All strains and plasmids used in this study are listed in Table 1.

Plasmids were constructed by sequence- and ligation-independent cloning (Li and Elledge, 2007; Jeong et al., 2012; Islam et al., 2017) and schematic diagrams of plasmid construction are shown in Figure 1 The primers used in this study are listed in Table 2. To construct the recombinant plasmid pVEC01, alp7 was amplified from pTKW106alp7 (Addgene, Watertown, United States) by polymerase chain reaction (PCR), and pUC19 was digested with XbaI and BamHI. To construct the recombinant plasmids pVEC02 and pVEC03, the hok/sok locus was amplified from pTKW106alp7 by PCR, and the vectors pUC19 and pVEC01 were digested with SalI. To construct the recombinant plasmid pPHB01, the phaCAB operon was amplified from the genomic DNA of C. manganoxidans and pVEC02 was digested with SphI. After the preparation of inserts and vectors, the DNA fragments were purified using the Gene-Spin™ 1-4-3 DNA Extraction Kit (PROTECH, Taipei, Taiwan). The insert and corresponding vector were mixed at an appropriate ratio and incubated at 37°C for 1 min with T4 DNA polymerase to generate 5′-overhangs. Thereafter, the reaction mixture was placed on ice for 20 min for single-strand annealing and was introduced into competent E. coli DH5α cells for transformation.

To construct the recombinant plasmid pPHB01-1, the Q5^® site-directed mutagenesis kit (NEB, Ipswich, United States) was used to insert a 57-bp E. coli ρ factor-independent terminator, thrLABC (ECK120033737) between the phaCAB operon and the hok/sok locus in pPHB01. The primers used for constructing pPHB01-1 were In/del-F- pPHB01-1 and In/del-R-pPHB01-1 (Table 2). After PCR, the amplified DNA was treated with a Kinase-Ligase-DpnI enzyme mix at room temperature for 5 min for circularization and template removal. Mutant selection was enriched by simultaneous ligation and DpnI treatment. Next, 5 μL of the KLD mix was directly introduced into competent E. coli DH5α cells. Colony PCR was used for pPHB01-1 screening with the primers CK-F-pPHB01-1 (in-del) and CK-R-pPHB01-1 (in-del) (Table 2). Finally, pPHB01-1 was confirmed by DNA sequencing. All recombinant plasmids constructed in this study were verified by colony PCR, restriction enzyme digestion, and DNA sequencing.

Escherichia coli strains DH5α, MG1655, BL21 (DE3), and MZLF (Yang et al., 2016) were used as hosts to test plasmid stability. Cells for the plasmid stability test were grown from -80°C stocks by streaking on an LB plate with ampicillin. After 16–24 h, a single colony was picked and used to inoculate LB containing antibiotics for 12 h. Then, 10 μL of the cell suspensions was inoculated into 3 ml LB without antibiotics to start the stability experiment. The inoculum was diluted and plated on LB plates supplemented with or without antibiotics.

After 12 h, 10 μL of the cell suspensions was transferred to a new test tube. Every 12 h, 10 μL of the sample was transferred to a new flask, and the cells were diluted and plated on LB plates supplemented with or without antibiotics. Plasmid stability (segregation) was estimated by the equation: Ampicillin − resistant cells  ( % )   = colonies on   L B + ampicillin colonies on LB × 100 % .

The putative promoter of the phaCAB operon was predicted by BPROM (http://www.softberry.com/berry.phtml) and putative ribosome binding sites (RBSs) were reported in a previous study (Lin et al., 2017).

The strains used for shake flask experiments were grown aerobically at 200 rpm and 37°C in fresh 25-ml LB medium supplemented with 20 g/L glucose, 0.6 g/L MgSO₄ 7H₂O, 0.07 g/L CaCl₂ 2H₂O, 0.04 g/L KH₂PO₄, 0.04 g/L K₂HPO₄, 0.4 g/L NaHCO₃ and 100 μg/ml ampicillin. In the shake flask experiments, the initial OD₆₀₀ was adjusted to 0.05. Samples were taken and stored every 12 h until 48 h for the subsequent analysis.

Cell density was monitored by measuring the optical density at 600 nm (GENESYS 10S, Thermo Scientific, United States). For the measurement of biomass concentration, 3 ml of the cultured cells were centrifuged for 1 min at 6,791 rcf. The cells were frozen at −20°C and lyophilized. After freeze-drying, the weight of the sample was recorded to calculate the biomass concentration. The PHB produced by the recombinant E. coli cells was quantified by gas chromatography (GC), and the samples were treated as described before (Hsiao et al., 2016). Briefly, an appropriate amount of biomass was transferred to a clean spiral test tube and 1 ml of chloroform, 0.85 ml of methanol, and 0.15 ml of sulfuric acid were added. The tube was incubated in a water bath for 140 min at 80°C and cooled down to room temperature. Then 1 ml of DI H₂O was mixed well by vortexing. After standing and layering, the organic phase was removed and filtered through a 0.2-μm PVDF filter and then analyzed by GC. The temperatures of the injector and detector were 230 and 275°C, respectively. The temperature of the column was set at 100°C and increased to 200°C at a rate of 10°C/min and maintained at 200°C for 2 min. The nitrogen was used as the carrier gas at 3 ml/min. The split mode with the split ratio of 1:10 was used (Chen S.-K. et al., 2013).

The metabolites were determined using a Thermo Scientific™ Dionex™ Ulitmate 3000 LC system equipped with an ORH-801 column. Sulfuric acid (5 mM) was used as the mobile phase, and the flow rate was maintained at 0.6 ml/min. Samples for quantification were collected from the culture media after the removal of the cells by centrifugation for 1 min at 6,791 rcf. The supernatant was passed through a 0.2-μm PVDF filter, and the sample injection (10 μL) was performed using an auto-sampler.

The bacterial culture solutions were centrifuged at 4°C and 9,072 × g for 10 min and resuspended in lysis buffer (50 mM Tris, 0.1 M NaCl, 1 mM EDTA, pH 7.8) to obtain a final OD of 20. The concentrated bacterial solutions were subjected to ultrasonication (10 s break and 5 s intermittent, total 60 cycle) (QSonica Q700, United States). The cell lysates were then diluted with 4x Laemmli sample buffer (Bio-Rad Laboratories, Inc., United States) and heated in a dry batch for 10 min. Samples (10 µL) were loaded into the wells for SDS-PAGE analysis (12% acrylamide).

To confirm in vivo PHB production, the recombinant E. coli BL21 (DE3)/pPHB01 was grown on LB plates containing 10 g/L glucose and 0.8 mg/ml nile red. After 24 h of cultivation at 37°C, the plates were exposed to monochromatic light at 470 nm for observation.

The recombinant plasmids pVEC01, pVEC02, and pVEC03 were tested for plasmid stability in different E. coli hosts. Compared to the parental plasmid pUC19, pVEC02 showed good stability in E. coli DH5α E. coli BL21 (DE3) and with time required to reach 50% ampicillin-resistant cells of 26 and 336 h, respectively (while pUC19 had times of 25 and 54 h, respectively, Figures 2A,B). The hok/sok system increased the segregational stability in BL21 (DE3) by 6 fold, but not in E. coli K strain. In fact, pVEC01, pVEC02, and pVEC03 were not stable in E. coli K strains DH5α. pVEC01 was slightly more stable than pUC19 in BL21 (DE3) cells and pVEC03 showed instability. Because of the stability of pVEC02 in this test, it was chosen as the new vector for the subsequent induction of PHB production.

We tested PHB production by conducting shake flask experiments at 37°C under the control of the native promoter of phaCAB operon from C. manganoxidans. Figure 3A shows that E. coli BL21 (DE3)/pPHB01 consumed 18 DE1 and 13 a1 g/L glucose with and without ampicillin, respectively. In the presence of ampicillin, E. coli BL21 (DE3) harboring pPHB01 was able to effectively produce PHB and achieved a PHB concentration of 3.1 DE0.6 g/L (Figure 3B), yield of 0.16 , 0.03 g/g-glucose, and content of 36 /g4% (Figure 3D). In contrast, in the absence of ampicillin, E. coli BL21 (DE3) harboring pPHB01 produced PHB with a decreased concentration of 0.5 w0.1 g/L, as shown in Figures 3B–D. The effective PHB production in E. coli BL21 (DE3)/pPHB01 justifies the high glucose consumption in the presence of ampicillin.

The capability of E. coli BL21 (DE3)/pPHB01-1 to produce PHB without the inducer and antibiotics was confirmed by the Nile red assay, as shown in Figure 4.

The putative promoter of the phaCAB operon was predicted by BPROM, and the sequences of the putative promoter, -35 box and -10 box sequences were TTC​GAT​TTC​GCA​AGG​CGG​CCG​GTG​TAA​AAA, TTCGAT, and GTGTAAAAA, respectively (Figure 5). Furthermore, RBSs of the phaCAB operon were labeled according to a previous study (Figure 5) (Lin et al., 2017).

While pVEC02 (hok/sok) was shown to be a stable vector in E. coli without ampicillin supplementation, the low PHB production by E. coli BL21 (DE3)/pPHB01 in the absence of ampicillin indicates the instability of pPHB01 in E. coli BL21 (DE3). It is speculated that the insertion of the phaCAB operon in pVEC02 disrupted the balance of hok/sok transcripts. In other words, the promoter of the phaCAB operon may carry an additional transcript of hok and make pPHB01 toxic to E. coli (Figure 6). To verify this speculation, an efficient E. coli ρ factor-independent terminator, thrLABC (ECK120033737) (Chen Y.-J. et al., 2013), was inserted between the phaCAB operon and the hok/sok locus to isolate the two genetic units (Figure 6). The newly constructed plasmid was named pPHB01-1.

E. coli BL21 (DE3) harboring the newly constructed plasmid pPHB01-1 in was tested for the PHB production with and without ampicillin supplementation (Figure 7). In the presence of ampicillin, E. coli BL21 (DE3)/pPHB01-1 was able to produce PHB effectively. All the fed glucose of 13 DE3 g/L was completely consumed in 48 h (Figure 7A), and the PHB concentration, yield, and content were 3.1 , 0.3 g/L, 0.26 /L0.08 g/g-glucose, and 45 /g4%, respectively (Figures 7B–D). In the absence of ampicillin supplementation, E. coli BL21 (DE3)/pPHB01-1 was still able to provide a comparable PHB performance, where the PHB concentration, yield, and content was 3.0 PH0.2 g/L, 0.26 /L0.07 g/g-glucose, and 44 /g3%, respectively (Figures 7B–D).

To characterize the long-term stability of PHB production in E. coli BL21 (DE3)/pPHB01-1, E. coli BL21 (DE3)/pPHB01-1 was first cultivated in the liquid culture without the ampicillin (the medium composition was described in 2.5) and transferred to a fresh medium every 12 h. During the transfer, the grown-culture solution was spread on the agar plate without the ampicillin but 0.8 mg/ml nile red. After the formation of CFU, the total CFU as well as the fluorescent CFU were counted to quantify the percentage of CFU that can produce PHB. It can be seen in Figure 8A that the OD₆₀₀, which is positively correlated to the PHB production, can be maintained in the range of 15 and 18 in the first 84 h and gradually dropped to 5.1 a0.1 at 132 h. Figure 8B shows that the E. coli BL21 (DE3)/pPHB01-1 maintained its ability to produce PHB in the first 96 h and less than 20% of cell population can produce PHB afterwards.

SDS-PAGE was used to confirm the expression of the phaCAB operon derived from C. manganoxidans in E. coli. It can be clearly seen in Figure 9 (lanes 7–12) that three overexpressed signals were found in both E. coli BL21 (DE3)/pPHB01 and E. coli BL21 (DE3)/pPHB01-1 samples, which correspond to PhaC (61 kDa), PhaA (43 kDa), and PhaB (27 kDa). PhaA and PhaB in E. coli BL21 (DE3)/pPHB01 and E. coli BL21 (DE3)/pPHB01-1 were mainly found in the soluble fractions (lanes 8 and 11 of Figure 9), while PhaC was mainly found in the insoluble fraction (lanes 9 and 12 of Figure 9).