Interactions and competition within the microbial community of the human colon: links between diet and health
Summary
The microbiota of the human intestinal tract play an important role in health, in particular by mediating many of the effects of diet upon gut health. Surveys of 16S rRNA sequence diversity in the human colon have emphasized the low proportion of sequences that match cultured bacterial species. This may reflect limited recent effort on cultivation rather than inherent unculturability, however, as anaerobic isolation methods can apparently recover a wide range of the diversity found. A combination of information from representative cultures, molecular tools for enumeration and tracking of bacterial metabolites offers the most powerful route to understanding the roles played by different groups of bacteria in the gut ecosystem. Progress is being made for example in defining key functional groups including primary colonizers of insoluble dietary substrates, and major contributors to metabolites such as butyrate that influence the health of the gut mucosa. There is increasing evidence that bacterial populations in the large intestine respond to changes in diet, in particular to the type and quantity of dietary carbohydrate. A general consequence of increased carbohydrate consumption is to reduce the pH of the gut lumen, which is likely to play a major role in determining bacterial metabolism and competition. Oligosaccharides used as dietary prebiotics must inevitably have complex effects upon the bacterial community that include non-target organisms and the consequences of metabolic cross-feeding and changes in the gut environment.
Introduction
The human large intestine is colonized by a dense and complex community composed largely of anaerobic bacteria, whose cell numbers can exceed 1011 per gram. The activities of these organisms have a major impact upon the nutrition and health of the host via the supply of nutrients, conversion of metabolites and interactions with host cells. Information on microbial diversity within this community has seen an explosion in recent years as a result of 16S rRNA-based analyses. Several recent reviews have discussed evolutionary and functional aspects of microbial diversity in the human large intestine (Dethlefsen et al., 2006; Flint, 2006; Ley et al., 2006a). The challenge now is to extend and exploit these new insights so as to provide a better understanding of the responsiveness, variability and resilience of the gut community, and its interactions with diet and with the human host. The energy sources that support the microbial community of the large intestine are dietary components that resist degradation in the upper intestinal tract, together with endogenous products such as mucin. Anaerobic metabolism by the microbial community in the colon produces short-chain fatty acids together with CO2, H2 and CH4 (Macfarlane and Gibson, 1997). These fermentation products have significant effects on the gut environment and on the host, as energy sources, regulators of gene expression and cell differentiation and anti-inflammatory agents; butyrate, for example, is considered to play a particularly important role as an energy source for colonocytes and in the maintenance of gut health (McIntyre et al., 1993; Pryde et al., 2002). This brief review focuses on the response of the bacterial community of the large intestine and its metabolic outputs to dietary change.
Cultivability of gut microorganisms
In soils and most aquatic environments it has been estimated that less than 0.5% of microbial diversity can be cultivated in the laboratory (Amann et al., 1995). In gut microbial ecosystems, non-culture-based surveys of 16S rRNA gene diversity have indicated that more than 75% of the phylotypes detected in the human large intestine (Suau et al., 1999; Hold et al., 2002; Eckburg et al., 2005) do not correspond closely to known cultured species, and the corresponding figures for the rumen, and for the pig, equine and rodent large intestine are similar, or higher. The gut is, however, a very different type of microbial ecosystem, as it is a nutrient-rich, open system with a constant temperature and continuous turnover. This means that, for survival, organisms in the gut lumen have to reproduce at a rate sufficient to avoid washout, or else have an ability to attach to or colonize host tissues. We might expect to find extensive nutritional interdependence, potentially leading to complex growth requirements and the possibility of obligate syntrophy (Schink, 1992). While some human colonic bacteria for example simply require acetate or branched chain fatty acids that are normally present in the gut environment for growth on carbohydrate substrates (Barcenilla et al., 2000) the detailed growth requirements for the majority of gut bacteria remain unknown. Although most bacteria in the large intestine are obligate anaerobes, classic anaerobic techniques (Bryant, 1972) have allowed the isolation of a range of highly oxygen-sensitive organisms. This is illustrated in Fig. 1 for recent isolates of butyrate-producing bacteria from human faeces that include Roseburia intestinalis (Duncan et al., 2006), which survives for less than 2 min when cells are exposed to air on an agar surface. Despite this sensitivity, however, a recent comparison of directly amplified 16S rRNA sequences with those from cultured strains indicates that the species diversity of Roseburia-related bacteria is now well represented by a small number of available isolates (Aminov et al., 2006). If this degree of culturability applies also to other groups of gut bacteria, then it seems probable that the small fraction of described species mainly reflects a lack of recent effort on strain isolation, rather than any inherent unculturability among gut bacteria (Duncan et al., 2007a). It is therefore unfortunate that the largest isolation studies (e.g. Finegold et al., 1983) were made before the use of 16S rRNA sequence information in phylogeny became routine.

Survival of four strains of butyrate-producing human gut bacteria following exposure to air. Strains tested were Eubacterium hallii L2-7, Roseburia intestinalis L1-82, Anaerostipes caccae L1-92 and Faecalibacterium prausnitzii A2-165 (Barcenilla et al., 2000). Dilutions were spread onto M2 agar plates in an anaerobic cabinet, exposed to air for varying time intervals, and then returned to the cabinet for incubation at 37°C (S.H. Duncan and H.J. Flint, unpubl. results).
Such exquisite oxygen sensitivity raises some interesting questions about how the initial colonization of the gut occurs. The conditions used in Fig. 1 were, however, quite extreme, as they involved exposure of single cells to 20% oxygen on an agar surface. Membership of microcolonies and mixed consortia, and association with biofilms on particles, are likely to confer significantly more protection in vivo. The fact that the populations of many strict anaerobes only become significant after weaning (Wang et al., 2004) suggests either that new acquisitions of gut bacteria can occur at that time, or that bacteria acquired in the immediate postnatal period can persist in low numbers until weaning.
Relating phylogeny to function in anaerobic metabolism
The dominant groups of human faecal and colonic bacteria, based on 16S rRNA sequence analyses, are low G + C content Gram-positives (Firmicutes) and Gram-negative Bacteroidetes (Suau et al., 1999; Hayashi et al., 2002; Hold et al., 2002; Eckburg et al., 2005). Seventy-two per cent of the 395 phylotypes detected by Eckburg and colleagues (2005) belonged to the clostridial group of Firmicute bacteria, most in clusters XIVa (also referred to as the Clostridium coccoides group) or IV (Clostridium leptum) (Collins et al., 1994) and the majority of these were either novel, or unrelated to species held in culture collections. Fluorescent in situ hybridization (FISH) also demonstrates significant populations of clostridial cluster IX bacteria (Walker et al., 2005) and of high G + C content Gram-positive Bifidobacterium and Atopobium that appear to be underestimated by clone library approaches (Harmsen et al., 2002; Flint, 2006; Mueller et al., 2006). It is clearly important to understand the roles of dominant, but little studied, groups in the gut community, and the analysis of available cultured isolates still offers the most comprehensive approach for relating the functional attributes of gut bacteria (as distinct from genomic fragments) to 16S rRNA sequences.
A few key metabolic characteristics can already be correlated with bacterial lineages defined by 16S rRNA sequencing. Butyrate production is distributed across many clostridial clusters (Pryde et al., 2002), but one particularly abundant group of butyrate-producers in human faeces, accounting for approximately 7% of total faecal bacteria, comprises relatives of Roseburia and Eubacterium rectale (Fig. 2) for which all available isolates show high levels of butyrate production together with net acetate utilization (Barcenilla et al., 2000; Pryde et al., 2002; Aminov et al., 2006). Another subgroup of the Clostridial cluster XIVa, related to Eubacterium hallii and Anaerostipes caccae, includes isolates that show the ability to convert acetate and D or L lactate into butyrate (Duncan et al., 2004a) (Fig. 2).

Phylogenetic tree based on 16S rRNA sequences of representative anaerobic bacteria belonging to the clostridial cluster XIVa, isolated from human faeces. The upper box includes lactate-utilizing relatives of Eubacterium hallii and Anaerostipes caccae (Duncan et al., 2004a). Members of the Roseburia/E. rectale group of butyrate-producing bacteria (Aminov et al., 2006) are shown in the lower box. Strains detected by the Rrec584 and Erec482 FISH probes (see Fig. 3) also are indicated.
16S rRNA-targeted FISH probes that recognize such functional groupings can provide important insights into their ecological roles in the complete microbial ecosystem. They have been used for example to study competition for carbohydrate substrates in fermentor systems (Duncan et al., 2003; Walker et al., 2005) (Fig. 3A), and to reveal the dependence of the Roseburia group, in particular, upon dietary carbohydrate intake in obese human volunteers in vivo (Duncan et al., 2007b) (Fig. 3B). Similarly, populations of the quercetin-degrading species Eubacterium ramulus have been shown to respond to the presence of quercetin in the diet in human subjects (Simmering et al., 2002). It is also possible to develop detection methods based on polymerase chain reaction (PCR) amplification for functionally important, diagnostic genes, such as butyryl CoA transferase (Louis and Flint, 2007) or APS reductase (Deplancke et al., 2000). Other functionally significant attributes that have been identified for clostridial cluster XIVa species include acetogenesis (Bernalier et al., 1996), utilization of aromatic compounds from the diet (Simmering et al., 2002), metabolism of linoleic acid (Devillard et al., 2007) and degradation of mucin (Hoskins, 1993) but the phylogenetic distribution of these attributes is not yet fully established.

A. Effect of pH on populations of major groups of human colonic bacteria, monitored by FISH, in mixed anaerobic continuous culture [results of the study by Walker and colleagues (2005)]. The detection limit (106 cells ml−1) has been subtracted from the log count shown on the vertical axis. Shifts in species composition were reflected in major changes in the concentrations of fermentation products in the fermentors (Walker et al., 2005). FISH probes used detected: most bacteria (Eub338); Bacteroides (Bac303); clostridial cluster XIVa (Erec482); Roseburia and Eubacterium rectale (Rrec584); Faecalibacterium (Fprau645); clostridial cluster IX (Prop853).B. Changes in dietary carbohydrate intake (g/10 per day), and consequent changes in faecal butyrate concentrations (mM) and bacterial populations (estimated by FISH), in a group of 19 obese human volunteers [results of the study by Duncan and colleagues (2007b)]. MC and LC refer to weight loss diets with a medium or low content of carbohydrate, respectively, that were consumed in turn for 4 weeks by each volunteer. HC refers to the initial weight maintenance (high carbohydrate) diet. % Ros refers to the Roseburia/E. rectale group (Rrec 584 probe),% Bac to the Bacteroides group (Bac303 probe) and % XIV-Ros to cluster XIVa bacteria other than the Roseburia group (Erec484 – Rrec584), expressed as a percentage of total bacterial populations estimated in faecal samples with the broad Eub338 probe.
Culture-independent approaches can be used in other ways to provide functional insights into as yet uncultured bacteria. It is apparent from Fig. 3B, for example, that faecal populations of cluster XIVa bacteria other than the Roseburia group were stimulated by a reduction in carbohydrate consumption by human volunteers. FISH probes can also reveal the localization and morphology of cells in situ, while information on metabolically active bacterial populations can be inferred from comparison of 16S rRNA- and DNA-based denaturing gradient gel electrophoresis (DGGE) (Zoetendal et al., 1998). Clone library analysis can be used to gain information on different gut compartments, or substrate-attached communities (McWilliam Leitch et al., 2007). Another important methodology, as yet little applied to gut ecosystems, is the use of stable isotope probing in conjunction with 16S rRNA analysis to reveal groups of bacteria that are active at a particular site in vivo, or that utilize a particular substrate in the mixed ecosystem (Egert et al., 2006). Metagenomics can potentially produce valuable information on the abundance of different classes of genes in gut communities, regardless of cultivability (Gill et al., 2006).
Insoluble substrates and microbial biofilms
Much dietary carbohydrate enters the large intestine in the form of insoluble fragments of plant material (fibre). The ability to colonize and degrade polymeric, insoluble substrates may be limited to a subset of specialized primary degraders. This appears to be the case for cellulose (Robert and Bernalier-Donadille, 2003) and mucin (Derrien et al., 2004) degradation by bacteria from the human large intestine, although bacteria associated with insoluble substrates may be particularly under-represented among cultured species. Recent in vitro studies examining the colonization of insoluble substrates also support the view that colonization of a given substrate can be highly species-specific. Thus four species (E. rectale, Ruminococcus bromii, belonging to the clostridial clusters XIVa and IV, respectively, and two Bifidobacterium spp.) accounted for 80% of 16S rRNA sequences recovered from faecal bacteria, from four volunteers, that adhered tightly to resistant starch (McWilliam Leitch et al., 2007). The reason for such specialization probably resides in the requirement for appropriate systems for substrate attachment, degradation and uptake. Cellulolytic strains of Ruminococcus flavefaciens from the rumen, for example, produce an elaborate cell surface-anchored cellulosome that is thought to play a key role in the breakdown of plant cell walls (Rincon et al., 2005). Related cellulose-degrading bacteria are known to occur in the human gut (Robert and Bernalier-Donadille, 2003).
Microbial consortia that involve primary degraders as well as secondary colonizers associated with insoluble substrates are likely to play an important role in gut microbial ecology. Mature biofilms involving co-aggregation between different species and extracellular matrix material are known to develop on inert surfaces such as tooth surfaces and catheters (Costerton et al., 2005). Surfaces provided by digesta particles, the mucin layer and the gut mucosa, differ in being transient, but may still allow biofilms to develop. Metabolic capabilities may differ between bacteria that are adherent and non-adherent to food residues (Macfarlane and Macfarlane, 2006). Microscopy has provided some evidence for the formation of surface-associated communities on mucosal surfaces in the colon, particularly in disease states such as inflammatory bowel disease (IBD) (Macfarlane et al., 2004; Swidsinski et al., 2005). Mechanisms involved in attachment, co-aggregation and quorum sensing that may be relevant to biofilm formation remain largely unexplored for dominant commensal gut bacteria. Such mechanisms have been investigated in pathogenic bacteria, where these may be important for survival in the outside environment or in alternative hosts as well as in the human gut.
Competition for carbohydrate substrates – predicting the impact of prebiotics
Changes in the type and quantity of non-digestible carbohydrates in the human diet influence both the metabolic products formed in the lower regions of the gastrointestinal (GI) tract and bacterial populations detected in faeces (Fig. 3B). Non-digestible dietary carbohydrates such as inulin and fructo-oligosaccharides are now widely used as prebiotics in order to manipulate the composition of the gut microbiota (Macfarlane et al., 2006). A range of other naturally occurring oligosaccharides, and also synthetic products, have selective effects in vitro (e.g. Manderson et al., 2005). Several studies have shown that inclusion of inulin as a dietary prebiotic for humans can increase the proportions of bifidobacteria in faeces (e.g. Kruse et al., 1999). Bifidobacteria are widely regarded as conferring benefits upon health because of their potential interactions with the immune system and with pathogens, and are seen as the main targets for prebiotics (Gibson and Roberfroid, 1995).
Because we know little about the substrate preferences of the majority of gut bacteria, and in view of the importance of cross-feeding in anaerobic communities (discussed below), it should not be surprising to find that prebiotics influence non-target populations within the gut community. Inulin, for example, has been shown to stimulate groups of bacteria other than bifidobacteria in vivo in animal models (Kleessen et al., 2001; Apalajahti et al., 2002). In in vitro gut simulations, two groups of Clostridium-related bacteria, and an added strain of Roseburia inulinivorans, were shown to be stimulated by inulin in a mixed faecal community (Duncan et al., 2003). As a butyrate-producer, R. inulinivorans may be regarded as potentially beneficial. In most cases, however, it is simply not known whether the other bacterial groups that are stimulated have a positive, or negative, impact on health. Some hypothetical responses are shown in Fig. 4 to illustrate how prebiotics may stimulate one or more unknown elements of the microbiota in addition to the targeted group whose stimulation is assumed to confer a health benefit. Figure 4 is a great oversimplification of a highly complex picture, however, because it is already clear that responses will differ at the level of individual strains and species, as well as groups. Moreover the nutritional versatility of certain gut bacteria in utilizing alternative substrates, including those of host origin (Sonnenburg et al., 2005; Scott et al., 2006) may be an important factor tending to reduce the influence of changes in carbohydrate supply upon bacterial populations.

Hypothetical responses in the populations (shown on an arbitrary vertical scale) of different groups of gut bacteria (numbered 1–15) to changes in dietary carbohydrate intake. An increase in the overall intake of fermentable, non-digestible carbohydrate (B, cf. A) will have effects on composition of the gut microbiota, in part through effects on the gut environment (see Fig. 3). Addition of prebiotics (C, D) will stimulate both target and non-target groups of microorganisms. These may include groups that respond to overall increases in carbohydrate intake, as well as ones that respond specifically to the prebiotic. For the sake of argument, group 4, for example, might represent Bifidobacterium spp.
Prebiotic effects are likely to be influenced by many features of the substrate, including solubility, the distribution of chain lengths, branching and substituents (Rossi et al., 2005). The complement and organization of degradative enzymes and transport systems appears to vary widely between different gut bacteria, and must play a key role in determining substrate preferences and competitive ability (Flint, 2004; Van der Meulen et al., 2006). Tests of the ability of isolated bacteria to utilize purified carbohydrates in vitro, however, can provide only a preliminary indication of substrate preferences in the mixed ecosystem in vivo, as they ignore the consequences of interspecies competition and cooperation. It is also to be expected that responses to prebiotics will depend on the dietary context and the gut environment, and will be influenced by variations in the species composition of the resident gut microbiota between individuals.
Metabolic cross-feeding
The breakdown of energy-rich complex carbohydrates creates opportunities not only for competition, but also for cooperation via metabolic cross-feeding (Fig. 5). Primary cell wall degraders, for example, may release a wider range of polysaccharides than they utilize, and these become available to bacteria that are less closely adherent to the substrate, or that are planktonic (Dehority, 1991). In addition, many gut bacteria have the capacity to utilize oligosaccharides, but not polysaccharides. In a recent study only eight of 55 bifidobacterial strains tested were able to degrade long-chain inulin, leading to the suggestion that their observed stimulation by dietary inulin in vivo is due mainly to the cross-feeding of products released by other inulin-degrading bacteria (Rossi et al., 2005). Meanwhile, strains of Bifidobacterium sp. that can degrade starch or fructooligosaccharides (FOS) can stimulate the growth of species in co-culture that cannot degrade these complex substrates (Belenguer et al., 2006; Falony et al., 2006).

The role of cross-feeding interactions in the conversion of non-digestible carbohydrates by the anaerobic microbial community of the large intestine (see text). SRB, sulfate-reducing bacteria.
Classic work on the rumen ecosystem revealed numerous types of cross-feeding interaction in which the products of one species are utilized, or provide growth requirements, for others, often influencing the energy metabolism of one or both partners (Wolin et al., 1997). These interactions include hydrogen transfer, and utilization of fermentation products such as lactate, succinate and branched chain fatty acids, and of partial breakdown products released from complex polymers. Reutilization of fermentation products is an equally important factor in the anaerobic community of the human large intestine. Butyrate-producing Roseburia and Faecalibacterium prausnitzii are net consumers of acetate (Barcenilla et al., 2000; Duncan et al., 2004b), which they require for optimal growth. Many of these bacteria are also hydrogen-producers, and in co-culture with an acetogen, hydrogen is consumed to produce acetate, which is in turn utilized by the butyrate-producer (Chassard and Bernalier-Donadille, 2006) (Fig. 6). Although acetogens show a lower affinity for hydrogen than methanogenic archaea or sulfate-reducing bacteria (Macfarlane and Gibson, 1997) they may be favoured by the slightly acidic conditions in the proximal colon (Bernalier et al., 1996). Sulfate reduction has potentially deleterious consequences for gut health via the formation of toxic sulfide (Roediger et al., 1997). It is not entirely clear to what extent sulfate reducers and methanogens can coexist, or are mutually exclusive, within the colonic community (Gibson et al., 1988; Dore et al., 1995). It is well established that the utilization of hydrogen can alter the metabolism of hydrogen- or formate-producing species in co-culture (Wolin et al., 1997; Samuel and Gordon, 2006). Interestingly, there are indications that the major routes of hydrogen disposal in a given individual may influence the competitive balance between other species of gut bacteria (Robert and Bernalier-Donadille, 2003).

Potential contribution of metabolic cross-feeding to butyrate formation in the human colon. Some interactions demonstrated in vitro in co-culture experiments:A. Interaction of a xylan-utilizing R. intestinalis strain with an acetogen (Chassard and Bernalier-Donadille, 2006).B. Two types of interaction between starch- or FOS-degrading B. adolescentis and strains of E. hallii or R. hominis that were unable to directly use these substrates (Belenguer et al., 2006).
The fermentation of carbohydrates by bifidobacteria is assumed to be a significant source of lactate, but lactate can be utilized by other species and generally does not accumulate in healthy subjects. Co-culture of a starch-degrading Bifidobacterium adolescentis strain with a lactate-utilizing E. hallii strain unable to use starch was shown to result in butyrate formation via cross-feeding (Belenguer et al., 2006). Other groups of human intestinal bacteria may use the acrylate pathway to convert lactate into propionate, described for cluster IX bacteria such as Megasphaera elsdenii. Stable isotopic tracer studies indicate that conversion of lactate both to butyrate and propionate can occur in the mixed gut community (Bourriaud et al., 2005; Morrison et al., 2006). Competition between different lactate-utilizers, which also include sulfate-reducing bacteria (MacFarlane and Gibson, 1997) may be an important factor in determining the balance of fermentation products.
Influence of the gut environment
The gut environment has the potential to greatly influence bacterial metabolism and competition. Factors include the concentrations of a wide range of growth inhibitors and growth factors of dietary, microbial or host origin, osmolarity, pH and turnover. A consequence of the fermentation of non-digestible carbohydrates in the proximal colon is to lower the lumenal pH (Bown et al., 1974) as a result of increasing concentrations of acidic fermentation products. A recent study examined pH-controlled continuous cultures supplied with a mixture of polysaccharide substrates (mainly potato starch) (Walker et al., 2005). A switch from pH 5.5 to pH 6.5 resulted in a marked shift from butyrate to acetate and propionate production; this was accompanied not only by increasing Bacteroides populations, but also by the virtual disappearance of Roseburia-related bacteria (Fig. 3A). In pure culture, Roseburia species grew equally well at pH 5.5 and 6.5, but the Bacteroides strains tested failed to grow at pH 5.5 (Walker et al., 2005). This implies that members of the Bacteroides group were able to outcompete butyrate-producing Roseburia at pH 6.5, and that the Roseburia group only became able to compete for soluble carbohydrate at the lower pH. These effects of pH are likely to be exerted in part through effects on bacterial susceptibility to inhibition by short-chain fatty acids. Many of the effects of dietary carbohydrate intake, including prebiotic consumption, may therefore be attributable to local lowering of the lumenal pH.
Interestingly, genetically obese (ob/ob) mice have been reported to show increased proportions of clostridial cluster XIVa, and decreased Bacteroidetes, by comparison with wild-type animals (Ley et al., 2005). Assuming that the increased dietary intake of the obese mice resulted in a reduced caecal or colonic pH, this pH change could be an important factor in the observed community shift. A recent report suggests that the proportion of Bacteroidetes 16S rRNA sequences, detected from PCR-amplified clone libraries, is also depressed in faeces from obese human subjects, but this proportion apparently increased gradually over a 52-week weight loss period. This would imply the involvement of some longer-term physiological mechanism rather than a short-term response to diet (Ley et al., 2006b). As noted earlier (Fig. 3B) a separate study on obese subjects, using FISH microscopy to enumerate faecal bacteria, detected short-term decreases in Roseburia and Bifidobacterium spp., but no change in Bacteroides spp., over a 4-week period of weight loss in response to reduced carbohydrate intake (Duncan et al., 2007b).
Population dynamics and individual variation
There is increasing evidence that shifts in the species composition of the gut microbiota occur in response to changes in the content of the diet. Such changes can be expected to result from differential effects of substrates on bacterial growth, but selective lysis under changing gut conditions may be at least of equal importance. We know rather little about the turnover of the dominant bacteria that colonize the human large intestine, but potential causes of cell lysis and death are numerous, including bacteriophage, host-derived antimicrobial compounds, and environmental stresses caused by pH, oxygen and host secretions such as bile. 16S rRNA-based FISH probing and clone library analyses detect both live and dead cells, as demonstrated recently by the use of fluorescence activated cell sorting (FACS) to recover FISH-labelled cells (Ben-Amor et al., 2005). While the detection of significant populations of a given bacterial group in faeces by FISH or PCR-based detection of 16S rRNA genes provides evidence for growth and replication somewhere in the gut, it provides little idea of the exact location or niche occupied.
These considerations suggest that the relative abundance of different bacterial strains, particularly within the lumen of the gut, is likely to be in a continual state of change in response to dietary intake and host physiology. Faecal samples provide some form of temporal and locational ‘averaging’ of the colonic microbiota, but their composition must also be expected to change constantly. There is evidence to suggest that the composition of an individual's colonic microbiota, as determined by 16S rRNA DGGE profiling, shows a degree of stability and differs consistently between individuals (Zoetendal et al., 1998; Green et al., 2006). On the other hand, bacterial strains that show very similar (> 99.5% identical) 16S rRNA sequences are regularly recovered from different individuals (e.g. Louis et al., 2004). This suggests that many of the dominant phylotypes of gut bacteria, as defined by 16S rRNA sequencing, may be common to different individuals. 16S rRNA sequencing does not, however, discriminate strain variation at a genomic level, and genome plasticity and clonal variation is likely to occur even following recruitment of a strain to an individual's gut microbiota. Thus it seems likely that dietary habits and life events contribute to unique combinations of bacteria within an individual that are derived from a very large set of common bacterial strains and species. Humans show apparently discrete variation in the production of methane. This is explained by the observation that methanogenic archaea can be detected in faeces from methane producers at > 108 ml−1, whereas in non-producers these levels are only 102−103 ml−1 (Dore et al., 1995). There is little evidence for a direct involvement of host genotype, however, in determining the numbers and activity of methanogens in the large intestine and environmental and dietary factors appear more likely to be responsible for this variation (Florin et al., 2000). More generally, the extent of control over the composition of the gut microbiota that is exerted by the host genotype, as distinct from the maternal or environmental inoculum and the influence of diet, lifestyle and medication, remains one of the major unanswered questions in gut microbiology (Dethlefsen et al., 2006; Ley et al., 2006a). A closely related, and also largely unanswered, question is the extent to which the immune system influences the species composition of the gut microbial community (Fagarasan, 2006).
Conclusions and future prospects
Detailed understanding of the physiology and genetics of individual gut microorganisms is currently limited to a few species that provide rather poor representation of the numerically predominant groups. There are now exciting opportunities to expand this knowledge through genome sequencing of a much wider range of cultured isolates, and through metagenomic approaches. Understanding the response of gut microbial communities to diet and other factors, however, presents an additional, and distinct, set of challenges. Significant progress has been made in relating 16S rRNA sequence information to function, aided by the availability of cultured isolates, and this is starting to reveal the niches occupied by certain key bacterial groups in vivo. Continued innovation in microbial ecology and systems analysis is needed, in particular to enable rapid analysis of microbial community composition in large sample sets, to monitor metabolic shifts and host interactions, and especially to track these events under in vivo conditions.
Acknowledgements
The Rowett Institute receives support from the Scottish Executive Environment and Rural Affairs Department.