Conversion of Pipecolic Acid into Lysine in Penicillium chrysogenum Requires Pipecolate Oxidase and Saccharopine Reductase: Characterization of the lys7 Gene Encoding Saccharopine Reductase

Abstract

Pipecolic acid is a component of several secondary metabolites in plants and fungi. This compound is useful as a precursor of nonribosomal peptides with novel pharmacological activities. In Penicillium chrysogenum pipecolic acid is converted into lysine and complements the lysine requirement of three different lysine auxotrophs with mutations in the lys1, lys2, or lys3 genes allowing a slow growth of these auxotrophs. We have isolated two P. chrysogenum mutants, named 7.2 and 10.25, that are unable to convert pipecolic acid into lysine. These mutants lacked, respectively, the pipecolate oxidase that converts pipecolic acid into piperideine-6-carboxylic acid and the saccharopine reductase that catalyzes the transformation of piperideine-6-carboxylic acid into saccharopine. The 10.25 mutant was unable to grow in Czapek medium supplemented with α-aminoadipic acid. A DNA fragment complementing the 10.25 mutation has been cloned; sequence analysis of the cloned gene (named lys7) revealed that it encoded a protein with high similarity to the saccharopine reductase from Neurospora crassa, Magnaporthe grisea, Saccharomyces cerevisiae, and Schizosaccharomyces pombe. Complementation of the 10.25 mutant with the cloned gene restored saccharopine reductase activity, confirming that lys7 encodes a functional saccharopine reductase. Our data suggest that in P. chrysogenum the conversion of pipecolic acid into lysine proceeds through the transformation of pipecolic acid into piperideine-6-carboxylic acid, saccharopine, and lysine by the consecutive action of pipecolate oxidase, saccharopine reductase, and saccharopine dehydrogenase.

Many secondary metabolites contain l-lysine, d-lysine, or pipecolic acid moieties (29, 36). In filamentous fungi, the biosynthesis of pipecolic acid is related to lysine metabolism. In Metarhizium anisopliae and Rhizoctonia leguminicola pipecolic acid is an intermediate in the biosynthesis of alkaloid compounds, such as swansonine and slaframine. In these fungi pipecolic acid is formed by catabolism of lysine (40–42). However, using ¹⁴C- and ¹⁵N-labeled α-aminoadipic acid and [¹⁴C]lysine, Aspen and Meister (2) showed in Aspergillus nidulans that the carbon chain of α-aminoadipic rather than that of lysine was the major precursor of pipecolic acid and the nitrogen atom of α-aminoadipic acid becomes the nitrogen atom of pipecolic acid. Furthermore, in Neurospora crassa kinetic studies with radioactively labeled d-lysine showed that pipecolic acid was an intermediate in the conversion of d-lysine into l-lysine (17).

In Penicillium chrysogenum the biosynthetic pathway of lysine and penicillin have several steps in common (reviewed in references 1 and 12) (Fig. 1). α-Aminoadipic acid is the intermediate where both branching routes diverge (11). α-Aminoadipic acid has a key function in penicillin biosynthesis, since addition of exogenous α-aminoadipic acid (21) or genetic modifications that increase the internal α-aminoadipic acid pool (11, 22, 23) lead to a stimulation of the rate of penicillin biosynthesis. α-Aminoadipic acid is converted into piperideine-6-carboxylic acid (P6C) by α-aminoadipate reductase (9), and the piperideine-6-carboxylate may be reduced to pipecolic acid.

FIG. 1.

Biosynthesis of lysine and penicillin G in P. chrysogenum showing the interconversion of pipecolic acid into lysine. lys2 and lys5 encode two different proteins required for α-aminoadipic acid reductase activity (Lys5 is the cognate phosphopantetheinyl transferase that introduces a phospantetheine group in Lys2). The conversion steps blocked in the 7.2 and 10.25 mutants are indicated by arrows with hatch marks through the stems.

Pipecolic acid serves as a substrate of some nonribosomal peptide and polyketide synthetases, resulting in the formation of secondary metabolites with interesting novel pharmacological activities, i.e., the immunosuppressors rapamycin and immunomycin (31, 36). It seems, therefore, possible to use P. chrysogenum as a host for producing pipecolate-containing secondary metabolites. Because of the ability of P. chrysogenum to accumulate α-aminoadipic acid, it was of interest to study the interconversion of lysine and pipecolic acid in this industrially important filamentous fungus.

MATERIALS AND METHODS

Strains.

Four different P. chrysogenum strains were used in this work: P. chrysogenum Wisconsin 54-1255, a low-level penicillin-producing strain that has a single copy of the penicillin gene cluster (18); P. chrysogenum HS1⁻, a lysine auxotroph disrupted in the lys1 gene (5a); P. chrysogenum TDX195, a lysine auxotroph with the lys2 gene disrupted (11); and P. chrysogenum L2, a lysine auxotroph blocked in the first half of the α-aminoadipate pathway (16, 27a). Escherichia coli DH5α was used as host for DNA plasmid manipulation. E. coli DH10B was used to recover plasmids from the genomic DNA of P. chrysogenum.

All P. chrysogenum strains were grown in Power medium (10) supplemented with the appropriate filter-sterilized amino acids at a final concentration of 1.75 mM. Spores were collected from Power medium plates after incubation during 5 days at 28°C. Growth tests were performed on Czapek minimal medium with the appropriate supplements (10).

Plasmid vector and plasmid genomic library.

pAMPF9L (19) is a shuttle plasmid between E. coli and P. chrysogenum; it contains the AMA1 region, which confers autonomous replication in P. chrysogenum, and pBluescript for replication in E. coli. pLARA is an autonomous replication derivative of the pAMPF9L bearing the lys1 gene under the control of its own promoter region (3, 4). The genomic library was constructed by ligating partially digested Sau3AI genomic DNA obtained from P. chrysogenum AS-P-78 at the BglII site of the pAMPF9L plasmid.

Isolation of P. chrysogenum genomic DNA and plasmid rescue in E. coli.

The genomic DNA of P. chrysogenum was obtained as described previously (10). For plasmid rescue, genomic DNA (1 μg of each transformant) was used to electroporate electrocompetent E. coli DH10B cells (39) in Bio-Rad Gene Pulser cuvettes (0.2-cm gap). Electroporation conditions were 100 Ω, 2,500 V, and 25 μF.

Mutant isolation.

P. chrysogenum HS1⁻ (a strain defective in the homocitrate synthase) spores (10⁶/ml) were incubated in 0.1 M Tris-maleate buffer, pH 9.0, for 12 to 15 h. To induce synchronous spore germination, which correlates with DNA synthesis (28), 1.75 mM lysine, glucose (15 g/liter), and (NH₄)₂SO₄ (1 g/liter) were added to the spore suspension. The mutation was carried out during 90 min with 0.5 mM nitrosoguanidine in the same Tris-maleate buffer, pH 9.0, to obtain a mortality rate of 90%. Serial dilutions of mutated spores were plated in Power medium plus lysine (1.75 mM). Lysine auxotrophs unable to grow on pipecolic acid (lys⁻ pip⁻) were screened by replicating each colony in Czapek minimal medium with pipecolic acid (100 mM) and Czapek minimal medium without pipecolic acid. Reversion rates of the mutations were determined for all the mutants by plating spore suspensions (with known viable spore concentrations) on Czapek minimal medium or Czapek medium plus pipecolic acid and counting the revertant colonies.

Enzyme activities.

Cultures of the different P. chrysogenum mutants were grown in defined production (DP) medium (11). Cell extracts were prepared by grinding frozen mycelium samples in a mortar. Disrupted cells (0.5 to 1 g) were mixed with 1 ml of a solution containing 20 mM Tris-HCl (pH 8.0), 1 mM phenylmethylsulfonyl fluoride, 10 mM β-mercaptoethanol, and 5 mM EDTA or 50 mM Tris-HCl–5 mM EDTA (pH 7.5) and centrifuged for 15 min at 4°C and 10,000 × g and the clarified supernatant was collected. Protein concentration was determined by the Bradford assay (7).

Saccharopine dehydrogenase (EC 1.6.1.7) activity was determined by measuring the consumption of NADH (0.15 mM) for 3 min at 37°C in the presence of 50 mM lysine, 10 mM α-ketoglutarate, and 50 μg of total protein in 250 mM Tris-HCl (pH 7.5). Controls were performed similarly but without lysine in the reaction mixture. Enzymatic activity was expressed in units (1 U = 1 μmol of NADH consumed/min · ml; molar extinction [ɛ] = 6,220 liters/mol · cm at 340 nm).

Saccharopine reductase (EC 1.5.1.10) activity was determined by measuring P6C formation in the presence of 4 mM saccharopine, 0.5 mM NADP, and 50 μg of total protein in 100 mM Tris-HCl (pH 9.0). Control reactions were made without NADP in the reaction mixture. Enzyme assays were performed for 1 h at 30°C and terminated by addition of 200 μl of 5% tricarboxylic acid (in ethanol). P6C was quantified by derivatization of 500 μl of reaction mixture with 750 μl of ortho-amino-benzaldehyde (OAB) (1 mg/ml in 2% ethanol) and incubated for 30 min at 37°C, and the absorbance at 465 nm was determined. The activity was expressed as units (1 U = 1 μmol of P6C formed/min · ml; molar extinction of P6C-OAB [ɛ] = 2,800 liters/mol · cm).

Pipecolic acid oxidase activity was quantified as described by Mihalik et al. (30). The reaction mixture contained 10 mM l-pipecolic acid, 0.64 mM o-dianisidine, 20 U of horseradish peroxidase, and 100 μg of total protein in buffer A (40 mM Tris-HCl, pH 8.5; 80 mM KCl; 0.8 mM EDTA). Blank controls lacked l-pipecolic acid. The assay was performed for 15 min at 37°C, and H₂O₂ formation was determined by measuring the oxidized o-dianisidine at 460 nm. The pipecolate oxidase activity was expressed as units (1 U =1 μmol of o-dianisidine oxidized/min · ml; molar extinction of the o-dianisidine = 11,300 liter/mol · cm).

Transformation of P. chrysogenum and DNA sequencing.

The transformation of P. chrysogenum protoplasts was performed as described previously (8). Transformants were selected in Czapek medium with 0.7 M KCl. DNA fragments were sequenced by the dideoxynucleotide chain termination method (38) using Sequenase (U.S. Biochemicals) in an ALF DNA sequencer (Pharmacia). All other nucleic acid manipulations were performed by using standard methods (37).

Total RNA isolation and intron analysis.

Total RNA was isolated with the RNeasy kit (Qiagen, Chatsworth, Calif.) from P. chrysogenum Wis 54-1255 grown in DP medium (11). To elucidate the presence of putative introns in the DNA sequence, the region containing the expected intron splicing sites was amplified from RNA with a SuperScript one-step RT-PCR system (Gibco), using as primers the following oligonucleotides: Ia (5′-TGCGATGTTCTCCAGTTG-3′) and Ib (5′-CGCATATACATCCCATTG-3′).

Nucleotide sequence accession number.

The lys7 nucleotide sequence has been deposited in the EMBL database under the accession number AJ319030.

RESULTS

Pipecolic acid is converted into lysine in P. chrysogenum.

As a first approach to study the relationship between pipecolic acid and lysine metabolism in P. chrysogenum, we tested if pipecolic acid could be converted into lysine. Three P. chrysogenum lysine auxotrophic mutants blocked (i) in the synthesis of α-aminoadipic, including P. chrysogenum HS1⁻ (with a targeted inactivation of the homocitrate synthase) and P. chrysogenum L2 (defective in the homoaconitase) or (ii) in the conversion of α-aminoadipic acid to α-aminoadipic semialdehyde, e.g., P. chrysogenum TDX195 (disrupted in the lys2 gene encoding α-aminoadipate reductase), were tested for their ability to grow in Czapek minimal medium supplemented with pipecolic acid. Results showed (Fig. 2) that all three different lysine auxotrophs were able to grow either with lysine or with pipecolic acid although growth on pipecolic acid was very slow, indicating that in P. chrysogenum, pipecolic acid is converted into lysine.

FIG. 2.

Growth of the P. chrysogenum Wis 54-1255 parental strain and the lysine auxotrophs derived from it on Czapek minimal medium MM, Czapek medium with pipecolic acid, and Czapek medium with l-lysine. Note that growth on pipecolic acid of the three auxotrophs is very slow compared with growth on lysine. Abbreviations: Wis, P. chrysogenum Wis 54-1255; TDX, P. chrysogenum TDX195; HS−, P. chrysogenum HS1⁻; L2, P. chrysogenum L2. Strain TDX has been disrupted in the lys2 gene, strain L2 is a lysine auxotroph obtained by ethylmethane sulfonate mutation defective in the homoaconitase gene (27a), and strain HS1⁻ has been disrupted in the lys1 gene (5a).

Isolation of lys⁻ pip⁻ mutants.

In order to study how pipecolic acid is converted into lysine, mutants impaired in the conversion of pipecolic acid into lysine were isolated from P. chrysogenum HS1⁻. These mutants should be altered in any of the enzymes leading to lysine formation from pipecolic acid.

We isolated three mutants named P. chrysogenum 8.46, 7.2, and 10.25 that did not grow in Czapek medium with pipecolic acid but were able to grow in Czapek medium plus lysine (the so-called Pip⁻ phenotype) after screening 5,000 colonies following mutation with nitrosoguanidine. Analysis of the reversion rate (Table 1) showed that mutants 7.2 and 10.25 were stable whereas mutant 8.46 was highly unstable. The first two mutants were used for further characterization studies.

TABLE 1.

Growth on pipecolic acid and α-aminoadipic acid of different mutants isolated from HSl⁻ showing their reversion rates

Mutant	Genetic defect	Growth on:				Reversion rate
		Czapek MMc	Czapek medium + lysine	Czapek medium + pipecolate	Czapek medium + α-aminoadipate
P. chrysogenum Wis 54-1255	None	Yes	Yes	Yes	Yes	NDa
P. chrysogenum TDX195	Lys2−	No	Yes	Yes	No	<4 × 10−9
P. chrysogenum HS1−	Lys1−	No	Yes	Yes	Yes	<3.3 × 10−8
P. chrysogenum 8.46	ND	No	Yes	No	ND	3.3 × 10−6
P. chrysogenum 7.2	p.o.−b	No	Yes	No	Yes	<6.2 × 10−9
P. chrysogenum 10.25	Lys7−	No	Yes	No	No	<5.8 × 10−9

^aND, not determined. 

^bp.o.⁻, pipecolate oxidase deficient. 

^cMM, minimal medium. 

Mutant P. chrysogenum 10.25 is altered in the second half of the lysine pathway.

To study whether pipecolic acid is converted directly into lysine or whether it is transformed in some intermediate of the lysine biosynthesis pathway, the isolated mutants were plated in Czapek medium plus α-aminoadipic acid to test if they have a functional conversion of α-aminoadipic acid into lysine. Results showed (Table 1) that mutant 10.25 was unable to grow in Czapek medium plus α-aminoadipic acid, indicating that the mutation of this strain alters one of the enzymes of the second half of the lysine biosynthetic pathway (i.e., the α-aminoadipate reductase, saccharopine reductase, or saccharopine dehydrogenase). The conversion of pipecolic acid into lysine proceeds, therefore, through the action of some of these enzymes. On the other hand mutant 7.2 was able to grow on α-aminoadipic acid (see below).

Cloning of the lys7 gene by complementation of the 10.25 mutant.

In order to further characterize the conversion of pipecolic acid into lysine, the gene complementing the mutation of the 10.25 strain was cloned. This mutant derives from the P. chrysogenum HS1⁻ strain that has the lys1 gene disrupted. For this reason, to clone the gene responsible for the pip⁻ phenotype a cotransformation was performed with pLARA (autonomous replicating plasmid bearing the lys1 gene) (4) and with a genomic library constructed in the pAMPF9L (a vector bearing the AMA1 region which confers autonomous replication in P. chrysogenum) (19). After cotransformation of the 10.25 mutant, two transformants (named T1 and T2) were selected directly in minimal Czapek medium. These transformants were prototrophs, indicating that the lys1 mutation was complemented; in addition they were able to convert pipecolic acid into lysine.

Rescue of the plasmids complementing the 10.25 mutant.

The plasmids present in the T1 and T2 transformants were rescued in E. coli. Analysis of the plasmid population present in these E. coli transformants showed that some of them (Fig. 3A, lanes 2, 3, 4, 5, 7, 11, and 15) were identical to the pLARA plasmid (Fig. 3, lane 1) whereas others contained DNA inserts that might correspond to the lys7 gene or to DNA fragments that have recombined. To identify which of these plasmids were recombinants bearing lys1 or plasmids with inserts from the genomic library, Southern blotting was performed with the lys1 gene as probe. Results (Fig. 3B) showed that most plasmids present in the blot, except those in lanes 18, 19, and 20, gave hybridization with the lys1 gene. The nonhybridizing plasmids should contain the DNA fragment complementing the pip⁻ mutation. One of these plasmids (Fig. 3, lanes 18 to 20) was named p10T1.

FIG. 3.

Plasmids present in the T1 and T2 transformants of P. chrysogenum 10.25 after they were rescued in E. coli DH10B. Plasmids were digested with HindIII. (A) Agarose gel. (B) Hybridization with an AccI probe (650 bp) internal to the lys1 gene. Lanes 1 to 13, plasmids isolated from transformant T2; lane 14, λHindIII (size marker); lanes 15 to 20, plasmids isolated from transformant T1. Note that plasmids in lanes 18, 19, and 20 do not hybridize with the lys1 probe. These plasmids contain inserts that complemented the 10.25 mutation.

Cotransformation of the 10.25 mutant with pLARA and p10T1 indicates that p10T1 contains the pip⁻-complementing gene.

To test whether plasmid p10T1 was responsible for the complementation of the pip⁻ mutation, a cotransformation of the 10.25 mutant was performed with the pLARA and p10T1 plasmids. A large number of positive transformants were obtained. Rescue of the plasmids from three transformants resulted in the recovery of plasmids pLARA and p10T1 (Fig. 4), showing that both plasmids were necessary to complement the double mutation (lys1⁻ and pip⁻) in the 10.25 strain.

FIG. 4.

Plasmids of strain 10.25 complemented by cotransformation with pLARA and p10T1 after rescuing in E. coli DH10B. Plasmids recovered from three 10.25 cotransformants are shown. Lanes 1 to 5, transformant 10.25-1; lanes 6 to 10, transformant 10.25-2; lanes 11 to 15, transformant 10.25-3; lane 16, size markers (HindIII-digested lambda phage); lane 17, pLARA control plasmid; lane 18, p10T1 plasmid. Note that both pLARA and p10T1 (arrows) are present in complemented 10.25 transformants (lysine prototrophs).

ORF1 of the cloned DNA fragment encodes a protein with high homology to saccharopine reductases.

A 2,987-bp region of the insert of the p10T1 plasmid was sequenced on both strands. An open reading frame (ORF) of 1,953 bp (ORF1) was found. ORF1 contains 10 putative introns of 104, 62, 53, 50, 52, 59, 52, 55, 53, and 63 nucleotides. RT-PCR using RNA from 24-h cultures of P. chrysogenum Wis 54-1255 and oligonucleotides Ia and Ib as primers confirmed the presence of the 10 introns. A DNA band of the expected size after the splicing of the 10 introns (1,350 bp) was obtained. The amplified DNA band was sequenced on both strands, confirming that the introns had been removed at the splicing sites corresponding to nucleotides 53 to 156, 211 to 272, 455 to 507, 629 to 678, 752 to 803, 847 to 905, 920 to 971, 1019 to 1073, 1178 to 1230, and 1795 to 1857 (numbered from the ATG translation initiation codon).

ORF1 encoded a protein of 449 amino acids starting with a methionine encoded by an in-frame ATG with a deduced molecular mass of 48.8 kDa. No other ATG that might correspond to putative translation initiation codons was found in that region. The encoded protein showed a high similarity to other saccharopine reductases (Fig. 5), particularly those of N. crassa (70.4% identical amino acids), Magnaporthe grisea (66.6% identity) (24), Saccharomyces cerevisiae (63.1% identity) (6), and Schizosaccharomyces pombe (60% identity). The amino-terminal region of the encoded protein was very similar to that of the corresponding proteins of N. crassa, M. grisea, and S. cerevisiae. These results suggest that the protein encoded by ORF1 corresponds to the P. chrysogenum saccharopine reductase, and the gene has been named lys7, according to the gene designation in Fig. 1.

FIG. 5.

Alignment of the amino acid sequence of the protein encoded by lys7 (cloned by complementation of the 10.25 mutation) with proteins in the EBI databases. Note that the protein encoded by lys7 is very similar to the saccharopine reductases of N. crassa, M. grisea, S. cerevisiae, and S. pombe. Identical amino acids are shaded.

Pipecolic acid is converted into lysine by the action of the saccharopine reductase and saccharopine dehydrogenase in the lysine biosynthetic pathway.

To further characterize the pip⁻ mutants we measured the saccharopine reductase and saccharopine dehydrogenase activities, which are involved in the conversion of α-aminoadipic acid-δ-semialdehyde into saccharopine and then into lysine, respectively. Results showed (Table 2) that saccharopine dehydrogenase was present in all strains studied. Mutant 10.25 lacked saccharopine reductase, in agreement with the DNA cloning experiments that indicated that the lys7 gene encodes a saccharopine reductase. These results confirmed that the 10.25 mutant is blocked in saccharopine reductase and suggested that pipecolic acid enters the lysine pathway at the level of the α-aminoadipic acid-δ-semialdehyde (Fig. 1).

TABLE 2.

Enzymatic activities in P. chrysogenum mutants 10.25 and 7.2

Mutant	Activity (mU/mg of protein)
	Saccharopine dehydrogenase	Saccharopine reductase	Pipecolate oxidase
P. chrysogenum Wis 54-1255	381	44	2.1
P. chrysogenum HS1−	402	46.1	1.9
P. chrysogenum 10.25	571	0	2.1
P. chrysogenum 7.2	450	35.4	0

Pipecolic acid is incorporated in the lysine pathway through its conversion in P6C by the pipecolate oxidase.

The pipecolate oxidase enzyme converts pipecolic acid into P6C which is the cyclic form of the α-aminoadipic acid-δ-semialdehyde.

Analysis of the pipecolate oxidase in P. chrysogenum showed that this activity was present in the parental P. chrysogenum Wis 54-1255 as well as in the 10.25 mutant (Fig. 6). Moreover, when P. chrysogenum Wis 54-1255 was grown with pipecolic acid as the sole nitrogen source, the pipecolate oxidase activity was induced in this strain (Fig. 6), suggesting that pipecolic acid is catabolized through its conversion into P6C by the action of pipecolate oxidase. The pipecolate oxidase induction by pipecolic acid was prevented when ammonium (37.8 mM) was added together with pipecolic acid, suggesting that ammonium prevents the uptake of pipecolic acid, as occurs with other amino acids in P. chrysogenum (5). An alternative explanation is that ammonium repressed the pipecolate oxidase, preventing pipecolate utilization. Interestingly, analysis of the pipecolate oxidase activity in different strains (Table 2) revealed that this activity was missing in the P. chrysogenum 7.2 mutant, which lacks the ability to complement the lysine auxotrophy with pipecolic acid. These results suggest that pipecolic acid is converted into P6C by the action of pipecolate oxidase, and then P6C (or its linear form, α-aminoadipic acid-δ-semialdehyde) is converted into saccharopine and lysine by the consecutive action of saccharopine dehydrogenase and saccharopine reductase.

FIG. 6.

Pipecolate oxidase activity in P. chrysogenum Wis 54-1255 in DP medium with pipecolic acid as sole nitrogen source (○), with pipecolic acid and ammonium (▴), and with only ammonium (●). Note that pipecolate oxidase activity is induced in medium with pipecolic acid as the sole nitrogen source and is repressed by ammonium.

DISCUSSION

Pipecolic acid is formed by the catabolism of lysine in humans. Although this pathway is considered to be secondary in most tissues, pipecolic acid is the main lysine catabolism product in the brain (34, 35), and its accumulation was one of the first biochemical abnormalities detected in the Zellweger syndrome. In the filamentous fungi M. anisopliae and R. leguminicola, pipecolic acid, an intermediate in the biosynthesis of alkaloid compounds, comes directly from lysine catabolism, but through a different pathway than in mammals, involving the intermediates saccharopine and P6C (40–42).

As shown in this work, P. chrysogenum can use pipecolic acid through the action of pipecolate oxidase, thus converting pipecolic acid into P6C. Pipecolate oxidase has been shown to occur in the brain and liver of mammals, including humans (15, 30, 33, 35), and in Rhodotorula glutinis (26, 27), Pseudomonas putida (13), and Pseudomonas sp. (32).

In P. chrysogenum pipecolic acid can be converted into lysine, through its transformation into P6C, saccharopine, and finally lysine, by the consecutive action of pipecolate oxidase, saccharopine reductase, and saccharopine dehydrogenase (Fig. 1). It is unclear if the conversion of pipecolic acid into lysine is reversible in P. chrysogenum and whether pipecolic acid is synthesized from lysine catabolism (14, 20). Indeed, the reverse conversion of lysine into pipecolic acid through saccharopine and P6C corresponds exactly to the proposed pipecolic acid biosynthetic pathway in R. leguminicola (41, 42).

In this work we have cloned the lys7 gene encoding the saccharopine reductase that is involved in the conversion of pipecolic acid to lysine. Saccharopine reductase (EC 1.5.1.10), also referred to as saccharopine dehydrogenase (glutamate forming) or aminoadipic semialdehyde-glutamate reductase, catalyzes the penultimate step in the lysine biosynthetic pathway by condensing the α-aminoadipic semialdehyde with the amino donor glutamic acid by a Schiff base mechanism in which the amino group of glutamic acid reacts with the aldehyde group of the α-aminoadipic semialdehyde, resulting in the formation of the pseudodipeptide saccharopine. Later, saccharopine releases α-ketoglutarate, yielding lysine in the last reaction of the lysine pathway catalyzed by saccharopine dehydrogenase.

The amino acid sequence of the protein encoded by lys7 of P. chrysogenum is very similar to the sequences of the saccharopine reductase from N. crassa, M. grisea (24), S. cerevisiae (6), and S. pombe. The C-terminal part of the protein is similar (about 38% identity) to the C-terminal region (residues 400 to 800) of the lysine-α-ketoglutarate reductase/lysine dehydrogenase from several eukaryotic organisms such as Homo sapiens, Mus musculus, and Arabidopsis thaliana, an enzyme with a similar molecular mechanism to that of saccharopine reductase.

The saccharopine reductase from M. grisea has been purified and crystallized (24). Saccharopine reductase is a homodimer, and each subunit consists of three domains that are also present in the P. chrysogenum Lys7 protein. Domain I contains a variant of the Rossmann fold that binds NADPH. Domain II folds into a mixed seven-stranded beta-sheet flanked by alpha-helices and is involved in substrate binding and dimer formation. Domain III is all helical. The structure analysis of the ternary complex reveals a large movement of domain III upon ligand binding (25).