Introduction

The rumen chamber in ruminant digestive tract has evolved into an efficient fermentation unit, with a remarkable ability for plant biomass utilization. The rumen is inhabited by specialized microflora that harbors complex system for the microbial attachment and hydrolysis of plant polysaccharides into their oligomers and monomers, which are else indigestible by the host (Pope et al. 2010; Yeoman and White 2014). Amid the various microbial domains (bacteria, archaea, fungi, protozoa and bacteriophage) inhabiting the rumen, bacteria are the most dominant representing about 95% of the total microbial population and contribute far more to ruminal fermentation than other inhabitants. These organisms produce a range of enzymes collectively known as CAZymes, including cellulases and hemicellulases attributed for recalcitrant plant polysaccharides deconstruction. Accordingly, rumen can be used as an excellent niche for exploring CAZymes to be used in economic and efficient conversion of plant biomass to biofuels and other value-added products (Brulc et al. 2009; Aggarwal et al. 2017). It has been stated that, the symbiotic microbial communities residing inside the rumen are characteristic of both, the concerned animal species and its diet (Smith et al. 2013; Reddy et al. 2014). When cow and buffalo are provided with same feed type, buffalo utilizes feed more proficiently due to a higher abundance of cellulolytic bacteria present in buffalo rumen as compared to cattle rumen (Wanapat et al. 2000; Chanthakhoun et al. 2012). Dietary interventions and nature of the forage (green or dry roughage) provided to the host have also been proposed to influence rumen microbiome for improved feed digestibility (Thomas et al. 2011; Stiverson et al. 2011). In tropical countries like India, ruminants such as Bos taurus (cow) and Bubalus bubalis (buffalo) represent a substantial proportion of domesticated animal species and are primarily fed on nutrient-poor, high fiber content seasonal crop residues, commonly termed as roughage. Roughages are further classified into two types; green and dry roughages depending on their quality. Green roughage (GR) generally has high moisture content, relatively soluble and is readily hydrolyzed by bacteria. Dry roughage (DR) has lesser moisture content, less soluble and rigid to hydrolyze but is mostly fed in the scarcity or unavailability of green roughage, i.e., during summer or drought condition (Hoffmann and Kovacs 2011; Prajapati et al. 2016). Therefore, it becomes essential to reveal differences prevailing between buffalo and cow rumen microbiome under different feeding scenario which is the basis to exploit rumen bacteria for plant biomass deconstruction. However, due to the absence of suitable techniques to analyze such composite ecosystem in the past, the impact of host and diet on composition and function of rumen microbiome is currently under the scanner. Application of small subunit (SSU) rRNA gene sequence analysis have demonstrated the inability of culture-dependent methods to represent the entire bacterial community as large number of the microbial component remains uncultured and uncharacterized, mostly a major fraction of the fibrolytic bacteria (Larue et al. 2005). Understanding carbohydrate utilization in ruminants is expanding with the advent of “Metagenomics”. The technique has allowed an in-depth understanding of rumen microbiome and uncovered diverse functional genes/enzymes in the system. Profound understanding of the ruminal microbial community can help in providing opportunities for the conversion of plant biomass into bio based products. In the view of the above, Illumina platform was used for sequencing rumen metagenomes of cow and buffalo, with the objective of exploring phylogenetic profile and CAZymes diversity in ruminants under the similar and varied (green and dry roughage) feeding scenarios. The data obtained in our study were analyzed comparatively for the presence of unique CAZyme families in the rumen ecosystem fed with either type of the diet. The results enhance our information on manipulating microbial communities of livestock to improve plant biomass deconstruction. We believe, study with the purpose to ascertain the effects of feed type and animal species on the rumen microbiota and CAZymes diversity between buffalo and cow, with modern microbial molecular techniques has not been reported.

Methodology

Experimental design and sample collection

The experimental animals, a cow and a buffalo, were selected at Gourakshan Kendra, Nagpur, India. Sampling was carried out at two different time intervals: post-monsoon and pre-monsoon. For post-monsoon sampling, both the animals were nourished with readily available green roughage, whereas for pre-monsoon sampling the animals were nourished with readily available dry roughage for nearly 4 months before sampling point. About 50 ml of the rumen digesta samples were acquired using a sterile stomach tube in 250 ml of autoclaved screw capped bottle using a vacuum pump as illustrated by Singh et al. (2014). Rumen samples were placed in dry ice, brought to the laboratory and immediately treated for DNA extraction.

Simultaneously, culturable approach was employed to screen potent cellulolytic bacteria from the collected rumen samples. Based on plate zymography, RAPD profiling, ability to produce diverse enzymes using chromogenic substrates and genome sequencing P. polymyxa ND25, P. polymyxa ND24, B. subtilis ND23, M. arborescence ND21 and E. cloacae ND22 were selected for designing the microbial consortia NDMC-1 (Bohra et al. 2019a).

Metagenomic DNA extraction

The metagenomic DNA extraction of rumen samples was accomplished by means of QIAmp DNA stool mini kit (Prajapati et al. 2016). DNA was quality checked on 0.8% agarose gel and quantified on Qubit® 2.0 Fluorometer (Life Technologies). A260/280 ratio was determined on Nanodrop ND-1000 spectrophotometer (Thermo Scientific, India).

Metagenome library construction, sequencing and assembly

NEB Next Ultra DNA Library Preparation Kit was utilized for constructing the paired-end sequencing library from 200 ng DNA. The library was quality and quantity checked in Bioanalyzer 2100 equipped with high sensitivity DNA chip (Agilent Technologies) and quantified again on Qubit fluorometer. Libraries were further laden onto Illumina platform for cluster generation and sequencing. CLC Genomics Workbench 6.0 was employed for De novo assembly of high quality PE reads by means of default parameters. The resulting metadata were deposited in the Metagenomics Rapid Annotation Using Subsystem Technology (MG-RAST) server at http://metagenomics.anl.gov/.

Bioinformatics analysis of metadata

Bioinformatics analysis of metadata from all the four samples was performed using MG-RAST server. Taxonomic and hierarchical classification of reads that passed the quality filters were executed using the RefSeq database (Pandit et al. 2018), whereas functional annotation was performed against the Subsystem database (Overbeek et al. 2013) with the maximum e value cutoff: e−5 and the minimum identity cutoff: 60%. Data normalization was performed before subsequent analysis. The significance of differences (q value) among microbial phyla, genera and various categories of functional gene was statistically tested using G test (w/Yates’) + Fisher’s test built-in Statistical Analysis of Metagenomic Profiles (STAMP) software package. An effect size of < 1.00 was fixed for “difference between the proportion’ of effect size filter 1 (Jose et al. 2017a).

Mining of carbohydrate-active enzyme using dbCAN CAZyme annotation algorithm

A gene-calling program, FragGeneScan (Ye 2015) was used for predicting putative protein-coding regions in the metadata. This program is put up on a hidden Markov model and combines sequencing error models for improved estimation of the protein-coding region in smaller reads. The identified coding regions were further examined for the occurrence of CAZymes by dbCAN carbohydrate-active enzymes (CAZy) annotation algorithm against the Carbohydrate-Active Enzymes database (Do et al. 2018; Huang et al. 2017). The output so obtained were analyzed manually for determining the diversity and abundance of the various CAZymes classes: glycoside hydrolases (GHs), carbohydrate esterases (CEs), polysaccharide lyases (PLs), glycosyl transferases (GTs), auxiliary activities (AAs) and carbohydrate-binding modules (CBMs) in the rumen metagenomes. The coding sequences predicted as GHs were further characterized by similarity search against NCBI’s protein data bank (pdb) and non-redundant (nr) database using the BLASTP algorithm (Bohra et al. 2018, 2019b). Venn diagram was constructed for the comparison of CAZyme, GHs and CBMs within NDC_GR, NDC_DR, NDB_GR and NDB_DR to show the number of proteins shared or unique within a particular relationship.

NCBI accession numbers of rumen microbiome samples

The sequencing data of the present study is submitted to MG-RAST with accession IDs: NDC_GR (mgm4826198.3), NDC_DR (mgm4826199.3), NDB_GR (mgm4826197.3) and NDB_DR (mgm4826196.3).

Results

Culturable approach for designing of microbial consortia

All the four rumen samples were explored for the potent cellulolytic bacterial population. The distinct isolates obtained were further screened using RAPD, plate zymography and were broadly classified based on enzyme activity index (EAI). The group exhibiting EAI > 3 were studied for their efficiency to produce diverse cellulolytic and hemicellulolytic enzymes on chromogenic substrate. Of these five isolates were selected for genome sequencing and designing microbial consortia NDMC-1 based on their diverse enzymes activities (Bohra et al. 2019a).

Yield and purity assessment of extracted metagenomic DNA

The extracted metagenomic DNA was quality checked by 0.8% agarose gel electrophoresis and quantified using ND1000 nanodrop spectrophotometry. The quantification of metagenomic DNA is commonly performed to determine the average DNA concentration and its purity in a solution. Generally, absorbance at 260 nm (A260) is used to determine the concentration of DNA and A260/280 ratio is calculated to assess purity. A260/280 ratio below 1.7 indicates the protein contamination. A260/280 ratio of all four extracted metagenomic DNA ranged between 1.78–1.89 and DNA concentration ranged between 204–338 ng/μL. DNA extraction results showed that the DNA isolation procedure that we opted gave the best output in terms of both enough and intact metagenomic DNA.

Metagenome sequence data statistics

Illumina next generation sequencing led to the thorough elucidation of rumen microbial diversity and their functional capacity to hydrolyze lignocellulosic biomass in cow and buffalo rumen under different (roughage) feeding scenario. Metagenome sequencing of DNA from all the four samples generated about 255.71 Mb metadata with 2,80,220 raw reads and 2,72,700 reads post quality filtering having an average read length ranging from 673 to 1343 base pairs. The metagenomic data analysis statistics of individual samples calculated by in-house Perl scripts are detailed in Table 1. As evident from the MG-RAST annotation, the metagenomic libraries were found to have very rare number of SSU rRNA hits, comprising 0.009–0.13% of the total reads per rumen microbiome sample while 47–66% of the reads were assigned to various functional subsystems. Similar data have been used earlier to inspect rumen microbial ecology with reference to aromatic compounds metabolism, resistome, mobilome and stress, dormancy and spore producing genes in bacteria (Singh et al. 2014; Prajapati et al. 2016, Reddy et al. 2014). In this study, the metagenomic sequences were explicitly used to assess lignocellulosic biomass hydrolyzing capabilities of two different ruminant organisms, i.e., cow and buffalo when fed with seasonal roughage diet.

Table 1 Sequence details and data assembly statistics of rumen metagenome from NDC_GR, NDB_GR, NDC_DR and NDB_DR by using in-house Perl scripts
Full size table

Taxonomic diversity of rumen microbiome samples

At the domain level, with the abundance of more than 93%, bacteria were found to be the most abundant domain in all the samples followed by archaea ranging between ~ 0.6 and 5.3%. The proportion of eukaryotes ranged between 0.4 and 1%, while the rest comprised virus and unassigned sequences (Fig. S1).Taxonomic assignment of bacterial taxa using RefSeq database showed the dominance of phylum Bacteroidetes in the rumen microbiome of both, green and dry roughage-fed animals. Bacteroidetes were followed by Firmicutes, Proteobacteria, Actinobacteria and Fibrobacter and constituted about 90–96% of the total phyla across each of the four rumen metagenomes. Rest 4–10% majorly comprised phyla Chlorobi, Cyanobacteria, Spirochaetes, Verrucomicrobia, Elusimicrobia, Fusobacteria, Chloroflexi, Lentisphaerae, Planctomycetes, Acidobacteria, Tenericutes, Thermotogae, Synergistetes and others (Fig. 1a).

Fig. 1
figure 1

a Taxonomic distribution of bacterial phyla based on predicted proteins and rRNA in metagenome sample from NDC_GR, NDB_GR, NDC_DR and NDB_DR. Statistical analysis using STAMP based on phylum level taxonomic assignments in b NDC_GR and NDC_DR, d NDB_GR and NDB_DR. Statistical analysis using STAMP based on genus-level taxonomic assignments in c NDC_GR and NDC_DR, e NDB_GR and NDB_DR. Brown color indicates dry roughage and green color indicates green roughage-fed animals

Full size image

Unfolding microbial diversity in response to dietary changes and ruminant animal

Effect of roughage type on the rumen microbial community of cow

As the feeding type was shifted from GR to DR in cow, significant decrease in the abundance of Bacteroidetes, Fibrobacteres coupled with an increase in Firmicutes, Proteobacteria, and Actinobacteria was observed (Fig. 1b). At genus level significant changes in 11 different genera were observed. Prevotella and Bacteroides were found to be the most abundant genera in both the samples from cow rumen. However, their abundance declined when the diet is switched from green to dry roughage type. Bacillus, Parabacteroides, Clostridium, Acinetobacter were found more abundant genera in NDC_DR when compared to NDC_GR (Fig. 1c).

Effect of roughage type on the rumen microbial community of buffalo

With the shift in feeding type from GR to DR in buffalo significant changes were observed in the abundance of 10 different phyla. An increase in the abundance of Firmicutes, Bacteroidetes, Fibrobacteres and Actinobacteria coupled with the decrease in Proteobacteria was observed (Fig. 1d). A further genus-level classification of the phyla revealed Prevotella, Bacteroides, Clostridium, and Parabacteroides to be the most abundant genera in both the samples from buffalo rumen. The abundance of Clostridium, Parabacteroides declined while Prevotella, Bacteroides increased as the diet is switched from green to dry roughage type (Fig. 1e). Buffalo fed on GR shared members of the genera Micrococcus with the buffalo fed on DR. The members of Tetragenococcus were only identified in the sample NDB_GR.

Interspecies differences in the ruminal microbial community under the same (green roughage) feeding scenario

Phylogenetic comparison of animals fed on green roughage type revealed the significant difference in the abundance of genera Bacteroidetes, Firmicutes, Proteobacteria and Fibrobacter in their metagenomes. Bacteroidetes and Fibrobacter were found to be abundant in cow rumen metagenome whereas Firmicutes and Proteobacteria were abundant in buffalo rumen fed on green roughage. At genus level Prevotella, Bacteroides, Fibrobacter were abundant genera in cow rumen while Clostridium, Parabacteroides abundant in buffalo rumen. The maximum difference was found in the abundance of genera Prevotella (Figs. S2a, S2b). In addition to this, members of Candidatus Zinderia and Intrasporangium were identified only in the samples of animal fed with GR.

Interspecies differences in the ruminal microbial community under the same (dry roughage) feeding scenario

Animals fed on dry roughage have a higher profusion of Bacteroidetes, Firmicutes, and Proteobacteria covering more than 60% of the bacterial community. Relatively higher abundance of Bacteroidetes and Firmicutes is apparent in buffalo rumen whereas Proteobacteria was abundant in cow rumen metagenome. At genus level Clostridium, Prevotella and Bacteroides dominated both the rumen metagenome with higher abundance in buffalo rumen (Figs. S2c, S2d). Members of Thalassobium were identified only in the samples of animal fed with DR. Members of Wigglesworthia were only identified in NDC_DR. Similarly, members of the genera Leptospirillum and Phaeobacter were found in the NDB_DR.

Functional profiling of the rumen metagenome datasets

Functional annotation of the rumen metagenomes was performed using subsystem database. Out of the 24 subsystems identified, the reads assigned to carbohydrates metabolism were most numerous followed by reads associated with clustering-based subsystems. The abundance of genes responsible for carbohydrates metabolism were found to be 14.52% in NDC_GR, 12.14% in NDB_GR, 11.99% in NDC_DR and 13.66% in NDB_DR as depicted in Fig. S3. A significant difference in the abundance of genes encoding various subsystems with the change in diet from green to dry roughage (Fig. 2) and change in ruminant animal (Fig. S4) was statistically tested using STAMP.

Fig. 2
figure 2

Statistical analysis using STAMP based on functional assignments at subsystem level in a NDC_GR and NDC_DR, b NDB_GR and NDB_DR. Brown color indicates dry roughage and green color indicates green roughage-fed animals

Full size image

Mining rumen metagenomes to identify CAZymes

The annotation for CAZymes present in metadata was performed against the CAZy database through dbCAN annotation algorithm. dbCAN analysis of assembled sequences identified 28.13 and 8.08% of the total contigs from NDC_GR and NDB_GR, respectively, having resemblance with at least one of the CAZyme. This was in contrast with samples NDC_DR and NDB_DR, exhibiting 10.93 and 12.53% abundance for CAZyme genes associated with the lignocellulosic biomass hydrolysis. The CAZyme encoding contigs were further categorized into diverse categories of CAZymes viz. GHs, CEs, GTs, PLs, CBMs and SLH. GHs catalytic modules were the most predominant category found in our dataset which included the genes involved in plant polysaccharide hydrolysis. GHs was found to be increased with the shift in the diet from green (NDC_GR = 39.51%, NDB_GR = 41.04%) to dry (NDC_DR = 41.80%, NDB_DR = 42.55%) roughage. The enzymes classified to functional group GTs were found to be the second most abundant group in all four samples representing 19–24% of the total CAZymes. CBM modules were the third dominant CAZyme class comprising 20.81, 16.14, 15.38 and 17.60%, respectively, in NDC_GR, NDB_GR, NDC_DR and NDB_DR sample. Approximately 11–14% of the contigs found to be associated with CE families. Of these, CE1 family encoding feruloyl esterases, crucial for lignin deconstruction, was the most prominent CE family with 23.44, 22.35, 25.43 and 25.85% abundance against NDC_GR, NDB_GR, NDC_DR and NDB_DR sample, respectively. The presence of nearly 2–5% PLs and 0.7–2.5% AAs were also detected in the metagenome of studied rumen sample. Dry roughage-fed rumen samples NDC_DR and NDB_DR displayed a greater abundance of genes associated with GHs and CEs than the green roughage NDC_GR, NDB_GR. The percentage abundance of various CAZyme classes is depicted in Fig. 3.

Fig. 3
figure 3

Comparison of predicted CAZymes classes: GHs, GTs, CEs, CBMs, PLs and AAs annotated in the rumen metagenome samples from NDC_GR, NDB_GR, and NDC_DR and NDB_DR

Full size image

Comparative jVenn plotting of CAZymes in rumen metagenomes

Venn diagram was constructed for the comparison of CAZyme (Fig. S5), GHs (Fig. S6) and CBMs (Fig. S7) annotated within NDC_GR, NDC_DR, NDB_GR and NDB_DR to show the number of proteins shared or unique within a particular relationship. Overall, 178, 295, 208 and 294 unique CAZyme were annotated in NDC_GR, NDC_DR, NDB_GR and NDB_DR metagenome. Of these 10, 7, 32 and 34 CAZyme were exclusively annotated in respective metagenomes. GHs accounted for about 39–43% of the total CAZyme identified in the studied metagenomes. 89, 155, 114 and 154 unique GHs were annotated in NDC_GR, NDC_DR, NDB_GR and NDB_DR metagenomes. Of these 3, 12, 2 and 14 CAZyme were exclusively present in respective metagenomes. Similar to GHs, the most diverse CBMs (49) with 8 exclusive CBMs were annotated in NDB_DR metagenome. Comparatively high CAZyme (294), GHs (154) and CBMs (49) along with the presence of elite CAZyme 34, GHs (14) and CBMs (8) indicates the proficiency of NDB_DR toward efficient biomass hydrolysis. The significantly enormous number of enzymes categorized into GH family in the rumen metagenome of NDB_DR (154 families) and NDC_DR (155 families) shows a more complex process of plant biomass breakdown in them as compared to others.

Comparative metagenomics of GHs and CBMs responsible for plant biomass hydrolysis in rumen metagenomes

Of the total GHs identified, 83.99, 73.84, 74.39 and 77.78% of them were identified to be directly involved in plant polysaccharide hydrolysis against NDC_GR, NDB_GR, NDC_DR and NDB_DR, respectively. Four families of GHs (GH5, GH9, GH44, GH45) identified to be majorly cellulolytic comprised 7.42, 4.01, 3.20 and 4.40% of the total GHs in NDC_GR, NDB_GR, NDC_DR and NDB_DR, respectively, with the dominance of GH5 and GH9. Enzymes dedicated for hydrolysis of oligosaccharides (GH1, GH2, GH3, GH4, GH13, GH18, GH20, GH27, GH29, GH30, GH31, GH32, GH35, GH36, GH38, GH39, GH42, GH43, GH52, GH59, GH57, GH92 GH95, GH97, GH116, GH120, GH130) were the most abundant representing 47.56, 46.87, 46.90, 47.66% of the total GHs, respectively. The analysis also revealed the presence of 9.8–14.6% GHs from GH1, GH3, GH5, and GH9 families, which have an essential role in the breakage of β-1,4-linkage in cellulose polymer and release of glucose residue. Enzymes chiefly accountable for the deconstruction of xylan main chains including, endo-1,4-β-xylanase and endo-1,3-β-xylanase are presented by GH8, GH10, GH11, GH12, GH26, GH28, GH53, GH98, GH113 and GH141 families. Debranching enzymes from family GH23, GH33, GH36, GH51, GH54, GH67, GH77, GH78, GH84, GH103, GH106, GH110 and GH127 responsible for the depolymerization of hemicellulose were also annotated. In total both of these accounted for about 29, 23, 24 and 25% of the total GHs, respectively. Out of the annotated GH families, sequences affiliated to GH3, GH13 and GH43 were most abundant and in total contributed 25.29, 18.94, 19.36 and 24.14% against NDC_GR, NDB_GR, NDC_DR and NDB_DR, respectively. GH3, GH13 and GH43 family encodes for glucan 1,3-β-glucosidase, β-glucosidase, xylan 1,4-β-xylosidase, α-l-arabinofuranosidase, glucan 1,4-β-glucosidase, β-xylosidase, oligo-α-glucosidase, α-l-arabinofuranosidase, arabinanase, β-N-acetyl hexosaminidase and xylanase (Fig. 4). CAZymes often exhibit linked modular structure known as CBMs which helps in adhesion to the carbohydrates. In total, fourteen CBMs were annotated in rumen microbiome that associates with both cellulase and hemicellulose. Additionally, 8 and 13 CBMs exclusively associating with cellulase and hemicellulose gene, respectively, were also found in the rumen microbiome. CBM4, CBM6, CBM32, CBM37 and CBM44 were found to be the most abundant covering 5–9% of the GHs (Fig. 5).

Fig. 4
figure 4

Distribution of predicted GH family responsible for cellulose and hemicellulose hydrolysis in rumen metagenome samples from NDC_GR, NDB_GR, NDC_DR and NDB_DR. GH families are assigned to different categories based on CAZy database (http://www.cazy.org). Heat Map lists the percentages of different GH families

Full size image
Fig. 5
figure 5

Comparison of CBM linked to various cellulases and hemicellulose in rumen metagenome samples from NDC_GR, NDB_GR, NDC_DR and NDB_DR. CBM families are assigned to different categories based on CAZy data base (http://www.cazy.org). Heat Map lists the percentages of different CBMs

Full size image

A detail of major enzymatic activity carried out by individual GH family, helpful in biomass deconstruction is provided in Table S1. Endo-β-1, 4-glucanases accountable for the breakage of β-1,4-linked glucose residues in long chain cellulose comprised only 0.2–0.4% contigs of the total GHs in the metagenome. In contrast, genes relating to hemicellulose hydrolysis dominated in the rumen samples in all treatment groups. A large proportion of hemicellulolytic enzymes; 36.89, 37.24, 27.81 and 38.18% were identified in sample NDC_GR, NDB_GR, NDC_DR and NDB_DR, respectively. Hemicellulose debranching enzymes like α-l-arabinofuranosidases, α-L-rhamnosidases, and α-glucuronidases that play an important role in depolymerization of hemicellulose were also identified. Endo-1, 4-β-xylanase and endo-1, 3-β-xylanase involved in deconstructing basic xylan chains were estimated to be 2.05 and 6.23% of the total GHs, however debranching enzymes, α-l-arabinofuranosidases, α-L-rhamnosidases, and α-glucuronidases and other oligosaccharide degrading enzyme exhibiting important role in hydrolyzing hemicellulose were profusely annotated in the metagenome samples (Table 2).

Table 2 Percentage abundance of predicted cellulase and hemicellulose enzymes annotated in the rumen metagenome of NDC_GR, NDB_GR, NDC_DR and NDB_DR
Full size table

Ruminant animal and diet-based alteration in CAZymes distribution

Intraspecies alteration of CaZyme composition in cow rumen with a change in roughage type

The abundance of GHs in cow rumen increased with the shift in the diet from green (39.51%) to dry (41.80%) roughage. As compared to NDC_GR, GH11, GH55, GH81, GH94, GH54, GH84, GH110, GH103, GH27, GH39, GH42, GH52, GH59, GH116 and GH120 involved in lignocellulosic biomass hydrolysis were exclusively annotated in NDC_DR metagenome samples. In addition to various other CAZYmes, oligo-1, 6-glucosidase, arabinan endo-1,5-α-l arabinosidase, mannan endo-1,4-β-mannosidase, α-glucuronidase were annotated only in NDC_DR metagenome sample.

Intraspecies alteration of CaZyme composition in cow rumen with a change in roughage type

With the shift in the diet from green to dry roughage, an abundance of GHs increased from 41.04 to 42.55%. However, the percentage of GHs directly involved in biomass hydrolysis was more in NDB_GR, NDB_ DR comprised members from diverse GHs family. GH45, GH8, GH11, GH12, GH141, GH81, GH28, GH54, GH67, GH27, and GH59 were reported only in NDB_ DR. whereas absent in NDB_GR samples. α-Glucuronidase was annotated only in NDB_ DR with various other CAZymes annotated in NDB_GR.

Interspecies differences in CaZyme profile under the same (green roughage) feeding scenario

Significant difference exists in CaZyme profile when both the animals were allowed to feed on green roughage. Green roughage-fed buffalo has higher proportion of GHs (41.04%) as compared to green roughage-fed cow (39.51%). The GH profile was more diverse in NDB _GR metagenome as GH39, GH42, GH52, GH55, GH64, GH84, GH94, GH103, GH110, GH113 families were exclusively annotated in them; whereas, GH8, GH44, GH67, GH98, GH141 were only annotated in NDC _GR metagenome. Also oligo-1, 6-glucosidase, mannan endo-1,4-β-mannosidase, arabinan endo-1,5-α-L-arabinosidase were annotated in NDB _GR metagenome.

Interspecies differences in CaZyme profile under the same (dry roughage) feeding scenario

Dry roughage displays similar effect on the CaZyme profile of cow and buffalo as observed with green roughage. NDB_DR metagenome display higher abundance of GHs (42.55%) as compared to NDC_DR (41.80%). GH12, GH45, GH113, GH128 were annotated only in NDB_DR metagenome. The diverse CaZyme profile of NDB_DR indicates its efficiency toward hydrolyzing complex biomass.

Discussion

Ruminants utilize carbohydrate-rich agricultural waste residues as substantial energy resources. The host itself is unable to synthesize any enzymes required for deconstructing the plant biomass and primarily depend on rumen microbiota to liberate the energy from plant polysaccharides in the form of carbohydrates and sugars. The fiber deconstructing enzymes produced by rumen microflora are of supreme importance in the research related to biofuel generation. Rumen microbial community is crucial for the hydrolysis of plant biomass and varies significantly with changes in the host’s age and diet (Kittelmann and Janssen 2011; Li et al. 2012). Substantial differences in the rumen microbiome have also been reported across discrete animal species (Jami and Mizrahi 2012). In this study, we compared both the rumen microflora and functional genes that putatively encode CAZymes in the rumen metadata from cow and buffalo fed on green and dry roughage diet. The metagenomic approach was used for unfolding the composite rumen biome at a level, not been accomplished previously using traditional culture-dependent studies. Previous studies on rumen ecology have reported the presence of core bacterial community chiefly from the phyla Firmicutes and Bacteroidetes, in spite of the variation in the breed, diet and other environmental aspects (Kittelmann and Janssen 2011; Petri et al. 2013; Thoetkiattikul et al. 2013; Henderson et al. 2015). These findings are consistent with the results obtained in our study, where phyla Firmicutes and Bacteroidetes dominated all four rumen samples irrespective of diet and species, further affirming the existence of core microbiome in the rumen. However, the diversity and abundance of bacterial community in buffalo rumen were found to be higher as compared to cow’s rumen similar to the study conducted by Chanthakhoun et al. (2012), who reported a high abundance of cellulolytic bacteria in buffalo rumen than cattle rumen. The difference in community composition between the two animal species is indicated by a significantly higher abundance of Firmicutes and lower abundance of Bacteroidetes in Buffalo rumen as compared to cow rumen. Higher Firmicutes abundance in host rumen is correlated with enhanced hydrolysis of plant biomass and efficient feed utilization (Myer et al. 2015). Accordingly, in this study, the higher richness of Firmicutes in buffalo rumen indicates its higher efficiency in hydrolyzing plant biomass over cow rumen microbiome. Parallel to the findings of Jose et al. (2017a), the abundance of bacterial genus Ruminococcus was relatively higher while the abundance of Prevotella was lesser in buffalo as compared to cows. Prevotella is reported for protein and polysaccharide digestion (Flint et al. 2012), whereas Ruminococcus is proficient in the deconstruction of cellulose and hemicellulose (Biddle et al. 2013). The high richness of Ruminococcus in buffalo rumen reflects its efficiency for biomass deconstruction. Cellulose and hemicellulose in plant biomass are considered as one of the prevalent renewable resources of fermentable sugars that could be utilized further in numerous industrial processes, like ethanol generation. However, the existence of extremely resilient lignin remains the major bottleneck in accessibility of these fermentable sugars (Alvira et al. 2010). Ruminants with their highly composite and proficient microbial populations can facilitate the hydrolysis of lignocellulose biomass (Hess et al. 2011). The natural ability of rumen microflora to produce a range of potential CAZymes that hydrolyze the harsh plant biomass has been effectively utilized for treating diverse agricultural waste. Ruminant’s capability to deconstruct the harsh feed into fermentable sugars is entirely accredited to polysaccharide hydrolyzing enzymes produced by rumen microflora. The complex lignocellulosic degradation machinery of rumen makes it an excellent source for mining CAZymes (Yue et al. 2013; Patel et al. 2014). The CAZy database (http://www.cazy.org) is an assemblage of a diverse group of enzymes exclusively responsible for carbohydrate metabolism. The database provides information concerning all CAZyme families; GHs, GTs, PLs, CEs, CBMs and AAs; thus allowing one to inspect all recognized families and enzymes accountable for cellulolytic and hemicellulolytic activity. Of various CAZyme family, GHs are the most abundant and diverse group of enzymes accountable for the breakage of glycosidic bonds in plant polysaccharide, comprising 50% of the enzyme classified in CAZyme database (Cantarel et al. 2009). GHs accounted for about 39–43% of the total CAZyme identified. Previous studies have reported the presence of 72 and 78 distinct families of GHs, respectively, from buffalo and cattle rumen that is responsible for the hydrolysis of various plant cell wall constituents (Patel et al. 2014; Jose et al. 2017a). The significantly enormous enzymes categorized into GH family in the rumen metagenome of NDB_DR (154 families) and NDC_GR (155 families) show a more complex process of lignocellulose breakdown compared to others. In line with the other studies (Wang et al. 2013; Pope et al. 2012), a higher proportion of genes from GH5 and GH9 families encoding endoglucanase were present than other GH families. GH5 family proteins are reported to exhibit endo-xylanase mannanase and exoglucanase activity in addition to endoglucanase activity (Dai et al. 2012). The abundance of genes encoding putative GH5 family in NDC_GR indicates that rumen microbiome of cow may comprise a unique approach for plant polycarbohydrate deconstruction. Similar to previously reported studies on ruminants metagenome (Pope et al. 2012; Dai et al. 2012), contigs associated with GH6 and GH7 families cannot be predicted in the metagenome explored in this study. However, Jose et al. (2017b) had identified contig belonging to GH6 family in Indian crossbred cattle rumen microbiome. Enzymes belonging to GH48 family are reported to be one of the components in most of the cellulosome system and therefore the most abundant enzyme subunit in cellulosome-producing bacteria (Artzi et al. 2018; Shinoda et al. 2018). Its absence in studied metagenome points the likelihoods of the non-cellulosomal mode of plant biomass deconstruction in ruminants. Other than GHs, activities of other groups of CAZymes including GTs, CBMs, CEs, PLs, and AA is required for complete depolymerization of plant polysaccharides. GTs, previously stated to be responsible for catalyzing the breakage of glycosidic linkage (Mota et al. 2018), were found to be the second most profuse CAZy family (19–24% of total CAZyme). Of various GT family recognized, GT2 and GT4 families (comprising chitin synthase, cellulose synthase, α-glucosyltransferase, etc.) were the most numerous comprising > 50% of the total GTs. CBMs which do not exhibit the enzymatic activity of their own but helps in binding CAZymes (GHs, CEs, and AAs) to polysaccharide, thus potentiating their activities (Du et al. 2018; Bernardes et al. 2019), accounted for 15–21% of CAZymes in our study. Nearly 11.2–14% of the CAZymes were found to have similarity with class CEs that facilitate the action of various GHs through ester hydrolysis of plant polysaccharides. PLs responsible for cleaving glycosidic bonds (Bertucci et al. 2019) and AAs targeting the insoluble polymers (Sandhu et al. 2018) were the lesser annotated CAZyme classes in studied metagenome.

Conclusion

Ruminants harbor a vast and diverse microbial community that functions in the utilization of lignocellulosic feedstuffs. In the present study, metagenomics approach was employed to portray the rumen microbial diversity and identify the major plant biomass hydrolyzing enzymes of two ruminant animals Bos taurus (cow) and Bubalus bubalis (buffalo) bred on either green or dry roughage. Results indicate that the rumen bacterial community develops to hydrolyze particular forage is quite specific and vary with variation in feed type and animal species. Phylogenetic profiling of rumen microbiome samples revealed the abundance of bacterial genera Bacteroides, Fibrobacter, Clostridium, Prevotella, and Ruminococcus in all the samples, however, significant differences were observed between the animals fed on different forage type. Further analysis revealed the existence of enormously diverse CAZyme belonging to classes: GHs, CEs, PLs, GTs, CBMs and AAs, with the dominance of GHs. Of the total GHs identified, 83.99, 73.84, 74.39 and 77.78% against NDC_GR, NDB_GR, NDC_DR and NDB_DR, respectively, were found to be responsible for direct plant polysaccharide deconstruction. The genes related to cellulose and hemicellulose deconstruction were found in the highest abundance in all the sequenced metagenomes. Due to the presence of higher bacterial population present study concludes that the rumen of buffalo as the most efficient plant biomass fermenter.

The study establishes rumen as an important reservoir for gene mining studies with the purpose of prospecting industrially important novel CAZymes simultaneously providing an extended collection of ruminal CAZymes that can be inspected further for obtaining potential enzymes for industrial applicability and development of efficient fermentation systems for bioconversion of plant biomass to high-value products.