Immune checkpoint dysfunction in large and medium vessel vasculitis

Abstract

Giant cell arteritis (GCA) is a granulomatous vasculitis of the aorta and its medium-sized branch vessels. CD4 T cells, macrophages, and dendritic cells (DCs) build granulomatous infiltrates that injure the vessel wall and elicit a maladaptive response to injury. Pathological consequences include fragmentation of elastic membranes, destruction of the medial layer, microvascular neoangiogenesis, massive outgrowth of myofibroblasts, and lumen-occlusive intimal hyperplasia. Antigens have been suspected to drive the local activation of vasculitogenic CD4 T cells, but recent data have suggested a more generalized defect in the threshold setting of such T cells, rendering them hyperreactive. Under physiological conditions, immune checkpoints provide negative signals to curb T cell activation and prevent inflammation-associated tissue destruction. This protective mechanism is disrupted in GCA. Vessel wall DCs fail to express the immunoinhibitory ligand programmed cell death ligand-1, leaving lesional T cells unchecked. Consequently, programmed cell death protein-1-positive CD4 T cells can enter the immunoprivileged vessel wall, where they produce a broad spectrum of inflammatory cytokines (interferon-γ, IL-17, and IL-21) and have a direct role in driving intimal hyperplasia and intramural neoangiogenesis. The deficiency of the programmed cell death protein-1 immune checkpoint in GCA, promoting unopposed T cell immunity, contrasts with checkpoint hyperactivity in cancer patients in whom excessive programmed cell death ligand-1 expression paralyzes the function of antitumor T cells. Excessive checkpoint activity is the principle underlying cancer-immune evasion and is therapeutically targeted by immunotherapy with checkpoint inhibitors. Such checkpoint inhibitors, which unleash anticancer T cells and induce immune-related toxicity, may lead to drug-induced vasculitis.

autoimmune inflammation, attacking the walls of the aorta and its side branches, results in aneurysm formation, rupture, and dissection and for medium-sized vessels in luminal occlusion. To protect the host from such life-threatening complications, the wall layers of vital arteries are immunoprivileged, rendering them resistant to localized immune and inflammatory responses (53). The role of the immunoprivilege in the aortic wall is broken in giant cell arteritis (GCA), an autoimmune and autoinflammatory disease causing aortitis and arteritis of the second to fifth aortic branches (3, 44, 52). Clinical consequences include wall damage to the aorta, aortic arch syndrome, and ischemic complications in the eye and the posterior brain. Almost always, the vascular inflammation in GCA patients is combined with a syndrome of systemic inflammation, manifesting as fever, failure to thrive, polymyalgia rheumatica, and exuberant production of acute-phase reactants (15). GCA is diagnosed by a biopsy of the temporal artery, typical findings on computer tomography, or magnetic resonance angiography. Therapeutically, induction therapy continues to rely on high doses of corticosteroids (52). Recently, newer immunosuppressants have been added to the therapeutic armamentarium, but they do not appear to induce durable disease remission (3, 50). Chronic smoldering vasculitis is now the most significant clinical challenge, complicated by the lack of data on the risk/benefit ratio of chronic immunosuppression and by incomplete understanding of the immunopathology of persistent vasculitis.

The Immune Lesion in GCA T Cells and Macrophages Rescinds the Artery’s Immune Privilege

The histomorphology of GCA-affected arteries describes intimal hyperplasia, fragmentation of the elastic lamina, and intramural neoangiogenesis, which are all understood as a maladaptive response to intramural granulomatous inflammation (Fig. 1). Typically, GCA-affected arteries are occupied by granulomatous infiltrates composed of CD4 T cells, highly activated macrophages, multinucleated giant cells, and dendritic cells (DCs) (53). CD4 T cell clones with identical T cell receptors have been isolated from independent tissue biopsies of GCA patients, nurturing the concept of a vasculitogenic antigen (13, 56). Functional analysis of tissue cytokines produced in vasculitic lesions has supported the notion of multifunctional T cells trapped in the lesions. Vascular wall resident T cells produce interferon (IFN)-γ, IL-17, IL-9, and IL-21, implicating T cell helper (Th)1, Th17, Th9, and follicular Th (T_FH) cell lineages in the disease process (Fig. 1) (5, 6, 46, 51). Notably, tissue IL-17 production is highly sensitive to corticosteroid therapy, whereas IFN-γ-producing T cells seem unaffected by high doses of systemic steroids (6, 53), supporting the concept of functional heterogeneity among lesional T cell populations. The absence of CD4 and CD8 regulatory T cells in the vascular wall is in line with the observation that GCA immune lesions are driven by multiple effector T cell types.

Fig. 1.

Immune lesion in giant cell arteritis (GCA). A: healthy medium-sized artery with open lumen and three wall layers. Vascular dendritic cells (DCs) are placed at the media-adventitia border, where they may function to guard the immune privilege of the vessel wall. B: GCA is caused by granulomatous inflammation within the arterial wall layers. The major cellular players are CD4 T cells, highly activated macrophages, and DCs. Different inflammatory pathways dominate in different territories of the wall. Boxes indicate the cell types and their disease-relevant products that form the adventitial infiltrate and the medial infiltrate. The adventitia is an important site of antigen presentation and T cell activation. Multiple types of T effector cells are placed in the adventitia and media. Macrophage effector functions are critically involved in the tissue damage and wound-healing response centered in the media. The artery responds to inflammation with a maladaptive restructuring program. Neoangiogenesis in the adventitia opens the floodgates for infiltrating inflammatory cells. Myofibroblast mobilization and migration create rapidly progressive intimal hyperplasia, leading to luminal occlusion and the ischemic complications of GCA. TGF-β, transforming growth factor-β; T, T cell.

Many of the granuloma-residing cells are macrophages, often called histiocytes. Their functional commitment is tightly correlated to their location in the vessel wall (Fig. 1) (48), emphasizing that different microenvironments serve different functions in the disease process. Macrophages in the hyperplastic intima are proinflammatory, producing inducible nitric oxide in the vasculitic lesions (57). Medial macrophages have been implicated in vessel wall remodeling and oxidative stress protection, producing growth factors (platelet-derived growth factor, VEGF, and FGF), metalloproteinases, and aldose reductase (17, 18, 39, 40). Adventitial macrophages coproduce transforming growth factor-β₁, IL-6, and IL-1β (57). Available data suggest that inhibition of vasculitic macrophages, effectively achieved by treating GCA patients with glucocorticoids, will not restore the vascular wall immune privilege (6). Studies of repeat biopsies in patients before and after glucocorticoid therapy have indicated that vasculitic infiltrates persist in most patients, even after 9–12 mo of therapy (27).

Evidence for antigen presentation within the granulomas derives from the prominent participation of DCs. As an indigenous population of the vessel wall (22, 26, 36), vascular DCs are believed to have a critical role in protecting the artery’s immune privilege. Equipped with costimulatory ligands and chemokines, they control cellular flux and amplify T cell responses (14, 22, 26, 32, 36, 55). DCs are also the source of immunoinhibitory ligands, providing negative signals to T cells, thus critically shaping the size and duration of immune responses (20, 59). Recent data demonstrate that the programmed cell death protein-1 (PD-1) immune checkpoint, through which programmed cell death ligand-1 (PD-L1)⁺ DCs provide a stop signal to PD-1⁺ T cells, is defective in GCA (62), emphasizing the regulatory importance of lesional DCs.

What Are Immune Checkpoints?

Protective and pathogenic T cell immunity is a multiple-step process centering on the clonal selection of antigen-specific cells. Upon encountering an antigen, T cells undergo activation and proliferate massively to expand the clonal size. Some cells will develop into memory T cells and enter storage sites in secondary lymphoid tissues. Others will traffic to peripheral tissues to reencounter the antigen, execute direct effector functions, or provide help (through cytokine release and membrane ligands) for a multitude of immune effector cells (45, 61). Each of these steps underlies regulation by counterbalancing stimulatory and inhibitory signals, ultimately to regulate the amplitude, durability, and quality of the immune response (Fig. 2) (12). Classical costimulatory signals derive from CD28-CD80 or CD28-CD86 interactions, which are an absolute requirement to drive the metabolic programs of expanding T cells (8). Incoming “go” signals are paired with “stop” signals, often called immune checkpoints. Well-recognized stop signals restraining the activation and clonal expansion of T cells are derived from cytotoxic T lymphocyte-associated antigen-4 (CTLA-4; CD152) and PD-1 (CD279). Both immune checkpoint receptors are critically involved in avoiding excessive immunity, protecting the host from collateral damage, and preventing autoimmunity (30, 38, 63). Both receptors are now recognized as targets for cancer immunotherapy (33, 35, 37), as tumors have learned how to usurp antitumor immunity by strengthening immune checkpoint signaling. Whereas CTLA-4 is believed to disrupt early steps in T cell activation, PD-1 is typically upregulated on T cells after activation, providing an opportunity to inhibit later stages of the clonal expansion program. PD-1 triggering, by its ligands PD-L1 and PD-L2, activates the phosphatase Src homology 2 domain-containing phosphatase-2 (1), thus inhibiting key kinases transmitting the T cell activation cascade. PD-L1 and PD-L2 are expressed on antigen-presenting cells, tumor cells, and tissue-residing cells, empowering them to determine the strength and longevity of immune responses (Fig. 2) (7, 9, 64).

Fig. 2.

Costimulatory and coinhibitory signals in T cell regulation. T cells interact with antigen-presenting cells to recognize their cognate antigen. The T cell receptor (TCR) binds to the human leukocyte antigen (HLA)-antigen complex to trigger the T cell activation cascade. Concomitant receptor-ligand interactions provide either positive signals (costimulation) or transmit negative signals (coinhibition), ultimately to adjust the quality, intensity, and duration of the T cell response. Shown is the major costimulatory molecule CD28, which binds to CD80 and CD86 to amplify the T cell activation cascade. Also shown are the two major inhibitory pathways. The inhibitory receptors cytotoxic T lymphocyte-associated antigen-4 (CTLA-4) and programmed cell death protein-1 (PD-1), expressed on T cells, are triggered by their ligands on the antigen-presenting cell and provide a “stop” signal to the T cell.

The PD-1 Immune Checkpoint in GCA

PD-L1 is expressed on antigen-presenting cells, some stromal cells, and tumor cells (7, 9, 64). Gene expression profiling has demonstrated high expression of PD-L1 transcripts in healthy, noninflamed human arteries (62), where this immunoprotective ligand may contribute to the immune privilege. Conversely, GCA-affected arteries had distinctly low PD-L1 transcript levels, and wall-embedded DCs had no demonstrable PD-L1 expression. Low PD-L1 protein expression held true for ex vivo-generated DCs from GCA patients, whereas DCs generated from patients with rheumatoid arthritis (RA) responded to activation signals with robust PD-L1 induction. Functional testing revealed that patient-derived PD-L1^low DCs provided strong T cell-activating signals, and blockade of PD-L1, an intervention known to amplify T cell stimulation, failed to enhance further T cell responses. In essence, PD-L1 deficiency of GCA DCs leaves the patient with unopposed T cell-activating signals (Fig. 3).

Fig. 3.

The defective PD-1 immune checkpoint in giant cell arteritis. DCs from patients with GCA express high levels of the costimulatory ligands CD80 and CD86 but low levels of inhibitory programmed cell death ligand-1 (PD-L1). As a consequence, they fail to inhibit interacting T cells. Uninhibited T cells proliferate and acquire multiple effector functions. In the vasculitic lesions of GCA, the majority of T cells carry the PD-1 receptor. Lesion-residing T cells are multifunctional, produce proinflammatory cytokines [e.g., interferon (IFN)-γ, IL-17, and IL-21] and promote vessel wall restructuring by enhancing microvascular neoangiogenesis and accelerating intimal hyperplasia.

How does weakness of PD-L1-induced stop signals affect the T cell response in GCA? In inflamed temporal arteries of GCA patients, essentially all T cells trapped in the granulomas are PD-1 positive. PD-1 is considered a delayed activation marker on T cells recognizing antigen. More importantly, the presence of this surface molecule has been associated with two outcomes: T cell exhaustion and Th activity in the germinal center, provided by so-called PD-1⁺ T_FH cells (43, 58). Neither of these two scenarios applies to the granulomatous lesions in GCA. T cells occupying the inflamed vessel wall are certainly not exhausted. They are highly activated, and local production of several powerful T cell cytokines has been described. IFN-γ appears to be particularly important, and tissue IFN-γ has been associated with the intensity of the wall remodeling process (18). Lesional T cells also do not behave like T_FH cells; they colocalize with macrophages but not with B cells. The granulomatous infiltrates are distinctly negative for B cells, and no autoantibody production has been identified for GCA patients (28).

The functional activity of PD-1⁺ T cells in the inflamed arterial wall has been examined in a model system of GCA, created by engrafting normal human arteries into immunocompromised mice (34, 62). Such chimeric mice are then reconstituted with peripheral blood mononuclear cells, and vasculitis develops within 1–2 wk. Treatment of such human artery-mouse chimeras with anti-PD-1 antibody resulted in substantial aggravation of vessel wall inflammation and T cell infiltration (Fig. 4). Tissue gene expression analysis revealed marked enhancement of innate and adaptive immunity after PD-1/PD-L1 interactions had been dismantled. Artery grafts from anti-PD-1-treated chimeras were enriched for a broad array of cytokine transcripts: IL-1β, IL-6, TNF-α, IL-23 p19, IL-27 p28, IL-7, and IL-15. These data are most compatible with PD-1-dependent signaling, regulating tissue-residing T cells and macrophages. The impairment of the PD-1-derived stop signal ultimately aggravated a multitude of effector functions and overall enhanced the intensity of the vasculitogenic immune response.

Fig. 4.

Checkpoint inhibition with anti-PD-1 antibody treatment exacerbates vasculitis. Human arteries were engrafted into immunodeficient nonobese diabetic NSG-γ mice. To induce vasculitis in the engrafted vessels, chimeric mice were reconstituted with peripheral blood mononuclear cells from patients with GCA. Two weeks later, the human arteries were explanted, and the intensity of vasculitis was determined by immunostaining for human CD3⁺ T cells in tissue sections. Before harvesting of the human arteries, the chimeric mice were treated with anti-PD-1 antibodies (100 µg) or control IgG by alternative-day intraperitoneal injections. Anti-CD3-binding T cells (brown) in the tissue were visualized with horseradish peroxidase-conjugated goat anti-rabbit secondary antibodies. Compared with the IgG control (left), PD-1 blockade (right) markedly increased the density of the vascular T cell infiltrate. Original magnification: ×600.

Arguably, the most important observation made in the human artery-mouse chimera model (62) was the impact of disrupting the PD-1 signal on the remodeling of the inflamed arterial wall. The increase of the density of PD-1⁺ T cells in the vasculitic lesions led to aggravation of intramural neoangiogenesis and the induction of endothelial cell activation markers. Typically, GCA-affected arteries have dense networks of newly formed microvessels, which are believed to enable the fast recruitment of inflammatory cells and supply the growing inflammatory lesion with oxygen and nutrients. Higher density of PD-1⁺ T cells in the arterial wall was also associated with the size of the intimal layer, doubling intimal thickness. Intimal hyperplasia causes the ischemic complications of GCA, such as blindness and stroke (52, 54). These data place PD-1 signaling and the intensity of T cell activation at the pinnacle of the wall remodeling process. Further studies need to identify the cellular partners of PD-1⁺ T cells, e.g., endothelial cells, vascular smooth muscle cells, and others, that ultimately execute the pathological events.

The placement of the deficiency of the immunoinhibitory ligand PD-L1 to vascular DCs raises the question of how the presentation of endogenous and exogenous antigens is affected. Vasculitogenic antigens in GCA have remained unknown, and episodic reports on infectious antigens expressed in vasculitis-affected arteries have not been confirmed in independent studies. An appealing concept for endogenous antigens driving vascular disease has come from studying hypertension (16, 23, 29). In hypertension, increased oxidative stress in DCs has been associated with the formation of highly reactive γ-ketoaldehydes (21, 60). These highly reactive protein modifications have been implicated in altering protein structure and function and in the generation of neoantigens to which the immune system is not tolerant. Consequences include strong activation of DCs and inducing production of IL-6, IL-1β, and IL-23, which, in turn, enable interacting T cells to secrete IL-17, IFN-γ, and TNF-α. Whether similar mechanisms are at work in medium and large arteries susceptible to vasculitis is not understood. Obviously, the weakening of checkpoint function by PD-L1 loss could act as an amplification loop, supporting immune recognition of otherwise ignored self-antigens.

It is currently unknown whether insufficiency of the PD-1 checkpoint is a selective defect in GCA or is also of relevance in other autoimmune diseases. Since the defective PD-1 checkpoint appears to unleash immunity in the vessel wall, it would be equally important to explore its role in other cardiovascular diseases, such as atherosclerosis and hypertension. Available data in mouse models of atherosclerosis support the concept that an intact PD-1/PD-L1 pathway is necessary to suppress T cells that promote atherosclerotic lesion formation and plaque inflammation (2, 11). The risk of patients diagnosed with GCA to develop aortic aneurysm is doubled compared with age-matched controls (42), likely a direct manifestation of vasculitic involvement of the aorta. Furthermore, individuals affected by GCA are more likely to be diagnosed with cerebrovascular and cardiovascular disease, more so for older men from disadvantaged socioeconomic areas (41). Intriguingly, a recent study (47) has identified lower body mass index as a predictor of GCA risk, suggesting divergent risk scenarios for those susceptible to large vessel vasculitis versus those with classical atherosclerosis.

Conclusions, Open Questions, and Clinical Implications

Under physiological conditions, the wall of large arteries is protected from immune attack. In patients with large vessel vasculitis, such as GCA, this privilege breaks down, and immune cells (specifically T cells and macrophages) enter the wall layers to initiate wall remodeling and eventually luminal occlusion. Recent studies have implicated defects in immunoinhibitory signaling as upstream pathologies in GCA. Specifically, tissue-residing and ex vivo-generated DCs from patients with GCA have a defect in upregulating the immunoinhibitory ligand PD-L1, thus failing to counterbalance stimulatory signals with inhibitory signals. Thus, T cells in GCA patients are unleashed and able to infiltrate into the vessel wall, recruit macrophages, and form granulomatous assemblies. Molecular mechanisms underlying the defective induction of PD-L1 remain to be clarified, but a close correlation between inflammatory markers and DC PD-L1 expression levels in individual patients emphasizes the immediate impact that this molecular defect may have (62). Indeed, the accumulation of PD-1⁺ T cells within the vascular lesions appeared to have direct consequences for the quality and intensity of the vasculitic response. The more PD-1⁺ T cells were admitted to the vessel wall, the more T cell cytokines were produced, the more microvessels were generated, and the thicker the intimal layer grew (Table 1). These findings support a novel pathogenic model for GCA, assigning ultimate control over wall remodeling processes to T cell activity. Equally transformative is the recognition that antigen-nonspecific immune regulation determines the outcome of vascular inflammation.

Table 1.

Consequences of an impaired PD-1 checkpoint in vasculitis

Biological Pathway	Clinical Consequence
T cell activation and polarization	Density of the T cell infiltrate
	Interferon-γ, IL-17, and IL-21 production
	IL-7 and IL-15 production
Macrophage activity	IL-1β, IL-6, IL-23, and TNF-α production
Intramural neoangiogenesis	Density of the microvascular lumina
Intimal hyperplasia	Thickness of the intimal layer

As the implication of immune checkpoint regulation in the pathogenic events of GCA broadens the mechanistic understanding of the disease, it also raises several questions that will be instrumental to guide further studies. The PD-L1^low phenotype of GCA DCs appears to be disease specific, not affecting DCs generated from patients with RA. Both disease states are considered autoimmune and inflammatory. These observations indicate that an inflammatory milieu, even when present over years, is insufficient to alter PD-L1 expression. Epigenetic studies are under way to assess the epigenetic landscape of the PD-L1 promoter in GCA, RA, and coronary artery disease. Most information about the regulation of PD-L1 expression is derived from investigations of tumor cells. So far, mechanisms driving PD-L1 expression are only superficially understood and appear to be strongly cell type specific (4, 19, 24, 25, 31, 49). Recent data collected in PD-L1-overexpressing tumors have suggested a role for novel structural variants in the 3′-untranslated region of the PD-L1 promoter to determine surface expression (19). Both IFN-γ and LPS are used in standard assay systems to upregulate the ligand, and DCs from GCA patients are particularly nonresponsive to IFN-γ (62). These data suggest that IFN-dependent signaling pathways may be primarily responsible for the defect.

An intriguing aspect of identifying immune checkpoint abnormalities in the pathogenesis of large vessel vasculitis is the molecular interlinking of antitumor and antivessel wall immunity. CD4 T cells in vasculitis patients have a low threshold setting for activation and can survive in tissue niches from which they are otherwise excluded. In the absence of PD-L1 signaling, they are free to commit to a broad range of effector functions, including communications with vascular cells, such as endothelial cells and myofibroblasts. In tumor patients, cytotoxic T cells are banned from the tumor microenvironment but regain access once the PD-1 checkpoint is inhibited (Fig. 5) (35, 37, 64). Immune-mediated side effects of checkpoint inhibitor therapy have been implicated in inducing vasculitis (10). Physicians and patients need to be alerted to the possible risk of antivascular immunity as part of the growing spectrum of immune-related adverse events in cancer immunotherapy.

Fig. 5.

The PD-1 immune checkpoint in cancer and GCA interactions between PD-L1 and PD-1 provides negative signals to T cells, effectively suppressing T cell effector functions (see Fig. 2). Tumor cells escape from immune recognition by expressing the immunoinhibitory ligand PD-L1. T cells attempting to kill cancer cells are paralyzed by receiving negative signals through their PD-1 receptor. Recent breakthroughs in cancer immunotherapy capitalize on targeting PD-L1/PD-1 interactions. Checkpoint inhibitors prevent PD-L1/PD-1 interactions and have led to remarkable success in enhancing tumor immunity. The unleashing of T cell immunity in checkpoint-treated patients is associated with immune-related toxicity, e.g., the induction of autoimmune disease. Drug-induced vasculitis has been reported. In GCA, checkpoint signaling is impaired due to low PD-L1 expression on vessel wall DCs. The loss of this “immune break” enables PD-1⁺ T cells to infiltrate into the vessel wall and coordinate a wall-damaging inflammatory response. PD-1⁺ T cells have been implicated in regulating lesional cytokine production, driving microvascular neoangiogenesis, and enhancing lumen-occluding intimal hyperplasia. Given the deficiency of negative immune signaling implicated in cancer-immune evasion, GCA patients may have very effective antitumor immune responses.

GRANTS

Support for this work was provided by National Institutes of Health (NIH) Grants R01-AR-042527, R01-HL-117913, P01-HL-129941, and R01-AI-108906 and the Govenar Discovery Fund (to C. M. Weyand) as well as by NIH Grants R01-AI-108891, U19-AI-057266, and R01-AG-045779 and Biomedical Laboratory Research & Development Grant I01-BX-001669 (to J. J. Goronzy).

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

G.B., J.J.G., and C.M.W. conceived and designed research; R.W. and H.Z. performed experiments; R.W., H.Z., and J.J.G. analyzed data; R.W., H.Z., J.J.G., and C.M.W. interpreted results of experiments; R.W., H.Z., J.J.G., and C.M.W. prepared figures; J.J.G. and C.M.W. drafted manuscript; R.W., G.B., J.J.G., and C.M.W. edited and revised manuscript; J.J.G. and C.M.W. approved final version of manuscript.