Community Structure Analysis of Methanogens Associated with Rumen Protozoa Reveals Bias in Universal Archaeal Primers

Protozoan sampling and DNA extraction.Rumen fluid was obtained from four 10-month-old rumen-cannulated Black Angus heifers that were fed grass hay. Samples (250 ml from each of four sites, including the reticulum, ventral and caudal sacs, and the dorsoventral midline) were pooled and strained twice through two layers of PTEX mesh (355-μm pore size; Sefar Inc., Kansas City, MO). To allow for separation, samples were incubated at 39°C for 30 min in a sealed container that was occasionally allowed to vent positive pressure. The top plant debris layer was discarded, and protozoa were separated by filtration through NITEX mesh (11-μm pore size; Sefar Inc.). Protozoa (∼10 ml of biomass per sample) were exhaustively washed (until free-living methanogens were no longer evident as assessed by microscopy) with sterile anoxic basal salt solution (pH 6.8 to 7.0; per liter, 2.0 g NaCl, 4.9 g K₂HPO₄, 3.8 g KH₂PO₄, 0.07 g MgSO₄ · 7H₂O, and 0.05 g CaCl₂ · 2H₂O), and protozoa were fixed in 70% (vol/vol) ethanol and stored at −20°C until processed for DNA extraction as previously described (9).

Construction and sequencing of clone libraries.Clone libraries were constructed separately for each of the four animals (numbered 5 to 8) with five different primer sets (designated A to E), resulting in a total of 20 sublibraries. Partial fragments of the SSU rRNA gene were amplified using universal archaeal primers as follows: library A, Met86f (25)/Met915r (22); library B, 21f/958r (3); library C, 1Af/1100Ar (24); library D, A109f (23)/Met915r; library E, A109f/958r. Each 50-μl PCR mixture contained 1.25 U Ex Taq DNA polymerase (TaKaRa Bio Inc., Japan), 1× PCR buffer with Mg²⁺, 0.8 mM deoxynucleoside triphosphates (dNTPs), 0.5 μmol of each primer, and 100 ng of DNA. Cycling conditions included an initial denaturation at 94°C for 30 s, followed by 30 cycles of 98°C for 10 s, the annealing temperature (55°C for libraries A, B, D, and E and 60°C for library C) for 30 s, and 72°C for 1 min, and a final elongation of 5 min at 72°C. Amplified products were cloned into the pCR2.1 TOPO vector (Invitrogen Canada Inc., Burlington, ON, Canada) and transformed into Escherichia coli Top10 cells according to the manufacturer's directions. Approximately equal numbers of randomly selected clones were sequenced from each sublibrary (a total of ∼60 clones per primer set). Plasmid DNA extraction and capillary sequencing were performed by Functional Biosciences (Madison, WI), using M13 forward and reverse primers.

Analysis of SSU rRNA clone libraries.Sequences were compared to entries in the GenBank nr database (filtered to exclude sequences from environmental and uncultured clones) by using BLASTn (27). Chimeric sequences were identified using Bellerophon v.3 (8). Nonmethanogen and chimeric sequences were eliminated from the data set. Sequences were aligned with ClustalW (20), and distance matrices were calculated according to the Kimura-2 parameter algorithm using Mega5.05 (18). Sequences were assigned to operational taxonomic units (OTUs) based on ≥97% sequence similarity using the furthest-neighbor algorithm as implemented in mothur v.1.20.0 (14). Phylogenetic trees were constructed in Mega5.05 using the neighbor-joining method and bootstrap analysis of 1,000 replicates. Community structures of the libraries were compared using ∫-Libshuff (13); P values were corrected using Bonferroni's correction, and significance was defined at a P value of <0.05.

Real-time qPCR.Partial fragments of the SSU rRNA gene were amplified using archaeal primers sets A to E as described above. The SSU rRNA gene copy numbers of total methanogens, Methanobrevibacter spp., RCC, and Methanomicrobium spp. were then quantified via nested quantitative PCR (qPCR) using the taxon-specific primer sets NestmetF (AMGWTCCAGGCCCTACGG)/NestmetR (TGGCACCSGTCTTRCCC), NestMbbF (TGGGAATTGCTGGWGATACTRTT)/NestMbbR (GGAGCRGCTCAAAGCCA), NestRCCF (TTCTGGGGTAGGGGTAAAATC)/NestRCCR (GTCTGCAGCGTTTACACCCT), and NestMmF (TGTTTAAAACACATGGGAAGA)/NestMmR (ATTCCCAGTATCTCTTAGACGC), respectively. All qPCRs were carried out in 20-μl volumes that contained 1× Brilliant II SYBR green qPCR master mix (Stratagene, Cedar Creek, TX), 0.5 μM respective forward/reverse primers, and 1 μl of the primary PCR product (diluted 1/10,000). Quantitative PCR was conducted using an Mx3005P Stratagene thermocycler. Amplification conditions were 1 cycle at 95°C for 10 min followed by 30 cycles of 95°C for 30 s, 60°C for 30 s, and 72°C for 30 s. Melting curve analysis was conducted over a range of 55 to 95°C to assess specificity of the amplification products.

In total, 298 sequences were obtained for all five libraries (i.e., approximately 60 clones per each of five primer sets). Twenty-two sequences (7.4%) were chimeras or nonmethanogens and were removed from the final data set (276 sequences). The distribution of the sequences according to genera for each library is shown in Fig. 1. Overall, 94 to 100% of the sequences in each library belonged to the three main genera/taxa which are most common in the rumen, namely, Methanobrevibacter, rumen cluster C (RCC), and Methanomicrobium (10). According to ∫-Libshuff analysis, the community structures of the libraries differed significantly from one another, with the exceptions of libraries A and D being similar and libraries B and C being similar. This indicates that the choice of primers used to construct the libraries vastly biased the species composition. For libraries A and D, sequences belonging to Methanobrevibacter (41 to 57%) and RCC (34 to 55%) were predominant (Fig. 1). Similarly, Methanobrevibacter and RCC sequences were the predominant sequences amplified by primer sets A and D as determined by qPCR analysis (Table 1). In contrast, sequences related to Methanomicrobium were predominant in libraries B and C (85 to 91%; Fig. 1) and were the predominant sequences amplified by these primer sets as determined by qPCR analysis (Table 1). Library E consisted almost exclusively of sequences belonging to Methanobrevibacter (98%; Fig. 1). Likewise, qPCR analysis indicated that Methanobrevibacter spp. represented the predominant sequence amplified by primer set E (Table 1). Similar to the results of other studies of PAMs, Methanobrevibacter spp. have been identified as the predominant methanogens (88 to 99%) associated with protozoa of sheep and goats in three different studies (2, 9, 21), while Methanomicrobium spp. were the predominant taxa (85%) associated with protozoa of sheep and cattle in one other study (12). Given the observed effect of primer bias on species composition, it is likely that differences in the methanogen species composition between these previous studies and this study could be related to the use of different primer sets among all of the studies rather than differences in host factors (e.g., animal species, genetic variation, and diet). To explore this further, the primers were analyzed for potential nucleotide mismatches with target sequences which could lead to biased amplification of certain species (Table 2). Target sequences included those from representative type strains of the main species for each genus. As RCC has yet to be taxonomically classified, target sequences included those from all cultivated members (only four to date) and representative sequences from uncultured clones associated with various livestock. It should be noted that primers incorporate themselves into DNA sequences during PCR amplification; hence, only sequence data in between the primers were included in Table 2.

• Open in new tab
• Download powerpoint

Fig 1

Distribution of protozoan-associated methanogen clones for five libraries constructed using different SSU rRNA universal archaeal primers. Each segment represents a different OTU that was defined at >97% sequence similarity. Each library consists of ∼48 clones. RCC, rumen cluster C.

Importantly, RCC was well represented only in libraries A and D, which were constructed using Met86f (or the highly similar primer A109f) paired with Met915r. These three primers exhibit perfect matches with target sequences from common rumen methanogen species. In contrast, the primers 21f, 958r, and 1100Ar have multiple mismatches with RCC sequences, which would account for the poor amplification of RCC by these primers (Table 2). Interestingly, many of the reference RCC sequences that we retrieved from GenBank were amplified using Met86f and Met1340r as the forward and reverse primers, respectively (REFs), (6, 11, 19, 26), suggesting that this may be one of the only universal archaeal primer sets that amplifies this group. RCC has been reported to associate with protozoa by only one other group, and in that study, RCC accounted for ∼1.5% of the methanogens (9). This previous study utilized primers ArcF915 (sequence identical to that of Met915r), which exhibits a perfect sequence match with RCC sequences, and ArcR1326, which contains a 1-bp mismatch that may be responsible for the low number of RCC sequences detected compared to that in this current study.

In contrast, Methanomicrobium appears to be overrepresented in libraries B and C. While the primer sets used to construct these libraries have not been used to characterize PAMs, they have been previously used to study rumen methanogens (15, 24). Using the 21f/958r primers, Shin et al. identified Methanomicrobium (86%) as the predominant methanogen in cattle (15). These primers exhibit greater mismatches for Methanobrevibacter and RCC than for Methanomicrobium, accounting for the observed amplification bias. Using 1Af/1100Ar primers, Whitford et al. (in 2001) reported that Methanobrevibacter (59%), Methanosphaera (27%), and Methanimicrococcus (15%) were present in the rumen of cattle, while Methanomicrobium and RCC were notably absent (24). As noted above, primer mismatching with primer 1100Ar would account for the lack of RCC. As 1100Ar exhibits a 100% sequence match to both Methanobrevibacter and Methanomicrobium, it is likely that the observed bias is associated with primer 1Af. Unfortunately, it is not possible to assess 1Af for potential bias, since sequence data for the 5′ end of the SSU rRNA gene is likely not reliable, due to the incorporation of primer sequences into the DNA during PCR amplification. As such, we cannot determine whether differences between our results and those of Whitford et al. (24) are related to potential bias with primer Af1 or other methodological differences (e.g., sample preparation, animal diet, and genetics).

While Methanobrevibacter spp. are the most abundant rumen methanogens (10), they were overrepresented in library E, where they comprised 96 to 98% of the methanogens, as determined via cloning and qPCR, respectively. Sequences belonging to this genus can be divided into two main clades: the RO clade (representing Mbb. ruminantium and Mbb. olleyae) and the SGMT clade (representing Mbb. smithii, Mbb. gottschalkii, Mbb. millerae, and Mbb. thaurei) (10, 11). Library E contained sequences belonging only to the SGMT clade (Fig. 2), indicating a bias for this particular clade. The opposite bias was noted for library C, which contained only sequences belonging to the RO clade. Similarly, Methanobrevibacter sequences from the study by Whitford et al., who used these primers to characterize rumen methanogens, belong only to the RO clade (24). On the other hand, libraries A and D contained members of both the RO and SGMT clades. While the primer sets used to construct these libraries are more comprehensive than the others, slight differences between the two libraries in terms of phylogenetic groups within the SGMT clade (Fig. 2) indicate that a multiple-primer approach may be required for thorough characterization of community diversity and structure.

• Open in new tab
• Download powerpoint

Fig 2

Phylogenetic analysis of methanogen SSU rRNA sequences from protozoan-associated methanogens. The tree was constructed using one representative sequence from each OTU that was defined at ≥97% sequence similarity. GenBank accession numbers are presented in parentheses. The scale bar represents the number of nucleotide substitutions per base. The bootstrap consensus tree was inferred from 1,000 replicates, and bootstrap values (≥50%) are indicated on the tree. Methanobrevibacter was divided into two major clades: the RO clade (representing Mbb. ruminantium and Mbb. olleyae) and the SGMT clade (representing Mbb. smithii, Mbb. gottschalkii, Mbb. millerae, and Mbb. thaurei). RCC, rumen cluster C. As an example of the naming scheme for clones, clone A5-1 refers to primer set A, animal number 5, and clone number 1.

In conclusion, this study emphasizes the importance of primer selection to minimize bias during PCR-based analysis of methanogen community structure. We identified methanogens belonging to Methanobrevibacter (41 to 57%) and RCC (34 to 55%) as the predominant taxa associated with protozoa from cattle. Our results differ from the results of previous studies of PAMs, and this is likely related to the use of primers that exhibit biased amplification of certain taxa in the other studies. Of the primer sets we tested, Met86f (or A109f) and Met915r were the most comprehensive and represented the only set to detect members of the RCC and members of both the Methanobrevibacter RO and SGMT clades. In silico analysis of primer sequence mismatch also suggests that Met1340R may also display minimal primer bias. An accurate description of the ecology of ruminal methanogens could be key to the successful development of chemogenomic or reverse vaccinology approaches to curtailing ruminal methane production.

Nucleotide sequence accession numbers.Nucleotide sequences generated during this study were deposited in GenBank under the accession numbers JN315156 to JN315255 and JN595893 to JN596068.