Search

Velcrin compounds are a class of small molecules that induce complex formation between PDE3A and SLFN12, killing cancer cells that express elevated levels of these two proteins by a mechanism independent of PDE3A enzymatic inhibition. Instead, PDE3A binding stimulates the RNase activity of SLFN12, resulting in cleavage of the specific SLFN12 substrate, tRNA-Leu-TAA. Cleavage of tRNA-Leu-TAA in turn causes ribosomal pausing, inhibition of protein synthesis, and cancer cell death. Unlike traditional targeted therapies that leverage dependencies created in cancer cells by genomic alterations, velcrins instead kill cancer cells by a gain-of-function mechanism dependent on the RNase activity of SLFN12. In a collaboration between the Broad Institute and Bayer Pharmaceuticals, we developed the first velcrin, BAY 2666605, to enter Phase I clinical trials. BAY 2666605 is active in cell line and patient-derived xenografts of several tumor types, specifically where elevated levels of the two biomarkers, PDE3A and SLFN12, are expressed. Biomarker-positive tumors are especially enriched among melanomas, and we have consistently observed tumor regression in biomarker-positive melanoma tumor models in vivo. BAY 2666605 furthermore shows drug-like properties, excellent brain penetration, increased stimulation of SLFN12 RNase activity, and reduced inhibition of PDE3A enzymatic activity compared with most other velcrins and approved PDE3A inhibitors. BAY 2666605 has recently entered a First-in-Human study (NCT04809805) in patients with advanced solid tumors that co-express PDE3A and SLFN12, including melanoma, ovarian cancer, and sarcoma. Citation Format: Stefan Gradl, Sooncheol Lee, Martin Lange, Xiaoyun Wu, Silvia Goldoni, Timothy Lewis, Charlotte Kopitz, Colin Garvie, Philip Lienau, Stephanie Hoyt, Henrik Seidel, Stephan Kaulfuss, Manuel Ellermann, Luc de Waal, Adrian Tersteegen, Sven Golfier, Detlev Suelzle, Christa Hegele-Hartung, James Carr, Frederick Brookfield, Michael Bruening, Melanie Berthold, Thibaud Jourdan, Monica Schenone, Galen Gao, Joseph McGaunn, Antje Wengner, Elisa Aquilanti, Franziska Siegel, Marine Garrido, Annette Walter, Isabelle Genvresse, Andrew Cherniack, Stuart Schreiber, Knut Eis, Ashley Eheim, Matthew Meyerson, Heidi Greulich. BAY 2666605: The first PDE3A-SLFN12 complex inducer for cancer therapy [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr ND04.

Velcrins are small molecules that induce apoptosis of cancer cells expressing high levels of phosphodiesterase 3A (PDE3A) and schlafen family member 12 (SLFN12) by inducing PDE3A-SLFN12 complex formation. Recently, we found that SLFN12 is an RNase and PDE3A binding dramatically activates SLFN12 RNase activity, resulting in velcrin-mediated apoptosis. However, the mechanism of cancer cell death by the PDE3A-SLFN12 complex is poorly understood. In this study, we found that SLFN12 specifically recognizes tRNA-Leu-TAA and induces its cleavage in velcrin-sensitive cell lines. According to tRNA-sequencing, tRNA-Leu-TAA isodecoders decreases by 2-fold to 5-fold in Hela cells treated with DNMDP compared to DMSO control, while other tRNAs are not differentially expressed. Down-regulation of tRNA-Leu-TAA by velcrin treatment is observed only in velcrin-sensitive cell lines but not insensitive cell lines. Furthermore, we tested whether SLFN12 digests tRNA-Leu-TAA in vitro. Purified wild-type SLFN12 cleaves tRNA-Leu-TAA but not heat-denatured or catalytically inactive SLFN12. Even though tRNA-Leu-TAA is weakly digested by low concentrations of SLFN12, co-incubation with purified PDE3A increases SLFN12 RNase activity, which is even further upregulated by velcrin treatment. To verify specificity for tRNA-Leu-TAA, we assessed activity of SLFN12 against a panel of in vitro transcribed tRNAs. We found that purified SLFN12 is significantly more efficient at degrading tRNA-Leu-TAA when compared with tRNA-Leu-TAG or tRNAs for other amino acids. Digestion of tRNA-leu-TAA by SLFN12 could inhibit mRNA translation due to the delivery of leucine amino acid to elongating ribosomes. To test this hypothesis, we designed the mutant tRNA-Leu-[CAG:TAA] in which the anticodon of tRNA-Leu-CAG was replaced by TAA anticodon. Ectopic expression of wild-type tRNA-Leu-TAA or tRNA-Leu-CAG has no effect on viability of DNMDP-treated cells, whereas expression of tRNA-Leu-[CAG:TAA] genes partially rescues the cells from DNMDP-induced cytotoxicity. These data confirm that impaired leucine delivery is responsible at least in part for the observed effects of DNMDP treatment. Citation Format: Sooncheol Lee, Stephanie Hoyt, Xiaoyun Wu, Colin Garvie, Joseph McGaunn, Andrew D. Cherniack, Matthew Meyerson, Heidi Greulich. Velcrin-induced cleavage of tRNA-Leu-TAA by SLFN12 RNase causes cancer cell death [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 867.

BAY 2666605, co-developed by the Broad Institute and Bayer Pharmaceuticals, is a selective and potent molecular glue, part of a family of small molecules recently baptized as ‘velcrins’, that induces complex formation between phosphodiesterase 3A (PDE3A) and SLFN12. BAY 2666605 has recently entered a First-in-Human study (NCT04809805) in patients with advanced solid tumors and here we describe its pre-clinical pharmacology profile. DNMDP, the precursor to BAY 2666605, was discovered in a phenotypic screen of genomically annotated cancer cell lines and sensitivity to treatment correlated to high expression of PDE3A (1). Upon treatment, SLFN12 is recruited into a stable complex with PDE3A where its RNase activity is enhanced and required for response (2). BAY 2666605 is a potent complex inducer (EC50 = 7 nM) and cytotoxic in vitro with nanomolar potency (IC50 = 1nM, in the most sensitive cell lines). Cancer cells with high expression of PDE3A and co-expression of SLFN12 are killed by a mechanism independent of PDE3A enzymatic inhibition. PDE3A-SLFN12 binding is required for cytotoxicity. Biomarker-positive lines are enriched in the melanoma lineage and show dose-dependent sensitivity to BAY 2666605 both in vitro and in vivo. Notably, we have consistently observed tumor regression in biomarker-positive melanoma models, including in PDX models (10mg/kg po BID). Based on target expression data from TCGA and tumor arrays, various other tumor types also co-express PDE3A and SLFN12, such as sarcomas and ovarian cancer. To this end, we show that BAY 2666605 inhibits tumor growth of PDX models of sarcoma and ovarian cancer in vivo. BAY 2666605 has excellent brain penetration, making glioblastoma a promising indication. Biomarker-positive GBM models are sensitive to BAY 2666605 both in vitro and in vivo. In a subset of orthotopic GBM models BAY 2666605 treatment has significant impact on survival. In BAY 2666605 treated models we have observed MCL1 downregulation and this biomarker will be evaluated in clinical settings. Our pre-clinical data indicate that BAY 2666605 is a potent anti-tumor agent with first-in-class potential and broad indication space. 1. de Waal et al. Identification of cancer-cytotoxic modulators of PDE3A by predictive chemogenomics, Nat. Chem. Biol. 12, 102-108 (2016) 2. Garvie et al. Structure of PDE3A-SLFN12 complex reveals requirements for activation of SLFN12 RNase, Nat. Commun. 12, 4375 (2021) Citation Format: Silvia Goldoni, Martin Lange, Charlotte Kopitz, Stefan Kaulfuss, Sven Golfier, Adrian Tersteegen, Stefanie Bunse, Melanie Berthold, Thibaud Jordan, Philip Lienau, Franziska Siegel, Annette Walter, Henrik Seidel, Elisa Aquilanti, Andrew Baker, Xiaoyun Wu, Sooncheol Lee, Stefan Gradl, Emmanuelle di Tomaso, Matthew Meyerson, Knut Eis, Ashley Eheim, Heidi Greulich. Preclinical profiling of BAY 2666605: The first PDE3A-SLFN12 complex inducer for cancer therapy [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 2663.

PDE3A-SLFN12 complex formation is induced by a class of compounds, now called “velcrins”, exemplified by the small molecule, DNMDP. Cancer cells that express elevated levels of PDE3A and SLFN12 are sensitive to a velcrin-mediated cytotoxic response, which is independent of PDE3A inhibition. However, the details of complex formation have not yet been revealed. We solved the crystal structure of PDE3A with a series of ligands bound to the active site and found that PDE3A exists as a dimer, and velcrin binding does not cause any obvious structural changes in the PDE3A protein structure. Hydrogen-deuterium exchange (HDX-MS) experiments with velcrin-bound PDE3A in the absence and presence of SLFN12 identified three regions of PDE3A that are shielded from solvent as a result of velcrin-induced SLFN12 binding. Two of these regions are near the velcrin binding site, and the third region lies at the PDE3A homodimerization interface. In order to further investigate the structural relationship between PDE3A, DNMDP, and SLFN12, we took a deep-mutation scanning (DMS) approach to identify residues of PDE3A that impact DNMDP sensitivity. A library of PDE3A alleles was developed in which the sequence encoding amino acids 668-1141, including the PDE3A catalytic domain, was substituted with a codon for every other possible amino acid or a stop codon in the context of the full-length cDNA. The library was transduced into PDE3A-knockout GB1 glioblastoma cells and assessed for survival in the presence of DMSO or DNMDP. Corroborating the HDX-MS data, we identified three regions of PDE3A in which missense mutations abrogated DNMDP response: the active site, the homodimerization surface, and an alpha helix containing amino acid F914. We confirmed that mutations of F914 and the homodimer interface retain the ability to bind resin-conjugated compound but fail to complex with SLFN12. Taken together, the HDX and DMS results suggest that PDE3A dimerization is required to stabilize velcrin-induced SLFN12 binding and implicate the alpha helix containing F914 as the SLFN12 binding interface of PDE3A. Citation Format: Xiaoyun Wu, Malvina Papanastasiou, Gavin Schnitzler, Colin Garvie, Stephanie Hoyt, Terry Zhang, James Mullahoo, Andrew Baker, Joseph McGaunn, Bethany Kaplan, Sooncheol Lee, Martin Lange, Steven Carr, Xiaoping Yang, Federica Piccioni, Andrew Cherniack, Matthew Meyerson, Heidi Greulich. Deep mutational scanning of PDE3A identifies residues required for DNMDP response [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 1219.

We have identified a novel mechanism of selective cancer cell-killing, whereby small molecules such as DNMDP induce complex formation between phosphodiesterase 3A (PDE3A) and a poorly characterized protein, SLFN12, that leads to downstream cell death (de Waal et al., Nat Chem Biol 2016; Lewis et al., ACS Med Chem Letters, 2019; Wu et al., J Bio Chem 2020). This process therefore relies on the presence and interaction of these two proteins, rather than a pre-existing dependency, as is the mechanism of many currently available cancer therapies. While complex formation is required for cancer cell killing, the molecular basis for complex induction and the activity of the complex that leads to cell death have not been described. In this study, we employed cryogenic electron microscopy (cryo-EM) to deduce the high-resolution structure of the DNMDP-induced PDE3A-SLFN12 complex. The solved structure includes DNMDP-bound PDE3A at 2.97 Å, as well as the structure of the full length SLFN12 at 2.76 Å. We found that the entire complex consists of a 2:2 hetero-tetramer with the C-terminal alpha-helix of SLFN12 reaching into the DNMDP-occupied substrate binding pocket of PDE3A and contacting both DNMDP and active site residues of PDE3A. The binding of DNMDP to PDE3A does not induce structural changes, but rather creates an adhesive surface at the entrance of the substrate binding pocket, which allows for stable binding of SLFN12. We have named DNMDP and similar PDE3A-SLFN12 complex inducing compounds “velcrins”, after the adhesive fabric. Site-directed mutagenesis confirmed that residues on the SLFN12 C-terminal helix are required for DNMDP-induced cell death. Based on homology to the SLFN13, we predicted that SLFN12 would also encode an RNase. Indeed, purified SLFN12 protein cleaved ribosomal RNA in vitro and mutation of the putative catalytic residues abolished this activity. Importantly, catalytically dead SLFN12 mutants failed to mediate DNMDP-induced cell killing while retaining the ability to complex with PDE3A. Consistent with the requirement of SLFN12 RNase activity for DNMDP-induced cell death, overexpression of wild-type SLFN12 alone was cytotoxic, whereas overexpression of catalytically dead SLFN12 was not. The RNase activity of SLFN12 in vitro was stimulated in the presence of DNMDP and PDE3A, suggesting that the mechanism of DNMDP-induced cell killing involves activation of the SLFN12 RNase. Intriguingly, SLFN12 homodimer interface mutants were no longer cytotoxic when overexpressed, but still supported DNMDP-induced cell killing. We hypothesize that PDE3A, an obligate dimer, activates SLFN12 RNase activity by stabilizing the dimeric form of SLFN12. Citation Format: Xiaoyun Wu, Colin Garvie, Sooncheol Lee, Steven Horner, Andrew Baker, Marcus Toetzl, Joseph McGaunn, Bethany Kaplan, Luc de Waal, Martin Lange, Timothy Lewis, Chris Lemke, Matthew Meyerson, Heidi Greulich. Cryo-EM structure of the PDE3A-SLFN12 complex reveals requirements for the activation of SLFN12 RNase and DNMDP-induced cancer cell killing [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2021; 2021 Apr 10-15 and May 17-21. Philadelphia (PA): AACR; Cancer Res 2021;81(13_Suppl):Abstract nr 54.

We have previously reported1 our results with a small molecule called DNMDP which kills certain cancer cells by modulating the interaction of PDE3A and Schlafen 12 (SLFN12), a more recently discovered protein of unknown function. While DNMDP selectively inhibits PDE3, most PDE3 inhibitors have no cell killing effects and in fact rescue cancer cells from DNMDP-induced death. DNMDP is not suitable for advanced studies due to structural liabilities. Optimization of DNMDP to both increase activity and improve pharmacokinetic properties resulted in enantiomerically pure, low molecular weight, metabolically stable compounds which are active at low doses in animal models of cervical cancer and melanoma. While these compounds are selective PDE3A inhibitors and their biochemical activity mirrors their cellular activity, the activity of the compounds results not from PDE3 inhibition and increased cAMP levels, but from increased compound-induced binding of PDE3A to SLFN12, which most PDE3 inhibitors do not effect. Our results suggest that small molecule modulators of PDE3/SLFN12 binding may provide a novel treatment for the treatment of certain cancers. 1Nat. Chem. Bio. 2016, 12, 102-108. Citation Format: Timothy A. Lewis, Luc de Waal, Xiaoyun Wu, Manuel Ellerman, Charlotte Kopitz, Antje Wengner, Knut Eis, Martin Lange, Adrian Tersteegen, Philip Lienau, Heidi Greulich, Matthew Meyerson. Small-molecule modulators of PDE3/SLFN12 to kill cancer cells [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2018; 2018 Apr 14-18; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2018;78(13 Suppl):Abstract nr 5880.

Perifosine is an alkylphospholipid exhibiting antitumor activity as shown in both preclinical studies and clinical trials. This activity is partly associated with its ability to inhibit Akt activity. It has been shown that the mammalian target of rapamycin (mTOR) axis plays a critical role in regulation of cell proliferation and survival primarily through functioning both downstream and upstream of Akt. The current study reveals a novel mechanism by which perifosine inhibits Akt and the mTOR axis. In addition to inhibition of Akt, perifosine inhibited the assembly of both mTOR/raptor and mTOR/rictor complexes. Strikingly, perifosine reduced the levels of Akt and other major components including mTOR, raptor, rictor, 70-kDa ribosomal S6 kinase, and 4E-binding protein 1 in the mTOR axis by promoting their degradation through a GSK3/FBW7-dependent mechanism. These results thus suggest that perifosine inhibits the mTOR axis through a different mechanism from inhibition of mTOR signaling by classic mTOR inhibitors such as rapamycin. Moreover, perifosine substantially increased the levels of type II light chain 3, a hallmark of autophagy, in addition to increasing poly(ADP-ribose) polymerase cleavage, suggesting that perifosine induces both apoptosis and autophagy. The combination of perifosine with a lysosomal inhibitor enhanced apoptosis and inhibited the growth of xenografts in nude mice, suggesting that perifosine-induced autophagy protects cells from undergoing apoptosis. Collectively, we conclude that perifosine inhibits mTOR signaling and induces autophagy, highlighting a novel mechanism accounting for the anticancer activity of perifosine and a potential strategy to enhance the anticancer efficacy of perifosine by preventing autophagy. [Cancer Res 2009;69(23):8967–76]

In a differential cytotoxicity screen, we identified a novel small molecule modulator of phosphodiesterase 3A (PDE3A) that kills cancer cells expressing elevated levels of PDE3A and SLFN12 (de Waal, Nat Chem Biol, 2016). Treatment with this cell-selective cytotoxic small molecule, DNMDP, induces complex formation between PDE3A and SLFN12, resulting in apoptosis. Inhibition of PDE3A enzymatic activity is not sufficient for cancer cell killing, and expression of both PDE3A and SLFN12 are required. Although the mechanism of signaling to the apoptosis machinery remains unclear, we examined more closely the role of the PDE3A-SLFN12 complex in cancer cell killing mediated by DNMDP. We found that cancer cell lines made resistant to DNMDP by persistent exposure downregulated SLFN12 expression and that re-expression of SLFN12 was sufficient to restore sensitivity. Furthermore, ectopic expression of PDE3A and SLFN12 are sufficient to sensitize cancer cells to DNMDP. These data underscore the tight correlation of PDE3A-SLFN12 complex formation and cancer cell killing mediated by DNMDP. Citation Format: Xiaoyun Wu, Timothy Lewis, Luc de Waal, Galen Gao, Jian Zhang, Monica Schenone, Colin Garvie, Brett Diamond, Selena Lorrey, Andrew Cherniack, Steven Corsello, Alex Burgin, Todd Golub, Stuart Schreiber, Matthew Meyerson, Heidi Greulich. PDE3A modulation for cancer therapy [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2017; 2017 Apr 1-5; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2017;77(13 Suppl):Abstract nr 2028. doi:10.1158/1538-7445.AM2017-2028

Recent cancer genome sequencing and analysis has identified millions of somatic mutations in cancer. However, the functional impact of most variants is poorly understood, limiting the use of this genetic knowledge for clinical decision-making. Here we describe a new high-throughput approach, expression-based variant impact phenotyping (eVIP), which uses gene expression changes to infer somatic mutation impact. We generated a lentiviral expression library representing 53 genes and 194 somatic mutations identified in primary lung adenocarcinomas. Next, we introduced this library into A549 lung adenocarcinoma cells and 96 hours later performed gene expression profiling using Luminex-based L1000 profiling. We built a computational pipeline, eVIP, to compare mutant and wild-type expression signatures to infer whether variants were gain-of-function, change-of-function, loss-of-function, or neutral. Overall, eVIP identified 69% of mutations as impactful whereas 31% appeared functionally neutral. A very high rate, 92%, of missense mutations in the KEAP1 and STK11 tumor suppressor genes were found to inactivate or diminish protein function. As a complementary approach, we assessed which mutations are epistatic to EGFR or capable of initiating xenograft tumor formation in vivo. A subset of the impactful mutations identified by eVIP could induce xenograft tumor formation in mice and/or confer resistance to cellular EGFR inhibition. Among these mutations were 20 rare or non-canonical somatic variants in clinically-actionable or -relevant oncogenes including EGFR S645C, ARAF S214C and S214F, ERBB2 S418T, and PIK3CA E600K. eVIP can, in principle, characterize any genetic variant, independent of prior knowledge of gene function. Further application of eVIP should significantly advance the pace of functional characterization of mutations identified from genome sequencing. Citation Format: Alice H. Berger, Angela N. Brooks, Xiaoyun Wu, Yashaswi Shrestha, Candace Chouinard, Federica Piccioni, Mukta Bagul, Atanas Kamburov, Marcin Imielinski, Larson Hogstrom, Cong Zhu, Xiaoping Yang, Sasha Pantel, Ryo Sakai, Nathan Kaplan, David Root, Rajiv Narayan, Ted Natoli, David Lahr, Itay Tirosh, Pablo Tamayo, Gad Getz, Bang Wong, John Doench, Aravind Subramanian, Todd R. Golub, Matthew Meyerson, Jesse S. Boehm. High-throughput phenotyping of lung cancer somatic mutations. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr 4368.

With the comprehensive analysis of cancer genomes approaching completion, the research community stands poised to rapidly advance genome-guided therapeutic hypotheses into clinical settings. However, for the vast majority of cancer patients, existing knowledge of the function(s) of the newly discovered mutant genes harbored by their tumor is incomplete or non-existent since most cancer mutations are exceedingly rare. As a result, we now have long lists of candidate alleles but a paucity of targets whose biology is sufficiently well understood to guide therapeutics. Here we present an interim progress report on a pilot effort aiming to create a generalizable framework to systematically map the molecular consequences of cancer variants at scale (Target Accelerator). First, we created an efficient pipeline to generate cancer variants and generated an initial library of 1300 mutant cDNA clones corresponding to variants in lung cancer and diffuse large B-cell lymphoma as well as those nominated by “pan-cancer” computational analyses. Second, we established an industry-scale, next-generation pipeline to generate new cancer models (Cell Line Factory) directly from patient samples. We have leveraged this pipeline to process over 330 samples from 208 patients across 16 cancer types, with over 60% growing through at least 5 population doublings. We show that tumor genomics can be retained in such patient-derived models and that drug testing to discover clinically validated dependencies within 3 months is feasible. In addition, we use combinatorial molecular barcoding to rapidly generate a panel of pathway-primed human tumorigenesis models that are suitable for massively parallel multiplexed tumorigenesis assays in vivo (TumorPlex). We hypothesized that this integrated framework could be utilized to generate meaningful functional hypotheses from cancer variants of unknown significance in a high-throughput manner. To test this hypothesis, we introduced over 1000 cancer mutations into cell models and created gene expression signatures together with phenotypic data. In lung cancer, we show that the mutational impact of mutant alleles with known and unknown functions can be rapidly assessed by comparing signatures of wild-type and mutant alleles. We show that this generalizable approach, which does not require prior knowledge, can place variants of unknown significance into dominant gain-of-function and loss-of-function categories. As a complementary approach, we have used TumorPlex assays to test the tumorigenic potential of 550 mutant alleles nominated by Pan-Cancer computational analyses and discovered unexpected variants in the KRAS, AKT1, MAP2K1, ERBB2, PIK3CB, NFE2L2, FAM200A and POT1 genes as being potently tumorigenic. These proof-of-concept studies demonstrate initial feasibility of mapping cancer variant function at scale. Importantly, they demarcate a path by which mapping variant function and predicting vulnerabilities might soon be possible on a patient-by-patient basis, achieving the promise of precision functional genomics. Citation Format: Alice H. Berger, Eejung Kim, Angela Brooks, Nina Ilic, Yashaswi Shrestha, Yuen-Yi Tseng, Xiaoyun Wu, Lihua Zou, Atanas Kamburov, Xiaoping Yang, Cong Zhu, Paula Keskula, Sara Seepo, Andrew Hong, Philip Kantoff, Keith L. Ligon, Levi A. Garraway, John G. Doench, David E. Root, Matthew Meyerson, William C. Hahn, Gad Getz, Todd R. Golub, Jesse S. Boehm. Towards precision functional genomics via next-generation functional mapping of cancer variants. [abstract]. In: Proceedings of the AACR Special Conference on Translation of the Cancer Genome; Feb 7-9, 2015; San Francisco, CA. Philadelphia (PA): AACR; Cancer Res 2015;75(22 Suppl 1):Abstract nr PR07.

Recently, the decline in the cost of genome sequencing has led to the rapid identification of thousands of cancer-associated somatic mutations. However, progress in characterization of these genetic events has lagged significantly behind. Understanding mutation function is critical not only for research purposes but also for determining targeted treatment strategies based on individual tumor genetic profiles, yet determination of mutation impact remains a significant bottleneck. Here we describe a high-throughput approach to classify somatic mutations that is robust, scalable, and requires no prior information of gene function. We generated a lentiviral cDNA expression library of ~550 mutated and wild-type alleles of genes mutated in lung adenocarcinoma and introduced these alleles into four human lung cell lines. 96 hours post-infection, gene expression profiles were generated using Luminex-based L1000 profiling. In total, more than 2000 gene expression signatures were generated. We discovered that gain-of-function mutants induce expression signatures with a greater signal strength or different identity than the corresponding wild-type gene signature. In contrast, loss-of-function mutants could be identified by their incapability to induce strong signatures. Based on these features of signature strength and signature identity, we developed a decision-tree approach to classify mutations as either dominant, loss-of-function, or likely inert. An orthogonal functional approach, an EGFR inhibitor resistance screen, was used as validation. The gene expression approach correctly classified known gain-of-function mutations in KRAS (13/13), EGFR (6/7), and ARAF (2/2) and identified dozens of never-characterized gain-of-function and loss-of-function missense mutations. In addition to rare, dominant mutations in clinically-actionable oncogenes such as PIK3CA and AKT1, we identified unexpected dominant mutations in the transcription factor MAX and the phosphatase subunit PPP2R1A, among others. We also observed a substantial enrichment of loss-of-function mutations in tumor suppressor genes such as STK11, KEAP1, FBXW7, and CASP8 as well as in genes not previously connected to lung adenocarcinoma, including GPR137B and MAPK7. Most genes assayed also harbored variants that are likely inert, further underscoring the importance of characterizing individual variant alleles. The method developed here can, in principle, characterize any genetic variant, independent of prior knowledge of gene function, and should significantly advance the pace of functional characterization of mutations identified from genome sequencing. Citation Format: Alice Berger, Angela Brooks, Xiaoyun Wu, Larson Hogstrom, Itay Tirosh, Federica Piccioni, Mukta Bagul, Cong Zhu, Yashaswi Shretha, David Root, Pablo Tamayo, Ryo Sakai, Bang Wong, Aravind Subramanian, Todd Golub, Matthew Meyerson, Jesse Boehm. High-throughput gene expression profiling as a generalizable assay for determination of mutation impact on gene function. [abstract]. In: Proceedings of the AACR Special Conference on Computational and Systems Biology of Cancer; Feb 8-11 2015; San Francisco, CA. Philadelphia (PA): AACR; Cancer Res 2015;75(22 Suppl 2):Abstract nr PR04.

Recently, the decline in the cost of genome sequencing has led to the rapid identification of thousands of cancer-associated somatic mutations. However, progress in characterization of these genetic events has lagged significantly behind. Understanding mutation function is critical not only for research purposes but also for determining targeted treatment strategies based on individual tumor genetic profiles, yet determination of mutation impact remains a significant bottleneck. Here we describe a high-throughput approach to classify somatic mutations that is robust, scalable, and requires no prior information of gene function. We generated a lentiviral cDNA expression library of ~550 mutated and wild-type alleles of genes mutated in lung adenocarcinoma and introduced these alleles into four human lung cell lines. 96 hours post-infection, gene expression profiles were generated using Luminex-based L1000 profiling. In total, more than 2000 gene expression signatures were generated. We discovered that gain-of-function mutants induce expression signatures with a greater signal strength or different identity than the corresponding wild-type gene signature. In contrast, loss-of-function mutants could be identified by their incapability to induce strong signatures. Based on these features of signature strength and signature identity, we developed a decision-tree approach to classify mutations as either dominant, loss-of-function, or likely inert. An orthogonal functional approach, an EGFR inhibitor resistance screen, was used as validation. The gene expression approach correctly classified known gain-of-function mutations in KRAS (13/13), EGFR (6/7), and ARAF (2/2) and identified dozens of never-characterized gain-of-function and loss-of-function missense mutations. In addition to rare, dominant mutations in clinically-actionable oncogenes such as PIK3CA and AKT1, we identified unexpected dominant mutations in the transcription factor MAX and the phosphatase subunit PPP2R1A, among others. We also observed a substantial enrichment of loss-of-function mutations in tumor suppressor genes such as STK11, KEAP1, FBXW7, and CASP8 as well as in genes not previously connected to lung adenocarcinoma, including GPR137B and MAPK7. Most genes assayed also harbored variants that are likely inert, further underscoring the importance of characterizing individual variant alleles. The method developed here can, in principle, characterize any genetic variant, independent of prior knowledge of gene function, and should significantly advance the pace of functional characterization of mutations identified from genome sequencing. Citation Format: Alice Berger, Angela Brooks, Xiaoyun Wu, Larson Hogstrom, Itay Tirosh, Federica Piccioni, Mukta Bagul, Cong Zhu, Yashaswi Shretha, David Root, Pablo Tamayo, Ryo Sakai, Bang Wong, Aravind Subramanian, Todd Golub, Matthew Meyerson, Jesse Boehm. High-throughput gene expression profiling as a generalizable assay for determination of mutation impact on gene function. [abstract]. In: Proceedings of the AACR Special Conference on Translation of the Cancer Genome; Feb 7-9, 2015; San Francisco, CA. Philadelphia (PA): AACR; Cancer Res 2015;75(22 Suppl 1):Abstract nr PR12.

For the vast majority of cancer patients, existing knowledge of the function(s) of the mutant genes harbored by their tumor and the dependencies they induce is incomplete or non-existent since most cancer mutations are exceedingly rare. As a result, we now have long lists of candidate alleles but a paucity of targets whose biology is sufficiently well understood to guide therapeutics. Here we present an interim progress report on a pilot effort aiming to create a generalizable framework to systematically map the molecular consequences of cancer variants at scale (Target Accelerator). First, we created an efficient pipeline to generate cancer variants and generated an initial library of 1300 mutant cDNA clones corresponding to variants in lung cancer and diffuse large B-cell lymphoma as well as those nominated by “pan-cancer” computational analyses. Second, we established an industry-scale, next-generation pipeline to generate new cancer models (Cell Line Factory) directly from patient samples. We have leveraged this pipeline to process over 330 samples from 208 patients across 16 cancer types, with over 60% growing through at least 5 population doublings. We show that tumor genomics can be retained in such patient-derived models and that drug testing within 3 months is feasible. In addition, we use combinatorial molecular barcoding to rapidly generate a panel of pathway-primed human tumorigenesis models that are suitable for massively parallel multiplexed tumorigenesis assays in vivo (TumorPlex). We hypothesized that this integrated framework could be utilized to generate meaningful functional hypotheses in a high-throughput manner. To test this hypothesis, we introduced over 1000 cancer mutations into cell models and created gene expression signatures together with phenotypic data. In lung cancer, we show that the mutational impact of mutant alleles with known and unknown functions can be rapidly assessed by comparing signatures of wild-type and mutant alleles. We show that this generalizable approach, which does not require prior knowledge, can place variants of unknown significance into dominant gain-of-function and loss-of-function categories. As a complementary approach, we have used TumorPlex assays to test the tumorigenic potential of 550 mutant alleles nominated by Pan-Cancer computational analyses and discovered unexpected variants in the KRAS, AKT1, MAP2K1, ERBB2, PIK3CB, NFE2L2, FAM200A and POT1 genes as being potently tumorigenic. These proof-of-concept studies demonstrate initial feasibility of mapping cancer variant function at scale. Importantly, they demarcate a path by which mapping variant function and predicting vulnerabilities might soon be possible on a patient-by-patient basis, achieving the promise of precision functional genomics. Citation Format: Alice Berger, Eejung Kim, Angela Brooks, Yashaswi Shrestha, Yuen-Yi Tseng, Xiaoyun Wu, Nina Ilic, Lihua Zou, Atanas Kamburov, Xiaoping Yang, Cong Zhu, Paula Keskula, Sara Seepo, Andrew Hong, John Doench, Aravind Subramanian, Keith Ligon, Philip Kantoff, Katherine Janeway, Levi Garraway, David Root, Todd Golub, Matthew Meyerson, William Hahn, Gad Getz, Jesse Boehm. Towards precision functional genomics via next-generation functional mapping of cancer variants. [abstract]. In: Proceedings of the 106th Annual Meeting of the American Association for Cancer Research; 2015 Apr 18-22; Philadelphia, PA. Philadelphia (PA): AACR; Cancer Res 2015;75(15 Suppl):Abstract nr 957. doi:10.1158/1538-7445.AM2015-957

Deregulated Wnt signaling has been implicated in a wide range of cancer types, both through truncating mutations in the tumor suppressor protein Adenomatous Polyposis Coli (APC), most prevalent in colorectal cancer, as well as overexpression of Wnt ligands and receptors. Canonical Wnt signaling involves the regulated degradation of the beta catenin protein. In the absence of Wnt signaling, cytosolic beta catenin levels are maintained at low levels, as a result of its phosphorylation-dependent ubiquitination and subsequent proteasomal degradation. This degradation is mediated by the ‘destruction complex’, whose constituents include Glycogen Synthase Kinase 3 (GSK3β), Casein Kinase 1 (CK1) and the scaffolding proteins Axin and APC. The efficient assembly of this complex depends upon the steady state levels of its components, such as Axin, which when overexpressed has been shown to decrease beta catenin levels and inhibit Wnt signaling. Recent reports suggest that the poly-ADP-ribosylating Tankyrase enzymes can bind to and regulate Axin levels, hence providing promising targets for treating Wnt-dependent tumors. We have identified potent, low nM small molecule inhibitors of the Wnt pathway, which inhibit Tankyrases 1 and 2, stabilize Axin, deplete beta catenin protein levels and modulate a set of Wnt-regulated genes in colorectal cell lines in a similar manner as seen when using siRNA to beta catenin. Furthermore, using siRNAs to Tankyrase 1 and 2 we have shown that co-depletion of both enzymes is required for Axin stabilization and Wnt pathway inhibition. We have also successfully demonstrated that our lead molecules cause prolonged Axin stabilization and Wnt pathway inhibition in vivo following oral BID dosing in colon xenografts carrying APC mutations. However, in contrast to previous reports, using molecular tools and our lead compounds (e.g. inducible Axin cell lines, compound rescue experiments using siRNAs to Axin 1/2), we have shown that Axin stabilization has no effect on the viability/proliferation of APC mutant colorectal cell lines, under conditions where siRNA to beta catenin was able to cause significant growth inhibition. Some potential mechanistic rationales for this lack of phenotypic effect will be presented. Understanding the consequences of Tankyrase inhibition in vivo in relevant disease models will be important to establish the clinical utility of these molecules in colorectal cancer and beyond. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 102nd Annual Meeting of the American Association for Cancer Research; 2011 Apr 2-6; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2011;71(8 Suppl):Abstract nr 5196. doi:10.1158/1538-7445.AM2011-5196