Constitutive STAT3 Phosphorylation in Circulating CD4+ T Lymphocytes Associates with Disease Activity and Treatment Response in Recent-Onset Rheumatoid Arthritis

The aim of the present study was to examine constitutive signal transducer and activator of transcription 3 (STAT3) phosphorylation in circulating leukocytes as a candidate biomarker in rheumatoid arthritis (RA). 25 patients with recent-onset, untreated RA provided samples for whole blood flow cytometric determination of intracellular STAT3 phosphorylation, expressed as relative fluorescence units. The occurrence of constitutive STAT3 phosphorylation was evaluated by determining proportion of STAT3-phosphorylated cells among different leukocyte subtypes. Plasma levels of interleukin (IL)-6, IL-17 and IL-21 were measured by immunoassay, radiographs of hands and feet were examined and disease activity score (DAS28) was determined. Biomarkers were restudied and treatment response (according to European League Against Rheumatism) was determined after 12 months of treatment with disease-modifying antirheumatic drugs. At baseline, constitutive phosphorylation of STAT3 occurred in CD4+ T cells of 14 (56%) patients, CD8+ T cells of 13 (52%) patients, in CD19+ B cells of 7 (28%) patients, and in CD14+ monocytes of 12 (48%) patients. STAT3 phosphorylation levels of CD4+ T cells associated with DAS28, and those of all leukocyte subtypes studied associated with erosive disease. The presence of constitutive STAT3 phosphorylation in CD4+ T lymphocytes, pSTAT3 fluorescence intensity of CD4+ and CD8+ T cells and C-reactive protein (CRP) levels at baseline associated with good treatment response. In conclusion, constitutive STAT3 phosphorylation in circulating CD4+ T cells is common in recent-onset untreated RA and associates with good treatment response in patients characterized by high disease activity and the presence of systemic inflammation.

Introduction corticosteroids. They did not take non-steroidal anti-inflammatory drugs regularly either before or after entering the study.
Laboratory and hospital personnel who did not have autoimmune diseases or immunosuppressive medication served as healthy reference subjects (n = 17).
The study protocol was approved by the Ethical Review Board of the Joint Authority for the Hospital District of Helsinki and Uusimaa, and written informed consent was obtained from each patient.

Clinical evaluation
A comprehensive clinical and laboratory evaluation was undertaken at entry concomitant to blood sampling, and after follow-up time (median 12 months, range 6 to 18 months) to assess outcome. 66/68 joints were evaluated for swelling and pain, patient's global assessment of disease activity was recorded on a 100 mm visual analogue scale, and laboratory measurements including ESR and serum CRP level were logged. Disease activity score using 28 joints (DAS28) was calculated [25]. Treatment response was evaluated using EULAR response criteria [26]. Postero-anterior radiographs of hands and feet were obtained at baseline and presence of erosions was evaluated by an experienced rheumatologist (MLR).

Blood samples
A 4-ml blood sample was taken, at baseline and at follow-up, by venipuncture from the antecubital vein into a Falcon polypropylene tube (Becton Dickinson) supplemented with 400 μl of pyrogen-free acid citrate dextrose solution A (ACD-A, Baxter). Cells for flow cytometry were prepared within 3 hours of blood sampling. Plasma was separated from the remaining blood sample by centrifugation and frozen in aliquots at -80°C.

Preparing leukocytes for flow cytometry
Blood samples were prepared according to protocol by Becton Dickinson [27]. All antibodies were also purchased from BD. The optimal antibody amounts and the compatibility of the antibodies with the permeabilization procedure were tested in preliminary experiments.

Flow cytometry
Flow cytometric data were acquired on FACSCantoII flow cytometer and analyzed with FACS-Diva software (BD BioSciences, San Jose, CA), as described previously [9]. CD4 + and CD8 + T lymphocytes, CD19 + B lymphocytes and CD14 + monocytes were identified and pSTAT3 fluorescence intensity of the respective histograms was determined as relative fluoresence units (RFU). The proportion of pSTAT3 positive (pSTAT3 + ) cells in the patients' samples was determined as described previously [9,10]. In brief, a marker was set on each histogram from a healthy subject's sample obtained within a week from that of a patient's sample so that it encompassed less than but as close as possible to 5% of the events, and the markers were then copied to the respective histograms from the patient's sample. If a marker encompassed more than 5% of the events, the patient was said to be pSTAT3 + for the corresponding leukocyte subset. Each healthy subject studied served as a reference for one to three patients. The use of the proportion of pSTAT3 + cells, as compared to that of fluorescence intensity values, aids to minimize the variation related to phosphospecific flow cytometry [28] and provides a more sensitive and illustrative way to reveal constitutive phosphorylation occurring within leukocyte populations of the patients.

Plasma IL-6, IL-17 and IL-21 determination
Plasma levels of IL-6, IL-17 and IL-21 were determined by enzyme-linked immunoassay (ELISA) by using the reagents from eBioscience Inc (San Diego, CA, USA). The detection limits and inter-assay coefficients of variation were 0.39 pg/ml and 1.9% for IL-6, 0.98 pg/ml and 6.3% for IL-17, and 15.6 pg/ml and 5.7% for IL-21. In the immunoassays, possible influence of rheumatoid factor and related compounds present in the patient samples was tested by adding commercially available heterophilic blocking agent HeteroBlock (Omega Biologicals, Bozeman, MT, USA) in the samples in increasing concentrations up to 600 μg/ml as recommended by Todd et al. [29]. No influence on IL-6 concentrations was found while when measuring IL-17 and IL-21 concentrations, the samples were treated with 300 μg/ml and 150 μg/ml of Hetero-Block blocking reagent, respectively.

Statistical analysis
The data are presented as means with standard deviations (SD), medians with interquartile range (IQR), or counts with percentages. Statistical comparisons were made by using bootstrap type t-test, permutation test, Mann-Whitney U test with exact p values and permutation type analysis of co-variance. However, as sample size was small and some variables were skewed, exact and resampling-based (Monte Carlo permutation and bootstrap) methods were used to achieve significance level and 95% confidence intervals (CI). Correlations were estimated by Spearman's correlation coefficient method. Exact logistic regression models [30] were used to investigate factors related to treatment response. Receiver operating characteristic (ROC) curves were constructed to determine the predictive ability of baseline markers that corresponds to the EULAR response, with bias corrected bootstrap CI. Stata 13.1 (StataCorp LP, College Station, TX, USA) statistical package was used for the analyses.
During the follow-up time, DAS28 score decreased significantly more in patients in whom CD4 + or CD8 + T cells were pSTAT3 + at baseline than in those in whom they were not: the mean ratio of the change in DAS28 (follow-up adjusted for baseline DAS28) was 3.24-fold (95% CI 1.28 to 16.14) higher in patients with pSTAT3 + CD4 + T cells than in patients with pSTAT3 -CD4 + T cells, and 1.61-fold (95% CI 1.09 to 2.56) higher in patients with pSTAT3 + CD8 + T cells than in patients with pSTAT3 -CD8 + T cells.
To further determine the actual improvement during follow-up, EULAR response criteria were used. Patients with good response were categorized as good responders and patients with moderate or no response as non-responders. Thirteen out of the 14 patients with pSTAT3 + CD4 + T cells at baseline reached a good EULAR treatment response and one patient was a   Fig 2) indicating that presence of pSTAT3 + CD4 + T cells in early RA may serve as predictor of good treatment response. Of the 11 patients not having pSTAT3 + CD4 + T cells at baseline, seven were non-responders to therapy and four had a good response (Fig 2, baseline data). As presented in the previous paragraphs, the presence of pSTAT3 + leukocytes was associated with disease activity and treatment response in early RA. Therefore we aimed to investigate if the intensity of pSTAT3 fluorescence (depicting the magnitude of STAT3 phosphorylation) is also a determinant of disease activity and treatment response in our patients proving evidence of quantitative association and potential causality. pSTAT3 fluorescence intensity in CD4 + T lymphocytes correlated positively with DAS28 score at baseline (Table 2). Furthermore, pSTAT3 fluorescence intensity of each leukocyte subtype showed a positive correlation with erosion score, CD8 + T cells showing the strongest correlation ( Table 2).
Circulating levels of CRP and proinflammatory cytokines, such as IL-6, are considered to serve as markers of systemic inflammation. At baseline, pSTAT3 fluorescence intensity of CD4 + T cells and monocytes correlated positively with IL-6 level, that of B cells correlated with IL-17 level, and that of each leukocyte subtype with IL-21 level (Table 2). pSTAT3 fluorescence intensities in CD4 + and CD8 + T lymphocytes at baseline were associated with good treatment response (Table 1), as was serum CRP level. Multivariate exact logistic regression analysis was used to test for the independent explanatory value of pSTAT3 fluorescence intensity for treatment response. After adjusting for CRP level, pSTAT3 fluorescence intensity in CD4 + T cells statistically significantly predicted treatment response but that in CD8 + T cells did not.
During the follow-up time, pSTAT3 fluorescence intensity generally decreased (Fig 2). The decrease was significant among good responders in CD4 + T cells (p = 0.01) and CD14 + monocytes (p = 0.02). Also, there was a significant decrease in ESR (p<0.001), CRP (p<0.001), IL-6 (p<0.001), IL-17 (p<0.001) and IL-21 levels (p = 0.04) among good responders and in IL-17 Table 2. Correlations between baseline characteristics and pSTAT3 fluorescence intensity of lymphocyte subsets and monocytes at baseline.  levels among non-responders (p = 0.01). At follow-up, the pSTAT3 fluorescence intensities in CD4 + T cells and CD14 + monocytes and as well as the levels of ESR, CRP, IL-6, IL-17 and IL-21 of good responders were close to those of non-responders ( Table 3).
The ability of cytokines, pSTAT3 fluorescence intensity, and C-reactive protein to predict EULAR treatment response To evaluate the ability of pSTAT3 fluorescence intensity to predict the EULAR response, ROC curves were generated (Fig 3). The AUC values of pSTAT3 fluorescence intensities of CD4 + and CD8 + T cells were close to the AUC of CRP and slightly higher than those of IL-6 and IL-17.

Discussion
The results show that STAT3 is frequently phosphorylated in circulating leukocyte subsets of patients with recent-onset untreated RA. Our findings confirm the results of previous reports [23,33] and extend them by showing that constitutive STAT3 phosphorylation of CD4 + T cells identifies a subgroup of RA patients characterized by high disease activity, as determined by DAS28, presence of systemic inflammation, as determined by plasma CRP level, and good EULAR response to treatment with DMARDs. Constitutive phosphorylation of STAT3 in CD4 + T cells was recently found to occur in patients with undifferentiated arthritis in whom it served as a biomarker of progression to RA [34]. If this finding and ours are confirmed by other research groups, constitutive STAT3 phosphorylation may aid in developing treatment strategies for individual patients with undifferentiated arthritis and recent-onset untreated RA. At present, constitutive STAT3 phosphorylation of the RA patients' CD4 + T cells is considered to be driven by IL-6 [33,34]. In accordance with this, our results showed a positive correlation between plasma IL-6 levels and pSTAT3 fluorescence intensity in CD4 + T cells and monocytes at baseline. IL-6, a multifunctional cytokine, is associated with several molecular and cellular characteristics of inflammation in RA [35]. Recent studies have further brought up the roles of IL-6 in systemic inflammation in RA, e.g. cytokine and chemokine production of peripheral blood mononuclear cells (PBMC) [36], and the potential of plasma IL-6 levels to serve as a biomarker of structural damage in the joints during the first years of RA (independently of CRP levels) [37]. Also, as IL-6/STAT3 signaling has been shown to promote the differentiation of naïve T cells into B cell helper T cells [38], the activation of STAT3 by IL-6 in CD4 + T cells may expand cellular activation to B cells to enhance the development of RA [37]. A novel finding in the present study was that plasma levels of IL-21 were positively correlated with pSTAT3 fluorescence intensity in all leukocyte subtypes studied. IL-21 modulates, for example, immune responses of both T cells and B cells [38,39]. Our results also showed that plasma level of IL-17 correlates with pSTAT3 fluorescence intensity in B cells. IL-17 is the major contributor in developing the Th17 type of immune response and considered to play numerous roles in the pathogenesis of RA [40]. Like IL21, IL17 is a direct target of pSTAT3 [15]. Clearly, the molecular relationships between constitutive STAT3 phosphorylation in circulating leukocyte subsets and plasma cytokine levels described in the present study, and systemic inflammation remain to be elucidated. In addition to altered cytokine milieu in peripheral blood, constitutive STAT3 phosphorylation in RA may involve aberrations in the expression or activation of intracellular kinases or negative regulators upstream of STAT3 or in transcriptional activity of STAT3. In this context it is of note that elevated levels of STAT3 mRNA in circulating monocytes and CD3 + T cells [33] and downregulation of suppressor of cytokine signaling 3 (a negative STAT3 pathway regulator) in CD4 + T cells [41] have been reported in RA patients. We observed radiological erosions in six of 25 patients with recent-onset untreated RA at the time of diagnosis. A novel finding was that the number of eroded joints correlated with pSTAT3 fluorescence intensity in all leukocyte subtypes studied, and the correlation was strongest for CD8 + T cells. This may reflect the influence of IL-21, as IL-21 has been reported to be up-regulated in the synovium and synovial fluid of patients with RA and to enhance osteoclastogenesis in vitro [42]. In this context it is of note that serum levels of granzyme B, an acidic protease expressed by cytotoxic T lymphocytes, have been reported to be increased in arthritis and positively correlated with erosive RA [4], and that IL-21 increases its expression in CD8 + T cells [43]. Also, additional cytokines may be able to activate both STAT3 activation and granzyme B expression in CD8 + T cells [44]. Altogether, regardless of the mediating mechanism, granzyme B could be one of the factors connecting pSTAT3 + CD8 + T cells and presence of erosions.
Another novel finding in the present study was that pSTAT3 positivity in CD4 + T cells in recent-onset untreated RA is in relationship with successful response to therapy with DMARDs, as evaluated by EULAR response and effectual decrease in DAS28 after one year of treatment. Furthermore, during the follow-up time, STAT3 phosphorylation level in CD4 + T cells and monocytes decreased significantly among good responders, whereas there was no statistical difference between baseline and follow-up STAT3 phosphorylation levels among nonresponders. Accordingly, Anderson et al. [34] have also reported that DMARD treatment depressed constitutive STAT3 phosphorylation. Even though the rationale of the result warrants further studies, an explanation may be provided by the molecular effects of DMARDs that directly counteract STAT3 activation-related phenomena. In this context it is noteworthy that activated STAT3 can activate over 3000 genes in CD4 + T cells [18], including genes involved in cell adhesion and migration to the joints [45,46], as well as genes important for proliferation and cell cycle regulation [17,18,47,48] analogously to malignant cells in which STAT3 is constitutively activated to promote prolonged functioning of dysregulated cells [15,16,[47][48][49][50]. Considering DMARDs, sulfasalazine, for example, interferes with CD11b/ CD18-mediated leukocyte recruitment [51] and chloroquine suppresses cell-cell adhesion mediated by β1 integrins [52]. Furthermore, considering that the levels of systemic inflammation biomarkers significantly decreased among good responders in our study, it is notable that methotrexate alone or in combination with sulfasalazine reduces circulating levels of IL-6 [53], and hydroxychloroquine inhibits IL-6, IL-17 and IL-22 production from PBMC [54]. Altogether, an interesting possibility introduced in the current study is that pSTAT3 positivity in peripheral blood CD4 + T cells could be a marker indicating a disease form of RA that is effectively alleviated by synthetic DMARDs.
Although constitutive STAT3 phosphorylation in CD4 + T cells of RA patients with recentonset disease was clear in the present study and the previous studies [23,33,34], it was not evident in a recent study of RA patients refractory to DMARD therapy [55]. In the study by Isomäki et al. [33], 11/15 patients had long-standing RA. Because DMARD therapy depresses STAT3 phosphorylation, as shown previously [34] and in the present study, differences in DMARDs taken by the patients and disease duration may explain the discrepancy. Another explanation for discrepancy involves the methods used. To minimize ex vivo leukocyte activation, we and the two other groups [33,34] studied whole blood samples whereas Ortiz et al. [55] separated mononuclear cells from blood by density gradient centrifugation. Sample handling ex vivo readily promotes leukocyte activation [56][57][58], which may distort constitutive STAT phosphorylation data. Meticulous sample handling is a prerequisite for the study of constitutive STAT3 phosphorylation.
The present results show that constitutive STAT3 phosphorylation is common in circulating leukocytes of patients with recent-onset untreated RA, associates with active disease when present in CD4 + T cells, associates with presence of erosions when present in CD8 + T cells, and identifies a patient group designated by the presence of systemic inflammation and good response to treatment with synthetic DMARDs when present in CD4 + T cells. Optimally, the whole blood flow cytometric method we used may provide a rapid and easy way to recognize the RA patients in whom aberrant STAT3 signaling takes place. In the future, STAT3 phosphorylation status of leukocytes may, either alone or as part of a larger signaling profile, be determined for designing most beneficial and precisely targeted personalized treatment for patients with recent-onset RA.

Conclusions
Constitutive STAT3 phosphorylation in circulating leukocytes is common in recent-onset untreated rheumatoid arthritis and identifies patients characterized by the presence of systemic inflammation and good EULAR response to DMARD treatment.
Measurement of intracellular STAT3 phosphorylation in circulating T lymphocytes may provide a novel predictive biomarker in RA and warrants further studies.