Researchers had every reason to expect that a compound called BKM120 (also known as buparlisib) would stifle glioblastoma brain tumors lacking the protein PTEN. After all, it was known to block the tumor-promoting PI3K protein and could easily pass through the blood-brain barrier — the dense layer of cells that guards entry to the brain — to make its way to glioblastoma tumor cells.
When the compound was tested in animal models of glioblastoma tumors, however, it was something of a letdown, prolonging the animals’ survival only modestly. The reason, scientists at Dana-Farber discovered in a new study, is that BKM120 is not the all-purpose PI3K blocker it was thought to be: instead of thwarting two subtypes of PI3K equally, it shackles one but barely restrains the other. The effect is to deprive glioblastoma cells of one aspect of their cancerous nature while giving almost free reign to another.
The finding, reported in a paper in the journal Cell Reports, underscores the challenge of developing a drug for PTEN-deficient glioblastoma that crosses the blood-brain barrier and targets both subtypes of PI3K. Existing compounds that cross the barrier target either subtype, but not both; and those that target both subtypes can’t make it past the barrier.
“Glioblastoma is the most common and most aggressive brain tumor in adults, but current treatments extend patients’ lives for a rather limited time,” says study lead author Jean Zhao, PhD, of Dana-Farber and the Broad Institute of Harvard and MIT. “Better treatments depend on a fuller understanding of the biology of the disease, particularly the molecular abnormalities that underlie it. In this study, we focused on glioblastoma marked by a loss of PTEN — something that occurs frequently in this disease and is associated with rapid growth and metastasis.”
A common loss
PTEN is a tumor-suppressor protein whose loss occurs in virtually every type of cancer. It is part of the same cell-growth pathway as PI3K: PI3K switches the pathway on, PTEN shuts it off. With PTEN absent, the normal controls on tumor cell growth slacken.
To counter the loss of PTEN, researchers have attempted to quash the “go” part of the growth pathway, developing and testing drugs that target PI3K. BKM120 looked like it might be the perfect candidate: not only could it slip through the blood-brain barrier but it was also known as a pan-PI3K inhibitor, able to block all four “isoforms” of PI3K, dubbed p110α, p110β, p110γ, and p110δ. (Isoforms are structurally similar proteins that originate from the same gene or gene family. In glioblastoma, p110α and p110β are the PI3K isoforms that need to be blocked.)
To probe glioblastoma at the molecular level, and tease out the roles of PI3K isoforms, Zhao and her colleagues created lines of neural stem cells that lacked the genes for PTEN and the fellow tumor-suppressor p53. Allowed to grow in laboratory cultures and then in animal models, the cells grew into high-grade glioblastoma tumors, providing useful models of the disease.
When researchers used genetic approaches to silence either the p110α or p110β isoform of PI3K, the animals still developed glioblastoma but more slowly than those that were untreated. When both isoforms were blocked, the animals didn’t develop glioblastoma tumors.
“To prevent glioblastoma or stop it from progressing, it appears to be necessary to shut down both PI3K isoforms,” Zhao says. “We next explored whether the alpha and beta isoforms are responsible for different aspects of tumor growth.”
Glioblastoma tumors normally have indefinite edges: instead of a clear-cut margin between tumor and normal tissue, tumors tend to blur into their surroundings. When researchers deleted the p110α isoform in animal models, the resulting glioblastoma tumors still had messy borders but the tumors themselves were noticeably small. When the p110β isoform was deleted, the tumors were full-sized, but their borders were distinct.
The implication was that p110α is responsible for tumor cell proliferation and p110β is responsible for migration, the ability to penetrate adjacent tissue. An effective drug for glioblastoma would need to target both.
When investigators treated the animal models with BKM120, which seemed to have all the qualifications for success against glioblastoma, the animals lived longer than they would have otherwise, but not by much. The compound’s reputation as a pan-PI3K inhibitor was, it turns out, unfounded.
“We found that BKM120 inhibits the p110α isoform far more than it does the p110β isoform,” Zhao says. “At the concentrations tested, it acts as a selective inhibitor for p110α. This explains its failure to bring glioblastoma under control.
“Our findings underscore the need to develop a true pan-PI3K inhibitor for glioblastoma or, alternatively, an inhibitor of p110β that can be used in combination with BKM120.”
The lead author of the study is Shaozhen Xie, PhD, of Dana-Farber. Co-authors are J. Ricardo McFaline-Figuero, MD, PhD; Yanzhi Wang, Roderick Bronson, DVM, Keith L. Ligon, MD, PhD; Patrick Y. Wen, MD; and Thomas M. Roberts, PhD, of Dana-Farber.
About the Medical Reviewer
Jean Zhao received her PhD from Tufts University School of Medicine in 1999. She did her postdoctoral work in the laboratory of Dr. Thomas M. Roberts, became an Instructor in Medicine at Harvard Medical School in 2003, and joined the faculty of DFCI and Harvard Medical School in 2006.