Enzymes (Phanta Max Super-Fidelity DNA polymerase, FastDigest restriction enzymes, and T4 DNA ligase) were purchased from Thermo Fisher Scientific (Shanghai, China). Kits for DNA purification, plasmid and genome extraction were supplied by TianGen (Beijing, China). All chemical reagents were purchased from Dingguo Changsheng Biotech (Tianjin, China). FAD, FMN, and RF were purchased from Sigma–Aldrich (Shanghai, China).

RF-overproducing S. oneidensis strain, namely C5 (Yang et al., 2015), was constructed previously and used as the base strain in this study. The plasmid PHG13 was used for gene expression (Cao et al., 2019). The strains and plasmids used in this study are listed in Table 1. For plasmid propagation, E. coli Trans1 T1 was grown in Luria-Bertani (LB) medium. For conjugation, strain E. coli WM3064 was cultivated in LB supplemented with 100 μg/mL 2, 6-diaminopimelic acid (DAP). E. coli strains were grown at 37°C with shaking at 220 rpm. S. oneidensis and derivative strains were cultivated in LB or anode solution (including M9 buffer, LB medium, MgSO₄, CaCl₂, and 20 mM lactate) at 30^°C with shaking at 200 rpm. When required, antibiotics were added to a final concentration with 50 μg/mL kanamycin and 34 μg/mL chloramphenicol.

The sequences of primers used for PCR amplification are listed in Supplementary Table 1. The genes ccpA, ahpF, and ahpC were amplified using corresponding primer pairs ccpA-F/R, ahpF-F/R, and ahpC-F/R with S. oneidensis MR-1 genome as a template. The resulting fragments and plasmids PHG13-CB, PHG13-CT, and PHG13-CL were digested with the restriction endonuclease NdeI/NheI, followed by ligation, leading to PHG13-CB-gene, PHG13-CT-gene, and PHG13-CL-gene, respectively. Genetic assembly of ahpC and ccpA was constructed by BioBrick standards (Cao et al., 2019). Plasmid PHG13-CL-ccpA was digested with AvrII and NheI, then ligated into the plasmid PHG13-CL-ahpC that was digested by NheI, resulting in plasmid PHG13-CL-ahpC-ccpA.

The CRISPR base editing mediated by dCas9 and activation-induced cytidine deaminase (AID) has a 5 nt target window that all C within −16 to −20 positions upstream of the PAM sequence can be edited to T, which results in the stop codons (TAG, TAA, and TGA) (Komor et al., 2016; Kim et al., 2017). For CRISPR/dCas9-AID genome editing in S. oneidensis, dCas9-AID and sgRNA in plasmid 104 were expressed to obtain the desirable mutations (Supplementary Figure 1). To construct sgRNA expression plasmids, the pairs of 24-bp primers (Supplementary Table 2) were annealed to form a double-strand DNA with cohesive ends, which were ligated to BsaI sites in plasmid 104 using Golden Gate assembly (Wang et al., 2018).

S. oneidensis strains transformed with the sgRNA expression plasmids were plated on LB-agar plates supplemented with 0.8 mM IPTG and 50 μg/mL kanamycin. Plates were incubated 18–24 h at 30^°C until colonies were visible. For the complete base mutation, colonies in plates were incubated in LB-liquid medium with IPTG and kanamycin for 12 h. For plasmid curing, cells were transferred to LB-liquid culture without antibiotic and incubated overnight at 30^°C. Subsequently, cell suspensions were streaked on fresh LB-agar plates and incubated overnight to generate single colonies. These colonies were plated on LB plates containing kanamycin and incubated at 30°C to test them for plasmid loss.

Fermentation was performed as follows: a cryopreservation stock stored at −80°C was streaked on fresh LB-agar plates and incubated at 30°C until colonies were visible. After that, the colonies on the plate were suspended in 3 mL LB medium and incubated for 12 h to acquire the seed suspension, which was used to inoculate 30 mL test tubes containing 5 mL of the fermentation medium, adjusting the initial OD₆₀₀ to 0.01. The fermentation medium was LB medium supplemented with 20 mM lactate and corresponding antibiotics depending on plasmids. The inducers were added at the early exponential phase (OD₆₀₀ = 0.5–0.6) with corresponding concentrations (0.75 mM IPTG for promoter placUV5, 10 mM arabinose for promoter pBad, and 1,000 ng/ml anhydrotetracycline hydrochloride for promoter pTet). The fermentation was conducted by inducing for 12 h, and the cell growth and the RF production were measured subsequently.

The production of extracellular RF was measured by using High Performance Liquid Chromatography (HPLC) with a UV detector (Shimadzu, Japan). Fresh samples of cell cultures were harvested by centrifugation, and all standard solutions and sample supernatants were filtered using a 0.45 μm syringe filter, and then assayed by a reverse phase C18 column (5 μm particle size, 250 mm × 4.6 mm, Thermo Fisher Scientific). The column was equilibrated into 30% (v/v) methanol and NaH₂PO₄, an isocratic eluent was applied over 25 min at a flow rate of 0.6 mL/min at 30°C with 20 μL injection volume. The range of concentration for standard solutions were 1, 10, 50, 100, 200, and 400 mg/L. The elution times were around 7.2 min for FAD, 11.9 min for FMN, and 19.1 min for RF, respectively.

To evaluate the efficiency of EET, the overnight S. oneidensis culture suspension (0.5 mL) was inoculated into 50 mL fresh LB medium supplemented with corresponding antibiotics and IPTG. After incubated for 12 h, the cell suspension concentration was adjusted OD₆₀₀ to 0.5 and dispersed into the bioanode of MFCs. When required, 50 μg/mL kanamycin and 1 mM IPTG were added in anolyte to ensure consistent culture conditions. All MFCs were cultivated in a 30°C incubator and each group was tripled for parallel experiments.

Dual-chamber MFCs with a working volume of 110 mL were applied in this study. Carbon cloth (CeTech, China) was used as the electrodes for both anode (1.0 cm × 1.0 cm) and cathode (2.5 cm × 3 cm). The Nafion 117 membrane (DuPont Inc., United States) was used to separate the anode and cathode. The anolyte consisted of M9 buffer (Na₂HPO₄, 42.3 mM; KH₂PO₄, 22.1 mM; NaCl, 8.5 mM; NH₄Cl, 18.7 mM; MgSO₄, 1 mM; CaCl₂, 0.1 mM), supplemented with 20 mM lactate and 5% (v/v) LB broth. The cathodic electrolyte was composed of 50 mM K₃[Fe (CN)₆], 50 mM KH₂PO₄, and 50 mM K₂HPO₄. Besides, to measure the voltage generated, a 2 kΩ external resistor was connected to the external circuit of MFCs.

The voltage output was recorded using a digital multimetera (DT9205A). The cyclic voltammetry (CV) curve and linear sweep voltammetry (LSV) curve were conducted during the plateau of voltage output of MFC. CV curve was performed in a three-electrode configuration included an Ag/AgCl reference electrode (Aida Hengsheng Co., China) with a 1 mV/s scan rate in the range from −0.7 to 0.1 V. Polarization curve and power density curve were obtained by LSV which was decreasing the potential at a rate of 0.1 mV/s from OCP (open circuit potential). Both the CV curve and LSV curve were monitored by an electrochemical workstation CHI1000C (CH Instrument, Shanghai, China).

The planktonic cells were harvested after 4 or 12 h of cultivation in MCFs by quick centrifugation at 10,000 × g for 1 min and immediately frozen in liquid nitrogen. Total RNA was extracted using the RNAprep pure Cell/Bacteria Kit (Tiangen) following lysozyme treatment. RNA degradation and contamination were monitored on 1% agarose gels. RNA integrity was assessed on a 2100 Bioanalyzer (Agilent Technologies, CA, United States). RNA concentration was measured by Qubit^® RNA Assay Kit in Qubit^® 2.0 Flurometer (Life Technologies, CA, United States). rRNA was removed using a Ribo-zero kit that left the mRNA. A total of 3 μg RNA was used as an input per sample. Sequencing libraries were generated using a NEBNext^® Ultra™ RNA Library Prep Kit for Illumina^® (NEB, United States) following the manufacturer’s instructions. Index codes were added to attribute sequences to each sample. According to the manufacturer’s instructions, clustering of the index-coded samples was performed on a cBot Cluster Generation System using the TruSeq PE Cluster Kit v3-cBot-HS (Illumina). The library preparations were sequenced on an Illumina Hiseq 4000 platform and paired-end reads were generated. The data were analyzed by Beijing Novogene Bioinformatics Technology Co., Ltd. (China). Specifically, raw reads were firstly processed through in-house perl scripts, generating clean data with high quality. The index of reference genome was built and clean reads were aligned to reference genome¹ by using Bowtie2 v2.2.3. HTSeq v0.6.1 was used to count the read numbers mapped to each gene. Then, differential expression analysis of the C5 strain compared with the PYYDT strain was performed using the DEGSeq v1.12.0. p-values were adjusted using the Benjamini and Hochberg method and corrected to q-value. Lists of differentially expressed genes [abs (log₂ fold change) > 1 and q-value < 0.005] in the C5 strain compared with the PYYDT strain at 4 and 12 h are finally presented in Data Sheet 1 and Data Sheet 2, respectively.

The engineered S. oneidensis strain C5 expressing the RF biosynthetic genes (ribA, ribD, ribE, ribH, and ribC) in plasmid PYYDT exhibited a dramatically improved EET capability in MFC, and the maximum power density reached 233.0 ± 24.9 mW/m², which was 14.2 folds of the PYYDT strain that transformed with an empty vector PYYDT (Yang et al., 2015). To exploit beneficial cellular responses associated with flavins-mediated EET, we performed a comparative transcriptome analysis of the C5 and the PYYDT strains. Cells incubated in MFC were sampled at 4 and 12 h, the early and middle exponential phase of the output voltage (Supplementary Figure 2). As shown in Figure 1A, there were 602 differently expressed genes in the C5 strain compared with the PYYDT strain at 4 h, including 372 upregulated genes [log₂ (fold change) > 1 and q-value < 0.005)] and 230 downregulated genes [log₂ (fold change) < −1 and q-value < 0.005] (Data Sheet 1). There were 501 differently expressed genes in the C5 strain compared with the PYYDT strain at 12 h, including 280 upregulated genes [log₂ (fold change) > 1 and q-value < 0.005] and 221 downregulated genes [log₂ (fold change) < −1 and q-value < 0.005] (Figure 1B, Data Sheet 2). The results indicated that in the C5 strain, higher EET capability may be affected not only by the engineering of flavin synthetic pathway but also secondary cellular responses to RF overproduction.

To investigate potential gene targets associated with the RF-dependent EET, we evaluated the effect of their overexpression/inactivation on the production of electricity. Considering the process of gene reverse-engineering, we selected downregulated and upregulated targets at an earlier (4 h) and later (12 h) time, respectively. The top four upregulated genes at 12 h included ccpA, katB, ahpF, and ahpC (Figure 1C). Gene ccpA encodes a peroxidase that involves in the metabolism of c-Cyts. Gene katB encodes the catalase that responses to hydrogen peroxide. Genes ahpC and ahpF are relevant to cellular redox reactions (Seaver and Imlay, 2001). Additionally, the top nine downregulated genes at 4 h included tonB, SO_3344, pubC, pubB, exbB, pubA, putA, putB, and hmuA (Figure 1D). All of these downregulated genes are involved in the process of extracellular iron uptake (Huang and Wilks, 2017). However, the EET capability of these differentially expressed genes has not been investigated in previous studies. Consequently, it is necessary to modulate these gene targets to validate whether the reinforcement of cellular response could enhance the flavins-mediated EET.

To evaluate if overexpression of the upregulated targets can increase EET, we expressed each of these genes by a second plasmid PHG13 in the C5 strain (Figure 2A). To avoid the metabolic burden caused by two plasmids and inappropriate expression intensity, expression of the upregulated genes was fine-tuned by different promoters, including the high-strength pBad promoter, the medium-strength pTet promoter, and the low-strength placUV5 promoter (Cao et al., 2019). Gene katB was failed to be amplified from the genome. Considering the importance of flavin in EET, we firstly assayed RF titer of the engineered strains. Tube fermentation was conducted in LB medium supplemented with 20 mM lactate for 12 h after induction. Cell growth of the engineered strains was measured (Supplementary Figure 3). As shown in Figure 2B, overexpression of the upregulated genes driven by the high-strength pBad or medium-strength pTet promoters severely reduced RF production, while their expression controlled by the low-strength placUV5 promoter had no adverse effect on the RF production. The DL0 strain produced slightly decreased RF compared to the C5 strain (Supplementary Figure 4), which might due to the expression of the empty plasmid PHG13. Particularly, the DL1 strain (overexpressing ccpA by placUV5 promoter in C5 strain) and the DL3 strain (overexpressing ahpC by placUV5 promoter in C5 strain) exhibited higher RF production than the DL0 strain (overexpressing empty plasmid PHG13 in C5 strain). Furthermore, these two targets were co-expressed and the resulted DL4 strain obtained a further enhanced RF production (Figures 2A,B).

To investigate the effect of these upregulated targets on electron transfer, the recombinant strains DL1, DL3, and DL4 were inoculated in anodic chambers of MFCs for measuring the voltage output. The linear sweep voltammetry (LSV) was then performed during the plateau of voltage to measure the power density. As shown in Figure 2C, the maximum power densities of the DL3, DL1, and DL4 strains reached 432.18 ± 131.18, 364.09 ± 16.20, and 380.22 ± 55.05 mW/m², respectively, which were 1.31, 1.10, and 1.15 folds of the DL0 strain. These results indicated that overexpression of genes ahpC and ccpA controlled by the low-strength placUV5 promoter further facilitated RF-mediated EET.

To assess the effect of downregulated targets on EET, genes pubB, pubC, pubA, tonB, exbB, hmuA, putA, and putB (Figure 1C) were inactivated through CRISPR base editing. CRISPR/Cas9-AID converts C base to T base and induces nonsense mutations (Wang et al., 2018). The inactivated sites of the selected genes were shown in Figure 3A. Gene SO_3344 was failed to design the inactivated sites. To evaluate whether these gene inactivations impair cell growth and RF metabolism, tube fermentation of the engineered strains was performed in LB medium with 20 mM lactate for 12 h. As shown in Figure 3B and Supplementary Figure 3, modification of these genes did not decrease RF production and cell growth. Particularly, the BE1 (inactivating pubA), BE5 (inactivating putB), and BE7 (inactivating tonB) strains obtained increased RF production. Genes pubA, putB, and tonB were further combinatorically inactivated, resulting in the BE9 strain. The combination of these genetic perturbations has no negative effect on cell growth (Supplementary Figure 5) and exhibits only a slightly decreased RF production (Figure 3B).

The BE1, BE5, BE7, and BE9 strains were applied for further detection of electrochemical characteristics. These engineered strains were inoculated in anodic chambers of MFCs and the voltage output was measured. The LSV was then performed during the plateau of voltage to measure the power density. As shown in Figure 3C, compared to the wild-type strain (WT), the BE1 and BE5 strains showed reduced electron transfer capability, while the BE7 and BE9 strains exhibited improved EET capability. The maximum power densities of the BE7 and BE9 strains reached 74.78 ± 4.93 and 62.19 ± 4.35 mW/m², which were 1.72 and 1.43 folds of the WT strain, respectively. These results confirmed that simultaneous inactivation of gene targets tonB, pubA, and putB, or inactivation of tonB, can greatly accelerate the electron transfer in S. oneidensis.

Our studies indicated that overexpression of ahpC and ccpA efficiently improved the EET capability of S. oneidensis (Figure 2C). To exploit the corresponding mechanism, we analyzed the functions of beneficial genes. A previous study indicated that the increased RF was prone to generate destructive ROS (Xu et al., 2015), which resulted in severe DNA damage and impaired cell viability (Imlay, 2003). Gene ahpC, encoding the alkyl hydroperoxide reductase peroxiredoxin, enables scavenging the intracellular ROS (Seaver and Imlay, 2001). Likewise, gene ccpA, encoding the periplasmic c-type cytochrome peroxidase, has been reported to catalyze the reduction of hydrogen peroxide, which results in detoxification of ROS (Hoffmann et al., 2009; Schutz et al., 2011). Consequently, we speculated that upregulation of aphC and ccpA could provide a favorable physiological status for EET.

Hemin in c-Cyts can be reduced through the oxidation of RF (Wang et al., 2017), which facilitates the generation of heme and then enhances EET capability. We found that inactivation of genes tonB, pubA, and putB also improved the electricity production of S. oneidensis (Figure 3C). TonB is an energy transducer, which is necessary to transport extracellular iron into intracellular (Minandri et al., 2016; Balhesteros et al., 2017). The siderophore putrebactin, encoded by gene pubA, involves ferric iron uptake (Kadi et al., 2008). The reductase encoded by putB can catalyze the generation of ferrous iron (Liu et al., 2018). Therefore, we speculate that the inactivation of tonB, pubA, and putB hinders ferric iron entering into cytoplasm and increases its extracellular availability. The increase of ferric iron could facilitate hemin accumulation and accelerate its conversion to heme, which promotes the transfer of the intracellular electron (Kranz et al., 2009).

In the above study of exploration of beneficial genes, the DL3 (expressing ahpC in C5 strain) and DL4 (expressing ahpC and ccpA in C5 strain) strains as well as the BE7 (inactivating tonB) and BE9 (inactivating tonB, pubA, and putB) strains obtained improved power density. To further accelerate the electron transfer, we combined gene perturbation in these strains and resulting in four recombinant strains, CM1 (expressing ahpC and inactivating tonB in C5 strain), CM2 (expressing ahpC and inactivating tonB, pubA, and putB in C5 strain), CM3 (expressing ahpC, ccpA, and inactivating tonB in C5 strain), and CM4 (expressing ahpC, ccpA, and inactivating tonB, pubA, and putB in C5 strain) (Figure 4A). We firstly assayed RF production of the engineered strains, and found that the combined gene perturbation reduced the RF titer compared to the DL0 strain (Figure 4B). We then conducted LSV to evaluate the electrochemical characteristics of these recombinant strains. The maximum power density of the CM4 strain reached 651.78 ± 124.60 mW/m², which was 1.97 and 1.51 folds of the DL0 and DL3 strains, respectively (Figure 4C). The results demonstrated that the combination of upregulating ahpC and ccpA and inactivating tonB, pubA, and putB significantly enhanced flavin-mediated EET capability in S. oneidensis.

EET is a sophisticated network that involves the coordination of multiple intermediates and complex interactions (Lubner et al., 2017; Ikeda et al., 2021). Thus, many in-depth targets might also lead to unexpected improvement for the desired phenotype. Transcriptome analysis has provided an engineering blueprint for metabolic rewiring and enabled unveiling non-intuitive key targets in the process of cellular metabolism (Ajjawi et al., 2017; Fang et al., 2021), which facilitate for exploring underlying mechanisms on cellular responses (Huang et al., 2017; Liu et al., 2019). In this study, we utilized transcriptome analysis to investigate beneficial gene targets in RF-producing strain. We found that the MtrABC complex genes were upregulated at the transcript level in the C5 strain compared with the PYYDT strain, while the transcript of riboflavin biosynthetic genes (ribA, ribD, ribE, and ribC) changed slightly (Supplementary Table 3). The results suggested the enhanced flavin-mediated EET capability in the C5 strain. We identified five novel targets ahpC, ccpA, putB, pubA, and tonB that were beneficial to flavins-dependent EET. Notably, these genes have not been engineered for EET in previous studies. Overexpression of ahpC and ccpA were speculated to reduce the destructive ROS. Inactivation of putB, pubA, and tonB might increase heme availability, which accelerate the electron transmission. Thus, we identified novel clues for improving power density and elucidating the mechanism of RF-mediated EET.