In This Issue

See article, p. 714.

Cancer genome sequencing projects have identified oncogenes and tumor suppressor genes that occur at high frequency but also thousands of mutant alleles that occur at low frequency. To systematically differentiate functionally relevant alleles from neutral, passenger mutations, Kim, Ilic, Shrestha, Zou, Kamburov, and colleagues tested a panel of 474 cancer-associated gene variants identified in a cohort of 5,338 tumors representing 27 different cancer types. In vivo pooled assessment of the tumorigenic potential of the candidate mutant alleles expressed in an immortalized human embryonic kidney cell line revealed that tumors derived from pools with known oncogenic alleles repeatedly demonstrated a similar pattern of allele representation, whereas tumors derived from other pools demonstrated a wide diversity of allele representation. Alleles that were found at a frequency of more than 1% in at least 2 tumors or more than 90% in at least 1 tumor were considered tumorigenic. In vitro gene expression profiling of cells expressing each allele identified alleles that induced changes in gene expression that correlated with or were distinct from those of wild-type or known gain-of-function counterparts. Together, these phenotypic assays identified 12 oncogenic alleles, including 1 rare KRAS allele (KRAS^D33E) as well as 2 alleles (PIK3CB^A1048V and POT1^G76V) in genes that had not previously been identified as oncogenic. This pilot study provides a framework for a systematic and scalable approach that can complement genomic sequencing efforts to characterize variants of unknown significance and distinguish driver and passenger mutations.

See article, p. 727.

Isocitrate dehydrogenase 1 (IDH1) and IDH2 mutations are among the most frequent genetic alterations in intrahepatic cholangiocarcinomas (ICC), which are highly lethal bile duct malignancies with a poor prognosis. To identify synthetic lethal interactions in IDH mutant ICCs, Saha and colleagues performed a high-throughput screen of clinically relevant compounds against 17 human biliary tract cancer cell lines, two of which harbor IDH1 mutations, and compared drug response profiles to those generated for 682 solid tumor cell lines. Both IDH1 mutant biliary tract cancer cell lines exhibited a distinct drug response profile, most strikingly hypersensitivity to the SRC family kinase inhibitor dasatinib, compared to IDH WT biliary tract cancer and solid tumor cell lines. IDH mutant–associated dasatinib hypersensitivity was confirmed in additional low-passage patient-derived ICC cell lines, and a cell line derived from a genetically engineered mouse model (GEMM) of ICC harboring Idh2 and Kras mutations. Consistent with these findings, an IDH mutant ICC PDX and the ICC GEMM exhibited extensive necrosis in response to dasatinib treatment. Proteomic characterization of the activated kinome and analysis of CRISPR/Cas9-generated gatekeeper mutations identified SRC as the main target of dasatinib in IDH mutant ICC, and depletion of SRC revealed that SRC drives mTOR-S6K–mediated growth and survival of IDH mutant ICC. Together, these results identify a critical driver of IDH mutant ICC and a potential new therapeutic target for patients with IDH mutant ICC.

See article, p. 740.

Cyclin-dependent kinases (CDK) 4 and 6 drive progression from G1- to S-phase through phosphorylation of the RB tumor suppressor protein and are frequently activated in cancer. Abemaciclib is a small-molecule CDK4/6 inhibitor that has shown preclinical antitumorigenic activity, prompting Patnaik and colleagues to evaluate the safety, tolerability, and efficacy of orally delivered abemaciclib in 225 patients with advanced solid tumors in a first-in-human multicenter phase I dose-escalation trial with tumor-specific cohorts. The primary objective was to evaluate safety and tolerability, and secondary objectives included analysis of pharmacokinetics, predictive biomarkers, and antitumor activity, and determination of a recommended dose range. Abemaciclib was generally well tolerated with few serious adverse events, which allowed continuous dosing. Reduced levels of phosphorylated RB and the S-phase marker topoisomerase II were observed in epidermal keratinocytes and in fresh tumor biopsies, and a ≥60% decrease in phosphorylated RB in skin was correlated with clinical benefit in patients with hormone receptor–positive breast cancer. In tumor-specific cohorts, abemaciclib monotherapy elicited radiographic responses and a 61% clinical benefit rate in HR-positive breast tumors, and had a better disease control rate in patients with non–small cell lung cancer (NSCLC) with KRAS mutations (55%) compared to those with wild-type KRAS (39%). A partial response as well as stable disease was observed among patients with melanoma; stable disease also occurred in patients with ovarian cancer, glioblastoma, and colorectal cancer. Together, these data highlight the overall safety and efficacy of abemaciclib in patients with solid tumors and support its further clinical development.

See article, p. 754.

Despite the frequent occurrence of activating KRAS mutations in non–small cell lung cancer (NSCLC), therapeutic strategies targeting KRAS effectors such as MEK have had limited efficacy due to adaptive resistance. Kitai, Ebi, and colleagues found that resistance to MEK inhibition in KRAS-mutant NSCLC was mediated by feedback reactivation of AKT and ERK signaling driven by differential receptor tyrosine kinase activation. In epithelial-like KRAS-mutant lung cancer, activation of ERBB3 in response to MEK inhibition enhanced AKT and ERK signaling. In contrast, mesenchymal-like KRAS-mutant cell lines lacked ERBB3 expression but exhibited high expression of FGFR1. MEK inhibitor treatment in mesenchymal-like cells suppressed the expression of Sprouty 4, an antagonist of FGFR1 signaling, leading to feedback induction of FGFR1-driven FGFR substrate 2 (FRS2) signaling and downstream reactivation of AKT and ERK. Consistent with these findings, combined treatment of mesenchymal-like, but not epithelial-like, KRAS-mutant cells with MEK and FGFR inhibitors resulted in greater suppression of ERK and AKT, increased apoptosis, and tumor regression in vivo compared with MEK inhibitor monotherapy. Furthermore, elevated FGFR1 expression was associated with the mesenchymal phenotype and inversely correlated with ERBB3 expression in a large cohort of primary KRAS-mutant lung adenocarcinomas. Taken together, these results demonstrate that MEK inhibition induces distinct feedback activation pathways in KRAS-mutant NSCLC depending on EMT status and suggests that EMT may be useful as a biomarker to predict response to combined MEK/FGFR1 blockade in patients with mesenchymal-like tumors.

See article, p. 770.

Translocations involving the mixed lineage leukemia (MLL) histone methyltransferase are frequently observed in leukemia and underscore the role of deregulated histone methylation in tumorigenesis. MLL-associated leukemia depends on lens epithelium-derived growth factor (LEDGF), which binds methylated histone H3 lysine 36 (H3K36), a marker of transcriptional activation and elongation. However, the role of H3K36 dimethylation (H3K36me2) in MLL leukemia remains unclear. Zhu and colleagues found that LEDGF preferentially bound H3K36me2 at nucleosomes, colocalized with MLL and H3K36me2 around the transcription start site of MLL target genes, and was required for chromatin retention of wild-type MLL and protein complexes that bind the MLL fusion protein. Reciprocally, LEDGF binding at proximal promoter regions of MLL target genes was mutually dependent on MLL. Knockdown of the histone methyltransferase absent, small, or homeotic discs 1–like (ASH1L) diminished H3K36me2 and binding of LEDGF and wild-type MLL at MLL target gene promoters. Consistent with this finding, ASH1L knockdown inhibited expression of the MLL target gene HOXA9 and suppressed the in vitro growth and in vivo leukemogenicity of MLL-transformed cells. In contrast, SET domain–containing 2 (SETD2)-mediated H3K36 trimethylation did not affect MLL/LEDGF recruitment or MLL-induced transformation. However, overexpression of the histone demethylase lysine (K)-specific demethylase 2A (KDM2A), but not KDM2B, reduced H3K36me2 and occupancy of LEDGF and wild-type MLL at MLL target genes and suppressed MLL-induced transformation. These data highlight the importance of ASH1L-mediated H3K36me2 in driving MLL-associated gene expression and leukemogenesis and support ASH1L as a potential therapeutic target.

See article, p. 784.

Multiple long noncoding RNAs (lncRNA) have been reported to play a role in carcinogenesis; however, the specific mechanisms of lncRNAs in gastric cancer remain unclear. Sun, He, and colleagues profiled lncRNA expression in gastric tumors and paired normal tissue, and compared the differentially expressed candidate lncRNAs with clinical outcome. The gastric cancer–associated lncRNA1 (GClnc1) was associated with poor prognosis in patients with gastric cancer, and its expression increased throughout tumor progression. GClnc1 depletion in gastric cancer cells reduced cell proliferation and invasive ability, and increased sensitivity to chemotherapeutics. Similarly, suppression of GClnc1 in tumor xenografts resulted in reduced tumor weight and decreased the number of metastatic lung foci. GClnc1 bound to the histone acetyltransferase KAT2A and the histone methyltransferase complex component WDR5, and was critical for their interaction. GClnc1 knockdown resulted in reduced WDR5/KAT2A co-occupancy of promoters, concordant with reduced trimethylation of histone 3 at lysine 4 and acetylation of histone 3 at lysine 9, and dysregulated expression of WDR5/KAT2A-bound genes including superoxide dismutase 2 (SOD2). Consistent with these findings, GClnc1 promoted SOD2 expression, and SOD2 knockdown reduced gastric cancer cell proliferation and inhibited tumor xenograft growth. Together, these findings indicate that the lncRNA GClnc1 promotes gastric cancer by acting as a scaffold for the recruitment of histone modifying complexes to upregulate the transcription of SOD2, and suggest that GClnc1 or SOD2 may be therapeutic targets in gastric cancer.

Note: In This Issue is written by Cancer Discovery editorial staff. Readers are encouraged to consult the original articles for full details.