|Home | About | Journals | Submit | Contact Us | Français|
Genome-wide analysis has revealed abundant FabG (β-ketoacyl-ACP reductase) paralogs, with uncharacterized biological functions, in several halophilic archaea. In this study, we identified for the first time that the fabG1 gene, but not the other five fabG paralogs, encodes the polyhydroxyalkanoate (PHA)-specific acetoacetyl coenzyme A (acetoacetyl-CoA) reductase in Haloarcula hispanica. Although all of the paralogous fabG genes were actively transcribed, only disruption or knockout of fabG1 abolished PHA synthesis, and complementation of the ΔfabG1 mutant with the fabG1 gene restored both PHA synthesis capability and the NADPH-dependent acetoacetyl-CoA reductase activity. In addition, heterologous coexpression of the PHA synthase genes (phaEC) together with fabG1, but not its five paralogs, reconstructed the PHA biosynthetic pathway in Haloferax volcanii, a PHA-defective haloarchaeon. Taken together, our results indicate that FabG1 in H. hispanica, and possibly its counterpart in Haloarcula marismortui, has evolved the distinct function of supplying precursors for PHA biosynthesis, like PhaB in bacteria. Hence, we suggest the renaming of FabG1 in both genomes as PhaB, the PHA-specific acetoacetyl-CoA reductase of halophilic archaea.
Several haloarchaeal species belonging to the genera Haloferax, Haloarcula, Natrialba, and Haloquadratum are capable of synthesizing short-chain-length polyhydroxyalkanoates (SCL-PHAs) (6, 8, 14, 16), a large family of biopolymers with desirable biodegradability, biocompatibility, and thermoplastic features (31). Although the metabolic pathways of PHAs in bacteria have been characterized in detail (10, 15, 20, 25, 26, 37), the genes involved in PHA biosynthesis in haloarchaea were not recognized until recently, when the PHA synthase genes were identified and characterized for Haloarcula marismortui, Haloarcula hispanica, and Haloferax mediterranei (6, 19). These archaeal PHA synthases are all composed of two subunits, PhaE and PhaC. They are homologous to the class III PHA synthases from bacteria but have a longer C-terminal extension in the PhaC subunit. Nevertheless, the pathway of supplying the PHA precursors has not yet been clarified for any haloarchaeal strain.
Both H. mediterranei and H. hispanica are able to synthesize poly(3-hydroxybutyrate-co-3-hydroxyvalerate) (PHBV) from unrelated carbon sources, despite the content of the (R)-3-hydroxyvalerate (3-HV) monomer of PHBV in H. mediterranei (10 to 13 mol%) (4, 19) being much higher than that in H. hispanica (~3 mol%) (19). Conversely, the bacteria Ralstonia eutropha and Synechocystis sp. strain PCC6803, which possess class I and III PHA synthases, respectively, accumulate just poly(3-hydroxybutyrate) (PHB) when the 3-HV-related carbon sources (i.e., propionate and valerate) are not supplied (30). In these two bacteria, the biosynthesis of the (R)-3-hydroxybutyrate coenzyme A [(R)-3-HB-CoA] precursor is conducted by two steps. First, two acetyl-CoA molecules are condensed into one acetoacetyl-CoA molecule by the enzyme β-ketothiolase (PhaA). The acetoacetyl-CoA is then reduced to (R)-3-HB-CoA by a PHA-specific acetoacetyl-CoA reductase (PhaB). The resulting (R)-3-HB-CoA is subsequently incorporated into PHB, catalyzed by PHA synthases (26, 36).
Both PhaB and FabG belong to the short-chain dehydrogenase/reductase (SDR) superfamily, whose members are homologous in sequence and have several conserved motifs (27, 29). Interestingly, although FabGs naturally reduce 3-ketoacyl-ACP to form (R)-3-hydroxyacyl-ACP in fatty acid biosynthesis, a few FabGs also recognize 3-ketoacyl-CoA and hence function in PHA biosynthesis. For example, the FabG proteins of Escherichia coli and Pseudomonas aeruginosa have been demonstrated to supply precursors for PHA biosynthesis in recombinant E. coli cells (21, 22, 32, 35). In addition, several FabG paralogs may have evolved a distinct function, to be responsible only for PHA accumulation. This situation was observed in Synechocystis sp. strain PCC6803, where the originally annotated FabG (12) was renamed PhaB after an understanding of its function in PHA biosynthesis (36).
Genome-wide analysis of H. marismortui ATCC 43049 (1) revealed eight FabG paralogs in this haloarchaeon. Similarly, multiple fabG paralog genes (fabG1 to fabG6) were also observed in the newly sequenced genome of H. hispanica (our unpublished data). In this study, we demonstrate that fabG1, but not the other five fabG paralogs, encodes the PHA-specific acetoacetyl-CoA reductase in H. hispanica. It is responsible for providing (R)-3-HB-CoA for PHA biosynthesis in Haloarcula species, and interestingly, this enzyme also functions well in Haloferax volcanii, endowing this PHA-defective strain with the ability to accumulate PHA when cotransformed with PHA synthase genes.
The strains and plasmids used in this study are listed in Table Table1.1. The primers are listed in Table Table2.2. Plasmids used for gene disruption/knockout and expression were derived from pUBP (6) and pWL102 (13) (Table (Table1),1), respectively. These plasmids were usually constructed in E. coli JM109 and then introduced into H. hispanica or H. volcanii (39) by a polyethylene glycol-mediated transformation method (3).
E. coli JM109 was grown in Luria-Bertani (LB) medium at 37°C (33). When needed, ampicillin was added to a final concentration of 100 mg/liter. Generally, H. hispanica and H. volcanii strains were cultivated at 37°C in a nutrient-rich medium (AS-168) (6). For PHA accumulation analysis, H. hispanica and H. volcanii cells were first cultured at 37°C for 48 h in AS-168 medium, and then 5% (vol/vol) inocula were transferred to 100 ml MG medium (10 g/liter glucose in minimal medium) in shaking flasks (6) and cultivated for an additional 96 h. The pH was maintained manually at about 7.2 in MG medium. When required, mevinolin was added to a final concentration of 5 mg/liter for H. volcanii and H. hispanica transformants.
Total RNA of H. hispanica AS2049 was extracted with Trizol reagent (Gibco BRL) from cells grown in MG medium for 72 h. Six primer pairs (fabG1RTF-fabG1RTR, fabG2RTF-fabG2RTR, fabG3RTF-fabG3RTR, fabG4RTF-fabG4RTR, fabG5RTF-fabG5RTR, and fabG6RTF-fabG6RTR) were designed specifically for the fabG1 to fabG6 genes, respectively (Table (Table2).2). The RNA samples were treated with RNase-free RQ1 DNase (Promega) to eliminate any DNA contamination, which was confirmed by a control PCR without prior reverse transcription (RT). The DNA-free RNA samples were then used as templates for RT-PCR with a OneStep RT-PCR kit (Qiagen) according to the manufacturer's instructions.
To distinguish the functions of the FabG paralogs in H. hispanica during PHA synthesis, the six fabG genes were disrupted by a single-crossover method. Briefly, a 475-bp DNA fragment located in the middle of the fabG1 gene was amplified with primer pair fabG1SF-fabG1SR (Table (Table2).2). The PCR product was sequenced and inserted into the suicide plasmid pUBP (Table (Table1).1). The resulting plasmid, pUBPSG1, was introduced into H. hispanica to disrupt the fabG1 gene by single-crossover homologous recombination. The resulting mutant strain was named H. hispanica fabG1-1. Similarly, the other five fabG-disrupted mutants, named H. hispanica fabG2-1 to fabG6-1, were constructed as described above, using primer pairs fabG2SF-fabG2SR, fabG3SF-fabG3SR, fabG4SF-fabG4SR, fabG5SF-fabG5SR, and fabG6SF-fabG6SR (Table (Table2),2), respectively.
To construct an in-frame fabG1 deletion mutant, a 465-bp DNA fragment spanning the 5′ region upstream of the fabG1 gene and a 467-bp fragment spanning the 3′ region downstream of the fabG1 gene were amplified by use of primer pairs fabG1F2-fabG1R2 and fabG1F3-fabG1R3 (Table (Table2),2), respectively. These two PCR products were sequenced and cloned into the plasmid pUBP (Table (Table1).1). The resulting plasmid, pUBPDG1 (Table (Table1),1), was then introduced into H. hispanica to knock out the fabG1 gene by double-crossover homologous recombination, generating a fabG1-deleted strain named H. hispanica ΔfabG1. Primers fabG1F2 and fabG1R3 were used for identification of the single-crossover and double-crossover mutants by PCR analysis. To construct the complementation strain of H. hispanica ΔfabG1, the fabG1 gene was amplified with primers fabG1F1 and fabG1R1 and cloned into pL52 under the control of the hsp5 promoter from Halobacterium (18). The resulting plasmid, pWLG1, was confirmed by sequencing and introduced into H. hispanica ΔfabG1 to check if the capability of PHA accumulation was restored.
For coexpression of fabG1 with PHA synthase genes, the phaEC genes of H. hispanica, including the native promoter, were amplified with primers phaECF and phaECR and inserted into plasmid pWLG1, resulting in plasmid pWLG1EC. For the fabG2 to fabG6 genes, the respective coding sequences were amplified from H. hispanica by PCR with primer pairs fabG2F1-fabG2R1, fabG3F1-fabG3R1, fabG4F1-fabG4R1, fabG5F1-fabG5R1, and fabG6F1-fabG6R1, respectively, and cloned into pL52, resulting in pWLG2 to pWLG6, respectively. Next, the phaEC genes were cloned into pWLG2 to pWLG6, generating pWLG2EC to pWLG6EC, respectively, as described for pWLG1EC. In addition, the same DNA fragment of phaEC was inserted into pWL102 to construct plasmid pWLEC, which was used as a control plasmid in testing of the functions of the fabG genes. All of these plasmids were confirmed by DNA sequencing, and the H. volcanii cells harboring these plasmids were subjected to analysis of PHA accumulation.
The PHA contents and compositions in dry cells were determined by gas chromatography (GC) with an Agilent GC-6820 instrument as described previously (6, 19). Benzoic acid was used as an internal standard to calculate the amount of PHA.
For enzyme activity assay, H. hispanica cells cultivated in AS-168 medium for 72 h were harvested by centrifugation and then suspended with 3.4 M KCl in 20 mM Tris-HCl (pH 7.5). Crude extracts were prepared by ultrasonic treatment of these cells, and the intact cells and debris were pelleted by centrifugation (15 min, 8,000 × g, 4°C). The concentration of cellular proteins was determined with a bicinchoninic acid protein assay kit (Pierce).
The activity of acetoacetyl-CoA reductase was measured spectrophotometrically at 340 nm by recording the oxidation of NADPH or NADH as described by Senior and Dawes (34), with some minor modifications. Briefly, assays were carried out at 25°C in a final volume of 200 μl containing 20 mM Tris-HCl (pH 7.5), 3.4 M KCl, 20 μM acetoacetyl-CoA (Sigma), 100 μM NADPH or NADH (Roche), and 50 μg of protein extracts. Negative controls in which acetoacetyl-CoA was omitted from the reaction mixture were performed to subtract the activities of any possible NAD(P)H oxidases in the protein extracts. The reaction was initiated by the addition of crude extract samples at 25°C. A decrease in the absorbance at 340 nm was recorded for 10 min. The initial rate was used to calculate the enzymatic activity. The oxidation of 1 μmol NADPH/NADH per minute corresponded to an enzyme activity of 1 U.
Both H. marismortui and H. hispanica are capable of synthesizing SCL-PHAs, and the PHA synthase genes in these two haloarchaeal strains were characterized recently (6). However, the PHA-specific acetoacetyl-CoA reductase (PhaB), which provides the (R)-3-HB-CoA precursor for biosynthesis of either PHB or PHBV in bacteria, has not yet been identified in these haloarchaea. Interestingly, multiple FabG paralogs are present in H. marismortui ATCC 43049 (1), and at least six homologous proteins, FabG1 to FabG6, named like their counterparts in H. marismortui, are also observed in the H. hispanica genome, with identities ranging from 95.0% to 97.6% for each pair of counterparts.
Like the case for PhaB (previously named FabG) in Synechocystis sp. strain PCC6803 (36), the catalytically active triad (Ser, Tyr, and Lys) and the N-terminal cofactor binding sequence (Gly motif [GlyXXXGlyXGly]) of SDRs (23) are present in all of the FabG paralogs from H. hispanica (Fig. (Fig.1A;1A; see Fig. S1 in the supplemental material). Moreover, the proposed signature sequence of the Gly motif of PHA-specific acetoacetyl-CoA reductases (ValThrGlyXXXGlyIleGly) (36) is completely conserved in FabG1, FabG2, and FabG5 (Fig. (Fig.1A1A).
To understand the phylogenetic relationships between these haloarchaeal FabG paralogs and well-known bacterial FabGs or PhaBs, a phylogenetic tree was constructed (Fig. (Fig.1B).1B). To our surprise, while the confirmed bacterial PhaBs (GenBank accession no. NP_441203.1, P45375.1, and YP_725942.1) (15, 28, 36), FabGs (NP_215999 and AAK04872.1) (24, 38), and those SDRs with both PhaB and FabG activities (AAA23739.1 and BAE19680.1) (21, 22, 35) were clustered in several specific subgroups, the haloarchaeal FabG1 to FabG6 proteins were highly divergent from their bacterial counterparts (Fig. (Fig.1B).1B). The levels of amino acid identity between PhaBs and the six haloarchaeal FabG paralogs ranged from 25% to 38%, with FabG1 exhibiting the highest level of homology. However, even FabG1 was only distantly related to the bacterial PhaBs (Fig. (Fig.1B),1B), and hence the involvement of FabG1 and other FabG paralogs in supplying (R)-3-HB-CoA precursors for PHA biosynthesis in Haloarcula species remained to be clarified experimentally.
Prior to exploring the possible involvements of the FabG paralogs during PHA synthesis in H. hispanica, RT-PCR using gene-specific primers was first performed to check whether they were transcribed under PHA-accumulating conditions (Table (Table2).2). The results demonstrated that all six genes were actively transcribed, but the transcription levels were relatively higher for the fabG1, fabG3, and fabG4 genes (Fig. (Fig.2).2). Hence, all of the FabG paralogs might play roles in metabolism of this archaeon.
In order to identify which FabG paralog might be involved in PHA synthesis, the fabG genes were then disrupted by a single-crossover recombination method (see Materials and Methods) (Fig. (Fig.3A).3A). The successful gene disruptions for fabG1 to fabG6 were confirmed by PCR analysis (data not shown). The resultant six fabG-interrupted mutants, named H. hispanica fabG1-1 to fabG6-1, were subjected to the following phenotype analyses.
Interestingly, all of the fabG mutants showed no significant difference in growth compared with the wild-type strain, as evaluated by their cell dry weight (CDW) (Table (Table3).3). Moreover, disruption of the fabG2 to fabG6 genes had no evident effects on PHA accumulation, including both the PHA content and monomer ratios (Table (Table3).3). Significantly, though, GC analysis revealed that PHA synthesis was completely abolished in H. hispanica fabG1-1 cells. These results inferred that FabG1, but not the other FabG paralogs, might be the PHA precursor supplying enzyme and might play an important role in PHA biosynthesis.
Analysis of the gene location of fabG1 in the chromosome of H. hispanica revealed two downstream genes encoding an acetyl-CoA carboxylase α subunit (AccA) and a conserved hypothetical protein. There was only a 1-base spacing between any two adjacent genes, suggesting that these three genes might be cotranscribed. Thus, the large insertion in the fabG1 gene in the fabG1-disrupted mutant might generate polar effects on downstream genes, which might also influence the metabolism of PHA synthesis. Therefore, gene knockout and complementation analysis was carried out to further clarify the function of FabG1 during PHA biosynthesis in H. hispanica.
With a double-crossover homologous recombination strategy (Fig. (Fig.3B),3B), the fabG1 gene was deleted in frame, resulting in a mutant strain named H. hispanica ΔfabG1. The genotype of the ΔfabG1 strain was confirmed by PCR analysis (Fig. (Fig.3C).3C). Significantly, the ΔfabG1 cells lost the ability to accumulate PHBV, as demonstrated by GC analysis (Fig. (Fig.4B).4B). In addition, when the ΔfabG1 strain was transformed with the fabG1 expression plasmid pWLG1, the capability of PHA accumulation was restored (Fig. (Fig.4C).4C). These results further demonstrated that the fabG1 gene was indeed indispensable for PHA synthesis and was the PHA precursor-supplying enzyme in H. hispanica.
H. volcanii DS70 is a model haloarchaeon which is easy to use for genetic manipulation but defective in PHA biosynthesis. Therefore, it is a suitable host for verification of PHA biosynthesis genes from other haloarchaea. BLAT analysis revealed that there was neither a PhaB nor PHA synthase homolog in the genome of H. volcanii (The UCSC Archaeal Genome Browser [http://archaea.ucsc.edu/]). Therefore, to verify the function of FabGs in H. volcanii, it is necessary to heterologously express the fabG gene with PHA synthase genes (phaEC). These kinds of expression plasmids, pWLG1EC to pWLG6EC (Table (Table1),1), were introduced into the H. volcanii DS70 cells, with pWLEC harboring phaEC genes as the control. Table Table33 shows the PHA accumulation in H. volcanii DS70 and its transformants. Notably, neither the wild-type strain nor the recombinant harboring plasmid pWLEC accumulated detectable PHA, confirming that H. volcanii DS70 was deficient not only in polymerization enzymes but also in PHA precursor-supplying enzymes. The seven recombinant strains harboring the heterologous fabG plus phaEC genes showed no significant difference in cell growth compared with the wild-type strain of H. volcanii (Table (Table3).3). Significantly, only the coexpression of fabG1 and phaEC in DS70 resulted in detectable PHA accumulation, with ~3% (wt/wt) CDW (Table (Table3).3). In contrast, the other five FabGs (FabG2 to FabG6) likely had no roles in supplying PHA precursor, as no PHA accumulation was detected after their expression in H. volcanii. These results demonstrated that FabG1 from H. hispanica, but not the other five FabG paralogs, could supply precursors for SCL-PHA synthesis in both Haloarcula and Haloferax species.
To biochemically confirm the function of FabG1, an enzyme activity assay was established to investigate whether it could convert acetoacetyl-CoA into (R)-3-HB-CoA, the main precursor for SCL-PHAs. Since the fabG genes were constitutively expressed under both nutrient-rich (AS-168 medium) and nutrient-limited (MG medium) conditions (data not shown), the cell crude extracts of H. hispanica strains cultivated in AS-168 medium were subjected to the acetoacetyl-CoA reductase activity assay, supplying NADPH or NADH as the cofactor (Table (Table4).4). Significantly, when the fabG1 gene was disrupted or knocked out, the NADPH-dependent activity of acetoacetyl-CoA reductase in the mutant, the ΔfabG1 or fabG1-1 strain, was greatly decreased, from ~20 to ~0.5 U/g crude extract proteins (Table (Table4).4). Correspondingly, when the fabG1 gene was reexpressed in the ΔfabG1 strain, the NADPH-dependent acetoacetyl-CoA reductase activity (25.6 U/g) (Table (Table4)4) was restored. Under both conditions, however, the NADH-dependent acetoacetyl-CoA reductase activities were not significantly affected. These results indicate that FabG1 is indeed a NADPH-dependent acetoacetyl-CoA reductase.
To our surprise, when the other five fabG paralogs (fabG2 to fabG6) were disrupted, neither the NADPH- nor NADH-dependent acetoacetyl-CoA reductase activity was significantly changed (Table (Table4).4). It is possible that these FabG paralogs prefer their natural substrates (i.e., 3-ketoacyl-ACP); hence, little acetoacetyl-CoA reductase activity would be present. It is also possible that these FabG paralogs evolved to serve other, unknown functions, which remain to be investigated in the future. In any case, our results incontrovertibly demonstrate that FabG1 is distinct from its five paralogs and evolved to be a PHA-specific acetoacetyl-CoA reductase, using NADPH as a cofactor, in the Haloarcula genus.
FabG and PhaB are members of a vast protein family, the SDR superfamily, which catalyze a wide variety of NAD(P)(H)-dependent oxidation/reduction reactions (23). Annotation of fabG- or phaB-homologous genes is particularly problematic (2, 17, 36). In the present study, we combined bioinformatics, classic genetic techniques, and biochemical methods to elucidate the functions of several fabG-paralogous genes in PHA biosynthesis in H. hispanica. We demonstrated that the fabG1 gene actually encodes a NADPH-dependent acetoacetyl-CoA reductase (PhaB) which is indispensable for converting acetoacetyl-CoA into (R)-3-HB-CoA for PHBV biosynthesis in this haloarchaeon. To our knowledge, this is the first PHA precursor-supplying gene experimentally identified so far in the domain of Archaea.
There are multiple FabG paralogs in the sequenced genomes of several haloarchaeal species, including H. marismortui (1), H. hispanica (our unpublished genome data), and H. volcanii (The UCSC Archaeal Genome Browser [http://archaea.ucsc.edu/]). Our results indicate that FabG1 in H. hispanica, and possibly its counterpart (with 97% identity) in H. marismortui, has evolved certain distinct functions in PHA biosynthesis. Biochemical assay revealed that only FabG1 among the six paralogs is a NADPH-dependent acetoacetyl-CoA reductase, and genetic and metabolic evidence supports the observation that only FabG1 could convert acetoacetyl-CoA into (R)-3-HB-CoA and provide this precursor for PHA synthesis. While the exact functions of FabG2 to FabG6 remain unclear, the NADH-dependent reductase activity (<3 U/g crude extract proteins) detected in all of the fabG mutants infers that there might be other NADH-dependent acetoacetyl-CoA reductases. These NADH-dependent enzymes, if any, might just convert acetoacetyl-CoA into (S)-3-HB-CoA (7), which is unable to be incorporated into PHBV by PHA synthases. Interestingly, BLAT analysis of the genome of H. volcanii DS2 revealed that there were no homologous genes encoding either PhaEC or FabG1, while those for the other five FabG paralogs were detected. This observation was consistent with our heterologous expression results showing that only the introduction of fabG1 and phaEC could confer on the PHA-defective strain the ability to synthesize PHA. The low content of PHA in the recombinant H. volcanii strain was likely due to its poor capability of carbohydrate utilization (5).
Genome-wide analysis revealed that there are several β-ketoacyl thiolase genes in H. marismortui as well as in H. hispanica. In the present study, we showed that the FabG1 protein was responsible for providing 3-HB-CoA, and coexpression of the fabG1 and phaEC genes was able to reestablish the PHA synthesis pathway in H. volcanii. These results suggested that β-ketoacyl thiolases do indeed exist in both H. hispanica and H. volcanii. In fact, we detected the activity of β-ketoacyl thiolase in the crude extracts of both strains (data not shown). At present, we are investigating which β-ketoacyl thiolase catalyzes acetoacetyl-CoA synthesis and hence is involved in PHA synthesis. Thus, the PHA biosynthesis pathway from acetyl-CoA, catalyzed by β-ketoacyl thiolase, acetoacetyl-CoA reductase, and PHA synthase, as distributed in Bacteria, likely also exists in the domain of Archaea. However, for PHB-accumulating haloarchaeal Natrialba strain 56, no enzyme activity of acetoacetyl-CoA reductase or β-ketoacyl thiolase was detected in the crude extract (8, 9), indicating that a different metabolic route toward PHB biosynthesis might be employed.
In bacteria, the PHA-specific acetoacetyl-CoA reductase, named PhaB, has two other conserved residues (Val and Ile) in the cofactor binding sequence (ValThrGlyXXXGlyIleGly). Similarly, FabG1 was also found to share these conserved amino acids. Most bacterial PhaB proteins use NADPH as the cofactor (7, 27), except that PhaB from Allochromatium vinosum (originally named Chromatium vinosum) has been reported to be NADH dependent (15). Likewise, the FabG1 protein was experimentally proven to be NADPH dependent and functioned as an (R)-3-HB-CoA-supplying enzyme. The cofactor of FabG1 could also be predicted by the method developed by Kallberg et al. (11). They proposed that if there is no acidic residue (Asp) at the end of the second β-strand and a basic residue (Arg or Lys) is located in the Gly motif or at the first loop position after the second β-strand, like in FabG1, NADPH will be used as the cofactor (Fig. (Fig.1A).1A). Therefore, based on their conserved motif, function, and enzyme activity, we recommend renaming the FabG1 proteins of both H. hispanica and H. marismortui as PhaB proteins, the PHA-specific acetoacetyl-CoA reductases of halophilic archaea.
This work was supported by grants from the National Natural Science Foundation of China (30621005 and 30830004), the National 863 Program of China (2006AA09Z401), and the Chinese Academy of Sciences (KSCX2-YW-G-023).
Published ahead of print on 31 July 2009.
†Supplemental material for this article may be found at http://aem.asm.org/.