For the present study, siRNA duplexes were sourced from Shanghai GenePharma Co., Ltd. A non-targeting control siRNA, labeled siNegative, was employed, with the following sequences: Sense, 5'-UUCUCCGAACGUGUCACGUTT-3' and antisense, 5'-ACGUGACACGUUCGGAGAATT-3'. Additionally, an siRNA designed specifically to target the PUM2 gene was employed, labeled siPUM2, with the following sequences: Sense, 5'-AGCAUUAGAAUCUAUUUCUUCUGAT-3' and antisense, 5-AUCAGAAGAAAUAGAUUCUAAUGCUUU-3.

H9C2 rat cardiomyocyte cells (cat. no. CL-0089; Procell Life Science & Technology Co., Ltd.) were cultured in a controlled environment at 37˚C with a humidified atmosphere containing 5% CO₂. The culture medium utilized was DMEM (cat. no. PM150210; Gibco; Thermo Fisher Scientific, Inc.), supplemented with 10% FBS (cat. no. 10091148; Gibco; Thermo Fisher Scientific, Inc.) and antibiotics including 100 µg/ml streptomycin and 100 U/ml penicillin (cat. no. SV30010; HyClone^™; Cytiva). Cells in 12-well plates were transfected with 160 pmol siRNA using Lipofectamine^™ RNAiMAX (cat. no. 13778150; Invitrogen; Thermo Fisher Scientific, Inc.) according to the manufacturer's protocol. Following a 48-h incubation at room temperature, the cells were harvested and analyzed by reverse transcription-quantitative PCR (RT-qPCR) and western blotting.

Total RNA was extracted from H9C2 cells by the TRIzol^™ (Thermo Fisher Scientific, Inc.). The RNA was further purified with two phenol-chloroform treatments and then treated with RQ1 DNase (Promega Corporation) to remove DNA. The quality and quantity of the purified RNA were redetermined by measuring the absorbance at 260/280 nm (A260/A280) using NanoPhotometer N50 (Implen NanoPhotometers). The integrity of RNA was further verified by 1.0% agarose gel electrophoresis. cDNA was synthesized utilizing a reverse transcription kit (cat. no. R323-01; Vazyme Biotech Co., Ltd.). The procedure entailed an initial incubation at 42˚C for 5 min, followed by 15 min at 37˚C and a final step at 85˚C for 5 sec, all executed using the T100 thermocycler (Bio-Rad Laboratories, Inc.). Subsequent to cDNA synthesis, qPCR was conducted using the ABI QuantStudio^™ 5 (conventional) system (Applied Biosystems; Thermo Fisher Scientific, Inc.). The amplification protocol included a denaturation step at 95˚C for 10 min, and 40 cycles comprising denaturation at 95˚C for 15 sec and a combined annealing and extension step at 60˚C for 1 min. Each sample was analyzed in triplicate to ensure data reliability. To quantify the expression levels of each transcript, normalization was performed using GAPDH, and the 2^-ΔΔCq method was used for data analysis (25). Detailed sequences of primers used for qPCR analysis are listed in Table SI.

H9C2 cells were lysed using ice-cold RIPA lysis buffer (cat. no. PR20001; Proteintech Group, Inc.) supplemented with protease inhibitors (cat. no. 4693116001; Sigma-Aldrich; Merck KGaA). The lysis process was carried out on ice for 30 min to disrupt the cells while simultaneously inhibiting any protease activities that might lead to protein degradation. Following lysis, cell samples underwent a heating step, during which they were boiled for 10 min in conjunction with a protein loading buffer (cat. no. P1040; Beijing Solarbio Science & Technology Co., Ltd.). This heating step was key for denaturing the proteins, thereby facilitating their separation during subsequent analysis. Prepared samples (25 µg protein/lane) were then carefully loaded onto a 10% SDS-PAGE gel.Electrophoresis was carried out to separate the proteins based on their molecular weights, allowing for effective analysis of the protein profiles. Once electrophoresis was completed, the separated proteins were transferred onto 0.45-µm PVDF membranes (cat. no. ISEQ00010; MilliporeSigma) to enable further examination. Following successful protein transfer, the PVDF membrane was blocked with 5% skimmed milk to reduce non-specific binding. This involved incubating the membranes at room temperature for 1 h, which helped to minimize background noise during the detection phase. Subsequently, the membranes were incubated overnight at 4˚C with a specific primary GAPDH antibody (1:1,000; cat. no. A19056; ABclonal Biotech Co., Ltd.) that targeted PUM2 (1:1,000; cat. no. Ab92390; Abcam) and an actin antibody (1:1,000; cat. no. 20536-1-AP; Proteintech Group, Inc.). This overnight incubation allowed the primary antibodies time to bind to their respective protein targets. Following this incubation period, membranes were treated with an HRP-conjugated secondary antibody, either anti-rabbit (1:10,000; cat. no. SA00001-2; Proteintech Group, Inc.) or anti-mouse (1:10,000; cat. no. AS003; ABclonal Biotech Co., Ltd.), for 45 min at room temperature. This step was essential for enhancing the detection of the bound primary antibodies. Finally, visualization of the proteins on the membranes was achieved using an enhanced ECL reagent (BeyoECL Moon; cat. no. P0018FM; Beyotime Biotechnology), which permitted chemiluminescent detection of the proteins of interest.

In the present study, a cell viability assay was conducted utilizing a Cell Counting Kit-8 (CCK-8; Shanghai Yeasen Biotechnology Co., Ltd.). The procedure involved plating H9C2 cells at a density of 10,000 cells/well in 96-well culture plates in a controlled environment (37˚C and 5% CO₂) for cell viability. Cells were transfected with either a non-targeting control siRNA [negative control (NC) group] or dicer-substrate small interfering RNA (DSI; experimental group), while vials lacking cells served as blank controls to ensure measurement accuracy. Cells were incubated at 37˚C in a 5% CO₂ atmosphere for designated time periods: 48 and 72 h. Following incubation, 10 µl CCK-8 solution was added to each well for further incubation for 3 h at 37˚C. This step allowed for the assessment of cell viability. After incubation, the optical density (OD) was measured using the ELx800 microplate reader (BioTek; Agilent Technologies, Inc.) at a wavelength of 450 nm. To quantify the viability rate of the cells, the following formaul was applied: viability rate=(experimental OD value-blank OD value)/(control OD value-blank OD value) x100. This formula enabled a clear assessment of cell viability in response to varying treatments.

For the assessment of apoptosis, an annexin V-allophycocyanin (APC)/7-aminoactinomycin D apoptosis detection kit (cat. no. 40304ES60; Shanghai Yeasen Biotechnology Co., Ltd.) was employed according to the manufacturer's instructions. H9C2 cells were cultivated in 6-well plates for 24 h at 37˚C and subsequently transfected with siRNAs for 48 h as aforementioned. The transfected and control siRNA cell populations were combined with 5 µl annexin V-APC and incubated in the dark at room temperature for 5 min, followed by the addition of 5 µl 7-amino-actinomycin for incubation for 5 min at room temperature. Samples were then analyzed using the BD FACSCanto^™ Clinical Flow Cytometry system (BD Biosciences) to evaluate the levels of apoptosis, and the data were analyzed with FlowJo^™ software (BD Biosciences).

Total RNA was isolated from H9C2 cells using TRIzol^® reagent (cat. no. 15596026; Invitrogen; Thermo Fisher Scientific, Inc.), following the protocol established by Chomczynski and Sacchi (26). Following RNA extraction, DNA digestion (37˚C for 10 min) was carried out using RQ1 RNase-Free DNase (cat. no. M6101; Promega Corporation). RNA quality was assessed by measuring the A260/A280 ratio with a Nanodrop^™ One^C spectrophotometer (Thermo Fisher Scientific, Inc.). Integrity was verified through 1.5% agarose gel electrophoresis. The concentration of selected RNA samples was subsequently quantified using the Qubit^™ 3.0 fluorometer (Thermo Fisher Scientific, Inc.) in combination with the Qubit^™ RNA broad range assay kit (cat. no. Q10210; Thermo Fisher Scientific, Inc.). A total of 2 µg RNA was then used for the preparation of stranded RNA-sequencing (RNA-seq) libraries, utilizing the KC^™ Stranded mRNA Library Prep Kit for Illumina (cat. no. DR08402; Wuhan Kangce Technology) in accordance with the manufacturer's protocol. PCR products with lengths between 200 and 500 base pairs were enriched, quantified and ultimately sequenced by Novogene (Tianjin Sequencing Center and CAP Certified Clinical Lab, Tianjin, China) using the NovaSeq 6000 sequencer (Illumina, Inc.) and 150-bp paired-end sequencing was performed.

Initially, raw sequencing reads that contained >2 ambiguous bases, denoted as ‘N’, were eliminated from the dataset to ensure data quality. Following this step, adaptors and low-quality base calls were systematically trimmed from the remaining raw sequencing reads. This processing was carried out using the FASTX-Toolkit (version 0.0.13; https://github.com/agordon/fastx_toolkit), which is specifically designed for efficient read preprocessing. Furthermore, any sequencing reads that were <16 nucleotides in length were discarded to maintain the integrity of the dataset. The clean reads that passed these quality control measures were then aligned to the mRatBN7.2 genome using the HISAT2 tool (v2.2.1; http://daehwankimlab.github.io/hisat2/), allowing for the accommodation of ≥4 mismatches during the alignment process. This alignment procedure ensured that the uniquely mapped reads could be accurately utilized for counting gene read numbers and for calculating the fragments per kilobase of transcript per million fragments mapped (FPKM). During analysis of gene expression, the ‘DESeq2’ package (v 1.38.3; http://www.bioconductor.org/packages/release/bioc/html/DESeq2.html) from R Bioconductor (v 3.22; https://www.bioconductor.org/) was employed to identify DEGs using the thresholds P<0.01 and fold change >1.5 or <0.67. These parameters were essential for pinpointing variations in gene expression that could have biological relevance.

For the analysis of alternative splicing events (ASEs) and specifically regulated ASEs (RASEs) among the samples, the ABLas pipeline (https://github.com/ablifedev/ablas) was employed. ABLas identifies 10 distinct forms of alternative splicing based on the analysis of splice junction reads. These forms include exon skipping (ES), intron retention (IntronR), alternative 5' splice sites (A5SS), alternative 3' splice sites (A3SS) and mutually exclusive exons (MXE), as well as variations in the 5' and 3' untranslated regions (UTRs), specifically mutually exclusive 5' UTRs and mutually exclusive 3' UTRs, in addition to cassette exons, and combinations of A3SS and ES, and A5SS and ES.

To evaluate the influence of PUM2 on the ASEs, statistical analysis was carried out using unpaired Student's t-test to determine the significance of changes in the ratio of ASEs across the samples. Events that demonstrated significant alterations, with a P-value corresponding to a threshold that aligned with a false discovery rate cut-off of 5%, were classified as PUM2-regulated ASEs. This approach ensured that only those events with statistical significance were recognized to be potentially influenced by PUM2.

For RT-qPCR verification with GAPDH as the control gene to measure relative target gene expression levels, total RNA was isolated from H9C2 cells using TRIzol reagent (cat. no. 15596026; Invitrogen; Thermo Fisher Scientific, Inc.). cDNA was synthesized with a reverse transcription kit (cat. no. R323-01; Vazyme Biotech Co., Ltd.) on a mycycler (cat. no. T100; Bio-Rad Laboratories, Inc.) at 42˚C for 5 min, 37˚C for 15 min and 85˚C for 5 sec. qPCR was performed on the ABI QuantStudio 5 with Hieff^® qPCR SYBR Green Master Mix, starting with denaturing at 95˚C for 10 min, followed by 40 cycles of 95˚C for 15 sec and 60˚C for 1 min, with three technical replicates per sample. Transcript levels were normalized to GAPDH using the 2^-ΔΔCq method (25). The primers for pre-mRNA splicing detection, detailed in Table SI, were designed to target splice junctions between constitutive and alternative exons for precise isoform amplification.

To organize the functional classifications of DEGs, the KOBAS 2.0 server (https://github.com/xmao/kobas) (19) was utilized to identify Gene Ontology (GO) terms as well as Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways. The enrichment of each term was determined using the hyper-geometric test along with the Benjamini-Hochberg false discovery rate-controlling method.

All plots were conducted in R (version 4.2.3; https://cran.r-project.org/bin/windows/base/old/4.2.3/) as implemented in RStudio, in addition to pattern diagrams and stacked bar charts. Pattern diagrams and stacked bar charts were generated using GraphPad Prism (version 8.0; Dotmatics). Data are presented as the mean ± standard error of the mean (SEM), with three independent biological replicates. The statistical difference between two groups for the RT-qPCR, cell viability and apoptosis assays was calculated using an unpaired Student's t-test, where P<0.05 was considered to indicate a statistically significant difference. Other statistical methods used for the bioinformatics analyses have been described in the corresponding sections; for example, DEGs were calculated using DESeq2, and ASEs were analyzed using unpaired t-tests.

To explore the role of PUM2 in MF, H9C2 cell lines with PUM2 knockdown were constructed using siRNA transfection. Both RT-qPCR and western blotting confirmed successful knockdown of PUM2 in H9C2 cells (Fig. 1A and B). These efficiency validation experiments demonstrated that mRNA expression was significantly reduced (by ~50%) but not eliminated, whereas protein expression was knocked down; however, this was not statistically assessed. The rat PUM2 gene encodes five protein-coding transcripts, and the DSI sequence used may specifically target only a subset of these transcripts (27), while the RT-qPCR assay measured total PUM2 expression at the gene level. This selective targeting may explain why a larger reduction was observed at the protein level compared with the mRNA level. Consequently, expression from non-targeted transcripts masks the extent of knockdown in the PCR results, leading to the observed discrepancy between the RT-qPCR and western blot analyses. The effect of PUM2 knockdown on viability and apoptosis of H9C2 cells was detected using a CCK-8 assay and flow cytometry. The CCK-8 assay demonstrated that the viability of H9C2 cells was significantly inhibited at both 48 and 72 h after transfection following PUM2 knockdown (Fig. 1C). Flow cytometry revealed that PUM2 knockdown did not significantly affect apoptosis at 48 h post-transfection (Fig. S1). The results indicated that PUM2 is key for the regulation of viability in H9C2 cells.

To explore the regulatory role of PUM2 in gene expression, RNA-seq was conducted to assess and compare the overall gene expression profiles between samples subjected to siPUM2 transfection and those in NC group. Utilizing the principal component method, the FPKM values of all DEGs were analyzed, with the first principal component (dimension 1) successfully distinguishing the DSI group from the control group (Fig. 2A). The volcano plot illustrates the DEGs identified between DSI and NC samples (Fig. 2B). A total of 147 DEGs influenced by PUM2 were identified, encompassing 95 upregulated and 52 downregulated genes. Additionally, analysis of the heat map depicting DEG expression patterns in the RNA-seq samples (Fig. 2C) revealed a high level of consistency among the three biological replicates regarding siPUM2-mediated transcription. Notably, the number of upregulated genes markedly exceeded that of downregulated genes, implying that PUM2 may exert a repressive effect on gene expression.

To evaluate the possible biological functions of these DEGs, GO and KEGG enrichment analyses were conducted. GO enrichment analysis indicated that the genes exhibiting increased expression were predominantly associated with two pathways: ‘Positive regulation of GTPase activity’ and ‘positive regulation of transcription from RNA polymerase II promoter’ (Fig. S2). KEGG analysis provided comprehensive insights into the pathways associated with the upregulated genes, indicating their notable involvement in a multitude of key biological processes. The genes were heavily represented in pathways related to various forms of cardiomyopathy, such as ‘hypertrophic cardiomyopathy’, ‘dilated cardiomyopathy’ and ‘arrhythmogenic right ventricular cardiomyopathy’. Additionally, analysis highlighted the inclusion of key signaling pathways, including the ‘Rap1 signaling pathway’ and the ‘PI3K-Akt signaling pathway’, both of which serve vital roles in cellular responses. Furthermore, the involvement of these genes in ‘pathways in cancer’, ‘focal adhesion’, the ‘thyroid hormone signaling pathway’, ‘vascular smooth muscle contraction’ and the ‘Ras signaling pathway’ underscores their relevance in the complex interplay of genetic and biochemical signals (Fig. 2D). Conversely, the examination of downregulated genes using GO enrichment analysis revealed a primary connection to ‘positive regulation of cell proliferation’ and ‘positive regulation of transcription from RNA polymerase II promoter’ (Fig. S2). KEGG analysis identified notable associations between the downregulated genes and a diverse array of biological pathways, including ‘focal adhesion’ and ‘ECM-receptor interaction’, both of which are key for cellular communication and structural integrity (28,29). Additionally, the ‘PI3K-Akt signaling pathway’ was revealed to serve a role in the regulation of these genes, along with pathways associated with ‘arrhythmogenic right ventricular cardiomyopathy’ and the ‘regulation of the actin cytoskeleton’. Other notable associations included pathways involved in ‘colorectal cancer’, ‘pantothenate and CoA biosynthesis’, ‘Parkinson disease’ and ‘human papillomavirus infection’ (Fig. 2E). Specific pathways such as ‘hypertrophic cardiomyopathy’, ‘dilated cardiomyopathy’, ‘arrhythmogenic right ventricular cardiomyopathy’, ‘vascular smooth muscle contraction’, ‘focal adhesion’ (Fig. 2D), ‘ECM-receptor interaction’ and ‘regulation of actin cytoskeleton’ (Fig. 2E) are particularly noteworthy as they relate to molecular function. RNA-seq data for a representative set of five upregulated genes, including adrenomedullin (Adm), NDRG family member 4 (Ndrg4), phospholipid phosphatase 3 (Plpp3), integrin subunit α 8 (Itga8) and regulator of G protein signaling 2 (Rgs2), and three downregulated genes, including cadherin 11 (Cdh11), integrin subunit α 11 (Itga11) and transforming growth factor β-induced (Tgfbi), can be found in Fig. 2F.

Pathway analysis revealed that several significantly dysregulated genes are involved in key molecular functions. Notably, the elevated genes Itga8 and Adm, along with the downregulated gene Itga11 (identified in Tables I and II), were central components of the enriched pathways related to cell adhesion and cardiovascular regulation. To confirm the findings of RNA-seq, RT-qPCR was conducted to assess the relative expression patterns of the genes Adm, Itga8, Rgs2, Plpp3, Ndrg4, Itga11, Tgfbi and Cdh11 in both the control and experimental groups. Results from the RT-qPCR validation aligned with those obtained from the RNA-seq analysis, affirming the reliability of the sequencing data (Fig. 2F). RT-qPCR amplification and melting curves for all targets confirmed assay specificity and reproducibility, validating the RNA-seq findings (data not shown).

To explore how PUM2 knockdown influences the regulation of alternative splicing, RNA-seq data were assessed to identify PUM2-regulated RASEs in the H9C2 cell line. The aforementioned ABLas pipeline was employed to characterize and quantify ASEs and RASEs across the different samples. The types and quantities of all notable RASEs between the NC and DSI samples are illustrated in Fig. 3A. The primary RASEs detected included IntronR (183 events), A3SS (129 events) and A5SS (100 events; Table SII). These findings indicate that PUM2 regulates ASEs globally in the H9C2 cell line. The heat map of splicing ratios for RASEs demonstrated high reproducibility across biological triplicates in siPUM2-treated cells (Fig. 3B). GO analysis of PUM2-regulated alternative splicing genes (PUM2-RAS) was carried out to explore the potential biological functions of PUM2-RAS. Analysis revealed that the enriched pathways included ‘brain development’, ‘peptidyl-tyrosine dephosphorylation’, ‘axonogenesis’, ‘neural tube closure’, ‘mRNA processing’, ‘protein stabilization’, ‘positive regulation of GTPase activity’, ‘negative regulation of cell proliferation’, ‘protein phosphorylation’ and ‘cell proliferation’ (Fig. 3C). Splicing ratios for several ASEs were significantly altered upon PUM2 knockdown. Representative examples are presented in Figs. 3D and S3, including tropomyosin 1 (Tpm1; MXE), actinin a 1 (Actn1; ES) and PPFIA binding protein 1 (Ppfibp1; ES). The reads distribution diagram showed the distribution of splicing junctions for Tpm1 and Actn1, with significant changes in the splicing ratio (Fig. 3D). In addition, the distribution of Ppfibp1 is also represented as a reads distribution diagram, with the splicing ratio of Ppfibp1 presented as a bar plot (Fig. S3). RT-qPCR validation of ASEs in the Ppfibp1, Actn1 and Tpm1 genes aligned with the RNA-seq results (Figs. 3D and S3). RT-qPCR amplification and melting curves for all targets confirmed assay specificity and reproducibility, therefore validating the RNA-seq findings (data not shown).