MyD88 Signaling Regulates Steady-State Migration of Intestinal CD103⁺ Dendritic Cells Independently of TNF-α and the Gut Microbiota

The gastrointestinal mucosa represents the body’s largest surface toward the external environment and is an important entry site for a large variety of enteric pathogens. The intestinal immune system must protect against these harmful microorganisms, but at the same time maintain tolerance to commensal bacteria and food Ags continuously present in the gut lumen. Dendritic cells (DC) represent a heterogeneous family of innate leukocytes with an exceptional ability to prime adaptive immune responses and are thought to play an important role in balancing adaptive immunity versus tolerance in the intestinal mucosa (1). Conventional DCs derive from committed bone marrow (BM) precursors that migrate out in the periphery to take up residence in either secondary lymphoid organs (SLO) or nonlymphoid tissues (2). Although DCs that directly enter SLO via high endothelial venules localize to the T cell zone in close proximity to the Ag-transporting conduit system (3), the nonlymphoid tissue DCs internalize Ags from their local environment and then migrate via afferent lymphatics into draining lymph nodes (LN) (4–6). In both the absence (i.e., in the steady state) and presence of overt inflammation, the mobilization of tissue-resident DCs into draining LN is governed by the chemokine receptor CCR7 that confers responsiveness to the ligand CCL21 and CCL19 expressed by lymphatic endothelial cells and LN stromal cells, respectively (7, 8).

We and others have identified a functional specialization of DCs in the gut-draining mesenteric LN (MLN) that express the integrin α-chain CD103 (9, 10). These DCs display an enhanced ability to induce the gut-homing molecules CCR9 and α₄β₇ on T cells and, in the presence of TGF-β, support the differentiation of Foxp3⁺ T regulatory (Treg) cells in vitro through mechanisms involving the vitamin A metabolite retinoic acid (RA) (9, 11, 12). The large majority of MLN CD103⁺ DCs appears to represent small intestinal (SI) lamina propria (LP)–derived migratory DCs as they are selectively reduced in MLN, but not in the SI LP, of CCR7-deficient mice (9, 13) and in BrdU pulse-chase experiments accumulate with delayed kinetics in the MLN compared with the SI LP (14). Consistent with this, the majority of DCs in thoracic duct lymph collected from mice after mesenteric lymphadenectomy express CD103 (15). However, a small proportion of DCs in such preparations lacks CD103 expression, for which reason CD103 appears less suitable to serve as a retrospective marker for all SI LP-derived DCs in the MLN (15). As DCs with high-level expression of MHC class II (MHC-II) are selectively absent from skin-draining LN in CCR7-deficient mice (8), and mesenteric lymph-borne DCs uniformly display a MHC-II^high phenotype (15), migratory and LN-resident DCs in the MLN under steady state are currently more often distinguished based on relative levels of MHC-II expression.

Whereas it is well recognized that TLR ligands as well as inflammatory mediators, including TNF-α, induce a rapid migration of DCs from tissues to draining SLO (16–18), the pathways involved in mobilizing DC migration from tissues to draining SLO in steady state remain largely unclear. In the gut this process seems to be intimately linked to intestinal homeostasis, as induction of oral tolerance to ingested proteins is defective after surgical removal of the MLN as well as in CCR7-deficient animals (13), and in vivo expansion of DC numbers by FMS-like tyrosine kinase 3 ligand augments oral tolerance (19). In the current study, we assessed the mechanism underlying the steady-state migration of CD103⁺ DCs from the SI LP to the MLN. Our results reveal a previously unrecognized role for the TLR intracellular adaptor molecule MyD88 in this process and demonstrate that MyD88 signaling in the DCs stimulates their migration independently of the microbiota, TNF-α, type I IFN, and caspase-1–dependent maturation of IL-1 cytokine family members.

C57BL/6 wild-type (WT) mice were purchased from Taconic or Charles River. Ifnar1^−/− mice (20), Tnfrsf1-dKO mice (TNFR1/2^−/−) (21), caspase-1^−/− (22), B6.129P2-Il18r1^tm1Aki/J (23), B6.129P2(SJL)-Myd88tm1Defr/J (24), Itgax-Cre (25), C57BL/6-Tg(TcrαTcrβ)1100Mjb/J (OT-I).CD45.1, and B6.Cg-Tg(TcrαTcrβ)425Cbn/J (OT-II).CD45.1 mice, all on a C57BL/6 background, were bred and maintained at the Biomedical Center animal facility (Lund University, Lund, Sweden). Myd88^−/− mice on the C57BL/6 background (26), C57BL/6 mice, Trif^−/− mice on a mixed C57BL/6/129 background (27), and corresponding control C57BL/6/129 mice were bred and maintained at the Department of Microbiology, Immunology, and Glycobiology animal facility (Lund University). Mice were used for experiments between 6 and 14 wk of age. Littermates or WT and knockout mice of the same sex and age were used within each experiment. Germ-free mice were fed an autoclaved chow diet and maintained in flexible film isolators at the Experimental Biomedicine Animal Facility, University of Gothenburg. Germ-free isolators were routinely tested for sterility by culturing and PCR analysis of feces amplifying the 16S rRNA gene. Animal experiments were performed after approval from the Malmö/Lund or Gothenburg Animal Ethics Committee.

The following Abs were used: CD103 biotin (M290), CD103 PE (M290), anti-rat IgG2b biotin (G15-337), and CD11c PE (HL3; BD Pharmingen); NK1.1 PE Cy5 (PK136), CD11b FITC (M1/70), MHC II Pacific blue (M5/114.15.2), CCR7 PE (4B12), CD40 PE (1C10), streptavidin allophycocyanin, CD45.1 Alexa Fluor 700 (A20), CD3ε Alexa Fluor 700 (17A2), NK1.1 Alexa Fluor 700 (PK136), CD19 Alexa Fluor 700 (6D5), IL-18R Alexa Fluor 647 (BG/IL-18RA), CD205 biotin (NLDC-145), and CD45.2 Alexa Fluor 700 (104; BioLegend); and CD11b allophycocyanin eFluor780 (M1/70), MHC II allophycocyanin (M5/114.15.2), CD3ε PE Cy5 (145-2C11), CD19 PE Cy5 (eBio1D3), CD11c PE Cy7 (N418), CD80 allophycocyanin (16-10A1), CD86 allophycocyanin (GL1), integrin α₄β₇ PE (DATK32), CD62L PE (MEL-14), B220 PE Cy7 (RA3-6B2), CD8 eFluor450 (53-6.7), CD4 allophycocyanin eFluor780 (RM4-5), CD4 PE (GK1.5), Foxp3 allophycocyanin (FJK-16s), CD8α FITC (53-6.7), and CD11c Alexa Fluor 700 (N418; eBioscience). Anti-CCR9 purified from the hybridoma 7E7 was provided by O. Pabst (Hannover, Germany).

Isolation of MLN and SI LP DCs was performed, as previously described (9). Briefly, for isolation of cells from MLN, LN were cut into pieces and incubated with collagenase IV (500 μg/ml) and DNase I (50 U/ml; Sigma-Aldrich) diluted in R10 media (RPMI 1640, 10% FCS, 100 U/ml penicillin, 100 μg/ml streptomycin, 50 μg/ml gentamicin, 10 mM HEPES, 1 mM sodium pyruvate, and 50 μM 2-ME; all from Invitrogen) for 45 min at 37°C on an orbital shaker. Remaining tissue was mashed and filtered through a 70-μm cell strainer (BD Falcon). Total cell counts were determined with a Sysmex KX-21N hematology analyzer (Sysmex). For isolation of cells from the SI LP, Peyer’s patches (PP) were removed and the intestine opened longitudinally, cut into pieces, and incubated 3 × 15 min in HBSS containing 10% FCS and 2 mM EDTA at 37°C to remove epithelial cells. After each incubation step, tubes were shaken for 10 s and media containing epithelial cells and debris was discarded. The remaining tissue was incubated for 1 h in R10 supplemented with collagenase VIII (0.25 mg/ml; Sigma-Aldrich) and CaCl₂ (50 mM) at 37°C with magnetic stirrers. The resulting cell suspension was filtered consecutively through 100- and 40-μm cell strainers. For isolation of BM cells, femur and tibia were collected and crushed using a mortar and pestle. The cells were collected in PBS, filtered through a 70-μm cell strainer, and counted.

Cells were stained with directly conjugated Abs, essentially as previously described (9). Abs to CD45 were included in the staining of LP cells to exclude cells of nonhematopoietic origin. Dead cells were identified using propidium iodide (PI; Invitrogen) or the Live Dead Fixable Violet Dead Cell stain kit (Invitrogen), according to the manufacturer’s instructions. For detection of BrdU, cells were stained with FITC-conjugated anti-BrdU Ab (FITC BrdU Flow kit; BD Pharmingen), according to the manufacturer’s instructions. Analysis of aldehyde dehydrogenase activity was performed using the ALDEFLUOR staining kit (STEMCELL Technologies), according to the manufacturer’s instructions. ALDEFLUOR-reactive cells were detected in the FITC channel. Data acquisition was performed on a FACSAria or LSRII (BD Biosciences) and analyzed using FlowJo software (Tree Star).

For BrdU pulse-chase experiments, mice were injected i.p. with BrdU in PBS (2 mg; Sigma-Aldrich), and organs were collected 12, 24, or 48 h later.

Mice were fed 10 μg R848 by oral gavage and sacrificed 16 h later. MLN were collected and analyzed by flow cytometry.

C57BL/6 and MyD88^−/− mice were irradiated with a single dose of 900 rad and reconstituted with BM cells (4 × 10⁶ cells/mouse) from C57BL/6 or MyD88^−/− mice. The mice received ciprofloxacin (100 mg/l; Bayer HealthCare AG) in the drinking water for 2 wk following irradiation and were allowed to reconstitute for 8 wk.

Mesenteric lymphadenectomy and thoracic duct cannulation procedures were performed according to established protocols (28). Mesenteric lymphadenectomy was performed on 6-wk-old C57BL/6 germ-free or control male mice by laparotomy and blunt dissection. Germ-free status was confirmed by PCR for bacterial 16S rRNA prior to cannulation. Six weeks postsurgery, mice were fed 0.2 ml olive oil to visualize the lymphatics, and the thoracic lymph duct was cannulated by the insertion of a polyurethane cannula (2Fr; Linton Instrumentation, Diss, U.K.). Lymph was collected in PBS with 20 U ml⁻¹ heparin sodium (Wockhardt UK, Wrexham, U.K.), on ice, for up to 16 h.

SI epithelial and LP fractions were prepared by treatment with EDTA, as described above, after which the media containing epithelial cells was collected and remaining LP tissue digested with collagenase. The LP and epithelial cell fractions were then enriched on 40:70 Percoll gradients. After centrifugation at 600 × g for 20 min, the interface was collected from the LP samples and the top layer was collected from the epithelial cell samples. RNA was prepared using RNeasy Mini Kit (Qiagen), and cDNA was generated using Superscript III First-Strand kit (Invitrogen) with random hexamers. Expression of β-actin, IL-18, and IL-1β was analyzed using an iCycler (Bio-Rad) and SYBR Green ER qPCR SuperMix (Invitrogen) and the following primers: β-actin forward, 5′-CCACAGCTGAGAGGGAAATC-3′; β-actin reverse, 5′-CTTCTCCAGGGAGGAAGAGG-3′; IL-18 forward, 5′-TGTTCGAGGATATGACTGATATTGA-3′; IL-18 reverse, 5′-CCAGTCCTCTTACTTCACTGTCTTT-3′; IL-1β forward, 5′-AGCACCTTCTTTTCCTTCATCTTT-3′; and IL-1β reverse, 5′-AGCCTGTAGTGCAGTTGTCTAATG-3′.

C57BL/6 mice were given drinking water supplemented with 1 g/L ampicillin, 0.5 g/L vancomycin, 1 g/L neomycin, and 1 g/L metronidazole for 2 wk before they were sacrificed, and DCs in the MLN and small intestine were analyzed by flow cytometry.

OT-I and OT-II cells were isolated from spleen and LN of CD45.1⁺ OT-I or OT-II mice by negative selection using MACS, and CFSE labeled according to standard protocols. For Foxp3⁺ Treg cell generation, 10⁶ OT-II cells were injected i.v. into each recipient mouse (day 0), followed by administration of 50 mg OVA per mouse by oral gavage on days 1 and 2. Mice were sacrificed on day 5, and MLN were collected for analysis of donor T cells (PI⁻ CD4⁺CD45.1⁺ B220⁻). Alternatively, 10⁶ OT-I and OT-II cells mixed at a 1:1 ratio were injected i.v. into each recipient mouse, followed by administration of 10 mg OVA by oral gavage. Mice were sacrificed 4 d later, and donor T cells (CD4⁺ or CD8⁺ in combination with a PI⁻ CD45.1⁺ B220⁻ phenotype) in the MLN were analyzed by flow cytometry.

Statistical analysis was performed with Prism software (GraphPad Software) using one-way ANOVA with Tukey’s multiple comparison test for comparison of three or more unpaired groups, or Mann–Whitney U test for two unpaired groups, in which *p < 0.05, **p < 0.01, and ***p < 0.001.

To determine whether SI LP-derived DCs can be distinguished from LN-resident subsets in the MLN on the basis of relative levels of MHC-II expression, we first compared the phenotype of MLN DCs expressing high and relatively lower levels of MHC-II (MHC-II^high and MHC-II^low, respectively). Whereas only a minor proportion of the MHC-II^low MLN DCs expressed CD103 (14 ± 3%, n = 24), the frequencies of CD103⁺CD11b⁺, CD103⁺CD11b⁻, and CD103⁻ (coexpressing CD11b or not) DCs within the MHC-II^high gate appeared to mirror the previously described composition of intestinal lymph DCs (15) (Fig. 1A, 1B). We, and others, have previously performed BrdU pulse-chase experiments to track the entry of DCs into the MLN in the steady state (14, 29). As it is reasonable to assume that BrdU-labeled SI LP-derived DCs accumulate in the MLN with a delayed kinetics compared with BrdU⁺ DCs proliferating in the LN or entering directly from the blood via high endothelial venules (14), we next compared the short-term accumulation of MHC-II^high and MHC-II^low DCs in the MLN. Following a single i.p. injection with BrdU, labeled MHC-II^high DCs appeared with delayed kinetics as compared with MHC-II^low DCs, and this was equally evident for all three aforementioned phenotypic subsets of MHC-II^high DCs (Fig. 1C, 1D). As expected, labeled MHC-II^low DCs (of which the large majority is CD103⁻) accumulated rapidly in the MLN, consistent with previous results indicating that most CD103⁻ DCs derive from precursors entering the LN directly from the blood (9, 14, 30), and can undergo proliferation within the LN (14). Collectively, these results confirm that the MHC-II^high phenotype marks SI LP-derived DCs in the gut-draining MLN, and we will onward use the terms MHC-II^high MLN DCs and SI LP-derived DCs interchangeably.

Similar to skin-draining LN (8), the frequency and number (not depicted) of MHC-II^high MLN DCs were strongly and selectively reduced in CCR7^−/− mice due to a dramatic decrease of the CD103⁺ subset (Fig. 2A). To determine to what extent the BrdU pulse-chase methodology represents a useful tool for addressing the mechanisms underlying steady-state migration of SI LP DCs to the MLN, we next compared the short-term accumulation of MHC-II^high DCs in the MLN of WT and CCR7-deficient mice. CCR7^−/− mice showed an almost complete block in BrdU⁺ CD103⁺ MHC-II^high DC accumulation in the MLN (Fig. 2B, 2C). Notably, we could, however, not detect a significant reduction in the mobilization of BrdU⁺ MHC-II^high DCs lacking CD103 expression in the CCR7-deficient mice (Fig. 2B, 2D), indicating that this relatively small subset may enter afferent lymphatic vessels at least partially by CCR7-independent mechanisms. Importantly, the appearance of BrdU-labeled CD103⁺ DCs in the SI LP did not differ between WT and CCR7^−/− mice (Fig. 2E, 2F), indicating that the impaired accumulation of labeled CD103⁺ MHC-II^high DCs in the MLN of CCR7^−/− mice was not caused by a failure of these cells to enter the gut mucosa in the absence of CCR7. Finally, the more rapid accumulation of BrdU⁺ MHC-II^low DCs in the MLN was also not affected by CCR7 deficiency (Fig. 2G), consistent with the normal number of these cells in the MLN of the CCR7^−/− mice (see Fig. 2A). Collectively, these results suggest that BrdU pulse-chase experiments represent a sensitive tool to track the steady-state mobilization of the CD103⁺ DCs from the SI LP into the MLN.

Oral delivery of the TLR7/8 agonist R848 induces a massive and rapid migration of SI LP DCs to the MLN that is dependent on TNF-α, whereas type I IFN, under this specific condition drives upregulation of costimulatory molecules by the DCs (17). A similar TNF-α–dependent SI LP DC mobilization occurs after i.p. injection of LPS (16). We thus next addressed the role of TNF-α and type I IFN signaling in SI LP CD103⁺ DC migration into the MLN in the steady state using mice genetically deficient in TNFR type 1 and 2 (TNFR1/2^−/− mice) or the type I IFN receptor (IFNAR^−/− mice). The absolute number of CD103⁺ MHC-II^high DCs in the MLN did not differ between WT and IFNAR^−/− or TNFR1/2^−/− mice (Fig. 3A) and in BrdU pulse-chase experiments CD103⁺ MHC-II^high DCs accumulated with similar kinetics in MLN of all three strains (Fig. 3B). Consistent with these results, CD103⁺ MHC-II^high MLN DCs in both TNFR1/2^−/− and IFNAR^−/− mice displayed similar levels of CCR7 expression as their counterparts in WT animals (Fig. 3C). In marked contrast, and in agreement with published results (17), the accumulation of DCs in the MLN after oral R848 administration was significantly impaired in TNFR1/2^−/− mice (analysis performed on total DCs as all MLN DCs become MHC-II^high after oral R848 treatment) (Fig. 3D). Collectively, these data indicate that neither TNF-α nor type I IFN signaling is required for the steady-state migration of CD103⁺ SI LP DCs into the MLN.

To determine whether the TLR signaling adapter molecules Toll/IL-1R domain–containing adapter inducing IFN-β (TRIF) and MyD88 are involved in driving steady-state migration of DCs from the gut mucosa to the draining MLN, we next performed essentially the same line of experiments as described above, now using TRIF^−/− and MyD88^−/− mice. Consistent with the functional role of TRIF upstream of IFN-β expression (27), TRIF^−/− mice displayed no signs of compromised SI LP DC migration to the MLN (Fig. 4A, 4B).

Although the proportion of CD205⁺CD8α⁻ cells among total MLN DCs has been shown to be comparable between MyD88^−/− and WT mice (31), flow cytometry analysis of MHC-II^high DCs revealed that the CD205⁺CD8α⁻ phenotype does not fully encompass all migratory DCs in the MLN (Supplemental Fig. 1). Because of this, and because assessment of subset frequencies within the DC lineage does not take possible alterations in total DC number and/or total MLN cellularity into account, we assessed how MyD88 deficiency impacts on the number and short-term accumulation of MHC-II^high DCs in the MLN in the steady state. Analysis of MyD88-deficient mice revealed an overall reduced cellularity in the MLN (Fig. 4C). However, the percentage of MHC-II^high, but not MHC-II^low, DCs among total viable cells was significantly reduced as a consequence of MyD88 deficiency (Fig. 4D). Notably, the frequency and absolute number of CD103⁺ SI LP-derived DCs were reduced by 43 ± 6% and 59 ± 11% (mean value ± SD, n = 27), respectively, in the MLN of MyD88^−/− mice (Fig. 4D, 4E). The percentage of CD103⁺ DCs among total viable leukocytes in the SI LP was still comparable between the two strains (Fig. 4F), indicating that the development and recruitment of these cells to the intestinal mucosa occur normally in the absence of MyD88. To assess whether the reduced number of SI LP-derived CD103⁺ DCs in the MLN of MyD88^−/− mice reflected deficient recruitment from the gut, we performed BrdU pulse-chase experiments, as above. By this approach, we detected a ∼50% reduction in the short-term accumulation of BrdU⁺ CD103⁺ MHC-II^high DCs in the MLN of the MyD88-deficient animals (Fig. 4G). This was accompanied by slightly, but significantly reduced levels of CCR7 on the migratory CD103⁺ DCs that had entered the MLN of these mice as compared with their counterparts present in MLN of WT mice (Fig. 4H). In contrast to the MLN, neither WT nor MyD88-deficient CD103⁺ SI LP DCs expressed detectable levels of CCR7 when analyzed directly ex vivo (data not shown). Culturing the SI LP cell preparations for 2 h at 37°C, however, resulted in detectable CCR7 expression by the CD103⁺ SI LP DCs that was similar for WT and MyD88^−/− cells (Supplemental Fig. 2), demonstrating MyD88-deficient SI LP DCs are not intrinsically impaired in their ability to express the receptor. Taken together, these results indicate that MyD88 signaling (independent of TNF-α) is required for optimal CCR7-dependent steady-state migration of CD103⁺ SI LP DCs to the MLN.

To assess the contribution of MyD88 signaling in stromal versus hematopoietic cells, BM chimeras (WT→WT; MyD88^−/−→WT; WT→MyD88^−/−; MyD88^−/−→MyD88^−/−) were injected with BrdU, and the number of BrdU⁺ CD103⁺ SI LP-derived DCs in the MLN was determined after 24 h (Fig. 5A). Whereas control chimeras in which all cells were either of WT (WT→WT) or MyD88^−/− (MyD88^−/−→MyD88^−/−) origin reproduced the results obtained with nonchimeric animals, we could not detect a significant reduction in SI LP CD103⁺ DC migration with MyD88 deficiency confined to the stromal cell compartment (p = 0.42, n = 7–8 per group). However, fewer BrdU-labeled CD103⁺ MHC-II^high DCs appeared to accumulate in MLN of chimeras lacking MyD88 selectively in hematopoietic cells than in WT→WT control chimeras (p = 0.039, n = 7–8 per group) (Fig. 5A), indicating that MyD88 signaling in BM-derived cells at least partially accounts for the MyD88 dependency in this process.

Given the results obtained with BM chimeric mice, we next assessed whether MyD88 expression in CD11c⁺ cells was required for optimal steady-state DC migration by comparing CD11c-Cre·MyD88^fl/fl mice with MyD88^fl/fl littermates. Analysis of MLN showed that the total cellularity was comparable between the two groups (Fig. 5B), indicating that MyD88 deficiency in cells other than DCs underlies reduced total MLN cellularity in mice with germline deletion of MyD88 (see Fig. 4C). However, the number of CD103⁺ MHCII^high DCs in the MLN of CD11c-Cre·MyD88^fl/fl mice was significantly lower than in MyD88^fl/fl littermates (Fig. 5C), and this was accompanied by a 60 ± 14% (mean value ± SD, n = 8–11, p = 0.0008) reduction in short-term mobilization of these cells from the SI LP to the MLN, as assessed by BrdU pulse-chase experiments (Fig. 5D). Finally, as LP-derived CD103⁺CD11b⁺ and CD103⁺CD11b⁻ MLN DCs have different ontogeny and function (32), we addressed these subsets separately. As shown in Fig. 5E and 5F, the total number and accumulation of BrdU-labeled CD103⁺CD11b⁺ and CD103⁺CD11b⁻ DCs in the MLN were similarly reduced in CD11c-Cre·MyD88^fl/fl mice compared with MyD88^fl/fl controls. Collectively, these results demonstrate that the selective depletion of MyD88 in CD11c⁺ cells fully reproduces the impaired migration of SI LP CD103⁺ DCs to the MLN observed in the complete MyD88 knockout, and that both CD11b⁺ and CD11b⁻ SI LP CD103⁺ DCs are dependent on MyD88 expression in CD11c⁺ cells for optimal steady-state migration to the MLN.

Similar to the TLRs, the cytokines IL-1β and IL-18 use the adapter MyD88 for intracellular downstream signaling events, and targeted disruption of the MyD88 gene results in loss of IL-1β– and IL-18–mediated function (26). IL-18 and IL-1β were constitutively expressed in the SI epithelium and LP, respectively (Fig. 6A, 6B), and mRNA levels for both cytokines were significantly reduced in MyD88^−/− mice (Fig. 6A, 6B). IL-1β and IL-18 are translated as immature proteins and must be enzymatically processed to gain biological activity (33). Inflammasome-dependent activation of caspase-1 represents a major pathway for proteolytic activation of these cytokines (33). Thus, to address a possible role for MyD88-dependent IL-1β/IL-18 signaling in steady-state SI LP CD103⁺ DC migration, we performed BrdU pulse-chase experiments in caspase-1^−/− mice. The accumulation of BrdU-labeled SI LP-derived CD103⁺ DCs in the MLN was, however, not significantly reduced in the absence of caspase-1 (Fig. 6C). Although alternative and caspase-1–independent pathways exist for generation of mature IL-1β and IL-18 (34), analysis of MHC-II^high MLN DCs in IL-18R^−/− mice revealed only a minor reduction in the number of CD103⁺CD11b⁺ and CD103⁺CD11b⁻ DCs as compared with WT littermates (Fig. 6D). Furthermore, the short-term accumulation of these subsets in the MLN was not significantly reduced in IL-18R^−/− mice as compared with WT littermates, as assessed by BrdU pulse-chase experiments (Fig. 6E). Consistent with this result, we could by flow cytometry not detect IL-18R expression on any of the DC subsets or on CD11c⁺ macrophages in the SI LP, whereas expression was clearly detectable on NK cells in the same LP cell preparations (Fig. 6F). Altogether, these results suggest that IL-18 and caspase-1–dependent IL-1β are unlikely to play a critical role in the MyD88-dependent steady-state SI LP CD103⁺ DC migration to the MLN.

To determine whether the intestinal microbiota is required for the homeostatic MyD88-dependent intestinal DC migration, we performed thoracic duct cannulation of germ-free and conventionally housed mice after mesenteric lymphadenectomy. Flow cytometry analysis of mesenteric lymph collected from these mice revealed a similar percentage of CD11c⁺ MHC-II⁺ DCs among total viable cells (Fig. 7A), and the frequencies of lymph-borne CD103⁺ and CD103⁺CD11b⁺ DCs were not reduced in the germ-free as compared with conventionally housed mice (Fig. 7B–D). To verify these results in an additional experimental system, WT mice were treated with broad-spectrum antibiotics for 2 wk, a protocol previously shown to dramatically reduce microbial loads in the intestinal lumen (35). Antibiotic-treated mice displayed considerably enlarged ceca, an anatomical alteration arising from the decreased microbial load and also observed in germ-free animals (36). BrdU-injected mice subjected to this treatment displayed a similar accumulation of BrdU-labeled CD103⁺ SI LP DCs in the MLN as compared with untreated controls (Fig. 7E). Altogether, these results suggest that the steady-state migration of SI LP DCs to the MLN is not driven by the intestinal microbiota.

Given the important role for SI LP DCs in presenting luminal Ags to I T cells in the draining MLN (13), we finally examined the impact of specific MyD88 deletion in CD11c⁺ cells on T cell responses to orally administered Ag. OVA-specific TCR transgenic CD4⁺ and CD8⁺ T cells (OT-II and OT-I cells, respectively) were labeled with CFSE and adoptively cotransferred to MyD88^fl/fl and CD11c-Cre·MyD88^fl/fl mice, which were subsequently fed 10 mg OVA by gavage (Fig. 8). FACS analysis of the MLN 4 d later revealed no apparent effect on T cell expansion, as judged from CFSE dilution and total number of donor T cells (Fig. 8A–D). However, the percentage of CD8⁺ OT-I cells expressing CCR9 was slightly, but significantly reduced, as was the downregulation of CD62L by the activated CD4⁺ OT-II cells (Fig. 8A–D). To assess the generation of Foxp3⁺ Treg cells in the MLN, OT-II cell recipients were fed 50 mg OVA on consecutive 2 d. Flow cytometry analysis 3 d later revealed similar frequencies of Foxp3⁺ OT-II cells in the presence or absence of MyD88 in CD11c⁺ cells (Fig. 8E). Furthermore, there was also no difference in the percentage of Foxp3⁺ cells among the recipients’ endogenous SI LP CD4⁺ T cells (Fig. 8F). RA stimulates both CCR9 expression and, in the presence of TGF-β, induction of Foxp3⁺ Treg cells and is produced from vitamin A under the control of aldehyde dehydrogenases (11, 12, 37). A reduced ability of MyD88-deficient CD103⁺ SI LP-derived DCs to produce RA appeared, however, not to underlie decreased CCR9 expression by CD8⁺ T cells in CD11c-Cre·MyD88^fl/fl mice, as assessed by aldehyde dehydrogenase activity using the Aldefluor assay (Supplemental Fig. 3). The steady-state expression of CD86, CD80, and CD40 was also comparable between CD103⁺ SI LP-derived DCs in MyD88^−/− and WT mice (Supplemental Fig. 3). Collectively, these results do not reveal dramatic changes in T cell activation and differentiation in response to an oral Ag in mice with specific deletion of MyD88 in DCs. Furthermore, they show that a slightly reduced ability of T cells to acquire CCR9 expression and to downregulate CD62L in these mice is unlikely to be caused by reduced aldehyde dehydrogenase activity or a general costimulatory defect of the MyD88-deficient DCs.

Ag presentation by DCs in the absence of overt inflammation leads to peripheral T cell tolerance (38), and intestinal CD103⁺ DCs have been reported to be particularly effective at supporting TGF-β–dependent differentiation of Foxp3⁺ Treg cells in vitro, a property at least partially conferred by their ability to generate the vitamin A metabolite, RA (11, 12). Because the induction of oral tolerance to fed soluble Ags requires migration of APCs from the SI LP to the draining MLN (13), and the majority of MHC-II–expressing cells entering the MLN via this route are CD103⁺ DCs (15, 30), we have in this study addressed the mechanism by which SI LP CD103⁺ DCs are recruited to the MLN in the steady state. Our results reveal that deficiency in the TLR signaling adapter molecule MyD88 is associated with a 50–60% reduction in CD103⁺ DC migration, affecting both the CD103⁺CD11b⁺ and CD103⁺CD11b⁻ DC subsets, and that MyD88-signaling in CD11c⁺ cells accounts for this effect. Notably, DC migration was not impaired after treatment with broad-spectrum antibiotics, and the frequencies of DCs were similar in thoracic lymph collected from germ-free and control mice after mesenteric lymphadenectomy. Collectively, these results indicate that MyD88 signaling independently of the commensal gut microbiota is involved in regulating steady-state migration of SI LP CD103⁺ DCs to the MLN.

Multiple parameters most likely influence the number of migratory DCs present in the draining LN in the steady state, including immigration, survival, and proliferation. We have previously demonstrated that CD103⁺ DCs undergo minimal proliferation within the MLN, as determined by their lack in expression of the proliferation-associated nuclear Ag Ki67 (14). This has allowed us to perform short-term in vivo BrdU pulse-chase experiments to assess the mechanisms regulating entry of CD103⁺ DCs into the MLN. In agreement with a critical role for CCR7 in this process (9, 13, 39), we show that the accumulation of BrdU-labeled CD103⁺ DCs in the MLN is essentially shut off in CCR7^−/− mice. Further to this, our BrdU pulse-chase experiments show that MHC-II^high DCs accumulate with a delayed kinetics in the MLN as compared with their MHC-II^low counterparts, which is consistent with the MHC-II^high phenotype of all DCs in mesenteric lymph (15, 40). This result therefore confirms that relative expression levels of MHC-II can be used as a criterion for discriminating between migratory and resident DCs in the MLN. The ability to directly track short-term accumulation of MHC-II^high DCs in the MLN has thus allowed us to identify migration per se as one of the MyD88-dependent effects regulating SI LP-derived CD103⁺ DC numbers in the MLN.

The finding that MyD88-dependent signaling significantly contributes to the homeostatic migration of CD103⁺ DCs into the MLN is in contrast with a previously published observation of a normal proportion and maturation status of migratory DCs within the MLN of MyD88^−/− mice (31). However, in this study, CD103⁺ MHC-II^high DCs were not enumerated, and instead the frequency of CD8α^low DEC205⁺ DCs among total MLN DCs was compared between WT and MyD88^−/− mice. Consistent with other more recent reports [reviewed in (1)], our flow cytometry results show a more complex phenotypic composition of the SI LP-derived DCs in the MLN. Thus, the previous finding of a normal SI LP DC migration in MyD88^−/− mice may relate to the assessment of frequency rather than absolute number of migratory DCs and/or the different markers used to identify these DCs in the MLN. Finally, intestinal steady state can probably represent quite different physiological conditions depending on variations in, for example, the gut microbiota, enteric viruses, and/or nutritional conditions between different animal facilities, and such parameters are likely to influence the gut immune system also in the absence of overt inflammation. Whereas the failure of BrdU-labeled CD103⁺ SI LP DCs to accumulate in the MLN of CCR7^−/− mice serves as an important internal control for our experimental approach, it is clear that the corresponding reduction in MyD88^−/− mice is not equally dramatic. Other and MyD88-independent mechanisms are hence in place (see below) and may play a more dominant role in other settings.

MyD88 conveys signals from the IL-1R family members and innate receptors, including the TLRs (33). Although compounds derived from the microbiota influence immune homeostasis in both MyD88-dependent and -independent manners (41, 42), we show in this work that mesenteric lymph from germ-free mice contains a normal proportion of DCs, and that treatment of conventionally housed WT mice with broad-spectrum antibiotics has no effect on the accumulation rate of intestinal CD103⁺ DCs in the MLN. These results are in agreement with previous observations of similar phenotype and function of DCs in MLN of germ-free and conventionally housed rodents (31, 43, 44), and collectively argue against a direct role for the gut microbiota in driving the MyD88-dependent steady-state migration of SI LP CD103⁺ DCs. It still, however, remains possible that TLR ligands present in the food and/or produced by enteric viruses could account for the MyD88 dependency in this process. Although we currently cannot exclude these possibilities, an alternative explanation is that endogenous damage-associated molecular patterns (DAMPs) released from dying cells instigate MyD88 signaling events in the DCs. In this regard, the intestinal surface is under continual renewal with rapid epithelial turnover in the steady state, and lymph-borne DCs have been shown to contain cytoplasmic apoptotic DNA, epithelial cell–restricted cytokeratins, and epithelial cell–associated nonspecific esterase⁺ inclusions—indicative of a continuous transport of apoptotic intestinal epithelial cell remnants to the MLN (43). As it more recently has become clear that intestinal epithelial cell death also can occur through necroptosis (45), a process associated with the release of a wide range of DAMPs including IL-1 cytokine family members (46), and because a number of DAMPs bind to receptors utilizing MyD88 as a signaling adaptor (47), it remains possible that this pathway plays a role in promoting steady-state migration of SI LP CD103⁺ DCs to the MLN. In this context, it would also be interesting to examine the impact of simultaneous genetic deficiency in IL-1β, IL-18, and IL-33.

Although the steady-state accumulation of CD103⁺ SI LP DCs in the MLN is critically dependent on CCR7, MyD88 deficiency had only a minor (although significant) impact on the steady-state levels of CCR7 expressed by CD103⁺ SI LP-derived DCs present in the MLN. The mechanism by which CCR7 mediates accumulation of migratory DCs in draining LN involves guidance of the DCs within the perilymphatic interstitium along gradients of the CCR7 ligand CCL21 that is produced by the lymphatic capillary endothelial cells (48). Although we could not detect CCR7 on SI LP DCs directly ex vivo, the moderate difference observed between WT and MyD88-deficient SI LP-derived CD103⁺ DCs in the MLN indicates that MyD88 is involved in regulating CCR7 expression in these cells also in the steady state. A reduced ability to gain CCR7 could therefore at least partially explain the decreased steady-state migration of CD103⁺ SI LP DCs in the absence of MyD88.

As mentioned earlier, the partial effect of MyD88 deficiency on the CCR7-dependent steady-state migration of CD103⁺ SI LP DCs indicates that alternative and MyD88-independent mechanisms contribute to this process. Whereas these molecular pathways remain enigmatic, we have in the current study ruled out an essential role for a number of MyD88-dependent and –independent pathways associated with DC activation under inflammatory conditions. We thus show that the TLR3 and TLR4 signaling adapter molecule TRIF and downstream type I IFN production are not required. We further demonstrate a normal migration in the absence of caspase-1–dependent IL-1β and IL-18 maturation. Finally, whereas TNF-α underlies exaggerated mobilization of SI LP DCs into the MLN in response to inflammatory stimuli, including TLR4 or TLR7 agonists (16, 17), our results rule out an essential role for TNF-α in this process during homeostatic conditions. Distinct molecular pathways therefore appear to govern SI LP DC migration during homeostatic and inflammatory conditions.

We have previously shown that the bulk population of CD103⁻ MLN DCs accumulates with a rapid kinetics in the MLN (14), indicating that most of them are not derived from the gut mucosa. However, taking the recently discovered lymph-borne CD103⁻ DCs into consideration (15), we have by restricting our analysis to cells expressing high levels of MHC-II in our BrdU-labeling experiments now been able to resolve the migratory CD103⁻ DC subset in the MLN. Although the slow accumulation of the MHC-II^high CD103⁻ DCs in the BrdU pulse-chase experiments is consistent with their entry via afferent lymphatics, our results surprisingly indicate that CCR7 could be dispensable for their accumulation in the MLN. As CD103⁻CD11b⁻ DCs are absent from mesenteric lymph in RORγt-deficient mice, which lack PP and isolated lymphoid follicles, it seems likely that a large proportion of these cells derives from GALT rather than from the LP (15). Cellular egress from PP into mesenteric lymph appears to occur mostly from the interfollicular region, where the absorbing lymphatic peripheral vessels form a dense network (49). One possible explanation for the apparently CCR7-independent accumulation of migratory CD103⁻ DCs in MLN is thus that the entry of these cells from the PP into the mesenteric lymphatic vessels instead is dependent on signals that govern leukocyte egress from the PP (50, 51).

The failure of SI LP DCs to migrate to the MLN in CCR7-deficient mice is associated with severely reduced T cell priming in the MLN and impaired induction of oral tolerance (13). Mice with targeted deletion of inhibitory κB kinase in DCs were recently reported to have a similarly strong defect in their accumulation of migratory DCs in cutaneous LN, leading to impaired generation of Treg cells in these LN and spontaneous development of autoimmunity (52). Deleting MyD88 specifically in CD11c⁺ cells, however, resulted in only modest alterations in T cell responses in MLN after oral administration of Ag in the absence of adjuvant. Early T cell expansion and generation of Foxp3⁺ Treg cells were not affected, indicating that the number of CD103⁺ DCs reaching the MLN independently of MyD88 is sufficiently high to sustain T cell expansion, and that MyD88 expression in DCs is not essential for Treg cell generation per se. The mechanism accounting for the reduced CCR9 expression in mice lacking MyD88 specifically in DCs is currently unclear. Although MyD88-deficient DCs have been proposed to produce reduced amounts of RA (53), similar to Molenaar et al. (54), we could not detect reduced aldehyde dehydrogenase activity in these cells, indicating that their ability to produce RA from retinol was not severely impaired. Also unclear is why CD4⁺ T cells downregulated CD62L less efficiently when primed by MyD88-deficient DCs. Although downregulation of CD62L in general seems to correlate with T cell activation signal strength (55, 56), the expression of costimulatory molecules CD80, CD86, and CD40 was, in contrast to CCR7, not significantly lower on MyD88-deficient than on WT DCs. It is, however, possible that the reduced number of Ag-presenting DCs in the MLN of mice with DC-specific deletion of MyD88 results in an increased competition between individual T cells for TCR ligands and/or costimulatory molecules, including RA, and that a limited availability to these stimuli underlies the altered expression of CD62L and CCR9 in these mice.

In summary, we have demonstrated that the steady-state migration of CD103⁺ SI LP DCs to the draining MLN involves MyD88 signaling in the DCs and occurs independently of the gut microbiota. In contrast to the pronounced effect that TNF-α has on SI DC migration in response to TLR ligands, the steady-state migration is also not dependent on TNF-α. As the continuous migration of DCs from the intestinal mucosa to the MLN appears to be instrumental in the regulation of the intestinal immune system, and for intestinal homeostasis, deciphering the MyD88-dependent mechanism governing this mobilization should aid future attempts to treat intestinal inflammation or to exploit the oral route for vaccination.

We thank Drs. Oliver Pabst, Boris Reizis, Mary-Jo Wick, and Fredrik Bäckhed for providing CCR7^−/−, CD11c-Cre, caspase-1^−/−, and germ-free mice, respectively.

The authors have no financial conflicts of interest.