healthmedicinet

Acetogenic community composition and dynamics

Acetogenic community dynamic traced by fhs-TRFLP profiling

T-RFLP fingerprinting was conducted to follow the dynamics of the acetogenic community
in digesters SAO1 and SAO3. SAO1 was operated with stable (control) and SAO3 with
gradually increasing ammonia concentrations, which provoked a shift from aceticlastic
methanogenesis to SAO after 225 days of operation in the latter 7].

In SAO1, the fhs profile was relatively stable over the entire operating period of 642 days and was
represented by the major terminal restriction fragments (T-RFs) 86, 238, 253, 309,
477 and 593 bp (Fig. 1, green bars) with two exceptions: T-RFs 239 and 332 bp emerged at the last two sampling
points, at a relative abundance of more than 30 %, while another abundant T-RF with
size 159 bp (Fig. 1, blue bar) declined during digester operation and was no longer detectable by day
442. The dominant T-RFs found in SAO1 were also present in the SAO3 digester on the
first sampling occasion (day 70; 0.02 g NH
₃
/L). However, on days 141 and 225, when the SAO3 digester was subjected to ammonia
levels of around 0.09 and 0.30 g NH
₃
/L, respectively (1.9 and 3.3 g NH
₄⁺
-N/L, respectively), the T-RFLP profile indicated a distinct shift in the community,
characterised by the major T-RFs 220, 236, 296, 339, 442, 553 and 590 bp (Fig. 1, orange bars). On days 442 and 630, at ammonia levels of around 0.62 and 0.96 g NH
₃
/L (5.5 and 6.9 g NH
₄⁺
-N/L), respectively, yet another community shift occurred whereby the previously observed
T-RFs disappeared and instead the major T-RFs 87, 283, 379, 446, 471 and 591 bp became
dominant (Fig. 1, red bars). A large proportion of the amplified partial fhs sequences remained uncut (Fig. 1, black bars). A few fragments were only observed on day 442 at an ammonia level of
around 0.62 g NH
₃
/L (Fig. 1, pink bars) including the T-RFs 60, 85, 296, 463, 495 and 520 bp.

Fig. 1. Dynamics of the acetogenic community in SAO1 (low-ammonia control digester) and SAO3
(high-ammonia experimental digester) traced by formyltetrahydrofolate synthetase (fhs) gene profiling using terminal restriction fragment length polymorphism (T-RFLP).
Terminal restriction fragments (T-RFs) that could be affiliated to fhs genotypes are labelled by fragment size (bp) and operational taxonomic unit (OTU)
affiliation or accession number. T-RFs were grouped into equally behaving fragments:
Not establishing under either condition (blue 50, 159 bp), stable fragments in the control but fading in the experimental digester
(green 86, 253, 238, 309, 477, 593 bp), establishing fragments up to 0.62 g NH
₃
/L in the experimental digester, not detected in the control (orange 76, 220, 236/7, 240, 339, 442, 553, 590 bp), establishing fragments up to 0.96 g
NH
₃
/L in the experimental digester, not detected in the control (red 283, 379, 446, 471, 591 bp), exclusively detected at sample point day 442 in experimental
digester (pink 60, 296, 463, 495/6, 520 bp). Fragments marked in black remained uncut. Peaks that emerged non-chronologically on one or two occasions are marked in grey. Days of operation are plotted on the x-axis, relative peak abundance on the y-axis

Clone-based comparison of acetogenic community composition

A total of 79 and 61 clones recovered from SAO1 and SAO3 digesters, respectively,
were sequenced. Sequences sharing 97–99 % nucleotide identity were considered an OTU;
otherwise identical fhs sequences or sequences detected only once were considered to be “fhs clones” and named according their accession numbers (JQ0822…). The OTU and clone
frequency (rank abundance) is presented
in Additional file 1: Figure S1. In the experimental digester SAO3, 13 fhs clones and 10 fhsOTUs were recovered, designated OTU1-OTU10. Of the total partial fhs sequences recovered, 13 % were represented by OTU10, followed by OTU8, OTU7 and OTU5
(9.1 % each). For the control digester SAO1, 17 fhs clones and 10 fhsOTUs were obtained, designated OTU11-OTU20. Comparing the deduced amino acid sequences,
OTU19 and OTU20 shared a pairwise identity of 98 %, representing together 27 % of
the total partial fhs sequences recovered from SAO1. OTU11 and OTU14 had a pairwise identity of 92 % and
represented in total 8 % of the fhs sequences recovered from SAO1. With the exception of OTU15 (4 % in SAO1), which was
also recovered with one clone from the library of the experimental digester, none
of the partial fhs sequences was recovered from SAO3 and vice versa. All OTUs found in the experimental
and control digesters, as well as nine fhs clones in SAO3 and seven fhs clones in SAO1, were allocated to the T-RFLP pattern (Fig. 1). However, OTUs and clones were often represented by one and the same T-RF, as illustrated
in Fig. 1. Moreover, the most highly abundant T-RF (332 bp) that emerged at the last two sampling
points in the control digester could not be allocated to any of the clones obtained,
and might thus be an enzymatic or technical artefact.

Abundance of ammonia-induced OTUs in other ammonia-stressed biogas processes

Based on their numerical dominance in the ammonia-stressed SAO3 digester, OTUs 3–10
were selected as representatives for potential novel SAOB. OTU2 was excluded from
further analyses since it appeared to be identical to the SAOB T. acetatoxydans. The abundance of this species has already been investigated in the digesters in
question 12], 46]. In order to assess the occurrence of these selected OTUs, specific primers were
designed for qPCR assays. The specificity of the primers and the qPCR efficiency are
shown in Additional file 2: Table S1. The primers were used for investigation of ten large-scale biogas plants
(B, C, D-H, J, L-M) and two parallel laboratory-scale digesters R1 and R2 (at day
390), operating at high ammonia level and dominated by SAO (Fig. 2a; Additional file 3: Table S2). The majority of the OTUs analysed were present in all SAO processes investigated.
OTUs 3, 6, 7 and 10 were even found in the two thermophilic SAO digesters (L and M),
but at lower abundance (Fig. 2a). In the low-ammonia, large-scale processes dominated by aceticlastic methanogenesis
(B, C), only OTU10 and OTU7 were found to be present, and at significantly lower abundance
than in the SAO-dominated processes. Relating the number of OTUs present to the total
number of OTUs analysed, the highest ratio among the SAO-dominated processes was obtained
for biogas plants G, H and J, and lower values for the thermophilic plants L and M.
The lowest ratio was obtained for the aceticlastic processes (Fig. 2b).

Fig. 2. Abundance of operational taxonomic units (OTUs) 3–10 in industrial and laboratory-scale
biogas processes, illustrated as: a log scale of gene copy numbers and b the ratio between number of OTUs present and total number of OTUs analysed. The presence
of all OTUs gave a ratio = 1, as illustrated for the control (SAO3 day 442). Log scale
values and standard deviation are summarised in Additional file 2: Table S1. Operation parameter and abundances of methanogens and known SAOB can be
found in supplementary Additional file 9: Table S6 and Additional file 10: Table S7

The abundances of OTU3-10 were also traced over the entire operating period (430 days)
in the two parallel laboratory-scale digesters R1 and R2 by analysing 11 sampling
points. Among all OTUs investigated, OTUs 5, 7, 8 and 10 were found to be part of
the microbial community (Fig. 3a). OTU5 and OTU8 were present from the beginning and increased slightly with increasing
ammonia level, as observed for the SAOB S. schinkii (Fig. 3b) 46]. However, OTU7 and OTU10 were only detectable when the ammonium-nitrogen level exceeded
3 g/L and SAO became the dominant methane production process, a pattern previously
reported for the SAOB T. acetatoxydans and C. ultunense (Fig. 3b) 46].

Fig. 3. a Abundance of operational taxonomic units (OTUs) 5, 8, 7 and 10 in two parallel mesophilic
laboratory-scale biogas digesters R1 and R2 subjected to increasing ammonia levels,
illustrated as: log scale of fhs gene copy numbers obtained for OTU 5, 8 (upper graph), 7 and 10 (lower graph) and b average log gene abundance of the SAOB S. schinkii (upper graph) T. acetatoxydans (middle graph), and C. ultunense (lower graph) in R1 and R2 (dashed lines). b Re-drawn from data in Westerholm et al. 46]

Considering all mesophilic, ammonia-rich samples (digesters D-H and J, SAO3 day 442,
R1 day 390, R2 day 390), the OTUs 5, 7, 8 and 10 represented the most frequently found
fhs genotypes
(Additional file 4: Figure S2).

Phylogenetic affiliation of the recovered partial fhs sequences

The deduced amino acid sequences of the recovered partial fhs sequences were phylogenetically analysed in relation to known acetogens, the current
closest relative if sharing at least 80 % identity (Additional file 5: Table S3), sulphate reducers, known SAOB (Additional file 6: Table S4) and fhs sequences obtained from other biogas processes 40], 47], 48]. The partial fhs clones recovered from the low-ammonia control digester SAO1 formed three distinct
clusters, designated AD (

A

aerobic

D

igestion) clusters I-III (Fig. 4). The largest AD cluster, cluster I, is primarily represented by OTU12, OTU16 and
OTU17. The closest relatives identified belong mainly to the families Thermoanaerobacteraceae
and Ruminococcaceae (Additional file 5: Table S3). AD cluster II consists of OTU11 and OTU14, both sharing more than 90 %
identity with a Sedimentibacter species, order Clostridiales (Additional file 5: Table S3). AD cluster III is mainly formed by OTU19 and OTU20, which cluster together
with a member of the family Porphyromonadaceae as closest relative (Additional file
5: Table S3). OTU13, OTU15 and OTU18 were found dispersed in the tree and did not belong
to any of the clusters identified.

Fig. 4. Phylogenetic placement in the maximum likelihood tree of the deduced formyltetrahydrofolate
synthetase (FTHFS) amino acid sequences. Reference strains are given together with
their accession numbers. AD anaerobic digestion, pSAOB potential syntrophic acetate-oxidising bacteria. The sulphate reducers formed a separate
clade including the SAOB S. schinkii and T. phaeum as shown by Müller et al. 31] and were therefore excluded from the alignment. The tree was then re-build in order
to reduce the size. (1) Westerholm et al. 48], (2) Moestedt et al. 47], (3) Westerholm et al. 43], (4) Hori et al. 40]

For the ammonia-rich experimental digester SAO3, another three distinct clusters,
designated potential Syntrophic Acetate-Oxidising Bacteria (pSAOB) clusters I–III,
were found (Fig. 4). In general, these clusters appear distantly related to those found in the control
digester SAO1. OUT9 and OTU5 form pSAOB cluster I, together with a bunch of fhs clones retrieved from SAO3 and other ammonia-rich processes 47], 48]. The current closest relatives identified affiliate to Thermoanaerobacterales families
and Peptococcaceae (Additional file 5: Table S3). pSAOB cluster II encompasses OTU3 and three more fhs genotypes recovered from SAO3. The closest relatives found belong to the families
Thermoanaerobacteraceae and Ruminococcaceae, but with identities below 80 % (Additional
file 5: Table S3). Another bunch of partial fhs sequences including OTU7 shared up to 90 % identity with the SAOB C. ultunense, family Clostridiaceae (Additional file 6: Table S4), which together form the pSAOB cluster III. OTU1, OTU4, OTU6, OTU8 and
OTU10 were found scattered throughout the tree. OTU10 shares high identity (96 %)
with the FTHFS of Pediococcus damnosus, family Lactobacillaceae, a lactic acid producing bacterium commonly found in beer
production 49] (Additional file 5: Table S3). OTU4, OTU8 and OTU6 are only distantly related to members of the Enterococcaceae,
Ruminococcaceae and Clostridiales, respectively (Additional file 5: Table S3). OTU1 appeared to be distantly related to any reference strain included
in the alignment. With the exception of OTU2 (97 % nucleotide identity to T. acetatoxydans) and AD cluster II, none of the OTUs or clones recovered branched together with characterised
acetogens.

Dynamic and taxonomic distribution of the bacterial community

MiSeq Illumina sequencing and T-RFLP profiling targeting the 16 s rRNA gene pool were
used to investigate the whole bacterial community composition and dynamics in order
to shed light on the phylogenetic affiliation of potential SAOB candidates, as indicated
by the acetogenic community composition. The numbers of OTUs, coverage, Chao1, Shannon
index and Simpson index values obtained per sample are summarised in Table 1.

Table 1. Summary of observed OTUs, Chao1, Shannon and Simpson index in digester SAO1 and SAO3

The rarefaction analysis is displayed in supplementary Additional file 7: Figure S3. The estimated coverage indicated that, in the case of the control digester
SAO1, the observed OTUs covered between 89 and 94 % of the bacterial community and
in the case of the experimental digester SAO3 83–90 %. The number of OTUs in the experimental
digester SAO3 decreased concurrently with rising ammonia level, whereas no similar
trend was observed in the control SAO1. The Simpson index indicated slightly lower
evenness with increasing ammonia level, with the lowest value (0.907) at sampling
point day 642. The community in the control digester remained more even (0.981) at
that sampling point. The Shannon index, accounting for both species abundance and
evenness, revealed a similar trend, with the lowest value at day 642 in SAO3 (5.26)
compared with the control at the same sampling point (7.46). Both weighted and unweighted
UniFrac matrices PCoA plots revealed that all five SAO1 sampling points and the first
SAO3 sampling point grouped close to each other, indicating similar phylogenetic composition.
However, a significant phylogenetic distance was observed between the samples withdrawn
from the control digester SAO1 and the samples from the experimental digester SAO3
subjected to increasing ammonia levels (Fig. 5).

Fig. 5. Principal coordinates analysis (PCoA) plot showing phylogenetic distances between
SAO digester samples as determined by a unweighted UniFrac principal coordinate analysis and b weighted UniFrac principal coordinate analysis. Red SAO1; blue SAO3

On average, taxonomic assignment using the Ribosomal Database Project (RDP) classifier
encompassed 85–97 % of the partial 16S rRNA gene sequences on phylum level and 45–77 %
on family level. The bacterial community in the control digester SAO1 consisted of
roughly equal proportions of the phyla Firmicutes (18–24 %), Bacteroidetes (23–31 %)
and Actinobacteria (17–31 %). However, the latter dropped to 7 % after day 225 and
was mainly replaced by Bacteroidetes (Fig. 6; Additional file 8: Table S5). The phylum Actinobacteria was strongly dominated by the family Actinomycetaceae
(7–30 %). Within the phylum Bacteroidetes, unclassified Bacteroidales were most frequent
(16–27 %), together with a smaller proportion of Porphyromonadaceae (2–7 %, Fig. 6). The highest diversity was observed within the phylum Firmicutes, in which most
of the families belonged to the order Clostridiales, including Clostridiaceae (1.0–1.3 %),
Syntrophomonadaceae (2–3 %), genus Syntrophomonas, Tissierellaceae (3–6 %), genus Sedimentibacter, and 1–3 % unclassified Clostridiales members. Members of the phyla Chloroflexi (2–5 %),
Synergistetes (3–6 %), Verrucomicrobia (2–8 %) Proteobacteria (1–4 %), Planctomycetes
(0.8–1.4 %), Spirochaetes (0.5–3 %), Tenericutes (0.4–1.5 %), Thermotogae (0.5–2.2 %)
and WWE1 (0.2–6 %) were also present, but in smaller proportions. These minor phyla
were mainly dominated by the families Anaerolinaceae (1–3.5 %), Syntrophaceae (0.2–2 %),
including the genus Syntrophus (0.1–2 %), Thermotogaceae (0.1–2 %), including the genus Kosmotoga (0.4–1.7), Dethiosulfovibrionaceae (2–4 %), Cloacamonaceae (0.1–6 %), unclassified
Bacilli (2–5 %) and 1–6 % unclassified Verrucomicrobia.

Fig. 6. OTU heatmap based on bacterial OTUs having relative abundance higher or equal to 2.5 %
with the process SAO1 and SAO3

The bacterial community in the experimental digester SAO3 encompassed the same phyla
as observed in the control digester at the first sampling point, but with increasing
ammonia levels the proportions were significantly affected (Fig. 6; Additional file 8: Table S5). In particular, OTUs affiliated to the phylum Firmicutes increased from
an initial relative abundance of 21 % up to 80 % between day 442 and day 642.

Conversely, the abundance of Bacteroidetes declined from 25% to 15–17 % at the last
two sampling points. The phylum Actinobacteria was already strongly affected by the
first ammonium increase at sampling point day 141, representing only 5 % of the bacterial
community.

On family level, the impact of elevated ammonia levels on community composition was
even more significant (Fig. 6; Additional file 8: Table S5) and can be summarised as follows: The gradual increase in ammonia level
was accompanied by two distinct changes in the community composition. The first change
occurred up to an ammonium level of 0.3 gNH
₃
-N/L at sampling point day 225 and a second change in community composition occurred
between day 225 and day 442, when ammonia level was further increased. The majority
of the families observed in SAO1 declined in the experimental digester SAO3 with increasing
ammonia levels and were completely replaced at the last two sampling points by other
families or another genus belonging to the same family.

The phylum Actinobacteria was mainly represented by the family Actinomycetaceae (genus
Actinomyces, 3–5 %). Similar low diversity on this taxonomic level was found for the phylum Bacteroidetes,
which was represented only by members of the family Porphyromonadaceae, but at a significantly
higher abundance (15–17 %) than in the control digester (1.6–6 %), at sampling points
day 442 and day 642. The highest diversity and the most pronounced community change
were observed within the phylum Firmicutes (Fig. 6; Additional file 8: Table S5). The first community change in response to increasing ammonium levels
was characterised by a rise of Clostridiaceae, dominated by genus Clostridium (up to 7 %), Eubacteriaceae, genus Eubacterium (up to 1.3 %), Ruminococcaceae (up to 2.5 %), Acidaminobacteraceae, genus Guggenheimella (up to 1 %), Tissierellaceae, unknown genus (up to 31 %), Erysipelotrichaceae (up
to 2.3 %), Lactobacillaceae (up to 2.3 %) and Leuconostocaceae, genus Leuconostoc (up to 1.2 %), as well as up to 12 % unclassified Clostridiales members. However,
members of the family Cloacamonaceae (up to 12 %) belonging to the WWE1 candidate
division also increased. The remaining changes were an increase of 0.4 %, at family
level (Additional file 8: Table S5). However, once ammonia concentration exceeded 0.30 g NH
₃
/L, this bacterial community was replaced by members of the following taxa: Clostridia
order SHA-98 (up to 31 %), and MBA08 (up to 8 %), Tissierellaceae, genus Tepidmicrobium (up to 19 %), Caldicoprobacteraceae, genus Caldicobacter (up to 5.0 %), Clostridiaceae, mainly genus Alkaliphilus (up to 1.8 %), Clostridiales (0.5–4.0 %), Thermoanaerobacterales (1.1–1.4 %), Thermoanaerobacteraceae,
genus Thermacetogenium (up to 0.9 %), and unclassified Firmicutes (up to 0.9 %), Enterococcaceae (up to
1.7 %), Lactobacillaceae, unknown genus (up to 6 %) and Acholeplasmataceae (up to
0.6 %), the latter belonging to the phylum Tenericutes (Figs. 6, 7; Additional file 8: Table S5). All other changes observed were below 0.1 %.

Fig. 7. Correlation between bacterial community and the level of ammonium and free ammonia
(g/L) within the process illustrated by using principal component analysis (PCA).
The dominant bacterial operational taxonomic units (OTUs) that have relative abundance
higher or equal to 2.5 % were selected as representative for bacterial community.
All variables were centred and scaled before analysis. Results of the first two PCs
were present within biplot. Sampling points day 70 (.0), day 141 (.2), day 225 (.3),
day 442 (.5), day 642 (.7)

A similar community dynamic was observed using T-RFLP analysis (Additional file 7: Figure S4). A cluster of bacteria represented by T-RFs labelled orange increased
until the ammonia concentration reached 0.30 g/L at day 225, but declined at higher
ammonia levels. Instead, another group of bacteria labelled red in the T-RFLP profile
emerged and replaced the bacterial community dominating under low-ammonia conditions
(labelled blue, Additional file 7: Figure S4).