*2.5. DAS-28, Anti-CCP and Rheumatoid Factor (RF) Parameters in Relation to miR-24, -26, -31, -146a, and -155 Expression Levels in Th17 and Treg Cells from RA Patients*

To assess the contribution of the examined miRNAs in the RA phenotype, we stratified our patients based on the disease activity, anti-CCP presence, and RF presence. Based on disease activity, we divided RA patients into patients with active disease with DAS-28 > 5.1 (n = 6) and patients with moderate or low disease activity with DAS-28 ≤ 5.1 (n = 8).

MiR-31 expression was higher in RA patients with DAS-28 ≤5.1. We noticed significantly higher expression levels of miR-31 in Th17 cells from RA patients with DAS-28 ≤ 5.1 (p = 0.002). In Treg cells from RA patients with DAS28 > 5.1, miR-24 expression levels is twice higher than in those with DAS28 ≤ 5.1 (p = 0.048). Other outcomes were not statistically significant, however, it is interesting that in RA Treg cells with DAS28 > 5.1, we observed higher expression levels of miR-146a and miR-155 (Figure 10A,B). Moreover, we also found that miR-24, miR-26, miR-31, and miR-155 were downregulated, while miR-146a were upregulated in Th17 cells from RA patients with DAS28 ≤ 5.1.

**Figure 10.** DAS-28 (**A**,**B**), anti-CCP (**C**,**D**), and rheumatoid factor (RF) (**E**,**F**) parameters in relation to miR-24, miR-26 miR-31, miR-146a, miR-155 expression levels in Th17 (**A**,**C**,**E**) and Treg (**B**,**D**,**F**) cells from RA patients. Data are presented as mean ± SEM. Expression of patients with DAS28 ≤5.1 (**A**,**B**), negative anti-CCP (**C**,**D**), and negative RF (**E**,**F**) was taken as 1. The Mann–Whitney test was used. \* p < 0.05, \*\* p < 0.01.

Downregulation of miR-31 in Treg cells obtained from anti-CCP positive RA patients. Our results demonstrated that in Treg cells from anti-CCP positive RA patients, expression of miR-31 was significantly lower than in anti-CCP negative RA patients (p = 0.006; Figure 10D). In contrast, in Treg cells from anti-CCP positive RA patients, we noticed twice higher expression level for miR-24, 13 times higher for miR-146a, and 5 times higher for miR-155 than in anti-CCP negative RA patients; however, results were not significant. We also noticed that in Treg cells from anti-CCP negative RA patients, expression level of miR-26 is twice more than in anti-CCP positive patients; these outcomes were not significant. We also observed that in Th17 cells from anti-CCP positive RA patients, expression of miR-146a was higher and miR-24 and miR-31 were lower than in anti-CCP negative RA patients; these results were also not significant.

The higher expression level of miR-146a in RA patients with RF. In the present study, we reported almost twice higher miR-24, 2.5 times higher miR-31, and 82 times higher miR-146a (p = 0.02; Figure 10E,F) expression levels in Treg cells from RA patients with RF than in Treg cells from RA patients without RF. We also observed that expression levels of miR-155 were almost two times lower in Treg cells from RA patients with RF than in RA patients without RF, but this difference was not statistically significant. In contrast, expression levels of miR-126 were similar in both RF-positive and RF-negative RA patients. In RA patients with RF, we also observed the upregulation of miR-31 in Treg cells compared to Th17 cells.

MiR-26 and miR-155 May Be Good Potential Biomarkers for RA and OA. To estimate the further potential value of the examined miRNAs individually or as a panel as RA and/or OA biomarkers, receiver operating characteristic (ROC)–AUC (Area Under Curve) analyses were performed. The outcomes are presented in Table 1. MiR-26 after comparison of RA and HCs in Th17 (AUC = 0.75, p = 0.02) and Treg cells (AUC = 0.92, p = 0.0002) showed a prognostic value. We also revealed this potential for miR-26 in OA vs. HCs in Treg cells (AUC = 0.86, p = 0.0013). MiR-155 after comparison of RA and HCs in Th17 (AUC = 0.80, p = 0.006) and Treg cells (AUC = 0.73, p= 0.048) revealed a prognostic value. We also exposed this potential for miR-155 in OA vs. HCs in Th17 cells (AUC = 0.75, p = 0.03).


**Table 1.** Prognostic values of the relative expression level of the analyzed microRNAs in RA and OA patients as well as healthy subjects (HCs) based on the area under the curve receiver operating characteristic (ROC)–AUC (Area Under Curve).

In the next step, we conducted the ROC curve analysis and estimated AUC to assess the diagnostic potential of candidate microRNAs that could distinguish RA and OA patients from healthy controls. Figure 11A–F shows the ROC curve analysis for miR-26 in Th17 and Treg cells in RA, OA, and HCs.

**Figure 11.** ROC curve analysis for miR-26. (**A**) RA vs. HC in Th17; (**B**) OA vs. HC in Th17; (**C**) RA vs. OA in Th17; (**D**) RA vs. HC in Treg; (**E**) OA vs. HC in Treg; (**F**) RA vs. OA in Treg.

The highest AUC value was obtained for miR-26 discriminating RA from healthy controls (AUC 0.92) and also differentiating OA from HCs (AUC 0.86) in Treg cells. The high AUC value was also obtained for miR-26 distinguishing RA from HCs (AUC 0.75) in Th17 cells.

Figure 12A–F shows the ROC curve analysis for miR-155 in Th17 and Treg cells in RA, OA, and HCs. The highest AUC value was obtained for miR-155 discriminating RA from healthy controls in Th17 cells (AUC 0.80) and Treg cells (AUC 0.73) and also differentiating OA from HCs in Th17 cells (AUC 0.75).

**Figure 12.** ROC curve analysis for miR-155. (**A**) RA vs. HC in Th17; (**B**) OA vs. HC in Th17; (**C**) RA vs. OA in Th17; (**D**) RA vs. HC in Treg; (**E**) OA vs. HC in Treg; (**F**) RA vs. OA in Treg.

Setting a cutoff value at the specific level permitted us to receive the maximum sensitivity and specificity values (Table S3). For miR-26 in RA vs. HC, we obtained 93.33% sensitivity and 53.85% specificity in Th17 (cutoff value = 0.022) and 84.62% sensitivity and 92.86% specificity in Treg cells (cutoff value = 0.025). In OA vs. HC, those values were 76.92% sensitivity and 61.54% specificity (cutoff value = 0.020) in Th17; and 84.62% sensitivity and 78.57% specificity (cutoff value = 0.051) in Treg cells. For miR-155 in RA vs. HC, we set a cutoff values at the levels 0.046 in Th17 (80% sensitivity, 61.54% specificity) and 0.118 in Treg cells (92.31% sensitivity, 30.77% specificity), respectively. For miR-155 in OA vs. HC, the cutoff values were 0.070 in Th17 (83.33% sensitivity, 46.16% specificity) and 0.075 in Treg cells (61.54% sensitivity, 30.77% specificity), respectively.

#### *2.6. Multivariable Logistic Regression*

Based on the results of the likelihood ratio test (LRT), multivariable logistic regression with a combination of miR-24, miR-26, miR-31, miR-146a, miR-155 as a predictor vector was used to assess their diagnostic accuracy for the possible diagnostic potential. Figure S2 presents ROC analysis for the combination of miR-26 and miR-155 in Th17 RA vs. HCs and miR-146a in Treg OA vs. HCs. Significant diagnostic potential in RA Th17 cells (AUC = 0.811) has only been demonstrated for the combination of miR-26 and miR-155. In Treg from OA patients, only miR-146a has been revealed as a significant estimator in the model (AUC = 0.4196).

#### **3. Discussion**

In the last years, we observe substantial progress in the field of miRNA-mediated regulation of immune function and T cell development. The clarification of miRNAs role in lymphocytes subtypes development/differentiation can be used to enhance our understanding of the molecular pathways that are involved in immune regulation. Currently, there is limited knowledge about the specific miRNAs participated in the regulation of immune-related molecular processes and to what extent their activity promotes rheumatoid arthritis pathogenesis. Rheumatoid arthritis is an autoimmune disease, which affects joints, and symptoms include pain and lead to disability through cartilage and bone destruction. This study confirmed the major imbalance between Th17 cells and Treg cells in patients with active RA, as was previously reported in the literature [12]. MicroRNAs play a multifaceted role in the development of RA and are part of the complex net of epigenetic interactions. Alternations to these interactions might be results of the disease or a contributing factor in RA pathogenesis [13]. Earlier studies on animal models have also confirmed the role of mir-155 as one of the factors contributing to arthritis development. Results of serum testing in arthritic rats showed an increased level of miR-155; on the other hand, mice lacking miR-155 were resistant to collagen-induced arthritis [14,15]. In the present study, we decided to focus on microRNAs since they are responsible for regulating gene expression of transcriptional factors that impact Th17/Treg balance [2,12]. Understanding the molecular pathway of genetic and epigenetic regulation of Treg and Th17 cells is critical to understand RA etiology and pathogenesis. We are the first, to our knowledge, who analyzed microRNAs from Th17 and Treg cells obtained from RA in comparison to control groups—OA patients also characterized among others by joint damage, but with different pathophysiology and without an autoimmune component and healthy subjects—and correlated them with some transcriptional factors important in Treg and Th17 cells differentiation.

The next focus of our study was to analyze the correlation between microRNAs and mRNAs in Treg and Th17 cells in tested groups used in this study: RA and controls; OA and HCs. In our study, we have shown a correlation between miR-155 and STAT3 as well as between miR-26 and STAT3, SMAD3, and SOCS1 in RA Th17 cells. Che et al. have shown that miR-155 shows the potential to be used as a biomarker [14]. For RA Treg cells, we found a correlation between miR-155 with SMAD3 and SMAD4, which again confirms the possibility to use miR-155 as a biomarker [16–19]. In the present paper, we have also shown a correlation between miR-31 and SMAD3 as well as between miR-26 and SOCS1; to our knowledge, our study is the first showing these correlations in RA patients. However, the correlation between miR-31 and SMAD3 has already been reported by Hu et al. [20]. In Th17 cells from OA patients and healthy subjects, we did not found a correlation between analyzed microRNAs and mRNAs. On the other hand, our results confirmed earlier studies that in OA Treg cells, there exists/occurs/is a correlation between miR-24 with SMAD3; miR-31 with SMAD4; miR146a with SMAD3; and miR-155 with SOCS1, SMAD3, and STAT3 [17,18,21,22]. In accordance with an earlier study, in Treg cells from healthy subjects, we also found a few correlations between miR-24 and SOCS1, miR-126 and SOCS1, miR-155 and SOCS1; and two negative correlations between miR-155 and SMAD3 and STAT3 [18,19,21]. Both earlier studies and our present study indicate miR-155 as a potential diagnostic biomarker for RA/OA detection, because miR-155 has a negative correlation with SMAD3 and STAT3 in healthy subjects, but positive in RA patients and OA patients.

Analysis of selected microRNAs' expression in Treg and Th17 cells confirmed differences between Treg and Th17 cells, which is consistent with other studies [13,23]. We observed that the expression of miR-26 is significantly higher in healthy subjects than in RA patients. MiR-31, like miR-24, demonstrated a significant change in expression between Treg and Th17 cells within the HCs group. Our results for miR-146a and miR-155 confirmed the results of previous studies on the expression of microRNAs in RA [23]. We observed that miR-146a expression level in RA and OA Treg was significantly higher than in Th17 cells. The last studies demonstrated that miR-146a plays a considerable role in various aspects of immunopathogenesis. Our results emphasize the role of miR-146a as an anti-inflammatory agent. The downregulating miR-146a expression in Treg cells may promote their proliferation and alleviate the inflammation process. In comparison with the fact that miR-146a preferably reduces the proinflammatory immune response, miR-155 strengthens inflammation, which is consistent with our observations. In the present study, we observed that miR-155 expression in Th17 cells was higher in HCs than in RA patients and that miR-155 expression was higher in Treg cells from RA and OA patients than in Th17 cells. Treg cells from active RA patients with high disease activity (we have 42% RA patients with DAS-28 > 5.1) expressed elevated levels of SOCS1—a target of miR-155—which negatively regulates inflammatory processes. Moreover, our study confirms the value of miR-146a and miR-155 as an important factor in early detection of RA, but we were also able to show the potential use of miR-26 as a biomarker [17,23–25].

The present study revealed significant correlations between particular microRNAs. In RA Treg cells, we noticed the correlation between miR-31 and miR-24 as well as between miR-31 and miR-155 levels. In RA Th17 cells, correlations were between miR-126 and miR-26, miR-24, and miR-31 levels. OA Treg cells showed correlations between miR-155 and miR-24, miR-155 and miR-26, miR-26, and miR-24. In OA Th17 cells we saw significant correlations between miR-24 and miR-146a, miR-26 and miR-146a, miR-26 and miR-155, miR-31 and miR-155, and miR-146a and miR-155. In Bae et al. meta-analysis, levels of circulating miR-146a were significantly higher in RA patients than in controls. This study suggests the noteworthy role of miR-146a levels in RA proinflammatory processes [26].

Our study revealed some connections between microRNAs' expression levels and clinical parameters. We observed in RA patients with DAS28 ≤ 5.1 significantly higher miR-31 expression levels in Th17 cells and contrarily, in RA patients with DAS28 > 5.1 twice higher miR-24 in Treg cells. We also noticed in RA Treg cells upregulation of miR-155 and miR-146a, when DAS28 > 5.1. MiR-26 was at a similar level, independently of DAS score; however, these results were not statistically significant. Li et al. found that in PBMCs from active RA patients, miR-155 expression is positively correlated with DAS28 [11]. Moreover, miRNA-146a in RA patients was also positively correlated with DAS28 [27]. We did not notice significant differences in the anti-CCP level in RA patients, excluding that miR-31 in RA with positive anti-CCP was downregulated in comparison to the level observed in anti-CCP negative RA patients. Treg cells of RA patients with RF-positive results were characterized by a significantly higher level of miR-146a expression.

Our additional analysis revealed that the miR-26 and miR-155 reveal significant diagnostic potential for RA differentiation from healthy subjects. Moreover, high diagnostic potential has been revealed for the combination of miR-155 and miR-26 in RA Th17 cells. MiR-26a was reported as overexpressed in RA PBMCs and plasma in comparison to HC [7,19,28]. During interleukin (IL)-17 differentiation and T CD4+ cells generation, upregulation of miR-26a occurs. T CD4+ cells are crucial in RA pathology [27]. MiR-155 insufficiency reduces the number of Treg cells and downregulates IL-2 receptor (IL-2R) signaling and phosphorylation of STAT5, causing deficient SOCS1 suppression [29]. Yao et al. found that STAT3 and STAT5 phosphorylations are positively regulated by miR-155. That is probably due to miR-155 blocking the inhibitory effect on phosphorylations of STAT3 and STAT5 mediated by SOCS1 [10]. MiR-155 was assessed as a potential biomarker of response on methotrexate (MTX) in RA patients by AUC analysis by Singh et al. [30].

Our study has some limitations. The sample size should be enlarged. Further studies are necessary to elucidate the exact microRNAs influence on transcriptional factors as well as on Th17 and Treg cells in the etiology and pathogenesis of rheumatoid arthritis. Additional studies can identify and describe the microRNAs and their target genes signaling pathways, and interactions between them and Th17 and Treg cells included in the inflammatory processes, which may cause the development of new RA therapeutics.

#### **4. Materials and Methods**

#### *4.1. Study Population*

In this study, we used blood from 14 patients with RA and controls: 11 patients with OA and the group of 15 healthy controls (HCs), which were age and gender-matched to patients we used in this study. This study meets all criteria contained in the Declaration of Helsinki and was approved by the Ethics Committee of the National Institute of Geriatrics, Rheumatology, and Rehabilitation, Warsaw, Poland (approval protocol number 29 June 2016). All participants gave their written informed consent before enrollment. Patients with RA were recruited from the National Institute of Geriatrics, Rheumatology and Rehabilitation in Warsaw, Poland; and from the Poznan University of Medical Sciences, Poland. All RA patients fulfilled the American College of Rheumatology (ACR 2010) criteria for RA. Patients with OA were recruited from the National Institute of Geriatrics, Rheumatology and Rehabilitation in Warsaw, Poland. OA patients were diagnosed based on characteristic x-ray findings and the absence of features suggestive of inflammatory arthritis and must meet the ACR criteria for OA of the knee. RAand OA patients with an active infection, cancer, or other rheumatological diseases were excluded from the study. The HCs consisted of volunteers without clinical or laboratory signs of autoimmune diseases. The control group was chosen randomly between blood bank donors to match the age, gender, and ethnicity of patients with RA and OA. All participants who donated blood for this study had the same socioeconomic status and were from the same geographic region.

RA patients were qualified based on their physical examination and laboratory tests. The main factor based on which patients were qualified was age, gender, disease duration, swollen joints number, C-reactive protein (CRP), erythrocyte sedimentation ratio (ESR), platelets (PLT), and creatinine. Additionally, we evaluated them to study by presence of rheumatoid factor (RF ≥ 34 IU/mL), presence of anti-CCP antibodies (anticyclic citrullinated peptide autoantibodies, aCCP ≥ 17 IU/mL), disease activity score in 28 joints (DAS-28), visual analog scale (VAS, range 0–100), Larsen score, and information about the treatment were collected at the time of obtaining samples from patients. Demographic and clinical characteristics of subjects are summarized in Table 2 and detailed clinical characteristics of RA patients are summarized in Table 3.

**Table 2.** Demographic and clinical characteristics of the study population. ESR—erythrocyte sedimentation ratio; CRP—C-reactive protein.



**Table 3.** Detailed clinical characteristics of RA patients.
