Regulation of autophagy and the ubiquitin-proteasome system by the FoxO transcriptional network during muscle atrophy

Abstract

Stresses like low nutrients, systemic inflammation, cancer or infections provoke a catabolic state characterized by enhanced muscle proteolysis and amino acid release to sustain liver gluconeogenesis and tissue protein synthesis. These conditions activate the family of Forkhead Box (Fox) O transcription factors. Here we report that muscle-specific deletion of FoxO members protects from muscle loss as a result of the role of FoxOs in the induction of autophagy-lysosome and ubiquitin-proteasome systems. Notably, in the setting of low nutrient signalling, we demonstrate that FoxOs are required for Akt activity but not for mTOR signalling. FoxOs control several stress-response pathways such as the unfolded protein response, ROS detoxification, DNA repair and translation. Finally, we identify FoxO-dependent ubiquitin ligases including MUSA1 and a previously uncharacterised ligase termed SMART (Specific of Muscle Atrophy and Regulated by Transcription). Our findings underscore the central function of FoxOs in coordinating a variety of stress-response genes during catabolic conditions.

Figure 1. Deletion of FoxOs is permissive for normal muscle function.

(a) PCR analysis with genomic DNA from FoxO1,3,4^f/f and FoxO1,3,4^−/− gastrocnemius muscles. (b) FoxO1, FoxO3 and FoxO4 mRNA expression were quantified by RT–PCR in Tibialis Anterior (TA) muscle of FoxO1,3,4^−/− and control mice. n=4 each group. (c) Immunoblot showing reduction of FoxO1 and FoxO3 proteins in homogenates of FoxO1,3,4^−/− gastrocnemius muscles. Data are representative of three independent experiments. (d) Haematoxylin and eosin (Scale bar, 100 μm), (e) SDH (Scale bar, 1 mm) and (f) PAS staining (Scale bar, 1 mm) showing normal morphology, fibre type and glycogen of FoxO1,3,4^−/− gastrocnemius muscle. (g) SDS–PAGE and (h) immunohistochemistry analysis of myosin heavy chain type I, IIA, IIB and IIX proteins in gastrocnemius muscles showing no differences between FoxO1,3,4^−/− and FoxO1,3,4^f/f mice. Data are representative of three independent experiments. (i) Frequency histograms showing the distribution of cross-sectional areas (μm²) in TA of FoxO1,3,4^f/f (white bars) and FoxO1,3,4^−/− (black bars) fibres, n=4, each group. Data are shown as mean±s.e.m. Error bars indicate s.e.m. **P<0.01 (Student's t-test).

Figure 2. Deletion of FoxOs prevents muscle loss and weakness during fasting.

(a) Frequency histograms of cross-sectional areas (μm²) of FoxO1,3,4^f/f (black bars) and FoxO1,3,4^−/− (magenta bars) fibres in fed (upper panel) and fasted (lower panel) conditions, n=4, each group. (b) Force measurements preformed in vivo on gastrocnemius showed that FoxO1,3,4^−/− muscles preserve maximal tetanic force after fasting; n=6 muscles in each group. (c) Force/frequency curve of starved gastrocnemius muscle underlines the important protection achieved by the absence of FoxOs; n=6 muscles in each group. (d) Immunoblot of protein extracts from gastrocnemius muscles. Phosphorylation of AKT and S6 is reduced in fed and starved FoxO1,3,4^−/− muscles when compared with controls. Data are representative of three independent experiments. (e) Immunoblot analysis of p62 and LC3 in homogenates of gastrocnemius muscles from fed and starved FoxO1,3,4^−/− or controls. Fasting did not induce LC3 lipidation and p62 upregulation in FoxO-deficient muscles. Data are representative of three independent experiments. (f) Quantification of GFP–LC3-positive vesicles in FoxO1,3,4^f/f and FoxO1,3,4^−/− TA muscles; n=4 muscles in each group (g) Autophagy flux is not increased in FoxO-deficient TA muscles. Inhibition of autophagy–lysosome fusion by colchicine treatment induces accumulation of LC3II band in starved control but not in starved FoxO1,3,4^−/− muscles. Upper panel: immunoblot analysis of gastrocnemius homogenates. Lower panel: quantification of LC3 lipidation. n=4 muscles in each group (h) The scheme shows the overlap between FoxO-dependent genes, identified by gene expression profiling of fed (n=4) and fasted (n=4) muscles from FoxO1,3,4^f/f and FoxO1,3,4^−/− and atrophy-related genes or atrogenes. The data in the graphs are shown as mean±s.e.m. Error bars indicate s.e.m. *P<0.05, **P<0.01 (Student's t-test).

Figure 3. Absence of FoxOs prevent the induction of critical atrogenes.

Quantitative RT–PCR of atrogenes from fed and 24-h starved tibialis anterior of control and FoxO1,3,4^−/− mice. Data are normalized to GAPDH and expressed as fold increase of control-fed animals. n=4 muscles in each group Values are mean±s.e.m. *P<0.05,**P<0.01. (Student's t-test).

Figure 4. Acute inhibition of FoxOs phenocopies the conditional FoxO1,3,4 knockout.

(a) Frequency histograms of gastrocnemius muscles showing the distribution of cross-sectional areas (μm²) of inducible muscle-specific FoxO1,3,4 mice after tamoxifen-dependent deletion of FoxO1,3,4 genes (FoxO1,3,4^f/f: black bars and FoxO1,3,4^−/−: magenta bars) in fed (upper panel) and fasted (lower panel) conditions, n=3, each group. (b) Force measurements preformed in vivo on gastrocnemius muscle showed that acute inhibition of FoxOs in adulthood prevents force drop during fasting. n=6 muscles in each group. Freq: Frequency. (c) Left, immunoblotting analyses of gastrocnemius homogenates after acute deletion of FoxO1,3,4^−/− and controls. Right, quantification of LC3 lipidation. Data are representative of three independent experiments. (d) Autophagy flux is not increased in FoxO-deficient TA muscles. Inhibition of autophagy–lysosome fusion by colchicine treatment induces accumulation of LC3II band in starved control but not in starved FoxO1,3,4^−/− muscles. Left, immunoblotting analyses of gastrocnemius homogenates. Right, quantification of LC3 lipidation. (e,f) ChIP quantitative RT–PCR shows the recruitment of FoxO3 and FoxO1 on the promoters of selected atrophy-related genes. ChIP assays were performed in starved control and FoxO1,3,4^−/− TA muscles. IgG was used as the reference. n=3 for each group. Data are shown as mean±s.e.m. Error bars indicate s.e.m. *P<0.05, **P<0.01 (Student's t-test). S1: FoxO binding site 1; S2: FoxO binding site 2.

Figure 5. Deletion of FoxOs in skeletal muscle partially prevents atrophy during denervation.

(a) Frequency histograms of gastrocnemius muscles from FoxO1,3,4^−/− and control mice showing the distribution of cross-sectional areas (μm²) of FoxO1,3,4^f/f (black bars) and FoxO1,3,4^−/− (magenta bars) in control (upper panel) or in denervation (lower panel). n=4 muscles each groups. (b) Force measurements preformed ex vivo on soleus muscles show that FoxO1,3,4^−/− muscles are stronger than controls, both in basal condition and after 14 days from denervation. n=6 muscles in each group. (c) Force/frequency curves of denervated soleus highlight the higher strength generated by FoxO1,3,4^−/− muscles when compared with controls. n=6 muscles in each group. (d) Immunoblots of gastrocnemius protein extracts reveal a decrease of AKT phosphorylation both in contralateral and in denervated muscles of FoxO1,3,4^−/− mice. The increase of 4EBP1 protein is blunted in FoxOs knockout mice. (e) Immunoblots of autophagy-related proteins. FoxOs are required for p62 induction, while LC3 is less lipidated after 3 days of denervation. Data are representative of three independent experiments. Data are shown as mean±s.e.m. Error bars indicate s.e.m. *P<0.05, **P<0.01 (Student's t-test). C, control; D, denervated.

Figure 6. FoxOs are required for expression of several atrogenes after denervation.

Quantitative RT–PCR of the indicated atrogenes after 3 days from denervation. Data are normalized to GAPDH and expressed as fold increase of control innervated muscles. Values are mean±s.e.m. *P<0,05, **P<0.01 (Student's t-test).; cont, control; den, denervated.

Figure 7. FoxO members are redundant and control a new set of ubiquitin ligases.

(a–c) Frequency histograms showing the distribution of cross-sectional areas (μm²) of gastrocnemius muscles from muscle-specific (a) FoxO1^−/−, (b) FoxO3^−/− and (c) FoxO4^−/− mice. Data are shown as mean±s.e.m. of four muscles each group. *P<0.05, **P<0.01. (d,e) qRT–PCR of the novel ubiquitin ligases MUSA1, Fbxo21/SMART, Fbxo31, Itch from 24 starved (d) or denervated (e) FoxO1,3,4^f/f and FoxO1,3,4^−/− mice. Data are normalized to GAPDH and expressed as fold increase of fed control mice. n=4 muscles for each group. (f) ChIP qPCR of FoxO3 and (g) FoxO1 on the promoters of MUSA1, Fbxo21/SMART and Itch. IgG was used as the reference. n=3. (h–j) qRT–PCR of MUSA1, Fbxo21/SMART, Fbxo31, Itch in (h) FoxO1, (i) FoxO3 and (j) FoxO4 knockout mice after 3 days of denervation. Values are normalized to GAPDH and expressed as fold increase of control mice. n=4 muscles for each group. Data are shown as mean±s.e.m. Error bars indicate s.e.m. *P<0.05, **P<0.01 (Student's t-test). con control; den, denervated. MUSA1 S1, FoxO-binding site1; MUSA1 S2, FoxO-binding site2; MUSA1 S3, FoxO-binding site3.

Figure 8. Smart is a novel ubiquitin ligase required for denervation-dependent atrophy.

(a) Co-immunoprecipitation experiment showing that SMART is a F-box protein that forms a SCF complex. C2C12 muscle cell lines were transfected with SMART, Skp1, Cul1 and Roc1 expression plasmids. After 24 h, cells were lysed and immunoprecipitation against FLAG-tag or control IgG was performed. Western blots for the different SCF components are shown. (b) RNAi-mediated knockdown of SMART revealed by quantitative RT–PCR (qRT–PCR). Adult TA muscles were transfected with bicistronic expressing vectors that encode either oligo 4 or scramble and GFP. Two weeks later TA muscles were collected, RNA extracted and endogenous SMART, MUSA1, Atrogin1, MuRF1 and Fbxo31 expression were analysed by qRT–PCR, n=4. (c) Inhibition of SMART prevents muscle atrophy in denervated muscles. Adult muscle fibres were co-transfected with bicistronic expressing vectors that encode shRNAs against SMART (oligo 4) or scramble and GFP and denervated. Two weeks later cross-sectional area of transfected fibres, identified by GFP fluorescence, was measured. n=6 muscles for each group. (d) Densitometric quantification of polyubiquitinated proteins in muscle extracts transfected with shRNAi against SMART or scramble. Values are normalized to GAPDH and expressed as fold increase of fed control mice. n=3 muscles for each group. Data are shown as mean±s.e.m. Error bars indicate s.e.m. *P<0.05, **P<0.01 (Student's t-test). con, control; den, denervated; IB, immunoblotting.