Abstract
Objective. To measure the frequencies of IL-4+ CD8+ T cells from patients with AS and RA, and to assess their clinical relevance and properties.
Methods. Peripheral blood (PB) and clinical data were obtained from 37 AS, 36 RA patients and 37 healthy controls. We also generated IL-4-producing CD8+ T cell lines and clones by co-culture with autologous dendritic cells. Using flow cytometry, we evaluated intracellular cytokine expression by T cells following stimulation with PMA and calcium ionophore. The phenotype and ability of the IL-4-producing CD8+ T cell clones to suppress IFN-γ production were examined.
Results. The percentages of IL-4+ CD8+ T cells were higher in PB of patients with AS and RA than controls (medians 0.90 and 0.84% vs 0.30%). In RA, patients with active inflammation had an increased percentage of IL-4+ CD8+ T cells. Higher frequencies of IL-4+ CD8+ T cells were also found in CD8+ T cell lines established from patients with arthritis. Interestingly, most IL-4+ CD8+ T cells produced TNF-α. Cloning the CD8+ T cell lines yielded more IL-4-producing clones from AS (23%) and RA patients (14%) than from controls (7%). The ability to suppress IFN-γ production was observed in 56% (AS) and 85% (RA) of IL-4-producing clones. Suppressive IL-4+ CD8+ T cell clones from RA patients showed a similar regulatory phenotype to the clones previously isolated from AS patients.
Conclusions. Expansion of IL-4+ CD8+ T cells, which may include precursors of a regulatory CD8+ T cell subset, may represent a general response to chronic joint inflammation.
Introduction
The role of CD8+ T cells in the pathogenesis of inflammatory arthritis remains undefined. AS is strongly associated with HLA-B27, leading to the assumption in the past that HLA-B27-restricted CD8+ T cells are important in the pathogenesis of disease. To date, however, arthritogenic HLA-B27-restricted CD8+ T cells have not been clearly identified in AS, and in the HLA-B27 transgenic rat model of AS, CD8+ T cells are not required for arthritis [1]. Nevertheless, several recent studies have still indicated a significant role of CD8+ T cells in the pathogenesis of AS. For example, CD8+CD28− T cells are more frequent in peripheral blood (PB) of AS patients than in healthy controls [2] and CD8+ effector and memory T-cells subsets have been characterized by a decreased production of pro-inflammatory cytokines in HLA-B27+ subjects including AS patients [3]. A CD8+ T-cell response to cartilage-derived peptides, has also been detected in AS [4].
In RA, CD8+ T cells have attracted less attention, but there is evidence that these cells may be involved in pathogenesis. In the synovial membrane, the most common IFN-γ-producing cell is the CD8+ T cell [5] and persistent clones of CD8+ memory T cells have been described in RA SF [6]. Subsets of CD8+ T cells may be preferentially recruited into the synovial tissue in a non-antigen-specific manner [7] by chemokines such as MIP-1α (macrophage inflammatory protein-1α) and RANTES (regulated upon activation, healthy T-cell expressed and secreted) expressed in RA synovial tissue [8, 9]. CD8+ T cells are also associated with the formation or maintenance of lymphoid aggregates in the RA synovium [10].
Although CD8+ T cells are predominately associated with the production of Th1 cytokines, typically IFN-γ, there is now good evidence that some subsets of these cells can also produce ‘Th2’ cytokines such as IL-4, IL5 and IL-10. IL-4-producing CD8+ T cells have been described in humans in several diseases including leprosy [11], HIV infection [12], Mycobacterium tuberculosis infection [13], tumours [14], SSc [15], chronic graft vs host disease [16], chronic obstructive pulmonary disease [17] and atopic asthma [18], and also as part of healthy ageing [19]. However, there are few data concerning IL-4-producing CD8+ T cells in inflammatory arthritis.
The potential functions associated with IL-4-producing CD8+ T cells are as yet unclear. Previously, we described novel human CD8+/TCR-αβ+ T cells with a regulatory phenotype and function, which were readily generated from PB of the patients with AS [20]. We expanded and cloned these cells using autologous lipopolysaccharide (LPS)-activated dendritic cells (DCs). The clones were not cytolytic, but responded in an autoreactive HLA class I-restricted fashion, by proliferation and production of IL-4, IL-5, IL-13 and TGF-β1, but not IFN-γ. They constitutively expressed CD69 and CD25 as well as molecules associated with CD4+CD25+ regulatory T (Treg) cells, including cytotoxic lymphocyte-associated antigen 4 (CTLA-4) and Foxp3. They suppressed IFN-γ production and proliferation by CD4+ T cells in vitro in a cell contact-dependent manner. We reasoned that CD8+ T cells that could be shown to produce IL-4 ex vivo might be precursors of these IL-4+ CD8+ Treg cells. Therefore, in this study, we measured the frequencies of IL-4+ CD8+ T cells in PB and autoreactive T cell lines of patients with AS, and compared them with the frequencies in RA and in healthy controls. We also investigated whether the frequencies were correlated with clinical features of disease and evaluated the properties of the IL-4-producing CD8+ T-cell clones.
Materials and methods
Subjects
PB was obtained from patients with AS (37) or RA (36), and from healthy controls (37). The patients with AS and RA met the Modified New York Criteria [21] and 1987 ACR Criteria [22], respectively. The Bath AS Disease Activity Index (BASDAI) [23] for AS patients and swollen joint count (SJC) for RA patients were calculated when their PB was obtained. Age, sex, disease duration, ESR, CRP, RF levels, HLA-B27 positivity, presence of radiographic bony erosions and current medication were assessed by reviewing medical records; ESR and CRP values were included if tested within 1 month of obtaining PB. The study was approved by Addenbrooke's Hospital Local Research Ethical Committee and written informed consent was given by all patients.
Purification of PB mononuclear cells
PB was collected into sterile, heparinized tubes. PB mononuclear cells (PBMCs) were purified from whole blood by centrifugation, using a Ficoll–Hypaque gradient (Amersham Pharmacia Biotech AB, Little Chalfont, Bucks).
Generation of IL-4-producing CD8+ T-cell lines and clones
CD8+ T-cell clones were generated as described [20]. First, DCs were differentiated from CD14+ cells as described [24]. In brief, PBMCs collected from patients and healthy controls were labelled with anti-CD14 mAb conjugated to microbeads (Miltenyi Biotech) and labelled cells were purified on separation columns (Miltenyi Biotech, Bisley, Surrey). CD14+ monocytes were cultured for 7 days in complete Rosewell Park Memorial Institute (RPMI) medium containing 10% human serum. Cultures were supplemented with 50 ng/ml human GM-CSF (Novartis, Frimley, Surrey) and 1000 U/ml human rIL-4 (PharMingen, San Diego, CA). DCs generated in vitro were activated with 1 μg/ml LPS at day 6. At 24 h post activation, DCs were harvested, washed at least three times with phosphate buffered saline (PBS) to remove IL-4 and GM-CSF and cultured at 104 cells/well in a 96-well plate.
Autologous CD8+ T cells were purified by negative selection using a cocktail of biotin-conjugated antibodies against CD4, CD14, CD16, CD19, CD36, CD56, CD123, TCR-γ/δ and CD235a (Glycophorin A). These cells are subsequently labelled with anti-biotin microbeads (Miltenyi Biotech) for depletion, and then CD8+ T cells were eluted from a separation column (Miltenyi Biotech).
Purified CD8+ T cells were added to DC cultures at 104 cells/well and cultured for 2 weeks in the presence of 50 U/ml human rIL-2 and 0.5 ng/ml IL-7. Cultures were supplemented with rIL-2 and IL-7 every 5–7 days. After 2 weeks of culture, the expanded CD8+ T-cell line was further purified by negative selection as described earlier, to obtain a pure CD8+ population for cloning. Following the second purification, the T-cell line contained 98% CD8+ T cells. The cells were seeded at 0.6 cells/well in 96-well plates with 105 irradiated allogeneic PBMC/well. Cells were cultured in complete medium containing 10% human AB serum supplemented with phytohaemagglutinin (PHA), 50 U/ml rIL-2 and 0.5 ng/ml IL-7. Cultures were supplemented with fresh rIL-2 and IL-7 every 5 days. After 2–3 weeks, the clones were established and were examined for IL-4 and IFN-γ production in response to autologous PBMCs by ELISA. Cytokine ELISA was performed using ELISA kits (Human CytoSets Antibody Pairs) obtained from Biosource International (Paisley, UK) according to the manufacturer's instructions. CD8+ T-cell clones were maintained by restimulation every 3 weeks with irradiated allogeneic PBMC plus PHA.
Flow cytometry and antibodies
Using flow cytometry, we evaluated the expression of surface markers and intracellular cytokines by T cells in PB, and in T-cell lines or clones. Antibodies used were anti-IL-4 phycoerythrin (PE), -IFN-γ (FITC), -TNF-α (FITC), -CD3 [peridin chlorphyll protein (PerCP)], -CD8 [allophycocyanin (APC) or PE], -TCR-αβ (FITC), -CD25 (FITC), -CD69 (FITC), -Foxp3 (PE), -CTLA4 (PE), -perforin (PE) and appropriate conjugated IgG isotype controls, obtained from BD PharMingen, San Diego, CA. Cell staining was analysed on a FACScan (Becton Dickinson, San Diego, California). For intracellular staining, cells were first stained with antibodies against surface antigens, then fixed and permeabilized using Perm/Fix solution (BD PharMingen). Cells were washed in buffer containing saponin (perm wash buffer; BD PharMingen) and stained with antibodies directed against intracellular cytokines. Antibodies for intracellular staining were prepared in perm wash buffer. To detect intracellular cytokines, cells were stimulated for 6 h with 50 ng/ml PMA and 1 μg/ml calcium ionophore in the presence of monensin (Golgi stop; BD PharMingen). Live CD3+ or CD3+ CD8+ T cells were gated and the percentage of these cells producing IL-4, IFN-γ or TNF-α was noted.
IFN-γ suppression assay
Allogeneic PBMCs were plated in duplicate in 96-well plates (105/well) in RPMI containing 10% human AB serum and stimulated with 2 μg/ml of PHA. In some wells, IL-4+ CD8+ T-cell clones were added at a 0.5 or 1 : 1 ratio with PBMC. At day 3, cell culture supernatants were analysed for IFN-γ by ELISA. Clones were defined as suppressive when the level of IFN-γ from the co-culture of PHA-stimulated allogeneic PBMC plus the CD8+ T-cell clone was <50% of that from PHA-stimulated allogeneic PBMCs alone.
Statistics
The median percentages of CD8+ T cells or their subsets according to cytokine expression were compared between subject groups or in terms of clinical status using the Mann–Whitney U-test. The Wilcoxon signed rank test was performed to compare the percentages of IL-4+ CD8+ T cells in PBMC and T cell lines expanded from the same donor. To test for correlations between the percentages of IL-4+ IFN-γ− CD8+ T cells and the percentages of IL-4+TNF-α+ T cells, Spearman's rank correlation coefficient was used. This was also used to ascertain any correlation between the percentages of IL-4+ CD8+ T cells and clinical parameters. All analyses were performed using Graphpad software (Prism, San Diego, CA).
Results
Increased frequency of IL-4+ CD8+ T cells in PB of patients with inflammatory arthritis
PB was obtained from 37 AS patients, 36 RA patients and 37 healthy controls whose characteristics are shown in Table 1. Using flow cytometry, we evaluated the intracellular expression of the cytokines IL-4 and IFN-γ by PB T cells following stimulation with phorbol 12-myristate 13-acetate (PMA) and calcium ionophore. Figure 1A–C shows three examples of double staining for IL-4 and IFN-γ, on CD8+ T cells. In each case, three major populations are visible—those making either IFN-γ or IL-4 and those not making either cytokine; cells making both IL-4 and IFN-γ were always rare. Figure 1D and E shows the results obtained from each of the patient and healthy control populations, with the percentages of either IL-4+IFN-γ− cells (referred to hereafter as IL-4+) or IL-4−IFNγ+ cells (referred to hereafter as IFNγ+) within the CD8+ T-cell population. To normalize the expression of IL-4, which might be influenced by variable activity of PMA in different experiments, the ratio of IL-4+ to IFNγ+ cells within the CD8+ T-cell population was also compared (Fig. 1F).
![Increased frequency of IL-4+ CD8+ T cells in PB of patients with inflammatory arthritis. (A–C) Examples of intracellular expression of IL-4 and IFN-γ by CD8+ T cells from patients with AS (A), RA (B) and from a healthy control (HC; C) after stimulation with PMA and calcium ionophore. (D) The percentages of IL-4+ CD8+ cells in T cells (median [interquartile range]): AS: 0.90% [0.44–2.19]; RA: 0.84% [0.45–1.85]; HC: 0.30% [0.12–0.54]). (E) The percentages of IFN-γ+ CD8+ cells in T cells in patients with AS or RA, and in HC—AS: 15.30% [5.81–36.98]; RA: 18.23% [5.29–50.45]; HC: 18.55% [11.80–39.83]. (F) The ratios of IL-4+ CD8+ cells to IFN-γ+ CD8+ cells in patients with AS: 0.065 [0.015–0.295]; RA: 0.055 [0.025–0.230] and HC: 0.020 [0.010–0.030]. Horizontal bars indicate the medians. IL-4+IFN-γ- and IL-4-IFN-γ+ cells are abbreviated as IL-4+ and IFN-γ+, respectively.](https://oup.silverchair-cdn.com/oup/backfile/Content_public/Journal/rheumatology/47/6/10.1093/rheumatology/ken089/2/m_ken089f1.jpeg?Expires=1709920517&Signature=LPfQFE-4p8TQOzebIWEo~4d4fXKRxriBI6LvPI4Rkb~TigEd2kynPjRXDjbHOmqXIY89OtLx9qBt1rSvzWI0FbBC17AnByE2IJvp0poTlGXer4upEsWtYh-NyNYr8l944-tAcDRGcVRcyiPJfOWfr6Gq3jYENU6JfyzALZoVysuB9~S2dDic-2ysUoXp54fivEyGSGyRarVXQgclOzAUsIDIxdGjkfJTMM839vOaaqvohhA-Isog~VgqeePMC7-Ce8xIvNRTcIxguT-adUd6MYxoqvAcDuA5A2MWtEzdewME1NcutRbywYtDzDIBweS9kBRY2Kp5LU5OEOliKTzV2A__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
Increased frequency of IL-4+ CD8+ T cells in PB of patients with inflammatory arthritis. (A–C) Examples of intracellular expression of IL-4 and IFN-γ by CD8+ T cells from patients with AS (A), RA (B) and from a healthy control (HC; C) after stimulation with PMA and calcium ionophore. (D) The percentages of IL-4+ CD8+ cells in T cells (median [interquartile range]): AS: 0.90% [0.44–2.19]; RA: 0.84% [0.45–1.85]; HC: 0.30% [0.12–0.54]). (E) The percentages of IFN-γ+ CD8+ cells in T cells in patients with AS or RA, and in HC—AS: 15.30% [5.81–36.98]; RA: 18.23% [5.29–50.45]; HC: 18.55% [11.80–39.83]. (F) The ratios of IL-4+ CD8+ cells to IFN-γ+ CD8+ cells in patients with AS: 0.065 [0.015–0.295]; RA: 0.055 [0.025–0.230] and HC: 0.020 [0.010–0.030]. Horizontal bars indicate the medians. IL-4+IFN-γ- and IL-4-IFN-γ+ cells are abbreviated as IL-4+ and IFN-γ+, respectively.
Demographic and clinical characteristics of patients and healthy controls
| AS (n = 37) | RA (n = 36) | HC (n = 37) | |
|---|---|---|---|
| Age (yrs)a | 46.0 [37.5–57.0] | 63.0 [57.0–71.0] | 38.5 [28.5–48.5] (n = 22) |
| Sex (Male : Female) | 25 : 12 | 14 : 22 | 8 : 14 (n = 22) |
| Disease duration (yrs)a | 11.0 [6.3–18.0] (n = 24) | 9.0 [3.0–14.0] | |
| BASDAIa | 3.6 [2.7–6.4] (n = 23) | ||
| Swollen joint counta | 2.0 [0.0–7.0] (n = 33) | ||
| Radiographic erosion | 22 (n = 32) | ||
| ESR (mm/h)a | 11.0 [6.0–21.0]/15 | 19.0 [9.0–28.0] (n = 35) | |
| CRP (mg/l)a | 7.0 [1.5–24.5]/13 | 10.0 [6.0–50.0] (n = 13) | |
| HLA-B27 | 13/17 | 9 (n = 20) | |
| RF+ | 23 (n = 33) | ||
| Current use of steroid | 3 (n = 25) | 20 | |
| Current use of TNF blocking agent | 6 (n = 25) | 3 |
| AS (n = 37) | RA (n = 36) | HC (n = 37) | |
|---|---|---|---|
| Age (yrs)a | 46.0 [37.5–57.0] | 63.0 [57.0–71.0] | 38.5 [28.5–48.5] (n = 22) |
| Sex (Male : Female) | 25 : 12 | 14 : 22 | 8 : 14 (n = 22) |
| Disease duration (yrs)a | 11.0 [6.3–18.0] (n = 24) | 9.0 [3.0–14.0] | |
| BASDAIa | 3.6 [2.7–6.4] (n = 23) | ||
| Swollen joint counta | 2.0 [0.0–7.0] (n = 33) | ||
| Radiographic erosion | 22 (n = 32) | ||
| ESR (mm/h)a | 11.0 [6.0–21.0]/15 | 19.0 [9.0–28.0] (n = 35) | |
| CRP (mg/l)a | 7.0 [1.5–24.5]/13 | 10.0 [6.0–50.0] (n = 13) | |
| HLA-B27 | 13/17 | 9 (n = 20) | |
| RF+ | 23 (n = 33) | ||
| Current use of steroid | 3 (n = 25) | 20 | |
| Current use of TNF blocking agent | 6 (n = 25) | 3 |
HC: healthy controls; aMedian [interquartile range]; n: the number of subjects on whom demographic or clinical details available. No details were available on healthy donors of buffy coats obtained from the National Blood transfusion service.
Demographic and clinical characteristics of patients and healthy controls
| AS (n = 37) | RA (n = 36) | HC (n = 37) | |
|---|---|---|---|
| Age (yrs)a | 46.0 [37.5–57.0] | 63.0 [57.0–71.0] | 38.5 [28.5–48.5] (n = 22) |
| Sex (Male : Female) | 25 : 12 | 14 : 22 | 8 : 14 (n = 22) |
| Disease duration (yrs)a | 11.0 [6.3–18.0] (n = 24) | 9.0 [3.0–14.0] | |
| BASDAIa | 3.6 [2.7–6.4] (n = 23) | ||
| Swollen joint counta | 2.0 [0.0–7.0] (n = 33) | ||
| Radiographic erosion | 22 (n = 32) | ||
| ESR (mm/h)a | 11.0 [6.0–21.0]/15 | 19.0 [9.0–28.0] (n = 35) | |
| CRP (mg/l)a | 7.0 [1.5–24.5]/13 | 10.0 [6.0–50.0] (n = 13) | |
| HLA-B27 | 13/17 | 9 (n = 20) | |
| RF+ | 23 (n = 33) | ||
| Current use of steroid | 3 (n = 25) | 20 | |
| Current use of TNF blocking agent | 6 (n = 25) | 3 |
| AS (n = 37) | RA (n = 36) | HC (n = 37) | |
|---|---|---|---|
| Age (yrs)a | 46.0 [37.5–57.0] | 63.0 [57.0–71.0] | 38.5 [28.5–48.5] (n = 22) |
| Sex (Male : Female) | 25 : 12 | 14 : 22 | 8 : 14 (n = 22) |
| Disease duration (yrs)a | 11.0 [6.3–18.0] (n = 24) | 9.0 [3.0–14.0] | |
| BASDAIa | 3.6 [2.7–6.4] (n = 23) | ||
| Swollen joint counta | 2.0 [0.0–7.0] (n = 33) | ||
| Radiographic erosion | 22 (n = 32) | ||
| ESR (mm/h)a | 11.0 [6.0–21.0]/15 | 19.0 [9.0–28.0] (n = 35) | |
| CRP (mg/l)a | 7.0 [1.5–24.5]/13 | 10.0 [6.0–50.0] (n = 13) | |
| HLA-B27 | 13/17 | 9 (n = 20) | |
| RF+ | 23 (n = 33) | ||
| Current use of steroid | 3 (n = 25) | 20 | |
| Current use of TNF blocking agent | 6 (n = 25) | 3 |
HC: healthy controls; aMedian [interquartile range]; n: the number of subjects on whom demographic or clinical details available. No details were available on healthy donors of buffy coats obtained from the National Blood transfusion service.
The percentages of IL-4+ CD8+ T cells were higher in AS and in RA patients than in healthy controls (medians 0.90 and 0.84% vs 0.30%; P < 0.0001 and P = 0.0002, respectively, Fig. 1D). Likewise the ratios of IL-4+ to IFNγ+ CD8+ T cells were also elevated in patients with AS (0.065) or RA (0.055) compared with controls (0.020; P = 0.0008 and P < 0.0001, respectively, Fig. 1F). In contrast, there were no differences amongst the three groups in the percentages of CD8+ cells staining for IFN-γ alone (Fig. 1E), or for both IL-4 and IFN-γ (data not shown).
Since the numbers of cytokine-producing CD8+ T cells would be affected by the absolute numbers of CD8+ T cells in PB, we examined the percentages of CD8+ T cells within the total T-cell populations (CD3+ cells) for each of the groups. The percentage of CD8+ cells within the CD3+ population of RA patients was lower than that of healthy controls (median [interquartile range] 18.80% [12.72–25.80] vs 25.16% [21.68–30.18]; P = 0.0072), but this was not the case for AS patients. When we expressed the numbers of IL-4+ CD8+ T cells as a percentage of the total CD3+ population, they were still significantly higher in AS patients (0.125% [0.075–0.415]) and RA patients (0.165% [0.045–0.320]) than in controls (0.075% [0.025–0.165]) (P = 0.0124 and P = 0.0238, respectively). Thus, the frequency of IL-4+ CD8+ T cells was increased in patients with inflammatory arthritis independent of the proportion of CD8+ cells within the CD3+ population.
When CD8− cells were examined (these are virtually all CD4+) the frequencies of cells staining for IL-4 were comparable in patients with AS and RA with those found in healthy controls (1.20, 1.06 and 0.85%, respectively; Fig. 2A). However, the ratios of IL-4+ to IFNγ+ cells were significantly higher for CD8– cells in both AS (0.165) and RA (0.260) than in controls (0.100; P = 0.0013 and P < 0.0001, respectively, Fig. 2C). In the case of RA, this was partially attributable to a lower percentage of IFNγ+ CD8− cells as compared with controls (5.13 vs 8.82%; P = 0.002).
![(A) The percentages of IL-4+ cells in CD8− T cells: median [interquartile range] AS: 1.20% [0.86–1.57]; RA: 1.06% [0.38–1.86]; HC: 0.85% [0.43–1.56]. (B) The percentages of IFN-γ+ CD8− cells in AS: 6.66% [4.35–12.32]; RA: 5.13% [1.77–8.59] and HC: 8.82% [5.77–13.20]. (C) The ratio of IL-4+ to IFN-γ+ cells for CD8− cells in AS: 0.165 [0.120–0.295]; RA: 0.260 [0.175–0.590] and HC: 0.100 [0.065–0.140]. Horizontal bars indicate the medians. IL-4+IFN-γ− and IL-4−IFN-γ+ are abbreviated as IL-4+ and IFN-γ+, respectively.](https://oup.silverchair-cdn.com/oup/backfile/Content_public/Journal/rheumatology/47/6/10.1093/rheumatology/ken089/2/m_ken089f2.jpeg?Expires=1709920517&Signature=kxZCXJEOlrNYm0QICzejlZe4q6~IAQ-kZ88Y2yIlVQDeziei8MYRW6oYKbbfUTdk6gd-JxDPL4q3VnbXizMTPhFSw8FZyH0G8cp95b6-wpBeMCliEP0G~Lqn2ebTwfXDcSMRWqeGx-WzO5Uk12Afuy9OfW~4f6aIgM5nB~fEJIrahJ6H-mSrwGtODqkz7dm9NjJkj69TfkH0od9B81cgfb-vL5IQjbEFxdzfrCCovWxRCPYDB71ikvfmdflTsHKeFRUGTsnOmsX6R7-WNuE-srssg4m4Itz584LkDVSsVVTS5hvzuhodnr0YDAa4t-qoVQgUvK-uzhonXOTD9GydVA__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
(A) The percentages of IL-4+ cells in CD8− T cells: median [interquartile range] AS: 1.20% [0.86–1.57]; RA: 1.06% [0.38–1.86]; HC: 0.85% [0.43–1.56]. (B) The percentages of IFN-γ+ CD8− cells in AS: 6.66% [4.35–12.32]; RA: 5.13% [1.77–8.59] and HC: 8.82% [5.77–13.20]. (C) The ratio of IL-4+ to IFN-γ+ cells for CD8− cells in AS: 0.165 [0.120–0.295]; RA: 0.260 [0.175–0.590] and HC: 0.100 [0.065–0.140]. Horizontal bars indicate the medians. IL-4+IFN-γ− and IL-4−IFN-γ+ are abbreviated as IL-4+ and IFN-γ+, respectively.
The patients with RA, and to a lesser extent AS, were somewhat older than the healthy controls in our sample (Table 1). In view of the report linking frequencies of IL-4+ CD8+ T cells and age [19], we compared only those subjects between 37 and 60 yrs of age in each group, to allow for the possible effect of age on the proportion of IL-4+ CD8+ T cells. In these cases, the AS and RA patient groups did not differ significantly by age from controls (46.5 [43.0–55.5] yrs and 54.0 [44.0–57.0] yrs vs 47.5 [39.0–53.5] yrs, respectively). Even following this adjustment, the AS patients still had higher percentages of IL-4+ CD8+ T cells than controls (0.900% [0.585–1.795] and 0.300% [0.105–1.150]; P = 0.0257), but this was not the case for the RA patients (0.765% [0.230–1.330]). However, the ratios of IL-4+ to IFNγ+ CD8+ cells were significantly higher in both AS and RA patients as compared with controls (0.080 [0.020–0.270] and 0.095 [0.045–0.280] vs 0.015 [0.000–0.030]; P = 0.0059 and P = 0.0014, respectively). This analysis suggests that the difference in the percentage of IL-4+ CD8+ T cells between the patient and controls groups could not simply be explained on the basis of differences in age. There was also no significant correlation between age and the percentage of IL-4+ CD8+ T cells in each group of subjects (data not shown).
The influence of disease characteristics on the frequency of IL-4+ CD8+ T cells
There was no significant correlation between the frequency of IL-4+ CD8+ T cells and disease duration in either AS or RA, although in AS, there was a suggestion of a negative correlation with disease duration (Spearman r = −0.3989, P = 0.0535). In AS, there was no correlation between the percentage of IL-4+ CD8+ T cells and indices of clinical activity—BASDAI, duration of morning stiffness, ESR or CRP, or with treatment with steroids or TNF-blocking drugs (data not shown). In RA, patients with five and more swollen joints had an increased percentage of IL-4+ CD8+ cells compared with those with <5 swollen joints (1.89 vs 0.69%; P = 0.0249; Fig. 3A). In addition, RA patients with an ESR >20 mm/h showed a higher percentage of IL-4+ CD8+ T cells than those with an ESR ≤20 mm/h (1.81 vs 0.52%; P = 0.0034, Fig. 3C). Likewise, the patients with more swollen joints or higher ESR had an increased percentage of IL-4+ CD8+ T cells as a proportion of CD3+ T cells (Fig. 3B and D). Considering the whole group of RA patients there was a significant correlation between numbers of IL-4+ CD8+ cells as a proportion of CD3+ cells and the SJC (r = 0.35, P < 0.05). Correlations with CRP were present (r = 0.4–0.5), but were not significant, but the number of observations of CRP was substantially less than that for ESR. There were no correlations with RF-positivity, the presence of radiographic erosions, or treatment with either steroids or TNF blocking drugs (data not shown).
![Correlations between inflammatory status and the frequency of IL-4+ CD8+ T cells in RA patients. (A and B) Patients with five and more swollen joints compared with those with <5 swollen joints for (A) percentage of IL-4+ cells in CD8+ T cells (median [interquartile range] 1.89% [0.74–4.36] vs 0.69% [0.22–1.39] or (B) percentage of IL-4+ CD8+ as a proportion of CD3+ T cells 0.35% [0.25–0.84] vs 0.13% [0.04–0.25]. (C and D) Patients with an ESR >20 mm/h compared with those with an ESR ≤20 mm/h for (C) percentage of IL-4+ cells in CD8+ T cells 1.81% [0.81–3.70] vs 0.52% [0.22–0.94] or (D) for IL-4+ CD8+ cells as a proportion of CD3+ T cells 0.33% [0.16–0.72] vs 0.10% [0.03–0.17]. Horizontal bars indicate the medians. IL-4+IFN-γ− and IL-4−IFN-γ+ are abbreviated as IL-4+ and IFN-γ+, respectively.](https://oup.silverchair-cdn.com/oup/backfile/Content_public/Journal/rheumatology/47/6/10.1093/rheumatology/ken089/2/m_ken089f3.jpeg?Expires=1709920517&Signature=mO1yYse5igu-KRmPFbjNTdFY-Mn0oLggsWfRGDNTIf2PMI9Cl2591tMAxZtbzcNWEGVSYAJi4wCK1EXExkBAJ0FVj8qZvMb1pvPJ1KhJUGG-M2Z~N2XpfnZEIUkteLOSLxHgnDcnAJJ3RnH6BnRBoObR6FJxJjgZv1dXFWpyP2OEgJaff5568vzGNAeWcKH89n5c21Q4uXAhBVf9g1QmF8rRKOq1rxKddCHjEo2Vig0VjI1bBXgVx~SJahWV8BrFXfucjXFmyWUcIw0X~pREMiYPZyUJ8Bay~AniacEfPZfTpmf6exa8HdMjLQtilDln6nXrQMWu79qT~4~bJzt9jg__&Key-Pair-Id=APKAIE5G5CRDK6RD3PGA)
Correlations between inflammatory status and the frequency of IL-4+ CD8+ T cells in RA patients. (A and B) Patients with five and more swollen joints compared with those with <5 swollen joints for (A) percentage of IL-4+ cells in CD8+ T cells (median [interquartile range] 1.89% [0.74–4.36] vs 0.69% [0.22–1.39] or (B) percentage of IL-4+ CD8+ as a proportion of CD3+ T cells 0.35% [0.25–0.84] vs 0.13% [0.04–0.25]. (C and D) Patients with an ESR >20 mm/h compared with those with an ESR ≤20 mm/h for (C) percentage of IL-4+ cells in CD8+ T cells 1.81% [0.81–3.70] vs 0.52% [0.22–0.94] or (D) for IL-4+ CD8+ cells as a proportion of CD3+ T cells 0.33% [0.16–0.72] vs 0.10% [0.03–0.17]. Horizontal bars indicate the medians. IL-4+IFN-γ− and IL-4−IFN-γ+ are abbreviated as IL-4+ and IFN-γ+, respectively.
Most IL-4+ CD8+ T cells also produce TNF-α
On intracellular staining PB CD8+ T cells for both IL-4 and TNF-α, the majority of the IL-4-producing cells were also positive for TNF-α (Fig. 4A–C). Indeed the frequency of IL-4+TNFα + CD8+ T cells was strongly positively correlated with the percentage of IL-4+ CD8+ T cells in all subjects studied (P < 0.0001, Spearman r = 0.8867) (Fig. 4D). This suggests that most IL-4+ CD8+ T cells also produce TNF-α.
(A–C) Examples of intracellular expression of IL-4 and TNF-α by PB CD8+ T cells from a patient with AS (A), RA (B) or from a healthy control (HC) (C) after stimulation of PMA and calcium ionophore. (D) Correlation between the percentages of IL-4+TNF-α+ CD8+ T cells and the percentages of IL-4+IFNγ−CD8+ T cells for all the subjects studied.
Properties of IL-4+ CD8+ T cell clones from AS and RA patients
Since we originally described regulatory CD8+ T cells in AS patients [20], we wished to determine whether similar cells were also present in RA, as suggested by the increased proportion of IL-4+ CD8+ cells in PB in RA patients. Accordingly, we established T-cell lines by co-culturing purified CD8+ T cells with autologous LPS-activated DC and measured intracellular expression of IL-4, IFN-γ and TNF-α by the resulting T cells. The percentages of IL-4+ CD8+ T cells were significantly increased in T-cell lines, as compared with the starting percentages in PBMC (data not shown), and were higher in lines derived from RA patients (15.05% [1.75–38.60], n = 10) as compared with those derived from healthy controls, (1.04% [0.32–6.10], n = 19; P = 0.0048); the percentages of IL-4+ CD8+ T cells were also higher in lines from AS patients but this difference was not significant (3.60% [1.84–10.43], n = 17; P = 0.0994). Likewise, the ratios of IL-4+ to IFN-γ+ CD8+ T cells were elevated in patients with AS (0.750 [0.225–1.400]) or RA (1.635 [0.205–5.575]) compared with controls (0.030 [0.020–0.680]; P = 0.0108 and P = 0.0109, respectively). IL-4+ cells in the lines were also found to be producing TNF-α. These lines were then cloned, as described in ‘Materials and methods’ section. This yielded a total of 291, 369 and 129 CD8+ T-cell clones from AS, RA and control subjects, respectively (Table 2). These clones were then screened for IL-4 production in response to autologous PBMC by ELISA; this physiological stimulus was used in preference to a non-specific stimulus such as PMA and ionomycin since we had previously shown CD8+ Treg clones to be autoreactive and class I MHC-restricted. This analysis revealed 66 (22.7%) and 50 (13.6%) of the clones to be IL-4-producing from AS and RA patients, respectively. All IL-4-producing cells tested (i.e. all clones obtained from two RA and two AS patients) were also shown to produce TNF-α. In contrast, nine (7%) of the clones obtained from the controls were IL-4-producing; as well as being present at low frequency these clones grew poorly and could not be characterized further. Most of the IL-4-producing CD8+ T-cell clones (84.8%) from RA patients (39 of 46 tested) had the ability to suppress IFN-γ production as compared with 55.6% (35 of 63) of the clones from AS patients (Fig. 5A and B), as compared with a cytotoxic CD8+ T-cell clone that did not produce IL-4 (Fig. 5C). Two IL-4-producing CD8+ T-cell clones from RA patients with the ability to inhibit IFN-γ production were characterized by flow cytometry and shown to have a phenotype indistinguishable from that previously found for clones from AS patients, i.e. TCR-αβ+, IL-4+, TNF-α+, CD25+, CD69+, CTLA4+ and Foxp3+, but negative for IFN-γ and perforin (one example shown in Fig. 5D).
(A–C) IFN-γ suppression assay for CD8+ T-cell clones from RA patients. Effects on IFN-γ production by PHA-stimulated allogeneic PBMCs when co-cultured with different numbers of cells of IL-4-producing CD8+ T-cell clones. (A) and (B) show the effects of non-cytolytic IL-4+ clones as compared with (C), which shows the effect of a cytotoxic IL-4− CD8+ T-cell clone. aThe ratios of PHA-stimulated PBMC to co-cultured clone cells. Error bars show s.d. (D) Flow cytometric analysis of a CD8+ T-cell clone from an RA patient. Thick black lines show staining with the antigen-specific antibody; filled grey histograms represent staining with isotype control antibodies.
CD8+ T-cell clones from patients with inflammatory arthritis and healthy controls
| AS (n = 5) | RA (n = 4) | HC (n = 3) | |
|---|---|---|---|
| No. of CD8+ | 291 | 369 | 129 |
| No. of IL-4-producing CD8+ | 66 | 50 | 9 |
| No. of suppressivea IL-4-producing CD8+(of those tested) | 35/63 | 39/46 | 3/3 |
| AS (n = 5) | RA (n = 4) | HC (n = 3) | |
|---|---|---|---|
| No. of CD8+ | 291 | 369 | 129 |
| No. of IL-4-producing CD8+ | 66 | 50 | 9 |
| No. of suppressivea IL-4-producing CD8+(of those tested) | 35/63 | 39/46 | 3/3 |
HC: healthy controls; aa clone was defined as suppressive when the level of IFN-γ from the co-culture of PHA-stimulated allogeneic PBMCs with the clone at a 1:1 ratio was <50% of that obtained from PHA-stimulated allogeneic PBMCs alone.
CD8+ T-cell clones from patients with inflammatory arthritis and healthy controls
| AS (n = 5) | RA (n = 4) | HC (n = 3) | |
|---|---|---|---|
| No. of CD8+ | 291 | 369 | 129 |
| No. of IL-4-producing CD8+ | 66 | 50 | 9 |
| No. of suppressivea IL-4-producing CD8+(of those tested) | 35/63 | 39/46 | 3/3 |
| AS (n = 5) | RA (n = 4) | HC (n = 3) | |
|---|---|---|---|
| No. of CD8+ | 291 | 369 | 129 |
| No. of IL-4-producing CD8+ | 66 | 50 | 9 |
| No. of suppressivea IL-4-producing CD8+(of those tested) | 35/63 | 39/46 | 3/3 |
HC: healthy controls; aa clone was defined as suppressive when the level of IFN-γ from the co-culture of PHA-stimulated allogeneic PBMCs with the clone at a 1:1 ratio was <50% of that obtained from PHA-stimulated allogeneic PBMCs alone.
Discussion
Although there has been a tendency to equate inflammation and the production of ‘Th1’ cytokines such as IFN-γ, it has recently been reported that in SpA, PB T cells secrete ‘Th2’ cytokines such as IL-4 and IL-10, with a relative lack of T cells that produce IFN-γ [25]; likewise, the synovium in SpA has been shown to contain IL-4-producing cells [26]. With regard to RA, patients with early arthritis who subsequently developed RA had elevated levels of IL-4 in SF as compared with those who did not develop RA, but predominance of IL-4 was not a feature of SF in patients with established RA [27]. In our study, an increase in IL-4+ T cells was observed particularly in the CD8+ subset in PB of patients with either AS or RA. IL-4-producing cells were also present at higher frequencies in T-cell lines derived from co-culture of CD8+ T cells with autologous DC, both in AS and in RA. In addition, the percentage of IL-4+ CD8+ T cells was increased in RA patients with active disease as judged by the number of swollen joints or ESR. Although, in a limited data set, such an effect was not obvious in AS when activity was assessed in terms of BASDAI and ESR, these measures are less accurate indicators of disease activity in AS, since the BASDAI is subjective [23], unlike the SJC, and many AS patients do not mount an impressive acute-phase response [28, 29]. Thus the increased frequency of IL-4+ CD8+ T cells in PB may in fact reflect active inflammation in both diseases.
It has also been reported that the frequency of IL-4+ CD8+ T cells is increased in healthy older adults [19]. In that study, IL-4+ CD8+ T cells did not occur in those under the age of 40, but were present in 36% of those >60 yrs of age. The authors suggested that these cells might counterbalance the overproduction of pro-inflammatory cytokines in old age. We could not confirm this phenomenon in our healthy controls because they had a skewed age distribution, being all under 60 yrs. Meanwhile, although the patients with AS and RA were somewhat older than healthy controls, the age-matched sub-group analysis suggested that the differences between the patient and control groups could not simply be explained on the basis of differences in age.
Until recently, CD8+ T cells that showed a ‘Th2’ pattern of cytokine secretion (i.e. IL-4, IL-5, IL-13 and IL-10) were considered a subset of cytotoxic T cells and the term ‘Tc2’ has been used [30]. However, there is growing evidence that many CD8+ T cells that produce these cytokines have reduced expression of perforin and do not function as cytolytic cells [12, 19]. The functions associated with these non-cytolytic CD8+ T cells are still unclear. They have been demonstrated to provide B cell help, as do Th2 cells [12, 31], and IL-4-producing CD8+ T cells may be involved in the humoral immune response following immunization [19]. IL-4 is an anti-inflammatory cytokine, which can therefore counteract both Th1, and the more recently described Th17, responses [32, 33]. IL-4-producing CD8+ T cells have also been reported to suppress the proliferative response of CD4+ T cells [11]. CD8+ T cells lacking IFN-γ production were associated with lack of pathogen clearance in some models of viral infection [34, 35], so that induction of CD8+ T cells with a Th2 pattern of cytokine secretion could be a mechanism employed by pathogens to evade cytotoxic T-cell mediated immune responses [11].
Previously we suggested that the IL-4+ CD8+ T cells seen following stimulation immediately ex vivo might be precursors of the IL-4+ CD8+ T cells with regulatory phenotype and function that we isolated from patients with AS following co-culture of CD8+ T cells with autologous DC [20], although it is important to note that there is no necessary connection between the IL-4+ CD8+ T cells identified directly ex vivo and those which emerge in the DC co-culture system. However, having observed significant numbers of IL-4+ CD8+ cells in RA as well as AS, we wished to determine whether we could also isolate CD8+ regulatory T cells from RA patients. This proved to be the case, with IL-4+ CD8+ clones from RA patients being able to suppress IFN-γ production by PHA-stimulated PBMCs to the same extent as the clones from AS patients. In addition, suppressive IL-4+ CD8+ T-cell clones from RA patients expressed the same phenotypic markers (CD25+, CTLA4+ and Foxp3+, whilst being negative for IFN-γ and perforin) as the clones previously isolated from AS patients. Few clones were obtained from normal subjects and they could not be expanded for further testing; recently, however, we have been able to characterize a few IL-4+ CD8+ clones from healthy subjects and they show the same phenotype and regulatory function as those obtained from AS and RA patients.
It remains possible therefore that the IL-4+ CD8+ T cells detectable ex vivo by stimulation with PMA and ionomycin represent precursors of a regulatory subset of CD8+ T cells in inflammatory arthritis. We attempted to demonstrate this point directly by isolating CD8+ cells that make IL-4 following stimulation by using a surface cytokine capture technique rather than intracellular staining (which kills the cells). Very small numbers of cells were obtained by this technique and we were not successful in cloning them. Thus we have not been able to prove conclusively that the IL-4+ CD8+ cells that we have demonstrated in PBMC ex vivo are precursors of the cells isolated by co-culture with DCs. Clones derived from the latter have a regulatory phenotype and function, but our recent experiments have suggested that not all of the IL-4+ CD8+ T cells in lines obtained by co-culture with autologous DC express Foxp3, whereas all of the Foxp3+ clones obtained from these lines are IL-4+ (L. Zhang and L.B. Jarvis, unpublished data). Thus, cells with regulatory capacity seem to be a subset of all those CD8+ cells capable of making IL-4.
Although it is paradoxical that we have isolated IL-4+ CD8+ regulatory cells from patients who have ongoing active inflammation, the expansion of IL-4+ CD8+ T cells might represent a general response to chronic joint inflammation as an anti-inflammatory feedback mechanism like the production of soluble TNF receptors or IL-1 receptor antagonist in active RA [36, 37]. Whilst the clones show potent regulatory properties in vitro it is not known whether they have this effect in vivo. Longitudinal studies to correlate the percentages of IL-4+ CD8+ T cells with the course of arthritis are needed to support this proposed role of these cells in inflammatory arthritis.
Whilst we have obtained IL-4-producing clones with regulatory function, it is also conceivable that IL-4-producing cells could drive inflammation, since IL-4 can be pro-inflammatory [38]. In this respect, it is of interest that most IL-4+ CD8+ T cells in PB and T-cell lines also produced TNF-α, and this applied even to clones with the ability to suppress IFN-γ production. TNF-α is a major mediator of inflammation and TNF blocking agents are widely used in treating RA and AS, hence again it seems paradoxical that suppressive IL-4+ CD8+ T cells produce TNF-α. However, TNF-α is known to have an immunoregulatory as well as a pro-inflammatory activity. Mice with low levels of TNF-α production develop enhanced autoimmunity and severe renal disease [39], whilst TNF-deficient mice develop severe inflammatory arthritis following immunization with type II collagen [40]. Anti-TNF treatment in patients commonly result in the appearance of ANA, whilst treatment of multiple sclerosis patients resulted in disease exacerbations rather than remission [41, 42]. The anti-inflammatory actions of TNF-α have been previously reported to be caused by inducing apoptosis of activated antigen-specific T cells [43, 44]. In addition, it has been recently suggested that the immunoregulatory function of TNF-α may be exerted by inhibition of IL-12 and IL-23 production by macrophages and DC [45]. Both T-cell apoptosis induction and the effect on antigen-presenting cells require the p55 TNF receptor Type I (TNFRI). However, in some experimental models, immunosuppression by TNF-α did not require TNFRI [46]; thus, experimental autoimmune encephalomyelitis (EAE) was induced in TNFRI−/− mice, the myelin-specific T-cell response declined with time in the same way as in wild-type controls, and resistance to re-induction of EAE was established normally. Therefore, TNFRII (p75) may also play a role; TNFRII is expressed on T cells including Treg, but the effects of TNF-α on regulation are controversial. Whilst there are reports of inhibition of Treg function by TNF-α [47] and improved Treg function in patients treated with TNF inhibitors [48], others report TNF-α promoting Treg proliferation and function [49]. Thus, although the role in vivo of TNF-α produced by suppressive IL-4+ CD8+ T cells is not known, it may in some circumstances be immunoregulatory. There are also additional precedents for IL-4-producing T cells also making TNF, as reported recently [50].
In conclusion, we have shown that the frequencies of IL-4+ CD8+ T cells were increased in PB of patients with inflammatory arthritis as compared with healthy controls. In RA, their frequency was clearly correlated with active inflammation. As we reported previously for AS [20], regulatory IL-4+ CD8+ T cells could be obtained from RA patients by co-culture with autologous DC, and the phenotype of these clones was identical to those previously characterized in AS. Thus we propose that the expansion of a regulatory subset of CD8+ T cells in inflammatory arthritis may represent a general response to chronic joint inflammation.
Acknowledgements
We are grateful to Dominique Raut-Roy for collecting blood samples and clinical data from patients.
Funding: This work was supported by grants from the Korea Research Foundation Grant of the Korean Government (MOEHRD) (KRF-2006-013-E00104), AstraZeneca plc, The Henry Smith Charity, and NIHR Cambridge Biomedical Research Centre.
Disclosure statement: The authors have declared no conflicts of interest.
Comments