The authors have declared that no competing interests exist.
Conceived and designed the experiments: SB BN IK AT VL BL PC AD. Performed the experiments: SB BN IK AT VL. Analyzed the data: SB FF BN IK. Contributed reagents/materials/analysis tools: SB FF BN IK BL PC AD. Wrote the paper: SB FF BN IK BL PC AD.
Insulin resistance (IR) is associated with increased production of triglyceride-rich lipoproteins of intestinal origin. In order to assess whether insulin resistance affects the proteins involved in lipid metabolism, we used two mass spectrometry based quantitative proteomics techniques to compare the intestinal proteome of 14 IR patients to that of 15 insulin sensitive (IS) control patients matched for age and waist circumference. A total of 3886 proteins were identified by the iTRAQ (Isobaric Tags for Relative and Absolute Quantitation) mass spectrometry approach and 2290 by the SWATH-MS strategy (Serial Window Acquisition of Theoretical Spectra). Using these two methods, 208 common proteins were identified with a confidence corresponding to FDR < 1%, and quantified with p-value < 0.05. The quantification of those 208 proteins has a Pearson correlation coefficient (r2) of 0.728 across the two techniques. Gene Ontology analyses of the differentially expressed proteins revealed that annotations related to lipid metabolic process and oxidation reduction process are overly represented in the set of under-expressed proteins in IR subjects. Furthermore, both methods quantified proteins of relevance to IR. These data also showed that SWATH-MS is a promising and compelling alternative to iTRAQ for protein quantitation of complex mixtures.
Insulin resistance (IR) is associated with dyslipidemia, which contributes to the pathogenesis of atherosclerosis and increases the risk of cardiovascular disease [
A number of techniques and strategies can be used for the relative comparison of protein expression between different conditions. Isotopic labeling and label-free methods are currently used for this task. There are some publications and reviews comparing both methods in the last years [
The most popular labeling approaches are iTRAQ (isobaric tag for relative and absolute quantitation), SILAC (stable isotope labeling by amino acids in cell culture) and TMT (tandem mass tag). The iTRAQ reagents as well as the TMT reagents react with peptide amino-termini or lysine residues, and hence label most peptides and proteins of the cells. Upon collision-induced dissociation (CID) or higher-energy collisional dissociation (HCD), iTRAQ or TMT reporter ions (4 or 8 for iTRAQ and 6 or 10 for TMT) are released in the MS/MS spectra. The intensity of these peaks will be used for the relative quantification of peptides and proteins.
In recent years, label-free quantification based on precursor signal intensity has gained popularity because of its fast and low-cost measurement. Peak intensity based comparative LC/MS and spectral count based LC-MS/MS are the most commonly used label-free quantification methods [
Most label-based and label-free approaches use data-dependent acquisition (DDA) where a survey scan is used to determine which precursor will be selected for product ion scanning [
There are some studies comparing labeled and label-free strategies. iTRAQ has been compared to gel and label-free LC-MS/MS [
Another group compared iTRAQ and peak-intensity-based label-free approaches on
A recent article [
Lambert et al [
In this study, we evaluated and compared the iTRAQ labeling technique and the label-free SWATH-MS strategy in duodenal biopsy of IR vs IS subjects. This picture of the differential proteome corroborates previous positions about the effects of insulin resistance and raises interesting hypothesis about the implication of other proteins such as PACAP and MARCKS in the onset of insulin resistance in duodenum.
Fourteen nondiabetic, insulin-resistant (IR) males and fifteen insulin-sensitive (IS) males matched for age and waist circumference were recruited in the Quebec City area to participate in the study. IR subjects had to have plasma triglyceride (TG) levels > 1.7 mmol/L, HDL-C < 1.1 mmol/L, plasma insulin levels > 90 ρmol/L and a waist circumference > 94 cm. Subjects were excluded if they had elevated blood pressure, monogenic hyperlipidemia such as familial hypercholesterolemia, plasma TG levels > 4.5 mmol/L, a recent history of alcohol or drug abuse, diabetes mellitus or a history of cancer. Furthermore, all participants were unrelated at the first and second degrees. The research protocol was approved by the Laval University Medical Center ethical review committee and written informed consent was obtained from each subject. Biopsies were obtained from the second portion of the duodenum during gastro-duodenoscopy. Six biopsy samples were collected using multiple sample single-use biopsy forceps and immediately flash frozen in liquid nitrogen and stored at -80°C before protein extraction.
Frozen duodenal tissue biopsies (15 IS and 14 IR patients) were weighted (6 to 20mg), and disrupted using a mortar and pestle. Samples were kept frozen on dry ice, and grinded to fine powder. Then lysis buffer (50mM ammonium bicarbonate, 50mM dithiothreitol (DTT), 0.5% sodium deoxycholate (SDC)) containing protease inhibitors cocktail (Roche) was added, and the sample preparation was homogenized on ice by sonication with a Sonic Dismembrator (Fisher) with 1 second pulse (20 times). Samples were centrifuged 10 min at 16000g. The supernatants were mixed with 5 volumes of acetone (stored at -20°C) and incubated overnight at -20°C. Precipitated proteins were centrifuged 15min at 16000g. Protein pellets were air dried, and then resuspended in 0.5M Triethylammonium bicarbonate (TEAB) containing 0.5% SDC. Finally, the protein concentration of each sample was determined by colorimetric Bradford assay.
Equal amounts of protein (50μg) from each control (IS) or IR sample were combined to give a control group and a IR group, respectively (
Duodenum biopsy samples were obtained from 15 IS volunteers and 14 IR patients. Samples were pooled and two technical replicates were generated.
Fractions collected from the isoelectric focusing were resuspended in 25 ul 0.1% formic acid. Mass spectrometry analysis was performed on a TripleTOF 5600 mass spectrometer fitted with a nanospray III ion source (AB SCIEX, Concord, ON) and coupled to an Agilent 1200 HPLC (Agilent, California). Two ul samples were injected by the Agilent 1200 autosampler onto a trapping column (Zorbax 300SB-C18) 5 u, 5 x 0.3mm at ten ul/min for desalting then onto a 0.075 mm (internal diameter) self-packed PicoFrit column (New Objective) packed with a isopropanol slurry of 5um Jupiter C18 (Phenomenex) stationary phase using a pressure vessel (Proxeon) set at 700 psi. The length of the column was 15 cm. Samples were run using a 90 min gradient from 5–35% solvent B (solvent A 0.1% formic acid in water; solvent B: 0.1% formic acid in acetonitrile) at a flow rate of 300 nl/min. Data were acquired using an ion spray voltage of 2.4 kV, curtain gaz of 30 psi, nebulizer gaz of 8 psi and an interface heater temperature of 125°C. A DDA method was set up with the MS survey range set between 400 amu and 1250 amu (250msec) followed by dependent MS/MS scans with a mass range set between 100 and 1800 amu (50m sec) of the 20 most intense ions in the high sensitivity mode with a 2+ to 5+ charge state. Dynamic exclusion was set for a period of 12 sec and a tolerance of 100 ppm. Rolling collision energy was used and iTRAQ reagent collision energy adjustment was on.
Data files were submitted for simultaneous searches using Protein Pilot version 4.5 software (AB SCIEX) utilizing the Paragon and Progroup algortihms [
The detected protein threshold in the software was set to the value which corresponded to 1% FDR. The following criteria were required to consider a protein for further statistical analysis: the
200 ug of tryptic peptides from IS and IR samples in which we spiked 20 pmol of bovine serum albumin peptides (as an internal standard) were fractionated by IEF into 14 fractions (as described above). Two hundred fifty ng from each fraction was injected on the TripleTOF 5600 (AB SCIEX) in DDA mode using the same chromatography conditions and the same acquisition parameters as used for iTRAQ (except for the extra CE for iTRAQ) and then searched against the uniprot ‘Complete Proteome’ human database (release of March 2013, 84848 sequences) with Protein Pilot 4.5. The resulting protein pilot. group file was used to generate the library which was used for SWATH processing and quantification.
Tryptic peptide samples from IR (200ng) and IS (200ng) were injected in 6 replicates in data independent acquisition (DIA) mode for SWATH analysis. 3 replicates of IR and IS contained 10 fmol of BSA and the 3 other replicates of IR and IS contained 50fmol of BSA (
Each condition is injected separately in a DDA mode to build the spectral library and in a SWATH mode. The spectral library built from the DDA runs is then used by PeakView and MarkerView to extract the peptide and the quantification information in each of the SWATH runs. BSA 1 fmol and 50 fmol was spiked in each SWATH run as an internal standard.
Data were processed with PeakView 2.0 and MarkerView 1.2. The result file from the DDA experiment used for the library generation was imported into PeakView with a protein FDR threshold of 1%. Ten peptides per protein and 10 transitions per peptide were extracted from the SWATH files. Shared peptides were excluded as well as peptides with modifications. Peptides with FDR lower than 1.0% were exported in MarkerView for the t-test.
In order to select tryptic peptides that are the most suitable for sensitive and selective protein detection, MRM analysis were performed on tryptic peptides unique to the studied proteins with length ranging from 5 to 25 amino acids. Peptides containing methionine and cysteine were eliminated. Peptides were selected using Skyline v2.5 [
Purified synthetic peptides containing 13C6 Lys and 13C6 Arg were obtained from JPT (Germany) and reconstituted in 0.1% formic acid to a final concentration of 500 pmol/μL. A solution containing 10 fmol/μL of each peptide was prepared from the stock solutions and used to reconstitute the samples after tryptic digestion for relative quantification. 1.25 ug of peptides (in 5 ul) were analyzed on a AB SCIEX 5500QTRAP hybrid triple quadrupole/linear ion trap mass spectrometer equipped with an Eksigent nanoLC AS2 cHiPLC nanoflex controlled by Analyst 1.6 and with a nanospray ionization source. MS analysis was conducted in positive ion mode with an ion spray voltage of 2300V. Peptides were desalted on a 200um x 0.5 mm chip trap column packed with ChromXP C18, 3 um, (Eksigent) at 2 ul/min of Solvent A (formic 0.1%) then switched in line at a flow rate of 300 nL/min on a 75um x 15 cm chip column packed with ChromXP C18, 3 um (Eksigent) with a 20 min linear gradient from 5 to 25% of solvent B (ACN 0.1% FA), then a 2 min linear gradient from 25 to 80% B, followed by a 8 min linear gradient. Nebulizer gas was set to 8 (Gas1), curtain gas to 20, heater to 150°C and declustering potential (DP) to 70 V. LC-MRM/MS analyses were performed using three transitions on two peptides for each of the target proteins and quantification done with MultiQuant 2.1 was based on the relative areas of the SIS and endogenous peptides. The MRM transition that gave the highest area counts was used for the quantitation, with the other two transitions acting as qualifier transitions to confirm peptide retention times and the fragment ion ratios. A blank solvent injection was run between biological samples to prevent sample carryover on the HPLC column and the samples were injected in random order. Samples were analyzed in duplicate. Samples containing 5 fmol of digested BSA were injected periodically in order to confirm system stability.
Statistical analysis was done on the R/Bioconductor plateform [
In this study, we evaluated and compared two quantitative proteomics strategies, the iTRAQ labeling technique and the label-free SWATH-MS strategy, to discover the proteins that are potentially associated with insulin resistance in duodenal biopsy samples from IR and IS subjects.
Tryptic peptides of proteins extracted from duodenal biopsy samples of IS and IR patients were labeled with the four-plex iTRAQ reagents in duplicates. After labeling, sample IR #1 was pooled with sample IS#1 and sample IR#2 was pooled with sample IS#2 creating two replicates. Those two technical replicates were treated separately for the subsequent steps according to
Using Protein Pilot, a total of 3886 protein groups were identified in at least one iTRAQ replicates with global FDR < 1% (
For the SWATH analysis, a spectral library of 2290 proteins was created with FDR <1% by injecting a new set of 14 fractions with a Data Dependent Acquistion method on the 5600 triple-TOF mass spectrometer (
A SWATH experiment was performed on 6 replicates of IS and 6 of IR from which 3 samples of IR and 3 of IS were spiked with BSA 1fmol and the 3 others with BSA 50 fmol (
The library and the 12 SWATH files were uploaded into PeakView. One thousand three hundred twenty-six proteins with at least one peptide with a FDR < 1% were exported into MarkerView. Marker View performed t-test comparing IS and IR groups for each protein: 798 proteins were quantified with p-value below 0.05 (
In this study, more proteins were detected by iTRAQ than in the SWATH experiment since the iTRAQ protein identifications came from the two replicates of 14 fractions whereas the SWATH library was created using 14 fractions. In order to increase the SWATH library, the iTRAQ dataset was used to build the SWATH library by using a conversion option designed for this in the PeakView software. Unfortunately, we observed that the peptides detected from the iTRAQ experiment often produced ions of higher charge state than the same peptide from a non-labeled experiment and were not the ones giving the best signals in the SWATH experiment. This probably explains why using the data from the iTRAQ experiment as a library for SWATH produced sub-optimal results. To increase the size of the SWATH library, more injections from duodenal extract would be needed.
Two thousand eighty-two proteins were commonly identified between the two techniques at FDR <1% (
(A) Comparison in the number of protein identification at FDR<1% (B) Comparison in the number of proteins quantified at p-value <0.05.
We next assessed the level of correlation between quantification by iTRAQ-label and by SWATH-MS (
A) Scatter plot analysis of the 208 quantified proteins by iTRAQ and SWATH B) Volcano plot of SWATH ratios vs iTRAQ C) Comparison between SWATH and iTRAQ ratios for the 208 quantified proteins D) Density plot of iTRAQ and SWATH ratios for all differential ratios with p-value <0.05.
The main drawbacks of the iTRAQ or TMT strategies are the underestimation of the fold change (compression effect) [
In SWATH, many precursors are selected in the same window for MS/MS but the chances that all the fragment ions have the same mass is slim and the ratio is not affected as in iTRAQ.
Since quantification experiments often produce a classification into ‘over-expressed’ and ‘under-expressed’ lists, the direction of the fold-change is especially important. Indeed, a large difference in quantitation of log2 ratio of 0.5 and 3 (corresponding to ‘fold changes’ of 1.4 and 8 respectively) will have little impact on downstream gene-ontology and pathway analyses. But a comparatively smaller difference in quantitation between reported log2 ratios of -0.4 and 0.4 will have important implications for downstream analyses. Our results show that iTRAQ and SWATH agree on the ‘direction’ of the fold change over 92% of the time when differential ratios are chosen solely on the basis of a t-test p-value threshold. (See
It should be noted that SWATH seems to have a higher precision than iTRAQ since for the same p-value threshold SWATH reports differential ratios that are closer to 0 (see the area around 0 in
To evaluate the compression effect of the SWATH quantitation, we spiked a bovine serum albumin digest into both conditions (IR and IS). 1fmol of BSA digest was spiked into 3 replicates of IS and 3 replicates of IR and 50 fmol of BSA was spiked in 3 other replicates of IS and IR. The SWATH experiment reported ratio values of 29.5 +/- 5 (mean +/- SD for the three replicates) (
We used the proteins reported as differentially expressed by both the iTRAQ and SWATH experiments and performed Gene Ontology analysis using the three main ontologies (Cellular Component, Biological Process and Molecular Function). We found that proteins from the under-expressed list were far more likely to be associated with metabolic process and oxydo-reduction process than a random selection of proteins (p-values of 3e-5 and 7e-5 respectively) (
Bioinformatic analyses reveal that proteins involved in lipid metabolic process and oxydation-reduction processed are mostly underexpressed in insulin resistant patients.
One of the objectives of the study was to find regulated proteins associated with insulin resistance and lipid metabolism,
Results of the different quantitative proteomics analyses on selected proteins involved in lipid metabolism and insulin regulation.
However, one can observe that the ratio difference is greater in the MRM targeted experiment than in the SWATH and iTRAQ experiments. As discussed earlier, it is due to the compression effect observed with iTRAQ [
Only a few human studies have examined the potential mechanisms underlying the oversecretion of intestinal lipoproteins in individuals with IR. Our previous study [
Our results shows that proteins associated with lipid metabolism and catabolic processes are clearly underexpressed in insulin resistance, thus confirming the results obtained in our previous paper [
Indeed, both iTRAQ and SWATH results reveal that the protein MTTP (Microsomal Triglyceride transfer protein) and MGAT2 are down regulated in insulin resistant patients which has been shown previously by real-time PCR quantitation and MRM analyses [
ACSL5 expression was also decreased in IR (
On the other hand, an increase in MARCKS levels (
The protein plasma cell-induced resident endoplasmic reticulum protein (also named PACAP) was found to be over-expressed in IR patients (
By using 2 different quantitative approaches, this study have shown that insulin resistance affects the proteins involved in lipid metabolism. A comparison of proteomics quantitation using iTRAQ and SWATH methods was presented. It was shown that results from both methods can have a high degree of correlation both in terms of actual quantitation and in terms of the ‘direction’ of the fold change. The methods also appear to have complementary aspects: for a given instrument time, the iTRAQ experiment will provide more protein identifications; however, the SWATH experiment can identify more differential ratios, being both more sensitive in the detection of small differential ratios and being able to cover a larger dynamic range. The differential proteins identified both confirm previous experiments (for example, the roles of MTTP, MGAT2 and ACSL5), and raises interesting new hypothesis, as with MARCKS and PACAP. Proteins of relevance to IR were mostly associated with metabolic and oxydo-reduction process and were found using both iTRAQ and SWATH-MS. These data also showed that SWATH-MS is a promising and compelling alternative to iTRAQ for protein quantitation of complex mixtures.
(TIFF)
(XLS)
(XLSX)
(XLSX)
(XLS)
(XLS)
(XLS)
The authors thank Brigitte Simons for helpful discussion on SWATH-MS and Joana Hunter for revision of the manuscript. Benoît Lamarche leads the Chair on Nutrition at Laval University.