Skip to main content

Stromal cells as primary drivers of immunopathology: towards targeted disease modification in spondyloarthritis

Periodic Reporting for period 3 - Inflammostrome (Stromal cells as primary drivers of immunopathology: towards targeted disease modification in spondyloarthritis)

Reporting period: 2018-07-01 to 2018-12-31

Over the last 2 decades, we have made a huge progress in the treatment of inflammatory diseases such as chronic arthritis by developing treatments that target immune cells and their products. A striking example of this is the major improvement in signs and symptoms of disease as well as in quality of life of patients with spondyloarthritis, a prototypical form of rheumatic inflammation affecting mainly the spine, when treated ith so-called biologic treatments (monoclonal antibodies blocking important inflammatory molecules such as TNF and IL-17). Despite these advances, however, a significant fraction of patients do not respond to these treatments, do only partially respond, or even when to do respond well still continue to accumulate organ damage over time. These clinical observations indicate that we have not yet found the primary cause of the disease.

In this project, we hypothesize that not only immune cells but also resident tissue cells contribute to the disease process. We propose that in same disorders tissue cells may even be the primary driver of chronic inflammation. Therefore, understanding the mechanisms by which tissue cells can drive these conditions is essential to find new targets for causal treatment.

Using spondyloarthritis as a model disease, this project aims to unravel the role of stromal cells in chronic tissue inflammatiom by defining a) the molecular profile of the stromal alterations, b) the mechanisms of pro-inflammatory cytokine production by stromal cells, c) the activation of downstream effector patwhays drivin inflammation as well as tissue remodeling, and d) the functional contribution of the tissue cells to experimental and human spondyloarthritis.

With out sigificant advances in symptomic treatment of many inflammatory conditions, this project is part of a large effort to move towards real disease-modifying treatments. The relevance relates to our ambition to modulate the natural course of those disabling conditions and thereby to maintain health in our patients.
Objective 1: Molecular characterization of the primary stromal alternations in SpA

The basic hypothesis of this work-package is that specific stromal alterations rather than immune dysregulation underlies the immunopathology of SpA, based on the strong myofibroblast signature seen in SpA but not RA synovitis (Yeremenko, Arthritis Rheum 2013). This signature, once established, appeared to be independent of inflammation as it was not modulated by anti-inflammatory treatment in vivo or exposure to inflammatory stimuli in vitro (Yeremenko, in preparation).
Screening for epigenetic factors that may impact the Thy-1 positive synovial myofibroblasts indicated HDACs as potential modulators. Extensive in vitro studies showed that, whereas HDAC5 may suppress the inflammatory stromal phenotype, HDAC3 is an important driver of pro-inflammatory mediator production by synovial fibroblasts (Angiolilli, Ann Rheum Dis 2016; Angiolilli, Ann Rheum Dis 2017; Angiolilli, Epigenetics 2017). Interestingly, the impact of these HDACs was much more profound on synovial fibroblasts than on prototypical inflammatory cells such as macrophages.
Investigating which factor may trigger such epigenetic changes, we focused our attention on ER stress as this type of stress, but not mechanical stress or hypoxia, was able to recapitulate the myogene signature in vitro (Yeremenko, in preparation). ER stress alone was not sufficient to activate the synovial fibroblasts, but there was a clear synergy with TLR stimulation leading to significant upregulation upregulation of the inflammation potential of these cells through mRNA stabilization (Kabala, Arthritis Res Ther 2017). These findings are consistent with the fact that TLR signalling is crucially modulated by ‘co-activation’, with as examples not only ER stress but also FcR signalling (Hansen, J Immunol 2017; Hansen, Nat Commun 2018). We also found that ER stress abrogates the anti-inflammatory effects of IL-10 on myeloid cells by inhibiting STAT3 phosphorylation (Hansen, Front Immunol in press).
As to the origin of ER stress in SpA, it has been suggested to non-conventional forms of HLA-B27 could induce this type of cellular stress. Investigating these nc HLA-B27 forms, we found there expression in synovial and gut tissue of human SpA patients; more importantly however, they were not only expressed by infiltrating cells but also by tissue-resident cells, supporting our hypothesis of stromal contribution to the disease process (Rysnik, J Autoimmun 2016). Moreover, using our HLA-B27 tg rat model, we could demonstrate the these nc HLA-B27 forms were already present prior to disease onset (Rysnik, Data Brief 2016). Also in this model where a fraction of the male rats spontaneously develop disease after 3 to 6 months, we demonstrated that innate immune stimulation using heat-killed M Tub lead to accelerated and synchronized disease in >90% of both males and female animals, confirming the role of synergy between ER stress and innate activation in the disease pathophysiology (van Tok, Front Immunol 2017).

Objective 2: Analysis of IL-23 and tmTNF production by stressed stromal cells
In our experiments with ER stress on stromal cells, in particular synovial fibroblasts, we noticed that one of the strongest increases in mRNA was for the IL-23 sub-unit p19. As described above, this was mainly observed upon co-stimulation with inflammatory cytokines or TLR-ligands together with ER stress. We could also demonstrate that p19 mRNA was paralleled by increased expression levels of p19 protein. However, we could not demonstrate p40 production and/or the presence of IL-23 protein, the p19/p40 heterodimer. This contrast sharply with what we see in macrophages, eg our recent demonstration of an increase in IL-23+ inflammatory macrophages in human SpA (Ciccia, Arthritis Rheum 2018).
We therefore assessed two other molecular pathways that could be related to p19 in stromal cells: a) detailed analysis of stressed synovial fibroblasts showed the upregulation of the the p40 homologue EBI3. P19 and EBI3 have been recently described to form the novel cytokine, IL-39. We are currently attempting to prove that IL-39 is produced by SpA synovial fibroblasts (Blijdorp, in preparation). b) we observed an unexpected p19 staining in the nuclei of the fibroblasts, which was confirmed by protein analysis of nuclear extracts. We therefore hypothesized that p19 could act as a transcription factor/transcriptional co-activation and could ‘intrinsically’ modulate synovial fibroblast behaviour. Indeed, silencing experiments demonstrated very poor viability of p19 deficient fibroblast. We are currently assessing how p19 overexpression in stromal cells affects survival, proliferation, and behaviour of these cells (Yeremenko, in preparation).
Another key finding is that SpA synovitis, in contrast to RA synovitis, is characterized by over-expression of tmTNF but not sTNF. We could demonstrate that this is related to decreased TACE activity in human SpA. Interestingly, the altered TACE activity and sTNF/tmTNF balance is not only observed on immune cells such as macrophage but also on synovial fibroblasts. Using mice that selectively over-express tmTNF, we could demonstrate that they develop a spontaneous spondyloarthritis with 100% incidence: bot the axial and peripheral disease was reminiscent of human SpA with a combination of inflammation (synovitis, osteitis, and enthesitis), modest bone destruction, and pronounced new bone formation leading to complete ankylosis over time. These animals did not develop extra-articular manifestations, and we showed formally that the phenotype and histology was completely different from mice over-expressing sTNF such as delta-ARE mice. Whereas the inflammatory part of the disease was dependent on TNF-R1, the bone remodelling was dependent on TNF-R2. Additionally, we could demonstrate that the SpA-phenotype observed in tmTNF tg mice is dependent on tmTNF expression by stromal cells but not by immune cells, underscoring the role of stromal cells in the pathophysiology of the diseases (van Duivenvoorde, J Exp Med, second revision).


Objective 3: characterization of downstream inflammatory pathways
Besides our studies focusing on the stromal cells that can initiate and/or drive the inflammatory responses, we also continued to investigate which downstream inflammatory mediators contribute the SpA inflammation. Continuing our studies on IL-17A, we are still investigating the exact cellular source of this cytokine in SpA. Based on our original observations that mast cells are the main IL-17A expressing population in SpA inflammation, we demonstrated that - although mast cells contained IL-17A protein - they do not produce this cytokine themselves but are capable of taking-up, storing, and releasing exogenous IL-17A (Noordenbos et al, J Leukocyt Biol 2016). This mechanism is not unique to SpA synovitis and IL-17A loaded mast cells are numerous at body surfaces such as skin and gut; these IL-17A loaded ‘sentinel cells’ are not activated themselves by IL-17A and are not modulated by anti-IL-17A treatment (Chen, submitted for publication).
Investigating other cellular sources of IL-17A, we used a combination of human biopsy work, flow cytometric sorting, and single cell transcriptional profiling to show that ILCs are not a major source of IL-17 in SpA synovitis (Yeremenko, submitted for publication). In contrast, our data seem to indicate that ILCs, in particular type II ILCs, have anti-inflammatory properties in arthritis (Zeiss, Cell Rep 2018).
Looking at other cytokines that may contribute to inflammation beyond TNF and IL-17A, we have extended the evidence of a key role of the IL-17 axis in SpA by demonstrating that, in contrast to mice studies, not only IL-17A but also its twin cytokine IL-17F drives inflammation in vitro and ex vivo (Glatt, Ann Rheum Dis 2018). We demonstrated clear co-expression in vivo and similar function ex vivo for IL-17A and IL-17F in human SpA: however, the IL-17A/IL-17F ratio seems to differ from one tissue to another, with potentially important therapeutic implications (Chen, submitted for publication). Ongoing work in a public-private collaboration are focused on the molecular mechanisms of regulation of IL-17A and IL-17F, and on the IL-23 dependency (or independency) of these cytokines.
One other cytokine of interest in this context is IL-9, as we demonstrated a strong increase in Th9 cells in the psoriatic subset of SpA (Ciccia, Arthritis Rheum 2016) and also observed intriguing modulation of this cytokine by therapeutic interventions in our HLA-B27 tg rat model (van Tok, in preparation).

Objective 4: characterization of stromal-driven remodelling pathways
In line with the studies described above, we have assessed how key inflammatory pathways may affect new bone formation by performing in vitro fibroblast-to-osteoblast differentiation assays. We confirmed that both TNF (through TNF-R2) and IL-17A can promote osteoblastic differentiation of mesenchymal precursor cells. IL-17F behaves similarly to IL-17A in these assays, and co-stimulation with TNF and IL-17A/F has even a greater impact. In contrast, IL-22 did not promote osteoblastic differentiation. As to the mechanisms, we observed a broad activation of canonical and non-canonical Wnt signalling pathways but failed to find specific mediators up to now. Interestingly, IL-17A blockade in vivo in our HLA-B27 tg rat model lead to suppression of inflammation, of Wnt activity, and of new bone formation (van Tok, submitted for publication).
A major finding of this part of the project is that, during our investigation of the cellular origin of IL-17A and F in SpA synovitis, we found that the major IL-17 family member expressed in this tissue is the poorly studied IL-17D isoform. Focusing on this novel finding, we demonstrated that, in contrast to IL-17A and F, IL-17D is not produced by leukocytes but by stromal cells, in particular osteocalcin-positive mesenchymal precursor cells. Interestingly, IL-17D production was suppressed by inflammatory cytokines in vitro as well as in vivo (using the HLA-B27 tg rat model). We could show that IL-17D is secreted and binds to IL-17RB but not the classical IL-17RA and RC. As to function, IL-17D expression seems to correlate with differentiation of mesenchymal precursors cell to osteoblasts but not to chondrocytes or mast cells (Chen, in preparation). In vitro silencing and over-expression experiments are currently performed to assess if IL-17D is a driver of new bone formation, and in vitro experiments in IL-17D knock-out mice are planned.


Objective 5: in vivo evaluation of the contribution of stromal pathways to experimental and human SpA
We performed a whole series of observational human studies and interventional rat and humans studies to understand how the pathways described above contribute to SpA and to chronic tissue inflammation in general.
For our observational studies, we focused mainly investigated the early and preclinical forms of SpA. We tested a few candidate biomarkers that have been proposed to have potential value for early diagnosis and/or stratification of SpA; unfortunately, most of these biomarkers such as anti-CD74 do not seem to hold up to promises (de Winter, Arthritis Res Ther 2017). Even MIR imaging of inflammatory lesions of the SI joints seems less specific than previously anticipated (de Winter, Arthritis Rheum 2018). We were successful, however, in the identification of a novel genetic risk factor, HLA-C*07, which confers risk independently of HLA-B27 and may identify a subset of patients with more structural damage (de Winter, submitted). We have also extended our prospective cohort study of first-degree relative to more than 200 individuals, and are currently assessing the biological and clinical features (de Jong, in preparation).
As to TNF, we demonstrated by an innovative PET-CT technique combined with spinal biopsy analysis that TNF blockade has minor – if any- impact on new bone formation in axial SpA despite good anti-inflammatory effect (Bruijnen, Rheumatology 2018). Transcriptional profiling of synovial biopsies before and after TNF blockade confirmed that this treatment, despite symptomatic efficacy, is not able to revert the molecular disease signature (Yeremenko, in preparation). However, we could also demonstrate that early initiation of TNF blockade in psoriatic SpA leads to high levels of clinical remission, upt to 80% (van Mens, submitted).
As to IL-17, we confirmed that clinical impact of IL-17A blockade on SpA (Dheodar, Arthritis Rheum 2016; Wei, Int J Rheum Dis 2017). Moreover, we demonstrate a clear impact in the synovial immunopathology without compromising the systemic immune responses (van Mens, Arthritis Rheum 2018). Finally, we provided prelimimary evidence that this treatment could retard or even halt new bone formation (Braun, Ann Rheum Dis 2017). We are currently performing detailed transcriptional analyses to confirm these clinical findings (Yeremenko, in preparation).
A major break-through in this field, based on our identification of IL-17F as an important contributor beyond IL-17A in human but not rodent tissue inflammation, is our study with the dual IL-17A and IL-17F inhibitor Bimekizumab. We demonstrated best-in-disease impact on joints and skin in a proof-of-concept study in psoriatic arthritis (Glatt, Ann Rheum Dis 2018) and confirmed now these data in larger phase II studies (in preparation).
Considering our in vitro findings that TNF and IL-17A/IL-17F synergize to drive both inflammatory responses and osteoblastic differentiation in vitro, we performed two combo-treatment in vivo studies. In our HLA-B27 tg rat model, we compared anti-TNF, anti-IL17A, and anti-TNF + anti-IL17A: we did not observe additional benefit of the combo treatment on clinical endpoints but are currently assessing the histology and new bone formation (van Duivenvoorde, in preparation). In humans, we performed a study in anti-TNF-incomplete responders patients in RA where we observed some benefit in terms of remission of anti-TNF + Bimekizumab versus anti-TNF alone (Glatt, in preparation). However, this effect was not consistent across all clinical endpoints, questioning the relevance of this combo-treatment in RA.
We next assessed the role of p19/IL-23 in SpA. In our HLA-B27 tg rat model, we demonstrated that targeting IL-23 (not pure p19) was effective in preventing disease but had no impact at all on established disease (van Tok, Front Immunol in press). In human axial SpA, p19 blockade with risankizumab had no impact on clinical endpoints and srum and imaging biomarkers (Baeten, Ann Rheum Dis 2018). These data confirm our preclinical data indicating that a) p19 may have an intrinsic intracellular function independently of secreted IL-23, and b) IL-17A and IL-17F production may be independent of IL-23, certainly in established inflammation. An additional piece of intriguing information supporting the concept that IL-23 and IL-17 biology are not linear, is our finding that RORgamma inhibition in the HLA-B27 tg rats fails to block IL-17 and exacerbates disease (van Tok, in preparation).
Finally, we also explored the potential to modulate ER stress in vivo. As autophagy has been described to temper the HLA-B27-induced ER stress, we tested the impact of rapamycine treatment in vitro and in vivo, demonstrating that this completely prevents and reverts the disease phenotype in HLA-B27 tg rats (Chen, in preparation). To what extend this is mediated solely by impacting ER stress and to what extend this is related to stromal versus immune cells, remains to be further investigated.