Design of a Novel Cancer Suicide Gene Therapy Regulated by Casein Kinase II (CK2) Activity

Viral vectors have long been pursued as attractive potential alternatives or adjuvants for chemotherapy-resistant cancers. A major focus of that work has been the development of retroviral vectors for suicide genes that encode non-mammalian enzymes capable of converting innocuous prodrugs into cytotoxic metabolites. Suicide gene therapies have thus far failed to reach the clinic despite more than two decades of clinical trials, which we hypothesize have involved inherently flawed vector designs that rely on either weak, cancer-specific promoters or strong, non-specific viral promoters. We previously designed a new suicide gene vector that overcomes this efficacy-limiting tradeoff by combining a strong viral promoter with a posttranslational cancer stability switch based on the activity of the oncogenic extracellular signal-regulated kinase (ERK) pathway. In that design, a suicide gene (herpes simplex virus thymidine kinase or yeast cytosine deaminase) was fused with a nuclear localization sequence and an unstable PEST domain (rich in proline, glutamic acid, serine, and threonine) from the FRA1 transcription factor, which is stabilized when phosphorylated by ERK. In this thesis, we sought to extend the success of that design by creating a suicide gene product stabilized by casein kinase II (CK2), a holoenzyme with constitute activity that is widely involved in disease progression and chemoresistance in cancers including glioblastoma. The major challenge in designing a CK2-regulated suicide gene is that PEST-like peptide sequences regulated specifically by CK2 are unknown. To address this, we implemented a two-stage design process beginning first with a small panel of peptide sequences based primarily on a FRA1 scaffold with incorporation of CK2 phosphorylation sites. The peptide sequences were intentionally designed to capture variation in properties including the number and strength of PEST domains and CK2 phosphorylation motifs, as well as the positioning of these along the peptide. FLAG-tagged fusions of these peptide sequences with herpes simplex virus thymidine kinase were stably expressed in glioblastoma cells and evaluated for their stability and response (degradation) to CK2 pharmacological inhibition. Because the most CK2-regulated peptide was also the least stably expressed, a machine learning model was constructed to predict the peptide features that most determine peptide stability. Based on the model, a second-generation of the most CK2-regulated peptide was engineered and found to have superior stability while retaining strong regulation by CK2 activity. Specificity of the second-generation suicide gene product for regulation by the two CK2α catalytic subunits was confirmed using stable shRNA-mediated knockdowns. Furthermore, we found that the second-generation suicide gene product was stabilized in its expression in response to the chemotherapeutic carboplatin in a CK2-dependent manner and that carboplatin and GCV synergistically killed glioblastoma cells engineered to express the suicide gene. Thus, this work produced a novel cancer suicide gene vector engineered for specific regulation by a kinase that is broadly important in disease progression and therapeutic resistance and provides a framework for the deployment of machine learning algorithms to design other kinase-regulated suicide genes in an iterative fashion.