Phenotypic, Functional, and Gene Expression Profiling of Peripheral CD45RA+ and CD45RO+ CD4+CD25+CD127low Treg Cells in Patients With Chronic Rheumatoid Arthritis

Objective Conflicting evidence exists regarding the suppressive capacity of Treg cells in the peripheral blood (PB) of patients with rheumatoid arthritis (RA). The aim of this study was to determine whether Treg cells are intrinsically defective in RA. Methods Using a range of assays on PB samples from patients with chronic RA and healthy controls, CD3+CD4+CD25+CD127low Treg cells from the CD45RO+ or CD45RA+ T cell compartments were analyzed for phenotype, cytokine expression (ex vivo and after in vitro stimulation), suppression of Teff cell proliferation and cytokine production, suppression of monocyte‐derived cytokine/chemokine production, and gene expression profiles. Results No differences between RA patients and healthy controls were observed with regard to the frequency of Treg cells, ex vivo phenotype (CD4, CD25, CD127, CD39, or CD161), or proinflammatory cytokine profile (interleukin‐17 [IL‐17], interferon‐γ [IFNγ], or tumor necrosis factor [TNF]). FoxP3 expression was slightly increased in Treg cells from RA patients. The ability of Treg cells to suppress the proliferation of T cells or the production of cytokines (IFNγ or TNF) upon coculture with autologous CD45RO+ Teff cells and monocytes was not significantly different between RA patients and healthy controls. In PB samples from some RA patients, CD45RO+ Treg cells showed an impaired ability to suppress the production of certain cytokines/chemokines (IL‐1β, IL‐1 receptor antagonist, IL‐7, CCL3, or CCL4) by autologous lipopolysaccharide‐activated monocytes. However, this was not observed in all patients, and other cytokines/chemokines (TNF, IL‐6, IL‐8, IL‐12, IL‐15, or CCL5) were generally suppressed. Finally, gene expression profiling of CD45RA+ or CD45RO+ Treg cells from the PB revealed no statistically significant differences between RA patients and healthy controls. Conclusion Our findings indicate that there is no global defect in either CD45RO+ or CD45RA+ Treg cells in the PB of patients with chronic RA.

Objective. Conflicting evidence exists regarding the suppressive capacity of Treg cells in the peripheral blood (PB) of patients with rheumatoid arthritis (RA). The aim of this study was to determine whether Treg cells are intrinsically defective in RA.
Methods. Using a range of assays on PB samples from patients with chronic RA and healthy controls, CD31CD41CD251CD127 low Treg cells from the CD45RO1 or CD45RA1 T cell compartments were analyzed for phenotype, cytokine expression (ex vivo and after in vitro stimulation), suppression of Teff cell proliferation and cytokine production, suppression of monocyte-derived cytokine/chemokine production, and gene expression profiles.
Results. No differences between RA patients and healthy controls were observed with regard to the frequency of Treg cells, ex vivo phenotype (CD4, CD25, CD127, CD39, or CD161), or proinflammatory cytokine profile (interleukin-17 , interferon-g [IFNg], or tumor necrosis factor [TNF]). FoxP3 expression was slightly increased in Treg cells from RA patients. The ability of Treg cells to suppress the proliferation of T cells or the production of cytokines (IFNg or TNF) upon coculture with autologous CD45RO1 Teff cells and monocytes was not significantly different between RA patients and healthy controls. In PB samples from some RA patients, CD45RO1 Treg cells showed an impaired ability to suppress the production of certain cytokines/chemokines (IL-1b, IL-1 receptor antagonist, IL-7, CCL3, or CCL4) by autologous lipopolysaccharide-activated monocytes. However, this was not observed in all patients, and other cytokines/chemokines (TNF, IL-6, IL-8, IL-12, IL-15, or CCL5) were generally suppressed. Finally, gene expression profiling of CD45RA1 or CD45RO1 Treg cells from the PB revealed no statistically significant differences between RA patients and healthy controls.
Conclusion. Our findings indicate that there is no global defect in either CD45RO1 or CD45RA1 Treg cells in the PB of patients with chronic RA.
T cells with a regulatory phenotype (i.e., CD41 CD251CD127 low FoxP31) are abundantly present in the inflamed joints of patients with rheumatoid arthritis (RA) (1)(2)(3)(4)(5)(6)(7)(8). However, despite their presence, inflammation persists, thus posing the question as to whether Treg cells are functionally impaired in RA. Evidence that CD41CD251 Treg cells are important in controlling the severity of arthritis comes from experimental mouse studies in which depletion of Treg cells using an anti-CD25-depleting antibody before immunization resulted in exacerbated disease (9,10). Conversely, adoptive transfer of CD41CD251 Treg cells in the early phase of the disease led to a reduction in disease severity (10,11). Additionally, earlier onset of disease and more aggressive disease progression were observed in the K/BxN model of spontaneous arthritis in scurfy mice, a mouse strain that is devoid of Treg cells due to a mutation in the Foxp3 gene and, consequently, develops severe multiorgan inflammation (12). These data suggest that a functional impairment of Treg cells may contribute to chronic joint inflammation.
Indeed, several groups of investigators have shown that peripheral Treg cell function is defective in RA patients (13)(14)(15)(16). It was reported that Treg cells from patients with active RA can suppress the proliferation of Teff cells, but the ability of Treg cells to inhibit proinflammatory cytokine production, such as production of interferon-g (IFNg) and tumor necrosis factor (TNF) by T cells and production of TNF by monocytes, is impaired (13). The inability of Treg cells from RA patients to suppress IFNg production in Teff cells has also been demonstrated by other groups (15)(16)(17). It was proposed that this functional defect may be caused by negative effects of TNF on Treg cell function (14,15), which was supported by the finding that TNF blockade could improve Treg cell function (13)(14)(15)18). However, results from several studies have contradicted the notion that defective Treg cell function contributes to inflammatory arthritis. In nude mice injected with CD25-depleted lymphocyte suspensions, relatively few animals developed signs of polyarthritis under nondisease-inducing conditions (19,20). In addition, in human studies, signs of arthritis were observed in only a few cases of X-linked syndrome of immune dysregulation, polyendocrinopathy, and enteropathy (IPEX), a disease that develops in individuals with a FOXP3 gene mutation (21,22); instead, patients with IPEX present with thrombocytopenia, insulin-dependent diabetes mellitus, diar- * In these analyses, data from the peripheral blood (PB) or synovial fluid (SF) of RA patients were not available for some of the variables, as follows: in the Treg cell frequency and phenotype assays, n 5 1 for the Disease Activity Score in 28 joints (DAS28), n 5 2 for erythrocyte sedimentation rate (ESR), n 5 5 for C-reactive protein (CRP) level and n 5 12 with a CRP level ,5 mg/liter, n 5 3 for rheumatoid factor (RF) status, and n 5 7 RA PB and n 5 3 RA SF for disease duration; in the functional assays, n 5 1 for age, n 5 1 for ESR, n 5 2 for CRP level and n 5 2 with a CRP level ,5 mg/liter, n 5 1 for RF status, and n 5 2 for disease duration. IQR 5 interquartile range; TNF 5 tumor necrosis factor. † Disease-modifying antirheumatic drug (DMARD) use was as follows: in patients analyzed for Treg cell frequency and phenotype, methotrexate (MTX) n 5 15, hydroxychloroquine (HCQ) n 5 3, sulfasalazine (SSZ) n 5 2, MTX/HCQ n 5 4, MTX/SSZ n 5 2, MTX/HCQ/SSZ n 5 4, MTX/HCQ/steroid n 5 1, MTX/nonsteroidal antiinflammatory drug (NSAID) n 5 1, and HCQ/SSZ n 5 1; in patients analyzed for Treg cell function, MTX n 5 8, SSZ n 5 1, MTX/HCQ n 5 1, MTX/ HCQ/SSZ n 5 1, and MTX/steroid n 5 1.
rhea, or thyroiditis (22). These findings suggest that there is no direct correlation between impaired Treg cell presence and/or function and the development of arthritis. Furthermore, several groups, including our own, have shown that Treg cells from the peripheral blood (PB) of patients with RA are intact in their capacity to suppress the proliferation of, or cytokine production by, Teff cells (2,3,5,7,23,24). Moreover, in all studies except one (14) that have investigated CD41CD251 Treg cells in the inflamed joints of patients with arthritis, the findings concur, showing that these cells are functionally intact and are fully capable of suppressing proliferation and cytokine production ex vivo (1)(2)(3)(4)(5)(6)(7)(25)(26)(27). Instead, the dysregulation of Teff cell function in arthritis may be attributed to resistance of activated Teff cells to Treg cell-mediated suppression, rather than to intrinsic defects in Treg cell function (2,26,(28)(29)(30)(31).
Reasons for the conflicting results obtained in the studies investigating PB-derived Treg cells in RA patients are neither well reported nor understood, but appear to not be simply explained by differences in cell isolation procedures or patient characteristics (for review, see ref. 32). Interpretation of the results may, however, be restricted by the fact that most of these earlier studies utilized only a limited range of readout assays, often focusing on suppression of either Teff cell proliferation or cytokine production (mostly, production of IFNg, TNF, or interleukin-17 ). We sought to shed light on whether PB Treg cells are truly dysfunctional in RA by performing an extensive analysis of CD31CD41CD251CD127 low cells (hereafter referred to as Treg cells) in both the CD45RO1 and CD45RA1 T cell compartments, using a wide range of readout systems. Our collective data indicate that there is no global defect in either CD45RO1 memory Treg cells or CD45RA1 naive Treg cells in the PB of patients with chronic RA.

PATIENTS AND METHODS
Patients and healthy volunteers. PB and synovial fluid (SF) samples were obtained from patients with RA recruited from the rheumatology outpatient clinic at Guy's and St. Thomas' Hospital NHS Trust (London, UK). Clinical parameters such as Disease Activity Score in 28 joints (DAS28) (33), erythrocyte sedimentation rate (ESR), C-reactive protein (CRP) level, rheumatoid factor status, medication, age, sex, and disease duration were documented on the day of sample collection ( Phenotype and cell frequency analyses. For analyses of phenotype and frequency of each cell subset, PE-conjugated CD25 (Miltenyi Biotec), APC/Cy7-conjugated CD3, PerCP/ Cy5.5-conjugated CD4, PE/Cy7-conjugated CD39, Alexa Fluor 488-conjugated CD127, Alexa Fluor 647-conjugated CD161, Pacific Blue-conjugated CD45RO, and Alexa Fluor 647conjugated FoxP3 (all from BioLegend) were used. For FoxP3 staining, cells were extracellularly stained, and were then fixed and permeabilized using 13 FoxP3 Perm buffer (BioLegend). Cells were acquired on a BD FACSCanto II and analyzed using FlowJo software (version 7.6.1; Tree Star).
T cell suppression assay. Freshly sorted CD252 Teff cells (2-5 3 10 6 /ml) or, where indicated, CD25 int Teff cells were labeled with 2 mM 5,6-carboxyfluorescein succinimidyl ester (CFSE; Molecular Probes) for 15 minutes at 378C. Labeled cells (5 3 10 4 ) were cultured with sorted Treg cells at various concentrations (1 3 10 4 25 3 10 4 ) and 5 3 10 4 monocytes, in the presence or absence of 100 ng/ml LPS, in a 96-well U-bottomed plate. On day 3, supernatants were collected for analysis of cytokine production. In addition, T cell proliferation was determined by flow cytometry, measuring CFSE fluorescence. Results are expressed as the percentage suppression of T cell proliferation, Figure 1. Frequency and phenotype of CD31CD41CD45RO1CD251CD127 low Treg cells in patients with rheumatoid arthritis (RA). A, Fluorescence-activated cell sorting analysis was used to gate on CD251CD127 low Treg cells (boxed area with percentage value shown) within the CD41CD45RO1 T cell population in peripheral blood mononuclear cells (PBMCs) from RA patients (left). Percentages of these cells were compared between healthy controls (HC) and RA patients (each n 5 42) (center) or between age-matched healthy controls (mean 6 SD age 49 6 2.5 years) and RA patients (mean 6 SD age 49 6 2.4 years) (each n 5 20) (right). B, Expression levels of the indicated surface markers in CD31CD41CD45RO1CD251CD127 low Treg cells were analyzed in PBMCs from healthy controls (n 5 40) and RA patients (n 5 36). FoxP3 expression was measured intracellularly (n 5 23 healthy controls, n 5 30 RA patients). C, The percentage of CD251CD127 low cells within CD41CD45RO1 T cells was determined in paired samples of freshly isolated PB and synovial fluid (SF) from RA patients (n 5 15) (top), and Treg cell frequencies in the RA SF (n 5 10) were assessed for correlation with the Disease Activity Score in 28 joints (DAS28), by Spearman's test (bottom). Symbols represent individual patients. D, PBMCs from healthy controls (n 5 7) and RA patients (n 5 9) were stimulated with phorbol myristate acetate and ionomycin for 3 hours in the presence of GolgiStop, and the percentage of interleukin-17 (IL-17)-positive or tumor necrosis factor (TNF)-positive cells (boxed areas with percentage values shown) within CD31CD41CD142CD45RO1FoxP31 T cells was determined. Representative dot plots are shown. In A, B, and D, symbols represent individual donors; horizontal bars show the mean. Groups were compared by either paired or unpaired t-test. MFI 5 mean fluorescence intensity. calculated using the following formula: (100 2 [% proliferation with Treg cells/% proliferation without Treg cells]) 3 100.
Monocyte suppression assay. Freshly isolated CD141 monocytes (1 3 10 5 ) were cultured with or without sorted CD41CD45RO1CD251CD127 low Treg cells at a 1:1 ratio in the presence of 100 ng/ml LPS and 100 ng/ml anti-CD3. Supernatants were collected on day 3 for analysis by multiplex assay.
Cytokine detection. Supernatants were stored at 2808C until analyzed by enzyme-linked immunosorbent assay (for IL-6 and IL-8; BioLegend) or by the Human Cytokine 25-plex assay (Invitrogen Life Technologies) on the Luminex platform.
Gene expression profiling. Sorted Treg and Teff cell samples were lysed in 500 ml of TRIzol (Invitrogen). Chloroform (100 ml) was added, and the samples were then whirl mixed and incubated for 2-3 minutes at room temperature. Following centrifugation (10,000g for 15 minutes at 4 C), the water phase was further purified using the RNeasy MinElute Cleanup kit (Qiagen). RNA integrity was confirmed on an Agilent Technologies 2100 bioanalyzer using total RNA Nano chips. One hundred nanograms of total RNA was used to prepare the targets, using the 3 0 IVT Express kit (Affymetrix) in accordance with the manufacturer's instructions. Hybridization cocktails were hybridized onto a Human Genome U219 Array plate in a GeneTitan instrument (both from Affymetrix). Chips were scanned and gene expression data were normalized using the RMA algorithm and Bioconductor package "Affy" (available at http://www.bioconductor.org). A custom chip definition file (available at http://brainarray.mbni.med.umich.edu) was used. Gene expression profiling analysis was performed using Qlucore Omics Explorer software, version 3.0. The microarray data have been deposited in the GEO database (GEO accession no. GSE65010).
Statistical analysis. Significance testing was performed with GraphPad Prism software, version 5. Data were tested for normality using the D'Agostino and Pearson omnibus normality test, followed by appropriate parametric or nonparametric testing for significance.

RESULTS
Comparing frequencies of PB CD41CD251 CD127 low Treg cells between RA patients and healthy controls. Frequencies of CD41CD251CD127 low Treg cells (34) in the PB from patients with RA were determined within the CD31CD41CD45RO1 (memory) T cell compartment, since CD41CD25 high Treg cells are predominantly memory cells (35) and because T cells in the SF are of the CD45RO1 memory phenotype (36). We found no significant differences in the percentage of CD251CD127 low Treg cells within CD41CD45RO1 T cells between PB samples from RA patients and those from healthy controls (each n 5 42), even when the groups were age-matched (mean 6 SD age 49 6 2.4 years in RA patients versus 49 6 2.5 years in healthy controls [each n 5 20]) ( Figure 1A). Treg cell frequencies did not correlate with age in either the RA cohort or the healthy control cohort (see Supplementary Figure 2A, available on the Arthritis & Rheumatology web site at http://online-library.wiley.com/doi/10.1002/art.39408/abstract). No sexspecific differences in Treg cell frequencies were found (results not shown). In addition, the frequencies of Treg cells within CD31CD41CD45RO2 T cells (naive Treg cells) were also similar between RA patients and healthy controls (see Supplementary Figure 2B, available on the Arthritis & Rheumatology web site at http://onlinelibrary. wiley.com/doi/10.1002/art.39408/abstract).
We investigated the ex vivo phenotype of CD251 CD127 low cells within the CD31CD41CD45RO1 T cell compartment in PBMCs from RA patients and healthy controls. CD45RO1 Treg cells from RA patients (n 5 36) showed similar expression levels of CD4, CD25, and CD127 as those from healthy controls (n 5 40) (Figure 1B). We found no significant differences in the expression levels or the percentage of cells expressing the ectoenzyme CD39, a Treg cell-specific surface marker (37), or CD161, which identifies a subpopulation of Treg cells with increased potential to express the proinflammatory cytokine IL-17 (38) (Figure 1B  Interestingly, when Treg cells were analyzed in paired PB and SF samples from patients with RA, we observed an increase in the frequency of CD45RO1 memory Treg cells in the SF compared to the PB, and this finding showed a positive correlation with the DAS28 (Spearman's r 5 0.7988, P 5 0.0072; n 5 10) ( Figure 1C). In contrast, the frequency of CD45RO1 Treg cells in the PB did not correlate with the DAS28 (Spearman's r 5 0.02, P 5 0.9; n 5 41) (results not shown).
Taken together, these data indicate that both the frequency and phenotype of naive and memory Treg cells in the PB were similar between RA patients and healthy controls. Memory Treg cells were, however, increased at the site of inflammation, i.e., in the SF relative to the PB, in patients with RA, and the frequency of synovial Treg cells correlated positively with the level of disease activity.
Cytokine profiles of PB Treg cells from RA patients and healthy controls. We investigated whether PB-derived Treg cells in RA patients showed a more proinflammatory cytokine profile, by determining the percentages of IL-17-and TNF-expressing Treg cells ex vivo. We focused on memory Treg cells, since this population is more plastic compared to the naive Treg cell population (39). No significant differences in the percentages of IL-171 or TNF1 cells within the CD31CD41CD45RO1 FoxP31 T cell population were found between PB samples from RA patients and those from healthy controls ( Figure 1D).
To investigate whether CD45RO1 Treg cells from RA patients would show increased cytokine expression following in vitro culture with CD141 monocytes, CD45RO1 Treg cells were sorted from the PB of RA patients (n 5 13) and healthy controls (n 5 10) and cultured in vitro with autologous CD141 monocytes (at a 1:1 ratio) in the presence of anti-CD3 mAb. No significant differences in the percentages of cytokineexpressing Treg cells (expressing IL-17, IFNg, TNF, or IL-10) were observed between RA patients and healthy controls (see Supplementary Figure 3A, available on the Arthritis & Rheumatology web site at http://onlinelibrary. wiley.com/doi/10.1002/art.39408/abstract).
We recently showed that LPS stimulation of CD141 monocytes results in increased secretion of IL-1b, IL-6, and TNF by the monocytes, which thereby drives the expression of IL-17 by Treg cells (8). However, the addition of LPS to the above cultures did not reveal an increased propensity of Treg cells from RA patients to express the proinflammatory cytokines IL-17, IFNg, or TNF as compared to Treg cells from healthy controls (see Supplementary Figure 3B, available on the Arthritis & Rheumatology web site at http:// onlinelibrary.wiley.com/doi/10.1002/art.39408/abstract). Taken together, these data indicate that CD45RO1 memory Treg cells in the PB of patients with RA do not show an enhanced potential to produce proinflammatory cytokines.
Treg cell suppression of T cell proliferation, as well as IFNg and TNF secretion, in autologous Teff cell-monocyte cocultures from patients with RA. We next investigated whether CD45RO1 Treg cells from RA patients are capable of suppressing the proliferation of Teff cells. CD45RA2CD45RO1 Treg cells were sorted from the PB of RA patients (n 5 9) and healthy controls (n 5 8). Autologous FACS-sorted CD41CD45RA2 CD45RO1CD252CD1271 cells (Teff cells) were labeled with CFSE and cocultured with autologous monocytes (at a 1:1 ratio) in the presence of anti-CD3 mAb. CD45RO1 Treg cells were added to the cultures at 3 different cell ratios (as indicated in Figures 2A-D), and Teff cell proliferation was assessed on day 3 by gating on CFSE1 Teff cells. The presence of CD45RO1 Treg cells, both from RA patients and from healthy controls, led to a cell ratio-dependent reduction in the prolifera-tion of autologous Teff cells (Figures 2A and B). Although in cocultures at the 1:0.5 and 1:1 cell ratios, the mean percentage of Treg cell-mediated suppression was slightly lower in RA PB, this was not significantly different from that in healthy control PB ( Figure 2B). The addition of LPS during coculture, to mimic a proinflammatory environment, did not abrogate Treg cell-mediated suppression of Teff cell proliferation in either RA patients or healthy controls (results not shown). CD45RO1 Treg cells from RA patients were also capable of suppressing the secretion of IFNg and TNF in these cocultures (Figures 2C and D). Treg cells from some patients appeared to be less efficient in suppressing cytokine secretion, especially at the lower cell ratio (1:0.5); however, the level of suppression was not significantly different from that mediated by Treg cells from healthy controls. IL-17 secretion was not detectable in any of the conditions. Interestingly, patients whose PB samples showed less efficient Treg cell-mediated suppression of Teff cell proliferation were not necessarily those whose PB samples showed less efficient suppression of cytokine production (e.g., RA patient 5 in Supplementary  Table 1, available on the Arthritis & Rheumatology web site at http://onlinelibrary.wiley.com/doi/10.1002/ art.39408/abstract). Furthermore, less efficient Treg cell-mediated suppression of Teff cell proliferation could also be observed in the PB of certain healthy control subjects (e.g., healthy controls 5 and 6 in Supplementary  Table 1). We did not find an association between clinical parameters (DAS28, CRP level, ESR, disease duration) and the ability of Treg cells from RA patients to suppress proliferation. Notably, Treg cells from the patient with the highest DAS28 were among the better suppressors (see RA patient 9 in Supplementary Table 1).
It was shown that in patients with multiple sclerosis, a functional defect resides within naive Treg cells (40,41). We found that CD45RA1 Treg cells from both RA patients and healthy controls suppressed less efficiently than did their CD45RO1 Treg cell counterparts. However, the overall ability of CD45RA1 Treg cells to suppress CD45RO1 Teff cell proliferation (at a 1:1 ratio of Teff cells to Treg cells) was not significantly different between RA patients (mean 6 SD 21 6 8% suppression; n 5 7) and healthy controls (33 6 8% suppression; n 5 6) (P 5 0.37) (see Supplementary Table 1).
Recently, it was shown that the ability of Treg cells to suppress proliferation of Teff cells with intermediate CD25 expression (CD25 int Teff cells) was reduced compared to their ability to suppress proliferation of CD252 Teff cells. This could be attributed to a more activated phenotype and higher proliferation rate of CD25 int Teff cells (42). We found that the frequencies of CD25 int Teff cells were similar between the PB of RA patients and the PB of healthy controls ( Figure 3A). In both RA patients and healthy controls, CD25 int Teff cells, compared to CD252 Teff cells, contained higher percentages of IL-17-expressing and TNF-expressing cells following stimulation ( Figure 3B), indicating that the CD25 int Teff cell population may contain more activated cells. Treg cells from RA patients were, however, able to effectively suppress the proliferation of autologous CD25 int Teff cells, similar to the findings in healthy controls ( Figure 3C). Taken together, these data show there is no significant difference in the ability of Treg cells to suppress autologous Teff cell proliferation or the secretion of IFNg or TNF in monocyte-T cell cocultures from the PB of RA patients compared to that from healthy controls.
Suppression of monocyte-derived cytokines and chemokines by Treg cells from RA patients. We recently demonstrated that CD141 PB-derived monocytes from RA patients show a more activated phenotype compared to PB monocytes from healthy controls (8). We therefore investigated the ability of Treg cells from RA patients to suppress LPS-induced monocyte-derived cyto-kine and chemokine production. PB monocytes from RA patients (n 5 9) and healthy controls (n 5 8) were cultured with LPS in the absence or presence of autologous CD45RO1 Treg cells for 3 days. LPS-activated PB monocytes from RA patients and those from healthy controls produced similar levels of cytokines/chemokines in the absence of Treg cells (Figure 4). In the presence of Treg cells, although considerable interdonor differences were observed, similar overall levels of suppression were found in the majority of RA patients and healthy controls with regard to suppression of IL-6, IL-8/CXCL8, IL-12, IL-15, IFNg-inducible protein 10/CXCL10, RANTES/ CCL5, and TNF. Although a lack of suppression of individual cytokines/chemokines was observed in some patients (e.g., IL-8 in a patient with RA [open downward triangles in Figure 4]), Treg cells from these same Figure 4. Ability of CD45RO1 Treg cells from patients with RA to suppress lipopolysaccharide (LPS)-induced monocyte (Mono)-derived cytokine/ chemokine production. CD141 monocytes from healthy controls (n 5 6-8) and RA patients (n 5 7-9) were cultured in the presence of 100 ng/ml LPS without or with sorted CD45RO1 Treg cells at a 1:1 ratio. After 3 days, the supernatants were collected and analyzed using a human 25-plex cytokine array or by enzyme-linked immunosorbent assay (for IL-6 and IL-8). Each line joined by symbols represents an individual donor. The broken horizontal line indicates the lower limit of detection. For monocyte chemotactic protein 1 (MCP-1), the dotted horizontal line indicates the upper limit of detection. Data were analyzed by Wilcoxon's matched pairs signed rank test. * 5 P , 0.05; ** 5 P , 0.01 versus absence of Treg cells. IL-1Ra 5 interleukin-1 receptor antagonist; IP-10 5 interferon-g-inducible protein 10; MIP-1a 5 macrophage inflammatory protein 1a; NS5 not significant (see Figure 1 for other definitions).
PERIPHERAL Treg CELLS IN CHRONIC RA Figure 5. Gene expression profiling of CD45RA1 and CD45RO1 Treg and Teff CD41 T cell subsets from the peripheral blood (PB) of healthy controls (HC) and patients with rheumatoid arthritis (RA). CD251CD127 low Treg cells and CD252CD1271 Teff cells were sorted from the CD45RA1CD45RO2 (naive) or CD45RA2CD45RO1 (memory) CD41 T cell compartments from the PB of age-and sex-matched healthy controls (mean age 54 years [range 34-69]) and RA patients (mean age 57 years [range 37-75]) (each n 5 6 female donors). Five of the RA patients were receiving disease-modifying antirheumatic drug treatment (n 5 3 methotrexate [MTX], n 5 1 hydroxychloroquine [HCQ], n 5 1 MTX, HCQ, and sulfasalazine); 1 patient had received no medication. All patients had moderate-to-active disease, with a mean 6 SEM Disease Activity Score in 28 joints of 4.8 6 0.3. A, Principal components analysis (PCA) based on global gene expression levels in the PB shows distinct clusters of the CD45RA1 Teff, CD45RA1 Treg, CD45RO1 Teff, and CD45RO1 Treg cell populations. B, PCA shows that the 4 T cell populations do not cluster differently between healthy controls and RA patients. C, The heatmap, representing results from hierarchical cluster analysis, shows the relative expression levels of previously described Treg signature-associated genes in the different T cell subsets from the PB of healthy controls and RA patients. patients efficiently suppressed the other cytokines/ chemokines. More notable differences were observed with regard to Treg cell suppression of IL-1b, IL-1 receptor antagonist (IL-1Ra), IL-7, macrophage inflammatory protein 1a (MIP-1a)/CCL3, and MIP-1b/CCL4, which were suppressed by CD45RO1 Treg cells from healthy controls, whereas this was not consistently observed in cultures with Treg cells from RA patients (Figure 4 and Supplementary Table 2, available on the Arthritis & Rheumatology web site at http://onlinelibrary.wiley.com/ doi/10.1002/art.39408/abstract). However, again this was not a global defect, as PB samples from ;44% of the RA patients did show an intact ability to suppress these chemokines and cytokines. Monocyte chemotactic protein 1 was not significantly suppressed by Treg cells from either RA patients or healthy controls, and IL-10 was mostly increased in the presence of Treg cells both from RA patients and from healthy controls, as has been previously shown in healthy individuals (43). Taken together, these data indicate that some differences can be observed in the capacity of Treg cells from RA patients to suppress secretion of certain cytokines/chemokines by LPS-activated monocytes. However, this impairment was not detected for all cytokines/chemokines and, furthermore, was not observed in all patients.
Gene expression profiles of PB CD45RA1 and CD45RO1 Treg cells and Teff cells from RA patients and healthy controls. Finally, we assessed whether Treg cells from RA patients are different at a molecular level, using genome-wide gene expression profiling of FACS-sorted PB-derived Treg cells from the CD45RA1 CD45RO2 (naive) and CD45RO1CD45RA2 (memory) T cell compartments from the PB of RA patients and healthy controls. CD45RA1 and CD45RO1 Teff cells were isolated in parallel. As shown by principal components analysis, each of the 4 CD41 T cell subsets was found to be grouped as a distinct cluster based on global messenger RNA expression levels ( Figure 5A). However, cells from RA patients did not cluster separately from cells derived from healthy controls, within either the Treg cell or Teff cell populations ( Figure 5B). In fact, no significant differentially expressed genes were found when comparing RA cells and healthy control cells in any of the 4 populations (false discovery rate 5%, fold change in gene expression 1.5).
To verify that the sorted Treg cells were truly of a regulatory cell type, we investigated the relative expression levels of genes previously associated with a Treg cell phenotype (44). As visualized in the heatmap shown in Figure 5C, CD45RO1 Treg cells as well as CD45RA1 Treg cells could be clearly distinguished from their Teff cell counterparts based on the expres-sion of well-described Treg signature genes, including expression of FOXP3, CTLA4, CD25, CD27, and CD127 ( Figure 5C). In summary, the ex vivo gene expression profiles of Treg cells from the PB of RA patients and healthy controls were highly similar, supporting the other observations made herein showing that Treg cells in the PB of RA patients are not intrinsically different from those in the PB of healthy individuals.

DISCUSSION
In this report we have presented a comprehensive analysis of the frequency, phenotype, cytokine profile, suppressive potential, and gene expression profile of CD41CD251CD127 low Treg cells from both the CD45RO1 and CD45RA1 T cell compartments of the PB from patients with established RA compared to ageand sex-matched healthy individuals. Collectively, our findings indicate that CD41CD251CD127 low Treg cells in the blood of patients with chronic RA are not intrinsically or globally defective.
We show that PB Treg cell frequencies are similar in RA patients and healthy controls in both the CD45RO1 and CD45RO2 T cell compartments, thus extending the findings from previous studies (1,3,4,17). Consistent with the results reported in earlier studies (1)(2)(3)(4)(5)(6)(7)(8)25), we found that the percentages of CD45RO1 Treg cells were significantly increased in the RA SF compared to paired RA PB samples, and notably, this positively correlated with the DAS28. Phenotypically, PB Treg cells showed similar expression levels of CD4, CD25, CD127, CD39, and CD161 in RA patients and healthy controls, with a slight, but significantly, increased expression of FoxP3 in RA patients, both in CD45RO1 Treg cells and in CD45RO2 Treg cells.
Recent data indicate that Treg cells, particularly those with a memory phenotype, can convert into IL-171 Treg cells under proinflammatory conditions (for review, see ref. 45). However, our data showed that CD45RO1 Treg cells from RA patients did not contain significantly enhanced percentages of IL-171, IFNg1, or TNF1 cells, when analyzed ex vivo or after in vitro culture. Although some groups have reported that Treg cells from the blood of RA patients are unable to suppress Teff cell proliferation (14,15,17), the majority of studies reporting on the suppressive function of PB-derived Treg cells in RA have demonstrated intact Treg cell function when the suppression of Teff cell proliferation was assessed (2,5,7,13,23,24). Consistent with these observations, our findings also indicated that in RA patients, PBderived Treg cells from either the CD45RO1 or the CD45RA1 T cell compartment were functionally intact with regard to suppression of Teff cell proliferation. In addition, we demonstrated that Treg cells from RA patients can suppress proliferation of both CD252 and CD25 int Teff cells to a similar extent as healthy control Treg cells, even though CD25 int Teff cells contained increased percentages of IL-171 and TNF1 cells, suggesting a potentially higher activation status of these cells.
Several groups have reported that in RA, PBderived Treg cells are defective in their ability to suppress IFNg and/or TNF production (13)(14)(15)(16)(17), whereas others have reported that Treg cells have intact suppressive function (2,3,7). Our current data indicate that the ability of Treg cells from the blood of RA patients to suppress these cytokines is not significantly impaired. It has been suggested that Treg cells from RA patients are impaired in their ability to suppress IL-17 (18,46). In our assays, IL-17 levels were below the limit of detection. Analysis of the ability of Treg cells to suppress monocyte-derived cytokines/chemokines revealed no consistent differences between RA patients and healthy controls with regard to the suppression of TNF, IL-6, IL-8, IL-12, IL-15, and CCL5, although suppression seemed reduced in some patients. We did find impaired suppression of IL-1b, IL-1Ra, IL-7, CCL3, and CCL4 in RA PB relative to healthy control PB; however, this was not consistently observed in all patients. Importantly, none of the patients displayed impairment in all tested aspects of Treg cell-mediated suppression. Additionally, in healthy controls, some degree of variation in Treg cell suppressive function was observed.
Findings in a recent study suggested that Treg cell suppressive function (defined by suppression of T cell proliferation and IFNg production) was defective in patients with early, active RA who had not yet received therapy (treatment-naive), whereas Treg cell suppressive function was not defective in patients with chronic RA who had been exposed to methotrexate (MTX), although the group numbers were small (n 5 4 per study group) (47). Another recent study in patients with RA suggested that Treg cell function was reduced in MTX nonresponders, and that this defect was related to low CD39 expression (48). The identification of intact peripheral Treg cell function in patients with chronic RA who had received therapy with disease-modifying antirheumatic drugs (DMARDs)/biologic drugs, as described herein and in previous studies (2,7,23,24), could thus be a consequence of effective DMARD therapy. However, other studies have shown no defects in the suppressive capacity of CD41CD25 high Treg cells from patients with early, active RA who were being treated with steroids and were DMARD-naive, as compared to patients who were being treated with DMARDs and had well-controlled disease or healthy controls (5,7). In our study, none of the patients had early RA or were treatment-naive, and therefore further studies are required before conclusive statements can be made regarding Treg cell function in treatment-naive patients with early RA. Nonetheless, the notion that DMARD or biologic therapy may restore defective Treg cell function is important to consider.
It is also possible that disease activity influences Treg cell function. However, when we compared the results with regard to suppression of IFNg and TNF production and T cell proliferation between those patients with highly active disease (DAS28 .5.1; n 5 4) and those with moderate disease activity (DAS28 3.2-5.1; n 5 4) (see Supplementary Table 1), we observed neither considerable differences nor consistent differences between these 2 groups with respect to Treg cell function. This finding clearly indicates that even in patients with active disease, Treg cells are still functional. Taken together with our finding that Treg cells are increased in the inflamed joint and this is correlated with the DAS28, it is reasonable to suggest that Treg cell activity is preserved, and may even be enhanced (8), during inflammation in chronic RA, as a means to dampen collateral damage. This is supported by experimental models showing that Treg cells are increased in number and function during inflammation (49) and that TNF can boost Treg cell function (50).
The notion that PB-derived Treg cells are not defective in patients with chronic RA is strengthened by our findings that neither CD45RA1 nor CD45RO1 Treg cells displayed significantly differentially expressed genes relative to their counterparts from the PB of sexand age-matched healthy controls. One limitation of in vitro suppression assays is that cells are taken out of their inflammatory environment when tested for suppressive activity. Our genome-wide expression arrays were performed on cells that were isolated from patients directly ex vivo, thus closely resembling the in vivo situation. An interesting side observation from our work was that a large proportion of Treg signature genes are predominantly expressed in CD45RO1, but not CD45RA1, Treg cells, suggesting that current investigations of Treg cell-specific signature genes may be biased toward features of the memory Treg cell population.
Importantly, our findings do not exclude the probability that cues from the local inflammatory environment can impede Treg cell-mediated suppression. Indeed, inflammatory cytokines like IL-6 and TNF can negatively modulate FoxP3 activity (14,51). A recent study showed that PB-derived Treg cells from RA patients were unable to suppress Teff cell proliferation or TNF or IFNg secretion by autologous PB Teff cells, and this was attributed to the increased resistance of Teff cells to Treg cell-mediated suppression, rather than to defective Treg cell function (31). Increased resistance of Teff cells to regulation was also reported in a spontaneous experimental arthritis model (49). Furthermore, using crossover experiments, it was shown that activated T cells from the inflamed joints of patients with RA were more resistant to Treg cell-mediated suppression than their PB counterparts (2,26,(28)(29)(30).
Collectively, our data indicate that Treg cells from the PB of patients with chronic RA are not intrinsically different from those in healthy individuals. Combined with previous studies showing intact suppressive function of Treg cells derived from the inflamed joint (1)(2)(3)(4)(5)(6)(7)(25)(26)(27), these findings imply that persistent disease activity in chronic RA cannot simply be attributed to defective CD41CD251CD127 low Treg cell function in the circulation.