Biochemical Society Transactions

Electron Transfer at the Microbe–Mineral Interface

Shewanella oneidensis MR-1 chemotaxis proteins and electron-transport chain components essential for congregation near insoluble electron acceptors

H. Wayne Harris , Mohamed Y. El-Naggar , Kenneth H. Nealson


Shewanella oneidensis MR-1 cells utilize a behaviour response called electrokinesis to increase their speed in the vicinity of IEAs (insoluble electron acceptors), including manganese oxides, iron oxides and poised electrodes [Harris, El-Naggar, Bretschger, Ward, Romine, Obraztsova and Nealson (2010) Proc. Natl. Acad. Sci. U.S.A. 107, 326–331]. However, it is not currently understood how bacteria remain in the vicinity of the IEA and accumulate both on the surface and in the surrounding medium. In the present paper, we provide results indicating that cells that have contacted the IEAs swim faster than those that have not recently made contact. In addition, fast-swimming cells exhibit an enhancement of swimming reversals leading to rapid non-random accumulation of cells on, and adjacent to, mineral particles. We call the observed accumulation near IEAs ‘congregation’. Congregation is eliminated by the loss of a critical gene involved with EET (extracellular electron transport) (cymA, SO_4591) and is altered or eliminated in several deletion mutants of homologues of genes that are involved with chemotaxis or energy taxis in Escherichia coli. These genes include chemotactic signal transduction protein (cheA-3, SO_3207), methyl-accepting chemotaxis proteins with the Cache domain (mcp_cache, SO_2240) or the PAS (Per/Arnt/Sim) domain (mcp_pas, SO_1385). In the present paper, we report studies of S. oneidensis MR-1 that lend some insight into how microbes in this group can ‘sense’ the presence of a solid substrate such as a mineral surface, and maintain themselves in the vicinity of the mineral (i.e. via congregation), which may ultimately lead to attachment and biofilm formation.

  • chemotaxis
  • electrokinesis
  • energy taxis
  • microbe–mineral interaction
  • microbial fuel cell (MFC)
  • Shewanella oneidensis MR-1

EET (extracellular electron transport)

Before the discovery of EET [1,2], electron transport was considered to be an intracellular phenomenon, occurring in the cytoplasm or on the cytoplasmic (or photosynthetic) membranes of mitochondria, bacteria and archaea. EET was first discovered in bacteria, referred to as DMRB (dissimilatory metal-reducing bacteria), a discovery that required a change in thinking, with the realization that bacteria (and archaea) are capable of electron transfer to solid substrates such as manganese and/or iron oxides and oxyhydroxides [1,2]. The mineral–microbe interface thus became the focus of intense efforts with regard to unravelling the mechanism(s) whereby microbes transport electrons across the outer membrane to IEAs (insoluble electron acceptors) that cannot be transported into the cell. In the subsequent years, it has been hypothesized that a number of electron-transfer mechanisms are used by cells, including: (i) direct reduction of minerals via extracellular multihaem cytochromes [37]; (ii) indirect reduction of minerals via soluble redox molecules (i.e. electron shuttles) [810]; (iii) electron transfer along extracellular appendages known as microbial nanowires [1113]; and (iv) extracellular matrices containing conductive or semi-conductive minerals [14]. Given that survival of the cells may well be dependent upon EET, it would be surprising if a number of solutions to this problem had not evolved; one might expect more to be discovered. This seems especially true when one considers that almost everything that is known about EET comes from studies of only two model systems, Shewanella [15,16] and Geobacter [17].

One area that has attracted only minimal attention is how microbes can locate IEAs. The observation that microbes utilize motility to accumulate on and in the vicinity of metal oxide particles [1820] implies that there are mechanisms for sensing and taxis. But, to date, there are no detailed explanations of how an organism ‘recognizes’ an IEA, and whereas the redox potential of a surface may be ideal in terms of electron flow, how does a microbe know that an IEA is present? It is this question that we focus on in the present paper.

Chemotaxis and energy taxis

In chemotaxis, cells swim up gradients of attractants using MCPs (methyl-accepting chemotaxis proteins) as receptors. These receptors bind the attractants directly at periplasmic ligand-binding domains or indirectly, using periplasmic binding proteins. Sensory information is routed through a two-component signal transduction system that includes a histidine protein kinase, CheA, and a response regulator, CheY, to the flagellar motor [21]. Responses require neither transport nor metabolism of the chemoattractant [2123]. Taxis to oxygen and soluble anaerobic electron acceptors, often referred to as energy taxis, involves a variation of bacterial chemotaxis, and has been observed in several other bacteria species [22,2427].

Energy taxis in Shewanella is not yet fully explained, and studies have shown that the so-called ‘chemical-in-plug’ assay can be unreliable for determining that the cells directly sense the electron acceptors using receptors [20,22,28,29]. Instead of using receptors that bind the chemoattractants as ligands, the cells may respond to a change in some energetic parameter, e.g. the redox state of an electron-transport protein or a change in the electrochemical gradient associated with the pmf (protonmotive force). Shewanella oneidensis MR-1 is rich in chemotaxis-related genes, suggesting that this bacterium is capable of a wide range of behavioural responses, and, indeed, S. oneidensis MR-1 has been shown to respond to a variety of different electron acceptors [19,30]. So far, only one of three putative CheA proteins, CheA-3, has been shown to be necessary for behavioural responses to anaerobic electron acceptors, including nitrate, nitrite, fumarate, DMSO, TMAO (trimethylamine N-oxide) and Fe(III) citrate [28,30]. A mutant lacking cheA-3 has been shown to be smooth-swimming, i.e. unable to change the direction of rotation of its flagellum [30]. An MCP with a Cache domain (SO_2240) has also been shown to be necessary for behavioural responses to a number of electron acceptors (TMAO, DMSO, nitrite, nitrate and fumarate), although deletion of this gene did not completely abolish the tactic responses [28]. Because deletion of the SO_2240 mcp resulted in loss of responses to a range of anaerobic electron acceptors, it was suggested that this MCP is an energy taxis receptor. Mutants lacking any one of the four mcp genes that encode MCPs with PAS (Per/Arnt/Sim) domains (SO_0584, SO_1385, SO_2123 and SO_3404) showed near-wild-type tactic responses to soluble electron acceptors in chemical-plug-in-pond assays [28]. However, double mutants lacking both SO_2240 and any one of the four MCPs that have PAS domains had slightly stronger phenotypes, perhaps indicating that the cells monitor more than one energetic parameter [28].

Thus it appears that S. oneidensis MR-1 cells have the necessary equipment for sensing and responding to soluble electron acceptors, but there is as yet no explanation for how they can sense and respond to insoluble metal oxides and electrodes.

Other reports suggest that riboflavin excreted by Shewanella cells is an attractant that mediates energy taxis [9]. In this case, the riboflavin is hypothesized to be excreted by electron acceptor-limited cells to create a chemical gradient for the taxis [10]. In general, unless the cells themselves create the gradient, chemical gradients do not agree with our observations of a rapid behavioural response around poised electrodes, since the latter do not release diffusing chemicals [18]. As noted above, there have been severe criticisms of the chemical-plug-in-pond and swim plate techniques used in previous studies [29]: criticisms that have motivated us to use video microscopy and cell tracking methods to study this process.

In a previous study, we used video microscopy to show that S. oneidensis MR-1 cells swim faster in the vicinity of IEAs (both metal oxides and charged electrodes) [18], a response perhaps indicative of a direct connection between electron acceptor reduction and pmf generation. In the present paper, we report, in addition to the observed increase in speed, an increase in flagellar reversal frequency, which leads to an accumulation of cells around metal oxide particles and electrodes, a response we refer to as ‘congregation’. A hypothetical model is presented to explain the phenomenon: a model that involves swimming speed enhancement upon contact with the IEA, and flagellar reversal at high swimming speeds, resulting in accumulation of cells in the vicinity of the electron acceptor, or congregation. We also report genes that, when deleted, results in phenotypes with defective abilities to congregate.


Cultivation and strains

S. oneidensis MR-1 and several deletion mutants originating from S. oneidensis MR-1 were examined in our study (Table 1, and Supplementary Figures S1 and S2 at All experiments were carried out using a previously described defined minimal medium, containing 18 mM sodium lactate as an energy source [31] (see Supplementary Table S1 at Strains were inoculated from freezer stocks on LB (Luria–Bertani) plates and then grown overnight at 30°C. Individual colonies were then selected and inoculated into defined minimal medium (M1). All strains and mutants were inoculated into 5 ml of M1 medium inside airtight 15 ml tubes (VWR International LLC) and incubated horizontally in a shaker at 180 rev./min for 48 h at 30°C (Amerex Instruments). Attenuance was measured using a spectrophotometer (Unico 1100RS spectrophotometer). Cells were sampled at a D600 of 0.48–0.52 (after ~48 h). For the metal oxide particle experiments, 5 μl of suspended mineral particles [300 mg/ml MnO2 or Fe(OH)3] was then mixed with the cell culture. Aerobic cells and minerals were mixed by inversion three times, and were then loaded by capillary action into rectangular capillary tubes (0.02 mm×0.20 mm) (Vitrocom). These tubes were then sealed (zero time) with silicone vacuum grease (Dow Corning) and then observed by microscopy. Metal oxides were synthesized as described in the Supplementary Online Data (at

View this table:
Table 1 Relevant genes used in our study

Miniature electrochemical cell and coated electrode

For electrode experiments, cells were loaded into a miniature electrochemical observation device and sealed with silicone vacuum grease at zero time (see Supplementary Figure S3 at [18]. The device was similar to a system described previously [18], but with several additional features (see Supplementary Figure S3). First, the graphite fibre electrode was coated with Teflon and then cut to expose a defined area of conductive graphite. For details regarding coated graphite electrodes, mineral synthesis, soluble electron acceptor chemicals and GFP (green fluorescent protein) time-lapse photography, see the Supplementary Online Data. Cells were placed in an electrochemical cell and the electrode first poised at +700 mV after 15 min or, alternatively, the potential was stepped up in 50 mV increments from 0 mV to +700 mV at 3 min intervals (i.e. 0, 50, 100, 150, etc. mV).

Hand-tracking analysis of cell movements

Bacterial swimming tracks (both computer and manual tracks) were calibrated using a microscope scale ruler (100 μm). From each experiment, the overall swimming activity within the video frame, equivalent to a 107 μm×193 μm field of view, was recorded and the video was time-normalized to give swimming speeds in μm/s. Several measurements were made for each bacterial swimming track: the total distance moved, the time of track (between when the bacteria first appear and disappear), the number of reversals, the distance between each reversal and the IEA, and the distance between the IEA and the start of bacteria track (see the Supplementary Online Data).


The starting position of the bacteria with respect to the nearest IEA surface was logged, and each bacterial reversal event was identified and logged with regard to the distance from the nearest IEA surface (Figures 1A and 1B). For a known time of swimming activity, the swimming cells were divided into two groups for analysis: cells that swam within 2 μm of a particle were considered ‘contacting’ and those that did not swim within 2 μm from the particle surface were considered ‘non-contacting’. In addition to the hand-tracking methods described above, some experiments (such as those in Figure 3) utilized a computer-tracking algorithm [32] (see the Supplementary Online Data for details).

Figure 1 Swimming S. oneidensis MR-1 congregate around IEAs because of reversals in swimming direction

Swimming tracks of S. oneidensis MR-1 around a stationary particle of manganese oxide (A) or an electrode poised at +700 mV compared with Ag/AgCl (B) were analysed. Particles of MnO2 were mixed with S. oneidensis MR-1 cells and then sealed in a capillary tube with lactate as the carbon source (A). Video data were recorded after the cells consumed all of the dissolved O2. The grey lines indicate the path of each individual bacterium that swam in 10 s. The red star signifies the location where a cell reversed direction. The distance from the start of the bacteria track and from any reversal event to the mineral was recorded (dash-dotted line and dotted line respectively). (C) Time-lapse fluorescent microscopy showing motile S. oneidensis MR-1 cells labelled with GFP around MnO2 (outlined with red dotted oval in leftmost frame). Each frame was taken 1 h apart (0–4 h, from left to right) and was photographed immediately after irreversibly photobleaching the protein at zero time. Scale bar, 100 μm. There appears to be accumulation of motile bright cells (from outside of the frame) near the mineral and attached to the surface.

S. oneidensis MR-1 cells swim faster in proximity to metal oxide particles and poised electrodes

Cells that contacted IEAs (either metal oxides or charged electrodes) swam at significantly higher speeds than cells that did not contact these surfaces (Figures 1A and 1B and Supplementary Movie S1 at Figure 2(A) shows that cell swimming speeds near the exposed tip of a Teflon-coated electrode poised at +700 mV compared with the Ag/AgCl electrode are similar to the swimming speed in response to a Fe(OH)3 or MnO2 mineral (Table 2 and Supplementary Movie S2 at Cells lack a swimming response during 0–500 mV applied potential (Supplementary Figure S2 and Supplementary Movie S3 at The coated electrode, with a defined area (~700 μm2) of conductive graphite at the tip (Figure 1B and Supplementary Figure S3B), allowed us to approximate the particle size and redox potential of small metal oxide particles found in sediment. Unlike the previous study that recorded only swimming speeds [18], our analysis (Figures 1A and 1B) includes the starting position, and the position of all reversal events (red star) of each individual cell (dot-dashed line and dotted line respectively).

Figure 2 Congregation behaviour of swimming S. oneidensis was characterized by increased swimming speed and increased reversal frequency of cells near IEAs

(A) Elevated swimming speeds of bacteria occur within 60 μm of the IEA surface [Δ, MnO2; ○, Fe(OH)3; ×, electrode poised at 700 mV compared with Ag/AgCl]. (B) The reversal timing allowed bacteria to congregate around the acceptor surface rather than swimming away. (C) A plot of average reversal frequency against binned speed of swimming S. oneidensis MR-1 indicates that stimulated cells (25–45 μm/s) reverse direction more often than less-stimulated cells (5–25 μm/s). (D) Total number of data points collected for each type of experiment, which were then used to generate the graphs (AC). Overlapping S.D. values in (AC) indicate a similar reversal and speed response in the vicinity of MnO2, Fe(OH)3 minerals or electrode poised at +600 mV compared with Ag/AgCl. This specialized reversal timing allowed the bacteria to congregate and attach to electron-accepting surfaces. The contacting S. oneidensis cells can transfer electrons directly to IEAs via specialized outer membrane cytochromes.

View this table:
Table 2 Comparison of reversal frequency and speed of swimming S. oneidensis within a population: those that contact insoluble electron acceptor surface compared with those which have not contacted

The data were collected after bacteria and mineral were sealed in the capillary for 15 min. The swimming tracks within the same experiment are sorted into two separate groups based on swimming path: those that contact insoluble electron acceptor surface (swam within ≤2 μm) compared with those which have not contacted (≥2 μm). Results are means ±2 S.D. The pairs of letters highlight measurements that are significantly different (P<0.05) from each other; those without are not statistically different. These paired measurements also correspond to the observation that strains successfully congregated around mineral after 30 min.

Swimming S. oneidensis MR-1 cells congregate and then attach to the MnO2 surface

Using strains labelled with GFP, it was demonstrated that many of these swimming cells eventually (in 0.5–4 h) become attached to mineral surface and electrode (Figure 1C and Supplementary Figure S4 at The time course in Figure 1(C) starts when the GFP-labelled cells in the tube are photobleached at zero time (leftmost frame). Then new motile cells move from outside of the bleached area into the darkened area, eventually attaching to the mineral surface (outlined in red dotted oval in leftmost frame).

In the proximity of metal oxides and charged electrodes, S. oneidensis MR-1 cells exhibits more swimming reversals

Changes in swimming speeds alone do not account for the congregation response. Video microscopy and manual tracking of cell motility around mineral surfaces and charged electrodes revealed that direction of swimming was also altered. Because S. oneidensis MR-1 has a single polar flagellum, reversal of swimming direction is accomplished simply by reversal of flagellar rotation. At 30 min after the aerobic cells were added to the capillary containing an IEA, most cells cease swimming, whereas a small percentage, located near the minerals continue swimming (Figures 3A and 3B, and Supplementary Movies S1, S4 and S5 at Video microscopy and manual tracking of cell motility around manganese oxide particles (at a 30 min time point), showed that cells, which contacted the MnO2 particle, had a significantly higher reversal frequency than cells swimming further away from the particle. Figure 2 shows hand-tracked data for S. oneidensis MR-1 cells swimming around a large MnO2 particle. The contacting cells and non-contacting cells for each experiment can be found in Supplementary Table S2 at The reversal frequency of the contacting cells was 0.944±0.53 reversals/s, whereas the reversal frequency of the non-contacting cells was 0.627±0.73 reversals/s. The increased reversal frequency of cells surrounding the IEA then provides a mechanism for the congregation.

Figure 3 Number of swimmers declines to near zero except for wild-type strains in the presence of IEAs

Mean (±4 S.D.) number of swimming bacteria (per s) after being sealed in a capillary (zero time) in the vicinity of insoluble MnO2 (A) or Fe(OH)3 (B) mineral particle or with no mineral added (C). The strains are labelled S. oneidensis MR-1 (●), ΔSO_2240 (□), ΔSO_1385 (Δ), ΔSO_3207(×) and ΔcymA (○). The Δmcp_cacheSO_2240), ΔcheA-3SO_3207) and Δmcp_pasSO_1385) mutants formed less effective congregations compared with wild-type around Fe(OH)3. This was apparent when comparing cell swimming 30 min after being sealed with the mineral. The ΔcheA-3 and Δmcp_cache strains were entirely unable to congregate around the mineral.

Electron-accepting electrodes also induce changes in the behaviour of S. oneidensis MR-1 cell swimming

A previous study showed that +700 mV compared with the Ag/AgCl electrode, a potential that mimics the redox potential of the Mn(IV)/Mn(II) couple, induced a rapid swimming response in S. oneidensis MR-1 cells [18]. This redox potential was therefore selected for additional cell tracking to determine changes in reversal frequency. Figure 1(B) shows the congregation response of S. oneidensis MR-1 cells swimming around an electrode poised at +700 mV after 1 h. However, if +700 mV is applied only 15 min after the cells are sealed in an anaerobic capillary, the congregation response is intensified (Table 2). Analysis of the tracks, after 15 min, indicated that swimming cells in the contacting group swam at a speed of 19.54±11.8 μm/s with a reversal frequency of 0.551±0.51 reversals/s, whereas non-contacting cells swam more slowly (8.28±3.27 μm/s) with far fewer reversals (0.042±0.08 reversals/s) (Table 2). Applied voltages of 550–800 mV elicited the maximum swimming response and reversal timing conducive to congregation in wild-type, whereas 0–500 mV applied resulted in fewer swimmers (Supplementary Figure S2). As can be seen, when the distance from the electrode increases, the swimming speed and the reversal frequency decrease (Figure 2).

Changes in reversal frequency correlate with changes in swimming speed in S. oneidensis MR-1 cells around IEAs

We plotted the average reversal frequencies against swimming speeds, grouped in 5 μm/s increments. This plot revealed a correlation between the frequency of reversals and the swimming speed: the faster the cells swam, the more often those cells reversed (Figure 2C). This relationship was seen for cells around the electrode at +700 mV, and around MnO2 and Fe(OH)3 mineral particles (Figure 2 and Table 2).

Cell response varies with time

Figures 3(A) and 3(B) depict swimming and motility after being sealed inside an anaerobic capillary tube where oxygen was consumed and cell motility persisted only around the minerals. As mentioned above, if cells were exposed to the charged electrode at 15 min after the capillary was sealed, the response was far greater than if the cells were allowed to sit anaerobically for 1 h before charging the electrode (Table 2, rows 3 and 4). This can also be seen directly in the Supplementary Movies, where in one experiment, the +700 mV compared with Ag/AgCl potential was applied 15 min after bacteria were sealed inside the capillary tube (Supplementary Movie S2), and in another the charge was applied 1 h after bacteria were sealed inside a capillary tube (Supplementary Movie S3). These data are summarized in Table 2.

This difference in timing is critical when mutant studies are being carried out, as many of these strains survive poorly in the absence of an electron acceptor, and if they cannot congregate or they cannot respire via EET, they will die rapidly. After 15 min, the cells had consumed the oxygen (as determined using an electrode), and most strains showed some motility around the IEAs; however, by 30 min, several strains, including ΔcymASO_4591) and Δmcp_cacheSO_2240), were completely non-motile around MnO2, Fe(OH)3 and poised electrodes (Figure 3). Interestingly, the Δmcp_pas mutant (ΔSO_1385) found to be present in many Shewanella strains, exhibited wild-type levels of motility and reversals with regard to MnO2, but irregular or no response to Fe(OH)3 particles or poised potentials (Supplementary Movie S5).

Increased reversal frequency after contacting IEAs is essential for congregation

The receptor protein and histidine protein kinase required for energy taxis in S. oneidensis MR-1 have been identified [28]. Mutants lacking the chemotaxis proteins, i.e. mcp_cacheSO_2240), mcp_pasSO_1385) and cheA-3SO_3207), or the EET cytochromes, i.e. cymASO_4591), were screened for their response to MnO2, Fe(OH)3 and the poised electrode. In response to Fe(OH)3, these mutants showed a significant (P<0.05; Student's t test) decrease in reversal frequency compared with wild-type S. oneidensis MR-1 (Figure 4). In the case of the ΔcheA-3 mutant, this behaviour was expected because the mutant is unable to reverse its direction of motion. In contrast, the ΔSO_2240 cells, although capable of reversal, showed no significant difference in reversal frequency or swimming speed compared with those in the non-contacting group (Figure 4 and Supplementary Table S2).

Figure 4 Deletion mutants lacking MCPs (ΔSO_1385 or ΔSO_2240), chemotaxis transduction protein (ΔSO_3207) or EET cytochromes (ΔSO_4591) are unable to congregate around MnO2 or Fe(OH)3

Reversal frequencies of swimming bacteria within a population: those that contacted insoluble mineral surface compared to those that did not. Experiment data, from added MnO2 (A) or Fe(OH)3 (B) to bacteria cultures of S. oneidensis MR-1, Δmcp_cacheSO_2240), Δmcp_pasSO_1385), ΔcymASO_4591) and ΔcheA-3 ΔSO_3207 were divided into two subpopulations for analysis. This allowed for a comparison of reversal frequency and speed of swimming bacteria within a population. A division was made between bacteria that contacted (light grey) insoluble acceptor surface (i.e. swam within ≤2 μm) compared with those that were swimming, but had not contacted the surface (dark grey). The contacting group was significantly faster (Supplementary Table S2 at and reversed direction more often than the non-contacting group, in the following experiments: S. oneidensis MR-1 with MnO2 (A), Δmcp_pasSO_1385) with MnO2 (A) and S. oneidensis MR-1 with Fe(OH)3 (B). ΔSO_3207 exhibited a smooth-swimming phenotype with no reversals, and is therefore not shown. Results are means ±4 S.D.

Figure 4 displays reversal frequencies of contacting and non-contacting cells in response to MnO2 or Fe(OH)3, which highlight significant reversal frequency timing irregularities compared with wild-type. The ΔSO_2240 mutant exhibited irregular reversal timing around both MnO2 and Fe(OH)3, whereas the ΔSO_1385 reversal phenotype is similar to wild-type around MnO2, but irregular around Fe(OH)3 (Figure 4). Neither the ΔSO_2240 nor the ΔSO_1385 mutant responded with swimming to any voltage, during the applied potential iterations, compared with the wild-type swimming response (Supplementary Figure S2). Additional deletion mutants of genes that contain the PAS domain, ΔSO_0584, ΔSO_2123 and ΔSO_3404 showed no difference in phenotype from wild-type in response to insoluble acceptors (results not shown).

EET is required for congregation in S. oneidensis MR-1

Mutants unable to perform EET are unable to congregate. For example, ΔcymA failed to congregate around minerals and poised electrodes. After 15 min, swimming ΔcymA cells contacting the MnO2 or Fe(OH)3 did not reverse significantly more than non-contacting cells (Figure 4). By 30 min, ΔcymA cells were completely non-motile around MnO2, Fe(OH)3 and poised electrodes. However, ΔfccA mutant strain (ΔSO_0970) lacking cytochromes which are non-essential for EET did not respond significantly differently from wild-type.

Model of congregation

The results of our study have led us to propose a simple hypothetical model that provides an explanation for how these microbes congregate near to IEAs (Figure 5). Congregation could be of substantial value in environments where rapid redox cycling occurs, particularly in sediments, where dissolved oxygen can change dramatically and quickly [33], not unlike the situation at the beginning of our experiment when the capillary was sealed with added MnO2.

Figure 5 Proposed congregation model

Aerobic cells swim stochastically in all directions, reversing 0.5 times per s (A). Within 10 min of being sealed in the capillary with metal oxide, motile cells consume all available dissolved O2 and begin to randomly contact mineral, passing electrons via the specially adapted EET chain and build a pmf. (B) Now anaerobic, the cells making contact congregate by sensing intercellular pmf (mediated by MCPs SO_2240 and SO_1385). Using appropriately timed reversals, via the chemotaxis pathway, stimulated cells often return to mineral resulting in congregation (C) and eventual attachment. A model of the process is shown on the right.

The simple model, shown in Figure 5, consists of the following steps. (i) Initially, the cells are highly motile, utilizing dissolved oxygen as the electron acceptor and seldom reversing direction (Figure 5A). As oxygen is depleted, swimming speed decreases, and after 15–30 min, all cells are non-motile except for those that have incidentally encountered an electron acceptor (metal oxide or poised electrode). This is a stochastic process that continues throughout the experiment. Of the total cells in the capillaries, only 1–3% are motile. (ii) These contacting cells interact with the particle via a series of electron carriers to the outer membrane (Mtr) protein complexes. The contacting cells are energized to swim, resulting in the previously described electrokinesis response [18]. (iii) These fast-swimming cells undergo rapid flagellar reversal and directional reversal, characteristic of a monotrichous cell [34], and, by this response, are maintained in the vicinity of the solid-state electron acceptor. This is a directed response that occurs in addition to the stochastic recruitment of new cells. (iv) Swimming cells continuously attach to the electron acceptor, eventually forming a biofilm.

Significance of PAS domain MCP (SO_1385)

Several studies have hypothesized that four MCPs with PAS domains are in some way responsible for energy taxis or response to IEAs [2830]. However, up to this point, no study has found SO_1385 genes to be essential for response to electron acceptors in Shewanella [28]. Our results, using analysis of cell swimming tracks, suggest that SO_1385 may have a role in congregation around IEAs, where the mechanism is fundamentally different from that needed to locate a soluble electron acceptor. We found that one of the four putative PAS domain-containing signal transducers (SO_1385) appears to play a significant role in congregation around IEAs with low redox potential such as Fe(OH)3, whereas the other three signal transducers (SO_0584, SO_2123 and SO_3404) appear to play no essential role in congregation. Genetic analysis shows that the SO_1385 gene is the most abundant in the 17 Shewanella species (present in 12 of the 17 Shewanella species analysed) and encodes a PAS domain-containing receptor [35]. These findings are noteworthy because transcriptomic analysis of wild-type S. oneidensis MR-1 revealed specific up-regulation of this SO_1385 gene under Fe(III)- or Mn(IV)-reducing conditions. Furthermore, the gene (SO_1385) has been shown to share 58% homology with Escherichia coli aerotaxis transducer [35] and is therefore a strong candidate as a flavin-containing redox/energy taxis transducer in S. oneidensis MR-1. Previously, this PAS sensory protein was thought to play only a minor or insignificant role in S. oneidensis MR-1 energy taxis in response to electron acceptors [28].

The need for EET

One prediction of this model is that it should require EET in order to activate the cells. Not surprisingly, mutation of any of the genes involved with EET abolished congregation, as they had been reported to abolish electrokinesis [18]. In addition, increasing the IEA surface area, for example by increasing the conductive surface area of an electrode by using less insulation coating, but applying the same surface charge also increased the speed of swimming (Table 2 and Supplementary Figure S5 at At first glance, it is tempting to define this congregation behaviour as a variation of energy taxis, as it is likely to be a metabolism-dependent response. But, given the possible involvement of the EET chain and multiple MCP interactions (receptors with both PAS domains and Cache domains), this behaviour seems to be distinct from all previously defined behaviours. Therefore we refrain from making this distinction until further research can be performed.

The need for a sensing mechanism

Another prediction of the congregation model is that there must be some sensing mechanism involved that can lead to control of flagellar reversal. The fact that reversal frequency is increased in bacteria with higher speed (and thus closer to IEA) provides a way for the bacteria to congregate around the IEAs, but what is the sensing mechanism? Experiments with several mutants known to be involved with chemotaxis in S. oneidensis [20,2830] suggest some answers. For example, in a mutant lacking a functional chemotaxis protein CheA-3 (ΔSO_3207), congregation is totally eliminated. A cheA3 mutant in E. coli leads to a phenotype in which flagellar reversal is inhibited [30]. Assuming this mutant is unable to reverse flagellar rotation, one would predict that cells would be stimulated to become motile by random contact with the electron acceptor, but would almost never return to the insoluble particle (Figure 4 and Supplementary Figure S1).

Similarly, a mutant lacking a functional MCP coded for by the gene SO_2240 was incapable of congregation. This MCP is a Cache domain-containing protein that is thought to act by sensing the transmembrane potential in E. coli [28]. It is easy to see how such ability could be coupled to the congregation response. One possibility is that a rapid increase in pmf that might occur upon contact with the IEA would stimulate flagellar reversal: another might be that the pmf is constantly monitored and, as it increases, the probability of flagellar reversal also increases. This possibility is now under investigation (Figure 4 and Supplementary Figure S1).

Mutation of genes coding for PAS domain proteins led to an interesting incongruity that was seen with regard to congregation around hydrous ferric oxide, in that one mutant was observed that blocked the congregation around iron, but showed no effect on the congregation around manganese particles. This was the MCP PAS domain-containing protein coded for by SO_1385. Given the low potential of iron oxide in comparison with MnO2, this might not be particularly surprising, but it may also provide some clues about the interaction of the PAS domain-containing MCPs, which are now under more detailed study (Figure 4).

Will this work in E. coli?

Engineered E. coli (E. coli mtrCAB) has been shown to be capable of EET [36]. Our results show that wild-type E. coli cannot congregate around IEAs. This raises the question of whether or not the engineered E. coli strain, with transplanted mtrCAB genes from Shewanella, is capable of this behaviour. Our hypothesis is that it would not be capable of congregation. Despite having mcp_pas, mcp_cache and mtrCAB (EET cytochromes) genes, it cannot congregate because the strain lacks a single polar flagellum. On the basis of our tracking data (Figures 1A and 1B), we believe that this response must be limited to monotrichous bacteria that will be capable of returning to the surface of IEAs by a series of runs and reversals [34]. According to our model, even with the functioning Shewanella mtrAmtrC genes expressed in E. coli, allowing this organism to reduce solid metal oxides [36], congregation behaviour should not occur. Although an E. coli with added Shewanella MCPs [30] cannot perform true congregation around IEAs not only because it lacks the mtrAmtrC genes, but also because its response to flagellar reversal will be to tumble rather than to reverse, and the probability of returning to the IEA surface will be vanishingly small.

Why congregation?

The term congregation describes the observed motility driven accumulation on the surface and in the vicinity of IEAs: a distinctive type of behaviour that cannot be put into any of the presently known bacterial response definitions. Our results, which characterized the response around IEAs, do not support the idea of chemotaxis towards a small amount of soluble electron acceptor [37]. Neither do our data fit the previously defined energy taxis paradigm [27]. As the molecular mechanism(s) involved becomes clear, the relationship between congregation and other more well-known tactic responses should become clear.

The biological rationale for this behaviour is also not yet clear, but it should be considered in the context of the kind of environment that these microbes encounter, where rapid limitation of either electron donors and/or electron acceptors can occur. Thus commitment to either an electron donor or an electron acceptor may constitute an important regulatory ‘decision’ retaining the capacity to move from the zone of electron donor excess to the zone of electron acceptor excess may be a very positive adaptive trait [27,38]. For example, in our experiments, we employed a high level of electron donor (18 mM lactate), which is almost certainly seldom encountered in Nature. If cells settle on the IEA surface in Nature, they will almost certainly become rapidly electron-donor-limited. Congregation provides a way to avoid both electron donor and IEA limitation. In fact, when metabolism of IEA particles is observed, they are often completely degraded with minimal cell attachment despite extensive congregation activity (Supplementary Movie S6 at

Post-congregation activities

Initial studies with different strains of Shewanella indicate that congregation is an important first step in the attachment and biofilm formation by several different microbes and that these processes are closely linked (H.W. Harris, J.S. McLean, M.Y. El-Naggar, E.C. Salas and K.H. Nealson, unpublished work); i.e. a strong congregation response under a given condition leads to attachment and biofilm formation. If so, then this mechanism is potentially of great importance with regard to natural environments where redox chemistry and electron exchange occur.


This work is supported by an Air Force Office of Scientific Research Award [grant number FA9550-06-1-0292].


Special thanks to Mandy J. Ward for advice on research and Jeff McLean for experiment design. We thank Meaghan Sullivan and William Tran for their manual tracking analyses. We thank Cécile Jourlin-Castelli, Samantha Reed, Jun Li and David Culley for supplying the Δmcp_cache, ΔmtrB, ΔmtrA, Δmcp_pas, ΔcheA3 and ΔcymA mutants.


  • Electron Transfer at the Microbe–Mineral Interface: A Biochemical Society Focused Meeting held at University of East Anglia, Norwich, U.K., 2–4 April 2012. Organized and Edited by Jim Fredrickson (Pacific Northwest National Laboratory, U.S.A.), David Richardson (University of East Anglia, U.K.) and John Zachara (Pacific Northwest National Laboratory, U.S.A.).

Abbreviations: EET, extracellular electron transport; GFP, green fluorescent protein; IEA, insoluble electron acceptor; MCP, methyl-accepting chemotaxis protein; PAS, Per/Arnt/Sim; pmf, protonmotive force; TMAO, trimethylamine N-oxide


View Abstract