RNA-Seq analysis of isolated satellite cells in Prmt5 deficient mice

Abstract

Satellite cells (SCs) represent a distinct population of stem cells, essential for maintenance, growth and regeneration of adult skeletal muscle. SCs are mononuclear and are located between the basal lamina and the plasma membrane of myofibers. They are typically characterized by presence of the transcription factor paired-box 7 (PAX7) that is widely used as a satellite cell marker. Under normal physiological conditions SCs are quiescent but are activated by insults such as injury, disease or exercise. Once activated, satellite cells proliferate and subsequently differentiate into myoblasts to finally fuse to form new myofibers or with preexisting myofibers to repair or rebuild the skeletal muscle. A minority of SCs retains stem cell characteristics and self-renews to assure future bouts of regeneration throughout most of adult life. While a comprehensive picture of the regulatory events controlling SC fate has not yet been achieved, several factors were recently identified playing important roles in functional processes. One example is the arginine methyltransferase Prmt5 that is known to have multiple roles in germ cells and is involved in the maintenance of ES cell pluripotency. We have previously shown that Prmt5 is required for muscle stem cell proliferation and regenerative myogenesis due to direct epigenetic regulation of the cell cycle inhibitor p21. Here we provide a dataset that investigates the loss of Prmt5 in isolated Pax7⁺ primary SCs using the Pax7^CreERT2/Prmt5^loxP/loxP knockout mouse model.

RNA-Seq raw and analyzed data have been deposited in GEO under accession code GSE66822.

Specifications
Organism/cell line/tissue	Murine satellite cells (Pax7CreERT2/Prmt5loxP/loxP)
Sequencer or array type	IonTorrent Proton
Data format	Raw and analyzed RNA-Seq data
Experimental factors	Wildtype, Prmt5-knockout
Experimental features	Arginine methyltransferase Prmt5 was inactivated in isolated MuSCs from adult mice (Pax7CreERT2/Prmt5 loxP/loxP + TAM) and compared to the control condition (Pax7CreERT2/Prmt5+/loxP + TAM) using RNA-Seq.
Consent	Not applicable
Sample source location	Not applicable

1. Direct link to deposited data

http://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE66822.

2. Experimental design, materials and methods

2.1. Experimental design

Two Pax7^CreERT2/Prmt5^loxP/loxP mice and two Pax7^CreERT2/Prmt5^+/loxP mice were used to isolate satellite cells by dissection of hind limb muscle followed by FACS sorting using genetic markers or staining of SC-specific surface proteins. All samples were treated in culture with 4-OH tamoxifen to produce Prmt5 gene knockouts while those with +/loxP represent the control situation. Five days later RNA was isolated, used for sequencing, and assessed for differentially expressed genes.

2.2. Sample collection and RNA analysis

SC cell isolation and in vitro ablation of Prmt5 were performed as previously described [1]. RNA of FACS sorted satellite cells from mutant (Pax7^CreERT2/Prmt5^loxP/loxP + TAM) and wild type mice (Pax7^CreERT2/Prmt5^+/loxP + TAM) was isolated using the miRNeasy micro Kit (Qiagen). On-column DNase digestion (DNase-Free DNase Set, Qiagen) was performed according to manufacturer protocols to prevent genomic DNA contamination. RNA and library preparation integrity were verified using a 2100 BioAnalyzer system (Agilent). Ribosomal depletion was performed by RiboMinus Eukaryote System v2 (Life Technologies) with 500 ng total RNA input following the low input protocol. Finally, libraries were prepared using the Ion Total RNA-Seq Kit v2 (Life Technologies) with minor changes of the standard protocol including lower amplification cycles and usage of entire fragmented RNA for cDNA synthesis. For the sequencing reactions an Ion Torrent Proton platform with V3 chemistry (Ion PI Template OT2 200 Kit v3, Life Technologies) and PIV2 Chips (Ion PI™ Chip Kit v2, Life Technologies) was used. The experiment was performed on two chips containing two biological replicates per condition (wt1/mt1 on chip A, wt2/mt2 on chip B) resulting in 163 M reads in total and at least 40 M reads per library (Table 1). According to previous studies, 10 million single-ended raw reads are sufficient for differential expression analyses of the most abundant transcripts, while 50 ⁺ million reads are desirable for maximum performance [2].

Table 1.

Overview of general sequencing and mapping statistics.

Sample	Proton chip id	Raw reads [#]	Mean raw read length [nt]	Trimmed reads [#]	Mean trimmed read length [nt]	STAR mapped reads [#]	Reads unambiguously assigned to genes [#]
wt1	A	40,761,503	96	30,762,527	85	29,050,721	13,242,187
wt2	B	41,299,259	87	33,600,312	83	28,763,316	14,438,850
mt1	A	40,342,148	93	32,394,674	89	28,504,113	16,192,323
mt2	B	41,805,514	135	38,973,691	119	33,516,399	20,050,595

2.3. QC, mapping and normalization

The resulting raw reads were assessed for quality, adapter content and duplication rates with FastQC 0.10.1 [3]. No remaining adapters could be identified. Reaper version 13-100 was employed to trim reads after a quality drop below a mean of Q20 in a window of 10 nucleotides [4]. Only reads between 40 and 150 nucleotides were cleared for further analyses (Parameters: -clean-length 40 -qqq-check 53/10 -trim-length 150). These values were chosen to permit a correct and unique mapping of reads versus the reference genome, avoiding known quality drops on longer Proton reads (Fig. 1). The resulting trimmed FASTQ files were uploaded to GEO.

Fig. 1.

Quality scores per base summarized over all reads plotted by FastQC. Reads were trimmed before the characteristic drop in quality at ~ 150 bp.

Trimmed and filtered reads were aligned versus the Ensembl mouse genome version mm10 (GRCm38) using STAR 2.4.0a to increase the maximum ratio of mismatches to mapped length to 10% (Parameters: –outFilterMismatchNoverLmax 0.1) [5]. The effectiveness of the Tamoxifen treatment and resulting knockout of Prmt5 was inspected with IGV (Fig. 2) [6].

Fig. 2.

Visualization of Prmt5 knockdown efficiency with genome browser IGV. The upper track per sample depicts the combined read coverage per genomic position, while the lower track shows singular reads. All tracks were scaled to the same absolute level. Normalized count values show a ~ 12-fold mean reduction of Prmt5 transcripts in knockout mice (mt: Pax7^CreERT2/Prmt5^loxP/loxP) when compared to the control (Pax7^CreERT2/Prmt5^+/loxP).

The number of reads aligning to genes was counted with featureCounts 1.4.5-p1 tool from the Subread package [7]. Only reads mapping at least partially inside exons were admitted and aggregated per gene (Parameters: -t exon -g gene_id). Reads overlapping multiple genes or aligning to multiple regions were excluded to avoid false assignments.

2.4. Differential expression analyses

Differentially expressed genes were identified using DESeq2 version 1.62 [8]. Only genes with a minimum fold change of +/− 2, a maximum Benjamini-Hochberg corrected p-value of 0.05, and a minimum combined mean of 5 reads were classified as significantly differentially expressed, resulting in 504 candidate genes. MA and Volcano plots were computed using R functions to assess the distribution of genes according to counts, fold change and FDR (Fig. 3).

Fig. 3.

Distribution of genes according to fold-change, counts and FDR. Significantly differentially expressed genes as classified by DESeq2 (red) typically show relatively high read counts due to the inherent insecurity associated with low expressed genes.

The DESeq2 result file was uploaded to GEO.

2.5. Data reproducibility

Correlations of raw gene counts per sample were assessed with the Spearman ranked correlation algorithm included in R 3.11 [9]. The log₂-transformed read counts of 20,065 expressed genes were utilized. All samples showed a high correlation of 0.94 to 0.97, with wt1/wt2 and wt1/mt1 being most similar (Fig. 4).

Fig. 4.

Pairwise Spearman correlation of log₂-transformed raw gene counts.