Introduction

Communication, the sending and receiving of information, is ubiquitous in natural biological system1,2. The critical biological information is typically transmitted as diffusion gradients of ions and small molecules during the communication process and plays vital roles in dynamically driving coordinating activities1,3. Microbial communication, prevalent among biological systems, has raised great attention owing to its potential in dynamically modifying microbial activities for a purpose in various areas such as biomanufacturing4,5,6. Specifically, microbial consortium can dynamically adjust the gene expression upon dynamically sensing and dealing with the external signal information through the cell-to-cell communication, thus globally redistributing the cellular resources and coordinate the microbial metabolism processes7,8,9. Consequently, it is of vital significance to have a deep understanding of microbial communication network, thus guiding the microbial metabolism coordination and optimization, offering a promising way for driving microbial industries and better understand living systems.

Currently, several biological communication strategies have emerged including quorum sensing, ion channel, redox-based molecular communication and etc1,10. Among the different communication modes, redox-based molecular communication has attracted increasing attention owing to its distinct modality for biological communication11. To be specific, molecular redox state can be switched dynamically and reversibly through electron transfer in the redox-based communication process, consequently providing the driving force for dynamically regulation of microbial metabolism12. In addition, the activity of signaling enzymes with redox-active functional groups can also be dynamically altered owing to the varied redox potential that is associated with the ratio of oxidized/reduced redox species1,13,14. Analogous to the communication network in electronic information technologies, redox-based molecular communication system also has the ability to sense, analyze and process signal information. Hence, constructing microbial redox-based communication network provides an expedient tool for intelligent self-adaptive regulation, enabling the dynamic resource reallocation and high flux in microbial biomanufacturing practice of value-added chemicals1,15,16.

The Fe3+/Fe2+ redox couple, which widely exists in nature, is critical in maintaining metabolic processes17,18. Here, we construct a microbial communication network model consisted of three building blocks including signal router, optical verifier and bio-actuator based on microbial electron transfer triggered Fe3+/Fe2+ redox reaction (Fig. 1). With the ability to sense, analyze and process the dynamic and reversible Fe redox species, the microbial communication network system is capable of coordinating the metabolism behavior of bio-actuator, leading to an increased pathway flux and improved biomanufacturing efficiency. Specifically, Fe3+ redox signal can be sensed and transduced by Shewanella putrefaciens CN32 (S. putrefaciens) signal router to Fe2+ with the metal-reducing (Mtr) pathway19. Rhodopseudomonas palustris TIE-1 (R. palustris), the bio-actuator that is capable to receive and utilize the electrons of the transduced Fe2+, has dual functions in Fe2+ output signal retranslating with the iron oxidation pathway and lycopene manufacturing with the mevalonate (MVA) and methylerythritol 4-phosphate (MEP) pathway20,21. In addition, optical verifier with high spatial and temporal resolution is introduced into the communication model, providing real-time and dynamic monitoring for the fluctuant Fe3+/Fe2+ redox signal22. We then capture the concept of biological local area network (LAN) which consists of “router” “actuator” and “verifier” as demonstrated in reference1, and apply the same methodology to a Fe signal based system that has these benefits. In this work, we find that electron transfer related genes are highly expressed in the biological LAN with redox communication. Actually, electrons from Fe2+ redox signal can be transferred into R. palustris bio-actuator cells with iron oxidation pathway through the electron transfer related proteins (Supplementary Fig. 4)19, channeling reducing power directionally into R. palustris bio-actuator, thus boosting the lycopene biomanufacturing and CO2 fixation efficiency. Further, we demonstrate that biological LAN with redox communication can both coordinate and regulate metabolism behaviors in microbial co-culture (S. putrefaciens-R. palustris) and microbial consortia (S. putrefaciens-Geobacter soli-R. palustris) system. Compared with the previous work on redox-based communication1,13, the electron transfer triggered redox communication strategy proposed in this study was independent of external electronics input and the intrinsic electron transfer triggered redox cycle could function as batteries that support microbial metabolism20. The redox-based communication strategy in this work may pave ways for programming biological systems with dynamic functions or behaviors and open doors for efficient biomanufacturing.

Fig. 1: Biological local area network (LAN) with dynamic and reversible redox signal transduction through electron transfer.
figure 1

The biological LAN consists of three building blocks including signal router, optical verifier and bio-actuator based on microbial electron transfer triggered Fe3+/Fe2+ redox reaction.

Results

Biological local area network based on redox communication

Actually, the established redox-based communication network above can serve as a compact biological local area network (LAN). Parallel to the electronic LAN, the redox-based communication network has intrinsic advantages including signal propagation, division of labor across populations and spatial organization1. Three components of “router” “actuator” and “verifier” in the biological LAN are responsible for the signal interception, relay and interpretation respectively. Specifically, signal router is hardwired to the active redox species via “electron transfer wire”, enabling the interception of Fe3+ redox signal and transduction into Fe2+. Subsequently, Fe2+ redox signal can be further transmitted toward the remaining bio-actuator and verifier, thus realizing the signal relay and interpretation independently.

Fe3+/Fe2+ couple, as one of the most common redox species in nature, plays crucial roles in geochemical cycle and life evolution through electron transfer triggered redox state switching18,19. S. putrefaciens is one of the earliest metal-reducing bacteria and is competent to transduce Fe3+ redox species into Fe2+ signal (Supplementary Fig. 1)19. The Mtr pathway in S. putrefaciens signal router can be termed as the molecular “central processing unit” (CPU) that is in charge of the redox signal operation and control17. By instinct, S. putrefaciens signal router transfers electrons to external Fe3+ species through the molecular “CPU” consisting of electron transfer proteins such as MtrC and CymA, accomplishing the redox signal transduction22 (Fig. 2a). To be specific, Mtrc, OmcA and OmcB locate on the bacteria surface and could transfer electrons directly22, and CymA is a key membrane-anchor protein that could transfer electrons from cytoplasm to periplasm in S. putrefaciens22. When Fe3+ incubated with the S. putrefaciens signal router, Fe2+ concentration increased over time (Fig. 2b), suggesting the successful redox signal transduction process. Further, mutants with MtrC-encoding gene deleted (ΔMtrC) and CymA-encoding gene deleted (ΔCymA) were constructed to investigate their roles in redox signal transduction process. As shown in Fig. 2c, it can be observed that Fe2+ signal transduction efficiency decreases about 1.61 and 1.69 times in ΔMtrC and ΔCymA mutant systems respectively compared with wild-type strain. While, after the addition of electron shuttle humic acid (HA) in S. putrefaciens system23, Fe2+ signal transduction efficiency increased 1.23 times (Supplementary Fig. 2). The above results indicate that the redox signal transduction of S. putrefaciens signal router is closely related with electron transfer processes.

Fig. 2: Biological LAN with signal router, optical verifier and bio-actuator.
figure 2

a Schematic illustration of Fe signal transduction with Mtr pathway in S. putrefaciens. b Fe2+ concentration in bare S. putrefaciens system at different incubation time with different initial Fe3+ addition. c Fe2+ concentration in ΔMtrC, ΔCymA mutants and S. putrefaciens system cultured at 48 h. Data presented as mean values ± SD., n = 3. d Schematic illustration of R. palustris with dual actuating functions including signal reconversion and chemical biosynthesis. e Fe2+ concentration in S. putrefaciens-R. palustris co-culture system at different incubation time with different initial Fe3+ addition. f Lycopene biosynthesis yield of R. palustris incubated at 96 h with or without Fe redox communication represented by “+” and “–” respectively. Data presented as mean values ± SD., n = 3. g Schematic illustration of the biological LAN. h The recorded persistent luminescence intensity of ZGO:Mn verifier and the absorbance measured with o-Phenanthroline in the biological LAN. P = 0.002, P values were determined by a two-tailed test. Data presented as mean values ± SD., n = 3. i Time course of model-predicted Fe2+ concentration and OD600 in the biological LAN. Experimental data are indicated by data points and fitted models of Fe2+ concentration (red) and OD600 (blue) are shown as solid lines. P = 0.02, P values were determined by a two-tailed test. Data presented as mean values ± SD., n = 3. Source data are provided as a Source Data file.

Next, the transduced Fe2+ signal can be further received by bio-actuator and verifier in the biological LAN, realizing the signal relay and interpretation. To be specific, electrons carried by Fe2+ signal can be channeled through PioABC complex that is directly involved in taking up extracellular electrons for photoferrotrophic and photoautorophic growth of R. palustris bio-actuator (Supplementary Fig. 3)19. Meanwhile, when R. palustris ingests the electrons, Fe2+ signal is reconverted back to Fe3+. In addition, R. palustris, reported as a microbial host for terpenoid production through the MVA and MEP pathways, possesses potential for high value-added chemical production (Supplementary Fig. 4)21,24. Overall, R. palustris in the constructed compact biological LAN possess dual actuating functions including signal reconversion and chemical biosynthesis (Fig. 2d). To investigate the functions of the R. palustris bio-actuator, we measured the Fe2+ concentrations and the biosynthetic yields throughout the incubation period. As shown in Fig. 2e, Fe2+ concentration first increases and then decreases over incubation time, mainly owing to the Fe2+ signal reconversion process activated by R. palustris bio-actuator (Supplementary Fig. 5). Moreover, the lycopene biosynthetic yield of R. palustris bio-actuator with Fe redox communication is about 1.12 times higher than the bare co-culture system without Fe redox communication, suggesting that the electron transfer triggered redox communication enhances the actuating function of biosynthesis (Fig. 2f, Supplementary Fig. 6). Considering that the reduced nicotinamide adenine dinucleotide phosphate (NADPH) is critical in biosynthetic pathways serving as the electron carrier for a large subset of oxidoreductases25, the increased lycopene yield is partially due to the enhanced NADPH in R. palustris by receiving electrons from the dynamic Fe3+/Fe2+ cyclic communication process through the iron oxidation pathway.(Supplementary Fig. 7)19. The Fe redox signal inducible verifier (Zn2GeO4:Mn) introduced into the biological LAN can not only confirm and track the relayed redox information, but also guide the optimization of biosynthesis process (Fig. 2g, Supplementary Fig. 8)22. Specifically, the luminescence of Zn2GeO4:Mn (ZGO:Mn) at 536 nm originates from the 4 T1(G) → 6A1(S) transition of Mn2+ luminescence center22. Owing to the reduction Fermi energy level locations of EFe3+,red and EFe2+,red, the nano verifier shows different luminescence responses toward Fe3+ and Fe2+, thus achieving a dynamic and reversal monitoring of the Fe redox signal conversion in real-time (Supplementary Fig. 9)22. In addition, the nano verifier exhibits low toxicity both against S. putrefaciens and R. palustris according to quantitative colony-forming unit (CFU) survival assays (Supplementary Fig. 10). To investigate the performance of the verifier in Fe redox signal monitoring in the biological LAN, we performed the correlation analysis between the luminescence intensity measured with the verifier method and the absorbance measured with the typical phenanthroline spectrophotometric method26. As shown in Fig. 2h, the luminescence intensity of verifier is highly correlated with the measured absorbance (R = 0.99, P = 0.002), further proving that verifier can provide an accurate and dynamic luminescent monitoring of the Fe redox signal. Furthermore, based on the recorded luminescence intensity of the verifier (Supplementary Fig. 11) and the linear relationship between logarithm of persistent luminescence intensity and Fe2+/Fetotal ratio (Supplementary Fig. 9d), the concentrations of Fe3+/Fe2+ redox couple were calculated. With the calculated values, we then simulated the kinetic transformation of the Fe3+/Fe2+ redox signal according to the equations (Supplemental methods) with MATLAB software1,27. The simulation results in Fig. 2i show that Fe3+ signal is transformed to Fe2+ first and the concentration of Fe2+ peaks around 0.76 mM, then the Fe2+ signal decreased gradually. The simulation results of Fe2+ concentration and OD600 in the biological LAN are demonstrated to be highly correlated with the experimental data (R = 0.90, P = 0.02 for Fe2+ concentration and R = 0.99, P < 10–4 for OD600), indicating that the simulation models are competent to assess or predict behaviors in the biological LAN. Overall, according to the above results, the constructed biological LAN with the ability to sense, analyze and process Fe redox signal demonstrate the achievement of social division of labor.

Redox communication coordinates microbial metabolism

The long-term evolution of the microbial communities enables cells to sense and response to the extracellular environment28. According to the varied environments, microbial communities are capable to dynamically allocate cellular resources and optimally control pathway expressions, thus maintaining the stability of the microbial community29,30. Similarly, microbial community behavior can be self-adapted in response to the extracellular dynamic switching redox state. The confocal images in Fig. 3a show rod-shaped morphologies of S. putrefaciens signal router (green) and R. palustris bio-actuator (red) labeled by fluorescent in situ hybridization (FISH) probes31,32. To explore the effects of Fe redox communication on microbial function and physiology, the functional potential of the microbial community based on 16 S rRNA gene sequencing were predicted with PICRUSt2. As shown in Fig. 3b, the abundance of major function groups involved in microbial growth such as amino acids metabolism and lipid metabolism is higher, while the abundance of genes involved in energy metabolism is lower in the biological LAN with redox communication compared with the bare co-culture system. The results suggest that the introduction of Fe redox signal into the biological LAN leads to enhanced growth metabolism and reduced energy metabolism16. This may be due to the syntrophic cross-feeding of electrons between S. putrefaciens signal router and R. palustris bio-actuator mediated with the Fe redox signal, saving the energetic cost and increasing the excretion of critical reduced compounds, resulting in enhanced physiological functions (Supplementary Fig. 12)33,34.

Fig. 3: Microbial behaviors in the biological LAN with redox communication.
figure 3

a Confocal fluorescence imaging of S. putrefaciens (FAM, green) and R. palustris (Cy5, red) in the biological LAN labelled with the FISH probes. Scar bar is 10 μm. At least three independent experiments were carried out with similar results. b Gene function prediction of S. putrefaciens-R. palustris co-culture in the biological LAN by PICRUSt2. The middle value represents the mean differences between the groups, and the error bar represents the 95% confidence intervals. c–e Log2 (NADH/NAD+) (c), Log2 (NADPH/NADP+) (d) ratio and relative ATP concentration (e) of S. putrefaciens-R. palustris co-culture in the biological LAN incubated at 96 h. Data presented as mean values ± SD, n = 3. f Relative abundance of S. putrefaciens and R. palustris over incubation time in the biological LAN. g Calculated OD600 of R. palustris in the biological LAN. Data presented as mean values ± SD, n = 3. Source data are provided as a Source Data file.

To validate the above hypothesis, the critical reduced compounds and energy rich compounds including NADPH, NADH and adenosine 5′-triphosphate (ATP) were tested. It was found that, the reducing compounds of NADPH and NADH accumulated correspondingly with the simultaneous generation of ATP in the biological LAN with redox communication (Fig. 3c–e), indicating that the electron transfer triggered redox communication could provide efficient reducing power and energy to the microbial community. In addition, it can be observed that the population composition differs between the biological LAN with redox communication and the bare co-culture system. As shown in Fig. 3f, the proportions of R. palustris bio-actuator at different incubation times in the biological LAN with redox communication are higher compared with the bare co-culture system (Supplementary Fig. 13). The OD600 values of R. palustris bio-actuator are about 2.0-fold higher compared with the values in bare co-culture system across time (Fig. 3g), while the OD600 values of S. putrefaciens are similar with the values in bare co-culture system across time (Supplementary Fig. 14). The above results indicate that the redox communication process both modulates the metabolism behaviors and the population compositions of the microbial community.

Redox communication regulates the gene expression of electron transfer related proteins

To have a further understanding of the redox communication mechanism in the biological LAN, transcriptomic analysis of S. putrefaciens signal router and R. palustris bio-actuator was performed35,36. Differentially expressed genes were evaluated according to the Log2 fold change (Log2 FC), which was calculated with {Log2 [(Gene read counts)(Experimental)] – Log2 [(Gene read counts)(Control)]}. Overrepresented (white to purple for S. putrefaciens and white to red for R. palustris) and underrepresented transcript (white to green for S. putrefaciens and white to blue for R. palustris) with the reference of gene expressions in bare co-culture shown as Log2 FC were exhibited in Fig. 4a. It can be observed that the Log2 FC of electron transfer proteins including flagellum, pilus and c-type cytochromes (c-Cyts) and PioABC complex are >0 on the whole, indicating that genes encoding electron transfer proteins were actively expressed in the biological LAN with redox communication. In addition, genes associated with reducing power substance also demonstrated high expression in the biological LAN with redox communication. The highly expressed genes typically represent the highly express of related proteins. Accordingly, it is suggested that the redox communication might improve the functions of electron transfer proteins, thus fascinating the electron transfer process between S. putrefaciens signal router and R. palustris bio-actuator. To validate the increased electron transfer function, current density of cells was measured with gold interdigitated microelectrode (Supplemental methods). The current density results demonstrate that cells in the biological LAN with redox communication have about 2.10 times higher current density compared with cells without redox communication, suggesting the improved electron transfer capability (Supplementary Fig. 15). Besides, Fig. 4a also shows that the Log2 FC of proteins associated with both the MEP & MVA pathways and CO2 fixation are above 0 generally, suggesting that the redox communication triggered with electron transfer facilitates the biosynthesis and CO2 fixation processes in microorganism (Supplementary Fig. 16).

Fig. 4: Differential gene expression analysis of S. putrefaciens and R. palustris in the biological LAN.
figure 4

a Schematic illustration of electron cross feeding between S. putrefaciens and R. palustris in the biological LAN with Fe redox communication. Inset heat maps show the differential gene expression presented as Log2 FC value. b A Volcano plot of differentially expressed genes in R. palustris. The threshold of Log2 FC is |0.58| (i.e., FC ≥ |1.5|), and that of P value is <0.05. There were 169 differentially expressed microbial transcripts that met these criteria. c,d Violin plots showing the significant differences in expression of (c) electron transfer related genes (including pilus, c-Cyts, flagellum, PioABC, flavin) and (d) reducing equivalent related genes of R. palustris (including NAD(P)H related proteins). P values were determined by an unpaired two-sided t-test. Source data are provided as a Source Data file.

Specifically, of 4562 transcripts detected in R. palustris, 169 were significantly associated with the redox communication (P value < 0.05 and FC ≥ |1.5 | ), indicating that the redox communication network influences the gene expressions of microorganism (Fig. 4b). The conclusion was further validated according to differentially expressed genes analysis in S. putrefaciens (Supplementary Fig. 17). Subsequently, by assessing the mRNA expressions of R. palustris and S. putrefaciens, it was found that the electron transfer and reducing power associated genes had higher expression level represented as Log2 FPKM value in the biological LAN with redox communication compared with the bare co-culture (Fig. 4c, d, Supplementary Fig. 18). The above results demonstrate that the Fe signal transduction in biological LAN could upregulate gene expressions of electron transfer and reducing power generation related proteins, thus regulating microbial metabolic behaviors.

Redox communication facilitates the biomanufacturing efficiency of R. palustris bio-actuator

The essential of the redox state dynamic conversion is the circulation flow of electrons. In the biological LAN with redox communication, the dynamic conversion of Fe3+/Fe2+ redox couple is able to channel electrons directly to R. palustris bio-actuator through the iron oxidation pathway (Supplementary Fig. 19)19,20,37. Specifically, PioA and PioB are proposed to oxidize Fe2+ extracellularly and then transfer the released electrons across the outer membrane to PioC, which is hypothesized to be in the periplasm19. Previous studies have proved that the oriented electron transfer is critical in promoting reducing power NADPH for high biomanufacturing efficiency16,38,39. Hence, we sought to investigate the influence of directed electron flow on lycopene biosynthesis yields of R. palustris bio-actuator in our constructed biological LAN.

Carbonyl cyanide m-chlorophenyl hydrazone (CCCP) is a lipid-soluble molecule and has been reported to uncouple the electron transfer from ATP synthesis (Supplementary Fig. 20). Rotenone, previously reported as an inhibitor for NADH dehydrogenase, is capable to block electron transfer from the iron-sulfur clusters in NADH dehydrogenase to ubiquinone (Supplementary Fig. 21)40. Both CCCP and rotenone are recognized to inhibit the electron transfer process in R. palustris bio-actuator40. It is observed that upon CCCP and rotenone treatment, the lycopene biosynthesis yields of R. palustris bio-actuator in the biological LAN with redox communication decreased 40.64% and 40.70% respectively compared with the untreated controls (Fig. 5a), demonstrating that the hindered electron transfer processes decrease the biosynthesis efficiency of R. palustris bio-actuator. Furthermore, upon the addition of electron shuttle of humic acid (HA) and anthraquinone-2,6-disulfonate (AQDS), the biosynthesis yields increase 54.05% and 88.86% respectively in the redox-based communication network, suggesting that accelerated electron transfer processes increase the biosynthesis efficiency (Fig. 5b). Similarly, the biosynthesis efficiency can be obviously improved through the addition of electron donors for microbial respiration including acetate and lactate20,41 (Supplementary Fig. 22). The observations intriguingly suggested that the electron transfer process may indeed be involved in regulating biosynthesis metabolism in R. palustris bio-actuator.

Fig. 5: Biomanufacturing function of R. palustris in the biological LAN.
figure 5

a Lycopene yields of R. palustris in the Fe redox based biological LAN with or without the addition of electron transfer related protein inhibitor. Data presented as mean values ± SD, n = 3. b Lycopene yields of R. palustris in the Fe redox based biological LAN with or without the addition of electron transfer shuttle. Data presented as mean values ± SD, n = 3. c,d Log2 (NADH/NAD+) (c) and Log2 (NADPH/NADP+) (d) of the isolated R. palustris from the biological LAN. Data presented as mean values ± SD, n = 3. e Schematic illustration of lycopene biomanufacturing process of R. palustris in the Fe redox based biological LAN. f CO2 fixing rate of R. palustris in the Fe redox based biological LAN. Data presented as mean values ± SD, n = 3. Box edges indicate the upper and lower quartiles, the centerlines indicate the mean value, the horizontal bars indicate the maximum and minimum values. g Schematic illustration of CO2 fixation of R. palustris in the Fe redox based biological LAN. Source data are provided as a Source Data file.

To validate the relationship between electron transfer and biosynthesis metabolism in R. palustris bio-actuator, fluorescence-activated cell sorting (FACS) was performed to isolate the R. palustris bio-actuator from S. putrefaciens-R. palustris co-culture for further metabolism investigation. As shown in Fig. 5c, d, the values of Log2 ratio of NADH/NAD+ (2.94 ± 0.24) and NADPH/NADP+ (3.71 ± 0.62) in R. palustris bio-actuator with Fe redox communication are higher than the values without Fe redox communication, largely due to the acceptance of external electrons of Fe2+ with the iron oxidation pathway in R. palustris22,25. Accordingly, as schemed in Fig. 5e, electrons can be supplied continuously and delivered specifically into R. palustris bio-actuator through the dynamic and reversible Fe3+/Fe2+ signal transduction process in the redox-based communication network, allowing to enhance the generation of reducing power NADPH. It is widely recognized that NADPH acts as critical electron carrier for a large subset of oxidoreductases to drive the microbial biosynthesis metabolism25. Hence, the lycopene biosynthesis yields can be improved through the MVA and MEP pathway in R. palustris with the continuous supply of electrons. Subsequently, fed-batch fermentation at laboratory scale was further performed to explore the applicability of the proposed redox based communication strategy. It is inspiring to find that the lycopene biosynthesis increased about 1.43 times in fed-batch fermentation compared with the shake-flask fermentation (Supplementary Fig. 23, Supplementary Methods). The above results indicated the potential application prospects of the proposed redox based communication strategy in practical lycopene biomanufacturing.

R. palustris bio-actuator is a typical photosynthetic bacterium in which CO2 can be fixed through the Calvin-Benson-Bassham (CBB) cycle. CO2 fixation process requires the investment of reducing power36,42,43. The constructed redox-based communication network has demonstrated to be able to supply electrons and subsequently accelerate reducing power in R. palustris. The possibility that redox communication process could drive the reducing power generation and therefore improving the efficiency of CO2 fixation was further investigated. As shown in Fig. 5f, the CO2 fixation rate of R. palustris in redox-based communication network reaches 5.55 Mm g–1 DCW–1 h–1, 1.30 times higher compared with the rate in bare co-culture, validating that the redox communication network improves efficiency of CO2 fixation in R. palustris. The enhanced CO2 fixation rate might be due to the directed electron transfer provided by the dynamic Fe3+/Fe2+ cyclic process in the redox communication network, affording the critical reducing power for CBB cycle (Supplementary Fig. 24).

Redox communication in microbial consortia

Microbial consortia are highly desired for efficient biomanufacturing with the distinct advantages of alleviating metabolic burden by division of labor through trading metabolites or exchanging signals and has recently attracts considerable interest44,45. The applicability of the redox communication strategy was further investigated in microbial consortia. As schemed in Fig. 6a, a Fe redox based biological LAN consisted of dual-band router, optical verifier and bio-actuator was constructed (Supplementary Fig. 25). Similar with S. putrefaciens, Geobacter soli (G. Soli) is also an iron-reducing bacterium that can convert Fe3+ redox signals into Fe2+. The dual-band router composed of S. putrefaciens and G. Soli can ensure the stably Fe2+ signal output according to their respective metabolic state, avoiding the occurrence of “unstable network” or “disconnected” situations46.

Fig. 6: Redox communication in microbial consortia.
figure 6

a Schematic illustration of the biological LAN consisted of dual-band router, optical verifier and bio-actuator. b The persistent luminescence of verifier in the microbial consortia. c Relative abundance of bacteria in the microbial consortia. Data presented as mean values ± SD, n = 3. d Relative lycopene biomanufacturing yield of R. palustris bio-actuator in the microbial consortia with or without Fe redox communication. Data presented as mean values ± SD, n = 3. e Relative CO2 fixing rate of R. palustris bio-actuator in the microbial consortia with or without Fe redox communication. Data presented as mean values ± SD, n = 3. f,g Log2 (NADH/NAD+) (f) and Log2 (NADPH/NADP+) (g) ratio of the isolated R. palustris from the microbial consortia. Data presented as mean values ± SD, n = 3. Source data are provided as a Source Data file.

As shown in Fig. 6b, the verifier exhibited increased luminescence intensity in S. putrefaciens and G. Soli co-culture compared with the bare S. putrefaciens and bare G. Soli, suggesting that the dual-band router is competent for efficient Fe2+ signal output (Supplementary Fig. 26). In addition, with Fe redox communication, the lycopene biosynthesis yield and CO2 fixation rate of R. palustris bio-actuator is 1.55 and 1.48 times higher respectively compared with the bare microbial consortia (Fig. 6d, e). The promoted biosynthesis and CO2 fixation efficiency is largely due to the directed and specific electron transfer from Fe2+ to R. palustris bio-actuator, leading to the accumulation of critical reducing power for metabolism (Fig. 6f, g). The above results demonstrate that the electron transfer triggered redox communication strategy is also capable to improve the biosynthesis and biological CO2 fixation efficiency in microbial consortia. The designed electron transfer triggered redox communication network proposes a general strategy for microbial metabolism coordination and regulation, offering insights into reducing power regeneration and metabolic optimization.

Discussion

Microbial communication, capable to drive coordinated functions through sensing, analyzing and processing signal information, plays critical roles in biomanufacturing and life evolution, and has attracted considerable recent interest. Redox-based communication provides an expedient tool for intelligent self-adaptive regulation and is promising in metabolism coordination. Much pioneering work has been performed on redox-based communication for guiding biological function1,13. For example, Bentley and the co-workers have long engaged in bio-electronic control of microbial metabolism with electrogenetic device that uses redox biomolecules to carry electronic information to engineered bacterial cells1,13. From another perspective, this work proposed a redox-based communication network strategy that was independent of external electronics input and the intrinsic electron transfer triggered redox cycle could function as batteries that facilitate the biosynthesis processes. Our study provides fundamental insights into the redox communication between microbial species and revealed the metabolism coordination mechanism. A biological local area network model consisted of three building blocks including signal routing cell (S. putrefaciens), optical verifier and bio-actuation cell (R. palustris) based on electron transfer triggered Fe redox communication strategy was constructed as a proof of concept. It is demonstrated that the redox communication network drives gene expressions of electron transfer proteins and simultaneously facilitates the critical reducing power regeneration in “bio-actuator cells”, enabling microbial metabolism regulation. In this way, the biomanufacturing yield of lycopene in “bio-actuator cells” increased with the mevalonate and methylerythritol 4-phosphate pathway. In addition, our work demonstrated that the proposed redox communication strategy is capable to facilitate CO2 fixation and is applicable both in co-culture and microbial consortia. The demonstration of microbial behavior coordination with the electron transfer triggered redox communication strategy represents a critical step to construct robust metabolism regulation platform for engineering microbial metabolic relevant activities. Looking forward, with the development of synthetic biology, we believe that the proposed redox communication strategy may pave ways for self-adaption and dynamic microbial metabolism regulation and is promising in facilitating the development of mixed fermentation industries with the distinct advantages of alleviating metabolic burden by division of labor.

Methods

Materials

Germanium oxide (GeO2, 99.99%), zinc nitrate hexahydrate (Zn(NO3)2·6H2O), manganese nitrate (Mn(NO3)2), manganese powder, manganese(II) chloride (Mn(Cl)2), manganese dioxide (MnO2), iron (III) chloride hexahydrate (FeCl3·6H2O), iron(II) chloride (FeCl2), O-Phenanthroline were purchased from Sinopharm Chemical Reagent Co., Ltd. Concentrated nitric acid (HNO3), concentrated sulfuric acid (H2SO4), sodium hydroxide (NaOH), lactate, sodium phosphate dibasic heptahydrate (Na2HPO4·7H2O), potassium dihydrogen phosphate (KH2PO4), sodium chloride (NaCl), ammonium chloride (NH4Cl), sodium fumarate, humic acid (HA), anthraquinone-2,6-disulfonate (AQDS), acetonitrile, methanol, chloroform, glutathione disulfide (GSSG) and glutathione (GSH) were purchased from Aladdin Biochemical Technology Co., Ltd. Hydroxylamine hydrochloride, sodium acetate, acetic acid, ammonium iron (II) sulfate (Fe(NH4)2·(SO4)2·6H2O) were purchased from Macklin Biochemical Co., Ltd. Glycerin, protonophore carbonyl cyanide m-chlorophenyl hydrazone (CCCP), rotenone, formamide were purchased from Sigma-Aldrich Trading Co., Ltd. Peptone, yeast extract, paraformaldehyde, 10% SDS solution, agar, PBS buffer (PH = 7.2–7.4), Tris-HCl buffer solution (PH = 7.5) were purchased from Solarbio Science & Technology Co., Ltd. The bacteria strains S. putrefaciens (BNCC 337021) and R. palustris (BNCC 232041) were obtained from Bnbio Co., Ltd. G. Soli (CCTCC AB 2014145) was provided by China Center for Type Culture Collection. The FISH probes for S. putrefaciens, R. palustris and G. Soli with the sequences of 5′-FAM- AGCTAATCCCACCTAGGTTCATC-3′, 5′-FAM-AGCTAATCCCACCTAGGTTCATC-3′ were synthesized by Shanghai Sangon Biotech Co., Ltd. NADP+/NADPH Assay Kit with WST-8, NAD+/NADH Assay Kit with WST-8 and Enhanced ATP Assay Kit were purchased from Beyotime Biotech. Inc. The ultra-pure water was obtained using a Millipore water purification system.

Characterization

The morphology of ZGO:Mn nano verifier was measured by a transmission electron microscope (TEM) (JEOL, JEM-2100, Japan) with a working voltage of 200 kV. Powder X-ray diffraction (XRD) patterns of ZGO:Mn nano verifier was measured on an X-ray diffractometer (Burker, D8 Advance, Germany) with Cu-Kα radiation (λ = 1.5406 Å) to determine the crystal structure. Absorbance was measured and recorded on a BioTeK Synergy H1 Hybrid Reader (BioTek, USA). Persistent luminescence performance of ZGO:Mn nano verifier was measured on a Hitachi FL4600 fluorescence spectrometer (Hitachi, Japan). The morphologies of S. putrefaciens and R. palustris were observed with a field emission scanning electron microscope (SEM) (Zeiss, SIGMA,Germany). Microbial consortia were imaged on the Olympus Fluoview FV3000 confocal microscope (Olympus, Japan). Current intensity of microbial cells was measured by a digital source meter (Keysight B1500A, Germany) connected to a probe station (PRCBE LAB, China). Fluorescence-activated cell sorting (FACS) was performed by BD FACSAria III Cell Sorter (BD Biosciences, USA). The Agilent 1260 Infinity HPLC system (Agilent, USA) was used to measure the concentration of lycopene.

Scanning electron microscopy characterization

The morphologies of S. putrefaciens signal router and R. palustris bio-actuator were observed with a field emission scanning electron microscope (Zeiss, Germany). Specifically, bacteria solution with OD600 of 0.6 was centrifuged and then washed twice with ultra-pure water. The washed bacteria were fixed in 10 mL 4% paraformaldehyde for 1.5 h. After the fixation process, the bacteria were washed twice with the rotation speed of 4000 rpm. Subsequently, 5 mL of 25, 50, 75, 100% ethanol solution was used in order to dehydrate the bacteria. Each dewatering time was 15 min. Finally, the dehydrated bacteria were centrifuged and dissolved in anhydrous ethanol for preparing SEM samples.

Performance measurements of the ZGO:Mn nano verifier

To investigate the relationship between the persistent luminescence intensity and Fe species concentration, standard solution with different Fe2+/Fetotal ratio was prepared with FeCl2 and FeCl3·6H2O (Fetotal was fixed at 2 mM, and the final concentration of ZGO:Mn nano verifier was 0.5 mg/mL). Then, the persistent luminescence intensity of the different solution was measured with a fluorescence spectrometer (Hitachi, FL4600, Japan). The excitation wavelength was 254 nm and the data mode was set as phosphorescence. Similarly, we measured the persistent luminescence intensity upon the addition of Mn (0.110 mg/mL), MnCl2 (2 mM), MnO2 (0.174 mg/mL), GSSG (2 mM), GSH (2 mM), supernatant and medium with fluorescence spectrometer (Hitachi, FL4600, Japan).

Monitoring of the Fe species in biological LAN with the nano verifier

To have a better understanding of the Fe signal transduction between S. putrefaciens and R. palustris, ZGO:Mn nano verifier was adopted. Specifically, ZGO:Mn with final concentration of 0.5 mg/mL was added into 5 mL sterilized tubes containing 2 mL LB medium and 2 mM Fe3+. Subsequently, 20 uL S. putrefaciens and 20 uL R. palustris with the OD600 around 0.8 was added. Then, the centrifuge tubes with mixed bacteria solution were placed into a shaker with a speed of 220 rpm at 30 °C. The persistent luminescence intensity of the mixed solution was measured at different culture time with a fluorescence spectrometer (Hitachi, FL4600, Japan). The excitation wavelength was 254 nm and the data mode was set as phosphorescence.

Cell viability tests

The colony-forming unit (CFU) was performed to determine the viability of S. putrefaciens signal router and R. palustris bio-actuator after incubation with the optical ZGO:Mn nano verifier. Specifically, diluted S. putrefaciens-ZGO:Mn and R. palustris-ZGO:Mn co-culture solution at different culture times was transferred into the Luria-Bertani (LB) solid medium respectively. After 24 h growth at 30 °C, white circular colonies can be observed. The CFU per milliliter can be determined by counting the colonies.

Fe signal transduction and bacterial growth model in biological LAN

Fe redox signal transduction processes in the biological LAN were modelled with Simulink 9.7 (R2019b). The systems of ordinary differential equations below were solved using ode45 solver. The mathematical model here can be used to investigate in silico the redox signal transduction kinetic behaviors in continuous cultures. In the constructed model, it is supposed that there are no delays in responses to signals to simplify the actual system. In addition, 1 OD was assumed to be 109 cells. The model parameters were calculated using a combination of the experimental data and the model.

Specifically, the concentrations of Fe3+ and Fe2+ were calculated based on the recorded luminescence intensity of the verifier according to the following equation (Supplementary Fig. 9d) in which FeTotal was 2 mM:

$${Lg}\left({Intensity}\right)=0.01764*{C}_{{{Fe}}^{2+}/{{Fe}}_{{Total}}}+2.126$$
(1)

Then, based on the calculated concentrations, the Fe2+ signal transduction from Fe3+ in bare S. putrefaciens system was simulated with Eq. (2) to fit and abstain the constants:

$${\left[{{Fe}}^{2+}\right]}_{{S}^{{\prime} }}={B}_{S}+\frac{{A}_{S}-{B}_{S}}{1+{\left(\frac{t}{{C}_{S}}\right)}^{{D}_{S}}}$$
(2)

Similarly, the Fe3+ signal transduction from Fe2+ in bare R. palustris system was simulated with Eq. (3) to fit and abstain the constants:

$${2-\left[{{Fe}}^{2+}\right]}_{{R}^{{\prime} }}={B}_{R}+\frac{{A}_{R}-{B}_{R}}{1+{\left(\frac{t}{{C}_{R}}\right)}^{{D}_{R}}}$$
(3)

Based on the above models of the bare S. putrefaciens and bare R. palustris system, the redox signal transduction and bacterial growth models of the co-culture system were established. Considering that S. putrefaciens transduces Fe3+ signal into Fe2+ while the Fe2+ can be transduced back to Fe3+ by R. palustris, a redox cycle module was introduced in the co-culture model to simulate Fe redox communication. The initial Fe3+ concentration was set at 2 mM which is consistent with the experimental data. In addition, Fe redox communication influence on bacterial growth was also considered. Overall, the differential equations were obtained as bellowing:

S. putrefaciens density in co-culture system:

$$\frac{d{[{Cells}]}_{S}}{{dt}}={\left[{{Fe}}^{2+}\right]}_{S}*{\left[{Cells}\right]}_{S}*\frac{{E}_{S}}{{F}_{S}+\frac{{G}_{S}-{F}_{S}}{1+{\left(\frac{t}{{H}_{S}}\right)}^{{I}_{S}}}}$$
(4)

R. palustris density in co-culture system:

$$\frac{d{[{Cells}]}_{R}}{{dt}}=\left(2-{\left[{{Fe}}^{2+}\right]}_{R}\right)*{\left[{Cells}\right]}_{R}*\frac{{E}_{R}}{{F}_{R}+\frac{{G}_{R}-{F}_{R}}{1+{\left(\frac{t}{{H}_{R}}\right)}^{{I}_{R}}}}$$
(5)

Fe2+ concentration in co-culture system:

$$\frac{d\left[{{Fe}}^{2+}\right]}{{dt}}= \frac{d{\left[{{Fe}}^{2+}\right]}_{S}}{{dt}}-\frac{d{\left[{{Fe}}^{2+}\right]}_{R}}{{dt}}={\left[{Cells}\right]}_{S}*\frac{{J}_{S}}{{K}_{S}+\frac{{L}_{S}-{K}_{S}}{1+{\left(\frac{t}{{M}_{S}}\right)}^{{N}_{S}}}}\\ -{\left[{Cells}\right]}_{R}*\frac{{J}_{R}}{{K}_{R}+\frac{{L}_{R}-{K}_{R}}{1+{\left(\frac{t}{{M}_{R}}\right)}^{{N}_{R}}}}$$
(6)

Bacteria culture

S. putrefaciens, R. palustris, S. putrefaciens-R. palustris co-culture and S. putrefaciens-R. palustris-G. Soli microbial consortia were all cultured in Luria-Bertani (LB) medium that contains (per litre) 10.0 g peptone, 5.0 g yeast extract and 5.0 g NaCl. G. Soli was cultured in N2-flushed LM medium that contains (per litre) 2.0 g lactate, 0.2 g yeast extract, 12.8 g Na2HPO4·7H2O, 3 g KH2PO4, 0.5 g NaCl and 1.0 g NH4Cl. Additional sodium acetate (20 mM) as the electron donor and sodium fumarate (50 mM) as the electron acceptor were added into the above LM medium. All the bacteria were cultured at 30 °C with a shaking rate of 220 rpm. R. palustris was photosynthetic bacteria and a full-spectrum LED (100 W) was used as its light source.

Preparation of the optical ZGO:Mn verifier

The optical ZGO:Mn verifier was synthesized by a hydrothermal method. Typically, a mixture solution was obtained by dissolving 0.01 mmol Ga(NO3)3, 2 mmol Zn(NO3)2, 0.005 mmol Mn(NO3)2 and 300 µL HNO3 into in 11 mL deionized water. Then 1 mmol Na2GeO3 solution was prepared by dissolving GeO2 powder in 2 mol L–1 NaOH solution. Subsequently, the prepared Na2GeO3 solution was added drop by drop into the above mixture solution. Ammonium hydroxide (28 wt%) was used to adjust the pH value of the mixture solution to 9.5. Then, the reaction was left at room temperature under stirring for 1 h. After that, the solution was transferred into a Teflon-lined autoclave and reacted at 220 °C for 6 h. The resultant ZGO:Mn nanorods were then obtained by centrifugation and washed three times with deionized water.

Construction of redox based biological LAN

The initial inoculation of S. putrefaciens and R. palustris were cultured in LB medium. To construct redox based biological LAN, S. putrefaciens signal router, optical verifier and R. palustris bio-actuator were introduced. Specifically, 1% of the seed culture of S. putrefaciens and R. palustris with initial OD600 around 0.8 were transferred respectively into sterilized centrifuge tubes with LB medium. Then, optical ZGO:Mn verifier (0.5 mg/mL) and Fe3+ (2 mM) were added. The biological LAN including the S. putrefaciens-G. Soli dual-band router was constructed as the above procedure except for the addition of G. Soli.

Redox signal verification in the biological LAN

The redox signal transduction can be monitored in real-time with the optical ZGO:Mn verifier. Briefly, the persistent luminescence intensity of the mixed solution at different time was measured with a Hitachi FL4600 fluorescence spectrometer (Hitachi, Japan). The excitation wavelength was 254 nm and the data mode was set as phosphorescence.

Construction of ΔCymA and ΔMtrC mutants

Mutants with MtrC-encoding gene deleted (ΔMtrC) and CymA-encoding gene deleted (ΔCymA) were constructed respectively as previous report. Specifically, an upstream and downstream fragment were amplified separately from the S. putrefaciens genomic DNA by PCR using the primer pair Mtrc-UP-smaI-F/Mtrc-UP-R, and the pair Mtrc-down-F/Mtrc-down-kpnI-R, respectively (Table S1). Then, the PCR products were purified using a QIAquick Gel Purification Kit (QIAGEN Inc.) and cloned into the SmaI/KpnI sites of pRE112 vector. Subsequently, the recombinant plasmid transformed in E. coli S17-1/λpir was transferred into S. putrefaciens via the E. coli-Shewanella conjugation. Finally, the Mtrc deletion mutants were screened and its genotype was confirmed by PCR using the primer pair Mtrc-JD-F/Mtrc-JD-R (Table S1) and DNA sequencing. Similarly, the CymA deletion mutants can be constructed according to the above protocol with its own primer pair (Table S1).

o-Phenanthroline spectrophotometric method for Fe2+ quantification

The well-known o-phenanthroline spectrophotometric method was adopted to quantify the concentration of Fe2+. Typically, 0.7 g Fe(NH4)2·(SO4)2·6H2O was dissolved in 1 M H2SO4 solution to prepare the Fe2+ standard solution (100 mg·L–1). 20 μL 10% (w/w) hydroxylamine hydrochloride solution, 50 uL of o-Phenanthroline (0.5 wt%) and 50 uL of sodium acetate buffer (pH = 5) were added into a 96-well plate. Subsequently, 0, 2, 4, 6, 8, 10, and 12 µL Fe2+ standard solution was added respectively into the 96-well plate. Each well was fixed to 200 µL with ultra-pure water. The above 96-well plate was left at room temperature for 15 min and the absorbance was measured at 510 nm to obtain the standard curve. Subsequently, the Fe2+ concentrations of different samples were measured. Specifically, 50 uL of o-Phenanthroline (0.5 wt%) and 50 uL of sodium acetate buffer (pH = 5) were added into a 96-well plate. Then, sample solution was added into the 96-well plate. Each well was fixed to 200 µL with ultra-pure water. The above 96-well plate was left at room temperature for 15 min and the absorbance was measured at 510 nm. Hence, the concentration of Fe2+ can be quantified according to the obtained standard curve.

Confocal fluorescence microscopy imaging

S. putrefaciens-R. palustris co-culture and S. putrefaciens-R. palustris-G. Soli microbial consortia were imaged with a Olympus Fluoview FV3000 confocal microscope (Olympus, Tokyo, Japan). Specifically, 10 mL cultured bacteria were first washed twice and resuspended in 4% paraformaldehyde solution for 1.5 h to fix the bacteria. The samples were then washed twice with PBS and resuspended in ethanol-PBS (V/V = 1:1) solution. Subsequently, the samples were storage at –20 °C for over 24 h. Subsequently, the samples were washed twice and resuspended in a hybridization buffer (30% formamide, 0.01% SDS, 20 mM Tris and 0.9 M NaCl). The FISH probes of S. putrefaciens (5′-FAM-AGCTAATCCCACCTAGGTTCATC-3′) and R. palustris (5′-Cy5-CCTCTGACTTAGAAACCCGC-3′) were then added. Both the FISH probes had a final concentration of 5 μg/mL. The mixed solution was then incubated overnight at 50 °C. After the hybridization, the samples were washed with washing buffer (0.01% SDS, 20 mM Tris and 0.9 M NaCl). Finally, the samples were prepared for the confocal imaging.

Fluorescence-activated cell sorting

As described above, fixed bacterial samples of S. putrefaciens-R. palustris co-culture and S. putrefaciens-R. palustris-G. Soli microbial consortia were resuspended in a hybridization buffer (30% formamide, 0.01% SDS, 20 mM Tris and 0.9 M NaCl). Then, The FISH probe of R. palustris (5′-Cy5-CCTCTGACTTAGAAACCCGC-3′) was added into the above prepared solution with a final concentration of 5 μg/mL. The mixed solution was then incubated overnight at 50 °C. After the hybridization, the samples were washed with washing buffer (0.01% SDS, 20 mM Tris and 0.9 M NaCl). Finally, the samples were prepared for flow cytometry to isolate the R. palustris bio-actuator cells.

Lycopene extraction and titration

The lycopene yields at different conditions were measured. Specifically, the bacteria cells cultured for 96 h at different conditions were harvested by centrifugation at 5000 × g for 5 min. The collected bacteria were washed twice with 1.0 g L–1 NaCl. The cells were dried at 70 °C overnight to remove the water. Subsequently, lycopene in R. palustris bio-actuator was extracted with 3 mL mixed reagent of n-hexane and methanol (1:1 v/v). The mixed solution was vortexed for 5 min and then centrifugated at 7500 × g for 10 min at 4 °C. Finally, the exacted lycopene in supernatant was collected. The concentration of lycopene was measured by the Agilent 1260 Infinity II HPLC system (Agilent, Palo Alto, CA, USA). This involved a C18 column (4.6 × 100 mm, Agilent) with the mobile phase consisted of acetonitrile/methanol/chloroform (42.5:42.5:15 v/v/v). The flow rate was set at 1.0 mL min–1 and the injection amount was set at 10 uL. The final concentrations of CCCP, rotenone, AQDS and HA used in this manuscript are 1 μM, 100 μM, 250 μM and 0.5 mg/mL respectively.

Determination of cellular redox state and ATP concentration

R. palustris bio-actuator was isolated from the S. putrefaciens-R. palustris co-culture and S. putrefaciens-R. palustris-G. Soli microbial consortia through flow cytometry by labeling FISH probe of R. palustris (5′- Cy5-CCTCTGACTTAGAAACCCGC-3′). The collected R. palustris bio-actuator through fluorescence-activated cell sorting were centrifuged and washed with PBS solution (0.01 M, PH = 7.2–7.4). In particular, NAD+/NADH and NADP+/NADPH ratios were measured with purchased NAD+/NADH Assay Kit with WST-8 and NADP+/NADPH Assay Kit with WST-8. According to the protocol, the NAD+/NADH and NADP+/NADPH ratios were determined with a colorimetric assay under a microplate reader by detecting wavelength of 450 nm. For ATP concentration measurements, an Enhanced ATP Assay kit was adopted. The ATP concentration can be obtained by detecting chemiluminescence with a luminometer plate reader.

Shake-flask fermentation

Shake-batch fermentations were conducted in 50.0 mL tube with the working volume of 30.0 mL. Firstly, S. putrefaciens and R. palustris were cultured in O2-free LB medium respectively. Then, 1% (v/v) seed culture of S. putrefaciens and R. palustris solution with the culture OD600 around 0.8 were transferred respectively into the tubes containing 30 mL LB medium with or without 2 mM Fe3+. The fermentation temperature was set at 30 °C and the agitation speed was controlled at 220 rpm. In addition, a full-spectrum LED (100 W) was used for external light illumination.

Fed-batch fermentation

Fed-batch fermentations were conducted in 5.0 L fermenter with the working volume of 3.0 L. Firstly, S. putrefaciens and R. palustris were cultured in O2-free LB medium respectively. Then, both the S. putrefaciens and R. palustris solution with the culture OD600 around 0.8 was transferred into sterilized culture bottle with 400 mL O2-free LB medium containing 2 mM Fe3+. The S. putrefaciens-R. palustris co-culture seed was then incubated at 30 °C with a shaking rate of 220 rpm for 96 h, and a full-spectrum LED (100 W) was used as its light source. Subsequently, 1% (v/v) seed culture was transferred into sterilized fermenter, and the fed-batch fermentation started with feeding. An anaerobic condition inside fermenter was created by flushing argon gas until concentration of dissolved oxygen dropped to zero. In order to maintain anaerobic conditions inside fermenter, the feed medium (LB medium containing 2 mM Fe3+) was also flushed with argon gas before being fed into fermenter. 3 g/L HCl and 3 g/L NaOH were used to control the pH of bacterial fermentation broth at ~7.0. The fermentation temperature was set at 30 °C and the agitation speed was controlled at 220 rpm. In addition, a full-spectrum LED (100 W) was used for external light illumination.

16S rRNA gene sequencing

Total genomic DNA was extracted from the S. putrefaciens-R. palustris co-culture and S. putrefaciens-R. palustris-G. Soli microbial consortia. Three independent biological replicates were performed. Specifically, V3 and V4 regions of 16S rRNA were amplified by PCR with primers of 341 F:5′-CCTA CGGG NGGC WGCA G-3′ and 805 R: 5′-GACT ACHV GGGT ATCT AATC C-3′. The amplified DNA was then processed for gene sequencing with Illumina MiSeq-PE250 system. QIIME2 was adopted to analyze the gene sequences and classify the operational taxonomic units (OTUs) on the basis of 97% sequence identity. Subsequently, the functional potential of the bacterial community was predicted by using PICRUSt2 based on the 16 S rRNA gene sequencing data. The predictive genes were then used to construct Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways based on KEGG database. Differential functional gene expression was analyzed with STAMP 2.0.

RNA extraction, RNA-Seq and transcriptomic analysis

Three independent biological replicates of S. putrefaciens-R. palustris co-culture with/without Fe redox communication were subjected to RNA-seq analysis. According to the manufacturer’s protocols, the total RNA was extracted using the Trizol reagent (Invitrogen, California, USA). RNA quantity was measured with the Qubit 2.0 fluorometer (ThermoFisher Scientific, Waltham, MA, USA). Obtained RNA was quality and integrity checked using the RNA 6000 Nano kit with an Agilent Bioanalyzer 2100 (Agilent Technologies). Then the rRNA was depleted with Ribo-Zero rRNA Removal Kit (bacteria) (Illumina, San Diego, CA, USA) and cDNA libraries were constructed with an Illumina NEBNext Ultra RNA Library Prep Kit (New England Biolabs, Ipswich, MA, USA). Subsequently, the cDNA libraries were pair-end-sequenced on an Illumina Novaseq 6000 platform (Illumina, San Diego, CA, USA). Afterward, Cutadapt software was adopted to clean the raw reads by removing adaptor sequences, ambiguous based annotated as N, and low-quality sequences (Qphred ≤ 20 bases, accounting for >50% of the total read length). The clean reads were obtained with the FastQC software and were then mapped to prjna881029 (S. putrefaciens) and prjna683609 (R. palustris) reference genome using STAR software. The read counts and fragments per kilobase of transcripts per million mapped reads were calculated using featureCounts (version 1.5.0-p3) and kallisto (version 0.46.1) respectively. Mapped reads were normalized with fragments per kilobase per million mapped reads (FPKM). Differential expression analysis of two groups was performed using DESeq2 (version 1.16.1), which is based on a negative binomial distribution model. The standardized method was DESeq. Genes with adjusted P value < 0.05 and Log2 fold change ≥ |0.58| were considered significantly differentially expressed.

Bioelectrochemical detection

The gold interdigitated microelectrode was adopted to measure electron transfer function of S. putrefaciens-R. palustris co-culture with or without Fe redox communication. The gold interdigitated microelectrode was fabricated on a silicon wafer with 300 nm thermal silicon oxide (SiO2/Si) by photolithography and metal evaporation47. To be specific, a 4 inch silicon wafer was washed sequentially in acetone, ethanol and deionized (DI) water. Subsequently, the silicon wafer was dried with a compressed nitrogen gun. The residual moisture and solvent were removed by baking the wafer at 180 °C for 5 min. After cooling down, 3 ml of S1813 resist was dispensed at the wafer center and the wafer was then baked on the hot plate at 115 °C for 1 min. Then, a direct-write optical lithography system (ABM. Inc, America) was adopted for the preparation of the gold interdigitated microelectrode. After the photolithography process, the wafer was then loaded face down onto the sample holder of a thermal evaporator system (Jiashuo JSD-300, China) to deposit the active material (The thickness of chromium and gold are set as 5 nm and 40 nm respectively). S. putrefaciens-R. palustris co-culture with or without Fe redox communication incubated at 96 h were washed twice with 1×PBS buffer. Then, we fixed the OD600 of S. putrefaciens-R. palustris co-culture to 3.0 with 1 × PBS buffer. Subsequently, the real-time current intensities of the co-culture were measured by injecting 10 µL bacterial suspension into the gold interdigitated microelectrode with the input voltage set at 0.2 V. Reported current densities (μA cm–2) were then obtained by averaging the ratio of the stable current intensity to the effective working-area (0.80 × 0.88 mm) of the gold interdigitated microelectrode.

Determination of CO2 fixation rate

The CO2 fixation rates in S. putrefaciens-R. palustris co-culture and S. putrefaciens-R. palustris-G. Soli microbial consortia were measured respectively. The bacteria were cultured in LB medium with the addition of NaHCO3 (30 mM). Then, the suspension was collected by centrifugation at 5000 × g for 5 min. H2SO4 solution (98 wt%) was used to prepare a 5 mM solution for the quantification of inorganic carbon by recording the mass variation. The CO2 fixation rate was calculated as: bicarbonate consumption (mM)/time interval (96 h)/DCW(g).

Statistics and reproducibility

All data are presented as mean ± standard deviation (SD) and statistical analyses and graphs were performed using OriginLab 9.0 and GraphPad Prism 8.0.1. For comparison of two groups, significance was determined by the two-tailed unpaired Student’s t test. Each n indicates the number of biologically independent samples. Data in Figs. 2c, f, h, 3c–e, 5a–d and 6d–g and Supplementary Figs. 2, 9b, e, f, 10a, b, 11, 22 and 26 were successfully replicated in two independent experiments.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.