To examine change in the gut community of rats fed high amylose maize starch (HAMS).
To examine change in the gut community of rats fed high amylose maize starch (HAMS).
Rats were fed AIN93G diets containing HAMS (5% resistant starch type 2) or alphacell (control). HAMS increased faecal short-chain fatty acid output, faecal propionate and total bacteria output but reduced gut pH and blood urea concentrations compared with rats ingesting the control diet. Feeding HAMS resulted in a gut community dominated by four phylotypes homologous with Ruminococcus bromii, Bacteroides uniformis and with yet to be cultivated organisms aligning into the Family Porphyromonadaceae. Enrichment of phylotypes aligning within the Bacteroidetes occurred primarily in the caecum, whereas those homologous with R. bromii were found primarily in the faeces. HAMS altered community structure such that the phylum Bacteroidetes represented the dominant gut lineage and progressively reduced faecal community phylotype richness over the duration of feeding.
Feeding HAMS resulted in a caecal and faecal community dominated by organisms that require ammonia as a primary nitrogen source. Gut ammonia derived from endogenous urea represents an important factor contributing to caecal community composition in addition to the ability to utilize HAMS. Increases in faecal propionate, rather than butyrate as is often observed following resistant starch feeding, reflected a gut community dominated by the Bacteroidetes.
Diet-mediated change is often viewed strictly in terms of available carbohydrate. Here, we have shown that ammonia derived from endogenous urea is an important factor contributing to gut community composition and structure in rats fed this substrate.
Resistant starches (RS) are starches or products of starch degradation, which are not digested or absorbed in the small intestine and pass into the colon where they become available as substrate for the resident microbiota (Asp et al. 1992). Resistant starches are currently classified into four categories according to their physical and chemical properties (Topping and Clifton 2001). RS type 1 (RS1) is physically inaccessible to the enzymes of the small intestine, RS type 2 (RS2) is native granular starch that is difficult to hydrate, RS type 3 (RS3) is retrograded amylose, and RS type 4 (RS4) is resistant due to chemical modification. The fermentation of RS, as with other carbohydrates, has been associated with various physiological benefits including increased faecal short-chain fatty acid output (SCFA; Topping and Clifton 2001), total butyrate output (McOrist et al. 2011), lower faecal pH (Phillips et al. 1995) and increased faecal wet weight and dry matter when compared to the faeces from individuals fed starch-free diets (Birkett et al. 1996). The fermentation of RS has also been shown to reduce ammonia levels, mediating the adverse effects arising from protein fermentation in the large intestine (Phillips et al. 1995; Birkett et al. 1996; Cummings et al. 1996; MacFarlane and MacFarlane 2012).
Starch can be utilized by a wide range of species isolated from the gut community (MacFarlane and Englyst 1986). Despite this, starches have had some very specific effects on the gut communities in humans and animals. For example, cultivation-based studies shown RS to be bifidogenic in rodents (Kleessen et al. 1997; Silvi et al. 1999; Wang et al. 2002), pigs (Brown et al. 1997) and humans (Bouhnik et al. 2004) and to stimulate the growth of additional cultivable gut species, including Lactobacillus, Eubacterium and Bacteroides (Kleessen et al. 1997; Wang et al. 2002; Le Blay et al. 2003). In contrast, molecular-based studies examining the effects of ingesting different forms of starch have consistently demonstrated substantive increases in taxa related to Ruminococcus bromii within the faecal communities of rodents (Abell et al. 2011), humans (Abell et al. 2008; Martinez et al. 2010; Walker et al. 2011), an artificial colon inoculated with human faeces (Kovatcheva-Datchary et al. 2009) and in the rumen of cattle-fed grain-based diets (Klieve et al. 2007). It is also known that not all humans ferment-resistant starch to the same degree (Birkett et al. 1996), and recently, the ability to do so was correlated with the presence of R. bromii-related phylotypes in their respective faecal communities (Walker et al. 2011). While the growth of R. bromii in response to various forms of starch appears to be a consistent and specific feature across different animals, additional taxa may also be stimulated. For example, in rats, RS2 increased the predominance of amplicons in denaturing gradient electrophoresis gels related to yet to be cultivated members of the Bacteroidetes (Abell et al. 2011), whereas in humans, RS2 or RS3 significantly increased phylotypes related to Eubacterium rectale (Martinez et al. 2010; Walker et al. 2011). Experiments using a model colon suggest that R. bromii represents the primary degrader of starch in human faeces with growth stimulation of additional taxa possibly resulting from a more terminal position in the trophic chain (Kovatcheva-Datchary et al. 2009).
While change in the faecal community of various animals in response to various forms of starch shares some common and quite selective effects, the basis for this is currently not well understood. Furthermore, different sources of RS are associated with different SCFA profiles (Ferguson et al. 2000; Martin et al. 2000; Conlon et al. 2012), suggesting that there may be differences in gut community composition depending on the type of RS present. Rodent models offer an important advantage in diet studies in that dietary composition can be completely controlled, allowing for a more precise determination on how different fermentable substrates may impact these microbiologically complex communities. Our aims were to obtain a more complete perspective on the totality of gut community change and to identify additional factors contributing to community diversity in response to the feeding RS2.
Sixteen 28–42-day-old male BioBreeding control rats (Animal Resource Division, Health Canada) were fed a control diet (AIN93G containing 5% alphacell; Reeves et al. 1993) for 2 weeks prior to the initiation of the study. Alphacell (cellulose) is poorly fermented (Davies et al. 1991) and is included in rodent diets as a faecal-bulking agent (Reeves et al. 1993). Rats were then randomly assigned to continue on the control diet (n = 8), or switched to a high amylose maize starch (AmyloGel 03003, Cargill, Hammond, IN) containing diet (AIN93G containing HAMS, sufficient to provide 5% w/w RS2, n = 8). The energy content of both diets was similar; compositions are listed in Table S1. Rats were housed in mesh-bottomed stainless steel cages, held on a 12-h light/dark cycle (21°C and 40% humidity), had free access to water (reverse osmosis treated to 95% purity) and feed. Environmental enrichment (glass balls, gnawing sticks and PVC pipe housing) was supplied.
Fresh faecal pellets were collected from each rat immediately prior to the start of the feeding trial (day 0) and then on days 7, 14, 21 and 28 under each diet. Samples for cultivation analyses were processed immediately and the remainder stored frozen at −20°C. This study was approved by the Health Canada Animal Care Committee.
Following completion of the feeding phase, rats were fed for an additional 14-day period. All faeces from this period were collected and stored frozen at −80°C. Total food consumption was also measured. Resistant starch (AOAC method 2002.02; Megazyme, Wicklow, Ireland) and dietary fibre (AOAC 992.16; Mongeau and Brassard 1993) were determined in each diet and corresponding faecal samples taken during the balance study. Faecal short-chain fatty acids and branch chain fatty acids were measured as previously described (SCFA, BCFA; Weaver et al. 1997). Daily faecal bacterial outputs under each diet over the balance phase were estimated by measuring the 16S rRNA gene content of dried faeces using a universal primer set (Table 1) and quantitative polymerase chain reaction (qPCR; see below). At necropsy, caecal contents were recovered from individual rats and stored frozen at −20°C.
|Application||Primers||Primer sequence||Annealing temperature (ºC)||Reference|
Abnous et al. (2009)
|Clone libraries|| |
|58°C||Abnous et al. (2009)|
|Total bacteria|| |
|60°C||Tannock et al. (2000)|
|Ruminococcus bromii|| |
|Unclassified Porphyromonadaceae|| |
|Lactobacillus acidophilus|| |
|Uncultured Lachnospiraceae|| |
Total cultivable anaerobes (cfu g−1 faeces wet weight) on day 0 were determined by dilution plating on L-10 medium (Caldwell and Bryant 1966). Faecal samples from rats fed HAMS (days 0, 14 and 28) were surveyed for bacteria able to grow on starch as a sole substrate using a modified L-10 anaerobic medium containing 1% (w/v) soluble starch (Difco, Detroit, MI, USA). All media were prepared under anaerobic conditions with the addition of 0·1% cysteine-HCl as a reductant, and cultivation was carried out in an anaerobic hood under an atmosphere of H2/CO2/N2 (10 : 20 : 70 v/v). Isolates were identified as previously described using a combination of denaturing gradient gel electrophoresis (DGGE) and sequencing of the full-length 16S rRNA gene (Brooks et al. 2009).
To reduce inter-rat faecal community variability, faecal pellets from each rat under each diet at each time point were pooled. This was also the case for subsamples of caecal contents. Faecal and caecal community DNA was isolated by grinding samples in liquid nitrogen and sterile sand then purified as previously described (Brooks et al. 2009). Genomic DNA from individual bacteria was prepared as previously described (Brooks et al. 2009) and stored at −20°C.
Bacterial tag-encoded FLX amplicon pyrosequencing of faecal DNA samples spanning the 28-day feeding trial was performed as previously described (Callaway et al. 2009) using the forward primer F44 and the reverse primer 519R (Table 1). Each DNA sample was subjected to a one-step PCR with a total of 30 cycles using a mixture of Hot Start and HotStar high fidelity Taq polymerases with amplicons originating and extending from the F44 primer. Tag-encoded FLX amplicon pyrosequencing analyses utilized Roche 454 FLX instrument using Titanium reagents and was performed at the Research and Testing Laboratory (Lubbock, TX, USA).
Sequence reads for each time point under each diet were initially analysed separately. Initial filtering and binning of DNA reads was carried out using ESPRIT (Sun et al. 2009). Phylotypes assigned using ESPRIT at each sampling point under each diet were screened for the presence of chimers using Chimeraslayer implemented through Mothur (Schloss et al. 2009), with suspect sequences being subsequently removed from each data set. Phylotypes identified under each diet and time point were classified using the Seq-match program of the RDP (Wang et al. 2007) and the resulting data used to construct heat plots illustrating the distribution of sequence reads through the phylogeny (King et al. 2005).
Phylotypes identified within each sample were aligned using clustalW as implemented using SeaView (Gouy et al. 2010) and Neighbour-joining trees generated (1000 iterations) using a Jukes Cantor correction. Sequences aligning within ≤2% sequence divergence were further binned together. This was repeated for each library until the core phylotypes were identified. Phylotypes from each sample and time point were then screened for best match against the RDP Seq-match program (Cole et al. 2009) and those having a similarity score of <0·80 or encompassing 3 or less clones were removed from each data set. Shared phylotypes across samples were identified based on the best match as determined by the RDP Seq-match program and confirmed by clustalW alignment.
Full-length 16S rRNA genes were amplified from the caecal community DNA samples, prepared as stated previously, using primers F44 and R1543, with 15 cycles of amplification, then cloned and sequenced as previously described (Brooks et al. 2009). Two separate libraries were constructed for each pooled caecal DNA sample under each diet. Full-length 16S rRNA clone sequences were aligned using the align function at GreenGenes (DeSantis et al. 2006). Chimeric sequences and putative chimers were eliminated using Bellerophon (Huber et al. 2004), and the remaining 16S rRNA sequences were aligned and edited using ARB (Ludwig et al. 2004). ARB-generated distance matrices were used to assign phylotypes (DOTUR; Schloss and Handelsman 2005) and to compare phylogenetic structure (∫-Libshuff; Schloss et al. 2004). Phylotype assignments were checked by generating Neighbour-joining trees using the ARB software with 1000 permutations using a Jukes Cantor correction. Clones found to align with <3% sequence divergence were binned together.
Total faecal bacteria and selected 16S rRNA target genes were quantified by qPCR (Table 1) as previously described (Brooks et al. 2009). Each determination was performed in triplicate on two independently derived DNA samples from each pooled faecal sample. Following amplification, melting temperature analysis of PCR products was performed to confirm the specificity of the reaction. Amplicons generated using phylotype-specific primer sets (Table 1) were cloned and sequenced to confirm the specificity of selected primer pairs.
The pH of caecal (day 42, n = 7) and faecal (day 28, n = 4) samples from individual rats fed either diet was determined. For caecal contents or faeces, samples from individual rats were each mixed into 5 ml of distilled water, allowed to sit for 5 min and then subjected to a low-speed centrifugation to remove particulate materials, prior to the determination of pH.
Individual serum samples (n = 6/diet) obtained at the termination of the balance study were analysed for blood urea nitrogen concentration using the urease–glutamate dehydrogenase method (Talke and Schubert 1965). Determinations were carried out using an ABX Pentra 400 Automated clinical chemistry analyser and ABX Pentra Urea CP test kits (Horiba Canada Inc., Burlington, ON, Canada).
Differences in phylogenetic structure among the 16S rRNA gene libraries were assessed using ∫-Libshuff (Schloss and Handelsman 2005). Matrices of phylotypes frequencies for each condition were used to create cluster diagrams using Sorensen distances (Beals 1984) and subsequently used to generate nonmetric multidimensional scaling analysis (NMS) diagrams using Statistica Software (Tulsa, OK, USA). Rarefaction curves, Choa1 estimates and Shannon indices were calculated using the on-line FastGroupII software using phylotype abundance data as the input (Yu et al. 2006). Differences in blood urea nitrogen concentrations, faecal and caecal pH, and SCFA were assessed by anova followed by Tukey's HSD test when warranted.
Body weight gain over the 14-day balance phase of the feeding trial did not differ significantly between diets (results not shown). Daily dry weight faecal outputs for each rat over the entire balance phase were lower in those fed HAMS (0·80 ± 0·08 g vs the control diet: 1·50 ± 0·08 g; P < 0·0001) as was daily dietary fibre excretion (0·06 ± 0·03 g HAMS vs 0·86 ± 0·04 g control; P < 0·05). The determined digestibility of the dietary HAMS was 99%. Daily faecal bacteria outputs over the balance phase were approximately 2-fold higher in rats fed HAMS (2·90 ± 0·03 × 1010 16S rRNA gene copies μg−1 faecal community DNA) compared with that found in the controls (1·01 ± 0·03 × 1010 16S rRNA gene copies μg−1 faecal community DNA). In rats fed HAMS, SCFA and BCFA outputs were significantly higher than found under the control diet (Table 2). Feeding HAMS increased the relative proportion of propionate by 2-fold, whereas relative levels of acetate decreased and those for butyrate remained similar to those found in rats fed the control diet (Table 2).
|Acetate||48·0 ± 3·4 (73·3)||79·7 ± 2·4 (57·4)|
|Propionic||6·9 ± 0·5 (10·5)||30·3 ± 1·7 (21·8)|
|Butyric||5·2 ± 1·0 (7·9)||14·8 ± 0·6 (10·7)|
|Iso-butyric||1·0 ± 0·2 (1·5)||2·2 ± 0·1 (1·6)|
|Iso-valeric||2·3 ± 0·2 (3·5)||3·9 ± 0·1 (2·8)|
|Valeric||1·6 ± 0·2 (2·4)||5·4 ± 0·2 (3·9)|
|Caproic||0·4 ± 0·1 (0·6)||1·9 ± 0·2 (1·4)|
|Heptanoic||0·0||0·7 ± 0·1 (0·5)|
|Total||65·5 ± 4·6||138·9 ± 3·8|
Differences in the pH of caecal contents and faeces were observed under each diet. Both the caecal contents and faeces of rats fed HAMS were more acidic (7·7 ± 0·4 and 7·0 ± 0·2, respectively) than those fed the control diet (8·3 ± 0·2 and 7·5 ± 0·1, respectively) and these differences were significant (P < 0·05). Blood urea nitrogen values in rats fed HAMS were lower (16·8 ± 0·7 mg dl−1) than found in rats fed the control diet (19·3 ± 0·5 mg dl−1), and this difference was also significant (P = 0·02).
The total number of readily cultivable bacteria able to grow using soluble starch as the sole carbohydrate source in the day-0 faeces of HAMS fed rats was threefold lower than the total cultivable community plated on standard L-10 medium (8·2 × 107 vs 2·3 × 108 cfu g−1 wet weight faeces). However, following day 14 and day 28, cultivable faecal bacteria able to grow using starch had increased in rats fed HAMS compared with day 0 (1·6 × 109 cfu g−1 wet weight faeces). Starch utilizers in the day-0 faecal community primarily consisted of two species, closely related to Bacteroides uniformis and Enterococcus faecium (Table 3). However, by day 28, the cultivable faecal community was dominated by a single strain of bifidobacteria.
|Isolate no.||Nearest relativea||S_abb||Day 0c||Day 14||Day 28|
|14-10||Uncultured bacterium R-9176; FJ878975 (genus Allobaculum)||0·99||1|
|28-9||Anaerostipes sp. AY833660||0·99||2|
|28-1||Bacteriodes uniformis AB050110||0·98||4||1|
|14-6||Uncultured bacterium R-8994; FJ880995 (genus Bifidobacterium)||0·99||1||3|
|14-2||Bifidobacterium animalis X70971||0·98||1||6|
|0-11||Enterococcus faecium FJ378893||0·99||3|
|0-5||Clostridium sporogenes CP000962||0·99||1|
|14-1||Lactobacillus acidophilus IMAU30066||0·99||3|
Temporal change in the whole faecal community in response to HAMS was determined through the analysis of 16S rRNA genes. Pyrosequencing of faecal community DNA under each diet over time (days 0, 7, 14, 21 and 28) yielded a total of 173 283 reads. Following an initial quality filtering, resulted in a total of 96 901 sequences with each sample containing on average 9690 reads. Collectively, these sequences resolved into a total of 206 phylotypes, the majority of which were homologous or closely related to previously reported rodent phylotypes. Rarefaction curves for each faecal sample had plateaued, indicating that sequence coverage was sufficient to encompass the majority of diversity contained within each sample (results not shown).
In rats fed the control diet, indices of faecal community richness and diversity remained constant over the duration of the feeding trial (Table 4). In contrast, those fed HAMS showed a marked reduction in both faecal community richness and diversity over time. A comparison of shared diversity (i.e. phylotype occurrence and abundance) as determined by NMS revealed that faecal samples from rats fed the control diet, along with the day-0 sample from rats fed HAMS, clustered together indicating that these samples were quite similar (Fig. 1). The distribution of 16S rRNA genes at the level of phylum and family (i.e. community structure) also remained relatively stable over time (Fig. 2). These faecal samples were dominated by the Firmicutes > Bacteroidetes > Actinobacteria (81 ± 9, 14 ± 10 and 5 ± 4%, respectively).
|Sample||Species richnessa||Community diversityb|
|Control day 0||179||4·4|
|Control day 7||146||3·6|
|Control day 14||169||3·8|
|Control day 21||161||3·9|
|Control day 28||159||4·0|
|HAMS day 0||183||4·1|
|HAMS day 7||151||3·2|
|HAMS day 14||115||3·0|
|HAMS day 21||115||3·2|
|HAMS day 28||87||2·5|
The day-7 to day-28 faecal samples from rats fed HAMS also clustered into a single group, distinct from the faecal samples from rats fed the control diet (Fig. 1). In addition, faecal samples from day 7 onwards were dominated by phylotypes aligning within the phylum Bacteroidetes (56 ± 8% of total phylotypes) and the family Ruminococcaceae (43 ± 9% of total phylotypes). Dominant taxa present in the faecal community of rats fed the control diet (phylum Actinobacteria, and families Erysipelotrichaceae, Lactobacillaceae and Lachnospiraceae) each represented less than 1% of the total phylotypes in the faecal community of rats fed HAMS (Fig. 2). Phylotypes exhibiting the most extensive changes in abundance in response to HAMS feeding are listed in Table 5. Two phylotypes (RScecalA12; R. bromii and RScecalA23; family Porphyromonadaceae) accounted for slightly more than half of the faecal community 16S rRNA gene content in samples spanning days 7–28.
|Phylotypea||RDP classification||% Faecal Abundanceb||% Cecal Abundancec|
|RScecalA161||Lactobacillus acidophilus AB680529||11·1 ± 6·0||<1·0||NPd||<1·0|
|AINcecalA127||Lachnospiraceae; uncultured bacterium FJ879729||6·5 ± 4·8||NP||31·9 ± 6·9||NP|
|RScecal A149||Uncultured bacterium FJ880497 (Genus Allobaculum)||11·9 ± 6·6||3·0 ± 1·3||<1·0||3·6 ± 0·9|
|RScecalA69||Bacterium ID4 AY960571 (Genus: Allobaculum)||5·1 ± 2·9||<1·0||NP||2·2 ± 0·5|
|RScecalA143||Ruminococcaceae; uncultured bacterium FJ880545||19·1 ± 8·1||7·5 ± 2·8||1·9 ± 1·8||2·2 ± 0·5|
|RScecalA165||Coriobacteriaceae: uncultured bacterium DQ014723||5·2 ± 4·0||<1·0||NP||<1·0|
|RScecalA21||Bacteroides uniformis AB510711||1·2 ± 0·7||5·9 ± 7·3||NP||5·0 ± 0·1|
|RScecalA23||Porphyromonadaceae; uncultured bacterium DQ777956||2·3 ± 1·0||31·9 ± 9·5||2·8 ± 3·9||24·7 ± 7·4|
|RScecalA35||Porphyromonadaceae; uncultured bacterium FJ880778||<1·0||13·5 ± 3·1||7·3 ± 9·4||7·9 ± 2·0|
|RScecalA12||Ruminococcus bromii EU266549||1·8 ± 0·9||22·1 ± 9·1||NP||5·8 ± 1·8|
To examine the impact of diet on the caecal communities, duplicate full-length 16S rRNA clone libraries were prepared from the caecal contents of rats fed either diet recovered following the termination of the balance phase. We chose to examine the caecal communities using full-length 16S rRNA genes, rather than pyrosequencing, to obtain higher quality DNA sequences to allow us to design specific primer sets targeting individual dominant phylotypes.
The distribution of caecal clones through the phylogeny in rats fed either diet is shown in Fig. 2. Community structure in the day-42 caecal sample under the control diet was different from the day-0 to day-28 faecal communities. Approximately, 50% of the clones aligned within the family Lachnospiraceae, although much of the diversity within this taxon was encompassed by a single phylotype (AINcecalA127: Table 5). In terms of community structure, dominant lineages present in the faecal community (Coriobacteriaceae, Lactobacillaceae and Ruminococcaceae) occurred at a much lower abundance in the caecal community. For example, a phylotype aligning with the Lactobacillaceae (RScecalA161) was very abundant in the control faecal community, but not the corresponding caecal community (Table 5). Despite the obvious differences in the abundance of various familial lineages and individual taxa, ∫-Libshuff analysis indicated no significant difference in the phylogenetic structure between the day-28 faecal and day-42 caecal communities of rats fed the control diet. Furthermore, community richness and evenness were also similar (Table 4).
In rats fed HAMS phylotypes aligning within the phylum Bacteroidetes dominated the day-42 caecal community (48% of total clones). A single phylotype accounted for 26% of the entire caecal community 16S rRNA gene pool (RScecalA23), homologous with the dominant phylotype found in the day-7 to day-28 HAMS fed faecal samples (Table 5). Phylotypes homologous with R. bromii formed a smaller portion of the caecal community than found in the faeces (5·8 vs 22·1%; Table 4). The caecal community in rats fed HAMS was much richer and more diverse than the day-28 faecal community (Table 4) and also differed in terms of phylogenetic structure as determined by ∫-Libshuff analysis (P < 0·0043).
Changes in the relative abundance of specific phylotypes based on 16S rRNA gene abundance were confirmed using qPCR and targeted oligonucleotide primer sets (Table 1). Dominant phylotypes aligning within the Lactobacillaceae (Lactobacillus acidophilus; RScecalA161, Table 5) and Lachnospiraceae (AINcecalA127, Table 5) decreased in abundance under the HAMS diet compared with those fed the control diet (Fig. 3, panels a and b, respectively), consistent with the observed change in 16S rRNA gene abundance over the course of the feeding trial (Table 5). Caecal 16S rRNA gene copy numbers for Lact. acidophilus (Fig. 3, panel a) were similar to estimates based on their abundance in the 16S rRNA caecal clone libraries (Table 5), although qPCR determinations for the dominant Lachnospiraceae phylotype (Fig. 3, panel b) were approximately 3-fold lower than found in the caecal clone libraries (Table 5).
Phylotypes homologous with R. bromii (RScecalA12, Table 5) and an unclassified Porphyromonadaceae (RScecalA23, Table 5) increased in abundance in the faecal community in response to HAMS (Fig. 3, panels c and d, respectively), in agreement with changes observed in 16S rRNA gene abundance over the course of the feeding trial. However, gene copy numbers determined by qPCR for the latter target under both diets were approximately 3-fold higher in all of the faecal and caecal samples than indicated by their abundance as determined by 16S rRNA sequence analysis (Table 5). Bifidobacteria 16S rRNA gene abundance was slightly higher in the faeces of rats fed HAMS compared with those on the control diet, although their contribution to total gene abundance was <1·0% across all samples (Fig. 3, panel e), consistent with that observed through analysis of 16S rRNA gene abundance (0·2 ± 0·2 and 0·6 ± 0·3% of total reads for the control and HAMS faecal community, respectively).
Supplementation with HAMS altered both the gut fermentation and the gut bacterial community compared with rats ingesting the control diet. First, HAMS significantly increased total SCFA output and in particular propionate, as found in a previous rat study using HAMS (Abell et al. 2011). However, increased propionate outputs are not always associated with the feeding of this substrate to rats (Ferguson et al. 2000; Conlon et al. 2012). Second, HAMS increased faecal bacterial excretion in agreement with previous data on swine fed this substrate (Bird et al. 2007). Increases in both SCFA and bacterial output are consistent with the near complete fermentation of the added HAMS (99% fermented) in contrast to alphacell (control diet), which undergoes a more limited fermentation (Davies et al. 1991). Finally, community change in response to HAMS was evident in the readily cultivable faecal community, and in common with previous studies, resulted in the growth stimulation of bifidobacteria (Brown et al. 1997; Silvi et al. 1999; Wang et al. 2002; Bird et al. 2007). However, in relative terms, the observed increase in bifidobacteria represented a minor community change, as phylotypes aligning within this genus accounted for less than 1% of the faecal community under either diet.
Community analysis based on 16S rRNA sequencing revealed that HAMS induced very substantive increases in the abundance of a limited number of phylotypes, with the result that the majority of faecal community diversity was encompassed by only two familial lineages (Porphyromonaceae and Ruminococcaceae). Dominant familial lineages present in the control faecal community (Lactobacillaceae, Lachnospiraceae, Coriobacteriaceae and Erysipelotrichaceae) were each reduced to levels of <1% of the total phylotypes (Fig. 2). This change was quite rapid, occurring within the initial 7 days of feeding. While the faecal community composition was stable from day 7 onwards (Fig. 2), faecal community richness progressively decreased over the duration of feeding (Table 4) so that by day 28, the phylogenetic structure of the faecal and caecal communities of rats fed HAMS were significantly different. Collectively, these changes resulted in a substantive distortion to the faecal community structure, with the contribution of the phylum Bacteroidetes exceeding that of the Firmicutes. These observations differ from the situation in humans, where supplementation with RS2 or RS3 has been shown to alter the abundance of individual faecal phylotypes but have little impact on the overall community structure (Martinez et al. 2010; Walker et al. 2011). However, our findings are similar to a recent DGGE-based study examining the impact of HAMS on Sprague-Dawley rats, which showed increases in the abundance of amplicons related to R. bromii and an unclassified Bacteroidetes (Abell et al. 2011). The relative abundance of specific phylotypes in the faecal community was confirmed using quantitative PCR, and the results showed that our analysis of temporal change based on 16S rRNA gene content represents a reasonable assessment of the impacts of this substrate on the gut community.
Typically, increases in faecal butyrate, rather than propionate, are more commonly associated with the feeding of RS in both rodent (Levrat et al. 1991; Ferguson et al. 2000; Wang et al. 2002; Le Blay et al. 1999) and human studies (McOrist et al. 2011). However, the increase in faecal propionate was consistent with the observed changes to the gut bacterial community. First, phylotypes related to Eu. rectale have been shown to increase in abundance following feeding of RS in humans (Martinez et al. 2010; Walker et al. 2011). The Eu. rectale group (Family Lachnospiraceae; also referred to as cluster XIVa) are important contributors to butyrate production in the human gut (Barcenilla et al. 2000). We found that the abundance of this taxon was much reduced in the faeces of rats fed HAMS compared with those fed the control diet (>0·2% vs 9%, respectively on day 28), consistent with the lack of a butyrogenic response following the feeding of this substrate. Second, feeding HAMS resulted in caecal and faecal communities dominated by phylotypes aligning within the Phylum Bacteroidetes (44 and 51% of total phylotypes, respectively). In the Bacteroides, propionate, succinate and acetate represent the major fermentation products (Smith et al. 2006), although their relative concentrations can change depending on the availability of carbohydrate. Succinate accumulates under conditions where carbohydrate is readily available, whereas propionate derived via the decarboxylation of succinate accumulates when carbohydrate becomes limited (MacFarlane and MacFarlane 2003). The increased abundance of Bact. uniformis is consistent with a previous study where elevated succinate levels were observed in the caecum of rats fed HAMS (Morita et al. 1998). However, the dominant gut phylotypes aligned within the Family Porphyromonadaceae and not within the Bacteriodes. Unfortunately, neither of these was related to previously characterized species, although some species within this Family are known to produce propionate (Fournier et al. 2001; MacFarlane and MacFarlane 2012). Based strictly on abundance, we suspect that these yet to be cultivated species must also be contributing to faecal propionate output. We also know that carbohydrate was limited in the colon of the HAMS-fed rats, based on the increased concentrations of BCFAs and the fact that the substrate was fermented to near completion (Table 2). Under these conditions, given the very high content of Bacteroidetes in the gut, it is not unexpected that the fermentative end products would shift towards increased propionate production.
While increases in faecal propionate are not generally associated with feeding RS, there are rat studies that have also observed this following the feeding of HAMS (Abell et al. 2011), retrograded potato starch (Kleessen et al. 1997) and with high amylose wheat flour but not with HAMS (Conlon et al. 2012). It is often difficult to directly compare feeding trials due to differences in diet formulation, source and structure of the RS fed, and perhaps differences in the gut communities associated with rat lines utilized across different studies. At this point, the influence of these additional factors on butyrate vs propionate production during fermentation is unknown.
The second most abundant faecal phylotype in rats fed HAMS was homologous with R. bromii, a response shared with other animal gut communities fed starch (Klieve et al. 2007; Abell et al. 2008, 2011; Kovatcheva-Datchary et al. 2009; Martinez et al. 2010; Walker et al. 2011). Interestingly, increases in abundance were detected primarily in the faeces, in contrast to those aligning in the phylum Bacteroidetes that increased primarily in the caecum (Table 5). Differences in the primary location of growth among these dominant organisms likely reflect the dependence of R. bromii for the branch-chained fatty acid iso-butyrate for growth (Herbeck and Bryant 1974). Branch-chained fatty acids are produced by other gut species from the fermentation of branched amino acids, which occurs to varying degree along the colon (Smith and MacFarlane 1998).
The specificity of the gut community response to HAMS was somewhat surprising, given that starch can be utilized by a wide variety of species commonly found in the gut (MacFarlane and Englyst 1986). Clearly, additional factors beyond the ability to utilize starch must be contributing to this response. One obvious factor is the availability of fixed nitrogen. Both R. bromii and Bact. uniformis are dependent on ammonia as either the sole or primary nitrogen source (Herbeck and Bryant 1974; Smith et al. 2006) and we suspect that this may also be true for the yet to be cultivated porphyromonads. Gut ammonia can be derived from the bacterial mediated breakdown of endogenous urea excreted into the proximal gut (Fuller and Reeds 1998) or by the fermentation of dietary or endogenous proteins and peptides, which occurs principally under conditions where carbohydrate becomes limiting (Smith and MacFarlane 1998; MacFarlane and MacFarlane 2012). A caecal fermentation, where starch fermenters having the ability to use ammonia are preferentially enriched is consistent with previous research demonstrating increased urea flux into the rat caecum (Younes et al. 1995) and reduced caecal ammonia concentrations resulting from feeding RS (Levrat et al. 1991; Younes et al. 1995; Silvi et al. 1999). Consistent with this altered nitrogen cycling, we found that blood urea levels were significantly lower in rats fed HAMS, indicating that not only is the caecal bacterial community incorporating more ammonia derived from endogenous urea than those fed the control diet, but that endogenous urea represents an important factor contributing to caecal community composition.
Feeding HAMS reduced the abundance of many dominant faecal taxa (Fig. 3) and gradually reduced faecal community richness (Table 4). A good example of this effect was found in the family Lactobacillaceae, a dominant lineage in control faeces (~15% of total phylotypes) but occurring at a much lower abundance in the faeces rats fed HAMS (<1·0% of total phylotypes). This marked reduction in faecal abundance was unusual as lactobacilli isolated from rats are often amylolytic (Kleessen et al. 1997; Wang et al. 2002; Table 3) and that RS must be available in the colon because the bulk of R. bromii growth occurred postcaecum (Table 5). Initially, we suspected that reductions in the abundance of the lactobacilli, other faecal taxa, and the gradual loss in community richness might reflect a decrease in colonic protein turnover, restricting the supply of peptidyl-nitrogen required to sustain the growth of the more fastidious gut species like the lactobacilli (De Man et al. 1960). Increases in colonic acidity resulting from the fermentation of carbohydrate are known to mediate the fermentation of protein (Birkett et al. 1996; Smith and MacFarlane 1998) and the loss of phylotype richness was not observed in the caecum of HAMS fed rats (Table 4) where sufficient peptidyl-nitrogen would likely be available from the diet. However, HAMS feeding only resulted in a moderate decrease in faecal pH, and we know that protein fermentation was occurring in the colon because faecal BCFAs were all significantly higher in rats fed HAMS (Table 2). A more likely explanation is that the dominant colonic species, having an ample supply of fermentable substrate and ammonia, have simply overgrown the colon reducing available niche space for organisms having slower growth rates or more fastidious nutritional requirements.
This research was funded by Health Canada (SPJB), Agriculture and Agri-Food Canada, the Advanced Foods and Materials Network, General Mills Inc., Alberta Life Sciences Institute and the Ontario Ministry of Agriculture Food and Rural Affairs (MK).