Insights into lipid accumulation in skeletal muscle in dysferlin-deficient mice[S]

Loss of dysferlin (DYSF) protein in humans results in limb-girdle muscular dystrophy 2B, characterized by progressive loss of muscles in the distal limbs with impaired locomotion. The DYSF-null (Bla/J) mouse develops severe steatotic muscles upon aging. Here, we report a marked increase in adipocytes, especially in the psoas and gluteus muscles but not in the soleus and tibialis anterior muscles in aged Bla/J mice compared with WT mice. There was a robust upregulation in the mRNA expression of enzymes involved in lipogenesis and triacylglycerol (TAG) synthesis pathways in the steatotic skeletal muscles. Lipidomic analysis of the steatotic skeletal muscles revealed an increase in several molecular species of TAG, although it is unclear whether it was at the expense of phosphatidylcholine and phosphatidylserine. The adipocytes in steatotic muscles were extramyocellular, as determined by the increased expression of caveolin 1 (a cellular marker for adipocytes) and lipid-droplet protein, perilipin 1. This increase in adipocytes occured as a consequence of the loss of myocytes.

such as fibrofatty replacement and inflammatory infiltrates with mononuclear cells. Affected patients have complete loss of DYSF protein in the skeletal muscle and present with significant ambulatory discomfort by 6 months of age (4). A defective DYSF protein results in abnormally high accumulation of lipid droplets in the muscle, seen both by Oil-Red-O (ORO) staining and by electron microscopy (5). This seems to be a feature of DYSF-deficient muscles, which is generally not observed in other types of muscular dystrophies. Previously, DYSF-deficient Bla/J mice were evaluated for muscle phenotype at 6-12 months of age using a noninvasive MRI technique. From 9 months onwards, these mice progressively show the presence of fat and a decrease in muscle volume, but primarily in the gluteus and psoas muscles, and a progressive loss of motor function by 12 months of age (6). These mice also show gait abnormalities when put on a treadmill (6). However, the mechanisms responsible for high lipid accumulation in DYSF-deficient muscles remain unclear.
DYSF is a type II transmembrane protein that localizes to the periphery of the muscle fibers. While DYSF is an important regulator of vesicle fusion at the sarcolemma and plays an essential role in muscle repair, it also plays a role in vesicle trafficking, endocytosis, membrane receptor recycling, membrane turnover, muscle regeneration, and T-tubule formation (7).
To further study the molecular mechanisms and pathways responsible for excessive muscle lipid accumulation in aged Bla/J mice, we studied six different skeletal muscles involved in locomotion [psoas, gluteus, quadriceps, gastrocnemius, soleus, and tibialis anterior (TA)] for the presence of lipids and markers of lipid synthesis.

Abstract Loss of dysferlin (DYSF) protein in humans results
in limb-girdle muscular dystrophy 2B, characterized by progressive loss of muscles in the distal limbs with impaired locomotion. The DYSF-null (Bla/J) mouse develops severe steatotic muscles upon aging. Here, we report a marked increase in adipocytes, especially in the psoas and gluteus muscles but not in the soleus and tibialis anterior muscles in aged Bla/J mice compared with WT mice. There was a robust upregulation in the mRNA expression of enzymes involved in lipogenesis and triacylglycerol (TAG) synthesis pathways in the steatotic skeletal muscles. Lipidomic analysis of the steatotic skeletal muscles revealed an increase in several molecular species of TAG, although it is unclear whether it was at the expense of phosphatidylcholine and phosphatidylserine. The adipocytes in steatotic muscles were extramyocellular, as determined by the increased expression of caveolin 1 (a cellular marker for adipocytes) and lipid-droplet protein, perilipin 1. This increase in adipocytes occured as a consequence of the loss of myocytes. - Supplementary key words limb-girdle muscular dystrophy • adipose tissue • lipidomic • extra-myocellular adipocytes • Bla/J mice The limb-girdle muscular dystrophies (LGMDs) are rare heterogeneous genetic disorders characterized by progressive loss of muscle tissue and weakness. Affected patients report weakness in the muscles of the hip girdle, thighs, shoulder girdle, and proximal arms (1). Mutations in the dysferlin (Dysf) gene (2) result in three different forms of autosomal recessive muscular dystrophies: LGMD2B, Miyoshi myopathy, and distal anterior compartment myopathies (3). Patients have marked elevations in serum creatine kinase levels, and muscle biopsies reveal dystrophic features

Transmission electron microscopy
Perfused muscle tissues were further fixed in osmium tetroxide (OsO 4 ), embedded in resin, and sectioned into thick (350 nm) sections for toluidine blue staining and thin (50 nm) sections for transmission electron microscopy (TEM) imaging. Tissues were post fixed in buffered 1% OsO 4 , stained in 4% uranyl acetate, dehydrated with a graded series of ethanol, and embedded in EMbed-812 resin. Thin sections were cut on a Leica Ultracut UCT ultramicrotome and stained with 2% uranyl acetate and lead citrate. Images were acquired on a FEI Tecnai G2 Spirit electron microscope equipped with a LaB6 source and operating at 120 kV.

RNA isolation
We have described a general method for total RNA extraction and RT-quantitative PCR (RT-qPCR) (8). Total RNA was extracted from mouse skeletal muscles (50-100 mg) using RNA STAT-60 (Tel-Test, Friendswood, TX). Because some muscle types have excessive fatty tissue, and due to a failure to reliably perform RT-qPCR, it became necessary to extract the total RNA using RNeasy® lipid tissue kit (catalog #74804; Qiagen, Valencia, CA) according to the manufacture's protocol with on-column DNase I treatment. RT-qPCR was carried out and analyzed as described below.

RT-qPCR
Total RNA, in equal quantity, was pooled from 6 to 10 of each skeletal muscle type of each genotype and sex, and RT-qPCR was carried out in a 20 l reaction volume. A total of 20 g of RNA was DNase I treated using the DNase-free kit from Ambion (Grand Island, NY). cDNA was made using 2 g of DNase I-treated RNA using a reverse-transcription kit from ABI (Carlsbad, CA). RT-qPCR was performed in duplicate using 2.5 mM primers, 20 ng cDNA, and SYBR Green. All RT-qPCRs were carried out in 96-well plates using the StepOnePlus real-time PCR system (Applied Biosystems, Foster City, CA). RT-qPCR was performed twice and in duplicate, and the transcript levels were normalized to eukaryotic elongation factor 2 (Eef2). We tested several housekeeping genes (Eef2, cyclophilin, and 18S), and Eef2 was less variable among the different muscle types than the other housekeeping genes. Eef2 was selected as a housekeeping gene for normalization in this study. The C t value for each sample was calculated as C t = [C t (gene of interest)  C t (Eef2)]. The C t value for each gene of interest was calculated as C t = [C t (gene of interest)  C t (WT)]. The fold change was calculated as fold change = 2 Ct . Primers used for gene amplification were obtained from Harvard primer bank and Integrated DNA Technologies (Coralville, IA). The primers used in this study are provided in supplemental  Table S1.
In a preliminary study, we amplified the expression of mRNA for several genes individually and in pooled samples and compared the raw C t values for each gene. As shown in supplemental Fig. S1, there was excellent correlation between the mean of individual samples to those when the samples were pooled. We now routinely pool the samples for measurements of mRNA expression (8). Furthermore, in the current study, in order to account for the experimental variance, we generated cDNA from the pooled samples two different times and amplified independently in duplicates.

Immunoblots
Protein (3-100 g) was resolved either on 10% or 4-20% Criterion™ TGX™ premade gels (Bio-Rad) and transferred onto a PVDF membrane (Millipore) using a semi-dry transfer protocol (Thermo Fisher). The protein blots were blocked overnight in TBS plus 0.2% of Tween-20 (TBS-T) containing 5% nonfat dry milk. Primary antibodies were prepared in TBS-T containing 1% fat-free dry milk. GAPDH (1:30,000, AM4300; Ambion) antibodies were incubated for 1-2 h at room temperature. The secondary antibodies were used at a dilution of either 1:2,500 or 1:10,000 at room temperature for 1 h. The antibodies used and their respective dilutions are listed in Table 1. The blots were developed using chemiluminescent reagents (Immobilon Western Chemiluminescent HRP Substrate; Millipore) and exposed to X-ray films. To quantify the bands obtained via Western blot analysis, Image Studio Lite software was utilized (https://www.licor.com/bio/image-studio-lite/). The blots were scanned and the density of each DYSF protein band was first normalized to an internal standard and then normalized to EEF2.

Triglyceride assay
Skeletal muscle triglyceride was measured using a triglyceride colorimetric assay kit from Cayman Chemical (catalog #10010303; Ann Arbor, MI) according to the manufacturer's protocol with a few minor adjustments. Approximately 100 mg of muscle tissue were weighed and homogenized in 1 ml of the buffer with two times the normal amount of protease inhibitor (Roche), except for soleus. The homogenate was first spun (3,000 g for 10 min) to break the excessive froth and transferred to a microfuge tube and spun again at 10,000 g for 10 min. The supernatant was transferred to another tube, including the fat layer, and the volume noted. A preliminary dilution of the samples allowed us to determine the dilution appropriate to be within the range of the standard. The triglycerides are expressed as micrograms per milligram of tissue.

Ponceau S staining
Immunoblots were incubated in a solution of 0.1% (w/v) Ponceau S in 5% (v/v) acetic acid for 1 h at room temperature with gentle rocking. Blots were then rinsed in distilled water until the background was clean.
Untargeted infusion lipidomic analysis. Details of MS have been reported before (10). Briefly, lipid extracts were directly infused over a 3 min interval into a SCIEX 6500 triple TOF mass spectrometer at a rate of 10 l/min. The infusion began with acquisition of TOF spectrum for 30 s. Then, precursor ions were sequentially collected for MS/MS ALL analysis where the production spectrum for each unit mass between 150 and 1,200 Da was collected. This process was repeated in positive and negative mode. However, negative mode data were not further analyzed. After acquisition, lipids were identified by a combination of molecular ion and unique product ion or neutral loss features. The intensity value of each lipid was then normalized to an internal standard for subclass of lipids. Lipids were normalized to 0.2 mg tissue weight (wet). For some analyses, the sum of lipids with common features (e.g., number of double bonds) was used. Lipids were ascertained based on their fragmentation pattern producing defined m/z (expressed in Da) and the intensity measured for each ion for their lipid class. The signals obtained in the mass analyzer were translated into m/z values and intensities (11). The m/z value and intensity peak of each lipid ion were used for identification and quantification of specific lipid species using in-house and commercially available lipid analysis platforms.
Fatty acid profiling by GC-MS. Fatty acids were measured by GC-MS, which has been described in detail before (10).

Statistical analysis
Mean and SD values were calculated using either MatLab or Excel (Microsoft). P values were determined from a two-tailed t-test using either Excel or MatLab. For some analysis, GraphPad Prism 5.01 (GraphPad Software, Inc.) was also used. Lipid species were identified using LipPy, an in-house script. A P value of <0.05 was considered significant.

Bla/J mice
A spontaneous insertion of a retrotransposon in intron 4 in the Dysf gene in the A/J strain of mice (discovered at the Jackson Laboratory) resulted in features of LGMD type 2B (LGMD2B), Miyoshi myopathy, or distal myopathy with anterior tibial onset. Subsequently, this mouse was backcrossed into the C57BL/6 background and designated B6.A-Dysf prmd /GeneJ (Bla/J mice) (12). These mice are no longer C5 complement deficient, which otherwise leaves the A/J mouse susceptible to infections. Dystrophic features in the A/J and BLA/J mice were reported to be very similar, although a small increase in central myonuclei was detected in some muscles at 1 month of age, slightly earlier than the original line. The psoas major and gastrocnemius muscles were the most severely affected followed by the TA and then the quadriceps. Interestingly, this is the only model where initial dystrophic involvement appears more distal than proximal.
Because only one retrotransposon insertion was noted in intron 4 (leaving the promoter intact), and because we did not know how this might affect the expression of the Dysf gene in specific skeletal muscle types, we amplified mRNA for Dysf from each of the muscle types studied using primers located before the retrotransposon insertion (exons 3-4) and those distal to this (primers located in exons 20-21) (supplemental Fig. S2A). Interestingly, we did note increased expression of Dysf mRNA compared with WT mice when amplified with primer pairs located in exons 3-4 (supplemental Fig. S2B), but not when amplified with primer pairs located in exons 20-21 (supplemental Fig. S2C). An increased detection of transcripts proximal to the retrotransposon insertion is likely because the Dysf gene promotor, which is still intact, drives the Dysf gene expression, while the transcripts that contain sequences distal to the retrotransposon will result in aberrant Dysf transcripts and could be degraded. Despite the detection of truncated Dysf transcript proximal to retrotransposon insertion, there was no detectable DYSF protein in any skeletal muscles tested (supplemental Fig. S2D).

Distal skeletal muscles
For locomotion, several skeletal muscles are involved, which include the psoas, gluteus, quadriceps, gastrocnemius, and TA (see supplemental Fig. S3A for the location of mouse muscles), although minor skeletal muscles like extensors and flexors of the foot are also used, but not included in this study. Shown are the images of 95-week-old WT and Bla/J mice of both sexes revealing the extensive degeneration and fatty infiltration of skeletal muscles (supplemental Fig. S3B).
Increased detection of lipids with ORO stain and biochemical measurement of triglycerides in skeletal muscles. The skeletal muscles of both sexes were first evaluated for the presence of lipids in situ by staining the perfused cryosections with ORO. As illustrated in Fig. 1, the most affected muscles (determined by the presence of lipids) were the gluteus and psoas of both sexes (Fig. 1A, B) compared with WT mice. This was followed by the gastrocnemius and quadriceps, which were more affected in female Bla/J mice than male mice, while the soleus and TA seem to be least affected. This pattern of ORO muscle staining was reflected in the biochemical measurement of triglycerides. As shown in Fig. 1C and D, there was a robust increase in the triglycerides in gluteus [male (M), 23-fold; female (F), 4-fold] and psoas (M, 6-fold; F, 8-fold) muscles, followed by an increase in the gastrocnemius (M, 2-fold; F, 4.5-fold) and quadriceps (M, 3-fold; F, 10-fold) muscles. We also noted that the female gastrocnemius and quadriceps muscles were more affected than the male and that female soleus showed a small increase in triglycerides, but not the male soleus. This analysis shows that while a few muscles like gluteus and psoas are equally affected in both sexes, there is some sex-related variation in other muscle types.
Expression of de novo lipogenesis and TAG synthesis pathway genes are increased in the steatotic skeletal muscles of Bla/J mice of both sexes. A schematic for several metabolic pathways examined in this study is provided in supplemental Fig. S4A. As expected, the muscles with the most lipid accumulation (psoas, gluteus, and quadriceps) showed the most robust increase in the expression of those genes involved in the lipid synthesis pathway (de novo lipogenesis and TAG synthesis) in both sexes ( Table 2). Gastrocnemius showed relatively less expression when compared with psoas, gluteus, and quadriceps muscles. We also noted that there was no significant increase in the expression of lipogenesis genes in soleus and TA. These two muscles were also the least affected in terms of lipid accumulation (ORO staining). The gene expression profiling of skeletal muscles with lipid accumulation from Bla/J mice revealed marked upregulation of several genes of the de novo lipogenesis pathwway (e.g., Acyl, Acc1, Fas, Evol6, and Scd1 were upregulated 30to 50-fold) and those of TAG synthesis (Gpat3/ Agpat10, Agpat2, Ppap2c, Dgat1, and Dgat2 were increased 10to 20-fold) ( Table 2). Interestingly, expression of lipin 1, which is reported to have phosphatidic acid phosphorylase activity, was not increased (13,14), but the expression of other phosphatidic acid phosphorylases (Ppap2a1, Ppap2b) were increased. No increase in the expression of Mogats was noted, indicating a lack of involvement of the monoacylglycerol acyltransferase pathway for TAG synthesis (15) in this disease condition. Srebp-1c, a transcription factor regulating de novo lipogenesis, was also increased.
Expression of gluconeogenesis and glycolysis genes are increased in the steatotic muscles of Bla/J mice of both sexes. We next measured the expression of key enzymes in the pathway for gluconeogenesis, as shown in Table 3. Chrebp is a key transcription factor regulating carbohydrate metabolism. It was increased 17-fold in the psoas muscle of male Bla/J mice but not in the other muscle types studied. However, the key to gluconeogenesis is the robust expression of phosphoenolpyruvate carboxykinase (Pepck) in all the muscles, mostly in the psoas and gluteus (20-to 100fold), but not G6pase. Dephosphorylation of G-6-P is the last step in releasing glucose. This suggests that the glucose synthesized in the affected tissue remains in the tissue, most likely stored as glycogen. However, this seems not to be a likely scenario because the mRNA expression of key enzymes for glycogen synthesis remained unaltered, or in fact were decreased, as in the psoas muscles of both sexes. Alternatively, increased expression of Pepck in these pathological tissues has an additional function. For example, Pepck has a cellular proliferation function utilizing glucose and glutamine and increases ribose synthesis using noncarbohydrate carbon sources like lactate. This also corroborates well with increased expression of Slc1a5, described below (16). The cause of such an increase in the mRNA expression of Pepck is unclear, and how its products are channeled in other pathway(s) remains unresolved.
Expression of genes involved in amino acid uptake and metabolism in the steatotic muscles of Bla/J mice in both sexes. We also measured the expression of Slc1a5 (a neutral amino acid transporter belonging to the solute carrier 1 family). Slc1a5 is highly expressed in nonneural tissues (17) including the lung, intestine, and adipocytes (18). The mRNA expression of Slc1a5 was robustly upregulated in the most steatotic muscles (psoas and gluteus), less in the gastrocnemius and quadriceps, and remained unchanged in the soleus and TA (Table 3). Slc1a5 has the highest affinity for glutamine (19), the most abundant nonessential amino acid in circulation (20). Glutamine is metabolized to glutamate by the enzyme glutaminase (Gls), a key precursor for many biosynthetic pathways. However, the expression of Gls was Note that in the gluteus and psoas muscles of both sexes, excessive lipids coalesce into oil droplets due to the hydrophobic nature of neutral lipids. Note also that in female mice, skeletal muscles obtained from psoas, gastrocnemius/plantaris (Gastroc), quadriceps (Quad), and gluteus (Glut) have significant lipid accumulation (as ascertained by ORO staining), and in male gastrocnemius/plantaris, it is less prominent. Furthermore, soleus muscle from both sexes lacks any ORO staining, suggesting undetectable lipid. Shown also is the tibialis anterior (TA) muscle stained for lipid. While there is some indication of ORO staining in TA muscle from female mice, this is absent in TA muscle from male mice. Images were captured as described in the Methods using two different cameras. The § indicates images captured with a Jenoptik Gryphax NAOS camera and those unmarked were captured with an Optronics Microfire camera. Scale bar 40 m. C, D: Biochemical measurement of triglycerides (TAG) expressed as micrograms per milligram of tissues. Individual values are shown in either open or filled circles and the bar represents the mean (n = 3-6). P values are shown above the bars. Relative fold changes in the expression of various genes (mRNA) of de novo lipogenesis and TAG synthesis pathways by RT-qPCR (see supplemental Fig. S4A and B for details). Relative mRNA expression for key enzymes involved in de novo lipogenesis, TAG synthesis pathway, dihydroxyacetone phosphate acyltransferase (DHAPAT) glycerolipid synthesis pathway, fatty acid uptake, and Our cut-off value for meaningful gene expression is 30 C Relative fold changes in the expression of various genes (mRNA) of gluconeogenesis, glycolysis, glycogen, amino acid uptake, mitochondrial, and cholesterol synthesis pathways by RT-qPCR (see supplemental Fig. S4A and B for details). Relative mRNA expression for key enzymes involved in gluconeogenesis, glycolysis, glycogen pathway, amino acid uptake and metabolism, mitochondrial pathway, and the cholesterol synthesis pathway. The C t value for each sample was calculated as C not increased in the muscles examined. As observed before, Pepck was significantly increased in the affected muscles and recent investigations reveal that Pepck can regulate glutamine utilization and enter into the TCA cycle and provide precursors for gluconeogenesis (21). Further studies will define the role of amino acid metabolism in these affected muscles. In addition, Slc1a5 has affinity for other amino acids as well. Therefore, it is unclear from this observation whether uptake of other amino acid(s) is increased in the diseased muscle tissues. Next, we determined the gene expression for mitochondrial fatty acid metabolism by measuring Cpt1a, Cpt1b (muscle specific), and Cpt2 (Table 3). We noted an increase in expression of Cpt1a and Cpt2 in the steatotic muscles, but the expression of Cpt1b, which is muscle specific, remained unchanged. This further reflects the decrease in the muscle tissue. Insig1 is a key protein in the pathway regulating cholesterol synthesis. We also noted an increase in Insig1 expression in the psoas and gluteus in both males and females, and in the gastrocnemius and quadriceps only in females.
Changes in various lipid species in the steatotic muscles of Bla/J mice in both sexes (lipidomics). We next surveyed the total lipids in the fatty skeletal muscles using an infusion MS approach. In this method, the total lipids are extracted in one solvent system and directly analyzed with MS. Using this unbiased approach, we detected several molecules with lipid-like features, anywhere from 293 molecules to 1,558 molecules, depending on the muscle type. There was also variation in detected lipids in the male versus female muscles. For example, while only 293 lipid species were detected in male gastrocnemius muscle, in the female, we could detect 1,136 lipid species. It is also interesting to note that the vast majority of lipid species (54-78% depending on the muscle types) remain unidentified, mostly due to incomplete annotation of the lipids or nonavailability of lipid standards. Furthermore, unknowns could also be adducts, isotopes, and other anthropogenic compounds like plastics, surfactants, etc. Shown are the volcano plots plotted for all lipid data (known and unknown) by plotting an arbitrary P value of <0.05 against the fold change between the WT and Bla/J muscles (Fig. 2). We did not detect a clear pattern of lipid species emerging from the various muscle types or between sexes studied. In male gastrocnemius muscle, more lipid species were increased than decreased, while the opposite was the case in psoas muscle, where a vast number of lipids were decreased. This was largely absent in female muscles except for female quadriceps muscle where, again, more lipids species were downregulated. As more lipid annotations are made available, the unidentified lipids will be determined and their functional aspects could then be ascertained.
The lipid species that changed 2-fold or more are presented in supplemental Tables S2 and S3. It is interesting to note that, in general, there were increases in TAGs, which corroborates the biochemical measurements of TAGs in the muscles. This is also substantiated with an increase of the mRNA expression of key enzymes for the TAG biosynthesis pathway. Most TAGs observed have total carbons from C48 to C58 with at least one saturated fatty acid in both sexes. A few TAG trends also emerged. TAGs with 48-50 carbons were only detected in the quadriceps muscle, while those with C52-C54 were only present in the psoas muscle of male mice. The presence of saturated fatty acids was also noted parallel to increased expression of desaturase. These TAG trends were less visible in female mice.
Other classes of lipids that were well-represented were PC and PE. These phospholipids are usually found in cellular membranes and are extracted well with the solvent system used in this study.
Collision-induced dissociation of TAGs using MS results in the loss of one fatty acid group. The remaining DAG-like structure is identified by the sum of the remaining two fatty acids. Identifying the individual fatty acids is not feasible without more advanced MS techniques. Fatty acids found in the fragmented TAGs are presented in supplemental Fig. S5. Generally, all the long fatty acids were increased more than 2-fold in the muscles of both sexes. A few of those (C18:1, C18:2, C18:3, and C20:4) were increased more than others.
We also noted increases in certain very long-chain fatty acids (VLCFAs) found in TAGs, mostly C22:5 or C22:6. Some of these VLCFAs are of dietary origin and some of them are de novo synthesized. The endogenous generation of these VLCFAs requires elongases [elongation of very long-chain fatty acids (Elovl)] (22). There are at least seven known elongases; the substrate specificities of all the known Elovl isoforms are unclear but most of these Elovls have substrate redundancy. The roles of these Elovls in the de novo synthesis of VLCFAs are shown in supplemental Fig. S4B. Based on the expression pattern extracted from the public database, BioGPS (http://biogps.org), and shown in supplemental Fig. S4C, it emerges that Elovl3 is poorly expressed in both white adipose tissue and skeletal muscle. We amplified the mRNA for Elovl3, and found that it is highly upregulated in the psoas muscle of both sexes of Bla/J mice (Table 2). Both Elovl5 and Elovl6 are expressed in white adipose tissue and skeletal muscle (supplemental Fig. S4C). When amplified in the muscles of Bla/J mice, we noted muscle-specific upregulation of these two Elovls. Elovl5 was upregulated more in those muscle types that have fatty infiltration, while Elovl6 was only noted in psoas muscle of both sexes.
Fatty acid levels in the psoas and gluteus muscles of Bla/J mice of both sexes. The observation that a few of the fatty acids found in TAGs were increased was further corroborated by measuring the fatty acids by GC-MS in the neutral lipids. We chose to measure fatty acids only in psoas and gluteus muscles, as these were the most affected muscles. As shown in supplemental Table S4, the majority of the fatty acids found in the neutral lipids by the previous method are also increased in this method. Although a few of these, such as C22:5(n3), increased 7-fold in male and 23-fold in female gluteus muscle, they did not reach statistical significance. However, in psoas muscle of both sexes, an 7-fold increase was noted, but it only reached statistical significance in male psoas muscle. Overall, this assay confirms the fatty acids detected in the TAGs.  2. Volcano plots of total lipids detected by MS in the aged skeletal muscles of WT and Bla/J muscles. Total lipids of all lipid classes identified (known and unknown) are plotted as a volcano plot of the fold change between WT and Bla/J versus significance P < 0.05. Each dot represents an individual lipid species. In the plots, we have identified some of the lipid species that are unique to the specific muscle type (for other lipid species, see supplemental Tables S1, S2). Those species are circled in red and identified on the graph. The total number of lipid species plotted, along with the number of unknown species, is mentioned on each graph. There are n = 9-11 samples per muscle for males and n = 7 for females. Lipids are normalized to 0.2 mg tissue weight (wet). The raw lipidomics data sets will be made available upon request.
An increase in adipocytes in the steatotic muscles of Bla/J mice. We next examined the presence of adipocytes in the muscle tissue. Displayed in Fig. 3A are the markers for adipocyte differentiation. Some of the key markers for adipocyte differentiation and maturation are shown. Cebp and Ppar, early markers for adipocyte differentiation, were significantly increased (20-fold) in steatotic muscles. The significant increase in the expression of Ap2, a fatty acid binding protein abundant in mature adipocytes, was also increased (12-to 20-fold), again only in steatotic muscles which are affected. The adipocyte cellular markers, caveolin 1 (Cav1) and Ptrf/Cavin1, were also robustly increased in the steatotic muscles of both sexes. Lipid droplets are decorated with multiple proteins, among which is perilipin (PLIN)1, which is specifically associated with the lipid droplets found in adipocytes. We measured the mRNA Fig. 3. Increased molecular fingerprints of adipocytes in the affected skeletal muscles of Bla/J mice of both sexes. A: mRNA expression for the key adipocyte differentiation markers. Shown are the mean ± SD from two independent experiments measured in duplicate. Values shown in red are 2-fold or more up-or downregulated in the BLa/J mice compared with WT. *WT value was above 30 C t , Bla/J value was not. B: Immunoblots for PLIN1 protein. Equal quantities (30 g) of total tissue lysates of various skeletal muscles were probed with PLIN1 antibody. The expected PLIN1 protein is marked in the red box. Included also is tissue from brown adipose tissue (BAT) as a positive control. The same blot was stripped and probed with EEF2, a housekeeping gene, and the blots were then stained with Ponceau S for detection of total protein transfer. White dashed lines are drawn to orient the gel lanes. Wider white space indicates different gels. Immunoblots for CAV1 protein. Equal quantities (30 g) of total tissue lysates of various skeletal muscles were probed with CAV1 antibody. Expected CAV1 protein is marked by a solid arrowhead. The same blot was stripped and probed with EEF2 and GAPDH housekeeping genes and the blots were then stained with Ponceau S for detection of total protein transfer. White dashed lines are drawn to orient the gel lanes. Wider white space indicates different gels. We performed two housekeeping genes, one (EEF2) more constantly expressed in muscle tissue and the other (GAPDH) a more general protein from the metabolic pathway. Because of extreme fatty tissue in the psoas and gluteus, the detection of proteins was not consistent. For this reason, we further stained the protein blots with Ponceau S to detect the total protein transfer. A significant increase of CAV1 is observed. However, because of inconsistent housekeeping protein detection, we are not showing protein quantitation. C: Immunoblots for PTRF protein. Equal quantities (30 g) of total tissue lysates of various skeletal muscles were probed with PTRF antibody. Expected PTRF protein is marked by solid arrowhead. The same blot was stripped and probed with EEF2, a housekeeping gene, and the blots were then stained with Ponceau S for detection of total protein transfer. Because of extreme fatty tissue in psoas and gluteus, the detection of proteins was not consistent. For this reason, we further stained the protein blots with Ponceau S, detecting the total protein transfer. A significant increase of PTRF is observed. However, because of inconsistent housekeeping protein detection, we are not showing protein quantitation. White dashed lines are drawn to orient the gel lanes. Wider white space indicates different gels. expression of Plin1-6 in the muscles (23). Plin1 was significantly increased (M, 33to 85-fold; F, 60-fold) (Fig. 3A). Additional isoforms (Plin2-5) were increased, but mostly in the psoas muscles of both sexes. This increase in mRNA expression was also corroborated at the protein level. Tissue lysates from all the muscles used in this study were analyzed using immunoblot. We detected the presence of PLIN1 in those muscle types that have excessive lipid accumulation (quadriceps and psoas muscles), but relatively less in gluteus muscle, further confirming the presence of lipid droplets (Fig. 3B). PLIN1 was only observed in those muscle types that are most affected. This again confirms the presence of lipid droplets.
Additionally, we then amplified the expression of two genes, Cav1 and Ptrf/Cavin1, both at the mRNA and protein level, which are part of caveolae, a small plasma membrane invagination found abundantly in adipocytes (which in the current study we take as cell markers for adipocytes). In contrast, muscles express cav3 instead of cav1. As shown in Fig. 3A, mRNA expression of both Cav1 and Ptrf was in-creased in muscle types showing increased fatty/adipose tissue. These adipocyte markers were increased (2-to 5-fold) in both sexes, but only in those muscles that are affected (Fig. 3A). This observation was further confirmed by immunoblots probed with antibodies for CAV1 (Fig. 3B) and PTRF (Fig. 3C, C i ). These observations confirm the presence of adipocytes in the degenerating muscle tissue.
Increased adipocytes are a consequence of myocyte degeneration in steatotic muscles of Bla/J mice in both sexes. The increase in adipocytes observed in the affected muscles seems to occur as a consequence of myocyte degeneration (24,25). We amplified key markers for the differentiation of myocytes and their mRNA expression molecular makers (26) (Fig.  4A). In our analysis, Pax3 was undetectable in all the muscles examined. However, another myocyte differentiation marker, Pax7, was decreased substantially in the psoas muscle or undetectable in muscles from male mice. However, this downregulation was not so robust in muscles from female mice. The molecular markers for myocytes, Myf5 and Mef2c, were again either undetectable or significantly Fig. 4. Decreased molecular fingerprints of muscle in the affected skeletal muscles of Bla/J mice of both sexes. A: mRNA expression for the key differentiation markers. Shown are the mean ± SD from two independent experiments measured in duplicate. Values shown in red are 2-fold or more up-or downregulated in the BLa/J mice compared with WT. Our cut-off value for meaningful gene expression is 30 C t and above. The C t values above 30 in both WT and Bla/J muscle samples were not used for calculating fold-change. B: Immunoblots for CAV3 protein. Equal quantities (30 g) of total tissue lysates of various skeletal muscles were probed with Cav3 antibody. Expected CAV3 protein is marked by a solid arrowhead. The same blot was stripped and probed with EEF2 and GAPDH housekeeping genes, and the blots were then stained with Ponceau S for detection of total protein transfer. The detection of proteins was not consistent in the extremely fatty psoas and gluteus tissues. For this reason, we further stained the protein blots with Ponceau S to detect the total protein transfer. CAV3 is undetectable in psoas and gluteus muscles of both the sexes, marked with a red box. However, because of inconsistent housekeeping protein detection, we are not showing protein quantitation. White dashed lines are drawn to orient the gel lanes. Wider white space indicates different gels. reduced. Finally, the markers for mature myofibers, MyoD and MyoG, were also downregulated. However, this effect was less pronounced in gluteus and quadriceps, despite having steatosis. Thus, myogenesis was substantially reduced in the psoas muscle of both sexes. This would suggest that the numbers of satellite cells available for differentiation are limited. We also measured the mRNA expression of Dmd and Cav3, specific for skeletal muscles. Both the markers were decreased in the affected psoas muscles of both sexes, although their expression did not change substantially compared with WT mice.
The observed decrease in the mRNA expression of Cav3 was further confirmed at the protein level (Fig. 4B). In those muscles which are affected the most (psoas and gluteus), CAV3 protein was undetectable. CAV3 protein was detected in other muscles that were less affected. This certainly suggests the degeneration of muscle in psoas and gluteus.
The observed adipocytes are extramyocellular in affected muscles as observed in light microscopy. After observing the presence of markers for adipocytes by mRNA and proteins, we set out to determine whether adipocytes are intra-or extramyocellular. As illustrated in Fig. 1, the ORO staining was intense only in those muscles that are most steatotic (namely, the psoas, gluteus, and quadriceps) but barely seen in the least affected muscles. However, the ORO staining of the muscle cryosections did not reveal a clear cellular localization of lipids (whether it is intra-or extramyocellular). We then followed this with examining paraffin embedded muscle tissue sections stained with H&E where clear outlines of adipocytes are discernable (Fig. 5). The H&E-stained muscle images clearly show that, when present, adipocytes are extramyocellular (Fig. 5).
Skeletal muscle-associated adipocytes are extramyocellular as confirmed by TEM. The gluteus and psoas muscles were then examined by TEM, and the images are presented in Fig. 6. Clearly, the adipocytes are extramyocellular.
Upon review of the literature and as defined by some investigators, the localization of fat in the muscle has been variously described. One study defines intermuscular adipose tissue as being fatty layers between the muscle fiber bundles, and intramyocellular lipid as being lipid droplets stored in the cytoplasm of muscle cells (27).
Our observation clearly detects the fat layer outside the muscles, which we refer to as extramyocellular adipose tissue [see whole animal skeletal muscle where only the skin is removed (supplemental Fig. S3B  All images were captured using the same camera/software, except as noted. Shown are the 40× images, scale bars are shown. Histological evaluation of paraffin embedded skeletal muscle stained with H&E from aged (95-99 weeks old) Bla/J mice shows the presence of intermuscular adipocytes in muscles from both sexes. As presented in this figure, the psoas, gastrocnemius/plantaris, quadriceps, and gluteus muscles show an extensive presence of intermuscular adipocytes defined by their cellular membranes. Such a prominent presence of adipocytes is absent from the heart and soleus muscle. Male TA muscle shows a few adipocytes in this image, although most of the muscle tissue examined does not. Images were captured as described in the Methods using two different cameras. Images with a § sign were taken with a Jenoptik Gryphax NAOS camera. All other images were taken with an Optronics Microfire camera. Scale bar 40 m. DYSF protein is more highly expressed in the psoas, gluteus, and quadriceps compared with that in soleus, gastrocnemius, and TA in aged WT mice. Among the different distal skeletal muscles we studied in Bla/J mice, the most affected for lipid/ adipocyte accumulation are psoas, gluteus, and quadriceps, and the least affected are soleus, gastrocnemius, and TA. There might be multiple reasons for this phenotype, namely, differential expression of Dysf, both at mRNA and protein levels. Additionally, it could be due to differential protein stability and a differential role of DYSF in unaffected muscles. Because there are multiple Dysf variants deposited in GenBank and the cellular function and differential expression patterns are unclear in these muscle types, we aligned all the variants and designed primers that will amplify all the variants. As shown in Fig. 7 (schematic for partial Dysf gene structure, Fig. 7A), we amplified Dysf transcripts both in the proximal (primer located between exon 3 and exon 4; Fig. 7B, D) and distal (primer located between exon 20 and exon 21; Fig. 7C, E) regions of the transcript. When quantifying the expression compared with the quadriceps, it seems that the expression of Dysf is 50% less in the soleus and psoas in both sexes when amplified using both primer pairs. To study the DYSF protein in these muscles, we initially characterized the commercially available antibodies to determine their suitability for immunoblots and immunofluorescence studies. Our characterization revealed that while the DYSF protein could be identified by molecular weight, the antibody has robust interaction with low molecular weight nonspecific proteins in the muscle lysates, and thus seems unsuitable for in situ immunofluorescence detection of DYSF (supplemental Methods and Results, supplemental Fig. S7). However, this mRNA expression of Dysf is not faithfully replicated at the protein level. Immunoblots of serial dilution of DYSF protein in the skeletal muscles (Fig. 7F) is parallel only for soleus muscle but was discordant for TA muscle. In TA muscle, mRNA expression for Dysf is high, yet the protein is substantially low. Psoas muscle does seem to correlate well between the expression of mRNA and protein (Fig.  7G, H). To quantitate the DYSF protein, we used a serial   7. Expression and quantitation of DYSF transcripts and protein in various distal skeletal muscles in mice. A: Schematic for partial Dysf gene showing the exons (boxed) and the location of primer pairs. B, C: Fold change in the expression of Dysf mRNA normalized to Eef2 and expressed compared with quadriceps as 1, amplified with primer pairs 3-4 and 20-21, respectively, in male mice. Bars represent the mean of two independent amplifications of pooled muscle samples (n = 6). D, E: Fold change in the expression of Dysf mRNA normalized to Eef2 and expressed compared with quadriceps as 1, amplified with primer pairs 3-4 and 20-21, respectively, in female mice. Bars represent the mean of two independent amplification of pooled muscle samples (n = 6). F: Immunoblots for DYSF protein. Several concentrations of total tissue lysates were resolved on 4-20% gradient gels and probed with DYSF antibody. Note that an additional slightly smaller protein band is also recognized by this antibody that could be a variously spliced Dysf transcript (see supplemental Fig. S7 for explanation). The immunoblot was stripped and reprobed with a housekeeping protein, EEF2, followed by staining the blot with Ponceau S for total protein. The DYSF protein was normalized to EEF2 and quantified and expressed compared with quadriceps as 1. H: Similar quantification for DYSF was carried out at individual protein concentration normalized to EEF2. Gastroc, gastrocnemius; Glut, gluteus; Quad, quadriceps; TA, tibialis anterior. dilution method (28) such that the analyzed protein level did not reach saturation and quantification was more accurate. From this DYSF protein quantification, it would seem that those muscles that are least affected also have reduced presence of protein, further suggesting the role of DYSF in the etiology of adipocyte/adipose tissue generation.

DISCUSSION
The key observations of this study are: 1) There is an extensive accumulation of adipocytes in the steatotic muscles of Bla/J mice of both the sexes. 2) Only a few of the muscles are severely affected. For example, the psoas, gluteus, and quadriceps are the most affected, followed by the gastrocnemius, but the soleus and TA are spared.
3) The steatotic muscles consist of adipocytes as determined by the increased expression for the cellular markers of adipocytes like CAV1 and lipid droplet protein PLIN1. 4) This increase in adipocytes appears to be a consequence of muscle degeneration. 5) The increased presence of muscular adipocytes is extramyocellular.
In mammals, there are three muscles that connect the spine to the legs: the psoas, gluteus maximus, and piriformis. The psoas muscle attaches to the vertebrae of the lumbar spine. It next joins with the iliacus muscle, gluteus muscle, and muscles of the hind limb (quadriceps, gastrocnemius, soleus, and TA). The biomechanics of locomotion in humans and mice are different: mice have a crouched quadrupedal gait compared with the upright bipedal gait of humans. Therefore, the biomechanics of muscle function also differ. The 3D model of the lower limbs (29) of humans and mice (30) was recently employed to simulate the locomotion in humans and mice (31). It was observed (31) that mice had an 48% decrease in muscle fiber extension compared with humans, suggesting that the human muscles should be more affected than those in mice. The observations in mice parallel the observations reported in human subjects affected with LGMD due to mutations in Dysf, except that in mice, the fatty infiltration was indistinguishable between the sexes, and in humans, females were more affected than the males.
We also found heterogeneity in the severity of muscle steatosis in the Bla/J mice, sparing the soleus and TA, while the psoas, gluteus, and quadriceps were severely affected. It is unclear why these muscles are spared in mice. This distinction was not revealed in humans, which could be due to the different modes of locomotion. We did attempt to discern this fact by examining DYSF mRNA and protein expression in various muscles. In our hands, the expression of DYSF was not altered among the various muscle types. Because DYSF is known to repair damaged muscle fibers, we speculate that there might be two plausible scenarios: 1) In mice, the soleus and TA do not undergo extensive stretch/extension and thus there is not extensive muscle fiber damage/repair. 2) There is an alternative mechanism in operation that helps to repair damaged muscle fibers. In this regard, it will be interesting to study the muscle-specific total protein (32) and mRNA expression. This might reveal additional muscle repair pathway(s) that might be exploited for muscle repair.
LGMD occurs due to mutations in several genes categorized in 34 different subtypes reviewed in (1,33,34). Fat deposition and accumulation have been reported in Duchenne muscular dystrophy increasing to 50% of muscle mass (35). Thus, muscular fatty infiltration might not be as unique to subjects affected by mutation in DYSF as initially thought. Patients with Becker muscular dystrophy do not show an early fat infiltration of gluteus medius and maximus muscles of the lower limb (36)(37)(38). The muscle phenotype in mice is also similar, if not identical to humans. A recent study evaluating LGMD due to mutations in DYSF reports similar fat infiltration in human muscles (39).
Lipidomic analysis of the lipids from the fatty muscle showed muscle-specific molecular species of TAGs (supplemental Tables 2, 3). For example, four different TAG species (C48:2, C48:3, C50:3, and C50:4) were not found in any other muscle type studied except in quadriceps. Likewise, TAG C52:2, C52:4, C54:4, and C54:5 were uniquely detected in the psoas muscle. In gastrocnemius, C54:6, C54:7, C54:8, and C58:10 were uniquely upregulated, while in gluteus muscle, no TAG species were found to be uniquely upregulated. Furthermore, these muscle-specific TAG changes were more prominently displayed in the male muscles compared with female muscles, where we did not find any TAG species unique to any one muscle type. It is unclear why the various skeletal muscles studied generate a few of the muscle-specific TAG species. As only two mammalian DGATs are described with no reported substrate specificities, it would suggest altered availability of fatty acid synthesis and the fatty acid pool used for TAG synthesis in these muscles. Although not clear, it is also possible that TAG remodeling can occur differently in these muscles. TAGs are usually stored in lipid droplets due to the extreme hydrophobic nature of TAGs. It has not yet been reported whether TAGs can also be embedded in the cellular membranes, like DAGs, and might have a physiological/signaling role. Specific TAGs can alter the physicochemical properties of the membranes where they reside.
Neutral lipids are differentiated by the number of carbon atoms, the position of acyl chains at the sn-1, sn-2, and sn-3 positions of the glycerol chain, and the number of double bonds in the acyl chains. We next determined the unique molecular species of TAGs in the muscle types studied (shown in supplemental Tables S2, S3). These TAGs range from C50:1 to C58:3. Earlier studies have determined that fragmentation at the sn-1 and sn-3 positions of the glycerol occurs before the loss of acyl chain at the sn-2 position (40). The precursor ions therefore identify the sn-2 acyl chain. Thus, based on the number of carbon atoms and the double bonds (shown in supplemental Tables S2, S3), we can speculate the various possible acyl chain combinations. To narrow down this list of possible combinations will require further targeted analysis for these molecular species of TAG using more advanced MS measurements (41). The precursor for DAG contains the two acyl chains, and, again, due to the fragmentation pattern of acyl chains from the sn-1 and sn-2 positions of the glycerol molecule, the