Access the full text.
Sign up today, get DeepDyve free for 14 days.
Loading next page...
/lp/ou_press/cold-adaptation-and-replicable-microbial-community-development-during-fbRJi09L6j
- Publisher
- Oxford University Press
- Copyright
- Copyright © 2022 Federation of European Microbiological Societies
- ISSN
- 0168-6496
- eISSN
- 1574-6941
- DOI
- 10.1093/femsec/fiy095
- Publisher site
- See Article on Publisher Site
Abstract
Downloaded from https://academic.oup.com/femsec/article/94/7/fiy095/5004848 by DeepDyve user on 14 July 2022 FEMS Microbiology Ecology, 94, 2018, fiy095 doi: 10.1093/femsec/fiy095 Advance Access Publication Date: 25 May 2018 Research Article RESEARCH ARTICLE Cold adaptation and replicable microbial community development during long-term low-temperature anaerobic digestion treatment of synthetic sewage 1, ,†,§ 1 1 1,‡ 2 C. Keating * ,D.Hughes , T. Mahony ,D.Cysneiros ,U.Z.Ijaz ,C. 1,† 1 J. Smith and V. O’Flaherty Microbiology, School of Natural Sciences and Ryan Institute, National University of Ireland, Galway, Ireland. and Infrastructure and Environment, School of Engineering, University of Glasgow, Rankine Building, 79-85 Oakfield Avenue, Glasgow, G12 8LT, UK Corresponding author: Infrastructure and Environment, School of Engineering, Rankine Building, 79-85 Oakfield Avenue, University of Glasgow, Glasgow, G12 8LT, UK. Tel: +441413306310; E-mail: ciara.keating@glasgow.ac.uk Present address: Infrastructure and Environment, School of Engineering, University of Glasgow, Glasgow, UK. Present address: Future Biogas, 10–12 Frederick Sanger Road, Guildford, GU2 7YD, UK. One sentence summary: This paper explores low-temperature anaerobic digestion of a synthetic sewage-based wastewater with a focus on microbial community adaptation when using a mesophilic starting community. Editor: Alfons Stams C. Keating, http://orcid.org/0000-0001-9199-3068 ABSTRACT The development and activity of a cold-adapting microbial community was monitored during low-temperature anaerobic digestion (LtAD) treatment of wastewater. Two replicate hybrid anaerobic sludge bed-fixed-film reactors treated a synthetic ◦ −3 −1 sewage wastewater at 12 C, at organic loading rates of 0.25–1.0 kg chemical oxygen demand (COD) m d , over 889 days. The inoculum was obtained from a full-scale anaerobic digestion reactor, which was operated at 37 C. Both LtAD reactors readily degraded the influent with COD removal efficiencies regularly exceeding 78% for both the total and soluble COD fractions. The biomass from both reactors was sampled temporally and tested for activity against hydrolytic and ◦ ◦ methanogenic substrates at 12 Cand 37 C. Data indicated that significantly enhanced low-temperature hydrolytic and methanogenic activity developed in both systems. For example, the hydrolysis rate constant (k)at12 C had increased 20–30-fold by comparison to the inoculum by day 500. Substrate affinity also increased for hydrolytic substrates at low temperature. Next generation sequencing demonstrated that a shift in a community structure occurred over the trial, involving a 1-log-fold change in 25 SEQS (OTU-free approach) from the inoculum. Microbial community structure changes and process performance were replicable in the LtAD reactors. Keywords: anaerobic digestion; psychrophilic, hydrolysis; microbial community structure; adaptation Received: 18 May 2018; Accepted: 24 May 2018 FEMS 2018. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. 1 Downloaded from https://academic.oup.com/femsec/article/94/7/fiy095/5004848 by DeepDyve user on 14 July 2022 2 FEMS Microbiology Ecology, 2018, Vol. 94, No. 7 community within an ecosystem is essential for stability, pro- INTRODUCTION ductivity and sustainability (Girvan et al. 2005). This is true for High-rate anaerobic digestion (AD) of domestic wastewaters is AD reactors, regardless of operating temperature. Leven, Eriks- both successful and well established at full scale in tropical son and Schnur ¨ er (2007) reported a higher diversity of species regions (Bowen et al. 2014). Low-strength anaerobic treatment at lower temperatures during operation during mesophilic and of wastewater at ambient temperatures in areas with a tem- thermophilic conditions. Authors have also described a fur- perate climate, however, calls for efficient AD processes capa- ther increase in microbial diversity from mesophilic to psy- ble of operating below 20 C. Numerous successful laboratory- chrophilic conditions (Bialek et al. 2012). The reproducibility of scale low-temperature [<20 C] anaerobic digestion (LtAD) tri- bacterial community structure in reactor systems is debated als have been undertaken over the past decade for a range of owing to high functional redundancy and microbial popula- waste streams (e.g. Connaughton, Collins and O’Flaherty 2006; tion disparity between reactors and waste streams. It has been Enright et al. 2009;McKeown et al. 2012; Gouveia et al. 2015). reported that changes in microbial community structure in sus- Yet, despite laboratory-scale success and the economical and pended biomass systems can occur, even during stable opera- environmental advantages of LtAD, full-scale implementation tion (Fernandez ´ et al. 1999), while other authors have reported has not yet come to fruition. Moreover, many successful LtAD no changes in a microbial structure, despite perturbation (Akar- studies have focused on less complex wastewater and, as such, subasi et al. 2005). In contrast, other authors Collins, Mahony do not address the issue of solids hydrolysis (Petropoulos et al. and O’Flaherty (2006) and Madden et al. (2010) found mirroring 2017). It has been reported that hydrolysis rates decrease as tem- microbial communities in identical parallel granular reactor set- peratures drop and suspended solids subsequently may accu- ups. mulate in AD reactors, causing a reduction in treatment effi- The objective of this study was two-fold: (i) to examine ciencies and biomass washout (Elmitwalli et al. 2001;Singh the development of microbial structure and activity of a cold- and Viraraghavan 2002;Lew et al. 2003). Recent studies have, adapting community in replicated parallel LtAD reactors treat- however, demonstrated efficient treatment of sewage by LtAD ing a complex, but defined, wastewater; and (ii) to investigate (Smith, Skerlos and Raskin 2013; Keating et al. 2016) using reac- if mirrored microbial community development occurred in the tors designed to retain biomass and particulates. separate LtAD reactors seeded with the same mesophilically Efficient long-term treatment cannot rely solely on physi- cultivated biomass. We hypothesised that a mesophilic granu- cal entrapment. The degradation of organic matter to methane lar inoculum would demonstrate cold adaptation as well repro- during LtAD is dependent on the microbial community struc- ducible reactor performance and reproducible microbial com- ture (Raskin et al. 1994) and the activity (Lettinga et al. 1999; munity development in LtAD reactors with stable input and Foresti, Zaiat and Vallero 2006; Cavicchioli 2015) of the reac- operational parameters. tor biomass, which are strongly influenced by temperature. The requirement for a psychrophilic, or psychrotolerant, inoculum for successful LtAD has been proposed as being advantageous. MATERIALS AND METHODS The use of a truly psychrophilic inoculum (from naturally cold environments) has been tested by Xing, Zhao and Zuo (2010) Reactor design, set-up and operation and Petropoulos et al. (2017) with promising results. A disad- vantage of this approach is that this type of biomass is non- This study employed two glass laboratory-scale hybrid sludge bed/fixed-film reactors (R1 and R2) [2.8 l working volume] as granular and may not have high levels of activity against some substrates. Granular seed inocula are particularly advantageous described by (Hughes et al. 2011). Both reactors were seeded with −1 20 g volatile suspended solids (VSS) l of anaerobic biomass. for biomass settling and retention in high-rate AD reactors (van Anaerobic sludge granules were obtained from a mesophilic, Lier et al. 2001; Sakar, Yetilmezsoy and Kocak 2009). Granular full-scale, internal circulation reactor, located at the Carbery biomass also provides a more rapid start-up time (Elmitwalli Milk Products plant in Co. Cork, Ireland. The VSS content of the et al. 2001) and can prevent acidification (Neves, Oliveira and −1 granules was 119 g VSS l . The substrate used was synthetic Alves 2004). Using mesophilically cultivated granular inocula for −1 sewage (SYNTHES; Aiyuk and Verstraete 2004) at 500 mg l psychrophilic treatment without prior efforts to cold-adapt has COD . The reactors were operated at 12 Cinatriallasting for been deployed in numerous studies, with varying degrees of Tot 889 days. The trial was divided into five phases, each represen- success (Rebac et al. 1995; Langenhoff and Stuckey 2000;Smith, tative of a different hydraulic retention time (HRT) and organic Skerlos and Raskin 2013). In light of this information, long-term loading rate (OLR; Table 1). The filter unit was replaced on Day assessments into cold acclimation (how a community adapts to this change in its environment), activity (how active this com- 434. munity will be) and maturation (the sustainability of this adap- tation and activity) of cold-adapting AD communities and how these impact treatment efficiencies warrants deeper investiga- Reactor effluent analyses tion. Reactor effluent was sampled daily and also combined into a In practice, full-scale treatment facilities still work as a clas- weekly composite sample for total, chemical oxygen demand sic ‘black-box’ systems with the microbial community struc- (COD) (COD ), soluble COD (COD ), suspended COD (COD ) Tot Sus Sol ture and diversity largely unknown. As of yet, there have not and colloidal COD (COD ) determinations according to Stan- Col been sufficient advances to link what we know about the micro- dard Methods (APHA 2005). Protein and polysaccharide concen- bial communities to process optimisation and bio-monitoring trations in the effluent were determined by the Lowry method of AD a larger scale. Identifying the structure of the micro- (Lowry et al. 1951) and the DuBois method (DuBois et al. 1956), bial community during stable and unstable periods of opera- respectively. The concentration of volatile fatty acids (VFAs) in tion is crucial to understanding treatment parameters, but this the effluent was determined by chromatographic analysis in a in itself is not straightforward. The diversity of the microbial Varian Saturn 2000 GC/MS system (Varian Inc., Walnut Creek, CA). Biogas analysis was performed by gas chromatography Downloaded from https://academic.oup.com/femsec/article/94/7/fiy095/5004848 by DeepDyve user on 14 July 2022 Keating et al. 3 Table 1. Reactor operation phases and associated operational conditions. Phase 1 2 3 4 5 Days 0–104 105–259 260–559 560–665 666–889 HRT 48 36 24 18 12 TEMP 12 12 12 12 12 OLR 0.25 0.33 0.5 0.63 1 VLR 0.5 0.67 1 1.33 2 SLR 0.03 0.03 0.05 0.06 0.1 SLR 0.01 0.02 0.03 0.03 0.05 UV 2.5 2.5 2.5 2.5 2.5 a ◦ Temperature ( C). Hydraulic retention time (h). c −3 −1 Organic loading rate (kg COD m d ∗. d 3 −3 −1 Volumetric loading rate (m Wastewater m Reactor d ). e −1 −1 Sludge loading rate (kg COD kg[VSS] d )∗. f 3 −1 −1 Sludge loading rate (m Wastewater kg[VSS] d ). g −1 −1 Up-flow velocity (m h ). ∗Values calculated based on influent concentration of 500 mg l COD . Tot (Varian Inc., Walnut Creek, CA) according to Standard Methods The primers 1369F and 1492R and Taqman probe TM1389F (APHA 2005) (5 -CTTGTACACACCGCCCGTA-3 ) were used for bacterial anal- ysis (Suzuki, Taylor and DeLong 2000). The primers 787F and 1059R and Taqman probe 915F (5 -AGGAATTGGC-GGGGGAGCAC- Biomass characterisation 3 ) were used for archaeal analysis (Yu et al. 2005). Stan- Specific methanogenic activity testing dard curves were prepared using plasmids containing the full- To evaluate changes in the hydrolytic and methanogenic capa- length 16S rRNA gene sequence from a representative bacterial bilities of the seed (Day 0) and reactor biomass (sampled on days (Escherichia coli) and archaeal (Methanosarcina bakeri)strain. The 105, 260, 666 and 889), samples were screened using the specific plasmids were extracted using a Plasmid Extraction kit (BIO- methanogenic activity (SMA) testing method using the pressure LINE). A PCR reaction was then carried out using the primer transducer technique, as described previously (Colleran et al. pairs described above. This product was cleaned using QIAQuick 1992; Keating et al. 2016). PCR Clean Up kit (Qiagen, Crawley, UK) according to manufac- turer’s instructions. To construct the RT-PCR cDNA standard Substrate (Protein) degradation assays to assess substrate depletion curves were produced from cDNA prior in vitro transcription of curve for the determination of K, A and K the target mRNA by using the MEGAshortscript T7 kit (Ambion) max m The maximum specific activity ( A ), the maximum initial as described by Smith et al. (2006). The concentration of stan- max velocity (Vmax), the apparent half-saturation constant (K )and dards was measured in duplicate using a Qubit system (Invitro- the first-order hydrolysis constant ( k) of the seed inoculum and gen) and converted into copy concentration. A 10-fold serial dilu- 9 1 −1 reactor biomass were evaluated on a protein source (solubilised tion series (10 –10 copies ml ) was generated for each stan- skimmed milk powder). These rates were determined using sub- dard solution and analysed, in duplicate, with its corresponding strate depletion assays, which were set up similarly to the SMA primer and probe set. qPCR cycling conditions can be found in test described above and the kinetic parameters calculated as Keating et al. (2016). described by Bialek, Cysneiros and O’Flaherty (2013). Tests were ◦ ◦ performed, in triplicate; at 12 Cand 37 C using biomass and pro- −1 −1 tein concentrations of2gVSS l and 2 g COD vial ,respec- Illumina Miseq analysis tively. Terminal Restriction Fragment Length Polymorphism was used as a screening step to select samples for next generation DNA/RNA co-extraction from biomass sequencing (data not shown)-outlined in Keating et al. (2016). Genomic DNA and RNA was extracted from granular biomass Subsequently, DNA and cDNA from reactor biomass sampled samples taken from R1 and R2 on Days 0 (I), Days 105 (P1), 236 on Day 0 (Seed), Days 296 (P2b), 531 (P3b), Take-Down (E- (P2a), 296 (P2b), 392 (P3a), 531 (P3b), 546 (P3c), 666 (P4) and at the Day 889) and from the filter upon take-down (FE) were sent end of the trial (Day 889). Biomass was sampled from the fixed- for Miseq Illumina analysis at MR DNA (Shallowater, Texas, film filter at two points:—mid-trial (Day 454) and at the end of the USA). Universal 16S rRNA primer pair targeting the V4 region were used, 515F (5 -GTGCCAGCMGCCGCGGTAA-3 ) and 806R trial (Day 889). The nucleic acids were co-extracted by a modifi- cation of a phenol extraction method and processed as outlined (5 GGACTACHVGGGTWTCT-AAT-3 )—for paired-end sequencing with the forward primer in each pair containing a barcode by Keating et al. (2016). sequence. Amplicons were pooled and purified using calibrated Ampure XP beads (Bechman Coulter). This product was pre- Quantitative-polymerase chain reaction pared using the Illumina TruSeq DNA library protocol. The Quantitative polymerase chain reaction (qPCR) was carried out DNA library was processed on a Solexa Miseq machine accord- for Archaeal and Bacterial domains using DNA and cDNA ing to the manufacturer’s instructions. Sequences were anal- generated from granular biomass sampled from R1 and R2 ysed using an OTU-free approach using the DADA2 algorithm and the fixed-film filter as described above. qPCR was per- (Callahan et al. 2016). We used the standard workflow given formed using a LightCycler 480 (Roche, Manheim, Germany). at http://benjjneb.github.io/dada2/tutorial.html that learns the Downloaded from https://academic.oup.com/femsec/article/94/7/fiy095/5004848 by DeepDyve user on 14 July 2022 4 FEMS Microbiology Ecology, 2018, Vol. 94, No. 7 error model from the data first, dereplicates the reads and Replicability of reactor performance then runs the DADA2 algorithm separately on forward and During Phase 1, a significant difference ( P < 0.05) in performance reverse reads. Finally, merging the overlapping reads from both was observed between the two reactors. Reactor 2 (R2) average forward reduced sequence variants and reverse reads to give COD concentrations were much higher than reactor 1 (R1) for 1396 unique sequences (SEQs), which were then used to cre- all COD fractions (Table 2). However, this can be attributed to ate sequence tables for the different samples. The representa- a start-up period of ∼56 days for R2, while no start-up period tive SEQs were then taxonomically classified against the Silva was observed for R1. Both systems performed well upon com- 123 database with assign taxonomy.py script from Qiime (Capo- mencement of the second phase. Transient increases in the raso et al. 2010). To find the phylogenetic distances between effluent concentrations of the COD ,COD and COD frac- Tot Sus Col SEQs, we multisequence aligned the SEQs against each other tions from both reactors were observed upon further reduction using mafft v7.040 (Katoh and Standley 2013) and FastTree v2.1.7 of the applied HRT during Phase 3 (Table 2). The COD removal, Sol (Price, Dehal and Arkin 2010). Finally, the make otu table.py however, was not noticeably affected by this change (Table 2). from Qiime was employed to combine abundance table with tax- The particulate proportion of the influent (COD )was Sus onomy information. Raw sequences were submitted to the SRA degraded/retained in both reactors until concentrations in efflu- database under bioproject submission number SUB3108010. ent from R1 increased from Day 329 (Phase 3) and subsequently effluent COD concentrations also increased. This suggested Col that suspended solids might have been degraded to colloidal particles. Similarly, effluent COD and COD in R2 increased Sus Col Statistical analysis during this period. These fractions of COD remained elevated in GraphPad Prism software (San Diego, California, USA) was used effluent from both reactors, until the filter matrix was changed for calculating Student’s t-test based on reactor effluent param- on Day 434 (Phase 4). The fourth period of reactor operation eters and qPCR data. A significance level of 95% ( P < 0.05) was was characterised by efficient and stable process performance selected. Further statistical analyses of the sequencing data by both systems, with removal efficiencies of COD and COD Tot Sol were performed via the software R, version 3.4.1 (http://www. routinely >75% (Table 2). However, colloidal particles were not R-project.org/) using the SEQS tables and data generated as degraded/retained by either R1 or R2 (0% removal). The removal described previously and metadata. For community analysis, of the COD ,COD and COD fractions was not significantly Tot Sol Sus we used the package ‘Vegan’ (Oksanen et al. 2013). The follow- different (P > 0.05) between the replicate reactors during phases ing alpha diversity measures were used: Fisher’s alpha; Pielou’s 2–4. evenness; Richness; Shannon and Simpson. We used Vegan’s The final operational phase (Phase 5) was defined by an HRT −3 −1 aov() to calculate pair-wise ANOVA P-values and drew these on of 12 h and an applied OLR of 1 kg COD m d . The response top of alpha diversity figures. To calculate Unifrac distances, we to this HRT change perturbation was distinct in both reactors. A used the package ‘Phyloseq’ (McMurdie and Holmes 2013). Prin- period of biomass washout, lasting two weeks, upon commence- cipal co-ordinate analysis (PCoA) plot of community data (SEQs) ment of the phase was observed in R1, with effluent COD con- Tot −1 were made using different distance measures (Vegan’s cap- centrations reaching 1 g l , composed primarily of suspended scale() function): Bray Curtis; Unweighted Unifrac; and Weighted solids, before slowly decreasing over a period of 20–25 days. In Unifrac. The samples were grouped for different treatments as contrast, R2 displayed no obvious response to the HRT change. well as the mean ordination value and spread of points (ellipses The effluent VFA to COD ratio was highest during this phase were drawn using Vegan’s ordiellipse() function that represent (Table 2). A period of ∼100 days of stable operation was then the 95% confidence interval of the standard errors). To find SEQs recorded in both reactors before R1 effluent values began to fluc- −1 that are significantly different between different conditions, we tuate again, with COD concentrations reaching 440 mg l on Sus used DESeqDataSetFromMatrix() function from DESeq2 (Love, Day 868. R2 also displayed a period of less efficient performance Huber and Anders 2014) package with adjusted P-value (after from Day 819, where COD and COD increased (reaching Sus Sol −1 accounting for all comparisons) cut-off of 0.01 and minimum log below 130, and 150 mg l , respectively). An increase in efflu- fold change of 1. After performing multiple testing corrections, it ent VFA concentrations was recorded from Day 805, to reach a −1 reports SEQs that have log-fold changes between multiple con- range of 10–20 mg l (data not shown). It was demonstrated that ditions. The statistical workflows for the above can be found at effluent COD and COD were significantly different ( P < 0.05) Tot Sol http://userweb.eng.gla.ac.uk/umer.ijaz#bioinformatics. between both systems during this final phase of the trial. Protein was completely hydrolysed/degraded in both reac- tors throughout the trial with removal efficiencies of c. 100% (Table 3). Carbohydrate (the polysaccharide portion) was also completely degraded/retained in both reactors (Table 3)until the RESULTS filter matrix was changed on Day 434. Following this, effluent Reactor performance −1 carbohydrate concentrations from R1 reached 34 mg l on Day 490 (data not shown). The removal of carbohydrates was not sig- Both reactors treated the synthetic sewage wastewater success- nificantly different between systems during phases 1–4, but dur- fully, with COD removal efficiencies in excess of 80% generally ing the final phase of operation P was < 0.05. recorded, corresponding to low effluent COD concentrations of −1 typically less than 120 mg COD l at applied OLRs up to 0.63 Tot −3 −1 kg COD m day (Table 2). The performance was sustained Tot during the long-term trial, with minor fluctuations, until the Microbial activity and cold adaptation −3 −1 loading rate was increased to 1.0 kg COD m day from Tot Day 666 (Table 1), at which point the efficiency of the process The granular biomass was sampled temporally from each reac- decreased somewhat in R1, although COD removal rates of c. Tot tor throughout the trial and tested for its activity against 60% were maintained (Table 2). hydrolytic and methanogenic substrates under mesophilic and Downloaded from https://academic.oup.com/femsec/article/94/7/fiy095/5004848 by DeepDyve user on 14 July 2022 Keating et al. 5 Table 2. Average effluent COD ,COD ,COD and COD values during the five phases of reactor operation for R1 and R2. a; concentration Tot Sus Col Sol −1 in mg l , b; removal efficiency percentage, c; standard deviation, d; VFA:COD ratio based on average VFA concentrations and COD for each Sol phase. Sample Total COD Suspended COD Colloidal COD Soluble COD a b c (Conc) (RE) (SD) (Conc) (RE) (SD) (Conc) (RE) (SD) (Conc) (RE) (SD) R1 Phase 1 73 86 ±11 22 71 ±811 60 ±0.1 41 84 ±2 R2 Phase 1 140 73 ±462 17 ±0.2 23 17 ±556 78 ±2 R1 Phase 2 61 88 ±418 75 ±64 84 ±240 83 ±1 R2 Phase 2 76 85 ±631 58 ±0.7 13 27 ±0.7 32 86 ±4 R1 Phase 3 110 79 ±341 46 ±625 9 ±245 82 ±1 R2 Phase 3 103 80 ±19 39 47 ±15 24 14 ±440 84 ±2 R1 Phase 4 124 75 ±25 46 36 ±13 39 0 ±644 84 ±3 R2 Phase 4 105 79 ±827 63 ±14 33 0 ±546 83 ±0.2 R1 Phase 5 215 59 ±12 125 0 ±0.5 25 9 ±167 73 ±2 R2 Phase 5 114 78 ±561 19 ±0.9 15 45 ±137 85 ±3 VFA:COD (Ratio) Phase 1 Phase 2 Phase 3 Phase 4 Phase 5 R1 0.07 0.1 0.14 – 1.47 R2 0.04 0.14 0.48 – 0.56 Table 3. Average effluent Carbohydrate and Protein values throughout the five phases of reactor operation for R1 and R2. a; concentration in −1 mg l , b; removal efficiency percentage, c; standard deviation. Sample Carbohydrate Protein a b c (Conc) (RE) (SD) (Conc) (RE) (SD) R1 Phase 1 0 100 ±0 0.05 100 ±0.04 R2 Phase 1 0 100 ±0 0.01 100 ±0.05 R1 Phase 2 0.3 100 ±10.08 99 ±0.05 R2 Phase 2 0.1 100 ±0.05 0.03 100 ±0.1 R1 Phase 3 7.08 91 ±10.09 99 ±0.04 R2 Phase 3 4.1 95 ±10.07 99 ±1 R1 Phase 4 12 84 ±1 0 100 ±0 R2 Phase 4 24.3 69 ±3 0.02 100 ±0.02 R1 Phase 5 9.4 88 ±8 0.01 100 ±0.02 R2 Phase 5 2.1 98 ±1 0 100 ±0.06 −1 −1 ◦ psychrophilic conditions to; assess: (i) the activity of the micro- Methane (CH ) g [VSS] d )at12 C, hydrogenotrophic activ- −1 −1 bial population; (ii) if the microbial populations were adapting to ity (39±19 ml Methane (CH ) g [VSS] d ) was comparable to psychrophilic conditions; and (iii) if the activity and adaptation that measured at 37 C(Table 5). By the end of the first phase of developed at the same rate in both reactor systems. the trial, reactor biomass SMA at 37 C had increased, with the hydrogenotrophic activity having increased to a level 10 times and 3 times greater than the seed biomass in R1 and R2, respec- Hydrolysis tively. At the lower temperature, there was little change in SMA The hydrolysis rate constant (k) results demonstrated that, values compared to the inoculum. throughout the trial, biomass activity increased when tests were R2 biomass sampled during the second phase had an carried out under both mesophilic and psychrophilic conditions. increased SMA against all substrates tested at 37 C, with In fact, the hydrolysis rate at 12 C during phase 4 was increased the exception of acetate. In contrast, R1 biomass displayed by ∼20 times in biomass from both reactors, compared to the decreased activity for all substrates tested, except for propi- seed inoculum. In biomass from both reactors, A (Table 4) max onate. Biomass from both systems tested at the psychrophilic increased at both temperatures tested, with psychrophilic activ- condition demonstrated similar activity ranges with increased ity at the end of the trial being comparable to (R1), or greater activity noted against propionate (Table 5). Interestingly, no ace- than (R2) the mesophilic activity. K for mesophilic hydroly- toclastic activity was detected in either reactor biomass at this sis increased throughout the trial for both reactors, indicating point. a decrease in substrate affinity at the higher temperature. K At the end of the fourth phase, the SMA of biomass from both at the lower temperature decreased over time, indicating an systems was within a similar range. Under psychrophilic condi- increase in substrate affinity for both biomass sources under tions, the SMA profile had increased for both R1 and R2, with these conditions. comparable levels of activity in both biomass sources (Table 5). At the end of trial, the SMA at 37 C was comparable for R1 and R2 Acetogenic and direct methanogenic substrates biomass samples, with the exception of the direct methanogenic The initial inoculum had a high SMA at 37 C, with acetoclas- substrates, for which SMA was significantly lower in R2 biomass. tic activity being six times higher than hydrogenotrophic activ- The SMA at 12 C against all substrates was in a similar range for ity (Table 5). While SMA against acetate was only slight (7±1ml Downloaded from https://academic.oup.com/femsec/article/94/7/fiy095/5004848 by DeepDyve user on 14 July 2022 6 FEMS Microbiology Ecology, 2018, Vol. 94, No. 7 ◦ ◦ Table 4. Hydrolysis kinetic assays of reactor biomass at 37 Cand 12 C, based on a skimmed milk protein source. a; Maximum substrate utilising −1 −1 −1 −1 −1 rate gCOD gProtein d . b; Apparent half-saturation rate constant gProtein l . c; Maximum initial velocity gProtein l d for R1 and R2, d: −1 Hydrolysis rate constant d . Values are the mean of triplicates ± standard deviation in brackets. a b c d Sample A K Vmax k max m Seed 37C15(±3) 1.1 (±0.41) 0.9 (±1.07) 0.9 (±0.58) Seed 12C18(±0.13) 2.5 (±0.03) 0.3 (±0.05) 0.3 (±0.06) R1 Phase 1 37C12(±1) 1.8 (±0.36) 2.2 (±1.13) 1 (±0.42) R2 Phase 1 37 C 104 (±10) 2.7 (±0.07) 0.8 (±0.23) 0.8 (±0.31) R1 Phase 1 12C40(±7) 1 (±0.34) 1.1 (±0.4) 1.3 (±0.06) R2 Phase 1 12C19(±2.52) 4.3 (±3) 2.7 (±2.19) 0.9 (±0.21) R1 Phase 2 37 C 145 (±26.5) 3.9 (±2.12) 8.8 (±8.39) 5.7 (±2.18) R2 Phase 2 37 C 164 (±16) 3.1 (±2.23) 7.5 (±6.72) 2.1 (±0.91) R1 Phase 2 12 C 127.5 (±45) 1.6 (±0.11) 0.1 (±0.04) 0.8 (±0.16) R2 Phase 2 12C35(±27) 1.4 (±0.13) 0.1 (±0.06) 1.2 (±0.58) R1 Phase 3 37C94(±26) 1.6 (±0.43) 3.3 (±3.94) 1.6 (±0.37) R2 Phase 3 37C62(±9.5) 2.8 (±0.14) 3.1 (±0.70) 2.2 (±0.41) R1 Phase 3 12C52(±32) 2.6 (±0.19) 0.3 (±0.09) 1.1 (±0.32) R2 Phase 3 12C91(±29) 1.9 (±0.22) 0.3 (±0.16) 1.5 (±0.33) R1 Phase 4 37 C 257 (±32) 1.8 (±0.27) 1 (±1.07) 0.9 (±0.58) R2 Phase 4 37C15(±1) 2.1 (±2.02) 2 (±2) 1.9 (±1.18) R1 Phase 4 12C40(±18) 0.5 (±0.04) 0.1 (±0.01) 3.2 (±0.89) R2 Phase 4 12C51(±5) 0.3 (±0.08) 1.7 (±1.57) 4.7 (±2.92) R1 End 37C89(±33) 2.8 (±1.58) 2.3 (±3.73) 1.4 (±0.787) R2 End 37C72(±8) 1.2 (±0.16) 1 (±0.27) 2.9 (±0.216) R1 End 12C68(±7) 1.8 (±0.37) 0.4 (±0.23) 1.8 (±0.714) R2 End 12C83(±22) 2.2 (±0.19) 0.6 (±0.15) 1.6 (±0.383) ◦ ◦ −1 −1 Table 5. Maximum specific methanogenic activity (SMA) of reactor biomass at 37 Cand 12 C presented as ml Methane (CH ) g [VSS] d for R1 and R2. Values are the mean of triplicates ± standard deviation in brackets. Sample Propionate Butyrate Ethanol Acetate H /CO 2 2 Seed 37C84(±9) 523 (±30) 561 (±140) 300 (±33) 50 (±5) Seed 12C5(±2) 3 (±2) 7 (±4) 7 (±1) 39 (±19) R1 Phase 1 37 C 100 (±26) 155 (±19) 470 (±39) 345 (±20) 570 (±115) R2 Phase 1 37C30(±3) 55 (±18) 166 (±45) 179 (±32) 134 (±6) R1 Phase 1 12C8(±3) 2 (±15) 33 (±13) 14 (±11) 15 (±2) R2 Phase 1 12C5(±3) 6 (±1) 51 (±18) 14 (±7) 31 (±3) R1 Phase 2 37 C 136 (±20) 147 (±35) 230 (±97) 79 (±47) 65 (±41) R2 Phase 2 37C87(±4) 187 (±10) 307 (±46) 11 (±6) 193 (±41) R1 Phase 2 12C17(±8) 27 (±21) 39 (±2) 0 11 (±3) R2 Phase 2 12C22(±2) 2 (±2) 12 (±3) 0 24 (±1) R1 Phase 4 37C23(±2) 65 (±9) 257 (±92) 201 (±18) 170 (±15) R2 Phase 4 37C40(±22) 94 (±7) 21 (±9) 353 (±50) 308 (±21) R1 Phase 4 12C1(±1) 3 (±1) 31 (±6) 51 (±22) 19 (±9) R2 Phase 4 12C3(±1) 12 (±8) 46 (±12) 42 (±6) 98 (±23) R1 End 37C23(±13) 91 (±13) 153 (±33) 329 (±66) 231 (±29) R2 End 37C34(±2) 73 (±14) 171 (±16) 181 (±29) 110 (±6) R1 End 12C2(±1) 9 (±1) 21 (±6) 26 (±7) 38 (±3) R2 End 12C4(±2) 8 (±3) 29 (±4) 28 (±3) 35 (±4) both R1 and R2 and compared to the initial inoculum activity on 4-start of Phase 5), whereby the bacterial population increased ethanol and acetate had increased by ∼3 times (Table 5). to 31% and 43% of the total population, respectively (1.2 × 10 8 1 and 2.3 × 10 copies g- ) and was greater than in R1 biomass during these times. However, no significant difference ( P > 0.05) Microbial community structure was found throughout the trial. A 1-log reduction of the total Bacterial and archaeal numbers were quantified throughout the bacterial and archaeal gene copy numbers was also observed in reactor trial. The bacterial and archaeal profiles were generally R2 biomass on Day 666 (Phase 4-start of Phase 5; Fig. 1A). The reproducible in both systems with copy numbers in the range of filter communities were distinct from the other sampling points 8 10 1 8 10 1 with a greater proportion of bacterial to archaeal cells. This was 3 × 10 –3.5 × 10 (copies g- )and 2.4 × 10 –1.2 × 10 (copies g- ), respectively (Fig. 1A). The ratio of bacteria to archaea in R1 and due to a reduction in the numbers of archaeal cells relative to the granular biomass (Fig. 1A). R2 biomass was broadly similar also, but deviations were noted, for example, in R2 on Day 296 (Phase 2b) and Day 666 (Phase Downloaded from https://academic.oup.com/femsec/article/94/7/fiy095/5004848 by DeepDyve user on 14 July 2022 Keating et al. 7 Figure 1. (A) qPCR data of Bacterial and Archaeal 16S copy numbers (per g biomass) on the right y-axis from biomass samples (x-axis) throughout the trial corresponding to the ratio of bacteria to archaea (expressed as a percentage) on the left y-axis. (B) qPCR data of Bacterial and Archaeal 16S rRNA transcripts copy numbers (per g biomass) on the right y-axis from biomass samples (x-axis) throughout the trial corresponding to the ratio of bacteria to archaea (expressed as a percentage) on the left y-axis. were reproducible between the systems and no significant dif- The 16S rRNA transcripts (Fig. 1B) varied from 6.5 × 10 to 15 1 1.15 × 10 copies g− (21%–69%) for bacterial cells and 1.4 × ference was found between the systems (P > 0.05). Nextgener- 14 15 1 ation sequencing was carried out to identify the bacterial and 10 to 2 × 10 copies g− (31%–79%) for archaeal cells, these archaeal populations. The major bacterial populations identi- numbers were greater than those observed through DNA-based analysis. Deviations were again noted in the proportion of bac- fied included representatives of the Proteobacteria (8%–52%), mainly Deltaproteobacteria based on a DNA-based analysis and teria to archaea between the systems in biomass sampled from Day 236 (Phase 2a), Day 296 (Phase 2b), Day 531 (Phase 3b), Day Gammaproteobacteria, based on the cDNA-derived sequences (Fig. 2). The Synergistetes (1.5%–44%) and the Bacteroidetes, 666 (Phase 4-start of Phase 5), End and Filter End (Day 889). The greatest deviation was noted in the biomass sampled on Day mainly Flavobacteria, Sphingobacteria and Bacteroidea (0%– 52%) were also present in the reactors throughout the trial. Chlo- 531 (Phase 3b) in which R1 had the highest bacterial transcripts 15 1 recorded (1.2 × 10 copies g− ), comprising 62% of the total sam- roflexi (0%–19%), Firmicutes comprised mainly Clostridia and ple pool (Fig. 1B). This contrasted with the same time point in 14 1 R2, when bacterial copy numbers were 2.3 × 10 (copies g− ), comprising just 32% of the total sample pool (Fig. 1A). Despite these deviations, the bacterial and archaeal transcript numbers Downloaded from https://academic.oup.com/femsec/article/94/7/fiy095/5004848 by DeepDyve user on 14 July 2022 8 FEMS Microbiology Ecology, 2018, Vol. 94, No. 7 Figure 2. Taxa-plot of the percentage abundance of bacterial and archaeal phyla identified per sample. Samples are grouped according to phase ‘Initial’, ‘Phas e2’, ‘Phase 3’, ‘End’ and ‘Filter’. Bacilli (0%–24%). Less abundant, or occasionally present, bacte- Microbial community development over time rial groups included the Fusobacteria (0%–11%), the Actinobac- Community comparisons teria (Actinobacteria and Coriobacteria; 0%–24%), the Plancto- To follow the bacterial and archaeal community over time and to mycetes (Phycisphaerae and Planctomycetales; 0%–14%), Aci- compare the development of the mesophilic ‘seed’ within each dobacteria (Halophagae; 0%–6.5%), and <5% abundance; Cald- reactor, alpha-diversity matrices (Richness, Shannon, Simpson, iserica, Chlorobi, Gemmatimonadetes, Hyd24–12, Omnitroph- Alpha and Evenness) were compared at the SEQ level. Samples ica, Spirochaetae, Thermotogae, TM6, WD272, Verrucomicrobia from the ‘Seed’, ‘R1’ and ‘R2’ demonstrated similar observed and rare phyla; Candidate division SR1, Cyanobacteria, Defer- values (Figure S1, Supporting Information). No significant dif- ribacteres, Dictyoglomi, Elusimicrobia, Gracilibacteria, Fibrobac- ference was found between the ‘Seed’, ‘R1’ and ‘R2’; how- teres, Hydrogenedentes, Lentisphaerae, Nitrospirae, Parcubacte- ever, a large variation could be observed within the values per ria, SHA-109. group. Subsequently, PCoA was carried out at SEQ level using Heatmap analysis was employed to visualise temporal vari- unweighted Unifrac (ß-diversity metric) on the phylogenetic dis- ations in the bacterial populations in both reactors and simi- tance of sequences to visualise the similarities and dissimi- larity matrices were used in tandem (Fig. 3). In the heatmap of larities in the microbial communities. Figure 5 demonstrates the bacterial genera, the sequences clustered together based on that the sequences from the replicated reactors group together time, and DNA or cDNA origin. An exception to this was the R2 based on time—‘Seed’, ‘Phase 2b’, ‘Phase 3b’, ‘Filter’ and ‘End’ DNA biomass sample from the pumice filter unit, which formed rather than reactor origin. However, it must be noted that while a separate branch. This was due to the apparent increased abun- the sequences grouped together based on sampling period they dance presence of Commamonas and Candidatus Caldatribacterium were not found to be significantly different from each other. and this sample also demonstrated decreased species richness. As the PCoA data indicated that the samples clustered based The archaeal portion of the community was dominated on time period, sampled analysis of significant species con- by sequences identified as Methanosaeta concilii strain X16932 tributing to beta-diversity was carried out to identify what throughout the trial (Fig. 4). Methanobacterium, Methanolinea species were responsible for differences in these groupings. and Methanospirillum sequences were also present. Biomass Analysis of the significant species contributing to beta-diversity from ‘Phase 3b’ branched separately due to apparent decreases was carried out at genus level at a 2-log and 1-log fold change in hydrogenotrophic methanogens. for ‘Phase 2b’, ‘Phase 3b’, ‘Filter’ and ‘End’ whereby direct Downloaded from https://academic.oup.com/femsec/article/94/7/fiy095/5004848 by DeepDyve user on 14 July 2022 Keating et al. 9 Figure 3. Heatmap analysis for the bacterial throughout the trial showing the dominant genera (>2%) and Bray–Curtis similarity between samples and between the dominant genera. comparisons could be made between R1 and R2. The results identities of all significant SEQs are described in Table S3 (Sup- demonstrated that there were no significantly different species porting Information). between the replicate reactors at each of these phases (data not shown). This analysis was then repeated at SEQ level for the DISCUSSION reactor phases (Figure S2a–c, Supporting Information). In the case of ‘Phase 2b’ SEQ4, SEQ5, SEQ6 and SEQ139 were greater Though we have not tested real sewage, we have demon- in R1 (S2a, Supporting Information). SEQ4, 5 and 6 were found to strated a sufficient capacity of the microbial community for be a Methanosaeta concilii strain X16932 and SEQ139 were found sustained low-temperature degradation of a complex wastew- to be an uncultured Anaerolinaceae bacterium clone (Table S3, ater. Indeed, the removal efficiencies of these systems exceeded Supporting Information). In ‘Phase 3b’ a total of 24 SEQs were those reported in previous low-temperature trials carried out in significantly different between R1 and R2 (S2b, Supporting Infor- a traditional UASB [upflow anaerobic sludge bed reactor] (Ban- mation). Of these 19 were greater in R1 (SEQS 36, 46, 184, 159, dara et al. 2012). This study demonstrated that a mesophilic 17, 45, 50, 122, 210, 27, 111, 281, 19, 320, 53, 67, 103 and 92) and inoculum rapidly acclimated to psychrophilic conditions to 5 (SEQs 69, 414, 512, 546, and 430) were greater in the R2 sam- allow efficient COD removal to occur in both reactors. There ples. There was no significant difference between the commu- were indications of a capacity for enhanced bacterial activity at nities in R1 and R2 filter unit communities. There were only two 12 C, as evidenced by the protein hydrolysis assays. K values ◦ ◦ sequences that were significantly different between R1 and R2 throughout the trial increased at 12 C and decreased at 37 C, biomasses at the end of the trial (Figure S2c, Supporting Infor- indicating an increase in substrate affinity at lower tempera- mation). These were SEQ44 that was greater in R1 and SEQ235 tures. The literature that substrate affinity will decrease at lower that was greater in R2. SEQ44 was identified as an uncultured temperatures for psychrophiles, mesophiles and thermophiles Synergistetes bacterium and SEQ235 was identified as Chryseobac- (Nedwell 1999), but this often reflects only short-term studies. terium species strain SE19. Significant species was also used to Our results point towards the emergence of psychrophilic pro- assess the maturation of the granular biofilm and the species teolytic activity that was mirrored in both systems. While psy- contributing at a 1-log fold difference between the seed inocu- chrophilic microorganisms may not be crucial for successful lum and the R1 and R2 end biomass (Fig. 6). From this 25 SEQS remediation of waste streams from a process steering aspect, were identified as significantly different. SEQs 218, 165, 104, 275, the possibility to develop truly psychrophilic consortia could 280, 301, 378, 379, 338, 139, 265, 273, 341, 361, 381, 383, 401, 580, open important new opportunities for AD technology (Sekiguchi 436, 490 and 513 were more abundant in the seed inoculum. et al. 2001). With respect to the archaeal populations, SMA data While SEQs 107, 197 and 207 were more abundant in the biomass revealed that the microbial consortia became psychrotolerant upon take down of the reactors (Fig. 6; S3, Supporting Informa- for methanogenic substrates, rather than truly psychrophilic, a tion). Interestingly, SEQ107 was identified as a psychrotolerant finding commonly reported in the literature (Lettinga et al. 1999; species—Flavobacterium sinopsychrotolerans (Xu et al. 2011). The O’Flaherty, Collins and Mahony 2006). Our study demonstrates Downloaded from https://academic.oup.com/femsec/article/94/7/fiy095/5004848 by DeepDyve user on 14 July 2022 10 FEMS Microbiology Ecology, 2018, Vol. 94, No. 7 Figure 4. Heatmap analysis for the archaeal fraction throughout the trial showing the dominant sequences (>2%) and Bray–Curtis similarity between samples. that a psychrophilic or cold-adapted ‘seed’ was not necessary as suggesting that COD in the synthetic wastewater were read- Sus a starting inoculum for successful stable anaerobic digestion at ily degraded to COD , despite the absence of wastewater-borne Col low temperatures. Bowen et al. (2014) reported that a mesophilic lipases associated with non-synthetic wastewaters (Petropou- inoculum from an anaerobic suspended biomass sewage sludge los et al. 2017). SYNTHES was used so we could strictly define reactor was not successful for LtAD, but this biomass had much the influent and remove the variability associated with using lower SMA than the high-rate granular sludges used as inocula real sewage—to be sure the microbial community development here, and in previous successful LtAD trials (e.g. Collins, Mahony was not impacted by external variables. While SYNTHES car- and O’Flaherty 2006; Madden et al. 2013; Keating et al. 2016). It ries a similar proportion of particulate COD (31%) to real sewage is likely that the retention of the anaerobic biomass in hybrid (30%) [Aiyuk and Verstraete 2004] a disadvantage is however, sludge bed fixed-film reactors supported the development of that starch comprises the complex carbohydrate portion, which the reactor microbial community to function efficiently at lower may be easier to degrade than complex cellulosic materials that temperatures. Moreover, the trial lasted 889 days, which may would be present in real sewage. No accumulation of solids was have provided sufficient time for the maturation of cold-adapted observed in the granular sludge bed in agreement with previ- populations to allow for increased loading rates to be applied. ous work (Keating et al. 2016). The physical entrapment of solids This strategy for low-temperature sewage treatment offers a within the pumice matrix of the hybrid reactor may have facili- significant advantage over suspended biomass systems. In sus- tated subsequent degradation. pended biomass systems biomass washout would occur and the Successful high-rate AD is contingent on well-functioning microbial population may be more sensitive to immigration and microbial communities. Stable community structures are main- selective pressures of the influent (Vanwonterghem et al. 2014). tained through syntrophic interactions between the bacterial We have demonstrated that stable, long-term, high-rate and archaeal communities (Schnur ¨ er, Zellner and Svensson anaerobic digestion of a relatively complex wastewater, in the 1999). Low temperatures had been thought to limit these syn- form of synthetic sewage, was possible and even efficient, at trophic interactions (Kotsyurbenko 2005). However, the commu- low-operating temperature. Reactor performance data indicated nities represented in our systems were well balanced from the that the systems were functionally robust and stable, via the commencement of the trial, as indicated by negligible VFA accu- efficient effluent degradation with COD removal efficiencies mulation in the reactor effluents and the diverse bacterial and for COD and COD of >73% (Table 2), despite incremental archaeal populations found in the active fraction (cDNA-based Tot Sol increases in the OLR applied over the course of the trial. Per- analysis) throughout the trial. Interestingly, members of the haps surprisingly, we have also shown that under these con- Synergistetes were dominant members of the AD community in ditions hydrolysis was not rate-limiting at 12 Cwithevidence this trial. These are generally only found in frequencies of 1% or Downloaded from https://academic.oup.com/femsec/article/94/7/fiy095/5004848 by DeepDyve user on 14 July 2022 Keating et al. 11 Figure 5. PCoA plot based on unweighted Unifrac of DNA and cDNA sequences from R1 and R2 biomass samples. For each group, the legends are drawn at the mean value of the samples of that group. less in most AD systems (Godon et al. 2005), but in this study their however, identified between the two reactors, based on pro- abundance increased up to 44% of bacterial sequences (Fig. 2). cess performance despite there being no significant differences Isolated members of the Synergistetes partner syntrophic rela- between the communities at genus level. Firstly, an initial vari- tionships with the methanogens in the degradation of amino ation was observed upon start-up of the replicated systems. An acids with the production of VFAs (Vartoukian, Palmer and Wade immediate start-up was observed in R1 whereby all COD frac- 2007). Thus, the Synergistetes may be important for LtAD reactor tions were degraded, while the start-up of R2 took considerably function and may play a role in the low-temperature metabolism longer (∼56 days). While this variation was found to be signifi- of proteins observed in reactor biomass. Perhaps, the nature of cant, no definitive cause could be identified, as molecular sam- SYNTHES selected for a protein/amino acid degrading commu- pling was not carried out during the cold-adapting period so nity or their high prevalence in the seed inoculum (coming from as not to disturb initial community development. Considering a dairy treatment facility) allowed for their development in this that COD removal was similar and highly efficient in both Sol trial. reactors during phase 1 (pointing to efficient microbial activity), Adaptation involved a temporal shift in the microbial com- the cause may have reflected a greater potential for leaching of munity structure over the course of the study. However, the COD particles or the loss of flocculent biomass from R2. Sec- Sus replicate reactors maintained a remarkably similar microbial ondly, the commencement of Phase 5 led to a period of ∼5 weeks community profile to each other and this development was, in perturbation in R1, which was not mirrored in R2. qPCR data fact, reproducible down to genus level with no significant differ- from the start of this phase showed a 1-log reduction of the total ence found between the reactors at each phase. Indeed, in the bacterial and archaeal gene copy numbers were observed in R2 take-down biomass only two sequences were significantly dif- (Fig. 1A). Changes in the microbial community structure were ferent between the reactors (S3, Supporting Information). This missed at this time point; however, sequencing results prior to was mirrored in the reactor performance data, whereby the reac- this (from Phase 3) indicated that samples from this time point tors exhibited significant long-term reproducibility (889 days) clustered together and no significant difference was found at during treatment of the synthetic sewage substrate (Table 2). genus level. Fluctuations in COD removal rates generally occurred at similar As stated previously there were no significant changes (1 or points in both reactors, indicating that degradation was occur- 2-log) in the microbial populations present between reactors at ring through biological activity, rather than by physical entrap- each time period at genus level as demonstrated by significant ment of the COD fractions. Two divergences in behaviour were, species contributing to beta-diversity analyses. This statistical measurement indicated that time was the driver of microbial Downloaded from https://academic.oup.com/femsec/article/94/7/fiy095/5004848 by DeepDyve user on 14 July 2022 12 FEMS Microbiology Ecology, 2018, Vol. 94, No. 7 Figure 6. Significant SEQs contributing to beta-diversity at an SEQ level at a 1-log fold change was assessed between the seed community (Initial) and the biomas s taken from the end of the trial for both R1 and R2 (End). community structure rather than reactor identity. Comparisons activity is impacted by lower temperatures and that under these were then made at a sequence level in order to elucidate fur- conditions hydrogenotrophic methanogens dominate and facil- ther the species that were different between the systems and the itate efficient VFA degradation (Nozhevnikova et al. 2000; Collins species diverging from the ‘seed’ inoculum. From the sequences et al. 2005; Connaughton, Collins and O’Flaherty 2006). In Phase outlined in Table S3 (Supporting Information), it is worth not- 3, Anaerolinea-like species were reduced in R1 in comparison to ing sequences associated with granule formation and granule R2 (following this R1 demonstrated biomass washout upon com- integrity (Methanosaeta concilii species and Anaerolinea species). mencement of Phase 5). Biomass sampled from Phase 2 demon- Anaerolinea species dominated the Chloroflexi phyla in the reac- strated that Methanosaeta like species were reduced in R2 in com- tor systems. The Chloroflexi metabolise primary substrates in parison to R1 (prior to this phase R2 demonstrated reduced per- wastewater such as carbohydrates and cellular matter (Yamada formance). While this data are not conclusive, the close moni- et al. 2005). Anaerolinea species belong to Subphylum 1 an elusive toring of such species is crucial to granule integrity may provide phylum comprising environmental clones (Hugenholtz, Goebel an opportunity to link granule health with process performance and Pace 1998). They form web-like structures on the outside of and to develop means to promote their growth in poorly per- granules in mesophilic and thermophilic systems and thus are forming systems. thought to be important for granule structure (Sekiguchi et al. 1998). Given their stable dominance in these reactors further characterisation of their role in low-temperature systems would be valuable. Methanosaeta dominated the archaeal communities CONCLUSIONS in both systems as demonstrated by sequencing analysis (Figs 2 and 4; Table S3, Supporting Information). Methanosaeta con- Overall this study revealed that a cold-adapted or psychrophilic cilii is a key organism in granulation in these anaerobic systems ‘seed’ inoculum was not necessary for efficient LtAD of wastew- (Hulshoff Pol et al. 2004). The distinctive solely acetate utilis- ater. Our work demonstrated reproducible process performance ing acetoclastic Methanosaeta are known to dominate in steady and mirrored microbial community development between repli- state reactors in which acetate concentrations are low (McMa- cated systems. The nature of the reactor system allowed for hon et al. 2001). VFA analyses indicated that in-reactor acetate the retention of biomass allowing sufficient cold-adapted com- values were negligible throughout the trial. Moreover, acetoclas- munities to develop and mature, to the point where increased tic methanogens have been seen to be dominant at low temper- activity at low temperature developed within the hydrolytic and atures (Chin, Lukow and Conrad 1999). However, this is in con- methanogenic populations. Next generation sequencing identi- trast to reports by several authors that suggest that acetoclastic fied a number of possible cold-adapted species and increased Downloaded from https://academic.oup.com/femsec/article/94/7/fiy095/5004848 by DeepDyve user on 14 July 2022 Keating et al. 13 abundance of the Synergistetes and Anaerolinea phyla that war- Chin KJ, Lukow T, Conrad R. Effect of temperature on structure rant further targeted investigations to determine their possible and function of the methanogenic archaeal community in an future biotechnological relevance. anoxic rice field soil. Appl Environ Microbiol 1999;65:2341–9. Colleran E, Concannon F, Golden T et al. Use of methanogenic activity tests to characterise anaerobic sludges, screen for SUPPLEMENTARY DATA anaerobic biodegradability and determine toxicity thresh- olds against individual anaerobic trophic groups and species. Supplementary data are available at FEMSEC online. Water Sci Technol 1992;25:31–40. Collins G, Foy C, McHugh S et al. Anaerobic biological treatment of phenolic wastewater at 15–18 C. Water Res 2005;39:1614– ACKNOWLEDGEMENTS The authors thank Dr. Aoife Duff and Dr. Eoin Gunnigle for the Collins G, Mahony T, O’Flaherty V. Stability and reproducibility provision of culture strains. of low-temperature anaerobic biological wastewater treat- ment. FEMS Microbiol Ecol 2006;55:449–58. Connaughton S, Collins G, O’Flaherty V. Development of micro- FUNDING bial community structure and activity in a high-rate anaero- bic bioreactor at 18 C. Water Res 2006;40:1009–17. This work was supported by Science Foundation Ireland, DuBois M, Gilles KA, Hamilton JK et al. Colorimetric method for through a Charles Parsons Award (06 CP E006) and an Inves- determination of sugars and related substances. Anal Chem tigator Programme Grant (14/IA/2371); and the Irish Environ- 1956;28:350–6. mental Protection Agency (2014-W-LS-7). CJS is supported by Elmitwalli TA, Soellner J, De Keizer A et al. Biodegradability and Science Foundation Ireland Starting Investigator-COFUND fel- change of physical characteristics of particles during anaer- lowship (11/SIRG/B2159). UZI is funded by a NERC fellowship obic digestion of domestic sewage. Water Res 2001;35:1311–7. NE/L011956/1. Enright A-M, McGrath V, Gill D et al. Effect of seed sludge and Conflicts of interest. None declared. operation conditions on performance and archaeal com- munity structure of low-temperature anaerobic solvent- degrading bioreactors. Syst Appl Microbiol 2009;32:65–79. REFERENCES Fernandez ´ A, Huang S, Seston S et al. How stable is stable? Func- Aiyuk S, Verstraete W. Sedimentological evolution in an UASB tion versus community composition. Appl Environ Microbiol treating SYNTHES, a new representative synthetic sewage, 1999;65:3697. Foresti E, Zaiat M, Vallero M. Anaerobic processes as the core at low loading rates. Bioresour Technol 2004;93:269–78. Akarsubasi AT, Ince O, Kirdar B et al. Effect of wastewater technology for sustainable domestic wastewater treatment: consolidated applications, new trends, perspectives, and composition on archaeal population diversity. Water Res 2005;39:1576–84. challenges. Rev Environ Sci Biotechnol 2006;5:3–19. Girvan MS, Campbell CD, Killham K et al. Bacterial diversity pro- APHA. In: Clesceri LS, Greenberg AE, Eaton AD (eds). Standard Methods for the Examination of Water and Wastewater, 20th ed. motes community stability and functional resilience after perturbation. Environ Microbiol 2005;7:301–13. American Public Health Association, Washington, DC, 2005. Bandara WMKRTW, Kindaichi T, Satoh H et al. Anaerobic treat- Godon J-J, Moriniere J, Moletta M et al. Rarity associated with spe- cific ecological niches in the bacterial world: the ‘Synergistes’ ment of municipal wastewater at ambient temperature: analysis of archaeal community structure and recovery of example. Environ Microbiol 2005;7:213–24. dissolved methane. Water Res 2012;46:5756–64. Gouveia J, Plaza F, Garralon G et al. Long-term operation of a pilot Bialek K, Cysneiros D, O’Flaherty V. Low-Temperature (10 C) scale anaerobic membrane bioreactor (AnMBR) for the treat- anaerobic digestion of dilute dairy wastewater in an ment of municipal wastewater under psychrophilic condi- tions. Bioresour Technol, 2015;185:225–33. EGSB bioreactor: microbial community structure, popula- tion dynamics, and kinetics of methanogenic populations. Hugenholtz P, Goebel BM, Pace NR. Impact of culture- independent studies on the emerging phylogenetic view of Archaea 2013;10:1–10. Bialek K, Kumar A, Mahony T et al. Microbial community struc- bacterial diverity. J Bacteriol 1998;180:4765–74. Hughes D, Enright A-M, Mahony T et al. Novel Anaerobic Sewage ture and dynamics in anaerobic fluidized-bed and granular sludge-bed reactors: influence of operational temperature Treatment and Bioenergy Production: High-Rate Anaerobic Diges- tion as a Core Technology for Sustainable Treatment of Munici- and reactor configuration. Microb Biotechnol 2012;5:738–52. Bowen EJ, Dolfing J, Davenport RJ et al. Low-temperature limita- pal and Low-Strength Industrial Wastewaters. EPA STRIVE Report Series No. 64. Johnstown Castle: Environmental Protection tion of bioreactor sludge in anaerobic treatment of domestic wastewater. Water Sci Technol 2014;69:1004–13. Agency, 2011. Callahan BJ, McMurdie PJ, Rosen MJ et al. DADA2: high-resolution Hulshoff Pol LW, de Castro Lopes SI, Lettinga G et al. Anaerobic sample inference from Illumina amplicon data. Nat Methods sludge granulation. Water Res 2004;38:1376–89. 2016;13:581–3. Katoh K, Standley DM. MAFFT multiple sequence alignment software version 7: improvements in performance and Caporaso JG, Kuczynski J, Stombaugh J et al. QIIME allows anal- ysis of high-throughput community sequencing data. Nat usability. Molecular Biol Evoluti 2013;30:772–80. Keating C, Chin JP, Hughes D et al. Biological phosphorus removal Methods 2010;7:335–6. Caporaso JG, Lauber CL, Walters WA et al. Ultra-high-throughput during high-rate, low-temperature, anaerobic digestion of wastewater. Front Microbiol 2016;7:226. microbial community analysis on the Illumina HiSeq and MiSeq platforms. ISME J 2012;6:1621–4. Kotsyurbenko OR. Trophic interactions in the methanogenic microbial community of low-temperature terrestrial ecosys- Cavicchioli R. Microbial ecology of Antarctic aquatic systems. Nat Rev Microbiol 2015;13:691–706. tems. FEMS Microbiol Ecol 2005;53:3–13. Downloaded from https://academic.oup.com/femsec/article/94/7/fiy095/5004848 by DeepDyve user on 14 July 2022 14 FEMS Microbiology Ecology, 2018, Vol. 94, No. 7 Langenhoff AAM, Stuckey DC. Treatment of dilute wastewater Raskin L, Poulsen LK, Noguera DR et al. Quantification of using an anaerobic baffled reactor: effect of low temperature. methanogenic groups in anaerobic biological reactors by Water Res 2000;34:3867–75. oligonucleotide probe hybridization. Appl Environ Microbiol Lettinga G, Rebac S, Parshina S et al. High-rate anaerobic treat- 1994;60:1241–8. ment of wastewater at low temperatures. Appl Environ Micro- Rebac S, Ruskova J, Gerbens S et al. High-rate anaerobic treat- biol 1999;65:1696–702. ment of wastewater under psychrophilic conditions. JFer- Leven ´ L, Eriksson ARB, Schnur ¨ er A. Effect of process temperature ment Bioeng 1995;80:499–506. on bacterial and archaeal communities in two methanogenic Sakar S, Yetilmezsoy K, Kocak E. Anaerobic digestion technol- bioreactors treating organic household waste. FEMS Microbiol ogy in poultry and livestock waste treatment—a literature Ecol 2007;59:683–93. review. Waste Manage Res 2009;27:3–18. Lew B , Beliavski M, Admon S et al. Temperature effect on Schnur ¨ er A, Zellner G, Svensson BH. Mesophilic syntrophic UASB reactor operation for domestic wastewater treatment acetate oxidation during methane formation in biogas reac- in temperate climate regions. Water Science and Technology tors. FEMS Microbiol Ecol 1999;29:249–61. 2003;48:25–30. Sekiguchi Y, Kamagata Y, Harada H, Recent advances in Love MI, Huber W, Anders S. Moderated estimation of fold methane fermentation technology. Current Opinion in Biotech- change and dispersion for RNA-seq data with DESeq2. nology 2001;12:277–282. Genome Biol 2014;15:550. Sekiguchi Y, Kamagata Y, Syutsubo K et al. Phylogenetic diversity Lowry OH, Rosebrough NJ, Farr AL et al. Protein measurements of mesophilic and thermophilic granular sludges determined with the folin phenol reagent. J Biol Chem 1951;193:265–75. by 16S rRNA gene analysis. Annu Rev Microbiol 1998;144:2655– Madden P, Al-Raei AM, Enright A-M et al. Effect of sulfate on low- temperature anaerobic digestion. Front Microbiol 2013;5:376. Singh KS, Viraraghavan T. Impact of temperature on perfor- Madden PD, Chinalia FA, Enright AM et al. Perturbation- mance, microbiological and hydrodynamic aspects of UASB independent community development in low-temperature reactors treating municipal wastewater. In: Proceedings of Sev- anaerobic biological wastewater treatment bioreactors. enth Latin American Workshop and Symposium on Anaerobic Biotechnol Bioeng 2010;105:79–87. Digestion. Merida, Mexico, 2002, pp. 613–20. McKeown RM, Hughes D, Collins G et al. Low-temperature anaer- Smith AL, Skerlos SJ, Raskin L. Psychrophilic anaerobic mem- obic digestion for wastewater treatment. Curr Opin Biotechnol brane bioreactor treatment of domestic wastewater. Water 2012;23:444–51. Res 2013;47:1655–65. McMahon KD, Stroot PG, Mackie RI et al. Anaerobic codigestion Smith CJ, Nedwell DB, Dong LF et al. Evalution of quantitative of municipal solid waste and biosolids under various mix- polymerase chain reaction-based approaches for determin- ing conditions e II: microbial population dynamics. Water Res ing gene copy and gene transcript numbers in environmental 2001;35:1817–27. samples. Environ Microbiol 2006;8:804–15. McMurdie PJ, Holmes S. phyloseq: an R package for reproducible Suzuki MT, Taylor LT, DeLong EF. Quantitative analysis of small- interactive analysis and graphics of microbiome census data. subunit rRNA genes in mixed microbial populations via 5 - PLoS One 2013;8:e61217. nuclease assays. Appl Environ Microbiol 2000;66:4605–14. Nedwell D, Effect of low temperature on microbial growth: low- Van Lier JB, Tilche A, Ahring BK et al. New perspectives in anaer- ered affinity for substrates limits growth at low temperature. obic digestion. Water Sci Technol 2001;43:1–18. FEMS Microbiology Ecology 1999;30:101–11. Vanwonterghem I, Jensen PD, Ho DP et al. Linking microbial Neves L, Oliveira R, Alves MM. Influence of inoculum activity community structure, interactions and function in anaerobic on the bio-methanization of a kitchen waste under different digesters using new molecular techniques. Curr Opin Biotech- waste/inoculum ratios. Process Biochem 2004;39:2019–24. nol 2014;27:55–64. Nozhevnikova AN, Rebak S, Kotsyurbenko OR et al. Anaerobic Vartoukian SR, Palmer RM, Wade WG. The division ‘Synergistes’. production and degradation of volatile fatty acids in low tem- Anaerobe 2007;13:99–106. perature environments. Water Sci Technol 2000;41:39–46. Xing W, Zhao Y, Zuo JE. Microbial activity and community struc- O’Flaherty V, Collins G, Mahony T. The microbiology and bio- ture in a lake sediment used for psychrophilic anaerobic chemistry of anaerobic bioreactors with relevance to domes- wastewater treatment. J Appl Microbiol 2010;109:1829–37. tic sewage treatment. Rev Environ Sci Biotechnol 2006;5:39–55. Xu M, Xin Y, Tian J et al. Flavobacterium sinopsychrotolerans sp. nov., Oksanen JF, Blanchet G, Kindt R et al. Vegan: Community Ecology isolated from a glacier. Int J Syst Evol Microbiol 2011;61:20–4. Package, version 2013 Dec 12;2 (9) R package version 3.0 2013. Yamada T, Sekiguchi Y, Imachi H et al. Diversity, localiza- Petropoulos E, Dolfing J, Davenport RJ et al. Developing cold- tion, and physiological properties of filamentous microbes adapted biomass for the anaerobic treatment of domestic belonging to chloroflexi subphylum i in mesophilic and ther- wastewater at low temperatures (4, 8 and 15 C) with inocula mophilic methanogenic sludge granules. Appl Environ Micro- from cold environments. Water Res 2017;112:100–9. biol 2005;71:7493–503. Price MN, Dehal PS, Arkin AP. FastTree 2–approximately Yu Y, Lee C, Kim J et al. Group-specific primer and probe sets maximum-likelihood trees for large alignments. PLoS One to detect methanogenic communities using qpcr. Biotechnol 2010;5:e9490. Bioeng 2005;89:670–9.
Journal
FEMS Microbiology Ecology – Oxford University Press
Published: Jul 1, 2018
Keywords: biomass