Metabolomic Analysis of Oxidative and Glycolytic Skeletal Muscles by Matrix-Assisted Laser Desorption/IonizationMass Spectrometric Imaging (MALDI MSI)
Skeletal muscles are composed of heterogeneous muscle fibers that have different physiological, morphological, biochemical, and histological characteristics. In this work, skeletal muscles extensor digitorum longus, soleus, and whole gastrocnemius were analyzed by matrix-assisted laser desorption/ionization mass spectrometry to characterize small molecule metabolites of oxidative and glycolytic muscle fiber types as well as to visualize biomarker localization. Multivariate data analysis such as principal component analysis (PCA) and partial least squares discriminant analysis (PLS-DA) were performed to extract significant features. Different metabolic fingerprints were observed from oxidative and glycolytic fibers. Higher abundances of biomolecules such as antioxidant anserine as well as acylcarnitines were observed in the glycolytic fibers, whereas taurine and some nucleotides were found to be localized in the oxidative fibers.
KeywordsMetabolomics MALDI Mass spectrometric imaging Multivariate data analysis Muscles
Skeletal muscles compose over 40% of the human body mass and are a major contributor to overall metabolic homeostasis [1, 2, 3, 4]. Various conditions and diseases affect skeletal muscle function and, thus, the physical performance and health of the organism. For instance, Duchene muscular dystrophy is found in 1 of 3500 males in their early childhood, in which rapid degeneration of the muscles occurs followed by death in the patient’s early twenties [1, 2, 5]. Decrease in muscle quality, mass, and strength arising from conditions such as aging and obesity have been observed as well [5, 6, 7, 8, 9, 10, 11]. Diagnostic imaging methodologies may enhance our ability to identify metabolic and structural abnormalities contributing to these diseases states.
Skeletal muscles are heterogeneous in nature, and are composed of a mixture of muscle fibers that can be distinguished based on different physiological, morphological, biochemical, and histological characteristics. Historically, skeletal muscle fibers were divided into two types: slow-twitch and fast-twitch fibers, which correspond to their speed of shortening [12, 13]. The two types are also known as red and white fibers, respectively, for their different appearance in color, where the red color arises from the higher amount of myoglobin and higher number of capillaries in slow-twitch fibers [13, 14]. In the 1960s, with the use of ATPase histochemical staining analysis, the two fiber types were found to have different ATPase activity. The slow-twitch fibers rely predominantly on oxidative metabolism, whereas the fast-twitch fibers have higher glycolytic capacity [12, 13]. Subsequently, subtypes (i.e., IIa and IIb) of fast-twitch fibers were found by the use of the succinate dehydrogenase staining method [12, 14]. In the late 1980s, the analysis of muscle myosin heavy chain (MyHC) composition with monoclonal antibodies revealed the four MyHC isoforms, type I, type IIa , type IIx , and type IIb [13, 14]. It is important to note that not all animals express all four types, nor does each fiber type express similar properties across different species [12, 13, 14].
Metabolomics, a comprehensive investigation of the metabolome perturbation resulting from the biological or environmental stimuli, has gained attention in the field of systems biology [15, 16]. This approach of study involves large-scale identification and quantitation of small molecules (metabolites) such as amino acids, nucleotides, fatty acids, and lipids [16, 17]. Although several analytical platforms and techniques are available for conducting metabolomics analysis, mass spectrometry has proven to be highly valuable because of its high selectivity and sensitivity [15, 16]. A general workflow of a metabolomics analysis involves (1) collecting the appropriate samples, (2) sample preparation for the analytical method of choice, (3) sample analysis, (4) data processing and analysis [which typically involves multivariate statistical analysis such as principal component analysis (PCA) or partial least squares discriminant analysis (PLS-DA)], (5) biomarker structural elucidation and identification, (6) and biological interpretation [16, 18]. Several mass spectrometry-based proteomic analysis of skeletal fiber types and their responses to conditions and disease such as aging  and immobilization-induced atrophy  have been reported; however, very limited metabolomics studies have been conducted. Thus, metabolomic characterization of skeletal muscle fiber types may provide valuable insight into biology and disease.
Mass spectrometric imaging (MSI) is a microprobe technique that detects and maps a large number of biomolecules directly from the tissue sections [21, 22]. More importantly, MSI is a label-free technique that does not require the use of radiolabeling or fluorescent tagging as in whole-body autoradiography and fluorescence microscopy, respectively . The utilization of matrix-assisted laser desorption/ionization (MALDI) is perhaps the most common ionization source used for biomolecule MSI analysis [21, 22]. Briefly, a typical MALDI MSI experiment involves (1) collecting appropriate samples (i.e., samples with structural or spatial information), (2) sample preparation (e.g., cryostat sectioning, drying, matrix coating), (3) tissue analysis by raster underneath the laser to collect mass spectra at discrete positions, (4) mass analysis by identifying mass-to-charge (m/z) values of analytes of interest, and (5) ion image generation by plotting the intensity of the m/z versus the X,Y position. This work reports the metabolomic analysis of oxidative and glycolytic skeletal muscles by MALDI MSI and the correlation to histochemical analysis.
Chemical and Material
2,5-Dihydroxybenzoic acid (DHB) and 9-aminoacridine (9AA) were purchased from Sigma-Aldrich (St. Louis, MO, USA). HPLC grade water (H2O), chloroform (CHCl3), methanol (MeOH), and phosphate buffered saline (PBS) were purchased from Fisher Scientific (Fair Lawn, NJ, USA), and 100% ethanol (EtOH) was purchased from Decon Labs (King of Prussia, PA, USA). Two different matrix solutions were prepared for positive and negative ion mode analyses. For positive ion mode, 0.5 M DHB in 100% MeOH was used, whereas 10 mg/mL 9AA in 70:30 (v:v) EtOH:H2O was prepared for negative ion mode analysis. For the multicolor immunofluorescence analysis, primary antibodies BA-F8, SC-71, and BF-F3 were purchased in 1 mL supernatant form from the Developmental Studies Hybridoma Bank at the University of Iowa (Iowa City, IA, USA). Secondary antibodies Alexa Fluor 350 Goat Anti-Mouse IgG2b (γ2b), Alexa Fluor 488 Goat Anti-Mouse IgG1 (γ1), Alexa Fluor 555 Goat Anti-Mouse IgM (μ chain), and Prolong Gold antifade reagent were purchased from Life Technologies (Grand Island, NY, USA).
Animals and Sample Preparation
Six-month-old male Fisher 344 x Brown Norway (F344BN) rats were purchased from the National Institute on Aging Colony at Harlan Laboratories (Indianapolis, IN, USA). This rat strain is known for its increased lifespan along with decreased cumulative lesion compared with other strains, as well as age-related increase in adiposity and decreased lean body mass [23, 24]. The animals were housed individually on a 12-hour light/dark cycle in a specific pathogen-free facility accredited by the American Association for Accreditation of Laboratory Animal Care. Animals were fed a standard rodent chow (18% kcal from fat, no sucrose, 3.1 kcal/g, diet 2018; Harlan Teklad, Madison, WI, USA), and were housed for 4 week to adapt to their housing conditions for establishing baseline health, body composition, and food intake prior to sacrifice. Health status, body weight, and food intake were monitored daily. All experimental protocols were approved by the University of Florida’s Animal Care and Use Committee and in accordance with the Guide for the Care and Use of Laboratory Animals. Rats were sacrificed by rapid decapitation using a guillotine; soleus (SOL), extensor digitorum longus (EDL), and gastrocnemius (GAS) muscles were removed and subsequently flash-frozen in liquid nitrogen. All muscle blocks were stored in a –80°C freezer until further preparation. SOL, EDL, and GAS muscles were selected for this study because of their characteristic fiber type compositions. SOL and EDL are mainly composed of type 1 and type 2 fibers, respectively [7, 25]. GAS is composed of a mixture of type 1 and type 2 fibers, with the regionalization of oxidative type 1 fibers in the deeper region of the muscle group .
For MSI analysis, tissue sections were placed in a vacuum desiccator for 1 h to remove moisture and subsequently coated with 0.5 M DHB in 100% MeOH or 10 mg/mL 9AA in 70:30 EtOH:H2O matrix solutions. The slides were spray-coated with matrix solutions using a glass Type A Meinhard nebulizer (Golden, CO, USA) using nitrogen gas at a pressure of 30 psi and a flow rate of 3 mL/min.
Primary and Secondary Antibody Cocktails and Concentrations for Rat Muscle Fiber Myosin Heavy Chain Immunofluorescence Staining. All cocktails are Prepared in 10% Normal Goat Serum in PBS
Primary antibody cocktail
Secondary antibody cocktail
Alexa Fluor 350 IgG2b
Alexa Fluor 488 IgG1
Alexa Fluor 555 IgM
Multicolor Immunofluorescence Analysis
An Aperio ScanScope FL (Vista, CA, USA) with an XCite mercury light source and an Olympus 20X/0.75 Plan Apo objective was used for scanning the fluorescent stained tissues and capturing the image at 20×; the resolution is approximately 0.468 μm/pixel.
Mass Spectrometry and Imaging Analysis
All MS experiments were performance using a Thermo LTQ XL (San Jose, CA, USA) equipped with a Lasertechnik Berlin MNL 106-LD N2 laser (Berlin, Germany) with a repetition rate of 60 Hz, wavelength at λ = 337 nm, and laser spot size at approximately 100 μm in diameter. All tissue sections were rastered with a step size of 100 μm. A laser energy of 3 μJ and 3 laser shots per spot were used for all MS experiments.
Image Generation and Data Analysis
Multicolor Immunofluorescence Analysis
All multicolor immunofluorescence images were generated using Aperio ImageScope (Vista, CA,USA). Image and contrast were adjusted to allow best visualization of different fluorescent color.
Multivariate data analysis was performed to extract features in the datasets. Comparison was made among four sample groups, SOL, EDL, oxidative region of GAS (GasOx), and glycolytic region of GAS (GasGly). PCA and PLS-DA were performed using the MetaboAnalyst web server [28, 29]. From centroid data, the m/z values and intensity lists were exported from QualBrowser into Microsoft Excel and saved as .csv files. Each m/z and intensity list was an average of 10 mass spectra (or scans). At least 10 lists per sample group were used for PCA and PLS-DA. A mass tolerance of 0.5 m/z was used to match peaks across samples. Known m/z values from the MALDI matrix were removed from the dataset. Finally, the dataset was auto-scaled and normalized by sum prior to PCA and PLSDA. Scores and loadings from PCA and PLS-DA were evaluated for significant m/z values for further analysis.
The MS images were extracted from position specific mass spectra using Thermo ImageQuest v1.0.1 software (San Jose, CA, USA). All MS images were generated by normalizing ion intensity to the total ion current (TIC) at each pixel.
Compounds were identified using an in-group database as well as by tandem MS (MSn) using collision-induced dissociation (CID) or pulsed Q collision-induced dissociation (PQD). For MS2, the isolation window was set at 1.5 u with a collision energy of 35 to 40 arbitrary units normalized to m/z 400.
Features Extracted and Putatively Identified from Positive Ion Mode Analysis and the Direction of Loading in the PCA
Positive ion mode
Higher loading area
[M + H]+
[M + H]+
[M + K]+
[M + H]+
[M + H]+
Fragment from phosphatidylcholine (18:1/18:1)
Fragment from phosphatidylcholine (18:2_20:4)
[M + Na]+
[M + Na]+
[M + K]+
[M + K]+
[M + Na]+
Features Extracted and Putatively Identified from Negative Ion Mode Analysis and the Direction of Loading in the PCA
Negative ion mode
Higher loading area
[M – H]–
[M – H]–
Glucose monophosphate (GMP)
[M – H]–
Fructose bisphosphate (FBP)
[M + Na-2H]–
Fructose bisphosphate (FBP)
[M – H]–
Uridine diphosphate (UDP)
[M – H]–
Inosine monophosphate (IMP)
[M – H]–
Adenosine diphosphate (ADP)
[M – H]–
Adenosine triphosphate (ATP)
[M – H]–
Adenosine diphosphate-ribose (ADP-ribose)
Discussion and Summary
Different metabolomics fingerprints were observed from oxidative and glycolytic skeletal muscles. Methyl histidine ([M + H]+ at m/z 170) and anserine ([M + H]+ at m/z 241) were found to be higher in glycolytic fibers. These two biomolecules are known to be present in skeletal muscles and are high in antioxidant activity [30, 31]. These two biomolecules and other histidine derivatives such as carnosine are known to be higher in animals with sprinter characteristic that require fast turnover of energy [30, 31], which explains the higher abundance in the glycolytic fibers that are responsible for fast, explosive movements. Other biomolecules localized in the glycolytic region include IMP, ADP, and ATP ([M – H]– ions at m/z 347, 426, and 506, respectively). Nucleotides can degrade to nucleosides by 5′-nucleotidase by dephosphorylation of AMP ribose and IMP, which can be produced from AMP by AMP deaminase. This leads to the accumulation of IMP in the skeletal muscles as the activity of AMP deaminase is higher than that of 5′-nucleotidase . The difference between the glycolytic and oxidative region in the abundance of IMP may be due to the slower salvage rate in the glycolytic fibers compared with oxidative fibers . In the case of ATP, this present study is in agreement with several studies on muscle biopsies of human and animals that showed higher ATP content per gram of wet weight in a predominantly fast-twitch muscle compared with muscle composed of predominantly slow-twitch fibers . Interestingly, based on our previous and current studies, we found that small metabolites such as ATP and ADP have limited stability, such that they are readily detectable by MALDI MS from fresh-cut cross sections that were immediately coated with matrix upon drying. If the cross sections were stored in the –80°C prior to coating, the signals would fluctuate. We are currently working on testing this observation.
For the oxidative region, palmitoyl (16:0) and oleoyl (18:1) carnitines ([M + H]+ at m/z 400 and 426, respectively) were found to be higher abundance when compared to the glycolytic region. This observation may be due to the fact that oxidative fibers generate most energy (i.e., ATP) by oxidative phosphorylation using fatty acid as the primary substrate . Fatty acids are supplied to the fiber either by fatty acids bound to albumin or by hydrolyzed triacylglycerides ; they were converted to acyl CoA, and then finally to acylcarnitines to enter the mitochondrial where they can undergo ß oxidation . In terms of phosphatidylcholines (PCs), most of the identified PCs showed localization and higher abundance in the oxidative region were composed of 18:1 (oleic) fatty acid tail; this observation is also in concordance with literatures [35, 36] that showed higher amount of oleic fatty acid as PC fatty acid tail in SOL. One explanation to this difference in PC fatty acid tail composition between oxidative and glycolytic regions is that there is a greater number of long-chain fatty acid transporters in oxidative muscles compared with glycolytic ; this difference in fatty acid transporter population also explains the differences in acylcarnitine abundance between the two regions.
Taurine is a sulfonic amino acid found in high concentrations in skeletal muscles and serves several functions in osmoregulation, nerve excitement modulation, antioxidation, and Ca2+ homeostasis [38, 39]. Although its abundance varies across muscles and animals species, taurine is found to be higher in slow-twitch fibers (i.e., oxidative fibers) [38, 39], and this pattern is also observed in this study. However, because the resting levels of GMP, FBP, and UDP are not documented in the literatures, their localization to the oxidative region remains an empirical question. One speculation may be that because oxidative fibers are able to generate almost all of the energy they need by oxidative phosphorylation , almost no glycolytic metabolism would occur in the oxidative fibers, leading to the accumulation of the precursor molecules GMP and FBP.
This work demonstrates the use of MALDI MSI in skeletal muscle composition analysis and reports different metabolic profiles of various muscle fiber types. MALDI MSI offers significant advantages over traditional histological staining methods since both the chemical and spatial information of the cross section are preserved and can be analyzed in the same cross section. Coupled with multivariate data analysis, MALDI MSI can be a useful tool in the study of skeletal muscle in biomarker discovery.
The authors acknowledge support for this work by the National Institute of Health grant P30 AG028740, Clinical and Translational Science Institute (CTSI, UL1 TR000064), and Southeast Center for Integrated Metabolomics (SECIM, NIH grant #U24 DK097209). The authors thank Dr. Paramita Chakrabarty from the Department of Neuroscience, Center for Translational Research in Neurodegenerative Disease (CTRND), University of Florida, for the assistance in acquiring immunofluorescence image.
- 1.Squire, J.M.: Muscle: Design, Diversity, and Disease, pp. 319–322. The Benjamin/Cummings Publishing Company, Inc, Menlo Park (1986)Google Scholar
- 2.Wortmann, R.L.: Diseases of Skeletal Muscle. Lippincott Williams and Wilkins: Philadelphia, PA, 8–12, pp. 208–209 (2000)Google Scholar
- 4.Schiller, J.: MALDI-MS of Lipids. In: MALDI MS, pp. 215–243. Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim (2007)Google Scholar
- 7.Alnaqeeb, M.A., Goldspink, G.: Changes in fiber type, number, and diameter in developing and ageing skeletal muscle. J. Anat. 153, 31–45 (1987)Google Scholar
- 13.Scott, W., Stevens, J., Binder-Macleod, S.A.: Human skeletal muscle fiber type classifications. Phys. Ther. 81(11), 1810–1816 (2001)Google Scholar
- 19.Chaves, D.F.S., Carvalho, P.C., Lima, D.B., Nicastro, H., Lorenzeti, F.M., Siqueira-Filho, M., Hirabara, S.M., Alves, P.H.M., Moresco, J.J., Yates, J.R., Lancha, A.H.: Comparative proteomic analysis of the aging soleus and extensor digitorum longus rat muscles using TMT labeling and mass spectrometry. J. Proteome Res. 12(10), 4532–4546 (2013)CrossRefGoogle Scholar
- 24.Rice, K., Linderman, J., Kinnard, R., Blough, E.: The Fischer 344/NNiaHSd × Brown Norway/BiNia is a better model of sarcopenia than the Fischer 344/NNiaHSd: a comparative analysis of muscle mass and contractile properties in aging male rat models. Biogerontology 6(5), 335–343 (2005)CrossRefGoogle Scholar
- 29.Xia, J., Mandal, R., Sinelnikov, I.V., Broadhurst, D., Wishart, D.S.: MetaboAnalyst 2.0—a comprehensive server for metabolomic data analysis. Nucleic Acids Res. 40(Web Server issue), W127–W133 (2012)Google Scholar
- 32.Brault, J.J., Terjung, R.L.: Purine salvage to adenine nucleotides in different skeletal muscle fiber types. J. Appl. Physiol. 91(1), 231–238 (2001)Google Scholar
- 34.Van Der Vusse, G.J., Reneman, R.S.: Lipid metabolism in muscle. In: Comprehensive Physiology, pp. 952–994. John Wiley and Sons, Inc. In: Ron Terjung, (ed.) 952–994. american Physiological Society by Wiley Blackwell (Methesda, MD), (2010)Google Scholar
- 35.Janovská, A., Hatzinikolas, G., Mano, M., Wittert, G.A.: The effect of dietary fat content on phospholipid fatty acid profile is muscle fiber type dependent. Am. J. Physiol. Endocrinol. Metab. 298(4), E770–E786 (2010)Google Scholar
- 37.Bonen, A., Luiken, J.J.F.P., Liu, S., Dyck, D.J., Kiens, B., Kristiansen, S., Turcotte, L.P., Van Der Vusse, G.J., Glatz, J.F.C.: Palmitate transport and fatty acid transporters in red and white muscles. Am. J. Phys. Endocrinol. Metab. 275(3), E471–E478 (1998)Google Scholar