Article pubs.acs.org/jpr
Standardized Protocols for Quality Control of MRM-based Plasma Proteomic Workflows Andrew J. Percy,† Andrew G. Chambers,† Derek S. Smith,† and Christoph H. Borchers*,†,‡ †
University of Victoria - Genome British Columbia Proteomics Centre, Vancouver Island Technology Park, #3101 − 4464 Markham Street, Victoria, BC V8Z 7X8, Canada ‡ Department of Biochemistry and Microbiology, University of Victoria, Petch Building Room 207, 3800 Finnerty Road, Victoria, BC V8P 5C2, Canada S Supporting Information *
ABSTRACT: Mass spectrometry (MS)-based proteomics is rapidly emerging as a viable technology for the identification and quantitation of biological samples, such as human plasmathe most complex yet commonly employed biofluid in clinical analyses. The transition from a qualitative to quantitative science is required if proteomics is going to successfully make the transition to a clinically useful technique. MS, however, has been criticized for a lack of reproducibility and interlaboratory transferability. Currently, the MS and plasma proteomics communities lack standardized protocols and reagents to ensure that high-quality quantitative data can be accurately and precisely reproduced by laboratories across the world using different MS technologies. Toward addressing this issue, we have developed standard protocols for multiple reaction monitoring (MRM)based assays with customized isotopically labeled internal standards for quality control of the sample preparation workflow and the MS platform in quantitative plasma proteomic analyses. The development of reference standards and their application to a single MS platform is discussed herein, along with the results from intralaboratory tests. The tests highlighted the importance of the reference standards in assessing the efficiency and reproducibility of the entire bottom-up proteomic workflow and revealed errors related to the sample preparation and performance quality and deficits of the MS and LC systems. Such evaluations are necessary if MRM-based quantitative plasma proteomics is to be used in verifying and validating putative disease biomarkers across different research laboratories and eventually in clinical laboratories. KEYWORDS: plasma proteomic workflows, multiple reaction monitoring, MRM, SIS peptides, quantitation, standard kits, quality control
■
INTRODUCTION Mass spectrometry (MS) is becoming increasingly important in biomedical and biological research. Its importance has been spurred by the launch of such large-scale research initiatives as the Human Proteome Project,1,2 which uses MS as one of the main technology platforms for identifying, quantifying, and characterizing the human proteome. In addition, efforts to verify and validate candidate disease biomarkers in human biofluids, such as blood plasma, toward clinical utility as diagnostic and prognostic indicators are also being performed through a targeted, multiplexed approach involving multiple reaction monitoring (MRM) in conjunction with stable isotope-labeled standards (SIS).3,4 The considerable reduction in development © XXXX American Chemical Society
time and cost, in addition to the ability to multiplex the protein assays and obtain accurate quantitative results with near absolute specificity, makes this quantitative approach particularly attractive compared to enzyme-linked immunosorbent assays (ELISAs), the current “gold standard” in clinical laboratories.5 The MS-based proteomics field, however, is plagued by questions with regard to the reproducibility and transferability of the developed MRM assays. These issues must be addressed if Special Issue: Chromosome-centric Human Proteome Project Received: September 22, 2012
A
dx.doi.org/10.1021/pr300893w | J. Proteome Res. XXXX, XXX, XXX−XXX
Journal of Proteome Research
Article
of quantitation, and protein concentration) to the “known” values for the proteins in the kits, which will enable system performance to be accredited. Doing so will highlight the presence of an instrument-specific or sample preparation-related issue that will need to be addressed prior to performing their quantitative research of interest. This was supported by an intralab reproducibility study performed at the Centre by researchers with varying backgrounds and expertise. The results of this investigation are also discussed herein.
MS-based assays are to eventually be utilized for patient screening in a clinical setting and if the proteomics community is to work together to verify and validate the true disease markers from a lengthy list of possible candidates. Blood plasma is the most complex human-derived proteome matrix,6 due to the presence of thousands of proteins, a dozen of which account for >90% of the total plasma protein mass.7 Moreover, these protein concentrations span a 10 order-ofmagnitude concentration range (albumin, 41 mg/mL; interleukin 1-beta, 1.2 pg/mL).8 Despite these problems, plasma remains a popular biofluid for screening as it is relatively inexpensive to sample, the collection is minimally invasive, and it contains a repository of circulating proteins that are secreted or leaked from surrounding cells, tissues, and organs. Thus, plasma proteins provide clues to the pathophysiology of the patient which is the impetus behind the use of plasma for disease biomarker discovery, verification, and validation. If we, as a proteomics community, are to use MRM methods to assess the candidate disease biomarkers in plasma samples and if these methods are to be translatable to clinical laboratories, these methods must be able to be reproduced when transferred to independent laboratories. In recent years, interlaboratory reproducibility has been investigated by various teams of researchers for standardizing methods9−11 and system performance12 within a MRM-based plasma proteomic approach involving SIS peptides. To reach this goal, Addona et al. have demonstrated high intra- and interlaboratory reproducibility (mean CVs 98% isotopic enrichment; Cambridge Isotope Laboratories; Andover, MA) on the C-terminal arginine or lysine residue. As described previously,20 after SIS peptide synthesis, purification was performed by reversed-phase high performance liquid chromatography (RP-HPLC) and then by matrix-assisted laser desorption/ionization-time-of-flight-MS (MALDI-TOF-MS) and capillary zone electrophoresis (CZE) analyses on fractions containing >80% of the target peptide. The absolute peptide concentrations were determined by amino acid analysis (AAA) analyses at the Hospital for Sick Children (Toronto, ON, Canada). The CZE and AAA values are used later to help correct for any possible contribution from incorrect or partial synthesis products. The average purity of the 43 target peptides was 95%. The SIS peptides are stored in aliquots as a lyophilizate or in solution (at 1 nmol/μL in 30% ACN and 0.1% FA) at −80 °C until use.
LC−MS Analytical Platform
The experiments described here were performed on Agilent’s standard-flow LC−MRM-MS platform (Agilent Technologies; Palo Alto, CA). This platform consists of a 1290 Infinity UHPLC system that is interfaced to a 6490 triple quadrupole mass spectrometer via a standard-flow ESI source. The LC−MS conditions employed here were similar to those described previously.15 Briefly, plasma tryptic digests containing added SIS peptides were separated by RP-UHPLC at 0.4 mL/min over a 30-min run. This was facilitated by a multistep gradient from 3 to 90% mobile phase B, which consisted of 0.1% FA in 90% ACN. The gradient used was as follows (time, %B): 0 min, 0.3%; 2 min, 11%; 15 min, 19%; 20 min, 29%; 22 min, 39%; 25 min, 45%; 27 min, 90%; 29, 90%; and 30, 3%. The amounts of digest and SIS mix injected on-column were 10 μg (the equivalent of 143 nL of neat plasma per analysis) and 100 fmol, respectively. Analytical and technical replicates were performed in triplicate for the optimization experiments (unless otherwise specified), and in quintuplicate for the analyses of the standard samples in the quantitation experiments. Peptide-specific retention times and collision energies (5−37 V, 15 V on average) were employed for the synthetic and natural (NAT) forms of each peptide. While 3 transitions for each of the 43 target peptides were monitored during method development, a single transition for each peptide was used for stability testing and protein quantitation in the final MRM assay. This resulted in a total of 43 transitions (SIS only) on which the stability study was performed, and 82 (SIS and NAT) interference-free transitions for protein quantitation (see the section below for details on the interference analysis). In all cases, a single peptide was monitored for each plasma protein. In the quantitation experiments, the transition dwell times ranged between 13 and 129 ms, with 18 transitions being the maximum number that could be monitored in a 260 ms cycle.
Solution Preparation
A stock solution of ammonium bicarbonate (25 mM, pH 8) was prepared for use as a buffer in the following solutions: 10× diluted raw human plasma, 10% sodium deoxycholate, 0.5 M TCEP, 100 mM iodoacetamide, and 100 mM DTT. Promega sequencing-grade modified trypsin (20 μg lyophilized powder/ vial) was reconstituted in 25 mM ammonium bicarbonate (50 μL) immediately prior to use. A solution of Worthington TPCKtreated trypsin at 0.9 mg/mL was prepared immediately prior to use, in phosphate or ammonium bicarbonate buffer (each at 25 mM), and in the absence or presence of calcium chloride (0−15 mM). Equimolar (100 fmol/μL) and concentration-balanced (250 fmol/μL) SIS peptide mixtures were prepared, then diluted to 25 fmol/μL (in 0.1% FA) for the optimization experiments. A series of concentrations in 0.1% FA (from 125 to 0.025 fmol/μL for standard samples F to A) were then prepared for quantitation. Aliquots of the SIS peptide mixes were stored at −80 °C until use, either for sample preparation or peptide stability testing. Various percent FA solutions were prepared for postdigestion acidification (1%), washing the solid phase extraction cartridges (0.1%), and eluting the peptides from the extraction cartridges (50% ACN in 0.1% FA). Sample Preparation
All samples (i.e., for optimization, reproducibility, and quantitation experiments) were prepared using a similar experimental procedure. Briefly, the diluted human plasma was buffered in 354 μL of 25 mM ammonium bicarbonate before being denatured and reduced for 30 min at 60 °C with sodium deoxycholate and TCEP for a final concentration of 1% and 5 mM, respectively. The reduced disulfide linkages were then alkylated with 100 mM iodoacetamide for 30 min at 37 °C in the dark for a final concentration of 10 mM. To prevent alkylation of other residues, excess iodoacetamide was inactivated with 100 mM dithiothreitol (final concentration: 10 mM) and incubated at 37 °C for an additional 30 min. Proteolytic digestion was then initiated by adding an aliquot of Promega trypsin at a 50:1 substrate/enzyme ratio or an aliquot of Worthington trypsin at a 20:1 substrate/enzyme ratio. The Worthington trypsin was added in two doses, with the second being added 4 h after the first, to compensate for its lower specific activity (271 vs 17294 units/mg protein for the Promega trypsin). Digestion was allowed to proceed for a total of 9 h at 37 °C before being terminated with a chilled, acidified SIS peptide mixture (final concentration: 0.5% FA for a pH < 3). The acidic pH precipitated the acid-insoluble surfactant so that it could be subsequently removed by centrifugation at 12000× g for 10 min. After
Interference Screening
From a previously curated list of optimized MRM transitions,15,17 the 3 most intense precursor/product ion pairs for each target peptide were selected, and these were screened for chemical interference. These evaluations were performed under matrix-free (i.e., mobile phase A: 0.1% FA) and with-matrix (i.e., digested blood plasma) conditions, with duplicate LC−MRMMS analyses being run for each condition, as described previously.20,23,25 Briefly, the response ratios of the SIS peptide in buffer, the SIS peptide in plasma, and the NAT peptide in plasma were calculated, relative to the most intense transition in the set, for the 3 transitions for each peptide. From these relative ratios, the average relative ratios and variabilities could be determined for each transition. If the transitions had an average coefficient of variation (CV) below 25% and the NAT peptide displayed identical chromatographic behavior to its synthetic analogue, then a rank was assigned to the transitions. The transition that produced the highest average relative ratio for each of the interference-free peptides was selected as the C
dx.doi.org/10.1021/pr300893w | J. Proteome Res. XXXX, XXX, XXX−XXX
Journal of Proteome Research
Article
and reagents that can be obtained to ensure that high quality data can be reproduced by laboratories across the world using different MS technologies. To address this issue, we have developed two reference kits to help verify both the analytical performance of the mass spectrometer, as well as the samplepreparation steps used for MRM-based plasma proteomic analyses. Briefly, kit development involved trypsin grade optimization and selection, concentration balancing of the SIS peptide mixture, SIS peptide stability testing, and interference screening. Using the optimal digestion conditions and the stable, interference-free peptides, protein quantitation was performed for the generation of the reference values and their validation through interlaboratory testing. The development and application of the kits are described and discussed below in the order in which it was performed.
representative transition to be used in the quantitative MRM assay. Data Analysis
All data was processed with Agilent’s MassHunter Quantitative Analysis software (version B.04.00), as described previously.15,17 Briefly, the data was preprocessed to verify peak selection and integration, before the samples were analyzed for peak area (i.e., response) measurements and ultimately, for the generation of calibration curves. The relative responses of either NAT/SIS (for the optimization experiments) or SIS/NAT (for the quantitation experiments) were used in the analyses. To determine the plasma protein concentrations, calibration curves were constructed with 7 concentration levels (with 5 replicates/level), spanning a 10000-fold concentration range (0.1−1000 fmol of concentration-balanced SIS mix on-column). Each calibration curve was fitted to a 1/x2 weighted linear regression model. The data points within a given concentration level had to be both precise (with an average CV of
