Energy Conversion and Management 244 (2021) 114491

Contents lists available at ScienceDirect

Energy Conversion and Management

journal homepage: www.elsevier.com/locate/enconman

Pilot-scale biomethanation in a trickle bed reactor: Process performance

and microbiome functional reconstruction
Panagiotis Tsapekos a, Laura Treu b, *, Stefano Campanaro b, *, Victor B. Centurion c, Xinyu Zhu a,
Maria Peprah a, Zengshuai Zhang a, Panagiotis G. Kougias d, Irini Angelidaki a
a
Department of Chemical and Biochemical Engineering, Technical University of Denmark, Kgs., Lyngby DK-2800, Denmark
b
Department of Biology, University of Padova, Via U. Bassi 58/b, 35121 Padova, Italy
c
Microbial Resources Division, Research Center for Chemistry, Biology, and Agriculture (CPQBA), State University of Campinas - UNICAMP, Paulínia, SP CEP 13081-
970, Brazil
d
Soil and Water Resources Institute, Hellenic Agricultural Organisation Demeter, Thermi-Thessaloniki 57001, Greece

A R T I C L E I N F O A B S T R A C T

Keywords: Biogas upgrading is an emerging technology offering unique opportunities for further exploitation of biomethane
Biomethanation as fuel for vehicles or direct injection into the gas grid, expanding the conventional use of biogas for combined
Biogas upgrade heat and electricity generation. Up to date, most of the studies exploring the potential of biological carbon di­
Trickle-bed reactor
oxide hydrogenation was performed at laboratory scale systems, hampering the evaluation of the process under
Functional metagenomics
Archaea
real environmental conditions. The current work demonstrates the performance of a pilot trickle bed reactor that
was fed with real biogas as CO2 source under progressively increased gas provision rates. Additionally, the study
is supported by a genome-centric metagenomic analysis to gain deep insights into the microbiome of the reactor.
A maximum methane content of 98.5% was achieved at a gas retention time of 5 h. Stand-by periods in which no
influent gas was provided in the reactor did not lead to fatal deterioration of the overall process, as the bio­
methanation efficiency was recovered after a certain period of time. Samples obtained from three different layers
of the packing material, the liquid phase of the reactor and the inoculum demonstrated a distinct clustering of
microbial members. The provision of the nutrient media from the top layer led to the enrichment of specific
bacteria, such as Clostridiaceae DTU-pt_113, whose genome profile contains Veg-family genes, which are known
to be associated with biofilm formation. Similarly, the injection of influent gases from the bottom of the reactor
favoured the proliferation of hydrogenotrophic methanogens, solely belonging to family Methanobacteriaceae.

1. Introduction carrier to ease Power to X technologies.

Among the conversion technologies for the deployment of H2 to the
With a view of achieving a climate neutral Europe by 2050 and to­ energy market, the biological coupling with CO2 to produce biomethane
wards deep decarbonisation, European Commission has lately is a promising alternative [3,4]. Biomethanation can be easily imple­
announced the key role of H2 produced from excess renewables elec­ mented at biogas plants to upcycle the CO2 in biogas but the current
tricity [1]. The surplus electricity, produced from fluctuating renew­ techno-economic metrics (e.g., production capacity, switch on/off
ables during the peak periods, must be consumed to avoid losses or grid flexibility, CAPEX) should be improved to compete with conventional
congestion [2]. Although the temporary surplus can be partially CO2 capturing technologies (e.g., amine scrubbing, pressure swing
exported to neighbouring countries, the potential of renewables is not adsorption). However, the traditional upgrading processes do not
fully utilized. It is anticipated that smart grid approaches should be upcycle CO2 to CH4 leading to limited environmental benefit and also, at
adopted and effective solutions for storage and utilization of cheap en­ low H2 prices are less economically advantageous compared to bio­
ergy should be developed. Hence, excess electricity can be used to methanation [5].
produce H2 appearing as an energy carrier by storing electrons in the Among the different reactor systems for biomethanation, trickle bed
form of stable chemical bonds that can be used as a fuel or serve as a reactor (TBR) appears as an efficient system to achieve high CH4 quality

* Corresponding authors.
E-mail addresses: laura.treu@unipd.it (L. Treu), stefano.campanaro@unipd.it (S. Campanaro).

https://doi.org/10.1016/j.enconman.2021.114491
Received 11 May 2021; Accepted 29 June 2021
Available online 8 July 2021
0196-8904/© 2021 Elsevier Ltd. All rights reserved.
P. Tsapekos et al. Energy Conversion and Management 244 (2021) 114491

and production capacities [6]. In TBRs the microbes can be immobilized 2. Materials and methods
on the packing material which should have a high surface-area for
gas–liquid mass transfer favouring the high density and activity of 2.1. Pilot trickle bed reactor
methanogenic archaea [7]. Focusing on packing, polyurethane foam is a
dense and cheap material posing some characteristics to be used in A trickle bed reactor (TBR) with an overall height of 2.1 m and active
biological upgrading. Specifically, it was lately associated with high H2 filling volume of 68 L was used. The schematic diagram of the TBR and
conversion and CH4 production rates at lab-test [8]. Despite the prom­ ancillary equipment are depicted in Fig. 1. The reactor shell was con­
ising performance, in the cited study the tests ran only for 60 days and structed of AISI 304 (ø273 × 2 mm; ID 269). The active volume was
thus, the tolerance of polyurethane against potential operational prob­ separated in three sections filled with polyurethane foam which was
lems (e.g., filter clogging due to excess biofilm) cannot be concluded. supported by a polyester grid (h: 25 mm, ø260 mm) to ensure avoidance
Considering the exploitation of TBR with different packing materials, of packing material displacement.
long-term pilot experiments were previously conducted. For instance, A screw-in resistance thermometer (JUMO Type PT 100) was placed
Strübing et al. [9] operated a pilot-scale TBR that was filled with two in the middle of the TBR to monitor and control the temperature at
different commercially available packing materials for more than 300 thermophilic conditions (52 ± 1 ◦ C) using a heater cable (Fluoropolymer
days using pure H2/CO2 and artificial nutrient media. Subsequently, the over jacket over tinned copper braid, EMTS2-CF). Since the process is
same research group followed a more dynamic approach applying a exothermic, a cooling system was prepared to maintain the temperature
repetitive standby/restart operation (at 25 ◦ C and 55 ◦ C) to evaluate the of the TBR on the desired level. Hence, when heat was produced as
biomethanation efficiency of a demand/oriented operation [10]. advised by an online thermometer, water at ambient temperature was
Research showed that the repetitive standby at 55 ◦ C could markedly automatically recirculated using a peristaltic pump (Watson-Marlow,
deteriorate TBR restart due to high inactivation of hydrogenotrophic 600 series) through soft copper tubing which was wrapped around the
archaea and volatile fatty acids (VFA) accumulation. However, rela­ outer surface of the TBR. Inlet and outlet ports were available at the
tively short standby periods were applied (up to 8 days) at the controlled bottom and top layer of the TBR to provide flexibility for either co– or
thermophilic conditions. In contrast, the effect of prolonged standby on counter-flow operation. The operating pressure was monitored using an
TBR restart procedure is not clear and on top of it, the impact of non- analogue positive pressure gauge that was connected to the upper sec­
controlled temperature has not been evaluated before. Moreover, none tion of the active volume.
of the above-mentioned very informative studies was validated in rele­ Biogas was supplied using a peristaltic pump (Watson-Marlow, 600
vant environments upgrading real biogas and using digestate as a series) and hydrogen was provided by a mass flow controller (Bronk­
nutrient source. horst High-Tech BV). Gases were injected below the lowest level of
Apart from reactor design, the microbial species involved in bio­ TBR’s active volume via either craft-made perforated stainless-steel in­
methanation markedly determine process efficiency. The microbial jectors with 5 sets of holes (3 × ø3.5 mm per set) or ceramic membranes
community responsible for organic matter to methane conversion is made from silicon carbide (SiC) with a 0.5 μm pore size. Sieved digested
composed of many species and can change in abundance and composi­ municipal biowaste was used as a nutrient medium and trickled once per
tion according to the environmental conditions [11]. Revealing the di­ day from the nutrient sump (7.5 L working volume) to the top of the
versity and dynamics of hydrogen-fuelled microbial communities it reactor using a peristaltic pump (Watson-Marlow, 600 series). A cylin­
would be feasible to acquire significant knowledge on how to optimize drical drip tray with multiple drain holes was placed at the top of the
the operational conditions. Due to the technical difficulties associated reactor to ensure proper distribution of liquid medium.
with the isolation process of anaerobic microbes, metagenomic ap­ A pressure relief valve was placed on the top of the TBR for safety.
proaches are more frequently applied to determine the functional The complete set-up was assembled and installed in a 40-foot container
properties of microbes involved in biogas upgrading [12,13]. Meth­ in line with ATEX regulations. All above mentioned electrical and me­
anogenic microbiome is known for its intricate metabolic networks and chanical equipment was connected to an electrical panel and controlled
complex consortium structure. Previous studies showed that the online via LabView (National Instruments, USA).
microbiome is typically dominated by uncharacterized species, leading
to difficulties on the functional characterization of the key members. 2.2. Inoculation
Thus, the de-novo assembly of individual genomes through genome-
centric metagenomics provides significantly deeper insights into the A continuously stirred tank reactor (CSTR) with 9.0 L total and 7.5 L
functionality of methanogenic microbes. In addition, the reconstruction working volume was employed to prepare an enriched hydro­
of metabolic pathways in metagenome-assembled genomes (MAG) un­ genotrophic seed. The CSTR was operated at thermophilic conditions
veiled the potential interspecies interaction among microbes in a (52 ± 1 ◦ C), the typical operation for biogas plants in Denmark, using a
methanogenic community [14]. In particular, genome-centric in­ silicone thermal jacket and was further used in the pilot test as nutrients
vestigations can provide a better characterization of the functional roles sump (Fig. 1). For inoculum preparation, the working volume of the
of microbial species and more qualitative information in comparison to CSTR consisted of 90% with the effluent of a pilot-scale biogas reactor
marker-gene based predictive approaches [15]. Nevertheless, there is (working volume: 500 L) fed with municipal biowaste and 10% with
scant research on biomethanation reactors operated in relevant envi­ hydrogenotrophic inoculum derived from a lab-scale upgrading reactor
ronments. Considering that biomethanation will markedly aid the green [16]. The municipal biowaste was pretreated in a pulping facility [17]
transition of the bioenergy sector, deeper insights on microbial aspects and then, was anaerobically digested at a hydraulic retention time of 20
are needed under more realistic conditions compared to the mainly days under mesophilic conditions. The raw municipal biowaste had the
examined lab-scale trials. following characteristics (g/kg): 166.34 ± 36.49 total solids (TS),
In the present study, the capability of a pilot-scale biomethanation 140.16 ± 28.84 volatile solids (VS), 4.27 ± 0.98 total nitrogen (TN),
TBR is evaluated using real biogas and digested municipal biowaste as 0.76 ± 0.20, ammonium nitrogen (NH4-N) and 4.4 pH. The effluent of
nutrient sources in a relevant environment. The TBR was operated the biogas reactor was collected in storage tanks and before usage, it was
varying the operational conditions and gradually increasing the H2 sieved and filtered to avoid operational problems and clogging during
feeding rate to assess process performance at long-term. To evaluate trickling. The finally used digested municipal waste had the following
process flexibility, the TBR was put twice on standby mode at different characteristics (g/kg): 14.23 ± 1.85 TS, 9.94 ± 1.36 VS, 1.09 ± 0.09
temperatures to evaluate restart efficiency. Moreover, microbial samples NH4-N, and 7.09 pH. The content of essential trace elements (i.e. Fe, Ni,
collected from different heights in the TBR and analysed to reveal Co) for methanogenesis was also defined and is presented below.
microbiome distribution. The CSTR was flushed with an 80:20 (v/v) mixture of H2 and CO2,

2
P. Tsapekos et al. Energy Conversion and Management 244 (2021) 114491

Fig. 1. Schematic diagram of the pilot-scale trickle bed reactor. Gray dashed arrow represents the points of sampling and biofilm appearance. PL: low point; PM:
middle point; PH: high point.

while a gasbag filled with the same gas feedstock was connected at the day 234 using biogas as CO2 source and operated at 10 h GRT up to day
headspace. The gas was continuously circulated using a peristaltic pump 256 (P-8). Then, the GRT was adjusted to 5 h (P-9) and finally, the TBR
and injected into the working volume using a ceramic diffuser. Due to H2 was operated at 2 h GRT (P-10). The detailed plan of the operating
and CO2 coupling a pressure drop appeared and fresh gas feedstock was conditions is presented in Table 1 and the GRT was calculated based on
daily added. During the cultivation period, 10% of liquid volume was reactor volume (VTBR) and the total feeding gas (Fin), as shown in Eq. (1).
replaced twice a week to ensure adequate nutrients provision. The CSTR
VTBR
was operated for one month and subsequently, the enriched hydro­ GRT = (1)
Fin
genotrophic culture was used to inoculate the pilot TBR. Prior to TBR
start-up, 40 L of sieved digested biowaste was continuously trickled to From day 0 to 193, the container with the integrated TBR unit was
ensure wetting of packing material and ease microbial access to nutri­ placed in BIOFOS WWTP (Avedøre, DK) and was fed with a mixture of
ents. Then, the 7.5 L of enriched inoculum was loaded into the upper CH4/CO2 at 60/40 (v/v) as an artificial biogas (Air Liquide A/S,
surface of the TBR. Denmark). At the end of the second standby period (day 194), the unit
was operated at Lemvig Biogas (Midtjylland, DK) and fed with biogas
produced in the plant derived after desulfurization. Microaerobic
2.3. Operating conditions and monitoring removal of H2S is applied in the biogas plant and thus, residual O2 (<0.2
vol%, monitored as safety limit) could be occasionally available on the
The TBR was flushed with N2 for 30 min before start-up to ensure outlet side of the biological process. CH4 and CO2 content in the biogas
anaerobic conditions. During the first 8 days (P-1), the H2:CO2 feedstock was regularly defined to adjust the value of added H2. During the whole
was settled to a 2:1 ratio to avoid H2 overfeeding and rapid pH increase. experimental period pure H2 was used (ALPHAGAZ™ 1 H2 ≥ 99.999
Subsequently, the feedstock was adjusted to the stoichiometrically mol%, Airliquide) and digested municipal biowaste was utilized as a
needed 4:1 up to day 27 reaching a gas retention time (GRT) of 10 h (P- nutrients source. Before usage, the digestate was sieved to remove large
2). To evaluate the effect of “hot standby” period (i.e., no feeding and particles that could provoke clogging. Apart from standby periods, the
keeping unchanged the temperature) [6], the gas provision was ceased nutrient solution was trickled once per day at a rate of 1.2 L/min and
from day 28 to day 53 (P-3). Then, the TBR was fed again at the same allowed to drain by gravity to return to the sump. Changes of the gas
feed rate (P-4). The gas feeding rate was increased again at day 81 to feed rate, temperature variation and operation/standby periods
reach a 5 h GRT (P-5) and at day 110 the craft-made perforated stainless- throughout the entire experiment are depicted listed in Fig. 2.
steel injectors were replaced with Sic membrane as means to improve H2
diffusion (P-6). The operational conditions were kept unchanged up to
day 118. Then, a “cold standby” period (i.e., no feeding and ambient
temperature) was established for 75 days (P-7). The TBR was restarted at

3
P. Tsapekos et al. Energy Conversion and Management 244 (2021) 114491

Table 1
Operating conditions and characteristics.
Period Days T (oC) H2: CO2 Fin (L/ GRT Gas dispersion system Notes
CO2 source day) (h)

1 0–8 52 ± 1 2:1 Pure 115 14 Craft-made perforated stainless-steel injectors with 5 sets of holes (3 ×
2 9–27 4:1 163 10 ø3.5 mm per set)
3 28–53 52 ± 1 – – “hot standby”
period
4 54–80 52 ± 1 4:1 Pure 163 10
5 81–109 52 ± 1 4:1 Pure 327 5

6 110–118 52 ± 1 4:1 Pure 327 5 Ceramic membranes made from silicon carbide (SiC) with a 0.5 μm pore
7 119–193 Ambient – – 327 – size. “cold standby”
period
8 194–233 52 ± 1 4:1 Biogas 163 10
9 234–256 52 ± 1 4:1 Biogas 327 5
10 255–276 52 ± 1 4:1 Biogas 817 2

Fig. 2. Simplified operational scheme indicating the operational and standby periods and the corresponding reactor temperatures lower and upper, H2 and total gas
feed rates. Yellow circles represent the sampling dates for DNA extraction. In the cold standby period, the dashed red lines depict the minimum and maximum
ambient temperatures, while the straight line stands for the average.

2.4. Analytical methods and data analysis GmbH, Hilden, Germany) with minor modifications as previously
described [16]; an initial cleaning step with Phenol:Chloroform: Isoamyl
The standard methods for the examination of water and wastewater Alcohol (25: 24: 1) was implemented to increase the purity of the
were followed to measure TS, VS, pH, and ammonium nitrogen (NH4-N) extracted nucleic acids. Quality control on extracted DNA was done by
[18]. The VFA composition was determined using a GC-TRACE 1300 means of NanoDrop (Thermo Fisher Scientific, Waltham, MA, USA) for
(Thermo Scientific) as previously described by Khoshnevisan et al. purity and Qubit (Thermo Fisher Scientific, Waltham, MA, USA) for
(2018a). The effluent gas was quantified using the water displacement concentration.
method. Gas samples were daily collected in vacuum vials (Exetainer®
5.9 mL Evacuated Flat Bottom Vial) to monitor the biomethanation ef­ 2.6. Metagenomic sequencing and binning analysis
ficiency. A gas chromatograph GC-TRACE 1310 (Thermo Scientific) was
used to determine the methane content and production capacity. All Library preparation was performed using the Nextera DNA Flex Li­
measurements were measured in duplicate and results significance (p < brary Prep Kit (Illumina Inc., San Diego CA) and sequenced with the
0.05) was defined by analyses of variance (ANOVAs) and Student’s t-test Illumina NovaSeq 6000 platform (2 × 150, paired end) at the CRIBI
using the OriginPro 9.0.0 SR2 software (OriginLab Corporation, USA). Biotechnology Center sequencing facility (University of Padova, Italy).
Raw sequences were uploaded to the Sequence Read Archive (NCBI)
2.5. Microbial sampling and DNA extraction under the project ID PRJNA694528.
Reads were filtered using Trimmomatic v 0.39–1 tool [19] to remove
To explore the microbial diversity and dynamicity along the TBR adapters and trim out low-quality bases (Phred score ≤ 20). BBDuck
height, three samples for genomic DNA extraction were collected from software (BBTools, jgi.doe.gov/data-and-tools/bbtools) was used for
the high, middle, and low points of the filter (PH, PM, and PL, respec­ eliminating bacteriophage phi X 174 contamination sequences. Co-
tively). Moreover, liquid sample (Li) was also extracted from the nutri­ assembly was carried out using MegaHit (v1.2.9) tool [20] with
ents sump to define the planktonic microbiome, and inoculum “meta-large” option. Contigs shorter than 1 kbp rarely encode complete
microbiome was also characterized. Hence, five samples were analysed genes and they provide little information on gene functions, for this
in total, including the inoculum sample (In). The samples from the TBR reason they were removed. Bowtie2 (v2.2.6) [21] and SAMTool (v.1.10)
were collected at day 118 and used for metagenomic analysis. Genomic [22] were used to generate the contigs coverage profiles for the subse­
DNA extraction was performed using DNeasy PowerSoil® (QIAGEN quent binning approach. Metagenome Assembled Genomes (MAGs)

4
P. Tsapekos et al. Energy Conversion and Management 244 (2021) 114491

were identified with MetaBAT2 (v2.12.1) [23]. CheckM (v1.1.2–1) [24] A remarkably high CH4 concentration (98%) was detected at P-4. A
was used to assess quality (completeness and contamination) and rela­ deviation to the H2 flow was faced from day 72 to 75 reducing the
tive abundance of the MAGs in each sample. MAGs were classified as upgrading efficiency. Once the H2 flow was adjusted to the correct value,
high quality (≥90% completeness, ≤ 5% contamination), medium the biomethane content reached again more than 95%. The lowering of
quality (≥50% completeness, ≤ 10% contamination), low quality the GRT at 5 h (P-5) by doubling the feeding load affected negatively the
(≤50% completeness, ≤ 10% contamination), and bad quality (>10% biomethanation performance. Specifically, the CH4 was decreased to
contamination). The taxonomic classification of the MAGs was made by 58% after 4 days of operation and stabilized to 76% at day 110. In
GTDB-Tk (v1.0.2) [25]. Finally, the prediction of protein encoding genes parallel, a slight pH drop and acetate accumulation was observed, sug­
in each MAG was performed with Prodigal (v2.6.3) [26]. Genes func­ gesting the presence of homo-acetogenic bacteria (Fig. 3b). Via the
tional annotation and functional investigation was performed with Wood-Ljungdahl, the homo-acetogens convert CO2 to acetate causing
DRAM (v1.1.1) [27] and EggNOG (v5.0) tools [28]. pH decrease. In accordance, a recent study in an up-flow bio­
methanation reactor filled with packing material found that the partial
utilization of CO2 and H2 for homo-acetogenesis instead of methano­
2.7. Statistics and data visualization
genesis led to pH reduction [34].
Another factor that might affected negatively the biomethanation
MAGs relative abundance was visualized as heatmap using R (v3.6.3)
process is the concentration of micronutrients that were found to be
(R core team, 2020) and the heatmaps of heatmaply (v1.1.1) package
significantly decreased during P5. Micronutrients, such as Ni, Co and Fe,
[29]. MAGs distribution in samples was checked using the Principal
are essential for basal microbial functions. It is well known that hy­
Component Analysis (PCA) of vegan (v2.5–6) package [30]. Out of the
drogenases are Ni-Fe enzymes and efficient cytochromes biosynthesis
DRAM tool results, only the Kyoto Encyclopedia of Gene and Genomes
requires iron within a defined concentration range. Additionally, it has
(KEGG) module database [31] and pathways of importance for biogas
been previously reported that low content of micronutrients has a
production were selected and visualized as heatmaps with ggplot2
detrimental role on the methanogenic activity of anaerobic biofilms
(v3.3.2) package [32]. Manual investigation of specific KEGG IDs was
[35]. In the current experiment, the concentrations of Co and Ni were
performed by colouring the genes in the pathway maps using KEGG
significantly lower (6 and 10 μM, respectively) at the beginning of P-5
mapper, and the metabolic reconstruction of MAGs was visually repre­
(day 80) compared to the beginning of the test (21 and 37 μМ, respec­
sented with BioRender (https://biorender.com/). Identification of ge­
tively) to. In accordance with the current work, the limited ability of
nomes belonging to the same MAGs previously reported was performed
archaea to perform a high biomethanation rate along with VFA accu­
with Average Nucleotide Identity calculation in the MiGA platform [33]
mulation was previously revealed due to shortage of Co and Ni [36].
using as reference the Bio-Gas microbiome database.
Similarly, the concentration of Fe in the nutrients’ sump was also
reduced from 302 to 51 μM compared to day 0 (Table 2). Based on
3. Results and discussion
literature, the optimal levels for Ni and Fe for hydrogenotrophic meth­
anogenesis are 0.2–1 and 15–500 μM, respectively [37]. Considering
3.1. Bioconversion of H2 and CO2 to CH4
that the micronutrient concentration was above the literature threshold
levels, even though they were found to be decreased, the strategy on
High biomethanation efficiency was achieved since the first week of
nutrients provision did not change through the experiment. Thus, it was
operation reaching more than 90% in the output (Fig. 3a). After feed
hypothesized that the limited biomethanation efficiency could be
regime adjustment at the stoichiometrically optimum at day 8, a slight
attributed to the fact that polyurethane foam was very thick filling
drop of upgraded methane was detected from day 10 to 18. However, at
material, and for these reasons the contact between nutrients and gases
the end of P-2 and prior to “hot standby” period, no residual H2 was
was not optimized leading to a non-homogenous biofilm creation
detected in the output, and the CH4 content was equal to 90%.

Fig. 3. Biomethane content (a), TVFA and acetic acid accumulation, NH4-N and pH fluctuation (b) of TBR.

5
P. Tsapekos et al. Energy Conversion and Management 244 (2021) 114491

Table 2 microbiome heatmap coverage shows nearly identical profiles of sam­

Content of crucial trace elements for enzymatic activity of methanogenic ples PL and PM (Supplementary Fig. 1), highlighting that the commu­
archaea. nities present in the medium–low part of the reactor were highly
Day Fe, μM Co, μM Ni, μM homogeneous. On the contrary, the remaining samples (In, Li, and PH)
0 302 21 37
had clearly distinct profiles and clustered individually. Firmicutes (n =
53 54 6 11 81) was the dominant phylum in all the samples and was represented
80 51 6 10 mainly by members of the class Clostridia (n = 37) (Supplementary
117 37 4 8 Table 1). Bacteroidetes (n = 15) and Proteobacteria (n = 11) were the
194 42 6 10
second and third most presented phyla, respectively. The methanogenic
234 31 5 9
256 43 8 12 archaea were represented by three MAGs assigned to the Meth­
276 53 10 16 anobacteriaceae family (Methanothermobacter wolfeii DTU-pt_43, Meth­
anobacterium DTU-pt_46, Methanobacteriaceae DTU-pt_063). For the sake
of clarity, only the ten more abundant MAGs of each sample were
throughout the TBR bed. This hypothesis was validated during sample selected and visualized in the heatmap (Fig. 4a). Clostridia DTU-pt_99
collection for microbiological analysis showing thicker biofilm forma­ was dominant in the inoculum (25%) and it was the third most abun­
tion at the top of the TBR, the position from where the nutrients were dant (6%) in the planktonic sample, followed by Firmicutes DTU-pt_168
trickled (Fig. 1). To overcome limited biomethanation, the perforated and Chromatiales DTU-pt_172 (both at 11%). Consequently, Clostridia
stainless-steel injectors were replaced with SiC membrane at day 110 as DTU-pt_99 was the main responsible for the bioprocess occurring in In
means to improve the gas–liquid contact. Indeed, the positive impact and Li samples (Fig. 4b). On the contrary, Clostridiaceae DTU-pt_113 was
was quickly revealed and biomethane content reached a value of 98% at the dominant microorganism (19%) in the PH sample, clearly demon­
day 110. In parallel, the acetate levels were not significantly increased strating a stratification of microbial members throughout the length of
(from 1732 to 1747 mg/L, p > 0.05), validating the improved hydro­ the TBR. Methanobacterium DTU-pt_46 was the dominant microorganism
genotrophic activity compared to the previous period. in the samples PL and PM, (20%) and the only MAG clearly associated
Subsequently, the “cold standby” followed, and after 75 days at with the medium–low part of the reactor, thus the main methanogen of
ambient temperature, the TBR initiated at 10 h GRT using real biogas. A the whole system. The other two identified methanogens were found in
small leakage did not allow proper H2 provision, and thus, almost 30 significantly lower relative abundance; specifically, Meth­
days were needed to adjust the feeding regime at the appropriate stoi­ anothermobacter wolfeii DTU-pt_43 was the second most abundant (0.8%
chiometry. Once the problem was fixed, more than 90% CH4 was ach­ average), while the abundance of Methanobacteriaceae DTU-pt_063 was
ieved similarly to the previous experimental period at 10 h GRT. less than the minimal threshold considered. Focusing on Meth­
Subsequently, by further increasing the feeding rate (P-9, GRT: 5 h) anothermobacter spp., their ability to efficiently form biofilm on ceramic
more than 98.5% CH4 content was observed which was the superior packing materials and then, couple H2 and CO2 for CH4 production was
upgrading efficiency during the whole experiment. Finally, by adjusting lately revealed [39]. Nevertheless, this recent study followed a pure
the feeding rate to the highest value, i.e., GRT: 2 h at P-10 a clear drop on culture gas fermentation which can be more susceptible to process dis­
biomethanation efficiency was detected, along with a concomitant drop turbances compared to the mixed culture of the present study. More­
in pH values below 8.0. This can be explained as the partial CO2 pressure over, Methanothermobacter spp. were also found to dominate the biogas
is in balance with the dissolved H2CO3 and the pH fluctuation is the upgrading community in up-flow system equipped with ceramic mem­
result of a bicarbonate buffer system. During efficient hydrogenotrophic branes [40].
methanogenesis, CO2 is coupled with H2 and the increased pH favours The 35 most abundant MAGs have, in general, the potential to
HCO–3 formation as shown in Eq. (2), while Eq. (3) is mainly favoured degrade carbohydrate through pentose phosphate, Emden-Meyerhof
above pH of 10. (EM), Wood-Ljungdahl (WL) pathways and citrate cycles (Fig. 5a) and,
H2CO3 → HCO−3 + H+ (2) consequently, to produce acetate and acetyl-CoA (Fig. 5b). The meta­
bolic annotation revealed a high abundance of putative homo-
HCO3- ↔ CO2−
3
+
+H (3) acetogenic bacteria (e.g., DTU-pt_99 and DTU-pt_113), which can use
H2 and CO2 and produce acetate through the WL pathway (Supple­
Nevertheless, the CO2 uptake rate is enhanced at alkaline conditions.
mentary Table 2). Despite difficulties in correlating gene content with
In accordance, the upgrading efficiency increased to the highest stan­
acetogenic behaviour, this finding, obtained from functional annotation,
dards (>95% CH4) when the pH increased to similar levels as before and
is a possible explanation for the acetate accumulation (1747.81 mg/L)
the major dissociated form of CO2 was HCO−3 . Despite optimal range for
observed after 80 days of operation (Fig. 3b). The high abundance of
AD process is considered from 7.0 to 8.0, the ex-situ methanogenesis
these bacteria agrees with previous findings in reactors fed with H2 and
typically occurs in a more alkaline environment due to external H2
CO2, and a syntrophic relationship with methanogenic archaea was
provision and CO2 uptake. Ashraf et al. found that the production ca­
previously hypothesized [41,42]. The three most abundant MAGs pre­
pacity of TBR was not decreased even at pH values higher than 8.5
sent in the planktonic sample (DTU-pt_99, DTU-pt_168 and DTU-pt_172)
showing tolerance of the community at more alkaline conditions [38]. In
have similar metabolic pathways, and this suggests a shared functional
the present study, the H2/CO2 fueled microbiome was also efficiently
behaviour. The only difference is in the Cytochrome complex, which is
functioning at such pH levels.
complete in Chromatiales DTU-pt_172, a purple sulfur bacteria of the
class Gammaproteobacteria characterized by reduced sulphur compounds
3.2. Microbiome in biogas upgrading TBR as electron donors [43].
To overcome the issue derived for incomplete MAG genomes, the
3.2.1. General microbial community profile most relevant MAGs identified according to PCA analysis (Fig. 4b) with
The genome-centric metagenomic analysis applied to all microbial values lower than 90% were compared with the MiGA Bio-Gas Micro­
samples generated 191 MAGs out of which 48 were of high quality and biome database through ANI analysis. Clostridia DTU-pt_99 and Clos­
75 of medium quality. Only the MAGs with high and medium quality (n tridiaceae DTU-pt_113 have high similarity with Firmicutes sp.
= 123) were retained for further analyses, such as statistics and bioin­ AS24abBPME_73 (99% ANI, 98.1% of completeness) and Clostridia sp.
formatics functional investigation. According to the percentage of AS23ysBPME_31 (98% ANI, 56.6% of completeness), respectively. Thus,
mapped reads (66 to 75% depending on the sample), the selected MAGs the more complete reference genome was used to reconstruct the
accounted for more than half of the total community. The global metabolic pathway (Fig. 6); this was done under the assumption that

6
P. Tsapekos et al. Energy Conversion and Management 244 (2021) 114491

Fig. 4. Selected MAGs abundance and distribution. (a) Heatmap of 35 high abundant MAGs. The taxon colors are the information of phylum level. (b) Principal
component analysis (PCA) of the different samples and MAGs. In: Inoculum; Li: Liquid; PL: low point; PM: middle point; PH: high point.

MAGs having high similarity at genome level were also sharing most methanogenesis and homo-acetogenesis are more energetically favor­
functional pathways. able compared to acetate oxidation at areas close to H2 supply and
proximity [46]. Thus, it is expected that the H2 is locally utilized for
3.2.2. Microbial spatial stratification these two processes, while syntrophic acetogenesis and homo-
As previously mentioned, the microbial community structure was acetogenic oxidation can occur in areas with lower H2 availability. At
divided in distinct clusters within the TBR. The microbial spatial strat­ a micro-scale methanogenic membrane reactor, it was found that
ification was attributed to two main factors as a result of the process hydrogenotrophic methanogenic biofilm was located above the gas
operation. The first factor is associated with the nutrient provision that permeable membrane while the syntrophic acetogens and homo-
was facilitated from the top of the reactor enabling the formation of a acetogenic oxidizers were mainly functional in the bulk phase away
thick biofilm with high affinity at the upper part of the packing material. from the H2 supplementation area [47]. In a higher scale, a similar
The biofilm was getting progressively thinner while moving to the lower stratification was observed herein. The relative abundance of hydro­
parts of the packing material. This observation is aligned and supported genotrophic methanogens was markedly high in the zones close to the
with the abundance profile of Clostridiaceae DTU-pt_113 which was added H2 (samples PL and PM), while potential homo-acetogens (Clos­
found to be greater enriched in PH samples rather than in PM and PL tridiales DTU-pt_099) were also detected in these samples. On the other
samples (Fig. 4). The sequenced data identified the presence of the gene hand, acetogenic Clostridiaceae DTU-pt_113 were found in the upper
family VEG (PF06257), which is associated with biofilm formation, in zone of the TBR where the available H2 was significantly lower than the
the genome of Clostridiaceae DTU-pt_113. More specifically, this family middle and low zones. As mentioned above, the utilized packing mate­
is highly conserved in gram-positive bacteria and has the function of rial did not allow a proper gas distribution and nutrients trickling
stimulating the production of the amyloid fibre component of the bio­ through reactor bed. To improve the homogeneity of microbial distri­
film [44]. The second factor for the spatial stratification was attributed bution, alternative packing materials of high surface area and porosity
to the provision of H2 and CO2, which were both injected from the lower could be helpful. In addition, continuous trickling can contribute on
part of the reactor. Such environmental conditions favoured the prolif­ higher nutrient availability compared to the applied once per day
eration of methanogens especially at PL and further to PM samples trickling. However, high trickling rates can allocate the packing mate­
indicating that methane was produced, mainly, in the low and middle rials and vigorously mix the community interrupting the syntrophic
points of the TBR (Fig. 4). It was found that methanogenesis was associations which are important for mixed fermentation as shown in
accomplished mainly by Methanobacterium DTU-pt_46; in fact, the different reactor systems [48]. On the other hand, absence of mixing can
taxonomic classification agreed with the presence of the complete clearly lead to stratification of full-scale digesters [49]. Thus, a contin­
pathway of hydrogenotrophic methane production identified in the uously slow trickling rate can be beneficial for biomethanation ensuring
genome of this MAG (Fig. 6). It must be noted that another microor­ nutrients availability and decreasing stratification.
ganism with high relative abundance in PL and PM samples was Clos­
tridiaceae DTU-pt_99. This MAG was the dominant microbial member in
inoculum (In) and it would have been expected that its abundance 3.3. Future perspective
would remain stable independently from the sampling point. Nonethe­
less, as shown in Fig. 4, the coverage profile of Clostridiaceae DTU-pt_99 Microbial analysis showed that the native AD microbiome can be
was correlated with the corresponding one of Methanobacterium DTU- naturally reconstructed and form a highly efficient community able to
pt_46, suggesting a putative syntrophic relationship due to this operate with raw biogas and nutrients from the digestate. The outcome
frequent co-occurrence. In fact, Clostridiaceae DTU-pt_99 belongs to the of this study is particularly important showing that the suggested
same species of unclassified Bacteria sp. DTU645 (>99.5% ANI) that was technology offers an alternative to other industrial upgrading processes
dominant in the microbiome together with several hydrogenotrophic relying on single-celled methanogenic archaea as biocatalyst and pure
archaea [45]. It has been previously reported that this bacterium per­ media for nutrients provision. Considering that the present study
forms an alternative pathway for acetate degradation to CO2 through demonstrated biomethanation using a pilot-scale reactor at relevant
reversed glycine reduction [45]. environment, system analysis of the whole concept should be now
Based on the thermodynamics, the hydrogenotrophic conducted. The feasibility of biomethanation compared to conventional
methods for biogas cleaning (i.e., water and amine scrubbing) should be

7
P. Tsapekos et al. Energy Conversion and Management 244 (2021) 114491

Fig. 5. Heatmap of DRAM results for 35 highly abundant MAGs. (a) Completeness level of KEGG modules functions is reported as percentage. (b) Short-chain fatty
acid (SCFA) and alcohol conversion functions are reported as absent (“0′′ - white) and present (“1” - blue).

evaluated. Since CO2 is biologically upcycled in the existing technology 4. Conclusions

to produce more energy, a more environmentally friendly approach is
achieved. This advantage must be confirmed via detailed techno- The current work evaluated for the first time the process perfor­
economic assessment which can also help the further up-scaling to in­ mance and the microbial spatial distribution of a pilot trickled bed
dustrial applications. Finally, business plan can further help to reveal the reactor for biomethanation using real biogas as CO2 source. Maximum
need for different kinds of subsidies and incentives to increase feasibility biomethanation efficiency led to an output gas with a methane content
of the developed biomethanation technology. When sustainability is of 98.5%. Additionally, genome-centric metagenomics revealed a strong
validated, biomethanation can be further exploited in other industries as stratification of the microbial community which is coherent with the
for example fermentation sites, biorefineries, cement, and lime in­ metabolic functions occurring in different layers. Methanogens and
dustries to capture the produced CO2. The potential usage of the native potential syntrophic bacteria proliferated close to the influent gas in­
AD microbiome in TBR will allow carbon-intensive industries to jection ports. On the contrary, microbes whose genome profile contains
improve environmental footprint, enhance productivity and profit­ genes associated with biofilm formation and stimulation dominated the
ability. Biomethanation can be a significant contributor to the green upper layer of the packing material, most probably due to the rich-
transition. nutrient media that was trickling from the top of the reactor. The

8
P. Tsapekos et al. Energy Conversion and Management 244 (2021) 114491

Fig. 6. Metabolic reconstruction of three selected MAGs based on results obtained from KEGG database. The cells represent the H2 and CO2 utilization pathways for
the production of acetate (green color: WL pathway incomplete, blue color: WL pathway complete) and methane (purple color). Histograms represent the selected
MAGs abundance in each sample. In: Inoculum; Li: Liquid; PL: low point; PM: middle point; PH: high point.

results from the current work are essential and can be directly exploited References
for elevating the technological maturity of biological biogas upgrading
process. [1] Communication COM/2020/301: A hydrogen strategy for a climate-neutral Europe
| Knowledge for policy n.d. https://knowledge4policy.ec.europa.eu/publication
/communication-com2020301-hydrogen-strategy-climate-neutral-europe_en
CRediT authorship contribution statement (accessed January 21, 2021).
[2] Szima S, Nazir SM, Cloete S, Amini S, Fogarasi S, Cormos A-M, et al. Gas switching
reforming for flexible power and hydrogen production to balance variable
Panagiotis Tsapekos: Investigation, Data curation, Visualization, renewables. Renew Sustain Energy Rev 2019;110:207–19. https://doi.org/
Writing – original draft. Laura Treu: Data curation, Supervision, 10.1016/j.rser.2019.03.061.
Writing – review & editing. Stefano Campanaro: Resources, Writing – [3] Angelidaki I, Treu L, Tsapekos P, Luo G, Campanaro S, Wenzel H, et al. Biogas
upgrading and utilization: Current status and perspectives. Biotechnol Adv 2018;
review & editing. Victor B. Centurion: Software, Methodology, Data 36:452–66.
curation, Writing – original draft. Xinyu Zhu: Investigation, Writing – [4] Alibardi L, Green K, Favaro L, Vale P, Soares A, Cartmell E, et al. Performance and
review & editing. Maria Peprah: Investigation, Methodology, Data stability of sewage sludge digestion under CO2 enrichment: A pilot study. Bioresour
Technol 2017;245:581–9. https://doi.org/10.1016/j.biortech.2017.08.071.
curation. Zengshuai Zhang: Investigation, Methodology, Data curation.
[5] Vo TTQ, Wall DM, Ring D, Rajendran K, Murphy JD. Techno-economic analysis of
Panagiotis G. Kougias: Conceptualization, Data curation, Writing – biogas upgrading via amine scrubber, carbon capture and ex-situ methanation.
review & editing. Irini Angelidaki: Funding acquisition, Resources, Appl Energy 2018;212:1191–202.
Conceptualization, Methodology, Writing – review & editing. [6] Strübing D, Moeller AB, Mößnang B, Lebuhn M, Drewes JE, Koch K. Load change
capability of an anaerobic thermophilic trickle bed reactor for dynamic H2/CO2
biomethanation. Bioresour Technol 2019;289:121735. https://doi.org/10.1016/j.
Declaration of Competing Interest biortech.2019.121735.
[7] Dupnock TL, Deshusses MA. High-performance biogas upgrading using a
biotrickling filter and hydrogenotrophic methanogens. Appl Biochem Biotechnol
The authors declare that they have no known competing financial 2017;183:488–502.
interests or personal relationships that could have appeared to influence [8] Sieborg MU, Jønson BD, Ashraf MT, Yde L, Triolo JM. Biomethanation in a
the work reported in this paper. thermophilic biotrickling filter using cattle manure as nutrient media. Bioresour
Technol Rep 2020;9:100391. https://doi.org/10.1016/j.biteb.2020.100391.
[9] Strübing D, Huber B, Lebuhn M, Drewes JE, Koch K. High performance biological
Acknowledgements methanation in a thermophilic anaerobic trickle bed reactor. Bioresour Technol
2017;245:1176–83.
[10] Strübing D, Moeller AB, Mößnang B, Lebuhn M, Drewes JE, Koch K. Anaerobic
We acknowledge financial support from projects BioUpgrade (For­ thermophilic trickle bed reactor as a promising technology for flexible and
skEL 2016-1-12465), eFUEL (EUDP 64018-0559), VARGA (MST-141- demand-oriented H2/CO2 biomethanation. Appl Energy 2018;232:543–54.
01377), FUBAF (MST-117-00508) and São Paulo Research Foundation - [11] Ács N, Szuhaj M, Wirth R, Bagi Z, Maróti G, Rákhely G, et al. Microbial community
rearrangements in power-to-biomethane reactors employing mesophilic biogas
FAPESP (Process no. 2019/22891-0).
digestate. Front Energy Res 2019;7:132. https://doi.org/10.3389/
fenrg.2019.00132.
Appendix A. Supplementary data [12] Fontana A, Kougias PG, Treu L, Kovalovszki A, Valle G, Cappa F, et al. Microbial
activity response to hydrogen injection in thermophilic anaerobic digesters
revealed by genome-centric metatranscriptomics. Microbiome 2018;6. https://doi.
Supplementary data to this article can be found online at https://doi. org/10.1186/s40168-018-0583-4.
org/10.1016/j.enconman.2021.114491. [13] Kougias PG, Campanaro S, Treu L, Zhu X, Angelidaki I. A novel archaeal species
belonging to Methanoculleus genus identified via de-novo assembly and

9
P. Tsapekos et al. Energy Conversion and Management 244 (2021) 114491

metagenomic binning process in biogas reactors. Anaerobe 2017;46:23–32. [32] Wickham H. ggplot2. New York, NY: Springer New York; 2009. https://doi.org/
https://doi.org/10.1016/j.anaerobe.2017.02.009. 10.1007/978-0-387-98141-3.
[14] Basile A, Campanaro S, Kovalovszki A, Zampieri G, Rossi A, Angelidaki I, et al. [33] Rodriguez-R LM, Gunturu S, Harvey WT, Rosselló-Mora R, Tiedje JM, Cole JR,
Revealing metabolic mechanisms of interaction in the anaerobic digestion et al. The Microbial Genomes Atlas (MiGA) webserver: taxonomic and gene
microbiome by flux balance analysis. Metab Eng 2020;62:138–49. https://doi.org/ diversity analysis of Archaea and Bacteria at the whole genome level. Nucleic Acids
10.1016/j.ymben.2020.08.013. Res 2018;46:W282–8. https://doi.org/10.1093/nar/gky467.
[15] Campanaro S, Treu L, Kougias PG, Zhu X, Angelidaki I. Taxonomy of anaerobic [34] Kougias PG, Tsapekos P, Treu L, Kostoula M, Campanaro S, Lyberatos G, et al.
digestion microbiome reveals biases associated with the applied high throughput Biological CO2 fixation in up-flow reactors via exogenous H2 addition.
sequencing strategies. Sci Rep 2018;8:1–12. J Biotechnol 2020;319:1–7. https://doi.org/10.1016/j.jbiotec.2020.05.012.
[16] Treu L, Kougias PG, de Diego-Díaz B, Campanaro S, Bassani I, Fernández- [35] Cresson R, Carrère H, Delgenès JP, Bernet N. Biofilm formation during the start-up
Rodríguez J, et al. Two-year microbial adaptation during hydrogen-mediated period of an anaerobic biofilm reactor—Impact of nutrient complementation.
biogas upgrading process in a serial reactor configuration. Bioresour Technol 2018; Biochem Eng J 2006;30:55–62.
264:140–7. https://doi.org/10.1016/j.biortech.2018.05.070. [36] Cresson R, Dabert P, Bernet N. Microbiology and performance of a methanogenic
[17] Tsapekos P, Kougias PG, Kuthiala S, Angelidaki I. Co-digestion and model biofilm reactor during the start-up period. J Appl Microbiol 2009;106:863–76.
simulations of source separated municipal organic waste with cattle manure under [37] Glass J, Orphan VJ. Trace metal requirements for microbial enzymes involved in
batch and continuously stirred tank reactors. Energy Convers Manage 2018;159: the production and consumption of methane and nitrous oxide. Front Microbiol
1–6. https://doi.org/10.1016/j.enconman.2018.01.002. 2012;3:61.
[18] Federation WE, Association APH. Standard methods for the examination of water [38] Ashraf MT, Sieborg MU, Yde L, Rhee C, Shin SG, Triolo JM. Biomethanation in a
and wastewater. Am Public Health Assoc APHA Wash DC USA 2005. thermophilic biotrickling filter — pH control and lessons from long-term operation.
[19] Bolger AM, Lohse M, Usadel B. Trimmomatic: a flexible trimmer for Illumina Bioresour Technol Rep 2020;11:100525. https://doi.org/10.1016/j.
sequence data. Bioinformatics 2014;30:2114–20. https://doi.org/10.1093/ biteb.2020.100525.
bioinformatics/btu170. [39] Thema M, Weidlich T, Kaul A, Böllmann A, Huber H, Bellack A, et al. Optimized
[20] Li D, Liu C-M, Luo R, Sadakane K, Lam T-W. MEGAHIT: an ultra-fast single-node biological CO2-methanation with a pure culture of thermophilic methanogenic
solution for large and complex metagenomics assembly via succinct de Bruijn archaea in a trickle-bed reactor. Bioresour Technol 2021;333:125135. https://doi.
graph. Bioinforma Oxf Engl 2015;31:1674–6. https://doi.org/10.1093/ org/10.1016/j.biortech.2021.125135.
bioinformatics/btv033. [40] Corbellini V, Kougias PG, Treu L, Bassani I, Malpei F, Angelidaki I. Hybrid biogas
[21] Langmead B, Salzberg SL. Fast gapped-read alignment with Bowtie 2. Nat Methods upgrading in a two-stage thermophilic reactor. Energy Convers Manag 2018;168:
2012;9:357–9. https://doi.org/10.1038/nmeth.1923. 1–10. https://doi.org/10.1016/j.enconman.2018.04.074.
[22] Li H, Handsaker B, Wysoker A, Fennell T, Ruan J, Homer N, et al. The Sequence [41] Yan M, Treu L, Zhu X, Tian H, Basile A, Fotidis IA, et al. Insights into ammonia
Alignment/Map format and SAMtools. Bioinformatics 2009;25:2078–9. https:// adaptation and methanogenic precursor oxidation by genome-centric analysis.
doi.org/10.1093/bioinformatics/btp352. Environ Sci Technol 2020;54:12568–82. https://doi.org/10.1021/acs.
[23] Kang DD, Li F, Kirton E, Thomas A, Egan R, An H, et al. MetaBAT 2: an adaptive est.0c0194510.1021/acs.est.0c01945.s002.
binning algorithm for robust and efficient genome reconstruction from [42] Zhu X, Campanaro S, Treu L, Kougias PG, Angelidaki I. Novel ecological insights
metagenome assemblies. PeerJ 2019;7:e7359. https://doi.org/10.7717/ and functional roles during anaerobic digestion of saccharides unveiled by
peerj.7359. genome-centric metagenomics. Water Res 2019;151:271–9.
[24] Parks DH, Imelfort M, Skennerton CT, Hugenholtz P, Tyson GW. CheckM: assessing [43] Du Z-J, Wang Z-J, Zhao J-X, Chen G-J. Woeseia oceani gen. nov., sp. nov., a
the quality of microbial genomes recovered from isolates, single cells, and chemoheterotrophic member of the order Chromatiales, and proposal of
metagenomes. Genome Res 2015;25:1043–55. https://doi.org/10.1101/ Woeseiaceae fam. nov. Int J Syst Evol Microbiol 2016;66:107–12. https://doi.org/
gr.186072.114. 10.1099/ijsem.0.000683.
[25] Chaumeil P-A, Mussig AJ, Hugenholtz P, Parks DH. GTDB-Tk: a toolkit to classify [44] Lei Y, Oshima T, Ogasawara N, Ishikawa S. Functional Analysis of the Protein Veg,
genomes with the Genome Taxonomy Database. Bioinformatics 2020;36:1925–7. Which Stimulates Biofilm Formation in Bacillus subtilis. J Bacteriol 2013;195:
https://doi.org/10.1093/bioinformatics/btz848. 1697–705. https://doi.org/10.1128/JB.02201-12.
[26] Hyatt D, Chen G-L, LoCascio PF, Land ML, Larimer FW, Hauser LJ. Prodigal: [45] Zhu X, Campanaro S, Treu L, Seshadri R, Ivanova N, Kougias PG, et al. Metabolic
prokaryotic gene recognition and translation initiation site identification. BMC dependencies govern microbial syntrophies during methanogenesis in an anaerobic
Bioinf 2010;11:119. https://doi.org/10.1186/1471-2105-11-119. digestion ecosystem. Microbiome 2020;8. https://doi.org/10.1186/s40168-019-
[27] Shaffer M, Borton MA, McGivern BB, Zayed AA, Rosa SLL, Solden LM, et al. DRAM 0780-9.
for distilling microbial metabolism to automate the curation of microbiome [46] Demirel B, Scherer P. The roles of acetotrophic and hydrogenotrophic
function. BioRxiv 2020:2020.06.29.177501. https://doi.org/10.1101/ methanogens during anaerobic conversion of biomass to methane: a review. Rev
2020.06.29.177501. Environ Sci Biotechnol 2008;7:173–90. https://doi.org/10.1007/s11157-008-
[28] Huerta-Cepas J, Szklarczyk D, Heller D, Hernández-Plaza A, Forslund SK, Cook H, 9131-1.
et al. eggNOG 5.0: a hierarchical, functionally and phylogenetically annotated [47] Garcia-Robledo E, Ottosen LDM, Voigt NV, Kofoed MW, Revsbech NP. Micro-scale
orthology resource based on 5090 organisms and 2502 viruses. Nucleic Acids Res H2–CO2 dynamics in a hydrogenotrophic methanogenic membrane reactor. Front
2019;47:D309–14. https://doi.org/10.1093/nar/gky1085. Microbiol 2016;7. https://doi.org/10.3389/fmicb.2016.01276.
[29] Galili T, O’Callaghan A, Sidi J, Sievert C. heatmaply: an R package for creating [48] McMahon K. Anaerobic codigestion of municipal solid waste and biosolids under
interactive cluster heatmaps for online publishing. Bioinformatics 2018;34: various mixing conditions—II: microbial population dynamics. Water Res 2001;35:
1600–2. https://doi.org/10.1093/bioinformatics/btx657. 1817–27. https://doi.org/10.1016/S0043-1354(00)00438-3.
[30] Oksanen J, Blanchet FG, Friendly M, Kindt R, Legendre P, McGlinn D, et al. vegan: [49] Nguyen LN, Johir MAH, Commault A, Bustamante H, Aurisch R, Lowrie R, et al.
Community Ecology Package. 2020. Impacts of mixing on foaming, methane production, stratification and microbial
[31] Kanehisa M, Goto S, Sato Y, Kawashima M, Furumichi M, Tanabe M. Data, community in full-scale anaerobic co-digestion process. Bioresour Technol 2019;
information, knowledge and principle: back to metabolism in KEGG. Nucleic Acids 281:226–33. https://doi.org/10.1016/j.biortech.2019.02.077.
Res 2014;42:D199–205. https://doi.org/10.1093/nar/gkt1076.

10