Inhibition of TGF-β Signaling Suppresses Th17 Differentiation and Promotes Treg Numbers but does not Reduces Experimental Arthritis

TGF-β is an important growth factor to promote the differentiation of T helper 17 (Th17) and regulatory T cells (Treg). The potential of TGF-β as therapeutic target in T cell-mediated diseases like rheumatoid arthritis (RA) is unclear. We investigated the effect of TGF-β inhibition on murine Th17 differentiation in vitro, on human RA synovial explants ex vivo, and on the development of experimental arthritis in vivo. Methods Murine splenocytes were differentiated into Th17 cells, and the effect of the TGF-βRI inhibitor SB-505124 was studied. Synovial biopsies were cultured in the presence or absence of SB-505124. Experimental arthritis was induced in C57Bl6 mice and treated daily with SB-505124. FACS analysis was performed to measure different T cell subsets. Histological sections were analysed to determine joint inammation and destruction.


Introduction
Rheumatoid arthritis (RA) is an autoimmune disease of unknown etiology, that is characterized by chronic joint in ammation leading to destruction of articular cartilage and bone. CD4+ T cells are found in in ammatory in ltrates of the rheumatoid synovium and are known to play a central role in the development of RA [1][2][3][4] . In patients with RA, the T helper 17 (Th17) and regulatory T cells (Treg) balance is skewed in favor of Th17 cells, which contributes to the development of autoimmunity and in ammation 5 .
Transforming growth factor beta (TGF-β) is a growth factor that can differentiate CD4+ T cells into Th17 cells in the presence of IL-6 or IL-21 6 . TGF-β signaling and STAT3 activate and regulate expression of the transcription factor RAR-related orphan receptor C (RORc), the master regulator of Th17 cells 7,8 . There are three isoforms of TGF-β: TGF-β1, TGF-β2, and TGF-β3 9 . Lymphoid cells produce mainly TGF-β1, which has a controversial role in in ammation and autoimmunity as it both suppresses and induces immune reactions 10 . Three canonical TGF-β receptors have been described: TGF-βRІ, TGF-βRІІ, and the co-receptor TGF-βRІІІ, that together form the TGF-β receptor complex. When TGF-β binds to this receptorcomplex, receptor-regulated Smad (R-Smad) proteins are phosphorylated and translocate to the nucleus to modulate target gene expression 11 .
In diseases such as cancer and atherosclerosis, the role of TGF-β is well de ned, but its role in (experimental) arthritis is still not clear. TGF-β is a regulatory cytokine, having pleiotropic functions on immunity including promoting the expansion of Tregs 12 and regulation of CD4+ T cell polarization 13 . TGF-β promotes the differentiation of Th17 as well as of Treg cells, depending on the cytokine environment 14 . This has resulted in quite some opposing phenotypes in mice transgenic for the TGF-β signaling pathway. On the one hand, mice with T cell-speci c deletion of TGF-βRII show uncontrolled T cell activation and massive and fatal multi-organ in ammation 15 . On the other hand, mice with a deletion of the Tgfb1 gene selectively in CD4+ and CD8+ T cells have a defect in the generation of Th17 cells that protected them from experimental encephalitis (EAE) 16 . In line with that, mice subcutaneously injected with 100 µg anti-TGF-β1,2,3 antibody 1D11 failed to differentiate naïve CD4+ T cells into Th17 cells and were also protected from EAE 17 .
Due to its dual role in regulating the immune system, the potential of targeting the TGF-β pathway as therapy in RA is unclear. In this study, we aimed to investigate the effect of inhibition of TGF-β signaling with the TGF-βRI inhibitor SB-505124 on murine Th17 differentiation in vitro, on cytokine production by human RA synovial explants ex vivo, and to study the effect of local SB-505124 treatment in vivo during Th17-driven experimental arthritis.

Patient donors
Synovial tissues from eight RA patients were obtained during joint replacement surgery from the orthopedics department of the Sint Maartenskliniek, Nijmegen, The Netherlands. All patients adhered to the American College of Rheumatology (ACR) criteria and were end-stage RA. Patients gave written informed consent for the use of their material for research. The patient material was pseudonymized. Procedures were performed in accordance to the code of conduct for responsible use of human tissue in medical research. The presence of a synovial lining was determined on 5 µm cryosections stained with hematoxylin and eosin (H&E) to con rm the synovial origin of the tissue.

Mice
Female C57BL/6N were purchased from Janvier-Elevage (Le Genest Saint Isle, France). Animals were used between 10 and 12 weeks. A standard diet and water were provided ad libitum. All animal procedures and experimental protocols were approved by the animal ethics committee of the Radboud University Nijmegen (permit RU-DEC 2018-0037). All procedures were performed according to the Institute of Laboratory Animal Research Guide for Laboratory Animals and were performed in compliance with the institutional and ARRIVE guidelines.

RNA isolation and quantitative real-time PCR
Cells from in vitro studies were collected in TRIreagent for further RNA isolation as previously described 18 . Gene expression levels were determined by QPCR on the StepOnePlus sequence detection system (Applied Biosystems) using SYBR Green (Applied Biosystems) and 0.2 µM primers (Biolegio). GAPDH Page 5/20 was used as reference genes. Relative quanti cation of the PCR signals was performed by comparing the cycle threshold (Ct) value, of the gene of interest for each sample with the Ct values of the reference gene. Primer sequences were as follows (Table 1): Methylated bovine serum albumin/interleukin-1 (mBSA/IL-1) induced-arthritis Acute in ammatory arthritis was induced as previously described 19 by intra-articular (i.a.) injection with 200 µg of mBSA (Sigma A-1009), followed by daily injections on days 0-2 with 250 ng of human IL-1β (Biolegend). At sacri ce, knee joints were isolated for histological analysis and QPCR. Draining popliteal and inguinal lymph nodes were isolated for FACS analysis. The TGF-βRI inhibitor SB-505124 was applied locally starting two hours before the induction of arthritis, and mice were injected i.a. on four consecutive days with either saline, 20% DMSO as vehicle control, or 75 nmol SB-505124 in 20% DMSO with an injection volume of 6 µl per joint (Sigma). Anti-IL-17a antibodies (BioXCell) were used as positive control, administered intra-peritoneally 50 µg/mouse injected every other day. Mice were sacri ced by cervical dislocation four days after arthritis induction, and knee joints were subsequently isolated for histological analysis.

Streptococcal cell wall (SCW) arthritis induction
Streptococcus pyogenes T12 organisms were cultured overnight in Todd-Hewitt broth. Cell wall fragments were prepared as described previously 20 . Arthritis was subsequently induced by injecting 25 µg SCW fragments into the knee joints of C57BL/6N mice, resulting in local, acute in ammation. Local SB-505124 treatment was applied as described above.

Histology
Histology was processed and scored in line with the 'SMASH' recommendations for standardized microscopic arthritis scoring 21 . In short, isolated joints were xed for at least four days in 4% formaldehyde, decalci ed in 5% formic acid, and subsequently dehydrated and embedded in para n.
Standard frontal sections of 7 µm were stained with H&E or Safranin O (SO) to study joint pathology. The severity of arthritis was scored on an arbitrary scale of 0-3, where 0 = no pathology and 3 = maximal pathology, for three different parameters (joint in ammation, cartilage proteoglycan (PG) depletion, and bone erosion), on three semi-serial sections of the joint, spaced 140 µm apart, in a blindfolded manner.

Immunohistochemistry
Protein expression of phosphorylated Smad2/3 was evaluated on 7 µm para n-embedded sections of murine synovial tissue after co-injection with recombinant 10 ng TGF-β and 75 nmol SB-505124. For immunohistochemistry, endogenous peroxidase activity was blocked with 3% H2O2 (Merck Millipore) in methanol, and antigen retrieval was performed in 10 mM citrate buffer, pH 6.0 at 60°C. Subsequently, sections were stained with primary antibodies: rabbit anti-mouse pSmad2/3 (Cell Signaling #3108) (1:300 for 60 min at RT), mouse or isotype. Subsequently, the primary antibodies were stained with biotinylated anti-mouse IgG H+L (1:100 for 30 min at RT). Next, a biotin-streptavidin detection system was used according to the manufacturer's protocol (PK-6101; Vector Laboratories). Peroxidase was developed with diaminobenzidine (Sigma Aldrich) and counterstained with hematoxylin for 60 sec.

Synovial explant culture
Synovial tissue was obtained during joint replacement surgery and tissue was freshly processed into standardized biopsies using a 3 mm biopsy punch (Stiefel). The synovial explants were subsequently cultured for 24h at 37°C and 5% CO 2 in 200 µl X-vivo serum-free medium in the presence or absence of 5µM SB-505124. After 24 hours, culture medium was centrifuged 10 min, 241x g to remove remaining cells. Before Luminex analysis, culture medium was centrifuged 10 minutes, 10.000 rpm.

Statistical analysis
To determine the level of statistical signi cance between means of experimental groups, the Student's t test or a one-or two-way analysis of variance (ANOVA) was used unless stated otherwise. This depended on the number of experimental groups and normality testing using GraphPad Prism version 5.03 (GraphPad Software). P-values <0.05 were considered signi cant.

TGFB-RI inhibitor SB-505124 prevents TGF-β-induced expression of Th17 genes and proteins
To investigate the effect of inhibition of TGF-β signaling on murine Th17 differentiation in vitro, murine splenocytes were cultured with different concentrations of TGF-β in the presence or absence of 5 µM SB-505124. First, cells cultured with increasing concentrations of TGF-β showed a dose-dependent increase in IL-17 secretion ( Figure 1A) and a dose-dependent decrease in IFNγ production ( Figure 1B). TGF-β also increased the percentage of Th17 cells (de ned as CD4+IL-17+ cells) when dosed 0.05 to 1 ng/ml ( Figure  1C). Remarkably, higher concentrations of 10 ng/ml and 100 ng/ml of TGF-β decreased the percentage of Th17 cells ( Figure 1C) without affecting the IL-17 production ( Figure 1F), demonstrating that IL-17 protein production not always correlates with Th17 differentiation levels.
Addition of SB-505124 to these Th17 differentiation conditions supplemented with 1 ng/ml TGF-β resulted in signi cantly decreased Il17a mRNA expression, and in a trend for suppressed IL17a levels in the 10 ng/ml TGF-β group ( Figure 1D). Even more pronounced effects were observed for the mRNA expression of the lineage-speci c transcription factor Rorc. In line with the data for Il17a, TGF-β signi cantly increased the expression of Rorc compared to the plain Th17 cocktail condition, and this TGF-mediated effect could be completely counteracted by SB-505124 ( Figure 1E). Also at protein level, the inhibiting effect of SB-505124 on TGF-β-mediated Th17 differentiation was observed. However, whereas the TGF-β-induced increase in IL-17 production was completely and signi cantly abolished by SB-505124 in the 1 ng/ml TGF-β group, the SB-505124-mediated suppression of IL-17 production in the 10 ng/ml TGF-β group is only a trend of 30% reduction. The opposite of the TGF-β-and SB-505124induced effects on Rorc and IL-17 expression and production were observed on IFNγ levels: the presence of TGF-β completely blocked IFNγ production, which was reversed by addition of SB-505124 to the Th17 culture ( Figure 1F and G). These results indicate an important role for TGF-β during murine Th17 differentiation and suggest potency for SB-505124 to intervene in this process.
In addition to the observed effect of TGF-β blocking by SB-505124 on murine Th17 differentiation, we investigated the potential of SB-505124 to inhibit the spontaneous production of in ammatory cytokines by human RA synovial explants. Interestingly, SB-505124 signi cantly inhibited the production of IL-6 ( Figure 2A) by the explants from 5,802 pg/ml +/-848 (mean +/-SEM) in the control group to 2,565 pg/ml +/-633 in the SB-505124 group. All our nine donor samples were active IL-6 producing explants that showed SB-505124-inhibited IL-6 production of on average 56%, suggesting that active TGF-β signaling contributes to the in ammation in this synovial tissue. The levels of IL-10 were relatively low ( Figure 2B) and remained unaffected by SB-505124 treatment. For TNFα production, not all donors responded similar to the SB-505124 treatment. Five donors showed decreased levels, two donors increased levels and two donors showed no difference by SB-505124 treatment ( Figure 2C). Other cytokines including IFN-γ, IL-4, GM-CSF, and IL-17 were not detectable.

Intra-articular injection of SB-505124 does not induce joint in ammation and cartilage PG depletion in naïve mice
Before inhibiting TGF-β signaling during experimental arthritis, we rst aimed to study the potential joint pathology of SB-505124 treatment in naïve mice, as TGF-β is not only important in T cell activation but also in cartilage homeostasis. Repeated intra-articular SB-505124 injections for four consecutive days did not induce synovial in ammation or cartilage proteoglycan depletion in naïve knee joints as observed on histology at day seven. Also the vehicle control 20% DMSO did not result in joint pathology, as these control mice showed a thin synovial lining and healthy cartilage similar to saline-injected knee joints ( Figure 3). Importantly, we demonstrated that our treatment regimen of daily intra-articular injections of SB-505124 can indeed inhibit local TGF-β signaling. By co-injection of recombinant TGF-β1 and the TGF-βRI inhibitor SB-505124 in naïve murine knee joints, we demonstrated that SB-505124 inhibits TGF-β activity in vivo, as indicated by reduced pSmad2/3 staining by immunohistochemistry in synovium as marker for active TGF-β signaling (Supplementary gure 1&2).

Intra-articular injection of SB-505124 enhances Tregs and suppresses Th17 cell levels in the draining lymph nodes of arthritic mice
To study the effect of inhibition of TGF-β signaling on Th17-driven experimental arthritis, we treated mice during IL-1/mBSA arthritis by daily intra-articular injections with SB-505124. Interestingly, in the draining lymph nodes of SB-505124-treated mice, we observed a signi cant reduction in CD3+CD4+ T cells at day 4 ( Figure 4A). Moreover, we observed that inhibition of TGF-β signaling resulted in increased numbers of FOXp3+ CD3+CD4+ T cells (Tregs) ( Figure 4B) and less RORγt+ CD3+CD4+ T cells ( Figure 4C), although IL-17+ CD3+CD4+ cells were not altered by SB-505124 treatment ( Figure 4D). This indicates that local treatment with SB-505124 inhibits an important role for TGF-β signaling during T cell proliferation and Th17/Treg differentiation in this experimental arthritis model.

TGF-β signaling inhibition with SB-505124 does not decrease arthritis joint pathology
By subsequent histological analysis, the effect of local TGF-β signaling inhibition on arthritis severity was studied in more detail. Despite the skewed Th17/Treg balance in the draining lymph nodes of these arthritic mice, no differences were observed on in ammation and PG depletion on day four ( Figure 5A) and day seven ( Figure 5B) of arthritis between the SB-505124 group and its vehicle control. As expected, the positive control treatment using anti-IL-17 antibodies signi cantly reduced arthritis pathology ( Figure  5B). Additionally, no effects were observed in T cell-independent SCW arthritis after local SB-505124 treatment (supplementary gure 3).

Discussion
RA patients may bene t from the inhibition of Th17 cells, either by neutralizing their main effector cytokine IL-17 22 , or by targeting further upstream via its transcription factor RORγt 23 or Th17-inducing cytokines 24,25 . In this study, we explored the potential of targeting TGF-β signaling to inhibit Th17 cells, this in view of the importance of this growth factor in the differentiation of these cells.
Previous studies in experimental arthritis models showed contradictory results with pro-or antiin ammatory roles for TGF-β. For instance, recombinant TGF-β1 injection in naïve rat and mouse joints induced joint in ammation with synovial in ltration of T lymphocytes and neutrophils [26][27][28] . Treatment of arthritic rats with anti-TGF-β antibodies inhibited in ammation and bone resorption 29 . Additionally, intraperitoneal (IP) injected HTS466284 (TGF-β type 1 receptor kinase inhibitor) prevented experimental arthritis in mice 30 . IP treatment with the speci c TGF-β blocking peptide p17 reduced severity of CIA, however these effects were not statistically signi cant 31 .
In contrast, other studies showed that IP administration of TGF-β1 reduced the incidence and severity of CIA, especially when injected in late disease 32,33 . In rats with SCW-induced arthritis, IP 34 and intramuscular injection 35 of TGF-β at peak of in ammation suppressed the development of arthritis, whereas anti-TGF-β in CIA mice increased pro-in ammatory cytokines and arthritis severity 36,37 . Furthermore, an increase in TGF-β during the remission phase of CIA suggests an important role for TGFβ in regulating the disease 38 . These opposing ndings indicate a differential role for TGF-β, probably depending on the type of animal model, route of administration, disease stage, and site of in ammation.
In our in vitro studies, we observed that blocking TGF-β during Th17 differentiation e ciently and signi cantly reduced Rorc mRNA expression, but not IL-17 protein production. TGF-β is important in T cell differentiation, but also in dampening of the immune response of various T and B lymphocytes 39 . TGF-β inhibits differentiation and activation of speci c T helper subsets by suppressing their lineage-speci c transcription factors such as T-Bet and GATA-3, which are critical for Th1 and Th2 responses, respectively 40 The importance of TGF-β in regulating T cell responses in vivo has been strengthened by the observation that mice lacking TGF-βRII speci cally on T cells develop lethal multi-organ in ammation 15 . Also TGF-β1−/− mice develop multi-organ in ammation due to an impaired control of T cell activation and differentiation 41 . The phenotype of the TGF-β1−/− mice is completely rescued if mice are crossed to an MHCII knockout background, highlighting a crucial role for TGF-β in regulating pathological CD4+ T cell responses 42 . Furthermore, subsequent studies using mice with T cell-speci c deletions of Smad2 and Smad3 showed that intracellular signaling via Smad2/3 is essential for the TGF-β-mediated inhibition of effector T cells 43 . In our in vivo experiment, intra-articular SB-505124 treatment during experimental arthritis caused a clear reduction in Th17 levels,. However, this did not result in suppression of arthritis, suggesting that (1) the Th17 pathway is not that important in this model, (2) the remaining (Th17) cells are highly active due to loss of TGF-β-mediated dampening, or (3) that the treatment with SB-505124 had a too limited duration in time on TGF-β signaling to suppress the arthritis measured at end stage of our in vivo experiment. Unfortunately, our study design did not enable us to investigate the activation of the local Th17 cells and other immune cells by checking cytokine levels in the joints.
However, this would be in line with our in vitro studies where we observed that blocking TGF-β during Th17 differentiation e ciently and signi cantly reduced the Rorc mRNA expression but not the IL-17 production. From the potent effects of the anti-IL-17 treatment group as reference control, we can conclude that IL-17 is an important cytokine to block during this phase of this arthritis model.
TGF-β is often considered as an immunosuppressive cytokine, inhibiting for instance TNF-α, IFN-γ and IL-1β 10 . Interestingly, when inhibiting the TGF-β signaling pathway by SB-505124 on human RA synovium explants, IL-6 protein production by RA synovium was signi cantly reduced, suggesting that TGF-β is an important inducer of IL-6 in arthritic synovium. In line with this observation, we have shown that TGF-β stimulates the production of IL-6 by primary articular chondrocytes 44 and by the chondrocyte G6 cell line 44 . TGF-β also stimulates IL-6 production in PBMCs 45 . The other way around, TGF-β down-regulates IL-6 signaling in intestinal epithelial cells 46 , showing that TGF-β regulation of IL-6 signaling is cell type, tissue and context dependent. In our ex vivo studies, synovial explants are originating from end-stage RA patients undergoing total knee or hip replacement, and thereby less active in secreting cytokines than early RA tissue. However, even with this less in amed tissue, we observe an interesting suppression of the proin ammatory cytokine IL-6 in all donors after inhibition of TGF-β signaling by SB-505124.
In our approach we chose to use the small molecule inhibitor SB-505124 to inhibit TGF-β signaling, which has an advantage over antibodies in tissue and cell penetration 47,48 . Disadvantage is the short half-life 49 , therefore SB-505124 must be administered frequently. With daily intra-articular injections, we observed that this compound highly e ciently blocked TGF-β-mediated Smad2/3 phosphorylation (supplementary Figure 1 and 2).
In our proof-of-concept study to demonstrate the potential of controlling Th17 cells by targeting TGF-β signalling, we included the positive control anti-IL17 treatment, since the arthritis model we used is known to be dependent on CD4+ T cells and IL-17 19,50 . Depletion of CD4+ T lymphocytes in mBSA/IL-1-induced arthritis led to dose-dependent T cell proliferation in draining lymph nodes in response to mBSA. Anti-IL17 antibody administered during our mBSA/IL-1-induced arthritis markedly reduced disease, con rming that the model is indeed IL-17 dependent 19,50 . However, our blocking of TGF-β signalling apparently did not su ciently suppress the formation or activity of Th17 cells, as arthritis pathology was not affected by SB-505124 treatment, showing that our strategy to target the Th17 pathway upstream is not as effective as blocking the main effector cytokine IL-17 itself.

Conclusions
We revealed suppressive effects of SB-505124 on Th17 differentiation in vitro and on the Th17/Treg balance in arthritic mice. However, SB-505124 did not suppress joint in ammation and destruction during experimental arthritis. This indicates that, despite the importance of TGF-β in Th17 differentiation, inhibiting TGF-β signalling is therefore not an adequate way to target Th17-driven in ammation.  51 . Procedures were performed in accordance to the code of conduct for responsible use of human tissue in medical research.

Consent for publication
All authors read the manuscript and gave their consent for publication. Figure 1 TGFB-RI inhibitor SB-505124 prevents TGF-β-induced expression of Th17 genes and proteins.

Figure 2
Inhibiting TGF-β signaling using SB-505124 signi cantly reduced IL-6 production by human RA synovial explants. Synovium biopsy punches were collected from synovium collected during joint replacement surgery. Explants were cultured for 24hr in the presence or absence of 5µM SB-505124 and cytokine levels in the supernatants were measured by Luminex. Graphs show results for IL-6 (A), IL-10 (B), and TNFα (C). N=9 donors, mean of 2-6 biopsies per donor per condition. **P < 0.01, as determined by paired Student's t-test.

Figure 3
Intra-articular injection of SB-505124 does not induce joint in ammation and cartilage PG depletion. Mice were daily injected with 75 nmol SB-505124 or 20% DMSO as vehicle control from day 0-4, and sacri ced after 7 days. Knee joints were subsequently isolated for histological analysis of in ammation (HE stain, original magni cation 100x) (B) and cartilage PG depletion (SafO stain, original magni cation 100x) (B) Representative pictures of n=4 joints per group.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.