STING activation enhances immunotherapy of AML

In this issue of Blood, Linder et al¹ demonstrate that the efficacy of bispecific T-cell engager AMG 330 directed against CD33 on acute myeloid leukemia (AML) cells can be improved by simultaneous application of a stimulator of interferon genes (STING) agonist. Bispecific T-cell engagers (BiTEs) targeting B-cell antigens have revolutionized the therapeutic options for B-cell malignancies, but similar strategies directed against myeloid lineage antigens have yielded much lower response rates in AML.² A BiTE termed AMG 330 that crosslinks CD33 on myeloid cells with the CD3ε unit of the CD3/T-cell receptor (TCR) complex is currently in clinical phase 1 studies of patients with relapsed/refractory AML.³ 

In their study, Linder et al have addressed the issue of whether the efficacy of AMG 330 in killing AML cells in vitro and in a preclinical in vivo model can be enhanced by simultaneous activation of the innate immune sensor STING. The cytosolic cyclic GMP-AMP synthase/STING pathway is a major activator of the type I interferon response but also triggers pro-inflammatory cytokine induction through activation of NF-κB.⁴ Recently, the potential application of STING agonists to enhance cancer immunotherapy has generated great attention.⁵ Activation of innate immunity through STING agonists has already been suggested to overcome immune evasion in AML.^6,7 

Linder et al observed that the addition of STING agonist cGAMP to cocultures of peripheral blood T cells with myeloid leukemia cells in the presence of AMG 330 greatly increased the killing of myeloid HL-60 cells. This was accompanied by increased degranulation and release of granzyme B (GZMB), interferon-γ (IFN-γ), and tumor necrosis factor-α (TNF-α). Transcriptome analysis of T cells and HL-60 cells separated after coculture in the presence of cGAMP and/or AMG 330 revealed upregulation of IFNG, TNF, and GZMB genes in T cells by AMG 330, whereas type I IFN-responsive genes were induced by cGAMP. The combination of AMG 330 and cGAMP led to further upregulation of IFNG and GZMB compared with single-agent exposure. They went on to investigate whether the increased killing of HL-60 cells in the presence of cGAMP and AMG 330 was dependent on STING and IFN-γ receptor expression. To this end, they deleted STING and IFN-γ receptor expression in HL-60 cells by a CRISPR-Cas9 knockout strategy. Killing of STING-deficient HL-60 cells was not enhanced in cocultures with T cells in the presence of AMG 330 and cGAMP, indicating that AML cells require intrinsic STING signaling to enhance AMG 330–mediated cytotoxicity. Importantly, however, in the same experimental setting, cGAMP also did not increase killing of HL-60 cells deficient in cell surface IFN-γ receptor. Further studies revealed that T-cell–derived cytokines were required for intrinsic STING signaling in AML cells; this effect could be verified by direct exposure of HL-60 cells to IFN-γ. Most experiments were performed with myeloid cell lines such as HL-60. However, using primary AML cells from 11 patients, they were able to confirm that STING agonists cGAMP and diABZI (a small non-dinucleotide molecule) efficiently enhanced killing of patient-derived AML cells by AMG 330 in the presence of T cells. Last but not least they evaluated the effect of STING agonist on the immunotherapy of AML in a preclinical in vivo model. Immunodeficient NSG mice were injected intravenously with MOLM-13 AML cells, followed by injection with human T cells and a more stable derivative of AMG 330 together or without diABZI. The combined application of BiTE and STING agonist proved to be more efficient in reducing tumor burden associated with longer survival than each compound alone.

Taken together, Linder et al have uncovered a new strategy of how the response rate to myeloid-specific BiTE may be increased in patients with refractory/relapsed AML. Although AMG 330 has some activity on its own as it activates T cells by cross-linking CD3/TCR on T cells with CD33 on AML cells (see figure panel A, left), the killing capacity is strongly enhanced by STING agonists (see figure panel A, right). From a translational viewpoint, the confirmation of the efficacy of STING costimulation with patient-derived AML cells and in the in vivo model suggest that this strategy could be rapidly explored in clinical studies.

In addition to BiTEs, chimeric antigen receptor T cells and other TCR therapies are in development for treatment of AML.⁸ Of special interest are Delta One T (DOT) cells, a subpopulation of human γδ T cells expressing Vδ1 as a variable element of the γδ TCR δ-chain. γδ T cells (including DOT cells) do not depend on HLA molecules for their activation; that is, their mode of action is HLA-independent. This offers a unique opportunity for γδ T cells to be used in a therapeutic setting across HLA barriers. DOT cells expanded from peripheral blood are potent effector cells in preclinical models of chronic lymphocytic leukemia and AML.^9,10 DOT cells should be suitable effector cells for BiTEs like AMG 330 because they express the CD3ε chain as part of their CD3/TCR complex. Therefore, we suggest exploring DOT cells as effector cells with AMG 330 (and possibly other BiTE)–mediated cytotoxicity of AML cells; the additional activation of DOT cells via the DNAM-1/poliovirus receptor axis may further boost the killing of AML cells (see figure panel B, left).¹⁰ Moreover, based on the findings by Linder and colleagues, the overall efficacy of this approach may be further enhanced by application of STING agonists (see figure panel B, right).

Conflict-of-interest disclosure: The authors declare no competing financial interests.