Constructing a GIT microbial gene catalog in dairy cattle
We collected 120 content samples covering 10 GIT regions from 12 dairy cattle fed forage- and grain-based diets. Over 2.7 terabytes (Tb) of metagenomic data was obtained with an average of 22.8 gigabytes (Gb) per sample, totaling 18.2 billion sequencing reads with a length of 150 bp (Additional file 1: Table S1). Following the removal of host and diet DNA contamination, 1.6 Tb of the GIT microbial data remained. We then obtained 90.7 million contigs and 153 million open reading frames (ORFs) via metagenomic assembly and ORF prediction (see “Methods” section). After clustering at 95% nucleotide sequence identity, we generated a non-redundant microbial gene catalog of 45,886,195 genes (average length 607 bp). According to the currently available databases, only half of the genes (48.8%, 22,398,032) were taxonomically classified, of which, 97.7% could be assigned to bacteria, and the remaining genes were classified as archaeal (0.58%), eukaryotic (0.30%), and viral species (0.17%). We also found that 61.4% (28,170,604), 62.3% (28,574,080), and 3.7% (1,692,7920) of genes were annotated as clusters of orthologous groups of proteins, KEGG orthologous groups (KOs), and carbohydrate-active enzymes (CAZymes), respectively. This gene catalog represents a comprehensive GIT microbiome of dairy cattle and was used in subsequent studies.
Microbial composition landscape along the GIT of dairy cattle
To explore the segmental organization of microbial communities, we compared our gene catalog in 10 different regions and found that the GIT samples were partitioned into three distinct clusters, corresponding to different physiological areas, including the four-chambered stomach, small intestine, and large intestine (Fig. 1a). Notably, we found that the gastrointestinal microbiota exhibited a stable differential distribution along the GIT without modification by diet. Principal coordinate analysis (PCoA) revealed that region exerted a more pronounced effect on the separation of GIT microbial composition than diet (PERMANOVA, Fregion = 27.2, Fdiet = 9.4, Fregion × Fdiet interaction = 1.5, p < 0.05). This diet-independent pattern was observed in the alpha diversity index, which was highest in the four-chambered stomach (rumen, reticulum, omasum, and abomasum) and lowest in the small intestine (duodenum, jejunum, and ileum) (Wilcoxon rank-sum test, p < 0.01; Fig. 1b and Additional file 2: Fig. S1a). In addition, beta diversity was lowest in the four-chambered stomach, demonstrating an opposite trend to alpha diversity across the GIT regions in the forage-fed cows. However, no difference was observed between the four-chambered stomach and large intestine (cecum, colon, and rectum) of cows fed a grain-based diet (Fig. 1c and Additional file 2: Fig. S1b), which may be due to greater microbiome dissimilarity in the four-chambered stomach of grain-fed cows.
We next characterized the prevalent taxa of the GIT microbiome and similarly found region-specific patterns independent of diet regimes (Additional file 1: Table S2, S3). The GIT microbiome was mostly assigned to bacterial taxa, which were dominated by Firmicutes (46.5%), Bacteroidetes (33.6%), Proteobacteria (7.5%), and Fibrobacteres (3.0%) (Additional file 2: Fig. S2). The taxa classified into Bacteroidetes and Fibrobacteres were dominant in the four-chambered stomach, whereas the Firmicutes and Proteobacteria taxa were dominant in the post-gastric intestine (the small and large intestine) (Fig. 1d). Prevotella spp. and Fibrobacter spp. were enriched in the four-chambered stomach, Phyllobacterium spp. and Eubacterium spp. were more abundant in the small intestine, and Bacteroides spp. and Alistipes spp. were prevalent in the large intestine (Wilcoxon rank-sum test, p < 0.05) (Fig. 1f and Additional file 1: Table S3). Euryarchaeota occupied 99.5% of the annotated sequences that were assigned to archaea among the GIT regions (Wilcoxon rank-sum test, p < 0.05) (Fig. 1d and Additional file 1: Table S2). Notably, Methanobrevibacter spp. were the most prevalent archaea across all GIT regions and were mainly enriched in the post-gastric intestine, whereas Methanocorpusculum spp. were enriched in the large intestine (Fig. 1f). Nematoda were the dominant eukaryotic taxa in the GIT, with a relatively higher abundance in the post-gastric intestine (Fig. 1e). The top genera belonging to ciliates were Oxytricha spp., Stylonychia spp., and Entodinium spp., which were prevalent in the four-chambered stomach, whereas Anaeromyces spp. and Caecomyces spp., classified as Neocallimastigomycota, were relatively enriched in the large intestine (Fig. 1f). These results indicate that the GIT microbiome has a region-specific distribution across the physiological regions of dairy cattle, which represents segmental enrichment of distinct functional taxonomic groups, and may be related to polysaccharide utilization, H2 transfer, and fiber degradation in the GIT.
Microbial functional landscape along the GIT of dairy cattle
To characterize the microbial function of the GIT microbiome in dairy cattle, we performed predictive analyses of metabolic enzymes, focusing on 7861 KOs and 335 CAZyme families from the GIT microbial gene catalog. We first investigated the core metabolic pathways, wherein KOs were present in at least 90% of the four-chambered stomach, small intestine, and large intestine samples. Of these, 377 core pathways from 1989 KOs were shared by the whole GIT microbiome, including biosynthesis of cofactors, ABC transporters, carbon metabolism, biosynthesis of amino acids, two-component systems, and methane metabolism (Additional file 1: Table S4). When exploring specific KOs in the four-chambered stomach (788), small intestine (78), and large intestine (193), we observed enrichment of purine metabolism, N-glycan biosynthesis, and glycerophospholipid metabolism in the four-chambered stomach; enrichment of oxidative phosphorylation, arachidonic acid metabolism, and tryptophan metabolism in the small intestine; and enrichment of antibiotic biosynthesis, galactose metabolism, carbon metabolism, and amino acid metabolism in the large intestine (Additional file 1: Table S4). These results suggest that the GIT microbiome exhibits substantial functional heterogeneity. Notably, we found segmental differences in the prokaryotic ABC transporters associated with transporting carbon-derived nutrients. For example, the four-chambered stomach microbiome seems to be particularly responsible for transporting cellobiose, L-arabinose, lactose, alpha-glucoside, sorbitol, mannitol, mannose, and glucose. Those in the large intestine were particularly associated with trehalose, maltose, and fructose transport, and no specific transporters were found in the small intestinal microbiome (Additional file 1: Table S4). These results highlight that the GIT microbiome of dairy cattle has a region-specific prevalence of trophic transport systems, which are associated with differences and complexities of nutrient substrates in distinct GIT regions.
Ruminant GITs possess an efficient microbial polysaccharide degradation system in which many glycoside hydrolases (GHs) and polysaccharide lyases are involved [11]. Thus, we characterized region-specific features of polysaccharide degradation among the GIT regions. We first assigned 761,245 CAZymes to 119 GH (95.7%) and 22 polysaccharide lyase (4.3%) families and then classified them into functional groups based on their main polysaccharide targets (Fig. 2a and Additional file 1: Table S5). We found that enzymes belonging to the amylase family, GH13, and hemicellulase families, GH43 and GH3, were the three most prevalent across all GIT segments. Interestingly, we found that most CAZyme families had a region-specific distribution, regardless of the diet regime. For example, the cellulase, hemicellulase, and pectinase families showed a higher representation in the four-chambered stomach (Wilcoxon rank-sum test, p < 0.001). Microbe-derived glycans (e.g., peptidoglycan and chitin) were likely important substrates for the microbiota in the small intestine, which was supported by the enrichment of lysozyme (GH25 and GH24) and chitosanase (GH19). Notably, most host glycan-degrading families (e.g., GH109, GH92, and GH20) were enriched in the large intestine. However, the major peptidoglycan-degrading family (GH23) did not differ between the microbiomes of the four-chambered stomach and large intestine in forage-fed cows but was most prevalent in the four-chambered stomach under a grain-based diet (Scheirer–Ray–Hare test; pRegion × Diet = 0.016), which may be due to the grain introduction, leading to the accumulation of microbe-derived glycans in the four-chambered stomach (Fig. 2a and Additional file 1: Table S5). Together, these findings indicate the region-specific features of polysaccharide degradation in the GIT microbiome caused by the geographic specialization in nutrient availability and taxonomic populations of different regions [3, 6].
We hypothesized that the carbohydrate fermentation strategies of region-specific microbiomes also have an adaptive pattern. The fermentation stoichiometries of volatile fatty acids were first detected, and a higher proportion of acetate in the large intestine than in the four-chambered stomach and greater proportions of propionate and butyrate in the four-chambered stomach were observed (Additional file 2: Fig. S3). We further explored the features of carbohydrate fermentation in the GIT microbiome (Fig. 2b and Additional file 1: Table S6). A relatively high abundance of genes involved in converting acetyl-CoA to acetate and the Wood–Ljungdahl pathway was found in the large intestine. In addition, genes (fucO and pduCDEP) related to the propanediol pathway, one of the propionate biosynthesis pathways via deoxyhexose [12], were elevated in abundance in the large intestine, which was linked to the enrichment of fucose-degrading enzymes (e.g., GH29 and GH37) and rhamnose-degrading enzymes (e.g., GH78 and GH88) in the large intestine (Additional file 1: Table S5). We also observed higher representation of the gene encoding phosphate butyryltransferase (ptb), which catalyzes butyryl-CoA to butyrate in the four-chambered stomach. Interestingly, methanogenesis genes, including fwdABCD, ftr, mch, mtd, mer, mtrABCDEFGH, and mcrABCDG, were more abundant in the jejunum and ileum, further emphasizing the possibility of methanogenesis in the small intestine. Together, these findings revealed segmental differences in microbial functional groups based on the fermentation patterns in the GIT.
Overview of 3079 draft microbial genomes constructed from dairy cattle GITs
To further clarify microbial organization and functionality at the genomic level, we performed contig binning based on single-sample assemblies and obtained 23,356 MAGs. After quality control and data filtration, 3079 MAGs exceeding medium quality (completeness ≥ 50% and contamination ≤ 10%) remained, of which 1450 were from the four-chambered stomach, 249 were from the small intestine, and 1380 were from the large intestine (Additional file 1: Table S7). We then compiled 1904 strain-level genome bins (SGBs) with a dereplication cutoff of 99% average nucleotide identity (Additional file 1: Table S8). Of these, 592 SGBs were of high quality (completeness > 80%, contamination < 10%, and quality score > 50). For taxonomic profiling, 323, 1507, and 1904 SGBs were classified into microbes at the species, genus, and phylum levels, respectively (Additional file 2: Fig. S4a). PCoA analysis showed that MAG profiling presented clear divergence among the four-chambered stomach, small intestine, and large intestine, as described in the gene catalog data (Additional file 2: Fig. S4b). At the phylum level, 15 phyla were annotated, mainly consisting of Firmicutes (861) and Bacteroidetes (747), followed by Spirochaetes (76), Proteobacteria (61), and Euryarchaeota (39) (Additional file 2: Fig. S4a). Among them, 1865 SGBs mainly belonged to bacteria (47.4% of MAGs from the four-chambered stomach, 8.1% from the small intestine, and 44.5% from the large intestine) and 39 SGBs (28.3% of MAGs from the four-chambered stomach, 8.7% from the small intestine, and 63.0% from the large intestine) were classified as members of archaea (Fig. 3). Of the bacterial SGBs, most of the genomes were classified into the genera Prevotella (142 SGBs; 74.9% of MAGs from the four-chambered stomach) and Alistipes (106 SGBs; all MAGs from the large intestine). Furthermore, most SGBs belonging to the class Negativicutes consisting of propionate-producing bacteria via the succinate pathway [13] were enriched in the four-chambered stomach. In contrast, all 16 SGBs assigned to Kiritimatiellae were over-represented in the small intestine (Additional file 1: Table S9). Notably, 33 Methanobrevibacterr-affiliated SGBs, hydrogenotrophic methanogens [14], were mainly retrieved from the large intestine (60.6%). However, their abundance was higher in the small intestine (Fig. 3). We also observed the prevalence of four Methanomethylophilus-affiliated SGBs (methylotrophic methanogens) in the four-chambered stomach and the representation of two Methanocorpusculum-affiliated SGBs (hydrogenotrophic methanogens) in the large intestine [14, 15] (Additional file 1: Table S9). These results suggest that different taxonomic groups have region-specific distributions across the GIT, and these inferences may be confirmed by genomic function analysis.
Genome-based characterization of glycan-degrading microbes across GIT regions
We further compared the CAZyme profiles of 592 high-quality SGBs from different GIT regions. The cellulosomal components were first investigated, and we found that dockerin-containing proteins were widely identified in five phyla, dominated by Firmicutes (51.7%) and Bacteroidetes (46.3%), and 36.9% of them possessed cohesion modules (Additional file 1: Table S10). By comparing the genomic abundance among the GIT microbiomes, we found that most four-chambered stomach-enriched SGBs were classified as Prevotella spp. (e.g., Prevotella sp. UBA3846) and Ruminococcus spp., which may have the potential to produce cellulosomes because of the prevalence of dockerin and cohesion, as well as cellulase (e.g., GH5 and GH9), and hemicellulose (e.g., GH43) (Additional file 1: Table S10), indicating that the microbes in the four-chambered stomach have a powerful potential to degrade fiber [7, 16, 17]. In addition, we found that the four-chambered stomach-enriched Muribaculaceae C941 spp. had a prevalence of dockerin and GH13 but without cohesion (Additional file 1: Table S10). Members of Muribaculaceae C941 likely contribute to starch degradation via a non-amylosomal approach in the four-chambered stomach. In contrast, the hindgut-enriched SGBs belonging to Alistipes spp. lacked dockerin modules, and the Kiritimatiellae-affiliated SGBs that were enriched in the small intestine encoded myriad GH109 enzymes to degrade glycans derived from host mucins [18] (Additional file 1: Table S10). These results indicate that microbes placed in the four-chambered stomach provide efficient hydrolysis for complex glycan breakdown to deal with primary dietary substrates.
We further predicted the polysaccharide utilization loci (PULs) for these SGBs and found that the four-chambered stomach-enriched Prevotella sp. UBA3846 also possessed more than 12 PULs (Additional file 1: Table S11) that were predicted to encode various polysaccharide-degrading enzymes, such as hemicellulase (e.g., GH43 and GH3), fructan-degradation (e.g., GH32), and pectinase (e.g., GH28) PULs (Fig. 4). These results highlight that the enrichment of Prevotella spp. in the four-chambered stomach may serve as a strong force for degrading complex glycans. Other four-chambered stomach-enriched SGBs (SGB9, SGB200, SGB357, and SGB627) belong to Vibrio cholerae RC9 encoded PULs that contained double cellulase GH26 (Fig. 4 and Additional file 2: Fig. S5), indicating its potential for degrading low-value plants in the four-chambered stomach. In contrast, the hindgut-enriched Alistipes-affiliated SGBs (SGB1850, SGB1637, SGB1230, and SGB1407) utilized host-derived glycans inferred by the enrichment of GH109, GH20, and GH92 genes and encoded putative PULs that also contained GH20, GH2, and GH27 CAZymes (Fig. 4 and Additional file 1: Table S10). These findings support the idea that many host mucous secretions and shed epithelial cells serve as additional energy sources for microbes in the hindgut of dairy cattle. Together, the four-chambered stomach microbiome orients toward the degradation of plant-derived glycans and the microbiome in the post-gastric intestine can utilize host-derived glycans.
Genome-based characterization of hydrogen metabolism–associated microbes across GIT regions
Hydrogen can be primarily produced via microbial fermentation processes by hydrogenases, which are also the major substances in hydrogenotrophic consortia, including methanogens, acetogens, fumarate reducers, sulfate reducers, and nitrate reducers [2]. Here, we focused on microbial populations possessing hydrogenases ([NiFe]-, [FeFe]-, and [Fe]-hydrogenases) among 592 high-quality SGBs. We found that 358 SGBs classified into 10 phyla encoded [FeFe]- and [NiFe]-hydrogenases (Fig. 5), suggesting that H2 metabolism is a widespread trait among the GIT microbiome of dairy cattle. We observed that more hydrogenase-encoding genomes were obtained from the large intestine (53.1%), followed by the four-chambered stomach (37.7%) and small intestine (9.2%). Of these 358 SGBs, 213 encoded groups A1, A2, and B [FeFe]-hydrogenases for H2 evolution during fermentation (Fig. 5 and Additional file 1: Table S12), and approximately 59.2% of the MAGs were obtained from the large intestine. Further considering the microbiome of the large intestine, we found that these fermentative hydrogenases in the large intestine were uniquely contributed by Alistipe spp. (Fig. 5). Interestingly, up to 54.5% of hydrogenases were group A3 [FeFe]-hydrogenase (Fig. 5 and Additional file 1: Table S13), suggesting that such electron-bifurcating hydrogenases appear to substantially mediate H2 production by the GIT microbiome in dairy cattle [2]. Over half of the MAGs (55.9%) within the coding group A3 [FeFe]-hydrogenase were obtained from the large intestine. The majority of electron-bifurcating hydrogenases were contributed by the genera Ruminococcus, Vibrio cholerae RC9, and Clostridiales bacterium Firm_07 (Fig. 5). The contributors also showed segmental heterogeneity, such as the main contribution of the facultative anaerobic fermentative bacteria Erysipelotrichaceae in the small intestine (Fig. 5). Therefore, the analysis of [FeFe]-hydrogenases indicates that H2 production in the GIT microbiome is primarily driven by fermentative and electron-bifurcating hydrogenases and that microbial populations in the large intestine may contribute to greater H2 evolution.
We further identified 11 methanogen SGBs that processed hydrogenotrophic methanogenesis by encoding H2-uptake [NiFe]-hydrogenases (groups 3a, 3c, 4 h, and 4i) and methyl-CoM reductases (mcrA), including Methanobrevibacter spp. (20), Methanocorpusculum spp. (2), and Methanomethylophilus (1) (Additional file 3: Table S14). Methanogenic hydrogenases were mainly contributed by the genomes of the large intestine (54.5%). We also noticed other hydrogen-utilizing functional groups that harbored the required terminal reductases and specific hydrogenases to compete for H2 with hydrogenotrophic methanogens. Two SGBs (SGB1248 and SGB1432) from the large intestine (Lachnospiraceae) encoded both [FeFe]-hydrogenases (A3 or A4) and the marker genes for hydrogenotrophic acetogenesis (acsB, CooS, cdhD, cdhE, or FdhA). Fumarate reduction was driven by Acetobacter spp., Selenomonas spp., and Escherichia spp. within the coding [NiFe]-hydrogenase (groups 1d or 1c) and fumarate reductase (SdhA), and 66.7% of the contributors were obtained from the four-chambered stomach. Desulfovibrio-affiliated SGB901 encoded group 1b [NiFe]-hydrogenase and terminal reductases AprA, DsrA, and NrfA for sulfate and nitrate reduction. Together, distinct GIT habitats support different hydrogen-utilizing functional groups.
Region-specific responses of microbial populations to a grain-based diet introduction
To further illustrate the region-specific responses of the GIT microbiota to a diet regime shift in dairy cattle, we compared the changes in microbial taxa at both the gene and genome levels between forage- (control) and grain-based diets. Although dietary regimes had little influence on the segmental dissimilarity of the GIT microbiome, we found that grain introduction significantly affected the GIT microbial structure in each region, particularly for the four-chambered stomach microbiome, which preferentially consumed dietary substrates (ANOSIM, p < 0.01; Fig. 6a). The decreased richness and increased inter-individual variability among the four-chambered stomach and large intestine (Fig. 6b) suggest that a grain-based diet drives the instability of a microbial community, resulting in the loss of specific taxa and greater microbiome dissimilarity between individual animals.
We found that a grain-based diet markedly affected taxonomic populations in the four-chambered stomach with the expansion of Proteobacteria and the reduction of Lentisphaerae (Fig. 6c; Additional file 1: Table S15). A significantly increased abundance of the dominant taxa Prevotella was observed in the four-chambered stomach of grain-fed cows. Elevation of Methanomicrobium spp. in the four-chambered stomach and reduction of Methanobrevibacter spp. in the abomasum occurred in grain-fed cows. A grain-based diet also reduced the relative abundance of Piromyces spp. and Caecomyces spp. classified as fiber-degrading fungi Neocallimastigomycota in the omasum. The jejunum was the most affected region in the small intestine, seeing effects such as evidenced by enrichment of the genus Ruminococcus and depletion of the genera Romboutsia, Phyllobacterium, and Turicibacter. The genera Faecalibacterium, Bifidobacterium, and Succiniclasticum were significantly elevated in the hindgut, whereas the dominant genus Alistipes was reduced in the rectum.
At the genome level, we observed that the genomic abundance of Prevotella spp. shifted across the GIT regions after grain introduction (Wilcoxon rank-sum test, log2fold−change > 1 and p < 0.05; Additional file 1: Table S16). In the four-chambered stomach, we found that a grain-based diet significantly increased the abundance of Prevotella and Ruminococcus-affiliated SGBs (Fig. 6d), whereas it generally decreased the abundance of SGBs belonging to Alistipes spp. and Peptococcaceae bacterium UBA7185 (Fig. 6e). In the post-gastric GIT, the abundance of Prevotella-affiliated SGBs increased (Fig. 6d), whereas the abundance of Alistipes and Saccharofermentans-affiliated SGBs decreased in the large intestine (Fig. 6e). These results indicated that a grain-based diet selectively alters diverse microbial populations in different GIT regions.
Region-specific modification of metabolic cascades of the GIT microbiome by a grain-based diet
We further outlined that segmental variations of major metabolic cascades from macromolecules undergo polysaccharide degradation and fermentation in which metabolites are transferred among microbes. We found that the microbiota in the four-chambered stomach and small intestine was oriented toward propionate-type fermentation (t-test, p < 0.05), whereas no change in fermentation type occurred in the hindgut (Additional file 2: Fig. S6). We hypothesized that the carbon-fueled trophic structure would change under a grain-based diet, including polysaccharide degradation and glucose fermentation.
To test this hypothesis, we explored the microbial potential for polysaccharide degradation after a diet regime shift. A grain-based diet caused a mass of changes in the abundance of CAZyme families across the GIT regions, most of which were associated with the degradation of starch, plant cell wall, and microbial cell wall (Additional file 1: Table S17). In the four-chambered stomach, a grain-based diet increased the abundance of peptidoglycan-degrading families GH73, GH103, GH104, GH23, GH24, GH25, alpha-amylase family GH119, and chitinase family GH19 (Additional file 1: Table S17). Higher peptidoglycan-degrading ability was also observed in the small intestine (GH104) and large intestine (GH24 and GH25) (Additional file 1: Table S18). In contrast, a grain-based diet reduced the abundance of the cellulose-binding enzyme CBM9 in both the four-chambered stomach and large intestine (Additional file 1: Table S19). Together, a grain-based diet greatly changed the degradation strategies from polysaccharide to glucose by the microbial enzymatic repertoire in the GIT of dairy cattle.
To further dissect the assignment of specific taxa to the polysaccharide degradation system after grain introduction, we focused on the microbial populations within the broader substrate-related enzymatic repertoire. Strikingly, a grain-based diet significantly depleted the abundance of SGBs affiliated with cellulose-degrading members of Prevotella sp. UBA3846 in the four-chambered stomach (Fig. 7a and Additional file 1: Table S16), implying a reduction in fiber degradation ability in the four-chambered stomach. A closer examination of Prevotella sp. UBA3846 revealed that the reduction of several genomes (SGB219, SGB190, and SGB406) in the four-chambered stomach possessed GH43-containing hemicellulase and GH28-containing pectinase PULs, whereas SGB488 and SGB713 were found to encode GH28-containing pectinase PUL and were elevated in the large intestine (Fig. 7b). In addition, the expansion of other Prevotella spp. (e.g., SGB847, SGB189, SGB728, and SGB66) in the four-chambered stomach was predicted to possess diverse amylase PULs (e.g., GH13-GH13_13-GH97-GH77 and GH13-GH13_1) (Fig. 7b). Therefore, a grain-based diet changes the microbial environment in the four-chambered stomach, which regulates taxonomic reassembly in favor of starch degradation by narrowing down plant biomass hydrolysis, and excessive pectin is mainly degraded in the large intestine.
Regarding glucose fermentation pathways, we found that a mass change in KOs occurred in the four-chambered stomach after feeding a grain-based diet (Additional file 2: Fig. S7). For the glycolytic pathways, the expansions of polyphosphate glucokinase (ppgK) involved in the embden-meyerhof-parnas pathway, and ribose 5-phosphate isomerase A (rpiA) and ribulose-phosphate 3-epimerase (rpe) involved in the hexose monophosphate pathway were observed in the four-chambered stomach after grain-based diet feeding (Additional file 1: Table S20). Notably, the expansion of genes involved in the tricarboxylic acid cycle was observed in the four-chambered stomach, which suggests plentiful production of dicarboxylic acids after grain introduction [19] (Additional file 1: Table S20). The decarboxylation of dicarboxylic acids serves as the sole energy source for the growth of fermenting bacteria [20]; therefore, the expansion of the tricarboxylic acid cycle may support the acceleration of the fermentation process by a grain-based diet.
Next, we assessed how microbial populations modified physiological fermentation schemes via representative segments (rumen, jejunum, and cecum) after grain-based diet feeding at the genome level (Fig. 8a). In the rumen, Dialister-affiliated SGB606 exhibited a 139-fold increase in abundance under the grain-based diet and possessed a complete succinate pathway (Additional file 4: Table S21), suggesting that it contributes largely to the higher propionate concentration after grain-based diet feeding (Additional file 2: Fig. S6). The substantial reduction in the abundance of members of Alistipes and Clostridiales bacterium Firm_07 also exhibited a high prevalence of genes controlling the conversion of acetyl-CoA to acetate (ackA) and genes involved in the Wood–Ljungdahl pathway (fhs, fold, and metF; Additional file 5: Table S22), thereby possibly creating significant effects on the reduction of acetate proportion (Additional file 2: Fig. S6). In the jejunum, we found that a decreased abundance of por-carrying SGBs (e.g., Vibrio cholerae RC9 and Clostridium) encoded the process of pyruvate conversion to acetyl-CoA (Additional file 6: Table S23). Thus, the suppressed microbial populations responsible for reducing pyruvate to acetyl-CoA were the main contributors to the decreased concentrations of acetate and butyrate in the jejunum (Additional file 2: Fig. S6). The cecum substantially increased the stoichiometry of volatile fatty acids, including the concentrations of acetate, propionate, and butyrate, which was different from that in the rumen and jejunum in that there were no changes in the ratio of acetate to propionate (Additional file 2: Fig. S6). Among the more abundant genomes, Methanobrevibacter-affiliated genomes encoded porABDG and Prevotella-affiliated genomes encoded por to promote the production of acetate and butyrate (Additional file 7: Table S24). Overall, grain-based diets influence region-specific microbial populations to reorganize microbial fermentation strategies in the GIT of dairy cattle.
Reshuffle of hydrogenogenic and hydrogenotrophic processes during a grain-based diet
Hydrogen metabolism is a key junction that connects different microbial functional groups in the GIT ecosystem [13]. Hence, we further decoded the segmental heterogeneity of interspecies hydrogen transfer during shifts in the fermentation type after a grain-based diet. We observed that 66.7% of diet-altered genomes coding fermentative hydrogenases and 71.6% coding electron-bifurcating hydrogenases were decreased in abundance in grain-fed cows and were mainly classified into members of Clostridiales bacterium Firm_07, Ruminococcus, and Alistipes (Additional file 8: Table S25). In addition, the genomes encoding hydrogenases were mainly reduced in the four-chambered stomach, followed by those in the jejunum and large intestine. These results suggest that a grain-based diet may cause fewer H2 sinks across GIT regions [3, 21]. We speculated that a large proportion of genomes encoding H2-evolving hydrogenases that decreased after grain diet feeding may affect hydrogenotrophic pathways. To test this, we focused on the change in the abundance of genomes coding H2-uptake modules, including methanogenic hydrogenases (groups 3a, 3c, 4 h, and 4i [NiFe]-hydrogenases) and respiratory hydrogenases (groups 1d, 1c, and 1b [NiFe]- hydrogenases), accompanied by the required terminal reductases. As expected, the marker gene for hydrogenotrophic acetogenesis (acetyl-CoA synthase, acsB) was reduced in the four-chambered stomach after grain introduction (Fig. 8c and Additional file 1: Table S20). This result also underpinned the reduction of acetate-type fermentation in the four-chambered stomach and small intestine (Additional file 2: Fig. S6). Moreover, no changes were observed in the abundance of mcrA in the methanogenic pathway (Additional file 1: Table S20). Notably, we observed that a grain-based diet promoted the gene abundance of frdBCD, which was involved in fumarate reduction in the four-chambered stomach that underpinned the process of propionate production (Fig. 8c and Additional file 1: Table S20). Therefore, a grain-based diet may cause a more significant H2 sink in the propionate pathway than in other fermentation pathways.