Streptomyces coelicolor RedP and FabH enzymes, initiating undecylprodiginine and fatty acid biosynthesis, exhibit distinct acyl-CoA and malonyl-acyl carrier protein substrate specificities


Correspondence: Kevin A. Reynolds, Office of Academic Affairs, Portland State University, PO Box 751, Portland, OR 97207-0751, USA. Tel.: +1 503 296 4829; fax: +1 503 725 5262; e-mail:


RedP is proposed to initiate undecylprodiginine biosynthesis in Streptomyces coelicolor by condensing an acyl-CoA with malonyl-ACP and is homologous to FabH that catalyzes the same reaction for initiation of fatty acid biosynthesis. Herein, we report the substrate specificities of RedP and FabH from assays using pairings of two acyl-CoA substrates (acetyl-CoA and isobutyryl-CoA) and two malonyl-ACP substrates (malonyl-RedQ and malonyl-FabC). RedP activity was observed only with a pairing of acetyl-CoA and malonyl-RedQ, consistent with its proposed role in initiating the formation of acetyl-CoA-derived prodiginines. Malonyl-FabC is not a substrate for RedP, indicating that ACP specificity is one of the factors that permit a separation between prodiginine and fatty acid biosynthetic processes. FabH demonstrated greater catalytic efficiency for isobutyryl-CoA in comparison with acetyl-CoA using malonyl-FabC, consistent with the observation that in streptomycetes, a broad mixture of fatty acids is synthesized, with those derived from branched-chain acyl-CoA starter units predominating. Diminished FabH activity was also observed using malonyl-RedQ with the same preference for isobutyryl-CoA, completing biochemical and genetic evidence that in the absence of RedP this enzyme can produce branched-chain alkyl prodiginines.


Plants and bacteria use a dissociated type II fatty acid synthase (FAS) to generate fatty acids (Heath et al., 2002). The first step in this process is catalyzed by FabH, which catalyzes a decarboxylative condensation of an acyl-CoA primer with malonyl-acyl carrier protein (ACP). The resulting 3-ketoacyl-ACP product is processed by the remaining enzymes of the type II FAS to the final elongated acyl-ACP (Fig. 1). FabH enzymes exhibit different acyl-CoA specificities. For organisms that generate only straight-chain fatty acids (such as Escherichia coli), the FabH has been shown to be specific for acetyl-CoA (Tsay et al., 1992). Many microorganisms, including bacilli and streptomycetes generate predominantly branched-chain fatty acids (Han et al., 1998). These fatty acids are generated typically using isobutyryl-CoA and methylbutyryl-CoA starter units, and FabH from some of these organisms has been shown to use these as substrates in addition to acetyl-CoA. Crystal structures of numerous FabH enzymes and examination of their acyl-binding pockets has provided a structural insight into the basis of this substrate specificity (Florova et al., 2002; Qiu et al., 2005; Sachdeva et al., 2008).

Figure 1.

Proposed role of FabH and RedP in initiating biosynthesis of fatty acids (a) and undecylprodiginine (b) in Streptomyces coelicolor. The type II FAS is responsible for subsequent conversion of 3-ketoacyl-ACP products to an acyl-ACP. A subsequent decarboxylative condensation with malonyl-ACP by either FabF or RedR elongates the acyl chain. The alkyl group (R) is typically C1–C5 and can be straight or branched.

A dramatic shift, from predominantly branched-chain fatty acids to straight-chain fatty acids, has been reported for the lipid profile of a Streptomyces coelicolor YL1 mutant, in which the natural FabH is replaced by the E. coli FabH (Li et al., 2005). This observation has provided clear evidence that the substrate specificity of a FabH plays a pivotal role in determining the type of fatty acid made by an organism.

In streptomycetes, FabH enzymes are also found in processes that generate secondary metabolites such as frenolicin, hedamycin, R1128, and undecylprodiginine (Bibb et al., 1994; Marti et al., 2000; Cerdeno et al., 2001 and Bililign et al., 2004). Undecylprodiginine, a tripyrrole red-pigmented compound, is known to exhibit a wide range of biological activities such as antibacterial, immunosuppressive, antimalarial, and anticancer (Williamson et al., 2007; Papireddy et al., 2011). For its biosynthesis in S. coelicolor, a FabH and a FabC homolog are encoded by redP and redQ in the undecylprodiginine biosynthetic gene cluster. It has been proposed that RedP catalyzes a decarboxylative condensation between acetyl-CoA and malonyl-RedQ, as the first step in generating dodecanoic acid (Fig. 1) (Cerdeno et al., 2001). This intermediate is then used to generate the alkyl side chain of the final undecylprodiginine product. A ΔredP mutant (SJM1) has been shown to produce about 80% less of this product and to produce very low levels of new branched-chain alkyl prodiginines (the straight-chain prodiginine product predominates). Evidence that in SJM1, undecylprodiginine biosynthesis is initiated by the fatty acid synthase FabH was provided by observation that higher levels of this enzyme led to a partial restoration of overall prodiginine yields (Mo et al., 2005).

The observations of fatty acid and prodiginine biosynthesis by the S. coelicolor wild type, and the YL1 and SJM1 mutants raise several questions regarding the role and specificities of RedP and FabH. For RedP, the proposed preference for acetyl-CoA and malonyl-RedQ has not been investigated. For FabH, the initial characterization of the Streptomyces glaucescens FabH (which has 100% amino acid sequence identity with S. coelicolor FabH) demonstrated comparable enzyme efficiencies for isobutyryl-CoA and acetyl-CoA. A preference for branched-chain acyl-CoA substrates would be predicted given that the corresponding long-chain fatty acids predominate in S. coelicolor (and are almost completely lost in the YL1 mutant) and that there is no evidence that these substrates are present at higher intracellular concentrations than acetyl-CoA in the cell. On the other hand, a FabH preference (or tolerance) for branched-chain acyl-CoA substrates does not readily explain why it initiates the formation of predominantly acetyl-CoA-derived prodiginines in the SJM1 mutant.

Herein reported is a characterization with respect to substrate specificity of both the S. coelicolor RedP and FabH enzymes. Kinetic studies demonstrate that RedP is specific for the straight-chain acetyl-CoA, and FabH for the branched-chain isobutyryl-CoA. Additionally, both enzymes are shown to have differing ACP specificities. These data provide answers to the questions arising from analyses of the YL1 and SJM1 mutants.

Materials and methods


[1-14C]Acetyl-CoA (60.4 mCi mmol−1) was purchased from Moravek Biochemicals, and [1-14C]isobutyryl-CoA (55 mCi mmol−1) was obtained from American Radiolabeled Chemicals Inc. Cosmids 3F7 and 4A7 containing S. coelicolor genomic DNA were kindly provided by the John Innes Institute.

Construction of expression plasmids

The redP gene was amplified from 3F7 cosmid using the forward primer 5′-CGTGCATGCATATGACCCGGGCGTCCGT-3′ and the reverse primer, 5′-GCTACTCGAGGACCGGATCGACGGCGG-3′. The scfabD gene was amplified from 4A7 using the forward primer 5′-GACTCATATGCTCGTACTCGTCGCTCC-3′ and the reverse primer 5′-GATTACTCGAGTCAGGCCTGGGTGT-3′ (restriction sites are underlined). The redP gene was cloned into expression vector pET28a to construct the plasmid pSJM3, and the scfabD gene was cloned into expression vector pET15b to give pSJM5.

Protein expression and purification

Both plasmids were used to transform E. coli BL21(DE3) cells. The resulting transformants were grown at 37 °C in LB medium containing either 50 μg mL−1 kanamycin for pSJM3 or 100 μg mL−1 ampicillin for pSJM5 to an A600 nm = 0.6, induced with 0.1 mM isopropyl-β-d-thiogalactopyranoside and incubated for approximately 12 h at 30 °C. Cells were harvested by centrifugation at 12 000 g for 10 min at 4 °C, and cell pellets were stored at −80 °C.

The appropriate E. coli cell pellets were suspended in lysis buffer-A (50 mM sodium phosphate buffer pH 7.2, 300 mM NaCl, 5 mM 2-mercaptoethanol, 10% glycerol, 0.05% (v/v) Tween-20) with 10 mM imidazole and lysozyme (1 mg mL−1). The resulting cell suspension was incubated on ice for 30 min, and cell lysate was cleared by centrifugation at 16 000 g for 20 min. The crude cell extract was loaded onto a Ni-NTA resin column. His-tagged protein was eluted using buffer-A with 300 mM imidazole. Fractions containing pure protein were pooled, exchanged with 50 mM sodium phosphate buffer pH 7.2, and stored in 20% glycerol at −80 °C.

Expression and purification of FabH, holo-FabC, and holo-RedQ were carried out in a similar way as previously described (He et al., 2000; Lobo et al., 2001; Whicher et al., 2011, respectively).

Preparation of malonyl-ACPs

The recombinant S. coelicolor His6-FabD was used to prepare malonyl-RedQ and malonyl-FabC (from holo-RedQ or holo-FabC) with a previously described protocol (He et al., 2000). The purity of each malonyl-ACP product was monitored using a microTOF-Q (QqTOF) (Bruker) mass spectrometer, with a similar method to that described previously (Whicher et al., 2011).

RedP and FabH assays

Enzyme activity was determined by monitoring conversion of radioactive acyl-CoA and malonyl-RedQ (or malonyl-FabC) substrates to a radiolabeled 3-ketoacyl-RedQ (or 3-ketoacyl-FabC) product using a standard TCA precipitation assay (Han et al., 1998). Briefly, the reaction mixture contained 50 mM sodium phosphate buffer (pH 7.2), 1 mM dithiothreitol, 40.0 μM of malonyl-RedQ (or malonyl-FabC), 40 μM [1-14C]acetyl-CoA (or [1-14C]isobutyryl-CoA), and 0.1 μg RedP (or FabH) in a final volume of 20 μL. The reaction mixture was incubated at 30 °C for 10 min and terminated by the addition of 10% (w/v) trichloroacetic acid. Precipitation was completed by incubation on ice, and the precipitate was collected by centrifugation. The pellets were resuspended in 200 μL of 2% SDS in 20 mM NaOH. The suspension was combined with scintillation fluid and analyzed with a scintillation counter. Steady-state kinetic parameters for acetyl-CoA and isobutyryl-CoA were obtained by the determination of RedP and FabH activity using various concentrations of [1-14C]acetyl-CoA (2.5–40 μM) or [1-14C]isobutyryl-CoA (0.25–10.0 μM) and a constant concentration (30 μM) of either malonyl-RedQ or malonyl-FabC. Similarly, an apparent Km for malonyl-RedQ and malonyl-FabC was obtained using a constant concentration of either 30 μM [1-14C]acetyl-CoA or 10 μM [1-14C]isobutyryl-CoA and variable concentrations of malonyl-RedQ (2.5–40 μM) and malonyl-FabC (1.0–25 μM).

Results and discussion

RedP specificity

RedP was expressed as a recombinant protein in E. coli and assayed using two acyl-CoA substrates (acetyl-CoA and isobutyryl-CoA) and two malonyl-ACP substrates (generated by FabD from RedQ and FabC using malonyl-CoA). The redQ gene has been predicted to encode a protein with ACP homology (Cerdeno et al., 2001), and is directly adjacent to redP in the prodiginine biosynthetic gene cluster, and thus the protein is a likely substrate for RedP. In contrast, the fabC gene product is unlikely to be a RedP substrate as this gene is located with fabH, fabF, and fabD in S. coelicolor (Revill et al., 1996) and other streptomycetes, and all current data indicate that this provides the ACP for fatty acid biosynthesis.

As predicted, RedP was active (Table 1) with an acetyl-CoA and malonyl-RedQ pairing (kcat 1.52 min−1, Km value for acetyl-CoA 10.35 ± 2.76 μM and for malonyl-RedQ 6.73 ± 0.31 μM). However, no detectable activities were observed with any other pairing (limit of detection was < 1% of activity observed with acetyl-CoA and malonyl-RedQ), demonstrating that neither isobutyryl-CoA nor malonyl-FabC are substrates for RedP.

Table 1. Kinetic data of RedP and FabH
Fixed concentrationsMalonyl-FabCMalonyl-RedQIsobutyryl-CoAAcetyl-CoA
Variable concentrationsIsobutyryl-CoAAcetyl-CoAIsobutyryl-CoAAcetyl-CoAMalonyl-FabCMalonyl-RedQMalonyl-FabCMalonyl-RedQ
  1. ND, not detectable.

kcat (min−1)17.171.720.660.0720.331.142.570.13
Km (μM)1.74 ± 0.328.36 ± 1.51.66 ± 0.377.13 ± 1.844.53 ± 0.877.8 ± 1.14.11 ± 1.177.93 ± 1.25
kcat/Km (μM−1min−1)9.840.200.400.014.840.140.660.017
kcat (min−1)NDNDND1.52NDNDND0.92
Km (μM)NDNDND10.35 ± 2.76NDNDND6.73 ± 0.31
kcat/Km (μM−1min−1)NDNDND0.147NDNDND0.13

These observations demonstrate a clear substrate preference for RedP and provide biochemical evidence to support the role of RedP catalyzing the first step in the biosynthesis of the undecylpyrrole component of undecylprodiginine. The specificity for acetyl-CoA plays a key role in controlling the formation of a straight-chain dodecanoyl-ACP and thus the formation of acetate-derived alkyl prodiginines in S. coelicolor. The RedP specificity for malonyl-RedQ demonstrates that the process to generate acetate-derived alkyl prodiginines via a dodecanoyl-ACP (Fig. 1) occurs using a dedicated ACP. We have recently demonstrated that RedJ is a thioesterase that can catalyze the hydrolysis of dodecanoyl-RedQ to provide dodecanoic acid (Whicher et al., 2011), and genetic evidence has shown that it is converted to undecylpyrrole by the actions of RedL and RedK (Mo et al., 2008). RedJ has been demonstrated to have much greater activity with longer-chain acyl substrates (up to C10 in length) and to efficiently discriminate between acyl-RedQ substrates and other acyl-ACPs. This ACP selectivity is thus observed at both the first (RedP) and the last step (RedJ) in the formation of dodecanoic acid for prodiginine biosynthesis and presumably plays a key role in keeping this process and the fatty acid biosynthetic process separate.

An 80% decrease in prodiginine production upon deletion of redP in S. coelicolor (SJM1) indicates that RedP is an important enzyme for prodiginine biosynthesis, but not essential (Mo et al., 2005). A significant restoration of prodiginine biosynthesis is observed in SJM1 with plasmid-based expression of FabH, indicating that FabH can function in place of RedP. The specificity of RedP and RedJ for malonyl-RedQ would predict that in order to support prodiginine biosynthesis, FabH should be able to utilize malonyl-RedQ as well as malonyl-FabC.

FabH specificity

The streptomycetes FabH was initially assayed using the E. coli ACP to generate the malonyl-ACP. The cognate ACP from streptomycetes (FabC) was not used in these assays. Isobutyryl-CoA was observed to have a threefold slower Vmax than acetyl-CoA, and a lower Km (Han et al., 1998). In this study, we sought to extend these analyses to include both the cognate ACP (malonyl-FabC) and malonyl-RedQ.

As shown in Table 1, a lower Km (1.74 μM) for isobutyryl-CoA than acetyl-CoA (8.36 μM) was also observed using malonyl-FabC. However, in this case, the overall reaction rate (kcat) was 10 times faster for isobutyryl-CoA in comparison with acetyl-CoA (Table 1 and Fig. 2). The FabH is approximately 50-fold more efficient using isobutyryl-CoA vs. acetyl-CoA using malonyl-FabC and provides a much a clearer demonstration of a) the role of the streptomycetes FabH in generating predominantly branched-chain fatty acids and b) the reason why replacement of S. coelicolor FabH with the acetyl-CoA-specific E. coli FabH (YL1/ecFabH mutant) results in a dramatic shift to a fatty acid profile of predominantly straight-chain fatty acids (Li et al., 2005).

Figure 2.

Comparison of FabH activity for isobutyryl-CoA (a) and acetyl-CoA (b) using either malonyl-FabC or malonyl-RedQ.

As predicted, FabH was able to use malonyl-RedQ in place of malonyl-FabC. Under saturating malonyl-RedQ conditions, FabH was able to use either acetyl-CoA or isobutyryl-CoA (Table 1). The Km values for each of these were comparable to those observed using malonyl-FabC, and again there was almost a 40-fold higher catalytic efficiency (kcat/Km) for isobutyryl-CoA compared to acetyl-CoA. However, for both acyl-CoA substrates, the reaction rate kcat was at least 20-fold less using malonyl-RedQ vs. malonyl-FabC (Fig. 2). At fixed isobutyryl-CoA and acetyl-CoA concentrations and variable malonyl-RedQ or malonyl-FabC concentrations, similar sets of observations were made. Greater catalytic efficiency was seen with isobutyryl-CoA relative to acetyl-CoA, and for each acyl-CoA substrate, the apparent reaction rate was much faster using malonyl-FabC than with malonyl-RedQ. This set of analyses also demonstrated that the apparent Km for malonyl-FabC (4.53 μM) and malonyl-RedQ (7.80 μM) was comparable. Thus, the difference in overall catalytic efficiency of FabH using malonyl-ACP substrates arises predominantly from differences in apparent catalytic rates rather than Km values. The ability of FabH to utilize malonyl-RedQ and to have a preference for isobutyryl-CoA is consistent with a) genetic data which suggest that FabH can initiate prodiginine biosynthesis in SJM1, the S. coelicolor redP deletion mutant, and b) the observation of a significant increase in branched-chain alkyl prodiginines in the SJM1 mutant relative to the wild-type S. coelicolor (Mo et al., 2005).

A final observation from these analyses is that the maximal kinetic efficiency of FabH (kcat/Km of 9.84 μM−1 min−1 using isobutyryl-CoA and malonyl-FabC) is 66-fold higher than that of RedP (kcat/Km of 0.147 μM−1 min−1 using acetyl-CoA and malonyl-RedQ). This difference might arise from the ability of FabH to utilize isobutyryl-CoA (the enzymes have comparable efficiencies using acetyl-CoA), or because FabH is a primary metabolic enzyme.

Role of ACP in apparent substrate specificity of FabH and RedP

Initial characterization of many FabH enzymes, including those from streptomycetes, was carried out with a commercially available E. coli ACP (Han et al., 1998; Choi et al., 2000ab; Khandekar et al., 2001). Subsequent work has revealed that these enzymes have ACP specificity. Improved catalytic activity and in some cases apparent changes in acyl group specificity can be observed when assays are performed using malonyl-ACP generated from the cognate ACP (Florova et al., 2002; Brown et al., 2005). In the case of the Mycobacterium tuberculosis FabH (mtFabH), which has a broad specificity for long-chain acyl-CoA substrates, the preferred substrate is dodecanoyl-CoA using the E. coli FAS ACP, and octadecanoyl-CoA using the M. tuberculosis FAS ACP enzyme (Brown et al., 2005). In the current study, the streptomycetes FabH has been shown to have both a much greater difference in catalytic efficiency between isobutyryl-CoA and acetyl-CoA using FabC (the cognate ACP) than was initially observed using the E. coli ACP, and a much greater catalytic efficiency using malonyl-FabC than with malonyl-RedQ. In addition, RedP has been shown to effectively discriminate between malonyl-RedQ and malonyl-FabC, using only acetyl-CoA as a substrate.

A recent model for FabH catalysis, based on experiments with the mtFabH, has indicated an open form of the enzyme, which orders around the acyl-CoA substrate and leads to the formation of an acyl-enzyme intermediate. In the case of the mtFabH, a long acyl-binding pocket to accommodate acyl chains has been identified from the X-ray crystal structure analyses. Similar structural analyses have shown a small acyl-binding pocket for the E. coli FabH, which is only able to utilize acetyl-CoA and propionyl-CoA substrates (Heath & Rock, 1996; Qiu et al., 1999; Davies et al., 2000), and a slightly larger acyl-binding pocket for the enzyme in Staphylococcus aureus, which uses branched substrates such as isobutyryl-CoA (Qiu et al., 2005). Thus, it is the acyl-binding channel which to some extent dictates FabH specificity. The data obtained in the current study would indicate that the acyl-binding channel of RedP (which utilized only acetyl-CoA) is likely to be more restrictive than the corresponding binding channel of the streptomycetes FabH enzyme (which also could utilize isobutyryl-CoA).

The mtFabH model also provides a rationale for how steps subsequent to the formation of the acyl-enzyme intermediate, involving the malonyl-ACP, also contribute to the overall catalytic reaction rate and differing reaction rates for various acyl-CoA substrates. Reaction of the acyl-enzyme intermediate with the malonyl-ACP leads to the formation of the 3-ketoacyl-ACP product and an open form of the enzyme, which permits egress of the product via binding of the acyl group to an appropriate region of the ACP (Sachdeva et al., 2008). Under certain conditions, this final step is the rate-determining step, and differences in the ability of ACPs to sequester the various acyl groups of the 3-ketoacyl-ACP products and to productively interact with the acyl-enzyme form of the FabH provide a basis for the observations regarding FabH specificity and activity. Thus, if FabC can sequester branched-chain acyl groups more effectively than the E. coli ACP, much faster reactions will be observed using this as the malonyl-ACP substrate with the streptomycetes FabH and isobutyryl-CoA. Slower overall rates observed with the streptomycetes FabH using malonyl-RedQ indicate that it can bind productively with the activated FabH, but there is a slower rate-limiting product release. The higher catalytic rate for isobutyryl-CoA under these conditions would suggest that RedQ, like FabC, has the ability to sequester straight and branched acyl chains. Finally, if malonyl-FabC cannot bind productively with the activated RedP, formation and release of a 3-ketoacyl-ACP product will only be observed with malonyl-RedQ.


This work was supported by a grant from the National Institutes of Health (GM 077147).