Background Therapeutic approaches utilizing IL2 or IL15 have suffered from low tolerability, fast clearance, and limited therapeutic window due to extensive activity on peripheral T and NK cells. Conversely, selective targeting of tumor-reactive lymphocytes (TIL) may improve tolerability and anti-tumor activity while simultaneously improving drug PK. We hypothesized that we could selectively target tumor-reactive TILs by combining a reduced potency IL15/IL15Rα heterocomplex with an immune checkpoint(CP)-targeting arm to bias binding and activation to CP-positive TILs, potentially improving therapeutic index. Lymphocyte activation gene 3 (LAG3) was chosen as the CP targeting-arm due to its frequent co-expression with PD1, bias to CD8+ T cells, ability to easily combine with anti-PD1 agents, and recent promising results with anti-LAG3 agents in the clinic.

Methods First, potency-reduced IL15/IL15Rα were created by engineering amino acid substitutions in IL15 – at the IL2Rβ/γc interface – that reduced in vitro potency by at least 1000-fold. We then designed LAG3 x IL15 fusion proteins containing single-chain IL15/IL15Rα and LAG3-targeting arms attached to a heterodimeric-Fc region, relying on targeting avidity (in cis) to recover potency on LAG3+ cells. In vitro proliferation of lymphocytes in human PBMCs, stimulated with sub-optimal concentrations of anti-CD3 to induce LAG3 expression, was monitored by CFSE dilution or by counting Ki67+ cells after incubation with LAG3 x IL15. In vivo activity was evaluated using a humanized mouse tumor model. Lead LAG3 x IL15 were evaluated in non-human primates (NHP).

Results LAG3 x IL15 showed >500-fold selectivity compared to a non-targeted IL15 in an in vitro proliferation assay of lymphocytes stimulated for induced LAG3 expression. In vitro potency was greatest on effector memory CD8 T cells. LAG3 x IL15 were 3-fold more potent on CD8 T cells compared to CD4 T cells and had very weak activity on NK cells, consistent with minimal LAG3 expression on this population. In mouse tumor models, LAG3 x IL15 had single-agent anti-tumor activity and promoted significantly increased numbers of T cells. Moreover, LAG3 x IL15 combined productively with anti-PD1 to promote additional anti-tumor activity and T cell expansion. In NHP, LAG3 x IL15 has superior PK compared to native IL15-Fc and showed selective targeting of LAG3+ peripheral T cells.

Conclusions LAG3 x IL15 shows a promising profile of selective delivery to LAG3+ cells with minimal peripheral activity and may help to preferentially expand LAG3+ TIL in patients with cancer, while potentially avoiding systemic toxicity due to off-target activation and expansion of peripheral lymphocytes.