Construction and optimization of the PCA biosynthesis pathway in S. oneidensis
Using S. oneidensis MR-1, we found that the addition of PCA at ~80 µM resulted in the highest power generation, which was significantly higher than that of either flavins or quinones when added at their optimal levels, respectively (Fig. 1a and Supplementary Figs. 2–5). Meanwhile, PCA exhibited an optimal level in mediating EET of S. oneidensis MR-1. As shown in Fig. 1b, the EET rate increased as the amount of added PCA increased in the beginning, while the EET rate peaked at the PCA concentration of ~80 µM and a further increase in PCA concentration resulted in suboptimal EET. This result suggested that the PCA level needed to be maintained at ~80 µM (an optimal level as an electron shuttle) to promote EET.
To establish a PCA biosynthesis pathway (Supplementary Fig. 6) and maintain the PCA level at ~80 µM in S. oneidensis MR-1, the native PCA biosynthesis operons phzABCDEFG from nine phenazine-producing species of four different genera45 (Streptomyces. cinnamonensis DSM1042 in Streptomyces (I), Burkholderia lata 383 in Burkholderia (II), Erwinia. carotovora ssp. Atroseptica SCRI1043 in Pectobacterium (III), and Pseudomonas. aeruginosa PA14, P. aeruginosa PAO1, P. fluorescens 2-79, P. aureofaciens 30-84, P. chlororaphis GP72, P. chlororaphis PCL1391 in Pseudomonas (IV)), were individually cloned into S. oneidensis MR-1 driven by the promoter PlacUV5, resulting in eleven recombinant strains SC1-SC11 (Fig. 1c, Supplementary Fig. 1, Supplementary Note 1, and Supplementary Data 1). Three strains (SC1, SC3, and SC11) that harbored the three PCA synthesis operons from S. cinnamonensis DSM1042, E. carotovora ssp. Atroseptica SCRI1043, and P. chlororaphis PCL1391 were unable to elicit PCA synthesis, while the other eight recombinant strains were capable of synthesizing detectable levels of PCA. Notably, the operons from Pseudomonas genera (P. aeruginosa PA14, P. aeruginosa PAO1, P. fluorescens 2-79, and P. aureofaciens 30-84) showed relatively high activity for synthesizing PCA. Strain SC7 bearing the operon phzABCDEFG from the PCA biosynthesis gene cluster in P. aeruginosa PAO1 produced the highest PCA level of 8.72 ± 0.76 µM (Fig. 1c). The EET rates of these strains were positively correlated with the synthesized PCA level, and strain SC7 (PlacUV5–phzABCDEFG) enabled the highest EET rate with the maximum power density of 1185.93 ± 63.46 mW m−2, 14.07-fold increase over WT (84.30 ± 2.53 mW m−2) (Fig. 1d and Supplementary Fig. 7). Strain SC7 was thus chosen for further analysis (Supplementary Fig. 1, Supplementary Note 1, and Supplementary Data 1).
To improve PCA synthesis, the promoter that drives the operon phzABCDEFG was optimized to upregulate the operon expression level in strain SC7. Six promoters (ParcA, Pae, Plaps, Ptac, PT7, and Pxyl) with gfp as a reporter gene were tested in S. oneidensis MR-1, revealing that Pae, Pxyl, and Ptac were stronger than PlacUV5 (Supplementary Fig. 8), and these three promoters were then individually assembled into strain SC7 to replace the promoter PlacUV5, resulting in three recombinant strains SO1-SO3 (Fig. 1e, Supplementary Fig. 1, Supplementary Note 1, and Supplementary Data 1). Strains SO2 and SO3 exhibited higher PCA levels than that of strain SC7 (Fig. 1e). Notably, strain SO3 (Ptac–phzABCDEFG) containing the Ptac-driven phzABCDEFG operon from P. aeruginosa PAO1 produced PCA at a level of 72.74 ± 4.30 µM (Fig. 1e), close to the optimal PCA level (~80 µM), which resulted in a maximum output power density of 1757.87 ± 28.91 mW m−2, 1.48-fold increase than that of strain SC7 (1185.93 ± 63.46 mW m−2) (Fig. 1f and Supplementary Fig. 7). Thus, strain SO3 was selected for further study (Supplementary Fig. 1, Supplementary Note 1, and Supplementary Data 1).
Engineering cell membrane permeability to facilitate PCA transport
To enhance EET rate, the intracellular PCA has to be transported out of cells in time, and it has been known that cellular permeability plays an important role in transmembrane transport of electron shuttles46. Thus, aiming to increase the cell membrane permeability of strain SO3 to facilitate PCA transport, the genes of outer membrane (OM) porin and inner membrane (IM) efflux pump were individually incorporated in S. oneidensis to regulate the porin-mediated passive diffusion and the efflux pump-mediated active transport of PCA, respectively (Fig. 2a).
a Schematic of cell membrane permeability consisting of porin-mediated passive diffusion in the outer membrane (OM) and efflux pump-mediated active transport in the inner membrane (IM). b Maximum power density (taken from Supplementary Fig. 9) of strains SP1-SP6 and SP7-SP10. c NPN uptake assay to measure cell permeability (arb. units: arbitrary units). d Schematic of the intracellular PCA detector (pYYD-PsoxR–soxR-PsoxS–gfp) for evaluating cellular membrane permeability to PCA. The redox stress-responsive sensor PsoxR–soxR-PsoxS was employed to develop an intracellular PCA detector, comprising the effector protein SoxR and the SoxR-regulated promoter PsoxS. In PsoxR–soxR-PsoxS, the [2Fe-2S] cluster within SoxR dimer could be oxidized by PCA to activate the transcription of PsoxS. Upon exogenous introduction of PCA to a strain containing the PCA detector but lacking the PCA biosynthesis operon, the externally supplied PCA is transported into the cell, thereby activating GFP expression. This transport process is positively correlated with cellular membrane permeability. Consequently, cellular membrane permeability to PCA can be assessed by measuring the relative fluorescence intensity of GFP. e The relative fluorescence intensity of GFP. f The synthesized PCA level. g Cell growth profiles. h The viable cell number assessed from the dilution plate method. Results in (b–h) from three independent experiments (n = 3) were expressed as means and standard errors, and data in (b–h) are shown as the mean ± SD. The result (c) has been checked for consistency with 3 individual experiments. Source data are provided as a Source Data file.
Six OM porins were individually expressed in strain SO3: OprF from P. aeruginosa PAO1, OmpPst1 from Providencia stuartii, OmpF from E. coli K12, OmpC from E. coli K12, Omp35 from Klebsiella. pneumoniae, and Omp36 from K. pneumoniae. A dual-plasmid system was used, in which the PCA biosynthesis operon phzABCDEFG driven by the promoter Ptac was expressed in the plasmid pYYD, and the porin gene driven by the promoter PlacUV5 was expressed in the other plasmid pHG13, eventually resulting in six strains SP1-SP6 (Fig. 2b, Supplementary Fig. 1, Supplementary Note 1, and Supplementary Data 1). In these six strains, strain SP1 (Ptac–phzABCDEFG and PlacUV5–oprF) exhibited the highest power density (2262.45 ± 10.92 mW m−2), 1.29-fold higher than that of SO3 (Fig. 2b and Supplementary Fig. 9).
In addition, we individually expressed four IM efflux pumps (PA3718 from P. aeruginosa PA14, MexG from P. aeruginosa PAO1, PA3523 from P. aeruginosa PAO1, and Bfe from S. oneidensis MR-1), resulting in four strains SP7-SP10 (Fig. 2b, Supplementary Fig. 1, Supplementary Note 1, and Supplementary Data 1). However, these strains with the IM efflux pumps exhibited lower EET than that of strain SO3. Strain SP1 was thus used for further study (Supplementary Fig. 1, Supplementary Note 1, and Supplementary Data 1).
To verify whether OprF enhanced cell envelope permeability to facilitate PCA transport and consequently increased the EET rate, we conducted an N-phenyl-1-naphthylamine (NPN) uptake assay to evaluate the cell membrane permeability. The non-polar NPN molecule would emit strong fluorescence when diffusing from the aqueous solution into non-polar cellular phospholipids. When the above strains were mixed with NPN, the SP1 + NPN mixture showed a stronger fluorescence intensity than that of SO3 + NPN (Fig. 2c), suggesting NPN in strain SP1 was higher than in strain SO3, an indication that strain SP1 possessed higher cell membrane permeability than that of strain SO3.
Given the difference in the molecular structure of PCA and NPN, to evaluate cellular membrane permeability to PCA molecule, we developed an intracellular PCA detector (pYYD-PsoxR–soxR-PsoxS–gfp) based on the redox stress-responsive biosensor PsoxR–soxR-PsoxS, which consists of the effector protein SoxR driven by the constitutive promoter PsoxR and the promoter PsoxS regulated by SoxR (Fig. 2d). In PsoxR–soxR-PsoxS, the SoxR dimer binds to the promoter PsoxS, in which the [2Fe-2S] cluster of SoxR can be oxidized by PCA, leading to change in SoxR conformation, thus activating the transcription of PsoxS47,48. Upon exogenously supplying PCA to the strain harboring the PCA detector (pYYD-PsoxR–soxR-PsoxS–gfp) but lacking the PCA biosynthesis operon, the externally supplied PCA is transported into the cell to activate the GFP expression, and the PCA transport is positively correlated with the cellular membrane permeability, which thus enables evaluation of cellular membrane permeability to PCA by measuring the relative fluorescence intensity of GFP. As shown in Fig. 2e, exogenous addition of PCA (80 µM) to the culture medium resulted in the relative fluorescence intensity of 17076.10 ± 356.18 arbitrary units (arb. units) for strain SP1-D (pYYD-PsoxR–soxR-PsoxS–gfp & pHG13-PlacUV5-oprF), which was 1.38-fold higher than that of strain SO3-D (pYYD-PsoxR–soxR-PsoxS–gfp & pHG13) (12353.93 ± 259.50 arb. units). This result suggests that more PCA molecules could be transported into the cell with OprF expression, thus demonstrating increased cellular membrane permeability to PCA in strain SP1 compared to strain SO3.
However, in comparison with strain SO3, the level of PCA in strain SP1 was slightly decreased from 72.74 ± 4.30 µM (SO3) to 63.23 ± 3.91 µM (SP1) (Fig. 2f), and strain SP1 exhibited a significant decrease in growth rate compared to strain SO3 and WT (Fig. 2g). We further assessed the viable cell number via colony-forming units (CFU). As shown in Fig. 2h, the number of viable cells was 3.27 ± 0.52 × 107 for strain SO3, 1.84-fold higher than that of strain SP1 (1.78 ± 0.01 × 107), indicating the lower cell viability of strain SP1 than that of strain SO3. These results suggested that overexpression of OprF inhibited cell growth and impaired cell viability.
Upon transcription and translation in cytoplasm, the porin OprF (a membrane protein) is transported and inserted into the cell membrane via the signal recognition particle pathway49 and the Sec-translocon50,51. A previous study showed that overexpression of exogenous membrane protein would overload the Sec-translocon system due to its limited capacity, thus inducing cytotoxicity52. Specifically, when the expression level of an exogenous membrane protein (e.g., OprF in S. oneidensis MR-1) exceeds the transport capacity of the Sec-translocon system, it creates bottlenecks in protein sorting and translocation, leading to intracellular misfolding and aggregation of both OprF precursors and other membrane protein precursors in cytoplasm. The misfolding and aggregation of protein precursors would disrupt cellular homeostasis and trigger stress response of the cells. In addition, the accumulation of OprF at high levels would inhibit the proper insertion and folding of other essential endogenous membrane proteins, thus inhibiting cell growth. Consequently, OprF expression facilitates PCA transmembrane transport to accelerate EET, while its cytotoxicity would impair cell growth and viability.
Dynamic decoupling PCA biosynthesis and transport relieved OprF-induced cytotoxicity
To relieve OprF cytotoxicity, we designed a dynamic regulatory approach to decouple PCA transport and biosynthesis. In the initial phase (low PCA level), the cell synthesized PCA without expressing the porin OprF; when PCA accumulated to a certain threshold level, as sensed by the PCA biosensor, OprF expression was then initiated to facilitate PCA transport. In this way, the cellular metabolic processes were not inhibited in the early phase of cell growth due to silencing of the oprF gene. This allowed effective cell growth and PCA synthesis. By the time OprF overexpression commenced, the cell growth had reached the exponential growth phase, and the cellular machineries responsible for endogenous membrane protein synthesis, translocation, insertion, and folding were already established, thereby mitigating the cytotoxicity caused by the excessive OprF expression.
To this end, the redox stress responding sensor PsoxR–soxR-PsoxS was further used to regulate the expression of the oprF gene. It was found that the GFP expression level was correlated with the added PCA level (Supplementary Fig. 10), suggesting that PsoxR–soxR-PsoxS could act as the PCA biosensor for sensing the PCA level. The gene circuit PsoxR–soxR-PsoxS–oprF was constructed in the plasmid pHG13 and transferred into strain SO3, resulting in the recombinant strain SD1 (Ptac–phzABCDEFG and PsoxR–soxR-PsoxS–oprF) that harbored the PCA synthesis operon Ptac–phzABCDEFG in the plasmid pYYD (pYYD-PCA), and the PCA biosensor-regulated transport circuit PsoxR–soxR-PsoxS–oprF in the plasmid pHG13 (pHG13-soxR-PsoxS–oprF) (Fig. 3a, top row; Supplementary Fig. 1, Supplementary Note 1, and Supplementary Data 1). In this way, strain SD1 possessed the capacity of dynamic regulation of PCA transport based upon sensing the intracellular PCA level, of which the EET rate was further assessed.
a Schematic of dynamic decoupling of PCA biosynthesis with transport (top row) and optimization of PCA biosensor (bottom row). The PCA biosensor PsoxR–soxR-PsoxS consisted of effector protein SoxR driven by constitutive promoter PsoxR, and promoter PsoxS regulated by SoxR, in which SoxR dimer bound with promoter PsoxS and its [2Fe-2S] cluster could be oxidized by PCA, leading to the change of SoxR conformation, thus activating transcription of PsoxS. In strain SD1 harboring the PCA synthesis operon Ptac–phzABCDEFG in pYYD (pYYD-PCA) and the PCA biosensor-regulated transport circuit PsoxR–soxR-PsoxS–oprF in pHG13 (pHG13-soxR-PsoxS–oprF) (top row), cell first synthesized PCA driven by operon Ptac–phzABCDEFG, then initiated OprF porin expression under the control of gene circuit PsoxR–soxR-PsoxS–oprF to achieve programmed PCA biosynthesis and transport. In the optimization of PCA biosensor (bottom row), site-specific mutagenesis was conducted in the SoxR DNA binding site, locating between the -35 and -10 elements in the promoter PsoxS of wild-type PCA biosensor (PsoxR–soxR-PsoxS), resulting in nine PCA biosensor variants PsoxR–soxR-PsoxS V1-V9, respectively. b Maximum power density (taken from Supplementary Fig. 11) of the strains SD1 and SDV1-SDV9. c The oprF gene expression profiles regulated by the promoters PlacUV5 in SP1, PsoxR–soxR-PsoxS in strain SD1, and PsoxR–soxR-PsoxS V3 in strain SDV3, respectively. d Cell growth curves of strains WT, SP1, and SDV3. e The synthesized PCA level. Results in (b–e) from three independent experiments (n = 3) were expressed as means and standard errors, and data in (b–e) are shown as the mean ± SD. Source data are provided as a Source Data file.
However, strain SD1 exhibited an unexpectedly lower output power density of 1797.64 ± 53.40 mW m−2 than its parental strain SP1 (2262.45 ± 10.92 mW m−2) (Fig. 3b and Supplementary Fig. 11), suggesting that the OprF expression under the control of the wild-type PCA biosensor impeded EET. To reveal the differential expression of OprF in strains SP1 and SD1, the corresponding expression levels were quantified by fusing green fluorescent protein (GFP) to OprF, resulting in PlacUV5–oprF–gfp in strain SP1 and PsoxR–soxR-PsoxS–oprF–gfp in strain SD1. We found that the time and intensity of the GFP expression under the control of PsoxR–soxR-PsoxS in strain SD1 were much delayed and lower than in strain SP1 under the control of the promoter PlacUV5 (Fig. 3c), explaining the tardiness and weakness of OprF expression in SD1, which incapacitated PCA transport, thus inhibiting the EET rate of strain SD1.
To address this issue, we developed a site-specific mutagenesis approach to optimize the PCA biosensor to achieve appropriate response strength and sensitivity. In the PCA-biosensor PsoxR–soxR-PsoxS, the SoxR dimer binds to the 19 bp binding sites located between the -35 and -10 elements of the promoter PsoxS. When SoxR is oxidized by PCA, its redox-dependent conformation is then altered, which results in DNA twist of the promoter PsoxS and allows RNA polymerase to bind to PsoxS to initiate the transcription of the downstream genes48. Therefore, the DNA sequence of the promoter PsoxS, acting as the SoxR binding site, plays a decisive role in the sensitivity of the PCA biosensor. We conducted single-, two-, and multi-base mutations in the SoxR DNA binding site in the promoter PsoxS, resulting in nine PCA biosensor variants PsoxR–soxR-PsoxS V1 – PsoxR–soxR-PsoxS V9, respectively (as shown in Fig. 3a, bottom row). By replacing the wild-type PCA biosensor PsoxR–soxR-PsoxS with the nine PCA biosensor variants in strain SD1, respectively, nine recombinant strains SDV1-SDV9 were obtained (Supplementary Fig. 1, Supplementary Note 1, and Supplementary Data 1). Bio-electrochemical characterizations of these strains showed that strain SDV3 (Ptac–phzABCDEFG and PsoxR–soxR-PsoxS V3-oprF) exhibited the highest EET rate with the maximum power density of 2845.26 ± 100.19 mW m−2, 1.26-fold higher than that of strain SP1 (2262.45 ± 10.92 mW m−2) (Fig. 3b and Supplementary Fig. 11), which, to the best of our knowledge, is one of the highest recorded power outputs by a recombinant exoelectrogen3,9,53,54. Strain SDV3 was thus chosen for further study (Supplementary Fig. 1, Supplementary Note 1, and Supplementary Data 1).
In addition, the gene expression profile under the control of the PCA biosensor variant PsoxR–soxR-PsoxS V3 (PsoxR–soxR-PsoxS V3-oprF–gfp in SDV3, Fig. 3c) showed that the oprF expression time and intensity in strain SDV3 were earlier and higher than those in strain SD1 (PsoxR–soxR-PsoxS–oprF–gfp in SD1), but still slower and significantly lower than those in SP1 (PlacUV5–oprF-gfp in SP1). This observation suggested that the response strength and sensitivity of the PCA biosensor variant PsoxR–soxR-PsoxS V3 was optimized in regulating the OprF expression. As shown in Fig. 3d, the growth rate and the final OD600 of strain SDV3 were similar to WT and higher than strain SP1, which showed the OprF cytotoxicity in strain SP1 was resolved in strain SDV3. Also, the biosynthesized PCA level in strain SDV3 was recovered from 63.23 ± 3.91 µM (SP1) to 78.08 ± 0.50 µM (SDV3) (Fig. 3e). Thus, the PCA biosensor-based approach to dynamically decouple PCA synthesis and transport relieved cytotoxicity caused by the OprF overexpression, enabling the synthesis of PCA at a level to achieve high output power density.
Elucidating molecular mechanisms of PCA-boosted EET
PCA shuttled electrons from OM c-Cyts MtrC and OmcA to electrode. A previous study showed that the electron shuttle as redox mediator could receive electrons from outer membrane c-type cytochromes (OM c-Cyts), which subsequently shuttled these electrons to electrode34. To elucidate how electrons were transferred from strain SDV3 to PCA molecules, the genes mtrC and omcA encoding the OM c-Cyts MtrC and OmcA were thus individually or simultaneously deleted in SDV3 and WT, respectively. As shown in Fig. 4a and Supplementary Fig. 12, strains SDV3∆mtrC (1221.94 ± 53.25 mW m−2) and SDV3∆omcA (474.52 ± 75.22 mW m−2) showed 2.33- and 6.00-fold decrease in power density compared to strain SDV3 (2845.26 ± 100.19 mW m−2), respectively. Moreover, SDV3∆mtrC∆omcA (53.15 ± 11.41 mW m−2) displayed minimal power generation, similar to the strain WT∆mtrC∆omcA (56.87 ± 7.73 mW m−2). These findings suggested the OM c-Cyts MtrC and OmcA were essential for PCA-mediated EET. Similarly to another phenazine derivative, phenazine methosulfate55,56, PCA could almost completely oxidize both MtrC and OmcA in vitro (Supplementary Fig. 13 and Supplementary Note 2). These results revealed that electrons were transferred from the SDV3 cell to PCA through OM c-Cyts MtrC and OmcA.
a Maximum output power density (taken from Supplementary Fig. 12) of the c-Cyts deletion mutants, including SDV3 (MtrC+OmcA+), SDV3∆mtrC (MtrC–OmcA+), SDV3∆omcA (MtrC+OmcA−), SDV3∆mtrC∆omcA (MtrC–OmcA−), and WT∆mtrC∆omcA (MtrC–OmcA–). b–f Interactions in the complexes of PCA-MtrC (b–d) and PCA-OmcA (e, f) were analyzed, in which hydrogen bonds were shown as green dotted lines and non-bond interactions (e.g. hydrophobic interactions) were shown as red radial lines. The gray, green, red, and black balls represented carbon, nitrogen, oxygen, and iron atoms, respectively. Results in (a) from three independent experiments (n = 3) were expressed as means and standard errors, and data in (a) are shown as the mean ± SD. The results in (b–f) have been checked for consistency with 3 individual experiments. Source data are provided as a Source Data file.
The electron shuttles (flavins) in Shewanella were observed to either bind with OM c-Cyts as a cofactor or diffuse freely along the redox gradient potential57,58,59,60. To explore the interaction mechanism between OM c-Cyts (MtrC and OmcA) and PCA, differential pulse voltammetry (DPV) analysis was performed. The shifts in PCA peak potential (Supplementary Figs. 14–15 and Supplementary Note 3–4) and the variations in the PCA half-width potential (Supplementary Fig. 16 and Supplementary Note 5) suggested that PCA could function as a cofactor, binding with MtrC and OmcA to form complexes to facilitate EET at low concentrations. As the PCA concentration increased, the PCA binding sites within MtrC and OmcA became saturated, thus the excess PCA adopted the diffusion-based shuttling mechanism for EET.
To further investigate the specific binding sites involved in the interaction between PCA and MtrC/OmcA, we conducted computational molecular docking simulation. The hemes 2, 7, and 10 of MtrC and the hemes 5 and 10 of OmcA, located at the termini of multiheme wires61,62,63, were thus selected as the box centers for docking simulation, respectively. The simulation results showed that PCA could form hydrogen bonds with residues Gly240, heme 2, His507, Thr628, and Thr631 in MtrC (Fig. 4b–d), and Ser356 in OmcA (Fig. 4f). In addition, PCA also formed non-bond interactions (e.g., hydrophobic interactions) with various residues in both MtrC (e.g., Lys220, His222, Trp239, Lys241, Asn243, His512, Val626, His627, Phe635, and hemes 6, 7, 10, etc. Fig. 4b–d) and OmcA (e.g., His358, His359, Ser695, His696, Thr699, and hemes 8, 10 etc. Fig. 4e, f). The shortest distances between the N5 atom of PCA and the iron atoms of heme 2, heme 7, and heme 10 in MtrC, as well as heme 5 and heme 10 in OmcA were measured to be 5.8, 6.2, 7.3, 5.8, and 6.8 Å, respectively (Supplementary Fig. 17 and Supplementary Table 1), which were all shorter than the maximum distance (11 Å) allowing for electron hopping to occur, indicating both MtrC and OmcA could transfer electrons to PCA via electron hopping from heme 2, heme 7 and/or heme 10 of MtrC, as well as heme 5 and/or heme 10 of OmcA.
To validate these simulation results, the dissociation constant (Kd) in the interactions between PCA and c-Cyts was calculated according to the protein-ligand binding model developed by Okamoto et al.64 (Eqs. 1–4). Accordingly, the amino acids that interact with PCA in MtrC and OmcA were firstly individually mutated to alanine (Ala), thus constructing 19 c-Cyts Ala mutants (Supplementary Data 1). According to Eqs. 3 and 4, the estimated Kd value for PCA was 0.36 µM when interacting with the wild-type (WT) S. oneidensis MR-1 (Table 1 and Supplementary Fig. 18). In comparison, by interacting with c-Cyts mutants, the estimated Kd value for PCA was significantly increased. Especially for strains WT-MtrCLys220Ala (8.20 µM), WT-MtrCHis222Ala (6.47 µM), WT-MtrCLys241Ala (2.50 µM), WT-MtrCHis507Ala (5.94 µM), WT-MtrCHis627Ala (3.32 µM), WT-MtrCThr631Ala (2.59 µM), WT-MtrCPhe635Ala (8.00 µM), WT-OmcAHis359Ala (12.94 µM), WT-OmcASer695Ala (6.00 µM), and WT-OmcAThr699Ala (4.31 µM), the estimated Kd value was 22.78-, 17.97-, 6.94-, 16.50-, 9.22-, 7.19-, 22.22-, 35.94-, 16.67-, and 11.97-fold higher than that of the WT, respectively, indicating the disruption of these amino acid would hinder the binding of PCA with c-Cyts (MtrC and OmcA). Besides, the c-Cyts Ala mutants all showed decreased maximum power density (Supplementary Fig. 19 and Supplementary Note 6). These results thus collectively confirmed the molecular docking simulations.
PCA enhanced carbon source catabolism, c-Cyts biosynthesis, and biofilm formation. Transcriptomic analysis (Supplementary Figs. 20-21 and Supplementary Table 2) showed significant alteration in the expression of genes related to carbon catabolism (e.g., gltA, SO_1053), cytochromes biosynthesis (e.g., mtrC, dmsAB65, yceJ66, etc.), iron uptake/transport (e.g., fbpA67), c-di-GMP synthesis and degradation (e.g., dgcS68), cellular motility (e.g., flgT69 and pilV70), cell energy and chemotaxis (e.g., SO_143471), as well as prophages (e.g., gpF and gp4672) in strain SDV3 with synthesized PCA in comparison to the WT strain (Supplementary Fig. 22, Supplementary Note 7, and Supplementary Data 2). These processes were tightly associated with the lactate catabolism, c-Cyts biosynthesis, and biofilm formation, suggesting PCA could modulate Shewanella cellular metabolism and behavior.
As shown in Fig. 5a, strain SDV3 exhibited a higher lactate consumption rate than that of WT, indicating PCA accelerated carbon source metabolism to enhance intracellular electron generation. Additionally, with heme staining, the sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) showed the bands corresponding to MtrC and OmcA in strain SDV3 were significantly broader than those in WT (Fig. 5b), indicating a significantly higher abundance of MtrC and OmcA in strain SDV3 than that in WT. Consistent with this observation, the results of Raman spectrum (Supplementary Fig. 23a and Supplementary Note 8) and UV-visible spectroscopy (Supplementary Fig. 23b and Supplementary Note 8) suggested that PCA facilitated the biosynthesis of MtrC and OmcA, thereby accelerating electron transmembrane transfer. Furthermore, the scanning electron microscopy (SEM) and confocal laser scanning microscope (CLSM) observations revealed that strain SDV3 exhibited a thicker biofilm with a higher cell density on electrode surface compared to those of WT (Fig. 5c). Quantitative assessment of biofilm biomass revealed the strain SDV3 biofilm reached 285.30 ± 4.12 µg cm−2, 4.24-fold higher than that of the WT biofilm (67.28 ± 4.64 µg cm−2) (Fig. 5d). And the viable cell number in the strain SDV3 biofilm was 8.47 ± 0.95 × 107 cm−2, 5.39-fold higher than that of the WT biofilm (1.57 ± 0.21 × 107 cm−2) (Fig. 5d). Moreover, the electrochemical impedance spectroscopy (EIS) analyses revealed the charge-transfer resistance of the strain SDV3 biofilm was 2113.74 Ω, 9.35-fold lower than that of WT (19756.41 Ω, Fig. 5e), indicating PCA significantly improved the biofilm conductivity, leading to reduction in electron transfer resistance.
a Lactate consumption. b Sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) with heme staining. c CLSM (upper row) and SEM (bottom row) images of anodic biofilms. Scale bars of SEM and CLSM images were 5 µm and 0.2 mm, respectively. For gel and micrographs, the reported results have been checked for consistency with 3 individual experiments. d Assay of biomass and colony-forming units (CFU) of viable cells attached on anode surfaces. e Nyquist plot of electrochemical impedance spectroscopy. f The cyclic adenosine 3’, 5’-monophosphate (cAMP) levels in strains SDV3 and WT. g Mechanisms on PCA-enhanced cellular metabolism and biofilm formation of S. oneidensis. In strain SDV3, the synthesized PCA enhances cAMP synthesis and improves the level of intracellular cAMP. The increased cAMP binds with CRP forming a complex, which activates the transcription of dld, lldP, mtrCAB, and omcA, thereby regulating the lactate catabolism and c-Cyts biosynthesis. In addition, cAMP-CRP complex binds with BpfD, thus facilitating the interaction of BpfD with protease BpfG, which reduced proteolytic processing of adhesin, BpfA, thus enhancing release of BpfA and biofilm formation. Abbreviations: cAMP cyclic adenosine 3’, 5’-monophosphate, CRP cyclic adenosine 3’, 5’-monophosphate receptor protein, MQ methyl naphthoquinone, ATP adenosine triphosphate, CyaC class III adenylate cyclase, MtrC and OmcA decaheme c-Cyts, MtrB β-barrel trans-OM protein, MtrA periplasmic decaheme c-Cyts, CymA inner membrane tetraheme c-Cyts, Dld D-lactate dehydrogenase, LldP lactate transport protein, NqrBCDEF Na+-translocating NADH-quinone reductase, BpfA cell surface-associated adhesin, AggA type I protein secretion system secretin component, BpfG protease, BpfD putative c-di-GMP effector. h Current density at different stages. Stage I: inoculation of the strain SDV3 cells poised at 0.24 V (vs. the standard hydrogen electrode) until the current was stable. Stage II: the strain SDV3 cells were reset into a fresh medium in the absence of the inducer IPTG, thus PCA biosynthesis ceased and the PCA-mediated EET pathway was thus disrupted. Results in (a, d, and f) from three independent experiments (n = 3) were expressed as means and standard errors, and data in (a, d, and f) are shown as the mean ± SD. The results in (b, c, e, and h) have been checked for consistency with 3 individual experiments. Source data are provided as a Source Data file.
The aforementioned biochemical characterizations demonstrated that PCA enhanced lactate consumption, c-Cyts biosynthesis, and biofilm formation. However, the mechanism through which PCA modulated metabolism and behavior of Shewanella remained unclear. According to a previous study73, the cyclic adenosine 3’, 5’-monophosphate (cAMP)-cyclic adenosine 3’, 5’-monophosphate receptor protein (CRP) complex could bind with the promoter regions of mtrCAB (encoding the outer membrane c-Cyts complex MtrCAB), omcA, dld (encoding D-lactate dehydrogenase), and lldP (encoding lactate transport protein) to stimulate their transcription, thus strengthening the c-Cyts biosynthesis and lactate catabolism. Meanwhile, the class III adenylate cyclase CyaC, which catalyzes cAMP synthesis from ATP, could regulate its catalytic activity in response to the changes in redox state74. We thus speculated that PCA could potentially enhance the synthesis of cAMP in strain SDV3, thus promoting lactate catabolism and c-Cyts biosynthesis. Subsequent analysis of intracellular cAMP levels in strains SDV3 and WT revealed a significantly higher concentration in strain SDV3 (3.65 ± 0.23 pmol mg−1 protein), representing a 2.74-fold increase compared to WT (1.33 ± 0.23 pmol mg−1 protein) (Fig. 5f). Through RT-qPCR analysis (Supplementary Fig. 24), the transcription levels of the genes dld, lldP, mtrCAB, and omcA were up-regulated in strain SDV3 compared to those of WT, further accounting for the enhanced lactate consumption and c-Cyts biosynthesis. Additionally, in the course of biofilm formation, a recent study reported that cAMP-CRP could directly interact with a putative c-di-GMP effector of BpfD in Shewanella75. This interaction between cAMP-CRP and BpfD could strengthen the existing interaction between BpfD and protease BpfG, ultimately inhibiting the proteolytic activity and releasing a cell surface-associated adhesin of BpfA, thereby promoting biofilm formation75. In summary, a model of PCA-enhanced cellular metabolism and biofilm formation in S. oneidensis was refined (Fig. 5g). In strain SDV3, the synthesized PCA facilitated synthesis of cAMP, which bound with its receptor protein (CRP) to form a complex. The cAMP-CRP complex activated the transcription of the genes encoding c-Cyts (mtrCAB, and omcA) and enzymes (dld and lldP) involved in lactate catabolism. The elevated level of cAMP also promoted the release of the cell surface-associated adhesin BpfA, thereby facilitating biofilm formation.
The dominant mechanism underlying PCA-boosted EET was elucidated. It was shown PCA not only mediated electron transfer acting as an electron shuttle, but also improved cellular metabolism and biofilm formation, both of which could contribute to the increased EET rate. To further identify the dominant mechanism of the PCA-boosted EET, we inoculated strain SDV3 with its synthesized PCA in the three-electrode BES system poised at 0.24 V (vs. standard hydrogen electrode). A stable oxidation current density of 443.53 µA cm−2 was observed after 32 h’ incubation (stage I, Fig. 5h). Upon transfer of the SDV3 cells into a new medium in the absence of the inducer IPTG, PCA biosynthesis ceased, which led to the disruption of PCA-mediated EET. Then, the oxidation current dramatically dropped to 129.10 µA cm−2 (stage II, Fig. 5h). It was thus indicated PCA acting as the electron shuttle was the dominant mechanism underlying PCA-boosted EET in S. oneidensis.
