Molecular Analysis of Commensal Host-Microbial Relationships in the Intestine

See allHide authors and affiliations

Science  02 Feb 2001:
Vol. 291, Issue 5505, pp. 881-884
DOI: 10.1126/science.291.5505.881


Human beings contain complex societies of indigenous microbes, yet little is known about how resident bacteria shape our physiology. We colonized germ-free mice with Bacteroides thetaiotaomicron, a prominent component of the normal mouse and human intestinal microflora. Global intestinal transcriptional responses to colonization were observed with DNA microarrays, and the cellular origins of selected responses were established by laser-capture microdissection. The results reveal that this commensal bacterium modulates expression of genes involved in several important intestinal functions, including nutrient absorption, mucosal barrier fortification, xenobiotic metabolism, angiogenesis, and postnatal intestinal maturation. These findings provide perspectives about the essential nature of the interactions between resident microorganisms and their hosts.

Human beings harbor an incredibly complex and abundant ensemble of microbes. We are in contact with components of this microflora from birth, yet little is known about their influence on normal development and physiology. The human intestine is more densely populated with microorganisms than any other organ and is a site where the microflora may have a pronounced impact on our biology. We tested this idea at a molecular level using a simplified mouse model of the interactions between gut commensals and their host. In this model, adult germ-free animals are colonized withBacteroides thetaiotaomicron. This anaerobe was chosen because it can be genetically manipulated and is a prominent member of the adult mouse and human gut microflora. Moreover, B. thetaiotaomicron normally colonizes the distal small intestine (ileum) during the suckling-weaning transition, a time of rapid and pronounced functional maturation of the gut (1,2). We previously used this model to show how B. thetaiotaomicron regulates production of ileal epithelial fucosylated glycans for its own nutritional benefit (3).

As so little is known about the range of intestinal functions that are shaped by components of the microflora, we used high-density oligonucleotide arrays for analysis of the host transcriptional responses evoked by B. thetaiotaomicron colonization. In addition, we defined the cellular origins of a subset of these responses using laser-capture microdissection and real-time quantitative reverse transcriptase polymerase chain reaction (qRT-PCR) and examined the specificity of selected B. thetaiotaomicron–elicited responses using other members of the normal flora.

Adult male germ-free mice were inoculated with B. thetaiotaomicron and killed 10 days later (4). Ileal RNA was isolated from mice with >107 colony-forming units (CFU) of bacteria per milliliter of ileal contents. Earlier studies had shown that 10 days was sufficient to produce robust colonization of the ileum and that ≥107 CFU/ml were necessary for full induction of fucosylated glycan expression in the ileal epithelium (3, 5). Affymetrix Mu11K and Mu19K GeneChips, representing ∼25,000 mouse genes, were used to compare ileal gene expression in age-matched germ-free and colonized animals.

mRNAs represented by 118 probe sets changed ≥twofold with colonization as defined by duplicate microarray hybridizations (6). Transcripts represented by 95 probe sets increased, whereas those represented by 23 probe sets decreased. Seventy-one genes, represented by 84 probe sets, were assigned to functional groups, whereas 34 transcripts were from uncharacterized genes or expressed sequence tag clusters (7). The microarray data reveal the unanticipated breadth of this commensal's impact on expression of genes involved in modulating fundamental intestinal functions.

Germ-free rodents require a higher caloric intake to maintain their weight than those with a microflora (8). Microarray analysis provided molecular insights into how B. thetaiotaomicronimproves host nutrient absorption and processing. Colonization led to increased ileal levels of Na+/glucose cotransporter (SGLT-1) mRNA (Table 1) (7,9). Concerted increases were seen in several components of the host's lipid absorption machinery, including pancreatic lipase-related protein-2 (PLRP-2), colipase, a fatty acid–binding protein (L-FABP), and apolipoprotein A-IV (7) (Table 1). The prominent decrease in expression of fasting-induced adipose factor, a PPARα target that is repressed with fat feeding (10), provided further evidence for augmented lipid uptake in colonized mice. Micronutrient absorption also appears to be augmented by colonization, as evidenced by a threefold increase in expression of the high-affinity epithelial copper transporter (CRT1) and five- to sixfold decreases in expression of metallothioneins I and II and ferritin heavy chain, which sequester heavy metals within cells (7) (Table 1).

Table 1

Real-time qRT-PCR verification of selected colonization-associated changes in gene expression. Assays were performed with the pooled ileal RNA samples used for the microarray studies. Mean values for triplicate determinations ±1 SD are shown. See (7) for a comprehensive list of 71 genes that change with colonization.

View this table:

The DNA microarray and confirmatory qRT-PCR analyses measure host responses in a complex tissue composed of multiple cell types. An in vivo model will, unlike cell culture–based models, preserve the contributions of other cell lineages and environmental factors in shaping the response. The challenge is to recover the responding cell population, without perturbing its expressed mRNAs, so that its reaction to the microorganism can be characterized quantitatively. To define the cellular responses in our in vivo model, we combined laser-capture microdissection (LCM) with qRT-PCR. LCM was used to recover three cell populations from frozen sections of ileum (11), including epithelium from crypts (containing proliferating undifferentiated cells plus differentiated members of the Paneth cell lineage), epithelium overlying villi (containing postmitotic, differentiated members of the other three intestinal epithelial lineages), and mesenchyme underlying crypt-villus units [see (7) for LCM images]. Specific mRNAs in the LCM populations were quantified by RT-PCR.

Colipase is produced by pancreatic exocrine acinar cells, but we discovered that colipase is also expressed in ileal crypt epithelium, where it increased 10-fold upon B. thetaiotaomicroncolonization (Fig. 1 and Table 1). Colipase plays a critical role in lipid metabolism by stimulating the activity of both pancreatic triglyceride lipase and pancreatic lipase-related protein-2 (12). Localization of colipase in the crypt epithelium and its regulation by a gut commensal reveal a previously unappreciated mechanism for lipid processing in the intestinal epithelium.

Figure 1

Real-time qRT-PCR analysis of colonization-associated changes in gene expression in laser-capture microdissected ileal cell populations. Values are expressed relative to levels in germ-free mesenchyme (9). Each mRNA was assayed in triplicate in three to four independent experiments. Representative results (mean ± 1 SD) from pairs of germ-free (gf) and colonized (col) mice are plotted.

An intact mucosal barrier is critical for constraining resident intestinal microbes. Barrier disruption can provoke immune responses that cause pathology such as inflammatory bowel disease (13). Bacteroides thetaiotaomicron colonization produces no detectable inflammatory response, as judged by histologic surveys (5) and by the absence of a discernible induction or repression of immune response genes represented on the microarrays. An influx of immunoglobulin A (IgA)–producing B cells does occur in the ileal mucosa 10 days after introduction of B. thetaiotaomicron(5), but similar commensal-induced IgA responses have been shown to be T cell–independent and to enforce barrier integrity (14).

The influx of IgA-producing B cells was accompanied by increased expression of the polymeric immunoglobulin receptor (pIgR) that transports IgA across the epithelium (Table 1). There was augmented expression of the CRP-ductin gene encoding both a mucus layer component (MUCLIN) and a putative receptor for trefoil peptides (15, 16). Decay-accelerating factor (DAF), an apical epithelial inhibitor of complement-mediated cytolysis, increased sixfold (7) (Table 1). Coincident enhancement of expression of these three genes should help prevent bacteria from crossing the epithelial barrier and avoid mucosal damage from activation of complement components in intestinal secretions.

The most pronounced response to B. thetaiotaomicronwas an increase in small proline-rich protein-2 (sprr2a) mRNA (Table 1). sprr family members contribute to the barrier functions of squamous epithelia, both as a component of the cornified cell envelope and as cross-bridging proteins linked to desmosomal desmoplakin (17). LCM/qRT-PCR revealed that sprr2a mRNA is present in the epithelium, primarily on the villus rather than crypt, and that B. thetaiotaomicron elicits a 280-fold increase in its villus epithelial expression (Fig. 1). The epithelial sprr2a response suggests that this protein participates in fortifying the intestinal epithelial barrier in response to bacterial colonization.

Environmental and dietary constituents, as well as drugs, are detoxified in the intestine by oxidation or conjugation. Colonization produced a twofold decrease in expression of glutathioneS-transferase, which conjugates glutathione to a variety of electrophiles, and a fourfold decrease in multidrug resistance protein–1a (Mdr1a), which exports glutathione-conjugated compounds from the epithelium (18) (Table 1). Debrisoquine hydroxylase (CYP2D2), which is involved in oxidative drug metabolism in humans (19), also declined threefold. These results indicate that colonization with B. thetaiotaomicron affects the host's capacity to metabolize xenobiotics and endogenous toxins.

Previous electrophysiological studies of germ-free and conventionally raised animals indicate that the microflora influences gut motility (20). Colonization led to two- to fivefold increases in mRNAs encoding l-glutamate transporter,l-glutamate decarboxylase (converts glutamate to γ-aminobutyric acid), vesicle-associated protein-33 (a synaptobrevin-binding protein involved in neurotransmitter release) (21), and enteric γ-actin and cysteine-rich protein-2 (muscle-specific proteins). These results suggest that B. thetaiotaomicron may affect components of the enteric nervous system (22) and motility.

Bacteroides thetaiotaomicron colonizes the mouse (and human) gut during the weaning period (1, 2). This stage in postnatal gut development is marked by several maturational changes, including reductions in ileal epithelial lactase, which hydrolyzes milk sugar (23). Ileal epithelial lactase levels also fall as adult germ-free mice are colonized with B. thetaiotaomicron(Table 1). Adenosine deaminase (ADA) and polyamines are effectors of postnatal intestinal maturation (24,25). Colonization increased expression of both ADA and ornithine decarboxylase (ODC) antizyme, a regulator of polyamine synthesis (26), suggesting that commensals may perform an instructive role in postnatal intestinal maturation (7) (Table 1).

Colonization increased expression of angiogenin-3, a secreted protein with known angiogenic activity (27, 28), and angiogenin-related protein, which does not have any apparent angiogenic activity (29). Angiogenin-3 was originally identified in NIH 3T3 fibroblasts (27), but little is known about its cellular origins in tissues. LCM/qRT-PCR revealed that angiogenin-3 mRNA is largely confined to the crypt epithelium and that colonization results in a sevenfold increase in its crypt expression (Fig. 1). This increase accounts for the change in expression defined by microarray and qRT-PCR analyses of total ileal RNA (Table 1). Localization of a secreted angiogenesis factor in the crypt epithelium puts it in a strategic position to function as an effector of several host responses to microbial colonization (e.g., enhanced absorption and distribution of nutrients).

Our findings raise two questions. Are similar host responses elicited by other components of the gut microflora, and do changes in the metabolic capacity of B. thetaiotaomicron affect ileal gene expression? To examine the first question, age-matched adult male mice (n = 4 to 8 mice per group) were colonized for 10 days with B. thetaiotaomicron, Bifidobacterium infantis (a prominent component of the preweaning human and mouse ileal flora and a commonly used probiotic), Escherichia coliK12 (a normal component of human intestinal flora), or a “complete” ileal/cecal microflora harvested from conventionally raised mice (5). qRT-PCR was used to compare mRNA levels in each group (all animals had ≥107 CFU/ml ileal contents). Ileal expression of colipase and angiogenin-3 was induced after colonization with each of the three organisms and by the complete ileal/cecal flora (Fig. 2). Moreover, the ileal levels of colipase and angiogenin-3 mRNAs in these ex–germ-free mice were comparable to those of age-matched mice conventionally raised since birth (Fig. 2). In contrast, the response of sprr2a to colonization depended on the colonizing species: B. infantis and E. coliproduced only small increases (Fig. 2). Mdr1a and glutathioneS-transferase also exhibited species-specific responses.Bacteroides thetaiotaomicron suppressed and E. coli and B. infantis stimulated expression of both genes, whereas the multicomponent ileal/cecal flora produced no significant (i.e., ≥twofold) change in levels of either mRNA compared with germ-free controls. The differing Mdr1a/GST responses suggest that variations in xenobiotic metabolism between individuals may arise, in part, from differences in their resident gut flora.

Figure 2

Specificity of host responses to colonization with different members of the microflora. Germ-free mice were inoculated with one of the indicated organisms or with a complete ileal/cecal flora from conventionally raised mice (Conv. μflora) (4). Ileal RNAs, prepared from animals colonized at ≥107 CFU/ml ileal contents 10 days after inoculation, were pooled, and levels of each mRNA shown were analyzed by real-time qRT-PCR. Mean values (±1 SD) for triplicate determinations are plotted.

The only B. thetaiotaomicron genes currently known to link changes in bacterial metabolism with host responses are those involved in fucose utilization (3). Transposon-mediated mutagenesis of fucI (encoding fucose isomerase) blocks the organism's ability to use fucose as a carbon source and to signal fucosylated glycan production in the ileal epithelium (3). Microarray analysis of the host response to colonization revealed no appreciable differences between isogenic mutant and wild-type strains. This similarity extends to all genes in Table 1. Future identification of microbial factors that interlink microbial and host physiology will require characterization of changes in B. thetaiotaomicrongene expression as a function of colonization.

In summary, the studies described above provide a broad-based in vivo characterization of transcriptional responses to colonization with a prototypic gut commensal. Our results reveal that commensals are able to modulate expression of host genes that participate in diverse and fundamental physiological functions. The species selectivity of some of the colonization-associated changes in gene expression emphasizes how our physiology can be affected by changes in the composition of our indigenous microflora. The fusion of germ-free technology, functional genomics, and LCM/qRT-PCR makes it possible to use in vivo systems to quantify the impact of a microbial population on host cell gene expression.

  • * To whom correspondence should be addressed: E-mail: jgordon{at}


View Abstract

Navigate This Article