Fractalkine mediates T celldependent proliferation of synovial fibroblasts in rheumatoid arthritis.код для вставкиСкачать
ARTHRITIS & RHEUMATISM Vol. 56, No. 10, October 2007, pp 3215–3225 DOI 10.1002/art.22919 © 2007, American College of Rheumatology Fractalkine Mediates T Cell–Dependent Proliferation of Synovial Fibroblasts in Rheumatoid Arthritis Hirokazu Sawai, Yong W. Park, Xiaowen He, Jörg J. Goronzy, and Cornelia M. Weyand ation. TNF␣ amplified the expansion of FLS by enhancing their expression of CX3CR1 and FKN. Conclusion. FKN–CX3CR1 receptor–ligand interactions regulate FLS growth and FLS-dependent T cell function. FLS stimulate autocrine growth by releasing FKN and triggering the activity of their own CX3CR1. This growth-promotion loop is amplified by TNF␣ produced by CX3CR1-expressing T cells upon stimulation by FKN-expressing FLS. These data assign a critical role to FKN and its receptor in fibroblast proliferation and pannus formation in RA. Objective. In rheumatoid arthritis (RA), synovial fibroblasts proliferate excessively, eventually eroding bone and cartilage. The aim of this study was to examine the mechanisms through which CD4 T cells, the dominant lymphocyte population in patients with rheumatoid synovitis, regulate synoviocyte proliferation. Methods. Fibroblast-like synoviocyte (FLS) lines were established from rheumatoid synovium. CD4 T cells from patients with RA and age-matched control subjects were cultured on FLS monolayers. FLS proliferation was quantified by cytometry, using carboxyfluorescein succinimidyl ester staining or microscopic enumeration of PKH26-stained FLS. Surface expression of the fractalkine (FKN) receptor CX3CR1 was monitored by fluorescence-activated cell sorting. The induction of CX3CR1 and its ligand FKN in FLS was quantified by real-time polymerase chain reaction. Results. The proliferation of FLS was significantly increased in the presence of CD4 T cells from patients with RA compared with control T cells. CD4ⴙ,CD28– T cells were particularly effective in supporting FLS growth, inducing a 25-fold expansion compared with a 5-fold expansion induced by CD4ⴙ,CD28ⴙ T cells. The growth-promoting activity of CD4ⴙ,CD28– T cells was mediated through CX3CR1, a chemokine receptor expressed on both T cells and FLS. AntiCX3CR1 antibodies inhibited T cell production of tumor necrosis factor ␣ (TNF␣) and suppressed FLS prolifer- Hyperplasia of the synovial layer is a principal disease mechanism in rheumatoid arthritis (RA). The expansion of synovial tissue generates pannus, a destructive tumorlike structure that penetrates into the cartilage and subchondral bone, leading to erosion of these structures. In the rheumatoid joint, the hyperplastic membrane is composed of a specialized type of fibroblasts, fibroblast-like synoviocytes (FLS), which grow in an anchorage-independent manner and are resistant to apoptosis (1). FLS from the rheumatoid synovium have long been recognized as a source of proinflammatory cytokines and proteases, functioning as an amplifier of inflammation and directly contributing to tissue damage (2,3). FLS derived from patients with RA were shown to attach to and invade normal human cartilage in a SCID mouse model (4). It has been suggested that the invasive behavior of FLS from patients with RA correlates with the rate of joint destruction as the disease progresses (5). FLS have also been implicated in regulating the fate of tissue-invasive lymphocytes, placing them in a critical position in the rheumatoid disease process. Specifically, FLS have been described to provide signals to T cells, altering their susceptibility to apoptosis and determining their survival and migration pattern in the inflamed lesions (6–8). Mutuality in the relationship between FLS and T cells has been suggested by the demonstration that activated T cells determine the functional profile of FLS in a contact-dependent manner (9,10). Supported in part by the NIH (grants R01-AR-42527, R01AI-44142, R01-AR-41974, and R01-AI-57266). Hirokazu Sawai, MD, Yong W. Park, MD, Xiaowen He, PhD, Jörg J. Goronzy, MD, PhD, Cornelia M. Weyand, MD, PhD: Emory University School of Medicine, Atlanta, Georgia. Drs. Sawai and Park contributed equally to this work. Address correspondence and reprint requests to Cornelia M. Weyand, MD, PhD, Lowance Center for Human Immunology, Emory University School of Medicine, Room 1003 Woodruff Memorial Research Building, 101 Woodruff Circle, Atlanta, GA 30322. E-mail: firstname.lastname@example.org. Submitted for publication December 12, 2006; accepted in revised form June 22, 2007. 3215 3216 CD4⫹ helper T cells are recognized as central players in the pathogenesis of RA (11). Not only are CD4 T cells the most prominent cell population in synovial infiltrates, they also contribute to several disease pathways, such as the process of lymphoid neogenesis. Evidence has accumulated that dysfunctional CD4 T cells in patients with RA are not restricted to the joint. Rather, the entire T cell population is abnormal. Specifically, CD4⫹ T cells from patients with RA display phenotypic and functional changes indicative of premature senescence (12,13). CD4⫹ T cells from patients with RA have shortened telomeres, suggesting an intense proliferative history. The senescence program in these cells is associated with loss of the costimulatory molecule CD28 and de novo expression of numerous immunoregulatory receptors, including molecules of the killer cell immunoglobulin-like receptor (KIR) family (14) and NKG2D (15). CD4⫹,CD28– T cells become resistant to apoptosis (16) and overexpress the Th1 cytokines interferon-␥ (IFN␥) and tumor necrosis factor ␣ (TNF␣). Also, CD4⫹,CD28– T cells acquire cytolytic activity and lyse target cells with efficiency similar to that of professional killer cells (17). With the shift in their receptor profile, CD4⫹,CD28– T cells are obviously regulated by a new set of environmental signals. Ligands for KIR and NKG2D are HLA class I molecules. Thus, CD4⫹,CD28– T cells no longer depend on professional antigen-presenting cells for activation but rather become responsive to a broad spectrum of cellular partners in inflamed lesions. This principle is exemplified by recent reports that CD4⫹,CD28– T cells in patients with RA express CX3CR1 (18), a receptor binding the chemokine fractalkine (FKN) and usually restricted to cytotoxic CD8 T cells and natural killer cells. Synovial CD4 T cells use CX3CR1 to costimulate T cell receptor–derived signals and respond to cell-bound FKN with the enhanced release of cytokines. CX3CR1 is also expressed on synovial cells other than T cells, such as macrophages, dendritic cells, and synovial fibroblasts (18). The synovial microenvironment is rich in FKN (19), an unusual chemokine that exists in both a membrane-integrated and a soluble form. Its chemokine domain is displayed on a long negatively charged mucin-rich stalk extending from the cell surface. When cleaved from the cell membrane, FKN yields a soluble form (sFKN) (19,20). In RA synovium, the major source of FKN is FLS (18), but synovial macrophages and endothelial cells may also produce this chemokine. It is currently believed that FKN mediates all of its biologic actions by binding CX3CR1. Soluble FKN acts as a chemoattractant; the SAWAI ET AL membrane-bound form functions as an adhesion molecule (21). It has been proposed that FKN promotes recruitment of monocytes and T cells into the rheumatoid synovium (22–24). Recent work by our group suggests that in RA, the role of FKN extends beyond chemoattraction and adhesion (18). FKN is abundantly expressed on cultured synovial fibroblasts and on hyperplastic synoviocytes in rheumatoid tissue. Among CD4⫹ T cells, only CD28⫺ T cells express the FKN receptor CX3CR1. Such T cells strongly adhere to FLS in a FKN/CX3CR1-dependent manner. More importantly, FKN displayed on the surface of FLS cooperates with T cell–activating signals and amplifies T cell proliferation, IFN␥ production, and cytoplasmic granule expulsion. In the present study, we investigated how CD4⫹,CD28– T cells affect the proliferation of rheumatoid FLS. We found that interactions between CX3CR1 and FKN not only regulate T cell function but, equally importantly, affect fundamental functions of synovial fibroblasts. Specifically, with CX3CR1 expressed on FLS and FKN secreted by FLS, the CX3CR1–FKN receptor–ligand interaction functions as an autocrine growth-promotion loop enhancing proliferative expansion of these fibroblasts. Expression of the receptor and the ligand on FLS is regulated through the T cell proinflammatory cytokine TNF␣, placing this cytokine at a critical checkpoint for controlling synovial hyperplasia. TNF␣ is provided from CD4⫹,CD28– T cells that receive amplifying signals from FKNexpressing FLS, triggering the CX3CR1 receptor on T cells. These data assign a dual regulatory role to FKN and its receptor CX3CR1 in controlling synoviocyte proliferation and T cell function in the synovial microenvironment. PATIENTS AND METHODS Sample collection. The study cohort included 22 patients with RA (73% women, mean ⫾ SD age 51 ⫾ 15 years, mean ⫾ SD disease duration 6.5 ⫾ 7 years) who met the American College of Rheumatology (formerly, the American Rheumatism Association) revised criteria for the classification of RA (25). All patients were rheumatoid factor positive, 96% had bony erosions and received disease-modifying therapy, and 32% had extraarticular manifestations. Individuals with no family history of a chronic inflammatory disease served as controls. All donors provided informed consent, and biologic specimens were handled according to institutional review board–approved protocols. Cells and cell lines. CD4⫹ T cells were isolated from the blood of patients and normal control subjects, using a CD4⫹ T cell–enrichment cocktail (StemCell Technologies, T CELL–MEDIATED FLS PROLIFERATION Vancouver, British Columbia, Canada). CD4⫹,CD28⫹ or CD4⫹,CD28⫺ T cells were obtained by further incubating the cells with biotinylated mouse anti-human CD28 monoclonal antibody (ID Labs, London, Ontario, Canada) and streptavidin-conjugated magnetic beads (Miltenyi Biotec, Sunnyvale, CA), and separating the CD28⫹ and CD28⫺ subpopulations using magnetic separation columns (Miltenyi Biotec) (18). Primary cultures of FLS were established using synovial tissue specimens obtained from patients with RA (18). Fresh tissue was digested with 500 g/ml collagenase (Sigma, St. Louis, MO), and single-cell suspensions were cultured in Dulbecco’s modified Eagle’s medium (DMEM; BioWhittaker, Walkersville, MD) supplemented with 10% fetal calf serum (FCS; Summit Biotech, Fort Collins, CO), 2 mM L-glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin sulfate (Life Technologies, Grand Island, NY). Nonadherent cells were washed off after 3 days, and plastic-adherent cells were collected using trypsin–EDTA (Sigma). FLS from the third to eighth passages were used. FLS coculture and proliferation assays. To examine T cell–induced FLS proliferation, 2 ⫻ 103 FLS stained with PKH26 (Sigma) were cocultured with 2 ⫻ 103 T cells in 200 l DMEM (1% FCS) on chamber slides (Nalge Nunc International, Rochester, NY). After 3–5 days, FLS proliferation was evaluated by counting the number of fluorescent FLS per visual field. Pilot experiments were performed to determine the optimal time points when FLS numbers had increased without fading of the stain. Alternatively, FLS were stained with carboxyfluorescein succinimidyl ester (CFSE) at 1 g/ml and cocultured with T cells in 24-well plates. T cells were harvested after 5–7 days to allow for sufficient cell divisions to assess CFSE dilution by flow cytometry. The proliferation index was calculated as the ratio of the peak fluorescence intensities of nonproliferating and proliferating cells. To examine the effects of growth factors on FLS proliferation, 3 ⫻ 103 FLS were labeled with CFSE or PKH26 and cultured in DMEM (1% FCS) in the presence of sFKN (R&D Systems, Minneapolis, MN) or TNF␣, at the concentrations indicated. After 72 hours, the proliferation of FLS was evaluated by counting labeled FLS, using fluorescence microscopy. To examine the role of FKN–CX3CR1 interactions, anti-CX3CR1 antibodies (Torrey Pines Biolab, La Jolla, CA) or control immunoglobulin was added to the cultures at a concentration of 10 g/ml. In some experiments, either FLS or T cells were preincubated with anti-CX3CR1 antibodies, and excess antibodies were removed before the culture setup. Analysis of CX3CR1, FKN, and TNF␣ expression. FLS (0.5 ⫻ 106) were cocultured with 0.5 ⫻ 106 T cells in a 24-well plate for 96 hours. Cells were collected and incubated with 1% purified rabbit anti-human CX3CR1 (Torrey Pines Biolab) for 30 minutes at 4°C. After washing, cells were stained with fluorescein isothiocyanate–linked anti-rabbit IgG for 30 minutes and analyzed by flow cytometry. To evaluate TNF␣ gene expression induced by FLS and T cell interaction, 5 ⫻ 105 FLS and 5 ⫻ 105 CD4⫹,CD28⫹ or CD4⫹,CD28⫺ T cells were cocultured for 24 hours. TNF␣-induced FKN and CX3CR1 gene expression was assessed by culturing 5 ⫻ 105 FLS in 6-well plates in the presence of 5 ng/ml TNF␣ (R&D Systems). Transcription of TNF␣, FKN, and CX3CR1 was quantified by real-time polymerase chain reaction, utilizing the Mx3000 system (Stratagene, La Jolla, CA). The following 3217 primer sequences were used: for TNF␣, 5⬘-CTTTGGGATCATTGCCCTGTG-3⬘ and 3⬘-TCGTTGTTCTGGTGGTGAAGC-5⬘; for FKN, 5⬘-ATGGACGAGTCTGTGGTCCTGGAG-3⬘ and 3⬘-AGGCAATCGGAAAAGGTCCG-5⬘; for CX3CR1, 5⬘-GGACATCGTGGTCTTTGGGACTGT-3⬘ and 3⬘-CTTGGGCTTCTTGCTGTTGGTGAG-5⬘; for ␤-actin, 5⬘GTCCTCTCCCAAGTCCACACA-3⬘ and 3⬘-CTGGTCTCAAGTCAGTGTACAGGTAA-5⬘. Results are expressed as the number of specific transcript copies per 200,000 ␤-actin copies. Statistical analysis. For each culture condition, the number of FLS in 10 visual fields was counted, and the Mann-Whitney rank sum test was used to compare absolute numbers of proliferating FLS. FLS proliferation indices and TNF␣, FKN, and CX3CR1 transcripts were compared by paired t-test, using SigmaStat software (SPSS, Chicago, IL). RESULTS Enhanced FLS proliferation by CD4ⴙ T cells from patients with RA. To examine whether CD4 T cells regulate FLS proliferation, we cocultured PKH26labeled FLS lines established from rheumatoid synovial tissue with CD4⫹ T cells isolated from the blood of patients with RA. T cells from healthy age-matched individuals served as controls. FLS growth was assessed by fluorescence microscopy. To reduce the spontaneous proliferation of FLS caused by FCS, the serum concentration in the culture medium was minimized to 1%. CD4⫹ T cells from both healthy individuals and patients with RA had a marked growth-promoting effect on rheumatoid FLS. Within 5 days, FLS numbers more than doubled in the presence of T cells. CD4 T cells derived from patients with RA outperformed those from control subjects in enhancing FLS outgrowth. As shown in Figure 1, the number of FLS was ⬎50% higher in the presence of RA-derived CD4 T cells compared with control T cells (P ⫽ 0.0017). Thus, the spectrum of CD4 effector functions includes regulating fibroblast growth. In particular, CD4 T cells from patients with RA provided a significant growth advantage compared with synoviocytes. Particular efficiency of CD4ⴙ,CD28ⴚ T cells in promoting FLS growth. Our group previously reported that the T cell pool from patients with RA is enriched for presenescent and senescent CD4 T cells that have lost expression of CD28 and gained alternative immunoregulatory receptors (26,27). CD4⫹,CD28⫺ T cells home to both lymph nodes and CCL5-producing synovial lesions (28) and utilize multiple nonconventional receptors to costimulate T cell receptor–mediated signals (11,18). To address the question of whether CD4⫹,CD28⫹ and CD4⫹,CD28⫺ T cells differ in terms of their interaction with FLS and differentially affect FLS proliferation, both T cell subpopulations were isolated from the blood 3218 Figure 1. Superiority of CD4⫹,CD28⫺ T cells in promoting fibroblast-like synoviocyte (FLS) proliferation. A, PKH26-labeled FLS were grown in the absence or presence of CD4⫹ T cells from a patient with rheumatoid arthritis (RA) for 5 days. Fibroblast numbers were assessed by fluorescence microscopy. B, The number of FLS per low-power field (lpf) was obtained by counting 10 random lpfs of triplicate cultures. Data are shown as box plots, where each box represents the 25th to 75th percentiles, lines outside the boxes represent the 10th and the 90th percentiles, and lines inside the boxes represent the median. Results are representative of experiments performed with CD4 T cells from 5 patients and 5 control subjects. C, The proliferation of FLS was determined by flow cytometry and carboxyfluorescein succinimidyl ester (CFSE) dilution. FLS were stained with CFSE and cocultured for 7 days with CD4⫹,CD28⫹ or CD4⫹,CD28⫺ T cells isolated from patients with RA. Shaded area represents FLS control culture without CD4⫹ T cells; solid line represents FLS plus CD4⫹,CD28⫺ T cells; broken line represents FLS plus CD4⫹,CD28⫹ T cells. D, The proliferation indices were determined, defined as the ratio of the CFSE peak fluorescence intensities in FLS cultured with or without CD4 T cells. Bars show the mean and SD results from triplicate cultures. of patients with RA and tested for their growthpromoting capabilities. To quantify FLS proliferation, the CFSE dilution of these cells was measured after 7 SAWAI ET AL Figure 2. Role of CX3CR1 in T cell–induced growth of FLS. A, FLS were labeled with PKH26 and cocultured with CD4⫹,CD28⫹ and CD4⫹,CD28– T cells in the absence or presence of anti-CX3CR1 antibodies (Ab) for 5 days, and the number of FLS was evaluated by fluorescence microscopy. B, CFSE-labeled FLS were grown in the absence or presence of CD4⫹,CD28⫹ and CD4⫹,CD28⫺ T cells. Anti-CX3CR1 antibodies were added at the initiation of culture, and FLS proliferation was assessed by flow cytometry after 5 days. Proliferation indices were defined as described in Figure 1. C, FLS were labeled with PKH26 and expanded in the absence or presence of CD4 T cells, and FLS growth was determined by fluorescence microscopy after 5 days. Either T cells or FLS were pretreated with anti-CX3CR1 antibodies and intensively washed before being added to the cocultures. Blocking of CX3CR1 on CD4⫹,CD28⫺ T cells and FLS reduced FLS proliferation; however, blocking of CX3CR1 was more effective on CD4⫹,CD28⫺ T cells than on synoviocytes. Data in A and C are shown as box plots, where each box represents the 25th to 75th percentiles, lines outside the boxes represent the 10th and 90th percentiles, and lines inside the boxes represent the median. Bars in B show the mean and SD results from triplicate cultures. See Figure 1 for other definitions. T CELL–MEDIATED FLS PROLIFERATION 3219 Figure 3. Role of CX3CR1 in fractalkine (FKN) stimulation of FLS growth. CFSE-labeled FLS were cultured for 72 hours in the absence or presence of soluble FKN (sFKN; 100 ng/ml) and anti-CX3CR1 antibodies (Ab) or isotype control antibodies. Control cultures were maintained in medium alone. A, Fibroblast numbers were determined using fluorescence microscopy (original magnification ⫻ 200). B, Cell numbers were counted on 10 randomly selected fields of triplicate cultures. Data are shown as box plots, where each box represents the 25th to 75th percentiles, lines outside the boxes represent the 10th and 90th percentiles, and lines inside the boxes represent the median. Results are representative of experiments with 4 different FLS lines. See Figure 1 for other definitions. days of T cell–FLS coculture. Proliferation indices demonstrated that FLS cocultured with CD4⫹,CD28⫹ T cells divided ⬃2–3 times (average proliferation index of 5) within 7 days. In contrast, FLS cocultured with CD4⫹,CD28⫺ T cells diluted the CFSE levels by almost 30-fold, indicating that they passed through 5 division cycles (Figure 1C). Thus, senescent CD4⫹,CD28⫺ T cells clearly outperformed their CD4⫹,CD28⫹ counterparts in enhancing fibroblast proliferation, providing an explanation for the increased capacity of RA-derived T cells to support FLS growth. Enhanced FLS growth by CX3CR1 on both T cells and FLS. One of the characteristic features of CD4⫹,CD28⫺ T cells is the spontaneous expression of CX3CR1, the receptor for FKN (18). When partnering with FKN-expressing FLS, CD4⫹,CD28⫺ T cells utilize this chemokine receptor to costimulate cytokine production, granule expulsion, and survival. To investigate whether CX3CR1–FKN interactions are involved in regulating FLS growth, antibodies specific to CX3CR1 were added to the T cell–FLS cocultures. As shown in Figure 2A, blocking FKN from binding to its receptor 3220 essentially did not affect the growth-promoting activity of CD4⫹,CD28⫹ T cells but markedly reduced FLS growth in the cocultures containing CD4⫹,CD28⫺ T cells (P ⫽ 0.001). Essentially, FLS growth rates in the presence of both T cell subpopulations were very similar if anti-CX 3 CR1 was added. The importance of CX3CR1–FKN interactions in mediating T cell–induced FLS proliferation was confirmed using the CFSE dilution as readout (Figure 2B). The addition of antiCX 3 CR1 antibodies reversed the superiority of CD4⫹,CD28⫺ T cells in driving FLS outgrowth. Although the blocking experiments established the relevance of CX3CR1 in regulating the outcome of T cell–FLS interactions, they did not address on which of the cellular partners FKN and its receptor were expressed. Previous studies have shown that FKN is found mainly in FLS (18). Besides expressing CD4⫹,CD28⫺ T cells, FLS also express CX3CR1. We therefore performed blocking experiments in which either T cells or FLS were preincubated with anti-CX3CR1 antibodies and washed extensively before being added to the coculture system (Figure 2C). These experiments confirmed the competence of CD4⫹,CD28⫺ T cells in enhancing FLS proliferation and the role of CX3CR1 blockade in disrupting this effect. Preincubation of CD4⫹,CD28⫹ T cells with antibodies did not have any effect, which is consistent with the observation that these cells do not express CX3CR1. Blocking CX3CR1 on CD4⫹,CD28⫺ T cells had the most significant impact on impairing FLS growth (P ⫽ 0.001). The consequences of blocking CX3CR1 on FLS were smaller but significant (P ⫽ 0.001) and were seen irrespective of whether FLS were cocultured with CD4⫹,CD28⫹ or CD4⫹,CD28⫺ T cells. Responsiveness of FLS to the growth factor FKN determined by FLS expression of CX3CR1. The role of CX3CR1 in regulating the proliferation behavior of FLS raised the question of whether FKN is a growth factor for these specialized fibroblasts. To address that question, early-passage rheumatoid FLS were expanded in the absence or presence of sFKN. CFSE-labeled FLS were cultured for 3 days with 100 ng/ml sFKN. As shown in Figure 3, the expansion rate more than doubled, establishing that the chemokine FKN enhances multiplication of FLS. The growth-promoting effect of sFKN was completely blocked when anti-CX3CR1 was added with sFKN. Even in the absence of exogenous sFKN, anti-CX3CR1 antibodies lowered the rate of FLS proliferation, indicating the effect of constitutive FKN secretion in an autocrine growth-promoting loop. These SAWAI ET AL results confirmed a critical role of the FKN–CX3CR1 pathway in regulating synovial fibroblast growth. Role of TNF␣ in T cell–induced FLS proliferation. The results of experiments described in Figures 1 and 2 demonstrated that T cells modulate the cell cycle behavior of FLS. The mechanism involved CX3CR1, which raised the possibility that T cells interfered with the FKN–CX3CR1 pathway. Our group previously reported that CX3CR1 stimulation on CD4⫹,CD28⫺ T cells augments IFN␥ and TNF␣ production as well as granule expulsion (18). To determine whether these T cell cytokines influence the growth behavior of FLS, early-passage FLS were exposed to increasing doses of TNF␣ and IFN␥, and their proliferative responses were quantified. IFN␥ failed to stimulate FLS growth, and high doses inhibited FLS proliferation (data not shown). In contrast, TNF␣ had excellent growth factor function for FLS (Figure 4B). Particularly at TNF␣ doses of 1–5 ng/ml, FLS responded with more than doubling of cell numbers by day 3. Higher doses of TNF␣ were not as effective; 100 ng/ml TNF␣ could no longer boost FLS growth. To examine the regulation of TNF␣ production in the T cell–FLS coculture system, TNF␣ messenger RNA was quantified in cocultures containing either CD4⫹,CD28⫹ or CD4⫹,CD28⫺ T cells. The allogeneic stimulation provided in the system was sufficient to induce TNF␣ transcription (Figure 4A), and TNF␣ levels were clearly higher in the cultures with CD4⫹,CD28⫺ T cells. The excess production of TNF␣ was blocked by adding anti-CX3CR1 antibodies at the initiation of the T cell–FLS coculture. These data confirmed our prior reports that CD4⫹,CD28⫺ T cells are an excellent source of TNF␣ and utilize CX3CR1 to costimulate the production of this cytokine. To examine whether TNF␣-driven FLS proliferation was dependent on CX3CR1, TNF␣ stimulation was performed in the absence or presence of antiCX3CR1 antibodies (Figure 4C). TNF␣ induced robust enhancement of FLS proliferation at doses of 1 ng/ml and 5 ng/ml. The addition of anti-CX3CR1 antibodies abrogated this enhancement. The results of these experiments established that TNF␣-driven proliferation of FLS depends on signaling through CX3CR1. TNF␣-regulated CX3CR1 expression and FKN production by FLS. Possible mechanisms through which TNF␣ could modulate the growth pattern of FLS include enhanced CX3CR1 expression as well as regulation of FKN production. Stimulation of early-passage FLS with TNF␣ at a dose of 5 ng/ml promptly induced up-regulation of FKN. The spontaneous production of T CELL–MEDIATED FLS PROLIFERATION 3221 Figure 4. Relationship between the CX3CR1–fractalkine (FKN) axis and tumor necrosis factor ␣ (TNF␣). A, CD4⫹,CD28⫹ or CD4⫹,CD28⫺ T cells were cultured with FLS for 24 hours. TNF␣ sequences were quantified by real-time polymerase chain reaction (PCR) and adjusted for ␤-actin copies. B, PKH26-labeled FLS were expanded in the presence of increasing doses of TNF␣, and densities of FLS were assessed after 72 hours. C, FLS proliferation was induced by TNF␣ in the absence or presence of anti-CX3CR1 antibody (Ab). FLS density was measured after 3 days, by fluorescence microscopy. D, FLS were seeded at a density of 5 ⫻ 105 cells and stimulated with 5 ng/ml TNF␣ for 24 hours. Total RNA was extracted, and FKN- and CX3CR1-specific transcripts were quantified by real-time PCR. Bars in A and D show the mean and SD results from triplicate cultures. Data in B and C are shown as box plots, where each box represents the 25th to 75th percentiles, lines outside the boxes represent the 10th and 90th percentiles, and lines inside the boxes represent the median. See Figure 1 for other definitions. FKN was minimal, but exposure to TNF␣ resulted in robust expression of FKN sequences (Figure 4D). CX3CR1 transcripts were detectable at low levels in the absence of stimulation. Within 24 hours, TNF␣ in- creased CX3CR1 transcript production by more than 3-fold (Figure 4D). To test whether this mechanism is functional in T cell–FLS interactions, we used fluorescence-activated 3222 SAWAI ET AL were explored by measuring the proliferation response of T cell–exposed FLS to FKN. For these experiments, FLS were cocultured with the 2 T cell subpopulations for 3 days. The T cells were removed, and sFKN was added. FLS that were exposed to CD4⫹,CD28⫺ T cells reacted with an enhanced proliferative response to FKN (Figure 5B). In contrast, FLS cultured in the presence of CD4⫹,CD28⫹ T cells showed only a modest enhancement of cell replication. These data support the concept that T cells determine FLS growth behavior by regulating the expression of growth factor receptors. DISCUSSION This study demonstrates that CD4 T cells regulate proliferation of synovial fibroblasts, which is a Figure 5. Role of CD4 T cells in up-regulating the expression of CX3CR1 on FLS and sensitization of FLS toward the growth-inducing effect of fractalkine (FKN). A, CX3CR1 expression on FLS was analyzed by fluorescence-activated cell sorting after coculture with CD4 T cells for 96 hours. Shaded area represents FLS cultured with medium only; bold line represents FLS cultured with CD4⫹,CD28⫺ T cells; broken line represents FLS cultured with CD4⫹,CD28⫹ T cells; black line represents isotype control antibody. Results are shown as histograms of mean fluorescence intensities and are representative of 5 experiments. B, CD4⫹,CD28⫹ (solid line) or CD4⫹,CD28⫺ (broken line) T cells were cocultured with FLS and washed away after 72 hours. Soluble FKN (sFKN) was added at doses of 50 ng/ml and 100 ng/ml. After an additional 96 hours, the number of FLS per lpf was obtained by counting 10 randomly selected fields of triplicate cultures. Results are shown as the mean ⫾ SD. See Figure 1 for other definitions. cell sorting to monitor the surface expression of CX3CR1 in FLS cocultured with either CD28⫹ or CD28⫺ CD4 T cells. FLS grown in the absence of either type of T cells expressed CX3CR1 on their surface. The mean fluorescence intensity of CX3CR1 staining doubled in the presence of CD4⫹,CD28⫹ T cells and increased almost 4-fold when FLS were exposed to CD4⫹,CD28⫺ T cells (Figure 5A). These experiments demonstrated that T cells regulate the expression of growth receptors and chemokine receptors on fibroblasts. The functional consequences of this T cell activity Figure 6. Fractalkine (FKN)–CX3CR1 interactions in rheumatoid synovitis. CX3CR1 on CD4⫹,CD28⫺ T cells recognizes FKN expressed on the surface of fibroblast-like synoviocytes (FLS) and costimulates production of tumor necrosis factor ␣ (TNF␣). FLS also have CX3CR1 on the surface and secrete FKN and utilize this receptor–ligand pair for autocrine growth stimulation. TNF␣ can amplify this autocrine loop by inducing CX3CR1 expression and enhancing FKN production. TCR ⫽ T cell receptor. Color figure can be viewed in the online issue, which is available at http://www. arthritisrheum.org. T CELL–MEDIATED FLS PROLIFERATION critical disease mechanism in the rheumatoid joint. Underlying molecular pathways involve FKN and its receptor CX3CR1, a receptor–ligand pair formerly implicated in chemotaxis and cell adhesion. Crosslinking of CX3CR1 on CD4 T cells amplifies the production of TNF␣. Interestingly, FLS also express CX3CR1 and, by releasing the ligand FKN, sustain an autocrine growthpromotion loop. Both components of this autocrine loop are enhanced by TNF␣, which connects T cell responses and FLS proliferation (Figure 6). CX3CR1 expression is a feature of a unique T cell subpopulation, senescent CD4⫹,CD28⫺ T cells, which accumulate in patients with RA and mediate many of the proinflammatory processes. Implicating CX3CR1 and FKN in 2 interconnected pathways of T cell–induced FLS proliferation identifies these molecules as pluripotent mediators and potential therapeutic targets in rheumatoid synovitis. CD4⫹,CD28⫺ T cells are end-differentiated with short telomeres and sluggish cell replication, identifying them as senescent cells (12,13). Besides loss of the costimulatory molecule CD28, they are functionally characterized by the de novo expression of a series of novel regulatory receptors (11,14,18). Equipped with an unusual set of receptors and signal transduction pathways (28), CD4⫹,CD28⫺ T cells are capable of communicating and interacting with cellular partners that are primarily not antigen-presenting cells but are mesenchymal cells localized at sites of tissue inflammation (29). This rule certainly holds for FLS, a cell population that builds the stroma of the synovial membrane and is meant to provide structural support and nutrition to the bony and cartilaginous elements of the joint. Our group recently demonstrated that CD4⫹,CD28⫺ T cells, as opposed to their CD28⫹ counterparts, express and use the chemokine receptor CX3CR1 to communicate with FLS (18). Those studies emphasized the role of FLSexpressed FKN in costimulating T cell activation, including CD69 induction, cytoplasmic granule expulsion, and IFN␥ release. The current study extends the role of this costimulatory signal beyond cytokine release by activated CD4⫹,CD28⫺ T cells. Instead, it becomes obvious that a CX3CR1-mediated signal enhances the ability of CD4⫹,CD28⫺ T cells to intensify the hyperproliferative response of synovial fibroblasts. Enhanced FLS replication is mediated by TNF␣, extending the list of disease-relevant functions of TNF␣ in RA by yet another process. An intriguing aspect of the current study was the finding that CX3CR1 was not restricted to T cells but was also expressed on FLS. Sequential antibodyblocking experiments clarified that CX3CR1 on both T 3223 cells and FLS was functionally relevant for fibroblast proliferation, although the impact of blocking CX3CR1 on the T cells was clearly more significant than hindering access to CX3CR1 on FLS. Blocking CX3CR1 on T cells markedly reduced TNF␣ production, confirming earlier data that TNF␣ release by CD4⫹,CD28⫺ T cells is modulated by a CX3CR1-derived signal (18). TNF␣, when dosed optimally, had a strong proliferative effect on FLS. The concentration of sFKN in the joint fluid of patients with RA is elevated (23,24). Blaschke and coworkers (30) reported that sFKN induces matrix metalloproteinase 2 production in synovial fibroblasts in vitro, suggesting that sFKN has broad proinflammatory activities in the rheumatoid joint. The relevance of FKN as a fibroblast growth factor was confirmed in experiments testing the growth-promoting effect of exogenous FKN. A proinflammatory role for this chemokine has been supported by gene expression experiments in rat adjuvant-induced arthritis (22). FKN and its receptor CX3CR1 were abundantly expressed on day 18, a time of intense inflammation in the rat joint. Soluble FKN is produced by proteolytic cleavage of membrane-bound FKN, and it has been suspected that its major functions are chemotactic and proadhesive (20). Accordingly, inhibition of FKN can ameliorate murine collageninduced arthritis (31). Treatment with anti-FKN monoclonal antibodies inhibited migration of adoptively transferred cells into the inflamed synovium. FKN is a unique member of the chemokine gene family, with a single transmembrane region and a short intracellular C terminus. We did consider the possibility that CX3CR1 expressed on the surface of T cells promotes FLS cell cycling by signaling through membrane-integrated FKN. However, the soluble form of FKN effectively enhanced FLS proliferation. When cocultured with CD4⫹,CD28⫺ T cells, FLS were rendered more susceptible to the growth-inducing action of sFKN, even after the removal of T cells. These data strongly support the concept that T cell–mediated FLS proliferation reflects a change in the responsiveness of FLS to the growth factor. This notion was strengthened by experiments demonstrating a TNF␣-driven enhancement in the expression of the FKN receptor CX3CR1 on FLS. In essence, selected T cells appear to regulate the threshold of FLS growth responses in the rheumatoid joint. The facts that FLS themselves supply the growth factor and that production of the growth factor is subject to T cell regulation add new twists to the T cell–FLS relationship. The interactions are contact dependent, as far as stimulation of 3224 CX3CR1 on T cells is involved (18). T cells reside in the sublining and are therefore distal from FLS in the synovial lining. However, cadherin 11–positive cells are also found in the sublining layer in inflamed synovium and likely represent FLS migrating to the synovial lining (32). In addition, T cells may secrete soluble mediators that up-regulate the FKN–CX3CR1 pathway in synovial lining cells. The current study did not examine how CX3CR1 interferes with cell cycle regulation in FLS. Reports by Lucas et al (33) have indicated that aortic smooth muscle cells, another type of specialized mesenchymal cells, utilize the CX3CR1–FKN pathway for proliferation control. In vitro, the soluble form of FKN promotes aortic smooth muscle cell proliferation through NF-B signaling (34). The possibility remains that membraneintegrated FKN is also capable of enhancing FLS and smooth muscle cell replication. In the tissue site, FLS may receive growth-promoting signals from neighboring fibroblasts as well as through the secreted form of the ligand. FLS have long been recognized as displaying almost autonomous features in terms of their proliferative response, a characteristic that has encouraged discussions on whether these cells are tumorlike (35,36). The autocrine growth loop described here, fueled by FKN, may well contribute to the hyperplastic reaction pattern of mesenchymal cells in sites of inflamed tissue. In that sense, intimal hyperplasia driven by vascular smooth muscle cells and synovial hyperplasia driven by FLS may share pathways of dysregulation. Vascular smooth muscle cells are now emerging as important partners of T cells in inflamed atherosclerotic plaques, not only because they are the subjects of T cell effector functions but also because they create synaptic communication platforms with tissue-infiltrating T cells (37,38). An unresolved issue of the current study is the question of how CD4⫹,CD28⫺ T cells are activated when interacting with FLS. Possibly, CD4⫹,CD28⫺ T cells recognize alloantigens on FLS lines. However, CD4⫹,CD28⫺ T cells are generally oligoclonal (39) and have been shown to be defective in alloreactive responses (40). In contrast, CD4⫹,CD28⫺ T cells respond vigorously in autologous mixed-lymphocyte reactions and may indeed be autoreactive (26). Evidence has been provided that synovial T cells are easier to stimulate, irrespective of their antigen specificity (41). This may point toward a fundamental abnormality in the response threshold of T cells in RA, an abnormality that may be a functional consequence of T cell senescence. Rheumatoid FLS play an important role in the pathogenesis of RA. They demonstrate pseudotumor- SAWAI ET AL like overgrowth and resist apoptosis (42–44). The present study demonstrated the pivotal roles the CX3CR1 pathway plays in senescent CD4⫹,CD28⫺ T cell and FLS interaction and in the T cell–induced overgrowth of rheumatoid FLS. Activation of the CX3CR1 pathway on CD4⫹,CD28⫺ T cells through T cell–FLS interactions leads to activation of T cells, enhanced TNF␣ production, and up-regulated FKN and CX3CR1 expression on FLS (Figure 6). The CX3CR1 pathway in FLS is directly linked to enhanced FLS proliferation and can be activated by both membranebound FKN and sFKN, which are present in high concentrations in rheumatoid joints (Figure 6). Because CX3CR1 stimulation is a common pathway used by T cells, exogenous TNF␣, and sFKN to stimulate FLS proliferation, FKN receptor blocking might serve as an effective therapeutic intervention and reduce hyperplasia of the synovium and joint inflammation in RA. ACKNOWLEDGMENTS We thank Dr. Sergey Pryshchep for help with preparing the figures and Tamela Yeargin for editorial support. AUTHOR CONTRIBUTIONS Dr. Weyand 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 design. Goronzy, Weyand. Acquisition of data. Sawai, Park, He. Analysis and interpretation of data. Goronzy, Weyand. Manuscript preparation. Goronzy, Weyand. Statistical analysis. Sawai, Park. REFERENCES 1. Yamanishi Y, Firestein GS. Pathogenesis of rheumatoid arthritis: the role of synoviocytes [review]. Rheum Dis Clin North Am 2001;27:355–71. 2. McInnes IB, Leung BP, Liew FY. Cell-cell interactions in synovitis: interactions between T lymphocytes and synovial cells [review]. Arthritis Res 2000;2:374–8. 3. Muller-Ladner U, Gay RE, Gay S. Activation of synoviocytes [review]. Curr Opin Rheumatol 2000;12:186–94. 4. Muller-Ladner U, Kriegsmann J, Franklin BN, Matsumoto S, Geiler T, Gay RE, et al. Synovial fibroblasts of patients with rheumatoid arthritis attach to and invade normal human cartilage when engrafted into SCID mice. Am J Pathol 1996;149:1607–15. 5. Tolboom TC, van der Helm-Van Mil AH, Nelissen RG, Breedveld FC, Toes RE, Huizinga TW. Invasiveness of fibroblast-like synoviocytes is an individual patient characteristic associated with the rate of joint destruction in patients with rheumatoid arthritis. Arthritis Rheum 2005;52:1999–2002. 6. Filer A, Parsonage G, Smith E, Osborne C, Thomas AM, Curnow SJ, et al. Differential survival of leukocyte subsets mediated by synovial, bone marrow, and skin fibroblasts: site-specific versus activation-dependent survival of T cells and neutrophils. Arthritis Rheum 2006;54:2096–108. 7. Salmon M, Scheel-Toellner D, Huissoon AP, Pilling D, Shamsadeen N, Hyde H, et al. Inhibition of T cell apoptosis in the rheumatoid synovium. J Clin Invest 1997;99:439–46. T CELL–MEDIATED FLS PROLIFERATION 8. Bradfield PF, Amft N, Vernon-Wilson E, Exley AE, Parsonage G, Rainger GE, et al. Rheumatoid fibroblast-like synoviocytes overexpress the chemokine stromal cell–derived factor 1 (CXCL12), which supports distinct patterns and rates of CD4⫹ and CD8⫹ T cell migration within synovial tissue. Arthritis Rheum 2003;48: 2472–82. 9. Rezzonico R, Burger D, Dayer JM. Direct contact between T lymphocytes and human dermal fibroblasts or synoviocytes downregulates types I and III collagen production via cell-associated cytokines. J Biol Chem 1998;273:18720–8. 10. Burger D, Rezzonico R, Li JM, Modoux C, Pierce RA, Welgus HG, et al. Imbalance between interstitial collagenase and tissue inhibitor of metalloproteinases 1 in synoviocytes and fibroblasts upon direct contact with stimulated T lymphocytes: involvement of membrane-associated cytokines. Arthritis Rheum 1998;41: 1748–59. 11. Goronzy JJ, Weyand CM. Rheumatoid arthritis [review]. Immunol Rev 2005;204:55–73. 12. Koetz K, Bryl E, Spickschen K, O’Fallon WM, Goronzy JJ, Weyand CM. T cell homeostasis in patients with rheumatoid arthritis. Proc Natl Acad Sci U S A 2000;97:9203–8. 13. Schonland SO, Lopez C, Widmann T, Zimmer J, Bryl E, Goronzy JJ, et al. Premature telomeric loss in rheumatoid arthritis is genetically determined and involves both myeloid and lymphoid cell lineages. Proc Natl Acad Sci U S A 2003;100:13471–6. 14. Yen JH, Moore BE, Nakajima T, Scholl D, Schaid DJ, Weyand CM, et al. Major histocompatibility complex class I-recognizing receptors are disease risk genes in rheumatoid arthritis. J Exp Med 2001;193:1159–67. 15. Groh V, Bruhl A, El-Gabalawy H, Nelson JL, Spies T. Stimulation of T cell autoreactivity by anomalous expression of NKG2D and its MIC ligands in rheumatoid arthritis. Proc Natl Acad Sci U S A 2003;100:9452–7. 16. Schirmer M, Vallejo AN, Weyand CM, Goronzy JJ. Resistance to apoptosis and elevated expression of Bcl-2 in clonally expanded CD4⫹CD28⫺ T cells from rheumatoid arthritis patients. J Immunol 1998;161:1018–25. 17. Nakajima T, Schulte S, Warrington KJ, Kopecky SL, Frye RL, Goronzy JJ, et al. T cell-mediated lysis of endothelial cells in acute coronary syndromes. Circulation 2002;105:570–5. 18. Sawai H, Park YW, Roberson J, Imai T, Goronzy JJ, Weyand CM. T cell costimulation by fractalkine-expressing synoviocytes in rheumatoid arthritis. Arthritis Rheum 2005;52:1392–401. 19. Bazan JF, Bacon KB, Hardiman G, Wang W, Soo K, Rossi D, et al. A new class of membrane-bound chemokine with a CX3C motif. Nature 1997;385:640–4. 20. Imai T, Hieshima K, Haskell C, Baba M, Nagira M, Nishimura M, et al. Identification and molecular characterization of fractalkine receptor CX3CR1, which mediates both leukocyte migration and adhesion. Cell 1997;91:521–30. 21. Tarrant TK, Patel DD. Chemokines and leukocyte trafficking in rheumatoid arthritis [review]. Pathophysiology 2006;13:1–14. 22. Ruth JH, Volin MV, Haines GK 3rd, Woodruff DC, Katschke KJ Jr, Woods JM, et al. Fractalkine, a novel chemokine in rheumatoid arthritis and in rat adjuvant-induced arthritis. Arthritis Rheum 2001;44:1568–81. 23. Umehara H, Tanaka M, Sawaki T, Jin ZX, Huang CR, Dong L, et al. Fractalkine in rheumatoid arthritis and allied conditions. Mod Rheumatol 2006;16:124–30. 24. Volin MV, Woods JM, Amin MA, Connors MA, Harlow LA, Koch AE. Fractalkine: a novel angiogenic chemokine in rheumatoid arthritis. Am J Pathol 2001;159:1521–30. 25. Arnett FC, Edworthy SM, Bloch DA, McShane DJ, Fries JF, Cooper NS, et al. The American Rheumatism Association 1987 revised criteria for the classification of rheumatoid arthritis. Arthritis Rheum 1988;31:315–24. 26. Schmidt D, Goronzy JJ, Weyand CM. CD4⫹ CD7⫺ CD28⫺ T 3225 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43. 44. cells are expanded in rheumatoid arthritis and are characterized by autoreactivity. J Clin Invest 1996;97:2027–37. Snyder MR, Lucas M, Vivier E, Weyand CM, Goronzy JJ. Selective activation of the c-Jun NH2-terminal protein kinase signaling pathway by stimulatory KIR in the absence of KARAP/ DAP12 in CD4⫹ T cells. J Exp Med 2003;197:437–49. Zhang X, Nakajima T, Goronzy JJ, Weyand CM. Tissue trafficking patterns of effector memory CD4⫹ T cells in rheumatoid arthritis. Arthritis Rheum 2005;52:3839–49. Weyand CM, Goronzy JJ. T-cell-targeted therapies in rheumatoid arthritis [review]. Nat Clin Pract Rheumatol 2006;2:201–10. Blaschke S, Koziolek M, Schwarz A, Benohr P, Middel P, Schwarz G, et al. Proinflammatory role of fractalkine (CX3CL1) in rheumatoid arthritis. J Rheumatol 2003;30:1918–27. Nanki T, Urasaki Y, Imai T, Nishimura M, Muramoto K, Kubota T, et al. Inhibition of fractalkine ameliorates murine collageninduced arthritis. J Immunol 2004;173:7010–6. Lee DM, Kiener HP, Agarwal SK, Noss EH, Watts GF, Chisaka O, et al. Cadherin-11 in synovial lining formation and pathology in arthritis. Science 2007;315:1006–10. Lucas AD, Bursill C, Guzik TJ, Sadowski J, Channon KM, Greaves DR. Smooth muscle cells in human atherosclerotic plaques express the fractalkine receptor CX3CR1 and undergo chemotaxis to the CX3C chemokine fractalkine (CX3CL1). Circulation 2003;108:2498–504. Chandrasekar B, Mummidi S, Perla RP, Bysani S, Dulin NO, Liu F, et al. Fractalkine (CX3CL1) stimulated by nuclear factor B (NF-B)-dependent inflammatory signals induces aortic smooth muscle cell proliferation through an autocrine pathway. Biochem J 2003;373:547–58. Li X, Makarov SS. An essential role of NF-B in the “tumor-like” phenotype of arthritic synoviocytes. Proc Natl Acad Sci U S A 2006;103:17432–7. Aupperle KR, Boyle DL, Hendrix M, Seftor EA, Zvaifler NJ, Barbosa M, et al. Regulation of synoviocyte proliferation, apoptosis, and invasion by the p53 tumor suppressor gene. Am J Pathol 1998;152:1091–8. Pryshchep S, Sato K, Goronzy JJ, Weyand CM. T cell recognition and killing of vascular smooth muscle cells in acute coronary syndrome. Circ Res 2006;98:1168–76. Sato K, Niessner A, Kopecky SL, Frye RL, Goronzy JJ, Weyand CM. TRAIL-expressing T cells induce apoptosis of vascular smooth muscle cells in the atherosclerotic plaque. J Exp Med 2006;203:239–50. Schmidt D, Martens PB, Weyand CM, Goronzy JJ. The repertoire of CD4⫹ CD28⫺ T cells in rheumatoid arthritis. Mol Med 1996;2:608–18. Morishita Y, Sao H, Hansen JA, Martin PJ. A distinct subset of human CD4⫹ cells with a limited alloreactive T cell receptor repertoire. J Immunol 1989;143:2783–9. Miranda-Carus ME, Balsa A, Benito-Miguel M, Perez de Ayala C, Martin-Mola E. IL-15 and the initiation of cell contact-dependent synovial fibroblast-T lymphocyte cross-talk in rheumatoid arthritis: effect of methotrexate. J Immunol 2004;173:1463–76. Devauchelle V, Essabbani A, De Pinieux G, Germain S, Tourneur L, Mistou S, et al. Characterization and functional consequences of underexpression of clusterin in rheumatoid arthritis. J Immunol 2006;177:6471–9. Uzan B, Ea HK, Launay JM, Garel JM, Champy R, Cressent M, et al. A critical role for adrenomedullin-calcitonin receptor-like receptor in regulating rheumatoid fibroblast-like synoviocyte apoptosis. J Immunol 2006;176:5548–58. Palao G, Santiago B, Galindo MA, Rullas JN, Alcami J, Ramirez JC, et al. Fas activation of a proinflammatory program in rheumatoid synoviocytes and its regulation by FLIP and caspase 8 signaling. Arthritis Rheum 2006;54:1473–81.