The transcription factor NR4A3 controls CD103⁺ dendritic cell migration

The intestine has a mucosal surface that directly interacts with large quantities of food antigens and microorganisms daily. As a consequence, the gut is the accessible entry site for many pathogens. Because most food antigens and microorganisms are not harmful to the host, the immune system must be able to distinguish invading pathogens from innocuous antigens. Also, tolerance to those innocuous antigens is crucial to maintaining the fine balance between immunity and tolerance. If this balance is skewed, inflammatory diseases such as inflammatory bowel disease (IBD) and food allergies can ensue. Dendritic cells (DCs) play important roles in initiating adaptive immunity against pathogens as well as inducing tolerance against unharmful antigens and commensal microorganisms required to maintain this homeostasis.

In homeostatic conditions, CD103⁺ migratory lamina propria DCs (LPDCs), including CD103⁺CD11b^– and CD103⁺CD11b⁺ DCs, sample antigens from the gut lumen and from apoptotic epithelial cells, then migrate to mesenteric lymph nodes (MLNs) (1, 2) to promote antigen-specific T cell activation, including activation of FOXP3⁺ Tregs (3). LPDCs include at least 3 subsets: CD103⁺CD11b^– DCs, which are derived from conventional DCs (cDCs), CD103^–CD11b⁺ DCs, which are derived from monocytes (4), and CD103⁺CD11b⁺ DCs, which are of heterogeneous origin (5). CD103⁺CD11b^– DCs promote FOXP3⁺ Tregs (6). CD103⁺CD11b⁺ DCs are known to activate protective Th17 responses in a Citrobacter rodentium–induced colitis model (5, 7).

Migration of CD103⁺CD11b^– and CD103⁺CD11b⁺ DCs to MLNs largely relies on expression of the chemokine receptor CCR7. Several transcription factors are known to regulate CCR7 expression directly, including RUNX3 (8), FOXO1 (9, 10), NF-κB, and AP1 (11). The regulation of CCR7 by LXR-α has been controversial, as Bruckner et al. showed CCR7 downregulation by LXR-α, and Feig el al. showed CCR7 upregulation by an LXR-α agonist (12, 13). However, this discrepancy was reconciled by the finding that the serine 198 phosphorylation status of LXR-α regulates CCR7 promoter activity (14). CCR7 expression can also be regulated by cytokines and lipid mediators in the tissue, including prostaglandin E2 (PGE2) (15, 16), thymic stromal lymphopoietin (TSLP) (17), and IFN-β (18).

Nr4a nuclear receptor family members include NR4A1, NR4A2, and NR4A3. These nuclear receptors function as transcription factors and are involved in various cellular processes including cell development and immune function. Nr4a family members can be induced by diverse stimuli including inflammatory cytokines, lipid mediators, calcium, and hormones in various tissues and cells including neurons, endothelial cells, adipocytes, muscles, macrophages, and T cells (19). Nr4a3 has been shown to play an important role in cell survival and proliferation in nonhematopoietic cells (20, 21). Nr4a3 expression in monocytes is probably antiatherogenic (22). In lymphocytes, Nr4a3 is involved in the negative selection of T cells, together with Nr4a1 (23). NR4A3 is expressed in macrophages (24, 25), but little is known about the role of Nr4a3 in DCs. Studies have shown that NR4A3 is expressed in BM-derived DCs (BMDCs)(26) and is induced by TLR stimulation in vitro (27). Recently, Grajales-Reyes et al. showed that CD24⁺ and CD172a⁺ cDCs from mice express high levels of Nr4a3 compared with their precursor cells pre-CD8 DCs and pre-CD4 DCs, respectively (28). However, a specific function for NR4A3 in DCs in vivo has not been defined. Therefore, in the present study, we investigated a role for Nr4a3 in DC function and determined that this transcription factor is essential for the migration of a subset of intestinal DCs.

Nr4a3 is highly expressed in migratory DCs in MLNs. To investigate a possible role for Nr4a family members in DC development and function, we analyzed lymphoid DC populations in C57BL/6J mice by flow cytometry. LN-resident DCs were divided into 3 subsets on the basis of CD4 and CD8 expression (29). Because CD8α⁺ rDCs are all CD24⁺, CD24⁺CD11b^– DCs represent CD8α⁺ DCs, and CD24^–CD11b⁺ DCs include CD4⁺ and CD4^–CD8^– DCs. Migratory DC (mDC) populations were gated using CD103 and CD11b. Detailed gating strategies for the flow cytometric analyses in this study are described in the Supplemental materials (Supplemental Figures 2–4; supplemental material available online with this article; doi:10.1172/JCI87081DS1). Unlike skin-draining LNs (SLNs), which contain only 2 mDC subsets of CD103⁺CD11b^– DCs and CD103^–CD11b⁺ DCs (30, 31), MLNs contain 3 subsets of mDC populations: CD103⁺CD11b^–, CD103⁺CD11b⁺, and CD103^–CD11b⁺ DCs (4). Given the low numbers of CD103⁺ DCs in MLNs, CD103⁺CD11b^– and CD103⁺CD11b⁺ DCs were pooled together (Supplemental Figure 1A). We measured mRNA expression of Nr4a family members in CD24⁺CD11b^– and CD24^–CD11b⁺ LN-resident DCs (rDCs) and CD103⁺ and CD103^– mDCs. Interestingly, both mDC populations expressed high levels of Nr4a3 mRNA (Figure 1A). Nr4a2 expression was extremely low in all DC populations tested. As NR4A1 and NR4A3 are also expressed in T cells, we compared NR4A1 and NR4A3 expression levels between MLN DC subsets and splenic CD4⁺ and CD8⁺ T cells. Although both CD4⁺ T cells and CD8⁺ T cells expressed NR4A1 and NR4A3, the expression levels of NR4A3 were much higher in the MLN DC subsets (Supplemental Figure 1B), suggesting that NR4A3 plays a specialized role in mDCs.

Figure 1

mDC numbers are significantly lower in MLNS from Nr4a3^–/– mice. (A) MLN DC subsets were sorted by FACS from a pool of 5 mice, and mRNA for Nr4a family members was measured by quantitative RT-PCR. Bar graph represents 2 measurements. (B) MLN DC subsets from Nr4a3^+/+ and Nr4a3^–/– mice were analyzed by flow cytometry. Data are representative of more than 4 experiments, with 3 to 6 mice per group. (C) Flow cytometric analysis of MLN mDC subsets from Nr3a3^+/+, Nr3a3^+/–, and Nr4a3^–/– mice. Data represent 1 of 2 independent experiments. Each dot represents a single mouse. Error bars indicate the mean + SD. *P < 0.05, **P < 0.01, and ***P < 0.005, by unpaired, 2-tailed Student’s t test.

On the basis of these data, we examined the number and frequency of DC populations in Nr4a1^–/– and Nr4a3^–/– mice. Although Nr4a1 expression in LN-resident DCs was quite high, we found no dramatic differences in the numbers of total LN-resident DC or mDC populations in Nr4a1^–/– mice (data not shown), so we did not explore this further.

However, we found that Nr4a3-deficient mice had significant reductions in mDC numbers (Figure 1B), as shown by the gating strategy described in Supplemental Figure 2. CD103⁺CD11b^– and CD103⁺CD11b⁺ subsets were significantly decreased, and CD103^–CD11b⁺ subsets were slightly lower in Nr4a3^–/– mice in terms of frequencies and cell numbers compared with WT mice. However, the number of total LN-resident DCs was not different, although we did detect a slight reduction in the CD8a⁺ DC subset numbers (Supplemental Figure 5A). The specific reduction of mDCs in LNs was also found in skin-draining LN (SLN) as shown in Supplemental Figure 7A (gating strategy shown in Supplemental Figure 6), and in para-aortic LNs (data not shown). Thus, the loss of mDCs in lymphoid tissue in NR4A3-deficient mice appears to be systemic. We also examined splenic DC populations, since the spleen is a special lymphoid organ that lacks mDCs. We analyzed splenic DC subsets using the gating strategy shown in Supplemental Figure 3 and found no differences in these splenic DC subsets (Supplemental Figure 5B), confirming that the defect in Nr4a3^–/– mice is selective for mDC subsets.

NR4A3-deficient mDCs have impaired migration to LNs. mDCs in the gut move from the small intestine and colon to MLNs to present antigens, including self and nonself, from tissues to T cells. To address whether the lack of mDCs in Nr4a3^-/- MLNs is caused by changes in migration or is a developmental issue, we first quantified migratory LPDCs in small intestine and colon (Supplemental Figure 5, C and D), using the gating strategy described in Supplemental Figure 4. Unlike in LNs, high expression of CD11c and MHC II was not sufficient to identify the LPDC population, given the presence of CD11c⁺MHC-II⁺ macrophages within the gut. To exclude macrophages from the LPDC population, CD64⁺ cells were excluded from LPDC gating (Supplemental Figure 4), because CD64 expression distinguishes macrophages from DCs (32, 33). Interestingly, no significant differences in small intestine or colon LPDC numbers or subsets were observed (Supplemental Figure 5, C and D). However, we found that langerin⁺CD103⁺ and CD103^–CD11b⁺ dermal DC subsets were slightly increased in ear skin (Supplemental Figure 7B). We observed no differences in the numbers of lamina propria (LP) macrophages (data not shown). These data indicate that NR4A3-deficient LPDCs are present in tissues at normal levels, suggesting that Nr4a3^–/– mDCs may have an impaired ability to migrate to LNs.

To determine whether Nr4a3 is a dominant driver of this defect, mDCs from MLNs of heterozygous Nr4a3^+/– mice were compared with DCs from WT Nr4a3^+/+ and homozygous Nr4a3^–/– mice. We found a gene dose–dependent effect of Nr4a3 on MLN total mDC and CD103⁺ DC subsets (Figure 1C), suggesting that changes in Nr4a3 expression probably drive CD103⁺ DC migration to LNs.

Defective migration of Nr4a3^–/–DCs is cell intrinsic. Since DCs acquire antigens from peripheral tissue, they express the chemokine receptor CCR7 to home to LNs in response to the CCL19/21 gradient present along lymphatic vessels. We first assessed the expression levels of CCL21 in Nr4a3^–/– mice. Mouse CCL21 includes 2 variants at position 65: CCL21-Leu and CCL21-Ser. CCL21-Leu is expressed on the initial lymphatic vessels in the periphery, and CCL21-Ser is expressed on the terminal lymphatic vessels of LN subcapsules (34). We used confocal microscopy to measure RNA (data not shown) and protein expression of CCL21-Leu on lymphatic vessels in the terminal ileum and observed no differences in CCL21 expression in the lymphatic vessels of Nr4a3^–/– mice (Supplemental Figure 8).

Next, to determine whether the loss of mDCs in lymphoid tissue was hematopoiesis derived, we performed a series of BM transplantation studies, in which BM from CD45.2 WT Nr4a3^+/+ or CD45.2 Nr4a3^–/– mice was transferred into irradiated recipient CD45.1 mice. After 8 weeks of reconstitution, we analyzed total mDCs and their subsets and found that mDCs were significantly reduced only in the mice that received NR4A3-deficient BM (Figure 2A), indicating that the defect was hematopoietic cell derived and suggesting that the defect in migration may be cell intrinsic. Next, we used a mixed-chimera approach, in which lineage-depleted hematopoietic stem cell progenitors were isolated from CD45.2 Nr4a3^–/– and CD45.1 mice, mixed at a 1:1 ratio, and transferred into sublethally irradiated CD45.1/2 heterozygous mice, as described in Methods. As a control, a 1:1 mixture of WT CD45.2 Nr4a3^+/+ and CD45.1 hematopoietic stem cell progenitors was transferred into sublethally irradiated CD45.1/2 heterozygous mice. After 10 days of reconstitution, tissues were analyzed for mDC subsets. As shown in Figure 2B, there was an equal reconstitution of CD45.1 and CD45.2 cells in the blood. Moreover, progenitors from Nr4a3^–/– mice reconstituted mDCs to normal numbers in the colon but not in the MLNs, again indicating a defective migration of these cells from tissues to LNs (Figure 2B). As a positive control, normal numbers of resident DCs from Nr4a3^–/– mice repopulated the MLNs (Supplemental Figure 9), suggesting that the defect was selective for mDCs. We conclude from these data that there is a cell-intrinsic defect in CD103⁺ mDCs in the absence of Nr4a3 that impairs their migration from tissues to LNs.

Figure 2

mDC defect in Nr4a3-deficient mice is hematopoiesis derived and cell intrinsic. (A) BM transplantation. BM from CD45.2 Nr4a3^+/+ or CD45.2 Nr4a3^–/– mice was transplanted into sublethally irradiated CD45.1 recipient mice. After 8 weeks of reconstitution, mDCs and MLN-resident DCs (rDCs) were analyzed by flow cytometry. Results represent 1 of 2 independent experiments (n = 4 mice per group). (B) Mixed BM chimeras. BM hematopoietic stem cell progenitor cells from CD45.2 Nr4a3^+/+ or CD45.2 Nr4a3^–/– mice were mixed 1:1 with CD45.1 cells to make CD45.2 Nr4a3^+/+ plus CD45.1 and CD45.2 Nr4a3^+/+ plus CD45.1 chimeras. 1:1 BM precursor mixtures were transferred into sublethally irradiated CD45.1/2 heterozygous mice. After 10 days, the mice were sacrificed for analysis. DC populations were analyzed in colon and MLNs. BM reconstitution was assessed in blood samples. The percentage of CD45.2 cells was calculated from total donor cells that included CD45.2 and CD45.1 cells. Results represent 1 of 2 independent experiments (n = 3–4 mice per group). *P < 0.05 and ** P < 0.01, by an unpaired 2-tailed Student’s t test.

Migration defect of Nr4a3^–/– DCs is due to impaired CCR7 expression. We next explored mechanisms that would explain the defective migration of CD103⁺ DCs in the absence of NR4A3, focusing on expression of CCR7 on mDCs. The migration of DCs from intestinal LP to MLNs is known to be CCR7 dependent (2). We measured the expression of Ccr7 mRNA and surface CCR7 in total mDCs sorted from MLNs. Although mDCs in WT mice expressed abundant Ccr7 mRNA, the few remaining mDCs in Nr4a3^–/– mice expressed almost no Ccr7 mRNA (Figure 3A). Furthermore, CCR7 expression was markedly reduced on the surface of Nr4a3^–/– mDCs in MLNs (Figure 3, B and C). CD103⁺CD11b^– and CD103⁺CD11b⁺ subsets of mDCs from Nr4a3^–/– mice showed similar results (data not shown).

Figure 3

The migration defect of Nr4a3^–/– DCs is CCR7 dependent. (A) Ccr7 mRNA expression. Plasmacytoid DCs (CD11c^med, MHC-II^med, PDCA-1⁺, CD3^–, CD19^-), LN-resident DCs (CD11c⁺, MHC-II^med, CD3^–, CD19^–), and mDCs (CD11c⁺, MHC-II^hi, CD3^–, CD19^–) from WT and Nr4a3^–/– mice were sorted by FACS. Data represent pools of 5 mice per group. Results represent 1 of 2 independent experiments. (B) Histogram of CCR7 surface expression on mDCs. (C) MFI of CCR7 on mDCs (n = 6 mice per group). Results represent 1 of 3 independent experiments. (D) CD11c^YFP WT or CD11c^YFP Nr4a3^–/– mice were treated with R848 for 2 hours in vivo. Confocal whole-mount images of mesenteric lymphatic vessels from R848-treated CD11c^YFP WT or CD11c^YFP Nr4a3^–/– mice were analyzed using Imaris software. Lymphatics (blue) were isosurfaced, and distance transformation was applied to create a new channel that encoded the 3D distance from the lymphatic vessels. CD11c^YFP cells (green) were isosurfaced and classified as being associated with lymphatics (red) when the distance from the lymphatics was less than 15 μm. To normalize between the different data sets, the number of cells associated with lymphatics was divided by the volume of lymphatic vessels (expressed in μm³ and divided by 100,000). (E) Numbers of YFP⁺ cells associated with lymphatic vessels (red). (F) Numbers of YFP⁺ cells away from lymphatic vessels (green). (G) Day-7 BMDCs were stimulated with 0.5 μg/ml LPS for 36 hours. CD11c⁺MHC-II⁺⁺ cells were gated for CCR7 expression. (H) MFI of CCR7 expression on CD11c⁺MHC-II⁺⁺ cells was analyzed by flow cytometry. Representative histogram (top) and dot plot of individual mouse samples (bottom). Max, maximum. (I) Expression of Foxo1 mRNA and FOXO1, p-AKT, AKT, and β-actin proteins from CD11c⁺ BMDCs, stimulated with or without LPS. Results from G and I represent 1 of 3 to 4 independent experiments. *P < 0.05, **P < 0.01, and ***P < 0.001, by unpaired, 2-tailed Student’s t test.

We next used CD11c^YFP mice that either expressed Nr4a3 (WT) or had been crossed with Nr4a3^-/- mice (KO) to visualize mDCs in lymphatic vessels of the gut. Migrating CCR7⁺ LPDCs attach to CCL21 to travel along the lymphatics. Since CCR7⁺ DCs from LP are difficult to detect by flow cytometry and we also wished to visualize the mDCs within the lymphatic vessels, we used confocal microscopy. To stimulate migration, CD11c^YFP WT and CD11c^YFP Nr4a3^–/– mice were treated in vivo with the TLR7 agonist R848, and the mesentery lymphatic vessels (Figure 3D, blue) were imaged 2 hours later. As shown in Figure 3, D and E, the number of YFP⁺CD11c⁺MHC-II⁺ cells attached to lymphatic vessels (red cells) was significantly lower in mice lacking NR4A3. As a control, we found that the numbers of cells located away from lymphatic vessels (green cells) by more than 15 μm were not significantly different (Figure 3F). These results support the idea that the migration defect of Nr4a3-deficient mDCs is caused by reduced CCR7 expression, although they do not rule out an activation defect.

We also assessed CCR7 expression on Nr4a3^–/– DCs using a BMDC culture with granulocyte macrophage–CSF (GM-CSF) that mimicked monocyte-derived inflammatory DCs, because we observed that Nr4a3 expression increased during the course of BMDC development with the addition of GM-CSF (data not shown). BMDCs cultured for 7 days in the presence of GM-CSF were stimulated with LPS for 36 hours to stimulate TLR4 signaling and were then analyzed for CCR7 and DC surface marker expression. First, we observed that the numbers of CD11c⁺MHC-II⁺⁺ cells from the Nr4a3^–/– cell culture were 40% lower than the numbers detected in WT cell culture. However, after LPS treatment, the fold increase in frequency was similar to that observed in WT cell culture, suggesting that the TLR-dependent DC activation pathway may be comparable to that in WT cells (Figure 3G). However, CCR7 expression on Nr4a3^–/– DCs was much lower than that on WT DCs, even at baseline (Figure 3H). Further, although LPS stimulation increased the number of CCR7⁺ BMDCs in Nr4a3^–/– mice, the mean fluorescence intensity (MFI) of CCR7 as well as the frequency of CCR7⁺ DCs were much lower than those in WT cells (Figure 3H).

To understand how Nr4a3 regulates Ccr7, we first performed a series of promoter-reporter assays. We analyzed 3.3 kb of the mouse Ccr7 promoter upstream from the transcriptional start site using MacVector software and found 2 NR4A-binding motifs (NBRE, monomer-binding motif) and 5 putative C/EBP α/β–binding motifs (Supplemental Figure 10A). We cloned this region into pGL4-luciferase and found no evidence for a direct regulation of Ccr7 promoter activity by NR4A3 in vitro (Supplemental Figure 10B). C/EBP β was used as a positive control for regulating Ccr7 promoter activity (15). We concluded that CCR7 expression must be regulated by NR4A3 through an indirect mechanism.

As FOXO1 has been shown to be a critical transcription factor for CCR7 expression in T cells (9) and DCs (10), we next measured Foxo1 mRNA and protein expression using CD11c⁺ BMDCs. As shown in Figure 3H, LPS drove the induction of CCR7 expression in DCs. Foxo1 mRNA and protein expression levels were increased by LPS stimulation in WT Nr4a3^+/+ cells (Figure 3I). Foxo1 mRNA was also slightly increased by LPS treatment of Nr4a3^–/– DCs, but to a lesser extent (Figure 3I). This did not correlate with FOXO1 protein expression, which was lower in Nr4a3^–/– mouse DCs, both at baseline and after treatment with LPS (Figure 3I). FOXO1 protein levels are primarily regulated through phosphorylation of the AKT pathway, which directs FOXO1 for polyubiquitination and degradation (35). We found increased AKT phosphorylation (at serine 473) in both unstimulated and LPS-stimulated Nr4a3^–/– DCs (Figure 3I), suggesting that the observed reduction in FOXO1 protein was likely caused by AKT activation, which triggered FOXO1 degradation.

Migration defect of Nr4a3^–/–DCs impairs T cell activation in MLNs. Retinoic acid plays an important role in MLN DCs to aid in the induction of FOXP3⁺ Tregs and to imprint activated T cells to migrate to gut tissue via CCR9 induction (36, 37). The key enzyme for these functions is retinaldehyde dehydrogenase 2 (RALDH2, encoded by Aldh1a2), which is highly expressed in LPDCs. First, we measured mRNA expression of Aldh1a2, encoding RALDH2, from flow cytometry–sorted total mDCs, LN-resident DCs, and plasmacytoid DCs (pDCs) from MLNs (Supplemental Figure 11A). Only WT mDCs expressed high levels of Aldh1a2. We found that Nr4a3^–/– mDCs did not express Aldh1a2. We next measured RALDH enzyme activity by ALDEFLUOR assay. LPDCs and migratory MLN mDCs from WT mice expressed high levels of RALDH. Although Nr4a3^–/– LPDCs in the small intestine expressed RALDH, we found that Nr4a3^–/– migratory MLN mDCs did not express RALDH. We also found that LP macrophages did not express RALDH (data not shown). Thus NR4A3-deficient CD103⁺ DCs express fully functional RALDH that potentially could allow them to induce iTregs and CCR9 expression, if they are able to successfully migrate to MLNs.

We next examined whether Tregs and T effector cells in Nr4a3^–/– mice are affected by the loss of CD103⁺ DCs in steady-state MLNs. We observed no differences in the composition of CD4⁺ T cells, CD8⁺ T cells, or B cells (Figure 4A). However, we detected reduced frequencies of FOXP3⁺ Tregs (Figure 4B) and CCR9⁺ T cells (Figure 4C) and significantly reduced frequencies of IFN-γ⁺CD4⁺ T cells (Figure 4D). We did not observe differences in IL-17A⁺CD4⁺ T cells (Figure 4D) in MLNs. As CCR7 is important for T cell homing to LNs, we assessed CCR7 expression on T cells and found that it was not affected by the absence of Nr4a3 expression (Figure 4E). Thus, we conclude that the CCR7 defect observed in Nr4a3^–/– mice is limited to mDCs and likely results in decreased percentages of CCR9⁺ T cells and IFN-γ⁺ T cells in MLNs.

Figure 4

Nr4a3 deficiency does not change lymphocyte composition or CCR7 expression on T cells but impairs steady-state T cell activation in MLNs. Percentages of T cells and B cells (A) and of CD4⁺CD25⁺FOXP3⁺ T cells (B) and CD4⁺CCR9⁺ T cells (C) were measured by flow cytometry. Steady-state IFN-γ⁺CD4⁺ T cells and IL-17A⁺CD4⁺ T cells were measured (D). Histogram showing CCR7 expression on CD4⁺ and CD8⁺ T cells and plot showing the MFI of CCR7 (E). Data represent 1 of 3 independent experiments. *P < 0.05 and ***P < 0.001, by unpaired, 2-tailed Student’s t test.

CCR7 expression is not induced on CD103⁺ DCs in the absence of Nr4a3. We next asked whether Nr4a3-deficient mDCs could be mobilized to LNs by the induction of inflammatory cytokine pathways. Oral administration of the TLR7 agonist R848 has been shown to mobilize LPDCs to MLNs through pDC-mediated type 1 IFN and TNF-α production (38), which causes the induction of CCR7 on DCs (1). To determine whether R848 oral gavage could mobilize CD103⁺ DCs, we treated mice with R848 and examined the mDC content in MLN DCs at 14 hours. Like Nr4a3^–/– mice, Ccr7^–/– mice have normal numbers and subsets of LPDCs in the gut, but deficient numbers of mDCs in MLNs (39). CD103⁺CD11b^– DC and CD103⁺CD11b⁺ mDC subsets from Ccr7^–/– mice were dramatically reduced at steady state (Supplemental Figure 12A), indicating that these DCs utilize CCR7 for LN homing. However, CD103^–CD11b⁺ DCs still migrated in the absence of CCR7 (Supplemental Figure 12A). We observed an increase in total mDC frequencies from all strains upon R848 treatment (Supplemental Figure 12B). Total numbers of mDCs from Nr4a3^–/– mice were similar to those from Ccr7^–/– mice (Supplemental Figure 12C). However, we did not observe an increase in CCR7 on mDCs from Nr4a3^–/– mice in response to R848 (Supplemental Figure 12D).

Finally, we asked whether bacterial infection could induce CCR7 and LPDC migration in Nr4a3^–/– mice by stimulating TLR-dependent CCR7 induction (40) in LPDCs. Nr4a3^–/– mice were infected with C. rodentium, a murine model of E. coli infection in humans (41). Nr4a3^–/– mice were more susceptible to C. rodentium infection than were WT mice (Figure 5A). Since most of the deaths occurred 8 days after infection, we asked whether the adaptive immune response in Nr4a3^–/– mice was unable to control the infection. We analyzed CD4⁺ T cells from MLNs and colon on day 8 after infection. In MLNs, IFN-γ⁺ Th1 and IL-17A⁺ Th17 cell numbers were significantly lower in Nr4a3^–/– mice (Supplemental Figure 13). However, these differences were not observed in the colon (Supplemental Figure 13). Similar to what we observed at steady state, the number of MLN DCs from Nr4a3^–/– mice after infection was significantly lower than that of MLN DCs from WT mice (Figure 5B). However, CD103⁺CD11b^– LPDC numbers from Nr4a3^–/– mice were slightly decreased (Figure 5C), unlike in the steady state, in which no differences were observed between WT and Nr4a3^–/– mice. Moreover, C. rodentium infection did not cause an increase in the surface expression of CCR7 on Nr4a3^–/– mDCs (Figure 5D). These data collectively suggest that CCR7 expression is markedly reduced on mDCs in Nr4a3^–/– mice under homeostatic conditions and cannot be increased by TLR-dependent (C. rodentium infection) or cytokine-dependent (R848 treatment) activation in vivo.

Figure 5

Nr4a3^–/– mice are susceptible to C. rodentiuminfection, but CCR7 expression on mDCs is not increased. (A) Nr4a3^–/– and Nr4a3^+/+ mice were infected with 1 × 10⁹ CFU C. rodentium by oral gavage. The appearance and survival of infected mice were monitored daily for 21 days.**P < 0.01, by log-rank (Mantel-Cox) test. Results represent 1 of 2 independent experiments (n = 9–10 mice per group). (B–D) MLN DC (B) and colon LPDC (C) subsets were analyzed by flow cytometry on day 8 after infection. The graphical summaries are representative of 4 mice per group. (D) Histograms plotting CCR7 expression of MLN DC subsets. A graphical summary of the MFI is shown. Results represent 1 of 2 independent experiments (n = 4 mice per group). *P < 0.05 and **P < 0.01, by unpaired, 2-tailed Student’s t test.

NR4A3 also regulates the mitochondrial function of DCs during maturation. Pearen and coauthors reported their transcriptomic analysis of muscle-specific Nr4a3-trangenic mice and showed that more than 25% of differentially regulated genes were mitochondrial genes (42). This would be consistent with the data of Everts and coauthors, who showed that blocking glycolysis inhibits TLR-dependent DC activation and CCR7 expression (43). Thus, we hypothesized that Nr4a3 plays a role as a metabolic regulator of DC maturation and activation.

To compare bioenergetic functions of WT and Nr4a3^–/– DCs, we used the Seahorse XF^e analyzer to perform a modification of the Mito Stress Test that probes different metabolic states of mitochondria using sequential additions of the ATP synthase inhibitor oligomycin, carbonilcyanide p-triflouromethoxyphenylhydrazone (FCCP) uncoupler, and the respiratory inhibitor antimycin A (44). Both WT and Nr4a3^–/– BMDCs were found to possess some spare respiratory capacity (Figure 6A). This means that these cells are capable of meeting their basic energy demands and compensate for a moderate increase under stress conditions. The observed lower level of basal respiration shown by Nr4a3^–/– BMDCs compared with that of the WT BMDCs is most likely due to a downregulated energy demand of the cells rather than a deficiency in the number of mitochondria. The above-mentioned preservation of the spare respiratory capacity, maintenance of normal ATP levels (Supplemental Figure 14), and lack of compensatory increase (in fact, a decrease) in the extracellular acid production rate (ECAR) (Figure 6B) all point to this conclusion. However, the maximal respiratory capacity of Nr4a3^–/– BMDCs (induced by FCCP) is approximately one-third lower than that of WT cells.

Figure 6

Nr4a3 regulates mitochondrial function in BMDCs. (A) CD11c⁺ BMDCs were plated onto a XF96 plate, and OCRs and ECARs were measured with sequential treatment of oligomycin A (Oligo), FCCP, and antimycin A (Ant A) using a SeaHorse XFe 96 analyzer. After analysis, the cell numbers were measured by Hoechst 33342 staining for normalization of the ECARs and OCRs. The mean value of the relative OCRs is shown. A diagram of the OCR plot is shown. (B) Relative OCRs and ECARs are shown. P < 0.0001, by 2-way ANOVA for genotype, treatment, and interaction. Results represent 1 of 3 independent experiments. Each data point represents 5 technical replicates.

Oxidative phosphorylation inevitably accompanies ROS production (45), and GM-CSF highly increases ROS production in BMDCs (46). We asked whether the lower basal respiratory capacity of Nr4a3^–/– BMDCs is due to a smaller quantity of mitochondria or to a mitochondrial functional defect. We measured mitochondrial content and ROS production of CD11c⁺ BMDCs with MitoTracker and 2′,7′-dichlorofluorescin diacetate (DCFDA), respectively (Figure 7A). Although mitochondrial content was not different between WT and Nr4a3^–/– BMDCs, ROS production was significantly lower in Nr4a3^–/– BMDCs (Figure 7B). Our data suggest that the observed lower level of basal respiration is probably due to suboptimal mitochondrial function in Nr4a3^–/– BMDCs, but not to the cellular quantity of mitochondria. Thus, our data indicate that NR4A3 also plays a role in maintaining homeostatic mitochondrial function in DCs. Therefore, in addition to being important for the regulation of CCR7 expression, NR4A3 also appears to aid in maintaining adequate cellular respiration of mDCs. However, whether this role for NR4A3 in maintaining cellular respiration has any direct impact on CCR7 expression in this cell type is unclear.

Figure 7

ROS production is lower in day-8 BMDCs from Nr4a3^–/– mice. (A) CD11c⁺ BMDCs were stained with MitoTracker Red CMXRos for mitochondria, DCFDA for ROS, and anti–DC-SIGN for the surface of BMDC and Hoechst 33342 for nuclei. Scale bars: 20 μM. (B) DC-SIGN⁺ cells were plotted for fluorescence intensity of MitoTracker and dichlorodihydrofluorescein diacetate (DCFDA). Each dot on the plots represents a single cell. ***P < 0.0001, by unpaired, 2-tailed Student’s t test. Results are representative of 2 independent experiments.

The Nr4a nuclear receptor family consists of 3 highly homologous proteins that can be simultaneously induced by similar stimuli. However, their expression pattern is both tissue and cell-type specific, indicating that some of their functions are nonredundant. Each Nr4a family member has its own specific, nonredundant function. For example, Nr4a1 plays a selective role in monocyte development (47) that cannot be compensated by Nr4a2 or Nr4a3. However, FOXP3⁺ Treg development is complemented by other family members, and only through deletion of Nr4a1, Nr4a2, and Nr4a3 together is FOXP3⁺ T cell development ablated in vivo (48). Here, we show that CD103⁺ mDCs express higher levels of Nr4a3 than either Nr4a1 or Nr4a2 and that the migratory capability of CD103⁺ DCs is uniquely lost in the absence of Nr4a3.

In the current study, we show that Nr4a3 expression in CD103⁺ mDCs is critical for regulating the migratory function of DCs from tissues to LNs in vivo. Nr4a3 expression is selectively high in the CD103⁺ mDC subset compared with other DC populations, including classical CD11c^hi DCs. In the absence of Nr4a3, there are markedly reduced numbers of CD103⁺ mDCs in the LNs. Interestingly, CD103⁺ DC numbers in tissues are relatively normal, indicating that Nr4a3 is not critical for the development of these cells from the precursors present in BM or blood.

CCR7 is required for the migration to LNs of lymphocytes, including T cells (49–51). CCR7 is also quite important for lymphocyte development in the thymus (52, 53). We found that the reduced levels of CCR7 expression in Nr4a3^–/– mice appeared to be confined to the mDC population, because we observed no differences in CCR7 expression on LN T cells (Figure 4E) or B cells or on developing T cells in the thymus (data not shown). The migratory CD103⁺ mDC subset contained much higher expression levels of Nr4a3 compared with levels in B or T cells, suggesting that Nr4a3 plays a major functional role in this DC subset, and perhaps not much of one in T or B cells.

Although Nr4a3^–/– mice have systemic mDC defects, we and others did not observe any major health problems, except spinning behavior due to an inner ear defect (54). We investigated gut pathology in the small intestine, cecum, and colon of these mice at approximately 8 to 12 weeks of age and found no evidence of colitis (data not shown). This might be partly due to the fact that the CCR7 defect is specific to DCs and not to T cells. Also, 12 weeks may not be enough time to show inflammatory disease phenotypes based on a DC migratory defect.

DCs mature through both TLR-independent and -dependent pathways, yet both maturation pathways involve the induction of CCR7 expression. To test whether the regulation of DC maturation by NR4A3 occurs through a TLR-dependent or -independent pathway, we performed a series of experiments to test the induction of CCR7 by these pathways both in vitro and in vivo. We found that neither pathway induced CCR7 expression or DC maturation in Nr4a3^–/– mice. Further, our data suggest that not only CD103⁺ mDCs, but also monocyte-derived inflammatory DCs from Nr4a3^–/– mice are unable to migrate to LNs in response to external inflammatory stimuli such as C. rodentium. Thus, CCR7 cannot be induced by several known activation pathways in Nr4a3-deficient mDCs, indicating that Nr4a3 is definitely required for CCR7 induction in CD103⁺ mDCs.

Interestingly, we found that a likely mechanism for defective CCR7 expression in Nr4a3^–/– DCs is the posttranscriptional degradation of FOXO1 protein. FOXO1 has been shown to be critical for the regulation of CCR7 activity in DCs through its direct binding to the Ccr7 promoter (10). Phosphorylation by activated AKT excludes FOXO1 from the nucleus and targets it for polyubiquitination for proteasomal degradation. FOXO1 protein was decreased in Nr4a3^–/– BMDCs, despite a slight increase in mRNA expression upon LPS stimulation of DCs. Concomitant with this, we observed an increase in phosphorylated AKT (p-AKT), both at baseline and especially after LPS stimulation. We speculate that NR4A3 regulates AKT phosphorylation in mDCs to prevent the proteasomal degradation of FOXO1, thereby controlling CCR7 expression and influencing the migratory capacity of CD103⁺ DCs. Alternatively, NR4A3 could possibly directly interact with FOXO1 to inhibit its degradation. Briand et al. (55) have shown that NR4A1, an NR4A family member, physically and directly binds to FOXO1 in pancreatic β cells to repress FOXO1 action on insulin secretion, suggesting that NR4A3 may also be able to interact with FOXO1.

Moreover, we observed that Nr4a3-deficient mice are more susceptible to death from C. rodentium. We anticipate that the sole loss of mDC function does not fully explain the susceptibility of Nr4a3^–/– mice to C. rodentium infection, because Ccr7^–/– mice are resistant to C. rodentium infection, even though they also have no mDCs present in MLNs (56). As such, the susceptibility of Nr4a3^–/– mice to C. rodentium infection suggests additional defective innate immune functions of LPDCs and MLN DCs in Nr4a3^–/– mice. It is likely, on the basis of these data, that fully functional gut-resident DCs may be crucial for resistance to C. rodentium infection in vivo.

The defect in mDC numbers in MLNs in the mixed BM chimeras (Figure 2B) was slightly less severe than that observed in both the global Nr4a3-KO and full chimeric mice (Figure 2A). We believe that this is probably due to the following scenario: CCR7 responds to 2 chemokines — CCL21, which is present on lymphatics, and CCL19, which is produced by the mDCs. In the case of the mixed chimera, the presence of WT DCs will generate some CCL19 that can be recognized by CCR7, although the expression of CCR7 on NR4A3-deficient mDCs is very low. In the case of the Nr4a3^–/– mouse, there is no CCL19 produced at all due to the absence of mDCs, so the phenotype may be more severe. Alternatively, it is quite likely that NR4A3 has additional extrinsic effects that influence the migration of DCs to LNs.

When pre-cDCs extravasate to tissues for terminal differentiation, they face an oxygen gradient from the blood vessel. Hypoxia has been shown to enhance both inflammatory and homeostatic DC maturation, including the regulation of CCR7 expression (57) via HIF-1α (58). Resting DCs rely more on oxidative phosphorylation than on glycolysis (59), which reduces energy expenditure and extends the lifespan of cells. However, DCs instantly turn on glycolysis and fatty acid synthesis when they are stimulated by TLR ligands (43). One possibility is that Nr4a3 functions as an important metabolic switch for CD103⁺ DC maturation and activation in response to hypoxia. Indeed, muscle-specific Nr4a3-transgenic mice showed increased expression of genes involved in energy metabolism, including those that regulate mitochondrial function, oxidative phosphorylation, the TCA cycle, and fatty acid metabolism (42). Concordantly, we found that the mitochondrial function (basal and maximal respiratory capacity) of Nr4a3^–/– BMDCs is decreased compared with that of WT DCs in the basal state. Lower levels of mitochondrial function of Nr4a3^–/– DCs were also confirmed by microscopy and ROS production. As such, in the absence of NR4A3, CD103⁺ DCs could not migrate in response to TLR ligands, suggesting impaired respiration in the absence of NR4A3. This is supported by studies by Mancino et al., who showed that DCs with impaired respiratory capacity that were cultured under hypoxic conditions and transferred into healthy recipients are unable to efficiently migrate to LNs (60). Although a link between energy metabolism and CCR7 regulation exists in the literature (57–59), it is not clear from our studies that NR4A3 directly regulates CCR7 through changes in energy metabolism. Rather, we anticipate that the regulation of CCR7 by NR4A3 is independent of the effect of NR4A3 on mitochondrial respiration. Thus, NR4A3 appears to control genes that happen to impact both mitochondrial respiration and the FOXO1-mediated regulation of CCR7. Both mitochondrial respiration and the FOXO1/CCR7 pathways clearly function to influence CD103⁺ DC migration, and both are regulated by NR4A3 but are probably independent of each other.

In summary, the transcription factor NR4A3 plays a critical role in CD103⁺ DC migration to LNs. Mice lacking Nr4a3 have reduced numbers of mDCs in lymphatics and LNs. Nr4a3 is highly expressed in mDCs and indirectly regulates CCR7 expression through FOXO1. In the absence of Nr4a3, mDCs are unable to increase CCR7 expression, either through TLR-dependent or -independent pathways, leading to impaired immune responses and survival against pathogens.

Mice. Nr4a3^–/– and littermate WT Nr4a3^+/+ mice were gifts of Dennis Bruemmer (University of Pittsburgh, Pittsburgh, Pennsylvania, USA) and Orla Conneely (Baylor College of Medicine, Houston, Texas, USA) and were originally generated and described by Orla Conneely (54). CD11c^YFP mice [B6.Cg-Tg(Itgax-Venus)1Mnz/J, stock 008829] and congenic CD45.1 (B6.SJL-Ptprc^a Pep3^b/BoyJ, stock 002014) mice were purchased from The Jackson Laboratory. CD11c^YFP Nr4a3^+/+ and CD11c^YFP Nr4a3^–/– mice were generated by crossing CD11c^YFP mice with Nr4a3^–/– mice. All mice were fed a standard chow diet containing 0% cholesterol and 5% calories from fat (PicoLab diet 5053; LabDiet). All mice were bred and housed in microisolator cages in a pathogen-free animal facility at the La Jolla Institute for Allergy and Immunology.

Flow cytometry. The following antibodies were used in the experiments: CD103 (clone 2E7, BioLegend); CCR7 (clone 4B12, BioLegend); CD64 (clone X54-5/7.1, BioLegend); CD24 (clone M1/69, BD); CD25 (clone 7D4, BD); I-A/I-E (clone M5/114.15.2, BioLegend); CD45.1 (clone A20, BD); PDCA1 (clone 927, BioLegend); CD45 (clone 30-F11, BioLegend); CD8α ( clone 53-6.7, BioLegend); CD3 (clone 145-2C11, BioLegend); CD19 (clone 6D5, Life Technologies); CD11c (clone N418, BioLegend); CD11b (clone M1/70, BD); CD45.2 (clone 104, BioLegend); FOXP3 (clone FJK-16S, eBioscience); CCR9 (clone CW-1.2, eBioscience); TCR-β (clone H57-597, BioLegend); CD4 (clone RM4-5, BioLegend); Thy1.2 (clone 53-2.1, BioLegend); IL-17A (clone eBio17B7, eBioscience); and IFN-γ (clone XMG1.2, BD).

For flow cytometric analysis, cells were incubated with 1 to 2 μg anti-mouse CD16/CD32 (BD) and LIVE/DEAD Fixable Yellow (Life Technologies, Thermo Fisher Scientific) for 15 minutes at 4°C, and the antibody mixture was added and incubated for 30 minutes at 4°C. Tregs were stained with a FOXP3 staining kit (eBioscience). For intracellular cytokine staining, cells were stimulated with 50 ng/ml PMA and 500 ng/ml ionomycin for 4 hours in the presence of GolgiPlug (BD). After surface marker staining, cells were fixed and permeabilized with Cytofix/Cytoperm Fixation/Permeabilization (BD) for cytokine staining. Data were collected using an LSR II Flow Cytometer (BD) and analyzed with FlowJo 9.8 software.

Quantitative RT-PCR. Taqman assays of Nr4a1, Nr4a2, Nr4a3, Ccr7, and Hprt were purchased from Applied Biosystems (Life Technologies, Thermo Fisher Scientific). Total RNA was isolated using the TRIzol method (Life Technologies, Thermo Fisher Scientific) and/or an RNeasy Mini Kit (QIAGEN). cDNA was synthesized by using the SuperScript II Reverse Transcriptase System (Life Technologies, Thermo Fisher Scientific). Gene expression reactions were performed using the Taqman Universal Assay Reagent (Applied Biosystems) and a My-iQ II instrument (Bio-Rad). The housekeeping gene Hprt was used to normalize the data.

Tissue digestion for intestine, LNs, and spleen. Small intestine (duodenum, jejunum, and ilium) and colon were excised. Food matter and feces were removed, washed with HBSS containing 5% FBS, and mesentery tissue, fat tissue, and Peyer’s patches were carefully removed. The small intestine and colon were opened by longitudinal excision. After washing with HBSS containing 5% FBS, the gut tissues were cut into 1-inch-long pieces. The gut tissues were incubated with 25 ml HBSS containing 5% FBS, 5 mM EDTA, 1 mM DTT, and 25 mM HEPES for 15 minutes at 37°C with shaking at 200 rpm. Epithelial cells and intraepithelial lymphocytes were removed, and the remaining tissue was washed 3 times with HBSS containing 5% FBS and 25 mM HEPES. The gut tissue was digested with 25 ml RPMI containing 5% FBS, 25 mM HEPES, 1 mg/ml collagenase D (Roche), and 50 μg/ml DNase I (Roche) for 40 minutes at 37°C with shaking at 200 rpm. After passing through a 100-μm cell strainer, cells were washed with HBSS containing 5% FBS and 25 mM HEPES. Next, cells were isolated by 80% and 40% Percoll (GE Healthcare) gradient centrifugation at 900 g for 20 minutes at room temperature, without a break. Interface cells were isolated, washed, and counted with a Vi-CELL counter (Beckman Coulter).

LNs and spleens were excised and chopped after removing the fat tissue. Chopped tissues were digested with RPMI containing 5% FBS, 25 mM HEPES, 1 mg/ml collagenase VIII (Sigma-Aldrich), and 50 μg/ml DNase I (Roche) for 30 minutes at 37°C, with occasional shaking. Cells were passed through a 70-μm cell strainer and washed with PBS containing 2% FBS. Splenic cells were treated with RBC Lysis Buffer (BioLegend) to remove rbc and washed with PBS containing 2% FBS. Total cell numbers were counted with a Coulter Z2 (Beckman Coulter) or a Vi-CELL Counter.

BM stem cell progenitor cell transfer. Stem cell progenitor cells from BM were isolated from CD45.2 Nr4a3^+/+, CD45.2 Nr4a3^–/–, and CD45.1 mice using a Mouse Hematopoietic Cell Isolation Kit (STEMCELL Technologies). Briefly, biotinylated antibodies against nonhematopoietic stem and progenitor cells (CD5, CD11b, CD19, CD45R, 7-4, Ly-6G/C [Gr-1], and TER119) were used to remove lineage-positive cells from total BM cells. Enriched lineage-negative cells were used as stem cell progenitor cells. A 1:1 mixture of CD45.1 and CD45.2 stem cell progenitor cells was adoptively transferred into sublethally irradiated (2.75 Gy ) CD45.1/2 heterozygous mice via the retro-orbital route. After 10 days, reconstitution of CD45.1 and CD45.2 in blood, colon, and LNs was analyzed by flow cytometry.

Confocal microscopy. CD11c^YFP Nr4a3^+/+ and CD11c^YFP Nr4a3^–/– mice were fed with 10 μg R848 and sacrificed 2 hours later. Mesentery tissue was collected and fixed with 4% paraformaldehyde for 16 hours. After washing out fixatives, the tissues were incubated with PBS containing 2% saponin, 2% FBS, and 0.1% sodium azide for at least 18 hours, as described by Chodaczek et al. (61). Then, the tissues were stained with phycoerythrin-conjugated (PE-conjugated) anti–MHC II (BD Biosciences) and eFlour 660–conjugated anti–Lyve-1 antibody (1:100; eBioscience) for 18 hours. After washing, the samples were mounted in Prolong Gold Antifade Reagent (Life Technologies, Thermo Fisher Scientific) with a number 1.5 borosilicate glass coverslip. An SP5 resonant laser-scanning confocal system mounted on a DM 6000 upright microscope (Leica Microsystems) with a ×25 water-immersion objective (numerical aperture, 0.95) was used for fluorescence microscopy. Image acquisition was done at room temperature using Leica Application Suite Advanced Fluorescence software. Images used for 3D reconstructions were acquired with a 2-μm Z-step size and 488-nm and 633-nm excitation wavelengths to visualize yellow fluorescent protein–labeled (YFP-labeled) cells and eFluor 660–labeled lymphatic vessels. Fluorescent signals were detected with internal photomultiplier tubes. Image analysis was done using Imaris 8.12 software (Bitplane). Lyve-1–labeled structures were isosurfaced, and distance transformation was applied to create a new channel that encoded the 3D distance from the lymphatic vessels. CD11c^YFP cells were isosurfaced and classified as being associated with lymphatics when the distance from the nearest Lyve-1–positive structure was less than 15 μm. To normalize between data sets, the number of cells associated with lymphatics was divided by the volume of lymphatic vessels (expressed in μm³ and divided by 100,000).

For ROS visualization, CD11c⁺ cells were isolated from day-7 BMDC cultures and plated on pre-cleaned 1-mm glass coverslips overnight. The cells were mounted on a microscope slide with HBSS and placed on a heated plate at 33.5°C. To visualize DCs, mitochondria, and oxidative function, all cells were stained with APC-conjugated APC Armenian Hamster Anti-mouse DC-SIGN (1:100; eBioscience); MitoTracker Red CMXRos (200 μm/ml; Thermo Fisher Scientific); and DCFDA (5 μm/ml; Abcam) for 30 minutes at 37°C, respectively. To visualize nuclei, Hoechst 33342 (Thermo Fisher Scientific) was subsequently added at 10 μl/ml for 5 minutes at 37°C. Cells were washed and imaged within the hour. Leica Acquisition Suite Advanced Fluorescence software was used to capture 3 images per slide, with visible lasers at 488-nm, 543-nm, 633-nm excitation wavelengths, an infrared laser tuned to 800 nm, and a 1.8-μm Z-step. Imaris 8 software (Bitplane) was used to mask DC-SIGN⁺ cells and analyze pixel intensity for each channel. The mean intensity for MitoTracker Red and DCFDA for each whole cell was plotted for each experiment.

BMDC culture. BM cells from Nr4a3^+/+ and Nr4a3^–/– mice were cultured by following a previously described method (62). Briefly, 2 million BM cells were plated onto a 6-well plate with RPMI containing 10% FBS, 20 ng/ml GM-CSF (Peprotech), 2 mM glutamine, 50 μM β-mercaptoethanol, and 1× antibiotic-antimycotic (Gibco, Life Technologies, Thermo Fisher Scientific). Fifty percent of the culture media was changed every other day. Cells cultured between days 7 and 9 were used for experiments.

Ccr7 promoter analysis and transcription factor binding site prediction. 3.3kb upstream DNA sequence from the translation initiation site of the mouse Ccr7 promoter was analyzed by MacVector software (version 12.7.5.). Transcription factor binding sites were predicted with 90% homology setting using JASPAR database. This 3.3kb of the murine Ccr7 promoter sequence was cloned into pGL4 luciferase vector (Promega). cDNA expression clones of Nr4a3 and C/EBP beta genes cloned into pCMV6-Entry vector were purchased from OriGene. Constitutive beta-galactosidase plasmid, pGL4 Ccr7 reporter plasmid and Nr4a3 or C/EBP plasmids were co-transfected into RAW264.7 or 293T cells. After 18 hrs, cells were lysed and a dual reporter assay was performed using a single automatic injection Mithras (Berthold technologies) luminometer following the manufacturer’s protocol (Promega). Ratios of Luciferase activity and beta-galactosidase activity were normalized to an empty reporter construct.

Immunoblotting. CD11c⁺ cells isolated from day-8 BMDC cultures were plated at 2 million cells per well and stimulated with or without 0.5 μg/ml LPS for 18 hours. Cells were lysed with RIPA buffer. After total protein was quantified by using a Pierce BCA Protein Assay Kit (Thermo Fisher Scientific), an equal amount of protein was loaded and separated in a 4% to 12% Bis-Tris NuPAGE Precast Gel (Invitrogen, Thermo Fisher Scientific). Protein was transferred to a 0.45-μm nitrocellulose membrane. The membrane was blocked and incubated with the following antibodies: anti-FOXO1, anti-AKT, anti–p-AKT (Ser473), and anti–β-actin (all from Cell Signaling Technology). Primary antibodies were detected by HRP-conjugated anti-rabbit IgG, and blots were developed using SuperSignal West Pico Chemiluminescent Substrate (Thermo Fisher Scientific).

Infection model. Nr4a3^–/– and Nr4a3^+/+ mice were cohoused and maintained on nonacidic water for more than 4 weeks before steady-state and infection experiments. Age- and sex-matched male and female mice were used for all experiments, except for the C. rodentium survival experiment, in which 8- to 12-week-old female mice were used. A chloramphenicol-resistant variant of the WT C. rodentium strain DBS100 was cultured as previously described (63). Nr4a3^+/+ and Nr4a3^–/– mice were orally infected with 1 × 10⁹ CFU C. rodentium. The appearance and survival of infected mice were monitored daily for 21 days. DCs, Th1 cells, and Th17 cells were analyzed 8 days after infection. Bacterial burden was measured by plating feces and splenic and liver tissue.

Energy metabolism of BMDC. Energy metabolism of BMDCs was assessed with an XF^e96 Flux Analyzer (Seahorse Bioscience). CD11c⁺ cells were selected from day-8 BMDCs using CD11C microbeads (Miltenyi Biotec). Cells were plated on Seahorse cell culture plates in their growth medium at a density of 10⁵ cells per well and then stimulated with LPS for 16 hours prior to the experiment. Immediately before measurement, the medium was removed, and the adherent cells were gently washed with assay buffer (unbuffered DMEM, prepared according to Seahorse Bioscience protocols, supplemented with 10 mM glucose, 10 mM sodium pyruvate, 1× GlutaMax, and 1 mM HEPES, pH 7.4). The wells were filled with 450 μl/well assay buffer, and the measurements were initiated. Oxygen consumption rates (OCRs) and ECARs were normalized by Hoechst 33342 fluorescence from nuclei staining.

Statistics. Data were analyzed with GraphPad Prism 6.0 (GraphPad Software). Statistical significance was calculated by an unpaired, 2-tailed Student’s t test or 2-way ANOVA. A P value of less than 0.05 was considered statistically significant.

Study approval. All experiments followed the guidelines of the IACUC of the La Jolla Institute for Allergy and Immunology. Approval of the use of rodents was obtained from the La Jolla Institute for Allergy and Immunology Animal Care and Use Committee, according to NIH guidelines for the care and use of laboratory animals.