Dicer knockdown inhibits endothelial cell tumor growth via microRNA 21a-3p targeting of Nox-4

Abstract

MicroRNAs (miR) are emerging as biomarkers and potential therapeutic targets in tumor management. Endothelial cell tumors are the most common soft tissue tumors in infants, yet little is known about the significance of miR in regulating their growth. A validated mouse endothelial cell (EOMA) tumor model was used to demonstrate that post-transcriptional gene silencing of dicer, the enzyme that converts pre-miR to mature miR, can prevent tumor formation in vivo. Tumors were formed in eight of eight mice injected with EOMA cells transfected with control shRNA but formed in only four of ten mice injected with EOMA cells transfected with dicer shRNA. Tumors that formed in the dicer shRNA group were significantly smaller than tumors in the control group. This response to dicer knockdown was mediated by up-regulated miR 21a-3p activity targeting the nox-4 3'-UTR. EOMA cells were transfected with miR 21a-3p mimic and luciferase reporter plasmids containing either intact nox-4 3'-UTR or with mutation of the proposed 3'-UTR miR21a-3p binding sites. Mean luciferase activity was decreased by 85% in the intact compared with the site mutated vectors (p < 0.01). Attenuated Nox-4 activity resulted in decreased cellular hydrogen peroxide production and decreased production of oxidant-inducible monocyte chemoattractant protein-1, which we have previously shown to be critically required for endothelial cell tumor formation. These findings provide the first evidence establishing the significance of dicer and microRNA in promoting endothelial cell tumor growth in vivo.

Keywords: Dicer; Endothelial Cell; Hemangioendothelioma; MicroRNA; Monocyte Chemoattractant Protein-1; NADPH Oxidase; Tumor; miR-21a-3p.

FIGURE 1.

MicroRNA biogenesis is essential for endothelial cell tumor formation. EOMA cell stable transfectants were generated using lentiviral shRNA delivery. A and B, targeted post-transcriptional gene silencing of dicer was confirmed by Western blot (A) and real time PCR (B). C and D, Matrigel angiogenesis assay was used to evaluate the functional effects of dicer knockdown in vitro. Cells were stained with calcein-AM (C), and the area within the formed tubes was quantitated by analyzing three high powered fields per well using AxioVision Rel 4.8 software (D). Matrigel tube formation was compromised in dicer knockdown EOMA cells. E, reduced tumor size in response to dicer knockdown. F, incidence of tumor formation in mice injected with EOMA cells transfected with controls shRNA or dicer shRNA. G, tumor volume as quantified using calipers (length × width × height). The results are means ± S.D. of at least three independent experiments. *, p < 0.05.

FIGURE 2.

Dicer knockdown attenuated Nox-4 activity in tumor forming EOMA cells. A, murine dicer sequence identification of dicer siRNA binding sites in green and shRNA binding site in yellow confirm distinct binding sites within dicer mRNA. B and C, transfection with dicer siRNA decreased dicer protein (B) and mRNA (C) expression. D, mRNA expression of gp91 isoforms present in EOMA cells. Only Nox-4 was present at significant levels and was affected significantly by dicer knockdown. E, dicer knockdown lowered Nox-4 protein expression. F and G, the effect of dicer knockdown on Nox-4 function was measured using flow cytometry (F) to detect oxidation of 5-(and-6)-carboxy-2′,7′-dichlorofluorescein diacetate (DCF) and by the Amplex Red (Invitrogen) assay to measure H₂O₂ production (G). The results are means ± S.D. of at least three independent experiments. *, p < 0.05.

FIGURE 3.

Dicer knockdown compromised MCP-1 protein expression and function. A and B, effect of dicer knockdown on MCP-1 mRNA expression (A) and MCP-1 protein expression (B). C, loss of MCP-1 function was confirmed using a Transwell migration assay. Lower chambers contained EOMA cells transduced with either control shRNA or dicer shRNA, and upper wells contained RAW cells, a murine macrophage cell line. D, migrated cells were fixed with paraformaldehyde and stained with DAPI, and the numbers of migrating cells were scored. Cells were counted using total cells from each image divided with total area (2.4 mm²) to obtain final migrated EOMA cells/mm². The results are means ± S.D. of at least three independent experiments. *, p < 0.05.

FIGURE 4.

Dicer knockdown compromised processing of pre-miR to mature miR, resulting in decreased abundance of angiogenic miRs 18a, 27b, and 32 but increased abundance of miR-21a-3p. A, the amount of RNA 50–100 nucleotides in size, consistent with the size of pre-miR, was measured in response to dicer knockdown. Prior to analysis, the RNA integrity (RIN) of the samples was analyzed on a scale of 1–10 (where 10 is the highest integrity) to ensure that fragments were not derived from sample degradation. Dicer knockdown resulted in the accumulation of RNA fragments the size of pre-miR. B and C, real time PCR analyses showed that dicer knockdown in EOMA cells resulted in decreased processing of pre-miR to mature miR as shown with miR18a, miR27b, and miR32 both in vitro (B) and in vivo (C) from HE tumor extracts generated using EOMA transfected with dicer shRNA. D and E, dicer knockdown did not alter cellular levels of miR21a but did increase miR21a-3p in dicer siRNA transfected (D) and dicer shRNA transfected cells (E). F and G, HE specimens showed increased miR-21a-3p expression between days 4 and 7 after injection with untreated EOMA cells with a corresponding decrease in dicer mRNA levels confirming in vitro findings. The results are means ± S.D. of at least three independent experiments. *, p < 0.05.

FIGURE 5.

miR-21a-3p loading into the RISC complex and half-life were increased in response to dicer knockdown. Ago-2 immunoprecipitation (IP) was performed to identify miR loaded into the RISC complex and ready to bind mRNA targets. A, immunoblot for Ago-2 immunoprecipitation showed significant levels of Ago-2 protein in the immunoprecipitate, but not in cell lysate or cell supernatant fractions. B, miR-21a-3p abundance after Ago-2 immunoprecipitation was significantly increased after dicer knockdown. C, real time PCR analysis of miR-21a after Ago-2 immunoprecipitation remained unchanged in response to dicer knockdown. D and E, measurement of pri-miR-21a-3p (D) and pre-miR-21 (E) by real time PCR did not show any significant increase in abundance following dicer knockdown. F, miR-21a-3p decay was examined 6 h after addition of actinomycin D (2.5 μg/ml) in control and dicer siRNA transfected EOMA cells. A significant decrease in miR-21a-3p in control siRNA transfected cells but not in dicer siRNA transfected cells was noted. The results are means ± S.D. of at least three independent experiments. *, p < 0.05.

FIGURE 6.

miR-21a-3p inhibitors rescued angiogenic phenotype of dicer knocked-down EOMA cells. A, effectiveness of miR-21a-3p mimic and inhibitor in EOMA cells transfected with control siRNA or dicer siRNA was confirmed by real time PCR. B, miR-21a-3p mimic and inhibitor did not have off target effects that alter dicer mRNA levels. C, addition of miR-21a-3p mimic to EOMA cells transfected with dicer siRNA resulted in decreased Nox-4 protein levels that were restored to basal (control siRNA) levels following addition of miR-21a-3p inhibitor. D, Matrigel^TM tube formation showed profound loss of angiogenic response in dicer siRNA transfected EOMA cells with addition of miR-21a-3p mimic and restoration of angiogenic response with addition of miR-21a-3p inhibitor. E, quantification of length of tube formation. The results are means ± S.D. of at least three independent experiments. *, p < 0.05.

FIGURE 7.

miR-21a-3p targets Nox-4 in EOMA cells. A, proposed binding sites for miR-21a-3p were identified in the Nox-4 3′-UTR that was inserted into a firefly luciferase plasmid vector (wild type). The binding sites were altered as indicated by yellow highlights to generate a second (mutated) firefly luciferase plasmid vector. B, EOMA cells were co-transfected with either the wild type or mutated Nox-4 3′-UTR firefly luciferase plasmids and a Renilla luciferase plasmid. Nox-4 transcriptional activation as measured by luciferase assay demonstrated significant loss of luciferase activity in wild type vectors in the presence of elevated miR21a-3p and no difference in the luciferase levels in cells transfected with the mutated Nox-4 3′-UTR vector. C, Amplex Red assay demonstrated loss of Nox-4 function with significantly decreased H₂O₂ levels observed in miR-21a-3p mimic delivered cells. The results are means ± S.D. of at least three independent experiments. *, p < 0.05.

FIGURE 8.

Lowered Nox-4 in response to treatment with miR-21a-3p mimic. A, EOMA cells with or without miR-21a-3p mimic were labeled with fluorescent probes DAPI (blue) for nucleus, phalloidin (red) for actin and (green) for Nox-4. Confocal microscopy showed that Nox-4 was localized in perinuclear area in EOMA cells, and there was significant decrease in Nox-4 fluorescence intensity in miR-21a-3p delivered cells. B, fluorescence intensity of images was quantified using Olympus FV10-ASW software. C, Z-stack image showing Nox-4 localized mostly in the perinuclear space. D, immunoblot analyses of nuclear and cytosolic fractions of EOMA cell lysates compare relative distribution of Nox-4 protein. The results are expressed as means ± S.D. of at least three independent experiments. *, p < 0.05.