Swine and poultry gut microbiomes

A complex and dense community of bacteria, archaea, fungi, protozoa and viruses inhabits the gut of the pig and the chicken. Indeed, the total number of microbial cells within the GIT of single-stomached animals, including man, exceeds that of the host cells by at least one order of magnitude⁽Reference Savage¹¹⁾. The GIT microbiome exhibits a gradient concentration, with numbers and diversity increasing from proximal to distal. For example, in pigs, the stomach and proximal small intestine contain relatively low numbers of bacteria (10³–10⁵ bacteria/g or ml of contents); with dominant bacteria being Lactobacillus spp. and Streptococcus spp.⁽Reference Jensen¹²⁾. In contrast, the distal small intestine harbours a more diverse and numerically greater (10⁸ bacteria/g or ml of contents) population of bacteria⁽Reference Gaskins, Lewis and Southern¹³⁾. In broiler chickens, analysis of 16S rDNA gene sequences revealed thirteen, eleven, fourteen, twelve, nine and fifty-one operational taxonomic units in the crop, gizzard, duodenum, jejunum, ileum and caecum, respectively⁽Reference Gong, Si and Forster¹⁴⁾. Radial distribution of microbes within specific GIT segment has also been described⁽Reference Gaskins, Lewis and Southern¹³⁾. The four micro-habitats include: the intestinal lumen; the unstirred mucus layer or layer that covers the mucosal epithelium; the deep mucus layer found in the crypts; and the surface of the intestinal epithelial cells. The diversity of bacterial populations within a particular micro-habitat in the GIT is influenced by factors such as digesta flow rate, pH, anoxic conditions, types of endogenous and dietary substrates, inhibitory factors such as bacteriocins and SCFA, and competitive advantage⁽Reference Gaskins, Lewis and Southern^13, Reference Pluske, Pethick and Hopwood¹⁵⁾. Given this scenario, it is likely that certain nutrients and their associated physico-chemical effects play a major role in maintaining the balance of the microflora in specific micro-habitats, and subsequently in determining whether a pathogenic bacterium proliferates.

A significant hindrance in studying the gut microbiome has been the inability to effectively identify and quantify microbial species, their metabolic endproducts, and mechanisms by which they affect host health⁽Reference Zoetendal, Collier and Koike^16, Reference Torok, Hughes and Mikkelsen¹⁷⁾. This is mainly because the bulk of available information is limited to cultivable microbiota. The fraction of micro-organisms that are cultivable remains low mainly because the growth requirements of most of the bacteria are unknown, but also because of the selectivity of the isolation media that are used, the stress imposed on bacteria by the cultivation procedures, the need for anoxic conditions and problems simulating the microbe–microbe and microbe–host interactions that occur naturally in the gut⁽Reference Zoetendal, Collier and Koike¹⁶⁾. Some of these issues have been overcome by the increased use of culture-independent, molecular-based techniques that use the sequence diversity of the 16S rDNA gene to explore the diversity and abundance of bacterial communities within the GIT of animals. The use of 16S rDNA sequencing has been applied to analyse the intestinal bacterial community in a number of species, including swine and poultry, and has shown that the majority of the bacterial species colonising the GIT have not been cultured and characterised⁽Reference Danzeisen, Kim and Isaacson^18, Reference Kim, Borewicz and White¹⁹⁾. These data provide crucial information on the community structure in the GIT, but they provide limited information on microbe–microbe and host–microbe interactions⁽Reference Danzeisen, Kim and Isaacson¹⁸⁾. Microbes in the GIT have evolved mechanisms to influence the intestinal environment for their own benefit, and potentially that of the host, by influencing epithelial host cell gene expression⁽Reference Gaskins, Lewis and Southern^13, Reference Zoetendal, Collier and Koike¹⁶⁾. Host–microbe interaction is currently a very active area of research and may help in identifying clusters of GIT bacteria that are consistently associated with better growth performance and health in animals raised in varied environments⁽Reference Torok, Hughes and Mikkelsen^17, Reference Stanley, Geier and Hughes²⁰⁾.

Concept of dietary approaches to gastrointestinal health

The primary functions of the GIT are to digest and absorb nutrients from the diet and to excrete waste products. As alluded to, the GIT also contain normal microbiota responsible for a plethora of functions including intestinal development and functionality (as evidenced by differences seen between gnotobiotic and conventional animals), nutrient digestion and absorption, mucus secretion, immune development and cytokine expression⁽Reference Gaskins, Lewis and Southern^13, Reference Klasing^21, Reference Niba, Beal and Kudi²²⁾. However, there are many specific bacterial pathogens that also inhabit the GIT, and they generally cause disease when the gut ecosystem is disturbed in some manner. For example, in the post-weaning period in pigs, numbers of pathogenic Escherichia coli proliferate to exceed those of other bacterial populations, resulting in clinical disease⁽Reference Fairbrother, Nadeau and Gyles²³⁾. Many factors influence the diversity and activity of the GIT microbiota, including the age of the animal and the environment it inhabits, antimicrobial agents (antibiotics and minerals such as Zn and Cu), dietary composition (for example, type and content of carbohydrates and protein), feed additives (for example, organic acids; FE), feeding methods (fermented liquid feeding), disease load, weaning, season, stress and genetics⁽Reference Gaskins, Lewis and Southern^13, Reference Pluske, Pethick and Hopwood^15, Reference Zoetendal, Collier and Koike^16, Reference Torok, Hughes and Mikkelsen¹⁷⁾. These factors, which usually interact, can make comparison of studies on gut microbiota a daunting task.

One question that generally arises in relation to the GIT microbiome, and particularly the area of gut ‘health’, is: what is ‘normal’ when referring to the health of the pig and chicken gut? Hillman⁽Reference Hillman²⁴⁾ suggested that emphasis should be placed on an ‘optimal’ GIT microbiota rather than a ‘normal’ microbiota being present, because it is very difficult to define what is ‘normal’ given the wide array of conditions that pigs and chickens are grown under. Producers strive to keep pigs and chickens free of infections (bacteria, viruses, parasites) and achieve the best utilisation of feed for muscle gain and egg production as possible. However, with at least 400 species of bacteria, with numbers as high as 10¹⁴ colony-forming units/g inhabiting the GIT⁽Reference Savage^11, Reference Mackie and White²⁵⁾, it is little wonder that perturbations sometimes occur to cause clinical disease and occasionally death⁽Reference Jensen^12, Reference de Lange, Pluske and Gong²⁶⁾. Specific enteric pathogens can cause enormous economic loss to pig and poultry enterprises; hence there is interest in being able to identify, quantify and track the different components of the microbiota (both pathogenic and non-pathogenic) to improve health and production. To check microbial perturbations the industry has long depended on dietary fortification with sub-therapeutic levels of antibiotics and antimicrobial growth promoters (AGP)⁽Reference de Lange, Pluske and Gong^26, Reference Heo, Opapeju and Pluske²⁷⁾. However, long-term use of AGP has been linked to the potential problem of increasing transferable resistance of bacteria to antimicrobial drugs⁽Reference Adjiri-Awere and Van Lunen²⁸⁾ and has been banned outright in some jurisdictions and increasingly under intense scrutiny in others.

Interest in alternatives to AGP has increased recently as there is a clear need to control enteric pathogens previously contained through the use of AGP in feeds. For example, in poultry there is greater risk of an outbreak of necrotic enteritis (NE) with the use of viscous grains (barley, wheat and rye)⁽Reference Riddell and Kong^29–Reference Jia, Slominski and Bruce³¹⁾. This has been associated with high digesta viscosity, decreased nutrient digestibility and prolonged intestinal transit time, thus favouring growth of Clostridiumperfringens in the upper gut⁽Reference Timbermont, Haesebrouck and Ducatelle³²⁾. In swine, viscous fibres have also been linked to exacerbation of post-weaning collibacillosis (for example, McDonald et al. ⁽Reference McDonald, Pethick and Pluske^33, Reference McDonald, Pethick and Mullan³⁴⁾, Hopwood et al. ⁽Reference Hopwood, Pethick and Pluske³⁵⁾ and Montagne et al. ⁽Reference Montagne, Cavaney and Hampson³⁶⁾) and swine dysentery (SD) (for example, Durmic et al. ⁽Reference Durmic, Pethick and Pluske^37, Reference Durmic, Pethick, Mullan and Cranwell³⁸⁾ and Pluske et al. ⁽Reference Pluske, Siba and Pethick^39, Reference Pluske, Durmic and Pethick⁴⁰⁾). The detrimental effects of soluble fibres in swine have been associated with increasing digesta viscosity, undigested nutrients in the GIT and endogenous secretions. Furthermore, an increased flow of ileal undigestible protein in the hindgut can result in proteolytic fermentation in the large intestine of pigs and the caecum of poultry that can negatively affect their performance and health⁽Reference Smulders, Veldman and Enting^41, Reference Cone, Jongbloed and Van Gelder⁴²⁾. Arguably, the use of AGP in the past markedly reduced the negative consequences of feeding such feedstuffs and as a result performance was maintained on diets that otherwise would be problematic. Indeed, earlier studies demonstrated that the growth-depressing effect of viscous rye was ameliorated by antibiotic supplementation⁽Reference Marquardt, Ward and Misir^43, Reference Antoniou and Marquardt⁴⁴⁾. Furthermore, Smulders et al. ⁽Reference Smulders, Veldman and Enting⁴¹⁾ showed that antibiotics were more effective in diets with low digestible protein content v. in diets with high digestible protein content. Consequently, there is considerable interest in identifying alternative nutritional strategies for managing the microbiota of pigs and chickens when fed antibiotic-free diets.

Influence of feed enzymes on the gastrointestinal microbiota

Enzymes are biological catalysts that speed up reactions and act on specific substrates or reactants. Arguably, characteristics of FE are designed and set by the producing organism given its ecology (substrate). However, the efficacy in animal feed applications depends very much on a completely different set of criteria which are based on the mechanism of action in animal nutrition⁽Reference Bedford and Schulze⁵⁾. The link between FE and the GIT microbiome can perhaps be better understood from two points of view (Fig. 1). On one hand are the effects of substrates by themselves on the digesta biochemical characteristics and GIT physiology and on the other hand is the modification of these effects by FE to the extent that the substrates are degraded or modified in the GIT. This view is important because the rationale for the development and application of FE is to target certain dietary substrates (such as phytate and NSP) that are not degraded sufficiently or indeed not at all by the endogenous digestive enzyme array in the GIT. For example, an accepted paradigm in broiler chickens is that increased intestinal viscosity due to soluble NSP is the most important mechanism for poor growth and feed utilisation⁽Reference Adeola and Cowieson^3, Reference Bedford and Schulze^5, Reference Choct, Hughes and Wang⁴⁵⁾. However, the viscous nature of soluble NSP is a digesta biochemical characteristic that is accompanied by or precipitates many small intestine physiological responses such as increased digesta transit times⁽Reference Almirall, Francesch and Perez-Vendrell⁴⁶⁾, intestinal mass and turnover rates of mucosa cells⁽Reference Gee, Lee-Finglas and Wortley^47–Reference Daenicke, Bottcher and Jeroch⁴⁹⁾, mucin and carbohydrate expression of goblet cell glycoconjugates⁽Reference Fernandez, Sharma and Hinton⁵⁰⁾ and undigested constituents⁽Reference Choct, Hughes and Wang^45, Reference Jørgensen, Zhao and Eggum⁵¹⁾. These responses have been shown to increase small intestine microbial activity, composition and size⁽Reference Choct, Hughes and Wang^45, Reference Hübener, Vahjen and Simon⁵²⁾. Conversely, supplemental NSP-degrading carbohydrases (NSPases, such as xylanase and β-glucanase) reduce digesta viscosity by partial or complete hydrolysis of soluble NSP, improving animal performance and nutrient utilisation but also effecting changes to the composition and metabolic potential of bacterial populations⁽Reference Choct, Hughes and Wang^45, Reference Hübener, Vahjen and Simon^52, Reference Torok, Ophel-Keller and Loo⁵³⁾.

Fig. 1 Link between feed enzymes and gut microbiota in poultry and swine. NSPases, NSP-degrading carbohydrases. (A colour version of this figure can be found online at http://www.journals.cambridge.org/nrr)

The literature evidence suggests that diet composition is a strong modulator of the composition of the gut microbiota. For example, the relative contributions of host genetics and diet in shaping the gut microbiota was investigated using wild-type mice and ApoA-I knockout counterparts (genetically modified for impaired glucose tolerance) fed different diets using DNA fingerprinting and bar-coded pyrosequencing of 16S rRNA genes⁽Reference Zhang, Zhang and Wang⁵⁴⁾. These investigations revealed that 57 and 12 % variation of gut microbiota in mice was accounted for by the diet and genetic background, respectively, suggesting the primacy of diet in determining the composition of the gut microbiota. Within the context of interaction between FE and GIT microbiota, the view (Fig. 1) that FE act on specific components of feed ingredients in most cases explains the role of FE in modulating the gut microbiota. However, it is also plausible that some enzymes such as alkaline phosphatase can act directly on the microbiota by dephosphorylating the outer membrane⁽Reference Chen, Malo and Moss^55, Reference Lallés⁵⁶⁾, but as far as we know alkaline phosphatases are not a major FE and there is a dearth of data demonstrating their effects on gut microbiota. In view of the foregoing, it has been suggested that FE might influence the intestinal microbiota through two main mechanisms: reducing the undigested substrates; and creating (in situ) short-chain oligosaccharides from cell wall NSP with potential prebiotic effects. These two mechanisms and the more recent findings will now be discussed.

Reducing undigested substrates

The diversity of the microbiome in a gut section reflects in part the types of nutrient substrates in those sections. Bacteria in the GIT derive most of their carbon and energy from luminal compounds (dietary and/or endogenous) which are either resistant to attack by digestive fluids or absorbed so slowly by the host that bacteria can successfully compete for them. Indeed, it has been suggested that the performance improvement attributes of AGP are due to factors such as reduced competition for nutrients in the small intestine, reduced local inflammation due to control of pathogens, and reduced size of intestine⁽Reference Visek^57, Reference Niewold⁵⁸⁾. Since bacterial species differ in their substrate preferences and growth requirements, the chemical composition and structure of the digesta largely determine the species distribution of the bacterial community in the GIT. Consequently, bacterial community structure and metabolic function are very much dependent upon digesta biochemical conditions, as a result of feed composition and attendant host physiological responses such as endogenous secretions. For example, Apajalahti et al. ⁽Reference Apajalahti, Kettunnen and Graham⁵⁹⁾ analysed bacteria chromosomal DNA content of guanine and cytosine (G+C) in caecal digesta samples from broiler chickens fed either maize- or wheat-based diets. Compared with wheat, maize favoured low %G+C microbes (20–34 %) at the expense of the higher %G+C bacteria (55–59 %). In swine, Drew et al. ⁽Reference Drew, Van Kessel and Estrada⁶⁰⁾ reported changes in gut bacterial populations in pigs fed diets containing maize, barley or wheat as the main carbohydrate source, and correlated some changes with the fibre content of the diets. From the foregoing, it is clear that bacterial species differ in their substrate preferences and growth requirements; the chemical composition and structure of the digesta largely determine the species distribution of the bacterial community in the GIT. Consequently, it is important to understand that FE will mediate or modulate microbial status preset by the main dietary ingredients (target substrates).

When digestion is compromised, the flow of undigested/unabsorbed nutrients into the hindgut increases⁽Reference Choct, Hughes and Wang^45, Reference Jørgensen, Zhao and Eggum⁵¹⁾ and the microbial ecology can change dramatically⁽Reference Hübener, Vahjen and Simon⁵²⁾. Bird et al. ⁽Reference Bird, Vuaran and Brown⁶¹⁾ compared pigs fed highly digestible starch (low amylose) and low digestible starch (mixture of non-heated and heated high amylose). Low starch digestibility had a dramatic effect on intestinal physico-chemical and microbial characteristics. The study also reported increased caecum weight and colon length in pigs fed low digestible starch and this was linked to a large flow of digesta into the hindgut that stimulated gut mass growth. These findings were replicated in a more recent pig study (Regmi et al. ⁽Reference Regmi, Metzler-Zebeli and Gänzle⁶²⁾) that showed that slowly digestible starch resulted in poor feed efficiency concomitant with an increased flow of DM, starch and protein in the hindgut. Dietary fibre and in particular soluble NSP have been shown to have dramatic effects on the flow of undigested DM into the hindgut. For example, Jørgensen et al. ⁽Reference Jørgensen, Zhao and Eggum⁵¹⁾ showed that increasing dietary fibre from 6 to 27 % resulted in a 5- to 6-fold increase in the flow of digesta through the terminal ileum of pigs. In broiler chickens, the addition of soluble NSP (40 g/kg) isolated from wheat to a sorghum-based diet reduced the ileal digestibility of starch and protein by more than 35 %⁽Reference Choct, Hughes and Wang⁴⁵⁾. The implications of increased concentration of undigested substrates in the GIT are such that the microbiota proliferates and competes with the host for the available nutrients, and increase the risk of microbial perturbation if the balance is tipped for pathogen growth⁽Reference Pluske, Pethick and Hopwood¹⁵⁾. Furthermore, the host responds to increased undigested substrates by increasing the gut mass which carries a cost in terms of increased maintenance requirements which ultimately compromise efficiency of growth⁽Reference Ferrell^63, Reference Agyekum, Slominski and Nyachoti⁶⁴⁾. It is inevitable that the use of any additive that influences the digestibility of the diet will change the selection pressures on the resident microbiota which in turn will moderate the efficiency with which the host utilises its feed.

Carbohydrases and proteases

It has recently been reported that the beneficial effects of FE are inextricably linked to the amount of the undigested fat, protein and starch in the ileum⁽Reference Romero and Plumstead⁸¹⁾. For example, Romero et al. ⁽Reference Romero, Plumstead and Ravindran⁸²⁾ fed broiler chicken two types of diets, a simple maize–soya diet (maize-SBM) and a complex diet in which maize-SBM was fortified with dried distillers grains with solubles and rapeseed meal in line with the current trends in the industry to seek alternative ingredients to curb feed costs. The results (Fig. 2) showed that a blend of xylanase, amylase and protease improved ileal digestibility of starch, fat and protein and thus significantly improved the bird's ability to extract energy in two distinctly different diets in terms of substrate complexity. Such accelerated intestinal digestion and removal of what would otherwise be apparently undigested without FE must clearly limit the nutrients available for the microbes.

Fig. 2 Ileal digestibility (%) of starch (A), fat (B), protein (C) and their contribution (kJ/kg diet) to dietary energy (D) in maize–soya (maize-SBM) or maize–soya diets with 10 % dried distillers grains and 5 % rapeseed meal (mixed) fed to broiler chickens as influenced by enzyme combinations containing xylanase and amylase (XA), or xylanase, amylase and protease (XAP). (D) , Protein; , fat; starch. Xylanase was from Trichoderma reesei at 2000 units/kg feed; amylase was from Bacillus lichiniformis at 200 units/kg feed; protease was from B. subtilis at 4000 units/kg feed. ^a,b,cMean values with unlike superscript letters within a subgrouping were significantly different (P< 0·05). Adapted from Romero et al. ⁽Reference Romero, Plumstead and Ravindran⁸²⁾ (A colour version of this figure can be found online at http://www.journals.cambridge.org/nrr).

Indeed, Torok et al. ⁽Reference Torok, Ophel-Keller and Loo⁵³⁾ used the terminal restriction fragment length polymorphism method to examine changes in gut microbial communities in response to the addition of an NSP-degrading enzyme product containing β-glucanase, xylanase and protease activities in a barley-based diet. The enzyme product improved growth performance and energy utilisation compared with the control. Further correlation analysis of the dietary apparent metabolisable energy and microbial community composition within the ileum and caeca revealed distinct clusters associated with control and enzyme-supplemented birds. This is one of few studies that directly linked differences in the composition of the gut microbial community with improved performance, implying that the presence of specific beneficial and/or absence of specific detrimental bacterial species may have contributed to the improved performance in birds receiving FE. Furthermore, this study is a clear indication that molecular techniques coupled with statistical methods are capable of identifying desirable and undesirable clusters of organisms as far as good performance is concerned. Previous reports showed that supplemental carbohydrases reduced caecal counts of Campylobacter jejuni in broilers⁽Reference Fernandez, Sharma and Hinton⁵⁰⁾ and Brachyspira intermedia in the laying hen⁽Reference Hampson, Phillips and Pluske⁸³⁾. Clearly, FE might have dramatically altered the caecum ecology to the extent that for these specific bacteria growth was not favourable.

Short-chain oligosaccharides with potential prebiotic effects

Usage of NSP-degrading enzymes (NSPases; xylanase, β-glucanase, β-mannanase, α-galactosidase and pectinase) is common in poultry and swine feeds. As reviewed by Simon⁽Reference Simon⁸⁴⁾, NSPases targeting NSP may have several modes of action: partial hydrolysis of NSP, decrease in digesta viscosity, and rupturing of NSP-containing cell walls, thereby making the encapsulated nutrients available for digestion. Partial hydrolysis of soluble and insoluble NSP and rupturing of NSP-containing cell walls have been associated with reduced recovery of NSP when feedstuffs were incubated with NSPases⁽Reference Castanon, Flores and Pattersson^{85, 86)}. Furthermore, studies investigating the efficacy of NSPases reported not only improvement of nutrient digestibilities but also digestibility of NSP⁽Reference Adeola and Cowieson^3, Reference Slominski⁷⁾. Although there are some studies in which no effect of added NSPases on nutrients digestibility was recorded, for example, as reviewed by Adeola & Cowieson⁽Reference Adeola and Cowieson³⁾, if the added NSPases in the diet do achieve the intended purpose they do so by partially hydrolysing the substrate, i.e. dietary NSP. It is possible that part of the variation in the response of NSPases on NSP utilisation is related to interactions with the gut microbiota, which have a role in the utilisation of undigested substrates in the intestinal lumen.

In commercial human food production, short-chain oligosaccharides are extracted from natural sources as is the case for the galacto-oligosaccharides from soyabeans, but they can also be obtained biochemically⁽Reference Voragen⁸⁷⁾. Indeed, in the food industry, enzymic hydrolysis processes are applied for production of a whole array of oligosaccharides from plant cell wall polysaccharides⁽Reference Voragen^87, Reference Vázquez, Alonso and Domínguez⁸⁸⁾. As outlined by Voragen⁽Reference Voragen⁸⁷⁾, the concept for manufacturing oligosaccharides from suitable polysaccharides is simple: starting from a polysaccharides-rich feedstock followed by controlled hydrolysis of some of the heterocyclic ester bonds of the main chain backbone by an exogenous enzyme to give compounds of a lower degree of polymerisation (DP). For example, the enzymic hydrolysis process is widely applied in the commercial production of fructo-oligosaccharides from inulin⁽Reference Chesson, Stewart, Piva, Bach Knudsen and Lindberg⁸⁹⁾. Manufactured sources of oligosaccharides may allow the dose to be more precisely defined and controlled, but only a very limited range of oligosaccharide structures is presently available. Most are linear, with short chain length and containing only one or two different sugar units, and are specially manufactured for the ever-increasing human market.

Plant and plant co-products used by the feed industry contain almost an unlimited range of polysaccharides (Table 1). The use of specific NSPases, singly or in combination, against a range of polysaccharides can generate very large numbers of oligomer mixtures⁽Reference Chesson, Stewart, Piva, Bach Knudsen and Lindberg⁸⁹⁾. However, the NSP listed in Table 1 represent general classes in which there is commonality of overall structure but considerable variation in fine structure. For example, while the underlying structure of most arabinoxylans is similar, i.e. a β-1,4-linked backbone of d-xylose residues, in practice the variety is enormous due to differences in backbone size and in type and degree of substitutions from the backbone, all of which depend on the source of arabinoxylan⁽Reference Sunna and Antranikian⁹⁰⁾. For example, xylan hydrolysis products from maize are different from those produced from wheat in terms of the size, degree of substitution and in quantity⁽Reference Hespell, O'Bryan and Moniruzzaman^91, Reference Li, Azadi and Collins⁹²⁾. Even when the source of polysaccharide is constant (i.e. same grain), fine structure can vary with age, environmental conditions of growing and between varieties. For example, Austin et al. ⁽Reference Austin, Wiseman and Chesson⁹³⁾ showed that partial degradation of arabinoxylan from twelve UK-grown wheat samples with an endoxylanase (single cloned) resulted in mixtures of up to twelve oligosaccharides, with each differing in the presence and relative proportions of these oligomers. Such an observation suggested that even when the enzyme has only one activity, it is likely that polysaccharides will uniquely release a mixture of oligomers. Importantly, such divergence in products of enzyme activity will probably influence the in vivo response observed.

Table 1 Carbohydrate contents (g/kg DM) and major NSP in common feedstuffs*

* Adapted from Bach Knudsen⁽Reference Bach Knudsen^9, Reference Bach Knudsen¹⁰³⁾ and Meng⁽⁸⁶⁾.

† Includes lignin.

Regardless of whether a crude or purified enzyme product is used to hydrolyse NSP in typical diets for non-ruminants, it follows that short-chain oligosaccharides must be generated on every occasion when NSPases are used as feed additives. In this context, Apajalahti & Bedford⁽Reference Apajalahti and Bedford⁹⁴⁾ showed that the addition of xylanase in wheat-based broiler diets resulted in a 5-fold increase in concentrations of short-chain xylo-oligomers ( < 10 DP) in the caecum. There are apparently few studies where the in situ generation of oligosaccharides has been monitored in pigs and chickens fed diets containing NSPases. However, since oligosaccharides will be among the hydrolysis products released when NSPases degrade NSP, a question has been raised as to whether the presence of such products would influence GIT ecology⁽Reference Pluske, Pethick and Hopwood^15, Reference Chesson, Stewart, Piva, Bach Knudsen and Lindberg^{89, 95)}. This is because the resulting NSP hydrolysis products may modulate intestinal microflora⁽Reference Pluske, Pethick and Hopwood^15, Reference Chesson, Stewart, Piva, Bach Knudsen and Lindberg⁸⁹⁾. Numerous reports in pigs indicated that supplemental NSPases increased intestinal SCFA and this was associated with increased abundance and activity of specific bacteria communities such lactobacilli⁽Reference Diebold, Mosenthin and Piepho^96–Reference Kiarie, Nyachoti and Slominski⁹⁸⁾. In chickens, abundance of Enterobactreacea was reduced in caeca when birds were offered cereal-based diets with supplemental NSPases⁽Reference Rosin, Blank and Slominski^99, Reference Jòzefiak, Rutkowski and Kaczmarek¹⁰⁰⁾. Furthermore, Choct et al. ⁽Reference Choct, Hughes and Wang^45, Reference Choct, Hughes and Bedford¹⁰¹⁾ linked increased concentration of SCFA in the caecum of chickens fed diets with supplemental xylanase to increased flow of xylo-oligomers into the caecum. However, as the in situ production of short-chain oligosaccharides was not monitored in the aforementioned studies, it is thus rather difficult to associate observed microbial activity with the presence or absence of NSP hydrolysis products.

Short-chain xylo-oligosaccharides (arabinoxylooligosaccharides; AXOS) derived from the in vitro hydrolysis of wheat bran with an endoxylanase and fed to broiler chickens resulted in increased bifidobacteria populations in the caecum and improvements in feed conversion ratio (FCR) on both wheat- and maize-based diets⁽Reference Courtin, Broekaert and Swennen¹⁰²⁾. Interestingly, in the same study some birds were fed diets supplemented with xylanase instead of AXOS and the scale of the improvement in FCR achieved was equal for both strategies. This suggests that the production of fermentable oligomers may well be a large part of the total response to FE supplementation, although it was noted that the xylanase did not influence the intestinal flora to the same extent as that of the oligosaccharides, suggesting the xylanase, at the levels used, may not have released similar quantities of AXOS in situ. As the majority of arabinoxylans in feedstuffs are insoluble⁽Reference Bach Knudsen¹⁰³⁾, perhaps the xylanase used was not as effective in solubilising and cleaving insoluble NSP to effective oligomer sizes under limiting gut conditions (for example, digesta transit time). The effects on FCR were of similar magnitude in both the maize- and wheat-based diets, suggesting that caecal fermentation may be of significance. Xylo-oligosaccharides are not digestible or are very poorly digestible up to the ileum–caecum junction, but are fermented in part by microbiota in the caecum with concomitant production of SCFA and thus provision of extra energy to the birds. This might partly explain the reason why FCR improved with feeding AXOS in both maize- and wheat-based diets. It is also possible that AXOS improved the uptake of minerals in the caecum, as has been observed for inulin via lower pH caused by enhanced fermentation⁽Reference Van Loo¹⁰⁴⁾. The production of butyrate, a fuel for the colonocytes, might have also improved the absorptive capacity of the intestinal mucosa⁽Reference Guilloteau, Martin and Eeckhaut¹⁰⁵⁾. Furthermore, these data indicated that AXOS modulated gut health by increasing bifidobacteria counts in line with observations in other animal models⁽Reference Damen, Verspreet and Pollet¹⁰⁶⁾; a benefit of bifidogenic effects is overall better gut health and thus better FCR. This observation perhaps suggested decreased energetic costs associated with constitutive low-level inflammation caused by enteropathogens or caused by reduced passage of pro-inflammatory components of (commensal) bacterial origin such as lipopolysaccharides⁽Reference Anderson, McCracken and Aminov¹⁰⁷⁾. However, it remains to be proven that AXOS are released by xylanases in the GIT at levels sufficient to exert microbiota-modulating effects.

A potential limitation for in situ generation of AXOS via supplemental NSPases such as xylanase is the fact that xylanases are inhibited by the presence of AXOS (negative feedback), and the shorter the chain length the greater the inhibition, so in many cases the depolymerisation of xylan is self-limiting⁽Reference Lo and Pickersgill¹⁰⁸⁾, with the result that in situ production of AXOS in the gut might be limited. However, a recent study in rats indicated that the prebiotic potential and fermentation characteristics of cereal AXOS depend strongly on their structural properties⁽Reference Damen, Verspreet and Pollet¹⁰⁶⁾. The study reported that feeding a mixture of arabinoxylans of different DP (5–284) resulted in a larger increment in bifidobacteria in the caecum without increasing total bacteria; however, SCFA levels were greater for AXOS with 284 DP than for AXOS of 5 DP. Furthermore, evaluation of structurally different AXOS showed that smaller AXOS resulted in higher intestinal butyrate concentrations and a pronounced bifidogenic effect, whereas larger compounds mainly led to lower branched SCFA concentrations, an indication of lower proteolytic fermentation (i.e. better indices of gut health)⁽Reference Van Craeyveld, Swennen and Dornez¹⁰⁹⁾. These insights into the structure–activity relationship of AXOS suggest tremendous opportunity for developing and evolving FE products capable of in situ production of AXOS with optimised prebiotic and fermentation properties. Furthermore, perhaps future research could explore co-supplementation of NSPases/AXOS with probiotics to synchronise production of oligomers and utilisation.

Influence of feed enzymes on the gastrointestinal health–pathogen challenge: the case for a disease challenge model

Choct et al. ⁽Reference Choct, Sinlae and Al-Jassim¹¹⁰⁾ reported that broiler chickens fed a wheat-based diet with xylanase had a negligible number of C. perfringens compared with control birds. Nian et al. ⁽Reference Nian, Guo and Ru¹¹¹⁾ showed that the addition of xylanase to a wheat-based diet improved performance and was accompanied by a reduction in the number of coliforms and Salmonella in the ileum. In piglets, there was a significant reduction in the frequency and severity of diarrhoea in piglets fed diets supplemented with fibre-degrading enzymes⁽Reference Inborr and Ogle¹¹²⁾. However, the cause of diarrhoea was not determined, which presents problems in interpreting these data, as the diarrhoea seen after weaning can be osmotic⁽Reference Pluske, Pethick and Hopwood¹⁵⁾ in nature rather than being of bacterial origin which might be influenced by FE⁽⁹⁵⁾. The presence or absence of a pathogenic organism may not necessarily predict that disease will occur unless pathogen numbers proliferate to such an extent to overwhelm the normal microbial population in the GIT. However, the study of a specific bacterial disease, particularly if it causes economic loss, offers a means of assessing the usefulness of nutritional interventions or strategies on the survival of that particular pathogen in the GIT, and its subsequent effects on production, morbidity and mortality⁽Reference Pluske, Pethick and Hopwood¹⁵⁾.

There are a number of well-known enteric bacterial diseases that occur throughout the world, and each is relatively unique in that it generally occurs at different phases of pig and chicken growth and/or in different regions of the GIT⁽Reference Pluske, Pethick and Hopwood^15, Reference Timbermont, Haesebrouck and Ducatelle³²⁾. Enteric diseases are an important concern to the poultry and pork industries because of production losses, increased mortality, reduced welfare of animals and increased risk of contamination of poultry and pork products for human consumption. Numerous reports have used experimental enteric pathogen challenge models to examine whether a particular feed additive or dietary strategy is effective in controlling pathogens under study⁽Reference Jia, Slominski and Bruce^{31, 95)}. A major advantage of using such an in vivo model is that the impact of the particular additive can be assessed in a context of an infectious pathogenic agent being part of the GIT ecosystem. The following sections deal with four enteric diseases that afflict chickens and pigs worldwide and report examples of investigations exploring how FE can modulate the expression of the pathogen and accompanying disease that can occur.

Post-weaning colibacillosis in piglets

Post-weaning colibacillosis is a disease largely afflicting the small intestine and is caused by certain enterotoxigenic strains of E. coli (ETEC). Despite a rapid digesta flow through the small intestine⁽Reference Turlington, Allee and Nelssen¹¹³⁾, the pathogenic E. coli possess fimbriae, or pili, that attach to the enterocytes lining the small-intestinal villi or to the mucus covering the villi⁽Reference Fairbrother, Nadeau and Gyles²³⁾. This attachment physically prevents the bacteria from being flushed through to the large intestine. E. coli fimbriae attach to glycoprotein receptors expressed in the brush border of cells lining the intestinal villi. The most common E. coli associated with causing post-weaning colibacillosis is K88, renamed as F4⁽Reference Fairbrother, Nadeau and Gyles²³⁾. After colonising the small intestine, ETEC provoke hypersecretory diarrhoea through the release of specific enterotoxins (heat-labile toxins) that cause secretion of ions and water into the lumen. Post-weaning colibacillosis is a major cause of mortality and morbidity worldwide⁽Reference Fairbrother, Nadeau and Gyles²³⁾. Colonisation of the small intestine and diarrhoea usually lasts between 4 and 14 d, with the strains being spread between animals primarily by the faecal–oral route, and also by aerosols⁽Reference Fairbrother, Nadeau and Gyles²³⁾. However, it is important to note that despite ETEC being identified as the primary infectious agent in this disease, there is abundant evidence to suggest that a plethora of other factors are necessary for post-weaning colibacillosis to occur⁽Reference Fairbrother, Nadeau and Gyles²³⁾. Given the economic importance of post-weaning colibacillosis to pig production, diets fed to newly weaned pigs generally contain antimicrobial agents (antibiotics or ZnO) to control post-weaning diarrhoea and reduce the negative impacts of this disease. Restriction on the use of antimicrobial agents in some parts of the world, however, has forced the pig industry to examine other ways of controlling this disease, with ‘nutrition’ being a prime and obvious area of investigation.

Thus, Kiarie et al. ⁽Reference Kiarie, Slominski and Krause^114, Reference Kiarie, Slominski and Nyachoti¹¹⁵⁾ showed that when certain feedstuffs were incubated with a mixture of enzymes (xylanase, cellulases and mannanase) a wide range of sugars was released (Table 2). It is interesting to note that mannose was released during the hydrolysis of soyabean meal and wheat middlings; mannose is a well-described immune stimulant⁽Reference Jensen, Patterson and Yoon¹¹⁶⁾. This indicated that when these ingredients were subjected to enzyme hydrolysis, potential exists for active prebiotic release. The potential of such enzyme hydrolysis products in influencing the secretory diarrhoea caused by ETEC K88 was assessed in an in situ model of secretory diarrhoea⁽Reference Nabuurs, Hoogendoorn and van Zijderveld¹¹⁷⁾. In this context, the enzyme hydrolysis products were infused into small-intestinal segments prepared in live anaesthetised pigs and experimentally infected with ETEC and fluid passage through the segments measured. The segments that were infused with soyabean, wheat or flaxseed hydrolysis products had greater fluid absorption than control segments that were infused with an isotonic saline solution. What this means is that the hydrolysis products allowed the intestinal segments to retain more fluid and mitigate the negative effects of the ETEC infection. The fact that soyabean meal and wheat middlings had the greatest effects suggests that mannose mediated these effects, thus corroborating previous evidence that products derived from partial enzymic hydrolysis of wheat arabinoxylans⁽Reference Alam, Sarker and Molla¹¹⁸⁾ and fermented soyabeans⁽Reference Kiers, Meijer and Nout^119–Reference Kiers, Nout and Rombouts¹²¹⁾ enhanced fluid and electrolyte balance in models of intestinal secretory diarrhoea.

Table 2 Monomeric sugar composition (mg/g) of enzyme hydrolysis products from common feedstuffs*

nd, Not detected.

* Adapted from Kiarie et al. ⁽Reference Kiarie, Slominski and Krause^114, Reference Kiarie, Slominski and Nyachoti¹¹⁵⁾.

The enzyme hydrolysis products were further tested in piglets fed either a control diet (devoid of ingredients used to generate the hydrolysis products) or a treatment diet with composite hydrolysis products from wheat middlings, soyabean meal, rapeseed meal and flaxseed at 5 g/kg⁽Reference Kiarie, Slominski and Krause^122, Reference Kiarie, Slominski and Krause¹²³⁾. Piglets received these diets for 9 d during which basic performance metrics were measured. After this initial period, all pigs were orally challenged with ETEC and monitored for another 48 h. During the pre-challenge period piglets receiving diets with hydrolysis products ate more (299 v. 269 g/d) and grew faster (252 v. 207 g/d) than control piglets. Furthermore, during the challenge period, piglets fed hydrolysis products consumed more feed, suggesting the ability of enzyme hydrolysis products to attenuate ETEC enteritis, the hallmark of anorexia during infections (Fig. 3(A)). This peculiarity is explained by the lower incidence of diarrhoea (Fig. 3(B)) observed for the pigs fed the hydrolysis products. Interestingly, the piglets fed the enzyme hydrolysis products had lower stomach pH (2·62 v. 3·86) than the control piglets. Maintaining low gastric pH is important in the control of enteric pathogens such as ETEC which are transmitted within the herd via oral–faecal cycling⁽Reference Fairbrother, Nadeau and Gyles²³⁾. The enzyme hydrolysis products mediated this effect by eliciting overall high fermentation as measured by increased organic acid production and in particular lactic acid concentration. Foregoing research indicates that enzyme hydrolysis products of common piglet feedstuffs were beneficial in minimising the negative effects of ETEC infection, affording the piglet a healthier gut as indicated by better appetite in the presence of active disease. It is relevant that viscous fibres have been linked with the exacerbation of colibacillosis in some studies⁽Reference Hopwood, Pethick and Pluske^35, Reference Montagne, Cavaney and Hampson³⁶⁾. It appears that there is tremendous potential in developing FE for degrading NSP to generate short-chain oligosaccharides capable of controlling enteric infections such as ETEC-secretory diarrhoea.

Fig. 3 Acute feed intake of piglets fed enzyme hydrolysis products () or a control diet () and challenged with enterotoxigenic Escherichia coli (ETEC) (A) and incidence and severity of diarrhoea (faecal scores) in piglets fed enzyme hydrolysis products () or a control diet () upon challenge with ETEC (B). Values are means, with standard errors represented by vertical bars. * Mean value was significantly different from that of the control (P< 0·05). Adapted from Kiarie et al. ⁽Reference Kiarie, Slominski and Krause^122, Reference Kiarie, Slominski and Krause¹²³⁾.

A core microbiome at a functional level

Reduction of undigested substrates in the small intestine and provision of fermentable NSP hydrolysis products are perhaps the probable means by which FE can affect gut microbiota. This is more so in instances where anti-nutrients such as phytic acids and viscous ingredients attenuate digestion such that significant quantities of starch and/or protein enter the large intestine, stimulating the activity of putrefactive bacteria and pre-disposing the animal to intestinal disorders. Clearly, under such circumstances the use of FE will improve small intestine digestion and as a result limit substrate availability in the large intestine, effectively mitigating any potential GIT microbial dysfunction⁽Reference Jia, Slominski and Bruce^31, Reference Durmic, Pethick, Mullan and Cranwell^38, Reference Inborr and Ogle¹¹²⁾. However, it is still unclear whether, on feeding FE, the shifts in microbial profiles in the hindgut have a greater impact on animal health and nutrition than shifts in ileal profiles. Indeed, in some cases there may be significant changes in hindgut or ileal microbial profiles with little or no associated response in performance, indicating that the two are not always linked⁽Reference Parker, Oviedo-Rondon and Clack¹³⁸⁾. Nevertheless, it is clear that the response of the microflora to FE probably depends on the initial GIT microbial status, which in turn depends on the form and digestibility of the diet and the extent of the microfloral challenge it evokes. Many of the factors governing the extent of FE responses are most probably similar to those influencing responses to in-feed antimicrobials. For example, the response to antimicrobials in the diet depends to a large degree on the conditions under which the animals are raised⁽Reference Gaskins, Lewis and Southern^13, Reference Niewold⁵⁸⁾. It is likely that any positive responses to FE will be dictated not only by the status of the microflora in the GIT, but also by environmental, management and dietary factors such as cereal type and quality, and processing⁽Reference Adeola and Cowieson^3, Reference Bedford and Cowieson¹³⁹⁾.

By fair estimation most pigs and chickens are fed mostly non-viscous maize-based diets; furthermore, quite a great deal of advances have been made in housing equipment and bio-security. Under such production systems it might be difficult for a nutritionist to economically justify an extra dose of FE to cater for gut health. However, a key metric for profitability in swine and poultry production is growth performance, which in turn is dependent on variability in animal health and efficient utilisation of feed. At times this variability is driven by genetics and it is therefore likely that responses to FE may also depend upon the genotype of the animal. For example, Garcia et al. ⁽Reference Garcia, Gomez and Mignon-Grasteau¹⁴⁰⁾ observed that xylanase use in birds selected over five generations for divergent digestion efficiency attenuated bile acid deconjugation (a measure of deleterious microbiota in the small intestines) in the less but not in the more efficient birds. But even so, when similar genotypes are fed the same diets variability in performance is expected and is indeed a major challenge for production systems employing all-in and all-out flock or herd flow management.

FE have a role in increasing uniformity of the flock or herd and thus improving profitability. However, the application of FE to manage poor-performing animals in a herd or flock would perhaps be more precise if such variability impinges on gut microbiota. Indeed, there is growing evidence linking dietary nutrient utilisation, growth performance and GIT microbiota composition⁽Reference Torok, Hughes and Mikkelsen^17, Reference Torok, Ophel-Keller and Loo^53, Reference Aphalajalati, Geier and Hughes¹⁴¹⁾. Classical work by Turnbugh et al. ⁽Reference Turnbaugh, Hamady and Yatsunenko¹⁴²⁾ demonstrated that a diversity of GIT microbiota assemblages can yield a core microbiome at a functional level, and that deviations from this core are associated with different physiological states (obese v. lean). This study corroborated earlier findings in mice by Turnbaugh et al. ⁽Reference Turnbaugh, Ley and Mahowald¹⁴³⁾ and collectively demonstrates that the microbiomes in obese mice and human individuals harbour a larger proportion of genes for digesting fat, protein and carbohydrates, which might make them better at extracting and storing energy from food. As the application of molecular microbiology techniques increases in animal nutrition research, a potential strategy can be the establishment of robust and repeatable associations between enrichment of particular strains or clusters of bacteria and animal performance. In this context, recent studies by Torok et al. ⁽Reference Torok, Hughes and Mikkelsen^17, Reference Torok, Ophel-Keller and Loo⁵³⁾ linking poorly performing and well-performing broilers to distinct GIT microbiome is a step in the right direction. It may be possible to evolve and develop FE to induce desirable changes in the gut microbiota for the enhancement of growth and uniformity of the production of commercial herds and flocks.

In studies involving microbial community analysis, the complexity of interpretation increases with the use of molecular techniques, all of which are subject to potential pitfalls of PCR and other limitations⁽Reference Marsh^144, Reference Highlander¹⁴⁵⁾. No method is ideal, and even with the most advanced PCR-based techniques it is possible that microbial groups representing a significant fraction of the total microbial community escape detection. Consequently, it is possible that a link between microbiota and performance can be found using one method but not confirmed using some other method, even if the trial setup is identical⁽Reference Aphalajalati, Geier and Hughes¹⁴¹⁾. Understanding the dynamics of the gut microbial community, or microbial balance, is necessary to establish or develop strategies to improve feed efficiency and growth rate, avoid intestinal diseases and proliferation of food-borne pathogens, and evolve efficacious FE systems. Undoubtedly, molecular microbiology techniques are an important strategy in such endeavours. The 16S rRNA gene sequences generated in various studies could be used as a basis for developing quantitative assays and thus allowing validation of the presence of performance-related gut bacteria; potentially, these assays could be used to monitor strategies to improve feed efficiency⁽Reference Torok, Hughes and Mikkelsen¹⁷⁾. Such advances will probably be slow but ultimately optimal and detrimental microbial populations should be easily quantified and strategies to avert disease and poor performance more easily formulated⁽Reference Aphalajalati, Geier and Hughes¹⁴¹⁾.