The MYCN 5' UTR as a therapeutic target in neuroblastoma

Abstract

Tumor MYCN amplification is seen in high-risk neuroblastoma, yet direct targeting of this oncogenic transcription factor has been challenging. Here, we take advantage of the dependence of MYCN-amplified neuroblastoma cells on increased protein synthesis to inhibit the activity of eukaryotic translation initiation factor 4A1 (eIF4A1) using an amidino-rocaglate, CMLD012824. Consistent with the role of this RNA helicase in resolving structural barriers in 5' untranslated regions (UTRs), CMLD012824 increased eIF4A1 affinity for polypurine-rich 5' UTRs, including that of the MYCN and associated transcripts with critical roles in cell proliferation. CMLD012824-mediated clamping of eIF4A1 spanned the full lengths of mRNAs, while translational inhibition was mediated through 5' UTR binding in a cap-dependent and -independent manner. Finally, CMLD012824 led to growth inhibition in MYCN-amplified neuroblastoma models without generalized toxicity. Our studies highlight the key role of eIF4A1 in MYCN-amplified neuroblastoma and demonstrate the therapeutic potential of disrupting its function.

Keywords: CP: Cancer; CP: Molecular biology; MYCN; MYCN amplification; PAR-CLIP; amidino-rocaglate; eIF4A1; neuroblastoma; ribosome profiling; rocaglate; translation; translation regulation.

Figure 1.. MYCN-amplified neuroblastomas exhibit translation initiation factor upregulation and are enriched for polypurine-rich 5′ UTR mRNAs

(A) Hierarchical clustering of translation initiation factor gene expression in primary neuroblastoma tumors (n = 498, GSE62564), ranked by MYCN expression. Z score = mean ± SD. (B) Violin plots showing expression of the indicated initiation factors in tumors with lowest and highest MYCN expression levels, as depicted in (A) (n = 30 each) (P, Student’s t test). (C) Hierarchical clustering of the tumors in (A) ranked by c-MYC expression. (D) Violin plots depicting the expression of the indicated initiation factors in primary tumors in (C) with the highest and lowest c-MYC (n = 30) expression levels. (E) Correlogram of MYCN and translation initiation factor gene expression in MYCN-amplified primary tumors (n = 92, GSE62564). Circles represent Spearman’s correlation coefficients, p < 0.01. (F) ChIP-seq profiles of MYCN binding at the indicated gene loci in Kelly neuroblastoma cells. X axis, genomic position; y axis, MYCN binding in units of reads per million (rpm). (G) Polypurine ranking of mRNAs expressed in primary neuroblastomas. Blue, bottom 25%; red, top 25%. (H) Fold change (FC) distributions of highly variable genes in tumors with the highest and lowest (top and bottom 10%) MYCN expression levels (n = 30 each) (p < 0.01, Student’s t test). (I) Polypurine rank distribution of the highly variable upregulated genes (high, FC > 5; low, FC < 2) (p < 0.01, Student’s t test). (J) Volcano plot of genes correlated with eIF4A1 in primary tumors (n = 498, FDR < 0.05). See also Figure S1 and Table S1.

Figure 2.. CMLD012824 exhibits differential cytotoxicity in neuroblastoma cells

(A) Cell viability of MYCN-amplified (red), nonamplified (blue) human neuroblastoma, and non-transformed (gray) cells, treated with varying concentrations of CMLD012824 (ADR-824) for 72 h. Data = mean ± SD, n = 3 replicates. Inset: chemical structure of ADR-824. (B) Upper, western blot (WB) analysis of PARP cleavage. GAPDH, loading control; middle, annexin V; lower, membrane integrity analyses in MYCN-amplified (Kelly) and nonamplified (SK-N-AS) cells exposed to ADR-824 at the indicated doses. Data = mean ± SD, n = 3. (C) Flow cytometry analysis of propidium iodide and EdU incorporation in the indicated neuroblastoma and non-transformed (HEK293) cells 24 h post exposure to ADR-824 (10 nM). Bottom, quantification of mean ± SD, n = 3 biological replicates. (D) WB analysis of cell cycle markers in the indicated cells 24 h after varying doses of ADR-824. eIF4A1, loading control. (E) Metabolic labeling of nascent protein synthesis in the indicated cells exposed to ADR-824, CHX, or DMSO for 1 h. Bottom, quantification of mean ± SD, n = 2. See also Figure S2.

Figure 3.. ADR-mediated inhibition of eIF4A1 impairs MYCN translation

(A) Immunofluorescence images of the MYCN protein in MYCN-amplified (Kelly) cells at 1 h post ADR-824 (10 nM) treatment. PTBP1, polypurine-poor control. Blue, DAPI nuclear stain. Scale bar, 10 μm. (B) WB analysis of MYCN/c-MYC expression at 4 h post ADR-824 treatment at the indicated doses in MYCN-amplified (+) and nonamplified (−) cell lines. (C) WB analysis of eIF4A1 protein levels after treatment as in (B). (D) WB analysis of MYCN expression in MYCN-amplified (Kelly) cells treated with CHX (10 μg/mL) or ADR-824 (10 nM) for the indicated times. XRN2, polypurine-poor control. (E) RT-qPCR analysis of MYCN mRNA in cells treated with CHX or ADR-824 as in (D). Data represent mean ± SD, n = 2. (F) WB analysis of MYCN expression in MYCN-amplified neuroblastoma cells treated with actinomycin D (1 μg/mL) with or without MG132 (100 mM) or ADR-824 (10 nM). GAPDH, loading control for (B), (C), (D), and (E). (G) ChIP-PCR analysis of MYCN and PHOX2B at the promoters of the indicated genes in MYCN-amplified neuroblastoma cells under DMSO- and ADR-824- (10 nM) treated conditions. Percent binding relative to input signal and IgG control is shown. Data = mean ± SD, n = 3.

Figure 4.. ADR-824 causes selective translation repression of long, polypurine-rich mRNAs

(A) Scatterplot of total vs. ribosome-associated mRNA changes in DMSO- vs. ADR-824-treated (10 nM × 1 h) MYCN-amplified Kelly neuroblastoma cells (n =3 biological replicates each). p < 0.1, Anota2seq analysis (see STAR Methods). Dotted black lines indicate 1.5 FC in total mRNA (x axis) and ribosome occupancies (y axis). (B) Functional enrichment of unique differentially regulated transcripts (FC > 1.5, p < 0.1, Fisher’s exact test in Enrichr). (C) Volcano plot of translationally regulated mRNAs in DMSO- vs. ADR-824-treated cells (FC > 1.5, p < 0.1, Anota2seq). (D) Motif enrichment analysis of the top motifs in the downregulated mRNA subset, trained against a background list of unregulated transcripts. E-values determined by MEME (see STAR Methods). (E and F) 5′ UTR length distribution (E) and polypurine rank distribution (F) of translationally regulated transcripts (p < 2.2e–16; Student’s t test each). (G and H) Scatterplots of polypurine (G) and GC content (H) changes by 5′ UTR length in upregulated versus downregulated mRNAs (Loess regression analysis, shaded regions, 95% confidence intervals). (I) Heatmaps of translational efficiency (TE) changes (n = 76; p < 0.1) and polypurine ranking of MYCN-regulated target genes in ADR-824 vs. DMSO-treated cells (Z score = mean ± SD; p values, Student’s t test). (J) Ribosome occupancy profiles of the polypurine-rich MYCN-regulated genes. GAPDH, polypurine-poor control. (K) RT-qPCR analysis of the indicated mRNA distributions in polysome fractions pooled according to polysome occupancy. Light: 1–3 polysomes; heavy: 4+ polysomes. Signal was calculated by 2^{^}-ΔΔCt method, normalized to total RNA in gradient and GAPDH controls. Data = mean ± SD, n = 3. See also Figures S3 and S4.

Figure 5.. ADR-824 augments mRNA binding of eIF4A1 along the full lengths of mRNAs

(A) RT-qPCR analysis of 5′ UTR polypurine-rich and -poor mRNAs bound to endogenous eIF4A1 protein immunoprecipitated in lysates from DMSO- or ADR-824-treated (10 nM × 1 h) MYCN-amplified cells. Data = mean ± SD, n = 3. ***p < 0.0001, Student’s t test. (B) Frequency distribution of eIF4A1-bound clusters by nucleotide cluster length in naive- and ADR-824-treated conditions. Data represent consensus clusters,n = 2 biological replicates per condition. The inset zooms into the range of cluster length with highest frequencies in both conditions. (C) Volcano plot of the relative changes in binding (FC > 1.5) of eIF4A1-bound mRNAs in DMSO- or ADR-824-treated cells as in (A) (p < 0.1, Anota2seq). (D) Metagene analysis of eIF4A1-bound clusters along the indicated mRNA regions in DMSO- and ADR-824-treated cells. Data represent mean coverage (RPM), n = 2 biological replicates. (E) Top motifs identified in eIF4A1 clusters that map to the indicated mRNA regions in DMSO- and ADR-824-treated cells. E-values adjusted to motif frequency are shown. (F) Representative tracks of eIF4A1 binding to the MYCN mRNA. Signal in units of reads per kilobase per million (RPKM). (G) Representative tracks of eIF4A1 binding (PAR-CLIP) and ribosome occupancy (RIBO-SEQ) profiles of polypurine-rich and -poor mRNAs. Black boxes, 5′ UTR regions. See also Figures S5 and S6.

Figure 6.. ADR-824 clamps eIF4A1 onto select polypurine-rich cellular mRNAs in a 5′ UTR-dependent and cap-independent manner

(A) WB analysis of exogenously expressed 5′ UTR-depleted MYCN in MYCN-nonamplified SK-N-AS neuroblastoma cells, treated with the indicated doses of ADR-824 (1 h). GAPDH, loading control. Schematic depicts MYCN 5′ UTR-deleted construct. (B) Renilla luciferase activity of in-vitro-translated endogenous 5′ UTR sequences cloned upstream of luciferase in the presence of DMSO or ADR-824 (25 nM). Signal is normalized to internal globin-firefly luciferase control. CKS2 and XRN2 RNAs, polypurine-poor controls; HCV IRES RNA, eIF4A-independent control. Data = mean ± SD, n = 3. ***p < 0.0001, Student’s t test. (C) Top: schematic representation of the WT MYCN 5′ UTR, 5′ deletion mutant (MYCN 5′ DEL), and 3′ deletion mutant (MYCN 3′ DEL). Bottom left (three panels), luciferase activity of in-vitro-translated RNAs generated with the canonical MYCN m⁷G-cap or nonfunctional ApppG analog (A-cap). Bottom right, percent suppression of translation. Data = mean ± SD, n = 2, representative of 3 independent experiments. ***p < 0.0001, Student’s t test. (D) Renilla luciferase activity from in vitro translation of indicated RNAs at the indicated concentrations in the presence of globin-firefly RNA (200 ng per reaction). Data = mean ± SD, n = 3.

Figure 7.. ADR-824 inhibits tumor growth and improves survival in neuroblastoma

(A) Tumor volumes of NB-9464 xenograft tumors in C57BL/6J mice (n = 10) treated 3 times weekly with the indicated doses of ADR-824. Dashed lines indicate beginning and end of treatment. Each curve corresponds to a separate animal (vehicle vs. 0.1 mg/kg, p < 0.25; vs. 0.2 mg/kg, p < 0.006, Welch’s test). (B) Tumor volumes of MYCN-amplified PDX models (COG-N-415x) treated 3 times weekly with vehicle (n = 7) or ADR-824 (n = 10). Data = mean ± SD. Statistically significant differences between treatment groups were observed on days 21, 23, and 25 (**p < 0.001, ***p < 0.0001, Student’s t test) after which no vehicle-treated animals survived. (C) Kaplan-Meier analysis of COG-N-415x PDX-bearing mice in (B) (p < 0.02, Mantel-Cox t test). (D) Representative images of hematoxylin and eosin (H&E) and immunohistochemistry analyses (IHC) of the indicated tumor markers (Ki67 (proliferation), CC3, cleaved caspase 3 [apoptosis]) in vehicle- (top) and ADR-824-treated (bottom) mice. Scale bar, 100 μm. (E) WB analysis of the indicated polypurine-rich, -poor, translation factor, and control proteins in COG-N-415x PDX tumors (t) in (D). See also Figure S7.