We focus on cancer

The focus of our research is cancer, one of the leading causes of death in the US and the world. Although people in their 50s and 60s are more likely to get cancer, children and young adults are not exempt from them and are often subject to long term side effects caused by treatments. Cancer is a genetic disease primarily associated with abnormal changes in DNA sequences and dosages (collectively referred to as mutations here). Most of the changes observed in a cancer genome do not result in uncontrolled cell growth. But if these changes occur in genes controlling for critical functions, cells start to behave erratically, disrupting the fine balance between growth and apoptosis (programmed cell death), eventually leading to a full-blown cancer. This process can take decades, for cancer cells have to identify the right “player” in order to highjack the usually tightly regulated molecular machinery, and to find ways to evade immune surveillance. Cancer cells do this through a trial and error approach – the 3 billion base pair long genome provides rich materials for these experiments. Along with each cell division, cells randomly acquire a few mutations in coding or non-coding regions of the genome. If these mutations do not do much harm to the cell, they will be cumulated or even enriched, and the process goes on. Eventually, one cell gets all the necessary “driver events” and gives rise to the neoplastic tissue mass that we call cancer.

What makes cancer so hard to treat? It is not hard to cure a cancer patient if the disease is well confined to one site, unless the location is inoperable (such as the brain stem). But once cancer cells start to invade the surrounding tissues, lymph nodes, or even spread to a distant organ (this is called metastasis), they become surgically impossible to eradicate. At this point, patients typically are given the options of chemotherapy, or molecularly targeted therapy, or some other forms of systemic therapy. Responses to these therapies vary greatly among patients, depending on their tumor’s genetic profiles. Nearly all patients develop resistance, because each of the millions of cancer cells is unique in their genetic makeup, and of them, one or a few are resistant and will take over the space left after most other cancer cells are killed by these therapies.

Is it possible to win the war against cancer, if they are genetically so heterogenous? The answer is yes. We now know cancer far better than 50 years ago, and we see great improvements in prognosis for some of the worst cancers seen before. However, we are far from winning the war. Resistance is a common issue, so is toxicity related to treatments. Some cancers such as pancreatic cancer and glioblastoma have such a dire outcome that patients diagnosed with these diseases are close to receiving a death penalty. This reality calls for continuous research in all fronts to understand the diseases and bring hope to patients.

We explore the cancer genome

Our main tool to study cancer is high throughput data/assay, mostly DNA/RNA sequencing and proteomics. These types of data give us the tools to measure every gene of a tumor, and sifting them through may allow us to identify driver genes, stratify molecular subtypes, and model significant events in cancer.

Example 1: mining the genome of glioblastoma. Glioblastoma, a.k.a GBM, is a deadly brain tumor affecting ~10,000 people annually in the US. It is one of the cancers that did not see much outcome improvement in the last decade. On average, patients live with the disease for only 15 months, despite surgery, radiation, and chemo. Around 2012, we launched a study to investigate the copy number landscape of GBM hoping to find a gene that has the similar genetic change to EGFR vIII, a very important signaling gene in controlling cell growth. Although we did find infrequent events like that, a surprising finding was frequent double minutes encompassing thus co-amplifying two oncogenes CDK4 and MDM2, both closely localized to chr12q (Zheng et al. Genes Dev, 2013). This phenomenon was later confirmed by other groups and was thought to result from chromothripsis.

Example 2: integrative analysis of adrenocortical carcinoma. This disease, short named ACC, is so rare that even its annual incidence couldn’t be accurately calculated (rough estimate is 1 of a million people). Because of its rarity, few progress has been made therapeutically. The only FDA approved drug for ACC is mitotane, an adrenal gland toxin (originally a pesticide). Since 2013, I worked with a group of disease experts (particularly closely with Tom Giordano and Gary Hammer from Univ Michigan), to dissect the molecular events and tumor heterogeneity of ACC (Zheng et al. Cancer Cell, 2016). Highlights of our findings include a new driver gene ZNRF3, a genomic event (whole genome doubling) commonly seen in ACC, telomere abnormalities during disease progression, and three consensus subtypes with distinct prognosis and genetics. This work has inspired efforts to develop assays to customize patient treatment plans.

Example 3: telomeres and telomerase. Most cells of our body are not immortal – they will stop proliferating once reaching a certain number of divisions (Hayflick limit). This is an intrinsic check mechanism preventing cancer development. What enables this check are telomeres and telomerase. The former is a repetitive segment of the hexamer TTAGGG, and erodes with every cell division. The latter, silent in somatic cells, is an enzyme that can add new TTAGGG to telomeres to either maintain or elongate it. Very short telomeres trigger senescence and crisis, as happened in most cells. However, cancer cells evade this fate by reactivating telomerase through expression of its coding gene TERT. Since 2015, we were intrigued by this telomere/telomerase tug-of-war, and designed a systematic study to measure telomere lengths, and enumerated what may help cancer cells regain TERT expression (Barthel et al. Nat Genet, 2017). The most important insight from this study was that promoter methylation may complement mutation in a tissue dependent manner in reactivating TERT transcription.

Other work. Over the years, we developed numerous tools that have received community attention. Before 2011, I developed network based tools including dmGWAS and GenRev. At MD Anderson, I helped develop the fusion calling tool PRADA and TCGA fusion portal. I was also involved in many collaborations. Our most recent work evaluated panel sequencing in the clinical management of glioma (Zheng, Mol Cancer Ther, 2019).

Where are we headed

Despite our broad interest in understanding the cancer genome, we pay special attention to telomeres and telomerase, and are continuing our exploration of their roles in cancer. They are not only key to cancer cell immortality, but also are intricately linked to genomic instability, most notably aneuploidy. How is telomerase activity regulated? Given its exclusive activity in cancer cells, why has its inhibition not shown satisfactory clinical response? The idea of targeting telomerase remains valid, perhaps it is our targeting strategy that should take the blame. Rather than directly targeting telomerase, can we identify indirect approaches that can overcome our limitations? From a computational perspective, it is imperative to develop methods that can accurately predict telomerase activity, a first step to unsupervised search for novel telomerase regulatory mechanisms that would enable indirect targeting. If we can successfully develop such a method , it would allow us to characterize telomerase activity in tissue development and cancer, and find molecular correlates that will provide new insights into the regulation of this critical complex.

We are also developing new tools that can help our bench colleagues tackle pediatric cancer. Cancers in children pose a unique challenge. They are usually more amenable to treatments, but are unmistakably lethal once first line treatment fails. Further, children are often subject to long term side effects. One initiative we are undertaking is developing a gene expression data based portal. This portal will enable users to navigate multi-modal molecular data and associate them with gene expression. This design is tailored for pediatric cancer because they present much lower genetic alterations (mutation and copy number change) than those in adults. We are working closely with our colleagues to build a tool not only useful to our institute but also to the whole research community.