Work Description

Date: 08 May, 2025

Dataset Title: The Impact of Acute Endurance Exercise on Alternative Splicing in Skeletal Muscle.

Dataset Creators: A. Ahn & A.T. Ludlow

Dataset Contact: Andrew Ludlow [email protected]

Funding:

Key Points:
- Acute endurance exercise can elicit sufficiently potent perturbations in AS-related responses and RBP expression in skeletal muscle.
- We identified 61 RNA-binding proteins (RBPs) with significant isoform switching (but no overall changes in gene expression), previously unknown to be related to exercise stimuli.
- Among the 61 RBPs detected, we validated our data using RT-PCR in mHnrnpa3, mHnrnpa1, and mTia1. We also focused further on hnRNP-A3, which underwent a significant isoform-switching event with nearly two-fold increases in protein levels post-exercise.

Research Overview:
Publicly accessible short-read RNA sequencing (SRS) of clinical exercise studies were extracted from the Gene Expression Omnibus. Oxford Nanopore long-read RNA sequencing (LRS) was performed on mouse gastrocnemius before and following treadmill exercise. Differential gene expression (DGE), differential alternative splicing (DAS), and differential isoform expression (DIE) were analyzed. Gel-based/droplet digital RT-PCR and western blots were performed to validate expression changes of select genes. Both SRS and LRS illustrated significant DGE in skeletal muscle post-exercise, whereby 89 RBPs were significantly up-/down-regulated. rMATS analysis of SRS data revealed that exon-skipping and intron-retaining splicing events were the most common. Swan analysis of LRS data revealed 61 RBPs with significant isoform switching: one of these RBPs, mHnrnpa3, underwent a significant non-coding to protein-coding switch. HnRNP-A3 protein levels validated nearly two-fold increases at 1 hour and 24 hours post-exercise.

Methodology:
FIGURE 1 Methods:
Short-Read Sequencing (SRS) Data Retrieval, Long-Read (LRS) Library Preparation, and MinION Nanopore Sequencing
This study illustrated that acute endurance exercise can elicit changes in AS-related responses and RNA-binding protein (RBP) expression in skeletal muscle using single-molecule, long-read RNA sequencing technologies.

Bioinformatics Analyses of human SRS Data (hsYS, hsAS, and hsAA): After data retrieval from the Gene Expression Omnibus, FastQC, Trimmomatic, RSEM-STAR, DESeq2, rMATS, and DAVID were used to visualize differential gene expression (DGE) and exon-level detection analyses. Raw sequence files of the human SRS datasets used can be found on the Gene Expression Omnibus (GSE87749 for Pattamaprapanont et al; GSE151066 for Rubenstein et al).

Bioinformatics Analyses of mouse LRS Library (mmYS): After sequencing, Minimap2, samtools, TranscriptClean, TALON, DESeq2, Swan, DAVID, and PANTHER were used to visualize DGE, differential transcript expressions, and isoform-switching analyses. Raw sequence files of the mouse LRS dataset used can be found on the Gene Expression Omnibus (GSE279359).

FIGURE 2 Methods:
Using the human SRS and mouse LRS data sets, DGE analyses of pre- and post-acute endurance exercise were visualized using volcano plots and multivariate plots of DAVID GO terms. Additionally, genes with significant DGE events from the mouse LRS data set were validated using droplet digital RT-PCR (RT-ddPCR) for each time point (mUbe2d1 in pre:ipe, mFbxo32 in pre:1pe, and mFos in pre:24pe).

FIGURE 3 Methods:
89 RBPs with significant DGE (|log2FC| > 0.5; raw p < .05 for mouse, p-adj < .05 for human) in response to an acute bout of exercise in skeletal muscle were visualized on the individual subject basis (raw counts converted to relative Z-scores) in the human and mouse data sets, and the RBP genes were overlapped across the cohorts to find common RBPs.

FIGURE 4 Methods:
Using the human SRS datasets, genes with detections of significant exon-level events were categorized into five types of alternative splicing (skipped exons, retained introns, mutually exclusive exons, alternative 5' splice site, and alternative 3' splice site) using rMATS and were visualized with pie charts. Subsequently, these genes that had at least one significant exon-level events from the three human cohorts were overlapped.

FIGURE 5 Methods:
Using the mouse LRS dataset, genes with significant DGE and/or DAS were visualized across the three different post-exercise time points (immediate, 1 hour, and 24 hours). Subsequently, the number of genes that had significant DAS without significant DGE for each post-exercise timepoint were graphed using GraphPad Prism 10. Additionally, 29 exercise-associated genes with significant DAS but no DGE were found and overlapped across the three post-exercise time points. Further, one of the genes, mSirt2, that had significant DAS was validated using gel-based RT-PCR compared between pre and 24pe. Using the mouse LRS dataset, transcript and isoform-switching profiles of mSirt2 was visualized using full count matrix found in "processed_counts.txt" and graphed using GraphPad Prism 10. Raw images of the gel-based RT-PCR to generate Figure 5G are included in "Raw Data for RT-PCR Figures.zip."

FIGURE 6 Methods:
The isoform-switching genes with significant DAS but no DGE of the three time points were overlapped to reveal 262 common genes. Additionally, these common 262 genes were analyzed with PANTHER GO enrichment and visualized using a multivariate plot. Finally, these 262 genes were overlapped with an in-house list of 1130 RBP genes to reveal 61 RBP genes had significant isoform-switching events across all three time points. Further, several of the common 262 RBP genes (including mHnrnpa3, mHnrnpa1, and mTia1) that had significant DAS were validated using RT-ddPCR and gel-based RT-PCR compared between pre and 24pe. Using the mouse LRS dataset, transcript and isoform-switching profiles of mHnrnpa3, mHnrnpa1, and mTia1 were visualized using full count matrix found in "processed_counts.txt" and graphed using GraphPad Prism 10. Finally, the isoform-switching events of Hnrnpa3 were validated using western blotting of the mouse gastrocnemius tissues with antibodies probing for anti-hnRNP-A3. Raw images of the western blots to generate figures are included in "Raw Data for WB Figures.zip."

Files contained here:
This dataset contains 8 files and 7 compressed folders related to the research. The 8 files consist of the present "README.txt" file, count matrix of mouse LRS dataset, R script to generate figures, transcriptomic gene names and ID downloaded from Ensembl, in-house list of RBP gene names, and saved Swan data with respective Python script. The 4 compressed folder with the prefix "Figure" are in an order of conducting bioinformatics data analyses and generating main figures using the R script file, "Figures_Script.R." Similarly, the compressed folder named "Supplemental Figures" was used to generate supplemental figures using the aforementioned R script file. The compressed folder named "Raw Data for RT-PCR Figures.zip" and "Raw Data for WB Figures.zip" contain all files of raw images and data that were used to create figures of gel-based RT-PCR/RT-ddPCR and western blots.

Formats of files in the dataset are as the following:
.zip: Compressed folders.
.RDS and .R: Bioinformatics data saved in R data serialization files to generate figures using R script.
.csv: Comma-separated values file.
.txt: Text file.
.tsv: Tab-separated values file.
.pdf: Portable document format file.
.prism: GraphPad Prism 10 data to conduct statistical analyses and generate bar chart figures.
.png: Portable network graphics image files of gel-based RT-PCR and western blot figures using Bio-Rad ImageLab software.
.p: Swan data saved in pickle file to generate figures as part of our analysis pipeline.
.zip: Compressed folders.

List of files and folders:
"processed_counts.txt" - Processed counts matrix of the mouse LRS data.
"Figures_Script.R" - R script used to perform bioinformatics analyses and generate figures.
"Figure 2.zip" - Saved RDS and data files to generate Figure 2 using R script.
"Figure 3.zip" - Saved RDS and data files to generate Figure 3 using R script.
"Figure 4.zip" - Saved RDS and data files to generate Figure 4 using R script.
"Figure 5.zip" - Saved RDS and data files to generate Figure 5 using R script.
"Raw Data for RT-PCR Figures.zip" - Prism and image files containing raw data to create main and supplementary figures.
"Raw Data for WB Figures.zip" - Prism and image files containing raw data to create main and supplementary figures.
"Supplementary Figures.zip" - Saved RDS and data files to generate supplementary figures using R script.
"hs.mm_ensembl.txt" - Ensembl transcriptome gene names and gene ID of homo sapiens and mus musculus.
"hs_ensembl.txt" - Ensembl transcriptome gene names and gene ID of homo sapiens only.
"mm_ensembl.txt" - Ensembl transcriptome gene names and gene ID of mus musculus only.
"RBPs_list.tsv" - In-house list of RBPs for overlap analyses.
"mmYS-pre.pe_Swan.all.5.p" - Saved Swan data file for DGE, DAS, and DIE analyses.

Related publication(s):
Ahn, A., Kim, J. J., Slusher, A. L., Ying, J. Y., Zhang, E. Y., & Ludlow, A. T. (2024). The Impact of Acute Endurance Exercise on Alternative Splicing in Skeletal Muscle. bioRxiv, 2024-11.

Use and Access:

To Cite Data:
Ahn, A., Kim, J. J., Slusher, A. L., Ying, J. Y., Zhang, E. Y., & Ludlow, A. T. (2025). The Impact of Acute Endurance Exercise on Alternative Splicing in Skeletal Muscle. [Data set]. University of Michigan - Deep Blue. https://doi.org/to_be_modified