Dendritic Cells of Mesenteric and Regional Lymph Nodes Contribute to Yersinia enterocolitica O:3–Induced Reactive Arthritis in TNFRp55^−/− Mice

Reactive arthritis (ReA) is a form of peripheral spondyloarthritis (SpA) that develops after a gastrointestinal or urogenital infection (1). The incidence range of ReA has been reported between 1 and 30 cases/100,000 per year, the frequency of which depends on the geographic area. ReA manifests ∼1–3 wk after the mucosal infection (2). Once ReA is detected, bacterial cultures of fluids are often negative, and usually, the Ag cannot be found in the joints (2). Although the coexistence of gut and joint inflammation is well established in SpA (3), the immunopathogenic mechanisms of this connection in ReA are currently unknown.

Yersinia enterocolitica are Gram-negative bacteria that cause food-born gastrointestinal diseases, including diarrhea, mesenteric lymphadenitis, and terminal ileitis. Moreover, Y. enterocolitica is a well-established trigger of ReA (4). Following the ingestion of contaminated food or water, Y. enterocolitica first colonize the lumen and transmigrate through Ag-sampling M cells across follicle-associated epithelium of the small intestine. This results in the colonization of Peyer's patches (PP) (5). Subsequently, Y. enterocolitica can spread via the lymph and/or blood into the mesenteric lymph nodes (MLN), where the bacteria cause inflammatory changes and from where Y. enterocolitica can invade internal organs, such as the liver and the spleen (5). Alternatively, the bacteria may use a PP-independent dissemination route and spread directly from the intestine to systemic tissues (6). Y. enterocolitica serotype O:3 is less virulent in mice than other serotypes, such as the serotype O:8. However, it is frequently isolated from patients and also associated with ReA (7).

TNF is a pleiotropic cytokine and a master regulator of the immune system, which plays important roles in homeostasis and inflammation (8). TNF mediates its function through two receptors: TNFRp55 (TNFRp55, TNFR1, or CD120a) and TNFRp75 (TNFR2 or CD120b). TNFRp55 is constitutively and ubiquitously expressed on a broad variety of cells, whereas expression of TNFR2 is inducible and tightly regulated (8). In previous studies, we have demonstrated that TNFRp55 is essential for the protection against both Y. enterocolitica infection and Y. enterocolitica–induced ReA through mechanisms that involve IL-12/23p40 production (9, 10).

Dendritic cells (DCs) are APCs that play powerful roles orchestrating innate and adaptive immune responses. They also maintain tissue tolerance or promote immune responses to pathogens (11). DCs are phagocytic cells that continuously sample Ags from their microenvironment. After the uptake of the Ag, DCs migrate to the draining lymph nodes, where they present this Ag to naive T lymphocytes. This migration is guided by the interaction of CCR7 on migrating DCs with its corresponding ligand CCL21, which is found on lymphatic endothelial cells (11).

A link between the intestine and joint inflammation is clear in ReA because this disease develops after a gastrointestinal infection (12). In previous research, bacterial products from a preceding gut infection were recovered from the joints of patients who developed ReA (12, 13). In addition, signaling through TNFRp55 has been previously shown to be essential for the development of both bowel inflammation and arthritis (14). However, the cellular targets of TNF remain poorly defined. DCs are also found throughout the intestine, including MLN, in which CD103⁺ DCs, cells absent in the spleen, represent a population that migrates from the intestine lamina propria (15). Recently, we have reported that DCs are the main cellular source of high IL-12/23p40 levels in the pathophysiological scenario of Y. enterocolitica–induced ReA in TNFRp55-deficient (TNFRp55^−/−) mice (16). However, to the best of our knowledge, the role of DCs in the intestine–joint connection in ReA has not been previously explored. In this study, we investigated DCs of MLN and joint-draining lymph nodes (regional lymph nodes [RLN]) during the development of Y. enterocolitica–induced ReA in TNFRp55^−/− mice. Moreover, we modulated DC trafficking to determine whether ReA can be suppressed by hindering the migration of DCs. Our data provide insights into the role of DCs in the joint–gut relationship in ReA. A deeper understanding of this connection could open new ways for controlling the development of ReA after an intestinal infection.

C57BL/6 TNFRp55^−/− mice were kindly provided by the Max von Pettenkofer-Institute (Munich, Germany). C57BL/6 wild-type (WT) mice were purchased from the Animal Facilities of the National University of La Plata (La Plata, Argentina). Breeding colonies were established at the Animal Facilities of the National University of San Luis. Mice were kept under specific pathogen-free conditions in positive-pressure cabinets (EHRET, Emmendingen, Germany) and provided with sterile food and water ad libitum. Six- to eight-week-old male mice were used for the experiments. Experimental protocols were approved by the Animal Care and Use Committee of the National University of San Luis (San Luis, Argentina) (protocols number B-163/13).

Strain MHC 700 Y. enterocolitica O:3 (kindly provided by Dr. G. Kapperud, Department of Bacteriology, National Institute of Public Health, Oslo, Norway) was used for infection. Bacteria were cultured as described by Di Genaro et al. (9). For tracking DC migration, GFP-expressing Y. enterocolitica O:3 (Y. enterocolitica O:3–GFP) was constructed by transforming this strain with the plasmid pACYC184 harboring the gfp gene (17). Transformation was carried out by electroporation as previously described (18). Mice were starved for 3 h and then infected orogastrically with 1–5 × 10⁸ yersiniae in 200 μl of saline using a gastric tube. The control group received saline instead. The number of inoculated bacteria was controlled by plating serial dilutions of the inoculated suspension on Mueller–Hinton agar and counting of CFU following incubation at 27°C for 48 h.

Bone marrow–derived DCs (BMDCs) were obtained according to Lutz et al. (19) protocols. Briefly, 2 × 10⁶ bone marrow cells from femurs and tibias of C57BL6 WT or C57BL6 TNFRp55^−/− mice were cultured in 100-mm dishes with supplemented RPMI 1640 medium: 10% FBS, 2 mM l-glutamine, 1 mM pyruvate, 100 IU/ml penicillin, and 100 μg/ml streptomycin. This contained 200 U of GM-CSF (Recombinant Murine GM-CSF; PeproTech, Rocky Hill, NJ). Three days after seeding, 10 ml of fresh medium with GM-CSF was added. Then, every 2 d, half of the medium is discarded and replaced by fresh medium. At days 8 and 9, the cells were analyzed by flow cytometry and more than 90% expressed CD11c marker. In addition, in vivo expansion and isolation of DCs from MLN (MLN DCs) were performed as previously described (16). For infection, 2 × 10⁵ BMDCs or MLN DCs were seeded in 24-well plates in supplemented RPMI 1640 medium without antibiotics and infected with Y. enterocolitica O:3 in a multiplicity of infection (moi) of 10. After 1 h, 0.1 μg/ml of gentamicin was added to kill extracellular bacteria. Then, the BMDCs were cultured by 24 h for additional experiments. To recapitulate MLN microenvironment (MLN me), BMDCs were infected as described above in the presence of the supernatants of MLN homogenates obtained from WT or TNFRp55^−/− mice on day 14 (MLN me on day 14 [MLN me-day 14]) in a dilution of 1/3 in supplemented RPMI 1640 medium.

On days 7, 14, and 21 postinfection (pi), MLN and RLN from C57BL6 WT mice or C57BL6 TNFRp55^−/− were collected. The organs were finely cut and digested for 20 min at 37°C in HBSS medium containing collagenase IV (0.5 mg/ml; Sigma-Aldrich, St. Louis, MO) and DNase I (15 μg/ml; Roche Molecular Biochemicals, Mannheim, Germany). Cells were incubated for 15 min at 4°C with anti-mouse CD16/32 (Fc blocking; BD Biosciences, San Diego, CA) and stained with FITC anti-mouse IA/IE (2G9; BD Biosciences), APC anti-mouse CD11c (HL3; BD Biosciences), PE anti-mouse CD11b (M1/70; BD Biosciences), and PerCP Cy5.5 anti-mouse CD103 (2E7; BioLegend) Abs. Data were obtained on a FACSCalibur flow cytometer (BD Biosciences) and analyzed with FlowJo software (Tree Star, Ashland, OR).

For intracellular staining, MLN or RLN cells from naive mice were obtained. These were stimulated with 1 μg/ml of LPS (Escherichia coli 0111: B4; Sigma, St. Louis, MO) for 12 h in a humidified atmosphere containing 5% CO₂ at 37°C in supplemented RPMI 1640 medium as described above in the presence of GolgiPlug (BD Biosciences), according to the manufacturer’s instructions. Then, the cells were stained with FITC anti-mouse IA/IE and APC anti-mouse CD11c Abs, fixed and permeabilized with BD Fixation and Permeabilization Solution (BD Cytofix/Cytoperm), and after that, they were stained with PE anti-mouse IL-12p40/p70 mAb (clone 15.6; BD Biosciences). Data were obtained with a FACSCalibur cytometer and analyzed using FlowJo software.

To study CCR7 expression, 2 × 10⁵ BMDCs or MLN DCs were infected as described above. At 24 h pi, the cells were collected and stained with FITC IA/IE, APC CD11c, and PerCpCy5.5 CCR7 (clone 4B12; BioLegend) anti-mouse Abs and analyzed by flow cytometry.

Fingolimod (FTY720) (Cayman Chemicals, Ann Arbor, MI), was prepared as described by St John et al. (20). FTY720 was dissolved in DMSO in a concentration of 25 mg/ml and stored at −80°C until use. This solution was diluted in sterile PBS containing 2% of hydroxypropyl-β-cyclodextrin (Sigma-Aldrich) for i.p. administration. C57BL6 WT or C57BL6 TNFRp55^−/− were infected as described above, and 14 d pi, FTY720 solution was administered via i.p. route in a dosage of 5 mg/kg every 3 d until day 21 pi (Fig. 5A). The control group received PBS.

Mice were examined daily for the visual appearance of arthritis in peripheral joints, and the arthritis score was calculated until day 21 pi, as described by Banda et al. (21). Clinical severity was classified for each paw, as follows: 0, normal joint; 1, slight inflammation and redness; 2, severe erythematic and swelling affecting the entire paw; and 3, deformed joint paw with ankylosis, joint rigidity, and loss of function. The total score was based on the assessment of all paws, with a maximum score of 12 per animal, as described (9, 10).

At day 21, knee joints were dissected. After routine fixation, decalcification, and paraffin embedding, 5-μm-thick sections were cut, stained with H&E, and examined with a light microscope. Photographs were taken using an Olympus BX40 light microscope equipped with Sony SSC-DC5OA camera.

C57BL6 WT or C57BL6 TNFRp55^−/− were infected and treated with FTY720 as described above. After 21 d pi, joint homogenates were prepared as previously described (10), and IL-12p40 (Mouse IL-12p40 ELISA Set; BD Biosciences) and IFN-γ (BioLegend ELISA Kit; BioLegend, San Diego, CA) were measured according to the manufacturer’s instructions. The cytokine concentrations were normalized to the total protein concentrations of the homogenates, which were measured by Qubit Protein Assay Kit (Thermo Fisher Scientific, Waltham, MA).

Differences between the groups were tested for significance using the Kruskal–Wallis and Mann–Whitney U test, as appropriate. A p value ≤ 0.05 was considered statistically significant. The results are expressed as the mean ± SEM.

DCs are central players of the immune response that uptake Ags and migrate to draining lymph nodes to prime naive T cells (11). First, we analyzed the recruitment of immune cells in RLN of TNFRp55^−/− and WT mice after oral Y. enterocolitica O:3 infection. We observed that RLN of TNFRp55^−/− mice were markedly swollen compared with RLN of WT mice (Fig. 1A). This observation correlated with an increase of the total cell number of RLN of Y. enterocolitica O:3–infected TNFRp55^−/− mice (Fig. 1B). These cells were characterized using CD11b and CD11c markers (22). It was detected that all CD11b⁺ populations—CD11c⁺, CD11c^int, and CD11c⁻—augmented in RLN under TNFR deficiency (Fig. 1C).

Conventional or classical DCs share certain surface markers with other mononuclear phagocytes, namely tissue-resident macrophages, monocytes, and various monocyte-derived cells; however, DCs are characterized as CD11c⁺ cells with a high expression of MHC class II molecules (IA/IE^high) (23). Moreover, migration of DCs into initial lymphatic vessels is induced by DC maturation, leading to upregulation of MHC molecules (23). In addition, largely sessile lymph node–resident DCs are present in the T cell area of the draining lymph node (24). Hence, we differentiated migratory (CD11c⁺IA/IE^high) and resident (CD11c⁺IA/IE^int) in the RLN (Fig. 1D, 1E). Recent discoveries have simplified the classification of DC subsets (25). We selected CD11b marker to analyze the two major subpopulations (CD11b⁻ and CD11b⁺ DCs) in migratory and resident DC compartments. In Y. enterocolitica O:3–induced ReA model in TNFRp55^−/− mice, clinical signs of arthritis began to develop on day 14 (arthritis onset) (9, 10). Interestingly, on day 14, the absolute numbers of both migratory and resident DCs were significantly increased in RLN of TNFRp55^−/− mice compared with the numbers found in RLN of WT mice (Fig. 1E, 1F). Early, on day 7, resident DCs also showed significant augment under TNFRp55 deficiency, and this increase was sustained until day 21 after Y. enterocolitica infection. The increase of migratory DCs was delayed in WT mice. In contrast, no modification in the number of resident DCs was detected in RLN of control mice (Fig. 1F). Higher cell numbers of both CD11b⁺ and CD11b⁻ subsets of migratory DCs were found in RLN of TNFRp55^−/− mice on day 14 (Fig. 1G). In contrast, the CD11b⁻ and CD11b⁺ subpopulations of resident DCs were increased in RLN under TNFRp55 deficiency from day 7 until day 21 (Fig. 1H).

These data demonstrate that TNFRp55 deficiency impacts the infiltration of DCs in joint-draining lymph nodes. The augment of resident DCs of TNFRp55^−/− mice may indicate early and sustained homing of the precursors of DCs (pre-DCs), via high endothelial venules (HEV), into RLN. Furthermore, marked DC migration through the lymphatics takes place on arthritis onset and is maintained under TNFRp55 deficiency.

A joint–gut axis has been suggested in SpA (12, 26). For T cell priming, after oral Ag administration, DCs transport the Ag from the intestine to MLN. In addition, in Y. pestis–lymph node remodeling, infected DCs have been involved in the trafficking of these bacteria through the lymphatic system (20).To know whether joint-draining lymph nodes mirror the changes of DCs of intestinal-draining lymph nodes, we analyzed DCs in MLN. As we detected in RLN, inflamed MLN, as well as more total cells per MLN, were observed in TNFRp55^−/− mice (Fig. 2A, 2B).

We next investigated the number of migratory and resident DCs in MLN of TNFRp55^−/−and WT mice at different time points after Y. enterocolitica O:3 infection. As shown in Fig. 2C and 2D, a peak of both migratory and resident DCs was detected in MLN of TNFRp55^−/−mice on the day of arthritis onset (day 14). Interestingly, migratory DCs significantly dropped in TNFRp55^−/−mice on day 21, reaching levels lower than MLN of WT mice (Fig. 2C). Similarly, resident DCs decreased markedly in their MLN for TNFRp55^−/− mice on day 21 (Fig. 2D).

When the CD11b⁺ and CD11b⁻ subpopulations were characterized, again both subsets of migratory and resident DCs reached a peak on day 14 in MLN of TNFRp55^−/− mice (Fig. 2E, 2F). Moreover, CD11b⁻ migratory DCs decreased on day 21 in TNFRp55^−/− MLN (Fig. 2E). Under gut inflammation, it has been reported that CD11b⁺ DCs have the ability to induce Th17 responses, which have antimicrobial activities in mucosa (11). In the intestine, CD11b⁻ DCs have been described in general terms as exhibiting a special propensity to induce regulatory T cells (11). It is interesting to notice that on day 21, CD11b⁻ DCs were markedly reduced in MLN of TNFRp55^−/−mice (Fig. 2E) in both migratory and resident DC compartments.

Overall, these results demonstrate changes in the number of DCs in MLN of TNFRp55^−/− mice. Considering that on day 14, the changes of migratory DCs in MLN mirror those observed in RLN, we suggest a DC trafficking from MLN to RLN. Migratory DCs peaked on arthritis onset in MLN under TNFRp55 deficiency. On day 21, these cells decreased only in TNFRp55^−/−mice. Therefore, it could be suggested that TNFRp55 signaling controls the migration of DCs to MLN, orchestrates a regulatory response in mucosa, and consequently contributes to avoid Y. enterocolitica–induced ReA development.

In previous studies, we have demonstrated increased levels of IL-12/23p40 in Y. enterocolitica–induced ReA in TNFRp55^−/− mice (10). Recently, we have confirmed that DCs participate in this IL-12/23p40 overproduction (16). Therefore, we evaluated whether migratory or resident DCs of MLN of TNFRp55^−/− mice are the main source of IL-12/23p40. To explore intracellular IL-12/23p40 production in the early hours, cells from MLN or RLN of naive TNFRp55^−/− and WT mice were LPS stimulated for 12 h, and then the frequencies of IL-12/23p40⁺ migratory or resident DCs were evaluated. We found a higher frequency of IL-12/23p40⁺ migratory DCs in TNFRp55^−/− MLN when compared with WT MLN (Fig. 3A). In contrast, neither migratory nor resident DCs from naive TNFRp55^−/− RLN showed differences when compared with DCs from WT RLN (Fig. 3B).

The chemokine receptor CCR7 is essentially involved in the migration of DCs toward or into lymphatics (27). Upregulation of CCR7 expression on DCs is induced by early recognition of the pathogen by these cells (11). Hence, BMDCs were obtained and infected with Y. enterocolitica O:3 at moi 10:1 for 1 h. After killing extracellular bacteria, BMDCs were cultured for 24 h, and the CCR7 expression was analyzed by flow cytometry. Uninfected TNFRp55^−/− BMDCs expressed higher levels of CCR7 than uninfected WT BMDCs (Fig. 3C). Y. enterocolitica O:3 infection induced an increase of CCR7 expression in BMDCs, which was significantly higher in TNFRp55^−/− BMDCs (Fig. 3C). To recapitulate the MLN me-day 14 in BMDCs, these cells were infected in the presence of supernatants of homogenates of MLN obtained from WT or TNFRp55^−/− on day 14. As expected, significant higher CCR7 expression was detected on Y. enterocolitica–infected BMDCs from TNFRp55^−/− mice in knockout MLN me-day 14 compared with the same cells in WT MLN me-day 14 (Fig. 3C). Interestingly, Y. enterocolitica–infected BMDCs from WT mice in knockout MLN me-day 14 increased CCR7 expression, reaching the same level of Y. enterocolitica–infected TNFRp55^−/− BMDCs (Fig. 3C).

These data suggest that migratory DCs from MLN are an essential source of IL-12/23p40 in TNFRp55^−/− mice. RLN could depend on DC migration from MLN. Furthermore, after Y. enterocolitica O:3 infection, TNFRp55 deficiency may promote DC migration into lymph nodes via lymphatics by inducing the CCR7 expression on DC.

Intestinal lamina propria DCs have the capacity to migrate in afferent lymphs to MLN (11). Although early work assumed that intestinal DCs comprised a homogeneous population of CD103⁺ cells, most recent studies have divided them on the basis of CD103 and CD11b expressions (28, 29). To track intestinal DC migration to RLN, gastrointestinal infection with Y. enterocolitica O:3–GFP was performed, and GFP⁺ CD103⁺ DCs were examined in RLN early pi (day 5) because bacteria are not found in the joint and RLN once ReA is detected (Fig. 4A). Furthermore, to analyze DC migration between MLN and RLN on arthritis onset, we evaluated the major subsets of the migratory DC compartment of MLN in RLN on day 14: CD103⁺CD11b⁻, CD103⁻CD11b⁺, and CD103⁺CD11b⁺. Five days after oral Y. enterocolitica O:3–GFP infection, GFP⁺ intestinal DC subpopulations were found in RLN of both WT and TNFRp55^−/− mice (Fig. 4A). In contrast with WT mice, TNFRp55^−/− mice presented a marked increase of the three major intestinal DC subsets in their RLN on arthritis onset (Fig. 4B–D).

Therefore, early pi, intestinal DCs travels with Y. enterocolitica O:3 to RLN. TNFRp55 deficiency enhances the recruitment of intestinal DCs into RLN, which supports the connection between intestinal and joint-draining lymph node. This may sustain Y. enterocolitica O:3–induced ReA in TNFRp55^−/− mice.

FTY720 is an immune modulator derived from myriocin that inhibits the functions of sphingosine-1 phosphate, a potent bioactive sphingolipid metabolite that regulates many cellular processes, including cell migration (30, 31). In previous studies, it has been demonstrated that FTY720 suppresses the development of collagen-induced arthritis by interfering with DC migration (30). Therefore, we assessed the effect of blocking DC migration by FTY720 on Y. enterocolitica O:3–induced ReA in TNFRp55^−/− mice. Because 50–60% of these knockout mice died because of Y. enterocolitica O:3 infection (9, 10), we started the FTY720 treatment in the surviving mice since day 14 pi every 3 d until day 21 (Fig. 5A). All the studies were performed on day 21 pi. We first examined DC populations in RLN and observed a slight reduction of the absolute number of total migratory DCs in FTY720-treated TNFRp55^−/− mice (p = 0.06) (Fig. 5B). Although this reduction is not significant, both CD11b⁺ and CD11b⁻ subsets of migratory DCs exhibited a significantly lower absolute number in RLN of TNFRp55^−/− mice after FTY720 treatment (Fig. 5C). In addition, FTY720 treatment significantly decreased the absolute number of total resident DCs and their subpopulation CD11b⁺ and CD11b⁻ in RLN of TNFRp55^−/− mice (Fig. 5D, 5E). Furthermore, FTY720 treatment reduced the clinical score of arthritis in TNFRp55^−/− mice (Fig. 6A). In agreement with these results, histological examination revealed that FTY720 treatment diminished synovial hyperplasia, cartilage erosion, and bone loss when compared with nontreated TNFRp55^−/− mice (Fig. 6B).

It has been reported that CD11b⁻ DCs are the primary subset that produces more IL-12, which promotes Th1 response under proinflammatory conditions (25). We detected a lower number of migratory CD11b⁻ DCs in RLN of FTY720-treated TNFRp55^−/−mice (Fig. 5C). Moreover, IL-12/23p40 and IFN-γ are key cytokines in Y. enterocolitica O:3–induced ReA in TNFRp55^−/− mice (10). Hence, we investigated the levels of these two proinflammatory cytokines in joint homogenates. High concentration of IL-12/23p40 and IFN-γ were detected in Y. enterocolitica O:3–infected TNFRp55^−/−mice when compared with Y. enterocolitica O:3–infected WT mice. On the contrary, both cytokines (Fig. 7A, 7B) significantly reduced to the levels found in the control mice in the joints of FTY720-treated TNFRp55^−/− mice.

CCR7 is a crucial chemokine receptor for DC migration. Therefore, we explored whether CCR7 is associated with the mechanisms underlying the FTY720 inhibitory effect on DC migration. We performed in vitro experiments to analyze the effect of FTY720 on the expression of CCR7 onto the DC membrane. As shown in Fig. 7C and 7D, FTY720 treatment downregulated CCR7 expression on Y. enterocolitica O:3–infected BMDCs or purified MLN DCs.

Our results demonstrate that CCR7-mediated DC migration into RLN plays an important role in the pathogenesis of Y. enterocolitica O:3–induced ReA under TNFRp55 deficiency.

The relationship between gut and joint is clear in ReA because this type of SpA develops from an antecedent gastrointestinal infection (12). However, how this immunopathogenic intestine–joint connection operates in this disease, to the best of our knowledge, has not been thoroughly described. DCs are powerful APCs whose appropriate localization is essential in the initiation or regulation of the immune response (27). Therefore, DC migration has been demonstrated to be fundamental in the maintenance of the tissue homeostasis and also of different pathogenic conditions (11, 27). Moreover, infected DCs are involved in the trafficking of Y. pestis among lymph nodes through the lymphatic system (20). In a previous study, we showed a pathogenic role of DCs as the main cellular source of IL-12/23p40 in Y. enterocolitica O:3–induced ReA in TNFRp55^−/−mice (16). In the present work, we examine DCs in RLN and MLN to investigate the gut–joint association in this murine model of ReA.

We initially observed inflammation and an increased absolute number of total cells in RLN of TNFRp55^−/− mice on day 14 after Y. enterocolitica infection. This may reflect an influx of proinflammatory cells to RLN on the day that we previously reported as the day of arthritis onset (10). The ability to migrate to draining lymph nodes is a key characteristic of DCs. After sampling the foreign Ag, an intrinsic program induces DC maturation, leading to upregulation of MHC class II (11). We found migratory DCs (CD11c⁺IA/IE^high) in RLN of the mice, indicating that DCs travel from the intestine to these lymph nodes. We detected a migration of DCs in TNFRp55^−/− RLN (day 14), which was earlier than in WT RLN (day 21). In fact, it has been reported that TNF through TNFRp55 and TNFRp75 regulates DC maturation and survival (32). In addition, DCs have been shown to be important host cells for the dissemination of various pathogens (33). Although no bacteria are present in the joints of TNFRp55^−/− mice with ReA (9), we detected that DCs traffic Y. enterocolitica O:3–GFP from the intestine to RLN early pi. Therefore, we suppose that DCs traffic bacterial Ags from the intestine to RLN, which is significant under TNFRp55 deficiency. However, whether Y. enterocolitica Ags are internalized in migratory DCs of RLN remains to be addressed. A second pathway of Ag delivery is mediated by resident DCs that present soluble Ags and enter via the afferent lymphatics in the T cell area of the draining lymph node. This happens before there is any detectable DC migration from the periphery (34). Accordingly, resident DCs increased in TNFRp55^−/− RLN from day 7–21 after Y. enterocolitica infection, but no changes in this DC population was detected in WT RLN. Because TNFRp55^−/− mice are susceptible to Y. enterocolitica infection (9), it is probable that from blood, pre-DCs gain access to lymph nodes via HEVs early pi, and they increase the resident DC network. In addition, TNFRp55^−/− mice exhibit a selective defect in the development of PP (35). As a result, more intestinal soluble Ags may reach RLN and then promote resident DC differentiation. Detailed analysis of the CD11b⁺ and CD11b⁻ subsets of both migratory and resident DC compartments revealed that both subtypes increased in TNFRp55^−/− RLN. Therefore, under TNFRp55 deficiency, both resident DC subsets may interact with the T cells retained in the RLN and cooperate to their expansion and differentiation.

It has been recently demonstrated that the transport of Y. enterocolitica to MLN is dependent on CCR7-expresing CD103 DCs (6). Similarly, other reports showed that after Salmonella oral infection, migratory DCs are the only immune cells that can carry the viable bacteria from the site of infection through migration in lymph to the MLN (36). Furthermore, acute mesenteric lymphadenitis is one of the most characteristic forms of Y. enterocolitica infection in humans (37), and clinical cases caused by these bacteria have identified the MLN as a primary site of infection (38). Indeed, we also found inflammation and cellular influx in MLN. A peak of both migratory and resident DCs was observed in TNFRp55^−/− MLN on arthritis onset (day 14), mirroring the changes of cells in RLN. In contrast with MLN of WT mice, MLN of TNFRp55^−/− mice showed a transient increase of migratory and resident DCs on day 14, but a considerable increase of migratory CD11b⁺ DCs was detected in MLN of WT mice on day 21. Following activation, this subset has been described to induce Th17 cells, which promotes protective responses against mucosal infection (11). This evidence is in agreement with our previous report, which showed a marked increase of IL-17 in MLN of WT mice on 21 d after Y. enterocolitica infection (10). Interestingly, CD11b⁻ DCs reduced in MLN of TNFRp55^−/−mice in both migratory and resident DC compartments (Fig. 2E, 2F). This DC subset has the ability to induce regulatory T cells (11, 29). Coincidentally, in our prior studies, both TGF-β and IL-10, which are critical immunosuppressive cytokines produced by regulatory T cells, were detected augmented in MLNs of WT mice only 21 d pi (10, 39). Therefore, DCs of MLN may promote regulatory mechanisms presumably mediated by TNFRp55, which is crucial for controlling ReA development.

We have previously reported data that strongly support a role of the TNFRp55 pathway in controlling IL-12/23p40 (16). A cross-talk between TNFRp55 and TLR4 signaling has been identified, and the candidate protein A20 has been associated as a negative regulator of the core cross-talk element TNFR-associated factor 6 (40). Thus, in a mouse model in which CD11c⁺ cells, including DCs, lack expression of A20, the A20^−/− DCs produce a higher concentration of IL-12 when they are stimulated with bacterial products (41). Interestingly, these mice spontaneously developed lymphocyte-dependent colitis, seronegative ankylosing arthritis, and enthesitis (41). Hence, DCs may drive Y. enterocolitica–induced ReA in TNFRp55^−/− mice through their overproduction of IL-12/23p40. When we stimulated naive DCs from MLN with LPS in vitro, we observed that migratory DCs were the source of IL-12/23p40 and that the frequency of IL-12/23p40⁺ migratory DCs was higher in cells of TNFRp55^−/− MLN. In contrast, naive DCs from TNFRp55^−/− RLN did not secrete IL-12/23p40 after in vitro LPS stimulation, supporting the requirement of MLN–RLN connection. The MLN and RLN link is also evident because the major subsets of the migratory DC compartment of MLN (CD103⁺CD11b⁻, CD103⁻CD11b⁺, and CD103⁺CD11b⁺) (28) were found in TNFRp55^−/− RLN on day 14. Moreover, when migration was tracked in the Y. enterocolitica–GFP infection, GFP⁺CD103⁺ DCs were found in TNFRp55^−/− RLN on day 5 pi. Therefore, it is tempting to speculate that DCs travel via lymph from MLN to RLN. Thus, DCs have been isolated from the efferent lymphs (42). Furthermore, Y. pestis disseminate intracellularly in DCs beyond draining lymph nodes to downstream lymph nodes (20). These cells may exit via efferent lymphs and travel through the lymphatic vasculature to drain to higher-order lymph nodes (20). Furthermore, we could show that both TNFRp55^−/− BMDCs in MLN environment (Fig. 3C) and DCs isolated from TNFRp55^−/− MLN (Fig. 7D) upregulated CCR7 under in vitro Y. enterocolitica O:3 infection. This chemokine receptor, expressed on migrating DCs, interacts with its ligand CCL21, which is found on lymphatic endothelial cells. This CCR7–CCL21 axis guides DC entry to afferent lymphatic vessels and through stromal networks into the T cell–rich paracortex in lymph nodes (27, 43). Therefore, our results support the hypothesis that TNFRp55 deficiency promotes enhanced CCR7-dependent DC migration into lymph nodes. Based on our data, we favor the notion that, at least in our model of TNFRp55^−/− mice, DCs participate in ReA pathogenicity, connecting intestine and joint-draining lymph nodes and secreting IL-12/23p40.

Finally, we abrogate DC migration by FTY720 treatment evidenced by the reduction of migratory DC subsets in TNFRp55^−/− RLN. This was associated with the decrease of IL-12/23p40 and IFN-γ in the joints of FTY720-treated TNFRp55^−/− mice. In the same line, CD11b⁻ DC subpopulation has been demonstrated to produce IL-12 and therefore promotes Th1 response under proinflammatory conditions (25). Similarly, FTY720 treatment decreased the production of IL-12 and IL-23 in LPS-stimulated DCs via the common subunit p40 as well as in the cross-talk with activated keratinocytes (44). In addition, FTY720 treatment reduced the severity of the ReA in TNFRp55^−/− mice. Additionally, we found that FTY720 treatment reduced the total resident DCs and their subpopulation CD11b⁺ and CD11b⁻ in RLN of TNFRp55^−/− mice. The factors that participate in the entry of the pre-DCs via HEVs from blood into the network of resident DCs in lymph nodes are not yet resolved. However, CCR7 internalization and/or desensitization might be involved (45). Therefore, we cannot exclude the contribution of resident DCs for Y. enterocolitica–induced ReA development under TNFRp55 deficiency. Consistent with our results, FTY720 has modulated DC migration to local lymph nodes in several mouse disease models, alleviating the clinical symptoms of the diseases (30, 46–48). In addition, we have shown that the mechanism underlying the effect of FTY720 on DC trafficking involves CCR7 expression because FTY720 treatment of Y. enterocolitica–infected TNFRp55^−/− BMDCs and purified MLN DCs decreased the expression of this chemokine receptor. This mechanism fully concurs with the published findings obtained in different settings (30, 47–49).

In conclusion, we provide evidence that DCs mediate the link between intestine and joints in ReA induced by oral Y. enterocolitica O:3 infection under TNFRp55 deficiency. Furthermore, our data indicate that FTY720 treatment blocks DC trafficking to local lymph nodes, which may represent a potential strategy to treat ReA induced by intestinal infection.

We acknowledge Gabinete de Asesoramiento en Escritura Científica en Inglés of Universidad Nacional de San Luis for improvement of the language of the manuscript. Moreover, the authors thank Marisol Velazquez for the preparation of Y. enterocolitica O:3–GFP.

The authors have no financial conflicts of interest.