The microbial world within us includes a vast array of gastrointestinal (GI) tract communities that play an important role in health and disease. Significant progress has been made in recent years in describing the intestinal microbial composition based on the application of 16S ribosomal RNA (rRNA)-based approaches. These were not only instrumental in providing a phylogenetic framework of the more than 1000 different intestinal species but also illustrated the temporal and spatial diversity of the microbial GI tract composition that is host-specific and affected by the genotype. However, our knowledge of the molecular and cellular bases of host–microbe interactions in the GI tract is still very limited. Here an overview is presented of the most recent developments and applications of novel culture-independent approaches that promise to unravel the mechanisms of GI tract functionality and subsequent possibilities to exploit specifically these mechanisms in order to improve gut health.
Microbial ecosystems are found all over the world. What a delight for microbiologists that one of the more complex ecosystems on this planet appears to lay within ourselves! Analysis of our microbiota along the highly variable niches within the human gastrointestinal (GI) tract has been a formidable task for microbiologists and is currently an energized topic in the scientific world (see Bäckhed et al., 2005 for a recent overview). The GI tract can be regarded as a very complex ecosystem, because it does not involve solely eukaryotic tissues like in other organs, but involves an interplay between food, host cells and microbes (see also Simpson et al., 2005). The challenge to study the interactions between the microbes residing in our GI tract and the rest of our body has gained impetus over the last decade due to the advent of molecular techniques. These methods, especially those rooted in the use of ribosomal RNA (rRNA) and the encoded genes have exposed an astonishing wealth of microbes within us. The composition of this microbiota can now be rapidly assessed by a choice of methods depending on the question and the answers that can link our microbial diversity to health, diet, geographic location, disease and genotype. While the importance of the intestinal microbiota to our health has been recognized for years, global technologies such as transcriptional profiling of gene expression are offering precious new insights into the communication between our microbes and us. Ultimately, this information might be used to influence the microbiota or the host in order to promote health, provide cures, or allow preventative measures against disease.
Evolutionary diversity in the GI tract
The use of rRNA has become established for classification and phylogeny of microbes. The 16S rRNA gene consisting of approximately 1.5 kb is a reasonable size for comparative sequence analysis, and harbours sufficient variable regions for identification and differentiation of specific species. Currently, nearly 200 000 16S rRNA sequences are available in the DNA databases, which is far more than for any other gene (see Cole et al., 2005 for recent review). The use of 16S rRNA sequence information to study microbial species diversity can be performed using several, often complementary, approaches such as sequencing of cloned 16S rRNA gene amplicons, 16S rRNA gene fingerprinting, quantitative dot blot- and fluorescent in situ hybridization (FISH), and quantitative PCR approaches (Table 1; Zoetendal et al., 2004). The choice of the approach used will mainly depend on the question to be answered.
Table 1. Overview of the approaches used in microbial ecology.
16S rRNA gene sequencing
16S rRNA genes
Collection of 16S rRNA gene sequences
Bias in DNA extraction, PCR and cloning
16S rRNA genes
Display of 16S rRNA gene diversity
In early stage of development
Dot blot hybridization
Relative abundance of 16S rRNA
Requires 16S rRNA gene sequence data
16S rRNA genes
Bias in DNA extraction and PCR
Enumeration of bacterial populations
Requires 16S rRNA gene sequence data
Non-16S rRNA gene Fingerprinting
Genomic DNA; cellular fatty acids
Identification of microbes
16S rRNA genes
Relative abundance of 16S rRNA genes
Bias in DNA extraction
Variation between genomes
Bias in DNA extraction and cloning efficiency
Probe-based cell sorting
Genomic DNA, plasmid DNA
Sorted cells containing gene of interest
Depends on sequence data
Unique gene sequences
Sensitive for false positives
Biological explanation of data
In situ isotope tracking
Identification of substrate-utilizing microbes
Only suitable for simple pathways
Real-time PCR (RT-PCR)
Specific gene expression
Only applicable to limited number of genes
Identification of induced promoters
Identification transcribed genes
Depends on selective amplification
Insert of transposons in genome
Depends on transformation ability of microbes
In several studies, fingerprinting of 16S rRNA genes has demonstrated that the composition of the predominant bacterial community is host-specific and stable over time in healthy adult individuals (Zoetendal et al., 1998; Tannock et al., 2000; Seksik et al., 2003; Vanhoute et al., 2004). In addition, sequencing of 16S rRNA gene clone libraries from the faeces of healthy adult humans has indicated that a significant fraction of the bacteria has not been described previously (Zoetendal et al., 1998; Suau et al., 1999; Eckburg et al., 2005). In addition, it was observed that the majority of sequences were derived from Gram-positive bacteria. Although these studies might have provided a biased representation of the bacterial community structure, a confirmation of major findings by FISH using 16S rRNA-targeted oligonucleotide probes has been performed. Harmsen et al. (2002) and Lay et al. (2005) demonstrated that the current probe sets cover approximately 80–90% of faecal bacteria in healthy adults, with 54–75% corresponding to Gram-positive bacteria (Fig. 1). Besides faecal samples, the bacterial community structure from other intestinal samples such as specimens from the human colon and ileum, has also been determined (Marteau et al., 2001; Hold et al., 2002; Wang et al., 2003; Eckburg et al., 2005). These studies suggested that faecal samples do not necessarily represent the bacterial community in other parts of the GI tract, as demonstrated by denaturing gradient gel electrophoretic (DGGE) analysis of 16S rRNA gene amplicons (Zoetendal et al., 2002; Lepage et al., 2005). Remarkably, this mucosa-associated bacterial community appeared uniformly distributed along the complete colon, indicative of interactions between host cells and microbes (Zoetendal et al., 2002; Nielsen et al., 2003; Lepage et al., 2005). In contrast, a recent paper by Van der Waaij et al. (2005) suggested that colonic microbes are not in direct contact with the mucosa as revealed by FISH microscopy of mucosal biopsy samples from the colon using 16S rRNA-targeted probes. In addition, no significant difference was found in community structure between colonic biopsies and faeces. This latter contrasting observation could, however, also be explained by methodological differences between the FISH and DGGE approach (Van der Waaij et al., 2005). Also samples from the upper intestine, the oesophagus and oral cavity have been characterized by cloning and sequencing of 16S rRNA genes and, as expected, the bacterial community found at these sites consists of completely different groups of bacteria compared with lower intestinal communities (Kroes et al., 1999; Paster et al., 2001; Pei et al., 2004). Interestingly, the same six groups of bacteria predominate in both locations, including firmicutes, bacteroidetes, actinobacteria, proteobacteria, fusobacteria and the uncultured group TM7 (Paster et al., 2001; Pei et al., 2004). Remarkably, the percentage of Gram-positive bacteria was ∼75% and ∼40% for oesophagus and oral cavity respectively, although different individuals were sampled for each location.
In summary, the application of culture-independent approaches to study the GI tract composition has expanded our knowledge of the microbial diversity present there. It is estimated that more than 1000 species inhabit our intestine (Whitfield, 2004), among which Gram-positive bacteria predominate. This indicates that novel high-throughput approaches which are 16S rRNA-based will become indispensable and argues for the development of high-diversity DNA microarrays.
The use of the 16S rRNA gene as a model for detection and identification of microbes has been widely applied in ecology. However, it does not give any indication of the metabolic potential in an ecosystem. Approaches based on genome sequences offer powerful insights into the physiological potential of microbes (Table 1). The increase in the number of completely sequenced microbial genomes makes the comparative genomics of microbes a reliable predictive tool for hypothesis-driven experimental design. The genomes of a small number of GI tract commensals have been sequenced, including Bacteroides thetaiotaomicron, several Lactobacillus species, Enterococcus faecalis and Bifidobacterium longum (Fig. 1). In silico comparisons between these and other bacteria have resulted in remarkable observations. For example, a very large number of phosphotransferase sugar transport systems were found in Lactobacillus plantarum, E. faecalis and Listeria monocytogenes, which are especially active in the small intestine, while genomes of bacteria predominating in the colon, such as B. thetaiotaomicron and Bi. longum contain many ABC transporters, permeases, proton symporters and an enormous number of genes involved in the utilization of complex carbohydrates (De Vos et al., 2004). These and other variations might reflect the ecological success of specific bacterial groups in the different parts of the GI tract. The availability of their genomes now allows for functional and other post-genomics approaches to validate these suggestions (Bron et al., 2004; Sonnenburg et al., 2005; De Vries, 2006).
Despite the value of comparative genomics, this approach is almost solely applied to well-studied isolates. The full extent of genomic diversity can be accessed using bacterial artificial chromosome (BAC) or related vectors to construct libraries of large genomic DNA fragments (> 100 kb) from communities. These are often termed metagenomic libraries. This approach provides direct access to large genomic fragments isolated directly from microbes in natural environments, which can be used to study the phylogenetic, physical and functional properties of the microbiome. Recently, a metagenomic approach has been performed to study the viral community in human faeces (Breitbart et al., 2003), which consisted of approximately 1200 viral genotypes. Interestingly, the majority of viral sequences in this collection showed highest sequence similarity to phages known to infect Gram-positive bacteria. This is in agreement with 16S rRNA gene-based studies, which indicate the predominance of Gram-positive bacteria in human faeces (Fig. 1). Metagenomic approaches have also been applied to prokaryotic communities and can be considered as the upcoming research activity in microbial ecology (for recent overviews see Handelsman, 2004; 2005). The first prokaryotic metagenomic libraries from intestinal samples of a variety of animals have recently been published and we expect to see an explosive increase in metagenomic data within the coming years (Ferrer et al., 2005; Walter et al., 2005; Manichanh et al., 2006). The first comparative metagenomic approach with faecal samples from humans indicated that the firmicutes phylum is reduced in complexity in Crohn's disease patients than in healthy subjects (Manichanh et al., 2006). A limitation of metagenomics is that the number of different genes within a community is so large that the number of cloned fragments must be enormous to get a reasonable coverage. In addition, the gene content of strains belonging to the same microbial species might already differ by as much as 20% (Boucher et al., 2001). Similar observations were made for 20 L. plantarum strains (Molenaar et al., 2005). This indicates that the genetic diversity within an ecosystem, such as the GI tract, is beyond our imagination. Furthermore, bacterial communities in the GI tract are host-specific, which makes it difficult to perform comparative genomics on an ecosystem level. Despite this limitation, initiatives to perform comparative metagenomics have been taken to compare and contrast libraries from different ecosystems (Tringe et al., 2005). Another approach to facilitate comparative metagenomics is to narrow down the number of clones. DNA microarray analysis and subtractive hybridization (SH) have been aimed at recovering unique genomic information by discriminating one microbe from another. SH is especially interesting as it reveals the differences between two organisms, rather than the similarities (Lisitsyn, 1995). This approach has mainly been used to compare pathogen and commensal isolates of the same genus or species. However, SH has also been successfully used to study genomic differences between two Ruminococcus flavefaciens strains and four Lactobacillus sobrius strains respectively (Antonopoulos et al., 2004; Konstantinov et al., 2005). Likewise, attempts to compare ruminal ecosystems with each other and to determine their uniqueness in terms of genetic potential that have been undertaken (Galbraith et al., 2004) reveal that it is possible to retrieve ‘ecosystem-specific’ genome fragments using SH.
As there are only a few (one to up to a dozen) rRNA operons in individual genomes and the library inserts are about 50–100 kb, only few clones of a metagenomic library include rRNA genes (Beja et al., 2000). This complicates linking specific gene sequences to microbes from which they originate. One way to enrich for metagenomic fragments containing rRNA genes is selection based on I-Ceu1 restriction site, which is unique in the 23S rRNA gene (Nesbøet al., 2005). Another approach for linking gene sequences to microbial origins is the use of a recently described cell sorting approach (Zwirglmaier et al., 2004). Here, cells are sorted using polynucleotide probes that hybridize to a target gene and, subsequently, these cells are captured on a microplate coated with DNA fragments complementary to the probes. It is assumed that some polynucleotide probes or parts of the probes will remain outside the target cells due to their size and network formation (Zwirglmaier et al., 2003).
Besides sequence-driven analysis, metagenomes offer great possibilities for function-driven analysis (see Handelsman, 2004 for recent review). Williamson et al. (2005) developed an intracellular approach to screening metagenomic libraries for clones that induced or inhibited quorum sensing. In this screen, DNA from pooled clones were transferred into cells containing a biosensor plasmid and subsequently detected and isolated by fluorescence-activated cell sorting (FACS). In a similar way, a substrate-induced gene-expression screen has been developed to identify genes in a groundwater metagenome that were induced by aromatic hydrocarbons (Uchiyama et al., 2005).
Genomics and metagenomics will provide a wealth of information and this requires continuous developments and improvements in high-throughput sequencing technologies and bioinformatics. It is obvious that there is a current need for genome sequences from the Clostridium leptum and Clostridium coccoides–Eubacterium rectale groups, because these are the most abundant low guanine and cytosine (GC) Gram-positive bacteria in the human GI tract as indicated in Fig. 1. In addition, combining sorting and complete genome amplification offers great possibilities to obtain genome sequences of uncultured bacteria (Raghunathan et al., 2005). However, as the diversity in the gut is enormous, and individual- and location-specific genome sequencing of selected and representative single isolates has to be complemented by sequence analysis of metagenomes from a variety of locations in the GI tract from different individuals.
Microbial diversity is host-mediated
Metagenomic approaches will provide a wealth of sequence information from single ecosystems, but they are not suitable for high-throughput monitoring. DGGE and related fingerprinting techniques have proven their power in comparing and monitoring ecosystems at the 16S rRNA gene or functional gene level. These approaches have been widely applied to study bacterial communities in the GI tract (Table 1, Zoetendal et al., 2004). The composition of the predominant bacterial community is stable over time and is host-specific in faeces from healthy adult individuals. Similar results have been described by Lay et al. (2005) based on the analysis of faecal samples from 91 individuals. They observed that there is large individual variability, which, however, could not be explained by geographic origin, age or gender. Moreover, the mucosa-associated bacterial community in the colon is also host-specific (Zoetendal et al., 2002; Nielsen et al., 2003; Lepage et al., 2005). In addition, independent studies indicate that the development of host-specific communities might already start after birth (Favier et al., 2002; Schwiertz et al., 2003; Wang et al., 2004). All these findings argue for a host-genotype impact on the GI community. To test this hypothesis, comparisons were made between DGGE profiles of faecal samples from human adults with differing genetic relatedness, varying from unrelated persons to monozygotic twins. This study indicated a significant impact of the host genotype on the bacterial composition in the GI tract, while there was hardly an environmental effect (Zoetendal et al., 2001). A recent study with children of varying ages and zygoty confirmed this finding (Steward et al., 2005). Remarkably, unstable faecal communities in adults can often be correlated with GI tract disorders, as in patients suffering from Crohn's disease (Seksik et al., 2003). In contrast, no differences were observed between ulcerated and non-ulcerated mucosa in Crohn's disease patients, which suggests a more prominent relation between the luminal populations and the disease (Seksik et al., 2005). A striking observation is that the number of bacteriophages in ulcerated mucosal tissues was very limited compared with healthy tissues from the same individuals (Lepage, 2005). Whether this reduced number is due to an increased induction of prophages or excretion of the phages into the lumen needs to be addressed in future. Although these studies suggest the presence of specific links between certain diseases and the microbial community in the GI tract, it will remain complicated to implicate bacterial populations specifically to GI tract disorders. First of all, it is difficult to determine whether bacterial shifts associated with a disease are the cause or the result of the disease. Second, the host-specificity of bacterial communities in the GI tract might result in individual variations among GI tract diseases. Searches for an association of GI infections and inflammatory bowel diseases, such as ulcerative colitis and Crohn's disease, with specific individuals, families or ethnic groups deserve special attention.
Viability and metabolic activity
While a first step in studying ecosystems is the identification of their members, a subsequent and important aspect is the analysis of what each organism is doing. Determining the in situ activity is a challenging issue, given that the majority of microbes escape cultivation. Microbes cannot be easy singled out and there is a wealth of biotic and abiotic interactions in the complex intestinal ecosystem.
Microbial activity can be measured using a variety of targets, such as cells, their constituent RNAs, proteins or metabolites, or using reporter systems (Table 1). Whatever target one selects, it must be realized that the detection spans of these targets differ considerably. Most transcriptome studies are, in fact, snapshots of the present as they monitor steady state concentrations of mRNAs. However, genetic approaches might also provide a picture of transcriptional activity in the past (see below). Similarly, proteins have a higher half-time than mRNAs, which can be exploited in promoter fusions. The detection range is more difficult to determine for the viability and metabolic parameters, because bacteria might multiply and metabolites might accumulate or be converted.
The basic question of ‘who is living’ can be answered for the easily cultivated microbes simply by culturing. However, alternative approaches are needed to determine the physiological status of the uncultured cells. It was recently shown that cells of bifidobacteria in pure culture can be enumerated and divided into active, injured and dead cell populations by flow cytometry (FCM) with live/dead staining probes (Ben-Amor et al., 2002). Using this FCM approach, up to one-third of all bacteria in human faeces were scored as dead (Apajalahti et al., 2003). A similar FCM study supported this conclusion and further showed that about 50% of the cells can be regarded as viable while 20% was found to be injured (Ben-Amor et al., 2005). These injured cells are characterized as cells that are detected with both life and dead staining probes and a previous study with bifidobacterial cultures demonstrated that they are viable, but difficult to cultivate (Ben-Amor et al., 2002). Remarkably, the composition of these three groups sorted by the FCM differed. Bacteria related to known butyrate-producing bacteria predominated in the active population, while bacteria affiliated to Bacteroides, Ruminococccus and Eubacterium were more abundant in the dead fractions (Ben-Amor et al., 2005). These are important findings, because they link phylogenetic information to activity. Butyrate-producing bacteria have been suggested to play an important role in the GI tract, because butyrate is one of the major carbon sources for the epithelium (Pryde et al., 2002), and the current FCM study confirmed that most of these cells are indeed active throughout the length of the colon (Ben-Amor et al., 2005).
Isolation and characterization of GI tract microbes have given us insight into their physiology and potential roles in the GI tract, but the ability to perform a certain function in culture does not mean that these microbes perform that function in situ. This is well demonstrated with environmental samples that have been incubated with substrates labelled with stable or radioactive isotopes. A convincing example was described by Manefield et al. (2002), who demonstrated that phenol degradation in a bioreactor was dominated by an uncultured member of the Thauera genus and not by the phenol-degrading isolates that could be cultured from the bioreactor. Notably, in view of the complex substrate conversions that occur in the colon, stable isotope probing of GI tract microbes might be an attractive approach for studying the in situ utilization of prebiotics and other carbohydrates (Egert et al., 2006).
It is much more difficult to study gene expression in situ in microbes than in eukaryotes, because most of the currently available methodology is applicable to poly A-tailed mRNA, which is rare in prokaryotes. Moreover, only a small fraction of total bacterial RNA consists of mRNA and prokaryotic mRNA is by far less stable than that from eukaryotes. However, a growing number of studies indicate that this approach can be successful. Deplancke et al. (2000) detected specific expression of adenosine-5′-phosphosulphate reductase at different locations in the mouse GI tract by using an RT-PCR approach. In addition, Fitzsimons et al. (2003) measured L. acidophilus slpA mRNA levels in human faeces samples spiked with this lactic acid bacterium. Moreover, the expression level of different Helicobacter pylori genes during colonization of the gastric mucosa in humans and mice, and the germination level of genetically engineered Bacillus subtilis spores in the mouse GI tract could be determined by real-time and competitive RT-PCR respectively (Rokbi et al., 2001; Casula and Cutting, 2002). A very recent study by Sonnenburg et al. (2005) demonstrated the adaptive responses of B. thetaiotaomicron in the GI tract of mice fed either with polysaccharide-rich or simple-sugar diets by transcriptional profiling. B. thetaiotaomicron expressed genes encoding outer-membrane polysaccharide binding proteins and glycoside hydrolases when mice were on the polysaccharide-rich diet, whereas in the absence of these polysaccharides, it switched to the expression of genes involved in utilization of host mucus glycans. This study demonstrates that transcriptional profiling can monitor the modulating capacities of diets on GI tract microbes, which is a promising tool for screening of novel functional foods. The first attempts to perform transcriptomics to study in vivo gene expression in L. plantarum in the human GI tract are underway. Despite the complications caused by several factors, including the availability of appropriate sample material, the isolation of intact mRNA and the background of the indigenous microbiota in the samples, the first results demonstrate that this approach works well in humans who have consumed L. plantarum (De Vries, 2006). The microarray data indicated that L. plantarum undertakes glycolytic degradation of simple carbohydrates in the small intestine and degrades the more difficult ones via the pentose-P pathway in the colon. Recently, a benchmark paper reported the combination of simultaneous FISH of mRNA and rRNA in samples from pure cultures, bivalve symbionts and sediment (Pernthaler and Amann, 2004) to monitor the expression of pmoA at a single cell level. The simultaneous detection of mRNA and rRNA in single cells is a very promising approach for linking phylogeny and activity, and opens up the way to characterize in situ gene expression of uncultured microbes in the GI tract.
A different approach for measuring gene expression in ecosystems is called in vivo expression technology (IVET). The IVET strategy allows identification of promoters that are specifically induced when bacteria are exposed to certain environmental conditions. Other strategies include selective capture of transcribed sequences (SCOTS) and signature-tagged mutagenesis (STM) (Table 1). IVET has mainly been used to study gene expression in pathogens, but it has also been used to study colonization by L. reuteri of the mouse GI tract (Walter et al., 2003). However, only three genes could be linked to Lactobacillus colonization in this experimental system that required continuous antibiotic feeding to the mice. These genes showed homology to xylose isomerase (xylA), peptide methionine sulphoxide reductase (msrB) and a gene with unknown function respectively. In a more comprehensive study, the resolvase-based IVET (R-IVET) method has been used to identify promoters induced in L. plantarum during its passage through the GI tract of conventional mice (Bron et al., 2004). An erythromycin resistance gene (ery) flanked by two loxP boxes was introduced into the genome of a rifampicin-resistant strain of L. plantarum WCFS1 and a vector containing the Cre resolvase (cre) was used to screen the promoter library. The result of specific promoter induction is the removal of ery from the chromosome by cre and subsequent loss of erythromycin resistance. This irreversible result of promoter induction makes the R-IVET method easier to apply than the standard IVET, because no selective pressure by antibiotic feeding is needed. A total of 72 L. plantarum genes were identified as induced during GI tract passage. The functions of these gene products include sugar transport, synthesis of amino acids, cofactors, nucleotides and vitamins and stress response. Interestingly, one of the hypothetical open reading frames found is a homologue of a gene with unknown function found in L. reuteri, which might indicate its importance during passage of lactobacilli through the GI tract. Remarkably, a large number of the functions and pathways identified as potentially GI-specific in L. plantarum are also found in pathogens in which they are important during infection, suggesting that survival, rather than virulence, explains their expression during disease-related infection. IVET and R-IVET are powerful approaches and, combined with transcriptome approaches, will provide more insight into gene expression in specific niches of a complex ecosystem such as the GI tract (Bron et al., 2004; De Vos et al., 2004).
Specific activity can also be analysed using proteomics and a promising metaproteomic approach has recently been reported for activated sludge and acid mine drainage (Wilmes and Bond, 2004; Ram et al., 2005). This provides possibilities to apply this technology to the GI tract ecosystem and the first characterization of the GI tract metaproteome of newborns indicated the presence of a new enzyme in bifidobacterial metabolism (E.S. Klaassens, W.M. de Vos and E.E. Vaughn, submitted). Metabolomics is another popular but complex ‘∼omics’ approach to profile activity. An excellent overview by Nicholson et al. (2005) discussed how metabolomics can be applied to study the impact of drugs and diets on the interactions between humans and the GI tract community. The authors concluded that appropriate consideration of individual human GI tract activities will be a necessary part for future personalized healthcare paradigms. In addition, metabolomics might be very useful in studying quorum sensing or other communication systems in the GI tract, because there is accumulating evidence to suggest that quorum sensing plays an important role and is involved in adhesion in Gram-positive bacteria (Sturme et al., 2002, 2005).
Despite the fact that ‘omics’-based studies in GI tract ecology are still in their infancy, their future application looks very promising. Combining all these approaches will enable us to make a substantial step forward into unravelling roles of microbes in the GI tract.
Impact of microbes on the host
Studies of germ-free mice revealed that the microbial community plays an important role in the GI tract health status (see review article by Hooper et al., 2002). On the one hand, the microbial community helps the host with the processing of nutrients. On the other hand, the microbial community can also be a source for (opportunistic) pathogens. The intestine of germ-free mice has several physiological differences and a less developed immune system than that of conventional mice. The microbial community also produces vitamins and other components, such as butyrate, which are utilized by the host. Despite this, only recently have insights into the molecular mechanisms underlying the interactions between host and bacteria been uncovered (Table 2) . A hallmark paper by Bry et al. (1996) showed that B. thetaiotaomicron was able to induce the production of fucosylated glycans in the gut of mono-associated mice. Interestingly, a B. thetaiotaomicron mutant which was disrupted in its fucose utilization pathway did not show this induction, indicating that specific host–microbe cross-talk takes place in the GI tract. After this exiting discovery, Hooper et al. (2001) studied the global transcriptional responses of germ-free mice to colonization by B. thetaiotaomicron using DNA microarrays and real-time RT-PCR. In addition, the specific cellular responses were determined by laser-capture microdissection. Their results demonstrated that B. thetaiotaomicron was able to modulate expression of genes with a variety of physiological functions, including nutrient absorption and immune responses. In addition, the host reacted differently to colonization of Escherichia coli and Bi. infantis, which indicated that microbial interactions with host cells in the GI tract might be microbe-specific. In a related study, it was shown that Paneth cells in the mouse intestine secrete a previously uncharacterized angiogenin, Ang4, after colonization by B. thetaiotaomicron (Hooper et al., 2003). This secreted peptide has bactericidal activity against intestinal microbes and is part of the innate immune response. In addition, these Paneth cells induce the development of the vascular network in the villi after bacterial colonization. These studies indicate that specific communication between host and microbes can shape the immunity and maturation of the GI tract of host and consequently, has a substantial impact on GI tract ecology.
Table 2. Described impact of microbes on host response.
It has been suggested that toll-like receptors (TLRs) could play an important role in host–microbe interactions (Rakoff-Nahoum et al., 2004). TLRs comprise a family of pattern-recognition receptors that function as sensors of microbial infection and are critical for the initiation of inflammatory and immune defence responses. This initiation by TLRs starts with the recognition of components of microbial origin (Takeda et al., 2003). Most well-known examples are the recognition of lipopolysaccharides (LPS), lipoteichoic acids (LTA) and DNA by TLR4, TLR2 and TLR9 respectively. It is still puzzling how TLR-mediated recognition allows differentiation between pathogens and commensal bacteria, because LPS and LTA are widely spread among pathogens and commensals. It was suggested that sequestration of the GI tract community by the epithelium prevents an inflammatory response to the commensals. However, it was recently discovered that commensal bacteria are recognized by TLRs under normal steady-state conditions and that this recognition is crucial for the maintenance of the intestinal epithelial homeostasis (Rakoff-Nahoum et al., 2004). In addition, it was observed that activation of TLRs by commensals is critical for the protection against gut injury. These are very important findings because they argue for a new role for TLRs.
Some recent studies have indicated the important recognition role of TLRs in the intestine. Rachmilewitz et al. (2004) demonstrated that DNA from probiotics VSL-3 was already sufficient to mediate an anti-inflammatory response in a colitis-induced mouse model via TLR9 signalling. More recently, it has been demonstrated that the molecular structure of the signalling component plays an essential role in the TLR-sensing and its subsequent response regulation. Comparative studies of L. plantarum wild-type strain and its isogenic Dlt– mutant, deficient in d-alanylation of LTA revealed a marked shift from pro- to anti-inflammatory response and a protective effect in a colitis mouse model (Grangette et al., 2005). These are very important findings and make a case for more comparative studies between wild-type and mutant strains (De Vos, 2005).
Microbes sometimes influence host physiology in ways not directly related to the GI tract function. The microbial community in the caecum of conventional mice was shown to be responsible for a 60% increase in body fat content, irrespective of the food intake by the mice (Bäckhed et al., 2004). The main mechanism behind this phenomenon is the regulation of fasting-induced adipocyte factor (FiaF). Expression of this enzyme is repressed by the microbial community in the caecum and, consequently, the adipocytes increase the triglyceride production, as demonstrated by comparisons of wild-type and fiaF knock-out mice. These studies indicate that the GI tract community affects not only the host immune system but also energy harvest and storage in the host. These types of host–microbe interactions definitely deserve more attention.
One should keep in mind that the above-mentioned studies are all based on animal model systems and we do not know how well these animal models reflect the human situation. The application of modern molecular techniques allows us to analyse the different model systems in more detail. However, whether the outcomes will be sufficient to validate these models remains to be answered in future.
This review provides a short summary of the microbial ecology of the human GI tract, with the main focus on the use of novel culture-independent approaches. It is well recognized that the application of culture-independent approaches has revolutionized microbial ecology and enables us to answer questions that cannot be addressed by culture methods. An overview of the present research position and possible future avenues is represented in Fig. 2.
So far, most research has focused on the detection of microbes using 16S rRNA-based approaches. Despite the enormous amount of data from this type of studies, there is a current need for more baseline studies. For example, little is known about the impact of different ages, and host genotypes, and their relation with diet and intestinal disorders. This definitely deserves attention in the future. In addition, the developments in ‘(meta-)∼omics’ approaches are needed. Although the number of (meta)genomes has drastically increased during the past years, we are still at the beginning of this novel era. Currently, the main problem is our inability to convert the mass of ‘∼omics’ data into biological meaningful information for instance by using systems approaches. This may be complicated by the observation that the intestinal bacterial community is host-specific and dependent on the host genotype. These differences in diversity make it even more challenging to find general functional patterns in the GI tract and this requires dedicated statistical tools for applying comparative ‘∼omics’.
Host–microbe interactions can be very complex. Therefore, the comparisons between gene expression profiles of different hosts and microbes should be extended with comparisons between wild-type organisms, their corresponding directed and isogenic mutants, and combinations of these (Hooper et al., 2001; 2003; De Vos, 2005; Grangette et al., 2005). In addition, the use of RNA for interfering gene regulation in animals looks very promising for studying host–microbe interactions. Using Drosophila S2 cells as a model system, Cheng et al. (2005) detected many previously uncharacterized host functions involved in various steps of listerial pathogenesis by genome-wide RNA interference screening. These types of approaches will simplify the comparisons and promises to expand our insight into the complexity of intestinal host–microbe interactions, because the precise genetic differences between the host and microbe combination, respectively, are known.
Another research area that deserves special attention in the near future is the microbe–microbe interactions. It is estimated that each species of microbe is surrounded by a myriad of other microbial and phage species. It is evident that the role of all these players is important in the functioning of the ecosystem and therefore, their interactions with each other should not be neglected.
In the coming years, we expect to see an increase in the development and application of novel molecular technologies and, as a result, increasing amounts of ‘∼omics’ data. These developments might ultimately help us to understand the microbial ecology of the GI tract and provide us with insight into the mechanisms underlying GI tract health and disease. This may subsequently enable us to find biomarkers for gut health and allow us to move from observation into prediction of GI tract health status that is impacted by the microbial world within us.
The authors thank MSc. Mirjana Rajilic for help with construction of the phylogenetic tree in Fig. 1. We are very grateful to Muriel Derrien, Eline Klaassens and Maaike de Vries for sharing some of their to-be-published data.