See article, p. 713.

  • EGFRT790M-positive rociletinib-resistant tumors were biopsied to identify resistance mechanisms.

  • Outgrowth of EGFRT790–wild-type subclones conferred acquired resistance in six of 13 cases.

  • Quantification of EGFRT790 heterogeneity may predict clinical response to rociletinib treatment.

Third-generation EGFR inhibitors such as rociletinib, which specifically target the EGFRT790M (T790M) gatekeeper mutation associated with resistance to first-generation inhibitors, have shown clinical activity in patients with T790M-positive lung tumors. However, the potential contribution of intratumor heterogeneity to the emergence of resistance to T790M-specific inhibitors remains unclear. Piotrowska and colleagues prospectively biopsied EGFR-mutant, T790M-positive lung tumors from patients treated with rociletinib as part of a phase I/II clinical trial, and compared paired pretreatment and posttreatment progressing samples to identify mechanisms of resistance. Seven of 13 rociletinib-resistant biopsies retained T790M mutations, with three of these tumors exhibiting acquired amplification of EGFR. However, the remaining six resistantbiopsies, two of which were characterized by transformation to small cell lung cancer, were T790–wild-type upon progression. Single-cell clone analysis of one pre-rociletinib biopsy revealed the coexistence of T790M-positive and T790–wild-type cells, indicating that the apparent loss of T790M observed in this cohort likely results from molecular heterogeneity in the original tumor and selection for preexisting T790–wild-type subclones that drive resistance. Furthermore, the degree of baseline T790M heterogeneity in pretreatment biopsies was predictive of response to rociletinib. Longitudinal analysis of plasma samples was consistent with biopsy results and supported the notion that rociletinib resistance can develop from the outgrowth of either T790M-positive or T790–wild-type cells. These results highlight intratumor heterogeneity as a driver of resistance and suggest that combined targeting of both subpopulations may improve clinical outcomes.

See article, p. 723.

  • Inactivating mutations in CDKN2B were detected in a subset of patients with familial RCC.

  • Missense mutations are predicted to destabilize CDKN2B or decrease its affinity for CDK4/6.

  • CDKN2B missense mutations impair the growth-suppressive activity of CDKN2B in RCC cell lines.

Germline mutations in various genes, including VHL, FLCN, SDHB, MET, and BAP1, are associated with inherited renal cell carcinoma (RCC), which comprises approximately 3% of RCC cases. However, germline mutations in known RCC-associated genes have not been detected in the majority of patients with familial RCC, prompting Jafri and colleagues to perform exome sequencing and targeted resequencing of patients with inherited RCC to identify additional potential pathogenic mutations. This analysis revealed a candidate nonsense mutation in the tumor suppressor gene CDKN2B, which encodes the cyclin-dependent kinase (CDK) inhibitor CDKN2B/p15INK, in individuals affected by RCC in a familial RCC kindred. Screening of an additional 84 individuals with features of familial RCC identified three missense mutations at evolutionarily conserved residues in CDKN2B, which were predicted to be pathogenic and included two previously unreported mutations. In silico structural analysis indicated that two of the mutations were predicted to localize to the binding interface with CDK4/6 and decrease the affinity of CDKN2B for CDKs, whereas the third mutation was predicted to destabilize the CDKN2B protein. Furthermore, each of the three CDKN2B missense mutations impaired the growth-suppressive activity of CDKN2B in RCC cell lines. These results implicate germline mutations of CDKN2B as a cause of familial RCC and support the investigation of genetic testing for germline CDKN2B mutations in individuals at risk for familial RCC.

See article, p. 730.

  • INPP4B directly dephosphorylates PI(3,4,5)P3 and prevents its accumulation in PTEN-deficient cells.

  • INPP4B and PTEN deletion induces PI3K–AKT2-driven aggressive follicular thyroid tumors in mice.

  • Human thyroid tumors frequently display concomitant loss of INPP4B and PTEN expression.

Thyroid cancer is a common endocrine disease that can be subdivided into nonaggressive papillary thyroid carcinoma and metastatic follicular thyroid carcinoma (FTC). Although the mechanisms driving thyroid tumorigenesis remain unclear, patients with germline mutations in the phosphatidylinositol (3,4,5) trisphosphate [PI(3,4,5)P3] phosphatase PTEN have a high propensity to form benign thyroid abnormalities, suggesting that aberrant PI3K–AKT signaling contributes to thyroid cancer formation. The inositol polyphosphate-4-phosphatase, type II (INPP4B) phosphatase has been implicated as a potential tumor suppressor that is frequently downregulated in cancers with PTEN mutations, prompting Kofuji and colleagues to assess whether inactivation of PTEN and INPP4B cooperate in tumorigenesis. INPP4B dephosphorylated PI(3,4,5)P3in vitro, suggesting that INPP4B may be required to control PI(3,4,5)P3 levels in PTEN-deficient cells. In support of this idea, codeletion of Inpp4b in Pten+/− mice induced the formation of aggressive follicular thyroid tumors and pulmonary metastases, which were not observed upon inactivation of either gene alone. Downregulation of INPP4B was detected in human FTC samples and frequently co-occurred with PTEN downregulation. Mechanistically, loss of INPP4B in human and mouse thyroid tumors with low PTEN expression led to enhanced PI(3,4,5)P3accumulation and hyperactivation of AKT signaling, and deletion of Akt2, but not Akt1, in Inpp4b−/−; Pten+/− mice led to increased survival. Together, these data indicate that, in the absence of PTEN, INPP4B prevents the accumulation of PI(3,4,5)P3, and that concomitant inactivation of these tumor suppressors promotes PI3K–AKT-driven thyroid tumorigenesis.

See article, p. 740.

  • Loss of Pten and Inpp4b cooperate to promote metastatic follicular thyroid carcinoma (FTC) in mice.

  • INPP4B is downregulated in human FTC cells and correlates with enhanced AKT activation.

  • INPP4B inhibits PI3K-mediated activation of AKT2 in early endosomes of thyroid cancer cells.

Loss of the PTEN tumor suppressor gene promotes cell transformation in vivo via accumulation of phosphatidylinositol (3,4,5) trisphosphate [PI(3,4,5)P3] and hyperactivation of the PI3K–AKT signaling pathway. Inositol polyphosphate-4-phosphatase, type II (INPP4B) has been shown to possess similar antagonistic function toward PI3K–AKT and is downregulated in a variety of primary and metastatic cancers, prompting Chew and colleagues to examine the role of INPP4B in PI3K–AKT-driven tumorigenesis in vivo. In contrast to Pten+/− mice, heterozygous or homozygous deletion of Inpp4b did not lead to overt tumor formation, suggesting that PTEN and INPP4B likely have different functions in cancer. However, codeletion of Pten and Inpp4b (Pten+/−;Inpp4b−/−) cooperated to promote the formation of thyroid tumors that closely resembled human follicular thyroid carcinoma (FTC) or follicular variant papillary thyroid carcinoma (FV-PTC), and induced pulmonary metastases in 50% of Pten+/−;Inpp4b−/− mice. Expression of INPP4B was downregulated in human FTC cell lines and tumor samples compared with normal thyroid tissues and correlated with hyperactivation of AKT signaling. Mechanistically, suppression of INPP4B promoted the accumulation of PI(3,4)P2 in early endosomes of thyroid cancer cells, which led to class II PI3Kα isoform–mediated activation of AKT2, but not AKT1. Functionally, depletion of INPP4B in thyroid cancer cells led to a growth advantage under specific serum conditions and increased anchorage-independent growth in vitro. Together, these results point to a role for INPP4B as a tumor suppressor in primary and metastatic FTC via endosome-specific regulation of PI3K–AKT signaling.

See article, p. 752.

  • ARID1A physically interacts with ATR and is recruited to DSBs in an ATR-dependent manner.

  • ARID1A facilitates DSB end resection and activation of ATR-mediated checkpoint signaling.

  • Inactivation of ARID1A sensitizes cancer cells to PARP inhibitors.

Recent genome sequencing studies have identified inactivating mutations in ARID1A, which encodes a subunit of SWI/SNF chromatin remodeling complexes, as one of the most frequent genetic events in human cancers, but the cellular consequences of ARID1A loss are unclear. Shen and colleagues made the unexpected observation that ARID1A binds to ATR, a kinase that plays a central role in activating the DNA damage response, and found that ATR recruits ARID1A to DNA double-strand breaks (DSB). In turn, ARID1A was required for activation of ATR in response to DSBs and for the initiation and maintenance of the G2/M checkpoint. ARID1A loss reduced the efficiency of DSB end resection to single-stranded DNA, which is necessary for activation of the ATR-mediated checkpoint response and for homologous recombination and single-strand annealing repair mechanisms, and led to an increase in histone deposition at DSBs, suggesting that ARID1A is needed to create a permissive chromatin environment for DSB end resection and DNA damage checkpoint signaling. Similar to BRCA1 and BRCA2, homologous recombination proteins whose loss confers sensitivity to DSBs induced by PARP inhibition, knockdown of ARID1A sensitized cancer cells to PARP inhibitors in vitro and in vivo. PARP inhibitor sensitivity could not be rescued by ARID1A mutants lacking the ATR interaction domain or by some patient-derived ARID1A mutants, indicating that a consequence of ARID1A inactivation may be impairment of the ATR-mediated DNA damage response and providing a rationale for evaluation of PARP inhibitors in patients with ARID1A-mutant cancers.

See article, p. 768.

  • The dual PI3K/mTOR inhibitor BEZ235 increases MYC levels in colorectal cancer cells.

  • FOXO–MAPK signaling and eIF4A-driven translation enhance MYC expression in BEZ235-treated cells.

  • Silvestrol inhibits eIF4A, suppressing MYC expression and colorectal cancer cell proliferation.

Enhanced expression of the MYC transcription factor is ubiquitous in colorectal cancers, suggesting that strategies to increase MYC protein turnover may be clinically beneficial. To test this possibility in colorectal cancer cells, Wiegering, Uthe, and colleagues utilized BEZ235 to target both PI3K and mTOR signaling, which are known to control MYC translation and turnover in other tumor types. However, BEZ235 treatment did not stimulate MYC degradation, but rather increased MYC mRNA and protein expression in colorectal cancer cells. Enhanced MYC expression in BEZ235-treated cells resulted in part from forkhead box protein 3A (FOXO3A)–dependent upregulation of growth factor receptor genes, resulting in MAPK activation and increased MYC transcription. In addition, MYC translation initiation was maintained in BEZ235-treated cells due to insufficient sequestration of eukaryotic initiation factor 4E (eIF4E) and eIF4A-dependent, but eIF4E-independent, translation of MYC via an internal ribosome entry site. Treatment with a small-molecule inhibitor of eIF4A, silvestrol, reduced both basal and BEZ235-induced MYC protein levels and suppressed the proliferation of colorectal cancer cells. Consistent with these data, silvestrol inhibited MYC protein expression and hyperproliferation in a murine colorectal cancer model, whereas BEZ235 treatment had no effect. Taken together, these results demonstrate that, whereas dual PI3K/mTOR inhibition paradoxically enhances MYC expression in colorectal cancer cells due to feedback mechanisms, blockade of MYC translation initiation via eIF4A inhibition suppresses colorectal cancer cell growth in vitro and in vivo and may be a therapeutically tractable strategy.

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