in which chargetransfer states play a role in preparing and stabilizing the product state, by filling the porphyrin ‘hole’.The proximity of anti-ferromagnetic reactant states (Table 1) and the possibility that the reactant state exists in a dynamic mixture of highspin and low-spin situations, will certainly assist the spin-state crossover along the reaction path 42 but will not altogether eliminate the requirement for spin–orbit coupling mediated transition from the high-spin ground state to the low-spin product-state.Let us then proceed to analyze some key issues projected by the results; the roles of the porphyrin ‘hole’ and of the excited states of the ferryl, as well as the eVect of the proximal ligand through the charge-transfer states. The role of the porphyrin ‘hole’ The importance of the porphyrin ‘hole’ may be understood by reference to the electronic structure of the water-complex in Scheme 5, which shows that the ‘a2u’ orbital is doubly occupied and the porphyrin is therefore closed-shell while the d-block of the metal contains five electrons in a dx2 2 y2 2dxz 2dyz 1 formal configuration.Thus, imagine a hydroxylation process starting with a ferryl-complex that already possesses a closed-shell porphyrin, as e.g., in the HRP(II) species. Since initially there is one additional electron, this would be required to populate in the hydroxylation-complex either the empty s*(dz2) orbital or the singly occupied dyz orbital.The first option would have led to the population of a high lying orbital, while the second option would have required a closed-shell dx2 2 y2 2dxz 2dyz 2 configuration (formal FeII) which is destabilized by virtue of electron– electron repulsnd is unfavorable unless the ligand field is very strong. Thus, the porphyrin’s ‘hole’ serves as an electronic sink which stabilizes the hydroxylation product state.This conclusion provides a theoretical rationale for the findings 9 that the compound (II)-type ferryl-complexes with the closed-shell porphyrin either do not exhibit an oxidative activity (see footnote 13 of ref. 9a) or are sluggish oxidants (ref. 9b). This is also in line with the fact that the HRP(II) compound does not participate in hydroxylation but rather in electron transfer en route to regeneration of the active species HRP(I).2,3,5,11b The situation of the ferryl embedded in the porphyrin complex can be further compared with that of bare Fe–O113 which performs hydroxylation without the presence of the porphyrin.Thus, bare FeO1 with the dangling d-orbitals of the coordinatively unsaturated iron has a built-in electron sink,13,43 whereas in the ferryl complex of e.g., P-450, in which the iron is coordinatively saturated, the sink has to be provided by the cation radical state of the porphyrin ligand.The role of the low-spin excited states of the ferryl-complex The calculations focused on two excited state types which have a role along the hydroxylation path; the O2-like excited states (Table 3) and the internal charge-transfer states (Table 4). Consider first the O2-like excited states. Initially in the highspin ground state the ferryl unit has a half-filled valence shell and no low lying empty orbitals which can create bonding interactions with the substrate undergoing hydroxylation.In contrast, the low-spin states (especially the 1D-types) possess a low lying empty orbital (p*) which can interact with the substrate (RH) and lead to formation of the new O–H and O–R bonds. Thus, the role of O2-like low-spin excited states is to activate the ferryl unit towards bond formation with the substrate. 13,15 The fact that these states are considerably lower than the corresponding states in the O2 molecule itself (Table 2 vs.Table 3) implies also that ferryl-complexes will be more powerful oxidants than O2. The role of the internal charge-transfer states is to promote the electron reorganization that fills the porphyrin ‘hole’ along the reaction pathway. A more detailed orbital origin of the charge-transfer state can be defined by a simple electron count of the valence electrons of the reactant ferryl-complex and the product water-complex. Since the p and p* orbitals of the ferryl unit have mixed Fe and O characters, all the corresponding electrons must be counted.This will require counting also the p-lone pair electrons of the water moiety in the product complex since this orbital mixes with the symmetry matched d-orbitals of the iron. Thus, initially the ferryl moiety involves eight electrons in the d/p block (d2p4p*2, Fig. 3) and one electron in the ‘a2u’ orbital. Finally, in the product complex, the Fe–O moiety contains seven electrons (pO 2dx2 2 y2 2dxz 2dyz 1), while ‘a2u’ is doubly filled (Scheme 5).Thus, while the precise details of the charge-transfer may be complex, it is apparent that the ferryl moiety loses one electron to the porphyrin during the reaction. This in turn means that the ferryl-to-ring charge-transfer states participate in the internal electron transfer which occurs during the reaction. A simple rationale for this internal reorganization is that in order for the ferryl oxygen to form two new bonds during the hydroxylation, and at the same time retain one Fe–O bond, one electron must be lost from the bonding block of the ferryl moiety and relayed by internal charge-transfer to the ‘hole’ in the porphyrin.It is important to recognize that along the reaction coordinate the charge-transfer states and the O2-like excited states will mix with each other until correlation is achieved to the product state. Thus, it is the combination of the excited states which will promote the bonding reorganization required to transform the reactant state to the product state.408 J.Chem. Soc., Perkin Trans. 2, 1999, 399–410 The eVect of the proximal ligand The lower the energy of the state combination which correlates to the product state along the reaction pathway, the lower would be the crossing point between the quartet and doublet states (see Fig. 1) and the smaller the barrier for the process is likely to be. In this sense, it is apparent that while the O2-like states (Table 3) are unaVected by the ligand, in contrast the charge-transfer states (Table 4) exhibit a significant dependence on the identity of the proximal ligand; the sulfur ligand being considerably more eVective than the nitrogen ligands.Let us therefore discuss the possible pathways along which the proximal ligand eVect may be expressed. Inspection of Fig. 8 shows that the low-spin product state possesses 2A0 symmetry. Tables 3 and 4 show that there exist two sets of O2-like and charge-transfer excited states with 2A9 and 2A0 symmetries.It is apparent therefore that it is the 2A0 combination of O2-like and charge-transfer states which will eventually correlate to the product state. Nevertheless, the proximal ligand eVect depends on the question of whether or not the hydroxylation reaction path conserves a plane of symmetry, i.e., if all the structures along the path belong to the Cs point group. If the reaction path conserves Cs symmetry, only the 2A0 states will aVect reactivity.As seen by comparing entries 3 and 4 in Table 4, the SH2 ligand stabilizes the 2A0 charge-transfer state by ca. 0.4 eV in comparison with the CH2]] NH ligand. Such a pathway will exhibit a moderate eVect of the proximal ligand making the thiolate-ferryl complex more reactive than its nitrogen-base alternative. If however, the reaction path does not conserve any symmetry, this will lead to the mixing of all the charge-transfer states, and since the 2A9 charge-transfer state (entries 4 vs. 3, and 6 vs. 5) is very strongly aVected by the identity of the ligand, we may anticipate a much more pronounced ligand eVect on reactivity. Thus, a ligand eVect is expected irrespective of the reaction path symmetry, and the thiolate ligand should exert a more pronounced eVect on the hydroxylation mechanism compared with the nitrogen-based ligand. Fig. 9 shows an orbital interaction which accounts for the proximal ligand eVect of the SH2 ligand.Sulfur has two lonepair orbitals (a9 and a0 in Cs point group symmetry), and each one of them can mix with the symmetry matched singly occupied p* orbital of the ferryl unit, and form bonding and antibonding combinations, the latter made primarily of the p*(FeO) orbital. The mixing depends on the Fe–S distance, which is seen from Fig. 8 to undergo significant shortening (2.471Æ2.18 Å) with concomitant pulling of Fe toward the sulfur from the position above the porphyrin (10.100 Å) to the position below it (20.169 Å/20.133 Å for 7LS and 7LS9 respectively).Clearly, then, as the Fe–S bond gets shorter along the reaction path, the orbital mixing increases and the antibonding orbital combin- Fig. 9 The ligand assisted mechanism of internal electron transfer from the ferryl to the porphyrin ring along the hydroxylation pathway. The lone-pair orbital (lp) of the sulfur ligand mixes with the p* orbital of the ferryl. The electron in the antibonding combination orbital eventually fills the singly occupied a2u orbital of the porphyrin. 1e– a2u lp S pFeO * ation will rise up to a point where it can depopulate its electron into the a2u orbital of the porphyrin. The net eVect is a signifi- cant stabilization of the system along the reaction coordinate. It follows therefore that the thiolate ligand stabilizes the chargetransfer state which is required to assist the transformation of reactants to products by providing a donor orbital that can mix with the ferryl p* orbitals and cause them to shift an electron to the porphyrin ‘hole’.44 The nitrogen base ligand HN]] CH2 has in comparison a pN]] C orbital which is significantly lower than the thiolate lone-pair orbitals, and therefore HN]] CH2 has an ineVective orbital interaction and high lying charge-transfer states.Thus, it is here in the charge-transfer excitation that we see the impact of the proximal ligand which distinguishes the P-450 species (4) from the HRP(I) species (31).Similar explanations were given before 11b,c to account for this diVerence and the associated ‘push’ eVect 1,3 of the proximal thiolate ligands. This electronic eVect can be moderate to significant depending on whether or not the reaction path conserves Cs symmetry. Moreover, the proximal ligand eVect is associated with the presence of the porphyrin hole, and as such the reduced ferryl complexes like HRP(II) and type (II)-ferryls would not be expected to exhibit a pronounced proximal ligand eVect.Summary and conclusions Three fundamental issues were raised at the outset concerning the reactivity patterns of model ferryl-complexes of P-450 and HRP(I)/HRP(II) species (1–4 in Scheme 3). What is the role of the cation radical state of the porphyrin in P-450 and type (I)- ferryl complexes? What is the eVect of the proximal axial ligand in these complexes? And whether there occurs indeed a spinstate crossing during their hydroxylation reactions, as postulated in reference 15? Our DFT study and theoretical analysis provide insight into these problems and demonstrate their intimate link.The calculations for P-450 and HRP(I)/HRP(II) species (1–4 in Scheme 3) reveal that all the ferryl-complex models possess a high-spin ground state, in which triplet ferryl-electrons are ferromagnetically coupled to a porphyrin electron in a singly occupied orbital (a2u) of the porphyrin cation radical.At the same time, the product state generated by P-450 oxidation of H2 (7 in Scheme 3) is in a low-spin doublet state. Thus, a case is made for a two-state-reactivity (TSR) 15 for P-450 alkane hydroxylation, where the initial high-spin surface is crossed by a low-spin surface along the reaction pathway. The low-spin surface which eventually correlates to the product state is influenced by an internal charge-transfer state in which a ferryl p* electron is relayed into the singly occupied orbital (a2u) of the porphyrin cation radical.This is mediated by the interaction of the ferryl moiety with both the proximal ligand and the substrate, along the hydroxylation pathway. Thus, the study provides supporting theoretical evidence that the cation radical state of the porphyrin is an essential ingredient required to accept the relayed electron and stabilize thereby the ground state of the hydroxylation product, and that the donor property of the proximal ligand has a significant influence on the energy of the ferryl-to-ring charge-transfer states which are essential to convert the reactant state to the hydroxylation product state. Thus, the role of the ‘hole’ state, the eVect of the proximal ligand and the TSR aspect are all linked.In this sense, our study sheds some light on the diVerence between HRP(I) and HRP(II) or in general between type (I) and type (II) ferryls,9 as well as on the eVect of the proximal ligand.12 One would expect that type (II)-ferryls which lack the porphyrin ‘hole’ will react generally slower than type (I) complexes, and will exhibit entirely diVerent axial ligand eVects.The present results must eventually be complemented by further quantum-chemical calculations which can elucidate the full mechanistic details of the hydroxylation by ferryl-J. Chem. Soc., Perkin Trans. 2, 1999, 399–410 409 porphyrins especially with regard to the transition states. Such studies are in progress.Finally, the study provides a few intriguing features which merit attention in future studies. One is concerned with the ligand-to-ring charge-transfer states in the P-450 ferryl-complexes. It is apparent from Table 4 (entries 5 and 6) that these states are very low lying for the SH2 ligand, and this is no doubt related to the high lying donor orbital (a0) of sulfur.45 It is reasonable to expect that in some thiolate derivatives the ground state of the ferryl-complex will involve a triplet ferryl coupled to a sulfur radical, like the charge-transfer states in Table 4 (entries 5 and 6).Trautwein et al.32 have indeed reported such a high-spin ground state for the CH3S2 ligand. In such an event, the ‘hole’ state shifts from the porphyrin to the proximal ligand and it plays a similar role to the porphyrin ‘hole’. Our calculations (Table 4, entry 6) however show that the ground state for this situation is more likely to be a low-spin state 2A0.Should this low-spin ground state be a common situation then the reactivity of thiolate derivatives will not follow TSR and will be distinct among the ferryl-complexes. The second intriguing feature is concerned with the dense manifold of ground states, which is exhibited in Table 1. Thus, within a range of a few kcal mol21, each ferryl complex possesses two diVerent high-spin ferromagnetic states (e.g., 4A9 and 4A0) and two antiferromagnetic twins. The oxidation process with background dynamics of all these states is a challenging problem to decipher.The loss of the Fe–O bond in the high-spin suggests that the high-spin and low-spin pathways will be diVerently aVected by the protein environment. Finally, the movement of the proximal ligand, exemplified by the shortening of the Fe–S bond during the reaction (2.471Æ2.18 Å) suggests a significant role for the protein pocket which provides the proximal ligand group. Acknowledgements This research is supported by the WV Stiftung and G.I.F.(The German-Israeli Foundation). Appendix The Appendix provides expressions used in applying the vector coupling scheme 25 to the excited states of the various species. Eqns. (1) and (2) were used to calculate the excited state energies for O2 from the corresponding Kohn–Sham determinants (Table 2). Here the p* orbitals refer to the real Cartesian orbitals in the x or y direction perpendicular to the molecular axis. With these equations, the low-spin excited states acquire the correct electronic symmetry.E(1Dg) = ��� [E(px*p� y*) 1 E(px*p� x*)] (1) E(1Sg 2) = E(px*p� y*) 1 E(px*p� x*) 2 E(px*py*) (2) The corresponding state energies for FeO21 are calculated (Table 2) using eqns. (3) and (4), where dx corresponds to the dx2 2 y2 d-type orbital and dy to dxy. E(3D) = ��� [E(dxdypx*p� y*) 1 E(dxdypx*p� x*)] (3) E(3S2) = 2E(3D) 2 E(5S1) (4) The ferryl states in C4v and Cs groups (Fig. 3) are very similar and their energies (Table 3) can be expressed via single determinant energies as given in eqns.(5)–(11). E(3A2) = E(exey); ex = px*, ey = py * (5) E(1B2) = 2E(exe�y) 2 E(3A2) (6) E(1B1) = E(exe�x) 1 E(exe�y) 2 E(1B2) (7) E(1A1) = E(1B2) 1 E(1B1) 2 E(3A2) (8) E(3A0) = E(a9a0); a9 = px*, a0 = py * (9) E(1A0) = 2E(a9a�0) 2 E(3A0) (10) E(a 2 1A9) = ��� [E(a9a�9) 1 E(a0a�0)] 2 ÷D2 1 K2 (11) E(b 2 1A9) = ��� [E(a9a�9) 1 E(a0a�0)] 1 ÷D2 1 K2 D = ��� [E(a9a�9) 2 E(a0a�0)] K = ��� [E(1A0) 2 E(3A0)] References 1 W.-D.Woggon, Top. Curr. Chem., 1996, 184, 40. 2 Cytochrome P-450: Structure, Mechanisms and Biochemistry, Ed.: P. R. Ortiz de Montellano, Plenum, New York, 1986. 3 M. Sono, M. P. Roach, E. D. Coulter and J. H. Dawson, Chem. Rev., 1996, 96, 2841. 4 (a) J. T. Groves and Y.-Z. Zhang Han, in Cytochrome P-450: Structure, Mechanisms and Biochemistry, Ed.: P. R. Ortiz de Montellano, Plenum, New York, ch. 1, 1986; (b) J. T. Groves, J. Chem. Educ., 1985, 62, 928. 5 B. Meunier, Chem. Rev., 1992, 92, 1411. 6 (a) M. Newcomb, M.-H. Le Tadic, D. A. Putt and P. F. Hollenberg, J. Am. Chem. Soc., 1995, 117, 3312; (b) M. Newcomb, M.-H. Le Tadic-Biadatti, D. L. Chestrey, E. S. Roberts and P. F. Hollenberg, J. Am. Chem. Soc., 1995, 117, 12085; (c) P. H. Toy and M. Newcomb, J. Am. Chem. Soc., 1998, 120, 9718. 7 (a) K. A. Lee and W. Nam, J. Am. Chem. Soc., 1997, 119, 116; (b) A. Vaz and J. Coon, Proc. Natl. Acad. Sci. USA, 1998, 95, 3555. 8 (a) H.-A. Wagenknecht and W.-D. Woggon, Chem. Biol., 1997, 4, 367; H.-A. Wagenknecht . Chem., 1997, 109, 404; (b) B. Stäubli, H. Fretz, U. Piantini and W.-D. Woggon, Helv. Chim. Acta, 1987, 70, 1173. 9 (a) K. Kamaraj and D. Bandyopadhyay, J. Am. Chem. Soc., 1997, 119, 8099. See especially footnote 13 therein; (b) J. T. Groves, Z. Gross and M. K. Stern, Inorg. Chem., 1994, 33, 5065. 10 D. Schröder and H. Schwarz, Angew.Chem., Int. Ed. Engl., 1995, 34, 1973. 11 (a) W.-D. Woggon, Nachr. Chem. Tech. Lab., 1988, 36, 890; (b) J. H. Dawson, Science, 1988, 240, 433; H. I. Liu, M. Sono, S. Kadkhodyan, L. P. Hager, B. Hedman, K. O. Hodgson and J. H. Dawson, J. Biol. Chem., 1995, 270, 10544; (c) J. Bernardou, A. S. Fabiano, A. Rober and B. Meunier, J. Am. Chem. Soc., 1994, 116, 9375; (d ) P. M. Champion, J. Am. Chem. Soc., 1989, 111, 3434. 12 For pronounced proximal ligand eVect on epoxidation in model compounds, see: (a) Z.Gross and S. Nimri, Inorg. Chem., 1994, 33, 1731; (b) K. Czarnecki, S. Nimri, Z. Gross, L. M. Proniewicz and J. Kincaid, J. Am. Chem. Soc., 1996, 118, 2929; (c) Z. Gross, J. Biol. Inorg. Chem., 1996, 1, 368; (d ) Z. Gross, S. Nimri and L. Simkhovich, J. Mol. Catal. A: Chem., 1996, 113, 231. 13 (a) A. Fiedler, D. Schröder, S. Shaik and H. Schwarz, J. Am. Chem. Soc., 1994, 116, 10734; (b) D. E. Clemmer, Y.-M. Chen, F. A. Khan and P. B. Armentrout, J.Phys. Chem., 1994, 98, 6522; (c) S. Shaik, D. Danovich, A. Fiedler, D. Schröder and H. Schwarz, Helv. Chim. Acta, 1995, 78, 1393; (d ) K. Yoshizawa, Y. Shiota and T. Yamabe, J. Am. Chem. Soc., 1998, 120, 564. 14 H. Li, S. Narashimhulu, L. M. Havran, J. D. Winkler and T. L. Poulos, J. Am. Chem. Soc., 1995, 117, 6297. 15 S. Shaik, M. Filatov, D. Schröder and H. Schwarz, Chem. Eur. J., 1998, 4, 193. 16 (a) For the role of agostic complexes, see: J. P. Collman, A. S. Chien, T.A. Eberspacher and J. I. Brauman, J. Am. Chem. Soc., 1998, 120, 425; (b) for the elucidation of the hydroxylation mechanism in amine oxidation, see: J. P. Dinnocenzo, S. B. Karki and J. P. Jones, J. Am. Chem. Soc., 1993, 115, 7111; S. B. Karki, J. P. Dinnocenzo, J. P. Jones and K. R. Korezkwa, J. Am. Chem. Soc., 1995, 117, 3658. 17 (a) T. Ziegler, Chem. Rev., 1991, 91, 651; (b) L. A. Eriksson, L. G. M. Petersson, P. E. M. Siegbahn and U. Wahlgren, J. Chem. Phys., 1995, 102, 872; (c) V.Jonas and W. Thiel, J. Chem. Phys., 1995, 102, 8474 (and references cited therein). 18 (a) A. D. Becke, Phys. Rev. A, 1988, 38, 3098; (b) J. P. Perdew, Phys. Rev. B, 1986, 33, 8822. 19 R. D. Amos, I. L. Alberts, J. S. Andrews, S. M. Collwell, N. C. Handy, D. Jayatilaka, P. J. Knowles, R. Kobayashi, N. Koga, K. E. Laidig, P. E. Maslen, C. W. Murray, J. E. Rice, J. Sanz, E. D. Simandrias, A. J. Stone and M.-D. Su, CADPAC5: The Cambridge Analytic Derivatives Package, Cambridge, UK, 1992.410 J. Chem.Soc., Perkin Trans. 2, 1999, 399–410 20 (a) H. Kuramochi, L. Noodleman and D. A. Case, J. Am. Chem. Soc., 1997, 119, 11442; (b) A. Ghosh, J. Almløf and L. Que, Jr., J. Phys. Chem., 1994, 98, 5576. 21 N. Godbout, D. R. Salahub, J. Andzelm and E. Wimmer, Can. J. Chem., 1992, 70, 560. 22 W. J. Hehre, L. Radom, P. v. R. Schleyer and J. A. Pople, Ab Initio Molecular Orbital Theory, Wiley-Interscience, New York, 1986. 23 M. Filatov and W. Thiel, Mol.Phys., 1997, 91, 847. 24 Computations have been carried out for ferryl-complexes and watercomplexes without and with a proximal ligand (NH3), using FT97 and BP86 with a Wachters basis set on Fe, 6-31G* on O, N (ligand), and N (ring) and STO-3G for C, H (or 6-31G** for H). The basis sets have been abandoned for computer time economy, and FT97 has been waived in favor of the more standard BP86 functional. We note though that these computations gave very similar results to the ones presented in the present paper. 25 (a) T. Ziegler, A. Rauk and E. J. Baerends, Theor. Chim. Acta, 1977, 43, 261; (b) U. von Barth, Phys. Rev A, 1979, 20, 1693; (c) C. Daul, Int. J. Quantum Chem., 1994, 52, 867. 26 (a) A. C. Stückl, C. Daul and H. U. Güdel, J. Chem. Phys., 1997, 107, 4606; (b) K. Doclo, C. Daul and S. Creve, Int. J. Quantum Chem., 1997, 61, 475; (c) F. Giraldoni, J. Weber, K. Bellafrouh, C. Daul and H. U. Güdel, J. Chem. Phys., 1996, 104, 7624; (d ) A. C. Stückl, C. Daul and H. U. Güdel, On the calculation of multiplets, in Recent Advances in Density Functional Methods (Part II), D. P. Chong, Ed., World Scientific Publishing Co., 1997. 27 C. J. Cramer, F. G. Dulles, D. J. Giesen and J. Almlöf, Chem. Phys. Lett., 1995, 245, 165. 28 A. A. Ovchinnikov and J. K. Labanowski, Phys. Rev. A, 1996, 53, 3946. 29 U. von Bart, Phys. Rev. A, 1979, 20, 1693. 30 S. Yamamoto, J. Teraoka and H. Kashiwagi, J. Chem. Phys., 1988, 88, 303. 31 (a) G. H. Loew, C. J. Kert, L. M. Hjemeland and R. F. Kirchner, J. Am. Chem. Soc., 1977, 99, 3534; (b) L. K. Hanson, C. K. Chang, M. S. Davis and J. Fajer, J. Am. Chem. Soc., 1981, 103, 663; (c) K. Tatsumi and R. HoVmann, Inorg. Chem., 1981, 20, 3771. 32 J. Antony, M. Grodzicki and A. X. Trautwein, J. Phys. Chem. A, 1997, 101, 2692. 33 A. Strich and A. Veillard, Nouv. J. Chim., 1983, 7, 347. 34 For a Ru analog of type-(II) ferryls, see: J. T. Groves and K. H. Anh, Inorg. Chem., 1987, 23, 3831. 35 (a) G. Loew and M. Dupuis, J. Am. Chem. Soc., 1996, 118, 10584; (b) ibid., 1996, 118, 10588. 36 The sextet high-spin state (S = 5/2) was computed at two diVerent geometries and found to be higher than the states shown in Fig. 8 here. In any event, the energy of the S = 5/2 state will not change any of the arguments. 37 H. Thomann, M. Bernardo, D. Goldfrab, P. M. H. Kroneck and V. Ulrich, J. Am. Chem. Soc., 1995, 117, 8243. 38 (a) See also, M. Unno, J. F. Christian, D. E. Benson, N. C. Gerber, S. G. Sligar and P. M. Champion, J. Am. Chem. Soc., 1997, 119, 6614; (b) H. Schappacher, L. Ricard, R. Weiss, E. Bill, R. M. Montoya, H. Winkler and A. X. Trautwein, Eur. J. Biol., 1987, 168, 419; (c) see however, a high-spin with a spatially fixed PhS2 ligand in a fenced prophyrin in ref. 8. 39 J. D. Lipscomb, Biochemistry, 1980, 19, 3590. 40 Gmelins Handbuch der Anorganischen Chemie. 8. Auflage, Verlag Chemie, GmbH, Weinheim, 1953. 41 See, M. Filatov and S. Shaik, J. Chem. Phys., 1999, 110, 116; M. Filatov and S. Shaik, Chem. Phys. Lett., 1998, 288, 689. 42 R. A. Harman and H. Eyring, J. Chem. Phys., 1942, 10, 557. 43 M. Filatov and S. Shaik, J. Phys. Chem. A, 1998, 102, 3835. 44 For similar conclusions on the role of thiolate, see: O. Zakharieva, M. Grodzicki, A. X. Trautwein, C. Veeger and I. M. C. M. Rietjens, J. Biol. Inorg. Chem., 1996, 1, 192. 45 For a discussion of various thiolate ligands and their eVect on spectra in FeIII complexes, see: K. K. Stavrev and M. C. Zerner, Int. J. Quantum. Chem.: Quantum Biol. Symp. 22, 1995, 155. Paper 8/09385G

ISSN:1472-779X    

DOI:10.1039/a809385g

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

J. Chem. Soc., Perkin Trans. 2, 1999, 411–417 411 Determination of absolute configuration of helicenes and related biaryls from calculation of helical twisting powers by the surface chirality model Alberta Ferrarini,a Giovanni Gottarelli,b Pier Luigi Nordio †a and Gian Piero Spada b a Dipartimento di Chimica Fisica “A. Miolati”, Università di Padova, Via Loredan 2, 35131 Padova, Italy b Dipartimento di Chimica Organica “A. Mangini”, Università di Bologna, Via S. Donato 15, 40127 Bologna, Italy Received (in Cambridge) 8th December 1998, Accepted 25th January 1999 The sign and the magnitude of the helical twisting power b for a series of helicenes have been calculated by the Surface Chirality model. The principal contribution to b derives from the helicity in the direction perpendicular to what can be defined as the main molecular plane; the cholesteric axis is also predicted to be along this direction, in agreement with a view of the cholesteric induction by helical molecules based on empirical observations. In the case of non-rigid biphenanthryl derivatives, the value of b is predicted to vary with the dihedral angle between the phenanthryl moieties, changing sign at about 908 where the conformation passes from s-cis to s-trans, i.e.when the stereochemical descriptor of the biaryl moiety exchanges the M and P indices. Comparison between experimental and calculated values indicates an s-trans conformation for the flexible dopants, in agreement with a previous conclusion drawn from empirical correlations.The Surface Chirality model appears to be a promising technique to assess the absolute configuration of rigid molecules by the comparison of experimental and calculated b values. For flexible molecules, the quality of the information depends critically on the degree of knowledge of their conformational freedom. Introduction Chiral resolved molecules are usually characterised by measuring their optical rotation, in general referred to the sodium D-line, or their circular dichroism spectra.Both methods are essentially spectroscopic and involve interaction of the electromagnetic radiation with the electrons of the molecules. An alternative non-spectroscopic way of characterising chiral molecules is the measurement of their twisting power in nematic liquid crystals.1,2 It is known that the addition of traces of a chiral solute to an achiral nematic liquid crystal induces the formation of a helical cholesteric structure, characterised by its handedness and pitch.Equal amounts of enantiomeric solutes induce helical structures with identical pitch and opposite handedness. DiVerent substances show diVerent abilities to twist the nematic phase, the helical twisting power being defined by eqn. (1),3 where p is the pitch (in micrometers), c the concenb = (p c r)21 (1) tration (moles of solute per mole of solvent) and r the enantiomeric purity of the dopant.The signs 1 or 2 indicate righthanded and left-handed cholesteric helices, respectively. The value of b is constant in a large range of concentrations and changes with the nature of the solvent. The twisting power can therefore characterise a chiral molecule in a similar way to optical rotation; as b is a non-spectroscopic quantity, it is expected to give diVerent and hopefully complementary stereochemical information.4–8 The measurements of pitch and handedness are simple and require only a common microscope with polarizers and a graduated scale in the ocular or an x 2 y translator.9–11 † Pier Luigi Nordio died on the 20th of October, 1998.His human and scientific presence will always be with us. Furthermore, the quantity of resolved compound needed is much smaller than for optical rotation. A quantitative relation between the twisting power and the molecular structure of the chiral dopant is derived by a theoretical method known as the Surface Chirality model.This is a model which, without dealing with the origin of the intermolecular forces, accounts in a phenomenological way for the short-range interactions of the solute molecule with the surrounding solvent, which are modulated by the solute molecular shape. Despite some approximations inherent in the derivation of the model 12–14 and the obvious diYculties in its application to molecules with many degrees of internal freedom, the method has been successfully employed to calculate b values for systems of various complexities, such as substituted biaryls and heptalenes.13,15,16 In this paper we show that it is possible to predict correctly the sign and magnitude of the twisting power for a group of helical-like molecules with rigid geometry [(M)-1–4].We have extended the investigation to analogous but conformationally flexible derivatives [(R)-5–8], and also in these cases, calculations of b for conformers predicted as the most stable are in agreement with experimental observation.This result confirms that for systems with internal flexibility the method can be applied with confidence to deduce information on the preferred conformations. Results and discussion The absolute configuration of helicenes Helicenes are characterised by a helical structure made up of ortho-condensed aromatic rings,17,18 the helical structure being a consequence of the steric interaction between terminal aromatic nuclei.The synthesis of the first optically active helicene (hexahelicene, 1) was reported by Newman19 in 1956 and a number of heterohelicenes, especially those containing thio-412 J. Chem. Soc., Perkin Trans. 2, 1999, 411–417 phene units,20 was subsequently described. The assignment of the absolute configuration of helicenes has mainly followed three distinct lines of approach: X-ray diVraction studies using Bijvoet’s method, chemical correlation, and calculations based on chiroptical properties.The absolute configuration of (2)-hexahelicene 1 has been unambiguously determined as (M) by a Bijvoet X-ray structure determination of the (2)-2-bromo derivative which was then chemically converted to the (2)-hexahelicene.21 This assignment is in agreement with the best available calculations,22,23 though not with the earlier ones performed by less refined methodology,24 and was finally confirmed by chemical correlation.25,26 The configuration of tetrahelicene 2 was deduced by SCF calculations.27 The configuration of the helicene-like compound 4 was determined by chemical correlation 28 with the precursor 4,49- biphenanthrene-3,39-diol whose configuration was already established.29,30 The configuration of heterohelicene 3 was deduced from chemical correlation 28 with the same biphenanthrenediol.This assignment is in agreement with what is expected from resolution with TAPA28,31,32 [2-(2,4,5,7-tetranitro- 9H-fluoren-9-ylideneaminooxy)propionic acid] and from qualitative comparison of its CD spectrum 33 with those of a series of heterohelicenes 34 whose absolute configurations were unambiguously assigned.The analogy of CD spectra34 of heterohelicenes containing thiophene moieties in diVerent positions and the behavior of the latter when treated with TAPA resolving agent 31,32 allow one to extend the configurational assignment to diVerent helicenes if the absolute configuration of one of them is known.The configuration of a hetero[6]helicene containing a benzodithiophene unit was assigned by X-ray diVraction 20 and confirmed by chemical correlation 25 and SCF calculation of the optical activity.20,34 The absolute configuration of 5–8 was determined by chemical correlation 28 with the precursor 4,49-biphenanthrene- 3,39-diol, whose configuration has been unambiguously established. 29,30 Although the absolute configurations of the helicenes investigated have been already obtained, the methods used are not of general application and suVer some limitations: X-ray diVraction requires the availability of crystals of appropriate characteristics, chemical correlation requires the existence of a stereocontrolled pathway, and empirical comparison of CD spectra is severely dependent on the nature of the chromophoric part of the molecule.In addition, calculations of the chiroptical properties cannot always be considered fully reliable. Despite initial diYculties in the calculation of the optical rotation of simple molecules,35,36 recently, substantial progress in this area has allowed the determination of absolute configuration of complex molecules.37 In the present paper, we show that it is possible to predict the helical twisting power of helicenes and related molecules and hence to determine their absolute configuration.The method presented here, based on the surface interaction between solvent molecules and a chiral probe, is independent of the solute electronic properties which determine the chiroptical behaviour.The surface chirality model The theoretical model has been presented in detail elsewhere, 13,14 and only the main physical aspects of the mathematical problem will be reviewed here. The model is based on the assumption that solely the molecular shape of a solute determines its alignment in the local nematic environment, and that the twisted shape of the chiral probe is able to exert a torque on the local nematic director, this eVect being transmitted at a distance of thousands of molecular lengths by virtue of the elastic properties of the nematic medium.14 Implementation of the model requires a mathematical description of the molecular shape, and a statistical mechanics treatment for the distortion free energy.This treatment is reminiscent of that presented in the de Gennes book to consider the competing eVect of wall alignment and field-alignment, for the case of pure twist.38 The molecular surface is constructed by considering the molecule as an assembly of van der Waals spheres centred at the atomic positions.‡ By exploring the molecular surface by unit vectors normal to each surface element, numerical values for two molecular properties, termed as “surface” T and “helicity” Q tensors respectively, are obtained.41 The tensors are defined in any arbitrary molecule-fixed axis system, but it is always possible to identify for each of them a principal axis system in which the tensor has only diagonal elements, called principal values. In general, the surface and helicity tensors have diVerent principal axes, unless rotational symmetry axes do exist, which, in this case, are principal axes for both tensors.42 The values (Txx, Tyy, Tzz) of the surface tensor define the tendency of the corresponding molecular axes to align along the nematic director; the values (Qxx, Qyy, Qzz) of the helicity tensor quantify the helicities as viewed along those axes.According to common convention, a positive (negative) Qii value indicates right (left)-handed helicity of the molecular surface along the i-th molecular axis. Both tensors are traceless, i.e. the sum of their diagonal elements is equal to zero. It is important to recall the diVerent mathematical properties of the surface and helicity tensors. The surface tensor is a second-rank tensor like the polarizability or inertia tensor, while the helicity tensor is a pseudo-tensor, whose components vanish for molecules having improper rotations as symmetry operations, with the exception of molecules belonging to non-enantiomorphous groups such as Ch, C2v, D2d and S4.42,43 A well-known property mathematically related to the surface tensor is the molecular ordering matrix S,44 generally determined by magnetic resonance experiments. The ordering matrix is also traceless, and its principal axes coincide with those of the surface tensor.The elements Sii have values ranging between 2��� and 1, a positive value denoting the tendency of the i-th molecular axis to align with the director, and a negative value the tendency to be aligned perpendicular to the director. The magnitude of Sii gives the degree of alignment of the i-th molecular axis; thus, the diVerence Sii 2 Sjj measures the diVerent tendency to alignment of the i-th and the j-th axes. Statistical methods allow one to derive an expression for the distortion free energy per unit volume of the sample, due to its elastic fluctuations and the twist deformation exerted by the chiral probe.Under the condition that the induced pitch is much larger than molecular dimensions, minimisation of the distortion free energy leads to the relation in eqn. (2) for the b = RTeQ/2pK22nm (2) twisting power of a specific dopant in a given nematic solvent,14 where T is the absolute temperature, and e, K22 and nm are respectively orienting strength, twist elastic constant and molar volume of the nematic solution (which for very low dopant concentrations are those of the solvent); the chirality order parameter Q is defined by eqn.(3), where x, y, z are the prin- Q = 2(2/3)1/2(QxxSxx 1 QyySyy 1 QzzSzz) (3) cipal axes of the ordering matrix S (and the surface tensor T). Thus the twisting power is proportional to the chirality order parameter, which is essentially a molecular property with a proportionality factor which depends on measurable properties of ‡ Other descriptions are possible, e.g.by defining the surface enclosing the electron density obtained by quantum mechanical calculations,39 or by considering the smoothed surface generated by rolling a sphere over the van der Waals envelope.40 However, in the present case, we have used the widely accepted molecular representation by van der Waals spheres, which has the advantage of being simple and has been used for a number of successful predictions.13,15,16J.Chem. Soc., Perkin Trans. 2, 1999, 411–417 413 the solvent. It should be remembered that the orienting strength parameter e at the temperature of the experiment can be derived from molecular order parameters in the nematic solvent.45,46 In the present case, the solvent is the nematic mixture E7 at T = 300 K, which corresponds to a reduced temperature T/Tc ª 0.9. The proportionality factor between Q and b has been calculated using the values K22 = 8 × 10212 newton and nm = 3 × 1024 m3, reported in the literature.47 In the absence of experimental data on the ordering matrix of our solutes, the parameter e has been taken to be equal to 5 nm22.This choice is justified by the fact that such a value provides an order parameter approximately equal to 0.5 for the long molecular axis of elongated molecules of the dimensions of cyanobiphenyl mesogens,13,45 at a reduced temperature of about 0.9.By using these solvent parameters, a value of about 1000 is predicted for the ratio b/Q, if the twisting power b is expressed in mm21 units and the chirality order parameter Q in nm3 units. In order to gain some understanding of the results expected by the model, we shall consider as an example the case of a disclike object, twisted along the (x,z) axes in the disc plane in such a way as to have the symmetry of a right-handed four-blade propeller. The y-axis is therefore the symmetry axis, and it is a principal axis for both the ordering matrix and the helicity tensor.Because of the axial symmetry and the traceless character of the tensors, the relations (4) and (5) must be obeyed, with (5) Sxx = Szz = 2Syy/2, Qxx = Qzz = 2Qyy/2 (4) Q = 2(3/2)1/2QyySyy (5) following from (4). When dissolved in a nematic environment formed by elongated mesogens, the y-axis of the twisted disc will preferentially align perpendicular to the director, so that Syy must be negative.Since Qyy is positive for a right-handed helical configuration of the probe, the chirality order parameter Q comes out to be positive, and the induced cholesteric helix is predicted to be right-handed according to eqn. (2). Calculated and experimental helical twisting powers Helicenes. The molecular structures were obtained by full geometry optimisation based on ab initio SCF-MO calculations at the 6-31G** level. The calculations were performed with the GAUSSIAN94 program package.48 In all cases overall-twisted structures with skewed phenanthryl units were obtained.The calculated structures are in good agreement with those determined from X-ray diVraction for 149 and other helicenes, 17,18,20,50 and the diVerences between chirality order parameters obtained with the two sets of structures do not diVer more than 10%. Given the mol geometries, surface and helicity tensors were calculated on the basis of molecular surfaces defined as assemblies of van der Waals spheres centred at nuclear positions, in the united atom approximation. Table 1 reports for (M)-1–4 the principal elements of the surface tensor T, in addition to the diagonal elements of the helicity tensor Q in the principal axis system of the surface tensor and in its own principal frame.Furthermore, Table 1 shows the principal values of the orientational ordering matrix S and of the chirality order parameter Q multiplied by 1000, which is the factor b/Q calculated for the E7 solvent.In addition, experimental helical twisting powers are also reported.33 We have labeled as x, y, z the principal axes of the surface tensor T, coincident with the principal axes of the ordering matrix S (see Chart 1). In particular, we have taken as z and y the axes with the most pronounced tendency to stay along the director (corresponding to the largest positive T and S components) and perpendicular to it (largest negative T and S components), respectively.For the molecules under consideration the y axis is perpendicular to the approximate molecular plane. The x and z axes lie on this plane, with x parallel to the C2 or quasi-C2-axis. As seen in Table 1, the preferential alignment of the z axis is not very pronounced in the case of (M)-1 and (M)-3, in contrast with the behaviour of Table 1 Principal elements of the surface tensor T, diagonal elements of the helicity tensor Q in the principal axis system of the surface tensor and in its principal frame, principal values of the ordering matrix S, chirality order parameter Q calculated with e = 5 nm22 and experimental helical twisting powers b taken from ref. 33, for derivatives (M)-1–4 Txx/nm2 Tyy/nm2 Tzz/nm2 Qxx/nm3 Qyy/nm3 Qzz/nm3 Qaa/nm3 Qbb/nm3 Qcc/nm3 Sxx Syy Szz 1000 Q/nm3 b/mm21 (M)-1 0.27 20.56 0.29 0.066 20.084 0.018 0.066 20.087 0.021 0.16 20.35 0.19 236 255 (M)-2 0.16 20.46 0.30 0.047 20.043 20.004 0.049 20.052 0.003 0.03 20.34 0.31 212 29 (M)-3 0.32 20.66 0.34 0.058 20.092 0.034 0.058 20.092 0.034 0.18 20.38 0.20 243 220 (M)-4 0.20 20.53 0.33 0.072 20.074 0.002 0.072 20.089 0.017 0.06 20.35 0.29 225 213 Chart 1 For the rigid compounds 1–4, the approximate directions of the principal axes of the surface tensor are indicated.414 J.Chem. Soc., Perkin Trans. 2, 1999, 411–417 Fig. 1 Projections of (M)-1 on the principal planes of the surface tensor. Under each projection the label of the axis perpendicular to the projection plane is reported.Along the y axis the molecule shows a left-handed screw-like structure (Qyy negative). (M)-2 and (M)-4. In other words, the orientational behaviour of (M)-1 and (M)-3 is that of disc-like molecules, although the orientation with the C2-axis perpendicular to the local director is slightly favoured. The molecules (M)-2 and (M)-4 are biaxial objects with little resemblance to discs or rods: they tend to stay with the director on the molecular plane, but in this plane there is a strongly preferred alignment axis, which for both molecules is perpendicular to the C2 or quasi C2-axis.The principal axes of the helicity tensor Q are denoted as a,b,c, and in general do not coincide with the x,y,z axes unless they are determined by symmetry. Therefore, the two reference frames are expected to share only the C2-axis in the molecules considered here. However, it results from calculations that the other principal axes of Q also lie very close to the corresponding principal axes of T.Fig. 1 shows, as an example, the projections of the molecular structure of (M)-1 on the principal planes of the surface tensor T. Table 1 shows that the four molecules are characterized by a negative helicity along the y-axis, while two positive values are predicted on this plane, one along an axis parallel to the C2 or quasi C2-axis, and the other along an axis perpendicular to it. The latter is in all cases the smallest in magnitude, in agreement with what has already been seen for model biphenyl and binaphthyl systems, where very low helicity is predicted along the bond connecting the two aromatic moieties.12 In agreement with the stereochemical descriptor (M) established for the molecules considered here, a negative value of the helicity tensor component Qyy is predicted. This means that a translation along the y direction of the molecular surface is associated with a left-handed rotation, as occurs for the translation of a lefthanded helix in the direction of the helical axis (see Fig. 1, view along y). The sign convention agrees with that used to describe the configuration of rigid helical molecules (P or M). According to eqn. (4) a negative chirality order parameter Q is calculated, in keeping with experiment. The behaviour of these molecules can be summarised by saying that in the cholesteric phase the axis perpendicular to the molecular plane tends to align along the helix axis, producing twisted phases with the same handedness as the helicity along that axis.The numerical values of Q agree, within a factor of the order 1000 as expected from solvent properties, with the experimental twisting powers. The discrepancies between theoretical and experimental values may be due to several factors: i) the uncertainty in the enantiomeric purity of the dopants; ii) the inadequacy of the geometries calculated for isolated molecules to represent actual structures in the liquid crystal environment; iii) the neglect of longer range interactions, involving electrostatic moments of the molecules, which can be more relevant in the presence of heteroatoms; 46 and iv) the simplified model for the solvent, treated as a continuum.Open-chain derivatives. Encouraged by the successful estimates of twisting powers for rigid helicenes, we attempted the extension to the open-chain derivatives (R)-5–8. The result in these cases depends not only on the absolute configuration, but also on the twist angle a between the phenanthryl units and on the conformation of the substituents. Let’s take as an example (R)-5, which is the simplest of the open-chain derivatives considered here.Prediction of the twisting power would require the following procedure: (i) derivation of the torsional potential, (ii) calculation of the helicity tensor and order parameters, and hence of the chirality order parameter, for various torsional angles, and (iii) a (weighted) average of the Q values over the torsional angle distribution.The crucial point is the first step, which is computationally diYcult, because accurate calculations are required to get torsional pro- files and, even though the results may be reliable for isolated molecules, there are no reasons to believe that they are realistic when the molecules are dissolved in condensed phases. The procedure becomes even more complex for flexible molecules with several torsional angles, such as (R)-6–8.For this reason we did not perform, in the case of the open-chain derivatives, the complete analysis leading to Q values to be directly compared with experimental twisting powers. Instead, with the aim of getting some insight into the complex behaviour of these systems and of highlighting the sensitivity of the Q values towards the molecular conformation, we limited ourselves to the investigation of the stable conformers obtained by full geometry optimisation.Ab initio calculations at the 6-31G** level were performed for the derivatives (R)-5 and (R)-6, while the more complex structures of (R)-7 and (R)-8 were obtained by the semiempirical method PM3. Again the software package GAUSSIAN94 was used.48 In all cases, with the only exception of one of the conformers of (R)-8, the energy minima corresponded to a geometry with the phenanthryl units approximately perpendicular to each other, and much less twisted than in the case of the rigid systems.For the conformers of the open-chain derivatives Table 2 reports the principal values of the surface tensor T and the diagonal elements of the helicity tensor Q in the principal axis system of the surface tensor and in its principal frame. In addition, the principal values of the ordering matrix S and the chirality order parameter Q are shown in the table. As for helicenes, the reference axes are labelled in such a way that z and y correspond to the principal axes of T, with the strongest tendency to being oriented parallel and perpendicular to the nematic director, respectively. It can be seen from Table 2 that the various conformers have similar properties, with the only exception being the A2 conformer of (R)-8, which will be discussed in more detail in the following.It appears that, in contrast with the rigid systems considered above, the open-chain derivatives show a rod-like behaviour with a net tendency to align the axis lying close to theJ. Chem.Soc., Perkin Trans. 2, 1999, 411–417 415 Table 2 Principal values of the surface tensor T, diagonal elements of the helicity tensor Q in the principal axis system of the surface tensor and in its principal frame, principal values of the ordering matrix S and chirality order parameter Q calculated with e = 5 nm22 for selected conformers of derivatives (R)-5–8. A2 and A1 refer to conformers with and without a C2 symmetry axis perpendicular to the phenanthryl–phenanthryl bond.The experimental helical twisting powers b are taken from ref. 33 (R)-7 (R)-8 Txx/nm2 Tyy/nm2 Tzz/nm2 Qxx/nm3 Qyy/nm3 Qzz/nm3 Qaa/nm3 Qbb/nm3 Qcc/nm3 Sxx Syy Szz 1000 Q/nm3 (R)-5 20.12 20.20 0.32 20.151 0.157 20.006 20.156 0.158 20.002 20.16 20.27 0.43 9 (R)-6 20.09 20.20 0.29 20.134 0.154 20.020 20.149 0.154 20.005 20.12 20.27 0.39 17 A2 0.9 20.39 0.30 20.213 0.239 20.026 20.218 0.239 20.021 0.17 20.51 0.34 62 A1 20.06 20.28 0.34 20.205 0.131 0.074 20.063 0.060 0.003 20.09 20.36 0.45 4 A2 0 20.26 0.26 0.162 20.021 20.141 0.162 0.046 20.208 20.04 20.37 0.33 45 A1 20.08 20.28 0.36 20.188 0.193 20.005 20.190 0.203 20.013 20.12 20.35 0.47 18 b/mm21 8 28 7.5 16 Fig. 2 Projections of (R)-5 on the principal planes of the surface tensor. Under each projection the label of the axis perpendicular to the projection plane is reported. Along the y axis the molecule shows a right-handed screw-like structure (Qyy positive). phenanthryl–phenanthryl bond, which is therefore denoted as z axis, parallel to the director.The axis with the strongest tendency to orient perpendicular to the director (y axis) is for these systems the C2 or quasi-C2 axis. It can also be seen that, in analogy with the rigid compounds, the helicity along the axis parallel to the C2 axis is large and positive and that along the axis perpendicular to both the C2 and the phenanthryl–- phenanthryl bond is large and negative.On the contrary, the helicity along the latter bond, which is much smaller in magnitude, changes its sign from positive to negative on passing from the rigid to the flexible systems, as expected in keeping with the change of the conformation from s-cis to s-trans. This dependence of the elements of the Q tensor on the dihedral angle characterising structures with axial chirality is rather general, and has been predicted also for biphenyl and binaphthyl, as well as for model systems.13,14 The simpler of the flexible derivatives is (R)-5, this stable conformer is predicted to have a twist angle a = 90.48 (hereafter a will be defined as the 3–4–49–39 dihedral angle).The general considerations made above appear in a clear way from the comparison of the molecules (R)-5, whose projections on the principal planes of the surface tensor are shown in Fig. 2, and (M)- 4, which is very similar, but is forced by the S–S bond to reduce the angle between the phenanthryl moieties to about 538.In order to show the angular dependence of the chirality order parameter Q for (R)-5, we have considered a number of conformers generated from the minimum energy structure by changing the torsional angle a in a restricted range about 90.48. As shown in Fig. 3, a significant dependence of the chirality order parameter appears, with sign inversion in the proximity of the perpendicular arrangement of the aromatic units.Analogous behaviour of the chirality order parameter as a function of the twist angle has been predicted for biphenyl and binaphthyl.13 In the case of (R)-6, the lowest energy conformer has a twist angle a = 90.68 between the phenanthryl units (and Cring–Cring–- S–CH3 dihedral angles of about 1808). The projections of the molecule on the principal planes of the T and Q tensors are similar to those of (R)-5. As can be seen from the values reported in Table 2, the methyl substituents have the eVect of increasing the chirality order parameter with respect to (R)-5.The presence of longer substituents makes the analysis even more complex in the case of (R)-7 and (R)-8. The lateral chains Fig. 3 Twist angle dependence of the chirality order parameter for (R)-5, calculated with e = 5 nm22.416 J. Chem. Soc., Perkin Trans. 2, 1999, 411–417 give a substantial contribution to the molecular chirality, which can substantially enhance or reduce that of the biphenanthryl skeleton, depending on their conformation.For each molecule geometry optimization was performed with and without C2 symmetry constraint. In the case of (R)-7 the more symmetric conformer (A2) has a = 92.98 (and Cring–Cring–S–CO = 1238), while the structure lacking any symmetry (A1) has a = 94.58 (and Cring–Cring–S–CO dihedral angles equal to 1238 and 21278). The exchange of oxygen and sulfur leads to conformers for (R)-8 quite diVerent from those obtained for (R)-7, with a CS bond longer than the CO bond and the CSC bond angle smaller (1068) than COC (1208).The lowest energy conformers of (R)-8 have a = 74.78 (and Cring–Cring–O–CS = 126.58) for A2 conformer and a = 958 (and Cring–Cring–O-CS equal to 1518 and 21138) for the A1 conformer. Significant diVerences in both orientational and chiral properties between the two conformers of each molecule appear from Table 2. The properties of the A1 conformers and those of the A2 conformer of (R)-7 are analogous to those of (R)-5 and (R)-6.For the A2 conformer of (R)-8 the conformation passes from s-trans to s-cis, and correspondingly the helicity tensor component along the phenanthryl–- phenanthryl bond becomes positive. The structure of this conformer is also reflected by its orienting behaviour, which is diVerent from that of both the other open-chain s-trans conformers and the (s-cis) rigid helicenes: the directions of highest and lowest alignment are respectively the phenanthryl–phenanthryl bond and the normal to the plane containing this bond and the A2 axis.Finally, it can be seen from Table 2 that the chirality order parameters predicted for the A2 structures are high in magnitude and positive in sign even in the case of (R)-8, in spite of its s-cis conformation. This shows how, in the presence of bulky substituents, the twisting power cannot be simply related to the configuration of the biaryl moiety. Conclusions From the comparison between the results of the theoretical calculations and the stereochemical data for the series of helicenes analysed here, a number of significant conclusions can be drawn.First, in all cases the correct handedness of the induced helical macrostructures is predicted unambiguously by the model. In particular, it turns out to be left-handed for the rigid chiral probes (M)-1–4, but it becomes right-handed for the open-chain systems (R)-5–8, where internal flexibility allows the molecules to adopt a s-trans conformation, with the phenanthryl rings at an angle slightly higher than 908. In other words, Q changes sign when the conformation passes from s-cis to s-trans, and the stereochemical descriptor of the biaryl moiety changes from M to P.Secondly, the chirality order parameter Q reproduces quite well, apart from a constant factor which depends on the solvent, the numerical values of the experimental twisting powers.To fully appreciate this result, it should be recalled that the computations are performed on molecular structures represented by van der Waals spheres of standard radii, with no adjustable parameters. The surface chirality model appears therefore a promising technique to assess the absolute configuration of rigid molecules. In the case of flexible systems, such as the (R)-6–8 derivatives, the numerical values calculated for the twisting power depend rather critically on the molecular conformations.In principle it is possible to perform statistical averages over all metastable conformers, but this may be prevented in practice by the diYculty of obtaining the full multidimensional internal potential surface of the molecules. Further complications arise when the barriers separating the conformational sites are too low, because the quantummechanical calculations may not converge to a well-defined geometry. Acknowledgements This work has been supported by the Italian Ministry of the Universities and the Scientific and Technological Research (MURST) for the Program “Liquid Crystals: Structures, Computer Simulations, Properties and Applications”, and by the National Research Council (CNR).References 1 H. J. Krabbe, H. Heggemeier, B. Schrader and E. H. Korte, J. Chem. Res. (M), 1978, 3020. 2 G. Gottarelli, B. Samorì, C. Stremmenos and G. Torre, Tetrahedron, 1981, 37, 395. 3 G. Solladié and R. G. Zimmermann, Angew. Chem., Int.Ed. Engl., 1984, 23, 348. 4 G. Gottarelli and G. P. Spada, Mol. Cryst. Liq. Cryst., 1985, 123, 377. 5 G. Gottarelli, G. P. Spada and G. Solladié, Nouv. J. Chim., 1986, 10, 691. 6 G. P. Spada and G. Proni, Enantiomer, 1998, 3, 301. 7 H.-G. Kuball, Th. Müller, H. Brüning and A. Schönhofer, Mol. Cryst. Liq. Cryst., 1995, 261, 205. 8 H.-G. Kuball, B. Weiß, A. K. Beck and D. Seebach, Helv. Chim. Acta, 1997, 80, 2507. 9 G. Heppke and F. Oestreicher, Z. Naturforsch., Teil A, 1977, 32, 899. 10 G. Heppke and F. Oestreicher, Mol. Cryst. Liq. Cryst. Lett., 1978, 245. 11 J. P. Berthault, J. Billard and J. Jacques, C. R. Seances Acad. Sci. Ser. 3, 1977, 284, 155. 12 A. Ferrarini, G. J. Moro and P. L. Nordio, Liq. Cryst., 1995, 19, 397. 13 A. Ferrarini, G. J. Moro and P. L. Nordio, Mol. Phys., 1996, 87, 485. 14 A. Ferrarini, G. J. Moro and P. L. Nordio, Phys. Rev. E, 1996, 53, 681. 15 L. Feltre, A. Ferrarini, F. Pacchiele and P. L. Nordio, Mol. Cryst. Liq.Cryst., 1996, 290, 109. 16 A. Ferrarini, P. L. Nordio, P. V. Shibaev and V. P. Shibaev, Liq. Cryst., 1998, 24, 219 17 R. H. Martin, Angew. Chem., Int. Ed. Engl., 1974, 13, 649. 18 K. P. Meuer and F. Voegtle, Top. Curr. Chem., 1985, 127, 1. 19 M. S. Newman and D. Lednicer, J. Am. Chem. Soc., 1956, 78, 4765. 20 M. B. Groen, G. Stulen, G. J. Visser and H. Wynberg, J. Am. Chem. Soc., 1970, 92, 7218. 21 D. A. Lightner, D. T. Helfenfinger, T. W. Powers, G. W. Frank and K. N. Trueblood, J.Am. Chem. Soc., 1972, 94, 3492. 22 W. S. Brickel, A. Brown, C. K. Kemp and S. F. Mason, J. Chem. Soc. (A), 1971, 756. 23 W. Hugand and G. Wagnière, Tetrahedron, 1972, 28, 1241. 24 A. J. Moscowitz, Tetrahedron, 1961, 13, 48. 25 J. Tribout, R. H. Martin, M. Doyle and H. Wynberg, Tetrahedron Lett., 1972, 2839. 26 M. Nakazaki, K. Yamamoto and M. Maeda, J. Org. Chem., 1981, 46, 1985. 27 C. M. Kemp and S. F. Mason, Tetrahedron, 1966, 22, 629. 28 A. Dore, D. Fabbri, S. Gladiali and G.Valle, Tetrahedron: Asymmetry, 1995, 6, 779. 29 T. Hayashi, H. Iwamura, M. Naito, Y. Matsumoto, Y. Uozumi, M. Miki and K. Yanagi, J. Am. Chem. Soc., 1994, 116, 775. 30 K. Yamamura, S. Ono, H. Ogoshi, H. Masuda and Y. Kuroda, Synlett, 1989, 18. 31 H. Nakagawa, S. Ogashiwa, H. Tanaka, K. Yamada and H. Kawazura, Bull. Chem. Soc. Jpn., 1981, 54, 1903. 32 F. Mikes, G. Boshart and E. Gil-Av, J. Chem. Soc., Chem. Commun., 1976, 99. 33 G. Gottarelli, G. Proni, G. P. Spada, D.Fabbri, S. Gladiali and C. Rosini, J. Org. Chem., 1996, 61, 2013. 34 M. B. Groen and H. Wynberg, J. Am. Chem. Soc., 1971, 93, 2968. 35 S. F. Mason, Molecular Optical Activity the Chiral Discrimination, Cambridge University Press, Cambridge, 1982, p. 175. 36 L. D. Barron, J. Chem. Soc., Faraday Trans. 2, 1975, 71, 293. 37 R. K. Kondru, P. Wipf and D. N. Beratan, J. Am. Chem. Soc., 1998, 120, 2204. 38 P. G. de Gennes, The Physics of Liquid Crystals, Oxford University Press, Oxford, 1974. 39 C. J. Adam, A. Ferrarini, M. R. Wilson, G. J. Ackland and J. Crain, Mol. Phys., submitted. 40 A. Ferrarini, F. Janssen, G. J. Moro and P. L. Nordio, Liq. Cryst., in the press. 41 Expressions for the elements of the surface and helicity tensors, as well as the details of the model and its implementation can be found in refs. 13 and 15.J. Chem. Soc., Perkin Trans. 2, 1999, 411–417 417 42 R. McWeeny, Symmetry, Pergamon Press, Oxford, 1963. 43 A. Ferrarini and P. L. Nordio, J. Chem. Soc., Perkin Trans. 2, 1998, 455. 44 G. R. Luckhurst, in The Molecular Physics of Liquid Crystals, G. R. Luckhurst and G. W. Gray, Eds., Academic Press, London, 1979. 45 A. Ferrarini, G. J. Moro, P. L. Nordio and G. R. Luckhurst, Mol. Phys., 1992, 77, 15. 46 G. Celebre, G. De Luca and A. Ferrarini, Mol. Phys., 1997, 92, 1039. 47 B. Badur, R. K. Sarna and V. G. Bhide, Mol. Cryst. Liq. Cryst. Lett., 1982, 72, 139; H. Hakemi, E. F. Jagodzinski and D. B. Dupré, Mol. Cryst. Liq. Cryst., 1983, 91, 129. 48 GAUSSIAN94, M. J. Frisch, G. W. Trucks, H. B. Schlegel, P. M. W. Gill, B. G. Johnson, M. A. Robb, J. R. Cheeseman, T. A. Keith, G. A. Petersson, J. A. Montgomery, K. Raghavachari, M. A. Al-Laham, V. G. Zakrewski, J. V. Ortiz, J. B. Foresman, J. Cioslowski, B. B. Stefanov, A. Nanayakkara, M. Challacombe, C. Y. Peng, P. Y. Ayala, W. Chen, M. W. Wong, J. L. Andres, E. S. Replogle, R. Gomperts, R. L. Martin, D. J. Fox, J. S. Blinkey, D. J. Defrees, J. Baker, J. P. Sewart, M. Head-Gordon, C. Gonzalez and J. A. Pople, Gaussian Inc., Pittsburgh, PA, 1995. 49 C. de Rango, G. Tsoucaris, J.-P. Declercq, G. Germain and J. P. Putzeys, Cryst. Struct. Commun., 1973, 2, 189. 50 A review of the crystal structures of helicenes reported in the literature is given in: J. C. Dewan, Acta Crystallogr., Sect. B, 1981, 37, 1421. Paper 8/09593K

ISSN:1472-779X    

DOI:10.1039/a809593k

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

J. Chem. Soc., Perkin Trans. 2, 1999, 419–423 419 Analyzing the origins of receptor–ligand adhesion forces measured by the scanning force microscope Adam Moore,a Philip M. Williams,*a Martyn C. Davies,*a David E. Jackson,b Clive J. Roberts *a and Saul J. B. Tendler *a a Laboratory of Biophysics and Surface Analysis, School of Pharmaceutical Sciences, The University of Nottingham, University Park, Nottingham, UK NG7 2RD. E-mail: Phil.Williams@nottingham.ac.uk; Phone: 144 (0)115 9515063; Fax: 144 (0)115 9515110 b Oxford Molecular Group plc, Medawar Centre, Oxford Science Park, Sandford-on-Thames, Oxon, UK OX4 4GA Received (in Cambridge) 23rd April, 1998, Accepted 15th December 1998 Enthalpic approaches have been shown to be of value in the simulation of scanning force microscope (SFM) force–- distance experiments.We show that for streptavidin, adiabatic mapping with lenient minimization convergence criteria can produce useful data for the comparative analysis of diVerent ligands.The lenient mapping protocol profiles the undocking pathway in a fraction of the time required for other methods presented in the literature. Unbinding pathways and hydrogen bonding patterns for three ligands are predicted in a total of 74 computer-hours using a single processor of a Hewlett-Packard J-210. Hence this method allows the analysis of SFM ligand rupture pathways with a low computational overhead and also importantly suggests further avenues of biophysical experimental investigation and data interpretation.Introduction Molecular interactions are key to the processes of life. It is important to understand the fundamental mechanisms of such interactions to further our comprehension of biological processes. One technique to study the strength of these interactions is the scanning force microscope (SFM) force–distance experiment, 1 which measures the force required to rupture the bonds between receptor and ligand.Utilising this technique intermolecular forces as low as 10 pN, corresponding to individual hydrogen bonds, have been resolved.2 Complementary techniques can be used to elucidate the origin of particular interactions, for example those between amino acid side chains and ligand chemical moieties. This is achieved by experimental techniques such as site directed mutagenesis (e.g. ref. 3) and comparative studies examining the behaviour of ligand analogues (e.g. ref. 4). Computer simulations of these interactions allow the investigation of atomistic behaviour, revealing unique insights into the contributions of individual atoms and bonds to complex formation.However, current techniques for these simulations suVer from an unacceptably heavy demand on computer time to model real systems (multimeric proteins, over millisecond time scales). Here we present a fast computational methodology for the rapid prediction and evaluation of molecular interactions in the scanning probe microscopy force–- distance experiment.The methodology also has broad applications for the study of receptor–ligand interactions, and will be a useful aid in a milieu of biophysical investigations, such as suggesting useful candidate amino acids for site-directed mutagenesis, and aiding in the design of novel drug–receptor interactions. Avidin is a glycoprotein obtained from egg white, streptavidin is a non-glycosylated protein derived from Streptomyces avidinii; these proteins are structurally similar and both bind the vitamin biotin, shown in Fig. 1(a), with very high aYnity.5 The streptavidin–biotin system has an aYnity constant of 1015 l mol21,6 the highest known: it contains four binding sites for biotin, one on each of the monomeric subunits. This high specificity and the ability to bind multiple ligands, coupled with the small size of biotin, has led to it being extensively utilised in bioscience applications. Weber et al.7 have derived the structure of the complex using X-ray crystallography and the binding pocket was shown to involve a number of direct biotin–- amino acid hydrogen bonds.This extensive ligand–receptor hydrogen bond network is seen as the main contributor to the extraordinarily high aYnity of the system. Numerous analogues of biotin exist naturally and many more have been generated through chemical modification. Two are shown in Fig. 1(b) and (c) respectively; desthiobiotin, which lacks the Fig. 1 The structure of biotin and two of its analogues. (a) Biotin (BTN). (b) Desthiobiotin (dBTN), where the lower ring is now open due to the lack of a sulfur atom. (c) Iminobiotin (iBTN), where the carbonyl oxygen of the ureido group is replaced by a nitrogen.420 J. Chem. Soc., Perkin Trans. 2, 1999, 419–423 sulfur atom, and 2-iminobiotin, which has an imino group replacing the keto group of the ureido ring of biotin. The aYnity constant for streptavidin–desthiobiotin is 5 × 1013 l mol21 8 and for iminobiotin it is approximately 108 l mol21.3 The iminobiotin was modelled unprotonated, as this is reported to be the only form bound by streptavidin.9 It is desirable to both predict the rupture forces for a series of ligands and explore the underlying mechanisms.Several workers have investigated experimental ligand rupture data through simulation. This has often been performed through a molecular dynamics (MD) protocol. Grubmüller et al.demonstrated a wide variety of biotin–side chain interactions in their MD investigations of biotin interacting with a water saturated streptavidin monomer.10 Also using MD, Izrailev predicted a slip-stick process of unbinding between avidin and biotin.11 Molecular dynamics have also been used to study the stretching by the SFM of dextran molecules,12 and similarly xanthan. Despite these successes the molecular dynamics approach is not an ideal technique to simulate the rupture experiment since the time scales of simulation and experiment diVer by at least five orders of magnitude.We have previously used adiabatic mapping to simulate the forced undocking of biotin from the streptavidin monomer.13 In adiabatic mapping the separation between receptor and ligand is incrementally increased and then the resulting structure minimized, the minimization being terminated according to a set convergence criterion. The value of the convergence criterion controls the amount of minimization performed.Previously we have used a very strict convergence criterion, demanding a high degree of minimization of the conformation before termination, which necessitates a high computational overhead. Here we further explore the adiabatic technique by investigating the eVects of relaxing the convergence criterion, thus lowering the computational overhead, but also leading to less minimized conformers. This allows the rapid exploration of unbinding phenomena in days, rather than the months of computational time necessitated by other techniques.Results and discussion With the strict convergence criterion the adiabatic mapping of the biotin rupture experiment took a total of 900 hours. With the less strict criterion the simulation took 25 hours. The adiabatic measurements of the rupture pathways of the desthiobiotin and iminobiotin ligands took 25 and 24 hours, respectively. In order to compare the behaviour of the biotin ligand during undocking with diVerent minimization conditions we calculated the root mean square diVerence (RMSD) of the ligand conformations after each increase in separation. For the strict minimization conditions the biotin quickly reaches a maximum RMSD of 0.32 nm at a separation increase of 1.39 nm before decreasing to 0.27 nm at 2 nm, after which it slowly rises to a final value of 0.28 nm.In contrast, with lenient minimization conditions the RMSD of biotin conformation is approximately linear, rising slowly to a maximum of 0.15 nm at the end of the simulation.These results demonstrate the greater degree of conformational flexibility allowed by stricter minimization conditions, as much more of the conformational space and energy landscape is explored. In fact, the RMSD results for the strict run show some correlation with the energy profile of the undocking, which would suggest that the biotin conformation provides a significant contribution to the undocking process.The diVerence in the results suggest that an alternative reaction coordinate was explored using the two diVerent conditions. This may in part be due to the extra degrees of freedom mentioned above, allowing the biotin and streptavidin to rearrange their conformations so that the initial contacts in the binding pocket are maintained for longer in the strict run. As the undocking protocol increases the separation between the constrained atoms it may be that there is a degree of accommodation in the strict run, resulting in both an elongation of the beta barrel of the streptavidin monomer and the alkyl chain of the biotin.In order to investigate the diVerences in the number of iterations needed to undock the biotin from the binding pocket and the conformational freedom of the biotin during undocking, the position of the biotin ring carbonyl and carboxyl moieties relative to the backbone Ca carbon of Phe130 was calculated for both the strict and lenient runs and can be seen respectively in Fig. 2(a) and (b). Comparing the two graphs we can see that the distance from carboxyl and carbonyl (the shaded area) is always greater in the strict run. This indicates that the biotin in the strict run is more extended. There is greater variance in position with respect to separation increase in the strict run, especially comparing the carbonyl atom positions, which shows greater conformational flexibility.We also see that in the lenient run the separation between the biotin and the binding pocket increases at a faster rate than in the strict run. This is illustrated by the diVerence in constrained separation increase that is necessary to increase the separation between the carbonyl oxygen of the ureido ring, O3, and the binding pocket. In the strict run this is 0.84 nm, whilst in the lenient run it is 0.62 nm. This again indicates that a diVerent reaction coordinate was traversed, or that changes in the conformation of the structures in the strict run were made to accommodate the distance restraints, allowing the persistence of initial binding pocket contacts.To assess whether the same interactions occur in each of the undocking pathways followed, we can compare the energy pro- files of the two runs. The strict run energy can simply be plotted against the increase in separation, however due to the diVerence of the conformational flexibility between this and the lenient run, it is necessary to have a consistent point of reference for the pathway.As we have demonstrated previously that the O3 moiety dominates the interaction energetics of this pathway,13 we chose to synchronize the pathways by adjusting for the difference in the position of the carbonyl oxygen (O3) of biotin. The lenient run energy was plotted against this adjusted value. The x axes of the two runs are displayed from the initial position to the moment of undocking for each run.This comparison is shown in Fig. 2(c). It can be seen from this comparison that there are many similar features in both energy profiles, such as the two peaks at 1 and 1.3 nm in the strict run and at 0.6 and 1 nm in the lenient run. The actual values of the energy profiles can be seen to be roughly comparable over the first two-thirds of the graph, but markedly diVerent at the end. We would expect to see the lenient energies being slightly higher due to the reduced minimization applied, and this is indeed the case.The large deviation at the end of the run is due to the streptavidin adopting a higher energy resting conformer, due to the less strict minimization criterion used. These results suggest that the same undocking pathway is followed using both convergence criteria. The hydrogen bonding interactions between the ligands and steptavidin during the unbinding simulations were calculated and can be seen in Fig. 3.If we first compare the strict and lenient biotin runs, we can see that the initial interactions of the carbonyl oxygen (O3) of the biotin ureido group to the binding pocket amino acids break in the same order, and that the interaction between the carboxyl group (O2) and Ser88 is almost identical. The O3–Ser45 interaction is also seen in both runs, however, the last two interactions of the strict run, between O3 and Ser88, and S1 and Thr115 are not seen. This may again indicate a diVerent reaction coordinate in the latter third of the simulation.In general, the O3 interactions break at a lower separation increase in the lenient run. As can be seen by a comparison of the carbonyl atom positions in Fig. 2(a) and (b), this is because the leniently minimized system does not allow the biotin to fully relax and stretch back into the binding pocketJ. Chem. Soc., Perkin Trans. 2, 1999, 419–423 421 Fig. 2 Comparison of strict and lenient biotin position and energies.The position of biotin relative to the distal end of the binding pocket in the strict (a) and lenient (b) runs. The position of the ureido carbonyl oxygen (O3) is indicated by solid lines, the position of the carboxyl moiety of the alkyl (O2) by dashed lines. The shaded area indicates the elongation of the biotin. A dotted line indicates the point where the separation between O3 and the binding pocket reaches 1 nm. The positions are all relative to the Ca of Phe130. (c) The system energy for the strict and lenient runs.The x axes are adjusted to cover the same unbinding pathway. after the ligand has been moved. These findings lead us to conclude that we are exploring approximately the same unbinding pathway in both runs. It can be seen from Figs. 3(b) and 3(c) that the interactions of biotin and desthiobiotin are very similar, as would be expected due to the fact that the O3 carbonyl atom is present in both. The fact that some hydrogen bonds break at a higher separation increase in the desthiobiotin undocking may be due to the extra degrees of freedom of rotation of the ureido group, due to the lack of the ring-closing sulfur atom.The interactions of iminobiotin [Fig. 3(d)] have many similarities to that of the other two ligands, but the initial structure of the binding pocket is diVerent, in that the main contacts made by the ureido ring are between Asp128 and the imino and N2 moieties. Initial contacts are also made between Ser27 and the ureido ring and Ser88 and the carboxyl group, as in the other ligands.During undocking we see contacts between N2 and Ser45 and Val47, and N3 and Tyr43, comparable to those made in the other ligands, but no contact between the imino group and Ser45. Due to the distortion of the energy curves for the lenient runs’ by the convergence criteria used, it is not possible to predict a maximum rupture force for the ligands.14 In order to gain some insight into the forces during the unbinding event we have investigated the energetic contributions of the hydrogen bonding interactions.The hydrogen bond interaction energy was modelled as a Lennard–Jones potential, i.e. a sixth power decaying field [eqn. (1)]. Ehbond = o 1 (Donor 2 Acceptor Separation)6 (1) The sum of the hydrogen bonding energies for each ligand were calculated and are shown in Fig. 3(e). The curves show that, as expected, the biotin and desthiobiotin curves are quite similar, with biotin generally having a higher interaction energy, and a significantly greater energy at the end of the undocking.Iminobiotin has a lower interaction energy, with the hydrogen bonds generally being of shorter duration and having higher donor–- acceptor separations. The total sum of interaction energies for biotin, desthiobiotin and iminobiotin were calculated and found to be in the ratio 1 : 0.7 : 0.5 respectively. The ratio of rupture forces found in experiment are 1 : 0.8 : 0.5,15 using avidin as a receptor.This data then shows qualitative agreement in the ratios, although a diVerent receptor is used they are structurally similar. The total time for this prediction of the three rupture pathways was 74 hours, a significant improvement in computational overhead. Previously we have shown that adiabatic mapping can successfully predict the rupture force for a ligand–receptor interaction. The results presented here indicate that a comparable undocking pathway can be explored using less strict convergence criteria and therefore be analyzed with much reduced computational overhead. Analysis of molecular interactions enforces the importance of hydrogen bonding in the rupture forces for the streptavidin system.The results also show that it is not possible to use this technique to directly predict the rupture force of an SFM experiment. Conclusions We have shown that adiabatic mapping with lenient minimization convergence criteria profiles the same rupture pathway as the slower, strict protocol.However, the energies determined are distorted by the magnitude of the criteria. The lenient mapping protocol profiles the undocking pathway with a much reduced computational overhead compared to other methods presented in the literature. Subsequent analysis of the hydrogen bonding patterns permits the qualitative prediction of interaction energies for the streptavidin protein, but not the prediction of the SFM rupture force.Relative rupture energies for422 J. Chem. Soc., Perkin Trans. 2, 1999, 419–423 Fig. 3 Hydrogen bonding patterns and energies. (a) Streptavidin–biotin undocking with strict convergence criterion. (b) Streptavidin–biotin undocking with lenient convergence criterion. (c) Streptavidin–desthiobiotin undocking with lenient convergence criterion. (d) Streptavidin– iminobiotin undocking with lenient convergence criterion.(e) The hydrogen bonding energy of the three ligands undocked with lenient convergence criterion. three diVerent ligands were predicted in a total simulation time of 74 computer-hours. We have demonstrated a method of improving the significant computational overhead involved in the analysis of SFM ligand rupture experiments. The method is a useful aid to suggest further avenues of experimental investigation and in the interpretation of the results. The techniques are also valid for studying the fundamental process of structural recognition.Experimental From the X-ray crystallographic, or other suitable starting structure, the ligand is undocked from its binding site by increasing the separation between it and its host in increments. After each increment the system is energy minimized with constraints applied to maintain the ligand/receptor separation. The convergence criterion for these minimizations during the undocking is specified as eqn. (2), where D is the current 0.1Dmax for I £ 0.1Imax D = ÏÔ Ì ÔÓ DmaxS I Imax D for I > 0.1Imax (2) convergence criterion, Dmax is the target maximum criterion, I is the iteration number and Imax is a target limit of iterations permitted.D is also constrained with a lower limit of Dmax/10. Convergence is measured as the maximum of the derivatives of the atomic energies (the forces). After each minimization the energy of the system and position of the atoms is calculated and recorded for analysis.The PULMIN algorithm and full details of the constraint terms and general method for this adiabatic mapping are detailed in our first report.13 The starting structures used here are based on the crystallographic data of the streptavidin (monomer)/biotin complex.7,16 Streptavidin/iminobiotin and streptavidin/ desthiobiotin, structures were generated by substitution of the biotin ligand in the crystallographic data. The biotin, desthiobiotin and iminobiotin complexes were energy minimized within the COSMIC(90)17 forcefield using 1461, 341 and 517 iterations, respectively, of a conjugate gradient minimization procedure.Using the PULMIN procedure with the COSMIC(90) force- field the ligands were removed from their receptors and the forces recorded. Biotin was pulled from streptavidin using twoJ. Chem. Soc., Perkin Trans. 2, 1999, 419–423 423 convergence criteria; with Dmax = 5 kcal mol21 Å21† and Imax = 500, and Dmax = 100 kcal mol21 Å21 and Imax = 1000.Desthiobiotin and aminobiotin were analyzed using only the second, less strict criterion. The increment used for each simulation was 0.01 nm and the undocking performed over a total of 3 nm. The mappings were performed using one processor of a Hewlett Packard J-210 workstation (HPUX 10.20). The hydrogen bonding interactions of all runs were analyzed using HBPlus.18 Acknowledgements We thank the University of Nottingham and Oxford Molecular Group plc for a research studentship for A.M. S. J. B. T. is a NuYeld Foundation Science Research Fellow. References 1 N. A. Burnham and R. J. Colton, J. Vac. Sci. Technol., A, 1989, 7, 2906. 2 J. H. Hoh, J. P. Cleveland, C. B. Pratter, J.-P. Ravel and P. K. Hansma, J. Am. Chem. Soc., 1993, 114, 4917. † 1 Calorie = 4.184 Joules (by definition). 3 A. Chilkoti, P. H. Tan and P. S. Stayton, Proc. Natl. Sci. U. S. A., 1995, 92, 1754. 4 G. U. Lee, D. A. Kidwell and R. J. Colton, Langmuir, 1994, 10, 354. 5 L. Chaiet and F. J. Wold, Arch. Biochem. Biophys., 1964, 106, 1. 6 E. P. Diamandis and T. K. Christopoulos, Clinical Chemistry, 1991, 37, 625. 7 P. C. Weber, D. H. Ohlendorf, J. J. Wendolowski and F. R. Salemme, Science, 1989, 243, 85. 8 R. Blankenburg, Biochemistry, 1989, 28, 8214. 9 F. K. Athappilly and W. A. Hendrickson, Protein Sci., 1997, 6, 1338. 10 H. Grubmüller, B. Heymann and P. Tavan, Science, 1996, 271, 997. 11 S. Izrailev, S. Stepaniants, M. Balsera, Y. Oona and K. Schulten, Biophys. J., 1997, 72, 1568. 12 M. Rief, F. Oesterhelt, B. Heymann and H. E. Gaub, Science, 1997, 275, 1295. 13 A. Moore, P. M. Williams, M. C. Davies, D. E. Jackson, C. J. Roberts and S. J. B. Tendler, J. Chem. Soc., Perkin Trans. 2, 1998, 2, 253. 14 E. Evans and K. Ritchie, Biophys. J., 1997, 72, 1541. 15 V. T. Moy, E.-L. Florin and H. E. Gaub, Science, 1994, 264, 415. 16 F. C. Bernstein, T. F. Koetzlw, G. J. B. Williams, E. F. Meyer, Jr., M. D. Brice, J. R. Rodgers, O. Kennard, T. Shimanouchi and M. Tasumi, J. Mol. Biol., 1977, 122, 535. 17 S. D. Morley, R. J. Abraham, I. S. Hawarth, D. E. Jackson, M. R. Saunders and J. G. Vinter, J. Comput. Aided Mol. Des., 1991, 5, 475. 18 I. K. McDonald and J. M. Thornton, J. Mol. Biol., 1994, 238, 777. Paper 8/03061H

ISSN:1472-779X    

DOI:10.1039/a803061h

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

J. Chem. Soc., Perkin Trans. 2, 1999, 425–429 425 Solvent eVects on redox properties of radical ions 1 Mats Jonsson,*† Abdelaziz Houmam, Gloria Jocys and Danial D. M. Wayner Steacie Institute for Molecular Sciences, National Research Council of Canada, Ottawa, Ontario, Canada K1A 0R6 Received (in Cambridge) 4th December 1998, Accepted 28th January 1999 The one-electron reduction potentials of the radical cations of 1,4-diazabicyclo[2.2.2]octane (DABCO) and N,N,N9,N9-tetramethylphenylene-1,4-diamine (TMPD) in propan-2-ol, ethanol, methanol, acetone, acetonitrile and dimethyl sulfoxide have been measured by cyclic voltammetry. Furthermore, the one-electron reduction potentials of 1,4-benzoquinone, 1,4-benzoquinone radical anion, methyl viologen dication and methyl viologen radical cation also have been measured in a number of solvents.The present results, together with previously published data on radical anions, have been used to evaluate solvent eVects in view of the Kamlet–Taft relationship.The main factors aVecting the magnitude of the solvent eVects are the gas-phase redox properties of the corresponding neutral molecule (ionization potential and electron aYnity) and the charge. In general, the magnitude of the solvent eVects on solution redox properties of radical ions decreases with increasing redox stability of the radical ion, reflected by low ionization potential of the corresponding neutral molecule for radical cations and by high electron aYnity of the corresponding molecule for radical anions.Introduction Solvent eVects on reaction kinetics and mechanisms have been a subject of interest for a number of years.2 In recent years, this has also become a topic of interest in the field of radical chemistry. 3,4 An understanding of the eVects of the local environment (which includes solvation) on both reaction kinetics and thermodynamic properties of reactants, products and intermediates is of vital importance for the interpretation and prediction of data for complex chemical systems such as those found in heterogeneous and interfacial systems (e.g., heterogeneous catalysts, enzymes, biological membranes).Furthermore, this knowledge provides a useful basis for the design or optimization of new chemical processes. Properties in solution, e.g., solubility, rates of reactions and free energy and enthalpy of equilibria, can often be described by so called linear free energy relationships (LFER) or linear solvation energy relationships (LSER).5 One of the most successful relationships has been found to be the Kamlet–Taft expression [eqn.(1)], where XYZ is the property of interest, XYZ = XYZ0 1 aa 1 bb 1 sp* 1 hdH (1) XYZ0, a, b, s and h are solvent-independent coeYcients characteristic of the process, a is the hydrogen bond donor (HBD) ability of the solvent, i.e., its ability to donate a proton in a solvent-to-solute hydrogen bond, b is the hydrogen bond acceptor (HBA) or electron pair donor ability to form a coordinative bond, p* is its dipolarity/polarizability parameter and dH is the Hildebrand solubility parameter which is a measure of the solvent–solvent interactions that are interrupted in creating a cavity for the solute.5,6 For some processes any of the coeYcients XYZ0, a, b, s and/or h may be negligibly small, so that the corresponding terms do not play a role in the characterization of the solvent eVects for these processes.This approach has been criticized for not separating specific and non-specific eVects.7 Alternative approaches which separate specific and non-specific eVects have also been elaborated, e.g., by Koppel and Palm2,7,8 and more recently by Drago and co-workers.9,10 † Visiting Scientist 1995/96. Present address: Department of Chemistry, Nuclear Chemistry, Royal Institute of Technology, S-100 44 Stockholm, Sweden. In a recent study we found that the one-electron reduction potentials of some aromatic amine radical cations appear to vary with the dipolarity/polarizability of the solvent (as given by the p* scale).11 Studies on the redox properties of other radical cations have revealed similar trends.12–14 Unfortunately, all these observations are based on measurements in only a few solvents where the diVerence in dipolarity/polarizability is rather small.In order to explore the generality of these eVects and other possible eVects we have extended the solvent dipolarity/polarizability range used in previous studies and have measured the reduction potentials of the radical cations of 1,4-diazabicyclo[2.2.2]octane (DABCO) and N,N,N9,N9-tetramethylphenylene- 1,4-diamine (TMPD) in propan-2-ol, ethanol, methanol, acetone, acetonitrile and dimethyl sulfoxide by cyclic voltammetry.In general, one-electron oxidation of organic molecules initially results in the formation of radical cations that are often very reactive.The relative stability of a radical cation in solution (which often parallels its reactivity) is reflected by its oneelectron reduction potential, E8, [reaction (2)] and, provided the RH RH?1 1 e2 (2) homolytic bond dissociation enthalpies D8(R–H) are similar, by its pKa [reaction (3)]. RH?1 R? 1 H1 (3) The E8 values are approximately related to the corresponding gas-phase ionization potentials, IP, via eqn. (4) where the con- IP ª 4.44(±0.02) 1 E8 1 DG8 solv(RH) 2 DG8 solv(RH~1) F (4) stant, 4.44 (±0.02) eV,15 is the absolute potential of the hydrogen electrode in water,16,17 DG8 solv(RH) and DG8 solv(RH~1) are the free energies of solvation of the neutral molecule and the radical cation, respectively, and F is the Faraday constant.It should be noted that the ionization potential is the enthalpy of ionization at 0 K; thus, the ionization entropy and the temperature correction are neglected in eqn. (4). However, these corrections are assumed to be fairly small.As can be seen from426 J. Chem. Soc., Perkin Trans. 2, 1999, 425–429 Table 1 Experimental and predicted one-electron reduction potentials in V relative to ferrocene Solvent i-PrOH EtOH THF MeOH Acetone MeCN DMF DMSO DABCO~1/DABCO 0.46 (0.46) a (0.44) b 0.43 (0.43) a (0.44) b — 0.44 (0.43) a (0.45) b 0.30 (0.30) a (0.31) b 0.35 (0.35) a (0.34) b — 0.25 (0.25) a (0.24) b TMPD~1/TMPD 20.24 (20.25) c (20.25) d 20.26 (20.26) c (20.26) d — 20.26 (20.27) c (20.26) d 20.28 (20.28) c (20.28) d 20.28 (20.28) c (20.28) d — 20.30 (20.30) c (20.31) d BQ/BQ~2 20.825 (20.75) e (20.82) f — 21.025 (21.04) e (21.11) f 20.565 (20.54) e (20.66) f — 20.885 (20.82) e (20.94) f 20.905 (20.87) e (20.96) f 20.835 (20.83) e (20.90) f BQ~2/BQ22 21.365 (21.46) g (21.37) h — 21.655 (21.75) g (21.66) h — — 21.610 (21.71) g (21.61) h 21.710 (21.81) g (21.70) h 21.700 (21.81) g (21.71) h MV21/MV~1 — — — 20.825 (20.81) i — 20.835 (20.85) i 20.900 (20.89) i 20.920 (20.92) i MV~1/MV — — — 21.282 (21.27) j — 21.240 (21.27) j 21.280 (21.27) j 21.270 (21.27) j a Predicted from eqn.(6). b Predicted from eqn. (8). c Predicted from eqn. (7). d Predicted from eqn. (9). e Predicted from eqn. (15). f Predicted from eqn. 17. g Predicted from eqn. (16). h Predicted from eqn. (18). i Predicted from eqn. (13). j Predicted from eqn. (14). the above equation, the parameters that govern the one-electron reduction potential of a radical cation are the diVerence in solvation free energy between the neutral molecule and the corresponding radical cation and the gas-phase ionization potential of the neutral molecule.Thus, diVerences in redox properties of radical cations in diVerent solvents are due to diVerences in solvation and it is therefore reasonable to expect that solvent eVects on redox potentials should follow linear solvation energy relationships. One-electron reduction of neutral molecules usually results in formation of radical anions [reaction (5)].RH 1 e2 RH~2 (5) The one-electron reduction potential is related to the gas-phase electron aYnity via an equation analogous to eqn. (4). Thus, solvent eVects on the redox properties of radical anions directly reflect diVerences in solvation. Solvent eVects on the oneelectron reduction potentials of a number of neutral species have been thoroughly studied by Shalev and Evans.18 In this work, we have used our redox data for radical cations together with data for radical anions from Shalev and Evans to explore the applicability of the Kamlet–Taft expression to redox properties.Furthermore, we have measured the reduction potentials of 1,4-benzoquinone and the 1,4-benzoquinone radical anion and the first and second reduction potentials of methyl viologen in various solvents in order to study the solvation eVects on the doubly charged anion and cation. We have focused our attention on the applicability of the Kamlet–Taft expression on radical ions without trying to separate specific and non-specific eVects.The objective of this work has been to find practically useful linear solvation energy relationships and to establish the physico-chemical properties of the solute radical ions governing the magnitude of these solvent eVects. Experimental Cyclic voltammetry was performed with a PAR 273A Potentiostat/Galvanostat interfaced to a base PC using the EG&G Model 270 software package.The cell was a standard three electrode setup using a 3 mm diameter glassy carbon working electrode, a platinum coil counter electrode and a reference electrode consisting of a silver wire in a glass tube containing a 0.1 M solution of tetrabutylammonium perchlorate (TBAP) in acetonitrile. All solvents were of the purest grade available (Omnisolv). Acetonitrile was distilled from CaH2 under 1 atm of argon prior to use. N,N-Dimethylformamide was distilled under reduced pressure from CaH2.Acetone and methanol were distilled. THF was refluxed and distilled from P2O5 then from KOH and finally distilled over potassium. Propan-2-ol was refluxed with CaO and distilled. The supporting electrolyte, TBAP, was recrystallized three times from 10% hexane in ethyl acetate and dried in a vacuum oven (40 8C, 10 Torr). TBAP (0.1 M) was used as supporting electrolyte in all solvents except in propan-2-ol where 0.1 M LiClO4 was used. All potentials were measured with respect to ferrocene as an internal reference.Full IR compensation was employed in all measurements. It was assumed that the solvent eVect on the potential of ferrocene was small. For potential calibration ferrocene was used (E8 = 0.69, 0.72 and 0.68 V vs. NHE in acetonitrile, DMF and DMSO, respectively).19 Results and discussion The measured one-electron redox potentials (peak oxidation potential of DABCO and half-wave oxidation/reduction potentials of TMPD, 1,4-benzoquinone, 1,4-benzoquinone radical anion, methyl viologen dication and methyl viologen radical cation) relative to ferrocene are given in Table 1.The solvatochromic parameters a, b, p* and dH for the solvents used in this study are given in Table 2. The standard potentials in Table 1 are reported with respect to ferrocenium/ferrocene which was used as an internal reference. It has been pointed out that there are only small variations in the standard potential of ferrocenium in a number of aprotic solvents (E8 = 0.71 ± 0.03 V vs.NHE).19 Small diVerences in the reported values have been attributed to changes in liquid junction potential. For the purposes of the following discussion, we will assume the redox potential of ferrocene to be invariant with solvent and use the average value suggested by Sawyer and co-workers.19 While this assumption is not rigorously correct, the diVerence in redox potential for ferroceneJ. Chem. Soc., Perkin Trans. 2, 1999, 425–429 427 between diVerent solvents is small compared to the potential diVerences relevant to the analysis below. Radical cations Admittedly, the number of solvents used in this study (six) is not statistically suYcient for a fit to a four parameter equation. Bearing this in mind, multilinear regression of the potentials of DABCO and TMPD result in eqn. (6) and (7). DE8D ABCO = 0.27(±0.02) 2 0.34(±0.04)a 1 0.17(±0.02)b 2 0.93(±0.06)p* 1 0.065(±0.005)dH (6) DE8 TMPD = 20.24(±0.04) 2 0.08(±0.08)a 1 0.05(±0.03)b 2 0.22(±0.12)p* 1 0.01(±0.01)dH (7) The redox potentials predicted from these equations are given in Table 1.For the multilinear regressions we get the following statistical parameters: r2 = 0.999 and F = 873 for DABCO and r2 = 0.98 and F = 13 for TMPD. The standard errors in the predicted potentials are 0.003 and 0.006 V for DABCO and TMPD, respectively. As can be seen in the equation for DABCO, the dominating contributions to the solvent eVect are those of the solvent dipolarity/polarizability and the Hildebrand solubility parameter.Note that the Hildebrand solubility parameter is 10–20 times larger than the other solvatochromic parameters, thus, even though the coeYcient, h, is smaller than the coeYcients a and b, the impact is higher. We have therefore also performed multilinear regression using a reduced form of the Kamlet–Taft expression containing only the two most significant parameters.This resulted in eqn. (8) and (9). DE8D ABCO = 0.41(±0.08) 2 0.40(±0.05)p* 1 0.019(±0.006)dH (8) DE8 TMPD = 20.21(±0.03) 2 0.11(±0.02)p* 1 0.001(±0.002)dH (9) The redox potentials predicted from these equations are also given in Table 1 for comparison. For the multilinear regressions we get the following statistical parameters: r2 = 0.97 and F = 51 for DABCO and r2 = 0.93 and F = 19 for TMPD. The standard errors in the predicted potentials are 0.02 and 0.007 V for DABCO and TMPD, respectively which are comparable to experimental errors in their determinations.In Fig. 1 we have plotted the one-electron reduction potentials of the radical cations of DABCO and TMPD (vs. Fc~1/Fc) predicted from eqn. (8) and (9) against the corresponding experimental numbers. Roughly, the one-electron reduction potentials of the radical cations of both DABCO and TMPD decrease with increasing solvent dipolarity/polarizability. It has previously been suggested that the higher the gas-phase ionization potential, the more localized is the charge on the radical cation which, in turn, leads to a more exergonic solvation of the radical cation.21 Thus, the solvent has a leveling eVect on the reduction potential, i.e.the diVerence in potential Table 2 Solvent parameters used in the Kamlet–Taft equation Solvent i-PrOH EtOH THF MeOH Acetone MeCN DMF DMSO p*a 0.48 0.54 0.58 0.60 0.71 0.75 0.88 1 a a 0.76 0.86 0 0.98 0.08 0.19 00 b a 0.84 0.75 0.55 0.66 0.43 0.40 0.69 0.76 dH b 11.5 12.7 9.1 14.5 9.9 11.9 12.1 12 a Ref. 6.b Ref. 20. between two radical cations decreases when going from the gas phase to solution. This is also reflected by the diVerence in the magnitude of substituent eVects on reduction potentials in diVerent solvents. It is reasonable to expect that a stronger solvation would also result in an increased sensitivity to changes in solvent properties. The gas-phase ionization potential of DABCO is 7.32 eV22 and that of TMPD is 6.1 eV22 and, as expected, the magnitude of the solvent eVects on the one-electron reduction potential of the corresponding radical cations is higher for DABCO.Using these data together with some previously published data 11,12,23 we can qualitatively check the relationship between ionization potential and the magnitude of the solvent eVect for nine radical cations. For most of these radical cations, redox data are only available for two or three solvents.To quantify the solvent eVects we have therefore simply used DE8 Dp* . The magnitude of the solvent eVects on the one-electron reduction potentials of the radical cations and the gas-phase ionization potentials of the corresponding neutral molecules are given in Table 3. The relationship between the magnitude of the substituent eVect and the ionization potential is illustrated in Fig. 2 and the linear trend is given by eqn. (10). DE8 Dp* = 2.8(±0.6) 2 0.46(±0.08)IP (10) The statistical parameters for the linear regression are r2 = 0.82, F = 33 and the standard deviation is 0.13.Despite Fig. 1 The one-electron reduction potentials of the radical cations of DABCO (squares) and TMPD (filled triangles) predicted from eqn. (8) and (9), respectively, plotted against the corresponding experimental values. Table 3 EVects of solvent polarity on the one-electron reduction potential of some radical cations Radical cation CH3OC6H5~1 1,2-(CH3O)2C6H4~1 1,4-(CH3O)2C6H4~1 1,2,4-(CH3O)3C6H3~1 DABCO~1 TMPD~1 (C6H5)2NH~1 C6H5NHCH3~1 C6H5NH2~1 IP 22 8.2 7.8 7.56 7.5 7.32 6.1 7.19 7.32 7.72 DE/Dp* 21.12 a 20.74 a 20.82 a 20.68 a 20.38 20.10 20.64 b 20.45 b 20.76 b a Calculated from data given in ref. 12 and 23. b Calculated from data given in ref. 11.428 J. Chem. Soc., Perkin Trans. 2, 1999, 425–429 the fact that these compounds are of a very diVerent nature (aromatic and non-aromatic amines and methoxybenzenes), we obtain a linear trend. Clearly, this trend agrees with the above suggestion.The two one-electron reduction steps of methyl viologen [reactions (11) and (12)] are relatively insensitive to diVerences in solvent properties. MV21 1 e2 MV~1 (11) MV~1 1 e2 MV (12) Multilinear regression of these two sets of data resulted in reduced Kamlet–Taft equations [eqn. (13) and (14)]. The redox DE8(MV21) = 20.6(±0.2) 2 0.28(±0.15)p* 2 0.001(±0.004)dH (13) DE8(MV~1) = 21.25(±0.2) 2 0.01(±0.2)p*20.0005(±0.006)dH (14) potentials predicted from these equations are also given in Table 1 for comparison.For the multilinear regressions we get the following statistical parameters: r2 = 0.91 and F = 5.2 for MV21 and r2 = 0 and F = 0 for MV~1, i.e. no correlation in the latter case. The standard error in the predicted potentials is 0.02 for MV21. For the first reduction step the solvent eVect is significant which indicates that, as expected, the solvation free energy of the dication, MV21 is more strongly solvent dependent than the solvation free energy of the radical cation, MV~1.Thus, the solvent dependence appears to increase with the charge of the cation. In fact, this agrees with the suggestion that the solvation energy of more strongly solvated species is more sensitive to diVerences in solvent properties. From eqn. (14) and the data in Table 1 we see that the second reduction step is virtually solvent independent which is well in line with the previously presented trend (Fig. 2) since this reduction step corresponds to a low ionization potential. Radical anions Applying the Kamlet–Taft equation to the first and second reduction steps for 1,4-benzoquinone (Table 1) results in eqn. (15) and (16). As for the radical cations, there are not enough Fig. 2 The magnitude of solvent eVects on the one-electron reduction potentials of radical cations plotted against the gas-phase ionization potentials of the corresponding neutral molecules.DE8(BQ) = 21.4(±0.5) 1 0.4(±0.5)a 2 0.09(±0.3)b 1 0.4(±0.8)p* 1 0.02(±0.08)dH (15) DE8(BQ2) = 21.5 1 0.5a 2 0.05b 1 0.09p* 2 0.03dH (16) data for a four parameter equation. For the multilinear regression of the first reduction step we get the following statistical parameters: r2 = 0.96 and F = 6.7. For the second reduction step we do not have enough data to get any statistics. The potentials predicted using eqn. (15) and (16) are also given in Table 1. As can be seen from eqn.(15), the hydrogen bond donor (HBD) ability, a, and the solvent dipolarity/polarizability, p*, are the dominating contributions to the solvent eVects. Using a reduced Kamlet–Taft equation containing only these two parameters we obtain eqn. (17) and (18). DE8(BQ) = 21.4(±0.1) 1 0.45(±0.06)a 1 0.5(±0.1)p* (17) DE8(BQ2) = 21.59(±0.04) 1 0.37(±0.03)a 2 0.12(±0.04)p* (18) For the multilinear regressions we get the following statistical parameters: r2 = 0.95 and F = 28 for the first reduction step and r2 = 0.996 and F = 242 for the second reduction step. The standard errors in the predicted potentials are 0.04 and 0.01 V for the first and second reduction step, respectively.The potentials predicted using eqn. (17) and (18) are also given in Table 1. From eqn. (18) we can draw the conclusion that the solvent dependence on the free energy of solvation of the dianion is stronger than for the monoanion, analogous to the cation case. In the original paper, Shalev and Evans found a correlation between the one-electron reduction potential and the solvent acceptor number (AN).18 Here, we have applied the reduced Kamlet–Taft equation (i.e.versus p* and a) to their original data on the one-electron reduction potentials of 22 nitrobenzenes in five diVerent solvents. The resulting parameters are given in Table 4 along with the corresponding electron aYnities. For the multilinear regressions we get the following range of statistical parameters: r2 = 0.991–0.999 and F = 108–721.As can be seen, the dipolarity/polarizability contribution is relatively invariant with electron aYnity whereas the contribution from the hydrogen bond donor (HBD) ability decreases significantly with increasing electron aYnity (Fig. 3). Fig. 3 The magnitude of the eVects of solvent hydrogen bond donation ability, a, and solvent dipolarity/polarizability, s, on the oneelectron reduction potentials of 22 nitrobenzenes plotted against the gas-phase electron aYnities, EA, of the nitrobenzenes.J.Chem. Soc., Perkin Trans. 2, 1999, 425–429 429 The trends can be described by eqn. (19) and (20). s = 0.55(±0.01) 2 0.06(±0.01)EA (19) (F = 40, r2 = 0.67 and standard error 0.02) a = 0.71(±0.02) 2 0.21(±0.01)EA (20) (F = 203, r2 = 0.91 and standard error 0.03) It is not unexpected that radical anions should be hydrogen bond acceptors. The extent to which they are should be related to the extent of delocalization of the charge; i.e. the charge on a radical anion corresponding to a molecule with high electron aYnity (more positive standard potential) is generally more delocalized than the charge on a radical anion with a lower electron aYnity (more negative standard potential).In the latter case, the more localized charge will lead to a stronger solvent-to-solute hydrogen bond and thereby also to increased sensitivity to the hydrogen bond donation ability of the solvent, a. From Table 4 it is also evident that at higher electron aYnities the dominating contributor to the solvent eVects changes from the hydrogen bond donation ability to the dipolarity/polarizability. The overall solvent eVect on the solvation free energy of the radical anion also decreases with increasing electron aYnity.This is well in line with the observations on the solvent eVects on the solvation free energy of radical cations, i.e. the more stable the radical ion is in itself, the less sensitive it is to variations in the solvent properties.This compensating eVect of p* and a can also be understood in terms of the more localized (i.e. lower EA) anions being better hydrogen bond acceptors. Conclusions In this work we have shown that the solvent eVects on oneelectron reduction potentials of dications, radical cations, neutral molecules and radical anions can be quantitatively Table 4 Electron aYnities and solvent independent coeYcients for nitrobenzenes Nitrobenzene 2,4,6-(CH3)3C6H2NO2 2,3-(CH3)2C6H3NO2 4-CH3OC6H4NO2 2-CH3C6H4NO2 4-CH3C6H4NO2 3-CH3C6H4NO2 C6H5NO2 3-CH3OC6H4NO2 2-FC6H4NO2 4-FC6H4NO2 2-ClC6H4NO2 3-FC6H4NO2 4-ClC6H4NO2 3-ClC6H4NO2 3-CF3C6H4NO2 3-CNC6H4NO2 2-CNC6H4NO2 1,3-(NO2)2C6H4 1,2-(NO2)2C6H4 4-CNC6H4NO2 1,4-(NO2)2C6H4 3,5-(NO2)2C6H3CN EA/eV24 0.70 0.84 0.88 0.90 0.92 0.96 0.99 1.02 1.05 1.08 1.11 1.20 1.22 1.25 1.37 1.51 1.55 1.60 1.60 1.68 1.92 2.19 a 0.61 0.52 0.54 0.51 0.51 0.50 0.48 0.49 0.48 0.46 0.45 0.44 0.44 0.45 0.41 0.41 0.36 0.37 0.30 0.35 0.36 0.24 s 0.52 0.48 0.50 0.48 0.51 0.50 0.50 0.52 0.51 0.49 0.48 0.49 0.48 0.47 0.48 0.44 0.43 0.44 0.45 0.45 0.47 0.43 described by reduced Kamlet–Taft relationships leading to simple predictive relationships.The main factors aVecting the magnitude of the solvent eVects are the gas-phase redox properties of the corresponding neutral molecule (ionization potential and electron aYnity) and the charge. In general, the magnitude of the solvent eVects on solution redox properties of radical ions decreases with increasing redox stability of the radical ion, reflected by low ionization potential of the corresponding neutral molecule for radical cations and by high electron aYnity of the corresponding molecule for radical anions.The solvent sensitivity also increases with charge of the ion. Acknowledgements M. J. thanks the Swedish Natural Science Research Council for financial support. We also thank the referees for pointing out some serious problems in the original manuscript.References 1 Issued as NRCC publication No. 40905. 2 C. Reichardt, in Solvents and Solvent Effects in Organic Chemistry, 2nd edn., VCH, Weinheim, 1988. 3 J. M. Tanko, in Chemistry of Double Bonded Functional Groups; Supplement A3, S. Patai (ed.), Wiley, New York, 1997. 4 L. Valgimigli, J. T. Banks, K. U. Ingold and J. Lusztyk, J. Am. Chem. Soc., 1995, 117, 9966. 5 Y. Marcus, Chem. Soc. Rev., 1993, 409 and refs.therein. 6 M. J. Kamlet, J.-L. M. Abboud, M. H. Abraham and R. W. Taft, J. Org. Chem., 1983, 48, 2877. 7 W. R. Fawcett, in Quantitative Treatments of Solute/Solvent Interactions, P. Politzer and J. S. Murray (ed.), Elsevier, Amsterdam, 1994, pp. 183–212. 8 I. A. Koppel and V. A. Palm, in Advances in Linear Free Energy Relationships, N. B. Chapman and J. Shorter, ed., Plenum, London, 1972, Ch. 5. 9 R. S. Drago, J. Chem. Soc., Perkin Trans. 2, 1992, 1827. 10 R. S. Drago, M. S. Hirsch, D. C. Ferris and C. W. Chronister, J. Chem. Soc., Perkin Trans. 2, 1994, 219. 11 M. Jonsson, D. D. M. Wayner and J. Lusztyk, J. Phys. Chem., 1996, 100, 17539. 12 M. Jonsson, J. Lind, T. Reitberger, T. E. Eriksen and G. Merényi, J. Phys. Chem., 1993, 97, 11278. 13 M. Jonsson, J. Lind, G. Merényi and T. E. Eriksen, J. Chem. Soc., Perkin Trans. 2, 1995, 67. 14 L. Engman, J. Persson, C. M. Andersson and M. Berglund, J. Chem. Soc., Perkin Trans. 2, 1992, 1309. 15 S. Trasatti, Pure Appl. Chem., 1986, 58, 955. 16 R. G. Pearson, J. Am. Chem. Soc., 1986, 108, 6109. 17 There are uncertainties with respect to liquid junction potentials and the standard state of the electron and therefore the term ‘absolute potential’ is not strictly correct. However, to avoid confusion we have adopted this terminology as suggested by Pearson (ref. 16). 18 H. Shalev and D. H. Evans, J. Am. Chem. Soc., 1989, 111, 2667. 19 W. C. Barrette, H. W. Johnson Jr. and D. T. Sawyer, Anal. Chem., 1984, 56, 1890. 20 A. F. M. Barton, Chem. Rev., 1975, 75, 731. 21 D. D. M. Wayner, B. A. Sim and J. J. Dannenberg, J. Org. Chem., 1991, 56, 4853. 22 NIST Standard Reference Database Number 69-March 1998 Release. 23 A. Zweig, W. G. Hodgson and W. H. Jura, J. Am. Chem. Soc., 1964, 86, 4124. 24 S. Chowdhury, E. P. Grimsrud and P. Kebarle, J. Phys. Chem., 1987, 91, 2551. Paper 8/09499C

ISSN:1472-779X    

DOI:10.1039/a809499c

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

J. Chem. Soc., Perkin Trans. 2, 1999, 431–439 431 Acyloxymethyl as a drug protecting group. Part 5.1 Kinetics and mechanism of the hydrolysis of tertiary N-acyloxymethylsulfonamides Francisca Lopes,a Rui Moreira *a and Jim Iley *b a CECF, Faculdade de Farmácia, Universidade de Lisboa, Avenida das Forças Armadas, 1600 Lisboa, Portugal b Department of Chemistry, The Open University, Milton Keynes, UK MK7 6AA Received (in Cambridge) 19th November 1998, Accepted 21st December 1998 Tertiary acyloxymethylsulfonamides undergo hydrolysis via pH-independent and acid- and base-catalysed processes.Reactions are also buVer catalysed for buVer species with pKa values > ca. 10.5. For the pH-independent pathway, hydrolysis takes place via formation of an N-sulfonyl iminium ion. There is no general-base or nucleophilic catalysis in this region. The mechanism of the acid-catalysed process involves pre-equilibrium protonation of the substrate followed by iminium ion formation.General-acid catalysis is not observed. The base-catalysed pathway involves the normal BAc2 mechanism of ester hydrolysis. The buVer-catalysed reaction gives rise to a curved Brønsted plot, with b values of 1.6 and 0.25 for nucleophiles with pKa values <12.5 and >13, respectively. This is indicative of nucleophilic catalysis associated with a change in rate-limiting step from formation of the tetrahedral intermediate for buVer species with pKa > 13 to decomposition of the tetrahedral intermediate for buVer species with pKa < 12.5.Acyloxymethylsulfonamides have similar reactivity to, and follow similar reaction mechanisms as, the corresponding carboxamide derivatives. Semi-empirical PM3 SCF-MO calculations of the heats of formation, DHf, and atomic charges, q, for acyloxymethyl- and chloromethyl-sulfonamides and amides and their derived iminium ions were performed. These reveal that (1) iminium formation from the sulfonamide series is thermodynamically slightly favoured, and (2) the charge diVerence at the nitrogen atom, DqN, between the neutral molecule and the iminium ion is similar for both sulfonamides and amides.Tertiary N-acyloxymethylamides 1 are of particular interest in drug chemistry because of their potential as prodrugs both of secondary amides and of carboxylic acids.1–5 These esters display two diVerent modes of decomposition: solvolysis of the ester moiety to generate the corresponding N-hydroxymethylamide, which can react further to liberate the secondary amide, and rate-limiting formation of a N-acyliminium ion via loss of the carboxylate group (Scheme 1).While the first mode operates in the HO2-catalysed hydrolysis of 1, both the H3O1- catalysed and pH-independent hydrolysis pathways involve iminium ion formation.1,4 Ester prodrugs 1, especially those containing electrondonating substituents on nitrogen or carboxylate leaving groups of pKa £ 3 (e.g.benzylpenicillin), hydrolyse very rapidly,5 making pharmaceutical formulation diYcult. We rationalised that replacing the amide moiety by the more powerfully electron-withdrawing sulfonamide group would identify N-acyloxymethylsulfonamides, 2, as prodrugs able to regenerate a carboxylic acid or sulfonamide parent drug at significantly slower hydrolysis rates. Scheme 1 These esters have the potential to act as prodrugs of secondary sulfonamides, especially those that have their therapeutic eVectiveness reduced as a result of unfavourable physicochemical properties.For example, sumatriptan 3, a 5-HT1 agonist used in the treatment of migraine,6a and some non-peptidic inhibitors of human leukocyte elastase, 4,6b exhibit oral absorption problems leading to sub-optimal concentrations of the drugs at their receptor sites. Unexpectedly, our initial experiments revealed that the rates at which tertiary N-acyloxymethyl-N-alkylsulfonamides 2 hydrolyse to liberate the carboxylic acid and secondary sulfonamide were similar to the corresponding amide derivatives.7 Typically, the ratios of the pH-independent rate constants of compounds 1 and 2, are within the range 1.1–1.3.7 We now report a more detailed kinetic study of compounds 2a–p that is directed towards clarifying the mechanisms of hydrolysis of tertiary N-acyloxymethyl-N-alkylsulfonamides.Compounds 2a–e explore the eVect of the nitrogen substituent, 2f–i the eVect of the arenesulfonyl fragment and 2i–p the eVect of the carboxylate group.Experimental Melting points were determined using a Kofler camera Bock- Monoscop M and are uncorrected. IR spectra were recorded using a Nicolet Impact 400 spectrophotometer. UV spectra were recorded using a Shimadzu UV2100 spectrophotometer. 1H-NMR spectra were recorded in CDCl3 solutions using JEOL JNM-EX 400 and FX90Q spectrometers; chemical shifts are given in ppm relative to Me4Si and J values are given in Hz.FAB mass spectra were recorded using a VG Mass Lab 25-250 spectrometer. Elemental analyses were obtained from Medac Ltd., Brunel Science Park, Englefield Green, Egham, Surrey, UK, and ITQB, Oeiras, Portugal.432 J. Chem. Soc., Perkin Trans. 2, 1999, 431–439 Synthesis N-Aryl-N-benzoyloxymethylsulfonamides 2a–e were synthesised by alkylation of the corresponding N-arylsulfonamide with chloromethyl benzoate. Typically, sodium hydride (1 mol equiv.) was added to a solution of the appropriate secondary sulfonamide in dry DMF.When liberation of hydrogen was complete this solution was injected slowly into a solution of chloromethyl benzoate (1 mol equiv.) in DMF. When the reaction was complete the solvent was removed under vacuum and the residue was subjected to column chromatography. The synthesis of N-alkyl-N-benzoyloxymethylsulfonamides 2f–p was achieved by sulfonamidomethylation of the carboxylic acid with the appropriate N-alkyl-N-chloromethylsulfonamides.7 The latter were prepared by reacting the corresponding secondary sulfonamide with paraformaldehyde in chlorotrimethylsilane. 8 A solution of the appropriate chloromethylsulfonamide (1.1 mol equiv.) in dry THF was added to a suspension of the sodium carboxylate (1 mol equiv.) in THF at room temperature. Upon completion of the reaction, the solvent was evaporated, and the residue treated with water and extracted with dichloromethane. The organic extracts were washed with water, sodium hydrogen carbonate and dried (MgSO4). Evaporation of the solvent gave the crude ester which was purified by column chromatography. 2a: Mp 71–76 8C; nmax/cm21 1710, 1365, 1160; dH 3.67 (3H, s), 5.82 (2H, s), 6.61–7.74 (14H, m); m/z 397 (M1), 276 (M 2 PhCO2), 141 (PhSO2 1), 105 (PhCO1). Found: C, 63.1; H, 4.6; N, 3.5%. C21H19NO5S requires: C, 63.5; H, 4.8; N, 3.5%. 2b: Mp 99–101 8C; nmax/cm21 1715, 1350, 1160, 930; dH 2.29 (3H, s), 5.82 (2H, s), 6.95–7.69 (14H, m); m/z 381 (M1), 260 (M 2 PhCO2), 141 (PhSO2 1), 105 (PhCO1).Found: C, 66.1; H, 5.0; N, 3.7%. C21H19NO4S requires: C, 66.1; H, 5.0; N, 3.7%. 2c: Mp 95–98 8C; nmax/cm21 1725, 1355, 1155; dH 5.85 (2H, s), 7.17–7.74 (15H, m); m/z 367 (M1), 246 (M 2 PhCO2), 141 (PhSO2 1), 105 (PhCO1). Found: C, 65.3; H, 4.7; N, 3.7%. C20H17NO4S requires: C, 65.4; H, 4.6; N, 3.8%. 2d: Mp 73–76 8C; nmax/cm21 1718, 1350, 1170; dH 5.90 (2H, s), 6.80–7.83 (14H, m); m/z 385 (M1), 264 (M 2 PhCO2), 141 (PhSO2 1), 105 (PhCO1).Found: C, 62.1; H, 4.3; N, 3.7%. C20H16FNO4S requires: C, 62.3; H, 4.2; N, 3.6%. 2e: Mp 108–111 8C; nmax/cm21 1715, 1350, 1160; dH 5.77 (2H, s), 7.03–7.66 (14H, m); m/z 403/401 (M1), 282/280 (M 2 PhCO2), 141 (PhSO2 1), 105 (PhCO1). Found: C, 59.5; H, 4.0; N, 3.5%. C20H16ClNO4S requires: C, 59.8; H, 4.0; N, 3.5%. 2f: Mp 65–68 8C; nmax/cm21 1723, 1354, 1170; dH 2.37 (3H, s), 2.97 (3H, s), 5.60 (2H, s), 7.07–7.74 (9H, m); m/z 198 (M 2 PhCO2), 155 (4-MeC6H4SO2 1), 105 (PhCO1).Found: C, 60.8; H, 5.6; N, 4.4%. C16H17NO4S requires: C, 60.2; H, 5.3; N, 4.4%. 2g: Mp 69–71 8C; nmax/cm21 1736, 1345, 1170; dH 2.95 (3H, s), 5.60 (2H, s), 7.07–7.88 (10H, m); m/z 184 (M 2 PhCO2), 141 (PhSO2 1), 105 (PhCO1). Found: C, 59.1; H, 5.0; N, 4.6%. C15H15NO4S requires: C, 59.0; H, 4.9; N, 4.6%. 2h: Mp 89–94 8C; nmax/cm21 1728, 1350, 1179; dH 2.97 (3H, s), 5.57 (2H, s), 7.17–7.77 (9H, m); m/z 341/339 (M1), 220/218 (M 2 PhCO2), 177/175 (4-ClC6H4SO2 1), 105 (PhCO1).Found: C, 52.6; H, 4.3; N, 4.3%. C15H14ClNO4S requires: C, 53.0; H, 4.1; N, 4.1%. 2i: Mp 162–165 8C; nmax/cm21 1726, 1354, 1171; dH 3.05 (3H, s), 5.62 (2H, s), 7.17–8.12 (9H, m); m/z 229 (M 2 PhCO2), 186 (4-NO2C6H4SO2 1), 105 (PhCO1). Found: C, 52.0; H, 3.9; N, 7.8%. C15H14N2O6S requires: C, 51.4; H, 4.0; N, 8.0%. 2j: Mp 144–145 8C; nmax/cm21 1720, 1360, 1170; dH 3.08 (3H, s), 3.85 (3H, s), 5.70 (2H, s), 6.73–7.83 (4H, AA9BB9), 7.93–8.43 (4H, AA9BB9); m/z 380 (M1), 229 (M 2 4-MeOC6H4CO2), 186 (4-NO2C6H4SO2 1), 135 (4-MeOC6H4CO1). Found: C, 50.7; H, 3.8; N, 7.6%.C16H16N2O7S requires: C, 50.5; H, 4.2; N, 7.4%. 2k: Mp 137–140 8C; nmax/cm21 1719, 1349, 1166; dH 2.41 (3H, s), 3.10 (3H, s), 5.73 (2H, s), 7.10–7.77 (4H, AA9BB9), 7.97–8.47 (4H, AA9BB9); m/z 364 (M1), 229 (M 2 4-MeC6H4CO2), 186 (4-NO2C6H4SO2 1), 119 (4-MeC6H4CO1). Found: C, 52.7; H, 4.5; N, 7.7%. C16H16N2O6S requires: C, 52.7; H, 4.4; N, 7.7%. 2l: Mp 167–170 8C; nmax/cm21 1727, 1382, 1177; dH 3.07 (3H, s), 5.77 (2H, s), 7.60–8.50 (8H, m); m/z 399 (M 2 F), 229 (M 2 4-CF3C6H4CO2), 186 (4-NO2C6H4SO2 1), 173 (4-CF3C6- H4CO1).Found: C, 45.9; H, 2.9; N, 6.7%. C16H13F3N2O6S requires: C, 45.9; H, 3.1; N, 6.7%. 2m: Mp 203–210 8C (decomp.); nmax/cm21 2219, 1723, 1381, 1177; dH (d6-DMSO) 3.10 (3H, s), 5.73 (2H, s), 7.80–8.47 (8H, m); m/z 375 (M1), 229 (M 2 4-CNC6H4CO2), 186 (4-NO2- C6H4SO2 1). Found: C, 51.2; H, 3.4; N, 11.1%. C16H13N3O6S requires: C, 51.2; H, 3.5; N, 11.2%. 2n: Mp 172–173.5 8C; nmax/cm21 1734, 1382, 1177; dH 3.13 (3H, s), 5.80 (2H, s), 7.83–8.53 (8H, m); m/z 229 (M 2 4- NO2C6H4CO2), 186 (4-NO2C6H4SO2 1), 150 (4-NO2C6H4CO1). Found: C, 45.8; H, 3.3; N, 9.2%. C15H13N3O8S requires: C, 45.6; H, 3.3; N, 10.6%. 2o: Mp 121–122 8C; nmax/cm21 1749, 1361, 1176; dH 1.42 (6H, s), 2.84 (3H, s), 5.58 (2H, s), 6.58–7.16 (4H, AA9BB9), 7.95–8.26 (4H, AA9BB9); m/z 442 (M1), 229 (M 2 4-ClC6H40OC(Me)2- CO2), 186 (4-NO2C6H4SO2 1), 169 (4-ClC6H4OC(Me)2 1). Found: C, 48.8; H, 4.3; N, 6.3%.C18H19Cl N2O7S requires: C, 48.8; H, 4.3; N, 6.3%. 2p: Mp 62–64 8C; nmax/cm21 1749, 1361, 1177; dH 0.79 (6H, t,J. Chem. Soc., Perkin Trans. 2, 1999, 431–439 433 J = 7.5), 1.09 (4H, sextet, J = 7.5), 1.20–1.43 (4H, m), 2.12 (1H, tt, J = 6.0, 9.0), 2.95 (3H, s), 5.49 (2H, s), 8.07–8.40 (4H, AA9BB9); m/z 372 (M1), 229 (M 2 Pr2CHCO2), 186 (4-NO2C6H4SO2 1). Found: C, 51.8; H, 7.2; N, 7.5%. C16H24N2O6S requires: C, 51.6; H, 6.5; N, 7.5%.Product analysis Products were identified by HPLC by comparison with standards. In selected cases, large scale reactions were carried out and the corresponding products isolated and identified by 1H-NMR. In all cases, the products were the secondary sulfonamide and carboxylic acid. Kinetics Hydrolyses were monitored using both UV spectroscopy and HPLC. In the UV method, reactions were initiated by addition of a 10–15 mm3 aliquot of a 1022 mol dm23 stock solution of substrate in acetonitrile to thermostatted UV cuvettes containing 3 cm3 of the required buVer solution comprising 10% (v/v) of acetonitrile.Ionic strength was maintained at 0.5 mol dm23 with NaClO4. The reactions were monitored at fixed wavelength by following the decrease in absorbance, and the pseudo-firstorder rate constants, kobs, were determined from plots of ln (At 2 A•) versus time. Alternatively, reactions were monitored using HPLC, following either the loss of substrate or the formation of products.Reactions were initiated by injecting ca. 50 mm3 of the appropriate substrate stock solution (1023 mol dm23) to 5 cm3 of the buVer solution. At regular intervals, samples of the reaction mixture were analysed using a system involving a Merck LiChrospher“ RP-8 5 mm 125 × 4 mm column, an isochratic mobile phase comprising acetonitrile–0.2 mol dm23 pH 4.0 sodium acetate buVer (varying from 55: 45 to 70 : 30 depending on the substrate), and a flow rate of 1.0 cm3 min21.Quantitation was obtained by comparison with standards analysed under identical conditions. Good agreement (±5%) between the rate constants determined by both UV and HPLC methods was obtained. SCF-MO calculations These were carried out using the PM3 procedure from the GAMESS suite of programs.9 All structures underwent an initial geometry optimization using the MM2 force field program. Complete geometry optimization (bond lengths, bond angles and dihedral angles) was achieved using the Broyden– Fletcher–Goldfarb–Shanno formulation.10 Results and discussion Kinetic data and pH–rate profiles The influence of pH on the rates of hydrolysis of compounds 2a, c, g and h is shown in Fig. 1. These pH–rate profiles are marked by a broad U-shape indicative of the presence of acidcatalysed, kH1, base-catalysed, kHO2, and pH-independent, ko, processes, corresponding to eqn. (1). Similar pH–rate profiles kobs = ko 1 kH1[H1] 1 kOH2[OH2] (1) have been previously described for the closely related tertiary N-acyloxymethylbenzamides 1.1,4 The rate constant for the pH-independent process, ko, was determined from the pH–rate profile, while the catalytic secondorder rate constants kH1 and kOH2 were obtained from the plots of kobs versus [H1] (e.g.Fig. 2) and [OH2] (e.g. Fig. 3), respectively. The intercepts of these plots were identical to the pHindependent rate constant, ko, obtained from the pH–rate pro- file. Values of ko, kH1 and kOH2 for the esters 2a–p are listed in Table 1.The pH-independent pathway Both the extension of the plateau in the pH–rate profile and the pH-independent pathway rate constant, ko, are clearly dependent on the polar eVect of the nitrogen substituent in compounds 2 (Fig. 1 and Table 1). Thus, the plateau for the N-methyl derivative, 2g, extends over a range of ca. 8 pH units, whereas for the N-phenyl derivative, 2c, the plateau extends over a range of ca. 6 pH units.This reflects the eVect of the N-substituent upon the rate of the pH-independent reaction relative to those of the acid- and base-catalysed processes. The possible pH-independent hydrolysis mechanisms are outlined in structure A. Of these, path (a) would be expected to be independent of, or slightly retarded by, electron-donating substituents in the arenesulfonamide and N-aryl rings, path (b) should be enhanced by electron-withdrawing substituents in the Fig. 1 pH–Rate profiles for compounds 2a (h), 2c (m), 2g (j) and 2h (d) at 25 8C, in aqueous buVers containing 10% (v/v) of acetonitrile.Fig. 2 Variation of pseudo-first-order rate constants, kobs, with [H1] for compounds 2e (s), 2f, (j), 2h (m) and 2p (d) at 25 8C.434 J. Chem. Soc., Perkin Trans. 2, 1999, 431–439 Table 1 Pseudo-first-order rate constants, ko, for the pH-independent, and second-order rate constants, kH1 and kOH2, for the acid- and basecatalysed hydrolyses of acyloxymethylsulfonamides 2 at 25 8C Compound 2a 2b 2c 2d 2e 2f 2g 2h 2i 2j 2k 2l 2m 2n 2o 2p ko/1025 s21 2.36 2.45 1.70; 1.61 a 2.63,b 4.85,c 7.85,d 14.6,e 23.5,f 77.8 g 0.827 0.526 87.0 51.8 17.5; 15.4 a 1.82 0.900 1.40, 0.73,h 2.64,b 4.19 c 8.68 19.4 24.2 9.72 0.118 kH1/1022 dm3 mol21 s21 0.883 1.03 0.455; 0.705 a 1.13,b 1.78,c, 2.67 d 0.373 0.169 11.0 7.34 4.01; 4.32 a 0.510 0.458 0.552, 0.316,h 0.927,b 2.29 c 0.401 0.527 0.479 0.886 1.14 kOH2/1022 dm3 mol21 s21 4.44 5.53 6.69; 8.72 a 11.0,b 19.5,c 23.0 d 5.92 7.66 10.3 12.2 12.3; 31.3 a 21.2 4.50 7.33 5.52,h 12.7,b 17.6 c 133 253 512 29.2 0.547 a In D2O.b 30 8C. c 35 8C. d 40 8C. e 45 8C. f 50 8C. g 60 8C. h 20 8C. arenesulfonamide and N-aryl rings, and path (c) should be enhanced by electron-donating substituents in the arenesulfonamide ring and in the N-aryl ring. For compounds 2f–i, corresponding to a substituent eVect in the arenesulfonamide ring, a Hammett plot (Fig. 4) yields a r value of 21.80 (r2 = 1.0).For the N-aryl derivatives, 2a–e, a Hammett correlation was possible only using the normalised s8 parameters, giving a r value of 21.61 (r2 = 0.99) (Fig. 4). These are consistent with path (c) above, that is, the mechanism presented in Scheme 2 in which the developing positive charge is stabilised by electron-donating substituents. Interestingly, the r value obtained for the arenesulfonamide substituents is of comparable magnitude to that (21.6) reported for the aroyl substituents in the corresponding reaction of tertiary N-aroyloxymethylbenzamides 1.4 This would imply that that the sulfonamidomethyl moiety has the same ability to stabilise the incipient carbocation as the amidomethyl moiety (see below).Fig. 3 Variation of pseudo-first-order rate constants, kobs, with [HO2] for compounds 2c (d), 2i, (j), 2m (m) and 2n (×) at 25 8C. Further evidence in support of an SN1-type mechanism for the pH-independent hydrolysis of acyloxymethylsulfonamides 2 is provided by the following observations.First, the temperature dependence of the solvolysis reaction for compounds 2c and 2k (Table 1) yields values of (243 ± 1) and (250 ± 2) J K21 mol21, respectively, for DS‡. These values, though negative, are within the range observed for other unimolecular ionisation reactions studied in aqueous buVers containing organic solvents. 11 SN2 reactions similar to path (b) in A above commonly have DS‡ values for mixed aqueous-organic solvents in the Fig. 4 Hammett plots for the pH-independent hydrolysis of 2a–e (m), 2f–i (d) and 2i–p (s), at 25 8C. Scheme 2J. Chem. Soc., Perkin Trans. 2, 1999, 431–439 435 region of 2100 J K21 mol21.11 Values of DS‡ for both SN1 and SN2 reactions are more positive in purely aqueous systems 11 and we have previously reported a DS‡ of (29 ± 2) J K21 mol21 in aqueous buVers for the corresponding reaction of 2 (R1 = 4-tol, R2 = Pr, R3 = Ph).7 Second, the solvent isotope eVect, ko H/ko D, of ca. 1.1 obtained for 2c and 2h (Table 1), is similar to those reported for typical ionisation reactions,12,13 though it is also consistent with the SN2 substitution, path (b). Third, no general-base or nucleophilic catalysis by the buVer species was observed (Table 2), in contrast to the general-base catalysis normally associated the pH-independent hydrolysis of simple alkyl esters.14 Fourth, ko values for the sulfonamidomethyl esters 2i–p (Table 1) display a high dependence on the pKa of the carboxylate leaving group, with a Brønsted blg value of 21.33 (r2 = 0.97) (Fig. 5). Significantly, this correlation includes compounds 2o,p which contain sterically hindered carboxylate groups. A similar blg value of ca. 21 was reported previously the esters 1, which also included highly hindered carboxylic acid moieties.1,5 For the pH-independent hydrolysis of RCO2- CH2OMe, for which a SN1 mechanism has been suggested,15 a blg of ca. 20.8 for the carboxylate leaving group can be calculated.The acid-catalysed pathway Using the kH1 values in Table 1 the following observations may be made. First, the acid-catalysed pathway is also sensitive to the electronic eVects of the substituents in the arenesulfonamide and N-aryl rings. Hammett plots (Fig. 6) for the acidcatalysed hydrolysis of esters 2a–e (again using s8) and 2f–i give rise to r values of 21.68 (r2 = 0.94) and 21.43 (R2 = 0.99), respectively. These values are similar to those for the pHindependent solvolysis and indicate development of positive charge in the sulfonamide group in the transition state.In con- Table 2 Dependence of the pseudo-first-order rate constants, kobs, on the buVer concentration for the hydrolysis of 2a and 2h at 25 8C Compound 2a 2h BuVer MeCO2H H2PO4 2 ClCH2CO2H (CF3)2CHOH PrnNH2 CF3CH2OH HC]] ] CCH2OH pH 4.83 7.15 2.20 9.45 10.83 11.19 11.63 11.53 11.79 [BuVer]tot/ 1022 mol dm23 0.10 0.20 0.50 0.02 0.04 0.10 0.05 0.10 0.30 0.50 0.01 0.03 0.05 0.10 0.01 0.03 0.05 0.10 0.01 0.03 0.05 0.07 0.01 0.03 0.05 0.07 0.01 0.03 0.05 0.07 0.01 0.03 0.05 0.07 kobs/ 1025 s21 2.78 3.71 2.76 2.17 3.00 2.42 38.6 41.4 40.6 33.4 24.5 22.6 22.0 21.7 29.6 29.8 29.3 27.8 48.0 72.2 104 155 104 200 293 363 66.7 101 148 237 110 268 491 622 trast, the Hammett plot for the esters 2i–n (Fig. 6) gives rise to a r value of almost zero. Second, the solvent isotope eVects, kH1/ kD1 for the acid-catalysed decomposition of 2c and 2h are 0.6 and 0.9, respectively. Third, the temperature dependence of the reaction for compounds 2c and 2k gives rise to values of DS‡ of (4 ± 3) and (28 ± 4) J K21 mol21, respectively. Furthermore, no general-acid-catalysed hydrolysis is observed (Table 2). These data provide strong support for a dissociative mechanism for the acid-catalysed hydrolysis of esters 2, identical to that for the pH-independent pathway, other than an extra step involving protonation of the substrate prior to iminium ion formation (Scheme 3).From our results it is not possible to ascertain the exact site of protonation—the sulfonamide oxygen, the sulfonamide nitrogen or the carbonyl oxygen atom. Sulfonamides are known to be very weak bases, with Ho values at half-neutralisation ranging from 25.0 to 26.0 depending on the measurement procedure.16 N-Arylsulfonamides are much less basic than their N-alkyl counterparts, with estimated Ho values <8.16 1H-NMR studies of secondary 17 and tertiary sulfonamides18 in fluorosulfonic acid point to the most likely site of protonation as the nitrogen, and not the oxygen, atom.Esters are also very weak bases, with pKa ranging from ca. 26.5 to 27.5.19 Thus, the two possible protonated species derived from 2 are B and C. The r values for the acid-catalysed pathway are derived from Fig. 5 Brønsted plot for the pH-independent hydrolysis of 2i–p at 25 8C. Fig. 6 Hammett plots for the acid-catalysed hydrolysis of 2a–e (m), 2f–i (d) and 2i–p (s), at 25 8C.436 J.Chem. Soc., Perkin Trans. 2, 1999, 431–439 a combination of the r for protonation and that for the decomposition of the protonated species. Protonation of compounds 2 to give the intermediate B would be expected to be enhanced by electron-donating substituents in the arenesulfonamide and N-aryl groups, whereas decomposition of B would not be expected to greatly aVected by such substituents since, on going from B to the sulfonyliminium ion, a significant proportion of the positive charge would remain on the N-atom.Conversely, if hydrolysis were to proceed via C, then substituents in the arenesulfonamide and N-aryl moieties would be expected to have little eVect on the protonation and the final r value would be expected to reflect the dissociation step. The r values of 21.71 and 21.43 observed for 2a–e and 2f–i, respectively, are consistent with either species, though their almost identical values to those for the corresponding pH-independent reaction might suggest C as the protonated species (though not unambiguously so).Unfortunately, the substituent eVect in the ester group is even less informative. The protonation step leading to B would not be expected to be aVected by substituents in the ester group, nor would the decomposition of B (as one oxygen acts as electron-acceptor and the other acts as electron-donor), resulting in an overall r of ca. zero, as observed.An identical outcome would be expected for C, as r values for the protonation and decomposition steps are likely to be of similar magnitude but opposite sign. In contrast, the solvent isotope eVect of ca. 0.8 is somewhat higher than the value of ca. 0.4 usually found for typical fast pre-equilibrium processes,12 and may be an argument in favour of the formation of species B, followed by a rate-limiting intramolecular proton transfer from nitrogen to the carbonyl oxygen.Such rate-limiting proton transfer is likely given that the diVerence in the pKa values of the proton donor and acceptor is small.20 Whichever protonated species is involved we are led to conclude that the acid-catalysed hydrolysis of 2 occurs via an AAl1 mechanism involving formation of a sulfonyliminium ion (Scheme 3). Interestingly, the acid-catalysed hydrolysis of aryloxyethyl propanoates 21 (5, R1 = Et, R2 = Me) and the hydrolysis of aryloxymethyl acetates (5, R1 = Me, R2 = H) in concentrated Scheme 3 sulfuric acid22 yield Hammett r values of 22.07 and 23.06, respectively.These values also are consistent with an AAl1 mechanism, involving the formation and subsequent decomposition of a carbonyl oxygen protonated intermediate. The diVering magnitudes of the r values have been ascribed to the additional inductive stabilisation of the incipient carbocation by the a-methyl, R2, group in 5. These values are somewhat more negative than the r value of 21.71 obtained here for N-aryl substitued compounds 2a–e, which may indicate that, if reaction procedes via the protonated form C, the nitrogen lone pair in 2 is more eYcient in stabilising the incipient carbocation than the phenolic oxygen in 5, despite the similar pKa values of the parent N-arylsulfonamides and phenols.16 This would suggest that NÆS delocalisation of electrons (i.e. 6�7) for compounds 2 is less important than anticipated.16 Base- and buVer-catalysed pathways In the base-catalysed region, the values of kOH2 for series 2a–e, 2f–i and 2i–n (Table 1) define linear Hammett plots with r values of 0.42 (r2 = 0.86), 0.32 (r2 = 0.95) and 1.86 (r2 = 0.99), respectively (Fig. 7). Thus, the rate of HO2-promoted hydrolysis increases with the electron-withdrawing ability of the substituents on the arenesulfonamide, N-aryl and benzoate groups, indicative of negative charge development in the transition state. The change in sign of the Hammett r value for the series 2a–e and 2f–i, as compared with the corresponding r values for the acid-catalysed and pH-independent pathways, points to a change in mechanism.The low magnitude of the r values for the arenesulfonamide and N-aryl groups implies that they are remote from the reaction centre and are consistent with a BAc2 mechanism (Scheme 4, Nu = HO). Similar r values (0.36 and 0.42) have been reported for the BAc2-type hydrolysis of compounds 5.21,22 In contrast, the r value for substituents in the benzoate group is much closer to the value of ca. 2.2 Fig. 7 Hammett plots for the base-catalysed hydrolysis of 2a–e (m), 2f–i (d) and 2i–p (s), at 25 8C.J. Chem. Soc., Perkin Trans. 2, 1999, 431–439 437 reported for the alkaline hydrolysis of benzoate esters in aqueous acetone and aqueous methanol.23 For compounds 2c and 2h, the solvent isotope eVects, kOH2/kOD2, are 0.8 and 0.4 (Table 1), respectively, and the DS‡ values are (253 ± 3) and (272 ± 2) J K21 mol21, respectively, consistent with the BAc2 mechanism depicted in Scheme 4.In the pH region where the reactions are HO2-catalysed, buVer catalysis is also observed (Table 2). The pseudosecond- order rate constants, kB9, obtained from plots of kobs vs. [buVer]t, where [buVer]t is the total buVer concentration, are proportional to the fraction of the free base form of the buVer, fb. Second-order rate constants, kB, for the buVer-catalysed reaction of 2h were calculated from the relationship kB = kB9/fb and are contained in Table 3. The Brønsted plot of these data (Fig. 8) is curved. Such plots are characteristic of nucleophilic catalysis of ester hydrolysis involving a change in rate limiting step (Scheme 4).24 According to this Scheme, the data may be analysed using eqn. (2),24 where b1 and b2 are, respectively, the log(kB/kB8) = b2(pKa 2 pKa8) 2 log{[1 1 10(b2 2 b1)(pKa 2 pKa8)]/2} (2) Brønsted exponents of the right- and left-hand branches of the plot, pKa8 is the pKa of a (theoretical) nucleophile for which k21 = k2 and kB8 is the corresponding rate constant for such a nucleophilic catalyst.The solid line in Fig. 8 is obtained using b1 = 0.25, b2 = 1.6, kB8 = 1 dm3 mol21 s21 and pKa8 = 12.8. For nucleophiles on the left-hand branch of the plot k21 > k2 and step k2 is rate limiting; the value for b of 1.6 is large, but a value of 1.3 has been reported for the nucleophile-catalysed hydrolysis of methyl phenyl carbonate.25 For nucleophiles on the right-hand branch of the plot k21 < k2 and step k1 is rate Scheme 4 Table 3 Second-order rate constants, kB, for the buVer-catalysed hydrolysis of 2h at 25 8C BuVer Propylamine Triethylamine Piperidine Pyrrolidine Trifluoroethanol Propargyl alcohol Chloroethanol Hydroxide pKa a 10.69 10.87 11.28 11.46 12.27 13.39 14.15 15.59 kB/1022 dm3 mol21 s21 0 0.131 0.513 1.58 26.0 271 446 12.3 a Corrected for I = 0.5 mol dm23.limiting; the b value of 0.25 is similar to those reported for ester and carbonate hydrolyses.26 The curvature between the two branches is centred on pKa8, the pKa value at which the leaving group ability of the nucleophile and the sulfonamidemethanol, 8, are equal.The pKa of 8 can be calculated to be ca. 12.9.27 Similar calculation of the pKa of PhCONHCH2OH yields a value of 13.5; this is ca. 0.4 units larger than the experimentally observed value of 13.1.28 Thus, the pKa of 8 is likely to be in the range 12.5–12.9, in excellent agreement with the pKa8 of 12.8 determined experimentally despite the diverse structures of the nucleophiles employed.It is clear that the value for HO2 lies well away from the Brønsted curve, being much less reactive than anticipated from the other nucleophiles. This eVect has been reported previously, for example in the hydrolysis of 4-nitrophenyl acetate.29 Molecular orbital calculations One of the outcomes of the present and previous 4 work is the similar reactivity and susceptibility to substituent eVects displayed by the acyloxymethylsulfonamides 2 and their amide counterparts 1.To gain further insight into these structural eVects on reactivity, semi-empirical SCF-MO calculations were carried out using the PM3 program. Atomic charges, q, and heats of formation, DHf, for the fully geometry optimised structures of acyloxymethylsulfonamides, chloromethylsulfonamides, and their amide counterparts, together with the diVerences, Dq and DDHf, between each molecule and its corresponding iminium ion, are contained in Table 4.For comparison, similar calculations were performed for N-(2- chloroethyl)-, N-(3-chloropropyl)-, and N-(4-chlorobutyl)- sulfonamides, their amide counterparts and their cyclisation products. From these data the following observations are worthy of note. First, iminium ion formation is slightly favoured (lower DDHf) from the sulfonamide derivatives by ca. 0.5–10 kJ mol21 as compared with the analogous amides. Second, iminium ion formation from N-methyl compounds is favoured compared with the N-phenyl compounds, consistent with the pattern of reactivity observed experimentally. Third, the enhanced reactivity of sulfonamides over amides is evident only for the acyloxymethyl and chloromethyl compounds.When the leaving group is more than one carbon atom away from the sulfonamide or amide nitrogen atom amides are predicted to be more reactive than sulfonamides.Fourth, the change in charge density at the N-atom (the N becomes more positive) upon formation of the iminium ion (from acyloxyand chloromethyl substrates), as well as the cyclic ammonium ions (from w-chloroalkyl systems) is almost the same for sulfonamides as for amides. The corresponding changes at the sulfonyl S- or carbonyl C- atoms are very small. This would imply that a sulfonamide N-atom is able to interact with the incipient sp2 carbocationic centre of the iminium ion (and also the sp3 Fig. 8 Brønsted plot for the buVer-catalysed hydrolysis of 2h at 25 8C.438 J. Chem. Soc., Perkin Trans. 2, 1999, 431–439 Table 4 PM3 SCF-MO calculated heats of formation, DHf, and atomic charges, qS/C and qN, for acyloxymethylsulfonamides, chloroalkylsulfonamides, acyloxymethylamides, and chloroalkylamides, and DDHf, DqS/C and DqN for the formation of the derived cations Molecule DHf/kJ mol21 2419.2 2417.6 2525.5 2272.9 2280.1 2387.1 292.9 292.5 2202.9 2106.2 2106.3 2217.0 2129.6 2131.2 2240.6 2153.2 2152.3 2264.5 qS/C 0.318 0.317 2.274 0.314 0.315 2.270 0.307 0.306 2.250 0.305 0.307 2.240 0.304 0.308 2.236 qN 20.076 20.099 20.528 20.068 20.067 20.515 20.069 20.061 20.507 20.066 20.063 20.498 20.067 20.064 20.498 Ion Df /kJ mol21 c 1145.2 1143.1 1131.8 1155.9 1164.4 1155.4 818.8 818.0 809.2 881.6 881.7 880.4 805.6 807.3 823.5 754.4 753.5 778.7 DqS/C 0.006 0.002 0.092 0.007 0.005 0.096 0.011 0.012 0.109 20.010 20.012 0.095 20.017 20.021 0.092 DqN 0.392 0.368 0.341 0.361 0.359 0.328 0.358 0.350 0.317 0.423 0.420 0.382 0.477 0.474 0.441 a Transoid rotamer; b Cisoid rotamer (cisoid and transoid designate the stereochemical relationship between the carbonyl oxygen and the N–CH2 group.c DDHf = DHf (iminium ion) 2 DHf (molecule).J. Chem. Soc., Perkin Trans. 2, 1999, 431–439 439 carbon of the w-chloroalkyl systems) to the same extent as an amide N-atom. Interestingly, a similar conclusion may be made for the ability of the sulfonamide group to interact with the sp2 centre of an aromatic system through comparison of the Hammett sp values 30 for the substituents N(R1)SO2R2 and N(R1)COR2; 0.00 vs. 0.03 for NHCOMe/NHSO2Me; 20.07 vs. 20.01 for NHCOPh/NHSO2Ph; 0.26 vs. 0.24 for N(Me)COMe/ N(Me)SO2Me; 0.27 vs. 0.39 for NHCOCF3/NHSO2CF3; and 0.39 vs. 0.44 for N(Me)COCF3/N(Me)SO2CF3. Thus, it can be concluded that sulfonamides and amides directly linked to an aromatic ring exert essentially the same electronic eVect.This suggests that delocalisation of the sulfonamide nitrogen lone pair to an adjacent sp2 hybridised carbon is more eYcient than NÆS delocalisation. Acknowledgements The authors thank the Fundação de Ciências e Tecnologia for financial support under project PBIC/CEN/1084/92. References 1 J. Iley, R. Moreira, T. Calheiros and E. Mendes, Pharm. Res., 1997, 14, 1634. 2 M. Johansen and H. Bundgaard, Arch. Pharm. Chemi, Sci.Ed., 1981, 9, 43. 3 H. Bundgaard and G. J. Rasmussen, Pharm. Res., 1991, 8, 1238. 4 J. Iley, R. Moreira and E. Rosa, J. Chem. Soc., Perkin Trans. 2, 1991, 563. 5 R. Moreira, T. Calheiros, J. Cabrita, E. Mendes, M. Pimentel and J. Iley, Pharm. Res., 1996, 13, 70. 6 (a) A. K. Scott, Clin. Pharmacokinet., 1994, 27, 337; (b) P. R. Bernstein, D. Andisik, P. K. Bradley, C. B. Bryant, C. Ceccarelli, J. R. Damewood, R. Earley, P. D. Edwards, S. Feeney, B. C. Gomes, B. J. Kosmider, G.B. Steelman, R. M. Thomas, E. P. Vacek, C. A. Veale, J. C. Williams, D. J. Wolanin and S. A. Woolson, J. Med. Chem., 1994, 37, 3313. 7 T. Calheiros, J. Iley, F. Lopes and R. Moreira, Bioorg. Med. Chem. Lett., 1995, 5, 937. 8 R. Moreira, E. Mendes, T. Calheiros, M. J. Bacelo and J. Iley, Tetrahedron Lett., 1994, 35, 7107. 9 M. W. Schmidt, K. K. Baldridge, J. A. Boatz, S. T. Elbert, M. S. Gordon, J. J. Jensen, S. Koseki, N. Matsunaga, K. A. Nguyen, S. Su, T. L. Windus, M. Dupuis and J.A. Montgomery, J. Comput. Chem., 1993, 14, 1347. 10 J. J. P. Stewart, J. Comput. Aided Mol. Des., 1990, 4, 1. 11 G. Kohnstam, Adv. Phys. Org. Chem., 1967, 5, 121. 12 E. K. Thornton and E. R. Thornton, in Isotope Effects In Chemical Reactions, eds. C. J. Collins and N. S. Bowman, Van Nostrand Reinhold, New York, 1970, p. 213. 13 S. L. Johnson, Adv. Phys. Org. Chem., 1967, 5, 273. 14 E. K. Euranto, in The Chemistry of Carboxylic Acids and Esters, ed. S. Patai, Wiley, Chichester, 1969, p. 505. 15 P. Salomaa, Acta Chem. Scand., 1965, 19, 1263. 16 J. F. King, in The Chemistry of Sulphonic Acids, Esters and their Derivatives, eds. S. Patai and Z. Rappoport, Wiley, Chichester, 1991, p. 249. 17 R. G. Laughlin, J. Am. Chem. Soc., 1967, 89, 4268. 18 F. M. Menger and L. Mandell, J. Am. Chem. Soc., 1967, 89, 4424. 19 J. March, Advanced Organic Chemistry, Wiley, New York, 1985, ch. 8. 20 M. Eigen, Angew. Chem., Int. Ed. Engl., 1964, 3, 1. 21 C. D. Hall and C. W. Goulding, J. Chem. Soc., Perkin Trans. 2, 1995, 1471. 22 R. A. McClelland, Can. J. Chem., 1975, 53, 2763. 23 N. S. Isaacs, Physical Organic Chemistry, Longman Scientific, Harlow, 1987, p. 470. 24 P. M. Bond, E. A. Castro and R. B. Moodie, J. Chem. Soc., Perkin Trans. 2, 1976, 68; D. J. Hupe and W. P. Jencks, J. Am. Chem. Soc., 1977, 99, 451. 25 E. A Castro and M. Freudenberg, J. Org. Chem., 1980, 45, 906. 26 E. A. Castro and C. L. Santander, J. Org. Chem., 1985, 50, 3595. 27 D. D. Perrin, B. Dempsey and E. P. Serjeant, pKa Prediction for Organic Acids and Bases, Chapman and Hall, London, 1981, p. 60. For alcohols (RCH2OH) pKa = 15.9 2 1.42Ss*. The s* value for 4-ClC6H4SO2NH can be computed as 2.14 from that of PhSO2NH (1.99) assuming the relative eVect of the 4-ClC6H4 and Ph groups is the same as for 4-ClC6H4SO2 (3.49) and PhSO2 (3.25). Further correction for the N-Me rather than N-H substituent was not attempted. 28 M. Johansen and H. Bundgaard, Acta Pharm. Chemi, Sci. Ed., 1979, 7, 175. 29 W. P. Jencks and M. Gilchrist, J. Am. Chem. Soc., 1962, 84, 2910. 30 O. Exner, in Correlation Analysis in Chemistry, eds. N. B. Chapman and J. Shorter, 1978, Plenum, New York, ch. 10. Paper 8/09047E

ISSN:1472-779X    

DOI:10.1039/a809047e

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

J. Chem. Soc., Perkin Trans. 2, 1999, 441–448 441 Proton chemical shifts in NMR. Part 13.1 Proton chemical shifts in ketones and the magnetic anisotropy and electric field eVect of the carbonyl group Raymond J. Abraham* and Nick J. Ainger Chemistry Department, The University of Liverpool, PO Box 147, Liverpool, UK L69 3BX Received (in Cambridge) 16th November 1998, Accepted 16th December 1998 The proton resonance spectra of a variety of cyclic ketones including 2-tert-butylcyclohexanone, 4-tertbutylcyclohexanone, fenchone, trans-1-decalone, androstan-3-one, androstan-17-one, androstane-3,17-dione and androstane-3,11,17-trione were obtained and completely assigned.These data together with previous literature data allowed the determination of the carbonyl substituent chemical shifts (SCS) in a variety of cyclic molecules. These SCS were analysed in terms of the carbonyl electric field, magnetic anisotropy and steric eVect for long-range protons together with a model (CHARGE6) for the calculation of the two-bond and three-bond eVects.The anisotropic eVect of the carbonyl bond was found to be well reproduced with an asymmetric magnetic anisotropy acting at the carbon atom with values of Dcparl and Dcperp of 17.1 and 3.2 (10230 cm3 molecule21). This together with the electric field eVect of the carbonyl group gave good agreement with the observed proton shifts without the need to invoke any steric eVects. The short range eVects of the carbonyl group (i.e.HCC]] O) were modelled by a cos q function which was found to be dependant on the ring size of the cyclic ketone via the CC(O)C bond angle. This model gives the first comprehensive calculation of the SCS of the carbonyl group. For the data set of ca. 200 proton chemical shifts spanning ca. 2 ppm the rms error of the observed vs. calculated shifts was 0.11 ppm. Introduction The influence of the carbonyl group on the chemical shifts of neighbouring protons has been the subject of considerable debate and some controversy since the beginning of organic NMR and the standard descripton of the C]] O anisotropy (Fig. 1) must be one of the most well known illustrations in NMR.2 Despite this interest there is still no definitive investigation of the carbonyl substituent chemical shifts (SCS) in a suYciently wide variety of compounds to rigorously test the known interactions determining the proton chemical shifts in simple ketones. The early investigations concentrated on the carbonyl anisotropy and Narasimhan and Rogers 3 concluded that the proton chemical shifts in formamide and DMF were entirely due to the C]] O anisotropy.However even the C]] O anisotropy was uncertain as Jackman4 suggested that there is a large diamagnetism in the direction normal to the nodal plane of the p-orbitals whereas Pople’s calculations 5 suggested a paramagnetism centred on the carbon atom, large in the y direction and the largest diamagnetism on the O atom in the x direction (i.e.along the C]] O bond). An authoritative review of these and other early investigations has been given by Pople and Bothner-By.6 In his pioneering treatment of proton chemical shifts, Zurcher 7 was limited to observing only the methyl groups in steroids but concluded that both the C]] O bond anisotropy and the electric field eVect were needed to explain the observed SCS. Zurcher used the McConnell equation 8 to calculate the C]] O anisotropy and also used the carbonyl dipole to calculate the electric field eVect.Due to lack of data Zurcher did not consider near (<4 bonds) protons nor did he need to invoke any steric eVects of the carbonyl group. ApSimon and co-workers,9 again using only the methyl groups of steroids for their data, reformulated the McConnell equation in order to obtain the anisotropy eVects on near nuclei (<3 Å away from the substituent). They also found that both anisotropy and electric field eVects were necessary to predict the SCS of the carbonyl group.Subsequently Homer et al.10 observed that the original McConnell equation was just as accurate in their investigations. Toyne 11 reviewed the literature calculations of the C]] O anisotropy in which the position of the magnetic dipole varied from the carbon atom to the oxygen atom. He concluded that taking the dipole to be approximately mid-way along the C]] O bond at 0.6 Å produced the best results.More recently Schneider et al.12 obtained all the proton shifts in three keto steroids and analysed these SCS in terms of both anisotropy and electric field eVects. They obtained rather large values for the carbonyl anisotropy (see later) and also they were not able to calculate the chemical shifts of the protons vicinal to the carbonyl group. Recently Williamson et al.13 performed similar calculations for the a C–H protons in proteins. They used the known crystal structures of the proteins and included electric field and anisotropic eVects, the latter from both the C]] O bonds and also from the aromatic residues present.They obtained good agreement with the observed data when both the electric field and anisotropy terms were included. As these proton shifts were measured in aqueous solution the electric field eVect is considerably diminished compared to non polar solvents. Again protons vicinal to the C]] O bond were excluded from their treatment.We give here the complete assignment of the proton spectra of 2-tert-butylcyclohexanone (1), 4-tert-butylcyclohexanone (2), fenchone (3), trans-1-decalone (4), androstan-3-one (5), androstan-17-one (6), androstane-3,17-dione (7) and androstane- 3,11,17-trione (8). In addition the spectra of norbornanone (9) and camphor (10) are remeasured. These plus previous literature results provide suYcient data for an analysis of carbonyl SCS based on a previous model of proton chemical shifts.1 In previous parts of this series this model, which is based on simple charge calculations over one, two and three bonds and steric, electric field and anisotropic contributions over > three bonds, has been applied successfully to a variety of saturated hydrocarbons,14,15 haloalkanes 16 and442 J.Chem. Soc., Perkin Trans. 2, 1999, 441–448 ethers.1 We shall show that this model provides a quantitative treatment for carbonyl SCS and that these are due to electric field and anisotropic eVects of which the electric field eVect is the major contributor.Theory As the theory has been detailed previously 1,17 only a brief summary of the latest version (CHARGE6) is given here. The theory distinguishes between substituent eVects over one, two and three bonds, which are attributed to the electronic eVects of the substituents, and longer range eVects due to the electric fields, steric eVects and anisotropy of the substituents.The CHARGE scheme calculates the eVects of atoms on the partial atomic charge of the atom under consideration, based upon classical concepts of inductive and resonance contributions. If we consider an atom I in a four atom fragment I-J-K-L the partial atomic charge on I is due to three eVects. There is an a eVect from atom J given by the diVerence in the electronegativity of atoms I and J, a b eVect from atom K proportional to both the electronegativity of atom K and the polarisability of atom I and a g eVect from atom L given by the product of the atomic polarisabilities of atoms I and L.The important carbon g eVect (i.e. CCCH) is parametrised separately and is given by a simple cosq dependance where q is the CCCH dihedral angle. There are also routines for the methyl g eVect and for the decrease in the g eVect of the electronegative oxygen and fluorine atoms for CX2 and CX3 groups. The total charge is given by summing these eVects and the partial atomic charges (q) converted to shift values using eqn.(1). d = 160.84q 2 6.68 (1) The eVects of more distant atoms on the proton chemical shifts are due to steric, anisotropic and electric field contributions. H ? ? ? H steric interactions were found to be shielding and X? ? ? H (X = C, F, Cl, Br, I) interactions deshielding according to a simple r26 dependance [eqn. (2)], where aS is a coeYcient for the steric eVect of the atom. dsteric = aS/r6 (2) Furthermore any X ? ? ? H steric contributions on a methylene or methyl proton resulted in a push-pull eVect (shielding) on the other proton(s) on the attached carbon.The eVects of the electric field of the C–X bonds (X = H, F, Cl, Br, I, O) were calculated from eqn. (3) where AZ was determined as 3.67 × 10212 esu (63 ppm au) and EZ is the component of the electric field along the C–H bond. The electric field for a univalent atom (e.g. fluorine) is calculated as due to the charge on the fluorine atom and an equal and opposite charge on the attached carbon atom.The vector sum gives the total electric field at the proton concerned and the component of the electric field along the C–H bond considered is EZ in eqn. (3). This del = AZEZ (3) procedure is both simpler and more accurate than the alternative calculation using bond dipoles. The magnetic anisotropy of the C–C bond was originally included using the McConnell equation [eqn. (4)] for a bond with cylindrical symmetry as illustrated in Fig. 1 for the carbonyl group. dan = DcC–C (3cos2f 2 1)/3R3 (4) In eqn. (4) R is the distance from the perturbing group to the nucleus of interest in Å and is taken from the mid-point of the perturbing group for a symmetric bond such as the C–C bond, f is the angle between the vector R and the symmetry axis and DcC–C the molar anisotropy of the C–C bond. (DcC–C = cC parl 2 cC perp) where cC parl and cC perp are the susceptibilities parallel and perpendicular to the symmetry axis respectively.These contributions were then added to the shifts of eqn. (1) to give the calculated shift of eqn. (5). dtotal = dcharge 1 dsteric 1 dan 1 del (5) Application to the carbonyl group The vicinal (HCC]] O) eVects are treated separately in CHARGE and these will need to be evaluated from the observed data. The carbonyl group also has in principle steric, electric field and anisotropic eVects on protons more than three bonds distant, thus all these have to be incorporated into the model.The steric eVects of both the carbonyl carbon and oxygen atoms are not known and therefore a value of the coeYcient aS in eqn. (2) for these atoms must be determined. We assume that the ketone carbon atom has a similar steric eVect to a saturated carbon, thus the same value of aS is used. The value of aS for the carbonyl oxygen atom is unknown and needs to be obtained. This and the associated push-pull coeYcient are the only additional parameters required for the steric eVect.The electric field of the carbonyl group is calculated in an identical manner to that for any C–X bond. The electric field is calculated as due to the charge on the oxygen atom and an equal and opposite charge on the carbon atom. As the oxygen charge is already calculated in CHARGE and the coeYcient in eqn. (3) is known the electric field eVect is given immediately without any further parametrisation. The anisotropic eVect of the carbonyl group also needs to be calculated.The C]] O group is not an axially symmetric group and has diVerent magnetic susceptibilities (c1,c2 and c3) along the X1, X2 and X3 axes respectively (Fig. 2). There are two anisotropy terms required for a non-axially symmetric group and thus the full McConnell equation [eqn. (6)] must be used. dan = [Dc1(3cos2q1 2 1) 1 Dc2(3cos2q2 2 1)]/3R3 (6) In eqn. (6) q1 and q2 are the angles between the radius vector R and c1 and c3 respectively and Dc1 (c1 2 c2) and Dc2 (c3 2 c2) are the two anisotropies for the C]] O bond which may be termed the parallel and perpendicular anisotropy respectively. In order to apply this calculation to ketones the two anisotropies need to be determined and also it is necessary to determine the eVect of the position of the magnetic dipole along the C]] O bond.Fig. 1 Representation of the anisotropy in an axially symmetric molecule. Note, the signs refer to the change in the d values, not to the shielding.Fig. 2 The principal axes of the carbonyl bond.J. Chem. Soc., Perkin Trans. 2, 1999, 441–448 443 Experimental Materials 2-tert-Butylcyclohexanone (1) was synthesised by the oxidation of 2-tert-butylcyclohexanol (Aldrich Chem. Co.) using chromic acid. 4-tert-Butylcyclohexanone (2), fenchone (3), trans-1- decalone (4), norbornanone (9) and camphor (10) were also obtained from Aldrich. 5a-Androstan-3-one (5), 5a-androstan- 17-one (6), 5a-androstane-3,17-dione (7) and 5a-androstane- 3,11,17-trione (8) were kindly donated by GlaxoWellcome.The solvents were obtained commercially, stored over molecular sieves and used without further purification. Spectrometers and spectral details 1H and 13C NMR spectra were obtained on a Bruker AMX400 spectrometer operating at 400.14 MHz for proton and 100.63 MHz for carbon. Spectra for 2, 4, 5 and 7 were recorded on a Varian 600 (EPSRC service, Edinburgh University) and 7 and 8 on a Varian 750 MHz spectrometer (GlaxoWellcome). HMQC, HMBC and NOE experiments were carried out on the Varian 750 MHz spectrometer.Spectra were recorded in 10 mg cm23 solutions (1H) and ca. 50 mg cm23 (13C) with a probe temperature of ca. 25 8C in CDCl3 and referenced to TMS unless otherwise stated. Typical 1H conditions were 128 transients, spectral width 3300 Hz, 32k data points, giving an acquisition time of 5 s and zero-filled to 128 k to give a digital resolution of 0.025 Hz. 2D Experiments were performed on the AMX400 and the Varian 750 MHz spectrometers using the standard Bruker COSY-DQF and HXCO-BI and the standard Varian HMQC and GHMQC-DA pulse sequences.18,19 The geometries of the compounds investigated were obtained by geometry optimizations using the GAUSSIAN94 programme at the RHF/6- 31G* level.20 Full details of these optimizations and geometries are given in ref. 21. The GAUSSIAN94 calculations were performed on the University of Liverpool Central Computing facility, and the CHARGE computations were performed on a PC.Compound assignments The assignments of compounds 1–10 are given in Tables 2 to 5. 2-tert-Butylcyclohexanone (1). The 1H spectra of 1 in CDCl3 consists of a number of complex patterns which were assigned from a HET-CORR experiment with the aid of a literature 13C assignment.22 This gave the assignment of H2 (d 2.15) uniquely and the assignments of the protons of the various CH2 groups. The assignments of the 3, 4 and 5 axial and equatorial protons were made on the basis of their fine structure.The 3e, 5e and 4e protons have complex splitting patterns centered at d 2.18, 2.06 and 1.90 respectively, the 4a, 5a and 3a protons have characteristically axial splitting patterns centered at d 1.64, 1.66 and 1.47 respectively. The 6e and 6a protons give a strongly coupled multiplet centred at d 2.28 and inspection of the splitting pattern suggests that H6e is to lower field. This provisional assignment contrasts with that predicted from the calculations (Table 3) in which the two protons are reversed. 4-tert-Butylcyclohexanone (2). The H2a and H2e protons are easily assigned as they are the most low field and further examination of the splitting pattern (again an AB type) shows that H2e is at d 2.356 and the H2a at d 2.272. The H3e proton is at d 2.079 but even at 600 MHz the H3a and H4a protons are coincident at d 1.450. Fenchone (1,3,3-trimethylbicyclo[2.2.1]heptan-2-one) (3).The assignment of this compound was straightforward, the only diYculty encountered was the assignment of the 7syn and 7anti protons. This was performed by examining the NOE from the 3exo methyl group, assuming that there would be an NOE to the 7syn proton but not to the 7anti which formed the basis of the assignment. From this experiment we assign 7syn at d 1.80 and 7anti at d 1.54. The H4 proton is a multiplet with integration 1 centered at d 2.14, the 5x, 5n, 6x and 6n protons were all assigned by analysis of splitting patterns and examination of a HET-CORR spectrum using a literature 13C assignment.23 trans-1-Decalone (4).The assignment of this compound was performed by a variety of methods, the analysis of the AB pattern at d 2.24–2.32 corresponding to the 2a and 2e protons was carried out using the LAOCOON programme.24 The results of these analyses are reported separately.25 The other protons were assigned by connectivity (HMBC), coupling (COSYDFTP) and H–C correlation (HMQC) experiments. 5·-Androstan-3-one (5). The 600 MHz spectrum of this compound consists of 30 closely coupled protons over a range of 2.4 ppm. Analysis of the multiplets between d 2.40 and 2.22 shows that unusually the axial 2b proton is downfield of the equatorial 2a proton, due to the combined deshielding eVects of the axial C19 methyl group and the vicinal 3-keto group. Further analysis of COSY and HET-CORR experiments at 750 MHz confirms the previous assignment given by Schneider 12 of the 400 MHz spectrum though in ref. 12 only the SCS were given. 5·-Androstan-17-one (6). The assignment of this compound has also been reported previously 12 though again only the SCS were given. Again analysis of COSY and HET-CORR experiments at 750 MHz confirms the assignment. 5·-Androstane-3,17-dione (7). The lowfield part of the 1H spectrum reveals two well separated AB patterns due to the C2 and C16 protons and a HET-CORR plot together with a previous 13C assignment 23 showed that the 16b proton is the most downfield.A strong correlation with this proton in the COSY plot identified the 16a and C15 protons. Analysis of the splitting patterns assigned 15a at d 1.946 and 15b at d 1.520. The COSY correlations of the C15 protons assigned the H14 at d 1.294 and this process was repeated for all the ring protons. These assignments were confirmed from a calculated spectrum using the Bruker WIN-DAISY programme18 of all the protons in this compound except the H6 and H7 protons which even at 600 MHz are a very strongly coupled multiplet.The results of this analysis are reported elsewhere.25 5·-Androstane-3,11,17-trione (8). Although this is the most substituted of the 5a-androstanes studied, the spectrum of this compound showed considerable overlap at 400 MHz and thus the spectrum was obtained at 750 MHz. This together with the 13C spectrum, COSY, HMQC and HMBC experiments were suYcient to obtain a complete assignment of this compound.Again a detailed analysis including the coupling constants is given in ref. 25. The spectra of 9 and 10 were also re-examined in detail because of the importance of these compounds in the parametrisation (see later). Norbornanone (9). The proton spectrum of 9 was given previously 26 and the assignment was confirmed by a COSY plot. Camphor (10). The assignment of the proton spectrum of 10 has been the subject of some controversy.26–28 Both COSY and HET-CORR experiments were performed in order to check the assignment.Sanders and Hunter 27 assigned all the protons in this molecule including the three methyl groups on the basis of a series of elegant NOE experiments and this assignment was subsequently confirmed by Kaiser et al.28 Our experiments also444 J. Chem. Soc., Perkin Trans. 2, 1999, 441–448 Table 1 Proton SCS for the 3-keto, 11-keto and 17-keto group in 5a-androstane 3-keto a 11-keto a 17-keto a Proton 1a 1b 2a 2b 3a 3b 4a 4b 5 (CH) 6a 6b 7a 7b 8 (CH) 9 (CH) 11a 11b 12a 12b 14 (CH) 15a 15b 16a 16b 17a 17b 18-Me 19-Me This work 0.48 1.67 0.81 0.99 —— 0.86 1.05 0.49 0.10 0.10 0.05 0.07 0.05 0.07 0.03 0.12 0.03 0.02 0.03 0.02 0.03 0.02 0.03 0.03 0.04 0.04 0.24 Ref. 12 0.45 1.66 0.77 0.96 —— 0.84 1.02 0.45 0.11 0.11 0.03 0.04 0.05 0.07 0.02 0.13 0.02 0.02 0.02 cc 0.03 0.03 c 20.02 0.03 0.23 This workb 20.15 0.74 0.04 0.06 ——c 0.02 20.05 0.03 c 0.19 0.15 0.34 0.94 —— 1.05 0.61 0.62 0.14 0.11 0.18 0.09 —— 20.07 0.18 Ref. 12 20.13 0.74 cc 20.04 20.02 cc 20.07 cc 0.19 0.10 0.36 1.00 —— 1.15 0.54 0.64 0.12 0.08 0.16 0.16 0.22 0.03 20.03 0.22 This work 0.04 20.02 cc 20.03 c 0.07 0.03 0.05 0.03 0.03 0.06 0.09 0.28 0.04 0.14 c 0.14 0.09 0.37 0.30 0.37 0.48 0.82 —— 0.17 0.02 Ref. 12 c 0.01 cccc 0.07 0.07 c 0.03 0.03 0.04 0.09 0.26 0.03 0.12 c 0.12 0.08 0.37 0.27 0.35 0.49 0.89 —— 0.17 0.02 a d(ketone) 2 d(androstane). b d(8) 2 d(7). c SCS < 0.01 ppm.confirmed this assignment though the chemical shifts measured here diVer slightly from those recorded previously. Results The above data combined with the proton chemical shifts of the parent compounds given previously15 allow the carbonyl SCS to be obtained in these compounds. The carbonyl SCS for 4 vs. trans-decalin and 9 vs. norbornane are given in Fig. 3. Also the SCS for the carbonyl group at the 3, 11 and 17 positions in the steroid nucleus obtained here from the data for compounds 6, 7 and 8 together with the proton chemical shifts of androstane are given and compared with the results obtained by Schneider et al.12 in Table 1.In ref. 12 only the SCS were tabulated not the actual proton chemical shifts. Also the SCS for the 11-keto group has been obtained in this investigation as d(8) 2 d(7) whereas Schneider et al.12 obtained this SCS directly from the analysis of 11-keto androstane. The excellent agreement of the two sets of results in Table 1 is impressive and the additivity of the SCS values in the steroid nucleus is very clearly shown by the agreement of the two sets of values for the SCS of the 11-keto group.The carbonyl SCS in these well defined systems are of some interest. In general the g eVect of the carbonyl oxygen atom (i.e. HCC]] O) is strongly deshielding with however an orientational dependance. For example, in trans-decalone the SCS of the carbonyl group on H2ax (1.07) and H9 (1.05) is significantly greater than on H2eq (0.69) and this pattern is reproduced in the cyclohexanes and steroids.In contrast in norbornanone the SCS of Fig. 3 Carbonyl SCS in trans-1-decalone and norbornanone. the carbonyl on H3endo (0.68) is similar to that on H3exo (0.59) and again this is observed in camphor. The long range (>3 bonds) eVects of the carbonyl group are also large and extend over both the bicycloheptene and decalin system. The eVects are usually deshielding with only the 5ax and 6ax protons in trans-decalone showing an upfield shift.This pattern is also observed in the steroid nucleus (Table 1) where very few of the protons show an upfield SCS and these shifts are usually very small with the proton far removed from the keto group. The only marked exception to this is the SCS of the 11-keto group at the 1a proton (20.15 ppm) and this is accompanied by a large positive SCS (0.74 ppm) at the 1b proton. The combined eVect of these shifts is so large that these two methylene protons occur at the two extremes of the proton spectrum in 8 (apart from the methyl groups).We shall show that these shifts may be completely explained by our present theories. The data collected in Tables 2–5 provide a rigorous test of the application of both the CHARGE model and also present theories of carbonyl SCS to these compounds. The compounds listed in the tables are all of fixed conformation with the possible exception of the five membered rings of cyclopentanone and ring D of the steroid nucleus, which may exhibit some conformational flexibility.The GAUSSIAN94 (6-31G*) calculations gave the cyclopentanone geometry as the half-chair (Cs) conformation in agreement with both molecular mechanics (PCMODEL) calculations 31 and the experimental gas phase geometries.32 Similar calculations for the saturated ring D of androstan-3-one gave the same geometry as obtained for androstane,15 i.e. as a C13-envelope with C14, C15, C16 and Fig. 4 Nomenclature used for 5a-androstan-17-one.J. Chem. Soc., Perkin Trans. 2, 1999, 441–448 445 Table 2 Observed vs. calculated proton chemical shifts (d) of acyclic and cyclic ketones Compound Acetaldehyde Acetone Cyclopentanone Pinacolone Cyclohexanone Me CHO Me Ha Hb Me tBu H2,6 H3,5 H4 Obs.a 2.20 9.78 2.17 2.17 1.98 2.14 1.13 2.33 1.88 1.71 Calc. 1.96 9.70 1.83 2.22 1.93 1.88 1.26 2.24 1.82 1.77 Compound trans-1-Decalone 2a 2e 3a 3e 4a 4e 5a 5e 6a 6e 7a 7e 8a 8e 9 10 Obs.b 2.33 2.36 1.67 2.05 1.43 1.77 1.15 1.79 1.18 1.70 1.14 1.79 1.25 1.91 1.95 1.37 Calc. 2.27 2.34 1.69 2.05 1.34 1.82 0.98 1.63 1.27 1.69 1.20 1.67 1.34 1.77 1.84 1.31 a Ref. 29. b This work. Table 3 Observed vs. calculated chemical shifts in substituted cyclohexanones 2-Methylcyclohexanone 3-Methylcyclohexanone 4-Methylcyclohexanone 2-tert-Butylcyclohexanone 4-tert-Butylcyclohexanone Proton 2a 2e 3a 3e 4a 4e 5a 5e 6a 6e tBu Obs.a 2.43 — 1.38 2.10 1.67 1.84 1.67 2.07 2.30 2.37 — Calc. 2.30 — 1.44 1.82 1.65 1.93 1.64 2.01 2.21 2.31 — Obs.a 2.01 2.35 1.89 — 1.34 1.89 1.66 2.01 2.25 2.35 — Calc. 1.84 2.27 1.77 — 1.31 1.90 1.66 2.02 2.21 2.31 — Obs.a 2.32 2.36 1.41 2.00 1.89 — 1.41 2.00 2.32 2.36 — Calc. 2.22 2.31 1.31 2.00 1.77 — 1.31 2.00 2.22 2.31 — Obs.b 2.15 — 1.47 2.18 1.64 1.90 1.66 2.06 2.32 2.26 0.99 Calc. 1.95 — 1.47 2.00 1.64 1.94 1.59 2.00 2.23 2.33 0.95 Obs.b 2.27 2.36 1.45 2.08 1.45 — 1.45 2.08 2.27 2.36 0.90 Calc. 2.23 2.33 1.32 2.07 1.45 — 1.32 2.07 2.23 2.33 0.91 a Ref. 30. b This work. Table 4 Calculated vs. observed chemical shifts in bicycloheptane systems Camphor 10 Norbornanone 9 Fenchone 3 Proton 1 3x 3n 4 5x 5n 6x 6n 7s 7a 8(Me) 9(Me) 10(Me) 3x(Me) 3n(Me) Obs.a — 2.35 1.84 2.09 1.95 1.34 1.68 1.40 —— 0.84 0.96 0.92 —— Calc. — 2.51 1.78 2.18 2.05 1.37 1.93 1.64 —— 0.98 0.95 1.05 —— Obs.b 2.60 2.06 1.84 2.67 1.79 1.45 1.81 1.53 1.73 1.56 ————— Calc. 2.62 2.28 1.89 2.61 1.85 1.46 1.78 1.58 1.76 1.63 ————— Obs.c 1.15(Me) —— 2.14 1.80 1.70 1.54 1.37 1.80 1.54 1.04 1.04 Calc. 1.00(Me) —— 2.16 1.90 1.52 1.75 1.66 1.96 1.36 1.07 0.99 a Data from ref. 25, assignments from refs. 26 and 27. b Ref. 25. c This work. C17 more or less in a plane with only a 9.58 twist. In the 17-keto compounds (6, 7 and 8) the GAUSSIAN (and PCMODEL) calculations gave the conformation of ring D as a C14 envelope with C13, C15, C16 and C17 almost coplanar and this is in agreement with the observed coupling constants for ring D.21 In the CHARGE model the g eVects of the substituents are considered to be due to electronic eVects and therefore they are modelled on a simple empirical basis.For the ketones studied here we initially made the assumption that the electronic g eVects of the carbonyl carbon (HCCC]] O) are the same as for a saturated carbon atom which is already incorporated into the CHARGE scheme. Subsequently a small correction (0.1 ppm) was added. However the g eVects of the carbonyl oxygen (HCC]] O) need to be determined.As mentioned earlier inspection of the data of Fig. 3 and Tables 1–5 shows that there is clearly an orientation dependance of the carbonyl g SCS. In the similar analysis of saturated carbon (HCCC) and oxygen (HCCO) g eVects a simple angular function (A 1 Bcos q) was found to be appropriate with values of the coeYcients A and B determined by the observed data. Thus this approach was initially used here. However more detailed inspection of the observed data showed that the carbonyl g SCS were also dependant on the bond angle (a) of the carbonyl group (CC(O)C).In particular the five-membered ring ketones with carbonyl bond angles ca. 106–1098 have quite diVerent SCS to the six-membered ketones with bond angles ca. 115–1168. This additional functionality was therefore incorporated into the carbonyl oxygen g eVect again as a simple cos a dependance. The coeYcients in this equation were then determined from the observed SCS by an iterative least mean squares calculation to give finally eqn.(7) for the carbonyl gamma eVect (GSEF).446 J. Chem. Soc., Perkin Trans. 2, 1999, 441–448 Table 5 Observed a vs. calculated chemical shifts in 5a-androstanones 3-one (5) 11-one 17-one (6) 3,17-dione (7) 3,11,17-trione (8) Proton 1a 1b 2a 2b 3a 3b 4a 4b 56 a 6b 7a 7b 89 11a 11b 12a 12b 14 15a 15b 16a 16b 17a 17b 18(Me) 19(Me) Obs. 1.35 2.03 2.29 2.39 —— 2.08 2.27 1.51 1.32 1.32 0.96 1.75 1.33 0.75 1.56 1.38 1.12 1.72 0.92 1.65 1.17 1.60 1.64 1.15 1.43 0.72 1.02 Calc. 1.44 2.00 2.21 2.44 —— 2.03 2.06 1.61 1.48 1.36 1.19 1.93 1.19 0.86 1.56 1.48 1.01 1.65 0.82 1.65 1.53 1.56 1.63 1.05 1.60 0.76 1.00 Obs.b 0.76 2.40 1.50 1.41 1.19 1.65 1.23 1.23 0.99 1.23 1.23 1.12 1.79 1.65 1.69 —— 2.25 2.25 1.54 1.77 1.23 1.72 1.72 1.35 1.45 0.66 1.01 Calc. 0.80 2.43 1.51 1.44 1.19 1.67 1.08 1.33 1.03 1.42 1.28 1.25 1.99 1.57 1.76 —— 2.01 2.28 1.27 1.75 1.61 1.62 1.69 1.14 1.68 0.87 0.93 Obs. 0.91 1.65 1.49 1.42 1.18 1.67 1.29 1.25 1.07 1.25 1.25 0.97 1.77 1.56 0.72 1.67 1.27 1.23 1.80 1.26 1.93 1.51 2.06 2.43 —— 0.86 0.81 Calc. 0.97 1.60 1.53 1.44 1.25 1.69 1.38 1.09 1.10 1.43 1.32 1.26 2.00 1.36 0.81 1.56 1.41 1.18 1.79 1.14 2.18 2.13 2.26 2.27 —— 1.03 0.77 Obs. 1.35 2.03 2.31 2.39 —— 2.11 2.26 1.56 1.38 1.38 1.01 1.84 1.59 0.80 1.70 1.40 1.27 1.83 1.29 1.95 1.52 2.08 2.45 —— 0.89 1.04 Calc. 1.44 2.00 2.21 2.44 —— 2.05 2.07 1.62 1.53 1.40 1.31 2.05 1.43 0.91 1.60 1.47 1.20 1.82 1.17 2.20 2.15 2.27 2.28 —— 1.05 1.01 Obs. 1.22 2.77 2.27 2.45 —— 2.12 2.28 1.51 1.41 1.37 1.20 1.99 1.93 1.74 —— 2.32 2.44 1.91 2.09 1.63 2.26 2.54 —— 0.82 1.22 Calc. 1.26 2.86 2.19 2.44 —— 2.02 2.08 1.56 1.56 1.40 2.16 1.40 1.87 1.91 —— 2.21 2.47 1.66 2.31 2.24 2.35 2.36 —— 1.19 1.18 a This work. b d values from SCS (Table 1) and d(5a-androstane), ref. 15. GSEF = 0.09 (2.0 2 3.0cosa)(2.0 2 cos q) (7) This equation gave generally good agreement for all the vicinal protons in the data set (a total of 50 protons).These results will be discussed later. Long-range eVects The interactions considered to be responsible for the long range eVects of the carbonyl group have been documented earlier as steric, electric field and magnetic anisotropy eVects. We are now in a position to test these theories against the observed data presented in the tables. It is convenient to consider first the electric field eVect as there are no additional parameters required to calculate the electric field eVects of the carbonyl group from eqn.(3). There is the implicit assumption that the charges used in eqn. (3) provide a reasonable measure of the electric field of the carbonyl group. The partial atomic charges calculated in the CHARGE routine have been derived from the observed molecular dipole moments and the extent of the agreement provides one check of the electric field calculation. The calculated and observed (in parenthesis) dipole moments (in debye) of formaldehyde, acetaldehyde, acetone and cyclohexanone are 2.28 (2.34), 2.68 (2.68), 3.03 (2.86) and 3.03 (3.08) and the excellent agreement provides strong support for the use of these charges in the calculations. As the coeYcient in eqn.(3) is known together with the molecular geometries the electric field eVect of the carbonyl group at any proton more than three bonds removed from the carbonyl oxygen atom is given immediately.These values will be discussed later. In contrast to the above, the steric and anisotropic terms are not known and both the steric coeYcient as [eqn. (2)] for the oxygen atom and the magnetic anisotropies Dc1 and Dc2 [eqn. (6)] need to be evaluated. In addition there is a push-pull coeYcient for the steric eVect and also the position of the magnetic anisotropy along the carbonyl bond needs to be determined. It is because of this multifunctional parametrisation that it is essential to have a large and diverse data set.The data set of the non-vicinal protons used here comprises 112 proton shifts and the iterations were achieved using a non-linear least mean squares programme (CHAP8).33 The iterations were initially carried out on the observed SCS in order to eliminate any errors in the calculated shifts of the parent hydrocarbons, but subsequently the observed chemical shifts were used. The results are of some interest. All the iterations including the steric term plus the anisotropy terms gave no better results than the corresponding iterations without the steric term.Thus the steric term for the carbonyl oxygen atom was removed. Also the values of the parallel anisotropy (Dc1) obtained from the iterations were always much larger than those for the perpendicular anisotropy Dc2. These calculations were all performed with the carbonyl anisotropies placed at the midpoint of the C]] O bond. It was found that the best iteration still gave significant errors for some protons in the bicycloheptanones (Table 4).In particular the observed 6exo and 6endo SCS were much smaller than calculated. However placing the anisotropy at the carbonyl carbon atom gave much better agreement for these protons without any significant eVect for the remaining protons in the data set. The final values of the anisotropies obtained were Dc1 17.1 and Dc2 3.2 (10230 cm3 molecule21) and these together with the results obtained can now be considered.The observed vs. calculated proton shifts for the ketones considered are given in Tables 2–5 and it is of some interest to consider these results. The general agreement of the observed vs. calculated shifts is very good and the great majority of the observed shifts are reproduced to better than 0.1 ppm. This is as good as could be expected as the observed vs. calculated proton shifts for the corresponding hydrocarbons are only to ca. 0.1 ppm. The agreement is particularly striking for the chair conformations of decalone (Table 2) and the methylcyclohexanones (Table 3) with no error larger than 0.2 ppm.Also the general agreement for the steroid ketones is encouraging though in this quite sterically compressed system there areJ. Chem. Soc., Perkin Trans. 2, 1999, 441–448 447 larger errors in the calculated shifts for some of the protons in the base hydrocarbon androstane. In particular the 7b and 15b protons are the only resolved protons in androstane with errors >0.2 ppm probably due to large steric interactions and this transfers to the steroid ketones.The good agreement for the C1 protons in the 3,11,17-trione (Table 5) is particularly noteworthy as the 1b proton in the 11-ketosteroids is very close to the 11-keto oxygen and the SCS for this proton provides a critical test of the model. Indeed Schneider et al.12 noted that the 1b proton deviated appreciably (by 0.6 ppm) from their calculated value, based on a dipole model of the electric field and ApSimon’s anisotropy equation.The calculated shifts in the bicycloheptanone systems are also in generally good agreement with the observed shifts (Table 4) though there are some significant errors. It may be significant that in the bicycloheptane system it was necessary to consider possible orbital interactions between the bridging C7 carbon and the ring carbons in order to reproduce the observed shifts in these molecules using the CHARGE model.17 However the largest errors in Table 4 are for the 6exo and 6endo protons in camphor and fenchone in which both the calculated proton shifts and the SCS are much less than the observed values (by ca. 0.2–0.3 ppm). This deviation does not appear to be a function of the bicyclic ring system as in norbornanone both the calculated shifts and the SCS at the C6 protons are in good agreement with the observed values. Why the introduction of methyl groups should aVect the SCS of the carbonyl group is not clear.The proton shifts of camphor were obtained in solvents of varying polarity (CCl4, CDCl3, acetone and methanol) in order to determine if any intramolecular hydrogen bonding between the carbonyl oxygen and the methyl protons was occurring but the shifts were as expected with no evidence of any such interaction. The only acyclic compounds investigated are the simple compounds in Table 2 as all other acyclic ketones exist in two or more conformations. The observed shifts for acetone and acetaldehyde are both slightly greater than calculated and this may be due to solvation eVects.On the reaction field model for any given solvent the solvation shifts are proportional to both the dipole moment and to the reciprocal of the volume of the solute.34 Thus in these small polar molecules solvent eVects will be most pronounced. The values of the carbonyl anisotropy determined here are also of interest.In all the iterations performed the value of the parallel anisotropy Dc1 remained reasonably constant at ca. 20 (10230 cm3 molecule21). In the final iteration the value obtained was 17.1. However the value of the perpendicular anisotropy Dc2 varied considerably with both positive and negative values obtained during the iterations. The last iteration gave a value of 3.2. The variability is a consequence of the small eVect this parameter has on the proton chemical shifts.The only definitive method of determining this parameter would be to obtain SCS from protons situated both at the sides and immediately above the carbonyl group. Although examples of the first type are present in the collected data (e.g. the C8 protons in 1-decalone ) we were unable to obtain suitable compounds in which protons were situated immediately above the carbonyl group. The value of the carbonyl anisotropy obtained here (cf. Fig. 2) is c1 2 c2 17.1 and c3 2 c2 3.2, hence c1 2 c3 equals 13.9.Comparison with the results of previous investigations is not facilitated by the diVerent nomenclatures used. Zurcher 7 defined Dc1 = c1 2 c3 and Dc2 = c2 2 c3. ApSimon9 and also Schneider 12 and Williamson 13 write the anisotropy equation [cf. eqn. (6)] as (1 2 3cos2 q) which merely reverses the sign of Dc. Also the definition of the x, y, and z axes diVers in these investigations. Converting to the nomenclature of Fig. 2 and eqn. (6) gives values of c1 2 c2, c3 2 c2 and c1 2 c3 of 17.1, 3.2, 13.9 (this work), 13.5, 212.2, 25.7 (ref. 7), 21, 26, 27 (ref. 9), 24, 212, 36 (ref. 12) and 4, 29, 13 (ref. 13). There is generally reasonable agreement for the value of the parallel anisotropy (c1 2 c2 or c1 2 c3) apart from Williamson’s value but the value of c3 2 c2 is not well defined. This reinforces the caveat above concerning the uncertainty in the value of Dc2. It is probable that Schneider 12 used the correction to the McConnell eqn.(6) given by ApSimon though this is not explicitly stated in ref. 12 and this may aVect the values of the anisotropies they obtained. It is of some interest to consider the actual magnitudes of the various contributions to the carbonyl SCS and Table 6 gives the observed vs. calculated SCS for trans-1-decalone with the calculated electric field and anisotropy contributions. The table clearly shows that both eVects are important in determining carbonyl SCS. The table also shows that other contributions are present in determining the SCS.For example, the sum of the electric field plus anisotropy contributions for the 8a and 8e protons are 20.04 and 20.01 ppm whereas the calculated SCS are 10.40 and 10.30 ppm. The additional contribution in this case stems from the H ? ? ? H steric interaction. The 8a and 8e protons in trans-decalin experience a large high-field shift due to the proximity of the 1a and 1e protons and these protons are upfield from axial cyclohexane (d 0.93 vs. 1.18 for the axial protons and d 1.54 vs. 1.68 for the equatorial protons) as a result. This steric interaction is removed when these protons are replaced by the carbonyl group giving an additional low-field shift. This eVect is also observed in the SCS of H10 in which there is a 1,3-diaxial H–H interaction with H1ax in transdecalin which is absent in 1-decalone. Apart from these special cases the anisotropic and electric field contributions determine the carbonyl SCS though the relative size of these contributions varies considerably with the orientation of the proton from the carbonyl group.Acknowledgements We thank the EPSRC and GlaxoWellcome Ltd. for a CASE research studentship (N. J. A.) and Drs Richard Upton and John Hollerton for the samples of the keto-steroids and for the 750 MHz spectrum of 4, 7 and 8 and for their continuing support during this work. We also thank Dr I. Sadler and the high-field NMR service at Edinburgh for the 600 MHz spectra of 2, 4, 5 and 7.We are pleased to acknowledge the assistance of Dr P. D. Mallinson and the University of Liverpool central computing facility for the operation of GAUSSIAN94. References 1 Part 12, R. J. Abraham, L. GriYths and M. A. Warne, J. Chem. Soc., Perkin Trans. 2, 1998, 1751. 2 R. J. Abraham, J. Fisher and P. Loftus, Introduction to NMR Spectroscopy, J. Wiley, 1988, p. 23. 3 P. T. Narasimhan and M. T. Rogers, J. Phys. Chem., 1959, 63, 1388.Table 6 Calculated vs. observed SCS for trans-1-decalone (4) with the electric field and anisotropy contributions Proton 2a 2e 3a 3e 4a 4e 5a 5e 6a 6e 7a 7e 8a 8e 9H 10H Obs. 1.07 0.69 0.42 0.38 0.51 0.23 0.22 0.26 20.07 0.02 20.11 0.11 0.32 0.37 1.07 0.49 Calc. 1.08 0.73 0.34 0.30 0.47 0.31 0.11 0.12 0.02 0.03 20.05 0.01 0.40 0.30 1.06 0.35 Electric field —— 0.23 0.18 0.12 0.15 0.04 0.06 0.05 0.03 0.01 0.06 0.18 0.20 — 0.24 Anisotropy —— 20.07 0.07 0.30 0.12 0.07 0.03 20.04 20.01 20.05 20.05 20.22 20.21 — 20.09448 J.Chem. Soc., Perkin Trans. 2, 1999, 441–448 4 L. M. Jackman, Nuclear Magnetic Resonance Spectroscopy, Pergamon Press, 1959, pp. 112–30. 5 J. A. Pople, J. Chem. Phys., 1962, 37, 53,60. 6 A. A. Bothner-By and J. A. Pople, Ann. Rev. Phys. Chem., 1965, 16, 43. 7 R. F. Zurcher, Prog. Nucl. Magn. Reson. Spectrosc., 1967, 2, 205. 8 H. M. McConnell, J. Chem. Phys., 1957, 27, 226. 9 J. W. ApSimon, P. V. DeMarco and D. W. Mathieson, Tetrahedron, 1970, 26, 119. 10 J. Homer and D. Callagham, J. Chem. Soc. A, 1968, 439. 11 K. J. Toyne, Tetrahedron, 1973, 29, 3889. 12 H. J. Schneider, U. Buchheit, N. Becker, G. Shmidt and U. Siehl, J. Am. Chem. Soc., 1985, 107, 7027. 13 (a) M. P. Williamson and T. Asakura, J. Magn. Reson., 1991, 94, 557; (b) ibid., 1993, B101,63; (c) M. P. Williamson, T. Asakura, E. Nakamura and M. Demura, J. Biomol. NMR, 1992, 2, 83. 14 R. J. Abraham, M. E. Edgar, R. P. Glover, M. A. Warne and L. GriYths, J. Chem. Soc., Perkin Trans. 2, 1996, 333. 15 R. J. Abraham, L. GriYths and M. A. Warne, J. Chem. Soc., Perkin Trans. 2, 1997, 31. 16 R. J. Abraham, L. GriYths and M. A. Warne, J. Chem. Soc., Perkin Trans. 2, 1997, 203, 881. 17 R. J. Abraham, L. GriYths and M. A. Warne, Magn. Reson. Chem., 1998, S179,36. 18 Bruker UXNMR version 010892, Bruker AM, Silbersteifen, D-7512 Germany. 19 Varian Associates, Palo Alta, CA, USA. 20 GAUSSIAN 94, Gaussian Inc., Pittsburgh PA, 1994. M. J. Frisch, G. W. Trucks, M. Head-Gordon, P. M. W. Gill, M. W. Wong, J. B. Foresman, B. G. Johnson, H. B. Schlegel, M. A. Robb, E. S. Replogle, R. Gomperts, J. L. Andres, K. Raghavachari, J. S. Binkley, C. Gonzalez, R. L. Martin, D. J. Fox, D. J. Defrees, J. Baker, J. J. P. Stewart and J. A. Pople. 21 N. J. Ainger, PhD thesis, Liverpool University, 1999. 22 F. J. Weigert and J. D. Roberts, J. Am. Chem. Soc., 1970, 92, 1347. 23 J. K. Whitesell and M. A. Minton, Stereochemical Analysis of Alicyclic Compounds by C-13 NMR, Chapman and Hall, London, 1987. 24 S. Castellano and A. A. Bothner-By, J. Chem. Phys., 1964, 41, 3863. 25 R. J. Abraham and N. J. Ainger, manuscript in preparation. 26 R. J. Abraham, A. P. Barlow and A. E. Rowan, Magn. Reson. Chem., 1989, 27, 1074. 27 J. K. M. Sanders and B. K. Hunter, Modern NMR Spectroscopy — A Guide for Chemists, OUP, Oxford, 1987, p. 308. 28 C. R. Kaiser, R. Rittner and E. A. Basso, Magn. Reson. Chem., 1994, 32, 503. 29 C. J. Puchert and J. Behnke, Aldrich Library of 13C and 1H FT NMR Spectra, Aldrich Chemical Company Inc. Milwaukee, USA, 1993. 30 L. GriYths, PhD thesis, Liverpool University, 1979. 31 PCModel version 5, Serena Software Ltd., PO BOX 3076, Bloomington, Indiana, 1994. 32 Landholt-Bornstein, Vol. 7, Structure Data of Free Polyatomic Molecules, ed. K. H. Hellwege and A. M. Hellwege, Springer, NY, 1976. 33 S. S. Kuo, Computer Applications of Numerical Methods, Addison- Wesley, London, 1972, ch. 8. 34 R. J. Abraham and E. Bretschneider, Internal Rotation in Molecules, ed. W. J. Orville-Thomas, Academic Press, NY, 1974, ch. 13. Paper 8/08908F

ISSN:1472-779X    

DOI:10.1039/a808908f

出版商:RSC    

年代:1999

数据来源: RSC

摘要:

J. Chem. Soc., Perkin Trans. 2, 1997 445 Determination of isotope eVects on acid–base equilibria by 13C NMR spectroscopy Tonis Pehk,*,a Ene Kiirend,a Endel Lippmaa,a Ulf Ragnarsson b and Leif Grehn b aInstitute of Chemical Physics and Biophysics, Estonian Academy of Sciences, Akadeemia tee 23, EE0026 Tallinn, Estonia bDepartment of Biochemistry, University of Uppsala, Biomedical Center, PO Box 576, S-751 23 Uppsala, Sweden New results from isotope effect measurements on acid–base equilibria by 13C NMR spectroscopy show that in fatty acids deuterium isotope effects extend at least up to seven bonds and have unexpectedly small attenuation.In deuterium-substituted benzoic acids the isotope effect on pKa is practically independent of the site of deuterium substitution. 2H isotope effects on carboxy ionization in glycine and alanine are smaller than in the corresponding carboxylic acids. Different protonation shifts in unlabelled and labelled compounds are recorded. 18O and 13C isotope effects in acetic acid and 15N effects in glycine isotopomers were measured. The application of auxiliary isotope substitution is demonstrated for the measurement of very small differences in pKa values. Introduction A general method for measuring pKa differences by 13C NMR spectroscopy was elaborated in our previous publication.1 In the present report, some additional aspects of the application of this method to isotope effect studies on acid–base equilibria are discussed.Pioneering work in this field 2 deserves more attention, since only a couple of research groups have used it mainly for 18O isotope effect studies 3–5 and, in one case, for the measurement of the 15N isotope effect in glycine.6 The deuterium isotope effect has been determined by this method only in formic acid.2 The conflicting results concerning the DpKa value of [2,2,2-2H3]acetic acid, measured more than 30 years ago7,8 have not found a reasonable explanation.Therefore a systematic study of deuterium isotope effects in carboxylic acids and in amino acids along with heavy atom isotope effects in these compounds was undertaken. Experimental Isotope effects on pKa values were obtained from measurements of 13C chemical shift differences on the atoms close to the ionisation site of labelled and unlabelled compounds. Such atoms generally behave as sensitive indicators during the titration of the mixtures. Ratios between corresponding dissociation constants were calculated from the plots of these chemical shift differences against the degree of protonation of the reference (unlabelled) compound.These plots give bell-shaped curves described by eqn. (1): d 2 da = dd 2 da d 2 n(dd 2 dp) + Rn(da d 2 da p)/[1 + (R 2 1)n] (1) where d is the chemical shift of the reference compound and dd and dp are the chemical shifts of the reference compound in the deprotonated and protonated states. The corresponding chemical shifts for the measured compound are given by da, da d and da p; n is the degree of protonation of the reference compound, calculated from the formula n = (dd 2 d)/(dd 2 dp), and R is the ratio between the dissociation constants of reference and measured compound, calculated by the data fitting programs Origin (Microcal.Inc.) or Grafit (Erithacus Software Ltd.). NMR studies were performed on a Bruker AMX-500 spectrometer at 293 K. Details of the method and experimental procedure are reported in our previous publication.1 For the study of higher fatty acids and benzoic acid isotopomers dioxane was required to increase the solubility of these acids in water. 18O-labelled compounds were prepared by 18O exchange under acidic conditions using 18O-enriched water. The mixture of all isotopomers of deuteriated acetic acids was obtained by heating a sodium acetate–D2O solution at 150 8C in a sealed glass tube for 24 h. [3,3,3-2H3]Propionic, [4,4,4-2H3]butyric, [6,6,6-2H3]- caproic and [2,3,4,5,6-2H5]benzoic acids originated from Cambridge Isotope Laboratories.The first of them was a gift from Larodan Fine Chemicals AB, Malmö, Sweden. Monodeuteriated benzoic acids and [2,3,5-2H3]benzoic acid were prepared from the corresponding monobromo or 2,3,5-triiodo derivative by reductive dehalogenation with Devarda’s (Cu-Al-Zn) alloy in NaOD–D2O solution.9 In contrast to Raney Ni-Al the use of Cu-Al alloy results in high purity of deuteriated isotopomers.[15N]Glycine was made from Boc- [15N]glycine 10 by deprotection with formic acid 11: 0.25–1 mmol of Boc-derivative was dissolved in formic acid (2–5 ml) in a small test tube in which all subsequent steps took place. After 3–6 h the acid was evaporated in vacuo. Upon addition of diethyl ether, a white, solid product was obtained which was carefully precipitated by centrifugation and the diethyl ether was decanted. Fresh diethyl ether was added twice followed by centrifugation. The yield of amino acid was 96–98% after drying.[2-13C,15N]Glycine was made from Boc-[2-13C,15N]glycine10 and L-[3,3,3-2H3]alanine from Boc-L-[3,3,3-2H3]alanine 12 using the same procedure. Results and discussion Long range (secondary) deuterium isotope eVects Long range deuterium isotope effects were studied in a series of fatty and benzoic acids with different numbers of intervening bonds between the label and the ionisation site. From the experimental point of view comparative measurements of deuteriated and undeuteriated compounds by 13C NMR spectroscopy are facilitated by significant isotope shifts. A comparison of deuterium isotope effects on 13C chemical shifts in some measured acids are presented in Table 1.Triple 2H wsubstitution in fatty acids results in quite large long range effects, e.g. 76 ppb (9.6 Hz at 11.7 T) over three bonds in [4,4,4- 2H3]butyric acid, which gives sufficient separation of signals. In446 J.Chem. Soc., Perkin Trans. 2, 1997 the study of isotopic shifts in monodeuteriated benzoic acids at 22.6 MHz,13 high field shifts were observed only on substituted carbon and its nearest neighbours. Our measurements at 125.7 MHz show some additional long range effects and lower the limit of measurable differential chemical shift values (3 ppb, in 13 15 ppb). Nevertheless, long range isotopic shifts in benzoic acid are definitely smaller than in fatty acids. This is best illustrated by the absence of an isotopic shift on the carboxy carbon even in the [2-2H] isotopomer.Measurements of differential shieldings are somewhat complicated by the broadening of 13C signals to several Hz from the coupling with deuterium nuclei over two and three bonds (in acetic acid 2JCH = 7.3 Hz,14 in aromatic compounds 3JCH is ca. 8 Hz). This broadening is clearly seen on many signals and it lowers the spectral resolution. The results of isotope effect measurements on acid–base equilibria in carboxylic acids are presented in Table 2 and in Figs. 1–5. The number of useful sites for the pKa studies depends on the molecule studied. It can be up to three, as in propionic and butyric acid (Figs. 3 and 4), but in several cases only a single site gives the desired information. Carboxy carbon is ruled out for caproic and benzoic acid due to the very small isotope shifts and spectral broadening. It is interesting to note that in benzoic acid the carboxy carbon shift (4.8 ppm), unlike that in fatty acids, is not the most sensitive one for the measurement of the degree of protonation.The most sensitive site is instead C]1 (ca. 6.8 ppm, Table 2) and third place is taken by the paraposition with a 2.65 ppm high field (!) protonation shift. The remaining ortho-(0.63 ppm) and meta-(0.48 ppm) positions are quite insensitive. The results based on various positions were practically the same (Table 2), although quite different curves Fig. 1 Measured 13C chemical shift differences on titration of a mixture of acetic and [2,2,2-2H3]acetic acid to determine the deuterium isotope effect on Ka: .-methyl carbons; m-carboxy carbons Table 1 2H isotope effects on chemical shiftsa of deuteriated aliphatic carboxylic acids and deuteriobenzoic acids Aliphatic w-[2H3] acids b [2H]Benzoic acids c Position Propionic Butyric Caproic o- m- pa 256 285 297 265 270 292 b 76 85 88 78/113 108/106 110 g — 25 28 <3/15 <3/4 <3 d — — 26 <3 <3 <3 e — — <1 — — — COO 210 28 22 <3 <3 <3 a In ppb, high field shifts positive, in aqueous solution.b Data reduced per single 2H substitution. c Ring carbon with lower/higher number. describe the shift difference variation. Protonation shifts are different not only at different positions from the ionisation site, but also at the same sites in the labelled and unlabelled molecules. This last phenomenon has not been noticed in isotope effect studies.Different deuterium isotope effects in trideuteriated acetic acid have previously been reported by Halevi et al. 7 (DpKa = 0.026) and Streitwieser and Klein 8 (DpKa = 0.014), who used potentiometric and conductometric methods, respectively. These authors were not able to find any reasonable explanation for the inconsistency of their results. The present data (Table 2, Fig. 1) are in good agreement with the latter,8 which are also quoted by Ingold in his monograph.15 Our binary mixture was about 0.1 M in each isotopomer, which is close to the upper concentration limit used previously.8 At concentrations down to 0.004 M the isotope effect was practically the same and the authors concluded that cancellation of activity coefficient effects might occur.8 Our experience with mixtures of carboxylic acids of higher concentration revealed a dependence of R on the total concentration of acids present in the sample.1 Another sample with all deuteriated isotopomers (each ca. 0.25 M) was measured to check for high concentration effects, additivity of isotope substitution effects and differential shielding effects in the protonated and deprotonated states.The titration curves for the methyl carbons are presented in Fig. 2, which give for monodeuteriated acid R = 1.0074 (DpKa = 0.0032), for dideuteriated acid R = 1.0144 (DpKa = 0.0062) and for trideuteriated acid R = 1.0229 (DpKa = 0.0098). Carboxy carbons cannot be used in this case because the four different broadened signals afford insufficient separation of the signals.The obtained data show good additivity in deuterium substitution effects and also in differential shielding between the protonated and deprotonated states. At the same time the R value is definitely smaller for the trideuteriated species at this higher concentration of acetic acid isotopomers as compared with the previous case (dashed curve in Fig. 2, Table 2). The deuterium isotope effect measurements for propionic and butyric acid at three different sites (carboxy, a- and b-carbons) give consistent results (Figs. 3 and 4). In contrast to those for acetic acid the data reported for propionic acid 7 agree with our results, especially considering that a sample with only partial deuterium substitution in the methyl group was previously available.7 In the mixture of [6,6,6-2H3]caproic acid and unlabelled isotopomer, carboxy carbons are separated at low and high pH values only by 0.8 Hz, the carboxy carbon of the deuteriated isotopomer being shifted to low field, as with all Fig. 2 Additivity of deuterium isotope effects on pKa of acetic acid, measured on methyl carbon chemical shifts: d-[2-2H], m-[2,2-2H2], .-[2,2,2-2H3] isotopomers; --- curve for [2,2,2-2H3] from Fig. 1J. Chem. Soc., Perkin Trans. 2, 1997 447 Table 2 Deuterium isotope effects on pKa in carboxylic acids Acid Measured site dd–dp Ra DpKa DDG‡ b [2H]Formic COOH 6.51 1.082c 0.0342 46.5 [2,2,2-2H3]Acetic COOH 4.60 1.0326 ± 0.0008 0.0139 18.9 CH3/CD3 2.903/2.853 1.0298 ± 0.0008 0.0128 17.3 [3,3,3-2H3]Propionic COOH 5.26 1.0191 ± 0.0006 0.0082 11.2 CH2 3.50 1.0188 ± 0.0005 0.0081 11.2 CH3/CD3 1.856/1.820 1.0172 ± 0.0009 0.0074 10.1 [4,4,4-2H3]Butyric COOH 4.84 1.0114 ± 0.0003 0.0049 6.7 a-CH2 3.81 1.0107 ± 0.0003 0.0046 6.3 b-CH2 1.38 1.0098 ± 0.0004 0.0042 5.8 [6,6,6-2H3]Caproic b-CH2 1.50 1.0012 ± 0.0003 0.0005 0.7 [2-2H]Benzoic C-1 6.783/6.779 1.0046 ± 0.0002 0.0020 2.7 [3-2H] C-4 22.65 1.0045 ± 0.0002 0.0019 2.6 [4-2H] C-4 22.627/22.623 1.0042 ± 0.0005 0.0018 2.5 [2,3,5-2H3] C-1 6.789/6.786 1.0137 ± 0.0002 0.0059 8.0 [2,3,4,5,6-2H5] C-1 6.784/6.775 1.0230 ± 0.0002 0.0099 13.4 a With standard errors.b In cal mol21. c From ref. 2. other fatty acids. During the titration exchange broadening is observed and the carboxy site cannot be used for the measurement. The same broadening also influences a-carbon atoms to carboxy group without any measurable isotope shift.Well separated resonances from g- and d-positions cannot be used for Fig. 3 13C NMR titration of a mixture of [3,3,3-2H3]propionic and unlabelled propionic acid: .-C]3; d-C]2; m-C]1 (carboxy) carbons Fig. 4 13C NMR titration of a mixture of [4,4,4-2H3]butyric and unlabelled butyric acid: .-C]3; d-C]2; m-C]1 (carboxy) carbons the DpKa measurement due to too small protonation shifts at these sites. The only suitable position in this case is the bmethylene carbon, which is shifted in the deuteriated isotopomer by 2.2 Hz to low (!) field, its protonation shift being only ca. 30% of that from the carboxy carbon. Nevertheless, this position can be exploited to determine the degree of protonation with the result showing that w-trideuteriated caproic acid is weaker than caproic acid by 0.0005 units. On the basis of the deuterium isotope effect in acetic acid, the corresponding values for higher fatty acids show unexpectedly small attenuation.In ref. 8 the authors have assumed that the magnitude of the deuterium isotope effects in fatty acids are consistent with the bond dipole changes and their attenuation obeys the regular fall-off of the inductive effect with distance by a factor of ca. 2.8 per carbon atom down the chain. These conclusions were based on the comparison of the DpKa values of pivalic and acetic acid and their deuterio-isotopomers. On this basis, DpKa for the propionic acid isotopomers should be 0.005, which is definitely lower than our measured value (0.008).The agreement is worse for the w-trideuteriated butyric acid (calculated 0.0018, experimental 0.005). Finally, for the wtrideuteriated caproic acid, the calculated value (DpKa = 0.0002) is again less than the measured one (DpKa = 0.0005). The reason for this unexpected behaviour is most probably connected with the nonbonded interactions, which have been the object of earlier discussions of deuterium isotope effects.8,16 More recently, isotopic perturbations of conformational equilibria have been observed in NMR spectra of various alicyclic compounds17 and they are likely to be present also in aliphatic chains.This also explains the good separation of all corresponding signals in the butyric acid isotopomers and the unusual low field isotope shift of g-carbon in the deuteriated caproic acid. The resulting variations of non-bonded interactions might also modify the pKa values. Deuteriated benzoic acids were studied to investigate the conclusions drawn only on the basis of [2,3,4,5,6-2H5] and [2,6-2H2] isotopomers of benzoic acid.8 The reported DpKa value for the pentadeuteriated isotopomer (0.010 ± 0.002) fits with the present result (Table 2).On the basis of additivity of isotope effects the reported value for the [2,6-2H2] isotopomer (0.003 ± 0.001) points to the possibility that in meta- or para-positions isotope effects can be even larger than in the ortho-position.Obviously, monodeuteriated isotopomers can clarify this situation. Measurement of the [2-2H] isotopomer (Fig. 5) shows that the value obtained for the [2,6-2H2] isotopomer is a bit too small, although within the reported error limits.8 Our results for the [3-2H] and [4-2H] (Fig. 5) give practically the same values as for the [2-2H] isotopomer. Experimentally, the [4-2H] isotopomer is the most difficult case because only C-4 can be exploited in this context, but C-4 gives a broadened triplet due to deuterium.The measurement of the [3-2H] isotopomer, where C-4 is also448 J. Chem. Soc., Perkin Trans. 2, 1997 Table 3 Deuterium isotope effects on pKa values of [2,2-2H2]glycine and [3,3,3-2H3]alanine in aqueous solutions Protonation Measured Deprotonation Acid site site shift (ppm) Ra DpKa DDG‡/cal mol21 Gly COO CH2/CD2 1.356/1.316 1.0056 ± 0.0002 0.0024 3.3 Ala COO CH 1.75 1.0142 ± 0.0003 0.006 8.3 Gly NH2 COO 7.66 1.054 ± 0.001 0.023 31.1 Gly NH2 CH2/CD2 2.683/2.558 1.051 ± 0.002 0.022 29.5 Ala NH2 CH3/CD3 4.177/4.090 1.0377 ± 0.0004 0.016 21.8 a With standard errors.used, is more straightforward. Additional experiments with the [2,3,5-2H3] and [2,3,4,5,6-2H5] isotopomers were also performed. These are more convenient cases, because C-1 can be used and the results confirm the additivity of isotope effects within an aromatic nucleus. The results of our study on deuteriated isotopomers of benzoic acid are in full agreement with the earlier conclusions about the even distribution of deuterium isotope effects within the ring.8 On the other hand, the predicted effect of one meta-deuterion on pKa on the basis of the inductive effect model8 0.0012 is not in excellent agreement with the experiment.Fig. 5 13C NMR titration of binary mixtures of [2-2H] (d, C-1), [3- 2H] (., C-4), [4-2H] (m, C-4) and undeuteriated benzoic acid. Concave curves for [3-2H] and [4-2H] isotopomers results from C-4 high field protonation shift Fig. 6 13C NMR titration of a mixture of [2,2-2H2]glycine and glycine for the determination of Ka ratios for carboxy (right axis, d-CH2/CD2; --- linear dependence) and amino (left axis, .-CH2/CD2, m-COO) groups Secondary deuterium isotope effects in amino acids may be of practical importance and as first examples in this field [2,2- 2H2]glycine and [3,3,3-2H3]alanine were measured in mixtures with the nondeuteriated counterparts. In these cases, the isotope effects on both pKa values can be measured and the results are presented in Table 3 and Figs. 6 and 7. In the glycine experiment carboxyl protonation was monitored at the methylene carbon, whereas for the amino group the carboxy carbon was used. This is, in fact, the only possibility for determining the pKa difference at the carboxy site because the carboxy carbons themselves are insufficiently resolved upon protonation, whereas the CH2 and CD2 signals are well separated due to the deuterium isotope effect on the shielding constant.For amino group protonation it is well known that the largest effect on chemical shifts are observed in the b-position and therefore the carboxy carbons are much more sensitive in this context. This does not rule out the possibility of using the methylene carbon for an additional check of the results. In the case of [3,3,3-2H3]alanine, the large protonation shifts of the carboxy carbon cannot be exploited because the deuterium isotope effect on the chemical shift is much smaller than in glycine and this results in overlapping signals.As noticed for acetic and propionic acid above, in alanine an unexpectedly large R value is observed for the carboxyl group protonation, more than twice the size of that for glycine. The isotope effect per deuterion on the pKa value of the carboxyl group in glycine is less than one third of that in acetic acid.The deuterium isotope effect on the amino group pKa in glycine is one order of magnitude higher as compared to that of the carboxy group. The deuterium isotope effect on the amino group pKa in alanine is also remarkably high. Nevertheless, no straightforward explanation can be given for the significant effect on the carboxy group pKa value in this amino acid. Heavy atom isotope eVects on pKa Among heavy nuclei studied in this context, 18O has been used most frequently.2–5 The relatively big change in atomic masses Fig. 7 13C NMR titration of a mixture of [3,3,3-2H3]alanine and alanine for the determination of Ka ratios for carboxy (right axis, d-CH) and amino (left axis, .-CH3/CD3) groupsJ.Chem. Soc., Perkin Trans. 2, 1997 449 within the ‘heavy atom scale’ (18O/16O = 1.125), and being itself a protonation site, resulted in R values for formic acid of about 1.015 (DpKa = 0.006) for each isotopic substitution. We have now measured the acetic acid 18O isotope effect and found that DpKa per single isotope substitution is nearly twice as small as in formic acid, as shown in Fig. 8 (R = 1.0086 ± 0.0002, DpKa = 0.0037), where the curve corresponding to single 18O substitution in formic acid (DpKa = 0.006 212) falls between the two curves for acetic acid. The 18O isotope effect in glycine was also determined and the value obtained for a single substitution (R = 1.0084 ± 0.0001) is the same as in acetic acid. Recently, the 15N/14N isotope effect (15N/14N = 1.07) on the pKa value of the amino group of glycine was determined by Rabenstein and Mariappan6 and the R value 1.0224 ± 0.0003 was reported. From previous synthetic work on 13C,15N labelled glycines 18 we had access to all glycine isotopomers, including the 15N labelled one.We have therefore been able to repeat the measurement and confirm the result. We obtained R = 1.0227 ± 0.0002, giving further confidence in the method used. The 15N isotope effect on the pKa value of the carboxy group Fig. 8 Effects of one (d) and two (.) 18O atoms on the dissociation of acetic acid. The dashed curve corresponds to single 18O substitution in formic acid.2 Fig. 9 Effect of 13C substitution in acetic acid carboxy group on its dissociation constant: R = 1.0021; for the comparison calculated curves with R = 1.003 (upper curve) and R = 1.001 (lower curve) are also given is too small even for the present method. The introduction of a 13C label into the carboxy group brings the heavy nucleus two additional bonds closer to the ionisation site.This also improves to some extent the prospect of observing the corresponding isotope effect, but raises the question of how to measure this effect using 13C NMR spectroscopy. Proper species in the sample must be selected for the isotope effect measurements and not all molecules can be studied in this context, e.g. not formic acid with its single carbon atom. Therefore, carboxygroup- labelled acetic acid was chosen. Consequently, in this sample one exploits only the methyl 13C signals of different multiplicity in the [2-13C]CH3COOH and [1,2-13C2]CH3COOH isotopomers.In the proton decoupled spectrum one obtains a singlet from the first species and a doublet from the C-C coupled doubly labelled counterpart, which means that the DpKa determination is carried out at natural abundance. The effect of protonation of 13C labelled acetate is presented in Fig. 9. Only very small differences within 0.4 Hz were observed, which are again modified by different protonation shifts. The obtained R value 1.0021 ± 0.0001 gives a DpKa value 0.0009 and free energy difference 1.2 cal mol21. For comparison two additional calculated curves with R values of 1.001 and 1.003 are also presented in Fig. 9 to show that they differ significantly from the observed dependence. In contrast to the previous examples, multiple labels give additional possibilities for DpKa studies.In Fig. 10 the results on protonation of the amino group in [2-13C,15N]- glycine are presented. In this experiment R = 1.0239 ± 0.0003, DpKa = 0.0103, DDG‡ = 14 cal mol21. The bell-shaped curve originating from this experiment is clearly shifted from that of [15N]glycine and thus gives the possibility of estimating the 13C isotope effect from the methylene carbon on the pKa value of the amino group: R = 1.0012 ± 0.0005, DpKa = 0.0005 and DDG‡ = 0.7 cal mol(!).The consequence of the auxiliary 15N labelling has, to our knowledge, not been exploited previously. All of the above examples of isotope effects and also literature data show that replacement of a lighter nucleus by a heavier one always results in smaller dissociation constants (higher pKa values). A summary of these data is presented in Table 4. A reasonable explanation of isotope effects involves changes in vibrational states of molecules.19 Finding quantitative relationships between these parameters is a problem which might be studied, together with the presented data, by the use of this very sensitive method based on 13C NMR spectroscopy.Fig. 10 Comparison of titration curves of [15N] (m) and [2-13C,15N]- glycine (d) in the mixtures with unlabelled glycine450 J. Chem. Soc., Perkin Trans. 2, 1997 Acknowledgements This work was supported by the Royal Swedish Academy of Sciences and the Estonian Science Foundation. References 1 T.Pehk, E. Kiirend, E. Lippmaa and U. Ragnarsson, J. Chem. Soc., Perkin Trans. 2, 1996, 2351. 2 S. L. R. Ellison and M. J. T. Robinson, J. Chem. Soc., Chem. Commun., 1983, 745. Table 4 Summary of long range isotope effects to pKa values in organic compounds (per single isotope atom substitution)a Effect Compound Nb R DpKa DDG‡c 2H/1H Formic acid 2 1.082 0.034 46.5 2H/1H Acetic acid 3 1.011 0.0047 6.5 2H/1H Propionic acid 4 1.0062 0.0027 3.6 2H/1H Butyric acid 5 1.0039 0.0017 2.3 2H/1H Caproic acid 7 1.0002 0.0002 0.2 2H/1H Benzoic acid 4,5,6 1.0045 0.0020 2.6 2H/1H Glycine -COO 3 1.0028 0.0012 1.7 2H/1H Glycine -NH2 2 1.026 0.011 15.2 2H/1H Alanine -COO 4 1.0047 0.002 2.8 2H/1H Alanine -NH2 3 1.0125 0.005 7.3 18O/16O Formic acid 0 1.015 0.006 8.2 18O/16O Acetic acid 0 1.0085 0.0037 5.0 18O/16O Glycine -COO 0 1.0084 0.0036 4.9 15N/14N Glycine -NH2 0 1.0227 0.010 13.3 13C/12C Acetic acid 1 1.0021 0.0009 1.2 13C/12C Glycine -NH2 1 1.0012 0.0005 0.7 a Results of present work except for formic acid.2 b Number of bonds to the protonation site.c In cal mol21. 3 C. L. Perrin and J. D. Thoburn, J. Am. Chem. Soc., 1989, 111, 8010. 4 W. B. Knight, P. M. Weiss and W. W. Cleland, J. Am. Chem. Soc., 1986, 108, 2759. 5 J. P. Jones, P. M. Weiss and W. W. Cleland, Biochemistry, 1991, 30, 3634. 6 D. L. Rabenstein and S. V. S. Mariappan, J. Org. Chem., 1993, 58, 4487. 7 E. A. Halevi, M. Nussim and A. Ron, J. Chem. Soc., 1963, 866. 8 A. Streitwieser Jr. and H. S. Klein, J. Am. Chem. Soc., 1963, 85, 2759. 9 M. Tashiro, K. Nakayama and G. Fukata, J. Chem. Soc., Perkin Trans. 1, 1983, 2315. 10 L. Grehn, T. Pehk and U. Ragnarsson, Acta Chem. Scand., 1993, 47, 1107. 11 B. Halpern and D. E. Nitecki, Tetrahedron Lett., 1967, 3031. 12 L. Lankiewicz, B. Nyasse, B. Fransson, L. Grehn and U. Ragnarsson, J. Chem. Soc., Perkin Trans. 1, 1994, 2503. 13 T. Yonemitsu, H. Tuzuki, S. Mataga and M. Tashiro, Kyushu Sangyo Daigaku Kogakubu Kenkyu Hokoku, 1993, 30, 145; Chem. Abstr., 1995, 122, 9376x. 14 E. Breitmaier, G. Jung, W. Voelter and L. Pohl, Tetrahedron, 1973, 29, 2485. 15 C. K. Ingold, Structure and Mechanism in Organic Chemistry, Cornell University Press, 1969, Table 57-8. 16 L. S. Bartell, J. Am. Chem. Soc., 1961, 83, 3567. 17 F. A. L. Anet, V. J. Basus, A. P. W. Hewett and M. Saunders, J. Am. Chem. Soc., 1980, 102, 3945; T. Pehk, A. Laht and E. Lippmaa, Org. Magn. Reson., 1982, 19, 21. 18 B. Nyasse, L. Grehn and U. Ragnarsson, J. Chem. Soc., Chem. Commun., 1994, 2005. 19 R. P. Bell and W. B. T. Miller, Trans. Faraday Soc., 1963, 59, 1147. Paper 6/06790E Received 3rd October 1996 Accepted 6th November 1996

ISSN:1472-779X    

DOI:10.1039/a606790e

出版商:RSC    

年代:1997

数据来源: RSC

摘要:

J. Chem. Soc., Perkin Trans. 2, 1999, 449–455 449 A fluorescent glucose sensor binding covalently to all five hydroxy groups of ·-D-glucofuranose. A reinvestigation Mia Bielecki, Hanne Eggert and Jens Chr. Norrild * Department of Chemistry, University of Copenhagen, Symbion, Fruebjergvej 3, DK-2100 Copenhagen, Denmark Received (in Cambridge) 13th November 1998, Accepted 21st January 1999 The structures of the complexes between a fluorescent bisboronic acid 7 and glucose have been determined.Shinkai et al.1 previously studied the complex between 7 and glucose and they deduced a 1,2 : 4,6-a-D-glucopyranose bisboronate structure. We have shown that this structure is only valid as an initial complex formed under completely nonaqueous conditions. In the presence of water the pyranose complex rearranges rapidly into an a-D-glucofuranose- 1,2 : 3,5,6-bisboronate in which all five free hydroxy groups of glucose are covalently bound by the sensor molecule. A favourable B–N interaction around the 1,2-binding site and the eVect of an o-ammoniomethyl group on the pKa value of the second boronic acid group allow for the observed binding at neutral pH.The structure evaluations are based on 1H and 13C NMR data as well as information obtained from 1JCC coupling constants. The fluorescence spectra of both complexes were measured and discussed. MALDI TOF-MS experiments showed competitive formation of 1 : 2 (boronic acid : glucose) complexes under conditions of physiological glucose levels.Introduction There is today a strong demand for the development of new eYcient, selective and cheap carbohydrate sensors and in this sense especially new glucose sensors. Recently boronic acid based carbohydrate sensors have been suggested as a new and very promising alternative to glucose oxidase based sensor systems.2–4 For more than 50 years studies on the interactions between boronic acids and carbohydrates have been performed and complexes between boronic acids and especially glucose and fructose are well recognised although their structures only just recently have been determined.5,6 To be able to distinguish between various carbohydrates, a concept of having a bisboronic acid capable of binding simultaneously to two specific sites in the carbohydrate was investigated. 3,7 This concept has now, through numerous interesting investigations by S. Shinkai’s group,2,8 been further developed into sensor molecules of which some are presently reported capable of giving a larger response to glucose relative to fructose1,3,9 and others of distinction between enantiomeric forms of certain carbohydrates.10 The detection methods demonstrated include e.g.circular dichroism,3,9 UV–vis absorbance, 11,12 electrochemistry,13 and fluorescence.4,8,14 For most sensors suggested the working range however has been limited to pH values between 9 and 12. We have recently studied the structures of the boronic acid complexes under both aqueous and non-aqueous conditions as this information to us seems essential for the future design of new custom-built sensor molecules.Our results show that for the binding of glucose there is a strong preference for binding of the a-glucofuranose form.6 When we read a recent paper by Shinkai and co-workers 1 concerning a new fluorescent anthracene based bisboronic acid we became very curious. In the design of this new bisboronic acid (7, Scheme 1) they ingeniously used a concept originally developed by WulV,15,16 where an intramolecular boron–nitrogen interaction lowers the pseudo pKa-value (pKa1, Fig. 1) of the boronic acid. As it is well known that significant complex formation between boronic acids and carbohydrates only occurs at pH � pKa of the boronic acid (i.e. with tetrahedral boronates),17 the low pKa1 value thus obtained expands the pH-window for strong binding of carbohydrates to include neutral pH values.From a 1H NMR study Shinkai and co-workers deduce the complex between 7 and glucose to be a 1,2: 4,6-a-glucopyranose complex which in our eyes was quite surprising with regard to our previous studies 6 and a very recent investigation by our lab of a water soluble bisboronic acid capable of binding at neutral pH in the furanose form.18 The 1H NMR evidence in the article was very convincing; however, experiments in our laboratory revealed that the 1H NMR spectrum, as presented without solvent specification, was obtained from pure methanol-d6 and not in the water–methanol mixture which was used in the fluorescence experiments.This was not clear from the article and it made us wonder whether the suggested pyranose complex would be representative also for the conditions under which the fluorescence experiments were performed. On this basis we synthesised 6 and 7 and reinvestigated the complexes with glucose as described in the following section.The results of this study are a reassignment of the 1H NMR spectrum previously reported 1 and a determination of the structure of the complex present under the conditions of the fluorescence measurements diVerent from the one reported in methanol. Furthermore the diVerences in Fig. 1 Boron–nitrogen interactions at varying pH in o-(N,N-dialkylaminomethyl) arylboronic acids. Approximate pK values refer to aryl = phenyl (ref. 16). Only structures with tetrahedral boron (structures B, C and E) form strong complexes with diols in aqueous solution.450 J.Chem. Soc., Perkin Trans. 2, 1999, 449–455 Scheme 1 fluorescence intensities of the two complexes have been measured and discussed. Results and discussion 9,10-Bis-[[N-[o-(5,5-dimethyl-1,3,2-dioxaborinan-2-yl)benzyl]- N-methylamino]methyl]anthracene (6) was synthesised and purified as described in the Experimental section. This procedure is slightly diVerent from the one used by Shinkai and co-workers and includes a final crystallisation of the compound. Compound 6 was deprotected under alkaline aqueous conditions followed by extraction of the liberated bisboronic acid into chloroform with the aid of a little DMSO followed by evaporation and drying.This gave a sample of the free bisboronic acid 7 suYciently pure for NMR studies. To reproduce the 1H NMR spectrum previously published, we freshly dissolved a 1 : 1 mixture of 6 and anhydrous a-D-glucose in methanol-d6 and recorded the 1H NMR spectrum instantaneously. The spectrum is shown in Fig. 2(a) and is in full agreement with the one previously reported.1 However, rerecording the spectrum after a few hours revealed several new peaks in the spectrum. After 20 h we observed a considerable change as seen by the appearance of a totally new set of signals (Fig. 2(b)). These signals are most obvious in the regions from Fig. 2 The spectrum of 6 and anhydrous a-D-glucopyranose (1 : 1) in methanol-d6 recorded (a) instantly, (b) after 20 h and (c) after 8 days.J.Chem. Soc., Perkin Trans. 2, 1999, 449–455 451 Table 1 13C Chemical shifts (ppm) for the aliphatic part of the boronic acid complexes and reference compoundsa Compound Complex-1 b Complex-2 b 8 d 9 e 10 e a-D-Glucopyranose e a-D-Glucopyranose d C-1 98.3 104.6 104.0 107.6 105.7 94.9 92.4 C-2 79.0 82.2 85.8 85.7 86.1 74.3 72.5 C-3 77.1 76.3 73.6 79.5 79.2 75.4 73.2 C-4 74.1 80.1 74.7 81.1 83.3 72.3 70.7 C-5 69.1 73.2 70.8 73.9 72.2 74.0 72.1 C-6 62.3 67.0 62.0 67.0 67.3 63.3 61.4 N–CH3 42.9 38.5 N–CH39 42.8 40.5 Ca 63.4 c 60.5 c Ca9 64.2 c 63.8 c Cb 47.6 c 51.5 c Cb9 47.8 c 53.7 c a The data are given relative to TMS.The assignments are in agreement with the information obtained from 1H–13C-heterocorrelated and 13C–13Ccorrelated spectra and with 1D selective decoupling experiments. b In CD3OD. c As the sugar’s orientation related to the ligand has not been determined the chemical shifts of the ligand CH2 groups marked respectively a/a9 and b/b9 can be interchanged. d In DMSO-d6.e In D2O at pD = 11–12. 5.3 to 5.7 ppm and from 8.5 to 8.9 ppm. After 8 days the signals from the original complex had disappeared and only the additional signals were left (Fig. 2(c)). In the following we denote the initially formed complex as Complex-1 and the complex formed later as Complex. Repeating the above experiment using a-D-glucopyranose monohydrate it was impossible to reproduce the spectrum of Fig. 2(a). After only 20 min the signals from Complex-2 were dominant and after 3 days Complex-1 could not be observed. The final mixture contained signals from Complex-2 (>90%) in mixture with other unidentified products. Our results show that the initially formed a-D-glucopyranose complex under the conditions is converted slowly to a thermodynamically more stable a-D-glucofuranose complex. This transformation has to comprise a mutarotation of the initial bound a-D-glucopyranose and here the solvent has a central role.The observed enhanced transformation rate in methanol of Complex-1 to Complex-2 by using the monohydrate of glucose compared to anhydrous glucose agrees with the well known eVect of water compared to methanol on the mutarotation of a-D-glucopyranose. The ~30 times faster mutarotation of glucose in water as compared to methanol19 thus suggests a very fast transformation of Complex-1 to Complex-2 in the presence of water.Therefore under the conditions for fluorescence measurements, as applied by S. Shinkai and co-workers (water–methanol (2 : 1), buVer pH 7.8), we would expect only the presence of Complex-2 and not Complex-1. In order to prove this assumption we recorded the 1H and 13C NMR spectra of a 1: 1 mixture of 6 and a-D-glucose in water– Fig. 3 Assigned structures of Complex-1 and Complex-2. methanol (1 : 2). Due to the low solubility of the ligand, the latter solvent mixture was the closest we could get to the above solvent conditions while still being able to obtain NMR spectra.The 13C NMR spectrum was obtained with uniformly 13C6 labelled a-D-glucose. Due to the still very low solubility of 6 under these solvent conditions, acceptable spectra were obtained only after very long data accumulation. However, in accordance with our expectations we were able to conclude the presence of Complex-2 only 10 min after mixing and longer accumulation did not unveil the presence of other complexes.Structure evaluation: Complex-1 We assign the structure of Complex-1 as shown in Fig. 3. This structure is in agreement with the one earlier deduced by Shinkai et al. based on 1H NMR data. However, our reinvestigation of the complex aVords several corrections to the assignments and provides new significant and consistent evidence for the structure. The a-D-glucopyranose structure of the glucose part of Complex-1 is substantiated by the relatively low 13C-chemical shift values for the glucose part compared to those of similar a-furanose complexes 6 (Scheme 2 and Table 1).The measured 3JH–H coupling constants (Table 4) show, except for J12, high values in agreement with an approximate trans axial arrangement of these protons as expected for an a-glucopyranose ring. Our assignment of the 1H NMR data of the complex diVers considerably from the one previously published. The H-5,6a/b protons have been reassigned and the chemical shifts of 5.78, 6.23, 6.81 and 6.88 ppm assigned by Shinkai et al.to the methylene protons (Hb and Hb9, Fig. 3) proved by the Scheme 2452 J. Chem. Soc., Perkin Trans. 2, 1999, 449–455 Table 2 1H Chemical shifts (ppm) in CD3OD for the aliphatic part of Complex-1 and Complex-2 a Complex-1 H-1 5.25 Halb 3.98 H-2 3.08 Ha2b 4.93 c H-3 20.21 Ha91 b 4.17 H-4 2.75 Ha92 b 4.66 H-5 2.75 Hbl b 4.85 c H-6a 3.48 Hb2b 4.92 c H-6b 3.73 Hb91 b 5.01 N–CH3 2.49 Hb92 b 5.11 N–CH39 2.75 2JHH/Hz 11.9 11.4 13.9 14.2 Complex-1-(3-O-Me) H-1 5.33 Halb 4.04 H-2 3.67 Ha2b 4.94 H-3 0.11 Ha91 b 4.21 H-4 3.20 Ha92 b 4.72 H-5 2.67 Hbl b —d H-6a 3.52 Hb2b —d H-6b 3.78 Hb91 b 5.04 N–CH3 Hb92 b 5.09 N–CH39 2JHH/Hz 11.7 11.9 —d 14.2 Complex-2 H-1 5.49 Halb 3.87 H-2 2.95 Ha2b 4.24 H-3 1.32 Ha91 b 4.24 H-4 3.40 Ha92 b 5.36 H-5 3.35 e Hbl b 5.54 H-6a 3.76 Hb2b 5.77 H-6b 3.65 Hb91 b 5.65 N–CH3 2.30 Hb92 b 5.79 N–CH39 2.95 2JHH/Hz 13.8 12.1 14.4 14.5 a The data are given relative to TMS.The assignments are in agreement with the information obtained from 1H–1H-COSY, 1H–13C-heterocorrelated and 13Cb13C-correlated spectra and with 1D selective decoupling experiments. The chemical shifts for the free 2,2-dimethylpropane-1,3-diol in CD3OD are 3.36 and 0.89 ppm. b As the sugar’s orientation related to the ligand has not been determined the chemical shifts of the ligand CH2 groups marked respectively a/a9 and b/b9 can be interchanged.Within each methylene group 1 and 2 can be interchanged. c The signals are partly hidden under the water signal and exact values were obtained by solvent inverse recovery decoupling. d Not determined. e The chemical shift of H-5 was disclosed by decoupling of the methanol signal giving an eVect at H-4 and H-6b. method of 1H–13C-heterocorrelated spectra and 1D selective decoupling experiments to be aromatic signals while the methylene protons had chemical shifts close to that of water and were consequently hidden under the broad water signal.An inverse recovery decoupling of the water signal unveiled the chemical shifts and coupling constants of these protons (See Table 2). A reassignment of the approximately triplet signals 6.81 and 6.88 ppm to neighbouring aromatic protons of one of the boron substituted phenyl rings explains their multiplicity in a consistent manner. To exclude hidden signals under those of the protecting group, an experiment with the free boronic acid 7 was performed and new data for Complex-1 did not appear.To determine the binding sites of the pyranose ring we prepared a 1: 1 solution of 6 and uniformly 13C6 labelled a-D-glucose and obtained the one bond C–C coupling constants within the sugar part of the complex (Table 3). The one bond C–C coupling constant, within a RO–C–C–OR9 fragment containing sp3 hybridised carbon atoms, depends a) on the O–C–C–O dihedral angle and b) on the R–O–C–C torsions, the latter showing the greater variations.According to calculations by Serianni et al.20 a minimum value of 1JC–C should be expected for an approximately all eclipsed geometry within such a fragment. In agreement with this we showed earlier that exceptionally low 1JC–C values are found when the two carbons are contained in five membered 1,3-dioxolane or 1,3,2- dioxaborinane rings relative to values for the free sugar.5,6 The measured value of 1JC1–C2 = 36 Hz compared to 44 Hz of the free a-D-glucopyranose (Table 3) thus strongly indicates a 1,2- Table 3 1JCC Coupling constants (Hz) Compound Complex-1 a Complex-2 a 8 b 9 c 11 b 13 b a-D-Glucopyranose a b-D-Glucopyranose a J1,2 36 35 34.4 35.7 34.0 33.9 44.4 45.6 J2,3 37 44 44.0 43.6 42.3 46.6 38.1 38.9 J3,4 38 33 34.4 34.3 38.2 31.8 35.9 39.6 J4,5 38 40 40.5 40.0 48.1 40.0 39.6 40.5 J4,5 43 35 40.9 34.5 34.2 33.4 42.7 43.4 a In CD3OD.b In DMSO-d6. c In D2O at pD = 11–12. boronate. The 1JC5–C6 of 43 Hz on the other hand excludes C-5 and C-6 from being members of such a five membered ring and thus eliminating a 5,6-bound furanose.This leaves three imaginable secondary binding sites, namely i) a trans vicinal pyranose 3,4-boronate, ii) a seven membered 3,6-boronate, and iii) a six membered pyranose 4,6-boronate. trans Vicinal boronic acid complexation of pyranose hydroxy groups was reported earlier but only under aprotic conditions and always comprising 1,3,5,2,4-trioxadiborepane-2,4-diyl structures which are not possible here.21,22 Seven membered boronate rings have been isolated from aprotic media but always under constrained conditions where five or six membered rings were not possible.23–25 We therefore, in accordance with the proposal of Shinkai and co-workers, conclude the secondary binding site to be 4,6 giving an a-D-glucopyranose 1,2 : 4,6 bisboronate structure for Complex-1.This conclusion is substantiated by the close resemblance of the measured 3JHH coupling constants to those of 1,2 : 4,6-di-O-benzylidene-a-D-glucopyranose (12) (Table 4) and by the fact that no complex formation was observed in an experiment with 6-deoxy-L-glucose.26 On the other hand 3-O-methyl-a-D-glucopyranose did indeed give the corresponding O-3 methylated Complex-1 (see Fig. 4 and Tables 2 and 4). Structure evaluation: Complex-2 Complex-2, which is the thermodynamically more stable under the conditions applied, is assigned the structure shown in Fig. 3. The 1H NMR data together with data from model compounds are compiled in Tables 2 and 4. The chemical shift of H-4 was best determined from an experiment with the free boronic acid 7 as the signal is partly hidden under the protection group signal at 3.36 ppm. The 3JH4–H5 coupling constant was estimated to be 2–3 Hz by selective decoupling of H-3 which transformed the H-4 triplet to a distorted doublet. In the proton spectrum H-2 and H-3 are both doublets indicating 3J23 ~ 0 as also found in similar boronic acid complexes of a-D-glucofuranose 6 (Table 4).In agreement with a furanose ring no large vicinal coupling constants are found. Furthermore, the glucose part of the complex shows relative high 13Cchemical shift values (Table 1) with e.g. the anomeric carbon signal less shielded by 6 ppm as compared to Complex-1.J. Chem. Soc., Perkin Trans. 2, 1999, 449–455 453 Table 4 JH–H Coupling constants (Hz) for the glucose part of boronic complexes and selected model compounds Compound Complex-1 a Complex-1-(3-O-Me) a Complex-2 a 8 d 9 e 10 e 11 e 12 f 13 d a-D-Glucopyranose g J1,2 5.5 6.1 4.2 4.1 3.6 4.0 3.6 4.9 3.6 3.8 J2,3 7.0 6.8 ~0 ~0 ~0 ~0 ~0 6.4 ~0 9.9 J3,4 9.6 10.2 3.1 2.4 2.8 2.4 2.8 9.4 3.0 9.6 J4,5 —b 9.2 2–3 c ~0 2.6 9.5 6.8 9.2 1.3 9.6 J5,6a 9.6 9.8 ~0 2.4 ~0 6.0 6.4 10.0 ~0 2.2 J5,6b 4.4 4.2 4.8 2.4 5.1 3.5 5.5 4.9 5.2 5.5 J6a,6b 9.6 9.8 8.0 m. 8.8 9.0 8.8 10.3 7.9 12.3 a In CD3OD. b Could not be determined.c Could not be precisely determined. d In DMSO-d6. e In D2O at pD = 11–12. f In CDCl3 according to Liptak et al.34 g In D2O according to Curatolo et al.35 From experiments with 1 : 1 mixtures of 6 and 13C labelled glucose it was possible to obtain the 1JC–C coupling constants as listed in Table 3. For comparison we have included 1JC–C values for the p-tolylboronates 8 and 9 obtained from 13C NMR spectra of samples prepared with uniformly 13C6 labelled glucose together with data for the model compounds 11 and 13.The 1JC–C coupling constants of 11 and 13 were measured in unlabeled samples using the INADEQUATE technique. For the reasons discussed above the values of 1JC1–C2 and 1JC5–C6 being relatively low strongly indicate a 1,2 : 5,6 or a 1,2 : 3,5,6 bound a-furanose complex. Comparing the measured 1JC–C coupling constants of Complex-2 with those of a-D-glucofuranose 1,2 : 3,5,6-bis(p-tolylboronate) 6 (9) one observes two almost identical data sets (see Table 3) whereas large diVerences are seen for the 1,2: 5,6-di-O-isopropylidene-a-D-glucofuranose (11).This does, on the basis of our earlier studies,5,6 provide very strong evidence of the 1,2 : 3,5,6 bound structure. Further evidence for 3,5,6 binding is the measured 3JHH coupling constants which correlate well with those of 9 and of the 3,5,6- orthoester 13 (Table 4). Therefore we conclude Complex-2 to be the 1,2 : 3,5,6 bound a-D-glucofuranose as shown in Fig. 3. Our assignment of Complex-2 is further substantiated by experiments substituting a-D-glucopyranose with respectively 6-deoxy-L-glucose, 3-O-methyl-a-D-glucopyranose 26 and 3- deoxy-b-D-ribo-hexopyranose (3-deoxy-D-glucose). These compounds do not form Complex-2 analogs in accordance with their lack of the appropriate hydroxy groups as summarized in Fig. 4. Instead, after a few days, non-resolvable mixtures of multiple components are found in all three experiments.As we observe no diVerences between the NMR data of Complex-2 in methanol and aqueous buVer solution, we deduce the structure of Complex-2 to be the N-protonated form as Fig. 4 Complex formation between bisboronic acid 6 and various glucose derivatives in methanol. (a) Formation of Complex-2 via Complex-1. In dry MeOH rearrangement takes 7–8 days. Traces of water dramatically enhance the rate of transformation. (b) Multiple complexes form. No NMR evidence of Complex-2 type structures within the mixtures.(c) Initial formation of Complex-1 (with 3-O-methyl group). Rearranges with time to a complicated mixture. (d) Due to use of 3-deoxy-D-glucose as the crystalline b-D-ribo-hexopyranose initial formation of the 1,2 bound pyranose is prevented. During the slow mutarotation in dry MeOH multiple complexes form and neither Complex-1 nor Complex-2 type structures can be identified. shown in Fig. 3 under both conditions. In MeOH this zwitterionic form is a prerequisite whereas in aqueous buVer other factors may favour this structure.Drawing from earlier work by Yurkevich et al.27 it can be anticipated that ammoniomethyl substituents generally have a notable lowering eVect on the pKa values of boronic acids, consequently allowing for structure E (Fig. 1) to be involved in binding at neutral pH. This eVect, together with a superior stabilisation of tridentate boronates, enables the one boronic acid group to bind in the observed 3,5,6 fashion at neutral pH.We stress that our conclusion is limited to a 1 : 1 mixture at pH 7.8. B–N interactions The B–N interactions in complexation between 6 and glucose under aqueous conditions were deduced by Shinkai and coworkers as a result of fluorescence based pH titrations. In accordance with the work of WulV15,16 they observed a low pKa1 value of the free boronic acid as a result of the formation of the tetrahedral N-bonded boronic acid (see B, Fig. 1). As the observed decrease of the fluorescence (i.e.N-lone pair PET quenching) at pH around pKa1 was absent when glucose was added they concluded an even stronger B–N interaction upon complexation. In the latter case only a decrease of the fluorescence corresponding to pKa2 was observed (see Fig. 1). As 11B NMR experiments, for reasons including low solubility and extensive line broadening, did not provide unambiguous information on the boron atom geometries, we have authenticated the B–N interactions with regard to our assigned structures of Complex-1 and -2 by using the available information from our 1H and 13C NMR experiments.The 1H NMR spectrum in CD3OD of the free ligand 7 shows broad lines (Dn1/2 = 13 and 15 Hz ) at room temperature for the methylene groups CH2(a) and CH2(b), respectively, in accordance with an intermediate fast exchange between the B–N bonded form (B, Fig. 1) and the non-bonded structure (D, Fig. 1). However, for all signals of Complex-1 and -2 only sharp lines are observed. Regarding the 1H NMR data of Complex-1 one observes large chemical shift diVerences (0.95 and 0.59 ppm) for the geminal protons Ha1 and Ha2 and for Ha91 and Ha92 in agreement with these protons being fixed in a five membered B–N containing ring.For comparison small chemical shift diVerences of 0.1 ppm are observed for geminal protons of the CH2(b) groups. A diVerence is also observed between the geminal coupling constants of the two types of methylene groups.The two CH2(a) and the two CH2(b) groups each have very similar 2JH–H values, however the “b-type” is (numerically) larger by ~2 Hz. The structure deduced for Complex-2 (Fig. 3) allows for only one B–N interaction, making the two alkylamine substituents on the anthracene ring clearly diVerent. The NMR data of the alkylamine groups reflect this diVerence as follows. In Complex-2 the chemical shift diVerences between non-equivalent geminal protons are increased relative to Complex-1 except for one of454 J.Chem. Soc., Perkin Trans. 2, 1999, 449–455 the CH2(a) groups, where it has decreased by 0.37 ppm as a consequence of the group not being part of a five membered ring. Furthermore this CH2 group shows a geminal coupling constant 1.7 Hz larger than that of the other CH2(a) group and 1.9 and 2.4 Hz larger than those of the two CH2(a) groups in Complex-1. For Complex-2 the 13C chemical shift diVerences between N–CH3, CH2(a) and CH2(b) groups at each side of the anthracene ring are 2.0, 3.3 and 2.2 ppm respectively.In Complex-1 the corresponding values are much smaller (0.1, 0.8 and 0.2 ppm). Fluorescence To confirm the results obtained from NMR spectroscopy, a series of fluorescence measurements were made under the same conditions with regard to solvent and reactants. The fluorescence of a 1025 M 1 : 1-solution of 6 and anhydrous glucose in methanol was measured as a function of time (Fig. 5). A clear drop in intensity from t = 0 to t = 25 h was observed and after 120 h the intensity was only 41% of the initial one (Fig. 6); however no significant shifts in the emission maxima were observed. The observed time dependence of the fluorescence is in full agreement with the time span of the transformation of Complex-1 to Complex-2 under similar conditions as seen from Fig. 2. Therefore the observed decrease certainly reflects the diVerences in the fluorescence of the two complexes.It is interesting to note that in methanol solution the transformation of Complex-1 to Complex-2 implements the breaking of one B–N bond leaving a protonated nitrogen atom (see Fig. 3). In an experiment using a 33% methanol–water buVered solution of pH = 7.80 (which are the conditions applied by Shinkai and co-workers for their fluorescence measurements) no time dependent intensity changes could be observed within 8 h. In agreement with the evidence obtained from the NMR experiments under closely related conditions (see above) we suggest Fig. 5 Relative fluorescence of a 1 : 1 mixture of 6 and anhydrous a-D-glucopyranose (1025 M) in pure methanol as function of time. (lex = 370 nm; lem = 425 nm). Fig. 6 Emission spectra of a 1: 1 mixture of 6 and anhydrous a-D-glucopyranose (1025 M) in pure methanol to (a) t = 0 h, (b) t = 72 h and (c) t = 5 days. (lex = 370 nm). that under the fluorescence conditions the thermodynamically more stable furanose complex (Complex-2) is instantly formed even though extrapolating to the very dilute conditions of the fluorescence experiment could be troublesome.In order to obtain further knowledge of the species present under the conditions of the fluorescence experiments we have performed MALDI TOF-MS experiments of bisboronic acid 6 (1025 M) and glucose in 33% methanol–water. For a 1 : 1 boronic acid : glucose ratio we observe the peak from the 1 : 1 complex but also significant amounts of the 1 : 2 (boronic acid : glucose) complex.The latter was not mentioned previously by Shinkai and co-workers.1 Going to ratios of 1 : 400 and 1 : 4000 the 1 : 2 peak increases to around 1/3 of the 1 : 1 peak. This suggests that the expected strong fluorescence of the 1 : 2 complex might interfere in the interpretation of the fluorescence titrations at the 1 : 4000 ratio in the original paper.1 From this follows that the titration curve for a 6 : glucose ratio of 1 : 4000 will be a sum of fluorescence of at least two species but also that formation of a 1 : 2 complex cannot be neglected for this system under conditions of physiological glucose levels (i.e. 3–20 mM). Conclusions We have shown that the earlier proposed structure of Complex- 1 is only valid as an initially formed complex between a-Dglucopyranose and 6 or 7 under completely nonaqueous conditions. We found that Complex-1 rearranges to the thermodynamically more stable 1,2 : 3,5,6 bound a-D-glucofuranose complex Complex-2 as a function of time and water content of the medium.Under conditions related to those of the fluorescence measurements in the original work by Shinkai and co-workers 1 (water–methanol, 1 :2) we have shown that Complex-2 is instantly formed. The structure of Complex-2 is to our knowledge the first example of a bisdentate 1,2 : 3,5,6 bound a-D-glucofuranose complex and it strongly indicates the superior stability of this type of complex over 1,2 : 3,5 or 1,2 : 5,6 bound complexes in agreement with our findings on the corresponding bis-(p-tolylboronates).6 In this manner Complex- 2 represents the first example of a sensor molecule which binds covalently to all five hydroxy groups of glucose.The binding to all five hydroxy groups is obtained by a delicate balance of the equilibria considered in Fig. 1, where the superior stabilisation of a trisdentate boronate clearly suppresses one of the otherwise favoured B–N interactions.We have shown that the rearrangement of Complex-1 to Complex-2 in methanol causes a decrease in the observed fluorescence and that a 1 : 2 ligand to sugar complex is present in substantial amounts under conditions of physiological glucose levels. We believe that further eVorts can be made to optimise sensor structures and properties towards this superior type of a-Dglucofuranose binding and hence low pKa values and increased water solubility should certainly be implemented.Experimental NMR spectra were recorded at 25 8C, on a Varian Unity 400 NMR spectrometer 13C (100 MHz) and 1H (400 MHz). Chemical shifts are reported in ppm and referenced to CD3OD, 49.03 ppm (13C NMR) and CHD2OD, 3.35 ppm (1H NMR). All coupling constants are given as numerical values. Column chromatography was performed on silica gel 60 PF254 1366. Evaporations were performed in vacuo on a rotary evaporator. Melting points are uncorrected. Fluorescence spectroscopy was performed on a Perkin-Elmer LS50B instrument.Mass spectrometry was performed on a Jeol JMS-HX/HX 110A instrument. Microanalyses were performed by Micro Analytical Laboratory, The Örsted Institute, University of Copenhagen, Denmark.J. Chem. Soc., Perkin Trans. 2, 1999, 449–455 455 Materials All chemicals used were of reagent grade and all solvents were of HPLC grade. 3-Deoxy-D-glucose 28,29 was crystallised as the 3-deoxy-b-D-ribo-hexopyranose according to Anet.30 9,10-Bis(chloromethyl)anthracene (1).Prepared according to Miller et al.31 CAUTION: This compound is highly allergenic and should be handled with care. 1H NMR (acetone-d6) d 8.40 (4H, anthracene-H1,4,5,8), 7.60 (4H, anthracene-H2,3,6,7), 5.70 (4H, s, CH2). Mp 254–256 8C (decomposes). MS: M1: 274 m/z. Anal. Calcd. for C16H12Cl2: C, 69.84; H, 4.40. Found: C, 69.66; H, 4.26%. 9,10-Bis[(methylamino)methyl]anthracene (2). Prepared according to T. D. James et al.1 and further purified by chromatography (silica gel; toluene–MeOH–NEt3, 94:5:1).Mp 149– 150 8C. 1H NMR (benzene-d6) d 8.32 (4H, anthracene- H1,4,5,8), 7.35 (4H, anthracene-H2,3,6,7), 4.44 (4H, s, anthr– CH2–N), 2.39 (6H, s, N–CH3), 0.90 (2H, bs, –NH–). MS: M1: 264 m/z. Mp 149–150 8C. Anal. Calcd. for C18H20N2: C, 81.78; H, 7.63; N, 10.60. Found: C, 80.74; H, 7.36; N, 9.18%. 2-Methylphenylboronic acid (o-tolylboronic acid) (3). Prepared according to König and Scharrnbeck.32 o-(Bromomethyl)phenylboronic acid (4).Prepared according to Takeuchi et al.33 2,2-Dimethylpropane-1,3-diyl [o-(bromomethyl)phenyl]- boronate (5).1 (o-Bromomethyl)phenylboronic acid (4, 4.79 g, 22 mmol) was dissolved in toluene (50 mL) upon heating. 2,2- Dimethylpropane-1,3-diol (4.30 g, 41 mmol, 1.9 equiv) was added and the orange solution refluxed for 2 h and left at room temperature overnight with stirring. The orange organic phase was washed with water (3 × 25 mL) and dried over MgSO4. Evaporation aVorded 5 as an orange oil.The yield was 5.91 g (94%). The crude 5 was >98% pure by 1H NMR and was used in the next step without further purification. 1H NMR (CDCl3) d 7.80 (1H, ArH), 7.34–7.37 (2H, ArH), 7.24–7.30 (1H, ArH), 4.93 (2H, s, Ar–CH2–Br), 3.81 (4H, s, O–CH2–C–CH2–O), 1.05 (6H, s, (CH3)2C). 9,10-Bis[[N-[o-(5,5-dimethyl-1,3,2-dioxaborinan-2-yl)benzyl]- N-methylamino]methyl]anthracene (6). Prepared according to T. D. James et al.1 but the orange powder thus obtained was recrystallised from ethyl acetate to aVord 6 as light yellow crystals. 1H NMR (CD3OD) d 8.33 (4H, ArH), 7.27–7.80 (12H, ArH), 5.00 (4H, bs, b: anthr–CH2–N), 4.29 (4H, bs, a: phen– CH2–N), 3.35 (8H, bs, O–CH2–C–CH2–O), 2.40 (6H, bs, e: N– CH3), 0.85 (12H, bs, (CH3)2C). 13C NMR (CD3OD, only proton bearing carbon atoms) d 135.0 (i: phenyl), 132.6 (f, g or h: phenyl), 131.3 (g, h or f: phenyl), 128.8 (h, f or g: phenyl), 127.9 (d: anthracene-C2,3,6,7), 126.2 (c: anthracene-C1,4,5,8), 69.7 (protecting group –CH2–), 64.3 (a: phen–CH2–N), 50.8 (b: anthr–CH2–N), 40.8 (e: N–CH3), 21.7 ((CH3)2C).MS: 668 m/z. Anal. Calcd. for C42H50N2B2O4: C, 75.46; H, 7.54; N, 4.19. Found: C, 75.14; H, 7.24; N, 4.03%. 9,10-Bis({N-methyl-N-[o-(dihydroxyboryl)benzyl]amino}- methyl)anthracene (7). 9,10-Bis[[N-[o-(5,5-dimethyl-1,3,2- dioxaborinan-2-yl)benzyl]-N-methylamino]methyl]anthracene was suspended in 2 M NaOH and vigorously stirred for 20 h. HCl (1 M) was added to pH 6.5 and the resulting clear yellow solution was extracted with CHCl3 (2×) and CHCl3–DMSO (5 : 1, 1×).The organic phase was washed with water and dried over MgSO4. Evaporation of solvents and drying in vacuo aVorded 7 as a yellow powder. 1H NMR (CD3OD) d 8.31 (4H, c: anthracene-H1,4,5,8), 7.71 (2H, i: phenyl-H), 7.60 (4H, d: anthracene-H2,3,6,7), 7.31–7.41 (6H, f, g, h: phenyl-H), 4.99 (4H, s, b: anthr–CH2–N), 4.30 (4H, s, a: phen–CH2–N), 2.41 (6H, s, e: N–CH3). 13C NMR (CD3OD, only proton bearing carbon atoms) d 137.1 (i: phenyl), 134.4 (f, g or h: phenyl), 133.4 (g, h or f: phenyl), 130.7 (h, f or g: phenyl), 129.7 (d: anthracene-C2,3,6,7), 128.0 (c: anthracene-C1,4,5,8), 66.4 (a: phen–CH2–N), 52.6 (b: anthr–CH2–N), 40.2 (e: N–CH3).Anal. Calcd. for C32H34N2B2O4: C, 72.11; H, 6.43; N, 5.26. Found: C, 71.79; H, 6.18; N, 4.93%. Acknowledgements Dr Mikkel Jørgensen and Risoe National Laboratory, Denmark are acknowledged for providing equipment and help for MALDI TOF-MS experiments. The Carlsberg Foundation, Denmark is gratefully acknowledged for financing Dr J.C. Norrild through post doctoral grant #960298/20. References 1 T. D. James, K. R. A. S. Sandanayake, R. Iguchi and S. Shinkai, J. Am. Chem. Soc., 1995, 117, 8982. 2 T. D. James, K. R. A. S. Sandanayake and S. Shinkai, Angew. Chem., Int. Ed. Engl., 1996, 35, 1910, and refs. therein. 3 K. Tsukagoshi and S. Shinkai, J. Org. Chem., 1991, 56, 4089. 4 J. Yoon and A.W. Czarnik, J. Am. Chem. Soc., 1992, 114, 5874. 5 J. C. Norrild and H. Eggert, J. Chem. Soc., Perkin Trans. 2, 1996, 2583. 6 J. C. Norrild and H. Eggert, J. Am. Chem. Soc., 1995, 117, 1479. 7 A. P. Russell. Patent. In WO 91/04488, PCT/US90/05401, 1991. 8 T. D. James, P. Linnane and S. Shinkai, J. Chem. Soc., Chem. Commun., 1996, 281, and refs. therein. 9 M. Takeuchi, T. Imada and S. Shinkai, J. Am. Chem. Soc., 1996, 118, 10658. 10 T. D. James, K. R. A. S. Sandanayake and S.Shinkai, Nature, 1995, 374, 345. 11 M. Takeuchi, M. Taguchi, H. Shinmori and S. Shinkai, Bull. Chem. Soc. Jpn., 1996, 69, 2613. 12 H. Shinmori, M. Takeuchi and S. Shinkai, Tetrahedron, 1995, 51, 1893. 13 A. Ori and S. Shinkai, J. Chem. Soc., Chem. Commun., 1995, 1771. 14 M. Yamamoto, M. Takeuchi and S. Shinkai, Tetrahedron, 1998, 54, 3125. 15 T. Burgemeister, R. Grobe-Einsler, R. Grotstollen, A. Mannschreck and G. Wullf, Chem. Ber., 1981, 114, 3403. 16 G. WulV, Pure Appl. Chem., 1982, 54, 2093. 17 J. P. Lorand and J. O. Edwards, J. Org. Chem., 1959, 24, 769. 18 We found the formation of a strong 1,2 : 3,5-a-D-glucofuranose complex between glucose and a water soluble bis(pyridineboronic) acid at physiological pH. Submitted for publication elsewhere. 19 F. P. Worley and J. C. Andrews, J. Phys. Chem., 1927, 742. 20 I. Carmichael, D. M. Chipman, C. A. Podlasek and A. S. Serianni, J. Am. Chem. Soc., 1993, 115, 10863. 21 R. J. Ferrier, A. J. Hannaford, W. G. Overend and B. C. Smith, Carbohydr. Res., 1965, 1, 38. 22 R. J. Ferrier, J. Chem. Soc., 1961, 2325. 23 J. M. Sugihara and C. M. Bowman, J. Am. Chem. Soc., 1958, 80, 2443. 24 R. A. Bowie and O. C. Musgrave, J. Chem. Soc., 1963, 3945. 25 E. J. Bourne, I. R. McKinley and H. Weigel, Carbohydr. Res., 1974, 35, 141. 26 6-Deoxy-L-glucose and 3-O-methyl-a-D-glucopyranose were kindly provided by Prof. Christian Pedersen, DTU, Denmark. 27 A. M. Yurkevich, I. I. Kolodkina, E. A. Ivanova and E. I. Pichuzhkina, Carbohydr. Res., 1975, 43, 215. 28 E. J. Hedgley, W. G. Overend and R. A. C. Rennie, J. Chem. Soc., 1963, 4701. 29 S. Iacono and J. R. Rasmussen, Org. Synth., 1990, Coll. Vol. 7, 139. 30 E. F. L. J. Anet, Chem. Ind. (London), 1960, 345. 31 M. W. Miller, R. W. Amidon and P. O. Tawney, J. Am. Chem. Soc., 1955, 77, 2845. 32 W. König and W. Scharrnbeck, J. Prakt. Chem., 1930, 128, 153. 33 M. Takeuchi, T. Mizuno, H. Shinmori, M. Nakashima and S. Shinkai, Tetrahedron, 1996, 52, 1195. 34 A. Liptak, J. Imre, J. Harangi and P. Nanasi, Carbohydr. Res., 1983, 116, 217. 35 W. Curatolo, L. J. Neuringer, D. Ruben and R. Haberkorn, Carbohydr. Res., 1983, 112, 297. Paper 8/08896I

ISSN:1472-779X    

DOI:10.1039/a808896i

出版商:RSC    

年代:1999

数据来源: RSC