A BLAST-protein (BLASTP) search was performed against the protein sub-sections of the National Center for Biotechnology Information (NCBI) and DNA Data Bank of Japan (DDBJ) databases using the PhaC_Ac amino sequence as a template (Accession No. BAA21815) (Altschul et al., 1990). PhaCs with more than 85% similarity index and an identity index of 50%–60% in the BLASTP search were targeted as potential PhaCs with broad substrate specificities. Among the various PhaCs from different organisms that satisfied the criteria in the BLASTP search, four PhaCs were selected based on the diversity of the N-terminal region for further evaluation. Phylogenetic analyses were performed using the maximum likelihood method in MEGA11 (Tamura et al., 2021) and the protein sequences were aligned using ClustalW. This analysis involved six amino acid sequences: PhaC_Ac, four newly selected PhaCs from BLASTP, and PhaC from Ralstonia eutropha (WP_011615085) as an outgroup.

Four PhaC amino acid sequences were chosen based on the BLASTP search results. These phaC genes were chemically synthesized with optimized codon usage in E. coli by Eurofins Genomics Co. Ltd. (Tokyo, Japan) for plasmid construction and evaluation. E. coli LSBJ, a fadB fadJ double-deletion strain of E. coli LS5218 [fadR601, atoC (Con)] (Tappel et al., 2012a), was used as the host strain for PHA biosynthesis. This strain is an ideal host for non-native PHA production because of its ability to take up a wide variety of substrates to be incorporated into PHA homo- and copolymers, and bench-level scale-up methodologies available for overall production (Tappel et al., 2012b; Levine et al., 2016; Pinto et al., 2016; Fadzil et al., 2018; Furutate et al., 2021; Scheel et al., 2021). A broad-host-range plasmid pBBR1MCS-2 (Kovach et al., 1995) harboring the genes encoding the PhaCs to be evaluated, the lac promoter region, the (R)-specific enoyl-CoA hydratase gene from A. caviae (phaJ _Ac), the 3-ketothiolase gene (phaA) from Ralstonia eutropha H16, and the acetoacetyl-CoA reductase gene (phaB) from R. eutropha H16, termed pBBR1-phaCsAB_ReJ_Ac, was used for the expression of PhaCs (Supplementary Figure S1). For phaAB expression, the R. eutropha pha promoter and terminator regions were located upstream and downstream of their genes, respectively. To enhance the supply of 3HHx, 3H4MV, and 3H2MB monomers, the plasmid pTTQ-PCT (Furutate et al., 2017) containing the propionyl-CoA transferase (PCT) gene from Megasphaera elsdenii (pct) (Taguchi et al., 2008) was introduced into the E. coli LSBJ strain (Supplementary Figure S1).

Initially, recombinant E. coli LSBJ was incubated overnight at 37°C with reciprocal shaking (160 rpm) in a 50 mL baffle flask containing 20 mL of lysogeny broth (LB) medium as a seed culture. The LB medium contained 10 g/L Bacto-tryptone (Difco Laboratories, Detroit, MI, United States), 5 g/L Bacto-yeast extract (Difco Laboratories), and 10 g/L NaCl. For plasmid maintenance throughout the initial incubation period, 50 mg/L of kanamycin and 50 mg/L of carbenicillin were added.

Inoculations for PHA production were started with 5 mL of seed culture added to 500 mL shake flasks containing 95 mL of modified M9 medium (Furutate et al., 2021) (final volume:100 mL and 5% inoculum). The modified M9 medium comprised of 17.1 g/L Na₂HPO₄·12H₂O, 3 g/L KH₂PO₄, 0.5 g/L NaCl, 2 mL of 1 M MgSO₄·7H₂O, 0.1 mL of 1 M CaCl₂, and 2.5 g/L Bacto-yeast extract. For plasmid maintenance during PHA production, 50 mg/L of kanamycin and 50 mg/L of carbenicillin were added. Additionally, 1 mM isopropyl-β-D-thiogalactopyranoside (IPTG) was used to induce phaJ and pct gene expression. The P(3HB) homopolymer was synthesized from 20 g/L glucose, which was added at the beginning of the culture at 30°C for 72 h. For the synthesis of 3HB-based copolymers, the total incubation time was set to 76 h, in which an initial step for 4 h at 30°C with reciprocal shaking (130 rpm) was performed before the addition of IPTG, precursors, and glucose, and further cultured for 72 h. Hexanoic acid, 4-methylvaleric acid, trans-2-methylbut-2-enoic acid (tiglic acid), and 2,2-dimethyl-3-hydroxypropionic acid (3-hydroxypivalic acid), which had previously been converted to their respective sodium salts, were used as precursors for the 3HHx, 3H4MV, 3H2MB, and 3HPi units, respectively (Füchtenbusch et al., 1998; Tanadchangsaeng et al., 2009; Watanabe et al., 2015). These precursors are known to inhibit cell growth, and a high concentration of glucose can repress phaJ and pct genes, otherwise induced by IPTG. Thus, lower concentrations of glucose and the precursors were added intermittently to the culture medium (at 4, 28, and 52 h). A total of 7.5 g/L glucose (2.5 g/L each time) and 0.6 g/L precursors (0.2 g/L each time) were added throughout the main incubation period. Finally, cells were harvested by centrifugation and lyophilized for further analysis. The relationship between the precursors used and biosynthesized polymers is shown in Figure 1.

To construct mutated phaC _Ps , a substitution (N175G) was introduced into the gene by overlap extension PCR (Supplementary Figure S2) (Warrens et al., 1997). The primers for amino acid substitution were designed and chemically synthesized as follows:5′-GGCGGCCGCTCTAGAACTAGTGGATCCCGGGGCAA-3′ and 5′- CAC​TAA​GTT​TTG​ACCGCCGTT​CTC​CAA​GGT-3′ for an amplification of the 1.4-kb fragment, 5′-GCG​CTT​GGA​GGC​CGG​CAC​CG-3′ and 5′- GTG​ACC​TTG​GAG​AACGGCGGT​CAA​AAC​TTA-3′ for an amplification of the 2.3-kb fragment. The underlined sequence in the primer indicates the codon used to replace Asn175 (AAT) with Gly (GGC). The resulting plasmid carrying the mutated gene was introduced into E. coli LSBJ along with pTTQ-PCT for PHA biosynthesis analysis.

The dry cell weight was gravimetrically measured after centrifuging the culture medium at 6,000 × g for 10 min at room temperature three times (once for collecting the cells, discarding the medium, and twice to wash away the remaining salts with water) and lyophilized for approximately 3 days.

PHA content, PHA yield, and 3HA monomer composition were determined by gas chromatography (GC) using a Shimadzu GC-2014s instrument (Shimadzu, Kyoto, Japan) with a flame ionization detector. Lyophilized cells were methanolyzed to convert PHA into 3HA-methyl ester constituents in the presence of 15% sulfuric acid for GC analysis. The methanolysis reaction was carried out at 100°C for 140 min, except for 3H2MB- and 3HPi-containing polymers, for which the reaction time was set to 8 h to increase the reaction yield. The methanolyzed samples were allowed to cool to room temperature, and 1 mL of deionized water was added to separate the polar components from the non-polar components. The non-polar fraction containing 3HA-methyl ester was filtered, and an equal volume of chloroform solution containing 0.1% (w/v) methyl-n-octanoate as an internal standard was added to prepare the final sample for GC analysis. The samples were injected through the GC capillary column InertCap 1 (30 m × 0.25 mm, GL Science, Tokyo, Japan). The column temperature was initially set at 90°C for 2 min, increased to 110°C at a rate of 5°C/min, and then increased to 280°C at a rate of 20°C/min. The signal peak areas obtained were calculated for the total PHA content and 3HA monomer composition.

The molecular weight of P(3HB) synthesized using various PhaC enzymes was determined by gel permeation chromatography (GPC) using a Shimadzu Nexera GPC system with an RI-504 refractive index detector (Shodex, Tokyo, Japan) equipped with two KF-406 LHQ joint-columns (at 40°C, Shodex, Tokyo, Japan). Chloroform was used as the mobile phase at a flow rate of 0.3 mL/min. The sample concentration and injection volume were set at 1 mg/mL and 10 μL, respectively. Polystyrene standards with low polydispersity were also analyzed as reference standards to construct a calibration curve.

A BLASTP search was performed against the protein sub-sections of the NCBI and DDBJ databases using the amino acid sequence of PhaC_Ac, the first enzyme characterized by the natural copolymerization of 3HB and 3HHx monomers to PHA copolymers. Four PhaCs from the bacteria Ferrimonas marina (Katsuta et al., 2005), Plesiomonas shigelloides (Ferguson and Henderson 1947; Janda et al., 2016), Shewanella pealeana (Leonardo et al., 1999), and Vibrio metschnikovii (Lee et al., 1978) were selected for further evaluation, because of the diversity of the N-terminal region such as positions 149 and 171 in PhaC_Ac. These PhaCs were identified as Class I PHA synthases, which have a high potential for synthesizing scl-mcl PHA copolymers in a manner similar to PhaC_Ac, based on their homology. Although these bacteria were discovered long ago, their ability to produce PHA has not yet been studied.

A comparison with the amino acid sequence of PhaC_Ac revealed that the four PhaC enzymes identified in this study shared 85%–91% similarity and approximately 55% identity with PhaC_Ac (Table 1). Multiple sequence alignment of PhaCs is shown in Figure 2. All new PhaCs have a PhaC box sequence at the active site, which is typically described as G-X-C-X-G-G (where X is an arbitrary amino acid), and cysteine (Cys³¹⁹ in PhaC_Ac) is the active center (Nambu et al., 2020). In PhaC_Ac, the active sites Cys³¹⁹, Asp⁴⁷⁵, and His⁵⁰³ have been proposed to form a catalytic triad (Tsuge et al., 2007a), which are all conserved in the newly identified PhaC enzymes. In contrast, PhaC from P. shigelloides has a primary sequence of approximately 30 amino acid residues greater than that of others and exhibits relatively low sequence homology in the C-terminal region. The phylogenetic tree shown in Figure 3 indicates that PhaC from F. marina is closely related to PhaC_Ac, whereas PhaC enzymes from S. pealeana and V. metchniskovii are evolutionarily distinct. PhaC from P. shigelloides is neither closely related nor evolutionarily distant from PhaC_Ac. To the best of our knowledge, no study has explored PhaC enzymes isolated from these bacteria for PHA production. Thus, genes encoding the four PhaC enzymes were chemically synthesized with optimized codon usage in E. coli. The DNA sequences are included in Supplementary Information.

The biosynthesis of P(3HB) from 20 g/L glucose using one of the four PhaC enzymes is summarized in Table 2. P(3HB) accumulation ranging from 38.4 wt% to 54.2 wt% was achieved using the new PhaC enzymes, which was comparable to PhaC_Ac and its variant PhaC_AcNSDG. Thus, all newly identified PhaC enzymes showed great potential as biocatalysts for P(3HB) production. However, PhaCs from P. shigelloides (PhaC_Ps) showed the highest P(3HB) accumulation among the wild-type PhaCs tested.

The molecular weight is a crucial aspect in determining the suitability of a material for various commercial uses (Sudesh et al., 2000). The weight-average molecular weight (M _w ) is more closely related to material properties than the number-average molecular weight (M _n ). For PHA, ultrahigh molecular weight polymers can form strong fibers (Tsuge, 2016), thus meeting the requirements for practical use. In addition, a low polydispersity index (PDI) (Tsuge, 2016) also plays a significant role in determining the suitability of PHA for specific applications. However, not all PhaC enzymes can synthesize PHAs with high M _w and low PDI. In this study, PhaC_Ps synthesized P(3HB) with an ultrahigh M _w , which exceeded 3 × 10⁶, with a relatively low PDI below 1.5 (Table 2). Moreover, the other two identified PhaC enzymes from F. marina and S. pealeana could also synthesize P(3HB) with M _w of approximately 2 × 10⁶ with PDIs ranging from 1.3 to 1.5. PhaC_Vm from V. metschnikovii proved to be an exception, with PDI >2.5. The currently available PhaC_Ac is highly sensitive to ethanol (Hiroe et al., 2015), which is a metabolite of some bacteria, including E. coli, and functions as a chain transfer agent to terminate polymerization reactions (Tsuge, 2016), resulting in the synthesis of relatively low-molecular-weight PHA. The new PhaCs reported in this study may be less sensitive toward ethanol, thereby producing PHA with high M _w and low PDI. These new PhaC enzymes, especially PhaC_Ps, exhibited superior M _w and PDI values compared with PhaC_Ac and its NSDG variant, which could benefit PHA processing and material properties.

The new PhaC enzymes were evaluated for their substrate specificities alongside PhaC_Ac and its NSDG variant for incorporating 3HHx, 3H4MV, 3H2MB, and 3HPi monomers. Biosynthesis was performed using four precursors (hexanoic acid, 4-methylvaleric acid, tiglic acid, and 3-hydroxypivalic acid) in the presence of glucose (Figure 1). These precursors are toxic to cells, thus inhibiting cell growth and subsequently lowering PHA accumulation in bacteria. As PHA production is associated with cell growth (Sudesh et al., 2000), it is imperative to eliminate or reduce the risk of toxicity induced by such precursors. Therefore, the precursors were introduced into the culture medium after 4 h, once substantial cell growth was achieved, mainly for better tolerance (Furutate et al., 2021). Meanwhile, a high glucose concentration can cause catabolic repression of phaJ and pct genes induced by IPTG (Furutate et al., 2021); thus, the glucose concentration was maintained at a minimum to promote cell growth only. Glucose and its precursors were intermittently added to allow for better uptake of the second monomer, with no or fewer unanticipated effects on the cells. The details of the biosynthesis results are summarized in Tables 3–6.

All PhaCs, except PhaC from S. pealeana (PhaC_Sp), were able to incorporate all targeted monomers (3HHx, 3H4MV, 3H2MB, and 3HPi). PhaC_Sp copolymerized 3HB with 3HHx or 3HPi, but not 3H4MV or 3H2MB. The number of PhaC enzymes with broad substrate specificity is scarce; thus, the new PhaC enzymes reported in this study are highly intriguing for future studies. Furthermore, PHAs containing α-carbon methylated units are potentially attractive bio-based materials (Füchtenbusch et al., 1998; Furutate et al., 2021); thus, PhaCs with the ability to polymerize 3H2MB and 3HPi are of great interest. PhaC_Ps demonstrated superior performance in the polymerization of 3HPi to PhaC_Ac. Additionally, all PhaC_Ps-expressing strains showed higher PHA content than PhaC_Ac-expressing strains. Therefore, the potential of PhaC_Ps was further explored using site-directed mutagenesis.

In vitro evolution of PhaC is a powerful approach for enhancing the productivity and quality of PHA (Kichise et al., 2002; Taguchi and Doi, 2004). For instance, PhaC_AcNSDG, a variant of PhaC_Ac, exhibits enhanced performance (such as production yield and substrate specificity) compared to that of the wild-type enzyme (Tsuge et al., 2007b). In addition, various studies have proven the efficacy of PhaC engineering in PHA production towards the formation of super biocatalysts for tailor-made PHAs (Taguchi and Doi, 2004; Nomura and Taguchi, 2007). Therefore, PhaC_Ps, which exhibited the best performance among the new PhaCs, were selected for site-directed mutagenesis to study their potential positive effects on PHA production. Considering that the double mutation of PhaC_AcNSDG, amino acid substitutions of N149S and D171G drastically enhanced the performance of the enzyme (Tsuge et al., 2007b; Harada et al., 2021), similar efforts were adopted to generate a PhaC_Ps variant. According to the alignment (Figure 2), PhaC_Ps naturally contain a serine residue at the corresponding position of 149 in PhaC_AcNSDG. Thus, a single amino acid substitution was performed in PhaC_Ps in which asparagine 175 was changed to glycine (N175G). The resultant variant was termed PhaC_PsNG, and its PHA production ability was examined.

Interestingly, PhaC_PsNG showed enhanced P(3HB) synthesis, while maintaining a high molecular weight (Table 2). PhaC_PsNG exhibited enhanced activity for the incorporation of the α-methylated monomer 3H2MB compared with the parent enzyme and PhaC_AcNSDG (Table 5). This indicates the potential of PhaC_PsNG to surpass the currently best-performing enzyme (PhaC_AcNSDG) for the incorporation of the 3H2MB monomer. Moreover, PhaC_PsNG was shown to have a better ability to incorporate 3H4MV and 3HPi than the parent enzyme but less so than PhaC_AcNSDG (Table 4 and Table 6). Finally, PhaC_PsNG exhibited an almost similar level of 3HHx incorporation as the parent enzyme, which was inferior to PhaC_AcNSDG (Table 3).