The chronic persistence of rheumatoid synovitis, an inflammation driven by activated T cells, macrophages, and fibroblasts causing irreversible joint damage, suggests a failure in physiologic mechanisms that down-regulate and terminate chronic immune responses. In vitro CD8+CD28CD56+ T cells tolerize APCs, prevent the priming of naive CD4+ T cells, and suppress memory CD4+ T cell responses. Therefore, we generated CD8+CD28CD56+ T cell clones from synovial tissues, expanded them in vitro, and adoptively transferred them into NOD-SCID mice engrafted with synovial tissues from patients with rheumatoid arthritis. Adoptively transferred CD8+CD28CD56+ T cells displayed strong anti-inflammatory activity. They inhibited production of IFN-γ, TNF-α, and chemokines in autologous and HLA class I-matched heterologous synovitis. Down-regulation of costimulatory ligands CD80 and CD86 on synovial fibroblasts was identified as one mechanism of immunosuppression. We propose that rheumatoid synovitis can be suppressed by cell-based immunotherapy with immunoregulatory CD8+ T cells.

Rheumatoid arthritis (RA)6 manifests through chronic inflammatory lesions that form in small and large joints. The current model holds that Ag-reactive T cells, B cells, and macrophages are recruited into the synovium sublining layer where they induce a hyperplastic reaction of resident fibroblasts, eventually causing tissue destruction and cartilage and bone invasion (1, 2, 3). The assemblage of inflammatory infiltrates in the rheumatoid lesions is a tightly organized process and results in the well-defined lymphoid architecture formations that lend stability to ongoing immune responses in the synovium (4, 5). The process of ectopic lymphoid neogenesis recapitulates pathways used in lymphoid organogenesis and optimizes Ag recognition in the synovium.

The dominant lymphocyte population in rheumatoid synovitis is CD4+ T cells; they clearly outnumber CD8+ T cells (6). The preponderance of CD4+ T cells in the disease lesions supports a model of disease-associated HLA class II molecules presenting arthritogenic Ags and promoting the chronic inflammation (7). A number of abnormalities have been described for CD4+ T cells accumulating in the joint, including slow responsiveness to Ag triggers and premature immunosenescence (8, 9, 10). Less attention has been given to CD8+ T cells, which were described originally as being preferentially localized in the intermediary zones between T cell-B cell clusters. It is now clear that synovial CD8+ T cells have several mechanisms through which they modulate the outcome of rheumatoid synovitis (11).

In a subset of rheumatoid patients, synovial CD8+ T cells express CD40L and produce IFN-γ (11, 12). These CD8+ T cells lack production of the pore-forming enzyme perforin and have a unique location in the mantle zone of ectopic germinal centers (GCs). They are critically involved in maintaining the lymphoid microstructures because their depletion leads to the loss of follicular dendritic cells (DCs), the collapse of the GCs, and failure of IgG production. In contrast to the helper function of such CD8+CD40L+IFN-γ+ T cells, CD8+ T cells can also provide inhibitory signals and suppress rheumatoid synovitis. Klimiuk et al. (13) have adoptively transferred CD8+ T cells into NOD-SCID mouse chimeras engrafted with synovial tissue and found prompt down-regulation of inflammatory cytokines. IL-16 was one critical mediator of these anti-inflammatory effects, but increased production of IL-10 and TGF-β was also found (13).

Selected CD8+ T cells reportedly function as regulatory cells in immune responses (14, 15), but it is not known whether any CD8+ T cell subsets act as anti-inflammatory regulators in RA tissue lesions. The idea that CD8+ T cells are capable of suppressing immune responses first emerged when it was found that adoptive transfer of T cells from mice tolerant to a known Ag into syngeneic hosts decreased subsequent immune responses to that same Ag (16). In vitro studies confirmed the existence of a subset of CD8+ T cells with the potential to suppress CD4+ T cell responses (17, 18). Subsequent studies have characterized human CD8+ suppressor T (Ts) cells as low responsive to TCR-mediated stimulation (19, 20). Although CD8+ Ts cells express moderate levels of intracytoplasmic IFN-γ, they fail to secrete this cytokine upon activation (21). CD8+ Ts cells are MHC class I restricted (14, 22); however, they appear to exhibit lower cytotoxicity than nonsuppressor CD8+ T cells (14, 21). Phenotypically, human CD8+ Ts cells are characterized generally as CD28, CD57high, CD94high, and CD11bhigh (20). More specific markers have yet to be determined. Mechanistic studies of CD8+ T cells suggest that various types of suppressor cells exist and that such CD8+ Ts cells are capable of using more than one mechanism to down-regulate CD4+ T cells. Specifically, CD8+ Ts cells have been reported to curb CD4+ T cell responses by decreasing the expression of costimulatory molecules on the surface of APCs (14, 22). Also, CD8+ Ts cells can decrease immune responses in a MHC class I-independent fashion through the secretion of soluble cytokines such as IL-6 (23). Finally, some CD8+ T cells mediate immunosuppressive activity through the secretion of IL-10 (21).

To address the hypothesis that rheumatoid synovitis is amenable to the suppressive effects delivered by CD8+ T cells, we determined the phenotype and function of CD8+ Ts cells in vitro and extended the findings to synovium-derived CD8+ T cell clones in an in vivo model of human rheumatoid synovitis. CD8+CD28CD56+ Ts cells isolated and expanded from synovial tissue effectively blocked rheumatoid synovitis when adoptively transferred into SCID mouse chimeras implanted with the donor tissue. One mechanism of CD8+CD28CD56+ Ts cell function appears to be a profound down-regulation of CD86 on fibroblast-like synoviocytes (FLS) both in vitro and in vivo. Thus, rheumatoid tissue lesions are responsive to the anti-inflammatory effects of adoptively transferred CD8+ Ts cells. We propose that patients with RA lack sufficient anti-inflammatory CD8+ Ts cells in the tissue lesion and that this deficiency can be overcome by cell-based immunotherapy.

Synovial tissues were obtained from patients with active rheumatoid lesions who met the 1987 American College of Rheumatology criteria for the diagnosis of RA. Peripheral blood was collected from patients with RA and healthy donors. Informed consent was obtained from all donors, and biological specimens were prepared according to the protocols approved by the Mayo Clinic Institutional Review Board.

PBMCs from healthy volunteers were purified from peripheral blood by Ficoll-Hypaque centrifugation. Alloantigen-responsive T cell lines were generated as described previously (15). Responding PBMCs (5 × 106/well) were stimulated in 24-well plates with irradiated (3000 rad) myelomonocytic THP-1 cells (2.5 × 106/well) (American Type Culture Collection) expressing HLA-A2, HLA-A9, HLA-B5, HLA-DR1, and HLA-DR2. Cells were cultured for 7 days in complete medium (RPMI 1640 supplemented with 10% FCS, 2 mM l-glutamine, and 50 μg/ml gentamicin) (Invitrogen Life Technologies). After 7 days, responding cells were collected, washed, and restimulated with the original stimulating cells for 7 additional days. On day 3, rIL-2 (Boehringer Mannheim) was added (50 U/ml), and the cultures were expanded for an additional 4 days. CD8+CD28, CD8+CD28+, and CD4+ T cells were separated via FACS using anti-CD4, anti-CD8, and anti-CD28 Abs (BD Pharmingen).

Proliferation assays were performed with cells that had undergone two or three cycles of T cell stimulation. CD4+ T cells (5 × 104 cells/well) were stimulated with equal numbers of irradiated allo-APCs in the presence of CD8+CD28 or CD8+CD28+ T cells (5 × 104 cells/well) in 96-well plates. In some experiments, T cells were stimulated with plate-bound anti-CD3 Ab (0.1 μg/ml). After 48 h of incubation, the cultures were pulsed with [3H]thymidine and collected 18 h later. [3H]Thymidine incorporation was determined by scintillation spectrometry. Alternatively, CD4+ T cells were labeled with CFSE (3 μM; Molecular Probes), and proliferation was assessed by FACS after 5 days. In some experiments, allo-APCs were preincubated with anti-MHC class I Abs (W6/32; Serotec) or control Ig before coculture with CD8 Ts cells.

Synovial tissues from RA patients were minced into small sections and cultured in complete medium with 50 U of IL-2/ml in the presence of irradiated EBV cells (10 Gy) as feeder cells. All EBV blasts were pretreated with neuraminidase and galactose oxidase. After ∼2 wk, CD8 T cells were isolated by cell sorting and cloned by limiting dilution on 10,000/well irradiated EBV blasts in 96-well round-bottom plates in the presence of 50 U/ml IL-2 (24, 25). Established T cell clones were maintained by biweekly restimulation with anti-CD3 in the presence of irradiated PBMCs, EBV blasts, and IL-2. T cell clones were evaluated for the expression of surface markers by FACS. Apoptotic cells were measured with the Annexin V apoptosis kit (BD Pharmingen), and necrotic cells were measured by staining with propidium iodide (PI).

Six- to 8-wk-old NOD.CB17-Prkdcscid/J mice (NOD-SCID) obtained from The Jackson Laboratory were anesthetized with 50 mg/kg pentobarbital (Abbott Laboratories) i.p. injection and methoxyflurane (Medical Developments Australia) inhalation. Pieces of human synovial tissue with inflammatory infiltrates were placed into a s.c. pocket on the upper dorsal midline. In this model, complete engraftment is reached within 7 days (11, 24, 26). On day 14, CD8+CD28 or CD8+CD28+ T cell clones were transferred by i.p. injection. On day 21, mice were sacrificed, and the synovial tissue grafts were explanted and embedded in OCT (Tissue-Tek; Sakura Finetek) for immunohistochemical studies. In some experiments, synovial tissue explants were shock-frozen in liquid nitrogen for RNA extraction.

Synovial tissues were embedded in OCT compound and processed for immunohistochemical staining, as previously described (11, 27). Briefly, 5-μm cryostat sections were stained with mouse mAb specific for human CD3 (1:300; UCHT1; DakoCytomation) or CD86 (1:100; BU63; DakoCytomation), treated with biotinylated rabbit anti-mouse Ig (1:300; DakoCytomation), and developed with Vectastain Elite ABC peroxidase reagents (Vector Laboratories) and 3,3′-diaminobenzidine tetrahydrochloride (DakoCytomation). Sections were counterstained with hematoxylin and mounted in VectaMount medium (Vector Laboratories).

Total RNA was extracted from synovial tissue with TRIzol (Invitrogen Life Technologies). cDNA was synthesized by avian myeloblastosis reverse transcriptase (Roche Molecular Biochemicals) and amplified by real-time PCR. Transcripts for each gene were quantified using the Mx4000 PCR instrument (Stratagene). The number of cytokine-specific transcripts was expressed as the number of cytokine transcripts per 2 × 106 β-actin transcripts. The following primer sequences were used: TNF-α (5′-CTTTGGGATCATTGCCCTGTG-3′, 5′-CGAAGTGGTGGTCTTGTTGCT-3′); IFN-γ (5′-ACCTTAAGAAATATTTTAATGC-3′, 5′-ACCGAATAATTAGTCAGCTT-3′); β-actin (5′-ATGGCCACGGCTGCTTCCAGC-3′, 5′-CATGGTGGTGCCGCCAGACAG-3′); monokine induced by IFN-γ (Mig) (5′-CCCAATTCATCCTCACTCAGTCC-3′, 5′-GCCATCCTGCCCATAACAACATC-3′); and IFN-γ-inducible protein 10 (CXCL10) (5′-CTCAGCACCATGAATCAAACTG-3′, 5′-CTCCCCTCTGGTTTTAAGGAG-3′).

To identify the phenotype of CD8+ T cells with suppressor activity and to define their mode of action, CD8+CD28 and CD8+CD28+ alloreactive T cell lines were generated by stimulating PBMCs from healthy HLA-A2 donors with irradiated HLA-A2+ myelomonocytic THP-1 cells. CD8+CD28 and CD8+CD28+ T cells, sorted from established alloreactive cell lines by FACS, were tested for their ability to inhibit the priming of autologous naive CD4+ T cells to alloantigens. CD8+CD28 Ts cells significantly decreased the response of CD4+ T cells to the allostimulator, whereas CD8+CD28+ Th cells augmented CD4+ T cell proliferation (Fig. 1). In the presence of CD8+CD28 Ts cells, the proliferation of CD4+ T cells was reduced by 70–80%. Both CD8+CD28+ and CD8+CD28 T cells had some proliferative activity when stimulated with THP-1 cells, typically at 25 and 15% of the CD4+ T cell proliferation, respectively (data not shown).

FIGURE 1.

CD8+CD28 Ts cells inhibit priming of naive CD4+ T cells in vitro. T cell lines were established by stimulating PBMCs from HLA-A2 donors with HLA-A2+ THP-1 cells. CD8+CD28 and CD8+CD28+ T cells were sorted and cocultured with equal numbers of naive CD4+ T cells from the same donor and irradiated HLA-A2+ allostimulators. The data shown are representative of three experiments. T cell proliferation was determined by [3H]thymidine incorporation in a 72-h assay. Error bars represent SD from the mean obtained from triplicate cultures. Proliferation of CD8+CD28+ Th cells and CD8+CD28 Ts cells responding to APCs in the absence of CD4+ T cells was 25 and 15%, respectively, of the CD4 T cell proliferation.

FIGURE 1.

CD8+CD28 Ts cells inhibit priming of naive CD4+ T cells in vitro. T cell lines were established by stimulating PBMCs from HLA-A2 donors with HLA-A2+ THP-1 cells. CD8+CD28 and CD8+CD28+ T cells were sorted and cocultured with equal numbers of naive CD4+ T cells from the same donor and irradiated HLA-A2+ allostimulators. The data shown are representative of three experiments. T cell proliferation was determined by [3H]thymidine incorporation in a 72-h assay. Error bars represent SD from the mean obtained from triplicate cultures. Proliferation of CD8+CD28+ Th cells and CD8+CD28 Ts cells responding to APCs in the absence of CD4+ T cells was 25 and 15%, respectively, of the CD4 T cell proliferation.

Close modal

To explore whether the inhibitory capacity of CD8+CD28 Ts cells was limited to primary responses of naive CD4+ T cells or whether it also affected memory CD4+ T cell responses, we tested the ability of CD8+CD28 Ts cells to repress the proliferation of Ag-experienced CD4+ T cells (Fig. 2). In all experiments, the addition of CD8+CD28 Ts cells suppressed the proliferation of memory CD4+ T cells to allotargets by 50–80%. To determine whether the suppressive effect of CD8+ T cells was related to the killing of CD4 T cells or APCs, we tested CD8+CD28+ and CD8+CD28 T cell-mediated cytotoxicity in cocultures containing CD4+ T cells, APCs, and CD8+CD28+ or CD8+CD28 T cells. The rate of apoptosis in the different cell populations was determined by Annexin-V staining. Necrotic cells were identified by incorporation of PI. No significant difference in the number of apoptotic CD4+ T cells was observed between cocultures with CD8+CD28+ or CD8+CD28 T cells. After coculture for 4 h, >80% of THP-1 cells were viable as determined by PI staining. The number of apoptotic annexin-positive cells was small and mostly contained in the PI-positive fraction (data not shown). Together, these data suggest that the inhibitory effect of CD8+CD28 Ts cells on Ag-specific CD4+ T cell responses cannot be explained by cytotoxic activity and the elimination of APCs or CD4+ T cells.

FIGURE 2.

CD8+CD28 Ts cells suppress memory CD4+ T cell responses in vitro. THP-1-specific T cell lines were established as described in Fig. 1. CD4+, CD8+CD28, and CD8+CD28+ T cells were isolated from these lines after two to three rounds of weekly restimulation. Purified CD8+CD28 or CD8+CD28+ T cells were cocultured with equal numbers of alloantigen-specific CD4+ T cells plus THP-1 cells. The data shown represent three of eight experiments conducted. Proliferation was determined by [3H]thymidine incorporation in a 72-h assay. Error bars represent SD from the mean obtained from triplicate cultures.

FIGURE 2.

CD8+CD28 Ts cells suppress memory CD4+ T cell responses in vitro. THP-1-specific T cell lines were established as described in Fig. 1. CD4+, CD8+CD28, and CD8+CD28+ T cells were isolated from these lines after two to three rounds of weekly restimulation. Purified CD8+CD28 or CD8+CD28+ T cells were cocultured with equal numbers of alloantigen-specific CD4+ T cells plus THP-1 cells. The data shown represent three of eight experiments conducted. Proliferation was determined by [3H]thymidine incorporation in a 72-h assay. Error bars represent SD from the mean obtained from triplicate cultures.

Close modal

To decipher how CD8+ Ts cells and CD4+ T cells communicate, we examined whether both T cells had to be autologous or whether CD8+CD28 Ts cells could inhibit heterologous CD4+ T cell reactivity. CD8+CD28 and CD8+CD28+ T cell lines were established by stimulating HLA-A2 T cells against HLA-A2+ THP-1 cells. CD4+ T cells were generated by priming T cells from HLA class II-mismatched donors against THP-1 cells. When cocultured with CD4+ T cells, CD8+CD28 Ts cells effectively diminished the proliferative responses of autologous as well as heterologous CD4+ T cells by >50% (Fig. 3). These results demonstrated that the communication between CD8+ Ts cells and effector CD4+ T cells was not restricted by HLA class I or class II molecules, as long as both cell types recognize alloantigens on a shared APC. These findings do not exclude that a direct recognition event between CD4+ and CD8+ T cells is required for the suppressive effect.

FIGURE 3.

CD8+CD28 Ts cells suppress responses of HLA class II-mismatched CD4+ T cells. Alloantigen-specific CD8+CD28, CD8+CD28+, and CD4+ T cell lines were generated from the same donor (upper panel, autologous) and from HLA class II-mismatched donors (lower panel, heterologous) by stimulation with irradiated THP-1 cells. Purified CD4+ T cells were cultured with an equal number of THP-1 cells in the absence or presence of CD8+CD28+ and CD8+CD28 T cells. Proliferation was determined by [3H]thymidine incorporation in a 72-h assay. Error bars represent SD from the mean obtained from triplicate cultures.

FIGURE 3.

CD8+CD28 Ts cells suppress responses of HLA class II-mismatched CD4+ T cells. Alloantigen-specific CD8+CD28, CD8+CD28+, and CD4+ T cell lines were generated from the same donor (upper panel, autologous) and from HLA class II-mismatched donors (lower panel, heterologous) by stimulation with irradiated THP-1 cells. Purified CD4+ T cells were cultured with an equal number of THP-1 cells in the absence or presence of CD8+CD28+ and CD8+CD28 T cells. Proliferation was determined by [3H]thymidine incorporation in a 72-h assay. Error bars represent SD from the mean obtained from triplicate cultures.

Close modal

To determine whether suppressor activity of CD8+CD28 Ts cells required TCR-mediated activation via the recognition of allo-MHC class I, blocking assays with the MHC class I-specific Ab W6/32 were performed. THP-1 cells preincubated with various concentrations of W6/32 or control Ab were cocultured with HLA-A2-specific CD8+CD28 Ts cells. Conditioned THP-1 cells were then used to stimulate CSFE-labeled allospecific CD4 T cells. MHC class I blocking abrogated suppressor activity in an Ab-dose-dependent manner. Complete restoration of CD4 T cell proliferation was seen with an Ab concentration of 20 μg/ml (Fig. 4).

FIGURE 4.

Ts cell activity requires MHC class I-restricted Ag recognition. THP-1-specific T cell lines were established as described in Figs. 1 and 2. THP-1 cells were incubated with anti-MHC class I Abs or control Ig before coculture with purified CD8+CD28 T cells. CSFE-labeled alloantigen-specific CD4+ T cells were added. Proliferative response was assessed after a 5-day culture by flow cytometry. Results are shown for an Ab concentration of 20 μg/ml, at which optimal reversal of inhibition was seen.

FIGURE 4.

Ts cell activity requires MHC class I-restricted Ag recognition. THP-1-specific T cell lines were established as described in Figs. 1 and 2. THP-1 cells were incubated with anti-MHC class I Abs or control Ig before coculture with purified CD8+CD28 T cells. CSFE-labeled alloantigen-specific CD4+ T cells were added. Proliferative response was assessed after a 5-day culture by flow cytometry. Results are shown for an Ab concentration of 20 μg/ml, at which optimal reversal of inhibition was seen.

Close modal

One mechanism through which regulatory T cells can modulate the functional activity of other T cells is the release of cytokines and other soluble factors (28, 29). A transwell system was used to explore whether CD8+CD28 Ts cells down-regulate CD4+ T cell activation via cytokine production. HLA-A2-specific CD8+ T cells and CD4+ T cells were cultured with irradiated THP-1 cells and separated by a membrane with 0.2-μm pores. Proliferation of CD4+ T cells was quantified by dilution of the cell membrane dye CFSE. The proliferative response of CD4+ T cells remained unaffected if a membrane was placed between the CD4+ and CD8+ T cells (Fig. 5 A). These experiments ruled out soluble suppressive factors as the sole underlying mechanism.

FIGURE 5.

CD8+CD28 Ts cells mediate suppression through the APCs. Allospecific CD8+CD28, CD8+CD28+, and CD4+ T cells were isolated from repetitively stimulated T cell lines. A, A total of 0.5 × 106 CFSE-labeled CD4+ T cells and 0.5 × 106 CD8+CD28 or CD8+CD28+ T cells was positioned in a transwell separated by a 0.2 μM membrane. A total of 0.5 × 106 allostimulators were added to each side of the transwell. A direct coculture of CD4+ T cells with CD8+CD28 Ts cells and APCs was included as a control. The data shown are representative of three independent experiments. B, Equal numbers of CD4+ T cells and either CD8+CD28 or CD8+CD28+ T cells from the same T cell line were activated with plate-bound (0.1 μg/ml) anti-CD3 Ab. Proliferation was determined by [3H]thymidine incorporation in a 72-h assay. Error bars represent SD from the mean obtained from triplicate cultures. C, CD86 and CD80 were determined on APCs cultured in the presence of CD8+CD28 or CD8+CD28+ T cells. MFI, mean fluorescent intensity.

FIGURE 5.

CD8+CD28 Ts cells mediate suppression through the APCs. Allospecific CD8+CD28, CD8+CD28+, and CD4+ T cells were isolated from repetitively stimulated T cell lines. A, A total of 0.5 × 106 CFSE-labeled CD4+ T cells and 0.5 × 106 CD8+CD28 or CD8+CD28+ T cells was positioned in a transwell separated by a 0.2 μM membrane. A total of 0.5 × 106 allostimulators were added to each side of the transwell. A direct coculture of CD4+ T cells with CD8+CD28 Ts cells and APCs was included as a control. The data shown are representative of three independent experiments. B, Equal numbers of CD4+ T cells and either CD8+CD28 or CD8+CD28+ T cells from the same T cell line were activated with plate-bound (0.1 μg/ml) anti-CD3 Ab. Proliferation was determined by [3H]thymidine incorporation in a 72-h assay. Error bars represent SD from the mean obtained from triplicate cultures. C, CD86 and CD80 were determined on APCs cultured in the presence of CD8+CD28 or CD8+CD28+ T cells. MFI, mean fluorescent intensity.

Close modal

If THP-1 cells were replaced by cross-linked anti-CD3 Abs as a means of T cell triggering, the inhibitory effects of CD8+CD28 Ts cells were minimal, leading only to approximately a 20% reduction in T cell proliferation (Fig. 5 B). Collectively, these results favored the interpretation that the suppressive effects of CD8+CD28 Ts cells involved Ag recognition of CD4+ and CD8+ T cells on a shared APC and that suppression was not a result of cytokine release.

Ts cells could function by tolerizing APCs, thereby reducing CD4 T cell responses. To address that possibility, we examined whether CD8+CD28 Ts cells modulate the surface expression of costimulatory molecules on APCs. Alloreactive CD8+CD28 and CD8+CD28+ T cell lines were established by repeated stimulation with THP-1 cells and tested for inhibitory activity when added to THP-1-reactive CD4+ T cells. The CD8+ T cell lines were cultured with THP-1 cells alone, and the allo-APCs were analyzed for the expression of CD80 and CD86. Within 12 h of coculture with CD8+CD28+ Th cells, expression of CD80 and CD86 increased, showing activation of the APC (Fig. 5 C). In contrast, CD8+CD28 Ts cells prevented the up-regulation of the costimulatory ligands. The allostimulatory THP-1 cells remained viable for the first 12 h and included very low frequencies of PI+ or Annexin-V+ cells (data not shown).

Induction of Ig-like transcript (ILT)-3 and ILT-4 on DCs has been implicated as a mechanism by which CD8+ T cells prevent DC maturation and indirectly reduce CD4+ T cell activation (15). This was not the case in the experimental system we used. We did not detect ILT-3 or ILT-4 on THP-1 cells; either spontaneously or when cocultured with CD8+CD28 Ts cells, THP-1 cells had no transcripts for ILT-3 or ILT-4 and was negative for surface expression by FACS (data not shown).

To identify markers suitable for positive selection of CD8+ Ts cells, we compared the profiles of cell surface molecules on CD8+ T cell lines that either displayed stimulatory or inhibitory function. The results from seven independent experiments are summarized in Table I. CD8+CD28 Ts cells were strongly positive for NK cell-associated markers CD56 and CD57 and, to a lesser extent, expressed CD94 and CD161, whereas CD8+CD28+ Th cells lacked CD56, CD57, CD94, and CD161 expression. CD154 (CD40L) was abundantly present on CD8+CD28+ Th cell lines but typically absent on CD8+CD28 Ts cell lines. The integrins CD11a and CD11b could not distinguish between CD8+ Ts and CD8+ Th cell lines. Additional comparison between purified CD8+CD28CD56+ and CD8+CD28CD56 T cells documented that most of the suppressive activity resided in the CD56+ subset (data not shown).

Table I.

Phenotypic profiles of alloantigen-specific CD8+ T cell lines with helper or suppressor activity

ActivityaSurface Marker
CD28CD154CD45ROCD56CD57CD94CD161CD69CD11aCD11bCD25CD122
Suppressor − − 
Helper − − − − 
ActivityaSurface Marker
CD28CD154CD45ROCD56CD57CD94CD161CD69CD11aCD11bCD25CD122
Suppressor − − 
Helper − − − − 
a

PBMCs were stimulated against irradiated allostimulators. CD8+CD28+ and CD8+CD28 T cell lines were isolated by FACS sorting and tested for their ability to enhance or suppress alloantigen-specific CD4+ T cell responses. Purified CD8+CD28+ and CD8+CD28 T cell lines were stained with fluorochrome-conjugated Ab and analyzed by FACS.

To evaluate potential differences in the level of activation between CD8+ Ts and CD8+ Th cell lines, surface expression of the activation marker CD69 was determined. Compared with CD8+CD28+ Th cells, CD8+CD28 Ts cells did not differ with regard to CD69 expression. However, we found a higher expression of the inducible IL-2R α-chain CD25 on CD8+CD28+ Th cells as compared with CD8+CD28 Ts cells, whereas the level of expression of the constitutively expressed IL-2R β-chain CD122 was indistinguishable. All CD8+ T cell lines, irrespective of CD28 and CD56 expression, were positive for intracellular perforin. These observations show that beyond the loss of CD28, CD8+ Ts cells have a distinct phenotype compared with nonsuppressor CD8+ T cells.

The in vitro studies confirmed that CD8+CD28 Ts cells reduced the expansion of naive and memory CD4+ T cells as well as their cytokine production. In previous reports, synovial CD8+ T cells had anti-inflammatory effects when adoptively transferred into the rheumatoid synovium (13), whereas other synovial CD8+ T cells that located in perifollicular region had proinflammatory functions (11). Thus, we hypothesized that Ts cells would be low in frequency within the heterogeneous population of synovial CD8+ T cells. Therefore, we built upon the in vitro studies and focused on CD8+ T cells that had lost CD28 expression but gained CD56 expression. To explore whether such CD8+ T cells were present and could be enriched to a therapeutically relevant population size, we analyzed the cell surface phenotypes of synovial CD8+ T cells. CD56 and CD28 were expressed on mutually exclusive populations of synovial CD8+ T cells (Fig. 6). The CD8+CD28CD56+ T cell population constituted the minority of the synovial infiltrate.

FIGURE 6.

CD28 and CD56 are expressed on mutually exclusive populations of tissue-infiltrating CD8+ T cells in rheumatoid synovitis. CD8+ T cells were isolated from synovial tissue. Expression of CD3, CD4, CD8, CD28, and CD56 was determined by multicolor FACS analysis. All CD8+ cells stained for CD3 and were negative for CD4 (left panel). Shaded areas indicate stains with isotype control Ab. CD28 and CD56 were detected on two mutually exclusive populations of CD3+CD4CD8+ T cells (right panel). The population of CD8+CD28CD56+ T cells represented the minority of the infiltrate (17% of all CD8 T cells).

FIGURE 6.

CD28 and CD56 are expressed on mutually exclusive populations of tissue-infiltrating CD8+ T cells in rheumatoid synovitis. CD8+ T cells were isolated from synovial tissue. Expression of CD3, CD4, CD8, CD28, and CD56 was determined by multicolor FACS analysis. All CD8+ cells stained for CD3 and were negative for CD4 (left panel). Shaded areas indicate stains with isotype control Ab. CD28 and CD56 were detected on two mutually exclusive populations of CD3+CD4CD8+ T cells (right panel). The population of CD8+CD28CD56+ T cells represented the minority of the infiltrate (17% of all CD8 T cells).

Close modal

We then proceeded with random cloning of CD8+ T cells derived from clinically active rheumatoid synovium. CD8+ T cell clones were phenotyped and expanded to sufficient numbers to test them for inhibitory activity in vivo. The phenotypic characteristics of CD3+CD8+ T cell clones that were isolated from tissue samples of four different patients and used for in vivo testing are summarized in Table II. CD3+CD8+ T cell clones were adoptively transferred into SCID chimeras that had been implanted with synovial tissue from the same patient. Control synovium-SCID mouse chimeras received sham injections with medium alone. Of nine adoptively transferred CD8+ T cell clones, three expressed high levels of CD28 and were also strongly positive for CD154 (Table II). These three T cell clones (BM253, BM214, and DE146) lacked surface expression of CD56. In contrast, six CD8+ T cell clones were negative for CD28 and stained positive for CD56.

Table II.

Phenotypes and functional activities of synovial CD8+ T cell clones with helper or suppressor activity in vivo

DonorClone NameClone Activity In VivoaCD28CD56CD154Perforin
BM BM253 Helper +b − 
 BM214 Helper − 
 BM288 Suppressor − − (+)c 
DE DE146 Helper − 
 DE231 Suppressor − (+) 
BD BD102 Suppressor − NDd 
 BD277 Suppressor − − (+) 
RK RK119 Suppressor − ND (+) 
 RK227 Suppressor − ND (+) 
DonorClone NameClone Activity In VivoaCD28CD56CD154Perforin
BM BM253 Helper +b − 
 BM214 Helper − 
 BM288 Suppressor − − (+)c 
DE DE146 Helper − 
 DE231 Suppressor − (+) 
BD BD102 Suppressor − NDd 
 BD277 Suppressor − − (+) 
RK RK119 Suppressor − ND (+) 
 RK227 Suppressor − ND (+) 
a

Synovium-derived T cell clones from four RA patients were stained with fluorochrome-conjugated Abs 2 wk following an in vitro cycle of stimulation. In parallel, each CD8+ T cell clone was tested for in vivo regulatory activity as described in Fig. 6 and given the designation of suppressor if transcription of IFN-γ or TNF-α declined by >50% compared with sham-treated controls.

b

+, >10-fold expression over background.

c

(+), ≤10-fold expression over background.

d

ND, Not done.

When injected into synovium-SCID mouse chimeras, all CD8+CD28+CD56 Th cell clones had proinflammatory activity. Tissue production of IFN-γ and TNF-α transcripts increased >3-fold when CD8+ T cell clone BM253 was administered (Fig. 7,A). Increased IFN-γ production was also a feature of CD8+ T cell clone DE146. Adoptive transfer of CD8+CD28CD56+ Ts cells had opposite effects and inhibited inflammatory activity in the synovial tissue. The most pronounced suppressive activity was seen for T cell clone BM288, which almost completely abrogated production of the two key cytokines, IFN-γ and TNF-α, in the rheumatoid lesion (Fig. 7 A). Clone BM288 gave optimal suppression with adoptive transfer of 8 × 106 cells, and all subsequent adoptive transfer experiments were therefore done with this cell number. CD8+CD28CD56+ Ts cell clones DE231 and BD277 were less efficacious but consistently reduced proinflammatory cytokines. It is possible that higher doses of adoptively transferred cells would have been more efficacious for these two clones.

FIGURE 7.

In vivo immunosuppressive activities of tissue-derived CD8+ T cell clones in rheumatoid synovitis. NOD-SCID mice were engrafted with rheumatoid synovium. CD8+ T cell clones were established from nonimplanted tissue and expanded. CD8+ T cells (8 × 106) from each of the T cell clones were adoptively transferred into the chimeras carrying autologous tissue via i.p. injection. As a control, animals received an injection of medium. Seven days after the adoptive transfer, tissue explants were collected. A, IFN-γ and TNF-α transcripts were quantified by real-time PCR and transcript levels were normalized to 2 × 106 β-actin copies. Error bars represent SD from the mean obtained from triplicate measurements. Representative experiments with five CD8+ T cell clones (BM253, BM 288, DE146, DE231, and BD277) are shown. The phenotype of these clones is given in Table II. B, In parallel, tissue explants were cryosectioned and stained with anti-CD3 Ab (brown). The sections were subsequently counterstained with hematoxylin. Adoptive transfer experiments with CD8+CD28 Ts cell clone BM288, but not with CD8+CD28+ Th cell clone BM253, induced a nearly complete loss of the CD3+ T cell infiltrates. Results are representative of three independent experiments using tissues from three different donors. Scale bar: 50 μm.

FIGURE 7.

In vivo immunosuppressive activities of tissue-derived CD8+ T cell clones in rheumatoid synovitis. NOD-SCID mice were engrafted with rheumatoid synovium. CD8+ T cell clones were established from nonimplanted tissue and expanded. CD8+ T cells (8 × 106) from each of the T cell clones were adoptively transferred into the chimeras carrying autologous tissue via i.p. injection. As a control, animals received an injection of medium. Seven days after the adoptive transfer, tissue explants were collected. A, IFN-γ and TNF-α transcripts were quantified by real-time PCR and transcript levels were normalized to 2 × 106 β-actin copies. Error bars represent SD from the mean obtained from triplicate measurements. Representative experiments with five CD8+ T cell clones (BM253, BM 288, DE146, DE231, and BD277) are shown. The phenotype of these clones is given in Table II. B, In parallel, tissue explants were cryosectioned and stained with anti-CD3 Ab (brown). The sections were subsequently counterstained with hematoxylin. Adoptive transfer experiments with CD8+CD28 Ts cell clone BM288, but not with CD8+CD28+ Th cell clone BM253, induced a nearly complete loss of the CD3+ T cell infiltrates. Results are representative of three independent experiments using tissues from three different donors. Scale bar: 50 μm.

Close modal

To identify the mechanism through which CD8+CD28CD56+ Ts cells down-regulated rheumatoid synovitis, tissue blocks from the synovial grafts were examined for the morphology and density of the lymphoid infiltrates. Immunohistochemical studies of synovial explants established that the transferred CD8+ T cells affected the cellularity of synovial infiltrates. After the adoptive transfer of T cell clones with inflammatory activity, such as clone BM253, the lymphoid microstructures of the synovial lesion were maintained (Fig. 7,B, middle panel). The density of tissue-infiltrating T cells, B cells, and macrophages markedly decreased after the transfer of anti-inflammatory T cells, such as BM288, and organized T-B cell clusters were no longer present (Fig. 7 B, right panel). Therefore, the reduction in IFN-γ and TNF-α transcripts could be attributed to the loss of CD3+ T cells in the synovial lesions, suggesting that CD8+CD28CD56+ Ts cells modulate T cell recruitment and retention.

The experiments in the synovium-SCID mouse chimera model showed that suppression of in vitro responses could be extended to autoreactive responses in vivo. If adoptive transfers of CD8+CD28CD56+ Ts cell clones were to develop into a therapeutic strategy, these immunoregulatory T cells would need to function in heterologous tissues from different patients. The in vitro data indicated that MHC class I-restricted recognition of the Ts cells was required but that MHC class II matching was not necessary. All patient samples from which the CD8+ T cell clones for the in vivo experiments were generated had been selected to be HLA-A2+. Therefore, in vivo adoptive transfer experiments were set up in which the biologic effects of adoptively transferred cells were assessed on synovial tissues from patients who were HLA-A2+ but expressed different MHC class II haplotypes. In these experiments, we also sought to examine whether other molecular mediators, in addition to IFN-γ and TNF-α, would be affected by the anti-inflammatory activity of the T cell clones. T cell-attracting chemokines CXCL10 and Mig, which are both known to be elevated in rheumatoid synovitis (30), were chosen as outcome parameters and were quantified in tissue extracts from explanted synovial grafts by real-time PCR.

Adoptive transfer of 8 × 106 cells of the CD8+CD28+CD56 Th cell clone BM253 caused a marked increase in the production of CXCL10 (2.8-fold compared with control) and Mig (1.8-fold compared with control) in autologous synovial tissue (Fig. 8, left panels). Similar amplification of these chemokines was also found in MHC class II-mismatched heterologous tissues (Fig. 8right panels). In contrast, transfer of the CD8+CD28CD56+ Ts cell clone BM288 resulted in marked reduction of both chemokines by 60–80% in autologous as well as heterologous tissue grafts (Fig. 8). Suppression of these chemokines approximated reduction of IFN-γ transcripts (Fig. 7 A). Similar results were obtained in adoptive transfer experiments of CD8+CD28CD56+ Ts cell clones DE231 and BD277 (data not shown). These experiments established that the anti-inflammatory activity of CD8+CD28CD56+ Ts cells was maintained in heterologous synovial tissues having different MHC class II haplotypes.

FIGURE 8.

Synovial CD8+CD28CD56+ Ts cell clones inhibit production of chemokines CXCL10 and Mig in autologous and heterologous synovitis. NOD-SCID mice were engrafted with pieces of synovial tissue derived either from the CD8+ T cell clone donor patient or HLA-A2-matched patients. Fourteen days after implantation, 8 × 106 CD8+CD28+CD56 Th cell clone BM253 and CD8+CD28CD56+ Ts cell clone BM288 were adoptively transferred by i.p. injection. Control chimeras received an injection of medium. Seven days after the adoptive transfer of T cells, synovial tissue grafts were explanted and analyzed by real-time RT-PCR. Transcript levels were adjusted to 2 × 106 β-actin copies. The data shown are representative of three independent experiments. Error bars represent SD from the mean obtained from triplicate measurements.

FIGURE 8.

Synovial CD8+CD28CD56+ Ts cell clones inhibit production of chemokines CXCL10 and Mig in autologous and heterologous synovitis. NOD-SCID mice were engrafted with pieces of synovial tissue derived either from the CD8+ T cell clone donor patient or HLA-A2-matched patients. Fourteen days after implantation, 8 × 106 CD8+CD28+CD56 Th cell clone BM253 and CD8+CD28CD56+ Ts cell clone BM288 were adoptively transferred by i.p. injection. Control chimeras received an injection of medium. Seven days after the adoptive transfer of T cells, synovial tissue grafts were explanted and analyzed by real-time RT-PCR. Transcript levels were adjusted to 2 × 106 β-actin copies. The data shown are representative of three independent experiments. Error bars represent SD from the mean obtained from triplicate measurements.

Close modal

To study whether the suppressive activity of CD8+CD28 Ts cells in rheumatoid synovitis was related to the ability of such T cells to modulate expression of costimulatory ligands on APCs, we explored whether CD8+CD28 Ts cells affect the function of APCs relevant in the rheumatoid disease process. FLS are known to have hybrid characteristics combining the growth pattern of fibroblasts and functional capabilities of macrophages. Also, they show an activated phenotype with increased expression of MHC class II, costimulatory, and adhesion molecules (31). To determine whether CD8+CD28 Ts cells affected the expression of CD80 and CD86 on FLS, alloantigen-specific T cells were generated using TNF-α-treated FLS derived from rheumatoid synovium as allo-APCs. FLS-specific CD8+CD28 or CD8+CD28+ T cells were cultured with synovial fibroblasts, and the surface expressions of CD80 and CD86 were determined by FACS analysis. CD8+CD28+ Th cells markedly up-regulated the surface density of CD80 and CD86 on the fibroblast lines; compared with controls, CD80 and CD86 expression increased by 25 and 100%, respectively (Fig. 9 A). In contrast, in the presence of CD8+CD28 Ts cells, the levels of CD80 and CD86 expression on FLS diminished greatly. Lack of CD80 on FLS has been associated previously with the induction of T cell anergy (32) and may be the mechanism through which CD8+CD28 Ts cells can ameliorate rheumatoid synovitis.

FIGURE 9.

CD8+ Ts cells down-regulate the expression of costimulatory ligands on FLS in vitro and in vivo. A, CD8+CD28+ and CD8+CD28 T cell lines were established by stimulating PBMCs with TNF-α-treated (10 ng/ml) FLS. After three rounds of stimulation, purified CD8+CD28+ and CD8+CD28 T cells were incubated with FLS for 12 h. The expression of CD80 and CD86 on FLS was evaluated via FACS. The data shown are representative of three independent experiments. MFI, mean fluorescent intensity. B, Synovial CD8+ T cell clones with helper or suppressor activity in vivo were selected and adoptively transferred into NOD-SCID mouse chimeras engrafted with autologous tissue by i.p. injection of 8 × 106 T cells. Control chimeras received an injection of medium alone. Seven days after the adoptive transfer, synovial grafts were explanted and stained with anti-CD86 Ab (brown). The results shown are from two of four independent experiments. Scale bar: 100 μm.

FIGURE 9.

CD8+ Ts cells down-regulate the expression of costimulatory ligands on FLS in vitro and in vivo. A, CD8+CD28+ and CD8+CD28 T cell lines were established by stimulating PBMCs with TNF-α-treated (10 ng/ml) FLS. After three rounds of stimulation, purified CD8+CD28+ and CD8+CD28 T cells were incubated with FLS for 12 h. The expression of CD80 and CD86 on FLS was evaluated via FACS. The data shown are representative of three independent experiments. MFI, mean fluorescent intensity. B, Synovial CD8+ T cell clones with helper or suppressor activity in vivo were selected and adoptively transferred into NOD-SCID mouse chimeras engrafted with autologous tissue by i.p. injection of 8 × 106 T cells. Control chimeras received an injection of medium alone. Seven days after the adoptive transfer, synovial grafts were explanted and stained with anti-CD86 Ab (brown). The results shown are from two of four independent experiments. Scale bar: 100 μm.

Close modal

To test whether this mechanism has relevance in vivo, CD86 expression was analyzed in synovial lesions that had been treated with adoptively transferred synovial CD8+ T cells. In sham-treated synovial grafts, CD86 expression was identical to nonimplanted synovial tissues and was detected on a subset of B cells, a subset of macrophages, and on synovial fibroblasts (Fig. 9,B, left panels; data not shown). Besides the lining layer, synoviocytes deep in the stroma were positive for CD86. Following adoptive transfer of CD8+CD28+CD56 Th cells with proinflammatory activity, CD86 expression increased markedly, including widespread expression on the fibroblasts (Fig. 9,B, middle panels). Adoptive transfer of CD8+CD28CD56+ Ts cells with anti-inflammatory activity yielded the opposite result. CD86 expression declined markedly to the extent that wide areas of the synovial lesion became negative for this costimulatory ligand (Fig. 9 B, right panels). These data confirmed that CD8+CD28CD56+ Ts cells were capable of modulating APC function in the disease lesion by abolishing the costimulatory function of tissue-residing stromal cells.

This study establishes that cell-based immunotherapy effectively blocks inflammation in rheumatoid synovial lesions. Adoptive transfer of CD8+CD28 Ts cells down-regulated the transcription of inflammatory cytokines IFN-γ and TNF-α and of the T cell-attracting chemokines CXCL10 and Mig. Selection of tissue-derived CD8+ T cell clones for cell-based immunotherapy was guided by examining the phenotype of CD8+ Ts cells that functioned as effective suppressors in an alloantigen-specific in vitro system. Expression of CD56, combined with the loss of CD28 and CD154, emerged as a useful profile for identifying CD8+ T cells with immunosuppressive activity in vivo. In vitro as well as in vivo CD8+CD28CD56+ Ts cells were highly effective in down-modulating CD86 expression on FLS. We propose that APC function of resident synoviocytes is critically important in maintaining rheumatoid inflammation. CD8+ Ts cells are capable of ameliorating rheumatoid synovitis by targeting the APC function of synoviocytes, revealing a new therapeutic approach for rheumatoid synovitis.

CD8+CD28 Ts cells were capable of inhibiting priming as well as recall responses of CD4+ T cells to alloantigen stimulation. The suppressive effects of CD8+ Ts cells on CD4+ T cells required that both of them recognize Ag on the same APCs. The CD8+ Ts cell-mediated suppression of CD4+ T cells correlated with the down-regulation of costimulatory molecules not only on DC-like myelomonocytic THP-1 cells but also on nonprofessional APCs such as FLS.

Mechanistically, our data reveal that CD8+CD28CD56+ Ts cells decrease CD4+ T cell responses through a contact-dependent manner that requires the interaction of CD8+ Ts cells, APCs, and CD4+ T cells. In the allogeneic system, the suppressive effects of CD8+ T cells did not appear to be mediated by idiotypic interactions between CD8+ Ts cells and CD4+ T cells, because CD8+CD28 Ts cells from one donor were able to suppress HLA-mismatched CD4+ T cells. Furthermore, suppression was no longer significant when CD8+CD28 Ts cells and CD4+ T cells were cultured together and stimulated with anti-CD3 Ab in an APC-free system. Taken together, these data indicate that cell-to-cell contact between the CD8+CD28 Ts cells and CD4+ T cells is insufficient to mediate significant suppression. Our data suggest that the APCs serve as a bridge between CD8+CD28 Ts cells and CD4+ T cells to eventually suppress CD4+ T cell responses. In the transwell experiments in which CD8+CD28 Ts cells and CD4+ T cells were separated by a semipermeable membrane, suppression was not observed, discrediting soluble factors as solely responsible for suppression. Thus, the suppression of CD4+ T cells was not mediated by lymphokines but required a tricellular interaction between CD8+CD28 Ts cells, CD4+ T cells, and APCs.

CD8+ Ts cells generated in this study exhibit many similarities with suppressor T cells previously described in allogeneic and xenogeneic systems (14). CD8+CD28 Ts cells, induced by multiple rounds of stimulating PBMCs against allogeneic or xenogeneic APCs, have been found to be MHC class I restricted and to inhibit the ability of CD4+ T cells to produce the T cell growth factor IL-2 (22, 33). Evidence has accumulated that the major mechanism through which such allogeneic CD8+ Ts cells suppress CD4+ T cell function is mediated by rendering APCs tolerogenic (14, 15, 22). One of the characteristics of tolerogenic APCs seems to be the inhibition of the NF-κB pathway, which subsequently reduces the transcription of costimulatory molecules (34). CD8+CD28 Ts cells primed against the allo-APC THP-1 cells resembled the Ts cells previously described in that they were HLA class I restricted, needed to be in contact with APCs, inhibited CD4+ T cell responses, and did not work through the release of cytokines.

However, there were also important dissimilarities between CD8+ Ts cells previously described and the suppressor cells generated in the current study. First, we demonstrated down-regulation of only CD80/CD86; expression of CD154 and CD40 on the APCs was not affected by CD8+CD28CD56+ Ts cells. Most importantly, we could not demonstrate induction of the inhibitory receptors ILT-3 and ILT-4; the transcriptional profile of CD8+CD28 Ts cell-treated allo-APCs has been described previously to include up-regulated ILT-3 and ILT-4 mRNA (15). In that report, blocking of ILT-3 and ILT-4 by mAbs partially reversed the suppressive effects of CD8+ Ts cells on CD4+ T cell function. In contrast, we could not demonstrate baseline or induced expression of ILT-3 or ILT-4 on THP-1 cells either by PCR or by FACS analysis. Also, RT-PCR amplification of synovial tissue extracts generated from CD8+CD28CD56+ Ts cell-treated synovial grafts could not detect ILT-3- or ILT-4-specific sequences. We conclude that alternative mechanisms of rendering APCs incapable of inducing and sustaining CD4+ Th cell activation must exist and are functional in the cell treatment system used in the current study.

CD56 emerged as a marker that was helpful in selecting CD8+ Ts cells with anti-inflammatory potential. Expression of CD56, a marker typical of NK cells, has been helpful in dissecting NK cell populations into CD56dim and CD56bright subsets (35, 36). The majority of NK cells have a CD56dim phenotype, which is associated with cytolytic function and Ab-dependent cytotoxicity. In contrast, CD56bright NK cells have been found to have immunomodulatory functions through the secretion of cytokines. It is possible that CD56 has a direct involvement in the specialized function of Ts cells. However, additional NK cell receptors that are of potential relevance were found on CD8+CD28CD56+ Ts cells, including CD57 and CD94. In the mouse, CD8+ Ts cell pathways have been found to be dependent on Qa-1 (37, 38), which is equivalent to HLA-E in humans. The known receptor for HLA-E is a complex of NKG and CD94.

CD8+ Ts cell clones were isolated from the synovial lesions and thus are part of the inflammatory infiltrates. CD8+ cells in the tissue are phenotypically and functionally heterogeneous, but almost all patients have CD8+ T cells represented in the lesions (11, 12). Depletion experiments with anti-CD8 Abs consistently decreased inflammation, suggesting that the majority of CD8+ T cells provide proinflammatory signals (11). It is believed that CD8+ T cells have a role in stabilizing ectopic GCs; their depletion causes collapse of the GC structure, loss of lymphotoxin-β production, and failure to secrete Ig (11). CD8+ T cells are also involved in supporting other functions in the lesions; in particular, the formation of new blood vessels (39). Interaction between CD8+ T cells and synovial fibroblasts regulates the production of the angioinhibitory mediator thrombospondin 2, giving CD8+ T cells control over the process of neoangiogenesis.

However, not all CD8+ T cells in the lesions amplify inflammation and support a proinflammatory environment. Selected CD8+ T cell clones that can be phenotypically defined provide negative signals and inhibit the disease process. Studies on CD8+ T cell lines established from the tissue demonstrate that CD8+CD28+ Th cells are higher in frequency than the CD28-deficient counterparts (Fig. 6). The CD8+CD28CD56+ phenotype identified CD8+ Ts cells, both in vitro as well as in vivo. Underrepresentation of Ts cells in the chronic inflammatory lesions seems logical because sufficient numbers of such cells should be capable of eliminating rheumatoid synovitis. The chronicity of the disease process clearly indicates the failure of physiologic anti-inflammatory mechanisms. Indeed, within the CD8+CD28 Ts cell subsets, CD56+ cells account for only a subpopulation (Fig. 6). Furthermore, compared with peripheral blood of patients with RA, the frequency of CD8+CD28CD56+ T cells in the synovial tissue is decreased.

In terms of developing new cell-based immunotherapies for RA, it is encouraging that even established disease lesions are responsive to CD8+ T cell-mediated suppression. In vitro assays showed that CD8 Ts cells could inhibit the proliferative response of primed CD4 T cells. Adoptive transfers into human synovium-SCID mouse chimeras demonstrated the efficacy on the rheumatoid lesion. Compared with animal models of arthritis, the SCID mouse chimera has the advantage of truly studying the rheumatoid lesion and of exploring mechanisms that are different or ill-defined in the mouse, such as CD8 suppressor populations. An obvious limitation of the model is that the synovial lesion is isolated, and therefore, the dynamics of the global disease cannot be studied. Because of the complexity of the disease and the difference of functional lymphocyte subsets in mice and humans, the efficacy of such a cell-based therapy will have to be finally determined in clinical studies.

CD8+CD28CD56+ Ts cells were effective in the tissues from HLA class II-mismatched patients, thereby permitting broader clinical application. All clones in this study were generated from patients that expressed HLA-A2, an allele that is rather frequent in the population. Theoretically, HLA-A2-restricted CD8+ Ts cells could be helpful for many patients. A more detailed understanding of the molecular mechanism through which CD8+ Ts cells mediate immunosuppressive effects could eventually lead to therapeutic interventions that do not require adoptive transfer of the cells. Data presented here indicate that a critical event in the CD8+ T cell-mediated suppression of rheumatoid synovitis lies in functionally altered APCs. Both in vivo and in vitro experimental evidence demonstrated that the costimulatory ligand CD86 was down-regulated in FLS. Other APC populations in the tissue, specifically DCs and macrophages (40, 41, 42), may also be amenable to CD8+ T cell-mediated conditioning, rendering them ineffective or tolerogenic. A number of modalities through which tolerogenic DCs can be generated have been described previously (43, 44, 45). Compared with DC-targeted therapeutic approaches, cell-based therapy with CD8+CD28CD56+ Ts cells may have the advantage that distinct cell populations with APC function are inhibited. Following the adoptive transfer of CD8+ Ts cells into the SCID chimeras, CD86 expression was down-regulated dramatically on FLS. Careful analysis of the signals exchanged between CD8+ T cells and FLS will be necessary to understand the molecular underpinnings of this potentially effective therapeutic intervention.

We thank T. Yeargin for editorial support and L. Arneson for secretarial support.

The authors have no financial conflict of interest.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1

This work was supported by National Institutes of Health Grants ROI AR 42527, ROI AI 44142, and ROI AR 41974.

6

Abbreviations used in this paper: RA, rheumatoid arthritis; GC, germinal center; Ts, suppressor T; DC, dendritic cell; FLS, fibroblast-like synoviocyte; PI, propidium iodide; Mig, monokine induced by IFN-γ; ILT, Ig-like transcript.

1
Feldmann, M..
2001
. Pathogenesis of arthritis: recent research progress.
Nat. Immunol.
2
:
771
-773.
2
McInnes, I. B..
2001
. Rheumatoid arthritis: from bench to bedside.
Rheum. Dis. Clin. N. Am.
27
:
373
-387.
3
Feldmann, M., R. N. Maini, J. Bondeson, P. Taylor, B. M. Foxwell, F. M. Brennan.
2001
. Cytokine blockade in rheumatoid arthritis.
Adv. Exp. Med. Biol.
490
:
119
-127.
4
Takemura, S., A. Braun, C. Crowson, P. J. Kurtin, R. H. Cofield, W. M. O’Fallon, J. J. Goronzy, C. M. Weyand.
2001
. Lymphoid neogenesis in rheumatoid synovitis.
J. Immunol.
167
:
1072
-1080.
5
Weyand, C. M., P. J. Kurtin, J. J. Goronzy.
2001
. Ectopic lymphoid organogenesis: a fast track for autoimmunity.
Am. J. Pathol.
159
:
787
-793.
6
Hitchon, C. A., H. S. el-Gabalawy.
2003
. The histopathology of early synovitis.
Clin. Exp. Rheumatol.
21
:(5 Suppl. 31):
S28
-S36.
7
Goronzy, J. J., C. M. Wey.
1995
. T cells in rheumatoid arthritis: paradigms and facts.
Rheum Dis. Clin. N. Am.
21
:
655
-674.
8
Cope, A. P., R. S. Liblau, X. D. Yang, M. Congia, C. Laudanna, R. D. Schreiber, L. Probert, G. Kollias, H. O. McDevitt.
1997
. Chronic tumor necrosis factor alters T cell responses by attenuating T cell receptor signaling.
J. Exp. Med.
185
:
1573
-1584.
9
Schonland, S. O., C. Lopez, T. Widmann, J. Zimmer, E. Bryl, J. J. Goronzy, C. M. Weyand.
2003
. Premature telomeric loss in rheumatoid arthritis is genetically determined and involves both myeloid and lymphoid cell lineages.
Proc. Natl. Acad. Sci. USA
100
:
13471
-13476.
10
Koetz, K., E. Bryl, K. Spickschen, W. M. O’Fallon, J. J. Goronzy, C. M. Weyand.
2000
. T cell homeostasis in patients with rheumatoid arthritis.
Proc. Natl. Acad. Sci. USA
97
:
9203
-9208.
11
Kang, Y. M., X. Zhang, U. G. Wagner, H. Yang, R. D. Beckenbaugh, P. J. Kurtin, J. J. Goronzy, C. M. Weyand.
2002
. CD8 T cells are required for the formation of ectopic germinal centers in rheumatoid synovitis.
J. Exp. Med.
195
:
1325
-1336.
12
Wagner, U. G., P. J. Kurtin, A. Wahner, M. Brackertz, D. J. Berry, J. J. Goronzy, C. M. Weyand.
1998
. The role of CD8+CD40L+ T cells in the formation of germinal centers in rheumatoid synovitis.
J. Immunol.
161
:
6390
-6397.
13
Klimiuk, P. A., J. J. Goronzy, C. M. Weyand.
1999
. IL-16 as an anti-inflammatory cytokine in rheumatoid synovitis.
J. Immunol.
162
:
4293
-4299.
14
Liu, Z., S. Tugulea, R. Cortesini, N. Suciu-Foca.
1998
. Specific suppression of T helper alloreactivity by allo-MHC class I-restricted CD8+CD28 T cells.
Int. Immunol.
10
:
775
-783.
15
Chang, C. C., R. Ciubotariu, J. S. Manavalan, J. Yuan, A. I. Colovai, F. Piazza, S. Lederman, M. Colonna, R. Cortesini, R. Dalla-Favera, N. Suciu-Foca.
2002
. Tolerization of dendritic cells by TS cells: the crucial role of inhibitory receptors ILT3 and ILT4.
Nat. Immunol.
3
:
237
-243.
16
Dorf, M. E., V. K. Kuchroo, M. Collins.
1992
. Suppressor T cells: some answers but more questions.
Immunol. Today
13
:
241
-243.
17
Cantor, H., F. W. Shen, E. A. Boyse.
1976
. Separation of helper T cells from suppressor T cells expressing different Ly components. II. Activation by antigen: after immunization, antigen-specific suppressor and helper activities are mediated by distinct T-cell subclasses.
J. Exp. Med.
143
:
1391
-1401.
18
Reinherz, E. L., S. F. Schlossman.
1980
. Current concepts in immunology: regulation of the immune response-inducer and suppressor T-lymphocyte subsets in human beings.
N. Engl. J. Med.
303
:
370
-373.
19
Azuma, M., J. H. Phillips, L. L. Lanier.
1993
. CD28 T lymphocytes: antigenic and functional properties.
J. Immunol.
150
:
1147
-1159.
20
Arosa, F. A..
2002
. CD8+CD28 T cells: certainties and uncertainties of a prevalent human T-cell subset.
Immunol. Cell Biol.
80
:
1
-13.
21
Gilliet, M., Y. J. Liu.
2002
. Generation of human CD8 T regulatory cells by CD40 ligand-activated plasmacytoid dendritic cells.
J. Exp. Med.
195
:
695
-704.
22
Jiang, S., S. Tugulea, G. Pennesi, Z. Liu, A. Mulder, S. Lederman, P. Harris, R. Cortesini, N. Suciu-Foca.
1998
. Induction of MHC-class I restricted human suppressor T cells by peptide priming in vitro.
Hum. Immunol.
59
:
690
-699.
23
Filaci, G., S. Bacilieri, M. Fravega, M. Monetti, P. Contini, M. Ghio, M. Setti, F. Puppo, F. Indiveri.
2001
. Impairment of CD8+ T suppressor cell function in patients with active systemic lupus erythematosus.
J. Immunol.
166
:
6452
-6457.
24
Takemura, S., P. A. Klimiuk, A. Braun, J. J. Goronzy, C. M. Weyand.
2001
. T cell activation in rheumatoid synovium is B cell dependent.
J. Immunol.
167
:
4710
-4718.
25
Goronzy, J. J., P. Bartz-Bazzanella, W. Hu, M. C. Jendro, D. R. Walser-Kuntz, C. M. Weyand.
1994
. Dominant clonotypes in the repertoire of peripheral CD4+ T cells in rheumatoid arthritis.
J. Clin. Invest.
94
:
2068
-2076.
26
Klimiuk, P. A., H. Yang, J. J. Goronzy, C. M. Weyand.
1999
. Production of cytokines and metalloproteinases in rheumatoid synovitis is T cell dependent.
Clin. Immunol.
90
:
65
-78.
27
Ma-Krupa, W., M. S. Jeon, S. Spoerl, T. F. Tedder, J. J. Goronzy, C. M. Weyand.
2004
. Activation of arterial wall dendritic cells and breakdown of self-tolerance in giant cell arteritis.
J. Exp. Med.
199
:
173
-183.
28
Bach, J. F., J. Francois Bach.
2003
. Regulatory T cells under scrutiny. [Published erratum in 2003 Nat. Rev. Immunol. 3: 509. Francois Bach, J. [corrected to Bach, J. F.]].
Nat. Rev. Immunol.
3
:
189
-198.
29
Shevach, E. M..
2002
. CD4+CD25+ suppressor T cells: more questions than answers.
Nat. Rev. Immunol.
2
:
389
-400.
30
Hanaoka, R., T. Kasama, M. Muramatsu, N. Yajima, F. Shiozawa, Y. Miwa, M. Negishi, H. Ide, H. Miyaoka, H. Uchida, M. Adachi.
2003
. A novel mechanism for the regulation of IFN-γ inducible protein-10 expression in rheumatoid arthritis.
Arthritis Res. Ther.
5
:
R74
-R81.
31
Mourad, W., K. Mehindate, T. J. Schall, S. R. McColl.
1992
. Engagement of major histocompatibility complex class II molecules by superantigen induces inflammatory cytokine gene expression in human rheumatoid fibroblast-like synoviocytes.
J. Exp. Med.
175
:
613
-616.
32
Corrigall, V. M., E. Solau-Gervais, G. S. Panayi.
2000
. Lack of CD80 expression by fibroblast-like synoviocytes leading to anergy in T lymphocytes.
Arthritis Rheum.
43
:
1606
-1615.
33
Liu, Z., S. Tugulea, R. Cortesini, S. Lederman, N. Suciu-Foca.
1999
. Inhibition of CD40 signaling pathway in antigen presenting cells by T suppressor cells.
Hum. Immunol.
60
:
568
-574.
34
Li, J., Z. Liu, S. Jiang, R. Cortesini, S. Lederman, N. Suciu-Foca.
1999
. T suppressor lymphocytes inhibit NF-κB-mediated transcription of CD86 gene in APC.
J. Immunol.
163
:
6386
-6392.
35
Cooper, M. A., T. A. Fehniger, M. A. Caligiuri.
2001
. The biology of human natural killer-cell subsets.
Trends Immunol.
22
:
633
-640.
36
Shibuya, A..
2003
. Development and functions of natural killer cells.
Int. J. Hematol.
78
:
1
-6.
37
Hu, D., K. Ikizawa, L. Lu, M. E. Sanchirico, M. L. Shinohara, H. Cantor.
2004
. Analysis of regulatory CD8 T cells in Qa-1-deficient mice.
Nat. Immunol.
5
:
516
-523.
38
Panoutsakopoulou, V., K. M. Huster, N. McCarty, E. Feinberg, R. Wang, K. W. Wucherpfennig, H. Cantor.
2004
. Suppression of autoimmune disease after vaccination with autoreactive T cells that express Qa-1 peptide complexes.
J. Clin. Invest.
113
:
1218
-1224.
39
Park, Y. W., Y. M. Kang, J. Butterfield, M. Detmar, J. J. Goronzy, C. M. Weyand.
2004
. Thrombospondin 2 functions as an endogenous regulator of angiogenesis and inflammation in rheumatoid arthritis.
Am. J. Pathol.
165
:
2087
-2098.
40
Aarvak, T., J. B. Natvig.
2001
. Cell-cell interactions in synovitis: antigen presenting cells and T cell interaction in rheumatoid arthritis.
Arthritis Res.
3
:
13
-17.
41
Thomas, R., C. Quinn.
1996
. Functional differentiation of dendritic cells in rheumatoid arthritis: role of CD86 in the synovium.
J. Immunol.
156
:
3074
-3086.
42
Spack, E. G..
1997
. Treatment of autoimmune diseases through manipulation of antigen presentation.
Crit. Rev. Immunol.
17
:
529
-536.
43
Hackstein, H., A. E. Morelli, A. W. Thomson.
2001
. Designer dendritic cells for tolerance induction: guided not misguided missiles.
Trends Immunol.
22
:
437
-442.
44
Steinman, R. M., D. Hawiger, M. C. Nussenzweig.
2003
. Tolerogenic dendritic cells.
Annu. Rev. Immunol.
21
:
685
-711.
45
Hackstein, H., A. W. Thomson.
2004
. Dendritic cells: emerging pharmacological targets of immunosuppressive drugs.
Nat. Rev. Immunol.
4
:
24
-34.