Chemokines and their receptors determine the distribution of leukocytes within tissues in health and disease. We have studied the role of the constitutive chemokine receptor CXCR4 and its ligand, stromal-derived factor-1 (SDF-1) in the perivascular accumulation of T cells in rheumatoid arthritis. We show that synovial T cells, which are primed CD45RO+CD45RBdull cells and consequently not expected to express constitutive chemokine receptors, have high levels of the chemokine receptor CXCR4. Sustained expression of CXCR4 was maintained on synovial T cells by specific factors present within the synovial microenvironment. Extensive screening revealed that TGF-β isoforms induce the expression of CXCR4 on CD4 T cells in vitro. Depletion studies using synovial fluid confirmed an important role for TGF-β1 in the induction of CXCR4 expression in vivo. The only known ligand for CXCR4 is SDF-1. We found SDF-1 on synovial endothelial cells and showed that SDF-1 was able to induce strong integrin-mediated adhesion of synovial fluid T cells to fibronectin and ICAM-1, confirming that CXCR4 expressed on synovial T cells was functional. These results suggest that the persistent induction of CXCR4 on synovial T cells by TGF-β1 leads to their active, SDF-1-mediated retention in a perivascular distribution within the rheumatoid synovium.

Rheumatoid arthritis (RA)4 is characterized by the persistent accumulation of T cells within the synovial compartment (1, 2). Synovial T cells are highly differentiated and are almost exclusively of the primed CD45RO+CD45RBdull phenotype (3). However there is little evidence for active T cell division in situ. Synovial T cells synthesize few cytokines at very low levels and respond poorly to stimulation in vitro (2). How these T cells persist within the rheumatoid synovium in the absence of proliferation has remained unclear.

The maintenance of an inflammatory infiltrate requires a distorted balance among leukocyte recruitment, retention, proliferation, and death. Rheumatoid synovial T cells are highly susceptible to apoptosis in vitro, which reflects their advanced state of differentiation (4). However, we have recently shown that T cell apoptosis is actively suppressed in the rheumatoid synovium by stromal derived factors such as IFN-β (5, 6). These observations led us to propose that failure of synovial T cell apoptosis coupled to enhanced recruitment leads to accumulation of these cells within the rheumatoid synovium (5). However the contribution of changes in the rate of T cell exit from the inflamed synovium to the persistence of the T cell infiltrate has not been addressed.

A striking feature of rheumatoid synovial T cells is their enhanced integrin function despite the fact that they do not appear to be activated, suggesting the presence of an integrin-activating microenvironment within the joint (7, 8). The molecular basis for this is obscure, but stromal-derived chemokines are attractive candidates, because they act as potent integrin-activating agents and play a key role in regulating the navigation and positioning of leukocyte subsets within tissue microenvironments (9, 10).

Chemokines are divided into two main functional groups (11, 12). Inducible, inflammatory chemokines are produced in peripheral tissue at sites of inflammation, where they function to recruit neutrophils, monocytes, immature dendritic cells, and activated T cells. In contrast, constitutive chemokines, produced in the bone marrow and secondary lymphoid organs, regulate leukocyte trafficking under noninflammatory, physiological conditions (11, 12). The spectrum of chemokine receptors expressed by different leukocyte subsets determines the anatomical site to which they are recruited. For example, naive T cells migrate almost exclusively between the peripheral circulation and secondary lymphoid organs using the constitutive chemokine receptor CCR7, whereas effector cells that lack CCR7 but express the inflammatory chemokines receptors CCR5 and CXCR3, migrate predominantly to inflammatory sites (13).

Synovial T cells express high levels of the inflammatory chemokine receptors CCR5 and CXCR3, which is consistent with their highly differentiated state (14). These receptors mark subsets of T cells with a propensity to migrate to inflammatory sites where their ligands are produced. In contrast, rheumatoid synovial T cells do not express the chemokine receptor CCR4, which is restricted to memory T cells at sites of cutaneous inflammation (15, 16).

In this study we have tested the hypothesis that the rheumatoid synovial microenvironment promotes T cell accumulation by an active chemokine-driven process. We found that rheumatoid synovial T cells express high, functionally relevant levels of the chemokine receptor CXCR4. This constitutive chemokine receptor was previously reported to be expressed predominantly on unprimed T cells and has been implicated in promoting precursor cell retention in fetal liver and bone marrow (17, 18, 19). We have identified TGF-β as an important synovial-derived factor involved in maintaining inappropriately high levels of CXCR4 on synovial T cells. In addition, synovial endothelial cells express SDF-1, the only known ligand for CXCR4, in close proximity to perivascular T cell aggregates within the rheumatoid synovium. These observations suggest that the active retention of T cells by an integrin-mediated, chemokine-driven mechanism directly contributes to their accumulation within the synovial compartment.

Cells were cultured in RPMI 1640 (Sigma-Aldrich, Irvine, U.K.) with 10% FCS supplemented with antibiotics as previously described (4, 5). Recombinant human IL-1β, IL-6, IL-10, IL-12, IL-15, TGF-β2, TGF-β3, stem cell factor, stromal cell-derived factor (SDF-1α), bone morphogenetic protein-4, macrophage inflammatory protein-1α, and macrophage chemotactic protein were purchased from R&D Systems (Abingdon, U.K.); IL-1α, IL-4, IL-8, IL-13, TNF-α, TGF-β1, basic fibroblast growth factor, vascular endothelial cell growth factor (VEGF), platelet-derived growth factor, SDF-1β, and epithelial neutrophil-activating peptide-78 were purchased from PeproTech (London, U.K.); IL-16 was obtained from Endogen (Woburn, MA); IFN-α and IFN-β were obtained from BioSource (Watford, U.K.); IFN-γ was purchased from Biogen (Bracknell, U.K.); PMA was obtained from Sigma; and IL-2 was purchased from Chiron (Harefield, U.K.). All cytokines were used at 10 ng/ml, except for IL-2 (25 U/ml), IFN-α (100 U/ml), TGF-β1 (2 ng/ml), TGF-β2 and TGF-β3 (1 ng/ml), stem cell factor and macrophage inflammatory protein-1α (5 ng/ml), macrophage chemotactic protein-1 (20 ng/ml), VEGF (25 ng/ml), SDF-1α and PMA (50 ng/ml), SDF-1β (80 ng/ml), and bone morphogenetic protein-4 (100 ng/ml). Fibronectin was obtained from Sigma-Aldrich and used at 100 μg/ml to coat parallel flow chambers. Recombinant ICAM-1 Fc fusion protein was used as previously described (20). The following primary mAbs were used for flow cytometry and immunohistochemistry: CD3-PE (UCHT1) IgG1 (Dako, Cambridge, U.K.), CD4-PE (Leu 3a) IgG1 (Becton Dickinson, Mountain View, CA), CD4-Bio (Leu 3a) IgG1 (Becton Dickinson), CD45RA-biotin (SN 130) IgG1 (Royal Free Hospital, London, U.K.), CD45RO-PE (UCHL1) IgG2a (Dako), CCR5 IgG2b, CXCR3 IgG1 (Serotech, Oxford, U.K.), CXCR4 (12G5) IgG2a (R&D Systems), and SDF-1 (K15C) IgG2a (21). Irrelevant conjugated and unconjugated mouse or rabbit Abs of each isotype were used to establish specificity of staining. Secondary Abs were obtained from Southern Biotechnology (Birmingham, AL).

Samples from peripheral venous blood and SF were collected into preservative-free heparin from patients who fulfilled the 1987 ACR criteria for rheumatoid arthritis (20). PBL and SF lymphocytes (SFL) were isolated as previously described (5). SF was immediately centrifuged to remove cells and debris before storage as aliquots at −70°C. Synovial tissue was taken from hip, knee, and elbow joints at the time of joint replacement with approval of the ethics committee. The isolation of synovial T cells from synovial tissue was performed as follows. Synovial tissue was diced using sterile blades into 1-mm3 pieces and incubated for 4–5 h in medium containing 0.2% collagenase (type 1A, Sigma). After enzymatic digestion, tissue was passed twice through a metal mesh filter (100 μm pore size) to remove cell clumps and undissociated tissue. Cells were washed three times in medium to remove collagenase before the adherence depletion of monocytes and fibroblasts by 2-h incubation on plastic. Matched samples of peripheral blood, synovial tissue, and SFL were studied for chemokine receptor expression. The digestion of tissue with collagenase had no effect on CXCR4 expression, judged by the inability of collagenase to affect CXCR4 expression on PBL and IL-2/PHA-activated T cells (data not shown). Synovial tissue (1 mm3) was snap-frozen in liquid nitrogen at the time of surgery and stored at −70°C.

CD4+CD45RA+ cells were enriched by negative selection with magnetic beads (Dynal, Oslo, Sweden) as described previously (4). To study the expression of CXCR4 on CD4 cells during differentiation from naive to primed cells CD4+CD45RA+ T cells were transferred to a 24-well flat-bottom plate (1 × 106 cells/well) and maintained in culture in the presence of IL-2. Cells were stimulated with PHA-H15 (Murex Biotech, Dartford, U.K.) at 10 μg/ml every 2 wk. To study the effect of the synovial microenvironment on CXCR4 expression, freshly isolated SF T cells were transferred to 24-well flat-bottom plates (1 × 106 cells/well) and maintained in culture in the presence of IFN-β. After 6–7 days autologous SF was added back at a 50% dilution. Cell surface expression of CXCR4 expression was measured on matched peripheral blood T cells at the time of SF aspiration. Primed, short-term CD4 lines used to study the effect of cytokines/chemokines on the surface expression of CXCR4 were cultured as previously described (5). Cytokines were added to the cells on day 0, and after 24 h a second equal dose was added to the cells. Surface expression of CXCR4 was calculated as (median fluorescence intensity (MFI)CXCR4 − MFIcontrol) of cytokine treated cells divided by (MFICXCR4 − MFIcontrol) of cells cultured in medium alone and expressed as a percentage of the control value. To examine the effect of rheumatoid SF on the expression of CXCR4, CD4 T cells were incubated overnight in medium alone or in SF at various dilutions).

Analysis of cell surface molecules was performed using single-, two-, or three-color immunofluorescence as previously described (5). For analysis of chemokine receptor expression on T cell lines, PBL, synovial tissue, and SFL, cells were stained with anti-CXCR4, CCR5, and CXCR3 mAbs followed by FITC-labeled secondary Abs (Southern Biotechnology). The samples were analyzed on a Coulter EPICS XL Flow cytometer (Hialeah, FL) according to standard procedures (5). Cytometer calibration was standardized using fluorospheres (Immunocheck and Standardbrite, Coulter).

The expressions of CXCR4, SDF-1, and CD3 were studied by double immunofluorescence on 4-μm cryostat sections. Briefly, sections were fixed in cold acetone for 10 min and then incubated with primary mAbs followed by relevant secondary Abs. To avoid any possibility of cross-reactivity we used primary Abs of different species (mouse and rabbit). T lymphocytes were detected with a rabbit anti-CD3 and FITC-labeled anti-rabbit IgG. For the expression of CXCR4 and SDF-1 a three-layer method was used. Color intensification was achieved with streptavidin-Texas Red. Except for the overnight incubation at 4°C of primary Abs, all incubations were conducted for 1 h at room temperature. Images were captured with a SPOT-2 (Diagnostic Instruments, MI) digital camera and Image-Pro software (Media Cybernetics, Silver Spring, MD). Serial sections were stained by Diff-Quik (Dade Behring, Marburg, Germany).

Controlled detachment assays were performed as previously described (20) with the following modifications. Five-microliter dots of fibronectin or ICAM-1-Fc recombinant fusion protein (20) (10 μg/ml) were spotted onto polystyrene plates that were assembled as a lower wall in a parallel flow chamber, blocked with 2% BSA in complete medium, and mounted on an inverted phase contrast microscope. Treated T cells were allowed to adhere to the coated fibronectin/ICAM-1 under static conditions for 5 min in the presence or the absence of SDF-1α (100 ng/ml) before washing with increasing shear flow. Cells remaining adherent were captured on videotape after each change in flow rate (every 1 min) and expressed as the percentage of cells remaining compared with the total before washing. Three separate regions were counted, and results were expressed as the mean and SD.

TGF-β1 was measured by ELISA following the manufacturer’s instructions (R&D Systems DuoSet). Synovial samples from patients with rheumatoid arthritis were acid activated to release total TGF-β following the manufacturer’s instructions. To deplete TGF-β1, -2, and -3 from SF, acid-activated synovial samples were mixed with protein G-Sepharose beads coated with anti-TGF-β1, -2, and -3 antiserum (monoclonal mouse TGF-β1, -2, and -3 and polyclonal rabbit pan-specific TGF-β, R&D clone 1D11 and Serotec, respectively) or as a control mouse anti-IL-8 (R&D clone 6217.111) and subjected to three rounds of bead depletion (30 min at 4°C).

Tissues undergoing chronic inflammatory reactions contain leukocyte infiltrates often organized into defined structures. To determine whether aberrant chemokine receptor expression might explain these lymphoid-like structures, we examined the distribution of chemokine receptors in tissue sections from patients with RA. We confirmed previous reports describing the distribution of CCR5 and CXCR3 on synovial T cells (data not shown) (14). However, we were surprised to find that large numbers of CD3-positive T cells, distributed around vessels and within lymphoid aggregates in the synovial membrane, expressed high levels of CXCR4 (Fig. 1, a and b). In addition, CXCR4 was expressed on other non-T cells in the synovium, including B cells, macrophages, and synoviocytes (data not shown). To confirm this finding, we isolated CD3-positive T cells from peripheral blood, synovial membrane, and SF of patients with RA and analyzed their CXCR4 expression by flow cytometry. Peripheral blood CD3 cells from patients with RA expressed low levels of CXCR4 (Fig. 1,c) in contrast to SF and synovial membrane T cells (Fig. 1, d and e). These findings suggested that T cells infiltrating the synovial compartment are enriched for the expression of CXCR4 compared with peripheral blood T cells.

Several reports suggest that CXCR4 is poorly expressed or even absent on CD45RO+-primed T cells (19, 23). Rheumatoid synovial T cells are almost exclusively of the CD45RO+ phenotype, so we were surprised to find high levels of CXCR4 on synovial T cells in situ. To clarify the relationship between CXCR4 expression and T cell differentiation, we examined the expression of CXCR4 on T cells undergoing differentiation from a CD45RA+ to a CD45RO+ state. CD4+CD45RA+ T cells isolated from peripheral blood were stimulated with PHA at 2-wk intervals, and CXCR4 expression was measured by flow cytometry (Fig. 2, a and b). These cells lost expression of CXCR4 within 48 h of initial stimulation, but then regained expression over the following 4–5 days. Intriguingly, after two rounds of stimulation the expression of CXCR4 declined, consistent with previous reports showing low levels of CXCR4 on long-term T cell clones (23). T cells within the synovial joint are highly differentiated CD45RObrightRBdull cells that have undergone multiple rounds of stimulation (3, 4, 5) yet express high levels of CXCR4. Cells of an identical phenotype in blood and in vitro culture do not express CXCR4. These results suggested that factors present within the synovial microenvironment might induce the high levels of CXCR4 observed on synovial T cells in situ.

To examine whether the rheumatoid synovial microenvironment contains factors that up-regulate the expression of CXCR4, we isolated SF T cells from patients with RA and measured the expression of CXCR4 on these cells over time in culture. Synovial T cells are fragile and undergo apoptosis during culture in vitro unless rescued by appropriate survival signals (5). We therefore cultured synovial T cells in the presence of IFN-β, which inhibits their apoptosis without inducing proliferation, mimicking the phenotype of synovial T cells in situ (5, 6). IFN-β does not affect CXCR4 expression (see Fig. 4). Over 6 days of culture, the level of CXCR4 on rheumatoid synovial T cells declined, approaching that observed in the peripheral blood from the same individual taken at the time of SF sampling. After 6 days, when autologous cell-free SF was added to the cells for 24 h, the level of CXCR4 doubled (Fig. 3 a), suggesting that rheumatoid SF contains a factor responsible for the induction of CXCR4 expression.

To test whether CXCR4 expressed on synovial T cells was functional, we compared the adhesion of SDF-1-stimulated SF and peripheral blood T cells to fibronectin, a prominent component of the extracellular matrix in the inflamed synovium. Increased integrin-mediated adhesion to matrix proteins is a characteristic outcome of chemokine receptor engagement. We used detachment assays in which increased cell adhesion, in response to exposure to chemokines, is measured after periods of increasing shear flow. This assay has proven to be a robust, functional measure of chemokine/chemokine receptor function (16, 20). As the level of CXCR4 on cultured synovial T cells declined, there was a concomitant reduction in SDF-1-triggered adhesion to fibronectin, which was restored in cells re-exposed to autologous SF (Fig. 3,b). Furthermore, freshly isolated synovial T cells, which express high levels of CXCR4, responded vigorously to SDF-1, with strong adhesion to fibronectin compared with peripheral blood T cells (Fig. 3 c). These results indicate that the synovial microenvironment is able to induce functional CXCR4 expression on synovial T cells, leading to increased SDF-1-triggered cell adhesion to fibronectin.

To identify potential factors within SF responsible for the induction of CXCR4 expression on synovial T cells, we screened a large panel of chemokines and cytokines for their effects on the induction of CXCR4 expression on short-term CD4 T cell lines (Fig. 4, a and b). As a control and to validate the culture system, we used IL-4, which is absent in SF, but is known to stimulate the expression of CXCR4 on T cells (24). We found that TGF-β isoforms induced a 10-fold increase in the level of CXCR4 expressed on CD4 T cells after a 48-h incubation. IL-2, IL-10, and IL-15 produced modest 2-fold increases in CXCR4 expression, with IL-4 inducing a 14-fold increase. In contrast, PMA stimulation and exposure to SDF-1 led to down-regulation of CXCR4, as previously described (25). Combinations of TGF-β1, -2, and -3 were not additive or synergistic (data not shown)

Having established that TGF-β1 could induce the expression of CXCR4 on CD4 T cells we next examined whether the CXCR4 was functionally active by assessing the ability of these cells to mediate SDF-1 triggered adhesion to ICAM-1 (Fig. 4,c) and fibronectin (Fig. 4 d). Only TGF-β1-stimulated cells expressing high levels of CXCR4 were able to interact with fibronectin and ICAM-1 when treated with SDF-1. Untreated T cells adhered weakly in the presence of SDF-1, reflecting the low level of CXCR4 expression on these cells. In the absence of SDF-1 there was no binding of T cells to either fibronectin or ICAM-1. Together these results demonstrate that TGF-β can induce functional CXCR4 expression on CD4 T cells in vitro, because stimulation of these cells with SDF-1 leads to increased cell attachment to fibronectin and ICAM-1.

Previous reports have shown that the level of TGF-β1 in SF is high, in the range 2–20 ng/ml (26, 27). To determine whether TGF-β1 present within SF was responsible for the elevated CXCR4 expression in vivo, we cultured CD4+ T cell lines in the presence of dilutions of SF taken from six different patients with RA. TGF-β1 levels in these SF samples were assayed by ELISA and compared with their ability to induce CXCR4 expression after 18 h (Fig. 5 a). There was a good correlation between the presence of TGF-β1 in the SF and its ability to induce CXCR4 expression. However, the levels of CXCR4 induced on T cells by SF consistently underestimated the actual levels of TGF-β1 compared with the standard curve for CXCR4 induction, generated using purified TGF-β1. This is probably due to the presence of latent forms of TGF-β1 that are detected by ELISA but remain biologically inactive unless liberated by acid activation.

To test whether synovial TGF-β is responsible for CXCR4 induction in vivo, we depleted TGF-β isoforms from rheumatoid SF before addition to CD4+ T cells. To ensure that all potential latent forms of TGF-β were activated, we exposed SF to acid activation before depletion of TGF-β1, -2, and -3. The results in Fig. 5,b show that removal of TGF-β1, -2, and -3 from SF in four patients with RA led to a significant reduction in CXCR4 expression. Furthermore, the reduction in CXCR4 expression paralleled the depletion in TGF-β1 from SF (Fig. 5 b). These results strongly imply that TGF-β1 is responsible for the high expression of CXCR4 found on synovial T cells in vivo.

Both synovial T cells and TGF-β-stimulated CD4 T cells express functionally active CXCR4 (Figs. 3 and 4). SDF-1 mRNA has previously been reported to be abundant in rheumatoid synovial tissue (28). It was therefore of interest to study the distribution of the SDF-1 within rheumatoid synovium to determine whether its distribution might explain the perivascular localization of CXCR4-expressing CD4 T cells in the rheumatoid synovium. Using an anti-SDF-1 mAb, we found that SDF-1 was predominantly expressed by endothelial venules within synovial tissue, although there was sparse staining of synovial tissue in a perivascular distribution (Fig. 6). SDF-1 expression correlated with the distribution of CD4+CXCR4+-positive T cells in perivascular lymphoid aggregates (Fig. 6 b), suggesting that SDF-1 is well positioned to activate the adhesion and therefore lead to the retention of CXCR4-bearing T cells around blood vessels within the rheumatoid synovium.

The mechanisms responsible for the accumulation and persistence of T cell infiltrates within the rheumatoid synovium are poorly understood. In this study we set out to explore whether the aberrant expression of chemokine receptors on synovial T cells contributes to their persistence in chronic rheumatoid synovitis. Rheumatoid synovial T cells are highly differentiated primed cells that express high levels of β1 and β2 integrins as well as the inflammatory chemokine receptors CXCR3 and CCR5 (5, 7, 8, 14). Surprisingly, they also express high levels of the constitutive chemokine receptor CXCR4, whose expression has previously been thought to be confined to unprimed CD45RA+ T cells (19, 23). Removal of T cells from the synovial microenvironment led to loss of CXCR4 expression, which was regained after exposure to autologous SF, suggesting that the synovial microenvironment directly modifies the character of infiltrating T cells in the joint.

CXCR4 is expressed on a wide variety of cells and plays an active role in hemopoiesis, angiogenesis, embryonic development, and infection by viruses such as HIV-1 (18). CXCR4 expression is up-regulated on a number of cell types by cytokines, including IL-2, IL-4 (24), IL-6 and stem cell factor (29), VEGF, basic fibroblast growth factor (30), and TGF-β (23). In addition, IFN-γ, PMA, and SDF-1 have been shown to down-regulate CXCR4 expression (12, 23, 25). We found that TGF-β1, -2, and -3 induce CXCR4 expression on cultured peripheral blood CD4 T cells. While IL-2 and IL-15 are also capable of inducing CXCR4 expression, they do so to a lesser extent. Therefore, a diverse set of cytokines, signaling through different pathways, is able to regulate CXCR4 expression in a variety of cell types.

Changes in the level of chemokine receptor expression are not the only method by which cells can modulate their responses to chemokines. The maturation of B cells decreases their responsiveness to SDF-1 despite mature B cells expressing higher levels of CXCR4 than pre-B cells (31). It was therefore important for us to test whether the high levels of CXCR4 observed on synovial T cells were functionally relevant. We found that rheumatoid synovial T cells adhere better to fibronectin in response to SDF-1 than peripheral blood T cells. Furthermore, the removal of T cells from the synovial compartment led to a decrease in their response to SDF-1 in parallel with a decrease in expression of CXCR4. Re-exposure of these T cells to autologous SF caused a rapid re-expression of CXCR4 and a corresponding increase in adhesion to fibronectin in response to SDF-1. These results suggest that the interaction of SDF-1 with CXCR4 contributes to the proadhesive phenotype of synovial T cells (7, 8) and consequently plays a role in regulating T cell accumulation within the synovial microenvironment.

We show that synovial-derived TGF-β is capable of inducing the expression of CXCR4 on T cells, and that the level of TGF-β1 present within SF directly correlates with its ability to induce CXCR4 expression. Moreover, the depletion of TGF-β from SF leads to a significant reduction in the ability of SF to induce CXCR4 expression, suggesting that TGF-β plays a key role in regulating the levels of CXCR4 on synovial T cells in vivo.

The TGF-β family of proteins, is an expanding group of secreted signaling molecules with unique and potent immunoregulatory properties (32). In vitro, TGF-β isoforms exert nearly identical effects that can be grouped into three broad areas: modulation of inflammatory cell function, growth inhibition and differentiation, and control of extracellular matrix production (32). Studies in animals and humans strongly suggest that TGF-β is important in the pathogenesis of several diseases, including RA (32, 33). TGF-β1 and -2 are expressed in rheumatoid synovial tissue and fluid and have been proposed to account for most of the immunosuppressive activity of SF in lymphocyte proliferation assays (26, 27). Injection of TGF-β1 into the joints of normal rats or into joints in murine models of established arthritis induced marked synovitis, but prevented cartilage damage, suggesting that whereas TGF-β can stimulate bone repair, it appears to actively contribute to synovitis (33, 34).

Recent studies have shown that SDF-1 is expressed by human bone marrow endothelium, biliary duct epithelial cells, and skin endothelium and may play a role in lymphocyte accumulation at these sites (35, 36). We show here that SDF-1 is expressed on synovial endothelial cells, suggesting that the persistent and inappropriate induction of CXCR4 by stromal-derived factors such as TGF-β leads to the active SDF-1-driven retention of CD4 T cells in a perivascular distribution within the rheumatoid joint.

The current view on how differential lymphocyte accumulation within tissues is regulated proposes that selectivity is determined during leukocyte recruitment (9). However, our observations suggest that T cell accumulation and persistence within the synovial compartment may also occur because T cells are actively prevented from emigrating as well as dying (5) in an abnormal inflammatory microenvironment. Intraepithelial lymphocytes within the gut provide a striking precedent for the active retention of T cells in tissue microenvironments. Here the local production of TGF-β induces the expression of the integrin αEβ7 at the expense of α4β7, leading to T cell adhesion to E-cadherin expressed by gut epithelial cells (37). This stromal, rather than endothelial, cell “area code” leads to physiological and appropriate T cell retention. However, in the chronically inflamed joint our findings suggest that TGF-β and SDF-1 contribute to the inappropriate retention of T cells within the synovial microenvironment. Recent studies have shown that synovial T cells also express αEβ7, reflecting the TGF-β-rich mircoenvironment of the joint (38). Thus, inflammation in RA appears to persist as a direct result of the sustained recruitment, inappropriate retention, and active survival of highly differentiated T cells mediated by stromal-derived factors associated with the inflamed synovial joint itself. Therefore, targeting the synovial stromal microenvironment represents an important alternative therapeutic strategy in chronically inflamed joints.

We thank our clinical colleagues, including Andrew Thomas (Royal Orthopaedic Hospital, Birmingham, U.K.) for providing clinical material, and Dr. G. B. Nash (Department of Physiology, University of Birmingham) for help with the controlled detachment assays.

1

This work was supported by the Wellcome Trust, the Medical Research Council, the Arthritis Research Campaign, and the Nuffield Foundation.

4

Abbreviations used in this paper: RA, rheumatoid arthritis; MFI, median fluorescence intensity; SDF-1, stromal-derived factor-1; VEGF, vascular endothelial cell growth factor; SF, synovial fluid; SFL, SF lymphocytes.

1
Buckley, C. D..
1997
. Science medicine and the future: treatment of rheumatoid arthritis.
Br. Med. J.
315
:
236
2
Salmon, M., J. S. H. Gaston.
1995
. The role of T-lymphocytes in rheumatoid arthritis.
Br. Med. Bull.
51
:
332
3
Matthews, N., P. Emery, D. Pilling, A. Akbar, M. Salmon.
1993
. Subpopulations of primed T helper cells in rheumatoid arthritis.
Arthritis Rheum.
36
:
603
4
Salmon, M., D. Pilling, N. J. Borthwick, N. Viner, G. Janossy, P. A. Bacon, A. N. Akbar.
1994
. The progressive differentiation of primed T cells is associated with an increasing susceptibility to apoptosis.
Eur. J. Immunol.
24
:
892
5
Salmon, M., D. Scheel-Toellner, A. P. Huissoon, D. Pilling, N. Shamsadeen, H. Hyde, A. Dupuy D’Angeac, P. A. Bacon, P. Emery, A. N. Akbar.
1997
. Inhibition of T cell apoptosis in the rheumatoid synovium.
J. Clin. Invest.
99
:
439
6
Pilling, D., A. N. Akbar, J. Girdlestone, C. H. Orteu, N. J. Borthwick, N. Amft, D. Scheel-Toellner, C. D. Buckley, M. Salmon.
1999
. Interferon-β mediates stromal cell rescue of T cells from apoptosis.
Eur. J. Immunol.
29
:
1041
7
Arroyo, A. G., R. Garcia-Vicuna, M. Marazuela, T. A. Yednock, R. Gonzalez-Amaro, F. Sanchez-Madrid.
1995
. Expression and functional significance of an activation-dependent epitope of the β1 integrins in chronic inflammatory diseases.
Eur. J. Immunol.
25
:
1720
8
Yokota, A., N. Murata, O. Saiki, M. Shimizu, T. A. Springer, T. Kishimoto.
1995
. High avidity state of leukocyte function-associated antigen-1 on rheumatoid synovial fluid T lymphocytes.
J. Immunol.
155
:
4118
9
Butcher, E. C., L. J. Picker.
1996
. Lymphocyte homing and homeostasis.
Science
272
:
60
10
Foxman, E. F., J. J. Campbell, E. C. Butcher.
1997
. Multistep navigation and the combinatorial control of leukocyte chemotaxis.
J. Cell Biol.
139
:
1349
11
Baggiolini, M..
1998
. Chemokines and leukocyte traffic.
Nature
392
:
565
12
Sallusto, F., A. Lanzavecchia, C. R. Mackay.
1998
. Chemokines and chemokine receptors in T-cell priming and Th1/Th2-mediated responses.
Immunol. Today
19
:
568
13
Cyster, J. G.
1999
. Chemokines and cell migration in secondary lymphoid organs.
Science
286
:
2098
14
Qin, S., J. B. Rottman, P. Myers, N. Kassam, M. Weinblatt, M. Loetscher, A. E. Koch, B. Moser, C. R. Mackay.
1998
. The chemokine receptors CXCR3 and CCR5 mark subsets of T cells associated with certain inflammatory reactions.
J. Clin. Invest.
101
:
746
15
Suzuki, N., A. Nakajima, S. Yoshino, K. Matsushima, Y. Hideo, K. Okumura.
1998
. Selective accumulation of CCR5+ T lymphocytes into inflamed joints of rheumatoid arthritis.
Int. Immunol.
11
:
553
16
Campbell, J. J., G. Haraldsen, J. Pan, J. Rottman, S. Qin, P. Ponath, D. P. Andrew, R. Warnke, N. Ruffing, N. Kassam, et al
1999
. The chemokine receptor CCR4 in vascular recognition by cutaneous but not intestinal memory T cells.
Nature
400
:
776
17
Ma, Q., D. Jones, T. A. Springer.
1999
. The chemokine receptor CXCR4 is required for the retention of B lineage and granulocytic precursors within the bone marrow environment.
Immunity
10
:
463
18
Nagasawa, T., K. Tachibana, T. Kishimoto.
1998
. A novel CXC chemokine PBSF/SDF-1 and its receptor CXCR4: their functions in development, haematopoiesis and HIV infection.
Semin. Immunol.
10
:
179
19
Bleul, C. C., L. Wu, J. A. Hoxie, T. A. Springer, C. R. Mackay.
1997
. The HIV coreceptors CXCR4 and CCR5 are differentially expressed and regulated on human T lymphocytes.
Proc. Natl. Acad. Sci. USA
94
:
1925
20
Gunn, M. D., K. Tangemann, C. Tam, J. G. Cyster, S. D. Rosen, L. T. Williams.
1998
. A chemokine expressed in lymphoid high endothelial venules promotes the adhesion and chemotaxis of naïve T lymphocytes.
Proc. Natl. Acad. Sci. USA
95
:
258
21
Coloumb-L’Hermine, A., A. Amara, C. Schiff, I. Durand-Gasselin, A. Foussat, T. Delaunay, G. Chaouat, F. Capron, N. Ledee, P. Galanaud, F. Arenzana-Seisdedos, et al
1999
. Stromal cell-derived factor 1 (SDF-1) and antenatal human B cell lymphopoiesis: expression of SDF-1 by mesothelial cells and biliary ductal plate epithelial cells.
Proc. Natl. Acad. Sci. USA
96
:
8585
22
Arnett, F. C., S. M. Edworthy, D. A. Bloch, D. J. McShane, J. F. Fries, N. S. Cooper, L. A. Healey, S. R. Kaplan, M. H. Laing, H. S. Luthra.
1988
. The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis.
Arthritis Rheum.
31
:
315
23
Sallusto, F., D. Lenig, C. R. Mackay, A. Lanzavecchia.
1998
. Flexible programs of chemokine receptor expression on human polarized T helper 1 and 2 lymphocytes.
J. Exp. Med.
187
:
875
24
Jourdon, P., C. Abbal, N. Nora, T. Hori, T. Uchiyama, J.-P Vendrell, J. Bousquet, N. Taylor, J. Pene, H. Yssel.
1998
. IL-4 induces functional cell surface expression of CXCR4 on human T cells.
J. Immunol.
160
:
4153
25
Amara, A., S. L. Gall, O. Schwartz, J. Salamero, M. Montes, P. Loetscher, M. Baggiolini, J. L Virelizier, F. Arenzana-Seisdos.
1997
. HIV co-receptor downregulation as antiviral principle: SDF-1α-dependent internalization of the chemokine receptor CXCR4 contributes to the inhibition of HIV replication.
J. Exp. Med.
186
:
139
26
Lotz, M., J. Kekow, D. A. Carson.
1990
. Transforming growth factor-β and cellular immune responses in synovial fluids.
J. Immunol.
144
:
4189
27
Fava, R., N. Olsen, J. Keski-Oja, H. Moses, T. Pincus.
1989
. Active and latent forms of transforming growth factor β activity in synovial effusions.
J. Exp. Med.
169
:
291
28
Seki, T., J. Selby, T. Haupl, R. Winchester.
1998
. Use of a differential subtraction method to identify genes that characterize the phenotype of cultured rheumatoid arthritis synoviocytes.
Arthritis Rheum.
41
:
1356
29
Peled, A., I. Petit, O. Kollet, M. Magid, T. Ponomaryov, T. Byk, A. Nagler, H. Ben-Hur, A. Many, L. Shultz, et al
1999
. Dependence of human stem cell engraftment and repopulation of NOD/SCID mice on CXCR4.
Science
283
:
845
30
Salcedo, R., K. Wasserman, H. A. Young, M. C. Grimm. O. M. Howard, M. R. Anver, H. K. Kleinman, W. J. Murphy, J. J. Oppenheim.
1999
. Vascular endothelial growth factor and basic fibroblast growth factor induce expression of CXCR4 on human endothelial cells: in vivo neovascularization induced by stromal-derived factor-1α.
Am. J. Pathol.
154
:
1125
31
Honczarenko, M., R. S. Douglas, C. Mathias, B. Lee, M. Z. Ratajczak, L. E Silberstein.
1999
. SDF-1 responsiveness does not correlate with CXCR4 expression levels of developing human bone marrow B cells.
Blood
94
:
2990
32
Letterio, J. J., A. B Roberts.
1998
. Regulation of immune responses by TGF-β.
Annu. Rev. Immunol.
16
:
137
33
van den Berg, W. B..
1998
. Joint inflammation and cartilage destruction may occur uncoupled.
Springer Semin. Immunopathol.
20
:
149
34
Allen, J. B., C. L. Manthey, A. R. Hand, K. Ohura, L. Ellingsworth, S. M. Wahl.
1990
. Rapid onset synovial inflammation and hyperplasia induced by transforming growth factor β.
J. Exp. Med.
171
:
231
35
Peled, A., V. Grabovsky, I. Habler, J. Sandbank, F. Arenzana-Seisdedos, I. Petit, H. Ben-Hur, T. Lapidot, R. Alon.
1999
. The chemokine SDF-1 stimulates integrin-mediated arrest of CD34+ cells on vascular endothelium under flow.
J. Clin. Invest.
104
:
1199
36
Pablos, J. L., A. Amara, A. Bouloc, B. Santiago, A. Caruz, M. Galindo, T. Delaunnay, J. L. Virelizer, F. Arenzana-Seisdedos.
1999
. Stromal-cell derived factor is expressed by dendritic cells and endothelium in human skin.
Am. J. Pathol.
155
:
1577
37
Kilshaw, P. J., S. J. Murant.
1991
. Expression and regulation of β7 integrins on mouse lymphocytes: relevance to the mucosal immune system.
Eur. J. Immunol.
21
:
2591
38
Trollmo, C., I. M. Nilsson, C. Sollerman, A. Tarkowski.
1996
. Expression of the mucosal lymphocyte integrin αEβ7 and its ligand E-cadherin in the synovium of patients with rheumatoid arthritis.
Scand. J. Immunol.
44
:
293