Meeting report – 9th IRIC International Symposium on Molecular Targets in Cancer Genomics

In normal cells, genomic integrity is controlled through four main mechanisms: high-fidelity DNA replication, faithful chromosome segregation during mitosis, error-free DNA repair machinery and well-coordinated cell cycle progression. However, genomic instability is one of the most commonly observed hallmarks across all cancer types and it favours the appearance of cancer-enabling alterations (Hanahan and Weinberg, 2011). These alterations range from simple point mutations to the loss or gain of entire chromosomes. In this session, invited speakers demonstrated how the identification of key players implicated in genome integrity maintenance can contribute to the development of novel and effective anti-cancer therapies.

In cells, genomic integrity is maintained by the coexistence of a high-fidelity DNA replication machinery and efficient DNA repair mechanisms. Although multiple DNA polymerases are required to replicate the genome, their distinct roles in DNA replication is not known. To answer this question, Thomas A. Kunkel (National Institute of Environmental Health Sciences, Research Triangle Park, USA) used hydrolytic end sequencing (HydEn-Seq) to study the incorporation of ribonucleotides during replication in different yeast strains, each mutated for a specific DNA polymerase. By demonstrating that ribonucleotides were incorporated in a polymerase- and strand-specific fashion, his team was able to conclude that DNA polymerase α and δ are responsible for the lagging strand synthesis, whereas DNA polymerase ε is mainly involved in the leading strand synthesis. This work contributes to our understanding of fundamental processes underlying the maintenance of genomic integrity and how their disturbance can lead to genomic instability in cancer.

Telomeres are specific DNA sequence repeats whose maintenance is essential to protect chromosome ends and prevent aberrant chromosome fusion events. Human somatic cells often do not express sufficient levels of telomerase reverse transcriptase (TERT), an enzyme responsible for telomere maintenance during DNA replication, which leads to a gradual shortening of telomeres and eventually cell death or cell arrest. In contrast, murine stem cells express TERT, and telomere shortening does not occur. Lea Harrington (IRIC, Université de Montréal, Montréal, Canada) presented recent findings on the unpredicted role of telomere integrity in stem cell differentiation. In embryonic stem cells (ESCs), enforced telomere shortening through disruption of TERT induces repression of DNA methyltransferase (DNMT), leading to a genome-wide hypomethylation (Pucci et al., 2013). This further alters the patterns of the repressive histone modification H3K27 trimethylation at distant genetic loci. Such drastic changes in the epigenetic landscape of ESCs alter their ability to repress pluripotency factors such as Nanog, which are necessary for stable differentiation. This finding suggests that short telomeres have a role in cell fate that extends beyond the telomeres themselves.

A genetic interaction between two genes is termed ‘synthetic lethal’ if the mutation or disruption of both genes results in loss of cell viability (whereas the loss of function at one locus is not lethal). Similarly, small molecules, through their action on a particular genetic network, can also display synthetic lethality. The elaboration of chemical–genetic and genetic–genetic synthetic lethal interactions (SLIs) in the budding yeast Saccharomyces cerevisiae has proven to be a powerful tool to dissect the mechanisms of gene action and drug resistance in diseases such as cancer. Philip Hieter (University of British Columbia, Vancouver, Canada) presented the systematic identification of SLIs that are related to genome instability, which found a high degree of conservation in the SLIs that affect chromosome instability between humans and yeast (van Pel et al., 2013a,b). Several interesting interactions emerged, including SLIs between the FEN1 flap endonuclease and genes whose homologs are frequently mutated in colorectal cancer, such as MRE11A and CDC4. Interestingly, small molecule inhibition of FEN1 in cancer cells with mutated MRE11A or CDC4 recapitulated the SLI phenotype identified in yeast (van Pel et al., 2013a). This result emphasizes the importance of model organisms in the identification of clinically relevant synthetic lethal interactions in human tumors.

Concerning the elucidation of drug resistance mechanisms using synthetic lethality, Angelos Constantinou (Institut de génétique humaine, Centre national de la recherche scientifique, Montpellier, France) presented data on camptothecin (CPT), an inhibitor of topoisomerase I that interferes with the progression of replication forks. CPT resistance in cancer cells is mediated by multiple pathways, which help overcome replication impediments. The systematic identification of proteins recruited at replication forks unveiled new determinants of the anti-proliferative capacity of topoisomerase I inhibitors. Collecting this type of information on synthetic lethality will help identify genetic landscapes that are targetable by existing drugs or play a role in chemotherapeutic resistance. This will guide new drug design and combinational therapies that synergistically target cancer-specific pathways and compensatory mechanisms.

Epigenetic regulators include different enzymes or enzyme complexes that modify DNA or histones without altering the genetic sequences, resulting in diverse modifications encompassing DNA methylation and histone modifications (methylation, acetylation, phosphorylation and ubiquitylation). These specific epigenetic marks play important roles in modulating transcriptional activity, DNA replication and repair in normal cells. Epigenetic aberrations often lead to expression of various cancer oncogenes and cancer development, making them valuable diagnostic indicators as well as potential therapeutic targets. Unlike genetic alterations, epigenetic modifications are potentially reversible, and drugs targeting epigenetic regulators could allow the reversion of a malignant population to a more normal state. Several successful epigenetic therapeutics (‘Epi-drugs') have already been approved for cancer therapy, including inhibitors against DNA methyltransferases (DNMTi) and histone deacetylases (HDACi), encouraging new research and development of epigenetic modulators.

The distinct and recurring patterns of histone modifications directing gene expression suggests that chromatin–DNA interactions are guided by combinations of histone modifications (the ‘histone code’). The complexity of this histone code is exemplified by the fact that some histone modifications serve to recruit other proteins by specific recognition of the modifications, which alter chromatin structure and elicit downstream biological processes. Many chromatin-associated proteins harbor multiple chromatin recognition domains or are in macromolecular complexes containing different functional proteins. Brian Strahl (University of North Carolina at Chapel Hill, Chapel Hill, USA) discussed ubiquitin-like PHD and RING finger domain-containing protein 1 (UHRF1), which is a unique chromatin effector that recognizes both histone methylation and DNA methylation. Using a histone peptide array containing more than 7000 combinatorially biotinylated modified histone peptides, he showed that the tandem Tudor and PHD domains (both methyl-lysine binding) operate as a functional unit, providing a defined combinatorial readout of a specific heterochromatin signature (H3K9 methylation) within a single histone H3 tail. This multivalent engagement of histone H3 by UHRF1 is essential to guide downstream DNA methylation in human cells. UHRF1 is often upregulated in human cancers, underscoring its functional importance and therefore its consideration as a therapeutic target. Another epigenetic regulator considered as a potential therapeutic target is a subcomplex of the SAGA coactivator and was the topic of the talk by Cynthia Wolberger (Johns Hopkins University, Baltimore, USA). The SAGA coactivator complex is a 1.8-MDa, macromolecular histone-binding complex that modulates transcriptional activities by carrying out multiple functions, including histone acetylation and deubiquitylation. Histone H2B lysine 123 monoubiquitination (H2BK123ub) is a mark for active transcription and occurs upstream of H3K4 di- and tri-methylation (another prominent mark for active transcription). Cynthia Wolberger showed beautiful structural data demonstrating that a subcomplex of SAGA called the DUB module (DUBm) has high substrate specificity for H2BK123ub and physically interacts with only H2A–H2B within a nucleosome (Samara et al., 2010). Detailed structure–activity analysis of human DUBm could guide the design of specific small-molecule inhibitors against DUBm and might lead to development of alternative cancer therapies to target tumors with an overexpression of the SAGA complex.

Stephen Frye (University of North Carolina at Chapel Hill, Chapel Hill, USA), who has extensive experience in both pharmaceutical industry and academia, gave a practical guide for drug discovery with examples of using chemical probes to target epigenetic regulators. His focus of research are the so-called methyl-lysine (Kme) ‘readers’, which form a class of epigenetic regulators that bind Kme and have crucial roles in oncogene transcriptional regulation. At least 200 Kme reader domains have been described and most contain a well-defined ‘Kme-binding pocket’ made up of aromatic amino acids. Only a few labs have focused on this novel class of targets and no compounds have advanced to clinical testing at this stage. Stephen Frye used computational modeling and molecular dynamics simulations to investigate the binding mode of histone peptides to Kme readers, as well as in silico screening to design potential ligands for the CBX chromodomains in the polycomb and heterochromatin protein 1 complexes that participate in gene repression. He then showed results from in vitro chemical screens followed by cell-based assays to characterize the mode-of-action of synthesized compounds. This target-class strategy has so far yielded a number of new chemical probes that target Kme readers and these compounds are freely available for the academic community. Chemical probes that elicit desired downstream cellular effects could validate novel targets and serve as a chemical scaffold starting point for further medicinal chemistry development in the context of new anti-cancer epigenetic therapeutics.

Over the past decades, scientists have built a binary view of the human genome in which 2% is functional, for example, protein-coding genes, whereas the remaining 98% is considered as junk DNA. But given the fact that 70% of the genome is transcribed and that 88% of disease-causing single nucleotide polymorphisms identified by genome-wide association studies lie in intronic and intergenic regions, this argues that ‘junk DNA’ has a functional role (Djebali et al., 2012; Edwards et al., 2013). Since these observations were made, the expanding family of so-called non-coding RNAs has been implicated in a wide variety of cellular processes ranging from regulation of epigenetic marks to transcript stability. The speakers in this session presented techniques to investigate functions of non-coding RNAs and discussed how this knowledge can be used to design cancer treatments.

The discovery of non-coding RNAs has induced a paradigm shift in cell biology: RNA is now considered as a functional unit rather than just an intermediate between DNA and protein. To allow the genome-wide functional annotation of coding and non-coding regions, Michael T. McManus (University of California San Francisco, San Francisco, USA) presented two ‘CRIPSR-Cas9 2.0’ libraries. In the creation of these libraries, each gene-specific guide RNA was coupled to a unique barcode and used to repress or upregulate gene transcription through heterochromatin formation or a unique activation system. Following deep-sequencing of drug-treated and untreated cells, genes involved in drug resistance could then be identified by computing the barcode enrichment for their respective guide RNA. These libraries represent a powerful tool to unravel the role of coding, as well as non-coding, transcripts to answer scientifically fundamental and clinically relevant questions.

RNA base pairing is at the basis of RNA function, and although this occurs naturally in the cell it can also be assisted by protein interactions. Sarah Woodson (Johns Hopkins University, Baltimore, USA) demonstrated how the bacterial stress response protein Hfq binds to regulatory RNAs in a complex with their target RNAs to facilitate base pair exchange by eliciting conformational changes in the tertiary structure of the RNA (Panja et al., 2015). These conformational changes place complementary strands on a positively charged arginine patch on the protein surface, which speeds up their base pairing. A better understanding of how proteins chaperone RNA structure and molecular interactions is particularly important as more RNA-based therapeutics are moving to the clinic.

With their altered expression pattern in most cancer types, microRNAs (miRNAs) appear to be important players in oncogenesis. In fact, they can have both oncogenic and tumor suppressor roles. For example, miR-155 is overexpressed in several lymphomas, whereas miR-34a is repressed in multiple myeloma. Therefore, anti-cancer therapeutic strategies targeting miRNAs could be aimed at re-expressing tumor suppressor miRNAs using miRNA mimetics and/or antagonizing oncogenic miRNAs with antagomiRs. The key challenge in designing RNA-based therapies is their delivery. Reasoning that most tumor microenvironments are hypoxic and acidic, Frank Slack (Harvard Medical School, Boston, USA) presented a novel method for the specific delivery of RNA therapeutics to cancer cells. Using an in vivo model of miR-155-induced mouse lymphoma, his team coupled an anti-miR-155 RNA to a pH-low insertion peptide. They show that the anti-miR was delivered successfully to cancer cells and effectively inhibited miR-155, thus leading to inhibition of tumor growth and metastasis (Cheng et al., 2015). Given these impressive results, it is hoped that this delivery platform could be applied to other types of non-coding RNAs that emerge as possible therapeutics.

Conventional therapy, which targets cancer-specific phenotypes such as a high proliferative rate, also eliminates normal cells that present similar phenotypes and thus has serious side effects. In contrast, targeted therapy and immunotherapy have the advantage of being specific for cancer cells. However, their specificity is also their main drawback as cancer is the ‘most Darwinian’ of all diseases. In fact, tumors are heterogeneous populations of cells containing subclones that are present at different frequencies and characterized by different mutational landscapes. Therefore, any selective pressure (such as treatment) exerted on them will modify the intra-tumor equilibrium and positively select for treatment-resistant subclones, thus increasing chances of resistance or relapse. Speakers in this session addressed questions such as how analysis of ‘-omics' data can help to identify actionable cancer-specific targets and how can they be used to predict the best treatment for a given patient.

The current paradigm in targeted therapy is to use cancer mutational data as a guide to choose the right treatment for the right patient. However, should mutation-negative cancer patients be excluded from targeted therapies? In his talk, Andrew Emili (University of Toronto, Toronto, Canada) presented a method that combines affinity purification and mass spectrometry to identify protein–protein interactions in metazoan cells. If applied to cancer–normal pairs, this method would enable researchers to map the interactomes of those cells and identify cancer-specific protein networks. As these networks reflect dependencies of cancer cells, their disruption should lead to cancer cell death. Moreover, identification of such cancer cell dependencies might be a better indicator of therapy efficiency than the mutational information per se. As cancer is a result of the rewiring of protein networks to gain selective advantages over normal cells, mutation-positive and mutation-negative patients might actually present identical network signatures and therefore could be cured efficiently by the same drug.

Another important parameter that needs to be taken into account while treating cancer is the intra-tumor heterogeneity because it can promote the emergence of treatment-resistant subclones. However, it is difficult for a clinician to fully appreciate the extent of this heterogeneity. To tackle this problem, Edwin Wang (McGill University, Montréal, Canada) presented a software, eTumorKiller, that uses cancer genomic data as input to predict the number and frequency of subclones present in the tumor and to further determine the ideal drug targets for each subclone (Wang et al., 2015). Using this tool could help clinicians to maximize their chances of killing all cancer cells in a given patient, while minimizing the risk of subsequent relapse.

The rationale behind CD8-positive T-cell (CD8T) based immunotherapy is the same as that underlying targeted therapy except that CD8T cells (rather than drugs) are used to specifically destroy cancer cells. At steady state, CD8T cells detect and eliminate virally infected or pre-cancerous cells by scanning their major histocompatibility complex (MHC)-I-associated peptides (MAPs). It has been shown that infusing patients with CD8T cells that have been extracted from their own tumor can promote regression. But which MAPs on cancer cells are recognized by these T cells? Part of the answer came from Pramod K. Srivastava's talk (University of Connecticut School of Medicine, Farmington, USA), who showed that some cancer-specific somatic mutations predicted to generate MAPs were able to elicit CD8T cells responses in mice. This is because they generated more stable MHC-I–peptide complexes than their wild-type counterpart (Duan et al., 2014). Reasoning that cancer-specific MAPs might be exceedingly rare on hematopoietic cancer cells (because of their low mutation load), Claude Perreault (IRIC, Université de Montréal, Montréal, Canada) presented a proteogenomic approach to identify MAPs that are derived from germline polymorphisms, also called minor histocompatibility antigens (MiHAs) (Granados et al., 2015). Injection of allogenic MiHA-specific CD8T cells in the recipient will specifically kill hematopoietic cancer cells when (1) the donor and recipient are HLA-matched, (2) the target MiHA is present in the recipient but not the donor, and (3) the MiHA is specific to (or enriched on) hematopoietic cells. Moreover, he showed that MAPs that are derived from non-coding regions of the genome, also called cryptic MAPs, represented ∼10% of all presented MAPs at the cell surface of a B-lymphoblastoid cell line. These MAPs were enriched in germline polymorphisms and able to elicit immune responses in vitro, making them interesting new targets for immunotherapy.

Over two days, the IRIC hosted 14 international speakers and 250 symposium attendees from across North America, who came together to share and discuss the latest research findings in diverse fields of cancer genomics with a common objective to uncover new therapeutic targets for cancer. The symposium highlighted three broad take-home messages: first, the detailed understanding of molecular mechanisms underlying faithful DNA replication and efficient repair is important for developing cancer therapy strategies, as these mechanisms are the first lines of defense against genomic instability leading to cancer. Second, an understanding of epigenetic regulators, non-coding RNAs, protein networks and how they interconnect is essential for specifically targeting cancer landscapes. Although single-target identification is as important as ever, a combinatorial approach is emerging as the ultimate way to tackle cancer. A tumor can be targeted at the level of the genome, transcriptome (both coding and non-coding), proteome and its protein interactome. Why should it not be targeted from all these angles at once? Finally, cancer must be considered in the context of the human body. What is happening in the microenvironment, and can the immune system be re-educated to recognize and kill tumors through personalized immunotherapy? Gaps remain in our knowledge in all areas of cancer biology, but cutting-edge research like that presented at this year's symposium contributes to the steady progress towards development of effective treatments for cancer.

We are grateful to the organizers for putting together a fantastic symposium. It was a vibrant success at every level and universally applauded by participants. We thank all the speakers who accepted our invitation and delivered outstanding lectures. This symposium was sponsored by The Company of Biologists, Canadian Institutes of Health Research (CIHR), Génome Québec, Université de Montréal (Faculté de médecine, Groupe de Recherche Universitaire sur le Médicament, Fédération des Associations Étudiantes du Campus de l'Université de Montréal, Fonds d'Investissements des Cycles Supérieurs de l'Université de Montréal, IRICoR), GE Healthcare, BioLegend, ThermoFisher Scientific, Bristol-Myers Squibb and Nikon. We apologize that we cannot include highlights from student presentations owing to space limitations.