Intratumor MAPK and PI3K signaling pathway heterogeneity in glioblastoma tissue correlates with CREB signaling and distinct target gene signatures.

Limitations in discovering useful tumor biomarkers and drug targets is not only due to patient-to-patient differences but also due to intratumor heterogeneity. Heterogeneity arises due to the genetic and epigenetic variation of tumor cells in response to microenvironmental interactions and cytotoxic therapy. We explored specific signaling pathway activation in glioblastoma (GBM) by investigating the intratumor activation of the MAPK and PI3K pathways. We present data demonstrating a striking preponderance for mutual exclusivity of MAPK and PI3K activation in GBM tissue, where MAPK activation correlates with proliferation and transcription factor CREB activation and PI3K activation correlates with CD44 expression. Bioinformatic analysis of signaling and CREB-regulated target genes supports the immunohistochemical data, showing that the MAPK-CREB activation correlates with proliferative regions. In-silico analysis suggests that MAPK-CREB signaling activates a pro-inflammatory molecular signature and correlates with a mesenchymal GBM subtype profile, while PI3K-CREB activation correlates with the proneural GBM subtype and a tumor cell invasive gene signature. Overall, the data suggests the existence of intratumor subtype heterogeneity in GBM and that using combinations of both MAPK and PI3K drug inhibitors is necessary for effective targeted therapy.

PI3K activation in neural stem cells drives tumorigenesis which can be ameliorated by targeting the cAMP response element binding protein.

Background: Hyperactivation of phosphoinositide 3-kinase (PI3K) signaling is common in cancers, but the precise role of the pathway in glioma biology remains to be determined. Some understanding of PI3K signaling mechanisms in brain cancer comes from studies on neural stem/progenitor cells (NSPCs), where signals transmitted via the PI3K pathway cooperate with other intracellular pathways and downstream transcription factors to regulate critical cell functions. Methods: To investigate the role of the PI3K pathway in glioma initiation and development, we generated a mouse model targeting the inducible expression of a PIK3CAH1047A oncogenic mutant and deletion of the PI3K negative regulator, phosphatase and tensin homolog (PTEN), to NSPCs. Results: Expression of a Pik3caH1047A was sufficient to generate tumors with oligodendroglial features, but simultaneous loss of PTEN was required for the development of invasive, high-grade glioma. Pik3caH1047A-PTEN mutant NSPCs exhibited enhanced neurosphere formation which correlated with increased Wnt signaling, while loss of cAMP response element binding protein (CREB) in Pik3caH1047A-Pten mutant tumors led to longer symptom-free survival in mice. Conclusion: Taken together, our findings present a novel mouse model for glioma demonstrating that the PI3K pathway is important for initiation of tumorigenesis and that disruption of downstream CREB signaling attenuates tumor expansion.

Investigating Neural Stem Cell and Glioma Stem Cell Self-renewal Potential Using Extreme Limiting Dilution Analysis (ELDA).

Glioma stem cells (GSC) grown as neurospheres exhibit similar characteristics to neural stem cells (NSC) grown as neurospheres, including the ability to self-renew and differentiate. GSCs are thought to play a role in cancer initiation and progression. Self-renewal potential of GSCs is thought to reflect many characteristics associated with malignancy, including tumor recurrence following cytotoxic therapy due to their proliferative dormancy and capacity to allow for the development of resistant tumor cell sub-clones due to mutations acquired during their differentiation. Here, we demonstrate that using extreme limiting dilution analysis (ELDA), subtle differences in the frequency of sphere-forming potential between PI3K-mutant oncogenic NSCs and non-oncogenic NSCs can be measured, in vitro. We further show how ELDA can be used on cells, before and after forced differentiation to amplify inherent differences in sphere-forming potential between mutant and control NSCs. Ultimately, ELDA exploits a difference in the ability of a single or a few seeded stem cells to self-renew, divide and form neurospheres. Importantly, the assay also allows a comparison between genetically distinct cells or between the same cells under different conditions, where the impact of target-specific drugs or other novel cancer stem cell therapies can be tested.

Expression of CD133 and CD44 in glioblastoma stem cells correlates with cell proliferation, phenotype stability and intra-tumor heterogeneity.

Glioblastoma (GBM) is a heterogeneous tumor of the brain with a poor prognosis due to recurrence and drug resistance following therapy. Genome-wide profiling has revealed the existence of distinct GBM molecular subtypes that respond differently to aggressive therapies. Despite this, molecular subtype does not predict recurrence or drug resistance and overall survival is similar across subtypes. One of the key features contributing to tumor recurrence and resistance to therapy is proposed to be an underlying subpopulation of resistant glioma stem cells (GSC). CD133 expression has been used as a marker of GSCs, however recent evidence suggests the relationship between CD133 expression, GSCs and molecular subtype is more complex than initially proposed. The expression of CD133, Olig2 and CD44 was investigated using patient derived glioma stem-like cells (PDGCs) in vitro and in vivo. Different PDGCs exhibited a characteristic equilibrium of distinct CD133+ and CD44+ subpopulations and the influence of environmental factors on the intra-tumor equilibrium of CD133+ and CD44+ cells in PDGCs was also investigated, with hypoxia inducing a CD44+ to CD133+ shift and chemo-radiotherapy inducing a CD133+ to CD44+ shift. These data suggest that surveillance and modulation of intra-tumor heterogeneity using molecular markers at initial surgery and surgery for recurrent GBM may be important for more effective management of GBM.

Sensitivity of GBM cells to cAMP agonist-mediated apoptosis correlates with CD44 expression and agonist resistance with MAPK signaling.

In some cell types, activation of the second messenger cAMP leads to increased expression of proapoptotic Bim and subsequent cell death. We demonstrate that suppression of the cAMP pathway is a common event across many cancers and that pharmacological activation of cAMP in glioblastoma (GBM) cells leads to enhanced BIM expression and apoptosis in specific GBM cell types. We identified the MAPK signaling axis as the determinant of cAMP agonist sensitivity in GBM cells, with high MAPK activity corresponding to cAMP resistance and low activity corresponding to sensitization to cAMP-induced apoptosis. Sensitive cells were efficiently killed by cAMP agonists alone, while targeting both the cAMP and MAPK pathways in resistant GBM cells resulted in efficient apoptosis. We also show that CD44 is differentially expressed in cAMP agonist-sensitive and -resistant cells. We thus propose that CD44 may be a useful biomarker for distinguishing tumors that may be sensitive to cAMP agonists alone or cAMP agonists in combination with other pathway inhibitors. This suggests that using existing chemotherapeutic compounds in combination with existing FDA-approved cAMP agonists may fast track trials toward improved therapies for difficult-to-treat cancers, such as GBM.

Coexpression analysis of CD133 and CD44 identifies proneural and mesenchymal subtypes of glioblastoma multiforme.

Accumulating evidence suggests that the stem cell markers CD133 and CD44 indicate molecular subtype in Glioblastoma Multiforme (GBM). Gene coexpression analysis of The Cancer Genome Atlas GBM dataset was undertaken to compare markers of the Glioblastoma Stem-Progenitor Cell (GSPC) phenotype. Pearson correlation identified genes coexpressed with stem cell markers, which were then used to build a gene signature that classifies patients based on a CD133 coexpression module signature (CD133-M) or CD44-M subtype. CD133-M tumors were enriched for the Proneural (PN) GBM subtype compared to Mesenchymal (MES) subtype for CD44-M tumors. Gene set enrichment identified DNA replication/cell cycle genes in the CD133-M and invasion/migration in CD44-M, while functional experiments showed enhanced cellular growth in CD133 expressing cells and enhanced invasion in cells expressing CD44. As with the 4 major molecular subtypes of GBM, there was no long-term survival difference between CD44-M and CD133-M patients, although CD44-M patients responded better to temozolomide while CD133-M patients benefited from radiotherapy. The use of a targeted coexpression approach to predict functional properties of surface marker expressing cells is novel, and in the context of GBM, supports accumulating evidence that CD133 and CD44 protein marker expression correlates with molecular subtype.