Trace element and temperature effects on microbial communities and links to biogas digester performance at high ammonia levels

Methanogenic abundance and community structure

Quantitative PCR (qPCR) results obtained with primers targeting the 16S rRNA gene of methanogens belonging to Methanomicrobiales, Methanosarcinaceae, Methanobacteriales and Methanoculleus bourgensis are shown in Fig. 2. Genes related to Methanosaetaceae did not appear above the detection limit in any digester. Methanobacteriales varied between 2 × 10⁵ and 8 × 10⁶ gene copies/mL in all sampling points and no significant difference was revealed between the digesters. M. bourgensis comprised the major part of Methanomicrobiales and both increased in abundance with TE addition at 37 °C. The levels of Methanosarcinaceae was relatively low at 37 °C independent of TE addition and detected values varied between 4 × 10³–6 × 10⁶ gene copies/mL. In D^TE37 Methanosarcinaceae was below detection limit of 1 × 10³ gene copies/mL at day 126, 168 and 280 (Fig. 2). Comparisons of the digesters operating at 42 °C with and without addition of TE revealed no significant difference in gene abundance of any methanogenic group. However, the enhanced temperature in the non-supplemented digester (D42) caused a significant increase of Methanomicrobiales and Methanosarcinaceae compared to D37, whereas the level of M. bourgensis decreased. Methanosarcinaceae was significantly higher in D^TE42 than in D^TE37, whereas the difference in temperature between these digesters did not affect the levels of Methanomicrobiales. However, as observed for the non-supplemented digesters, M. bourgensis was lower at 42 °C than at 37 °C. Furthermore, in D42 and D^TE42 the gene abundance of Methanomicrobiales and M. bourgensis appeared to be less dynamic compared to the 37 °C digesters (Fig. 2).

Over 300 partial mcrA (methyl coenzyme-M reductase) sequences were recovered and analysed in the clone libraries. About 4 % of the total clones originated from unspecific amplification and were therefore excluded from further analysis. The recovery rate of the libraries was 74 % in D^TE37, 83 % in D37, 97 % in D42 and 97 % in D^TE42. The relative abundances of partial genes recovered are shown in Fig. 3, either grouped as operational taxonomic units (OTUs) or as single clones. The grouping of mcrA-gene sequences based on ≥99 % identity, or on ≥97 % (data not shown), provided a higher number of OTUs and higher number of separate single sequences in D^TE37 than in D37. The profiles of the 42 °C digesters were identical irrespective of the identity cut-off value chosen (≥97 or ≥99 %).

The position of each OTU and the accession number of reference sequences are shown in the phylogenetic tree (Fig. 4), while the closest relative(s) based on nucleotide sequence and deduced amino acid sequence identity are assigned in Additional file 1: Table S2. The brackets in Fig. 4 bundle those OTUs sharing at least ≥97 % identity and the respective closest relative. The assemblages are designated A (OTU2, 15), B (OTU4, 8–11) and C (OTU3, 5–7). OTUs in assemblage C appeared in both D37 and D^TE37 with equal order of magnitude (25–30 % of total clones). Group B was also represented in both these digesters, but with lower relative abundance in D37 (16 % of total clones) than in D^TE37 (28 %). Sequences included in group A were found only in D^TE37 (24 % of total clones).

OTU1, highly dominant in D37 (41 % of the total clones), had low nucleotide sequence identity (<79 %) to previously available mcrA genes. However, the deduced amino acid was revealed to have 88 % similarity to Methanosarcina acetivorans and Methanosarcina mazei. The other highly prevalent OTUs (3–6, 10 and 12) in digester D37 were closely related to M. bourgensis (97–99 % nucleotide identity, ≥97 % mcrA amino acid similarity), and together accounted for 50 % of the total clones. About 5 % of the gene sequences retrieved in D37 (OTU17) most closely resembled Methanobrevibacter smithii (89 % nucleotide identity, 97 % mcrA amino acid similarity). In D^TE37, 19 % of the retrieved clones assembled into OTU2 (92 % nucleotide identity, 97 % amino acid similarity to M. bourgensis). OTU1, highly dominant in D37, also appeared in D^TE37, but was of considerably lower relevance (7 %) than in the non-supplemented digester. In D^TE37, the OTUs closely affiliating to M. bourgensis (97–99 % nucleotide identity; OTUs 3–12) comprised 64 % of the total clones, but as many as 96 % of the clones had M. bourgensis as closest relative (92–99 % nucleotide identity; OTUs 2–12 and 15). The closest relative to OTU14 (4 % of clones from D^TE37) was Methanomassiliicoccus luminyensis (83 % nucleotide identity, 94 % similarity). OTU20 dominated the methanogenic community in both D42 and D^TE42, comprising 91–93 % of the total clones, respectively. Based on the nucleotide sequence, this OTU was most closely related to M. bourgensis (93 %), whereas the deduced amino acid sequence had highest identity (97 %) to Methanoculleus chikugoensis. mcrA genes (OTU19) with low identity (<79 %) to previously available sequences and 97 % amino acid identity to M. luminyensis were also identified in the 42 °C digesters.

Acetogenic community

The profiles of the acetogenic communities in the digesters on days 77, 168 and 280, obtained by terminal restriction fragment length polymorphism (T-RFLP) analysis using the functional gene fhs encoding the enzyme formyltetrahydrofolate synthetase (FTHFS), are displayed in Additional file 1: Figure S2. Ordination of terminal restriction fragment (T-RF) frequency and abundance by using a paired t test illustrated that a few of the most abundant T-RFs differed from the digester core community, apparently significantly affected by TE addition (x axis) and/or increased temperature (y axis) (Fig. 5). T-RF 65 bp positively correlated with TE addition at 37 °C (T-RFLP gene frequencies between 10 and 31 % in D37; 33–44 % in D^TE37), whereas at this temperature TE addition had a negative impact on the abundance of T-RF 283 bp (22–33 % in D37; 3 % in D^TE37). TR-F 283 bp was also negatively affected by operation at 42 °C instead of 37 °C, both with and without TE addition (22–33 % in D37; 6–17 % in D42; >17 % in D^TE42). In contrast, temperature had a strong positive influence on T-RF 116 bp (>16 % in D37; 33–72 % D^TE42) and a moderate positive influence on the abundance of T-RF 581 bp (>3 % at 37 °C and >8 % at 42 °C).

The majority of the most abundant T-RFs were assigned to partial fhs sequences recovered in the clone libraries of D^TE37 and D^TE42 at day 168. The relative OTU distribution is displayed in Additional file 1: Figure S3 and the accession numbers and the assignments of retrieved fhs sequences are listed in Additional file 1: Table S3. In total, 23 OTUs and genotypes were recovered from D^TE37, which was twice as much as from D^TE42. Of these, only OTU _fhs 32, 29, 30, 26 and the genotype KJ_701058 were recovered from both temperature conditions (Additional file 1: Table S3, Figure S3). OTU _fhs 31, 32, 33, 21 and 25 lacked the restriction site for Hpy188III and are thereby all represented by T-RF 635 bp in Additional file 1: Figure S2. Consequently, temporal dynamics or differences between digesters as regards these OTUs were not tracked by T-RFLP. However, the fhs clone library disclosed higher frequency of OTU _fhs 32 in D^TE37 than in D^TE42, whereas OTU _fhs 31, 33, 21 and 25 were only retrieved from D^TE37.

The recovered fhs genes distinguished phenotypically from each other and had partly low identities to known acetogens (Fig. 6). However, the phylogenetic tree constructed from deduced FTHFS amino acid sequences illustrated similarity with partial fhs sequences recovered from another previously investigated mesophilic AD process operating with high ammonia levels [16, 33] (Genbank Popset JQ082255-97, SAO3 clones; submitted by B. Müller and A. Schnürer). One partial fhs sequences retrieved from D^TE37 (KJ_701058 T-RF 309 bp) positioned together with the FTHFS1 of Clostridium ultunense and was to 99 % identical to one of the previously identified fhs clones (JQ082286, HM365339). None of the sequences grouped together with Syntrophaceticus schinkii or Thermacetogenium phaeum, or seemed to have close relatedness to FTHFS1 and FTHFS2 of Tepidanaerobacter acetatoxydans. OTU _fhs 28 (T-RF 116), highly abundant at 42 °C, and OTU _fhs 32, present in both D^TE37 and D^TE42, were in the tree closely related to each other, and the latter was identical to fhs clone JQ082293. Also OTU _fhs 30 retrieved from both D^TE37 and D^TE42 and OTU _fhs 31 highly presented in D^TE42 showed 99 % sequence identities to fhs clone JQ082296 and fhs clone JQ082263, respectively.

The T-RFs 65, 283 and 581 bp, which were significantly affected by temperature or/and TE addition, likely represent restriction fragments of OTU _fhs 23, OTU _fhs 29 and OTU _fhs 27 (Figs. 5, 6 and Additional file 1: Table S3). T-RF 283 bp (OTU _fhs 29), negatively correlating with TE addition and increased temperature shared 99 % identity to fhs clone JQ082274. T-RF 65 bp (OTU _fhs 23), only recovered from D^TE37 and positively affected by TE, was also branching together with one of the previously recovered fhs clones (JQ082278). T-RF 581 bp positively correlating with temperature can either affiliate to OTU _fhs 27 or to genotype KJ_701077. Since OTU _fhs 27 was exclusively recovered from D^TE37, KJ_701077, which was only recovered from D^TE42, might rather correspond to T-RF 581 bp. The positioning in the tree revealed the thermophilic acetogen Moorella thermoacetica as a close relative to OTU _fhs 27, whereas the genotype KJ_701077 was grouped in the same branch as two members of the halotolerant Haloanaerobiales.

Abundance of total bacteria and syntrophic acetate-oxidising bacteria (SAOB)

Comparisons of the abundance of C. ultunense in digesters with and without addition of TE revealed no significant difference during operation at 37 or 42 °C. However, S. schinkii abundance was significantly (P < 0.05) lower in D^TE37 and D^TE42 than in D37 and D42, respectively. T. acetatoxydans appeared at lower levels in D^TE42 than in D42, whereas the qPCR data indicated no substantial disparity between D37 and D^TE37.

Comparisons of digesters operating at 37 °C and at 42 °C revealed significantly (P < 0.05) higher levels of C. ultunense, S. schinkii and T. acetatoxydans at the higher temperature, in digesters with and without addition of TE. No products were generated in end-point PCR analyses of digester sludge with primers targeting T. phaeum and Thermotoga lettingae.

Presence of syntrophic propionate-oxidising bacteria (SPOB)

PCR analyses of SPOB were conducted with primers targeting the propionate CoA-transferase gene (pct). No specific products were generated from digester sludge in PCR with the general pct primer or with the specific primers (pct-c1 to c7). In PCR with general pct primers and pure culture DNA of Syntrophobacter fumaroxidans, Syntrophobacter sulfatireducens, Syntrophobacter wolinii and Pelotomaculum propionicicum, products of different lengths were generated. Altered PCR conditions (decreased annealing temperature from 62 to 61 °C; exclusion of bovine serum albumin, BSA) enhanced the retrieved amount of amplicons of predicted length, but gel purification and sequencing showed that this was not a specific product. In PCR using the primer pair pct-c1 or pct-c6, designed to target S. fumaroxidans MPOB and Tepidanaerobacter sp.-related pct, respectively, no products were formed from genomic DNA of S. fumaroxidans and T. acetatoxydans. In PCR with the 16S rRNA targeting primer pair Synt737-f/906-r, amplicons of the correct size were formed from pure culture DNA of S. fumaroxidans, S. sulfatireducens and S. wolinii, but not from Syntrophobacter pfennigii. However, with these primers no products were generated in PCR analyses of the digester samples. Inhibition of PCR could be excluded by spiking experiments using pure culture DNA. The specificity of Smi103-f/442-r could not be practically assessed, due to lack of growth of Smithella propionica (DSM 16934, obtained from DSMZ) and absence of product formation with this primer pair in digester samples.

Digester performance, production kinetics and batch degradation assays

The disparity in performance of the four digesters studies reflected the positive impact on methane yield of addition of TE, as reported in other studies investigating anaerobic degradation of food industry and household waste [27, 29, 36, 37]. Since iron was included in the TE mixture used in this study this could also have contributed to the enhanced performance observed in supplemented digesters. Iron constrains the sulphide precipitation of available trace metals in sludge, due to the primary removal of released sulphide by the iron [2, 29, 38]. The increase in methane yield (30–37 %) achieved by TE addition would substantially increase the profits in a commercial biogas plant, although the cost of the TE mixture would need to be included. Furthermore, the methane production rate was higher already in the early part of the feeding cycle in digester D^TE37 (all periods) and in D^TE42 (period 3), which implies potential to increase the organic loading rate (OLR) with maintained efficiency of methane production.

Another finding supporting previous studies [27, 29, 36, 37] was that addition of TE proved to be a potent intervention against VFA accumulation, both in the semi-continuous digesters and in the batch assays. Elevated VFA levels are highly undesirable, since they represent significant loss of biogas. This was indicated by the lower methane yield in digesters D37 and D42 with their high VFA concentrations compared with D^TE37, which operated with VFA levels below the detection limit. A probable cause of the high VFA in D37 and D42 was lack of trace elements required for enzymes involved in the removal of hydrogen, since that affects both acetate and propionate oxidation by product-induced feedback inhibition [39]. Co and Se are required for synthesis of the enzymes needed for hydrogenotrophic methanogenesis [40, 41] and deficiency of these elements has previously been reported to induce propionate accumulation in a high-ammonia mesophilic process indicatively dominated by SAO [27]. Other studies suggest a prominent role of Ni and Fe in enzyme activity in the reduction of carbon dioxide with hydrogen to methane [42]. Furthermore, availability of these nutrients could also be critical for SAOB, thereby making them necessary for efficient SAO-mediated degradation.

Interestingly, despite the slow reduction in propionate in the batch assay without TE addition, acetate remained at relatively low levels, indicating the presence of efficient acetate-degrading communities. Thus, it seems likely that the accumulation of propionate was a consequence of a less efficient propionate-degrading microbial community in the unsupplemented digesters compared with the TE digesters. However, the picture becomes more complex when considering that low concentrations of hydrogen and acetate are also known to shift several of the primary fermentation pathways towards the formation of hydrogen, carbon dioxide and acetate, which results in lower production of side products such as propionate [43]. Consequently, the low VFA level in the TE-supplemented digesters might be a combined effect of reduced formation of VFA and enhanced conversion efficiency.

Elevated VFA levels have been suggested to indicate process disturbance [44]. However, despite the high VFA levels in D37 and D42, relatively stable operation was maintained throughout the 320-day study. In this context, it should be borne in mind that the consistent operation maintained in the experimental digesters could have increased their functionality. The larger variations in input material and load typically seen at large-scale production plants would most likely compromise the functional stability observed in the experimental digesters.

Abundance and composition of methanogens, acetogens and SAOB

Addition of trace elements significantly increased the abundance of the hydrogenotrophic Methanomicrobiales, specifically M. bourgensis, at 37 °C, whereas it had no effect on the level of Methanosarcinaceae. This contradicts previous findings that addition of TE increases the abundance of Methanosarcinaceae, allowing it to outcompete members of Methanomicrobiales and Methanobacteriales [29]. That study defined SAO as the dominant pathway for methanogenesis of acetate, which led the authors to suggest involvement of Methanosarcinaceae as a hydrogenotrophic partner in SAO. Differences in operating conditions, in particular the lower ammonia concentration (0.4 g NH₃/L) in that study compared with the present investigation, is a likely reason for the different effect of TE addition on Methanosarcinaceae.

The mcrA gene clone libraries confirmed the qPCR results by demonstrating that 50 and 64–96 % (based on >97–92 % nucleotide identity) of total recovered genes affiliated to M. bourgensis in D37 and D^TE37, respectively. Methanogenic assemblages obtained by grouping based on ≥97 % sequence identity (shown in brackets in Fig. 4) visualised further structural differences between the 37 °C digesters. The higher relative abundance of strains represented by assembly B and C in D^TE37 could be the cause, or the effect, of the improved efficiency and the low hydrogen partial pressure of this specific digester. Furthermore, several of the mcrA genes, in particular from D^TE37, related to M. bourgensis strains MAB1, MAB2, MAB3 and BA1 [20], which have been identified as important hydrogen-utilising microorganisms in association with SAO, possibly owing to tolerance to elevated levels of ammonia and a high affinity for hydrogen [17]. This would contribute to lower hydrogen partial pressure in the system, in particular that prevailing in D^TE37, which would positively influence the upstream fermenting microbes and subsequently the complete process.

The relatively lower abundance of S. schinkii in both digesters with TE addition and of T. acetatoxydans in D^TE42 compared with the other digesters was interesting, particularly since this occurred concurrently with persistent dominance of the SAO pathway and enhanced performance of these digesters. Fhs gene-based analyses were carried out in order to investigate the presence and impact of operating parameters on bacteria with the potential to be significant contributors to the SAO process in the digesters. Statistical analysis revealed that a few T-RFs were significantly influenced by addition of TE. TE addition might have increased the competiveness of members of the SAOB community and led to reduced abundance of a particular member, as observed for S. schinkii and T-RF 286 bp when comparing D37 and D^TE37. In contrast, T-RF 65 bp seemed to have a competitive advantage. At higher temperature, the effect of TE addition on the community dynamics appeared clearly reduced, most likely due to co-occurrence of loss of species richness. Another impact factor to consider might be the prevailing VFA levels. The T-RFLP profile indicated a link between VFA and T-RF 283 bp. This fragment was relatively abundant in D37 and D42 (which had high VFA levels) and also appeared in D^TE42 at first, but then gradually decreased in abundance concurrently with the decline in VFA in this digester (Additional file 1: Figure S3).

Digester performance, production kinetics and batch degradation assays

Operation at 42 °C, instead of the commercially more common range 35–38 °C, had a low (D42) or transient negative (D^TE42, periods 1–2) impact on methane yield. Similarly, in previous studies on the processing of energy crops, temperature fluctuations within correlating ranges had a low impact on methane yield [45], whereas increasing the temperature from 38 to 44 °C considerably increased the methane yield from nitrogen-rich distiller’s waste [28]. An important aspect when considering the impact of operating temperature on digester performance is the effect on the ammonia level. Increased temperature shifts the equilibrium between NH₄ ⁺ and NH₃ towards the latter, which is mainly responsible for inhibition of the microbial community [6]. Consequently, process disturbance following a temperature increase may be caused by the higher ammonia level, even though the nitrogen load caused no disturbance at the lower temperature. Accordingly, the higher ammonia concentration in D42 and D^TE42 compared with D37 and D^TE37 (Table 2) could be a possible contributor to the differences observed between these digesters.

Induced accumulation of propionate seems to be a frequent concern during operation within the higher mesophilic temperature range [28, 45, 46]. In accordance higher propionate levels were throughout the course of the present study detected in D42 compared to D37. The propionate to acetate ratio has been cited as an early indicator of imminent process failure [47], which indicates that the 42 °C digester would be less resistant to the appearance of stress or increased loading than D37. The reason for the elevated propionate level throughout in D42 and in the initial period in digester D^TE42 was not fully established, but could be due to sub-optimal temperature conditions for growth of SPOB (reviewed in [48]) or increased inhibition of the active propionate-degrading community by the increasing ammonia level. However, the decrease in propionate level after day 120 in D^TE42 rather implies slow microbial acclimatisation to the prevailing operating conditions, which was obviously possible when there was adequate access to trace elements.

Moreover, the disappearance of accumulated propionate, the increased methane yield and the differences in the kinetic profile in D^TE42 after the extended operating period at constant parameters (including the initial acclimatisation period of 3 months) emphasise the importance of having an extended operating period in experimental trials before claiming stable digester performance.

Abundance and composition of methanogens, acetogens and SAOB

The methanogenic and acetogenic community structure and the abundance of SAOB and methanogens were strongly influenced by the operating temperature. The higher level of Methanomicrobiales and Methanosarcinaceae observed with increased operating temperature is in line with previous observations [28]. However, the impact of ammonia, which was present at considerably higher levels in the 42 °C digesters, needs to be considered in this context. For example, the high abundance of OTU20 in the mcrA-gene profile and T-RF 116 bp (OTU _fhs 28) in the putative acetogenic community profile not only indicates favourable temperature conditions for the populations present, but also their considerably high ammonia tolerance. Constrained competition for substrate induced by the higher ammonia level could have been another advantage for these genotypes. Similarly, the lower competitiveness from other more ammonia-sensitive acetate consumers in the 42 °C digesters probably promoted increased abundance of the ammonia-tolerant SAOB. However, the higher growth rate at 42 °C than at 37 °C of known syntrophic acetate-degrading cultures [26] implies that the populations directly benefited from the increased temperature. In a previous study, a temperature rise increased the abundance of T. acetatoxydans in a high-ammonia (0.6 g NH₃/L) digester, but S. schinkii and C. ultunense were not affected [28].

Quantitative PCR analyses

Construction of DNA standards and qPCR analyses performed on syntrophic acetate oxidisers (C. ultunense, S. schinkii and T. acetatoxydans) and the methanogenic population (Methanomicrobiales, M. bourgensis, Methanosarcinaceae, Methanosaetaceae and Methanobacteriales) as described previously [20, 35]. Total bacteria were assayed as reported elsewhere [64]. Triplicate samples from each digester and sampling point were included in the analyses. The mean similarities of gene abundances were compared using Student’s t test and P < 0.05 was regarded as statistically significant. In order to monitor artefacts such as primer dimer formation and to assess whether non-specific amplification had occurred, a temperature melt curve was performed at the end of each qPCR assay (55–95 °C, ΔT = 0.1 °C/s). Non-specific amplification was further evaluated by analysing the products formed, using gel electrophoresis. All standard curves for the quantitative PCR analyses had a linear correlation coefficient (r ²) of 0.98–1.0 and the calculated qPCR efficiency of the reactions varied between 80 and 100 %. The detection limit for the different groups was about 60–560 gene copy numbers/mL digester sludge.

Presence of the thermophilic SAOB T. phaeum and T. lettingae was analysed with end-point PCR as described by Sun et al. [19]. Due to generation of unspecific amplification by primer pair Mst in the absence of strains related to Methanosaetaceae, the products formed in the qPCR analyses were checked by performance of clone library as described by Sun et al. [19].

Terminal restriction fragment length polymorphism and statistical analysis

The acetogenic community was profiled with DNA samples extracted on days 77, 168 and 280, in order to represent the three operating periods of the digesters. T-RFLP was performed using primers targeting the functional gene encoding the enzyme FTHFS. The recently designed degenerated 3-SAOfhs primer pair, optimised to include SAOB, was used in PCR as described previously [22]. The 3-SAOfhs forward primer was labelled with 6-carboxyfluorescein (FAM). Each preparation was then subjected to triplicate PCR reactions and the corresponding PCR products were pooled to reduce potential bias. To reduce background noise, the respective band at approx. 635 bp was gel-purified using the QIAquick Gel Extraction Kit (Qiagen, Hilden, Germany). The PCR fragment was digested with Hpy188III (Fermentas) at 37 °C for 3 h and T-RFs were separated by ABI3730XL DNA analyser (Applied Biosystems) at Uppsala Genome Center (Sweden), using MapMarker 1000 (Rox) as size standard. The fragment data obtained were evaluated with Peak Scanner software (Applied Biosystems) including peaks ranging from 50 to 635 bp. The resulting profiles were further evaluated in Excel by setting the threshold value for peak abundance at 1 % of total peak abundance.

In order to formally assess which T-RFs were influenced by TE, temperature or a combination of these, the following statistical analysis was performed. Four environmental settings were considered: core (D37), addition of TE (D^TE37), temperature (D42) and TE addition combined with temperature (D^TE42). For each of these settings, the frequency of each T-RF was observed at three different time points, days 77, 168 and 280. Each T-RF in the core was assessed for association with TE or temperature by comparing frequency at all sampling points using a paired t test. This pairing eliminated the effect of time in the dataset. The t test was actively chosen instead of a non-parametric test, because it evaluated the actual numerical frequency levels rather than just their ranks, which was deemed more relevant to the data. P values <0.05 indicated TE or temperature influence. The combined effect of TE and temperature was assessed by a t test investigating whether the frequency was similar to the expected effect of TE (as estimated with the TE t test) plus the expected effect of temperature (as estimated with the temperature t test). A low interaction (P < 0.05) indicated that the effect of the combined treatment was significantly different from that of the individual treatments.

Construction of clone libraries and phylogenetic trees

To further investigate the microbial communities, clone libraries from digesters D^TE37 and D^TE42 (day 168) for the acetogenic communities and all four digesters (day 280) for the methanogenic communities, were constructed. Triplicate DNA extractions were used as template in PCR amplification with the 3-SAOfhs primer pair [22], or with primers mlas and mcrA-rev using recommended reaction conditions [65]. In both surveys, amplicons of correct size from the triplicate DNA extractions were pooled, gel-purified and subsequently cloned into the pGEMT vector system (Promega, Madison, WI, USA) as recommended by the manufacturer. Cells of CaCl₂-competent Escherichia coli JM109 (Promega) were transformed by the ligation mixture according to the manufacturer`s protocol. Clones were randomly selected and plasmid inserts were verified with colony PCR using the vector-specific primer pair M13. The amplicons were subsequently sequenced with the M13 primer (Macrogen Inc. Seoul, Korea). Quality checks, editing and sequence assembly were performed with Geneious v6.1 [66]. FTHFS homologues with nucleotide identity above 97 % were considered to represent the same genotype and assigned to an OTU _fhs . In order to reveal differences within the archaeal community on strain level, the mcrA gene sequences were assigned to an OTU based on 99 % nucleotide identity. Phylogenetic analyses were conducted with the BlastP search algorithm (National Library of Medicine) using a representative sequence from each OTU. Transformation of the sequences into amino acid sequences, followed by homology search ≥89 % amino acid sequence similarity, identified affiliation to methanogenic species [65, 67].

The mcrA sequences were aligned by MUSCEL v5.2.2 [67] and a maximum likelihood tree was constructed using PhyML v3.0 [68] choosing the WAD substitution matric. Both MUSCEL v3.8.31 and PhyML v3.0 are available on the Mobyle platform of Institute Pasteur. FTHFS multiple sequence alignment was performed using MAFFT v7.017 [69] and the maximum likelihood tree was constructed using PhyML v3.0, both implemented in Geneious v6.1.8 (Biomatters Ltd., USA). The tree includes partial fhs sequences obtained in this study, obtained by [33] and [70], selected public available partial fhs sequences (Genbank Popsets JQ082213-54; JQ082255-97; submitted by B. Müller and A. Schnürer) and reference strains (accession numbers given in brackets).

Molecular assessment of syntrophic propionate-oxidising bacteria and establishment of mcrA genes of methanogenic isolates

End-point PCR with the general pct primers and the specific primers pct-c1 to -c7 designed to target the propionate CoA-transferase gene (pct) [60] were used for molecular assessment of syntrophic propionate oxidisers in the digesters. Genomic DNA of P. propionicicum (DSM 15578), S. fumaroxidans (DSM 10017), S. pfennigii (DSM 10092), S. wolinii (DSM 2805M), S. sulfatireducens (DSM 16706) and T. acetatoxydans (isolated at the Dept. of Microbiology, Swedish University of Agricultural Sciences) was subjected to PCR with the pct primers to check possible product formation. The PCR procedure was initially performed as described elsewhere [60]. However, due to low concentration of products formed on applying the recommended conditions with pure culture DNA, changes in the PCR reaction and in the protocol were tested in order to improve the outcome. Alterations in concentrations of DNA, primers, BSA or MgCl₂ were assessed, and different annealing temperatures and a touchdown programme (95 °C for 5 min, 10 cycles of 30 s at 95 °C, 45 s at 62 °C (decreased by 1 °C per cycle to 53 °C), and 72 °C for 90 s, followed by 30 cycles of 30 s at 95 °C, 45 s at 53 °C, and 72 °C for 90 s) were evaluated. PCR products were verified by agarose gel electrophoresis, and expected bands (~1300 bp) were gel-purified and sequenced as described above. In order to include the strains using the alternative dismutation pathway for syntrophic propionate oxidation [71], primers targeting Smithella-related species (Smi103-f: 5′-CGCGTGGATAATCTACCCCTC-3′; Smi442-r: 5′-GCTCGACAGAGCTTTACGGT-3′) were designed. 16S rRNA sequences of known syntrophs and closely related species were aligned in Geneious v6.1 with ClustalW, group- or species-specific primers were designed and the specificity was checked by Blast. The 16S rRNA specific primer pair Synt737-f (5′-CTGGAGAGGAAGGGGGAATT-3′) and Synt906-r (5′-ATGAGTACCCGCTACACCT-3′) was also designed to specifically study eventual presence of S. wolinii, S. pfennigii, S. fumaroxidans and S. sulfatireducens. In targeting the 16S rRNA gene, the PCR protocol consisted of initial denaturation at 94 °C for 5 min, 40 cycles of denaturation at 94 °C for 30 s, annealing at 61 °C for 40 s, and elongation at 72 °C for 30 s, followed by a final extension step of 4 min at 72 °C.

The mcrA genes of the isolates M. bourgensis sp. BA1, MAB1, MAB2 and MAB3 [17] were obtained using the forward primer ME1 [72] or primer mlas together with the reverse primer mcrA-rev. The amplification was performed with genomic DNA extractions from pure cultures, using the PCR manual described elsewhere [65].

Nucleotide sequence accession numbers

All sequences were deposited in the NCBI GenBank database under the following accession numbers: GenBank: KJ701041-KJ701114 for all fhs sequences and GenBank: KJ701115-KJ701185 for all mcrA sequences. The mcrA sequence data for M. bourgensis strains BA1, MAB1, MAB2 and MAB3 are available at the accession numbers GenBank: KJ708787-KJ708790.