Sheri Holmen

Our lab is focused on defining critical targets in cancer cells that can become the focus for therapeutic intervention. Because of the high cost of developing new therapies, it is essential to first identify which genetic alterations can be productively targeted. We are concentrating our initial efforts on using a genetic approach in tumors that are refractory to conventional therapies including metastatic melanoma and glioblastoma. We plan to further validate these targets using pharmacological inhibitors of clinical importance.

Melanoma

The incidence of melanoma has been increasing at an alarming rate over the past thirty years. Approximately 90,000 new cases occur each year with nearly 10,000 resulting in death. Melanoma accounts for the majority of skin cancer deaths due to its propensity to metastasize to distant organs, including the lung and brain, and brain metastases are responsible for up to half of all melanoma deaths. Among all cancers that frequently metastasize to the brain, including breast, lung, colon, and renal cancer, melanomas have the highest frequency for colonizing this organ. As many as 43% of stage IV melanoma patients present with brain metastases at diagnosis and nearly 75% of autopsy reports identify central nervous system (CNS) involvement. Melanoma patients with brain metastases have a poor prognosis and overall survival for these patients ranges between 4 and 9 months from the time of diagnosis. Since 2011, several new therapies have been FDA approved for this disease but brain metastases are often a major component of treatment failure. Given the poor prognosis for these patients, improved prognostic biomarkers are desperately needed to identify those patients at highest risk for disease progression and development of brain metastases as early as possible. To date, multiple clinical, pathologic, and gene expression features associated with higher risk of development of metastases have been published. However, many individual published metastatic prognostic factors have failed validation in independent studies, and no detailed, integrated clinical and molecular predictor for brain metastases has been published. Development of more sensitive and specific predictors of disease spread to the brain would enable the implementation of clinical trials testing the effectiveness of directed surveillance and/or therapeutic prevention strategies in those patients at highest risk of developing future brain metastases. While earlier detection may have been of little consequence in the era prior to approval of effective systemic therapies, this is likely not the case today where multiple treatment options exist.

Further advances in the management of this disease require model systems that aid in the understanding of the behavior of melanoma and assist in the identification of mechanisms of resistance. A major initiative in our research program has been to develop a novel high-throughput mouse model of melanoma based on retroviral-mediated gene delivery to melanocytes. We have successfully developed this model and further employed the use of this model to better understand melanoma biology and response to therapy. We have identified genes and proteins with differential roles in melanoma initiation and progression as well as intrinsic resistance to mitogen-activated protein kinase (MAPK) inhibition. Our group has also extended the utility of this mouse model system by engineering the viruses to be responsive to doxycycline in the presence of Tet-off or Tet-on proteins. This allows us to regulate the expression of the delivered genes post-infection in vivo, define the role of specific genes in tumor maintenance, and develop models of resistance. We hope to use these tools to design rational combination therapies to improve outcome in patients with advanced melanoma.

In the melanoma model we developed, tumors evolve from gene mutations in developmentally normal somatic cells in the context of an unaltered microenvironment and therefore more closely mimic the human disease. Tumor development is a dynamic process that depends on the interactions between the tumor and its microenvironment. In our model, only a small number of cells are modified and therefore the cells surrounding the tumor are normal. Using this system, newly identified genes can be rapidly validated for their role(s) in tumor formation, progression, maintenance, and resistance to therapy. This method is based on the RCAS/TVA retroviral vector system that allows for tissue- and cell-specific targeted infection of mammalian cells through ectopic expression of the viral receptor. This system utilizes a viral vector, RCASBP(A), derived from the avian leukosis virus (ALV). The receptor for RCASBP(A) is encoded by the TVA gene and is normally expressed in avian cells but not mammalian cells. In mammalian cells engineered to express TVA, the viral vector is capable of stably integrating into the DNA and expressing the inserted experimental gene, but the virus is replication-defective, which allows for multiple rounds of infection. The ability of TVA-expressing mammalian cells to be infected by multiple ALV-derived viruses allows efficient modeling of human melanoma because multiple genetic alterations can be introduced into the same animal without the expense or time associated with creating new strains of mice. In addition, by restricting the expression of the viral receptor to melanocytes and by delivering the virus through subcutaneous injection, two levels of targeting are achieved.

The dopachrome tautomerase (DCT) promoter, also known as tyrosinase-related protein 2 (TRP2), was chosen to drive expression of the viral receptor TVA specifically in melanocytes since this gene is expressed early in melanocyte development when the cells are mitotically active. Because a significant percentage of both familial and sporadic melanomas have mutations that functionally inactivate both INK4a and ARF, DCT-TVA mice were crossed to Ink4a/Arf ^lox/loxmice to generate DCT-TVA;Ink4a/Arf ^lox/lox mice. As proof-of-principle, newborn mice were injected subcutaneously with RCAS viruses containing Cre-recombinase and NRAS^Q61R. Whereas no tumors were detected in TVA-negative mice, melanomas were visible in DCT-TVA;Ink4a/Arf ^lox/lox mice as early as three weeks of age. Within twelve weeks, more than one-third of DCT-TVA;Ink4a/Arf ^lox/lox mice developed melanoma that was histologically similar to the human disease. Delivery of a virus in which NRAS^Q61R and Cre expression was linked by an internal ribosomal entry site (IRES) resulted in tumor formation in nearly two-thirds of TVA positive mice. However, these tumors were not metastatic.

Metastasis is responsible for over 90% of all cancer-related deaths and unfortunately melanoma has a propensity to metastasize early in disease progression. In melanoma patients, the most common sites of metastases are skin, lung, brain, liver, bone, and intestine. Brain metastases are associated with extremely poor prognosis and are often responsible for treatment failure. Studies have suggested that as many as half of all melanoma deaths are due to brain metastases. Therefore, it is critical that we further our understanding of the metastatic process such that improved therapies can be developed for these patients. To this end, we have used our novel mouse model of melanoma to identify genes that promote metastasis. We have induced melanoma in this model by expression of mutant NRAS or BRAF in the context of Ink4a/Arf loss, Pten loss or AKT activation. Interestingly, tumors induced with mutant NRAS in the context of Ink4a/Arf loss are not metastatic tumors induced with mutant BRAF in the context of Ink4a/Arf loss and activated AKT1 metastasize to the lungs and brain similar to the human disease. We are currently evaluating the contribution of both Pten loss and activation of different AKT isoforms to melanoma metastasis in vivo with the goal of identifying more effective ways to target this pathway in tumor cells while limiting toxicity to normal cells.

Glioma

There has been a 35% increase in the incidence of pediatric brain tumors over the last thirty five years and brain tumors have become the single most common cause of cancer-related mortality in children. The primary treatment is surgical excision, but this is challenging in cases where the tumor is located within the optic pathways, the brain stem, or diencephalon (located near the midline of the brain). Prior to the recent discovery of BRAF mutations in these tumors, very little was known about the genetic alterations in pediatric malignant gliomas. These gliomas comprise World Health Organization (WHO) grades II-IV and account for 50% of cerebral tumors. Despite the addition of radiotherapy and chemotherapy, the prognosis for pediatric high-grade gliomas remains poor and 5-year survival rates are less than 20%. Recently, BRAF^V600E mutations were identified in 18% of WHO grade II, 33% of WHO grade III, and 18% of WHO grade IV pediatric gliomas (a combined 23% for grades II-IV). While these findings represent a significant breakthrough, the precise role of this alteration in the formation and maintenance of this disease remains to be defined. Whether BRAF or relevant downstream signaling mediators can be productively targeted for therapeutic intervention in glioma patients has yet to be determined.  

BRAF mutations have also been observed in adult grade III and IV gliomas. Glioblastoma multiforme (GBM) (WHO grade IV astrocytoma) is the most common and aggressive primary brain tumor in adults. It is also the most fatal. Despite major improvements in treatment, the prognosis for patients with this disease has not changed in the last twenty years. The current standard of care for this disease includes surgical resection followed by adjuvant radiotherapy and temozolomide (TMZ) chemotherapy. However, with standard treatment mean survival is only 15 months. Novel treatment strategies are badly needed to improve the outcome for these patients. We are working to better understand the biology of these gliomas to guide our development of new therapies, which we hope will improve survival and reduce morbidity in these patients.

High-throughput efforts such as The Cancer Genome Atlas (TCGA) pilot project and an extensive sequencing project have yielded novel information about GBM. In 2008, a multi-group collaboration sequenced >20,000 genes in 22 GBMs and identified a common point mutation in the metabolic gene isocitrate dehydrogenase 1 (IDH1) in secondary GBM. Numerous publications have since followed demonstrating that the majority of low-grade gliomas and secondary GBMs possess mutations in IDH1. The few exceptions that do not contain a mutation in IDH1 have an equivalent mutation in the related gene IDH2. This mutation had never before been linked to cancer and the function remains unclear. There have been reports suggesting the gene functions as a tumor suppressor and others proposing the mutation renders it oncogenic. At the heart of the debate are the findings that IDH1 mutations are associated with better prognosis. While these findings represent a significant breakthrough, they remain to be validated to demonstrate a clear role for IDH1 and IDH2 in the etiology of this disease. Whether mutant IDH or products of its activity can be productively targeted for therapeutic intervention in glioma patients has yet to be determined. A better understanding of the biology of these gliomas will guide the development of new therapies to improve survival and reduce morbidity in these patients.

Our laboratory utilizes a robust somatic cell gene delivery glioma mouse model based on the RCAS/TVA system originally pioneered by Dr. Harold Varmus and Dr. Eric Holland.  In this model, the retroviral receptor TVA is expressed under the control of the Nestin promoter, which is active in neural and glial progenitors. This mouse model allows efficient and cost-effective modeling of gliomas because testing of a new gene simply requires generating a new retroviral vector not a new transgenic mouse. Because a new virus can be made and evaluated quickly (<4 weeks) in comparison to the time required to make a new transgenic mouse, we were able to quickly demonstrate a role for BRAF^V600E in the etiology of this disease. In addition, multiple genetic alterations can be introduced into the same animal without the cost associated with mating multiple strains of transgenic or knockout mice allowing us to assess cooperating events. Our group also extended the utility of this mouse model system by engineering the viruses to be responsive to doxycycline in the presence of Tet-off or Tet-on proteins. This allows us to regulate the expression of the delivered genes post-infection in vivo and define the role of specific genes in tumor maintenance. This genetic approach is extremely useful for target identification because pharmacological inhibition of a target can introduce additional variables in the experiment. If tumor growth is not affected following drug treatment it may be unclear if this is due to the target or the drug. We have successfully used this approach to assess the role of mutant KRAS in the context of active AKT and Ink4a/Arf deficiency. We plan to apply this approach to validate the genes identified by TCGA.

RCAS System

We use a series of replication-competent retroviral vectors based on Rous Sarcoma Virus (RSV), a member of the Avian Leukosis Virus (ALV) family, to study the role of different genes in tumor initiation and progression. RSV is the only known naturally occurring, replication-competent retrovirus that carries an oncogene, src. In the RCAS vectors, the region encoding src, which is dispensable for viral replication, has been replaced by a gateway cassette. Foreign genes inserted into this region are expressed from the viral LTR promoter via a subgenomic splice site (just as src is in RSV). RCAN vectors differ from RCAS vectors in that they lack the src splice acceptor; the gene of interest is inserted along with an internal promoter. Higher-titer viruses subsequently have been generated by replacing the RSV polgene with the pol gene of the Bryan strain of RSV. These vectors are termed RCASBP or RCANBP. The ability of these vectors to infect non-avian cells relies on expression of the corresponding receptor, TVA, on the cell surface.

The TVA receptor is typically introduced into mammalian cells (or mice) via an inducible and/or tissue-specific promoter. Therefore, this system allows for tissue- and cell-specific targeted infection of mammalian cells through ectopic expression of the viral receptor. Alternatively, when targeted infection of mammalian cells is not required (e.g., in cell culture), infection can be achieved through the use of non-avian envelopes, such as the amphotropic envelope from murine leukemia virus. The receptor for this envelope is endogenously expressed on almost all mammalian cells. We have used the RCASBP/ RCANBP family of retroviral vectors extensively in both cultured cells and in vivo systems for studies of viral replication and for cancer modeling. Most of these studies have analyzed gain-of-function phenotypes by delivering and overexpressing a particular gene of interest.

Recently, we engineered the RCASBP vector to reduce the expression of specific genes through the delivery of short hairpin RNA sequences in the context of an endogenous microRNA (miRNA). We also engineered the RCANBP vector to control the expression of the inserted sequences using the tetracycline (tet)-regulated system. Sequences inserted into this region are transcribed from a tet-responsive element and not the viral LTR. This virus allows inserted sequences to be turned on and off, in order to determine if their expression is required for tumor initiation, maintenance, and progression. The ability to turn off expression after tumors develop helps determine whether that gene or miRNA is a good target for therapy.