Author Correction: Adiponectin promotes muscle regeneration through binding to T-cadherin

An amendment to this paper has been published and can be accessed via a link at the top of the paper.


Results
APN accumulation in regenerating muscle fibers. Firstly, we have characterized a human iliopsoas muscle autopsy specimen (Fig. 1). H&E staining results indicated almost intact muscle with tightly associated myofibers (Fig. 1A). Immunohistochemical studies showed that APN was found at the surface of myofibers and well colocalized with T-cad in these serial sections (Fig. 1B). Such accumulation of APN at the surface of myofibers was also observed in WT mouse tibialis anterior (TA) muscle, in which APN well colocalized with basal membrane marker, laminin α2 ( Fig. 1C upper panels). However, the accumulation of APN was not observed in T-cad KO mouse (Fig. 1C bottom panels), similarly in APN KO mouse (Fig. 1C middle panels), suggesting that such ectopic accumulation of APN other than its producing adipose tissues requires T-cad.
Next, we compared the protein amount of APN and T-cad in TA muscles during the process of skeletal muscle regeneration following cardiotoxin (CTX)-injury (Fig. 2). T-cad protein expression once decreased between day1 and day3 after CTX injection, and then it turned to increase to normal level after day7 ( Fig. 2A). The amount of APN protein in TA muscles also decreased after CTX injection and recovered to normal levels, in line with the change of T-cad ( Fig. 2A).
To examine the localization of APN in the regenerating muscle, TA muscles from WT were resected 7days after CTX injection and then subjected to immunofluorescence. APN accumulation significantly decreased in such regenerating muscle at day7 after CTX-injury in T-cad KO mice than in WT mice, indicating T-cad dependent mechanism for APN accumulation in myofibers (Fig. 2B). CTX injury is known to induce massive necrosis of myofibers. Upon necrosis, migrating neutrophils, and macrophages elicit inflammation 34 . This activates muscle stem cells, known as satellite cells residing under the basal lamina, to grow into regenerating myofibers 35 . Such actively regenerating myofibers can be distinguished from intact or already regenerated mature fibers by their  characteristic centrally located nuclei. We precisely investigated different phases of regenerating myofibers if APN might accumulate differentially (Figs 2C and S1 for higher magnifications). APN was found intracellularly in some endosome structures as well as at the cell surface in emerging crescentic myotubes in necrotic region with balloon like myofibers lacking nucleus ( Fig. 2C upper panels), and in actively regenerating myotubes with centrally located nuclei ( Fig. 2C middle panels), while APN was not found intracellularly but only at the surface of myofibers in intact fibers with peripherally located nuclei ( Fig. 2C bottom panels). In a time-course experiment, T-cad at the surface of myofibers in intact muscle was almost lost at day3 after CTX-injury. Regenerating myofibers which appeared at day5 and grew at day7 had intracellularly distributed T-cad in some endosome structures as well as in cell surface (Fig. 2D). Immunofluorescence colocalization experiments indicated that APN highly colocalized with a multivesicular body/exosome marker CD63 (Fig. 2E), suggesting endocytosis of APN into CD63-positive endosomes.
Effect of APN on muscle regeneration. Next, we have tested the effect of APN on muscle regeneration in mice. We administered APN-expressing adenovirus intravenously. Four-days after infection, CTX was injected into mouse TA muscles to induce muscle regeneration process (Fig. 3A). APN overexpression successfully increased serum APN concentrations (Fig. S2A). Although necrotic region area (%) in total muscle section area showed tendency of reduction by APN-overexpression after 7 days of CTX-injury (p = 0.06), myofiber area (%) in total muscle section area was not changed compared to control β-gal-overexpression group (Fig. 3B). Since C57BL/6J mouse strain is known to have a very high regenerating capacity, we performed the experiment under angiotensin II (AII)-infusion using implanted osmotic pump, which can partially mimic aging 36 and chronic heart failure condition ( Fig. 3A) 11,12 . AII-infused mice showed significant reduction of the ratio of myofiber area (%) in response to CTX-injury ( Fig. 3B), showing that AII impaired regenerating capacity. Under such suppressed regenerating condition, APN overexpression significantly increased the myofiber area (%) and decreased remaining necrotic region area (%), compared to those of β-gal-overexpression ( ) and not significantly altered by APN overexpression (Fig. 3D), suggesting that APN did not attenuate AII signaling itself. Canonical AII-signaling including phosphorylated ERK, p38MAPK, p65NFκB, and Nox2 protein was not significantly altered by AII-infusion nor by APN ( Fig. S2C-F).

Effect of APN overexpression on muscle regeneration in genetic absence of APN or T-cad.
Next, we conducted the same AII-loaded CTX-induced muscle regeneration experiment in APN KO mice and T-cad KO mice (Fig. 4). Loss of APN or T-cad did not significantly affect muscle regeneration, when we compared myofiber area (%) and necrotic area (%) between β-gal overexpressing WT mice (Fig. 3B) and APN KO mice (Fig. 4A) or T-cad KO mice (Fig. 4B). APN overexpression significantly improved muscle regeneration in APN KO mice, as judged by enhanced myofiber area and decreased necrotic region area (%) (Fig. 4A). Importantly, such improvement was not observed in T-cad KO mice, demonstrating the importance of T-cad in APN-mediated effect on muscle regeneration (Fig. 4B).

Differentiation dependent intracellular localization of APN in C2C12.
To investigate APN accumulation more precisely, we examined the expression of T-cad protein during the differentiation process of C2C12 myocytes (Fig. 5A). T-cad was little expressed in undifferentiated myoblasts and gradually increased along with their differentiation into myotubes (Fig. 5A). Purified high-molecular multimeric APN, included in differentiation medium, accumulated in each stage of differentiation in accordance with the expression levels of T-cad ( Fig. 5A). In agreement with such differentiation-dependent accumulation, myotubes differentiation accompanied enhanced exosome production by APN (Fig. 5B). Next, knockdown of T-cad significantly decreased APN accumulation in differentiating myotubes (Fig. 5C), while knockdown of AdipoR1 and AdipoR2 did not affect it (Fig. 5D). Furthermore, knockdown of T-cad attenuated APN-mediated increase of exosome secretion (Fig. 5E). In accordance with above quantitative studies, little accumulation of APN in undifferentiated myoblasts was seen with immunofluorescence ( Fig. 6A). After starting differentiation, however, APN was selectively found in tubular myocytes undergoing differentiation (Fig. 6B). In such differentiating myocytes, APN was mainly recognized as dotted structures, highly colocalizing with exosome/multivesicular body marker CD63 (Fig. 6B, bottom panels). After 5 days of differentiation, differentiated myocytes assembled and fused into multinuclear bundles, large myotubes. Again, APN accumulated inside of differentiating myocytes (arrow heads in Fig. 6C), even in such late period of differentiation. Knockdown of T-cad but not of AdipoR1/AdipoR2 strongly attenuated such intracellular distribution of APN in differentiating myocytes (Fig. 6D).

Discussion
Our study addressed the impact of high levels of adiponectin (APN) on injury-induced muscle regeneration and the role of T-cad mediating the function of APN. APN overexpression in mice decreased necrotic region and increased regenerating myotubes in angiotensin II (AII)-infusion mice. Such enhanced regeneration by excess APN was also observed in APN null mice, but not in T-cad null mice. T-cad was decreased by muscle injury and gradually restored with muscle regeneration. The ectopic accumulation of APN in muscle well correlated with this. APN accumulated on plasma membrane of myofibers both in mice and human. Importantly, APN accumulated in endosome-like structures positive for CD63 inside of regenerating myotubes but not in unaffected intact myofibers nor in necrotic ones. Such intracellular localization specifically in regenerating myofibers coincided with the changes of T-cad localization during regeneration. Purified high-molecular multimeric APN similarly   accumulated intracellularly and colocalized with a multivesicular bodies/exosomes marker CD63 in differentiating myocytes but not undifferentiated myocytes. In agreement with these differentiation-dependent endosomal accumulation, APN-mediated exosome production was also differentiation dependent and attenuated by T-cad knockdown experiment. It was reported that elevated circulating APN levels were an independent markers of both myocardial infarction (MI) and all-cause mortality in male patients undergoing coronary angiography 37 . Such inverse correlation with circulating APN levels has been also reported on muscle weakness in elderly people 31,32 , muscle mass and strength in elderly heart failure patients 33 . Here we have revealed that high circulating APN levels in mice were not inhibitory but promotive for muscle regeneration at least under high angiotensin II levels, suggesting that above mentioned higher circulating APN in elderly patients especially with heart failure may not be causative for their associated muscle weakness and loss of muscle mass. Our conclusion agrees with the notions obtained in Duchenne muscular dystrophy model mice, where overexpression of APN decreased muscle damage 38 and loss of APN caused lower muscle force/endurance 39 . It was reported that APN levels are positively associated with age, even after adjustment for visceral adiposity 40,41 . Regulation of circulating level of APN has not been fully understood. A decrease in its clearance in the kidney may be the cause of high levels of APN in the elderly 42,43 . Circulating APN decreases in visceral adiposity and insulin-resistance, likely because of decreased production and secretion from white adipose tissues. Conversely, leanness is associated with higher circulating APN. The production from increased bone marrow adipose tissue in addition to the increased production of APN from white adipose tissue contributes to hyperadiponectinemia in leanness 44 . Bone marrow adipose tissue increases with aging and may account for higher level of APN in the elderly. We and others recently reported that loss of T-cad in mice resulted in more than three-fold increase of plasma APN 19,21 . SNP around T-cad gene was reported to associate with plasma APN levels in human beings [22][23][24][25][26][27] . Thus, if the T-cad expression is attenuated by aging, it may contribute to higher APN levels in elderly. In our study, loss of APN did not significantly affect muscle regeneration, suggesting physiological level of APN has little role on such short-term regeneration reaction. We employed adenovirus-mediated overexpression of APN to delineate whether experimentally high APN (>10-fold increase) is deleterious or promotive for muscle regeneration in mice. The results suggested that high APN is promotive for muscle regeneration through T-cad.
Here we employed AII-infusion to mimic aging and chronic heart failure condition 11,12,36 . This treatment attenuated regeneration, otherwise it would be difficult to see the improvement of muscle regeneration in C57BL/6J mice, which have a better regeneration potential 45 . In agreement with those notions, AII-infusion, also in this study, up-regulated Axin2 expression, a downstream gene of the aging-promoting Wnt/β-catenin signaling pathway 36 . APN overexpression did not attenuate Axin2 expression, suggesting it had no significant effect on this signaling. However, AII-signaling on capillary endothelium in muscle tissue could be significantly attenuated by APN as we reported on cardiac tissue in AII induced cardiac hypertrophy model 46,47 . Along with this context, our current study has not evaluated the importance of angiogenesis after muscle injury. It was reported that APN promoted angiogenesis in a model of hind limb ischemia 20 . Because angiogenesis is also important for tissue repairing, our study cannot exclude that angiogenesis enhancement by APN resulted in faster muscle regeneration.
T-cad is abundantly expressed in skeletal muscle in addition to vascular endothelium and heart. We identified regeneration-induced unique intracellular localization of APN specifically in regenerating myofibers and colocalization with CD63, an exosome and multivesicular body (MVB) marker. We recently reported that APN accumulated in multivesicular bodies and stimulated exosome biogenesis, which accompanied excretion of unnecessary or harmful materials such as ceramide 30 . APN rescued the impaired regeneration caused by AII infusion. This required the presence of T-cad. Muscle regeneration accompanied the change of distribution of APN from cell surface into CD63-positive endosomes, which possibly reflects its accumulation into MVBs. MVBs are the endosomes producing exosomes. We recently reported that APN accumulated in MVBs in cultured endothelial cells and in vivo aortic endothelial cells 30 . Collectively, our study clearly indicated that T-cad mediated the promoting function of APN in muscle regeneration, and T-cad in regenerating myofibers had some roles in this process.

Methods
Study approval. Human iliopsoas muscle autopsy specimens were obtained after written informed consent for postmortem investigation was obtained from the subject prior to his or her death or afterward from the subject's relatives. Angiotensin II (AII) was infused as described previously 47 . Briefly, AII (Sigma) was dissolved in 0.01 M acetic acid. Mice were anesthetized and implanted osmotic minipumps (Alzet TM mini-osmotic pump model 2002, Durect Corp.) containing 2.4 mg/kg/day of AII (Sigma) dissolved in 0.01 M acetic acid, in midscapular region of mice at 12-16 weeks of age. To induce muscle regeneration, 50 µL of 10 µM cardiotoxin (CTX) (Latoxan) was injected to each tibialis anterior (TA) muscle. Supplementation of APN was performed as described previously 47 . Adenovirus expressing the full-length mouse APN (Ad-APN) or the β-galactosidase (Ad-βgal) was prepared using the Adenovirus Standard Purification Virakit TM (Virapur). Then, they were injected to mice via the tail vein at 4 days before cardiotoxin injection and pump implantation. At sacrifice, mice were anesthetized by intraperitoneal injection of medetomidine (0.3 mg/kg body weight), midazolam (4 mg/kg body weight), and butorphanol (5 mg/kg body weight) and transcardially perfused with cold saline to wash out circulating APN. Serum APN levels were measured by ELISA (Otsuka Pharmaceutical Co.).

Exosome preparation.
Exosome isolation from the cell culture supernatant was performed as described previously 30 , with some modifications. Briefly, differentiated C2C12 cells were cultured with DMEM containing indicated concentration of exosome-free FBS for 48 hours. Then, the conditioned medium was collected and centrifuged at 800 × g for 10 minutes to deplete floating cells, and at 10,000 × g for 30 minutes to remove cell debris. For exosome isolation, the supernatant was ultracentrifuged at average 110,000 × g for 2 hours, followed by a washing step of the exosome pellet with Dulbecco's phosphate-buffered saline with calcium and magnesium (PBS(+)) at average 110,000 × g for 2 hours (TLA100.1 rotor, Beckman Coulter). The exosome pellets were directly solubilized in Laemmli sample buffer. For comparative analysis, exosomes were collected from equivalent amounts of culture medium, conditioned by equivalent numbers of cells. Western blotting. Exosomes or whole cell lysates were loaded onto 4-20% gradient SDS-PAGE gels (Bio-Rad) and transferred onto nitrocellulose membranes. The membranes were blocked with Block-One TM blocking reagent (Nacalai Tesque) and then incubated with primary antibodies using Can Get Signal TM solution 1 (TOYOBO) overnight at 4 °C and followed by incubation with secondary antibodies conjugated with HRP using Can Get Signal TM solution 2 (TOYOBO) for 60 minutes at room temperature. The following primary antibodies were used: goat polyclonal anti-adiponectin (AF1119, R&D); goat polyclonal anti-T-cadherin (AF3264, R&D); rabbit monoclonal anti-α-tubulin (11H10, Cell Signaling); rat monoclonal anti-mouse CD63 (clone R5G2, MBL); rabbit polyclonal anti-syntenin (ab19903, abcam); rabbit polyclonal anti-Tsg101 (ab125611, abcam); goat polyclonal anti-MFG-E8 (AF2805, R&D); mouse monoclonal anti-Alix (ab117600, abcam); rabbit polyclonal anti-Nox2/gp91 phox (ab80508, abcam). CD63 was detected under non-reducing conditions. Chemiluminescence signals developed with Chemi-Lumi One Super TM (Nacalai Tesque) were visualized by ChemiDoc Touch TM and quantitated using Image Lab software (Bio-Rad).

Muscle fixation and histological analysis.
Formalin-fixed, paraffin-embedded human autopsy specimens were deparaffinized, sectioned (2-μm thick) using a cryostat (Leica Microsystems) and stained with H&E. Mouse tibialis anterior (TA) muscles were resected and frozen in isopentane cooled in liquid nitrogen. Frozen tissue was sectioned (10-μm thick) and stained with H&E. Myofiber area and necrotic area were quantified with a BZ-X700 microscope and its software (Keyence), according to the following procedure. For myofiber area quantification, we isolated eosin red stained regenerating myofibers from whole sectional image, using color extraction function of BZ-X analyzer. Then, the ratio of the obtained myofiber area to total sectional area was calculated. For necrotic area analyses, we first defined necrotic area as containing more than three clustering of necrotic fibers. Necrotic fibers can be identified as round-shaped, eosin stained myofibers with peripherally located nuclei. They can accompany crescent-shaped, centrally nucleated, regenerating myocytes emerging around them. They are distinguished from intact fibers by surrounding infiltrating immunocytes or proliferating interstitial tissues. Also using BZ-X analyzer, above defined necrotic areas are extracted from whole sectional image. Then, the ratio of the obtained necrotic area to total sectional area was calculated.