Electronic control of gene expression and cell behaviour in Escherichia coli through redox signalling

The ability to interconvert information between electronic and ionic modalities has transformed our ability to record and actuate biological function. Synthetic biology offers the potential to expand communication ‘bandwidth' by using biomolecules and providing electrochemical access to redox-based cell signals and behaviours. While engineered cells have transmitted molecular information to electronic devices, the potential for bidirectional communication stands largely untapped. Here we present a simple electrogenetic device that uses redox biomolecules to carry electronic information to engineered bacterial cells in order to control transcription from a simple synthetic gene circuit. Electronic actuation of the native transcriptional regulator SoxR and transcription from the PsoxS promoter allows cell response that is quick, reversible and dependent on the amplitude and frequency of the imposed electronic signals. Further, induction of bacterial motility and population based cell-to-cell communication demonstrates the versatility of our approach and potential to drive intricate biological behaviours.

Glucose consumed (normalized to initial measurement) plotted vs. the acetate produced for DJ901 cells with or without the pTT03 plasmid treated with the indicated mediators. Green and blue shaded regions represent 95% confidence intervals for green and blue trendlines, respectively. b. The percent of DJ901 pTT03 cells stained with propidium iodide, indicating dead cells. Error bars show s.d. of biological triplicates. Figure 7. Cell reduction of ferricyanide: a. Scheme of spectrophotometric and electrochemical methods for measuring ferricyanide reduction by cells. b. Absorbance at 420 nm correlates with ferricyanide concentration. c. Reduction of different ferricyanide concentrations by cells results in absorbance decrease (measured at 420 nm, starting cell OD600 of 2.0) over time. d. Oxidation of ferrocyanide shows more negative current (measuring ferrocyanide) when ferricyanide is reduced by cells. e. Reduction of ferricyanide, as measured by absorbance, is higher with higher cell amounts.

Supplementary
Multiple cycles of oxidation/reduction on the same solution and corresponding changes in absorbance over time.   Increasing concentrations of pyocyanin added to the culture resulted in higher phiLOV fluorescence as measured by flow cytometry (Supplementary Fig. 4 a insert) from samples induced for 1.5 hours before fixing. The addition of 5 mM of ferricyanide increased fluorescence to a much higher degree than just pyocyanin (up to ~17-fold in the case of 5 µM of pyocyanin) ( Supplementary Fig. 4 a). Increasing ferricyanide concentration also increased the fluorescence when pyocyanin was 5 µM ( Supplementary Fig. 4 b). When phiLOV-DAS fluorescence was measured over time, increasing ferricyanide sustained a higher fluorescence over a longer period of time ( Supplementary Figs. 4 c). Negative controls of ferrocyanide or ferricyanide alone were also tested, and as can be seen from Supplementary Figure 4 b, did not result in an increase of fluorescence. Therefore, pyocyanin and ferricyanide were both necessary for the amplified protein production (beyond pyocyanin at 5 µM) from the PsoxS promoter, with pyocyanin initiating the protein production, and ferricyanide amplifying it.

Supplementary Note 2
Our system is concentration-dependent on both pyocyanin and ferricyanide ( Supplementary Fig. 2 & 4) anaerobically. However, when oxygen is present, the effect of ferricyanide (increase of fluorescence) is either negated (in aerated conditions) or decreased (non-aerated) in the conditions tested ( Supplementary Fig. 5). Fluorescence was measured after 1.5 hours of aerobic culture at 37 °C with or without 250 rpm shaking. When any pyocyanin is present, aeration allows oxygen to amplify the response to a maximum value, and the concentration of ferricyanide does not affect fluorescence at the tested conditions. Non-aerated conditions allow for some cells to not be exposed to high oxygen over the course of the experiment. This drops the Pyo and SoxR-induced fluorescence response. Ferricyanide addition recovers a maximum response due to its high concentration throughout the sample. More optimization and studies would have to be conducted in conditions with various oxygen amounts to find those in which the electrogenetic device would function non-anaerobically, but here we present preliminary studies.

Supplementary Note 3
Since ferricyanide acts as an electron acceptor in anaerobic conditions, and pyocyanin can be toxic to cells, we wanted to check the metabolic and toxicity effects of our treatments. This included the measurement of glucose consumption, acetate production, and propidium iodide staining of treated cells. To tease apart metabolic effects of our treatments vs. those due to protein production, we used both the DJ901 cells with the pTT03 plasmid and those without.
Propidium iodide staining over 9 hours of growth with the indicated mediators shows that 5 µM pyocyanin and 5 mM ferricyanide, which are the concentrations used throughout the majority of the manuscript, do not induce cell death ( Supplementary Fig. 6 b). We found concentrations of ~ 50 mM Fcn (O/R) were toxic.
We measured the anaerobic consumption of glucose and production of acetate due to various treatments of both DJ901 cells with and without the pTT03 plasmid. Cells were inoculated at OD600 of 0.25 with the mediators in the anaerobic chamber. As can be seen from data in Supplementary Figure 6 a, whether the plasmid is present or not, adding both pyocyanin and ferricyanide to the cultures results in significantly lower production of acetate per glucose consumed. This was not the case for either ferricyanide or pyocyanin alone. Specifically, while the with-plasmid cases resulted in a slightly lower specific acetate yield on glucose than those cases without plasmid, there was no apparent influence of either redox mediator. In sum, the apparent amplification of pyocyanin-induced gene expression coincided with suppressed acetate production, but was not the cause of the suppressed acetate production (because the noplasmid case also had low acetate production). Thus, the apparent amplification in gene expression may have been assisted by reduced metabolic byproducts, but further experimentation would be needed.

Supplementary Note 4
Supplementary Figure 8 a shows the proposed intracellular interactions. Pyocyanin oxidizes SoxR and results in phiLOV protein production. Ferricyanide interacts with the cell by acting as an electron acceptor for one or more parts of the electron transport machinery. While a well-known microbial electron acceptor, ferricyanide's exact interactions within the electron chain remain unclear 9-11 .
To test whether ferricyanide specifically, by acting as a terminal electron acceptor, amplifies the production of pyocyanin-induced protein, we used an alternative electron acceptor in a similar experiment. As can be seen in Supplementary Figure 8, nitrate, a common anaerobic electron acceptor, also results in an increase in phiLOV fluorescence when pyocyanin is added to cells with pTT03. Correspondingly, nitrite, a reduced form of nitrate and a less efficient electron acceptor, does not result in as high an increase in protein production. We next used phenazine metho-sulfate (PMS) instead of pyocyanin, in similar concentrations, to induce a fluorescent response with ferricyanide addition to cells with pTT03. Supplementary  Figure 8 c shows that PMS behaves similarly to pyocyanin and induces fluorescence over time in a ferricyanide-dependent manner. Importantly, PMS behaves similarly to pyocyanin when cells are electronically induced with various charges (oxidizing potential, various times, as in Methods). We see an average cell fluorescence that correlates well with the applied charge ( Supplementary Figs. 8d).
These results substantiate the basic premise that Pyo-induced gene expression may be augmented due to Fcn(O)'s role as an electron acceptor and/or the involvement of electron transport mechanisms. Additional studies will need to be performed to systematically elucidate the mechanism of our electrogenetic device and the relevant redox interactions.

Supplementary Note 5
In order to separate synthesis (from promoter activity) from protein degradation, the following model was applied to the changing fluorescence levels with respect to charge: dF/dt = r(t) -k d F, where F = fluorescence, r(t) = rate of synthesis, and k d = first order degradation constant. All calculations were performed using Matlab.The degradation rate of the fluorescence was measured using degradation tags in separate experiments ( Supplementary Fig.   12 b). A first-order model for degradation was assumed and the exponential was fitted to the data using the Levenberg-Marquardt nonlinear least squares method. The resulting best-fit degradation constant was found (k d = 0.0362/min).
The fluorescence data was shifted to an initial value of 0 and interpolated using piecewise cubic hermite interpolation. Combining these data with the calculated first derivatives of the interpolation, r(t) was extracted. The integral of r(t) was then calculated to observe the dynamics of accumulated rate with respect to charge using adaptive quadrature (ie integral function in Matlab). This accumulated rate (from left to right, top to bottom; increasing in positive charge) is shown in red in Supplementary Figure 13, along with the interpolation (black line) of the original data (circle). The integral is calculated up to its maximum value, which we denote "integrated protein synthesis". This terminal integrated rate is plotted with respect to the corresponding charge in Figure 3 d. The resulting relationship between the terminal integrated rate and charges produced the following fitted line y(t) = (−1.543 * 10 5 ) + 3.814, meaning that as the charge increases, the terminal integrated rate increases linearly. The Matlab code for the model is available upon request.

Supplementary Note 6
Solution-based induction of the control WT cells (W3110), CheZ KO cells (W3110 cheZ -) without plasmid pHW01, and CheZ KO cells with the pHW01 plasmid were performed before measurement of CheZ protein and cell velocities. Western blots performed on WT and CheZ KO cells without pHW01 ( Supplementary Fig. 14 a) show that all WT cells have CheZ and all KO cells without pHW01 show absence of CheZ regardless of the inducers used. These results demonstrate that changes in CheZ protein in the CheZ KO cells with pHW01 are due to the addition of the genetically engineered elements. Additionally, the treatments do not affect the presence (WT) or absence (CheZ KO) of CheZ. Supplementary Figure 14 c shows expanded blots of the same results as in Figure 4 b. Since we pre-incubated polyclonal antibody, additional uncharacterized bands were present.
The results in Supplementary Figure 15 show that the WT cells retain a higher relative velocity that is not affected by the mediator treatments. The CheZ KO cells that do not have pHW01 show an expected lower velocity that is not affected by mediator addition. The CheZ KO cells with pHW01 show a low velocity, similar to that of the CheZ KO cells without the plasmid when no mediators are present. Pyocyanin does not affect the velocity, but ferricyanide alone did show a slight increase. Both mediators together show a concentration-dependent increase in velocity. These solution-based controls indicate that electronic induction as described in the paper enables CheZ induction and increased velocity relative to the non-motile knockouts.

Supplementary Note 7
The bioelectronic relay cells (DJ901 with the plasmid pTT05) and the biosensor cells (DJ901 with the plasmid pTT06 ) were co-cultured as described in the Methods and induced with the indicated mediators in solution to test initial effects on production of fluorescent protein by the biosensor cells. As can be seen in Supplementary Figure 16, a trend similar to when DJ901 with pTT03 cells were used is seen. That is, increasing amounts of fluorescence were measured from cells induced with Fcn (O). Pyocyanin is also needed for an amplified response. We can see that fluorescence is higher in cells that are co-cultured and induced with more negative charges (Supplementry Fig. 16 b) but that these charges are relatively closer to zero than those needed to similarly induce DJ901 cells with pTT03. This is due to LuxI & AHL's amplification of the initial Pyo and Fcn(O) signal induction, and allows for a method to tweak the sensitivity and response of the electrogenetic device.
Additionally, we performed experiments in which the relay and biosensor cells were not cocultured. Both cells were grown as before and placed in the anaerobic chamber. The relay cells alone were electrochemically induced by oxidation of pyocyanin and ferrocyanide for various times, which resulted in the charges indicated in Supplementary Figure 16 c. After a half hour of total induction + culture time, the cells were spun down, and the supernatant filtered through a 0.22 µm filter. The supernatant was then added to the biosensor cells at OD600 of 0.25. These cells were then grown for 2 hours, spun down and fixed, and then fluorescence was measured with flow cytometry. As can be seen from Supplementary Figure 16 c, there is a linear correlation between the induction charge applied to the relay cells and the fluorescent response of the biosensor cells. These results indicate that an electronic induction can be translated through cell-communication molecules (AHL) into a correlated response by a second set of cells. If we can thus separate the electronic signaling and the output response while still maintaining a good correlation, we can use such bio-electronic relay cells as translators between electronics and other cells that do not have to be under anaerobic conditions or be exposed to any of the mediators.