Inverse correlation between TP53 gene status and PD-L1 protein levels in a melanoma cell model depends on an IRF1/SOX10 regulatory axis

PD-L1 expression on cancer cells is an important mechanism of tumor immune escape, and immunotherapy targeting the PD-L1/PD1 interaction is a common treatment option for patients with melanoma. However, many patients do not respond to treatment and novel predictors of response are emerging. One suggested modifier of PD-L1 is the p53 pathway, although the relationship of p53 pathway function and activation is poorly understood.

We show that p53 loss influences the level of PD-L1 through IRF1 and SOX10 in an isogenic melanoma cell model, and that p53 loss affects NK-cell cytotoxicity toward tumor cells. Moreover, activation of p53 by MDM2 inhibition has a complex effect on IRF1/PD-L1 activation. These findings indicate that evaluation of p53 status in patients with melanoma will be important for predicting the response to PD-L1 monotherapy and/or dual treatments where p53 pathways participate in the overall response.

The study was performed on human melanoma cell lines with various p53 status. We investigated PD-L1 and proteins involved in IFNγ signaling by immunoblotting and mRNA expression, as well as membrane expression of PD-L1 by flow cytometry. We evaluated differences in the ability of NK cells to recognize and kill target tumor cells on the basis of p53 status. We also investigated the influence of proteasomal degradation and protein half-life, IFNγ signaling and p53 activation on biological outcomes, and performed bioinformatic analysis using available data for melanoma cell lines and melanoma patients.

We demonstrate that p53 status changes the level of membrane and total PD-L1 protein through IRF1 regulation and show that p53 loss influences the recently discovered SOX10/IRF1 regulatory axis. Bioinformatic analysis identified a dependency of SOX10 on p53 status in melanoma, and a co-regulation of immune signaling by both transcription factors. However, IRF1/PD-L1 regulation by p53 activation revealed complicated regulatory mechanisms that alter IRF1 mRNA but not protein levels. IFNγ activation revealed no dramatic differences based on TP53 status, although dual p53 activation and IFNγ treatment confirmed a complex regulatory loop between p53 and the IRF1/PD-L1 axis. Conclusions: We show that p53 loss influences the level of PD-L1 through IRF1 and SOX10 in an isogenic melanoma cell model, and that p53 loss affects NK-cell cytotoxicity toward tumor cells. Moreover, activation of p53 by MDM2 inhibition has a complex effect on IRF1/PD-L1 activation. These findings indicate that evaluation of p53 status in patients with melanoma will be important for predicting the response to PD-L1 monotherapy and/or dual treatments where p53 pathways participate in the overall response.