Haloarchaeal-Type β-Ketothiolases Involved in Poly(3-Hydroxybutyrate-co-3-Hydroxyvalerate) Synthesis in Haloferax mediterranei

Abstract

The key enzymes for poly(3-hydroxybutyrate-co-3-hydroxyvalerate) (PHBV) biosynthesis in haloarchaea have been identified except the β-ketothiolase(s), which condense two acetyl coenzyme A (acetyl-CoA) molecules to acetoacetyl-CoA, or one acetyl-CoA and one propionyl-CoA to 3-ketovaleryl-CoA. Whole-genome analysis has revealed eight potential β-ketothiolase genes in the haloarchaeon Haloferax mediterranei, among which the PHBV-specific BktB and PhaA were identified by gene knockout and complementation analysis. Unlike all known bacterial counterparts encoded by a single gene, the haloarchaeal PhaA that was involved in acetoacetyl-CoA generation, was composed of two different types of subunits (PhaAα and PhaAβ) and encoded by the cotranscribed HFX_1023 (phaAα) and HFX_1022 (phaAβ) genes. Similarly, the BktB that was involved in generation of acetoacetyl-CoA and 3-ketovaleryl-CoA, was also composed of two different types of subunits (BktBα and BktBβ) and encoded by cotranscribed HFX_6004 (bktBα) and HFX_6003 (bktBβ). BktBα and PhaAα were the catalytic subunits and determined substrate specificities of BktB and PhaA, respectively. Their catalytic triad “Ser-His-His” was distinct from the bacterial “Cys-His-Cys.” BktBβ and PhaAβ both contained an oligosaccharide-binding fold domain, which was essential for the β-ketothiolase activity. Interestingly, BktBβ and PhaAβ were functionally interchangeable, although PhaAβ preferred functioning with PhaAα. In addition, BktB showed biotechnological potential for the production of PHBV with the desired 3-hydroxyvalerate fraction in haloarchaea. This is the first report of the haloarchaeal type of PHBV-specific β-ketothiolases, which are distinct from their bacterial counterparts in both subunit composition and catalytic residues.

INTRODUCTION

Polyhydroxyalkanoates (PHAs) are accumulated intracellularly by many bacteria and some haloarchaea under unbalanced growth conditions as carbon, energy, and reducing power storage (1, 2). Since these biopolymers are biodegradable and biocompatible thermoplastics and can be produced from renewable resources, PHAs have attracted much attention as environmentally friendly alternatives to conventional petroleum-derived plastics (1). More than 150 different monomer constituents have been identified in biosynthesized PHAs (3), and the monomer composition is a key factor affecting the material properties of these polyesters. Poly(3-hydroxybutyrate) (PHB) and poly(3-hydroxybutyrate-co-3-hydroxyvalerate) (PHBV) are the two most extensively researched PHAs. Due to the presence of the 3-hydroxyvalerate (3HV) monomer units, PHBV exhibits improved material properties for biomedical and industrial applications over PHB (4, 5).

Since the first discovery of PHB granules in Bacillus megaterium in 1926 (6), an increasing number of bacterial genera have been found to accumulate PHAs. In most bacteria, PHB is produced from unrelated carbon sources via a three-step process (7). In the first step, β-ketothiolase catalyzes a biological Claisen condensation of two acetyl coenzyme A (acetyl-CoA) molecules into acetoacetyl-CoA. Acetoacetyl-CoA is then reduced to (R)-3-hydroxybutyl-CoA by a stereospecific β-ketoacyl-CoA reductase. Finally, PHB synthase polymerizes the 3-hydroxybutyrate (3HB) moiety of (R)-3-hydroxybutyl-CoA into PHB. It is noteworthy that the β-ketothiolases that catalyze the first step of PHA biosynthesis are biosynthetic β-ketothiolases. In general, β-ketothiolases are ubiquitous in all of the organisms and are classified as either biosynthetic or degradative β-ketothiolases (8). Biosynthetic β-ketothiolases catalyze condensation reactions and are involved in various biosynthetic pathways, such as the biosynthesis of PHAs or steroids (9). Degradative β-ketothiolases work preferably in the thiolysis direction and are involved in degradative pathways such as fatty acid β-oxidation (9). The PHA biosynthetic β-ketothiolases have been identified and studied extensively in a wide range of bacteria. All of the bacterial PHA biosynthetic β-ketothiolases reported thus far function as homotetramers (9–13). The crystal structure for the PHA biosynthetic β-ketothiolase from Zoogloea ramigera has been determined, which reveals a two-step “ping pong” reaction mechanism involving the catalytic residues Cys-His-Cys (14). The “ping pong” mechanism is commonly used by bacterial biosynthetic β-ketothiolases (9, 10, 13).

Haloarchaea represent another important group of PHA-accumulating organisms. However, PHA biosynthetic β-ketothiolase has not yet been reported in any haloarchaeon. Haloferax mediterranei is the most commonly used model organism for studies of haloarchaeal PHBV biosynthesis. It can produce a large amount of PHBV with ca. 10 mol% 3HV from unrelated cheap carbon sources (15, 16). Thus far, the key enzymes of β-ketoacyl-CoA reductase (PhaB1/PhaB2) (17) and PHA synthase (PhaE/PhaC) (15), the PHA granule-associated structural protein PhaP (18), and the four propionyl-CoA generation pathways (19) have been discovered for PHBV synthesis in H. mediterranei. Although bioinformatic analyses of the H. mediterranei genome sequence (20) revealed eight potential β-ketothiolase genes, the gene(s) encoding the PHA biosynthetic β-ketothiolase(s) that are responsible for providing acetoacetyl-CoA and 3-ketovaleryl-CoA remained unknown.

In the present study, we discovered two β-ketothiolases, PhaA and BktB, which were responsible for supplying the precursors for PHBV biosynthesis in the haloarchaeon H. mediterranei. PhaA was limited to the condensation of two acetyl-CoA molecules to generate acetoacetyl-CoA, while BktB produced both acetoacetyl-CoA and 3-ketovaleryl-CoA. Interestingly, PhaA and BktB were both composed of two different types of subunits and encoded by two cotranscribed genes, while their bacterial counterparts reported thus far are all encoded by a single gene (9–13). In addition, both PhaA and BktB harbored catalytic residues distinct from those found in bacterial biosynthetic β-ketothiolases. Therefore, haloarchaeal PHA biosynthetic β-ketothiolases might comprise a novel group of β-ketothiolases. Notably, BktB showed biotechnological potential in promoting the production of PHBV with controllable 3HV content from unrelated cheap carbon sources in haloarchaea.

MATERIALS AND METHODS

Strains and culture conditions.

All of the strains used in the present study are listed in Table S1 in the supplemental material. Escherichia coli JM109 was grown in lysogeny broth (LB) (21) at 37°C. When necessary, ampicillin was added to the media at a final concentration of 100 μg/ml. For PHBV accumulation analysis, H. mediterranei and Haloarcula hispanica strains were first cultured in nutrient-rich AS-168 medium (22) at 37°C for 36 h as seed culture. A 2.5-ml aliquot of the seed culture was then transferred into 50 ml of MG medium (per liter, 200 g of NaCl, 20 g of MgSO₄·7H₂O, 2 g of KCl, 37.5 mg of KH₂PO₄, 5 mg of FeSO₄·7H₂O, 0.036 mg of MnCl₂·4H₂O, 10 g of glucose, 1 g of yeast extract, 15 g of PIPES; pH 7.2) for PHBV accumulation in shaking flasks until the culture reached stationary phase. For the strains carrying the plasmid pWL502 or its derivatives, 5 g of yeast extract/liter was omitted from the AS-168 medium, and 2 g/liter NH₄Cl, instead of 1 g/liter yeast extract, was used in the MG medium. For the disruptants carrying the integration plasmid pUBP, mevinolin was added to the media at a final concentration of 3 μg/ml (15). For the H. mediterranei DF50 and DF50-based knockout mutants, 50 μg of uracil/ml was supplied in the media (23).

Mutant construction.

The primers and plasmids used for mutant construction and verification are listed in Tables S2 and S1, respectively, in the supplemental material. Construction of the recombinant strains was performed as previously described (22, 24). For disruption of the eight potential β-ketothiolase genes, a fragment (∼500 bp) located internally within the target gene was amplified, sequenced, and inserted into the pUBP plasmid (22). The resulting plasmid was transformed into H. mediterranei to disrupt the target gene through single-crossover homologous recombination. For construction of the knockout mutants, two fragments (∼650 bp each) located immediately upstream and downstream of the target gene were amplified, sequenced, and inserted into the pHFX plasmid (23). The pHFX-derived plasmid was then introduced into the H. mediterranei DF50 strain to delete the target gene through homologous recombination (23). Gene complementation was performed by transforming the corresponding expression plasmid derived from pWL502 into the host strain (18). The polyethylene glycol-mediated transformation method was used for transformation of haloarchaeal strains (25).

PHBV analysis.

The PHBV content in the cells and the monomer composition of PHBV were detected by gas chromatography (GC) analysis as previously described (22). Benzoic acid was used as the internal standard for the quantitative analysis.

Site-directed mutagenesis.

To investigate the catalytic residues of BktB and PhaA, site-directed mutagenesis was performed as previously described (26). First, the gene sequences of bktB and phaA were cloned into the commercially available cloning vector pUCm-T (Sangon, China) separately to yield the plasmids T-bktB and T-phaA, respectively. The T-bktB or T-phaA plasmid was then used as PCR template to introduce the designed point mutations using the mutagenic primers listed in Table S2 in the supplemental material. After the PCR, the PCR product was digested with DpnI to remove the template plasmid and then introduced into E. coli JM109. After sequencing, T-bktB- or T-phaA-derived plasmid with the desired point mutation was obtained, and the corresponding bktB or phaA fragment was subcloned into pWL502. The resulting plasmids were then transformed into H. mediterranei via the polyethylene glycol-mediated transformation method (25).

RT-PCR.

H. mediterranei cells were cultivated in MG medium (as described under “Strains and culture conditions” above) until mid-exponential phase and collected for RNA preparation. Total RNA was prepared using the TRIzol reagent (Invitrogen, USA). Reverse transcription-PCR (RT-PCR) was performed as previously described (22), except that the primers used for the RT reaction were a mixture of random hexanucleotides (Thermo Scientific); the primers for PCR are listed in Table S2 in the supplemental material.

Protein sequence analysis and phylogenetic tree construction.

Domain annotation was carried out by using Pfam family search (http://pfam.sanger.ac.uk/), the InterProScan program (http://www.ebi.ac.uk/Tools/pfa/iprscan/), and the NCBI conserved domain search (CDD) service (http://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi). Sequence homology was analyzed by using the BLAST service (http://blast.ncbi.nlm.nih.gov/Blast.cgi) and the GeneDoc Program (http://www.nrbsc.org/gfx/genedoc/). The phylogenetic tree was constructed using the neighbor-joining method with MEGA4 software based on amino acid sequence analysis (27). The topology of the phylogenetic tree was evaluated by performing a bootstrap analysis with 1,000 replicates.

RESULTS

Bioinformatic analysis of the β-ketothiolase homologs in H. mediterranei.

Bioinformatic analysis of the H. mediterranei genome sequence revealed eight potential β-ketothiolase genes (Table 1), whose encoding products all belonged to the thiolase family (InterPro accession number IPR002155). Five of the eight potential genes might share a promoter with the adjacent genes that lie immediately downstream, since these genes either overlapped or had intergenic regions that were too small (≤4 bp) to harbor a promoter. Interestingly, four of the five adjacent genes encoded small proteins with unknown function (Table 1 and see Fig. S1 and Table S3 in the supplemental material). Protein sequence analysis revealed that these small proteins all contained a putative OB-fold (oligonucleotide/oligosaccharide-binding fold; InterPro accession number IPR002878) domain. The OB-fold, a closed β-barrel with five-stranded β-sheet, is prevalent in all of the organisms and has been reported to be able to bind various ligands, including oligonucleotides, proteins, oligosaccharides, metal ions, and catalytic substrates (28). These small OB-fold-containing proteins may therefore be functionally related to β-ketothiolase activity. This would be different from all known bacterial PHA biosynthetic β-ketothiolases, which are all composed of identical subunits (9–13). In addition, phylogenetic analysis of the eight β-ketothiolase homologs from H. mediterranei and the bacterial PHA biosynthetic β-ketothiolases revealed that five of the eight β-ketothiolase homologs from H. mediterranei clustered together (group I), while the other three seemed to be more closely related to the bacterial PHA biosynthetic β-ketothiolases (group II) (Fig. 1). This phylogenetic classification was coincidentally consistent with the gene organization; that is, each of the genes encoding the group I β-ketothiolase homologs might form an operon with the adjacent gene, whereas the genes for group II appeared to be transcribed individually (Fig. 1 and Table 1).

Table 1.

Overview of the eight potential β-ketothiolase genes in H. mediterranei

Locus tag	Identity (%)a	Adjacent gene (description)b	Intergenic regionc
HFX_6004	100	HFX_6003 (protein with unknown function)	4-bp overlap
HFX_1023	74	HFX_1022 (protein with unknown function)	1-bp overlap
HFX_6015	38	HFX_6014 (superoxide dismutase)	138 bp
		HFX_6016 (protein with unknown function)	2 bp
HFX_0804	37	HFX_0803 (protein with unknown function)	1-bp overlap
HFX_6358	37	HFX_6357 (HMG-CoA synthase)d	4-bp overlap
HFX_2006	26	HFX_2005 (nitrite reductase)	285 bp
HFX_6356	24	—e
HFX_6051	19	HFX_6050 (anion transport ATPase)	373 bp

^aThe identities are shown with respect to HFX_6004 at the amino acid level.

^bThe adjacent genes listed here only include those oriented in the same direction as the β-ketothiolase genes. The annotation of the encoded protein is presented in parentheses. More details are given in Fig. S1 and Table S3 in the supplemental material.

^cThe intergenic region between the adjacent gene and the corresponding β-ketothiolase gene.

^dHMG-CoA synthase, hydroxymethylglutaryl-coenzyme A synthase.

^e—, the two adjacent genes of HFX_6356 are both divergently oriented to HFX_6356.

Fig 1.

Phylogenetic tree of the eight β-ketothiolase homologs from H. mediterranei and the bacterial PHA biosynthetic β-ketothiolases. The β-ketothiolase homologs from H. mediterranei were classified into two groups (group I and group II). Numbers at the nodes indicate the bootstrap percentage values based on 1,000 replicates. Only values >70% were considered significant and are shown here. The scale bar represents a difference of 0.2 substitution per site. The NCBI protein database accession numbers for the bacterial PHA biosynthetic β-ketothiolases are shown in parentheses.

Identification of the PHBV biosynthetic β-ketothiolases in H. mediterranei.

To identify the β-ketothiolase(s) involved in the biosynthesis of PHBV in H. mediterranei, we disrupted these eight putative β-ketothiolase genes individually. GC analysis revealed that disruption of one of the genes, HFX_6004, led to the accumulation of PHB only without the 3HV monomer, whereas disruption of the remaining seven genes had no significant effect on PHBV accumulation (data not shown). Therefore, we constructed an in-frame markerless deletion mutant of HFX_6004-6003. Consistent with the HFX_6004 disruption mutant, the knockout mutant of HFX_6004-6003 lost the ability to produce the 3HV monomer but had no significant effect on the supply of the 3HB monomer (Fig. 2A and B). These data indicated that the encoding product of either HFX_6004 or HFX_6004-6003 was at least partially responsible for the synthesis of 3-ketovaleryl-CoA, the precursor for the 3HV monomer. However, the β-ketothiolase(s) responsible for providing acetoacetyl-CoA for the 3HB monomer had not been determined yet. Therefore, we compared the protein sequence of HFX_6004 against the other seven β-ketothiolase protein sequences. HFX_1023 showed the highest level of identity, 74%, to HFX_6004 (Table 1). Hence, we constructed the knockout mutant of HFX_1023-1022. This knockout mutant, like the HFX_1023 disruption mutant, showed no significant effect on PHBV accumulation (Fig. 2A and C). Subsequently, an HFX_6004-6003 and HFX_1023-1022 double deletion mutant was constructed. Remarkably, the simultaneous loss of HFX_6004-6003 and HFX_1023-1022 caused the strain to lose the ability to accumulate PHA (Fig. 2D). Based on these results, we concluded that two β-ketothiolases were involved in PHBV biosynthesis in H. mediterranei. The encoding product of either HFX_6004 or HFX_6004-6003 was able to catalyze the synthesis of acetoacetyl-CoA from two acetyl-CoA molecules and also produce 3-ketovaleryl-CoA from acetyl-CoA and propionyl-CoA, while the encoding product of either HFX_1023 or HFX_1023-1022 was restricted to the biosynthesis of acetoacetyl-CoA. This is the first report of PHBV biosynthetic β-ketothiolases in H. mediterranei. However, further experiments were needed to determine whether the active β-ketothiolase was encoded by either one (HFX_6004 or HFX_1023) or two (HFX_6004-6003 or HFX_1023-1022) genes.

Fig 2.

GC analysis of PHA accumulation in H. mediterranei strains. (A) H. mediterranei DF50 strain; (B) H. mediterranei DF50 ΔHFX_6004-6003 strain; (C) H. mediterranei DF50 ΔHFX_1023-1022 strain; (D) H. mediterranei DF50 ΔHFX_6004-6003 HFX_1023-1022 strain. The peaks at 4.85 min represent the methyl ester product of benzoic acid, which serves as an internal standard. The data shown are representative of three independent experiments.

PHBV biosynthetic β-ketothiolases are composed of two different types of subunits in H. mediterranei.

Analysis of the gene organization revealed that HFX_6004 and HFX_6003 overlapped by 4 bp and that HFX_1023 and HFX_1022 overlapped by 1 bp (Fig. 3A and Table 1), which indicated that the neighboring genes might be cotranscribed. Therefore, RT-PCR was used to examine the transcription profiles of HFX_6004-6003 and HFX_1023-1022. The results showed that HFX_6004 and HFX_6003 were indeed cotranscribed in the mid-exponential-phase cells grown in the MG medium (see Materials and Methods), as were HFX_1023 and HFX_1022 (Fig. 3B), indicating that the neighboring genes were at least functionally related and might both be important for β-ketothiolase activity.

Fig 3.

RT-PCR analysis of the transcription of the HFX_6004-6003 and HFX_1023-1022 gene clusters. (A) Genetic organization of the HFX_6004-6003 or HFX_1023-1022 gene cluster. The arrows indicate the position of the primers (RT-F1/RT-R1 and RT-F2/RT-R2) used for RT-PCR. (B) RT-PCR determination of the cotranscription pattern of the HFX_6004-6003 (6004-6003) or HFX_1023-1022 (1023-1022) gene cluster. For cDNA lanes, the cDNA from the RNA of H. mediterranei is the template; for RNA lanes, the RNA extracted from H. mediterranei is the template; for “+” lanes, the genome is the template (positive control); “–” lanes represent the negative control; “M.” lanes contain the marker. The primers used here are listed in Table S2 in the supplemental material.

To confirm the indispensability of the cotranscribed genes for the activity of β-ketothiolase, a series of genetic complementation experiments using the double deletion mutant of HFX_6004-6003 and HFX_1023-1022 (HFXΔ2 strain, abbreviation of the double mutant) were performed. These experiments revealed that expression of only HFX_6004-6003 or HFX_1023-1022 in the HFXΔ2 strain was sufficient to accumulate PHBV or PHB; however, the HFXΔ2 strain harboring a single gene of HFX_6004, HFX_6003, HFX_1023, or HFX_1022 was deficient in PHA accumulation (Table 2). Thus, we concluded that the two cotranscribed genes, either HFX_6004 and HFX_6003 or HFX_1023 and HFX_1022, encoded the active PHA biosynthetic β-ketothiolase, which was designated as BktB or PhaA, respectively, according to their substrate specificity. In contrast, the bacterial PHA biosynthetic β-ketothiolases reported thus far are all encoded by a single gene (9–13).

Table 2.

PHA production by H. mediterranei strains^a

Strain	Mean ± SD
	CDWb (g/liter)	PHA content (wt%)	3HV fraction (mol%)	PHA concn (g/liter)
HFXΔ2(pWL502)	1.5 ± 0.20	ND	ND	ND
HFXΔ2(p6004)	0.9 ± 0.01	ND	ND	ND
HFXΔ2(p6003)	1.8 ± 0.05	ND	ND	ND
HFXΔ2(p1023)	1.2 ± 0.19	ND	ND	ND
HFXΔ2(p1022)	1.3 ± 0.34	ND	ND	ND
HFXΔ2(p6004-6003)	3.1 ± 0.11	32 ± 1.3	8.4 ± 0.54	1.0 ± 0.05
HFXΔ2(p1023-1022)	4.0 ± 0.64	32 ± 1.6	ND	1.3 ± 0.15
HFXΔ2(pbktBα-phaAβ)	3.4 ± 0.03	38 ± 1.3	8.7 ± 0.20	1.3 ± 0.03
HFXΔ2(pphaAα-bktBβ)	3.4 ± 0.34	38 ± 2.5	ND	1.3 ± 0.05
ΔbktBβ(pWL502)	6.8 ± 0.23	20 ± 0.5	ND	1.4 ± 0.07
ΔbktBβ(pphaAβ)	4.8 ± 0.55	24 ± 1.5	8.0 ± 0.33	1.2 ± 0.06

^aPHA accumulation was detected when cells reached the stationary phase. All the data are shown as means of three independent experiments. ND, not detectable.

^bCDW, cell dry weight.

Determination of the catalytic residues of BktB and PhaA.

As demonstrated above, BktB is composed of two different types of subunits that are encoded by bktBα (HFX_6004) and bktBβ (HFX_6003), and PhaA is also composed of two different types of subunits that are encoded by phaAα (HFX_1023) and phaAβ (HFX_1022). BktBα and PhaAα both contain a thiolase domain and are thus thought to be the catalytic subunits of BktB and PhaA, respectively. A catalytic triad composed of “Cys-His-Cys” has been verified in bacterial PHA biosynthetic β-ketothiolases (14). To identify the catalytic residues of BktBα and PhaAα, a multiple alignment of the sequences of BktBα and PhaAα and those of their bacterial counterparts was performed. The multiple sequence alignment clearly revealed some conserved amino acid residues between the haloarchaeal and bacterial PHA biosynthetic β-ketothiolases in spite of the low overall sequence identity (Fig. 4). The three amino acid residues Ser83-His331-Cys365 of BktBα and Ser84-His332-Cys366 of PhaAα corresponded to the bacterial catalytic triad Cys-His-Cys (Fig. 4). The catalytic residues of BktBα and PhaAα were also predicted using the NCBI CDD service. Interestingly, the NCBI CDD service predicted the catalytic triad to be Ser83-His281-His331 for BktBα and Ser84-His282-His332 for PhaAα, respectively, differing from the results obtained from the sequence comparison (Fig. 4). Therefore, the potential catalytic residues obtained using both the multiple sequence alignment and the NCBI CDD service were all investigated by site-directed mutagenesis. To ensure that the respective mutant proteins were expressed at a comparable level to the wild-type proteins, the genes of wild-type and site-directed mutant were, respectively, cloned into the same vector pWL502 and transferred into the same host H. mediterranei HFXΔ2. After being cultivated to stationary phase, the maximum PHA production of these strains was detected to reflect the relative activity of the BktB, PhaA, or their site-directed mutants. The HFXΔ2 strain expressing BktB with an S83A, H281Q, or H331Q mutation in BktBα did not accumulate PHA at all, whereas strain expressing BktB with a C365A or C365S mutation in BktBα had no significant effect on PHBV accumulation (Table 3). These data indicated that the S83A, H281Q, or H331Q mutation in BktBα abolished the activity of BktB in vivo, while the C365A or C365S mutation did not significantly affect the BktB activity. Similarly, the introduction of the S84A, H282Q, or H332Q mutation into PhaAα resulted in inactive PhaA, while the C366A or C366S mutation did not obviously affect PhaA activity (Table 3). These results demonstrated that the amino acid residues S83-H281-H331 of BktBα and S84-H282-H332 of PhaAα were the catalytic residues, while C365 of BktBα and C366 of PhaAα were not involved in the catalytic reaction (Fig. 4). It is noteworthy that the HFXΔ2 strain that expressed BktB carrying the S83C mutation in BktBα displayed no significant change in PHBV accumulation; however, expression of PhaA with the S84C mutation in PhaAα reduced the final PHB concentration by ca. 68% (Table 3). These data indicated that the catalytic sites S83 of BktBα and S84 of PhaAα were differentially stringent. Taken together, these mutagenesis studies demonstrate that the catalytic triad of BktBα and PhaAα is likely to be Ser-His-His, which is distinct from the Cys-His-Cys triad of bacterial biosynthetic β-ketothiolases (14).

Fig 4.

Alignment of the partial sequences of BktB and PhaA and four bacterial PHA biosynthetic β-ketothiolases. The three determined catalytic residues of BktB and PhaA are marked with asterisks above the alignment columns, and the three determined bacterial catalytic residues are labeled with brackets below the alignment columns. The point mutations introduced into BktB and PhaA are also indicated with the arrows to show amino acid substitutions. The numbers to the right of the alignment columns represent the position numbers of the last amino acid residues in the protein sequences. The NCBI protein database accession numbers for the bacterial PHA biosynthetic β-ketothiolases are as follows: Sm (Sinorhizobium meliloti 1021), NP_387368; Zr (Zoogloea ramigera), P07097; Pd (Paracoccus denitrificans), P54810; and Re (Ralstonia eutropha H16), YP_725941.

Table 3.

PHA accumulation in H. mediterranei HFXΔ2 strains harboring either bktB or phaA with point mutations^a

Strain	Mean ± SD
	PHA content (wt%)	PHA concn (g/liter)
HFXΔ2(pbktB)	24 ± 1.6	1.4 ± 0.19
HFXΔ2(pbktB S83A)	ND	ND
HFXΔ2(pbktB S83C)	27 ± 1.7	1.5 ± 0.29
HFXΔ2(pbktB H281Q)	ND	ND
HFXΔ2(pbktB H331Q)	ND	ND
HFXΔ2(pbktB C365A)	34 ± 1.2	1.4 ± 0.05
HFXΔ2(pbktB C365S)	33 ± 0.8	1.6 ± 0.10
HFXΔ2(pphaA)	35 ± 2.3	1.5 ± 0.14
HFXΔ2(pphaA S84A)	ND	ND
HFXΔ2(pphaA S84C)	15 ± 1.8	0.5 ± 0.07
HFXΔ2(pphaA H282Q)	ND	ND
HFXΔ2(pphaA H332Q)	ND	ND
HFXΔ2(pphaA C366A)	33 ± 6.0	1.6 ± 0.21
HFXΔ2(pphaA C366S)	33 ± 3.2	1.6 ± 0.15

^aPHA accumulation was detected when cells reached the stationary phase. All the data are shown as means of three independent experiments. ND, not detectable.

Determination of the key regions of BktBβ and PhaAβ.

BktBα and PhaAα appear to be the catalytic subunits of BktB and PhaA, respectively. However, the function of BktBβ and PhaAβ, which are the other essential subunits of BktB and PhaA, remains unknown. An OB-fold was found in both BktBβ and PhaAβ and spanned amino acid residues 32 to 94 in BktBβ and 35–98 in PhaAβ, respectively (Fig. 5A). The flanking regions of the OB-fold in BktBβ and PhaAβ were designated as the N terminus and the C terminus, respectively (Fig. 5A). To identify the key regions of BktBβ and PhaAβ, we constructed HFXΔ2 recombinant strains carrying either the bktB operon with region-deleted bktBβ or the phaA operon with region-deleted phaAβ. The final PHA accumulation of these strains at stationary phase (maximum PHA production) was compared to that of the HFXΔ2 strain harboring the entire bktB or phaA operon to indicate the relative enzyme activity of BktB or PhaA. The double mutant strain expressing BktB with the N terminus of BktBβ (amino acid residues 5 to 28) deleted displayed no significant change in PHBV production. The strain expressing BktB with the C terminus of BktBβ (amino acid residues 98 to 120) deleted displayed a ca. 70% reduction in PHBV production, and the HFXΔ2 strain expressing BktB with the OB-fold of BktBβ (amino acid residues 35 to 91) deleted did not accumulate PHA at all (Fig. 5B). These results showed that the OB-fold of BktBβ was indispensable for BktB activity, the C terminus of BktBβ was required for full BktB activity, and the N terminus of BktBβ was not essential for BktB activity. The HFXΔ2 strain expressing PhaA with the N terminus of PhaAβ (amino acid residues 5 to 31) deleted displayed a reduction in PHB production of ca. 93% compared to the control strain, while the HFXΔ2 strain expressing PhaA with either the OB-fold (amino acid residues 38 to 95) or the C terminus (amino acid residues 102 to 124) of PhaAβ deleted did not accumulate any PHA (Fig. 5B). These data suggested that the OB-fold and C terminus of PhaAβ were both indispensable for the activity of PhaA and that the N terminus was also essential for full PhaA activity. Taken together, the OB-fold is vital for the activity of both BktB and PhaA, while the N-terminal or C-terminal regions of BktBβ and PhaAβ are differentially important for the enzymatic activities of BktB and PhaA.

Fig 5.

Determination of the key regions of BktBβ and PhaAβ. (A) Domain prediction of BktBβ and PhaAβ by Pfam family search. BktBβ and PhaAβ both harbored an OB-fold, which spanned amino acid residues 32 to 94 in BktBβ and residues 35 to 98 in PhaAβ, respectively. The flanking regions of the OB-fold in either BktBβ or PhaAβ were designated as the N terminus (N-ter) and the C terminus (C-ter). (B) PHA production in the H. mediterranei HFXΔ2 strains carrying either the plasmid containing the bktB operon with region-deleted bktBβ (dark gray columns) or the plasmid containing the phaA operon with region-deleted phaAβ (light gray columns). bktB, the whole bktB operon; bktB βΔN, bktB operon with N-ter-deleted bktBβ; bktB βΔOB, bktB operon with OB-fold-deleted bktBβ; bktB βΔC, bktB operon with C-ter-deleted bktBβ; phaA, the whole phaA operon; phaA βΔN, phaA operon with N-ter-deleted phaAβ; phaA βΔOB, phaA operon with OB-fold-deleted phaAβ; phaA βΔC, phaA operon with C-ter-deleted phaAβ. PHA accumulation was detected when cells reached the stationary phase. All the data are shown as the mean values ± the SD of three independent experiments.

Functional interchangeability of BktBβ and PhaAβ.

In an attempt to examine the functional interchangeability of BktBβ and PhaAβ, we constructed different subunit combinations in vivo. The HFXΔ2 strain containing the bktBα and phaAβ genes was able to accumulate PHBV, and the HFXΔ2 strain containing the phaAα and bktBβ genes was able to accumulate PHB (Table 2). These data indicated that the hybrid enzyme composed of the BktBα and PhaAβ subunits had the same substrate specificity as BktB and that the enzyme composed of the PhaAα and BktBβ subunits was able to catalyze the synthesis of acetoacetyl-CoA like PhaA. These results suggested that the BktBα and PhaAα subunits demonstrated substrate specificity and the BktBβ and PhaAβ subunits were functionally interchangeable.

To further investigate whether there was a preference for combining the BktBα, BktBβ, PhaAα, and PhaAβ subunits into an active enzyme in vivo, the ΔbktBβ strain was constructed. The knockout strain of bktBβ only accumulated PHB without the 3HV monomer (Table 2). It appeared to have lost the BktB activity and retained only the PhaA activity. We then transformed a plasmid carrying the phaAβ gene under the control of the strong hsp5 promoter (29) into the ΔbktBβ strain. Surprisingly, increasing the copy number of phaAβ in the ΔbktBβ strain restored the ability of the strain to accumulate PHBV (Table 2). This phenomenon can be explained by assuming that PhaAβ prefers to interact with PhaAα. In the ΔbktBβ strain, PhaAα and PhaAβ could join together to form functional PhaA, leaving BktBα alone. However, when more PhaAβ is available in the ΔbktBβ strain, the excess PhaAβ could then interact with BktBα to form a hybrid enzyme with the same substrate specificity as BktB. These data suggest that the BktBβ and PhaAβ subunits are functionally interchangeable, but the PhaAβ subunit prefers to interact with the PhaAα subunit.

Potential application of BktB in PHBV production in haloarchaea.

Since BktB is involved in supplying 3-hydroxyvaleryl-CoA (3HV-CoA) during PHBV biosynthesis in H. mediterranei, it might have biotechnological potential for producing PHBV with high 3HV content in haloarchaea. H. hispanica is also a PHBV-producing haloarchaeon. However, the 3HV molar fraction of PHBV accumulated by this strain is very low (ca. 3 mol%) compared to that of H. mediterranei (24). To investigate the role of BktB in the supply of 3HV-CoA in H. hispanica, the bktB operon was introduced into the H. hispanica DF60 strain (a uracil auxotroph mutant of H. hispanica). GC analysis revealed that the 3HV molar fraction of PHBV accumulated by H. hispanica DF60 carrying the bktB operon reached a value of 9.7 mol%, ∼2.7 times of that observed in H. hispanica DF60 carrying the plasmid pWL502 (18) (negative control, 3.6 mol% of 3HV fraction). Therefore, BktB from H. mediterranei was also able to supply 3HV-CoA for PHBV synthesis in the Haloarcula species. Our results indicate that BktB has biotechnological potential in haloarchaea for production of PHBV with controllable 3HV content, from unrelated cheap carbon sources.

DISCUSSION

In this study, we identified two novel β-ketothiolases (BktB and PhaA) with different substrate specificities involved in PHBV biosynthesis in H. mediterranei. The discovery of BktB and PhaA has completed the identification of the enzymes involved in the PHBV biosynthesis pathway in H. mediterranei using glucose as the carbon source (see Fig. S2 in the supplemental material). Acetyl-CoA is produced from glucose via glycolysis, and propionyl-CoA can be produced through four different generation pathways (19). These two important metabolites, acetyl-CoA and propionyl-CoA, are then condensed by BktB and PhaA into acetoacetyl-CoA and 3-ketovaleryl-CoA, which are further reduced to 3-hydroxybutyl-CoA (3HB-CoA) and 3HV-CoA, respectively, via catalysis by PhaB1 and PhaB2 (17). The two monomers are finally polymerized into PHBV by PhaEC (15). Compared to the key enzymes of PHBV biosynthesis from unrelated carbon sources in bacteria, BktB and PhaA have been determined to harbor several distinct characteristics. The haloarchaeal PHA synthases (PhaEC), although similar to the class III PHA synthases from bacteria, also exhibit several unique features, including a longer C terminus in the PhaC subunit and certain conserved motifs (2).

β-Ketothiolases usually occur in multiple isozymes (30); eight β-ketothiolase homologs were found in H. mediterranei. Phylogenetic analysis revealed that the eight β-ketothiolase homologs could be classified into two groups (group I and group II) (Fig. 1). One possible explanation for this phenomenon is that β-ketothiolase gene duplication occurred in the haloarchaeal ancestor, and the two groups of β-ketothiolases then evolved independently to fulfill different physiological functions. Further gene duplication that occurred during these two independent evolution events may have allowed for adaptation to different environmental conditions (31), such as producing BktB and PhaA, which are in the same group (group I) but have different substrate specificities. In the present study, we have demonstrated the role of BktB and PhaA in PHBV biosynthesis in H. mediterranei, but the role of the remaining six β-ketothiolase homologs remains unknown.

The catalytic residues of BktBα and PhaAα have been determined to be S83-H281-H331 and S84-H282-H332, respectively (Table 3 and Fig. 4), which represents a distinct catalytic triad compared to that found in the bacterial biosynthetic β-ketothiolases (14). For example, the crystal structure of the PHA biosynthetic β-ketothiolase from Z. ramigera (PhbA_Zra) has been determined, and the catalytic residues have been identified as C89-H348-C378 (32). According to multiple sequence alignment, residue C89 in PhbA_Zra corresponded to residues S83 in BktBα and S84 in PhaAα, and residue H348 in PhbA_Zra corresponded to residues H331 in BktBα and H332 in PhaAα (Fig. 4). However, residue C378 in PhbA_Zra did not correspond to any of the catalytic residues in BktBα and PhaAα, nor did residues H281 and H282 in BktBα and PhaAα, respectively, correspond to any of the catalytic residues in PhbA_Zra (Fig. 4). These data demonstrated the differences between the catalytic triads of PHA biosynthetic β-ketothiolases from H. mediterranei and Z. ramigera. The catalytic mechanism of PhbA_Zra has been proposed to be a “ping pong” mechanism, which is commonly used by bacterial biosynthetic β-ketothiolases (9, 10, 13, 14, 32). In the first half of the reaction, residue H348 deprotonates and activates residue C89, and the activated C89 residue then forms a covalent acyl-enzyme intermediate via a nucleophilic attack on the substrate. In the second half of the reaction, the C378 residue activates another substrate molecule by abstracting a proton, and the activated substrate then nucleophilically attacks the acyl-enzyme intermediate to form the final product and the free enzyme. Mutation of residue C89 to S leads to a substantial loss of PhbA_Zra enzymatic activity (>100-fold lower) (33), even though this residue in PhbA_Zra exactly corresponded to serine residues (S83 and S84, respectively) in BktBα and PhaAα (Fig. 4). In addition, mutation of residue S84 to C in PhaAα reduced the PHB concentration by nearly 68%, but mutation of residue S83 to C in BktBα did not significantly affect the activity of BktB in PHBV biosynthesis (Table 3). These data demonstrated that the catalytic residues S83 in BktBα and S84 in PhaAα were not as stringent as residue C89 in PhbA_Zra. Interestingly, the flanking regions of residues H348 in PhbA_Zra, H331 in BktBα, and H332 in PhaAα displayed high sequence identity (Fig. 4), indicating the role of this catalytic residue might be conserved in both haloarchaeal and bacterial PHA biosynthetic β-ketothiolases.

Interestingly, the β-ketothiolase (BktB or PhaA) reported here contains two different types of subunits, the large catalytic subunit (BktBα or PhaAα) that determines the substrate specificity and the small subunit (BktBβ or PhaAβ) that is necessary for the enzyme activity. In contrast, the bacterial PHA biosynthetic β-ketothiolases reported thus far appear to be a tetramer with identical subunits (9–13). Since both BktBβ and PhaAβ harbor an OB-fold domain that is indispensable for the enzyme activity and the OB-fold is able to bind oligonucleotides, proteins, oligosaccharides, metal ions, and catalytic substrates (28), the BktBβ and PhaAβ may be involved in interaction with BktBα and PhaAα to form active enzymes. In addition, BktBβ and PhaAβ may interact with PHA-granule-associated proteins such as PhaEC and PhaP (18) to concentrate BktB and PhaA onto the PHA granules, and they may also help to deliver the substrates to the catalytic sites of BktBα and PhaAα. Interestingly, BLASTP analysis of BktBβ and PhaAβ indicated that, with the exception of one bacterial homolog in Fischerella sp. strain JSC-11 (NCBI protein database accession number WP_009460387), all of the BktBβ and PhaAβ homologs were archaeal, with a majority being haloarchaeal (see Table S4 in the supplemental material). Therefore, understanding the role of BktBβ and PhaAβ might shed light on the mechanisms underlying the adaptation of archaeal enzymes to specific extreme environments.

In contrast to the β-ketothiolases from bacteria, haloarchaeal β-ketothiolases are less understood. To our knowledge, only one β-ketothiolase with an unknown physiological role from Halobacterium strain ZP-6 has been purified and partially characterized (34). In the present study, we identified the first PHBV biosynthetic β-ketothiolases in haloarchaea, which are distinct from their bacterial counterparts in both subunit composition and catalytic residues. However, several interesting questions still need to be answered. Why is the small subunit (BktBβ or PhaAβ) necessary? Which key amino acid residues determine the different substrate specificities between BktB and PhaA? What is the catalytic mechanism of haloarchaeal β-ketothiolases? The answers to these questions will provide insight into the adaptation mechanisms of haloarchaeal enzymes to extreme environments and into the evolutionary processes of β-ketothiolases.