Tactic Response of Shewanella oneidensis MR-1 toward Insoluble Electron Acceptors

Previous hypotheses of tactic behavior of exoelectrogenic bacteria are based on techniques that do not accurately control the electrochemical potential, such as chemical-in-plug assays or microscopy tracking experiments in two-electrode cells. Here, we have revisited previous experiments and, for the first time, performed microscopy cell-tracking experiments in three-electrode electrochemical cells, with defined electrode potentials. Based on these experiments, taxis toward electrodes is observed to switch at about −0.2 V versus standard hydrogen electrode (SHE), coinciding with the reduction potential of flavins.

Here, we decided to revisit the function of electrochemical potentials in the tactic behavior of MR-1 toward electrode surfaces. An experimental platform was developed that can specifically inspect taxis of MR-1 around electrodes with defined electrode potentials (in a three-electrode system), such that the redox state of extracellular redox compounds such as flavins can be controlled. The set-up was designed with the aim to analyze both individual-and population-level behaviors of MR-1 and ⌬mtrC/⌬omcA MR-1 (⌬mtrC/⌬omcA MR-1 is a mutant strain deficient in outer membrane cytochromes that are required for EET).
Tracking analysis. Within the first 20 min after loading and sealing an electrochemical cell with MR-1, a high proportion (Ͼ70%) of bacteria displayed high motility throughout the cell with no relationship toward the electrode. This is expected as after sealing the cell, MR-1 will consume oxygen still dissolved in the medium. Between 20 and 60 min the motility gradually decreases as oxygen is consumed.
Microscope videos (at 10 FPS) were recorded of the electrochemical cell while applying various potentials to the gold working electrode. Motility of bacteria was analyzed using a tracking algorithm described in the supplemental material (Text S1). Similar to previous observations (12)(13)(14), only a subpopulation of the bacteria was observed to be motile. Here, only motile bacteria are considered, defined as bacteria with velocities Ͼ4 m/s, and it is observed that 50% Ϯ 10% of the bacteria are motile. Two key parameters are presented here: average velocity of all motile bacteria within an area (for bacteria with velocities Ͼ4 m/s) and motile population density, which is the density of bacteria with mean velocities Ͼ4 m/s in a specified area.
Taxis is dependent on electrode potential. Tracking analysis shows that when oxidative potentials are applied (0.3 V versus SHE), motile MR-1 bacteria accumulate close to the electrode (Fig. 1). These effects are stronger if an additional 2 M riboflavin is added to the capillary cell (see Fig. S1). Importantly, ⌬mtrC/⌬omcA MR-1 does not respond to the applied potential, indicating accumulation and motility are dependent on MR-1's ability for EET (Fig. S1).
Switching to a relatively reducing potential of Ϫ0.3 V results in a sudden drop in average velocity and motile population density close to the electrode (3 to 23 m, red lines in Fig. 1A and B). Somewhat unexpectedly, at first glance, is the increase in motile population density further away from the electrode (83 to 103 m, black lines in Fig. 1B). On further analysis, however, this is due to MR-1 migrating away from the electrode upon applying a reducing potential. In doing so, motile MR-1 bacteria pass through areas further away from the electrode, explaining the increase in motile population density further away from the electrode. The latter is also apparent when analyzing the directional bias of motile bacteria (Fig. 1C), which is quantified by treating the tracked trajectories as vectors with a magnitude and a directional component. Here the changes in net swimming direction can be clearly seen and coincide well with the changes in applied potential.
To validate the tracking analyses, videos were also analyzed with a method based on Shannon's entropy, which does not require locating bacteria and their trajectories, both of which can be biased depending on the tracking algorithm used. Shannon's entropy is the entropy for each pixel in the video based on intensity fluctuations over a specified duration. Shannon entropy values for pixels in areas of high bacterial motility will thus be higher than areas of low or no motility. Shannon's entropy has previously been used in the analysis of time series in numerous fields, including bacterial population motility by Nisenbaum et al. (15). Qualitatively identical results were obtained using the two unrelated techniques (Fig. S2). Agreement between results obtained from tracking analysis and Shannon's entropy confirms the validity of the analytical methods.
Tactic behavior switches at about ؊0.2 V. By applying a series of potentials, while measuring the bacterial response through quantifiable parameters (average velocity, motile population density, and directional bias), the potential can be identified at which the change in bacterial behaviors occurs. Throughout all the video microscopy experiments carried out using MR-1, with or without added riboflavin, the shift in motile behavior is observed between Ϫ0.25 and Ϫ0.15 V. For MR-1 with added riboflavin, the window where the change occurred is narrower and between Ϫ0.25 V and Ϫ0.205 V, which coincides with the reduction potential of riboflavin (Fig. 2).
Here, tactic behavior was mostly absent in ⌬mtrC/⌬omcA MR-1. Although in this work only riboflavin was added, it is well known that MtrC reduces other flavins besides riboflavin, in particular flavin mononucleotide (FMN) (16, 17). MR-1 is known to excrete flavins, reaching concentrations between 100 and 200 nM in "closed" electrochemical  (17). Besides secreted flavins, flavins are found in yeast extract (18) and vitamin mixes used to prepare growth media. We estimate that in our experiments (using a one-in-five dilution of LB), even without any additional riboflavin, flavin concentrations could reach up 160 to 240 nM, explaining why tactic behavior is also observed in the absence of added riboflavin (Fig. S2).
Flavins, including riboflavin (vitamin B 2 ), are a vital metabolite for MR-1, and control experiments in the complete absence of (ribo)flavin are not feasible. In spite of this, we propose that MR-1 utilizes riboflavin, and possibly other flavins, as a means to migrate toward insoluble electron acceptors (e.g., metal oxides) and initiating colonization of the surface. This function would be in addition to the multiple roles of flavins in EET respiration. First, flavins act as an electron shuttle for mediated electron transfer, enabling respiration to continue in situations where the bacterium is in a planktonic state with limited soluble electron acceptors and no insoluble electron acceptors in its immediate vicinity (17). Second, flavins act as cofactors for MtrC and OmcA, enabling enhanced direct electron transfer (16). The concentrations required for efficient energy taxis could be much lower than for substantial mediated electron transfer. We thus propose that self-secreted flavins are not only redox mediators, but also redox sensors, and could be an important factor for MR-1's survival in electron acceptor-limited niches.
The MATLAB codes used for the tracking and Shannon entropy analysis are openly available from the University of Leeds Data Repository (https://doi.org/10.5518/409).