Inhibition of JAK-STAT signaling suppresses pathogenic immune responses in medium and large vessel vasculitis

Abstract

Background

Giant cell arteritis (GCA), a chronic autoimmune disease of the aorta and its large branches, is complicated by aneurysm formation, dissection, and arterial occlusions. Arterial wall dendritic cells (DC) attract CD4⁺ T-cells and macrophages (Mo), to form prototypic granulomatous infiltrates. Vasculitic lesions contain a diverse array of effector T-cells that persist despite corticosteroid therapy and sustain chronic, smoldering vasculitis. Transmural inflammation induces microvascular neoangiogenesis and results in lumen-occlusive intimal hyperplasia. We have examined whether persistent vessel wall inflammation is maintained by lesional T-cells, including the newly identified tissue-resident memory T cells (T_RM) and whether such T-cells are sensitive to the cytokine signaling inhibitor tofacitinib, a JAK inhibitor (Jakinib) targeting the Janus kinase (JAK) 3 and JAK1.

Methods

Vascular inflammation was induced in human arteries engrafted into immunodeficient mice that were reconstituted with T-cells and monocytes from GCA patients. Mice carrying inflamed human arteries were treated with tofacitinib or vehicle. Vasculitic arteries were examined for gene expression (RT-PCR), protein expression (immunohistochemistry) and infiltrating cell populations (flow cytometry).

Results

Tofacitinib effectively suppressed innate and adaptive immunity in the vessel wall. Lesional T-cells responded to tofacitinib with reduced proliferation rates (<10%) and minimal production of the effector molecules IFN-γ, IL-17 and IL-21. Tofacitinib disrupted adventitial microvascular angiogenesis, reduced outgrowth of hyperplastic intima and minimized CD4⁺CD103⁺ tissue-resident memory T-cells.

Conclusions

Cytokine signaling dependent on JAK3 and JAK1 is critically important in chronic inflammation of medium and large arteries. The Jakinib tofacitinib effectively suppresses tissue-resident memory T-cells and inhibits core vasculitogenic effector pathways.

Introduction

Vasculitides of large elastic arteries are infrequent, but potentially fatal diseases, damaging vital blood vessels, such as the aorta, the subclavian-axillary bed, the carotid branches, and mesenteric arteries. Giant cell arteritis (GCA) accounts for most cases of autoimmune large vessel vasculitis; typically causing vision loss, aortic arch syndrome, aortic dissection, and aortic aneurysms. Extravascular GCA, consistent of an intense hepatic acute phase response (APR) gives rise to highly elevated laboratory markers of inflammation. Whether the hepatic APR precedes or follows vascular inflammation is unresolved. GCA is a chronic condition, which persists despite long-term therapy with high-dose corticosteroids^1,2 and disease risk genes have been localized to multiple biologic pathways³.

CD4⁺ T-cells and macrophages dominate the transmural lesions of this granulomatous vasculitis. Arterial wall dendritic cells (DC) function as gatekeepers and by providing access to invading T-cells and macrophages fail to protect the artery's immune privilege^4,5. In GCA arteries, wall-resident DC express low concentrations of the immunoinhibitory ligand PD-L1, disarming the protective PD-1 immune checkpoint^6,7. Tissue-infiltrating CD4⁺ T-cells are PD-1⁺, yet are highly activated, non-exhausted and cover multiple effector functions. Most prominent are tissue Th1 and Th17 cells, but IL-21- and IL-9-producing T-cells are also present^8,9. Heterogeneous T-cell effector populations in the lesions are indicative of an unopposed T-cell response.

GCA's chronicity suggests a role for tissue-resident memory T-cells (T_RM), a recently discovered T-cell lineage residing in tissues, where they provide fast and powerful helper functions¹⁰. Different from central memory and effector memory T-cells, T_RM cells receive localizing signals in the tissue niche and do not recirculate to secondary lymphoid organs. Two phenotypic markers, CD69 and CD103 (a receptor recognizing E-cadherin), have been identified¹¹. T_RM cells in E-cadherin^low tissues lacking epithelium possibly express alternative markers, such as type I collagen receptors. Originally considered crucial for rapid anti-pathogen responses, T_RM cells may also drive autoimmune tissue inflammation¹². Functional heterogeneity, being able to release IFN-γ, IL-17, IL-9, and TNF-α, enables pro-inflammatory effector functions of T_RM cells¹³. Tissue-derived IL-7, IL-15 and TGF-β are believed to guide T_RM recruitment, differentiation and maintenance¹³. Whether T_RM are involved in building and sustaining GCA's granulomatous lesions and the arterial wall remodeling process is unknown.

T-cells depend on signals through their T-cell receptor (TCR), but require input from the cytokine milieu to direct their clonal expansion, persistence, and functional differentiation. Environment-cell communications rely on cytokine signals that trigger the Janus kinase (JAK) and signal transducer and activator of transcription (STAT) pathway¹⁴. The JAK/STAT signaling pathway has been implicated in cancer cell growth and autoimmunity¹⁵. Th1 lineage commitment is strictly linked to STAT1- and STAT4-mediated gene induction¹⁶. STAT3 is considered the master regulator for Th17 cell differentiation. Gene polymorphisms encoding type I cytokine receptors and their signaling elements (IL-23R, IL-12B, JAK2 and STAT3) are linked to inflammatory bowel diseases and psoriasis¹⁷. STAT4 polymorphisms are associated with rheumatoid arthritis (RA), Sjogren's syndrome and systemic lupus erythematosus^18,19. JAK/STAT's critical role in immune-mediated disease has been therapeutically exploited with the development of Jakininbs, small molecule inhibitors that block the action of type I/II cytokines. The JAK3/1 inhibitor Tofacitinib²⁰ has been approved for RA treatment²¹. JAK3-activating mutations in T-cell acute lymphoblastic leukemia and JAK3-inactivating mutations in severe immunodeficiency emphasize the critical role of JAK3 in T-cell biology^22,23. Most prominently, JAK3 inactivation results in loss-of-function of the common gamma chain (γc); causing X-linked severe combined immunodeficiency^23,24. Cytokines central in regulating T-cell activation and survival (IL-2, IL-4, IL-7, IL-9, IL-15, IL-21) all employ JAK3 to mediate their effect²⁵.

Chronic GCA requires T-cells persisting in the arterial wall; to orchestrate macrophage activation, drive inflammation-associated neo-angiogenesis and promote intimal hyperplasia^26,27. Current therapies focus on pro-inflammatory innate cytokines^28,29, but fail to eliminate wall-infiltrating T-cells². We have explored whether targeting JAK3 with the JAK inhibitor (Jakinib) tofacitinib can remove disease-relevant T-cells from inflamed arteries. To overcome limitations in examining in-situ T-cells, we have employed a human-artery SCID chimera model, in which vascular inflammation is induced in engrafted human arteries by reconstituting such chimeras with peripheral blood mononuclear cells (PBMCs) from GCA patients^6,30. In this model system, alloantigens serve as model antigens to probe immunoregulatory defects in GCA patients that are relevant for immune responses in the microenvironment of the vessel wall. Tofacitinib therapy was highly effective in suppressing vasculitogenic immunity by inhibiting the clonal expansion of wall-residing T-cells. Processes downstream of vasculitogenic T-cell activity, e.g. microangiogenic growth of capillary networks and intimal hyperplasia were effectively disrupted. Most importantly, we demonstrate the existence of tissue-anchored CD4⁺ CD103⁺ T-cells in chronically inflamed arteries. Keeping such CD103⁺ T_RM cells alive and functional required JAK/STAT signaling and was successfully inhibited in tofacitinib-treated chimeras. Selective Jakinibs may provide new avenues to manage acute and chronic large vessel vasculitis.

Methods

All data supporting this study's findings are available from the corresponding author upon request.

Patients and tissue samples

Temporal arteries were collected from diagnostic biopsies. GCA was diagnosed based on typical histologic findings². Noninflamed human axillary arteries free of wall-infiltrating T-cells and macrophages derived from tissue donors (12-18 hours postmortem). Patients with a tissue diagnosis of GCA and active disease donated blood for the study. The Stanford Blood Bank Research Program supplied age and sex-matched healthy controls. Pre-existing cancer, autoimmune disease, or chronic viral infection were exclusion criteria. Clinical characteristics of the study patients are summarized in Table 1.

Table 1. Clinical characteristics of patients with GCA.

Parameters	Patients (n = 54)
Age (year, mean ± SD)	70.87 ± 8.7
Female	38/54
Ethnicity
Caucasian	47
Hispanic	4
Asian	2
African-American	1
Disease duration (month, mean ± SD)	6.0 ± 9.55
ESR (mean ± SD, mm/h)	52.09±26.02
CRP (mean ± SD, mg/dL)	8.93±10.64
Aortic/large vessel involvement	23/54
Polymyalgia rheumatica	30/54
Untreated	19/54
Prednisone (mg/day, mean ± SD)	12.07±14.3

PBMCs were isolated by density-gradient centrifugation (Lymphoprep, Oslo, Norway). Naïve CD4⁺ T-cells were purified by negative selection (EasySep™ naïve CD4⁺ T-Cell enrichment kits, STEMCELL Technologies, Vancouver, BC, Canada).

Human Artery-Severe Combined Immunodeficiency Mouse Chimeras

NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ (NSG) mice were purchased from Jackson Laboratory (Sacramento, CA). Human artery-mouse chimeras were generated as published^6,31. Normal temporal or axillary arteries were cut into rings (3 mm) and engrafted into a midline subcutaneous pocket of the mice. Six days later, mice received 10 μg LPS. The next day, PBMCs (1×10⁷) from a patient with active GCA or an age-matched healthy individual were adoptively transferred into each mouse.

For model validation we compared induction of vessel wall inflammation after transfer of PBMC from age-matched controls or GCA patients into artery-engrafted NSG mice (Suppl.Fig.1). Transferred human cells establish organized lymphoid structures in the murine spleen³². Spleen/body weight ratios were similar in mice receiving control or patient-derived PBMC (Suppl.Fig.1A); indicating similar immuno-reconstitution of paired mice carrying the same arterial tissue, but receiving healthy or GCA PBMC. Numbers of transferred human cells were chosen such that allogeneic recognition of arterial cells was avoided. To detect even subtle inflammation in the artery grafts, marker gene transcripts were monitored in the tissue transcriptome of explanted grafts (n=10 each from control PBMC or GCA PBMC-reconstituted mice). Based on tissue transcripts for the monocyte/macrophage marker CD163 and the DC marker CD83, only patient-derived PBMC were successful in inducing inflammation (Suppl.Fig.1B, C). As expected, arterial grafts were free of transcripts for TCR, T-bet and IFN-γ in the absence of human PBMC. No signal for tissue-infiltrating T-cells was recorded in mice reconstituted with control PBMC, but GCA-PBMC produced robust T-cell infiltrates (Suppl.Fig.1D). VEGF is an excellent marker of endothelial cell activation and neoangiogenesis in GCA³¹. Tissue VEGF transcripts were indistinguishable in grafts from PBMC-free mice and control-PBMC-reconstituted mice, but were upregulated in the presence of patient-derived PBMC (Suppl.Fig.1E).

To examine the impact of the artery graft on immune reconstitution, NSG mice were assigned to parallel treatment arms: Arm 1 received an artery as in Suppl.Fig.1; Arm 2 was sham operated. Mice were immune-reconstituted with 1x10⁷ GCA PBMC. Immune reconstitution was assessed after 7 days in the chimeras' spleen and circulating blood. Human T-cells were quantified by flow cytometry of CD45⁺CD3⁺ cells (Suppl.Fig.2).

Tofacitinib treatment

Mice carrying arterial tissue from the same donor and immuno-reconstituted with PBMCs from the same patient were randomly assigned to the vehicle or tofacitinib-treatment group. Tofacitinib dosing was assessed in pilot experiments to avoid T-cell and monocyte cytopenia³³. In experiments shown here, mice received 3 mg/kg tofacitinib or equal volume of vehicle orally daily over one to two weeks as indicated³⁴. Grafts were harvested at the end of the experiment, snap-frozen for RNA isolation or OCT-embedded for H&E and immunostaining. To prepare single-cell populations, explanted spleens or arteries were minced into pieces and incubated with collagenase I (2 mg/ml) and Dnase (40 U/ml) at 37°C. Cells were filtered through a cell strainer to prepare single-cell suspensions.

Statistics

Statistical analysis was performed using GraphPad Prism 5.0. All data are expressed as mean±SEM and p<0.05 was considered statistically significant. Populations were compared using t-test or Wilcoxon test as appropriate. To adjust for multiple testing, in addition to individual p-values, we used Hochberg's step-down method to control for a family-wise-error rate at the 0.05 level. To assess interventions, we applied paired t-test or paired Wilcoxon test. The Wilcoxon test was used for robustness when the sample size per group was >5. Where appropriate, one-way ANOVA was used and pair-wise comparison using Tukey's method to adjust for multiple testing was applied.

Study approval

All procedures and biospecimen collections were approved by the Institutional Review Board and informed consent was obtained as appropriate. The animal protocol was approved by the Animal Care and Use Committee.

Additional methods available in the supplementary file. PCR primers are listed in Suppl. Table 1.

Results

Expression of STAT target genes in GCA-affected arteries

Temporal arteries with the typical granulomatous lesions of GCA are a cytokine-rich tissue environment³⁵, with a multitude of innate and adaptive effector molecules. Th1-cell derived IFN-γ is a critical disease driver^36,37, but less is known about tissue concentrations of other adaptive and innate cytokines, including IL-21, IL-1, IL-6 and TNF-α. In search for clues which cytokines may be key activators in GCA arteries, we made use of the fact that many cytokines signal through JAK/STAT pathways to mediate their pro-inflammatory functions. To support the concept that selected JAK-STAT signaling pathways are particularly relevant for GCA inflammation, we evaluated the expression of selected STAT target genes (Fig.1A). STAT1- and STAT2-dependent target genes (T-bet, CXCL9, ISG15, OAS1) were strongly upregulated in the vasculitic arteries. Target genes induced by STAT3, STAT5a and STAT6 signaling were consistently low; indistinguishable from normal, noninflamed temporal arteries (Fig.1A). Typically, each of the STAT family members is a target gene induced when the respective transcription factor is activated¹⁷. Thus, we quantified transcripts for STAT1, STAT2, STAT3, STAT4, STAT5a and STAT6 in 8 diagnostic biopsies from GCA-affected and 8 noninflamed specimens (Fig.1B-1G). Transcripts for all STAT family members were at low abundance in normal arteries and STAT3, STAT5a and STAT6 remained sparsely expressed in GCA-affected arteries. In contrast, STAT1, STAT2 and STAT4 transcripts were abundantly present in vasculitic arteries (Fig.1B-1D). After adjustment for multiple testing using the Benjamini–Hochberg method, the STAT1 increase remained statistically significant.STAT1 and STAT2 are the key transcription factors triggered by interferons; with type 1 interferons activating STAT1/2 heterodimers and IFN-γ activating STAT1 homodimers³⁸. We examined the tissue transcriptome of noninflamed and inflamed arteries for IFN-α, IFN-β and IFN-γ transcripts (Fig.1H-1J) and found significantly elevated concentrations of IFN-α and IFN-γ specific sequences; suggesting that both type I and II interferons are actively participating in the vasculitic lesions. The STAT4 target gene furin and STAT4 transcripts were borderline elevated in vasculitic versus normal arteries. Target genes for STAT3 and STAT6 were distinctly low; questioning a role for IL-6 and IL-4 in the tissue microenvironment. The STAT5 target gene FoxP3 was barely detectable in the tissue transcriptome.

Figure 1. Expression profiles of STAT family transcription factors and their respective target genes in arteries affected by GCA.

Temporal artery biopsies were categorized according to histology into those free of vasculitis and those with classical findings of giant cell arteritis (GCA). Tissue extracts were processed for gene expression profiling applying RT-PCR and expression levels for individual genes were adjusted to β-actin. (A) Expression levels of selected target genes for each of the STAT family members shown as a heat map. Each lane represents one arterial sample. (B-G) Expression levels for transcription factors of the STAT family. (H-J) Tissue transcript concentrations for type I (IFN-α, IFN-β) and type II interferon (IFN-γ). Data from 8 vasculitis-free and 8 GCA-affected arteries were compared by Wilcoxon test and are presented as mean ± SEM. *P<0.05, **P<0.01, ***<0.001.

Albeit providing only indirect evidence, these gene expression data suggested predominance of STAT1 and STAT1/2 activation events in GCA lesions, focusing attention onto type I and type II interferons.

IFN-γ production in GCA T-cells is susceptible to the JAK/STAT inhibitor tofacitinib

Considering the predominance of STAT1 in the vasculitic infiltrates, we explored whether CD4⁺ T-cells from GCA patients are biased towards IFN-γ production and whether Th1 lineage commitment involves JAK3 activation. We stimulated CD4⁺ T-cells from patients and age-matched controls under non-skewing conditions and tested whether Th1 commitment was sensitive to the JAK3 inhibitor tofacitinib (Fig.2). As antigen-presenting cells were excluded from the cultures, stimulation conditions relied exclusively on T-cell receptor signaling. Tofacitinib preferentially suppresses signals dependent on cytokine receptors containing the γc chain (IL-2, IL-4, IL-7, IL-9, IL-15, IL-21)³⁹. Over 6 days, an average of 18% of healthy CD4⁺ T-cells became IFN-γ producers. Tofacitinib (30 nM) decreased the proportion of Th1-lineage-committed cells to 13%; indicating that one-third of the type II interferon-producing T-cells depended on JAK/STAT-derived signals (Fig.2A, 2B). CD4⁺ T-cells from GCA patients were spontaneously biased towards IFN-γ production. TCR activation induced intracellular IFN-γ stores in 25% of CD4⁺ T-cells. A substantial proportion of these cells were susceptible to JAK/STAT inhibition, which reduced frequencies of IFN-γ⁺ CD4⁺ T-cells to 15 % (Fig.2A, 2B).

Figure 2. CD4 effector T-cells from GCA patients are sensitive to JAK-STAT inhibition.

Peripheral blood CD4⁺ T-cells from GCA patients and age-matched healthy controls (HC) were stimulated with anti-CD3/CD28 in the absence or presence of the jakinib tofacitinib (10 nM; 30 nM). On day 6, intracellular IFN-γ was analyzed by flow cytometry. Fluorescence minus one (FMO) and isotype control were used as controls. (A) Representative dot blots. (B) Frequencies of IFN-γ⁺ CD4⁺ T-cells. Mean ± SEM from 12 healthy controls and 12 patients. One way ANOVA and pair-wise comparison using Tukey's method to adjust for multiple testing was applied. *P<0.05, ***P<0.001.

These data indicated that GCA patients guide CD4⁺ T-cells towards Th1 effector functions, specifically IFN-γ production, and that pharmacologic inhibition of JAK3/1 effectively suppresses this abnormality.

Tofacitinib inhibits inflammatory responses in the vessel wall

To investigate whether the pathogenic immune responses leading to vascular inflammation depend on γc-containing cytokine receptors, inflammation was induced in human arteries engrafted into NSG mice by immunoreconstituting such mice with PBMCs from GCA patients (Fig.3).

Figure 3. Tofacitinib suppresses T-cell expansion in the vasculitic lesions.

NSG mice engrafted with human medium-sized arteries were immunoreconstituted with PBMCs from GCA patients as in Suppl. Fig. 1. Vasculitic lesions accumulate within the human artery within 7-10 days. Subsequently, chimeric mice were treated with tofacitinib (3 mg/kg orally/day) or vehicle for one week. mRNA was prepared from explanted arteries for gene expression profiling (RT-PCR). Tissue sections were analyzed by H&E staining and immunohistochemistry (IHC). Tissue-residing cells were extracted from digested tissue. Each point represents one artery, paired arteries are line-connected. (A) Representative H&E stains from explanted arteries (600×). (B) IHC staining for CD3⁺ T-cells in the tissue sections (600×). (C) Density of the T-cell infiltrates measured by either quantification of TCR transcripts or enumeration of CD3⁺ cells by IHC (paired Wilcoxon test). (D) Tissue macrophage density estimated by CD163 transcript quantification (paired Wilcoxon test) (E, F) Proliferating T-cells (CD3⁺Ki-67⁺) in the vasculitic lesions detected by dual-color staining. Representative images (scale bar: 200 μm) and CD3⁺Ki-67⁺ cell percentages (compared by t- test). (G, H) CD4⁺CD103⁺ tissue-resident memory T-cells quantified by flow cytometry of tissue-extracted cells. Representative dot blots and absolute numbers per 100 mg of arterial tissue (p by t-test). **P<0.01.

Vessel wall infiltrates were markedly reduced in tofacitinib-treated grafts (Fig.3A). Especially, the density of tissue-infiltrating CD3⁺ T-cells declined. The concentration of TCR transcript was consistently lower after blockade of γc-chain–containing cytokine receptors. In a series of 8 independent experiments, numbers of tissue-residing T-cells reached 51.00±6.2/high-powered field and declined to 22.7±3.3/high-powered field in tofacitinib-treated mice (Fig.3B, 3C). Tofacitinib's T-cell–depleting effect was examined by enumerating T-cell counts in single-cell preparations extracted from digested arteries (Suppl.Fig.3). Resembling GCA-affected temporal arteries, CD4⁺ T-cells dominated the wall infiltrates and tofacitinib reduced the number of tissue-infiltrating CD4 and CD8 T-cells by about 80%. The immunosuppressive effect of the JAK3/1 inhibitor was not limited to T-cells. With reduced T-cell accumulation in the tissue, tissue macrophages declined as well; as indicated by the consistently lowered expression of the macrophage marker CD163 (Fig.3D). We examined whether JAK3/1 blockade affected recruitment and retention versus in situ expansion of tissue-infiltrating T-cells (Fig.3E, 3F). Tissue-residing T-cells were highly proliferative in vehicle-treated arteries. About 40% of tissue-residing CD3 T-cells expressed the proliferation marker Ki-67, suggesting that clonal expansion contributes significantly to the formation of the intramural lesions. As JAK3-dependent signaling was disrupted with tofacitinib treatment, <10% of tissue-entrapped T-cells were dividing (Fig.3F). Tofacitinib's anti-proliferative effect was not restricted to T-cells recruited to the vessel wall. Human T-cells in the blood and in the murine spleen also declined, implicating γ-chain⁺ cytokine receptors in homeostatic T-cell expansion in the murine host. Since the γc chain is mostly utilized by T-cell growth and differentiation factors (IL-2, IL-4, IL-7, IL-9, IL-15, IL-21), reduced frequencies of human CD14⁺ monocytes in tofacitinib-treated mice may be the result of impaired T-cell–derived survival signals for these myeloid cells (Suppl.Fig.4A-4D).

To understand how suppression of T-cell growth affects T-cell effector functions at the site of tissue inflammation, we analyzed gene expression patterns for lineage-determining transcription factors (T-bet, RORC, Bcl-6, FoxP3) as well as transcripts for the T effector cytokines IFN-γ, IL-17 and IL-21. Th1, Th17 and Tfh cells have all been localized to the vasculitic lesions and have been implicated in tissue-damaging inflammation⁴⁰. Tofacitinib treatment left expression of FoxP3 transcripts unaffected, but reduced concentrations of T-bet, RORC and BCL-6 by more than 40% (Fig.4). The immunosuppressive effect of the JAK3 inhibitor was even more pronounced for effector cytokines. IFN-γ and IL-17 mRNA was reduced to minimal levels and IL-21 mRNA declined by 58% (Fig.4).

Figure 4. Tofacitinib suppresses multiple vasculitogenic effector T-cell lineages.

Vasculitis was induced in human arteries and chimeric mice were treated with vehicle or tofacitinib as in Figure 3. Gene expression of transcription factors (A-D) and T-cell effector cytokines (E-G) was quantified by RT-PCR of tissue-extracted mRNA. Each data point represents one artery and paired arteries are connected by a line. **p<0.01 by paired Wilcoxon test. ns: not significant. Comparisons of T-bet, RORC, BCL-6, IFN-γ, IL-17 and IL-21 are statistically significant after controlling the family-wise error rate at the 0.05 level using Hochberg's step-down adjustment.

Taken together, common γ-chain function appears critical in the pathogenic cascade of vascular inflammation and JAK1/3 inhibition effectively interferes with the expansion of granulomatous lesions and the pro-survival signals produced in the tissue microenvironment.

Tissue-resident memory T cells (T_RM) require JAK1/3 function to survive

T_RM have recently been recognized as a transcriptionally, phenotypically and functionally distinct lymphocyte lineage, distinct from classical memory T-cells⁴¹. Specialized to keep residency in the tissue microenvironment, T_RM are believed to provide fast and effective immune responses when reencountering antigens¹¹. Expression of the alpha E integrin CD103, which combines with the integrin beta 7 to form the ligand for E-cadherin, has been mechanistically linked to tissue anchoring⁴². Given their persistence in peripheral tissues, T_RM are considered essential pathogenic elements in chronic inflammatory disease¹⁰. In support of the concept, that intramural vasculitic foci contain T_RM, we found that tissue-extracted T-cells included a population of CD4⁺CD103⁺ T-cells, that was essentially exclusively present in the arterial wall and was barely detectable in the peripheral blood (Fig.3G, 3H and Suppl.Fig.5). This population of CD4 T memory cells was highly sensitive to tofacitinib, implicating γc chain-signaling cytokines in its survival.

To further study T_RM in vasculitic lesions, we developed a transengraftment model, in which arterial inflammation was first induced in engrafted human arteries, before they were transplanted into an “empty” NSG mouse (Fig.5A). This model system avoided competition between continuous recruitment of new T-cell specificities into the vessel wall with T-cells residing and proliferating in the lesion. Following the removal from continuous T-cell supply, the arterial wall infiltrates maintained high proliferative activity, the density of tissue-invading T-cells grew and the expanding T-cell populations moved deeper into the vascular wall layers (Fig.5B, 5C). A subset of 5-8% of T-cells acquired expression of the CD103 marker, as identified by flow cytometric analysis of cells isolated out of the human arteries (Fig.5G, 5H). About 25% of the tissue-residing T-cells were Ki-67 positive, identifying them as dividing cells (Figure 5E, 5F). Over a period of 2 weeks, the number of T-cells seen in tissue sections doubled (Fig.5C, 5D). Growth in the vasculitic infiltrates was associated with 2-3–fold increase in the expression of transcripts specific for T-bet, RORC, IFN-γ, IL-17 and IL-21 (Fig.5I). If the “empty” NSG mice engrafted with the already inflamed arteries were treated with a daily oral dose of tofacitinib (3 mg/kg), the density of tissue-infiltrating T-cells declined markedly, returning to levels seen in nontransplanted arteries (Fig.5C,5D). In parallel, accumulation of cytokine-producing T-cells was effectively stopped, as demonstrated by the marked reduction in T-bet, RORC, IFN-γ, IL-17 and IL-21 mRNA (Fig.5I). Most importantly, tofacitinib effectively disrupted survival signals for CD4⁺CD103⁺ T_RM cells, which sank to low frequencies in artery grafts treated with the jakinib (Fig.5G, 5H).

Figure 5. Tofacitinib shortens the survival of tissue-resident memory T-cells.

Vasculitis was induced in engrafted human arteries as in Figure 3. After vasculitis induction, arteries were transengrafted into an “empty” NSG mouse. Chimeras with arteries from the same donor were assigned to parallel treatment arms: oral tofacitinib (3 mg/kg/day) or vehicle for 7 days. Explanted arteries were sectioned for H&E and immunofluorescence staining, or tissue-residing cells were extracted from digested explants and mRNA was extracted for transcriptome analysis by RT-PCR (n=8 arteries). (A) Scheme of the animal experiments. (B) H&E sections from pretransengraftment and posttransengraftment arteries. Representative images (scale bar: 200 μm). (C) Representative IHC images from pretransengraftment and posttransengraftment arteries stained with anti-CD3. (D) T cell numbers in tissue sections compared by paired Wilcoxon test. (E, F) Proliferating CD3⁺Ki-67⁺ T-cells in transengrafted arteries. Representative images and frequency of Ki-67⁺ T-cells. (G, H) Flow cytometry of tissue-extracted CD4⁺CD103⁺ tissue-resident memory T-cells. Representative dot blots and frequencies (t-test performed). (I) Gene expression profiling in arterial grafts before and after transengraftment assessed by RT-PCR. **p<0.01 by paired Wilcoxon testing. Comparisons of TCR, T-bet, RORC, IFN-γ, IL-17 and IL-21 remain statistically significant after controlling the family-wise error rate at the 0.05 level using Hochberg's step-down adjustment.

To examine whether the microenvironment of the inflamed arterial wall is permissive to support the induction and survival of T_RM, we searched for the presence of IL-7, IL-9 and IL-15 in the tissue microenvironment (Suppl.Fig.6), as these cytokines have been implicated in supporting T_RM⁴³. Induction of vessel wall inflammation was associated with significant upregulation of IL-7, IL-9 and IL-15 (Suppl.Fig.6A-6C); supplying critical ingredients for local T-cell persistence and growth.

In essence, T_RM in the inflamed vessel wall require JAK3/1-dependent signaling to survive and inhibiting the kinase activity appears to be an effective strategy in halting vasculitogenic immune responses.

Jak3-STAT signaling drives inflammation-associated microangiogenesis in the inflamed vessel wall

When affected by GCA, large and medium arteries develop a dense network of microvessels, leading to adventitial thickening and giving rise to the enhancement effect on imaging studies¹. Newly formed microvessels are identified by tissue stains for α-SMA–expressing pericytes and vWF⁺ endothelial cells⁶. Adventitial vasa vasorum are present in healthy, noninflamed arteries (Fig.6A, left) and remained essentially unchanged, if arteries were engrafted into mice lacking patient PBMC (Fig.6A central, 6B). Induction of vasculitic wall infiltrates in PBMC-reconstituted chimeras was associated with robust neoangiogenesis (Fig.6A right, 6B), indicating the critical role of the donor immune cells in the process of neoangiogenesis. The numbers of microvessels identified by immunostaining in tissue sections were more than 10-fold higher in arterial grafts after successful induction of transmural inflammation (Fig.6B).

Figure 6. Tofacitinib inhibits inflammation-associated microangiogenesis.

Human artery-SCID chimeras were generated as in Figure 3 and treated with vehicle or tofacitinib (3 mg/kg/day) as described above. Explanted arteries were analyzed for microangiogenesis by immunofluorescence staining of α-SMA, vWF and DAPI. Gene expression was quantified by RT-PCR. (A) Representative images of α-SMA⁺vWF⁺ adventitial microvessels. Left: healthy medium-sized artery; middle: artery explanted from mice that did not receive human PBMCs; right: artery explanted from chimeras that were immunoreconstituted with GCA PBMCs (scale bar: 200 μm). (B) Quantification of adventitial microvessels (n=4 each; comparison by t-test). (C, D) α-SMA⁺/vWF⁺ microvessels in vehicle and tofacitinib-treated grafts. Representative images (scale bar: 200 μm) and microvessel density in paired arterial grafts. Paired Wilcoxon test applied. (E) Tissue gene expression of angiogenic cytokines in vehicle and tofacitinib-treated explants. Paired arterial grafts are line-connected. *P<0.05, **P<0.01, ***P<0.001 by paired Wilcoxon test.

To explore whether γ-chain–dependent signaling had a role in this damage-associated pattern, chimeras with inflamed human arteries were assigned to a vehicle or a tofacitinib treatment arm (Fig.6C-6E). Grafts explanted from vehicle treated animals had dense networks of newly formed microvessels (Fig.6C) and tissue-extracted RNA contained high transcript levels for the angiogenic factors PDGF, FGF2 and VEGF (Fig.6E). Inhibition of JAK3 activity with tofacitinib robustly suppressed the neoangiogenic response. Numbers of microvessels were less than 50% of those in the vehicle control arteries (Fig.6C, 6D). Tissue concentrations of transcripts specific for PDGF, FGF2 and VEGF (Fig.6E) were strongly tofacitinib-responsive.

These experiments implicated signaling through the γc chain in the disease-relevant process of microangiogenesis, which supplies oxygen and nutrients to the thickening arterial wall and the expanding intimal layer. The Jakinib tofacitinib was sufficient to arrest the neoangiogenic program, in support of a direct role for activated T-cells in the growth of new vasa vasora.

JAK1/3-STAT signaling promotes intimal hyperplasia

The ischemic complications of GCA, above all the ischemic damage to the optic nerve, result from arterial stenosis/occlusion, caused by fast and concentric intimal hyperplasia⁴⁴. Induction of transmural inflammation in engrafted human arteries was associated with high proliferative activity in vascular cells, both in the media and the intima, leading to growth of the intimal layer (Fig.7A, 7B). Over a disease period of 2 weeks, the tunica intima grew to a thickness of 30-40 μm, equivalent to 15-20 cell layers. Proliferative activity was encountered in two cellular populations: (1) cells expressing the smooth muscle cell/fibroblast marker α-SMA, which were enriched in the adventitia and intima; (2) α-SMA^neg cells localized within the immune cell infiltrates and centering on the media (Fig.7C, 7D). Such proliferating α-SMA-negative cells represent T-cells and macrophages (Fig.3). Tofacitinib treatment essentially curbed the proliferative activity of all dividing cell populations (Fig.7C, 7D). The hyperplastic reaction of the intima was aborted, thickness measurements fell from 35 to 10 μm (Fig.7A, 7B). The anti-proliferative effect of tofacitinib equally affected α-SMA-expressing vascular cells and infiltrating immune cells; which may reflect direct and indirect inhibition of growth signals. Few dividing cells remained in the arteries treated with tofacitinib (Fig.7C, 7D).

Figure 7. Tofacitinib inhibits intimal hyperplasia.

Human artery-SCID chimeras were generated as in Figure 3 and treated with vehicle or tofacitinib (3 mg/kg/day) for 2 weeks. (A) H&E staining of tissue sections from vehicle or tofacitinib-treated arteries. Representative images (100×). (B) Intimal layer thickness measured in arterial cross-sections (n=5) Data from paired grafts are connected by a line. T-test was applied. (C, D) α-SMA⁺ and α-SMA^– cells expressing the proliferation marker Ki-67 in vehicle and tofacitinib-treated arteries. (C) Effect of tofacitinib treatment on the frequencies of Ki-67⁺ proliferating cells assessed amongst α-SMA⁺ and α-SMA^– cells. Data are from 6 paired grafts. p by paired Wilcoxon test. (D) Representative images showing intimal α-SMA⁺/Ki-67⁺ cells and medial α-SMA^-Ki-67⁺ cells (scale bar: 200 μm). *P<0.05, **P<0.01.

These data connected γc chain cytokine responses to the damage pattern of intimal hyperplasia. Blocking Jak3-STAT signaling suppressed proliferative activity of immune and nonimmune cells, successfully protecting the vascular wall from inflammation-induced remodeling.

Discussion

GCA is a prototypical large vessel vasculitis, caused by vasculitogenic T-cells and macrophages building organized lymphoid microstructures in the otherwise immuno- privileged niche of the vessel wall. Long-lived T-cells coordinate effector functions of accompanying macrophages and vessel wall cells react with a maladaptive repair response, ultimately promoting vascular failure. Acutely, GCA is treated with high-dose corticosteroids, which suppress nonspecific innate immunity, but spare the multifunctional T-cells trapped in the lesions^2,40. Persistent functional activity of lesional T-cells creates smoldering vasculitis, the ultimate challenge in the management of affected patients.

Here, we have characterized the proliferative capacity of the vasculitogenic T-cells and have identified CD4⁺CD103⁺ T_RM anchored in inflammatory wall lesions. CD4⁺CD103⁺ T_RM function as key drivers of long-lasting inflammation and represent preferred therapeutic targets to eradicate inflammatory vasculopathy. Tissue-resident and mobile CD4⁺ memory T-cells both relied on survival signals through γc chain-containing cytokine receptors; as documented by the powerful immunosuppressive effects of the JAK1/3 inhibitor tofacitinib. The therapeutic effectiveness of this jakinib focusses attention towards cytokines binding type I and type II cytokine receptors, in particular interferons.

Early studies in GCA temporal arteries emphasized the sharing of T-cell receptor sequences in physically separated vasculitic foci, compatible with antigen driving lesional T- cells. However, the concept of a single vasculitogenic autoantigen has recently been replaced by a disease model that focusses on antigen-nonspecific defects driving vasculitis. The emerging paradigm recognizes broad immuno-regulatory defects in GCA, indicative of T cell hyperreactivity due to altered activation threshold setting. Hyperreactive T-cells in GCA patients result from failed immune checkpoints that control the response to many antigens. Specifically, NOTCH1⁺ CD4⁺ T-cells in GCA patients receive activating signals from Jagged1⁺ microendothelial cells in the arterial adventitia³¹. Also, a defect in the immunoinhibitory PD-1 checkpoint enables unopposed T-cell activity in the vascular lesions⁶. And, patients' CD4⁺ T-cells are insufficiently suppressed due to the failure of immunoinhibitory CD8 Treg cells⁴⁵. These data emphasize the importance of the tissue environment in enabling inappropriate T-cell immunity and reveal abnormalities in T cell responsiveness that are applicable to a broad spectrum of antigens. Which antigens are recognized in the human artery-SCID mice is not known, but alloantigens serve as model antigens to probe T-cell reactivity in GCA. Insufficient control of a broad range of adaptive immune responses in GCA patients appears best matched by therapeutically targeting the T cell's signaling machinery; in line with the anti-inflammatory effects of the JAK1/3 inhibitor tofacitinib demonstrated in the current study.

Several lines of evidence support the concept that interferons, including type I (IFN-α/β) and type II (IFN-γ) are of particular relevance in GCA. IFN-α/β and IFN-γ trigger JAK-STAT signaling, with both types of interferon activating STAT1⁴⁶. STAT2 functions as a transcription factor for type I interferon-dependent responses⁴⁷. The tissue transcriptome of GCA-affected arteries contained a signature indicative for in situ interferon responses: IFN-α and IFN-γ transcripts were significantly increased, target genes for STAT1 (T-bet, CXCL9) and STAT2 (ISG15, OAS1) were enriched and STAT1 and STAT2 were selectively upregulated; supporting the pinnacle role of T-cells in the disease process. The IFN-γ–rich environment provides ideal conditions for highly activated and functionally diverse macrophages arranged in the typical granulomatous infiltrates³⁷ How this IFN-γ–rich environment promotes other disease components, especially the formation of new blood vessels and the hyperplasia of the intima, is more difficult to understand. The current paradigm holds that both types of interferon are anti-angiogenic. Previously, macrophages and multinucleated giant cells were identified as the producers of the angiogenic factor VEGF in GCA-affected temporal arteries⁴⁸. Indeed, in temporal artery biopsies, tissue IFN-γ and VEGF were strongly associated with the degree of intramural vessel formation; demonstrating that in this vasculitis IFN-γ promotes and not suppresses neoangiogenesis⁴⁸.

Unexpectedly, STAT3 target genes were at low abundance in the temporal artery biopsies; questioning in-situ action of IL-6, which may mostly function in extravascular sites. Also, the low expression of STAT6 target genes is in line with a lack of IL-4 production in GCA lesions⁴⁹.

The current treatment paradigm in GCA is the use of corticosteroids, which preferentially target innate cytokines, such as IL-1β, IL-12 and IL-6³⁶, but have little effect on tissue-residing T-cells^2,36. Therapeutic needs therefore lie in eliminating the T-cell infiltrates. Functionally, lesional T-cells are highly diverse⁴⁰, but data presented here indicate that they dependent on common survival signals. Inflamed arteries were rich in T-cell growth and survival cytokines (Suppl.Fig.6) and inhibiting the activity of JAK1/3 was sufficient to broadly diminish lineage-determining transcription factors (T-bet, RORC and BCL-6) and T-cell effector molecules (IFN-γ, IL-17 and IL-21). Tofacitinib minimized the in-situ proliferation of lesional T-cells and flow cytometric analysis confirmed that the T-cell suppressive effect of disrupting JAK-STAT signaling included CD4⁺CD103⁺T_RM. Tofacitinib's powerful immunosuppressive effect is encouraging, as it may be able to disrupt upstream pathogenic drivers and turn the disease from a long-lasting wall inflammation into an acute and treatable condition.

Tissue-resident memory T-cells (T_RM) have recently been described as a specialized T-cell subset defined by its exquisite ability to anchor in peripheral tissues instead of returning to secondary lymphoid organs. For protective immunity, T_RM provide fast and locally effective protection. For vasculitogenic immunity, such T-cells may hold the key to chronicity and persistent wall injury, even after good control of peripheral inflammatory markers has been reached. We confirmed the tissue residence of CD4⁺CD103⁺ T-cells by transengrafting inflamed arteries into “empty” mice, making vascular lesions the sole reservoir for human cells. CD4⁺CD103⁺ T-cells were explicitly infrequent in the periphery, but populated the transmural vascular lesions. Besides their role in driving chronic vascular inflammation, T_RM may provide important clues on pathogenic principles in inflammatory vasculopathies. In the vascular tree, GCA is characterized by stringent tissue tropism. Parameters defining the susceptibility of certain arteries to the disease, while others are protected, have not been identified. The ascending aorta, the aortic arch, the distal subclavian arteries, and the temporal arteries are particularly susceptible. Considering the critical role of T_RM in driving persistent inflammation, the local microenvironment may dictate the survival of T_RM. In this model, nonpermissive microenvironments would protect the immunoprivilege of many arterial walls by failing to support the persistence of tissue-residing T_RM. Molecular definition of such microenvironmental niches could greatly enhance our understanding of the disease process.

In conclusion, persistence of vessel wall inflammation in medium and large vessels appears to be dependent on a small population of highly specialized T-cells, characterized by their ability to survive in the tissue microenvironment. The longevity of such CD4⁺CD103⁺ T_RM requires cytokine signals that are mediated through the common γ-chain; rendering such T_RM susceptible to kinase inhibitors that can disrupt γ-chain–dependent signaling. Depletion of such CD4⁺CD103⁺ T_RM was sufficient to inhibit a cascade of pathogenic events: the production of effector cytokines (IFN-γ, IL-17 and IL-21), the formation of inflammation-induced microvascular networks and the expansion of the intimal layer. The JAK inhibitor tested here is approved for treatment of patients with rheumatoid arthritis and has been shown to have acceptable toxicity risks⁵⁰. Targeting dysregulated T-cells instead of suppressing innate immunity (e.g. through corticosteroids) promises to fulfill a gap in the management of patients with vasculitis, while also providing informative insights into basic immune abnormalities of patients with aortitis and medium-vessel vasculitis.

Clinical Perspective.

What is New?

• STAT1 and STAT2 target genes, the transcription factors STAT1 and STAT2 and the STAT-pathway activators type I and type II interferon are abundant in the tissue transcriptome of arteries with GCA.

• The JAK/STAT inhibitor Tofacitinib suppresses the ex-vivo induction of IFN-γ⁺ T-cells in GCA patients.

• Chimeric mice carrying human arteries and immune cells from GCA patients develop persistent vasculitis.

• In the chimeras, Tofacitinib efficiently suppresses T-cell invasion into the artery, inhibits proliferation and cytokine production of vasculitogenic T-cells and curbs survival of artery-resident T-cells.

• Tofacitinib treatment prevents neoangiogenesis and intimal hyperplasia in inflamed arteries.

What are the Clinical Implications?

• The JAK/STAT inhibitor Tofacitinib effectively targets multiple disease-relevant processes in inflammatory vasculopathy and represents a potential disease-modifying agent.