Catch and release: freeing up PD-L1 ameliorates autoimmunity

Abstract

The binding of PD-L1 to CD80 on antigen-presenting cells prevents PD-1 ligation on T cells. Therapeutic blockade of the cis-PD-L1–CD80 interaction liberates PD-L1 to bind to PD-1, inhibits autoreactive T cells and robustly alleviates autoimmune symptoms.

Autoimmune diseases, caused by a break in immune tolerance to self and subsequent immune-mediated destruction of targeted tissues, are estimated to affect 5–8% of the world population¹. Cancer immunotherapies that block the ‘checkpoint’ receptor–ligand pair PD-1– PD-L1 have been remarkably efficacious in the clinic and provide clear evidence that targeting inhibitory receptor pathways can influence disease outcome². Although efforts to promote the PD-1–PD-L1 pathway in autoimmunity to limit disease have drawn much interest, approved agonistic therapies are lacking in the clinic³. In the current issue of Nature Immunology, Sugiura et al.⁴ describe their development of mouse and human antibodies that elicit PD-1–PD-L1 signaling to ameliorate autoimmunity.

PD-1 ligation (on T cells) by PD-L1 (on antigen-presenting cells (APCs) and other cell types) directly antagonizes T cell receptor (TCR) and co-stimulatory receptor (CD28) signaling by recruiting protein tyrosine phosphatases, such as SHP2, to dephosphorylate downstream kinases⁵. The PD-1 and PD-L1 axis is crucial for the maintenance of immune homeostasis^3,6,7. Some of the most compelling evidence for the importance of the PD-1–PD-L1 axis in peripheral tolerance is highlighted by the fact that up to 37% and 24% of patients treated with anti-PD-1 and anti-PD-L1 antibodies, respectively, experience immune-related adverse events of any grade, with 7% and 4%, respectively, experiencing grade 3 or 4 adverse events^7,8.

The CD28 (on T cells) and CD80 or CD86 (on APCs) axis represents one of the most important co-stimulatory signals required for the propagation of TCR signals and activation of T cells⁹. However, this signal is antagonized by CTLA4 binding to CD80, which has a higher affinity interaction than CD28 and CD80 (Fig. 1)¹⁰. Several hours after TCR activation, PD-1 is upregulated; however, PD-1⁺ T cells initially exhibit effector function and are not functionally restricted. Previous studies have attributed this to a duplex that is formed between CD80 and PD-L1 on the same APC (in cis)^11,12. This interaction temporarily sequesters PD-L1 and prevents binding to PD-1, thus enabling T cell activation upon initial antigen presentation. Key immune regulatory axes (PD-1–PD-L1 and CTLA4–CD80) inhibit transduction of TCRs and co-stimulatory receptors, thereby controlling T cell activation and function. These immune regulatory axes are crucial pathways for the maintenance of immune tolerance to self and prevention of autoimmunity¹³. By contrast, the presence of this cis-CD80–PD-L1 duplex inhibits both the PD-L1–PD-1 and the CTLA4–CD80 axes, thereby adding another layer of immune fine-tuning and enabling initial TCR signal propagation^11,12.

Fig. 1 |. The cis-PD-L1–CD80 duplex can be disrupted with an anti-CD80 antibody to elicit PD-1 function and ameliorate autoimmunity, and acts synergistically in the presence of CTLA4-Ig.

a, Under autoimmune conditions, autoreactive CD4⁺ and CD8⁺ T cells recognize self-antigen presented by dendritic cells (DCs) or other antigen-presenting cells (APCs). These autoreactive T cells receive positive co-stimulation signals through the CD80–CD28 pathway, thus initiating downstream signal transduction and T cell activation. Mechanisms of peripheral tolerance, such as the inhibitory pathways PD-1–PD-L1 or CTLA4, are present but are insufficient to limit signal transduction. One explanation for this constraint is the formation of a cis-PD-L1–CD80 duplex on dendritic cells, which limits PD-L1 availability to bind to PD-1 and elicit an inhibitory signal. In this scenario, positive co-stimulation successfully activates the T cell. b, Sugiura et al.⁴ created an anti-mouse CD80 antibody (TKMG48) that limits the formation of cis-PD-L1–CD80 duplexes on APCs. TKMG48 binds to CD80 at the PD-L1 interacting site, thus inhibiting duplex formation. Inhibition of the duplex relieves PD-L1 to interact with PD-1 on T cells, thus triggering inhibitory signals to counteract TCR and CD28 signaling and T cell activation. c, TKMG48 can be combined with CTLA4-Ig therapy to elicit synergistic effects, thus agonizing two independent inhibitory pathways to strengthen the inhibitory signals present to counteract TCR and CD28 signaling. Collectively, these methods may be efficacious in restricting autoimmune reactions.

The goal of the study by Sugiura et al.⁴ was to develop a therapeutic approach to disrupt the PD-L1–CD80 duplex in the hopes of promoting PD-L1 binding to PD-1, thus enhancing the inhibition of T cell function and alleviating autoimmune disease. Dissociation of the PD-L1–CD80 duplex was detected using the soluble form of the extracellular region of PD-1 (PD-1-EC) which binds to PD-L1 and serves as a mimetic for PD-1–PD-L1 ligation. As expected, PD-1-EC bound to T cell lymphoma (IIAdL) cells that expressed PD-L1, and PD-L1-knockout dendritic cells were unable to bind PD-1-EC. PD-L2-knockout dendritic cells still bound PD-1-EC, which suggests that PD-L1 is the dominant ligand for PD-1. T cell lymphoma cells that express PD-L1 and CD80 in cis were used to screen anti-CD80 antibodies that blocked PD-L1–CD80 binding and enabled PD-1-EC binding to PD-L1. More than 3,000 hybridoma-derived monoclonal antibodies were subjected to this flow cytometric screening. One mouse IgG1 clone, TKMG48, facilitated PD-1-EC binding to PD-L1 in the presence of CD80 in a dose-dependent manner. Notably, TKMG48 did not disrupt the binding of CD80 to CTLA4 and only weakly decreased CD80 binding to CD28, which supports previous reports of unique CD28–CTLA4 (AGFCC’C’’ face) and PD-L1 (DEB face) binding sites on CD80¹⁴. This was confirmed with a competitive binding assay in which TKMG48 did not disrupt the binding of the anti-CD80 antibody clone 16–10A1, which is known to recognize the AGFCC’C’’ face of CD80¹⁵.

To assess the effect of TKMG48 on T cell function, DO11.10 T cell hybridoma cells were stimulated with ovalbumin peptide-pulsed B lymphoma (IIAdL) cells and the production of IL-2 was used as a readout. As predicted, TKMG48 decreased IL-2 production in a dose-dependent manner, but only when the T cell hybridomas expressed PD-1. Importantly, cells that expressed signaling-deficient mutants of PD-1 — PD-1(Y248F) and PD-1(Y225F) — were resistant to TKMG48.

The production of IL-2 and IFN-γ by primary T cells from DO11.10 transgenic mice was also decreased by TKMG48 when APCs expressing CD80, CD86 and PD-L1 were used; however, this effect was lost when PD-L1 was absent. IL-2 production by primary CD4⁺ and CD8⁺ T cells was also decreased by TKMG48 when primary bone marrow-derived dendritic cells, or CD8α⁺ or CD11b⁺ dendritic cells were used as APCs. These data demonstrate that disruption of the CD80–PD-L1 duplex with TKMG48 promotes PD-L1–PD-1 ligation, resulting in inhibition of T cell effector function.

Given that TKMG48 binds to CD80, it was important for Sugiura et al.⁴ to demonstrate that CD80–CD28 binding was not disrupted. TKMG48 weakly decreased T cell production of IL-2 when APCs (IIAdL) expressed CD80, but not CD86. However, when APCs expressed both CD80 and CD86, TKMG48 did not decrease IL-2 production, which suggests that CD86 provides sufficient ligation to facilitate CD28-mediated co-stimulation.

To ensure that TKMG48 did not elicit intrinsic effects on APCs, the authors assessed the effects of TKMG48 in vivo. Using a vaccination model (OVA peptide and poly(I:C)), TKMG48 did not affect the ability of dendritic cells to differentiate, migrate or present antigen or the expression levels of PD-L1, PD-L2, CD80 or CD86 on migratory dendritic cells. However, during vaccination, TKMG48 reduced the number of antigen-specific CD8⁺ T cells and the production of IFNγ and IL-2, which suggests that the PD-1 inhibitory pathway may have been elicited.

These promising in vivo observations served as a premise to determine the effect of TKMG48 on T cell function in several mouse experimental models of autoimmunity, including diseases mediated by T_H17 and T_H1 (such as arthritis, spondyloarthritis, multiple sclerosis and Sjögren’s syndrome). Clinical scores, gross swelling and immune infiltration in mice immunized with human glucose6-phosphate isomerase peptide to induce arthritis were markedly reduced when TKMG48 treatment was administered in both a prophylactic (one day after arthritis induction) and, importantly, a therapeutic (after overt symptoms were present) setting. The production of IFN-γ and IL-17A by autoreactive T cells was also reduced by TKMG48 in arthritic mice. As expected, blockade with anti-PD-L1 or anti-PD-1 antibodies prevented the therapeutic effect of TKMG48.

Furthermore, using a mouse model of spondyloarthritis (zymosan-induced SKG mice), which represents a severe and chronic form of arthritis, treatment with TKMG48 during the early stages of disease (day 5) markedly reduced clinical and histological scores, CD4⁺ T cell activation, and IFN-γ and IL-17A production. Notably, TKMG48 efficacy was robust as a curative treatment (when treated during the chronic stage, day 30) of spondyloarthritis. In addition, clinical scores in a mouse model of multiple sclerosis (experimental autoimmune encephalomyelitis) were reduced in a PD-L1-dependent manner with TKMG48. Histological scores of sialadenitis and dacryadenitis in a mouse model of Sjögren’s syndrome (NFS/sld mice) were also reduced with TKMG48 treatment.

Opportunities for the combinatorial therapeutic treatment of autoimmunity were highlighted by the authors in experiments combining TKMG48 with CTLA4-Ig (also known as abatacept), in which a notable decrease in clinical arthritis scores (induced by glucose-6-phosphate isomerase peptide) was observed. This synergistic effect was attributed to their distinct mechanisms of action. Collectively, Sugiura et al.⁴ have clearly demonstrated the therapeutic benefits of disrupting the CD80–PD-L1 duplex with TKMG48 in various inducible mouse models of autoimmune disease.

These promising in vivo studies prompted the authors to screen for a human anti-CD80 antibody with a similar mechanism of action to TKMG48. The human IgG2a clone TKMF5 was extensively validated with similar rigor to the mouse studies and shown to bind to the CD80–PD-L1 interacting domain of CD80. Together, these findings provide a premise to investigate the benefit of anti-CD80 antibody (TKMF5) alone or in combination with other therapies, such as CTLA4-Ig, for the treatment of autoimmune diseases in humans.

In contrast to more ‘conventional’ agonist approaches, which do not require ligand availability, a unique feature and requirement for the efficacy of TKMG48 is PD-1–PD-L1 expression at the site of autoimmune or inflammatory disease. Indeed, PD-L1 is expressed on immune and non-immune cells and induced by IFNγ during inflammation¹⁶. Therefore, patients with certain autoimmune disorders in which PD-L1 is upregulated on target tissues may be better candidates for TKMF5 therapy than others. Finally, future studies will be required to determine whether there are any off-target or compensatory effects of long-term treatment with TKMG48 or TKMF5, as autoimmune disorders are chronic and treatment may be continuous. Would the long-term induction of the PD-1 pathway impair immunosurveillance and affect the ability of patients to clear tumors or viral infections, or limit their ability to mount a sufficient immune response after vaccination? Finally, compensatory mechanisms to inhibition of the cis-CD80–PD-L1 duplex should be considered. For example, to what extent might the release of CD80 promote CD28 signaling, given that TKMG48 only weakly decreased CD80–CD28 interaction?

To conclude, Sugiura et al.⁴ have elegantly demonstrated the therapeutic potential for disrupting the CD80–PD-L1 duplex to liberate PD-L1 and enable PD-L1–PD-1-induced T cell anergy. This method of eliciting PD-1 signal propagation could represent a new and impactful strategy in the treatment of autoimmunity.