Methane Production Using Brewery Spent Grain: Optimal Hydrothermolysis, Fermentation of Waste and Role of Microbial Populations

The hydrothermolysis variables temperature (150–210 °C) and time (10–20 min) were assessed to improve hydrolysis efficiency of brewery spent grain (BSG) for renewable energy generation. The intensification of the pretreatment was expressed by the severity variation (2.8–4.5) and the process was optimized with methane production of 411.6 ± 7.2 mL g−1 TVS (severity 4.2). The fermentation-methanogenesis of BSG and hydrolysate resulting from BSG hydrothermolysis process under severity of 4.2 (210 °C for 10 min) was evaluated by central composite design with the variables operation temperature (30–60 °C), BSG concentration (7.3–20.7 g L−1) and hydrolysate (0–12.4 mL). The higher methane production observed was 305.8 mL g−1 TVS, with 14 g L−1 of BSG, without hydrolysate at 45 °C. The main soluble metabolites were acetic acid (233.17 mg L−1) and butyric acid (156.0 mg L−1). On other hand, the lower methane production (108.5 ± 2.0 mL g−1 TVS) verified was 14 g L−1 of BSG, 6.5 mL of hydrolysate at 60 °C, which revealed that in this condition propionic acid (947.4 mg L−1) and acetic acid (599.2 mg L−1) were expressive. In the optimal fermentation-methanogenic condition of pretreated BSG, Macellibacteroides and Sphaerochaeta (15.9 and 14.7%, respectively) were identified, as well as archaea similar to Methanosaeta (80.4%), favoring the acetoclastic methanogenic pathway.


Introduction
Due to population growth, energy consumption is expected to increase by approximately 50% from 2018 to 2050, reaching 1090 quadrillion British thermal units (Btu) [1]. The highest energy consumption is within the industrial sector, which is responsible for more than 50% of the total in relation to other sectors [1]. Therefore, national energy security, continuous use of renewable energy sources and energy recovery from waste are very relevant subjects [2,3].
The conversion of residual lignocellulosic biomass, from various processes, into bioproducts and biofuels is environmentally suitable and in agreement with the concept of sustainability [4]. Whether biodiesel, bioethanol, hydrogen or methane, these biofuels can replace conventional fossil sources and potentially reduce carbon emissions [5]. Methane can be produced from diverse feedstock, including lignocellulosic materials, such as municipal, food and agroindustry wastes, considering availability, abundance and organic matter content.
Beer is one of the most appreciated and consumed drinks in several countries and despite the technological advances in large-scale beverage production, some residues are process inherent, highlighting the brewery spent grain (BSG), which ranges from 14 to 20 kg for every 100 L of beer produced [6]. Because of the wide, year-round availability of BSG, anaerobic digestion from this material for biofuel production is interesting and promising. The BSG use for bioproducts process like sugars extraction, proteins, enzymes and antioxidant acids have been evaluated, although the main destination is still animal feed [7]. BSG lignocellulosic structure, whose composition is 17-26% of cellulose, 19-42% of hemicellulose and 12-28% of lignin [8], can provide resistance to its degradation.
Hydrolysis of lignocellulosic material is a determining step of anaerobic digestion, since hydrolytic enzymes released by microorganisms are limited to completely disrupt the complex lignocellulosic structure. This step emerges in few days for proteins and lipids, and can occur within several days for lignocellulosic components [9]. Thus, the study of anaerobic digestion under different conditions and pretreatments applied in lignocellulosic biomass are necessary.
The pretreatment is employed to prepare the lignocellulosic substrate for microbial activity, in order to make it more accessible to the microbial consortium [10]. Therefore, some viability criteria must be considered, such as effective reduction of particle size, increased substrate porosity, degradability and solubility, production inhibitors and energy consumption [11]. Depending on the treatment applied, they can be classified as physical, chemical or biological.
Among the physical pretreatments, hydrothermolysis is related to the application of high temperatures, pressure and time, through which the rupture of the chemical bonds of the cell wall occurs and the release of cellular components in the liquid phase, making soluble and biodegradable portions that were previously insoluble [12]. The combination of the factors temperature and time of hydrothermolysis results in different pretreatment severities.
Antwi et al. [13] pretreated cocoa pods residues by hydrothermolysis with temperature and reaction time varying between 150-220 °C and 5-15 min, respectively. The optimal biogas yield (526.38 mL g −1 VS) was obtained under severity of 2.65 (150 °C-15 min) while biogas potential of 357 mL g −1 VS (55% of methane) was verified with untreated residues. Ahmad et al. [14] evaluated the factors temperature (160, 180 and 200 °C) and time (5, 12 and 19 min) of hydrothermolysis from sugarcane bagasse pretreated by hydrogen peroxide. According to Ahmad et al. [14], 180 °C for 12 min were the optimized factors to increase the methane production up to 118.64%. Thus, it can be observed that optimal hydrothermolysis severity ranges according to the type of substrate.
Bochmann et al. [15] evaluated the methane production effects from BSG under 100 to 200 °C for 15 min, and verified optimal production at 140 °C (467.6 NmL g −1 VS), 14% more than the methane production obtained from the in natura substrate (409.8 NmL g −1 VS). Menardo et al. [16] also evaluated the temperature range (90 and 120 °C) of rice straw pretreatment for 30 min, and obtained increase of 24.5% on methane production with pretreated substrate (261 NmL g −1 VS) in relation to in natura (197 NmL g −1 VS).
[17] related greater methanogenic potential from the liquid fraction (376 ± 22 mL CH 4 g −1 VS) compared to the solid fraction (75 ± 6 mL CH 4 g −1 VS) resulting from the sugarcane bagasse hydrothermolysis process (150 °C for 40 min). Although the solid fraction was rich in sugars, the authors related the reduced methane production to hydroxymethylfurfural (HMF) and furfural (0.11 and 0.42 g L −1 , respectively) verified in the assays with bagasse, while high organic matter content was observed in the assays with hydrolysate.
During the disruption of lignin-hemicellulose-cellulose structure under high temperatures, furanic and phenolic compounds are inevitable, which are toxic and can inhibit the growth of bacteria and archaea. Phenolic compounds are released from the partial lignin rupture, while the 5-HMF and furfural are formed by the dehydration of pentoses and hexoses resulting from hemicellulose and cellulose degradation. In addition, formic acid is produced from 5-HMF and furfural, while levulinic acid is due to 5-HMF degradation [18]. However, contradictory, delignification is necessary to increase the biodegradability of waste [19].
Thus, to obtain reduced release of inhibitors concurrently with increasing solubilization of sugars, and consequently, the fermentation-methanogenesis efficiency, the lignocellulosic biomass pretreatment parameters can be statistically optimized [20]. In this study, BSG hydrothermolysis temperature (150-210 °C) and time (10-20 min) factors were evaluated regarding the maximum methane production. Then, the fermentation process was optimized through the central rotational composite design (CCD) and response surface methodology (RSM) in order to increase methane production. The effect of the hydrolysate from the hydrothermolysis process (0.6-12.4 mL), operation temperature (30-60 °C) and BSG concentration (7.3-20.7 g L −1 ) was evaluated. Subsequently, bacteria and archaea taxonomic characterization was carried out to map the BSG anaerobic digestion under optimal methane production conditions.

Substrate and Inoculum
Brewery spent grain (BSG) used as substrate in the methane production assays was donated by Kirchen (São Carlos, SP, Brazil). The BSG was washed, dried at natural air temperature and milled with knife mill (kind Willey SL-31, trademark Solab) [21] and stored at 4 °C.
The hydrothermal pretreatment (hydrothermolysis) was carried out by the addition of milled BSG (5.0 g) and 210 mL of water in the hydrothermal system. The hydrothermal reactor was built with stainless steel, consisting of reservoir (reactor) cyclone, integral proportional derivative controller (PID), pressure gauge and explosion valve [22]. At the end of the reaction, the solid fraction (pretreated BSG) was separated from the liquid fraction (hydrolysate) with a 2 mm diameter pore sieve. In the optimization assays under hydrothermolysis conditions (B1-B7) only the solid fraction (10 g L −1 ) was used, while in the optimization assays under fermentative-methanogenic conditions (C1-C17), the solid and liquid fractions were used at different ratios ( Table 2). The severity of hydrothermal pretreatment was determined according to Jacquet et al. [23] (Eq. 1): where S severity factor; t time (minutes); T temperature (°C).
Thus, as result of the temperature operational (150, 180 and 210 °C) and time of hydrothermolysis evaluated (10, 15 and 20 min), the severity ranged from 2.5 to 4.5.
Granular sludge from upflow anaerobic sludge blanket reactor (UASB) from a poultry slaughterhouse (Ideal Poultry, Pereiras, SP, Brazil) was used as inoculum source. The inoculum was kept at 4 °C and, prior to batch reactors inoculation, the granules were ground in a domestic blender, kept for 7 days at 37 °C for the recovery of microbial activity and removal of biodegradable organic matter, eliminating possible methane production from sludge. The control assays (C) were conducted under the same conditions as the others, without BSG, to verify the methanogenic potential from sludge.
According to analysis of the solid series [24], the sludge used as inoculum was composed of 60.7 g L −1 total solids (TS) and 56.0 g L −1 of total volatile solids (TVS).

Batch Reactors
The assays for the optimization of the hydrothermolysis conditions (B1-B7, Table S1, Supplementary Material) and the assays for the optimization of the fermentativemethanogenic conditions (C1-C17) were performed in triplicates of batch reactors, using Duran flasks of 250 mL (125 ml reaction medium and 125 mL headspace) and modified Zinder [25] culture medium. The Zinder medium [25] was modified with addition of yeast extract (1 g L −1 ) and sodium bicarbonate (10% w/v), and it was prepared according to Ahmad et al. [14]. The pH of the reaction medium was adjusted to 7.0 with addition of HCl.
Modified Zinder medium [25], inoculum (2 g kg −1 TVS) and substrate (1 g kg −1 TVS of BSG) were added to the reactors, always keeping the inoculum/substrate ratio equal to 2, according to Ahmad et al. [14], who applied inoculum to substrate ratio of 2/1 based on TVS in methanogenic reactors. Therefore, since the DCCR condition had a different concentration of BSG (286.6 g TVS Kg −1 ), the inoculum concentrations were also different. In Table 1 of central composite design (CCD) matrix, the substrate concentration and inoculum amount of each condition are presented.
After preparing the reaction medium, the reactors were subjected to an atmosphere of N 2 (100%) for 10 min, closed with a butyl cap and plastic screw, and incubated in an acclimatized oven. The operating temperature of the B1-B7 assays was 37 °C, while the operating temperature of C1-C17 varied according to the CCD matrix combinations (30-60 °C, Table 1).
Interactive effects between the variables maximum methane production response (P) were analyzed by Statistica 9.0 software package. The significance of experimental results (90% confidence level) was validated with F test analysis of variance (ANOVA). For the statistical model validation, reactors were operated in triplicates for optimal conditions of hydrothermolysis and fermentative-methanogenesis. The hydrolysate had pH 4.3, and organic matter concentrations, total soluble carbohydrates, phenols and acetic acid, respectively, of 17,554 ± 21.6 mg L −1 , 4240.0 ± 16.97 mg L −1 , 767.0 ± 10.6 mg L −1 and 93.0 ± 14.1 mg L −1 .

Physicochemical, chromatographic and microscopy analysis
In natura and pretreated BSG (210 °C for 10 min) were analyzed relative to insoluble lignin content (Klason lignin) and total soluble carbohydrates according to National Renewable Energy Laboratory protocol (NREL) [26] and by the phenol-sulfuric acid colorimetric method [27], with glucose as the standard. Morphological structure was analyzed by scanning electron microscopy (SEM) and samples were prepared in accordance with Nation [28], and the observation was performed using the Digital Scanning Microscope (DSM), Zeiss, model DSM-960.
Organic matter concentration was measured by Chemical Oxygen Demand (COD), phenols concentration by the 4-aminoantipyrine method and pH according to Standard Methods for the Examination of Water and Wastewater [24]. The total soluble carbohydrates analysis used the phenol-sulfuric acid colorimetric method [27], with glucose as the standard.
Organic acids were quantified by a Shimadzu® (GC-2010) gas chromatograph according to Adorno et al. [29] and biogas composition was determined using a Shimadzu® gas chromatograph (GC-2010) equipped with a thermal conductivity detector (DCT). A Carboxen 1010 PLOT column (30 m × 0.53 mm) was used. Temperature of injector, furnace and detector were 220 °C, 130 °C and 230 °C, respectively. Argon was used as drag gas, with a column flow of 5.66 mL min −1 with a volume of 12 mL. min [30]. The cumulative methane production data were adjusted using the Gompertz model (Eq. 2) modified [31]: where CH 4 cumulative methane production, P maximum methane production (mL de CH 4 g −1 TVS); R m maximum methane production rate (mL de CH 4 g −1 TVS. h); λ time to start methane production (h); e Euler number (2.71828182). The adjusted experimental data was performed for the average of the values obtained by triplicate reactors using the software Origin Pro 9.0 software.

Microbial Community Analysis
The microbial characterization by massive sequencing was carried out with samples collected from the batch reactors at the end of the operational period. The samples were washed with PBS buffer (NaCl 8%, KCl 0.2%, Na 2 HPO 4 1.44%, KH 2 PO 4 0.24%) and total DNA was extracted using protocol according to Sakamoto et al. [32]. Purity (260/280 nm ratio) and quantity (ng L −1 ) of the extracted DNA were measured in a Nanodrop spectrophotometer (Thermo Fisher Scientific, USA).
Operational taxonomic units (OTUs) were clustered considering 97% of similarity and the bacteria and archaea populations with relative abundance greater than 1% were considered in the results and discussion section.
Methanogenic assays to anaerobic degradation from unpretreated BSG were conducted using batch reactors with anaerobic sludge as inoculum source, 10 g BSG L −1 at 35 °C, where it was observed 171.6 ± 4.9 mL CH 4 g −1 TVS (P) and 1.2 ± 0.43 mL g −1 TVS h (Rm) [33]. From 1 3 these results it was possible to conclude that even under the lowest severity applied (2.5), the milling process associated with hydrothermal pretreatment of BSG was favorable to methane production, since under this condition was verified 250.0 ± 9.9 mL g −1 TVS (P) and 0.9 ± 0.06 mL g −1 TVS h (Rm), representing improvement of 45.7%. The effects of biomass particle size on anaerobic digestion were previously described by Dai et al. [34], who related that the particle size reduction of rice straw (from 20 to 0.075 mm) improved methane yield from 107 to 197 mL g −1 VS. Menardo et al. [16] also observed increase of methane production from 240 to 370 mL g −1 VS using reduced particle size of barley straw (from 5 to 0.5 cm). Therefore, increasing the surface available to microbiota by milling, the digestibility of BSG and hydrolytic process were improved [16]. Besides, the association of milling with the hydrothermal pretreatment, even under the lower temperature (150 °C), probably favored the organic matter hydrolysis and breakdown during anaerobic digestion [35].
In light of this, it was decided to continue this study with pretreated BSG by hydrothermolysis under severity of 4.2 (210 °C for 10 min). When the hydrothermolysis effects on BSG structure was evaluated under the abovementioned condition, there was no significant change on the lignin content of the material (29.1%) when compared to BSG in natura (26.5%). However, after the pretreatment, an increase of total carbohydrates solubilization from 33.5% (BSG in natura) to 53.5% (pretreated BSG) was confirmed, as well as morphological modification on the structure of BSG fibers, verified by Digital Scanning Microscope (DSM) (Fig. S5, Supplementary Material).
Kinetic parameters P, Rm and λ were obtained (Table 2) under different conditions of operation temperature (30 to 60 °C), pretreated BSG concentration (7.3 to 20.7 g L −1 ) and hydrolysate volume (0.6 to 12.4 mL). In relation to P, the greater results were obtained for C13 and C3, with 305.8 ± 7.79 and 286.6 ± 3.47 mL g −1 TVS, respectively. In these assays, the start time of methane production (λ) were similar, 16.5 ± 7.4 and 14.3 ± 7.83 h, Rm of 1.7 ± 0.1 and 1.7 ± 0.2 mL g −1 TVS h, respectively. In C13 and C3, lower hydrolysate volumes were added, 0.6 and 3.0 mL, respectively, supporting the negative effect of this variable.
The effect of hydrolysate addition on P can also be observed when comparing assays C13, C14, C15, C16 and C17, with 14 g BSG L −1 , operated at 45 °C and with different hydrolysate volumes. In C13, with 0.6 mL of hydrolysate, 305.8 ± 7.79 mL CH 4 g −1 TVS was produced. Concurrently to the increase of hydrolysate volume to 6.5 mL (C15, C16 and C17) and 12.4 mL (C14), there was a decrease of 32.4 and 38.3% in maximum methane production, respectively. Hydrolysate produced after hydrothermolysis may contain inhibiting compounds, like furfural and 5-HMF, therefore the microbial community requires a longer adaptation period [36], which may have compromised the BSG anaerobic digestion, causing lower methane production in C14, C15, C16 and C17 compared to C13.
After performing assays with concentrations of 0.4, 0.8 and 2.0 g L −1 of furfural and 5-HMF, Ghasimi et al. [19] reported total inhibition of methanogenesis with 2.0 g L −1 of both aldehydes, and partial inhibition with 0.8 g L −1 of 5-HMF, under mesophilic (35 °C) and thermophilic (55 °C) conditions. According to the authors, the thermophilic microbial consortium was more sensitive to the increase of intermediate compound concentrations released from the hydrolysis of lignocellulosic biomass when compared to mesophilic consortium, characterized by reduced methane production. These results corroborate those verified in this study, where assays with lower hydrolysate volumes (0.6 and 3.0 mL) and lower concentrations of compounds derived from BSG lignocellulosic disruption, were favorable conditions to anaerobic digestion, since there was greater methane production (305.8 ± 7.79 and 286.6 ± 3.47 mL g −1 TVS).
Lower P was observed in C10 (108.5 ± 2.0 mL g −1 TVS), where 14 g BSG L −1 and 6.5 mL of hydrolysate were added, operated at 60 °C, evidencing the negative effect of high temperature on microbial consortium. Additionally, higher λ was found in this assay, of 51.8 ± 2.9 h and Rm of 1.50 ± 0.13 mL g −1 TVS h, demonstrating the difficult microbiota adaptation to 60 °C. Similarly, Kim et al. [37] related the temperature influence on methane production from food wastes in reactors inoculated with mesophilic sludge, showing greater stability and methanogenic potential at 35 °C (230 mL g −1 VS) when compared to 55 °C (170 mL g −1 VS). For the same substrate, Zamanzadeh et al. [38] also observed better methane production (480 mL g −1 VS) at 37 °C when compared to reactors operated at 55 °C (448 mL g −1 VS). Various authors have reported superior performance of fermentativemethanogenic reactors conducted at mesophilic temperature in relation to thermophilic temperature, mainly when inoculated with a mesophilic microbial consortium, which may be related to the reduced adaptation ability of microbial groups to wide-ranging temperature variations [39,40]. For Chen and Chang [41], the ideal temperature range for the methanogenesis stage is from 35 to 42 °C, and an increase to 55 °C can especially inhibit the activity of methanogenic archaea, even after the acclimatization of mesophilic inoculum at 55 °C for 13 days. Thus, in this study, the methanogenic potential of the inoculum from UASB operated at mesophilic temperature was affected in reactors conducted at thermophilic temperature, probably due to the greater abundance of protein denaturation of mesophilic bacteria in the inoculum.
With regard to soluble metabolites in C1-C17, acetic acid concentrations were verified in the initial samples, from 35.0 to 301.1 mg L −1 (Fig. 1). Disruption of the lignocellulosic structure by hydrothermolysis contributed to the acetic acid content in the initial samples, since this acid is a product from hemicellulose degradation. In fact, it was confirmed that there was 93.0 ± 14.1 mg L −1 of acetic acid in the hydrolysate from BSG hydrothermolysis, which surely contributed to the respective initial concentrations. Thus, the assays with higher acetic acid concentrations in the initial sample were C7 (279.0 mg L −1 ) and C14 (301.1 mg L −1 ) with 10 and 12.4 mL of hydrolysate, respectively.
During the exponential phase of methane production ( Fig. 1 II) The highest concentration of propionic acid (836.1 mg L −1 ) in the intermediate phase was found in C10, the assay with the lowest methane production (108.5 ± 2.0 mL g −1 TSV). The accumulation of this acid indicates reactor disorder, since the overload is a consequence of the high partial pressure of hydrogen, which in turn suggests an unfavorable condition for acetogenesis, where a concentration between 900 and 2000 mg L −1 is considered toxic to the process [42]. Therefore, at the end of the assays (Fig. 1 III), in C10 there was an increase of propionic acid (947.4 mg L −1 ) and decrease of acetic acid (599.2 mg L −1 ), demonstrating the unfavorable condition of acetogenesis and methanogenesis at 60 °C.
The organic matter data (Fig. S6, Supplementary material) corroborates the decrease in total organic acids concentration from the exponential phase (II) to the final methane production phase (III). The highest concentration of total organic acids at the end of operation corresponded to assay C8 (5613 mg L −1 ), where the increase of acetic acid (2909.8 mg L −1 ) was verified. Probably, the low methane production observed in C8 (145.1 ± 5.2 mL g −1 TVS) occurred due to the inhibition of acetoclastic methanogenic activity, indicated by the acetic acid stock and caused due to the high temperature (54 °C) and hydrolysate volume (10 mL).
Depending on the condition, there was final pH ranging from 5.7 ± 0.07 to 7.2 ± 0.07. The lowest values were related to C8 (5.7 ± 0.07) and C10 (5.9 ± 0.10), where the high final concentrations of total organic matter were measured (5613.3 and 1639.0 g L −1 , respectively), which certainly contributed to the lower methane recovery (145.1 ± 5.2 and 108.5 ± 2.0 mL g −1 TVS, respectively).

Statistical Optimization of Methane Production with BSG Pretreated by Hydrothermolysis, Severity 4.2
In the optimization of methane production from pretreated BSG by hydrothermolysis (severity 4.2), the effects of the variables operation temperature (x 1 ), BSG concentration (x 2 ) and hydrolysate volume (x 3 ) were statistically analyzed. The methane production profiles (P) resulted from x 1 , x 2 and x 3 variations (shown in Table 3), which were delineated according to the central composite design (CCD) in Fig. 2.

3
The statistical significance of the model was determined by ANOVA (Table 3) and the correlation coefficient calculated (R 2 ) was 89.9%, describing the quality of the obtained adjustment. The second order polynomial equation (Eq. 3) represents methane production as a function of significant variables (p > 0.10): where Y predicted methane production (mL g −1 TVS), x 1 temperature operation (°C), x 2 BSG concentration (g L −1 ) and x 3 hydrolysate volume (mL).
For temperature (x 1 ), a significant negative effect was verified (square and linear). BSG concentration variable (x 2 ) had significant positive linear effect, and the variable hydrolysate volume (x 3 ) presented significant negative linear effect. Significant effects were also observed for the interactions between factors (x 1 .x 2 , x 2 .x 3 and x 1 .x 3 ). The operation temperature (negative square effect) represented the greatest magnitude effect on methane production, followed by hydrolysate volume (negative linear effect) and BSG concentration (linear effect). Based on the significant factors, the response surface for maximum methane production was developed, evidencing the optimal points resulting from the interaction of variables (Fig. 3). Throughout the statistical model, the optimal predicted operation temperature was at 36 °C, with reduced hydrolysate volumes (lower than 3.0 mL), and 18 g L −1 of BSG. In this conditions, optimal predicted methane production was 302.4 mL g −1 TVS.
Triplicates of reactors (assay C18) were carried out under these optimal predicted conditions for the respective variables, to validate the statistical model. 274.2 ± 5.6 mL g −1 TVS of methane production, Rm of 1.91 ± 0.10 mL g −1 TVS h, λ of 18.6 ± 3.9 h and R 2 of 0.99 were obtained. Therefore, there was 90.6% of accuracy in relation to the predicted values, thus, the data fit the model.

Taxonomy of Microbial Communities and Potential BSG Conversion Pathways
The structure of microbial communities (Bacteria and Archaea domains) was performed based on two samples, one of them was from the hydrothermolysis optimization assay (B2), and the other from the optimization assay of fermentative-methanogenic activity (C18). B2 assay was operated with 10 g L −1 of pretreated BSG by hydrothermolysis (severity of 4.2) at 37 °C, while C18 was operated with 18 g L −1 of pretreated BSG by hydrothermolysis (severity of 4.2) plus 3 mL of hydrolysate, at 36 °C. Based on the taxonomic characterization of both samples, it was possible to infer the effect of substrate concentration increase of 80% from B2 to C18 and the hydrolysate (3 mL) addition in C18.
Bacteroidetes and Firmicutes are widely found in mesophilic reactors and are related to acidogenesis and the ability to metabolize organic molecules, such as carbohydrates and proteins, to organic acids [43]. Paranhos et al. [44] reported predominance of the phylum Firmicutes (43%) and Bacteroidetes (37%) when characterizing poultry manure sample applied as inoculum in anaerobic reactors for lignocellulosic biomass co-digestion. In the present study, bacteria belonging to the phylum Bacteroidetes and Firmicutes may also have resulted from poultry manure, since the inoculum used was anaerobic sludge from UASB reactor fed with poultry slaughterhouse wastewater. Thus, in this study, members of this phylum probably converted complex organic molecules from BSG to soluble carbohydrates and organic acids.
Bacteria belonging to Spirochaetes are often found in anaerobic reactors and also in natural aerobic environments. They are associated to acetate, ethanol and lactic fermentation from sugars [45]. Lee et al. [46] related Spirochaetes to acetate oxidation, a thermodynamically unfavorable reaction that results in hydrogen production, through a syntropic relationship with hydrogenotrophic methanogenic archaea. This metabolic pathway probably occurred in B2 and C18, since hydrogenotrophic methanogenic archaea (Methanolinea, Methanoregula and Methanospirillum) were identified.
Members of the phylum Chloroflexi (7.6 and 4.3% in B2 and C18 samples, respectively) are able to metabolize carbohydrates and amino acids as well as oxidize hydrogen via homoacetogenesis [47]. Cloacimonetes (8.2 and 7.0%, in B2 and C18, respectively) are proteolytic and amino acids consumers, commonly found in wastewater treatment plants [48]. Additionally, it is reported that the phylum Firmicutes, Chloroflexi and Cloacimonetes are resistant to toxic compounds resulting from corn straw pyrolysis, such as phenolic compounds 5-HMF and furfural [49], like in this study, this phylum was identified in the assays where the substrate was subjected to extreme temperature and pressure conditions, which may have released 5-HMF and furfural in the reaction medium [14,50].
The difference between the bacterial genera identified in B2 and C18 was even more significant. This difference is shown in the circular ideogram, where each bacterial genus is represented by a colored segment that correlates it with the specific sample (B2 and/or C18) (Fig. 4).

and 3.2%.
Treponema, dominant genus in B2 (11.8%), use a wide variety of carbohydrates and amino acids as carbon source and can metabolize short and long organic acids chains [51]. Furthermore, they accomplish homoacetogenesis, through which hydrogen and carbonic gas consumption occurs with acetate production [52]. This metabolic characteristic is probably associated with the low relative abundance of methanogenic hydrogenotrophic archaea, since hydrogen was converted to acetate and methane by homoacetogenesis-acetoclastic methanogenesis. Syntrophorhabdus (7.5% in B2) allows the breakdown of compounds such as phenols, p-cresol, 4-hydroxybenzoate, isophthalate and benzoate in syntrophy with hydrogenotrophic methanogenic archaea [53]. Probably, this organism is involved in the conversion of phenolic compounds released from BSG lignocellulosic degradation, which favored the methanogenic activity in B2, the highest methane production (411.6 ± 7.2 mL g −1 TVS) in assays B1 to B7.
Macellibacteroides and Sphaerochaeta, belonging to the Spirochaetaceae family, are fermenters identified in both reactors. Spirochaetaceae (6.0% in B2 and 2.1% in C18) convert carbohydrates and amino acids into organic acids, mainly into acetic and butyric acids [54]. Usually found in mesophilic anaerobic reactors, bacteria similar to Macellibacteroides (7.4% in B2 and 15.9% in C18) have optimal growth at 35-40 °C and pH 5.0-8.5, which can convert sugar monomers primarily into lactate, acetate, butyrate and isobutyrate [55]. Similarly, Sphaerochaeta (3.8% in B12 and 14.7% in C18) can consume carbohydrates and produce, predominantly, acetate, formate and ethanol [56]. Regarding C18, it is likely that the sugar concentrations for BSG hydrolysis and hydrolysate favored Macellibacteroides and Sphaerochaeta, which were predominant in the reactor.
Under C18 conditions, Pelotomaculum were verified (2.3%), a acidogenic bacteria which are strictly anaerobic and metabolize a limited number of compounds, including propionate, primary alcohols, low molecular weight aromatics and lactate [57]. Both Hydrogenispora and Lutispora are endospore-forming bacteria, found in C18 (4.7 and 1.1%, respectively). Lutispora consume, primarily, amino acids and release amino acids into iso-butyrate, propionate and iso-valerate [58]. Hydrogenispora grows on medium rich in sugars, starch and yeast extract, forming acetic acid, ethanol and hydrogen, and are found in both anaerobic sludge and vegetables [59,60]. Thus, in this study, Hydrogenispora may have resulted from both BSG or anaerobic sludge.
Clostridium, identified in C18 (8.1%) and in B2 (1%) has been used as inoculum in reactors fed with lignocellulosic biomass due to its cellulolytic-fermentative ability, resulting in an increase of substrate degradability and, consequently, also due to the intermediates availability for methanogenesis [61]. Rabelo et al. [30] identified great quantities of Clostridium in reactors fed with in natura sugarcane bagasse (19%), as well as in reactors with the bagasse after hydrothermolysis (12.2%), which can metabolize organic compounds, such as carbohydrates and peptones, into diverse organic acids, acetones and alcohols. Moreover, Clostridium species are known for their resistance to high furan concentrations, as well as their ability to degrade them [36,62]. In this study, the bacteria of the genus Clostridium probably degraded these compounds, whose presence is undeniable because of the hydrothermolysis severity applied to the BSG (4.2).
Other groups characterized by cellulolytic-fermentative activity were also identified, such as Pseudomonas, Bacteroides, Gracilibacter, Ruminofilibacter, Ruminoclostridium. Pseudomonas were identified in C18 (4.5%) and also in B2 (1.1%), a bacterial group characterized by cellulolytic activity to degrade cellulose, hemicellulose and lignin, in addition to consuming biphenols [63]. In addition to BSG polysaccharide degradation, Pseudomonas were probably involved in the conversion of long chain organic acids in the reactor, making substrates available for fermentative bacteria [64]. Synergistaceae (2.0% in B2 and 1.7% in C18), although belonging to the phylum with lower relative abundance in both reactors (Synergistes), probably acted on amino acids metabolization producing short chain organic acids, such as acetic and butyric acids, releasing substrate for methanogenesis. Bacteroides (4.4% in C18) are found in the intestinal gut of animals [43] and, therefore, their probable presence in the sludge used as inoculum and collected from a poultry slaughterhouse wastewater reactor. Members of Bacteroides are related to carbohydrate hydrolysis and proteins and lipid degradation, becoming the available acetic, isovaleric, isobutyric and succinic acids [43,65].
Gracilibacter, Ruminifilibacter, Ruminoclostridium and Petrimonas were observed in lower relative abundance (lower than 3.2%). Gracilibacter are saccharolytic and produces acetate, lactate and ethanol [66]. Ruminofilibacter and Ruminoclostridium are cellulolytic and available hydrogen, carbonic gas, ethanol, acetate as fermentation products, readily consumed in the following anaerobic digestion steps [67]. Ruminoclostridium are generally related to enriched microbial culture with potential for cellulose degradation and hydrogen production [67]. Petrimonas (3.2% in C18 and 1.4% in B2) ferments a large number of monomers, with acetate and propionate as the main products, in addition to low levels of succinate [68].
Concerning the Archaea domain, the predominance of acetoclastic methanogenesis pathway is reiterated by the high relative abundance of the genus Methanosaeta (with 87.2 and 80.4%, respectively, in B2 and C18) (Fig. S9, Supplementary  Material). These archaea are consumers of electrons derived from propionate and butyrate oxidized into acetate [72], which may have been favored to more expressive concentrations of certain metabolites, such as acetic acid.
This configuration with a predominant acetoclastic pathway and less significant hydrogenotrophic methanogenic activity was previously described by Leclerc et al. [73] as the minimum microbiota necessary for the stability of anaerobic reactor. Stability was verified for both conditions, with methane production of 411 ± 7.2 and 274.2 ± 5.6 mL g −1 TSV in B2 and C18, respectively. Archaea similar to Methanolinea (6.2 and 5.5% in B2 and C18, respectively), Methanoregula (5.4 and 8.3% in B2 and C18, respectively) and Methanospirilum (2.4% in C18) were identified at lower relative abundance. These are hydrogenotrophic methanogenic archaea, which reduce carbonic gas into methane using hydrogen molecules as electrons donors.
The degradation of recalcitrant compounds, such as lignocellulosic materials, occurs through various biochemical process and microbial interactions, and based on metabolic characteristics reported in the literature and on the results presented in this study, it was possible to correlate the metabolic pathways that appeared in the BSG anaerobic digestion (Fig. 5).
Briefly, the BSG hydrothermolysis (severity 4.2) caused the rupture of BSG fibers, which favored substrate hydrolysis by Clostridium and Pseudomonas. BSG fermentation was also favored due to the higher release of soluble sugars which were converted into organic acids, mainly acetic acid, by Macellibacteroides, Bacteroides, Sphaerochaeta and Clostridium. Phenolic compounds resulting from hydrolysate and pretreated BSG did not compromise the fermentative-methanogenic process, since they were degraded by Syntrophorhabdus. Finally, acetic acid resulting from hydrolysate, released from the fermentation of soluble sugars, and formed from homoacetogenesis by Treponema, was converted into methane by Methanosaeta.

Conclusion
The application of hydrothermolysis to disrupt the lignocellulosic fibers of brewery spent grain (BSG) was crucial to increase methane production. Both temperature and time of BSG hydrothermolysis were statistically significant to increase the methane production to 75.1% with severity of 4.2.
Concerning the variables that can impact the pretreated BSG fermentation-methanogenesis, greater statistical significance was observed for the operational temperature, followed by hydrolysate volume and BSG concentration. According to response surface methodology, the optimal conditions predicted for methane production (302.4 mL g −1 TVS) was at 36 °C, with 3 mL of hydrolysate and 18 g BSG L −1 .
Changes were observed in the predominant bacterial phylum due to operational differences between B2 (10 g L −1 of BSG) and C18 (18 g L −1 of BSG and 3 mL of hydrolysate). In B2, Bacteroidetes (28.7%) was predominant, and in C18, Firmicutes endospore forming bacteria were dominant (37.6%). With regard to methanogenesis, acetoclastic methanogenesis was the main pathway, with Methanoseta prevailing in both conditions (87.2 and 80.4% in B2 and C18, respectively).

Conflict of 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.