Comparison of four phaC genes from Haloferax mediterranei and their function in different PHBV copolymer biosyntheses in Haloarcula hispanica

Background The halophilic archaeon Haloferax mediterranei is able to accumulate large amounts of poly(3-hydroxybutyrate-co-3-hydroxyvalerate) (PHBV) with high molar fraction of 3-hydroxyvalerate (3HV) from unrelated carbon sources. A Polyhydroxyalkanoate (PHA) synthase composed of two subunits, PhaCHme and PhaEHme, has been identified in this strain, and shown to account for the PHBV biosynthesis. Results With the aid of the genome sequence of Hfx. mediterranei CGMCC 1.2087, three additional phaC genes (designated phaC1, phaC2, and phaC3) were identified, which encoded putative PhaCs. Like PhaCHme (54.8 kDa), PhaC1 (49.7 kDa) and PhaC3 (62.5 kDa) possessed the conserved motifs of type III PHA synthase, which was not observed in PhaC2 (40.4 kDa). Furthermore, the longer C terminus found in the other three PhaCs was also absent in PhaC2. Reverse transcription PCR (RT-PCR) revealed that, among the four genes, only phaCHme was transcribed under PHA-accumulating conditions in the wild-type strain. However, heterologous coexpression of phaEHme with each phaC gene in Haloarcula hispanica PHB-1 showed that all PhaCs, except PhaC2, could lead to PHBV accumulation with various 3HV fractions. The three kinds of copolymers were characterized using gel-permeation chromatography (GPC), differential scanning calorimetry (DSC), and thermogravimetric analysis (TGA). Their thermal properties changed with the variations in monomer composition as well as the different molecular weights (Mw), thus might meet various application requirements. Conclusion We discover three cryptic phaC genes in Hfx. mediterranei, and demonstrate that genetic engineering of these newly identified phaC genes has biotechnological potential for PHBV production with tailor-made material properties.


Background
Polyhydroxyalkanoates (PHAs) are polymers of hydroxyalkanoates and are synthesized by various bacteria and some haloarchaeal strains [1]. The molecular weight of PHAs is generally distributed in the range of 50,000-1,000,000 Da, which is sufficient to provide these polymers with similar characteristics as conventional plastics [2]. However, due to the high production cost of PHAs, semi-commercial production has been realized only for the two common polyesters, poly(3-hydroxybutyrate) (PHB) and poly(3-hydroxybutyrate-co-3-hydroxyvalerate) (PHBV) [3][4][5]. Recently, halophilic Archaea have been attracting strong interest in PHBV research due to their comparable advantages as PHA producers, including utilizing unrelated cheap carbon sources, non-strict sterilization and easy method for PHA extraction [6][7][8]. A few haloarchaeal strains have been found to synthesize intracellular short-chain-length PHAs (scl-PHAs), PHB and PHBV, to serve as carbon and energy storage [8][9][10][11][12][13][14][15][16]. Compared with PHB, PHBV possesses better polymer properties and thus has wider applications in commodity materials [17][18][19]. Among the halophilic archaea, Haloferax mediterranei is one of the best-studied strains due to its capability to accumulate large amounts of PHBV from inexpensive carbon sources, including glucose, starch, and whey [6][7][8]20,21]. Therefore, it has been evaluated as one of the most promising candidate prokaryotes for the industrial production of PHBV [7].
PHA synthases are the key enzymes of PHA synthesis and catalyze the incorporation of 3-hydroxyacyl-CoA into PHAs. Over the past few decades, extensive research has been devoted to the study of PHA synthases in the domain of Bacteria. This enzyme is grouped into four types in bacteria, according to its substrate specificity and subunit composition [22]. The typical bacterial type III synthase is composed of two subunits, PhaE (~40 kDa) and PhaC (~40 kDa) [23]. In the domain of Archaea, a novel PHA synthase homologous to the bacterial type III PHA synthase, with subunits designated PhaC Hme and PhaE Hme , has been identified and characterized in Hfx. mediterranei [6]. As in bacteria, the PHA synthase-encoding genes (phaEC Hme ) in Hfx. mediterranei are clustered and cotranscribed, and both PhaC Hme and PhaE Hme are indispensable for the PHBV synthesis from multiple unrelated carbon sources. The haloarchaeal PHA synthase and bacterial type III synthase share some conserved residues (e.g., the Cys-Asp-His catalytic triad). However, they are quite different in molecular weights and several characteristic motifs. The PhaC Hme (54.8 kDa) is much larger than its bacterial counterparts, in which the longer sequence at the C terminus has been shown to be indispensable for enzyme activity [6]. In contrast, the PhaE Hme (~20 kDa) subunit is much smaller than its bacterial counterparts. Moreover, although "PhaE box" is strongly conserved in PhaEs of bacteria, it is missing in PhaE Hme of Hfx. mediterranei.
Recently, the genome sequence of Hfx. mediterranei CGMCC 1.2087 has been completed in our laboratory (Han J. et al: Bioplastic production in the domain of Archaea: insights from the genome sequence of Haloferax mediterranei, in preparation). Intriguingly, sequence analysis revealed that three additional phaC paralogs (named phaC1, phaC2, and phaC3) were present and scattered throughout the whole genome. In this study, we compared the features of PhaC Hme and these three putative PhaC proteins (PhaC1-PhaC3), and determined their capabilities of PHA polymerizing in Haloarcula hispanica. Furthermore, the material properties of the produced polymers were also characterized using multiple methods.

Results
Identification and analysis of the phaC genes in Hfx. mediterranei Hfx. mediterranei is comprised of one chromosome and three extrachromosomal elements (megaplasmids) [24]. The genome sequence of Hfx. mediterranei CGMCC 1.2087 has been recently completed by our group. The genes encoding its PHA synthase, named phaE Hme and phaC Hme , have already been cloned and characterized [6] and were found to be located on megaplasmid pHM300. Interestingly, analysis of the whole genome sequence revealed that three other phaC paralogs existed in this strain and were designated phaC1, phaC2, and phaC3.
The phaC1 was mapped to the chromosome, whereas phaC2 and phaC3 were both mapped to megaplasmid pHM500 ( Figure 1). Most of the genes adjacent to the four phaCs encoded conserved hypothetical proteins with unknown functions (Figure 1). The four PhaC proteins of Hfx. mediterranei exhibited homology to each other (43%-58% identity), and they all contained the catalytic triad residues (Cys-Asp-His) ( Figure 2). Both the "Lipase box-like" sequence (Gly-X-Cys-X-Gly-Gly) and the conserved motif of type III PHA synthase were present in PhaC Hme , PhaC1, and PhaC3 ( Figure 2). In the case of PhaC2, the last "Gly" residue of the "Lipase boxlike" sequence was replaced by an "Ala", and the conserved motif of the type III PHA synthase was also not strongly conserved ( Figure 2). The deduced molecular weights of the four PhaCs were as follows: PhaC Hme -54.8 kDa, PhaC1-49.7 kDa, PhaC2-40.4 kDa, and PhaC3-62.5 kDa. Only the molecular weight of PhaC Hme was equivalent to that of the PhaCs from Haloarcula marismortui (53.1 kDa; Accession no., YP_137339), Haloarcula hispanica (53.0 kDa; Accession no., ABV71394), and Haloquadratum walsbyi (52.1 kDa, Accession no., YP_658052). When compared with the primary sequences of the other three PhaCs, the C terminal sequence was missing in PhaC2, which resulted in the smallest molecular weight.
Only phaC Hme was transcribed in Hfx. mediterranei in laboratory conditions In previous work, we found that the phaEC Hme -deleted strain Hfx. mediterranei PHBV-1 could not accumulate detectable scl-PHAs and expression of phaE Hme alone could not restore the ability to accumulate scl-PHAs [6], which indicated that there were no additional functional PhaC subunits in Hfx. mediterranei PHBV-1. To provide some clues to this phenomenon, RT-PCR with gene-specific primer pairs for the four phaC genes (Table 1) was employed to determine whether they were transcribed in the wild-type strain. The lengths of the PCR products of phaC Hme , phaC1, phaC2, and phaC3 were 382 bp, 391 bp, 408 bp, and 412 bp, respectively. The results demonstrated that of the four genes, only phaC Hme was actively transcribed in MG medium ( Figure 3A), which was consistent with our previous observation that PhaC Hme was indispensable for PHBV biosynthesis in Hfx. mediterranei, as the other three phaCs were not expressed.

Genetic analysis of the functions of the four PhaCs in Haloarcula
To determine if phaC1, phaC2, and phaC3 could encode functional proteins, a PHA synthase genedeleted strain, named Har. hispanica PHB-1 [13], was used to heterologously express these proteins. Because PhaC proteins were proposed to be functional combined with the PhaE subunit, the three expression plasmids designated pWL3EC1, pWL3EC2, and pWL3EC3 (Table 1) were constructed based on the PhaE Hmeexpression plasmid pWL3E [6] and individually introduced into Har. hispanica PHB-1. During fermentation, Har. hispanica PHB-1/pWL3E [6] and the wild-type strain were used as the negative and positive control,  respectively. To compare the PHBV synthesized by the four PHA synthases with distinct PhaC subunits, Har. hispanica PHB-1/pWL3EC [6] harboring the phaEC Hme genes of Hfx. mediterranei was also employed. Table 2 summarized the results of PHA accumulation in Har. hispanica and its five transformants. Compared with the wild-type strain, there was no significant difference in final cell yield as evaluated by cell dry weight (CDW). The expression of PhaE Hme alone or coexpression with PhaC2 did not lead to the synthesis of any PHA. However, the other three PhaC proteins (PhaC Hme , PhaC1 and PhaC2) were functional during PHBV polymerization when excess glucose was supplied. PHBV content accumulated in recombinant PHB-1 strains harboring PhaEC Hme (17.33%) or PhaEC1 (15.61%) was a little higher than that of wild type (12.26%), and resulted in a much higher PHBV concentration of 0.99 g/l or 0.94 g/l, respectively, as compared to 0.58 g/l in wild type. In the case of PhaC3, the PHBV content (10.02%) and concentration (0.53 g/l) were Sequences representing the restriction sites are underlined. similar with that of wild type. Compared to that of wild type, the 3HV monomer composition of PHBV was a little higher for PhaEC1 (4.06 mol %), a little lower for PhaEC3 (1.95 mol %), and equivalent for PhaEC Hme (3.14 mol %). In short, three types of PHBV with altered 3HV fractions had been synthesized by the Har. hispanica PHB-1 recombinants with phaE Hme and different phaC genes.

Properties of PHBV synthesized by different PhaC subunits
A 400 MHz 1 H NMR study showed that the monomer composition of PHBV produced by the wild-type strain of Har. hispanica and its recombinants (data not shown) was in agreement with that determined by the GC analysis (Table 2). Table 3 Table 2). DSC analysis showed that the glass-transition temperatures (T g ) of the PHBV samples were 2.3°C (PhaEC Hme -PHBV), -0.1°C (PhaEC1-PHBV), and 5.7°C (PhaEC3-PHBV). This result was consistent with the 3HV monomer composition of the three types of PHBV; a larger amount of 3HV incorporated into PHBV could lead to lower T g . A similar melting temperature (T m ) was observed for all three PHBV samples, of which the PHBV with the lowest T m synthesized by PhaEC1 could facilitate its processing ability. Compared with the other two polymers, no cold crystallization temperature (T c ) was detected for the PHBV accumulated by PhaEC3. This result demonstrated that this polymer (with a high 3HB fraction) possessed a fast crystallization speed, which could also provide some advantages in processing.
Thermal stability plays an important role in polymer melt processing. This characteristic of the three types of PHBV was determined by TGA (Table 3 and Figure 4). The temperature at 5% weight loss (T d ) was used to evaluate their polymer thermal stability. The T d of the three polymers ranged from 252°C to 257°C. Our results showed that the three types of PHBV, with different monomer composition, changed obviously in their thermal properties.

Discussion
Multiple phaC genes in a same genome have been found in the domain of Bacteria. The complete genome sequence of Ralstonia eutropha has revealed the existence of a second phaC gene, which is not clustered with the phaCAB operon, and its function has not been clarified [25]. Additionally, Pseudomonas strains synthesizing medium-chain-length PHAs (mcl-PHAs) usually have two different phaC genes, which are separated by a phaZ gene that encodes an intracellular PHA depolymerase [1]. To date, the haloarchaeon Hfx. mediterranei has been found to possess the largest number of phaC genes. Compared with the other PhaC subunits from haloarchaeal strains, the three newly identified PhaCs  showed less homology with PhaC Hme . A phylogenetic tree was constructed based on the four PhaCs from Hfx. mediterranei and seven other PhaCs from Halobacteriaceae; 3 PhaC sequences from Thaumarchaeota were used as the outgroup ( Figure 5). The PhaCs from Halobacteriaceae and those from Thaumarchaeota clustered separately. PhaC Hme was more closely related to the PhaCs from other haloarchaea than to its three paralogs ( Figure 5). It seemed likely that phaC1, phaC2 and phaC3 were not directly acquired from a common source of PhaC Hme , but possibly by horizontal DNA transfer from other sources during evolution.
Of the four paralogs in Hfx. mediterranei, only phaC Hme was found to be transcribed under PHA-accumulating conditions by RT-PCR assays ( Figure 3A). However, the putative promoters (including the TATA box and BRE element) of the other three phaC genes, were also detected ( Figure 3B). As both phaC1 and phaC3 encoded functional proteins, it is conceivable that these silent genes in Hfx. mediterranei might be activated under specific ecological conditions to increase the environmental distribution of this haloarchaeon.
Regardless of the conserved catalytic triad residues (Cys-Asp-His) in the primary sequence, PhaC2 was found to have no PHA polymerization activity. This might be the result of the truncated C terminus, which has been shown to be indispensable for PHA synthase enzyme activity [6]. However, a recombinant PhaC2 containing the C terminus from PhaC3 does not have the ability to synthesize PHAs (data not shown). Alternatively, less conserved motifs such as the "Ala" residue in the "Lipase box-like" sequence or the lack of the conserved motif of PhaC might lead to its inactivity.
The incorporation of 3HV into PHB could improve material flexibility. Three types of PHBV with different monomer compositions were produced from the Har. hispanica PHB-1 recombinants, and the type of PHBV was supposed to be determined by the different substrate specificities of PhaC Hme , PhaC1, and PhaC3. Hfx. mediterranei has been found to be the best PHBV producer in the family Halobacteriaceae, and can accumulate the highest PHBV content (>60%) with the highest 3HV fraction from many unrelated cheap carbon sources [7,20,21]. Compared with Har. hispanica, Hfx. mediterranei has a stronger ability to provide more 3HB-CoA and 3HV-CoA monomers for PHA synthase. Therefore, the use of these phaC genes in Hfx. mediterranei PHBV-1 to produce PHBV with higher 3HV fractions and different properties would have much more promise, which could extend the applications of PHAs in the future.

Conclusions
In this study, we mined three silent phaC genes (phaC1, phaC2, and phaC3) from Hfx. mediterranei, which encoded potential PhaCs subunits of PHA synthase. In Har. hispanica PHB-1, the heterologous coexpression of phaE Hme with phaC Hme , phaC1, or phaC3 could lead to the accumulation of PHBV polymers with different molecular weights and thermal properties from excess glucose. Therefore, we concluded that genetic engineering of these phaC genes in Hfx. mediterranei would produce PHBV with desirable material properties, at lower production costs.

Strains, plasmids, and growth conditions
The plasmids used in this study were listed in Table 1. All of the plasmids were derivatives of pWL3E [6].
E. coli JM109 was cultivated in Luria-Bertani (LB) medium at 37°C [27]. When needed, ampicillin (100 mg/l) was added to the medium. Har. hispanica and Hfx. mediterranei were grown in AS-168 medium. For growth on plates, the media described above were supplemented with 1.2% (w/v) agar. For PHA accumulation, the haloarchaeal strains were transferred from AS-168 medium to MG medium and cultivated as previously described [6]. When required, mevinolin was added to a final concentration of 5 mg/l for H. hispanica recombinants. All of the strains were cultured at 37°C.

Primers and plasmid construction
All of the primers used in this study were summarized in Table 1. To determine the capabilities of PhaC1, PhaC2, and PhaC3 to synthesize PHAs, three expression plasmids were constructed. The expression vectors were derived from pWL3E, which contained the phaE Hme gene of Hfx. mediterranei with its native promoter [6]. The coding sequences of phaC1, phaC2, and phaC3 were amplified from Hfx. mediterranei by PCR with primer pairs phaC1-BamHI-F/phaC1-KpnI-R, phaC2-BamHI-F/phaC2-KpnI-R, and phaC3-BamHI-F/phaC3-KpnI-R, respectively. The purified PCR products (phaC1, 1.3 kb; phaC2, 1.1 kb; phaC3, 1.7 kb) were digested with BamHI and KpnI and inserted into BamHI and KpnI-treated pWL3E. The resulting expression plasmids pWL3EC1, pWL3EC2, and pWL3EC3 were confirmed by DNA sequencing and individually introduced into Har. hispanica PHB-1. The haloarchaeal recombinants harboring pWL3EC [6], pWL3EC1, pWL3EC2, and pWL3EC3 were subjected to PHA accumulation analysis in which Har. hispanica PHB-1 harboring pWL3E and the wild-type strain were used as the negative and positive control, respectively.

Isolation of total RNA and reverse transcription PCR (RT-PCR)
Total RNA of Hfx. mediterranei was extracted using TRIzol reagent (Invitrogen) from a logarithmic-phase culture grown in MG medium. Four specific primer pairs, phaC-RT-F/phaC-RT-F, phaC1-RT-F/phaC1-RT-F, phaC2-RT-F/phaC2-RT-F, and phaC3-RT-F/phaC3-RT-F, were designed for the phaC Hme , phaC1, phaC2, and phaC3 genes, respectively (Table 1). Prior to RT-PCR, the specificity of these primers were confirmed by sequencing of the corresponding PCR products using the genomic DNA of Hfx. mediterranei as template. For RT-PCR, the RNA sample was digested with RNase-free RQ1 DNase (Promega) overnight to eliminate any DNA contamination, which was subsequently confirmed by a control PCR without prior reverse transcription as previously described [28]. The DNA-free RNA sample was then used as the template for RT-PCR using the One- Step RT-PCR Kit (Qiagen) according to the manufacturer's instruction. The PCR products were electrophoresed in a 1% agarose gel.

Measurement of PHA content and composition
Scl-PHAs were qualitatively and quantitatively analyzed by gas chromatography (GC) with the Agilent GC-6820 as described previously [6,13]. Benzonic acid was used as an internal standard to calculate the amount of PHAs. PHBV (Sigma) was used as the standard for peak identification. The average results of three parallel experiments were recorded.

Extraction and purification of PHAs
After fermentation, the liquid cultures were harvested by centrifugation at 9000 rpm for 10 min. The cell pellets were resuspended in distilled water containing 0.1% SDS (w/v) and agitated for 20 min. The suspension was centrifuged at 5000 rpm for 5 min and yielded whitish, compact sediment. The sediment was resuspended and washed twice with 0.1% SDS solution (w/v) and distilled water, respectively. The resulting paste was dried in an oven at 70°C for 2 h for further purification. The extracted PHAs were treated with hot chloroform at 90°C for 4 h, followed by precipitation with 15 volume of pre-chilled ethanol. To obtain pure product, the precipitate was then redissolved in chloroform and the process was repeated. The collected PHAs were dried at 40°C for 24 h to remove all residual solvent.

Molecular weight determination of PHBV
The molecular weight of PHBV was estimated via gel permeation chromatography (GPC) with a Waters 1515 system at 30°C using a Waters 1525 pump equipped with a four Styragel column series (Styragel HR, 5 μm). A differential refractive index detector and a UV detector were both employed. Chloroform was used as the elution solvent at a flow rate of 1.0 ml/min. The concentration of the samples was 2 mg/ml and the injection volume was 50 μl. The calibration curve was generated with polystyrene standards (Sigma).

Characterization of PHBV thermal properties
Differential scanning calorimetry (DSC) was performed using a TA Instruments Q100 with an autocool accessory calibrated with indium to characterize thermal transitions of the polymers. The following protocol was used for each sample. A 3-5 mg sample in an aluminum pan was cooled from room temperature to -60°C by the autocool accessory. The pan was heated from -60°C to 180°C at a rate of 10°C/min, isothermally maintained at 180°C for 5 min, quenched to -60°C, and reheated from -60°C to 180°C at 10°C/min under a nitrogen flow rate of 50 ml/min. Data were collected during the second heating run.
Thermal stability of the PHA polymer was determined via thermogravimetric analysis using a Q50 thermogravimetric analyzer (TGA) (TA Instruments). The procedures used were as the same as those described by Xie et al. [29]. Briefly, samples were analyzed at a heating rate of 10°C/min under a nitrogen atmosphere with a N2 flow rate of 60 ml/min.