Conversion of carbon dioxide into valencene and other sesquiterpenes with metabolic engineered Synechocystis sp. PCC 6803 cell factories

Valencene is a natural sesquiterpene with desirable bioactivity and aroma, making it a valuable ingredient in the food and cosmetics industries. Traditionally, valencene was extracted from the citrus fruits, and its applications were restricted by the low concentrations in natural sources and high costs for extraction. Photosynthetic biomanufacturing represents a promising route for efficient and stable production of valencene, while cyanobacteria have been considered one of the most promising platforms regarding biotechnological routes for the direct conversion of CO2. In this work, we engineered Synechocystis sp. PCC 6803 to synthesize valencene. By introducing a heterologous valencene synthase and modifying the native MEP pathway, we obtained an efficient cyanobacterial cell factory that produced 154 mg/L valencene during a semi‐continual cultivation, with an average productivity of 4.3 mg/L/day, and the cell factory exhibited robust growth and production in non‐sterilized conditions. We also achieved the production of other sesquiterpenes including bisabolene, amorpha‐4,11‐diene, farnesene, and nerolidol by engineered cyanobacteria with enhanced MEP pathway flux, showing promising potentials as a universal chassis.


| INTRODUCTION
Valencene is a natural sesquiterpene, synthesized and accumulated as a secondary metabolite in fruits of various citrus species, such as Cyperus rotundus (Dantas et al., 2022). Due to its unique odor and bioactivity, valencene has been extensively used in food, beverages, perfumes, and cosmetics industries for flavoring and fragrance purposes (Frohwitter et al., 2014;Ouyang et al., 2019;Sharon-Asa et al., 2003). Besides, valencene could be used as a substrate for the production of other terpenes, for example, nootkatone, which has promising application prospects in medicine (Meng et al., 2020). At present, industrial production of valencene is mainly achieved by distilling citrus oil. However, the contents of valencene in natural plant resources are quite low, requiring over 2.5 million kilograms of oranges to produce a single kilogram of valencene (<0.001 ‰), which led to the complex extraction processes and high price of commercial products (Ouyang et al., 2019;Wang et al., 2015;Zhang et al., 2021).
From the source analysis, valencene extracted from plant biomass is synthesized with carbon dioxide fixed by photosynthesis, possessing the advantages of low raw materials cost over the newly developed microbial fermentation routes based on artificial heterotrophic cell factories (Chen et al., 2019;Li et al., 2021;Meng et al., 2020;Ouyang et al., 2019;Ye et al., 2022). However, the complex life histories, highly differentiated organizations and structures, and long cultivation terms of higher plants determined that the conversion ratio from carbon dioxide to valencene as the final product could hardly be significantly improved, which fundamentally limits the optimization potential of this plant biomass extraction route. Engineering other photoautotrophic platforms with simpler structures and life histories, for example, microalgae, for more direct and efficient conversion of carbon dioxide, represents a more promising route for photosynthetic production of valencene, as well as other sesquiterpenes (Einhaus et al., 2021;Lauersen et al., 2016;Lin & Pakrasi, 2019). Comparing with higher plants, microalgae possess more efficient photosynthesis, more rapid growth, and smaller sizes, and they are also more suitable to be engineered for direct conversion of carbon dioxide into desired products (Fernandez et al., 2021;Rahul et al., 2021).
Cyanobacteria are a group of prokaryotic microalgae, widely distributed in diverse ecosystems on Earth, and they have been traditionally selected as model systems for deciphering the photosynthesis mechanisms. In recent years, cyanobacteria are also supposed to be promising microbial photosynthetic platforms, due to the characteristics of simple structure, efficient photosynthesis, and mature genetic manipulation toolboxes (Gao, Sun, et al., 2016;Luan & Lu, 2018;Santos-Merino et al., 2019). Carrying a naturally existing 2-methyl-erythritol-4-phosphate (MEP) pathway, synthesis of diverse sesquiterpenes could be achieved in cyanobacteria by introduction of heterologous terpenesynthase genes, utilizing the farnesyl pyrophosphate (FPP) as the general substrate (Yu et al., 2019). In multiple chassis, including Synechocystis sp. PCC 6803, Synechococcus elongatus PCC 7942, Synechococcus sp. PCC 7002, and Anabaena sp. PCC 7120, representative sesquiterpene metabolites, including farnesene (Halfmann, Gu, Gibbons, & Zhou, 2014), caryophyllene (Reinsvold et al., 2011), and bisabolene (Dienst et al., 2020), have been successfully synthesized. Recently, the production of valencene in cyanobacteria has also been achieved. Matsudaira et al. introduced the valencene synthase from Callitropsis nootkatensis and the farnesyl diphosphate synthase (IspA) from Escherichia coli onto chromosome of Synechocystis sp. PCC 6803, and the obtained strain could produce 9.6 mg/L valencene in 166 h in shaking flasks (Matsudaira et al., 2020). In a more recent report, Dietch et al. expressed the same synthase gene in Synechocystis sp. PCC 6803 and optimized the carbon flux by reducing competitive flows toward other terpenes and carotenoids, and the valencene titer was increased to 17.6 mg/L (4 mL broth in six-well plates) (Dietsch et al., 2021). However, for industrial application, the valencene titers of the engineered cyanobacteria strains should be further improved and the performances of strains (such as growth, genetic stability) in long-term cultivation require more systematic evaluations.
In this work, we aimed to engineer more efficient and stable cyanobacterial photosynthetic production systems for valencene. By introduction of heterologous valencene synthase and optimization of the MEP pathway and in Synechocystis sp. PCC 6803 (hereafter termed as PCC 6803 for short), valencene productivity of the cell factories reached up to 7.58 mg/L/day in column photobioreactors. Through adoption of semi-continuous culture strategies, 154.81 mg/L of valencene was accumulated by the engineered strain in 36 days. Under non-sterilized conditions, valencene production could still be well maintained, showing the significant robustness of the cell factories. In addition, utilizing the strain with enhanced MEP pathway flux as a chassis, we successfully developed cell factories for photosynthetic production of four other sesquiterpenes (α-bisabolene, amorpha-4,11-diene, αfarnesene, transnerolidol), indicating the significance of the chassis engineering for photosynthetic production of terpenes.
The plasmids were transformed into PCC 6803 cells at mid-log phase. After washed with fresh BG11 medium, the cells were resuspended and then incubated overnight with 4 μg linear DNA at 30°C and 100 μmol photons/ m 2 /s. Cells were then cultivated on BG11 plates supplemented with corresponding antibiotics for selection, and transformants would be identified by PCR. In each batch of transformation, 21 transformants would be screened, and the transformation success rate is usually more than 50%. The transformants with fully segregated chromosomes would be used for subsequent evaluation. The constructed strains were listed in Table S2.

| Cultivation conditions
For evaluations of cell growth and sesquiterpene production, PCC 6803 and the derived strains were cultivated with 1X BG11 in shake flasks, and 1X/2X BG11 in column photobioreactor (with a diameter of 3 cm). Cells cultivated to logarithmic growth phase would be resuspended with 100 mL fresh culture medium to the initial OD 730 of 1.0, and then be cultivated at 30°C under constant white-light illumination of 150 μmol photons/m 2 /s, aerated with 3% (vol/vol) CO 2 -supplemented air. For semi-continuous cultivation, 50% volume of cells are removed periodically (samples were taken every 3 days from day 6 to day 12, and taken every 4 days from day 12), and equal volume of 2X BG11 culture medium would be supplemented, to maintain the cultivation with optimal growth rates and light availability. Cultures broths were overlaid with 10% dodecane to cover the liquid surface to avoid loss of the volatile product valencene according to previous reports (Davies et al., 2014). Dodecane would be replaced at each sampling point during the cultivation process to guarantee the recovery efficiency.

| Sesquiterpene content determination
In the photobioreactors, sesquiterpenes were accumulated in the dodecane phase. Thus, 1 mL of dodecane covered on culture broth would be assayed using gas chromatography-mass spectrometry (GC-MS). GC-MS assays were carried out by Agilent 7890, equipped with Agilent 5977 mass detector and HP-INNOWax column (30 m × 0.32 mm × 0.25 μm). Ultra-pure helium was used as carrier gas at a constant flow rate of 3 mL/min. The temperature of the column box was initially kept at 60°C for 5 min, increased to 260°C at a rate of 10°C/ min, and then kept at 260°C for 10 min. The concentration of sesquiterpenes was determined via external calibration using corresponding standard substances. Using the SIM mode, the ion peak with a specific charge-mass ratio of M/Z was identified between the sample and the standard, and a standard curve was made to quantify the sample concentration. Figure S1 is the comparison of retention time and mass spectrum of different samples and standards, with retention time (min) on X-axis and peak area of chromatographic response value on Y-axis. Among them, βfarnesene standard was used to quantify αfarnesene samples according to reported articles (Pattharaprachayakul et al., 2019). The ion peaks of bisabolene, amorpha-4,11-diene, valencene, and nerolidol with specific charge-mass ratios of 93, 119, 161, and 69, respectively. Since 10%(v/v) dodecane is used to concentrate and collect the product of 100 mL algae liquid, thus dividing the measured titers in the dodecane overlay by 10 reflected the actual production yields. The total valencene accumulation in the cultivation process is calculated by adding the yield of all sampling points.

| Expression of heterologous valencene synthases in PCC 6803 facilitated photosynthetic production of valencene
The MEP pathway exists in plant plastids, cyanobacteria, and eukaryotic microalgae, in which glyceraldehyde-3phosphate and pyruvate are stepwisely converted into isopentenyl diphosphate (IPP), dimethylallyl diphosphate (DMAPP), geranyl diphosphate (GPP), and farnesyl diphosphate (FPP), as the common precursors of all subsequent terpenoids . Sesquiterpenes would be synthesized by sesquiterpene synthases with FPP as the substrate (Figure 1a) (Zhu et al., 2014). Although some cyanobacteria sourced sesquiterpene synthases have recently been identified in some filamentous species (Nostoc punctiforme PCC 73102 and Nostoc sp. PCC 7120), a majority of cyanobacteria, such as PCC 6803, a representative cyanobacteria strain for constructing photosynthetic cell factory, cannot naturally produce sesquiterpenes (Agger et al., 2008;Choi et al., 2016). Thus, it is usually necessary to introduce heterologous sesquiterpene synthase into cyanobacteria strains for cell factories construction. In this work, we selected the valencene synthases from C. nootkatensis (Matsudaira et al., 2020) and C. sinensis Matsudaira et al., 2020), put the codon-optimized genes under the control of a native strong promoter P cpcB560 , and integrated the artificial cassette on a neutral site of PCC 6803 chromosome, finally obtaining the strain JS142 and JS136, respectively (Figure 1b).
The cell growth and valencene yield of JS136 and JS142 were firstly evaluated in 1X BG11 in shake flasks, and significant production of valencene could be detected in both the two engineered PCC 6803 strains, with the titers of 7.53 and 0.16 mg/L in JS142 and JS136, respectively ( Figure S2). After preliminary experiments, the valenceneproducing strain could maintain a stable growth and production condition in 2X BG11 medium ( Figure S3). Thus, in the subsequent experiments, all strains were evaluated in 2X BG11 medium culture mode to optimize and prolong the valencene production process. JS136 and JS142 were then further cultivated in column photobioreactors aerated with 3% (vol/vol) CO 2 , and the valencene tilters were improved to 12.04 and 0.71 mg/L in JS142 and JS136, respectively (Figure 1c,d). The above results indicated that unlike the previous report (Matsudaira et al., 2020), valencene production could be achieved in Synechocystis through just overexpressing the valencene synthases independent of other genetic modifications. Similar to the evaluation results of the two sources of valencene synthase by Matsudaira et al. (Matsudaira et al., 2020), the C. nootkatensis sourced enzyme could work more effectively than synthase from C. sinensis in PCC 6803. In the following work, JS142 overexpressing the valencene synthase from C. nootkatensis (CnValS) was utilized as a chassis for further improvement.

| MEP pathway optimization enhanced valencene production in PCC 6803
Since most cyanobacteria do not have sesquiterpene synthesis ability, it is necessary to introduce exogenous synthase into wild PCC 6803 to realize the accumulation of valencene, thus, synthase is the last key rate-limiting enzyme in the sesquiterpene synthesis pathway. However, when the sesquiterpene synthase is expressed, how to strengthen the carbon flow to the target product synthesis is extremely important. The direct substrate for sesquiterpenes biosynthesis is FPP, which is an important intermediate metabolite in MEP, and the optimization/ enhancement of MEP pathway has proved crucial for improving the performance of sesquiterpene synthesizing cyanobacterial cell factories (Luo et al., 2020;Zhu et al., 2014). Thus, we further performed stepwise optimizations of the MEP pathway in JS142 strain to improve the productivity of valencene (Figure 2a). Firstly, the E. coli sourced farnesyl pyrophosphate synthase gene (ispA) was introduced into the slr9394 neutral site on chromosome of JS142 for directly enhancing the conversion of IPP and DMAPP into GPP, and then to FPP (Pattharaprachayakul et al., 2019). Comparing with the native GppS enzyme, the introduction of IspA led to enhanced accumulation of FPP rather than GGPP in the MEP pathway (Blanc-Garin, Chenebault, et al., 2022;. In addition, the S. cerevisiae sourced isopentenyl diphosphate δisomerase (IPP isomerase, Idi) was also introduced in chromosome of JS142, aiming to optimize the DMAPP:IPP ratio for increased FPP synthesis Luo et al., 2020). As shown in Figure 2b,c, the JS142-derived strains (JS147 and JS150) overexpressing ispA or idi showed 10% (13.2 mg/L) and 20% (14.3 mg/L) improved valencene titers, respectively, than the initial strain (11.94 mg/L in 9 days). However, when both heterologous genes were overexpressed in strain JS148, no additional improvements (14.1 mg/L) on valencene production could be observed, indicating that the bottleneck holding FPP synthesis in MEP has been removed from the conversion process from IPP/DMAPP to FPP. Thus, we aimed to enlarge the MEP pathway flux for more FPP supply. The 1-deoxy-d-xylulose-5-phosphate synthase (Dxs) has been considered to be the rate-limiting enzymes controlling the flux of the MEP pathway (Englund et al., 2018), so the E. coli-sourced dxs gene was overexpressed in the strain JS148 under the control of another native strong promoter P rbcL6803 . This strategy showed significant effects, and the obtained strain JS133 (PCC 6803-CnValS-IspA-Idi-Dxs) could synthesize 39.6 mg/L valencene in 9 days, meaning over threefold higher than that of the initial strain JS142 (PCC 6803-CnValS). During the initial 3 days of cultivation, valencene productivity of JS133 reached up to 7.58 mg/L/day, indicating that the overexpression of dxs dramatically enlarged the carbon flux in MEP pathway as well as the valencene production as the final product in the engineered strain. In contrast, when Dxr, the subsequent enzyme following Dxs in MEP pathway, was also overexpressed in JS133, the valencene titer of the strain was not further improved ( Figure S4), indicating that the factor holding control over the MEP pathway flux in PCC 6803 indeed relies on the initial enzymatic reaction catalyzed by Dxs.

| Cultivation optimization prolonged the valencene production in PCC 6803
As shown in Figure 2b,c, cell growth and valencene production of JS133 could only be maintained for 3-6 days, and in the initial 3 days when cell densities were increasing rapidly, valencene was produced with a relatively higher productivity. Thus, we designed to optimize the cultivation strategy by maintaining the cell growth status of JS133 in column photobioreactors so as to keep a high valencene productivity (Fernandes et al., 2015). Previously, a semi-continuous cultivation mode, meaning routinely replacing the culture broth with fresh culture medium, has been utilized to prolong the growth of cyanobacterial factories for high production of sucrose and trehalose (Qiao et al., 2020a;Song et al., 2016). Thus, we designed and adopted a similar strategy for cultivating the valencene synthesizing JS133 strain. As shown in Figure 3a, the semi-continuous cultivation process was adopted and 50% of the culture broth would be removed and replaced with fresh 2X BG11 culture medium starting from Day 6 of the cultivation. As shown in Figure 3b, the semi-continuous cultivation strategy restrained the JS133 cells in a rapidly growing status in a 36-day long-term cultivation process. During this process, 154 mg/L valencene was produced by the JS133 cells, with an average productivity of 4.3 mg/L/day (Figure 3c). In contrast, densities F I G U R E 2 Engineering the MEP pathway to improve photosynthetic production of valencene in PCC 6803. (a) The schematic diagram for construction of the JS142 derived strains with engineered MEP pathway. (b, c) Cell growth (b) and valencene production of the JS142 derived strains. Cells of the valencene synthesizing strains were resuspended with 100 mL 2X BG11 culture medium at the initial OD 730 of 1.0, and then be cultivated at 30°C under constant white-light illumination of 150 μmol photons/m 2 /s, aerated with 3% (vol/vol) CO 2supplemented air.

F I G U R E 3
The semi-continuous cultivation of JS133 for prolonged valencene production. (a) The schematic diagram for semicontinuous cultivation of JS133 by 50% culture medium replacement. (b, c) Comparisons of cell growth (b) and valencene production (c) in semi-continuous mode and single batch mode of JS133 cultivation. Samples were taken every 3 days from day 6 to day 12, and taken every 4 days from day 12. of the cells cultivated in the single batch mode declined since Day 9, and thereafter, the valencene accumulation was not further increased.

| JS133 maintained stable cell growth and valencene production in non-sterilized conditions
As for the cyanobacterial and microalgal cell factories obtained through metabolic engineering, an essential bottleneck restricting their potentials for industrial application is the weak resistance toward environmental disturbance and low genetic stability for maintaining capacities for producing non-natural metabolites (Luan & Lu, 2018). Obviously, it is impossible to use distilled water for future industrialization of valencene production in cyanobacteria. Furthermore, it has been proved that using TAP water for cultivation of ethanol-producing strain has no negative effect on growth and ethanol production of PCC 6803 mutants (Gao et al., 2012). Thus, it is important to know whether the yield of valencene production can be affected by using tap water. We further evaluated the stability and robustness of valencene producing capacity of JS133 in mimicked "open" environments. Non-sterilized tap water was used for preparing 2X BG11 medium, and no antibiotics were supplemented. As shown in Figure 4a,b, comparing with the standard lab conditions (sterilized ddH 2 O for culture medium preparation and supplemented antibiotics), neither cell growth nor valencene production of JS133 cells was influenced in non-sterilized and no antibiotics culture broth, and 64.0 mg/L of valencene was produced in the 15-day cultivation process. Thus, it could be supposed that the photosynthetic production of valencene in JS133 was not dependent on strict sterilized conditions, and the potential biocontaminants invading risks from tap water and the aerated CO 2 -air could be "tolerated" by the PCC 6803 cell factory. The non-sterilized and antibiotics-independent cultivation would significantly reduce the costs for cultivation and improve the economic feasibility of cyanobacterial valencene production, indicating that the strain JS133 would be suitable for industrial cultivation.

| The valencene synthesizing strain could be engineered as a universal chassis for production of other sesquiterpenes
In this work, we engineered an efficient valencene synthesizing strain using PCC 6803 as a chassis cell, which possessing valencene productivity of as high as 7.58 mg/L/ day, through introduction of a valencene synthase and improvement of the native MEP pathway. As mentioned above, biosynthesis of diverse sesquiterpenes was catalyzed by specific synthases utilizing FPP as the common precursor. Thus, on the basis of PCC 6803, we expressed IspA, Idi, and Dxs genes to construct an engineering strain F I G U R E 4 Robustness evaluations of JS133 cell growth and valencene production. Comparisons of cell growth (a) and valencene production (b) of JS133 in sterilized/non-sterilized conditions with or without antibiotics (for maintaining the genetic stabilities). JS119 with enhanced terpenoid synthesis flux. As shown in Figure 5a, we introduced several plant-sourced sesquiterpene synthases (Choi et al., 2016;Davies et al., 2014;Lee et al., 2017;Li et al., 2021) onto the slr0168-neutral site of JS119 chromosome. Using this approach, we finally obtained the strains JS145, JS138, JS185, JS135 for the production of bisabolene, amorpha-4,11-diene, farnesene, and nerolidol, respectively.
As shown in Figure 5b,c, the overexpression of sesquiterpene synthases caused no significant disturbance on the cell growth of the recombinant strains, while the respective sesquiterpenes were successfully produced as designed. After 6 days cultivation, titers of the farnesene (in JS185), nerolidol (in JS135), bisabolene (in JS145), and amorpha-4,11-diene (in JS138) reached up to 9.99, 8.21, 1.26, and 1.36 mg/L, respectively, and the difference in productivity of diverse sesquiterpenes should be resulted from the difference of catalytic activities of the respective synthases.
As summarized in Table 1, lots of progress has been achieved on engineering cyanobacteria for photosynthetic production of sesquiterpenes, and this work made beneficial supplementations. Notably, photosynthetic production of nerolidol in cyanobacteria was reported for the first time in this work, and the titers of valencene and farnesene reached to similarly high levels as previously reported, confirming the effects of MEP pathway optimization on improving sesquiterpene production in cyanobacteria. As for production of amorpha-4,11-diene, it is interesting that although nearly the same genetic modifications were performed, the productivity achieved in this work with PCC 6803 (0.18 mg/L/day) was much lower than the previously reported level in PCC 7942 (1.98 mg/L/day) (Choi et al., 2016). This discrepancy indicated that the adaptabilities between terpene synthases and the chassis cells would serve as key determinants for the performances of the cyanobacteria cell factories. As for the production of bisabolene, although PCC 6803 was used as the chassis, in the previously published research with similar culture conditions to our experiment (excluding high-density culture), the yield of bisabolene was 0.67 mg/L/day when only bisabolene synthase was F I G U R E 5 Engineering cyanobacteria for other kinds of sesquiterpene production. Construction strategy for the production of other sesquiterpenes (a). Growth and sesquiterpenes productivities in different engineered strains (b, c). The cyanobacterial production of sesquiterpenes in this study (d). AG1, αbisabolene synthase; AMS1, amorpha-4,11-diene synthase; AFS1, αfarnesene synthase; NES, nerolidol synthase. This study overexpressed (i.e., ispA, dxs, and idi were not expressed) (Blanc-Garin, Chenebault, et al., 2022), which was higher than the bisabolene production in our study (0.23 mg/L/ day). Besides, although the yield of strains is much higher under high-density culture , it cannot be directly compared with our study because of the change of culture conditions. Thus, it is worth noting that, we overexpressed the sesquiterpene synthases on the basis of JS119, just in order to prove that this strain can be served as a chassis for engineering cell factories to produce diverse sesquiterpenes, thus, the obtained strains are not necessarily the optimal metabolic transformation for the production of corresponding products.

| DISCUSSION
Sesquiterpenes are an important group of terpenoid metabolites with diverse application prospects in the area of fuels, materials, and medicine, and photosynthetic production of valuable terpenoids using carbon dioxide and solar energy represents a promising direction for the development of cyanobacterial synthetic biology. In the previous reports on the production of valencene in cyanobacteria, Matsudaira et al. introduced the CnValS and IspA into PCC 6803, and the obtained strain produced 9.6 mg/L valencene in 166 h in shaking flasks, however, the yield of valencene is relatively low (1.37 mg/L/day) (Matsudaira et al., 2020). Although Dietch et al. achieved a valencene yield of 8.8 mg/L/ day, strains were cultured in six-well plates with 4 mL of BG11 (Dietsch et al., 2021); from the perspective of industrial application, the performances of strain in long-term cultivation requires more systematic evaluations. In this work, we performed systematic metabolic engineering and cultivation optimization to achieve efficient photosynthetic production of valencene, an important sesquiterpene, with a model chassis Synechocystis sp. PCC 6803. Through introduction of heterologous valencene synthase and modifications of the upstream MEP pathway (overexpression of heterologous IspA, Idi, and Dxs), valencene could be synthesized with a productivity of 7.58 mg/L/day. In this study, it was proved that the rate-limiting enzyme Dxs in MEP pathway could significantly enhance the carbon flux of valencene synthesis pathway. Compared with the previous articles on the production of valencene by cyanobacteria, we further evaluated the genotypic stability and verified the potential of the engineered strain in the long-term cultivation. The valencene synthesizing cell factory also showed significant robustness and could maintain a rapid growing status, accumulating over 150 mg/L valencene in 36-day semi-continuous cultivation. Remarkably, even in non-sterilized environments without antibiotic supplementation, cell growth and valencene production would still be well maintained, showing promising potentials for industrial application. Besides, the engineering strain JS119 with enhanced MEP pathway flux also demonstrated potentials as a chassis for engineering cell factories to produce other sesquiterpenes, cell factories converting carbon dioxide into bisabolene, amorpha-4,11-diene, farnesene, and nerolidol could be conveniently obtained, among which the photosynthetic production of nerolidol was reported for the first time.