Genome-wide shRNA screen revealed integrated mitogenic signaling between dopamine receptor D2 (DRD2) and epidermal growth factor receptor (EGFR) in glioblastoma

Abstract

Glioblastoma remains one of the deadliest of human cancers, with most patients succumbing to the disease within two years of diagnosis. The available data suggest that simultaneous inactivation of critical nodes within the glioblastoma molecular circuitry will be required for meaningful clinical efficacy. We conducted parallel genome-wide shRNA screens to identify such nodes and uncovered a number of G-Protein Coupled Receptor (GPCR) neurotransmitter pathways, including the Dopamine Receptor D2 (DRD2) signaling pathway. Supporting the importance of DRD2 in glioblastoma, DRD2 mRNA and protein expression were elevated in clinical glioblastoma specimens relative to matched non-neoplastic cerebrum. Treatment with independent si-/shRNAs against DRD2 or with DRD2 antagonists suppressed the growth of patient-derived glioblastoma lines both in vitro and in vivo. Importantly, glioblastoma lines derived from independent genetically engineered mouse models (GEMMs) were more sensitive to haloperidol, an FDA approved DRD2 antagonist, than the premalignant astrocyte lines by approximately an order of magnitude. The pro-proliferative effect of DRD2 was, in part, mediated through a GNAI2/Rap1/Ras/ERK signaling axis. Combined inhibition of DRD2 and Epidermal Growth Factor Receptor (EGFR) led to synergistic tumoricidal activity as well as ERK suppression in independent in vivo and in vitro glioblastoma models. Our results suggest combined EGFR and DRD2 inhibition as a promising strategy for glioblastoma treatment.

Figure 1. DRD2 is required for glioblastoma growth

(A) Schema of genome-wide shRNA screen. Cells were serially passaged after infection with virus containing shRNA constructs. The relative abundance of each shRNA prior to and after passages was subsequently determined (see Methods). (B) Cy5 signal for each gene from three independent U87MG screens (denoted as A, B or C) was plotted against each other (on x-, y-, and z- axes, respectively). High degrees of correlation were seen (R²=0.89-0.92). Similar degrees of correlations were seen for Cy3 signals and for other cell lines. (C) PANTHER pathway analysis of pro-proliferative genes for U87MG and A172. * denotes p < 0.05. (D) PANTHER curated DRD2 pathway genes. * denotes pro-proliferative genes identified in our screen. (E) DRD2 silencing by independent sh-/siRNAs compromised clonogenic U87MG growth in vitro. Efficiency of DRD2 silencing was determined by RT-PCR (bottom panel). Cells were seeded at ~50% confluency, transfected or infected with siRNA or shRNA constructs for 48 hrs and re-plated in serial dilution for clonogenic survival assessment. (F) Clonogenic growth inhibition of long- and short-term passaged, patient-derived glioblastoma lines by haloperidol. Cells were incubated with haloperidol at the indicated concentrations for the duration of the clonogenic assay. (G) Multiple DRD2 antagonists suppressed U87MG clonogenic growth. S: spiperone (5 μM, 10 μM), H: haloperidol (5 μM, 10 μM), R: risperidone (10 μM, 25 μM), L: L-741,626 (10 μM, 25 μM). Cells were incubated with DRD2 antagonists at the indicated doses for the duration of the clonogenic assay. (H) DRD2 knockdown abolished U87MG subcutaneous xenograft growth. Top: 3×10⁵ cells transduced with inducible shRNA2 against DRD2 were injected subcutaneously into the flanks of Nu/Nu mice. 7 days later, the mice were randomized to drinking water with or without doxycycline (1 mg/ml). Tumor size was monitored every 5-7 days. Bottom: Inducible shRNA silencing of DRD2. U87MG cells with a stably integrated, inducible shRNA construct were treated with doxycycline or vehicle for 72 hrs. RNA was then extracted and DRD2 RT-PCR was performed. (I) Expression of an shRNA resistant form of DRD2 (termed DRD2RR) suppressed the anti-proliferative effects of shDRD2 in vivo. Top: U87MG cells transduced with shRNA2 were transfected with DRD2 or DRD2RR, and injected subcutaneously. 7 days post-injection, the mice were fed doxycycline (1 mg/ml) containing water. Tumor size was monitored every 5-7 days. Bottom: DRD2RR expression is resistant to the DRD2 silencing effects of shRNA2. U87MG cells with stably integrated shRNA2 and transfected with DRD2 or DRD2RR (noted here as RR) were treated with doxycycline for 72 hrs. Whole cell lysates were prepared from these cells for DRD2 immuno-blotting.

Figure 2. Increased DRD2 expression in glioblastoma specimens

(A) Overexpression of DRD2 mRNA in glioblastoma specimens relative to surrounding normal brain tissues. DRD2 mRNA expression was analyzed using qPCR; matched normal-glioblastoma specimens from five patients were tested. T: Tumor; N: Normal brain. (B) DRD2 protein expression was confirmed using three additional matched glioblastoma/normal brain pairs by immuno-blotting. T: Tumor; N: Normal brain. Tubulin: loading control. The ratio of DRD2 to tubulin was quantitated and shown in the bottom panel. (C) Increased expression of DRD2 in GEMM glioblastoma lines. DRD2 mRNA expression was assessed by qPCR. DRD2 mRNA level in a glioblastoma line derived from hGFAP-Cre+; p53^−/−; NF1^flox/flox (noted as p53^−/−NF1^−/−) GEMM [29] was compared to that in an astrocyte line derived from the hGFAP-Cre+; NF1^flox/flox (noted as NF1−/−) GEMM. DRD2 mRNA level in a glioblastoma line derived from hGFAP-Cre+; p53^lox/lox; Pten^lox/+GEMM (noted as p53 ^−/−Pten ^+/−) GEMM [30] was compared to that in an astrocyte line derived from an hGFAP-Cre+; p53^lox/lox GEMM (noted as p53^−/−). (D) Elevated DRD2 expression in glioblastomas formed in vivo in Gtv-a Ink4a-Arf−/− mice stereotactically injected with RCAS-PDGFB-HA [31]. This expression level was compared to the contra-lateral normal cortex. Three sets of matched cortex/glioblastoma specimens are shown. For all qPCRs, the results were normalized to 18S rRNA. Comparable results were obtained when normalized to actin or GAPDH. (E) Sensitivity of GEMM derived glioblastoma and astrocyte lines to haloperidol. Glioblastoma lines were more sensitive to haloperidol relative to astrocyte lines. Cells were seeded at ~50% confluency and treated with 10 μM haloperidol for 5 days. Viability was determined using the CellTiter-Blue viability assay (Promega).

Figure 3. DRD2 signaling through a GNAI2-Rap1-ERK axis

(A) DRD2 antagonists decreased pERK accumulation in the U87MG glioblastoma cell line. D: DMSO. S: spiperone (5 μM), H: haloperidol (5 μM), R: risperidone (10 μM), L: L-741,626 (10 μM). Cells were treated with the various antagonists for 1 hr before lysis/immuno-blotting. (B) Decreased pERK activation by DRD2 shRNA can be rescued by expression of an RNAi resistant form of DRD2 (DRD2RR). U87MG lines harboring a doxycycline inducible shDRD2 construct and pCMV6-DRD2RR or pCMV6 were treated with vehicle or doxycycline (1 μg/ml) for 48 hrs before lysis. Bottom panel: quantitative ratio of pERK to ERK. (C) DRD2 antagonists decreased pERK accumulation in the patient-derived CMK3 GSC line in a dose-dependent manner. D: DMSO. S: spiperone. H: haloperidol. CMK3 was treated with the various antagonists for 1 hr before lysis. (D) Treatment with a DRD2 agonist, quinpirole, enhanced pERK accumulation (left upper). CMK3 was treated with quinpirole for 1 hr prior to lysis/immuno-blotting. Lower panel: quantitative ratio of pERK to ERK. Quinpirole promoted cellular growth in the CMK3 GSC line detected by CellTiter-Blue (right). Cells were treated with 10 μM quinpirole for 5 days. (E) qPCR analysis revealed increased GNAI2 mRNA expression in glioblastoma specimens relative to the surrounding normal brain. 5 matched normal-glioblastoma specimens are shown. N: normal brain; T: glioblastoma tumor specimens. GNAI2 mRNA level was normalized to 18S rRNA mRNA. Similar results were obtained when normalized to actin or GAPDH. (F) GNAI2 siRNA transfection compromised proliferation of U87MG cells (top) and reduced pERK accumulation (bottom). #1 and #2 denotes two independent siRNAs against GNAI2. Cells were seeded at ~50% confluency, transfected with siRNAs for 48 hrs, and re-plated in serial dilution for clonogenic survival assessment. (G) The pro-proliferative effect of quinpirole in the patient-derived 1123 GSC line was inhibited by GNAI2 silencing. Cells were transfected with siGNAI2s for 48 hrs prior to treatment with 10 μM quinpirole for 5 days. Viability was determined using CellTiter-Blue (Promega). (H) GNAI2 silencing inhibited quinpirole-induced pERK accumulation and reversed quinpirole-suppression of Rap1-GTP accumulation in U87MG cells. Cells were transfected with siRNAs for 48 hrs prior to the treatment with 10 μM quinpirole for 1 hr.

Figure 4. Haloperidol enhances the tumoricidal effect of the EGFR inhibitor, AG1478

(A) Haloperidol enhanced the tumoricidal effect of the EGFR inhibitor, AG1478, against the U87MG glioblastoma line in vitro (top). Cells were treated with the various agents for 5 days; viability was determined using CellTiter-Glo (Promega). The enhanced tumor kill correlated with pERK reduction (bottom). For the pERK analysis, cells were treated with the various agents for 1 hr before lysis/immuno-blotting. Combination index (CI) value was 0.32 in U87MG cells using the Chou-Talalay method. (B) Haloperidol enhanced the tumoricidal effect of the EGFR inhibitor, AG1478, against the 1123 GSC line in vitro (top). The enhanced tumor kill correlated with pERK reduction (bottom). Cells were treated as described above and viability was measured using CellTiter-Blue (Promega). CI value was 0.29 in 1123 GSCs using the Chou-Talalay method. (C) Combined effect of haloperidol and AG1478 on subcutaneous U87MG xenograft growth. Nu/Nu mice were injected with U87MG cells subcutaneously. 8 days later the mice were randomized to: vehicle (n=7), haloperidol (10 mg/kg, n=8), AG1478 (10 mg/kg, n=8) or combined haloperidol/AG1478 treatment (n=8). *: P<0.05. (D) Combined effect of haloperidol and AG1478 on an orthotopic 1123 mouse model in vivo. 7 days after intracranial injection with the 1123 GSCs (1x10⁴ cells), Nu/Nu mice were treated daily by oral gavage with vehicle (n=5), AG1478 (10 mg/kg, n=5), haloperidol (10 mg/kg, n=5) or combined AG1478/haloperidol (n=5). All xenograft experiments were repeated three times. Representative data are shown. *: P<0.05. (E) Synergistic effect of AG1478 and haloperidol in suppressing pERK in vivo. 3 weeks after intracranial injection with 1123, Nu/Nu mice were treated with vehicle, AG1478 (10 mg/kg), haloperidol (10 mg/kg), or combined AG1478/haloperidol for 2 hrs, followed by tumor dissection. Whole cell lysates were analyzed for pERK and pAKT levels, with Ku86 as loading control. The quantitative ratios of pERK to ERK are shown in the right panel.