Purification and Characterization of Eucaryotic Alanine Racemase Acting as Key Enzyme in Cyclosporin Biosynthesis*

A specific alanine racemase, which is a key enzyme in the biosynthesis of the undecapeptide cyclosporin A, was purified to electrophoretic homogeneity from the fungus lblypocladium niveum. This is the first enzyme of this kind isolated from a eucaryotic organism. The en- zyme catalyzes the reversible racemization of alanine and requires pyridoxal phosphate as the exclusive co- factor. K,,, values for L- and D-alanine were found to be 38 and 2 IILM, respectively. Maximal reaction velocity was observed at 42 "C and pH 8.8 for the L to D direction. Molecular mass determinations the en- by SDS-polyacrylamide gel electrophoresis gave a value of 37 kDa, whereas gel filtration calibration stud- ies yielded a value between 120 and 150 m a , indicating an oligomeric native structure.

represses the cellular immune response to foreign antigens (2) and is used worldwide to prevent allograft rejection.
Cyclosporin A is produced in a non-ribosomal manner by the fungus Tolypocladium niveum. The biosynthesis involves at least 40 different reaction steps (5, 6) and is catalyzed by the cyclosporin synthetase, a multifunctional enzyme with a molecular mass of about 1,500 kDa (3, 4). This enzyme activates the constituent amino acids of cyclosporin A to amino acyl adenylates and binds them covalently via thioester linkages. At this stage, seven of the substrate amino acids become N-methylated by a methyltransferase function, an integral activity of the cyclosporin synthetase, using AdoMet' as a methyl group donor (3). Finally peptide bond formation is facilitated by a prosthetic 4'-phosphopantetheine group, and the cyclosporin molecule is released from the enzyme (6). In addition to Lvaline, L-leucine, and glycine, 2-aminobutyric acid, ( 4 R )-4-[(E)-2-butenyl]-4-methyl-~-threonine, and D-alanine are precursors for this process. Whereas first insights into the biosynthetic pathways of ( 4 R )-4-[(E)-2-butenyl3-4-methyl-~-threonine have recently been achieved (71, the origin of D-alanine in T niveum is not understood so far. This paper describes the identification and purification of an alanine racemase from T niveum. We discuss the function of the racemase within the biosynthesis of * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.  Fig. 21, which is a further product of cyclosporin synthetase, and we compare the properties of the enzyme with its procaryotic counterparts. EXPERIMENTAL PROCEDURES Materials4hemicals were of the highest purity commercially available. Radiolabeled compounds were purchased from Amersham Corp. Amino acids were obtained from Fluka Chemie AG. o-Amino acid oxidase, L-alanine dehydrogenase, and lactate dehydrogenase are products of Boehringer Mannheim. EHC peptone was procured from Amber, and casein peptone was obtained from Difco. Cyclosporin synthetase and by Zocher et al. (3) cyclosporin A were isolated and purified using the procedures described Growth of Organisms--I: niveum strain 7939145 was donated by Sandoz Ltd. and maintained on agar slants (1% malt extract, 0.5% yeast extract, 1.5% agar). For precultures we used chemically defined medium (MCP 75) as described by Billich and Zocher (6). MCP 75 contains the following ingredients per liter of distilled water: 75 g of maltose, 25 g of casein peptone, 1 g of KH,PO, and 2.5 g of KCl; pH of the medium was 5.5. After 48 h 10 ml of the preculture were used as inoculum for 10 flasks with 200 ml of MCPA medium (same composition as MCP 75, but containing EHC peptone instead of casein peptone). Cultures were maintained on a rotary shaker (180 rpm) at 26 "C and harvested after 145 h by suction filtration. The mycelial cake was washed with distilled water, shock-frozen at -80 "C, and lyophilized.
Purification ofAlanim Racemase-All operations were carried out at 4 "C. Buffer A used throughout the purification procedure was 50 m Tris, pH 8.5, containing 4 m M EDTA, 20 m dithiothreitol, and 50 w pyridoxal phosphate unless otherwise stated. The lyophilized mycelium was suspended in buffer A with 300 mM KC1 and 200 m Tris. After disruption by French press (16,000 p.s.i.), cell debris was removed by centrifugation at 20,000 x g twice for 20 min. The resultant extract was dialyzed against buffer A and loaded onto a 60-ml QAE-Sepharose column (Pharmacia LKB Biotechnology Inc.) with a flow rate of 2 mumin. The column was washed with buffer A and eluted with the same buffer containing 150 m NaCl. Active fractions were pooled and applied to hydrophobe interaction chromatography on a 30-ml phenyl-Sepharose column (Pharmacia). After washing with buffer A the protein was eluted by a linear gradient of 0-7% Triton X at 1 ml/min. Active fractions were concentrated by ultrafiltration (Centricon 30 from Amicon, Inc.) to give a final volume of 5 ml and subjected to fast performance liquid chromatography on a Superdex 200 16/60 column (Pharmacia), using buffer B (50 m HEPES, pH 8.5,4 m EDTA, 20 m dithiothreitol, and 50 p pyridoxal phosphate). The flow rate was 0.5 d m i n with a fraction size of 2 ml. After addition of glycerol (10% final concentration), protein was loaded on a 2-ml cellulose phosphate column (E. Merck, Darmstadt, Germany) and washed extensively with buffer C (buffer B, pH 7.5, containing 10% glycerol). The enzyme was eluted by a linear NaCl gradient from 0 to 220 m in 5 min and from 220 to 400 m in 20 min (flow rate, 1 d m i n ; fraction size, 2 ml). Active fractions were diluted with 1 volume of distilled water and bound to a MonoQ anion exchange column (Pharmacia) on fast performance liquid chromatography. The gradient used for elution was 0-70 m NaCl (buffer B) in 10 min and 70-120 m in 60 min at 1 mumin. Active fractions were collected and concentrated by lyophilization.
Enzyme Assays-The standard alanine racemase assay mixture con- the addition of substrate at 42 "C. One unit was defined as the amount of enzyme catalyzing the formation of 1 m o l of epimerized product! min. Specific activity was expressed as units of enzyme activitylmg of protein.
Racemase activity was measured spectrophotometrically using either o-amino acid oxidase coupled to lactate dehydrogenase or L-alanine dehydrogenase. In the L to D direction we used 0.3 units of wamino acid oxidase, 25 units of lactate dehydrogenase, and 0.2 m~ NADH in a final volume of 350 pl. The reaction mixture for the reverse direction contained 0.1 unit of L-alanine dehydrogenase and 10 m~ NAD+ in the same volume. The change of absorbance at 340 nm due to the formation or consumption of NADH was followed by a Uvicon 930 recording spectrophotometer. A molar extinction coefficient of 6.22 cm2/pmol of NADH was used for calculation of enzyme activity.
To investigate substrate specificities, formation of D-amino acids was detected by the 2,4-dinitrophenylhydrazine assay described elsewhere (13). In this method a-keto acids, derived from D-amino acids by D-amino acid oxidase, are measured as hydrazones.
To demonstrate the formation of D-amino acid, the alanine racemase assay was carried out with %-labeled precursors (120-180 mCi/mmol). After incubation, the proteins were precipitated by adding acetone. The supernatant was evaporated and redissolved in 0.1 ml of HCVmethanol (1:l). For separation of the isomers, enantioselective TLC was used.
At the nanomolar range, formation of D-alanine was detected using the enzymatic biosynthesis of L-and D-DKF' as an assay. For this method, 100 pl of the sample were brought to pH 8 and 10% glycerol and then used instead of alanine in the reaction mixture for DKF' synthesis as described below.
Enzymatic Formation of Diketopiperazines-For synthesis of cyclo(alany1-N-methylleucine), 50 pl of enriched cyclosporin synthetase (in 100 m~ Tris-HC1, pH 8, 4 m~ EDTA, 4 m~ dithioerythritol, 10% glycerol, isolated as described previously (6)) were incubated in the presence of 1 m~ constituent amino acids (D-or L-alanine and L-leucine), 20 m~ ATP, 10 m~ MgC1, and 1 m~ AdoMet for 60 min at 25 "C. For detection, 0.5 pCi of [14Clleucine (312 pCi/mmol) was used. 1 ml of water was added to stop the reaction, and the mixture was extracted with 1 ml of ethyl acetate. The organic layer was evaporated, and the residue was dissolved in 10 p1 of ethyl acetate for TLC. This is a variation of the method for in vitro synthesis of cyclosporins as established by Billich and Zocher (6).
Gel Electrophoresis-SDS-polyacrylamide gel electrophoresis was done as described by Laemmli (8). Gels contained 10% acrylamide and 0.2% bisacrylamide and were silver stained using the procedure described by Blum et al. (9). The relative mass of enzyme bands was determinated from their mobility related to those of standard proteins.
Isoelectric focusing was carried out by using Servalyt Precotes (Serva) according to the instructions of the manufacturer. The isoelectric point of the racemase was determined by comparison with marker proteins.
Protein Determination-Protein concentrations were determined by using a modified Bradford procedure (10) with bovine serum albumin as a standard.

RESULTS
Purification of Alanine Racemase-!I! niveum was harvested at various times, and the racemase activities were measured in crude extracts. Racemase activity reached its maximum after 142 h and decreased slowly after 150 h. For purification of the enzyme, 145-h cultures were used. Among several disruption techniques, French press gave the best results. Ammonium sulfate precipitation, usually a good method for protein fractionation, resulted in loss of activity. A similar behavior was reported for the dad B alanine racemase from Salmonella typhimurium (11). Ion exchange chromatography on QAE-Sepharose was found to be much more effective, giving about a 10-fold purification. The total procedure is summarized in Table I. QAE-Sepharose was followed by hydrophobic interaction chromatography on phenyl-Sepharose, Superdex 200 gel filtration, and cation exchange chromatography on cellulose phosphate. Anion exchange chromatography on MonoQ was used as the last step in the purification procedure and yielded a single band in SDS-PAGE (Fig. 3, lane A). Starting with 25 g of lyophilized mycelium, the typical yield was 1 pg of homogenous enzyme with a specific activity of 47.7 unitdmg; this represented an over 6500-fold purification related to the specific activity of the crude extract.

Enzymatic Properties
Molecular Mass Determination-The molecular mass of the denatured enzyme was estimated by SDS-PAGE as shown in Fig. 3. The protein migrated as a single band with a molecular mass of about 37 kDa compared with standard proteins (myosin, 205 kDa; P-galactosidase, 116 kDa; phosphorylase, 97 kDa; egg albumen, 45 kDa; carbonic anhydrase, 29 kDa).
Molecular mass determination of the native enzyme was carried out by means of size exclusion chromatography using a Superdex 200 16/60 column (Pharmacia), which was previously calibrated with standard proteins (thyroglobulin, 660 kDa; ferritin, 440 kDa; catalase, 240 kDa; aldolase, 155 kDa; bovine serum albumin, 66 kDa). At a flow rate of 0.5 mVmin in buffer A, activity eluted a t 67 ml. Assuming a globular structure, this indicated a molecular mass of about 150 kDa (Fig. 4). A similar result was obtained by gel filtration on a high pressure liquid chromatography column (TSK 3000-SWG), where the maximum activity of the enzyme eluted according to a molecular mass of about 120 kDa (data not shown).
The different M, values from SDS-PAGE and gel filtration suggest a trimeric or tetrameric structure of the native racemase, composed of identical subunits.
Cofactor Requirement-Pyridoxal phosphate could be identified as the exclusive cofactor of the racemase. The Kd value was estimated to be 12 p~. When incubated with 2 m M NH,OH a t pH 7.5, all activity was lost but could be partly restored by the addition of 30 1.1~ pyridoxal phosphate. After Superdex gel filtration with pyridoxal phosphate-free buffer A, activity of racemase decreased by as much as 10%. Similar results were obtained after dialyzing the enzyme sample overnight. Activity was restored to 100% by addition of 30 1.1~ pyridoxal phosphate.
Kinetic Parameters-A wide range of amino acids, including 2-aminobutyric acid, was used to investigate the substrate specificity of the racemase under saturating conditions. The enzyme exhibits high specificity with respect to L-and D-alanine. Highest relative activities toward alternative substrates measured a t pH 8.0 were found with L-serine (23%), 2-~-aminobutyric acid (E%), and L-leucine (13%) (L-alanine = 100%). standard racemase assay was carried out at different temperatures. A broad optimum was found with a maximum around 42 "C. The enzyme showed maximal activity at pH 8.8 for the L to D direction and at pH 9.5 for the reversed reaction. Further Properties of Alanine Racemase-Enzyme activity was increased in the presence of sulfhydryl-containing compounds like glutathione, mercaptoethanol, or dithiothreitol. The latter compound provides the best conditions for enzyme activity a t a concentration of 20 m~. In contrast to these results, 2 m M L-as well as D-cysteine caused a complete inhibition. The enzyme showed no requirement for ATP, flavins, or metal ions.
Analytical isoelectrical focusing on a nondenaturing polyacrylamide gel yielded one single protein band with a PI of about 7.5. To prove the identity of this protein, the band was extracted from the gel and was found to have alanine racemase activity (not shown).
Formation of Diketopiperazines by a Coupled Reaction of Cyclosporin Synthetase and Alanine Racemase-It has been demonstrated in our laboratory that cyclosporin synthetase is able to synthesize cyclo(D-alanyl-N-methylleucine) (D-DKP) (3).
Here we demonstrate the formation of cyclo(L-alanyl-N-methylleucine) (L-DKP) using L-instead of D-alanine as a precursor. Offering both L-and D-alanine as substrates, we obtained a product mixture of D-DKP and L-DKP. In contrast to D-and L-alanine, these are stereoisomers and can easily be separated by TLC as described under "Experimental Procedures." Interestingly D-DKP formation is very sensitive. Using 14C-labeled leucine, we could detect the formation of D-DKP up from 10 pmol of D-alanine. Furthermore we could demonstrate the formation of D-DKP by the coupled reaction from alanine race- mase and cyclosporin synthetase after incubation with L-alanine, AdoMet, and 14C-labeled L-leucine. In the control reaction without alanine racemase only L-DKP was formed (not shown).

DISCUSSION
Alanine racemases (EC 5.1.1.1) are well known as typical procaryotic enzymes catalyzing the interconversion between Land D-alanine (for review see Refs. 16, 17, and 19). Because of their involvement in cell wall biosynthesis (20), they are essential for bacteria and potential targets for antibiotics (21). The kinetic and structural properties of the T. niveum enzyme indicate a close relationship to the procaryotic enzymes. Alanine racemases are typically monomers or dimers of identical subunits with molecular masses of about 40 kDa (22). The subunit of the T. niveum enzyme has a similar molecular mass of 37 ma, but the holoenzyme seems to be composed of at least three subunits as indicated by gel filtration. This is similar to the alanine racemase from Pseudomonas striata, where Roise et al. (18) observed a concentration-dependent assembling of three or more subunits.
All known alanine racemases utilize pyridoxal phosphate to build an intermediary SchiWs base with the substrate (19). We demonstrate that the activity of T. niveum racemase also depends on pyridoxal phosphate as the sole cofactor. Pyridoxal phosphate is loosely bound to the enzyme and can be removed by gel filtration or dialysis. We also have tested several possible racemase inhibitors that target the pyridoxal phosphate group.
Strong inhibition was caused by hydroxylamine (K. = 60 ~1~1 , probably by formation of an oxime with the pyridoxal phosphate aldehyde group. Additional well known inhibitors of alanine racemases are D-and L-(1-aminoethyll-phosphonate (Ala-P) (11). Incubation of the racemase for 30 min with 2 rn Ala-P resulted in decreased activities (60% for D-Ala-P and 40% for L-Ala-P). The inhibition is time-dependent and similar to the racemases from the Gram-positive organisms Staphylococ- Enzymatic Properties of Alanine Racemase from ' I : niueum cus aureus, Streptococcus faecalis (ll), and Bacillus stearotherrnophilus (23). The enzyme plays a key role in the biosynthesis of cyclosporins. The current knowledge of procaryotic non-ribosomal peptide synthesis is mainly based on studies focusing on the building mechanisms of the two procaryotic decapeptides, Gramicidin S (14) and Tyrocidine (15). Both compounds contain D-phenylalanine, and it has been shown that integrated domains of the peptide synthetases are responsible for the formation of these o-amino acids. L-Phenylalanine is accepted as a substrate and is isomerized after it is bound a s a thioester. Previous attempts to find a similar behavior in the case of cyclosporin showed that cyclosporin synthetase is not able to isomerize L-alanine and did not accept L-alanine as a substrate for the D-alanine position (3). Here we identify an alanine racemase responsible for the formation of D-alanine, the first example of a distinct racemase involved in non-ribosomal biosynthesis of peptides.
Recently linear peptides of different length were isolated from the cyclosporin synthetase by performic acid treatment (24). These peptides have been identified as thioester-bound intermediates of cyclosporin biosynthesis. All amino acids in all of the peptides represented partial sequences of cyclosporin A, starting with D-alanine as the N-terminal amino acid. In contrast, no formation of these intermediates could be detected in the absence of D-alanine, indicating that D-alanine acts as a starter amino acid and is required to initiate the formation of cyclosporin. Additionally, the formation of D-DKP in the presence of very low amounts of D-alanine demonstrates the high affinity of the cyclosporin synthetase for this starting amino acid. For these reasons production of D-alanine seems to be a rate-limiting step in cyclosporin biosynthesis.
In vivo, the cyclosporin high-producing strain l! niveurn 7939145 produces cyclosporin A in the range of 50 pg/ml of culture medium. The cellular concentration of racemase estimated from activity of crude extracts was found to be about 0.01% of the total cytosolic protein. Together with the high K,,, value for L-alanine, this indicates a very low cellular concentration of o-alanine. Compared with cyclosporin synthetase, the in vitro activity of the racemase measured under optimal conditions is sufficient to provide D-alanine for the cyclosporin synthesis in the fungal cell. We also determined the racemase activity in crude extracts of the l! niveurn wild type strain and furthermore in crude extracts of the I: niueurn related strain Cylindrosporum oligosperrnurn. The wild type strain, which produces relatively low amounts of cyclosporin A, showed only 30% of the racemase activity we found in the high producer, indicating that increased racemase activity could be responsible for an improved yield of cyclosporin. C. oligosperrnurn is a producer of a cyclic peptolide that is closely related to cyclosporin A. The main structural difference is the ester-linked D-2-hydroxyisovaleric acid that replaces the D-alanine of cyclosporin. Interestingly, we found no racemase activity in this strain. Previous attempts to characterize the enzyme system led to the identification of the diketopiperazine cyclo(D-alanine-N-methylleucine) as a further product of the cyclosporin synthetase (3). DKP represents a partial sequence of cyclosporin A and is the main reaction product only if D-alanine and L-leucine are available as amino acid substrates. We used and optimized this method as an in vitro assay and could show that cyclosporin synthetase is also able to synthesize cyclo(L-alany1-Nmethylleucine) (L-DKP), dependent on whether D-or L-alanine is offered as a substrate amino acid. When both isomers are offered, cyclosporin synthetase prefers the D-form. Even with a 10,000-fold molar surplus of L-alanine, D-DKP was the main reaction product reflecting the higher aftinity of the synthetase for D-alanine. This method also provides a very sensitive and selective enzymatic assay for the detection of D-alanine at a nanomolar level.