JAK2V617F Mutations Promote PD-L1 Expression to Mediate Immune Escape in Myeloproliferative Neoplasms

Programmed death ligand 1 and 2 (PD-L1 and PD-L2) engage the programmed death receptor 1 (PD-1) on T cells and induce T-cell exhaustion. Pharmacologic blockage of PD-1 signaling has resulted in numerous therapeutic advances for patients with cancer and may be particularly effective in tumors with a high mutational burden such as melanoma and non–small-cell lung cancer. In contrast to these tumor types, however, most hematopoietic malignancies have a lower mutational burden, and consequently, many patients with refractory leukemias and lymphomas do not seem to benefit from blockade of PD-1. One notable exception to this paradigm is Hodgkin lymphoma. For classical Hodgkin lymphoma (cHL), the PD-1-blocking antibodies pembrolizumab and nivolumab have received U.S. Food and Drug Administration approval for patients who have relapsed or progressed after autologous hematopoietic stem-cell transplantation.^1,2  This sensitivity is thought to be due to the fact that Reed–Sternberg cells of cHL are characterized by genomic amplification of the 9p24.1 locus, leading to overexpression of PD-L1 and PD-L2. Interestingly, the kinase JAK2 is also encoded at 9p24.1, approximately 322kb upstream of PD-L1.³  So, does upregulation of JAK2 activity also contribute to immune escape, or is JAK2 amplification simply a “bystander” of being adjacent to PD-L1?

In the article by Dr. Alessandro Prestipino and colleagues, the authors identify that expression of constitutively active JAK2 mutations as seen in patients with myeloproliferative neoplasms (MPNs) upregulates PD-L1 expression and confers responsiveness to PD-1 targeting in preclinical MPN models. To understand if there was a relationship between constitutively active forms of JAK2 and PD-1 expression, the authors tested PD-1 expression in isogenic cell lines with overexpression of wild-type JAK2 or JAK2V617F, mouse models with knock-in of the Jak2V617F mutation into the endogenous locus of Jak2, and cells from patients with JAK2-mutant MPNs versus healthy controls. This effort led to the consistent observation that the JAK2V617F mutation was associated with increased PD-L1 expression. Prior work has suggested that the transcription factor STAT3, and possibly STAT5, which are activated by JAK2, may directly bind to the promoter of PD-L1 and promote its expression. Consistent with this hypothesis, expression of a constitutively active STAT3 (STAT3 Y640F) also upregulated PD-L1 expression.

While the above results are exciting, it is important to note that other frequently mutated kinases that have been linked to STAT3 activation did not seem to be associated with PD-L1 expression. For example, FLT3 internal tandem duplications or tyrosine kinase domain mutations, alterations in PDGFR (the FIP1L1-PDGFRa translocation), and EGFR mutations (Del19, L861Q, and L858R) did not seem to upregulate PD-L1 protein expression in isogenic systems. Even more curiously, mutations in CALR, which are mutually exclusive with JAK2 mutations in MPN and are found in the majority of MPN patients lacking the JAK2V617F mutation,^4,5  were also not associated with increased PD-L1 expression. Thus, further mechanistic work to understand the precise relationship between STAT3 signaling and PD-L1 expression is clearly needed.

In addition to identifying a link between the JAK2 mutation and induction of PD-L1 expression, the authors found that pharmacologic JAK2 inhibition reduced both STAT3 activation as well as PD-L1 expression. This was identified by testing the effects of anti–PD-1 and anti–PD-L1 antibodies in syngeneic, immunocompetent mouse models of Jak2V617F-mutant MPNs as well as in xenografts of peripheral blood mononuclear cells from JAK2-mutant MPN patients. In these latter models, depletion of CD3+ T cells removed the survival advantage of anti–PD-1 antibody, indicating that response to anti–PD-1 therapy depends on an intact T cell response.