Analysis of electron transfer dynamics in mixed community electroactive microbial biofilms

Abstract

Mixed community electroactive organisms form multi-layered biofilms that are able to produce current densities comparable to those of pure Geobacter sulfurreducens, an extensively studied metal-reducing organism. The long-range electron transfer (ET) inside the biofilms and at the biofilm/electrode interface was proven to be promoted by a network of outer membrane cytochromes (OMCs). In the present work, we investigate the electron transfer process in mixed community biofilms grown on Indium Tin Oxide (ITO) electrodes by combining electrochemical measurements with Confocal Resonance Raman Microscopy (CRRM) under potentiostatic control and during chronoamperometry (CA). This approach allowed direct comparison of the heterogeneous redox process at the biofilm/electrode interface with the long-range OMCs-mediated ET inside the bulk biofilm. Our work shows that: (i) during substrate oxidation, all OMCs are in the reduced state at any distance from the electrode, and no concentration gradient of oxidized OMCs is observed; (ii) the rate constant for the long-range, homogeneous ET (k⁰_hom) is 0.028 s⁻¹, which is considerably lower than that predicted by others under the hypothesis that homogeneous ET is promoted by OMCs alone, and may thus indicate the contribution of alternative fast electron transfer processes; (iii) the metabolic respiration rate is much faster compared to both homogeneous and heterogeneous ET, which have similar rate constants. All in all, our results suggest that differences exist in electron transfer mechanisms between mixed community and G. sulfurreducens electroactive biofilms.

---

Introduction

Investigation of electron transfer (ET) mechanisms in electroactive microbial biofilms (that is, biofilms capable of extracellular electron communication with electrodes) has motivated substantial research efforts in recent years. Unraveling electron conduction in microbes is relevant not only to engineered devices such as biosensors and bioelectrochemical systems, but it can also help to explain important microbial physiological processes governing natural geochemical cycles in sediments. Amongst the most studied microbial players in electroactive biofilms are bacteria belonging to the genus Geobacter and Shewanella, because of the high current densities achievable at electrodes, and their importance as model organisms of ET modes. Geobacter sulfurreducens can develop thick biofilms on electrode surfaces, spanning distances exceeding 100 μm. The mechanisms by means of which electrons are transported through such long distances is the subject of intense debate.^1–3 Proposed models include (a) redox conduction, that is, electron transfer via incoherent multistep electron hopping between discrete redox cofactors bound to the biofilms,⁴ (b) metallic-like conductivity, according to which electron transfer occurs through intrinsic conducting properties of extracellular appendages (called nanowires) specific to Geobacter sulfurreducens.⁵ Redox conduction is based on the relatively high abundance of multiheme ctype cytochromes located on the outer membrane and periplasmic space, as well as along extracellular filaments, and dispersed in the extracellular polymeric substance.⁶ On the contrary, metallic-like conduction excludes hopping between cytochromes,^7,8 since the inter-heme spacing – measured by immune-gold labelling outside of the bioelectrochemical system – is too large for electron hopping to occur.⁹ While sufficient structural information regarding the spatial organization of heme-proteins in biofilms under physiologically relevant conditions is still missing, independent electrochemical and spectroscopy measurements seems to agree on the important role played by ctype cytochromes in wiring the electroactive biofilms to electrodes and providing electric conductivity.^1,10–21

Redox conduction is described by a detailed model consisting of a fast heterogeneous electron transfer at the biofilm/electrode interface governed by electron tunnelling, coupled with a slower long-range homogeneous electron transfer occurring via hopping mechanism.^17,18 This model requires a redox gradient as driving force to allow ET to a given direction.²² Recently, Bonanni et al.²³ proposed a model where the contribution of both OMCs and nanowires are considered. While this seems to explain electron conduction in G. sulfurreducens biofilms, electron transfer may be different in mixed culture films, where the presence of multiple species may add complexity to the already complicated spectrum of synergies and competitions that are expected to arise for available substrates and the terminal electron acceptor (i.e., the anode). While studying conduction in model organisms such as G. sulfurreducens is important from a physiological and mechanistic point of view, understanding ET in broader mixed culture microbial aggregates is essential also because of the role they play in existing and potential applications in environmental and industrial biotechnologies.^24–26

Spectroelectrochemical methods that derive from the combination of electrochemical techniques with various spectroscopies provide great opportunities for the study of electron conduction in these systems. For instance, compared to established electrochemical methods such as cyclic voltammetry (CV), techniques based on Raman scattering provide important structural information of the matrix based on spectral fingerprint of key molecules.²⁷ Resonance Raman (RR) spectroscopy is particularly suitable for the study of electron transfer involving cytochromes because of the signal enhancement achieved when the frequency of the excitation laser line is in proximity to an electronic transition of the heme group in a cytochrome. This selectivity towards heme-containing proteins together with the spatial resolution achievable by combining RR spectroscopy with microscopy (yielding confocal resonance Raman microscopy, CRRM), allows the use of CRRM to monitor surface-exposed cytochromes promoting the long-range ET at any distance from the electrode surface, as well as periplasmic cytochromes involved in cells metabolism and extracellular ET. However, due to their abundance and accessibility, surface-exposed proteins, hereby denoted as OMCs, dominate the spectrum.¹³ Using this approach, we have previously resolved the spatial distribution²⁸ and redox state²⁰ of OMCs of biofilms in vivo, without impacting on the catalytic activity of mixed culture biofilms.

Herein, we improved our experimental set up and devised a miniature spectroelectrochemical cell (schematized in Fig. S1†) equipped with transparent Indium Tin Oxide (ITO) working electrodes specifically designed to allow simultaneous electrochemical and CRRM measurements on the biofilms directly (that is, in situ) without the need to transfer the biofilm/electrodes from the electrochemical cell to the microscope stage; a procedure that inevitably exposes the biofilms to air, albeit briefly. This new set up allows to compare our results on mixed cultures to what others have done on G. sulfurreducens biofilms.¹⁶ This approach excludes that differences observed for mixed cultures and G. sulfurreducens are due to the experimental configuration such as the focusing of the laser beam and the electrode material. Moreover, CRRM hereby applied in the time-resolved mode, was used to support the interpretation of simultaneous chronoamperometry (CA) and CRRM measurements. To our knowledge, this combined approach was never attempted before, and it turned out to provide much direct information about the rate limiting steps of electron transfer processes in electroactive microbial biofilms.

Material and methods

Biofilms formation and growth medium

Primary biofilms were enriched using domestic wastewater from a local wet well as the inoculum and incubated in a sealed single-chambered bioelectrochemical system consisting of two carbon rods (Morgan AM&T, Australia) serving as working and counter electrodes, and a Ag/AgCl reference electrode in 3 M KCl (MF-2052, Basi, USA). All potentials herein are reported with respect to this reference electrode (+0.210 V vs. the standard hydrogen electrode, SHE). The electrodes were immersed into 400 mL sterile anaerobic media (composition below) purged with pure nitrogen for at least 30 minutes to ensure anoxic conditions. The working electrode was poised at 0 V using a potentiostat (Potentiostat/Galvanostat VMP3, BioLogic Science Instruments, France). Biofilm growth was monitored by measurement of bioelectrocatalytic current production and by regular cyclic voltammetry analysis. The media was regularly exchanged (usually once per week). Primary biofilms were scraped off the rods using a sterile blade and used as inoculum for the formation of secondary biofilms that were used in this study. Both primary and secondary biofilms were grown at room temperature (22 ± 1) °C. The growth medium consisted of autoclaved deionized water containing: Na₂HPO₄ (6.0 g L⁻¹), KH₂PO₄ (3.0 g L⁻¹), NH₄Cl (0.1 g L⁻¹), NaCl (0.5 g L⁻¹), MgSO₄·7H₂O (0.1 g L⁻¹), CaCl₂·2H₂O (0.015 g L⁻¹), trace elements solution (1 mL L⁻¹, composition in Lu et al.²⁹). Sodium acetate (CH₃COONa) was used as metabolic substrate at concentrations that ranged between 1 and 20 mM as indicated in the text.

Spectroelectrochemical cells

The spectroelectrochemical cells used for CRRM measurements were designed to allow for observations of electroactive biofilms in vivo and in situ. A schematic cell is provided in Fig. S1 in the ESI.† It consists of single chambered miniature BESs obtained by etching a 10 × 10 mm well into a 10 mm thick polycarbonate frame. Small channels were drilled on the sides of the frame to accommodate the reference electrode and the counter electrode (consisting of a platinum wire). The working electrode consisted of 20 × 20 mm sodalime glass cover slip (thickness 0.5 mm) coated with 100 nm of Indium Tin Oxide (ITO 90/10 wt%, 99.99%, Testbourne Ltd., UK) deposited using an electron beam evaporator (BJD2000, Temescal, USA). The use of ITO allowed to perform confocal analysis directly from the outside of the electrochemical cells, without the need to remove the biofilm from the medium and expose it to air, albeit for a brief period of time. The glass/ITO sandwich was plasma-treated to render the surface highly hydrophilic and favour bacterial attachment,³⁰ and it was then glued onto one side of the well using silicone glue, resulting in an effective (exposed) area of the electrode of 1 cm². Finally, a glass cover slip was glued to the opposite side of the well to close the electrochemical chamber, resulting in an internal volume of 1 mL. The electrochemical cells were fed using a multichannel syringe pump (NE-1600, New Era Pump Systems, USA) at flow rates ranging from 0.1 to of 1 mL h⁻¹. External connection of the ITO was obtained by gluing titanium wires on the portion of the ITO layer outside of the chamber using conductive epoxy glue (MG chemicals, USA). Working, counter, and reference electrodes were typically connected to a multichannel potentiostat (CHI1000B, CH Instruments, USA). For the electrochemical measurements during Raman spectra acquisition, a VMP3 Potentiostat/Galvanostat (BioLogic Science Instruments, France) was used. Biofilms were grown at a potential of 0 V. Measurements of catalytic current production over time and of cyclic voltammograms were used to monitor the electrochemical activity of the biofilms.

Confocal resonance Raman microscopy measurements

All CRRM measurements were performed at room temperature (22 ± 1 °C) using an Alpha 300 Raman/AFM (WITec GmbH, Ulm, Germany) equipped with a frequency-doubled continuous-wave Nd:YAG laser to obtain a 532 nm excitation line. The laser beam was focused by an objective lens (Nikon 40X, N.A. 0.6, CFI S Plan Fluor ELWD objective). The back-scattered Raman light from the sample was collected with a 100 μm optical fibre employing a Raman spectrometer (1800 grooves per mm grating) with a charge-coupled device (EMCCD) spectroscopic detector. Biofilm stability upon exposure to the 532 nm laser was tested by measuring changes of the signal intensity of the vibrational mode ν₁₅ (sensitive to ctype heme groups, vide infra) at 750 cm⁻¹ during a 300 s continuous exposure test (Fig. S2†). Measurements confirmed that changes of signal intensity for the mode ν₁₅ were negligible (<2.4 CCD counts s⁻¹) for laser powers less than 430 μW (measured at the sample using a power meter (Thor Labs, USA)). Project FOUR software (WITec GmbH, Ulm, Germany) was used for spectra processing and image reconstruction. OriginPro 9.1 software (OriginLab, Northampton, USA) was used for data fitting. Depth measurements on biofilm sections were done by collecting individual spectra over a total of 12 points spaced 5 μm from each other (covering 60 μm-thick sections), and using an integration time of 20 s per point (refer to the ESI† for additional information regarding this measurement).

Time-resolved simultaneous CRRM and CA measurements were conducted by continuously collecting RR spectra from a single biofilm location 5 μm in depth from the electrode surface. During the tests, the potential of the working electrode was stepped from an initial value (E_i) to a final value (E_f) to perturb the original redox equilibrium E_i. While E_i was set at −0.5 V, E_f of −0.2 V, 0 V, and +0.2 V where used. The E_i was kept for a total of 300 s, after which the potential was instantaneously shifted to the E_f and kept for additional 240 s. The subsequent relaxation profiles of current and intensity of the Raman vibrational mode ν₁₅ at about 750 cm⁻¹ (used as marker for the redox state of OMCs²⁰) was used to monitor the redox processes occurring during the potential transition experiment. To avoid over exposure of the biofilm to the laser beam, the collection of the RR spectra started at t = 240 s, that is, 60 s before the transition, and continued for the remaining of the test. Spectral acquisition was done at integration time of 0.2 s. To further confirm the redox state of the probed OMCs during the simultaneous CRRM and CA measurements, additional experiments were carried out where single spectra were collected during 60 s prior to the transition, and then again 180 s after the transition. The integration time for these tests was 60 s.

Results and discussion

Analysis of the voltammetric response of electroactive biofilms

Biofilms were incubated at a potential of 0 V until stable catalytic substrate oxidation was achieved as indicated by current profiles (data not shown). Confocal laser scanning microscopy (CLSM) micrographs using fluorescent in situ hybridization (FISH) labeling showed the electroactive organisms forming a homogeneous coverage of the electrode surface with an average thickness of (17 ± 4) μm (Fig. S5†). Analysis of the individual cell clusters scraped off the electrode confirmed the presence of Geobacter spp. (Text S2†). The voltammetric response of the biofilms was monitored under both turnover and non-turnover conditions, that is, in the presence and in the absence of the metabolic substrate (acetate), respectively. Typical voltammograms are reported in Fig. 1. Measurements in turnover conditions revealed the presence of one catalytic redox-active site E^f centered at −0.350 V. This value is very close to the arithmetic average (E_1/2 at −0.346 V) of the two redox couples E^f,1 and E^f,2 revealed under non-turnover conditions and centered at formal potentials of −0.389 V and −0.303 V, respectively, thus suggesting the involvement of both redox couples in the catalytic current. This voltammetric response has been typically observed in Geobacter enriched electroactive biofilms on graphite, roughened silver and glassy carbon,^14,20,31 demonstrating that ITO represents a suitable electrode material for electroactive biofilms, as previously shown by others.^{12,13,16,32,33}

Fig. 1 Characteristic cyclic voltammograms of the electroactive biofilms recorded at the scan rate of 1 mV s⁻¹. (a) Turnover CV recorded in the presence of acetate (20 mM), where E^f indicates the putative electron transfer site centered at −0.350 V. Inset indicates the first derivative of the turnover CV. (b) Non-turnover CV in acetate-depleted medium, where E_1/2 centered at −0.346 V indicates the arithmetic average of two redox couples E^f,1 and E^f,2.

Analysis of the RR spectra of biofilms in completely reduced or oxidized electrochemical conditions in the absence of metabolic substrate

The spectroelectrochemical cells used in this study allowed for CRRM measurements to be performed directly on the biofilms in their culturing media without the need to expose them to air while transferring onto the microscope stage. This represented a major advancement from our previous set ups.^20,28 The use of transparent ITO electrodes deposited on a glass substrate allowed for the Raman observations to be performed non-invasively from the outside of the spectroelectrochemical cell, providing minimal disturbance to the biofilms.

RR spectra of biofilms collected under potentiostatic conditions in the absence of acetate are reported in Fig. 2. Bands assignment is shown in Table 1. Consistently to what we reported previously, the vibrational bands ascribed to ctype heme groups of the cytochromes embedded in the biofilms dominate the RR spectra.²⁰ The mode ν₁₅, observed at around 750 cm⁻¹ is ascribed to the pyrrole breathing. The intensity of this band is proportional not only to the presence of ctype hemes,³⁴ but also to the amount of cytochromes in the reduced state.^20,35–37 Modes in the mid-frequency region (1100–1700 cm⁻¹), ascribed primarily to stretching vibrations of the porphyrin ring, are indicative of oxidation-, spin-, and coordination-state of the Fe atom.³⁸ With the anode poised at 0 V – hence more positive than the average of the macroscopic redox potentials E_1/2 as determined by non-turnover voltammetry discussed above – the modes ν₂₁, ν₄, ν₂₀, ν₂, and ν₁₀ were centered respectively at 1313, 1369, 1400, 1583, and 1635 cm⁻¹ and a shoulder appeared in correspondence of the mode ν₁₁ at 1563 cm⁻¹, consistently with an oxidized heme group. Conversely, the application the negative potential of −0.6 V (i.e., more negative than the E_1/2) caused the modes ν₂₁, ν₄, and ν₂₀, to downshift to 1310, 1360, 1391, respectively, while the ν₃ (not resolvable at 0 V) appeared at 1494 cm⁻¹. The mode ν₁₀ did not shift although it reduced considerably its intensity, and the mode ν₁₁ was no longer resolvable. The band ν₁₅ also did not shift significantly in position, but its intensity was considerably enhanced by the application of the negative potential. These changes are consistent with an oxidized (at 0 V) and reduced (at −0.6 V) heme group having a sixcoordinated iron atom in low-spin state. Even if mixed culture biofilms on graphite, glassy carbon and silver display the typical His-Fe-His axial ligation in both oxidation states,^14,20,28 our data on ITO show a different axial ligation at different oxidation states. In fact, while the ν₃ of the reduced state is consistent with the His-Fe-His axial ligation, the ν₄ and the ν₁₀ of the oxidized heme at 1369 and 1635 cm⁻¹ indicate the His-Fe-Met axial ligation.³⁹ This is the first time such a ligation is observed for mixed culture biofilms. However, since CRRM spectra ascribable to the His-Fe-Met ligation were reported by others for G. sulfurreducens biofilms on ITO,¹⁶ we argue that the electrode material contributed to select for bacterial species displaying the same axial ligation as G. sulfurreducens on the same electrode. This observation reinforces the conclusions of a recent report on the impact of surface composition on the redox properties of microbial biofilms.⁴⁰

Fig. 2 RR spectra of electroactive biofilms in acetate-depleted medium (average of multiple measurements) in (a) the spectral region between 600 and 1700 cm⁻¹ and (b) magnification on the region between 1200 and 1700 cm⁻¹. RR spectra recorded with the working electrode poised at −0.6 V (blue line) and 0 V (grey line). Relative positions of the marker bands indicated in the figure are reported in Table 1.

Table 1 Normal mode assignment of the most prominent bands from the averaged Raman spectra obtained with the working electrode poised at 0 V and −0.6 V shown in Fig. 2

Band assignment^a	RR bands at 0 V (cm^−1)	RR bands at −0.6 V (cm^−1)
a Assignment accordingly to Hu et al.^52b Shoulder.
ν15	747	746
ν22	1127	1127
ν30	1167	1165
ν21	1313	1310
ν4	1369	1360
ν20	1400	1391
ν3	—	1494
ν11	1563^b	—
ν2	1583	1582
ν10	1635	1635

---

Measurements of the redox state of cytochromes across biofilm sections in the presence of metabolic substrate

The RR spectra recorded under non-turnover condition at the potentials of −0.6 V and 0 V will serve as reference spectra for the reduced and the oxidized OMCs, respectively. In fact, the spectra discussed in this study do not deviate from those shown in Fig. 2 significantly. For this reason, a qualitative assignment of the dominant redox state is possible by tracking the spectral position and intensity of redox-sensitive vibrational modes ν₁₅, ν₂₁, ν₄, ν₂₀, ν₃, and ν₁₀, without the need to adopt laborious fitting procedures. The knowledge of the redox state of OMCs in actively respiring biofilms have been subject of intense research in recent years, since it provides important information on the ratelimiting steps controlling the catalytic current generation in electroactive biofilms. According to the redox conduction model developed by Strycharz and coworkers for a system not limited by ET at the biofilm/electrode interface (that is, a system where the ET rate constant for heterogeneous electron transfer, k_het, is very large), in the presence of oxidizing electrode potential, the RR spectra of a metabolizing biofilm is expected to include the contribution of oxidized cytochromes, and the establishment of a measurable redox gradient.¹⁸ This was recently reported for G. sulfurreducens biofilms studied with confocal Raman spectroscopy.^11,16

To study possible local changes of redox state of OMCs in our mixed community biofilms, we used the confocal capabilities of our system to collect RR spectra from different focal planes within the biofilms. We performed measurements with the anode poised at the potentials of 0 V and +0.2 V, and in the presence of different levels of acetate (i.e., 1 mM, 5 mM, and 20 mM). Each spectrum was acquired sequentially every 5 μm along a 60 μm line extending in the Z-direction, using an integration time of 20 seconds. An example of redox profiling measurement in the presence of 20 mM acetate and with the anode poised at +0.2 V is included in the ESI (Fig. S4),† together with additional details on the measurement (Text S1†). Each spectrum was evaluated individually to assess the position of the redox marker bands ν₂₁, ν₄, and ν₂₀, which are ascribed to the largest spectral shifts upon changes in redox state (vide supra). Per each set of experiments, the respective spectral positions were grouped and averaged per discrete depth step (5 μm). Actual values are summarized in Tables S1 and S2 in the ESI,† while Fig. 3 shows a visual representation of the results. The figure depicts the average positions of the modes ν₂₁, ν₄, and ν₂₀ at increasing distances from the electrode surface. In order to assist the assessment of the redox state, the figure also indicates the average position of the bands ν₂₁, ν₄, and ν₂₀ recorded in non-turnover conditions with the anode poised at 0 V or −0.6 V (attributed to OMCs in completely oxidized or reduced state, respectively), thus depicting the full range of variations for these redox markers upon shifts in redox state. Results in Fig. 3a show that in the presence of non-limiting levels of metabolic substrate, that is at 5 and 20 mM sodium acetate, and with the electrode poised at 0 V, the OMCs are observed mostly in the reduced redox state. This is supported by the position of the redox markers ν₂₁, ν₄, and ν₂₀ centered at around 1311 cm⁻¹, 1360 cm⁻¹, and 1392 cm⁻¹ (Table S1†), which match closely the position of the same marker bands as recorded in the absence of metabolic substrate and with the electrode poised at −0.6 V, conditions at which the OMCs are mostly in the reduced form (see Table 1). Interestingly, the position of the marker bands is virtually identical at any distance from the electrode surface, suggesting the absence of a measurable redox gradient across the biofilms, even in proximity to the electrode. Increasing the driving force by increasing the electrode potential to +0.2 V did not resulted in appreciable changes in the redox state or in the appearance of a measurable redox gradient (Fig. 3b). In fact, on average the bands ν₂₁, ν₄, and ν₂₀ were centered at around of 1311, 1359, 1392 cm⁻¹ in 5 mM acetate, and 1311, 1360, and 1392 cm⁻¹ in 20 mM acetate (Table S2†). It was not until we performed measurements in the presence 1 mM of acetate in the bulk liquid that we noticed appreciable shifts in position of the redox marker bands consistently with changes in the redox state of the OMCs (Fig. 3). In fact, in this case the relative position of the bands ν₂₁, ν₄, and ν₂₀ is consistent with cytochromes in the oxidized state (Tables S1 and S2†). Diffusional limitations for metabolic substrate within the biofilm were likely irrelevant for acetate levels in the bulk liquid of 5 mM and 20 mM, as it was also suggested previously.⁴¹ It is worth noting that no significant increase of steady-state current followed the increase in the acetate concentration from 5 to 20 mM at both employed potentials (Fig. S6†), confirming that the whole biofilm performed at the maximal allowed conversion rates and therefore, local variations of the respiration rates due to low substrate availability can be excluded.

Fig. 3 Depth profiling under turnover conditions in media containing acetate: 1 mM (red squares), 5 mM (green circles), and 20 mM (blue diamonds). (a) Measurements performed with the working electrode poised at 0 V, and (b) at +0.2 V. The symbols depict the position of the redox markers ν₂₁, ν₄, and ν₂₀ as a function of the distance from the electrode surface. RR spectra were collected as 12 single spectrum acquisitions along a 60 μm line in the Z direction with an integration time of 20 s per point. Points were then grouped in discrete intervals of 5 μm and averaged (refer to the ESI† for details). Note that point Z = 0 μm corresponds to the biofilm/ITO interface, determined as described in the ESI.† Note that symbols for the data at 5 and 20 mM overlap in some instances. Standard deviations values are in many instances smaller than the symbols, hence, average and standard deviations are also reported in Tables S1 and S2 in the ESI.†

Analysis of electron transfer kinetics

The results presented above, specifically, the fact that the oxidation of OMCs lags significantly behind changes in electrode potential, and the absence of a measurable concentration gradient of oxidized (and reduced) OMCs across the biofilms under turnover conditions, suggests that the limiting step for the electron transfer process in the electroactive microbial community studied here may lay at the biofilm/electrode interface, contrary with what has been observed using CRRM on G. sulfurreducens films grown on gold,¹¹ and ITO electrodes.¹⁶ In these studies the presence of a measurable redox gradient suggested that the step that limits the overall electron transport is the long-range, homogeneous ET, as opposed to a very fast heterogeneous ET at the biofilm/electrode interface. Currently accepted models for electron transfer across a conductive biofilm matrix describe the catalytic acetate oxidation as the combination of numerous processes, each accounting for a particular electron transfer step.^18,42 Steps include (1) the mass transport of acetate into the microbial cells, (2) its internal turnover and (3) the reduction of ET mediators (identified here as OMCs) (steps (1) to (3) in Strycharz et al.¹⁸); followed by (4) homogeneous electron transfer through the biofilm matrix via a sequence of ET reactions between fixed mediators (e.g., electron hopping), and (5) the final heterogeneous electron transfer to the electrode by the oxidation of mediators at the biofilm/electrode interface (steps (4) and (5) in Strycharz et al.¹⁸). When used to describe catalytic acetate oxidation of G. sulfurreducens biofilms, this model identified non-limiting heterogeneous electron transfer kinetics (step (5) above), while identifying the long-range electron transfer across the biofilm matrix (step (4)) and metabolic substrate conversions (steps (1) to (3)) as the main rate-limiting steps.^18,42 Conversely, results presented recently by others, suggest that differences might exist between mixed and pure culture electroactive biofilms in the identity of the rate-limiting ET steps. For example, using time-resolved surface-enhanced resonance Raman scattering (SERRS), Ly et al.⁴³ probed selectively the redox states of the OMCs at the biofilm/electrode interface, reporting that the heterogeneous ET was a very slow process (having a k_ET = 0.03 s⁻¹) that was coupled with a slightly faster long-range ET (homogeneous) having a predicted rate constant k_hom of 1.2 s⁻¹.

To provide a deeper insight into the ET rates of mixed culture biofilms, we applied a combined experimental approach consisting of monitoring the changes of oxidation states of the OMCs in time by means of CRRM during the application of a potential step from an initial (E_i) to a final potential (E_f), the latter process being probed by chronoamperometry (CA). We used the subsequent relaxation of the current profile as indicative of all redox processes occurring in the bulk biofilm and at the biofilm/electrode interface, while we used the relaxation of the intensity of the RR spectra binned at the band ν₁₅ at 750 cm⁻¹ to monitor long-range electron transfer involving homogeneous transfer between OMCs at a portion of the biofilm at a specific distance from the electrode (fixed at 5 μm from the electrode surface on all measurements). The choice of the marker band ν₁₅ at 750 cm⁻¹ was dictated by its relation to the amount of cytochromes in the reduced state. We used a similar approach previously to monitor dynamic variations of heme redox-state during combined voltammetry and RR measurements.²⁰

Fig. 4 reports typical measurements performed using a potential step with amplitude of 0.7 V from an initial E_i of −0.5 V (which was maintained for 300 s prior to the transition) to a final E_f of +0.2 V (maintained for 240 s after the transition). The E_i was intentionally chosen lower than the average E_1/2 as determined by non-turnover voltammetry and equal to −0.346 V (vide supra), to initiate the transition from a largely reduced redox state of the OMCs. The CA traces of current vs. time in the absence of acetate after the transition (t > 300 s) display an exponential relaxation phase which levels off towards zero (Fig. 4b). Conversely, in the presence of acetate the current traces display a very fast discharge process in the instants immediately after the transition (within a few seconds past t = 300 s), which is then superimposed by an electron producing process that levels off towards an average steady-state catalytic current of (41.8 ± 0.1) μA in the later stages of the measurements (Fig. 4a). Interestingly, analogous profiles were obtained also during CA tests using a final E_f of 0 V, hence a smaller driving force (Fig. S7†). Busalmen and coworkers obtained similar profiles of current vs. time on G. sulfurreducens grown on graphite electrodes under turnover conditions, albeit with different time spans.^42,44 The authors described the turnover current as due to the simultaneous contribution of two processes: (1) the reoxidation of the cytochromes reduced in the phases preceding the transition, which is responsible for the initial fast discharge, and (2) the current deriving from the metabolism of acetate, which lags behind the first process, at least for the first instants after the transition, probably due to the requirement for acetate to be transported into the microbial cells before being metabolized. Using a similar conceptual approach, we can consider the turnover current (i_turnover) as due to the combination of all electron transfer steps (i.e., steps 1 to 5 in Strycharz et al.¹⁸), while the non-turnover current (i_non-turnover) as due to the combination of homogeneous and heterogeneous ET alone because of the absence of acetate in the medium (i.e., steps (4) and (5) in Strycharz et al.¹⁸). Therefore, the shape of the difference curve i_turnover − i_non-turnover accounts for the contributions of acetate metabolism alone (Fig. 5). The shape of this amperometric trace accounts for steps (1) to (3) in Strycharz et al.,¹⁸ that is, acetate uptake into the microbial cells, its metabolic oxidation, and ET to the matrix OMCs. The analysis of this amperometric profile suggests that the discharge current prevails over the metabolic current for a very short time after the transition. At about t = 301 s – thus much earlier than the complete oxidation of the OMCs occurs – the current increases due to the microbial metabolism of acetate. The trace also displays the presence of a local maximum at about 309 s (that is 9 s after the transition). It is possible that this maximum is due to the electrons generated by the oxidation of the acetate that had already entered the microbial cells in the phases prior the transition, when the potential was more negative than that required for the catalytic acetate oxidation. After this internal pool is exhausted (at t > 309 s), however, the production rate of additional electrons will depend on the rate-limiting steps between the transport of new acetate into the cells, its subsequent oxidation, and ET to OMCs. This process is probably regulated by the internal metabolism of the microbes, since changes in gene expression are unlikely within the timeframes observed. Hence, is possible that the presence of the minimum in electric current output at about 319 s is due to the initial need by the organisms to sense the change in potential, and adjust the internal metabolic machinery to the new redox conditions as imposed by the transition to E_f. Recent findings reporting the ability of metal-respiring bacteria to sense the electric field – a property called electrokinesis – forces us to consider that this possibility occurs also in the case of our mixed cultures.⁴⁵

Fig. 4 CRRM and CA measurements for potential steps with E_i = −0.5 V to E_f = +0.2 V. (a) Current vs. time and intensity of the band ν₁₅ vs. time during the test under turnover and (b) non-turnover conditions. (c and d) RR spectra recorded during a 60 s accumulation before the transition (spectra 1 and 3) and again 180 s after the transition (spectra 2 and 4) during (c) turnover and (d) non-turnover measurements. Insets in (c) and (d) represent magnification on the redox markers region and shows relative position of the bands ν₂₁, ν₄, ν₂₀, and ν₁₀.

Fig. 5 Turnover (i_turnover) and non-turnover (i_non-turnover) current vs. time traces (averages and standard deviations of the three CA profiles reported in Fig. 4) for a potential step with E_i = −0.5 V to E_f = +0.2 V. The calculated trace i_turnover − i_non-turnover represents the current due to acetate metabolism alone (see text for details).

Examination of the CRRM spectra corroborates the interpretation provided above. First of all, analysis of the intensity of the band ν₁₅ (I_ν15) during the transition experiments (also reported in Fig. 4) shows that prior to the potential step (t < 300 s) under both turnover and non-turnover conditions, the redox state of OMCs is mostly reduced. This is inferred by the high intensity of the band ν₁₅ (Fig. 4a and b), as well as the position of the redox markers ν₂₁, ν₄ and ν₂₀ in the RR spectra collected during the 60 seconds prior the application of the transition (spectrum 1 and 3 in Fig. 4c and d), which proves that the transition starts with OMCs mostly in the reduced state. This is not surprising since OMCs can store negative charge from the intracellular metabolism under conditions at which the anode cannot act as the electron sink.^42,44,46,47 Our data show that the cell metabolism refills the OMCs before their complete discharge occurs at the electrode. In other words, steps (1–3) described above (i.e., transport of acetate into the cells, its internal turnover, and reduction of matrix OMCs) are faster than steps (4 and 5) (i.e., homogeneous and heterogeneous ET, respectively). Our CRRM data confirm this interpretation. In fact, upon the application of the oxidizing potential E_f (+0.2 V), the profile of I_ν15 vs. time under non-turnover condition relaxes towards zero very quickly, indicating a shift in the redox state of the OMCs in the biofilm region at 5 μm from the electrode surface to mostly the oxidized state. This is confirmed by shifts of the bands ν₂₁, ν₄ and ν₂₀ to 1314, 1367, 1401 cm⁻¹, respectively, as well as by the presence of the band ν₁₀ at 1634 cm⁻¹ (Fig. 4d). This observation is coherent with the complete discharge of the OMCs in the absence of acetate (vide supra). Conversely, under turnover conditions, instead of the exponential decay as displayed in the absence of acetate, the profiles of the I_ν15 vs. time show a very sharp initial drop occurring within seconds after the transition (Fig. 4a), after which, instead of relaxing to zero as observed in non-turnover, the I_ν15 rests at an intermediate value ≠ 0 for the remaining of the observations, consistent with an incomplete discharge of the OMCs. Analysis of position and intensity of the bands ν₂₁, ν₄, ν₂₀, and ν₃ (virtually unchanged after the transition), and the absence of the band ν₁₀ confirmed the redox state of the OMCs as predominantly reduced, in spite of the partial reduction of intensity of the ν₁₅ band (spectra 1 and 2 in Fig. 4c). The initial drop observed in the I_ν15 is likely ascribed to the initial discharge that prevails in the instants right after the transition over the electrons produced by the microbial metabolism, as discussed earlier. However, after this initial phase, electrons from the microbial metabolism will feed to the bulk OMCs, keeping them (mostly) in the reduced redox state. The absence of a transient phase in the I_ν15 profile similar to that observed for the profile of i_turnover (and the i_turnover − i_non-turnover) before reaching steady-state, suggests that the two processes are not perfectly coupled. Even if the reason for this is unknown, we cannot exclude the contributions of other redox mediators promoting the electron transfer and thus increasing the electrocatalytic current.

Overall, these observations are in line with a scenario whereby in the presence of acetate (and under potentiostatic conditions), electrons are transferred from the microbial central metabolism to the external OMCs at a rate that is faster than the rate at which electrons travel between adjacent OMCs across the conducting biofilm matrix. This scenario requires the rate constant for homogeneous ET k_hom to be smaller than the rate constant for the combined metabolic turnover rates k_mic (which is controlled by the rate-limiting step between acetate diffusion, acetate oxidation, and electron transfer to the OMCs), that is k_mic > k_hom. In fact, a scenario characterized by sluggish microbial kinetics combined to a fast homogeneous transfer (i.e., k_mic < k_hom) is not consistent to the presence of cytochromes mostly in the reduced redox state as we observed during the transient CA test under turnover conditions (I_ν15 trace in Fig. 4a). Under these conditions, the rate-limiting process for ET would be either the homogeneous ET or the heterogeneous ET, or the combination of both. Previous analysis by Ly et al.⁴³ suggested that the bulk ET is faster than the interfacial process (k_hom > k_het). This is consistent with our measurements (which cannot exclude, however, k_hom ≈ k_het). In fact, a scenario consisting of a sluggish ET in the biofilm and a much faster interfacial electron transfer (i.e., k_hom ≪ k_het) would not have made possible for the bulk OMCs as probed by CRRM to be completely oxidized during the CA test in non-turnover conditions, and it would have resulted in the generation of an appreciable concentration gradient of oxidized OMCs within the biofilms, consistent to CRRM observations on G. sulfurreducens.^11,16 Both our measurements at steady-state and during the transient CA experiments do not seem to support such a scenario (vide supra). CRRM measurements performed on biofilms comprised of individual cell clusters, which therefore exclude the contribution of multiple cell layers to the redox state of the OMCs in proximity of the interface, are also consistent to the proposed scenario of a sluggish heterogeneous ET (data not shown).

Analysis of the apparent electron transfer rate constant for homogeneous ET

Contrary to the method by Ly et al.⁴³ based on SERRS, our approach based on CRRM does not allow for the direct measurement of the interfacial ET process due to the physical impossibility to focus the laser beam only on the interfacial OMCs: in fact, being the thickness of the beam typically in the micrometer range, it exceeds by far the thickness of the surface-confined OMCs (<10 nm). Discerning the contribution of the homogeneous ET from the heterogeneous ET from the current vs. time trace in non-turnover conditions was also not possible, since the current traces results from the contribution of all ET processes in the whole biofilm. However, we focused here on the examination of the long-range ET process using the analysis of the relaxation profiles of the I_ν15 vs. time after the transition. In fact, the relaxation constant for the biofilm-embedded OMCs during the transient CA tests described above can be determined by fitting of a single exponential decay function to the profile of the I_ν15 vs. time. By applying different E_f, it was possible to determine the apparent k for homogeneous ET relatively to different jump amplitudes (ΔV). The fitting exercise yielded k_hom of (0.038 ± 0.004), (0.043 ± 0.021), and (0.066 ± 0.026) s⁻¹ (averages ± standard deviation for triplicate measurements) for the potential steps with E_f of −0.2 V, 0 V, and +0.2 V, respectively. Results are presented in Fig. 6. Best fitting of the data was achieved with an exponential function (coefficient of determination R² = 0.93). The theoretical k value determined by imposing the condition of zero overpotential to the fitting curve (that is, for a hypothetical potential transition where E_f = E_1/2 = 0.346 V (vide supra) is, according to the Butler–Volmer formalism, the rate constant at equilibrium, k⁰_hom. The analysis yielded a k⁰_hom of 0.028 ± 0.069 s⁻¹ (mean ± 95% confidence interval). This is 60 times lower than that predicted by Ly and coworkers under the hypothesis that long-range electron transfer is promoted by only OMCs in the bulk biofilm,⁴³ and may indicate the contribution of redox mediators other than OMCs to the long range ET. In our systems, comparison of the measured k_hom vs. ΔV with the hypothetical profile of k^BV_hom vs. ΔV as predicted by the Butler–Volmer equation for the same value of k⁰_hom shows that the observed homogeneous ET is indeed much slower than that predicted by the model (see Fig. S8 in the ESI†). While a weak dependency of ET rates with driving force is characteristic of electron transfer via a hopping mechanism⁴⁸ and it was used previously to establish the dominance of inelastic hopping over tunneling,^43,49,50 our results can be considered consistent to the simultaneous interplay of a fast ET process with a much slower ET mechanism. While the slow ET process could still be ascribed to hopping between adjacent cytochromes, alternative fast ET pathways can perhaps include tunneling between cytochromes with more favorable orientation and/or at closer proximity with each other, or the involvement of alternative ET modes including bacterial nanowires with metallic-like conducting properties as proposed for networks of G. sulfurreducens.⁵¹ It is important to note that the large signal-to-noise ratio that characterizes the real-time CRRM measurements (see for example the I_ν15 profiles in Fig. 4, 6 and S7†) may not permit to distinguish with sufficient accuracy between fast and slow ET processes involving OMCs (i.e., the I_ν15 traces account for the macroscopic redox response of the probed sample). On the other hand, the lack of additional information on the biofilm architecture and composition does not permit to be resolute on this aspect at this stage. For example, CRRM is not capable to detect the presence of cellular components such as nanowires or other potentially important redox active molecules in the biofilms examined unless they generate a detectable Raman signal.

Fig. 6 CA performed in non-turnover conditions. (a) Relaxation profiles of the intensity of the band ν₁₅ vs. time after potential transitions from E_i to E_f where E_i = −0.5 V and E_f = −0.2 V, 0 V, and +0.2 V. Values of k_hom are evaluated by fitting the I_ν15 vs. time traces with a monoexponential decay function (an example for E_f = 0 V is indicated by a red line in the figure). (b) Dependence of the apparent electron-transfer rate constants for homogeneous ET (k_hom) on the driving force (ΔE = E_f − E_i). Values are reported as averages ± standard deviations for triplicate measurements. The values of k_hom as a function of the driving force were fit with an exponential function to determine the k_hom at zero overpotential (η = 0) – that is, for an hypothetical potential transition with E_f = E_1/2 = −0.346 V – which corresponds to the rate constant at equilibrium, k⁰_hom. The fitting yielded k⁰_hom = (0.028 ± 0.069) s⁻¹ (average ± 95% confidence interval).

Conclusions

By using specifically designed spectroelectrochemical cells that utilized transparent ITO electrodes, it was possible to assess the redox chemistry of mixed culture electroactive biofilms in situ and in vivo. Our results show that in the presence of non-limiting levels of metabolic substrate, the oxidation state of OMCs embedded in the biofilm matrix lags significantly behind the electrode potential (contrary to measurements in substrate-depleted medium). Under the same conditions, the redox state of OMCs is mostly reduced at any distance from the electrode surface, even when the electrode is poised at potentials sufficiently high to promote complete oxidation of the OMCs in the absence of substrate. CRRM analysis during potential transitions under turnover and non-turnover conditions suggested that the microbial respiration rate is a fast process compared to both long-range and interfacial electron transfer. Thus, in mixed community biofilms grown on ITO, a fast metabolic acetate respiration feeds electrons to the OMCs. This process is coupled with much slower or comparable homogeneous ET in the biofilm and heterogeneous electron transfer at the interface.

Acknowledgements

This work was performed in part at the Queensland node of the Australian National Fabrication Facility, a company established under the National Collaborative Research Infrastructure Strategy to provide nano- and micro-fabrication facilities for Australia's researchers. BV, BCD, and JOK acknowledge the financial support for CEMES through The University of Queensland. BV acknowledges the support of the UQ Biomedical ECR grant. DM acknowledges the Netherlands Organisation for Scientific Research (NWO) grant 722.011.003.

References

1. S. M. Strycharz, R. M. Snider, A. Guiseppi-Elie and L. M. Tender, Energy Environ. Sci., 2011, 4, 4366–4379.
2. N. Malvankar, M. T. Tuominen and D. Lovley, Energy Environ. Sci., 2012, 5, 6247–6249.
3. S. M. Strycharz and L. M. Tender, Energy Environ. Sci., 2012, 5, 6250–6255.
4. S. Pirbadian and M. El-Naggar, Phys. Chem. Chem. Phys., 2012, 14, 13802–13808.
5. N. Malvankar, M. Vargas, K. P. Nevin, A. Franks, C. Leang, B.-C. Kim, K. Inoue, T. Mester, S. F. Covalla, J. P. Johnson, V. M. Rotello, M. T. Tuominen and D. Lovley, Nat. Nanotechnol., 2011, 6, 573–579.
6. D. R. Bond, S. M. Strycharz, L. M. Tender and C. I. Torres, ChemSusChem, 2012, 5(6), 1099–1105.
7. N. Malvankar, M. T. Tuominen and D. Lovley, Energy Environ. Sci., 2012, 5(9), 8651–8659.
8. M. Vargas, N. Malvankar, P.-L. Tremblay, C. Leang, J. A. Smith, P. Patel, O. Synoeyenbos-West, K. P. Nevin and D. Lovley, mBio, 2013, 4(2), e00105.
9. C. Leang, X. Qian, T. Mester and D. Lovley, Appl. Environ. Microbiol., 2010, 76, 4080–4084.
10. J. P. Busalmen, A. Esteve-Nunez, A. Berna and J. M. Feliu, Angew. Chem., Int. Ed. Engl., 2008, 47, 4874–4877.
11. N. Lebedev, S. M. Strycharz and L. M. Tender, ChemPhysChem, 2014, 15(2), 320–327.
12. Y. Liu and D. R. Bond, ChemSusChem, 2012, 5, 1047–1053.
13. Y. Liu, H. Kim, R. R. Franklin and D. R. Bond, ChemPhysChem, 2011, 12, 2235–2241.
14. D. Millo, F. Harnisch, S. A. Patil, H. K. Ly, U. Schröder and P. Hildebrandt, Angew. Chem., Int. Ed. Engl., 2011, 50, 2625–2627.
15. H. Richter, K. P. Nevin, H. Jia, D. A. Lowy, D. Lovley and L. M. Tender, Energy Environ. Sci., 2009, 2, 506–516.
16. L. Robuschi, J. P. Tomba, G. D. Schrott, P. S. Bonanni, P. M. Desimone and J. P. Busalmen, Angew. Chem., Int. Ed. Engl., 2013, 52, 925–928.
17. R. M. Snider, S. M. Strycharz, S. D. Tsoi, J. S. Erickson and L. M. Tender, Proc. Natl. Acad. Sci. U. S. A., 2012, 109, 15467–15472.
18. S. M. Strycharz, A. P. Malanoski, R. M. Snider, H. Yi, D. Lovley and L. M. Tender, Energy Environ. Sci., 2011, 4, 896–913.
19. S. M. Strycharz and L. M. Tender, ChemSusChem, 2012, 5, 1106–1118.
20. B. Virdis, D. Millo, B. C. Donose and D. J. Batstone, PLoS One, 2014, 9, e89918.
21. M. Estevez Canales, A. Kuzume, Z. Borjas, M. Füeg, D. Lovley, T. Wandlowski and A. Esteve-Nunez, Environ. Microbiol. Rep., 2014, 7, 219–226.
22. S. M. Strycharz, J. Roy, D. Boyd, R. Snider, J. S. Erickson and L. M. Tender, ChemElectroChem, 2014, 1, 1957–1965.
23. P. S. Bonanni, D. Massazza and J. P. Busalmen, Phys. Chem. Chem. Phys., 2013, 15, 10300–10306.
24. P. Ledezma, P. Kuntke, C. J. N. Buisman, J. Keller and S. Freguia, Trends Biotechnol., 2015, 33, 214–220.
25. B. E. Logan and K. Rabaey, Science, 2012, 337, 686–690.
26. M. T. Agler, B. A. Wrenn, S. H. Zinder and L. T. Angenent, Trends Biotechnol., 2011, 29, 70–78.
27. D. Millo, Biochem. Soc. Trans., 2012, 40, 1284–1290.
28. B. Virdis, F. Harnisch, D. J. Batstone, K. Rabaey and B. C. Donose, Energy Environ. Sci., 2012, 5, 7017–7024.
29. H. Lu, A. Oehmen, B. Virdis, J. Keller and Z. Yuan, Water Res., 2006, 40, 3838–3848.
30. V. Flexer, M. Marque, B. C. Donose, B. Virdis and J. Keller, Electrochim. Acta, 2013, 108, 566–574.
31. K. Fricke, F. Harnisch and U. Schröder, Energy Environ. Sci., 2008, 1, 144–147.
32. A. Jain, G. Gazzola, A. Panzera, M. Zanoni and E. Marsili, Electrochim. Acta, 2011, 56, 10776–10785.
33. A. Okamoto, K. Hashimoto and R. Nakamura, Bioelectrochemistry, 2012, 85, 61–65.
34. B.-S. Yeo, S. Maedler, T. Schmid, W. Zhang and R. Zenobi, J. Phys. Chem. C, 2008, 112, 4867–4873.
35. N. A. Brazhe, M. Treiman, A. R. Brazhe, N. L. Find, G. V. Maksimov and O. V. Sosnovtseva, PLoS One, 2012, 7, e41990.
36. F. Adar, J. Phys. Chem., 1978, 82, 230–234.
37. A. Kuzume, U. Zhumaev, J. Li, Y. Fu, M. Füeg, M. Estévez, Z. Borjas, T. Wandlowski and A. Esteve-Nunez, Phys. Chem. Chem. Phys., 2014, 16, 22229–22236.
38. M. Abe, T. Kitagawa and Y. Kyogoku, J. Chem. Phys., 1978, 69, 4526.
39. S. Oellerich, H. Wackerbarth and P. Hildebrandt, J. Phys. Chem. B, 2002, 106, 6566–6580.
40. K. Artyushkova, J. A. Cornejo, L. K. Ista, S. Babanova, C. Santoro, P. Atanassov and A. J. Schuler, Biointerphases, 2015, 10, 019013.
41. E. Marsili, J. B. Rollefson, D. B. Baron, R. M. Hozalski and D. R. Bond, Appl. Environ. Microbiol., 2008, 74, 7329–7337.
42. P. S. Bonanni, G. D. Schrott, L. Robuschi and J. P. Busalmen, Energy Environ. Sci., 2012, 5, 6188–6195.
43. H. K. Ly, F. Harnisch, S.-F. Hong, U. Schröder, P. Hildebrandt and D. Millo, ChemSusChem, 2013, 6, 487–492.
44. G. D. Schrott, P. S. Bonanni, L. Robuschi, A. Esteve-Nunez and J. P. Busalmen, Electrochim. Acta, 2011, 56, 10791–10795.
45. H. W. Harris, M. Y. El-Naggar, O. Bretschger, M. J. Ward, M. F. Romine, A. Y. Obraztsova and K. H. Nealson, Proc. Natl. Acad. Sci. U. S. A., 2010, 107, 326–331.
46. A. Esteve-Nunez, J. Sosnik, P. Visconti and D. Lovley, Environ. Microbiol., 2008, 10, 497–505.
47. N. Malvankar, T. Mester, M. T. Tuominen and D. Lovley, ChemPhysChem, 2012, 13, 463–468.
48. L. J. C. Jeuken, A. K. Jones, S. K. Chapman, G. Cecchini and F. A. Armstrong, J. Am. Chem. Soc., 2002, 124, 5702–5713.
49. T. Morita and S. Kimura, J. Am. Chem. Soc., 2003, 125, 8732–8733.
50. J. Hrabakova, K. Ataka, J. Heberle, P. Hildebrandt and D. H. Murgida, Phys. Chem. Chem. Phys., 2006, 8, 759–766.
51. N. Malvankar, S. E. Yalcin, M. T. Tuominen and D. Lovley, Nat. Nanotechnol., 2014, 9(12), 1012–1017.
52. S. Hu, I. K. Morris, J. P. Singh, K. M. Smith and T. G. Spiro, J. Am. Chem. Soc., 1993, 115, 12446–12458.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c5ra15676a

---