Continuous Anaerobic Digestion of Wood Vinegar Wastewater From Pyrolysis: Microbial Diversity and Functional Genes Prediction

Wood vinegar wastewater (WVWW) is the main by-product of biomass pyrolysis process, which is more suitable to use anaerobic digestion (AD) to achieve energy recovery due to its large amount of organic matter. In this study, the up-flow anaerobic sludge bed (UASB) reactor was used to investigate the continuous anaerobic transformation of WVWW with gradient concentrations (0.3, 0.675, 1, 2, 3, 4, 5, 6, and 7 g COD/L). Then, the changes of microbial community, diversity index and functional gene were analyzed in detail. The results revealed that WVWW showed good AD performance in continuous fermentation. WVWW of organic loading rate (OLR) of >8.58 g COD/L⋅d showed severe inhibition on biodegradability and methane production, which is mainly due to the toxic substances as compared with the control group. The bacterial communities were dominated by phyla of Chloroflexi, Firmicutes, Proteobacteria, Acidobacteria, Synergistetes, and Actinobacteria. The gene abundances related to energy production, carbohydrate transport and metabolism were relatively high, which are mainly responsible for carbon forms conversion and carbohydrate degradation. This study will provide a basis for the screening and enrichment of functional bacteria and genes.


INTRODUCTION
Pyrolysis of biomass is currently the most attractive and prospective technology for converting biomass into renewable energy such as bio-oil and biochar (Zhu et al., 2019;Shen et al., 2020). Nevertheless, improper disposal of the pyrolysis by-product, wood vinegar wastewater (WVWW), causes environmental pollution and waste of resources, considering its high organic concentration and complex composition (Zhang et al., 2019a). Anaerobic digestion (AD) is a preferred method for adequately disposing organic waste because it involves high controllability, low operating cost, and production of methane with low carbon footprint. WVWW is a high-strength organic wastewater characterized by high chemical oxygen demand (COD, 110-120 g/L) (Zhang et al., 2019b), with acetic acid commonly accounting for about 30-70% of the total organic compounds (Theapparat et al., 2015). Since acetic acid is an essential precursor of anaerobic methanogenesis, AD of WVWW is justified.
According to (Pimenta et al., 2018) and Si et al. (2018) the organic characteristics of WVWW are similar to those of hydrothermal liquefaction wastewater (HTLWW), suggesting comparable anaerobic digestibility. Currently, the biological treatment of WVWW mainly involves aerobic technology, serving as an additive for investigating the effect of aerobic microbial activity and competitive adsorption of heavy metals Zhang et al., 2019a,b), whereas HTLWW is primarily degraded by AD for recovering energy. In addition to small biodegradable organic molecules (organic acids, alcohols, etc.) HTLWW also contains refractory molecules including phenols and N/O heterocyclic compounds which inhibit the methanogenic activity of anaerobic microorganisms (Wirth et al., 2015;Zheng et al., 2016). Tommaso et al. (2015) studied the anaerobic digestibility of HTLWW, concluding that the wastewater produced at high temperatures exhibited better cumulative methane production and anaerobic performance. Nevertheless, the high concentration (≥13.3%) of HTLWW, severely inhibits anaerobic microbial activity (Zheng et al., 2016;Fan et al., 2020). Diverse methods have been employed to improve the anaerobic methanogenic potential of HTLWW including physical absorption, photocatalytic oxidation, and ozonation. Several studies have also revealed that physical adsorbents such as activated carbon and zeolite increase the cumulative methane yield and mitigate inhibition by toxicants (Zheng et al., 2016;Usman et al., 2019). Photocatalyzed HTLWW can enhance methane production and the COD removal rate, increasing the total energy recovery to 66.7% (Quispe-Arpasi et al., 2018). Yang et al. (2018) utilized ozonation to efficiently convert phenols and N-heterocyclic compounds in HTLWW, with BOD 5 /COD increasing by 32.36% (maximum of 0.41). These studies demonstrated that AD is a feasible method for degrading HTLWW that can significantly improve the energy recovery of the hydrothermal liquefaction process system (Gu et al., 2019). Additionally, the microbial community structures were directly related to the degradation efficiency of the organic wastewater. Microbial community analysis revealed high amounts of bacteria capable of degrading N-heterocyclic and aromatic compounds such as Syntrophorhabdus and Synergistes (Usman et al., 2019). Si et al. (2016) also reported numerous microorganisms associated with acidification and detoxification in long-term continuous AD experiments.
In the current study, we focus on the anaerobic digestibility of WVWW with gradient organic lading rates (OLRs). The daily methane production and biodegradability (BD) are measured to evaluate continuous methanogenesis of the gradient OLRs. Since the microbial community exerts influence on organic wastewater degradation, the structures and diversity of the community under diverse OLRs are analyzed by high-throughput sequencing technology. The aim is to elucidate the function of and abundance of different genes, suitable for enriching detoxifying bacteria and screening of functional genes.

Materials
The WVWW was produced by condensation of steam from peanut shell pyrolysis at 500 • C, and the physicochemical characteristics involved a COD concentration of 110 g/L, ammonia concentration of 45.3 g/L, and pH of 3-4. The WVWW was stored at 4 • C. And granular sludge containing 9.74% total solids (TS, % w/w) and 75.35% volatile solids (VS, % of TS) was collected from up-flow anaerobic sludge bed (UASB) reactor treating sugar processing wastewater of the organic wastewater treatment plant in Shandong, China. It was continuously fed with 1 g COD/L sodium acetate in UASB reactor for 30 days to maintain microbial activity. All chemical reagents used in the experiments were of analytical grade.

Experimental Design
The UASB reactor was used to culture anaerobic microorganisms and enrich functional genes with increasing OLRs. The effective volume of the 7 cm diameter and 41 cm high reactor was 2.8 L. Half of its effective volume was filled with granular sludge and maintained at 38 • C by constant temperature circulating water, and the hydraulic retention time (HRT) was set at 48 h through a peristaltic pump (BT100-2J, LONGER, China). Samples of WVWW were prepared by diluting to 0.3, 0.675, 1, 2, 3, 4, 5, 6, and 7 g COD/L, with corresponding OLRs of 0.43, 0.97, 1. 43, 2.86, 4.29, 5.72, 7.15, 8.58, and 10.01 g COD/L·d. The pH of the influent was adjusted to approximately 7.0 using 3 mol/L NaOH and trace elements essential for microbial metabolism were added to retain 1 mL/g COD. The components and concentrations were as follows: H 3 BO 3 (50 mg/L), NiCl 2 ·6H 2 O (50 mg/L), MnCl 2 ·4H 2 O (50 mg/L), (NH 4 ) 6 Mo 7 O 24 ·4H 2 O (50 mg/L), AlCl 3 ·6H 2 O (50 mg/L), CuSO 4 ·5H 2 O (80 mg/L), ZnSO 4 ·7H 2 O (80 mg/L), CoCl 2 ·6H 2 O (200 mg/L), MgCl 2 (4000 mg/L), and CaCl 2 (4000 mg/L). A control experiment was conducted simultaneously using sodium acetate as the substrate, while other operating procedures were similar to that for WVWW samples. Each OLR stage lasted 15 days, with the gas volume generated and its composition measured. The granular sludge was sampled from the bottom of the UASB reactor for high-throughput sequencing analysis at the end of OLRs for the 1. 43, 2.86, 4.29, 5.72, 7.15, 8.58, and 10.01 g COD/L·d experiments, corresponding to influent organic concentrations of 1, 2, 3, 4, 5, 6, and 7 g COD/L. The microbial samples were labeled as IC-1, IC-2, IC-3, IC-4, IC-5, IC-6, and IC-7. At the end of the 10.01 g COD/L OLR experiment, the granular sludge was sampled from the bottom and the middle of the UASB reactor to evaluate the distribution of functional genes, with the samples named as a and b, respectively.
The experimental data were expressed as mean ± standard deviations.
Statistical analyses were carried out using Origin v9.1.

Biodegradability
Biodegradability can be used to analyze the AD performance of WVWW. It is calculated from the experimental methane production (EMP) and the theoretical methane production [TMP, 350 mL CH 4 /g COD (Lin et al., 1999) expressed in ] as follows: Analytical Methods

Generated Gas Analysis
The volume of biogas collected daily was measured by a medical syringe at room temperature, and the components quantitatively analyzed by gas chromatography (GC-6890, Agilent, United States). About 0.4 mL gas sample was injected into a GC equipped with a thermal conductivity detector and a packed TDX-01 stainless-steel column, with argon, the carrier gas, flowing at 50 mL/min. The inlet temperature, detector temperature, and constant oven temperature of the GC were 120, 150, and 100 • C, respectively.

Bacterial Functional Gene Prediction
The collected granular sludge samples were analyzed by highthroughput sequencing technology for bacterial functional genes prediction. The DNA of each microbial sample was extracted by the PowerSoil DNA Isolation Kit (Mo Bio Laboratories Inc., Carlsbad, CA, United States). The extracted genomic DNA was detected by electrophoresis on 1% agarose gel. The DNA molecules were amplified using TransStart Fastpfu DNA Polymerase (TransGen AP221-02) by a PCR Amplifier (9700, ABI GeneAmp R , United States). The V3 and V4 regions of the 16S rRNA genes of bacteria were amplified with (5 -ACTCCTACGGGAGGCAGCAG-3 )/806R and (5 -GGACTACHVGGGTWTCTAAT-3 ) as primers. The PCR amplifying procedure involved denaturation at 94 • C for 5 min followed by a total of 25 cycles, including cycles at 50 and 95 • C for 30 s and 72 • C for 40 s. The final extension program was performed at 72 • C for 7 min. The amplified genome was transferred to an Illumina MiSeq platform (Illumina, CA, United States) for sequencing by Majorbio Bio-Pharm Technology Co. Ltd., Shanghai, China. The samples were differentiated using barcodes and primers at the ends of sequences to obtain effective sequences. The sequence directions were corrected to produce optimized data, with all sequences classified at the 97% similarity level. Using the Silva (Release 128) database as the reference standard, the acquired sequences were taxonomically analyzed by the ribosomal database project (RDP) classifier at a confidence threshold of 0.7. The abundance-related indexes (ACE) and Chao, and diversity-related indexes, Shannon and Simpson, were calculated based on operational taxonomic units (OTUs). Rarefaction curves were produced to demonstrate the reliability of the sequencing data. The results of the high-throughput sequencing were then uploaded to the Sequence Read Archive (SRA) database of the National Center for Biotechnology Information (NCBI).
The sequencing results were evaluated by OTU cluster analysis. The functional genes of the bacterial communities were predicted by PICRUSt, with the abundance of relevant functional genes also calculated. The greengene ID corresponding to the OTU was compared using the COG and KEGG databases to obtain the COG and KEGG Ontology (KO) family information, with the abundances also recorded. The relevant information and function of the COG were analyzed through the eggnog database according to the COG code.

Effects of WVWW on Methanogenesis
The daily methane production and BD for the samples and control group experiments for diverse OLRs are displayed in Figures 1, 2. The daily methane production changes several times as the OLRs increase. At the early continuous AD (OLR ≤ 5.72 g COD/L·d) stage, the daily methane production values for the experimental samples are comparable to that of the control experiment. At an OLR of 7.15 g COD/L·d, the daily methane production of the WVWW sample is slightly higher than that of the control sample, suggesting that the essential elements in WVWW and biodegradable organic matter are conducive for methanogenesis. Zheng et al. (2016) also reported that HTLWW with a low volume ratio is non-toxic for anaerobic microorganisms, possibly enhancing small organics (organic acids, alcohols, etc.) supply and nutrients (N, P, etc.) for microbial metabolism. The trend of daily methane production then reverses; that is, results for the experimental samples are lower than that for the control sample, indicating that toxicants progressively hinder the methane production. The data for the experimental samples exhibit a cliff-like decline, mainly because high toxicants concentrations are severely inhibiting the methane  production activity of the archaea (Tommaso et al., 2015). Also, the BD values in the first phase are very high (Figure 2), possibly due to residual nutrients in the granular sludge. Overall, the BD shows an upward trend as the OLRs increase from 0.97 to 7.15 g COD/L·d. Further, a BD value > 1 indicates that WVWW with low concentration and high organic matter likely promotes methane production (Zheng et al., 2016). At OLR > 8.58 g COD/L·d, the BD variation is similar to that of the daily methane production, further demonstrating the inhibitory effect of high toxicants concentration on the activity of methanogens (Chen et al., 2017a). In addition, the BD at 10.01 g COD/L·d sharply decrease by approximately 20%, highlighting refractory WVWW at this OLR.

Effects of WVWW on Abundance and Diversity of Microbial Community
The high throughput sequencing of Illumina MiSeq was employed in analyzing bacterial structure changes during continuous AD of WVWW. The indexes of phylogenetic abundance and diversity of the microbial communities are presented in Table 1.  examined the relationships between the alpha analysis indexes and the diversity and richness of microbial communities. The Chao and ACE indexes reflect the richness of the bacterial community, with higher values corresponding to higher richness. Also, the Simpson and Shannon indexes are closely related to species diversity of a bacterial community; that is, a low Simpson index and high Shannon index mean high species diversity. Therefore, the sample IC-4 showing the highest ACE (685.57) and Chao (700.32) reflects the highest bacterial abundance. This indicates that the WVWW, under the OLR condition, is conducive for growth and reproduction of bacteria. The highest OTU value (584) for the sample also supports the earlier observation. However, sample IC-7 exhibits the highest Shannon index (4.44) and the lowest Simpson index (0.25), highlighting the highest species diversity. This suggests that specific microorganisms with strong toxicity resistance are enriched during the longterm culture.
Based on OTUs at 97% similarity, the rarefaction curves were produced (Supplementary Figure S1), enabling species abundance comparison and reflection on the rationality of the sequencing data of samples. The flat shapes of the rarefaction curves indicate reliability of the sequencing results.

Venn Diagrams Comparison
A Venn diagram is used for counting the common and unique species of the seven samples (Figure 3), thereby intuitively expressing species composition and species sharing similarity for the OTU level at 97% similarity (Chen J. et al., 2019). The total OTU for the seven samples indicate significant variation with increasing OLRs (from 499 to 584). Throughout the continuous AD, the shared OTU value is 274, with the unique OTU values generally increasing initially, and then decreasing with increasing OLRs, consistent with the total OTU. This indicates that the complex organic compounds in WVWW under low OLRs promote microorganism enrichment. Meanwhile, these organisms improve methanogenesis of wastewater and degradation of organic matter. Previous studies also suggest that bacterial abundance is closely related to the degradation efficiency of organics (Zhang et al., 2019c). Nevertheless, increasing OLRs create higher concentrations of toxic substances, and the toxic effects hamper the metabolism and reproduction of common microorganisms (Zheng et al., 2016). Evidently, selective enrichment of toxicity-tolerant microorganisms occurs during long-term AD culture.

Functional Genes Comparison
Functional Genes Comparison at Diverse OLRS According to Figure 5A, 25 COG functional genes related to RNA processing and modification, energy production and conversion, carbohydrate transport, and metabolism, and coenzyme transport and metabolism are detected. The seven samples are in similar COG functional gene classification, but their relative abundance change significantly with increasing OLRs. The gene abundance related to energy production and conversion is relatively high, such as transport and metabolism of amino acid, coenzyme, and lipid ( Figure 5A). These are mainly responsible for carbon forms conversion, playing an essential role in the methanogenic process during AD. The metabolism of coenzyme and amino acid can synthesize protein and polypeptide, which are essential for the growth and reproduction of microorganism. However, lipid was not detected in the wastewater, and genes associated with lipid metabolism may be responsible for cell membrane structure. Genes associated with carbohydrate transport and metabolism can promote carbohydrate degradation . The trace carbohydrate in WVWW may account for the scarcity of related functional genes. The low genes linked to defense mechanisms also demonstrate that anaerobic microorganisms are susceptible to toxicants, which is consistent with the observation that high OLRs reflect low methane production (Figure 1). Cell growth and reproduction are primarily related to protein turnover, replication, and recombination and repair genes.

Functional Genes Comparison at Different UASB Locations
From Figure 5B, the distribution of genes along the longitudinal direction of the UASB reactor look similar. Moreover, few differences exist in the species of related functional genes, indicating that the functional  43, 2.86, 4.29, 5.72, 7.15, 8.58, and 10.01 g COD/L·d. a and b represent sludge samples at different locations of UASB at OLR of 10.01 g COD/L·d. microorganisms are homogeneous. However, the functional genes abundance in the middle of the UASB reactor is slightly higher than that at the bottom, suggesting that the concentration of toxic substances decreases in the longitudinal direction of the reactor. Functional analyses of predictive genes also support this viewpoint such as amino acid transport and metabolism, carbohydrate transport and metabolism, inorganic ion transport, and metabolism, and general function prediction. And the decreased abundance of bacteria such as Anaerolineae, Aminicenantales, etc., which are associated with the detoxification of wood vinegar, also indicated lower toxicity in the middle of UASB (Figure 4). Low toxicity can accelerate the transformation and metabolism of inorganic ions, amino acids, and other small molecules. The bacteria at the bottom of the UASB reactor are mainly related to the acidification and detoxification of organisms, which is consistent with the higher abundance of genes related to defense mechanisms in the microbial samples at the bottom of the reactor than in the middle.

CONCLUSION
In this study, we investigated methane production, microbial communities, and functional genes prediction associated with continuous AD of WVWW. When the OLR was >8.58 g COD/L·d, the AD performance deterioration was mainly manifested by decreased daily methane production and BD. The common dominant bacteria for the seven samples at the phylum level were Chloroflexi, Firmicutes, Proteobacteria, Acidobacteria, Synergistetes, and Actinobacteria, and these functioned mainly in the acidification and detoxification of substrates. However, the structure and abundance of bacteria significantly changed at the genus level with increasing OLRs, suggesting that longterm continuous AD can promote the selective enrichment of functional bacteria. The functional genes associated with energy production and conversion, amino acid transport and metabolism, carbohydrate transport and metabolism, and inorganic ion transport and metabolism were predicted to dominate. Predicting genes functions provides a basis for further screening of functional genes.

DATA AVAILABILITY STATEMENT
All datasets presented in this study are included in the article/Supplementary Material.

AUTHOR CONTRIBUTIONS
DH and QF performed the experiments and wrote the manuscript. YZ and HX designed the research. LC and HS analyzed the data. YL analyzed the data and reviewed the manuscript. All authors read and approved the final manuscript.

ACKNOWLEDGMENTS
We would like to thank Editage (www.editage.cn) for English language editing.