Reducing dietary cation–anion difference seldom affects rumen fermentation, cellulolytic bacteria populations, and microbiota in goats

Background: Dietary cation–anion difference (DCAD) has been receiving increased attention in recent years; however, information on the rumen fermentation, cellulolytic bacteria populations, and microbiota of goats fed a low-DCAD diet is less. This study aimed to evaluate the feasibility of feeding a low-DCAD diet for goats with emphasis on rumen fermentation parameters, cellulolytic bacteria populations and microbiota. Growth performance, urine pH, and plasma metabolites were also analyzed as well. Materials and method: Eighteen goats were randomly allocated to 3 treatments with six replicates of each treatment and 1 goat per replicate. Animals were fed diets with varying DCAD levels at +338 (High DCAD; HD), +152 (Control; CON), and −181 (Low DCAD; LD). This study includes 15-d experimental period and 30-d adaption period. Results: The DCAD level did not affect the rumen fermentation parameters including pH, buffering capability, acetic acid, propionic acid, butyric acid, total volatile fatty acids, and ratio of acetic acid/propionic acid (P > 0.05). The 4 main ruminal cellulolytic bacteria populations including Fibrobacter succinogenes, Ruminococcus avefaciens, Butyrivibrio brisolvens and Ruminococcus albus did not differ from DCAD treatments (P > 0.05). The DCAD levels did not affect bacterial richness and diversity indicated by the indices Chao, Ace and Simpson and Shannon, respectively (P > 0.05). Both weighted UniFrac and unweighted UniFrac showed no difference in the composition of rumen microbiota for CON, HD and LD (P > 0.05). At the phylum level, Bacteroidetes was the predominant phylum followed by Firmicutes, Synergistetes, Proteobacteria, Spirochaetae, and Tenericutes, and they showed no difference (P > 0.05) in relative abundances except for Firmicutes, which was higher in HD and LD compared to CON (P < 0.05). At the genus level, relative abundance of 11 genera were not affected by DCAD treatments (P > 0.05). Level of DCAD had no effect (P > 0.05) on growth performance including dry matter intake, average net gain, average daily gain, and feed conversion ratio; and nutrients digestibility of crude protein, neutral detergent ber, acid detergent ber, and organic matter (P > 0.05). Urine pH in LD was lower than HD and CON (P < 0.05). LD resulted in higher plasma calcium than HD and CON (P < 0.05) but not for other plasma metabolites (P > 0.05). the above-mentioned parameters of rumen pH, buffering capability, volatile fatty acids of acid, propionic acid, butyric acid, total volatile fatty acids, and acetic acid/propionic acid proles; ruminal cellulolytic bacteria populations of Fibrobacter succinogenes, Ruminococcus avefaciens, Butyrivibrio brisolvens and Ruminococcus albus, and microbiota. Growth performance, acid-base balance, plasma calcium level and metabolites were also measured. The results should provide a comprehensive evaluation of the feasibility of feeding a low-DCAD diet to goats.

Ruminant diets usually include up to 70% roughage and so are ber-rich. As the most important digestive organ, the rumen is the main site for digestion and utilization of nutrient, especially for degradation of cellulose by cellulolytic bacteria. Among rumen measurements, pH is the key parameter determining rumen fermentation status and is often lowered when ruminants are fed high levels of concentrate, resulting in ruminal acidosis [21][22][23]. Bacteria account for 95% of the total amount of rumen microorganisms [24], and thus are the main factor to modulate digestive and metabolic activity of the rumen [25]. Butyrivibrio brisolvens, Fibrobacter succinogenes, Ruminococcus avefaciens, and Ruminococcus albus are reported the four most important cellulolytic bacteria for digestion and utilization of ber in the rumen [26][27][28][29].
However, to our knowledge, there is less information on the rumen fermentation, cellulolytic bacteria populations and microbiota for goats fed a low-DCAD diet, therefore, the present study was conducted to evaluate the effect of a low-DCAD on the above-mentioned parameters of rumen pH, buffering capability, volatile fatty acids of acid, propionic acid, butyric acid, total volatile fatty acids, and acetic acid/propionic acid pro les; ruminal cellulolytic bacteria populations of Fibrobacter succinogenes, Ruminococcus avefaciens, Butyrivibrio brisolvens and Ruminococcus albus, and microbiota. Growth performance, acid-base balance, plasma calcium level and metabolites were also measured. The results should provide a comprehensive evaluation of the feasibility of feeding a low-DCAD diet to goats.

Experimental design and animal management
The animal treatment procedures were approved by the local animal care and use committee. Using a randomized block design, 18 Qianbei miscellaneous goats (a native goat breed in the southwest of China; 30.07 kg initial weight and aged 13 months) were allotted to 3 treatments of six replicates of 1 goat per replicate. Animals were fed one of three diets with different DCAD levels (mmol/kg DM): +350 (HD), +100 (CON), and −150 (LD). Diets consisted of peanut straw (Arachis hypogaea), faba bean straw (Vicia faba) and concentrate which were mixed with NaHCO 3 (HD) or NH 4 Cl (LD) and were pelleted (4 mm diameter), with a concentrate: roughage ratio of 30:70.
Goats were fed in separate metabolic cages. The experiment duration was 45 d including a 30-d adaption period and 15-d trial period. The adaption period was divided into three stages. In the rst stage (1-12 d), goats were observed for health condition and treated for parasites and disinfected. During the second stage (13-18 d), goats were allowed to adjust to their respective diets. In the third stage (19-30 d), steady DM intake (DMI) was determined for individual goats. After that, in the trial period (31-45 d), goats were fed the treatment diets strictly according to the established DMI at 09:00 and 18:00. All goats had free access to water during the whole experiment. Ingredients and nutrient levels of diets for goats are shown in Table 1. The actual DCAD levels were measured as +338, +152, and −181, showing slight differences from the designed values of +350, +100, and −150 for HD, CON, and LD, respectively.  At the end of the feeding  experiment, dietary samples were composited and dried at 65°C and were ground to pass a 1-mm screen for proximate chemical composition determination of DM, crude protein (CP), organic matter (OM), Ca and P [30]; and neutral detergent ber (NDF) and acid detergent ber (ADF) [31]. An atomic absorption spectrophotometer (iCE 3000 SERIES, Thermo Fisher Scienti c, USA) was used to measure Na and K contents. Silver nitrate titration was used to determine Cl concentration. The S level was determined using the magnesium nitrate method as described by Wang and Beede [32]. The DCAD was calculated using the following equation: All goats were weighed on d 32 as the initial weight and on 46 d as the nal weight. The DMI was recorded daily for each goat calculated by allowance of refusals. The average net gain (ANG) was determined by subtraction of initial weight from nal weight. The average daily gain (ADG) was determined by dividing ANG with the trial period (15 d). The feed conversion ratio (FCR) was the ratio of DMI to ADG. Total feces was collected during the last ve consecutive days for nutrients digestibility analysis. Feces samples from individual goats were dried in an oven at 65°C for 48 h and ground to pass a 1-mm screen for approximate nutrient measurements: CP and OM [30] and NDF and ADF [31].

Rumen status
At 9:00, 13:00, and 17:00 on d 44, rumen uid was collected through the esophageal cannula via a vacuum pump (VP30, Labtech Instrument Co. Ltd, Beijing, China) and was measured pH using a pH meter (PHS-3C, Youke Instrument Co. Ltd, Shanghai, China). Sample collected at 17:00 was used to assess rumen buffering capability (BC) as described by Tucker et al. [33], and the operational procedure was titrating a 10-mL aliquot of the sample from the original pH dropping to pH 5 using 1 N HCl.
Rumen uid (20 mL) was used to detect volatile fatty acids (VFA), and the remaining sample was used for microbial high-throughput sequencing analysis. To determine the VFA, acidi ed samples were centrifuged at 4°C for 10 min (Thermo Fisher-ST 16R), and the supernatant fraction ltered through a 0.45-µm lter. The 1280 μL of ltrate was mixed with 600 μL of 20% metaphosphoric acid and 120 μL of crotonic acid (internal standard). The VFA concentrations in ltered samples were determined by gas chromatography (GC-2010-plus, Shimadzu, Japan) equipped with a ame ionization detector and a capillary column (SH-Rtx-Wax, Shimadzu), and nitrogen was used as the carrier gas. Rumen microbial high-throughput sequencing analysis was performed by TinyGen Bio-Tech (Shanghai) Co. Ltd.
PCR ampli cation of 16S rRNA genes and Miseq sequencing Ruminal uid samples were stored at −80°C and the DNA extracted from 200-mg samples using a QIAamp DNA Stool Mini Kit (Qiagen, Hilden, Germany) following the manufacturer's instructions. The DNA concentration and purity were checked by running samples on 1.0% agarose gels.
The PCR ampli cation of 16S rRNA genes was performed using general bacterial primers: 515F 5 -GTGCCAGCMGCCGCGGTAA-3 and 926R 5 -CCGTCAATTCMTTTGAGTTT-3 . The primers also contained the Illumina 5 -overhang adapter sequences for two-step amplicon library building, following the manufacturer's instructions for the overhang sequences. The initial PCR reactions were carried out in 25

Statistical analysis
The MIXED module in SAS 9.4 (SAS Institute Inc, Cary, NC, USA) was applied for analysis of experimental data. The DCAD levels (+338, +152, and −181) were designated as xed effects, and goats as the random effect, then Tukey's method was adopted to determine differences among means of the three DCAD treatments. The experiment results were expressed as mean ± standard deviation. Statistical signi cance was de ned as P < 0.05.

Rumen fermentation
There were no signi cant differences in rumen pH for the same sampling intervals (0, 4, and 8 h; P > 0.05; Table 2) and all the collected mean among the HD, CON, and LD treatments (0, 4, and 8 h; P > 0.05). The variation of DCAD had no effect on ruminal BC and levels of acetic acid, propionic acid, butyric acid, total VFA (TVFA), and acetic acid/propionic acid (A/P) in the goats (P > 0.05). 3 TVFA = acetic acid + propionic acid + butyric acid.

Rumen cellulolytic bacteria
The relative contents of F. succinogenes, R. avefaciens, B. brisolvens, and R. albus were not signi cantly affected among goats fed HD, CON, and LD diets (P > 0.05; Table 3). The proportions of F. succinogenes and R. avefaciens were markedly increased with lower DCAD compared to B. brisolvens and R. albus.  1). Rarefaction curves were established to quantify the OUT coverage of sampling and each rarefaction tended to be gentle with the increase of sequence number, and meanwhile, the OTU rank abundance in the 3 groups exhibited a gentler slope and wider distribution on the horizontal axis (Fig. 2).  Alpha diversity results showed that DCAD levels did not affect Chao, Ace, Simpson, and Shannon as listed in Table 4 (P > 0.05). The Chao, Ace, Shanno and Simpson chart of each group was also tended to be gentle corresponding to the increase of sequence number (Fig. 3).  According to Fig. 4, both weighted UniFrac (axis 1 + axis 2 = 66.7%, Fig. 4a) and unweighted UniFrac (axis 1 + axis 2 = 38.37%, Fig. 4b) were observed no difference in the composition of rumen microbiota for CON, HD and LD (P > 0.05). Taxonomic classi cation summary indicated that 16 phyla were tested in all samples (Fig. 5A). At the phylum level, Bacteroidetes (61.60%) was the predominant phylum followed by Firmicutes, Synergistetes, Proteobacteria, Spirochaetae, Tenericutes with average relative abundances of 25.32%, 5.84%, 1.82%, 2.08%, 1.2%, respectively, but there was no difference (P > 0.05) among the groups on the above phylum levels except for Firmicutes, which was signi cantly higher in HD and LD compared to CON (P=0.008, Table 5).
At the genus level, taxon displayed that the relative abundance of 11 genera were not affected by DCAD among all samples (P > 0.05; Fig. 5B). At the same time, Prevotella, Paraprevotella, Selenomonas, Ruminococcus, Ruminococcus, Butyrivibrio, Quinella, Fretibacterium and Treponema showed no grouping difference of the genera across treatments (P > 0.05). Among the genera with relative abundance exceeded 0.1%, prevotella was the dominant genus in each group with the highest proportion (Table 5).  Growth performance Levels of DMI were unaffected by DCAD variations (P > 0.05; Table 6). Lower DCAD had no effect (P > 0.05) on gwowth performance of nal weight, ANG, ADG, and FCR and digestibility of crude protein, NDF, ADF, and OM for goats.  Fig. 6b). Feeding of the LD diet resulted in the highest plasma Ca level (Table 7), which was signi cantly higher than both HD and CON (P < 0.05).There were no signi cant differences in plasma Glu, UN, ALT, AST, AKP, TP, Alb, GSH-Px, CAT, SOD, and MDA among the DCAD treatments (P > 0.05; Table 7).

Discussion
Rumen pH has a key role in measuring rumen status and can re ect the composition and abundance of rumen micro ora. Rumen health is maintained by an appropriate pH which is in range of 6.26 ~ 6.79 [34]. Rumen pH decreases when ruminants are fed high levels of concentrate. This is quite different from the acidi ed rumen status induced by increases in anion (Cl − and S 2− ) concentrations in the diet and which elevates Cl − and S 2− in the urine and so reduces urine pH. In this study, rumen pH was unaffected for all DCAD treatments and ruminal uid sampling time points. This was also shown in the study of Apper-Bossard [35].
Briggs et al. [36] found that rumen BC was closely related to rumen pH, and Church [37] argued that the rumen buffer system was controlled by pH, pCO 2 , and VFA. The rumen BC can maintain a stable rumen status by keeping any sudden rise or fall in rumen pH within a certain range [38]. The rumen has a relatively stable buffer system, which is closely related to the feed, saliva, and secretion of the rumen wall and maintains the rumen internal environment. In our study, the lack of signi cant difference in rumen BC and pH showed that DCAD reduction did not in uence the rumen internal environment. However, the rumen BC was numerically elevated with increased DCAD level on cows when adding NaHCO 3 [33,39,40]. The explanation could be attributed to the use of a different animal species and different dietary ingredients. Generally, up to now there has been insu cient information on rumen BC in goats, and further study is needed due to the importance of BC for rumen status.
The rumen VFA pro le is mainly impacted by the proportion of concentrate and forage in the diet.
Increasing DCAD, by adding K and Na, had no effect on rumen VFA concentration [41]. Tucker et al. [42] reported that the rumen VFA pro le was unaffected by DCAD levels of − 100, 0, 100, and 200. Apper-Bossard et al. [35] found no signi cant difference in VFA concentration with varying DCAD. These results are consistent with the present study in which rumen VFA concentration was not signi cantly affected by DCAD level. Correspondingly, the VFA pro les and A/P levels were unaffected by DCAD. This indicates that the rumen fermentation pattern was unaltered by the three DCAD treatments.
In the current study, DCAD variation had no in uence on the populations of B. brisolvens, F. succinogenes, R. avefaciens, and R. albus. This result indicates that reducing DCAD would not affect the growth and colonization of rumen cellulolytic bacteria. This is likely associated with stable rumen pH and VFA maintained by the constant ratio of concentrate to roughage (30:70) for the three DCAD diets used in this study. Grilli et al. [43] reported that a high proportion of maize disturbed the ruminal bacterial ecosystem of goats. Wang et al. [44] and Li et al. [45] noted that the relative quantity of ruminal B. brisolvens, F. succinogenes, R. avefaciens, and R. albus was improved with decreasing rumen pH. In our most recent study (unpublished data) with dairy goats as experimental animals fed 4 DCAD levels at 349, 120, and − 167, respectively, rumen BC was not in uenced. This result also supports the conclusions in the present study.
The microbiota composition of the gastrointestinal tract (GIT) in uence the health of animals as well as productivity [46], the structure and composition of ruminant microbial community are often devoted to evaluating the health status of the host and ensuring the healthy development of the rumen. Furthermore, the diversity and composition of the GIT microbiota can be in uenced by many factors including age, diet, feeding management and feed additives [47,48]. Microbial richness in rumen will be altered with dietary composition [49,50]. Dietary nutrients are fermented by rumen rumen microbial, such as bacteria, fungi and protozoa, and then degraded into VFA and MCP to provide energy for ruminants [51] and to guarantee the healthy and stable rumen environment [52,53]. In such reactions, rumen pH plays a decisive role in the composition and abundance of rumen micro ora, and is an important indicator re ecting whether the composition and abundance of rumen micro ora are normal [54,55]. Similarly, rumen microbial is also in uenced by animal species, diet composition and different ages [56][57][58].
The study of Zhang et al. [59] showed that rumen bacterial species, Chao index and Ace index were affected by rumen pH which can alter the bacterial community structure. Accordingly, Guo [60] found that decreased rumen pH would up-regulate bacterial diversity, composition and abundance of bacteria. In this experiment, rumen pH was not impacted by DCAD level, which exactly explained why the Chao index, Ace index, Simpson index and Beta diversity of the three groups were homogeneous. The Shannon index of the HD group was the highest, possibly because NaHCO 3 was added as a buffer to neutralize gastric acid and was essential for stomach health by creating a suitable internal environment for rumen microorganisms. Reducing DCAD had little effect on Shannon index. In addition, the dominant bacteria of the three groups in this experiment were Bacteroidetes followed by Firmicutes and Synergistetes, which was coincide with the results of previous studies on the basis of phylum [61][62][63]. In terms of genus level, the relative abundance of Prevotella, one of the primary protein-degrading microorganisms of Qianbei miscellaneous goats, was the highest, which was supported by the results of other ruminant studies [64]. The Firmicutes are a kind of intestinal bacteria related to obesity can degrade insoluble ber and play an important role in the process of substance metabolism in rumen. The abundance of Firmicutes in HD and LD were 51.95% and 53.98% higher than CON, respectively in the study. The up-regulation in Firmicutes abundance might be due to our application of a high forage diet. In summary, neither the reduction of DCAD have a negative effect on the bacterial diversity nor changed the content of the main bacterial genus.
As described above, rumen pH, BC, VFA, cellulolytic bacteria population and rumen micro ora diversity are crucial in the fermentation status of ruminants. The unaffected these parameters in the present study provide further reliable information and the feasibility on feeding a low-DCAD diet to goats.
Level of feed intake is the most important prerequisite for animal growth performance. Generally, pure anionic salt exerts some reduction on feed intake when it was simply mixed into diet, due to its bitter taste and poor palatability [65,66]. Therefore, improving palatability of anionic salt is important for DMI and growth performance. Our previous study [67] showed that DMI of goats fed a low-DCAD did not decrease because the anionic salts were mixed with molasses and dried distillers grains with solubles. Takagi and Block [68,69] also observed that reducing DCAD did not impact DMI containing anionic salts when feeding a total mixture ration. Diets were pelleted in our study, and therefore, they were unaffected for goats fed diet HD, CON, and LD. This indicates that DMI is unaffected by anionic salts inclusion as long as the bitter taste is concealed.
The levels of DCAD had no effect on goat nal weights. This can be attributed to the similar DMI level and possibly because the goats in this experiment were a local breed, with their adult steady body weight averaging as much as 35 kg, and thus there was limited potential for body weight gain. Accordingly, ADG and FCR did not show difference among goats provided HD, CON, or LD diets.
Nutrients digestibility can be used to measure the digestion and absorption degree of a diet. Correlative analysis in the study demonstrated the digestibility of CP, NDF, ANF, and OM were unaffected for goats fed the three DCAD diets. Apper-Bossard et al. [35] reached similar conclusions. Positive DCAD diet is bene cial to the growth and reproduction of rumen cellulolytic bacteria, and can maintain the activity of digestive enzymes in the digestive tract, thus improving nutrient digestibility [70]. However, different levels of DCAD did not alter the digestibility of CP, NDF, ADF and OM with the rumen pH and rumen cellulolytic bacteria community remained unchanged in this research.
Urine pH is a useful indicator to monitor the effect of a reduced DCAD diet on acid-base balance in goats and sheep [71,72], dairy cows [8,73,74], and buffalo [4]. This phenomenon can be explained using the strong ion difference theory of Stewart [75], who argued that with reductions of DCAD, the concentration of anions in blood would increase and cause the kidney to expel redundant H + in urine, resulting in lower urine pH. The recommended urine pH is 6.5-6.8, because too low a level would exert a burden on the kidneys [76][77][78]. Our results showed that urine pH value in goats fed LD was lower than HD and CON. This is accordant with the recommended level, and there is a strong association between DCAD and urine pH in the trial period, suggesting that the LD level is appropriate for the diet of goats.
Muscle contraction, conduction of nervous impulses, and signal transduction are closely dependent on blood Ca homeostasis. Following the study of Block [2] and subsequent results of ruminant researchers [71,79,80], reducing the DCAD level has been the most commonly used strategy to increase blood Ca levels in transition mammary animals [81]. Our previous study showed that reducing DCAD could increase the plasma Ca concentration of female goats [67]. In the current experiment, the LD caused higher plasma Ca level than HD and CON by 25.97% and 22.27%, respectively, indicating more stable blood Ca homeostasis. Horst et al. [82] and Goff and Horst [83] claimed this may be because LD-induced acidic status enhanced Ca absorption in the gastro-intestine and also increased Ca resorption in the bone, facilitating Ca matrix ow into blood for easier transfer from lumen to blood.
In the review of Khanal and Nemere [84] who described three steps for Ca absorption. First, Ca 2+ in ux occurs at the apical membrane via epithelial Ca 2+ channels, which include the transient receptor potential vanilloid receptor 6 (TRPV6). This step is considered to be rate-limiting for transcellular Ca 2+ transport [85]. Second, intracellular diffusion is facilitated by the vitamin D-dependent calcium binding protein D9k (CaBP-D9k) [86]. Last, extrusion at the basolateral membrane is achieved by either the Na + /Ca 2+ exchanger1 (NCX1) or the plasma membrane calcium ATPase 1b (PMCA1b) [87]. The proteins of TRPV6, CaBP-D9k, and PMCA1b are believed to be the three key factors involved in the Ca absorption process, therefore, we speculate that TRPV6, CaBP-D9k, and PMCA1b expression levels in the intestine might be upregulated for animals fed the LD diet. However, a lack of data on TRPV6, CaBP-D9k, and PMCA1b expression in the intestine limits our further discussion. Future work is needed in this area.
Blood measurements are useful to re ect the metabolic status of animals. Our results showed that all plasma levels of Glu, UN, ALT, AST, AKP, TP, Alb, SOD, GSH-Px, MDA, and CAT were unaffected by DCAD variation. This indicates that reducing DCAD had no effect on nutrient metabolite processes in the plasma. This is supported by previous study. Melendez and Poock [9] reported that lowering DCAD had little effect on blood Alb. Wu et al. [67] found that DCAD level (+ 300, + 150, 0, and − 150) had no signi cant effect on plasma GSH-Px and MDA content in female goats.

Conclusion
Reducing DCAD has little in uence on rumen status and rumen microbiota, showing no harmful to rumen fermentation of goats. Blood Ca level is increased and urine pH is decreased by DCAD reduction. These results indicate the feasibility of feeding a low-DCAD diets to goats. Availability of data and materials All data generated or analyzed during this study are included in this published article.

Ethics approval
All procedures with animals received prior approval from the Animal Care and Use Committee of the Guizhou University and followed the regulations and guidelines for animal care and welfare established by the committee.

Consent for publication
Not applicable.

Competing interests
All authors declare that there are no present or potential con icts of interest among the authors and other people or organizations that could inappropriately bias their work.  The VENN graph of rumen bacterial of goats fed diets with varying cation-anion difference levels, which displaying that the disposition of operational taxonomic units (OTUs) among the control group (CON, n=6), high dietary cation-anion difference (HD, n=6) and low dietary cation-anion difference (LD, n=6) in rumen uid microbiota.