The ME-FBR was designed and constructed as previously reported (Tejedor-Sanz et al., 2017a,b, 2018). The tubular ME-FBR had a total volume of 1.2 L, including the recirculation pipe and the electroconductive bed volume. The working electrode (anode) consisted of a bed composed of electroconductive vitreous carbon particles (20% V_{electroconductive material}/V_{net reactor volume}) with a diameter range of 0.6–1 mm (Chemviron Carbon ^®, Belgium). A graphite plate (20 × 80 mm) was vertically immersed into the electroconductive bed material, acting as the current collector.

A Ti/Pt mesh (30 × 40 mm) was used as cathode material (Inagasa, Spain) at the top of the ME-FBR reactor.

The ME-FBR was operated as a three-electrode electrochemical cell. The fluidized bed potential was fixed to different values. A potentiostat (Nanoelectra NEV3, Spain) was used for poisoning the working electrode from −200 to + 600 mV [all anode potentials were reported vs. Ag/AgCl electrode (HANNA)]. Additionally, an electrochemical control test was performed under abiotic conditions. The data logger installed in the NEV3 potentiostat registered the output current during wastewater treatment. The current density values for the ME-FBR are given as current per net reactor volume (NRV). Cyclic voltammograms were performed at a scan rate of 0.005 V s^–1.

Finally, two control tests (see Supplementary Figures 1, 3B) were performed in reactors with the same architecture and volume as the ME-FBR described above. The first one was an electrochemical control tested under abiotic conditions, while the second control was an electrode-free anaerobic bioreactor (UASB).

Granular anaerobic sludge from a wastewater treatment plant (Guadalajara, Spain) was used as the initial inoculum. This anaerobic sludge was diluted 1:3 with synthetic wastewater. The cell suspension was incubated in the bioelectrochemical reactors for 4 days, without aeration, to encourage the growth of the anaerobic culture. During such inoculation action, no pollutants were supplied to the reactors. After this period, pollutant-spiked wastewater was fed into the reactors.

The synthetic wastewater contained 2.5 g L^–1 NaHCO₃, 0.75 g L^–1 NH₄Cl, 0.60 g L^–1 NaH₂PO₄, 0.1 g L^–1, KCl 4.1 g L^–1 NaC₂H₃O₂, and 0.1 ml L^–1 mineral stock solution and a vitamin stock solution as previously described in previous work (Tejedor-Sanz et al., 2016). The synthetic wastewater was enriched with two pharmaceutical compounds: CBZ (Sigma Aldrich, CAS 298-46-4) and SMX (Sigma Aldrich, CAS 723-46-6). The standard synthetic wastewater contained 10 and 3 mg L^–1 of CBZ and SMX, respectively. Nevertheless, to evaluate the biodegradation capacity from the initial inoculum, different solutions of CBZ and SMX were used.

The ME-FBR was fed with synthetic wastewater once a day, so its operation mode can be considered semi-continuous with 3.2 days of hydraulic retention time (HRT). The other parameters, temperature (37°C) and N₂:CO₂ (80:20) headspace gas, were kept constant. The electrolyte recirculation velocity used during the experiments was 0.68 cm s^–1. Effluent samples were taken daily.

Samples were filtered with 0.45-μm nylon membranes prior to storage at −20°C. TOC was analyzed in both feed stream and effluents after 1:10 dilution using a TOC-VCSH Shimadzu analyzer. Acetate concentration was measured by an HP series 1,100 high-pressure liquid chromatograph coupled with a UV detector (210 nm), equipped with a Supelco C-610H column and using 0.1% H₃PO₄ as the mobile phase with a flow rate of 0.5 ml min^–1. In addition, pH and conductivity were measured daily by a HACH HQ40D ^® multiparametric probe to control the technical performance of the anaerobic systems. CBZ and SMX concentrations were measured using an HP series 1,100 high-pressure liquid chromatograph equipped with an automation injection, a diode array detector (280 nm), and a Kromasil column (150 × 4.6 × 1.5 μm).

Measurement of the luminescent decay of the bacteria Vibrio fischeri NRRL-B 11177 was used to test synthetic wastewater and treated effluent toxicity. Bioassays were carried out using lyophilized bacteria after their reactivation following instruction from BioFix ^® Lumi (Macherey-Nagel, Germany). Tests were conducted at 15°C with minimal modification of the standard ISO 11348-3:2007 to measure light emission in 96-well microplates, using Fluoroskan Ascent FL (Thermo Fisher Scientific).

The impact of pharmaceuticals on the bioluminescence emission of bacteria was measured by triplicate as the inhibition percentage regarding the light emission of the control test. Toxicity values were calculated after 28 min of exposure time.

The adsorption capacity from a fluid-like electrode to retain CBZ and SMX was assessed by continuously feeding under gravity 50 ml of synthetic wastewater through 10 g of vitreous carbon until the final concentration of the pharmaceutical compounds was stable for three consecutive cycles. Finally, 10 additional feeding steps were performed. Additionally, it is also important to remark the operation of an electrochemical abiotic control test (described in section “Design and Construction of Bioelectrochemical Reactors: Microbial Electrochemical Fluidized Bed Reactor”), which also provided adsorption data under electrode polarization.

The anaerobic removal of CBZ and SMX by our microbial consortium was evaluated during 120 h in batch mode. During this assay, electrochemistry was not involved in order to evaluate the capacity of the anaerobic inoculum to remove CBZ and SMX (Figure 1). Our results revealed that CBZ was partially removed by the anaerobic consortium regardless of the concentration tested (4.09 and 1.88 mg L^–1 of CBZ). On the contrary, anaerobic bacteria were inefficient in biodegrading SMX at low concentrations (1.14 mg L^–1 of SMX), showing a minor SMX removal capacity at high concentrations (4.73 mg L^–1 of SMX).

In order to assess the biodegradation capacity of the microbial consortium, the synthetic wastewater was spiked with 50 mM acetate as the sole electron donor in addition to either CBZ or SMX.

Two different stages could be identified during the acetate and pollutant removal. During the first 48 h, anaerobic bacteria simultaneously consumed both the acetate and the pharmaceutical compounds. The removal rates for CBZ were higher (∼1.3 mg L^–1 day^–1 of CBZ) when the pollutant was given at the highest doses (∼4 mg L^–1). The acetate removal during both experiments was ∼20% regardless of the doses of CBZ. After the first stage, no further removal of CBZ or acetate was observed, suggesting a stationary phase.

On the contrary, in the case of SMX-polluted wastewater, a different pattern was observed. A lower SMX removal was achieved compared to CBZ oxidation. However, cells were metabolically active since a higher acetate removal was observed. Thus, wastewater spiked with 4.7 mg L^–1 of SMX showed a removal of 0.74 mg L^–1 day^–1 of SMX in contrast to a null removal after supplying wastewater with ∼1.1 mg L^–1 SMX.

Low removal of both CBZ and SMX at ∼1 mg L^–1 pollutant concentration could be related to mass transfer limitations. Nevertheless, CBZ and SMX may not have been entirely mineralized due to the low capacity of conventional anaerobic treatments, as anaerobic digesters, to remove these complex organic molecules (Ali, 2019; Angeles et al., 2019; Mestre and Carvalho, 2019; Pun et al., 2019). CBZ and SMX were expected to be partially oxidized to form intermediates that may not further be converted into methane and carbon dioxide using this conventional anaerobic condition. Usually, intermediates from pharmaceuticals can be more toxic than the initial organic micropollutants, so effective treatments should be performed to reach complete mineralization of pharmaceutical chemicals (Rodrigo et al., 2010; Song et al., 2020). According to previous results, higher metabolic oxidation rates should be achieved in order to remove CBZ and SMX completely.

Once we demonstrated the natural capacity of our consortium to remove CBZ and SMX, then we proceed to grow such microbial population using an ME-FBR. In order to discard the role of our electroconductive material for retaining CBZ and SMX, we performed a number of assays (Supplementary Figure 1) that showed a minor impact (9.42 ± 0.09 and 7.92 ± 0.05%, respectively). The retention tests were performed in triplicate. In parallel, we designed two key control reactors to (a) evaluate the impact of our material to remove CBZ and SMX under electrochemical operation in the absence of microorganisms and (b) evaluate the biodegradation capacity of our microbial consortium (UASB) in the absence of fluidized electrode. Indeed, CBZ and SMX pollutants were removed just in the range of 14–16% after the abiotic electrochemical control (see Supplementary Figure 3B). Regarding the electrode-free control, our UASB was able to remove just 28% for CBZ and 36% for SMX (Figure 2). In contrast, the full operation of our microbial electrochemical reactor ME-FBR under the semi-continuous mode for ∼30 days outperformed the UASB wastewater treatment capacity in terms of acetate, TOC, CBZ, and SMX removal.

Precisely, the ME-FBR validation was performed under different anode potentials to monitor the removal of acetate, CBZ, and SMX. An optimal removal of all pollutants including acetate, CBZ (−r CBZ = 2.75 mg L^–1 day^–1 of CBZ; 88% CBZ removal), and SMX (−r SMX = 0.79 mg L^–1 day^–1 of SMX; 85% SMX removal) was observed when the anode was acting as the electron acceptor at + 200 mV (vs. Ag/AgCl).

Furthermore, the removal was kept in the range of 70–80% at anode potentials as high as + 400 mV. At higher anode polarization, + 600 mV (vs. Ag/AgCl), the TOC removal decreased to 63%, while CBZ and SMX showed a slight decrease compared to the operation at lower anode potentials. The reduced electrobioremediation efficiency at potentials as high as + 600 mV has been previously reported and is most likely due to an alteration in the permeability of the bacterial membrane (Rodrigo Quejigo et al., 2019), resulting in lower microbial viability.

Finally, CBZ and SMX removal was evaluated under an anode potential with a negative value (−200 mV). Under potentiostatic control at negative potential value, organic matter and pharmaceutical compound removal was not favored according to the low TOC removal (63%), including just 50% removal of an easily biodegradable compound like acetate. Based on our results, the optimization of the anode potential is key in order to maximize pollutant removal.

During the whole ME-FBR operation, the current density was continuously monitored. It can be observed that at + 400 mV (vs. Ag/AgCl), some current data are missing. During those days, the potentiostat described in section “Materials and Methods” (NEV3) was replaced by a conventional power supply due to operational problems in order to continue polarizing the fluid-like anode. The conventional power supply operated in that period did not include a data logger. A cell-free abiotic assay revealed negligible current density in contrast to values as high as 20 A m^–3 obtained when electroactive bacteria coupled the oxidation of pollutants and electron transfer to the electroconductive particles. Higher current densities were obtained under the most positive anode, while it was minimized at values as low as −200 mV (Figure 3).

The electrochemical operation of our ME-FBR requires low energy consumption, approximately 2.3 ± 0.4 Wh according to the chronoamperometric figures observed in Figure 3. Additionally, our systems share all the OPEX advantage of any anaerobic treatment performed with our reduced sludge production and with no need of artificial aeration.

An irregular chronoamperometric profile was obtained during the semi-continuous operation of ME-FBR. Current density typically achieved maximum values when TOC was available for electroactive bacteria, while it reached minimum values when TOC was biologically consumed.

The polarization potential of the fluid-like electrode drove the metabolic activity of the biofilm. In order to study the electrochemical interaction between the microbial community and the electrode, cyclic voltammograms were performed after operating the system under different anode potentials for at least 1 week (Figure 4).

The cyclic voltammograms were performed under catalytic conditions (turnover), and they showed oxidation–reduction peaks that were not present in the cell-free abiotic assays.

At negative potential (−200 mV), the voltammogram showed no peaks, while the electric capacity was significantly less compared to the voltammograms obtained at positive potentials. The absence of peaks pointed out the weak interaction between electroactive bacteria and the fluid-like electrode, according to current density (Figure 3). The low electrical capacity for this electrode was almost identical to the cell-free abiotic assay. Accordingly, the low values for both TOC and antibiotic removal were consistent with the poor electron transfer among bacteria and the fluid-like electrode at −200 mV.

However, this interaction between electroactive bacteria and fluid-like electrode was considerably increased at positive anode potentials (+ 200, +400, and + 600 mV). The higher interaction could be correlated with the width of the obtained voltammograms and the presence of redox peaks during the vast range of tested potentials. Those redox peaks were observed at −500 mV (reduction peak) and −200 mV (oxidation peak), achieving a maximum current density of 13 A m^–3_NRV when the anode was poised at + 400 mV. It is important to note that voltammograms were similar when fluid-like anode was polarized either at + 200 or + 400 mV. In contrast, the voltammogram at + 600 mV (vs. Ag/AgCl) showed redox peaks that suggested less electroactivity in the biofilm. Instead, two new peaks were observed, a reduction peak at −200 mV and an oxidation peak at −400 mV, suggesting a shift in the ETT mechanism previously reported by Geobacter sulfurreducens when anode polarization was observed upgraded from + 200 to + 600 mV (Busalmen et al., 2008).

Cyclic voltammogram analysis revealed that interaction between electroactive bacteria and the working electrode was optimized by applying an anode potential of + 400 mV. Nevertheless, CBZ removal was slightly higher at + 200 mV (Figure 2), which suggests that non-electroactive communities from ME-FBR may also have a role in biodegrading certain pharmaceutical pollutants.

After optimization of the applied anode potential, the achieved ME-FBR bioremediation was compared with the UASB reactor’s wastewater treatment capacity.

As previously shown, TOC removal was remarkably higher in the ME-FBR case than in the UASB treatment where the anaerobic oxidation capacity of the anaerobic consortium was tested for pharmaceutical compound removal, with a %TOC removal of ∼85%, indicating higher removal of the pharmaceutical compounds compared with the conventional anaerobic treatment (43.18%). Higher TOC removal was obtained with ME-FBR due to (i) the cooperation between an active electrode biofilm and planktonic bacteria not limited to mass transfer issues and (ii) the continuous polarization of the fluidized bed that boosts the pharmaceutical compound removal, compared to conventional anaerobic digestion technologies.

As shown in the preceding sections, CBZ and SMX oxidation led to the formation of chemical intermediate species. This fact is easily observed in the partial removal of TOC (Figures 2, 5). In order to evaluate the potential ecotoxicity of such intermediates, we measured the antibacterial activity of our effluents toward V. fischeri.

The luminescence inhibition of the polluted synthetic wastewater reached 100% after 15 min of exposure time, clearly indicating its high ecotoxicity (Figure 6A). That luminescence inhibition was further translated into an environmental indicator, which is the detoxification of the wastewater (Figure 6B). Interestingly, the ME-FBR operation rapidly decreased the ecotoxicity of the effluent (Figure 6B) regardless of the electrode potential tested. This suggested the absence of toxic metabolites after microbial metabolism of CBZ and SMX. In other words, the high detoxification achieved during the ME-FBR operation denotes the depletion of the environmental risk associated with the initial wastewater polluted with CBZ and SMX after the ME-FBR operation. In addition, among all potential tests, those enhancing oxidative conditions [especially + 400 mV (vs. Ag/Ag)] showed detoxification as high as 70%. We can conclude that such high-value redox polarizations severely impact the removal of toxic metabolites from CBZ and SMX; a similar trend was observed for transforming CBZ and SMX (Figure 2). This high detoxification is clearly due to the action of electroactive bacteria in ME-FBR according to the low CBZ and SMX removal obtained in the electrochemical control test.