Comprehensive 3′UTR Mapping Uncovers Widespread Alternative Polyadenylation Remodeling in Human Right Ventricular Failure

OverviewThis dataset accompanies the manuscript “Comprehensive 3′UTR Mapping Reveals Widespread Alternative Polyadenylation Remodeling in Human Right Ventricular Failure” (submitted to Circulation Research). The study uses Poly(A)-ClickSeq (PAC-seq) to define alternative polyadenylation (APA) changes in failing human right ventricular (RV) tissue and integrates these findings with external transcriptomic datasets to establish disease-stage specificity and mechanistic relevance. Human SamplesPAC-seq was performed on snap-frozen RV tissue obtained from four WHO Group I pulmonary hypertension patients undergoing heart–lung transplantation and four non-failing donor hearts (IRB-approved). All patients exhibited severe RV dysfunction with pulmonary artery systolic pressures ranging from 70–140 mmHg. Data Processing and AnalysisPAC-seq libraries were analyzed using PolyA-miner2 to identify gene-level APA remodeling. APA directionality was quantified using the PolyA index (log₂ proximal/distal PAS usage), where negative values indicate 3′UTR shortening and positive values indicate lengthening. Statistical significance was defined as an adjusted gene-level P value (AdjG-Pval) ≤ 0.05. Contents of This RecordThis Zenodo record contains Tables 2–8 referenced in the manuscript, including: Table S2. PAC-seq identification of significantly shortened and lengthened transcripts by PolyA-miner Poly(A)-ClickSeq (PAC-seq) data from failing human right ventricular tissue were analyzed using PolyA-miner to identify genes undergoing significant alternative polyadenylation (APA) remodeling. Genes were classified as shortened or lengthened based on the direction of the PolyA index (log₂ proximal/distal PAS usage). Statistical significance was defined as adjusted gene-level P value (AdjG-Pval) ≤ 0.05, with negative PolyA index values indicating 3′UTR shortening and positive values indicating 3′UTR lengthening. Table S3. Pathway enrichment analysis of APA-shortened transcripts in failing human RV Pathway enrichment analysis of genes exhibiting significant 3′UTR shortening identified by global PAC-seq profiling of failing human right ventricles (RV). A total of 509 significantly shortened transcripts were analyzed using Enrichr against the MSigDB Hallmark 2020 gene set library. Table S4. Left ventricular (LV) and right ventricular (RV) APA gene lists used for overlap analysis Gene lists used to compute overlap between left ventricular (LV) and right ventricular (RV) alternative polyadenylation (APA) maps shown in Figure C.LV shortened (LV_S) = 671 genes; LV lengthened (LV_L) = 642 genes; RV shortened (RV_S) = 509 genes; RV lengthened (RV_L) = 140 genes.LV APA gene lists were obtained from the published left-ventricular failure dataset by Creemers et al. (Circulation Research, 2016), using gene-level APA calls reported in the original study. RV APA gene lists were derived from the present human RV PAC-seq analysis using PolyA-miner with an adjusted gene-level significance threshold of P ≤ 0.05. APA direction was defined based on the PolyA index, with negative values indicating 3′UTR shortening and positive values indicating 3′UTR lengthening. Each column represents an independent gene list; rows do not imply gene-to-gene correspondence across LV and RV datasets. Table S5. Integration of RV PAC-seq APA remodeling with differential gene expression in an external right ventricular dataset (GSE198618) Genes showing overlap between RV PAC-seq–derived alternative polyadenylation (APA) changes and differentially expressed genes (DEGs) identified in the independent GSE198618 right-ventricular transcriptomic dataset (Control, Compensated RV hypertrophy, and Decompensated RV failure). Genes are grouped by comparison and APA–DEG directionality. APA shortening corresponds to negative PolyA index values, and APA lengthening corresponds to positive PolyA index values. Table S6. Integrated pathway enrichment of APA-shortened, upregulated genes across RV disease progression Pathway enrichment analysis of genes exhibiting concordant 3′UTR shortening and transcriptional upregulation across right ventricular (RV) disease stages using the MSigDB Hallmark 2020 library. PAC-seq–derived APA-shortened genes were integrated with differential expression results from the GSE198618 dataset. Pathways are shown for Control vs Decompensated and Compensated vs Decompensated comparisons. Odds ratios and combined scores were used for visualization in Figure D–E. Control vs Compensated comparisons showed no significant enrichment. Table S7. CPSF6 knockdown–induced alternative polyadenylation remodeling in human cardiac fibroblasts Primary human cardiac fibroblasts were transfected with control or CPSF6 siRNA and subjected to Poly(A)-ClickSeq (PAC-seq). Alternative polyadenylation (APA) changes were quantified using PolyA-miner at the gene level. CPSF6 knockdown resulted in widespread 3′UTR remodeling, dominated by shortening (PolyAIndex ≤ −0.2) with a smaller subset of lengthened transcripts (PolyAIndex ≥ +0.2), defined at an adjusted gene-level P value ≤ 0.05. The PolyA index reflects the log₂ ratio of proximal versus distal polyadenylation site usage. Table S8. Pathway enrichment of APA-shortened transcripts following CPSF6 knockdown in human cardiac fibroblasts Pathway enrichment analysis of genes exhibiting significant 3′UTR shortening following CPSF6 silencing in primary human cardiac fibroblasts. PAC-seq–derived APA-shortened genes (negative PolyA index; adjusted gene-level P ≤ 0.05) were analyzed using the Elsevier Pathway Collection. The table reports adjusted P values, odds ratios, combined scores, and the full list of contributing genes for each enriched pathway, highlighting activation of TGF-β signaling, epithelial–mesenchymal transition, and pro-fibrotic pathways. Reviewer-requested data accessibility These tables provide the full gene-level and pathway-level data underlying Figures B, C, D–E, and K and are deposited to enable reuse, independent mining, and citation by the cardiovascular research community, as requested during peer review. SUPPLEMENTAL MATERIAL Human Right Ventricular Samples Right ventricular (RV) tissue was obtained from four female patients with advanced pulmonary hypertension (mean age, 45 years) who underwent combined heart–lung transplantation at Houston Methodist Hospital. All patients exhibited severe RV dysfunction with markedly elevated pulmonary artery systolic pressures (PASP, 70–140 mmHg). Control RV tissue was obtained from four nonfailing donor hearts from age-unmatched individuals without known cardiac disease. Fresh tissue was obtained directly from the surgeon and immediately processed. Samples were dissected and allocated for (i) snap-freezing for RNA and protein isolation, (ii) paraffin embedding for pathological and histological analyses, or (iii) isolation of primary cardiac fibroblasts. Tissue designated for molecular analyses was flash-frozen in liquid nitrogen and stored at −80 °C until use. Tissue designated for histological analyses was fixed in 2% paraformaldehyde, dehydrated through graded alcohols, cleared in xylene, and embedded in paraffin using standard protocols. Paraffin blocks were sectioned at 5 µm thickness, mounted on glass slides, and rehydrated prior to histological and immunohistochemical staining. All human tissue collection and experimental procedures were performed in accordance with protocols approved by the Houston Methodist Hospital Institutional Review Board. Poly(A)-Click Sequencing Library Construction Total RNA was extracted using the RNeasy Kit (Qiagen). For Poly(A)-ClickSeq (PAC-Seq), 1.3µg total RNA was directly mixed with 1µl 100µM 3’ Illumina_4N_21T_VN primer (5'-GTGACTGGAGTTCAGACGTGTGCTCTTCCGATCTNNNNTTTTTTTTTTTTTTTTTTTTTVN-3'), 2µl 5mM AzNTP (Baseclick, #BCT-25~ -28)/dNTP mixture in a 1:5 ratio, and H2O to have 13 µl total in volume. The 13 µl mixture was heated at 65°C/5min, and snap-cooled on ice for 3min, and then subjected to reverse transcription. The RT was carried out using SuperScript III (ThermoFisher Scientific, 18080093). In brief, 7µl master mix containing 4µl 5X Superscript First Strand Buffer, 1µl 0.1M DTT, 1µl RNaseOUT (ThermoFisher Scientific, #10777019), and 1µl Superscript III were mixed with the RNA sample. Heating was performed as follows: 25°C/10min, 50°C/40min, and 75°C/15min sequentially in thermocycler followed by RNase H (ThermoFisher Scientific, AM2293) treatment using 1U per reaction for 37°C/30min and then 80°C/10min. The cDNA was then purified using Sera-Mag Speedbeads (Cytiva, 65152105050250). Speedbeads working solution was made by washing 1ml beads slurry twice in 1XTE buffer and then resuspending in 50ml of 1XTE buffer containing 9g PEG-8000, 1M NaCl, and 0.05% Tween-20. Speedbeads, 1.8x reaction volume, were mixed with cDNA followed by a 5min incubation in room temperature. The beads were then pelleted using a magnetic bead collector and the supernatant was discarded. Two washes with 80% of ethanol were performed while the beads were pelleted on the magnet. The beads were then dried, and the cDNA was eluted by resuspending the beads in 22µl 50mM HEPES pH 7.2 for 2min at room temperature. Then, 20µl of the cDNA was mixed with 11µl Click Mix (2.5M NaCl & 30% EtOH in H20) and 4µl of 5µM UMI-click-adapter (5'-Hexynyl-NNNNNNNNNNNNAGATCGGAAGAGCGTCGTGTAGGGAAAGAGTGT-3', HPLC purified). To activate the reaction, 4µl of 50mM vitamin C and 1µl of Click Catalyst (4mM CuSO4 in H20 & 20mM THPTA (Baseclick, BCMI-006) in H20) were pre-mixed before being added into the cDNA solution for 60min incubation at room temperature in the dark. Clicked cDNA was then purified in the same manner as post-RT purification except eluted in 22µl of 10mM Tris pH7.4. Then, half of eluted cDNA was subjected to PCR amplification with indexing primers i5 (5'-AATGATACGGCGACCACCGAGATCTACAC[Index]ACACTCTTTCCCTACACGACGCTCTTCCGATC*T-3') and i7 (5'-CAAGCAGAAGACGGCATACGAGAT[Index]GTGACTGGAGTTCAGACGTGTGCTCTTCCGAT*C-3'). Both primers have a phosphorothioate bond added at 3’ end (denoted as *) to increase stability. The PCR reaction was carried out in the volume of 50µl including 25µl 2X OneTaq Master Mix (NEB, #M0482), 2µl of 5µM i5 and i7 each, and 10µl cDNA with total 13~17 cycles of amplification depending on the Ct value of 1µl cDNA qPCR test run. The program was run at 94°C/4min; 53°C/30sec; 68°C/10min; 12~16X [94°C/30sec, 53°C/30sec; 68°C/2min]; 68°C/5min; hold at 4°C. The library was then purified and size-selected with Speedbeads. In brief, 45µl beads (0.9X volume of PCR reaction) were mixed with the reaction to remove larger fragments. Supernatant was collected and mixed with 10µl beads (0.2X volume of PCR reaction) to remove smaller fragments and primer dimers. Two washes with 100% ethanol were then performed while the beads were on the magnet. Beads were dried before resuspended in 12µl of 10mM Tris pH7.4. The library was then collected and subjected to quality control and analysis at Genomics Research Center at the University of Rochester. Poly (A) index Poly (A) index provides a single quantitative measure of APA shifts that reflects the balance between proximal and distal cleavage site usage. It quantifies changes in polyadenylation site usage between two groups, with a focus on the relative use of proximal versus distal polyadenylation sites. For each gene, read counts supporting the proximal PAS (cp, tp) and distal PAS (cd, td) are obtained for the control (c) and treatment (t) conditions. These counts are converted to usage fractions by normalizing the total PAS-supporting reads within each condition: The Poly(A) index is then defined as the log₂ fold change in proximal PAS usage between conditions: This metric captures the direction and magnitude of 3′UTR length changes: Positive values indicate that the control samples use the proximal PAS more frequently than treated samples, implying more distal usage (3′UTR lengthening) in the treated condition. Negative values indicate increased proximal PAS usage in the treated condition, consistent with 3′UTR shortening. A value near zero denotes little or no change in APA between conditions. Comparison of LV and RV ventricular APA datasets APA-regulated transcripts in RV failure patients were identified using PolyA-miner (adjusted P ≤ 0.05). For comparison, the left-ventricular failure (DCM) APA dataset from Creemers et al. (Circ Res. 2016) was used, extracting gene-level APA from Creemers et al. Supplementary Table S1. Overlap was determined by gene-symbol matching, and APA directionality was classified as concordant or opposite. LV–RV APA overlap analysis Alternative polyadenylation (APA) changes in right ventricular (RV) failure were identified from human RV Poly(A)-ClickSeq (PAC-seq) data and analyzed using PolyA-miner. APA direction was summarized at the gene level, with genes classified as RV shortened (RV_S) or RV lengthened (RV_L) based on the direction of the Poly(A) index. Statistical significance was defined as an adjusted gene-level P value (AdjG-Pval ≤ 0.05). For comparison, left ventricular (LV) APA gene lists—LV shortened (LV_S) and LV lengthened (LV_L)—were obtained from the published LV failure dataset by Creemers et al. (Circulation Research, 2016), using the study’s supplementary gene-level APA table. Overlap between LV and RV APA datasets (shown in Figure C) was determined by gene-symbol matching. The complete LV and RV gene lists used for overlap analysis are provided in Table S1. Statistical analysis: Unless otherwise stated, all quantitative data are presented as mean ± SEM from independent biological replicates. Statistical analyses were performed using GraphPad Prism 10, and the number of biological replicates (n) for each experiment is reported in the corresponding figure legends. Normality was assessed using the Shapiro–Wilk test. For datasets that passed normality testing and had adequate sample size (n ≥ 6), parametric tests were applied, including unpaired two‑tailed Student’s t‑test for two‑group comparisons or one‑way ANOVA followed by Tukey’s post hoc test for comparisons involving three or more groups. For datasets that did not meet normality assumptions or had small sample sizes (n < 6), non‑parametric tests were used, including the Mann–Whitney U test for unpaired comparisons, the Wilcoxon matched‑pairs signed‑rank test for paired data, or the one‑sample Wilcoxon signed‑rank test for comparisons against a theoretical value (e.g., zero change in dPAS usage). Statistical testing was not performed on technical replicates. A P value < 0.05 was considered statistically significant. Detailed statistical approaches, including sample sizes, normality testing, and exact tests used for each experiment, are summarized in the Summary of Statistics Table (Supplementary). Summary of Statistics Table Figure: Alternative Polyadenylation Remodeling Drives Right Ventricular Failure Progression Through CPSF6-Dependent 3′UTR Regulation. Figure F. Distal/proximal ratio of CPSF6 in Ctrl RV and Failing RV by PAC-seq Sample Size Shapiro-Wilk Test Statistics test P Value CTRL vs RVF n=4 per group Not Applicable Mann Whitney U test 2.9 x 10-2 Figure F. dPAS usage analysis of CPSF6 in Ctrl and RVF human tissues by qPCR Sample Size Shapiro-Wilk Test Statistics test P Value CTRL vs RVF n=4 per group Not Applicable Mann Whitney U test 2.9 x 10-2 Figure G. CPSF6 protein expression in failing RV human tissues by Western Blot Sample Size Shapiro-Wilk Test Statistics test P Value CTRL vs RVF n=4 per group Not Applicable Mann Whitney U test 2.9 x 10-2 Figure H. CPSF6 protein expression in human failing LV–RV tissues by Western Blot Sample Size Shapiro-Wilk Test Statistics test P Value LV vs RV n=4 per group Not Applicable Wilcoxon matched pairs signed rank test 1.3 x 10-1 Figure I. External validation of CPSF6 reduction using GSE198618 Sample Size Shapiro-Wilk Test Statistics test P Value CTRL vs COMP CTRL n=14 COMP n=11 Normal One-way Anova with Tukey multiple comparison 4.8 x 10-1 CTRL vs DECOMP CTRL n=14 DECOMP n=7 Normal One-way Anova with Tukey multiple comparison 5.3 x 10-5 COMP vs DECOMP COMP n=11 DECOMP n=7 Normal One-way Anova with Tukey multiple comparison 1.4 x 10-3 Figure J: CPSF6 knockdown validation in human cardiac fibroblasts by western blot Sample Size Shapiro-Wilk Test Statistics test P Value CTRL siRNA vs CPSF6 siRNA n=6 per group Normal Unpaired t-test 2.3 x 10-8 Figure J: CPSF6 knockdown validation in human cardiac fibroblasts by qPCR Sample Size Shapiro-Wilk Test Statistics test P Value CTRL siRNA vs CPSF6 siRNA n=6 per group Normal Unpaired t-test 1.2 x 10-9 Figure N: dPAS usage of COL1A1 and TGFβR1 in CPSF6 knockout cardiac fibroblasts Sample Size Shapiro-Wilk Test Statistics test P Value COL1A1 CTRL siRNA vs CPSF6 siRNA n=3 per group Not Applicable One sample Wilcoxon -test versus 0 2.5 x 10-1 TGFβR1 CTRL siRNA vs CPSF6 siRNA n=3 per group Not Applicable One sample Wilcoxon -test versus 0 2.5 x 10-1 Figure O: Immunofluorescence and quantitative analysis of COL1A1 in CPSF6-knockout cardiac fibroblasts Sample Size Scale Bar Shapiro-Wilk Test Statistics test P Value CTRL siRNA vs CPSF6 siRNA n=6 per group 150µm Non-normal Mann Whitney U test 2.2 x 10-3 Figure P: Gel Contraction and quantitative analysis in CPSF6-knockout cardiac fibroblasts Sample Size Shapiro-Wilk Test Statistics test P Value CTRL siRNA vs CPSF6 siRNA n=6 Normal Unpaired t-test P = 8.3 x 10-9 Figure Q: Representative image of Masson Trichrome staining of Human RV tissues Sample Size Scale Bar Shapiro-Wilk Test Statistics test P Value CTRL vs RVF n=4 per group 50µm Not Applicable Mann Whitney U test 2.9 x 10-2 Figure Q: Representative Immunofluorescence image of GFP and COL1A1 in Human RV failure cardiac fibroblasts Scale Bar RVF+Ad GFP vs RVF+Ad CPSF6 GFP 150µm Figure Q: Overexpression validation of CPSF6 in Human RV failure cardiac fibroblasts Sample Size Shapiro-Wilk Test Statistics test P value RVF+Ad GFP vs RVF+Ad CPSF6 GFP n=4 per group Not Applicable Mann Whitney U test 2.9 x 10-2 Figure Q: Analysis of dPAS usage of COL1A1 in Human RV failure cardiac fibroblasts Sample Size Shapiro-Wilk Test Statistics test P value RVF+Ad GFP vs RVF+Ad CPSF6 GFP n=3 per group Not Applicable One sample Wilcoxon -test versus 0 2.5 x 10-1 Related ManuscriptData deposited in response to reviewer request to ensure accessibility and reuse of primary sequencing results associated with a Circulation Research Research Letter.