Inhibitory activity of FOXP3+ regulatory T cells reveals high specificity for displaying immune tolerance in remission state rheumatoid arthritis

Immune regulation status may indicate immunological remission in rheumatoid arthritis (RA). This cross-sectional study aimed to determine the Regulatory T cell (Treg) properties, together with 14 plasma cytokines levels between active RA and clinical remission patients. Peripheral blood (PB) Foxp3+ Treg was collected from RA patients for determination of Treg inhibitory activity using a co-culture system. Other PB T cell types and plasma cytokines were determined by flow-cytometry. The Treg results were analyzed according to the disease activity score-28 (DAS28). Then sensitivity and specificity were calculated for the indication of the remission status. The number and inhibitory activity of Treg are higher in the clinical remission as compared to the active RA (p value < 0.0001). Also, Treg: CD4+CD25+CD127+ cell ratio demonstrates the similar result (p value < 0.05). Treg inhibitory activity is inversely correlated with the DAS28. Specificity and positive likelihood ratio of inhibitory activity for indicating remission status are 92.31% (95% CI 63.97–99.81) and 11.14 (95% CI 1.67–74.14), respectively. Treg inhibitory activity is a promising prognostic marker and probably represents the immunological remission status in RA.

Cell sorting and Foxp3 Treg, T cell subset identification. In brief, the plasma from fresh EDTAblood of each participant was prepared by centrifuge at 1500xg, 4 °C for 5 min and stored at − 80 °C until assayed for cytokines. Peripheral blood mononuclear cells (PBMCs) were isolated using the BD Vacutainer CPT tube for PBMC isolation (centrifuged at 1700×g, 20 °C for 20 min). After washed twice with PBS that PBMC (10-15 × 10 6 ) were incubated with anti-CD8 antibodies, carried by the magnetic column kit for CD8 depletion (X-Zell Biotech, (Thailand)), for 5 min, then washed with PBS, centrifuged at 450×g for 10 min at 4ºC, and incubated with nano-magnetic bead reagent for 10 min. After that, the cells were suspended with 1 mL of the commercial kit buffer, transferred to flow-through CD8+ cell depletion employing the commercial kit magnetic column (X-Zell Biotech, (Thailand)). The CD4+ enrichment was obtained and centrifuged at 500 g for 10 min at 4ºC. Following the CD4+-enriched cells were treated with mouse anti-human fluorescent dye-conjugated antibodies [CD4PEcy7, CD25PEcy5 and CD127PE (Becton Dickinson, USA)] for cell sorting and isolation of Treg (CD4+CD25high+CD127low−) and conventional T cell or Tconv (CD4+CD25−CD127+) with FACSAria III. Both T cell populations were used for co-culture inhibition assay. www.nature.com/scientificreports/ In order to determine by flow-cytometry, the numbers of Foxp3+ Treg cells and T cell subset originally present in whole blood, PBMCs without CD8+ depletion were stained for presence of intracellular protein Foxp3 with rat fluorescent dye-conjugated anti-human Foxp3eflu660. CD4+CD25+Foxp3+ population is defined as Foxp3+ Treg (intra-maker Treg) while T cell sub-populations were extracellular stained as CD4+CD25+CD127+ (activate T cells) and CD4+CD25high+CD127low− (extra-maker Treg) populations. In addition, both ratio of Foxp3 Treg:CD4+CD25+CD127+ population (Foxp3+ Treg ratio) and CD4+CD25high+CD127low−:CD4+CD 25+CD127+(CD4+CD25high+CD127low− ratio) were performed. Overall protocol steps and gating strategies were demonstrated in Fig. 1. Full descriptions of the protocol is given in our previous study (12).

Figure 1.
Method framework and flowcytometry gating strategies. Peripheral blood mononuclear cells (PBMCs) were isolated by their intensity-centrifugation, followed by CD8+ cell depletion employing a magnetic column method. CD4+-enriched cells were stained with mouse anti-human fluorescent dye-conjugated antibodies [CD4PEcy7, CD25PEcy5 and CD127PE for isolation of Treg (CD4+CD25high+CD127low−) and conventional T cell or Tconv (CD4+CD25−CD127+) with FACSAria III. The gating strategies started from debris discrimination (A), followed by lymphocyte (B) and CD4+ gating (C). Treg population (D) and Tconv gating were shown (E, followed by F). Short-term Treg expansion assay was performed for gaining Treg cell number. Beside, Tconv cells were rested by resting assay parallel with the Treg expansion assay for four days at 37 °C under 5% CO 2 atmosphere. Followed by phenotype checking on Treg and Tconv by intracellular Foxp3+ expression using flow-cytometry method. Afterward, Carboxyfluorescein succinimidyl ester (CFSE) was stained in Tconv population for following-up T cell proliferation. Followed by Treg co-culture assay for three days at 37 °C under 5% CO 2 atmosphere, under stimulation of anti-CD3/28 beads and supplemented by 500 IU of Interleukin 2. CFSE stained Tconv proliferation was determined by flowcytometry. The formula for percent Treg inhibition was provided in the figure. This figure was created officially by first author.

Results
Patient's characteristics. The average age of the RA patients is not different between those in clinical remission and active RA ( Table 1). All 14 patients in the clinical remission group were female while only one male was present in the active disease group. In the former group, the range of DAS28 scores were lower than the latter group. The majority of RA patients in remission was symptom-free for > 6 months prior to the recruitment. Three patients were symptom-free for > 2 years. VAS score were 0-30 in the RA remission group and 0-60 in the active RA group. Serology test (RF or ACPA level) was positive for 64% of remission and 85% of active RA groups. The majority of both groups were treated with three-DMARDs and the remaining with two. In the remission group, only one patient was treated with methotrexate alone and no NSAID was prescribed. No biological DMARDs used in all patients. The patients were categorized into four group depending on their DAS28 score as described above. The remission (n = 11), low (n = 8), moderate (n = 6), and high DAS28 (n = 2) groups have their DAS28 score (median ± IQR) followed; 2.24 ± 1.02, 2.75 ± 0.30, 3.90 ± 0.97, and 4.84 ± 0.41, respectively. There was no statistical difference in DAS28 score between the negative vs positive serology patients (2.3 ± 0.7 vs 3.2 ± 1.4, p value = 0.111), especially when separated the patients based on DAS28 score as described above. The DAS28 scores between negative and positive serology in the remission and the low-DAS28 score group were 2.0 ± 0.7 versus 1.9 ± 0.6 (p value = 0.541) and 2.6 ± 0.0 vs 2.8 ± 0.2 (p value = 0.899), respectively. However, all patients in the moderate and high-DAS28 score groups had positive serology results (Supplementary Table 1). In addition, the correlation of DAS28 score and serology was performed (r = 0.311, p value = 0.115). The ESR values (median ± IQR) between the negative and positive serology patients were 24.0 ± 13.3 and 21.7 ± 14.18 mm/h, p value = 0.768. For the three patients who were followed-up at active state and after entered into remission RA, DAS28 score at active state and remission state are 2.99 ± 0.18 and 1.95 ± 0.62, respectively. The age-range is 45-67 years and one of them has negative serology. www.nature.com/scientificreports/ Percent Foxp3 Treg and T cell sub-populations in the whole blood. Intracellular Foxp3 staining was performed for Treg identification and quantification by flow-cytometry. The average percentage of Foxp3 Treg from the total CD4+ T cells in clinical remission group was higher than those of the active RA group, p value < 0.001 (Table 2). In addition, percentage (Median) of Foxp3 Treg cells were 3.07 ± 1.57, 2.35 ± 1.85, 2.09 ± 1.09, and 2.64 ± 1.23 in patients with DAS28 score < 2.6 (remission), ≥ 2.6-3.2 (low DAS28 score), ≥ 3.2-5.0 (moderate DAS28 score; active RA), and ≥ 5.1 (high DAS28 score; active RA), respectively. However, a significant difference (p value < 0.05) in percentage of Foxp3 Treg cells was found only between remission and moderate DAS28 score groups (Supplementary Figure S1). Among the other T cell sub-populations, the only significant finding was the lower average percent of CD4+CD25+CD127+ cells in the remission group as compared to RA active group. Additionally, both Foxp3+ Treg ratio and CD4+CD25high+CD127low− ratio were higher in the remission as compared to the active RA group, p value < 0.05 (Table 2). We found that there are higher Treg and ratios in healthy controls, but lower active T cells, indicating that healthy controls have more regulatory function when compared to active RA. Moreover, the percentage of Foxp3+ Treg and CD4+CD25high+CD127low− were not significantly different when compared with remission RA. These findings consisted that the both parameters in remission RA were not different when compared to healthy controls. Besides, even the CD4+CD25high+CD127low− population in active RA was lower than the healthy controls, but it was not significantly different ( Table 2). In addition, the positive correlation (r = 0.878) of Foxp3+ Treg ratio and CD4+CD25high+CD127low− ratio in all patients was shown in Supplementary Figure S2.
Inhibitory activity of Treg cells. Foxp3 Treg following a short-term expansion period and concurrently rested Tconv cells were co-cultivated for three days. Histograms of inhibitory activity analysis in healthy control, remission RA and active RA patients of Tconv:Treg ratio at 1:10 were shown in Fig. 2A (Fig. 2B).  www.nature.com/scientificreports/ When Foxp3 Treg inhibitory activity was compared within the same patient, inhibitory activity in active RA state (12.2 ± 4.6) was lower than the remission state (39.9 ± 11.9), p value < 0.05 (Supplementary Figure S3B). Moreover, analysis of percent inhibition according to the DAS28 score showed the Treg inhibitory activity is lowest in high disease activity patients (Fig. 2C). For cross-tabulation analysis, a cutoff of Treg inhibitory activity for separating patient-outcome was 35%, which was calculated from the mean difference of percent inhibition between active RA and remission state (Fig. 2D) (Fig. 2E, F). Furthermore, the Treg inhibitory activity was observed in RA patients who were classified based on their received doses of MTX. Percent Treg inhibitory activity (Median ± IQR) in patient who received < 10 mg/week (low dose), 11-17.5 mg/week (intermediate dose) and ≥ 20 mg/week (high dose) are 49.4 ± 20.7, 35.2 ± 25.4 and 18.4 ± 10.6, respectively. We found that Treg activity in < 10 mg/week MTX group is higher than in the other groups. A significant difference was found when compared to ≥ 20 mg/week MTX group (p < 0.01) (Supplementary Figure S4A). Besides, the RA patients who received high and intermediate doses of MTX had more DAS28 score when compared to the low dose group (p < 0.05) (Supplementary Figure S4B). Taken together, the lowest Treg inhibitory activity could be found in the RA patients who had high disease-severity. The levels of those four cytokines and TGF-β in active patient group were significantly higher than healthy control group. Also, the TGF-β in the remission group was significantly higher than those of healthy controls ( Table 2). Box Plot graphic of the significant cytokines is given in Supplementary Figure S5. When compared to the DAS28 scores, the IL-10 and TGF-β level were significantly lower between the remission group and low (≥ 2.  Figure S1). Besides, the other cytokines (IL-2, IL-4, IL-5, IL-9, IL-13, IL-17F, IL-21, IL-22 and INF-γ) are not significantly difference among these groups comparison.

Discussion
Previously, a study in Lupus patients demonstrated that Treg associated with immunological remission of Lupus 14 . However, no study has been discovered in RA patients, in terms of indicating immunological remission status. In our previous study, the method of Treg inhibitory detection was validated for studying the peripheral blood Treg inhibitory function in RA 12 . However, no study has used the Treg inhibitory function to demonstrate the immunological tolerance in remission RA. Foxp3+ Treg properties may reflect the prognosis or prediction of alteration of clinical symptom of RA patients. This study shows that the number of peripheral blood Foxp3+ Treg cells is significantly lower in clinically active as compared to remission RA patients while both demonstrate lower quantity than that of the healthy controls, which is in accordance with the other studies 5 . A previous study reported a negative correlation of Foxp3+ Treg number and DAS28 score in RA patients 15 . However, in our study, when the patients were separated based on their DAS28 score, the number of Foxp3+ Treg cells is significantly decreased only in the moderate DAS28 group (≥ 3.2-5.0) compared to remission RA (Supplementary Figure S1). We discuss that not only the number of DAS28 high group (n = 2) is limitation of this study, but also the plasticity of Treg might also be a critical factor of the Foxp3+ measurement 16 . Nevertheless, the Treg function of high DAS28 group is clearly decreased when compared to low and moderate groups (Fig. 2C). Moreover, the negative correlation between the Treg inhibition vs VAS (r = −0.60) or DAS28 score (r = −0.513) was found in this study (Fig. 1E, F). These correlations clarified that Treg function might correlated with clinical severity rather than the number of Foxp3+ Treg.
Accordingly, The sole numbers of Foxp3+ Treg cells may be insufficient to determine the status of immune suppression in RA patients because Treg plasticity can occur at sites of inflammation 6 . Myeloid-derived suppressor cells (CD14+ HLA-DR-low MDSCs) and cytokines, such as IL-1β, IL-6, IL-17, and TGF-β, are involved in the inter-conversion between Foxp3+-and Th17-Treg cells 6,17 . In addition, increased numbers of IL-17-producing Treg cells are present in RA patients' peripheral blood as compared to healthy individuals 18 . Although this study does not measure Th17-Treg cell numbers, the decreased Foxp3+ Treg: CD4+CD25+CD127+ cell ratio (termed Foxp3+ Treg ratio) was observed in most RA patients, especially in active RA when compared to healthy individuals ( Table 2). The CD4+CD25+CD127+ population indicates not only activating T cells but also effector memory T cells 19 . Such finding additionally indicated an increase in numbers of other T or effector T cells and/ or a decrease in Foxp3+ Treg numbers occurring in RA patients. In addition, a significant correlation of Foxp3+ Treg ratio with CD4+CD25high+CD127low−:CD4+CD25+CD127+ cell ratio (r = 0.878, p value < 0.0001; Supplementary Figure S2) was found in this study. The correlation indicated that using either the intracellular (Foxp3) or extracellular markers for Treg identification can demonstrate the imbalance of Treg and/or effector T cells in RA patients. Moreover, the imbalance of Treg and others T cell was reported in other autoimmune diseases 20 .
Co-culture inhibition assay of Treg cells has been used for determining the function of Treg in many diseases 7,21,22 . Foxp3+ Treg inhibitory activity in such system was lowest in active RA patients and lower in the remission RA patients than the healthy individuals. The inhibitory activity in patients with DAS28 score ≤ 2.6 is significantly higher than patients with moderate and high DAS28 scores (> 3.2). Furthermore, the lowest Treg inhibitory activity was observed in patients who received high dose of MTX (≥ 20 mg/week) when compared  Figure S4B). However, these patients had different combined-DMARDs used.
Currently, no study have tested the effect of MTX on functional peripheral blood Treg isolated from of RA patient. However, an in vitro study shown that MTX has no effects on Foxp3+ expression in healthy controls. Perhaps, the study in vitro may not demonstrating the effect of MTX in RA patient because drug mechanism is different when compared to in vivo study 23 . Also, the Foxp3 expression study is not enough for indicating the Treg function because of T cell plasticity 16 .
In terms of indicating the immunological remission, this study shows high sensitivity (85.71%), specificity (92.31%), LR+ (11.14) and LR− (0.15) of Treg inhibitory activity to differentiate between the remission and active status. As an illustration, those patients with ≥ 35% Treg inhibitory activity are in remission state. In contrast, those with Treg activity less than 35%, are in active state RA.
Previous studies showed slightly impaired inhibitory activity in both early active and active RA patients, which is not significantly different from those well-controlled RA patients 7,8 . These observations might be due to a small number of patients studied and that Treg isolation is performed using only CD4 and CD25 markers. Ehrenstein et al. 24 noted that suppressive function is recovered in RA patients following anti-TNF therapy, inhibiting IFN-γ and TNF-α secretion from autologous effector T cells. In this study, Foxp3+ Treg inhibition property is determined in the same RA patients in both active and remission states following DMARDs treatment. The results clearly show a decreased inhibitory activity of Foxp3+ Treg when RA clinical status alters from remission to active (relapsed) disease. In a follow-up study, it is reported that the Treg number is improved in patients treated with conventional and biological DMARDs. However, the Treg inhibitory activity is not mentioned 15,25 .
Treg function in RA also depends on the presence of a variety of pro-inflammatory cytokines 26 . That is the reason why this study also measured the plasma cytokines. According to our results, the levels of IL-6, IL-10, IL-17A, and TNF-α are elevated in patients with active RA, with significantly higher IL-10 and IL-17A level in patients in the low and moderate disease activity, respectively, as compared to those in remission. While inhibitory function of Treg is decreased in active RA. Reasonably, A study shown that TNF-α binds to Treg tumor necrosis factor receptor type II and down-regulates FOXP3 expression by inducing the expression of protein phosphatase 1. This specifically dephosphorylates Ser418, leading to impaired FOXP3 phosphorylation, and thereby loss of Treg function 26 . IL-10 is an anti-inflammatory cytokine that is secreted by Treg and released as a consequence of high levels of inflammatory cytokines. As an evidence, a supporting study shows that amounts of mRNA expression of IL-10 cytokine family from the PB and synovial fluid mononuclear cells in RA were increased 27 . This is not a surprising finding that the higher levels of IL-6 and IL-17A are found in the active disease state. Both of them are essential mediators for auto-antibody production and Th17 differentiation 28 . Differing from the healthy controls, no significant difference is discerned about TGF-β level between the active and remission RA patients although a slightly but significantly higher level is observed between low disease activity and remission. This finding is well supported by a result from a previous study illustrating that TGF-β signaling in RA synovial tissue is increased in a murine collagen-induced arthritis model 29 . Although, only DAS28 high group (N = 2) is limitation of this study for measured the plasma cytokines, but the level of every cytokines are same approach in between active RA and remission RA groups.
In terms of serology for indicating the immunological remission, this study demonstrated that ACPA/RF positivity do not correlate with DAS28 score (r = 0.311, p value = 0.115). However, all patients with DAS28 ≥ 3.2-5.0 are having ACPA/RF positivity (Supplementary Table 1). A two-year follow-up study on RA has suggested that levels of ACPA and RF do not associate with RA outcome. At one year follow-up, ACPA and RF levels decrease only by 31 and 56%, respectively 3 . In a large cohort survey, ACPA and RF levels are only moderately correlated with the markers of inflammation 30 . A randomized study demonstrates that the ACPA and RF levels are not predictive of joint damage progression in early RA patients 31 . Recently, a multi-biomarker disease activity (MBDA) scoring system has been developed. This score can be used to monitor the disease activity over time. Despite its significant correlation with DAS28, MBDS scores requires measurement of 12 serum proteins as contributing parameters for score calculation 32 .
In summary, this study demonstrates that the degree of Foxp3+ Treg inhibitory activity of RA patients in remission is higher than that of active RA. As well, it is inversely correlated to the disease activity score-28 in patients with moderate and high disease severity. Moreover, it also provides a high specificity and LR+ for indicating the remission status in RA patients. TGF-β and IL-10 levels are significantly elevated in active state. In addition, the number of Foxp3+ Treg cells and Foxp3+ Treg ratio are higher in remission as compared to active state. Taken altogether, Foxp3+ Treg inhibitory activity reflects the restoration of the immune system imbalance in patients with active RA, rendering its potential as a prognostic marker and representative of immunological remission in RA. There may be some possible limitations in this study, the age median of healthy control is younger than the remission RA and active RA patients which probably influences the human peripheral blood naturally occurring Treg declines with age including the cytokine expression pattern called inflamm-aging that underpin most major age-related disease. Fairly, for indicating the immunological remission status, we mainly focus on the Treg inhibitory activity of the remission RA and active RA patients in which the age-range of these groups is the same. However, further study involving a larger cross-sectional or cohort as well as age-matched control are warranted given the promising results achieved.