Our research direction originated when we first identified a spontaneously occurring mutant EGF receptor in glioblastoma. Known as EGFRvIII (pronounced “E-G-F-R-v-three), this molecule represents a deletion of exons 2 through 7 in the extracellular domain of the EGF receptor. This removes 273 amino acids and creates a novel glycine at the fusion junction. This mutant is expressed on tumor tissues, and only very rarely found in any normal tissue, making it an exceptional target for anti-cancer therapeutics. We have engaged in both very basic studies on the signal transduction pathways initiated by EGFRvIII, as well as translational work to create diagnostic tools and therapies around EGFRvIII.

EGFRvIII is constitutively active thus it does not require the ligand EGF for activation. Our work demonstrates that it also highly activates the PI3K pathway, where the wildtype EGFR is better known for its activation of ERK. Our efforts on signaling pathways have discovered that glioblastoma tumors preferentially utilize the JNK pathway, unlike many other tumors that use the ERK pathway. More specifically, only one of the 12 known JNK isoforms is highly expressed, JNK2a2. Co-incidentally, we have uncovered that this isoform is constitutively active and that it has a specific activation domain. JNK2a2 also upregulates TGF-a, a ligand for the EGF receptor that is frequently overexpressed in glioblastomas. Our additional basic science studies to understand EGFRvIII signaling led us to discover the Gab1 docking protein. Originally discovered as a docking protein substrate for both EGF and Insulin receptors, it is now known as a substrate for multiple receptors vital for growth factor and cytokine pathways in diverse phenotypes such as cell survival, the cellular response to stress, and the cellular entry of parasites.

Our translational work has encompassed both diagnostics and therapeutics. We have developed antibodies that specifically recognize EGFRvIII and not EGF receptor and have also developed rapid RT-PCR based assays to detect EGFRvIII. Using these tools, we have discovered that expression of EGFRvIII is not limited just to glioblastoma tumors but it is also found in breast, lung, prostate and ovarian tumors. Despite early studies that have shown that the rearranged and amplified EGFR gene that results in EGFRvIII protein is seen in most cells in GBM tumor samples, our more recent expression-based studies have shown the intriguing phenotype of sporadic/focal expression, including expression in tumor vessels.

We were the first to show that a peptide vaccine based on the cancer specific epitope found in EGFRvIII is an effective anti-tumor drug (ALT-110, CDX-110, Rindopepimut). In animal models, this vaccine can both prevent tumors and induce the regression of existing tumors. We also played a major role in initiating the first clinical trial in humans using this peptide vaccine. Three trials have now been completed with highly interesting results. A Phase II clinical trial at Duke and MD Anderson for glioblastoma patients has shown median survival of 30 months vs. 14 months for conventional therapy in a matched cohort. An expanded Phase II/III trial for brain tumors using this vaccine is currently enrolling which is sponsored by Celldex Therapeutics.

The interesting discovery of focal/sporadic EGFRvIII expression in primary GBM samples prompted us to investigate the possible relationship of this sporadic expression with the cancer stem cell (CSC) theory. CSCs have some essential properties in common with stem cells in that they are capable of self-renewal and are toti-potent (tumor potent). Additionally, cancer stem cells can express stem cell markers such as CD133, differentiation is progressive and cell division can be asymmetric, and they create a cellular hierarchy. Numerous cancers have been suggested to contain CSCs including colon, breast, and glioblastoma tumors.

The CSC theory suggests that not all cells in the tumor mass are causative for the cancer, and brings to light the need to target a specific, relatively small stem cell like subpopulation to prevent tumor progression. There are a few characteristics of CSCs that are different from stem cells and are important to understand. Cancer stem cells do not necessarily arise from normal stem cells, and likewise, differentiated cells can revert to CSC like state. CSCs can arise independently in a tumor mass, and the presence of markers can identify CSCs, but they are not necessarily causative.
If CSCs are truly the causative element in tumors, then one would expect them to harbor key genetic tumor forming mutations seen in the tumor as well as genes for self-renewal. Our most recent work focused on the hypothesis that EGFRvIII may be found in GBM CSCs. Initial work identified a single rearranged band corresponding to EGFRvIII in GBMs, suggesting that it arise from a single clone.

Flow cytometric, RT-PRC, and immunofluorescent analysis of primary and cultured human GBM tumor spheres indeed identified the presence of EGFRvIII expression, which showed a significant co-expression with the stem cell marker CD133. The addition of EGFRvIII expression along with CD133 characterized a sub-population that had the highest rate of tumor sphere formation, the highest self renewal, and the highest tumor formation in mice.

Additionally we found that EGFRvIII expression is associated with markers of stem cells such as SSEA-1 and Nestin, but is not associated with markers of differentiation such as TUJ-1, Vimentin, and GFAP. A very interesting discovery that is still under investigation is the finding that in GBM CSCs, there is a significant lack of co-expression of EGFRvIII with the wildtype EGFR.

These discoveries prompted us to generate a reagent targeting this population, with hopes that it might pave the way for improved treatment of GBM, either alone or in combination with other therapies. We developed a bi-specific antibody (BsAb) that targets both EGFRvIII and CD133, and is capable of activation of ADCC. Preliminary studies demonstrated that this reagent is effective in dual epitope expressing cells, but does not affect single epitope, or non-expressing cells.

Furthermore, this reagent inhibits EGFRvIII/CD133 tumor sphere tumors in mice, as well as prolonging survival.