In vivo activated T cells in rheumatoid synovitis. Analysis of Th1- and Th2-type cytokine production at clonal level in different stages of disease


Prof RobertoGerli Department of Clinical and Experimental Medicine, Section of Internal Medicine and Oncologic Sciences, Rheumatology Unit, Policlinico di Monteluce, I-06122 Perugia, Italy E-mail:


T-cell cytokines play a crucial role in the pathogenesis and progression of rheumatoid arthritis (RA). Their detection in the joint, however, is impaired by the complex network present in the synovium. Although many synovial T cells show signs of previous activation, only a few express interleukin (IL)-2 receptor, marker of recent activation. The aim of this study was to analyse the cytokine production by in vivo activated (IL-2R +) T cells from RA at different stages of the disease. For this purpose, T cells were isolated from peripheral blood and synovial fluid of four patients with active RA, two at the onset of the disease, one in the early phase during treatment, one in long-lasting chronic phase. One patient was studied at the onset of the disease and 52 months later. Cells were initially expanded with a low dose of IL-2, cloned and analysed for cytokine production. The results showed a strong predominance of T helper (Th) 1 clones in the blood and a slight prevalence of Th0 clones in the joint of all the four patients. Interferon-γ and IL-2 production was higher in the long-lasting RA, whereas IL-4 synthesis was prevalent in early RA. Enrichment in IL-10-producing clones was present only in the joint of the untreated patients. The longitudinal study confirmed the differences in cytokine production between early and late phases of disease. These data confirm that RA is mainly a Th1-driven condition. However, in vivo activated synovial T cells produce also Th2-type anti-inflammatory cytokines, such as IL-4 and IL-10. The synthesis of both cytokines is a feature of the very early phase of RA, although the selective recruitment of IL-10-producing T cells is quickly lost.


Rheumatoid arthritis (RA) is characterized by inflammatory and proliferative synovitis that exhibits invasive and destructive features [1,2]. So far, it has been assumed that most of the tissue damage associated with rheumatoid synovitis is due to activated macrophages and synoviocytes [3], whereas tissue infiltrating T lymphocytes are believed to play a pivotal role in orchestrating the inflammatory response [4–6]. This concept is supported by the abundance, in RA synovium, of T cells expressing differentiation and activation markers on their surface [5]. This phenotypic evidence strongly suggests that antigen-specific T lymphocytes act as initiators of the inflammatory process and that T-cell-derived cytokines play a major role in maintaining inflammation and in mediating matrix destruction of the synovium [2,4–6].

As reported for murine T cells, human T lymphocytes may be identified as T helper (Th)1- or Th2-type cells depending on their cytokine production profile [7]. It has been postulated that the balance between cells producing Th1-type cytokines with pro-inflammatory properties, such as interferon (IFN)-γ and interleukin (IL)-2, and those producing Th2-type cytokines which down-modulate inflammation, including IL-4 and IL-10, may be important in regulating both development and persistence of chronic inflammatory processes [8,9]. In addition, a regulatory role in inflammation has been recently described for T-cell subsets mainly producing transforming growth factor-β (Th3-type cells) or IL-10 (T regulatory cells type 1 – Tr1) [10–13].

In RA, the levels of T-cell-derived cytokines are very low in the synovial fluid (SF) as well as in the supernatants of unstimulated T-cell cultures from rheumatoid joints [2–5]. Therefore, most of the data on T-cell cytokine production in RA synovitis derive from information drawn by stimulation of fresh or cloned T cells. These studies suggest that RA synovitis is a Th1-driven process, because of the prevalence of IFN-γ production by synovial T cells [14–16]. The picture, however, seems to be much more complex because of the extensive heterogeneity of the cytokine profiles secreted by the different T-cell subsets present in the inflamed joints [2,8,17]. In this setting, the complex cytokine network may be influenced by the different stages of the disease, flares and remissions, and by the treatment employed [18–22]. Taken together, these observations may account for the conflicting data reported on T-cell-derived lymphokine production in RA.

The majority of T lymphocytes are bystander cells randomly recruited into the RA joint from the peripheral blood (PB) [5,6] and display phenotypic markers suggestive of an incomplete activation, while only a minority of them express high-affinity IL-2 receptors (IL-2R) which indicate recent activation [5,17,23,24]. Analysis of this last T-cell subset may provide useful information on the potential pattern of cytokines produced by T cells which, being already activated in vivo, should be directly involved in the immune response of RA synovitis [5]. Previous analysis of cytokine production by T cells cloned from a pool of in vivo activated cells obtained from RA synovium, showed the predominance of high IFN-γ/low IL-4-producing clones and revealed the presence of IL-10-secreting clones [25]. However some authors have reported that the method of deriving human T cell clones may alter the proportion of IL-10 producing cells, thus giving an erroneous picture of the pattern of in vivo cytokine secretion [26].

Our aim was to analyse the lymphokine production by T-cell clones generated from RA SF and PB T cells expanded in vitro with low concentrations of rIL-2. This method has been used to study the pattern of cytokines secreted by clones obtained from four patients with active RA at different stages of the disease; one of the patients was studied at disease onset, before treatment, and 52 months later, on treatment.



Paired PB and knee SF samples were obtained from four patients who fulfilled the American College of Rheumatology diagnostic criteria for RA [27] (Table 1). According to disease duration, three patients had early RA (A1, A2 and B) and the other one had long-standing RA (C). Patients A1 and A2 were studied at diagnosis, before treatment. Both were then followed up for at least one year to confirm the diagnosis. Patient A1 was evaluated again 52 months later when on methotrexate (10 mg/week) and persistent active disease. Patient B had disease duration shorter than 2 years and was under treatment at the moment of the study. All patients were female, rheumatoid factor positive and had active disease as defined by the presence of six or more tender joints, and two of the following conditions: nine or more swollen joints, morning stiffness lasting for more than 45 min, or a Westergren erythrocyte sedimentation rate of 28 mm/h or more. Written informed consent was obtained from each one. As controls, PB samples were obtained from 3 age-matched healthy women.

Table 1.   Clinical features of the four RA patients included in the study
at onset
A1F41  4 months+YesNone
A2F37  4 months+YesNone
BF3814 months+YesPrednisone (5mg/day), Hydroxychloroquine (200mg/day)
CF4712 years+YesPrednisone (5mg/day), Methotrexate (10mg/week)

General reagents and monoclonal antibodies (mAbs)

ELISA kits were used to measure the levels of IFN-γ, IL-4, and IL-10 in T-cell clone supernatants (Genzyme Diagnostics, Cambridge, MA, USA), and a radioimmunoassay kit was used to detect IL-2 supernatant levels (Medgenix Diagnostic SA, Fleurus, Belgium). Anti-CD3 mAb was purified from supernatant of hybridoma cells obtained from the American Type Cell Culture Collection (Rockville, MD, USA). Phytohemagglutinin (PHA) was purchased from Gibco-Brl (Gaithersburg, MD, USA), phorbol myristate acetate (PMA) from Sigma (St. Louis, MO, USA) and human recombinant (r) IL-2 (specific activity, 9 × 106U/mg) from Janssen (Beerse, Belgium).

Cell separation, T cell cloning and cytokine production pattern analysis

Control PB and paired PB and SF samples from the patients were collected in preservative free heparin (5U/ml). SF samples were then incubated with 3000 U of hyaluronidase (Sigma) for 30min at room temperature. PB mononuclear cells were isolated by density gradient centrifugation on Ficoll-Hypaque (Lymphoprep, Nycomed, Oslo) and resuspended in RPMI supplemented with 10% foetal calf serum, 4 mmol/l L-glutamine, 100U/ml penicillin, and 100μg/ml streptomycin (complete medium) (Gibco).

Mononuclear cells from both SF and PB were cultured in complete medium with low doses of rIL-2 (25IU/ml) for 14 days before cloning. T cells were then cloned by limiting dilution, as previously described [28]. Briefly, viable expanding cells were isolated by Ficoll-Hypaque density gradient, resuspended in complete medium and finally seeded at 0·3cell/well in 96-well round-bottomed plates (Nunclon, Nunc, Kamstrup, Denmark) in the presence of 1 × 105 irradiated (5000rads) allogeneic PB mononuclear cells as feeder cells, PHA (1:200), and rIL-2 (25IU/ml). Growing microcultures were further expanded at weekly intervals with irradiated feeder cells and rIL-2 for 14 days. Viable cloned T cells were finally transferred into fresh complete medium. In order to measure cytokine levels in the supernatants, T cell clones (1 × 106/ml) were cultured for 24h in complete medium in the presence of PMA (10ng/ml) and anti-CD3 (50ng/ml). Cultures were then centrifuged and supernatants collected, filtered through a 0·22-μm filter and stored at –20°C, until use. The levels of IFN-γ, IL-2, IL-4 and IL-10 in the T-cell clone supernatants were detected as previously described. According to the production of IFN-γ and/or IL-4 above or below 200pg/ml, T-cell clones were identified as Th1 (IFN-γ alone), Th2 (IL-4 alone) or Th0 (both IFN-γ and IL-4) type cytokine-secreting clones. Moreover, supernatant levels of IL-10 > 200pg/ml and of IL-2 > 2IU were considered significant. Cytokine analysis was repeated for the same T-cell clones obtained from the patients A1 and A2. The results were overlapping, thereby providing evidence for functional stability of each clone.

Statistical analysis

The Wilcoxon's two tailed test – normal approximation for paired data, the Spearman's rank correlation coefficient and the simple linear regression were adopted for the statistical analysis of the results. Values of P < 0·05 were chosen for rejection of the null hypothesis.


The clinical features of the patients studied is reported in Table 1. A total of 123 T-cell clones were generated from the four patients (SF/PB: 69/54) and were evaluated for cytokine production. Thirty-six and 27 clones were obtained from the untreated patients with early RA (A1 and A2, respectively), 26 from the treated patient with early RA (B) and 34 from the long-standing RA patient (C). No Th2-type clones (i.e. producing IL-4, but not IFN-γ) were generated from SF or PB; the majority of the clones were Th1 (i.e. secreting only IFN-γ, 61/123, 49·6%) or Th0 (secreting both IL-4 and IFN-γ, 46/123, 37·4%). Th1 clones were prevalent in the PB (Th1/Th0: 37/9), while Th0 clones prevailed in the SF (Th1/Th0: 24/37). In addition, there was a small number of clones (16/123, 13·0%) that did not secrete IL-4 or IFN-γ and that were similarly distributed in PB and SF (8 and 8, respectively).

Significant amounts of IL-2 were produced by almost all Th1-cell clones (57/61, 93·4%) and by all the clones that produced neither IL-4 nor IFN-γ (16/16, 100%). On the opposite, IL-2 was secreted only by few Th0 clones (7/46, 15·2%), all generated from the patient C and which displayed a very high IFN-γ/IL-4 ratio due to strong production of IFN-γ (data not shown).

High levels of IL-10 were produced by the majority of Th0 (37/46, 80·4%), essentially generated from SF, and by half (8/16) IL-4/IFN-γ-not secreting clones (Fig. 1). Only a minority of Th1 clones (16/61, 26·2%) secreted IL-10.

Figure 1.

. Production of IL-10 by T-cell clones from synovial fluid (○) and peripheral blood (•) of the four RA patients according to the Th-phenotype. Horizontal bars indicate the median.

As Fig. 2 shows, the production of IFN-γ and IL-2 was significantly higher in the patient C than in the patients A1, A2 and B in both PB and SF clones. IL-4 was higher in SF than in PB in the early RA patients (A1, A2 and B). The synthesis of IL-2 was lower and that of IL-10 higher in the SF compared to PB clones of the untreated patient A1, while the patient A2 showed significant differences between PB and SF only for IL-10 levels. Furthermore, Fig. 2 also demonstrates that the levels of IL-10 produced by the SF clones of the patients A1 and A2 were higher than those produced by the clones of the other two patients (B and C).

Figure 2.

. Production of (a) IFN-γ, (b) IL-4, (c) IL-2 and (d) IL-10 by T-cell clones from synovial fluid (○) and peripheral blood (•) of the two untreated (A1 and A2) and the treated (B) patient with early RA and of the patient with long-standing RA (C). Horizontal bars indicate the median. IFN-γ, *P < 0·001 versus SF A1, A2, B; ‡P < 0·001 versus PB A1, A2, B. IL-4, *P < 0·05 versus PB A1; †P < 0·05 versus PB A2; ‡P < 0·01 versus PB B. IL-2, *P < 0·001 versus SF A1, A2 B; †P < 0·03 versus SF A1; ‡P < 0·001 versus PB A1, A2, B. IL-10: *P < 0·04 versus SF B,C, PB A1; ‡P < 0·04 versus SF B, C, PB A2.

Figure 3 shows the different pattern of cytokine production by clones obtained in early and long standing disease in patient A1 studied twice: the results are similar to those obtained in the other patients.

Figure 3.

. Production of (a) IFN-γ, (b) IL-4, (c) IL-2 and (d) IL-10 by T-cell clones from synovial fluid (○) and peripheral blood (•) of patient A1 studied at the disease onset (0) and 52 months later (1). IFN-γ, *P < 0·001. IL-4, *P < 0·03, †P < 0·05. IL-2, *P < 0·03. IL-10, * and †P < 0·04.

The procedure employed in this study did not allow to obtain a sufficient number of clones from PB samples of the controls, to evaluate the cytokine production.


The remarkable dichotomy between the presence of an extensive T-cell infiltration of the synovium and the relative failure to detect T-cell derived cytokines is the main reason for questioning the direct contribution of T cells to the pathogenesis of RA. It is possible that the low levels of T-cell-derived cytokines detected in RA SF may reflect an in vivo suppression exerted by the large amounts of cytokines produced by macrophages and synoviocytes [3]. Therefore, the cloning of SF and PB T cells represents one of the best methods that allows the characterization of the cytokine pattern produced by activated T lymphocytes in RA patients. However, it is possible that the method used to expand in vitro T cells from RA PB and synovium can drive a Th1-orientated T-cell differentiation or create in vitro artefacts of cytokine synthesis [26,29]. For example, high concentrations of rIL-2 can favour the proliferation not only of in vivo activated T cells, but also of heterogeneous T-cell populations, such as T-cell receptor γ/δ + cells or α/β + CD8 + lymphokine-activated killer cell precursors, which express constitutively the β-, but not the α-chain, of IL-2R [29–32]. On the opposite, low concentrations of rIL-2, as we have used in this study, may selectively expand T cells expressing high-affinity IL-2R, i.e. mounting both the α- and the β-chain. Following this procedure, already adopted in a previous study investigating oligoarticular juvenile arthritis [33], we obtained clones from both SF and PB of RA patients, but not from the PB of three normal subjects. These findings support our hypothesis that the method used allows the cloning of cells already activated in vivo[34] and the concept that activated T cells are present not only in the inflamed joints, but also in the PB of RA patients, probably indicating a circulation of T cells between the two compartments [5,35–38].

Our data confirmed that both PB and SF derived T-cell clones in RA are in prevalence Th1, since they produce high levels of IFN-γ and IL-2, particularly in patients with long-standing RA. These findings are in agreement with previous studies that reported high percentages of IFN-γ-producing T cells in RA and support the concept of a predominance of pro-inflammatory cytokines, such as IFN-γ, in the chronic phases of the disease [14–16,39,40].

However, although none of the clones generated from either SF or PB was of the Th2 phenotype, i.e. producing IL-4 but not IFN-γ, we found that many of the IFN-γ-producing clones produced also IL-4. The combined production of IFN-γ and IL-4 (Th0 phenotype) was associated with a poor production of the Th1-type associated cytokine IL-2, particularly by the clones with low IFN-γ/IL-4 ratio, and was prevalent in the T-cell clones from SF, thus suggesting a selective recruitment of Th0 cells into the joint.

The demonstration of IL-4-producing T cells in RA joint is of particular interest because of the anti-inflammatory activity of this cytokine [8]. In this setting, our finding of a higher IL-4 production by the SF compared to the PB derived clones from the three patients with early RA, but not from the patient with long-standing disease is very intriguing. This datum, confirmed by a longitudinal study performed in an early RA patient evaluated again years later, suggests a joint enrichment of in vivo activated T cells that produce IL-4 in the first phases of RA.

The possibility that the T-cell infiltrate and cytokine production in early rheumatoid synovitis may be different from those characterizing chronic phases of disease is still a debated question [41–45]. The previously reported conflicting data on this topic may depend, at least in part, by the definition of early disease adopted in the different studies [46]. In the present investigation, we examined clones from three subjects with early RA, as usually defined [47]. Two of them were selected at the onset of the disease before treatment (patient A1 and A2) and the other one 14 months after disease onset and on treatment (patient B). The importance of this distinction is underlined by the finding that synovial T-cell clones from the two untreated patients secreted higher amounts of IL-10 compared to those from PB and SF derived clones of patient B.The results of our study demonstrate that IL-10 is essentially produced by Th0 clones, but also by clones secreting IL-2 and very low levels of both IL-4 and IFN-γ. The production of IL-10 and IL-2 is reminiscent of a distinct T-helper subset recently described and for which the label of ‘follicular helper T cells’ have been proposed [48,49]. Such T-cell subpopulation is able to provide efficient help for B cells immunoglobulin secretion. It is worth mentioning that T cells characterized by this peculiar cytokine production pattern have been previously described in RA synovitis [25].

The selective accumulation in SF of IL-10-producing T cells in very early RA showed in our study is of particular relevance, since this cytokine is known to down-modulate inflammatory processes [50–52]. The hypothesis that the increased number of IL-10-producing clones might have been an in vitro artefact generated by the cloning procedure, as suggested by previous investigations [26], is unlikely, since we used low concentrations of rIL-2 and no IL-4 and we found different results in the SF and PB derived clones, generated with the same methodology.

In RA chronic synovitis, IL-10 is mainly produced by non-T cells of the macrophage-lineage, but there is evidence for IL-10 secretion also by synovial T lymphocytes [25,34,53,54]. Since it is believed that T cells play a major role in the initial steps of rheumatoid synovitis, while macrophage and synoviocyte activities prevail during the late phases of disease [2,4,5,8,55], we may postulate that IL-10 secreted by activated T cells in the very early phases of RA synovitis exerts a key role in the modulation of the initial inflammatory process.

Taken together, our results show that in vivo activated T cells in RA joint produce significant amounts of IL-4 and IL-10 in the early phases of the disease. However, the selective recruitment into the joint of IL-10-producing T cells appears to be progressively lost, whereas a significant production of IL-4 is maintained even after months of persistent synovitis. Several considerations may be drawn from these findings. First of all, the cytokine profile of the T cells in the RA joint differs from the one present in the PB supporting the idea that the T-cell subsets that accumulate into the joint are not a random sample [38]. Secondly, the demonstration that RA synovial T cells produce cytokines with anti-inflammatory properties confirms the hypothesis that the course of RA synovitis may be expression of a balance between pro- (Th1) and anti (Th2) inflammatory cytokines [8]. At the beginning of inflammatory process, both kind of cytokines would be produced, whereas, with the persistence of the disease, there would be a progressive reduction of Th2-type cytokines with a prevalence of Th1-type cytokines. We could therefore hypothesize that the eventual predominance of Th2- over Th1-type cytokines may lead to spontaneous resolution of synovitis, although we have no data to support this hypothesis. The combined secretion of IL-4 and IL-10 at the onset of synovitis, however, may have important implications in understanding both the beneficial effects of some disease modifying antirheumatic drugs [45,56,57] and the pathogenic mechanisms triggering RA synovitis: indeed the anti-inflammatory activity of IL-4 can be synergistic with that exerted by IL-10 in order to prevent cartilage damage [9,58,59]. The observation that IL-10 production is very high in the SF of the untreated patients with early RA suggests that it may be a major contributor to the endogenous immune suppression that occurs in the first phases of the disease. Our data, however, do not clarify whether these IL-10-producing cells belong to the described regulatory T cell subsets, as the surface expression of the IL-2R might suggest. The in vitro procedure adopted in the present investigation, indeed, is not suitable for the complete growth and development of T cells subpopulations exerting regulatory activity and defined by the production of IL-10, but not IL-4 and IL-2, or by the constitutive expression of IL-2R-α chain (CD25) on CD4 + cells [60,61].

In conclusion, we have shown in the present study that the cells generated from RA joint produce a range of cytokines with pro- and anti-inflammatory properties. Th1-type cytokines appear to prevail in the chronic phases of the disease, whereas activated synovial T cells produce high amounts of the anti-inflammatory cytokines IL-10 and IL-4 at the beginning of the disease. On the other hand, the diminished IL-10 and IL-4 production in chronic RA synovitis may favour disease progression and eventually lead to joint destruction [62]. We believe that our findings help clarifying some features of pathogenesis and progression of RA and may have implications for the design of new therapeutic strategies.