|Home | About | Journals | Submit | Contact Us | Français|
The faecal microbiota of muskoxen (n=3) pasturing on Ryøya (69° 33′ N 18° 43′ E), Norway, in late September was characterized using high-throughput sequencing of partial 16S rRNA gene regions. A total of 16 209 high-quality sequence reads from bacterial domains and 19 462 from archaea were generated. Preliminary taxonomic classifications of 806 bacterial operational taxonomic units (OTUs) resulted in 53.7–59.3 % of the total sequences being without designations beyond the family level. Firmicutes (70.7–81.1 % of the total sequences) and Bacteroidetes (16.8–25.3 %) constituted the two major bacterial phyla, with uncharacterized members within the family Ruminococcaceae (28.9–40.9 %) as the major phylotype. Multiple-library comparisons between muskoxen and other ruminants indicated a higher similarity for muskoxen faeces and reindeer caecum (P>0.05) and some samples from cattle faeces. The archaeal sequences clustered into 37 OTUs, with dominating phylotypes affiliated to the methane-producing genus Methanobrevibacter (80–92 % of the total sequences). UniFrac analysis demonstrated heterogeneity between muskoxen archaeal libraries and those from reindeer and roe deer (P=1.0e-02, Bonferroni corrected), but not with foregut fermenters. The high proportion of cellulose-degrading Ruminococcus-affiliated bacteria agrees with the ingestion of a highly fibrous diet. Further experiments are required to elucidate the role played by these novel bacteria in the digestion of this fibrous Artic diet eaten by muskoxen.
This study gives a first glimpse into the faecal bacterial and archaeal microbiota harboured in faeces from muskoxen, giving a broad view of this highly complex microbial environment. Only one previous study has been conducted on the gut microbiome of muskoxen, describing their rumen eukaryotes. The results presented here are consequently an interesting complement to gain further insight into the microbial diversity housed in the gastrointestinal tract of muskoxen. The relatively high percentage of novel bacterial and archaeal phylotypes in the current study points to a need for new experiments to identify this putatively novel microbiota and their related functional gene contents. Taken together, the results presented here not only help increase knowledge on the microbial diversity occurring with highly fibrous diets but also on the microbiological aspects related to methanogenesis in Arctic ruminants.
Muskoxen (Ovibos moschatus) are one of few large, terrestrial mammals adapted to a high Arctic environment. It was once holarctically distributed, but became extinct over most of its Eurasian range at the beginning of the Holocene, and is currently found as relict natural populations mainly in Greenland, northern Canada and Alaska (Campos et al., 2010). In their natural habitat, muskoxen are typical grass and roughage eaters, feeding on grass heath communities or willows found on exposed ridges, slopes or plains. Their anatomy consists of a large rumen–reticulum (containing 34–78 % of the total alimentary tract contents), a large omasum and a relatively small caecal–colon complex, all contributing to extremely long retention times (Thing et al., 1987; Staaland & Thing, 1991; Staaland et al., 1997). For most of the year the animals graze on forages high in lignocellulose, whereas high-quality forages are only available during the short Arctic summer (Thing et al., 1987; Forchhammer, 1995 ).
Symbiotic microbial fermentation accounts for 79 % of the dry matter digestion in muskoxen and is consequently the main source of energy for the animal (Barboza et al., 2006). Only one study targeting polyadenylated eukaryotic mRNA in rumen samples from muskoxen fed a highly lignified diet has been performed thus far to characterize their microbiome (Qi et al., 2011). Only the eukaryotic fraction was analysed in that study, describing a very high percentage of cellulolytic enzymes, but the presence of bacteria and archaea (3.4 and 0.1 % of total RNA reads, respectively) was also reported (Qi et al., 2011). In the gut system, methanogenic archaea are the only micro-organisms responsible for methane production. Enteric methane emissions from muskoxen consuming brome hay comprised only 2.0–3.2 % of the gross energy intake (White & Lawler, 2002). Still, very little is known about the microbial ecosystem in the rumen and the hindgut of muskoxen. Considering the lack of knowledge regarding the microbial diversity in the digestive tract of this arctic ruminant, the present study aims to examine the diversity of archaea and bacteria in muskoxen faeces by applying a metagenomics approach. In addition, we include a comparison of the results obtained here with bacterial and archaeal datasets from other ruminant and non-ruminant herbivores to assess potential similarities/dissimilarities at a microbiota level.
Previous studies have proven the reliability of using faeces as a proxy for describing the hindgut microbiome (Steelman et al., 2012; de Oliveira et al., 2013), at least for bacteria. Faecal samples [Muskoxen Faecal Sample (MkFS)1, MkFS2 and MkFS3] were collected immediately after dropping from three adult muskoxen grazing on patches of cultivated grassland and heather (September pasture) in open birch and pine forest on the island of Ryøya (69° 33′ N 18° 43′ E) near Tromsø, Norway (Fig. 1a). The animals belong to a herd of muskoxen, kept at Ryøya since the arrival of the first animals in 1979/1980, after originally being imported from East Greenland to an inland location in northern Norway (Blix et al., 2011). Due to the territorial nature of this ruminant, samples were collected from a distance and therefore no gender or health status could be determined for each sample. The muskoxen faecal samples were immediately stored on ice and after no more than 3 h frozen at −80 °C prior to DNA extraction and molecular analysis.
DNA extraction was based on the protocol of the Repeated Bead Beating plus Column (RBB+C) method developed by Yu & Morrison (2004)), with minor modifications. DNA quantification was done using a NanoDrop 2000c spectrophotometer and solutions were stored at −20 °C until amplification by PCR.
PCR amplifications for Bacteria and Archaea were performed in an Eppendorf Mastercycler Gradient in 25 µl reaction volumes, with 12.5 µl of iProof High-Fidelity Master Mix (BioRad), 1 µl of each primer (400 nM), 1 µl of the corresponding DNA template and 1.25 µl DMSO to increase PCR efficiency.
Bacterial and archaeal 16S rRNA amplification was carried out with the bacterial primer set 27F and 515R (Pope et al., 2012), giving an around 500 bp amplicon product; and the archaeal primer set 340F and 1000R (Gantner et al., 2011), yielding an around 650 bp amplicon product. The reverse primer included an 8 nt Multiplex Identifier (MID) (Hamady et al., 2008) for sample identification in downstream analysis. Both primer sets contained the Life Sciences primer A and B sequences necessary for pyrosequencing. PCRs were run with an initial denaturation step at 98 °C for 30 s; followed by 25/35 cycles (Bacteria/Archaea) consisting of 98 °C for 10 s, 60 °C/58 °C (Bacteria/Archaea) for 30 s and 72 °C for 45 s, completed with a final extension step at 72 °C for 7 min. Amplicon size was assessed by 1.5 % agarose gel and DNA concentration was quantified using a Qubit fluorimeter (Invitrogen). Sample products were then pooled in equimolar amounts, run in a 1 % agarose gel electrophoresis, excised and purified from the gel using NucleoSpin Gel and PCR Clean-up kit (Macherey-Nagel). The resulting DNA was stored at −20 °C until sequencing. PCR amplicons were sequenced with a 454/Roche GS FLX device using LIB-L titanium chemistry, at the Norwegian Sequencing Centre (NSC) in Oslo.
Sequences for 16S rRNA genes from both microbial groups were analysed using the Quantitative Insights Into Microbial Ecology (QIIME) pipeline (Caporaso et al., 2010a). Firstly, barcode and primer sequences were removed and sequences were discarded when: length was < 350 or > 650 nt; the number of homopolymer runs exceeded 6 nt; average quality score was below 25; and mismatches in primers occurred, thus ensuring high-quality sequences. The operational taxonomic unit (OTU)-clustering criterion was set on a 3 % genetic distance cutoff using the QIIME-incorporated version of usearch (Edgar, 2010), with a word length of 64 and discarding those OTUs below four reads. To avoid any potential bias at the OTU-clustering step, all sequences were trimmed to a similar 500 nt length. Chimeric sequences were identified using the uchime (Edgar et al., 2011) tool in QIIME and discarding any sequence flagged as a putative chimera.
For interspecies comparisons between muskoxen faecal libraries and other datasets, the sequences were previously edited to obtain similar length and orientation. Muskoxen and Norwegian reindeer (Table S1, available with the online Supplementary Material) bacterial and archaeal libraries were reversed in orientation and we took the complementary sequence as sequencing was applied only on amplicons obtained from the reverse primer. In some instances, samples from the rumen were included for comparison among the bacterial libraries due to the scarcity of publicly available datasets from faeces or the lower intestine (Table S1). Samples with substantial differences in their total counts were finally included in the comparisons for the archaeal libraries. Accordingly, OTU clustering for these archaeal libraries included singletons (i.e. OTUs containing a single sequence) in the analyses to include as much microbial information as possible.
The most abundant sequence for each OTU was taken as representative and subsequently aligned against the Greengenes core-set reference database using the Python-based version of the Nearest Alignment Space Termination (NAST) algorithm (Caporaso et al., 2010b), with a minimum length of 150 nt and 75 % similarity cutoff. Taxonomic assignment down to the genus level for all the aligned chimera-free OTUs was performed using the RDP classifier tool (Cole et al., 2003) in QIIME, which applies a Naïve-Bayesian algorithm on 8 k-mers at a default 80 % identity cutoff using the RDP-II database as reference taxonomy. Rank abundance plots evaluating sample richness and evenness were generated with the plot_rank_abundance_graph.py script in QIIME. Alpha-diversity analyses assessing species richness (Chao1) (Chao, 1984), evenness (Shannon-Wiener) (Shannon, 1948), total observed species and sample coverage (Good’s coverage) (Good, 1953) were performed on randomly subsampled datasets from each sample, and resulting rarefaction curves were obtained with the make_rarefaction_plots.py script. Pairwise sample dissimilarity analyses (beta diversity) were performed using unweighted UniFrac distance matrices (Lozupone & Knight, 2005) calculated with subsampled datasets adjusted to the sample yielding the lowest sequence counts. For interspecies comparisons between archaeal datasets from different animals, no rarefaction was performed due to the considerable differences in dataset size mentioned above. Principal coordinates for each dataset were calculated based on UniFrac distance matrices and principal coordinates analysis (PCoA) plots were generated. OTU network maps were created using the make_otu_network.py script in QIIME, and visualized with the Cytoscape (v3.1.1) platform (Shannon et al., 2003). Statistical differences between pairs of sample datasets were assessed with unweighted UniFrac phylogenetic tree distances calculated based on iteration (Monte Carlo randomizations, 100 times) using the beta_significance.py script in QIIME. Heatmap analysis for comparisons with the different archaeal and bacterial datasets was done calculating the standard score (z-score) of the different phylotypes. Plots were generated with a customized version of the heatmap.2 script within the ‘gplots’ package in the ‘R’ software (R Development Core Team, 2008).
A total of 16 209 500 bp-trimmed high-quality 16S rRNA gene sequences were obtained from the faeces of three adult muskoxen (MkFS1: 6 527; MkFS2: 4 987; MkFS3: 4 695). OTU clustering based on a 97 % similarity criterion resulted in 806 chimera-free OTUs, with 393 OTUs [74.2 % of the total sequences (12 031 seqs)] shared by the three animals (Fig. S1). These shared OTUs comprised 76.3 % (4 982 seqs), 68.1 % (3 398) and 77.7 % (3 651) of the total sequences in the MkFS1, MkFS2 and MkFS3 libraries, respectively (Table 1). Rank abundance plots showed steep curves for the three samples, thus indicating low sample evenness (Fig. 2a). Similar trends were also seen with rarefaction curves based on alpha diversity parameters (Fig. S2). Pairwise comparisons with unweighted UniFrac indicated statistical differences between the three samples (P<0.05) although not between MkFS1 and MkFS3 when corrected (P=0.078, Bonferroni corrected).
The bacterial communities were mostly dominated by Firmicutes (70.7–81.1 % of the total sequences) and Bacteroidetes (16.8–25.3 %), whereas Tenericutes (0.9–2.5 %) and Cyanobacteria (0.3–0.5 %) were also represented, albeit in minor proportions (Fig. 1b and Table S2). The Firmicutes and Bacteroidetes are also dominant in faeces from the cattle and horse hindgut (Dowd et al., 2008; Steelman et al., 2012). At family/genus level, Ruminococcaceae constituted the major Firmicutes-affiliated family, ranging from 43.8 to 51.9 % of the total sequences (Fig. 1b; Table S2). Around 28.9–40.9 % of the total sequences could not be designated to any characterized genus within this family. Of the shared sequences, uncharacterized genera belonging to Ruminococcaceae (24.9–39.3 %) and Lachnospiraceae (3.9–9.8 %) were dominant (Table 1). Several members related to Ruminococcaceae play a key role in the degradation of recalcitrant polysaccharides such as crystalline cellulose (Rincón et al., 2001; Ben David et al., 2015). Free-ranging muskoxen typically graze on highly lignified plants during winter, which may explain the high proportion of Ruminococcaceae-related constituents observed in this study (Fig. 1b and Table S2). Phylotypes affiliated to the family Lachnospiraceae constituted the second major group within Firmicutes (9.4–19.8 %), with Roseburia as the major genus (0.4–1.8 %). Uncharacterized genera within Lachnospiraceae constituted 5.2 % of total bacterial counts, on average (3.4–7.5 %). Several Bacteroidetes-related families such as Bacteroidaceae (3.8–4.2 %), Rikenellaceae (1.7–2.5 %) and Prevotellaceae (0.5–1.1 %) were also identified, with uncharacterized lineages affiliated to the order Bacteroidales accounting for 6.5–12.3 % of the total sequences (Fig. 1b and Table S2). Uncharacterized phylotypes within the order Bacteroidales accounted for 3.3–8.2 % of shared sequences (Table 1).
Overall, uncharacterized bacteria constituted 53.7–59.3 % of the total bacterial sequences. These relative proportions were higher than reported for steer fed diets with different fibre contents (Fernando et al., 2010; De Oliveira et al., 2013). A high proportion of novel bacterial phylotypes has also been described in the rumen of Norwegian and Svalbard reindeer (Pope et al., 2012). Metatranscriptomics describing the eukaryotic fraction in the rumen of muskoxen reported a remarkable 17 % of carbohydrate active enzyme (CAZy) genes identified as ‘putative’ or ‘predictive proteins’, not found in any other rumen metagenome (Qi et al., 2011). The high relative proportion of uncharacterized bacteria found in muskoxen faeces suggests the existence of novel bacterial phylotypes, which may be involved in the digestion of fibrous plants found at Arctic latitudes.
A total of 19 462 500 bp-trimmed high-quality 16S rRNA gene sequences were obtained from the three faecal samples (MkFS1: 6 320 sequences; MkFS2: 5 736; MkFS3: 7 406). OTU clustering gave 37 chimera-free OTUs, and 34 of these were shared by the three animals (99.1 % of the total sequences) (Fig. S1). Rank abundance plots showed low sample evenness and sequence abundance dominated by few archaeal phylotypes (Fig. 2b). This was corroborated by rarefaction curves based on several alpha diversity parameters (Fig. S2). Euryarchaeota was the only phylum detected in this study, with Methanobacteriaceae as the major family encompassing 83.6–96.3 % of the total sequences (Fig. 1c). Within this family, 80–92 % of the total sequences were designated to the genus Methanobrevibacter and 3.3–4.4 % as Methanosphaera-related phylotypes (Fig. 1c and Table S3). A dominance of Methanobrevibacter species is common in several other ruminant and non-ruminant herbivores in both rumen and faeces, without diet specificity (Sundset et al., 2009a; Liu et al., 2012; Turnbull et al., 2012; Li et al., 2014; Cersosimo et al., 2015). High abundances of Methanobrevibacter species have been associated with high methane outputs (Wallace et al., 2015). In contrast, muskoxen were reported to possess net methane emissions generally lower than for domesticated animals such as sheep and cattle (Blaxter & Clapperton, 1965; Johnson & Ward, 1996). The consumption of diets rich in plant secondary metabolites, such as the diet commonly eaten by muskoxen, may also negatively affect methanogenesis (Bodas et al., 2012). Considering that enteric methane is mostly produced in the rumen (Murray et al., 1976), any potential estimation of the methanogenesis ratio for comparisons with other ruminants based solely upon archaeal community structure compositions from faeces should be attempted with caution.
Of the total sequences, 3.7–16.3 % were allocated to uncharacterized archaeal members of the family Methanomassiliicoccaceae (Fig. 1c and Table S3), representing the greatest source of variation between the samples; however, UniFrac-based beta diversity comparisons showed no statistical differences among the samples (P=1.0, Bonferroni corrected). This family belongs to group E2 within the class Thermoplasmata (Dridi et al., 2012; Paul et al., 2012). Several members of this class were also found at high proportion in other Arctic ruminants, such as Norwegian and Svalbard reindeer rumen (Sundset et al., 2009a, b). Only one study has reported the presence of Methanomassiliicoccaceae-related phylotypes in the large intestine of herbivores (Luo et al., 2013). In particular, group E2 methanogens produce methane mainly using methanol as carbon source (Gorlas et al., 2012; Iino et al., 2013). Methanol can be produced by the hydrolysis of methyl esters from pectin, and this can be degraded by genera associated with the Bacteroides, Lachnospira and Ruminococcus (Gradel & Dehority, 1972; Osborne & Dehority, 1989). The presence of Methanomassiliicoccaceae-related members in muskoxen faeces may partly be explained by the metabolism of methanol produced by fibre-degrading bacteria, mostly associated with the Firmicutes.
Comparative analyses were conducted between our muskoxen faecal bacterial and archaeal datasets with libraries from other ruminants to search for differences in their microbial traits (Fig. 3 and Table S1).
OTU Network mapping and PCoA (Fig. 3a,b) showed a higher number of bacterial OTUs shared between the muskoxen and reindeer datasets compared with other ruminants included in the analysis. Shannon index values were higher in muskoxen and reindeer libraries than in the other datasets, except rumen samples from sheep (fed grains or pellets) (Kittelman et al., 2013) (Fig. S3). Sampling site (faeces, caecum or rumen) also had a stronger influence on sample clustering with no significant differences observed between muskoxen faeces, reindeer caecum and cattle faeces (Durso et al., 2010;Klein-Jöbstl et al., 2014-)(Table S4). Instead, when considering the libraries together, only reindeer libraries did not show statistical differences from muskoxen (Table S4). OTU heatmap analysis with hierarchical clustering based on z-score profiles of the major bacterial phyla from each animal supported these findings, showing sub-clustering with samples from muskoxen and reindeer (Fig. 4). These results suggest that muskoxen and reindeer possessed comparable bacterial profiles. Despite the differences in diet (muskoxen: pasture; reindeer: grain-based diet), these animals shared a similar Arctic environment, which may account for their similar bacterial profiles; however, the dataset from moose from northern Norway was different from that from muskoxen. The moose samples were originally collected from the rumen instead of the caecum or faeces (Ishaq & Wright, 2014). The difference in sampling site would partly explain the dissimilarities of moose libraries with those from ruminants co-habiting similar environments (muskoxen and reindeer). The same was observed for other animals phylogenetically related to muskoxen, such as sheep or goats (Lee et al., 2012; Kittelmann et al., 2013)(Table S4). Sampling site has been shown to greatly influence microbial profiles (de Oliveira et al., 2013). Future comparative analysis should consider the effects produced by this parameter on interpretation of the results.
OTU Network mapping revealed a higher degree of OTUs shared by muskoxen and Norwegian reindeer fed with pellets concentrate (Fig. 3c). Unweighted UniFrac-based PCoA plots also displayed closer clustering for samples from these two Arctic ruminants (Fig. 3d), with overall diversity being higher in both datasets compared with other libraries (Fig. S4). In contrast, OTU-Heatmap analysis resulted in muskoxen faecal samples branching separately (Fig. 5). UniFrac-based statistical tests with libraries treated separately or together corroborated the differences observed between the two Arctic ruminants (P=1.0e-02, Bonferroni corrected) (Table S4). Unexpectedly, no statistical differences were found between muskoxen faeces and samples from hindgut fermenters such as horse, pony and white rhinoceros (P>0.05). These libraries were obtained with samples collected from animals fed diets constituted by a high proportion of fibre supplemented with concentrated feed similar to that for muskoxen, and different from that for reindeer (grains). Nonetheless, Methanocorpusculum labreanum-related phylotypes were dominant in these hindgut fermenters (Luo et al., 2013; Lwin & Matsui, 2014), whereas no representatives of Methanomicrobia were found in muskoxen. The libraries from hindgut fermenters possessed substantially lower sequence counts compared with muskoxen (Table S1), with lower sequencing depths; however, a small sample size was also observed in rumen dataset from Tibetan yak and cattle (Huang et al., 2012) (Table S1), which yielded statistical differences with muskoxen libraries (Table S6). Whether sequencing depth may have driven the lack of statistical differences observed among these datasets (muskoxen and hindgut fermenters) remains to be clarified.
This study is the first investigation into the faecal microbiome structure in the muskoxen. The large number of uncharacterized bacteria described here along with previous reports of novel features described for the eukaryotic fraction in their rumen microbiome (Qi et al., 2011) emphasizes the potential for the bacterial microbiome housed by this Arctic ruminant. Interest in the mining of these novel enzymes involved in polysaccharide degradation is also conceivable, given the high efficiency of these microbiomes to deconstruct low-quality forages and enabling the host to survive under austere nutritional conditions.
Hans E. Lian and Elin Olsson are thanked for their help with sampling the muskoxen at Ryøya, as well as Lorenzo Ragazzi for providing the picture of the semi-domesticated Muskoxen. Funding by UiT - The Arctic University of Tromsø, Norway. PBP was supported by a grant from the European Research Council (336355-MicroDE).
Supplementary File 1