Katsumitsu Arai MD Department of Orthopaedic Surgery, Niigata University School of Medicine, Asahimachi-dori 1, Niigata 951, Japan.
The distribution of CD57+ T and CD56+ T cells in patients with RA was examined. In control osteoarthritis patients, these cells exist as a minor population in the peripheral blood. Our data show that in patients with RA, CD57+ T cell levels are elevated in peripheral blood, knee joint fluid, knee synovial membrane and bone marrow (BM), compared with peripheral blood of controls. CD57+ T cells are especially high in knee joint fluid and joint-adjacent BM, while CD56+ T cells show no such increase. CD57+ T cells contain a major population of CD8+ cells and higher proportions of CD4−8− cells and γδ T cells than do CD57−T cells. CD57+T cells in peripheral blood and joint fluid increase with the duration of disease. Erythrocyte sedimentation rate (ESR) is inversely correlated with the proportion of CD57+T cells in the joint fluid. Although RA frequently occurrs in patients with CD3+57+ cell leukaemia, and some CD57+T cells are likely to be involved in the onset of RA, we suggest that CD57+T cells may rather suppress inflammation of RA, and other cellular components (e. g. granulocytes) may govern the severity of the inflammation of RA. These CD57+ T cells are probably generated extrathymically in the adjacent BM or joint space.
RA is a chronic inflammatory autoimmune disease in which some autoreactive T cells and autoantibodies are reportedly associated with the inflammation of the joint [1, 2]. A small number of T cells with CD56 or CD57 are normally present in periphery of humans [3–5]. These T cells with natural killer (NK) markers are abundant in the liver and are present in tumour-infiltrating lymphocytes [6, 7]. Their proportions in the peripheral blood also increase in patients with colorectal cancer [6, 7]. CD57+ T cells reportedly increase in patients with AIDS [8, 9] and those subjected to kidney, heart and bone marrow (BM) transplantation [10–12]. It is suggested that these T cells with NK cell markers may be counterparts of extrathymic T cells in mice , which have intermediate T cell receptor (TCR) (a major proportion also have NK cell marker, NK1.1), and contain potentially self-minor lymphocyte stimulatory (Mls) antigen–reactive clones [14, 15] and thereby may play a role in inducing autoimmune diseases [14–16]. Accordingly, these T cells with an NK cell marker in humans might also be important in RA.
RA frequently occurs in patients with CD3+57+ cell leukaemia , and CD57+ T cells increase in peripheral blood of RA patients . These findings suggest that CD57+ T cells play an important role in the onset and pathogenesis of RA. To study further the pathogenesis of RA, we examine how CD57+ and CD56+ T cells are distributed in various organs. Further, to determine the function of CD57+ T cells, we examine the relationship between CD57+ T cells and clinical features of RA.
PATIENTS AND METHODS
Subjects consisted of 45 RA patients (five males, 40 females) from the arthritis clinic of the Orthopaedic Surgery, Niigata University School of Medicine. Ages ranged from 32 to 72 years (59.1 ± 1.5 years, mean±s.e.m.). They all fulfilled the 1987 American College of Rheumatology (formerly the American Rheumatism Association) criteria for the diagnosis of RA . Age-matched controls (n = 17) were otherwise healthy patients with osteoarthritis (OA) (three males, 14 females) whose ages ranged from 32 to 72 years (54.7 ± 2.6 years).
Peripheral blood mononuclear cells (PBMC) were isolated from heparinized blood and were derived from the interface by Ficoll–Paque gradient (1.077) centrifugation. Thymus was obtained from resected specimens from patients who had undergone cardiac and lung surgery. To prepare MNC from the thymus, samples were cut into small pieces with scissors and pressed through a 200-gauge stainless steel mesh. Knee joint fluid was obtained from RA out-patients or from RA patients undergoing total knee arthroplasty. Synovial membranes and joint-adjacent BM were obtained during the operation for total knee arthroplasty using an inflated pneumatic tourniquet. Iliac BM aspirate was obtained from the iliac crest of patients with RA; since aspirated amounts were <2.5 ml in each area, the probability of contamination by peripheral blood was considered to be negligible . MNC of the joint fluid and adjacent BM were derived by Ficoll–Paque gradient centrifugation (similar to the procedure followed with peripheral blood). To obtain MNC from synovial membrane, a minor modification of the technique previously described by Thomas et al. was used . Briefly, samples were cut into small pieces with scissors and then digested with 1 mg/ml collagenase (Wako, Osaka, Japan) at 37°C for 60 min. Treated samples were pressed through a 200-gauge stainless steel mesh and suspended in Eagles' minimal essential medium (MEM) with 5 mm HEPES (Nissui Pharmaceutical, Tokyo, Japan) supplemented with 2% new born calf serum. The suspension was centrifuged to remove the supernatant and then purified by Ficoll–Paque gradient (1.077) centrifugation. All materials were used with the permission of the patients.
Phenotype of cells by flow cytometory
MNC from peripheral blood, joint fluid, synovial membrane and BM were incubated with FITC-, PE- or biotin-labelled MoAbs. The MoAbs used were CD3 (NU-T3; Nichirei, Tokyo, Japan), CD8 (NU-Ts/c; Nichirei), CD4 (Leu-3a), CD56 (Leu-19), CD57 (Leu-7), and TCR γδ (TCRδ1) (T Cell Sciences, Cambridge, MA). Two- or three-colour flow cytometric analysis was performed using a fluorescence-activated cell analyser (FACScan; Becton Dickinson, Mountain View, CA). Cells in the two-colour staining (1 × 104) and 3 × 104 cells in the three-colour staining were analysed.
Staining for the morphology of CD57+ T cells
Morphological observation by light microscopy was carried out for CD57+ T cells in joint fluid. CD57+ T cells were isolated from joint fluid by cell sorter (FACS Vantage; Becton Dickinson) after being stained with MoAbs and then re-stained by the May–Grünwald method .
When distributions were normal and variances approximately equal, differences between means were compared by Student's-t-test. If either criterion was violated, the Mann–Whitney U-test was employed to compare means. Differences were considered significant when P ≤ 0.05.
Abundance of CD57+ T cells in joint fluid and joint-adjacent BM in patients
The distribution of CD57+ T cells in peripheral blood, joint fluid, synovial membrane, joint-adjacent BM and iliac BM was examined. A considerably larger population of CD3+CD57+ cells was identified in joint fluid and joint-adjacent BM (Fig. 1a). To confirm this finding, the population of CD3+CD57+ cells among the total CD3+ cells was further examined in peripheral blood, joint fluid, synovial membrane, joint-adjacent BM, and iliac BM of RA patients, and peripheral blood and thymus of controls. In RA patients, CD57+ T cells were abundant in joint fluid (25.6 ± 3.2% (mean ± s.e.m.)) and joint-adjacent BM (36.8 ± 8.5%) compared with peripheral blood (18.2 ± 2.0%) (P < 0.05) (Fig. 1b). CD57+ T cells significantly more abundant in every RA patient site compared with peripheral blood of controls (11.7 ± 1.5%) (P < 0.05) (Fig. 1b). However, CD57+ T cells were not significantly more abundant in synovial membrane and iliac BM compared with peripheral blood in RA patients (Fig. 1b).
We also looked for CD57+ T cells in joint-adjacent BM of OA patients, but there were few (data not shown). It should be noted that CD57+ T cells were not present in the thymus controls (Fig. 1b).
Population of CD56+ T cells among various organs in patients with RA
The distribution of CD56+ T cells was examined in parallel with the examination of CD57+ T cells. The population of CD56+ T cells was not increased in peripheral blood, joint fluid, synovial membrane and joint-adjacent BM of two RA patients (Fig. 2a). Similarly, the population of CD3+CD56+ cells among the total CD3+ cells of RA patients in every site was not greater than that in peripheral blood of controls (Fig. 2b). It should also be noted that, as with CD57+ T cells, CD56+ T cells were not present in the thymus of controls (Fig. 2b).
The composition of CD4−8−, CD4+, CD8+ and TCRγδ cells in CD57+ T cells
We used three-colour staining for CD57, CD4, CD8, and TCRγδ to determine the proportions of CD4−8−, CD4+, CD8+ and TCRγδ in the peripheral blood of two RA patients. CD8+ cells were dominant in CD57+ T cells, but CD4+ cells were dominant in CD57− T cells (Fig. 3). CD57+ T cells also contained higher populations of CD4−8− and TCRγδ cells than did CD57−T cells (Fig. 3). To confirm these findings, the population of CD4−8− and TCRγδ cells in CD57+ T cells was further examined in peripheral blood and joint fluid of RA patients. In peripheral blood, CD57+ T cells contained significantly (P<0.05) more CD4−8− cells (7.9 ± 1.3) than did CD57− T cells (3.5 ± 0.44) (Fig. 4a), and TCRγδ cells were also significantly (P < 0.01) more abundant in CD57+ T cells (14.2 ± 3.5) compared with CD57−T cells (3.6 ± 0.8) (Fig. 4b). In joint fluid, CD4−8− cells were significantly more abundant in CD57+ T cells (7.0 ± 1.5) than in CD57−T cells (2.9 ± 0.49) (P < 0.05) (Fig. 4a). Although CD57+ NK cells show a large granular lymphocyte morphology , sorted CD3+ CD57+ T cells in the joint fluid of RA patients were agranular lymphocytes (data not shown).
Relationship between proportions of CD57+ T cells or CD56+ T cells and clinical data
In confirmation of an earlier  report, a correlation was observed between duration of the disease and the proportion of CD57+ T cells in peripheral blood (r = 0.58, P < 0.05) (Fig. 5a). A statistically significant correlation was not observed between the proportions of CD57+ T cells in peripheral blood and ages of patients (not shown). We also found an inverse relationship between erythrocyte sedimentation rate (ESR) and the proportion of CD57+ T cells in joint fluid (r = −0.50, P < 0.05) (Fig. 5b). The relationships between joint fluid CD57+ T cells and C-reactive protein (CRP) (Fig. 5c) and between the proportion of CD57+ T cells in peripheral blood and ESR (Fig. 5d) were not statistically significant.
Proportions of CD56+ T cells were not correlated with any of these clinical parameters (data not shown).
Since the identification of NK cell markers CD57  and CD56 , investigators have noticed that there is a small population of T cells with these NK cell antigens [3–5]. CD56 is now also known as a neural cell adhesion molecule-1 , and CD57 is a sulfated carbohydrate determinant on glycoprotein of neural cells . RA occurs frequently in patients with CD3+ CD57+ T cell leukaemia . Further, the number of CD57+ T cells in peripheral blood of RA patients is higher than in normals, and correlates with the duration of RA . These findings suggest that CD57+ T cells play an important role in the onset and pathogenesis of RA.
The present study reveals that the number of CD57+ T cells is elevated in RA patients, especially in joint fluids and in joint-adjacent BM, but not in iliac BM. These cells are composed mainly of CD8+ cells and significant high populations of CD4−CD8− T cells and γδ T cells. Proportions of these cells in joint fluids are inversely correlated with ESR of RA patients, which is well known as an inflammatory marker. On the other hand, CD56+ T cells are unlikely to be involoved in the pathogenesis of RA.
Liver and BM are known to be the sites where T cells differentiate extrathymically in mice [13, 14,27–29]. The extrathymic T cells are intermediate TCR cells, including γδ T cells and αβ T cells [30–37]. Intermediate αβ T cells are exclusively CD44high and IL-2Rβhigh, both of which are shared by NK cells and consist of an NK1.1 antigen+ (NK1+) subset and another NK1− subset . NK1+ intermediate T cells predominate in the liver, while NK1− int. T cells predominate in BM. Human liver also contains a large population of CD3+CD56+αβ T cells, and T cells in human BM are mainly CD3+CD57+αβ T cells [7, 37]. Both NK1+ T cells of mice and CD3+CD56+αβ T cells of humans can be potent anti-tumour effectors when stimulated by IL-12 [37–40]. Further, NK1− int. T cells in mice as well as CD3+CD57+αβ T cells in humans increase with ageing  and after BM transplantation [10, 29, 41]. From these findings, we propose that although human T cells with NK cell markers exclusively express high TCR, human CD56+αβ T cells and CD57+αβ T cells are counterparts of NK1+ and NK1− int. T cells in mice, respectively [13, 37].
There are characteristic changes in joint-adjacent BM of RA patients, because CD57+αβ T cells are abundant, and myeloid lineage cells (which can differentiate into polymorphonuclear leucocytes) are present . This finding is consistent with a report that extrathymic T cells are activated in parallel with granulocytes in mice . In mice with collagen-induced arthritis, although extrathymic T cells are higher in joints and might play an important role in the onset of the disease, these cells seem to have a suppressive effect on arthritis, because in vivo treatment of mice with anti-γδ TCR antibody significantly increased the severity of the arthritis . These findings suggest that some of the extrathymic T cell populations might expand and serve to suppress inflammation rather than to aggravate arhtritis.
This situation may be applicable to humans. In fact, CD57+ T cells in joint fluid of RA patients show an agranular lymphocyte morphology, and the proportion of CD57+ T cells within joint fluid is inversely correlated with ESR. It was reported that CD57+ T cells suppress the differentiation of haematopoietic stem cells , the proliferation of B cells , and induction of cytotoxic T lymphocytes [9, 47]. It was also reported that populations of these cells are elevated in the peripheral blood of long-term survivors after organ transplantations , including BM transplantation. It raises the possibility that these donor-derived cells suppress the immune function of other cell populations and thereby inhibit immune response in graft-versus-host disease (GVHD) after organ transplantation. Thus, T cells with NK cell markers seem to have unique immunosuppressive functions different from those of conventional T cells and NK cells.
It should also be noted that in RA patients, population of CD57+ T cells are elevated in joint-adjacent BM but not in iliac BM. This finding suggests that these cells do not increase as a result of systemic inflammation, but may rather be the result of local inflammation of the diseased joints. CD57+ T cells may differentiate in adjacent BM or joint cavities in situ and may suppress the local inflammation of RA. One might speculate that other cells, such as cytotoxic T cells or granulocytes, may be important in the pathogenesis of acute phase of RA or RA with strong clinical features. In fact, many clinicians know that active RA patients always show an increase of granulocytes in peripheral blood and joints, a fact that has recently been shown to play an important role in damaging various tissues [43, 47]. We propose that, although CD57+ T cells are certainly associated with the onset of RA, these cells may in turn suppress strong inflammation rather than lead to further deterioration in RA. Since CD57+ T cells consist of heterogeneous populations, the precise role of each population in RA remains to be elucidated in further investigations.
Finally, although T cells with NK cell markers in both humans and mice can develop extrathymically, it should be noted that thymic hormones or thymus-derived T cells are required for effective expansion of these extrathymic T cells .
This study was supported by Grants-in-Aid for Scientific Research (C) from The Ministry of Education, Science, Sports and Culture, and Grants-in-Aid from Niigata Medical Association and Tsukada Medical Science Foundation.