Diet dominates age in shaping the rumen bacteria community and function in dairy cattle


 Background

To understand the effects of diet and age on the rumen bacterial community and function, forty-eight dairy cattle at 1.5 (M1.5), 6 (M6), 9 (M9), 18 (M18), 23 (M23), and 27 (M27) months old were selected. The M1.5, M6, and M27 had the high protein and starch dietary, while the M9, M18, and M23 had the high fiber dietary. Fermentation profile, enzyme activity, and bacteria community in rumen fluid were measured.
Results

The acetate to propionate ratio (A/P) at M9, M18, and M23 (high fiber diet) was higher than other ages, and M6 was the lowest (P < 0.05). The total volatile fatty acid (TVFA) at M23 and M27 was higher than other ages (P < 0.05). The urease at M18 was lower than M1.5, M6, and M9, and the xylanase at M18 was higher than M1.5, M23, and M27 (P < 0.05). The α-diversity indexes (Ace and Chao1) of ruminal bacteria increased from M1.5 to M23, while they decreased from M23 to M27 (P < 0.05). Thirty-three bacteria were identified as biomarkers of the different groups based on the linear discriminant analysis (LDA) when the LDA score > 4. The variation partitioning approach analysis showed that the age and diet had a 7.98% and 32.49% contribution to the rumen bacteria community variation, respectively. The richness of Succinivibrionaceae_UCG-002 and Fibrobacter were positive correlated with age (r > 0.60, P < 0.01) and also positively correlated with TVFA and acetate (r > 0.50, P < 0.01). The Lachnospiraceae_AC2044_group, Pseudobutyrivibrio, and Saccharofermentans has a positive correlation (r > 0.80, P < 0.05) with diet NDF and negative correlation (r < -0.80, P < 0.05) with diet CP and starch, which were also positively correlated with the acetate and A/P (r > 0.50, P < 0.01).
Conclusion

These findings indicated that the quantitative effect of diet and age on the rumen bacteria were 7.98% and 32.49%, respectively. The genera of Lachnospiraceae_AC2044_group, Pseudobutyrivibrio, and Saccharofermentans could be worked as the target bacteria to modulate the rumen fermentation by diet; meanwhile, the high age-correlated bacteria such as Succinivibrionaceae_UCG-002 and Fibrobacter also should be considered when shaping the rumen function.


Introduction
Ruminant animals can capture nutrients from roughage by the digestion process of ruminal microorganisms to the cell wall components. The rumen is a complex microbial ecosystem containing a great diversity of bacteria, archaea, viruses, protozoa, and fungi [1,2]. Within the rumen microorganisms, bacteria are the most abundant species and are the major contributor to digest plants [2]. The bacteria could convert the feed into volatile fatty acids (VFA), ammonia, and microbial crude protein (MCP), which could further supply nutrients for ruminants [3,4]. The key role during the degradation of plant is the enzymes, which were encoded and secreted by the microorganisms [5,6]. The digestive enzyme could catalyze and decompose feedstuff into molecular for animals to use, like amylase could decompose the starch into glucose and futher enhance the ruminants starch digestibility [7]. The exogenous protease could alert the amino acid composition and improve the starch digestibility of corn silage [8]. Bacteria, enzymes, and the VFA, MCP, etc., are closely related and jointly assist in completing the rumen digestive function.
The rumen function and ecosystem stability largely depend on the diversity and complexity of microorganisms [9]. The study about the rumen bacteria has been explored further from the application of high-throughput sequencing technology. Fonty et al. had found that the rumen cellulolytic bacteria of lambs reach a comparable level of the mature rumen at the end of a week after birth [10]. Jami et al.
found that the calf was born with some rumen bacteria essential for mature rumen function [11]. The ruminal bacterial community is established before the intake of solid feed, and the increased intake of starter could, in turn, shape this community [12]. Anderson et al. also indicated with the solid feed intaking, and the proteolytic bacteria increased from 1-2% (at the delivery) to 10% (12weeks); meanwhile, the amylolytic bacteria also increased with age [13]. From 6-month to 2-year old, the rumen bacteria community was signi cantly different with the same diet [11]. However, Bohra et al. showed the rumen bacteria composition varied with the dietary nutritional level [14]. The different roughage sources also altered the rumen microbiome and carbohydrate-active enzyme pro le [14], and the change is associated with feedstuffs' nutrients [15]. A meta-analysis showed the bacteria might exert independent effects on various aspects of ruminant performance [16]. Ming-yuan et al. indicated the richness of Prevotella in rumen, contributing to improved functions related to branched-chain amino acid biosynthesis and then enhanced the dairy cows' milk protein yield [17]. The bacteria composition, metabolic pathway, and metabolite also different between high and low-yield dairy cows [18]. Diet and age, who is the main driving force to change the rumen bacteria and affect animal performance, has not yet been revealed.
The bacteria composition, enzyme, and the endproduct of VFA, MCP, etc., are the crucially factors shaping the rumen functions and characteristics. Although it has been demonstrated that age and diet in uence the ruminal bacterial community. There is no information about the correlation of these elements in different dairy cattle stages and which one is the main pusher to alter the rumen bacteria community and function. Therefore, this study investigates the ruminal bacteria pro le and its products in six production stages under different ages and diets. We hope to illustrate the speci c diet or age-related bacteria and its' production. Ultimately, to provide the theory basis for dairy cattle precisely feeding and management.

Ethics statements
The experimental procedures used in the present study were approved by the Ethical Committee of the College of Animal Science and Technology, China Agriculture University (Protocol number: 2013-5-LZ).

Animals and sample collection
Animals were selected from a farm in Beijing, China, with the same management system. At the 1.5 (M1.5), 6 (M6), 9 (M9), 18 (M18), 23 (M23), and 27 (M27) months, eight cattle were selected at each period and collected the rumen uid samples. The animal feed formula and chemical composition of these diets are shown in Table S1. In brief, the M1.5, M6, and M27 had a relatively high diet starch and protein content, while the M9, M18, and M23 had a relatively high ber diet.
Rumen uid sample was collected by oral intubation before morning feeding. About 50 mL of rumen liquid from each animal was obtained, with the initial 50 mL (approximately) discarded to avoid saliva contamination. Each sample was separated into two sterile tubes. One was immediately placed in liquid nitrogen and stored at -80℃ for 16S rRNA sequencing and enzyme activity analysis. Another was ltered by four cheesecloth and then stored at -20℃ for fermentation pro le analysis.

16S rRNA sequencing
The DNA of rumen uid samples was extracted using HiPure Stool DNA Kits (Magen company, Guang zhou, China). DNA concentration and purity were monitored on 1% agarose gels. According to the concentration, DNA was diluted to 1ng/µL using sterile water. The V3-V4 region of the 16s rRNA gene was ampli ed by PCR (denaturation: 94°C for 2 min, followed by 30 cycles at 98°C for 10 s, annealing reaction: 62°C for 30 s and 68°C for 30 s and a nal extension at 68°C for 5 min) using speci c primer: former primer 341F (CCTACGGGNGGCWGCAG), reverse primer 806R (GGACTACHVGGGTATCTAAT) [27].
Amplicons were extracted from 2% agarose gels and puri ed using the AxyPrep DNA Gel Extraction Kit (Axygen Biosciences, Union City, CA, U.S.) according to the manufacturer's instructions. The amplicons were quanti ed using ABI StepOnePlus Real-Time PCR System (Life Technologies, Foster City, USA). The puri ed amplicons were pooled in equimolar and paired-end sequenced on a PE250 Illumina platform. Paired-end reads were merged using FLASH (V1.2.7,http://ccb.jhu.edu/software/FLASH/) [28]. Low quality (score ≤ 20) short reads (< 200 bp) and reads containing ambiguous bases or unmatched to primer sequences and barcode tags were ltered to obtain the high-quality clean tags [29] according to the QIIME (V1.9.1, http://qiime.org/scripts/split_libraries_fastq.html) [30] quality-controlled process. The tags were compared with the reference database (Silva database, https://www.arb-silva.de/) using the UCHIME algorithm (UCHIME Algorithm, http://www.drive5.com/usearch/manual/uchime_algo.html) [31] to detect chimera sequences. Then the chimera sequences were removed [32], and the Effective Tags were nally obtained. Sequences analysis was performed by Uparse software (Uparse v7.0.1001, http://drive5.com/uparse/) [33]. Sequences with ≥ 97% similarity were assigned to the same OTUs. The representative sequence for each OTU was screened for further annotation. OTUs abundance information was normalized using a standard of sequence number corresponding to the sample with the least sequences. Subsequent analysis of alpha diversity and beta diversity was performed basing on this output normalized data. For each representative sequence, the Silva Database 132 (http://www.arbsilva.de/) [34] was used based on Mothur algorithm to annotate taxonomic information.

Statistics
The rumen fermentation pro le and enzyme activities were subjected to One-Way ANOVA by SAS (version 9.4, SAS Institute Inc., Cary, NC, USA). Alpha-diversity indices were calculated with QIIME (Version 1.7.0) and analyzed using the Kruskal-Wallis test and Wilcoxon rank test using the "dplyr" package in R.
Principal Co-ordinates Analysis (PCoA) and analysis of similarities (ANOSIM) (999 permutations) was performed and visualized using the "ggplot2" package in R (Version 3.6.1). Spearman's rank correlation was used to identify the relationship between the enzyme activity and rumen fermentation pro le (VFA, NH 3 -N, and MCP); the top 50 abundance bacteria at genus level and its' byproducts (enzyme, VFA, NH 3 -N, and MCP) using the "corrplot" package in R. The result was visualized as a heatmap using the R package "pheatmap." All P-value was corrected using a false discovery rate of 0.05, as described by Benjamini and Hochberg [35]. The false discovery rate corrected P < 0.05 was considered signi cant. The linear discriminant analysis effect size (LEfSe) [36] was used to determine the difference of rumen bacteria among ages and diets by coupling Kruskal-Wallis Test for statistical signi cance with additional tests assessing biological consistency and effect relevance. Variation partitioning approach (VPA) was used to evaluate the relative importance of age and dietary nutrients on rumen bacteria community using "vegan" package in R [37]. Spearman's rank correlation and liner regression also used to analysed the relationship between the PC1 (Principal component 1 of the principal coordinate analysis' axis of rumen bacteria) and age or diet nutrients [38].

Rumen fermentation pro le and enzyme activity
The rumen pH among the 1.5 M, 6 M, 9 M, and 18 M dairy cattle had no difference, while the ruminal pH value in these groups was higher than the M27 (P < 0.05). The NH 3 -N content in the M1.5 was the highest (P < 0.05). The MCP content at the M6, M23, and M27 was higher than M1.5 and M9, and the M18 was the lowest (P < 0.05). The acetate concentration at M6, M9, and M18 was no different from others, while that at M6, M9, and M18 was signi cantly higher than M1.5 (P < 0.05). Propionate content at M6 and M27 was higher than that at M1.5 and M18 (P < 0.05). The rumen butyrate concentration at M1.5 was lower than others (P < 0.05). The rumen butyrate at M27 and M6 was signi cantly higher than M18 (P < 0.05). The total VFA concentration at M27 and M23 were higher than others (P < 0.05). The acetate to propionate ratio (A/P) at M23, M18, and M9 (high ber diet) was higher than other ages, and M6 was the lowest (P < 0.05).

Rumen bacteria diversity analysis
After sequence trimming, quality ltering, and chimeras removing, a total of 2,575,670 high-quality sequence tags was obtained from all samples. The M1. 5 (Table S2). The Good's coverages for all samples were more than 99.70%. The alpha-diversity indices, including Chao1, ACE, observed OTUs, Shannon, and Simpson index, were compared among six groups (Fig.S2). Interestingly, the observed OTUs, ACE, and Chao1 values at M18 and M23 were signi cantly higher than M6 and M9; the M1.5 and M6 were lower than others (P < 0.05). Shannon index was increased from M1.5 to M18 but showed no difference between the M18 and M23. Shannon index at the M27 was lower than M18 and M23 (P < 0.05). The Simpson index at the M1.5 was higher than other groups (P < 0.05). These indexes showed the M18 and M23 had the highest bacteria diversity.

Rumen bacteria composition analysis
The top ten phyla account for more than 99.9% of bacteria (Fig. S3B). Twenty-two genera were identi ed as core bacteria, which were identi ed with a relative abundance > 1% and present in at least 80% of all samples (File. S2). Bacteria with LDA scores higher than four were speculated to have a different abundance across the different groups (Fig. 4A). Finally, 33 bacteria were identi ed as biomarkers of the various groups, respectively. The unique bacteria at M1.5 were Proteobacteria, Gammaproteobacteria, Succinivibrio, Lachnospiraceae, and Bacteroidaceae (genus level). Prevotellaceae, Veillonellaceae, Selenomonadales, Negativicutes, and Muribaculaceae were higher at M6. Prevotella_ruminicola and Lachnospiraceae (family level) were higher at M9. Some Firmicutes phylum bacteria could be the biomarker at M18 (Fig. 4B), like Clostridia, Firmicutes, Ruminococcaceae, and Christensenellaceae. Rikenellaceae and Bacteroidales. The unique bacteria at M23 was Fibrobacter. Succinivibrionaceae and Aeromonadales were higher at M27 (P < 0.05).

The correlation between bacteria and its main byproducts
To explore the potential roles of ruminal bacteria on enzyme activity and fermentation pro le, we analyzed the relationship between the top 50 abundant genera and their main byproducts (enzyme, VFA, NH 3 -N, and MCP) using Spearman correlation analysis (Fig. 6). We found that 23 bacteria were signi cantly correlated with A/P, acetate, and TVFA (P < 0.05). Five genera belong to phyla Firmicutes, and eight genera belong to phyla Bacteroidota. 13 bacteria genera were signi cantly correlated with NH 3 -N, valerate, and urease (P < 0.05). Four genera belong to the Firmicutes phyla; ve genera belong to the Bacteroidota phyla; two genera belong to the Proteobacteria phyla. The genera of Shuttleworthia, Oribacterium, Prevotellacear_YAB2003_group, Succinivibrionaceae_UCG-001 were positively correlated (r > 0.5, P < 0.05) with the dehydrogenase, isovalerate, MCP, and propionate, while genera of Prevotellaceae_NK3B31_group, UCG-005, Butyrivibrio, and Rikenellaceae_RC9_gut-group were negatively correlated (r < -0.5, P < 0.05) with them. Speci cally, the genus of Succinivibrionaceae_UCG-002, Treponema, and Eubacterium_ruminantium_group were strongly positive-correlated with acetate (r > 0.73, P < 0.01).
The Spearman's correlation coe cient of the top 50 genera with age or diet was in Table S3. We selected ve bacteria genera highly correlated with age and diet and also correlated with the VFA (Fig. 7). The Succinivibrionaceae_UCG-002 and Fibrobacter were positively correlated with age (r > 0.60, P < 0.01) and positively correlated with TVFA and acetate (r > 0.50, P < 0.01). The Lachnospiraceae_AC2044_group, Pseudobutyrivibrio, and Saccharofermentans has a Spearman's correlation coe cient value > 0.80 with diet NDF and < -0.80 with diet CP and starch (P < 0.01), which also positively correlated with the acetate and A/P (r > 0.50, P < 0.01). These bacteria should be targeted goal when regulated the rumen function based on different ages and diet backgrounds.

Rumen fermentation pro le and enzyme activity
Rumen pH was affected by the diet chemical composition, and high dietary NDF content could increase the rumen pH (Jiang et al., 2017), while the high grain diet could produce more fatty acids and further reduce the rumen pH [39]. M1.5 group received the lowest NDF and highest grain content diet (Table S1). Still, the incomplete rumen function couldn't produce enough fatty acids making the rumen pH decreased. VFAs are the endproducts of diets' fermentation, and they are also essential for rumen development, production performance, and body metabolism [40][41][42]. Previous studies indicated diet chemical composition could alter the rumen VFA production [21,41,42]. High diet starch content could enhance the rumen propionate concentration [41]. The calf at the age of 1.5M had a lower propionate concentration was due to the immature rumen function, which couldn't produce enough enzyme to degrade the starch into propionate. At 18M, heifer with the lowest diet starch content also had less propionate, caused by the lack of substance, such as starch. A high ber content diet could enhance the rumen acetate concentration and the A/P value [43]. Also, the A/P is age-related [44]. The M27 group had a lower A/P value than M9 and M18, and the discrepancy indicated the diet takes on a more important role in shape the rumen fermentation.
Non-protein nitrogen could be hydrolyzed into ammonia by urease produced by microbes [45]. The protein is hydrolyzed into amino acids and peptides by protease, and then parts of amino acids also became ammonia by microbial deaminating [45]. A portion of ammonia synthesis MCP via microorganism, the other parts be absorbed into the blood, participating in the rumen nitrogen cycle [46]. Our results indicated with the high protein diet, unmatured rumen absorption function [47] at M1.5 lead to the high NH 3 -N content in rumen. MCP acted as a signi cantly important role in the ruminants production performance and diet CP utilization e ciency. In our study, low CP and energy levels in diet inhibited rumen synthesis MCP (M9 and M18) [48]. Our results indicated dietary protein level, enzyme activity, and matured rumen function were three critical factors for rumen utilization of protein.

Rumen bacteria composition
Although the rumen bacteria community has been established in the calf period, the change of rumen bacteria still age-related in 6 to 120 months [44]. The observed OTUs and diversity index were increased with age; however, the decrease from 23M to 27M indicated the dietary had a more decisive in uence on rumen bacteria diversity. The transition of rumen bacteria from 23M to 27M was consistent with the Zhigang et al. [49], which indicated the change from high ber to low ber diet decreased the rumen bacteria diversity. Jami et al. indicated rumen bacteria community was affected by age and diet [11]. The genus with a relative abundance > 1% and present in at least 80% of all samples was de ned as core bacteria (Sli erz et al., 2015). The core bacteria were established during the calf stage, testi ed by Figure  S3A, S3B, and File. S2. However, under the speci c age and diet condition, rumen cultured unique genera to nish the particular rumen function in this stage.
Our result indicated the Gammaproteobacteria was rich in 1.5M. It was consistent with Rey et al., which stated the Gammaproteobacteria was the dominant bacteria (24% relative abundance) in calf at the age of 15-83 d [50]. Firmicutes strongly correlated with ber digestion and could degrade the complex carbohydrates, like cell surface [51,52]. The Firmicutes, Clostridia, Ruminococcaceae were rich in M18, digesting the high ber diet [52,53]. Huws et al. indicated the Fibrobacteria was abundant in the rumen bacteria community under the ryegrass diet, which also plays a vital role in forage degradation [54]. M23 had different ber correlated bacteria, like Fibrobacteria, from M18; it's because the roughage type affected these bacteria [55]. Our results also indicated the rumen bacteria composition was concerned with the nutrient level and the feedstuff species [55,56]. Diet supplement with nitrate could increase the Succinivibrio, which worked e ciently in the nitrogen utilization [57,58]. Under the high CP diet condition, M27 was rich in Succinivibrionaceae, while M1.5 and M6 were not. It was due to the Succinivibrionaceae was also age-related [44], which may reach a certain abundance under the speci c age and dietary conditions to come into play.

The relationship within the rumen bacteria, enzyme, and VFA
Rumen was the most important workshop for the digestion of the nutritional substance of ruminants. Bacteria played a crucial role in digest and convert plant materials to VFA and MCP [59]. The enzyme, which was secreted by bacteria, could catalyze feedstuff decomposition and nutrients turnover [5]. The acetate, TVFA, A/P, NH3-N, urease, valerate, and xylanase strongly correlated with rumen bacteria in our study. Bacteria act as a processor to connect the diets and these end products. The genus Fibrobacter plays a vital role in cellulolytic and converted the feeds into VFA [54,60]. Pseudobutyrivibrio could degrade the complex plant polysaccharides and produce VFA for ruminants to utilize [61]. Saccharofermentans belongs to the Bacteroidetes phylum, including 116 genes encoding glycosyl hydrolases involved in hemicellulose, pectin, arabinogalactan, starch, fructan, and chitin degradation [62]. These age or diet-related genera could work as the target bacteria to regulated the rumen function under different feeding backgrounds. The age-related bacteria affected the TVFA and acetate, while the dietrelated bacteria affected the A/P and acetate. From the age and diet-related bacteria and their relationship with TVFA and A/P, it can be concluded that the diet could change the rumen fermentation type. In contrast, age in uences the rumen fermentation ability.

Conclusion
Although the rumen bacteria community has already been established at the calf stage, the rumen bacteria composition still changes along with age and diet variation. Our study gave the quantitative effect of diet and age on the rumen bacteria (explained 32.49% vs. 7.98% bacterial community variation, respectively). Comprehensive correlations were observed between rumen bacterial community, microbiota functions, and rumen fermentation capacities. Our results reveal targeting the bacterial community by diet to regulate rumen fermentation is an e cient way, but dairy cattle's age should also be considered. Besides the diet and age, there are more unknown factors affecting the rumen bacteria community of dairy cows, which need to be further explored.

Declarations Con ict of interest
The authors declare no con ict of interest.  Figure 1 The rumen uid fermentation pro le of dairy cattle NH3-N: ammonium nitrogen; MCP: microbial crude protein; TVFA: total volatile fatty acid; A/P: the ratio of acetate to propionate. The different letters mean that the difference is signi cant (P < 0.05).

Figure 2
The enzyme activity in rumen uid of calves and heifers The different letters mean that the difference is signi cant (P < 0.05).    The correlation between bacteria (genus level) and its byproduct. Cells are colored based on Spearman's correlation coe cient: red represents a positive correlation, and blue represents a negative correlation. "*,"