The rumen microbiome is responsible for methane emissions from ruminants, including the huge number (> 1 billion) of sheep across the globe . The growth, abundance, and metabolic activities are reflected by DNA replication at the DNA level and transcription at the RNA level. This study explored the potential associations of methane emissions from two groups of sheep (with low or high methane yield) with the growth rate of rumen bacteria, in addition to their abundance and transcriptional activities. The elevated relative abundance of M. elsdenii (a lactate utilizer ) in the LMY sheep reflects its enrichment in the rumen of LMY sheep, as reported by Kamke J, et al. . In addition, our result showed that I. porci, which is a lactate producer recently isolated from the intestine of swine , was substantially more predominant in the LMY sheep than in the HMY sheep (21.5% vs. 1.5%). Kamke J, et al.  reported that the LMY rumen was highly enriched with Sharpea azabuensis, which was initially isolated from the feces of thoroughbred horses and produced lactate . They used 16S rRNA gene sequences to calculate the relative abundance of individual bacteria, whereas in the present study we calculated the relative abundance of MAGs that were at least 70% complete, and none of the MAGs that met the MAG criteria was classified as S. azabuensis. The classification based on 16S rRNA gene sequences vs. MAGs might also be a reason for the discrepancy. S. azabuensis and I. porci are within the same order (Erysipelotrichales), and they share 91% similarity in their 16S rRNA gene sequences . They also share the major fermentation product, lactic acid [43, 44]. Thus, even though the study presented here did not identify any MAGs (> 70% complete) that could be assigned to S. azabuensis, the enrichment of lactate-producing bacteria of the Erysipelotrichales in the rumen of the LMY sheep was consistent in both studies. In addition to M. elsdenii and S. azabuensis, F. prausnitzii and I. porci may serve as potential biomarkers of methane production by the rumen microbiome.
F. prausnitzii (formerly Fusobacterium prausnitzii) is a butyrate producer, and it does not produce hydrogen , which is the primary reducing power of hydrogenotrophic methanogenesis in the rumen when fermenting sugars. It can also utilize lactate and produce butyrate consuming hydrogen . Intriguingly, the LMY sheep had a lower relative abundance and growth rate (as estimated as PTR from metagenomic sequence data) of F. prausnitzii compared to the HMY sheep. F. prausnitzii has not been detected as a major species of the rumen bacteria in metataxonomic studies. However, it was represented by contigs in a metagenomic study of the rumen microbiome of Holstein cows, and it was shown to decrease in response to methane-mitigating diets , which agrees with the low abundance of this species in the LMY sheep. Our metatranscriptomic analysis, interestingly, revealed the upregulation of genes involved in carbohydrate metabolism, including pyruvate-formate lyase, enol-CoA dehydrogenase, both of which are involved in butyrate production, alcohol dehydrogenase, and oxaloacetate decarboxylase in the LMY sheep compared to the HMY sheep. It is also intriguing that the LMY sheep had significantly upregulated expression of the gene encoding the bacteriophage protein gp37 (Table 2), which belongs to the superfamily Radical S-adenosylmethionine  that was reported to activate pyruvate formate-lyase . The PTS was also upregulated in the LMY sheep. It is not known if all the genes of this butyrate fermentation pathway need to be upregulated to increase butyrate production or these two enzymes mediate the rate-limiting steps of the pathway, but the upregulation of these two enzymes might suggest increased butyrate production by F. prausnitzii in the rumen of the LMY sheep. Nevertheless, the upregulated expression of the F. prausnitzii genes involved in carbohydrate metabolism is consistent with an in-vitro study , which showed that carbohydrate metabolism was upregulated while DNA replication was downregulated in F. prausnitzii when expressed to cell-free supernatant of lactic acid bacteria. Although the present study did not detect any significant downregulation of growth-related genes of F. prausnitzii, the lower growth rate of F. prausnitzii in the LMY sheep than in the HMY sheep suggests that the growth-related genes of F. prausnitzii were downregulated in the LMY sheep. Also, the expressions of two tRNA synthetase (i.e., isoleucyl − tRNA synthetase, EC: 184.108.40.206; and tyrosyl − tRNA synthetase, EC: 220.127.116.11) genes selected by the TopKLists analysis (Suppl Table 3) showed downregulation in the LMY group, which concur with the downregulation of growth-related genes of F. prausnitzii. The lactate concentration in the rumen of the LMY sheep was much higher than in the HMY sheep (~ 0.9 vs. 0.014 mM) . F. prausnitzii, as shown in the present study and the study by Lebas M, et al. , might adjust its metabolism towards carbohydrate metabolism when exposed to lactate. Its ability to utilize lactate  might be one plausible explanation.
Based on the results of the present study and previous studies [8, 50], we proposed a working model to illustrate the growth-related mechanism to explain promoted butyrate formation by F. prausnitzii in the rumen enriched with lactate-producing bacteria and how they collectively influence methane production (Fig. 4). In this mechanism, lactate-producing bacteria, such as S. azabuensis and I. porci, were enriched in the rumen microbiome of the LMY sheep. The lactate produced (~ 0.9 mM) was then utilized by butyrate producers, such as M. elsdenii and F. prausnitzii, in producing butyrate, which reduced the metabolic hydrogen available for methanogenesis. Specifically, after metabolites (probably lactate) were produced by lactate-producing bacteria, the expression of carbohydrate metabolism-related genes of F. prausnitzii was a high priority, and butyrate production was increased, but the growth of F. prausnitzii might have been simultaneously suppressed, as demonstrated previously . In contrast, in the rumen of the HMY sheep, butyrate formation was not promoted because the abundance of lactate-producing bacteria and lactate concentration (0.014 mM) were low. This could result in increased acetate production and methanogenesis. Indeed, the rumen of the HMY sheep had a higher concentration of acetate but a lower concentration of butyrate compared to the rumen of the LMY sheep, even though the differences did not reach statistical significance .
As sequencing cost continues to decrease, sequencing coverage depth in metagenomic studies has increased substantially, allowing for more and more MAGs to be obtained from complex microbiomes. Indeed, nearly 5,000 MAGs were recovered from the rumen  and more than 10,000 MAGs were recovered from different segments of the gastrointestinal tract, including the rumen . The approach used in this present study can be used to infer the growth of individual bacteria represented by MAGs in metagenomes in future studies so that their population dynamics can be determined and taken into consideration in interpreting microbiome data. A recent study showed that the accuracy of the DEMIC method was high for fast-growing bacteria, but relatively low for slow-growing bacteria . Continued improvement in bioinformatics and statistical tools can further improve the estimation of bacterial growth using metagenomic data.