See article, p. 492.

  • The BET inhibitor OTX015/MK-8628 was administered to four patients with NMC with BRD4–NUT fusions.

  • Two patients rapidly responded with tumor regression and symptomatic relief.

  • Treatment interruption reversed side effects but resulted in NMC progression.

Two thirds of patients with NUT midline carcinomas (NMC), which have a median overall survival of 6.7 months, harbor an in-frame fusion of NUT with the bromodomain and extraterminal (BET) family member bromodomain containing 4 (BRD4), which epigenetically drives cell transformation via aberrant expression of pro-growth genes. Stathis and colleagues investigated the potential antitumor activity of the BET inhibitor OTX015/MK-8628 compassionately administered to four previously treated patients with advanced NMC. All patients harbored the BRD4–NUT fusion as confirmed by FISH, and molecular profiling of NMC tissue from three patients showed that minimal alternate genetic aberrations were present. Two patients exhibited a rapid response to once-daily OTX015/MK-8628, including tumor regression as determined by physical exam and PET-CT and alleviation of disease-related symptoms, and had an overall survival of 19 and 7 months post-diagnosis. A third patient had disease stabilization with a minor metabolic response; following disease progression the patient was treated with paclitaxel and achieved an overall survival of 18 months post-diagnosis. OTX015/MK-8628 treatment side effects included but were not limited to nausea, fatigue, hyperglycemia, and thrombocytopenia, which were reversible after dose lowering and brief treatment cessation, the latter of which resulted in rapid disease progression. These results highlight the potential for BET inhibitor use in NMC and potentially other solid malignancies alone or in combination with other therapies.

See article, p. 501.

  • FLT3-mutant/DNMT3A-haploinsufficient mice recapitulate cytogenetically normal AML.

  • DNMT3A reduction transforms FLT3ITD cells via hypomethylation of CpG shores and enhancers.

  • The effects of DNMT3A haploinsufficiency on methylation and clonogenicity are reversible.

FLT3 internal tandem duplication (ITD) mutations are common in patients with cytogenetically normal acute myeloid leukemia (CN-AML), and are associated with a poor survival and increased risk of relapse. Mutations in the DNA methyltransferase DNMT3A are also associated with a poor prognosis, which is even poorer in patients harboring both FLT3 and DNMT3A mutations. However, Flt3ITD mice develop myeloproliferative neoplasia that does not progress to AML, and the mechanism by which DNMT3A mutations may promote AML remains unknown. To determine if FLT3 and DNMT3A mutations cooperate to induce AML, Meyer and colleagues generated Flt3ITD knock-in mice with inducible deletion of Dnmt3a. This resulted in the rapid development of CN-AML and death; most AML cells deleted only one Dnmt3a allele, suggesting that Dnmt3a haploinsufficiency contributes to oncogenesis. Patients with FLT3ITD/DNMT3A-mutant AML had hypomethylated genomes, with enhanced methylation at CpG shores and enhancer region. Similar hypomethylation was observed in Flt3ITD/Dnmt3a haploinsufficient mice. RNA sequencing indicated that the majority of genes overexpressed in DNMT3A-mutant AML corresponded with DNA hypomethylated sites. Forced expression of DNMT3A in haploinsufficient cells resulted in reversion of the hypomethylation, and reduced clonogenicity, indicating that DNA hypomethylation resulting from DNMT3A loss is reversible. Further, single-cell RNA sequencing and complementary single-cell assays mapped the biological architecture of AML. This study generated a mouse model of CN-AML, and indicates that DNMT3A haploinsufficiency is sufficient to transform FLT3ITD myeloid progenitor cells via reversible hypomethylation and that DNMT3A may be therapeutically targeted in AML.

See article, p. 516.

  • MENA promotes haptotaxis, a type of cell migration guided by surface-bound molecule gradients.

  • The MENAINV isoform promotes tumor-cell mediated remodeling of the extracellular matrix.

  • High MENAINV correlates with tumor recurrence and poor outcome in human breast cancer cohorts.

Chemotaxis, the migration of cells along soluble molecule gradients, can promote metastasis, but haptotaxis, cell migration guided by gradients of surface-bound molecules such as extracellular matrix (ECM) proteins, is less well understood. In breast tumors, fibronectin (FN) is a major ECM component, and high levels are associated with disease progression. MENAINV is a splice variant of the actin regulator MENA that binds to integrin α5β1 more robustly than MENA and regulates integrin signaling, and is associated with breast cancer metastatic progression. Oudin and colleagues hypothesized that the ECM may promote MENA/MENAINV-driven metastasis. Full-length MENA induced robust breast cancer cell haptotaxis along low-concentration gradients of FN. MENA-driven haptotaxis was abolished in response to high FN concentrations, inhibition of the FN-binding integrin α5β1, or deletion of either the MENA integrin or F-actin binding domains. In comparison to MENA, MENAINV had increased association with integrin α5β1, and MENAINV expression induced haptotaxis toward high FN concentrations representative of levels found near blood vessels and in metastatic tumor sites. MENAINV-driven haptotaxis required integrin outside-in signaling, tumor cell–mediated collagen reorganization, and induction of FN fibrillogenesis, which collectively resulted in highly invasive tumors. In patients with breast cancer, MENA and MENAINV expression correlated with FN and α5β1 expression, but only MENAINV correlated with decreased time to disease recurrence and poor outcome. These results reveal a tumor cell–intrinsic mechanism by which MENAINV drives reorganization of the tumor ECM and haptotaxis along FN gradients to promote breast cancer metastasis.

See article, p. 532.

  • Glioblastoma TICs exhibit chromosomal instability, which generates karyotypic diversity.

  • Experimental elevation of chromosomal instability reduces TIC self-renewal and growth in vitro.

  • Increasing mitotic errors in TICs abolishes tumorigenesis in murine models.

Glioblastoma tumor-initiating cells (TIC) may be responsible for tumorigenesis and are a source of functional intratumor heterogeneity, whereas chromosomal instability (CIN) is a source of genetic heterogeneity. Both TICs and CIN may contribute to therapeutic resistance; however, the relationship between TICs and CIN has not been established. Godek and colleagues determined that TIC lines isolated from multiple glioblastomas, known as glioma neural stem (GNS) cell lines, had an increased frequency of lagging chromosomes at anaphase compared to stable diploid cells, which is characteristic of CIN. Furthermore, FISH analysis revealed GNS cells to be aneuploid for most chromosomes, indicating the ability of GNS cells to propagate despite abnormal karyotypes. Further, there was extensive karyotypic diversity in individual cell clones as well as within populations. Exogenous expression of dominant-negative mitotic centromere-associated kinesin (MCAK) in GNS cell lines increased chromosomal instability, likely via the stabilization of aberrant chromosome microtubule attachments. TICs with elevated chromosomal instability due to dominant-negative MCAK expression had reduced self-renewal, differentiation, and proliferative capabilities in vitro, and were unable to form tumors upon intracranial injection in vivo. Taken together, these findings establish a critical link between TICs and CIN, which combine to drive malignancy through functional and genetic heterogeneity; however, TICs have a tolerable upper CIN limit which highlights a potential therapeutic vulnerability.

See article, p. 546.

  • Human HCCs harbor a subset of PD-1hi B cells that correlate with disease grade and recurrence.

  • TLR4-driven upregulation of BCL6 induces PD-1 in HCC-derived B cells.

  • PD-1hi B-cell interaction with PD-L1 promotes IL10-driven T-cell dysfunction and tumor progression.

Upregulation of programmed death-1 (PD-1) receptor on T cells leads to T-cell exhaustion and protumorigenic immune evasion and has been therapeutically targeted by blocking the interaction between PD-1 and its ligands PD-L1 and PD-L2. PD-1 can also be upregulated in other tumor-infiltrating immune cell types, suggesting that non–T cell–derived inhibitory signals may contribute to tumor immune evasion. Xiao, Lao, Chen, and colleagues showed that a subset of B cells in human hepatocellular carcinomas (HCC) displayed high levels of PD-1 (PD-1hi), and infiltration of PD-1hi B cells correlated with disease progression and early recurrence. Characterization of HCC-derived PD-1hi B cells revealed a unique CD5hiCD24+/−CD27hi/+CD38dim phenotype that was distinct from other known B-cell populations with immunosuppressive functions. The HCC microenvironment induced activation of toll-like receptor 4 (TLR4) on B cells, which in turn induced PD-1 expression in a BCL6-dependent manner. Stimulation of HCC-derived PD-1hi B cells with a PD-1 agonist induced IL10 production ex vivo, and co-culture of tumor-derived PD-L1+ monocytes with PD-1hi B cells led to IL10 production in a PD-L1–dependent manner. Importantly, transfer of PD-1hi B cells into recipient mice promoted tumor growth and impaired T-cell function, which was reversed upon treatment with antibodies targeting IL10 or PD-L1. The identification of a T cell–independent mechanism by which PD-1 mediates immunosuppression provides additional insight into the function of tumor-infiltrating B cells and raises the possibility that immune checkpoint inhibitors may exert effects on more immune cell types than previously appreciated.

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