Article pubs.acs.org/jpr
Quantitative Analytical Method for Determining the Levels of Gastric Inhibitory Polypeptides GIP1−42 and GIP3−42 in Human Plasma Using LC−MS/MS/MS Atsushi Miyachi,*,† Takayo Murase,† Yuichiro Yamada,‡ Takeshi Osonoi,§ and Ken-ichi Harada∥ †
Laboratory Management Department, Sanwa Kagaku Kenkyusho Co., Ltd., Mie 511-0406, Japan Department of Endocrinology and Diabetes and Geriatric Medicine, Akita University Graduate School of Medicine, Akita 010-8543, Japan § Naka Memorial Clinic, Ibaraki 311-0113, Japan ∥ Graduate School of Environmental and Human Science and Faculty of Pharmacy, Meijo University, Nagoya, Aichi 468-8503, Japan ‡
S Supporting Information *
ABSTRACT: Gastric inhibitory polypeptide (GIP), an incretin, is an important subject in endocrinology. Some LC−MS assays have been proposed; however, their sensitivities are insufficient for the study of endogenous human incretin. Here, we describe a nanoflow LC hybrid triple quadrupole/linear ion trap MS assay for the simultaneous quantification of GIP1−42 and GIP3−42 from human plasma. We selected the surrogate peptide to avoid oxidative modification, and the endoproteinase Asp-N was selected for the proteolysis of GIP1−42 and GIP3−42. The phenylalanine residue at position 6 in both GIP1−42 and GIP3−42 was substituted with 13C9,15Nlabeled phenylalanine, and these substituted GIPs were used as the internal standards. This facilitated accurate and precise quantification because large corrections are possible at all steps of sample pretreatment and ionization efficiency. The lower limit of quantification was 1 pM for GIP1−42 and 10 pM for GIP3−42 by using 200 μL of plasma. Quantification of GIP1−42 and GIP3−42 in plasma from patients with type 2 diabetes was possible using this method, which included protein precipitation, Asp-N proteolysis, solid-phase extraction, nanoflow LC, and positive-ion multiple reaction monitoring cubed (MRM3) for GIP1−8, and MRM for GIP3−8 to achieve accurate, precise, and quantitative analysis that can be validated to support large clinical trials. KEYWORDS: gastric inhibitory polypeptide, incretin, absolute quantification, endoproteinase Asp-N, hybrid triple quadrupole/linear ion trap mass spectrometer, MS/MS/MS, MRM3, charcoal-stripped plasma
■
to yield inactive GIP3−42.6 Therefore, knowledge of the concentrations of active GIP1−42 and total GIP (the sum of active GIP1−42 and inactive GIP3−42) is essential to confirm the efficacy of not only these incretin-related drugs but also other antidiabetic drugs such as α-glucosidase inhibitors (α-GI), which inhibit the absorption of glucose from the intestinal gut and result in decreased secretion of GIP. Currently, immunoassay- and mass spectrometry-based methods have been used for the absolute quantification of peptide and protein. Although the former is sensitive, its selectivity is not enough to distinguish intact and posttranslationally modified peptides and metabolites, and it is difficult to simultaneously quantify several numbers of target peptides or proteins. In many cases, suitable antibodies may not be available for each protein analyte and are more unlikely to be
INTRODUCTION
Gastric inhibitory polypeptide (GIP) is a 42-amino-acid peptide hormone released from duodenal K-cells after the absorption of glucose or fat. Similar to its structurally related peptide glucagon-like peptide-1 (GLP-1), which is released from intestinal L-cells, GIP is considered to be an incretin hormone.1−3 In addition to its incretin effect, GIP is believed to increase the activity of lipoprotein lipase (LPL), which is a key enzyme of lipid metabolism, as well as to promote the glucose absorption of adipose cells in the presence of insulin;4 it is also associated with bone metabolism.5 Because some physiological roles of GIP are still unknown, the quantitative evaluation of GIP secretion and active GIP1−42 levels is important. Recently, incretin-related drugs, such as GLP-1 receptor agonists and dipeptidyl peptidase-4 (DPP-4) inhibitors, have been launched and have attracted attention as antidiabetic drugs that are not prone to causing hypoglycemia. Active GIP1−42 as well as GLP-1 are degraded rapidly by DPP-4 © 2013 American Chemical Society
Received: January 23, 2013 Published: April 23, 2013 2690
dx.doi.org/10.1021/pr400069f | J. Proteome Res. 2013, 12, 2690−2699
Journal of Proteome Research
Article
treated with dipeptidyl peptidase-4 (DPP-4) inhibitor and αGI.
obtained for active GIP1−42. Additionally, it should be noted that the plasma GLP-1 levels measured using different immunoassay kits are not consistent, and pretreatment of plasma samples has been reported to be important.7 Plasma levels of active GIP1−42 and total GIP have been quantified using some immunoassays in many clinical trials.8−10 Although an anti-GIP1−42 antibody has been developed by Deacon et al.,11 it is not commercially available. Therefore, a specific and sensitive method for quantification of active GIP and total GIP is required. Owing to its high specificity, sensitivity, and relatively short method development time, there has been an increasing interest in using liquid chromatography mass spectrometry (LC−MS) for the quantification of peptides from complex biological matrices.12−14 There have been several reports on the quantification of GIP using LC−MS (or LC−MS/MS).15−17 Wolf et al. has successfully quantified GIP1−42 and GIP3−42 in human plasma by LC−MS after enrichment using immunoprecipitation (IP) with an originally developed antibody that recognizes GIP1−42 and GIP3−42.15 However, they needed 1.9 mL of plasma for quantification, and in cases where there are other measurement items by a clinical trial, this amount would not be practical. Subsequently, they produced an antibody that recognizes GLP-17−36 and GLP-19−36 in addition to GIP1−42 and GIP3−42 and developed a LC−MS method to quantify them simultaneously after enrichment with this antibody.16 However, 1.0 mL of plasma was used in this method, which is still not a practical amount. Moreover, although the technique involving pretreatment by IP is excellent, it is time-consuming in terms of the development of the antibody. Siskos et al. reported that the GIP1−42 and GIP3−42 were quantified simultaneously using LC−MS/MS via their tryptic surrogate peptides.17 Larger intact peptides and proteins are more difficult to analyze by electrospray ionization (ESI) because of the occurrence of many multiple charge states and sometimes cause adsorption problems in experimental vessels. To avoid this problem, tryptic digestion, which is routinely used in proteomics research for peptide mapping and protein sequence work, due to its highly specific cleavage resulting in a limited number of tryptic peptides, has often been employed for the generation of shorter fragment peptides. In particular, the careful choice of the targeted peptides is essential for the success of the multiple reaction monitoring (MRM) experiments, and the selection criteria have been reported.18,19 Additionally, the sensitivity in the method developed by Sikos et al. was not enough for determination of the levels of GIP1−42 and GIP3−42 in human plasma. For this subject, caution is needed not only in the selection of LC parameters, including type of column and mobile phase, and MS parameters but also in the selection of an extraction method, assessing adsorption problems during the extraction process, etc.20,21 In the present study, an analytical procedure has been developed for the simultaneous quantification of active GIP1−42 and inactive GIP3−42 as follows: protein precipitation (PP) was performed during sample pretreatment before digestion and solid-phase extraction (SPE) was performed after digestion. In order to obtain desirable peptides including the N-terminus, trypsin and endoproteinase Asp-N were compared for use as the protease. As a result, the proteolytic peptide fragments, GIP1−8 and GIP3−8, were used as the corresponding surrogates result in from digestion by Asp-N and were sensitively analyzed by a nanoLC−MS/MS/MS. Finally, this developed method was successfully applied to a clinical trial for diabetic subjects
■
MATERIALS AND METHODS
Standards
The human GIP1−42 peptide standard (MWav 4983.5) was purchased from Peptide Institute, Inc. (Osaka, Japan). GIP3−42 (MWav 4749.3) was purchased from Bachem AG (Bubendorf, Switzerland). GIP1−8 (MWav 886.9), GIP1−16 (MWav 1811.0), GIP3−8 (MWav 652.7), and pyroglutamyl GIP3−8 (MWav 634.7) were purchased from BEX Co, Ltd. (Tokyo, Japan). The internal standards (ISs) 6-Phe-[13C9,15N]-GIP1−42 and 6-Phe[13C9,15N]-GIP3−42 (isotopic purity >99% 13C; > 99% 15N) were purchased from Thermo Fisher Scientific GmbH (Ulm, Germany). Chemicals
Acetonitrile (ACN, LC−MS grade), methanol (MeOH, LC− MS grade), ethanol (EtOH, special grade), formic acid (FA, LC−MS grade), acetic acid (AcOH, LC−MS grade), trifluoroacetic acid (TFA, special grade), ammonium carbonate (special grade), and 28% ammonia solution (NH4OH, special grade) were purchased from Wako Pure Chemical Industries, Ltd. (Osaka, Japan). Ammonium formate was purchased from Fluka (Buchs, Switzerland). Water used for the preparation of all solutions was purified through a Milli-Q apparatus (Millipore, Billerica, MA). Charcoal (charcoal activated A, powder) was purchased from Nacalai Tesque, Inc. (Kyoto, Japan). Materials
Blank human plasma (sterile human plasma EDTA, sodium) was purchased from Rockland Immunochemical, Inc. (Gilbertsville, PA). Diprotin A, an inhibitor of DPP-4, was purchased from Peptide Institute, Inc. (Osaka, Japan). Endoproteinase Asp-N and 2× Asp-N reaction buffer (100 mM Tris-HCl, 5 mM Zinc Sulfate, pH 8.0) were purchased from New England Biolabs Inc. (Ipswich, MA). Oasis MAX 96-well plates (30-μm particle size, 30 mg) and the analytical column for the recovery studies after PP (Acquity BEH C18 1.0 mm × 50 mm, 1.7 μm particle size) were purchased from Waters (Milford, MA). The trap column (Acclaim PepMap C18, 5 μm particle size, 300 μm i.d. × 1 mm) and analytical column (Acclaim PepMap C18, 3 μm particle size, 75 μm i.d. × 150 mm) were purchased from Dionex Softron GmbH (Germering, Germany). The ESI emitter (PicoTip EMITTER, FS360-20-10-N-20-C12) was purchased from New Objective (Woburn, MA). Ultrafree centrifugal filter (Durapore PVDF; 0.22 μm) and low-binding hydrophilic poly(tetrafluoroethylene) (0.20 μm) were purchased from Merck Millipore (Darmstadt, Germany). Instrumentation
Liquid Chromatography−Mass Spectrometry. For the evaluation of the preservation stability at 4 °C of the peptide GIP1−8 and the peptide GIP1−16, the automated nanoLC−MS system consisted of an Ultimate 3000 Series nanoLC system and an QSTAR Elite quadrupole/time-of-flight mass spectrometer (Q/TOF MS, MDS Sciex, Toronto, Canada). For the recovery determinations of GIP1−42 and GIP3−42 from human plasma after PP, an ultraperformance liquid chromatography (UPLC−MS) system was used, consisting of a UPLC column (Waters, Milford, MA) and an LTQ Orbitrap linear ion trap high-resolution mass spectrometer (Thermo Fisher Scientific GmbH, Bremen, Germany). For the quantification of GIP1−42 and GIP3−42 using the corresponding surrogate peptide, GIP1−8 2691
dx.doi.org/10.1021/pr400069f | J. Proteome Res. 2013, 12, 2690−2699
Journal of Proteome Research
Article
Figure 1. Flowchart for the extraction and quantification of GIP1−42 and GIP3−42 in human plasma.
10 min at 4 °C and 20,400 × g, the supernatant was evaporated to dryness under a stream of nitrogen at 40 °C. Digestion by Endoproteinase Asp-N. The concentrated extracts obtained by PP were dissolved in 100 μL of Asp-N reaction buffer (50 mM Tris-HCl, 2.5 mM zinc sulfate, pH 8.0). Then, 10 μL of a solution containing 0.08 μg/μL of Asp-N was added, and samples were incubated at 37 °C for 15 h. After incubation, analytes were extracted using Oasis MAX 96-well plates. Solid-Phase Extraction (SPE). The digested samples were then resuspended in 1 mL of 5% NH4OH and loaded onto an OASIS MAX 96-well plate that had been washed with 1 mL of MeOH and 1 mL of 75% ACN containing 0.2% TFA and equilibrated with 1 mL of water. The wells were subsequently washed with 500 μL of 5% NH4OH, ammonium formate (pH 8.0), ammonium formate (pH 5.0), 0.1% acetic acid (pH 3.0), and 75% ACN. GIPs that had been digested by Asp-N were eluted from the wells with 1 mL of 75% ACN containing 0.1% TFA, and the eluents were evaporated to dryness. The dried residues were reconstituted with 25 μL of 10% ACN containing 0.1% TFA and then filtered. Following filtration with a 0.20-μm filter, 5 μL of extracted sample was injected for analysis. LC−MS/MS/MS Conditions. LC−MS/MS(/MS) analyses were performed on an Ultimate 3000 series nanoLC coupled to an API QTRAP 5500 hybrid triple quadrupole/linear ion trap mass spectrometer (described above in detail). Instrument control, data acquisition, and processing were performed using the associated Analyst 1.5 software. The mobile phase for the trapping column consisted of 2% ACN containing 0.1% TFA. Mobile phase A consisted of 2% MeOH containing 0.1% FA. Mobile phase B consisted of 95% MeOH containing 0.1% FA. The analytical gradient profile with a flow rate of 250 nL/min was as follows (min/% B): 0/0, 2/0, 3/25, 21/60, 22/100, 27/ 100, 27.1/0. A subsequent period of 18 min at 0% B for system re-equilibration was added. The ambient temperature of the analytical column and that of the autosampler was set at 50 and 10 °C, respectively. MS analysis was carried out in positive ionization mode using an ion spray voltage of 2300 V. Since GIP is secreted after a meal, the concentration of time before a meal is low. It has been reported that the plasma concentrations of GIP1−42 and GIP3−42 before the meal tolerance test in patients with type 2 diabetes are about 14 pM and 16 pM, respectively.8 It has been also reported that LC−MS in MRM3 mode using hybrid triple quadrupole/linear ion trap mass spectrometer is a powerful tool for targeted peptide quantification, and the MRM3 mode is superior to the MRM mode in specificity and sensitivity.23 Therefore, in this
and GIP3−8, respectively, the automated nanoLC−MS system consisted of an Ultimate 3000 Series nanoLC system (Dionex Softron GmbH, Germering, Germany) and an API QTRAP 5500 hybrid triple quadrupole/linear ion trap mass spectrometer (QqQ/LIT-MS, MDS Sciex, Toronto, Canada) equipped with a nano-ESI interface NANO Spray III (AB Sciex, Toronto, Canada). For all experiments, the tray temperature of each autosampler was set at 10 °C. Others. Evaporation of samples was performed using TurboVap LV (Biotage, Uppsala, Sweden) or EZ-2 Plus (Genevac Inc., SP Industries, Gardiner, NY). Preservation Stability Studies of N-Terminal Peptides
GIP1−8 and GIP1−16 were dissolved in water/ACN/FA (20/80/ 0.1, v/v/v) to a final concentration of 1 μM each. The mixed solution was stored at 4 °C and injected into nanoLC-Q/TOF, and the peaks corresponding to GIP1−8 and GIP1−16 were analyzed at day 0 and day 7. Because the resolution of the quadrupole mass spectrometer was not enough to allow detailed evaluation of modifications such as oxidation as well as unexpected modifications of peptides, high-resolution and high-accuracy analyses of the modifications of GIP1−8 and GIP1−16 were conducted by QSTAR Elite. Sample Preparation for Quantification of Human GIP1−42/GIP3−42 (Figure 1)
Preparation of Plasma without GIPs (GIP-Free Plasma). Two grams of charcoal was added to every 10.0 mL of plasma, and this mixture was agitated for 12 h at 4 °C. After centrifuging for 1 h at 4 °C and 105,000 × g, the supernatant was filtered using a 0.22-μm filter.22 Protein Precipitation (PP). In order to prepare a standard curve, a charcoal-stripped plasma sample aliquot, 200 μL in volume, and 10 μL of a solution containing 0.06 μmol/μL of diprotin A were taken in a 2-mL tube. Then aliquots of GIP1−42/GIP3−42 standard solutions and IS solution were added, in order to prepare a standard curve ranging in concentration from 1 to 500 pM for GIP1−42 and from 10 to 500 pM for GIP3−42. The calibration curve for GIP1−42 consisted of 11 calibrators as follows: blank, 0, 1, 2, 5, 10, 20, 50, 100, 200, and 500 pM. The standard curve for GIP3−42 consisted of 8 calibrators as follows: blank, 0, 10, 20, 50, 100, 200, and 500 pM. Twenty microliters of IS solution was added to all samples resulting in a final plasma concentration of 100 pM for 6-Phe-[13C9,15N]-GIP1−42 and 1000 pM for 6-Phe[13C9,15N]-GIP3−42. Subsequently, 100 μL of 180 mM ammonium carbonate solution and 900 μL of cold EtOH were added and stored for 20 min on ice. After centrifuging for 2692
dx.doi.org/10.1021/pr400069f | J. Proteome Res. 2013, 12, 2690−2699
Journal of Proteome Research
Article
replicates of samples spiked with low, medium, and high levels of GIP1−42 and GIP3−42: 1, 10, 100, and 500 pM for GIP1−42 and 10, 100, and 500 pM for GIP3−42.
study, the selectivity and sensitivity of GIP1−8 extracted from human plasma were evaluated in MRM3 mode and in MRM mode. According to the results of this evaluation and considering the scan speed of mass spectrometry, the quantification of GIP1−8 was performed in the MRM3 mode, and the quantification of GIP3−8 was performed in the MRM mode. An MRM3 ion chromatogram is reconstructed from selected specific fragment ions produced from a primary product ion trapped in the Q3 linear ion trap and subsequently activated by resonant excitation. This primary product ion is selected among the most intense MRM transitions observed for a proteolytic peptide. The Q1 resolution was adjusted to 0.7 ± 0.1 amu fwhm for both MRM and MRM3 modes, referred to as unit resolution. Q3 was also set to unit resolution in the MRM mode. In the MRM3 mode, ion trap fill time performed with dynamic fill time, and excitation time was set to 25 ms. In all MRM3 experiments, the scan rate was set to 10,000 Da/s. The curtain gas flow was set at 10 psi using nitrogen, and ion source gas 1 flow was set at 5 psi using air. The selected transitions for the peptide GIP1−8 were m/z 887.4 → 782.4 → 764.2, those for 6-Phe-[13C9,15N]-GIP1−8 were m/z 897.4 → 792.4 → 774.2 in the MRM3 mode, those for GIP3−8 were m/z 653.2 → 288.1 + 653.2 → 435.3 + 653.2 → 548.3 (summed transitions of 3 product ions), and those for 6-Phe-[13C9,15N]-GIP3−8 were m/z 663.2 → 288.1 + 663.2 → 445.2 + 663.2 → 558.3. MRM transitions were optimized by adjustments of collision energy, declustering potential, collision cell exit potential, entrance potential, and dwell time. Additionally, the MRM3 transition was optimized by adjustment of excitation energy.
Plasma Levels of GIP1−42/GIP3−42 in Patients with Type 2 Diabetes
The present method was applied to determine the plasma levels of GIP1−42 and GIP3−42 obtained from patients with type 2 diabetes treated with the DPP-4 inhibitor sitagliptin or the α-GI miglitol in order to confirm the usefulness of the developed assay method and to clarify the effect on the GIP concentration levels in plasma of each antidiabetic drug. Fourteen patients underwent DPP-4 inhibitor monotherapy at a dose of 50 mg per day for 1 month. Eighteen patients underwent α-GI monotherapy at a dose of 50 mg per meal for 1 month. Blood samples (approximately 3 mL) were collected into blood collection tubes containing DPP-4 inhibitor and ethylenediaminetetraacetic acid dipotassium salt (BD P-700, Becton, Dickinson and Company, NJ), before and 1 h and 2 h after the meal tolerance test in the morning. Then the blood samples were immediately centrifuged, and the obtained plasma samples were stored at −80 °C. The blood samples were obtained in an exploratory trial conducted in a hospital setting (Naka Memorial Clinic, Ibaraki) in Japan. The study protocol was approved by the Ethics Committee of Naka Memorial Clinic. All statistical analyses were performed using EXSUS ver. 7.7.1 (CAC EXICARE Corporation, Osaka, Japan). The significance of differences between 2 groups was determined by unpaired t test. Values of p < 0.05 were considered significant.
■
RESULTS AND DISCUSSION Siskos et al. have reported that the N-terminal peptide of GIP1−16/GIP3−16 obtained by tryptic digestion is able to be a surrogate peptide for intact GIP1−42/GIP3−42 in a simultaneous quantification method for GIP1−42 and GIP3−42 in mouse plasma, and the surrogate peptide is more sensitive compared to the intact GIPs in LC/ESI−MS/MS analysis. This proposal suggests a method for the quantification of low-abundance peptides. Additionally, in our experience, surrogate peptides are not only sensitive but also useful by virtue of being difficult to adsorb. Trypsin is the most popular protease in proteome research. However, in this study, we have evaluated whether it is an appropriate protease for the generation of the surrogate peptide for the quantification of GIPs. We developed a quantification method for use as a practical and robust GIP assay in a clinical trial, which has the following limitations: (i) the sample volume for quantification and (ii) the use of commercially available blood collection tubes that are commonly used for incretin research in the hospital. Because of the limitations of sample volume and lower concentration of GIPs present, we further pursued a more sensitive method than that developed by Siskos et al. The equipment that we used differs from theirs. Furthermore, the two methods differ in the use of isotope-labeled full-length internal standards, MRM3 mode, mobile phase, and nanoLC.
Assay Validation for the Quantification of Human GIP1−42/GIP3−42
Recovery Studies of GIP1−42/GIP3−42 after Protein Precipitation (PP). The recoveries of GIP1−42 and GIP3−42 after PP were determined at a concentration of 100 nM in plasma (n = 3). To determine the recovery rate of human GIP1−42 and GIP3−42, 3 charcoal-stripped plasma samples were spiked with 100 nM of GIP1−42 and GIP3−42 either prior to or after the extraction. The corresponding ISs were spiked postextraction. The recovery rates were calculated by comparing the mean peak area ratio of the analyte to the IS of the samples. Since the supernatant after PP includes a large number of peptides and small molecules, it appears that these contents would disturb the detection of GIP1−42 and GIP3−42. Therefore, ESI-MS experiments for mass spectrometric quantification for the recovery were conducted on an Orbitrap mass spectrometer with high-resolution/high-accuracy mass measurement.24 The mass tolerance of the ion peaks was set to 10 ppm. Recovery Studies of GIP1−8/GIP3−8 after Solid-Phase Extraction (SPE). The recoveries of GIP1−8 and GIP3−8 after SPE were determined at a concentration of 10, 100, and 500 pM in plasma (n = 3). To determine the recovery rate of human GIP1−8 and GIP3−8, extracted residues of human plasma obtained after PP and digestion by Asp-N were spiked with each concentrations of GIP1−8 and GIP3−8 either prior to or after the extraction. The corresponding ISs were spiked after the extraction. The recovery rates were calculated by comparing the mean peak area ratio of the analyte to the IS of samples. Accuracy and Precision. For evaluation of the accuracy and precision of the procedure, three validation batches were prepared. Each batch included a double blank (no analyte and IS), a blank (no analyte), a set of calibration standards, and 3−5
Stability of Fragment Peptide
There are two ways to quantify peptides using LC−MS. One is to quantify intact target peptide and another is to quantify a surrogate peptide, which is a fragment peptide from the target peptide. In general, it is possible to quantify peptides composed of about 40 amino acid residue, i.e., GIP, using intact peptides. However, due to the adsorption and solubility of the peptide, the method of quantification using the shorter surrogate 2693
dx.doi.org/10.1021/pr400069f | J. Proteome Res. 2013, 12, 2690−2699
Journal of Proteome Research
Article
containing 0.1% FA, followed by storage at 4 °C for 7 days. Before and after the storage, the solution was injected into the nanoLC system coupled to Q/TOF mass spectrometry for the detailed evaluation of peptide modification (Figure 3). The peak corresponding to oxidized GIP1−16 was found together with GIP1−16 in the total ion current chromatogram, whereas GIP1−8 was stable in this solution. LC−MS/MS analysis of oxidized GIP1−16 was performed for the identification, confirming that the methionine at 14 had been oxidized. Regarding GIP3−8, one more modification should be considered. It is possible that the peptide containing glutamic acid at the N-terminal forms pyroglutamyl peptide.27 Therefore, we looked for pyroglutamyl GIP3−8 in the extracted plasma sample by nanoLC coupled to a hybrid triple quadrupole/linear ion trap mass spectrometer in MRM mode. However, the corresponding peak was not found in the sample (data not shown). These results strongly indicate that GIP1−8 and GIP3−8, which were obtained after proteolysis by endopeptidase Asp-N, are suitable as surrogate peptides corresponding to GIP1−42 and GIP3−42, respectively.
peptide would be superior to that of quantification using intact peptide in terms of the sensitivity of mass spectrometry. The surrogate peptide should be unique and also should be resistant to any modification. It has been reported that GIP1−42 and GIP3−42 are substrates of DPP-4; however, they are not a substrate of neutral endopeptidase 24.11.2,25 Siskos et al. reported that GIP1−42 and GIP3−42 were quantified simultaneously using LC−MS/MS via their tryptic surrogate peptides. Trypsin is routinely used in proteomics research for peptide mapping and protein sequence work, due to its highly specific cleavage resulting in a limited number of tryptic peptides. However, the tryptic surrogate peptides GIP1−16 and GIP3−16 contain methionine, which can be easily oxidized, at position 14.18,26 Therefore, the use of endoproteinase Asp-N to obtain surrogate peptides that do not contain methionine was highly recommended. Proteolysis of GIP1−42 and GIP3−42 by Asp-N yielded GIP1−8 and GIP3−8, respectively (Figure 2). Since both
Selection of Internal Standard (IS) Peptides
The addition of isotopically labeled ISs effectively overcomes the problems of fluctuations in signal intensity. Therefore, many quantifications of peptides or proteins using LC−MS are performed with an isotope-labeled peptide as the IS.26 In this approach an IS, an isotopically labeled peptide, is added to the samples after the completion of the digestion step. However, the digestive efficiency could not be corrected by this method when the digestive efficiency differs for individual samples, reducing the accuracy of the quantification method. Brun et al. have developed the protein standard absolute quantification
Figure 2. GIP1−42 and GIP3−42 amino acid sequences and sequences of the GIP1−8 and GIP3−8 surrogate fragments. Solid arrows show the cleavage site of Asp-N. Dotted arrows show the cleavage site of trypsin.
of the fragment peptides do not have methionine, they were expected to remain stable throughout the experiment. To evaluate the stability of the fragment peptides in solution, synthetic GIP1−8 and GIP1−16 were mixed in 80% ACN
Figure 3. Total ion current chromatograms obtained from the mixed standard solution of GIP1−8 and GIP1−16 before and after storage for 7 days at 4 °C. (A) Oxidized GIP1−16 was not observed at day 0, whereas (B) it was observed clearly at day 7. M(Ox): methionine sulfoxide. 2694
dx.doi.org/10.1021/pr400069f | J. Proteome Res. 2013, 12, 2690−2699
Journal of Proteome Research
Article
Figure 4. Comparison of specificity and sensitivity of detection between nanoflow LC−MRM and LC−MRM3 experiments for the fragment peptide of GIP, GIP1−8, spiked at 2 pM into charcoal-stripped human plasma. (A) MRM-extracted ion chromatogram of summed transition 887.4/782.4 + 887.4/669.3 + 887.4/522.2. (B) MRM3-extracted ion chromatogram of transition 887.4/782.4/764.2.
Selection of Mobile Phase for High Sensitivity Quantification
(PSAQ) method, which uses in vitro synthesized isotopelabeled full-length proteins as the ISs and MS for the absolute quantification of biomarkers.28 Their results demonstrated the advantage of the PSAQ approach over conventional methods that use labeled proteolytic peptide as the IS. In this study, 6Phe-[13C9,15N]-GIP1−42 corresponding to GIP1−42 and 6-Phe[13C9,15N]-GIP3−42 corresponding to GIP3−42 were added to samples as the ISs. Recovery after PP and digestive efficiency by Asp-N were significantly corrected for by these developed isotope-labeled full-length IS peptides, indicating that the method was highly effective as an accurate quantification method.
The intensity of GIP1−8 in some mobile phases was evaluated as follows: 20 nM GIP1−8 solutions were prepared in (1) MeOH/ H2O/FA (50/50/0.1, v/v/v), (2) ACN/H2O/FA (50/50/0.1, v/v/v), (3) MeOH/2 mM ammonium formate aqueous solution (50/50, v/v), and (4) ACN/2 mM ammonium formate aqueous solution (50/50, v/v) and were injected into a mass spectrometer at flow rate of 250 nL/min using an infusion pump. Data were acquired in full scan mode and in a range of m/z from 800 to 1000 to evaluate the intensity of a single charged ion detected at m/z 887.4 corresponding to GIP1−8 and noise peaks around the peak (Table S1, Supporting Information). These data demonstrated that a higher signal-tonoise ratio (S/N) for GIP1−8 was achieved using the MeOHbased solvent system, and FA and ammonium formate did not influence the S/N. Comparison between MeOH and ACN as the organic modifier has been reported for the analysis of peptide mixtures to maximize the limits of detection by nanoLC−ESI-MS/MS.29 Their data demonstrated that MeOH reproducibly yielded a higher S/N for each peptide. We also examined the sensitivity of the peak corresponding to GIP1−8 in the MRM3 mode by nanoLC coupled to a hybrid triple quadrupole/linear ion trap mass spectrometer and confirmed that MeOH yielded 2-fold higher sensitivity than did ACN in the analysis of GIP1−8 (data not shown). ON the basis of these data, we used MeOH as the organic modifier of the mobile phase for high-sensitivity quantification of GIP1−42 and GIP3−42 using nanoLC−MS/MS/MS.
Detection in MRM3 Mode
As described above, plasma concentrations of GIP1−42 and GIP3−42 in patients with type 2 diabetes before a meal are lower than 20 pM.8 It was also reported that the concentrations of GIP1−42 and GIP3−42 in healthy subjects are less than those concentrations in patients with type 2 diabetes.3 Based on this information, we set the target lower limit of quantification (LLOQ) for GIP1−42 and GIP3−42 to 1 pM and 10 pM, respectively, and the amount of the plasma for an extraction was set to 200 μL in consideration of measuring other vital signs in a clinical test. This setup required a sensitivity of less than 200 amol on the column for LC−MS analysis. For such highly sensitive quantification of GIP1−42, nanoLC coupled to a hybrid triple quadrupole/linear ion trap mass spectrometer was selected. To confirm the performance of the MRM3 mode of determination, which is expected to quantify some minor proteins or peptides in complex biological matrixes like plasma or serum, GIP1−8 extracted from human plasma by PP, followed by proteolysis by Asp-N and SPE, was detected in the MRM mode and in the MRM3 mode (Figure 4). Three transitions were summed (887.4 → 782.4 + 887.4 → 669.3 + 887.4 → 522.2) in the MRM mode to enhance detection levels. Transitions 887.4 → 782.4 → 764.2 were performed in the MRM3 mode. As revealed in Figure 4A, the MRM chromatogram of the targeted transitions for GIP1−8 spiked at 2 pM exhibits numerous interfering peaks. In contrast, the MRM3 chromatogram (Figure 4B) clearly displays a peak corresponding to GIP1−8. Although an extract from plasma by only PP and SPE includes a large number of peptides and small molecules, the MRM3 showed improvement in the limit of detection. The performance of the MRM3 mode in quantification and its practical use with plasma obtained from patients with type 2 diabetes are shown below.
Assay Validation
In order to use the established procedure for the evaluation of the profile of GIP1−42 and GIP3−42 in human plasma (Figure 1), a validation was conducted to cover its linearity, specificity, LLOQ, recovery, intraday validation, and interday validation. Two steps, PP and SPE, were evaluated for the recovery rate (Tables 1 and 2). The best recovery rates of GIP1−42 and GIP3−42 after PP were achieved under basic conditions (pH 9.0) and were 78.1% and 75.7%, respectively. Under acidic conditions with hydrochloric acid, the amount of precipitated protein was small and many noise peaks were observed in the extracted ion chromatogram for GIP1−42 and GIP3−42. The recovery rate of GIP1−8 and GIP3−8 after SPE was more than 59.2%. These recovery rates are good but not perfect; however, the isotope-labeled full-length IS peptides described above will surely compensate for the insufficiency of the recovery rates. 2695
dx.doi.org/10.1021/pr400069f | J. Proteome Res. 2013, 12, 2690−2699
Journal of Proteome Research
Article
matrices for the standard curve is an important aspect. As for a matrix, it is desirable not to include the target peptide and to have a composition similar to that of the sample, if possible. Adsorption of the target peptide to the surface of a vial or microcentrifuge tube has also been shown to affect the quantitative analysis of peptides at less than micromolar concentrations.21 In immunoassays, surfactants are often added to prevent such adsorption.30,31 For the analysis of peptides in plasma using LC−MS, however, surfactants may not be compatible with MS measurements. Lame et al. prepared amyloid beta standards in artificial cerebrospinal fluid (CSF) containing 5% rat plasma, which improved detection and recovery.14 In our study, we also observed problems with recovery when we used phosphate-buffered saline or 5% bovine serum albumin (BSA). However, the use of charcoal-stripped plasma solved this problem, and excellent calibration curves were achieved for GIP1−42 and GIP3−42 as a result. These results suggest that charcoal-stripped human plasma is excellent with respect to preventing adsorption loss and being devoid of endogenous target peptides that influence quantification and is widely available for other quantification methods for endogenous peptides or proteins.
Table 1. Recovery of GIP1‑42 and GIP3‑42 after Protein Precipitation (n = 3) recoverya (%)
pH GIP1−42
GIP3−42
2.0 7.4 9.0 2.0 7.4 9.0
nc 47.5 78.1 nc 45.1 75.7
± 1.3 ± 3.1 ± 3.0 ± 2.5
Data represent the mean ± SD. nc: peak shapes were not acceptable for the calculation of recovery because of interference by noise peaks.
a
Table 2. Recovery of GIP1‑8 and GIP3‑8 after Solid-Phase Extraction (n = 3)a recovery (%)
a
concn (pM)
GIP1−8
GIP3−8
10 100 500
65.1 ± 12.7 74.5 ± 7.1 59.2 ± 2.9
72.5 ± 22.2 71.6 ± 11.3 66.3 ± 18.1
Data represent the mean ± SD.
Clinical Sample Analysis
Further, this minimized losses during the pretreatment process, and high sensitivity and precise and accurate quantification were accomplished. GIP1−42 validation samples were analyzed at concentrations of 1, 10, 100, and 500 pM and showed precision (% CV) not exceeding 15% and accuracy (% RE) within 11%. Similarly, GIP3−42 validation samples were prepared and analyzed at 10, 100, and 500 pM and showed % CV
