CX3CR1-Mediated Dendritic Cell Access to the Intestinal Lumen and Bacterial Clearance

See allHide authors and affiliations

Science  14 Jan 2005:
Vol. 307, Issue 5707, pp. 254-258
DOI: 10.1126/science.1102901


Dendritic cells (DCs) and macrophages are critical to innate and adaptive immunity to the intestinal bacterial microbiota. Here, we identify a myeloid-derived mucosal DC in mice, which populates the entire lamina propria of the small intestine. Lamina propria DCs were found to depend on the chemokine receptor CX3CR1 to form transepithelial dendrites, which enable the cells to directly sample luminal antigens. CX3CR1 was also found to control the clearance of entero-invasive pathogens by DCs. Thus, CX3CR1-dependent processes, which control host interactions of specialized DCs with commensal and pathogenic bacteria, may regulate immunological tolerance and inflammation.

Antigen-presenting dendritic cells (DCs) play a crucial role in establishing immunity to pathogens, tolerance to self-antigens, and the induction of organ-specific autoimmunity (15). Through their interactions with antigen-specific T lymphocytes, DCs induce adaptive immune responses and are also critical in controlling tissue inflammation. At mucosal interfaces, DCs constantly survey and process commensal bacteria and pathogens as a first step in the development of immunity, and recent studies have suggested that distinct DC subsets achieve this through diverse means (68). In addition to acquiring antigens in the lamina propria, intestinal DCs located below specialized intestinal epithelial cells (IECs), called M cells, also detect incoming pathogens. Furthermore, intestinal DCs transport bacteria and capture proteins from the gut lumen (9). Recent evidence has suggested that intestinal DCs also directly monitor the content of the intestinal lumen by either entering or extending dendrites into the epithelium (10, 11). However, both the molecular basis and the physiological relevance of this process remain unclear.

To characterize the intestinal mononuclear phagocytes that are responsible for the recognition of intestinal pathogens, we used mice in which one or both copies of the gene encoding the chemokine receptor CX3CR1 (cx3cr1) was replaced with green fluorescent protein (GFP) reporter cDNA (12, 13). In these mice, GFP expression is under the control of the CX3CR1 promoter and, consequently, heterozygous mice (cx3cr1GFP/+) express both the receptor and GFP, whereas homozygous mice (cx3cr1GFP/GFP) express GFP but are CX3CR1-deficient. As a result, GFP expression can be used to identify, isolate, and characterize intestinal cell populations that express the chemokine receptor. These cell populations are targets for CX3CL1/fractalkine, a transmembrane chemokine expressed at the surface of IECs and endothelial cells in the intestine (14, 15). In earlier studies, we found GFP and CX3CR1 expressed in circulating monocytes and their progeny within tissues (16). Although CX3CR1 was found to contribute to the migration of blood monocytes into tissues, no phenotypic or functional differences were detected in the peripheral lymphoid system between heterozygous and wild-type mice (16).

Examination of gut-associated tissues from cx3cr1GFP/+ and cx3cr1GFP/GFP mice by fluorescence microscopy revealed GFP-positive mononuclear cells in the lamina propria, Peyer's patches (PPs), and mesenteric lymph nodes (MLNs) (Fig. 1A and fig. S1). Upon isolation, lamina propria leukocytes from cx3cr1GFP/+ and cx3cr1GFP/GFP mice were found to represent different developmental stages of immature and mature DCs, which expressed the proteins CD11c and CD11b (Fig. 1B and fig. S1). Lamina propria DCs from cx3cr1GFP/+ and cx3cr1GFP/GFP mice also expressed major histocompatibility complex (MHC) class II, CD80, and CD86 proteins at levels that were comparable to those found in DCs isolated from wild-type mice (fig. S1). Lamina propria CX3CR1-positive DCs lacked CCR6 expression, in contrast to DC subsets that had been associated with PPs (1720). Expression of the CX3CR1 ligand CX3CL1 was at its highest level in IECs that were isolated from the terminal ileum, where the ligand was detected in basolateral membrane compartments (Fig. 1, C and D).

Fig. 1.

(A) Fluorescence imaging of leukocytes in the lamina propria of the small intestine of cx3cr1+/+, cx3cr1GFP/+, and cx3cr1GFP/GFP mice, and in PPs and MLNs of cx3cr1GFP/+ mice. (B) Flow cytometry of lamina propria leukocytes from cx3cr1GFP/+ and cx3cr1GFP/GFP mice with indicated antibodies. (C) Immunostaining of CX3CL1 in the terminal (ter) ileum and jejunum of wild-type cx3cr1+/+ mice. (D) Immunoblot analysis of CX CL1 expression in membrane (P) and cytosolic (S) protein fractions from isolated IECs of the indicated small intestinal regions. L, lumen; L.P. lamina propria; N, nuclei. Arrows indicate the basolateral compartment of IECs.

Living intestinal and lymphoid tissues were examined using confocal microscopy to determine the distribution and morphology of intestinal DCs and their interaction with IECs. In cx3cr1GFP/+ mice, lamina propria DCs extended dendrites into and through the intestinal epithelium (Fig. 2, A to D). However, these transepithelial dendrites were only observed in villi of the terminal ileum and were absent in the duodenum, jejunum, proximal ileum, and mid-ileum of the same mice. In contrast, ileal villi in cx3cr1GFP/GFP mice lacked these intraepithelial DC extensions, suggesting that their formation is dependent on a CX3CR1-mediated interaction with IECs (Fig. 2, E to H). In cx3cr1GFP/+ mice, 1.74 ± 0.2 intraepithelial dendrites were observed per villus, but only 0.07 ± 0.02 were observed per villus in the cx3cr1GFP/GFPmice (n = 40 villi from each of three mice). Nevertheless, DCs in the lamina propria of cx3cr1GFP/GFP mice retained the ability to form dendrites, although these were confined to the lamina propria, which is consistent with their impaired capacity to traverse the epithelial cell monolayer (Fig. 2, G and H).

Fig. 2.

(A to D) Confocal microscopic analysis of ileal mucosa of cx3cr1GFP/+ and (E to H) cx3cr1GFP/GFP mice. (B) 3D tissue reconstruction of small intestinal villi from cx3cr1GFP/+ and (F) cx3cr1GFP/GFP mice. In (C), (D), (G), and (H) the epithelium was counterstained with orange CMTMR [5-(and-6)-(((4-chloromethyl)benzoyl)amino)tetramethylrhodamine]. (I) 3D analysis of a villus or (J) of the apical part of a villus in cx3cr1GFP/+ mice. (K) 3D analysis of a villus in cx cr1GFP/+ mice after staining of the microvascular system with Texas Red Dextran. (L) Model representation of the morphologic relationship of DCs, microvasculature, and IECs. (M and N) 3D analysis of a villus in cx3cr1GFP/+ mice after oral administration of DsRed2-expressing E. coli. The GFP signal from CX3CR1-positive DCs was removed in (N) to reveal the presence of E. coli. (O) Merged dual- or (P) single-channel confocal microscopy of a PP in cx3cr1GFP/+ mice after uptake of DsRed2-expressing E. coli. BM, basal membrane; Lu, lumen; LP, lamina propria; TJ, tight junctions; CV, capillary vessels. Asterisks indicate luminal dendritic cell compartments. Arrow heads indicate transepithilial dendrites in (A), (B), and (M). Scale bars, 10 μm.

The complex structure of transepithelial processes was revealed by three-dimensional (3D) tissue reconstructions of intestinal villi. Intraepithelial dendrites were seen to originate in the lamina propria, either from the DCs that directly underlie the epithelium (Fig. 2, D and I) or from the DCs that are separated from the epithelium by the capillary blood vessel system (Fig. 2K). Intraepithelial dendrites extended through the IEC monolayer to end in mono- or multiglobular structures outside the epithelium (Fig. 2, I and J). The diameter of dendrites was reduced to 0.2 to 0.5 μm in regions where they crossed the basal membrane and the epithelial tight junction, possibly as a means of limiting disturbances to intestinal barrier function. This analysis suggests that the interaction of CX3CL1 expressed on IECs with CX3CR1 on intestinal DCs has a critical role in transepithelial dendrite formation.

Because the extension of transepithelial dendrites has been shown to be involved in luminal antigen sampling (11), we examined the uptake of commensal and enteropathogenic bacteria by lamina propria DCs in cx3cr1GFP/+ and cx3cr1GFP/GFP mice. In cx3cr1GFP/+ mice, DsRed2-labeled nonpathogenic Escherichia coli were observed in DCs that extended dendrites into the IEC layer and in DCs within interfolicular regions (IFRs). DsRed2-expressing E. coli could be cultured from isolated PPs and MLNs of these mice (Fig. 2 and Table 1). In contrast, the uptake of E. coli into the lamina propria was impaired in the cx3cr1GFP/GFP mice, and E. coli could not be cultured from isolated MLNs of these mice (Table 1). However, equal numbers of E. coli were recovered from PPs of cx3cr1GFP/GFP, cx3cr1GFP/+, and cx3cr1+/+ mice, indicating that bacterial sampling by M cells remained intact in these animals. The sampling defect in cx3cr1GFP/GFP mice may thus specifically affect lamina propria DCs that are responsible for the transport of commensal bacteria to the MLNs.

Table 1.

CX3CR1-dependent uptake of nonpathogenic E. coli into intestine-associated lymphoid tissues. The number of colony-forming units (CFUs) of DsRed2-expressing E. coli F-18 in MLNs and PPs of cx3cr1+/+, cx3cr1GFP/+, and cx3cr1GFP/GFP mice (n = 5 mice in each group) is shown 18 hours after oral administration.

Mice Mesenteric lymph node (CFU per g of tissue) Peyer's patch (CFU per g of tissue)
cx3cr1+/+ 912 ± 256 97,444 ± 53,690
cx3cr1GFP/+ 1,664 ± 1,300 102,221 ± 68,380
cx3cr1GFP/GFP None 79,377 ± 76,076

To determine whether bacterial sampling and transport have consequences for the ability of cx3cr1GFP/GFP mice to cope with bacterial infection, we examined mice infected with the enteroinvasive pathogen Salmonella typhimurium. Dendrite formation in cx3cr1GFP/+ mice was enhanced (3.8 ± 1.1 dendrites per villus, n = 20 villi from four mice in response to S. typhimurium infection, whereas the number of transepithelial processes did not increase in infected cx3cr1GFP/GFP mice (0.08 ± 0.01 dendrites per villus, n = 20 villi from four mice) (Fig. 3A, panels a to d). In response to S. typhimurium infection, DCs in cx3cr1GFP/GFP mice formed only globular structures at the basal surface of IECs, which failed to cross the epithelium (Fig. 3A, panel c). An invasion-deficient mutant of S. typhimurium was observed in DCs that extended transepithelial dendrites in the terminal small intestine 12 hours after oral administration in cx3cr1GFP/+ but not in cx3cr1GFP/GFP mice (Fig. 3A, panels e to h). In contrast, invasive S. typhimurium traversed the intestinal epithelium in the cx3cr1GFP/GFP mice independently of dendrite formation, and were subsequently phagocytosed by lamina propria DCs (Fig. 3A, panels c and d). Uptake of S. typhimurium by lamina propria DCs was also observed in MHC class II–GFP transgenic mice (fig. S2). In cx3cr1GFP/+ mice, invasive S. typhimurium were found confined to CX3CR1/GFP–positive DCs in the IFRs and MLNs. In contrast, large numbers of bacteria were observed in a GFP-negative phagocyte subset in the cx3cr1GFP/GFP mice (Fig. 3, B and C). This suggests that in the absence of CX3CR1 expression, an additional lamina propria phagocyte subset may have been recruited to defend against invasive S. typhimurium but was unable to compensate for the loss of DC function in the cx3cr1GFP/GFP mice. The number of invasive and noninvasive S. typhimurium that were recovered from CX3CR1-deficient lamina propria DCs after in vitro uptake was reduced when compared to DCs from cx3cr1GFP+ and wild-type control mice (Fig. 4, A and B). These results suggest that CX3CR1 is required for the control of the uptake of S. typhimurium by DCs within the intestinal mucosa. Most likely as a consequence of impaired bacterial sampling by DCs, cx3cr1GFP/GFP mice displayed enhanced susceptibility to S. typhimurium infection. Thus, they succumbed within 6 days of oral administration of 1 × 109 bacteria and displayed significantly higher bacterial loads in their organs, compared with the relative resistance of their cx3cr1GFP/+ and wild-type counterparts (Fig. 4, C and D). Delayed pathogen uptake by DCs from the lumen and the lamina propria may thus form the mechanistic foundation for the impaired antibacterial defense in CX3CR1-deficient mice.

Fig. 3.

(A) Confocal microscopic analysis of living small intestinal mucosa after oral administration of DsRed2-expressing invasive and noninvasive S. typhimurium in cx3cr1GFP/+ and cx3cr1GFP/GFP mice. Arrow heads indicate transepithelial dendrites. (B) Confocal microscopic analysis of IFRs and (C) MLNs in cx cr1GFP/+ and cx3cr1GFP/GFP mice after oral administration of DsRed2-expressing invasive S. typhimurium. Nuclei of CX3CR1/GFP-negative phagocytes containing S. typhimurium are indicated by asterisks.

Fig. 4.

(A and B) In vitro uptake of invasive and noninvasive S. typhimurium by lamina propria DCs after isolation from cx3cr1GFP/+, cx3cr1GFP/GFP, and cx3cr1+/+mice (**P < 0.001, cx3cr1GFP/+ DCs compared to cx3cr1GFP/GFP DCs). (C) Survival of indicated mouse strains after oral infection with 109 S. typhimurium. (D) Organ load with S. typhimurium in cx3cr1GFP/+ and cx cr1GFP/GFP mice and their cx cr1+/+ littermates (*P < 0.001, cx3cr1GFP/+ mice compared to cx3cr1GFP/GFP mice).

We have demonstrated an extensive intestinal DC network that serves as a gateway for the uptake and transport of the intestinal microbiota. We have identified and characterized CX3CR1-positive lamina propria DCs, a major component of this system, which are capable of taking up bacteria by way of transepithelial dendrites in order to provide defense against pathogenic microorganisms. CX3CR1 deficiency results in a defect of lamina propria DCs that impairs the sampling of bacteria from the intestinal lumen and impedes their ability to take up invasive pathogens in vitro. The intrinsic functional programs and subspecifications of CX3CR1-positive and -negative DCs in mucosal innate and adaptive immune responses will need to be further defined. Nevertheless, CX3CR1-dependent regulation of DCs appears to provide a central mechanism for the control of the mucosal defense against entero-invasive bacteria.

The interaction of DCs with the intestinal microbiota by the way of CX3CR1-dependent transepithelial dendrites could activate an innate immune pathway that protects the mucosa from pathogenic bacteria. The formation of these dendrites may be linked to the immature phenotype of lamina propria DCs and associated with their phagocytic function. Their use thus constitutes a mechanism by which DCs could take up intestinal antigens, which is distinct from previously established systems involving M cells. We propose that the CX3CR1-dependent and the M cell–dependent systems could thus be associated with specific DC subsets. It will be important to determine whether such networks operate synergistically as redundant systems or if they have distinct functions in the recognition of commensal and pathogenic bacteria. Luminal sampling by CX3CR1-positive DCs occurs by globular structures formed at the end of transepithelial dendrites, which could serve as luminal sensors for the mucosal immune system to continually monitor intestinal content. Characterization of the surface components that facilitate antigen uptake through this specialized cellular compartment may aid in developing strategies to prevent bacterial and viral pathogens from co-opting this route during infection. Furthermore, targeting of antigens to transepithelial dendrites could be used to directly engage the function of intestinal CX3CR1-positive DCs in vaccine development.

Supporting Online Material

Materials and Methods

Figs. S1 and S2


Movies S1 to S4

References and Notes

View Abstract

Navigate This Article