Beyond checkpoint inhibition: PD-1 cis-targeting of an IL-2Rβγ-biased interleukin-2 variant as a novel approach to build on checkpoint inhibition

ABSTRACT

The immunocytokine PD1-IL2v was designed to overcome liabilities and improve efficacy of IL-2 therapies. PD1-IL2v preferentially targets PD-1⁺ T-cells and acts on antigen-specific stem-like PD-1⁺ TCF-1⁺ CD8⁺ T-cells expanding and differentiating them towards better effectors resulting in superior efficacy in LCMV chronic infection and tumor models compared to checkpoint inhibition.

KEYWORDS:

• Eciskafusp alfa
• IL-2
• PD-1
• PD1-IL2v
• RG6279
• TCF-1

In our recent publication in Nature, we describe PD1-IL2v (RG6279, eciskafusp alfa), a novel programmed cell death 1 (PDCD1, best known as PD-1) cis-targeted IL2v immunocytokine that preferentially targets antigen-specific stem-like PD-1⁺ TCF-1⁺ CD8⁺ T cells and differentiates them towards a novel population of better effectors for the treatment of cancer and chronic infections¹.

In spite of its approval in 1998 as first cancer immunotherapy, clinical use of Interleukin-2 (IL-2, aldesleukin) has been hampered by severe toxicities and the detrimental expansion of regulatory T cells (Tregs). To overcome these liabilities IL-2 Rβγ biased agonists with retained binding to the signaling IL-2 Rβγ heterodimer, but abolished binding to the IL-2 Rα (CD25) subunit were developed e.g. bempegaldesleukin (NKTR-214)², SAR444245/THOR-707³ or the fibroblast alpha protein (FAP) targeted immunocytokine simlukafusp alfa (FAP-IL2v)⁴. However, while these molecules exhibit improved safety over IL-2, their clinical activity as monotherapy and in combination with PD-1/PD-ligand 1 (PD-L1) checkpoint inhibitors (CPIs) is limited, emphasizing the need to pursue novel strategies to agonize the IL-2R pathway⁵.

Previous work showed that anti-PD-L1 and IL-2 therapies synergize in the LCMV (lymphocytic choriomeningitis virus) chronic infection model⁶. In our accompanying Nature publication by Hashimoto et al., we demonstrated that IL-2 combined with a CPI prevented the differentiation program of stem-like PD-1⁺ TCF-1⁺ CD8⁺ T cells towards exhausted T cells to generate instead a novel population of virus-specific better effector CD8⁺ T cells. In addition, we provide a mechanistic rationale for the modest clinical efficacy of IL-2 Rβγ biased agonists by using either antibodies blocking CD25 or an IL-2 variant devoid of CD25 binding, which prevent the generation of these better effector T cells⁷.

We reasoned early on that delivering IL2v in cis to PD-1, as a proxy marker of antigen-specific T cells, could be a way to improve the potency on PD-1⁺ T cells without the need to bind to CD25, thus preventing CD25-related toxicities and Treg expansion¹. For this purpose, we designed the PD1-IL2v immunocytokine which comprises a bivalent pM-affinity PD-L1/PD-L2 blocking PD-1 antibody and a single IL2v moiety fused to the C-terminus of the heterodimeric human IgG1 Fc-portion bearing P329G LALA mutations to prevent FcγR binding. In vitro assays showed that as a consequence of PD-1 mediated delivery, PD1-IL2v mediated ca. 40-fold higher potency in inducing STAT5 phosphorylation on PD-1⁺ T cells as opposed to a control IL-2 v immunocytokine targeted to FAP as a tumor stroma marker, FAP-IL2v. By using differentially labeled PD-1⁺ and PD-1⁻ T cell populations it was proven that PD1-IL2v acts through cis-targeting on PD-1⁺ T cells. Furthermore, PD1-IL2v induced more potent effector functions e.g. granulocyte-macrophage colony-stimulating factor (GM-CSF) and granzyme B (GrzB) secretion. Importantly, in spite of PD-1 being expressed on Tregs, PD1-IL2v preferentially bound T effector cells and rescued them from Treg suppression. Remarkably, because of simultaneous PD-1 and IL-2Rβγ binding, rapid internalization of the bound PD-1 was observed.

For in vivo studies muPD1-IL2v, a mouse specific surrogate that showed comparable cis-targeting as PD1-IL2v on muPD-1⁺ murine splenocytes, was applied. In the LCMV model of chronic infection muPD1-IL2v improved LCMV-specific CD8⁺ T cell responses leading to viral control while expanding and differentiating stem-like PD-1⁺ TCF-1⁺ CD8⁺ T cells. When combined with anti-muPD-L1 the number of polyfunctional CD8⁺ T cells was further enhanced. Adoptive transfer experiments established that muPD1-IL2v mediates superior efficacy via a proliferative burst and differentiation of stem-like CD8⁺ T cells into better effectors whereas exhausted CD8⁺ T cells did not expand.

In C57/BL6 mice bearing syngeneic Panc02 tumors muPD1-IL2v mediated superior tumor control benefit including complete tumor eradication/long term survival compared to the combination of anti-muPD-1 and muFAP-IL2v. Notably, this was accompanied by a strong intratumoral increase in the CD8⁺/CD4⁺ T cell ratio and in PD-1⁺ GrzB⁺ T cells. A subsequent in-depth pharmacodynamics analysis of tumor infiltrating lymphocytes in the Panc02 model by flow cytometry and single cell RNA sequencing confirmed that, similar to chronic LCMV infection, muPD1-IL2v expanded and differentiated PD-1⁺ TCF-1⁺ stem-like CD8+ T cells into a new better effector CD8⁺ T cell subset. Conversely, anti-muPD-1 alone or combined with muFAP-IL2v resulted in the accumulation of terminally differentiated, exhausted CD8⁺ T cells (Figure 1). In contrast to muPD1-IL2v targeting/acting on PD-1⁺ T cell, tumor stroma targeted muFAP-IL2v mainly acts on PD-1⁻ T cells resulting in accumulation of non-tumor specific CD8⁺ T cells. Likewise, in huPD-1 transgenic C57/BL6 mice, PD1-IL2v mediated superior efficacy compared to the combination of pembrolizumab with aldesleukin in the Panc02 model.

Figure 1. Mechanism of action of PD-1 cis-targeted PD1-IL2v as opposed to the combination of anti-PD-1 with FAP-IL2v. PD1-IL2v acts by differentiating stem-like CD8⁺ T cells into better T effector cells whereas PD-1/PD-L1 checkpoint inhibition combined with IL-2Rβγ biased FAP-IL2v does not expand better effector T cells, but via a transitory stage rather result in the accumulation of terminally differentiated, exhausted T cells. Created with BioRender.Com.

Finally, muPD1-IL2v’s superior anti-tumor efficacy over anti-muPD-1 plus muFAP-IL2v was confirmed in (i) the syngeneic MCA205 model, (ii) the B16-F10-OVA model where muPD1-IL2v resulted in a strong expansion of OVA-specific better effector CD8⁺ T cells, and (iii) in the genetically engineered, autochthonous pancreatic neuroendocrine tumor RipTag5 model which is refractory to CPI. Subsequent work by Tichet et al. confirmed that muPD1-IL2v resulted in infiltration by stem-like CD8⁺ T cells, subsequent tumor regression and enhanced survival, however, for sustained responses combination with anti-muPD-L1 was required to overcome an emerging adaptive resistance mechanism⁸.

Taken together, these preclinical experiments unequivocally demonstrate that cis-targeting of IL-2 variant devoid of CD25 binding to PD-1 results in superior anti-viral and anti-tumor efficacy in models of chronic infection and cancer by differentiating stem-like CD8⁺ T cells into a novel population of better effector T cells as opposed to conventional checkpoint inhibition, and provides the rationale for clinical testing of PD1-IL2v. PD1-IL2v (RG6279, eciskafusp alfa) became the first PD-1 targeted immunocytokine entering clinical trials 2020, and is currently in Phase 1b clinical trials as a monotherapy in CPI relapsed/refractory patients and in combination with atezolizumab. Similar approaches have been reported for PD-1 cis-targeting of low affinity IL-2⁹ or IL-15¹⁰ to tumor-reactive T cells further supporting the mechanistic rationale described for PD1-IL2v. We believe that PD-1 cis-targeted IL-2R agonists represent one of the most promising strategies to further build on the efficacy of PD-1/PD-L1 checkpoint inhibition in patients that have raised an endogenous anti-tumor immune response.

Acknowledgments

The authors want to thank specifically Rafi Ahmed, Emory University School of Medicine and all co-authors and contributors to the publications and research on PD1-IL2v described.

Disclosure statement

LCD, CK and PU declare employment, stock ownership and patents with Roche related to this work.

Additional information

Funding

Part of the work by MH was co-funded by Roche.