D-Lysergyl peptide synthetase from the ergot fungus Claviceps purpurea.

The ergot fungus Claviceps purpurea produces the medically important ergopeptines, which consist of a cyclol-structured tripeptide and D-lysergic acid linked by an amide bond. An enzyme activity capable of non-ribosomal synthesis of D-lysergyl-L-alanyl-L-phenylalanyl-L-proline lactam, the non-cyclol precursor of the ergopeptine ergotamine, has been purified about 18-fold from the ergotamine-producing C. purpurea strain D1. Analysis of radioactively labeled enzyme-substrate complexes revealed a 370-kDa lysergyl peptide synthetase 1 (LPS 1) carrying the amino acid activation domains for alanine, phenylalanine, and proline. The activation of D-lysergic acid is catalyzed by a 140-kDa peptide synthetase (LPS 2) copurifying with LPS 1. LPS 1 and LPS 2 contain 4′-phosphopantetheine and bind their substrates covalently by thioester linkage. Kinetic analysis of the synthesis reaction revealed a Km of ∼1.4 µM for both D-lysergic acid and its structural homolog dihydrolysergic acid, which is one to two orders of magnitude lower than the Km values for the other amino acids involved. The Km values for the amino acids reflect their relative concentrations in the cellular pool of C. purpurea. This may indicate that in in vivo conditions D-lysergyl peptide formation is limited by the D-lysergic acid concentration in the cell. In vitro, the multienzyme preparation catalyzes the formation of several different D-lysergyl peptide lactams according to the amino acids supplied. Specific antiserum was used to detect LPS 1 in various C. purpurea strains. In C. purpurea wild type, the enzyme was expressed at all stages of cultivation and in different media, suggesting that it is produced constitutively.

The ergot fungi growing on their host plants develop characteristic sclerotia containing ergot alkaloids. The basic structure of these compounds is a characteristic tetracyclic ring system (ergoline unit). The various ergolines are distinguished from each other by various side groups and modifications, and the most important among them is D-lysergic acid (for review, see Refs. 1 and 2). Natural and synthetic D-lysergic acid-derived compounds have various pharmacological activities that are used for human therapy (3). Knowledge about the biosynthesis of ergot alkaloids may help the development of new variants, and it may help in the suppression of their formation in the environment where they pose a serious risk for human and animal health.
Ergot peptide alkaloids, also called ergopeptines, are produced by Claviceps purpurea and consist of cyclol-structured tripeptides attached by amide linkage to D-lysergic acid (Fig.  1a). Variations in the structures of naturally occurring ergopeptines arise by substitutions of the amino acid positions I and II, while amino acid III is always L-proline. Related to the ergopeptines are the ergopeptams (4), which accumulate as by-products of ergopeptines in saprophytic cultures of some C. purpurea strains (5) (reviewed in Refs. 2 and 4). They contain D-proline, and their tripeptide chain is a non-cyclol lactam (Fig.  1b). It has been postulated that the cyclol in the ergopeptines arises by introduction of a hydroxyl group to the ␣-C of the amino acid I of a putative D-lysergyl (L,L,L)tripeptide lactam (6,7). The reaction product is proposed to undergo ring closure to yield the corresponding ergopeptine most probably in a nonenzymatic reaction (8). Ergopeptams arise by spontaneous epimerization of D-lysergyl (L,L,L)tripeptide lactams in the proline residue (5). Because of the D-configuration of proline, they cannot be converted into the corresponding ergopeptines and ergopeptams thus accumulate in cultures (4).
We have previously reported a cell-free system capable of forming the novel D-lysergyl peptides D-lysergyl-L-alanyl-L-phenylalanyl-L-proline lactam and the corresponding D-proline containing stereoisomer (9). They are synthesized from free D-lysergic acid, L-alanine, L-phenylalanine and L-proline with consumption of ATP (9). The formation of the (L,L,L) compound has not been detected in in vivo conditions as yet, because it appears to be converted rapidly to the corresponding ergopeptine or ergopeptam. The in vitro formation, therefore, confirms the previous hypotheses on ergopeptine formation from the D-lysergyl peptide lactam as well as the intermediacy of free D-lysergic acid in the biosynthetic process. In the present report we describe the purification and characterization of the more than 500-kDa enzyme system catalyzing D-lysergyl peptide lactam synthesis. It will be shown that the enzyme system has four peptide synthetase activities. They activate and incorporate into product D-lysergic acid and the three amino acids of the tripeptide moiety. A thiol template mechanism is used as for other non-ribosomal peptide syntheses (10).

MATERIALS AND METHODS
Strains and Cultures-C. purpurea wild type ATCC 20102, its mutant derivative D1 selected for high production of ergotamine, and strain 1029, were described previously (11). C. purpurea strain Ecc93 (kindly supplied by Dr. H. Kobel, Sandoz AG) is a high producer of ergocristine. Maintenance and culture conditions for ATCC 20102-derived strains were as described previously (11,12). Strain Ecc93 was cultivated according to Kobel and Sanglier (13). Inoculum medium and medium T25 were used for alkaloid fermentations (14,15). In some occasions a modified Vogel's medium was used (12,16).
Radiochemicals and Chemicals-[9,10-3 H]-9,10-Dihydroergocryp-* This work was supported by the Deutsche Forschungsgemeinschaft (Grant Ke 452/1-2). 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.
Methods of Analysis-Amino acid analysis of enzymatically formed radioactive peptides was performed as described previously (9). Quantitation and identification of enzymatically formed radioactive compounds were done either by liquid scintillation counting of ethyl acetate extracts of radioactive materials and by radioscanning of TLC separations, respectively. SDS-PAGE gels containing radioactively labeled proteins bands were first stained with Coomassie Blue and than were soaked with Amplify (Amersham International) according to the manufacturer's protocol and dried in a gel dryer. The gels were then either subjected to radioscanning or to autofluorography by exposure to x-ray film for several weeks. Similarly, for visualization of radioactive bands on TLC plates, exposure to x-ray film was for several days. In the case of tritium-labeled material, prior to exposure to x-ray film, plates were treated with ENHANCE (DuPont NEN) spray according to the manufacturer's instructions. Protein concentrations were determined according to Bradford (18). SDS-polyacrylamide gel electrophoresis was performed according to Laemmli (19). Staining of gels was either with Coomassie Blue or with silver (20). For molecular mass determinations, 4% SDS-PAGE gels in a mini-gel apparatus (Hoefer) were used. Molecular mass markers were enniatin synthetase (350 kDa), surfactin synthetase II (405 kDa), actinomycin synthetase II (280 kDa), gramicidin synthetase 2 (510 kDa), and myosin (205 kDa). Western blots were performed with 5% SDS-gels essentially according to Towbin et al. (21) in a tank blotter. Detection of immunopositive bands was performed by standard techniques using 1:10,000 dilution of primary antibody, phosphatase-conjugated goat anti-rabbit antibody (Sigma) and nitro blue tetrazolium and 5-bromo-4-chloro-3-indolyl phosphate as reagents. 4Ј-Phosphopantetheine determinations of protein fractions (liquid samples) were done by the method of Pugh and Wakil (22) modified according to Zocher et al. (23). Alternatively determinations of 4Ј-phosphopantetheine in protein bands were done according to a method of Stindl (24). To this end, enzyme concentrated after the DEAE-cellulose step (Table I) was electrophoresed in 1.5-mm thick SDS-PAGE gels and blotted onto glassy bond membranes. After brief staining with Coomassie Blue, bands of interest were cut out (in general from 5 to 10 lanes run in parallel), combined, and transferred to Eppendorf tubes. After treatment with 1 M NaOH in a boiling water bath for 1 h, samples were neutralized in 1 M Tris-HCl, pH 8. Alkaline phosphatase was added, and incubation was resumed for a further hour at 37°C. After this, the sample was subjected to a microbiological assay for the presence of panthetheine based on growth dependence of Lactobacillus plantarum DSM 20205. Controls involved treatment of glassy bond strips from the same lanes but devoid of protein bands and strips with protein bands known to contain no 4Ј-phosphopantetheine.
Enzyme Assays-Amino acid and ergoline carboxylic acid dependent ATP-pyrophosphate exchange reactions were performed as described previously (25,26). Assays for measuring D-lysergyl peptide lactam synthesis were as described elsewhere (9). In the case of testing fractions eluting from column separations, 1-ml portions from fractions of Ultrogel separations were concentrated about 10-fold on small DEAEcellulose columns in Pasteur pipettes, equilibrated in buffer B (see "Buffers and Solvent Systems"). After elution with 0.2 M NaCl (in buffer B), enzyme was desalted on small AcA 202 Pasteur pipette columns eqilibarated with buffer B without dithioerythritol and assayed immediately. In the case of DEAE-cellulose fractions, the same procedure was used except that prior to concentration, samples were diluted 2-fold with buffer B to allow enzyme to adsorb to DEAE-cellulose. Reaction products were extracted into ethyl acetate and, after evaporation of solvent to dryness, analyzed by TLC in various solvent systems (see below). The assay for measuring thioester formation was as described previously (25).
Buffers and Solvent Systems-Buffer A for the preparation of extracts of broken cell of C. purpurea was described previously (9). Buffer B was 0.1 M Tris⅐HCl, pH 8.0, 15% (w/v) glycerol, 10 mM dithioerythritol, 1 mM benzamidine, 2 mM EDTA, 2 mM phenylmethylsulfonyl fluoride. For the preparation of buffers, bidistilled water was used that had been previously autoclaved at 121°C for 20 min.
Enzyme Purification-All operations were performed at 2-4°C. Approximately 50 g of freshly harvested mycelium of C. purpurea were suspended in 150 ml of buffer A and passed through a French press at 10,000 p.s.i. This method of cell disintegration was less time-consuming than the previously used procedure (9) and yielded more units of enzyme. The resultant homogenate was centrifuged at 15,000 rpm for 30 min in a SS34 rotor. To the supernatant was added a 15% Polymin P solution to give 0.3% final concentration. After 30 min, the sample was centrifuged as above. The clear supernatant was precipitated by fractionated ammonium sulfate from 35 to 55% saturation for 3 h standing on ice. After centrifugation as above, the pellet was dissolved in buffer B (designated crude extract). 3-4-ml portions of crude extract were subjected to Ultrogel AcA 34 gel chromatography on a column of bed dimensions 48 ϫ 1.8 cm. 4-ml fractions were collected, and the various fractions were assayed for D-lysergyl peptide lactam synthesis or for peptide synthetase activities such thioester formation (25) or the amino acid-dependent ATP-pyrophosphate exchange (26). Fractions catalyzing D-lysergyl peptide lactam synthesis were pooled and applied onto a DEAE-cellulose column (bed dimensions 16 ϫ 1.8 cm) that had been preequilibrated with buffer B. After washing the column with buffer B, the enzyme was eluted with a linear gradient from 0 to 0.3 M NaCl. Under these conditions, the enzyme elutes between 0.13 and 0.15 M NaCl. The peak located by assaying for D-lysergyl peptide lactam synthesis was diluted with two volumes of buffer B and concentrated by application onto a small DEAE-cellulose column (bed dimensions 1 ϫ 5 cm) and subsequent elution with 0.2 M NaCl (see Table I). The 3-4 ml of the resultant eluate were applied onto a Ultrogel AcA 22 column (bed FIG. 1. Panel a, structure of ergopeptines. Natural ergopeptines vary in the amino acid residues in positions I (alanine, aminobutyric acid, and valine) and II (aminobutyric acid, phenylalanine, valine, leucine, and isoleucine). There is always proline in position III. Examples are: Substituents are as in the ergopeptines (a). Ergopeptams isolated from C. purpurea cultures contain D-proline. L-Proline-containing D-lysergyl peptide lactam has been obtained by enzymatic synthesis as described in the text. dimension 50 ϫ 1.8 cm) that had been previously equilibrated with buffer B, and 3-ml fractions were collected. The enzyme eluted as a single peak. At this point, the enzyme was not further purified and immediately used up for investigation. Freezing the enzyme preparation led to heavy activity loss.
Preparation of Antibodies-The band of D-lysergyl peptide synthetase (LPS 1) 1 was cut after brief staining in 8-anilino-1-naphthelenesulfonate from 20 lanes of a 3-mm thick 5% SDS-gel, each lane containing a total of 400 g of protein from the DEAE-cellulose step. After maceration of the gel pieces the protein was electroeluted in an Elutrap (Schleicher & Schuell) overnight using Laemmli electrophoresis buffer containing SDS. The electro-eluate (700 l) contained about 150 g of protein and, after mixing in complete Freund's adjuvant, was administered to a rabbit. A second injection of 150 g was given 4 weeks after the first injection in incomplete Freund's adjuvant. Serum was taken 6 weeks after the first injection, and final bleeding was after 3 months. The resultant antiserum was used for Western blotting without further purification.

RESULTS
Purification of D-Lysergyl Peptide-synthesizing Enzyme-C. purpurea strain D1 produces ergotamine and minor amounts of Leu-ergokryptine. (Later, when antibodies were available (see below) it became clear that other strains in our laboratory such as Ecc93 (kindly provided by Dr. H. Kobel, Sandoz AG, Basel) or 1029 (11) would also have been suitable for enzyme preparation.) For enzyme extraction from strain D1, cells were harvested after 70 h of growth in inoculum medium or 120 h of growth in T25 medium. At these times the cultures were starting the main ergotamine accumulation phase (15). Ergotamine was already detectable at ϳ10 mg/liter after 36 h of growth in T25 medium. The extract from freshly harvested cells was fractionated by Polymin P precipitation and ammonium sulfate precipitation. Instead of total loss after standing on ice for 1 day (9), this preparation had a half-life of 16 -20 h in buffer containing 15% glycerol, 10 mM dithioerythritol (or dithiothreitol), and various protease inhibitors. In the absence of dithioerythritol, the half-life of enzyme activity was reduced to 2-4 h in different preparations. The enzyme activity is, however, strongly inhibited by dithioerythritol and similar reagents (9). Therefore, enzyme preparations were always desalted into dithioerythritol-free buffer prior to assaying. The assay reaction then proceeded fairly linearly for about 10 min at 25°C, then continued at a gradually reducing rate for up to 30 min.
Further fractionation of crude extract was done with small portions (3-4 ml) on relatively small, fast-running Ultrogel AcA 34 columns. D-Lysergyl peptide-synthesizing activity appeared immediately behind the void volume, indicating that the protein was larger than 300 kDa, as was to be expected for a multifunctional protein involved in a non-ribosomal peptide synthesis (Fig. 2). The activity profile of D-lysergyl peptide formation showed a single peak which coincided with the peak for D-lysergic acid activation, measured by ATP-pyrophosphate exchange, and the binding of the three amino acids alanine, phenylalanine, and proline as thioesters (curves for alanine and proline are not shown). Interestingly, in these fractions dihydrolysergic acid was activated to nearly the same level as D-lysergic acid, which is consistent with the in vivo incorporation of this compound into the corresponding dihydroergopeptines (27). Further purification of enzyme was achieved by chromatography on DEAE-cellulose from which it eluted as a single peak between 130 and 150 mM NaCl (not shown). Finally, the enzyme preparation was subjected to gel filtration on Ultrogel AcA 22. The overall purification was about 18-fold and the yield was 1.7%. The data for the individual steps are listed in Table I. Attempts to achieve higher specific activity were unsuccessful because of severe activity losses. It was noted that phenylmethylsulfonyl fluoride improved enzyme stability at all stages of purification. SDS-PAGE revealed that the preparation contained at all stages multiple protein bands. The two strongest bands in the M r range higher than that of myosin (205,000) can be seen in Fig. 3 (indicated by arrows). They appear to become enriched in the course of the activity purification and were therefore considered as candidates for the putative D-lysergyl peptide synthetase.
Identification of D-Lysergyl Peptide Synthetase-To see which one of the two prominent large proteins seen in Fig. 3 was responsible for synthesis of the D-lysergyl peptide lactam, concentrated enzyme purified on Ultrogel AcA 22 was incubated with radioactively labeled phenylalanine, proline, alanine, or dihydrolysergic acid each in the presence of ATP. The samples were fractionated by SDS-PAGE, and the radioactive bands were identified by autofluorography. Fig. 4 shows that [ 14 C]phenylalanine labeled the slowest migrating protein (des- 1 The abbreviations used are: LPS, D-lysergyl peptide synthetase; PAGE, polyacrylamide gel electrophoresis. ignated LPS 1) and a protein of ϳ120 kDa which was presumed to be a degradation product of the larger protein. Interestingly, labeling with [ 3 H]dihydrolysergic acid revealed a 140-kDa protein, LPS 2, which is responsible for D-lysergic acid activation and incorporation into the peptide.
Labeling of LPS 1 by [ 14 C]proline and [ 14 C]alanine yielded only a faint band in the autofluorograms. Besides the weak response of 14 C-labeled compounds to the fluorophor used in these experiments, the faint labeling may be due to the high K m values for alanine and proline (see below) and the fact that purified enzyme preparations usually contain varying amounts of amino acids already bound to the enzyme. This can also explain uneven product labeling (9). Furthermore, instability of thioesters during electrophoresis might also influence labeling.
Preparative electrophoresis was used to obtain LPS 1 to prepare specific antiserum. Western blots of samples from the various purification steps show many of the lower bands crossreacting with the antiserum. The increase of these bands during isolation (Fig. 3, lane 3Ј) suggests that LPS 1 is, at least in part, cleaved after cell disruption. Interestingly, the antibodies show no significant cross-reaction with the strong band copurifying with LPS 1 and migrating between myosin and LPS 1, suggesting that this band represents a protein unrelated to LPS 1.

Molecular Mass Determination of D-Lysergyl
Peptide Synthetase-The molecular mass of LPS 1 was determined using 4% SDS-PAGE gels using peptide synthetases of known molecular mass as reference markers. Fig. 5 shows that LPS 1 has a molecular mass of 370 kDa, which should be sufficient for three amino acid activation domains (28,29). From these considerations it is unlikely, however, that the 370-kDa protein could contain all four amino acid domains required for synthesis of the whole D-lysergyl peptide lactam. LPS 2, which was found in the same protein fraction, is most probably carrying the residual fourth domain required for tetrapeptide synthesis.
Characteristics of the Substrate Binding Reaction-Covalent binding of substrate amino acids to peptide synthetases in thioester-linkage is a prerequisite for peptide formation in nonribosomal peptide synthesis (28,29). Accordingly, when testing LPS 1 we saw thioester formation of alanine, phenylalanine, and proline (as measured by conversion of substrates into a trichloroacetic acid-precipitable form) which was very rapid and required ATP and Mg 2ϩ . Analysis of these covalent enzyme-substrate complexes showed that they could be cleaved by performic acid but not by formic acid. Cleavage was also seen upon treatment with mild alkali such as 0.1 N NaOH for 10 min at room temperature. 2 Fig. 6 shows the results of performic acid oxidations in the case of phenylalanine and proline (alanine not shown).
Presence of 4Ј-Phosphopantetheine in LPS-Portions of enzyme fractions purified on Ultrogel AcA 34 or DEAE-cellulose were subjected to determination of 4Ј-phosphopantetheine, which is a covalently bound cofactor in every amino acid activation domain of all peptide synthetases (30). The analysis showed that the peak of LPS activity coincided with that of 4Ј-phosphopantetheine contents of the various fractions (results not shown). To test whether LPS 1 and LPS 2 contained 4Ј-phosphopantetheine, concentrated enzyme from the DEAEcellulose step was subjected to SDS-PAGE, and the separated proteins were transferred to membranes by electroblotting. After isolation, bands of LPS 1 and LPS 2 were subjected to determination of 4Ј-phosphopantetheine using a microbiological assay with L. plantarum. Hydrolysates of LPS 1 and LPS 2 stimulated growth of the Lactobacillus and therefore contained 4Ј-phosphopantetheine (not shown). Controls using empty 2 B. Walzel, B. Riederer, and U. Keller, unpublished data. a One unit is the amount of enzyme catalyzing the formation of 1 nmol of D-lysergyl-alanyl-phenylalanyl-proline lactam at 26°C under the conditions described under "Materials and Methods." b ND, not determined.

FIG. 3. SDS-PAGE and Western blot analysis analysis of various steps in the purification of D-lysergyl peptide synthetase.
Samples from the various purification steps indicated in Table I  membrane strips or with strips with bovine serum albumin or lysozyme gave no growth of the test organism.
Kinetic Properties of D-Lysergyl Peptide Synthetase-Determinations of K m values of the various substrates of the multienzyme were partly hampered by the fact that the partially purified enzyme preparations contain varying (small) amounts of free amino acids associated with the protein. We occasionally even detected formation of D-lysergyl peptide lactam in the absence of externally added D-lysergic acid when using enzyme preparations from cultures with exceptionally high alkaloid yield. The apparent K m values for proline and alanine were ϳ125 and ϳ190 M, respectively. The K m for phenylalanine was ϳ15 M, reflecting the lower concentration of this amino acid in the free amino acid pool of C. purpurea ATCC 20102 (31). D-Lysergic acid and dihydrolysergic acid gave the same K m of ϳ1.4 M. Judging from the substrate velocity curves, dihydrolysergic acid acts competitively with endogenous D-lysergic acid and has a 2.5-fold lower V max (not shown). The K m for ATP was ϳ210 M. The reactions were absolutely dependent on supplementary ATP, indicating that none was present in the enzyme preparations.
Substrate Specificity of D-Lysergyl Peptide Synthetase-C. purpurea D1 produces ergotamine together with minor amounts of Leu-ergokryptine. Besides the ergotamine group of ergopeptines with alanine in amino acid position I, there exist also the ergoxine and ergotoxine groups which contain aminobutyric acid or valine in position I, respectively (1,4). To test whether our enzyme preparation could also synthesize these peptide lactams it was incubated with [ 3 H]dihydrolysergic acid, phenylalanine, proline, alanine, aminobutyric acid, or valine. Fig. 7 (lanes 1-3) shows TLC separations of ethyl acetate extracts from these reactions each of which had produced the expected dihydrolysergyl peptide lactam. Moreover, when phenylalanine was replaced by leucine, dihydrolysergyl peptide lactam homologs of Leu-ergokryptine and Leu-ergoptine were formed (lane 4 and 5, respectively). The identity of all of these compounds was confirmed by alkaline and acid hydrolysis after labeling with [ 3 H]dihydrolysergic acid, [ 14 C]phenylalanine, [ 14 C]leucine, or [ 14 C]proline. The formation of these compounds was ATP-dependent (lane 6). The results shown here suggest that LPS 1 and LPS 2 can synthesize various different Dlysergyl peptides and that the diversity of structures elaborated by the various C. purpurea strains, at least in part, reflects the different actual concentrations of substrate amino acids in their free cellular pools.
Immunological Screening-Western blots were performed of extracts of broken cells of a number of peptide alkaloid-produc-   a and b) in the presence of MgATP as described. After precipitation and washing with 7% trichloroacetic acid, protein was subjected to treatment with formic acid (lanes a and c) and performic acid (lanes b and d). Reaction products were chromatographed on silica gel plates using solvent system V and visualized by autofluorography. The data show that amino acids are released from the enzyme by treatment with performic acid, but not by formic acid. Note that the additional band in lane d is the formylation product of phenylalanine. ing C. purpurea strains from this laboratory such as ATCC 20102 (wild type), strain D1, strain 1029, and C. purpurea strain Ecc93 cultured under different conditions and for different times. Antibody to LPS 1 recognized a 370-kDa band in all cases together with lower bands (raised by proteolysis), which were present in appreciable amounts when the strains were grown in media favoring ergot peptide alkaloid synthesis (such as inoculum medium, medium T25, or production medium of Ecc93) (not shown). Despite the different ergotamine productivities of the C. purpurea strains (e.g. wild type strain ATCC 20102 produces ϳ10 -15 mg/liter, and strain D1 produces ϳ700 -1000 mg of ergotamine/liter after 14 days of cultivation), the amount of immunoreactive material in all of the strains was fairly comparable (not shown). Obviously, from these results all strains would have the ability to synthesize appreciable amounts of ergopeptines. A possible reason for the different productivities may be the limited production of D-lysergic acid. In fact, short term productivity measurements in C. purpurea wild type strain ATCC 20102 revealed a 4 -5-fold stimulation of ergotamine formation by the addition of D-lysergic acid to protoplasts or intact mycelium (31). Similar experiments with strain D1 did not show a response upon addition of D-lysergic acid, which may indicate that in this strain the intracellular level of D-lysergic acid was at saturation and therefore considerably higher than in its parent. 3 Correlation of Ergotamine Production and Levels of LPS-When grown in a modified Vogel's medium (12,16) with a high phosphate content, C. purpurea wild type strain or strain 1029 developed long, slim vegetative-type mycelium, which did not produce ergot peptide alkaloid (12). In production medium where phosphate is limited, productivity is associated with a type of morphology called sclerotia-like cells (12) (for review, see Ref. 32). Surprisingly, Western blots with cell extracts of wild type strain grown in the two different media showed the presence of comparable levels of LPS 1 in both vegetative and sclerotia-like cells (not shown). Thus, LPS 1 is synthesized constitutively and thus not responsible for growth-linked repression of peptide alkaloid synthesis by phosphate (33,34). DISCUSSION The work presented here characterizes the enzyme system catalyzing the formation of D-lysergyl-alanyl-phenylalanyl-proline lactam, the non-cyclol peptide precursor of ergotamine in the ergot fungus C. purpurea. Conditions of isolation and purification were established that enabled us to identify two proteins responsible for the activation and condensation of the building blocks of the peptide alkaloid. As it appears, LPS 1 is a multifunctional polypeptide chain of 370 kDa that activates the amino acids of the tripeptide portion of the peptide alkaloid (alanine, phenylalanine, and proline) in covalent thioester linkage. LPS 2 is a 140-kDa protein-activating D-lysergic acid only. The presence of 4Ј-phosphopantetheine covalently bound to enzyme in peptide alkaloid-synthesizing fractions has been demonstrated, and the cofactor could be detected unambiguously in both proteins. This identifies them as the thiol template of ergot peptide alkaloid synthesis. Such thiol templates involved in the biosynthesis of various antibiotic peptides from bacteria and fungi have been shown to consist of repeating units of approximately 1000-amino acid length (equivalent to 120 kDa each) (28,29). These units contain regions with homology to adenylating enzymes as well as regions with homology to acyl carrier proteins containing 4Ј-phosphopantetheine as prosthetic groups and are referred to as peptide synthetase domains. After activation as adenylates, the amino acids become attached to the 4Ј-phosphopantetheines of the peptide synthetase domains in thioester linkage, thus serving as carriers in amino acid and peptidyl transfer during their polymerization (10).
Enzymes LPS 1 and LPS 2 comprise a total length of more than 500 kDa, which should be sufficient to activate and polymerize four acyl or aminoacyl residues. Indeed, the purified enzyme catalyzes the formation of the whole D-lysergyl peptide lactam containing four peptide bonds. Similar sizes are seen in the case of gramicidin S synthetase 2 (510 kDa, activating four and polymerizing five amino acids) (35); HC toxin synthetase (550 kDa, activating, modifying, and polymerizing four amino acid residues) (36); or aminoadipyl-cysteinyl-valine synthetase (421 kDa, activating, modifying, and polymerizing three amino acids) (37).
The attachment of D-lysergic acid to the tripeptide portion in the ergopeptines is through an amide bond which requires activation of the carboxyl group at least as adenylate. Conflicting results have been obtained previously because of the occurrence of a D-lysergic acid-activating enzyme in extracts of C. purpurea, which catalyses the formation of D-lysergyl adenylate but not of dihydrolysergyl adenylate and which does not form thioester (9,26). On the basis of these findings it was presumed that ergot peptide synthesis would proceed in a similar fashion as in the case of the class of acylpeptide lactones from streptomycetes (38). Here, aromatic carboxylic acids are activated by relatively small enzymes (45-60 kDa) as adenylates but not as thioesters. The adenylate reacts later with the next amino acid in the reaction chain on the surface of a large multifunctional peptide synthetase that carries this amino acid as thioester. The reaction product is the corresponding acylamino acid that can react further with the next amino acid of the peptide lactone ring as in the case of actinomycin (39 -41). The data described here clearly show that this type of reaction will not occur in D-lysergyl peptide formation because LPS 2 activates D-lysergic acid (or dihydrolysergic acid) as a thioester and thus resembles other single amino acid-activating enzymes with a known acyl carrier-like module such as gramicidin S synthetase 1 (42). Furthermore, recent data from this laboratory indicate that the D-lysergic acid-activating enzyme could be separated from the peptide-synthesizing multienzyme, proving their independence from each other. 2 An important difference between bacterial and fungal peptide synthetases is that the latter contain all domains on one single polypeptide chain, such as in the case of enniatin synthetase (43) or cyclosporin synthetase (44), while in the bacterial ones they are contained in more than one protein. Because of the strong degradation of LPS 1 during isolation, LPS 1 and LPS 2 could be produced from a larger polypeptide by proteolytic cleavage. The gene sequence for the HC-toxin synthetase of Cochliobolus carboneum predicts a 550-kDa protein, but two separate proteins always have been isolated (36). These proteins activated three of the four amino acid constituents of the toxin. They arose by fragmentation of the 550-kDa polypeptide that was barely visible in SDS-PAGE gels (36). No Ͼ500-kDa protein has been detected yet in cell extracts from C. purpurea, and the sequence of the LPS gene is not yet available.
The constitutive expression of LPS and finely tuned dependence of enzyme activity on the supply of D-lysergic suggests that the regulation may occur at the earlier steps, maybe at the synthesis of the ergoline ring carboxylic acid. Dimethylallyl tryptophan synthase, an enzyme catalyzing the first step in the ergoline ring synthesis, has been described to be inducible through tryptophan and repressible through phosphate as is alkaloid production (33). It would therefore be an appropriate target for the observed repression of ergotamine synthesis in C. purpurea wild type in Vogel's medium (12). In the cases of enniatin synthetase in Fusarium scirpi, the producer of the cyclohexadepsipeptide enniatin B and of other toxins of phytopathogenic fungi, constitutive peptide synthetase expression has also been demonstrated (45). It may be argued that the constitutive expression of peptide synthetases may allow rapid conversion of key metabolites such as D-lysergic acid into the peptide alkaloids. D-Lysergic acid, in contrast to clavines or ergot peptides, may be too unstable to be deposited safely in the growing sclerotium.
An interesting insight into the enzymatic regulation of ergot peptide synthesis came from kinetic characterization of the enzyme complex, which indicates that D-lysergyl peptide synthesis is strongly controlled by the available free D-lysergic acid due to the low apparent K m value, which is ϳ1.4 M. Limitation of ergotamine production by low D-lysergic acid concentration of ergotamine production was seen in C. purpurea wild type (31) but not in the ergotamine high producing derivative strain D1. Furthermore, the rate of synthesis and the spectrum of products is controlled by the nature and concentration of the amino acids present in the cell-free incubations, which is consistent with earlier in vivo data concerning the role of amino acids in the cellular pool of C. purpurea (31,46). From the data presented here it follows that LPS has a broad amino acid substrate specificity giving it the capacity to synthesize naturally many different ergopeptine structures, probably according to fluctuations in the free amino acid pool. These properties make LPS a promising tool for the development of novel Dlysergic acid-containing compounds in the future.