A Novel α9 Integrin Ligand, XCL1/Lymphotactin, Is Involved in the Development of Murine Models of Autoimmune Diseases

Integrins are transmembrane receptors involved in a wide range of cellular processes, including cell adhesion, migration, differentiation, proliferation, and apoptosis, and cancer metastasis (1–5). α9 integrin binds to various extracellular matrix proteins, including osteopontin (OPN), tenascin-C (TN-C), and vascular endothelial growth factor C and D, and is associated with autoimmune diseases such as arthritis (6–9). An anti-α9 integrin–blocking Ab was reported to ameliorate the disease scores in collagen Ab–induced arthritis (CAIA) (10) and experimental autoimmune encephalomyelitis (EAE) (11). However, inhibition or knockout of OPN has lesser effects than the use of the anti-α9 integrin Ab (12–14), and the involvement of other α9 integrin ligands in CAIA remains unknown, indicating that there may be other players in the development of autoimmune diseases. Therefore, we searched for a candidate α9 integrin ligand by liquid chromatography tandem-mass spectrometry (LC-MS/MS) analysis and identified XCL1/lymphotactin.

In this study, we demonstrate that XCL1 functions in autoimmune diseases as a novel α9 integrin ligand. A physiological interaction between α9 integrin and XCL1 was confirmed by pull-down assays. XCL1 belongs to the C-class of chemokines, and its only known receptor is XCR1 (15). We found that XCL1 stimulation enhanced cell migration in α9 integrin–expressing cells, and that this XCL1-dependent cell migration was suppressed by a function-blocking anti-α9 integrin Ab. To further investigate the role of XCL1 in vivo, we generated an XCL1-neutralizing mAb, 1A3A. The 1A3A Ab successfully protected mice against CAIA, indicating that XCL1 is involved in the development of this disease. We also administered the Ab to mice with EAE, and observed amelioration of the disease. Recently, XCL1 expression was demonstrated to be elevated in patients with rheumatoid arthritis (RA) (16), whereas α9 integrin expression was reported to be increased in the RA synovium (9). Taken together, it is suggested that the interaction between XCL1 and α9 integrin can be a new therapeutic target for autoimmune diseases.

NIH3T3 cells, Chinese hamster ovary (CHO) cells, HEK293T cells, Plat-GP cells, rhabdomyosarcoma (RD) cells (derived from a human rhabdomyosarcoma), and fibroblast-like synoviocytes (FLS) were cultured in DMEM containing 10% FBS (HyClone, Logan, UT). Ba/F3, an IL-3–dependent murine pro-B cell line, was maintained in RPMI 1640 medium supplemented with 10% FCS and 10% WEHI-3B conditioned medium as a source of IL-3. HRP-conjugated anti-human IgG (Jackson ImmunoResearch, West Grove, PA) was used for ELISA and Western blot analysis. Anti-rat IgG (Jackson ImmunoResearch) was used for ELISA. Anti-His-tag Ab OGHis (MBL, Nagoya, Japan) and anti-FLAG-tag Ab 1E6 (WAKO, Osaka, Japan) were used for Western blot analyses. Anti-human α9 integrin mAb Y9A2 (Chemicon, Temecula, CA) was used for flow cytometry and cell migration assays. An anti-α9 integrin polyclonal Ab for Western blot analyses was generated by immunizing a rabbit with a peptide derived from the C-terminal domain of human α9 integrin, CEAEKNRKENEDSWDWVQKNQ. IgG from human serum (Sigma-Aldrich, St. Louis, MO), and recombinant His-tagged human XCL1 protein (hXCL1-His) (PeproTech, Rocky Hill, NJ) were used for cell migration assays.

CHO cells expressing α9 integrin or α4 integrin, and NIH3T3 cells expressing α9 integrin or SFα9, or coexpressing α9 integrin and SFα9 were generated as previously described (17). Briefly, α9-FLAG-pBabepuro, α4-FLAG-pBabepuro, α9-pWZLBlast2, or SFα9-FLAG-pBabepuro was cotransfected with pCMV-VSV-G into Plat-GP packaging cells using Lipofectamine 2000 (Invitrogen, Carlsbad, CA). At 3 d after transfection, virus-containing supernatants were harvested, passed through a 0.45-μm filter, added to 40% confluent CHO cells or NIH3T3 cells in the presence of 8 μg/ml polybrene, and cultured for 1 d. The virus-containing medium was removed and the cells were cultured in 10% FBS/DMEM supplemented with 10 μg/ml puromycin (Sigma-Aldrich) or 5 μg/ml blasticidin (Invitrogen) for NIH3T3 cells and CHO cells. For generation of cells coexpressing α9 integrin and SFα9, retroviruses were generated by transfection of SFα9-FLAG-pBabepuro, and then applied to α9/NIH3T3 cells, followed by selection with 10% FBS/DMEM supplemented with 10 μg/ml puromycin and 5 μg/ml blasticidin. Cells expressing α9 integrin and/or SFα9 were identified by flow cytometry with anti-α9 Ab Y9A2, and Western blot with the rabbit anti-α9 integrin polyclonal Ab. For α9-FLAG/CHO and α4-FLAG/CHO cells, expression was confirmed by Western blot with anti-FLAG Ab 1E6. Stable Ba/F3 transformants were generated as follows. A cDNA for human αL integrin was PCR-amplified from human B lymphoblast cell line SKW6-CL4 with following primers: 5′-GGCGCCGGCCGGATCCGCCACCATGAAGGATTCCTGCATCAC-3′ (sense) and 5′-ATTCCACAGGGTCGACTTACTTGTCATCGTCATCCTTGTAGTCCTTGCCACCACCACTC-3′ (antisense). Amplicon was cloned into BamHI (Takara, Kusatsu, Japan)- and SalI (Takara)-digested pBabepuro vector using the In-Fusion system (Takara) to construct αL-FLAG-pBabepuro. α9-FLAG-pBabepuro, α4-FLAG-pBabepuro, and αL-FLAG-pBabepuro were then transfected to Ba/F3 by electroporation using Gene Pulser II (Bio-Rad, Richmond, CA), followed by selection with Ba/F3 maintaining medium supplemented with 2 μg/ml puromycin.

NIH3T3 cells expressing SFα9-FLAG-His were exposed to hypotonic buffer (25 mM NaCl, 0.5 mM CaCl₂, 18 mM Tris–HCl pH 8) and the membrane fraction was collected by centrifugation at 4000 × g for 10 min at 4°C. After removal of the supernatant, the membrane fraction was lysed with CelLytic M (Sigma-Aldrich) containing a protease inhibitor mixture 1× Complete Mini Protease Inhibitor Cocktail (Roche, Basel, Switzerland), followed by centrifugation at 15,000 × g for 10 min at 4°C. The supernatant was applied to Anti-FLAG M2 Affinity Gel (Sigma-Aldrich). After washing with TBS, the bound proteins were eluted with TBS containing a FLAG peptide. The eluted proteins were checked by Western blot, and then concentrated by 20-fold using a Vivaspin (Sartorius, Göttingen, Germany) for LC-MS/MS analysis.

XCL1-Ig fusion proteins were generated as follows. A cDNA for mouse XCL1 was PCR-amplified from the thymi of C57BL/6 mice using the following primers: 5′-AGTGAATTCGCCACCATGAGACTTCTCCTCCTGA-3′ (sense) and 5′-TGTGGATCCCCAGTCAGGGTTATCGCTGT-3′ (antisense). cDNAs for human XCL1 and human XCL2 were purchased from Open Biosystems, and PCR amplified using the following primers: 5′-GGTGAATTCGCCACCATGAGACTTCTCATCCTGG-3′ (sense) and 5′-ATGGGATCCCCAGTCAGAGTCACAGCTGTA-3′ (antisense). The PCR products were digested with EcoRI (Takara) and BamHI and inserted into an EcoRI- and BamHI-digested mammalian expression vector containing the Fc portion of human IgG1. The resulting expression vectors were transfected into HEK293T cells, and recombinant Ig-tagged mouse XCL1 (mXCL1-Ig), human XCL1 (hXCL1-Ig), and human XCL2 (hXCL2-Ig) proteins were purified from the supernatants with protein A-Sepharose beads (GE Healthcare, Little Chalfont, U.K.). For the generation of various His-SUMO-tagged mouse XCL1 variants, the target genes were PCR amplified using the following primers: 5′-TATTGAGGCTCATCGCGAACAGATTGGAGGTGTGGGGACTGAAGTCCTA-3′ (sense) and 5′-ATGCCTGCAGGTCGACTTACCCAGTCAGGGTTATCGCTG-3′ (antisense) for full, 5′-TATTGAGGCTCATCGCGAACAGATTGGAGGTGTGGGGACTGAAGTCCTA-3′ (sense) and 5′-ATGCCTGCAGGTCGACTTAAGTCTTGATCGCTGCTTTCA-3′ (antisense) for 22–83, 5′-TATTGAGGCTCATCGCGAACAGATTGGAGGTGGGGCCATGAGAGCTGTA-3′ (sense) and 5′-ATGCCTGCAGGTCGACTTACCCAGTCAGGGTTATCGCTG-3′ (antisense) for 53–114, 5′-TATTGAGGCTCATCGCGAACAGATTGGAGGTATTTGTGCTGATCCAGAA-3′ (sense) and 5′-ATGCCTGCAGGTCGACTTACCCAGTCAGGGTTATCGCTG-3′ (antisense) for 68–114, 5′-TATTGAGGCTCATCGCGAACAGATTGGAGGTGTGGATGGCAGGGCCAGT-3′ (sense) and 5′-ATGCCTGCAGGTCGACTTACCCAGTCAGGGTTATCGCTG-3′ (antisense) for 84–114. The PCR products were subcloned into the NruI- and SalI-digested pMsec SUMOstar vector (LifeSensors, Malvern, PA), a mammalian expression vector containing a His-SUMO tag, using the In-Fusion system. HEK293T cells were transfected with the individual expression vectors, and the supernatants were collected at 48 h after transfection. Secreted recombinant XCL1 variants were confirmed by Western blot.

NIH3T3, CHO, and RD cells were lysed with CelLytic M by gentle rotation at 4°C for 15 min. The lysates were clarified by centrifugation at 15,000 × g for 15 min at 4°C, and then incubated with recombinant human Fc-G1 (Bio X Cell, West Lebanon, NH), mXCL1-Ig, hXCL1-Ig, or hXCL2-Ig (5 μg/tube) for 30 min at 4°C. Protein A-Sepharose (50% slurry, 12 μl/tube) was applied to each tube and incubated for another 1 h at 4°C. After three washes of the beads with TBS containing 0.05% Tween-20, the precipitated polypeptides were extracted in SDS sample buffer (93.75 mM Tris–HCl pH 6.8, 30% glycerol, 15% 2-ME, 7.5% SDS), separated by SDS–PAGE under reducing conditions, transferred to polyvinylidene fluoride membrane, probed with the anti-α9 integrin Ab or anti-FLAG Ab 1E6, and detected by Plus-ECL (PerkinElmer, Waltham, MA).

For adherent cells, PBS containing 0.5 mM EDTA was used to dissociate cells from the culture dish, followed by blocking with FBS, and then sequentially incubated with anti-α9 integrin Ab Y9A2 and PE-labeled goat anti-mouse Ab (Jackson ImmunoResearch) for human α9 integrin expression, APC-labeled anti-α9 integrin Ab 55A2C for murine α9 integrin expression (10), and PE-labeled mouse anti-XCR1 Ab ZET (BioLegend, San Diego, CA) for XCR1 expression before flow cytometry analysis. Ba/F3 cells were washed with PBS, blocked by FBS, then incubated with PE-labeled anti-αL integrin (BioLegend), PE-labeled anti-α4 integrin 9F10 (BioLegend), and anti-α9 integrin Ab Y9A2 and PE-labeled goat anti-mouse Ab.

Anti-XCL1 Abs (clones 1A3A and 3P11R) were generated in Sprague-Dawley rats immunized with recombinant XCL1 emulsified in Freund’s adjuvant. Their splenocytes were fused with ×63-Ag8-653 mouse myeloma cells as described previously (18). The resulting hybridoma cells were screened by ELISA. For characterization of the cloned Abs, ELISA and Western blot analyses were performed using Ig-tagged or His-tagged mouse and human XCL1, and His-SUMO-tagged XCL1 variants.

Nunc-immuno plates (Thermo Fisher Scientific, Roskilde, Denmark) were coated with either 1 μg/ml recombinant mouse integrin α9β1 (R&D Systems, Minneapolis, MN) or 10 μg/ml XCL1 proteins overnight at 4°C in 0.1 M carbonate-bicarbonate buffer (pH 9.2). After PBS wash, α9β1- and XCL1-coated plates were blocked with 5% skim milk and 0.1% BSA in PBS respectively for 1 h at room temperature. Plates were then washed once with washing buffer (PBS containing 0.05% Tween-20), followed by application of 5 μg/ml recombinant human Fc-G1 and XCL1-Ig to α9β1-coated plate and 10 μg/ml anti-mouse XCL1 Abs 1A3A and 3P11R to XCL1-coated plates. After incubation for 1 h at room temperature, plates were washed three times with washing buffer, then HRP-conjugated anti-human IgG and HRP-conjugated anti-rat IgG were applied to α9β1-coated plates and XCL1-coated plates respectively for another 30 min incubation at room temperature. Plates were then washed four times with washing buffer and the substrate 3,3′,5,5′-tetramethylbenzidine (KPL, Gaithersburg, MD) was added. After 15 min of incubation at room temperature, an equal amount of 1 N sulfuric acid was applied to stop the reaction. Absorbance was measured at 450 nm (reference wavelength of 620 nm) using an iMark microplate reader (Bio-Rad).

For chemotactic migration assays, 24-well Transwell plates (Corning, Corning, NY) and RPMI 1640 medium containing 1% FBS were used. Transwell filters (5.0-μm pore size) were equilibrated at 37°C overnight. The equilibration medium was then removed, and medium containing human IgG as an Ig control, mXCL1-Ig, or hXCL1-His (indicated concentration for Fig. 2A and 300 ng/ml each for other experiments) was added to the bottom chambers. Next, 1 × 10⁶ cells were added to the upper chambers and allowed to migrate for 5 h at 37°C. After unmigrated cells were gently removed from the membranes with a cotton swab, the membranes were fixed and stained with Diff-Quik (Sysmex, Kobe, Japan). The stained cells were counted in four randomly chosen high-power fields per filter. For anti-α9 integrin Ab treatment, cells were incubated with 10 μg/ml Y9A2 or control IgG prior to addition to the upper chambers. In a checkerboard type migration assay, cells were treated with either PBS of mXCL1-Ig, then applied to the upper chamber and allowed to migrate toward the lower chamber containing media supplemented with PBS, hIgG1 Fc control, or mXCL1-Ig. For barium treatment, barium chloride was added to cells and applied to the upper chamber. For anti-mXCL1 Ab treatment, medium containing mXCL1-Ig was treated with PBS, 1A3A, or 3P11R (10 μg/ml) and incubated for 15 min on ice prior to addition to the bottom chambers.

BALB/c and C57BL/6 mice were kept under specific pathogen-free conditions and provided with food and water ad libitum. Sprague-Dawley rats were purchased from Charles River Japan (Yokohama, Japan) and kept under conventional conditions. All experiments were conducted in accordance with the guidelines of the Institutional Animal Care and Use Committee of Hokkaido University.

Hind limbs were surgically cut from the ankle joint, followed by surgical separation of the dermal, s.c., tendinous, and muscle tissues from joints. The remaining soft tissues were removed from bone tissues and homogenized roughly with dissecting scissors. Homogenates were washed once with culture medium and treated with 3 mg/ml type II collagenase (Worthington Biochemical, Lakewood, NJ) in FBS-free DMEM at 37°C by vigorous stirring. After washing and straining with a 70 μm cell strainer (Greiner Bio-One, Frickenhausen, Germany), cells were cultured overnight. Non-adherent cells were removed, and attached cells were regarded as FLS.

Arthritis was induced using an arthritogenic mAb mixture kit (Chondrex, Redmond, WA). Briefly, 6–8 wk-old female BALB/c mice (Charles River) were injected intravenously with a mixture of anti-type II collagen mAbs on day 0, followed by i.p. injection of 50 μg of LPS (0111:B4) on day 3. FLS were obtained from the joints and expression of XCL1 was evaluated by quantitative PCR (qPCR). For XCL1 treatment, Ig-fused mouse XCL1 or Ig control was administered i.p. at a dose of 300 ng per mouse from days −1 to 9. For α9 integrin inhibition, anti-α9 integrin Abs 18R18D or 55A2C (10) were i.p. administered to XCL1-treated mice on days −1 and 2 as a control IgG and α9 integrin-neutralizing Ab, respectively. For XCL1 inhibition, anti-mouse XCL1 Ab 3P11R or 1A3A was i.p. injected on days −1 and 2 as a control IgG and XCL1-blocking Ab, respectively. The clinical severity of arthritis was graded for up to 16 d after Ab administration in each of the four paws on a scale of 0–4 as follows: 0, no signs; 1, slight focal swelling and/or redness in one digit; 2, moderate swelling and erythema of more than two digits; 3, marked swelling and erythema of the limb; 4, maximal swelling, erythema, deformity, and/or ankyloses. Mice were sacrificed, and then joint sections were stained with H&E and safranin-O for immunohistochemical evaluation.

EAE was induced in 8 wk-old C57BL/6 mice by s.c. injection with 100 μg of myelin oligodendrocyte glycoprotein (MOG) 35–55 peptide (MEVGWYRSPFSRVVHLYRNGK) emulsified with CFA on day 0 and i.v. injection with 400 ng of pertussis toxin (List Biological Laboratories, Campbell, CA) on days 0 and 2. Anti-mouse XCL1 Ab 1A3A or 3P11R, or PBS as a control, was administered at a dose of 400 μg per mouse at 1 d before and 2 d after MOG 35–55 immunization. The clinical scores of EAE severity were assessed daily as previously described (19): 0, no clinical signs; 0.5, partially limp tail; 1, paralyzed tail; 2, loss of coordinated movement, hind limb paresis; 2.5, one hind limb paralyzed; 3, both hind limbs paralyzed; 3.5, hind limbs paralyzed, weakness in forelimbs; 4, forelimbs paralyzed; and 5, moribund.

Total RNA from short hairpin RNA–transduced cells, FLS, and the spinal cords of EAE mice at days 0, 7, 14, 21, and 28 was extracted with TRIzol (Invitrogen). Random primers were used for RT-PCR and specific primers were used for qPCR assays to amplify hypoxanthine-guanine phosphoribosyltransferase (HPRT) (5′-TCCTCCTCAGACCGCTTTT-3′ and 5′-CCTGGTTCATCATCGCTAATC-3′), β-actin (5′-TGACAGGATGCAGAAGGAGA-3′ and 5′-CGCTCAGGAGGAGCAATG-3′), XCL1 (5′-AGACTTCTCCTCCTGACTTTCCT-3′ and 5′-GGACTTCAGTCCCCACACC-3′), XCR1 (5′-GCACTGGAGGAGATCAAAGG-3′ and 5′-CGGGATGCAGGGATACTGAG-3′), and α9 integrin (5′-ATGACGGGTTCCCAGATG-3′ and 5′-TGTAGACTGCGCCAGCAA-3′). The qPCR assays were conducted in a CFX96 Touch (Bio-Rad). The amplified cDNAs was detected using SYBR Green (Kapa Biosystems, Woburn, MA) and standardized by the ROX dye levels. The cDNA concentrations were expressed as the number of cycles to threshold, and the threshold values were normalized by the HPRT cDNA levels in the same samples.

Data are presented as mean ± SEM and are representative of at least three independent experiments. The statistical significance of differences between groups was calculated using a two-tailed Student t test or nonparametric Wilcoxon Mann–Whitney U test where applicable. Differences were considered significant for p < 0.05 or p < 0.01.

First, we determined candidates for novel ligands of α9 integrin by LC-MS/MS using α9 integrin–splicing variant SFα9, which contains only the β-propeller domain and thigh domain (17). We previously reported that SFα9 does not interact with known ligands of α9 integrin, thus SFα9 ligands remain unknown. Because SFα9 contains the β-propeller domain, which is important for ligand binging of α9 integrin, it is possible that novel ligands of SFα9 also have a binding capability to native α9 integrin and regulate cellular functions. Therefore, we postulated that either selective ligands of SFα9 or common ligands of α9 integrin and SFα9 could be novel ligands involved in autoimmune diseases by regulating α9 integrin activity. Among many candidates (Supplemental Fig. 1A), we focused on the C-class chemokine XCL1 as a functional ligand of α9 integrin. A previous study has demonstrated that secretory proteins such as vascular endothelial growth factor C and D are the ligand of α9 integrin (8), suggesting the secretory protein XCL1 may also be an α9 integrin ligand. Because secreted chemokines induce chemotaxis through binding to chemokine receptors (20), we hypothesized that XCL1 may act as a chemokine for α9 integrin. To verify the interactions between XCL1 and SFα9, and XCL1 and α9 integrin, we generated recombinant XCL1 and XCL2 proteins fused with IgG1 Fc (Supplemental Fig. 1B). XCL2 is a homolog of XCL1 in humans with only 2 aa differences (21), suggesting that XCL2 could also be a ligand of α9 integrin. In pull-down assays using recombinant XCL1-Ig and XCL2-Ig, we confirmed the interactions of XCL1 with both α9 integrin and SFα9 in cells overexpressing α9 integrin, SFα9, or both (Fig. 1A). XCL2 also bound to α9 integrin and SFα9, but its interactions were weaker than those of XCL1. NIH3T3 cells did not express endogenous α9 integrin (Supplemental Fig. 1C), and showed no enhanced cell migration toward XCL1 (Fig. 2B). We also observed the interaction of XCL1 with α9 integrin in RD cells, the human rhabdomyosarcoma cell line that endogenously expresses α9 integrin (Fig. 1B), indicating interaction between α9 integrin and XCL1 may occur endogenously. α9 integrin is structurally and functionally similar to α4 integrin, and shares some common ligands such as OPN and propolypeptide of von Willebrand factor (22, 23). Therefore, we examined whether the interaction between XCL1 and α9 integrin is selective. In pull-down assays using α4 integrin– and α9 integrin–overexpressing cells, α4 integrin showed no interaction with XCL1 (Fig. 1C), indicating that XCL1 selectively binds to α9 integrin. We also determined the selective binding of XCL1 with αL integrin. Because αL integrin requires β2 integrin to be expressed on the cell surface as a heterodimer, we transduced αL integrins into murine pro-B cell line Ba/F3, which expresses β2 integrin endogenously. Ba/F3 transformants express these integrin heterodimers on the cell surface (Supplemental Fig. 1D). Pull-down assays with these cells showed selective interaction between XCL1 and α9 integrin (Supplemental Fig. 1E). Physical interaction of α9β1 integrin with XCL1 was also observed in direct ELISA (Fig. 1D), suggesting XCL1 is a novel α9 integrin ligand.

XCL1 is a C-class chemokine that induces chemotaxis through its known receptor XCR1 (24, 25). Therefore, we examined whether XCL1 induces migration of cells expressing α9 integrin. In cells overexpressing α9 integrin, XCL1 showed a biphasic chemotactic response as previously reported (21), with the best response observed at 300 ng/ml XCL1 concentration (Fig. 2A). XCL1 did not induce cell migration in parent cells (Fig. 2B), and NIH3T3 cells did not express XCR1 (data not shown), indicating XCL1 acted on α9 integrin. We then treated the cells with anti-α9 integrin–neutralizing Ab Y9A2. As a result, XCL1-induced cell migration in α9 integrin–expressing cells was significantly inhibited (Fig. 2B). Checkerboard-type migration assay also confirmed this migration is directional, indicating XCL1 induced chemotaxis on α9 integrin–expressing cells (Fig. 2C). We also examined the effects of XCL1 and Y9A2 in RD cells. XCL1 successfully enhanced the cell migration of RD cells, and Y9A2 treatment caused inhibition of this migration (Fig. 2C). In a previous paper, it was reported that α9 integrin–dependent cell migration is dependent on inward-rectifier potassium (Kir) channels and that Kir channels can be blocked by Ba²⁺ (26). Therefore, we investigated whether XCL1-induced migration is dependent on α9 integrin by using Ba²⁺. We confirmed diminished cell migration toward XCL1 in the presence of Ba²⁺ (Fig. 2D). These results suggest that XCL1-induced migration is α9 integrin–dependent.

It was previously reported that α9β1 integrin is involved in the development of CAIA (10). Therefore, we investigated the in vivo role of XCL1 in CAIA. Real-time PCR of FLS revealed that XCL1 expression was increased in CAIA mice (Fig. 3A), suggesting local expression of α9 integrin and XCL1 may be involved in the development of inflammatory arthritis. To examine the effect of XCL1 in CAIA, we administrated recombinant Ig-fused XCL1 consecutively and observed significant exacerbation of arthritis in XCL1-treated mice (Fig. 3B, 3C). H&E staining revealed severe erosion of the joint cartilage (Fig. 3D, box 1) and infiltration of inflammatory cells (Fig. 3D, box 2) in XCL1-treated mice. We then stained the joints with safranin-O to visualize proteoglycans in the arthritic cartilage, and observed more extensive destruction of proteoglycans in XCL1-treated mice (Fig. 3D). From these results, we confirmed that XCL1 deteriorates arthritis in CAIA.

Blocking of α9 integrin was reported to ameliorate CAIA (10). Thus, we treated XCL1-administered CAIA mice with α9-integrin neutralizing Ab 55A2C (Fig. 3E). Inhibition of α9 integrin with the Ab significantly reduced the clinical score of the CAIA by XCL1 (Fig. 3F). We then stained the joint with H&E and safranin-O, and revealed that the erosion of the joint cartilage, infiltration of inflammatory cells, and destruction of proteoglycans were improved by blocking α9 integrin in XCL1-treated CAIA mice (Fig. 3G). To further investigate the role of XCL1 in CAIA, we isolated FLS from CAIA mice. We confirmed that FLS expressed α9 integrin (Supplemental Fig. 2A) as reported previously (10). Upon FLS stimulation with XCL1, various inflammatory cytokines were upregulated (Supplemental Fig. 2C), and this response was successfully inhibited by treating FLS with α9 integrin–neutralizing Ab (Supplemental Fig. 2D). Therefore, it is suggested that XCL1 aggravated CAIA in an α9 integrin–dependent manner by inducing cell migration of α9 integrin–expressing cells as well as increasing the mRNA expression level of inflammatory cytokines.

To further evaluate the effects of XCL1, we generated anti-XCL1 mAbs that block XCL1 function. First, we characterized two of our cloned anti-XCL1 mAbs, 1A3A and 3P11R, for their specificity. In ELISA, both 1A3A and 3P11R recognized Ig-tagged and His-tagged mouse XCL1, but showed no binding toward human XCL1 (Supplemental Fig. 3A). Thus, we confirmed that the generated Abs were specific for mouse XCL1 and showed no cross-reactivity with human XCL1. We then determined the binding regions of these mAbs. For this, we generated various recombinant XCL1 proteins with different lengths (Fig. 4A). When the secretion of recombinant XCL1 proteins was examined by Western blot, we noticed that proteins with C-terminal region of XCL1 showed smear bands at higher m.w. than expected. The C-terminal region of XCL1 exists as an α-helix, resulting in the formation of complexes causing the smear bands. In ELISA, both 1A3A and 3P11R recognized the closed C-terminal region of XCL1, at least from amino acid positions 68–144 (Fig. 4B, Supplemental Fig. 3B). Of note, when the C terminus of XCL1 was shortened to aa 84–114, the two Abs did not detect the protein, suggesting that the epitopes of these two mAbs were dependent on the conformation of the C terminus of XCL1. To determine their neutralizing effects, we treated α9 integrin–expressing cells with 1A3A and 3P11R. The 1A3A-treated cells showed significant inhibition of cell migration toward XCL1, whereas the 3P11R-treated cells exhibited no effect (Fig. 4C). We then examined whether these observations arose through enhanced cell death caused by 1A3A. In cell viability assays, neither 1A3A nor 3P11R had any effect on cell viability (Fig. 4D). Therefore, we successfully established 1A3A as an XCL1-neutralizing Ab.

Next, we investigated the involvement of XCL1 in CAIA using our anti-XCL1 mAbs. Mice were treated with 1A3A or 3P11R during CAIA induction (Fig. 5A). Because 3P11R showed no inhibitory effect in migration assays (Fig. 4C), we used this clone as a control Ab. In mice administered XCL1-neutralizing Ab 1A3A, clinical score and immunohistochemical staining showed that the severity of arthritis was significantly decreased compared with that in mice treated with 3P11R (Fig. 5B, 5C). Last, we determined the association of XCL1 with EAE, another autoimmune disease model involving α9 integrin (11). As the disease progressed, the gene expression levels of XCL1 and XCR1 were significantly increased, together with elevated α9 integrin expression in the spinal cord at day 7 (Fig. 5D). We then administered our anti-XCL1 mAbs to EAE mice (Fig. 5E). Treatment with α9 integrin–neutralizing Ab 1A3A was observed to suppress the disease progression (Fig. 5F). Taken together, these findings show XCL1 is involved in the development of CAIA and EAE.

α9 integrin is expressed on various cells including airway epithelial cells and smooth muscle cells (27), as well as cancer cells such as melanoma and breast tumor cells (28, 29). The known ligands for α9 integrin are OPN, TN-C, propolypeptide of von Willebrand factor, fibronectin-EIIIA, and polydom (6, 7, 23, 30, 31). α9 integrin is involved in autoimmune diseases such as RA, and in mouse models of inflammatory arthritis and EAE (9–11). A neutralizing Ab for α9 integrin was reported to inhibit the disease scores in CAIA and EAE (10, 11). However, investigations of CAIA using OPN-knockout mice or anti-OPN Ab treatment showed lesser effects compared with anti-α9 integrin Ab treatment (12–14). Therefore, we hypothesized that other ligands of α9 integrin are also involved in autoimmune diseases.

To search for novel α9 integrin ligands, we performed LC-MS/MS analyses using SFα9 and identified XCL1/lymphotactin as one of the candidates for an SFα9-binding partner. We showed that XCL1 selectively interacted with α9 integrin and SFα9 by pull-down assays. XCL1 is a C-class chemokine secreted by various immune cells including CD8⁺ T, CD4⁺ T, γδ T, NK, NKT, and medullary thymic epithelial cells (20, 25, 32–36). It acts on the known receptor XCR1, which is mainly expressed on CD8α⁺ dendritic cells to induce cell migration (37). Of note, XCL2, another member of the C-class chemokines that differs from XCL1 by only 2 aa (38), also bound to α9 integrin and SFα9. Compared with XCL1, the interactions of XCL2 with α9 integrin and SFα9 were weaker, suggesting the 2 aa difference between XCL1 and XCL2 may alter their functions. Considering that XCL2 showed a weaker interaction with α9 integrin than XCL1, it is possible that the cell migration of α9 integrin–expressing cells toward XCL2 may be diminished. More detailed experiments are necessary to elucidate the functional differences between XCL1 and XCL2. We also confirmed that XCL1 could directly interact with α9 integrin using ELISA. To our knowledge, our result is the first to demonstrate the direct binding of secretory chemokine to the integrin receptor. We found that XCL1 enhanced the cell migration of α9 integrin–expressing cells, and that XCL1-induced cell migration was inhibited by an α9 integrin–neutralizing Ab. α9 integrin–dependent migration involves Kir channels, which can be blocked by Ba²⁺ with a consequent decrease in the migration ability of α9 integrin–expressing cells (26). We successfully showed that the enhanced migration of α9 integrin–expressing cells induced by XCL1 was blocked by Ba²⁺. These results suggested that the XCL1-induced cell migration occurred in an α9 integrin–dependent manner. Real-time qPCR and flow cytometry analyses revealed there was little expression of XCR1 on NIH3T3 cells used in the migration assays, indicating that XCL1 directly interacted with α9 integrin to induce cell migration. We further knocked down XCR1 by a short hairpin RNA (Supplemental Fig. 4A, 4B) and found that cell migration toward XCL1 was not altered (Supplemental Fig. 4C), thereby supporting the notion that the enhanced cell migration was independent of XCR1.

In our in vivo experiments, we examined whether XCL1 is associated with CAIA and found that continuous administration of recombinant XCL1 during the development of CAIA exacerbated the disease. In a previous report, α9 integrin expressed on synovial fibroblasts was found to interact with locally produced OPN and TN-C, followed by secretion of various chemokines that resulted in the induction of inflammatory arthritis (10). Based on our results, it is suggested that XCL1 possibly acted on synovial fibroblasts and caused local infiltration of cells. In fact, we showed a blockade of α9 integrin by Ab treatment ameliorated the disease severity in XCL1-treated CAIA mice. We determined our hypothesis that FLS could be one of the cells where XCL1-α9 integrin interaction occurs by ex vivo analysis. As previously reported (10), FLS isolated from CAIA mice expressed α9 integrin, but not XCR1 (Supplemental Fig. 2A). Stimulation of FLS with XCL1 enhanced mRNA expression of inflammatory cytokines such as TNF-α, TGF-β, IL-6, and others (Supplemental Fig. 2C). Because FLS expressed no XCR1, it was suggested that XCL1 interaction with α9 integrin upregulated the expression of these cytokines. A previous study showed extracellular matrix proteins such as OPN and TN-C, which are the known ligands of α9 integrin, induced mRNA expression of IL-6 and IL-1α in FLS (10). This suggests that XCL1 bound to α9 integrin as a ligand and induced mRNA expression of inflammatory cytokines in a similar manner to the previous report. In addition, neutralizing α9 integrin Ab 55A2C inhibited the XCL1 stimulation on FLS (Supplemental Fig. 2D), supporting that XCL1 is a novel ligand involved in the development of autoimmune disease. It has also been reported that α9 integrin is associated with RA (9), and that XCL1 is elevated in plasma from RA patients (16). These studies suggest that XCL1 could be involved in the development of RA in an α9 integrin–dependent manner. However, we could not determine the involvement of XCL1-XCR1 interaction in vivo. Further study using either mutant XCL1, which has no binding capability to α9 integrin, or an Ab that specifically blocks interaction between α9 integrin and XCL1 will clarify the detailed mechanism.

We further examined whether aggravation of CAIA could be treated by the blockade of XCL1. However, there were no available functional mAbs against murine XCL1. Therefore, we generated Abs against XCL1, and established two clones, 1A3A and 3P11R. We examined the epitopes of these Abs, and found that both Abs recognized the C terminus of XCL1. Further investigations are required to fully identify the epitopes of these generated Abs. We also confirmed that 1A3A had a neutralizing ability against XCL1 in migration assays using α9 integrin–expressing NIH3T3, and thus used 1A3A and 3P11R for in vivo analyses as a functional XCL1-blocking Ab and its control Ab, respectively. Using the two generated Abs, we showed that XCL1-neutralizing Ab 1A3A diminished the severity of arthritis in CAIA mice. Our results indicated that XCL1 is involved in the development of CAIA. However, this study could not clarify the involvement of XCR1 in CAIA development.

We further demonstrated that anti-XCL1 Ab 1A3A inhibited the progression of EAE, another autoimmune disease model. In EAE, activated α9 integrin expressed on lymphatic endothelial cells enhances the egress of lymphocytes by interacting with TN-C (11). XCL1 could play a role in this process by binding to α9 integrin as a novel ligand and inducing the cell migration of lymphocytes. It is also possible that local elevation of XCL1 could directly enhance the infiltration of α9 integrin–expressing lymphocytes such as dendritic cells and macrophages, resulting in the aggravation of EAE. Further investigations using XCR1-neutralizing Abs or XCR1-deficient mice are required to fully elucidate the relationships between α9 integrin, XCL1, and XCR1 in various autoimmune diseases.

In conclusion, we have shown the novel interaction of α9 integrin and XCL1 and XCL1 involvement in the development of autoimmune diseases. In general, integrins interact with various matricellular proteins such as fibronectin, OPN, and TN-C to promote various cellular functions. Chemokines induce integrin-mediated signals through indirect mechanisms, wherein they bind to chemokine receptors and transduce their signals intracellularly to activate integrins, known as inside-out activation (39, 40). To our knowledge, our study is the first to demonstrate a direct interaction between an integrin and a chemokine, and to suggest that XCL1 is a functional ligand of α9 integrin in autoimmune diseases. Therefore, the XCL1–α9 integrin axis may be a potential therapeutic target for treating autoimmune diseases.

We thank D. Sheppard (University of California, San Francisco) for providing the pBabepuro and pWZL-blast2 vectors.

The authors have no financial conflicts of interest.