In This Issue

See article, p. 1155.

Oncogenic mutations in the EGFR kinase domain frequently occur in non–small cell lung cancers (NSCLC) and confer sensitivity to EGFR tyrosine kinase inhibitors (TKI). In an effort to identify additional therapeutically actionable genetic alterations, Gallant and colleagues performed next-generation sequencing of a metastatic lung adenocarcinoma obtained from a 33-year-old male never-smoker and identified an in-frame tandem kinase domain duplication in EGFR (EGFR-KDD) involving exons 18-25. Analysis of targeted sequencing results from large tumor databases revealed recurrent EGFR-KDD mutations in various cancers, including lung cancer and glioma. Similar to cells expressing the known oncogenic EGFR^L858R mutation, overexpression of EGFR-KDD led to constitutive autophosphorylation, enhanced activation of downstream signaling, and increased anchorage-independent growth compared with cells expressing wild-type EGFR. Computational modeling suggested that the tandem kinase domains could form asymmetric dimers, thereby potentially leading to intramolecular EGFR activation. Furthermore, treatment of EGFR-KDD–expressing cells with EGFR TKIs, including erlotinib, afatinib, and AZD9291, inhibited downstream MAPK signaling and cell proliferation in a dose-dependent manner. In line with this finding, treatment of the index patient with afatinib promoted significant tumor shrinkage and symptomatic improvement until disease progression, at which time the resistant tumor was genomically characterized by further amplification of the EGFR-KDD allele. Together, these data identify EGFR-KDD as an oncogenic driver in NSCLC that confers sensitivity to EGFR inhibitors.

See article, p. 1164.

Brain metastases frequently occur in patients with melanoma, breast cancer, or lung cancer and are associated with a high morbidity rate due to limited treatment options. Moreover, the degree of heterogeneity between primary tumors and metastatic brain lesions remains unknown, prompting Brastianos, Carter, and colleagues to perform whole-exome sequencing on 86 trios of patient-matched primary tumors, brain metastases, and normal samples, including 15 cases consisting of multiple brain lesions, distal extracranial metastases, or associated regional lymph nodes. Consistent with previous reports, branched evolution patterns were observed in all clonally related primary tumors and brain metastases. In addition, potentially therapeutically tractable alterations were detected specifically in brain metastases in 46 (53%) of 86 patients, including alterations that predicted sensitivity to inhibitors targeting cyclin-dependent kinases, the PI3K–AKT–mTOR pathway, HER2/EGFR signaling, and the MAPK pathway. Furthermore, analysis of multiple biopsy sites from single brain metastases and anatomically or temporally distinct brain metastases from the same patient revealed shared mutations, including driver and potentially targetable alterations, which were not detected in the primary tumor sample, suggesting that brain metastases are genetically homogenous. In contrast, extracranial and regional lymph node metastases were characterized by varying degrees of relatedness to and genetic divergence from brain metastases. Together, these data suggest that genomic profiling of brain metastases may help guide therapeutic strategy by highlighting potentially clinically actionable mutations that are absent from the primary tumor.

See article, p. 1178.

Papillary thyroid cancers (PTC) harbor mutually exclusive activating mutations in BRAF, RAS, or receptor tyrosine kinase oncogenes that drive tumorigenesis. In addition, The Cancer Genome Atlas recently identified ch22q loss as a high-frequency event in RAS-mutant PTC. The deleted 22q region contains the tumor suppressor NF2, which is mutated in various cancers and encodes merlin, a protein that inhibits cell growth in response to cell contact, suggesting that NF2 loss contributes to PTC tumorigenesis. Garcia-Rendueles, Ricarte-Filho, and colleagues analyzed 83 advanced thyroid tumors and 43 thyroid cancer cell lines by exon capture next-generation sequencing and found that 22q LOH is also a common event in poorly differentiated thyroid cancer (PDTC) and is preferentially associated with RAS-mutant tumors. In mice, neither thyroid-specific activation of Hras nor Nf2 deletion was sufficient for transformation, but their combined disruption led to highly penetrant PDTC characterized by increased MAPK signaling. Inactivation of the Hippo pathway in NF2/merlin–deficient cells allowed YAP to form transcriptionally active complexes with TEAD, leading to increased YAP-mediated transcription of the three RAS isoforms. Inhibition of the YAP–TEAD association with verteporfin blocked YAP-dependent oncogenic and wild-type RAS transcription and MAPK signaling and suppressed cell growth. Additionally, NF2 loss sensitized RAS-mutant thyroid cancer cells to MEK inhibitors, which diminished the growth of Hras-mutant; Nf2-null murine PDTCs in mice. These results identify YAP as a critical effector of RAS-induced tumorigenesis and suggest that inhibition of YAP or MEK may be effective in RAS-driven thyroid cancer.

See article, p. 1194.

The mTOR complex 2 (mTORC2) is hyperactivated in many cancers, but the mechanism of mTORC2 activation leading to increased AKT signaling is unclear. Liu and colleagues found that MAPK-associated protein 1 (MAPKAP1, also known as SIN1), a unique mTORC2 component, bound the mTOR kinase domain through its pleckstrin homology (PH) domain, resulting in suppression of mTORC2 kinase activity. In addition, the SIN1 PH domain interacted with phosphatidylinositol (3,4,5)-trisphosphate [PtdIns(3,4,5)P₃] at the plasma membrane upon growth factor stimulation. Association of the SIN1 PH domain with PtdIns(3,4,5)P₃ relieved SIN1 PH domain–dependent mTORC2 inhibition, indicating that the mTOR kinase domain and PtdIns(3,4,5)P₃ may bind SIN1 in a mutually exclusive manner. Mutation of three SIN1 residues predicted to be critical for SIN1 PH domain–PtdIns(3,4,5)P₃ binding largely inactivated mTORC2 and impaired downstream AKT signaling in response to growth factor stimulation. Furthermore, ovarian cancer cells expressing triple-mutant SIN1 in the absence of endogenous SIN1 exhibited attenuated mTORC2 activity and reduced colony growth in soft agar. Conversely, several cancer patient–derived SIN1 PH domain mutants were associated with compromised mTOR kinase domain binding, which resulted in enhanced mTORC2 activity, and subsequently elevated AKT phosphorylation, reduced apoptosis in response to DNA-damaging agents, and increased growth in soft-agar and xenograft assays. These results indicate that the SIN1 PH domain interacts with and inhibits mTORC2, which can be released by SIN1–PtdIns(3,4,5)P₃ binding or patient-derived SIN1 PH domain mutations to drive cell growth and transformation.

See article, p. 1210.

Heterogeneous responses to targeted therapies and acquired drug resistance emphasize the need to better understand the specific contexts that define cancer cell responses to small molecules. To identify genetic predictors of small-molecule sensitivity, Seashore-Ludlow and colleagues screened a panel of 860 genetically characterized cancer cell lines with 481 well-annotated FDA-approved drugs, clinical agents, and small-molecule probes. Unsupervised hierarchical clustering and annotated cluster multidimensional enrichment (ACME) analysis of growth curves systematically revealed sensitivity “hotspots,” namely groups of cell lines with cellular or genetic features in common that showed similar responses to several compounds having the same protein target. Previously reported cancer cell line vulnerabilities were validated using this approach, including differential sensitivity of BRAF-mutant melanoma cell lines to BRAF and MEK inhibitors and sensitivity of ERRB2-amplified breast cancer cell lines to ERBB2/HER2 inhibitors. Sensitivity clustering within this dataset also identified mechanisms of action of various small molecules, including compounds with high specificity, drugs targeting different nodes within the same pathway, and clusters of compounds with previously unappreciated activities as mitotic inhibitors or dual kinase–bromodomain inhibitors. Furthermore, ACME analysis also suggested possible cell context–specific effects, including enhanced sensitivity of ALK-driven neuroblastomas to dual inhibition of ALK and IGF1R, and synergistic drug combinations, such as IGF1R and MEK inhibitors in KRAS-mutant cancer cells. Taken together, this sensitivity dataset and analytical methodology provide a powerful approach to uncover cancer cell line vulnerabilities and provide insight into small-molecule mechanisms of action.

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