Mutational analysis of the C-domain in nonribosomal peptide synthesis Veit Bergendahl*, Uwe Linne and Mohamed A. Marahiel Biochemie/Fachbereich Chemie, Philipps-Universita È t Marburg, Germany The initial condensation event in the nonribosomal biosyn- thesis of the peptide antibiotics gramicidin S and tyrocidine A takes place between a phenylalanine activating racemase GrsA/TycA and the ®rst proline-activating module of GrsB/ TycB. Recently we established a minimal in v itro model system for NRPS with recombinant His 6 -tagged GrsA (GrsA Phe -ATE; 127 kDa) and T ycB1 (TycB1 Pro -CAT; 120 kDa) and demonstrated the catalytic function of the C-domaininTycB1 Pro -CAT to form a peptide bond between phenylalanine and proline during diketopiperazine formation (DKP). In this work we took advantage of this system to identify catalytically important residues in the C-domain of TycB1 Pro -CAT using site-directed mutagenesis and peptid e mapping. Mutations in TycB1 Pro -CAT of 10 strictly conserved residues among 80 other C-domains with potential catalytic function, revealed that only R62A, H147R and D151N are impaired in peptide-bond formation. All o ther mutations led to either unaected (Q19A, C154A/S, Y166F/W and R284A) or insoluble proteins (H146A, R67A and W202L). Although 100 n M of theserineproteaseinhibitorsN-a-tosyl- L -phenylalanylchlo- romethane or phenylmethanesulfonyl ¯uoride completely abolished DKP synthesis, no covalently bound inhibitor derivatives in the C-domain could be identi®ed by peptide mapping using HPLC-MS. Though the results do not reveal a particular mechanism for the C-domain, they exhibit a possible way of catalysis analogous to the functionally related enzymes chloramph enicol acetyltransferase and dihydrolipoyl transacetylase. Based on this, we propose a mechanism in which one catalytic residue (H147) and two other structural residues (R62 and D151) are involved in amino-acid condensation. Keywords: nonribosomal peptide synthesis; nonribosomal peptide synthetases; peptide synthetases; condensation domain; chloramphenicolacetyltransferase. A broad range of organisms utilize nonribosomal peptide synthesis to produce an immense spectrum of bioactive peptides (antibiotics, siderophores, biosurfactants and immunosuppressants, as well as antitumor and antiviral agents). For that purpose, they avail themselves a large number of amino and carboxy acids as substrates. The biosynthesis of these pharmacological signi®cant agents is performed by nonribosomal peptide synthetases (NRPS), which in their modular organ ization are related to poly- ketide synthases ( PKS) [1,2]. These large multifunctional enzymes are arranged in assembly lines with specialized units completely equipped for the correct activation and incorporation of a single substrate. Such catalytic units, referred to as m odules, are composed of functionally speci®c and independent domains, each of them responsible for catalyzing one single reaction. Remarkably, the order of the modules (with a repetitious assembly of domains) is predominantly colinear to the ®nal product [3,4]. The C-domain, a 450-amino-acid expanding region at the N-terminus of each elongating module, was attributed after extensive sequence analysis with the condensation activity. It was previously con®rmed to be responsible for the catalysis of peptide bond f ormation by the development of a minimized system representative for this family of enzymes [5]. Furthermore r ecent ®ndings indicate that the C-domains are bearing signi®cant substrate selectivity for the nucleo- philic accepto r amino acid and a n enatiosele ctivity for the electrophilic donor substrate [6]. The inherent selectivity at the accep tor site has been shown t o p revent internal misinitiation of the biosynthetic process and to control the timing of substrate epimerization [7]. Recognition and activation of the substrate amino acid are facilitated by the A-domain [8] through carboxy adenylation of the substrate. Hence, the substrate selectivity of the A-domains [9] simultaneously determines the primary sequence of the product. Subsequently the a ctivated amino acid i s tethered to the terminal thiol moiety of a 4¢-phosphopantetheinyl (Ppant) group [10,11]. This Ppant-cofactor itself is post- translationally transferred to a conserved serine residue of the T-domain a lso designated a s PCP (peptidyl c arrier protein) by a special class of CoASH binding 4¢-Ppant- transferases [12±14]. Besides these three domains (C-A-T), which are essential for a functional elongating m odule, there are some optional domains for further modi®cation of t he Correspondence to M. A. Marahiel, Biochemie/Fachbereich Chemie, Philipps-Universita È t Marburg, H ans-Meerwein-Straûe, 35032 Mar- burg, Germany. Fax: + 49 6421 2822191, Tel.: + 49 6421 2825722, E-mail: marahiel@chemie.uni-marburg.de Abbreviations: A-domain, adenylation domain; C-domain, condensa- tion domain, DKP, D -Phe- L -Pro-diketopiperazine; E-domain, epi- merization domain; IPTG, isopropyl thio-b- D -galactoside; LSC, liquid scintillation counting; NRPS, nonribosomal p eptide synthetases; PKS, polyketide synthases; Ppant, 4¢-phosphopantetheine; PP i ,inor- ganic pyrophosphate; T-domain, thiolation domain (also described as PCP, peptidyl-carrier-protein). *Present address: McArdle Laboratory for Cancer Research, Univer- sity of Wisconsin, Medical School, 1400 University Avenue, Madison, WI 53706, USA. (Received 16 August 2001, revised 15 November 2001, accepted 20 November 2001) Eur. J. Biochem. 269, 620±629 (2002) Ó FEBS 2002 substrate amino acids. Those domains are the epimerization domain (E-domain) for C a -epimerization [15], the methyl- transferase domain (M-domain) for N-methylation [16] and the cyclization domain (Cy-domain) [17]. These latter domains are related to the C-domains, as they catalyze the simultaneous condensation and heterocyclization of two aminoacyl or p eptidyl s ubstrates. Release of the ®nal product is catalyzed by a thioesterase (Te)-like domain found at the C-terminal t erminating module of NRPSs templates [18,19]. The enormous size and complexity of most peptide synthetases (up to 1.6 MDa [20]) have signi®cantly restrained a more detailed study of the C-domain f unction in the past. Therefore, we previously established a mini- mized NRPS in vitro system [5], which comprises of the initiation module GrsA Phe -ATE (phenylalanine-activating module; A-, T- and E-domain) from the gramicidin S system and the ®rst module i n the second peptide synthetase of the tyrocidine A system TycB1 Pro -CAT (proline-activat- ing module; C-, A-, and T-domain; see Fig. 1). Both proteins can be obtained in active form as recombinant His 6 -tag fusions by overexpression in E. coli [5]. By applying previously described in vitro assays, it is now possible to monitor the condensation of their cognate substrates L -Phe and L -Pro, and the presumably uncatalyzed intramolecular cyclization that ends up in the release of the cyclic dipeptide D -Phe- L -Pro-diketopiperazine (DKP; see Fig. 2). The same system was also utilized to demonstrate that C-domains possess an in trinsic editing function for the incoming aminoacyl moiety [6]. By using aminoacyl-S-CoA as probes it was shown that the ®rst C-domain of the tyrocidine synthetase complex possesses an enantioselectivity a t the electrophilic donor s ite ( D -Phe) and a s ubstrate speci®city at the nucleophilic acceptor site ( L -Pro) in the formation o f t he chain-initiating D -Phe- L -Pro dipeptidyl intermediate. The knowledge about the architecture that creates this selectivity and the residues, which are involved in catalysis of peptide- bond formation remained largely unclear. Sequence analysis revealed a highly conserved motif HHxxxDGx(S /C), commonly called ÔHis-motifÕ,thatwas suspected to participate in the catalysis. This hypothesis was supported by the ®nding that a single mutation of the Fig. 1. Genes and domain organizations of gramicidin S and tyrocidine A synthetases. These genes lead to the minimal syste m com- prising the NRP Ss G rsA Phe -ATE (from grsA) and TycB1 Pro -CAT (from tycA). Domains are depicted to illustrate the module structure. The amino-acid selectivity of each A-domain is indicated using the three-letter code. TycA and GrsA are highly s imilar to each other and can be used interchangeably, just as GrsB1 and TycB1. Fig. 2. Currently accepted model of the con- densation of L -Phe and L -Pro as catalyzed by the peptide synthetases GrsA Phe -ATE and TycB1 Pro -CAT. The crucial known step s of peptide bond formation in NRPSs are illus- trated for the current system using the texture code of Fig. 1. First the substrate amino acids are activated under ATP hydrolysis as an adenylate and then enzyme bound on the PpantmoietyofeachT-domainasa thioester. L -Phe-S-Ppa nt is then epimerized by the E-domain of GrsA Phe -ATE, but only D -Phe-S-Ppant undergoes condensation with L -Pro-S-Ppant presented at the C-do- main of TycB1 Pro -CAT. Free GrsA Phe -ATE can now be reloaded w hereas D -Phe- L -Pro- DKP can be released by intramolecular cycli- zation on TycB1 Pro -CAT. Ó FEBS 2002 Catalysis of nonribosomal peptide-bond formation (Eur. J. Biochem. 269) 621 second histidine residue (italic) in TycB1 Pro -CAT (H147V) was suf®cient to abolish peptide bond and DKP formation with GrsA Phe -ATE [5]. Furthermore, a chromosomal point mutatio n in srfA-B, changing a D to A in the His-motif (italic) of the corresponding Asp domain of the surfactin synthetase was found to abolish surfactin production in the B. subtilis producer strain [21]. Accordingly, the pres- ence of a catalytic diad or triad [22] was discussed on the basis of sequence analysis and in a proposed analogy to the family of dihydrolipoyl transacetylases or chloramphenicol acetyltransferases [23,24]. Among the questions about C-domain function are (a) what (additional) residues are important for catalysis of peptide-bond formation, (b) how are these residues arranged in the catalytic center, and (c) whether the growing aminoacyl (or peptidyl) donor is getting (at least at same point) covalently tethered to the C-domain? To address these questions, mutants in residues conserved across 80 NRPS C-domains were constructed in the C-domain of TycB1 Pro -CAT and assayed for their ability to catalyze peptide bond formation between D -Phe- S-Ppant-GrsA Phe -ATE and L -Pro-S-Ppant-TycB1 Pro - CAT. Furthermore, the minimal system GrsA Phe -ATE/ TycB1 Pro -CAT was compromised with two inhibitors (phenylmethanesulfonyl ¯uoride and N-a-tosyl- L -phenyl- alanylchloromethane), and analyzed for the appearance of possible covalent C-domain/inhibitor complexes in order to ®nd a catalytic triade similar t o the one in serine proteases. EXPERIMENTAL PROCEDURES Sequence alignments for the identi®cation of highly conserved residues The sequences of more than 80 C-domains we re retrieved from publicly accessable databases (NCBI, SwissProt, etc.). Sequences used were derived from biosynthesis systems of gramicidin S (Grs), tyrocidin A (Tyc), s urfactin (Srf), lichenysin (Lic), bacitracin ( Bac), fengycin (Fen), ente ro- bactin (Ent), chloroeremomycin (Cep), pristinamycin (Snb), enniatin (Esyn), HC-toxin (Hts1), penicillin (Acv) and cyclosporin (SimA). After outlining the  450-amino-acid stretches, the sequences were aligned using the program MEGALIGN from the DNA Star package, applying the Jotun Hein algorithm with default parameters. Site-directed mutagenesis and cloning All TycB1 Pro -CAT mutants were constructed by site- directed mutagenesis of pProCAT [5] using either the so- called ÔMegaprimer-PCRÕ method [25] with the Expand long-range PCR system TM (Bo È hringer Mannheim, Germa- ny) or t he QuickChange TM Site-Directed Mutagenesis K it (Stratagene). Restriction sites f or subsequent cloning and screening were i ntroduced with PCR oligonucleotides (MWG-Biotech, Germany) summarized in Table 1. Quick- Change TM mutagenesis was carried out in accordance with the m anufacturer's protocol. For the ÔMegaprimer-PCRÕ, PCR products were puri®ed with the QIAquick-spin PCR puri®cation kit (Qiagen, Germany), digested with NcoIand ligated. Standard procedures were applied for all DNA manipulations [26] and the Escherichia coli strain XL1-Blue [27] was used for cloning. The mutant TycB1 Pro - CAT(H147V) was obtained by site-directed mutagenesis as described pre viously [5]. Together with the desired point mutations, additional silent mutations were introduced in order to g enerate a new restriction site, which allowed a simple detection of all mutated plasmids. The fusion sites between vector and insert, as well as th e mutation-site of the plasmids pProCAT were con®rmed by DNA sequencing using an ABI-Prism 310 Genetic Analyzer with standard protocols described by ABI (Applied Biosystems, Germany). Expression and puri®cation of functional holo-peptide synthetase fragments To achieve in vivo production of functional holo-peptide synthetase modules, the Ppant-transferase gene gsp was coexpressed with the pQE60 plasmids carrying the DNA fragments for the TycB1 Pro -CAT mutants. Expression and puri®cation using single-step Ni 2+ -af®nity chromatography was performed according to previously published proce- dures [5]. Purity of the proteins was judged by SDS/PAGE (data not shown). Fractions containing the recombinant proteins were pooled and dialyzed against assay buffer (50 m M Hepes, pH 8.0, 100 m M sodium chloride, 10 m M magnesium c hloride, 2 m M dithioerythritol and 1 m M EDTA). After addition of 10% glycerol (v/v), the proteins could be stored at )80 °C. Protein concentrations were determined using the calculated extinction coef®cients for the A 280 of the proteins: 138 690 M )1 ácm )1 for GrsA Phe - ATE and 92 230 M )1 ácm )1 for TycB1 Pro -CAT and its mutants . ATP-pyrophosphate exchange assay The ATP-pyrophosphate exchange reaction was carried out to examine the adenylation activity of all recombinant peptide synthetase fragments puri®ed. Reaction mixtures contained (®nal volume: 100 lL): 50 m M Hepes, pH 8.0, 100 m M sodium chloride, 10 m M magnesium chloride, 1m M EDTA, 1 m M amino acid and 300 n M enzyme. The reaction was i nitiated by the addition of 2 m M ATP, 0.2 m M tetrasodium pyrophosphate and 0.15 lCi of tetrasodium [ 32 P]pyrophosphate (NEN/DuPont) a nd incubated at 37 °C for 10 min. Reactions were quenched by adding 0.5 m L of a stop mix c ontaining 1.2% (w/v) activated charcoal, 100 m M tetrasodium pyrophosphate and 3 50 m M perchloric acid. Subsequently, the charcoal was pelleted by centrifugation, washed once with 1 mL water and resuspended in 0.5 mL water. After addition of 3.5 mL of liquid scintillation ¯uid (Rotiscint Eco Plus; Roth, Germany), the charcoal-bound radioactivity was determined by liquid scintillation c ounting (LSC) with a 1900CA Tri-Carb liquid scintilliaton analyzer (Packard). Thioester formation: radioassay for the detection of covalent amino-acid incorporation Reaction mixtures in assay buffer (50 m M Hepes, pH 8.0, 100 m M sodium chloride, 10 m M magnesium chloride, 1m M EDTA) contained 500 n M enzyme, 2 m M ATP and 2 l M [ 14 C]-amino acid (Hartmann, Germany). The thio- esteri®cation was initiated upon the addition of the radio- labeled amino acid and the reaction was quenched after 622 V. Bergendahl et al. (Eur. J. Biochem. 269) Ó FEBS 2002 10 min by the addition of 1 mL chilled 10% (w/v) trichloroacetic acid and incubated on ice for 15 min. The precipitates were pelleted by centrifugation (4 °C, 16 000 g) and washed two times with 1 mL 10% trichloroacetic acid (w/v). The pellet containing the acid-stable label was then dissolved in 150 lL formic acid and quanti®ed by LSC as described above. Method of normalization for values in condensation assays A common margin of error is made when determining a protein concentration b y the calculated extinction coef®- cient. According to our experience, the current set o f proteins has s hown that thioesteri®cation activities may vary signi®cantly (up to threefold) depending on the b atch of protein utilized in the assays. I n the present work, we tried to tak e this behaviour into consideration by normalizing the values with the thiolation activ ity as an internal standard for amount of active protein. We normalized the values obtained in all the assays for dipeptide formation a nd for DKP production by multiplying the counts in the radioac- tive assays and the area in the HPLC-assays by the ratio of counts in the thiolation assay of mutant over wild-type. Values in the elongation a ssay were expressed as relative values to the value at t  0 which was set as 100%. DKP amounts were also expressed a s relative values (percent of wild-type value). Radio assay for the detection of elongation 500 n M of holo-enzyme (GrsA Phe -ATE and TycB1 Pro - CAT) were preincubated seperately in assay buffer with their substrate amino acids [2 l M [ 14 C] L -Phe (450 mCiám- mol )1 ), 100 l ML -Pro] and ATP ( 2 m M ). After 3 min, product formation was initiated by mixing equal volumes of reaction mixtures. At various time-points, 200 lL aliquots were taken and immediately quenched by addition of 1 mL ice-cold trichloroacetic acid (10%). After 15 min on ice, samples were centrifuged (4 °C, 16 000 g)for 20 min, w ashed two times with 1 mL ice-cold trichloro- acetic acid, redissolved in 150 lL formic acid and quan- ti®ed by LSC. DKP formation: indirect assay for D -Phe- L -Pro dipeptide formation The formation of the dipeptide was analyzed using G rsA Phe - ATE and the different C-domain mutants of TycB1 Pro - CAT. To ensure a complete acylation of the p eptide synthetase fragments with their cognate amino acids, a preincubation was carried out for 3 min at 37 °C; 1 l M GrsA Phe -ATE was incubated in assay b uffer containing 2m M ATP and 0.5 M phenylalanine (mixture A), and TycB1 Pro -CAT mutants were incubated in assay buffer containing 1 l M enzyme, 2 m M ATP a nd 0.5 m M proline (mixture B). For reactions with N-a-tosyl- L -phenylalanyl- chloromethane and phenylmethanesulfonyl ¯uoride the inhibitor was suspended in ethanol and added to mixture B without exceeding 1% ethanol content in the reaction mixture. The condensation reaction was initiated by the addition of 1 vol. of mixture B to 1 vol. of mixture A and incubated 45 min at 37 °C. In order to analyze the nature of the product(s), the reaction mixture (1 mL) was diluted by adding 4 mL water and immediately extracted with buta- nol/chloroform [4 : 1; (v/v)]. The organic phases w ere Table 1. Primers used for mutagenesis destinguished by the two dierent PCR techniques applied. The restriction sites indicated were introd uced to ease the screening for mutated plasmids after cloning. OP refers to outer primer and MP to megaprim er. Also the desired mutation is ind icated in the name of each primer. (lower case: modi®ed se quences; bold: re striction sites). Name of primer Sequence Restriction enzyme Megaprimer PCR OP1 5¢-cat gCC ATG GGT GTA TTT AGC-3¢ NcoI OP2 5¢-cat gCC ATG GTT AAT TTC TCC TCT TTA ATG-3¢ NcoI MP(D151N) 5¢-GAA GCA CCA GCC GTt CAT GAG GAT GTG-3¢ BspHI MP(C154S) 5¢-AAT GCT GAA GgA CCA GCC GTC CAT GAG-3¢ AvaII MP(C154A) 5¢-AAT GCT GAA tgc CCA GCC GTC CAT GAG-3¢ BsmI MP(H146A) 5¢-CCA TGA GGA TGT Gcg cAA AGC TCC A-3¢ FspI MP(H147R) 5¢-CCA TGA GGA Tcc GAT GAA AGC TCC A-3¢ BamHI MP(Y166W) 5¢-GCA AGG ACA Acc AtA TGG CAA GCA AGT C-3¢ NdeI MP(Y166F) 5¢-GCA AGG ACA Aga AgA TGG CAA GCA AGT C-3¢ QuickChange TM 3¢Q19A 5¢-GCG TTG ACC CCG ATG gcc GAG GGG ATG CTG TTT CAC-3¢ CfrI 5¢Q19A 5¢-GTG AAA CAG CAT CCC CTC ggc CAT CGG GGT CAA CGC-3¢ CfrI 5¢R62A 5¢-CTG CAT GTG CTG GTA GAG gcc TAC GAT GTA TTC CGC ACG-3¢ AatI 3¢R62A 5¢-CGT GCG GAA TAC ATC GTA ggc CTC TAC CAG CAC ATG CAG-3¢ AatI 5¢R67A 5¢-GGT AGA GAG ATA CGA TGT ATT Cgc gAC GTT GTT TAT CTA TGA AAA GC-3¢ NruI 3¢R67A 5¢-GCT TTT CAT AGA GAA ACA ACG Tcg cGA ATA CAT CGT ATC TCT CTA CC-3¢ NruI 5¢W202L 5¢-CAG GCC GCT CTC AAC TAc Tta AGC GAC TAT CTG GAA GCC-3¢ A¯II 3¢W202L 5¢-GGC TTC CAG ATA GTC GCT taA gTA GTT GAG AGC GGC CTG-3¢ A¯II 5¢R284A 5¢-GGC TCT GTT GTA TCC GGA gct CCT ACA GAC ATC GTC GG-3¢ SacI 3¢R284A 5¢-CCG ACG ATG TCT GTA GGa gcT CCG GAT ACA ACA GAG CC-3¢ SacI Ó FEBS 2002 Catalysis of nonribosomal peptide-bond formation (Eur. J. Biochem. 269) 623 transferred to fresh tubes and washed once with 5 mL of 0.1 M sodium chloride. A fter removal of the solvent under vacuum, the reminders of each reaction were further investigated by HPLC or HPLC-MS. The samples prepared were resolved in 200 lL 10% of buffer B and the products separated by using a C18 reversed-phase column (Nucleosil 3mm´ 250 mm, pore-size 120 A Ê , particle-size 3 lm; Macherey & N agel) o n a HP1100 HPLC-MS system (Agilent Technologies) with simultaneous monitoring at detector wavelengths of 214 and 256 nm. The following gradient pro®le was used at a ¯ow-rate of 0.35 mLámin )1 : loading (10% buffer B), linear gradient to 30% buffer B in 1 m in, followed by a linear gradient to 100% buffer B in 20 min, and then holding 100% buffer B for 10 min (buffer A, 0.05% formic acid in H 2 O; buffer B, 0.04% formic acid in methanol). Peptide mapping of trypsin-digested TcyB1 Pro -CAT Peptide mapping of TycB1 Pro -CAT was performed accord- ing to the manufacturer's p rotocol using the Sequencing Grade Modi®ed T rypsin Kit (Promega). Treatment of protein with N-a-tosyl- L -phenylalanylchloromethane or phenylmethanesulfonyl ¯uoride was performed before the digest by incubating a 10-fold excess of inhibitor with TycB1 Pro -CAT (1 l M )for10minat25°C. N-a-Tosyl- L -phenylalanylchloromethane and phenylmethanesulfonyl ¯uoride treated protein (1 mg) was precipitated and washed once with aceto ne to eliminate residual inhibitor that might interfere with the digest [28]. HPLC conditions were as described by Promega except for using a HP1100 HPLC-MS system and a C18 reversed-phase column (Nucleosil 3 ´ 250 mm, pore size 120 A Ê ,particlesize3lm; Macherey & Nagel). RESULTS Homology searches to select targets for the site-directed mutagenesis of the C-domain in TycB1 Pro -CAT For a fairly long time, virtually no biochemical d ata were available on C domains, and consequently, this  450- amino-acid stretch was considered to be only a spacer between consecutive, amino-acid-activating A-T bi-domains [29]. Recently, howe ver, it could be demonstrated that this region is actually responsible for catalysis of peptide bond- formation [5]. Furthermore, it was noted that C-domains share the signature sequence motif HHxxxDGxSW (the so-called ÔHis m otifÕ) with a superfamily of acyl transferases, and that the second His and the Asp residues are indispensable for C-domain activity [ 5,21]. To identify additional catalytic key residues, we started our study with alignments of the primary sequences of NRPSs C-domains. A scan of 80 C-domains revealed an overall similarity ranking from 60% (between TycB1 and GrsB1) to < 20% (between TycB1 and Hts1), with an average percentage of similarity of 35% (data not shown). Among all C-domains investigated, we f ound eight absolutely i nvariant residues, o f which all have functionalized side chains (carboxyl, amino, amine, guanidino, sulfhydryl or hydroxyl groups). These latter residues (namely Q19, R62, R67, H146, H147, D151, W202 and R284 within TycB1 Pro -CAT) and additionally C154 and Y166 ( > 95% cons erved and discussed as potentially involved in catalysis) were selected as targets for the subsequent mutational analysis. Generation and puri®cation of the recombinant enzymes In this study, we constructed a set of 12 TycB1 Pro -CAT [5] single mutants (Q19A, R62A, R67A, H 146A, H147R, D151N, C154S, C154A, Y166W, Y151F, W202L and R284A). Mutations other than to alanine were intentionally designed to show residual activity for similar functional groups (H/R, D/N, C /S and Y/W) as opposed to residues with no functionalized group (H/V, C/A and Y/F). In case of a catalytic triade Asp-His-Ser or Asp-His-Tyr similar functionalized groups were expected to exhibit residual activity, whereas unfunctionalized groups would have none. All m utants were individually expressed as C -terminal His6- tag fusions in the heterologous host E. coli and puri®ed by Ni 2+ -af®nity chromatography. As judged by SDS, a ll proteins could be puri®ed to homogeneity (data not shown), although in gene ral, the solu bility of the re combinant proteins appeared not to be very high (as estimated on < 30% by comparison after SDS/PAGE of pellet and supernatant after cell lysis). Highest amounts of soluble protein comparable to wild-type level were obtained for mutants Q19A, R62A, H147V, H147R, C154S and C154A. Preparation of mutants D151N, Y166W, Y151F, R284A revealed slightly lower solubilities. Mutants R67A, H146A and W202L gave less than 0.5 mg soluble protein per L culture indicating a misfolding induced by the mutation. In order to test for correct folding, all mutants were subjected to ATP±PP i exchange assay, which a ssesses the a ctivity and selectivity of the A domain embedded within the C- and T-domains of all t ridomain TycB1 Pro - CAT derivatives. Although A domain activity indicates a correct folding of the entire TycB1 Pro -CAT enzymes, it cannot be excluded that the connected C- and T-domains domains are somehow impaired in folding. The assay revealed that within an error-margin of < 5%, the same wild-type amino-acid dependent activity in the ATP±PP i - exchange assay could be obtained for most mutants, indicating that their structure cannot have changed much if at all. Three TycB1 Pro -CAT mutants, R67A, H146A and W202l sustained a drop in adenylation activity to less than 5% wild-type activity and therefore are likely t o be affected in folding (see below). Dipeptidyl- S -Ppant- and DKP-formation Prior studies revealed that C-domains are the peptide bond- forming catalysts, which raises the question of how ef®ciently the C-domain mutants of TycB1 Pro -CAT t ransfer the D -Phe moiety from GrsA Phe -ATE. To assay for the formation of the dipeptidyl-S-Ppant nascent product, L -Pro was allowed to load onto holo-TycB1 Pro -CAT mutants in the p resence of ATP. Sub sequently, t he L -Pro-S-Ppant- enzymes were mixed, respectively, with holo-GrsA Phe -ATE, which had been loaded in a preincubation with ATP and radiolabeled L -[ 14 C]Phe. Once translocation of D -[ 14 C]Phe from GrsA Phe -ATE to TycB1 Pro -CAT occurs by C-domain catalysis, the vacant holo-GrsA Phe -ATE is rapidly reloaded with surplus L -[ 14 C]Phe. Thus, if samples are taken at de®ned time points and immediately quenched b y the 624 V. Bergendahl et al. (Eur. J. Biochem. 269) Ó FEBS 2002 addition of 10% (w/v) trichloroacetic acid, the measurable amount of acid-stable label should increase f rom one equivalent to two equivalents, as at that poin t GrsA Phe -ATE and TycB1 Pro -CAT are both radiolabeled. Simultaneously, the D -[ 14 C]Phe- L -Pro dipeptide is slowly autoreleased from TycB1 Pro -CAT by cyclization to DKP. Consequently, as soon as all L -[ 14 C]Phe has been consumed, the amount of acid-stable label will decrease again until all enzyme has been liberated of radiolabeled substrate. This assay for dipeptide and DKP formation was performed w ith wild-type GrsA Phe -ATE an d all TycB1 Pro - CAT C-domain mutants except mutants R67A, H146A and W202L. The latter ones were not tested further, as they proved to be insoluble o r misfolded. As positive a nd negative controls we performed the assay with wild-type TycB1 Pro -CAT instead of mutants and without TycB1 Pro - CAT [5]. On the basis o f the results obtained by LSC (summarized in Fig. 3), the TycB1 Pro -CAT mutants can be categorized in three groups. Three mutants (Q19A, Y166W and R284A) revealed no effect on dipeptide-formation and release of DKP (Fig. 3A). A second group (C154A, C 154S, Y166F) was asymmetrically impaired in DKP release, implying that the release of D -Phe- L -Pro-DKP is somehow in¯uenced by the functional side-chain moieties of C154 and Y166 (Fig. 3B). The most interesting group, how- ever, consists of four TycB1 Pro -CAT C-domain mutants (R62A, H147R, H147V and D151N) that are completely impaired in dipeptide formation (Fig. 3C). This i ndicates that the three residues affected are essential for the translocation of D -Phe-S-Ppant from GrsA Phe -ATE to TycB1 Pro -CAT. The product of the condensation assay, DKP, is readily extractable into organic solvent [5]. Thus, for a more detailed analysis of the product(s) formed, the assays were performed with nonradiolabeled substrates. After 45 min, reactions were extracted with butanol/chloroform and analyzed by HPLC-MS. The results of the HPLC-MS analysis, which are summarized in Fig. 4, revealed that D -Phe- L -Pro-DKP was synthesized as the only product in the case of the mutants Q19A, C154A/S, Y166W/F and R284A. The amount of DKP produced ranged from 30 to 100% of the wild-type TycB1 Pro -CAT (normalized for thiolation activity). Changes in turnover are evident for C154S and R284A, but most signi®cantly for Y166F and Y166W, by a decreased a mount of DKP synthesized. In contrast, no DKP at all could be detected when using R62A, H147V/R or D151A. Peptide mapping of N -a-tosyl- L -phenylalanylchloro- methane and phenylmethanesulfonyl ¯uoride treated TycB1 Pro -CAT Possible catalytic mechanisms for the formation of pep- tide bonds would be for example a catalytic triade as in serine proteases or catalytic diade as found in Fig. 3. Time dependence of dipeptide and DKP formation between GrsA Phe -ATE and C-do main mutants of T ycB1 Pro -CAT. Normalized counts are plotted against the time of sampling. The increase in acid- stable label accounts to the translocation of labeled Phe onto TycB1 Pro -CAT and rapid reloading of GrsA Phe -ATE. Both labeled proteins can be precipitated by 10% trichloroacetic acid solution. Mutants Q19A, Y166W and R284A revealed no eect on dipeptide- formation and release of DKP represente d by the plot of Q19A. The second group (C154A, C154S, Y166F), represented by the plot of Y166F, is only impaired in DKP f ormation, implied by the slow decresae of the acid stable label. The most in teresting g roup, however, is represented by the plot of R62A. It consists of four TycB1 Pro -CAT C-domain mutants (R62A, H147R, H147V and D151N) that are signi®cantly impaired in dipeptide formation, as there is no increase in acid stable label which could indicate a formation of a Phe-Pro dipeptide. Ó FEBS 2002 Catalysis of nonribosomal peptide-bond formation (Eur. J. Biochem. 269) 625 chloramphenicolacetyltransferases [23,24]. Most strikingly, all results obtained so far were in f avor of the latter mode of action, and we a lso were not able to detect a conserved serine residue that might act in the catalytic center of C-domains. To further rule out the possibility of a catalytic triade, we n ext investigated the inhibition of dipeptide and DKP formation in the presence of N-a-tosyl- L -phenylalanyl- chloromethane and phenylmethanesulfonyl ¯uoride. Both inhibitors form covalent complexes with their target enzymes via catalytically exposed serine and histidine residues, respectively. We found that the minimal system was susceptible to complete inhibition of DKP synthesis by 100 n M N-a-t osyl- L -phenylalanylchloromethane and phen - ylmethanesulfonyl ¯uoride (data not shown). Consequently, we next performed a partial digest with trypsin of TycB1 Pro - CAT, which had been pretreated with inhibitor. Tryptic digests were applied to HPLC and analyzed by a coupled MS. Focusing only on tryptic fragments derived from the C-domain, we were able to separate the resulting peptides by reversed-phase chromatography and to trace out theoretical fragments larger than 500 m/z by MS (calculated by peptide tools, Agilent Technologies; data not shown). However, none of them showed an increase in mass that would occur by covalent derivatization with N-a-tosyl- L -phenylalanyl- chloromethane or phenylmethanesulfonyl ¯uoride one would expect in the case of a catalytic triade. No further efforts were made to track down the actual target of the inhibitors. DISCUSSION The purpose of t his work was to identify the key residues in the C-domain a nd their potential role in the catalytic mechanism. For t his purpose, sequence alignments of 80 C-domains were made and lead to 10 invariant residues (100% identity except C154 and Y166, who showed 100% similarity by variation with serine in four cases and phenylalanine in one case). These residues were Q19, R62, R67, H146, H147, D151, C154, Y166, W202 and R284. Figure 5 summarizes the results of the site-directed mutagenesis. Mutations of residues R67, H146 and W202 seemed to be structurally important as judged by their lack of solubility. Residues Q19, C154, Y166 and R284 were found to be not essential to the condensation activity. The condensation reaction requires the action of various domains, so we a re dealing w ith sets o f coupled catalyzed and noncatalyzed reactions (adenylation, thiolation, epi- merization and noncatalyzed cyclization) resulting in c om- plex kinetic i n¯uences. Consequently, a more thorough analysis of the mutants does not seem to be possible with the current methods. In contrast, mutation of residues R 62, Fig. 5. Summary of all observed eects for every TycB1 Pro -CAT mutant examin ed. The diagram shows a schematic presentation of the 450-amino-acid C-domain of TycB1 Pro -CAT, along with the approximate locatio n of core- motifs C1±7 [3]. Underneath, the eects of each mutation on solubility and various activities of each domain are rated by Ô+Õ (soluble or active) and Ô±Õ (insoluble or in- active). A gray background emphasizes the mutants R62A, H147V/R and D151N in particular, as they showed only a loss in con- densation activity, while the other domains appear to be uneected. Fig. 4. Relative amount of DKP generated with GrsA Phe -ATE and C-domain mutants of TycB1 Pro -CAT. The quantity of D KP synthe- sized (determined by HPLC-MS) using wild-type TycB1 Pro -CAT was set 100% and errors were calculated from a series of three consecutive experiments with the same batch of protein. 626 V. Bergendahl et al. (Eur. J. Biochem. 269) Ó FEBS 2002 H147 and D151 were found to abolish C-domain activity completely, while being still active in adenylation a nd thiolation activities. In p revious work, His147 was already found to be essential for condensation [5] and the corresponding residue of D151 in the SrfB2 Asp -CAT surfactin synthetase essential for the synthesis of surfactin in B. subtilis, shown in vivo [21,30]. In the present study we additionally found R62 to be essential for condensation activity. D151 was veri®ed in vitro to be essential in a second C-domain. Previously, based on sequence alignments, i t had been suggested, that residues C154 or Y166 could also be involved in the catalytic mechanism and possibly f orm a catalytic triad [22] analogous to serine proteases (Asp-His- Ser). However, the mutation of these two residues led to enzymes still active in the condensation assay. The C154A mutant showed a D KP-synthesis activity comparable to th e wild-type. The C154S mutant showed a reduced amount of total DKP synthesized when compared to the correspond- ing alanine mutant. In the case of the second potential key residue, Y166, the two mutants constructed (Y166W and Y166F) seem to be affected in the DKP-synthesis, but no signi®cant effect abolishing condensation activity was seen. So we tried to show an inhibitory effect of the typical serine protease inhibitors N-a-tosyl- L -phenylalanylchloromethane and phenylmethanesulfonyl ¯uoride in the condensation assay. However, N-a-tosy l- L -phenylalanylchloromethane and phenylmethan esulfonyl ¯uoride abolished product formation at concentrations of 100 n M ,wewerenotable to identify their target in the C-domain using peptide mapping. T hus the inhibitory effect may be due to interference with other reactive residues in NRPS modules like the thiol g roup of the 4 ¢-phosphopantethein-moiety or the aminoacyladenylate of the activated amino acid. Fur- thermore, N-a-tosyl- L -phenylalanylchloromethane could also act as a competitive inhibitor by competing with Phe in binding to the donor site in the C-domain. As we could not detect any covalent intermediate of N-a-tosyl- L -phenyl- alanylchloromethane or phenylmethanesulfonyl ¯uoride with residues of the C-domain of TycB1 Pro -CAT, an analogous mechanism as observed in serine proteases is not supported by our data. On the contrary, our data excludes the direct involvement o f serine residues i n catalysis of peptide-bond formation as postulated by deCrecy- Lagard et al. [22]. Interestingly, in the case o f serine proteases, a mutation in the catalytic triade (Asp-His-Ser ® Asp-His-Gly) was recently found by Elliott and coworkers to switch the protease activity into a ligase activity [31]. Therefore, a catalytic diad (Asp-His) for condensation domains of NRPSs would be in good agreement with our results and hence we suggest an alternative mechanism. There is a strong analogy to some a cyl transferases, such as the chloramphenicol acetyltransferase o r the dihydrolipoyl transacetylase. It is the participation of a histidine residue in almost the same motif [HHxxx(D/N)G] [23,24]. These enzymes use Acyl-CoA as a substrate in an acylation reaction and hence share t he ability of utilizing a Ppant moiety as an acyl carrier and donor. In the crystal structures of both r epresentatives only the second histid ine (His195 in chloramphenicol acetyltransferase) was observed to play a central role in catalysis and both sho w a very similar architecture (Fig. 6). In chloramphenicol acetyltransferase other residues such as Arg18 and Asp199 were found to be signi®cantly distant to the reactive center with the cocrystallized chloramphenicol and therefore were interpreted as less c atalytically then structurally important, although an electrostatic interaction t o the catalytic center could not have been excluded. It is conceivable to ®nd a similar architecture for the C-domain of peptide Fig. 6. Structure of chloramphenicol acetyl and dihydrolipoyl transacetylase. The horizontal dark gray a helix of th e structure includes the His-motif of which residues His195 and Asp199 are explicitly shown in all pictures. A very similar structure can be f ound in the dihydrolipoyl transacetylase. Especially the structure around the catalytic center may be similar to the structure in the C-domain of TycB1 Pro -CAT. Residues Arg18, His195 and Asp19 are emphasized in the enlarged portion of chloramphenicol acetyltransferase. Arg18 (at the light a helix), which forms a salt bridge with Asp199, can be found at a dierent position in the dihydrolipoyl transacetylase structure and may play a dierent role there. The catalytic centers of chloramphenicol a cetyltransferase and dihydrolipoyl transacetylase i nteract with their substrates bound in the binding poc ket o f a nother e nzyme molecule within the active homomultimer. This could explain the distance in sequence homology with NRPSs C-domains displayed in the alignment of the according regions in chloramphenicol acetyltransferase and dihydrolipoyl transacetylase with the proposed homolog region of four representative C-domains (from tyrocidine, lichenysin and surfactin synthetases), as latter presumably act as in a dierent structual assambly andon dierent substrates. Ó FEBS 2002 Catalysis of nonribosomal peptide-bond formation (Eur. J. Biochem. 269) 627 synthetases. Accordingly, the residues R62 and D151 in TycB1 Pro -CAT could form a similar ion pair as fou nd f or R18 and D199 in chloramphenicol acetyltransferase. There, these residues were found to be Ôstructurally important by virtue of being 8 A Ê of the catalytic HisÕ. Following this analogy His147 could act as a general base. It could be similarly stabilized via a tautomeric stabilization provided by its own carbonyl oxygen from the backbone rather than a side-chain carboxylate group from another residue [ 32], such as D151. A ccordingly, the e ffect of mutations Y166W/F could be interpreted by Y166 being involved in stabilizing the tetrahedral transition state of the thioester during condensation. Perhaps in the mutants a nearby hydroxyl group could restore this function (Fig. 7) or the stabilization is not absolutely necessary. Moreover, no covalently enzyme bound waiting position of the activated a cylsubstrate was f ound. Thus, f or the C-domains of peptide synthetases, no such covalent waiting position should be expected, either. The conserved serine in the crystal structure of c hloramphenicol acetyl- transferase was 4.3 A Ê away from the reactive hydroxyl group of chloramphenicol, ruling out its direct involvemen t in catalysis. Interestingly, two arginine residues were also discussed as possible participants, but wer e found n ot to be suf®ciently near to the catalytic center of the reaction. Chloramphenicol acetyltransferase, dihydrolipoyl transacetylase and TycB1 Pro -CAT show no signi®cant sequence homology between each other, which is also t rue for chloramphenicol acetyltransferase and dihydrolipoyl transacetylase when compared separately. They do though, show signi®cant structural homology, suggesting a common structure±function motif for this type of enzymes. Future structure investigations will have to determine, weather the C-domain also belongs t o this family. The prospect o f using NRPSs to produce new bioactive peptides has inevitably led to the demand for a more detailed understanding of each process involved. Although it is possible to direct the biosynthesis to new derivatives by reprogramming the substrate speci®city of A -domains and reorganizing the assembly of modules, we still have to ®ght with loss of enzyme activity and speci®city. A milestone in the study of peptide s ynthetases was the determination of the A-domain structure of GrsA Phe -A [8]. Aspects of speci®city and binding were superimposed on other A-domains and with the help of eight residues in the primary sequence the substrate can now be predicted and even altered [9]. Recent results in this ®eld have now demonstrated that substrate selectivity in the C-domain can be another obstacle to be overcome on the way to controlled and effective engineering o f peptide synthetases [6,7,33]. In consequence a more detailed analysis of the structurally related functions inside th e C -domain s till demands our efforts for the future and probably requires structural data from crystals. Nevertheless, the data presented in this s tudy excludes the idea of a c atalytic triad involved in C-domain catalysis of NRPSs. With the ®nding of the additional residue R62 and con®rming the residues H147 and D151 abolishing condensation activity of TycB1 Pro -CAT, the idea of a s imilar mechanism to chloramphenicol acetyltransferases, with the direct involve- ment of only one residue (H147) in catalysis, is strongly supported. Fig. 7. Proposed model for the catalytic mechanism at the C-domain. This improved model is based on the proposed analogy to chloram- phenicol acetyltransferase, for which the mechanism has be en estab- lished b y the crystal struc ture and kinetic s tudies. Residue H147 of TycB1 Pro -CAT plays the key role in t he catalytic mechanism and is essential, whereas Y 166 may b e replaced b y an other nea rby hydrox yl group, ac cording to our data. A ccording to the model R62 and D151 play a more structurally important role rather than being involved with the catalytic mechanism (i.e. displaying the backbone carbonyl in position). 628 V. Bergendahl et al. (Eur. J. Biochem. 269) Ó FEBS 2002 ACKNOWLEDGEMENTS We thank Jo È rg Tost and Robert Finking for their valuable contribution during their undergraduate studies and Dr Torsten Stachelhaus for critical reading of th e manuscript. This work was supported by the Deutsche Forschungsgemeinschaft and the Fond der chemischen Industrie. For their reliable te chnical supp ort w e a lso t hank Inge Schu È ler and Gabi Schimp-Weiland. REFERENCES 1. Schwarzer, D. & Marahiel, M.A. (200 1) M ultimodular biocatalysts for natural product assembly. Naturwissenschaften 88, 93±101. 2. Cane, D.E. & Walsh, C.T. (1999) The parallel and convergent universes of polyketide synthases and nonribosomal peptide syn- thetases. Chem. Biol. 6, R319±R325. 3. Marahiel, M.A., Stachelhaus, T. & M ootz, H.D. ( 1997) Modular peptide synth etases invo lved in non-ribosomal peptide synth esis. Chem. Rev. 97, 2651±2673. 4. Konz,D.&Marahiel,M.A.(1999)Howdopeptidesynthetases generate structural diversity? Chem. B iol. 6, R39±R48. 5. Stachelhaus, T., Mootz, H.D., Bergendahl, V. & Marahiel, M.A. (1998) Peptide bond formation in nonribosomal peptide biosyn- thesis. Catalytic role of th e condensation domain. J. Biol. Chem. 273, 22773±22781. 6. Belshaw, P.J., Walsh, C.T. & Stachelhaus, T. (1999) Aminoacyl- CoAs as probes of conden sation domain selectivity in nonribos- omal peptide synthesis. Science 284, 486±489. 7. Linne, U. & Marahiel, M.A. (2000) Control of directionality in nonribosomal peptide synthesis: role of the condensation domain in preventing misinitiation and timing of epimerization. Biochemistry 39, 10439±10447. 8. Conti, E., Stachelhaus, T., Marahiel, M.A. & Brick, P. (1997) Structural basis for the activation of phenylalanine in the non- ribosomal biosynthesis of gramicidin S. EMBO J. 16 , 4174±4183. 9.Stachelhaus,T.,Mootz,H.D.&Marahiel,M.A.(1999)The speci®city-conferring code of adenylation domains in nonriboso- mal peptide synthetases. Chem. Biol. 6, 493±505. 10. Stachelhaus, T., Hu È ser, A. & Marahiel, M.A. ( 1996) Biochemical characterization of peptidyl carrier protein (PCP), the thiolation domain of multifunctional peptide synthetases. Chem. Biol. 3, 913±921. 11. Weber, T., Baumgartner, R., Renner, C., Marahiel, M .A. & Holak, T.A. (2000) Solution structure of PCP, a prototype for th e peptidyl carrier domains of modular peptide synthetases. Struct. Fold. Des. 8, 407±418. 12. Lambalot, R.H., Gehring, A.M., Flugel, R .S., Zuber, P., LaCelle, M., Marahiel, M.A., Reid, R., Khosla, C. & Walsh, C.T. (1996) A new enzyme superfamily ± the phosphopantethe inyl transferases. Chem. Biol. 3, 923±936. 13. Walsh, C.T., Gehring, A.M., Weinreb, P.H., Luis, E .N. & Flugel, R.S. (1997) Post-translational modi®cation of polyketide and n onriboso mal peptide synthet ases. Curr. Opin. Chem. Biol. 1, 309±315. 14. Reuter, K., Mo®d, M.R., Marahiel, M.A. & Ficner, R. (1999) Crystal structure of the surfactin synthetase-activating enzyme Sfp: a prototype of the 4¢-phosphopantetheinyl transferase superfamily. EMBO J. 18, 6823±6831. 15. Stein, T., Kluge, B., Vater, J., Franke, P., Otto, A. & Wittmann- Liebold, B. (1995) Gramicidin S synthetase 1 (phenylalanine racemase), a prototype of amino acid r acemases containing the cofactor 4¢-phosphopantethein e. Biochemistry 34, 4633±4642. 16.Zocher,R.,Nihira,T.,Paul,E.,Madry,N.,Peeters,H., Kleinkauf, H. & Keller, U. (1986) Biosynthesis of cyclosporin A: partial puri®cation and properties of a multifunctional enzyme from Tolypocladium in¯atum. Biochemistry 25, 550±553. 17. Konz, D., Klens, A., Schorgendorfer, K. & Marahiel, M.A. (1997) The bacitracin biosynthesis operon of Bacillus licheniformis ATCC 10716: molecular characterization o f three multi-modular peptide synthetases. Chem. Biol. 4, 927±937. 18. Shaw-Reid, C.A., Kelleher, N.L., Losey, H.C., Gehring, A.M., Berg, C. & Walsh, C.T. (1999) Assembly line enzymology by multimodular nonribosomal peptide synth etases: the thioesterase domain of E. coli EntF catalyzes both elongation and cyclolact- onization. Chem. Biol. 6, 385±400. 19. Trauger, J., Kohli, R., Mootz, H., Marahiel, M. & Walsh, C. (2000) Peptide cyclization catalysed by the thioesterase domain of tyrocidine synthetase. Nature 407, 215±218. 20. Weber, G ., Scho È rgendorfer, K., Schneider-Scherzer, E. & Leitner, E. (1994) The peptide synthetase catalyzing cyclosporine produc- tion in Tolypocladium niveum is encoded by a giant 45.8-kilobase open reading frame. Curr. Genet. 26, 120±125. 21. Vollenbroich,D.,Kluge,B.,D'Souza,C.,Zuber,P.&Vater,J. (1993) Analysis of a mutant amino acid-activating domain of surfactin synthetase bearing a serine-to-alanine substitution at the site of carboxylthioester formation. FEBS Lett. 325, 220±224. 22. DeCrecy-Lagard,V.,Marliere,P.&Saurin,W.(1995)Multien- zymatic non ribosomal peptide biosynthesis: identi®cation of the functional do mains catalysing peptide elongation and epimerisa- tion. C. R. Acad. Sci. III 318, 927±936. 23. Leslie, A.G.W. (1990) Re®ned crystal structure of type III chloramphenicol acetyltransferase at a 1.75 A Ê resolution. J. Mol. Biol. 213, 167±186. 24. Mattevi, A., Obmolova, G., Schulze, E., Kalk, K.H., Westphal, A.H., de Kok, A . & Hol, W.G.J. ( 1992) Atomic structure of cubic core of the pyruvate dehydrogenase multienzyme complex. Science 255, 1544±1550. 25. Barik, S . (1997) Mutagenesis and gene fusion by megaprimer PCR. Methods Mol. Biol. 67, 173±182. 26. Sambrook, J., Fritsch, E.F. & Maniatis, T. (1989) Molecular Cloning: a Laboratory Manual, Cold Spring Harbor Laboratory Press, Cold Spring Harbor, New York. 27. Bullock, W.O., Fernandez, J.M. & Short, J.M. (1987) XL1-Blue: a high eciency plasmid transforming recA E. coli strain with beta- galactosidase selection. Biotechniques 5, 376±379. 28. Scaloni, A., Jones, W.M., Barra, D., Pospischil, M., Sassa, S., Popowicz,A.,Manning,L.R.,Schneewind,O.&Manning,J.M. (1992) Acylpeptide hydrolase: inhibitors and some active site res- idues of the human enzyme. J. Biol. Chem. 267, 3811±3818. 29. Stachelhaus, T . & Marahiel, M .A. (1995) Mod ular structure of genes encoding multifunctional peptide synthetases required for non-ribosomal pe ptide synt hesis. FEMS Microbiol. Lett. 125, 3±14. 30. Weinreb, P.H., Quadri, L.E., Walsh, C.T. & Zuber, P. (1998) Stoichiometry and speci®city of in vitro phosphopantetheinylation and aminoacylation of the valine-activating module of surfactin synthetase. Biochemistry 37, 1575±1584. 31.Elliott,R.J.,Bennet,A.J.,Braun,C.A.,MacLeod,A.M.& Borgford, T.J. (2000) Active-site variants of Streptomyces griseus protease B with peptide-ligation activity. Chem. Biol. 7, 163±171. 32. Ellis, J., Bagshaw, C. & Shaw, W. (1995) Kinetic mechanism of chloramphenicol acetyltransferase: the role of ternary complex interconversion in rate determination. Bioc hem istry 34, 16852±16859. 33. Ehmann, D.E., Trauger, J.W ., Stachelhaus, T. & Walsh , C.T. (2000) Aminoacyl-SNACs as small-molecule sub strates for the condensation domains of nonribosomal peptide synthetases. Chem. Biol. 7, 765±772. Ó FEBS 2002 Catalysis of nonribosomal peptide-bond formation (Eur. J. Biochem. 269) 629 . assembly of domains) is predominantly colinear to the ®nal product [3,4]. The C-domain, a 450-amino-acid expanding region at the N-terminus of each elongating. [5] and the corresponding residue of D151 in the SrfB2 Asp -CAT surfactin synthetase essential for the synthesis of surfactin in B. subtilis, shown in vivo