Decreased Th17 and antigen-specific humoral responses in CX3CR1-deficient mice in the collagen-induced arthritis model




CX3CR1 is a chemokine receptor that uniquely binds to its ligand fractalkine (CX3CL1) and has been shown to be important in inflammatory arthritis responses, largely due to its effects on cellular migration. This study was undertaken to test the hypothesis that genetic deficiency of CX3CR1 is protective in the chronic inflammatory arthritis model collagen-induced arthritis (CIA). Because CX3CR1 is expressed on T cells and antigen-presenting cells, we also examined adaptive immune functions in this model.


Autoantibody formation, clinical, histologic, T cell proliferative, and cytokine responses were evaluated in wild-type and CX3CR1-deficient DBA/1J mice after immunization with heterologous type II collagen (CII).


CX3CR1−/− mice had an ∼30% reduction in arthritis severity compared to wild-type mice, as determined by 2 independent measures, paw swelling (P < 0.01) and clinical disease score (P < 0.0001). Additionally, compared to wild-type mice, CX3CR1−/− mice had an ∼50% decrease in anti-CII autoantibody formation (P < 0.05), decreased Th17 intraarticular cytokine expression (P < 0.01 for interleukin-17 [IL-17] and P < 0.001 for IL-23), and decreased total numbers of Th17 cells in inflamed joints (P < 0.05).


Our findings indicate that CX3CR1 deficiency is protective in inflammatory arthritis and may have effects that extend beyond migration that involve adaptive immune responses in autoimmune disease.

Many chemokine–receptor interactions have been implicated in inflammatory cellular trafficking in rheumatoid arthritis (RA) (for review, see ref.1). However, the promiscuity of ligand–receptor interactions seen within most chemokine receptor families has been difficult to overcome therapeutically in clinical trials targeting the blockade of an individual chemokine or its receptor in arthritis patients (2, 3). The solitary member of the CX3CR family, CX3CR1, is unique in that it has only one known ligand, fractalkine (FKN; CX3CL1) (4), and blockade of the CX3CL1/CX3CR1 signaling axis has been shown to be efficacious in several models of preclinical inflammation (for review, see ref.5).

With particular relevance to RA, CX3CL1 and CX3CR1 are up-regulated in inflammatory cells within the synovial tissue in the rat model of adjuvant-induced arthritis (AIA) (6), and CX3CL1 mediates T cell–dependent proliferation of synovial fibroblasts from RA patients (7). In the collagen-induced arthritis (CIA) model of chronic disease, mice treated with a neutralizing antibody to CX3CL1 have lower clinical scores, improved histologic features, and decreased migration of adoptively transferred splenic monocytes to the joint (8). Additionally, patients with RA have increased levels of CX3CR1+ T cells circulating in the peripheral blood (6), and increasing levels of CX3CR1+ T cells and monocytes in the synovial fluid that correlate with disease activity (6). These data suggest that CX3CL1/CX3CR1 signaling plays an important role in the trafficking and function of inflammatory cell subsets in RA.

CX3CR1 signaling is also important in the pathogenesis of inflammatory vascular disease and atherosclerosis (9–12), which is a complication of RA of longstanding duration (13). Our group has shown that CX3CR1 deficiency is protective against intimal hyperplasia after arterial injury in mice as a result of decreased monocyte trafficking (9) and decreased dendritic cell accumulation (11) in atherosclerotic plaques. In humans, a naturally occurring gene polymorphism (CX3CR1-M280) correlates with a lower prevalence of atherosclerosis (10, 12), which could potentially be explained by reduced CX3CL1-dependent cellular adhesion in inflammatory cells expressing CX3CR1-M280 (10). These data suggest that blockade of CX3CR1 interactions may be an important therapeutic target for the treatment of RA and the inflammatory sequelae that arise from it, such as atherosclerosis.

Because CX3CR1 is predominantly expressed on T cells and antigen-presenting cells (APCs) (11, 14, 15), we hypothesized that adaptive immune responses may be affected beyond the migration abnormalities seen with blockade of the ligand CX3CL1 (8) in an immunization model of inflammatory arthritis (CIA). Consequently, we investigated clinical disease outcomes, autoantibody formation, T cell responses, histopathologic features, and cytokine responses in mice with CIA with a gene deletion of CX3CR1 (CX3CR1−/− mice) as compared to wild-type controls. Our results suggest that inhibition of CX3CR1 may have beneficial effects in inflammatory arthritis beyond that of migration, since decreased autoantibody levels and proinflammatory Th17 responses were observed in CX3CR1-deficient animals.



All animals were bred, housed, and cared for under specific pathogen–free conditions in Division of Laboratory Animal Medicine facilities under the approved Institutional Animal Care and Use Committee protocol number 09-245.0.


Anti-CD3 and anti-CD28, which were used for T cell proliferation studies, and eFluor 450–conjugated anti-CD4 and allophycocyanin-conjugated anti–interferon-γ (IFNγ), which were used in flow cytometry, were purchased from eBioscience. Anti-CX3CR1 antibodies (R&D Systems) and phycoerythrin-conjugated anti–interleukin-17A (anti–IL-17A) antibodies (BD PharMingen) were also used in flow cytometry.

Induction and evaluation of CIA.

CX3CR1−/− mice and wild-type controls were backcrossed ≥7 generations onto the DBA/1J background, which is highly susceptible to CIA (16). Eight-week-old male mice were immunized on day −21 with a 1:1 mixture of Freund's complete adjuvant (Sigma-Aldrich) and 100 μg per mouse of bovine type II collagen (CII; Chondrex) injected subcutaneously into the base of the tail. A second injection was given on day 0 with a 1:1 mixture of Freund's incomplete adjuvant and 100 μg per mouse of bovine CII subcutaneously into the base of the tail. Our protocol used a second, booster injection for 3 reasons: a single injection has more variability in disease onset; after 2 injections, most animals develop maximum disease by 6 weeks compared to only 40% of animals after a single injection; and a single injection results in less severe disease overall. This protocol has been well characterized by Wooley and others (16).

An observer (TKT) who was blinded with regard to mouse strain evaluated the mice for clinical disease severity and paw swelling from baseline. Disease severity was scored using a clinical disease index with a scale of 0–3, where 0 = normal paw, 1 = mild but definite swelling of either the ankle or digits, 2 = moderate redness and swelling of an ankle and/or any number of digits, and 3 = maximal redness and swelling of the entire paw and digits with or without ankylosis. The maximum score per paw was 3 with a maximum possible score of 12 per mouse. This scoring system has been validated by our group previously (17, 18). Paw swelling measurements were obtained by measuring the thickness of the fore and hind limbs at the wrist and ankle, respectively. Paw swelling is presented as the change in the mean thickness of the fore and hind limbs (in mm) from the baseline average.

Histopathologic analysis.

At experiment termination, hind limbs were fixed in 4% paraformaldehyde, decalcified in formic acid, and embedded in paraffin. Serial 5-μm sections were cut and stained with hematoxylin and eosin according to standard protocols for morphologic analysis.

Anticollagen antibody enzyme-linked immunosorbent assay (ELISA).

IgG anti-CII antibodies were measured by a standard sandwich ELISA (Chondrex). Autoantibodies in CIA can be assessed at many time points, but are typically analyzed between days −7 and 14, when their levels are most elevated (19). We chose to analyze autoantibodies before the second injection and before disease onset (day −7 and day 0) to examine the increase in the production of autoantibodies, which is thought to precede clinical illness and to be directly pathogenic (20). Serum was collected (on days −7 and 0) from the tail veins of mice immunized to develop CIA according to previously published protocols (21) and was diluted 1:5,000 for autoantibody measurement by ELISA, performed according to the recommendations of the manufacturer.

Assessment of lymph node T cell proliferation and intracytokine staining for IL-17 and IFNγ.

In CIA, autoantigen recognition and T cell proliferation and differentiation are initiated in the first 2 weeks after initial immunization (22). This is typically the optimal time to assess T cell proliferative responses and T cell cytokine production upon in vitro stimulation. We isolated the draining lymph nodes (iliac and inguinal) from the site of immunization (the base of the tail) where we expected to see the most potent immune response (on day −7 [7 days before the second injection]). For proliferation studies, cells were isolated in RPMI, 1% naive syngeneic mouse sera, 10 mM HEPES, and 25 units/ml heparin, and then further enriched for CD4+ T cells using a negative selection magnetic-activated cell sorting mouse CD4+ T cell isolation kit, according to the recommendations of the manufacturer (Miltenyi Biotec). Enriched T cells from wild-type or CX3CR1−/− mice were stimulated in 96-well flat-bottomed plates under 1 of the following 3 conditions: 1) with anti-CD3 (1 μg/ml) and anti-CD28 (1 μg/ml), 2) with irradiated APCs from the spleen of a naive syngeneic mouse and T cell proliferation–grade bovine CII (100 μg/ml; Chondrex), or 3) with irradiated APCs in media alone. Cultures were incubated at 37°C for 48 hours, then 1 μCi of 3H-thymidine was added to each well, and cultures were incubated an additional 15–18 hours prior to scintillation counting. For intracytokine staining of IL-17 or IFNγ, lymphocytes were stimulated with anti-CD3 (5 μg/ml coated plate) and anti-CD28 (1 μg/ml) in 24-well cell culture for 6 hours, and GolgiStop (1 μl/ml; BD Biosciences) was added to the culture for the final 2 hours. Cells were harvested and stained for surface CD4 and intracellular IL-17 and IFNγ, and then analyzed by flow cytometry.

Real-time quantitative polymerase chain reaction (PCR) analysis of intraarticular cytokine, matrix metalloproteinase (MMP), and chemokine/receptor levels.

Paws from arthritic mice (score of 2) were collected for RNA analysis on day 10, which corresponded to early onset of arthritis, and were compared to paws from naive, unimmunized mice (score of 0). All paw samples were homogenized using an Omni TH homogenizer. Total RNA was extracted with TRIzol reagent (Invitrogen Life Technologies), and complementary DNA (cDNA) synthesis was carried out with Superscript II reverse transcriptase according to the recommendations of the manufacturer (Invitrogen Life Technologies). Real-time quantitative PCR was performed using a Bio-Rad iCycler and SensiMix SYBR and Fluorescein kit (Quantace). The total volume of each reaction was 25 μl including 12.5 μl of 2× SYBR Green Supermix, 0.625 μl of each primer at a concentration of 20 μM, 10.25 μl of RNase-free water, and 1 μl of cDNA. The PCR was carried out using the following thermocycling conditions: 95°C for 3 minutes, 40 cycles at 60°C for 30 seconds, 95°C for 1 minute, and 60°C for 1 minute. IDUA ribosomal RNA was used as a control.

The following sets of primers were used: for IDUA, forward GCATCCAAGTGGGTGAAGTT and reverse CATTGAGCAGGTCCGGATAC; for FKN, forward CGCGTTCTTCCATTTGTGTA and reverse GTCTGTGCTGTGTCGTCTCC; for MMP-1, forward AACTACATTTAGGGGAGAGGTGT and reverse GCAGCGTCAAGTTTAACTGGAA; for MMP-2, forward CGGAGATCTGCAAACAGGACA and reverse CGCCAAATAAACCGGTCCTT; for MMP-9, forward GCGTGTCTGGAGATTCGACTT and reverse TATCCACGCGAATGACGCT; for MMP-13, forward CTTCTTCTTGTTGAGCTGGACTC and reverse CTGTGGAGGTCACTGTAGACT; for tumor necrosis factor α (TNFα), forward CATCTTCTCAAAATTCGAGTGACAA and reverse TGGGAGTAGACAAGGTACAACCC; for IL-1β, forward GGTCAAAGGTTTGGAAGCAG and reverse TGTGAAATGCCACCTTTTGA; for IL-6, forward CAAAGCCAGAGTCCTTCAGAG and reverse GGATGGTCTTGGTCCTTAGC; for IL-17, forward CTCCAGAAGGCCCTCAGACTAC and reverse AGCTTTCCCTCCGCATTGACACAG; and for IL-23p19, forward AGCGGGACATATGAATCTACTAAGAGA and reverse GTCCTAGTAGGGAGGTGTGAAGTTG-3. The real-time quantitative PCR primers that were used for IL-17 detection in the joint are specifically directed toward the family member IL-17A, which is expressed by T cells (23).

Intraarticular determination of Th17 cells.

Inflammatory cells were released from the inflamed joint at an early stage of inflammation, according to our previously published protocol (18), by dissection of the joint capsule followed by collagenase D (Sigma-Aldrich) digestion for 30 minutes at 2 mg/ml in warmed phosphate buffered saline. Afterward, cells were strained through a 70μ filter to inhibit contamination with stromal cells, stained for surface CD4 expression, fixed and permeabilized with BD Cytofix/Cytoperm (BD Bioscience), and then stained for intracellular IL-17A.

Statistical analysis.

For clinical disease and paw swelling curves, a statistical curve fit was used as in our previous studies (17, 18) to determine whether significant differences existed in disease over time between CX3CR1−/− mice versus wild-type controls. The advantage of this statistical test is its effectiveness for studying change. A backward selection (α = 0.05) procedure was used to select a linear mixed model with the best fit for the individual curves. Statistical variables included group, time, and experiment effect. The longitudinal analysis (mixed model) affords us the ability to distinguish variation that may be observed across time for a particular mouse from the variation among the group (24). The overall group effect was assessed using a likelihood ratio test (α = 0.05). The best-fit curves were plotted using predicted values calculated using the fixed effects from the models, averaging across experiment, which was controlled for if the curves were a significant (α = 0.05) predictor in the model using SAS version 9.1 software. For T cell proliferation, anticollagen antibody assays, intraarticular cytokine analysis by real-time quantitative PCR, and quantification of Th17 cells in inflamed joints by flow cytometry, an unpaired 2-tailed t-test was used to compare the means between groups.


Decreased severity of CIA in CX3CR1−/− mice compared to controls.

In humans with RA, increased CX3CR1 expression on circulating T cells (6, 25) and on T cells and monocytes in the synovial fluid (6) correlates with disease activity. To investigate whether or not the absence of CX3CR1 expression confers protection in inflammatory arthritis, we examined CX3CR1−/− mice and their wild-type controls in a model of chronic inflammatory arthritis (CIA) that is similar to RA in its histopathologic features, class II major histocompatibility complex restriction, and waxing–waning clinical course (16). We found that animals with a targeted genetic deletion of CX3CR1 had an ∼30% reduction in disease severity when compared to controls by 2 independent measures, paw swelling (P < 0.01) and clinical disease severity index (P < 0.0001) (Figures 1A and B). Although there was less inflammation in CX3CR1−/− mice (Figures 1A and B), they developed erosions (Figure 1D) that were histopathologically similar to those seen in the wild-type controls (Figure 1C).

Figure 1.

Decreased clinical inflammation in CX3CR1−/− mice with collagen-induced arthritis (CIA) compared to wild-type (WT) controls with CIA, by 2 independent measures. A and B, Changes in paw swelling (A) and the clinical disease index (B) in CX3CR1−/− mice (n = 14) and wild-type mice (n = 13). CX3CR1−/− mice had decreased paw swelling and clinical disease assessment scores compared to wild-type controls (P < 0.01 and P < 0.0001, respectively). Values are the mean ± SEM. C and D, Similar erosions at experiment termination on day 35 after the second booster immunization in wild-type mice (C) and CX3CR1−/− mice (D). Arrows indicate erosive pannus and inflammation.

Decreased formation of autoantibodies to CII in CX3CR1−/− mice compared to controls and similar T cell proliferative responses in the two groups.

CX3CR1 is expressed on monocytes (15), B cells (26), and dendritic cells (6, 27), which are APCs important for the induction of antibody responses. Additionally, the percent of dendritic cells expressing CX3CR1 is known to increase during joint inflammation in the rat model of AIA (6), which could affect antigen presentation and subsequent autoantibody formation. For these reasons, we hypothesized that adaptive immune responses may be affected as a result of the absence of CX3CR1 on immune cells, and the effects on CIA could extend beyond the migration abnormalities previously seen with blockade of the ligand CX3CL1 (8). Consequently, we examined levels of preclinical anti–CII autoantibodies, which are known to be pathogenic in CIA and to develop prior to the onset of clinical arthritis (20).

In CIA, T cell activation, proliferation, and initiation of B cell responses are initiated in the first 2 weeks after initial immunization (22); therefore, we examined lymph node–derived T cell proliferative responses as well as autoantibody formation at this time point. Anti-CII antibody responses in CX3CR1−/− mice were ∼50% lower than in wild-type controls on both day −7 (P < 0.05) and day 0 (P < 0.01), with substantially higher autoantibody levels in wild-type animals as they were approaching the onset of clinical disease (day 0) (Figure 2). In contrast, no significant differences were noted between the 2 groups in T cell activation and proliferation (Table 1).

Figure 2.

Decreased autoantibody production in CX3CR1−/− mice with CIA compared to wild-type controls with CIA. Formation of anti–type II collagen autoantibodies was substantially decreased in CX3CR1−/− mice (n = 13) compared to wild-type controls (n = 13) at time points that preceded clinical disease, and this difference was more pronounced over time. Values are the mean ± SEM. ∗ = P < 0.05; ∗∗ = P < 0.01, versus CX3CR1−/− mice. See Figure 1 for definitions.

Table 1. T cell proliferative responses*
 Wild-type miceCX3CR1−/− mice
  • *

    Enriched T cells from wild-type and CX3CR1−/− mice were stimulated with anti-CD3 and anti-CD28, with type II collagen (CII) and antigen-presenting cells (APCs), or with media alone. Values are the mean ± SEM.

T cells stimulated with anti-CD3/CD28, cpm31,850 ± 9,17828,240 ± 5,504
T cells stimulated with CII and APCs, cpm533.4 ± 105.9466.6 ± 26.34
T cells stimulated with media alone, cpm422.9 ± 78.91447.0 ± 85.41
T cells stimulated with CII, % of total stimulated T cells2.695 ± 0.94132.034 ± 0.3508
T cells stimulated with media alone, % of total stimulated T cells2.286 ± 0.91521.718 ± 0.2193

Selective decrease in Th17 cytokines and decreased total numbers of Th17 cells in the arthritic paws of CX3CR1−/− mice compared to controls.

Proinflammatory cytokines such as IL-1β, TNFα, IL-6, IL-17, IL-23, and MMPs have been shown to play an important role in the pathophysiology of inflammatory arthritis in both humans and animal models (28–32). Consequently, we examined the intraarticular production of these key proinflammatory mediators by real-time quantitative PCR (33) using severely inflamed joints (clinical score of 2) from animals with early inflammation (day 10). MMP-1, MMP-2, and MMP-9 levels were similar in the 2 groups (Figure 3B) as were CX3CL1 levels (Figure 3A). In CX3CR1−/− mice, levels of proinflammatory MMP-13 (Figure 3B) and cytokines IL-6, IL-1β, and TNFα (Figure 3A) were decreased compared to wild-type mice, but the differences were not statistically significant. In contrast, IL-17 levels were reduced >5- fold in CX3CR1−/− mice (P < 0.01 versus wild-type mice), and IL-23 levels were reduced >3-fold in CX3CR1−/− mice (P < 0.001 versus wild-type mice) (Figure 3A).

Figure 3.

Statistically significant decrease in relative gene expression of interleukin-17 (IL-17) and IL-23 in paws from CX3CR1−/− mice with early inflammatory arthritis. Paws from both CX3CR1−/− mice (n = 8) and wild-type (WT) mice (n = 8) with substantial clinical disease (score of 2) were collected at the earliest signs of clinical arthritis and analyzed for levels of A, proinflammatory cytokines and B, matrix metalloproteinases (MMPs) implicated in inflammatory arthritis. Only IL-17 and IL-23 showed a significantly reduced amount of relative gene expression in CX3CR1−/− mice compared to wild-type controls. Bars show the mean ± SEM. ∗∗ = P < 0.01; ∗∗∗ = P < 0.001. TNFα = tumor necrosis factor α.

To determine whether there were fewer Th17 cells in the inflamed joints of CX3CR1−/− mice, we dissected and collagenase-digested the joint capsules of severely inflamed paws (clinical score of 2) from wild-type mice (n = 10) and CX3CR1−/− mice (n = 7) with early arthritis (days 14–16) and analyzed the cellular infiltrate by flow cytometry. We chose the end points of days 14–16 for flow cytometry as compared to day 10 for quantitative real-time PCR to ensure that protein levels of IL-17 were elevated above the limit of detection after transcriptional up-regulation. CD4+IL-17+ T lymphocytes were identified in the joints of mice immunized to develop CIA, and CX3CR1−/− mice had a 3-fold decrease in the absolute number of articular Th17 lymphocytes as compared to wild-type mice (P < 0.05) (Figure 4).

Figure 4.

Reduced number of Th17 cells in the inflamed arthritic joints of CX3CR1−/− mice compared to wild-type (WT) mice. The joint capsule of inflamed paws (score of 2) from CX3CR1−/− mice (n = 7) and wild-type mice (n = 10) was dissected and dissolved with collagenase to release inflammatory cells, which were then stained to identify Th17 cells based on lymphocytes that had CD4+ surface expression and were positive for intracellular interleukin-17 by flow cytometry. Bars show the mean ± SEM absolute number of intraarticular Th17 cells per paw ∗ = P < 0.05.

To further examine the relationship between CX3CR1 and IL-17 on T cells, we analyzed lymphocytes from the draining lymph nodes (iliac and inguinal) of the mice by flow cytometry and identified a small subpopulation of cells expressing CX3CR1 above background levels that also expressed CD4 and IL-17. Specifically, of the wild-type mouse lymphocytes that stained positive for both CD4 and CX3CR1 (mean ± SEM 0.96 ± 0.15%; n = 5 wild-type mice), 5.14 ± 1.4% were also positive for IL-17. This population was too small to reliably perform migration studies. To investigate whether CX3CR1 deficiency had a functional effect on IL-17 production during antigen presentation in the lymph nodes after immunization, we harvested draining lymph nodes (iliac and inguinal) from mice 2 weeks after the initial immunization (day −7) and stimulated isolated T cells in vitro with anti-CD3/CD28 antibodies. CD4+ T cells were selected and analyzed for intracellular expression of IFNγ and IL-17, and no significant differences were seen between CX3CR1−/− mice and controls (Figure 5B). This may be due in part to the small fraction of CD4+CX3CR1+ T cells detected over background immunostaining for CX3CR1 as compared to the total population of CD4+ T cells (Figure 5A).

Figure 5.

Expression of CX3CR1 on CD4+ T cells in wild-type mice with CIA and production of Th1 and Th17 cytokines by stimulated CD4+ T cells in both wild-type and CX3CR1−/− mice. Draining lymph nodes (iliac and inguinal) were harvested from individual mice that had been immunized to develop CIA. A, Wild-type mouse lymphocytes were gated by forward scatter and side scatter and then for CD4+CX3CR1+ surface expression. CX3CR1−/− mouse lymphocytes were similarly gated as a staining control. Representative histograms are shown. Numbers inside the boxes are the percentages of dual-stained CD4+CX3CR1+ cells. B, Total lymphocytes isolated from draining lymph nodes were stimulated with anti-CD3 and anti-CD28 in culture, stained for surface CD4 and intracellular interleukin-17 (IL-17) or interferon-γ (IFNγ), and then analyzed by flow cytometry. Bars show the mean ± SEM (n = 5 mice per group). See Figure 1 for other definitions.


CX3CL1/CX3CR1 signaling has been established as an important proinflammatory chemokine receptor signaling interaction in chronic inflammatory diseases (for review, see ref.5), including RA (1, 6, 8, 34, 35) and atherosclerosis (10–12). Correlation studies have shown that CX3CL1 and/or CX3CR1 expression is increased in RA (34, 35) and is more elevated in patients with severe disease (35). In a mouse model of CIA, treatment of arthritic mice with a neutralizing antibody to CX3CL1, the unique ligand for CX3CR1, showed protection against disease that was mediated by inhibition of macrophage/monocyte trafficking to the joint (8). However, beyond the impaired macrophage/monocyte trafficking, changes in B and T cell function were not observed in those studies (8).

CX3CL1 is highly expressed on endothelial cells, intestinal epithelial cells, synoviocyte-like fibroblasts, and on some dendritic cells (6, 36–38). In contrast, its receptor CX3CR1 is expressed mostly on immune cells, such as monocytes, macrophages, dendritic cells, natural killer cells, and a subset of T cells, and recent reports suggest that it is expressed on a subset of B cells (11, 14, 15, 26). CX3CL1/CX3CR1 signaling activates the proinflammatory NF-κB pathway (39), and CX3CR1 deficiency is associated with decreased IL-6 and TNFα production by macrophages and dendritic cells (40). Because of its predominant expression on immune cells and multiple implicated mechanisms of immune regulation, we examined the genetic deletion of CX3CR1 in the CIA model to determine whether or not additional mechanisms beyond that of cellular trafficking may be affected in inflammatory arthritis when the receptor, as opposed to the ligand, is targeted.

The results of the present study show that CX3CR1 deficiency confers protection against CIA, in part, through decreased humoral and T cell responses. Since professional APCs (i.e., monocytes, macrophages, B cells, and dendritic cells) have CX3CR1 on their cell surface (9, 11, 15, 26, 27), it is conceivable that alteration of the CX3CR1+ subset of these APCs could lead to decreased production of autoantibodies, which are known to be pathogenic (20). Recently, Corcione et al (26) immunized CX3CR1−/− and wild-type mice with ovalbumin (OVA) to determine whether or not CX3CR1 deficiency affected antigen-specific antibody formation. The findings of that study were similar to the findings of the present study, in that OVA-specific IgG production was decreased (26). Lymphoid follicle architecture, B cell, and T cell enumeration did not differ; thus, the authors concluded that differences in autoantibodies could not be directly determined. Interestingly, Bar-On et al (27) have identified a specific subpopulation of CX3CR1+CD8β+ dendritic cells that share a gene signature overlapping with plasmacytoid dendritic cells. Since plasmacytoid dendritic cells are known to regulate B cell differentiation and antibody production (41), this CX3CR1+ dendritic cell subset may have functions that affect antibody production in inflammation.

The Th17 subset has recently been recognized as an important proinflammatory mediator in RA (30). Niess et al (40) showed that CX3CR1+ peptide-pulsed dendritic cells preferentially supported the differentiation of CD4+ Th17 cells in vitro in a model of inflammatory bowel disease. In the present study, we did not find a difference between CX3CR1−/− mice and wild-type mice in the ability to develop Th17 skewing in vitro, albeit there was a trend toward decreased Th1 IFNγ production in CX3CR1−/− mice, which would support the conclusions of Niess et al (40). We also acknowledge that only a small subset of CD4+ T cells express CX3CR1 (Figure 5A) after immunization to develop CIA, which may have limited our ability to detect a difference in the cytokine production assays.

An additional or alternative mechanism beyond that of impaired Th17 induction in CX3CR1−/− mouse lymph node cells could be the selective impairment of this particular T helper subset to migrate to the joint. Although we detected a subpopulation of CD4+T cells that also expressed CX3CR1 above background levels in wild-type mice (Figure 5A), the total numbers of IL-17+CD4+CX3CR1+ cells were low enough to raise questions as to the accuracy and reproducibility of this measurement. (Of the 1% of lymphocytes that were positive for CD4 and CX3CR1+, 5% were IL-17 positive.) The ideal experiment would be to examine CIA responses and CD4+IL-17+ cells using CX3CR1GFP reporter mice (4), which have green fluorescent protein knocked into the CX3CR1 gene locus. In this way, a more sensitive and accurate examination of the CX3CR1 expression pattern and migration of small cellular subsets that normally express this receptor (such as Th17 cells) could be more reliably conducted.

Although direct migration and enumeration of CD4+IL-17+CX3CR1+ cells in CIA was limited, we found a 3-fold decrease in numbers of CD4+IL-17+ cells in the inflamed paws of CX3CR1−/− mice compared to wild-type mice (Figure 4). Altered migration is suggested by the fact that there are fewer Th17 cells in the inflamed paws of CX3CR1−/− mice, but the evidence is indirect, and we cannot exclude the possibility of other mechanisms, such as enhanced apoptosis or impaired proliferation of Th17 cells.

Based on our findings of decreased levels of IL-23 in the joints of CX3CR1−/− mice, we additionally postulate that within the joint cytokine milieu, there may be environmental differences that affect local T cell cytokine secretion and function. Specifically, IL-23 is secreted by dendritic cells and induces the production of IL-17 by T cells (42). IL-23 is not needed for the de novo generation of Th17 cells but can augment IL-17 production from already generated Th17 memory cells (43). Consequently, decreased IL-23 levels within the local microenvironment of the joint may additionally affect the function of Th17 cells in CX3CR1−/− mice.

The CIA model is an approximation of human RA and does have limitations. Particularly, the inflammatory reaction is robust, and even in modest disease, extensive damage to the joint is seen at early and late time points. Therefore, a reduction in severity, as opposed to the elimination of disease, may not lead to statistically significant differences in histopathology scores in this model. One of our proposed mechanisms of protection is that loss of CX3CR1 impairs cellular trafficking to the joint, particularly of cells involved in the Th17 axis. However, inflammatory cells are not completely inhibited in trafficking to the joint, as was seen on histopathologic analysis (Figure 1), which may explain why disease severity is not significantly different at later time points. Additionally, levels of MMPs, which implement much of the direct tissue destruction in CIA, did not differ between CX3CR1−/− and wild-type mice, which could explain why bony erosions and cartilage loss did not differ significantly between the 2 groups.

We have previously noted the importance of CX3CL1/CX3CR1 signaling in atherosclerosis, which is attributed in part to alterations in inflammatory monocyte and dendritic cell trafficking to affected lesions (9–11). Accelerated atherosclerotic disease is becoming a widely recognized long-term complication of RA (13). In a recent observational study, Pingiotti et al (34) found that CD4+CX3CR1+ T cells isolated from the peripheral blood of RA patients were expanded when compared to healthy controls. Further, this increase in CD4+CX3CR1+ T cells in the RA patients correlated with increased carotid intima-media thickness and the Disease Activity Score in 28 joints (34). These data suggest that CX3CL1/CX3CR1 blockade may have long-term benefits that extend beyond inflammatory arthritis and into the prevention of early endothelial dysfunction that leads to atherosclerotic disease in these patients.


All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be published. Dr. Tarrant had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study conception and design. Tarrant, Liu, Patel, Fong.

Acquisition of data. Tarrant, Liu, Rampersad, Rothlein, Timoshchenko, McGinnis.

Analysis and interpretation of data. Tarrant, Liu, Esserman, Fitzhugh, Fong.


Author Patel is an employee of Novartis Institutes for BioMedical Research.


The authors would like to thank James Ellinger for technical assistance with the arthritis disease models and Dr. David Siderovski for critical review of the manuscript.