A Study on the Mechanism Regulating Acetate to Propionate Ratio in Rumen Fermentation by Dietary Carbohydrate Type

The research direction of our team is nutrition and physiology of ruminants, including dietary nutrition metabolism and rumen microorganisms. Previous research has shown that ruminal acetate-to-propionate ratio is related to diet utilization efficiency. At present, it is believed that the main factors affecting the ruminal acetate-to-propionate ratio are the degradation rate of the diet and the rumen microbial structure, but the main mechanism is unclear. This study found that the effect of ruminal acetate-to-propionate ratio was not affected by the concentration of the fermentation substrate, but was affected by the structure of the rumen microbiota. We believe that changes in the rumen microflora structure are the main mechanism for regulating the ruminal ace-tate-to-propionate ratio. This will help people to further understand the rumen physiology, thereby gradually improving feed conversion efficiency and reducing production costs. Abstract: In order to explore the mechanism by which diet regulates the acetate-to-propionate molar ratio (A: P ratio), we compared the effect on rumen fermentation parameters and the microbiome by altering the ratio of dietary concentrates to roughage ratio and calcium pyruvate infusion. The test animals were Laoshan dairy goats, and were fed continuously through an automatic feeder. The test groups were fed a base diet of low concentrates, and intraruminally infused with calcium pyruvate at two concentrations. The infusion concentrations were derived from the difference in the rate of carbohydrate degradation of the high and low concentrate diets, and they were artificially set such that the high concentration infusion group was infused with twice the concentration as the low concentration infusion group. The control groups were fed high concentrate (6:4) and low concentrate (3:7) diets, respectively. The following results were obtained by measuring rumen fermentation parameters and microbial composition: the rumen A: P ratio was significantly lower in the high-concentrate diet group than in the low concentrate diet group (P < 0.05). Infusion of low concentration calcium pyruvate had no significant effect on rumen A: P ratio (P > 0.05), while infusion of high concentration calcium pyruvate significantly increased the rumen A: P ratio (P < 0.05). Relative to goats fed the low concentrate diet, those fed the high concentrate diet had a greater abundance of microbes related to propionate production and a reduced abundance of microbes related to fiber degradation. Infusion of pyruvate had no significant effect on rumen microbial structure. The above results indicate that increasing the concentration of the fermentation substrate without affecting the composition of the microflora does not reduce the A: P ratio. Microbiological results showed that the A: P ratio was more closely related to the rumen microflora structure. Therefore, it is believed that rumen microflora structure is the main mechanism regulating A: P ratio in rumen fermentation.


Introduction
Reducing the ruminal acetate-to-propionate ratio (A: P ratio) can improve the efficiency of dietary energy utilization. Early studies have found that ruminal A: P ratio is related to dietary structure [1] [2]. However, a large number of studies have also shown that under the same diet structure, the ruminal A: P ratio can be changed by the rate of dietary degradation [3] [4]. If changes in the rate of dietary degradation also affect the ruminal microbial composition, what is the main mechanism of the dietary regulation of rumen-fermented A: P ratio?
Before a carbohydrate becomes a volatile fatty acid (VFA), it must be converted to pyruvic acid [5]. Pyruvate molecules are small and can easily pass through the cell membrane and are readily utilized by most microorganisms in the rumen. Compared to other hydrolytic products of carbohydrates, pyruvate is more evenly distributed among microorganisms. In this study, different carbohydrate degradation rates were simulated by ruminal infusion of high and low concentrations of calcium pyruvate, and the pyruvate level was adjusted such that animals had similar pyruvate level regardless of their dietary concentrate-to-roughage ratio. By comparing the effects of diets with different concentrate-to-roughage ratios and different concentrations of calcium pyruvate on ruminal VFA and microflora structure, the main mechanism of dietary regulation of rumen A: P ratio was explored.

Animals
Eight Laoshan dairy goats were selected that were in good health, had similar

Trial Diets
Trial diets were prepared according to UK AFRC (1993). The high concentrate diet contained 6 parts of concentrate, and 4 parts of roughage, while the low concentrate diet contained 3 parts of concentrate and 7 parts of roughage. The diets were pelletized with a granulator (Table 1).

Feeding Management
Goats were kept individually in single-animal cages, fed 350 g through an automatic feeder every 4 hours, and had free access to drinking water. Goats were dewormed and hoofed according to routine management procedures before entering their respective cages. Fistulated animals were given 30 days to recover from the rumen surgery. During the first week of recovery, high quality hay was provided. The amount of feeding increased subsequently and a small amount of drinking water was given. In the later stage of recovery, goats were given free access to sweet potato vine.

Experimental Design
The effective ruminal degradation rates of the two diets (as measured with the nylon bag method) were 33.48% and 33.42% in the high and low concentrate groups, respectively ( Table 2 and Table 3). The difference in fermentable carbohydrate degradation between the two diets in the rumen was calculated to be 5.27 g/d, which corresponds to 6.47 g/d crude pyruvate, as calculated by molecular formula and purity. Therefore, the low concentration group was infused with 6.47 g/d calcium pyruvate through a constant flow pump. Referring to the dietary components, the high-concentrate diet group was infused with twice as much calcium pyruvate as the low-concentrate diet group. The calculated fermentable carbohydrate difference between high and low concentrate diets was equivalent to 6.46 g/d pyruvate.  A 2 × 2 interactive design was used for the rumen calcium pyruvate infusion test. The test had two treatment groups. Each group contained two fistulated goats, which were fed with only low-concentration diets. Different concentrations of calcium pyruvate solution were infused at a constant rate throughout the day through a constant flow pump via the rumen fistula. The other 4 non-fistulated goats were randomly divided into two control groups, and the high-and lowconcentrate diets were fed according to the 2 × 2 interactive design. Each trial period lasted 20 days, with a pre-feeding period of 16 days and a sampling period of 4 days. At the end of each trial, animals were transferred to the next trial.

Sample Collection
During each sampling period, rumen chyme was collected through the rumen fistula after the morning feeding. A rumen fluid sample was obtained by pressurized filtration through 2 layers of gauze. Immediately after the rumen fluid was collected, the pH was measured with a pH meter. The collected rumen fluid samples were then stored as 4 mL aliquots in 5 mL cryotubes. Two of them were used at −20˚C for the determination of volatile fatty acids. One aliquot was frozen in liquid nitrogen and then stored at −80˚C for microbiological determination, and the remainder was frozen at −80˚C for backup.

Sample Measurement
The dry matter (DM) of the diet was determined according to GB 6435-86. The crude protein (CP) was determined by following GB/T 6432-94. The determination of lignin (ADL) and neutral detergent fiber (NDF) was based on GB/T 20806-2006. The acid detergent fiber (ADF) was determined as described in NY/T 1459-2007. The coarse ash (Ash) was determined according to GB/T 6438-86. The crude fat (EE) was determined according to GB/T 6433-2006. Rumen fluid VFA was determined by gas chromatography. The rumen fluid microflora composition was determined using MiSeq high-throughput sequencing technology.
The work of microbial sequencing and otu map was entrusted to Shandong Kaiyuan Gene Technology Co., Ltd. The raw data were filtered to eliminate the adapter pollution and low-quality reads to obtain clean sequences. Sequence reads with an average quality of under 20 over a 30-bp sliding window as per the phred algorithm were truncated, and trimmed reads having less than 75% of their original length, as well as its paired read, were removed. Then, paired-end reads with overlap were merged into tags using FLASH Tags were clustered to OTU at 97% sequence similarity by scripts of software USEARCH. Venn Diagram and package "ade4" of software R (v3.0.3) were used separately in Venn diagram and OTU PCA analysis. The tag numbers of each taxonomic rank (phylum, class, order, family, genus, and species) or OTU in different samples were summarized in a profiling table. The species with abundances of less than 0.5% were classified into "others" in other ranks for all samples.

Data Analysis
Rumen pH, VFA and other data were analyzed by SAS 9.1 software MIXED process for analysis of variance. The statistical model is: µ α β γ = + + + + where μ = population mean; α i = fixed effect of the ith experimental period; β j = continuous effect of calcium pyruvate dose; γ k = random effect of the kth animal; e ijk = random error. The linear and secondary effects of the calcium pyruvate dose were initially tested. When the secondary effects were found to be not significant, they were removed from the model. The Person's correlation test procedure was performed. The effect was considered significant at P < 0.05, and an influential trend at 0.05 < P < 0.1.

OTU Clustering and Abundance
The abundance of each operational taxonomic unit (OTU) in each sample was obtained by clustering analysis. The OTU abundance preliminarily illustrates the species richness of the sample. In order to obtain species classification information corresponding to each OTU, the RDP classifier Bayesian algorithm was used to classify the OTU representative sequence at a 97% similarity level, and the community composition for each sample was calculated at each taxonomic level: domain, kingdom, phylum, class, order, family, genus, and species. The  Overall, the rumen microbial composition of high-concentrate diets was significantly different those of the low-concentrate diet and calcium pyruvate-infused groups.

Community Analysis
The sequencing results showed that the rumen fluid samples covered a total of 28 phyla, 62 classes, 114 orders, 185 families, 414 genera, and 752 species of bacteria. Among the genera that contributed greater than 1% of the microbiome included Prevotella (

Differences in Species Diversity
The rumen microbial diversity indices of all groups are shown in Table 5. There were no significant differences in rumen microbial diversity indices among groups (P > 0.05). Therefore, these differences were not marked in the table.
However, the total reads of each group increased after the infusion of calcium pyruvate (P = 0.076), indicating that the infusion of calcium pyruvate had a tendency to increase total bacterial numbers; however, these differences were not significant. According to the Shannon and Simpson indices, it can be seen that the species diversity of the low concentrate diet group was higher than that of the high concentrate diet group. The rumen microbial diversity was further increased after rumen infusion of calcium pyruvate. However, the differences between groups were not significant. Table 6 shows the results of Wilcoxon rank sum test at the genus level between treatment groups. It focuses on species with abundances greater than 1% and significant differences. Overall, the difference was significantly concentrated in   the comparison of the high concentrate diet group with the other three groups. The vast majority had no significant differences between the low concentrate diet group and the calcium pyruvate infusion group. In the four groups of samples, those with abundances always greater than 1% were Prevotella (Prevotella_1 and Prevotellaceae_UCG−001), Rikenellaceae_RC9_gut_group, Selenomonas_1 and Succiniclasticum, and there were no significant difference among groups in these genera except Rikenonella. The abundance of Prevotella_1 was not significantly different among the four treatment groups and it was always the most abundant in the rumen. The abundance of Prevotella_1 was consistently above 23% in the four groups of samples. As shown in Table 6, the high concentrate diet group had significantly lower contents of Eubacterium (coprostanoligenes_group) and Ruminococcus (Ru-minococcaceae_UCG-002 and Ruminococcaceae_ V9D2013_group) than the low concentrate diet group, and it had significantly higher contents of Fibrobacte, Prevotellaceae_YAB2003_group and Roseburia. Compared with the low concentration pyruvate infusion group, the high concentrate diet group had significantly lower contents of coprostanoligenes_group, Anaeroplasma, Butyrivi-brio_2, Lachnospiraceae_ND3007_group, Prevotellaceae_UCG−003, Rikenella-ceae_RC9_gut_group, Ruminococcaceae_UCG−002, Ruminococcaceae_V9D2013_ group, Selenomonas_3, Veillonellaceae_UCG−001, and Tyzzerella_3, and it had significantly higher contents of Fibrobacter, Megasphaera, Prevotellaceae_ YAB2003_group, Ruminococcus_2, Roseburia and Succinivibrionaceae_UCG−002. Compared with the high concentration pyruvate infusion group, the high concentrate diet group had significantly lower contents of coprostanoligenes_group, Butyrivibrio_2, Lachnospiraceae_ND3007_group, Prevotellaceae_UCG−003, Ri-kenellaceae_RC9_gut_group, Ruminococcaceae_UCG −002, Ruminococcaceae_ V9D2013_group, Selenomonas_3 and Veillonellaceae_UCG−001, and it had higher contents of Fibrobacter, Megasphaera, Prevotellaceae_YAB2003_group, Ruminococcaceae_UCG−014, Ruminococcus_2, Succinivibrionaceae_UCG−002, and Treponema_2.

Significant Analysis of Differences between Treatments
There were no significant differences between the two calcium pyruvate infusion groups and the low concentrate diet group, and only the low concentrate diet group had significantly lower Christensenellaceae_R−7_group than the low concentration pyruvate infusion group. Ruminococcaceae_UCG-002 and Ruminococcaceae _V9D2013_group) and the eubacteria coprostanoligenes_group were significantly higher in the low concentrate diet group and the calcium pyruvate infusion group than in the high concentrate diet group, and the contents of Prevotellaceae genus Prevotellaceae_ YAB2003_group and Fibrobacter were significantly lower in the low concentrate diet group and the calcium pyruvate infusion group than in the high concentrate diet group.

Effect of Pyruvate Infusion on Ruminal A: P Ratio
An increase in the dietary NDF level will reduce DMI. The main reason for this X. Y. Lin et al. Advances in Bioscience and Biotechnology is that excessive intake of NDF increases the volume of rumen chyme and increases satiety, which reduces DMI. Arelovich et al. summarized a number of studies which demonstrated a significant inverse correlation between DMI and NDF content if the NDF content of dairy cows changed between 22.5% and 45% [6]. In addition to affecting feed intake, dietary NDF levels also affect feeding behavior and nutrient digestibility [7]. Increased feed intake may introduce other factors into our research, such as feeding time, chewing time, salivation, and so forth. The way in which feed is processed can also be an influencing factor. Although Jiang's research found that the roughage particle size has little effect on salivation [8], we know that changing the size of roughage grains will affect the degradation rate. Therefore, in order to eliminate the interference of factors which may affect the test, such as feed intake, picky feeding behavior and dietary processing methods, this experiment used continuous feeding to keep the rumen environment as stable as possible; each test animal was fed the same amount of feed, which was processed in the same way. The results showed that there was no significant difference in the effective degradation rate of fermentable carbohydrates between the two diets. This may be related to the way the diets were processed. Pulverization and granulation increase the contact area between dietary fiber and rumen microorganisms and promotes the degradation of dietary fiber. Studies have shown a linear decrease in the degradation rate occurs with increases in the average particle sizes of barley, wheat, rye, and pea flour [9]. Therefore, the effective degradation rate in the rumen of low concentrate diet was not much worse than that of the high concentrate diet. Although the difference in the effective degradation rates between the two diets was not significant, the difference in fermentable carbohydrate content was significantly higher in the high concentrate diet than that in the low concentrate diet. Therefore, the concentration of the fermented substrate produced per unit time from the high concentrate diet was significantly higher than that from the low concentrate diet. Our rumen fermentation results were identical to those of most studies. The increased in dietary concentrate proportion was accompanied by a decreased rumen A: P ratio. For the low concentrate diet, rumen infusion of low concentration pyruvate did not significantly change the rumen A: P ratio, while a significant increase in A: P ratio did occur with infusion of high concentration pyruvate. Chen et al. also found that the addition of pyruvate increased the ruminal A: P ratio [10].
After ruminal infusion of calcium pyruvate in the low concentrate diet group, the proportion of ruminal propionate was significantly reduced, the ruminal acetate-to-propylene ratio was significantly increased, and the proportion of hexanoic acid was significantly reduced. Meanwhile, the differences in ruminal pH were not significant, although the total volatile fatty acid concentration increased. In the process of producing VFAs, carbohydrate is first degraded into pyruvic acid by microbial cells. Infusion of calcium pyruvate is equivalent to increasing the substrate concentration of the fermentation. After infusion of calcium pyruvate in this experiment, the substrate level in the low concentrate diet  [14]. However, the changes in A: P ratio from the high-concentrate diet group followed the opposite trend compared to results of previous studies after the calcium pyruvate infusion. Ruminal infusion of calcium pyruvate may not mimic the true pyruvate distribution, and this may result in the differences. As pyruvic acid is mainly present in microbial cells, the concentration of pyruvate that can be detected in the rumen fluid is very low. In the high concentrate diet, the NFC level was high. Oligosaccharides and monosaccharides produced by NFC degradation can only be utilized by some microorganisms. For example, rumen Ruminococcus flavefaciens cannot utilize glucose, while rumen Ruminococcus albus preferentially utilize cellobiose. As a result, the fermentation substrates are distributed differently in the rumen, which affects the composition of the microflora. In the high-concentrate diet group, microorganisms associated with NFC degradation are multiplied, and monosaccharides or oligosaccharides produced by NFC degradation are rapidly captured and utilized by such microorganisms. Such microorganisms are often closely related to the formation of propionate, such as starch-degrading bacteria and lactic acid-producing bacteria. In the weakly acidic rumen environment, pyruvate becomes pyruvic acid. As a substrate of fermentation, pyruvic acid may be evenly or slightly unevenly distributed, absorbed and utilized by a wide range of microorganisms. For a low-concentrate diet, rumen fermentation occurs mainly as the acetic acid type. Therefore, after pyruvate infusion, calcium pyruvate is used by a large number of microorganisms related to acetic acid production, which promotes the growth of this group of microorganisms. Subsequently, the ruminal A: P ratio does not decrease, but increases. The results of rumen VFAs do not support one of the previous assumptions. The rate of carbohydrate fermentation may not be a major factor in regulating the A: P ratio. Therefore, there is skepticism about this hypothesis, which is because pyruvate infusion does not truthfully simulate the true distribution of pyruvate among rumen microorganisms under differential rates of degradation.

Differences in Microbial Flora between Rumen Fluid of High Concentrate and Low Concentrate Diets
The three most abundant phyla in the rumen fluid samples were Bacteroidetes, Firmicutes and Proteobacteria, and this result is consistent with those of previously studies [15]. Firmicutes were significantly more abundant in the low-concentrate diet group than in the high-concentrate diet group. Conversely, Proteobacteria and Bacteroides were significantly lower in the low-concentrate diet group than in the high-concentration diet group. At the genus level, Eubacteria (coprostanoligenes_group) and Ruminococcus (Ruminococcaceae _UCG-002 and Ruminococcaceae_V9D2013_group) were significantly lower in the high-concentrate diet group than in the low concentrate diet group, while Fi-

Changes in Rumen Flora after Infusion of Different Concentrations of Calcium Pyruvate
After rumen infusion of different concentrations of calcium pyruvate, the content of Proteobacteria was significantly lower than that of both the high concentrate and low concentrate groups. In the two calcium pyruvate infusion groups, Bacteroidetes were most common, followed by Firmicutes. The content of Euryarcheaota was also significantly higher in these two groups than in the high concentrate diet group. Euryarcheaota are a major category otherwise known as archaea, and include many methanogens in the rumen. At the genus level, the Wilcoxon rank sum test showed that the ruminal microbes in the pyruvate infusion group were not significantly different from the low concentrate diet group and significantly different from the high concentrate diet group. It is likely that compared to different types of dietary carbohydrates, pyruvic acid enters easily the cell membrane and is more easily utilized by most microorganisms, which results in smaller differences in distribution between microorganisms.
The high-concentrate diet group had significantly higher contents of Megasphaera, Succinivibrio (Succinivibrionaceae_UCG-002), Rumenococcus (Rumi-nococcus_2), Prevotella (Prevotellaceae_YAB2003_group), and Fibrobacter than did the two pyruvate infusion groups. Megasphaera elsdenii is the common Megasphaera species found in the rumen of juvenile and adult animals fed high-grain diets. As a lactic acid-utilizing bacteria, its main role is the fermenta-Advances in Bioscience and Biotechnology tion of D-type and L-type lactic acid, and the fermentation products are mainly propionic acid, butyric acid and the like. A previous study found that increases in dietary NFC/NDF significantly increased the number of M. eldenii [16]. Suc-cinivibrionaceae_UCG-002 is a hemicellulose-degrading bacterium. The fermentation products of Succinivibrio are mainly acetic acid and succinic acid. A study by R. John Wallace et al. found that ruminal contents of Succinivibrio were four times higher in low-methane-producing cattle than in high methane-producing cattle [17]. Based on this observation, it is a plausible assumption that Succinivibrio may help reduce inter-species hydrogen transfer and reduce methane production. At the phylum level, we found that Euryarcheaota was indeed less abundant in the high-concentrate diet group than in the lowconcentrate diet group. This means that the high concentrate diet group will produce less methane than the pyruvate infusion group. This also provides an explanation for the lower A: P ratio in the high-concentrate diet group than in the pyruvate-infused group.
The two pyruvate infusion groups had significantly higher contents of Sele-nomonas_3, Butyrivibrio_2, Russococcus (Ruminococcaceae_UCG-002, Rumi-nococcaceae_V9D2013_group), Lacnospiraceae_ND3007_group, Prevotella-ceae_UCG-003, Rikenellaceae_RC9_gut_group and Veillonellaceae_UCG-001 relative to the high concentrate diet group, but there were not significantly different from the low concentrate group. Selenomonas are starch-degrading bacteria; All Selenomonas strains can ferment starch and produce acetic acid and propionic acid, and cannot ferment structural carbohydrates; they can effectively utilize degradation products of structural carbohydrates, such as cellobiose. In addition, Selenomonas can also utilize lactic acid. However, unlike M. elsdenii, their ability to ferment lactic acid is inhibited by increased soluble sugar. Lachnospira are hemicellulose degrading bacteria, and their fermentation products are mainly formic acid, acetic acid, lactic acid and the like. Other studies have shown that Lachnospira can produce butyric acid [18]. It is currently believed that bacteria of genus Rikenella are widely distributed in the digestive tracts of animals. They also produce short chain fatty acids, which mainly include acetic acid and propionic acid. Zhang Ke et al. found that the proportion of Rikenella-ceae_RC9_gut_group in the rumen of Shanbei White Cashmere Goats increased from 28 days of age, and this was thought to be related to the digestion of cellulose [19]. Veillonella are lactic acid-utilizing bacteria that cannot ferment sugars, but can ferment lactic acid, pyruvic acid, fumaric acid, L-malic acid, and the like. These bacteria were less abundant in the rumen and do not act like M. elsdenii. In this test, the difference in Veillonella is probably due to the fact that the infusion of calcium pyruvate increases the fermentation substrate and benefits the reproduction of these bacteria.

Comparison of Effects of Rumen Microbes and Pyruvate Infusion Concentration on A: P Ratio
By comparison, it can be seen that calcium pyruvate infusion does not signifi- cantly affect rumen microorganisms. However, the flora associated with propionic acid production was significantly higher in the high concentrate diet group than in the two pyruvate infusion groups, which might result in the significantly higher A: P ratio in the high concentrate diet group than was seen in the two pyruvate infusion groups. The high concentrate diet group had a lower concentration of fermentation substrate than the other groups, and in particular, the high-concentration calcium pyruvate infusion group. However, greater numbers of microorganisms associated with propionic acid production are associated with a lower rumen A: P ratio. In contrast, when comparing the two calcium pyruvate infusion groups with the low concentrate group, calcium pyruvate infusion did not result in the expected decrease in the A: P ratio, and the A: P ratio was significantly increased in the high-concentration group. This result does not support the notion that the rate of carbohydrate fermentation is the main mechanism regulating A: P ratio, but the microbial composition results support the hypothesis that A: P ratio is regulated by microorganisms.
This is different from the conclusions of previous studies, which have mostly shown that elevated dietary NFC levels or promotion of carbohydrate fermentation can reduce A: P ratio [8] [20]. This may be due to an alteration of rumen flora structure through increases in NFC levels. However, some studies have found that a reduction in dietary NFC levels resulted in no change to A: P ratio, or even an elevated rumen A: P ratio [21] [22]. The reason may be that the magnitude of NFC change was insufficient to cause a change in the structure of the relevant microorganisms due to the redundancy of rumen microbial function and the strong elasticity and resilience of rumen microorganisms [23]. The rate of carbohydrate fermentation is likely to have greater effects on rumen microbes, rather than the insignificant effects from calcium pyruvate in this experiment.
This may be the reason that the results of this test differed from results from other studies, where the A: P ratio was reduced by changing feed source and processing. It can be seen that changes in the rumen A: P ratio are more closely related to changes of the rumen microflora structure.

Conclusion
The results showed that the main regulatory mechanism of rumen A: P ratio is more closely related to the structure of rumen microflora, and does not support the hypothesis that the rate of carbohydrate degradation regulates rumen A: P ratio.