Introduction

The past few decades have witnessed a continuous increase in CO2 emissions due to the use of nonrenewable hydrocarbon fuels [1]. The ever-increasing anthropogenic energy demand has further aggravated the CO2 content in the atmosphere [2], exacerbating global warming and climate change [3, 4]. This issue has led to a number of approaches for the direct capture and conversion of CO2, including physical, chemical, and biological methods [5,6,7]. Among them, the direct conversion of CO2 to value-added compounds, such as energy fuels, shows great promise. In these approaches, catalysts, such as rare metals and enzymes, are usually required to facilitate the conversion of CO2 [8,9,10]. However, low product selectivity, poor sustainability, and intensive energy investment have often limited their applications [11, 12]. Another option is to use whole-microbe biocatalysts, taking advantage of the highly organized pathways that favor the multistep CO2 fixation process and operate under the protection of the cell envelope [8, 13].

Energy investment is necessary to drive CO2 fixation in whole-cell biocatalytic systems [14]. Green energy, such as solar light, is thought to be the best choice [15]. Initial attempts have involved the construction of photoelectrochemical semiconductor biohybrids [16,17,18]. In these systems, semiconductor electrodes are used to harvest light and produce photoexcited electrons that are then “fed” to autotrophic bacteria via the cathode and serve as reducing power for CO2 fixation. However, strong illumination is required to excite the semiconductor electrode for electron generation, and the inefficient electron transfer between the electrode and microbes often limits the operational efficiency of the system [15]. In contrast, photosynthetic semiconductor biohybrids display efficient electron transfer and contribute to high solar-to-chemical efficiency [13, 19,20,21]. In biohybrids, light-harvesting semiconducting nanomaterials integrate whole-cell biocatalysts by forming intimate connections on/in microbial cells [20, 22, 23]. Thereafter, photoelectrons directly enter the reductive pathway and drive reducing power generation. Such biohybrids represent the best available mimics of photosynthesis in photosynthetic bacteria and provide potential solutions for nonphotosynthetic autotrophic bacteria to combine solar energy harvesting and CO2 fixation. Unfortunately, the innate defects of semiconductors, including cytotoxicity, photocorrosion, photodegradation, and the generation of photoexcited radicals, and their irreproducibility impair the stability and sustainability of biohybrids [15, 24, 25].

Photosynthetic bacteria house both light-harvesting elements and CO2-fixation machinery. Several photosynthetic bacteria, such as cyanobacteria and microalgae, have been selected as platforms to perform photosynthesis for CO2 to value-added product conversion [7, 26, 27]. Even though the natural photoreaction center in a photosynthetic bacterium is much more efficient in light absorption and energy conversion than any of the man-made photosensitizers, with an ability to be excited by a wide light spectrum and intensity, the carbon fixation pathway in photosynthetic bacteria is often less energetically favorable than that in lithoautotrophs [28, 29]. One potential solution to this problem involves the combination of the light-harvesting center of photosynthetic bacteria with the CO2-fixation machinery of nonphotosynthetic bacteria via the construction of a hybrid system to achieve efficient solar light-driven CO2 fixation. Notably, pioneering attempts have been performed by expressing the photoreaction center and allowing photophosphorylation in a nonphotosynthetic heterotrophic bacterium [30].

Rhodopseudomonas palustris is a purple nonsulfur photosynthetic bacterium that has been reported to generate current or reduce iron(III) oxides under illumination [31, 32], suggesting an ability to generate photoelectrons and transfer these electrons extracellularly. We speculated that R. palustris might be able to be used as a “natural whole-cell photosensitizer.” Here, we report the verification of this speculation by coculturing R. palustris with Methanosarcina barkeri to drive the conversion of CO2 to CH4. M. barkeri was chosen for study because: (1) it has been previously reported to directly accept electrons from other bacterial species, such as the exoelectrogen Geobacter metallireducens [33], after forming a direct electron syntrophic coculture to reduce CO2 for CH4 generation, and (2) because CH4 is a potential renewable energy source. In the coculture, both time-dependent and light-dependent methanogenesis were evaluated, and it was concluded that R. palustris and M. barkeri formed electron syntrophic cocultures to perform CO2-to-CH4 conversion. That is, R. palustris acted as a photosensitizer harvesting solar energy and producing electrons to drive the formation of methane in M. barkeri, and suggesting that such a photosynthetic coculture may provide an alternate strategy to replace semiconductors with photosynthetic bacteria in biohybrids.

Materials and methods

Bacterial strains and cultivation condition

R. palustris strain CGMCC 1.2180 was purchased from the China General Microbiological Culture Collection Center. M. barkeri MS (DSM 800) was purchased from the German Collection of Microorganisms and Cell Cultures. Strains of R. palustris and M. barkeri were routinely cultured in 0259 medium and CSM medium [34], respectively, at 35 °C under anaerobic conditions (N2:CO2, 80:20). In particular, R. palustris was cultured under illumination by LED light (15 W, with a wavelength of 590–592 nm and optical power density of 20 W m−2 measured by an optical power meter (CEL-NP2000, CEAULIGHT, Beijing)). When R. palustris and M. barkeri grew to their late exponential stages, cells were collected and washed three times using 0.9% NaCl solution by centrifuging at 6500 ×g for 15 min at 4 °C. The washed cells were used to initiate coculture by inoculating R. palustris and M. barkeri cells into 30 mL USM medium [34] without cysteine but supplied with 2 mM thiosulfate, achieving a final OD600 of 0.1, with thiosulfate as the only electron donor and CO2 as the sole electron acceptor. CH4 and H2 production by coculture were monitored as previously described using a gas chromatograph (Agilent 7890A, USA) with an HP-5 column and flame ionization detector [34]. The lowest detection limit for H2 is 50 ppm.

Operation of photosynthetic microbial fuel cells (photo-MFCs)

Two-chamber H-cells were employed to construct photosynthetic microbial fuel cells (photo-MFCs) as previously reported [35]. Briefly, the cell had a liquid volume of 25 mL and a headspace volume of 5 mL for each chamber. Graphite plates with a size of 1.5 cm × 1 cm × 0.5 cm were used as the electrodes. A proton exchange membrane (Nafion 117, DuPont Co., USA) was used to separate the anodic and cathodic chambers. An external resistance (1 MΩ) was applied between the anode and cathode, and the voltage across the resistance was recorded every 2 min by a data acquisition system (model 2700, Keithley Instruments, Ohio, USA), which was used to calculate the current density. USM medium (without cysteine) supplied with 2 mM thiosulfate served as the electrolyte. To initiate the reactions, 1 mL of R. palustris and M. barkeri cells (OD600 = 0.3) were inoculated into the anodic and cathodic chambers separately. The photo-MFCs were incubated at 30 °C under illumination by LED light, except the cathodic chamber containing M. barkeri was wrapped with aluminum foil to maintain darkness. The biofilms formed on anodic and cathodic electrodes were stained using a BacLight Live/Dead kit following the manufacturer’s instructions (Life Technologies, USA).

Electrochemical measurements

Linear sweep voltammetry (LSV) analysis of R. palustris and M. barkeri was performed using a single-chamber three-electrode cell as previously described [36]. A 4 cm × 5 cm indium tin-oxide glass was placed at the bottom and served as the working electrode. A platinum wire and Hg|Hg2Cl2 (sat. KCl) electrode were used as the counter and reference electrodes, respectively, and the electrolyte was USM medium (without cysteine) except supplying 2 mM thiosulfate for the R. palustris culture. LSV was conducted using an electrochemical workstation (660E, CH instruments, USA) and scanning over the range from −1 to 0.6 V at a rate of 1 mV s−1. Differential pulse voltammetry (DPV) analysis of the culture supernatant was conducted in the same setup. The parameters were Ei = −0.8 V, Ef = 0.3 V, pulse height 50 mV, pulse width 250 ms, step height 2 mV, step time 500 ms, and scan rate 5 mV s−1.

Physically separated culture cell setup

The physically separated culture cell was a modification of the photo-MFC setup with a semipermeable membrane (molecular weight cutoff of 1 kDa) replacing the proton exchange membrane and without electrodes (Supplementary Fig. S1). To start the operation of the cell, 0.5 mL of R. palustris and M. barkeri cells (OD600 = 0.3) were separately injected into the anodic and cathodic chambers. Both chambers were stirred continuously with stir bars.

Microscopy and fluorescent in situ hybridization

An optical microscope (Nikon Eclipse E200, Japan) operating in phase-contrast mode was employed to observe the cocultured cells grown at their logarithmic phase under a 100× objective lens. Fluorescent in situ hybridization was performed as previously described [37]. The probe 5’-[CY3]GCTGCCTCCCGTAGGAGT-3’ was selected for R. palustris, and the probe 5’-[CY5]GTGCTCCCCCGCCAATTCCT-3’ was used to label M. barkeri. Images were obtained using a confocal laser scanning microscope with a 40× objective lens (LSM880 NLO, ZEISS).

Quantification of R. palustris

The amount of R. palustris was absolutely quantified by quantitative PCR. Briefly, genomic DNA was extracted using TIANamp Bacteria DNA kit (Tiangen, Beijing, China). R. palustris-specific gene was amplified using the primer pair RpFor (5’-CTGGAAGTCTTGAGTATGGC-3’) and RpRev (5’-GCCTCAGCGTCAGTAATGGC-3’), and was cloned into the pMD19-T vector for a standard curve calculation. Quantitative PCR using the same primer pair was performed as previously described via a LightCycler 480 System (Roche Applied Science, Penzberg, Germany) [37].

Transcriptomic analysis

Cocultured cells at the early mid-log phase were collected by centrifugation at 5000 ×g for 15 min at 4 °C. Total RNA was extracted using TRIzol reagent (Invitrogen, California, USA) as previously described [38]. Directional libraries were prepared with the NEBNext® Ultra II™ Directional RNA Library Prep Kit (New England Biolabs Ltd., Beijing, China). mRNA was then sequenced by performing paired-end sequencing with the Illumina HiSeq/MiSeq paltformsOpen. All raw sequencing data were quality checked and filtered as previously described [39]. All reads matching 16S and 23S rRNA genes were removed, and the remaining reads were then used to map against the published genome of R. palustris CGMCC 1.2180 (NZ_CP058907.1) and M. barkeri MS (NZ_CP009528.1). Mapped reads were normalized with fragments per kilobase per million mapped reads (FPKM) as previously reported [39, 40].

Results

Light-driven methanogenesis in R. palustris and M. barkeri coculture

R. palustris and M. barkeri were cocultured in a defined coculture medium with thiosulfate as the sole electron donor, carbon dioxide as the only electron acceptor and light as the energy source. As illustrated in Fig. 1A, neither R. palustris nor M. barkeri could produce methane alone in the coculture medium. In contrast, their coculture generated methane steadily, approaching 357 ± 14 μM after 36 days (Fig. 1A). After continuous transfer, the methane generation rate increased, achieving 362 ± 11 μM after 9 days (Fig. 1B) in the sixth transfer with a calculated rate of 4.73 ± 0.23 μM/h, suggesting the formation of a stable methanogenic coculture. Furthermore, methane generation in the coculture was light-dependent, being activated by light and inhibited in darkness (Fig. 1C). R. palustris is in the purple nonsulfur group of bacteria that perform cyclic photophosphorylation and are incapable of methane formation [41]. Thus, R. palustris could not produce methane in the coculture medium. Instead, it performed as an anaerobic anoxygenic photoautotroph by oxidizing thiosulfate for CO2 fixation in the coculture medium (Supplementary Fig. S2A) [42]. In particular, under illumination, R. palustris could oxidize thiosulfate and transfer electrons extracellularly when an electron acceptor, such as an anode, was present (Supplementary Fig. S2B), as had been previously reported [32]. Furthermore, in the coculture, the presence of the extracellular electron acceptor in the form of M. barkeri promoted the photoautotrophy of R. palustris (Supplementary Fig. S3), probably by balancing the intracellular redox status, suggesting a syntrophic relationship between those two species.

Fig. 1: Methanogenesis in R. palustris (R.p) and M. barkeri (M.b) cocultures.
figure 1

Methane accumulation after initial inoculation (A) and at the sixth transfer after being transferred five times (B). C Methane production by R. palustris and M. barkeri coculture during continuous light-dark cycling.

Full size image

M. barkeri can use H2, acetate, and some one-carbon compounds to produce methane [43]. However, H2 should not be produced by R. palustris in the system, as H2 was shown to be the byproduct of biological nitrogen fixation in R. palustris, which would be inhibited in the presence of abundant ammonium in the culture medium [44]. As expected, no H2 was detected in the system (data not shown). Furthermore, neither acetate nor one-carbon compounds were generated by R. palustris as evidence by the fact that no methanogenesis was seen when M. barkeri was incubated in the filtrate of R. palustris monoculture spent medium (data not shown). Given that M. barkeri has been reported to take up electrons from electron-donating Geobacter species to reduce CO2 for methane production [33, 45], we propose that R. palustris acted as an electron donor in the light to drive the methanogenesis of M. barkeri in the coculture.

To establish the origin of the methane, 13C isotope labeling analysis was performed by replacing the 12C-NaHCO3 in the coculture medium with 13C-labeled NaHCO3, and the headspace gas was collected for gas chromatography–mass spectrometry analysis. As expected, 13C appeared in CH4, suggesting the direct CO2 to CH4 conversion in the coculture (Supplementary Fig. S4).

LSV analysis of R. palustris and M. barkeri demonstrated that R. palustris could transfer electrons outwards in a potential range of greater than −0.62 V (vs. SHE), while M. barkeri could accept electrons inwards in a potential range of less than 0.57 V. Thus, theoretically, electrons could transfer from R. palustris to M. barkeri in a potential window from −0.62 to 0.57 V (vs. SHE) (Fig. 3A) when applicable electrical conduits existed. These results suggest the possibility of electromethanogenesis of M. barkeri driven by electrons produced by R. palustris in the coculture. Actually, the two species formed aggregates (Fig. 2), a characteristic of electrosyntrophic coculture, in which microbial species are directly connected with each other providing an electron connection for direct interspecies electron transfer [37, 38].

Fig. 2: R. palustris and M. barkeri coculture aggregates.
figure 2

A Phase-contrast microscopy image and B fluorescence in situ hybridization (FISH) image. R. palustris was detected by a green probe, and M. barkeri was labeled by a red probe.

Full size image

Interspecies electron transfer between R. palustris and M. barkeri

To identify the process of electromethanogenesis in the coculture, R. palustris and M. barkeri were separately inoculated into each chamber of a MFC and then electrically connected by an external circuit as previously reported [35]. Both species formed biofilms on the electrodes (Supplementary Fig. S5), indicating direct electron transfer between either species and the electrode, as previously reported [35], and produced a maximum current of approximately 0.6 mA/m2 (Fig. 3B) with an accumulated CH4 production of 50.25 ± 3.32 μM after 30 days (Fig. 3C). As expected, this current production showed positive light activation (Fig. 3D). Notably, under darkness, a lower current was still produced (Fig. 3D), which was attributed to the consumption of stored energy by R. palustris since R. palustris could not oxidize thiosulfate by performing anaerobic chemotrophy under darkness (data not shown). In contrast, neither monocultures nor heat-killed cocultures produced current and methane (Fig. 3B, C) or displayed a current response to light (Fig. 3D). These results demonstrate that R. palustris acted as a photosensitizer, harvesting light energy and releasing electrons that M. barkeri could use for methane generation and suggesting direct interspecies electron transfer between those two species in coculture, since the proton exchange membrane excluded possible electron exchange for methanogenesis between the two species.

Fig. 3: Direct electrosyntrophic methanogenesis between R. palustris and M. barkeri.
figure 3

A Linear sweep voltammetry analysis. R. palustris generated anodic current in a potential range of greater than −0.62 V, and M. barkeri produced a cathodic current in a potential range of less than 0.57 V. B Current generation of photo-MFCs with an external resistance of 1 MΩ, showing direct electron transfer from the R. palustris anode to the M. barkeri cathode. The current generation of monocultures of R. palustris in the anode chamber and M. barkeri in the cathode chamber and of the two species in the separated chambers were measured. C Corresponding methane production in the photo-MFCs, showing that electrons produced by R. palustris could directly drive methanogenesis of M. barkeri. D Current generation of photo-MFCs during alternate light-dark cycles, showing light-driven electron transfer between R. palustris and M. barkeri. The test was performed after a steady current was generated.

Full size image

Microbial cells have been reported to secrete electron carriers or electron mediators to facilitate interspecies electron transfer [46, 47]. The possibility of meditated interspecies electron transfer between the two species was tested in a two-chamber cell separated by a semipermeable membrane that only allowed molecules with a molecular weight lower than 1 kDa to pass (Supplementary Fig. S1). Under these circumstances, these two species could not directly connect to transfer electrons. As indicated in Fig. 4A, 47.05 ± 7.36 μM of CH4 was produced by separated R. palustris and M. barkeri cultures in 35 days. In contrast, neither single species nor dead cells produced methane (Fig. 4A). The results demonstrate the possibility of indirect interspecies electron transfer facilitated by electron mediators between R. palustris and M. barkeri to support methanogenesis.

Fig. 4: Indirect interspecies electron transfer between R. palustris and M. barkeri.
figure 4

A Methane production by physically separated R. palustris and/or M. barkeri cultures, indicating the possibility of indirect electron transfer between these two species. B Baseline subtracted differential pulse voltammogram of the cell-free culture supernatant from physically separated culture cells, showing the presence of at least two redox-active compounds that could act as electron shuttles to facilitate electron transfer between R. palustris and M. barkeri. C Three-dimensional fluorescence spectrum of the culture supernatants from physically separated R. palustris and M. barkeri cultures. The two fluorescence peaks (I and II) correspond to the fluorescence of humic-like compounds.

Full size image

DPV analysis of cell-free culture medium was performed further to identify potential redox electron mediators. As shown in Fig. 4B, two redox compounds were detected with peak potentials at −0.28 and 0.01 V (vs. SHE) in the physically separated cultures, both of which lie in the potential window (−0.62 to 0.57 V vs. SHE), which would allow the electron transfer from R. palustris to M. barkeri. In contrast, no redox peaks were identified in the culture medium from monocultures of either R. palustris or M. barkeri or from the heat-killed separated cultures. However, the redox compound of 0.01 V should not contribute to the methanogenesis of M. barkeri since the potential is too high to drive the reduction of CO2 to methane (E0 of CO2/CH4 = −0.24 V) [48]. These results indicate that when separated, the methanogenic culture could produce redox compounds to mediate interspecies electron transfer. Coincidentally, two humic-like compounds (corresponding to fluorescence peaks I and II with excitation/emission of 270/435 and 340/430, respectively) were also identified by three-dimensional excitation–emission matrix fluorescence spectroscopy (Fig. 4C), but they were absent in the fluorescence spectrum of the plain culture medium and the culture media having grown with either species (Supplementary Fig. S6). Humic-like compounds have been reported to be redox active and can be secreted by microbial cells, facilitating extracellular electron transfer (EET) and mediating interspecies electron transfer [49, 50]. Therefore, redox-active compounds, such as humic-like substances, mediated interspecies electron transfer between physically separated R. palustris and M. barkeri. The same tests were also performed on the culture medium from aggregated coculture. Only a small redox peak with a potential at 0.01 V was detected in the differential pulse voltammogram (Supplementary Fig. S7), and a faint emission peak that was assigned to the 340/430 excitation/emission pair appeared (Supplementary Fig. S6), indicating that humic-like compounds were not highly secreted and did not significantly promote methanogenesis of the coculture. Therefore, direct interspecies electron transfer dominated in the aggregated coculture.

Electrosyntrophic CO2 reduction to CH4 by M. barkeri in coculture

Transcriptomic analyses were performed to dissect the light-driven electrosyntrophic methanogenesis in the coculture. The log2 FPKM value of each gene transcript was calculated and compared with the median (6.6 and 7.1 for R. palustris and M. barkeri, respectively) to evaluate gene expression (Supplementary Table S1). In R. palustris, sulfite oxidase (Sox) is responsible for thiosulfate oxidation, and ribulose-1,5-bisphosphate carboxylase/oxygenase (RubisCO) and enzymes involved in the Calvin–Benson–Bassham cycle are responsible for autotrophic CO2 fixation. The cyclic photophosphorylation pathway for energy generation was highly expressed (Fig. 5), consistent with R. palustris performing anoxygenic photoautotrophic growth by oxidizing thiosulfate under light in coculture. Previous reports have demonstrated that the membrane protein complex of PioAB and periplasmic protein PioC serve as part of the EET conduit to accept electrons from Fe(II) minerals and cathode in R. palustris [51, 52]. Surprisingly, the gene cluster pioABC was not highly expressed, suggesting that the pathways utilized for electron output by R. palustris were different from the one utilized for electron uptake. In contrast, flagella, pili, and some outer membrane c-type cytochromes were highly expressed (Fig. 5). The nanofilaments of flagella or pili in R. palustris have been reported to be conductive [31], and both conductive nanofilaments and cytochromes could participate in direct interspecies electron transfer [33, 37]. Therefore, the results are consistent with and suggestive of the direct electron interaction between R. palustris and M. barkeri and suggest that nanofilaments and cytochromes are involved in interspecies electron transfer in the coculture.

Fig. 5: The mechanism of light-driven CO2-to-CH4 conversion in R. palustris and M. barkeri coculture as revealed by metatranscriptomics.
figure 5

Genes involved in the processes include those that encode (1) Sox proteins, (2) proteins involved in cyclic photophosphorylation, (3) NADH-quinone oxidoreductase, (4) Ribulose-1,5-bisphosphate carboxylase/oxygenase, (5) and (23) ATP synthase, (6) enzymes participating in the Calvin–Benson–Bassham cycle, (7) PioABC, (8) nitrogenases, (9) c-type cytochromes, (10) flagellar proteins, (11) pilus proteins, (12) F420H2 dehydrogenases Fpo, (13) and (16) proteins involved in methanophenazine biosynthesis, (14) EchA-F hydrogenases, (15) MP-reducing hydrogenases VhtACG, (17) heterodisulfide reductases HdrDE, (18) heterodisulfide reductases HdrABC, (19) Formylmethanofuran dehydrogenases, (20) F420 hydrogenases, (21) Methenyl cyclohydrolases, (22) Methyl-CoM reductases, (24) methyltransferases MtrA-H, (25) Na+/H+ antiporter Nha, (26) acyl-CoA synthetases. Inset heat maps show the average transcript abundance of genes from triplicate independent cocultures presented as log2 FPKM values. The median log2 FPKM was 6.6 for R. palustris and 7.1 for M. barkeri. The blue and red lines indicate the speculated electron flux and CO2 fixation pathway, respectively, during light-driven methanogenesis in the coculture.

Full size image

Two membrane-bound hydrogenases, ferredoxin-dependent hydrogenase (Ech) and methanophenazine-dependent hydrogenase (Vht), and the membrane-bound heterodisulfide reductase complex (HdrDE) were highly expressed in M. barkeri (Fig. 5). Ech and Vht represent the energy-converting mechanism of M. barkeri by running intracellular H2 cycling to drive the reduction of methanophenazine (MP) [45, 53]. Thereafter, the redox cycling of MP/MPH2 provides electrons to the membrane-bound heterodisulfide reductase complex (HdrDE) for the reduction of CoM-S-S-CoB [45, 54]. In particular, the membrane-bound Ech complex was proposed to provide the interface to receive photoelectrons released by the semiconductor to drive CO2-to-CH4 conversion in the methanogenesis biohybrid system [34]. Therefore, membrane hydrogenase-dependent intracellular H2 cycling could provide an electron transfer mechanism for methanogenesis of M. barkeri in the coculture. In addition, both the proton-pumping F420-dehydrogenase (Fpo) and methanophenazine biosynthesis genes were highly expressed. In the absence of hydrogenase, the Fpo-dependent electron transfer pathway has been reported to support the growth of M. barkeri, providing a hydrogenase-independent electron transport pathway of direct electron uptake from extracellular electron sources [55, 56]. Under these circumstances, membrane-bound methanophenazine could contribute not only to the electrical conductivity of the membrane but also to MP/MPH2 cycling and thereafter mediate electron transfer from membrane-bound redox active protein to the Fpo-complex [57, 58]. Therefore, MP-Fpo could also serve as the electron uptake channel for M. barkeri in the coculture. Furthermore, Fpo could catalyze the generation of F420H2, which would participate in the electron bifurcation reaction catalyzed by the cytoplasmic heterodisulfide reductase complex (HdrABC) and contribute to the generation of reduced ferredoxin and the reduction of CoM-S-S-CoB, as indicated by the high expression of HdrABC. Consistent with these results, the CO2-to-CH4 conversion pathway was highly expressed, while the acetate degradation pathway was not (Fig. 5), which was also in agreement with electrosyntrophic CO2 reduction to CH4 by M. barkeri, as previously reported [39].

Discussions

Here we have provided the first proof that R. palustris can be used as a whole-cell photosensitizer to harvest light energy to drive the methanogenesis of M. barkeri. Under illumination, R. palustris and M. barkeri aggregated and formed an electrosyntrophic coculture in which R. palustris harvested light energy and performed photoautotrophy by oxidizing thiosulfate and transferring electrons extracellularly. Subsequently, M. barkeri accepted these electrons and used them for the reduction of CO2 to CH4. Furthermore, the electrons could be transferred between the two species either directly by cytochromes and conductive filaments or indirectly via electron shuttles in the form of humic-like substances. Such versatility of interspecies electron transfer may suggest that there will be a broad spectrum of electrosyntrophic partners for R. palustris.

The coculture system combines the strengths of both microbial species, which includes the highly efficient light-harvesting center in R. palustris and the highly specific methanogenic apparatus in M. barkeri, not only showing a higher thermodynamic favorability than hydrogen-dependent methanogenesis [59] but also having advantages over other light-driven methanogenic systems. For example, an abiotic artificial photosynthetic cell was designed employing the cathode catalyst and a platinum decorated semiconductive anode that harvested light energy for water splitting and provided electrons for CO2 to CH4 reduction on the cathode [60]. However, this system is not environmental and energy benign since precious metal and an external electric potential had to be applied in the system. In addition, even though the cathode catalyst has a relative high selectivity, without regarding its stability and persistence after long-time operation, its synthetic process is complex and harsh needing a high temperature of 800 °C. Similarly, in inorganic-biological hybrid systems, conventional semiconductor photosensitizers, such as the widely used CdS, demonstrate inherent defects in terms of photocorrosion, photodegradation, and toxicity, limiting their sustainability. In contrast, in the coculture system, the living photosensitizer of R. palustris confers sustainability, biocompatibility, self-replication, and self-repair. Moreover, the coculture produced methane at a highest rate of 4.73 ± 0.23 μM/h, which is higher than all reported biohybrid methanogenesis systems (3.46 µM/h for CdS@M. barkeri and approximately 3.81–4.36 µM/h for CdS:Ni@M. barkeri) [34, 61] and even higher than the engineered R. palustris methanogenic mutant (approximately 3 µM/h) [62], indicating effectiveness and efficiency. Notably, sacrificial organic reagents, such as alcohol and cysteine, are necessary to act as ultimate electron donors in biohybrid systems. In contrast, the inorganic substance thiosulfate donated electrons in our methanogenic coculture, which provided electrons for CO2 fixation to generate organic matter in R. palustris and for CO2 reduction to CH4 in M. barkeri, suggesting zero CO2 emissions in the R. palustris and M. barkeri electrosyntrophic coculture. Furthermore, previous attempts have employed R. palustris as a photosensitizer to harvest solar energy for current generation [32]. However, in our study, solar energy was stored as chemical energy in biomass and methane, representing a more efficient energy storage strategy.

Our study suggested the proof-of-concept of using living photosensitizers of anoxygenic photosynthetic bacteria to drive nonphotosynthetic microbial CO2 fixation. While thiosulfate was used as the electron donor in our methanogenic coculture, it is possible to use other electron donors, such as the universal electron donors H2 and Fe2+ minerals [63], or other low-value organics such as acetate or glycerol [32] to perform EET under light. Meanwhile, some nonphotosynthetic autotrophic bacteria, such as acetogens like Clostridium ljungdahlii or Moorella thermoacetica [64], have been shown to use extracellular electrons for CO2 reduction. Actually, those autotrophic bacteria being the workhorse of synthetic biology have shown great promise to act as platforms for CO2 being used in the production of tailored value-added compounds in microbial electrosynthesis and have been used in constructing biohybrids [15]. Thus, it can be predicted that R. palustris should be capable of serving as a living photosensitizer to drive CO2 conversion to value-added products, not limited to CH4, in those autotrophic bacteria via the formation of the electrosyntrophic coculture. Specially, photosynthetic organisms, such as algae and plant, have been shown to achieve direct light-harvesting and CO2 reduction to conversion to value-added compounds [65]. However, model prediction indicated that algal and cellulosic bioenergy programs aggravate global warming caused by nitrous oxide arising from the extensive use of fertilizers [65, 66]. Furthermore, thermodynamic comparison indicated that the carbon fixation pathway in those photosynthetic organisms is often less energetically favorable [29], which is in contrast to versatile autotrophic carbon fixation pathways with cheaper energy-consumption in nonphotosynthetic autotrophic bacteria. Notably, in the coculture system, solar energy and extracellular electrons were just partially flowed into those value-added products since they also supported the photoautotrophic growth of R. palustris in the electrosyntrophic coculture. Modulating the electron flow toward EET in R. palustris should increase the productivity of the coculture system.

The formation of an electrosyntrophic interaction benefits both species in the coculture. R. palustris is a metabolically versatile microbe that uses a wide electron donor spectrum ranging from inorganic substances to complex organic matter. In contrast, M. barkeri is only able to use several one- to two-carbon compounds or hydrogen. Considering that the presence of the electron acceptor strain M. barkeri promoted the anoxygenic photosynthesis of R. palustris, the electrosyntrophic interaction could improve the resilience of both species in nature. Actually, in some natural methanogenesis hotspots, such as paddy soil [67], wetland [68], and coastal zones [69], anaerobic photosynthetic bacteria prevail. It can be speculated that the solar light-driven methanogenesis between anaerobic photosynthetic bacteria and methanogens should also prevail in these environments, contributing to global carbon cycling. Repeatedly, the emergence of EET has been shown to open up a wide variety of new metabolic “windows” that simply were not imagined until a few years ago. In particular, EET between different microbial species has expanded the range of electron donor or acceptor being possible from one species to another species, contributing the formation of highly cooperated microbial communities with specific functions. Recently, electrosyntrophy has been discovered in various microbial communities, from bacteria [37] to archaea [70] and even in some conventional hydrogen syntrophic communities [71]. Our study has further extended the family of electrosyntrophic species and we are confident will inspire the discovery of electrosyntrophy in other microbial communities.