Naturwissenschaften (2001) 88:93–101 DOI 10.1007/s001140100211
R E V I E W A RT I C L E
Dirk Schwarzer · Mohamed A. Marahiel
Multimodular biocatalysts for natural product assembly
Published online: 13 March 2001 © Springer-Verlag 2001
Abstract Nonribosomal peptides and polyketides represent a large class of natural products that show an extreme structural diversity and broad pharmacological relevance. They are synthesized from simple building blocks such as amino or carboxy acids and malonate derivatives on multimodular enzymes called nonribosomal peptide synthetases (NRPSs) and polyketide synthases (PKSs), respectively. Although utilizing different substrates, NRPSs and PKSs show striking similarities in the modular architecture of their catalytic domains and product assembly-line mechanism. Among these compounds are well known antibiotics (penicillin, vancomycin and erythromycin) as well as potent immunosuppressive agents (cyclosporin, rapamycin and FK 506). This review focuses on the modular organization of NRPSs, PKSs and mixed NRPS/PKS systems and how modules and domains that build up the biosynthetic templates can be exploited for the rational design of recombinant enzymes capable of synthesizing novel compounds.
Nonribosomal peptide synthesis: from products to genes Since the discovery of the first antibiotic, penicillin (Fig. 1a), the biosynthesis of these pharmacologically important drugs has been of continual interest. Following this milestone, many other antibiotics were identified and used as drugs in modern medicine. Among them are the peptide antibiotics that comprise substances such as the penicillin and cephalosporin precursor δ-L-α-aminoadipyl-L-cysteinyl-D-valine (ACV) and the immunosuppressant cyclosporin, as well as the vancomycin-group of antibiotics that often represent the “last line of defense” against multi-resistant pathogenic bacteria (Healy et al. 2000). These low molecular weight compounds freD. Schwarzer · M.A. Marahiel (✉) Fachbereich Chemie/Biochemie, Philipps-Universität Marburg, Hans-Meerwein-Strasse, 35032 Marburg, Germany e-mail:
[email protected] Tel.: +49-6421-2825722
quently possess cyclic or branched cyclic peptide backbones that may also contain non-proteinogenic amino acids. Their diverse structural features suggest that these peptides are not synthesized by the nucleic acid-dependent ribosomal machinery. This hypothesis was confirmed in the early 1960s by biochemical studies on cellfree extracts from several peptide antibiotic producers. Cell-free production of a particular antibiotic could be restored even in presence of RNase or ribosome inhibitors (Gevers et al. 1968). Further biochemical investigation led to the identification of biosynthetic protein templates which were found to be of enormous size. Like the amino acyl tRNA synthetases of the ribosomal system, these enzymes, today referred to as nonribosomal peptide synthetases (NRPSs), activate their substrate amino acids as amino acyl adenylates at the expense of ATP. The NRPSs responsible for the biosynthesis of the cyclic decapeptide antibiotics gramicidin S and tyrocidine were among the first to be purified and used in vitro to synthesize the intact and active products (Tomino et al. 1967). More than 20 years later, the first genes encoding NRPSs were cloned and sequenced, revealing the modular organization of these megaenzymes and their independent evolution from the ribosomal machinery (Weckermann et al. 1988). The biosynthetic operon for gramicidin S synthesis encodes two open reading frames for the gramicidin S NRPSs GrsA (122 kDa) and GrsB (570 kDa), comprising one and four repetitive building units (modules), respectively (Turgay et al. 1992; Saito et al. 1994). Sequence data of more than 40 complete gene clusters encoding NRPSs are in the current databases and detailed biochemical studies have allowed the elucidation of many mechanistic details.
Understanding the nonribosomal machinery Biochemical and genetic studies of NRPSs have led to the so-called “multiple carrier thio-template mechanism”
94
being described, which represents the current model of this mode of peptide synthesis (Stein et al. 1996). According to this model, NRPSs are built of repetitive catalytic units called modules (Fig. 2), each with about 1,000–1,500 amino acids in length, which are responsible for the incorporation of one residue into the growing peptide chain. The number and order of modules within a NRPS matches the number and sequence of amino acids incorporated into the peptide product. Today some exceptions to these rules are known and will be discussed later. In all fungal NRPSs so far reported, these modules are part of one large polypeptide chain, whereas in bacterial systems these modules are usually distributed on several enzymes, with the largest one currently known containing eight modules (Guenzi et al. 1998). Each module is responsible for catalyzing one complete cycle of chain elongation, comprising the three basic reactions: substrate recognition, activation as acyl adenylate, and covalent binding as thioester. These enzymatic activities are embedded in distinct catalytic domains within the module (Fig. 2). Consequently, domains are usually part of a multifunctional NRPS, they partially exist as separate enzymes only in certain bacterial systems. Domains were found to be distinct catalytic folds with highly conserved core motifs, important for their catalytic activities (Stachelhaus and Marahiel 1995). Three domains are necessary as the basic equipment of a NRPS elongation module: an adenylation (A)domain that selects the substrate amino acid and activates it as amino acyl adenylate; a peptidyl carrier protein (PCP)-domain that binds the co-factor 4′-phosphopantetheine (4′-PP) to which the activated amino acid is covalently attached; and a condensation (C)-domain that catalyzes peptide bond formation. A-domains
Fig. 1 Structures of some bioactive secondary metabolites: a peptides synthesized by NRPS-systems, b products synthesized by mixed NRPS/PKS systems
These control the first step of nonribosomal peptide synthesis, namely the selection of an amino acid substrate and its subsequent activation as amino acyl adenylate (Fig. 3a; Dieckmann et al. 1995). About 550 amino acid residues in length, the substrate-dependent ATP-hydrolyzing activities observed in purified NRPSs correspond to the substrate specificities of the resident A-domains. However, compared with amino acyl tRNA synthetases which carry out the same chemical reaction, the Adomains of NRPSs often seem to have a relaxed substrate specificity. Recently, the successful crystallization and structural analysis of a prototype A-domain with the substrate amino acid and ATP led to the identification of the enzyme's active site and revealed the structural basis for substrate recognition and activation (Conti et al. 1997). Analysis of the binding pocket in combination with thorough sequence comparison with other Adomains allowed the identification of the substrate-binding pocket with all the residues needed for substrate coordination. The importance of these residues was confirmed by mutational analysis, leading to predicted alter-
95 Fig. 2 From modules to products: the modules of NRPS and PKS can be subdivided into domains that catalyze the single enzymatic reactions. The composition of the products is determined by the assembly of active domains found in the corresponding modules
PCP-domains The second step of the nonribosomal assembly of peptides is the transfer of the activated amino acid from the A-domains to the module's transport domain; the peptidyl carrier protein (PCP) or thiolation (T)-domain. Each PCP is equipped with a covalently-bound 4′-phosphopantetheine (4′-PP) co-factor, which is post-translationally attached to a highly conserved serine residue. This apo-to-holo conversion of all PCPs is catalyzed by dedicated phosphopantetheinyl transferases (4′-PP-transferases; Lambalot et al. 1996). PCPs, which are about 80–100 amino acid residues in length and are located downstream of A-domains, covalently bind the amino acid as a thioester to the terminal thiol moiety of the cofactor 4′-PP as aminoacyl-S-PCP, which can now serve as a swinging arm to reach various catalytic centers (e.g., for condensation or modification domains; Fig. 3a). C-domains
Fig. 3 Reactions catalyzed by single NRPS domains: a the catalytic activities of the essential domains, b catalyzed reactions by recently discovered auxiliary domains in NRPS
ations in specificity. Consequently, these residues are defined as the codons for substrate recognition by the Adomain within NRPSs. Using these residues, it is now possible to predict the specificity of biochemically uncharacterized NRPSs simply by sequence analysis (Stachelhaus et al. 1999).
The third domain of the minimal functional module is the condensation-domain (about 450 amino acids in length). The C-domain catalyzes the peptide bond formation between an amino acyl or peptidyl-S-PCP from the upstream module and the amino acyl moiety attached to the corresponding downstream module (Fig. 3a; Marahiel et al. 1997). Recently, it has been demonstrated that C-domains are selective to some degree for the incoming amino acyl-S-PP nucleophile. A minimal bimodular apoNRPS capable of carrying out one elongation reaction was mis-acylated in vitro with modified 4′-PP derivatives of the upstream and downstream PCPs using amino acyl-CoAs and a 4′-PP-transferase. By means of this mis-priming, the A-domains could be bypassed and the substrate selectivity of the C-domain assayed. The use of different amino acyl-CoAs led to the identification of a sieve in the acceptor site discriminating against size and stereochemistry of the incoming nucleophile. In contrast, little or no sieving could be observed for the electrophilic amino acyl or peptidyl donor of the preceding module.
96
Thus, C-domains apparently only verify the monomeric amino acyl acceptor (Belshaw et al. 1999). Although the importance of the C-domain in the elongation reaction was demonstrated by deletion and mutation experiments (Stachelhaus et al. 1998), the mechanism of peptide bond formation catalysis is as yet unknown. Te-domains In most bacterial systems, a fourth domain, the termination/thioesterase (Te)-domain (about 250 residues) has been found to be essential for product release (Fig. 3a; Schneider and Marahiel 1998). The Te-domain is located at the extreme C-terminal module of the corresponding biosynthetic template and catalyzes the release of the biosynthesized peptide in either linear, cyclic or branched cyclic form (Trauger et al. 2000). Auxiliary-domains In addition to the domains described above, several other modifying domains have been found in NRPS modules, which enlarge the structural diversity of the synthesized peptides. One example is the epimerization (E)-domain (about 400 residues) that converts the thioester-bound amino acid of an amino acyl-S-PCP or a peptidyl-S-PCP from the L- into the D- configuration (Fig. 3b; Stindl and Keller 1994; Stachelhaus and Walsh 2000). E-domains are usually located downstream of the C-terminus of the corresponding PCP. In some NRPS modules an N-methylation (M)-domain (about 450 residues) is inserted into the C-terminal end of an A-domain (Haese et al. 1993), catalyzing the transfer of a methyl group from co-factor S-adenosylmethionine (SAM) to the amino group of a PCP-bound amino acid. An alternative mode of elongation has recently been observed in some NRPS systems. Here, a cyclization (Cy)-domain substitutes for the usual C-domain in modules incorporating cysteine, serine or threonine residues. Instead of a simple peptide bond formation, the module mediates the formation of heterocyclic rings such as oxazolines or thiazolines. Although little is known about the catalytic mechanism and timing of condensation and heterocyclization, the catalytic role of the Cy-domain has been demonstrated in vitro in the yersiniabactin (Suo et al. 1999) and pyochelin systems (Quadri et al. 1999). The number of new catalytic domains is increasing. Newly discovered domains include a putative oxidation (Ox)-domain (about 250 amino acids) that was found strictly associated with Cy-domains. The Ox-domain is believed to oxidize the thiazoline ring formed into the aromatic thiazol by using FMN as co-factor (Fig. 3b; Du et al. 2000). Intriguingly, Ox-domains can be found in two different locations within the NRPS (Fig. 3b): downstream of the PCP or inserted into the C-terminal part of an A-domain. The broad substrate specificity of A-domains and the presence of several auxiliary domains in NRPS modules
facilitates the synthesis of small peptides with a high level of structural diversity.
Mechanism of polyketide biosynthesis Polyketides represent another important group of natural products, many of which are also synthesized on multimodular enzymes called type I polyketide synthases (PKSs). Among these polyketides are products of great pharmacological importance, such as the antibiotic, erythromycin and the immunosuppressive agents, rapamycin and FK 506. In contrast to the structural diversity of these products, virtually only two building blocks, acetate and propionate, are used in polyketide synthesis. The high level of structural diversity found in these compounds is generated by optional reduction of the keto groups and further tailoring steps such as hydroxylation and glycosylation (Hopwood 1997). Similarly to NRPS, each PKS module is responsible for the incorporation of one acetate or propionate unit into the growing ketide chain. The building blocks are selected in activated forms as malonyl- and methylmalonyl-CoA. A PKS module can also be further subdivided into domains, some of which are essential and some optional for chain elongation (Fig. 2; Cane and Walsh 1999). The four essential domains are the ketosynthase (KS)-domain, the acetyltransferase (AT)-domain, the acyl carrier protein (ACP)-domain and the thioesterase (Te)-domain (Khosla 1997). AT-domains The first step in polyketide assembly, catalyzed by the AT-domain (about 350 residues), is the transfer of activated building blocks to the corresponding ACP-domain (Fig. 4a). From this point of view, the AT-domain can be compared to the A-domain of the NRPS. However, the A-domain of NRPS fulfills two roles, in both selecting and activating the substrate. In contrast the AT-domain of a PKS receives an activated substrate either as malonylor methylmalonyl-CoA. ACP-domains The acyl carrier proteins (ACP)-domain (about 80–100 residues) have the same function as the PCP-domains in NRPSs, serving as transport units for the building blocks and elongation intermediates (Keating and Walsh 1999). KS-domains The ketosynthase (KS)-domain (about 450 residues) catalyzes the elongation reaction in polyketide synthesis (Fig. 4a). The reaction starts with the transfer of the ketide chain from the ACP of an upstream module to the
97
After an elongation step and before the transfer onto the KS-domain of the next module, the β-keto moiety of the intermediate can be reduced to a hydroxyl group by the KR-domain. This reduction is dependent on the cosubstrate NADPH (Fig. 4b). Dehydratization of the βhydroxyl group by a DH-domain results in double bond formation between positions 2 and 3 in the growing ketide chain (Fig. 4b). Further reduction occurs in modules containing an ER-domain which also uses NADPH as co-substrate (Fig. 4b). The product of the last step in the reductive cycle is a saturated ketide in positions 2 and 3 that is subsequently used for further elongation (Katz 1997).
Mixed NRPS/PKS systems
Fig. 4 Essential PKS domains (a), reactions catalyzed by PKS auxiliary domains (b)
cysteine residue in the active site of the KS-domain. A malonyl unit tethered to the ACP of the corresponding module is then decarboxylated by the KS-domain to give the nucleophile that reacts with the thioester group of the ketide chain attached to the KS-domain. This condensation step yields the ketide chain elongated by an acetate unit with a β-keto group. The ketide chain is now attached to the ACP of that module (Katz 1997). Te-domains Upon complete elongation and reduction, the ketide chain is transferred to the thioesterase domain (Te; about 250 residues) which is responsible for product release at the C-terminus of the last PKS-module (Katz 1997). Products are released as macrolactones (Fig. 4a), after head-to-tail cyclization of the ketide chain. Auxiliary domains The auxiliary domains of the PKS are the ketoreductase (KR)-domain, the dehydratase (DH)-domain and the enoylreductase (ER)-domain.
With the increasing number of reported NRPS and PKS gene clusters, mixed NRPS/PKS clusters can be identified, combining both strategies for the creation of structural diversity in natural products. One of the first sequenced PKS clusters, the rapamycin cluster from Streptomyces hygroscopicus (structure Fig. 1b), belongs to this group (Schwecke et al. 1995). The ketide chain is assembled by three multimodular enzymes, RAPS1–3, containing a total of 14 PKS modules and a CoA ligase as loading domain. After 14 elongation reactions, the ketide chain is transferred onto RapP (König et al. 1997), a NRPS module containing a C-domain, an A-domain specific for pipecolic acid, a PCP, and a second C-domain (C-A-PCP-C). RapP elongates the ketide chain with pipecolic acid and releases the final product by cyclization, presumably catalyzed by the second C-domain. Another prominent example for a mixed NRPS/PKS system is the recently reported epothilone cluster. Epothilone (Fig. 1b) is an important anticancer drug with cytostatic activity similar to taxol through stabilization of microtubules, produced by the myxobacterium Sorangium cellulosum So ce90 (Molnar et al. 2000; Tang et al. 2000). Analysis of the cluster revealed the presence of six enzymes containing nine modules and one loading module (Fig. 5g). EpoB is a cysteine-specific NRPS module (Cy-A-Ox-PCP), which comes into play during epothilone assembly as the first elongation step downstream of the loading module. It catalyzes peptide-bond formation between the acetyl moiety of the loader and the activated cysteinyl group, followed by heterocyclization to a thiazoline ring and further oxidation to give the thiazole product. This intermediate is then transferred to the remaining PKS modules that effect further elongation and final macrolactonization. The myxothiazol cluster from the myxobaterium Stigmatella aurantica DW 4/3-1 represents another interesting example of mixed NRPS/PKS systems, in particular as it contains NRPS and PKS modules directly connected on the same polypeptide chain (Silakowski et al. 1999). Six genes encode enzymes with a total of eight modules and one loader (Fig. 5f). Three modules in positions 3, 4 and 8 are the NRPS modules. Module 3 is a
98
Fig. 5 Examples for the organization of NRPS and NRPS/PKS gene clusters encoding NRPS enzymes that can assemble the following products: a surfactin from Bacillus subtilis ATCC 21332 (Cosmina et al. 1993), b tyrocidine from Bacillus brevis ATCC 8185 (Mootz and Marahiel 1997), c cyclosporin from Tolypocladium inflatum (Weber et al. 1994), d chloroeremomycin from Amycolatopsis orientalis (van Wageningen et al. 1998), e actinomycin from Streptomyces chrysomallus (Schauwecker et al. 2000). Mixed NRPS/PKS cluster for the following products: f myxothiazol from Stigmatella aurantiaca (Silakowski et al. 1999), g epothilone from Sorangium cellulosum So ce90 (Molnar et al. 2000; Tang et al. 2000), h yersiniabactin from Yersinia pestis (Gehring et al. 1998), i mycobactin from Mycobacterium tuberculosis (Quadri et al. 1998), j antibiotic TA from Myxoccocus xanthus (Paitan et al. 1999)
separate enzyme, MtaC, showing the same net domain composition as EpoB (Cy-A-PCP-Ox; Fig. 5f). In contrast to EpoB, however, the Ox-domain of MtaC is located downstream of the PCP and is not inserted into the Adomain. The second NRPS module is part of the MtaD enzyme (Fig. 5f) which can be seen as a “true” NRPS/PKS hybrid, consisting of two modules with a total of nine domains (Cy-A-Ox-PCP-KS-AT-DH-KRACP). The role of MtaG, the last NRPS module, again a separate enzyme, is unclear. It consists of a C-domain, an A-domain of unknown specificity, an inserted domain that shares homology to monooxygenases, followed by a PCP- and a Te-domain (C-A-monoOx-PCP-Te; Fig. 5f). It is presumably responsible for a so far unique termination mode in myxothiazol biosynthesis leading to the Cterminal amide function found in the product (Fig. 1b). In the antibiotic TA (Fig. 1b), produced by Myxoccocus xanthus, glycine is inserted into the ketide backbone. So far, only a fragment of the TA gene cluster has been cloned and sequenced (Fig. 5j; Paitan et al. 1999).
The encoded TA protein harbors an NRPS and a PKS module on one polypeptide chain, beginning with a Cdomain followed by an A-domain with predicted glycine specificity and a PCP (Fig. 5j). The PKS module starts with a KS-domain followed by a KR-domain and an ACP. Interestingly, this PKS module does not contain the AT-domain normally belonging to the minimal equipment of functional PKS modules. It has been proposed that in this case a separate acetyl transferase is responsible for the transfer of the extender unit to the ACP. The basic rules of polyketide assembly with respect to the organization of PKS domains, are not valid for the TA fragment, with one essential domain missing. A very interesting and unusual arrangement of NRPS and PKS domains can be found in the yersiniabactin biosynthetic gene cluster of Yersinsia pestis (Fig. 5h; Gehring et al. 1998). Yersiniabactin is a siderophore essential for the uptake of iron ions under iron deprivation conditions and is therefore essential for the survival of this pathogenic bacteria. The cluster contains three genes, ybtE, irp2 and irp1, encoding the biosynthetic enzymes YbtE, HMWP2 and HMWP1. In yersiniabactin assembly, salicylic acid is first activated by YbtE, a separate A-domain (Fig. 5h), and then transferred onto the first PCP of HMWP2, where two elongation steps with cysteine and subsequent heterocyclization occur. The resulting intermediate is transferred to HMWP1, and further extended by a ketide unit and a third cysteine for formation of a thiazoline ring. The product is released by the C-terminal Te-domain of HMWP1. Intriguingly, three cysteines are necessary for the synthesis of the three heterocyclic rings found in the siderophore, but only one cysteine-activating A-domain is included in the enzymes, namely in HMWP2 (Fig. 5h). This A-domain was shown to load
99
cysteine on all three PCPs located on two different enzymes (Keating et al. 2000; Quadri 2000). This unusual arrangement requires intermodular interactions between A-domains and PCPs and thus represents an exception to the established rules for the nonribosomal assembly-line mechanism, with one module strictly responsible for the incorporation of one amino acid. As the last example for mixed NRPS/PKS, the mycobactin system of Mycobacterium tuberculosis exhibits several unusual features of domain organization that are not yet understood (Fig. 5i; Quadri et al. 1998; Quadri 2000). Six enzymes seem to be responsible for biosynthesis of mycobactin (Fig. 1b), namely MbtA–MbtF. A PKS-module is distributed between two enzymes, MbtC and MbtD; the first is a separate KS-domain and the latter an ACP-AT-KR-ACP unit. In addition, MbtE (CA-PCP-C-PCP) with a PCP directly downstream of a Cdomain, also possesses an unusual domain composition. The role of the second PCP of MbtD in mycobactin assembly is unclear. These unusual arrangements of domains show that the simple basic rules for nonribosomal peptide synthesis can be circumvented by Nature in order to produce some natural products with a biosynthetic logic that we do not understand as yet.
Utilization of NRPSs as a molecular tool box As previously discussed, NRPSs are a powerful device for the synthesis of peptides that cannot be made on the ribosome. The variety of domains that incorporate and modify proteinogenic as well as non-proteinogenic amino acids can be regarded as the molecular tool box of NRPSs. Bacteria and fungi use this tool box for the synthesis of a wide spectra of secondary metabolites. The modular architecture of NRPSs allows alteration of the arrangement of modules or domains to create hybrid enzymes capable of synthesizing new peptides with a defined amino acid sequence. A straightforward approach to utilizing the “NRPS tool box” for biotechnological purposes would be the rearrangement of the DNA coding for the catalytic units of these biosynthetic enzymes. This can be done by fusing heterologous gene fragments of modules or domains with desired specificities to generate hybrid genes. The minimal domain equipment of an elongation module (C-A-PCP), suggests three approaches for such recombination of functional units (Mootz and Marahiel 1999): (1) intra-modular fusion between the domain boundaries of a C-domain and an A-domain, (2) intramodular fusions between an A-domain and a PCP, and (3) inter-modular fusions by connecting complete modules between a PCP and C-domain. Examples of all three fusion strategies have been reported recently and will be discussed. Crucial to all approaches is the knowledge of linker regions between the domains suitable for artificial fusion, i.e. such that the domains remain catalytically active and interact readily across the fusion site.
The first reported engineering experiment was an exchange of two domains in a surfactin NRPS, namely SrfA–C (Fig. 5a). Surfactin is a cyclic lipo-heptapeptide produced by Bacillus subtilis (Fig. 1a; Cosmina et al. 1993). The exchanges were done at the seventh module of surfactin synthetase. This terminal module is composed of a C-domain, a leucine-activating A-domain, a PCP, and a Te-domain. The A-domain and the PCP, representing a unit for substrate-activating and covalent binding, were replaced by different A-PCP units from bacterial and fungal origin of different amino acid specificities. The production of the novel surfactin derivatives was confirmed by mass analysis; however, the yield was found to be very poor (Stachelhaus et al. 1995). Nowadays, this type of fusion does not seem to be the best choice, as recently a certain selectivity of the C-domains towards their cognate amino acid at the acceptor site has been suggested (Belshaw et al. 1999; Ehmann et al. 2000). This selectivity might explain why additional C–A fusion experiments within the surfactin gene cluster were not successful (Schneider et al. 1998). Recently, a successful A–PCP unit substitution has been reported in the actinomycin synthetases (Fig. 5e; Schauwecker et al. 2000). The actinomycin synthetase II (ACMS II) consists of two modules responsible for the incorporation of threonine and valine, whereas the latter is epimerized by the E-domain of the second module (CA-PCP-C-A-PCP-E; Fig. 5e). A set of two hybrid enzymes based on ACMS II was constructed. The valineactivating and binding A–PCP unit was replaced by the valine-activating, N-methylating and binding A-M-PCP unit of ACMS III, with either deleting (C-A-PCP-C-AM-PCP) or restoring the C-terminal E-domain (C-APCP-C-A-M-PCP-E) of the module. In product assays, the synthesis of the predicted peptides was observed with the M-domain containing constructs. Epimerization, however, did not occur in the hybrid enzymes. Interestingly, the C-domain of the second ACMS II module that usually uses the primary amino group of valine for peptide bond formation can also process the secondary amino group of N-methyl-valine. The second possible inter-modular fusion site in NRPS is between an A-domain and a PCP, leaving the unit of C- and A-domains intact. This type of fusion was expected not to interfere with the C-domain's selectivity. Fusions of this type have been reported and a set of bimodular hybrid enzymes was constructed and characterized in vitro (Doekel and Marahiel 2000). In these experiments, two different A-domains were fused to a PCP that is followed by a complete elongation module including a terminal Te-domain. The critical point in this kind of fusion is the restoration of the A-domain's ability to transfer the activated amino acid to the PCP's co-factor. This transfer was monitored in loading and elongation assays, as well as in formation and release of the predicted dipeptides (Doekel and Marahiel 2000). The success of this fusion type indicates that a NRPS module can alternatively be defined as a unit with the domain order: PCP-C-A.
100
The third possible fusion site is located between a PCP and a C-domain, representing a whole-module fusion. Fusions of this type used modules from the tyrocidine synthetases of Bacillus brevis (Mootz and Marahiel 1997; Fig. 5b). It was shown previously that the first two modules of the tyrocidine synthetase (TycA and TycB1) could be reconstituted and dipeptide formation of DPhePro detected in vitro (Stachelhaus et al. 1998). This model system was then extended by fusing modules 9 and 10 of the tyrocidine synthetases to the C-terminus of TycB1 (Mootz et al. 2000). Module 9 is an ornithine-specific elongation module (C-A-PCP) and module 10 is a leucine-specific one which also harbors a Te-domain (C-APCP-Te). When using module 10, the predicted tripeptide DPhe-Pro-Leu could be detected. The turnover rate of tripeptide production was found to be dependent on the presence of the Te-domain, since the truncated module 10 lacking the Te-domain does not promote tripeptide release in this system. Fusions with module 9 were successful as well, but again catalyzed tripeptide release of DPhe-Pro-Orn was observed only when the Te-domain was fused additionally to the C-terminus of module 9. These experiments demonstrated the feasibility of whole-module fusion and that a Te-domain is necessary for product release. In summary, the modular architecture of the NRPS presents a great potential for the rational design of hybrid enzymes. Examples of different ways of constructing such hybrid enzymes in vitro and in vivo have been reported. The choice of the fusion site between the domains is still one of the most critical points. The substrate specificity of the C-domain at the nucleophilic acceptor site must be taken into account, and an optimal domain interaction must be achieved. Although several successful examples have been presented, the rules governing domain interactions still remain a mystery. Acknowledgements Work in the MAM laboratory is supported by the Deutsche Forschungsgemeinschaft and Fonds der Chemischen Industrie. We would like to thank H.D. Mootz, W. Klein, T. Stachelhaus and M.T. Stubbs for carefully reading the manuscript.
References Belshaw PJ, Walsh CT, Stachelhaus T (1999) Aminoacyl-CoAs as probes of condensation domain selectivity in nonribosomal peptide synthesis. Science 284:486–489 Cane DE, Walsh CT (1999) The parallel and convergent universes of polyketide synthases and nonribosomal peptide synthetases. Chem Biol 6:R319–R325 Conti E, Stachelhaus T, Marahiel MA, Brick P (1997) Structural basis for the activation of phenylalanine in the non-ribosomal biosynthesis of gramicidin S. EMBO J 16:4174–4183 Cosmina P, Rodriguez F, Ferra F de, Grandi G, Perego M, Venema G, Sinderen D van (1993) Sequence and analysis of the genetic locus responsible for surfactin synthesis in Bacillus subtilis. Mol Microbiol 8:821–831 Dieckmann R, Lee YO, Liempt H van, Döhren H von, Kleinkauf H (1995) Expression of an active adenylate-forming domain of peptide synthetases corresponding to acyl-CoA-synthetases. FEBS Lett 357:212–216
Doekel S, Marahiel MA (2000) Dipeptide formation on engineered hybrid peptide synthetases. Chem Biol 7:373–384 Du L, Chen M, Sanchez C, Shen B (2000) An oxidation domain in the BlmIII non-ribosomal peptide synthetase probably catalyzing thiazole formation in the biosynthesis of the anti-tumor drug bleomycin in Streptomyces verticillus ATCC15003. FEMS Microbiol Lett 189:171–175 Ehmann DE, Trauger JW, Stachelhaus T, Walsh CT (2000) Aminoacyl-SNACs as small-molecule substrates for the condensation domains of nonribosomal peptide synthetases. Chem Biol 7:765–772 Gehring AM, DeMoll E, Fetherston JD, Mori I, Mayhew GF, Blattner FR, Walsh CT, Perry RD (1998) Iron acquisition in plague: modular logic in enzymatic biogenesis of yersiniabactin by Yersinia pestis. Chem Biol 5:573–586 Gevers W, Kleinkauf H, Lipmann F (1968) The activation of amino acids for biosynthesis of gramicidin S. Proc Natl Acad Sci USA 60:269–276 Guenzi E, Galli G, Grgurina I, Gross DC, Grandi G (1998) Characterization of the syringomycin synthetase gene cluster: a link between prokaryotic and eukaryotic peptide synthetases. J Biol Chem 273:32857–32863 Haese A, Schubert M, Herrmann M, Zocher R (1993) Molecular characterization of the enniatin synthetase gene encoding a multifunctional enzyme catalysing N-methyldepsipeptide formation in Fusarium scirpi. Mol Microbiol 7:905–914 Healy VL, Lessard IA, Roper DI, Knox JR, Walsh CT (2000) Vancomycin resistance in enterococci: reprogramming of the D-ala-D-Ala ligases in bacterial peptidoglycan biosynthesis. Chem Biol 7:R109–R119 Hopwood DA (1997) Genetic contributions to understanding polyketide synthases. Chem Rev 97:2465–2497 Katz L (1997) Manipulation of modular polyketide synthases. Chem Rev 97:2557–2575 Keating TA, Walsh CT (1999) Initiation, elongation, and termination strategies in polyketide and polypeptide antibiotic biosynthesis. Curr Opin Chem Biol 3:598–606 Keating TA, Suo Z, Ehmann DE, Walsh CT (2000) Selectivity of the yersiniabactin synthetase adenylation domain in the twostep process of amino acid activation and transfer to a holocarrier protein domain. Biochemistry 39:2297–2306 Khosla C (1997) Harnessing the biosynthetic potential of modular polyketide synthases. Chem Rev 97:2577–2590 König A, Schwecke T, Molnar I, Bohm GA, Lowden PA, Staunton J, Leadlay PF (1997) The pipecolate-incorporating enzyme for the biosynthesis of the immunosuppressant rapamycin–nucleotide sequence analysis, disruption and heterologous expression of rapP from Streptomyces hygroscopicus. Eur J Biochem 247:526–534 Lambalot RH, Gehring AM, Flugel RS, Zuber P, LaCelle M, Marahiel MA, Reid R, Khosla C, Walsh CT (1996) A new enzyme superfamily – the phosphopantetheinyl transferases. Chem Biol 3:923–936 Marahiel MA, Stachelhaus T, Mootz HD (1997) Modular peptide synthetases involved in nonribosomal peptide synthesis. Chem Rev 97:2651–2673 Molnar I, Schupp T, Ono M, Zirkle R, Milnamow M, NowakThompson B, Engel N, Toupet C, Stratmann A, Cyr DD, Gorlach J, Mayo JM, Hu A, Goff S, Schmid J, Ligon JM (2000) The biosynthetic gene cluster for the microtubule-stabilizing agents epothilones A and B from Sorangium cellulosum So ce90. Chem Biol 7:97–109 Mootz HD, Marahiel MA (1997) The tyrocidine biosynthesis operon of Bacillus brevis: complete nucleotide sequence and biochemical characterization of functional internal adenylation domains. J Bacteriol 179:6843–6850 Mootz HD, Marahiel MA (1999) Design and application of multimodular peptide synthetases. Curr Opin Biotechnol 10:341– 348 Mootz HD, Schwarzer D, Marahiel MA (2000) Construction of hybrid peptide synthetases by module and domain fusions. Proc Natl Acad Sci USA 97:5848–5853
101 Paitan Y, Alon G, Orr E, Ron EZ, Rosenberg E (1999) The first gene in the biosynthesis of the polyketide antibiotic TA of Myxococcus xanthus codes for a unique PKS module coupled to a peptide synthetase. J Mol Biol 286:465–474 Quadri LE (2000) Assembly of aryl-capped siderophores by modular peptide synthetases and polyketide synthases. Mol Microbiol 37:1–12 Quadri LE, Sello J, Keating TA, Weinreb PH, Walsh CT (1998) Identification of a Mycobacterium tuberculosis gene cluster encoding the biosynthetic enzymes for assembly of the virulence-conferring siderophore mycobactin. Chem Biol 5:631– 645 Quadri LE, Keating TA, Patel HM, Walsh CT (1999) Assembly of the Pseudomonas aeruginosa nonribosomal peptide siderophore pyochelin: in vitro reconstitution of aryl-4, 2-bisthiazoline synthetase activity from PchD, PchE, and PchF. Biochemistry 38:14941–14954 Saito F, Hori K, Kanda M, Kurotsu T, Saito Y (1994) Entire nucleotide sequence for Bacillus brevis Nagano Grs2 gene encoding gramicidin S synthetase 2: a multifunctional peptide synthetase. J Biochem (Tokyo) 116:357–367 Schauwecker F, Pfennig F, Grammel N, Keller U (2000) Construction and in vitro analysis of a new bi-modular polypeptide synthetase for synthesis of N-methylated acyl peptides. Chem Biol 7:287–297 Schneider A, Marahiel MA (1998) Genetic evidence for a role of thioesterase domains, integrated in or associated with peptide synthetases, in non-ribosomal peptide biosynthesis in Bacillus subtilis. Arch Microbiol 169:404–410 Schneider A, Stachelhaus T, Marahiel MA (1998) Targeted alteration of the substrate specificity of peptide synthetases by rational module swapping. Mol Gen Genet 257:308–318 Schwecke T, Aparicio JF, Molnar I, Konig A, Khaw LE, Haydock SF, Oliynyk M, Caffrey P, Cortes J, Lester JB, et al (1995) The biosynthetic gene cluster for the polyketide immunosuppressant rapamycin. Proc Natl Acad Sci USA 92:7839–7843 Silakowski B, Schairer HU, Ehret H, Kunze B, Weinig S, Nordsiek G, Brandt P, Blöcker H, Höfle G, Beyer S, Müller R (1999) New lessons for combinatorial biosynthesis from myxobacteria: the myxothiazol biosynthetic gene cluster of Stigmatella aurantiaca DW4/3-1. J Biol Chem 274:37391– 37399 Stachelhaus T, Marahiel MA (1995) Modular structure of peptide synthetases revealed by dissection of the multifunctional enzyme GrsA. J Biol Chem 270:6163–6169 Stachelhaus T, Walsh CT (2000) Mutational analysis of the epimerization domain in the initiation module PheATE of gramicidin S synthetase. Biochemistry 39:5775–5787
Stachelhaus T, Schneider A, Marahiel MA (1995) Rational design of peptide antibiotics by targeted replacement of bacterial and fungal domains. Science 269:69–72 Stachelhaus T, Mootz HD, Bergendahl V, Marahiel MA (1998) Peptide bond formation in nonribosomal peptide biosynthesis: catalytic role of the condensation domain. J Biol Chem 273:22773–22781 Stachelhaus T, Mootz HD, Marahiel MA (1999) The specificityconferring code of adenylation domains in nonribosomal peptide synthetases. Chem Biol 6:493–505 Stein T, Vater J, Kruft V, Otto A, Wittmann-Liebold B, Franke P, Panico M, McDowell R, Morris HR (1996) The multiple carrier model of nonribosomal peptide biosynthesis at modular multienzymatic templates. J Biol Chem 271:15428–15435 Stindl A, Keller U (1994) Epimerization of the D-valine portion in the biosynthesis of actinomycin D. Biochemistry 33:9358– 9364 Suo Z, Walsh CT, Miller DA (1999) Tandem heterocyclization activity of the multidomain 230 kDa HMWP2 subunit of Yersinia pestis yersiniabactin synthetase: interaction of the 1-1382 and 1383-2035 fragments. Biochemistry 38:14023–14035 [published erratum Biochemistry 38:17000, 21 December 1999] Tang L, Shah S, Chung L, Carney J, Katz L, Khosla C, Julien B (2000) Cloning and heterologous expression of the epothilone gene cluster. Science 287:640–642 Tomino S, Yamada M, Itoh H, Kurahashi K (1967) Cell-free synthesis of gramicidin S. Biochemistry 6:2552–2560 Trauger JW, Kohli RM, Mootz HD, Marahiel MA, Walsh CT (2000) Peptide cyclization catalysed by the thioesterase domain of tyrocidine synthetase. Nature 407:215–218 Turgay K, Krause M, Marahiel MA (1992) Four homologous domains in the primary structure of GrsB are related to domains in a superfamily of adenylate-forming enzymes. Mol Microbiol 6:529–546 Wageningen AM van, Kirkpatrick PN, Williams DH, Harris BR, Kershaw JK, Lennard NJ, Jones M, Jones SJ, Solenberg PJ (1998) Sequencing and analysis of genes involved in the biosynthesis of a vancomycin group antibiotic. Chem Biol 5:155–162 Weber G, Schorgendorfer K, Schneider-Scherzer E, Leitner E (1994) The peptide synthetase catalyzing cyclosporine production in Tolypocladium niveum is encoded by a giant 45.8-kilobase open reading frame. Curr Genet 26:120–125 Weckermann R, Furbass R, Marahiel MA (1988) Complete nucleotide sequence of the tycA gene coding the tyrocidine synthetase 1 from Bacillus brevis. Nucleic Acids Res 16:11841
