Lysine biosynthesis and metabolism in fungi T. Mark Zabriskie* and Michael D. Jackson College of Pharmacy Oregon State University Corvallis OR 97331 USA Received (in Cambridge) 15th July 1999 Covering 1985–April 1999 Role of a-aminoadipate pathway intermediates in 6 1 Introduction lysine biosynthesis 1.1 The a-aminoadipate pathway to l-lysine 2 Enzymes of the a-aminoadipate pathway 2.1 Homocitrate synthase 2.2 Homoaconitate hydratase 2.3 Homoisocitrate dehydrogenase 2.4 a-Aminoadipate aminotransferase 2.5 a-Aminoadipate reductase 2.6 Saccharopine reductase 2.7 Saccharopine dehydrogenase 345 Role of pipecolic acid in lysine biosynthesis Lysine catabolism secondary metabolism The a-aminoadipate pathway in archaea and bacteria Conclusion Acknowledgements References 789 1 Introduction lysine biosynthesis Among the 20 common proteinogenic amino acids L-lysine is unusual in that two diverse pathways have evolved for its biosynthesis.In bacteria lower fungi (some Phycomycetes) and green plants l-lysine is synthesized via the diaminopimelate Mark Zabriskie was born in Salt Lake City in 1960. He attended the University of Utah where he received his Bachelors Degree in Chemistry in 1985. Continuing at Utah he studied marine natural products chemistry with Prof Chris Ireland and obtained a PhD in Medicinal Chemistry. In 1989 he began a postdoctoral fellowship at the University of Alberta working with Prof John Vederas on the mechanism and inhibition of peptide amide hormone formation.He joined the medicinal chemistry faculty of the College of Pharmacy at Oregon State University in 1992 where he now holds the rank of Associate Professor and is also a faculty member of the Center for Gene Research and Biotechnology and the program in Molecular and Cellular Biology. His research interests center on primary and secondary metabolism of amino acids with particular focus on transformations of lysine and arginine. His group is also involved in studying the molecular genetics and enzymology of nonribosomal peptide biosynthesis. Dr Zabriskie lives in Corvallis Oregon with his wife and their two sons with whom he enjoys mountain sports travel and baseball. This journal is © The Royal Society of Chemistry 2000 pathway in seven steps starting with aspartate semialdehyde and pyruvate.In addition to the lysine required for protein biosynthesis this pathway is the source of the diaminopimelate (DAP) and lysine incorporated into bacterial cell wall peptidoglycan. The pivotal role that enzymes of the DAP pathway play in cell viability fuelled extensive investigation into targeting these enzymes for the development of new antibacterial agents and the area has been recently reviewed.1,2 Likewise the molecular genetics of the bacterial diaminopimelate pathway has received a great deal of attention. With the recent finding that the argD-encoded N-acetylornithine aminotransferase also exhibits N-succinyl-l,l-diaminopimelate+a-ketoglutarate aminotransferase (DapC) activity all the genes in the DAP pathway have been cloned.2,3 In Euglenoids and higher fungi (Ascomycetes and Basidomycetes) de novo l-lysine biosynthesis proceeds through the intermediacy of l-a-aminoadipate in a series of transformations entirely unrelated to the bacterial diaminopimelate route.Work with auxotrophic mutants during the 1950’s and 60’s established the identity of the intermediates in the pathway and characterization of the biochemical steps proceeded simultaneously. Most of the biochemical and genetic understanding of the pathway results from studies with the yeast Saccharomyces cerevisiae. Significant findings also stemmed from work with Yarrowia lipolytica Neurospora crassa and some Candida species. Research in the aminoadipate pathway area has witnessed a recent resurgence in activity due to in a large part heightened efforts in cloning and characterizing fungal genes.The availability of these genes has facilitated a number of significant advances in recent years. Several important insights Michael D. Jackson was born in Ogden Utah in 1968. He studied both chemistry and biology at Western Oregon State College and became interested in doing graduate work at the interface between chemistry and biology. During his undergraduate studies he received a WOSC/Teledyne Wah Chang internship in analytical chemistry. Michael entered the graduate program in the Chemistry Department at Oregon State University in 1992 where he worked initially with Dr Steven J.Gould and later joined Dr Mark Zabriskie’s group in the College of Pharmacy. His graduate work entails studying arginine secondary metabolism in the antibiotic-producing soil bacteria Streptomyces lavendulae and S. capreolus. Upon completing his PhD in late 1999 he will join the laboratory of Dr John M. Denu at Oregon Health Sciences University as a postdoctoral fellow working on the mechanism of protein phosphatases. When not busy with his research Michael likes spending time in the great outdoors golfing and building remote-controlled airplanes. 85 Nat. Prod. Rep. 2000 17 85–97 into catalytic and regulatory mechanisms in the pathway are directly attributed to leads identified through sequence homology with other genes in the databases.The main purpose of this review is to provide an overview of the aminoadipate pathway for lysine biosynthesis in fungi and present the properties of the individual enzymes and associated genes. The reader is referred to an excellent review for coverage of progress in the field prior to 1985.4 Bhattacharjee has more recently published an overview of the aminoadipate pathway with special consideration of the evolutionary features.5 Where relevant direct reference to the older literature will be cited. Different ways of designating genes appear in the lysine molecular genetics literature. Because this overview is not targeted towards genetics specialists a common representation will be used here in order to prevent confusion. Mutants are identified by lower case letters in italics (e.g.lys1) genes are denoted by upper case italic letters (e.g. LYS1) and gene products are symbolized by the same three letters as the gene with only the first letter capitalized (e.g. Lys1). Another area of potential confusion arises in instances where genes encoding the same enzyme in different organisms will have different names. Table 1 correlates each enzyme of the pathway with the various genes and organisms. Table 1 Correlation of enzymes of the a-aminoadipate pathway with the respective genes and producing organisms Enzyme Homocitrate synthase Homoaconitate hydratase Homoisocitrate dehydrogenase a-Aminoadipate aminotransferase a-Aminoadipate reductase Saccharopine reductase Saccharopine dehydrogenase Scheme 1 Enzymes of the fungal a-aminoadipate pathway to lysine i homocitrate synthase EC 4.1.3.21; ii & iii homoaconitase EC 4.2.1.36; iv homoisocitrate dehydrogenase EC 1.1.1.87; v aminoadipate aminotransferase EC 2.6.1.39; vi aminoadipate reductase EC 1.2.1.31; vii saccharopine reductase EC 1.5.1.10; viii saccharopine dehydrogenase EC 1.5.1.7.Nat. Prod. Rep. 2000 17 85–97 86 Yeast/Fungus Gene S. cerevisiae Y. lipolytica P. chrysogenum S. cerevisiae A. nidulans LYS20 LYS1 LYS1 LYS4 LYSF S. cerevisiae S. pombe C. albicans P. chrysogenum S. cerevisiae S. cerevisiae Y. lipolytica C. albicans LYS2 LYS1 LYS2 LYS2 LYS9 LYS1 LYS5 LYS1 1.1 The a-aminoadipate pathway to l-lysine Lysine biosynthesis in fungi requires eight steps involving seven free intermediates.Several steps involve transient or enzyme-bound intermediates and some transformations require the action of two gene products. The overall process is detailed in Scheme 1. Steps comprising the first half of the pathway leading to a-aminoadipate take place in the mitochondria; the exception being the synthesis of homocitrate in the nucleus. The latter steps converting a-aminoadipate to l-lysine are carried out in the cytoplasm. The first half of the aminoadipate pathway shares many similarities with the tricarboxylic acid (TCA) cycle; the early intermediates in the lysine pathway are simply one carbon higher homologs. As such the enzymology of much of the aminoadipate pathway is quite similar to that found in the TCA cycle.Fungal lysine biosynthesis begins with the condensation of aketoglutarate (1) and acetyl-CoA catalyzed by homocitrate synthase. The resulting homocitric acid (2) undergoes a dehydration yielding cis-homoaconitic acid (3) which in turn is converted to homoisocitric acid (4) by homoaconitase. Oxidation of 4 by homoisocitrate dehydrogenase results in the transient formation of oxaloglutarate followed by loss of carbon dioxide to yield a-ketoadipic acid (5). Glutamate-dependent transamination of a-ketoadipate by aminoadipate aminotransferase gives rise to l-a-aminoadipic acid (6). The second half of the pathway begins with the reduction of the side chain carboxy of 6 to form l-a-aminoadipic acid-d-semialdehyde (7) by aminoadipate reductase in a process requiring ATP and NADPH.Saccharopine reductase then catalyzes the condensation of 7 with l-glutamate and subsequent reduction of the imine to give l-saccharopine (8). The last step in the pathway is the cleavage of the carbon–nitrogen bond within the glutamate moiety of 8 carried out by saccharopine dehydrogenase. The final products are l-lysine (9) and a-ketoglutarate (1). 2 Enzymes of the a-aminoadipate pathway 2.1 Homocitrate synthase The condensation of one molecule of acetyl CoA with aketoglutarate to form homocitrate catalyzed by homocitrate synthase (acetyl-CoA+2-ketoglutarate C-acetyl transferase; EC 4.1.3.21) represents the first committed step of fungal lysine biosynthesis (Scheme 2).Enyzmatic formation of homocitrate in Saccharomyces cerevisiae was first reported in 1964.6 It was later demonstrated that two homocitrate synthase isozymes could be separated by isoelectric focussing.7 Both isozymes were inhibited by lysine but only one form was transcriptionally repressed when the yeast was grown in medium supplemented with lysine.8 Scheme 2 An early subcellular localization study reported that 80% of the total homocitrate synthase activity in a S. cerevisiae cell lysate resided in the particulate fraction and was presumed to be associated with the mitochondria. It was not determined if the observed activity was due to both isozymes.9 This report also suggested that an observed inhibition of homocitrate synthase by low concentrations of coenzyme-A could be a regulatory mechanism.A more recent investigation into the subcellular forms of homocitrate synthase in the penicillin G producer Penicillium chrysogenum Q176 found the mitochondrial enzyme to be highly unstable whereas the cytosolic form could be purified 500-fold to near homogeneity.10 The cytosolic isozyme exhibited a native molecular mass of 155 kDa by gel filtration chromatography and a subunit size of 54 kDa by SDSPAGE. Contrary to findings in S. cerevisiae 75% of the total enzyme activity was located in the cytosol and no inhibition by coenyzme-A was observed. During the sequencing of chromosome IV from S. cerevisiae an open reading frame (ORF D1298) was identified that shared significant sequence similarity with the NIFV gene of Azotobacter vinelandii and a-isopropylmalate isomerase an enzyme involved in leucine biosynthesis.11 NifV catalyzes the formation of homocitrate not as a precursor to lysine but rather as an element of the iron–molybdenum cofactor in nitrogenase.This cofactor is composed of a Fe7S9Mo metal sulfur core and one molecule of homocitrate. Because S. cerevisiae has no known ability to fix nitrogen ORF D1298 is presumed to encode a homocitrate synthase functioning in lysine biosynthesis. Disruption of this gene named LYS20 in S. cerevisiae resulted in transformants exhibiting a three-fold decrease in homocitrate synthase activity.12 The failure to completely eliminate homocitrate formation by disrupting a single gene is consistent with earlier findings of two homocitrate synthase isozymes in this organism.Because the N-terminus of the predicted amino acid sequence contained no mitochondrial targeting sequence it was proposed that Lys20 is cytosolic. Expression of the homocitrate synthase isoform encoded by LYS20 was shown to be repressed by lysine. Using monoclonal antibodies against nuclear proteins from S. cerevisiae Chen et al. made the surprising discovery that two isozymes of homocitrate synthase are localized in the nucleus.13 Immunoscreening a yeast expression library allowed identification of two ORFs from chromosome IV corresponding to YDL182w (LYS20) and YDL131w in the S. cerevisiae database. LYS20 and YDL131w were shown to direct the expression of 47 and 49 kDa proteins respectively.The sequence of the YDL131w product is predicted to be 90% identical to Lys20. Localization of the proteins to the nucleus was based upon cell and nuclear fractionation and immunofluorescence studies. This finding was quite unexpected given the number of earlier reports that identified homocitrate synthase activity in the mitochondria and cytoplasm. While most of the homocitrate synthase was localized to the nucleus the remainder could account for the activity detected elsewhere in the cell. Furthermore it was suggested that crudely prepared mitochondria fractions may have contained nuclear fragments. The LYS1 gene encoding homocitrate synthase has been cloned and sequenced from the yeast Yarrowia lipolytica.14 Northern blot analyses indicate a single homocitrate synthase transcript is produced and Southern hybridization revealed a single copy of the gene is found in this species.The Y. lipolytica LYS1 is predicted to encode a 446 residue protein (48 kDa) possessing 84% identity at the amino acid level with the predicted sequence of S. cerevisiae Lys20. Deletion of Y. lipolytica chromosomal LYS1 produced lysine auxotrophs that regained lysine independent growth when further transformed with a self-replicating plasmid carrying LYS1. The LYS1 gene from P. chrysogenum was recently cloned and shown to be 71.1% and 71.7% similar in amino acid sequence with homocitrate synthases from Y. lipolytica and S. cerevisiae respectively.15 The gene codes for a 474 amino acid protein with an expected mass of 52 kDa.Expression of LYS1 was mildly repressed by high lysine concentrations in the growth medium and it was proposed that repression is a weak regulatory mechanism in comparison to lysine feedback inhibition of homocitrate synthase. 2.2 Homoaconitate hydratase In early studies on the aminoadipate pathway S. cerevisiae lys4 mutants were characterized that were auxotrophic for lysine accumulated homocitrate and cis-homoaconitate and lacked homoaconitate hydratase activity.16 Also identified were lys7 mutants which accumulated homocitrate and exhibited homocitrate synthase and homoaconitate hydratase activity. Based on these findings it was proposed that two distinct enzymes are needed for the conversion of homocitrate to homoisocitrate (Scheme 3).The enzymatic conversion of homoisocitrate to cis- Scheme 3 homoaconitate was first demonstrated in 1966 by following the formation of a chromophore at 240 nm.17 The transformation of cis-homoaconitate but not the trans-isomer to homoisocitrate by cell free extracts was also documented. The enzyme responsible for the activity homoaconitate hydratase (homoaconitase; EC 4.2.1.36) could be separated from the highly similar cis-aconitate hydratase by ammonium sulfate precipitation. Subsequent studies showed that cis-homoaconitate hydratase is a mitochondrial enzyme that is repressed by both lysine and glucose and established homoisocitrate as an intermediate in the biosynthesis of lysine and cis-homoaconitate as a potential intermediate.18 References to homocitrate dehydrase an enzyme proposed to form cis-homoaconitate from homocitrate appear in the literature but the activity has not been correlated with a purified protein.The two-step conversion of homocitrate to homoisocitrate is analogous to the aconitase reaction for the transformation of citrate to isocitrate in the TCA cycle. The reaction proceeds via enzyme bound cis-aconitate and is carried out by a single protein. Isopropylmalate isomerase catalyzes similar sequential elimination–addition reactions in the leucine biosynthetic 87 Nat. Prod. Rep. 2000 17 85–97 pathway. Thus while cis-homoaconitate can be detected in the cells and serves as a precursor to homoisocitrate the majority of cis-homoaconitate may be enzyme bound during the normal reaction.A small amount of free cis-aconitate is detectable in cells and in labeled form and it has been shown to serve as a precursor to isocitrate. The LYS4 gene was originally cloned by functional complementation on a 7.8 kb fragment of S. cerevisiae genomic DNA that also contained the linked LYS15 gene.19 When inserted into the YEp13 vector this fragment complemented lys4 lys15 and lys4lys15 mutations. Gamonet and Lauquin also cloned LYS4 and submitted the sequence of the homoaconitase gene to GenBank (accession X93502). Weidner et al. cloned the LYSF gene from the filamentous fungus Aspergillus nidulans and showed it encodes homoaconitase.20 The gene is 2397 bp in size contains a single 72 bp intron and the ORF encodes a 775 amino acid protein of approximately 84 kDa.Growing A. nidulans in the presence of lysine decreased homoaconitase specific activity by six-fold suggesting regulation by lysine. The A. nidulans enzyme exhibits 55% sequence identity at the amino acid level with S. cerevisiae homoaconitase. The greatest similarity is seen in regions containing cysteines believed to be involved in forming an Fe–S cluster analogous to that found in aconitase. Both the A. nidulans and S. cerevisiae homoaconitases share only 20% similarity with S. cerevisiae aconitase.20,21 A sequence analysis study of homoaconitase and members of the aconitase-family proteins appeared that explores the evolutionary relationship of the enzyme from the lysine biosynthetic pathway with aconitase isopropylmalate isomerase and iron-responsive element binding proteins from eukaryotes archaea and eubacteria.22 Sequence similarities are found primarily in residues associated with the Fe4–S4 cluster and the findings are proposed to provide an evolutionary link between genes found in the aminoadipate pathway and organisms other than fungi.It was suggested that homoaconitase be added to the aconitase-family proteins. As mentioned above lys7 mutants accumulate homocitrate while exhibiting homocitrate synthase and homoaconitate hydratase activity. This was interpreted as evidence that LYS7 may code for a homocitrate dehydrase. The S. cerevisiae LYS7 gene has now been cloned and is predicted to encode a unique 249 amino acid protein.23 Transcription was not regulated by lysine-specific or general amino acid control mechanisms and mutants lacking the entire LYS7 gene displayed phenotypes including pH and temperature sensitivity in addition to requiring lysine for growth.The multiple phenotypes observed in LYS7 null mutants and the constitutive expression pattern suggested that the gene product does not have a role confined to lysine biosynthesis. Insight into the actual role of Lys7 came from studies on copper trafficking in yeast. Culotta et al. identified Lys7 as a Cu–Zn superoxide dismutase-specific copper chaperone.24 S. cerevisiae mutants lacking the LYS7 gene had normal amounts of Cu–Zn superoxide dismutase (SOD1) but lacked enzyme activity.When 64Cu was used to label SOD1 in wild-type yeast a single radioactive band was detected. The lys7 null mutants were not radiolabeled with 64Cu but the defect could be complemented by the human copper chaperone for SOD1. Gamonet et al. observed similar results in their work with lys7 mutants.21 S. cerevisiae mutants in which LYS7 had been deleted were auxotrophic for lysine and methionine sensitive to oxygen- and superoxide-generating agents and light irradiation. 21 The latter finding suggests that Lys7 is involved in oxidative stress protection. Normal amounts of Cu–Zn superoxide dismutase (SOD1) were found but there was no detectable dismutase activity. Addition of Cu2+ to the growth medium of cell-free extracts restored enzyme activity.Subcellular localization revealed that the Lys7 is cytosolic and sequence analysis showed it possessed a metal-binding domain similar to other Cu-transporting ATPases. Stretches of the Lys7 sequence share Nat. Prod. Rep. 2000 17 85–97 88 similarity with the dimerization domains of superoxide dismutase but yeast two-hybrid screening failed to confirm an interaction between the two proteins. 2.3 Homoisocitrate dehydrogenase Homoisocitrate dehydrogenase (3-carboxy-2-hydroxyadipate dehydrogenase; EC 1.1.1.87) catalyzes the NAD+-dependent conversion of homoisocitrate to a-ketoadipate (Scheme 4) and Scheme 4 activity is monitored spectrophotometrically by following the reduction of NAD+ at 340 nm.25 The S.cerevisiae enzyme has been purified approximately 500-fold and has an estimated molecular mass of 48 kDa. The enzyme is separable from isocitrate dehydrogenase and has a KM for homoisocitrate of 10 mM a pH optimum of 8.5 for the oxidative decarboxylation of homoisocitrate and a pH optimum of 7.0 for the reductive carboxylation of a-ketoadipate. The Y. lipolytica dehydrogenase has been purified to electrophoretic homogeneity and appears as a single 48 kDa band under denaturing and nondenaturing conditions.26 Native molecular weight estimation by gel filtration revealed several peaks of activity between 49 and 90 kDa. Addition of homoisocitrate to the buffer caused most of the activity to elute at a volume corresponding to 90 kDa and indicates that native homoisocitrate dehydrogenase may function as a dimer.Two mutants lys9 and lys10 were identified which lacked homoisocitrate dehydrogenase activity. The lys9 mutant exhibited some carboxylating activity whereas the lys10 mutant retained NAD+-reducing activity. Other fungi in which homoisocitrate dehydrogenase activity has been detected include the pathogens Candida albicans Filobasidiella neoformans Aspergillus fumigatus and the fission yeast Schizosaccharomyces pombe.27 Levels of homoisocitrate dehydrogenase activity are repressed in S. cerevisiae 7305d (lys14) mutants whether grown in lysine-supplemented or minimal media.28 The activity is constitutively derepressed in lys9 mutants which are deficient in levels of saccharopine reductase.29 The relationship of lys9 and lys14 and the role of Lys14 in regulating lysine biosynthesis gene expression is discussed in Section 2.6.2.4 a-Aminoadipate aminotransferase a-Aminoadipate aminotransferase (EC 2.6.1.39) is a pyridoxal 5A-phosphate-requiring enzyme that catalyzes the l-glutamatedependent reversible formation of a-aminoadipate from aketoadipate (Scheme 5). a-Aminoadipic acid is a key branch Scheme 5 point metabolite that can enter into secondary metabolism leading to the b-lactam antibiotics in some fungi and actinomy-cetes or continue on in the biosynthesis of lysine. Aminotransferase activity has been identified in Torulopsis utilis N. crassa S. cerevisiae and a S. cerevisiae threonine auxotroph thr5.30,31 The majority of information on this step comes from early studies conducted by Matsuda and Ogur in S.cerevisiae. 32,33 The aminoadipate aminotransferase activity is associated with two isozymes that are separable by ion exchange and size exclusion column chromatography. a- Aminoadipate aminotransferase I has a molecular mass of 100 kDa is located in the mitochondria and is noncompetitively inhibited by a-ketoglutarate. This isozyme accepts either glutamate or aspartate as the amine source in the forward reaction but shows weak activity in the reverse direction wherein a-aminoadipate serves as amine donor. a-Aminoadipate aminotransferase II is a cytosolic enzyme with a molecular mass of 140 kDa and a strict requirement for glutamate as the amine source.Unlike a-aminoadipate aminotransferase I the forward reaction catalyzed by isozyme II is not inhibited by aketoglutarate. Neither enzyme was significantly affected by lysine. In analyzing the effect of media composition on enzyme activity a-aminoadipate aminotransferase II was found to be derepressed by a-aminoadipate and slightly repressed by glucose while a-aminoadipate aminotransferase I activity varied only slightly under different growth conditions. A kynurenine aminotransferase (EC 2.6.1.7) was isolated from the yeast Hansenula schneggii which had physical characteristics comparable to those of S. cerevisiae aminoadipate aminotransferase I and utilized a-aminoadipate as a modest alternate substrate.34 This enzyme is associated with tryptophan catabolism and catalyzes the transamination between l-kynurenine and a-ketoglutarate to form glutamate and kynurenic acid (Scheme 6).When a-ketoadipate is used as the amine acceptor activity is nearly identical to that observed with a-ketoglutarate. Scheme 6 Kynurenine and a-aminoadipate aminotransferase activities are well documented in mammalian systems and the reactions appear to be catalyzed by a number of different pyridoxal 5Aphosphate-dependent aminotransferases. Several studies suggested that kynurenine and a-aminoadipate aminotransferase activities are attributed to the same enzyme when efforts to separate the activities were unsuccessful. Mawal and Deshmukh provided evidence that the two activities in rat kidney are associated with two separate proteins whereas Buchli et al.cloned and expressed a soluble protein from the same tissue able to catalyze both reactions.35,36 Cloning of an a-aminoadipate aminotransferase gene solely associated with lysine biosynthesis in fungi remains to be reported. Taking this into consideration with the broad substrate specificity observed for several yeast aminotransferases suggests the transamination reaction converting a-ketoadipate to aaminoadipate may be conducted by an enzyme with dual functions. Aromatic aminotransferase I from S. cerevisiae has been suggested as possibly having such a role.37 This constitutively expressed form of aromatic aminotransferase is active not only with the aromatic amino acids but also with aaminoadipate methionine and leucine when phenylpyruvate is the amine acceptor.The action of aromatic aminotransferase I was reported to account for half of the glutamate:a-ketoadipate transaminase activity detected in cell-free extracts and the enzyme was suggested to be identical to one of the two known a-aminoadipate aminotransferases. Aromatic aminotransferase I has properties of a general aminotransferase which like several aminotransferases of E. coli may be able to play a role in several otherwise unrelated metabolic pathways. 2.5 a-Aminoadipate reductase Reduction of a-aminoadipate to the semialdehyde by aaminoadipate reductase (a-aminoadipate-d-semialdehyde dehydrogenase; EC 1.2.1.31) is arguably the most fascinating transformation in the fungal lysine biosynthetic pathway and is an entirely unique mechanism in primary metabolism (Scheme 7).It was reported as early as 1959 that the reduction Scheme 7 of a-aminoadipate required ATP Mg2+ NADPH and glutathione. Later work suggested that an adenylated intermediate is formed prior to reduction as opposed to the generation of a phosphate mixed anhydride which precedes carboxy reductions in proline and arginine biosynthesis. The activation of amino acid carboxy groups as acyladenylates is most commonly associated with the protein synthesis on ribosomes. It is also the means of activating carboxy groups utilized by nonribosomal peptide synthetases involved in secondary metabolism. Evidence that the reduction required two distinct gene products in S.cerevisiae was first reported by Bhattacharjee’s group in 1970.38 It was observed that both lys2 and lys5 mutants of S. cerevisiae lacked a-aminoadipate reductase activity. Subsequent in vitro studies using crude cell free extracts of lys2 and lys5 mutants revealed that reductase activity could be regained by complementation and established that LYS2 and LYS5 were unlinked structural genes.39 In the same study the S. cerevisiae a-aminoadipate reductase was partially purified and found to have an estimated Mr of 180 kDa based on calibrated gel-filtration analysis. The earliest mechanistic proposal for the reduction involved a process whereby the d-carboxy was activated as the adenylate in the first step thus facilitating hydride transfer from NADPH (Scheme 8).Decomposition of the hemiacetal in a third step Scheme 8 yields AMP and a-aminoadipic-d-semialdehyde (7). Support for this mechanism came from demonstrating an a-aminoadipate-dependent ATP–PPi exchange activity and from the observed accumulation of an adenylated a-aminoadipate derivative in a lysine auxotroph unable to grow on a-aminoadipate. 89 Nat. Prod. Rep. 2000 17 85–97 The original mechanistic proposal suggested the first two steps were catalyzed by Lys2 and the third step was catalyzed by Lys5.40 These gene products were proposed to form a heterodimeric multifunctional a-aminoadipate reductase. Wild-type S. cerevisiae cannot use a-aminoadipate as a sole nitrogen source whereas lys2 and lys5 mutants can.In fact growth of wild-type cells is inhibited by a-aminoadipate when the yeast is grown in a medium containing other amino acids as the primary nitrogen source. The mutant strains are able to use the exogenous a-aminoadipate as amine donor in the reversal of the normal a-aminoadipate aminotransferase reaction to yield l-glutamate which then serves as a more general nitrogen donor. Zaret and Sherman investigated why normal strains of S. cerevisiae were unable to use a-aminoadipate as a nitrogen source and found that high levels of 6 resulted in the accumulation of a toxic metabolite.41 The accumulated intermediate was not structurally characterized but was suggested to be the semialdehyde 7. They also demonstrated that the product did not accumulate in lys2 and lys5 mutants.Growth of lys9 mutants which are deficient in saccharopine reductase the enzyme that uses 7 as a substrate was inhibited to a greater extent by a-aminoadipate than that of wild-type cells. Growth of lys2lys9 and lys5lys9 double mutants was unaffected by aaminoadipate. Insight into the roles of the LYS2 and LYS5 gene products emerged with the cloning and sequencing of the respective genes. The high frequency of obtaining lys2 mutants coupled with the ease of positive selection using a-aminoadipate as the sole nitrogen source facilitated the cloning of LYS2 by functional complementation. The first reported cloning of LYS2 from S. cerevisiae appeared in 1983 and was followed shortly after by two other reports.42–44 The complete nucleotide sequence of S.cerevisiae LYS2 was reported in 1991 and showed the gene encoded a 1392 amino acid protein having a calculated molecular weight of 155 kDa.45 When the translated sequence of the LYS2 ORF was used to search the GenBank database for proteins with similar amino acid sequences Lys2 showed the greatest homology with the adenylation and thiolation domains of tyrocidine synthetase 1. This enzyme is the product of the Bacillus brevis tycA gene and a component of the nonribosomal peptide synthetase complex involved in the assembly of the peptide antibiotic tyrocidine.46 Nonribosomal peptide synthetases are multifunctional modular proteins that catalyze the sequence-specific condensation of amino acids into peptides without involvement of a nucleic acid template.The ATP-dependent activation of the amino acid carboxy group results in the formation of an amino acyladenylate. This prepares the amino acid for transfer to the thiol group of a 4A-phosphopantetheine (Ppant) cofactor to form a covalent thioester intermediate. The Ppant cofactor is covalently attached to a strictly conserved serine in the thiolation domain of the enzyme. At this stage the amino acid may be structurally modified (e.g. epimerized or N-methylated) or condense to form a peptide bond with an amino acid on an adjacent module which has been similarly activated and attached to the peptide synthetase. During studies on the biosynthesis of saframycin Mx1 in the myxobacterium Myxococcus xanthus Pospiech et al.identified a peptide synthetase ORF (SafA2) coding for two amino acid activation modules one of which is followed by a 370 amino acid region exhibiting 33% identity to the carboxy terminus of S. cerevisiae Lys2.47 The region is similar to ketoreductase domains associated with polyketide synthases and possesses a characteristic sequence motif for a NAD(P)H binding site. This represented the first report of an oxidoreductase domain associated within a peptide synthetase module. The LYS5 gene was subsequently cloned by Borell and Bhattacharjee in 1988 by functional complementation of a S. cerevisiae lys5 mutant.48 The gene was contained on a 7.5 kb DNA fragment and shown by restriction analysis to have no Nat. Prod. Rep.2000 17 85–97 90 homology with LYS2. Further subcloning established that the LYS5 gene could not be any greater than 1.65 kb.49 Hence the encoded protein must be significantly smaller than the 155 kDa LYS2 product. In 1996 the complete nucleotide sequence of the S. cerevisiae LYS5 gene was determined and predicted to encode a 272 amino acid 31 kDa protein.50 In combination with the 155 kDa Lys2 protein this was viewed as supporting evidence of a 180 kDa heterodimeric reductase. Database searching at the time failed to identify any proteins with significant homology to Lys5. The fission yeast S. pombe also requires two distinct genes for the ATP-dependent reduction of a-aminoadipate. The S. pombe LYS1 and LYS7 genes correspond to S. cerevisiae LYS2 and LYS5 respectively.The LYS1 gene was cloned and partially characterized and the predicted amino acid sequence from a fragment of the gene was found to share 49% amino acid sequence identity with the amino terminal region of S. cerevisiae Lys2.51 More recently the S. pombe LYS1 gene has been shown to encode a 1415 residue protein with a calculated molecular weight of 155.8 kDa.52 Analysis of the deduced amino acid sequence of S. pombe Lys1 revealed a 52% overall identity to S. cerevisiae Lys2 and a strong homology to peptide synthetases including ACV synthetases from P. chrysogenum and A. nidulans. ACV synthetase adenylates the d-carboxylate of a-aminoadipate in the process of forming the tripeptide d-(la-aminoadipyl)-l-cysteinyl-D-valine (ACV) the linear precursor to the penicillins and cephalosporins (Scheme 9).Scheme 9 Initial steps of penicillin biosynthesis assembly of the primary amino acid precursors into ACV and conversion to isopenicillin N. A 4.8 kb fragment of C. albicans DNA able to complement lys2 mutants of both C. albicans and S. cerevisiae has been cloned and characterized.53 Sequence analysis of this DNA fragment revealed an ORF corresponding to a 1391 residue protein with an estimated Mr of 154 kDa. This C. albicans LYS2 ORF showed 63% identity to the corresponding S. cerevisiae gene at the nucleotide level. Analysis of the deduced amino acid sequence revealed regions with a high level of similarity to ACV synthetases from several species and a region toward the C-terminus similar to a signature sequence found in short chain alcohol dehydrogenases.The study reporting the cloning of C. albicans LYS2 also provided insight into the repression of gene transcription and regulation of a-aminoadipate reductase activity by lysine. Northern analysis of mRNA levels in cells grown in minimal medium showed that transcription of LYS2 is substantially depressed when the medium is supplemented with lysine. When the cells were grown in a rich medium the LYS2 mRNA could not be detected suggesting the gene is under general amino acid control and also regulated to a certain degree by lysine. The same trend was observed for a-aminoadipate reductase activity. Enzyme activity in cells grown in minimal medium supplemented with lysine was 40% lower and that from cells grown in rich medium was 66% lower compared to cells grown in minimal medium.Lysine also causes feedback inhibition of a-aminoadipate reductase. Addition of 50 mM lysine to assay mixtures reduced a-aminoadipate reductase activity by 70% and a 1 mM concentration of the lysine analog (S)-2-aminoethyl-l-cysteine reduced activity by 92%. The P. chrysogenum LYS2 gene has also been cloned and characterized.54 The gene codes for a 1409 residue protein and exhibits 49.9% identity to the S. cerevisiae gene and 51.3% and 48.1% identity to the S. pombe and C. albicans a-aminoadipate reductase genes respectively. When compared with the aaminoadipate activating region of ACV synthetase a modest 12.4% identity at the amino acid level was observed.The P. chrysogenum LYS2 gene is located on chromosome III in strain AS-P-78 and on chromosome IV in strain P2 whereas the penicillin biosynthetic gene cluster is located on chromosome I in both strains. This is consistent with evidence that the aminoadipate pool required for penicillin production originates from lysine degradation rather than from the pool generated during lysine biosynthesis.55 A 160 kDa a-aminoadipate reductase has been partially characterized from Candida maltosa and exhibits properties similar to those observed for the enzyme from S. cerevisiae.56 A subsequent report by the same group revealed that C. maltosa mutants deficient in a-aminoadipate reductase activity could have the activity partially restored by transformation with plasmids carrying S.cerevisiae LYS2.57 The partial (8–22%) recovery of activity was attributed to unstable plasmids rapid degradation of S. cerevisiae Lys2 weak transcription of the S. cerevisiae gene and/or incorrect processing. With the finding that Lys2 must be post-translationally modified by Lys5 (see below) the latter explanation may include S. cerevisiae Lys2 serving only as a poor substrate for C. maltosa Lys5. In 1996 Lambalot et al. described a new enzyme family involved in the post-translational modification of enzymes requiring a covalent Ppant cofactor.58 These phosphopantetheinyl transferases (PPTases) catalyze the transfer of the 4Aphosphopantetheine moiety from coenzyme A to a conserved serine residue found in all peptidyl carrier protein (PCP) domains of nonribosomal peptide synthetases.Similar PPTases are involved in cofactor attachment to the acyl carrier domains of polyketide synthases thereby converting the inactive apo form of the enzyme to the active holo form. Sequence homology searching revealed that Lys5 showed a strong similarity to members of this newly described family of enzymes and the proposal was made that Lys5 promoted the phosphopantetheinylation of Lys2 and that Lys2 carries out both the adenylation and reduction of the a-aminoadipate side chain carboxy. Very recently Walsh and co-workers further dissected the relationship of Lys2 and Lys5.59 Sequence analysis of Lys2 revealed three functional domains; a 60 kDa peptide synthetaselike adenylation (A) domain (residues 225–808) a 14 kDa peptidyl carrier protein (PCP) domain (residues 809–924) and the reductase (R) domain (amino acids 925–1392).The first 224 residues of the protein may constitute a fourth domain but this region has not been assigned a function. Recombinant Lys5 Lys2 and truncated fragments of Lys2 were all overproduced in E. coli and used to assess the function of Lys5 and the roles of the A PCP and R domains. Using autoradiography and mass spectrometry Lys5 was shown to be capable of transferring [3H]coenzyme A to the 14 kDa Lys2-PCP domain. To obtain evidence that the Lys2 A domain adenylates a-aminoadipate a 105 kDa Lys2 A/PCP fragment was used in ATP–[32P]PPi exchange assays. This truncated Lys2 catalyzed the l-aaminoadipate-dependent incorporation of [32P]PPi into ATP whereas D-a-aminoadipate and the homologs DL-diaminopimelate and l-glutamate were poor substrates or produced no exchange activity.The aminoadipate analog (S)-carboxymethyl-l-cysteine served as a modest alternate substrate for the adenylation reaction and was used to demonstrate the formation of a covalent thioester intermediate. Thus incubation of [35S]- (S)-carboxymethyl-l-cysteine with Lys2 A/PCP followed by protein precipitation resulted in the association of 35S with the protein. Covalent radiolabeling could also be demonstrated by autoradiography after gel electrophoresis. However when [35S](S)-carboxymethyl-l-cysteine was incubated with holo- Lys2 followed by NADPH addition and protein precipitation the 35S was no longer associated with the protein thus illustrating a unique function for the reductase domain.Most enzymes employing covalent thioester intermediates release an acid product resulting from hydrolysis of the thioester. Alternately an intramolecular nucleophile such as an amine or hydroxy can attack the thioester and result in the release of a lactam or lactone respectively (e.g. cyclic peptides and depsipeptides). In the case of a-aminoadipate reductase reduction of the thioester yields a thiohemiacetal that decomposes to release the aldehyde. Hence LYS2 is the gene coding for apo-a-aminoadipate reductase and LYS5 encodes the requisite Lys2-PPTase. The overall process is outlined in Scheme 10.The extensive investigation of this step in lysine biosynthesis is in part due to the novel mechanism and also because of the usefulness of LYS2 as a tool for genetic studies. Because lys2 mutants require exogenous lysine for growth this property serves as a useful selection marker to identify strains possessing a wild-type gene. Furthermore lys2 mutants can be identified by their ability to grow on synthetic media having aaminoadipate as the sole nitrogen source. This ability to select for and against a given yeast gene is a trait shared only with the URA3 gene involved in uracil biosynthesis. The LYS2 gene has been used in this capacity in the construction of both replicating and integrative yeast transformation vectors.42–44 2.6 Saccharopine reductase Saccharopine reductase [saccharopine dehydrogenase (NADP+ l-glutamate forming) EC 1.5.1.10] catalyzes the condensation of a-aminoadipate-d-semialdehyde with l-glutamate and in the presence of NADPH produces l-saccharopine the penultimate product of the aminoadipate pathway (Scheme 11).Saccharopine reductase has been purified to near homogeneity from S. cerevisiae and shown to be a monomer of approximately 50 kDa.60 The reaction is reversible with the forward direction being favored at physiological pH. Because the semialdehyde is not readily available the enzyme is typically assayed in the reverse direction at pH 9.5 using saccharopine and NAD+ or NADP+ as substrates. The KMs for saccharopine and NAD+ were estimated at 2.3 mM and 0.05 mM respectively.Saccharopine formation in S. cerevisiae also requires the products of two unlinked genes LYS9 and LYS14.61 Both lys14 and lys9 mutants accumulate a-aminoadipate-d-semialdehyde and lack significant levels of saccharopine reductase. Hence it was concluded that Lys9 and Lys14 were both necessary for the biosynthesis of saccharopine reductase in wild-type cells. S. cerevisiae lys9 mutants are auxotrophic for lysine and do not exhibit significant saccharopine reductase activity whereas lys14 mutants grow slowly on media lacking lysine and retain low levels of reductase activity.62 The introduction of a plasmid carrying LYS9 into a lys14 mutant restored the saccharopine reductase activity to wild-type levels and conferred the ability to grow on minimal media.However the introduction of a plasmid harboring the LYS14 gene into a lys9 mutant did not complement the mutation and indicates that LYS9 is a structural protein and Lys14 is required for the expression of LYS9. Northern analysis to quantitate LYS9 mRNA in a lys14 mutant supports 91 Nat. Prod. Rep. 2000 17 85–97 Scheme 10 Illustrated summary of the proposed roles of Lys2 and Lys5 in the activation and reduction of a-aminoadipic acid. Scheme 11 this conclusion. In addition to the identification of a regulatory role for Lys14 a-aminoadipate-d-semialdehyde was found to serve as a coinducer of transcriptional activation. Low levels of saccharopine reductase activity in lys14 mutants can be explained by weak expression of LYS9 in the absence of induction by the semialdehyde.2.7 Saccharopine dehydrogenase The final step of the lysine biosynthetic pathway in fungi is the cleavage of saccharopine to yield a-ketoglutarate and l-lysine (Scheme 12). The NAD+-dependent oxidation is catalyzed by Scheme 12 saccharopine dehydrogenase (NAD+ l-lysine forming); EC 1.5.1.7. Fujioka and co-workers conducted extensive studies on S. cerevisiae saccharopine dehydrogenase during the 1970’s and early 1980’s. Their work revealed saccharopine dehydrogenase is a basic protein (isoelectric pH = 10.1) that exists as a 39 kDa monomer containing a single active site.63 The oxidation is reversible with the reaction yielding lysine exhibiting a maximum rate at pH 10 while the reverse reaction is favored at pH 7.The KMs for saccharopine and NAD+ were estimated to be 1.7 mM and 0.1 mM respectively. When measuring saccharopine formation the Michaelis constants for lysine a-ketoglutarate and NADH were determined to be 2.0 mM 0.55 mM and 0.089 mM respectively. A mechanism for the catalytic reaction based on initial rate pH studies and product/dead end inhibition studies has been proposed to involve initial oxidation to an iminium ion followed by addition of water to give the hemiaminal which cleaves to products.64 The reaction appears to follow a Bi-Ter mechanism in which Nat. Prod. Rep. 2000 17 85–97 92 binding of NAD+ precedes that of saccharopine. Lysine is the first product released followed by a-ketoglutarate and then NADH.65 Fujioka and Takata showed that the dehydrogenase catalyzes the stereospecific transfer of the hydrogen on C-2 of the saccharopine glutaryl moiety to the pro-R position at C-4 of NAD+.66 A search for alternate substrates of the S.cerevisiae dehydrogenase revealed a very high degree of substrate specificity. The only a-ketoacid evaluated that was able to substitute for a-ketoglutarate in the reverse reaction was pyruvate.67 A series of chemical modification experiments revealed that the active site of the enzyme possesses essential histidine,68 lysine,69 arginine,70 and cysteine71 residues. An eleven amino acid peptide containing the cysteine residue necessary for catalysis has been isolated and sequenced.72 The LYS1 gene encoding the S.cerevisiae dehydrogenase was cloned during studies on the regulatory role of LYS14 and the sequence deposited in GenBank (accession X77632).62 The corresponding gene has also been cloned from a number of other yeasts. In Y. lipolytica the LYS5 gene was cloned by complementation and shown to encode saccharopine dehydrogenase. 73 A single 1.5 kb RNA transcript was identified that hybridized with an internal fragment of LYS5 consistent with the 40 kDa molecular weight of the enzyme. Subsequent sequencing of a 2.5 kb DNA fragment complementing lys5 mutants of Y. lipolytica revealed two completely overlapping antiparallel ORFs. Site-directed mutagenesis studies and gene fusion experiments addressing the transcription and translation of the two ORFs established that ORF2 encodes saccharopine dehydrogenase.A function for the ORF1 product is unknown although the possibility that the RNA serves as an antisense regulator of ORF2 expression similar to situations found in prokaryotes was suggested. In N. crassa lys4 mutants are auxotrophic for lysine do not exhibit saccharopine dehydrogenase activity and accumulate saccharopine.74 Similar observations are seen in S. pombe lys3 mutants75 and C. albicans lys1 mutants.76 The cloned LYS1 gene from C. albicans has been localized to a 1.8 kb EcoR V-EcoR I restriction fragment. Transformation of C. albicans or S. cerevisiae lys1 mutants with plasmids harboring this DNA fragment results in prototrophs having significant saccharopine dehydrogenase activity.76 Following up on their earlier work with C.albicans Garrad et al. sequenced a DNA fragment carrying the LYS1 gene and showed it contains a 1.1 kb ORF encoding a 382 amino acid protein.77 C. albicans LYS1 exhibits 60% and 69% similarity at the nucleotide level with Y. lipolytica LYS5 and S. cerevisiae LYS1 respectively. C. albicans lysine auxotrophs are nonpathogenic in experimental infections in mice.78 Because the uniqueness of the saccharopine structure may lead to target specificity inhibition of saccharopine dehydrogenase might be a viable means of controlling opportunistic fungal pathogens. To test this hypothesis amide analogs of saccharopine (Fig. 1) were prepared and Fig. 1 Substrate analogs prepared as inhibitors of saccharopine dehydrogenase.evaluated as inhibitors of the commercially available S. cerevisiae enzyme.79 Each analog was evaluated at an initial concentration of 2 mM in the presence of 1.7 mM saccharopine and 0.33 mM NAD+. Compound 10 in which the overall number of atoms and the disposition of the carboxy groups and nitrogens is the same as saccharopine showed significant inhibition of the enzyme. The a-aminopimelate analog 11 produced a very modest inhibition. Further characterization of 10 established a Ki of 0.12 mM. None of the compounds in Fig. 1 served as alternate substrates for saccharopine dehydrogenase. Thus the substrate specificity of the dehydrogenase for saccharopine in the forward direction appears to be nearly as strict as the a-ketoacid specificity in the reverse reaction.Also none of the derivatives in Fig. 1 were successful at affecting the growth of S. cerevisiae or C. albicans on solid medium. This may be in part due to poor uptake of these triacids into the cell. 3 Role of pipecolic acid in lysine metabolism l-Pipecolic acid is a common lysine metabolite found in various organisms including bacteria yeast plants and mammals.80,81 Numerous studies have demonstrated that l-pipecolic acid (l- PA) is primarily a product of lysine degradation although it can serve a nutritional role in some species of bacteria and yeast. Certain lysine auxotrophs of the aerobic red yeast Rhodotorula glutinis can grow on a minimal medium supplemented with l- PA.82,83 This yeast is able to use pipecolate as the sole nitrogen source but not as a carbon source.The studies showed that l-PA (16) serves as a precursor to lysine through oxidation to D1- piperideine-6-carboxylate (D1-P6C 17) and hydration to aaminoadipate-d-semialdehyde (7) which is converted to saccharopine (8) in the presence of glutamate and NADPH (Scheme 13).83 Following up on this finding Kinzel and Bhattacharjee reported the purification to near homogeneity and characterization of pipecolic acid oxidase from R. glutinis.84 The enzyme is a 43 kDa monomer exhibiting optimum activity at pH 8.5 with an apparent KM for l-PA of 1.67 mM. Molecular oxygen is required for activity and H2O2 is produced along with D1-P6C. Surprisingly the authors were unable to identify a flavin metal ion or any other redox cofactor to be associated with the enzyme.Pipecolic acid occurs as a minor intermediate of lysine metabolism in most mammalian tissues85–87 and a pipecolate oxidase has been purified from primate liver that shares many similarities with the yeast enzyme. The primate flavoenzyme is a membrane-associated 46 kDa monomer possessing a covalent Scheme 13 flavin with an apparent KM for pipecolic acid of 3.7 mM.88 Unlike the yeast enzyme which is reversibly inhibited by proline the primate oxidase can utilize l-proline as a poor alternate substrate. Evidence that pipecolic acid has neuromodulatory properties in the central nervous system (CNS) mediated through interactions with GABAA receptors has prompted the preparation of specific inhibitors of the primate enzyme.89–92 Differences between the yeast and primate enzymes are also observed in their ability to be inhibited by various pipecolate analogs and recognize others as alternate substrates.93 Furthermore the R.glutinis system is unique in that l-PA serves as a precursor to lysine rather than a lysine catabolite. The formation of pipecolic acid in fungi and yeast has been suggested to originate from both d- and l-lysine. Involvement of a lysine racemase was proposed by one group to account for this result94,95 while the intermediacy of l-pipecolate in the conversion of d-lysine to l-lysine was suggested by another.96 Fangmeier and Leistner sought to clarify the issue using d-[a- 15N]lysine and D-[e-15N]lysine in incorporation studies with N.crassa.97 When D-[e-15N]lysine (18) was administered to cultures of the fungus the label was detected in the a position of l-lysine (9) as well as in l-PA (16). However the l-PA and llysine isolated from cultures receiving d-[a-15N]lysine were not labeled with 15N. If an amino acid racemase was operating the isotopic label from 18 should be retained in the conversion to 9. When alanine and glutamate were isolated from the d-[a- 15N]lysine incorporation experiments they were enriched in 15N indicating a-ketoglutarate and pyruvate had served as nitrogen acceptors in transamination reactions. These observations support the pathway illustrated in Scheme 14 wherein a four step sequence converts d-lysine to l-lysine via the intermediacy of l-pipecolate.Scheme 14 The parasitic fungus Rhizoctonia leguminicola incorporates l-PA into the toxic octahydroindolizine alkaloids slaframine 93 Nat. Prod. Rep. 2000 17 85–97 and swainsonine the latter being a potent a-mannosidase inhibitor.98 Early studies on the biosynthesis of these alkaloids reported that pipecolate derived primarily from l-lysine via 6-amino-2-oxocaproic acid (19) and the imine D1-P2C (20).94 Because some pipecolate was found to originate from d-lysine the action of a lysine racemase was proposed. Reinvestigation of l-PA biosynthesis in R. leguminicola led to the unexpected finding that l-lysine was actually incorporated first into saccharopine (8) and then converted to 7.99 The enzyme converting 8 to 7 saccharopine oxidase is a 45 kDa monomeric flavoenzyme requiring O2 and producing the semialdehyde glutamate and H2O2.100 Semialdehyde 7 spontaneously cyclizes and dehydrates to form D1-P6C (17) which serves as substrate for an NADPH-requiring reductase yielding l-pipecolate (16) (Scheme 15).The l-PA produced by this route would possess a nitrogen originating from the a-amine of l-lysine and may explain findings that both d- and l-lysine can serve as precursors to 16. 4 Lysine catabolism Scheme 15 Role of saccharopine and pipecolic acid in R. leguminacola alkaloid biosynthesis. Lysine degradation is extremely varied in Nature and the nine known catabolic fates are depicted in Scheme 16. In mammals the major pathway of l-lysine degradation is through steps formally equivalent to the reversal of the fungal biosynthetic pathway.101–103 The first two steps of the process involve the sequential action of l-lysine-a-ketoglutarate reductase (Scheme 16; xii) and saccharopine dehydrogenase (glutamate forming) (Scheme 16; xiii).The net process is effectively a transamination with the e-amino group of l-lysine being transferred to a-ketoglutarate producing a-aminoadipate-dsemialdehyde and l-glutamate. In bovine and baboon liver and in human placenta both of these enzyme activities are associated with aminoadipic semialdehyde synthase a large (470–480 kDa) bifunctional tetramer composed of four identical subunits.104 In rat the activities are separable; the reductase purified from liver mitochondria is a tetramer with an apparent Scheme 16 Enzymes involved in the initial step of lysine degradation in various species i lysine 6-dehydrogenase EC 1.4.1.18; ii lysine:a-ketoglutarate e-aminotransferase EC 2.6.1.36; iii lysine:pyruvate 6-aminotransferase EC 2.6.1.71; iv lysine decarboxylase EC 4.1.1.18; v lysine oxidase EC 1.4.3.14; vi lysine dehydrogenase EC 1.4.1.15; vii lysine 2,3-aminomutase EC 5.4.3.2; viii lysine N6-hydroxylase EC 1.14.13.59 (lysine 6-monooxygenase (NADPH)); ix lysine N6-acetyltransferase EC 2.3.1.32; x lysine racemase EC 5.1.1.5; xi lysine 2-monooxygenase EC 1.13.12.2 (lysine oxygenase); xii saccharopine dehydrogenase (NADP+ L-lysine forming) EC 1.5.1.8 (lysine-a-ketoglutarate reductase); xiii saccharopine dehydrogenase (NAD+ L-glutamate forming) EC 1.5.1.9.Nat. Prod. Rep. 2000 17 85–97 94 MW 230 kDa and a subunit MW of 52 kDa while the dehydrogenase is a monomer of 43 kDa.105 Interestingly saccharopine is not produced in the CNS. In rat monkey and human brain l-lysine is specifically metabolized to l-pipecolate.86,87 In other tissues notably liver and kidney where the a-aminoadipate pathway is functional l-PA formation is a secondary process and D-lysine appears to be the precursor.86,106,107 Nitrogen-15 labeling studies showed that formation of l-pipecolate in mammals proceeds through the oxidative deamination of both d- and l-lysine to 6-amino- 2-oxocaproic acid.108 To date a lysine oxidase (Scheme 16; v) has not been characterized from brain. Similar to mammals plants degrade lysine primarily via saccharopine109 with some species also producing pipecolate.110,111 Bacteria alter l-lysine in the greatest number of ways.Clostridia and several other bacteria are able to process a-llysine to b-l-lysine through the action of lysine 2,3-aminomutase (Scheme 16; vii).112 The oxidative decarboxylation of lysine to 5-aminovaleramide has been observed in Pseudomonads possessing lysine 2-monooxygenase (Scheme 16; xi).113,114 Biosynthesis of the E. coli iron siderophore aerobactin begins with the formation of N6-hydroxylysine by a FAD-dependent monooxygenase (Scheme 16; viii). Some Pseudomonas species are able to decarboxylate lysine to yield cadaverine (Scheme 16; iv) and the decarboxylase has been purified from E.coli115 and Bacillus cadaveris.116 Lysine 6-dehydrogenase (Scheme 16; i) has been isolated from Agrobacterium tumefaciens and the product shown to be a-aminoadipate-d-semialdehyde.117 The same product is formed during lysine catabolism by several Pseudomonas species that possess lysine 6-aminotransferase activity (Scheme 16; ii).114 a-Aminoadipate-d-semialdehyde is the first intermediate in the utilization of lysine for b-lactam biosynthesis and lysine 6-aminotransferase has been identified in the b-lactam producers Streptomyces clavuligerus118 and Nocardia lactamdurans. 119 A survey of 28 yeast strains identified two lysine degradative pathways and separated the organisms into three groups depending on the first enzymatic step of the catabolism.120 Entry into the first pathway begins with formation of aaminoadipate-d-semialdehyde by either lysine 6-dehydrogenase (Scheme 16; i)121 or lysine 6-aminotransferase.Aminotransferases utilizing either pyruvate (Scheme 16; iii)122 or a-ketoglutarate (Scheme 16; ii)123 as amine acceptor have been found. Lysine 6-dehydrogenase activity was observed only in C. albicans and Kluyveromyces marxianus. Subsequent studies with the pathogen C. albicans revealed the enzyme accepts alternate substrates such as 4- and 5-hydroxylysine and thialysine.124 A second degradative route proceeds through acetylated intermediates. In some species the first step is catalyzed by lysine N6-acetyltransferase (Scheme 16; ix) followed by loss of the a-amine through transamination with aketoglutarate.125 The gene for a 391 residue lysine N6- acetyltransferase LYC1 from Y. lipolytica has been cloned and functionally expressed.126 Lysine racemase (Scheme 16; x) was suggested to function during the formation of pipecolic acid in R. leguminicola (Section 3).94 The activity was sensitive to hydroxylamine suggesting involvement of a pyridoxyl 5A-phosphate cofactor. This species also catabolizes lysine via acetylated intermediates. 95 The fungus Trichoderma viride produces a lysine a-oxidase that generates 6-amino-2-oxocaproic acid from l-lysine (Scheme 16; v).127 The enzyme is a FAD-containing homodimer comprised of two 56 kDa subunits and reportedly possesses in vitro and in vivo antitumor activity.DNA synthesis was more greatly affected than protein synthesis and the antiproliferative property was suggested to arise from the combined deprivation of lysine and formation of D1-P2C and H2O2.128 The red yeast R. glutinis can utilize lysine as the sole nitrogen source and lysine:a-ketoglutarate aminotransferase was shown to catalyze the transamination of the e-nitrogen of lysine to the a-position of glutamate with the concomitant formation of aaminoadipate-d-semialdehyde (Scheme 16; ii).123 Glutamate then serves as a more widely used nitrogen donor. Enzyme activity is markedly increased when cells were grown on lysine as the sole nitrogen source but is also detectable when cells received ammonia as their only source of nitrogen. This enzyme has been purified from C.utilis and characterized as a 83 kDa dimer possessing two identical 40 kDa subunits and requiring pyridoxal 5A-phosphate.129 Whereas a-ketoglutarate was the preferred amine acceptor oxaloacetate pyruvate and 2-oxoadipate can serve in this role. 5 Role of a-aminoadipate pathway intermediates in secondary metabolism There are numerous examples of fungal alkaloids or peptides that have lysine as a structural element or biosynthetic precursor and a survey of these falls outside the scope of this review. There are also cases where aminoadipate pathway intermediates are incorporated into secondary metabolites; one example is saccharopine serving as a precursor to slaframine and swainsonine cited above. Certainly the most well known and thoroughly studied case is the incorporation of a-aminoadipate into ACV the linear tripeptide precursor to the penicillins (Scheme 9 Section 2.5).The importance of a-aminoadipate availability on penicillin production is well documented. Adding a-aminoadipate to fermentations of P. chrysogenum can elevate the rate of ACV formation and levels of penicillin production.130,131 Furthermore lysine causes feedback inhibition of homocitrate synthase resulting in a reduced a-aminoadipate pool and a concomitant decrease in penicillin production.132 More recently Martin and co-workers demonstrated that disruption of the LYS2 gene encoding a-aminoadipate reductase in P. chrysogenum leads to penicillin overproduction.133 Both single and double crossover strategies were used to target the disruption of LYS2.The mutants lacked detectable a-aminoadipate reductase activity and produced penicillin levels more than double those of the parent strain. The increased production likely results from diminished feedback inhibition by lysine and an increased flux of aminoadipate towards penicillin production. 134 Martin’s group also presented evidence that it is lysine catabolism in P. chrysogenum that yields the a-aminoadipate entering the penicillin biosynthetic pathway.135 They identified a P. chrysogenum lysine auxotroph unable to form aaminoadipate but which could still produce penicillin when supplemented with lysine.136 Further studies with this mutant indicated that [U-14C]lysine was incorporated into saccharopine and a-aminoadipic acid providing evidence that the catabolism proceeded by a reversal of the sequential actions of saccharopine dehydrogenase and saccharopine reductase.The aaminoadipate pool for penicillin production in P. chrysogenum is sequestered from the cytosol.137 Vacuoles were later identified as the sequestration site and shown to contain cysteine and valine as well as having ACV synthetase loosely associated with the vacuolar membrane.55 Hence the substrates and enzyme for the assembly of the first committed precursor to penicillin are compartmentalized although the enzyme using ACV as substrate isopenicillin N synthetase is cytosolic.55 Interestingly lysine 6-aminotransferase the enzyme that converts lysine directly to a-aminoadipate-d-semialdehyde was also detected in the P.chrysogenum mutant and a NAD+- dependent oxidation of a-aminoadipate-d-semialdehyde to aaminoadipate was observed. This particular catabolic route is well known in actinomycetes that produce penicillins.138 The gene for lysine 6-aminotransferase is associated with the cephamycin biosynthetic gene cluster in N. lactamdurans119 95 Nat. Prod. Rep. 2000 17 85–97 and the gene encoding piperideine-6-carboxylate dehydrogenase the enzyme oxidizing the semialdehyde to a-aminoadipate is found with the cephamycin gene cluster in S. clavuligerus.139 6 The a-aminoadipate pathway in archaea and bacteria The first evidence that the a-aminoadipate pathway may operate in certain bacteria was encountered in acetate incorporation studies with the thermophilic anaerobic archaeon Thermoproteus neutrophilus.140 Labeling patterns seen in lysine were consistent with a biosynthesis proceeding through the intermediacy of aminoadipate rather than diaminopimelate.This represented the first evidence for operation of the pathway outside of higher fungi and Euglenoids. In the bacterial biosynthesis of lysine threonine and methionine the first step is catalyzed by aspartate kinase and this enzyme is sensitive to feedback inhibition by each of these amino acids. Observations that aspartate kinase from the aerobic Gram-negative thermophile Thermus flavus was inhibited by threonine and methionine but not by lysine suggested that this organism may not use the diaminopimelate pathway.141 Kosuge and Hoshino analyzed the predicted amino acid sequences of two ORFs residing on a 3.8 kb fragment of DNA from the bacterium Thermus thermophilus and found 55% and 45% identity with S.cerevisiae homocitrate synthase and homoaconitase respectively.142 Gene disruption experiments resulted in lysine auxotrophy which could be complemented by aminoadipate but not by diaminopimelate. Similarly Kobashi et al. generated lysine auxotrophs of T. thermophilus and found none could survive on minimal medium supplemented with diaminopimelate but growth was seen when aminoadipate was added.143 A 4.3 kb DNA fragment was cloned that complemented the lysine auxotrophy and sequence analysis identified two genes HACA and HACB that were proposed to encode subunits of homoaconitate hydratase.Mutants with a disrupted HACA gene could not grow on minimal medium unless lysine was added but growth was observed if the disruptants received supplemental a-aminoadipate or a-ketoglutarate. Homocitrate synthase activity was observed in the wild-type T. thermophilus. 7 Conclusion Systemic fungal infections are among the most difficult infectious diseases to treat. The uniqueness of the aminoadipate pathway has prompted speculation that these enzymes may be viable targets for selective antifungal agents. Similarly the novelty of several genes in the pathway may permit them to serve as molecular markers to facilitate rapid identification of fungal pathogens. Bhattacharjee has already demonstrated the feasibility of using PCR to amplify conserved regions in lysine biosynthetic genes to identify C.albicans.53,144 Inasmuch as plants utilize the diaminopimelate pathway to lysine targeting fungal lysine biosynthesis may prove to be an effective fungicidal tactic. Our understanding of the a-aminoadipate pathway to lysine at the molecular level has increased dramatically in recent years paralleling the growth in genomic information. This trend is likely to continue and one can anticipate resolution of the unanswered biochemical questions in the pathway and additional insight into the evolution of the multiple routes for lysine formation and degradation in diverse organisms. 8 Acknowledgements Prof. Christopher Walsh is gratefully acknowledged for providing a preprint of a paper describing the role of Lys 5.Dr Philip Proteau is thanked for critically reading the manuscript. Research in the author’s laboratory on lysine metabolism has Nat. Prod. Rep. 2000 17 85–97 96 been supported in part by NIH grant NS 32421 and the Medical Research Foundation of Oregon. 9 References 1 R. J. Cox Nat. Prod. Rep. 1996 13 29. 2 G. Scapin and J. S. Blanchard Adv. Enzymol. Relat. Areas Mol. Biol. 1998 72 279. 3 R. Ledwidge and J. S. Blanchard Biochemistry 1999 38 3019. 4 J. K. Bhattacharjee Crit. Rev. Microbiol. 1985 12 131. 5 J. K. Bhattacharjee in Evolution of a-Aminoadipate Pathway for the Synthesis of Lysine in Fungi ed. R. P. Mortlock CRC Press Boca 6 M.Strassman and L. N. Ceci Biochem. Biophys. Res. Commun. 1964 7 A. F. Tucci and L. N. Ceci Arch. Biochem. Biophys. 1972 153 8 A. F. Tucci and L. N. Ceci Arch. Biochem. Biophys. 1972 153 9 J. W. Tracy and G. B. Kohlhaw Proc. Natl. Acad. Sci. U.S.A. 1975 Raton FL 1992 pp. 47. 14 262. 742. 751. 72 1802. 10 W. M. Jaklitsch and C. P. Kubicek Biochem. J. 1990 269 247. 11 P. Verhasselt M. Voet and G. Volckaert Yeast 1995 11 961. 12 F. Ramos P. Verhasselt A. Feller P. Peeters A. Wach E. Dubois and G. Volckaert Yeast 1996 12 1315. 13 S. Chen S. Brockenbrough J. E. Dove and J. P. Aris J. Biol. Chem. 1997 272 10839. 14 F. M. Perez-Campo J. M. Nicaud C. Gaillardin and A. Dominguez Yeast 1996 12 1459. 15 O. Banuelos J. Casqueiro F. Fierro M.J. Hijarrubia S. Gutierrez and J. F. Martin Gene 1999 226 51. 16 J. K. Bhattacharjee A. F. Tucci and M. Strassman Arch. Biochem. Biophys. 1968 123 235. 17 M. Strassman and L. N. Ceci J. Biol. Chem. 1966 241 5401. 18 H. Betterton T. Fjellstedt M. Matsuda M. Ogur and R. Tate Biochim. Biophys. Acta 1968 170 459. 19 L. Wang S. Okamoto and J. K. Bhattacharjee Curr. Genet. 1989 16 7. 20 G. Weidner B. Steffan and A. A. Brakhage Mol. Gen. Genet. 1997 255 237. 21 F. Gamonet and G. J. Lauquin Eur. J. Biochem. 1998 251 716. 22 S. D. Irvin and J. K. Bhattacharjee J. Mol. Evol. 1998 46 401. 23 J. Horecka P. T. Kinsey and G. F. J. Sprague Gene 1995 162 87. 24 V. C. Culotta L. W. Klomp J. Strain R. L. Casareno B. Krems and J. D. Gitlin J. Biol. Chem. 1997 272 23469.25 M. Strassman and L. N. Ceci J. Biol. Chem. 1965 240 4357. 26 C. M. Gaillardin A. M. Ribet and H. Heslot Eur. J. Biochem. 1982 128 489. 27 Z. H. Ye R. C. Garrad M. K. Winston and J. K. Bhattacharjee J. Basic Microbiol. 1991 31 149. 28 L. A. Urrestarazu C. W. Borell and J. K. Bhattacharjee Curr. Genet. 1985 9 341. 29 M. K. Winston and J. K. Bhattacharjee Curr. Genet. 1987 11 393. 30 H. P. Broquist A. V. Stiffey and A. M. Albrecht J. Biol. Chem. 1961 9 1. 31 N. Piediscalzi T. Fjellstedt and M. Ogur Biochem. Biophys. Res. Commun. 1968 32 380. 32 M. Matsuda and M. Ogur J. Biol. Chem. 1969 244 3352. 33 M. Matsuda and M. Ogur J. Biol. Chem. 1969 244 5153. 34 Y. Asada Y. Sawa and K. Tanizawa J. Biochem. 1986 99 1101. 35 M. R.Mawal and D. R. Deshmukh J. Biol. Chem. 1991 266 2573. 36 R. Buchli D. Alberati-Giani P. Malherbe C. Kohler C. Broger and A. M. Cesura J. Biol. Chem. 1995 270 29330. 37 A. Urrestarazu S. Vissers I. Iraqui and M. Grenson Mol. Gen. Genet. 1998 257 230. 38 A. K. Sinha and J. K. Bhattacharjee Biochem. Biophys. Res. Commun. 1970 39 1205. 39 D. R. Storts and J. K. Bhattacharjee Biochem. Biophys. Res. Commun. 1989 161 182. 40 A. K. Sinha and J. K. Bhattacharjee Biochem. J. 1971 125 743. 41 K. S. Zaret and F. Sherman J. Bacteriol. 1985 162 579. 42 H. Eibel and P. Philippsen Mol. Gen. Genet. 1983 191 66. 43 D. A. Barnes and J. Thorner Mol. Cell. Biol. 1986 6 2828. 44 U. N. Fleig R. D. Pridmore and P. Philippsen Gene 1986 46 237. 45 M. E. Morris and S.Jinks-Robertson Gene 1991 98 141. 46 R. Weckermann R. Furbass and M. A. Marahiel Nucleic Acids Res. 1988 16 11841. 47 A. Pospiech J. Bietenhader and T. Schupp Microbiology 1996 142 741. 48 C. W. Borell and J. K. Bhattacharjee Curr. Genet. 1988 13 299. 49 S. Rajnarayan J. C. Vaughn and J. K. Bhattacharjee Curr. Genet. 1992 21 13. 50 K. G. Miller and J. K. Bhattacharjee Gene 1996 172 167. 51 R. A. Ford and J. K. Bhattacharjee Curr. Genet. 1995 28 131. 52 V. Bhattacherjee and J. K. Bhattacharjee Yeast 1998 14 479. 53 K. Suvarna L. Seah V. Bhattacherjee and J. K. Bhattacharjee Curr. Genet. 1998 33 268. 54 J. Casqueiro S. Gutierrez O. Banuelos F. Fierro J. Velasco and J. F. Martin Mol. Gen. Genet. 1998 259 549. 55 T. Lendenfeld D. Ghali M. Wolschek E.M. Kubicek-Pranz and C. P. Kubicek J. Biol. Chem. 1993 268 665. 56 H. Schmidt R. Bode M. Lindner and D. Birnbaum J. Basic Microbiol. 1985 25 675. 57 G. Kunze R. Bode H. Schmidt I. A. Samsonova and D. Birnbaum Curr. Genet. 1987 11 385. 58 R. H. Lambalot A. M. Gehring R. S. Flugel P. Zuber M. LaCelle M. A. Maraheil R. Reid C. Khosla and C. T. Walsh Chem. Biol. 1996 3 923. 59 D. E. Ehmann A. M. Gehring and C. T. Walsh Biochemistry 1999 38 6171. 60 D. R. Storts and J. K. Bhattacharjee J. Bacteriol. 1987 169 416. 61 C. W. Borell L. A. Urrestarazu and J. K. Bhattacharjee J. Bacteriol. 1984 159 429. 62 F. Ramos E. Dubois and A. Pierard Eur. J. Biochem. 1988 171 171. 63 H. Ogawa and M. Fujioka J. Biol. Chem. 1978 253 3666. 64 M. Fujioka Arch.Biochem. Biophys. 1984 230 553 and references cited therein. 65 M. Fujioka and Y. Nakatani Eur. J. Biochem. 1970 16 180. 66 M. Fujioka and Y. Takata Biochim. Biophys. Acta 1979 570 210. 67 M. Fujioka and M. Tanaka Eur. J. Biochem. 1978 90 297. 68 M. Fujioka Y. Takata H. Ogawa and M. Okamoto J. Biol. Chem. 1980 255 937. 69 H. Ogawa and M. Fujioka J. Biol. Chem. 1980 255 7420. 70 M. Fujioka and Y. Takata Biochemistry 1981 20 468. 71 H. Ogawa M. Okamoto and M. Fujioka J. Biol. Chem. 1979 254 7030. 72 H. Ogawa T. Hase and M. Fujioka Biochim. Biophys. Acta 1980 623 225. 73 J. W. Xuan P. Fournier and C. Gaillardin Curr. Genet. 1988 14 15. 74 J. S. Trupin and H. P. Broquist J. Biol. Chem. 1965 240 2524. 75 Z. H. Ye and J. K. Bhattacharjee J.Bacteriol. 1988 170 5968. 76 R. C. Garrad and J. K. Bhattacharjee J. Bacteriol. 1992 174 7379. 77 R. Garrad T. M. Schmidt and J. K. Bhattacharjee Infect. Immun. 1994 62 5027. 78 M. G. Shepard Infect. Immun. 1985 50 541. 79 B. Ho and T. M. Zabriskie unpublished results. 80 V. W. Rodwell Methods. Enzymol. 1971 17B 174. 81 H. P. Broquist Annu. Rev. Nutr. 1991 11 435. 82 M. Kurtz and J. K. Bhattacharjee J. Gen. Microbiol. 1975. 83 J. J. Kinzel and J. K. Bhattacharjee J. Bacteriol. 1979 138 410. 84 J. J. Kinzel and J. K. Bhattacharjee J. Bacteriol. 1982 151 1073. 85 S. J. Mihalik and W. J. Rhead J. Comp. Physiol. B 1991 160 671. 86 Y. F. Chang Neurochem. Res. 1982 7 577. 87 Y. F. Chang J. Neurochem. 1978 30 347. 88 S. J. Mihalik M. McGuinness and P.A. Watkins J. Biol. Chem. 1991 266 4822. 89 B. Ho and T. M. Zabriskie Bioorg. Med. Chem. Lett. 1998 8 739. 90 T. M. Zabriskie J. Med. Chem. 1996 39 3046. 91 T. M. Zabriskie and X. Liang Bioorg. Med. Chem. Lett. 1997 7 457. 92 T. M. Zabriskie W. L. Kelly and X. Liang J. Am. Chem. Soc. 1997 119 6446. 93 T. M. Zabriskie unpublished data. 94 F. P. Guengerich and H. P. Broquist Biochemistry 1973 12 4270. 95 F. P. Guengerich and H. P. Broquist J. Bacteriol. 1976 126 338. 96 W.-U. Müller and E. Leistner Z. Naturforsch. 1975 30c 253. 97 N. Fangmeier and E. Leistner J. Biol. Chem. 1980 255 10205. 98 H. P. Broquist P. S. Mason B. Wickwire R. Homann M. J. Schneider and T. M. Harris in Swainsonine Production in the Mold Rhizoctonia leguminicola ed.A. A. Seawright M. P. Hegarty L. F. James and R. F. Keeler Yeerongpilly Australia 1985 pp. 301. 99 B. M. Wickwire C. M. Harris T. M. Harris and H. P. Broquist J. Biol. Chem. 1990 265 14742. 100 B. M. Wickwire C. Wagner and H. P. Broquist J. Biol. Chem. 1990 265 14748. 101 K. Higashino K. Tsukada and I. Lieberman Biochem. Biophys. Res. Commun. 1965 20 285. 102 J. Hutzler and J. Dancis Biochim. Biophys. Acta 1968 158 62. 103 J. Hutzler and J. Dancis Biochim. Biophys. Acta 1975 377 42. 104 P. J. Markovitz D. T. Chuang and R. P. Cox J. Biol. Chem. 1984 259 11643. 105 C. Noda and A. Ichihara Biochim. Biophys. Acta 1978 525 307. 106 J. Dancis and J. Hutzler Biochim. Biophys. Acta 1981 675 411. 107 J. Dancis and J. Hutzler Comp. Biochem.Physiol. 1982 73B 1011. 108 J. A. Grove T. J. Gilbertson R. H. Hammerstedt and L. M. Henderson Biochim. Biophys. Acta 1969 184 329. 109 G. Tang D. Miron J. X. Zhu-Shimoni and G. Galili Plant Cell 1997 9 1305. 110 H. R. Schuette and G. Seelig Z. Naturforsch. 1967 22b 824. 111 A. Cincerova and E. Cerna Z. Pflanzenphysiol. 1974 74 366. 112 T. P. Chirpich V. Zappia R. N. Costilow and H. A. Barker J. Biol. Chem. 1970 245 1778. 113 H. Takeda S. Yamamoto Y. Kojima and O. Hayaishi J. Biol. Chem. 1969 244 2935. 114 J. C. Fothergill and J. R. Guest J. Gen. Microbiol. 1977 99 139. 115 E. A. Boeker and E. H. Fisher Methods Enzymol. 1983 94 180. 116 K. Soda and M. Moriguchi Biochem. Biophys. Res. Commun. 1969 34 34. 117 H. Misono H. Uehigashi E.Morimoto and S. Nagasaki Agric. Biol. Chem. 1985 49 2253. 118 J. Romero J. F. Martin P. Liras A. L. Demain and N. Rius J. Ind. Microbiol. Biotechnol. 1997 18 241. 119 J. J. Coque P. Liras L. Laiz and J. F. Martin J. Bacteriol. 1991 173 6258. 120 T. Hammer R. Bode H. Schmidt and D. Birnbaum J. Basic Microbiol. 1991 31 43. 121 T. Hammer R. Bode and D. Birnbaum J. Gen. Microbiol. 1991 137 711. 122 H. Schmidt R. Bode and D. Birnbaum FEMS Microbiol. Lett. 1988 49 203. 123 J. J. Kinzel M. K. Winston and J. K. Bhattacharjee J. Bacteriol. 1983 155 417. 124 T. Hammer and R. Bode Zentralbl. Mikrobiol. 1992 147 65. 125 P. J. Large and A. Robertson FEMS Microbiol. Lett. 1991 66 209. 126 J. M. Beckerich M. Lambert and C. Gaillardin Curr. Genet. 1994 25 24. 127 H. Kusakabe K. Kodama A. Kuninaka H. Yoshino H. Misono and K. Soda J. Biol. Chem. 1980 255 976. 128 H. Kusakabe K. Kodama A. Kuninaka H. Yoshino and K. Soda Agric. Biol. Chem. 1980 44 387. 129 T. Hammer and R. Bode J. Basic Microbiol. 1992 32 21. 130 G. C. Friedrich and A. L. Demain Arch. Microbiol. 1978 119 43. 131 W. M. Jaklitsch W. Hampel M. Rohr C. P. Kubicek and G. Gamerith Can. J. Microbiol. 1986 32 473. 132 A. A. Brakhage and G. G. Turner FEMS Microbiol. 1992 12 123. 133 J. Casqueiro S. Gutierrez O. Banuelos M. J. Hijarrubia and J. F. Martin J. Bacteriol. 1999 181 1181. 134 Y. Lu R. L. Mach K. Affenzeller and C. P. Kubicek Can. J. Microbiol. 1992 38 758. 135 C. Esmahan E. Alvarez E. Montenegro and J. F. Martin Appl. Environ. Microbiol. 1994 60 1705. 136 J. M. Luengo J. R. Revilla M. J. López J. R. Villanueva and J. F. Martin J. Bacteriol. 1980 144 869. 137 C. Hönlinger and C. P. Kubicek Biochim. Biophys. Acta 1989 993 204. 138 K. Madduri C. Stuttard and L. C. Vining J. Bacteriol. 1989 171 299. 139 F.J. Pérez-Llarena A. Rodriguez-Garcia F. J. Enguita J. F. Martin and P. Liras J. Bacteriol. 1998 180 4753. 140 S. Schafer T. Paalme R. Vilu and G. Fuchs Eur. J. Biochem. 1989 186 695. 141 M. Nishiyama M. Kukimoto T. Beppu and S. Horinouchi Microbiology 1995 141 1211. 142 T. Kosuge and T. Hoshino FEMS Microbiol. Lett. 1998 169 361. 143 N. Kobashi M. Nishiyama and M. Tanokura J. Bacteriol. 1999 181 1713. 144 V. Bhattacherjee and J. K. Bhattacharjee 96th Annu. Meet. Am. Soc. Microbiol. 1996 Abstr F. Review a801345d 97 Nat. Prod. Rep. 2000 17 85–97