CELF1 Downregulation Promotes Cardiomyocyte Hypertrophy via Regulating Alternative Splicing of Tead1

TL;DR

This study shows that reducing CELF1 in heart muscle cells leads to cell enlargement and changes in how the Tead1 gene is spliced, which may contribute to heart disease.

Contribution

The novel finding is that CELF1 regulates cardiomyocyte hypertrophy through alternative splicing of Tead1, involving interactions with hnRNPC and m6A modifications.

Findings

CELF1 knockdown increases cardiomyocyte size and upregulates hypertrophic markers.

CELF1 knockdown shifts Tead1 splicing from the long isoform (Tead1-L) to the short isoform (Tead1-S).

CELF1 interacts with hnRNPC and influences global m6A abundance in HeLa cells.

Abstract

Background/Objectives: The RNA-binding protein CELF1 is crucial for cardiac development, but its role in cardiomyocyte hypertrophy is unclear. This study investigates the effects of acute CELF1 knockdown on alternative splicing and hypertrophic growth in cardiomyocytes. Methods: Neonatal rat cardiomyocytes (NRCMs) were transfected with two siRNAs targeting CELF1. Hypertrophy was assessed by cell size and expression of hypertrophic markers via qPCR and Western blot. RNA sequencing was performed in NRCMs to identify alternative splicing events. Tead1 function was tested by knockdown in NRCMs. Selected mechanistic assays were performed primarily in HeLa cells. Results: CELF1 knockdown in NRCMs increased cardiomyocyte size and upregulated hypertrophic markers, while its overexpression restored the phenotype. RNA-seq revealed that CELF1 knockdown alters the alternative splicing pattern. Specifically, the splicing of the transcription factor Tead1 shifted from the full-length long Tead1 isoform (Tead1-L) to the exon 4-skipped short isoform (Tead1-S). In HeLa cells, CELF1 interacted with hnRNPC, an m6A reader and splicing factor, and CELF1 perturbation correlated with changes in global m6A abundance. Conclusions: These findings suggest that CELF1 regulates hypertrophic phenotypes in cardiomyocytes and is associated with alternative splicing of Tead1.

Figures

Full text

1. Introduction

Cardiovascular disease is the leading cause of global mortality, posing a major challenge to public health systems [1,2]. Heart failure represents the common terminal stage in the progression of various cardiovascular disorders [3]. Pathological cardiac hypertrophy, which is a transitional maladaptive stage characterized by extensive reprogramming of gene expression and biological pathways, serves as a critical driver of this progression [4,5]. Consequently, elucidating and targeting the fundamental molecular mechanisms underlying cardiac hypertrophy offers a promising strategy for preventing or delaying the onset of heart failure [6,7].

The CUG-BP, Elav-like Family (CELF) consists of two subfamilies: CELF1–2 and CELF3–6 [8]. CELF1 is highly expressed in the heart and is crucial for cardiac development [9,10,11]. Structurally, it contains three RNA recognition motifs (RRMs), with a divergent domain intervening between RRM2 and RRM3 [12,13]. As an RNA-binding protein that recognizes U/G-rich motifs, CELF1 regulates key post-transcriptional processes, including pre-mRNA splicing [14,15,16], deadenylation [17], and mRNA decay [18,19,20]. However, the mechanisms governing CELF1-mediated splicing regulation remain unclear.

Heterogeneous nuclear ribonucleoproteins (hnRNPs) are canonical splicing factors and m^6^A readers, modulating mRNA alternative splicing through the m^6^A switch mechanism [21,22]. Furthermore, both alternative splicing and m^6^A RNA modification have been recognized as critical mechanisms underlying the pathogenesis of heart failure and various cardiomyopathies [23,24,25,26,27]. Despite its established role as a crucial post-transcriptional regulator, the exact mechanism of CELF1 in cardiac hypertrophy, especially its potential interplay with the m^6^A pathway, remains undefined.

In this study, we demonstrated that acute siRNA-mediated CELF1 knockdown induced cardiomyocyte hypertrophy in NRCMs. RNA sequencing revealed that CELF1 knockdown altered alternative splicing patterns, with Tead1 identified as a prominent target. Functional experiments showed that Tead1 knockdown partially attenuated the induction of hypertrophic markers. CELF1 was found to associate with hnRNPC and was correlated with changes in global m^6^A abundance in HeLa cells, suggesting a potential m^6^A-mediated regulatory role. Based on these observations, we propose a hypothetical pathway in which CELF1, in collaboration with m^6^A reader hnRNPC, regulates the alternative splicing of Tead1 through an m^6^A-associated mechanism, potentially contributing to the development of cardiomyocyte hypertrophy.

Previous studies on CELF proteins, largely conducted in the context of myotonic dystrophy and cardiac development, have primarily focused on antagonistic interplay with MBNL proteins to regulate specific splicing targets, such as cTnT and IR [9,10,28,29]. Here, we extend the functional scope of CELF1 to cardiomyocyte hypertrophy by demonstrating that acute loss of CELF1 is sufficient to trigger hypertrophic growth. We further identify Tead1 as a novel, heart-relevant splicing target that mediates the hypertrophic response. In addition, beyond the canonical CELF-MBNL axis, our data indicate that CELF1 associates with the m^6^A reader hnRNPC, and that CELF1 perturbation is correlated with changes in global m^6^A abundance.

2. Materials and Methods

2.1. Isolation and Culture, and Experimental Design of NRCMs

Primary NRCMs were isolated from 1–3-day-old neonatal Sprague–Dawley rats (purchased from Shanghai Bikai Keyi Biotechnology Co., Ltd., Shanghai, China) for in vitro studies of cellular hypertrophy [30]. Neonatal rats of mixed sex were used, as sex cannot be reliably determined at this developmental stage. Each independent biological replicate was defined as a complete isolation cycle starting from a unique cohort of 4–6 pups. The sample size of 15–20 pups per experiment was determined based on prior studies, which suggested this number would provide sufficient statistical power. In total, 250–280 pups were used for the study. Pups were euthanized by rapid decapitation, and hearts from 1–3-day-old neonatal Sprague–Dawley rats were aseptically excised, rinsed in cold PBS, and minced into approximately 1 mm^3^ fragments. The tissues were digested overnight at 4 °C in 0.25% trypsin, followed by repeated 15-min digestions at 37 °C in 0.1% type II collagenase containing 1% BSA. To obtain sufficient cell yield, tissue from all pups within an isolation batch was processed and the resulting cell suspensions were pooled, filtered, and centrifuged at 1000× g for 10 min. The pellet was resuspended in DMEM/M199 (3:1) supplemented with 10% fetal bovine serum (FBS). Cells were plated in 10 cm dishes and incubated for 1.5 h at 37 °C with 5% CO_2_ to allow attachment of non-cardiomyocytes. The supernatant containing cardiomyocytes was collected and transferred to new plates. After 24 h, the medium was replaced with DMEM/M199 (3:1) containing 1% FBS to remove non-adherent cells and maintain low-serum culture conditions.

For experimental treatments, culture wells were randomly assigned to different groups (e.g., control siRNA, siCELF1-1, or siCELF1-2). The experimenter was blinded to group allocation during all subsequent quantitative image and data analyses. All successfully isolated and adherent cardiomyocytes were included in the study without post hoc exclusion. Primary outcome measures were cardiomyocyte surface area (via cTnT immunofluorescence) and mRNA expression of hypertrophy markers (ANP, BNP, β-MHC). All data are presented as mean ± SEM from at least n = 3 independent biological replicates (isolations).

2.2. Cell Transfection

For knockdown experiments, NRCMs seeded in 12-well plates were transfected with CELF1-targeting siRNAs using Lipofectamine RNAiMAX. For each well, 2.5 μL Lipofectamine RNAiMAX was diluted in 50 μL Opti-MEM. Separately, 2.5 μL of a 10 μM siRNA stock was diluted in another 50 μL Opti-MEM. The two solutions were incubated separately at room temperature for 5 min, then combined, gently mixed, and further incubated at room temperature for 15 min before being added dropwise to the cells. The siRNA sequences were:

siCELF1-1: sense 5′-CACCCGUAAAGCUGCAUUATT-3′ and antisense 5′-UAAUGCAGCUUUACGGGUGTT-3′.

siCELF1-2: sense 5′-CCCAAUGGGAGGGUUAAAUTT-3′ and antisense 5′-AUUUAACCCUCCCAUUGGGTT-3′.

siTead1: sense 5′-CAGACUCGUACAACAAACATT-3′ and antisense 5′-UGUUUGUUGUACGAGUCUGTT-3′.

The culture medium was replaced 24 h post-transfection, and subsequent functional assays were conducted 72 h later. Knockdown efficiency was validated at both the protein (by Western blotting) and mRNA (by qRT-PCR) levels 72 h post-transfection.

For overexpression experiments, NRCMs seeded in 12-well plates were transfected with plasmids using Lipofectamine 3000 and P3000 reagent. For each well, the transfection complexes were prepared as follows: 2.5 μL Lipofectamine 3000 was diluted in 50 μL Opti-MEM; separately, 0.75 μg of plasmid DNA and 2.5 μL of P3000 reagent were diluted in another 50 μL Opti-MEM. After initial incubation at room temperature for 5 min, the two solutions were gently mixed and incubated for 15 min before being added dropwise to the cells. After 6 h, the transfection mixture was replaced with fresh pre-warmed culture medium. Related phenotypic or molecular analyses were conducted 72 h post-transfection. Transfection efficiency was validated at the protein level by Western blotting 72 h post-transfection.

To ensure that rescue experiments were performed under near-physiological conditions and to avoid artifacts from overexpression, the protein expression level of the siRNA-resistant CELF1 construct was quantified by Western blot and compared directly to endogenous CELF1 levels in untransfected control cells.

2.3. Immunofluorescence Staining

Cells were fixed with 4% paraformaldehyde for 15 min, permeabilized with 0.1% Triton X-100 in PBS (PBST) for 13 min, and blocked with 10% goat serum at room temperature (RT) for 1 h. After overnight incubation with primary antibodies (mouse anti-Cardiac Troponin T, Abcam, Cambridge, UK, Cat# ab8295, 1:500) at 4 °C, cells were washed with PBST and then incubated with Alexa Fluor-conjugated secondary antibodies (goat anti-mouse IgG (H&L), Abcam, Cambridge, UK, Cat# ab150113, 1:100) for 1 h at RT in the dark. Nuclei were stained with DAPI for 10 min and washed with PBS.

For cell size experiments, image acquisition and quantitative analysis were performed. Images were acquired using a fluorescence microscope (Leica, Wetzlar, Germany) with a 10× objective. For cell size analysis, cardiomyocyte surface area was quantified using ImageJ software (version 1.53). For each condition, 80–100 cardiomyocytes were measured per independent NRCM isolation, and three independent isolations were analyzed. To minimize bias, image acquisition and quantitative analysis were performed by an experimenter blinded to the experimental group allocation. For statistical analysis, the mean cell area from each independent isolation was treated as one biological replicate.

For quantification of CELF1 subcellular localization, GFP fluorescence intensity was measured using ImageJ. Briefly, nuclei were defined based on DAPI staining, and cytoplasmic regions were delineated by excluding the nuclear area within cTnT-positive cardiomyocytes. Mean GFP fluorescence intensities in nuclear and cytoplasmic compartments were measured for each cell, and nuclear-to-cytoplasmic (N/C) fluorescence intensity ratios were calculated. At least 20 cells from 2 independent experiments were quantified per condition.

2.4. Plasmid Construction

All plasmids were constructed using standard molecular cloning procedures: Total cDNA from NRCMs served as the template for PCR amplification of target gene fragments. Recipient vectors were linearized by restriction enzyme digestion and purified by gel extraction. Target fragments were assembled into the vector using a recombinase-based cloning system. The recombination products were transformed into competent Escherichia coli cells. Individual colonies were picked the next day and verified by Sanger sequencing to correct plasmid construction.

2.5. RNA Extraction and Quantitative Real-Time PCR (qRT-PCR)

Total RNA was extracted from cells using TRIzol reagent (Takara, Dalian, China, Cat# 9109). Briefly, 1 mL TRIzol was added to each well to lyse the cells completely. After adding 200 μL chloroform, samples were vortexed vigorously and incubated at RT for 15 min, then centrifuged at 12,000× g for 15 min at 4 °C. The upper aqueous phase was carefully transferred to a new tube, mixed with an equal volume of isopropyl alcohol, and incubated at RT for 15 min. The RNA pellet was collected by centrifugation, washed twice with 75% ethanol, air-dried, and dissolved in DEPC-treated water.

RNA concentration and purity were measured using a NanoDrop spectrophotometer. For each sample, 500 ng of total RNA was reverse-transcribed into cDNA using a PrimeScript RT reagent kit. qRT-PCR reactions were performed in 384-well plates and run on a QuantStudio 6 Real-Time PCR System. Gene expression quantification was performed using the comparative ΔΔCt method. The Ct values of target genes were normalized to the geometric mean of the stably expressed housekeeping gene Gapdh to calculate ΔCt. The ΔΔCt was then determined by comparing the ΔCt of experimental groups to that of the control group. Relative mRNA expression levels were calculated as 2^−ΔΔCt^. Primer sequences used for qRT–PCR analysis of hypertrophic markers and CELF1 are listed in Supplementary Table S1.

2.6. Western Blot

Total protein was extracted using RIPA buffer containing protease inhibitors. Following 30-min incubation on ice and centrifugation at 12,000× g for 15 min at 4 °C, supernatants were collected. Protein concentrations were measured with a BCA assay kit (Beyotime Biotechnology, Shanghai, China, Cat# P0009), and samples were denatured in Laemmli buffer with DTT at 100 °C for 5 min. Proteins were separated by NuPAGE Bis-Tris gels in MOPS running buffer and transferred to methanol-activated PVDF membranes at 100 V for 90 min in an ice bath. Membranes were blocked with 5% skim milk in TBST for 1 h at RT.

After blocking, membranes were incubated overnight at 4 °C with primary antibodies against CELF1 (Proteintech, Wuhan, China, Cat# 13002-1-AP, 1:1000), ANP (Santa Cruz Biotechnology, Dallas, TX, USA, Cat# sc-515701, 1:250), β-MHC (Proteintech, Cat# 22280-1-AP, 1:2000), RCAN1.4 (Sigma, St. Louis, MO, USA, Cat# D6694, 1:1000), TEAD1 (Santa Cruz Biotechnology, Cat# sc-393976, 1:250), hnRNPC (Invitrogen, Carlsbad, CA, USA, Cat# MA532270, 1:1000) and β-ACTIN (Santa Cruz Biotechnology, Cat#, sc-47778, 1:500). The TEAD1 antibody recognizes an internal epitope shared by both the full-length (Tead1-L) and exon 4–skipped (Tead1-S) isoforms; therefore, Western blotting reflects total TEAD1 protein abundance rather than isoform-specific expression. After TBST washes, membranes were incubated with HRP-conjugated secondary antibodies for 45 min at RT. Protein bands were visualized using a ChemiDoc^TM^ MP Imaging System (Bio-Rad, Hercules, CA, USA) and quantified with ImageJ software.

2.7. DNA Electrophoresis for Tead1 Splice Isoform Analysis

To resolve Tead1 alternative splice isoforms, RT-PCR products amplified using primers flanking exon 4 of the Tead1 transcript were separated by polyacrylamide gel electrophoresis. A 15% polyacrylamide gel was prepared according to the manufacturer’s instructions. After polymerization, the gel was placed in an electrophoresis chamber and immersed in 1X TBE buffer. RT-PCR products were mixed with loading buffer, loaded into the wells, and electrophoresed at 180 V for 30 min. Under these conditions, the full-length Tead1-L isoform and the exon 4–skipped Tead1-S isoform were resolved based on size differences. Following electrophoresis, the gel was carefully removed and stained in the dark for 10 min with SYBR Gold nucleic acid gel stain diluted 1:10,000 in 1×TBE buffer. The gel was then washed three times with 1×TBE buffer. SYBR Gold-stained DNA bands were visualized using the ChemiDoc^TM^ MP Imaging System. Primer sequences used for Tead1 alternative splicing RT–PCR are provided in Supplementary Table S3.

2.8. Protein Immunoprecipitation (IP)

Dynabeads^TM^ Protein G magnetic beads were pre-coated with anti-Flag or control IgG antibodies by rotating at 4 °C for 2 h. Total protein was extracted using lysis buffer containing protease inhibitors and incubated on ice for 30 min, centrifuged at 13,000× g for 15 min at 4 °C, and 10% supernatant was saved as the input.

The remaining lysate was incubated with antibody-conjugated beads at 4 °C with rotation for 3 h. Following immunoprecipitation, the bead complexes were washed four times. Both the input and IP samples were resuspended in 2× Laemmli loading buffer and denatured at 100 °C for 5 min. Denatured proteins were then analyzed by SDS-PAGE and Western blotting. Note: Lysates were not treated with RNase; therefore, the co-immunoprecipitation results reflect the association between CELF1 and hnRNPC under native conditions, which may include both RNA-mediated and direct protein–protein interactions.

2.9. RIP-qPCR

Dynabeads^TM^ Protein G magnetic beads were incubated with anti-Flag or control IgG antibodies on a rotator at 4 °C for 2 h. After washing, the antibody-bound beads were resuspended for subsequent use. Cells were lysed in 300 μL RIP lysis buffer (150 mM KCL, 25 mM Tris, PH 7.4, 5 mM EDTA, 0.5 mM DTT, 0.5% Triton-X100) containing fresh RNase and protease inhibitors and centrifuged at 13,000× g for 15 min at 4 °C. A 10% aliquot of the supernatant was saved as Input, while the remainder was incubated with the antibody-bead complexes at 4 °C for 3 h with rotation.

RNA was extracted from all samples using 500 μL TRIzol reagent and purified with the Zymo RNA Clean & Concentrator-5 kit (Zymo Research, Irvine, CA, USA) according to the manufacturer’s protocol. RNA concentration was measured using a NanoDrop spectrophotometer, and cDNA was synthesized by reverse transcription. Quantitative PCR was performed to assess the enrichment of CELF1-associated transcripts (Fhod3, R3hdm2, Rab30, Ubn1, Mbnl1, Myh14, Tead1). Primer sequences are listed in Supplementary Table S2.

For Tead1, the RIP-qPCR primers were designed with the forward primer located in exon 1 and the reverse primer located in exon 2, both of which are constitutive exons upstream of the alternatively spliced exon 4. This primer design detects total Tead1 transcripts shared by both Tead1-L and Tead1-S isoforms and therefore does not preferentially enrich a specific splice variant.

2.10. Generation of CELF1-Knockdown HeLa Cell Lines

HeLa cells were cultured in high-glucose DMEM supplemented with 10% FBS and 1% penicillin-streptomycin. To establish stable knockdown cell lines, HeLa cells were infected with three distinct lentiviruses expressing CELF1-specific shRNAs or an empty PLKO.1 control. Infections were carried out in medium containing polybrene at a final concentration of 8 μg/mL. Following infection, the virus-containing medium was replaced with fresh culture medium. To select for successfully transduced cells, puromycin was added to the culture medium at a final concentration of 2–3 μg/mL. Cells were maintained under puromycin selection until all cells in the non-infected control group were completely eliminated, indicating effective selection of stably transfected cells.

HeLa cells were used for selected mechanistic assays (co-immunoprecipitation and m^6^A-related analyses) because of their high transfection efficiency and suitability for biochemical interaction studies. These experiments were designed to examine general molecular interactions involving CELF1 and hnRNPC, rather than cardiomyocyte-specific functional outcomes, which were exclusively assessed in NRCMs.

2.11. RNA Sequencing and Bioinformatics Analysis

Total RNA was extracted from NRCMs isolated from three independent preparations and transfected with CELF1-targeting siRNAs. RNA quality and concentration were assessed using a NanoDrop spectrophotometer and agarose gel electrophoresis. For each condition, RNA samples from three independent NRCM isolations were subjected to library construction. Ribosomal RNA was removed from 1 μg of total RNA using the RiboMinus^TM^ Eukaryote Kit. Libraries were prepared with the NEBNext^®^ Ultra ^TM^ II Directional RNA Library Prep Kit, incorporating Illumina adapters, and sequenced on an Illumina HiSeq X Ten platform to generate 150 bp paired-end reads, yielding approximately 40–50 million reads per sample.

Raw sequencing reads were trimmed to remove adapters and low-quality bases using Trim Galore! (version 0.6.4_dev). Clean reads were aligned to the rat reference genome (Rnor_6.0) using HISAT2 (version 2.2.1) [31]. Mapping quality and alignment statistics were assessed, with mapping rates exceeding 90% for all samples.

For alternative splicing analysis, rMATS (v4.1.0) was used under the following parameters: read length 150, cstat 0.001, libType fr-firststrand [32]. Significant splicing events were defined as those with FDR <0.05. For gene expression analysis, reads were quantified using featureCounts, and differential expression was assessed with DESeq2 in R, considering genes with an adjusted p value (false discovery rate, FDR) <0.05 and |Fold Change| >1.5 as significant. Gene Ontology (GO) enrichment analysis was performed using Metascape.

All RNA-seq samples were processed in parallel using the same experimental and computational pipeline. To ensure robustness, potential batch effects arising from independent cell isolations were accounted for by including ‘Replicate’ as a covariate in the DESeq2 linear model.

2.12. Statistical Analysis

All data were analyzed using GraphPad Prism software (version 10.4.1) and presented as mean ± SEM. Comparisons between two groups were conducted using unpaired Student’s t-tests. For comparisons across three or more groups (e.g., control, siCELF1-1, siCELF1-2), one-way analysis of variance (ANOVA) was first applied to determine overall significance. If the ANOVA indicated a significant difference (p < 0.05), Tukey’s post hoc test was used for specific pairwise comparisons while controlling for the family-wise error rate. A p-value (or adjusted p-value for post hoc tests) of less than 0.05 was considered statistically significant.

3. Results

3.1. CELF1 Downregulation Promotes Cardiomyocyte Hypertrophy

To investigate the role of CELF1 in cardiomyocyte phenotypes, we first isolated neonatal rat cardiomyocytes (NRCMs) and screened two effective siRNAs for CELF1 knockdown, designated siCELF1-1 and siCELF1-2. Both siRNAs achieved over 80% knockdown efficiency as assessed by qRT-PCR (Figure 1B). For subsequent functional assays, the same siRNA transfection protocol was used consistently. in all experiments. Next, we transfected NRCMs with siRNA targeting CELF1 for 96 h and performed cytoskeletal staining with cTnT. We observed a significant increase in cell size upon CELF1 knockdown (Figure 1A–C). Additionally, we examined the expression of hypertrophy markers (ANP, BNP, β-MHC, and RCAN1.4) in NRCMs. CELF1 knockdown significantly increased the expression of these markers at both mRNA and protein levels (Figure 1D–F).

Further, to verify whether CELF1 overexpression can rescue the cardiomyocyte hypertrophy phenotype driven by its knockdown, we constructed a synonymous mutant CELF1 plasmid that was resistant to siCELF1-1. This mutant contained seven silent mutations from 67 to 73, the target site of siCELF1-1 (Figure 2A). Western blot analysis confirmed that the expression level of this siRNA-resistant CELF1 mutant was comparable to endogenous CELF1 in control cells, ensuring a near-physiological rescue condition (Figure 2B,C). Under this condition, the mutant CELF1 effectively reversed the upregulation of hypertrophy markers caused by acute siRNA-mediated CELF1 knockdown (Figure 2B,C). These results indicate that CELF1 knockdown significantly promotes cardiomyocyte hypertrophy. This cell-autonomous increase in cardiomyocyte size reflects a fundamental cellular process that contributes to pathological cardiac remodeling at the tissue level.

3.2. CELF1 Depletion Induces Alternative Splicing Changes

Based on these observations, we next hypothesized that CELF1 loss affects alternative splicing in NRCMs. RNA sequencing (RNA-seq) was then conducted on CELF1 knockdown NRCMs using siCELF1-1 and siCELF1-2. Subsequent bioinformatics analysis using rMATS software revealed 443 significant differential alternative splicing events in CELF1-knockdown cardiomyocytes compared to controls (FDR < 0.05) (Figure 3A).

To characterize these events, we categorized them into five major types: skipped exon (SE), retained intron (RI), mutually exclusive exons (MXE), alternative 3′ splice site (A3SS), and alternative 5′ splice site (A5SS) (Figure 3B,C). Among these, SE events accounted for the largest proportion of alternative splicing events (Figure 3D). Gene Ontology (GO) enrichment analysis showed that these genes were significantly enriched in biological processes related to supramolecular fiber organization, cellular component morphogenesis, actin filament-based movement, and microtubule cytoskeleton organization. These results indicate that CELF1-mediated alternative splicing regulations help maintain the cardiomyocyte structural and functional phenotype, and its disruption may contribute to cardiomyocyte hypertrophy.

3.3. CELF1 Knockdown Regulates the Alternative Splicing of Tead1

To identify robust CELF1-dependent alternative splicing targets, we compared RNA-seq data from multiple independent CELF1 knockdown conditions and focused on splicing events consistently altered across all siRNA treatments (Figure 4A). We identified 71 genes whose transcripts were concurrently subject to CELF1-dependent splicing changes across all siRNA groups (Figure 4A). These genes therefore represent a set of reproducible CELF1-associated splicing events. Based on their functional relevance, seven candidates were selected for further validation: Ubn1 (chromatin assembly), R3hdm2, Rab30, and Mbnl1 (RNA splicing), as well as Fhod3, Tead1, and Myh14 (cardiac pathophysiology). To determine whether CELF1 directly binds to its transcripts, we performed RNA immunoprecipitation-qPCR (RIP-qPCR) in NRCMs overexpressing Flag-tagged CELF1 (Figure 4B). The results confirmed that CELF1 was associated with five of the seven candidates (Ubn1, R3hdm2, Rab30, Fhod3, and Tead1). Notably, the Tead1 transcript exhibited the most significant enrichment following CELF1 overexpression (Figure 4C).

TEAD1, the predominant TEAD family member in the heart, is a key transcription factor of heart development that functions with nuclear co-activators YAP/TAZ [33]. TEAD1 exhibits high expression during embryonic and early postnatal stages but declines in adulthood, a pattern mirroring that of CELF1 [34]. We subsequently performed bioinformatics analysis of alternative splicing and identified a significant Tead1 exon skipping event. Further analysis showed that CELF1 knockdown markedly shifted the mRNA isoform balance from the full-length long Tead1 isoform Tead1-L to the shorter isoform Tead1-S, which lacks exon 4 (Figure 4D). To experimentally validate this splicing event, RT-PCR products spanning exon 4 were resolved by polyacrylamide gel electrophoresis, allowing separation of the Tead1-L and Tead1-S splice isoforms based on size differences. We confirmed that CELF1 depletion altered the relative abundance of Tead1-S and Tead1-L transcripts at the RNA level. The expected band sizes were approximately 210 bp for Tead1-L and 199bp for Tead1-S. Specifically, CELF1 knockdown increased the proportion of Tead1-S while decreasing that of Tead1-L (Figure 4E). Given the established role of TEAD1 in regulating transcriptional programs associated with cardiomyocyte growth and stress adaptation, this isoform switch is likely to influence downstream gene expression relevant to cardiac structural remodeling.

We next investigated if modulating TEAD1 expression could rescue the hypertrophic phenotype induced by CELF1 knockdown. To differentiate a downstream effector from an independent pathway, we employed a sequential knockdown: NRCMs were first transfected with si-CELF1 and, after 48 h, transfected with siTead1, followed by a further 48-h incubation. This design ensures that Tead1 manipulation occurs after CELF1 loss-of-function is established, thereby minimizing the possibility of Tead1 acting in a parallel, confounding pathway. Under this condition, TEAD1 knockdown effectively reversed the upregulation of hypertrophic markers induced by acute siRNA-mediated CELF1 knockdown (Figure 4F–H). These results indicate that CELF1 influences cardiomyocyte hypertrophy by regulating the alternative splicing of Tead1, thereby linking post-transcriptional regulation to structural remodeling of cardiomyocytes.

3.4. The Key Functional Domains of CELF1

To delineate the role of the RNA recognition motifs (RRMs) in CELF1 function, we constructed a series of CELF1 truncation plasmids (Figure 5A): GFP-CELF1^FL^ (full-length), GFP-CELF1^Δ1/2^ (lacking RRM1/2), GFP-CELF1^Δ3^ (lacking RRM3), and GFP-CELF1^Δ1/2/3^ (lacking all three RRMs).

We then investigated the contribution of each RRM to the subcellular localization of CELF1. Immunofluorescence staining revealed that the RRM3 domain is critical for the subcellular localization of CELF1 (Figure 5B). While the full-length CELF1 exhibited predominant nuclear localization, deletion of RRM3 resulted in a uniform distribution throughout both the nucleus and cytoplasm. To avoid bias from representative images, nuclear-to-cytoplasmic (N/C) GFP fluorescence intensity ratios were quantified across multiple cells under identical imaging and exposure conditions. Consistent with visual inspection, quantitative analysis confirmed a significant reduction in nuclear enrichment upon RRM3 deletion (Figure 5C). This finding suggests that RRM3 is essential for the proper nuclear enrichment of CELF1, which is critical for its role in regulating RNA alternative splicing.

We next transfected NRCMs with a series of CELF1 truncation mutants and performed RIP-qPCR to determine RRMs required for binding Tead1 mRNA. The results showed that deletion of any single RRM domain significantly reduced the enrichment of Tead1 mRNA bound to CELF1 (Figure 5C). This result suggests that the three RRMs function cooperatively to ensure stable and specific binding of CELF1 to the Tead1 transcript.

3.5. CELF1 Interacts with hnRNPC and Is Associated with Changes in Global mRNA m6A Levels

To investigate how CELF1 deficiency leads to the alternative splicing of Tead1, we performed mass spectrometry in CELF1-overexpressing HeLa cells to identify CELF1-interacting proteins. We identified several RNA-binding proteins that interact with CELF1, including HuR, Musashi, QKI, and hnRNPC (Figure 6A). These partners share a common feature that they recognize U-rich sequence motifs, which aligns with CELF1’s known function in binding U/G-rich elements.

Given the established role as an m^6^A-sensitive splicing regulator, we hypothesized that hnRNPC acts as a key cofactor mediating CELF1-dependent alternative splicing [35,36]. To validate the proteomic findings, Flag-tagged CELF1 was expressed in HeLa cells and analyzed by co-immunoprecipitation, revealing a specific interaction with hnRNPC (Figure 6B). We note that lysates were not treated with RNase, and therefore the interaction may be mediated by RNA and/or direct protein-protein binding. These data support a cooperative association between CELF1 and hnRNPC in splicing regulation, while the precise contribution of RNA-mediated versus direct binding remains to be further explored.

To functionally characterize this pathway, we generated CELF1-knockdown HeLa cells using pLKO.1-shRNA lentiviruses. Quantitative RT-PCR (qRT-PCR) analysis confirmed that all three shRNAs reduced CELF1 levels, with only shRNA-1 causing a statistically significant reduction and showing the highest efficiency (Figure 6C). To determine whether CELF1 influences m^6^A modification, we compared global m^6^A levels between control and CELF1-knockdown HeLa cells. The efficiency of CELF1 knockdown correlated with reduced global m^6^A levels, and the most potent shRNA (shRNA-1) elicited the strongest effect, lowering the m^6^A/A ratio from 0.22 to 0.16 (Figure 6D). Supporting this, we next employed an m^6^A-specific immunoprecipitation assay in CELF1-overexpressing HeLa cells. As a result, CELF1 overexpression successfully enriched m^6^A-modified RNA (Figure 6E). Collectively, these data demonstrate that CELF1 positively regulates the m^6^A modification of mRNA and that CELF1 correlates with global mRNA m^6^A abundance. While these findings do not establish a direct regulatory role for CELF1 in the m^6^A machinery, they suggest a potential link between CELF1, hnRNPC, and m^6^A-associated post-transcriptional regulation, which may contribute to splicing programs relevant to cardiomyocyte hypertrophy.

Based on our findings, we propose a hypothetical mechanistic model. Under physiological conditions, CELF1 is suggested to coordinate the alternative splicing of Tead1 pre-mRNA in concert with hnRNPC, thereby favoring the production of the full-length Tead1-L isoform. Upon CELF1 depletion, we hypothesize that the reduced m^6^A modification and weakened hnRNPC binding lead to increased exon skipping, shifting the isoform balance from Tead1-L to Tead1-S, and ultimately promoting the development of cardiomyocyte hypertrophy (Figure 7).

4. Discussion

Our study revealed that acute siRNA-mediated CELF1 knockdown can induce a hypertrophic response in NRCMs under baseline conditions. CELF1 knockdown was sufficient to promote cellular enlargement and upregulate multiple hypertrophic markers in NRCMs. In addition, this phenotype was effectively rescued by overexpressing an siRNA-resistant CELF1 mutant, confirming CELF1’s critical role in maintaining cardiomyocyte homeostasis.

Mechanistically, transcriptomic analysis confirmed that CELF1 depletion led to 443 significant alternative splicing events. Among the candidate targets, the transcription factor Tead1 emerged as a downstream target, confirmed by RIP-qPCR analysis as being highly enriched in CELF1. Notably, acute siRNA-mediated CELF1 knockdown promoted exon 4 skipping in Tead1 pre-mRNA, shifting the isoform balance from Tead1-L to the Tead1-S variant. Given TEAD1’s crucial role as a terminal effector of the Hippo/YAP signaling pathway in regulating cell growth and organ size [37,38,39,40], this splicing switch represents a maladaptive mechanism driving hypertrophic progression. We speculate that the Tead1-S isoform may exhibit dominant-negative transcriptional activity or control a distinct pro-hypertrophic gene network. In cardiomyocytes, Hippo/YAP/TEAD1 signaling is a key determinant of growth-related transcriptional programs. Thus, a CELF1-mediated shift in Tead1 isoforms may alter TEAD1-dependent gene expression, linking post-transcriptional regulation to cellular processes underlying pathological cardiac remodeling.

Importantly, while our data establish a robust association between CELF1 loss, Tead1 alternative splicing, and hypertrophic phenotypes, the causal contribution of individual Tead1 isoforms has not yet been directly dissected. In particular, whether the Tead1-S isoform is sufficient to drive cardiomyocyte hypertrophy, or whether hypertrophic signaling primarily arises from the loss of Tead1-L–mediated transcriptional activity, remains to be determined. Future studies employing isoform-specific gain- and loss-of-function approaches will be essential to resolve the relative contributions of Tead1-L and Tead1-S to cardiomyocyte growth control.

We also identified hnRNPC, the m^6^A reader and splicing factor as a novel CELF1-interacting partner. Furthermore, we observed that manipulation of CELF1 expression was associated with changes in global mRNA m^6^A abundance, as its knockdown reduced m^6^A levels, while overexpression enhanced m^6^A abundance. It therefore remains unclear whether CELF1 influences m^6^A deposition or removal through direct or indirect mechanisms, or whether changes in m^6^A levels reflect broader alterations in RNA metabolism following CELF1 perturbation. In particular, the potential involvement of canonical m^6^A writers and erasers, such as METTL3/METTL14/WTAP and FTO/ALKBH5, has not yet been examined. Future studies directly manipulating the m^6^A regulatory machinery will be required to determine whether m^6^A modulation is sufficient to recapitulate the Tead1 splicing changes observed upon CELF1 depletion.

These findings lead us to hypothesize that CELF1, in cooperation with hnRNPC, may regulate splicing through the m^6^A-associated mechanisms, but this remains hypothetical and requires further experimental confirmation. Within this framework, Tead1 alternative splicing represents a largely unexplored layer of post-transcriptional regulation with direct relevance to hypertrophic signaling.

Overall, our findings significantly advance the understanding of molecular mechanisms in cardiac hypertrophy. Our work suggests translational implications, highlighting the CELF1-Tead1 axis as a potential therapeutic target and Tead1-S/Tead1-L ratio as a novel disease progression biomarker. However, several limitations are worthy of further investigation. First, all phenotypes and functional analyses were performed in NRCMs, which differ from adult cardiomyocytes in their transcriptional and splicing landscapes. Consequently, validation in adult cardiomyocytes and in vivo models will be required to establish relevance to pathological cardiac remodeling in the adult heart. Second, selected mechanistic assays were performed in HeLa cells, which offer experimental robustness but do not fully reflect the cardiomyocyte-specific context. Future validation in cardiomyocytes and in vivo models will be important. Third, the mechanism by which CELF1, through the m^6^A-hnRNPC axis, regulates Tead1 alternative splicing remains incompletely understood. Fourth, the identities of the Tead1-L and Tead1-S isoforms, though clearly separated by size, await definitive confirmation by direct sequencing of the excised bands, which will be addressed in future studies. Finally, other CELF1 target genes may coordinate a pro-hypertrophic splicing program beyond Tead1.

In conclusion, our work elucidates a previously unknown pathway in which CELF1 regulates Tead1 alternative splicing to maintain cardiomyocyte homeostasis. These findings unveil the central role of splicing regulation in controlling cardiac function, thereby providing new mechanistic perspectives for heart failure prevention and treatment.