Gabarap Loss Mediates Immune Escape in High Risk Multiple Myeloma

Immune therapies including CAR T cells and bispecific T cell engagers are demonstrating remarkable efficacy in relapsed refractory myeloma (MM).

In this context, we have recently shown that proteasome inhibitor bortezomib (BTZ) results in immunogenic cell death (ICD) and in a viral mimicry state in MM cells, allowing for immune recognition of tumor cells. Induction of a robust anti-MM immune response after BTZ was confirmed both in vitro and in vivo: treatment of 5TGM1 MM cells with BTZ induced tumor regression associated with memory immune response, confirmed by ELISPOT of mouse splenocytes. We have confirmed the obligate role of calreticulin (CALR) exposure in phagocytosis and the ICD process, since BTZ-induced ICD is impaired in CALR ^KO MM cells both in vitro and in vivo. We further showed that the therapeutic efficacy of BTZ in patients was correlated with ICD induction: BTZ-induced ICD signature was positively correlated with OS (p=0.01) in patients enrolled in the IFM/DFCI 2009 study. Together, these studies indicate that ICD is associated with long-term response after BTZ treatment.

In this work, we reasoned that genomic or transcriptomic alterations associated with shorter survival of MM patients after BTZ treatment may impair activation of the ICD pathway. To this aim, we performed a transcriptomic analysis of purified CD138+ cells from 360 newly diagnosed, clinically-annotated MM patients enrolled in the IFM/DFCI 2009 study. By focusing on genes involved in the ICD process, we found that low levels of GABA Type A Receptor-Associated Protein (GABARAP) were associated with inferior clinical outcome (EFS, p=0.0055). GABARAP gene locus is located on chr17p13.1, a region deleted in high risk (HR) MM with unfavorable prognosis. Remarkably, we found that correlation of low GABARAP levels with shorter EFS was significant (p=0.018) even after excluding MM patients with del17p; and GABARAP is therefore an independent predictor of clinical outcome.

GABARAP is a regulator of autophagy and vesicular trafficking, and a putative CALR binding partner. Interestingly, among a panel of MM cell lines (n=6), BTZ treatment failed to induce exposure of CALR and MM cell phagocytosis by DCs in KMS11 cells, which carry a monoallelic deletion of GABARAP. This effect was rescued by stable overexpression of GABARAP. Moreover, CRISPR/Cas9-mediated KO of GABARAP in 3 ICD-sensitive cell lines (AMO1, H929, 5TGM1) abrogated CALR exposure and ICD induction by BTZ. GABARAP add-back by stable overexpression in KO clones restored both CALR exposure and induction of ICD, confirming GABARAP on-target activity. Similarly, pre-treatment of GABARAP ^KO cells with recombinant CALR restored MM phagocytosis, further confirming that GABARAP impairs ICD via inhibition of CALR exposure.

Based on these findings, we hypothesized that GABARAP loss may alter the ICD pathway via CALR trapping, resulting in the ICD resistant phenotype observed in GABARAP null and del17p cells. To this end, we explored the impact of GABARAP ^KO on the CALR protein interactome, in the presence or absence of BTZ. Importantly, GABARAP ^KO produced a significant increase of CALR binding to stanniocalcin 1 (STC1), a phagocytosis checkpoint that mediates the mitochondrial trapping of CALR, thereby minimizing its exposure upon ICD. Consistently, GABARAP ^KO also affected CALR interactome in BTZ-treated cells, which was significantly enriched in mitochondrial proteins. Importantly, co-IP experiments confirmed GABARAP interaction with STC1. These data indicate a molecular scenario whereby GABARAP interacts with STC1 to avoid STC1-mediated trapping of CALR, allowing for the induction of ICD after treatment with ICD inducers; on the other hand, this mechanism is compromised in GABARAP null or del17p cells, and the STC1-CALR complex remains trapped in the mitochondria, resulting in ICD resistance.

To functionally validate our findings in the context of the immune microenvironment, we performed mass Cytometry after T cell co-culture with DCs primed by both WT and GABARAP ^KO AMO1 clones. And we confirmed that treatment of GABARAP ^KO clones with BTZ failed to activate an efficient T cell response.

In conclusion, our work identifies a unique mechanism of immune escape which may contribute to the poor clinical outcome observed in del17p HR MM patients. It further suggests that novel therapies to restore GABARAP may allow for the induction of ICD and improved patient outcome in MM.