Bas Boots, School of Biosystems Engineering, Agriculture and Food Science Centre, University College Dublin, Belfield, Dublin 4, Ireland. E-mail: email@example.com
Anaerobic rumen fungi (Neocallimastigales) play important roles in the breakdown of complex, cellulose-rich material. Subsequent decomposition products are utilized by other microbes, including methanogens. The aim of this study was to determine the effects of dietary changes on anaerobic rumen fungi diversity.
Methods and Results
Altered diets through increasing concentrate/forage (50 : 50 vs 90 : 10) ratios and/or the addition of 6% soya oil were offered to steers and the Neocallimastigales community was assessed by PCR-based fingerprinting with specific primers within the barcode region. Both a decrease in fibre content and the addition of 6% soya oil affected Neocallimastigales diversity within solid and liquid rumen phases. The addition of 6% soya oil decreased species richness. Assemblages were strongly affected by the addition of 6% soya oil, whereas unexpectedly, the fibre decrease had less effect. Differences in volatile fatty acid contents (acetate, propionate and butyrate) were significantly associated with changes in Neocallimastigales assemblages between the treatments.
Diet clearly influences Neocallimastigales assemblages. The data are interpreted in terms of interactions with other microbial groups involved in fermentation processes within the rumen.
Significance and Impact of the Study
Knowledge on the influence of diet on anaerobic fungi is necessary to understand changes in microbial processes occurring within the rumen as this may impact on other rumen processes such as methane production.
The bovine rumen harbours complex anaerobic microbial communities (Russell and Rychlik 2001; Zoetendal et al. 2004). These communities are responsible for the hydrolysis of fibrous, cellulose-rich plant material to mono- and disaccharides, and their subsequent fermentation to short-chain fatty acids, including propionate, acetate, butyrate, lactate and succinate (Hobson and Stewart 1997), which serve as major carbon and energy sources for the ruminant. These communities generally consist of a consortium of phylogenetically and functionally distinct groups, including bacteria, archaea, chytrid fungi and protozoal ciliates, adapted to a strict anaerobic environment (Hobson and Stewart 1997). The contribution of anaerobic fungi to the solubilization of plant material has been shown to be more significant than that of cellulolytic rumen bacteria (Joblin et al. 1989), possibly due to the broader range of enzymes produced combined with physical alteration of the substrate during fungal colonization (Joblin et al. 1989; Orpin and Joblin 1997). All anaerobic gut fungi are now classified in a single order (Neocallimastigales) within the phylum Neocallimastigomycota (Hibbett et al. 2007), and are found in the rumen, hindgut and faeces of ruminant and nonruminant herbivorous mammals as well as in herbivorous reptiles (Liggenstoffer et al. 2010). Six genera and around 20 species have now been described (Griffith et al. 2010), although a large number of uncharacterized isolates has been reported (Griffith et al. 2010; Liggenstoffer et al. 2010; Kittelmann et al. 2012).
During anaerobic hydrolysis and fermentation processes, H2 and CO2 are generated as by-products of microbial metabolism. H2 and CO2 are substrates for methanogenic archaea, which convert them to the greenhouse gas methane. CH4 production from ruminants is a significant contributor to global greenhouse gas emissions, particularly in economies reliant on livestock agriculture (Johnson and Johnson 1995). Methane is 23 times more potent than CO2 as a greenhouse gas (Wuebbles and Hayhoe 2002), and there is now considerable focus on understanding the processes underlying enteric CH4 production, potentially leading to mitigation strategies for agricultural systems (Martin et al. 2010).
Due to the complex and interlinked nature of rumen microbial communities, changing diet (and thus substrates for primary degradation) can have cascading effects on rumen microbial metabolism, with consequent changes in both rumen organic acid profiles and methane levels produced (Wolin and Miller 1997). It has been shown that methane emissions from cattle can be decreased by supplementing diets with certain organic oils (e.g. Jordan et al. 2006; Calsamiglia et al. 2007; Grainger et al. 2010; Lillis et al. 2011). Essentially, methane production can be lowered by reducing the biological availability of H2 (Joblin 1999), accomplished by increasing the concentration of substrates that can act as alternative hydrogen sinks, such as propionate. Altering substrate availability through dietary modifications has been shown to affect anaerobic gut fungi. For example, using automated ribosomal intergenic spacer analysis (ARISA), Denman et al. (2008) observed changes in the composition of anaerobic rumen fungi in animals fed high-fibre diets compared with those fed with diets based on grains, with a diet high in plant fibre maintaining greater fungal diversity within the rumen than a high-grain-based diet.
DNA-based fingerprinting techniques have been shown to be highly useful for assessing effects of dietary changes on microbial communities within the rumen (Lillis et al. 2011; deMenezes et al. 2011). In a parallel study, Petrie et al. (2010) showed that dietary supplementation (alteration of forage to concentrate ratio and addition of soya oil) significantly reduced methane emissions from bovines, and this was linked to changes in bacterial community structure (Lillis et al. 2011). With the known role of anaerobic rumen fungi in polysaccharide degradation and production of substrates for methanogenic activity, it is hypothesized that the same dietary regime also influences the composition of anaerobic fungal assemblages (members of the order Neocallimastigales) within the rumen resulting in altered fungal biodiversity.
Materials and methods
The design of the experiment was as detailed in the parallel study described by Petrie et al. (2010). Briefly, four rumen-cannulated Limousin X steers with a mean body weight of 484 ± 26·4 kg were allocated in a latin square arrangement with four 28-day periods. Animals were allocated to one of two levels of dietary concentrate/forage ratio (50 : 50 or 90 : 10; with barley straw as the forage source) and one of two levels of dietary soya oil inclusion (0 or 60 g soya oil per kg dry matter) in a two × two factorial design. Diets were offered at 95% of voluntary daily mean intake and were isonitrogenous (140 g crude protein per kg dry matter). Feed intake was measured daily (with a mean of 9·2 kg day−1 observed) and here was no effect of any dietary factor on the daily feed intake. Over a period of 10 days, the concentrate/forage ratio was gradually increased from 0 : 100 to 50 : 50, with barley straw replacing grass silage as the forage source. The diet containing 0 g soya oil per kg dry matter (offered to animals on the 50 : 50 concentrate/forage ratio during the main experiment) was offered to all animals during this acclimatization period. From day one to seven of each experimental period, dietary concentrate/forage was increased in a stepwise fashion, and animals were then maintained on the appropriate concentrate/forage ratio and soya oil level for the following 21 days. The concentrate and forage were offered in two equal portions at 08:00 and 16:00 h daily. All animals used in this experiment were cared for under license in accordance with the European Community Directive, 86-609-EEC. Fresh drinking water was continually available throughout the experiment.
Rumen sampling, volatile fatty acids and CH4 measurements
After 21 days, rumen samples were collected 4 h postfeeding. Samples from the fluid digesta were taken using RT Rumen Fluid Collector tubes (Bar Diamond Inc., Parma, ID, USA) connected to a 50-ml syringe. The collection tubes had a stainless steel filter (~0·36-mm pore size) attached to the tip to omit solid fragments. Samples were taken from three locations within the rumen cavity, combined to form a single sample and 5 ml was taken for microbial analyses. In addition, rumen solids were collected from three locations within the fibrous mat (the layer of long, fibrous material floating on top of liquid digests, which is regurgitated and masticated) using sterile gloves. These were then combined to form a single sample, passed through a double layer of muslin to separate any liquid from the solid phase and 10 g of the solid matter was retained for microbial analyses. All samples for microbial analyses were immediately stored at −80°C. Ammonia and volatile fatty acid content (including acetate, butyrate, propionate, isobutyrate and isovalerate) were determined as reported in Petrie et al. (2010). Because reliable methane measurements cannot be obtained from cannulated animals due to potential gas leaks, methane emission measurements were obtained from a parallel set of noncannulated Limousin X steers of the same age, treated identically at the same time. Methane emissions were measured using a modification of the SF6 tracer gas technique of Johnson et al. (1994), as previously described by Lovett et al. (2003). Detailed procedures are described in the parallel study by Petrie et al. (2010).
DNA extraction and Neocallimastigales-specific fingerprinting
DNA was extracted separately from liquid and solid rumen samples via the method of Yu and Morrison (2004), and checked for quality and quantified using a NanoDrop™ spectrometer (Thermo Scientific, Wilmington, DE, USA) prior to standardisation to 50 ng DNA per μl. Each sample was extracted in triplicate. Amplification of the Neocallimastigales-specific internal transcribed spacer (ITS) region was performed with the forward (Neo-18S, 6FAM-labelled; 5′-AATCCTTCGGATTGGCT-3′) and reverse (Neo-5·8S; 5′-CGAGAACCAAGAGATCCA-3′) primer pair (Edwards et al. 2008), producing fragments lengths of 350–450 bp. All amplifications were carried out in 50 μl volumes containing 10 μl of 5 × Mg-free PCR buffer, MgCl2 (1·5 mmol l−1), 15 pmol of each primer, 200 μmol l−1 of each dNTP, 25 μg BSA, ~10 ng of extracted total DNA and 2·5 U Go-Taq DNA polymerase (Promega, Southampton, UK). The PCR conditions included an initial denaturation at 95°C for 5 min, followed by 10 cycles of 95°C (30 s), 68°C (45 s, −1°C each cycle) and 71°C (30 s). After this, 35 cycles of 95°C (30 s), 58°C (30 s) and 72°C (30 s), followed by a final extension step at 72°C (6 min). Successful PCR amplification was checked by agarose gel electrophoresis. After purification using a High Pure PCR Purification Kit (Roche, Mannheim, Germany), 1·0 μl of the PCR product was mixed with 0·25 μl of 600LIZ size standard (Applied Biosystems, Foster City, CA, USA) and 9·75 μl formamide (Applied Biosystems), and was denatured at 95°C (5 min). All fragment lengths were determined by electrophoresis using an AB3031 automated sequencer (Applied Biosystems). Electrophoresis was carried out on a 36 cm capillary, and fragments were separated at 60°C and 4 kV for 120 min to allow for separation of the larger fragments. Analysis of fragment profiles was performed using the Genemapper (Applied Biosystems) software. Fragments were sorted using the Ribosort package (Scallan et al. 2008) in R (ver. 2·15·0) (R Development Core Team 2010) and fragments with relative abundance <1% of the total abundance within a sample were regarded as background noise and excluded from further analyses (Gleeson et al. 2005). Fragments that differed by <0·5 bp in different profiles were considered identical, and those occurring in two of the three extractions were included and merged and averaged (Dunbar et al. 2001).
All data were screened for outliers and homogeneity of variance (Brown and Forsythe's Test), and no transformation was necessary for Neocallimastigales diversity and associated diversity indices (richness, evenness and Shannon index). Analyses of variance were performed and included the fixed factors ‘phase’ (solid and liquid), ‘concentrate/forage ratio’ (90 : 10 and 50 : 50) and ‘soya oil’ (0 and 6%). All univariate statistical analyses were carried out using sas for Windows ver. 9·1·3 SP4 (SAS Institute, Chicago, IL, USA).
Multivariate methods were used to analyse the effect of concentrate/forage ratio and soya oil dietary treatments on Neocallimastigales assemblages using the same model for the univariate data analysis. Similarity matrices were computed using Bray-Curtis dissimilarities (Bray and Curtis 1957) between samples on fourth-root transformed abundance data. Permutational multivariate analyses of variance (permanova, Anderson 2001) were computed to test the null hypotheses of no differences between assemblages across the treatments (at a significance level of α = 0·05) after testing for homogeneity of multivariate dispersions with deviations from centroids. All P-values were based on 9999 permutations of residuals under the reduced model. In addition, to visualize multivariate patterns, canonical analyses of principal coordinates (CAP, Anderson and Willis 2003) were used as a constrained method of ordination. Similarity percentages (simper) using Bray-Curtis similarities were computed to examine the contribution of each fragment to dissimilarities between, and similarities within the treatments. Furthermore, associated data on ammonia and volatile fatty acid contents were used to explain differences between Neocallimastigales assemblages in both the liquid and solid samples by computing multiple regression models as described by others (Legendre and Anderson 1999; McArdle and Anderson 2001). This procedure tests the null hypothesis that there is no relationship between (individual or sets of) ‘predictor’ variables. For this, ammonia and volatile fatty acid data were square root transformed to account for outliers, and normalized to account for the different scales of measurement. Proportions of variance explained by the predictor variables are expressed by adjusted-R2, with higher numbers representing more of the variability explained. Redundancy analyses (RDA) were then computed to visualize relations between microbial diversity and abiotic factors. All multivariate statistical analyses were computed using Primer ver. 6 with the permanova add-on (Primer Ltd, Plymouth, UK).
Neocallimastigales diversity in rumen solid and liquid phases
A total of 64 Neogallimastigales fragments were amplified from DNA of the 32 rumen samples obtained from cannulated steers offered diets differing in concentrate/forage ratio (with or without 6% soya oil) using primers specific for Neocallimastigales (Edwards et al. 2008). A species accumulation curve (Fig. 1) was constructed including both solid and liquid samples, revealing that the majority of Neocallimastigales diversity in the rumen was assessed using the ARISA fingerprinting procedures. Profiles of average relative fragment abundances for each treatment are shown in the Supporting Information (Fig. S1), which outline the effects of diet on the presence and relative abundance of individual Neocallimastigales fragments.
Neocallimastigales diversity was expressed as fragment number (S), Margalef's richness (d) and Shannon index (H') (Table 1). Interestingly, there were no significant differences in diversity parameters either between rumen solid and liquid phases or between diets differing in concentrate/forage ratio. There was, however, a significant reduction in both fragment number and Margalef's species richness when diets were supplemented with 6% soya oil in both solid and liquid rumen phases (Table 1). Differences in Neocallimastigales assemblages between treatments were elucidated using permanova on the fingerprint data. There was no significant difference between either animals (P =0·184) or rumen solid and liquid phases (P =0·438), but concentrate/forage and soya oil inclusion had significant effects on Neocallimastigales assemblages (P =0·036 and P <0·001, respectively). When the rumen phases were analysed separately, concentrate/forage ratio did significantly affect Neocallimastigales assemblages in the liquid phase (P =0·029), but not in the solid phase (P =0·228). The addition of 6% soya oil, however, significantly affected Neocallimastigales assemblages in both the solid (P =0·001) and liquid (P =0·001) phases. Specifically, addition of 6% soya oil reduced Neocallimastigales diversity. CAP ordination was performed to visualize these differences for both phases separately (Fig. 2a,b) and shows that assemblages clearly separate on the basis of 6% soya oil addition to the diet. To a lesser extent, the increase of concentrate/forage ratio separated Neocallismastigales assemblages, in particular in diets where soya oil was absent.
Table 1. Diversity indices of Neocallimastigales in rumen solid and liquid phases of cattle offered diets differing in concentrate/forage ratio (50 : 50 vs 90 : 10), with or without 6% soya oil supplementation
S is the amount of fragments, d is Margalef's richness and H' is the Shannon index with e as log base. Values are mean ± SEM, n = 4.
50 : 50
37 ± 5
3·74 ± 0·54
2·36 ± 0·14
26 ± 3
2·65 ± 0·31
2·81 ± 0·12
90 : 10
32 ± 2
3·21 ± 0·18
2·23 ± 0·08
18 ± 4
1·77 ± 0·42
2·19 ± 0·20
50 : 50
26 ± 4
2·64 ± 0·41
2·59 ± 0·14
23 ± 4
2·28 ± 0·40
2·37 ± 0·22
90 : 10
32 ± 3
3·22 ± 0·30
2·96 ± 0·08
24 ± 4
2·51 ± 0·42
2·54 ± 0·08
C : F
The contribution of individual Neocallimastigales fragments to these differences was examined using simper analysis (Table 2). In both the liquid and solid phases, simper indicated that the fragment of 415 bp contributed to most of the dissimilarity when comparing cattle offered diets containing different concentrate/forage ratios. When comparing cattle fed diets with or without soya oil, fragment 393 bp accounted for most of the dissimilarity. The fragment of 415 bp was significantly (P =0·005) less abundant in the 90 : 10 concentrate/forage samples (219 ± 103 RFU) compared with the 50 : 50 concentrate/forage samples (848 ± 184 RFU), whereas the fragment of 393 bp was most abundant in the 0% soya oil samples (868 ± 270 RFU), but almost absent (21 ± 21 RFU) in any of the 6% soya oil samples (P =0·011).
Table 2. Similarity percentages analysis of Neocallimastigales fragments in base pairs (bp) contributing approx. 45% to the dissimilarity between concentrate/forage ratio (50 : 50 vs 90 : 10) and soya oil (0 vs 6%) treatments within the solid and liquid phases
Associations between rumen metabolites and Neocallimastigales assemblages
Relationships between Neocallimastigales assemblages and rumen metabolites (Petrie et al. 2010) in solid and liquid phases were tested using multiple linear regression models. In the solid phase, a significant proportion of the variability in Neocallimastigales assemblages was explained by ammonia (P <0·01), acetate (P <0·01) and propionate (P <0·01) (Table 3). A supporting two-dimensional ordination using RDA (Fig. 3a) explained 76·8% of the fitted variability, and 48·8% of the total variation (first two axes). In the solid phase, Neocallimastigales assemblages in animals fed a low concentrate/forage ratio (high-fibre) diet associated positively with propionate, whereas there was a negative association between assemblages in the presence of 6% soya oil with ammonia and acetate. Similarly, a significant proportion of the variability in Neocallimastigales assemblages was explained by ammonia (P <0·05), acetate (P <0·01), propionate (P <0·01) and butyrate (P <0·05) (Table 3). A two-dimensional RDA ordination of the liquid phase (Fig. 3b, first two axes explained 66·6% of the fitted variability and 32·0% of the total variation) suggests that Neocallimastigales assemblages in animals fed a diet without soya oil associated negatively with propionate, whereas those in animals supplemented with 6% soya oil positively associated with acetate, butyrate and ammonia.
Table 3. Marginal tests of variance explained (%) by single VFA's and ammonia with Neocallimastigales assemblages in the solid and liquid phases obtained from multiple linear regressions
Values are proportions of the variance explained with *P <0·05, **P <0·01.
Anaerobic rumen fungi have not been extensively studied using molecular community (fingerprinting) approaches. Using ARISA, 64 different fragments representing members of the Neocallimastigales were successfully amplified. The primer pair used has been shown to be specific to genus level for the order Neocallimastigales (Edwards et al. 2008), and rarefaction suggested that the experimental design used has revealed the majority of fragments associated with the Neocallimastigales in the bovines used. The number of fragments found in this study was broadly in line with Fliegerova et al. (2010) who found 84 ITS sequences in bovine manure by ARISA using the same pair of primers.
The main finding of this study was that the addition of soya oil strongly influenced Neocallimastigales assemblage structure by reducing species (fragment) number, particularly where the diet was low in fibre (90 : 10 concentrate/forage ratio), and thereby influenced anaerobic rumen fungal diversity as was supported by generally lower Margalef's diversity indices. The SIMPER analysis suggested that several members of the Neocallimastigales may be affected by the concentrate/forage and soya oil treatments more strongly than others. Indeed, Neocallimastigales associated with fragment of 415 bp was almost absent in the rumen at 90 : 10 concentrate/forage ratio, and the species associated with fragment of 393 bp was almost reduced at the 6% soya oil dietary supplementation. The fingerprinting data did not reveal any species identities, and the effects of the dietary alteration should be interpreted as representing Neocallimastigales assemblages. Soya oil probably has influenced rumen Neocallimastigales community structure either directly through inhibition of specific species, or indirectly by changing general microbial metabolism within the rumen, which will impact on Neocallimastigales activity. It is not possible to pinpoint which of these mechanisms determines Neocallimastigales assemblage structure, although it might be a combination of both. Soya oil typically is rich in polyunsaturated lipids, dominated by linoleic (~55%) and oleic (~25%) acids (Hammond et al. 2005) and is known to have differential effects on rumen microbial diversity (e.g. Galbraith et al. 1971; Matsumoto et al. 1991; Lillis et al. 2011). It has been shown to be inhibitory to several species of bacteria in bovine rumen (Yang et al. 2009), has been used as a defaunation agent to reduce protozoa abundance in the rumen of ovines (Broudiscou et al. 1990) and has been also shown to reduce fibre digestibility in ovines (Machmüller et al. 2000) and methane outputs from the rumen of bovines (Jordan et al. 2006).
Rumen fibre contents clearly had an influence on Neocallimastigales community structure, although to a lower extent than soya oil in the diet. Rumen fungi are known to be particularly associated with the digestion of fibre, although, interestingly, there was little difference between Neocallimastigales community structure in either solid or liquid phases. Even though samples were appropriately filtered to separate the fractions, solid samples may still have contained some fungi of the liquid phase. Rumen fungi are also found extensively as zoospores in the rumen liquid phase. However, those fungi which do not readily release zoospores, such as members of the polycentric taxa Orpinomyces, may be underrepresented in the liquid phase when separated from the solids. Nevertheless, the rumen fungi are particularly associated with plant material in the rumen, with solid material being extensively colonized during the thallus stage with rhizoid attachment to vegetative material (Gordon and Phillips 1998). Rumen fungi produce a wide range of polysaccharide degrading enzymes including cellulases, hemicellulases, pectinases and lignin solubilizing enzymes (Gordon and Phillips 1998), all produced extracellularly and either associated with the colonizing thalli or liberated more widely into the rumen liquid phase (Lowe et al. 1987; Williams and Orpin 1987).
Clearly, Neocallimastigales communities cannot be looked at in isolation, as they form part of a much wider rumen community comprising archaea, bacteria and protists, interacting with the bovine host. In the same group of experimental animals, Lillis et al. (2011) showed that the addition of soya oil markedly altered rumen bacterial community structures, but had little effect on rumen methanogen community structure, although methanogen abundance was reduced. Additionally, Petrie et al. (2010) found that the addition of soya oil reduced acetate/propionate ratios and the levels of ammonia within the rumen in the same animals. The effect of raising rumen propionate levels is known to reduce H2 levels in the rumen and fermentation products are known to influence the activity of rumen fungi. Joblin and Naylor (1993) found that H2, formate, lactate and ethanol were strong inhibitors of cellulose degradation by Neocallimastix frontalis in culture, whereas acetate was mildly inhibitory with butyrate and propionate having no effect.
Multivariate analysis was used to explore relationships between Neocallimastigales diversity and rumen fermentation metabolites, and it was likely that some metabolites influenced rumen fungal community structure. Neocallimastigales community structure in animals fed diets supplemented with soya oil appeared to be positively correlated with propionate. Propionate is known to be a sink for H2 in the rumen (Hobson and Stewart 1997). Lillis et al. (2011) hypothesized that the presence of soya oil resulted in the selection of bacterial species that favoured the production of propionate. This would reduce H2 concentrations in the rumen and would potentially favour rumen fungal activity possibly leading to increased fibre degradation. These findings are in contrast to results obtained by Machmüller et al. (2000) who found that the addition of dietary oils reduces fibre digestibility in lambs. Processes affected by dietary treatments within the rumen of bovines may different from those occurring in ovines, and requires further study to provide more ruminant-specific information. Nevertheless, regardless of the ruminant involved, altering fibre digestibility may have a downstream effect on substrate availability for other rumen microbial communities and animal nutrition and highlights the complex interdependencies within the rumen.
In conclusion, alteration of bovine diet by altering fibre content or including a soya oil addition to the diet did affect Neocallimastigales community structures, particularly by reducing Neocallimastigales diversity especially in the case of soya oil addition. In understanding the relationships between diet and methane emission in the bovine rumen, it is necessary to take into account all microbial components contributing to enteric fermentation. In an earlier paper, Lillis et al. (2011) linked changes in rumen bacterial community structure (rather than methanogens) to a reduction in methane emission in the presence of soya oil. This view could now be extended to include a role for the Neocallimastigales in regulating methane production in the rumen.
This study was supported by the Irish Department of Agriculture Fisheries and Food Research Stimulus Fund (Grant no. RSF 06 361).