Issue 11 : March, 2014

In 2008, the respective laboratories of Dr. Marco Marra and Dr. Randy Gascoyne at Canada’s Michael Smith Genome Sciences Centre (GSC), British Columbia Cancer Agency (BCCA), partnered together to sequence the DNA and RNA of the two most common non-Hodgkin lymphomas (NHLs). Their objective was to diagnostically characterize the NHL tumors by using advanced sequencing techniques. Previous molecular studies had identified a few hallmark abnormalities in these cancers, but a more detailed picture of the genomic landscape was needed. Their comprehensive analysis, supported by the Cancer Genome Characterization Initiative (CGCI), revealed a surprising array of novel alterations^1,2,3. Among the most interesting were alterations that pointed to the epigenome as an influential player in lymphomagenesis. 

The two NHLs, follicular lymphoma (FL) and diffuse large B cell lymphoma (DLBCL), are cancers of mature B cells. FL is slow-growing and difficult to treat; DLBCL is often aggressive and clinically diverse. In both cancers, processes involved in B cell differentiation are frequently perturbed⁴. Normal differentiation occurs when an antigen-activated B cell induces formation of a germinal center in the lymph nodes. The B cell localizes to the germinal center and then proliferates, while simultaneously undergoing antibody diversification. The resulting outcome is a diverse collection of B cells with antigen receptors of varying affinities. The mature B cells with the highest affinity receptors are selected to become memory or antibody-secreting plasma cells and exit the germinal center. Those not selected self-destruct through apoptosis. 

Lymphomagenesis can occur when acquired alterations block completion of differentiation and cause B cells to divide uncontrollably and evade cell death. Consequently, FL and DLBCL may resemble B cells at a particular stage of differentiation⁴. Gene expression profiling revealed FL tumors have expression signatures like those of germinal center B cells. Additionally, DLBCL tumors can have signatures like those of germinal center B cells or activated B cells (B cells that have exited the germinal center, but have not fully matured into memory or plasma cells). Accordingly, DLBCL tumors are classified into at least two subgroups: germinal center B cell-like (GCB) and activated B cell-like (ABC). These subgroups show differences in growth behavior and response to treatment, suggesting differences in pathogenesis as well. 

Separate genome-wide characterization studies provided additional clues that the underlying pathogenesis differs for each NHL subtype: several genetic abnormalities are specific to lymphomas derived from either germinal center B cells (i.e. FL and GCB DLBCL) or activated B cells (i.e. ABC DLBCL). For example, overexpression of the BCL2 oncoprotein, a result of a translocation between chromosomes 14 and 18, is only found in GCB-derived lymphomas (89% FL and 30-40% GCB DLBCL)^5,6. NF-kB signaling dysregulation, on the other hand, is only found in the ABC subtype^7,8. These initial molecular profiles were very insightful, but limited in granularity. Dr. Marra and the CGCI-supported investigators at the GSC reasoned that exposing underlying genetic alterations to inform more precise treatments for FL and DLBCL would require higher-resolution, integrative “omic” analyses. 

By applying their expertise in sequencing and bioinformatics, the CGCI investigators examined the whole genomes, exomes, and transcriptomes of a large number of FL and DLBCL tissues. They published their results in three papers over the course of several years^1,2,3. Integrating the data revealed an interesting trend in these lymphomas: genes encoding histone-modifying proteins and histone proteins themselves were frequently mutated. Of those genes, EZH2 and MLL2 had the most recurrent mutations. Both encode enzymes called methyltransferases that regulate gene expression by adding methyl groups to histones. By methylating histones, these enzymes modify the degree to which the DNA is packed and, thus, the degree to which genes and their promoter regions are accessible to transcriptional machinery. The EZH2 and MLL2 mutations are particularly compelling, because they are both predicted to function in malignancy by turning down gene expression. 

EZH2

EZH2, an enzymatic subunit of the polycomb repressor complex 2, plays an important role in a variety of developmental processes. It tri-methylates lysine 27 on histone 3 (H3K27me3), which generally represses transcription. Expressed in a specific time window during B cell differentiation, EZH2 is required for the formation and functions of the germinal center⁹. Late-stage differentiating B cells must turn down EZH2 expression, so they can exit the germinal center to become memory or plasma cells. Before the publication of Morin et al. 2010, overexpression of EZH2 was linked to breast, prostate, and other cancers^10,11, but mutations in the EZH2 gene itself had not yet been associated with any cancer type. 

The CGCI investigators, including Dr. Marra’s then graduate student Ryan Morin, discovered a mutation hot spot in the EZH2 gene in their sequencing of mRNAs of DLBCL and FL cases¹. The mutations recurrently affected tyrosine 641 (Y641) in the catalytic domain of EZH2. Interestingly, the Y641 mutations were restricted to GCB-derived lymphomas (21% GCB DLBCL and 7.2% FL), suggesting they are important in GCB-derived B cell malignancy. 

To determine the effect of Y641 mutations on EZH2 function, the investigators performed in vitro experiments on several purified mutant proteins. The mutations disrupted the enzyme’s ability to add the first methyl group onto a histone peptide, indicating they were loss of function (inactivating). However, they tested the EZH2 mutant proteins singularly, and not in the presence of the wildtype EZH2. Y641 mutations were heterozygous in all cases identified, meaning each tumor had one copy each of the mutant and wildtype EZH2. Later experiments done by a different group showed that EZH2 mutations were actually gain of function (activating) when paired with the wildtype^12,13. Furthermore, EZH2-mutant cell lines have more tri-methylated H3K27 sites as compared to wildtype EZH2 cells, supporting the notion the mutations are indeed gain of function^12,14. Reconciling the two in vitro experimental results, researchers theorized that wildtype EZH2 efficiently adds the first methyl group to H3K27, which enables mutated EZH2 to add the second and third methyl groups more readily than the wildtype. The end result is more tri-methylated H3K27 sites and, thus, more repressive chromatin.  

The question of precisely how activating mutations in EZH2 contribute to B cell lymphomagenesis has elicited great interest from different research groups and is currently being explored. One recent study in mouse models and human cell lines indicates that mutant EZH2 may continuously silence genes antagonistic to malignancy (e.g. germinal center exit and proliferation checkpoint genes), which are normally transiently repressed in germinal center B cells⁹. As a result, mature B cells remain locked in a state of incomplete differentiation in the germinal center, where they rapidly grow and divide unhindered. Taking this into consideration, it would make sense that EZH2 mutations would not participate in the pathogenesis of ABC DLBCL, because EZH2 is no longer expressed in activated B cells that have left the germinal center. 

MLL2

MLL2 methylates the fourth lysine of histone 3 (H3K4), which promotes transcription. MLL2 regulates a diverse array of cellular processes and signaling pathways, including the tumor suppressor p53 pathway¹⁵, and is frequently mutated in certain cancers. For example, another group of CGCI investigators studying medulloblastoma identified recurrent mutations in MLL2 in 14% of all cases¹⁶. The majority of MLL2 mutations identified in this study were predicted to be loss of function, hinting that it may be a tumor suppressor. However, little is known about how MLL2 participates in oncogenesis. 

In their analysis of sequence from the genomes, exomes, and transcriptomes of 117 FL and DLBCL tumors, Morin and the CGCI-supported investigators identified MLL2 as one of 109 recurrently mutated genes². MLL2 showed the greatest mutational frequency, with the majority of identified mutations predicted to disrupt MLL2 function.  These largely inactivating MLL2 mutations were found in 89% of FL cases and 32% of DLBCL cases, a prevalence on par with the t(14;18) translocation. Unlike EZH2, MLL2 did not show bias towards any DLBCL subtype.  Because the role of MLL2 in B cell differentiation is not well-established, the function of MLL2 mutations in lymphomagenesis remains unclear. However, it is possible that impairment of its H3K4 methylating activity alters gene expression in favor of malignancy. This remains to be determined. 

So, What’s Next?

Sequencing the DNA and RNA of FL and DLBCL tumors revealed many frequent alterations in histone-modifying genes, most notably in genes affecting H3K27 (EZH2) and H3K4 (MLL2) methylation. The investigators at the GSC, along with several other groups, are following up by studying how the EZH2 and MLL2 alterations specifically contribute to lymphoma biology and how they may be exploited to improve treatment strategies for these cancers. Because both genes might regulate chromatin marks in the same promoter regions and H3K4 methylation opposes H3K27 methylation¹⁷, researchers speculate that both alterations produce the same outcome in differentiating B cells: increased H3K27-methylated chromatin and repressed transcription of genes antagonistic to malignancy¹⁸. However, additional and/or alternative theories for their roles in lymphomagenesis are entirely plausible. Both EZH2 and MLL2 have widespread functions that are not fully characterized and their exact mechanisms in B cell differentiation are not completely or clearly defined. 

For EZH2, the path to drug development is more straightforward, because the recurrent mutations result in increased activity, which can be counteracted by chemical inhibition. The biopharmaceutical company, Epizyme, recently reported tumor growth inhibition in NHL mouse models bearing the cancer-causing EZH2 mutations when treated with a small molecule inhibitor of EZH (EPZ-6438)¹⁹. They have since initiated a phase I/II clinical trial to test the safety and efficacy of using EPZ-6438 to treat lymphomas harboring those mutations²⁰. For MLL2, therapeutic development is trickier, because the recurrent mutations result in decreased activity, rendering chemical inhibition a non-starter. To circumvent this problem, Dr. Marra’s lab is working on identifying synthetic lethal interactions in MLL2 mutant tumors as potential alternative drug targets²¹.

The Emerging Role of the Epigenome

In addition to EZH2 and MLL2 mutations, the CGCI-supported investigators and other groups detected frequent mutations in other genes encoding previously identified epigenetic modifiers (e.g. CREBBP/EP300)^{2,3, 22}. They also found novel mutations in genes encoding histone proteins (e.g. HIST1H1C) and transcription factors that recruit epigenetic modifiers (e.g. MEF2B), confirming the importance of epigenetic dysregulation in B cell lymphoma biology. Clearly, more research must be done to elucidate the epigenome’s specific mechanistic contributions. 

Dr. Marra and colleagues are currently applying their savvy sequencing and bioinformatics approaches to examine the epigenome of NHL to gain deeper insight into its role in lymphomagenesis. Considering the mysteries that have been revealed in their previous studies of the genome and transcriptome, the NHL epigenome could prove equally illuminating.

References

1. Morin RD, Johnson NA, Severson TM, Mungall AJ, An J, Goya R, et al. (2010). Somatic mutations altering EZH2 (Tyr641) in follicular and diffuse large B-cell lymphomas of germinal-center origin. Nature Genetics 42(2):181–5 (PMID: 20081860)
2. Morin RD, Mendez-Lago M, Mungall AJ, Goya R, Mungall KL, Corbett RD, et al. (2011). Frequent mutation of histone-modifying genes in non-Hodgkin lymphoma. Nature 476(7360):298–303 (PMID: 21796119)
3. Morin RD, Mungall K, Pleasance E, Mungall AJ, Goya R, Huff R, et al. (2013). Mutational and structural analysis of diffuse large B-cell lymphoma using whole genome sequencing. Blood 122(7):1256-65 (PMID: 23699601)
4. Shaffer AL, Young RM, Staudt LM (2012). Pathogenesis of human B cell lymphomas. Annual Review of Immunology 30:565–610. (PMID: 22224767)
5. Horsman DE, Okamoto I, Ludkovski O, Le N, Harder L, Gesk S, Siebert R, et al. (2003). Follicular lymphoma lacking the t(14;18)(q32;q21): identification of two disease subtypes. Br. J. Haematol. 120(3):424–433 (PMID: 12580956)
6. Iqbal J, Sanger WG, Horsman DE, Rosenwald A, Pickering DL, Dave B, et al. (2004). BCL2 translocation defines a unique tumor subset within the germinal center B-cell-like diffuse large B-cell lymphoma. Am. J. Pathol. 165(1):159–166 (PMID: 16418494)
7. Davis RE, Brown KD, Siebenlist U, Staudt LM (2001) Constitutive nuclear factor κB activity is required for survival of activated B cell-like diffuse large B cell lymphoma cells. J. Exp. Med. 194(12):1861–1874 (PMID: 11748286)
8. Lenz G, Davis RE, Ngo VN, Lam L, George TC, Wright GW, Dave SS, et al. (2008). Oncogenic CARD11 mutations in human diffuse large B cell lymphoma. Science 319(5870):1676–1679 (PMID: 18323416)
9. Béguelin W, Popovic R, Teater M, Jiang Y, Bunting KL, Rosen M, et al. (2013). EZH2 is required for germinal center formation and somatic EZH2 mutations promote lymphoid transformation. Cancer Cell 23(5):677–92 (PMID: 23680150)
10. Varambally S, Dhanasekaran SM, Zhou M, Barrette TR, Kumar-Sinha C, et al. (2002). The polycomb group protein EZH2 is involved in progression of prostate cancer. Nature 419(6907):624–629 (PMID: 12374981)
11. Kleer CG, Cao Q, Varambally S, Shen R, Ota I, Tomlins SA, Ghosh D, et al. (2003). EZH2 is a marker of aggressive breast cancer and promotes neoplastic transformation of breast epithelial cells. Proc. Natl. Acad. Sci. USA 100(20):11606–11611 (PMID: 14500907)
12. Sneeringer CJ, Scott MP, Kuntz KW, Knutson SK, Pollock RM, Richon VM, Copeland RA (2010). Coordinated activities of wild-type plus mutant EZH2 drive tumor-associated hypertrimethylation of lysine 27 on histone H3 (H3K27) in human B-cell lymphomas. Proc. Natl Acad. Sci. USA 107(49):20980–20985 (PMID: 21078963)
13. Yap DB, Chu J, Berg T, Schapira M, Cheng SW, Moradian A, Morin RD, et al. (2011). Somatic mutations at EZH2 Y641 act dominantly through a mechanism of selectively altered PRC2 catalytic activity, to increase H3K27 trimethylation. Blood 117:2451–2459 (PMID: 21190999)
14. McCabe MT, Graves AP, Ganji G, Diaz E, Halsey WS, Jiang Y, et al. (2012). Mutation of A677 in  histone methyltransferase EZH2 in human B-cell lymphoma promotes hypertrimethylation of histone H3 on lysine 27 (H3K27). Proc. Natl Acad. Sci. USA 109(8):2989–2994 (PMID: 22323599)
15. Guo C, Chang CC, Wortham M, Chen LH, Kernagis DN, Qin X, Cho YW, et al. (2012). Global identification of MLL2-targeted loci reveals MLL2’s role in diverse signaling pathways Proc. Natl Acad. Sci. USA 109(43):17603-17608 (PMID: 23045699)
16. Parsons DW, Li M, Zhang X, Jones S, Leary RJ, Lin JC, Boca SM, Carter H, et al. (2011).The genetic landscape of the childhood cancer medulloblastoma. Science 331(6016):435–439 (PMID: 21163964)
17. Shaknovich R, Melnick A (2011). Epigenetics and B-cell lymphoma. Current Opinion in Hematology 18(4):293–9 (PMID: 21577103)
18. Mills AA (2010). Throwing the cancer switch: reciprocal roles of polycomb and trithorax proteins. Nature Reviews Cancer 10(10):669–82 (PMID: 20865010)
19. Knutson SK, Kawano S, Minoshima Y, Warholic N, Huang KC, et al. (2014). Selective inhibition of EZH2 by EPZ-6438 leads to potent antitumor activity in EZH2 mutant non-Hodgkin lymphoma. Mol. Cancer Therapeutics. Epub ahead of print, Feb 21, 2014 (PMID: 24563539)
20. http://clinicaltrials.gov/ct2/show/NCT01897571
21. http://bcgsc.ca/faculty/mmarra/marra-research-activities/trainee-projects-1
22. Pasqualucci L1, Dominguez-Sola D, Chiarenza A, Fabbri G, Grunn A, et al. (2011). Inactivating mutations of acetyltransferase genes in B-cell lymphoma. Nature 471(7337):189-95 (PMID: 21390126)

In the process of malignancy, cells acquire multiple genetic alterations, such as focal mutations, translocations, amplifications, or deletions, which allow them to survive and divide uncontrollably. In addition to these genetic changes, normal functioning proteins that play roles in innocuous cellular processes may become essential to cancer cell survival when they get co-opted into partnerships with oncogene proteins. The latter is a phenomenon known as non-oncogene addiction or oncogene-induced dependency. Insights into cancer genomes have led to novel clinical therapies that target oncogene proteins, and such targeted therapies have yielded high patient response rates. However, less than one percent of patients suffering from cancer benefit from these therapies, and patient-matched therapies that target essential non-oncogenic proteins or their related pathways have been difficult to determine. To accelerate discovery of patient-matched therapies, systematic approaches are needed to identify oncogene-induced dependencies that cancers acquire as a result of specific cellular or genetic features and small-molecule drugs that target them.

As part of the National Cancer Institute’s Cancer Target Discovery and Development (CTD²) Network, we profiled a large number of human cancer cell lines using a novel “Informer Set” of more than 350 small molecules to reveal such dependencies and their inhibitors. We deposited the dataset into a publicly accessible data portal, called the Cancer Therapeutics Response Portal (CTRP; www.broadinstitute.org/ctrp; Figure 1), so that other researchers can make connections between the genetic and lineage features of cancer cell lines and small-molecule sensitivities. The CTRP, along with findings from an initial analysis, are described in a recent publication (http://www.ncbi.nlm.nih.gov/pubmed/23993102)¹.

Figure 1. The Cancer Therapeutics Response Portal (CTRP; http://www.broadinstitute.org/ctrp) homepage (top left) allows users to access data through three entry points: small molecules, enriched features, or targets. Once inside the portal, users can mine this resource for novel and therapeutically exploitable vulnerabilities in different cancer types across ~185 small molecules. For example, a user might enter through the “Compounds” link then search for a molecule of interest, like navitoclax. Under the “General” tab, the navitoclax entry shows the chemical structure of the compound and provides other general information (top right). Under the “Enrichment Analysis” tab, users can select or exclude specific cell line subtypes and datasets. For example, CTNNB1-mutant cancer cell lines are sensitive to navitoclax (bottom).

Historically, human cancer cell lines derived from many different types of tumors have been used in high-throughput profiling efforts to reveal patterns of small-molecule sensitivities. These studies have been limited in the number, diversity, or level of molecular characterization of cancer cell lines or small molecules used. One of the earliest profiling efforts, the NCI-60² probed a limited set of 59 cancer cell lines from various lineages with more than 100,000 diverse small molecules and identified mostly lineage-specific small-molecule sensitivities. More recently, the Cancer Cell Line Encyclopedia (CCLE), a joint effort between the Broad Institute and Novartis Institutes for BioMedical Research, profiled 479 cancer cell lines with significant genomic characterization using 24 anti-cancer drugs³. Massachusetts General Hospital and the Wellcome Trust Sanger Institute profiled 350 cancer cell lines against 130 pre-clinical or clinical anti-cancer agents, though the set of genomic alterations correlated to sensitivity was limited to ~70 genes⁴. The Lawrence Berkeley National Laboratory and MD Anderson Cancer Center profiled 77 therapeutic compounds against a panel of ~50 breast cancer cell lines and linked various subsets, pathways and genetic features of these lines to drug sensitivity⁵.

By leveraging work from previous projects and recent advances in high-throughput technologies, we addressed previous shortcomings of high-throughput cancer profiling studies. We used 242 of ~1000 cancer cell lines from the CCLE, which had been genomically characterized for gene expression, amplification/deletion, and somatic mutation in 1,645 cancer genes, and had lineage or histology annotations. These cell lines were established from many different types of tumors. Furthermore, we used an Informer Set of 354 small molecules comprising 35 FDA-approved drugs, 54 clinical candidates, and 266 probes – tool compounds used in biological research to help illuminate pathways and mechanisms of protein function. These probes and drugs each selectively target distinct nodes in cancer cell circuitry and collectively modulate a broad array of cellular processes. Each cell line was grown in its preferred media, and each compound tested at eight different concentrations, with the starting concentration of each compound selected based on literature review. We generated concentration-response curves that determined each compound’s potency and efficacy for each cell line. The area under the concentration-response curve was used to calculate compound (i.e., small-molecule) sensitivities. We input the resulting data as well as previously determined molecular characterizations/annotations into one database and created the CTRP for access, query, and visualization.

The current version of CTRP (Figure 1) uses pre-computed visualizations to display more than 75,000 statistically significant connections between 185 small molecules and cancer cell lines of a particular lineage or with particular mutations. The portal also allows users to customize the query by eliminating cancer cell line subtypes, such as frequently sensitive or highly mutated ones, from analyses. Specific details on how to make novel connections between molecular features of cancer cell lines and the sensitivity profiles are found in the corresponding publication (http://www.ncbi.nlm.nih.gov/pubmed/23993102)¹. Of note, researchers may use the CTRP to identify drugs that could rapidly be tested clinically, because FDA-approved drugs and clinical candidates are included the Informer Set. This critical new resource may aid in advancing discovery of potential cancer drugs matched to the patient populations most likely to benefit from them.

The CTRP is a living resource that will grow over time. We recently completed a second phase of new small-molecule sensitivity data collection. This phase expanded the number of genetically characterized cancer cell lines to ~900 and the Informer Set to ~545 molecules. The new, enlarged Informer Set not only targets a larger swath of proteins, but also includes novel chemical entities, such as stapled-helical peptides, more FDA-approved drugs and clinical candidates, and rational combinations of small molecules. We are starting to plan a new generation of CTRP to accompany our new cancer cell-line profiling datasets. In particular, we anticipate three functionalities in a future CTRP version: (1) application of single-feature lineage or mutation enrichment analyses to new datasets with improved computational methods; (2) single-gene correlation analyses between small-molecule sensitivity and gene-expression or copy-number features; and (3) compound and cell-line correlation and clustering analyses using small-molecule sensitivity data alone. We are also developing additional methods to analyze and visualize the data. For example, we are incorporating the ability to correlate small-molecule sensitivity to combinations of features or to genetic vulnerabilities uncovered by Project Achilles⁶, another CTD²-sponsored initiative that uses a genome-wide shRNA library to silence individual genes and identify those genes that affect cell survival.

We strongly encourage the scientific community to use the CTRP to mine for novel and therapeutically exploitable vulnerabilities in different cancer types. Many have already started using this hypothesis-generating tool¹. Preliminary user-trend analysis for the month of January indicates about 32 users log in per day, with each user viewing an average of >170 results. We hope to attract more users as we increase the inputs, features, and functionality of this resource. As the volume and diversity of data in publicly available resources like the CTRP continue to grow, the catalog of cancer genes and related dependencies as well as the drugs that target them will likewise grow. Thus, we anticipate the ability to match patients with potentially effective drugs and that improved patient outcomes will dramatically increase.

References

1. Basu A, Bodycombe NE, Cheah JH, Price EV, Liu K, Schaefer GI, et al. (2013) An interactive resource to identify cancer genetic and lineage dependencies targeted by small molecules. Cell 154(5):1151-61 (PMID: 23993102)
2. Shoemaker RH (2006) The NCI60 human tumour cell line anticancer drug screen. Nature Reviews Cancer 6(10):813-23 (PMID: 16990858)
3. Barretina J, Caponigro G, Stransky N, Venkatesan K, Margolin AA, Kim S, et al. (2012) The Cancer Cell Line Encyclopedia enables predictive modelling of anticancer drug sensitivity. Nature 483(7391):603-7 (PMID: 22460905)
4. Garnett MJ, Edelman EJ, Heidorn SJ, Greenman CD, Dastur A, Lau KW, et al. (2012) Systematic identification of genomic markers of drug sensitivity in cancer cells. Nature 483(7391):570-5 (PMID: 22460902)
5. Heiser LM, Sadanandam A, Kuo WL, Benz SC, Goldstein TC, Ng S, et al. (2012) Subtype and pathway specific responses to anticancer compounds in breast cancer. Proc Natl Acad Sci USA 109(8):2724-9 (PMID: 22003129)
6. Cheung HW, Cowley GS, Weir BA, Boehm JS, Rusin S, Scott JA, et al. (2011) Systematic investigation of genetic vulnerabilities across cancer cell lines reveals lineage-specific dependencies in ovarian cancer. Proc Natl Acad Sci USA 108(30):12372-7 (PMID: 21746896)