Enhanced methane production from cellulose using a two-stage process involving a bioelectrochemical system and a fixed film reactor

Background It is desirable to improve the anaerobic digestion processes of recalcitrant materials, such as cellulose. Enhancement of methane (CH4) production from organic molecules was previously accomplished through coupling a bioelectrochemical system (BES); however, scaling-up BES-based production is difficult. Here, we developed a two-stage process consisting of a BES using low-cost and low-reactive carbon sheets as the cathode and anode, and a fixed film reactor (FFR) containing conductive material, i.e., carbon fiber textiles (CFTs) (:BES → FFR). By controlling the cathodic current at 2.7 μA/cm2 without abiotic H2 production, the three-electrode BES system was operated to mimic a microbial electrolysis cell. Results The thermophilic BES (inlet pH: 6.1) and FFR (inlet pH: 7.5) were operated using hydraulic retention times (HRTs) of 2.5 and 4.2 days, respectively, corresponding to a cellulose load of 3555.6 mg-carbon (C)/(L day). The BES → FFR process achieved a higher CH4 yield (37.5%) with 52.8 vol% CH4 in the product gas compared to the non-bioelectrochemical system (NBES) → FFR process, which showed a CH4 yield of 22.1% with 46.8 vol% CH4. The CH4 production rate (67.5 mM/day) obtained with the BER → FFR process was much higher than that obtained using electrochemical methanogenesis (0.27 mM/day). Application of the electrochemical system or CFTs improved the yields of CH4 with the NBES → FFR or BES → non-fixed film reactor process, respectively. Meta 16S rRNA sequencing revealed that putative cellulolytic bacteria (identified as Clostridium species) were present in the BES and NBES, and followed (BES→ and NBES→) FFR. Notably, H2-consuming methanogens, Methanobacterium sp. and Methanosarcina sp., showed increased relative abundances in the suspended fraction and attached fraction of (BES→) FFR, respectively, compared to that of (NBES→) FFR, although these methanogens were observed at trace levels in the BES and NBES. Conclusions These results indicate that bioelectrochemical preprocessing at a low current effectively induces interspecies H2 transfer in the FFR with conductive material. Sufficient electrochemical preprocessing was observed using a relatively short HRT. This type of two-stage process, BES → FFR, is useful for stabilization and improvement of the biogas (CH4) production from cellulosic material, and our results imply that the two-stage system developed here may be useful with other recalcitrant materials.


Background
Anaerobic digestion is one of the most promising biotechnologies for decomposition and stabilization of diverse organic substrates [1]. Currently, anaerobic digestion processes are frequently employed to produce renewable bioenergy in the form of biogas, such as CH 4 , which can be used as a replacement for fossil fuels to generate heat or electricity [2]. An important challenge of using anaerobic digesters is enhancing the overall process stability and achieving consistent biogas production, with a high percentage of substrate utilization. Many efforts have been made to control the various microorganisms involved in the multi-stage biochemical processes, including hydrolysis, acidogenesis, acetogenesis, and methanogenesis [3]. To improve the efficiency of digestion by microbial retention, various reactor types have been developed, such as fixed film reactors (FFRs) (previously referred to as packedbed reactor in our study) and fluidized bed reactors [4]. Additionally, two-stage processes to separate the acidogenic phase and methanogenic phases have been proposed [5] to increase total biogas production and demonstrate better process stability than single-stage processes [6].
An bioelectrochemical system (BES) uses microorganisms that transfer electrons to an electrode or that receive electrons from the electrode, which can be used for applications, such as wastewater treatment with simultaneous electricity production, electrochemical production of hydrogen (H 2 ) or methane (CH 4 ), and desalination [7]. A combination of anaerobic digestion and BES was previously used to develop a more robust waste treatment design [8]. Essentially, an electrochemical system consists of an anode and cathode, optionally separated by an ion-selective membrane to create two separate compartments through which a current can be applied or produced spontaneously; however, using a membrane may add to the overall cost of the system [9]. Thus, in many cases, bioelectrochemically assisted anaerobic digestion is performed without a membrane, and the externally applied potential can enhance CH 4 production, which makes in situ biogas upgradation possible [10][11][12]. This may be due to alterations in microbial metabolism and interspecies interactions, particularly affecting hydrogenotrophic methanogens [13,14]; because the H 2 concentration, i.e., H 2 consumption by methanogens, influences the biodegradation of organic compounds [15]. The electrode surface area and material are limiting factors for scaling up an integrated anaerobic digester and a BES [8]; however, anaerobic digesters are configured for large-scale production. Therefore, in this study, the BES and the anaerobic digester were separated and operated in a two-stage process. Moreover, we aimed to extend the effects of the first stage (the BES) to the second stage (an anaerobic digester using a fixed film).
In this study, a three-electrode BES was used in the first stage as a microbial electrolysis cell mode, where abiotic H 2 production was inhibited at a low current of 2.7 μA/cm 2 with low-cost and low-reactive carbon (C) electrodes, for both the cathode and anode. The main purpose of this design strategy was to boost CH 4 production, particularly in the latter FFR stage, which involved fixation of conductive carbon fiber textiles (CFTs) using a two-stage process (BES → FFR). Thus, inlet pH (pH in ) in the latter FFR or non-fixed film reactor (NFFR) stage was set to a neutral pH (7.5); however, the pH in in the BES or non-bioelectrochemical system (NBES) was set to an acidic pH (pH 6.1) to suppress biogas production in the former stage (Table 1). The NBES (without an external electrochemical system) and NFFR (without CFTs) were used as control reactors for BES and FFR, respectively. We propose utilization of a BES-FFR combination reactor to efficiently recover biogas from a model substrate containing cellulose as the major C source. Cellulose biodegradation has been evaluated because cellulose is the rate-limiting substrate in the anaerobic digestion [16].

Bioelectrochemical preprocessing accelerates methanogenesis from cellulose
First, we compared microbial CH 4 production from the BES followed by the FFR (BES → FFR) with that from Keywords: Bioelectrochemical system, Fixed film reactor, Two-stage process, Methane, Hydrogenotrophic methanogen, Cellulose  (Fig. 1). In this study, the less-reactive C sheet (i.e., a graphite block) was used as the electrode, and the current of the cathodic working electrode in the BES was set to 2.7 μA/cm 2 , low enough to eliminate abiotic H 2 production. This design enabled us to investigate the impact of electrochemical reactions on microbial CH 4 production from cellulose substrates in the two-stage process. Both types of twostage processes (BES → FFR and NBES → FFR) were performed for 17 days [twice the hydraulic retention time (HRT) of FFR] to acclimate the system to relatively long HRTs and establish a microbial consortium on CFT. Biofilms were rarely formed on the electrodes in BES and NBES. The results were compared with those found after 13 days of operation at a relatively short HRT ( Table 1). The reactor performances were compared after the former and latter stages were operated at an HRT of 2.5 and 4.2 days, respectively (Fig. 2). The reactor performances were similar at both sides of the H-type reactors in the BES (the cathodic and anodic sides) and NBES (the left and right sides); therefore, the average values of both sides are reported for the BES and NBES in the present Fig. 1 Schematic representation of the two-stage processes used in this study. The two-stage processes included a the BES followed by an FFR or an NFFR without CFTs, or b an NBES without a potentiostat followed by an FFR study. At this HRT (2.5 days), the synthetic medium containing cellulose as the major C source was added to the BES and NBES, at a cellulose load of 3555.6 mg C/(L day) (Fig. 2c). During the operation, pH in (pH 6.1) decreased, and the outlet pH was 5.2 ± 0.3 and 5.2 ± 0.2 in the BES and NBES, respectively. Therefore, due to the low pH, most cellulose remained without being consumed, corresponding to the removal of 11.4 ± 4.3% and 11.6 ± 4.3% of the suspended solids (SSs) in BES and NBES, respectively. Biogas production was suppressed (Fig. 2a), and most of the microbial products were represented by short-chain fatty acids (SCFAs), such as acetate, propionate, and butyrate, and ethanol (Fig. 2b).
The effluents of the BES and NBES were added to the FFR, which implied that the residual C (consisting of SSs and SCFAs + ethanol) was the influent for the FFR (Fig. 2c). During the operation, the pH in (pH 7.5) dropped, and the outlet FFR pH values were 6.8 ± 0.0 and 6.4 ± 0.4 with the [(BES→) FFR] and [(NBES→) FFR] systems, respectively. Interestingly, SS removal in the (BES→) FFR system (90.7 ± 2.2%) was much higher than that in the (NBES→) FFR system (66.9 ± 1.1%). Accordingly, the rate of biogas production (including CH 4 ) was much higher in the (BES→) FFR system than in the (NBES→) FFR system (Fig. 2a). The SCFAs and ethanol concentrations were lower in (BES→) FFR than that in (NBES→) FFR. The efficient removal of intermediates from the fermentation process enhanced SS removal in (BES→) FFR. As a result, the two-stage BES → FFR process achieved a CH 4 yield of 37.5 ± 1.4% with 52.8 ± 0.3 vol% CH 4 in the product gas, whereas the NBES → FFR process demonstrated a 22.1 ± 1.4% CH 4 yield with a 46.8 ± 1.3 vol% CH 4 in the product gas.

Microbial 16S rRNA gene sequence analysis of the microbial consortium in the reactors
The microbial communities developed during the twostage processes, BES → FFR and NBES → FFR, were examined at an HRT of 2.5 and 4.2 days, respectively (Table 1). Microbial abundances were determined by Fig. 2 Performances of the two-stage BES → FFR and NBES → FFR processes using HRTs of 2.5 → 4.2 days. a Rates of gas production in terms of CH 4 , CO 2 , and H 2 . b Production of SCFAs and ethanol. c C balances of the two-stage processes. The cellulose loads of the BES and NBES were both 3555.6 mg C/(L day). The residual C in all reactors was calculated based on the SSs
Using a combination of prokaryotic universal primers and the MiSeq platform, we obtained an average of 196,444 (± 29,391) reads for each sequencing reaction ( Table 3). The number of operational taxonomic units (OTUs) describes the approximate number of species. OTU numbers in the second stages, (BES→ and NBES→) FFRs, were higher than those in the first stages, BES and NBES, due to pH differences. Microbial species diversity is represented by the Shannon index [17], and Faith's phylogenetic diversity correlates with species richness [18]. The Shannon index value of the (BES→) FFR process was higher than that of the BES; however, the Shannon index value was lower with (NBES→) FFR than that with NBES. Consequently, the Shannon index value of (BES→) FFR process was higher than that of the Microbial compositions were examined at the species level ( Fig. 3). Most of the microorganisms were assigned to three phyla (Firmicutes, Chloroflexi, and Euryarchaeota) in the suspended fractions of two two-stage processes (BES → FFR and NBES → FFR) and in the CFT-attached fractions of FFRs [(BES→ and NBES→) FFR-C]. The dominant phylum was Firmicutes in the BES and NBES, consisting mainly of bacteria related to Thermoanaerobacterium saccharolyticum. Bacteria related to Clostridium sp. and Ruminococcus sp. constituted communities of predominant microorganisms in the BES, and bacteria related to Unclassified Clostridiales constituted communities of predominant microorganisms in the NBES. Almost no methanogens were detected in the BES and NBES, due to the low pH. In the suspended and CFT-attached fractions following the (BES→ and NBES→) FFR processes, bacteria related to Clostridium sp. constituted the major microbial community and were most abundant in (NBES→) FFR, corresponding to a low microbial diversity. With the (BES→) FFR process, bacteria related to unclassified species in the Clostridiales and MBA08 orders of Clostridia were the other predominant microorganisms. Interestingly, the relative abundances of methanogens related to Methanosarcina sp. increased after the (BES→) FFR-C process, compared to those found after the (NBES→) FFR-C process. The relative abundances of methanogens related to Methanobacterium sp. were higher in the suspended fraction of the (BES→) FFR process than those in the (NBES→) FFR process.

Application of a current stabilizes the two-stage BES → FFR process
We aimed to investigate whether application of a current as a bioelectrochemical-preprocessing step would improve the CH 4 yield of the two-stage NBES → FFR process. First, we operated each stage in the two-stage NBES → FFR process at an HRT of 2.5 and 4.2 days, respectively, after acclimatization ( Fig. 1 and Table 1).

Discussion
In this study, we developed a thermophilic two-stage process consisting of BES and anaerobic digester-fixing CFTs. Conversion of cellulose to CH 4 by microorganisms was enhanced using this type of two-stage process, compared to the process without an external electrochemical system in the first stage or without CFT in the second stage. We compared the input energy in terms of the applied electric current with the energy of CH 4 produced by microorganisms. CH 4 can be produced electrochemically through carbon dioxide (CO 2 ) reduction [19] via the following reaction: where 8 mol of electrons, that is 771,882 C (eight times the Faraday constant), are needed to produce of 1 mol of CH 4 . In the BES, 8.64 C (= 100 × 10 -6 × 86,400 s) of current per L was used for 1 day. If this was utilized entirely for CH 4 production, a maximum of 0.27 mM/day of CH 4 should be produced electrochemically under standard conditions. However, we obtained considerable amounts of CH 4 (53.9-70.2 mM/day) with our two-stage process containing BES. Evidently, a small amount of electric current affected the activity of the microbial community in our two-stage process. Bacteria related to Clostridium sp., T. saccharolyticum, Ruminococcus sp., and unclassified Veillonellaceae were predominant during the first stage (BES or NBES), after loading with cellulose (Fig. 3). Cellulose is highly recalcitrant to biodegradation, although cellulolytic microorganisms have been reported from the strains of Clostridium spp. and Ruminococcus spp. [20]. Additionally, uncultured Clostridia MBA08, mainly found with the (BES→) FFR process, was previously discovered in a thermophilic laboratory-scale digester used to process municipal solid waste [21]. Bacteria, such as Clostridium sp. and Ruminococcus sp., were present during both the first stage (BES or NBES) and the second stage (FFR) and potentially consumed cellulose. Members of the Veillonellaceae family were present in paddy soil amended with rice straw [22], and these microorganisms utilized cellulose degradation products. Cellulose biodegradation was limited in the first stage, probably due to the short HRT of 2.5 days. T. saccharolyticum can ferment a wide array of carbohydrates, including starch, glucose, xylose, and arabinose, but it cannot degrade cellulose [23], and T. saccharolyticum present in the former and latter stages would expect to utilize carbohydrates other than cellulose.
Notably, the relative abundances of the H 2 -utilizing methanogen, Methanobacterium sp. [24], increased in the suspended fraction and acetate-utilizing methanogen, Methanosarcina sp. [25], increased in the CFT-attached fraction of the (BES→) FFR process. It should be taken into consideration that thermophilic Methanosarcina sp. can perform hydrogenotrophic methanogenesis [25]. Another hydrogenotrophic methanogen, Methanothermobacter thermautotrophicus [26], was present at low levels in both (BES→ and NBES→) FFRs. Data from our previous study showed that syntrophy of cellulolytic bacteria with hydrogenotrophic methanogens accelerated cellulose decomposition [27]. Thus, an increase in the abundance of H 2 -scavenging methanogens contributed to the efficient conversion of cellulose to CH 4 in the (BES→) FFR process. The mechanism of this phenomenon in our two-stage process is discussed below. Data reported by Zhao et al. [28] indicated that a microbial electrolysis cell containing a conductive material (C felt) enhanced direct interspecies electron transfer to increase CH 4 production. Additionally, supplementation of conductive materials, such as C cloth, shifted interspecies H 2 transfer to direct interspecies electron transfer [29]. In this study, interspecies H 2 transfer is expected to dominate in the second stage, when the FFR contained the conductive material, CFT, following the first (BES) stage. Interspecies H 2 transfer would be promoted in both the first stage, when BES operated in microbial electrolysis cell mode (but abiotic H 2 production was prevented) and in the latter stage, with the (BES→) FFR process. That is, the applied electrical current in the initial BES stage would favor interspecies H 2 transfer in the following FFR stage. Findings on the dominance of acetate-removing Methanosarcina sp. on CFTs agreed with those of our previous study showing that Methanosarcina sp. were enriched on CFT because of the enhanced retention of these methanogens [30].
The abundance of suspended microbes and the microbial diversity in the CFT-attached fraction were higher in the (BES→) FFR system than those in the (NBES→) FFR system. Bacteria related to Anaerolineae CFB-26, Lutispora sp., Coprococcus sp., and Syntrophomonas sp. were particularly abundant in the (BES→) FFR system, probably due to the application of current to the first reactor. Anaerolineae demonstrate versatile metabolic abilities in terms of carbohydrate fermentation, but do not present any cellulolytic ability [31]. Additionally, Coprococcus can ferment carbohydrate [32]. Lutispora thermophila was capable of utilizing some amino acids for growth in the presence of yeast extract, but carbohydrates were not utilized [33]. The above-mentioned microorganisms produce acids. Bacteria, such as Syntrophomonas, can oxidize butyrate in syntrophic cooperation with methanogenic archaea [34]. These results are consistent with those of a previous research showing the accelerating effect of increasing microbial diversity on CH 4 production with laboratory anaerobic digesters [35]. Ecosystems with higher microbial diversity would have a higher probability of functional redundancy or functional stability [36]. Our data suggest that increased microbial concentration and diversity contributed to the efficient microbial conversion of cellulose to acids. Additionally, efficient removal of acids with the aid of methanogens provided an environment that increased biomass and biodiversity in the reactor.

Conclusions
In this study, we operated a thermophilic BES in microbial electrolysis cell mode (with a low current and without abiotic H 2 production) using low-cost C sheets as the cathode and anode in the first stage. In the second stage, we operated a thermophilic FFR containing the conductive material, CFT. This type of two-stage process, BES → FFR, accelerated the decomposition of cellulose by microorganisms and stimulated CH 4 production by decreasing the accumulation of acids and ethanol, when compared to the two-stage process without an electrochemical system (NBES → FFR) or without CFT (BES → NFFR). Electrochemical CH 4 produced due to the applied current was negligible, compared to the overall level of CH 4 production. Thus, both the electrochemical system and CFT were necessary for stable operation of the BES → FFR. The abundances of suspended microbes and the associated microbial diversity were higher in the (BES→) FFR process than in the (NBES→) FFR process. Bacteria with cellulolytic and saccharolytic abilities were present in the BES, NBES, and following (BES→ and NBES→) FFRs processes. Moreover, H 2 -consuming methanogens (Methanobacterium sp. and Methanosarcina sp.) showed increased relative abundances in the suspended fraction and attached fraction of the (BES→) FFR process, respectively, compared to those in the (NBES→) FFR process. Thus, interspecies H 2 transfer was promoted in the FFR to accelerate the conversion of cellulose to CH 4 by bioelectrochemical preprocessing. These results suggest that the BES → FFR process represents an advantageous system for the anaerobic digestion of cellulose, with a short HRT with the BES, thereby implying that BES and FFR can be subjected to small-and large-scale operations, respectively.

Configurations of the BES → FFR and NBES → FFR processes
BES was used for the first stage of fermentation. As shown in Fig. 1, BES was designed to have an H-type two-glass reactor (working volume of each reactor: 250 mL; total working volume: 500 mL) and a three-electrode system. Both the working electrode (cathode) and counter electrode (anode) were composed of C sheets (graphite blocks with dimensions of 25 mm × 75 mm × 2 mm), as shown in Fig. 1a. An Ag/AgCl reference electrode (saturated KCl) was inserted at the cathodic working side. Two C sheets were fixed in each reactor using stainless steel wire and connected to a potentiostat (PS-08; Tohogiken, Japan). The current of the working electrode was electrochemically regulated to 2.7 μA/cm 2 . Each glass reactor had one gas outlet port connected to a gas sampling collection bag. In contrast, the C sheets were allowed to remain open in the NBES under open-circuit operation and served as a control reactor for the BES (Fig. 1b).
FFR was used for the second stage after the BES or NBES. Each FFR was designed to have a glass reactor (working volume: 250 mL) equipped with two CFT sheets (DONACARBO paper; diameter: 25.0 mm; height: 70.0 mm; thickness: 2.4 mm; Osaka Gas Chemical Co., Ltd., Osaka, Japan) as support material. This glass reactor was connected to a gas sampling collection bag, as described above.

Start-up and operation
The BES or NBES were inoculated with 500 mL of sludge from the CH 4 fermenter and allowed to degrade 1% glucose at 55 °C. Next, the following FFR or NFFR was inoculated with 250 mL of the above-mentioned sludge. After inoculation, a predetermined volume of the fermentation liquid in both the cathodic and anodic sides of the BES or both the left and right sides of the NBES was discharged through the sampling port [37], and the same volume of fresh artificial medium was added once a day. The artificial medium contained the following (in deionized water): 20 g/L cellulose powder (Code 07748-75, Nacalai Tesque, Kyoto, Japan), 0.1 g/L KH 2 PO 4 , 0.2 g/L K 2 HPO 4 , 1 g/L yeast extract, 2 g/L NaHCO 3 , 1 g/L NH 4 Cl, 0.1 g/L MgCl 2 ⋅6H 2 O, 0.1 g/L CaCl 2 ⋅2H 2 O, 0.6 g/L NaCl, 10 mL/L trace element solution (Deutsche Sammlung von Mikroorganismen und Zellkulturen [DSMZ] 141 medium), and 1 mL/L vitamin solution (DSMZ 141 medium). A predetermined volume of fermentation liquid in the subsequent FFR or NFFR stage was discharged, and the same volume of the above-mentioned effluent (mixed in the cathodic and anodic sides, or the left and right sides) of the BES or NBES was added once a day. The contents of both the cathodic working and the anodic counter sides in the BES, or the left and right sides in the NBES, were mixed thoroughly using a magnetic stirrer before added to the FFR. The HRT was determined as the time required to replace the entire working volumes in the reactor. The temperature of the culture was maintained at 55 °C. The experiments were performed in triplicate.

Analysis
The volume of gas produced was measured with a water displacement method using a graduated cylinder. The CH 4 , CO 2 , and H 2 contents of the gas were measured using a gas chromatograph equipped with a thermal conductivity detector (GC390B; GL Sciences, Tokyo, Japan) and a stainless steel column packed with active C (30/60 mesh; GL Science), as per methods described previously [37]. The concentrations of lactate, acetate, propionate, butyrate, and ethanol were measured using high-pressure liquid chromatography (Shimadzu, Kyoto, Japan) with an Aminex HPX-87H column (Bio-Rad Laboratories, Hercules, CA, USA) and a RID-10A refractive index detector (Shimadzu). The suspension was filtered through a glass fiber filter (0.45 μm) to determine the SS contents in the cultures; then, the residue on the fiber was dried at 105 °C for 120 min, and the dry weight was measured. The removal of SS was determined using the following formula by considering the weight difference before and after fermentation: where SS inlet is the loaded SS weight and SS outlet is the residual SS weight after fermentation. Residual C in SSs was calculated by multiplying the SS weight by 72/162, which reflects the molecular weight of 6 C atoms divided by the molecular weight of cellulose, assuming that the amount of microorganisms are much lower than the residual cellulose.
The yield of CH 4 was calculated on the basis of g of C, as follows:

Microbial DNA extraction
Whole genomic DNA was prepared from the cultures and CFT-attached fractions, as per methods described previously [37]. A 5000-μL aliquot of each culture was centrifuged at 5000×g, and the pelleted material was suspended in 200 μL of Tris-ethylenediaminetetraacetic acid (EDTA) buffer (10 mM Tris-HCl, 1 mM EDTA, pH 8.0). Alternatively, the attached fraction was vortexed and recovered in 5000 μL of Tris-EDTA buffer, and then the pelleted material after centrifugation was suspended in 200 μL of Tris-EDTA buffer, as described above. Each of these suspended materials was transferred to a sterilized and DNA-free bead-beating tube containing SS removal (%) = SS inlet (g) − SS outlet (g) SS inlet (g) , CH 4 yield (%) = g of C in CH4 g of C in cellulose put into the reactor .
300 mg of glass beads (diameter: 0.1 mm). Five hundred microliters of Tris-EDTA-saturated phenol, two hundred fifty microliters of lysis buffer, and fifty microliters of 10% (w/v) sodium dodecyl sulfate were added to each tube. The mixtures were then shaken vigorously for 30 s at 5.0 m/s using the FastPrep-24 instrument (MP Biomedicals, USA). Next, each mixture was centrifuged at 22,000×g for 5 min. Each upper aqueous layer was transferred to a fresh tube containing 275 μL of isopropyl alcohol and a 1/10 volume of 3 M sodium acetate, and then chilled at − 20 °C for 10-15 min. The extracted DNA precipitate was pelleted by centrifugation at 22,000×g for 5 min, washed with 70% ethanol, and then dried under a vacuum. The DNA was subsequently dissolved in Tris-EDTA buffer.