circSamd4 represses myogenic transcriptional activity of PUR proteins

Abstract

By interacting with proteins and nucleic acids, the vast family of mammalian circRNAs is proposed to influence many biological processes. Here, RNA sequencing analysis of circRNAs differentially expressed during myogenesis revealed that circSamd4 expression increased robustly in mouse C2C12 myoblasts differentiating into myotubes. Moreover, silencing circSamd4, which is conserved between human and mouse, delayed myogenesis and lowered the expression of myogenic markers in cultured myoblasts from both species. Affinity pulldown followed by mass spectrometry revealed that circSamd4 associated with PURA and PURB, two repressors of myogenesis that inhibit transcription of the myosin heavy chain (MHC) protein family. Supporting the hypothesis that circSamd4 might complex with PUR proteins and thereby prevent their interaction with DNA, silencing circSamd4 enhanced the association of PUR proteins with the Mhc promoter, while overexpressing circSamd4 interfered with the binding of PUR proteins to the Mhc promoter. These effects were abrogated when using a mutant circSamd4 lacking the PUR binding site. Our results indicate that the association of PUR proteins with circSamd4 enhances myogenesis by contributing to the derepression of MHC transcription.

Figure 1.

circRNAs expressed differentially during myogenesis. (A) Myoblast differentiation program; phase-contrast micrographs depicting cultures of C2C12 mouse myoblasts in proliferating conditions [growth medium (GM)], in early myogenesis after 24 h in differentiation medium (DM), and in late myogenesis, including myotubes, after 72 h in DM. Schematic, total RNA was prepared from proliferating C2C12 cells (GM) and from C2C12 cultures differentiating for 72 h (DM); after digestion of linear RNA, RNA pools enriched in circRNAs were sequenced (GSE92632 and GSE136004), and the circRNAs most highly expressed and most differentially abundant between GM and DM were validated by reverse transcription (RT) followed by real-time quantitative (q)PCR analysis. (B) Expression levels of myogenic markers Myog, Mef2c, Mhc mRNAs in C2C12 cultures, as assessed by RT-qPCR analysis. mRNA levels were normalized to the levels of housekeeping transcript Gapdh mRNA. (C) Expression levels of the indicated circRNAs, chosen among the most differentially abundant in DM relative to GM and most highly expressed overall. RT-qPCR analysis was performed by using divergent primers (Materials and Methods) in proliferating (GM) and differentiated (DM, 72 h) cultures. mRNA levels were normalized to the levels of housekeeping transcript Gapdh mRNA. Bottom, representative PCR products on 2% agarose gels; product specificity was assessed from reactions carried out in the presence or absence of reverse transcriptase (RT). (D) Quantification of copy numbers of circSamd4 in C2C12 cultures in GM (per cell) and DM (per cell/nucleus-equivalent) by RT-qPCR analysis. Data in (B, C) are the means ± SEM from three independent experiments. *P < 0.05, **P < 0.005.

Figure 2.

Characterization of circSamd4. (A) Schematic of exon 3 of Samd4 mRNA, located on mouse chromosome 14 and giving rise to a 519-nucleotide long circSamd4, and detail of the sequence junction. (B) RT-qPCR results showing the abundance of circSamd4 and Samd4 mRNA in total RNA from C2C12 cells (GM) that were either left untreated or treated with RNase R. RNA levels were normalized to the levels of Gapdh mRNA. (C) Relative levels of Samd4 mRNAs as measured by RT-qPCR analysis in GM and DM (72 h) C2C12 cultures. RNA levels were normalized to the levels of Gapdh mRNA in the same samples. (D) Relative levels of circSamd4 in the nuclear (Nuc) and cytoplasmic/sarcoplasmic (Cyto) compartments of C2C12 GM and DM (72 h) cultures; the quality of the fractionation was assessed by monitoring the levels of a predominantly cytoplasmic transcript (Actb mRNA) and a predominantly nuclear lncRNA (7sk). (E) Brightfield micrographs to visualize circSamd4 (red) and Samd4 mRNA (blue) in proliferating (GM) and differentiated (DM) C2C12 cultures using the BaseScope single-molecule detection system (Advanced Cell Diagnostics, CA) (Materials and Methods). Nuclei are counterstained with hematoxylin. Data in (B, C) are the means ±SEM from three or more independent experiments. **P < 0.005.

Figure 3.

Silencing circSamd4 reduces the levels of myogenic markers. (A) Schematic representation of 3 siRNAs designed to target the circSamd4 junction. RT-qPCR analysis of the indicated circular and linear RNAs; C2C12 in GM were transfected with siRNAs, 8 h later the medium was switched to DM and 48 h after that, cells were collected for analysis. RNA levels were normalized to the levels of Gapdh mRNA. (B–D) Effect of silencing circSamd4 on C2C12 differentiation. Ctrl and circSamd4 siRNAs were transfected into C2C12 (GM) cells and differentiation was initiated 24 h later (time 0 h); 48 and 72 h into DM, phase-contrast micrographs were taken to monitor myotube formation (B), the levels of creatine kinase (CK) were measured at 0, 24, 48 and 72 h (C), and RNA was extracted to measure by RT-qPCR analysis the levels of circSamd4 as well as myogenic markers Myog, Mef2c, and Mhc mRNAs, normalized to Gapdh mRNA levels (D). (E, F) The effect of silencing circSAMD4 in human KM155 myoblasts (Supplementary Figure S1) on myogenesis was measured by immunofluorescence analysis of the myotube marker MHC in DM (72 h); nuclei were labeled with DAPI (E), and the levels of the myogenic proteins MHC, MYOG, and MEF2C were measured by western blot analysis (F). (G, H) Forty-eight hours after transfecting pcDNA3 (empty control) or pcDNA3-circSamd4 into C2C12 cells, RT-qPCR analysis was used to assess the expression levels of circSamd4 and Samd4 mRNA (G) as well as those of myogenic markers Myog, Mef2c, and Mhc mRNAs, which were normalized to the levels of Gapdh mRNA (H). Data in (A, C, D, G, H) are the means ± SEM from three independent experiments. *P < 0.05, **P < 0.005.

Figure 4.

circSamd4 interacts with transcription factors PURA and PURB. (A) Left, biotinylated antisense oligomers (ASOs) complementary to the junction of circSamd4 and control (Ctrl) were incubated with C2C12 in GM as well as in DM. After affinity pulldown using streptavidin beads, circSamd4-interacting proteins were identified by mass spectroscopy. Right, levels of circSamd4 enrichment in ASO pulldown samples, as assessed by RT-qPCR analysis of circSamd4 (relative to the enrichment of Gapdh mRNA, a transcript that does not bind the ASOs and encodes a housekeeping protein) in the pulldown samples. Enrichment in Samd4 mRNA was monitored in parallel. (B) Venn diagram representation of the proteins significantly enriched in the mass spec datasets identified after pulldown of circSamd4 ASO vs Ctrl ASOs in mouse C2C12 myoblasts (left) and in human KM155 myoblasts (right) (Supplementary Figure S3; Materials and Methods). 53 proteins were shared in both mass spec datasets (intersection). (C) Pie chart represents GO functional categories of the 53 proteins at the intersection of both datasets. (D) Top 7 proteins shared in the mass spec datasets from mouse and human myoblast pulldowns (Supplementary Table S2 and Supplementary Figure S3A and S3B). (E) Western blot analysis of the interaction of circSamd4 with PURA and PURB present in DM and GM lysates, as assessed after biotin-ASO pulldown followed by western blot analysis of PURA and PURB levels in the pulldown material. Input (In), 5 μg of GM or DM lysates (C2C12). (F) The interaction of PURA and PURB with circSamd4 was assessed as explained in panel (E), except that lysates were treated with RNase A and/or RNase R before pulldown followed by western blot analysis. (G) For RIP (ribonucleoprotein immunoprecipitation) analysis, IP was performed using DM C2C12 lysates and either IgG (control) antibodies or antibodies recognizing PURA or PURB; the presence of circSamd4 and Samd4 mRNA (as well as control Gapdh mRNA) in the IP material was measured by RT-qPCR analysis. Enrichments were normalized to the levels of 18s rRNA. (H) After fractionating nuclei and cytoplasm, RIP analysis was carried out as explained in (G) to measure the interaction of PURA and PURB with circSamd4 in the nucleus and the cytoplasm. Data in (A,G,H) are the means ±SEM from three independent experiments. *P < 0.05, **P < 0.005.

Figure 5.

PUR proteins predominantly associate with a specific region of circSamd4. (A) Schematic of biotin pulldown of PURA and PURB using nonoverlapping biotinylated RNAs (each ∼65 nt long) tiling the length of circSamd4. Following incubation of the biotinylated RNAs with C2C12 DM lysates, bound proteins were pulled down using streptavidin beads and the presence of PURA and PURB was assessed by western blot analysis. Input (In) was 2 μg of DM lysates; Beads only (Beads), no biotinylated RNA. (B, C) Schematic, circSamd4Δ3 bearing a deletion of segment 3. By 24 h after transfecting HeLa cells with vectors (2 μg) to overexpress circSamd4 and circSamd4(Δ3), RIP analysis was used to measure circSamd4 (left graph) and SAMD4 mRNA (right graph) bound to PURA and PURB relative to the levels of GAPDH mRNA (B). The levels of circSamd4Δ3 and circSamd4 expressed in HeLa cells 24 h after transfection of plasmids pcDNA3-circSamd4Δ3 and pcDNA3-circSamd4, respectively, were measured by RT-qPCR analysis and normalized to GAPDH mRNA levels (C). (D) Following overexpression of circSamd4 and circSamd4Δ3 as explained in panel (B), biotin-ASO pulldown was carried out to confirm the interaction of PURA and PURB with circSamd4 in C2C12 lysates as shown by Western blot analysis. The levels of the housekeeping protein HSP90 were assessed as a background control protein. Input was 2 μg of DM lysates. Data in (B,C) are the means ±SEM from three independent experiments. **P < 0.005.

Figure 6.

PUR proteins suppress MHC production transcriptionally. (A, B) C2C12 myoblasts were transfected with siRNAs to silence PURA and PURB; 48 h later, the levels of circSamd4, Samd4 mRNA, and Mhc mRNA were assessed by RT-qPCR analysis and normalized to Gapdh mRNA (A). Western blot analysis was used to evaluate the levels of PURA, PURB, MEF2C, MHC, and loading control HSP90 (B). (C) Schematic of the Mhc promoter reporter construct (pmh2-GLuc-RLuc) (GeneCopoeia Inc.) used to evaluate MHC transcription. GM C2C12 cells were transfected with 1 μg of plasmid pmh2-GLuc-RLuc in cells expressing normal levels or silenced PURA and PURB as explained in panel (A); 8 hours later, cells were switched to DM media, and 24 h after that, promoter activity was measured. GLuc levels were normalized to RLuc levels in each group; data were then plotted as the difference in GLuc/RLuc ratios in the Ctrl siRNA transfection group compared with PURA or PURB silencing groups. (D) C2C12 cells were transfected with control (Ctrl) siRNA or circSamd4 siRNA; 48 h later, the expression levels of Pura and Purb mRNAs, normalized to the levels of Gapdh RNA, were measured by RT-qPCR analysis and plotted. (E, F) Effect of silencing circSamd4 on the expression of PURA, PURB, and MHC; 48 h after transfecting GM C2C12 cells with control (Ctrl) siRNA or circSamd4 siRNA, the levels of circSamd4, and Samd4, Pura, and Purb mRNAs were assessed by RT-qPCR analysis (E) and the levels of proteins PURA, PURB, MHC, and loading control HSP90 by western blot analysis (F). Data in (A,C,D) are the means ±SEM from three or more independent experiments. *P < 0.05, **P < 0.005.

Figure 7.

circSamd4 binds PUR proteins, derepressing Mhc gene transcription. (A) Activity of the Mhc promoter, as determined by monitoring luciferase activity after transfecting 500 ng of reporter pMH2-LUC-Renilla (Figure 6C) into C2C12 cells that had been transfected 24 h earlier with parent vector pcDNA3 or with vectors expressing circSamd4 or circSamd4-Δ3 (left graph). Eight h after transfection of the reporter construct, the cells were switched to DM media for 24 h and promoter activity was measured; as above (Figure 6C), GLuc was normalized to RLuc and these ratios were plotted relative to the ratios measured in the empty vector group (right graph). (B) The levels of MHC in the different transfection groups shown in panel (A) were assessed by western blot analysis and HSP90 levels were shown as the loading control. (C–F) ChIP-qPCR analysis of the association of PURA and PURB with the Mhc proximal promoter region (−270/−450). Following crosslinking (Materials and Methods), antibodies recognizing PURA (C, E) or PURB (D, F) or control nonspecific rabbit IgG antibodies were used to immunoprecipitate sheared chromatin from C2C12 myoblasts in which circSamd4 was silenced (C, D) or overexpressed (E, F). Specific oligonucleotide primers were used to amplify the region of interaction with PURA and PURB. ChIP-qPCR data were calculated using percentage of input; the TTS region of the neuron-specific gene Syn2 (region +200 TTS), which encodes the protein SYN2 (Synapsin), was assayed as a non-target negative control. (G) Summary model. In myoblasts (GM), virtually undetectable levels of circSamd4 permit the repression of the Mhc promoter by PURA and PURB. As myogenesis progresses (DM), the rise in circSamd4 leads to binding of a fraction of PURA and PURB away from the Mhc promoter, thereby helping to derepress Mhc transcription and enabling the accumulation of Mhc mRNA and the production of MHC, a key protein in mature myotubes. Data in (A, C–F) are the means ±SEM from three or more independent experiments. *P < 0.05, **P < 0.005.