Biosynthesis of the Pseudomonas polyketide coronafacic acid requires monofunctional and multifunctional polyketide synthase proteins

RESULTS AND DISCUSSION

Previous sequencing of the 5′ end of SstI fragment 4 (Fig. 1A) revealed a 230-bp region with relatedness to the eryA genes involved in the synthesis of erythromycin (20). This prompted us to investigate whether there was any antigenic similarity between the proteins involved in CFA biosynthesis and DEBS. Since very large enzymes (>200 kDa) are associated with modular PKSs, we analyzed the cellular proteins of P. s. glycinea N9 (COR⁺ CFA⁺) and several COR⁻ CFA⁻ mutants for the presence of large proteins when grown under conditions optimal for COR production. SDS/PAGE analyses of total cellular proteins in N9 revealed two large proteins of approximately 220 and 280 kDa when this strain was grown under conditions that stimulated COR production (Fig. 2A, lane 1). These proteins were not present when cells were grown in rich medium or at 28°C, conditions unfavorable for COR production. Mutant D5, which maps at the 3′ end of the CFA gene cluster in cfa8 (20), produced the 220- and 280-kDa proteins (Fig. 2A, lane 4). However, the CFA⁻ mutants B2 and B21, which mapped within SstI fragment 5, produced only the 280-kDa protein (Fig. 2A, lanes 2 and 3). Mutants E9 and G2, which mapped to cfa5, produced neither protein (Fig. 2A, lanes 5 and 6), presumably because of the polarity of Tn5 on downstream genes. Mutants A1, A2, A25, and A28, which mapped within SstI fragment 6, also failed to synthesize the two large proteins associated with CFA synthesis (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

Figure 2

Detection of proteins related to DEBS in P. s. glycinea PG4180. (A) Total cellular proteins from the following strains: 1, N9; 2, B2; 3, B21; 4, D5; 5, E9; 6, G2. The migration of molecular size markers (in kDa) is shown on the left. (B) Western blot reactivity of DEBS 2 antisera with proteins shown in A. Arrows on the right indicate the migration of the 200- and 280-kDa proteins that reacted with the DEBS 2 antisera.

The antigenic relatedness of the 220- and 280-kDa proteins to DEBS was investigated in immunoblotting experiments using DEBS 2 antisera. The 220- and 280-kDa proteins in N9 and D5 (Fig. 2B, lanes 1 and 4) and the 280-kDa protein in B2 and B21 (Fig. 2B, lanes 2 and 3) showed immunoreactivity to the DEBS 2 antisera. There was no reactivity with similar-sized proteins in E9, G2 (Fig. 2B, lanes 5 and 6), A1, A2, A25, or A28 (data not shown). These observations suggested that both the 220- and 280-kDa proteins are modular PKSs encoded by DNA largely present in SstI fragments 5 and 6, respectively. Although the exact location of the start site was unknown, we reasoned that the 280-kDa protein was encoded by DNA in SstI fragment 6. This hypothesis was tested by subcloning a 12-kb HindIII/EcoRI fragment encompassing SstI fragment 6 into pBluescript SK(+) and KS(−) to form pCfa6A and pCfa6I, respectively (Fig. 1B). When induced with isopropyl β-d-thiogalactoside, E. coli BL21(pCfa6A) cells contained a 280-kDa protein (Fig. 3, lane 2), which was absent from BL21 cells containing pCfa6I (Fig. 3, lane 3) and BL21[pBluescript SK(+)] (Fig. 3, lane 1). In a parallel experiment, the 6.0-kb SstI/BamHI fragment containing SstI fragment 5 was subcloned in pBluescript SK(+) and KS(−) to form pCfa7A and pCfa7I, respectively (Fig. 1B). After induction, E. coli BL21(pCfa7A) cells contained a 220-kDa protein (Fig. 3, lane 4), which was absent from BL21(pCfa7I) cells (data not shown) and from BL21 containing the other constructs used in this experiment (Fig. 3, lanes 1–3). These results confirmed that pCfa6A and pCfa7A contained ORFs encoding the 280- and 220-kDa proteins, respectively.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3

Overexpression of cfa6 and cfa7. Proteins were labeled with [³⁵S]methionine and [³⁵S]cysteine as described in Experimental Procedures. Constructs were transformed into E. coli BL21(DE3) and overexpressed from the T7 promoter. Lanes: 1, BL21[pBluescript SK(+)]; 2, BL21(pCfa6A); 3, BL21(pCfa6I); and 4, BL21(pCfa7A). Molecular size markers (in kDa) are shown on the left; the migration of Cfa6 and Cfa7 is indicated on the right.

The 14.2-kb region beginning at the BamHI site separating fragments 6 and 7 and extending to the end of SstI fragment 5 (see shaded region, Fig. 1A) was sequenced, and the deduced organization of this region is shown in Fig. 1C. The most notable feature was the presence of two large ORFs designated cfa6 (8.0 kb) and cfa7 (6.2 kb). Each ORF had a potential ribosome binding site 9–12 bp upstream of the translational start site. The authenticity of the ORFs was validated by codon preference analysis with the UWGCG program codonpreference and a Pseudomonas codon usage table. The predicted size of the translational products of cfa6 and cfa7 were 284 and 221 kDa, respectively. Sequence analysis using the blastx and fastembl programs revealed a high degree of similarity to DEBS, and each gene contained one PKS module. The spatial location of the AT, ACP, KS, AT, DH, ER, KR, and ACP domains in Cfa6 and the KS, AT, DH, KR, ACP, and thioesterase (TE) domains in Cfa7 are shown in Fig. 1C. The catalytic sites in the ACP, DH, KS, and KR domains of Cfa6 and Cfa7 were closely related to those previously identified in the DEBS and rapamycin synthase modules (3, 33). The ER domain in Cfa6 was closely related to the ER domain in module 4 of DEBS 2, whereas Cfa7 contained a TE domain at the C terminus that retained the active site residues conserved in the TE domain of DEBS 3 (2). The N-terminal AT and ACP domains of Cfa6 resembled the analogous loading domain of DEBS 1, module 1. The internal AT domain in Cfa6 contained a motif related to that in AT domains which incorporate propionate (34); however, it is likely that this domain is involved in the introduction of butyrate, and a similar observation has been reported for the niddamycin PKS (4). The AT domain in Cfa7 is similar to that observed in ATs which incorporate acetate (34).

Complementation analysis and primer extension experiments previously demonstrated that the CFA region was organized as a single 19-kb transcriptional unit (35). The 5′ end of the transcript contained six discrete ORFs including cfl and cfa1–5 (18, 19) (Fig. 1C). The translational products of cfl and cfa5 showed relatedness to enzymes that activate carboxylic acids by adenylation (acyl-CoA ligases) (18, 19). The protein products of cfa1, cfa2, and cfa3 showed similarity to ACP, DH, and KS, respectively (19). The function of cfa4 could not be predicted from database searches. Interestingly, the following genes were coupled translationally: cfa2-cfa3, cfa3-cfa4, and cfa4-cfa5, ensuring equimolar production of the resulting gene products (19).

In the present study, we show that SstI fragments 5 and 6 contain two large ORFs, cfa6 and cfa7, and that the translational products show strong similarity to DEBS. Cfa6 contains all domains required for extension of the polyketide by butyrate followed by complete reduction, whereas Cfa7 contained the domains needed for extension of the polyketide by acetate followed by reduction and dehydration. Two remaining ORFs in the CFA cluster mapped downstream of cfa7 (Fig. 1C). The translational product of cfa8 showed similarity to crotonyl-CoA reductases from Streptomyces spp. (20). Crotonyl-CoA reductase catalyzes one step in the conversion of acetoacetyl-CoA to butyryl-CoA, and the latter product, as ethylmalonyl-CoA, is used as a four-carbon extender unit in polyketide synthesis (36). A ccr gene is also part of the biosynthetic gene cluster encoding the polyketide tylosin (36), and the overexpression of ccr in Saccharopolyspora erythraea has been utilized to produce novel derivatives of erythromycin (37). The recruitment of a ccr gene into the CFA gene cluster may reflect the requirement of butyryl-CoA as a precursor for CFA synthesis. It is also important to note that cfa7 and cfa8 are coupled translationally (20). The protein product of cfa9 showed relatedness to TEs involved in the synthesis of gramicidin, tyrocidine, and tylosin (20). Analysis of a cfa9 mutant indicated that this gene is dispensable for CFA and COR production but may increase the release of enzyme-bound products from the COR pathway (20).

Precursor feeding studies with ¹³C-labeled substrates demonstrated that CFA is a novel polyketide synthesized from C-2 and C-3 of pyruvate, three acetate units, and one butyrate unit (7). Further studies indicated that the pyruvate used for CFA biosynthesis is converted to α-ketoglutarate before incorporation into CFA (38). Since C-1 of pyruvate is not incorporated into CFA, C-1 of α-ketoglutarate must be lost at some stage in CFA biosynthesis. The most reasonable way for this loss to occur would be by decarboxylation of α-ketoglutarate before it is incorporated into CFA. Precedent for this transformation is provided by the occurrence of a thiamine-dependent α-ketoglutarate decarboxylase in the Gram-positive bacterium Leuconostoc oenos (39). The decarboxylation of α-ketoglutarate produces succinic semialdehyde, which presumably would need to be converted into a CoA ester before it could be used as the starter unit for polyketide assembly. The isolation of 2-[1-oxo-2-cyclopenten-2-ylmethyl]butanoic acid (CPE) from the fermentation broth of COR producers has been reported (40). This suggests that a cyclopentenone ring may be formed early in the biosynthetic pathway leading to CFA. These considerations allow the formulation of a working hypothesis for the role of most of the genes present in the CFA gene cluster (Fig. 4).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4

Hypothetical scheme for the biosynthesis of CFA.

The hypothetical pathway begins with conversion of succinic semialdehyde into its CoA ester. This process might require one of the two ligase genes (Cfl or Cfa5) located in the CFA gene cluster or it might utilize a gene that maps outside of the COR biosynthetic region. Cfa1 (ACP), Cfa3 (KS), and Cfa4 then are postulated to create enzyme-bound 2-carboxy-3-hydroxycyclopentanone, which subsequently is dehydrated by Cfa2 (DH) to give enzyme-bound 2-carboxy-2-cyclopentenone (CPC). Cfa4 is assigned the role of a cyclase by default since none of the ORFs in the Cfl-Cfa5 region exhibits a cyclase signature motif. A noteworthy feature of this portion of the pathway is that the ACP, KS, and DH would be used nonreiteratively, in contrast to proteins involved in the type II bacterial PKSs and FASs. Furthermore, these proteins would catalyze the potential synthesis of a unique starter, possibly CPC. This sequence of events is reminiscent of the synthesis of norsoloronic acid, in which hexanoate is utilized as a starter unit (41).

The next stage in the pathway could proceed in one of two ways. The first would involve the release of free CPC from Cfa1 (ACP) with the aid of Cfa9 (TE), followed by conversion of CPC to its CoA ester using Cfl or Cfa5. The CoA ester of CPC then could be loaded onto Cfa6, which possesses a loading domain similar to that found at the beginning of DEBS 1. Alternatively, CPC could be transferred directly from Cfa1 to Cfa6. Cfa6 contains all of the domains required for the extension of CPC by a butyrate unit followed by complete reduction at the β-keto ester position to give enzyme-bound CPE. The presence of trace amounts of CPE in the fermentation broth of P. s. glycinea could be explained by the release of CPE from Cfa6 by the TE encoded by cfa9. Transfer of enzyme-bound CPE from Cfa6 to Cfa7 would allow completion of CFA biosynthesis. Cfa7 possesses all domains required for the extension of CPE by malonate followed by reduction and dehydration of the β-keto ester; it also contains a C-terminal TE domain to unload free CFA. Cfa7 does not contain a region that can be assigned a cyclase function. The presence of enone functionality in CPE suggests that the formation of the six-membered ring of CFA probably occurs via a Michael addition rather than an aldol condensation. This, in turn, suggests that the KR and DH domains of Cfa7 may operate after the cyclization since the β-keto ester would be the preferred substrate for the Michael addition. The final stage in COR biosynthesis would involve the formation of an amide bond between CFA and CMA. It seems likely that this process is catalyzed by Cfl or Cfa5.

In summary, we have described an example of a polyketide biosynthetic pathway that requires both monofunctional and multifunctional PKS proteins and have presented a model for CFA and COR biosynthesis that correlates well with the enzymatic activities implied by sequence analysis. More detailed investigations of CFA biosynthesis using precursor incorporation experiments and assays of overproduced proteins are underway.