Mr. Michel and Dr. Turesson contributed equally to this work.
CD56-expressing T cells that have features of senescence are expanded in rheumatoid arthritis
Version of Record online: 28 DEC 2006
Copyright © 2006 by the American College of Rheumatology
Arthritis & Rheumatism
Volume 56, Issue 1, pages 43–57, January 2007
How to Cite
Michel, J. J., Turesson, C., Lemster, B., Atkins, S. R., Iclozan, C., Bongartz, T., Wasko, M. C., Matteson, E. L. and Vallejo, A. N. (2007), CD56-expressing T cells that have features of senescence are expanded in rheumatoid arthritis. Arthritis & Rheumatism, 56: 43–57. doi: 10.1002/art.22310
- Issue online: 2 JAN 2007
- Version of Record online: 28 DEC 2006
- Manuscript Accepted: 29 SEP 2006
- Manuscript Received: 8 JUN 2006
- Human Genome Sciences
- Consultancy awarded by Amgen and Centocor
- NIH. Grant Numbers: K24-AR-047578, R01-AG-022379
- David A. Oreck Research award
- L. A. Eaton Research award
- Western Pennsylvania Chapter of the Arthritis Foundation
- University of Pittsburgh Cancer and Aging Program
- The Children's Hospital of Pittsburgh Rangos Research Center is an NIH-supported facility (National Center for Research Resources grant). Grant Number: C06-RR-14489
- Top of page
- PATIENTS AND METHODS
- AUTHOR CONTRIBUTIONS
T cells deficient in CD28 expression have been implicated in the pathogenesis of rheumatoid arthritis (RA). Given that CD28-null T cells are functionally heterogeneous, we undertook this study to screen for novel receptors on these cells.
Seventy-two patients with RA (ages 35–84 years) and 53 healthy persons (32 young controls ages 19–34 years, 21 older controls ages 39–86 years) were recruited. Phenotypes and proliferative capacity of T cells from fresh leukocytes and of long-term cultures were monitored by flow cytometry. Lung biopsy specimens from patients with RA-associated interstitial pneumonitis (IP) were examined by immunohistochemistry. Receptor functionality was assessed by crosslinking bioassays.
Chronic stimulation of CD28+ T cells in vitro yielded progenies that lacked CD28 but that gained CD56. Ex vivo analysis of leukocytes from patients with extraarticular RA showed a higher frequency of CD56+,CD28-null T cells than in patients with disease confined to the joints or in healthy controls. CD56+,CD28-null T cells had nil capacity for proliferation, consistent with cellular senescence. CD56+ T cells had skewed T cell receptor (TCR) α/β-chain usage and restricted TCR third complementarity-determining region spectra. Histologic studies showed that CD56+ T cells were components of cellular infiltrates in RA-associated IP. CD56 crosslinking on T cells sufficiently induced cytokine production, although CD56/TCR coligation induced higher production levels.
Chronic activation of T cells induces counterregulation of CD28 and CD56 expression. The loss of CD28 is accompanied by the gain of CD56 that confers TCR-independent and TCR-dependent activation pathways. We propose that accumulation of CD56+ T cells in RA contributes to maladaptive immune responses and that CD56+ T cells are potential targets for therapy.
Clonal expansion of T cells is characteristic of patients with rheumatoid arthritis (RA). T cell receptor (TCR) sequences of oligoclonal cells have been reported (1–3), but common sequences among patients are scarcely indicated, and the identity of antigens driving T cell oligoclonality in RA remains elusive.
Oligoclonal T cells have unique properties. The most prominent of these is a deficiency in the expression of CD28 (4, 5), the major costimulatory receptor required to sustain activation and differentiation of T cell effector function (6). CD28 loss is generally irreversible due to transcriptional inactivation (7). It occurs in both CD4 and CD8 compartments, and the frequency of CD28-null T cells correlates with disease activity of RA (8–10). CD4+,CD28-null T cells are associated with severe extraarticular disease manifestations (8) that are known to be risk factors for excess/accelerated mortality in RA (11).
CD28-null T cells are chronically activated lymphocytes derived from the repeated stimulation of CD28+ precursors (12). They emerge more rapidly with prolonged exposure to tumor necrosis factor α (TNFα) (13). They are long-lived clonal lymphocytes with eroded telomeres, indicating an extensive replicative history consistent with cells approaching the end stages of senescence (14, 15).
CD4+,CD28-null T cells have lost their classic helper function, and subsets of CD8+,CD28-null T cells have been shown to lack cytotoxic function (for review, see ref.12). These lymphocytes are nonetheless functionally active. Certain subsets of CD4+,CD28-null T cells have autoreactive properties; some have gained perforin and are capable of lysing specific targets. Conversely, certain noncytotoxic subsets of CD8+,CD28-null T cells have gained a suppressive function that is elicited through inhibitory receptors and/or humoral factors. Other subsets of CD28-null T cells have large cytoplasmic stores of interferon-γ relative to their CD28+ counterparts. Some express the C-type lectin receptor CD161 that is known to facilitate tissue invasion. Still others also express the inflammatory chemokine receptor CCR5 (16). Clearly, CD28-null T cells are armed to invade sites of inflammation where they participate in maladaptive immune responses (17).
Because of the functional heterogeneity of CD28-null T cells, we screened for novel cell surface molecules. We examined lymphocytes from patients with RA, who are expected to display T cell phenotypes that reflect persistent immune activation in vivo (9, 18). We also examined an in vitro chronic activation system to monitor the development of senescent CD28-null T cells from polyclonal CD28+ precursors (12). Since CD28 is subject to modulation either in response to activating stimuli in vitro (19, 20) or as a consequence of persistent immune activation in vivo (9, 10), this experimental strategy is a reasonable approach to identifying receptors on chronically activated T cells. Here we document the expansion of CD28-null T cells that express CD56, the prototypical receptor of natural killer (NK) cells (21). Studies were conducted to examine functional outcomes of de novo expression of CD56 on T cells.
PATIENTS AND METHODS
- Top of page
- PATIENTS AND METHODS
- AUTHOR CONTRIBUTIONS
We recruited 72 patients with RA (median age 64 years [age range 35–84 years], female:male ratio 2.6:1, and median disease duration 11 years [disease duration range 1–55 years]) who met the 1987 revised classification criteria of the American College of Rheumatology (ACR; formerly, the American Rheumatism Association) (22). Clinical data were obtained from medical records. Except for 1 patient who was newly diagnosed as having RA, patients had established disease.
Patients were classified into 2 groups based on clinical features: limited RA for disease confined to the joints, and extraarticular RA for disease with major organ involvement in addition to the joints (Table 1). Patients with extraarticular RA had longer disease duration but did not differ significantly from patients with limited RA in age/sex distribution, rheumatoid factor (RF) positivity, and presence of joint erosions. Nine patients had a history of severe extraarticular RA (i.e., vasculitis, neuropathy, interstitial pneumonitis [IP], or scleritis) according to predefined criteria (11). All patients were being treated with steroids and/or disease-modifying antirheumatic drugs (DMARDs) such as methotrexate, hydroxychloroquine, sulfasalazine, and azathioprine. A few patients (10 of 72) were receiving combination therapy with DMARDs and TNFα blockers (etanercept, infliximab, adalimumab).
|Extraarticular RA (n = 45)†||Limited RA (n = 27)||P‡|
|No. of men/no. of women||11/34||9/18||0.41|
|Age, mean ± SD years||63.9 ± 11.5||60.7 ± 13.6||0.28|
|Disease duration, median (IQR) years||16.6 (8.9–28.2)||6.4 (2.2–11.8)||0.001|
|RF positive||39 (87)||22 (81)||0.86|
|Erosive disease||35 (78)||14 (52)||0.09|
|ESR, mean ± SD mm/hour||31.3 ± 26.0||26.8 ± 20.4||0.56|
|Currently receiving MTX||27 (60)||15 (56)||0.88|
|Currently receiving TNFα blockers||8 (18)||2 (7)||0.26|
|Currently receiving DMARDs§||38 (84)||21 (78)||0.73|
|Currently receiving steroids||27 (60)||13 (48)||0.64|
Fifty-three healthy controls (age range 19–86 years and female:male ratio 2:1) were also recruited. Since loss of CD28 expression occurs naturally with age in the absence of clinical disease (23, 24), the control subjects were classified into 2 age groups (ages 19–34 years [n = 32] and ages ≥39 years [n = 21]). The older group of healthy subjects (median age 60 years [age range 39–86 years] and female:male ratio 2:1) served as the age-matched controls for patients with RA. The age cutoff of 39 years for the older group was based on a previous study by our group (23) indicating an increase in CD4+,CD28-null T cells in vivo (at frequencies ≥2%) at this approximate age.
Subjects were recruited from the Pittsburgh, PA, and Rochester, MN, areas. All participants, including those with biopsy samples stored in tissue archives (see below), signed a written informed consent document. Clinical information and biologic specimens were deidentified and coded. Research protocols involving human subjects were approved by Institutional Review Boards of the University of Pittsburgh and the Mayo Foundation.
Collection of biologic specimens.
Blood was collected by venipuncture. Peripheral blood mononuclear cells (PBMCs) were isolated by isopyknic centrifugation over Ficoll and subjected to phenotypic analysis (see below). PBMC aliquots were also suspended in 90% fetal calf serum (FCS; Hyclone, Logan, UT) and 10% DMSO and cryopreserved for functional studies.
Open-lung biopsy specimens were selected from the Mayo Clinic tissue archives for patients with a documented diagnosis of RA-associated IP (n = 15) or idiopathic IP only (n = 21) and for normal controls (n = 6). These patients were in addition to the cohort described above. Patients with RA-associated IP met the ACR criteria for the classification of RA (22) and had a confirmed diagnosis of IP.
Medical records of patients from whom these biopsy specimens were obtained were reviewed. The age and sex distributions were similar in all 3 groups. A history of smoking was noted for 62% of patients with RA-associated IP, 38% of those with idiopathic IP, and 100% of controls. Pulmonary function measurements had yielded very similar results in patients with RA-associated IP and idiopathic IP (mean forced vital capacity [FVC] 67% versus 70%, respectively, of predicted values), but these values were lower than those in the controls (mean FVC 94% of predicted value). Fifty-four percent of patients with RA-associated IP had used steroids compared with 38% of patients with idiopathic IP. Four patients with RA-associated IP and 1 with idiopathic IP had been treated with methotrexate. The clinical course and the histopathologic picture of lung biopsy specimens were incompatible with methotrexate-induced IP in any of these cases.
Biopsy specimens were processed for histologic analysis (see below). Stained specimens were reviewed by pathologists blinded to the clinical diagnoses who verified the histologic diagnosis of IP or normal.
Phenotypes of PBMCs and cultured T cells were monitored by multicolor flow cytometry. To examine phenotypic changes that T cells undergo during their replicative lifespans, we first examined cell phenotypes in a long-term chronic activation culture system (see below). Cultures were periodically analyzed for the expression of classic T cell antigens (e.g., CD3, CD4, CD8, CD28), NK-related antigens (e.g., CD16, CD56, GL183), B cell marker (CD20), monocyte/macrophage antigen (CD14), and activation markers (e.g., CD7, CD25, CD27, CD69). Cultures were periodically examined by immunostaining with fluorochrome-conjugated antibodies (BD Biosciences, San Diego, CA). T cells were defined as CD3+,CD14−,CD16−,CD20− cells. These screening assays consistently showed progressive loss of CD28 with a concomitant induction of CD56 on cultured T cells. Subsequent phenotyping of T cells in culture and in fresh PBMCs was therefore focused on CD28 and CD56.
Raw cytometric data were acquired using the LSRII (for 6-color analysis) or the FACSCalibur (for 4-color analysis) flow cytometers (Becton Dickinson, San Jose, CA). Cytometer efficiency was ascertained for each experiment by compensation calibration using a kit (CompBeads; BD Biosciences). Analyses of cell populations were done offline using the FlowJo software (Tree Star, Ashland, OR). Cell population analyses were done by gating on live cells using forward and side scatter profiles. Height versus width or height versus area of the forward scatter signal was examined to discriminate single cells from cell doublets or aggregates. Signal intensities of antigen-specific staining were normalized against a compensation matrix of control cells singly stained with a specific antibody conjugated with each of the fluorochromes.
Cell culture and chronic activation system.
The T cell line JT is a Jurkat derivative that has low-to-nil expression of CD28 (25). In empirical studies, JT was found to express CD56 when incubated overnight with 50 ng/ml phorbol myristyl acetate (Sigma-Aldrich, St. Louis, MO), and it maintained a high expression level of CD56 for at least 2 days in the absence of phorbol myristyl acetate. JT cells were used as experimental controls for cytokine production studies (see below).
An in vitro chronic activation system was established as described previously (12). CD28+,CD3+ cells were isolated from PBMCs of healthy donors (ages 20–25 years) by a 2-step positive selection procedure using the RosetteSep and EasySep systems (StemCell Technologies, Vancouver, British Columbia, Canada). Purity was ≥98% as examined by flow cytometry. Cells were transferred to culture plates with immobilized anti-CD3 (OKT3; Janssen-Ortho, Toronto, Ontario, Canada) and cultured in RPMI 1640 (Mediatech, Herndon, VA) containing 10% FCS, 2 mML-glutamine, 50 units/ml penicillin, and 5 μg/ml streptomycin (Gibco Invitrogen, Carlsbad, CA), and supplemented with 50 ng/ml recombinant interleukin-2 (IL-2) (Proleukin; Chiron, Emeryville, CA). Cell lines were subsequently generated by stimulation with soluble OKT3, irradiated allogeneic PBMCs, and Epstein-Barr virus–transformed B-lymphoblastoid cells every 12–15 days. Culture media containing IL-2 were replenished every 3–5 days. In empirical studies, separate cultures of CD4+ and CD8+ T cells showed slower growth curves compared with mixed cultures. Subsequent cultures were therefore maintained as mixed CD4+/CD8+ cultures.
Cultures were maintained between 5 × 106 and 1 × 107 cells/ml. Cell phenotypes on days 7–10 following the last stimulation were monitored by flow cytometry as described above. Depending on the donor, senescent cells were generally produced between the fifth and the tenth stimulation cycles, at which time they were no longer able to divide as judged by 5,6-carboxyfluorescein diacetate succinimidyl ester (CFSE-DA) fluorescence assay (26).
CFSE-DA assays were done to determine the proliferative capacity of primary T cell lines, PBMCs, or freshly isolated T cells enriched for CD28 expression by the EasySep system (see above). Two million cells were labeled with 2 μM CFSE-DA (Molecular Probes, Eugene, OR) for 10 minutes at 37°C. Cells were washed and then cultured in RPMI 1640 medium supplemented with IL-2 in the presence or absence of plate-immobilized and soluble OKT3 for purified T cells and PBMCs, respectively. Cells were harvested after 5 days, washed, and stained with fluorochrome-conjugated antibodies to CD3, CD4, CD8, CD28, and CD56. Cells were permeabilized using the BD Cytofix-Cytoperm reagent (BD Biosciences) followed by incubation with 6 μM 7-aminoactinomycin D (Sigma-Aldrich) for 30 minutes at 37°C, washed, and analyzed by flow cytometry.
Measurement of T cell clonality.
Oligoclonality was assessed by flow cytometric analysis of cells immunostained with fluorochrome-conjugated antibodies to various TCR AV and BV families/segments (BD Biosciences and Beckman Coulter, Miami, FL) in addition to typical T cell markers. TCR clonality was also assessed by spectratyping procedures (27). In this case, total RNA was prepared from freshly sorted CD3+,CD56+ T cells and subjected to reverse transcription–polymerase chain reaction experiments for the amplification of the third complementarity-determining region (CDR3) sequences of the indicated TCR BV or AV.
Immunohistochemistry and computer-assisted imaging.
Paraffin-embedded sections of lung biopsy specimens were processed for histology as described previously (28). Antibodies to CD3, CD4, and CD56 (Novocastra, Newcastle-upon-Tyne, UK) and isotype controls (Dako, Glostrup, Denmark) were used according to the manufacturer's specification. Specific antibody binding was detected by immunoperoxidase reaction using a kit (Vectastain Elite ABC; Vector, Burlingame, CA) and counterstaining with 1% Mayer's hematoxylin.
Only peroxidase-conjugated antibodies were found to stain paraffin sections satisfactorily. Therefore, immunohistochemical studies were done using multiple serial sections. All biopsy specimens were analyzed for the copositioning of CD3, CD4, and CD56 staining in consecutive sections. These biopsy specimens were the same specimens analyzed in a previous study (28) in which we found increased CD4+ T cell infiltrates in lung tissues of patients with RA-associated IP. In the present study, new serial sections of tissue blocks were analyzed.
Slides were scanned using the BLISS Slide Scanner (Bacus Laboratories, Chicago, IL). The number of stained cells was quantified by one of the investigators (SRA) who was blinded to the diagnosis. The total lung tissue area was assessed by computer-assisted image analysis (28) using the IHCScore software (Bacus Laboratories), and the number of stained cells per mm2 was calculated for each slide.
Measurement of cytokine production.
CD56+ T cells used were JT cells and the short-term cell lines described above. Cells were washed extensively, and 2 × 106 cells were transferred to culture plates with 50 μg/ml immobilized anti-CD56 (C218; Beckman Coulter) or an IgG isotype control combined with varying concentrations (0–50 μg/ml) of OKT3. Plate immobilization of antibodies was carried out as described previously (29). Culture supernatants were harvested after 24 hours, and cytokine content was analyzed by multiplex analysis using the Luminex system (Luminex, Austin, TX). All assays for cytokine production were carried out in cultures without exogenous IL-2.
Quantitative data were analyzed nonparametrically at a nondirectional alpha level of 0.05 using SigmaStat software (SPSS, Chicago, IL) or StatsDirect software (StatsDirect, Cheshire, UK). T cell phenotypes of the entire population cohort were examined in 4 groups: young controls, older controls, patients with limited RA, and patients with extraarticular RA. The statistical significance of values for the entire cohort was determined by Kruskal-Wallis ranked analysis of variance (ANOVA) where P values less than or equal to 0.01 were considered significant. Comparison between any 2 groups was performed with Mann-Whitney U test, chi-square test, Dunn's method, Student's t-test, or Conover-Inman method, as appropriate, and P values less than 0.05 were considered significant. Correlations between clinical variables and cell frequencies were assessed using the Spearman rank test. Kinetics of in vitro cytokine production were examined by regression analysis.
- Top of page
- PATIENTS AND METHODS
- AUTHOR CONTRIBUTIONS
Loss of CD28 and gain of CD56 on chronically activated T cells.
CD28 is normally subject to down-modulation by activating signals (19, 20). We therefore established an in vitro chronic activation system to examine whether the irreversible loss of CD28 in human T cells (12) is accompanied by the gain of other receptors. This culture system involved repeated stimulation of CD28+,CD56−,CD3+ cells with anti-CD3 and irradiated allogeneic feeder cells in the presence of exogenous IL-2.
Figure 1A illustrates a representative experiment for the generation of a chronically activated T cell line from a young healthy donor. At approximately the fifth stimulation cycle, the majority of the cultured T cells lost expression of CD28 but acquired CD56. In the CD8 compartment, ∼90% were CD28-null, with ∼51% being CD28-null,CD56+. In the CD4 compartment, ∼72% were CD28-null, with ∼26% being CD28-null,CD56+. This differential pattern of CD28 loss and CD56 gain between CD4+ and CD8+ T cells was observed in 5 independent experiments. Although T cell lines derived from different donors showed variation in the numbers of CD56+,CD28-null T cells, we consistently found this subset in a larger proportion in CD8+ cells than in CD4+ cells. For the small subset of T cells that had retained CD28, CD8+ cells also showed significantly lower levels of CD28 expression than CD4+ cells.
In all experiments, we observed up to 4 subsets of CD4+ and CD8+ T cells with varying levels of CD28 and CD56 expression (i.e., CD28+,CD56−, CD28+,CD56+, CD28-null,CD56−, and CD28-null, CD56+). The time of emergence and proportions of these subsets varied between cell lines (donors) and the number of stimulation cycle. Irrespective of individual variation, this chronic activation system showed the development of CD28-null,CD56+ T cells from CD28+,CD56− precursors, with CD28+,CD56+ and CD28-null,CD56− cells likely to be transitional subsets. In 5 experiments, high frequencies of CD28-null,CD56+, CD8+, and CD4+ T cells were invariably found beginning at the fifth and continuing until the tenth stimulation cycle, at which time there was no significant population doubling (data not shown) as judged by routine trypan blue exclusion staining.
The irreversible loss of CD28 expression is a feature of T cells in advanced stages of senescence (12), indicated by the limitation or cessation of cell division (26). Therefore, proliferation of T cell subsets in the chronic activation system was assessed by the CFSE-DA assay, in which diminishing fluorescence signals of CFSE-DA–loaded cells measures the extent of proliferation. We compared fresh PBMCs from young healthy persons (ages 19–25 years) and primary T cell lines at the fifth stimulation cycle, the stage at which we found varying levels of CD28 and CD56 expression on T cells. Results of 4 independent experiments showed varying peaks of CFSE-DA fluorescence of CD3+ T cells, typical of unsynchronized cell populations undergoing cell division (Figure 1B). As expected, CD28+,CD56− T cells in CFSE-DA–loaded PBMCs incubated in OKT3 yielded fluorescence profiles dominated by low-intensity CFSE-DA peaks indicative of active proliferation. Phenotypic analysis of these actively dividing progeny showed CD4+ and CD8+ T cells that had varying levels of CD28. Progeny that either lacked CD28 or gained CD56 were not detected.
In contrast, primary T cell lines (Figure 1B) showed CFSE-DA fluorescence profiles consisting of higher intensity peaks, with a single peak of the highest CFSE-DA fluorescence representing a subset of cells that had undergone limited or nil cell division. Phenotypic analysis of cells in this highest CFSE-DA peak showed predominance of CD56+ cells and a significantly smaller proportion of CD28low cells; ∼71% of these cells were CD56+,CD28-null. Depending on the cell line, this CFSE-DAhigh,CD56+,CD28-null fraction consisted of ≥75% CD8+ T cells and ≥15% CD4+ T cells. In stark contrast, cells in the region of low CFSE-DA fluorescence, representing dividing cells, were predominantly the transitional subset CD28low/null,CD56− (∼81%), with ∼28% being CD28low,CD56−. Only ∼19% of CFSE-DAlow cells were CD28-null,CD56+ T cells. The total dividing CFSE-DAlow cells were mostly CD4+ cells (≥80%), with a smaller proportion of CD8+ cells (≥10%).
Loss of CD28 and gain of CD56 on T cells from patients with RA.
Ex vivo phenotypic analysis of PBMCs from a cohort of patients and healthy persons was conducted. Patients were stratified into 2 groups, namely, those with limited RA and those with extraarticular RA, based on diagnoses of extraarticular disease (Table 1). Results showed that all patients with RA had discrete populations of CD28+,CD56−, CD28-null,CD56+, and CD28+,CD56+ T cells in vivo (data not shown), reminiscent of chronically stimulated T cells in vitro (Figure 1A). Since the dynamic of the temporal regulation for CD28 loss and CD56 gain is unknown at this time, the frequencies of CD28-null, CD28+, and CD56+ T cells were independently determined.
Figure 2A shows that all patients had an overall increased frequency of CD4+,CD28-null T cells compared with healthy controls. Statistical analysis of the median frequencies of these cells showed highly significant differences among the 4 groups of the entire cohort (P < 0.001 by ranked ANOVA). However, comparison of median frequencies between patients with limited RA and healthy controls ages ≥39 years did not yield a statistically significant difference (P ≥ 0.05 by Dunn's method and Mann-Whitney U test). In contrast, patients with extraarticular RA had a significantly higher frequency of CD4+,CD28-null T cells than either those with limited RA or controls ages ≥39 years (P = 0.009), consistent with previous reports (8).
Analysis of a second control group consisting of younger subjects (ages 19–34 years) showed a significantly lower frequency of CD4+,CD28-null T cells, and a corresponding higher frequency of CD4+,CD28+ T cells, than either the older control group (ages ≥39 years) or the RA patient groups (P ≤ 0.008). This corroborates previous data (23) indicating the influence of chronologic age on CD28 loss in CD4+ T cells.
In the CD8 compartment, there was broad distribution of CD28-null cells in all subjects. Overall median frequencies of CD8+,CD28-null T cells among patients and controls were statistically different (P < 0.001 by ranked ANOVA). However, comparison of median frequencies showed no significant difference between healthy controls ages ≥39 years and either RA patient group (P ≥ 0.07 by Dunn's method and Mann-Whitney U test). In contrast, comparison between the patient groups showed a higher frequency of CD8+,CD28-null T cells in patients with extraarticular RA than in those with limited RA (P < 0.05). Consistent with previous reports (24), younger healthy persons (ages 19–34 years) had a significantly lower frequency of CD8+,CD28-null T cells than older healthy persons ages ≥39 years (P ≤ 0.008). The latter finding also indicates an influence of age on CD28 loss in CD8+ T cells.
Among patients, frequencies of CD8+,CD28-null and CD4+,CD28-null T cells were unrelated to age, disease duration, RF positivity, erosive disease, and treatment. There was also no significant difference in the frequency of these cells between patients with a history of severe extraarticular RA (n = 9) and the rest of the patients with extraarticular RA (data not shown).
The frequency of CD56+ T cells was similarly examined. Results showed a higher overall median frequency among all RA patients than among healthy persons (P < 0.001 by ranked ANOVA) (Figure 2A). CD56+ T cells were highly abundant in both CD4 and CD8 compartments.
Among RA patients, CD56+ T cells were generally CD28-null. The proportion of total CD56+,CD28-null T cells was higher in patients with extraarticular RA (means of 22.67% and 45.28% for CD4 and CD8, respectively) than in those with limited RA (means of 6.9% and 31.9% for CD4 and CD8, respectively) (P < 0.05 for both). In both patient groups, the number of CD56+,CD28-null cells was higher for CD8+ T cells than for CD4+ T cells. This dichotomous pattern of CD28 and CD56 expression in vivo between CD4+ and CD8+ T cells was reminiscent of that seen in the in vitro chronic activation system (Figure 1A).
The frequencies of CD4+,CD56+ (r = −0.28, P = 0.02) and CD8+,CD56+ (r = −0.21, P = 0.11) T cells tended to be negatively correlated with disease duration, although the latter subset did not reach statistical significance (Figure 2B). Expansion of CD56+ T cells was not associated with RF positivity, erosive disease, or medication (data not shown).
There was also a differential frequency of CD56+ T cells among healthy controls (Figure 2A). Although controls showed a lower frequency of these cells than patients with RA, healthy persons ages 19–34 years displayed an even lower frequency of CD56+ T cells compared with healthy persons ages ≥39 years (P < 0.001 by Mann-Whitney U test).
Down-regulation of CD4 and CD8 in patients with RA.
We observed that T cells of patients with RA showed discrete populations that were either dimly or brightly staining for CD4 and CD8 coreceptors (Figure 2C). CD4dim and CD8dim T cells were generally CD28- null,CD56+, although CD8dim cells tended to have variable levels of CD28 and CD56. Approximately 65% of patients had discrete subsets of these cells (Figure 2D). The 2 patient groups had significantly more CD4dim T cells compared with healthy older controls (P = 0.002), with a similar trend for CD8dim T cells. Such cells were not found in the younger control group.
Patients with extraarticular RA tended to have a higher frequency of CD4dim T cells than patients with limited RA, with a broad distribution and very high frequencies in some patients (P = 0.097). All patients also tended to have higher frequencies of CD8dim T cells. Both in controls and in patients, larger proportions of CD8dim T cells were found relative to CD4dim T cells. These T cell subsets were found among healthy persons ages ≥39 years, but were absent in healthy persons younger than 35 years (Figure 2D). There was no apparent relationship between the frequency of CD4dim T cells or CD8dim T cells and either age or sex of RA patients.
CD56+ T cells are oligoclonal lymphocytes.
Oligoclonal CD28-null T cells are a hallmark of RA (4, 8). We therefore verified whether these same cells expressing CD56 were clonal by measuring their relative representation in 4 TCRα/β families: AV12, BV2, BV6S7, and BV13S3. These TCR families are the most common TCRs, each one comprising ∼1.5% of normal circulating T cells (30, 31). To ensure high numbers of cells for TCR analysis, patients (see Figure 2A) who had ≥50% CD56+ T cells were chosen.
Figure 3A shows the results from 3 representative patients who had high frequencies of each of the 4 TCRs. There were only 2 cases (BV2 and AV12 for patients 2 and 3, respectively) in which the specific TCRα/β chain remained unexpanded at a frequency of ≤1.5%. More significantly, CD8+,CD56+ T cells tended to have a higher degree of representation of specific TCR AV and BV chains than CD4+,CD56+ T cells (Figure 3B). Patients had varying patterns of TCR expansion, consistent with studies indicating that T cell oligoclonality in RA may not be associated with any particular antigen (1–5).
Oligoclonality was further evaluated by CDR3 spectratyping of sorted T cells from PBMCs. Results showed that CD56−,CD28+ T cells consistently showed a Gaussian distribution of CDR3 lengths (Figure 3C). In contrast, CD56+ T cells showed varying degrees of restriction of CDR3 spectra, with dominance of between 1 and 3 sizes of CDR3.
Reduced proliferative capacity of CD56+ T cells.
The in vitro chronic activation system showed that gain of CD56 was associated with nil capacity for proliferation (Figure 1B), a characteristic of cells in advanced stages of senescence (26, 32). We therefore evaluated the proliferative capacity of CD56+ T cells derived directly from patients.
Three independent experiments on 9 randomly selected PBMC samples were carried out using 2 types of T cell stimulation, either by OKT3 or by irradiated allogeneic cells. Figure 4A shows results from CFSE-DA–loaded PBMCs incubated with OKT3 or IgG. CD3+ cells in PBMCs incubated with IgG had a single peak of high CFSE-DA fluorescence indicating lack of proliferation. In contrast, CD3+ cells in OKT3-stimulated PBMCs had a CFSE-DA fluorescence profile consisting of multiple peaks of low fluorescence, representing actively dividing cells, and one high-fluorescence peak representing cells that underwent little or nil proliferation.
Figure 4B shows the CFSE-DA profiles of subsets of OKT3-activated T cells that were distinguished by their expression of CD56 and CD28. The CD28+,CD56− subset (box c) were predominantly CFSE-DAlow cells, indicating that they were actively dividing. In contrast, the CD28-null,CD56+ subset (box a) were predominantly CFSE-DAhigh cells, indicating that they underwent little or nil proliferation. Whereas the former subset (box c) had multiple peaks of low-intensity CFSE-DA, the latter subset (box a) had a single peak of high-intensity CFSE-DA. In addition, a small proportion of T cells were CD56+,CD28+ (box b). These cells were in various stages of activation and cell division as indicated by multiple peaks of low CFSE-DA staining. Similar results (data not shown) were found in experiments in which stimulation was carried out with allogeneic feeders without OKT3.
CD56+ T cells in rheumatoid lesions.
Excess mortality in RA is associated with extraarticular RA manifestations such as IP (33). We previously reported higher levels of CD4+ T cell infiltration in the lungs of patients with RA-associated IP than in those of patients with idiopathic IP only (28). We therefore examined the same cohort of patients with lung disease and assessed whether CD56+ T cells were also components of the lung cellular infiltrates.
Figure 5A shows a typical immunostaining of CD4 and CD56 expression in lung tissues of a patient with RA-associated IP. Analysis of serial sections showed copositioning of CD4+ and CD56+ cells in the same biopsy specimen. These cells further colocalized with CD3-stained cells in a consecutive third serial section (results not shown). CD56+ cells were found in significantly greater numbers in the lungs of patients with RA-associated IP than in those of normal controls (Figure 5B). Moreover, the frequency of these cells was also higher in patients with RA-associated IP than in those with idiopathic IP only. We had previously examined these same tissue blocks and reported that CD16+,CD4−,CD3− NK cells were rarely found in the lung tissues (28). The present data indicate that the serially positioned CD4+,CD56+,CD3+ cells in the lung lesions are T cells.
The frequency of lung-infiltrating CD56+ T cells in RA-associated IP and in idiopathic IP was not related to sex, pulmonary function, or medication (data not shown). CD56+ cells were rarely seen in smokers with normal lungs, suggesting that they may not be directly induced by smoking.
Functionality of CD56 on T cells.
To examine the role of CD56 in T cells, activation studies were carried out. Results showed that CD56 ligation by itself was sufficient to induce production of IL-2 (400 pg/ml), TNFα (180 pg/ml), and macrophage inflammatory protein 1β (MIP-1β) (1,100 pg/ml) by CD56+ T cells (Figure 5C). CD3 ligation alone induced significantly lower levels of cytokine production. At the indicated increasing concentrations of OKT3 without anti-CD56, the ranges of cytokine production were 15.3–102.4 pg/ml IL-2, 10.5–67.0 pg/ml TNFα, and 125.8–680.0 pg/ml MIP-1β.
Interestingly, coligation of CD56 and CD3 resulted in higher magnitudes of cytokine production. Cells stimulated with a constant amount of anti-CD56 and increasing amounts of OKT3 showed 1.5–3.5-fold and 1.5–5.0-fold increased production of IL-2 and MIP-1β, respectively, relative to the levels of production by cells stimulated with anti-CD56 alone. However, TNFα production with CD3/CD56 coligation was increased only modestly (0.8–1.8-fold) over that seen with CD56 ligation alone. CD56 ligation, with or without CD3 ligation, did not induce IL-4 and MIP-1α production (data not shown). These results were consistently found in 6 independent bioassays.
- Top of page
- PATIENTS AND METHODS
- AUTHOR CONTRIBUTIONS
The data presented here show that a high frequency of CD4+,CD28-null T cells is characteristic of patients with extraarticular RA (Figure 2), which is consistent with findings of previous studies (8, 9). Patients with disease limited to the joints have a lower frequency, at levels equivalent to those seen among healthy age-matched persons. Furthermore, patients with extraarticular RA tend to have higher levels of CD8+,CD28-null T cells. These results are predictable, since CD28 is highly sensitive to down-modulation following cellular activation (19, 20). Extraarticular RA confers an increased burden of inflammation, with overexpression of endothelial HLA–DQ and IL-1, and with high serum levels of TNFα (34, 35) known to directly inhibit CD28 gene transcription (13, 25). Systemic inflammation and endothelial activation suggest an environment permissive of chronic T cell activation. These probably explain the overrepresentation of CD28-null T cells in the lymphocyte repertoire of patients with extraarticular RA relative to patients with limited RA. Similar populations of CD28-null T cells have been reported in other chronic diseases, such as Wegener's granulomatosis, ankylosing spondylitis, and multiple sclerosis (for review, see ref.17), in which there is underlying persistent immune activation.
We have been interested in identifying receptors that become expressed concomitant with the loss of CD28. Previous studies suggested a role for killer cell immunoglobulin-like receptors (KIRs), a family of highly polymorphic NK-related receptors (26, 36). One of these, KIR2DS2, is thought to be a risk factor for RA-associated vasculitis (37). Since KIR2DS2 is rarely represented in the general population (38), we focused on the expression of monomorphic receptors. Our data (Figure 2A) show de novo expression of CD56, the prototypic NK receptor (21), on T cells of all RA patients. A high frequency of CD56+ T cells is characteristic of patients with or without extraarticular RA. Acquisition of CD56 occurs in both CD4 and CD8 compartments. These findings suggest that abundance of CD56+ T cells in vivo is a more universal indicator of T cell repertoire perturbation than the simple loss of CD28 expression, although increased frequency of CD4+,CD28-null,CD56+ T cells is clearly associated with extraarticular RA.
CD28 loss and CD56 gain in T cells is probably a consequence of chronic immune activation. We and others have reported that CD28-null T cells are derived from the repeated stimulation of CD28+ precursors (12) and that their production can be accelerated by TNFα (13). Here we present evidence for the de novo acquisition of CD56 concomitant with the loss of CD28, a constitutively expressed receptor, during chronic activation of T cells (Figure 1).
The rates of CD28 loss and CD56 gain in vitro were faster for CD8+ T cells than for CD4+ T cells (Figure 1A). Similarly, the frequency of CD28-null and CD56+ T cells in vivo was greater for CD8+ T cells (Figure 2A). At any given time, a larger proportion of CD4+ T cells coexpressed low levels of CD28 and CD56. Although the kinetics of CD28 loss and CD56 gain have yet to be ascertained, our experimental approach suggests that CD28+/low,CD56+/low cells represent a transitional subset in the development of CD28-null,CD56+ T cells from CD28+,CD56− precursors. The molecular basis for the differential rates of CD28 down-regulation and CD56 up-regulation between CD4+ and CD8+ T cells is unknown. However, CD8+ T cells are known for their more rapid cell division and faster extinction of telomerase activity (39). Whether the counterregulation of CD28 and CD56 is coupled to cell division needs to be examined. While the independent regulation of CD56 induction in T cells also needs to be investigated, it has been reported that the irreversible loss of CD28 is due to transcriptional inactivation (7).
The notion that CD28-null,CD56+ T cells in RA are a consequence of chronic activation is further supported by their clonal characteristics. CD56+ T cells comprise a large proportion of at least 4 TCR families (AV12, BV2, BV6S7, and BV13S3) in vivo (Figure 3). Clonality of CD56+ T cells extends to the level of the CDR3, a region of the TCR that defines antigenic specificity. Whether a particular antigen(s) and/or defects in apoptosis are responsible for the clonal accumulation of CD56+ T cells is unknown. However, it has been reported that oligoclonality of CD28-null T cells is associated with resistance to cell death due to the overexpression of the antiapoptotic molecules (14).
T cell oligoclonality in RA has been shown to profoundly restrict diversity of the immune repertoire of patients (40). Whereas T cell diversity is critical to normal host defense, T cell repertoire contraction is thought to underlie the immunocompromised phenotype of patients with RA (41), possibly leading to the increased risk of infection (42) and increase in other comorbidities including lymphoma and atherosclerotic heart disease (43, 44). The degree to which clonal expansion of CD56+ T cells is associated with specific comorbid conditions in RA remains to be examined.
The physiologic environment in RA is unequivocally proinflammatory. Dysregulation of TNFα in RA is well documented (45). Here we report down-regulation of CD4 and CD8 on T cells in a majority of patients (Figures 2B and C). Although the functional impact of CD4/CD8 down-regulation is unknown, our data show that CD4dim and CD8dim T cells are predominantly CD56+ and/or CD28-null. CD4 and CD8 are normally subject to down-modulation (46), suggesting that CD4dim and CD8dim CD56+,CD28-null T cells in RA probably represent chronically activated cells.
CD28-null T cells have shortened telomeres and reduced telomerase activity (15, 39). Here we report that CD28-null,CD56+ T cells have nil capacity for proliferation, consistent with cells that are approaching the end stages of senescence (32). CD56+ T cell lines generated in vitro (Figure 1B) or fresh CD56+ T cells (Figure 4) exhibit nil cell division. Our data show that nondividing CFSE-DAhigh cells are predominantly CD56+,CD28-null T cells, whereas actively dividing CFSE-DAlow cells are generally CD28+ T cells with varying levels of CD56 expression.
Our data also show the influence of chronologic age on the in vivo accumulation of CD56+ T cells (Figure 2A). Consistent with previous reports indicating increased numbers of CD56+ T cells in elderly persons (particularly centenarians) (47), our data show that healthy persons ages ≥39 years have a significantly higher frequency of these cells than younger individuals. In contrast, there is an even higher level of in vivo accumulation of CD56+ T cells among RA patients, which is disproportionate with age. Similar age-related in vivo accumulation of CD28-null T cells has been reported (23, 24), whereas RA patients have even higher frequencies irrespective of age (8, 9) (Figure 2A). These findings support the idea of accelerated senescence in RA (17), with CD28-null,CD56+ T cells being prototypic senescent or presenescent lymphocytes.
CD56+ T cells contribute to the pathogenesis of RA. Our data show that they populate rheumatoid lesions outside the joint, such as in the lung (Figures 5A and B). We previously reported that the lungs of patients with RA-associated IP are characterized by infiltration of CD4+ T cells (28). Our present data further show that such lung-infiltrating cells are also CD56+. They are found in larger numbers in patients with RA-associated IP than in those with idiopathic IP despite more extensive immunosuppressive treatment in the group with RA-associated IP. The majority of the patients with RA-associated IP were smokers. Due to the reported role of smoking in RA pathogenesis (48), lung abnormalities in patients with RA are of interest. However, our study did not show any significant infiltration of CD56+ cells in smokers with otherwise normal lungs. More definitive correlations in RA-associated IP between smoking habit, medication, and the frequency of lung-infiltrating CD56+ T cells will require analysis of a larger patient cohort. In addition, the relationship between T cell subsets in the blood and in the lungs requires further study.
The mechanism(s) by which infiltrating CD56+ T cells evoke injury in the lung remains to be elucidated. In a recent study (49), investigators reported the isolation of CD8+,CD56+ T cell clones from the rheumatoid synovium. In a human synovium–engrafted SCID mouse model, infusion of these cells suppressed CD4+ T cell–driven responses in the implanted synovial tissue. Suppression was indirect through the inhibition of antigen-presenting cells that interact with both CD8+,CD56+ suppressors and CD4+ responders. The functional impact of CD56 in this situation was not studied. Whether a similar suppression mechanism operates specifically in RA-associated IP, or in the general setting of extraarticular RA, remains to be examined. Nonetheless, our data showing CD56+ T cells in lung lesions support the notion that RA-associated IP is T cell mediated (28). This suggests that RA-associated IP may require different management from idiopathic IP (50).
Our data show that CD56 is a functionally competent receptor on T cells. CD56 ligation is sufficient to induce production of IL-2, TNFα, and MIP-1β (Figure 5C). IL-2 is the key cytokine for T cell survival (51). MIP-1β is an inflammatory chemokine that recruits macrophages and granulocytes, both of which are involved in the establishment of rheumatoid lesions (52, 53). TNFα is the dominant pathogenic effector cytokine in RA (45). Production levels of these cytokines are further increased by coligation of CD56 and the TCR–CD3 complex. These findings indicate that CD56 can function either as an independent stimulatory receptor or as a costimulatory receptor. Whether the dual function of CD56 involves distinct signaling pathways is unknown. Nevertheless, this functional versatility of CD56 suggests an intriguing notion that CD56+ T cells are potentially more pathogenic than their CD56− counterparts, and are candidate targets for therapy. Because CD56 by itself can signal TNFα production, it is tantalizing to speculate about the clinical benefits of either the selective elimination of CD56+ T cells or the specific disruption of the TCR-independent signaling triggered by CD56. The latter will depend on the elucidation of the pertinent signaling pathway.
Whether CD56 triggers a distinct signaling pathway in NK cells is unknown. Its function in NK cells is also unknown. Although CD56 is an adhesion molecule and a biomarker for cytotoxic NK cells (21), cognate interactions between NK cells or between NK cells and their targets are not mediated by CD56 (54). To our knowledge, our data showing CD56-mediated cytokine production are the first documentation for the functionality of CD56 in cells of the immune system.
In summary, this work documents 3 aspects of T cell biology in the setting of RA. First, the physiologic inflammatory environment of RA is clearly permissive of the pervasive oligoclonal expansion of T cells that lack CD28, the major T cell costimulatory receptor, but that have acquired CD56, the prototypic NK receptor. Second, the inverse expression of CD28 and CD56 on T cells is probably related to the acceleration of cell senescence as a consequence of chronic activation, with the senescent or presenescent state of CD56+ T cells being indicated by their nil capacity for cell division. Third, the de novo expression of CD56 on oligoclonal T cells confers TCR-independent and TCR-dependent immune effector pathways that could contribute to immune anomalies in RA.
- Top of page
- PATIENTS AND METHODS
- AUTHOR CONTRIBUTIONS
Dr. Vallejo 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. Dr. Vallejo.
Acquisition of data. Mr. Michel, Drs. Turesson, Iclozan, Bongartz, Wasko, Matteson, and Vallejo, Ms Lemster, and Ms Atkins.
Analysis and interpretation of data. Mr. Michel, Drs. Turesson, Iclozan, Bongartz, Wasko, Matteson, and Vallejo, Ms Lemster, and Ms Atkins.
Manuscript preparation. Mr. Michel, Ms Lemster, and Drs. Matteson and Vallejo.
Statistical analysis. Mr. Michel, Drs. Turesson and Vallejo, and Ms Lemster.
Patient recruitment. Mr. Michel and Drs. Wasko, Matteson, and Vallejo.
Institutional Review Board protocol. Drs. Wasko, Matteson, and Vallejo.
Review of medical records. Dr. Turesson.
Oversight and direction of study. Dr. Vallejo.
- Top of page
- PATIENTS AND METHODS
- AUTHOR CONTRIBUTIONS
We thank Jane Jacquith and Lindy Grone Kilby for assistance in patient recruitment and sample collection, Drs. Henry Tazelaar, Jeffrey Myers, Jay Ryu, and Thomas Colby for providing lung biopsy specimens and assisting in the classification of lung disease, and Linda Murphy and Darren Riehle for assistance in processing of tissue specimens and in computer-assisted imaging.
- Top of page
- PATIENTS AND METHODS
- AUTHOR CONTRIBUTIONS
- 31The leucocyte antigen facts book. 2nd ed. London: Academic Press; 1997., , , , , .
- 34Tumor necrosis factor α, its soluble receptor I, and −308 gene promoter polymorphism in patients with rheumatoid arthritis with or without amyloidosis: implications for the pathogenesis of nephropathy and anemia of chronic disease in reactive amyloidosis. Arthritis Rheum 2003; 48: 3068–76., , , , , .
- 48A new model for an etiology of rheumatoid arthritis: smoking may trigger HLA–DR (shared epitope)– restricted immune reactions to autoantigens modified by citrullination. Arthritis Rheum 2006; 54: 38–46., , , , , , et al.