Single-cell genomics analysis reveals complex genetic interactions in an in vivo model of acquired BRAF inhibitor resistance.

The evolution of therapeutic resistance is a major obstacle to the success of targeted oncology drugs. While both inter- and intratumoral heterogeneity limit our ability to detect resistant subpopulations that pre-exist or emerge during treatment, our ability to analyze tumors with single-cell resolution is limited. Here, we utilized a cell-based transposon mutagenesis method to identify mechanisms of BRAF inhibitor resistance in a model of cutaneous melanoma. This screen identified overexpression of NEDD4L and VGLL3 as significant drivers of BRAF inhibitor resistance in vivo. In addition, we describe a novel single-cell genomics profiling method to genotype thousands of individual cells within tumors driven by transposon mutagenesis. This approach revealed a surprising genetic diversity among xenograft tumors and identified recurrent co-occurring mutations that emerge within distinct tumor subclones. Taken together, these observations reveal an unappreciated genetic complexity that drives BRAF inhibitor resistance.

Understanding cancer drug resistance with Sleeping Beauty functional genomic screens: Application to MAPK inhibition in cutaneous melanoma.

Combined BRAF and MEK inhibition is an effective treatment for BRAF-mutant cutaneous melanoma. However, most patients progress on this treatment due to drug resistance. Here, we applied the Sleeping Beauty transposon system to understand how melanoma evades MAPK inhibition. We found that the specific drug resistance mechanisms differed across melanomas in our genetic screens of five cutaneous melanoma cell lines. While drivers that reactivated MAPK were highly conserved, many others were cell-line specific. One such driver, VAV1, activated a de-differentiated transcriptional program like that of hyperactive RAC1, RAC1P29S. To target this mechanism, we showed that an inhibitor of SRC, saracatinib, blunts the VAV1-induced transcriptional reprogramming. Overall, we highlighted the importance of accounting for melanoma heterogeneity in treating cutaneous melanoma with MAPK inhibitors. Moreover, we demonstrated the utility of the Sleeping Beauty transposon system in understanding cancer drug resistance.

SRC-RAC1 signaling drives drug resistance to BRAF inhibition in de-differentiated cutaneous melanomas.

Rare gain-of-function mutations in RAC1 drive drug resistance to targeted BRAF inhibition in cutaneous melanoma. Here, we show that wildtype RAC1 is a critical driver of growth and drug resistance, but only in a subset of melanomas with elevated markers of de-differentiation. Similarly, SRC inhibition also selectively sensitized de-differentiated melanomas to BRAF inhibition. One possible mechanism may be the suppression of the de-differentiated state, as SRC and RAC1 maintained markers of de-differentiation in human melanoma cells. The functional differences between melanoma subtypes suggest that the clinical management of cutaneous melanoma can be enhanced by the knowledge of differentiation status. To simplify the task of classification, we developed a binary classification strategy based on a small set of ten genes. Using this gene set, we reliably determined the differentiation status previously defined by hundreds of genes. Overall, our study informs strategies that enhance the precision of BRAFi by discovering unique vulnerabilities of the de-differentiated cutaneous melanoma subtype and creating a practical method to resolve differentiation status.

Transposon Mutagenesis Reveals RBMS3 Silencing as a Promoter of Malignant Progression of BRAFV600E-Driven Lung Tumorigenesis.

Mutationally activated BRAF is detected in approximately 7% of human lung adenocarcinomas, with BRAFT1799A serving as a predictive biomarker for treatment of patients with FDA-approved inhibitors of BRAFV600E oncoprotein signaling. In genetically engineered mouse (GEM) models, expression of BRAFV600E in the lung epithelium initiates growth of benign lung tumors that, without additional genetic alterations, rarely progress to malignant lung adenocarcinoma. To identify genes that cooperate with BRAFV600E for malignant progression, we used Sleeping Beauty-mediated transposon mutagenesis, which dramatically accelerated the emergence of lethal lung cancers. Among the genes identified was Rbms3, which encodes an RNA-binding protein previously implicated as a putative tumor suppressor. Silencing of RBMS3 via CRISPR/Cas9 gene editing promoted growth of BRAFV600E lung organoids and promoted development of malignant lung cancers with a distinct micropapillary architecture in BRAFV600E and EGFRL858R GEM models. BRAFV600E/RBMS3Null lung tumors displayed elevated expression of Ctnnb1, Ccnd1, Axin2, Lgr5, and c-Myc mRNAs, suggesting that RBMS3 silencing elevates signaling through the WNT/ß-catenin signaling axis. Although RBMS3 silencing rendered BRAFV600E-driven lung tumors resistant to the effects of dabrafenib plus trametinib, the tumors were sensitive to inhibition of porcupine, an acyltransferase of WNT ligands necessary for their secretion. Analysis of The Cancer Genome Atlas patient samples revealed that chromosome 3p24, which encompasses RBMS3, is frequently lost in non-small cell lung cancer and correlates with poor prognosis. Collectively, these data reveal the role of RBMS3 as a lung cancer suppressor and suggest that RBMS3 silencing may contribute to malignant NSCLC progression. SIGNIFICANCE: Loss of RBMS3 cooperates with BRAFV600E to induce lung tumorigenesis, providing a deeper understanding of the molecular mechanisms underlying mutant BRAF-driven lung cancer and potential strategies to more effectively target this disease.

Machine learning approach informs biology of cancer drug response.

BACKGROUND: The mechanism of action for most cancer drugs is not clear. Large-scale pharmacogenomic cancer cell line datasets offer a rich resource to obtain this knowledge. Here, we present an analysis strategy for revealing biological pathways that contribute to drug response using publicly available pharmacogenomic cancer cell line datasets. METHODS: We present a custom machine-learning based approach for identifying biological pathways involved in cancer drug response. We test the utility of our approach with a pan-cancer analysis of ML210, an inhibitor of GPX4, and a melanoma-focused analysis of inhibitors of BRAFV600. We apply our approach to reveal determinants of drug resistance to microtubule inhibitors. RESULTS: Our method implicated lipid metabolism and Rac1/cytoskeleton signaling in the context of ML210 and BRAF inhibitor response, respectively. These findings are consistent with current knowledge of how these drugs work. For microtubule inhibitors, our approach implicated Notch and Akt signaling as pathways that associated with response. CONCLUSIONS: Our results demonstrate the utility of combining informed feature selection and machine learning algorithms in understanding cancer drug response.

Single allele loss-of-function mutations select and sculpt conditional cooperative networks in breast cancer.

The most common events in breast cancer (BC) involve chromosome arm losses and gains. Here we describe identification of 1089 gene-centric common insertion sites (gCIS) from transposon-based screens in 8 mouse models of BC. Some gCIS are driver-specific, others driver non-specific, and still others associated with tumor histology. Processes affected by driver-specific and histology-specific mutations include well-known cancer pathways. Driver non-specific gCIS target the Mediator complex, Ca++ signaling, Cyclin D turnover, RNA-metabolism among other processes. Most gCIS show single allele disruption and many map to genomic regions showing high-frequency hemizygous loss in human BC. Two gCIS, Nf1 and Trps1, show synthetic haploinsufficient tumor suppressor activity. Many gCIS act on the same pathway responsible for tumor initiation, thereby selecting and sculpting just enough and just right signaling. These data highlight ~1000 genes with predicted conditional haploinsufficient tumor suppressor function and the potential to promote chromosome arm loss in BC.

A Genetic Screen to Identify Gain- and Loss-of-Function Modifications that Enhance T-cell Infiltration into Tumors.

T-cell-mediated cancer immunotherapies, including anti-PD-1 and T cells expressing chimeric antigen receptors (CAR-T cells), are becoming standard treatments for many cancer types. CAR-T therapy, in particular, has been successful in treating circulating, but not solid, tumors. One challenge limiting immunotherapy success is that tumors lacking T-cell infiltration do not respond to treatment. Therefore, one potential strategy to overcome resistance is to enhance the ability of T cells to traffic into tumors. Here, we describe an unbiased in vivo genetic screen approach utilizing the Sleeping Beauty mutagenesis system to identify candidate genes in T cells that might be modified to drive intratumoral T-cell accumulation. This screen identified over 400 candidate genes in three tumor models. These results indicated substantial variation in gene candidate selection, depending on the tumor model and whether or not mice were treated with anti-PD-1, yet some candidate genes were identified in all tumor models and with anti-PD-1 therapy. Inhibition of the most frequently mutated gene, Aak1, affected chemokine receptor expression and enhanced T-cell trafficking in vitro and in vivo Screen candidates should be further validated as therapeutic targets, with particular relevance to enhancing infiltration of adoptively transferred T cells into solid tumors.

Functional Genomic Screening Independently Identifies CUL3 as a Mediator of Vemurafenib Resistance via Src-Rac1 Signaling Axis.

Patients with malignant melanoma have a 5-year survival rate of only 15-20% once the tumor has metastasized to distant tissues. While MAP kinase pathway inhibitors (MAPKi) are initially effective for the majority of patients with melanoma harboring BRAFV600E mutation, over 90% of patients relapse within 2 years. Thus, there is a critical need for understanding MAPKi resistance mechanisms. In this manuscript, we performed a forward genetic screen using a whole genome shRNA library to identify negative regulators of vemurafenib resistance. We identified loss of NF1 and CUL3 as drivers of vemurafenib resistance. NF1 is a known driver of vemurafenib resistance in melanoma through its action as a negative regulator of RAS. However, the mechanism by which CUL3, a key protein in E3 ubiquitin ligase complexes, is involved in vemurafenib resistance was unknown. We found that loss of CUL3 was associated with an increase in RAC1 activity and MEKS298 phosphorylation. However, the addition of the Src family inhibitor saracatinib prevented resistance to vemurafenib in CUL3KD cells and reversed RAC1 activation. This finding suggests that inhibition of the Src family suppresses MAPKi resistance in CUL3KD cells by inactivation of RAC1. Our results also indicated that the loss of CUL3 does not promote the activation of RAC1 through stabilization, suggesting that CUL3 is involved in the stability of upstream regulators of RAC1. Collectively, our study identifies the loss of CUL3 as a driver of MAPKi resistance through activation of RAC1 and demonstrates that inhibition of the Src family can suppress the MAPKi resistance phenotype in CUL3KD cells by inactivating RAC1 protein.

Disrupting Mitochondrial Pyruvate Uptake Directs Glutamine into the TCA Cycle away from Glutathione Synthesis and Impairs Hepatocellular Tumorigenesis.

Hepatocellular carcinoma (HCC) is a devastating cancer increasingly caused by non-alcoholic fatty liver disease (NAFLD). Disrupting the liver Mitochondrial Pyruvate Carrier (MPC) in mice attenuates NAFLD. Thus, we considered whether liver MPC disruption also prevents HCC. Here, we use the N-nitrosodiethylamine plus carbon tetrachloride model of HCC development to test how liver-specific MPC knock out affects hepatocellular tumorigenesis. Our data show that liver MPC ablation markedly decreases tumorigenesis and that MPC-deficient tumors transcriptomically downregulate glutathione metabolism. We observe that MPC disruption and glutathione depletion in cultured hepatomas are synthetically lethal. Stable isotope tracing shows that hepatocyte MPC disruption reroutes glutamine from glutathione synthesis into the tricarboxylic acid (TCA) cycle. These results support a model where inducing metabolic competition for glutamine by MPC disruption impairs hepatocellular tumorigenesis by limiting glutathione synthesis. These findings raise the possibility that combining MPC disruption and glutathione stress may be therapeutically useful in HCC and additional cancers.

Src-Dependent DBL Family Members Drive Resistance to Vemurafenib in Human Melanoma.

The use of selective BRAF inhibitors (BRAFi) has produced remarkable outcomes for patients with advanced cutaneous melanoma harboring a BRAFV600E mutation. Unfortunately, the majority of patients eventually develop drug-resistant disease. We employed a genetic screening approach to identify gain-of-function mechanisms of BRAFi resistance in two independent melanoma cell lines. Our screens identified both known and unappreciated drivers of BRAFi resistance, including multiple members of the DBL family. Mechanistic studies identified a DBL/RAC1/PAK signaling axis capable of driving resistance to both current and next-generation BRAFis. However, we show that the SRC inhibitor, saracatinib, can block the DBL-driven resistance. Our work highlights the utility of our straightforward genetic screening method in identifying new drug combinations to combat acquired BRAFi resistance. SIGNIFICANCE: A simple, rapid, and flexible genetic screening approach identifies genes that drive resistance to MAPK inhibitors when overexpressed in human melanoma cells.

Sleeping Beauty Mouse Models of Cancer: Microenvironmental Influences on Cancer Genetics.

The Sleeping Beauty (SB) transposon insertional mutagenesis system offers a streamlined approach to identify genetic drivers of cancer. With a relatively random insertion profile, SB is uniquely positioned for conducting unbiased forward genetic screens. Indeed, SB mouse models of cancer have revealed insights into the genetics of tumorigenesis. In this review, we highlight experiments that have exploited the SB system to interrogate the genetics of cancer in distinct biological contexts. We also propose experimental designs that could further our understanding of the relationship between tumor microenvironment and tumor progression.

A simplified transposon mutagenesis method to perform phenotypic forward genetic screens in cultured cells.

BACKGROUND: The introduction of genome-wide shRNA and CRISPR libraries has facilitated cell-based screens to identify loss-of-function mutations associated with a phenotype of interest. Approaches to perform analogous gain-of-function screens are less common, although some reports have utilized arrayed viral expression libraries or the CRISPR activation system. However, a variety of technical and logistical challenges make these approaches difficult for many labs to execute. In addition, genome-wide shRNA or CRISPR libraries typically contain of hundreds of thousands of individual engineered elements, and the associated complexity creates issues with replication and reproducibility for these methods. RESULTS: Here we describe a simple, reproducible approach using the SB transposon system to perform phenotypic cell-based genetic screens. This approach employs only three plasmids to perform unbiased, whole-genome transposon mutagenesis. We also describe a ligation-mediated PCR method that can be used in conjunction with the included software tools to map raw sequence data, identify candidate genes associated with phenotypes of interest, and predict the impact of recurrent transposon insertions on candidate gene function. Finally, we demonstrate the high reproducibility of our approach by having three individuals perform independent replicates of a mutagenesis screen to identify drivers of vemurafenib resistance in cultured melanoma cells. CONCLUSIONS: Collectively, our work establishes a facile, adaptable method that can be performed by labs of any size to perform robust, genome-wide screens to identify genes that influence phenotypes of interest.

Insertional mutagenesis using the Sleeping Beauty transposon system identifies drivers of erythroleukemia in mice.

Insertional mutagenesis is a powerful means of identifying cancer drivers in animal models. We used the Sleeping Beauty (SB) transposon/transposase system to identify activated oncogenes in hematologic cancers in wild-type mice and mice that express a stabilized cyclin E protein (termed cyclin ET74AT393A). Cyclin E governs cell division and is misregulated in human cancers. Cyclin ET74AT393A mice develop ineffective erythropoiesis that resembles early-stage human myelodysplastic syndrome, and we sought to identify oncogenes that might cooperate with cyclin E hyperactivity in leukemogenesis. SB activation in hematopoietic precursors caused T-cell leukemia/lymphomas (T-ALL) and pure red blood cell erythroleukemias (EL). Analysis of >12,000 SB integration sites revealed markedly different oncogene activations in EL and T-ALL: Notch1 and Ikaros were most common in T-ALL, whereas ETS transcription factors (Erg and Ets1) were targeted in most ELs. Cyclin E status did not impact leukemogenesis or oncogene activations. Whereas most SB insertions were lost during culture of EL cell lines, Erg insertions were retained, indicating Erg's key role in these neoplasms. Surprisingly, cyclin ET74AT393A conferred growth factor independence and altered Erg-dependent differentiation in EL cell lines. These studies provide new molecular insights into erythroid leukemia and suggest potential therapeutic targets for human leukemia.

A comprehensive evaluation of Hippo pathway silencing in sarcomas.

TAZ and YAP are transcriptional coactivators negatively regulated by the Hippo pathway that have emerged as key oncoproteins in several cancers including sarcomas. We hypothesized that loss of expression of the Hippo kinases might be a mechanism of activating TAZ and YAP. By immunohistochemistry, TAZ/YAP activated clinical sarcoma samples demonstrated loss of MST1 (47%), MST2 (26%), LATS1 (19%), and LATS2 (27%). Western blot similarly demonstrated loss of MST1 (58%), MST2 (25%), and LATS2 (17%). Treatment with MG132 demonstrated an accumulation of MST2 in 25% of sarcoma cell lines, indicating that proteosomal degradation regulates MST2 expression. qRT-PCR in sarcoma cell lines demonstrated loss of expression of the Hippo kinases at the RNA level, most pronounced in MST1 (42%) and MST2 (25%). 5-azacytidine treatment in sarcoma cell lines modestly reversed expression of predominantly MST1 (8%) and MST2 (17%), indicating CpG island hypermethylation can silence expression of MST1 and MST2. Trichostatin A treatment reversed expression of MST1 (58%) and MST2 (67%), indicating histone deacetylation also plays a role in silencing expression of MST1 and MST2. Loss of expression of the Hippo kinases is frequent in sarcomas and is due to a variety of mechanisms including regulation at the post-translational level and epigenetic silencing.

A Forward Genetic Screen Targeting the Endothelium Reveals a Regulatory Role for the Lipid Kinase Pi4ka in Myelo- and Erythropoiesis.

Given its role as the source of definitive hematopoietic cells, we sought to determine whether mutations initiated in the hemogenic endothelium would yield hematopoietic abnormalities or malignancies. Here, we find that endothelium-specific transposon mutagenesis in mice promotes hematopoietic pathologies that are both myeloid and lymphoid in nature. Frequently mutated genes included previously recognized cancer drivers and additional candidates, such as Pi4ka, a lipid kinase whose mutation was found to promote myeloid and erythroid dysfunction. Subsequent validation experiments showed that targeted inactivation of the Pi4ka catalytic domain or reduction in mRNA expression inhibited myeloid and erythroid cell differentiation in vitro and promoted anemia in vivo through a mechanism involving deregulation of AKT, MAPK, SRC, and JAK-STAT signaling. Finally, we provide evidence linking PI4KAP2, previously considered a pseudogene, to human myeloid and erythroid leukemia.

Chronic liver injury alters driver mutation profiles in hepatocellular carcinoma in mice.

Most hepatocellular carcinomas (HCCs) develop in a chronically injured liver, yet the extent to which this microenvironment promotes neoplastic transformation or influences selective pressures for genetic drivers of HCC remains unclear. We sought to determine the impact of hepatic injury in an established mouse model of HCC induced by Sleeping Beauty transposon mutagenesis. Chemically induced chronic liver injury dramatically increased tumor penetrance and significantly altered driver mutation profiles, likely reflecting distinct selective pressures. In addition to established human HCC genes and pathways, we identified several injury-associated candidates that represent promising loci for further study. Among them, we found that FIGN is overexpressed in human HCC and promotes hepatocyte invasion. We also validated Gli2's oncogenic potential in vivo, providing direct evidence that Hedgehog signaling can drive liver tumorigenesis in the context of chronic injury. Finally, we show that a subset of injury-associated candidate genes identifies two distinct classes of human HCCs. Further analysis of these two subclasses revealed significant trends among common molecular classification schemes of HCC. The genes and mechanisms identified here provide functional insights into the origin of HCC in a chronic liver damage environment. CONCLUSION: A chronically damaged liver microenvironment influences the genetic mechanisms that drive hepatocarcinogenesis. (Hepatology 2018;67:924-939).

Sleeping Beauty Insertional Mutagenesis in Mice Identifies Drivers of Steatosis-Associated Hepatic Tumors.

Hepatic steatosis is a strong risk factor for the development of hepatocellular carcinoma (HCC), yet little is known about the molecular pathology associated with this factor. In this study, we performed a forward genetic screen using Sleeping Beauty (SB) transposon insertional mutagenesis in mice treated to induce hepatic steatosis and compared the results to human HCC data. In humans, we determined that steatosis increased the proportion of female HCC patients, a pattern also reflected in mice. Our genetic screen identified 203 candidate steatosis-associated HCC genes, many of which are altered in human HCC and are members of established HCC-driving signaling pathways. The protein kinase A/cyclic AMP signaling pathway was altered frequently in mouse and human steatosis-associated HCC. We found that activated PKA expression drove steatosis-specific liver tumorigenesis in a mouse model. Another candidate HCC driver, the N-acetyltransferase NAT10, which we found to be overexpressed in human steatosis-associated HCC and associated with decreased survival in human HCC, also drove liver tumorigenesis in a steatotic mouse model. This study identifies genes and pathways promoting HCC that may represent novel targets for prevention and treatment in the context of hepatic steatosis, an area of rapidly growing clinical significance. Cancer Res; 77(23); 6576-88. ©2017 AACR.

A Trp53fl/flPtenfl/fl mouse model of undifferentiated pleomorphic sarcoma mediated by adeno-Cre injection and in vivo bioluminescence imaging.

Genetic mouse models of soft tissue sarcoma provide critical insights into disease pathophysiology, which are oftentimes unable to be extracted from human tumor samples or xenograft models. In this study we describe a mouse model of soft tissue sarcoma mediated by adenoviral-Cre recombinase injection into Trp53fl/fl/Ptenfl/fl lox-stop-lox luciferase mice. Injection of adenovirus expressing Cre recombinase, either subcutaneously or intramuscularly in two experimental groups, results in viral infection and gene recombination with 100% penetrance within the first 24 hours following injection. Luciferase expression measured by real-time bioluminescence imaging increases over time, with an initial robust increase following viral injection, followed by a steady rise over the next several weeks as primary tumors develop and grow. Intramuscular injections were more commonly associated with evidence of systemic viral distribution than subcutaneous injections. All mice developed soft tissue sarcomas at the primary injection site, with histological examination identifying 93% of tumors as invasive pleomorphic sarcomas based on microscopic morphology and immunohistochemical expression of sarcoma markers. A lymphocytic infiltrate was present in 64% of the sarcomas in this immunocompetent model and 71% of tumors expressed PD-L1. This is the first report of a viral-Cre mediated Trp53/Pten mouse model of undifferentiated pleomorphic sarcoma. The bioluminescence imaging feature, along with high penetrance of the model and its immunological characteristics, makes it suited for pre-clinical studies of soft tissue sarcoma.

Transposon mutagenesis identifies chromatin modifiers cooperating with Ras in thyroid tumorigenesis and detects ATXN7 as a cancer gene.

Oncogenic RAS mutations are present in 15-30% of thyroid carcinomas. Endogenous expression of mutant Ras is insufficient to initiate thyroid tumorigenesis in murine models, indicating that additional genetic alterations are required. We used Sleeping Beauty (SB) transposon mutagenesis to identify events that cooperate with HrasG12V in thyroid tumor development. Random genomic integration of SB transposons primarily generated loss-of-function events that significantly increased thyroid tumor penetrance in Tpo-Cre/homozygous FR-HrasG12V mice. The thyroid tumors closely phenocopied the histological features of human RAS-driven, poorly differentiated thyroid cancers. Characterization of transposon insertion sites in the SB-induced tumors identified 45 recurrently mutated candidate cancer genes. These mutation profiles were remarkably concordant with mutated cancer genes identified in a large series of human poorly differentiated and anaplastic thyroid cancers screened by next-generation sequencing using the MSK-IMPACT panel of cancer genes, which we modified to include all SB candidates. The disrupted genes primarily clustered in chromatin remodeling functional nodes and in the PI3K pathway. ATXN7, a component of a multiprotein complex with histone acetylase activity, scored as a significant SB hit. It was recurrently mutated in advanced human cancers and significantly co-occurred with RAS or NF1 mutations. Expression of ATXN7 mutants cooperated with oncogenic RAS to induce thyroid cell proliferation, pointing to ATXN7 as a previously unrecognized cancer gene.

A Transposon-based Analysis Reveals RASA1 Is Involved in Triple-Negative Breast Cancer.

RAS genes are mutated in 20% of human tumors, but these mutations are very rare in breast cancer. Here, we used a mouse model to generate tumors upon activation of a mutagenic T2Onc2 transposon via expression of a transposase driven by the keratin K5 promoter in a p53+/- background. These animals mainly developed mammary tumors, most of which had transposon insertions in one of two RASGAP genes, neurofibromin1 (Nf1) and RAS p21 protein activator (Rasa1). Immunohistochemical analysis of a collection of human breast tumors confirmed that low expression of RASA1 is frequent in basal (triple-negative) and estrogen receptor negative tumors. Bioinformatic analysis of human breast tumors in The Cancer Genome Atlas database showed that although RASA1 mutations are rare, allelic loss is frequent, particularly in basal tumors (80%) and in association with TP53 mutation. Inactivation of RASA1 in MCF10A cells resulted in the appearance of a malignant phenotype in the context of mutated p53. Our results suggest that alterations in the Ras pathway due to the loss of negative regulators of RAS may be a common event in basal breast cancer. Cancer Res; 77(6); 1357-68. ©2017 AACR.