Satellite cell-derived TRIM28 is pivotal for mechanical load- and injury-induced myogenesis

Abstract

Satellite cells are skeletal muscle stem cells that contribute to postnatal muscle growth, and they endow skeletal muscle with the ability to regenerate after a severe injury. Here we discover that this myogenic potential of satellite cells requires a protein called tripartite motif-containing 28 (TRIM28). Interestingly, different from the role reported in a previous study based on C2C12 myoblasts, multiple lines of both in vitro and in vivo evidence reveal that the myogenic function of TRIM28 is not dependent on changes in the phosphorylation of its serine 473 residue. Moreover, the functions of TRIM28 are not mediated through the regulation of satellite cell proliferation or differentiation. Instead, our findings indicate that TRIM28 regulates the ability of satellite cells to progress through the process of fusion. Specifically, we discover that TRIM28 controls the expression of a fusogenic protein called myomixer and concomitant fusion pore formation. Collectively, the outcomes of this study expose the framework of a novel regulatory pathway that is essential for myogenesis.

Keywords: Cell Fusion; Hypertrophy; Regeneration; Skeletal Muscle.

Figure 1. Maximal intensity contractions induce TRIM28(S473) phosphorylation in satellite cells.

Tibialis anterior muscles of male wild-type (WT) mice were subjected to a bout of maximal-intensity contractions (MIC+) or the control condition (MIC−) and collected at 1 h post-treatment. (A) Mid-belly cross-sections of the tibialis anterior muscles were subjected to immunohistochemistry for phosphorylated (P) TRIM28(S473), dystrophin (to identify the outer boundary of the myofibers), and nuclei. White arrows indicate P-TRIM28 positive myonuclei and the yellow arrow indicates a P-TRIM28 positive interstitial nucleus. (B) Quantification of the % of P-TRIM28(S473) positive (+) interstitial nuclei from (A) (paired Student’s t-test, n = 3/group, *p = 0.0245). (C) Mid-belly cross-sections of MIC− and MIC+ tibialis anterior muscles were subjected to immunohistochemistry for P-TRIM28(S473), Pax7 (to identify satellite cells), and nuclei. White arrows indicate P-TRIM28 negative satellite cells and the orange arrow indicates a P-TRIM28 positive satellite cell. (D) Images in (C) were used to quantify the percentage of satellite cells that were P-TRIM28(S473) positive (paired Student’s t-test, n = 4/group, *p = 0.0103). (E) WT mice and tamoxifen-inducible satellite cell-specific TRIM28 knockout mice (KO) were treated with tamoxifen. At 14 days post tamoxifen, mid-belly cross-sections from the plantaris muscle were subjected to immunohistochemistry for TRIM28, Pax7, and nuclei. White arrows indicate satellite cells. (F–H) Quantification of the percentage of satellite cells that were TRIM28 positive (unpaired Student’s t-test, n = 4/group, *p < 0.0001) (F) the percentage of Pax7 negative nuclei that were TRIM28 positive (unpaired Student’s t-test, n = 7–8/group) (G), and the satellite cell to myofiber ratio (unpaired Student’s t-test, n = 4/group) (H). All values are group means ± SEM. * indicates significant difference between groups, P < 0.05. Scale bars = 25 μm. Source data are available online for this figure.

Figure 2. The loss of TRIM28 does not alter the mechanical load-induced proliferation or differentiation of satellite cells.

Wild-type (WT) mice and tamoxifen-inducible satellite cell-specific TRIM28 knockout mice (KO) were treated with tamoxifen. At 14 days post tamoxifen, the plantaris muscles of mice were subjected to a unilateral synergist ablation surgery (SA+), with the non-ablated limb serving as a sham control (SA−). The plantaris muscles were collected 4 days post-surgery and subjected to the following analyses. (A) Mid-belly cross-sections were subjected to immunohistochemistry for Pax7, Ki67 (a marker of proliferating cells), and nuclei. White arrows indicate Ki67 negative satellite cells and orange arrows indicate Ki67 positive satellite cells. (B) Quantification of the percentage of satellite cells in A that were Ki67 positive (two-way ANOVA, n = 4/group, *p < 0.0001). (C) Mid-belly cross-sections were subjected to immunohistochemistry for MYOG (a marker of differentiated satellite cells), dystrophin, and nuclei. (D) Quantification of the MYOG positive cell to myofiber ratio in (C) (two-way ANOVA, n = 8/group, *p = 0.009 or 0.0002). Values are means ± SEM. * indicates a significant effect of SA within the given genotype, p < 0.05. Scale bars = 50 μm. Source data are available online for this figure.

Figure 3. The loss of TRIM28 in satellite cells attenuates mechanical load-induced myonuclear accretion but it does not alter the increase in muscle mass or myofiber size.

Wild-type (WT) mice and tamoxifen-inducible satellite cell-specific TRIM28 knockout mice (KO) were treated with tamoxifen. At 14 days post tamoxifen, mice were subjected to unilateral synergist ablation surgery (SA+), with the non-ablated limb serving as a sham control (SA−). The mice were given daily intraperitoneal injections of BrdU for 14 days to label cells with replicated DNA and then the plantaris muscles were collected. (A) Mid-belly cross-sections were subjected to immunohistochemistry for BrdU, dystrophin, and nuclei. Yellow arrows indicate BrdU-positive myonuclei. (B) Quantification of the BrdU-positive myonuclei to myofiber ratio (two-way ANOVA, n = 6–9/group, *p < 0.0001 or = 0.0015, †p = 0.0002) and (C) the total myonuclei to myofiber ratio in A (two-way ANOVA, n = 12/group, * and †p < 0.0001). (D) Mid-belly cross-sections were subjected to immunohistochemistry for laminin and the identification of myofiber type. (E) The average CSA of the type IIa, IIx, and IIb myofibers were determined and reported as type II myofiber CSA (two-way ANOVA, n = 11–12/group, *p = 0.0013 or 0.0164). (F) The muscle mass to body mass ratio (two-way ANOVA, n = 11–12/group, *p < 0.0001). Note: half of the mice in (E) and (F) were BrdU injected and the other half were not. Values are group means ± SEM.* indicates a significant effect of SA within the given genotype, † indicates a significant difference between the SA groups, p < 0.05. Scale bars = 50 μm. Source data are available online for this figure.

Figure 4. The loss of TRIM28 in satellite cells leads to impaired fusion during injury-induced myogenesis/regeneration.

Wild-type (WT) mice and tamoxifen-inducible satellite cell-specific TRIM28 knockout mice (KO) mice were treated with tamoxifen. At 14 days post tamoxifen, their tibialis anterior muscles were injected with BaCl₂ to induce injury or PBS as a control condition. (A) At 5, 7, 10, and 21 days post-injury (dpi), tibialis anterior muscles were collected and mid-belly cross-sections were subjected to immunohistochemistry for eMHC, laminin, and nuclei. (B–E) The myoblasts/myofibers in (A) were categorized as being in one of the following stages: fusing, fused, mature, or pre-existing. (B) Pre-existing (gray arrows) were defined as eMHC negative myofibers that did not contain centrally located nuclei and were likely not damaged by BaCl₂. (C) Mature (cyan arrows) were defined as eMHC-negative myofibers with centrally located nuclei. Fused (yellow arrows) were defined as eMHC-positive myoblasts/myofibers that did not contain interior laminin borders. (D, E) Fusing (white arrows) were defined as myoblasts/myofibers that contained clusters of eMHC positive cells that possessed individual laminin borders and were surrounded by a thicker outer laminin border (white dashed line in E). (F) Proportion of myoblasts/myofibers at 5, 7, and 10 dpi that were at the stage of fusing, fused, or mature (5 dpi: two-way ANOVA, n = 4/group; 7 dpi: two-way ANOVA, n = 6/group, *p < 0.0001; 10 dpi: two-way ANOVA, n = 4/group, *p < 0.0001). Values are means ± SEM, * indicates a significant difference from WT within the same category of myoblasts/myofibers, p < 0.05. Scale bars = 50 μm. Source data are available online for this figure.

Figure 5. The loss of TRIM28 in satellite cells leads to a profound functional deficit following BaCl₂-induced injury.

Wild-type (WT) mice and tamoxifen-inducible satellite cell-specific TRIM28 knockout mice (KO) mice were treated with tamoxifen. At 14 days post tamoxifen, their tibialis anterior muscles were injected with BaCl₂ (+) to induce injury or PBS (−) as a control condition. At 7 days post-injury (dpi), (A) the muscle mass to body mass ratio (two-way repeated-measures ANOVA, n = 6/group) and (B) in situ peak tetanic force production were measured (two-way repeated-measures ANOVA, n = 6–7/group, *p < 0.001, †p < 0.001). At 21 dpi, (C) the muscle mass to body mass ratio (two-way repeated-measures ANOVA, n = 6–7/group, *p = 0.0002 or <0.0001, †p < 0.0001) and (D) in situ peak tetanic force production were measured (two-way repeated-measures ANOVA, n = 6/group, * and †p < 0.0001). (E) The gross appearance of tibialis anterior muscles at 21 dpi. Values are group means ± SEM. * indicates a significant effect of injury, † indicates a significant difference between the BaCl₂ treated groups, p < 0.05. Source data are available online for this figure.

Figure 6. Knockdown of TRIM28 leads to a fusion defect during in vitro myotube formation.

(A–D) Primary myoblasts were infected with lentivirus expressing scrambled shRNA (CNT) or shRNA targeting Trim28 mRNA (KD). The infected cells were cultured in growth media (GM) or differentiation media (DM) for 2 days and subjected to western blot analysis of TRIM28 (two-way repeated-measures ANOVA, n = 4/group, *p < 0.001 or = 0.005), MYOD (two-way repeated-measures ANOVA, n = 4/group), and MYOG (two-way repeated-measures ANOVA, n = 4/group). (E–G) Infected myoblasts were subjected to a myotube formation assay and immunohistochemistry for MHC and nuclei. (F) The differentiation index (% of nuclei inside MHC positive cells) (unpaired Student’s t-tests, n = 4/group), and (G) the fusion index (% of nuclei inside MHC positive multinucleated cells) were quantified (unpaired Student’s t-tests, n = 4/group, *p < 0.0001). (H) 2 × 10⁴ infected myoblasts were seeded on day 0 and quantified on days 1 and 3 (two-way repeated-measures ANOVA, n = 4/group). Values are group means ± SEM, each sample representing an independent line of isolated primary myoblasts. * indicates a significant effect of genotype within a given condition. P < 0.05. Horizontal line indicates a significant effect of DM. Scale bars = 50 μm. Source data are available online for this figure.

Figure 7. TRIM28(S473) phosphorylation in satellite cells is not required for injury-induced fusion in vivo.

Strategy for creating mice that allowed for tamoxifen-inducible and satellite cell-specific knockout of endogenous TRIM28 along with tamoxifen-inducible and satellite cell-specific “rescue” expression of (A) human TRIM28 (KO + hTRIM28) or (B) a S473A phosphodefective mutant of hTRIM28 (KO + hTRIM28(S473A)). (C) Wild-type (WT), tamoxifen-inducible satellite cell-specific TRIM28 knockout (KO), KO + hTRIM28, and KO + hTRIM28(S473A) mice were treated with tamoxifen. At 14 days post-tamoxifen, mid-belly cross sections from the tibialis anterior muscles were subjected to immunohistochemistry for Pax7, HA, and nuclei. White arrows indicate satellite cells. (D) At 14 days post tamoxifen tibialis anterior muscles were injected with BaCl₂ to induce injury or PBS as a control condition. At 10 days post-injury (dpi), mid-belly cross sections of the tibialis anterior muscles were subjected to immunohistochemistry for eMHC, laminin, and nuclei. As defined in Fig. 4, myoblasts/myofibers in the injured muscles were assigned to one of the following four stages: fusing, fused, mature, or pre-existing. (E) Proportion of myoblasts/myofibers that were at the stage of fusing, fused, or mature (two-way repeated-measures ANOVA, n = 6/group, *p < 0.001, = 0.001, = 0.004, = 0.009, = 0.029, or = 0.049). Values are presented as the group mean ± SEM. Data were analyzed with two-way repeated-measures ANOVA. * indicates a significant difference from WT, p < 0.05. Scale bars = 50 µm. Source data are available online for this figure.

Figure 8. TRIM28(S473) phosphorylation in satellite cells is not required for injury-induced myogenesis/regeneration.

At 14 days post tamoxifen the tibialis anterior muscles of wild-type (WT) mice, tamoxifen-inducible and satellite cell-specific TRIM28 knockout mice (KO) mice, as well as KO mice that contain tamoxifen-inducible “rescue” expression of hTRIM28 (KO + hTRIM28) or phosphodefective hTRIM28 (KO + hTRIM28(S473A)) were injected with BaCl₂ (+) to induce injury or PBS (−) as a control condition. The tibialis anterior muscles were collected at 21 days post-injury. (A) Gross appearance of tibialis anterior muscles. Scale bars = 2 mm (see Fig. 5 for representative images of the WT and KO muscles). (B) The muscle mass to body mass ratios (two-way repeated-measures ANOVA, n = 6-9/group, *p < 0.0001 or = 0.046, †p < 0.0001). (C) Mid-belly cross sections of the muscles were subjected to immunohistochemistry for the HA tag, dystrophin, and nuclei. Scale bars = 50 µm. (D) In situ peak tetanic force production (two-way repeated-measures ANOVA, n = 6–7/group, * and † p < 0.0001). Values are presented as the group mean ± SEM. * indicates a significant effect of BaCl₂ within a given genotype, † indicates significant difference from the BaCl₂ treated WT condition, p < 0.05. Source data are available online for this figure.

Figure 9. The loss of TRIM28 inhibits the process of fusion pore formation but not hemifusion.

Primary myoblasts were infected with a lentivirus expressing either scrambled shRNA (CNT) or shRNA targeting Trim28 (KD) and then differentiated for 24 h. (A, B) CNT and KD myoblasts were detached and briefly labeled with either an intracellular content probe or a lipid membrane probe. The separately labeled cells were then mixed and differentiated for an additional 30 h. Mononucleated cells that were positive for both probes were classified as having only undergone hemifusion (yellow arrows), whereas multinucleated were classified as having undergone fusion pore formation (pink arrows). (C, D) Quantification of the proportion of nuclei in cells that had undergone (C) only hemifusion, or (D) fusion pore formation (unpaired Student’s t-test, n = 4/group, *p < 0.0001). (E, F) CNT and KD myoblasts were detached, separated into two sub-groups, and labeled with a red- or green-fluorescent intracellular content probe. The separately labeled cells were mixed as indicated in (F) and differentiated for an additional 24 h. Multinucleated cells that were positive for both probes (white arrows) were defined as having undergone fusion pore formation. (G) Quantification of the proportion of nuclei in cells that had undergone fusion pore formation during heterologous fusion (one-way ANOVA, n = 4/group, *p = 0.0144, = 0.0090, or <0.0001). Values are group means ± SEM, each sample representing an independent line of isolated primary myoblasts. * indicates a significant difference between the indicated groups, p < 0.05. Scale bars = 50 μm. Source data are available online for this figure.

Figure 10. The loss of TRIM28 inhibits the induction of MYMX expression during differentiation and injury-induced myogenesis.

(A–C) As in Fig. 6, primary myoblasts were infected with lentivirus expressing scrambled shRNA (CNT) or shRNA targeting Trim28 mRNA (KD). The infected cells were cultured in growth media (GM) or differentiation media (DM) for 2 days and subjected to western blot analysis of myomaker (MYMK) (two-way repeated-measures ANOVA, n = 4/group) and myomixer (MYMX) (two-way repeated-measures ANOVA, n = 4/group, *p < 0.0011). (D, E) Simple linear regression analysis of TRIM28 and MYMK expression (D) or MYMX (E) from the infected cells that had been cultured in DM for 2 days (dashed lines represent the 95% confidence intervals). See Fig. 6 for the western blot analysis of TRIM28. (F–J) Tibialis anterior muscles from wild-type (WT) and tamoxifen-inducible satellite cell-specific TRIM28 knockout (KO) mice were injected with BaCl₂ to induce injury or PBS as a control condition. At 3 days post-injury the muscles were subjected to western blot analysis of MYOD (two-way repeated-measures ANOVA, n = 6/group), MYOG (two-way repeated-measures ANOVA, n = 5–6/group, *p = 0.001), MYMK (two-way repeated-measures ANOVA, n = 6/group), and MYMX (two-way repeated-measures ANOVA, n = 5–6/group, *p = 0.001). Values are group means ± SEM, each sample representing an independent animal or line of isolated primary myoblasts. * indicates a significant effect of genotype within a given condition, p < 0.05. Horizontal line indicates a significant effect of DM or BaCl₂. Source data are available online for this figure.

Figure EV1. The loss of TRIM28 in satellite cells does not impact the mechanical load-induced increase in the size of Type IIa or IIx myofibers.

Wild-type (WT) mice and tamoxifen-inducible satellite cell-specific TRIM28 knockout mice (KO) mice were treated with tamoxifen. At 14 days post tamoxifen, mice were subjected to unilateral synergist ablation surgery (SA+), with the non-ablated limb serving as a sham control (SA-). The mice were treated as described in Fig. 3 and the plantaris muscles were collected at 14 days after the SA surgery. (A) Mid-belly cross-sections were subjected to immunohistochemistry for laminin and type IIx myofibers. (B–D) Quantification of the type IIa (two-way ANOVA, n = 11–12/group, *p < 0.001), type IIx (two-way ANOVA, n = 11–12/group, *p = 0.023 or 0.0025), and type IIb (two-way ANOVA, n = 11–12/group) myofiber cross-sectional area, respectively. Values are group means + SEM. * indicates a significant effect of SA within the given genotype, p < 0.05. Scale bars = 50 μm.

Figure EV2. The loss of TRIM28 in satellite cells leads to excessive fibrosis following BaCl2-induced injury, but reducing the fibrosis only minimally improves the impairment in fusion.

Wild-type (WT) mice and tamoxifen-inducible satellite cell-specific TRIM28 knockout mice (KO) mice were treated with tamoxifen. At 14 days post tamoxifen, their tibialis anterior (TA) muscles were injected with BaCl2 to induce injury or PBS as a control condition. (A) At 5, 7, 10, and 21 days post-injury (dpi), TA muscles were collected and mid-belly cross-sections were subjected to Sirius Red staining as a marker of collagen deposition. (B) Proportion of the muscle cross-section that stained positive for Sirius Red in (A) (two-way ANOVA, n = 4–6/group, *p = 0.0001 or <0.0001). (C–F) Daily intraperitoneal injections of Nilotinib (20 mg/kg/day) or DMSO were administered at 3 to 7 dpi. (C) TA muscles were collected at 10 dpi and mid-belly cross-sections were subjected to Sirius Red staining. (D) Proportion of the muscle cross-sections that stained positive for Sirius Red in (C) (two-way ANOVA, n = 4/group, *p < 0.0001, †p = 0.0016). (E) Mid-belly cross-sections of the TA muscles at 10 dpi were subjected to immunohistochemistry for eMHC, laminin, and nuclei. (F) Proportion of myoblasts/myofibers that were at the stage of fusing, fused, or mature as described in Fig. 4 (two-way ANOVA, n = 4/group, †p = 0.0006 or 0.0145). Values are group means + SEM. * indicates a significant difference between genotypes at the given condition, † significant effect of Nilotinib within the given genotype or stage p < 0.05. Scale bars = 100 μm in (A) and (C), and 50 μm in (E).

Figure EV3. The loss of TRIM28 in satellite cells leads to a fusion defect during in vitro myotube formation.

Wild-type (WT) mice and tamoxifen-inducible satellite cell-specific TRIM28 knockout mice (KO) mice were treated with tamoxifen. (A) At 14 days post tamoxifen, primary myoblasts were isolated, cultured in growth medium, and then subjected to immunohistochemistry for Pax7, TRIM28, and nuclei. (B) The proportion of the Pax7 positive primary myoblasts that expressed TRIM28 in (A) (unpaired Student’s t-test, n = 6/group, *p < 0.0001). (C–E) WT and KO primary myoblasts were subjected to a myotube formation assay and immunohistochemistry for myosin heavy chain (MHC) and nuclei. (D) The differentiation index (% of nuclei inside MHC positive cells) (unpaired Student’s t-test, n = 6/group, *p < 0.0001), and (E) the fusion index (% of nuclei inside MHC positive multinucleated cells) (unpaired Student’s t-test, n = 6/group, *p < 0.0001) were quantified. (F) 3 × 10⁴ primary myoblasts were seeded on day 0 and cultured in growth medium. The number of primary myoblasts was quantified on day 3 (two-way repeated-measures ANOVA, n = 6/group, * and † p < 0.0001). Values are group means + SEM, each sample representing an independent line of isolated primary myoblasts. * indicates a significant difference between genotypes within a given condition, † indicates a significant difference from day 0, p < 0.05. Scale bars = 50 μm.

Figure EV4. Strategy for creating mice that allow for the tamoxifen-inducible expression of human TRIM28 or a S473A phosphodefective mutant of hTRIM28.

(A) Embryos from TARGATT mice that contained an attP integration site in the H11 locus were injected with ΦC31 integrase and an HA-hTRIM28 TARGATT vector that contained a LoxP-Stop-LoxP (LSL) cassette and an attB integration sequence. (B) As illustrated in A, three primers (p1, p2, and p3) were used to confirm the successful integration of the HA-hTRIM28 vector into the genomic DNA of the offspring. (C) CRISPR-Cas9-mediated homologous-directed repair was used to make a single point mutation in TARGATT H11 : HA-hTRIM28 mice that switched the serine 473 residue of hTRIM28 to a non-phosphorylatable alanine (HA-hTRIM28(S473A). (D) Primers p4 and p5 along with Bpu10I digestion were used to confirm the differences in the genomic DNA of the mice that expressed HA-hTRIM28 versus HA-hTRIM28(S473A).

Figure EV5. TRIM28(S473) phosphorylation in satellite cells is not required for the restoration of muscle CSA or myofiber CSA following BaCl₂-induced injury.

At 14 days post tamoxifen the tibialis anterior muscles of wild-type (WT) mice, tamoxifen-inducible and satellite cell-specific TRIM28 knockout mice (KO) mice, as well as KO mice that contain tamoxifen-inducible “rescue” expression of hTRIM28 (KO + hTRIM28) or phosphodefective hTRIM28 (KO + hTRIM28(S473A)) were injected with BaCl₂ (+) to induce injury or PBS (−) as a control condition. The tibialis anterior muscles were collected after a 21-day recovery period. (A) Mid-belly cross sections of the muscles were subjected to immunohistochemistry for eMHC, laminin, and nuclei. Scale bar = 500 µm. (B) Measurements of whole muscle cross-sectional area (CSA) (two-way repeated-measures ANOVA, n = 6/group, *p < 0.001, = 0.002, = 0.036, †p < 0.001), and (C) the mean myofiber CSA per muscle (two-way repeated-measures ANOVA, n = 6/group). Values are presented as the group mean ± SEM, n = 6/group. Due to the absence of clear myofibers in the BaCl₂-treated muscles of KO mice, myofiber CSA data for these mice was not determined (ND). * indicates a significant effect of BaCl₂ within a given genotype, † indicates significant difference from the BaCl₂ treated WT condition, p < 0.05.