Engineering Cupriavidus necator H16 for the autotrophic production of (R) -1,3-butanediol

Butanediols are widely used in the synthesis of polymers, specialty chemicals and important chemical in- termediates. Optically pure R -form of 1,3-butanediol (1,3-BDO) is required for the synthesis of several industrial compounds and as a key intermediate of β -lactam antibiotic production. The ( R )-1,3-BDO can only be produced by application of a biocatalytic process. Cupriavidus necator H16 is an established production host for biosynthesis of biodegradable polymer poly-3-hydroxybutryate (PHB) via acetyl-CoA intermediate. Therefore, the utilisation of acetyl-CoA or its upstream precursors offers a promising strategy for engineering biosynthesis of value-added products such as ( R )-1,3-BDO in this bacterium. Notably, C. necator H16 is known for its natural capacity to fix carbon dioxide (CO 2 ) using hydrogen as an electron donor. Here, we report engineering of this facultative lithoautotrophic bacterium for heterotrophic and autotrophic production of ( R )-1,3-BDO. Imple- mentation of ( R )-3-hydroxybutyraldehyde-CoA- and pyruvate-dependent biosynthetic pathways in combination with abolishing PHB biosynthesis and reducing flux through the tricarboxylic acid cycle enabled to engineer strain, which produced 2.97 g/L of ( R )-1,3-BDO and achieved production rate of nearly 0.4 Cmol Cmol (cid:0) 1 h (cid:0) 1 autotrophically. This is first report of ( R )-1,3-BDO production from CO 2 .


Introduction
1,3-butanediol (1,3-BDO) is an important platform chemical used in a variety of industrial applications including production of 1,3-butadiene, a precursor of synthetic rubber (Duan et al., 2016). Amongst other applications, 1,3-BDO is mainly employed in the production of unsaturated polyester resins, plasticizers, and industrial dehydrating agents. Owing to the low toxicity, and good water solubility, it can serve as a humectant and emollient in personal care products. The optically active R-form of 1,3-BDO is used in the production of pheromones, fragrances and insecticides (Matsuyama et al., 1993). (R)-1,3-BDO is also known as a precursor for most widely prescribed antimicrobial drugs, β-lactam antibiotics (Llarrull et al., 2010). Noteworthy, the 1, 3-BDO can be oxidized to its ketone form 4-hydroxy-2-butanone (4H2B), an important precursor for the synthesis of pesticides, steroids, and anticancer drug doxorubicin (Zhang et al., 2010).
An alternative microbial chassis that has shown great promise is chemolithoautotroph Cupriavidus necator H16 (formerly Ralstonia eutropha H16). This bacterium is able to grow aerobically and accumulate biomass to a very high level, competitive with E. coli, and exhibits a faster growth rate than cyanobacteria, high chemosynthetic efficiency and genetic tractability. C. necator H16 has been widely studied for its natural ability to produce the biodegradable polymer poly (3-hydroxybutryate) (PHB), used by this bacterium as a storage compound and accumulated in the presence of excess carbon and limited macro-elements such as nitrogen, phosphorus or oxygen . C. necator H16 is an ideal candidate to produce platform chemicals with its ability not only to metabolise a wide range of organic compounds but more importantly to recycle CO 2 by using the Calvin-Benson-Bassham (CBB) Cycle (Bowien and Kusian, 2002;Pohlmann et al., 2006). With the ability to fix CO 2 as a feedstock, C. necator provides a significant advantage compared to the sugar-based fermentation. Besides the gasification of plant's waste, which allows the complete utilisation of carbon contained within the biomass, CO 2 , suitable for gas fermentation, can be captured from chemical plants and steel mills reducing its emission to limit the climate change (Liew et al., 2016). Considering these advantages, C. necator H16 has been engineered to produce a wide range of commodity chemicals including methyl ketones, alcohols, terpenes, and alka(e)nes (Bommareddy et al., 2020;Chakravarty and Brigham, 2018;Crepin et al., 2016;Grousseau et al., 2014;Krieg et al., 2018;Lu et al., 2012;Müller et al., 2013) demonstrating its versatility and potential as an industrial chassis.
In this study, we aimed to engineer C. necator H16 for (R)-1,3-BDO production. Based on high availability of either (R)-3-hydroxybutyraldehyde-CoA ((R)-3HBCoA) or pyruvate precursors, two alternative (R)-1,3-BDO biosynthetic pathways were explored (Fig. 1). To increase (R)-1,3-BDO yield, a number of genetic improvements including PHB biosynthesis inactivation, redirection of the carbon flux through deletion of TCA cycle genes, and increase of the copy number of biosynthetic pathway genes were implemented. To ensure the genetic stability, both (R)-1,3-BDO biosynthetic pathways were chromosomally integrated in the engineered strains. Finally, autotrophic fermentation using CO 2 as sole carbon source was demonstrated for (R)-1,3-BDO production.

Plasmid construction
All plasmids and oligonucleotide primers used in this study are listed in Supplementary Tables 1 and 2, respectively. Plasmids were assembled using either the USER cloning method (Bitinaite et al., 2007), NEBuilder HiFi DNA assembly method (New England Biolabs) or restriction enzyme-based cloning techniques (Sambrook et al., 1989). Plasmid DNA preparation was carried out using the QIAprep® Spin Miniprep Kit (Qiagen). Gel purified linearized DNA was extracted using the QIA-quick® Gel Extraction Kit (Qiagen). Genomic DNA was isolated with the GenElute™ Bacterial Kit (Sigma-Aldrich). All restriction endonucleases, T4 DNA ligase and NEBuilder® HiFi DNA Assembly Master Mix were acquired from New England Biolabs. DNA sequences were verified by Sanger sequencing (Eurofins Genomics). A detailed assembly description for each plasmid is provided in the Supplementary information.

Strains, transformation and media
All bacterial strains used in this study are listed in Table 1. For strain transformation, E. coli DH5α, MG1655 and S17-1 competent cells were prepared according to (Sambrook et al., 1989), while electrocompetent C. necator cells were prepared as described in (Ausubel et al., 2003).

Gene knockout and knock-in generation in C. necator
Gene knockout and knock-in were performed using the pLO3 suicide vector exhibiting selection through tetracycline resistance (tetR) and counter-selection in the presence of sucrose (sacB). Chromosomal gene deletion was introduced by preserving start and stop codons of the gene. Where endogenous genes were replaced by introducing exogenous genes under control of the araC/P araBAD inducible system, to eliminate potential transcriptional read-through, rrnB T2 and rrnB T1 terminators were incorporated upstream and downstream to the heterologous DNA pLO3 suicide vector-based plasmids were transformed into E. coli strain S17-1 (ATCC 47055) suitable for conjugative plasmid transfer to C. necator H16. E. coli and C. necator strains were cultivated overnight in Luria-Bertani (LB) medium supplemented with 15 μg/mL tetracycline and 10 μg/mL gentamicin, respectively. Cells were harvested by centrifugation (5000×g for 10 min) and washed for mating on a LB-agar plate for 6 h at 30 • C. C. necator H16 transconjugants resulting from a first homologous recombination were isolated by plating onto MM-agar plates supplemented with 0.4% (w/v) sodium gluconate, 10 μg/mL gentamicin and 15 μg/mL tetracycline. Single colonies were then purified by re-streaking twice onto MM-agar plates containing gentamicin and tetracycline. Single colonies were used to inoculate 5 mL LB supplemented with gentamicin and tetracycline and cultivated overnight. Cultures were then used to inoculate 5 mL low sodium-LB (2.5 g/L NaCl) supplemented with 15% (w/v) sucrose for overnight growth. Cells were then plated onto low sodium-LB-agar plates supplemented with 15% (w/ v) sucrose and single colonies were streaked onto LB-agar plates containing 15 μg/mL tetracycline and no antibiotic to establish loss of integrated chromosomal pLO3 DNA by a second homologous recombination. Cells were then screened by PCR for successful gene deletions or integrations.

Two-stage batch fermentation in shake-flasks
A two-stage batch fermentation in shake-flasks was employed for the production of (R)-1,3-BDO in E. coli or C. necator. Biomass and synthetic pathway related proteins were generated by growing cells in rich media (LB) before transferring them to nutrient limited minimal media with excess carbon. 50 μg/mL or 300 μg/mL kanamycin was used throughout for E. coli or C. necator, respectively. Freshly transformed cells from single colonies were inoculated in 5 mL of LB medium and incubated for 18 h at 30 • C and 200 rpm with orbital diameter of 1.9 cm. Subsequently, cultures of E. coli or C. necator strains were resuspended to an OD 600 of 0.1 or 0.2 in 50 mL LB supplemented with 0.2% (w/v) glucose or 0.2% (w/v) sodium gluconate, respectively. The cultures were grown in 250 mL baffled shake-flasks at 30 • C and 200 rpm with orbital diameter of 1.9 cm. At an OD 600 of 0.6-0.8, 0.25% (w/v) L-arabinose was added and cultures were allowed to grow further for 4-6 h enabling heterologous gene expression. Then, E. coli cells were harvested by centrifugation (1700g for 6 min), resuspended in 25 mL M9 minimal medium (0.24 mg/mL MgSO 4 , 0.011 mg/mL CaCl 2 and M9 salts) (Sambrook et al., 1989) supplemented with 3% (w/v) glucose, 1 μg/mL thiamine and 20 μg/mL uracil (Jensen, 1993) to an OD 600 of 10 and incubated in 250 mL baffled shake-flasks at 30 • C and 200 rpm with orbital diameter of 1.9 cm. Whereas, C. necator cells were harvested by centrifugation for 10 min at 6600 g, resuspended in 25 mL MM (excluding NH 4 Cl) supplemented with 2% (w/v) sodium gluconate to an OD 600 of 7 and incubated in 250 mL baffled shake-flasks at 30 • C and 200 rpm with orbital diameter of 1.9 cm. Samples of 0.5 mL were taken immediately, 12 and 48 h after L-arabinose supplementation, centrifuged for 5 min at 17,000 g, and the cell-free supernatant was subjected to HPLC-UV/RI analysis.

HPLC-UV/RI analysis and chemical compound yield quantification
Prior subjecting to the HPLC-UV/RI analysis, the cell-free supernatant samples were combined with an equal volume of mobile phase (5 Table 1 Strains used in this study. P denotes P araBAD promoter with square brackets showing genes under promoter control. and stored in 2 mL snap cap vial closed with cap containing septa (Thames Restek; cat. no. SR-0101102-AL and SR-01011TSIT, respectively). Samples were analysed using a Thermo Scientific UltiMate 3000 HPLC system equipped with a diode array detector DAD-3000 with the wavelengths set at 210 nm and 280 nm, a refractive index detector RefractoMax 521 (Thermo Fisher Scientific), and Phenomenex Rezex ROA-organic acid H+ (8%) 150 mm × 7.8 mm × 8 μm column (Phenomenex). The column was operated at 35 • C with an isocratic flow rate of 0.5 mL/min. Samples were run for 30 min and the injection volume was 20 μl. Chromeleon Chromatography Data System software was used for HPLC system control, data processing and analysis. The concentrations of chemical compounds were estimated from standard calibration curves generated by analysing known concentrations of sodium gluconate (cat. no. 10356290) and ethanol (cat. no. 10437341) from Fisher Scientific; 4-hydroxy-2-butanone (Alfa Aesar; cat. no. L11456); 3hydroxybutyraldehyde (Aldol; cat. no. CDS019977) and acetic acid (cat. no. A6283) from Sigma-Aldrich; L-arabinose (cat. no. 365185000), 1,3-butanediol (99% purity, Cat. No. 107622500) and pyruvic acid (cat. no. 132145000) from Arcos Organics. Chemical compound yields per biomass (Y P/X ) and substrate (Y P/S ) were calculated using equations (1) and (2), respectively: where P* t and P* t-1 are concentrations of chemical compound (e.g. 1,3-BDO) in g/L for time points t and t-1, X t and X t-1 are dry cell weight concentrations in g/L for time points t and t-1.
where P t and P t-1 are concentrations of chemical compound in carbon mole (Cmol) for time points t and t-1, S t and S t-1 are concentrations for substrate sodium gluconate in Cmol for time points t and t-1.
To estimate dry cell weight (DCW), 1 mL of cell culture was centrifuged in pre-dried and pre-weighed 1.5 mL Eppendorf tubes for 2 min at 17000 g and the supernatant was discarded. The cell pellet was dried for 48 h at 120 • C in a Heratherm OGH60 gravity convection oven (Thermo Fisher Scientific). Subsequently, samples were cooled in a desiccator and the DCW was determined using an analytical balance with accuracy to 0.1 mg (SI-234, Denver Instrument). DCW was calculated as grams per litre.

Specific cell growth rate
Cell growth was monitored by measuring the OD 600 using a Bio-Mate™ 3S UV-Visible Spectrophotometer (Thermo Fisher Scientific, MA, USA). Specific growth rate (μ) was calculated using the following equation (Widdel, 2007).
where lnOD 1 and lnOD 0 are the calculated natural logarithm values of measured OD 600 for time points t 1 and t 0 .

Fermentation
Autotrophic fermentation was carried out in 1.3 L vessel using a DASGIP® parallel bioreactor 4-fold system with Bioblock for microbiology including control modules CWD4, MP8, PH4PO4L, PH4PO4RD4, OD4, MX4/4, TC4SC4 (Eppendorf) equipped with probes to measure dissolved oxygen (DO) (optical DO probe, Mettler Toledo), pH (405-DPAS-SC-K8S pH Probe, Mettler Toledo) and temperature Platinum RTD Temperature Sensor (Eppendorf). DASware® control software was used for automated control of DO, temperature, and pH. The preculture was prepared and fermentation was performed as described previously (Bommareddy et al., 2020) with some modifications. Briefly, The first seed culture was grown overnight at 30 • C with 200 rpm shaking in 10 mL of LB from a single colony. Subsequently, this culture was reseeded to 120 mL of LB and grown for another 24 h as above. Resulting cells were harvested by centrifugation for 10 min at 6600g, washed with 10 mL of FMM to remove residual LB, resuspended in 50 mL FMM and used to inoculate 700 mL FMM. If appropriate, antibiotics were added to the growth medium at the following concentrations: 10 μg/mL gentamicin or 300 μg/mL kanamycin. When cells reached DCW greater than 1 g/L protein expression was induced by addition of L-arabinose. pH was controlled at 6.9 by the addition of 1 M NH 3 OH until a DCW of 0.75 g/L was achieved, changing to 1 M KOH to limit nitrogen availability. DO was maintained at 10% (v/v) by increasing air flow (8.5-9.5 L/h) and agitation with a Rushton-type impeller (400-1600 rpm) and temperature at 30 • C. Using the DASGIP MX 4/4 Gas Mixing Module CO 2 , H 2 and air were continuously sparged through 0.22 μm membrane filters into the bioreactors. Gas outflow composition was analysed using a Bioprocess R&D Lab Gas Analyser, Model RLGA-9804 (Atmosphere Recovery Inc.). 2 mL samples were taken immediately after addition of L-arabinose and then every 12 h for 120 h and subjected to the HPLC-UV/RI analysis.

Choice of (R)-1,3-BDO biosynthetic pathways
The systematic approach to engineer C. necator H16 for 1,3-BDO production was based on the following design and experimental rationale: 1) considering alternative biosynthetic pathways which enable to utilise pyruvate and its downstream anabolic products as precursors; 2) screening enzymes with butanal dehydrogenase and aldehyde reductase activities enabling biosynthesis of 1,3-BDO from (R)-3-hydroxybutyraldehyde-CoA, the natural pyruvate's anabolic product in C. necator; 3) engineering C. necator H16 strain to improve the flux towards precursors required for 1,3-BDO biosynthesis; 4) establishing fermentation conditions and strain engineering to reduce the by-product biosynthesis; 5) ultimately, developing C. necator H16 strain suitable for production 1,3-BDO from CO 2 .
C. necator H16 lacks any phosphofructokinase (2.7.1.11; 2.7.1.90 or 2.70.1.146) of the Embden-Meyerhoff-Parnas (EMP) pathway and 6phosphogluconate dehydrogenase (1.1.1.44 or 1.1.1.343) of the oxidative pentose phosphate (OPP) pathway. Such organisation of metabolism restricts the flux through OPP and forward-EMP pathways and instead directs it through the Entner-Doudoroff pathway under heterotrophic growth conditions. Under autotrophic conditions, CO 2 is fixed by the reductive pentose phosphate cycle into the glyceraldehyde-3-phosphate and can increase the carbon flux through the reversed-EMP and ED pathways, as this has been observed under mixotrophic growth conditions (Alagesan et al., 2018b). The resultant flux distribution increases the availability of pyruvate that is used as a precursor for PHB synthesis in C. necator H16 under excess carbon and limited macro-elements conditions .
The (R)-3HBCoA pathway requires three NAD(P)H, whereas the pyruvate pathway utilises one NADPH with two NAD + molecules remaining in oxidized form due to the direct conversion of pyruvate into acetaldehyde. Both pathways are NAD(P)H-consuming with net use of three reducing cofactor molecules for each (R)-1,3-BDO synthesised, and are, therefore, heavily reliant on the efficient regeneration and balance of reducing equivalent within the cell. Indeed, Bld protein contains a proline and a nonpolar/aliphatic amino acid in sequence positions that correspond to the residues P222 and I257 of structurally similar PduP (Supplementary Figure 1), which are implicated in the selectivity for NADH over NADPH (Trudeau et al., 2018). Moreover, in vitro assays have shown that Bld possess the NADH-dependent activity (Hwang et al., 2014).
Previous research has shown that key TCA cycle genes (sucC, fumA, mdh1) are downregulated when C. necator cells transition from exponential to stationary growth phase alongside the upregulation of PHB required genes phaAB (Peplinski et al., 2010): as one of a key nutrient is depleted and biomass production becomes restricted, the flux through (R)-3HBCoA is increased and the carbon is accumulated in the form of PHB. This involves β-ketothiolase (PhaA), NADP-dependent acetoacetyl-CoA reductase (PhaB) and poly(3-hydroxyalkanoate) polymerase (PhaC) activities. Notably, the PHB can constitute up to 90% of the DCW, if the excess carbon is available under nitrogen-limiting conditions . This strongly suggests that a sufficiently large pool of precursor in form of 3HBCoA can be generated under nutrient-limiting conditions generating a driving force for (R)-1,3-BDO biosynthesis when the (R)-3HBCoA-dependent pathway is utilised. Moreover, the deletion of phaC1 gene significantly reduces the poly (3-hydroxyalkanoate) polymerase activity enabling accumulation of (R)-3HBCoA, which can be utilised for biosynthesis of (R)-1,3-BDO.

Screening of biosynthetic pathway variants
To enable implementation of (R)-3HBCoA-dependent (R)-1,3-BDO biosynthetic pathway, a screening of gene combinations, encoding enzymes with butanal dehydrogenase and aldehyde reductase activities, was performed (Supplementary Figure 2).
In this screen, as a substitute for the bifunctional AdhE2 from C. acetobutylicum (Fontaine et al., 2002), a butanal dehydrogenase (Bld) from C. saccharoperbutylacetonicum (Kosaka et al., 2007;Nair et al., 1994) was combined with a number of aldehyde reductases, including widely utilised YqhD from E. coli (Jarboe, 2011). The bld gene possessing a very low GC content of 32.8% was codon-optimised for expression in C. necator H16 (66.3% average GC content). Aldehyde reductase candidates were selected based on protein homology to YqhD or enzymatic activity on similar compounds reported previously, such as the conversion of 4-hydroxybutyraldehyde to 1,4-butanediol (Wang et al., 2017), acetoin to 2,3-butanediol (Yan et al., 2009) or the in vitro conversion of 3-hydroxybutyraldehyde to 1,3-butanediol (Kim et al., 2017). Furthermore, the yqhD gene was combined with eutE from two different species, as well as adhE2 and adhE1 from C. acetobutylicum and E. coli adhE were included.
All pathway variants were tested in C. necator H16 wild-type and PHB deficient mutant with the (R)-1,3-BDO biosynthesis observed only in the latter. (R)-1,3-BDO was produced in strains H16ΔC-p2, H16ΔC-p15 and H16ΔC-p26 expressing bld with yqhD from E. coli MG1655 (hereafter denoted as yqhD Ec ) or P. putida KT2440 (yqhD Pp ), and bifunctional adhE2 from C. acetobutylicum, respectively (Supplementary Figure 2). Other biosynthetic pathway variants did not show detectable quantities of the diol by HPLC-RI. Biosynthesis of (R)-1,3-BDO in C. necator H16 obtained using bifunctional adhE2 on its own or bld in combination with yqhD is consistent with previously reported activities of these enzymes (Hwang et al., 2014;Kataoka et al., 2013) confirming their indispensable role.
It should be noted that (R)-1,3-BDO exhibited only a minor toxic effect on C. necator H16 with no growth inhibition in the presence of up to 83.2 mM (Supplementary Table 3).

Evaluation of (R)-1,3-BDO biosynthesis
Previous research has shown that the PHB synthesis in C. necator H16 is increased under nitrogen limiting conditions with excess carbon available (Tian et al., 2005). The nitrogen limitation effect on (R)-1, 3-BDO yield was investigated H16ΔC-p26 by changing carbon/nitrogen (C/N) ratio in culture minimal medium from 6 to 50. In spite of decrease in growth rate, more than 2-fold higher yield of (R)-1,3-BDO was observed using C/N ratio of 50 (Supplementary Figure 3). Furthermore, Y 1,3-BDO/S of 0.018 was measured 24 h after induction under nitrogen limitation, whereas in non-limiting nitrogen conditions (R)-1,3-BDO became detectable only after 48 h. These results demonstrate that (R)-1, 3-BDO biosynthesis in C. necator H16 can be improved by limiting nitrogen availability.
The AraC/P araBAD -arabinose inducible system is relatively well repressed under uninduced state and can be fine-tuned in the range from 0.00117 to 0.15% (w/v) of L-arabinose allowing to achieve more than 1000-fold induction in C. necator H16 (Alagesan et al., 2018a). Importantly, the L-arabinose is not metabolised by this bacterium and does not exhibit any adverse effect on the cell growth (data not shown). Therefore, AraC/P araBAD inducible system was selected to drive overexpression of (R)-1,3-BDO biosynthesis genes.
To establish an optimal gene expression level of (R)-3HBCoAdependent (R)-1,3-BDO biosynthetic pathway, induction conditions using a range of L-arabinose concentrations (from 0.005 to 0.2% (w/v)) were investigated. For H16ΔC-p2 strain expressing bld and yqhD Ec , a direct correlation between the biosynthesis levels of (R)-1,3-BDO and concentration of inducer was observed 24 h after induction with the highest quantity of (R)-1,3-BDO produced in the cell culture that was supplemented with 0.2% of L-arabinose (Supplementary Figure 4). However, at the later stages of induction, the specific (R)-1,3-BDO production was reduced and a strong growth inhibition observed in cultures supplemented with higher than 0.045% concentrations of Larabinose. Altogether these results revealed that 0.01-0.045% concentrations of L-arabinose are optimal for induction of (R)-1,3-BDO biosynthetic pathway genes when the plasmid-based expression system is used in C. necator H16.
Next, (R)-1,3-BDO-producing C. necator strains H16ΔC-p2, H16ΔC-p15 and H16ΔC-p26 were compared under heterotrophic nitrogenlimited growth conditions (Fig. 2). Cumulative yields of (R)-1,3-BDO were steady for the duration of 96-h cell growth period ranging from 0.035 to 0.055 Cmol Cmol − 1 , and comparable between all three strains. Increase in biomass and 1,3-BDO was greatest during initial 24-h post induction period with a highest yield of 0.055 ± 0.003 Cmol Cmol − 1 obtained using strain H16ΔC-p2. It can be concluded that of three strains possessing alternative combinations of genes of (R)-3HBCoA-dependent pathway, the strain H16ΔC-p2 performed marginally better producing highest yields of (R)-1,3-BDO during early logarithmic and stationary growth periods while exhibiting the least growth impairment. Furthermore, YqhD Ec aldehyde reductase specificity on butanal is higher (Km = 0.67) than that of AdhE2 (Km = 1.60) as reported previously (Palosaari and Rogers, 1988;Pérez et al., 2008). Therefore, the combination of genes bld and yqhD Ec was chosen to be utilised for (R)-3HBCoA-dependent biosynthetic pathway in next stages of this study.
Overexpression of yqhD has a clear adverse effect on the (R)-1,3-BDO yield in E. coli but not in C. necator. This is likely due to acetaldehyde dehydrogenase activity causing production of ethanol as reported previously (Nemr et al., 2018) and with this associated depletion of NADPH.

Metabolic by-products of the (R)-3HBCoA-dependent pathway
As indicated in section 3.1, C. necator H16 strains with ΔphaC1 background were primarily used for biosynthesis of (R)-1,3-BDO. Further analysis of extracellular metabolite composition revealed that, alongside the (R)-1,3-BDO, large amounts of pyruvate, representing yields of 0.419 ± 0.003 Cmol Cmol − 1 , 0.505 ± 0.008 Cmol Cmol − 1 and 0.415 ± 0.011 Cmol Cmol − 1 , were respectively excreted from strains H16ΔC-p2, H16ΔC-p15 and H16ΔC-p26, containing (R)-3HBCoAdependent pathway variants. Whereas only negligible quantities of acetate and ethanol were detected in these strains. The pyruvate was completely absent in cultures of wild-type background strains harbouring same biosynthetic pathway variants. The accumulation and excretion of pyruvate has been reported previously in C. necator H16 ΔpdhL and PHB − 4 (DSM541) strains (Raberg et al., 2011;Steinbüchel and Schlegel, 1989). The former is deficient of the dihydrolipoamide dehydrogenase (E3) component of pyruvate dehydrogenase complex. The accumulation of pyruvate in PHB − strains indicates that the deficiency of poly(3-hydroxyalkanoate) polymerase activity causes the build-up of upstream metabolites of the PHB pathway and that the increase in acetyl-CoA level inhibits the pyruvate dehydrogenase activity, as postulated previously (Jung and Lee, 1997;Raberg et al., 2011;Steinbüchel and Schlegel, 1989). Simultaneously, the pyruvate accumulation suggests that the (R)-3HBCoA-dependent pathway exhibits limited capacity to drive carbon flux towards the (R)-1,3-BDO.

Overexpression of endogenous phaA and phaB1
The endogenous C. necator H16 genes phaA and phaB are essential for biosynthesis of 3-HBCoA from acetyl-CoA (Fig. 1). In order to assess whether enhanced expression of phaA and phaB by increasing their copy number and expression level can improve (R)-1,3-BDO production, Fig. 3. Biosynthesis of 4H2B in C. necator expressing heterologous bld and yqhD genes. (A) Schematic of the 4H2B biosynthetic pathway. Acetoacetyl-CoA is converted to 3-oxobutanal by Bld exhibiting promiscuous acylating dehydrogenase properties. Then, 3-oxobutanal is subsequently reduced to 4H2B by YqhD. (B). 1,3-BDO carbon yield (bars) and 4H2B carbon yield (striped bars) in batch fermentation cultures of H16ΔCB-p14 (i), H16ΔCB-p2 (ii), H16ΔCB-p44 (iii) and H16ΔC-p2 (iv). Plus or minus sign indicates the presence or absence of a gene. Results represent the average of three biological replicates and error bars show standard deviation.
phaA and phaB in addition to bld and yqhD genes were included in the plasmid-based overexpression system yielding pJLG35. The yield of (R)-1,3-BDO in H16ΔC-p35 containing chromosomal and plasmid-based copies of phaAB was compared to that in H16ΔC-p15 (chromosomal copy of phaAB), H16ΔCAB-p15 (no phaAB) and H16ΔCAB-p35 (plasmid-based only copy of phaAB) (Fig. 4). Of all strains, H16ΔC-p35 and H16ΔCAB-p35 exhibited the diol production within the first 24 h, whereas the former maintained highest yield (approximately 0.045 Cmol Cmol − 1 ) throughout the rest of 120-h fermentation. Evidently, expression of plasmid-based phaAB genes encoding acetoacetyl-CoA reductase improved utilisation of carbon source and conversion of pyruvate at the later stages of fermentation. Interestingly, despite the lack of phaAB in strain H16ΔCAB-p15, the production of (R)-1,3-BDO was still observed, albeit at much lower yields of 0.008 ± 0.003 Cmol Cmol − 1 . Specific production of 1,3-BDO by strain H16ΔCAB-p15 after 120 h indicates combined activity of one or multiple β-ketothiolase homologues reported in the C. necator genome (Lindenkamp et al., 2012) and acetoacetyl-CoA reductase PhaB3 (H16_A2171) possessing reduced rate compared to PhaB1 .
As predicted, deletion of sucCD helped to improve (R)-1,3-BDO yield likely through increased acetyl-CoA pool. Despite the loss of ATP generation by the deletion of sucCD, all strains exhibited similar specific growth rates. Indistinctly, iclAB deletion strains showed no improvement in (R)-1,3-BDO biosynthesis as the sole deletion or when combined with sucCD deletion. Since iclAB has been previously reported to be Fig. 4. Improvement of (R)-1,3-BDO production by overexpression of phaAB. Batch fermentation profile data for strains H16ΔC-p15 (i); H16ΔC-p35 (ii); H16ΔCAB-p15 (iii) and H16ΔCAB-p35 (iv) are presented as following: (A) (R)-1,3-BDO yield (bars), (B) biomass DCW (circles) sodium gluconate concentration (triangles) and pyruvate yield (upside down triangles). Cells were grown in NLMM supplemented with 2% (w/v) sodium gluconate. The gene expression was induced by addition of 0.01% (w/v) arabinose. Results represent the average of three biological replicates and error bars show standard deviation.
primarily involved in β-oxidation pathways (Brigham et al., 2010;Sharma et al., 2016) and only low expression level was observed under heterotrophic growth (Alagesan et al., 2018b) this indicates its reduced involvement in gluconate metabolism and flux through the glyoxylate bypass. Fig. 5. Improvement of (R)-1,3-BDO yields by sucCD deletion. Yields of (R)-1,3-BDO (bars) and pyruvate (upside down triangles), and DCW obtained using strains H16ΔC-p2 (i), H16Δ2-p2 (ii), H16Δ3-p2 (iii), H16Δ4-p2 (iv) are shown. Cells were grown in NLMM supplemented with 2% (w/v) sodium gluconate. The biosynthetic pathway gene expression was induced by addition of 0.01% (w/v) arabinose. Results represent the average of at least two biological replicates and error bars show standard deviation. The yield (Cmol Cmol − 1 of sodium gluconate) of major byproducts excreted by the engineered C. necator H16 are highlighted as following: pyruvate (upside down triangles), acetate (crosses) and ethanol (diamonds). Strain H16ΔC-p306 was cultivated in NLMM supplemented with 2% (w/v) sodium gluconate and biosynthetic pathway gene expression was induced by addition of 0.01% (w/v) arabinose. Results represent the average of three biological replicates and error bars show standard deviation.

Evaluation of pyruvate-dependent pathway
The pyruvate-dependent biosynthetic pathway has been recently developed for (R)-1,3-BDO production in E. coli (Kim et al., 2017;Nemr et al., 2018). This pathway consisting of pyruvate decarboxylase (PDC), deoxyribose-5-phosphate aldolase (Dra) and aldo/keto reductase (AKR) enables to convert pyruvate to (R)-1,3-BDO through acetaldehyde and (R)-3HBA intermediates. To utilise the pyruvate that accumulates in C. necator ΔphaC strains, the pyruvate-dependent pathway was implemented in this study. With YqhD Ec proved suitable for conversion of (R)-3HBA to (R)-1,3-BDO in engineered C. necator, the gene of this enzyme was combined with PDC from Z. mobilis ZM4 and dra from B. halodurans into the plasmid pJLG306 yielding strain H16ΔC_p306 (Fig. 6A). However, similarly to H16ΔC-p26, this strain did not produce detectable quantities of (R)-1,3-BDO by HPLC-RI analysis under heterotrophic growth conditions (Fig. 6B). The further metabolite analysis revealed no accumulation of pyruvate, indicating that it is completely converted into acetaldehyde by PDC (Fig. 6C). However, high yields of acetate and ethanol suggest that Dra is ineffective in converting acetaldehyde to (R)-3HBA and causes a bottleneck in the pyruvate-dependent biosynthetic pathway. This is also supported by previous results showing that the gene copy number and expression level of dra contribute to the increase of (R)-1,3-BDO yield (Nemr et al., 2018). Furthermore, a rapid acetate synthesis from acetaldehyde is likely to be associated with acetaldehyde dehydrogenase AcoD activity in C. necator H16 (Priefert et al., 1992), whereas a low affinity of YqhD Ec towards acetaldehyde (Pérez et al., 2008) can contribute to the gradual increase in the ethanol yield during the 120-h fermentation.

Combining (R)-3HBCoA-and pyruvate-dependent pathways
Considering the absence of any detectable (R)-1,3-BDO production by the pyruvate-dependent pathway in C. necator and aiming to reduce accumulation of pyruvate and improve carbon flux through acetyl-CoA node, it was reasoned that the combination of both, pyruvate-and (R)-3HBCoA-dependent pathways, may improve (R)-1,3-BDO biosynthesis. As postulated previously, the acetoin dehydrogenase bypass can counteract an accumulation of pyruvate and utilise acetaldehyde that is generated as pyruvate-dependent pathway intermediate, by offering an alternative route to acetyl-CoA, especially, when pyruvate dehydrogenase complex is inhibited by elevated concentration of acetyl-CoA (Raberg et al., 2014).

Chromosomal integration of biosynthetic pathway
To improve genetic stability ensuring stable expression of (R)-1,3-BDO biosynthetic pathway genes, chromosomal integration of the constructs into phaCAB loci was performed. Simultaneously, heterologous genes for either 3-HBCoA-dependent pathway or combining the 3-HBCoA-dependent and pyruvate-dependent pathways were introduced. To ensure tuneable expression of chromosomally integrated heterologous genes, an arabinose inducible system (araC/P araBAD ) preceded with terminator was integrated into the phaC locus upstream of Fig. 7. Improvement of (R)-1,3-BDO yield by combining 3-HBCoA-dependent and pyruvate-dependent pathways. (A) Schematic of cumulative biosynthetic pathway indicating routes of (R)-1,3-BDO and by-product formation. Batch fermentation product yields (Cmol Cmol − 1 of sodium gluconate) for strains H16ΔC-p2 (i) and H16ΔC-p304 (ii) are presented as following: (R)-1,3-BDO (solid bars) (B); and 4H2B (squares), pyruvate (upside down triangles), acetate (crosses) and ethanol (diamonds) (C). Engineered strains were cultivated in NLMM supplemented with 2% (w/v) sodium gluconate and biosynthetic pathway gene expression was induced by addition of 0.01% (w/v) arabinose. Results represent the average of three biological replicates and error bars show standard deviation. either bld and yqhD Ec (strain H16Δ1::54) or bld, yqhD Ec , dra, and PDC (strain H16Δ1::56). By design, utilisation of the phaC locus as a target integration site not only abolished the PHB synthesis but also ensured a controllable expression of phaA and phaB, which are required for (R)-1,3-BDO production. Nonetheless, engineered strains contained only a single copy of chromosomally integrated biosynthetic pathway genes, and despite a significant reduction in gene copy number comparing to the plasmid-based expression system, a detectable level of (R)-1,3-BDO was observed for both strains H16Δ1::54 and H16Δ1::56 under nonoptimal growth conditions with limited aeration (Table 3).
Earlier results indicated that limited expression of either bld or dra can create a bottleneck in the (R)-1,3-BDO biosynthetic pathways. Moreover, bld from C. saccharoperbutylacetonicum is potentially an oxygen-sensitive enzyme, similarly to its homologue from C. beijerinckii (Yan and Chen, 1990). Therefore, to further improve strains H16Δ1::54 and H16Δ1::56, a second copy of these genes was introduced by replacing sucCD, deletion of which was identified in this study as beneficial for improving (R)-1,3-BDO yield. An additional copy of bld was integrated into the strains containing either the 3HBCoA-dependent pathway (H16Δ1::54/Δ3::58) or the combined 3HBCoA-and pyruvate-dependent pathway (H16Δ1::56/Δ3::58). The bld gene was placed under the control of a strong constitutive promoter (P 8 ) (Alagesan et al., 2018a). The same strategy was employed for integration bld and dra into the strain with combined 3HBCoA-and pyruvate-dependent pathway (H16Δ1::56/Δ3::60). All engineered strains were screened by measuring (R)-1,3-BDO and by-products yields ( Table 3). As expected, for all strains, diol yield was reduced compared with plasmid-based expression system. Nonetheless, a clear improvement of (R)-1,3-BDO biosynthesis was achieved by introducing additional copies of dra and/or bld.

Further improvement of autotrophic (R)-1,3-BDO production by increasing bld copy number
Moreover, to further evaluate if the increase in the copy number of biosynthetic pathways genes can improve (R)-1,3-BDO production, strains containing chromosomally integrated bld, yqhD Ec , dra, and PDC, were transformed with plasmid carrying bld and dra copies. (R)-1,3-BDO and by-product profiles of resulting strains H16Δ1::54_p14 (chromosomal bld and yqhD Ec ; plasmid bld), H16Δ1::56_p14 (chromosomal bld, yqhD Ec , dra, and PDC; plasmid bld) and H16Δ1::56_p45 (chromosomal bld, yqhD Ec , dra, and PDC; plasmid bld and dra) were compared to earlier characterised strains H16ΔC_p2 and H16ΔC_p304 (Supplementary Figure 5). A significant improvement of (R)-1,3-BDO yield was observed in strains (H16Δ1::54_p14 and H16Δ1::56_p14) with additional copy of bld on the plasmid. Whereas, the addition of dra had only a marginal effect on the (R)-1,3-BDO yield. As observed previously, by introducing the pyruvate-dependent pathway, no pyruvate accumulation is observed demonstrating efficient metabolism of pyruvate to acetaldehyde facilitated by PDC.
Highest producing strains H16Δ1::56_p14 and H16Δ1::56_p45 were subjected to autotrophic fermentation using CO 2 as a sole carbon source. Despite successful production of (R)-1,3-BDO in shake-flask mode, strain H16Δ1::56_p45 was genetically unstable due to the plasmid pJLG45 loss, which was observed at the early stage of fermentation by plating cell culture on non-selective medium and selective medium with chloramphenicol antibiotic. Therefore, the (R)-1,3-BDO or another metabolite production was inconsistent and was not subjected to further analysis. Nonetheless, H16Δ1::56_p14 achieved the highest reported (R)-1,3-BDO titre of 33 mmol L − 1 (2.97 g/L) (Fig. 9). With theoretical yield of 1.00 for (R)-1,3-BDO production from CO 2 , a yield of 0.77 Cmol Cmol − 1 for 72-to 84-h fermentation period and average yield of 0.4 Cmol Cmol − 1 were obtained. Furthermore, 4H2B production was high (19.7 mmol L − 1 titer and average yield close to 0.3). 4H2B yield increased during later stage of fermentation indicating insufficient conversion of acetoacetyl-CoA to 3-hydroxybutanal facilitated by PhaB, despite being under the control of the arabinose inducible system. With such high yields of (R)-1,3-BDO and 4H2B there was no other byproducts detectable.

Conclusions
Here, we report the stepwise engineering of C. necator H16 for production of (R)-1,3-BDO from CO 2 . To achieve this, two alternative heterologous (R)-1,3-BDO biosynthetic pathways, based on utilisation of either (R)-3HBCoA or pyruvate as precursors, were investigated. Initially, the (R)-1,3-BDO biosynthesis was achieved by heterologous gene expression of either C. saccharoperbutylacetonicum bld in combination with E. coli yqhD or C. acetobutylicum adhE2. The (R)-1,3-BDO yield was improved through the genetic inactivation of the PHB biosynthesis by deletion of either phaC1 gene or phaCAB operon and redirecting excess carbon toward the diol production. (R)-1,3-BDOproducing strains were further improved by introducing extra copies of phaA, phaB1, bld and dra, as well as by deleting sucCD genes. An alternative (R)-1,3-BDO biosynthetic pathway was implemented by heterologous expression of PDC from Z. mobilis, and dra and yqhD from E. coli. The introduction of this biosynthetic pathway did not yield a detectable level of (R)-1,3-BDO, whereas the combination of both biosynthetic pathways resulted in a highest diol production. Further to this, genes of both (R)-1,3-BDO biosynthetic pathways were chromosomally integrated ensuring the genetic stability of engineered strains. Application of (R)-3HBCoA-and pyruvate-dependent pathways, in combination with abolishing the PHB biosynthesis and reducing the flux through the tricarboxylic acid cycle, enabled to engineer a strain that was able to produce more than 2.97 g/L of (R)-1,3-BDO via autotrophic fermentation from CO 2 . In this fermentation mode a large proportion of carbon Table 3 (R)-1,3-BDO and by-product yields in engineered C. necator strains. Cells were grown in 10 mL of 2 nitrogen-limiting minimal media supplemented with 2% sodium gluconate and 0.1% arabinose for 72 h.  Fig. 8. Autotrophic fed-batch fermentation of CO 2 for (R)-1,3-BDO production using DASGIP parallel bioreactor system. Data for strains H16Δ1::54 (i), H16Δ1::54/Δ3::58 (ii), H16Δ1::56 (iii) and H16Δ1::56/Δ3::60 (iv) represented as following: production rate of (R)-1,3-BDO (solid bars) and CUR (triangles) (A); (R)-1,3-BDO titer (solid bars) (B); 4H2B yield (squares) and DCW (circles) (C); and acetate (squares), ethanol (diamonds), and pyruvate (upside down triangles) yields (D). Due to the continuous supply of carbon source, metabolite Cmol Cmol − 1 yields were calculated by dividing metabolite production within a 12 h time period by average carbon uptake rate (CUR mmol h − 1 ) for the identical 12-h time period. Results represent the average of three technical replicates (sampling) that were taken from single reactor for each strain.