A 35-fold enhancement in photocurrent generation of Synechocystis sp. PCC 6803 by outer membrane deprivation

Biophotovoltaics (BPV) is an emerging technology developed to utilize reducing equivalents produced by photosynthetic organisms. It generates electrical power by exploiting a phenomenon called extracellular electron transfer (EET), where reducing equivalents are transferred extracellularly to exogenous electron acceptors. Although cyanobacteria have been extensively studied in BPV because of their high photosynthetic activity and ease of handling, their extremely low EET activity remains a limitation. In this study, we achieved a 35-fold enhancement in photocurrent generation of the cyanobacterium Synechocystis sp. PCC 6803 by deprivation of the outer membrane, where electrons are suggested to stem from NADPH; this, along with a signicantly higher rate of exogenous ferricyanide reduction, veried that low permeability of the outer membrane contributues to low cyanobacterial EET activity. Moreover, outer membrane deprivation enhanced extracellular derivation of reducing equivalents serving as respiratory substrates for other heterotrophic bacteria, making it promising and useful for effective derivation of reducing equivalents.


Introduction
Recently, various technologies have been developed to utilize reducing equivalents produced by oxygenic photosynthesis, which absorbs light and oxidizes water to produce high energy electrons. Emerging among them is biophotovoltaics (BPV), in which the reducing equivalents are transferred extracellularly to electrodes or to other bacteria, a phenomenon called extracellular electron transfer (EET), eventually generating electrical power [1][2][3][4] . Cyanobacteria, which are gram-negative bacteria capable of performing oxygenic photosynthesis, have been extensively studied in the eld of BPV research [1][2][3] . This is because cyanobacteria have higher photosynthetic activity than terrestrial plants, much like eukaryotic algae [5][6][7] .
Moreover, they are easy to culture, amenable to genetic manipulation 8 , grow fast, and possess a simple cell structure compared with eukaryotes. Many studies on the mechanism of EET have been conducted using cyanobacteria [9][10][11][12][13][14][15] , and so far, direct EET via conductive nanowires 16 and indirect, mediated EET by endogenous mediators 14,15,17 have been put forward as possible EET pathways; some suggest that the latter is more likely than the former to occur in the case of cyanobacteria 2,18 .
Although much progress has been made in the eld of BPV in the past decade, the extremely low EET activity of cyanobacteria remains a limitation. The EET activity of cyanobacteria, both in the dark and under illumination, is extremely low compared with that of mineral-reducing, electricity-generating bacteria, e.g. the genera Shewanella and Geobacter 2 , which are capable of utilizing diverse electron acceptors including an anode 19 . As previously pointed out, the main reason for the low EET activity of cyanobacteria is their autotrophy, in which EET could be totally useless and wasteful because electrons originating from phosynthesis should be fully utilized to provide enough reducing equivalents and energy to x carbon 1 . Here, we hypothesize an additional reason for the extremely low EET activity of cyanobacteria: the low permeability of the outer membrane. The outer membrane of cyanobacteria exhibits more than 20-fold lower permeability to organic substrates than that of Escherichia coli, the model gram-negative bacteria 20 . This low permeability is thought to re ect its autotrophic life style 20,21 , which does not necessarily rely on uptake of extracellular nutrients, although various transport systems do exist and function in cyanobacteria 22-24 . In this study, using an outer membrane-deprived Synechocystis sp. PCC 6803 (hereafter Synechocystis) mutant, slr0688i 25 , we achieved a signi cant enhancement in cyanobacterial EET activity. EET activity was evaluated as extensively as possible in terms of photocurrent generation, ferricyanide reduction, and electron donation capacity as respiratory substrates (Fig. 1). This study clearly veri ed our hypothesis that the low permeability of the outer membrane contributes to the extremely low EET activity of cyanobacteria.

Results
Mediated EET was enhanced by outer membrane deprivation.
First, to compare the EET activity of the outer membrane-deprived mutant, slr0688i 25 , with that of the wild type, dCas9, the photocurrent from both strains was measured. Slr0688i and dCas9 resuspended in their respective supernatants were injected by gravity onto at indium tin oxide (ITO) electrodes to which + 0.25 V vs. Ag/AgCl was applied, and the current generated under illumination was recorded. Slr0688i generated 35 times as much photocurrent as dCas9 on average ( Fig. 2(a)), and approx. 100 times as much photocurrent (0.4 µA/cm 2 ) as dCas9 (0.007 µA/cm 2 ) at maximum ( Fig. 2(b)) with at ITO electrodes. To con rm that the observed enhancement in photocurrent generation of slr0688i did not stem from increased photosynthetic activity, the oxygen evolving activities of slr0688i and dCas9 were measured and compared. Under illumination with high light (750 µmol photons m − 2 s − 1 ), neither photosynthetic nor respiratory activity was signi cantly different between them on a chlorophyll basis ( Supplementary Fig. 1); moreover, under illumination with moderate light (120 µmol photons m − 2 s − 1 ), the photosynthetic activity of slr0688i was slightly lower than dCas9 ( Supplementary Fig. 1). These results indicated that the enhanced photocurrent generation of slr0688i is not attributable to increased photosynthetic activity but to its outer membrane deprivation.
Next, to examine the necessity of supernatants in photocurrent generation, the supernatant of slr0688i was substituted with that of dCas9 or fresh BG11, and the photocurrent was measured. Photocurrent generation from slr0688i was not affected by substitution with the supernatant of dCas9 (Fig. 2(c); orange solid versus dotted line) but was abolished by substitution with fresh BG11 (Fig. 2(c); orange versus red solid line). In contrast, dCas9 did not generate a photocurrent with the supernatant of either slr0688i (( Fig. 2(c); black dotted line) or dCas9 (( Fig. 2(c); black solid line) under our experimental conditions. These results demonstrated that redox-active, secreted compounds included in the supernatants of slr0688i and dCas9 serve as electron mediators in photocurrent generation. Some anodic peaks were observable in cyclic voltammograms taken from slr0688i resuspended in its supernatant ( Supplementary Fig. 2), which are possibily attributable to redox-active compounds secreted in the supernatant. These observations are consistent with recent suggestions that cyanobacterial secreted compounds serve as endogenous electron mediators in photocurrent generation from wild-type cells 2,14,15 . In addition, a photocurrent was observed with the supernatant of slr0688i containing compounds with MW < 3,000 but not with MW > 3,000 ( Supplementary Fig. 3), which is consistent with the widely accepted prediction that small molecular compounds such as quinones and avines might be responsible for photocurrent generation 2 . Taken together, these results indicated that outer membrane deprivation facilitates mediated EET, which requires secreted redox-active compounds regardless of deprivation of the outer membrane.
Electrons are suggested to stem from NADPH.
To identify the main site where the reducing equivalents involved in photocurrent generation are derived from slr0688i, several photosynthesis inhibitors were applied to Synechocystis cells and their effects on current generation were examined. The effects of the inhibitors were evaluated by total coulombs produced from both the photosynthetic and respiratory chains of the cells during a 10-min illumination, because the photosynthetic chain shares some components with the respiratory chain in Synechocystis cells [26][27][28] . slr0688i was treated with p-chloromercuribenzoate (pCMB), 3-(3,4-dichlorophenyl)-1,1dimethylurea (DCMU), or potassium cyanide (KCN) prior to injection into the chambers, and the photocurrent was recorded at + 0.25 V vs. Ag/AgCl. Treatment with pCMB led to a 30-fold increase in the total coulombs (% vs controls) ( Fig. 3(a)), most of which were produced light-dependently (Supplementary Fig. 4(a)). When slr0688i was treated with pCMB, oxygen evolution under illumination was inhibited ( Supplementary Fig. 5(a)). Then, the effect of pCMB on changes in NADPH uorescence was examined with a pulse modulation uorometer (Dual-PAM-100); pCMB-treated slr0688i cells generated NADPH (approximate half-time, 0.21 sec) as fast as cells incubated only with the solvent DMSO (approximate half-time, 0.17 sec), under illumination with actinic light (Fig. 4). This indicated that pCMB inhibits photosynthesis (Supplementary Fig. 5(a)), but its target site is not located before generation of NADPH. When treated with DCMU, which unequivocally inhibits electron transfer from Q A to Q B in PSII 29,30 , the total coulombs produced by slr0688i dropped to 52% of that of the control ( Fig. 3(a)), where only residual current was generated light-dependently (Supplementary Fig. 4(b)). These results indicated that electrons are derived from somewhere after Q A .
When treated with KCN, the total coulombs produced by slr0688i under illumination was only 14% of that of the control ( Fig. 3(a)), and no photocurrent was generated ( Supplementary Fig. 4(c)). KCN inhibits all the respiratory terminal oxidases 31-33 as well as plastocyanin [34][35][36] ; indeed, KCN-treated slr0688i cells did not exhibit oxygen evolution under illumination (Supplementary Fig. 5 Consistent with previous reports 2,11,12 , the results described above suggested that the electrons involved in photocurrent generation stem from NADPH ( Fig. 3(b)). This was further supported by an additional experiment using phenylmercuric acetate (PMA), an inhibitor of ferredoxin and/or FNR 37,38 ; treatment with PMA abolished the photocurrent of slr0688i ( Supplementary Fig. 6). When steady-state NADP(H) contents and the NADP + /NADPH ratio were quanti ed, there were no signi cant differences between slr0688i and dCas9 ( Supplementary Fig. 7); this indicated that the enhanced photocurrent generation from slr0688i is not due to an increase in the amount of reducing equivalents but to outer membrane deprivation.
Ferricyanide-mediated EET was enhanced by outer membrane deprivation.
Because the catalytic activity of a at ITO electrode is very poor, the EET activity of Synechocystis cells could not be fully evaluated by photocurrent measurements. Therefore, the rate of EET to a hydrophilic electron acceptor, ferricyanide, was measured next. Cell suspensions of slr0688i and dCas9 were mixed with 1 mM potassium ferricyanide, and changes in the ferricyanide concentration under illumination or in the dark were monitored spectroscopically. Under illumination, slr0688i reduced the extracellularly added ferricyanide at a rate of 2.3 µM/OD min, whereas dCas9 exhibited no reduction of ferricyanide in the tested time range (Fig. 5(a)(c)). In the dark, slr0688i reduced ferricyanide at a lower rate (0.8 µM/OD min) than under illumination but still at a signi cantly higher rate than dCas9 in the same conditions ( Fig. 5(b) (d)). Substitution of the supernatant with fresh BG11 did not affect the rate of ferricyanide reduction either under illumination or in the dark (Fig. 5), indicating that ferricyanide-mediated EET, which could not be detected by the photocurrent measurements ( Fig. 2(c)), was detected by this assay. It is widely accepted that, with its hydrophilicity, ferricyanide freely passes the outer membrane and accepts electrons from Synechocystis cells 39 via some transmembrane protein at the cytoplasmic membrane 11,12 ; however, these results indicated that the outer membrane does serve as a barrier to ferricyanide access to the cytoplasmic membrane. Together with the results of the photocurrent measurements, these results clearly showed that deprivation of the outer membrane leads to a signi cant increase of both mediated and ferricyanide-mediated EET.
Extracellular derivation of reducing equivalents was enhanced by outer membrane deprivation.
In an attempt to detect and evaluate EET activity more extensively, we examined the electron donation capacity of extracellularly derived reducing equivalents as respiratory substrates for other bacteria. These extracellular reducing equivalents, which are considered energy carriers, i.e. electron donors, in the BPV system 4 , cannot be detected and evaluated by photocurrent measurements or a ferricyanide reduction assay. Therefore, Bacillus cereus (hereafter Bacillus), which is a gram-positive model organism, was chosen as a recipient of these reducing equivalents. Changes in its membrane potential were monitored as an indicator of electron donation to the respiratory chain because membrane potential generation should be accompanied by extrusion of protons due to electron transfer through the respiratory chain.
Supernatants of slr0688i and dCas9 were added to carbonyl cyanide m-chlorophenylhydrazone (CCCP)treated Bacillus cells and the generation of membrane potential in the Bacillus cells was monitored using the voltage-sensitive dye 3,3′-dipropylthiadicarbocyanine iodide (DiSC 3 (5)). To generate enough membrane potential to calculate the rates of formation, the supernatants were concentrated to 5× by ultra ltration and the obtained retentates (containing compounds with MW > 3,000) were added to Bacillus cells. Generation of membrane potential was twice as fast with the supernatant of slr0688i as that of dCas9 ( Fig. 6(a)), indicating that the slr0688i supernatant possessed higher electron donation capacity as respiratory substrates compared with dCas9. Experiments with membrane vesicles isolated from Bacillus subtilis indicated that the compounds included in the supernatant of slr0688i do not donate electrons directly to the respiratory chain of Bacillus cells as a mixture of PMS/ascorbate does ( Supplementary Fig. 8); therefore, the observed generation of membrane potential is suggested to result from metabolization of compounds in the supernatant of slr0688i by Bacillus cells, which feeds electrons to their respiratory chain. Consistent with this, compounds with MW < 3,000 were capable of generating membrane potential more than ve times as fast as compounds with MW > 3,000 ( Supplementary Fig. 9), re ecting that simpler, lower molecular weight compounds are metabolized more e ciently and thus donate electrons to the respiratory chain at a faster rate.
To quantify the electron donation capacity as respiratory substrates, the rate of membrane potential generation was measured and compared with those of glucose solutions of different concentrations; a 0.5× concentration of slr0688i supernatant was revealed to generate membrane potential as fast as 4 µM glucose ( Fig. 6(b)); that is, the electron donation capacity of slr0688i supernatant was equal to 8 µM glucose.

Discussion
In this study, using an outer membrane-deprived Synechocystis sp. PCC 6803 (hereafter Synechocystis) mutant, slr0688i 25 , we achieved a 35-fold enhancement in photocurrent generation and a signi cantly higher rate of exogenous ferricyanide reduction. These results clearly indicate that deprivation of the outer membrane enhances EET, i.e. extracellular derivation of reducing equivalents originating from photosynthesis, and veri ed our hypothesis that the low permeability of the outer membrane contributes to low cyanobacterial EET activity. This is the rst report that as much as a 35-fold increase in EET activity can be achieved by genetic engineering 25 , providing clues for further improvement and development of BPV systems.
This study revealed that deprivation of the outer membrane is one of the most promising technologies to improve the e ciency of BPV systems. Based on the data shown in Fig. 5 and the chlorophyll content of slr0688i (5.6 µg Chl (mL OD 730 ) −1 ), the rate of EET from slr0688i to ferricyanide is calculated to be 14.6 nmol electrons nmol chl − 1 h − 1 under illumination. This is higher than the fastest rate reported with terminal oxidase mutants; 2.7 nmol electrons nmol chl − 1 h − 1 in the Cyd/ARTO double mutant, which we calculated using the data reported in Table S2 in Bradley et al., 2013 12 . What is more promising about slr0688i is that the photocurrent generation does not need special treatments other than simple application of planktonic cells by gravity onto a at, unmodi ed ITO electrode. Therefore, further enhancement in photocurrent generation may be achievable by various methods that have already proven effective in other studies: optimization 40 or modi cation of the electrodes 14,41,42 , formation of bio lms on electrodes 14,43,44 , deposition of concentrated biomass onto an electrode followed by dessication 13 , mild micro uidization of the cells 15 , utilization of aged cell cultures de cient in iron 45 , or digital printing of cells on paper 46 . Genetic engineering in addition to outer membrane deprivation would unequivocally be bene cial to further improvement, as already demonstrated in several studies 12,47 .
This study suggests that electrons originating from NADPH are involved in the enhanced photocurrent generation from the outer membrane-deprived mutant. This is consistent with previous studies suggesting that the electrons involved in photocurrent generation stem from PSI electron acceptors, i.e. ferredoxin and/or NADP(H) 2,11,12 . One of the strongest pieces of evidence of EET originating from NADPH is the observation that M55, a mutant lacking functional NDH complexes, exhibited a signi cant enhancement in EET to ferricyanide both in the dark and under illumination 12 , which should result from the overaccumulation of NAD(P)H.
In the NADPH uorescence measurements (Fig. 4), the decrease of uorescence after switching off the actinic light was slower in pCMB-treated cells than in non-treated cells; the approximate half-time of the decrease was 1.38 s and 0.66 s, respectively. This sugests that pCMB-treated cells do not consume NADPH as fast as non-treated cells because some reaction(s) related to the Calvin cycle is inhibited by pCMB. This is consistent with its inhibitory effect on oxygen evolution ( Supplementary Fig. 5(a)), because inhibition of the Calvin cycle leads to a halt of oxygen evolution as shown in, e.g., Table 1 in Takahashi   and Murata, 2005 48 . In addition, a transient increase in uorescence after switching off the light due to the pentose phosphate pathway was observed in non-treated cells, but not in pCMB-treated cells (Fig. 4); this further suggests that a Calvin cycle-related reaction(s) is somehow inhibited in pCMB-treated cells and thus organic compounds that are metabolized by the pentose phosphate pathway and yield NADPH are not generated. Previous studies suggested that the target sites of pCMB are iron-sulfur clusters of isolated ferredoxin 49 as well as of PSI 50 , NAD(P)H dehydrogenase (NDH) 1 and 2 in algal thylakoid membranes 51 , or enzymes involved in the Calvin cycle in intact cyanobacterial cells 52 . A more precise ideti cation of the target sites of pCMB in vivo is now in progress.
Deprivation of the outer membrane is undoubtedly bene cial not only for the improvement of BPV systems but also for enhancing production of various chemicals ranging from biofuels to high-value compounds [53][54][55][56][57][58][59][60] . Supporting this is a previous study showing that slr0688i secretes enough nutrients to support the growth of heterotrophs 25 ; moreover, this study also showed that the supernatant of slr0688i provides respiratory substrates to Bacillus cells as e ciently as 8 µM glucose. Many bene cial aspects of the outer membrane-deprived mutant remain to be explored, and full utilization of its photosynthetic reaction is a promising way to achieve a clean and sustainable future.

Materials And Methods
Bacterial strains and general growth conditions. The Synechocystis strains used in this study were slr0688i, whose outer membrane is deprived due to conditional repression of slr0688 by the CRISPRi system regulated by both dCas9 and sgRNA, and dCas9, a control strain (serving as the wild-type strain in this study) whose chromosomal DNA contains the same genetic construct as slr0688i without the sgRNA 25 . These strains were grown photoautotrophically at 30 °C, 100 µmol photons m − 2 s − 1 , with constant shaking at 140 rpm in BG11 medium consisting of the following ingredients (per L): 2 mL of Solution I (0.5 g/L Na 2 EDTA·2H 2 O, 3 g/L ammonium iron (III) citrate, 3 g/L citric acid), 25 mL of Solution II-a (60 g/L NaNO 3  Common procedures for electrochemical measurements performed in the current study were as follows: Synechocystis cell suspensions of 4 mL were separated into sedimented cells and supernatant by centrifugation at 2,500 × g for 10 min, and the supernatant was rst injected into the electrochemical chamber. The sedimented cells were then resuspended with the rest of the supernatant, drawn up into a syringe, injected by gravity onto the ITO glass at the bottom of the chamber, and incubated to settle the cells down. To avoid detection of a pseudo-photocurrent, which is explained in detail below, the pH of Synechocystis cell suspensions was always con rmed and adjusted when necessary prior to electrochemical measurements. The pH of the suspension preferably should be below approximately 8.0, and should never be above 8.5. Otherwise, the buffering capacity of TES (pH 6.8-8.2) is outcompeted by drastic changes (increases) in the pH of the medium due to photosynthesis [61][62][63] , resulting in detection of a pseudo-photocurrent of the following reaction: where Mn 2+ comes from MnSO 4 , a component of the BG11 medium. Some examples of detected pseudophotocurrents of manganese are shown and described in detail in Supplementary Fig. 10. Therefore, before electrochemical measurements, the pH of cell suspensions was always con rmed to be below approximately 8.0, and when necessary, fresh BG11 or HCl was added not only to adjust the optical density, but also the pH of cell suspensions below approx. 8.0. In chronoamperometry, the applied voltage was xed to + 0.25 V vs. Ag/AgCl, at which su cient current generation occurred ( Supplementary Fig. 11) and a psedo-photocurrent was never detected under an appropriate pH (Supplementary Fig. 10).
When the supernatant needed to be replaced, it was removed after centrifugation and substituted with the supernatant of interest, which was prepared by ltration with an Ultrafree-MC-GV, 0.22 µm (Millipore), or with fresh BG11.
Size-fractionated supernatants were prepared as follows: the supernatant was rst ltrated from a cell suspension of slr0688i with a Millex-GV Syringe Filter Unit, 0.22 µm (Millipore) and freeze-dried with a freeze dryer (TOKYO RIKAKIKAI). The obtained powder was suspended with water to yield a 10× concentrated supernatant and size fractionated by ultra ltration with Amicon Ultra Centrifugal Filters (Millipore) to yield a supernatant containing compounds with MW > 50,000, > 30,000, > 3,000 and < 3,000.
Each fraction was then diluted to 0.1× concentration with fresh BG11, and mixed with slr0688i cells to measure photocurrent generation.
For experiments using pCMB, DCMU, and KCN, cell suspensions adjusted to OD 730 = 1.5 and pH 7.6 by adding fresh BG11 were incubated with 100 µM pCMB, 10 µM DCMU, or 5 mM KCN for 1.5 h in the dark on ITO in the electrochemical chambers prior to electrochemical measurements. Additional experiments using 50 µM PMA were performed with cell suspensions whose chlorophyll concentration was 12 µg Chl/mL and pH was adjusted to 7.6 by adding HCl. Steady-state oxygen evolution and uptake measurements. The photosynthetic activity of Synechocystis cells was evaluated with a Clark-type electrode (Hansatech) at 25 °C. To measure steady state photosynthetic activity, 2 mL of cell suspension (12 µg Chl/mL) containing 5 mM NaHCO 3 was illuminated with a CoolLED pE-100 LED at PPFD of 750 µmol photons m − 2 s − 1 . Dark respiration was evaluated from the oxygen consumption rate before switching on the light, and photosynthetic activity was calculated by adding the rate of dark respiration to that of oxygen evolution under illumination.
The same oxygen electrode setup was used to analyze the effects of KCN and pCMB on photosynthetic activity. In the case of KCN, Synechocystis cells were collected and resuspended with fresh BG11 to a nal concentration of 12 µg Chl/mL, and mixed with 5 mM NaHCO 3 and 5 mM KCN, immediately after which oxygen evolving activity was measured. In the case of pCMB, Synechocystis cells collected and resuspended with fresh BG11 (24 µg Chl/mL) were incubated either with 100 µM pCMB or with its solvent DMSO for 1.5 h in the dark. The cell suspensions were then mixed with 5 mM NaHCO 3 and diluted to 12 µg Chl/mL with fresh BG11, followed by immediate measurement with an oxygen electrode.
NADPH uorescence was measured with a DUAL-PAM-100 (Walz) instrument as described previously 64,65 . Slr0688i cells collected and resuspended with fresh BG11 (24 µg Chl/mL) were incubated either with 100 µM pCMB or with its solvent DMSO for at least 1.5 h in the dark. The cell suspensions were then injected and diluted to 2.4 µg Chl/mL in a 1 × 1 cm cuvette installed in the DUAL-PAM. The measuring light of 365 nm was set at an intensity of 10 on the DUAL-PAM software, and the measuring frequency was programmed to increase from 500 to 5000 Hz during illumination with actinic light of 635 nm at 785 µmol photons m − 2 s − 1 .
The amount of NADP + /NADPH in Synechocystis cells was quanti ed using an NADP + /NADPH Assay Kit-WST (DOJINDO), following the manufacturer's manual.
Fluorometric measurement of membrane potential of Bacillus cells. Bacillus cereus (hereafter Bacillus) cells were cultured at 30 °C with constant shaking at 140 rpm in Luria-Bertani (LB) medium. An overnight preculture was diluted 100-fold in fresh LB medium and grown for another 5 h before analysis. Bacillus cells were then collected by centrifugation at 10,000 × g for 4 min and resuspended in fresh BG11 medium, whose OD 600 was adjusted to approx. 0.3. CCCP dissolved in DMSO was added to a nal concentration of 30 µM, and the cell suspension was incubated for 30 min at 30 °C with constant shaking at 140 rpm. The cells were then collected and washed once with BG11 to remove the CCCP, followed by nal resuspension with BG11, in which the cells were allowed to rest for about 40 min (here, the OD 600 of the cell suspention was supposed to be approx. 0.6). A voltage-sensitive dye, DiSC 3 (5), was used to monitor changes in the membrane potential, following Winkel et al. 66 ; depolarization of the membrane is re ected as an increase in uorescence originating from the dye, whereas repolarization leads to a decrease in uorescence 67 . Initial uorescence levels were rst recorded with the cell suspension incubated with 2 µM DiSC 3 (5) for about 5 min prior to the measurement. After the addition of supernatant of either slr0688i or dCas9 of the same volume as the cell suspension, making the concentration of DiSC 3 (5) 1 µM, the uorescence emission spectra of the cell suspension, excited at 610 nm, were recorded from 630 to 700 nm time-dependently with an FP-8500 spectro uorometer (JASCO). Changes in the uorescence intensity at 674 nm were plottted to analyze changes in membrane potential across the cytoplasmic membrane of Bacillus cells.
Casamino acid and glucose dissolved in fresh BG11 were used as positive controls, and fresh BG11 was used as a negative control.
Supernatants were prepared by ltration from cell suspensions of slr0688i or dCas9 with a Millex-GV Syringe Filter Unit, 0.22 µm (Millipore). Concentration and size fractionation were done by ltration with Amicon Ultra Centrifugal Filters (Millipore). All the supernatants and control samples were con rmed to be at the same pH (within approx. pH 7.6-7.8) before the measurements, and when necessary, were adjusted to pH 7.6-7.8 by addition of HCl.
Membrane vesicles free of intracellular components were prepared from Bacillus subtilis according to the method described in Konings et al., 1973 68 .
Note that because the supernatant of slr0688i contained phycocyanins 25 , it yielded unignorable uorescence in the tested range from 630 to 700 nm when excited at 610 nm ( Supplementary Fig. 12). The uorescence from phycocyanins disappeared when purged with oxygen, which leads to selfsensitized bleaching of phycocyanins 69 , or treated with heat ( Supplementary Fig. 12). It was con rmed that phycocyanins did not interefere with changes in uorescence from DiSC 3 (5) (data not shown).