Facilitated expansion of Th17 cells in lupus nephritis patients

Summary

The objective of this study was to investigate the mechanisms of T helper type 17 (Th17) expansion in lupus nephritis (LN) patients, and to determine whether or not it is associated with impaired function of regulatory T cells (T_reg). Major effector subsets of peripheral blood CD4⁺ T cells were assessed by flow cytometry in 33 LN patients with different activity of the disease and 19 healthy controls. The percentage of circulating Th17 cells was increased in LN (median = 1·2% of CD4⁺ compared to 0·6% in the control group, P < 0·01), while T_reg cells remained unchanged (12·3 versus 12·1% in controls), resulting in a significantly lower T_reg/Th17 ratio. Th17 expansion in the patient group was not related to LN activity, renal histology or blood and urine inflammatory biomarkers, but has been associated with a higher cumulative dose of cyclophosphamide. T_reg cells in LN displayed mainly effector memory phenotype and expressed higher levels of transforming growth factor (TGF)‐β; however, their suppressant activity in lymphocyte proliferation assay was diminished compared to controls (~fourfold, P < 0·05). Co‐culture of T_reg and conventional CD4⁺ T cells resulted in marked suppression of the Th1 subset in both of the groups studied, but also in a potent expansion of Th17 cells, which in LN was twofold higher, as in controls (P < 0·05). In conclusion, our results demonstrate that Th17 expansion in LN is not increased during disease exacerbation, but is related to chronic immunosuppressive therapy. This immune signature is probably linked to the abnormal function of T_reg cells, which were less suppressive in LN patients and even facilitated differentiation of Th17 cells.

Introduction

Systemic lupus erythematosus (SLE) is an autoimmune disorder characterized by aberrant activation of autoreactive T and B cells, tissue deposition of immune complexes and subsequent inflammation with organ damage [1]. Renal involvement, called lupus nephritis (LN), remains the most important predictor of disease severity and mortality in SLE [2]. Recent evidence has suggested that T helper type 17 (Th17)‐mediated responses might be involved in the pathomechanism of immune‐mediated glomerular diseases [3]. Th17 cells constitute a distinct lineage of T helper cells characterized by the expression of transcription factor RAR‐related orphan nuclear receptor (ROR)γt and by the production of proinflammatory cytokines [e.g. interleukin (IL)‐17A and IL‐17F] [4]. Apart from their protective role in infections, Th17 cells have been linked with pathogenesis of various autoimmune disorders [5]. Nevertheless, the role of the Th17‐axis in systemic lupus erythematosus (SLE) has not been resolved, due to large discrepancies in both experimental models and human studies. Animal models of LN have documented involvement of the Th17 response in autoimmune renal damage [6]; however, if repeated on an IL‐17A‐deficent background they did not confirm the crucial role of Th17 cells in nephritis [7]. Similarly, several reports have demonstrated increased systemic levels of IL‐17A [8, 9, 10, 11, 12] and a higher fraction of circulating Th17 cells in human SLE [13, 14, 15]. However, there were also studies which did not show significant over‐expansion of Th17 cells [16, 17, 18, 19]. The role of the IL‐17/Th17 axis in LN is not well understood, as there are contradictory results concerning the contribution of Th17 cells in active renal lesions [19, 20]. Recently, we compared urine mRNA expression in LN, but did not show a clear up‐regulation of signature genes of the Th17 axis in patients with renal flare [21].

As the development of Th17 cells shares many similarities with regulatory T cells (T_reg) and is reciprocally regulated, recent studies point to the imbalance between these two cell subsets as a major immune deregulation occurring in SLE [17, 18]. T_reg lymphocytes constitute a separate lineage of CD4⁺ T cells which maintain immunotolerance by a direct suppressant effect on conventional effector T cells or by secretion of immunoregulatory cytokines [22]. It has been suggested that a T_reg/Th17 imbalance could lead to a putative escape of autoreactive Th17 cells from T_reg control, and thus facilitate autoimmunization [23]. Little is known, however, about how such imbalance develops and whether expansion of Th17 cells is an intrinsic trait of LN or if it develops gradually during the course of the disease.

The aim of our study was to evaluate a fraction of Th17 and T_reg cells in the peripheral blood of LN patients, in search of a potential link between the Th17 immune axis and clinical manifestation of SLE and nephritis activity. Additionally, we investigated the in‐vitro cross‐talk between Th17 and T_reg cells in LN patients in order to determine whether the potentially enhanced systemic Th17 response in LN results from quantitative or qualitative deficiency in the T_reg subset.

Materials and methods

Characteristics of the patients

We enrolled 33 SLE patients who fulfilled the American College of Rheumatology criteria [24] and had clinically overt renal symptoms (clinical characteristics summarized in Table 1). In 29 subjects (88%) LN was confirmed by renal biopsy and staged according to International Society of Nephrology (ISN)/Renal Pathology Society (RPS) criteria [25]. In total, we analysed 16 patients with active LN [active urine sediment, proteinuria > 1 g/24 h or > twofold increase, SLE Disease Activity Index (SLEDAI) > 6 and 17 with inactive disease (stable proteinuria  < 1 g/24 h, SLEDAI  < 6]. In inactive LN the disease was quiescent for > 6 months. All patients were not treated with cyclophosphamide (CTX) or mycophenolate mofetil (MMF) in the preceding 6 months. Nineteen healthy individuals served as a control group. The study was approved by the Ethics Committee of the Jagiellonian University and informed written consent was obtained from all participants.

Table 1.

Characteristics of the subjects studied

	Active LN (n = 16)	Inactive LN (n = 17)	Healthy controls (n = 19)
Age (years)	37 (34–47)	35 (29–47)	34 (27–41)
Sex (female, %)	12 (75%)	14 (82%)	15 (79%)
Age at diagnosis (years)	27.5 (23–32)	23 (22–33)	n.a.
SLE duration (years)	11 (5·5–16·5)	8 (4–16)	n.a.
SLEDAI (score)	21 (14–23)##	2 (0–4)	n.a.
First renal exac./relapse (n)	7/9	n.a.	n.a.
Previous renal exac. (n) range1	3 (1–6)	1 (1–5)	n.a.
LN class II/III/IV/V (unknown)2	0/2/10/2(2)	2/1/11/1(2)	n.a.
Proteinuria (g/day)	3.1 (1.9–5.2)##	0.11 (0.06–0.32)	n.a.
Creatinine (serum, μmol/l)	72 (66–99)	65 (60–82)	71 (65–79)
GFR (ml/min/1·73m2)3	88.4 (71–97)	96 (80–115)	94 (85–99)
C3c (g/l)	0·560 (0·48–0·780)**##	1·04 (0·83–1·19)	1·15 (0·96–1·41)
C4 (g/l)	0·070 (0·04–0·1)**##	0·165 (0·130–0·19)*	0·205 (0·164–0·308)
CRP (mg/l)	17·1 (9·1–25·8)**##	1·8 (0·65–8·1)*	0·74 (0·32–1·1)
ANA (titre), range	10 240 (1280–20 480)#	5120 (160–20 480)	None
Anti‐dsDNA (titre), range	640 ( < 10–10 240)##	20 ( < 10–320)	None
Current systemic GCS (mg/day)4	10 (5–16)#	4 (0–4)	None
Current AZA therapy (%)5	3 (19%)	2 (12%)	None
Previous CTX treatment (yes, %)	7 (44%)##	15 (88%)	None
Cumulative CTX dose (g)	0 (0–9·8)#	9 (5·8–25)	n.a.

LN  =  lupus nephritis; SLE  =  systemic lupus erythematosus; SLEDAI  =  SLE Disease Activity Index; GFR  =  glomerular filtration rate; C3c  =  complement component C3c; C4  =  complement component C4; CRP  =  C‐reactive protein; ANA  =  anti‐nuclear antibodies; GCS  =  glucocorticoids; AZA  =  azathioprine; CTX  =  cyclophosphamide. ¹ Including current in case of active LN group; ² according to International Society of Nephrology (ISN)/Renal Pathology Society (RPS) [25]; ³ according to the Modification of Diet in Renal Disease Study (MDRD) formula; ⁴ adjusted to methylprednisolone; ⁵ at the study entry, and during the preceding 6 months, patients were not treated with other IS drugs [e.g. CTX or mycophenolate mofetil (MMF)]. *P < 0·05 and **P < 0·01 in comparison to controls; ^# P < 0·05 and ^## P < 0·01 in comparison to inactive LN.

Isolation of peripheral blood lymphocytes and cell cultures

Peripheral blood mononuclear cells (PBMC) were isolated from heparinized blood using Histopaque gradient centrifugation (Sigma‐Aldrich, St Louis, MO, USA). For detection of intracellular cytokines, PBMC were stimulated for 5 h with phorbol myristate acetate (PMA) (50 ng/ml) and ionomycin (0·7 μg/ml) in the presence of brefeldin A (10 μg/ml, all reagents from Sigma‐Aldrich) in RPMI‐1640 supplemented with 10% fetal calf serum (Pan Biotech, Aidenbach, Germany). Cells were stained for effector cytokines and analysed by flow cytometry (FC).

For the in‐vitro suppression assay, immunomagnetically separated (Miltenyi Biotec, Bergisch‐Gladbach, Germany) CD4⁺CD25^– (T_conv, conventional) and CD4⁺CD25⁺ (T_reg) cells were cultured (at ratios of 2 : 1, 1 : 1 and 1 : 2, final 0·2 × 10⁶/well) in aCD3‐coated (BioLegend, San Diego, CA, USA) 96‐well plates in X‐VIVO‐15 medium (Lonza, Basel, Switzerland) with 2·5% human AB‐serum (Pan Biotech) and aCD28 (1 μg/ml, BioLegend). Viable cells [7‐aminoactinomycin D (7‐AAD), BD Biosciences, San Jose, CA, USA] were counted at baseline and after 5 days by FC. To reproduce Th17‐like differentiation, T_conv were stimulated with aCD3/aCD28 and cultured without cytokines (control) or in the presence of cytokine mixture (all reagents from R&D Systems, Minneapolis, MN, USA): IL‐1β (final 10 ng/ml), IL‐6 (10 ng/ml), IL‐23 (50 ng/ml) and transforming growth factor (TGF)‐β₁ (1 ng/ml), anti‐interferon (IFN)‐γ and anti‐IL‐4 (6 μg/ml each). In a parallel set of wells purified T_reg cells were added at different ratios. Lymphocytes were restimulated on day 5 with PMA/ionomycin (as described) and analysed by FC. To analyse latent TGF‐β expression by T_reg, PBMC were stimulated for 24 h with aCD3/aCD28 and stained for FC.

Flow cytometry

Aliquots of blood were stained with combinations of the following antibodies (all from BD Biosciences, if not specified): CD45‐V450, CD3‐fluorescein isothiocyanate (FITC), CD4‐peridinin chlorophyll‐cyanin (PerCP‐Cy)5·5, CD8‐phycoerythrin (PE)‐Cy7, CD8‐allophycocyanin (APC)‐Cy7 (BioLegend), CD16/CD56‐PE, CD19‐APC, CD45RA‐PE, CD45RO‐FITC, CD25‐PE, CD127‐PE‐Cy7 (BioLegend), CXCR3‐PE‐Cy7, CCR4‐AlexaFluor‐647, CCR6‐V450, CCR7‐AlexaFluor‐647 and CCR10‐PE (R&D Systems). T_reg cells were identified as CD4⁺CD25^highCD127^low, which was confirmed by co‐expression of forkhead box protein 3 (FoxP3) transcription factor. Absolute cell numbers were calculated based on white blood cell (WBC) counts and cell differential measured by automated haematology analyser. To detect intracellular cytokines, lymphocytes were labelled with CD4‐PerCP‐Cy5.5 and CD8‐APC‐Cy7 (BioLegend), stained for viability (FVS450, BD Biosciences), fixed/permeabilized (Cytofix/Cytoperm Kit, BD Biosciences), and stained for cytokines: IL‐4‐AlexaFluor467 (BioLegend), IFN‐γ‐FITC (BioLegend), IL‐22‐PE‐Cy7 (eBioScience, San Diego, CA, USA) and IL‐17A‐PE (eBioScience). To detect a latent‐TGF‐β/glycoprotein A repetitions predominant (GARP) complex, activated PBMC were surface‐stained with CD4‐PerCP‐Cy5·5, CD127‐FITC, CD25‐APC‐Cy7 (BioLegend), GARP‐eFluor660 (eBioScience) and latency‐associated peptide (LAP)‐PE‐Cy7 (eBioScience) and viable (FVS450 exclusion) T_reg were identified by intracellular staining with FoxP3‐PE (eBioScience). Samples were analysed by multi‐colour FC [FACS Canto II, BD Biosciences] and FACS Diva Software (BD Biosciences).

Serum and urine biomarkers

Levels of serum and urine cytokines were measured by Luminex (Magpix; Luminex Corp., Austin, TX, USA) using predesigned multiplex panels (eBioscience). Urinary sediment mRNA expression was assessed by real‐time polymerase chain reaction (PCR) (43 genes; Applied Biosystems, Foster City, CA, USA) on a 7900HT real‐time PCR platform (Applied Biosystems). Detailed methods were presented in our previous study, which focused on molecular biomarkers of LN [21].

Statistical methods

Statistical analysis was performed with Statistica version 13.1 (Tibco Software Inc., Palo Alto, CA, USA) and GraphPad Prism version 7.0 (GraphPad Software, Inc., La Jolla, CA, USA). Data are presented as medians and 0·25–0·75 quartile range, if not stated otherwise. Differences between groups were determined by a Mann–Whitney U‐ or Fisher’s exact test. In in‐vitro experiments assessing the influence of T_reg or cytokines, statistical significance was determined by analysis of variance (anova) and repeated‐measures two‐way anova or Kruskal–Wallis test, if justified. Spearman’s rank sum test was used to analyse correlations; a P‐value  < 0·05 was considered significant.

Results

Increased fraction of Th17 cells in LN

All LN patients were characterized by a considerable lymphopenia affecting all lymphocyte subsets (Supporting information, Table S1). We observed large variability in the fraction of cytokine‐producing CD4⁺ T cells both in LN patients and controls (Fig. 1a,b). However, only a percentage of Th17 cells (i.e. producing IL‐17A but not IFN‐γ) was increased significantly in both active and inactive LN (median 1·2%) compared to control (median 0·6%, Fig. 1c, Supporting information, Table S2). Nevertheless, a marked increase in Th17 cells (i.e. > 90th percentile as determined in the control group) was seen only in four active and three inactive LN patients. These patients did not differ in demographic and clinical characteristics, except for a ~twofold higher cumulative CTX dose compared to patients with a normal percentage of Th17 cells (Supporting information, Table S2). As expected, absolute counts of cytokine‐producing CD4⁺ T cells were decreased in LN patients due to SLE‐related lymphopenia (Fig. 1d, Supporting information, Table S3). The proliferation potential of Th17 cells was investigated using short‐term cultures of T_conv cells in control conditions or in the presence of cytokines promoting type 17 differentiation (Fig. 1e). Cytokine treatment resulted in a significant increase in the fraction of Th17 cells in LN patients (Fig. 1f,g). Interestingly, the effect of cytokines (i.e. fold increase in the number of Th17 cells compared to control conditions) was exactly the same in LN patients and controls (Fig. 1h). These data were confirmed by FC analysis of the chemokine receptor signature of CD4⁺ memory T cells (Supporting information, Fig. S1a), which showed an increase in the percentage of the Th17‐like subset (i.e. CXCR3^–CCR4⁺CCR6⁺) in LN patients (median 8·4%) compared to controls (5·9%, P < 0·05). There was a marked correlation between corresponding CD4⁺ subsets (e.g. Th17 cells identified either by intracellular cytokines or chemokine receptors) and a similar decline in cell numbers in different study groups (Supporting information, Fig. S1b,c).

Figure 1.

Cytokine production by in‐vitro activated CD4⁺ T cells. (a) Peripheral blood mononuclear cells (PBMC) were stimulated with phorbol myristate acetate (PMA)/ionomycin and intracellulary stained for T effector signature cytokines. Representative flow cytometry dot‐plots and gating strategy used to identify cytokine producing CD4⁺ T cells. (b) Graph showing cumulative percentages of major functional subsets of CD4⁺ T cells in individual lupus nephritis (LN) patients and controls. (c) The percentage of CD4⁺ T cells with certain functional phenotypes: T helper type 2 (Th2), Th1, Th1/17, Th17 (dashed line  =  90th percentile of the control group) and Th22 (Th1/2 cells not shown). Horizontal bars represent medians. *P < 0·05, **P < 0·01. (d) Absolute numbers of CD4⁺ T cell subtypes (peripheral blood). Data presented as fold difference in comparison to median value in the healthy controls group. #P < 0·05 in comparison to the control group; *P < 0·05 in comparison to inactive LN. (e) Th17‐expansion assay. Magnetically separated CD4⁺CD25^– T cells (i.e. conventional CD4⁺ T cells) were stimulated in vitro with aCD3/aCD28 and cultured for 5 days in medium alone or in the presence of cytokines inducing Th17‐differentiation and blocking immunoglobulin (Ig)G (as indicated). Representative flow cytometry dot plots showing distribution of Th17 [interferon (IFN)‐γ IL‐17A⁺], Th1 (IFN‐γ⁺ IL‐17A‒) and Th1/17 (both positive) cells upon culture in Th17‐prone conditions [both graphs show 50 000 events (viable CD4⁺ gate)]. (f) The cumulative numbers (mean values) of cytokine positive CD4⁺ T cells (neg.  =  cytokine‐negative) in different culture conditions (data presented as relative to baseline, i.e. value 2 means twofold increase in the cell count after 5 days of culture). *P < 0·05 significant difference in the number of cytokine‐positive cells (all combined). (g) Percentage of Th17 cells after in‐vitro expansion in cytokine milieu [as in (e)]. *P < 0·05. (h) Comparable increase in the number of Th17 cells during in‐vitro cell culture of CD4⁺ T_conv cells from LN patients and controls. Data presented as mean change (95% confidence interval) in comparison to control conditions. There was a significant (P < 0·01) increase in the number of Th17 cells and a decrease of Th1, Th2 (except active LN) and Th22 cells in cytokine conditions (statistics symbols not shown for clarity).

No linkage between peripheral blood Th17 and LN activity and renal histology status

To investigate the association between the systemic Th17 response and LN clinics we stratified patients based on the median percentage of circulating Th17 cells. Th17^high (≥ 1·2% of CD4⁺) and Th17^low ( < 1·2%) patients did not differ in clinical characteristics (Supporting information, Table S4), except for a trend towards lower proteinuria (P = 0·052) and higher cumulative CTX dose (P = 0·056) in the case of the Th17^high endotype. Similarly, a significant correlation (r = 0.59, P < 0·01) was found between the percentage of Th17‐like cells and cumulative CTX dose (Supporting information, Fig. S1d–f). There was no association between percentage of circulating Th17 cells and other clinical characteristics of LN (Supporting information, Fig. S2) or renal histology (Supporting information, Table S5, Fig. S3). There was only a weak positive correlation between percentage of circulating Th17 cells and Th17‐axis gene expression (e.g. IL17A, CCL20) in the urine sediment (Supporting information, Fig. S4A). Additionally, serum IL‐17A was detected only in two active LN patients; both had an increased percentage of circulating Th17 cells (data not shown). Nevertheless, the most reliable cytokine [e.g. urine chemokine C‐X‐C motif ligand 10 (CXCL10) and chemokine C‐C motif ligand 2 (CCL2)] and mRNA (TBX21, CCL2, CD3G) biomarkers of active LN [21] were not elevated in patients with the Th17^high endotype (Supporting information, Fig. S4b,c). These results indicate that an increased systemic Th17 response in LN only weakly translates into the type 17 signature in the urine, and is not linked with more active disease or certain patterns of renal pathology.

Increased fraction of effector memory T_reg cells in LN

In the next part of the study we analysed whether or not expansion of Th17 subset in LN is associated with functional abnormality of T_reg cells. As shown in Fig. 2a, we did not detect any decline in the percentage of T_reg cells (CD4⁺CD25^highCD127^low) in LN compared to controls. A fraction of CD25^highCD127^low cells expressing the FoxP3 transcription factor was also not decreased in active LN (median 6·9% of CD4⁺) compared to inactive disease (6·5%) and normal donors (5·4%). Interestingly, we observed a switch towards a more mature phenotype of T_reg cells in LN, as the percentage of T_reg with effector memory (EM) phenotype (Fig. 2b,c) was increased both in active (median 47%) and inactive LN (36%) compared to control (28%, P < 0·01 both). As expected, there was a marked decline (~fourfold) in the absolute count of peripheral blood T_reg cells in all LN subjects compared to control (Fig. 2d).

Figure 2.

Regulatory T cell (T_reg) phenotype and function. (a) No difference in the percentage of peripheral blood CD25^hiCD127^lo CD4⁺ cells (T_reg) between lupus nephritis (LN) patients and control subjects. (b) Flow cytometry (FC) gating strategy to identify memory (EM  =  effector memory, CM  =  central memory) subsets of T_reg and conventional T cells (T_conv). (c) Decreased fraction of naive and increase in EM T_reg cells (gated on CD4⁺CD25^hiCD127^lo) in LN patients. *P < 0·05, **P < 0·01. (d) T_reg cell counts (peripheral blood) in LN patients and control subjects. (e) Transforming growth factor (TGF)‐β expression by in‐vitro activated T_reg cells. Peripheral blood mononuclear cells (PBMC) were activated (aCD3/aCD28) for 24 h in the presence of interleukin (IL)‐2, stained for surface latency‐associated peptide (LAP) and glycoprotein A repetitions predominant (GARP) and intracellular forkhead box protein 3 (FoxP3), and analysed by flow cytometry (FC). Dot‐plots show expression of LAP and GARP in CD4⁺FoxP3⁺ cells in different culture conditions. (f) Higher surface expression of GARP and LAP by in‐vitro stimulated lymphocytes from active LN patients. Data presented as mean fluorescence intensity (MFI) in the FoxP3⁺CD4⁺ gate. *P < 0·05, **P < 0·01. (g) Matrix of correlation coefficients (R _S – Spearman) in cross‐comparison of T_reg phenotype [naive, CM and effector memory (EM) (% of T_reg)], and expression of LAP/GARP, with major clinical and laboratory measures of LN. Data from active and inactive LN were combined. Numerical values indicate significant (P < 0·05) correlations.

Surprisingly, in‐vitro activated T_reg cells from LN patients showed very high expression of both latent TGF‐β and its coupling receptor GARP (Fig. 2e,f). This was due probably to a higher abundance of EM cells within the T_reg fraction, as we have shown a positive correlation between percentage of EM T_reg cells and relative expression of latent TGF‐β (r = 0·56, P < 0·01). Additionally, we showed associations between expression of latent TGF‐β and several clinical and laboratory parameters of the disease (Fig. 2g): for example, C3c levels (r = –0·49), SLEDAI score (r = 0·55) or urine protein excretion rate (r = 0·45).

T_reg/Th17 imbalance in LN is not related to laboratory and clinical disease parameters

Comparing the numbers of T_reg and conventional T helper subsets we found that only the ratio of T_reg to Th17 cells was decreased significantly in LN (Supporting information, Table S6). When stratified based on the median value of the T_reg/Th17 ratio, LN patients with lower T_reg/Th17 values (i.e. imbalance promoting Th17 response) did not differ in clinical parameters (not shown) and renal biopsy status (Supporting information, Fig. S5). Levels of serum and urine biomarkers (e.g. urine mRNA expression) were also comparable in patients with high and low T_reg/Th17 ratios ( Supporting information, Fig. S6, transcriptome data not shown).

T_reg cells are less suppressive and facilitate proliferation of Th17 lymphocytes in LN patients

To investigate potential defects in T_reg suppression we first analysed the capacity of T_reg cells to modulate proliferation of conventional CD4⁺ T cells. As shown in Fig. 3a, the suppressant activity of T_reg was significantly lower in LN patients (~10% decreased proliferation in T_conv: T_reg 1 : 1 ratio) compared to healthy subjects (~35% decrease, P < 0·01), with no difference between active and inactive LN (not shown). To further verify which T_conv fraction is predominantly inhibited, we analysed cytokine production by proliferating cells after restimulation with PMA/ionomycin (Fig. 3b). Co‐culture of T_conv and T_reg cells resulted in a marked increase in the fraction of Th17 cells, which was more pronounced in LN patients compared to the control group (Fig. 3c). In control subjects, T_reg cells inhibited mainly the proliferation of CD4⁺ T cells not expressing effector cytokines (i.e. naive), Th1 cells and, to a lesser extent, Th2 cells, while in LN patients we showed predominantly inhibition of Th1 cells, with a somewhat opposite effect on Th2 cells (Fig. 3d). In both study groups T_reg cells markedly facilitated the expansion of Th17 cells (three‐ to eightfold increase), but this effect was ~twofold stronger in LN patients compared to controls (Fig. 3d). A similar but smaller effect of T_reg was seen in cells cultured in the presence of cytokines facilitating Th17 expansion (not shown). Together, these data suggest that expansion of Th17 cells in LN could be related in part to the skewed memory phenotype of T_reg cells or their plasticity. This was supported by a positive correlation of the baseline percentage of EM T_reg with the fraction of Th17 cells in T_conv plus T_reg co‐cultures (Fig. 3e).

Figure 3.

Regulatory T cell (T_reg) suppression in lupus nephritis (LN). (a) Suppression assay. Activated (aCD3/aCD28) CD4⁺CD25^– (T_conv) and CD4⁺CD25⁺ (T_reg) cells were cultured for 5 days at different ratios (triangle). Results are presented as % inhibition of proliferation (compared to T_conv alone). LN data were combined, as there was no difference between active and inactive patients. **P < 0·01 compared to control subjects. (b) Influence of T_reg on CD4⁺ T cell proliferation and cytokine production. Stacked graph summarizes change in numbers of cytokine‐producing T_conv cells (neg.  =  cytokine negative) at day 5 compared to baseline (day 0). Experimental setup as in (a). (c) Percentage of T helper type 17 (Th17) cells in CD4⁺ T cell cultures (day 5) in LN patients (data combined) and control. Two‐way analysis of variance (anova) statistics and post‐hoc tests: *P < 0·05, **P < 0·01 in comparison to the control group. (d) Fold change in the number of cytokine producing CD4⁺ T cell subsets at different T_conv : T_reg ratios [as in (a)]. Data were standardized to experimental control (T_conv alone). *P < 0·05, **P < 0·01 compared to control group (anova and post‐hoc test). (e) Correlation between the percentage of effector memory (EM) T_reg cells and the fraction of Th17 cells in the cell culture at a 1 : 1 T_conv:T_reg ratio.

Discussion

In this study, we analysed functional subsets of peripheral blood T helper cells in a well‐defined cohort of SLE patients with renal disease. LN patients showed an increased fraction of circulating Th17 cells and lower T_reg/Th17 ratio compared to normal donors, regardless of disease activity and leading clinical symptoms. We also demonstrated that expansion of Th17 cells in LN might be related to the abnormal function of T_reg cells.

In our study, the LN endotype with a higher systemic percentage of Th17 cells was not related to more active disease, a specific pattern in renal histology or blood and urine inflammatory biomarkers. Our results are in accordance with previous studies which demonstrated up‐regulation of the Th17‐axis in SLE compared to control, but failed to find an association between Th17 responses and disease activity [9, 10, 11]. Moreover, in the available literature there was usually no association between increased levels of IL‐17A or a higher fraction of Th17 cells and the presence of nephritis in SLE patients [8, 9, 10, 11, 14] with only few studies demonstrating clear expansion of Th17 in more severe proliferative LN [26, 27]. Inconsistencies between various studies may be attributed to different strategies in detection of the Th17 subset, e.g. cytokine expression versus surface phenotype. Here, we used both methodological approaches, and the results were convergent. Altogether, our study indicates that Th17 expansion in LN is a general feature of SLE, but not a characteristic trait of active lupus nephritis.

The contribution of Th17 cells in the pathogenesis of LN is still not fully understood. In previous reports, IL‐17‐producing lymphocytes were detected easily in the kidneys of LN patients and correlated with some urinalysis markers of kidney involvement [20, 28]. The kidney expression of IL‐17A cells was usually increased in patients characterized by severe disease or proliferative nephropathy [12, 15]. Nevertheless, Th17 cells did not constitute the predominant effector T cell fraction in LN lesions, as they were largely outnumbered by Th1 cells both in renal tissues and urine samples [29, 30]. This was confirmed by gene expression studies, which demonstrated a predominant type‐1 immune pattern both in renal biopsies [31] and in the urine sediment of active LN patients [32]. Our recent results corroborated these findings, showing higher mRNA expression of Th1 signature genes (e.g. TBX21) in the urine samples from active LN, with no evident up‐regulation of the Th17 axis [21]. In the current study we re‐analysed urine biomarker data in search for a potential linkage between systemic Th17 response and renal inflammation. It emerged that the percentage of circulating Th17 cells correlated only weakly with urinary expression of Th17‐related genes, with no linkage between systemic Th17 cells and recognized urinary biomarkers of active LN. Additionally, recent studies showed a somewhat inverse correlation between urine biomarkers of the Th17‐axis and LN activity [19, 33, 34]. Taken together, these data indicate that even though the Th17 signature can be detected in renal lesions of LN patients, there is no clear evidence of the essential contribution of Th17 cells in LN exacerbations. It is interesting in this respect that recent blood transcriptome profiling in a large cohort of paediatric SLE patients showed predominant type I IFN and plasmablast signatures in the early stage of the disease, with gradual progression towards a full‐blown inflammatory pattern, marked by neutrophil signature, during the development of nephritis [35]. This observation, together with earlier data on the Th17 to Th1 shift during experimental glomerulonephritis [3], suggests that different immune mechanisms are operational at different stages of LN, with a predominance of Th1‐mediated responses in advanced disease.

The driving force of Th17 cell expansion in only some SLE patients remains unclear. This may be due to repeated stimulation with autoantigens and chronic inflammatory conditions (e.g. increased levels of IL‐1β and IL‐6), which favours Th17 differentiation. Here it is of interest that when we created in‐vitro conditions promoting Th17 differentiation it resulted in marked expansion of Th17 cells in LN. Nevertheless, we used an entire fraction of T_conv cells (both naive and memory cells) for this experiment, therefore the rate of Th17 expansion in vitro probably reflected baseline abundance of this memory subset. As a consequence, the fold increase in the number of Th17 cells in cytokine conditions was exactly the same in LN patients and healthy controls. Alternatively, Th17 expansion in SLE may also be a result of a skewed balance between Th17 and T_reg cells. Both subsets require TGF‐β to induce the expression of their master transcription controllers, RORγt and FoxP3, but their final lineage commitment seems to be regulated reciprocally [36]. Similarly, IL‐2, which is required for the expansion of T_reg cells, inhibits the differentiation of Th17 cells [37]. Additionally, recent studies have provided evidence for considerable phenotype plasticity of both Th17 and T_reg subsets [38]. For example, under pathogenic inflammatory conditions T_reg are able to lose FoxP3 expression and transdifferentiate into Th17 cells [39].

Whereas it is generally accepted that T_reg cells play a crucial role in maintaining self‐tolerance and preventing autoimmunity, the data on their deficiency or potential dysfunction in SLE are largely confounding. Similarly to our previous study in SLE [18], and many other reports [16, 17, 19, 40], we did not observe any decline in the percentage of T_reg cells in LN. In fact, in some patients the fraction of T_reg cells was even expanded, as has been described previously [41, 42]. In this study, SLE‐related lymphopenia affected both conventional CD4⁺ effector T cells and T_reg cells, but the absolute numbers of T_reg cells were reduced markedly in LN, which resulted in a significantly lower T_reg/Th17 ratio. Earlier studies showed similar data, pointing to the imbalance between the two subsets as a phenomenon linked with active disease [17, 27]. Nevertheless, in our study the ratio of T_reg to Th17 cells was not related to disease exacerbation, and did not differ in patients stratified based on clinical data or renal biopsy status.

Another interesting observation of our study is that T_reg cells from active LN patients showed a marked transition towards the EM phenotype compared to inactive disease and control subjects. Effector T_reg cells exert their regulatory actions by direct receptor contact with target cells and partly by secretion of immunosuppressive cytokines, TGF‐β being the most important one [22]. In the current study we showed higher expression of the TGF‐β complex by in‐vitro‐activated FoxP3⁺ T cells from LN patients. This phenomenon was related to the T_reg EM phenotype, and was more pronounced in the active disease. A similar feature of T_reg cells was described by Henriques et al. [16], who showed higher mRNA expression of TGF‐β in sorted T_reg cells from SLE patients. Together, these data suggest that circulating T_reg cells in LN, although limited in number, are still capable of production of high amounts of TGF‐β, and they should have correct suppressant functions. Surprisingly, the T_reg suppression assay revealed significantly lower inhibition of T cell proliferation in LN patients compared to healthy subjects, which is in line with earlier reports showing impaired T_reg functions in SLE subjects [43]. At first sight these observations are counterintuitive; however, they suggest potential differences in sensitivity of effector T cells subsets to the inhibitory effects of T_reg.

To verify this hypothesis, we next determined whether Th17 cells in LN are actually prone to T_reg‐mediated suppression. We cultured conventional CD4⁺ T cells in the presence of increasing numbers of T_reg cells, and next determined the proliferation potential of various functional (i.e. producing effector cytokines) T helper subsets. This showed that T_reg cells efficiently inhibited the proliferation of Th1 cells both in LN patients and controls. In contrast, Th17 subset expanded significantly in the presence of T_reg cells, and this effect was more pronounced in LN. As we adjusted the number of CD4⁺ T cells in the culture this effect cannot be attributed to SLE‐related lymphopenia, but is probably a characteristic feature of T_reg–Th17 cross‐talk in SLE. Additionally, due to considerable phenotypical heterogeneity and plasticity of T_reg cells, some in‐vitro expanded cells with the Th17 phenotype were probably derived from the T_reg fraction. Unfortunately, we did not track individual cells, so it was not possible to confirm to what extent facilitated in‐vitro expansion of IL‐17A‐expressing cells was due to enhanced proliferation of bona fide Th17 cells, or facilitated transdifferentiation of T_reg cells into the IL‐17A‐producing ‘ex‐T_reg’ subset. Regardless of which of these mechanisms dominated, these data indicated that SLE‐related expansion of Th17 cells and imbalance in T_reg/Th17 may result in part from the facilitated differentiation of cells with Th17‐like effector functions. Interestingly, recent studies confirmed that in certain conditions T_reg actually promoted Th17 cells and enhanced their proinflammatory functions [44]. For example, during infections T_reg cells not only induced expansion of Th17 cells, but also converted themselves into IL‐17A‐producing lymphocytes which, together, facilitated subsequent clearance of various pathogens [45, 46].

We did not detect any significant linkage between the frequency of circulating Th17 cells (or T_reg/Th17 cell imbalance) and disease activity status. Additionally, we showed large heterogeneity in fraction of functional T helper cell subsets, with only a small group of LN patients who showed a truly high expansion of Th17. Such heterogeneity is due probably to a natural variance in LN, or may arise from enrolment of patients at different disease stages or treatment modes. Interestingly, we noted a positive correlation between Th17 percentage and cumulative CTX dose. As in our study patients were not treated with CTX in the recent past (< 6 months), this result suggests that previous courses of induction immunosuppressive therapy might have a far‐reaching effect on the differentiation of Th17 and T_reg cells and their respective balance. This is in line with a recent report in LN, which demonstrated a significant increase in the frequency of Th17 but no change in ratio to T_reg cells after 6 months of standard induction immunosuppressive therapy with CTX or MMF [19]. Additionally, studies in vasculitis syndromes revealed that initial expansion of Th17 cells did not normalize long after completion of induction therapy, despite achieving complete remission [47, 48]. These effects could be attributed to the different susceptibility of Th17 and T_reg cells to CTX. For example, low‐dose CTX was able to promote expansion of Th17 cells both in vitro and during treatment of cancer patients [49]. Conversely, T_reg cells were found to be extremely sensitive to CTX‐mediated apoptosis [50]. These data suggest that changes in the rate of Th17 expansion, and subsequent T_reg to Th17 imbalance in LN, may depend upon certain therapeutic regimens or on the time that has elapsed since withdrawal from treatment. Moreover, once achieved, imbalance in memory T cell subsets (i.e. after repeated exacerbations and treatment courses) becomes a stable feature of SLE that persists in sustained remission.

In summary, our results indicate that pronounced Th17 expansion and related decline in the T_reg to Th17 ratio affects only a subgroup of LN patients, and is not associated with specific clinical pattern or disease activity. Th17 expansion in LN may be related to the altered function of T_reg cells, which displayed a less suppressive phenotype in vitro and facilitated differentiation of Th17 cells. Our data do not support a concept of a crucial role of Th17‐axis in development of SLE nephropathy, as there was no linkage with renal histology status or LN activity. Th17‐axis is probably involved in the pathogenesis of SLE, but it does not play a major role in renal exacerbation of the disease. We have found a relationship between the Th17 subset and intensity of immunosuppressive therapy in the past. However, the exact mechanisms underlying this observation require detailed examination. Further prospective studies are needed to determine whether Th17 expansion in some LN patients is associated with worse long‐term prognosis, and if novel therapies targeting Th17 axis would bring clinical benefits in this subset of patients.

Author contributions

B. J., J. K. and J. M. designed the study; J. K. and S. B. recruited patients and performed clinical data analysis; B. J., J. K. and H. P. performed the experiments; B. J., J. K. and M. S. analysed the data; B. J., S. B. and J. M. wrote the paper.

Disclosures

The authors declare they have no conflicts of interest.

Supporting information

Fig. S1. Subsets of CD4+ memory T‐cells in LN. (a) Major memory CD4+ T‐cell subsets were identified by the surface expression of chemokine receptors (flow cytometry, additional staining for CD45RA not shown for clarity). (b) Cells were classified as: Th1‐like (CXCR3+ CCR6‐), Th1/17‐like (CXCR3+ CCR6+), Th2‐like (CXCR3‐ CCR4+ CCR6‐ CCR10‐), Th17‐like (CXCR3‐ CCR4+ CCR6+ CCR10‐), and Th22‐like (CXCR3‐ CCR4+ CCR6+ CCR22+). Data presented as medians (quartiles). *P < 0·05. (c) Correlation between percentage of type‐17 T‐cells identified based on intracellular cytokine detection, or chemokine receptor staining. (d) Decline in the number of memory CD4+ T‐cells (peripheral blood) in LN patients. Data presented as fold difference (mean and 95%CI) in comparison to median value in the healthy controls group. #P < 0·05 in comparison to control group. *P < 0·05 in comparison to inactive LN. (e) Matrix of correlation coefficients (R_S – Spearman) in cross‐comparison of CD4+ T‐cell subsets distinguished by chemokine receptors (% of CD4+), with major clinical and laboratory measures of LN. Active and inactive LN data were combined. Numeric values mark significant (P < 0.05) correlations. (f) Correlation between percentage of Th17‐like cells (% of CD4+) and cumulative dose of cyclophosphamide (CTX, g). (g) Percentage of Th17‐like cells in LN patients (active and inactive LN combined) stratified based on cumulative CTX dose. Horizontal lines represent median values. **P < 0·01, *P < 0·05.

Fig. S2. Matrix of correlation coefficients (R_S – Spearman) in cross‐comparison of cytokine expressing CD4+ T‐cell subsets with major clinical and laboratory measures of LN. Active and inactive LN data were combined. Numeric values mark significant (P < 0·05) correlations.

Fig. S3. No linkage between fraction of systemic Th17 cells and renal histology status. (a) Percentage of circulating Th17 cells (IL‐17A+ IFNγ‐) in LN patients stratified based on disease activity status and renal histology class (ISN/RPS [Ref. 25 in the main body of manuscript]). (b) Th17 cell fraction in LN patients with different activity and chronicity of glomerular lesions (L – low, M – moderate, H – high). (c) Histology patterns in LN patients with contrasting numbers of Th17 cells. LN patients were stratified as Th17‐low and Th17‐high, based on the median percentage of circulating Th17 cells (< or > 1.2% of CD4+). There was no statistical difference in the frequency of certain lesions between the two designated groups. (d) Th17 percentages in patients with crescents and leukocyte infiltration in glomeruli as opposed to those without a lesion. On each graph horizontal bars represent median values. Only patients with III and IV ISN/RPS classes were shown in (b), (c) and (d).

Fig. S4. Correlation between systemic Th17 response and markers of renal inflammation. (a) Matrix of correlation coefficients (R_S – Spearman) in cross‐correlation of major CD4+ T‐cell subsets (% of CD4+) and Treg/Th17 ratio with serum and urine biomarkers. Only most important cytokines linked with active LN were included in the analysis (based on our previous study [Ref. 21 in the main body of the manuscript]). Lower part shows a list of gene transcripts significantly upregulated (red) or downregulated (blue) in urine sediment from active LN patients. Just for comparison we also show genes (black) linked with Th17 response (e.g. RORC stands for RORγc transcription factor). Active and inactive LN data were combined. There was no significant correlation in majority of cross‐comparisons except for weak (R_S~0.4) linkage between systemic Th17 (and Treg/Th17 ratio) and urine gene expression of CCL20, CXCL8 and IL17A. Numeric values mark significant (P<0.05) correlations. (b) Volcano plot showing differential expression of genes (panel of 43 immune genes as listed in Ref. 21) in the urine sediment cells from patients with Th17‐high (i.e. Th17 >median value 1.2% of CD4+) as compared to Th17‐low (<1.2%) endotype, separately for active and inactive disease. The transcriptome profiles did not differ except for moderately increased expression of TGFB1 (red dot) in inactive‐LN patients with Th17‐high endotype. (c) Relative mRNA expression (expressed as relative to GAPDH x10³) of TBX21, CCL2, CD3G, CXCL8, IL17A, and TGFB1 in LN patients with Th17‐high and –low endotypes. TBX21 and CCL2 mRNA were previously found to be the most reliable biomarkers of active LN. Horizontal lines represent medians. *P < 0·05.

Fig. S5. No linkage between ratio of systemic Treg to Th17 cells and renal histology status. (a) Treg/Th17 ratio in LN patients stratified based on disease activity status and renal histology class (ISN/RPS). (b) Treg/Th17 ratios in LN patients with different activity and chronicity of glomerular lesions (L – low, M – moderate, H – high). (c) Histology patterns in LN patients with contrasting values of Treg/Th17 ratio (low <10, high >10). There was no difference in the frequency of lesions between the two groups. (d) Treg/Th17 ratios in patients with or without crescents and leukocyte infiltration in glomeruli.

Fig. S6. No difference in serum (a) and urine (b) levels of cytokines in LN patients with low or high systemic Treg/Th17 ratios. Horizontal lines represent median values. (c) Matrix of correlation coefficients (R_S – Spearman) in cross‐comparison of major T‐helper cell subset ratios with major clinical and laboratory measures of LN. Active and inactive LN data were combined. Most important biomarkers of active LN are marked red (i.e. urine CXCL10 and CCL2). Numeric values mark significant (P < 0·05) correlations. There was no linkage between Treg/Th17 ratio and LN clinics.

Table S1. Basic subsets of peripheral blood T‐cells in LN patients (also stratified into active and inactive groups) and control.

Table S2. Clinical characteristics of active and inactive LN patients with markedly elevated Th17 cells (i.e. >1.7% of CD4+) in peripheral blood.

Table S3. Functional subsets of peripheral blood T‐cells in LN patients and control group.

Table S4. Clinical characteristics of active and inactive LN patients stratified based on the percentage of Th17 cells in peripheral blood.

Table S5. No linkage between percentage of circulating Th17 cells and renal biopsy ststus.

Table S6. Ratio of main subsets of cytokine producing CD4+ T‐cells and Treg cells.