Modern agriculture operates within 2 conceptual spheres with oftentimes competing objectives: health (i.e., provision of nutrition) and economics (i.e., production, distribution, and consumption of goods and services). For the human health professional, interest in agriculture is primarily directed toward ensuring that agricultural practices and resulting products promote the health and welfare of human populations. From this standpoint, human health intersects with agriculture along at least 2 distinct lines (Figure 1). Our lives are sustained by consumption of foods that are produced almost exclusively by modern agricultural practices. Ideally, these foods are of high nutritional value and free of potentially deleterious compounds or agents (e.g., toxins, pathogens, parasites, chemicals). The act of food production takes place in, and inevitably influences, a natural environment that is inhabited by humans. We may thus be exposed to, and our health compromised by, pollutants (e.g., pesticide-laden farm runoff), infectious agents (e.g., antibiotic-resistant pathogens), and degraded resources (e.g., nutrient-depleted field) as a result of livestock and crop production. Minimizing these environmental effects clearly is a highly desirable outcome.
Because agriculture operates under serious economic constraints, the ideals of the health professional often must be balanced against other interests, most notably those of the producer and business owner. Therefore, a critical goal of both agricultural scientists and health professionals is reconciling the sometimes competing priorities of consumers, who desire foods at the least price possible to enhance their quality of life, and producers, who must maximize economic returns. Many of the problems surrounding the nutritional quality and safety of our foods, as well as the growth and development of livestock, ultimately are affected by interactions between animals and microbes. As such, solutions may be found through better understanding of the microbiology of livestock. The thesis of this paper is that recent scientific developments in the study of the human-microbe ecosystem can foster more healthful animal agriculture. Furthermore, we posit that modification of the microbial constituents of the animal microbiome could elicit greater social acceptance than would chemical or genetic modification of the animal itself.
A NEW ERA OF CULTURE-INDEPENDENT MICROBIAL ECOLOGY
Medical microbiology matured and flourished within a paradigm in which diseases were associated with a single species of pathogens. The research of Koch and Pasteur formalized the logic of establishing infectious disease causation through implementation of procedures centered about pure microbial culture using Koch’s Postulates (Fredericks and Relman, 1996; Kaufmann, 2005). However, methods attuned to the study of acute diseases caused by professional pathogens (e.g., Vibrio cholerae, Mycobacterium tuberculosis) may be much less effective in analyzing more complex conditions, such as those in which microbial species or entire communities are covariates with environmental and host factors in determining disease risk. Furthermore, comprehensive studies of the numerous and diverse microbial communities that normally inhabit healthy individuals are not amenable to one-species-at-a-time approaches, as exemplified by microbial culture; however, we fully acknowledge that many applications of clinical microbiology, such as antibiotic susceptibility testing and strain-typing, currently demand axenic cultures. Advances in molecular biology and environmental microbiology over the last quarter century have ushered in a new era of culture-independent microbiology that is capable of disentangling the tightly interwoven metabolic, genetic, and biochemical networks that characterize complex microbial communities in their natural environments. Culture-independent methodologies have been developed to analyze the structure and function of these microbial communities through identification and categorization of the biomolecules they produce. A common attribute of these approaches (Figure 2) is the use of high-throughput, culture-independent analyses of many target molecules through parallelized specimen interrogation. Many excellent reviews of the field can provide more complete discussion of specific theories and practices of culture-independent microbiology (e.g., Relman, 1993; Falk et al., 1998; Vaughan et al., 2000; Handelsman, 2004; Zoetendal et al., 2004; Bäckhed et al., 2005; Kelly et al., 2005; Frank and Feldman, 2007; Frank and Pace, 2008; Tringe and Hugenholtz, 2008).
Two fundamental questions of community ecology can be addressed through culture-independent microbiology. First and foremost for any study is the question of the kinds and frequencies of microbial taxa in a community—“Who is there?” This is most often answered through molecular-phylogenetic analysis of small-subunit (SSU)-rRNA gene sequences. The SSU-rRNA gene has become the gold standard for identification of microbes through phylogenetic sequence analysis because of its presence in all cellular organisms, phylogenetic robustness (i.e., resolution of species- and domain-level affiliations), and relative ease of isolation by PCR amplification with broadly specific, universal primers (Lane et al., 1985; Lane, 1991). More nuanced characterization of microbial communities—“What are the microbes doing?”—can be achieved by expanding the target molecules from single genes to whole genomes or their products or both (Figure 2). The choice of methodological approach largely depends on the precise functional question under consideration. Cataloging genes in a community can be accomplished through mixed-genome sequencing [i.e., metagenomics (Handelsman, 2004) or microbiomics (Gill et al., 2006)], whereas phenotypic patterns can be elucidated by focus on mRNA biosynthesis (i.e., transcriptomics), protein expression (i.e., proteomics), or metabolite production (i.e., metabolomics). For the purposes of this paper, the exact technology deployed is less important than the general rubric embodied by the various approaches, which we will refer to as metagenomics.
An overview of metagenomic workflows is presented in Figure 3. In general, the biomolecule of interest is purified from a specimen, which could contain heterogeneous communities of organisms, for instance, host and microbes. Ensembles of informational macromolecules (i.e., all DNA, RNA, or protein in a sample) are then sequenced in parallel using appropriate technological platforms, and small molecules are analyzed through technologies such as mass spectrometry and nucleic magnetic resonance imaging. Massively parallel sequencing platforms now have capacities to generate millions to billions of sequence reads in a single experiment, thus permitting deep analysis of communities (Mardis, 2008). The molecular sequences or structures are then compared with analogous, previously analyzed data stored in public (e.g., GenBank) or proprietary databases to identify and classify the phylogenies and functions of the biomolecules present in a specimen.
Phylogenetic analysis of SSU-rRNA sequences usually entails multiple-sequence alignment (DeSantis et al., 2006; Cole et al., 2009; Nawrocki et al., 2009), followed by inference of phylogenetic trees (Ludwig et al., 2004; Stamatakis et al., 2005; Pruesse et al., 2007; Yarza et al., 2008), although more heuristic approaches have been developed to analyze large sequence data sets (Wang et al., 2007). Identification of microorganisms by SSU-rRNA phylogenetics provides a basic framework through which to organize and interpret the results of other metagenomic technologies. Depending on the extent of relatedness between an uncultured microorganism (perhaps known only through the macromolecules assayed in a given experiment) and its cultured relatives, the physiology and role of the environmental microbe can be assigned more or less accurately. Traits that are common to a given clade are inferred to be present in uncultured members of the taxonomic group as well.
More detailed characterizations of microbial communities can be accomplished by analysis of genes or gene products (e.g., mRNA, protein, metabolites) other than rDNA, for instance the occurrence and types of antibiotic resistance genes. Functional classification of genomic sequences and products, by comparative sequence analysis, reveals the types and functions of genes present in a specimen. The resulting information describes the collective genetic content of the community, from which physiological and metabolic lifestyles can be inferred.
MICROBIAL ECOLOGY OF HUMAN AND ANIMAL ECOSYSTEMS
Under normal circumstances, humans and other animals are not colonized in utero by microorganisms. Parturition begins a remarkable process in which polymicrobial communities are rapidly and permanently established on all bodily surfaces exposed to the environment (Edwards and Parrett, 2002). Although the factors that guide microbial succession patterns have not been elucidated in full, nutrition clearly affects the kinds of microbes and population dynamics of gastrointestinal (GI)-tract colonization. Infants who are breastfed or formula-fed tend to have gut communities that differ in a variety of microbial taxa (Penders et al., 2005, 2006; Newburg and Walker, 2007). For example, Clostridium difficile and Escherichia coli were observed at significantly greater loads in formula-fed infants relative to breastfed infants (Penders et al., 2005, 2006). After weaning, strict anaerobes, such as clostridia and bacteroides, gradually gain dominance, and by 2 yr of age, the microbiota reaches an adult-like climax community (Palmer et al., 2007). In the absence of extrinsic factors, such as antibiotic treatment (Dethlefsen et al., 2008) or substantial changes in diet (Ley et al., 2006b), the kinds and relative abundances of predominant human-associated microbes remain fairly stable (Dethlefsen et al., 2008; Frank et al., 2010) due to as yet poorly defined homeostatic mechanisms operating between host, microbes, and the environment (Allison and Martiny, 2008). A primary focus of future work on the microbial community ecology of humans and animals will be directed toward elucidating these mechanisms with the practical objective of improving health by manipulation of microorganisms.
An often-cited estimate is that the human cells that constitute our bodies are outnumbered 10 to 1 by resident microbes (Savage, 1977). We are, in effect, embedded in a cloud of microorganisms that forms one facet of the human “supra-organism” (Figure 4A; Turnbaugh et al., 2007). The GI tract, especially the lower bowel, is home to the majority of these microorganisms, which can reach cell densities of 1011 cells/g in the lumen of the colon. Hundreds to thousands of bacterial species inhabit the intestinal tract of a healthy adult (Eckburg et al., 2005; Ley et al., 2006b; Frank et al., 2007; Peterson et al., 2008), and their livelihoods likely are linked to each other by mutually dependent properties (syntrophies). The bacterial biodiversity of the GI tract is limited to only a few phyla, mainly Firmicutes (low G+C gram-positives; e.g., Clostridia and Lactobacillus spp.) and Bacteroidetes, with lesser frequencies of Proteobacteria (e.g., Escherichia coli), Actinobacteria (high G+C gram-positives; e.g., Bifidobacteria spp.), and Fusobacteria. Archaea, primarily methanogenic species, also are present in the GI tract but at much less frequency than are observed in ruminants (Qin et al., 2010). Commensal Eucarya remain largely unstudied in all GI environments.
Recent work by the Metagenomics of the Human Intestinal Tract initiative indicates that the gut microbial community of a human adult encodes, on average, approximately 3 million genes, whereas the human genome itself encodes a relatively paltry 25,000 genes (Qin et al., 2010). Approximately 90% of these microbial genes were of bacterial origin, whereas archaeal and eucaryal genes were observed at frequencies of 9 and 1%, respectively. The overlap in bacterial genes between individual humans was estimated to be about 40% for the most prevalent kinds of genes in the GI microbiota; this overlap defines a pan-microbial metagenome (or core microbiome) that is hypothesized to encode the minimal set of genes required to sustain the community (Qin et al., 2010). If commensal microbial communities have coevolved with their hosts (Ley et al., 2006a), this core microbiome is likely to enhance the phenotypic complexity of our extended selves (e.g., human plus microbes) by providing functional capacities not implicit in our genomes.
In the normal state of health, microbial inhabitants of animal and human hosts probably participate in mutualistic (i.e., beneficial to host and microbe) or commensal (i.e., beneficial to host or microbe, neutral to other partner) interactions with the host or other microbes or both. Recent studies in humans and animal models have demonstrated myriad mutualistic relationships in which benefits accrue to both host and microbe (Figure 4A). Several excellent reviews of these functional interrelationships have been published (Hooper et al., 2002; Duerkop et al., 2009; Round and Mazmanian, 2009; Hooper and Macpherson, 2010), and research into this topic is expanding. The commensal microbiota provides important cues that guide development of the GI tract (Bry et al., 1996; Stappenbeck et al., 2002; Mazmanian et al., 2005) and helps maintain immune homeostasis within the gut-associated lymphoid tissues (Hooper and Macpherson, 2010). In addition, nutritional webs in which the by-products of microbial metabolism are used by the host are a well-established example of mutualism (Hooper et al., 2002). For instance, bacterial fermentation of polysaccharides that are indigestible by humans provides an important source of energy to enterocytes in the form of short-chain fatty acids (e.g., butyrate; Pryde et al., 2002; Scheppach and Weiler, 2004; Wong et al., 2006).
Disruptions in any of the varied and extensive beneficial relationships that intertwine a host with its microbiota may cause the health of the former to deteriorate, potentially to the point of clinical disease (Figure 4B). Indeed, recent applications of culture-independent microbiology to human and animal diseases have revealed several instances in which alterations in the makeup of the commensal microbiota are associated with disease (Frank and Pace, 2008). These cases typically present as whole-scale imbalances in microbial populations, “dysbiosis” (Tamboli et al., 2004), rather than loss or gain of particular species. Conditions such as inflammatory bowel disease (Frank et al., 2007; Sokol et al., 2008a), obesity (Ley et al., 2005, 2006b), nonbacterial prostatitis (Tanner et al., 1999; Krieger and Riley, 2002), antibiotic-associated diarrhea (Young and Schmidt, 2004), and bacterial vaginosis (Fredricks et al., 2005; Oakley et al., 2008) all may have etiologies that involve dysbiosis. A critical challenge in studies of human-associated microbiotas is to establish causation, specifically whether dysbiosis causes or is a consequence of more fundamental etiologies. Even if not the ultimate cause of disease, dysbiosis may have clinical relevance if it intensifies the extent or duration of disease. In such cases, remediation of the microbial imbalance may prove palliative, if not curative. Additional work is required in most cases to determine whether pathology results from either the loss of community members or the gain of normally absent or low-abundance microbes. In the latter instance, the resulting community as a whole may function as pathogen.
APPLICATIONS TO ANIMAL AGRICULTURE: CASE STUDIES OF HUMAN DYSBIOSIS AND DISEASE
This section presents 3 case studies of scientific developments in human metagenomic research. Each case study may inform the study of host–microbe interactions in animal agriculture.
Many human pathogens gain entry into their hosts by traversing mucosal surfaces. However, the commensal microorganisms that coat these surfaces can deter pathogen function through several mechanisms, including selective modulation of the immune system (Lysenko et al., 2005; Lai et al., 2009) and direct interference with pathogen physiology (Uehara et al., 2001; Xavier and Bassler, 2005). Thus, loss of protective capacities exerted by commensals against pathogens is hypothesized to increase a host’s susceptibility to infectious diseases. This type of experiment is carried out each day in human medicine as well as the agricultural industry through prophylactic administration of antibiotics and dietary changes. In humans and murine models, systemic antibiotic treatment elicits profound changes in GI biodiversity characterized by decreased species richness and more uneven distributions of the remaining microorganisms (Young and Schmidt, 2004; Flanagan et al., 2007; Dethlefsen et al., 2008; Antonopoulos et al., 2009; Garner et al., 2009). These are typical hallmarks of an unstable ecological community that is at risk of invasion by other organisms (Peterson et al., 1998; Allison and Martiny, 2008). Indeed, GI dysbiosis has been observed both in human antibiotic-associated diarrhea (including that associated with chronic Clostridium difficile infection; Young and Schmidt, 2004; Chang et al., 2008) and murine susceptibility to Salmonella enterica serovar Typhimurium infection (Garner et al., 2009). The capacity of the GI microbiota to recover its predisturbance state (i.e., its resiliency) upon cessation of antibiotic treatment remains an open question because conflicting results have been observed in humans (Dethlefsen et al., 2008) and mice (Antonopoulos et al., 2009).
For the animal agriculture industry, the take-home message is that practices that disturb the microbiota, such as antibiotic use or dietary manipulation, can lead to dysfunction and increased susceptibility of livestock to infectious diseases. For example, the practice of withdrawing feed from laying hens to induce molting and enter the next laying cycle more quickly increases susceptibility to Salmonella enterica serovar Enteridis infection, presumably through disturbance of intestinal microbial communities that abrogate competitive exclusion (Callaway et al., 2009). Interestingly, Callaway et al. (2009) demonstrated that an alfalfa-based diet, which also induces molting, not only was less detrimental than feed withdrawal to the normal cecal GI microbiota of laying hens, but also provided resistance to Salmonella colonization. Thus, judicious manipulation of diet (e.g., by provision of prebiotics, probiotics, or nutrient supplements) and other management factors (e.g., litter material) can promote and sustain more stable microbiotas and, thus, reduce the need for prophylactic antibiotics (Torok et al., 2009). Alternatively, specific antibiotics or treatment regimens could be evaluated using metagenomic techniques to select ones that improve or leave relatively unmodified the GI microbiota (Collier et al., 2003).
As is the case with herbivores and other omnivores, humans are limited in their native ability to digest and absorb complex, insoluble plant polysaccharides. Rather, microbial fermentation transforms these polysaccharides into soluble short-chain fatty acids (e.g., acetate, butyrate, propionate) that are more readily absorbed and metabolized by the host (Flint et al., 2008). Butyrate, for instance, functions as a critical source of energy for enterocytes (Pryde et al., 2002), an inducer of antimicrobial peptides (Raqib et al., 2006), and an antiinflammatory agent (Segain et al., 2000). In herbivores, microbe-driven metabolism has been estimated to provide 70% of dietary calories (Flint et al., 2008), whereas mice and humans derive approximately 30 and 10%, respectively, of their energy intake from microbial fermentation products (Hooper et al., 2002). Beyond functioning as a bioreactor, the intestine also contains enteroendocrine cells that secrete hormones to regulate appetite and energy balance (Field et al., 2010). Interplay between the gut microbiota and the enteroendocrine system likely provides an important, but relatively unexplored, feedback mechanism between microbial communities and host metabolism, perturbation of which could lead to pathologies (Bäckhed et al., 2004; Cani and Delzenne, 2009; Bäckhed and Crawford, 2010; Field et al., 2010).
Humans being overweight and obesity have become epidemic in the United States (Flegal et al., 2010; Ogden et al., 2010) and in resource-replete societies worldwide (Hill et al., 2008). Because of the central role played by the GI microbiota in our energy status, its implication in obesity is not particularly surprising. Interestingly though, both clinical studies of humans and murine experiments have reported significant phylum-level shifts in enteric commensal communities in obese individuals compared with lean controls (Ley et al., 2005, 2006b; Zhang et al., 2009; Ley, 2010). Moreover, alterations in gut microbiota may precede a gain in BW (Kalliomäki et al., 2008). In these studies, obesity is associated with greater abundance of a variety of Firmicutes species, relative to Bacteroidetes, in intestinal specimens (Ley et al., 2005, 2006b). Although not proving a causative effect, both fat- and carbohydrate-restricted diets reverse the trend between obesity and greater Firmicutes:Bacteroidetes ratios, indicating that the composition of the human GI microbiota can indeed be manipulated by changes in diet (Ley et al., 2006b).
That the microbiota itself can promote obesity was demonstrated in a study by Turnbaugh et al. (2008) in which transplantation of the cecal microbiota of diet-induced obese mice to lean gnotobiotic mice induced greater fat deposition in the recipients than did transplantation of the microbiota of lean mice. The recipients of obese-microbiotas and lean-microbiotas consumed equivalent amounts of food, indicating that metabolic properties of these microbial communities were directly responsible for the observed phenotypes. Indeed, metagenomic comparison of obese and lean individuals indicates that the microbiomes of the former are enriched in genes encoding a variety of metabolic capacities (Turnbaugh et al., 2006, 2008, 2009). These findings support the hypothesis (Turnbaugh et al., 2006) that the enteric microbiotas of obese individuals provide more efficient dietary energy inputs than do the microbiotas of lean individuals. Thus, from equivalent dietary inputs, an obesity-associated microbiota can shunt more calories into the host, compared with lean-associated communities.
Unlike the case of humans, rapid and efficient BW gain is a clear goal in rearing livestock. Although rumen health has long been tied to intestinal microbiology, research on human obesity indicates that dietary effects on the composition of the GI microbiota can strongly influence host metabolism in ruminants. Therefore, manipulation of the GI microbiota, through particular feeds (Metzler-Zebeli et al., 2010) or probiotic supplements, could be used to tune the physiology and metabolism of livestock. For instance, enhancing the energy-harvesting capacity and efficiency of microbiota in an animal could promote faster BW gain. In support of this hypothesis, metagenomic analysis of the porcine ileal microbiota published by Collier et al. (2003) indicated that the growth-promoting activity of tylosin may result from its selection for probiotic Lactobacillus spp. Similarly, metagenomic analyses of the bovine rumen microbiome are now beginning to elucidate how diverse microbes and their gene products function in concert to degrade plant biomass (Brulc et al., 2009). Deeper understanding of the metabolic capacities of the GI microbiota, and their regulation through host–microbe and environment–microbe interactions, holds great promise for application of dietary and genetic modifications designed to fully exploit and manipulate microbial fermentation.
On a more speculative note, increased understanding of the links between diet, microbial functions in the gut, and host metabolism might be exploited to improve the nutritional content of the animal products we consume. Fat content and quality (Bäckhed and Crawford, 2010) or vitamin content (Hooper et al., 2002) in these products could be improved through manipulation of microbial communities and their metagenomes.
Breeding a Better Microbiota
Although many studies have revealed microbial community imbalances in association with human disease, the timing of these dysbioses with respect to disease onset typically is unknown, unless experimentally induced (e.g., by antibiotic treatment). Particularly in complex milieus such as the gut, where microbes and host are linked through multiple dependencies, disentangling the roles of host, microbe, and environment can be difficult, especially in clinical studies. Work on human inflammatory bowel disease is a case in point. Several studies have reported phylum-level disturbances in the fecal or mucosal microbiota or both of individuals with Crohn’s disease (CD) or ulcerative colitis (UC; Krook et al., 1981; Van de Merwe et al., 1988; Giaffer et al., 1991; Seksik et al., 2003; Prindiville et al., 2004; Gophna et al., 2006; Manichanh et al., 2006; Frank et al., 2007, 2011; Sokol et al., 2009; Qin et al., 2010). In our work, substantially less relative abundances of many clostridial species (e.g., of clusters IV and XIVa) were observed in a subset of CD and UC patients, compared with other CD, UC, and control subjects (Frank et al., 2007). However, because of the difficulties in diagnosing and accurately determining the time of onset of CD and UC, samples usually are collected after disease is overt. Oftentimes, metagenomic analyses are performed years after onset of disease and treatment with steroids, antibiotics, or surgery (Frank et al., 2007). Thus, it is reasonable to question if dysbiosis is simply a result of prolonged disruption of the intestinal environment by inflammatory conditions and antimicrobial products. Our recent work, however, indicates that genetic loci associated with increased risks of CD (Barrett et al., 2008), specifically alleles of the genes NOD2 and ATG16L1, are significantly associated with changes in the intestinal microbiota (Frank et al., 2011). Several clostridial genera, including Faecalibacterium, along with the genera Escherichia were correlated with the genotypes of one or both of these genes, independently of disease phenotype. Thus, host genetics can affect the kinds and abundances of microbes that comprise the commensal microbiota. In the case of CD, the presence of risk susceptibility alleles may promote formation of a microbiota that is predisposed to pathogenesis, perhaps by disfavoring the growth of particular species that normally provide beneficial services to the host (Sokol et al., 2008b, 2009). Alternatively, ecological properties such as microbial community stability and diversity may be compromised in association with CD risk alleles and thus produce a microbiota that is less resilient to changes in intestinal physiology than the microbiotas of individuals who are not genetically susceptible to CD.
Genetic contributions to the development of intestinal microbiota could be exploited in animal agriculture to select for greater magnitude and breadth of microbial activities in service of their hosts. Either individual or collective functions of the intestinal microbiota can be thought of as extended phenotypes that could be modified by selective livestock breeding. Thus, host genetic loci that promote desired microbiological traits, such as more resilient intestinal communities or those with enhanced energy harvesting capacities, could be selected for, provided that measurable phenotypic outcomes could be established. For example, molding a bovine intestinal microbiota that is less susceptible to ruminal acidosis (Khafipour et al., 2009) or that produces less methane could be accomplished through breeding or dietary modification.
SUMMARY AND CONCLUSIONS
Technological advances in culture-independent microbiology have revolutionized the study of complex microbial communities. Dysbiosis, characterized by imbalances in the distribution of microbes in a community, increasingly is noted in chronic and acute disease states. Because human-associated and animal-associated microbiotas contribute a variety of beneficial services to their hosts, dysbiosis may be a significant contributing factor to disease pathophysiology. As such, interventions designed to improve health, such as antibiotic treatment or dietary manipulation, may have unintended side effects if treatment also promotes dysbiosis. By the same token, remediation of dysbiosis and promotion of the stability and resilience of commensal communities in the face of outside influences (e.g., pathogens) may provide novel routes for improving livestock health, through reduced dependence on prophylactic antibiotics and GH, as well as lessened environmental effect. Transfer of insights gained from studying the human microbiota to the agricultural animal sciences may, therefore, provide socially acceptable means of modifying livestock and food products for the purpose of improving human health.