Abstract
Tetracyclines are clinically important aromatic polyketides whose biosynthesis is catalysed by bacterial type II polyketide synthases (PKSs). Tetracyclines are biosynthesized starting with an amide-containing malonamate starter unit and the resulting C-2 carboxyamide is critical for the antibiotic activities. In this work, we genetically verified that an amidotransferase, OxyD, and a thiolase, OxyP, are involved in the biosynthesis and incorporation of the starter unit. First, two mutations, R248T and D268N, were found to be present in OxyD* encoded in Streptomyces rimosus ATCC 13224, a strain that produces the acetate-primed 2-acetyl-2-decarboxyamido-oxytetracycline (ADOTC) instead of the malonamate-primed oxytetracycline (OTC). Homology modelling suggested that in particular D268N may inactivate OxyD. Complementation of S. rimosus ATCC 13224 with wild-type OxyD restored OTC biosynthesis, thereby confirming the essential role of OxyD in the synthesis of the amide starter unit. Second, using a series of knockout and complementation approaches, we demonstrated that OxyP is most likely involved in maintaining fidelity of the amide-priming process via hydrolysis of the competing acetate priming starter units. While the inactivation of OxyP does not eliminate OTC biosynthesis, the ratio of acetate-primed ADOTC to malonamate-primed OTC is significantly increased. This suggests that OxyP plays an ancillary role in OTC biosynthesis and is important for minimizing the levels of ADOTC, a shunt product that has much weaker antibiotic activities than OTC.
-
Supplementary material is available with the online version of this paper.
-
Edited by: K. Flardh
Introduction
Tetracyclines are broad-spectrum antibiotics that have antimicrobial activities toward both Gram-positive and Gram-negative bacteria (Chopra & Roberts, 2001). To combat the emergence of antimicrobial resistance mechanisms, the 2-naphthacenecarboxyamide carbon skeleton of the polyketide natural product has served as a useful starting point for the discovery of second- and third-generation tetracyclines, such as minocycline (Church et al., 1971), doxycycline (Martell & Boothe, 1967) and tigecycline (Sum et al., 1994). These sustained successes suggest that the tetracyclines indeed have an ‘evolutionarily privileged scaffold’ and may continue to be the starting point for generation of new antibiotics.
Natural tetracyclines such as oxytetracycline (OTC) and chlortetracycline (CTC) are aromatic polyketides synthesized by bacterial type II polyketide synthases (PKSs) from malonyl-CoA building blocks, through successive Claisen-like decarboxylative condensations (Hertweck et al., 2007). The minimal PKS that synthesizes the polyketide backbone comprises a ketosynthase-chain length factor heterodimer (KS-CLF), an acyl carrier protein (ACP) and a malonyl-CoA : ACP transacylase (MAT) (Hertweck et al., 2007). Following canonical PKS biosynthetic logic, a set of cyclases specifically cyclize the polyketide backbone regioselectively, followed by a dedicated set of tailoring enzymes that produce the richly substituted, heavily oxidized tetracycline molecule. However, the biosynthesis of tetracyclines differs from that of the other aromatic polyketide families in one significant way: the starter unit of the polyketide chain is a malonamate instead of the commonly found acetate or fatty acids (Thomas & Williams, 1983). The malonamate group is reflected in the final tetracycline structures as the C-2 amide, a structural feature that is conserved among all tetracyclines and is critical for the antibiotic activities of this family (Lykkeberg et al., 2004).
The biosynthesis and incorporation of the malonamate starter unit in tetracyclines have remained unresolved. It is believed that the malonamate is incorporated as a primer unit in the form of either malonamyl-CoA or malonamyl-ACP, which can be derived from the amidation of malonyl-CoA or malonyl-ACP, respectively, by the action of an amidotransferase (McDowall et al., 1991; Thomas & Williams, 1983). Our previous heterologous experiments with the oxytetracycline (oxy gene cluster) minimal PKS in Streptomyces coelicolor demonstrated that an ATP-dependent amidotransferase, OxyD, is required in addition to the minimal PKS (OxyAB and OxyC) to generate the amidated backbone (Zhang et al., 2006). Despite this finding, the exact substrate of OxyD has not been confirmed, and more importantly, its function in the native OTC-producing S. rimosus has not been confirmed genetically. If the role of OxyD in S. rimosus is indeed that of a malonamyl synthetase, we hypothesize that mutations in OxyD in S. rimosus should lead to a tetracycline-like shunt product, 2-acetyl-2-decarboxyamidooxytetracycline (ADOTC) in which the C-2 carboxyamide is replaced with an acetate unit. In fact, ADOTC is produced as the only tetracycline-like compound by S. rimosus ATCC 13224 (Tanner et al., 1962) and Streptomyces psammoticus (Lancini & Sensi, 1964). S. rimosus ATCC 13224 is a mutant of S. rimosus ATCC 10970 isolated via random mutagenesis (Tanner et al., 1962). However, the genetic difference between these strains and the OTC-producing S. rimosus has not been elucidated.
From the sequenced oxy gene cluster in S. rimosus, an additional gene, oxyP, was also proposed to be involved in the priming steps during OTC biosynthesis. OxyP is a putative acyltransferase that displays strong sequence similarity to ZhuC from the R1128 biosynthetic pathway. ZhuC was determined in vitro to be an acetyl-ACP thioesterase that removes the competing acetate-containing starter unit to enable priming of the R1128 minimal PKS by the various medium-chain fatty acids (Tang et al., 2004). Therefore, OxyP was proposed to perform a similar function in S. rimosus to attenuate the acetate priming pathway that may lead to the biosynthesis of ADOTC (Zhang et al., 2006). However, heterologous experiments with the oxy minimal PKS and OxyD did not show any marked difference in product phenotypes with or without coexpression of OxyP. Therefore, loss-of-function genetic characterization of OxyP in S. rimosus may provide the most direct insight into whether OxyP is involved in the priming steps.
In this study, we performed genetic analysis of oxyD and oxyP in S. rimosus. We show here that both OxyD and OxyP are intricately involved in the priming steps of OTC biosynthesis. While OxyD is an essential enzyme required for the synthesis of the amide starter unit, OxyP appears to function as an ancillary enzyme that suppresses the competing acetate priming mechanism (Fig. 1).
Priming mechanism of OTC and ADOTC biosynthesis in S. rimosus. OxyD is involved in the amidation of malonyl-CoA or malonyl-ACP, leading to the synthesis of a malonamate-containing starter unit. The amide of the malonamate primer is incorporated into OTC as the C-2 amide. When acetate is used as the starter unit, the biosynthetic pathway produces ADOTC as a shunt product. OxyP works as a thiolase that eliminates the acetyl-ACP starter units to eliminate ADOTC biosynthesis. Alternatively, acetoacetate could be used to initiate the synthesis of ADOTC followed by eight iterative condensation reactions (Fu et al., 1994).
Methods
Culture techniques and genetic manipulations.
Escherichia coli strains were cultivated on Luria–Bertani (LB) agar or in LB liquid at 37 °C for subcloning and plasmid manipulations. Streptomyces strains were grown on SFM plates (2 % mannitol, 2 % soybean powder, 2 % agar, pH 7.2) for conjugation between E. coli ET12567 and Streptomyces strains according to Kieser et al. (2000). For E. coli, kanamycin and ampicillin were used at 35 µg ml−1 and 100 µg ml−1, respectively, in both solid and liquid LB medium. For Streptomyces, apramycin and thiostrepton were used at 50 µg ml−1 and 25 µg ml−1 respectively. Genomic DNA extraction was performed as described by Kieser et al. (2000).
Construction of plasmids.
PCR was performed with Platinum Pfx DNA polymerase (Invitrogen) using cosmid pYT264 as template (Zhang et al., 2006). The PCR products were ligated into pCR-Blunt vector (Invitrogen) for further cloning. The recombined DNA fragments were cloned into pJTU870 shuttle vector for gene inactivation and into pJTU1278 shuttle vector (He et al., 2010) for complementation experiments. Ligation was performed by using T4 DNA ligase (Invitrogen). Synthesis of oligonucleotide primers and DNA sequencing were performed by Integrated DNA Technologies and Laragen respectively.
Complementation of S. rimosus ATCC 13224 with oxyD from S. rimosus ATCC 10970.
The oxyD gene was recovered from a XbaI–NheI fragment of pWJ7 and cloned into the XbaI site of pPW189 to replace oxyP and generate pDP12. Then pDP12 was introduced into S. rimosus ATCC 13224 via conjugation to generate ATCC 13224/pDP12. The exconjugants were selected with thiostrepton. The presence of the plasmid in ATCC 13224/pDP12 was confirmed via PCR (see Fig. S1 in the supplementary material available with the online version of this paper). The plasmid was extracted from the complemented S. rimosus strain and retransformed into E. coli XL-1. High concentrations of plasmid were prepared from the transformed E. coli strain and used for check digests.
Inactivation of oxyP.
The vector pPW173 was constructed by inserting a 12.7 kb AfeI–XmnI fragment of pYT264 into pCR-Blunt. This plasmid was transformed into E. coli BW25113/pIJ790 by electroporation. The acc(3)IV cassette, amplified from pIJ773 using primers oxyP-KO-F (5′-CTGCTGCTGCCCGGCCAGGGCTCCCAGTACCGGCGGATGATTCCGGGGATCCGTCGACC-3′) and oxyP-KO-R (5′-ATCCTCCGGCCCATCCGCCCGCTTAGGTGACAGCGGTACTGTAGGCTGGAGCTGCTTC-3′), was used to replace an 825 bp oxyP internal portion in pPW173 to generate pPW174 by ReDirect technology (Gust et al., 2003). A 13.3 kb EcoRI fragment from pPW174 was cloned into the same site of pJTU870 to yield pPW177. pPW177 was introduced into S. rimosus ATCC 10970 by conjugation from E. coli ET 12567/pUZ8002 (Paget et al., 1999). A set of putative mutants, termed WP3, were isolated by cultivating exconjugant strains without selection for the plasmid vector. This allowed segregation of plasmid-free derivatives, and those that still retained the aac(3)IV marker were identified. In this way, mutants were obtained in which the oxyP allele on the chromosome had been exchanged for the deletion allele that was on the plasmid. The mutations were confirmed by PCR using primers oxyP-T1 (5′-TCCTGTACCGGGCGAAACTG-3′) and oxyP-T2 (5′-ATCTCGGCATCCTCGTCGCA-3′) and Southern blot analysis (Supplementary Fig. S2).
Complementation of the ΔoxyP mutant strain.
The oxyP gene was recovered from a XbaI fragment of pYT313 and cloned into the same site of pPW167 to replace ssfL1 and generate pPW189, containing the ermE* promoter and oxyP. The ermE* promoter was recovered from a KpnI–EcoRI fragment of pUWL201pW (Doumith et al., 2000) and cloned into the same sites of pJTU1278. The oxyP* gene was generated via site-directed PCR mutagenesis. Mutation in oxyP was introduced with primers oxyP*-1 (5′-GCTCGGCCACGCCATCGGCGAGATGGCGGCCGCC-3′) and oxyP*-2 (5′-TCTCGCCGATGGCGTGGCCGAGCAGCGCCACCGG-3′). oxyP* was ligated into pCR-Blunt to generate pDP13. A 1.0 kb DNA fragment containing oxyP* was recovered from a XbaI–NheI fragment of pDP13 and ligated into the XbaI site of pPW189 to replace oxyP and generate pDP33. Both pPW189 and pDP33 were introduced independently into WP3 by conjugation to generate WP3/pDP189 and WP3/pDP33, respectively. The exconjugants were selected with thiostrepton. The presence of the respective plasmids in strains WP3/pPW189 and WP3/pDP33 was confirmed via PCR (Supplementary Fig. S1) followed by plasmid extraction and digestion as described above for pDP12.
Overexpression of OxyP, ZhuC and OxyD in S. rimosus ATCC 10970.
The zhuC gene was recovered from a XbaI–SpeI fragment of pYT106 and ligated into the XbaI site of pPW189 to replace oxyP and generate pDP1, containing the ermE* promoter and zhuC. oxyP was recovered from a XbaI–SpeI fragment of pYT313 and cloned into the XbaI site of pDP12 to generate pDP83, containing the ermE* promoter and oxyP and oxyD. Plasmids pPW189, pDP1, pDP12 and pDP83 were independently introduced into S. rimosus ATCC 10970 via conjugation. The exconjugants were selected with thiostrepton.
HPLC analysis of metabolites.
A quarter of a well-grown plate was chopped into fine pieces and extracted with 6 ml ethyl acetate/methanol/acetic acid (89%/10%/1%). One millilitre of extract was dried in vacuo and dissolved in 50 µl methanol for LC-MS analysis. Analyses were performed on a Shimadzu 2010 EV liquid chromatography mass spectrometer by using positive and negative electrospray ionization and a Phenomenex Luna 5 µm, 2.0×100 mm C18 reverse-phase column with a linear gradient of 5 % CH3CN in water (0.1 % formic acid) to 95 % CH3CN in water (0.1 % formic acid) for 30 min at a flow rate of 0.1 ml min−1.
Results
Genetic basis of ADOTC biosynthesis
To investigate the role of OxyD in OTC biosynthesis in the producing host S. rimosus ATCC 10970, we first attempted to determine the genetic basis for the exclusive biosynthesis of ADOTC in S. rimosus ATCC 13224. We hypothesized that since the functions of the minimal PKS and remaining tetracycline tailoring modifications are all intact to synthesize ADOTC, the genetic difference that contributes to the lack of the amide starter unit may be found in oxyD. Determining the genetic and biochemical basis for biosynthesis of the acetate-primed tetracyclines also has practical implications, as ADOTC and 2-acetyl-2-decarboxyamidotetracycline (ADTC) are found as undesirable impurities during fermentation of OTC (Kersey, 1950) and CTC producers (Hochstein et al., 1960), respectively. ADOTC only exhibits 10 % of the antimicrobial activity of OTC (Hochstein et al., 1960).
The genomic DNA of S. rimosus ATCC 13224 was isolated and regions of the oxy gene cluster were sequenced, including non-coding and regulatory sequences. Genes encoding biosynthetic enzymes that may be involved in starter unit biosynthesis and selection were sequenced fully, including oxyA, oxyB, oxyC, oxyD and oxyP. While the other gene sequences all matched that reported for the oxy gene cluster from S. rimosus ATCC 10970, the oxyD gene from the ADOTC producer (oxyD*) was found to contain two nonconserved mutations: positions 248 and 268 were mutated from Arg and Asp to Thr and Asn, respectively.
OxyD is homologous to the E. coli asparagine synthetases B (AsnB) (PDB ID 1CT9, 21 % sequence identity), which synthesizes asparagines from aspartate in an ATP-dependent fashion using glutamine or ammonia as the amine donor (Milman & Cooney, 1979). Sequence alignment of OxyD (or OxyD*) with eleven different AsnB-like amidotransferases (Supplementary Fig. S3a) confirms the highly conserved catalytic residues (Cys2, Arg50, Glu80, Asp103, Leu263 and Gly377 (Larsen et al., 1999), with Asp268 of OxyD located in a conserved region ‘LLSGGLDSS’. We further constructed a homology structure model of OxyD using the program PHYRE (Protein Homology/analogY Recognition Engine) (Bennett-Lovsey et al., 2008) based on the crystal structure of E. coli AsnB (Supplementary Fig. S4). Comparison between the OxyD homology structure and the AsnB structure shows that Asp268 and the conserved patch reside in the AMP-binding pocket. As is proposed for AsnB (Larsen et al., 1999), it is reasonable to speculate that the electrostatic interaction between the carboxylate group of Asp268 and the phosphoryl oxygen of the AMP moiety via a metal ion such as Mg2+ is essential for AMP binding. The Asp268 to Asn268 mutation therefore may eliminate the electrostatic interaction, resulting in the loss of function of OxyD*. On the other hand, Arg248 is mapped to an α-helix far away from the OxyD catalytic site and is relatively unconserved in the sequence alignment. The mutation Arg248 to Thr248 therefore may not impair the activities of OxyD.
The observed mutations in oxyD* in S. rimosus ATCC 13224 therefore very probably abolished the biosynthesis of the malonamate starter unit and led to the biosynthesis of ADOTC. To confirm this genetically, we constructed a complementation vector pDP12 which harbours the wild-type oxyD from S. rimosus ATCC 10970 under control of the ermE* promoter. Conjugation of pDP12 into S. rimosus ATCC 13224 resulted in the strain ATCC 13224/pDP12. ATCC 13224/pDP12 was cultured in parallel with S. rimosus ATCC 10970 and S. rimosus ATCC 13224 and analysed for polyketide production. Whereas S. rimosus ATCC 10970 produced mostly OTC, with 10 % ADOTC (Fig. 2a, Supplementary Fig. S5), S. rimosus ATCC 13224 produced no traces of OTC and entirely ADOTC (Fig. 2e). Analysis of the organic extract from ATCC 13224/pDP12 showed that OTC production was restored to a level comparable to that of the S. rimosus ATCC 10970 extract (Fig. 2f), thereby unambiguously confirming the essential role of OxyD in the OTC biosynthetic pathway. This result also revealed that the genetic basis of ADOTC-producing mutants is entirely due to deleterious point mutations in oxyD*.
HPLC analysis of OTC and ADOTC production from organic extracts from different S. rimosus strains: (a) ATCC 10970; (b) WP3; (c) WP3/pPW189; (d) WP3/pDP33; (e) ATCC 13224; (f) ATCC 13224/pDP12; (g) OTC and ADOTC standards.
The ancillary role of OxyP in OTC biosynthesis
The product profile trace shown in Fig. 2(a) suggested that even with a functioning OxyD, the priming of the oxy minimal PKS by the malonamate starter unit is incomplete and ~10 % of the products remain acetate-primed. This indicates that acetate and malonamate can prime the KS-CLF in a competitive fashion. In order to suppress the competitive binding of acetate during biosynthesis of the polyketide R1128, the gene cluster encodes an acetyl-ACP thiolase, ZhuC, that rapidly eliminates the acetyl-ACP starter units (Tang et al., 2004). In the absence of ZhuC, the KS-CLF is exclusively primed with acetate instead of the medium-chain fatty acyl starter units, suggesting an essential role of ZhuC in enabling loading of the nonacetate starter units. In the oxy cluster, oxyP was proposed to encode a ZhuC homologue that may perform a similar function during the priming steps of OTC biosynthesis and suppress ADOTC biosynthesis (Zhang et al., 2006). Therefore, confirming the role of OxyP may lead to useful strategies to completely suppress the ADOTC levels seen in Fig. 2(a).
To confirm the function of OxyP, we inactivated oxyP in S. rimosus ATCC 10970 by homologous recombination, generating a set of mutants termed WP3. The deletion of oxyP and introduction of the aac(3)IV selectable marker in three independent mutants were confirmed by PCR amplification (Fig. 3b). These WP3 mutants were then cultured in parallel with S. rimosus ATCC 10970 and the metabolites were extracted and separated by LC-MS. As shown in Fig. 2(b), inactivation of oxyP resulted in an increase in titre of ADOTC with a corresponding decrease in titre of OTC in WP3. Quantification of the metabolites based on mass ion intensity (Fig. S5b) showed that the OTC : ADOTC ratio decreased from ~11 : 1 to ~3 : 1 upon deletion of oxyP while the total amount of OTC and ADOTC remained the same, which corresponds to an approximately threefold increase in the level of ADOTC.
Inactivation and complementation of oxyP. (a) Schematic representation of oxyP inactivation; (b) PCR confirmation of oxyP mutants. Lanes: 1, 1 kb plus ladder; 2, S. rimosus ATCC 10970; 3–5, three independent mutants WP3-1; WP3-2; WP3-3. (c) Ratio of OTC to ADOTC in S. rimosus ATCC 10970, WP3, WP3/pPW189 and WP3/pDP33.
To confirm that the increase in amount of acetate-primed polyketides is indeed due to the loss of OxyP function, we constructed a complementation vector pPW189 in which oxyP was placed under the control of the ermE* promoter. Integration of pPW189 into WP3 resulted in the strain WP3/pDP189, and complementation of oxyP resulted in partial restoration of the titre of OTC relative to ADOTC with a ratio of ~5 : 1 (Figs 2c and 3c). Finally, to demonstrate that OxyP plays a catalytic role in suppressing the acetate starter unit, we performed site-directed mutagenesis of the OxyP active site. OxyP contains the signature thioesterase/acyltransferase active-site motif GxSxG, in which the side-chain hydroxyl of serine serves as the nucleophile in the proposed hydrolysis of acetyl-ACP (Serre et al., 1995). We introduced the mutant oxyP* gene, in which the active-site serine was mutated to an alanine, into the ΔoxyP strain WP3 using the same complementation strategy as described above to generate WP3/pDP33. As expected, the catalytically inactivated OxyP* was unable to suppress acetate priming and the level of OTC to ADOTC remained at ~3 : 1 (Fig. 2d). These results indicate that OxyP plays an ancillary role in the biosynthesis of OTC. Unlike ZhuC, OxyP is not an essential enzyme for priming of the oxy minimal PKS with the malonamate starter unit. However, the presence of functional copies of OxyP significantly suppresses the biosynthesis of acetate-primed ADOTC in favour of malonamate-primed OTC. Therefore, the oxy PKS utilizes a similar proofreading strategy to remove competing species and ensure that only the correct starter unit is loaded on the PKS machinery for the subsequent elaborate biosynthetic pathway.
Attempts to minimize ADOTC production in S. rimosus ATCC 10970
The above results indicated that two enzymes, OxyD and OxyP, are important for the biosynthesis of OTC in S. rimosus. Therefore, manipulation of the expression levels of these two enzymes in S. rimosus ATCC 10970 may lead to an engineered strain that has further suppressed levels of ADOTC, a phenotype that is desirable for the fermentative production of OTC. We constructed four different overexpression vectors attempting to minimize ADOTC levels (Table 1): pPW189 (overexpression of OxyP in an attempt to further eliminate the residual competing acetyl-ACP species; pDP1 (overexpression of ZhuC, which was shown to be highly efficient as an acetyl-ACP thiolase in vitro); pDP12 (overexpression of OxyD to increase the level of malonamate starter unit); and pDP83 (simultaneous overexpression of both OxyP and OxyD to achieve a combined maximum in malonamate priming). In each case, the genes were placed under control of the ermE* promoter and conjugated into S. rimosus ATCC 10970. Unfortunately, the relative levels of OTC and ADOTC remained unchanged at ~11 : 1 in all of these engineered strains based on LC-MS analysis (data not shown).
Discussion
One of the key sources of structural variations in type II aromatic polyketides is starter unit selection. While many minimal PKSs use acetate, in the form of acetyl-ACP, to initiate polyketide biosynthesis, a large number of minimal PKSs, such as those from the pathways of enterocin (Piel et al., 2000), R1128 (Marti et al., 2000), doxorubicin (Ye et al., 1994), hedamycin (Bililign et al., 2004; Das & Khosla, 2009) and fredericamycin (Wendt-Pienkowski et al., 2005), have been found to use a variety of non-acetate starter units. In most cases, a dedicated set of enzymes is employed not only to synthesize, but also to aid in the incorporation of the non-acetate acyl starter unit (Hertweck, 2009). One of the unique structural features of tetracyclines is that this family of compounds all use malonamate, a highly polar starter unit, to prime polyketide biosynthesis. The amide unit is important for the observed antibacterial activities of tetracyclines, evident by the 10-fold decrease in bioactivity of ADOTC when compared to OTC (Hochstein et al., 1960). In this study, we used genetics to confirm that two enzymes, OxyD and OxyP, constitute the initiation unit of the oxy pathway. OxyD is an essential amidotransferase in the biosynthesis of the malonamate starter unit, as two mutations, D268N and R248T, in OxyD* completely abolished OTC biosynthesis and resulted in only ADOTC biosynthesis in S. rimosus ATCC 13224. On the other hand, OxyP is not absolutely essential for OTC biosynthesis; however, its thiolase activity towards acetyl-ACP significantly attenuates the levels of the undesirable ADOTC.
OxyD belongs to the broader family of N-terminal nucleophile enzymes in which an N-terminal cysteine is the nucleophile that hydrolyses l-glutamine to supply a free amine group for the formation of asparagine from aspartic acid (Larsen et al., 1999) by the action of a C-terminal amide synthase domain (Richards & Schuster, 1998). Homologues of OxyD are found in a number of type II PKS pathways, including those of pradimicin (Kim et al., 2007; Zhan et al., 2009), rubromycin (Martin et al., 2001) and fredericamycin (Chen et al., 2010). However, in these gene clusters, the role of the amidotransferase is to amidate the free acid moiety after chain release with either free amine or an amino acid. The role of OxyD is therefore clearly different in OTC biosynthesis. While the involvement of OxyD in the biosynthesis of an amidated polyketide was demonstrated in a heterologous host (Zhang et al., 2006), we unambiguously confirmed the necessity of OxyD in OTC biosynthesis by locating the genotype of the ADOTC-producing S. rimosus strain to a mutant oxyD*. We identified two mutations, D268N and R248T, in OxyD* that may affect the binding of AMP to the active site. Complementation of the mutant with an intact oxyD gene completely restored the ability of the mutant strain to synthesize OTC. Our work therefore reveals that the ADOTC-producing S. rimosus ATCC 13224 is mutated in oxyD compared to the wild-type S. rimosus ATCC 10970.
OxyP displays high sequence homology to ZhuC (50 % identity) (Tang et al., 2004) and other acyltransferase-like enzymes in different type II PKSs, including DpsD (54 % identity) (Grimm et al., 1994), San2 (52 % identity) (Zaleta-Rivera et al., 2010) and SsfV (48 % identity) (Pickens et al., 2009). While the acetyl-ACP thiolase role of OxyP was first proposed when the gene cluster was sequenced, its function was not confirmed through heterologous reconstitution experiments. Exclusion of oxyP from the minimal PKS and oxyD did not lead to attenuated levels of amidated polyketides. Expression of OxyP from either E. coli or Streptomyces lividans was not successful, thereby precluding a direct assay of OxyP activities as was performed with ZhuC. The effect of OxyP on the relative levels of OTC and ADOTC was measured systematically by (1) deletion of oxyP; (2) complementation of ΔoxyP with oxyP; and (3) complementation of ΔoxyP with a catalytically inactivated oxyP*. We demonstrated that removal of OxyP activities led to significant accumulation of the acetate-primed ADOTC, while the nucleophilic serine in the conserved thiolase active site is essential for oxyP complementation. Taken together, these results confirmed the analogous role of OxyP in suppressing acetate priming in the oxy pathway to that of ZhuC in the R1128 pathway, and the recently confirmed EncL in the enterocin pathway (Kalaitzis et al., 2011). OxyP was, in contrast to ZhuC, not essential for OTC biosynthesis, suggesting that the minimal PKS has high enough affinity for the malonamyl starter unit, even in the presence of a competing acetyl starter unit. Indeed, homologues of OxyP are not always found in tetracycline biosynthetic pathways. While the SF2575 pathway contains a copy of the oxyP gene (Pickens et al., 2009), the closely related CTC biosynthetic pathway does not include an OxyP equivalent (Ryan et al., 1996). In the latter case, the CTC minimal PKS may have a very high specificity for the malonamyl starter unit, and no thiolase activity is required to remove the acetyl units. Attempts to minimize ADOTC production in S. rimosus ATCC 10970 via overexpresion of OxyD and OxyP did not decrease the relative levels of OTC and ADOTC compared to the wild-type strain. Under in vivo conditions, the potential protein–protein interactions between OxyD, OxyP and other PKS components may determine the final ratio of OTC to ADOTC. Therefore, overexpression of OxyD and OxyP may not alone be sufficient to further improve the ratio of OTC to ADOTC in S. rimosus ATCC 10970. It is also likely that the OTC minimal PKS may have retained a strong preference for the acetyl starter unit. Therefore, given the high abundance of acetyl species in the cell, complete elimination of these competing starter units is impossible and acetyl priming cannot be avoided.
In conclusion, the work reported here provides an additional genetic basis for the origin of the C-2 carboxyamide moiety in the tetracycline family of compounds. Our findings reveal a microbial strategy to balance the levels of the desirable malonamyl primer units versus the undesirable, but always abundant, acetyl primer units.
Acknowledgements
We thank Lauren B. Pickens for critical reading of our manuscript and insightful discussion. This work was supported by NSF CBET grants 0545860 and 1033070.