Tailored glycosylated anode surfaces: Addressing the exoelectrogen bacterial community via functional layers for microbial fuel cell applications

https://doi.org/10.1016/j.bioelechem.2020.107621 1567-5394/ 2020 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). ⇑ Corresponding authors. E-mail addresses: frederic.barriere@univ-rennes1.fr (F. Barrière), scanlae@tcd.ie (E.M. Scanlan), colavitp@tcd.ie (P.E. Colavita). Alessandro Iannaci , Adam Myles , Thomas Flinois , James A. Behan , Frédéric Barrière b,⇑, Eoin M. Scanlan a,⇑, Paula E. Colavita a,⇑


Introduction
The discovery of electroactive microorganisms by Potter in 1911 [1,2] opened new frontiers in alternative energy technologies and environmental chemistry and led to the development of Microbial Fuel Cells (MFCs). MFCs and related microbial bioelectrochemical systems, such as plant microbial fuel cells and microbial electrolysis cells, are some of the most interesting energy conversion devices with potential to facilitate the transition to a green economy. MFCs can generate electrical power using low-value bio-resources as fuels, e.g. wastewater, via bio-catalytic processes at room temperature thus transforming a waste management problem into a high-value product [3,4].
Colonisation of electrodes by suitable bacterial consortia and effective biofilm-electrode coupling are critical for the success of MFC devices and bioelectrochemical systems in general [5][6][7][8]. Therefore, approaches aimed at selecting for robust bacterial com-munities and improving biofilm-electrode coupling have been the subject of intense investigation for the effective development of bioelectrochemical technologies and their further implementation beyond the laboratory [3,9,10]. Interfacial modifications of solid electrodes are particularly interesting to address this challenge, as they offer the possibility of tailoring the electrode surface to enhance charge transfer rates while preserving the bulk mechanical/conducting properties of the solid electrode material. Recent examples of interfacial electrode modifications explored include the use of inorganic nanoparticles and nanocarbons [6,8,11], nanopatterning and nanoroughness modifications [12], ionic liquids [13] and conductive polymers [8,14].
Carbon-based materials are widely used as bioanode materials because of their high conductivity, good stability and low cost, hence a variety of strategies have been investigated to improve their performance. For instance, tuning of carbon morphology to achieve high surface area can enhance power densities; this has been ascribed to faster colonisation times and reductions in fouling, as suggested by several groups [12,[15][16][17][18]. Surface grafting with composites based on polyaniline/carbon allotropes demonstrates electron transfer promotion and better adhesion between https://doi.org/10.1016/j.bioelechem.2020.107621 1567-5394/Ó 2020 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/). biofilm and anode surface [19][20][21]. Chemical modifications of the carbon surface via oxidative treatments have also been observed to result in increased power densities [22][23][24][25][26]; improvements are generally attributed to enhanced adhesion and growth of the biofilm after introduction of hydrophilic oxidised groups, although chemical effects are often coupled with changes in morphology caused by oxidative etching [26,27]. Thermal treatment in ammonia atmosphere of graphite and carbon felts has been similarly found to improve performance [25,28].
Surface modification reactions that result in immobilisation of chemical moieties without affecting carbon morphology are of particular interest as they can be leveraged to investigate the role of interfacial functional groups on biofilm development and on the regulation of its coupling to the electrode surface. Functionalisation using aryldiazonium salts [29] offers such a possibility: aryldiazonium cations bearing a variety of functions can react under mild conditions via spontaneous or electrochemically assisted reactions from solution, yielding covalently attached moieties on carbons [30][31][32][33][34][35], as illustrated in Fig. 1 for the specific cations used in this work. Picot et al. [33] used this methodology to investigate the effect of functional groups including amine, carboxylate and triphenylphosphonium on the performance of MFC anodes. They found that the functional group net charge and hydrophilicity can affect charge transfer at the bio-electrode interface; importantly, they observed that the functional group affected the diversity in the biofilm. This group as well as others [36][37][38] have observed that anode surface pre-modification through electrografting of aryldiazonium bearing a p-amino functionality results in increased power densities, although careful control of molecular coverage is necessary to maintain fast charge transfer to the carbon electrode. Finally, a unique advantage offered by aryldiazonium chemistry is that of enabling the immobilisation of functional groups to explore specific affinity interactions with electroactive microorganisms. For instance, immobilisation of phenylboronic acid groups was used to target affinity binding interactions of these moieties with lipopolysaccharide diols in bacterial membranes, and resulted in significantly faster start-up times after the adlayer thickness was optimised [32]. More recently, diazopyridinium cations were used to investigate whether immobilised pyridine could mediate charge transfer between the electrode and the heme groups of c-type cytochromes in the biofilm, as is the case for cytochromes in solution [30].
In this work we investigate, for the first time, the application of aryldiazonium salts bearing glycan functionalities for the modulation of MFC bioanode performance. Glycans are known to regulate specific and non-specific interactions in nature and are involved in a wide range of biological functions, including control of conformational stability, cell recognition, cell adhesion, infection and biofilm formation. Past work from our group has shown that aryldiazonium cations bearing glycosides can be used to prepare functional layers of mono-and di-saccharides on carbon and other substrates [39][40][41][42][43]. Interestingly, we observed that these saccharide adlayers reduce unspecific protein adsorption as well as retention of foulants in a range of complex biomass-rich media [42][43][44]. This behaviour can be attributed to the ability of glycans to inhibit unspecific binding, similar to the glycocalix that surrounds and protects many cells. However, glycans can also display specific binding interactions with lectins which promote bacterial adhesion and attachment through recognition events at fimbriae and pili [45][46][47]. For instance, mannose, galactose, fucose, glucosamine and sialic acid units are all known to signal bacterial adhesion and promote surface colonisation. This suggests that immobilisation of specific glycans at carbon electrodes might offer new safe and sustainable routes for regulating colonisation of selected bacteria at bioanodes, ultimately influencing the composition of the catalytic biofilm community obtained at the electrode interface from a mixed inoculum.
Electrode materials were modified via electrografting of aryldiazonium bearing a p-mannoside functionality; covalent attachment was achieved via spontaneous reaction and via electrochemically assisted grafting using potential steps of different durations. We first discuss the effect of the different functionalisation protocols on wettability, roughness and electrochemical response of modified surfaces. Then we evaluate the performance of graphite rod electrodes to determine the effect of phenylmannoside adlayers on bioanode development, power output and overall electrochemical performance in an operational MFC.

Substrate preparation and functionalization
Prior to surface modifications and electrochemical studies GC electrodes were polished with progressively finer grades of alumina slurry (Buehler) according to published protocols [48][49][50], whereas GR electrodes were polished with sand paper (1200 grit) and subsequently sonicated in deionised water, acetone and methanol (5 min in each solvent). Surface modifications were carried out in a degassed 1.0 mM solution of the aryldiazonium salt prepared in situ from its p-aminophenyl precursor (Fig. 1). Briefly, 10.0 mL of 10.0 mM NaNO 2 were added dropwise to a solution of 14 mg of 4-aminophenol-a-D-mannopyranose precursor in 40 mL of 1.8 mM HCl in an ice bath, yielding a 1.0 mM aqueous solution of the diazonium product. Two different grafting methods were explored, as in Fig. 1: spontaneous reaction with carbon surfaces (Method 1) and electrochemically assisted grafting via potential step (Method 2). In Method 1, substrates were immersed in the cation solution for 2 h at room temperature in the dark [42,51]. In Method 2, substrates were used as the working electrode in a 3electrode cell (see Fig. 1) and a potential of À0.7 V vs SCE was applied for 35 s or 350 s [52]. After modification GC electrodes were rinsed as in previous work [41], while GR and ITO samples were rinsed in deionized water prior to further use. GC disks (Ø 0.50 cm) had an area defined by their geometry (0.196 cm 2 ) while the area of GR and ITO electrodes was defined at 2.2 cm 2 and 0.91 cm 2 , respectively, using a teflon tape to enable current and power density normalizations.

Characterization
Electrochemical characterization and potential step functionalizations were carried out using a potentiostat (Metrohm Autolab AUT50324) and a 3-electrode cell with a reference electrode, a graphite rod as counter electrode, and the substrate under study (GC, ITO or GR) as the working electrode. Water contact angle (WCA) measurements were performed using the sessile drop method (FTA1000). Atomic force microscopy (AFM, Asylum Research) was carried out using Au-coated cantilevers (obtained from NT-MDT Spectrum Instruments) in tapping mode. All errors on numerical values are calculated as standard deviations for n = 3, unless otherwise noted. Scanning electron microscopy (SEM) images were acquired on a Zeiss Ultra Plus field emission scanning electron microscope. Biofilm SEM imaging was performed following the protocol reported by Picot et al. [33]: briefly, the anodes were cut to lengths of approx. 0.5 cm using pliers; subsequently, the samples were collected and fixed overnight with a solution made of 0.1 M phosphate buffer and 2.5% vol. of glutaraldehyde; finally, they were washed with phosphate buffer and immersed in aqueous solutions containing progressively increasing ethanol concentrations (60, 70, 80, 90 and 100% ethanol), for 15 min in each solution. After this procedure the samples were critical point dried to replace the remaining water with CO 2 and they were Au-coated with a sputtering time of 20 s before SEM analysis.

Microbial fuel cell (MFC) studies
Double chamber MFC devices were fabricated from polycarbonate with ca. 85 mL capacity in each compartment. A circular membrane port Ø1.8 cm connected the anodic and cathodic compartments which were sealed with independent covers, each equipped with 3 testing ports for electrode mounting and characterization. GR electrodes were fitted and suspended in the two compartments through rubber stoppers. The anodic solution consisted of 50/50 vol% wastewater (Beaurade Wastewater Treatment Plant, Rennes, France [53]) and phosphate buffer saline (PBS) solution consisting of 0.032 M Na 2 HPO 4 , 0.018 M NaH 2 PO 4 , 6.0 mM NH 4 Cl and 2.0 mM KCl [54]. Sodium acetate was added to the resulting mixture of PBS and wastewater, resulting in a final sodium acetate concentration of 0.012 M in the final solution. The cathodic solution consisted of 0.1 M K 3 [Fe(CN) 6 ] in 0.032 M Na 2 HPO 4 and 0.018 M NaH 2 PO 4 ; the hexacyanoferrate (III) served as the electron acceptor and the 0.1 M concentration was confirmed to ensure a stable cathodic potential over the duration of MFC experiments [33]. After setup, the anode compartment was left under anaerobic conditions; MFCs were connected to 1000 X load and kept in a thermostated bath at 25.0 ± 0.1°C over the duration of experiments. To ensure nutrient availability, sodium acetate (equivalent to an additional 0.012 M concentration) was added to the anodic compartment of all cells after 14 days of operation as discussed in the main text.
Power density curves and cell potentials were recorded over time to monitor MFC performance. Each MFC was connected to a decade resistance box (Elenco) and MFC output potentials were acquired with a multimeter (Keithley 2700) connected to a PC, as previously described [51,[55][56][57]. Power density curves were obtained by first connecting each MFC to a 4 MX resistor for 2 h to acclimate the cells to the open circuit condition; then the resistance was progressively decreased in the range 10000-10 X at 30 min steps, while the output potential was recorded by the multimeter [55,56]. Cyclic voltammetry (CV) of MFC anodes was obtained in a 3-electrode configuration at 5 mV s À1 : the GR anode was used as working electrode, a saturated calomel reference electrode (SCE, IJ Cambria) was inserted in the anodic compartment through a testing port and the GR cathode was used as counter electrode. Two cycles were recorded; the second cycle of each measurement is shown in all cases.

Mannoside immobilization via spontaneous and electrochemical grafting
Aryl-mannoside layers were immobilized on electrode surfaces using either spontaneous covalent attachment or electrochemically assisted grafting from aqueous solutions. The aryldiazonium salt derivative of 4-aminophenol-a-D-mannopyranose was prepared via a diazotization reaction as shown in Fig. 1; the covalent attachment was carried out in situ in the diazotization solution using a 1.0 mM concentration of the aminophenol precursor in all cases. The characterization of aryl-glycoside adlayers prepared via aryldiazonium cation reactions at carbon and at a variety of other substrates has been reported in detail in previous work from our group [39][40][41][42][43]58,59]. Briefly, a combination of microscopy and spectroscopic methods provides evidence that aryl-glycosides can be immobilized via either spontaneous or electrochemically assisted methods at carbon. Binding assays using fluorescently labeled lectins also demonstrate that grafted aryl-glycosides present at the surface can be specifically recognised by the corresponding lectins. Spontaneous reaction leads to the formation of homogeneous sub-nm adlayers (0.5-0.9 nm), while electrochemical grafting enables the formation of denser adlayers that remain however of average thickness <2 nm [58,59] on smooth substrates.
To investigate the effects of density and morphology of the arylmannoside adlayer on bioanodes, three grafting protocols were explored: spontaneous attachment (2 h) or À0.7 V vs SCE applied potential for 35 or 350 s. Three different substrates were used to characterise the resulting glycan adlayers, as illustrated in Scheme 1: glassy carbon (GC) and indium tin oxide (ITO) electrodes were used to probe the effects of grafting protocol on well characterized electrode substrates, whereas graphite rods (GR) were used for all microbial fuel cell experiments as discussed in later sections. , in agreement with previous findings obtained using a lactoside analog of the precursor in Fig. 1 [59].
Potential steps of 35 and 350 s in the 1.0 mM diazoniation solution resulted in electrografted GC electrodes, indicated as GC35 and GC350, respectively. These surfaces were found to display greater passivating properties than those obtained via spontaneous reaction. This is evidenced in Fig. 2a by DE p values larger than 100 mV for both GC35 and GC350 that increase rapidly with scan rate (see Supporting Information), and by a significant depression in the current peaks clearly observed for GC350. This is diagnostic of an additional impedance to charge transfer characteristic of the presence of a passive film [49,60]. The total integrated cathodic charge densities associated with the generation of aryl radicals during the potential step experiments were 26 ± 5 and 98 ± 30 C m À2 for GC35 and GC350, respectively (see Supporting Information). These values are well in excess of the total charge density required for electrografting an aryldiazonium monolayer (ca. 10 À6 -10 À5 mol m À2 equivalent to 0.1-1.0 C m À2 ) at a topographically smooth surface [39,41]. Therefore, CV results indicate that GC35 and GC350 samples are likely to display denser arylmannoside adlayers than those obtained spontaneously, in agreement with recent reports using aryl-lactoside precursors [59]. The trends in CV waveforms also suggest that denser films are obtained after electrografting for the longer potential step, as longer grafting times result in enhanced current depression and increase of DE p values of Fe(CN) 6 À3/À4 peaks. experiments were carried out in this case using ITO substrates because of their smoother topography (rms roughness 3.01 ± 0.0 3 nm; see Supporting Information) relative to that of polished GC [48] or GR (see Supporting Information). AFM imaging shows that the functionalization method significantly affects the morphology  of the aryl-mannoside adlayer. The spontaneously modified surface displays an rms roughness of 1.44 ± 0.08 nm, a value lower than that of the bare substrate (see Supporting Information). These results indicate that spontaneous grafting yields conformal saccharide layers that are smoother than the original substrate, in agreement with previous findings using nitrobenzene diazonium salts [51]. Electrografting for 35 and 350 s results in a significant increase in the rms roughness to 4.3 ± 0.2 nm and 3.7 ± 0.3 nm, respectively, that is indicative of the presence of disordered glycan adlayers [61][62][63].
Microbial fuel cells were assembled using graphite rods (GR) as electrodes in both the anodic and cathodic compartments, as these carbon materials are among the most popular for MFC anodes due to their low resistivity, widespread availability and low cost [64]. The properties of aryl-mannoside adlayers were therefore also studied on these carbon substrates. Fig. 3a shows a CV recorded at 10 mV s À1 in 1.0 mM K 4 [Fe(CN) 6 ] solution in 0.5 M KCl of a bare GR and of mannoside-modified GR via 35 s (GR35) or 350 s (GR350) electrografting at À0.7 V. After electrografting, negligible passivation to Fe(CN) 6 À3/À4 charge transfer is observed at GR electrodes, as evidenced by the presence of quasi-reversible peaks with similar DE p compared to that on bare GR. This indicates that the adlayers formed on GR substrates are likely to be sparse and defective, despite application of identical electrografting protocols as for GC. The total integrated cathodic charge densities associated with the generation of aryl radicals during the potential steps were 64 ± 14 and 447 ± 124 C m À2 for GR35 and GR350, respectively (see also Supporting Information). Despite the difference in total cathodic charge density, the two electrografted GR electrodes result in very similar electrochemical passivation. Interestingly, the values of integrated cathodic charge density are significantly larger than those observed in the case of GC, thus indicating that the near reversible response at GR cannot be attributed to a lower concentration of reactive radical species during grafting relative to GC. Therefore, it appears more likely that differences arise instead from a larger specific surface area available at GR vs GC surfaces, a hypothesis supported by roughness determinations of polished GR (see Supporting Information) and significantly larger current densities than for GC electrodes. Aryldiazonium modification with glycan derivatives has been previously shown to result in a significant increase in surface free energy and hydrophilic character [58]. Given that wettability is typically highlighted as a factor controlling the activity of bioanodes [34,65], we examined the effect of covalent attachment of mannoside groups at the GR surface using water contact angle (WCA) measurements. WCA results are shown in Fig. 3b and in Supporting Information. The polished GR surface yields a value of 94°± 5°, close to that reported for pyrocarbon and pyrolyzed resist carbon films [66] but lower relative to typical carbon materials used in MFC studies, such as carbon papers [67,68], carbon cloths [69,70] and felts [71] (WCA greater than 100°). Grafting of arylmannosides results in a decrease in water contact angle to 87°± 3°, 67°± 6°and 70°± 9°for spontaneously grafted, 35 s electrografted and 350 s electrografted adlayers, respectively. This shows that spontaneous grafting results in a slight increase in hydrophilic character whereas a very significant increase in hydrophilic character is observed at the GR surface after electrografting during 35 or 350 s. This supports the conclusion that electrografting yields denser adlayers compared to spontaneous grafting despite the level of electrochemical passivation being similar on GR after both of these treatments (spontaneous or electrografting modification, see Fig. 3a).

Microbial fuel cell studies of glycan modified anodes
The performance of GR electrodes as bioanodes was evaluated in double-chamber MFC devices; all cells possessed identical geometry and are shown in Supporting Information. GR electrodes that had undergone one of the three treatments were tested: conventional polishing (GR), spontaneous grafting with arylmannosides (GR_S), or electrografting with aryl-mannosides for 350 s (GR350). During the initial start-up, the cells were all inoculated with 50/50 vol% wastewater/PBS containing 0.012 M sodium acetate, following previous protocols. [54] All cells were kept in a thermostated bath (25°C) over the duration of experiments with a 1000 X load connected to the leads when not under testing.
The potential of the MFC was acquired over time by connecting the leads to a 1000 X resistance [72][73][74]. Fig. 4 shows the evolution of the potential over time; after 10 days all cells equipped with aryl-mannoside modified GR anodes develop significantly higher potentials than cells equipped with bare anodes, thus indicating faster start-up time of the electroactive anodic biofilm. This was further confirmed via power density curves obtained by varying the cell load, shown in Fig. 5a and b at day 12 and 17, respectively; characteristic values of power density are also summarized in Table 1. The power density curves show that MFCs prepared with aryl-mannoside adlayers at the anodes deliver a maximum power density at day 12 that is 20-80 times larger than that of MFCs with bare GR. This difference can be confirmed to result from a faster rate of biofilm development on aryl-mannoside adlayers and not from a failure of colonization, given that all cells eventually converge to similar maximum power outputs at day 17.
The activity of the anodic biofilms was further investigated via CV using the anode as the working electrode in a three-electrode configuration. Fig. 6a shows examples of CVs obtained for a bare GR, GR_S and GR350 anodes at 11 days of operation; the CVs of all anodes tested are reported in Supporting Information. The CVs of anodes modified with aryl-mannoside groups show the characteristic waveform of the biocatalyzed oxidation of acetate at a mature biofilm, with an onset potential at ca. À0.45 V vs SCE [30,33,75]. The anodes that consist of bare GR show lower currents and an absence of biocatalytic activity suggests that the bioanode is still under development [76]. The small redox peaks around À0.1 V to À0.3 V vs SCE are tentatively assigned to electroactive pioneer species not fully capable of catalytic activity towards acetate oxidation [76][77][78]. After 14 days of operation, a drop was observed in the current densities of CVs for almost all anodes (Fig. 6b); small redox peaks due to cross-contamination from the catholyte were also observed at ca. 0.1 V. The anodic compartment was therefore replenished with sodium acetate (addition of 0.012 M equivalent) to continue biofilm growth. CVs in Fig. 6c obtained at day 18 show a clear onset for acetate oxidation at ca. À0.45 V vs SCE and high anodic current density for bare GR anodes, thus confirming that eventually all GR electrodes develop biocatalytic activity. This is also evident from maximum power densities reported in Table 1, which indicate that after 18 days all bioanodes display comparable performance. Biofilm morphology obtained at day 18 over the different bioanodes was then evaluated with SEM. Fig. 7 shows SEM images of three representative samples: GR#1, GR_350#1, and GR_S#1. The biofilms are homogeneous over all three considered electrodes; a mature biofilm is observed over the three different surfaces, in agreement with the comparable polarization performances obtained at day 17 (Fig. 5b, Table 1). From the SEM analysis of the three samples it is also possible to conclude that the grafted phenyl-mannoside layer is compatible with electroactive bacteria due to the presence of similar biofilm over the surface of the considered electrodes.

Discussion
Graphite rod electrodes were modified with phenyl-mannoside adlayers using two different protocols and were subsequently used as MFC anodes to test the effect of the glycan functional layer on MFC performance. Modification with mannoside groups was found to offer a facile method for regulating the wettability of anode materials. Despite observing differences in the WCA at graphite rods between spontaneously and electrochemically modified electrodes, no changes in the impedance to charge transfer were observed when using passivation tests with the ferri/ferrocyanide redox couple. This indicates that aryldiazonium reactions can be used to regulate wettability of graphite anodes without a tradeoff in electrochemical response.
The presence of phenyl-mannosides is clearly beneficial for accelerating cell start-up. This is evident from the rapid rise of cell potential under charge transfer conditions (1 kO load) in the case of mannoside-modified anodes after only 5 days from inoculation with wastewater, independently of modification protocol. By contrast, bare graphite rods do not show significant cell potentials until day 14. Further evidence of faster start-up emerges from power density curves which confirm that bioanode evolution is accelerated for modified graphite electrodes: the maximum power density on functionalized bioanodes was in fact found to be 20-80 times larger for mannoside-modified materials relative to bare graphite rods at day 11. These differences cannot be ascribed to failed colonization of control anodes, as eventually all MFCs converge to similar power density outputs at day 17 and to similar biofilm den-   sities as confirmed via SEM imaging of bioanodes. These results therefore indicate that modification with glycans can be leveraged to enhance the rate of biofilm colonization from a mixed inoculum without any adverse effects on final MFC outputs. Cyclic voltammetry of bioanodes shows that the power density increases at day 11 are due to the presence of a mature biofilm. The bio-catalytic wave observed at À0.45 V vs. SCE is diagnostic of electrode coupling to outer membrane cytochromes [33,79,80]. This biocatalytic wave is absent at bare graphite electrodes but eventually becomes evident at day 18, thus indicating that biofilm maturation is slower on the unmodified electrode surfaces.
It is interesting to note that differences in aryldiazonium grafting methods do not translate into significant differences in the maximum power densities observed for the MFCs, i.e. the best performances with spontaneously and electrochemically grafted mannosides are 397 and 347 mW m À2 , respectively. This is in contrast with previous observations on the effect of aryldiazonium electrografting charge reported by Picot et al. [33] using a range of chemical functionalities and could be explained by the lower temperature used in that work and with the thinner coatings obtained in this work with the saccharide functionalized aryldiazonium [33,41,58]. In the case of the MFC anodes in our work, we were not able to observe significant differences in passivation between spontaneously and electrochemically modified electrodes despite electrografting typically resulting in much denser films. The lack of passivation is likely due to the significant specific surface area and roughness of polished graphite rods (see Supporting Information) which results in a high density of sites available for grafting. However, we observed noticeable differences in the replicability of the performance at day 11 between the two grafting methods: at day 11, the two electrografted bioanodes display a relative variability of only 25% in maximum power density, whereas spontaneously grafted anodes yield a 40% variability in peak values. This suggests that spontaneous grafting is more likely to result in performances intermediate to those of bare and electrografted graphite anodes. We propose that the higher glycan densities achievable with electrografted methods are beneficial in presenting homogeneous mannoside adlayers at these highly rough electrode surfaces, in agreement with trends of increasing wettability observed via WCA measurements.
Finally, it is interesting to consider the origin of faster start-up times observed for mannoside-modified graphite anodes. Bioanode development has been shown to be affected by changes in surface charge, morphology/porosity and hydrophilicity in the anode car- bon material resulting from surface modifications. Prior work from our group demonstrated that, unlike oxidative treatments [81], glycoside immobilization via aryldiazonium chemistry does not result in changes in the surface electrostatic charge [58]. Therefore, attractive/repulsive interactions are not expected to be significantly different for biofilm/mannoside-electrode vs. biofilm/unmodified graphite interfaces. Also, the increase in surface roughness resulting from aryldiazonium electrografting is negligible compared to the original roughness of polished graphite rod surfaces; therefore, changes in electrode morphology can be excluded as a factor in accelerating bioanode development. Mannoside-immobilization introduces polyhydroxylated moieties that increase the hydrophilicity, which has been proposed to correlate with faster start up times [67,82]. The polished graphite rods display a WCA value that is slightly above the limit of the conventional hydrophobic/hydrophilic threshold (90°) [83], and are hydrophilic relative to the most common MFC anode materials. Spontaneous modification with mannosides results in a minor reduction in water contact angle (from 94°to 87°), while development of a mature biofilm is substantially accelerated. Although it is not possible to completely exclude that the increase in hydrophilic character contributes to faster biofilm development, the change in WCA appears small (<10°) compared to those reported in the MFC literature to affect bioanode performance. Therefore, it appears likely that the functional role of glycans in biofilm development might also contribute to the observed enhancements. Cell agglutination and surface adhesion can be favoured via lectincarbohydrate binding interactions; these interactions are known to play a role in bacterial adhesion, biofilm formation and surface colonization [84]. It is known that bacteria such as E. Coli show specific binding to glycosides, in the case of mannosides it is mediated by the FimH adhesin [63,64]. Therefore it appears likely that these bio-functional phenyl-mannoside adlayers might contribute to the regulation of bioanode development through signaling pathways that directly facilitate adhesion of exoelectrogeneous species, or that condition the surface with bacterial communities that promote adhesion of bioelectroactive species at a later stage. Further studies into the role of specific, surface-bound glycans in modulating bioanode development are ongoing in our labs.

Conclusions
Surface functionalization of graphite anode materials with mannoside adlyers was found to result in accelerated start-up of microbial fuel cells. Modified graphite materials were found to develop a catalytic biofilm faster than control bare graphite surfaces. Modified bioanodes yielded the characteristic catalytic waves of acetate oxidation in cyclic voltammograms at an estimated 40% faster rate than bare anodes. Similarly, the power density performances developed faster and reached their limiting values earlier under working operative conditions at glycanmodified anodes.
The effects of surface modification protocol and final glycan density were also investigated. Cell anodes were obtained by functionalizing MFC anodes with aryl-mannoside adlayers using two different grafting strategies: spontaneous and electrochemically driven grafting. Even if both of these strategies result in high power outputs we observed noticeable differences in the replicability of the performance at day 12 between the two grafting methods: the two electrografted bioanodes display lower variability in power density peaks than the spontaneously grafted anodes. We propose that the higher glycan densities achieved via electrografting are beneficial in presenting homogeneous mannoside adlayers at these highly rough electrode surfaces, in agreement with trends of increasing wettability observed via WCA measurements.
Our results demonstrate that functionalization of carbon electrodes with glycan adlyers via aryldiazonium reactions offers a method for enhancing the development of electroactive microbial biofilms. The enhancement in the MFC start-up times is likely to arise from a combination of increased wettability and binding affinity towards bacterial receptors in the biofilm. Cell agglutination and surface adhesion might be mediated via lectincarbohydrate interactions which are known to play a role in cell adhesion, biofilm formation and surface colonization. Further enhancements in MFC performance might be possible through immobilization of biomimetic oligosaccharide moieties that can elicit specific cell behaviour at the anode interface.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.