Spotlight: Altering the fingerprint of cancer
Image courtesy of Bang Wong, ӳý
It has become clear over the past few decades that cancer is a genetic disease, with vast combinations of gene mutations, translocations, additions, and deletions contributing to the diverse cancers that afflict humans. Given this knowledge, a sea change is brewing in the search for new cancer treatments. In the past, the guiding principle behind most cancer drugs was to poison rapidly dividing cells, which include tumor cells but also healthy cells too. Now, rather than searching for cytotoxic therapies that kill cells somewhat indiscriminately, researchers are taking cues from the forensic practice of fingerprinting in the quest for therapies that target cancers by their distinct molecular properties. Empowered by new genomic tools and the ability to analyze collections or “libraries” of chemicals in high capacity, we are now working to identify novel, more effective drugs for previously intractable cancers. We are doing this by taking advantage of cancers’ unique patterns of gene expression, known as genetic signatures.
As a pediatric oncologist who treats children with cancer, I recognize the painfully obvious fact that better treatments are needed. Although significant strides have been made in the last fifty years, there is still more room for improvement. For example, although progress has been made in the treatment of acute myelogenous leukemia (AML), the majority of these patients eventually relapse, and current treatments trigger toxic side effects in patients. For some diseases, such as osteosarcoma, the most common bone tumor in children and young adults, little has changed in therapy during the last two decades. A primary reason is that these diseases are rare. Thus, there is a reduced market incentive for the pharmaceutical industry to develop pediatric drugs and a general paucity of research funds to study these rare conditions. Moreover, even with drug in hand, clinical trials in pediatric and other rare cancers are difficult. A rare disease incidence means fewer patients enter the trials, necessitating larger cooperative groups to conduct these studies. Furthermore, safety concerns have resulted in a reluctance to test new drugs in children.
Although new treatments developed specifically for pediatric oncology are hard to come by, progress has been made in understanding the molecular basis of childhood cancers. For example, over a decade ago, researchers identified a genetic translocation that appeared in most patients with Ewing sarcoma, a primary bone tumor seen predominantly in children and adolescents. The translocation results in an abnormal fusion of the EWS and FLI proteins. Because FLI normally acts as a transcription factor — a “master switch” that controls where and when certain genes are active — it is thought that the union of EWS and FLI promotes cancer by inappropriately switching on or off genes that are critical for tumor formation. Despite this molecular knowledge, the current therapy choices for Ewing patients consist of combinations of non-targeted cytotoxic agents. This therapy is curative for roughly 70% of patients with disease that has not yet spread, but also comes with significant side effects. However, for patients with metastatic disease or for those who relapse, these conventional chemotherapy drugs are often not effective.
Paradoxically, a major stumbling block in identifying new targeted treatments for Ewing sarcoma stems from its own molecular origins. Transcription factors such as FLI are not well suited for traditional high-throughput chemical screening methods, in which thousands of chemicals are tested in parallel for their ability to interfere with a protein of interest. Because of this, many cancer-related proteins, including EWS/FLI, are considered “undruggable,” with no obvious inroads to pharmacologically inhibiting their function.
To overcome these many challenges to drug discovery, I have been working with Todd Golub and colleagues in the ӳý’s Cancer and Chemical Biology Programs and at Dana-Farber Cancer Institute to develop a genomic method for discovering compounds that could serve as research tools or new potential therapies. While conventional screening methods center on a specific cellular state or “phenotype” (e.g., cell death) or a specific molecule within cells, our approach focuses instead on how genes are expressed in cancerous and non-cancerous cells. Just as fingerprints give a much more complex and unique measure of a person’s identity than height or weight, a pattern of gene expression can provide richer information about a cell’s reaction to certain compounds than simply determining if the cell lives or dies. As such, this approach enables the kind of small molecule screening that was previously considered impossible, such as modulating the effects of transcription factors, and does not require a priori knowledge of a specific molecular target.
Kim StegmaierPhoto by Maria Nemchuk
Our method of gene expression-based high-throughput screening (GE-HTS) uses these genetic signatures in an attempt to discover new potential cancer therapies and tool compounds with which to dissect cancers’ molecular origins. We begin by identifying genes whose expression distinguishes different cellular states; for example, we compare cancerous versus normal cells, abnormal versus normal transcription factors, and active versus inactive kinases. We then screen a library of compounds to identify ones that turn the cancer cell’s gene signature into that of the desired state.
The GE-HTS method is powerful because it requires little “behind the scenes” knowledge. Forensic scientists matching a fingerprint found at a crime scene to one in a database do not need to know how or why the evidence was left, merely that the two prints are statistically indistinguishable. Similarly, we need not know a particular compound’s molecular target or the mechanism by which it induces a normal gene expression signature to identify it as a new potential therapy.
We first developed and validated GE-HTS by studying acute myelogenous leukemia (AML), a malignancy believed to result from defects in both cell proliferation and maturation. While most existing therapies for AML focus on killing cancer cells, we decided to follow a different approach — coaxing the cells to differentiate and hence, to stop proliferating abnormally.
Without knowing the complete biological basis of AML, we compared the patterns of gene expression in AML cells to those of mature, normal white blood cells. A compound that induced a normal, mature gene expression signature in AML cells could represent a potential treatment in patients. Eight compounds out of our screen of nearly 2,000 compounds fulfilled this criterion, which we then confirmed through functional studies. One compound identified as a result of this work is now in clinical trial for patients with relapsed or refractory AML.
In society, fingerprints have uses beyond forensic science, such as to search for lost children and control access to computers. One of the great strengths of the GE-HTS method, too, is its generic nature. Expression-based screening can identify molecules that affect a wide variety of biological states, which would have each required a specialized assay if we took the traditional phenotype-based screening approach.
An illustration of this power comes from our application of the GE-HTS method to studies of Ewing sarcoma. By replacing our original detection system with one employing fluorescent beads, we increased our ability to detect more complex gene expression signatures. Instead of reading signatures composed of only a handful of genes, we can now analyze signatures composed of 100 genes. (The next-generation technology promises to push that number up to 500.) With this enhanced capability, we set out to identify compounds that might reverse the signature of EWS/FLI activity and potentially block the fusion protein’s cancer-promoting effects in Ewing sarcoma cell lines.
We focused our initial effort on a collection of compounds enriched for drugs previously approved by the FDA for treating a variety of diseases. We hoped that an existing drug might prove to be a potential therapy for Ewing sarcoma, because that could help expedite the clinical testing process. Ironically, the top hit from our Ewing screen turned out to be cytosine arabinoside (Ara-C), a medication currently used to treat AML. Ara-C induced an expression fingerprint of “inactive” EWS/FLI, reduced levels of the EWS/FLI protein, and reduced the viability and cancer-like behavior of Ewing sarcoma cells.
We are now beginning a phase II clinical trial to evaluate the effects of Ara-C in children with relapsed or refractory Ewing sarcoma. Because the drug is routinely used to treat children with leukemia, the testing of Ara-C in children with Ewing sarcoma was greatly expedited. That is because safe regimens for using the drug in children had already been established, cutting years off the time from identification of the compound to actual clinical trial. In addition to testing Ara-C’s ability to reduce tumor size, in the trial we’ll also determine whether the drug can reduce EWS/FLI transcript levels in peripheral blood.
In the meantime, our group is beginning to explore the application of GE-HTS to the differentiation of other pediatric cancers, such as neuroblastoma and osteosarcoma, and to the modulation of other abnormal transcription factors, such as AML1-ETO, Notch1, and MLL rearrangements. One of the biggest challenges we’re facing now is how to pinpoint the proteins that are targeted by the compounds identified in our screens. Indeed, discovering a compound that could benefit patients is certainly important, but an in-depth understanding of how it works can lead to a better understanding of the underlying disease and to improved drugs. Hopefully, the story about how we successfully tackle this problem will be chronicled in a future Spotlight.
Further reading:
Stegmaier K, Wong JS, Ross KN, Chow KT, Peck D, Wright RD, Lessnick SL, Kung AL, Golub TR. (2007) . PLoS Medicine; 4:702-714. DOI:10.1371/journal.pmed.0040122
Peck D, Crawford ED, Ross KN, Stegmaier K, Golub TR, Lamb J. (2006) . Genome Biology; 7:R61. DOI:10.1186/gb-2006-7-7-r61
Stegmaier K, Ross KN, Colavito SA, O'Malley S, Stockwell BR, Golub TR. (2004) . Nature Genetics; 36:257-263. DOI:10.1038/ng1305
