Date of Award

Spring 5-15-2022

Author's School

Graduate School of Arts and Sciences

Author's Department

Biology & Biomedical Sciences (Immunology)

Degree Name

Doctor of Philosophy (PhD)

Degree Type



Glioblastoma multiforme (GBM) patients face limited treatment options and poor outcomes. The median survival is less than two years, and there are no FDA approved immune therapies. Although GBM itself is an immune-suppressive, heterogeneous tumor, the lack of FDA approved immune therapies might be in part because the cancer immunity cycle is less well understood for GBM than for other tumor types. My studies focused on developing mouse models of malignant glioma that more faithfully recapitulate human GBM from an immunologic perspective, and on defining the role of the conventional dendritic cell 1 subset (cDC1) and lymphatic drainage in central nervous system (CNS) antitumor immunity.

While genetically engineered mouse models of glioma have been described, they are for various reasons unsuitable to study the immune system’s reaction against the tumor, due to their use of outbred mice, immunologically immature mice, human oncogenes to drive transformation, or highly inflammatory initiation events. Furthermore, the most commonly deleted tumor suppressors in GBM are underrepresented in existing models. Thus, we engineered the tumor suppressor genes p16INK4a and p19ARF (INK4a/ARF; CDKN2A/B in humans) and phosphate and tensin homolog (PTEN) to be loxP-flanked on a pure C57BL/6 background. We used lentiviral transduction of Cre and the murine oncogene platelet derived growth factor beta (PDGFβ) to conditionally delete these tumor suppressors and transform target cells in brains of immunologically mature mice, which resulted in brain tumor formation.

With the standard treatment, GBM invariably recurs, with 20%-30% of cases hypermutated. It is often the loss of mutS homolog 6 (MSH6), a mismatch repair protein, that confers resistance to temozolomide (standard-treatment) and leads to treatment-induced hypermutations. We developed the tools to model this phenomenon in a preclinically. We isolated astrocytes from the B6 INK4a/ARFfl/fl x PTENfl/fl mice and transformed them with the Cre/mPDGFβ lentivirus constructs. We used CRISPR to delete the mismatch repair protein MSH6 in these ex-vivo transformed astrocytes. We characterized their resistance to temozolomide and successfully induced hypermutation with long term temozolomide treatment and inhibition.

Within the immunologically distinct location of the CNS the type of antigen presenting cell (APC) responsible for priming T cell responses against brain tumors remains undefined. In other non-CNS tumors, the conventional dendritic cell 1 (cDC1) subset cross-presents tumor-derived and cell-associated tumor antigen to generate antitumor CD8+ and CD4+ T cell responses. However, the homeostatic brain parenchyma is largely devoid of cDC1—their steady state location is restricted to the choroid plexus and the dura. Using orthotopic, syngeneic transplant models of murine glioblastoma, we investigated the roles of cDC1 and other antigen presenting cells in antitumor immunity of the CNS. We used the cDC1-deficient interferon regulatory factory 8-deficient (IRF8+32-/-) mice to determine that cDC1 are required to mediated αPD-L1 induced survival benefit as well as to generate neoantigen-specific CD8+ T cell responses against the brain tumors. Furthermore, using a fluorescent tracking system, we observed that dendritic cells (including the cDC1 subset) isolated from the tumor, the lymphatic vessel-containing dura, and the cervical lymph nodes harbored tumor-derived antigen. We extended these findings to humans. We identified several subsets of conventional dendritic cells, including the CD141+ cDC1 equivalent, in the immune cell infiltrate of a variety of human brain tumor types (including GBM), as well as in the tumor-adjacent dura. We determined tumor-infiltrating dendritic cells, including the CD141+ subset (equivalent to the mouse cDC1), contained the tumor-specific fluorescent metabolite of 5-aminolevulinic acid (5-ALA), protoporphyrin IX (PPIX), which is used for fluorescence guided resection of malignant glioma. The PPIX signal was absent in both tumor-infiltrating T cells and equivalent dendritic cell subsets isolated from intraoperatively harvested peripheral blood, which indicates that this phenomenon was specific to antigen presenting cells that had infiltrated the tumor. To our knowledge, this is the first observation in humans of antigen presenting cells ingesting tumor-derived material.

Together, these data provide evidence that cDC1 play a significant role in CNS antitumor immunity in mice and humans. Collectively, these studies have yielded improved tools to study the immunity cycle in GBM and have shed light on some of the elements regarding the nature and mechanism of antigen presentation in CNS antitumor immunity.


English (en)

Chair and Committee

Gavin P. Dunn

Committee Members

David De Nardo

Included in

Biology Commons