Secondary structures that regulate mRNA translation provide insights for ASO-mediated modulation of cardiac hypertrophy

Abstract

Translation of upstream open reading frames (uORFs) typically abrogates translation of main (m)ORFs. The molecular mechanism of uORF regulation in cells is not well understood. Here, we data-mined human and mouse heart ribosome profiling analyses and identified a double-stranded RNA (dsRNA) structure within the GATA4 uORF that cooperates with the start codon to augment uORF translation and inhibits mORF translation. A trans-acting RNA helicase DDX3X inhibits the GATA4 uORF-dsRNA activity and modulates the translational balance of uORF and mORF. Antisense oligonucleotides (ASOs) that disrupt this dsRNA structure promote mORF translation, while ASOs that base-pair immediately downstream (i.e., forming a bimolecular double-stranded region) of either the uORF or mORF start codon enhance uORF or mORF translation, respectively. Human cardiomyocytes and mice treated with a uORF-enhancing ASO showed reduced cardiac GATA4 protein levels and increased resistance to cardiomyocyte hypertrophy. We further show the broad utility of uORF-dsRNA- or mORF-targeting ASO to regulate mORF translation for other mRNAs. This work demonstrates that the uORF-dsRNA element regulates the translation of multiple mRNAs as a generalizable translational control mechanism. Moreover, we develop a valuable strategy to alter protein expression and cellular phenotypes by targeting or generating dsRNA downstream of a uORF or mORF start codon.

Fig. 1. Crosstalk between a uORF and an adjacent double-stranded RNA structural element inhibits mORF translation.

a, b Sequence analysis of Ensemble transcripts demonstrates the abundance of 5′ UTR start codons (a) and 5′ UTR GC content relative to the CDS or 3′ UTR (b). The solid line indicates the median, while the dotted lines indicate the interquartile range. More than half the human mRNA 5′ UTRs contain AUG start codons. c Left panel: schematic of FLuc reporter constructs. Right panel: dual luciferase reporter assay using a series of constructs that contain uORF start codon and adjacent dsRNA KanHP1 located at increasing distances in 3 nt intervals. d RT-qPCR of FLuc mRNA normalized to ACTB from cells in (c). No AUG control: ATG-to-TTG mutation in the −2 construct. e Dual luciferase reporter assay using mutant constructs. No AUG: ATG-to-TTG mutation. AUG −2: start codon is located at the −2 position relative to the hairpin (sequences available in Supplementary Information). WT: stable hairpin. MM: three mismatched mutations were introduced in the hairpin to disrupt the dsRNA structure. f RT-qPCR of FLuc mRNA normalized to ACTB from cells in e. Data are represented as mean ± SD. **P < 0.01, ***P < 0.001; Statistical significance was confirmed by unpaired two-tailed Student t test for c–f (N = 3 biological replicates). Source data are provided as a Source Data file.

Fig. 2. Double-stranded RNA element cooperates with initiation codon to activate GATA4 uORF translation.

a Schematic highlighting the key steps of RNA SHAPE workflow used for the human GATA4 uORF-dsRNA-bearing RNA sequence. b RNA SHAPE analysis of human GATA4 uORF-dsRNA-bearing RNA region. A sequencing gel resolved individual nucleotides of the SHAPE assay products. Primers 1 and 2 were used to detect the downstream and upstream dsRNA strands in the GATA4 RNA. Experiments in b were carried out twice, and one representative result is shown. c Densitometric quantification of band intensity corresponding to the structured nature of a given nucleotide using SAFA. The upper inserts represent two strands of the dsRNA. Lower NAI SHAPE reactivity (band intensity) indicates dsRNA structure. Two biological replicates from b were plotted. d The predicted lowest free energy secondary structure of human GATA4 uORF surrounding region based on SHAPE reactivity data using RNAstructure. The resolved region with detectable SHAPE activity is at 252-440 nt. e Left: Schematic of firefly luciferase (FLuc) reporters that include variants of the full-length RNA sequences of human GATA4 mRNA before the start codon of mORF (including uORF-dsRNA). ΔuORF ATG-to-TTG start codon mutation, MM mismatch mutations in dsRNA, Rescue:mismatch and compensatory mutations. Middle: dual luciferase reporter assay with WT and mutants. Following transfection into HEK293T cells, FLuc levels were normalized to a control RLuc reporter. Right: RT-qPCR of FLuc mRNA normalized to ACTB. Data are represented as mean ± SD. **P < 0.01; Statistical significance was confirmed by an unpaired two-tailed Student t test for e (N = 3 biological replicates). Source data are provided as a Source Data file.

Fig. 3. Mechanism-based design of two types of ASOs for regulating GATA4 mRNA translation.

a Left: Schematic of screening of several 16-nt 2′-O-methylated uORF-targeting ASOs that emulates the presence of dsRNA upstream, at, or downstream of the human GATA4 uORF start codon. The control ASO used was the mismatch version derived from the ASO_-3. Middle: Western blot analysis of GATA4 protein expression normalized by β-actin in human immortalized AC16 cardiomyocyte cells transfected with 50 nM of ASOs. Right: Quantification of GATA4 mRNA expression from the middle panel and ACTB mRNA is used for normalization. b Schematic of two types of ASOs targeting GATA4 uORF-dsRNA element. Class I (ASO1; the same as ASO_ + 2 in a) is designed for forming artificial dsRNA downstream of the uORF strand, while Class II (ASO2) is intended to prevent endogenous dsRNA formation and free up the uORF strand. c, d Dual luciferase reporter assay with WT and ΔuORF reporters after co-transfection of ASO1 (c) and ASO2 (d) in HEK293T cells. FLuc activity was normalized to WT and ΔuORF reporter activity with no ASO1 and ASO2 treatment, respectively. e Left and middle: Western blot analysis of GATA4 protein levels normalized to β-actin in human AC16 cells transfected with 10 or 50 nM of ASO1 and ASO2. Total ASO concentrations were equalized to 50 nM using the control ASO. Right: RT-qPCR measurement of relative GATA4 mRNA level normalized to ACTB. f Polysome profiles generated from absorbance readings at 253 nm for lysates of AC16 cells transfected with either control ASO, ASO1, or ASO2. g RT-qPCR measurement of GATA4 mRNA distributions across various fractions in polysome profiles. Experiments in (f) and (g) were repeated three times, and representative data were shown. Data are represented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001; Statistical significance was confirmed by unpaired two-tailed Student t test for a, c, d, and 1-way ANOVA followed by Holm-Sidak post hoc test for e (N = 3 biological replicates). Source data are provided as a Source Data file.

Fig. 4. uORF-targeting ASOs regulate GATA4 protein expression and cardiomyocyte hypertrophy in human ESC-derived cardiomyocytes.

a Western blot analysis of GATA4 protein expression in human ESC-derived WT and homozygous ∆uORF CMs with ASO1 and ASO2 (50 nM) or control ASO treatment. b RT-qPCR measurement of GATA4 mRNA normalized to ACTB. c Representative images of α-Actinin (green) and NKX2-5 (red) immunostaining in addition to DAPI (blue) in ESC-derived CMs treated with control ASO, ASO1, or ASO2. Scale bar: 100 μm. Co-staining of α-actinin (green) and NKX2-5 (red) discerns CMs from mis-differentiated cells. Cell surface area was measured for five different clumps of cells as the total surface area was divided by the number of cells. d MYH6 mRNA expression in ESC-derived WT and homozygous ∆uORF CMs at baseline. e MYH6 mRNA expression in ESC-derived WT and homozygous ∆uORF CMs with ASO1 and ASO2 or control ASO (50 nM) treatment. Data are represented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001; Statistical significance was confirmed by unpaired two-tailed Student t test for c and d, and one-way ANOVA followed by Holm–Sidak post hoc test for a, b, and e (N = 3 biological replicates). Source data are provided as a Source Data file.

Fig. 5. GATA4 uORF-targeting ASOs counteract TAC-induced cardiac hypertrophy.

a Schematic of TAC surgery-induced cardiac hypertrophy mouse model for tail-vein-injected nanoparticle-encapsulated control or uORF-ASOs. Heart samples were then harvested and analyzed. Both sexes were almost equally represented, with male and female dots depicted as black and pink, respectively (b–h). b Cardiac hypertrophy was measured as heart weight normalized to tibia length (HW/TL) in harvested hearts. Ctrl ASO-Sham: N = 9; Gata4 ASO-Sham: N = 10; Ctrl ASO-TAC: N = 10; Gata4 ASO-TAC: N = 13. c Representative images of wheat germ agglutinin-fluorescein in mouse heart transverse sections highlighting CM perimeter. Scale bars, 20 μm. N = 3 hearts with >250 CMs quantified per heart. d Representative images of picrosirius red staining of heart sections (left) to quantify the fibrotic area (right). Scale bar: 1 mm. Ctrl ASO-Sham: N = 9; Gata4 ASO-Sham: N = 10; Ctrl ASO-TAC: N = 10; Gata4 ASO-TAC: N = 12. e Echocardiographic cardiac function measurements of ejection fraction (EF). Ctrl ASO-Sham: N = 9; Gata4 ASO-Sham: N = 10; Ctrl ASO-TAC: N = 10; Gata4 ASO-TAC: N = 13. f Western blot analysis of GATA4 protein expression normalized to β-actin in the hearts samples. Ctrl ASO-Sham: N = 9; Gata4 ASO-Sham: N = 10; Ctrl ASO-TAC: N = 10; Gata4 ASO-TAC: N = 10. g RT-qPCR measurement of Gata4 mRNA expression normalized to Actb mRNA in the hearts. Ctrl ASO-Sham: N = 9; Gata4 ASO-Sham: N = 10; Ctrl ASO-TAC: N = 10; Gata4 ASO-TAC: N = 12. h RT-qPCR measurement of hypertrophy marker gene mRNA Nppa and Nppb in mouse heart samples. Actb mRNA was used as a normalizer. Ctrl ASO-Sham: N = 9; Gata4 ASO-Sham: N = 10; Ctrl ASO-TAC: N = 10; Gata4 ASO-TAC: N = 12. Data are represented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001; Statistical significance was confirmed by two-way ANOVA followed by Holm–Sidak post hoc test for b–h. Source data are provided as a Source Data file.

Fig. 6. Enhancing and expanding the functionality of Class I ASOs to modulate the translation of various mRNAs.

a–d Enhancing the uORF-targeting ASO1 through various nucleotide chemistries. Locked nucleic acid (LNA) bases produced greater suppression of GATA4 protein levels compared to 2′-O-methylated, 2′-O-methoxy-ethyled (MOE), and phosphorothioate (PS) backbone (a). The combination of 2′-O-methyl and LNA (4 LNA at ASO 3′ end) is superior to 2′-O-methyl alone (c). RT-qPCR measurement of GATA4 mRNA in (a) or (c) with ACTB mRNA used as a normalizer (b, d). e–g Examining effects of class I mORF-enhancing ASOs (mORF-ASOs) on mORF translation. GATA4 mORF-targeting ASOs enhance its mORF protein levels (e). Like with uORF-specific ASOs, the combination of 2′-O-methyl and LNA is superior to 2′-O-methyl alone and does not change mRNA levels (f). 2′-O-methyl and LNA (4 LNA at ASO 3′ end) mORF ASOs (50 nM) increase the protein levels of mRNAs with cognate start codons, in the case of MEF2C (in AC16 cells) and NKX2-5, or near-cognate GUG start codons as with EIF4G2 (in HEK293T cells) (g). Data are represented as mean ± SD. *P < 0.05, **P < 0.01, ***P < 0.001; Statistical significance was confirmed by unpaired two-tailed Student t test for a–g (N = 3 biological replicates). Source data are provided as a Source Data file.