Ex-vivo cow rumen fluid fermentation: changes in microbial populations and fermentation products with different forages

ABSTRACT This study investigated the effects of three forage treatments [herbal plantain (PL), ryegrass-white clover pasture (RW) and root crop fodder beet (FB)] on ex-vivo rumen fluid fermentations. Quantitative real-time Polymerase Chain Reaction (qPCR) and PCR-single-stranded conformation polymorphism (PCR-SSCP) analyses were undertaken for aliquots collected from the fermentations. The FB treatment had the highest volatile fatty acid (VFA) and lowest pH compared with other treatments. The RW culture that had a higher fibre content, had higher Fibrobacter succinogenes(F. succinogenes) and anaerobic fungi levels when compared with the FB and PL. Protozoans were however the most abundant microorganisms in the FB cultures, and they had the highest water-soluble carbohydrate (WSC) content. The abundance of F. succinogenes increased in the RW fermentation up to 12 h, and the abundance of anaerobic fungi increased in the RW fermentation up to 24 h. In all the treatments, no effect was found on the overall abundance of bacteria. The findings confirmed that changes in rumen microbial community and fermentation products are partly related to the WSC and fibre content of two novel forages (i.e. PL and FB) increasingly used in animal production.


Introduction
The rumen is a diverse microbial ecosystem comprising bacteria, archaea, protozoa, and fungi (Jami and Mizrahi 2012; Newbold et al. 2015). The microorganisms play essential roles in digesting feeds and supplying nutrients and energy to host ruminant animals (Janssen and Kirs 2008;Belanche et al. 2012). It is accepted that feeding different forages to ruminant animals can induce changes in ruminal fermentation products and microbial populations (Burke et al. 2006;Busquet et al. 2006). However, despite numerous in-vitro studies describing ruminal fermentation profiles [e.g. volatile fatty acid (VFA) production] of commonly used forages in New Zealand (Roca et al. 2010), little research has been conducted to understand and compare how microbial populations and communities change during the fermentation of different forages.
Exploring changes in the structure, function and diversity of the rumen microbial populations in response to different feeding regimes, is an important part of assessing strategies to mitigate adverse effects and adapt to changes in the host's diet (Pitta et al. 2010;Stiverson et al. 2011). The use of traditional culture techniques to analyse the rumen microbiome and specific microbial populations are both time consuming and cumbersome. Tajima et al. (2001) made the first attempt to quantify ruminal microorganism populations using a real-time Polymerase Chain Reaction (PCR) technique. They concluded the approach could offer a sensitive, efficient, simple and reliable means of quantifying populations. This PCR approach is now widely used (Fernando et al. 2010;Leng et al. 2011;Thoetkiattikul et al. 2013) and adapted for use in this study to quantify ruminal microorganism changes induced by three different feeds. PCR-single-stranded conformational polymorphism (PCR-SSCP) offers a simple, inexpensive and sensitive method for differentiating multiple DNA fragments containing different nucleotide sequences (Zhou et al. 2014). A PCR-SSCP approach allows investigation of the rumen microbial community and complements the microbial population quantification results derived from the use of realtime PCR. Together, PCR-SSCP and real-time PCR techniques would provide an improved understanding of how rumen microbial populations and communities change when incubating ex-vivo derived rumen fluid in a liquor with different forages. This study was therefore designed to compare diverse forages used in New Zealand animal production: commonly used perennial ryegrass/white clover pasture (RW; Cheng et al. 2016) and with two increasingly popular forages [herbal plantain (PL; Cheng et al. 2017) and root crop fodder beet (FB; Jenkinson et al. 2014)], for in-vitro fermentation and microbial dynamics.

Forage harvesting
The FB (Beta vulgaris L. var. iutea Dc), PL (Plantago lanceolata L.) and RW (Lolium perenne L. /Trifolium repens L.) were harvested from the Lincoln University Research Dairy Farm (LURDF) Lincoln, New Zealand (43°38 ′ S, 172°27 ′ E). The PL and RW herbage were harvested to approximately 5 cm above the ground level, and the bulb of the fodder beet was chopped into small pieces. The forages were then freeze-dried and then ground through a 1 mm sieve (ZM200, Retsch).

Rumen fluid collection, preparation and in-vitro fermentations
Rumen fluid (approximately 4000 mL) was collected from four fistulated Friesian × Jersey-cross dairy cows that were lactating and had been grazing PL at the LURDF. The fluid was collected into two pre-heated 2000 mL thermos bottles and rapidly transferred to the laboratory where it was blended and then filtered through four layers of cheesecloth into a pre-heated (39.5°C) flask. The flask was continuously purged with CO 2 .
To prepare the artificial rumen liquor (Buffer + Rumen fluid) for the fermentations, 400 mL of the blended and strained rumen fluid was added to 1596 mL of the buffer in each fermentation jar. Freeze-dried FB, PL and RW (25 g) was assigned to three of the fermentation jars, with the remaining jar containing only rumen fluid liquor (made up to the same volume) as a control. The fermentations were then carried out at 39.5°C in the DAISY II-200/220 using an approach described by Martinez et al. (2006) and Anassori et al. (2012). The fermentations were all duplicated to enable assessment of experimental error.

Measurements and analysis
The three forage types were also sub-sampled (100 g) and analysed for chemical composition using wet chemistry. The fibre content of forage was determined using an enzymatic-gravimetric method (McCleary et al. 2012) and the degradability of acid detergent fibre (ADF) and neutral detergent fibre (NDF) of the substrate was analysed using the Van Soest detergent procedure (Van Soest et al. 1991;Hindrichsen et al. 2006). Total soluble carbohydrate content was determined using the Anthrone reaction (Jermyn 1956;Pollock and Jones 1979) and digestibility was assayed using an in-vitro pepsin cellulase method (Clark et al. 1982).
The pH of the fermentations was recorded for each forage using a benchtop pH meter (Orion 2-Star, Thermo Scientific, Beverly, USA), and the fermentations were sampled after 0, 4, 8, 12, 24, and 48 h of incubation. Samples were collected at each incubation time, two 1.5 mL samples for VFA and ammonia analysis, and one 18 mL sample for microbial DNA extraction. The fermentation jars were flushed with CO 2 and re-sealed after each sampling.
The 1.5 mL sample for ammonia-N (NH 3 -N) determination was immediately acidified, and centrifuged at 3000 g for 20 min at 4°C. The supernatants were then transferred into new tubes and frozen at −20°C until analysis. NH 3 -N was determined by an enzymatic UV method using a Randox ammonia kit and the Randox Rx Daytona analyser (United Kingdom). The VFA concentrations (total, acetate, propionate and butyrate) were measured using a Gas Chromatograph (GC: Shimadzu GC-2010, Japan) fitted with an SGE BP21 30 m × 530 µm × 1.0 µm wide-bore capillary column.

DNA extraction from rumen fluid samples
Microbial genomic DNA was extracted from the 18 mL samples using a Bioline Isolated Fecal DNA kit (Bioline Reagent Ltd., UK) and following the manufacturer's instructions.

PCR-single-stranded conformational polymorphism analysis
The microbial community in each of the cultures was analysed using a PCR-SSCP approach. A fragment (approximately 194 bp) of prokaryote 16s rRNA gene was amplified using total bacteria primers (Carberry et al. 2012) while a fragment (approximately 357 bp) of the eukaryote 18S rRNA gene was amplified using universal primers (see Table 1). Amplifications were performed in a 20 μL reaction containing 1 μL genomic DNA, 0.25 mM of each primer, 150 mM of each dNTP (Eppendorf, Hamburg, Germany), 3.5 mM Mg 2+ , 0.5 U of Taq DNA polymerase (Qiagen, Hilden, Germany), and 1× reaction buffer supplied with the polymerase enzyme. Amplification was carried out in an iCycler (Bio-Rad, Hercules, CA), with the thermal profiles shown in Table 1. A negative control without template DNA was included in all amplifications.
After confirming amplification using electrophoresis in 1% agarose gels, the PCR products were analysed using an SSCP approach. A 1.5 μL aliquot of each amplicon was mixed with 10 μL of loading dye (98% formamide, 10 mM EDTA, 0.025% bromophenol blue, 0.025% xylene-cyanol), and after denaturation at 95 C for 5 min, samples were rapidly cooled on wet ice and then loaded on 16 cm × 20 cm, 14% polyacrylamide gels. Electrophoresis was performed in 0.5× TBE buffer for 18 h using Protean II xi cells, either at 280 V and 22°C for 16S rRNA amplicons, or at 250 V and 25 C for 18s rRNA amplicons. The gels were silver-stained by the method of Byun et al. (2009).

DNA sequencing and sequence analysis
Representative single-stranded DNA bands were excised as gel slices from the polyacrylamide gel, macerated, and then used as templates for re-amplification with the original rRNA primers. These amplicons were then sequenced in both directions at the DNA Sequencing Facility at Lincoln University, New Zealand. Sequence analysis was performed using DNAMAN (v. 6.0, Lynnon BioSoft, Vaudreuil, Canada). The BLAST algorithm was used to search the NCBI GenBank databases (http://www.ncbi. nlm.nih.gov/) for homologous sequences.

Quantitative PCR
Primers and quantitative real-time PCR (qPCR) cycling conditions are given in Table 1. All primers were synthesized by Integrated DNA Technologies (IA, USA). Genomic DNA from the rumen fluid samples were diluted 1 in 10 with nucleasefree deionized water. Additionally, 5 μL aliquots from each genomic DNA sample were combined to create a pooled sample. This pooled DNA was used to generate standard curves by preparing 1 in 5 dilutions and including the dilution series with each qPCR run.
A typical 16 μL qPCR reaction contained 8.0 μL of SYBR®Premix Ex Taq TM (TaKaRa, Norrie Biotech, Auckland, New Zealand), 0.8 μL each 5 μM primer (250 nM final concentration), 1.5 μL of template DNA and nuclease-free deionized water to the final volume. No template control (NTC) samples were included with each run. All reactions were set up using the CAS-1200 Robotic liquid handling system (Corbett Life Science, BioStrategy, Auckland, New Zealand) and relative qPCR analysis was performed on a Rotor-Gene TM 6000 real-time rotary analyser (Corbett Life Science). Fluorescence intensity was measured at the end of each extension step at 72 C and to confirm the specificity of the reaction, a melting curve analysis was performed at the end of each run. Data analysis was carried out using Rotor-Gene TM 6000 series software 1.7 (Corbett Life Science). The abundances for each gene were expressed as fold-changes relative to the control samples.

Statistical analysis
All data were analysed using GenStat 16 (Lawes Agricultural Trust, Rothamsted, UK). Analysis of variance (ANOVA) was conducted to examine associations among rumen fermentation variables, and measured relative microbial abundance values, with the three dietary treatments included as fixed effects in the analysis. Significance differences between treatment means were identified using protected least significant differences (LSD) (p < .05) following a significant ANOVA result. Mean separation was tested using the LSD method at the 5% level.

Results
The composition of RW, PL and FB forages are presented in Table 2. RW had 208 g and 303 g NDF/kg DM more than PL and FB, respectively. The water-soluble carbohydrate (WSC) was 455 and 508 g/kg DM higher for FB than RW and PL, respectively.
The average pH, and concentrations of NH 3 -N and VFA produced across the period of incubation for the control, RW, PL and FB are presented in Table 3. The average pH (P = .024) and the ratio of acetate to propionate (P = .008) were highest, and concentration of total VFA (P = .015), propionate (P = .009) and acetate (P = .004) were lowest, for the control compared with the other treatments. The average concentration of NH 3 -N was higher (P = .021) for PL than other treatments, but similar in the control, RW and FB fermentations. The concentration of butyrate was lower (P = .004) for the control compared to the other treatments, and higher (P = .004) for FB, when compared to RW and PL. There was a significant treatment × time of incubation interaction (Table 3) for pH (P = .014), concentration of ammonia-N (P = .05), total VFA concentration (P < .001), and propionate (P < .001), acetate (P < .001), and butyrate (P < .001) concentration. The pH of all forages over 48 h incubation averaged at 6.37 (Figure 1(a)), while NH 3 -N and total VFA averaged at 15.6 mmol/l (Figure 1(b)) and 41.5 mmol/l (Figure 1(c)). The pH was higher and concentrations of total VFA, acetate and propionate were lower for control than other treatments at 4, 8, 12, 24 and 48 h of incubation (Figures 1(a,c) and 2(a,b), respectively). The concentration of NH 3 -N was lower for PL than other treatments at incubation time of 4, 24 and 48 h, and NH 3 -N was also lower for PL than control and RW at an incubation time of 8 h (Figure 1(b)). The butyrate concentration was the highest for FB and the lowest for control at 4, 8, 12, 24 and 48 h of incubation (Figure 2(c)). The butyrate concentration increased from 1.9 at 0 h to 11.2 mmol/l at 48 h (Figure 2(c)).
The F. succinogenes population in the RW fermentations increased at least 9-fold between 0 and 8 h, while it tended to be lower in PL than in the other two treatment groups (Figure 3(a)). Total bacterial counts increased for all three treatments between 0 and 8 h (Figure 3(b)), but declined after that. Anaerobic fungi levels in the RW fermentations increased at least 28-fold from 0 to 24 h, but then declined. In the other feed incubations, there was little difference between anaerobic fungi levels at different incubation times. Protozoa levels increased up to 8 h of incubation for the three feed supplemented fermentations, but then all declined as did the control levels of protozoa. The results suggest protozoans were the dominant type of microorganism in the FB cultures, but that anaerobic fungi were the dominant microorganisms in the RW and PL cultures.
The SSCP banding pattern for 16s rRNA was very complex and thus no attempt was made to sequence any of the bands (results not shown). Analysis then focussed on the SSCP banding pattern for 18s rRNA, with the PCR-SSCP profiles indicating that FB in the cultures resulted in different microbial communities growing (Figure 4). The microbial community appeared to have a large number of dominant groups, mainly protists and anaerobic fungal communities. The sequences obtained from sequencing SSCP bands were 100% similar to known sequences in GenBank and the species identified were distinct. They included: Eremoplastron dilobum, Entodinium caudatum, Dasytricha ruminantium and other uncultured protists, ciliates and fungi, including a Piromyces species.

Discussion
The chemical composition of three feed types used in this study were in line with previous publications (Cheng et al. 2016(Cheng et al. , 2017. The relative abundance of a microbial group in a given culture generally reflects its ability to adapt to the conditions and compete for the nutrients available (Belanche et al. 2012). In this study, microbial community dynamics in response to forage fibre or fermentable carbohydrates changes were assessed by quantitative real-time PCR. The abundance of measured microorganisms from all treatment groups in this study reached a peak prior to 24 h of incubation, indicating an active fermentation process took place. The difference in relative abundance and diversity of microbial species likely reflects the difference in the fibre degradation and carbohydrate fermentation for the three forages.
Anaerobic fungi and anaerobic bacteria, including F. succinogenes, are commonly found in the rumen (Krause et al. 2003) and they can release cellulolytic enzymes to break Table 3. Ammonia (NH 3 -N) concentration, pH and volatile fatty acid (VFA) production of rumen liquid incubated in vitro without herbage or crop (control), or with perennial ryegrass/white clover (RW), plantain (PL) or fodder beet (FB) added. .121 a-c Means within a row with different superscripts differ (p < .05). d Sampling points, during which rumen liquor was sampled (0,4,8,12,24, and 48 h of incubation). down fibre (Kamra 2005;Sun et al. 2008;Kebreab et al. 2009). In the current study, both anaerobic fungi and F. succinogenes were found to be the dominant microorganisms present in RW supplemented rumen liquor fermentations, and this finding is consistent with RW having the highest fibre (i.e. NDF) content compared with the other treatments. The fermentation product of F. succinogenes should mainly be acetate and propionate (Koike and Kobayashi 2001), and this likely explains the lack of a difference observed between acetate and propionate in PL and FB supplemented fermentations in this study.
The relative abundance of F. succinogenes and anaerobic fungi in RW peaked at 8 and 24 h respectively, and this is in agreement with Paul et al. (2010). It seems likely that the total bacteria population may have had a limited contribution to the overall difference in fermentation between the three treatment groups, as the total bacterial abundance was similar at all sampling times. Further, the three treatments produced a similar response, with their total bacterial population peaking at 8 h and then dropping.
Protozoans live as commensals in the rumen, consuming bacteria and anaerobic fungi. They are mostly associated with the presence of easily fermentable carbohydrates and proteins (Sylvester et al. 2009;Jami and Mizrahi 2012;Parmar et al. 2015). Protozoan populations have been observed to have positive relationships with butyrate and isobutyrate production in the rumen (Carberry et al. 2012;Newbold et al. 2015). It is clear that the greater abundance of protozoa and high butyrate levels may be due to the higher concentrations of readily fermentable carbohydrates present in FB (Newbold et al. 2015). An increased abundance of protozoa was found in FB supplemented fermentations, and these contained higher levels of WSC than the other forage treatments at both 8 and 12 h. The abundance of protozoa in RW was similar to those in PL at all sampling times and this is consistent with PL and RW having similar WSC levels. Butyrate is the main product of WSC fermentation in the rumen, and it was higher in FB, compared to both PL and RW between 8 to 48 h of incubation. As might be expected, WSC fermentation also resulted in the lowest pH and highest VFA concentration being detected in the FB samples.
rumen fluid was harvested from cows that had ingested RW previously, the seven distinct strains found in the RW group suggest that the fermentation process may have enhanced the abundance of some protozoan species compared with the control. Interestingly, it was observed that only protozoan communities exhibited distinctive SSCP bands, and that there were more protozoan bands in FB and RW, than in PL and the control. This is likely to reflect a higher diversity of protozoan communities in the FB and RW supplemented cultures. Cost and time constraint precluded analysis of all the PCR-SSCP bands, and thus only distinct single bands and common patterns were selected for sequencing.
Interestingly, a Piromyces species was found. This organism can degrade cellulose in culture (Roger et al. 1992), and this supports the contention that RW and thus high NDF levels may have stimulated the growth of fibre degradation related microorganisms, such as the Piromyces spp., during the fermentation process. In comparison, the increased protozoa numbers and protozoan communities, and apparent dominance of Entodinium spp. are often found to be associated with an increased starch content in ruminant diets, and in general as the proportion of butyrate increases with increasing protozoa numbers in the rumen (Newbold et al. 2015). This agrees with the positive association between protozoa abundance and butyrate production in FB in this study. The high abundance of unclassified protozoa in the FB group could also reflect the highest content of soluble carbohydrate in this fermentation.

Conclusions
The abundance of microbes in in-vitro cultures showed changes in all treatment groups. F. succinogenes and anaerobic fungi were found to be the dominant strains in the higher fibre RW treatment group compared to the FB treatment group. The populations of anaerobic fungi were higher in RW and PL treatment groups in the middle and late stage of culture, respectively. The FB treatment had the highest protozoa population, concentration of VFAs and NH 3 -N, and lower pH than RW and PL. The findings further confirm that changes in rumen microbial community and fermentation products are related to the WSC and fibre content of forages.

Disclosure statement
No potential conflict of interest was reported by the authors.