Abstract 3518: AB248 is a CD8+ T cell selective IL-2 designed for superior safety and anti-tumor efficacy

High dose IL-2 was the first modern immunotherapy to show complete responses in a subset of cancer patients. The liabilities of IL-2 prompted development of second-generation molecules, which signal through the IL-2Rβγ (“not-α” IL-2 and IL-15 variants). Although these molecules avoid vascular leak syndrome, their clinical efficacy appears to be suboptimal compared to high dose IL-2 [1]. Furthermore, not-α IL-2s show biased expansion of NK cells over CD8+ T cells in patients due to their high expression of IL-2Rβ on NK cells, which can act as a sink and contribute to toxicity, and these molecules do not eliminate IL-2Rβγ-mediated activation of Tregs [2, 3]. To maximize the therapeutic potential of IL-2, we developed AB248, a cis-targeted IL-2 fusion protein that selectively signals on CD8+ T-cells with limited activity on the highly IL-2/IL-15-sensitive NK cells and immunosuppressive Tregs. We have previously shown that selective expansion of CD8+ T cells over NK cells and Tregs is achieved in vivo using muAB248, AB248’s murine surrogate. Furthermore, a single dose of muAB248 elicits strong efficacy in multiple tumor models without body weight loss. Here, we detail the properties of AB248, a fusion of a human CD8-targeting antibody that selectively binds to CD8+ T cells over NK cells, and an IL-2 mutein with reduced affinity to IL-2Rα and IL-2Rβ. In STAT5 assays, AB248 showed approximately 1000-fold preference for the activation of CD8+ T cells over NK cells and Tregs. AB248 mediated a dose-dependent selective expansion of CD8 T cells over NK and Tregs in non-human primates. Furthermore, AB248 reproduced the expected in vitro gene signatures of an IL2Rβγ agonist, demonstrating that AB248 recapitulates native IL-2Rβγ signaling selectively on CD8+ T cells.We previously demonstrated in mice that while the efficacy of not-α IL-2 was mediated by CD8+ T cells and not NK cells, toxicity as measured by body weight loss was dependent upon NK cells but not CD8+ T cells. Here, we show that human NK cells may also drive IL-2βγ agonist-induced toxicities. Both IL-2 and not-α IL-2 induced IFN-γ secretion from human PBMCs, whereas AB248 did not. Strikingly, depletion of CD56+ NK cells eliminated IL-2-induced cytokine secretion, demonstrating that human NK cells are capable of spontaneously secreting IFN-γ in response to IL-2 signaling. In the context of TCR stimulation, AB248 induced robust secretion of effector cytokines from CD8+ T cells, but no cytokine secretion was s