In This Issue

See article by Sun et al., p. 122.

Regulatory T cells (Treg) suppress immune responses and thus represent a major obstacle to cancer immunotherapies. The current gold standard for Treg identification is staining for FOXP3, an intracellular marker that requires cell permeabilization and therefore precludes isolation of viable Tregs and subsequent functional studies. Another limitation of FOXP3 as a Treg marker is that FOXP3 expression also increases during immunotherapy-induced T-cell activation in humans, hindering efforts to distinguish Treg and effector T cell (Teff) populations and accurately gauge the effects of therapy in tumors and peripheral blood. Low expression of the cell surface marker CD127 was previously reported to be an appropriate surrogate for identifying Tregs, but Sun and colleagues report that expression of an extracellular protein, latency-associated peptide (LAP), correlates more closely with Tregs than FOXP3-positivity or low expression of CD127. Compared with CD127^low T cells, LAP expression more accurately identified Tregs among in vitro activated human peripheral blood mononuclear cells (PBMC) or PBMCs derived from patients treated with ipilimumab, the anti–CTLA-4 immunotherapeutic recently approved as a standard-of-care treatment for patients with metastatic melanoma. Furthermore, LAP staining consistently correlated with a functional Treg population, unlike the CD127^low populations, which did not consistently represent immunosuppressive cells. These findings will aid efforts to monitor the various T-cell populations in patients undergoing treatment with ipilimumab or other immunotherapy agents and to understand the immune responses elicited by these therapies.

See article by Shi et al., p. 131.

Genome-wide association studies (GWAS) have had limited success in identifying lung cancer susceptibility loci, potentially because heterogeneity among histologic subtypes hinders the detection of associations between polymorphic variants and overall lung cancer risk. However, the relative rarity of some disease subtypes can make it difficult to accrue sample sizes necessary to identify histology-specific single-nucleotide polymorphisms (SNP) if the whole genome is analyzed. To address these limitations, Shi and colleagues analyzed the association between SNPs mapping to genes involved in inflammation, which is induced by smoking, and risk of small cell lung cancer, lung adenocarcinoma, or squamous cell lung carcinoma in a previous GWAS of smoking-related lung cancers and smoking controls. Using this method, the authors identified a strong association between variation in the inflammation pathway and squamous cell carcinoma, and the top 55 genes were genotyped in squamous cell carcinoma cases from 2 independent studies. Only one SNP, which mapped to the RAD52 gene region on chromosome 12p13.33, was replicated in both studies and in a third sample set subsequently analyzed, and was statistically significant on a genome-wide basis when the data from all 4 studies were combined. These findings demonstrate that subtype-specific analyses can improve the detection of association signals in GWAS studies and that their combination with pathway-based approaches can facilitate the identification of risk loci.

See article by Bellail et al., p. 140.

Agonists of the TNF-related apoptosis-inducing ligand (TRAIL) pathway, which activates caspase-8–dependent extrinsic apoptosis, have been touted as potential cancer therapeutics, but clinical trials have indicated that tumors are frequently resistant to TRAIL pathway–targeted therapies. To better understand mechanisms of TRAIL resistance, Bellail and colleagues analyzed the expression of several ubiquitin ligases that are known upstream regulators of caspase-8 activation and found that A20 (also known as TNF-α–induced protein 3) was specifically overexpressed in glioblastomas and TRAIL-resistant glioblastoma cell lines compared with normal brain tissue and TRAIL-sensitive cell lines. Furthermore, A20 specifically forms a complex with DR5, RIP1, and TRAF2 even in the absence of TRAIL stimulation in the tumor tissues and TRAIL-resistant cell lines, which the authors named the preligand assembly complex (PLAC). As part of the PLAC, A20 polyubiquitinates RIP1, resulting in recruitment of caspase-8, formation of the death-inducing signaling complex (DISC), and inhibition of caspase-8 dimerization and cleavage in the DISC. The clinical relevance of A20-dependent inhibition of TRAIL-induced apoptosis was supported by the presence of both the PLAC and DISC in patient-derived glioblastoma tissues and glioblastoma-initiating cells and the ability of A20 knockdown to increase glioblastoma-initiating cell apoptosis in the presence of TRAIL. These findings establish a cellular resistance mechanism that may potentially explain why some tumors do not respond to TRAIL pathway induction and suggest that targeting of A20-dependent RIP1 ubiquitination may potentiate the effect of TRAIL-targeted therapies.

See article by McKie et al., p. 156.

Loss of opioid binding protein cell adhesion molecule (OPCML) expression has been observed in multiple cancer types and its forced expression can inhibit the growth and tumorigenicity of ovarian cancer cell lines, but the mechanism of OPCML-mediated tumor suppression remains unclear. McKie and colleagues establish the clinical relevance of OPCML inactivation with the observations that OPCML expression is significantly reduced in the vast majority of high-grade serous ovarian cancers and that low OPCML expression is a significantly poor prognostic factor for ovarian and breast cancer. Because OPCML is an extracellular, lipid raft-associated membrane protein that cannot directly affect intracellular signaling, the authors hypothesized that OPCML interacts with receptor tyrosine kinases (RTK) to negatively regulate mitogenic signaling. In epithelial ovarian cancer cell lines, OPCML negatively regulates a specific subset of RTKs including EPHA2, FGFR1, FGFR3, HER2, and HER4 and phosphorylation of downstream substrates. Mechanistically, binding of OPCML to RTK extracellular domains redirects the RTKs to a lipid raft-dependent bulk endocytic pathway that ultimately leads to their polyubiquitination and proteasomal degradation. Exogenous recombinant OPCML downregulated the same spectrum of RTKs in ovarian cancer cells and was associated with growth inhibition in 6 of 7 ovarian cancer cell lines and suppression of tumor formation following intraperitoneal injection into mice bearing ovarian tumor xenografts. The unique mechanism of action of OPCML-mediated tumor suppression suggests that extracellular protein therapy may be a viable strategy for the treatment of OPCML-deficient cancers.

See article by Marcotte et al., p. 172.

Functional genomic approaches have identified unanticipated synthetic lethal interactions in a limited number of cancer cell lines and have provided useful information on the vulnerabilities of cancer cells, but a comprehensive analysis of essential genes in major tumor types has been lacking. To address this, Marcotte and colleagues performed near genome-wide pooled short hairpin RNA (shRNA) screens on 72 breast, pancreatic, and ovarian cancer cell lines to identify shRNAs that are selectively depleted from cell populations and thus target essential genes. The authors quantified hairpin dropout at multiple timepoints, in contrast to current methods for scoring synthetic lethal screens, which analyze “dropouts” at only a single timepoint. This scoring metric classified shRNAs as fast, continuous, or slow dropouts and assigned scores to individual genes based on the behavior of their corresponding shRNAs. Fast and continuous dropout shRNAs targeted highly conserved genes involved in housekeeping functions and were more likely to be generally essential in multiple cell lines. The screens also identified tissue-specific essential genes that only affected proliferation of cancer cells derived from one tissue type, and subtype-specific essential genes that could robustly distinguish basal and luminal breast cancer cell lines. The functional genomic data were also integrated with publicly available copy number information for breast, ovarian, and pancreatic tumors to determine whether the essential genes reside within recurrently amplified regions and identify potential tissue-specific oncogenic driver genes.

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