Cancer Genetics Broadens Its Quest
A revolution in cancer began almost 50 years ago when Dana-Farber researcher Frederick Li, MD, proposed that genes can raise the risk of inheriting cancer. The search then expanded to non-inherited (somatic) mutations that cause cancers to develop, and that drive metastasis, drug resistance, and relapse. Now the exploration has broadened from genes to whole genomes, including the vast expanses of DNA outside of genes, some times called 'dark matter' by analogy to cosmological studies of the universe.
Cancer genetics has launched a new therapeutic approach that targets a tumor’s genetic vulnerabilities, and a new way of classifying and treating cancers according to their molecular subtypes rather than the tissue in which they arise. For example, 14 different genetic mutations have been identified in lung cancer so far that have treatment implications. Thus, one lung tumor may have more in common molecularly with a breast tumor than with another lung tumor, with profound implications for treatment choices.
DF/HCC’s large and vibrant Cancer Genetics Program remains at the forefront of this revolution. “Our largest effort is to identify mutations that activate cancer-promoting oncogenes or deactivate genes that suppress tumors,” says David Kwiatkowski, MD, PhD (BWH), who with Levi Garraway, MD, PhD (DFCI), leads the program. “Our ultimate goal is to develop personalized cancer therapy for patients based on the genetic characteristics of their tumor.” To this end, the program is using ‘next gen’ DNA sequencing to rapidly and accurately characterize a patient’s cancer genetic subtype and guide the choice of the best therapy. Eventually, researchers may use cells and molecules circulating in the bloodstream to non-invasively monitor a cancer’s response to therapy and detect drug resistance, metastasis, or relapse – in real time.
Cancer genetics has had its most direct translation into clinical practice in the area of inhibiting oncogenic mutations – like EGFR, BRAF, ALK, and MEK – that promote the unbridled growth and proliferation of tumor cells. A growing roster of inhibitors, or targeted therapies, are either approved – like Gleevec (imatinib) and Tarceva (erlotinib) – or in clinical development.
But one common mutation has eluded inhibition: KRAS, which is mutated in about half of colon cancers and in many tumor types. “Figuring out how to target KRAS is one of the biggest challenges in cancer today,” says thoracic oncologist Jeffrey Engelman, MD, PhD (MGH). Engelman recently developed a new strategy that foils KRAS indirectly by double-teaming what it activates. Mutant KRAS activates the MEK growth signaling pathway. However, MEK inhibitors alone have not proven effective in KRAS mutated cancers, so another gene might be conspiring with MEK to evade cell death. Engelman and his colleagues decided to use MEK inhibition as a backbone for building a combination therapy.
They used a technique called RNA interference (RNAi) to silence the expression of different genes in different cells in an unbiased fashion. They grew these cells with and without exposure to a MEK inhibitor, reasoning that the cells that died with MEK inhibition had lost a MEK co-conspirator gene. This led to the BCL-XL gene, which researchers had previously shown allows cells to override apoptosis, the cell death pathway. Engelman’s team demonstrated in cell cultures and then in mice that inhibiting both MEK and BCL-XL was much more lethal to cancer cells than inhibiting just one or the other. In genetically engineered mouse models of human KRAS mutant cancer developed by Kwok Wong, MD, PhD (DFCI), the combined therapy demonstrated significant activity in shrinking tumors, providing greater support for the promise of this therapeutic approach.
Drugs that inhibit MEK (trametinib) and BCL-XL (navitoclax) are in clinical testing separately, which should accelerate the path toward a combination trial. If this strategy succeeds, it would be a big advance for the many cancer patients with KRAS mutations. Still, it may not benefit all KRAS mutant cancers equally. The combined inhibition worked better at shrinking tumors comprised of differentiated epithelial cells compared to those that displayed mesenchymal features. This distinction could provide a clinical biomarker for predicting which patients may respond to the combination – predictability that is key to personalized cancer therapy.
A Mixed and Evolving Population
Just as lung tumors, for example, have genetic heterogeneity, an individual tumor has cellular heterogeneity, with different cells playing different roles in growth, invasion, drug sensitivity, and relapse. David Langenau, PhD (MGH), is exploring this concept using zebrafish to visualize the dynamic process of tumor development and relapse. (See accompanying article Visualizing the Birth of Cancer in Zebrafish.) “In assessing genes associated with relapse in humans, we often compare the primary tumor to the relapsed tumor and assess how the whole bulk of the tumor is different at relapse,” he explains. “But each cancer cell could have different mutations at relapse, and we don’t know which mutations are drivers of relapse growth and which are passenger mutations that have no effect on tumor growth."
To gain a more linear view, Langenau transplanted a single leukemia cell into zebrafish and tracked its evolution into the full spectrum of leukemia cells over time. “Amazingly only a handful of genes – about 19 – seemed important to the evolution,” he says, and they were active in only a minority of tumor cells. Those cells may be tumor-propagating cells, with some properties of stem cells in their ability to differentiate into the other heterogeneous cell types of the tumor, and they may be responsible for tumorigenesis, metastasis, resistance, and/or relapse. Langenau is now testing these genes to see if they are potential targets for therapy in leukemia.
Dark Matter Does Matter
Until recently, most gene sequencing projects focused on protein-coding regions, to great payoff. For example, determining that the majority of metastatic melanomas had a mutation affecting BRAF led to a new targeted therapy (vemurafenib) for this aggressive disease. But genes actually account for only about 2% of the genome. With the increased power of whole genome sequencing technology, the importance of some of the non-coding ‘dark matter’ is coming to light. David Altshuler, MD, PhD (MGH, Broad), helped develop technologies to probe the whole genome for single DNA changes, called single nucleotide polymorphisms or SNPs, that influence hereditary human traits, ranging from height and weight to the risk of developing heart disease, diabetes, or cancer. Surprisingly, most of these risk variants were outside of genes, and many appeared to be in regulatory elements known as promoters and enhancers. Promoters essentially prepare genes for expression, while enhancers are dimmer dials for the level of expression.
How Enhancers Interact With Their Target Genes
Matthew Freedman studies enhancers, which are regulatory elements that often occur a long distance from the gene they regulate. Enhancers and genes are thought to interact when they come into contact because of how the DNA strands loop inside the nucleus. Chromosome Conformation Capture (3C) is a technique that determines which regions of the genome physically interact with each other, and then re-configures the DNA sequence into a format amenable to copying by polymerase chain reaction (PCR) for further study. Credit: Matthew Freedman.
As a postdoctoral researcher, Matthew Freedman, MD (DFCI), worked with Altshuler on whole genome sequencing, and now in his own lab he studies SNPs affecting enhancers. Freedman wants to connect cancer-risk enhancers to the genes they regulate. But whereas promoters can be found adjacent to the gene they regulate, enhancers can occur a long distance from the gene they influence, even on a different chromosome, which complicates the task.
However, enhancers interact with genes by coming into physical contact as DNA strands loop and twist in the nucleus, and a method called Chromosome Conformation Capture (3C) can visualize where a regulatory element touches the DNA loop, providing a clue of where to look in more detail. Freedman also compares the number of RNA transcripts of genes in cells with and without a SNP, looking for correlations that suggest a candidate gene for further study. Applying these methods to prostate and breast cancer, Freedman identified novel genes not previously associated with cancer risk. He hopes such work can provide new genes to target therapeutically, and insights into prevention for those with inherited cancer risks.
A Promoter Linked to Melanoma
Several years ago, Levi Garraway’s team had generated enough whole genome sequences of melanoma that they decided to look for cancer risk mutations in the non-coding regions of the genome. A very simple first pass identified two novel single-DNA changes. One or the other mutation occurred in 17 of the 19 samples they could examine in detail, and they each occurred in the same promoter. “We didn’t believe it at first because the mutations were so prevalent,” recalls Garraway. “We had imagined that we might someday find a non-coding mutation that was as important as a coding mutation, but together these were more prevalent than the BRAF mutation in melanoma.”
But it began to make sense. They were somatic mutations, not hereditary, and they involved a C to T change in DNA base pairs – typical of the ultraviolet light-induced DNA damage that is important to melanoma development. Also, the mutations occurred in a promoter that regulates the TERT gene. TERT produces part of the telomerase enzyme that maintains the ends of chromosomes, called telomeres. Telomeres normally shorten with each cell cycle, eventually causing the cell to undergo apoptosis. By activating telomerase and maintaining these tips, cancer cells become “immortalized” – able to divide indefinitely.
The two mutations were also common in a larger sample set Garraway analyzed later (and in some bladder and liver cancers). But is telomerase dysregulation a target for new therapeutic intervention? “Telomerase has not been a major focus of anti-cancer drug development,” says Garraway, “but perhaps we should re-evaluate the premise empirically in cancers with mutations affecting telomerase.” In general, discovering links between regulatory elements and cancer risk promises to generate many new leads for researchers to follow in the coming years.
Full Speed Ahead
When targeted therapies were first introduced, the approval process for each took years. Now an existing drug can gain approval for a new cancer with a specific mutation relatively quickly. However, it often takes a national effort to identify enough patients with rare mutations to perform clinical trials properly. That difficulty motivated 15 institutions to form the Lung Cancer Mutation Consortium three years ago, with Bruce Johnson, MD (DFCI) as a co-principal investigator.
“The new paradigm in clinical trails is a genotype-specific and cancer-non-specific trial,” says David Kwiatkowski, who is planning such a trial to test an existing drug, the mTOR inhibitor everolimus, in cancers with mutations in either of two genes (TSC1 and 2) that regulate mTOR, a master kinase. Kwiatkowski first discovered one of these genes as the cause of an inherited cancer syndrome, Tuberous Sclerosis Complex, and the mutations have been identified in bladder cancer, a type of sarcoma, and pancreatic neuroendocrine tumors. Now the Cancer Genome Atlas (TCGA) project (Matthew Meyerson, MD, PhD (DFCI), co-principal investigator) has discovered low frequencies of these mutations in many types of cancer. Kwiatkowski wants to see if everolimus can be repurposed for treating patients with any cancer that has mutations in TSC1 or TSC2.
Like Moore’s law in electronics, technological innovations are helping cancer genetics research progress exponentially. Thanks in part to DF/HCC’s involvement in large collaborations like the Lung Cancer Mutation Consortium and TGCA, its discoveries are being rapidly translated into the clinic.
Research detailed in this article was funded in part by NIH grants, including CA065164, CA12700, CA120060, CA137008, CA141457-01, CA090578, CA140594, CA122794, CA163896, HG006097-01, CA148537, CA131341, CA148065, CA009172, GM07753, OD002750, CA126674, CA120964, NS031535, AR055619-06, and CA154923-02.
— Cathryn Delude