Thbs1 regulates skeletal muscle mass in a TGFβ-Smad2/3-ATF4-dependent manner

Abstract

Loss of muscle mass is a feature of chronic illness and aging. Here, we report that skeletal muscle-specific thrombospondin-1 transgenic mice (Thbs1 Tg) have profound muscle atrophy with age-dependent decreases in exercise capacity and premature lethality. Mechanistically, Thbs1 activates transforming growth factor β (TGFβ)-Smad2/3 signaling, which also induces activating transcription factor 4 (ATF4) expression that together modulates the autophagy-lysosomal pathway (ALP) and ubiquitin-proteasome system (UPS) to facilitate muscle atrophy. Indeed, myofiber-specific inhibition of TGFβ-receptor signaling represses the induction of ATF4, normalizes ALP and UPS, and partially restores muscle mass in Thbs1 Tg mice. Similarly, myofiber-specific deletion of Smad2 and Smad3 or the Atf4 gene antagonizes Thbs1-induced muscle atrophy. More importantly, Thbs1^-/- mice show significantly reduced levels of denervation- and caloric restriction-mediated muscle atrophy, along with blunted TGFβ-Smad3-ATF4 signaling. Thus, Thbs1-mediated TGFβ-Smad3-ATF4 signaling in skeletal muscle regulates tissue rarefaction, suggesting a target for atrophy-based muscle diseases and sarcopenia with aging.

Keywords: ATF4; CP: Metabolism; TGFβ; atrophy; autophagy; extracellular matrix; proteasome; skeletal muscle; thrombospondin.

Figure 1.. Thbs1 is induced in mouse models of skeletal muscle atrophy

(A) Fold change in the expression of Thbs1 mRNA in soleus following 30 days of microgravity exposure (space flight) compared to control. Microarray dataset GEO: GSE80223 was analyzed from the Gene Expression Omnibus repository. Error bars denote ±SEM from n = 3 biologically independent animals per group; **p < 0.01 by two-tailed unpaired Student’s t test. (B) RT-qPCR for Thbs1, Thbs2, Thbs3, Thbs4, and Thbs5 mRNA in the right tibialis anterior (TA) of 12-week-old wild-type (WT) mice subjected to 3 or 6 days (d) of unilateral hindlimb denervation, compared contralateral sham-operated leg. Error denote ±SEM from n = 3–5 biologically independent animals per group. *p < 0.05 by two-tailed unpaired Student’s t test. (C) Western blot for Thbs1 and Gapdh control in right TA of 12-week-old WT mice subjected to 3 or 10 days (d) of unilateral hindlimb denervation, compared to contralateral, sham-operated TA. n = 2 biologically independent animals per time point. (D) Representative immunohistochemistry for endogenous Thbs1 (green) and BiP (red) to visualize the endoplasmic reticulum (ER) on cryo-embedded TA of WT mice subjected to 10 days (d) of denervation compared to sham-operated controls at 12 weeks of age. Nuclei are shown in blue with DAPI. Scale bars represent 50 μm. (E) RT-qPCR for Thbs1 mRNA from TA of 8-week-old mice fed ad libitum or fasted for 48 h. Data are presented as fold expression over fed WT; error bars denote ±SEM from n = 4 biologically independent animals analyzed per group. *p < 0.05 by two-tailed unpaired Student’s t test. (F and G) Western blot for Thbs1 and Gapdh in TA of 8-week-old mice fed ad libitum or fasted for 48 h (F), and in young (12 weeks of age) and old (24 months of age) WT quadriceps (G). (H) Representative immunohistochemistry for endogenous Thbs1 (green), BiP (red), and the nucleus (DAPI, blue) on cryo-embedded 12-week-old (“young”) and 24-month-old (“old”) WT quadriceps. Scale bars represent 50 μm.

Figure 2.. Transgene-mediated overexpression of Thbs1 induces skeletal muscle atrophy

(A) Schematic diagram of the Thbs1 skeletal muscle-specific transgene. (B) Western blot analyses for Thbs1 or Gapdh control from quadriceps from non-transgenic (Ntg) and skeletal muscle-specific Thbs1 transgenic (Tg) mice at 6 and 52 weeks of age (n = 2 biological replicates per genotype and per time point). (C) Thbs1 protein levels (ng/mL) determined by ELISA in quadriceps (Quad) and plasma obtained from Ntg and Thbs1 Tg mice at 6 weeks of age. Error bars denote ±SEM from n = 4 biologically independent animals per genotype. *p < 0.0001 by two-tailed unpaired Student’s t test). (D) Immunohistochemical fluorescent labeling of Thbs1 (green) and BiP to show the ER (red) on cryo-embedded quadriceps (Quad.) from Ntg and Thbs1 Tg mice at 6 weeks of age. Nuclei are visualized in blue (DAPI). Representative images of four mice per genotype are shown. Scale bars represent 50 μm. (E and F) Body weight (E) and quadriceps (Quad) weight normalized to tibia length (F) of Ntg and Thbs1 Tg mice at the indicated ages in weeks (wks). Error bars denote ±SEM from n = 5–9 biologically independent animals per genotype and per age. *p < 0.001 (E) and *p < 0.0001 (F) by two-tailed unpaired Student’s t test). (G) Representative Masson’s trichrome-stained histological sections of Ntg and Thbs1 Tg quadriceps (Quad) at 1 year of age. Representative images of five mice per genotype are shown. Scale bar represents 1 mm. (H) H&E histological staining on quadriceps sections from Ntg and Thbs1 Tg at 6 weeks of age. Representative images of 4–5 mice per genotype are shown. Scale bars represent 50 μm. (I) Myofiber cross-sectional area (CSA) in quadriceps (Quad) from Ntg and Thbs1 Tg mice at 6 weeks of age (n = 4–5 biologically independent animals per genotype; *p < 0.01 by two-tailed unpaired Student’s t test). The legend above the bar plot also applies to (J) and (K). (J and K) Running time to fatigue in seconds (sec) with a forced downhill treadmill (J) and forelimb grip strength (kg/kg body weight; K) from Ntg and Thbs1 Tg mice at the indicated ages in weeks (n = 5–18 biologically independent animals per genotype and per indicated age; *p < 0.01 and **p < 0.001 by two-tailed unpaired Student’s t test). (L) Kaplan-Meier survival plot of Ntg (n = 17) and Thbs1 Tg (n = 21) mice at the indicated ages. *p < 0.01, analyzed by two-tailed log-rank test. See also Figures S1 and S2.

Figure 3.. Thbs1 induces PERK, ATF4, and protein degradation in skeletal muscle

(A) Representative western blots for Thbs1, Thbs4, ATF6α-N (50 kDa, nuclear), BiP, calreticulin (Calret.), PERK, ATF4, IRE1α, fibroblast growth factor 21 (FGF21), LC3b, p62, LAMP1, LAMP2, and LIMPII from quadriceps protein extracts from Ntg and skeletal muscle-specific Thbs1 Tg, Thbs4 Tg, and ATF6α Tg mice at 6 weeks of age. Gapdh serves as a processing and loading control; n = 3 biologically independent animals per genotype. Red boxed areas show specific expression changes in Thbs1 Tg muscle. (B and C) Representative western blots for LC3b, p62, and Gapdh as a loading control in 6-week-old quadriceps protein extracts from Ntg and Thbs1 Tg mice treated with water as vehicle (Veh.) or 0.4 mg/kg/day colchicine (Colch.) for 2 days to impact autolysosome content/activity. (C) Quantitative analysis of LC3b-II protein levels relative to Gapdh from the experiment shown in (B). Error bars denote ±SEM from n = 3 biologically independent animals per genotype and per treatment. *p < 0.05, **p < 0.01, ***p < 0.001 by one-way ANOVA and Tukey’s multiple comparison test. (D) 20S chymotrypsin-like proteasome activity (pmol/[min·μg]) in quadriceps tissue of Ntg and skeletal muscle-specific Thbs1 Tg, Thbs4 Tg, and ATF6α Tg mice at 6 weeks of age. Error bars denote ±SEM from n = 4–7 biologically independent animals per indicated genotype. *p < 0.05, **p < 0.001 by one-way ANOVA and Tukey’s multiple comparison test. (E) Representative western blot for ubiquitin-conjugated proteins (Ubiq.) in quadriceps tissue of the indicated genotypes of mice at 6 weeks of age. Gapdh serves as loading control; n = 3 biologically independent animals per genotype. Quantitative analysis of this experiment is presented in Figure S4A. (F) Representative western blots for Thbs1, Thbs2, ATF6α-N (50 kDa, nuclear), PERK, ATF4, and LAMP2 from quadriceps protein extracts from Ntg and skeletal muscle-specific Thbs1 Tg and Thbs2 Tg mice at 12 weeks of age. Gapdh serves as loading control; n = 3 biologically independent animals per genotype. (G) Immunohistochemistry micrographs showing wheat germ agglutinin (WGA, green) and DAPI (blue) in quadriceps of Ntg, Thbs1 Tg, and Thbs2 Tg at 12 weeks of age. Scale bars represent 100 μm. (H) Myofiber cross-sectional area (CSA) in Ntg, Thbs1 Tg, and Thbs2 Tg quadriceps at 12 weeks of age as shown in (G). Error bars denote ±SEM from n = 4–5 biologically independent animals per genotype. ***p < 0.001 by one-way ANOVA and Tukey’s multiple comparison test. See also Figures S3 and S4.

Figure 4.. Loss of Eif2ak3 (PERK), Atf6, Cd36, or Cd47 does not rescue Thbs1-mediated skeletal muscle atrophy

(A) Breeding scheme to generate muscle-specific deletion of Eif2ak3 (PERK; referred to as Eif2ak3^mKO) in Thbs1 Tg mice using Eif2ak3-LoxP-targeted (Eif2ak3^fl/fl) mice crossed with Myl1-cre gene-targeted mice. (B) Western blots for Thbs1, PERK, ATF4, and FGF21 from quadriceps protein extracts from Eif2ak3^fl/fl, Eif2ak3^mKO, Thbs1 Tg Eif2ak3 ^fl/fl, and Thbs1 Tg Eif2ak3^mKO mice at 12 weeks of age. Gapdh serves as loading control; n = 3 biologically independent animals per genotype. (C) Representative Masson’s trichrome-stained histological sections of quadriceps from the indicated genotypes at 12 weeks of age. Representative images of six mice per genotype are shown. Scale bars represent 1 mm. (D–G) Quadriceps (Quad) weight normalized to tibia length of (D) Eif2ak3^fl/fl, Eif2ak3^mKO, Thbs1 Tg Eif2ak3^fl/fl, and Thbs1 Tg Eif2ak3^mKO mice, (E) wild-type (WT), Atf6^−/−, Thbs1 Tg, and Thbs1 Tg Atf6^−/− mice, (F) WT, Cd36^−/−, Thbs1 Tg, and Thbs1 Tg Cd36^−/− mice, and (G) WT, Cd47^−/−, Thbs1 Tg, and Thbs1 Tg Cd47^−/− mice at 12 weeks of age for (D), (F), and (G) and 8 weeks of age for (E). Error bars denote ±SEM. *p < 0.05, **p < 0.0001 by one-way ANOVA and Tukey’s multiple comparison test. See also Figure S5.

Figure 5.. Thbs1 activates TGFβ-Smad3-ATF4 signaling to facilitate skeletal muscle atrophy

(A) Log₂ fold changes of mRNA expression in TGFβ signaling-associated genes between 6-week-old Thbs1 Tg and Ntg control quadriceps assessed by RNA sequencing. Full RNA-sequencing data were deposited in the GEO (GEO: GSE245663). Gdf5, growth differentiation factor 5; Ltbp2, latent transforming growth factor β binding protein 2; Acvr1c, activating receptor type 1C; Gdf15, growth differentiation factor 15; Postn, periostin; Col10a1, collagen type 10 alpha 1; Fbn2, fibrillin 2, Col6a5, collagen type 6 alpha 5; Tgif1, TGFβ-induced factor homeobox 1; Loxl1, lysyl oxidase-like 1; Tgfb1, transforming growth factor β1. (B) Endogenous TGFβ activity levels in quadriceps protein extracts from Ntg, Thbs1 Tg, Thbs4 Tg, and ATF6α Tg mice at 6 weeks of age. Error bars denote ±SEM. *p < 0.05 by one-way ANOVA and Tukey’s multiple comparison test. (C and D) Representative western blots for phospho-Smad3 (Ser423/425) and total Smad3 in 6-week-old quadriceps protein extracts from the indicated skeletal muscle-specific transgenic mice. Gapdh serves as loading control; n = 3 biologically independent animals per genotype. (E) Schematic diagram of wild-type Thbs1 domain structure and the Thbs1Δt1 mutant lacking the type-1 repeat domain. Each wild-type Thbs1 monomer consists of an N-terminal laminin G (LamG) domain, a coiled-coil domain (cc), a procollagen homology domain (PC), type-1, −2, and −3 repeats (T1, T2 and T3), and a carboxy-terminal domain (CTD). (F) Representative western blot analysis for Thbs1, PERK, and ATF4 in gastrocnemius protein extracts from 16-day-old rat pups injected with Adβgal, AdThbs1, or AdThbs1Δt1 mutant on postnatal day 2, followed by a second “boost” injection 72 h later (3E8 viral particles per injection). Gapdh serves as loading control; n = 3 biologically independent animals per genotype. (G and H) Gastrocnemius (Gastroc) weight normalized to tibia length (TL; G) and myofiber cross-sectional area (CSA; H) from 16-day-old rat pups injected with Adβgal, AdThbs1, or AdThbs1Δt1 mutant following the timeline described for (F) (all 3E8 viral particles per injection). Error bars denote ±SEM from n = 6–8 biologically independent animals per treatment group. ***p < 0.001, ****p < 0.0001 by one-way ANOVA and Tukey’s multiple comparison test. (I) Western blots for Thbs1, dominant-negative TGFβ type II receptor (dnTGFβRII), phospho-Smad3 (Ser423/425), total Smad3, ATF4, and Gapdh control in 12-week-old quadriceps obtained from Ntg, Thbs1 Tg, skeletal muscle-specific Tg mice expressing dnTGFβRII (dnTGFβRII Tg), and Thbs1-dnTGFβRII double Tg (DTg) mice. n = 2 biologically independent animals per genotype. (J and K) Quadriceps (Quad) weight normalized to tibia length (J) and myofiber cross-sectional area (CSA; K) in Ntg, Thbs1 Tg, dnTGFβRII Tg, and Thbs1-dnTGFβRII double Tg (DTg) mice at 3 months of age. Error bars denote ±SEM from n = 4–7 biologically independent animals per genotype. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 by one-way ANOVA and Tukey’s multiple comparison test. See also Figures S6 and S7.

Figure 6.. Canonical TGFβ effector Smad2/3 and downstream ATF4 mediate Thbs1-induced muscle atrophy

(A) Representative western blot analysis of Thbs1, Smad3, Smad2, ATF4, FGF21, LC3b, p62, LAMP2, and LIMPII in 12-week-old quadriceps protein extracts from Smad2/3^fl/fl, Smad2/3^fl/fl-Myl1-Cre (Smad2/3^mKO), Thbs1 Tg Smad2/3^fl/fl, and Thbs1 Tg Smad2/3^mKO mice. Gapdh serves as loading control; n = 2 biologically independent animals per genotype. Red arrowhead indicates ATF4. (B and C) 20S chymotrypsin-like proteasome activity (pmol/[min·μg]; B) and quadriceps (Quad) weight normalized to tibia length (TL; C) from the indicated genotypes of mice at 12 weeks of age based on the color-coded legend shown on the right. Error bars denote ±SEM. The number of biologically independent animals per genotype are indicated in the graphs. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 by one-way ANOVA and Tukey’s multiple comparison test. (D) Western blots for phospho-Smad3 (Ser423/425), total Smad3, and ATF4 in gastrocnemius from 16-day-old rat pups injected with Adβgal or AdSmad3 on postnatal day 2, followed by a second “boost” injection 72 h later (3E8 viral particles per injection). Gapdh serves as loading control; n = 3 biologically independent animals per genotype. (E and F) Gastrocnemius (Gastroc) weight normalized to tibia length (TL) from 16-day-old rat pups injected with Adβgal or AdSmad3 (E) or injected with Adβgal, AdThbs1, AdSmad6, and/or AdSmad7 (F), following the timeline described for (D) (all 3E8 viral particles per injection). Error bars denote ±SEM from n = 5–6 biologically independent animals per treatment group. *p < 0.05 by two-tailed unpaired Student’s t test for (E); *p < 0.05, **p < 0.01, ***p < 0.001 by one-way ANOVA and Tukey’s multiple comparison test for (F). (G) Western blot analysis for Thbs1, ATF4, FGF21, LC3b, p62, LAMP2, and Gapdh in 12-week-old quadriceps protein extracts from Atf4^fl/fl, Atf4^fl/fl-Myl1-Cre (Atf4^mKO), Thbs1 Tg Atf4^fl/fl, and Thbs1 Tg Atf4^mKO mice; n = 2 biologically independent animals per genotype. (H and I) 20S chymotrypsin-like proteasome activity (pmol/[min·μg]; H) and quadriceps (Quad) weight normalized to tibia length (TL; I) from the indicated genotypes at 12 weeks of age. Error bars denote ±SEM from n = 3 biologically independent animals per genotype in (H) and n = 6–8 biologically independent animals per genotype for (I). *p < 0.05, **p < 0.01, ***p < 0.001, by one-way ANOVA and Tukey’s multiple comparison test. See also Figure S7.

Figure 7.. Loss of Thbs1 blunts muscle atrophy induced by denervation and fasting

(A) Representative western blot analysis of phospho-Smad2 (Ser465/467), total Smad2, ATF4, LC3b, p62, LAMP1, LAMP2, LIMPII, and Gapdh control in the tibialis anterior (TA) of wild-type (WT) and Thbs1^−/− mice subjected to 10 days of unilateral hindlimb denervation, compared to the sham-operated control contralateral TAs from the same mice. (B–E) Tibialis anterior (TA) weight normalized to tibia length (TL; B, D, and E) and TA myofiber cell surface area (CSA; C) after 10 days of denervation (Den.) in the indicated lines of mice, as compared to the non-denervated contralateral muscle (Ctl.) in the same mice. The legend to the left of (B) refers to all panels in the figure. Error bars denote ±SEM. The number of biologically independent animals per genotype are indicated in the graphs. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 by two-tailed unpaired Student’s t test. (F–J) Tibialis anterior (TA) weight normalized to tibia length (TL; F and I), TA myofiber cell surface area (CSA; G and J), and liver weight normalized to TL (H) from 8-week-old mice of indicated genotypes shown in the legend to the left of (B), which were fed ad libitum or fasted for 48 h. Error bars denote ±SEM. The number of biologically independent animals per genotype are indicated in the graphs. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001 by one-way ANOVA and Tukey’s multiple comparison test.