Growth performance and nutrient digestibility
We measured and calculated some of the metrics to estimate the performance of the production quota. We found that the ADG of sheep significantly decreased (P<0.05) with increasing intensity of cold stress, and after initiating wind treatment, sheep weight began to decrease (Fig. 1a).
To study whether changes in energy intake or energyoutput were responsible for the change in BW, sheep were fed in the metabolic cage. We found that Dry matter intake (DMI) varied among the treatment groups. DMI was low in the C group and lowest in the LT group. DMI then increased with increasing wind velocity, being similar between the LW and MW groups and highest in the HW group (P< 0.05) (Fig. 1b). The apparent digestibility of dry matter was much lower in LT and LW than C (P<0.05), whereas that in MW and HW was similar to that in C(Fig. 1c). As cold stimulation increased, the amount of crude fiber (CF) in feces increased significantly (Fig. 1d), possibly due to the degradation of primarily carbohydrate rather than cellulose by the rumen microbiota in the MW and HW groups. The levels of metabolic energy and digestibility energy were significantly lower in the LT and LW groups than the other groups (See Additional file 1: Fig.S2). These data indicated that the cold environment led to weight loss in the sheep and reduced digestion of the fiber in their feed.
Rumen microbiota changes with environmental changes
We used 16S rRNA gene sequence technology to analyze the abundance of rumen microbiota. The coverage index indicated that cold temperatures influenced microbial diversity (See Additional file 2: Table S1, S2).Through partial least squares discriminant analysis (PLS-DA), we found that microbiota community structures differed among the treatments(Fig. 2a).
We investigated the α diversity of the microbiota. Eighteen taxa were identified at the phylum level. Among the phyla, Bacteroidetes(58.7962±2.0898%) had the highest diversity, followed by Firmicutes (34.2305±1.6897%), Proteobacteria(2.7963±0.8040%), and Fibrobacteres (1.0466±0.2894%). The remaining taxa had values less than 1%. The diversity of Verrucomicrobia (0.3078±0.0486%) was significantly higher in the HW group than in the C and LT groups (P = 0.043).
At the genus level, 216 taxa were identified. Univariate ANOVA of the bacterial abundances revealed several significant differences in the rumen microbes among treatments(Fig. 2). The relative abundance of Prevotellaceae_UCG-003(Bacteroidetes, P = 0.045)and Solobacterium(Firmicutes,P = 0.017)were decreased in the LT, LW, MW and HW groups compared with the C group. Furthermore, the abundance of Brachybacterium(Actinobacteria, P = 0.040), Devosia(Proteobacteria, P = 0.008), Sphingomonas(Proteobacteria, P = 0.000) and unclassified_f__Enterobacteriaceae(Proteobacteria,P = 0.050) were higher in the LT group than in the other groups. The abundance of Rhizobium(Proteobacteria,P = 0.007)was higher in the LT and LW groups than the other groups. In contrast, the abundance of Sphaerochaeta(Spirochaetae,P = 0.044)was decreased in the LT and LW groups compared with the other groups. Furthermore, the abundance of Pseudobutyrivibrio(Firmicutes,P = 0.039) was higher in the MW group than in the other groups. The abundance of Ruminiclostridium_1(Firmicutes,P = 0.020), Ruminococcaceae_UCG-005(Firmicutes,P = 0.044), norank_c__WCHB1-41(Verrucomicrobia,P = 0.043)was increased in the HW group relative to the other groups. Interestingly, the abundance of Lachnospiraceae_XPB1014(Firmicutes,P = 0.012)exhibited highest levels in the LT group and the lowest levels in the LW group (Fig. 2).
We used qPCR to verify the changes in some bacterial groups (See Additional file 2: Table S3). The diversity of the dominant bacteriadid not differ significantly (See Additional file 2: Table S4).
Bacterial function prediction and molecular pathways in the rumen
We predicted the functions of the rumen bacteria and the associated molecular pathways in sheep to assess the impact of wind treatment. At KEGG level 1, there were 7 major categories, including Metabolism (49.76±0.22%), Genetic Information Processing (21.32±0.06%), Unclassified (13.95±0.04%), Environmental Information Processing (10.58±0.23%), Cellular Processes (2.68±0.08%), Organism System (0.77±0.01%), and Human Diseases (0.75±0.01%). The gene abundance of Unclassified was significantly lower in the HW group than the other groups (P<0.05); no other significant differences were observed (See Additional file 2: Table S5).
At KEGG level 2, a large percentage of 41 gene families were found to have correlations with Amino Acid Metabolism (10.57 ± 0.56%), Carbohydrate Metabolism (10.12±0.05%), Replication and Repair (9.88±0.04%), Membrane Transport (9.13±0.21%), Translation (6.00±0.02%), andEnergy Metabolism (6.03±0.03%) (See Additional file 2: Table S6). The gene abundance of Nervous System was significantly lower in the MW group than in the other groups (P<0.05). However, the gene abundance of Excretory System was lower in the LT, LW, and MW groups than in the C and HW groups, which had similar abundance (P= 0.06). At KEGG level 3, 328 KEGG orthology (KO) pathways were identified. The top 44 pathways with high expression are shown (See Additional file 2: Table S7). The gene abundance did not differ significantly among groups. These findings showed that LT and wind speed may affect the nervous system in the rumen.
Changes in the concentration of VFAs and cellulase activity in the rumen
Studies have shown that VFAs could provide energy to the host and participate in the host metabolism[4]. After discovering the changes in the rumen microbiota, we explored the levels of VFAs.We found that the concentration of total VFAs was significantly reduced in the MW and LW groups relative to the other groups (P<0.05) (Fig. 3a). Accordingly, the concentrations of acetic acid and propionic acid were decreased significantly in the MW and HW groups (P<0.05). However, the ratio of acetic acid to propionic acid did not markedly differ among the groups (See Additional file 2:Table S8). In addition, butyrate level was significantly reduced in the HW group relative to the other groups (P<0.05), whereas isobutyricacid and isovalerate levels were significantly increased in the MW and HW groups compared with the other groups (P<0.05).Cellulase activity in the rumen contents decreased with increasing wind speed, being significantly lower in the HW group than in the other groups(P<0.05) (Fig. 3b). These findings suggested that in the cold environment, the rumen microbes reduced their digestion of CF.
Changes in inflammatory factors and antioxidant enzymes
As we expected, the contents of proinflammatory factors, such as IL-2, IL-6, and IFN-γ, in plasma were reduced in the wind-treatment groups relative to the C group (P<0.05). In contrast, the contents of anti-inflammatory factors, such as IL-4, were increased in the wind-treatment groups relative to the C group (P<0.05) (Fig. 4).
We found that MDA content was significantly decreased in the wind-treatment groups compared with the C group (P<0.05) (Fig. 5a). In contrast, the serum concentrations of SOD, CAT and GSH-PX showed similar trends as T-AOC (Fig. 5), being increased in the wind-treatment groups relative to the C group (P<0.05). The ratio of T-AOC to MDA reflects the relationship between the body’s antioxidant capacity and oxidative damage. Low-temperature treatment significantly increased the ratio of T-AOC to MDA in serum (P<0.05), and this ratio increased significantly with increasing WV (P<0.05). However,the ratio of T-AOC to MDAwas significantly lower in theLW group than in all of the other groupsexcept the C group, for which no significant difference was observed. These data showed that cold stimulation led the sheep to enter an immunosuppressive and antioxidant state.
Associations of rumen microbiota with host phenotype
We used correlation analysis to research the associations between microbiota and host phenotype (Fig. 6, Additional file 3: Table S9). ADG was negatively correlated with Ruminiclostridium_1 (r = -0.525, P<0.05), Ruminococcaceae_UCG-005 (r = -0.480, P<0.05), Sphaerochaeta (r = -0.479, P<0.05), and norank_c__WCHB1-41 (r = -0.519, P<0.05) and the levels ofisobutyric acid (r = -0.500, P<0.01) and isovalerate (r = -0.553, P<0.05).Positive relationships were detected between ADG and Solobacterium (r = 0.583, P<0.01) and the levels ofacetic acid (r = 0.583, P<0.01), propionic acid (r = 0.523, P<0.05), butyrate (r = 0.638, P<0.01), valeric acid (r = 0.521, P<0.05), and total VFA (r = 0.499, P<0.05).
In addition, we found that some microbiota were associated with the levels of certain inflammatory factors. For example, the levels of the proinflammatory factors IL-2 and IFN-γ were positively associated withPrevotellaceae_UCG-003 (r = 0.618, P<0.01; r = 0.708, P<0.01.respectively) andSolobacterium (r = 0.541, P<0.05; r = 0.675, P<0.01.respectively). Furthermore, a significant negative association was detected between IFN-γ level and norank_c__WCHB1-41 (r= -0.494, P<0.05), and a positive relationship was found between IL-4 leveland Lachnospiraceae_XPB1014 (r= 553, P<0.05).
Furthermore, correlations between microbiota and oxidative stress markers were identified.TPrevotellacee_UCG-003had a significant positive correlation with MDA level (r = 0.534, P< 0.05) and negative correlations with T-AOC (r = -0.451, P<0.05), SOD (r = -0.646, P< 0.01), CAT (r = -0.664, P<0.01), and GSH-PX levels (r = -0.532, P<0.05). TSphingomonashad significant positive correlations with T-AOC (r =0.602, P< 0.01), CAT (r = 0.608, P<0.01), and GSH-PX levels (r = 0.506, P<0.05)