Modeling Of Cns Cancer With A Focus On The Immune Component

Author

Daniel ZamlerFollow

Author ORCID Identifier

0000-0003-0375-8119

Date of Graduation

5-2022

Document Type

Dissertation (PhD)

Program Affiliation

Cancer Biology

Degree Name

Doctor of Philosophy (PhD)

Advisor/Committee Chair

Giulio F. Draetta M.D. Ph.D.

Committee Member

Jian Hu Ph.D.

Committee Member

Ronald A DePinho M.D.

Committee Member

Jason Huse M.D. Ph.D.

Committee Member

Paul Scheet Ph.D.

Abstract

The knowledge surrounding cancers of the central nervous system remains poorly developed, in particular with regard to the immune component. The works contained in this thesis look at craniopharyngioma, glioblastoma, and several forms of brain metastasis. While some attention is given to the tumor cells themselves, as well as the patient setting which these studies model, the immune component of disease progression and treatment plays a strong role in each and is the primary focus of the works contained.

Craniopharyngioma is a relatively rare tumor in adults. Although histologically benign, it can be locally aggressive and may require additional therapeutic modalities to surgical resection. In the first set of experiments contained within, multiplatform analyses including next generation sequencing, chromogenic and in situ hybridization, immunohistochemistry, and gene amplification were used to profile craniopharyngiomas (n=6) to identify frequent therapeutic targets. Sixty-seven percent of patients had the BRAF V600E missense mutation, frequent in the papillary craniopharyngioma subtype. One patient had a missense mutation in the WNT pathway, specifically a mutation in CTNNB1 associated with the adamantinomatous subtype. Craniopharyngiomas lacked microsatellite instability, had relatively low tumor mutational burden, but did express PD-L1 protein, indicating potential use therapeutic use for immune checkpoint inhibition. We identified mutations not previously described, including an E318K missense mutation in the MITF gene, an R1407 frameshift in the SETD2 gene of the PIK3CA pathway, R462H in the NF2 gene, and a I463V mutation in TSC2. Two patients testing positive for epidermal growth factor receptor (EGFR) expression were negative for the EGFRvIII variant. Herein, we identified several alterations such as those in BRAF V600E and PD-L1, which may be considered as targets for combination therapy of residual craniopharygiomas. We hope that these insights may lead to better treatment of patients with this disease in the future.

Novel therapeutic strategies, including immunotherapeutics, targeting glioblastoma (GBM) often fail in the clinic, at least partly because available preclinical models do not recapitulate the human disease. To address this challenge in our second set of experiments, we took advantage of our previously developed spontaneous Qk/trp53/Pten (QPP) triple-knockout model of human GBM and compared its immune microenvironment components with those of patient-derived tumors in effort to determine whether this model might provide an opportunity for gaining insights into tumor physiopathology as well as for preclinical evaluation of therapeutic agents. Immune profiling analyses and single-cell sequencing of implanted and spontaneous tumors from QPP mice as well as from GBM patients revealed intratumoral immune components that were predominantly myeloid cells (e.g., monocytes, macrophages, and microglia) with minor populations of T, B, and NK cells. When comparing spontaneous and implanted mouse samples, we found that there were more neutrophils, T and NK cells in the implanted model. Neutrophils, T and NK cells were increased in abundance in samples derived from human high-grade glioma (HGG) compared to those derived from low grade glioma (LGG). Overall, our data demonstrate that our implanted and spontaneous QPP models recapitulate the immunosuppressive myeloid dominant nature of the tumor microenvironment of human gliomas. Our model provides a suitable tool for investigating the complex immune compartment of gliomas and it may contribute to a better understanding of the resistance of human glioblastoma to currently available immunotherapeutics. Given that we established the QPP model as viable for immune studies we next sought to pursue a therapeutic target in the form of Arginase-1 using this system.

An average GBM has roughly 30% infiltration of myeloid cells, the highest reported case at 70% infiltration, with some variation dependent on subtype. When we consider this fact in combination with the advent and success of immunotherapies in similar cancer types, the logical next step follows that myeloid cells, which are of the immune lineage, have the potential to clear this aberrant growth. Known for their incredible phagocytic capacity, myeloid cells are among the first responders to an injury to control pathogens and clear apoptotic cells from the site of an insult. Myeloid cells, through poorly understood mechanisms, undergo a switch after 3-5 days from attacking pathogens and clearing dead cells to secretion of growth cytokines to promote wound healing. We posit these myeloid cells may be tricked into erroneously supporting the tumor cells in an attempt to “heal the wound” instead of mounting an immune response. The enzyme responsible for this switch from pathogen response to wound healing is called Arginase-1 (Arg1). When Arg1 is upregulated, it has a two-pronged effect, the first is creation of peroxynitrite which is actively immunosuppressive to both myeloid and T-cells. The second is an increase in ornithine when Arg1 cleaves arginine into ornithine and urea, which can be used to create the building blocks for cellular growth as well as to create collagen, which is necessary for wound repair and found in high levels in brain tumors. Our hypothesis is that through manipulation of the Arg1 axis in myeloid cells we will increase anti-tumor response. We show the groundwork for evaluating whether or not this is a viable therapeutic target in GBM as well as assess currently available compounds for inhibition or Arg1 in the CNS.

It is a truism in cancer medicine that 90% of deaths are caused by cancer metastasis. While unfortunately difficult to collect information on, there are datasets that suggest the trend in increased metastatic cancer deaths has outpaced primary cancer deaths in recent decades. Metastasis to the brain is a large cause of mortality and roughly 300,000 cases are diagnosed annually. In an effort to address this urgent patient need we aimed to build a library of metastatic tumor models that were representative of the patient population and determine whether the models could be used for preclinical utility. We first assessed the growth patterns, latency and penetrance of our models and found that in general models tended to grow or not with little impact on survival from starting cell number. We came across several modeling problems including lack of engraftment and extracranial growth of tumors that presented challenges for modeling certain types of metastases to the CNS. We next chose to put several of our models through a battery of therapeutic regimens including Csfr1 inhibition, checkpoint blockade in the forms of anti-Pd-1 and anti-Lag-3 monoclonal antibodies, as well as radiotherapy. We used these single agent studies to inform upon potentially synergistic combinations that would then inform upon clinical trials.

Keywords

Glioblastoma, Glioma, CNS, Immune System, QPP, Brain Metastasis, Craniopharyngioma, Arginase 1, CNS Immune System, Immunity