Inhibitory receptor agonists: The future of autoimmune disease therapeutics?

Abstract

Central and peripheral tolerance both contribute to protection against autoimmunity. The pathogenesis of autoimmunity, however, can result from critical deficits or limitations in peripheral and/or central tolerance mechanisms, presenting an opportunity for therapeutic intervention. Recent advances highlight the substantial impact of inhibitory receptors (IRs), which mediate peripheral tolerance, in autoimmunity. Deletion and blockade studies in mice, IR disruption in humans, and correlation with positive disease outcomes all highlight potential clinical benefits of enhancing IR signaling (agonism) - specifically CTLA4, PD1, LAG3, TIM3 and TIGIT - to treat autoimmune disease. Although critical questions remain, IR agonists represent an unappreciated and untapped opportunity for the treatment of autoimmune and inflammatory diseases.

Introduction

Multiple mechanisms of self-tolerance protect against autoimmunity, including central and peripheral tolerance. While central tolerance is critical to delete cells expressing autoreactive B and T cell receptors, it is often imperfect [1]. Peripheral tolerance mechanisms, including ignorance, anergy or apoptosis, and regulatory T cells (Tregs), are therefore necessary to restrain autoreactivity [1]. Recent reports estimate up to ~30% CD4⁺Foxp3⁻ cells in healthy mice are able to respond to ‘self’ but are restrained by Tregs, and posits a compensatory role for inhibitory receptor (IR) expression in their model lacking Tregs, implying that >30% of CD4⁺ effectors could be autoreactive [2]. Critical deficits or limitations in tolerance mechanisms, rather than complete failure, can contribute to autoimmunity (Fig. 1). Evidence suggests that the majority of these deficits affect peripheral tolerance, presenting an opportunity for therapeutic intervention [1].

Figure 1 legend:

The role of IRs under normal conditions is to limit TCR signaling and terminate an immune response post-infection or to self-antigen. In autoimmune conditions, IRs are present and limit autoimmunity, but are ultimately insufficient to completely prevent autoimmunity, opening up an opportunity for therapeutic intervention. If IR expression or signaling and subsequent downstream effects is enforced therapeutically, autoimmunity may be prevented or managed more easily by patients.

Recent advances in cancer immunotherapy indicate that IR modulation affects T cell function and disease outcome, which merited the 2018 Nobel Prize in Medicine [3]. IRs mediate peripheral tolerance by antagonizing T cell receptor (TCR) signaling and/or signal propagation in autoreactive T cells, resulting in a dysfunctional state [4]. Evolutionarily, IRs are upregulated in response TCR signaling to dampen the immune response post infection or upon recognition of self-antigen (Fig. 1). Under chronic conditions, IR expression is heightened and T cell effector function is inhibited [5]. Furthermore, evidence suggests that IRs are often upregulated as a module (ie multiple IRs at the same time), implying cooperative and/or synergistic activity [6]. IRs have multiple ligands and different signaling mechanisms (reviewed in [7–9]), but overall result in dampening immune activation. An increasing appreciation for the role of IRs in autoimmunity suggests that IR agonism (increasing inhibitory signaling and its downstream consequences) may help prevent and manage autoimmune disease (Table 1–3). Such evidence is exemplified by deletion or blockade studies in mice, outcomes of IR disruption in humans, and positive correlation of IR expression with favorable disease outcomes. This review briefly summarizes the current state of the field in understanding the role of IRs in autoimmunity, with a focus on the IRs currently in clinical research: cytotoxic T-lymphocyte-associated protein 4 (CTLA4), Programed cell death protein 1 (PD1), Lymphocyte Activating Gene 3 (LAG3), T cell immunoglobulin and mucin domain-containing protein 3 (TIM3), and T cell immunoreceptor with Ig and ITIM domains (TIGIT).

Table 1:

IR agonists showing therapeutic efficacy in mouse models of autoimmunity. MHC – Major histocompatibility complex, MOG - Myelin Oligodendrocyte Glycoprotein, Trail - TNF-related apoptosis-inducing ligand, CD253, ICOS - Inducible T-cell costimulatory, Ig – immunoglobin, Fc – refers to the constant region of an antibody, RA – Rheumatoid Arthritis, DSS – Dextran Sodium Sulfate, CIA – Collagen induced Arthritis

Enforcing IR engagement in autoimmune mouse models in vivo
IR targeted:	Method of reinforcement:	Disease setting:	Ref:
PD1	Dendritic cells transduced to expressed MHC carrying MOG, PDL1, and Trail.	improved EAE outcomes	[53]
	Adenovirus expressing PDL1 + ICOS inhibitor	suppresses lupus symptoms in disease prone BXSB mice	[54]
	Adenovirus expressing PDL1-Fc	ameliorates DSS Colitis	[55]
	PDL1-Ig Fusion recombinant proteins	ameliorates T cell mediated Colitis	[55]
	PDL1 -Fc fusion protein	PDL1-fc prevented activation in synovial fluid mononuclear cells of RA patients and ameliorated CIA	[56]
	PDL1-Ig fusion protein	Ameliorated CIA	[57]
	PDL1-Ig fusion protein + anti-CD154	Improves Islet transplantation tolerance	[58]
CTLA4	Transgenic mouse expressing membrane bound agonist anti-CTLA4 on B cells	protects NOD mice from autoimmune diabetes onset	[59]
TIM3	Galectin-9 (ligand for TIM3) was administered as a soluble protein	ameliorates CIA	[60]
	Interferon beta and Galectin-9 fusion proteins	ameliorates EAE	[61]
TIGIT	agonist anti-TIGIT antibody	Improves disease outcomes in EAE	[62]

Table 3:

Clinical trials/approvals aimed at reinforcing IRs to treat autoimmunity. GVHD – Graph versus host disease

Current trials or approvals with IR modulators in patients with autoimmune disease
Therapeutic targets:	Method of reinforcement:	Diseases being studied:	Ref:
PD1	PD1 agonist antibody ANB030, phase 1 trials predicted to begin soon	Human alopecia PBMC, humanized model of GVHD	AnaptysBioANB030
	PD1 agonist antibody in phase 1 at Eli Lily: Patent US2019/0270818 A1	Under investigation for autoimmune diseases such as RA and transplant rejection	[67]
			https://www.lilly.com/discovery/clinicaldevelopmentpipeline#/
	CC-90006 Celgene agonist antibody: NCT03337022	Investigated for psoriatic arthritis	Clinicaltrials.gov
Blocking CD80/86 co-stimulation	Abatacept: CTLA4-ig, blocks CD28 co-stimulation	Summary of selected studies in autoimmune disease settings:
		RA (approved)	[68]
		SLE (so far failed to meet clinical trial outcomes)	[69,
		Sojourns Syndrome (promising results in phase 3) NCT02067910	[70]
		Autoimmune Hepatitis (recruiting for phase 1) NCT04203875	Clinicaltrials.gov
		Alopecia Areata (phase 2, positive outcomes) NCT02018042	Clinicaltrials.gov
		Prevention of T1D (Phase 2, delayed, but did not prevent T1 D) NCT01773707	[71]
		Systemic Sclerosis (Phase 2, well tolerated, not statistically significant outcomes) NCT02161406	[72]
		Relapsing Remitting MS (Phase 2, well tolerated, not statistically significant outcomes) NCT01116427	[73]
		Ulcerative Colitis/Crohn’s disease (Failed to show efficacy in Phase 3) NCT00410410	[74]
CD80/86 and B cells	Abatacept +Rituxumab: CTLA4-ig blocking CD2P co-stimulation + B cell depletion	T1D (Recruiting for phase 2) NCT03929601	Clinicaltrials.gov
CD80/86 and inflammatory cytokines IL-12/23	Abatacept + Ustekinumab: anti- IL-12/23 + CTLA4-ig blocking CD28 co-stimulation	Psoriasis Vulgaris (Phase 2, some clinical benefit) CT01999868	Clinicaltrials.gov

Genetic deletion and antibody blockade studies in mice

Some of the most compelling evidence for the role of IRs in autoimmunity comes from genetic deletion or antibody blockade studies. Under non-autoimmune prone conditions, genetic ablation of either CTLA4 or PD1 results in development of spontaneous autoimmune symptoms, albeit with slightly different manifestations. Mice deficient in CTLA4 rapidly develop severe lymphoproliferative disease resulting in death by 3–4 weeks of age, whereas PD1-deficient mice develop lupus-like symptoms over time, implying non-redundant functions in maintenance of immune tolerance [10–12]. These data illustrate that simply eliminating PD1 or CTLA4 is sufficient to disrupt baseline immune homeostasis and maintenance of peripheral tolerance. Alternatively, while genetic ablation of either LAG3, TIGIT or TIM3 in mice does not precipitate spontaneous autoimmunity, LAG3/PD1 double-knockout (KO) mice experience multiple organ autoimmunity that manifests more aggressively than PD1 KO alone, suggesting a synergistic effect in limiting autoimmunity [13,14]. This difference in disease progression and manifestation highlights the complexity of targeting each IR in a disease setting, the hierarchy of importance in limiting autoreactivity beginning with CTLA4 and PD1, and possible synergistic effects of ‘tertiary’ IRs such as LAG3, TIGIT and TIM3.

Regardless of their role in baseline immune homeostasis or spontaneous autoimmunity, blocking or genetically deleting IRs exacerbates disease in autoimmune-prone backgrounds or induced disease models (reviewed in [15]). For example, in the non-obese diabetic (NOD) mouse model of autoimmune diabetes, a spontaneous model in which 80% of female mice develop diabetes between 10–30 weeks of age, treatment with anti-PD1 or anti-LAG3 induces autoimmune diabetes with 100% penetrance within only a few weeks post administration, regardless of timing [16]. Similar results are obtained in global PD1, LAG3 or CTLA4 KOs on a NOD background [16–20]. Conceptually, these results are mirrored on the MRL-lpr background, which is susceptible to spontaneous lupus-like autoimmunity. On the MRL-lpr background, PD1, CTLA4 and B- and T-lymphocyte attenuator (BTLA) have been shown to limit disease progression in studies with KO mice and/or antibody blockade [15,21].

Additionally, IR blockade or deletion worsens autoimmunity in inducible settings. Studies using the experimental autoimmune encephalomyelitis model (EAE; a mouse model of multiple sclerosis), demonstrated that treatment with anti-PD1 or genetic deletion of PD1, TIM3, and TIGIT exacerbated disease symptoms [22–24]. Similarly, IRs may control environmentally-induced autoimmunity, as studies with LAG3 KO mice highlight a role for this IR in limiting mercury-induced autoimmune dysfunction [25].

To add further complexity, IRs play different roles depending on the cell type. For example, Tregs can use IRs such as LAG3 and CTLA4 as mechanisms of suppression [26,27]. Indeed, non-autoimmune prone C57Bl/6 mice with Treg-restricted deletion of CTLA4 develop extensive lymphoproliferative disease, similar to Ctla4 global KO or Foxp3^−/− mice, implying Treg suppressive capacity is severely impaired in the absence of CTLA4 and that Tregs may be driving the phenotype behind the global CTLA4 KO [28]. Conversely, however, Treg-restricted deletion of LAG3 improves disease outcomes in autoimmune diabetes, as LAG3 was shown to limit Treg proliferation in this disease setting [29].

Finally, the timing of IR blockade administration or induction of IR deletion can give rise to differing disease outcomes, highlighting another layer to consider when targeting IRs in an autoimmune setting. For example, anti-CTLA4 treatment of NOD mice induces autoimmune diabetes with 100% penetrance only when administered prior to insulitis onset [18,19]. Similarly, Treg-restricted temporal deletion of Ctla4 in C57Bl/6 mice in adulthood does not cause the hallmark systemic autoimmunity of global or Treg specific CTLA4 KO [30]. Both observations suggest that the suppressive capacity of CTLA4 may be more important during the T cell priming phase of disease [18,19,31]. Unexpectedly, although quite similar to Treg-restricted LAG3 deletion in the NOD model, adulthood Treg-restricted CTLA4 deletion confers protection from EAE, which is attributed to increased Treg numbers and an upregulation of compensatory IRs on CD4⁺ effector T cells [30]. The aforementioned dichotomous results imply a need to further understand the role of these IRs temporally, and in a cell type and disease specific manner prior to targeting them therapeutically.

IR disruption in humans

While animal studies are useful to gain a mechanistic understanding of IRs, there is good evidence that IR disruption also contributes to autoimmunity in humans. Notably, many large-scale genome-wide association studies (GWAS) have linked hundreds of SNPs to specific autoimmune diseases, many of which are immune related and are within genes encoding IRs [1]. This implies that IR function is critical to maintain self-tolerance, although these studies only inform correlation and not causation. SNPs have varying levels of association in autoimmune-prone individuals compared to the general population. Databases compiling GWAS studies in conjunction with other published data can now score genes for association with a particular disease [32] (Open Target Platforms URL: Targetvalidation.org). Interestingly, there is an almost hierarchical level of association of IRs with autoimmune disease. CTLA4 is very strongly implicated in almost every autoimmune disease, and thus could be considered a primary IR. The strongest associations linked to CTLA4 SNPs are type 1 diabetes (T1D), Graves’ disease, and systemic lupus erythematosus (SLE) [33]. PDCD1 (gene for PD1) is associated with many diseases but not all, specifically SLE [34], and thus could be considered a secondary IR. Tertiary IRs, such as TIGIT, LAG3 and HAVCR2 (TIM3), have lower association scores but are still linked to certain autoimmune diseases [9].

The use of IR blockade as a cancer therapeutic provides a unique setting to study the role of IRs in humans. Many patients with no prior history of autoimmunity who receive treatment with anti-CTLA4, anti-PD1, or anti-PDL1 experience autoimmune side-effects; as many as 85%, 37%, and 24% of patients, respectively [31]. The majority of IR blockade-induced autoimmunity results in organ specific immune infiltration and damage, and often occurs in a location distal to that of the primary malignancy (reviewed in [31,35]). These autoimmune side effects are difficult to predict and interpret, but they posit the idea that individuals may be poised for or experiencing subclinical autoreactivity, which is restrained by IRs, and only by blocking IR function is clinical autoimmunity revealed.

Signatures of IR expression are predictive of prognosis in human autoimmunity

Signatures of IR expression and inhibition are actively being implicated in autoimmune disease progression and outcomes. IRs are particularly upregulated in chronic diseases, and often result in a state termed “exhaustion” in which T cells lose effector functions and experience genome wide transcriptional and epigenetic changes [5]. Most autoimmune diseases are chronic and thus T cells from these patients show an IR-rich phenotype. The dichotomy of expressing IRs while still remaining pathogenic is somewhat of an enigma in autoimmunity; yet, this phenomenon highlights the fact that normal regulatory processes are present but insufficient to prevent autoimmunity (Fig. 1). This provides a potential window for therapeutic intervention, possibly even supporting the idea of therapeutically induced exhaustion to treat autoimmunity [36].

Given the recent advances in systems immunology and in tracking patients over time, we can now better investigate the relationship between IR expression and disease outcomes in humans. In 2015, the first implication of IR induced T cell exhaustion was linked to better prognosis in a long-term study of patients with antibody-associated vasculitis (AAV) [37]. Indeed, gene set enrichment for an exhaustion signature predicted better flare-free survival in this AAV cohort, as well as in inflammatory bowel disease (IBD) and SLE cohorts [37]. Patients were followed for up to 2,000 days post initial analysis, providing valuable information on flare-free survival as opposed to disease activity at time of blood draw, for which IR signature is not predictive [37]. Similarly, recently diagnosed T1D patients with an enriched exhausted phenotype on CD8⁺ T cells are reported to have slower disease progression and longer maintenance of c-peptide levels, a biomarker of insulin secretion [38]. Both of these studies identified an “exhaustion” signature based on gene or protein expression from peripheral blood, and relied mainly on IR expression, though it should be noted that this is just one aspect of an exhausted T cell. A current research limitation is obtaining adequate numbers of IR-expressing cells from human peripheral blood for analysis that would prove that these cells are functionally exhausted or for epigenetic analysis. Furthermore, an even more distinct exhausted signature may be evident in the target organ of an autoimmune disease, which presents another limitation of human studies.

Finally, there are novel immunotherapeutic strategies that show efficacy by indirectly modulating IR expression. T1D studies make an interesting case for targeting IR expression and function therapeutically, as many novel therapeutics directly target T cell function. Novel immunotherapies for T1D are limited in that they must be more safe and efficacious than daily insulin injections. Therefore, novel T1D therapies must specifically target T cell function and several therapeutics in trial modulate IRs to meet this goal.

The first example is teplizumab, an anti-CD3 blocking antibody, which is in clinical trial to prevent T1D in at-risk individuals (NCT03875729) [39]. Antagonizing CD3 blocks TCR signal propagation and subsequent activation. In both mice and humans, anti-CD3 induces remission and delays the onset of diabetes [39]. The mechanism of action of anti-CD3 therapy has been attributed not only to preserving Tregs, but also to inducing a “partially exhausted” or “potentially anergic” state in effector CD4⁺ and CD8⁺ T cells [40,41]. In fact, patients treated with anti-CD3 in trial showed multiple IR modulating effects, including transiently increased PD1⁺ Tregs, increased potentially anergic CD4⁺ effectors (CD57⁻KLRG1⁻PD1⁺) and increased partially exhausted CD8⁺ T cell populations (CD57⁻KLRG1⁺PD1⁺ and TIGIT⁺KLRG1⁺), characterized by increased exhaustion related transcription factor, EOMES, and downregulation of memory marker CD127 [40]. Furthermore, those TIGIT⁺KLRG1⁺ CD8⁺ T cells from responding patients showed transcriptional enrichment for markers of exhaustion including EOMES and IRs Tigit, Lag3, Cd160, and Tim3 [41]. Although anti-CD3 does not directly target IRs, anti-CD3 modulates IR expression and thus T cell function for therapeutic benefit, revealing the potential untapped power that reinforcing IRs may have on disease outcome. In summary, signatures of IR expression, either naturally or therapeutically induced, have been correlated to positive disease outcomes in T1D, AAV, SLE, and IBD in humans.

Similar to anti-CD3, though yet to be translated to humans, nanoparticles containing an insulin–ChgA hybrid peptide show therapeutic induction of anergy in the NOD model [42]. Nanoparticles can induce an antigen-specific response by delivering disease-specific antigens to Tregs. Interestingly, insulin–ChgA hybrid peptides not only increase the Treg:CD4⁺Foxp3⁻ T effector (Teff) ratio, but also alter the Teff phenotype. RNAseq shows this modulated phenotype is characterized by high expression of IRs Lag3 and Pdcd1, as well as several markers of anergy, thus contributing to disease protection [42].

Finally, signatures of IR upregulation and T cell dysfunction have been shown in other autoimmune settings. For example, features of T cell exhaustion, including IR upregulation, have also been observed in the kidney of MRL-lpr mice with nephritis [43]. Similarly, anergic CD4⁺ T cells can differentiate into Tregs and prevent autoimmune arthritis and IBD in an adoptive transfer settings, in which IR upregulation is observed [44]. However, an expanded PD1⁺CD8⁺ metabolically active population was recently found in IBD, juvenile idiopathic arthritis and atopic dermatitis, suggesting that single IR expression may simply mark an activated state, as opposed to a truly dysfunctional state [45]. Identifying these IR-expressing populations is the first step to determine clinical impact and provides evidence to warrant deeper investigation into their function in autoimmunity. This poses the question - can IR agonism, in monotherapy or in combination with other therapeutics, delay or prevent the onset of autoimmunity and/or limit symptoms?

Therapeutically targeting IRs

The success of cancer immunotherapy highlighted IRs as a meaningful and significant target to impact disease outcome. In autoimmunity, novel therapies that induce IR upregulation prevent or delay disease. However, therapeutics that directly agonize or enhance IR signaling are very limited and are, at most, in early phases of clinical trial (reviewed in [46]) (Fig. 1, Table 3). However, successful mouse studies and in vitro human studies have highlighted the potential therapeutic benefits that IR agonists may have in the treatment of autoimmunity (Table 1 and Table 2).

Table 2:

IR reinforcement studies that will be translated to treat autoimmune patients. FGL1 – Fibrinogen-like protein 1, PBMC – Peripheral blood mononuclear cells, MS – multiple sclerosis.

Observations that may be translated to autoimmune disease
IR targeted:	Method of reinforcement:	Setting studied:	Ref:
LAG3	FGL1 -Ig suppresses T cell proliferation in vitro	Treatment with FGL1-ig prevents Lag3+ T cell activation	[63]
	Crosslinking LAG3 with CD3 prevented T cell activation	Cross-linking prevents activation of human PBMC	[64]
	LAG3 agonist antibody IMP761	prevents activation of human PBMC, and delayed type hypersensitivity in cynomolgus macaque	[65]
TIGIT	Agonist anti-TIGIT	decreases proliferation and cytokine production in vitro from CD4+ T cells collected from MS patients	[66]

Despite some success, key questions remain that will inform the development of IR-agonist therapeutics. (1) Which cell types are targeted by IR agonism, and what is the impact on disease outcomes? While such therapeutics aim to target self-reactive CD4⁺ and CD8⁺ effector T cells, inevitably Tregs will also be influenced by IR agonists, the results of which are largely unknown. Additionally, agonizing certain IRs over others may preferentially affect effector T cell activation over Tregs, for example anti-CTLA4, as evidenced by temporal Treg-specific deletion studies [30].

(2) At what timepoint do IR agonists best influence disease outcome? This question is complicated because the answer may depend on the disease state as well as the IR in question. For example, CTLA4 functions in a time-restricted window in T1D; this window may be shifted or protracted in other immune settings. In many autoimmune cases, we may need to administer immunotherapeutic treatment prior to disease onset to reveal the true potential in disease prevention, which raises further questions regarding identification of patients at high risk for disease [39]. We must begin to assess timepoints for efficacy as well as how best to incorporate IR agonism into current treatments for autoimmunity, for example, as a monotherapy or in combination with other therapeutics, either dual IR-agonism or standard of care treatments.

(3) Are there potentially detrimental off-target effects of IR agonism through general immune suppression? IR blockade-induced autoimmunity occurs in a significant proportion of cancer patients [31]. One can envision opposing effects in autoimmune patients treated with IR agonists, such as poor response to vaccinations, impaired viral clearance, or even increased risk of cancer. Thus, the goal must be to minimize these possible adverse events for clinical treatment.

(4) What is the best way to target IRs in an agonistic manner (Fig. 1, Tables 1–3)? Agonist antibodies can be challenging to generate, but potent examples exist, such as the CD28 agonist TGN1412, which caused detrimental cytokine storms in phase 1 trials [47]. Further, there is great potential and value in producing novel IR agonists (Tables 1–3), some of which are showing promising therapeutic efficacy. Can similarly potent agonistic effects be achieved using small-molecule agonists or ligand (e.g., PDL1-Ig fusion to induce PD1 signaling)? Can IR fusion proteins (e.g., CTLA4-Ig, abatacept, Table 3) be used to adequately block co-stimulation? Can we achieve similar results by simply increasing expression of IRs? These issues are complicated further by an incomplete understanding of IR transcriptional regulation, ligands, and signaling pathways [15]. Furthermore, certain ligands, for example CD80/86 or CD155, may be shared between IRs and co-stimulatory molecules, which complicates the use of ligands to enhance IR signaling. Lastly, in choosing an approach to enhance IR signaling and function, one must consider that many IRs can also be produced and/or shed as soluble receptors in serum, such as LAG3, TIM3, CTLA4 and PD1, and may regulate T cell activity and/or interfere with any therapeutic approach [48–51]. Of note, a soluble CTLA4 isoform has been found to be elevated in many autoimmune settings and influences T cell proliferation and cytokine production [50]. In contrast, soluble LAG3, which is generated by ADAM metalloprotease-mediated cell surface shedding, does not appear to impact T cell function [51,52]. Thus care must be taken when using IR enforcement strategies to evaluate any potential impact of soluble IRs on efficacy and/or dosing. Levels of soluble IRs may vary between patients, which also may complicate evaluating the appropriate therapeutic window for IR agonists, and may require analysis on a per-patient basis. In summary, we need a better understanding of whether there are specific IRs expressed in unique disease settings and/or on unique cell types, the temporal usage of IRs, IR signaling and function, and patient prognosis information.

The role of IRs in autoimmunity holds promise for the development of novel therapeutic approaches to treat autoimmunity. There is strong rationale and supportive evidence for targeting IRs in autoimmunity through deletion or blockade studies, information gathered from human IR disruptions, and attempts to identify IR correlations with disease progression and outcomes. Although many questions remain, the future of immunotherapy for autoimmune disease may involve the use of IR agonists.