Introduction

Integrating electronic networks with biological systems generates powerful opportunities to reveal insights about complex natural systems1 and provides novel methods to confer precise control over such systems2. Connecting molecular communication networks to electronics is of emerging interest as it takes advantage of the well-established, simple, modular, and programmable nature of electronic devices. Facilitating electronic communication with biology allows for finely programmed behavior of responsive cells. Owing to the ubiquity of electronic devices, we believe that electrogenetic technologies, where electronics are used to control gene expression, can potentially revolutionize synthetic biology2,3 and bioelectronic technologies4,5. By developing the field of electrogenetics, designing gene circuits with increasingly complex behaviors will become more facile, expanding the capabilities of controlling living cells6.

A key challenge in connecting biology and electronics lies in their disparate communication modalities: molecular signals carry information based on structure on the cellular level, whereas the flow of electrons governs electronic communication. However, electronic and cellular circuits operate with similar principles and thus could have compatible signal processing7,8. Because redox activity is one of biology’s most prevalent signaling modalities8, redox activity provides a biologically natural way to interconvert between electronic and molecular, cellular systems. Microbes may be well-suited to convert redox signals into cellular pathway signals, as they contain a multitude of sensors that transduce redox signals into structural changes9. Electrogenetics may repurpose these redox sensors to control gene expression. For instance, redox-active diffusible molecules, or mediators, can transport electrons between an electrode and cells harboring redox-responsive promoters10.

Previous demonstrations using redox for electronic control of gene expression rewired redox-responsive regulons such as soxR, that is responsive to redox-cycling drugs11; oxyR, that is responsive to hydrogen peroxide12; and KEAP1 and NRF2, responsive to reactive oxygen species13. By applying an oxidative potential to mediators ferricyanide and pyocyanin, Tschirhart and others14 toggled bacterial gene expression that is regulated by transcriptional regulator SoxR. With this system, electronic inputs controlled phenotypes such as swimming14 and microbial signaling13, as well as CRISPR-based signal amplification and noise reduction15. For a second, distinct mechanism for electrogenetic control, electrode-generated hydrogen peroxide from the reduction of dissolved oxygen16 was used to electrogenetically modulate the oxyR regulon, enabling control of consortia composition17,18 and improvement of small molecule production in a co-culture19. Electrogenetic signaling is a newly emerging field, and the repertoire of electrochemical inducers and their respective genetic parts is limited. Development of new parts for the electrogenetic toolkit, such as redox-linked elements, will be crucial for designing more complex electronics-controlled genetic circuits, increasing the range of specificity, building bio-electronic devices, and connecting to environments rich in redox signaling.

That is, redox signaling is abundant in biology, in contexts including the gut microbiome20, soil rhizosphere21, and disease and inflammation22. Despite its ubiquitous nature, the mechanisms and networks of redox signaling remain poorly understood23,24,25, and the chemical tools for studying redox species often suffer from limitations26,27. Electrochemical tools offer a different approach for studying redox, and indeed have already shown promise by revealing new insights, through methods such as mediated electrochemical probing28,29.

The oxyRS regulon is a global stress response regulon that enables bacterial cells to adapt to and survive oxidative stresses, typically peroxides. Electronically controlling expression of the oxyRS regulon was recently demonstrated30 and involves the OxyR protein, which in its native state is reduced and inactive. When oxidized by hydrogen peroxide, OxyR undergoes a conformational shift owing to restructuring of disulfide switches31, binds with RNA polymerase, and positively regulates transcription of its dependent promoters, including the promoters for oxyS and oxyR32. We hypothesized that redox mediators in their oxidized state might also directly oxidize OxyR, which has a redox potential (E0) of − 185 millivolts33, or might indirectly lead to oxidized OxyR through the generation of peroxides that, in turn, restructure OxyR disulfide switches.

Here, we investigated redox mediator 1-(4-Hydroxy-3,5-dimethoxyphenyl)ethan-1-one (acetosyringone), a plant-derived phenolic signaling compound34 (E0 =  + 0.5 V35 vs. Ag/AgCl), for its ability to induce OxyR-regulated gene expression. Like many phenolic compounds, acetosyringone (AS) is produced by plant tissues in response to stress and pathogen infection36. The observation that AS contributes to a redox potential burst during the plant’s oxidative response to infection37 suggests that it acts as a redox-modulating agent affecting pathogen responses. Interestingly, oxidized AS was also demonstrated to mediate disulfide bond formation between cysteine molecules38 as well as the thiol groups of thiolated poly(ethylene glycol)35,38, providing a putative mechanism by which AS could interact with the thiol groups of the OxyR subunits and activate OxyR transcriptional responses if it were transported into the cells in an oxidative state or otherwise oxidized intracellularly. Such a phenomenon would indicate AS acting as a redox signal, in contrast to the well-studied AS signaling by molecular structure, as in the vir regulon. Interestingly, the two-component signal transduction process of Agrobacterium tumefacians39 consists of a membrane bound receptor (VirA), and its cognate transcriptional regulator (VirG); these have been transformed into E. coli enabling AS-activated gene expression40, but the oxidation state of the AS signal was not reported. Since AS in the native form is reduced we expect reduced AS is the activating signal. E. coli homologues to the VirA receptor and VirG transcriptional regulator have not been reported. Moreover, the Agrobacterium system does not involve OxyR-based disulfide switches.

In this work, we demonstrate electronic control of gene expression of the oxyS promoter by addition of oxidized acetosyringone, in addition to induction via production of hydrogen peroxide. That is, we show electrogenetic control of oxyRS by bolus addition of oxidized acetosyringone to reporter cells and by applying oxidative potential to a mixture of reporter cells and AS. We further demonstrate that distinct electrochemical mechanisms (oxidation of AS at oxidizing potentials, and the reduction reaction for H2O2 production at reducing potentials) can actuate a single promoter. That is, we show duality in the approach. By applying charge at + 0.5 V, one can oxidize AS and subsequently stimulate oxyRS gene expression. Transitioning towards negative (reduced) voltages has no effect, then transitioning further to a more reducing voltage (~ − 0.5 V) again stimulates expression via the identical promoter. This duality will enable more sophisticated means for electronically programming genetic circuits, potentially including real time measurement and control. Finally, we also show how dynamic application of these transient induction mechanisms can be overlaid to build complex cellular responses. In sum, our findings reveal that oxidized AS is a novel inducer of OxyR, opening up a richer set of possibilities for the prototypical stress-response mechanism and electrogenetics.

Results

Characterization of acetosyringone oxidation

We oxidized a 2 mM solution of acetosyringone (AS) in phosphate buffer (PB) by applying a positive potential to the solution in a half-cell setup41 (Fig. 1a). Applying a potential of + 0.7 V—higher than the E0 of AS—for 0 to 60 min was equivalent to application of 0 to -1.74 coulombs of charge. This gradually turned the AS solution from colorless to brownish-orange, indicating its oxidation42,43 (Fig. 1b) and increasing levels of AS oxidation were characterized spectrophotometrically with an increasing absorbance peak at 490 nm (Fig. 1c). Interestingly, the color change was linear with applied charge (Fig. 1d). AS oxidation was also characterized electrochemically using cyclic voltammetry. When scanning from 0 to 0.7 V, AS had peak currents at 0.47 V (oxidation peak or Epa, peak anodic potential) and 0.417 V (reduction peak or Epc, peak cathodic potential), and both peak currents were attenuated with increasing application of oxidative charge (Fig. 1e). Again, current attenuation was linear with applied charge (Fig. 1f). Oxidation of AS when suspended in phosphate buffer (PB) at pH 7.4 can thus be characterized by absorbance at 490 nm, current at the oxidative peak, and current at the reductive peak, as each of these three metrics correlated to the applied potential.

Figure 1
figure 1

Acetosyringone (AS) electrochemical oxidation is robustly characterized. (a) Electrochemical setup for application of charge to AS solution. Gold wire working electrode, Ag/AgCl reference electrode, and platinum counter electrode (separated by a salt bridge) were submerged in AS solution and connected to a potentiostat. Accumulated charge, optical images of AS solution in a cuvette (b), absorbance spectra (c), and cyclic voltammogram (CV, e) of 2 mM AS which was charged (oxidized) with + 0.7 V for 0–60 min. Images are taken immediately post applied charge using Olympus MVX10 Macro Zoom Fluorescence Microscope and cellSens Standard 2.1. Reduction peak (‘R’) and oxidation peak (‘O’) are marked on CV. Arrows indicate application of charge to AS for a longer duration of time. (d) Correlation of applied charge to absorbance at 490 nm and (f) to currents at oxidation and reduction peaks. Schematic (a) was generated using Microsoft PowerPoint Version 2401 and PubChem Sketcher V2.4. (e) was generated in Matlab R2020a. All other plots were generated in Microsoft Excel Version 2401.

Full size image

Oxidized acetosyringone imparts oxidative stress on E. coli

Perhaps oxidized AS inhibits pathogen invasion of plants by increasing toxicity towards microbes44. While E. coli is not a typical plant pathogen, we performed a colony count assay after treating E. coli with PB, AS, and oxidized AS (Fig. 2a), and we found oxidized AS inhibited E. coli growth more strongly than AS in its natural (reduced) state (Fig. 2b,c), suggesting a negative impact on the cell’s metabolic state. We hypothesized that the OxyR hydrogen peroxide-responsive stress regulon might be involved in the cellular response to oxidized AS. The OxyR protein is intracellularly oxidized by hydrogen peroxide and activates transcription of genes encoding catalase (katG), glutaredoxin (grxA), small noncoding transcriptional regulator OxyS RNA (oxyS), and other responding molecules to the H2O2 stress (Fig. 2d). Indeed, qPCR revealed that oxidized AS upregulated expression of two OxyR-regulated genes, oxyR and katG (Fig. 2e). Upregulation of genes in the oxyR regulon by oxidized AS suggests the potential to use oxidized AS for synthetic gene expression through the OxyR-regulated oxyS promoter.

Figure 2
figure 2

Oxidized acetosyringone elicits cellular oxidative stress response. (a) Scheme of colony count assay. (b) Colonies and (c) colony forming units (CFU) observed after cells were mixed with oxidized and reduced AS. Cells in LB were mixed with AS or phosphate buffer (PB, as a negative control), incubated for 1.5 h with shaking at 37 °C, and plated in 10× dilutions. Images were taken using an Amersham Imager 680. (d) The OxyR regulon in E. coli. The OxyR regulator is inactive in its reduced state. Upon oxidation by H2O2, oxidized OxyR binds to DNA at the promoter region of cognate genes and activates transcription. Adapted from Pomposiello and Demple31. (e) qPCR data showing effect of oxidized AS on cellular expression of stress-related genes oxyR and OxyR-regulated katG, with standard deviation of fold difference shown in parenthesis. Cells were grown to OD600 = 0.4, treated with the respective inducer, and pelleted at an OD600 of 0.8–0.9 for RNA extraction. Schematics (a) and (d) were generated in Microsoft PowerPoint Version 2401. (c) was generated in Microsoft Excel Version 2401.

Full size image

Setting the redox state of acetosyringone enables control of gene expression

To investigate whether oxidized AS could induce expression via hydrogen peroxide-inducible OxyR, we used previously engineered E. coli OxyRS-sfGFP reporters which express sfGFP downstream of the oxyS promoter and OxyR downstream of the constitutive proD promoter30 (Fig. 3a). Applying to AS potentials below its Epc had little effect on OxyRS-sfGFP reporters, while oxidizing AS by applying oxidizing potentials higher than its Epa (0.47 V) induced sfGFP expression resulting in measurable green fluorescence (Fig. 3b). The charge accumulated while applying potential to AS served as a gate for induction of the oxyS promoter. Applying potentials below the Epc of AS caused minimal accumulation of charge (< − 0.086 C), and the resulting AS did not induce gene expression. Applying oxidizing potentials caused charge accumulation over − 1.04 C, yielding oxidized acetosyringone which could induce the oxyS promoter (Fig. 3c). Expression levels were then shown to be modulated by the concentration of oxidized AS and the magnitude of potential applied to AS (Fig. 3d), with appreciable fluorescence induced only at higher concentrations and oxidizing potentials (> ~ 0.5 mM, >  ~ 0.5 V). We also found that expression level could be modulated by the duration of the applied oxidative potential (Fig. 3e), indicating that both the potential and charge duration applied to AS contribute toward accumulated charge and consequently the extent of AS oxidation. To compare the specificity of oxidized AS as a mediator inducing oxyR regulon expression, mediators ferrocene and iridium (E0 of + 0.25 and + 0.6735 vs Ag/AgCl, respectively) were also tested. Ferrocene and iridium were oxidized by application of + 0.3 V or + 0.9 V, respectively35, then added to OxyRS-sfGFP reporter cells in both the reduced and oxidized states. We also measured the final OD600 of the cultures to establish effects on cell growth. The two mediators at the concentration ranges tested either did not induce appreciable induction of the reporter cells and had little effect on growth or strongly induced expression but greatly inhibited growth (Fig. 3f), suggesting that oxidized AS selectively induces the oxyS promoter in a concentration dependent manner and at the same time has minimal effect on growth at concentrations leading to induced expression (i.e. < 750 μM).

Figure 3
figure 3

Oxidized acetosyringone (AS) induces the OxyRS promoter. (a) Scheme of experiment and gene circuit of OxyR reporter cells. (b) The indicated potentials were applied to AS for twenty minutes. OxyR reporter cells were mixed with 0.5 mM AS (to which the potential was applied) and incubated, and fluorescence was measured. Cyclic voltammogram shows oxidative (O) and reductive (R) peaks of AS. (c) Accumulated charge after applying indicated potentials to AS for 20 min. (d) AS was charged at varying potentials for 20 min and added to reporter cells at varying concentrations. Reporter cell fluorescence was measured. (e) Cell fluorescence resulting from induction by AS (oxidized for varying durations of time at + 0.7 V) at varying concentrations. (f) Fluorescence and final OD600 of cells treated with H2O2, AS, ferrocene (Fc), and iridium (Ir) in reduced and oxidized states. (b–f) show data after three hours of incubation; fluorescence is normalized to OD600 and reported as fold change relative to untreated cells. Schematic (a) was generated using Microsoft PowerPoint Version 2401 and PubChem Sketcher V2.4. (b), (d), and (e) were generated in Matlab R2020a. (c) and (f) were generated in Microsoft Excel Version 2401.

Full size image

Distinct oxyS promoter induction dynamics by H2O2 and oxidized acetosyringone

These results show that oxidized AS induces oxyRS, and while it is well-known that H2O2 also induces oxyRS, oxidized AS and H2O2 are electrochemically generated at dramatically different potentials. In Fig. 4, we found that both oxidized AS and hydrogen peroxide can induce expression through the OxyR-regulated oxyS promoter, but the dynamic responses were quite distinct. Fluorescence from the OxyRS-sfGFP reporter induced by H2O2 started to rise immediately after addition of H2O2 and reached a peak 60–80 min afterwards (Fig. 4a). In contrast, sfGFP fluorescence induced by oxidized AS rose more slowly and leveled off after 70–80 min without a subsequent decrease (Fig. 4b). The pretreated cells in both cases were identically cultured hence similarly metabolically active. Fluorescence-activated cell sorting confirmed the distinct dynamics and extent of induction elicited by H2O2 and oxidized AS (Supplementary Fig. S1). The different responses to H2O2 and oxidized AS were further highlighted by using OxyRS-sfGFP-AAV reporter cells in which sfGFP was fused with an ssRA45 degradation tag on its C-terminus. In this way, the sfGFP measurement is more reflective of the rate of generation as opposed to the absolute level. H2O2-induced fluorescence in the OxyRS-sfGFP-AAV reporters rose rapidly immediately after H2O2 addition, peaked after 20 min, and subsequently fell to the earlier uninduced reporter cell levels (Fig. 4c). These data indicated that the response to hydrogen peroxide addition was strong and finite in time, likely owing to the degradation of inducer, H2O2, at these cell densities46. Oxidized AS, on the other hand, induced a slower rise and a more sustained fluorescence (Fig. 4d), indicating continued oxidative stress (and expression) throughout.

Figure 4
figure 4

Oxidized AS elicits a distinct response from the OxyRS promoter. OD600 and fluorescence of cell reporters induced with H2O2 or with AS oxidized for 20 min at + 0.7 V. Cell reporters expressed sfGFP under control of the OxyRS promoter with (a,b) or without (c,d) a degradation tag (AAV). (a–d) show fluorescence normalized to OD600 and were generated in Microsoft Excel Version 2401. Schematics were generated in Microsoft Powerpoint Version 2401.

Full size image

We found minimal differences in growth rate between the different induction methods at the selected H2O2 and oxidized AS concentrations. Further, we performed a simple dynamic analysis of this system, calculating a zeroth order maximum rate expression and a first-order GFP decay rate spanning the appropriate times in Fig. 4a–d (Supplementary Fig. S2 and Supplementary Table S1). We found the maximum rate of synthesis was nearly identical for all hydrogen peroxide addition cases without a degradation tag (Fig. 4a) and the oxidized AS case at the 0.75 mM level (Fig. 4b). Interestingly, we also found nearly linear dependence of this maximum synthesis rate with AS concentration. We next found that the first-order degradation rate for the H2O2-induced sfGFP-AAV case (with the degradation tag) was roughly threefold larger than the largest degradation rate for the AS-induced sfGFP-AAV case (Supplementary Fig. S2 and Supplementary Table S1). This was somewhat surprising and suggested that the response to hydrogen peroxide elicited more proteolytic activity than the AS addition, although this was not tested. These results do show, however, that the responses of the cells to the different methods of induction were different.

Electronic control of gene expression

A promising outlook in electrogenetic control is to turn gene expression “on” by simple application of an electronic cue14. To demonstrate, we added AS to reporter cells and subsequently applied an oxidizing potential. This differs from the oxidized AS induction described in the previous experiments, where a bolus addition of oxidized AS induced fluorescence. Instead, here we show direct electrogenetic control: the signal for inducing expression of the OxyRS-sfGFP reporter cells is the application of an oxidative potential to an electrode immersed in a growing cell culture, rather than bolus addition of an oxidized chemical inducer (Fig. 5a). The electrochemical half-cell setup was similar to the setup for oxidation of AS, but we used a 1:1 mix of reporter cell culture and AS diluted in PB to the desired concentration (Fig. 5b). When an oxidizing potential was applied to the mixture of cell culture and AS, an increase in fluorescence was observed. This was directly analogous to the results from bolus addition of oxidized AS to reporter cells (Fig. 5c vs Fig. 4b). Gene expression and fluorescence could be tuned by varying the concentration of AS added as well as the duration of the applied oxidative voltage (Fig. 5d). Higher AS concentrations and application of oxidizing potential for a longer duration increasingly inhibited cell growth (Fig. 5e), such that inducing reporter cells using this method requires optimizing AS concentration and applied charge to avoid altered cell growth. Varying the potential applied also provides significant power to modulate gene expression (Fig. 5f). Importantly, we found that the accumulated applied charge correlated linearly with GFP fluorescence in all cases, whether the parameter being varied was charge duration or potential (Fig. 5g). The correlation between charge and gene expression supports the argument that applied charge is the effective parameter controlling gene expression, such that normalizing gene expression to charge enables consistency for comparison across distinct experiments. That is, we show here that gene expression can be reliably tuned based on total accumulated charge, and either potential or charge duration can be adjusted for fine-tuned modulation of gene expression.

Figure 5
figure 5

Acetosyringone (AS) can be applied for electronic control of cell fluorescence. (a) Direct electrogenetic control scheme. A potential applied through the gold electrode causes oxidation of AS in solution, which induces a cellular response in the cell reporters via the oxyS promoter. (b) Scheme of the electrochemical setup. A gold wire working electrode, Ag/AgCl reference electrode, and platinum counter (separated by a salt bridge) are submerged in a 2.4 mL mixture of liquid cell culture and AS. (c) OD600 and fluorescence of cell reporters induced by direct electrogenetic control. 100 μM AS was added to reporter cells, and an oxidative potential of + 0.7 V was applied for the indicated duration. Fluorescence is reported as the moving average of fluorescence over 30 min with 3 technical replicates. (d) Fluorescence of reporter cells after addition of AS at varying concentrations and application of + 0.7 V for varying durations of time. Fluorescence was normalized to that of untreated cells, and (e) the difference in OD600 between treated and untreated cells three hours after induction is shown. (f) Fluorescence of reporter cells, normalized to untreated, after addition of AS at varying concentrations and application of varying potentials for 7.5 min. (g) Correlation of applied charge to cell fluorescence. Cells were mixed with the indicated concentrations of AS. A constant potential of + 0.7 V was applied for varying durations from 0 to 10 min (red), or the potential was varied from 0 to + 0.7 V and applied for 20 min (green). The linear regression line, calculated for all values whether varying time or varying potential, and R2 value are shown. (d), (f), and (g) show fluorescence after three hours of incubation normalized to OD600 and are reported as fold change relative to untreated cells. Schematics (a) and (b) were generated in Microsoft PowerPoint Version 2401. (d), (e), and (f) were generated in Matlab R2020a. (c) and (g) were generated in Microsoft Excel Version 2401.

Full size image

An electronic switch: a duality in controlling gene expression

It is important to note that electrogenetic control of the oxyS promoter is also achieved using the 2-electron oxygen-reduction reaction (ORR) to convert dissolved oxygen into H2O2, thereby inducing OxyR-regulated gene expression by electronically generating one of its native signals15,17,41. The ORR is carried out at reducing potentials (− 0.5 V with a gold electrode 18,41) whereas oxidation of AS occurs at oxidizing potentials, and both H2O2 and oxidized AS are now shown to induce expression via OxyR (Fig. 6a). To verify induction of OxyRS-sfGFP reporters by electrode-generated H2O2 and to explore the extent to which one can use varied inputs, we applied potentials varying from + 0.8 to − 0.7 V to reporter cell cultures in the absence of any mediator. As expected, only application of the reducing potentials required for ORR (from − 0.5 to − 0.7 V41) accumulated charge (Fig. 6b), consistent with generation of H2O2 at the electrode41. Interestingly, when the same experiment was repeated in the presence of acetosyringone, charge accumulation was observed both at oxidizing potentials (+ 0.5 to + 0.8 V) and reducing potentials (− 0.5 to − 0.7 V, Fig. 6b). While most of the applied potentials resulted in comparable growth rates and final OD600 values, applying oxidizing potentials inhibited growth only in the presence of AS, consistent with earlier observations (Fig. 6c). Gene expression followed the same trends as the accumulated charge. Without AS present, only reducing potentials elicited fluorescence (“reductive activation”), whereas when AS was present, fluorescence was observed at both reducing and oxidizing potentials (“oxidative activation”) (Fig. 6d and Supplementary Fig. S3). Interestingly, induction by oxidation of AS led to higher fluorescence per unit charge (Fig. 6e). These results suggest that AS can facilitate gene expression at two distinct potential ranges: at oxidizing potentials, as oxidized AS induces the oxyS promoter; and at reducing potentials, as the presence of AS does not interfere with ORR-based H2O2 generation, allowing for induction of the oxyS promoter by H2O2.

Figure 6
figure 6

Biological responses (growth and gene expression) are elicited by electrochemical signals generated at distinct potential ranges. (a) Result of applying electronic signals to cells in the presence and absence of AS. Application of a reducing potential leads to H2O2 generation due to the oxidation–reduction reaction. Application of an oxidizing potential generates oxidized AS. Both H2O2 and oxidized AS induce gene expression via the OxyR regulon. (b) Cumulative applied charge, (c) OD600 after three hours of incubation normalized to untreated cells, and (d) fold fluorescence of POxyRS-sfGFP reporter cells after varying potentials are applied for 7.5 min in the absence or presence of 500 μM AS. Fluorescence is reported after three hours of incubation and is normalized to OD600, then the RFU/OD600 value is normalized to that of untreated cells. (e) Correlation of cumulative applied charge to gene expression for cells to which AS was added or omitted and reducing (− 0.5 to − 0.7) or oxidizing (+ 0.5 to + 0.7) potentials were applied. Linear regressions have a slope of 0.25 (− AS, reducing potentials), 0.81 (+ AS, oxidizing potentials), and 43.0 (+ AS, oxidizing potentials). Schematic (a) was generated using Microsoft PowerPoint Version 2401 and PubChem Sketcher V2.4. (b–e) were generated in Microsoft Excel Version 2401 and show the average of three biological replicates.

Full size image

Dynamic electrogenetic programming

To compare our acetosyringone-based electrogenetic approach with ORR-based H2O2 production, we mixed acetosyringone with OxyRS-sfGFP-AAV reporter cells and treated in varied electronic modalities by applying 5-min “pulses” of either oxidizing potential (+ 0.7 V) or reducing potential (− 0.5 V) and different combinations thereof (Fig. 7a). In Fig. 7b, we pulsed + 0.7 V in a repeated fashion for 5 min (− 160 ± 47 mC charge) every 45 min. It was readily evident that the added oxidizing potential repeatedly induced the cells, and interestingly, nearly equivalently at each step. The degradation-tagged sfGFP showed peaks in fluorescence followed by decreasing levels starting from 30 to 90 min after the pulse, back towards a still induced but lower level. With multiple applications of a reducing potential (− 0.5 V for 5 min, or + 191 ± 55 mC charge), we found sharp peaks followed by more rapid decay until all reached similar levels typically an hour after the peak in fluorescence (Fig. 7c). Subsequent tests with alternating oxidation and reduction pulses resulted in varied responses with different patterns, peaks, and decay times (Fig. 7d–f). Interestingly, in each case where + 0.7 V was applied, there was a relatively slower response (in both increase and decrease of signal), while in each case where − 0.5 V was applied, there was a faster, sharper increase in fluorescence (Fig. 7b–f). Also, we note that typically, the resultant steady state values were apparently additive, such that a culture with three pulses reached higher levels than those with two and these were seen higher than those with a single pulse, even without normalizing expression to OD600 (Supplementary Fig. S4). The dynamic trends resulting from the oxidation and reduction pulses were consistent across biological replicates (Supplementary Fig. S5). Hence, applying multiple pulses to the reporter cell culture over time yielded somewhat predictable and additive fluorescence outputs, demonstrating dynamic temporal control of gene expression based on the type (potential), duration, and timing of the electronic input.

Figure 7
figure 7

Electrochemically generated signals influence gene expression dynamics. (a) Scheme of electrochemical signals generated at reducing and oxidizing potentials in the presence of AS and reporter cells expressing degradation-tagged fluorophores. Fluorescence of POxyRS-sfGFP-AAV reporter cells, in the presence of 500 μM AS, after application of “pulses” of potential at the indicated voltage and time for 7.5 min. If the value of fluorescence difference between the sample and the untreated cell negative control was less than 5 RFU, it was denoted as 0 RFU. The applied potential “pulses” were (b) three oxidizing pulses, (c) three reducing pulses, (d) an oxidizing, reducing, then oxidizing pulse, (e) one reducing followed by two oxidizing pulses, or (f) two oxidizing pulses followed by a reducing pulse. Schematic (a) was generated using Microsoft PowerPoint Version 2401. (b–f) were generated in Microsoft Excel Version 2401.

Full size image

Discussion

Electronic control of gene expression expands our ability to program and manipulate cellular behavior. In this work, we demonstrate that redox electrogenetics can be facilitated by acetosyringone via transcriptional regulator OxyR and the oxyS promoter. OxyR has been described to be primarily activated by hydrogen peroxide and partially activated by certain S-nitrosothiols33 and decreases in the cellular thiol-disulfide ratio47. Oxidized acetosyringone, a phenolic plant-produced redox mediator, is reported here for the first time to be an inducer of OxyR. Importantly, this work introduces a new mechanism and electronic potential range for tunable electronic control of gene expression. We show that OxyR-mediated expression can be electronically induced both at reducing potentials (< − 0.5 V17) by electrode-generated H2O2, and at oxidative potentials (> 0.7 V) by oxidized acetosyringone (Fig. 6a). Induction of SoxR-mediated gene expression has previously been achieved at oxidative potentials only (> 0.3 V14), and the electrochemical configuration was based on two mediators—pyocyanin and ferricyanide/ferrocyanide—as well as the maintenance of an anaerobic environment. Acetosyringone-facilitated induction, on the other hand, provides a simple, tunable, and accessible electrogenetic scheme based on a single mediator that works in an aerobic environment, thus expanding the range over which electronic activation can be applied. In addition, because each of the described methods for electronic induction are accessed at distinct potential ranges (more negative than − 0.5 V, more positive than 0.3 V, more positive than 0.7 V), future work can multiplex electronic inputs by utilizing applied potential as a “knob” to selectively toggle expression of desired genes.

To our knowledge, this is the first demonstration that the redox signaling capabilities of acetosyringone can be harnessed to electronically actuate cellular function. Acetosyringone is produced by plants during wound stress and has been well studied for its signaling mechanism involving molecular recognition. Specifically, AS interacts with vir two-component receptor systems, inducing chemotaxis and virulence gene expression in Agrobacterium via the vir regulon48. AS can also act as a signaling molecule through another, distinct mechanism: it can interact through redox activity. The oxidized form of AS can be generated enzymatically49 or electrochemically50, and previous work suggested that oxidized AS could elicit stress responses in Pseudomonas syringae, upregulating expression of genes involved in metabolism, energy generation, and cell wall components and inducing a viable but not culturable (VBNC) state51. Our work builds upon such insights by demonstrating that oxidized AS, but not reduced AS, specifically interacts with the OxyR signal transduction pathway. In this capacity, AS serves as a “pro-signal”, wherein its redox-based messaging is conveyed when it is in the oxidized form. “Pro-signals” are consistent with the concept that redox signaling depends more on molecular activities than molecular structure8. While we showed that AS can act as a redox-state-dependent signal, we did not attempt to study the intracellular signal transduction mechanisms. As OxyR and its hydrogen peroxide-responsive homologs are widely conserved across many bacteria32, future applications of this work may involve sensing or actuation of other, rhizosphere-relevant, bacteria.

Interestingly, we observed a difference in the dynamics of the OxyR-regulated responses elicited by H2O2 and oxidized AS. Importantly, we show that a single promoter can be toggled using two distinct mechanisms simply by varying the type of potential applied to cells in the presence of AS. The imposed electronic input, or redox context, defines the duration and magnitude of the biological response, thus further demonstrating the importance of context for redox activity52,53. Both H2O2 and oxidized AS are signaling molecules present in the rhizosphere after plant infection. The difference in the dynamic responses to the two signaling molecules may have implications relative to the bacterial response to plant-produced oxidative stressors. We note further that physicochemical or spatial effects may arise from gradients projecting from the electrode to the solution41,54 (analogous to the plant and rhizosphere). Such gradients are a characteristic feature of redox signaling, and have been partially mitigated in electrogenetics by using gold-binding peptide surface display that serves to localize cells onto gold electrodes and the electrode-generated signal30 or by using an intermediate “transmitter” cell population that transduces a transient electrode-generated signal (H2O2) into a more sustained acyl homoserine lactone quorum sensing signal41.

We also found that by selectively tuning and applying oxidizing and reducing potentials in pulses, we could generate patterns of oscillating gene expression (Fig. 7). Oscillatory patterns are natively found in biology, including signaling molecules that are expressed in pulses55 and plant responses to multiple stresses56. As such, our electrochemical setup using acetosyringone could be used to recapitulate natural oscillatory activity. At the same time, biomanufacturing applications could involve synthetic strains to improve production of relevant molecules57, or activate a particular cellular response that is present for either a transient period or for a sustained duration depending on whether the inducer present is H2O2 or oxidized AS.

Understanding the signaling characteristics of oxidized acetosyringone provides new avenues for learning and building technologies for the soil rhizosphere as well as for engineered systems. Using the acetosyringone induction system, one could envision biosensors, cells and cell networks that collect new types of information about their local redox state wherein they are programmed to carry out designer functions (e.g., generate nutrients for root structures, degrade recalcitrant contaminants, or otherwise synthesize value-added products).

Materials and methods

Strains and plasmids

The strains, plasmids, and primers used in this work are listed in Table 1. Plasmid pOxyRS-sfGFP-AAV is derived from pOxyRS-sfGFP38 and incorporates a ssrA degradation tag encoding AANDENYLAAAV45 (“AAV”) at the 3ʹ end of sfGFP.

Table 1 Strains, plasmids, and primers used in this study.
Full size table

Cell culture

Plasmid pOxyRS-sfGFP was cloned into strain SW102 (ZK126 ΔoxyRS) and plasmid pOxyRS-sfgFP-AAV was cloned into NEB10-beta. All experiments involving these two plasmids were performed in the aforementioned strains, respectively. Cells were cultured at 37 °C with shaking at 250 rpm in an incubator or in a TECAN SPARK microplate reader. LB media was used for cloning and for growing 4–5 mL overnight cultures. For cell experiments, overnight cultures were diluted 100× into M9 media (1× M9 salts, 0.1 mM CaCl2, 2 mM MgSO4, 0.4% glucose, and 0.2% casamino acids) with the appropriate antibiotics and cultured until the OD600 reached ~ 0.4 before addition of inducer(s). Unless indicated, all other cell experiments used a 1:1 mixture of M9 media and 0.1 M phosphate buffer (PB, pH = 7.2) (with acetosyringone when indicated). For ampicillin-resistant plasmids (pOxyRS-sfGFP and pOxyRS-sfGFP-AAV), ampicillin was used at a concentration of 100 μg/mL. Microplate reader experiments were conducted using 96-well plates and 200 μL sample volume, and samples were measured in triplicate.

Electrochemical setup

Electrochemical techniques were run using a CHI 6273 C electrochemical analyzer (CH Instruments). All electrochemical methods used a three-electrode setup in a 17 mm diameter glass vial with an Ag/AgCl reference electrode, platinum wire counter electrode, gold working electrode, and small stir bar. For cyclic voltammetry, a 2 mm gold standard working electrode (CH Instruments) was used, and potential was scanned from 0 to 0.7 V and back at a scan rate of 0.2 V/s with a 0.001 V sample interval and 1 × 10–5 A/V sensitivity. For routine oxidation of acetosyringone with a half-cell setup41, a 4.5 mL, 2 mM solution of AS (Sigma-Aldrich) in PB was prepared, and the electrochemical setup used a salt bridge as well as a 0.5 mm diameter gold wire (Sigma-Aldrich) approximately 75.4 cm long, coiled to fit in the vial. Using the amperometric i-t curve technique, a constant potential of 0.7 V was applied for 20 min with a 0.1 s sample interval and 1 × 10–3 A/V sensitivity. The potential and duration of applied charge were varied as described in the experimental section. For oxidation of cell culture, the same setup was used, but with 2.4 mL of a 1:1 mixture of cell culture (OD600 of 0.4) and either PB only or AS and PB. AS concentration, potential, and duration of applied charge were varied as described in the experimental section. For application of “pulses” of potential, a 2.4 mL of a 1:1 mixture of cell culture (OD600 of 0.4) and PB with AS (final AS concentration of 500 μM) was prepared and the indicated potential was applied for 7.5 min. 50 μL samples were extricated in triplicate, diluted to 200 μL with equal volumes of M9 media and PB, and placed in a 96-well plate for incubation and measurement in the plate reader. Simultaneously, the remaining cell solution was incubated in a shaking incubator at the same temperature and shaking speed. After 45 min of incubation, the cell solution was replenished with 150 μL of a mixture of M9, PB, and AS to the same concentration as the initial setup. The same steps were repeated for a second and third “pulse” of potential.

Spectrophotometric and fluorometric readings

Absorbance and fluorescence were measured using a TECAN Spark microplate reader. Absorbance spectra of AS were measured with a clear 96-well plate. For cell experiments, 200 μL samples were loaded in triplicate in a black 96-well plate with a clear bottom. The plate was placed in a humidity cassette in the microplate reader, set at 37 °C with shaking, and OD600 and sfGFP fluorescence (485 nm excitation and 520 nm emission) were continually measured. Fluorescence (relative fluorescence units, or RFU) and absorbance units were normalized by subtracting measurements of the PB/M9 blank sample. Fluorescence was further normalized by dividing by OD600 (RFU/OD600); when indicated, RFU/OD600 was normalized to the RFU/OD600 of untreated cell samples (Fold change relative to untreated). Background fluorescence from oxidized AS at the concentrations utilized was negligible (Supplementary Fig. S6), so fluorescence was normalized by subtraction of the PB/M9 blank sample rather than subtraction of oxidized AS samples.

qPCR assay

6 mL of NEB10β cells were grown to an OD600 of 0.4 and induced with PB (negative control), 100 μM H2O2, or 0.75 mM of oxidized AS. After 30 min incubation, 1 mL of cells was pelleted in triplicate for biological replicates. RNA was isolated with the TRIzol® Max™ Bacterial RNA Isolation Kit (ThermoFisher) according to the manufacturer’s protocol and treated with DNAse I (New England Biolabs). Quantitative real time PCR was run on Applied Biosystems QuantStudio 7 Flex (ThermoFisher) using ~ 200 ng RNA, 1.5 μM of the qPCR primers listed in Table 1, and the Power SYBR® Green PCR RT-PCR mix following the manufacturer’s protocol (ThermoFisher). Gene expression fold change averages and standard deviations were calculated using ΔΔCT relative to E. coli 16s rRNA as an internal control and compared to the negative control of untreated cells.

CFU count assay

250 μL of NEB10β cells grown to an OD600 of 0.4 were transferred to microcentrifuge tubes and treated with PB (negative control), AS, or oxidized AS. Tubes were incubated for 1 h at 37 °C, then four 10× serial dilutions were prepared in PB. 5 μL of each dilution was pipetted onto an LB agar plate in triplicate and incubated overnight. Individual colonies were counted at the highest dilution and colony forming units (CFU) were determined based on the dilution factor.