Lipid Dynamics due to Muscle Atrophy Induced by Immobilization

peroxide dismutase is aggregated and mitochondrial func-tion is impaired, thereby causing lipid changes. In other words, lipid metabolism is considered to significantly affect muscle atrophy 3 ） . models induced lipids insulin the accumulation of lipids is in insulin muscle atrophy ） 7 ） intramyocellular lipids neutral lipids, triacylglycerol diacylglycerol ） impaired insulin sensitivity and alter lipid metabolic genes 8, 9 ） and are associated with muscle atrophy. phosphatidylserine （ PS ） is increased with muscle atrophy when the gene expressing PS-decarboxylase is in mice 10 ） . only a few studies have accurately analyzed the Abstract: Muscle atrophy refers to skeletal muscle loss and dysfunction that affects glucose and lipid metabolism. Moreover, muscle atrophy is manifested in cancer, diabetes, and obesity. In this study, we focused on lipid metabolism during muscle atrophy. We observed that the gastrocnemius muscle was associated with significant atrophy with 8 days of immobilization of hind limb joints and that muscle atrophy occurred regardless of the muscle fiber type. Further, we performed lipid analyses using thin layer chromatography, liquid chromatography–mass spectrometry, and mass spectrometry imaging. Total amounts of triacylglycerol, phosphatidylserine, and sphingomyelin were found to be increased in the immobilized muscle. Additionally, we found that specific molecular species of phosphatidylserine, phosphatidylcholine, and sphingomyelin were increased by immobilization. Furthermore, the expression of adipose triglyceride lipase and the activity of cyclooxygenase-2 were significantly reduced by atrophy. From these results, it was revealed that lipid accumulation and metabolic changes in specific fatty acids occur during disuse muscle atrophy. The present study holds implications in validating preventive treatment strategies for muscle atrophy.


Introduction
Skeletal muscles comprise approximately 40 of the total body weight and play fundamental roles in exercise, posture control, and energy metabolism 1 . These muscles can reconstitute their mass by accelerating protein synthesis and disassembly. However, when this system is distorted to lose its mass, muscle atrophy occurs 2 . Muscle atrophy is associated with the pathogenesis of cancer, diabetes, obesity, and aging. Sarcopenia, which refers to the loss of muscle mass and strength with aging, is one of the most important hindrances in maintaining quality of life. Patients developing sarcopenia show abnormal lipid metabolism such as dyslipidemias; however, the underlying mechanism is not fully understood. Patients with amyotrophic lateral sclerosis ALS , which causes skeletal muscle decline and muscle atrophy through the progressive loss of motor neurons, are known to have severe muscle atrophy 3 . Moreover, it is considered that in ALS, mutant Cu/Zn su-peroxide dismutase is aggregated and mitochondrial function is impaired, thereby causing lipid changes. In other words, lipid metabolism is considered to significantly affect muscle atrophy 3 .
Previously, studies on muscle atrophy models induced by high-fat diet 4 , muscle immobilization 5 , and denervation 6 clarified that the accumulation of excess lipids is not just associated with muscle atrophy but insulin resistance as well it has been reported that the accumulation of excess lipids is involved in insulin resistance and muscle atrophy 7 . Of special interest is increased intramyocellular lipids comprising neutral lipids, such as triacylglycerol TG and diacylglycerol DG , which lead to impaired insulin sensitivity and alter lipid metabolic genes 8,9 and are associated with muscle atrophy. Reportedly, phosphatidylserine PS is increased with muscle atrophy when the gene expressing PS-decarboxylase is knocked out in mice 10 . However, only a few studies have accurately analyzed the molecular species of these lipids. In this study, we aimed to determine the lipid changes caused by muscle atrophy, thereby demonstrating their roles in muscle atrophy at the molecular species level.
There are four types of muscle fibers found in skeletal muscles, namely myosin heavy chain MyHC type I, type IIa slow-twitch , type IIx, and type IIb fast-twitch . These myofibers have different contractile and metabolic characteristics. The gastrocnemius muscle contains all of these myofiber types and exhibits their characteristic localization. We developed an experimental model based on 8 days of immobilization of one limb to pinpoint lipid changes in the muscles due to muscle atrophy. We compared lipid content between the immobilized limb right and non-immobilized limb left collected from the same rat.

Animals
Five-week-old male Sprague-Dawley rats Nihon SLC, Ltd. Shizuoka, Japan were housed in stainless steel wire mesh cages in a temperature-controlled room 24 1 under a 12/12 h light/dark cycle dark phase: 15:00-3:00 . After 7 days of acclimatization, 12 rats were subjected anesthetic immobilization using isoflurane; the right limb was splinted with a bent plastic plate cast with a wetted plaster 3M Soft Elastic Tape, width: 5 cm, 3M Japan, Tokyo and wrapped with a steel wire mesh to inhibit gnawing. After the 8-day immobilization period, the rats were sacrificed by decapitation and the gastrocnemius muscles were removed and weighed. This study was conducted in accordance with the ethical guidelines of the Utsunomiya University Animal Experimentation Committee Approval No. A20-0016 and in complete compliance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. All efforts were made to minimize the number of animals used and limit experimentation sufficient to produce reliable scientific information.

Cryosectioning and hematoxylin and eosin H&E
staining Serial cross-sections of the muscles 10 μm thick were produced using a cryostat CM 1950; Leica Microsystems, Wetzlar, Germany and mounted onto Matsunami adhesive coated slides Matsunami, Osaka, Japan for histochemical staining or indium-tin-oxide-coated glass slides Bruker Daltonics, Bremen, Germany for mass spectrometry MS imaging. For morphological observation, the sections were stained with hematoxylin and eosin H&E , followed by measuring the cross-sectional area CSA of each muscle fiber total, 600 .

Fluorescence staining and immunoassays
To differentiate between muscle fiber types, the sections were subjected to immunofluorescence staining using fiber type-specific antibodies. Briefly, tissue sections were fixed with 100 acetone and incubated with primary antibodies against MyHC type I #BA-F8 , MyHC type IIa #SC-71 , and MyHC type IIb #BF-F3 , all obtained from the Development Studies Hybridoma Bank, University of Iowa USA . Further, the sections were incubated with Alexa Fluor 350 #A21040 -, Alexa Fluor 488 #A21042 -, and Alexa Fluor 555 #A21127 -conjugated secondary antibodies Thermo Fisher Scientific K.K. Tokyo, JAPAN . In addition, unstained black muscle fibers were assigned as MyHC type IIx. Fluorescence signals were detected with a widefield fluorescence microscope BZ-9000; KEYENCE, Osaka, Japan .

Thin-layer chromatography TLC
Total lipids were extracted from the gastrocnemius muscle with chloroform:methanol 2:1, v/v ; lipid fractions were extracted by the Bligh and Dyer method as described previously 7 . Equal amounts of the extracts were manually applied to silica gel 60 high-performance TLC plates Merck, Darmstadt, Germany . The plates were developed with the solvent systems methyl acetate/1-propanol/chloroform/methanol/0.25 aqueous potassium chloride 25:25:25:10:9, v/v/v/v/v and n-hexane/diethyl ether/acetic acid 80:30:1, v/v/v for the separation of polar lipids and neutral lipids, respectively. The chromatograms were sprayed with primuline reagent, and lipid bands were visualized with a Gel Doc EZ Imager Bio-Rad Laboratories, CA, USA under ultraviolet light. Relative densities of each lipid were determined quantitatively using ImageJ software http://rsbweb.nih.gov/ij/ . For TLC imaging, lipid bands developed on the TLC plates were transferred to a polyvinyl difluoride PVDF membrane, as described previously 11 . TLC-imaging was performed using a TOF/TOF 5800 mass spectrometer AB SCIEX, Tokyo, Japan . Briefly, the transferred PVDF membranes were attached to the MALDI target plate and deposited matrix for ionization. Positive ion-mode analyses were applied to obtain the signal, and m/z values in the range of 400-1000 were measured. To create two-dimensional ion-density maps, we used Datacube Explorer AMOLF, Amsterdam, Netherlands . Lipids were identified by referring to a previous report and by tandem MS analysis.

Liquid chromatography-mass spectrometry LC-MS
analysis Lipid profiles were acquired with HPLC-electrospray ionization ESI -orbitrap MS LTQ Orbitrap Discovery, Thermo Scientific . The stored samples 80 were thawed at 23 1 and resolved with 200 μL chloroform: methanol 1:2, v/v ; 5 μL of each sample solution was applied to the LC-MS system. The samples were separated with a reverse phase C18 column 1.9 μm, 2.1 50 mm, JASCO Corporation, Tokyo, Japan . The mobile phase was composed of A acetonitrile:methanol:water 9:9:2, v/v/v containing 0.1 formic acid and 0.028 ammonia, and B isopropanol containing 0.1 formic acid and 0.028 ammonia. The gradient program was as follows: 0 min, 0 B; 5 min, 100 B; 10 min, 100 B; 10.1 min, 0 B; and 13 min, 0 B. The column temperature was maintained at 45 and the flow rate was set to 0.20 mL/min. To wash the flow channel including the column, 5 μL of mobile phase B was injected after each analysis and the same gradient program was adopted.
MS detection was performed in the positive ion mode. ESI source conditions were set as follows: source voltage and capillary voltage were maintained at 4.0 kV and 20 V, respectively; capillary temperature, sheath gas flow rate, and aux gas flow rate were set at 275 , 20 arb, and 0 arb, respectively. Data were collected in a data-independent top 5 scan mode. Surveyed full scan MS spectra were acquired with a resolution of R 30,000 in the mass range of m/z 200-1,000. For molecular identification, tandem mass spectrometry MS/MS was performed using collision-induced dissociation, whereby the collision energy was normalized at 35. An analysis of each sample was repeated three times.

Western blot analysis
Following tissue homogenization in lysis buffer and sonication, the lysate was centrifuged at 14,800 rpm for 40 min at 4 to pellet the debris. The total protein concentration of the supernatant was determined using a Bradford assay. The supernatant sample was stored at 30 .
SDS-PAGE analysis was performed as described previously 12 . Samples were diluted in Laemmli sample buffer and boiled for 10 min at 98 . Protein samples were separated by 10 SDS-PAGE and transferred onto PVDF membranes Millipore, Bedford, MA, USA . The membranes were blocked for 1 h in tris-buffered saline containing tween TBST; 20 mM Tris-HCl pH 7.6 , 137 mM NaCl, 0.1 tween-20 and 5 skim milk. Thereafter, membranes were incubated overnight at 4 with primary antibodies against ATGL #2138, Cell Signaling Technology, MA, USA and GAPDH #2708, Cell Signaling Technology, MA, USA . The next day, the membranes were washed four times with TBST and incubated with horseradish peroxidase HRPconjugated secondary antibodies anti-rabbit IgG: 1:1000, Cell Signaling Technology, MA, USA for 1 h at room temperature. The protein bands were detected using enhanced chemiluminescence Luminata Forte Western HRP Substrate; Merck Millipore, Bedford, MA, USA with an Im-ageQuant LAS 500 imaging system GE Healthcare, Amersham, UK . Quantification was performed using ImageJ software http://rsbweb.nih.gov/ij/ .

Spectrophotometric analysis of cyclooxygenase
COX activity COX activity was determined using a commercial kit COX Activity Assay Kit, Cayman Chemical, MI, USA . The gastrocnemius muscle was sonicated in 5-10 mL cold buffer i.e. 0.1 M tris-HCl, pH 7.8, containing 1 mM EDTA per g of tissue, followed by centrifugation at 10,000 g for 15 min at 4 . The supernatant was collected and used for the assay. Briefly, the sample supernatant , buffer, arachidonic acid AA , and hemin were placed in a 96-well plate. Samples with inhibitors including 5-bromo-2-4-fluorophenyl -3-4-methylsulfonyl phenyl -thiophene Dup-697 and 5-4-chlorophenyl -1-4-methoxyphenyl -3-trifluoromethylpyrazole SC-560 were also placed in other wells. Dup-697 and SC-560 were used as inhibitors of COX-2 and COX-1, respectively. Absorbance was measured at 560 nm Molecular Devices, CA, USA , and COX activity was calculated from the obtained absorbance.

MS imaging
MS imaging was performed using the SolariX XR Bruker Daltonics, Bremen, Germany . Analyses were performed for masses in the range of m/z 400-1200. The matrices for ionization used 40 mg/mL 2,5-dihydroxybenzoic acid in methanol/water 8:2, v/v for the positive ion mode. Ion images were constructed using flexImaging Bruker Daltonics, Bremen, Germany . The MS parameters were set to obtain the highest sensitivity. Cryosections as described in section 2.2, for MS imaging, were prepared from three independent rats n 3 .

Statistical analysis
All data are expressed as the mean standard error of the mean. Statistical analyses were performed using a Student s t-test with StatView 5.0 SAS Institute, Tokyo, Japan . The level of significance was set at p 0.05 * or p 0.01 ** .

Induction of muscle atrophy by immobilization
Previously, it was reported that the weights of the soleus and gastrocnemius muscle decrease after 2 weeks of immobilization of the hind limb joint 13 . In the present study, we tried to analyze the lipid changes in the early stage of muscle atrophy; hence, we immobilized the hind limb joint for 8 days using the same method Fig. 1A . The weight of the muscle collected from the right limb immobilized was reduced as compared to that of the left limb control; Fig.  1B . Further, immobilized muscle exhibited significantly reduced transverse section areas Figs. 1C and 1D . We concluded that even an 8-day immobilization procedure can significantly reduce skeletal muscle masses.

Analyses of muscle atrophy based on each ber type
We observed muscle atrophic features in the immobilized muscle Fig. 1 . Further, we divided the gastrocnemius muscle into three parts depending on the muscle fiber composition LGasS: lateral head gastrocnemius muscle of the surface; LGasD: lateral head gastrocnemius muscle of the deep layer; MGasD: medial head gastrocnemius muscle of the deep layer to determine the level of atrophy in each fiber type Fig. 2A and measure the CSA of each muscle fiber 14 . We found that muscle atrophy occurred regardless of the site and muscle fiber type and that all parts and all muscle fibers were significantly atrophied by 0.5 to 0.7 times in the immobilized group as compared to those of the control group Figs. 2B and 2C . More specifically, the average atrophy rates for each muscle fiber type were 0.53 to 0.59 times for slow-twitch fibers MyHC types I and IIa, respectively and 0.60 to 0.61 times for fast-twitch fiber MyHC types IIx and IIb, respectively .

Accumulation of neutral lipids by muscle atrophy
We next measured total lipid content, which did not differ between control and immobilized groups data not shown . To determine lipid changes during skeletal muscle atrophy in detail, we analyzed differences in lipid content/ type between control and immobilized muscles using TLC. Of the neutral lipids, TGs tended to increase in immobilized muscles Fig. 3A , which was similar to that observed in a previous report 5 . Therefore, we conducted LC-MS analyses and examined the molecular species to understand the quantitative and qualitative changes that occurred with immobilization. We found 24 kinds of TG molecules, wherein signal ratios of the top 10 molecular species are shown in Fig. 3B. As shown in Fig. 3B, all molecular species showed a significant increase in immobilized muscles. To assess the mechanism of TG accumulation, we measured protein expression levels of adipose triglyceride lipase ATGL , a major lipase that catalyzes TG to DG. The expression of ATGL was significantly reduced in immobilized muscles as compared to that in control muscles Fig. 3C .

Accumulation of polar lipids with muscle atrophy
A previous study reported that phosphatidylethanolamine PE and PS increased in patients with spinal muscular atrophy 15 . To elucidate if our atrophy model exhibited such a trend, we performed TLC for phospholipid separation. Consequently, we detected PE, phosphatidylinositol PI , PS, phosphatidylcholine PC , sphingomyelin SM , and lysophosphatidylcholine LPC Fig. 4A . Quantitative analysis revealed that there was an increasing tendency for PS p 0.0755 and a significant increase in SM p 0.0061 . To clarify the kinds of molecular species of PS that changed, we analyzed PS bands using TLC imaging. As a result, we found that the molecular ion of PS 18:0/18:1 was the main molecular species and tended to increase p 0.0528 in the immobilized muscle compared to that in the control muscle. Reportedly, PC 18:0/20:4 and SM  d18:1/24:1 were increased in a muscle atrophy model of a high-fat diet 7 . Hence, we analyzed molecular species of PC and SM in a similar manner. We found that AA-containing PC 18:0/20:4 and SM d18:1/24:1 were significantly increased in immobilized muscle as compared to levels in control muscle Fig. 4B .

COX activity with muscle atrophy
As seen in Fig. 4B, AA-containing phospholipids, such as PC 18:0/20:4 , were significantly increased during muscle atrophy, a finding that is in accordance with our previous report 7 . To determine the physiological significance of AA accumulation, we measured COX total, COX-1, and COX-2 activity in the control and immobilized gastrocnemius muscles. Consequently, we found no change in the activities of total COX and COX-1; however, COX-2 activity tended to be reduced in immobilized muscle Fig. 5 . Hence, we speculate that joint immobilization modulates metabolism of the AA cascade, which might prevent muscle hypertrophy via a COX-2-dependent pathway.

Speci c localization of PC
We finally performed MS imaging to visually pinpoint lipid localization. We selected three molecular species of PCs, namely PC 16  ubiquitously. We observed that PC 16:0/18:1 was predominantly localized to the surface area rather than the deep area of the gastrocnemius muscle. In contrast, PC 18:0/20:4 was mainly localized to the deep area Fig. 6 . These specific localization patterns might be related to lipid dynamics specific to muscle fiber types. Not only quantitative differences in lipid components, but also their differential localization, are essential to elucidate lipid dynamics.

Discussion
We previously demonstrated the dynamic lipid changes due to muscle atrophy induced by a high-fat diet, in addition to revealing the characteristic accumulation of lipids 7 . However, these models differed in terms of the intake of lipids, and it was very difficult to reveal lipid dynamics in the muscle. Therefore, we realized that we needed to develop another animal model to efficiently elucidate lipid dynamics in the muscle. Further, in our previous study, we developed a disuse muscle atrophy model via 2 weeks of immobilization of the animal s hind limb joints 13 . We observed a significant reduction 0.6-fold in the gastrocnemius muscle of this model. In the present study, we reduced the period of immobilization to 8 days to validate lipid changes involved in the early stages of muscle atrophy. Regardless of the short period of immobilization, the muscle weight and CSA of myofibers were significantly reduced in immobilized muscle as compared to those in controls. In various muscle atrophy models, the effect of atrophy has been reported to be muscle fiber-specific; for example, only MyHC type IIx is significantly reduced in quadriceps of patients with chronic obstructive pulmonary disease 16 and MyHC type IIx CSA is significantly reduced in the muscles of spinal cord injury models 17 . To pinpoint which of these muscle fibers exhibited atrophic features, we utilized multiple immunofluorescence studies. Interestingly, all muscle fibers of all types MyHC type I, IIa, IIx, and IIb were reduced approximately 0.5-0.7-fold in our model. Furthermore, some reports represented muscle fiber shift from slow to fast 18 and/or fast to slow 19 by muscle atrophy. In our model, the muscle fiber ratio did not change by immobilization data not shown . We thought that only 8 days were not sufficient to exhibit different protein expression pattern. Therefore, we concluded that the disuse atrophy model led to severe muscle atrophy, wherein all myofibers showed atrophic features.
Intramuscular TG accumulation is known to be involved in insulin resistance and obesity 8 ; it reportedly occurs in a muscle atrophy model 20 . Further, the amount of intramuscular TG was found to decrease after exercise, indicating its association with increased insulin sensitivity due to ex-  ercise 21 . In the present study, we observed that TG tended to increase with immobilization Fig. 3A , similar to that in a previous report 5 . Therefore, we suggest that the accumulation of TG is a typical feature of muscle atrophy. We know that lipid bioactivity differs depending upon the type of acylated fatty acid. For example, palmitoleic acid is involved in the activation of phosphatidylinositol-3 kinase signaling and contributes to an improvement in insulin resistance 22 . However, AA inhibits insulin signaling by stimulating the p38 Mitogen-activated Protein Kinase pathway 23 . Hence, we analyzed the lipids that were increased with respect to their molecular species. Our LC-MS analyses detected 24 kinds of TG molecular species, of which almost all were significantly increased, but no molecular specificity in the accumulated TGs was observed. Furthermore, we elucidated the mechanism underlying lipid accumulation by studying the expression level of ATGL. We found that ATGL was significantly reduced with immobilization Fig.  3C . Previous studies have reported that exercise training increases the expression of ATGL, which promotes the lipolysis of TGs 21 . We conclude that all TG molecular species were accumulated owing to suppressed ATGL expression triggered by inactivity and reduced energy metabolism. We consider that further investigations to determine the regulation of CGI-58 and G0S2, which modulate the expression of ATGL, are necessary. PS is a membrane-constituting lipid that is known to play an important role during muscle fusion 24 . Previous studies have shown that PS is increased in muscle atrophy models 15 . However, these reports did not distinguish the molecular species that increased. Subsequently, we analyzed molecular species of polar lipids that increased in the present model. Some reports suggested that PS and PSbinding proteins such as Stabilin-2 are very important for muscle fusion 24 , and we will evaluate the increase in PS molecular species that could bind to such proteins in future study.
In the same manner, we detected a specific increase in AA-containing PCs, for which results were consistent with those of the high-fat fed atrophy model 7 , and hence, we suggest that the increase in AA-containing PCs is one of the universal lipid changes that occurs due to muscle atrophy. Previous reports demonstrated that supplementation with AA to C2C12 cells promotes myoblast fusion and is associated with hypertrophic myofibers 25 . In our model, total intake of fatty acids was equal between control left and immobilized right limbs, and we realized that the increase in AA-containing PCs led to a decrease in free AA, which might be related to the suppression of cell proliferation.
A previous report found that AA supplementation contributes to the activation of COX-2 25 . COX catalyzes the synthesis of prostaglandin PG from AA, which is produced from phospholipids via the action of phospholipase A2 25 .
COX has two isotypes, namely COX-1 and COX-2. COX-1 is ubiquitously present and involved in cell homeostasis, whereas COX-2 is involved in hypertrophy in muscle 25 and is related to chronic exercise 26 . From previous studies, it is known that Prostaglandin E2 levels are increased in muscle atrophy models, such as rheumatoid arthritis. However, it has been confirmed that Prostaglandin F2a PGF2a levels increase with the increase in muscle protein synthesis due to exercise 27 . This suggests that PGF2a is involved in muscle hypertrophy, which is catalyzed by the COX-2 pathway. Our model investigating the early stage of immobilization revealed downregulation of the COX-2 pathway. We hypothesized that this phenomenon demonstrated the decrease in free AA in the immobilized limb. Alvarez et al. showed that inflammation induced by atrophy triggers the activation of COX-2 to promote proliferation via interleukin-1β and tumor necrosis factor-α 28 . We postulated that the activity of COX-2 might increase in the later phase of immobilization.
In addition, we observed the specific localization of PCs by MS imaging, since each PC molecule is distributed differently through the muscle, as described in our study previously 29 . We found that AA-containing PC, one of the lipid molecules, is significantly modulated by muscle atrophy, showing specific localization in the deep region that consists of slow-twitch fibers Fig. 6 . Hence, further studies are required to evaluate location-specific lipid changes due to muscle atrophy. Furthermore, muscle fiber shifts might occur in some muscle hypertrophy/atrophy models 30 . However, we acknowledge the limitation that in our model, only 8 days of immobilization was studied, which is not enough time to shift muscle fibers; however, monitoring the phospholipid composition of these myofibers might support the possibility of muscle fiber shifts with our short immobilization model.

Conclusion
We demonstrated that muscle atrophy develops with 8 days of immobilization of the limb, which was associated with significant changes in lipid dynamics, with the accumulation of all TGs and specific phospholipids. In summary, we revealed the accumulation of specific lipids and determined their changes at the molecular species level. Further, we suggest that a reduction in ATGL is related to TG accumulation and that the inhibition of COX-2 is related to the accumulation of AA-containing phospholipids and the downregulation of protein synthesis. Our study holds implications in validating preventive treatments for muscle atrophy.