Engineering a Rhodopsin-Based Photo-Electrosynthetic System in Bacteria for CO2 Fixation

A key goal of synthetic biology is to engineer organisms that can use solar energy to convert CO2 to biomass, chemicals, and fuels. We engineered a light-dependent electron transfer chain by integrating rhodopsin and an electron donor to form a closed redox loop, which drives rhodopsin-dependent CO2 fixation. A light-driven proton pump comprising Gloeobacter rhodopsin (GR) and its cofactor retinal have been assembled in Ralstonia eutropha (Cupriavidus necator) H16. In the presence of light, this strain fixed inorganic carbon (or bicarbonate) leading to 20% growth enhancement, when formate was used as an electron donor. We found that an electrode from a solar panel can replace organic compounds to serve as the electron donor, mediated by the electron shuttle molecule riboflavin. In this new autotrophic and photo-electrosynthetic system, GR is augmented by an external photocell for reductive CO2 fixation. We demonstrated that this hybrid photo-electrosynthetic pathway can drive the engineered R. eutropha strain to grow using CO2 as the sole carbon source. In this system, a bioreactor with only two inputs, light and CO2, enables the R. eutropha strain to perform a rhodopsin-dependent autotrophic growth. Light energy alone, supplied by a solar panel, can drive the conversion of CO2 into biomass with a maximum electron transfer efficiency of 20%.


S4
Blh_R and inserted into the HindIII site of pLO11a-CRT to create pLO11a-blhDxrCRT. The GR gene with its own PBAD promoter was created by PCR from pLO11a-GR using primers pBAD_F and LOterm_R and inserted into the EcoRI site of pLO11a-blhDxrCRT to create pLO11a-blhDxrCRT-GR.

Carotenoid extraction and analysis
Strains transformed with the pLO11aa expression vector constructs were grown in 50ml LB with 12.5 µg ml -1 tetracycline to log phase and either induced with 0.1% (w/v) L-arabinose (Sigma-Aldrich) and overnight growth at 30 °C for R. eutropha or 0.2% (w/v) L-arabinose and overnight growth at 37 °C for E. coli. Carotenoids were solvent extracted from the 50ml cell culture pellets with 1 ml of 7:2 acetone:methanol (v/v), clarified by centrifugation and an absorption spectrum taken using an Agilent Cary 60 UV-Vis spectrophotometer. For more detailed analysis the solvent extract was dried down under nitrogen, re-dissolved in methanol and carotenoids separated by reversed-phase HPLC on an Agilent 1100 HPLC system using a Supelco Discovery HS C18 column as previously described (5). Elution of β-carotene and retinal were monitored at 450 nm and 380 nm respectively. An all trans-β-carotene standard (Sigma-Aldrich) of known concentration was used for the quantification of integrated peak areas. A pure all trans-retinal standard (Sigma-Aldrich) was used to assign the retinal peak.

Membrane purification
A 20 ml LB overnight culture with appropriate antibiotic selection of R. eutropha was used to inoculate 500 ml LB with selection in a 2.5 L conical flask. This was grown at 150 rpm and 30ºC to OD600 = 0.5 -07, induced with 0.1% (w/v) L-arabinose and 5 µg ml -1 all trans-retinal and grown overnight at 30ºC. The culture was pelleted and resuspended in approximately 10 ml of buffer A (25 ml K2HPO4/KH2PO4 pH 7.4) and lysed by two cycles of French pressing at a pressure of 18 000 psi. Membrane fractions were isolated on a 10-50% continuous sucrose gradient (the sucrose was made up in buffer A and 1.5 ml broken cells loaded per gradient) spun at 30,000 rpm for 2 hours at 4°C. A 1 ml membrane fraction was removed and an absorption spectrum taken using an Agilent Cary 60 UV-Vis spectrophotometer.

S5
Subsequently, the cell pellet was resuspended in the nitrogen-limited minimal medium, and the OD was adjusted to 1 before being transferred into flasks with a total volume of 250 mL.
Formate (80 mM) was added into baffled flasks with a working volume of 200 mL for microaerobic PHB synthesis. The flasks were illuminated with a white LED light (~50 μmol/s/m 2 ) and dark wrapped in foil. Unless otherwise stated, all cultures were grown at 30 °C and 150 rpm.

Identification of genes for beta-carotene synthesis in R. eutropha H16
Analysis of the KEGG pathway for R. eutropha H16 (Cupriavidus nectar H16; https://www.genome.jp/kegg-bin/show_pathway?reh00906) shows that it contains homologues of crtE and crtB (Genbank accession numbers CAJ92601.1 and CAJ93812.1 respectively) but not crtI and crtY. To see whether these homologous crtE and crtB genes were functional, and their products could be combined with exogenous CrtY and CrtI enzymes to make β-carotene, the crtY and crtI genes were isolated as one fragment by PCR from the crt operon of Erwinia herbicola, where they are situated adjacent to each other in the crt operon, and inserted into the modified pLO11 (Tc r , RK2 ori, Mob + ) vector containing the arabinose inducible PBAD promoter (pLO11a) specifically designed for expression in R. eutropha (1).

Overexpression of the dxr gene in R. eutropha enhanced β-carotene production
Isopentenyl diphosphate (IPP) is the common, five-carbon building block in the biosynthesis of all carotenoids with the second reaction in its synthesis being the reduction of 1-deoxy-Dxylulose 5-phosphate (DXP) to 2-C-methyl-D-erythritol-4-phosphate, catalysed by DXP reductoisomerase and encoded by dxr ( Fig. 2A). Overexpression of the Synechocystis dxr gene in tobacco and the maize dxr gene in Zea mays has been shown to increase carotenoid content in tobacco and maize plants (6,7). In order to try and increase the levels of β-carotene in R.

Redirection of carbon flux promoted β-carotene production in R. eutropha
The pDxrCRT construct was transformed into the RHM5 ΔphaCAB strain (henceforth referred to as H16Δpha) as it has been shown that deleting the phaCAB operon encoding the three metabolic pathway genes from acetyl-CoA to polyhydroxybutyrate (PHB) increases production in R. eutropha of ethanol from acetate (8). The resulting strain looked demonstrably more yellow after induction with arabinose (Suppl Fig. 2D) and β-carotene could easily be extracted into solvent and detected using an absorbance scan (Suppl Fig 2E) where the effect of induction on β-carotene production was clearly observable; subtraction of the uninduced S7 scan from the induced gave a difference spectrum identical to that of β-carotene (Suppl Fig.   4F); HPLC analysis confirmed the presence in the solvent extract of trans-and cis-β-carotene observed as peaks eluting at about 16.9 minutes and 17.1 minutes respectively (Suppl Fig. 2G).
Three weighed cell pellets from 50-ml cultures of R. eutropha H16 pCRT and H16 pDxrCRT were each solvent extracted, the extract analysed by HPLC to determine the mean integrated β-carotene peak areas (as determined by Agilent ChemStation HPLC software) and compared against the peak areas obtained from several dilutions of a 1 mg/ml β-carotene standard dissolved in methanol. Yield values for R. eutropha H16 pCRT and pDxrCRT of approximately 0.6 ± 0.2 µg and 0.9 ± 0.1 µg β-carotene per g wet weight of pellet respectively were obtained showing that more β-carotene was produced in the presence of dxr. A further improvement in yield was obtained with R. eutropha H16Δpha pDxrCRT where a value of approximately 2 ± 0.5 µg β-carotene per g wet weight of pellet was obtained (Suppl Fig. 2H).

Validation of GR biosynthesis in E. coli and R. eutropha H16 using single cell Raman microspectroscopy
In order to determine the presence in vivo of the GR-retinal complex in cells expressing the GR gene, samples of uninduced and arabinose induced H16Δpha GR, JM109 blhDxrCRT-GR and H16Δpha blhDxrCRT-GR (labelled H16); this band was also present in the H16 blhDxrCRT-GR strain (where the pha operon has not been deleted), appearing upon arabinose induction (Suppl Fig S4B). This characteristic Raman band was attributed to ethylenic stretching (νC=C) vibrations in retinal-protein complexes (9), which depending on the carbon chain length and structure, can shift between 1505-1530 cm -1 (10). To date, Raman spectroscopy has only been carried out on purified GR protein either as crystals (11) or reconstituted into liposomes (12) where the ethylenic stretching (νC=C) band was typically seen between 1524-1535 cm -1 depending on the pH conditions.

S8
The percentage of cells expressing GR-retinal in a population was calculated by counting the number of SCRS that contained the 1530 cm -1 GR band (Table S2). The induced JM109 blhDxrCRT-GR sample had 41% of its population present with GR-retinal whilst 9% of the uninduced population also had GR-retinal present which is presumably a reflection of some 'leakiness' in the PBAD promoter. Induced samples of H16Δpha blhDxrCRT-GR and H16Δpha GR showed 8% and 89% respectively of the populations expressing GR-retinal; no Raman band at 1530 cm -1 was observed in the uninduced samples. The percentage of cells expressing GR-retinal (Table S2) was highest in the H16Δpha GR strain, to which exogenous retinal had been added, and is presumably a reflection of the reduced amount of endogenous retinal produced in E. coli JM109 and R. eutropha H16Δpha by the pLO11a blhDxrCRT-GR construct, although the higher percentage of cells in the former indicates better expression of the construct in E. coli. As the amount of endogenous retinal is the limiting factor in the production of GR-retinal there is certainly more scope for improving its production, for example, through codon optimisation of all the genes, the use of different promoters with differing induction conditions and molecular evolution/screening techniques.   19.8% increase