Characterization of senescent stromal cells in the tumor microenvironment treated with immune checkpoint blockade

Introduction: The resistance of solid tumors to immune checkpoint blockade (ICB) has made it increasingly clear that unleashing endogeneous effector T cells is not sufficient for clinical response and targeting the tumor stroma may improve sensitivity to ICB. Recently much attention has been given to senescent cells (SnCs), which are most commonly defined through their state of persistent cell cycle arrest. Despite being a relatively infrequent population, SnCs can mediate potent inflammatory and ECM perturbation through their senescence associated secretory phenotype (SASP). However, senescence is a complex cellular state; it can occur in many cell types, has various inducers, and imperfect biomarkers. Whereas most of the studies on senescence and cancer focuses on pre-neoplastic cells or tumor cells, stromal senescence remains largely understudied. Therefore, our goal is to profile SnCs in responsive and non-responsive TMEs to verify their role in mediating ICB response. Our hypothesis is that senescent cells play various roles in TME; specifically, angiogenic senescence has an anti-tumorigenic effect and fibrogenic senescence has a pro-tumorigenic effect.

Materials and

Methods: MC38 cells are used as our immunotherapy responsive model and B16F10 cells are used as our non-responsive model. MC38 tumors were treated with 4 doses of PDL1 every 3 days at 5 mg/kg or isotype given IP. B16F10 tumors were treated with 4 does of PD1/CTLA4 every 3 days at 5 mg/kg or isotype given IP. For SnC co-implantation studies, SnCs were generated in vitro by irradiating dorsal skin fibroblasts with 10 Gy of irradiation from both WT and IL-6 KO mice and co-injected with B16F10 tumor cells.
The INKA mouse model uses a tamoxifen-inducible, cre-lox system to report p16 gene expression as tdTom florescence. INKA mice were given 70mg/kg of tamoxifen in corn oil, IP for five consecutive days starting a week before harvest. The INKATTAC model deletes p16+ cells through a caspase system inducible by a small molecule, AP20187. Mice were treated daily with high-dose AP20187 (10mk/kg) for 5 days given IP, followed by low-dose AP20187 (2 mg/kg) twice a week.
For flow cytometry, tumors were processed into a single cell suspension and stained with antibodies against the following markers, CD45,CD3,CD11b,CD31,CD29,CD90. For SnC reporting, the INKA model was used to detect SnCs through expression of tdTOM. Samples were run on a Attune flow cytometry and analyzed used Flowjo. For histology, tumors were FFPE treated; slides were generated; stained using the opal system to detect nuclei, p16, aSMA, and CD31; imaged on a Vectra Polaris; and analyzed using Halo.
Patient NSCLC tumors were treated with Nivolumab.

Results, Conclusions, and Discussions:

Results:
We observed an increase in SnCs following ICB treatment in MC38 tumors [Figure 1A] and these cells were closely associated with CD31+ vasculature. We are performing additional IF staining to confirm the identity of these cells. Preliminary flow cytometry on ICB treated MC38 tumors in INKA reporter mice revealed the majority of SnCs in the TME were endothelial cells. We are currently performing an INKATTAC deletion study in the MC38 model to determine if these cells play a functional role in immunotherapy responsiveness.
B16F10 tumors co-implanted with senescent WT fibroblasts (irradiation induced) grew faster than B16F10 cells alone. This effect was reduced when irradiated IL-6 KO fibroblasts were used instead, indicating that a key SASP factor plays a crucial role in mediating the effect. [Figure 1B]. The impact of co-implanted SnC fibroblasts on tumor growth could also be reversed by co-delivery of a senolytic (navitoclax) and immunotherapy. [Figure 1C]
IF staining in human NSCLC treated with immunotherapy revealed a large heterogeneity of SnCs, including fibroblastic and pericyte-like SnCs [Figure 1D].

Conclusions:
We found that ICB treatment and immunotherapy responsiveness led to an increase of SnCs in the TME of MC38 tumors. Profiling of these cells revealed that they closely associate with vasculature and flow cytometry confirmed that they were endothelial and peri-vascular. Further, co-implanting tumors with in vitro induced fibroblastic senescent cells was seen to worsen tumor outcomes and deletion of these senescent cells or a key SASP factor, IL-6, partially reversed this phenotype. In human NSCLC tumors, both fibroblastic and vasculature associated SnCs were apparent.

Discussions:
Non-tumoral SnCs might be useful in ICB treatment as either a target for deletion or a possible biomarker in predicting responsiveness. The work completed helps establish both an association between ICB responsiveness and SnCs, and the opposing nature of fibroblastic and vasculature-associated SnCs. Future experiments in the INKATTAC mouse model will delete SnCs in the TME of a responsive tumor to determine the function impact of these senescent subsets on immunotherapy responsiveness.

Acknowledgements (Optional): INKA and INKATTAC mice were kindly provided by Dr. Darren Baker

MDPhD Program at Hopkins – MSTP T32 (3T32GM136577, PI Cox)