Targeting PD-L1 interactome signaling to enhance immunotherapy efficacy in head and neck squamous cell cancer

SUMMARY. The programmed death receptor 1 (PD-1) drives immune escape in head and neck squamous cell cancer (HNSCC), whose incidence is rising due to human papillomavirus (HPV). Inhibitors of PD-1 (PD-1i) in T cells, or its ligand in cancer cells (PD-L1i), are in use in many cancers including recurrent/metastatic (R/M) HSNCC, but most patients do not respond. While the PD-L1 extracellular domain (ECD) is a ligand mediating T- cell anergy, the intracellular domain (ICD) role is less known. We found that PD-L1 over-expression increased key pro-tumorigenic properties. ICD deletion disrupted this signaling and reversed the pro-growth features. We hypothesized that PD-L1 ICD interacts with intracellular proteins, and to define the PD-L1 interactome we used a proximity-dependent labeling assay to measure interacting proteins with/without PD-1 binding. The PD-L1 interactome was consistent across cell lines, and top partners included nuclear factor of activated T cells (NFAT) subunits interleukin enhancer-binding factor 2 and 3 (ILF2/ILF3), that stabilized the signal transducer and activator of transcription 3 (STAT3) to induce pro-growth properties. In humanized mouse models of HPV+ and HPV- HNSCC, PD-1 binding induced PD-L1 signaling, and combined PD-L1i and STAT3 inhibitors (STAT3i) were required to achieve durable tumor control. In sum, PD-1 binding to PD-L1 led to an elegantly coordinated effect of simultaneous immune evasion and tumor progression, the latter driven by STAT3. In this project we will study the full spectrum of the PD-L1 interactome in HNSCC from its mechanistic basis to in vivo validation, and then patient clinical testing. First, we will identify the ICD precise domain responsible for the PD-L1 interactions with its cytoplasmic targets, by deploying HNSCC (and melanoma as control) cell lines with a set of mutated ICD constructs on a PD-L1 knockout by CRISPR/Cas9 backbone. PD-L1 interactome and signaling disruption will be assessed by changes in intrinsic and extrinsic properties. This will be followed by structural modelling and analyses of PD-L1 interactions that will enable discovering pharmacological disruptors. Secondly, we will test the role of STAT3 signaling in PD-1i response in a HNSCC humanized model with patient-matched, autologous thymic education capability. To elucidate complex immune events, in vivo testing is key, and to be relevant, mouse models need to be able to elicit a true, immune-based antitumor response. Thirdly, we will conduct a proof-of-concept trial of the STAT3i NT219 combined with the approved PD-1i pembrolizumab in PD-1i-naïve HNSCC patients. The planned immuno/pharmacodynamic and efficacy testing will explore if combined therapy increments response, and if baseline parameters including STAT3 signaling expression predict efficacy, setting the stage for practice-changing trials. The notion that the PD-1:PD-L1 axis, the most validated checkpoint inhibitor pathway in human cancer, is a bi-directional and synchronized process rather than a simple ligand- receptor step is a paradigm shift. By understanding canonical PD-L1 signaling and the coordinated role of the PD-L1:PD-1 interaction we will unlock the full potential of immune modulation as an anticancer therapy.