DNA damage repair deficiencies (DDRD) can elicit an innate immune response via activation of the cytosolic dsDNA-sensing cGAS-STING pathway. However, activation of this pathway can induce the up regulation of immunosuppressive checkpoint proteins, promoting an inflamed but immunosuppressed state. This provides a rationale for the use of immune checkpoint inhibitors (ICIs) for DDRD-positive tumours. Interestingly, mutations in the spliceosome, a premature-mRNA splicing complex, are commonly found in myeloid malignancies. Previously, we described how mutations in a key spliceosome component, SF3B1, conferred a DDRD. Therefore, this study aims to explore how common spliceosome mutations in SRSF2 and SF3B1 can be exploited to guide the use of ICIs in myeloid malignancies.

Using CRISPR/Cas9-generated isogenic models harbouring the SRSF2P95H and SF3B1K700E mutations, DNA damage repair kinetics, micronuclei induction and cytosolic dsDNA/dsRNA accumulation, were assessed using immunofluorescent labelling. qRT-PCR and ELISA analysis were used to assess expression and secretion of inflammatory cytokines, respectively, following treatment with daunorubicin, a topoisomerase II inhibitor. The immune competent embryonic chick chorioallantoic membrane (CAM) model was used for in vivo validation of the benefit of ICIs in spliceosome-mutant models.

Similar to previous findings in our SF3B1K700E model, the SRSF2P95H mutation conferred a reduced ability to repair DNA damage, indicating a potential DDRD. DNA damaging agents have been shown to boost cGAS-STING activation, particularly in DDRD tumours. DNA damage induction by daunorubicin induced significant increases in micronuclei and cytosolic dsDNA in SRSF2P95H and SF3B1K700E models compared to WT controls. Significant increases in expression and secretion of inflammatory cytokines, CCL5 and CXCL10, were observed in both our SRSF2 and SF3B1-mutant models, compared to WT counterparts post-daunorubicin. Interestingly, no significant reductions in CCL5 transcription or secretion were observed following STING-depletion in the SRSF2P95H model but were significantly reduced following depletion of MAVS, an adaptor protein involved in cytosolic dsRNA sensing. Indeed, cytosolic dsRNA accumulation was significantly increased in SRSF2P95H cells post-daunorubicin. Furthermore, ICI with nivolumab resulted in SF3B1K700E tumour control, but interestingly induced more aggressive SRSF2P95H tumours in the immune competent CAM model.

To date, our data suggests that the SRSF2P95H and SF3B1K700E mutations confer an increased capacity to induce an innate immune response compared to their WT counterparts. However, this response may be reliant on the dsRNA-sensing MAVS pathway rather than the STING-dependent pathway. Together, this data is pathing the way for use of SRSF2 and SF3B1 mutational status as a biomarker for the use of ICIs.