Abstract

Introduction

Development of the mammalian gastrointestinal tract is divided into three major phases demarcated by birth and weaning: in utero development, the suckling period, and the juvenile phase. The transition from one phase to another is precisely coordinated and tightly regulated by intrinsic and extrinsic factors. The first major transition at birth prepares fetal gut for digestion and absorption of milk. The second transition at weaning prepares the gut for digestion and absorption of solid food (Henning 1987; Montgomery et al. 1999; Nanthakumar 2001). The first transition is often assessed by the expression of lactase, responsible for digesting the milk sugar lactose, a suckling phenotype marker (Henning and Nanthakumar 1994). The second transition is typically measured as the induction of disaccharidases, such as sucrase, isomaltase, maltase, etc., which are responsible for terminal digestion of carbohydrates of adult food. These markers are traditionally used to measure the onset of intestinal maturity (Henning and Nanthakumar 1994; Nanthakumar 2001). The maturational changes of these enzyme markers are regulated at the level of transcription by intrinsic and extrinsic mechanisms.

Critical intrinsic and extrinsic factors that regulate intestinal maturation are circulating glucocorticoids and gut microbiota, respectively. Using sucrase induction as an outcome, exogenous and endogenous circulating glucocorticoid has been shown to precociously induce intestinal maturation (Nanthakumar and Henning 1993). The in vivo importance of glucocorticoids was reaffirmed by the discovery of a developmental surge of circulating corticosterone in the rat and mouse 2 days prior to the initiation of sucrase induction (Martin and Henning 1984). Elimination of this endogenous surge by adrenalectomy confirms the importance of glucocorticoids for coordination of this developmental transition (Martin and Henning 1984; Nanthakumar and Henning 1993; Henning and Nanthakumar 1994). Both exogenous and endogenous glucocorticoids administered during the first week of post-natal life can also precociously induce the adult phenotype of rodent gut (Martin and Henning 1984; Nanthakumar and Henning 1993; Henning and Nanthakumar 1994). In addition to inducing digestive and absorptive capacity to adult food, glucocorticoids can also accelerate the proliferative and migratory capacity of the newly differentiated cells in the small intestine (Henning 1987; Montgomery et al. 1999). The direct effect of glucocorticoids on the proliferative and maturational capacity of the intestine has been demonstrated in explant cultures (Simo et al. 1992). However, intestinal responsiveness to glucocorticoids is lost by the fourth post-natal week, after the intestine matures into the adult intestinal phenotype (Nanthakumar and Henning 1993); the mechanism responsible for the loss of glucocorticoid responsiveness is not known. Disaccharidases are widely used as tissue-specific markers to study the mechanism of steroid action on initiating the onset of the adult phenotype (Nanthakumar 2001; Nanthakumar et al. 2005). However, disaccharidase maturation is insensitive to luminal factors, including microflora (Henning 1987; Nanthakumar 2001). Developmentally dependent expression of glycosyltransferases can be influenced by the microbiome (Dai et al. 2000; Nanthakumar et al. 2003) and are also good markers of mucosal maturation. Thus, studying the interplay between steroids and luminal microbes in the ontogeny of gut maturation can be achieved by monitoring both a representative disaccharidase and glycosyltransferase.

The terminal ends of glycoconjugates in the suckling gut are predominantly sialic acid, whereas in the adult they are predominantly fucose (Biol et al. 1987; Dai et al. 2000). Therefore, the ratio of sialic acid to fucose in non-reducing terminal residues of glycoconjugates in the rodent gut epithelium reverses from high (sialic acid dominant) to low (fucose dominant) at the time of weaning (Biol et al. 1987; Dai et al. 2000, 2002; Nanthakumar et al. 2003). These changes in the composition of the terminal ends of glycans are due to decreasing levels of sialyltransferase activity with concurrent increases in the level of fucosyltransferase activity (Biol et al. 1987; Bry et al. 1996; Dai et al. 2000, 2002; Nanthakumar et al. 2003). The developmental decline in α2,3/6-sialyltransferase activity are observed in conventionally raised mice. However, in the absence of microbes in germ-free mice, these developmental changes of α2,3/6-sialyltransferase enzyme activities are not observed (Dai et al. 2002), and the suckling pattern of glycosylation persists (Dai et al. 2002; Nanthakumar et al. 2003). However, introduction of a mixture of adult fecal microbiota or Bacteroides thetaiotaomicron, one of its predominant species, can initiate fucosyltransferase expression to adult type within 2 weeks (Bry et al. 1996; Nanthakumar et al. 2003). Thus, differential glycosylation from the sialic acid predominant suckling phenotype to the fucose dominant adult phenotype of the mucosa defines the transition from pre- to post-weaning mouse intestine.

α1,2-Fucosyltransferase and sucrase activities were used as markers to delineate the role of extrinsic and intrinsic factors on this post-natal transition. The studies specifically address whether regulation of developmental α1,2-fucosyltransferase (FucT II) expression is independent of sucrase ontogeny, glucocorticoids regulate these processes, gut colonization influences this regulation, and post-weaning mucosal fucosylation influences homeostasis of adult gut. The ontogeny of α1,2-fucosyltransferase (GDP-Fuc: α1,2-fucosyltransferase) and sucrase-isomaltase (SI) gene expression was determined by measuring their enzyme activities, mRNA levels, and glycoconjugate expression in the mouse gut. These developmental studies were carried out using conventionally colonized mice, uncolonized germ-free and bacteria-depleted mice, and re-colonized mice. Recolonized mice were previously germ-free and bacteria-depleted mice, which had been inoculated with normal gut microbes freshly collected from age-matched colonized mice. The roles of intrinsic (glucocorticoid) and extrinsic (gut microbiota) factors were monitored in duodenum, jejunum, ileum and colon to determine their interactions on gut maturation pre- and post-weaning. Finally, the role of bacteria-induced fucosyltransferase activity on gut homeostasis was investigated. These experiments are designed to help understand the physiology of controlling gut development thought to underlie emerging inflammatory diseases whose risk factors include FUT2 polymorphisms.

Results

In conventionally colonized mice, the developmental expression of α1,2/3-fucosyltransferase activity and of sucrase activity was measured in the intestine for the first 6 weeks of life. Week 2 represents the preweaning gut, and week 6 represents the mature gut. The α1,2/3-fucosyltransferase assay measured changes in α1,2- or 1,3-fucosyltransferase activity, simultaneously screening for changes in the expression of several fucosyltransferase genes. We devised the assay with phenyl-β-d-galactoside, a substrate that measures broad α1,2- and 1,3-fucosyltransfease activity, as we did not know a priori which might be affected. The fucosyltransferase activity increased in all regions of the intestine between the 2nd and 3rd weeks of life and reached the mature level found in the adults by the fourth week (P < 0.001; Figure ​Figure1).1). The sucrase activity followed a parallel increase in the maturing duodenum, jejunum, and ileum (P < 0.001); sucrase is not expressed in the colon. Although all four regions displayed maturation-dependent increases in fucosyltransferase activity, absolute activities were much lower in the proximal gut than in the distal gut. SI activity was highest in the jejunum and lowest in the duodenum and ileum. These systematic regional differences follow their function: the digestion of disaccharides occurs mostly in the jejunum, and this is where the disaccharidases are expressed most heavily. In contrast, if the fucosylated cell surface of the mucosa relates to their interaction with the microbiota (Nanthakumar et al. 2003), the increasing proximal-to-distal gradient follows that of the density of the microbiota. Both of these activities increase with the maturation of the gut at weaning. Note that fucosyltransferase activity is initially high in colon, even during the suckling period, and the colon is also heavily colonized at this time.

Open in a separate window

Fig. 1.

Post-natal development of α1,2/3-fucosyltransferase activity (FucT ) (nmol/mg protein/h) and sucrase activity (µmol/mg protein/h) in the mouse (A) duodenum, (B) jejunum, (C) ileum and (D) colon. Six to eight litters of pups were used, and age denotes the post-natal week. Results are expressed as the mean ± SEM compiled from three samples. Tissues were pooled from six to eight mice per sample for 1- and 2-week-old mice and at three tissues per sample for 4- and 6-week-old mice.

Mice were treated with CA on day 10; on day 14, fucosyltransferase and sucrase activities were measured to determine glucocorticoid effect on intestinal development. CA administration accelerated the maturation of intestinal fucosyltransferase and sucrase activities in all regions of the small and large intestine (Figure ​(Figure22A).

Open in a separate window

Fig. 2.

Effect of CA on precocious maturation of (A) α1,2/3-fucosyltransferase activity (nmol/mg protein/h) and sucrase activity (µmol/mg protein/h) and (B) fut2 mRNA and SI mRNA in the suckling mouse gastrointestinal tract. Four litters of pups were used; on post-natal day 10, half the pups in each litter were either treated with CA or vehicle (VEH; saline) and tissue were harvested and assayed on day 14. Each bar represents the mean ± SE of 3-4 samples. *P < 0.05 vs control (VEH) group.

The change in the α1,2/3-fucosyltransferase activity could be due to changes in the expression of any of several fucosyltransferase genes. To determine which of the α1,2- or α1,3-fucosyltransferase genes were driving this change, expression of fut1 and fut2 (the α1,2 gene family) and fut4 or fut7 (representing commonly expressed members of the α1,3 gene family) were measured as levels of their respective mRNAs. Under all of our experimental conditions that resulted in induction of α1,2/3-fucosyltransferase activity, only expression of fut2 mRNA increased, and the magnitude of change was greater than that of the enzyme activity, consistent with the changes in the mixed fucosylation activity of the intestinal mucosa being driven primarily by altered transcription of the fut2 gene.

In Figure ​Figure2B,2B, the increase in fucosyltransferase activity and sucrase activity in response to CA corresponded to an even more pronounced increase in fut2 mRNA and SI mRNA, respectively. In contrast, these activities do not change in other organs such as the liver (not shown), indicating that cortisone induces precocious maturation specifically in the maturing intestine.

The duration of sensitivity to cortisone induction was measured for both fucosyltransferase and sucrase. Mice were treated on days 9, 13, 17, and 23 with CA or vehicle control, and the pups were sacrificed on postnatal days 14, 18, 22, and 28, respectively. Figure ​Figure33 shows data from the ileum, but all three regions of the small intestine displayed similar changes. The sensitivity of sucrase activity to glucocorticoid-mediated precocious induction was lost by day 17, whereas the sensitivity of fut2 induction to CA persisted until day 22. Changes in enzyme activity were accompanied by parallel changes in fut2 and SI mRNA expression, consistent with the cortisone effect on enzyme activity occurring at the level of transcription. The changes in fucosyltransferase activity coincided with even stronger parallel changes in fut2 mRNA expression, but no changes were evident in expression of fut1, fut4, or fut7. These data are consistent with the changes in intestinal fucosylation being specifically mediated by changes in fut2 gene expression. The changes in sensitivity for SI coincide with weaning and for fut2 with the rapid change in the intestinal microflora that follows weaning.

Open in a separate window

Fig. 3.

Loss of glucocorticoid responsiveness of α1,2/3-fucosyltransferase activity (nmol/mg protein/h); and sucrase activity (µmol/mg protein/h); and fut2 mRNA and SI mRNA in the developing mice ileum. Six litters of pups were used; on appropriate post-natal ages, pups were treated either with CA or vehicle (VEH; saline) and subsequently ileum was harvested and assayed. Each bar represents the mean ± SE of 3-4 samples. Each sample was pooled from six mice, *P < 0.05 vs control (VEH) group.

The role of intestinal microbes per se on the ontogeny of intestinal fut2 at weaning was investigated. To differentiate the effects of any programmed changes in the internal milieu at weaning from the concurrent changes in the external environment, diet, and colonization, post-natal development in conventionally colonized mice was compared with that of germ-free mice. Distinct from the intestinal increase in fut2 during development observed in conventionally colonized mice (Figure ​(Figure1),1), germ-free mice did not display age-related changes in fucosyltransferase activity through week 6, except for a modest increase in the colon (Figure ​(Figure4).4). FucT II activity remained at the low levels typical of the immature mucosal phenotype, suggesting that luminal microbes might play an essential role in the normal induction of intestinal fut2 gene expression. To directly test this hypothesis, germ-free mice were inoculated with microflora from conventionally colonized 6-week-old mice (ex-germ-free adult) and sacrificed 2 weeks after inoculation. Upon the introduction of luminal bacteria from aged-matched conventionally colonized mice, the levels of fucosyltransferase activity increased to that of conventionally colonized mice in the duodenum (P < 0.01), jejunum (P < 0.01), ileum (P < 0.001), and colon (P < 0.001) (Figure ​(Figure4).4). In contrast, sucrase enzyme activities were similar in conventionally colonized, germ-free, and ex-germ-free mice in all regions of the small intestine, whether inoculated at 4 (Figure ​(Figure4)4) or 6 (not shown) weeks of age. The next experiment was to determine if the microbiota of the suckling mouse is capable of inducing adult-type fucosylation. Bacteria-depleted mice were inoculated with the suckling microbiota isolated from a 7-day-old pup. Two weeks after inoculation, the microbiota of the adult mouse (P < 0.001), but not the microbiota of the suckling mouse (P > 0.8), was able to induce adult-type fucosylation in all segments of the small and large intestine (data not shown).

Open in a separate window

Fig. 4.

Post-natal development of α1,2/3-fucosyltransferase activity (nmol/mg protein/h) in mouse (A) duodenum, (B) jejunum, (C) ileum and (D) colon of the germ-free (GF), conventional (CONV) and newly colonized germ-free (XGF) mice. Each bar represents the mean ± SE of three samples. Tissues were pooled from six to eight mice per sample for 1- and 2-week-old mice and at three tissues per sample for 4- and 6-week-old mice. **P < 0.001 vs CONV mice.

Confirming these results, levels of fut2 mRNA (Figure ​(Figure5)5) remained low in germ-free mice until inoculation with adult mouse microbiota, whereupon the fut2 mRNA levels rose to those of conventionally colonized mice. In stark contrast, SI mRNA levels remained unchanged by colonization (Figure ​(Figure5H).5H). Thus, changes in intestinal FucT II activity are controlled by the induction of fut2 mRNA by adult microbiota. No such changes were evident for the constitutive low levels of fut1 mRNA (Figure ​(Figure5E),5E), fut4 mRNA (Figure ​(Figure5F),5F), or fut7 mRNA (Figure ​(Figure55G).

Open in a separate window

Fig. 5.

Post-natal development of fut2 mRNA in the mouse (A) duodenum, (B) jejunum, (C) ileum and (D) colon of germ-free (GF) and conventional (CONV) mice. Each bar represents the mean ± SE of three samples. Tissues were pooled from six to eight mice per sample for 1- and 2-week-old mice and at three tissues per sample for 4- and 6-week-old mice. **P < 0.001 vs CONV mice.

To determine the interaction between CA and colonization by microbiota in immature mice, the response of conventional mice to CA treatment was compared with that of germ-free mice (Figure ​(Figure6).6). On day 10, conventional and germ-free mice were treated either with CA or vehicle control, and on day 14, the levels of SI and fut2 were assessed in each region of the intestine. As before, CA treatment precociously induced both sucrase and fucosyltransferase activities (P < 0.001), and their mRNA levels (P < 0.01), in conventional mice. However, only sucrase activity (P < 0.001) and mRNA (P < 0.01) were induced by CA in germ-free mice. The fucosyltransferase activity and fut2 mRNA levels did not respond to CA and remained at the suckling level. This implies that signaling to the intestinal mucosa by the microbiota may be required for cortisone to induce precocious fucosylation of the gut mucosa.

Open in a separate window

Fig. 6.

Precocious induction of α1,2/3-fucosyltransferase activity (nmol/mg protein/h) and sucrase activity (µmol/mg protein/h) as well as the fut2 mRNA and SI mRNA in mouse ileum of germ-free (GF) and conventional (CONV) mice treated with vehicle control or CA. Each bar represents the mean ± SE of three samples. Tissues were pooled from six to eight mice per sample for 1- and 2-week-old mice and at three tissues per sample for 4- and 6-week-old mice. **P < 0.001 vs CONV mice.

Expression of α1,2-fucosylated-glycans was visualized in frozen sections by the α1,2-fucose-specific lectin, UEA-I, conjugated with Texas Red. Identical sections in which 100 mM l-fucose accompanied the Texas Red-UEA-1 during staining did not display any binding, indicating that the UEA-1 binding was specific to fucosylated moieties. Terminal α1,2-fucose-containing glycans were diminished in the colon of bacteria-depleted (Figure ​(Figure7A)7A) and germ-free (not shown) mice; however, reintroduction of the microbiota, or one of its major bacteria (B. fragilis), induced recovery of α1,2-fucosylated-glycans to the higher levels found in conventionally colonized mice. The changes in α1,2-fucosylated glycoconjugate expression were prominent on the cell surface, but also were apparent in the mucus of the goblet cells.

Open in a separate window

Fig. 7.

(A) UEA-1 lectin staining of α1,2-fucosylted-conjugates in adult colonic tissue of conventional (CONV), bacteria depleted (BD) and ex-bacteria depleted (XBD) mice. The lack of fucosylation due to the lack of bacteria (BD) is reversed by recolonization (XBD) or by colonization by B. fragilis, a dominant bacteria of the microbiome. (B) Fucosyltransferase expression is lost with the loss of bacteria (BD), but is restored upon recolonization (XBD); fucosyltransferase expression is also restored by colonization with one of the dominant fucose utilizers of the adult microbiome, B. fragilis^WT, but not by the same bacterium with a mutation that deprives it of the ability to utilize fucose, B. frag^Mut. Thus, B. fragilis induces expression of the fucosylated phenotype (seen in panel A) through the induction of fucosyltransferase activity. (C) Recovery from DSS-induced mucosal injury occurs in mice that are colonized (CONV), recolonized (XBD) or colonized by wild-type B. fragilis, but not in mice that lack colonization (BD), or are colonized by the mutant B. fragilis (B. frag^Mut) and hence lack fucosylation. These data link the expression of the fucosylated phenotype with intestinal homeostasis, manifest as mucosal recovery from injury.

To determine whether the loss of the fucosylated phenotype of bacterially depleted mice was mediated by the loss of fucosyltransferase expression, fucosyltransferase activity and fut2 mRNA levels were measured in these tissues. Adult levels of fucosyltransferase activity and fut2 mRNA are lost with lack of bacterial colonization (BD; Figure ​Figure7B),7B), but are restored upon recolonization (XBD); they are also restored by colonization with one of the dominant fucose utilizing bacteria of the adult microbiome, B. fragilis^WT, but not by the same bacterium with a mutation that deprives it of the ability to utilize fucose, B. frag^Mut. Thus, B. fragilis induces the expression of the fucosylated phenotype through the induction of fucosyltransferase activity (Figure ​(Figure77B).

The functional consequence of this immature, low fucosylation in bacteria-depleted mouse gut was investigated through the induction of mucosal injury by the chemical irritant, DSS. Recovery from DSS colonic inflammation was compared between conventionally colonized and bacteria-depleted mice. Conventionally colonized mice fully recover from the mucosal injury over 2 weeks, but in the absence of colonization (bacteria-depleted mice), recovery does not occur (P < 0.01). Recovery from DSS-induced mucosal injury occurs in mice that are colonized (CONV), recolonized (XBD), or colonized by wild-type B. fragilis. Recovery does not occur in mice that lack fucosylation due to a lack of colonization or are colonized by the mutant B. fragilis (B. frag^Mut; Figure ​Figure7C),7C), which is unable to utilize fucose or induce mucosal fucosylation. These data link the expression of the fucosylated phenotype with intestinal homeostasis involved in mucosal recovery from injury. Recovery from the colonic damage induced by TNBS and OXA also required bacteria-dependent fucosylation (not shown). These results strongly suggest that fucosylation is dependent on colonization, and that proper fut2-dependent FucT II fucosylation of the intestinal mucosa is required to maintain tissue homeostasis and recovery from inflammatory chronic injury.

Discussion

Many intrinsic signals mediate and modulate the genetic program of development. Glucocorticoids are common intrinsic signals that mediate genetically programmed milestones of tissue development, including mammary gland development at puberty (Cunha and Hom 1996) and lactogenesis (Convey 1974), early lung maturation (Farrell 1977; Henning 1987), and post-natal development of the intestine (Henning 1987). Thus, the glucocorticoid surge that accompanies parturition (Farrell 1977; Henning 1987) mediates the rapid maturation of the lung needed for the immediate transition to breathing, the maturation of digestive function for the transition to obtaining nutrients directly from food (Henning 1987), preferably milk, and the modulation of the mucosal immune function (Rock 1999) that permits colonization without excess inflammation. Similarly, the glucocorticoid surge at weaning mediates intestinal adaptation to solid foods; it is modeled in the laboratory either as exogenous administration of steroids or as elicited surges in endogenous circulating glucocorticoid, and precociously induces the suckling to adult transition of the gut (Henning 1987; Henning and Nanthakumar 1994; Montgomery et al. 1999).

In the experiments herein, the initial definition of the mouse gut model involved measuring levels of mucosal fucosylation of the pre- and post-weaning small and large intestine, and defining which of the fucosyltransferase genes was responsible for any differences. All changes were compared with the changes in SI expression, a standard marker of maturation of developing intestine. As expected, glucocorticoids accelerated the maturation of gut, accompanied by precocious expression of both SI and fucosyltransferase activity. Of fucosyltransferase genes 1, 2, 4, and 7, only the gene responsible for fucosyltransferase II expression (fut2) was specifically induced by glucocorticoids (Figure ​(Figure2).2). This accelerated expression by glucocorticoids could only occur in gut prior to weaning, consistent with glucocorticoids being mediators of specific maturational events in immature gut. In the mature post-weaning gut, the glucocorticoid had no further effect, as the milestones of maturation had already been achieved in the course of normal development (Figure ​(Figure33).

Likewise, in the germ-free animal, SI is only influenced by glucocorticoids before weaning (Figure ​(Figure3);3); glucocorticoids after day 22 have no further effect. However, in the germ-free animal, fucosylation is not induced upon weaning, and the immature state of fucosyltransferase gene expression is maintained despite any surge of glucocorticoids at weaning. This is consistent with the lack of effect by exogenous glucocorticoids on fucosyltranferase expression observed in the conventionally colonized mouse after day 28 (Figure ​(Figure3).3). These findings suggest a distinct type of developmental control of fucosyltransferase, and the difference between the germ-free and conventionally colonized mouse points to bacterial colonization as a prime candidate.

In the absence of a microbiota, the germ-free mouse gut retains its pre-weaning state of high sialyltransferase and low fucosyltransferase; with the introduction of the adult microbiota, the gut glycosylation rapidly shifts to high fucosyltransferase. This mimics the change at weaning for a normal conventionally colonized control. This is consistent with the primary control for the qualitative shift in gut glycosylation being signaling by the microbiota. Consistent with this explanation is that the adult microbiota, but not the distinctly different microbiota of the pre-weaning mouse, has the ability to induce fucosyltransferase activity in mature gut (Figure ​(Figure4).4). Moreover, the colonization of germ-free mature gut with only one of the major mutualist symbionts of adult gut, B. fragilis, is sufficient to induce the adult pattern of fusosyltransferase expression resulting in the typical highly fucosylated mucosal surface (Figure ​(Figure7).7). Moreover, the B. fragilis mutant unable to utilize fucose is unable to restore this fucosylation. Consistent with these data, SI is expressed only in the small intestine, the site of terminal carbohydrate digestion and absorption, whereas fucosyltransferase is expressed most heavily in the distal alimentary tract, the site of heaviest bacterial colonization. The developmental induction of fut2 mRNA expression is accompanied by the induction of α2,3-sialyltransferase and β1,4 galactosyltransferase mRNA expression, but these are induced to a lesser extent and may represent the adaptation toward synthesizing the glycan scaffolds on the surface of the intestinal mucosa to which the fucose must be attached. In contrast to the α2,3 sialyltransferase, α2,6 sialyltransferase mRNA expression displayed a robust reciprocal decrease, consistent with the reduced sialylated phenotype in colonized mature intestinal mucosa.

Fucosylation of the intestinal mucosa is stimulated by the colonization of adult colon by adult microbiota. Mice that lack colonization, either by lack of inoculation by microbiota (germ-free mice) or by oral antibiotic treatment, lack the fucosylated mucosal phenotype. However, the colonization of germ-free mice and the recolonization of bacteria-depleted mice results in the expression of fucosylated glycans on the surface of the intestinal mucosa. This induction of fucosylation depends upon the activation of the ERK and JNK signal transduction pathways, which result in specific transcription of fut2 mRNA, induction of fucosyltransferase activity, and the fucosylation of mucosal glycoconjugates (Meng et al. 2007). Inhibition of ERK and JNK activation in mucosa during colonization (germ-free) or recolonization (bacteria-depleted) prevents fut2 mRNA induction and subsequent mucosal fucosylation. This suggests that constant communication between the microbiota and the epithelium is necessary for sustained expression of fut2 mRNA transcription in the adult intestinal mucosa. Fut1, fut4 and fut7 did not display any such relationship with the microbiota, consistent with a highly specific relationship between the microbiota and fut2. Therefore, colonization by adult microbiota may be the primary regulator of mucosal fucosylation in the intestinal mucosa and, in turn, fucosylation of the mucosa may provide the niche for maintaining the stable microbial ecosystem in the adult gut. The interdependence of fucosyltransferase expression and the microbiota is consistent with the complex interkingdom communication typical of mutualist symbionts.

Mutualism includes both reciprocal communication and reciprocal benefit to both the microbiota and the host. Specific features of mutualism include the prokaryote microbe-inducing production of molecules in the eukaryote host thereby producing a niche that facilitates one side of the specific mutualistic relationship. The other side of the relationship is the benefit to the eukaryotic host. Experiments were designed to investigate the possibility that proper colonization would protect the host intestinal epithelium from damage or facilitate recovery from injury. Chemical irritants were used to cause mucosal injury under conditions that included or excluded the mutualist symbiotic relationship, and the consequence of colonization was measured. Normally colonized mice fully recover from the colitis caused by DSS or other chemical irritants (TNBS and OXA). In mice whose microbiome was greatly reduced, 40% did not recover and died. However, if the microbiota was replaced before treatment, recovery was complete (Figure ​(Figure7C).7C). It is noteworthy that fut2 expression was restored even when the mice were recolonized with only one major bacterium of the normal adult microbiota (B. fragilis, Figure ​Figure7A7A and B), and this recolonization also supported recovery from mucosal injury. Recolonization with a mutant B. fragilis unable to utilize fucose did not induce fut2 expression (Figure ​(Figure7B),7B), and was unable to support recovery from mucosal injury (Figure ​(Figure7C).7C). This suggests that bacteria-induced fucosylation is essential to cause a change in the internal state of the intestinal epithelial cell, rather than the microbiota per se, that is necessary for this recovery. The treatment with B. fragilis induced the same fucosylation through induction of fut2 expression as recolonization with the entire microbiota. This implies that some microbes of the microbiota, especially predominant species like B. fragilis, might be the critical triggers of fucosylation during recolonization. Preliminary data indicate that monocolonization by any of several Gram-negative components of the adult microbiota can induce fucosylation, suggesting that fut2 induction may be a response of uncolonized mature gut to Gram-negative prototypic molecules. The definition of the specific trigger and its ligand is ongoing and beyond the scope of this report.

These results strongly suggest that fucosylation is dependent on colonization and that proper fucosylation of the intestinal mucosa is required to maintain tissue homeostasis and recovery from inflammatory chronic injury. This conclusion is also consistent with previous observations in genetically reconstructed mice. FX mutant mice (FX^−/−) are unable to generate fucosylated glycans. They are born normal but shortly after weaning they die from colitis and diarrhea (Smith et al. 2002). Onset of colitis in FX^−/− mice can be prevented by maintaining them in germ-free conditions or by feeding them fucose. In light of the DSS data, the fucose can be seen as stimulating the colonization of their gut with fucose-dependent mutualists that can then protect the fucose-deficient mutant mice. Taken together, these data indicate a central and essential role of intestinal fucosylation induced by the gut microbiota for maintaining homeostasis in the host as well as mediating the mutualistic relationship between the adult gut microbiome and the adult gut.

Human populations contain a high frequency of a mutation in the FUT2 gene that renders the gene product inactive. In most populations of European, Asian and African descent, ~20% are homozygous recessive for the FUT2 allele and are known as non-secretors, because their exocrine secretions (milk, mucus, saliva etc.) lack the α1,2-fucosylated epitope; their intestinal mucosa also lacks this moiety. Thus, these individuals cannot respond to intestinal colonization with an induction of the gut α1,2-fucosylated phenotype. The colonization of non-secretors differs from that of secretors (data not shown), and non-secretors are predisposed to inflammatory bowel diseases such as necrotizing enterocolitis (Morrow et al. 2011) and Crohn's disease (McGovern et al. 2010). Thus, the data from our DSS colitis model, in conjunction with these clinical data, suggest that bacterially induced fucosylation can be of fundamental importance to human health. Conversely, polymorphisms in FUT2 that result in low levels of fucosylation of secretions and intestinal mucosa are associated with dysbiosis of the gut microbiota (unpublished data). In clinical studies, dysbiosis of the human microbiota is implicated in various severe inflammatory conditions, such as necrotizing enterocolitis (Morrow et al. 2011), Crohn's disease (McGovern et al. 2010) and type 1 diabetes (Smyth et al. 2011; Morrow et al. 2013).

The negative consequences of the non-secretor phenotype raise the question of why this trait persists in the population. This can be considered a subset of why there is so much diversity in the human glycome. Many pathogens bind to specific glycoconjugate receptors. Humans are social beings. Hence, if an emerging human pathogen arises with high lethality, human populations would be at risk of extinction unless a portion of the population did not express the specific receptor for the lethal pathogen and hence could escape infection and ensure their survival. This could account for the extreme diversity of cell surface glycan expression in humans, and the survival of traits that otherwise have negative consequences.

The intestinal mucosa is predominantly sialylated before weaning and fucosylated after weaning. The ability of adult, but not suckling microbiota to stimulate fucosylation provides the basis for hypothesizing two distinct phases of mutualism. The adult fucose-centered mutualism is described above. It is interesting to speculate that the predominance of sialylation from birth to weaning might be supporting colonization by pioneering sialic acid-dependent species of our microbiota.

The maturation of the intestine, its colonization and its adaptation to various diets is increasingly recognized as being central to human health and wellbeing. Conversely, intestinal dysbiosis contributes to a variety of human diseases and chronic conditions in both infants and adults. Advances in metagenomic analysis now allow characterization of microbiome in normal and disease states, and distinct differences have been reported. The dearth of information on the major factors that direct early gut colonization by the microbiome, which include genetic variation of the host, and intrinsic and extrinsic signaling, have been limiting intervention strategies. Deeper knowledge of the interdependence and integration of factors that contribute to initiation and succession of a healthy microbiome and how dysbiosis leads to immunologic and other pathologies provides an attractive target for new rational therapies for diseases that are currently intractable.

Materials and methods

Funding

This work was supported by the National Institutes of Health (HD059126; to NN, HD059140; to DSN, HD013021; to DSN, AI075563; to DSN and HD061930 to DSN).

Conflict of interest

None declared.

Abbreviations

BSA, bovine serum albumin; CA, cortisone acetate; DSS, dextran sulfate sodium; GAPDH, glyceraldehyde 3-phosphate dehydrogenase; GF, germ-free; OXA, oxazalone; OCT, optimal cutting temperature; PBS, phosphate-buffered saline; PCR, polymerase chain reaction; qRT, quantitative reverse transcription; SI, sucrase-isomaltase; TNBS, 2,4,6-trinitrobenzene sulfonic acid; UEA-I, Ulex europaeus agglutinin-1 FucT, fucosyltransferase.

Acknowledgements

References