Reversal of pre-existing NGFR-driven tumor and immune therapy resistance

nature COMMUNICATIONS

ARTICLE

Reversal of pre-existing NGFR-driven tumor and immune therapy resistance

Melanomas can switch to a dedifferentiated cell state upon exposure to cytotoxic T cells. However, it is unclear whether such tumor cells pre-exist in patients and whether they can be resensitized to immunotherapy. Here, we chronically expose (patient-derived) melanoma cell lines to differentiation antigen-specific cytotoxic T cells and observe strong enrichment of a pre-existing NGFR high population. These fractions are refractory also to T cells recognizing non-differentiation antigens, as well as to BRAF plus MEK inhibitors. NGFR high cells induce the neurotrophic factor BDNF, which contributes to T cell resistance, as does NGFR. In melanoma patients, a tumor-intrinsic NGFR signature predicts anti-PD-one therapy resistance, and NGFR high tumor fractions are associated with immune exclusion. Lastly, pharmacologic NGFR inhibition restores tumor sensitivity to T cell attack in vitro and in melanoma xenografts. These findings demonstrate the existence of a stable and pre-existing NGFR high multitherapy-refractory melanoma subpopulation, which ought to be eliminated to revert intrinsic resistance to immunotherapeutic intervention.

The landscape of treatment regimens for late stage melanoma patients has shifted progressively in recent years, largely owing to the development of therapies designed to generate or enhance T cell-mediated tumor killing. Clinical benefit of immune checkpoint blocking antibodies, such as anti-PD-one and anti-CTLA-four, has been reported to be over fifty percent in melanoma. Also other T cell-based therapeutic modalities, such as adoptive T cell transfer of either endogenous or genetically engineered T cells, have shown durable responses in a subset of melanoma patients. However, the majority of tumors display either innate or acquired resistance to these therapies, due to highly pleiotropic mechanisms including lack of actionable and clonal antigens and tumor heterogeneity. Currently, there is still no full understanding of how these mechanisms contribute to immunotherapy resistance, especially in the context of intratumor heterogeneity.

This is of particular importance in melanoma, a highly heterogeneous cancer type exemplified not only by the frequent presence of subclonal genetic alterations, but also by intratumoral transcriptional differences between melanoma cells, corresponding to different cell states. We and others previously described one such cell state, which is characterized by high expression of the receptor tyrosine kinase AXL and low expression of the master melanocyte transcription factor microphtalmia-associated transcription factor and its downstream target MART-one/Melan-A. Functionally, this process of dedifferentiation or phenotype switching is associated with both enhanced tumor invasion and resistance to MAPK pathway inhibitors. We found that AXL was commonly expressed in heterogeneous patterns in clinical tumor samples, indicating that cell states in melanoma are also highly heterogeneous in patients.

Phenotype switching and dedifferentiation have been linked also to acquired T cell resistance, given that microenvironment-derived cytokines such as tumor necrosis factor can cause reversible downregulation of melanocytic differentiation antigens and resistance to cognate T cells. This plasticity is seen both in animal models and melanoma patients and is associated with expression of Nerve Growth Factor Receptor, a protein originally identified as a putative melanoma stem cell marker. NGFR has also been suggested to be a key regulator of phenotype switching. Recent reports indicate that cell state heterogeneity in melanoma may be even more complex, with at least four dynamic cell states that can follow reversible trajectories.

In spite of these advances, it is currently unknown whether resistant melanoma cells that emerge upon chronic exposure to cytotoxic T cells pre-exist in patients prior to treatment, which we examined here. We also investigated whether such T cell-resistant melanoma cells display broader therapy resistance and whether they could be resensitized to T cells in vitro and in vivo. Lastly, we examined any association between the T cell-resistant cell state and the response to immunotherapy in melanoma patients. We describe a pre-existing cell population characterized by high expression of NGFR, which displays innate resistance to cytotoxic T cells as well as their cytokines. Moreover, NGFR high cells display cross-resistance to a variety of other therapies, including BRAF and MEK inhibition, in vitro, in mice and in patients. These melanoma cells express high levels of the neurotrophic factor brain-derived neurotrophic factor, and inhibition of either BDNF or NGFR enhances sensitivity to T cell-mediated tumor killing. Conversely, elevating the levels of NGFR in NGFR ten cells confers T cell resistance. Lastly, pharmacological inhibition of NGFR restores T cell sensitivity in tumor cells. These findings considerably extend our understanding of the importance of distinct melanoma cell states, in particular for NGFR high cell fractions, in governing immune resistance. Our results may warrant further exploration in a clinical, therapeutic setting.