Strains and plasmids used in this study are listed in Table 1. The AgPS gene was amplified from pAgGPPS-(GSG)₂-AgPS by PCR with specific primers AgPS_F: 5′-CAG​CAT​ATG​CGT​CGT​G-GTA​AAT​CTA​TCA​C-3′ and AgPS_R: 5′-GCG​GGA​TCC​TTA-CAG​CGG​AAC​AGA​TTC​CAG-3′. The PCR product was purified and ligated into the cloning vector pBZ (TransGen Biotech, China) to perform sequencing. The AgPS gene with the correct sequence was inserted into pAQ1EX-PcpcBA to generate pAQ1EX-AgPS plasmid via NdeI and BamHI (Thermo Scientific) digestion. The plasmid pAQ1EX-AgPS was aiming at integrating the functional genes into the endogenous high-copy plasmid pAQ1 through homologous recombination. A modified protocol following that described previously (Stevens and Porter, 1980; Frigaard et al., 2004) was used to transform pAQ1EX-AgPS into Synechococcus. To select the transformant strain, 100 μg ml⁻¹ streptomycin was applied, and the successful transgene incorporation was confirmed by colony PCR and sequencing using the primers PAQIN_F: 5′-GGA​ATT​GTG​CGT​GTG​GTT​TC-3′ and PAQIN_R: 5′-CTA​ACG​ATC​AGC​GCG​AAA​AG-3′.

Synechococcus wild-type (WT) and AgPS transformant (7002-PS) strains were grown in liquid medium A+ containing the following: 18 g L⁻¹ NaCl, 5 g L⁻¹ MgSO₄ 7H₂O, 1 g L⁻¹ NaNO₃, 0.6 g L⁻¹ KCl, 0.050 g L⁻¹ KH₂PO₄, 0.270 g L⁻¹ CaCl₂, 0.020 g L⁻¹ Na₂CO₃, 0.030 g L⁻¹ Na₂EDTA·2H₂O, 3.890 mg L⁻¹ FeCl₃·6H₂O, 1 g L⁻¹ Tris HCl (pH 8.2), 2.860 mg L⁻¹ H₃BO₃, 1.810 mg L⁻¹ MnCl₂·4H₂O, 0.220 mg L⁻¹ ZnSO₄·7H₂O, 0.390 mg L⁻¹ Na₂MoO₄·2H₂O, 0.100 mg L⁻¹ CuSO₄·5H₂O, 0.050 mg L⁻¹ Co(NO₃)₂·6H₂O, and 4 μg L⁻¹ vitamin B12 (Stevens and Porter, 1980; Ludwig and Bryant, 2011). The 7002-PS strain containing Sm^R cassette was selected on solid A+ medium, adding 1.2% (w/v) Bacto Agar (BD) and streptomycin with the required concentration. Seed cultures were performed in 50-ml Erlenmeyer flasks using an orbital shaker at 100 r min⁻¹. The flasks contained 30 ml of liquid A+ medium in the presence of antibiotic. All Synechococcus strains were grown at 30°C under continuous illumination of 100 μmol photons m⁻² s⁻¹, and optical density (OD) was measured to monitor cell growth using a spectrophotometer (Varian) at 730 nm.

Synechococcus cultures used for western blotting were inoculated into 50 ml fresh A+ medium with an initial OD₇₃₀ = 0.1 and grown for 24 h under continuous aeration supplement with 1% (v/v) CO₂. Cells were harvested by centrifugation and disrupted with a motor-driven tissue grinder (Sangon Biotech). Equal amounts of total proteins in crude cell lysates were subjected to western blotting analysis. AgPS was identified with anti-His mouse monoclonal antibody (TransGen Biotech, China) followed by goat anti-mouse IgG (H + L) (TransGen Biotech, China) and visualized using a DAB horseradish peroxidase color development kit (Beyotime, China).

The Water-PAM (Walz) was used to measure the variable chlorophyll fluorescence with a Water-S stirring device capable of keeping the samples homogenous and preventing sedimentation of the cells. Light-emitting diodes supplied red measuring light (spectral peak at 650 nm), actinic light, and saturation pulses (spectral peak at 660 nm). After 20 min of dark adaption, a 2-ml sample was used to analyze fluorescence parameters Fv/Fm by measuring Fo and Fm. Light treatment strategy and stirrer operation in this study were according to the previous reports (Campbell et al., 1998; Cosgrove and Borowitzka, 2006).

Synechococcus cultures used for pinene production were inoculated into 60 ml fresh A+ medium without antibiotic with an initial OD₇₃₀ = 1.0. The cultures were grown for 72 h under continuous aeration supplement with 1% (v/v) CO₂, and a 6-ml dodecane (Aladdin) overlay was applied at the beginning to trap the pinene excreted from the cells. Samples of 500 μl dodecane overlay were harvested at the end of the production period. The procedure of sample pretreatment was following a modified method as described previously (Sarria et al., 2014), adding (R)-(+)-limonene (Aladdin) as an internal standard. These samples were analyzed on a gas chromatography–mass spectrometer (GC-MS) (Agilent 7890A with Agilent 5975C MS detector) by a standard curve of (−)-α- and β-pinene (Aladdin). The GC-MS was equipped with a DB-5MS column (30 m × 0.25 mm × 0.25 µm) to separate hydrophobic molecules. The analysis conditions were as follows: He (1 ml min⁻¹) as a carrier gas, split ratio of 5:1; an injector temperature of 300°C; and an oven program of 50°C for 5 min, ramp at 10°C min⁻¹ to 150°C, ramp at 30°C min⁻¹ to 280°C and held for 5 min.

Synechococcus is capable of utilizing the endogenous MEP pathway to synthesize GPP. In order to build Synechococcus a platform for pinene production, we need to integrate an exogenous pinene synthase into the genome of Synechococcus to convert GPP into α- or β-pinene with high fidelity. Although both pinene isomers can be found in turpentine (Behr and Johnen, 2009), the β-pinene isomer is considered preferable for higher economic value (Sarria et al., 2014) and dimerization efficiency (Walls and Rios-Solis, 2020). Thus, the AgPS gene was selected to produce pinene with more proportion of β-pinene (Bohlmann et al., 1997) and had shown the highest activity for pinene synthesis in E. coli among the high-fidelity pinene synthases (Sarria et al., 2014).

The nucleotide sequence of the AgPS gene from Sarria et al. (2014) had been modified and expressed according to the codon usage preference of E. coli. A comparison of the modified AgPS gene sequence with the codon usage of Synechococcus was predicted with the Graphical Codon Usage Analyser (Fuhrmann et al., 2004), and the results showed that the relative adaptiveness values are all higher than 30% (Supplementary Material S1). We therefore directly amplified the modified AgPS gene from plasmid pAgGPPS-(GSG)₂-AgPS and therewith integrated it into the pAQ1EX vector between the NdeI and BamHI sites. The insertion of AgPS gene into the neutral region of pAQ1 between two open-reading frames encoding hypothetical proteins (Figure 1A) was confirmed by colony PCR and sequencing in numerous transformants of the 7002-PS strain (Figure 1B). Interestingly, although the 7002-PS strain was isolated using the streak plate method three times, as the colony PCR results showed, the neutral region of pAQ1 between two open-reading frames had still not been entirely replaced by the integrated sequence. We supposed that the region replaced by the transgenes in pAQ1EX-PS was essential for the growth of Synechococcus (Xu et al., 2011), and the complete integration of the exogenous genes in pAQ1EX was hard to realize. The potential different amount of the transgenic pAQ1 might eventually lead to the differences in pinene production performance among the independent transformants of the 7002-PS strain. The expression of AgPS was confirmed by western blotting (Figure 1C).

Dodecane overlay has been turned out as an effective method for terpenes harvesting (Davies et al., 2014; Sarria et al., 2014; Niu et al., 2018; Dienst et al., 2020). This article utilized 1% (v/v) CO₂ bubbles to supply inorganic carbon for pinene production in the 7002-PS strain. Photoautotrophic cultivation of cyanobacteria with sodium bicarbonate as an inorganic carbon source and using orbital shakers or magnetic stirrers seem adequate for terpenes production (Silva et al., 2016). However, CO₂ used as the inorganic carbon source has environmentally friendly advantages in integrating atmospheric CO₂ into biomass. Although filling the upper space of the medium with CO₂ could maintain continuous production of terpene (Davies et al., 2014), we need to develop deeper insight into the production performance of engineered strains grown with CO₂ bubble aeration.

To investigate whether dodecane overlay limits the photoautotrophic growth of cyanobacteria ventilated with CO₂, we inspected the photoautotrophic growth of the WT strain overlaid with different dodecane concentrations over a 72-h time course at various aeration rates. Cell growth rates were measured by OD_730, and the photosynthetic performance was evaluated by the chlorophyll fluorescence parameter Fv/Fm. The chlorophyll fluorescence parameter Fv/Fm and OD₇₃₀ of the WT cells were compared when growing with 0%, 5%, 10%, and 15% (v/v) dodecane overlay. As the organic phase could not maintain stability with an aeration rate higher than 0.12 vvm (volume of gas per volume of liquid per minute) in this work, the aeration rates were initially set to 0.02, 0.03, and 0.12 vvm. Apparently, the limited CO₂ provision caused by the lower aeration rates (0.02 and 0.03 vvm) had an adverse effect on the photoautotrophic growth of Synechococcus. Meanwhile, the OD₇₃₀ of the WT cells cultivated with aeration rate at 0.12 vvm exhibited obvious variation (Figure 2A). The chlorophyll fluorescence parameter Fv/Fm of the cells without dodecane overlay was higher than the that of the cells overlaid with dodecane after 2 days, and the cells aerating at 0.12 vvm presented the greatest gap, which was supposed to aggravate the difference of OD₇₃₀ (Figure 2B). However, neither the Fv/Fm nor OD₇₃₀ of the WT cells showed apparent difference whenever the aeration rate was 0.02 or 0.03 vvm. The results indicated that the dodecane overlay led to a slight restriction to the photosynthetic growth of Synechococcus cultivated with CO₂ bubble aeration at a high rate (0.12 vvm), and the three concentrations (5%, 10%, and 15%) of the dodecane overlay had no significant difference in the limitation effect. However, the dramatic variance of Fv/Fm between the WT cells overlaid with and without dodecane demonstrated that cultivation with dodecane overlay at a high aeration rate enables to impair the photosystem II efficiency of Synechococcus, which suggests that dodecane overlay is not the best choice for long-term product harvest in the condition of high gas aeration rate for Synechococcus.

The different concentrations of dodecane overlay resulted in the liquid level differences of the organic phase in the tube photobioreactor (Figure 2C). Considering the disturbance of the interface between two phases caused by bubble breakage, the moderate thickness of the organic phase is supposed to be of great significance for reducing pinene loss and the inhibition effect of dodecane overlay on cell growth. Besides, a slight loss of dodecane volume caused by gas aeration was observed on the tube wall (data not shown), which demanded an appropriate volume to reduce the calculation error caused by the volume change. Thus, we determined that a 10% (v/v) dodecane overlay was suitable for pinene harvesting from the culture medium in our tube photobioreactor.

To investigate the influence of genetic modification on the photoautotrophic growth of Synechococcus, we compared the photoautotrophic growth rates measured by daily OD₇₃₀ between the WT strain and three transformants of the 7002-PS strain grown in tube photobioreactor for 6 days (Figure 3A). No significant variation was observed among the examined strains under culture conditions of 30°C, 100 μmol photons m⁻² s⁻¹, and 1% (v/v) CO₂ aeration. Obviously, metabolic stress caused by the overexpression of transgene and the engineered flux of GPP towards pinene production has no adverse effect on the growth of cyanobacteria grown under the conditions described above.

As the colony PCR results confirmed that the integrated sequence had not entirely replaced the neutral region of pAQ1 between two open-reading frames (Figure 1B), numerous transformants of the 7002-PS strain were supposed to exhibit discrepant production performance. In this work, aeration rates at 0.12 and 0.03 vvm were applied to realize a 72-h and a 6-day period of pinene production, respectively. Two independent transformants of the 7002-PS strain were selected for 72-h pinene production with 10% (v/v) dodecane and 1% (v/v) CO₂ aeration at 0.12 vvm, resulting in the pinene titer at 1.362 ± 0.148 mg L⁻¹ and 1.525 ± 0.l45 mg L⁻¹ (pinene productivity at 0.110 ± 0.018 mg L⁻¹ OD₇₃₀ ⁻¹ and 0.114 ± 0.013 mg L⁻¹ OD₇₃₀ ⁻¹), respectively (Figure 3B). The transformant with higher pinene-producing performance, named 7002-PS7, showed relatively stable pinene productivity of 0.246 ± 0.05 mg L⁻¹ day⁻¹ for a 6-day period of pinene production with 10% (v/v) dodecane and 1% (v/v) CO₂ aeration at 0.03 vvm (Figure 3C). Besides, the α/β isomer ratios exhibited by the 7002-PS strain were also investigated (Figure 3B). Interestingly, AgPS expressed in Synechococcus resulted in a ∼33:67 mixture of α- to β-pinene, in which a significant increase in β-pinene production was realized compared to that produced in E. coli (Sarria et al., 2014). We hypothesized that the cellular metabolic environment of Synechococcus is suitable for forming the β-pinene isomer.