In This Issue

See article, p. 1511

Genome-wide association studies have identified a locus in GRM4 (encoding glutamate metabotropic receptor 4) linked to an increased risk of osteosarcoma, but the reason for this association is unknown. In a mouse model of osteosarcoma, Kansara and colleagues found that Grm4-knockout mice had accelerated tumor development. Dendritic cells (DC) from the bone marrow of the Grm4-knockout mice displayed increased expression of IL23, an inflammatory cytokine found to exhibit increased expression in many human cancers. Further experiments revealed that exposure to conditioned media from cultured mouse osteosarcoma cells led to increased Il23 expression by DCs derived from bone marrow, and Il23-knockout mice had markedly reduced osteosarcoma development. Demonstrating the potential clinical relevance of these findings, in situ hybridization experiments using human osteosarcomas indicated that elevated IL23 expression in the tumors was common, and high IL23 expression was correlated with decreased survival. Treating mice implanted with osteosarcoma cells with an antibody against IL23 led to a moderate reduction in tumor growth and increase in the duration of survival, and the treatment showed a synergistic effect with doxorubicin; notably, IL23 antagonists are already in use for other indications. Further, treatment with a positive allosteric modulator of GRM4 significantly reduced tumor growth, matching the efficacy of doxorubicin without causing the weight loss associated with doxorubicin treatment. Collectively, these results provide mechanistic insight into the role of GRM4 and IL23 in osteosarcoma and imply that these proteins may be promising therapeutic targets.

See article, p. 1520

In CAR-T and other T-cell therapies, T-cell transfer into lymphodepleted patients results in homeostatic activation of the T cells, which promotes their antitumor response. In mice and humans, Marshall and colleagues found that transfer of CD8⁺ T cells into lymphodepleted recipients triggered increased T-cell expression of immune-checkpoint proteins, including CTLA4 and PD-1, a phenomenon they termed homeostatic inhibition. In patients receiving autologous bone-marrow transplants (BMT) and in a mouse BMT model, homeostatic activation and homeostatic inhibition were coupled. In the mouse model, dual checkpoint blockade with PD-1 and CTLA4 inhibitors before and after BMT—which the authors termed immunotransplant—prevented checkpoint-induced suppression of CD8⁺ T cells. Immunotransplant triggered tumor-specific immune responses in a mouse model of B-cell lymphoma, eliminating T-cell exhaustion and causing tumor regression, whereas BMT or dual checkpoint blockade alone did not have the same efficacy. Blockade of both PD-1 and CLTA4 and recipient lymphodepletion were required for the antitumor efficacy of immunotransplant, and efficacy was reduced by depletion of CD8⁺ T cells and IFNγ, but not by depletion of CD4⁺ T cells. Signaling by the cytokine γ_c was also required for antitumor efficacy of immunotransplant, and this signaling was increased by the procedure. Indicating the possible clinical relevance of this finding, γ_c-receptor expression and γ_c signaling were increased in patients with lymphoma following BMT. Additionally, homeostatic activation and inhibition caused by BMT and γ_c were dependent on JAK3. Together, these results provide a rationale for further testing of immunotransplant along with mechanistic data to guide such studies.

See article, p. 1538

Metastatic castration-resistant prostate cancer (CRPC) is frequently characterized by maintenance of androgen receptor (AR) signaling, but treatment with currently available drugs that target AR signaling results in resistance. In AR-amplified prostate cancer cells stimulated with the androgen dihydrotestosterone (DHT), Rasool, Natesan, Deng, and colleagues found that AR and the transcriptional coactivator MED1 (a component of the transcription-activating Mediator complex) colocalized on chromatin, particularly at superenhancer loci. The interaction between MED1 and AR was dependent on MED1 phosphorylation at T1457 by the cyclin-dependent kinase CDK7, and inhibition of CDK7 by the specific CDK7/12 inhibitor THZ1 abolished DHT-induced recruitment of MED1 to chromatin. Further, AR-dependent prostate-cancer cell lines—but not other prostate-cancer cell lines—exhibited sensitivity to THZ1 treatment, as did prostate-cancer cell lines made resistant to the second-generation antiandrogen enzalutamide. Notably, enzalutamide-resistant prostate cancer cells had elevated levels of phosphorylated MED1. In a mouse xenograft model of relapsed CRPC, THZ1 treatment led to substantial tumor regression and appeared to be tolerated well. These results support the notion that CDK7 inhibitors—which have already been demonstrated to have preclinical efficacy in several other cancer types and are in clinical development—may be worth investigating for CRPC, alone or in combination with second-generation antiandrogens to overcome AR-dependent resistance mechanisms.

See article, p. 1556

Underlying the deadliness of glioblastoma are glioblastoma stem cells (GSC), a heterogeneous group of poorly differentiated cells that drive recurrence. Dong, Zhang, Qu, and colleagues discovered that knockdown of the core circadian-clock genes BMAL1 or CLOCK inhibited the proliferation of patient-derived GSCs but not nonmalignant brain cells. This reduction in proliferation appeared to be a result of cell-cycle arrest and an increase in apoptosis. BMAL1 or CLOCK knockdown also impaired sphere formation and resulted in reduced expression of GSC-maintenance transcription-factor genes, including SOX2, OLIG2, and MYC, indicating a reduction in GSC stemness. Compared to neural stem cells, GSCs exhibited increased BMAL1 chromatin occupancy at regions surrounding the transcriptional start sites, coinciding with an increase in histone marks that indicate active chromatin. Further investigation revealed that BMAL1 and CLOCK regulated expression of genes involved in glycolysis and the tricarboxylic acid cycle in GSCs and promoted mitochondrial oxidative phosphorylation. Treating GSCs with negative regulators of BMAL1 transcription (small-molecule agonists of REV-ERBα/β) reduced proliferation, as did treatment with a carbazole derivative (KL001) that stabilizes the BMAL1–CLOCK repressors CRY1 and CRY2; combination treatment increased this effect. In mice with tumors grown from intracranially implanted GSCs, BMAL1 or CLOCK knockdown nearly doubled life span and resulted in reduced tumor size. Data from The Cancer Genome Atlas revealed that high BMAL1 mRNA levels were correlated with higher-grade gliomas and glioblastoma and were associated with poorer prognosis. Collectively, these findings suggest that targeting BMAL1–CLOCK may represent a promising strategy in glioblastoma.

See article, p. 1574

Recurrence and poor outcome in glioblastoma are thought to result from glioblastoma stem cell (GSC) populations. Jimenez-Pascual, Hale, and colleagues discovered that the disintegrin metalloproteinase ADAMDEC1 was among the most upregulated proteases in glioblastoma, and ADAMDEC1 was the only protease for which increased expression was correlated with worse prognosis. In patient-derived xenografts, ADAMDEC1 was confirmed to be expressed by tumor cells rather than host microglia, and experiments using patient-derived cell lines revealed that GSCs secreted ADAMDEC1. Additionally, ADAMDEC1 was associated with stemness in this context. Treatment with recombinant ADAMDEC1 resulted in release of FGF2 from the extracellular matrix in GSC cultures, but not non-stem tumor cell cultures, and treatment of glioblastoma cells with recombinant FGF2 increased expression of GSC-associated transcription factors in a dose-dependent manner. Further experiments indicated that FGF2 signaling through FGFR1, which is located on the surface of GSCs, triggered ERK1/2 signaling; this led to increased expression of ADAMDEC1 and the stemness-associated transcription factors ZEB1, SOX2, and OLIG2, creating a positive feedback loop that resulted in maintenance of GSC stemness. Reinforcing the loop, ZEB1 also regulated expression of ADAMDEC1 and the effects of FGF2 and FGFR1 on GSC stemness. These findings may have clinical relevance, as trials of FGFR inhibitors in glioma are under way, but only patients with mutations in FGFR genes—representing only 3% to 5% of patients with glioblastoma—are enrolled. These data suggest that other patients may also benefit from FGFR inhibitors due to targeting of treatment-resistant GSCs.

See article, p. 1590

Inactivation of the kinase LKB1 is implicated in lung cancer, but the pathway through which it suppresses cancer growth is not known. Using an in vivo CRISPR screen in a mouse lung cancer model, Murray and colleagues identified the downstream kinases SIK1 and SIK3 as major tumor-suppressive LKB1 substrates. Tumors lacking SIK1 or SIK3 had increased growth, and this effect was seen even in the context of p53 loss. Simultaneous targeting of Sik1 and Sik3 revealed a synergistic effect, yielding tumors comparable in size to those produced by targeting Lkb1. Immunohistochemistry showed an increase in proliferation but no decrease in cell death in the tumors and revealed that Sik1- and Sik3-targeted tumors histologically resembled Lkb1-deficient tumors. There was also substantial overlap (greater than 70%) in the genes dysregulated because of the loss of either Sik1/Sik3 or Lkb1, and Sik1/Sik3 targeting induced a transcriptional state similar to that of human LKB1-deficient lung adenocarcinoma. Collectively, these results identify a previously unknown role for SIK1 and SIK3 in lung cancer pathogenesis, prompting a need to further investigate this pathway to potentially uncover therapeutic vulnerabilities for LKB1-deficient tumors.

See article, p. 1606

Many non–small cell lung cancers (NSCLC) are characterized by inactivating mutations in STK11, which encodes the serine/threonine kinase LKB1. Fourteen AMPK-related kinases lie downstream of LKB1, and it remains unclear which are responsible for the tumor-suppressor function of LKB1. Using a targeted CRISPR screen in human NSCLC cells, Hollstein and colleagues observed that kinases in the SIK subfamily, particularly SIK1, may be the relevant LKB1 targets in NSCLC. Further investigation using a mouse model of NSCLC revealed that disruption of SIK1 resulted in increased tumor burden, and cell lines derived from the tumors exhibited greater proliferation compared to controls. Inactivation of SIK1 alone did not fully recapitulate the tumorigenic effects of loss of LKB1; however, simultaneous inactivation of SIK1 and SIK3 more substantially accelerated tumor growth in mice, resulting in tumors similar in size to those seen with LKB1 knockout. Differential expression analysis of lung tumors from the mice showed that combined loss of SIK1 and SIK3 may account for 40% of the transcriptional alterations seen with LKB1 knockout. Notably, combined loss of SIK1 and SIK3, like loss of LKB1, resulted in altered expression of genes related to epithelial–mesenchymal transition and IL6–JAK–STAT3 signaling, a finding that was corroborated in human NSCLC cells. Additional experiments implied that the transcriptional regulator CRTC2 may mediate these transcriptional effects. Together, these results elucidate details of the pathway through which LKB1 loss promotes NSCLC, implicating the little-studied kinases SIK1 and SIK3 in the process.

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