Propionate CoA-transferase from Clostridium propionicum has been purified and the gene encoding the enzyme has been cloned and sequenced. The enzyme was rapidly and irreversibly inactivated by sodium borohydride or hydroxylamine in the presence of propionyl-CoA. The reduction of the thiol ester between a catalytic site glutamate and CoA with borohydride and the cleavage by hydroxylamine were used to introduce a site-specific label, which was followed by␣MALDI-TOF-MS. This allowed the identification of glutamate 324 at the active site. Propionate CoA-transferase and similar proteins deduced from the genomes of Escherichia coli, Staphylococcus aureus, Bacillus halodurans and Aeropyrum pernix are proposed to form a novel subclass of CoA-transferases. Secondary structure element predictions were generated and compared to known crystal structures in the databases. A high degree of structural similarity was observed between the arrangement of secondary structure elements in these proteins and glutaconate CoA-transferase from Acidaminococcus fermentans.
Clostridium propionicum has been isolated as an alanine fermenting organism from the black mud of San Francisco bay . The fermentation products were acetate, ammonia, carbon dioxide and propionate . In contrast to other organisms, which ferment alanine according to the so-called randomising pathway with succinate␣as␣a␣symmetric intermediate, C. propionicum ferments alanine via the nonrandomising pathway with acrylyl-CoA as characteristic intermediate. This pathway seems to be restricted to a limited number of organisms, including Megasphaera elsdenii, Bacteroides ruminicola, C. propionicum and Clostridium homopropionicum. Serine and cysteine are fermented in a similar manner by C. propionicum yielding acetate and propionate. Likewise, threonine is fermented to propionate and butyrate as volatile fatty acid end products.
As outlined in Fig. 1, alanine undergoes an initial oxidative cleavage to ammonia and pyruvate, the latter being either oxidized to carbon dioxide and acetate or reduced to (R)-lactate. (R)-Lactate is subsequently reduced to propionate by reactions carried out at the coenzyme A thiol ester level rather than using the free carboxylates. A critical step in the reductive branch of this pathway is the activation of (R)-lactate as its (R)-lactoyl-CoA derivative. This reaction is carried out by the enzyme propionate:acetyl-CoA CoA-transferase (EC 18.104.22.168, also known as propionate CoA-transferase), which has been demonstrated to activate (R)-lactate using the end product of the reduction, propionyl-CoA, or acetyl-CoA as a coenzyme A donor  (Eqn 1).
The enzyme has been previously purified and characterized as a homotetrameric enzyme (α4) with an apparent molecular subunit mass of 67 kDa . Although a preference of (R)-lactate over (S)-lactate was observed, the enzyme exhibited a rather broad substrate specificity for monocarboxylic acids including acrylate, propionate and butyrate whereas dicarboxylic acids were not used.
The general mechanism for the CoA-transferases has been suggested to proceed via the successive formation of a mixed anhydride between the CoA-donor carboxylic acid and an essential glutamate residue of the enzyme, followed by the formation of an enzyme-CoA thiol ester intermediate. The product is then formed by an inverted sequence of these steps with the acceptor carboxylate . More recently, a number of CoA-transferases have been discovered, which apparently do not follow this general mechanism [8,9]. Formation of an enzyme-CoA thiol ester can be detected either by site-specific cleavage of the polypeptide chain , reduction of the thiol ester with sodium borohydride [11–13] or directly by mass spectrometry . Alternatively the mixed anhydride intermediates can be detected by the formation of stable derivatives of the catalytic glutamate and hydroxamic acids  or indirectly by oxygen exchange experiments [14,16].
To date, only the crystal structure of glutaconate CoA-transferase (EC 22.214.171.124) from Acidaminococcus fermentans has been solved. In contrast to the homotetrameric propionate CoA-transferase of C. propionicum, the former enzyme is a hetero-octameric protein (α4β4), whose structure belongs to the open α helix/β sheet protein family and has been described as four-layered α helix/α helix/β sheet/α helix␣type with a novel topology . This topology is found in both subunits and differs considerably from other α helix/β sheet proteins including nucleotide-binding domains.
In this communication, we report the cloning and sequencing of the gene encoding propionate CoA-transferase from C. propionicum and present experimental evidence that glutamate 324 acts as the catalytic carboxylate.
Materials and methods
Sequencing grade proteases were purchased from Boehringer Mannheim (Germany). Coenzyme A (trilithium salt) was from ICN Biomedicals (Eschwege, Germany). All other chemicals were of the highest grade available and from common commercial sources. C. propionicum (DSMZ 1682) was purchased from the German collection of microorganisms and cell cultures (DSMZ, Braunschweig, Germany).
Synthesis of acyl-CoA substrates
Acetyl-, butyryl- and propionyl-CoA were prepared from the corresponding anhydrides and CoA by the method of Simon & Shemin . All CoA-derivatives were purified as described previously .
The enzyme test for propionate CoA-transferase activity was carried out at 25 °C as described previously .
Cultivation and storage of microorganisms
C. propionicum was cultivated in a complex medium containing d,l-alanine as the sole source of energy, as described previously . Freshly prepared anaerobic media were inoculated with 5 to 20% stationary or late exponential precultures and grown for 24 to 36 h at 37 °C. The cells were harvested by centrifugation and stored at −80 °C.
Preparation of cell free extracts of C. propionicum
Frozen cells (20 g) were suspended in 100 mL of 25 mm potassium phosphate, 1 mm dithiothreitol, 1 mm EDTA, 1 mm MgCl2, pH 7.0 (Buffer A). The suspension was homogenized by sonication on ice for 15 min. Poly(ethyleneimine) (0.2% m/v final concentration) was added and the crude extract was centrifuged for 45 min at 100 000 g. The clear supernatant was stored on ice until used.
Purification of propionate CoA-transferase
All purification steps were carried out at 4 °C. The clear␣extract was applied to a Q-Sepharose™ column (2.6 × 10 cm, Pharmacia) equilibrated with buffer A. The column was washed with 50 mL of buffer A and developed by a 500-mL linear gradient of 0 to 500 mm NaCl in buffer A. Fractions containing CoA-transferase activity were pooled and adjusted to a final concentration of 1 m (NH4)2SO4. The solution was centrifuged for 45 min at 100 000 g and applied in four aliquots to a Resource-Phe™ column (1 mL vol., Pharmacia) equilibrated with 1 m (NH4)2SO4 in buffer A. The column was washed with 5 mL of the starting buffer and the proteins were eluted in a 50-mL linear gradient from 1 to 0 m (NH4)2SO4. The pooled enzyme was dialysed overnight against 5 mm of each boric, citric and phosphoric acids and 5 mm Tris adjusted to pH 7.0 with KOH (buffer B) and then applied on a Resource-Q™ column (1 mL vol., Pharmacia) equilibrated with buffer B. The protein was eluted by a linear gradient (50 mL) from 0% to 100% of buffer B adjusted to pH 2.0 with HCl. The collected fractions at pH 5.0–5.1 were immediately neutralized by the addition of potassium phosphate, pH 7.5. When required, the protein solution was adjusted to a final concentration of 200 mm sodium acetate and subjected to gel filtration on Sephadex G25, equilibrated with buffer A, to obtain CoA-free protein. The purified enzyme was filter-sterilized and stored at 4 °C for several months without significant loss in activity. The purity of the enzyme was confirmed by SDS/PAGE with Coomassie-staining of the proteins and by RP-HPLC followed by mass spectrometry .
Purified enzyme (20 µg) was subjected to SDS/PAGE and blotted onto a poly(vinylidene difluoride) membrane. After staining of the membrane with Coomassie blue R450, the protein band was cut out and subjected to N-terminal sequencing by gas-phase Edman degradation.
Generation and purification of internal peptides
The purified enzyme (200 µg) was freeze-dried, reductively carboxymethylated and digested with trypsin as described previously . The peptides were purified by RP-HPLC using a Supelcosil-LC318 column (4.6 × 250 mm, 5 µm, 300 Å) equilibrated with 0.1% (v/v) trifluoroacetic acid and eluted with a linear gradient of 0–42% (v/v) acetonitrile within 42 min. The elution of the peptides was monitored simultaneously at 210 and 280 nm. Peptides exhibiting an absorbance at 280 nm were re-applied to the same column using an identical gradient in the presence of 0.1% (v/v) hexafluoroacetone-ammonia, pH 6.0, and analysed by MALDI-TOF-MS as described previously . Two peptides of suitable size and purity were subjected to N-terminal sequencing.
Cloning and DNA-sequencing
A degenerated primer pair was deduced from the N-terminus of the enzyme and from an internal peptide (Table 2) and used to amplify a ≈ 300-bp PCR product from genomic DNA of C. propionicum. The PCR product was cloned and sequenced using the TOPO TA-cloning kit according to the manufacturer's instructions.
The PCR-product was labelled with digoxygenin and used to screen EcoR1 fragments of genomic DNA from C. propionicum cloned in a λ-ZAP-Express phage. Two positive clones were isolated, plasmids were excised from the vector and the inserts were sequenced using standard laboratory protocols. The sequence data were analysed using the expasy (Expert Protein Analysis System) server of the Swiss Institute of Bioinformatics, the computational molecular biology facilities provided by the National Institute of Health and the 3d-pssm Web Server V2.0 provided by the Imperial Cancer Research Fold Recognition Server [20,21].
Site-specific label of the catalytic glutamate residue
Propionate CoA-transferase (50 µg) was incubated for 2 min at 25 °C in 50 mm potassium phosphate, pH 7.0, either in the presence or absence of 100 µm propionyl-CoA. Either hydroxylamine hydrochloride (pH 7.5, 200 mm final concentration) or sodium borohydride (20 mm final concentration) were added and the reaction was allowed to proceed for another 10 min at 25 °C. Aliquots of the samples were assayed for CoA-transferase activity. The samples were reductively carboxymethylated and the buffer was exchanged by gel filtration to yield 50 mm ammonium acetate, pH 8.0, 10% (v/v) acetonitrile. Each sample was split into four equal volumes and digested for 16 h at 37 °C in the presence of either 2% (w/w) chymotrypsin, endoproteinase AspN, endoproteinase GluC (V8 protease) or 2% (w/w) trypsin. The samples were acidified with 0.01 vol. of 2 m trifluoroacetic acid and analysed by MALDI-TOF-MS.
Purification of propionate CoA-transferase
The propionate CoA-transferase was purified from C. propionicum grown on alanine as the sole source of energy and carbon. Ion exchange chromatography on Q-Sepharose completely separated the propionate CoA-transferase (elution at 270–285 mm NaCl) from phosphotrans-acetylase (elution at 130–150 mm NaCl), which contributed to 50% of the apparent transferase activity in cell-free extracts. Most of the contaminating proteins were removed by hydrophobic interaction chromatography on a Resource-Phe column [elution at 700 mm (NH4)2SO4]. The remaining impurities were removed by ion exchange chromatography on Resource-Q using a decreasing pH-gradient (elution at pH 5.1–5.0).
As demonstrated in Fig. 2 and Table 1, the enzyme was essentially pure after these purification steps. The enzyme was 37-fold enriched to a specific activity of 85 U·mg protein−1. In addition to the predominating polypeptide with an apparent molecular mass of 60 kDa in SDS/PAGE, two faint bands were observed around 40 and 20 kDa. These additional bands were completely absent when the purified enzyme was incubated in the presence of sodium acetate followed by gel filtration on Sephadex G25, but accounted for up to 30% of the total protein when the enzyme was incubated with 100 µm propionyl-CoA, prior to sample preparation. These data indicated that a small but significant fraction of the purified CoA-transferase was trapped as the enzyme-CoA thiol ester intermediate. It has been previously shown that glutamate thiol ester-containing proteins, such as CoA-transferases and α2-macroglobulin, are site-specifically cleaved at elevated temperature ; a nucleophilic attack of the peptidyl amide nitrogen of the glutamyl residue to the thiol ester carbonyl, releases the thiol forming a protein-bound 4-oxoproline residue. The peptide bond between this residue and the preceding amino acid is easily hydrolysed to yield a truncated protein and a C-terminal polypeptide, which is N-terminally blocked␣by␣a pyroglutamyl residue . To our knowledge, the propionate CoA-transferase is the first CoA-transferase that has been partially purified in this catalytic intermediate form.
Table 1. Purification of propionate CoA-transferase from C. propionicum. The enzyme Activity was measured as described earlier . One unit of propionate CoA-transferase activity corresponds to the formation of 1 µmol acetyl-CoA per min from propionyl-CoA (100 µm) and acetate (200 mm) at 25 °C. Note that the activity in the cell free extract is the sum of CoA-transferase and phosphotransacetylase and that the latter enzyme is completely separated from the CoA-transferase by the first column.
Total activity (U)
Specific activity (U·mg−1)
Purification factor (fold)
Cell free extract
The protein exhibited rather broad signals for the single and double protonated molecular ions in MALDI-TOF-MS measurements, indicating a molecular mass of 56 607 ± 60 Da. The molecular mass of the transferase increased by 750 Da when the enzyme was incubated in the presence of 100 µm propionyl-CoA prior to the measurements, indicating the formation of an enzyme CoA-thiol ester as a covalent catalytic intermediate.
Cloning and sequencing of the gene encoding propionate CoA-transferase
The enzyme was blotted onto a poly(vinylidene difluoride) membrane and subjected to Edman degradation. Internal peptides were generated by cleavage with trypsin and purified to homogeneity by RP-HPLC. Two of these peptides were sequenced by Edman degradation. A degenerated primer pair was deduced from the N-terminus and from one of these peptides (Table 2), which was used to amplify a 300-bp fragment of genomic DNA from C. propionicum by PCR. The fragment was cloned into a TOPO TA vector and sequenced. The sequence was in accordance with the amino-acid sequences used for primer deduction and was similar to other CoA-transferases in the databases. A labelled PCR-product was used to screen a library of genomic DNA from C. propionicum in a λ-ZAP-Express vector. Two clones were isolated and excised from the phage to yield the corresponding plasmid, which was subsequently sequenced. The clones contained identical inserts of 2.7-kb and encoded the complete 524 amino-acid ORF corresponding to the propionate CoA-transferase (Fig. 4). According to the amino-acid sequence, an average molecular mass of 56 553 Da and an isoelectric point of 4.91 was predicted for the encoded protein. Both values were in agreement with the observed molecular mass and the elution of the enzyme during the final purification step.
Table 2. N-terminal sequencing and PCR-primer deduction. The N-terminal amino acid sequences of the purified, blotted protein and of two internal peptides are shown. These sequences have been used to deduce a degenerated primer pair for amplification of a propionate CoA-transferase specific probe from C. propionicum genomic DNA. M = A or C, N = A, G, C or T, R = A or G, H = A, C or T, Y = C or T.
Deduced PCR primer
In addition to the CoA-transferase gene, a second ORF (lcdB), encoding 122 C-terminal amino acids of a protein similar to the 2-hydroxyglutaryl-CoA dehydratase β subunit of A. fermentans, was detected upstream of the transferase gene. This gene probably encodes one subunit of the (R)-lactoyl-CoA dehydratase required in the reductive pathway from alanine to propionate. Directly downstream of the propionate CoA-transferase gene (pct), an AT-rich region was found that resembles rho-independent termination signals. The nucleotide sequence of the full insert has been deposited under the accession number AJ276553 in the EMBL nucleotide sequence database.
The amino-acid sequence of propionate CoA-transferase was compared to other proteins in the database using the blast algorithm [24,25]. The protein was most similar to a putative acetoacetate:acetyl-CoA CoA-transferase from B. halodurans (B84137, 56% identity, 519 amino-acid overlap ), and hypothetical proteins from E. coli (E85777 ydiF, 45% identity, 519 amino-acid overlap ), Aeropyrum pernix (D72478, 38% identity, 541 amino-acid overlap ), and Staphylococcus aureus (F89786, 36% identity, 519 amino-acid overlap ). Other hits were CoA-transferases from various microorganisms including Deinococcus radiodurans, Bacillus subtilis, Streptomyces coelicolor, Heliobacter pylori, Mycobacterium tuberculosis,␣Haemophilus influenzae and Clostridium acetobutylicum. These latter enzymes belong to the 3-oxoadipate CoA-transferase protein superfamily and consist of two different subunits. The similarity of these latter sequences to propionate CoA-tansferase was lower (23–28%) and restricted to the larger subunit of these enzymes (232–255 amino-acids overlap). However, when the C-terminal half of the amino-acid sequence of propionate-CoA-transferase was used for database search, this part of the sequence showed similarity to the smaller subunits of the latter enzymes.
The catalytic glutamate residue of hetero-oligomeric enzymes belonging to the 3-oxoadipate CoA-transferase superfamily is found in the small subunit and is located within a so-called (S)ENG motif . This characteristic motif is not found in the sequence of propionate CoA-transferase or in any of the putative proteins from B. halodurans, E. coli, A. pernix or S. aureus. Therefore, a multiple sequence alignment of the β subunits of 3-oxoadipate CoA-transferase and the C-terminal half (starting with Leu276) of propionate CoA-transferase from C. propionicum was generated using clustalw. The most likely candidate for the catalytic glutamate of propionate CoA-transferase based on these data was glutamate 324 (Fig. 3).
Detection of glutamate 324 as the catalytic carboxylate of propionate CoA-transferase
As the sequence analysis did not allow an unequivocal identification of the catalytic glutamate of propionate CoA-transferase, this residue was specifically labelled. The thiol ester in the proposed enzyme-CoA intermediate of the CoA-transferase reaction cycle is more reactive than, for example, free propionyl-CoA. This higher reactivity allows the reduction of the thiol ester with sodium borohydride to yield a protein-bound 2-amino-5-hydroxyvaleryl residue . In addition, it has been shown that nucleophiles such as methylamine or hydroxylamine can cleave enzyme-bound thiol esters , yielding N-methylglutamine or the corresponding hydroxamic acid. These reactions are useful tools for identifying the catalytic residue, as the derivatives give rise to changes in the molecular masses of peptides that originate from the protein inactivated by either borohydride (−14 Da) or hydroxylamine (+15 Da).
When propionate CoA-transferase was incubated with 100 µm propionyl-CoA in the presence of either sodium borohydride (20 mm) or hydroxylamine (pH 7.5, 200 mm), the enzyme was rapidly and irreversibly inactivated. The inactivation was strictly dependent on the presence of propionyl-CoA.
The inactivated proteins and controls, which had been incubated with the reagents but without propionyl-CoA, were subjected to reductive carboxymethylation and desalted by gel filtration. Aliquots of these samples were digested for 16 h at 37 °C with 2% (w/w) of either chymotrypsin, endoprotease-AspN, endoprotease-GluC or trypsin. The peptides were analysed by MALDI-TOF-MS without purification. Although only around 30–50% of the predicted peptides were detected in one particular digest, all four samples together covered the full amino-acid sequence predicted by the gene.
The molecular masses of peptides carrying the proposed catalytic glutamate 324 were found in all samples except the endoprotease-GluC digest. The masses of these peptides exclusively showed differences for inactivated samples and controls. As summarized in Table 3, the derivatives showed the predicted mass differences of −14 Da and +15 Da for sodium borohydride and hydroxylamine inactivated enzyme, respectively. In particular the observation of a chymotryptic peptide comprising amino acids 322–338 was very crucial, since this peptide contains the glutamate 324 as the sole carboxylate. Therefore, the tentative assignment of glutamate 324 has been confirmed by these experiments.
Table 3. Mass spectrometry of peptides derived from controls and inactivated propionate CoA-transferase. Propionate CoA-transferase was incubated in the absence (control) or presence of propionyl-CoA with either hydroxylamine (200 mm) or sodium borohydride (20 mm) as described in Material and methods. The protein was reductively carboxymethylated and subsequently digested with the proteases as indicated. The peptides were analysed by MALDI-TOF-MS as described elsewhere . The molecular masses expected for the controls (E324 = COOH), hydroxylamine- (E324 = CONHOH) or borohydride-treated (E324 = CH2OH) peptides is given in parentheses. The catalytic glutamate residue is shown in bold. Unless stated otherwise, the measured and calculated molecular masses are given as monoisotopic masses. For trypsin, both calculated and measured values refer to the average mass.
Propionate CoA-transferase from C. propionicum has been purified and initially characterized previously . In this communication we report an improved purification protocol for the enzyme. The gene encoding the protein was cloned, sequenced and glutamate 324 was identified as the active site glutamate residue.
The gene encoding propionate CoA-transferase from C. propionicum was cloned and sequenced. The encoded protein was similar to CoA-transferases belonging to the 3-oxoadipate CoA-transferase superfamily. However, whereas these proteins consist of two subunits and contain a highly indicative fingerprint motif [(S)ENG] surrounding the active centre glutamate, propionate CoA-transferase consists of one polypeptide and the fingerprint motif is not found in this protein. A site-specific label of the catalytic glutamate via the thiol ester catalytic intermediate, either by reductive cleavage with borohydride or by cleavage with hydroxylamine allowed the identification of the active site carboxylate. The predicted derivatives were located exclusively on glutamate 324, which led us to conclude that this residue is the active site carboxylate.
The proteins most similar to propionate CoA-transferase in the databases are a putative acetoacetate CoA-transferase from Bacillus halodurans and other proteins with as yet unknown function from Escherichia coli, Aeropyrum pernix and Staphylococcus aureus. Our data strongly suggest that these genes encode CoA-transferases. As shown in Fig. 4, the proteins align well, and in particular the glutamate residue 324 is conserved among all these proteins. It seems therefore reasonable to conclude that these proteins form a novel subclass of CoA-transferases. These enzymes lack the characteristic (S)ENG consensus motif of members of the 3-oxoadipate CoA-transferase superfamily and exhibit either a homooligomeric or monomeric quarternary structure. It is reasonable to suggest that a gene fusion could have occurred during the evolution of the former enzymes. Such a natural gene fusion has also been suggested for the mammalian oxoadipate CoA-transferase . In agreement with this proposal, it has been shown that the two subunits of glutaconate CoA-transferase from A. fermentans could be fused with genetic tools to yield an active enzyme composed of a single polypeptide .
Despite the low sequence similarities of different CoA-transferases on the amino acid sequence level, CoA-transferases have been predicted to have a very similar fold . In agreement with this proposal, we found that the secondary structural elements predicted for the amino-acid sequence of propionate CoA-transferase superimpose very well with the known elements in the crystal structure of glutaconate CoA-transferase (Fig. 5). Nevertheless, there are some striking differences in the arrangement of the secondary structural elements, which can partially be explained by the known biochemical properties of the enzymes. As shown in Fig. 5A,B, the secondary structure elements in glutaconate CoA-transferase form two outer layers of α helices followed by one layer of β sheets and an inner layer of α helices. This arrangement is found in both subunits of the protein. It is remarkable that the outer layer of α helices in the large subunit of glutaconate CoA-transferase (Fig. 5A) is apparently missing in propionate CoA-transferase (Fig. 5C). The crystal structure has shown that two antiparallel β sheets (Fig. 5B, triangles 6 and 7, respectively) of the β subunit of glutaconate CoA-transferase protrude into a cleft on the surface of the α subunit and are involved in mediation of subunit interactions. Therefore, the lack of this element (Fig. 5C) is not surprising when the homooligomeric structure of propionate CoA-transferase is taken into account. An additional interesting difference between both structures is the lack of a subdomain formed by two β sheets and one α helix connecting the β sheet 2 of the β subunit, which carries the catalytic glutamate, with the inner layer of α helices (Fig. 5B, triangles 3 and 4 and circle 3, respectively). This subdomain is thought to be involved in the substrate binding of glutaconate CoA-transferase. It has been suggested that two specific serine residues, Ser78 (in subunit A) and Ser68 (in subunit B), are involved in the formation of hydrogen bonds with the ε-carboxylate of glutaryl-CoA and that the latter serine is located within this subdomain. It is remarkable that both residues are apparently replaced by stretches of rather hydrophobic residues in propionate CoA-transferase and it is reasonable to conclude that the additional subdomain in glutaconate CoA-transferase represents an adaptation for the binding of a dicarboxyl-CoA by the latter enzyme.
During the course of our research, an interesting observation was made. All attempts to express the cloned gene from C. propionicum in E. coli failed (A. Willanzheimer, unpublished observations); the transformed E. coli cells carrying an isopropyl thio-β-d-galactoside-inducible expression vector exhibited no growth defect until the protein was induced by isopropyl thio-β-d-galactoside. Upon induction of the protein, E. coli stopped growing. These observations may point to a severe impairment of the cellular metabolism of E. coli by the C. propionicum enzyme. Although the reason for this impairment of growth for the host has not been elucidated as yet, a likely explanation could be the formation of lactoyl-CoA and other short-chain fatty acid CoA-thiolesters by the enzyme. Such reactions are predicted to significantly lower the intracellular acetyl-CoA pool. Therefore, growth inhibition by the formation of products, which interfere with essential metabolic pathways , may occur. However, further experiments will be required to establish the reason for the observed growth defects.
We are very grateful to W. Buckel for his constant support throughout this project. K. Neifer and B. Schmidt from the Zentrum für Molekulare Biologie und Biochemie of the George-August University, Göttingen, we thank for protein sequencing. This work was supported by grants from the Deutsche Forschungsgemeinschaft and the Fonds der Chemischen Industrie.