Background
Adipose triglyceride lipase (ATGL) deficiency causes the onset of neutral lipid storage disease with myopathy (NLSDM), a rare genetic disorder which is transmitted as an autosomal recessive trait [
1,
2]. Patients with NLSDM have either homozygous or compound heterozygous mutations in the ATGL gene. ATGL catalyzes the rate-limiting step in the hydrolysis of TG stored in lipid droplets [
3,
4]. ATGL deficiency is characterized by the presence of intracellular triglyceride (TG) deposition in most tissues, including leukocytes (Jordans’ anomaly), skeletal muscles and the heart [
1,
2]. Clinically, the patients reported so far show primarily skeletal muscle weakness associated with myopathy, with highly elevated creatine kinase levels, and show cardiomyopathy usually observed at later stages of the disease [
2,
5].
One of the NLSDM phenotypes, discovered in Japan, shows massive TG accumulation in both the coronary artery and myocardium, resulting in severe heart failure. This symptom has been designated “Triglyceride deposit cardiomyovasculopathy (TGCV)” [
6,
7]. Cardiomyopathy is lethal in patients with TGCV and necessitates cardiac transplantation; thus, therapeutic methods for reducing the burden on patients with this intractable disease are highly desirable. In addition, sensitive noninvasive biomarkers are needed for monitoring the therapeutic response rather than diagnostic biomarkers for detecting this disease. A clear understanding of the pathogenesis of TGCV is essential to exploit the therapeutic methods. However, the molecular nature of TGCV remains unknown to date.
As proteins are almost always the effectors of cellular functions, the use of proteomic approaches to decipher the molecular basis of diseases might offer new insight. Therefore, in this study, we performed proteomic analysis to examine differentially expressed proteins in TGCV patient cells. We employed the stable isotope labeling with amino acids in cell culture (SILAC) method coupled with LC-MS/MS [
8] and performed large-scale quantitative proteomic analysis.
The list of differentially expressed proteins obtained by comprehensive quantitative analysis needs to be validated, but this step is an expensive and time-consuming process requiring a pre-existing antibody. A more recent mass spectrometry-based technique called selected/multiple reaction monitoring (SRM/MRM) is a useful method for the validation of proteins without antibodies [
9]. In this study, we confirmed differentially expressed proteins identified by SILAC proteomics using SRM/MRM. As a result, we identified for the first time differentially expressed proteins between TGCV patient and healthy volunteer cells. Our identified proteins will be useful for elucidation of the pathogenesis of TGCV and the exploitation of therapeutic methods for TGCV in the future.
Methods
Cell culture
Skin fibroblast cells were isolated from the skin of two TGCV patients in the Department of Cardiovascular Medicine, Osaka University Graduate School of Medicine, Japan. The local ethics committee approved the study and informed consent was obtained from the donors. Cells were cultured in Iscove’s modified Dulbecco’s medium (IMDM) supplemented with 10% fetal bovine serum (FBS; Invitrogen, Carlsbad, CA, USA) and penicillin/streptomycin (100 IU/ml). For SILAC experiments, Dulbecco's modified Eagle’s medium without L-arginie and L-lysine (Invitrogen) was supplemented with dialyzed FBS (Invitrogen). The medium was then divided into three portions and supplemented with 13C6,15 N4 L-arginine and 13C6, 15 N2 L-lysine or 13C6 L-arginine and 4,4,5,5D4 L-lysine or normal L-arginine and L-lysine, to produce “heavy” or “medium” or “light” SILAC medium, respectively. All isotopes labeled L-arginine and L-lysine were purchased from Cambridge Isotope Laboratories (Tewksbury, MA, USA). Skin fibroblast cells were grown in SILAC medium for at least 6 doubling times to ensure that the amino acids had been fully incorporated.
Oil red O lipid staining
Cells were cultured on chamber slides (BD Biosciences, San Jose, CA, USA). The staining solution was prepared by dissolving 0.5 g Oil Red O powder (Sigma-Aldrich, St Louis, MO, USA) in 100 ml isopropanol, followed by 1:1 dilution with distilled water. Then, the solution was allowed to stand for 10 min before it was filtered through filter paper. The cells were washed in PBS, fixed with 10% formalin at room temperature for 10 min, and stained with Oil Red O at room temperature for 20 min. Samples were washed with distilled water and counterstained with Mayer's hematoxylin (Wako Pure Chemical Industries, Osaka, Japan) for 2 min.
Preparation of protein samples, 1-D SDS-PAGE separation and in-gel trypsin digestion
Cells were scraped into ice-cold RIPA buffer containing 50 mM Tris–HCl, pH 8.0, 150 mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate, 0.1% SDS, with 1 × protease inhibitor, compete Mini (Roche Applied Science, Indianapolis, IN, USA) and agitated at 4°C for 30 min. After centrifugation at 14000 rpm at 4°C for 20 min, lysates were collected and protein concentrations were determined using a DC Protein Assay (Bio-Rad, Hercules, CA, USA). Equal amounts (in weight) of lysates from three light (Arg0, Lys0) labeled control cells were pooled to generate the control, and then lysates of three labeled types, light (Arg0, Lys0), medium (Arg6, Lys4) and heavy (Arg10, Lys8), were mixed in equal amount (in weight). A 100 μg sample of mixed proteins was separated by SDS-PAGE using NuPAGE Novex 4-12% Bis-Tris (Invitrogen). Bands were visualized with Gels Simply Blue Safe Stain (Invitrogen) and lanes were sliced into 46 sections. After distaining with 50% ethanol in 50 mM ammonium bicarbonate (ABC), proteins in gel pieces were reduced with 10 mM DTT in ABC and alkylated by 50 mM iodoacetamide in ABC. After gel dehydration with 100% ethanol, the gel pieces were covered with approximately ~40 μl of 12.5 μg/mL trypsin in ABC and in-gel digestion was performed at 37°C for 16 h. Peptides were extracted from gels with 3% trifluoroacetic acid (TFA), 30% acetonitrile (ACN) and then 100% ACN. The resulting peptide mixtures were dried and resolved with 0.1% TFA and 2% ACN, and desalted using C18 stage Tips [
10].
NanoLC-MS/MS
NanoLC-MS/MS analysis was conducted using an LTQ-Orbitrap Velos mass spectrometer (Thermo Fisher Scientific, Bremen, Germany) equipped with a nanoLC interface (AMR, Tokyo, Japan), a nano HPLC system (Paradigm MS2; Michrom Bioresources, Auburn, CA, USA), and an HTC-PAL autosampler (CTC Analytics, Zwingen, Switzerland). L-column2 C18 particles (Chemicals Evaluation and Research Institute, Kurume, Japan) were packed into a self-pulled needle (200 mm length × 100 μm inner diameter) using a Nanobaume capillary column packer (Western Fluids Engineering, Wildomar, CA, USA). The mobile phases consisted of (A) 0.1% formic acid and 2% ACN and (B) 0.1% formic acid and 90% ACN. The peptides dissolved in 2% ACN and 0.1% TFA were loaded onto a trap column (0.3 × 5 mm, L-column ODS; CERI, Japan). The nanoLC gradient was delivered at 500 nl/min and consisted of a linear gradient of mobile phase B developed from 5 to 35% in 135 min. A spray voltage of 2000 V was applied.
Data acquisition with LTQ-Orbitrap Velos
Full MS scans were performed in the orbitrap mass analyzer of LTQ-Orbitrap Velos (scan range 350–1500 m/z, with 30 K full width at half maximum (FWHM) resolution at 400 m/z). In MS scans, the 10 most intense precursor ions were selected for MS/MS scans with the LTQ-Orbitrap Velos. The dynamic exclusion option was implemented with a repeat count of 1 and exclusion duration of 60 s. This was followed by collision-induced dissociation (CID) MS/MS scans of the selected ions performed in the linear ion trap mass analyzer. The values of automated gain control (AGC) were set to 1.00 × 106 for full MS and 1.00 × 104 for CID MS/MS. Normalized collision energy values were set to 35%.
Protein identification and quantification using MaxQuant software
The resulting mass spectra were analyzed using MaxQuant Software (version 1.2.2.5) [
11]. In short, raw MS files were loaded directly into MaxQuant, and identification and quantitation of individual peptides were generated in protein groups. The MaxQuant searches were executed using the International Protein Index (IPI) human protein database (version 3.67, 87302 forward and 87302 reverse protein sequences). All entries were filtered using a false positive rate of 1%. The following search parameters were used: two missed cleavages permitted, carbamidomethylation on cysteine fixed modification, and oxidation (methionine) and acetyl (N-terminus proteins) variable modifications. The mass tolerances for precursor ions and fragment ions were 20 ppm and 0.5 Da, respectively. Quantification of proteins was based on the normalized heavy/light (H/L) and medium/light (M/L) ratios as determined by MaxQuant.
Stable isotope-labeled peptides
Proteotypic peptides were chosen based on SILAC proteomics data. For SRM/MRM analysis of the 20 target proteins, 42 stable isotope-labeled peptides (SI-peptides, crude peptide: approximately 50% peptide purity and >99% isotope purity; Greiner Bio One, Frickenhausen, Germany) were synthesized. SI-peptides had isotope-labeled lysine or arginine at their C-terminus. Each SI-peptide was dissolved in distilled 40% ACN and 0.1% TFA, and stored at −80°C.
Creation of SRM/MRM transition list using SI-peptides
The mixture of SI-peptides was analyzed by LC-MS/MS using LTQ-Orbitrap XL (CID mode), and an msf file was generated using Proteome Discoverer and Mascot. The msf file was opened with Pinpoint software (version 2.3.0; Thermo Fisher Scientific), and a list of MS/MS fragment ions derived from SI-peptides was generated. Four MS/MS fragment ions were selected for SRM/MRM transitions of each targeted peptide based on the following criteria: y-ion series, strong ion intensity and at least 2 amino acids in length.
Optimization and operation of SRM/MRM method
SRM/MRM methods for each SI-peptide were created by Pinpoint 1.0, which included SRM/MRM transition lists and the instrument method with the following parameters: scan width of 0.002 m/z, Q1 resolution of 0.7 FWHM, cycle time of 1 s, and gas pressure of 1.8 mTorr. The SI-peptide mixture was analyzed by LC-SRM/MRM on the TSQ-Vantage triple quadruple mass spectrometer (Thermo Fisher Scientific) equipped with the parameters mentioned above. The nanoLC gradient was delivered at 500 nl/min and a spray voltage of 1900–2000 V was applied. Test runs of the SI-peptide mixture were performed to establish the retention time window (±2 min) for each peptide ion and optimize the collision energy for each transition. Four transitions were chosen for each peptide and all fragment ions were y-ions. When possible, two peptides were used per protein and all SRM/MRM analyses were run in duplicate. For SRM/MRM analyses, proteins were prepared as follows. Skin fibroblast cells cultured in normal IMDM medium were harvested, and proteolytic digestion were performed by a phase-transfer surfactant (PTS) protocol using an MPEX PTS reagent kit (GL Sciences, Tokyo, Japan) [
12]. Briefly, pellets were lysed with PTS B buffer followed by sonication for 5 min using a Bioruptor sonicator (Cosmo Bio, Tokyo, Japan). Lysed proteins were reduced with 10 mM DTT, alkylated with 50 mM iodoacetamide and sequentially digested by 1:100 (w/w) trypsin for 16 h at 37°C. An equal volume of ethylacetate was added to the digested samples, and the mixtures were acidified by 1% TFA followed by vortexing to transfer the detergents to the organic phase. After centrifugation, the aqueous phase containing peptides was collected. The resulting peptide mixtures were desalted using C18 stage Tips and resolved with 0.1% TFA and 2% ACN. Two micrograms of digested sample were transferred to a new tube and the SI-peptide mixture was added. The amount of SI-peptide was optimized to achieve similar ion intensity to the corresponding endogenous peptide, if possible. Samples were analyzed by LC-SRM/MRM on the TSQ-Vantage using the optimized SRM/MRM method.
SRM/MRM data analysis
SRM/MRM data were processed using Pinpoint 1.0. The peak area in the chromatogram of each SRM/MRM transition was calculated, and the values of endogenous targeted peptides were normalized to those of the corresponding SI-peptides. When the transition profile was different between endogenous peptides and SI-peptides, the transition was excluded from the quantification process. The different transition profile might have been caused by the detection of untargeted peptides. In addition, transitions having signal-to-noise ratios (S/N) of <10 were discarded from this study. In such cases, only peptides having more than one transition were used.
Protein accession numbers identified by SILAC proteomics analysis and their corresponding fold changes were imported into Ingenuity Pathway Analysis (IPA) software (Ingenuity Systems, Redwood City, CA, USA,
http://www.ingenuity.com) for network and functional analysis. IPA is web-based software that constructs protein interaction networks in silico based upon published associations that have been collated from the literature.
Microarray analysis
Skin fibroblast cells derived from two TGCV patient cells and two control cells (LC1, LC2, control1 and control2) were cultured in SILAC medium. Cells were harvested, and total RNA was isolated using a QIA shredder (Qiagen, Valencia, CA, USA) and an RNeasy Mini Kit (Qiagen) according to the manufacturer’s instructions. The quantity and purity of RNA were evaluated using a NanoDrop (Nanodrop Technologies, Wilmington, DE, USA) and an Agilent 2100 Bioanalyzer (Agilent Biotechnologies, Santa Clara, CA, USA). RNA from two control cells was equally mixed prior to array analysis. Analysis using Human Genome U133 Plus 2.0 microarray (Affymetrix, Santa Clara, CA, USA) was performed according to the standard Affymetrix protocol. Briefly, total RNA was reverse-transcribed into complementary DNA (cDNA) and cDNA was in vitro transcribed to biotinylated RNA. After fragmentation, 12.5 μg biotinylated RNA was hybridized overnight to the array, and then arrays were washed, stained with streptavidin–phycoerythrin and scanned using a GeneChip 3000 7G scanner (Affymetrix). The acquisition and initial quantification of array images were conducted using GeneChip Command Console Software (Affymetrix), then data were analyzed by GeneSpring GX Software (Affymetrix). We used present and absent call filtering to determine which transcripts were able to be considered accurately measured, which is a standard method for microarray analysis [
13].
Real-time reverse transcription PCR
Total RNA was isolated from skin fibroblast cells using the same method as described above, except that cells were cultured in normal medium. Complementary DNA was synthesized using the First Strand cDNA Synthesis Kit for RT-PCR (AMV) (Roche Applied Science). Real-time reverse transcription (RT) -PCR was performed using the Power SYBR Green PCR Master Mix (Applied Biosystems, Foster City, CA, USA) and ABI Prism 7900HT (Applied Biosystems). The following PCR primers were used: FBN2, 5′-GAAGTATTCATGAACCTGATC-3′ (forward) and 5′-GGTTGAACTTCATGTTGACGG-3′ (reverse); CXCR7, 5′-ATGTCACACAGTGCCTGTCGC-3′ (forward) and 5′-ATGAAGGCCTTCATCAGCTCG-3′ (reverse); PLA2G4A, 5′-ACCCAAGAATCCTGATATGGAG-3′ (forward) and 5′-CCTGGAGCCTTGTACTTTCTG-3′ (reverse); FLG, 5′-TAGACACTCTCAGCACGGAAGTG-3′ (forward) and 5′-CCTGGGTCCTTATTAATATACG-3′ (reverse); FBL2, 5′-GGGAACTGTGGGCTGTACTAC-3′ (forward) and 5′-AAATCCCATTACGGACACCTCT-3′ (reverse); RAB27B, 5′-CCTACCAGATCAGAGGGAAGTC-3′ (forward) and 5′-CATTCTGTCCAGTTGCTGCAC-3′ (reverse); MC4R, 5′-AGGTGCCAATATGAAGGGAGCG-3′ (forward) and 5′-GGATTCTGAGGACAAGAGATGTAG-3′ (reverse); SLC16A6, 5′-TCAGAGCATAGCAGGACTGGC-3′ (forward) and 5′-AGGCCCTGCTGTAGATCTTAC-3′ (reverse); TNFRSF21, 5′-TCTCCGCTGTGACTCTACATC-3′ (forward) and 5′-TACCTGCCGCAACACTGTGTC-3′ (reverse); NCAM1, 5′-GAAAGATGAGTCCAAGGAGCC-3′ (forward) and 5′-TCCGTCAGTGGCGTGGTCTCG-3′ (reverse); PLIN2, 5′-TTGGATATGATGATACTGATG-3′ (forward) and 5′-ACGTGGTCTGGAGCTGCTGAG-3′ (reverse); RPS18, 5′-TTTGCGAGTACTCAACACCAAC-3′ (forward) and 5′-AGCATATCTTCGGCCCACACC-3′ (reverse). The relative mRNA levels of each gene were normalized to RPS18 expression.
Discussion
Genetic mutations in ATGL are known causes of TGCV, but the pathogenesis mechanism remains unclear. Here we examined proteomic profiles of skin fibroblast cells derived from patients to elucidate pathogenesis processes and find clinically relevant proteins for TGCV. We identified over 4500 proteins, and specified over 50 proteins to be differentially expressed between TGCV and healthy control cells. To our knowledge, this is the most comprehensive quantitative study to date, aiming to understand the pathogenesis mechanisms of ATGL deficiency. We have provided a comprehensive proteome database that is essential to discover the key molecules underlying ATGL deficiency.
The main findings of our study are as follows. (1) Bio-function analysis revealed that proteins associated with “Lipid Metabolism” were affected by the cellular status of TGCV. Those proteins, such as DPP4, PON2, PTGS1 and PLIN2, are known to be key molecules in the metabolic process. (2) Network analysis revealed that the main network of altered proteins and genes in TGCV cells was tightly associated with connective tissue disorders. (3) Candidate proteins responsible for TGCV pathogenesis were discovered, such as PLIN2, CTHRC1 and FLG, and their altered expressions were confirmed by SRM/MRM and qPCR.
IPA analysis identified “Lipid Metabolism” as the top-ranked bio-function of altered proteins in TGCV cells. Proteins involved in this function are well-known key molecules in the metabolic process, as described in the Results section (Table
2) and are therefore candidate targets of therapy. For example, DPP4 is a serine protease, that specifically degrades incretin hormones; thus, its inhibitor is considered as a useful drug for type 2 diabetes mellitus [
37]. Recently, DPP4 inhibition was also reported to be associated with an improved cardiovascular profile, although its precise role is still unknown [
38]. PON2 possesses anti-oxidant properties, protecting cells from oxidative stress [
39] and has been suggested to exert protection against macrophage TG accumulation and oxidative stress [
40]. PTGS1, also known as cyclooxygenase 1, plays a pivotal role in the biosynthesis of prostanoids and thus regulation of PTGS1 activity has been considered in the therapy for various diseases [
23,
41]. Therefore, these proteins are considerable candidates as therapy targets for TGCV, even though their involvement in TGCV pathogenesis is unknown.
Reduced expression of PTGS1 was identified in both proteomic and transcriptomic analysis (Table
1 and Additional file
7: Table S6). In addition, reduced expression of PTGS2, which is the key enzyme in prostaglandin biosynthesis similar to PTGS1 [
42], was also detected at the transcription level (Additional file
7: Table S6). In contrast, over-expression of PLA2G4A, which is the enzyme that releases arachidonic acid from the phospholipid [
43], was observed at the transcription level and confirmed by qPCR (Additional file
7: Table S6 and Additional file
8: Figure S2). From these results, we speculate that aberration of the arachidonic acid cascade may occur in TGCV cells. However, further investigation is needed for elucidation.
IPA network analysis also revealed that the top-ranked networks of differentially expressed proteins and genes were tightly associated with connective tissue-related molecules (Additional file
2: Table S2 and Additional file
9: Table S7). For example, COL18A1, COL6A3, COL8A1, FBLN1, LAMA1, LAMA2 and CSPG4 are well-known extracellular matrix proteins [
24‐
26,
30,
31,
34]. TGCV patients showed marked characteristics, such as massive accumulation of TG in the skeletal, heart muscles and coronary artery. Aberration of molecules composed of connective tissues appeared to be involved in this symptom, but its precise role in the pathogenesis is still unclear and awaits further investigation.
Of the top-ranked over-expressed proteins, we focused on three proteins, PLIN2, CTHRC1 and FLG, whose over-expression was successfully confirmed by SRM/MRM. These proteins are interesting in the context of lipid metabolism disorders as discussed below.
PLIN2, is well-known to play a role in the formation of intracellular lipid droplets, and therefore to promote neutral lipid stores, particularly in non-adipose tissues [
22]. It exhibited a 2.1-fold increase in TGCV patient cells and was successfully confirmed in SRM/MRM analysis (Table
1, Figure
4A and D). Straub et al. described that PLIN2 was a general marker for a variety of human diseases associated with lipid droplet accumulation [
28]. In this study, we showed that PLIN2 was over-expressed in TGCV patient cells at the protein level, which was in good agreement with previous studies in the context of lipid storage diseases. In contrast, results from both array and qPCR analysis indicated that PLIN2 expression did not differ at mRNA levels between TGCV patient and control cells (Figure
6). From these results, we suggest that PLIN2 may be stabilized at the protein level during the progression of lipid droplets in TGCV patient cells. A possible reason is that PLIN2 was protected from the proteolysis pathway through the interaction with lipid droplets. However, more studies are needed to reach a firm conclusion.
One of the top-ranked over-expressed proteins, CTHRC1, which was 3-fold elevated in patient cells, was successfully confirmed in SRM/MRM analysis (Figure
4A and C). CTHRC1 is a secreted protein that has activity to repress collagen matrix synthesis during vascular remodeling [
44]. Therefore, it appears to be involved in the generation of atherosclerosis lesions in TGCV. Interestingly, a recent study by Stohn et al. identified CTHRC1 as a novel circulating protein having hormone-like metabolic effects [
27]. Livers from CTHRC1 null mice accumulated vast quantities of lipid, leading to extensive macrovesicular steatosis. Thus, CTHRC1 is likely to have the ability to repress the generation of lipid droplets and be over-expressed as it reflects the deposition of TG in TGCV patient cells. It should be noted that CTHRC1 may be a possible candidate biomarker for TGCV as it is a secreted protein. However, its precise role in TGCV pathogenesis and value as a biomarker for TGCV await further investigation of patients.
The other top-ranked over-expressed protein, FLG, was 8-fold elevated (Table
1). FLG is a key molecule that facilitates terminal differentiation of the epidermis and formation of the protective skin barrier [
45]. Lipid organization is well-known to play an important role in epidermal differentiation [
46,
47]. Thus, one question is whether FLG over-expression occurs as a result of a disorder of epidermal differentiation, which is affected by lipid metabolism failure. However, this might be contradictory. FLG acts as a major component in the outer layer of the epidermis [
29,
45]. On the other hand, skin fibroblast cells exist in the dermis, separated from the epidermis by a basement membrane. Therefore, it is unlikely that FLG over-expression in skin fibroblast cells is a result of a disorder of epidermal differentiation. Over-expression of FLG may rather be involved in TGCV pathological conditions, although the contributions of FLG to the TGCV pathogenesis process remain to be elucidated.
Competing interests
The authors declare no conflicts of interest.
Authors’ contributions
YH performed experiments and analysis, and drafted the manuscript. NK performed experiments. YH and SW operated the SRM instrument. JA participated in the design of the study. KH participated in the design of the study and coordination, and helped to draft the manuscript. TT designed the study and wrote the manuscript. All authors read and approved the final manuscript.