Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 7-Day Trial for You or Your Team.

Learn More →

Quantitative Mass Spectrometric Multiple Reaction Monitoring Assays for Major Plasma Proteins *

Quantitative Mass Spectrometric Multiple Reaction Monitoring Assays for Major Plasma Proteins * Research Quantitative Mass Spectrometric Multiple Reaction Monitoring Assays for Major Plasma Proteins* Leigh Anderson‡§ and Christie L. Hunter¶ creased sensitivity (into the pg/ml range) and precision (CVs Quantitative LC-MS/MS assays were designed for tryptic peptides representing 53 high and medium abundance  5–10%) at the cost of restricting discovery potential toward proteins in human plasma using a multiplexed multiple novel proteins (3). In practice, a combination of these ap- reaction monitoring (MRM) approach. Of these, 47 pro- proaches (one or more survey approaches for de novo bi- duced acceptable quantitative data, demonstrating with- omarker discovery coupled with a candidate-based approach in-run coefficients of variation (CVs) (n  10) of 2–22% to biomarker validation in large sample sets) appears to pro- (78% of assays had CV <10%). A number of peptides gave vide an effective staged pipeline (4) for generation of valid CVs in the range 2–7% in five experiments (10 replicate plasma biomarkers of disease, risk, and therapeutic response. runs each) continuously measuring 137 MRMs, demon- Candidate-based specific assays rely on the specificity of strating the precision achievable in complex digests. De- capture or detection methods to select a specific molecule as pletion of six high abundance proteins by immunosub- analyte. Capture reagents such as antibodies can provide traction significantly improved CVs compared with whole extreme specificity (particularly when two different antibodies plasma, but analytes could be detected in both sample types. Replicate digest and depletion/digest runs yielded are used as in a sandwich immunoassay) and form the basis correlation coefficients (R ) of 0.995 and 0.989, respec- of most existing clinical protein assays. There is intense in- tively. Absolute analyte specificity for each peptide was terest in miniaturizing sets of such assays (5, 6) in array demonstrated using MRM-triggered MS/MS scans. Reli- formats (on planar substrates, beads, etc.), although signifi- able detection of L-selectin (measured at 0.67 g/ml) in- cant problems remain in the production of suitable antibodies dicates that proteins down to the g/ml level can be and in the simultaneous optimization of multiple assays in one quantitated in plasma with minimal sample preparation, fluid volume. yielding a dynamic range of 4.5 orders of magnitude in a Mass spectrometry provides an alternative assay approach, single experiment. Peptide MRM measurements in plasma relying on the discriminating power of mass analyzers to digests thus provide a rapid and specific assay platform for select a specific analyte and on ion current measurements for biomarker validation, one that can be extended to lower quantitation. In the field of analytical chemistry, many small abundance proteins by enrichment of specific target pep- tides (stable isotope standards and capture by anti-peptide molecule analytes (e.g. drug metabolites (7), hormones (8), antibodies (SISCAPA)). Molecular & Cellular Proteomics protein degradation products (9), and pesticides (10)) are 5:573–588, 2006. routinely measured using this approach at high throughput with great precision (CV  5%). Most such assays use elec- trospray ionization followed by two stages of mass selection: Accurate quantitation of proteins and peptides in plasma a first stage (MS1) selecting the mass of the intact analyte and serum is a challenging problem because of the complex- (parent ion) and, after fragmentation of the parent by collision ity and extreme dynamic range that characterize these sam- with gas atoms, a second stage (MS2) selecting a specific ples (1). The widely adopted separative (survey) approach to fragment of the parent, collectively generating a selected proteomics, in which an attempt is made to detect all com- reaction monitoring (plural MRM) assay. The two mass filters ponents, has proven to be limited in sensitivity toward low produce a very specific and sensitive response for the se- abundance proteins (2) and typically provides limited quanti- lected analyte that can be used to detect and integrate a peak tative precision. The alternative candidate-based approach, in a simple one-dimensional chromatographic separation of which relies on specific assays optimized for quantitative the sample. In principle, this MS-based approach can provide detection of selected proteins, can provide significantly in- The abbreviations used are: CV, coefficient of variation; QqQ, From the ‡The Plasma Proteome Institute, Washington, D. C. triple quadrupole; MRM, multiple reaction monitoring; S/N, signal-to- 20009-3450 and ¶Applied Biosystems, Foster City, California 94404 noise; Nat, natural sample-derived peptide; SIS, stable isotope-la- Received, October 7, 2005, and in revised form, November 28, beled internal standard; polySIS, polyprotein SIS; SISCAPA, stable 2005 isotope standards and capture by anti-peptide antibodies; MIDAS, Published, MCP Papers in Press, December 6, 2005, DOI 10.1074/ multiple reaction monitoring-initiated detection and sequencing; mcp.M500331-MCP200 MARS, multiple affinity removal system. © 2006 by The American Society for Biochemistry and Molecular Biology, Inc. Molecular & Cellular Proteomics 5.4 573 This paper is available on line at http://www.mcponline.org This is an Open Access article under the CC BY license. MS Assays for Plasma Proteins EXPERIMENTAL PROCEDURES absolute structural specificity for the analyte, and in combi- nation with appropriate stable isotope-labeled internal stand- Peptide Selection and MRM Design—We developed a set of MRMs iteratively using three basic approaches: pure in silico design from ards (SISs), it can provide absolute quantitation of analyte sequence databases, design from available LC-MS/MS proteomic concentration. These measurements have been multiplexed survey data, and comprehensive MRM testing of all of the candidate to provide 30 or more specific assays in one run (11). Such peptides of a protein. methods are slowly gaining acceptance in the clinical labora- In our initial attempt to generate MRMs by purely in silico methods, tory for the routine measurement of endogenous metabolites a set of 177 proteins and protein forms was assembled that are (e.g. in screening newborns for a panel of inborn errors of demonstrated or have potential to be plasma markers of some aspect of cardiovascular disease (20), and a subset of 62 proteins was metabolism (12)) and some drugs (e.g. immunosuppressants selected for which an estimate of normal plasma abundance was (13)). available. Predicted tryptic peptides for each of these were generated Recently the MRM assay approach has been applied to the along with relevant Swiss-Prot annotations and a series of computed measurement of specific peptides in complex mixtures such physicochemical parameters: e.g. amino acid composition, peptide as tryptic digests of plasma (3). In this case, a specific tryptic mass, Hopp-Woods hydrophilicity (21), and predicted retention time in reversed-phase (C ) chromatography (22). An index of the likeli- peptide can be selected as a stoichiometric representative of 18 hood of experimental detection was derived from a data set reported the protein from which it is cleaved and quantitated against a by Adkins et al. (23) by counting the number of separate “hits” for the spiked internal standard (a synthetic stable isotope-labeled peptide in the data set divided by the number of hits for the most peptide (14)) to yield a measure of protein concentration. In frequently detected peptide from the same protein. An overall index of principle, such an assay requires only knowledge of the peptide quality was generated according to a formula that gave positive weights to Pro, KP, RP, and DP content and negative weights masses of the selected peptide and its fragment ions and an to Cys, Trp, Met, chymotrypsin sites, certain Swiss-Prot features ability to make the stable isotope-labeled version. C-reactive (carbohydrate attachment, modified residues, sequence conflicts, or protein (3), apolipoprotein A-I (15), human growth hormone genetic variants), and mass less than 800 or greater than 2000. The (16), and prostate-specific antigen (17) have been measured 3619 tryptic peptides predicted for the 62 protein marker candidates in plasma or serum using this approach. Because the sensi- (6–497 peptides per target) ranged in length from 1 to 285 amino acids. Within the useful range of 8–24 amino acids, 721 peptides had tivity of these assays is limited by mass spectrometer dy- a C-terminal Lys and 690 had a C-terminal Arg. In this report, pep- namic range and by the capacity and resolution of the assist- tides from 30 of these target proteins ending in C-terminal Lys were ing chromatography separation(s), hybrid methods have also selected for further study. been developed coupling MRM assays with enrichment of We also selected peptides based on two types of direct proteomic proteins by immunodepletion and size exclusion chromatog- survey experiments. In the first case we carried out classical LC- MS/MS analysis of plasma digests in which the major ions observed raphy (18) or enrichment of peptides by antibody capture were subjected to MS/MS using the ion trap capabilities of the 4000 (SISCAPA (19)). In essence the latter approach uses the mass Q TRAP instrument. The identified peptides showing the best signal spectrometer as a “second antibody” that has absolute struc- intensity and chromatographic peak shape for a given parent protein tural specificity. SISCAPA has been shown to extend the were selected. In addition, we used the Global Proteome Machine sensitivity of a peptide assay by at least 2 orders of magnitude database of Craig et al. (24) to select peptides from target proteins that were frequently detected (multiple experiments). (19) and with further development appears capable of extend- Finally we used an adaptation of the MIDAS workflow, described ing the MRM method to cover the full known dynamic range of previously for discovering post-translational modifications (25, 26), to plasma (i.e. to the pg/ml level). look for measurable tryptic peptides from a variety of plasma proteins. There is compelling evidence that high and medium abun- In this approach, the protein sequence is digested in silico, likely y-ion dance plasma proteins have value as clinical biomarkers and fragments are predicted, and theoretical MRMs are generated for all the thus that there may be applications for specific MRM assays peptides in an acceptable size window. These MRMs are then used as a survey scan in a data-dependent experiment to detect specific pep- even without antibody enrichment. One may therefore ask how tide peaks, and each resulting MRM peak is examined by full scan many plasma proteins can be measured by quantitating their MS/MS to obtain sequence verification of the hypothesized peptide. To peptides in a plasma digest, and how precise could these verify peptide specificity in designated protein targets, selected pep- measurements be? If the measurement strategy proves to be tides were searched with BLASTP for exact matches against the ge- nome-derived human, mouse, and rat Ensembl peptides using Ensembl robust, could it be carried out using existing high throughput Blastview (www.ensembl.org/Homo_sapiens/blastview). LC-MS/MS platforms? To address these questions we gener- “Random” MRMs—Two approaches were used to generate pseu- ated and tested MRM assays based on peptides from a variety dorandom MRMs. In the first case we used 100 MS1 values distrib- of high to medium abundance plasma proteins to see how many uted randomly (by the Excel RAND function) between 408.5 and could be measured effectively by LC-MS/MS with and without 1290.2 (the maximum and minimum of an early set of real MRMs we tested) paired with MS2 values chosen randomly between this MS1 subtraction of the most abundant proteins. An understanding of and the maximum of the real MRMs (1495.6), thus mimicking the the performance of MS/MS in this application could enable properties of our real MRMs (which are generally 2 charge state routine and relatively inexpensive measurement of classical peptides and 1 charge fragments). In a second set we paired 131 plasma proteins and also provide a foundation for use of MS1 values chosen randomly from among MS1 values in a large table MS/MS in more sensitive anti-peptide antibody-enhanced SIS- of real MRMs with MS2 values chosen randomly from the real MS2 CAPA assays for low abundance proteins. values of the same list, imposing only the constraint that each MS2 574 Molecular & Cellular Proteomics 5.4 MS Assays for Plasma Proteins had to be between 1 and 2 times the paired MS1 mass (to approxi- RESULTS mate our selection criteria for real MRMs). Peaks detected in these Design of MRM Assays for Abundant Plasma Proteins—In MRMs were examined by triggering MS/MS (the MIDAS workflow). an initial approach to the selection of representative peptides Reagents—The following chemicals were used: trypsin (Promega), sodium dodecyl sulfate (Bio-Rad), iodoacetamide (Sigma), formic for MRM assays, a single peptide of 8–18 amino acids was acid (Sigma), tris-(2-carboxyethyl)phosphine (Sigma), and acetonitrile chosen from each of 30 proteins spanning a broad range of (Burdick and Jackson). plasma concentrations (6.6  10 down to 1 fmol/ml normal Plasma Depletion and Digestion—All experiments were performed concentration) based on computed characteristics alone (19). on aliquots of a single human plasma sample from a normal volunteer. MRMs were designed assuming doubly charged peptide ions The six highest abundance proteins were depleted from plasma using the multiple affinity removal system (“MARS”; Agilent Technologies and using fragments selected as likely y-ions above the m/z of spin columns) according to the manufacturer’s protocol. Depleted the 2 parent ion with collision energies assigned by a ge- sample was then exchanged into 50 mM ammonium bicarbonate neric formula (CE  0.05  m/z  5), and the peptides were using a VivaSpin concentrator (5000 molecular weight cutoff, Viva- expressed as a concatamer polySIS protein containing single Science). Undepleted plasma was also desalted before digestion. 13 15 Both depleted or undepleted plasma samples were denatured and copies of each peptide labeled with [U- C ,U- N ]lysine. 6 2 reduced by incubating proteins in 0.05% SDS and 5 mM tris-(2- When a tryptic digest of the polySIS was analyzed, all 30 carboxyethyl)phosphine at 60 °C for 15 min. The sample was then peptides were detected by MRMs. When a digest of whole adjusted to 10 mM with iodoacetamide and incubated for 15 min at human plasma was added to the polySIS peptides, 19 of the 25 °C in the dark. Trypsin was added in one aliquot (protease:protein labeled polySIS peptides were still detected by the same ratio of 1:20) and incubated for5hat37 °C. Labeled Peptide Internal Standards: polySIS—A series of SIS pep- MRMs, but only 11 of the plasma digest-derived unlabeled tides was added to samples in selected experiments by spiking with cognate peptides were detected (by the same MRMs ad- the tryptic digest of a “polySIS” polyprotein. Briefly this protein was justed for isotope label masses). produced by cell-free transcription and translation of a synthetic gene Because different peptides from a single protein can vary coding for 30 concatenated tryptic peptide sequences (derived from 13 15 widely in detectability by ESI-MS, we attempted to improve 30 plasma proteins) in the presence of U- C ,U- N -labeled lysine 6 2 (a total mass increment of 8 amu compared with the natural peptide). upon the in silico approach to MRM design using experimen- The 30 peptides were selected based on our initial in silico MRM tal data from a conventional peptide survey scan of a human design approaches and are thus not fully optimized using experimen- plasma digest and applying the selection criteria to peptides tal data. Of these peptides, 13 were used in the present studies (the with demonstrated detectability. Using a 3-h LC gradient, remainder were not reproducibly detected in digested plasma with peak area 1E04). The positioning of the label atoms at the extreme MS/MS scans were collected for the major doubly or triply C terminus of each peptide has the effect that all fragments that charged ions across the separation using information-de- contain the C terminus (i.e. the y-ions) will show the mass shift due to pendent data acquisition, and a second run was performed the label, whereas all the fragments that contain the N terminus (and using time-filtered exclusions of the peptide ions detected in hence have lost one of more C-terminal residues: the b-series ions) the first run. The combined results identified 54 plasma pro- will have the same masses as the corresponding fragments from the natural (sample-derived) target protein. These features (shifted y-ions teins ranging in abundance from albumin down to fibronectin and normal b-ions) provide a simplification in interpreting the frag- (normal plasma concentration of 300 g/ml). This experi- mentation patterns of the SIS peptides in comparison with the similar mental MS/MS data provided explicit information for peptide QCAT concept described recently by Beynon et al. (27). To determine selection, charge state, and most abundant y-ion m/z value the absolute concentration of polySIS protein, an aliquot was diluted with 1 M urea, 0.05% SDS, and 50 mM Tris, pH 8 and subjected to under the specific conditions used (i.e. electrospray ionization N-terminal Edman sequencing, yielding an initial concentration of 5 with collisional peptide fragmentation), allowing improved de- 1 pmol/l. A tryptic digest of the polySIS protein was spiked into sign of MRMs. When these MRMs were then used to analyze whole and depleted human plasma digests at the final concentrations the same sample in a subsequent run, triggering MS/MS shown in Table I. scans at any MRM signal, most of the peptides were detected LC-MS/MS Analysis—Plasma digests with and without added polySIS peptides were analyzed by electrospray LC-MS/MS using LC as peaks in the chromatogram and identified by database Packings (a division of Dionex, Synnyvale, CA) or Eksigent nanoflow search as expected. In most of these MRM chromatograms, LC systems (Table I) with 75-m-diameter C PepMap reversed- only a single peak was detected. phase columns (LC Packings) and eluted with gradients of 3–30% Because peptide detection sensitivity using MRM is ex- acetonitrile with 0.1% formic acid. A column oven (Keystone Scien- pected to be greater than that achieved in a full scan MS tific, Inc.) was used to maintain the column temperature at 35 °C. Electrospray MS data were collected using the NanoSpray source survey approach, a comprehensive de novo MRM design on a 4000 Q TRAP hybrid triple quadrupole/linear ion trap instrument method was explored for those proteins not detected in the (Applied Biosystems/MDS Sciex), and the peaks were integrated above survey experiment. Using a novel software tool, a large using quantitation procedures in the Analyst software 1.4.1 (Intelli- set of MRMs was generated for each of a series of target Quan algorithm). MRM transitions were acquired at unit resolution in proteins by selecting all predicted tryptic peptides in a useful both the Q1 and Q3 quadrupoles to maximize specificity. size range together with multiple high mass y-ion fragments of each (the “MIDAS” workflow (25, 26)). These MRMs were then L. Anderson and C. L. Hunter, manuscript in preparation . tested in LC-MS/MS runs of the unfractionated plasma digest, Molecular & Cellular Proteomics 5.4 575 MS Assays for Plasma Proteins FIG.1. MIDAS workflow (MRM-triggered MS/MS) verification of the identity of peptide DLQFVEVTDVK representing fibronectin (normal concentration, 300 g/ml) in a digest of depleted human plasma. A shows the ion current profile of MRM 647.3/789.4 (MS1/MS2), and B shows the MS/MS spectrum of peptide fragments taken at the time of the peak (* marks the y7 fragment monitored in this MRM). cps, counts per second. grouped in panels that included all the predicted tryptic pep- proach (indicated by an X in the SIS column of Table II): eight tides of one or two proteins at a time (50–100 MRMs per run), of the initial 30 in silico peptides were eliminated as likely to be with MS/MS scans triggered on any peaks observed. Of 12 too low abundance for detection, and better alternative pep- proteins examined, nine produced at least one usable MRM tides were selected from experimental data for four others. (signal-to-noise (S/N) ratio 20). For all but one of the peptides we elected to measure two MRM results from the above approaches were pooled, and fragments (i.e. using two MRMs per peptide), yielding 119 a set of optimized MRMs was assembled that covered a total MRMs. Finally we included MRMs for 18 stable isotope-la- of 60 peptides representing 53 proteins in human plasma beled internal standard (“SIS”) versions of target peptides (i.e. (Table II; seven proteins were represented by two peptides). the tryptic digest of the polySIS protein) spiked into the digest This set includes 18 peptides selected by the in silico ap- plasma samples. The resulting set of 137 MRMs was meas- 576 Molecular & Cellular Proteomics 5.4 MS Assays for Plasma Proteins TABLE I Summary design of data sets (experiments A–F) Replicate runs were performed in series with 30-min washes between runs. Load is expressed as the equivalent volume of plasma from which the sample was derived. Load factors express total or non-depleted (MARS flow-through) loads relative to experiment A. Load factor Replicate Equivalent plasma PolySIS Experiment Sample LC system runs volume loaded spike Total protein Non-depleted proteins l fmol A 10 Depleted plasma digest LC Packings 0.01 1 1 1.3 B 10 Whole plasma digest LC Packings 0.01 10 1 1.3 C 10 Whole plasma digest Eksigent 0.001 6 0.1 2.0 D 10 Depleted plasma digest Eksigent 0.01 0.6 1 2.0 E 10 Depleted plasma digest Eksigent 0.033 3.3 3.3 6.0 F1_1 4 Depletion 1, digest 1 Eksigent 0.01 1 1 F1_2 4 Depletion 1, digest 2 Eksigent 0.01 1 1 F2_1 4 Depletion 2, digest 1 Eksigent 0.01 1 1 F2_2 4 Depletion 2, digest 2 Eksigent 0.01 1 1 ured in all the replicate runs described below using an 18-ms dwell time per MRM and a resulting cycle time of 3s between measurements. After the final MRM method was constructed, each MRM transition and respective retention time were validated again as indicative of each specific peptide. Full scan MS/MS was acquired upon the appearance of the MRM signal, and each resultant spectrum was manually inspected to determine matching to the specific peptide (Fig. 1). Application to Digests of Whole and Depleted Plasma—Six experiments (A–F) were performed in each of which the same set of 137 MRMs was measured during sequential replicate LC-MS/MS runs of a single sample (same injection volume), and the appropriate peaks were integrated by the Analyst software to yield a value (peak area) for each MRM in each run. Experiments A–E (10 replicate runs each) are summarized in Table I. These experiments included tryptic digests of both whole (unfractionated) human plasma (experiments B and C) and plasma depleted of abundant proteins by MARS immu- FIG.2. Histogram of CVs of MRM values (peak areas) for five noaffinity subtraction (experiments A, D, and E), high (exper- experiments (experiments A–E) across all 137 MRMs (A)or iments B and E) and low (experiments A, C, and D) total across the 47 “best” MRMs (B). peptide loadings, and different chromatographic setups. Our objective was to assess the performance of the MRMs in various typical plasma digest experiments. The reference Fig. 2A shows histograms of the coefficients of variation for peptide load (experiment A) was derived from tryptic digestion the five experiments (individual values for each MRM are of the protein contained in 10 nl of plasma (700 ng of total contained in Table II). In analyses of depleted plasma, more protein assuming an average plasma concentration of 70 than 60% of the MRMs show within-run CVs less than 10%, mg/ml (28)) after depletion of the most abundant proteins using and almost half have CVs below 5%. A number of these an Agilent MARS spin column (84% of protein mass should be MRMs (e.g.  -antichymotrypsin, apolipoprotein E, he- removed based on a calculation using normal abundances). mopexin, heparin cofactor II, plasminogen, prothrombin, fi- This loading, comprising an estimated 110 ng of total peptides, brinogen  chain, complement C4, and factor B) showed an proved to be a loading compatible with very good nanoflow average within-run CV of 3–4% across three experiments, chromatography of the MRM peptides. Experiments B and E precision equivalent to that of good clinical immunoassays. used higher loadings to explore the trade-off between peak Analyses of whole (undepleted) plasma digests showed gen- stability (chromatographic quality, adversely affected by in- erally higher CVs (20–50% of MRMs had CV 10%). It is creased load) and S/N ratio (improved by increased analyte important to note that these reproducibility measures are quantity). We concluded that the 110 ng loading was optimal. computed on raw peak areas without correction using internal Chromatographic elution times were quite reproducible, show- standards. Four of the measured proteins were expected to ing average CVs of 2% (experiment D) and 2.5% (experiment E). be removed by the depletion process used (Agilent MARS Molecular & Cellular Proteomics 5.4 577 MS Assays for Plasma Proteins TABLE II A set of MRMs designed and tested for the detection of 53 proteins in human plasma or serum Protein name, tryptic peptide sequence, retention time in experiment D, peptide mass (MS1, m/z), singly charged fragment MS2 (m/z), mean peak area values, and CVs for 10 replicate analyses across five experiments are shown. 578 Molecular & Cellular Proteomics 5.4 MS Assays for Plasma Proteins TABLE II—continued Molecular & Cellular Proteomics 5.4 579 MS Assays for Plasma Proteins TABLE II—continued spin column). In comparing average peak areas obtained in compared with the lower of the two fragment CVs for each analyses of digests of whole and depleted samples, we found MRM, the average reduction in CV in experiments A–E ranges substantial reductions in albumin (1.3E08 reduced to from 0.7% to 0.1%. These small improvements in CV come 1E04), transferrin (1.5E05 reduced to 5E03), and at the cost of doubling the measurement time (or halving the haptoglobin (4.6E06 reduced to 1E05).  -Antitrypsin number of peptides monitored) and thus are unlikely to be was not detected reliably in any of the runs (presumably a bad useful in routine operation for improving reproducibility. peptide choice for MRM assay), and so its removal could not The relationship between CV and peak area, shown in Fig. be confirmed. We did not incorporate MRMs to measure the 3 for experiments A–E, indicates that peak areas below two immunoglobulins subtracted by the MARS column. 1E04 are unlikely to yield acceptable CVs (i.e. below 10%). Multiple measurements of an MRM would be expected to A cutoff of 1E04 corresponds to a signal-to-noise ratio of improve CVs, and we thus examined whether the sum of the 10, which is consistent with the quantitative requirement of two fragment MRMs measured separately for 59 of the pep- a S/N ratio of 10 for a reported lower limit of quantitation. The tides exhibited a smaller CV across 10 replicate runs than the highest peak areas measured (albumin peptides in whole individual MRMs. As shown in Table III, the average CV for the plasma digest samples) are above 1E08, demonstrating a summed MRMs across 59 peptides is 1–3% lower than the maximal working dynamic range of 4 orders of magnitude averages of either individual MRM. If the summed CV is above this cutoff. 580 Molecular & Cellular Proteomics 5.4 MS Assays for Plasma Proteins Calibration with Internal Standard Peptides—A set of 13 SIS slightly higher than CVs of the component measurements (80 peptides having the same sequences as 13 of the MRM- of 130 comparisons). measured tryptic peptides were used to assess the reproduc- Sensitivity—Two proteins with relatively low normal con- ibility of the quantitative measurements (five of the 18 SISs centrations in plasma were clearly detected among the MRMs measured by MRM yielded unusable quantitative information tested: L-selectin and fibronectin. The soluble form of L- because the peak area signal of either the standard or its selectin is a 33-kDa protein present in plasma at a normal cognate sample digest peptide fell below 1E04). The SIS concentration of 0.67 g/ml (29) or 20.3 pmol/ml. Fibronec- 13 15 tin is a 260-kDa protein present in plasma at a normal con- peptides were labeled with [U- C ,U- N ]lysine during syn- 6 2 centration of 300 g/ml (30) or 1200 pmol/ml. Given that an thesis from a synthetic DNA template in an in vitro transcrip- amount of digest corresponding to 0.01 l of plasma was tion/translation system and were cleaved from the polySIS loaded on column in experiment D, these peptides would be protein by trypsin. Samples spiked with SIS peptides were expected to be present on column at 200 and 12,000 amol, analyzed in experiments D and E (Table I) at relative peptide loadings of 1 and 3.3, respectively (SIS peptides at relative respectively. In the case of L-selectin we had spiked a SIS loadings of 1 and 3). Measured peak areas of MRM pep- peptide at 2.0 fmol and thus could determine that the natural tides showed an average 2.7 difference between these ex- (sample-derived) peptide was present at 0.1 times the amount of SIS (single point quantitation), yielding a measured 200 periments. Ratios between the natural sample-derived (Nat) amol and implied plasma concentration of 0.6–0.67 g/ml in and labeled (SIS) versions of these peptides were extremely good agreement with expectation. CVs for fibronectin in ex- similar in the two experiments (Fig. 4) and maintained a linear periments D and E were 4 and 4%, respectively, and for relationship (R  0.998) over almost 3 orders of magnitude in L-selectin were 22 and 11%, respectively, presumably reflect- concentration. Because the Nat:SIS ratios combine the vari- ing the fact that L-selectin was near the lower limit (1E04) ance of two measurements, the CVs of the ratios are typically for high quality detection. Six of the 53 selected target proteins were not reliably TABLE III observed, however. Three are probably explained by low nor- Improvement of CV obtained by combining peak areas of two mal abundances: we did not obtain a reproducible signal for fragment MRMs versus keeping them separate the selected peptides from coagulation factor V, vitamin K- Avg, average; frag(s), fragment(s). dependent protein C, or C4b-binding protein. There were also Avg CV Experiment instances in which peptides from more abundant proteins Sum of frags Frag 1 Frag 2 were not reliably detected: the inter- trypsin inhibitor light chain (despite the fact that a peptide from the heavy chain of A 10.5 11.8 14.8 this protein gave a good quality MRM), apolipoprotein C-II, B 16.2 20.0 19.4 C 11.0 13.0 14.4 and  -antitrypsin. In these latter cases, the problem is most D 8.0 9.4 12.3 likely poor choice of peptides: numerous alternative peptides E 8.5 9.4 11.9 exist for both the inter- trypsin inhibitor light chain and FIG.3. Plot of peak area versus CV for all MRMs in experiments A–E. Molecular & Cellular Proteomics 5.4 581 MS Assays for Plasma Proteins FIG.4. Comparison of computed amounts (based on ratios between Nat and SIS peptides) for 13 peptides in experiments D and E. Data are in- cluded for those ratios where both nu- merator and denominator peak areas were 1E04 in experiment D (13 peptides). -antitrypsin, but for small proteins, such as apolipoprotein The Density of Signals in MRM Space: Behavior of Random C-II, there may be no better alternative, and additional enrich- MRMs—Most of the MRMs we designed appeared to detect ment of these peptides will be necessary. only a single peak during the LC run of a complex digest such In general, depletion of the most abundant proteins using as depleted plasma (e.g. Fig. 1): whereas 73% had a robust the Agilent MARS column improved the performance of peak (defined as area 32,000) corresponding to the target MRMs for non-subtracted proteins substantially. This effect peptide analyte, only about 8% had a second peak meeting appeared to be due to improved detection sensitivity and the same peak area criterion (histograms in Fig. 7). We there- improved chromatographic peak shape (achieved by de- fore attempted to confirm that the density of peptide peaks in creasing the total peptide loaded by 6-fold), both of which “MRM space” was indeed low (equivalent to high MS/MS contribute to improved MRM peak signal-to-noise and lower detector specificity relative to sample complexity) by exam- CVs. Fig. 5 illustrates the benefit of depletion in removing the ining two types of randomized MRMs in the same depleted albumin peptide (major peak in Fig. 5A) and thus boosting the plasma digest sample. In a first set, 100 MRMs were gener- minor peaks in the depleted sample (Fig. 5B). At very high ated with “parent” masses randomly distributed over the loading of undepleted plasma digest, we noticed large shifts mass range of real peptides used in the 137 designed MRMs in peak retention times, but at loadings in the region of our and “fragment” masses randomly distributed between the nominal load the effect of high abundance peptides on MRM parent mass and the maximum fragment mass among the retention times was minor. designed MRMs (“random MRMs”). Of the 100 random Reproducibility of Immunodepletion and Tryptic Digestion— MRMs, only six showed a peak with area 32,000, and none In a sixth experiment (F series), we performed MARS deple- of these MRM peaks produced MS/MS spectra that led to a tion on two aliquots of the same plasma sample and then protein identification when searched with Mascot against separately digested two aliquots of each depleted sample Swiss-Prot. A second set of 131 random MRMs was gener- (total of four samples; e.g. F1_2 refers to the second digest of ated by randomly pairing parent and fragment masses from the first depletion). Four replicate runs of the 137 MRMs were the set of designed MRMs detectable in plasma, excluding carried out for each sample in randomized order to avoid any those cases where the fragment mass was lower than the sequence effect. Fig. 6 compares the mean peak areas of two parent (“random combination MRMs”). By using real peptide digests of a single depleted sample (Fig. 6A, F1_1 versus and fragment masses, these MRMs avoided any potential F1_2) and two depletions (Fig. 6B, F1_1 versus F2_1). The bias arising from the tendency of real peptide masses to peak area values span 4 orders of magnitude of which 2.5 cluster around integral masses (the mass defect). In this sec- orders is usable (i.e. beginning at peak area 1E04, the ond set, about 12% of the MRMs exhibited a peak with peak approximate cutoff below which CVs become large). Dupli- area 32,000, and once again none of these peaks gave cate digests show excellent comparability (R  0.995 and MS/MS spectra yielding a protein identification. All the peaks 0.998 for F1_1 versus F1_2 and F2_1 versus F2_2, respec- observed in the random MRM sets occurred late in the LC tively). Duplicate depletions (which necessarily include the gradient (after 100 min) after the elution of a large majority of effects of different digests as well) are only slightly worse (e.g. the designed plasma protein MRMs. These results suggest R  0.989 and 0.991 for F2_1 versus F1_1 or F2_2 versus that the density of quantifiable features in MRM space at our F1_2, respectively). current sensitivity, even for a very complex peptide sample 582 Molecular & Cellular Proteomics 5.4 MS Assays for Plasma Proteins FIG.5. Total ion current profiles across the chromatographic peptide separation for digests of undepleted (A, whole) and depleted (B) plasma from experiments C and D, respec- tively. Pie charts represent the protein composition of the samples: whole plasma contains 50% albumin (stip- pled region), whereas the proteins re- maining after MARS depletion include fibrinogen,  -macroglobulin, comple- ment C3, and all other lower abundance proteins (four remaining pie slices or- dered clockwise from 12 o’clock), show- ing the proportion of protein removed by depletion (white segments in B). The ma- jor peptide in A at 20.3 min is derived from albumin (whose abundance is shown by the black pie segment with white dots). cps, counts per second. and using unit resolution in both mass analyzers, is only eight were dropped before testing because of expected in- 6–12% of which a minority may be canonical tryptic peptides. sufficient abundance. Thus 13 in silico selections survived, The distribution of peak areas observed for random MRMs whereas 10 were replaced in testing. closely matches the distribution for second (non-target pep- Of the six failures (see above), we expect three proteins to tide) peaks in our designed MRMs, indicating that these ad- be within the concentration range that should be measurable ditional peaks represent a random background. and are selecting substitute peptides for these. The distribu- Selection of Best MRMs for Plasma Proteins—From the 119 tion of CVs for the 47 best MRMs is shown in Fig. 2B:in tested MRMs for sample peptides, we selected the best frag- experiment D, 40 of these had CVs below 10%, and 19 had ment ions (from 59 cases of two tested fragment ions) and the CVs below 5%. best peptide (from seven cases where two peptides were DISCUSSION tested per protein). A usable MRM resulted for 47 of the 53 initial proteins (indicated by an X in the Best MRM column in We designed and characterized mass spectrometric MRM Table II). Each of the 47 peptide sequences was verified as assays for tryptic peptides representing 53 high-to-medium unique in the human proteome (represented by the Ensembl abundance proteins of human plasma. The resulting panel of peptides) and occurred only once in the target protein. Three 47 successful MRMs, having within-run CVs ranging from 2 to of the peptides (representing antithrombin III, apolipoprotein 22% in depleted plasma, should be useful in the exploration E, and vitamin K-dependent protein C) occur in the mouse of protein-disease relationships and expression control in homologs as well, and seven (apolipoprotein E, vitamin K-de- clinical serum and plasma specimens. The fact that MRM pendent protein C, complement C4  and , fibronectin, assays, once developed, can be implemented on triple qua- haptoglobin , and inter- trypsin inhibitor heavy chain) occur drupole (QqQ)-MS instruments, widely used for precise in rat homologs (all the other human sequences did not occur quantitation of small molecules at high throughput, offers in the other species’ Ensembl peptides). the prospect of data collection across very large clinical Of the final 47 MRMs, 12 were contributed by the in silico sample sets. Our approach exemplifies the development of approach leading to the 30 polySIS peptides, and one (he- targeted specific assays needed to enable the statistical val- mopexin) was contributed by an earlier in silico effort (19): idation of proposed biomarkers in plasma (“candidate-based these are indicated in Table II by an X in the SIS column for the proteomics” (4)). associated Lys-labeled internal standard. A total of eight in Our initial attempts to select usable tryptic peptides by silico selections were replaced by better performing peptides purely in silico means were reasonably successful as approx- from the same target protein as a result of experimental imately half of the peptides chosen (13 of 23) produced ac- testing (four before and four after selection of the 137 MRMs), ceptable MS signals in plasma. Although the prediction of two subsequently failed and have not yet been replaced, and ionization properties of tryptic peptides can be expected to Molecular & Cellular Proteomics 5.4 583 MS Assays for Plasma Proteins general process for MRM design using these sources of in- formation is shown in Fig. 8 in which bold arrows indicate the preferred approach. The final step in the process, selection among alternative candidate MRMs for a given protein based on measured CVs in replicate runs, is recommended because of the significant differences in CV observed among MRMs with similar signal strength: CV appears to be a characteristic of a peptide to some extent separate from ion current. Despite the complexity of plasma digests (particularly those of depleted plasma where a small number of superabundant peptides have been removed), most MRMs exhibited only a single peak across the peptide LC chromatogram. This ob- servation is consistent with the low density of peaks in two sets of randomly distributed MRMs measured in depleted plasma digests and demonstrates the exquisite specificity of the two-stage QqQ-MS selection process used as the detec- tor. Nevertheless the existence of secondary peaks (whether or not they are actually tryptic peptides) in a small subset (10%) of MRMs indicates that chromatographic elution time is an important additional factor in providing the absolute analyte specificity desired in these assays. Following more extensive experience with specific MRMs and any variation that may occur in them due to shifts in peak elution times, it may be possible to establish a set of plasma MRMs that are truly free of secondary peaks and that could therefore poten- tially be measured using short LC separations (constrained only by increases in ion suppression effects as peptides crowd together). The unambiguous measurement of a peptide derived from FIG.6. Comparison of mean peak areas values for all MRMs in L-selectin (normal concentration of 0.67) suggests that pro- two replicate digests (A) and two depletion runs (and subsequent digests) (B). Error bars show 1 standard deviation computed from teins present in normal plasma at 1 g/ml are measurable replicate runs. by MRM in depleted and undepleted samples with minimal upfront sample fractionation. Different tryptic peptides from the same protein can produce ion currents differing by factors improve substantially in the future, we used experimental of at least 1E03 in LC-MS/MS experiments, excluding the MS/MS data, in combination with computational methods, to peptides that are not detected at all. This variation is due to select more successful target peptides. Two experimental multiple factors, including propensity to ionize in the electro- methods proved particularly useful. High abundance peptides spray source, coincidence in elution time with other easily were detected in conventional LC-MS/MS data-dependent ionizing peptides, efficiency of release during tryptic diges- full scan MS experiments in which a subset of high signal tion, and presence of unrecognized post-translational modi- peptides seen in MS1 are subjected to MS/MS. Lower abun- fications arising due to biology or during sample preparation. dance peptides were detected by constructing lists of candi- Although we do not yet know the frequency of high efficiency date MRMs to all appropriately sized predicted tryptic pep- peptides among the tryptic products of each protein, we tides from a target protein and then characterizing any detected MRM peaks by MS/MS (the MIDAS workflow in believe that numerous additional plasma proteins can be which MRM methods are designed using a specifically de- measured down to 1 g/ml using MRMs and estimate that signed script within the Analyst software). Because MRMs are 50–100 proteins may be added to our list as we extend this typically more sensitive than full scan survey MS for detection effort to additional candidates from the plasma component of very low abundance components, the MIDAS approach database (2, 20). allowed us to find successful MRMs for more lower abun- Given that albumin, the most abundant protein in plasma, is dance peptides and represents the preferred strategy for measurable by MRM in undepleted plasma in the same ex- MRM design going forward. This process was facilitated by periment that detects L-selectin, the dynamic range of quan- the combination of high sensitivity triple quadrupole MRM and titatively measurable proteins appears to be 1 g/ml to 55 ion trap MS/MS scan capabilities on the hybrid triple quadru- mg/ml or 5E04. This dynamic range of 4–5 orders of pole linear ion trap 4000 Q TRAP mass spectrometer. A magnitude is consistent with typical quantitation experiments 584 Molecular & Cellular Proteomics 5.4 MS Assays for Plasma Proteins FIG.7. Histograms of peak areas obtained with four sets of MRMs in a digest of depleted plasma. it is easily possible to select lower signal peptides from high abundance proteins (and high signal peptides from low abun- dance proteins) to diminish the dynamic range requirement at the peptide level. As long as stable isotope-labeled internal standards are used, the quantitative information contained in Nat:SIS ratios should be unaffected by peptide choice. The reproducibility of peptide MRM measurements was impressive in many cases. For the three experiments analyz- ing depleted plasma digests, an average of 78% of the 47 best MRMs had within-run CVs of 10% or less, and 36% had CVs less than 5%. A number of MRMs gave low CVs across five experiments (i.e. in depleted and whole plasma digests at widely varying total loads) including peptides from hemopexin (CVs ranging between 2 and 6%), vitronectin (4–7%), kinino- gen (5–6%), complement factor B (3–6%), complement C4 (3–5%), apolipoprotein B100 (5–7%), antithrombin III (5–7%), and  -antichymotrypsin (2–7%). These precision values are comparable to many small molecule MRM measurements and approach the results of conventional immunoassays used in clinical diagnostics (32). They demonstrate clearly that pep- tide MRM measurements can be used in high precision quan- FIG.8. Schematic diagram of a process for selecting MRMs. titative assays. Based on reproducibility at this level, it ap- Thicker arrows represent the preferred process. Square-cornered pears likely that useful comparative data across samples boxes represent informatics components, and rounded-corner boxes might be obtained by comparing simple peak area measure- are experimental steps. ments (i.e. without SIS standardization) and that relatively small quantitative differences could be detected (differences performed using MRM on this instrument and covers almost of 20% would be 3 standard deviations from the mean for half the logwise dynamic range of currently known and meas- many of the best MRMs and thus very easily detected). The ured proteins in human plasma (1). The dynamic range of peak observation that individual peptides showed consistently area measurements appears to be similar, extending from higher or lower CVs suggests that quantitative precision de- 1.3E08  5% for undepleted albumin to less than 1E04 pends on characteristics of the peptide and perhaps its envi- (the approximate cutoff below which CVs become 10%). In ronment of co-eluting peptides. Therefore it will be useful in this context it is important to note that measurement of pro- future work to select MRM peptides based on CV across teins over this dynamic range does not require measuring peptides at such a broad range of signal intensities because replicate runs in addition to the characteristics normally con- Molecular & Cellular Proteomics 5.4 585 MS Assays for Plasma Proteins sidered (e.g. ion current and chromatographic peak shape). set. It thus appears that both depletion and digestion can be MRMs designed for absolute quantitation using a set of 13 carried out with sufficient reproducibility to provide useful SIS peptides (with sequences identical to 13 of the selected measurements and that MRM assays provide an ideal method tryptic peptides) performed well. When spiked at known load- for assessing sample preparation reproducibility. ing these provided internal standards for quantitation and The multiplexing capability of LC-QqQ-MS platforms for resulted in a measured value for L-selectin in almost exact measuring peptides in complex digests is substantial, provid- agreement with the literature value for normal subjects (0.67 ing an opportunity to measure large panels of proteins accu- g/ml). This result is encouraging but not definitive as we did rately in each run. Based on the performance of the present not measure target protein concentrations in our sample by set of 137 MRMs, which were all monitored continuously independent methods, and the individual SIS peptides (de- across the entire LC gradient as 18-ms sequential measure- rived by digestion of a quantitated precursor protein) were not ments, it is clear that 100–200 MRMs might be used routinely individually quantitated. Thus not all of our endogenous: to measure peptides in long LC gradients. However, given standard ratio measurements agreed so well with literature reproducible chromatographic elution times, it is possible with values. In addition, single point concentration curves as used existing systems to measure each MRM only during a short here do not provide the same accuracy in quantitation as time window when the peak is expected to occur (e.g. a multiple point curves. Nevertheless the extreme reproducibil- window of 10% of total run length given an average 2–2.5% ity (R  0.991) of the measured ratios (and implied absolute CV in peak elution time measured in our experiments D and peptide concentrations) between experiments D and E (Fig. 4) E). Once this approach is implemented, based on extensive across almost 3 orders of magnitude in peak area and 3-fold knowledge of elution time and column reproducibility, and difference in total peptide load are striking. To effectively provided that MRMs are selected that do not cluster too much standardize peptides (and proteins) over wider ranges, a more in elution time, substantially more MRMs could be used in a sophisticated strategy will be utilized using smaller equimolar single LC MRM experiment. groups of peptides selected to represent each decade of An additional important consideration for throughput of peptide concentration (i.e. one polySIS product or a set of MRM measurements is the duration of the chromatography equimolar SIS peptides for each decade of the abundance run. In our replicate experiments D and E, a 30-min gradient scale). In addition, the completeness of polySIS digestion, was used that led to a total cycle time (including intersample and thus the relative stoichiometry of the resulting SIS pep- wash) of 75 min. It would clearly be advantageous if this could tides, must be rigorously characterized. be reduced, allowing more samples to be run per day. The Depletion of the highest abundance plasma proteins using extreme analyte specificity indicated by the low density of an immobilized antibody column proved to be very useful in peaks in MRM space suggests that most of our MRMs should measuring our MRMs. We expected to be able to load a perform well with less benefit from chromatographic separa- digest of 6 times as much depleted plasma as undepleted tion, and the ability to focus the MRM measurements in dis- plasma because these two samples would contain approxi- crete time windows may allow more MRMs to be brought mately equal amounts of total peptides, and this was con- closer together in elution time without sacrificing the required firmed. We also confirmed that the reference loading chosen multiple measurements across each peak. Hence we expect (110 ng of peptides derived from digestion of 10 nl of that substantial improvements in run time should be possible, depleted plasma) was near but not above the limit for good probably in conjunction with a shift to higher flow rate (e.g. quality LC peak shape and reproducible chromatography in a capillary flow) systems providing increased robustness in rou- nanoflow system. However, we had few if any MRMs that tine operation. Higher flow rate LC systems have significantly were detectable in the depleted sample but not the unde- lower (10) absolute sensitivity than nanoflow, but in the pleted sample. The major difference emerged instead in the current application it would be easily possible to load 10 lower CVs of peaks measured in the depleted sample due more sample (i.e. 100 nl) as sample at these loadings is rarely presumably to the decreased level of competing peptides and limiting. consequent higher signal to noise. In future extensions of this Considered in a broad context, MRM assays appear to method to lower abundance proteins, it is likely that depletion have several advantages in addition to those described will expose numerous targets not otherwise detectable. De- above. 1) The instrumentation used to measure peptide pletion significantly reduced the levels of albumin, transferrin, MRMs is very similar to that used in existing robust platforms and haptoglobin as expected. for high throughput quantitative measurement of drug metab- Highly reproducible depletion and tryptic digestion are a olites in plasma, for the detection of inborn errors of metab- necessity as steps in routine sample preparation for MRM olism in newborns and for pesticide analysis. A large base of analysis. In a small study (experiment F) designed to look at compatible instrumentation and expertise thus already exists. these effects, R values 0.995 were obtained for peak areas 2) Because the target proteins are detected via tryptic peptide in duplicate digests, and R values 0.989 were obtained for surrogates after sample denaturation and digestion, peptide duplicate depletion and digestion in the same experimental MRM results should be insensitive to alterations in protein 586 Molecular & Cellular Proteomics 5.4 MS Assays for Plasma Proteins cation of C-reactive protein in the serum of patients with rheumatoid folding and intersubunit associations (factors that can nega- arthritis using multiple reaction monitoring mass spectrometry and C- tively affect immunoassays). 3) MRM results are obtained by labeled peptide standards. Proteomics 4, 1175–1186 integrating discrete, anticipated peaks in simple ion chro- 4. Anderson, N. L. (2005) The roles of multiple proteomics platforms in a pipeline for new diagnostics. Mol. Cell. Proteomics 4, 1441–1444 matograms using well established and widely used commer- 5. Joos, T. O., Stoll, D., and Templin, M. F. (2002) Miniaturised multiplexed cial software, and thus the method is not dependent on the immunoassays. Curr. Opin. Chem. Biol. 6, 76–80 much more complex data processing infrastructure required 6. Haab, B. B. (2005) Antibody arrays in cancer research. Mol. Cell. Proteom- to handle the peak matching and spectral identification chal- ics 4, 377–383 7. Kostiainen, R., Kotiaho, T., Kuuranne, T., and Auriola, S. (2003) Liquid lenges of a discovery proteomic pipeline. 4) It is probable, chromatography/atmospheric pressure ionization-mass spectrometry in although not yet demonstrated, that the cost per high quality drug metabolism studies. J. Mass. Spectrom. 38, 357–372 measurement will be lower than for immunoassays (both to 8. Tai, S. S., Bunk, D. M., White, E. T., and Welch, M. J. (2004) Development and evaluation of a reference measurement procedure for the determi- create the assay to begin with and to apply it to large sample nation of total 3,3 ,5-triiodothyronine in human serum using isotope- sets). dilution liquid chromatography-tandem mass spectrometry. Anal. Chem. Important limitations of peptide MRMs must also be recog- 76, 5092–5096 9. Ahmed, N., and Thornalley, P. J. (2003) Quantitative screening of protein nized. 1) Although the method is applicable to post-transla- biomarkers of early glycation, advanced glycation, oxidation and nitro- tionally modified peptides, the post-translational modification sation in cellular and extracellular proteins by tandem mass spectrometry must be known or specifically hypothesized, and its proper- multiple reaction monitoring. Biochem. Soc. Trans. 31, 1417–1422 10. Sannino, A., Bolzoni, L., and Bandini, M. (2004) Application of liquid chro- ties must be designed into the assay up front (25) (and nor- matography with electrospray tandem mass spectrometry to the deter- mally cannot be discovered de novo by our approach). 2) mination of a new generation of pesticides in processed fruits and Some proteins may not readily produce usable peptides. The vegetables. J. Chromatogr. A 1036, 161–169 11. Barr, D. B., Barr, J. R., Maggio, V. L., Whitehead, R. D., Jr., Sadowski, M. A., protein may be too short and hence produce few candidates Whyatt, R. M., and Needham, L. L. (2002) A multi-analyte method for the to begin with or may, like many immunoglobulin sequences, quantification of contemporary pesticides in human serum and plasma be too variable. These cases will require further exploration using high-resolution mass spectrometry. J. Chromatogr. B. Anal. Tech- nol. Biomed. Life Sci. 778, 99–111 (alternative proteolytic enzymes or sample derivatization pro- 12. Roschinger, W., Olgemoller, B., Fingerhut, R., Liebl, B., and Roscher, A. A. cedures). 3) Genetic variants altering a single amino acid in (2003) Advances in analytical mass spectrometry to improve screening the selected peptide will prevent its determination by the for inherited metabolic diseases. Eur. J. Pediatr. 162, Suppl. 1, S67–S76 13. Streit, F., Armstrong, V. W., and Oellerich, M. (2002) Rapid liquid chroma- wild-type MRM; hence these must be recognized in advance tography-tandem mass spectrometry routine method for simultaneous and designed specifically. Provided these issues are taken determination of sirolimus, everolimus, tacrolimus, and cyclosporin a in into account, none appears to prevent productive use of whole blood. Clin. Chem. 48, 955–958 peptide MRMs for measuring abundances of plasma proteins. 14. Gerber, S. A., Rush, J., Stemman, O., Kirschner, M. W., and Gygi, S. P. (2003) Absolute quantification of proteins and phosphoproteins from cell Based on the results presented here, numerous clinically lysates by tandem MS. Proc. Natl. Acad. Sci. U. S. A. 100, 6940–6945 important plasma proteins can be measured by peptide 15. Barr, J. R., Maggio, V. L., Patterson, D. G., Jr., Cooper, G. R., Henderson, MRMs with precision comparable to current clinical immuno- L. O., Turner, W. E., Smith, S. J., Hannon, W. H., Needham, L. L., and Sampson, E. J. (1996) Isotope dilution-mass spectrometric quantification assays. The panel of assays presented should have general of specific proteins: model application with apolipoprotein A-I. Clin. use as the nucleus of candidate-based biomarker validation Chem. 42, 1676–1682 approach and should lead ultimately toward a comprehensive 16. Wu, S. L., Amato, H., Biringer, R., Choudhary, G., Shieh, P., and Hancock, W. S. (2002) Targeted proteomics of low-level proteins in human plasma assay platform for all human proteins in plasma. by LC/MSn: using human growth hormone as a model system. J. Pro- teome Res. 1, 459–465 Acknowledgment—We thank Dr. Tina Settineri (Applied Biosys- 17. Barnidge, D. R., Goodmanson, M. K., Klee, G. G., and Muddiman, D. C. tems) for advice and support. (2004) Absolute quantification of the model biomarker prostate-specific * The costs of publication of this article were defrayed in part by the antigen in serum by LC-MS/MS using protein cleavage and isotope payment of page charges. This article must therefore be hereby dilution mass spectrometry. J. Proteome Res. 3, 644–652 marked “advertisement” in accordance with 18 U.S.C. Section 1734 18. Liao, H., Wu, J., Kuhn, E., Chin, W., Chang, B., Jones, M. D., O’Neil, S., solely to indicate this fact. Clauser, K. R., Karl, J., Hasler, F., Roubenoff, R., Zolg, W., and Guild, § To whom correspondence should be addressed: The Plasma B. C. (2004) Use of mass spectrometry to identify protein biomarkers of Proteome Inst., P. O. Box 53450, Washington, D. C. 20009-3450. disease severity in the synovial fluid and serum of patients with rheuma- toid arthritis. Arthritis Rheum. 50, 3792–3803 Tel.: 301-728-1451; Fax: 202-234-9175; E-mail: leighanderson@ 19. Anderson, N. L., Anderson, N. G., Haines, L. R., Hardie, D. B., Olafson, plasmaproteome.org. R. W., and Pearson, T. W. (2004) Mass spectrometric quantitation of peptides and proteins using stable isotope standards and capture by REFERENCES anti-peptide antibodies (SISCAPA). J. Proteome Res. 3, 235–244 20. Anderson, N. L. (2005) Candidate-based proteomics in the search for 1. Anderson, N. L., and Anderson, N. G. (2002) The human plasma proteome: biomarkers of cardiovascular disease. J. Physiol. 563, 23–60 history, character, and diagnostic prospects. Mol. Cell. Proteomics 1, 21. Hopp, T. P., and Woods, K. R. (1981) Prediction of protein antigenic 845–867 determinants from amino acid sequences. Proc. Natl. Acad. Sci. U. S. A. 2. Anderson, N. L., Polanski, M., Pieper, R., Gatlin, T., Tirumalai, R. S., 78, 3824–3828 Conrads, T. P., Veenstra, T. D., Adkins, J. N., Pounds, J. G., Fagan, R., 22. Krokhin, O. V., Craig, R., Spicer, V., Ens, W., Standing, K. G., Beavis, R. C., and Lobley, A. (2004) The human plasma proteome: a non-redundant list and Wilkins, J. A. (2004) An improved model for prediction of retention developed by combination of four separate sources. Mol. Cell. Proteom- times of tryptic peptides in ion pair reversed-phase HPLC: its application ics 3, 311–326 3. Kuhn, E., Wu, J., Karl, J., Liao, H., Zolg, W., and Guild, B. (2004) Quantifi- to protein peptide mapping by off-line HPLC-MALDI MS. Mol. Cell. Molecular & Cellular Proteomics 5.4 587 MS Assays for Plasma Proteins Proteomics 3, 908–919 Guillet, P. (1986) Differences between young and elderly subjects in 23. Adkins, J. N., Varnum, S. M., Auberry, K. J., Moore, R. J., Angell, N. H., seasonal and circadian variations of total plasma proteins and blood Smith, R. D., Springer, D. L., and Pounds, J. G. (2002) Toward a human volume as reflected by hemoglobin, hematocrit, and erythrocyte counts. blood serum proteome: analysis by multidimensional separation coupled Clin. Chem. 32, 801–804 with mass spectrometry. Mol. Cell. Proteomics 1, 947–955 29. Atalar, E., Ozmen, F., Haznedaroglu, I., Acil, T., Ozer, N., Ovunc, K., 24. Craig, R., Cortens, J. P., and Beavis, R. C. (2004) Open source system for Aksoyek, S., and Kes, S. (2002) Effects of short-term atorvastatin treat- analyzing, validating, and storing protein identification data. J. Proteome ment on global fibrinolytic capacity, and sL-selectin and sFas levels in Res. 3, 1234–1242 hyperlipidemic patients with coronary artery disease. Int. J. Cardiol. 84, 25. Unwin, R. D., Griffiths, J. R., Leverentz, M. K., Grallert, A., Hagan, I. M., and 227–231 Whetton, A. D. (2005) Multiple reaction monitoring to identify sites of 30. van Helden, W. C., Kok-Verspuy, A., Harff, G. A., and van Kamp, G. J. protein phosphorylation with high sensitivity. Mol. Cell. Proteomics 4, (1985) Rate-nephelometric determination of fibronectin in plasma. Clin. 1134–1144 Chem. 31, 1182–1184 26. Cox, D. M., Zhong, F., Du, M., Duchoslav, E., Sakuma, T., and McDermott, 31. Deleted in proof J. C. (2005) Multiple reaction monitoring as a method for identifying 32. Lestin, M., Hergert, M., Lestin, H. G., Brinker, C., Storm, H., Kuhrt, E., Kuhrt, protein posttranslational modifications. J. Biomol. Tech. 16, 83–90 B., Lambrecht, H. G., and Kuhnel, W. (2002) Evaluation of the chemilu- 27. Beynon, R. J., Doherty, M. K., Pratt, J. M., and Gaskell, S. J. (2005) minescence immunoassays for the measurement of troponin I, myoglo- Multiplexed absolute quantification in proteomics using artificial QCAT bin and CK-MB using the IMMULITE system in comparison to other proteins of concatenated signature peptides. Nat. Methods 2, 587–589 measuring systems. Clin. Lab. 48, 211–221 28. Touitou, Y., Touitou, C., Bogdan, A., Reinberg, A., Auzeby, A., Beck, H., and 588 Molecular & Cellular Proteomics 5.4 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Molecular & Cellular Proteomics American Society for Biochemistry and Molecular Biology

Quantitative Mass Spectrometric Multiple Reaction Monitoring Assays for Major Plasma Proteins *

Download PDF
16 pages

Loading next page...
 
/lp/american-society-for-biochemistry-and-molecular-biology/quantitative-mass-spectrometric-multiple-reaction-monitoring-assays-NvB0TlaaQo

References (35)

Publisher
American Society for Biochemistry and Molecular Biology
Copyright
Copyright © 2006 Elsevier Inc.
ISSN
1535-9476
eISSN
1535-9484
DOI
10.1074/mcp.m500331-mcp200
Publisher site
See Article on Publisher Site

Abstract

Research Quantitative Mass Spectrometric Multiple Reaction Monitoring Assays for Major Plasma Proteins* Leigh Anderson‡§ and Christie L. Hunter¶ creased sensitivity (into the pg/ml range) and precision (CVs Quantitative LC-MS/MS assays were designed for tryptic peptides representing 53 high and medium abundance  5–10%) at the cost of restricting discovery potential toward proteins in human plasma using a multiplexed multiple novel proteins (3). In practice, a combination of these ap- reaction monitoring (MRM) approach. Of these, 47 pro- proaches (one or more survey approaches for de novo bi- duced acceptable quantitative data, demonstrating with- omarker discovery coupled with a candidate-based approach in-run coefficients of variation (CVs) (n  10) of 2–22% to biomarker validation in large sample sets) appears to pro- (78% of assays had CV <10%). A number of peptides gave vide an effective staged pipeline (4) for generation of valid CVs in the range 2–7% in five experiments (10 replicate plasma biomarkers of disease, risk, and therapeutic response. runs each) continuously measuring 137 MRMs, demon- Candidate-based specific assays rely on the specificity of strating the precision achievable in complex digests. De- capture or detection methods to select a specific molecule as pletion of six high abundance proteins by immunosub- analyte. Capture reagents such as antibodies can provide traction significantly improved CVs compared with whole extreme specificity (particularly when two different antibodies plasma, but analytes could be detected in both sample types. Replicate digest and depletion/digest runs yielded are used as in a sandwich immunoassay) and form the basis correlation coefficients (R ) of 0.995 and 0.989, respec- of most existing clinical protein assays. There is intense in- tively. Absolute analyte specificity for each peptide was terest in miniaturizing sets of such assays (5, 6) in array demonstrated using MRM-triggered MS/MS scans. Reli- formats (on planar substrates, beads, etc.), although signifi- able detection of L-selectin (measured at 0.67 g/ml) in- cant problems remain in the production of suitable antibodies dicates that proteins down to the g/ml level can be and in the simultaneous optimization of multiple assays in one quantitated in plasma with minimal sample preparation, fluid volume. yielding a dynamic range of 4.5 orders of magnitude in a Mass spectrometry provides an alternative assay approach, single experiment. Peptide MRM measurements in plasma relying on the discriminating power of mass analyzers to digests thus provide a rapid and specific assay platform for select a specific analyte and on ion current measurements for biomarker validation, one that can be extended to lower quantitation. In the field of analytical chemistry, many small abundance proteins by enrichment of specific target pep- tides (stable isotope standards and capture by anti-peptide molecule analytes (e.g. drug metabolites (7), hormones (8), antibodies (SISCAPA)). Molecular & Cellular Proteomics protein degradation products (9), and pesticides (10)) are 5:573–588, 2006. routinely measured using this approach at high throughput with great precision (CV  5%). Most such assays use elec- trospray ionization followed by two stages of mass selection: Accurate quantitation of proteins and peptides in plasma a first stage (MS1) selecting the mass of the intact analyte and serum is a challenging problem because of the complex- (parent ion) and, after fragmentation of the parent by collision ity and extreme dynamic range that characterize these sam- with gas atoms, a second stage (MS2) selecting a specific ples (1). The widely adopted separative (survey) approach to fragment of the parent, collectively generating a selected proteomics, in which an attempt is made to detect all com- reaction monitoring (plural MRM) assay. The two mass filters ponents, has proven to be limited in sensitivity toward low produce a very specific and sensitive response for the se- abundance proteins (2) and typically provides limited quanti- lected analyte that can be used to detect and integrate a peak tative precision. The alternative candidate-based approach, in a simple one-dimensional chromatographic separation of which relies on specific assays optimized for quantitative the sample. In principle, this MS-based approach can provide detection of selected proteins, can provide significantly in- The abbreviations used are: CV, coefficient of variation; QqQ, From the ‡The Plasma Proteome Institute, Washington, D. C. triple quadrupole; MRM, multiple reaction monitoring; S/N, signal-to- 20009-3450 and ¶Applied Biosystems, Foster City, California 94404 noise; Nat, natural sample-derived peptide; SIS, stable isotope-la- Received, October 7, 2005, and in revised form, November 28, beled internal standard; polySIS, polyprotein SIS; SISCAPA, stable 2005 isotope standards and capture by anti-peptide antibodies; MIDAS, Published, MCP Papers in Press, December 6, 2005, DOI 10.1074/ multiple reaction monitoring-initiated detection and sequencing; mcp.M500331-MCP200 MARS, multiple affinity removal system. © 2006 by The American Society for Biochemistry and Molecular Biology, Inc. Molecular & Cellular Proteomics 5.4 573 This paper is available on line at http://www.mcponline.org This is an Open Access article under the CC BY license. MS Assays for Plasma Proteins EXPERIMENTAL PROCEDURES absolute structural specificity for the analyte, and in combi- nation with appropriate stable isotope-labeled internal stand- Peptide Selection and MRM Design—We developed a set of MRMs iteratively using three basic approaches: pure in silico design from ards (SISs), it can provide absolute quantitation of analyte sequence databases, design from available LC-MS/MS proteomic concentration. These measurements have been multiplexed survey data, and comprehensive MRM testing of all of the candidate to provide 30 or more specific assays in one run (11). Such peptides of a protein. methods are slowly gaining acceptance in the clinical labora- In our initial attempt to generate MRMs by purely in silico methods, tory for the routine measurement of endogenous metabolites a set of 177 proteins and protein forms was assembled that are (e.g. in screening newborns for a panel of inborn errors of demonstrated or have potential to be plasma markers of some aspect of cardiovascular disease (20), and a subset of 62 proteins was metabolism (12)) and some drugs (e.g. immunosuppressants selected for which an estimate of normal plasma abundance was (13)). available. Predicted tryptic peptides for each of these were generated Recently the MRM assay approach has been applied to the along with relevant Swiss-Prot annotations and a series of computed measurement of specific peptides in complex mixtures such physicochemical parameters: e.g. amino acid composition, peptide as tryptic digests of plasma (3). In this case, a specific tryptic mass, Hopp-Woods hydrophilicity (21), and predicted retention time in reversed-phase (C ) chromatography (22). An index of the likeli- peptide can be selected as a stoichiometric representative of 18 hood of experimental detection was derived from a data set reported the protein from which it is cleaved and quantitated against a by Adkins et al. (23) by counting the number of separate “hits” for the spiked internal standard (a synthetic stable isotope-labeled peptide in the data set divided by the number of hits for the most peptide (14)) to yield a measure of protein concentration. In frequently detected peptide from the same protein. An overall index of principle, such an assay requires only knowledge of the peptide quality was generated according to a formula that gave positive weights to Pro, KP, RP, and DP content and negative weights masses of the selected peptide and its fragment ions and an to Cys, Trp, Met, chymotrypsin sites, certain Swiss-Prot features ability to make the stable isotope-labeled version. C-reactive (carbohydrate attachment, modified residues, sequence conflicts, or protein (3), apolipoprotein A-I (15), human growth hormone genetic variants), and mass less than 800 or greater than 2000. The (16), and prostate-specific antigen (17) have been measured 3619 tryptic peptides predicted for the 62 protein marker candidates in plasma or serum using this approach. Because the sensi- (6–497 peptides per target) ranged in length from 1 to 285 amino acids. Within the useful range of 8–24 amino acids, 721 peptides had tivity of these assays is limited by mass spectrometer dy- a C-terminal Lys and 690 had a C-terminal Arg. In this report, pep- namic range and by the capacity and resolution of the assist- tides from 30 of these target proteins ending in C-terminal Lys were ing chromatography separation(s), hybrid methods have also selected for further study. been developed coupling MRM assays with enrichment of We also selected peptides based on two types of direct proteomic proteins by immunodepletion and size exclusion chromatog- survey experiments. In the first case we carried out classical LC- MS/MS analysis of plasma digests in which the major ions observed raphy (18) or enrichment of peptides by antibody capture were subjected to MS/MS using the ion trap capabilities of the 4000 (SISCAPA (19)). In essence the latter approach uses the mass Q TRAP instrument. The identified peptides showing the best signal spectrometer as a “second antibody” that has absolute struc- intensity and chromatographic peak shape for a given parent protein tural specificity. SISCAPA has been shown to extend the were selected. In addition, we used the Global Proteome Machine sensitivity of a peptide assay by at least 2 orders of magnitude database of Craig et al. (24) to select peptides from target proteins that were frequently detected (multiple experiments). (19) and with further development appears capable of extend- Finally we used an adaptation of the MIDAS workflow, described ing the MRM method to cover the full known dynamic range of previously for discovering post-translational modifications (25, 26), to plasma (i.e. to the pg/ml level). look for measurable tryptic peptides from a variety of plasma proteins. There is compelling evidence that high and medium abun- In this approach, the protein sequence is digested in silico, likely y-ion dance plasma proteins have value as clinical biomarkers and fragments are predicted, and theoretical MRMs are generated for all the thus that there may be applications for specific MRM assays peptides in an acceptable size window. These MRMs are then used as a survey scan in a data-dependent experiment to detect specific pep- even without antibody enrichment. One may therefore ask how tide peaks, and each resulting MRM peak is examined by full scan many plasma proteins can be measured by quantitating their MS/MS to obtain sequence verification of the hypothesized peptide. To peptides in a plasma digest, and how precise could these verify peptide specificity in designated protein targets, selected pep- measurements be? If the measurement strategy proves to be tides were searched with BLASTP for exact matches against the ge- nome-derived human, mouse, and rat Ensembl peptides using Ensembl robust, could it be carried out using existing high throughput Blastview (www.ensembl.org/Homo_sapiens/blastview). LC-MS/MS platforms? To address these questions we gener- “Random” MRMs—Two approaches were used to generate pseu- ated and tested MRM assays based on peptides from a variety dorandom MRMs. In the first case we used 100 MS1 values distrib- of high to medium abundance plasma proteins to see how many uted randomly (by the Excel RAND function) between 408.5 and could be measured effectively by LC-MS/MS with and without 1290.2 (the maximum and minimum of an early set of real MRMs we tested) paired with MS2 values chosen randomly between this MS1 subtraction of the most abundant proteins. An understanding of and the maximum of the real MRMs (1495.6), thus mimicking the the performance of MS/MS in this application could enable properties of our real MRMs (which are generally 2 charge state routine and relatively inexpensive measurement of classical peptides and 1 charge fragments). In a second set we paired 131 plasma proteins and also provide a foundation for use of MS1 values chosen randomly from among MS1 values in a large table MS/MS in more sensitive anti-peptide antibody-enhanced SIS- of real MRMs with MS2 values chosen randomly from the real MS2 CAPA assays for low abundance proteins. values of the same list, imposing only the constraint that each MS2 574 Molecular & Cellular Proteomics 5.4 MS Assays for Plasma Proteins had to be between 1 and 2 times the paired MS1 mass (to approxi- RESULTS mate our selection criteria for real MRMs). Peaks detected in these Design of MRM Assays for Abundant Plasma Proteins—In MRMs were examined by triggering MS/MS (the MIDAS workflow). an initial approach to the selection of representative peptides Reagents—The following chemicals were used: trypsin (Promega), sodium dodecyl sulfate (Bio-Rad), iodoacetamide (Sigma), formic for MRM assays, a single peptide of 8–18 amino acids was acid (Sigma), tris-(2-carboxyethyl)phosphine (Sigma), and acetonitrile chosen from each of 30 proteins spanning a broad range of (Burdick and Jackson). plasma concentrations (6.6  10 down to 1 fmol/ml normal Plasma Depletion and Digestion—All experiments were performed concentration) based on computed characteristics alone (19). on aliquots of a single human plasma sample from a normal volunteer. MRMs were designed assuming doubly charged peptide ions The six highest abundance proteins were depleted from plasma using the multiple affinity removal system (“MARS”; Agilent Technologies and using fragments selected as likely y-ions above the m/z of spin columns) according to the manufacturer’s protocol. Depleted the 2 parent ion with collision energies assigned by a ge- sample was then exchanged into 50 mM ammonium bicarbonate neric formula (CE  0.05  m/z  5), and the peptides were using a VivaSpin concentrator (5000 molecular weight cutoff, Viva- expressed as a concatamer polySIS protein containing single Science). Undepleted plasma was also desalted before digestion. 13 15 Both depleted or undepleted plasma samples were denatured and copies of each peptide labeled with [U- C ,U- N ]lysine. 6 2 reduced by incubating proteins in 0.05% SDS and 5 mM tris-(2- When a tryptic digest of the polySIS was analyzed, all 30 carboxyethyl)phosphine at 60 °C for 15 min. The sample was then peptides were detected by MRMs. When a digest of whole adjusted to 10 mM with iodoacetamide and incubated for 15 min at human plasma was added to the polySIS peptides, 19 of the 25 °C in the dark. Trypsin was added in one aliquot (protease:protein labeled polySIS peptides were still detected by the same ratio of 1:20) and incubated for5hat37 °C. Labeled Peptide Internal Standards: polySIS—A series of SIS pep- MRMs, but only 11 of the plasma digest-derived unlabeled tides was added to samples in selected experiments by spiking with cognate peptides were detected (by the same MRMs ad- the tryptic digest of a “polySIS” polyprotein. Briefly this protein was justed for isotope label masses). produced by cell-free transcription and translation of a synthetic gene Because different peptides from a single protein can vary coding for 30 concatenated tryptic peptide sequences (derived from 13 15 widely in detectability by ESI-MS, we attempted to improve 30 plasma proteins) in the presence of U- C ,U- N -labeled lysine 6 2 (a total mass increment of 8 amu compared with the natural peptide). upon the in silico approach to MRM design using experimen- The 30 peptides were selected based on our initial in silico MRM tal data from a conventional peptide survey scan of a human design approaches and are thus not fully optimized using experimen- plasma digest and applying the selection criteria to peptides tal data. Of these peptides, 13 were used in the present studies (the with demonstrated detectability. Using a 3-h LC gradient, remainder were not reproducibly detected in digested plasma with peak area 1E04). The positioning of the label atoms at the extreme MS/MS scans were collected for the major doubly or triply C terminus of each peptide has the effect that all fragments that charged ions across the separation using information-de- contain the C terminus (i.e. the y-ions) will show the mass shift due to pendent data acquisition, and a second run was performed the label, whereas all the fragments that contain the N terminus (and using time-filtered exclusions of the peptide ions detected in hence have lost one of more C-terminal residues: the b-series ions) the first run. The combined results identified 54 plasma pro- will have the same masses as the corresponding fragments from the natural (sample-derived) target protein. These features (shifted y-ions teins ranging in abundance from albumin down to fibronectin and normal b-ions) provide a simplification in interpreting the frag- (normal plasma concentration of 300 g/ml). This experi- mentation patterns of the SIS peptides in comparison with the similar mental MS/MS data provided explicit information for peptide QCAT concept described recently by Beynon et al. (27). To determine selection, charge state, and most abundant y-ion m/z value the absolute concentration of polySIS protein, an aliquot was diluted with 1 M urea, 0.05% SDS, and 50 mM Tris, pH 8 and subjected to under the specific conditions used (i.e. electrospray ionization N-terminal Edman sequencing, yielding an initial concentration of 5 with collisional peptide fragmentation), allowing improved de- 1 pmol/l. A tryptic digest of the polySIS protein was spiked into sign of MRMs. When these MRMs were then used to analyze whole and depleted human plasma digests at the final concentrations the same sample in a subsequent run, triggering MS/MS shown in Table I. scans at any MRM signal, most of the peptides were detected LC-MS/MS Analysis—Plasma digests with and without added polySIS peptides were analyzed by electrospray LC-MS/MS using LC as peaks in the chromatogram and identified by database Packings (a division of Dionex, Synnyvale, CA) or Eksigent nanoflow search as expected. In most of these MRM chromatograms, LC systems (Table I) with 75-m-diameter C PepMap reversed- only a single peak was detected. phase columns (LC Packings) and eluted with gradients of 3–30% Because peptide detection sensitivity using MRM is ex- acetonitrile with 0.1% formic acid. A column oven (Keystone Scien- pected to be greater than that achieved in a full scan MS tific, Inc.) was used to maintain the column temperature at 35 °C. Electrospray MS data were collected using the NanoSpray source survey approach, a comprehensive de novo MRM design on a 4000 Q TRAP hybrid triple quadrupole/linear ion trap instrument method was explored for those proteins not detected in the (Applied Biosystems/MDS Sciex), and the peaks were integrated above survey experiment. Using a novel software tool, a large using quantitation procedures in the Analyst software 1.4.1 (Intelli- set of MRMs was generated for each of a series of target Quan algorithm). MRM transitions were acquired at unit resolution in proteins by selecting all predicted tryptic peptides in a useful both the Q1 and Q3 quadrupoles to maximize specificity. size range together with multiple high mass y-ion fragments of each (the “MIDAS” workflow (25, 26)). These MRMs were then L. Anderson and C. L. Hunter, manuscript in preparation . tested in LC-MS/MS runs of the unfractionated plasma digest, Molecular & Cellular Proteomics 5.4 575 MS Assays for Plasma Proteins FIG.1. MIDAS workflow (MRM-triggered MS/MS) verification of the identity of peptide DLQFVEVTDVK representing fibronectin (normal concentration, 300 g/ml) in a digest of depleted human plasma. A shows the ion current profile of MRM 647.3/789.4 (MS1/MS2), and B shows the MS/MS spectrum of peptide fragments taken at the time of the peak (* marks the y7 fragment monitored in this MRM). cps, counts per second. grouped in panels that included all the predicted tryptic pep- proach (indicated by an X in the SIS column of Table II): eight tides of one or two proteins at a time (50–100 MRMs per run), of the initial 30 in silico peptides were eliminated as likely to be with MS/MS scans triggered on any peaks observed. Of 12 too low abundance for detection, and better alternative pep- proteins examined, nine produced at least one usable MRM tides were selected from experimental data for four others. (signal-to-noise (S/N) ratio 20). For all but one of the peptides we elected to measure two MRM results from the above approaches were pooled, and fragments (i.e. using two MRMs per peptide), yielding 119 a set of optimized MRMs was assembled that covered a total MRMs. Finally we included MRMs for 18 stable isotope-la- of 60 peptides representing 53 proteins in human plasma beled internal standard (“SIS”) versions of target peptides (i.e. (Table II; seven proteins were represented by two peptides). the tryptic digest of the polySIS protein) spiked into the digest This set includes 18 peptides selected by the in silico ap- plasma samples. The resulting set of 137 MRMs was meas- 576 Molecular & Cellular Proteomics 5.4 MS Assays for Plasma Proteins TABLE I Summary design of data sets (experiments A–F) Replicate runs were performed in series with 30-min washes between runs. Load is expressed as the equivalent volume of plasma from which the sample was derived. Load factors express total or non-depleted (MARS flow-through) loads relative to experiment A. Load factor Replicate Equivalent plasma PolySIS Experiment Sample LC system runs volume loaded spike Total protein Non-depleted proteins l fmol A 10 Depleted plasma digest LC Packings 0.01 1 1 1.3 B 10 Whole plasma digest LC Packings 0.01 10 1 1.3 C 10 Whole plasma digest Eksigent 0.001 6 0.1 2.0 D 10 Depleted plasma digest Eksigent 0.01 0.6 1 2.0 E 10 Depleted plasma digest Eksigent 0.033 3.3 3.3 6.0 F1_1 4 Depletion 1, digest 1 Eksigent 0.01 1 1 F1_2 4 Depletion 1, digest 2 Eksigent 0.01 1 1 F2_1 4 Depletion 2, digest 1 Eksigent 0.01 1 1 F2_2 4 Depletion 2, digest 2 Eksigent 0.01 1 1 ured in all the replicate runs described below using an 18-ms dwell time per MRM and a resulting cycle time of 3s between measurements. After the final MRM method was constructed, each MRM transition and respective retention time were validated again as indicative of each specific peptide. Full scan MS/MS was acquired upon the appearance of the MRM signal, and each resultant spectrum was manually inspected to determine matching to the specific peptide (Fig. 1). Application to Digests of Whole and Depleted Plasma—Six experiments (A–F) were performed in each of which the same set of 137 MRMs was measured during sequential replicate LC-MS/MS runs of a single sample (same injection volume), and the appropriate peaks were integrated by the Analyst software to yield a value (peak area) for each MRM in each run. Experiments A–E (10 replicate runs each) are summarized in Table I. These experiments included tryptic digests of both whole (unfractionated) human plasma (experiments B and C) and plasma depleted of abundant proteins by MARS immu- FIG.2. Histogram of CVs of MRM values (peak areas) for five noaffinity subtraction (experiments A, D, and E), high (exper- experiments (experiments A–E) across all 137 MRMs (A)or iments B and E) and low (experiments A, C, and D) total across the 47 “best” MRMs (B). peptide loadings, and different chromatographic setups. Our objective was to assess the performance of the MRMs in various typical plasma digest experiments. The reference Fig. 2A shows histograms of the coefficients of variation for peptide load (experiment A) was derived from tryptic digestion the five experiments (individual values for each MRM are of the protein contained in 10 nl of plasma (700 ng of total contained in Table II). In analyses of depleted plasma, more protein assuming an average plasma concentration of 70 than 60% of the MRMs show within-run CVs less than 10%, mg/ml (28)) after depletion of the most abundant proteins using and almost half have CVs below 5%. A number of these an Agilent MARS spin column (84% of protein mass should be MRMs (e.g.  -antichymotrypsin, apolipoprotein E, he- removed based on a calculation using normal abundances). mopexin, heparin cofactor II, plasminogen, prothrombin, fi- This loading, comprising an estimated 110 ng of total peptides, brinogen  chain, complement C4, and factor B) showed an proved to be a loading compatible with very good nanoflow average within-run CV of 3–4% across three experiments, chromatography of the MRM peptides. Experiments B and E precision equivalent to that of good clinical immunoassays. used higher loadings to explore the trade-off between peak Analyses of whole (undepleted) plasma digests showed gen- stability (chromatographic quality, adversely affected by in- erally higher CVs (20–50% of MRMs had CV 10%). It is creased load) and S/N ratio (improved by increased analyte important to note that these reproducibility measures are quantity). We concluded that the 110 ng loading was optimal. computed on raw peak areas without correction using internal Chromatographic elution times were quite reproducible, show- standards. Four of the measured proteins were expected to ing average CVs of 2% (experiment D) and 2.5% (experiment E). be removed by the depletion process used (Agilent MARS Molecular & Cellular Proteomics 5.4 577 MS Assays for Plasma Proteins TABLE II A set of MRMs designed and tested for the detection of 53 proteins in human plasma or serum Protein name, tryptic peptide sequence, retention time in experiment D, peptide mass (MS1, m/z), singly charged fragment MS2 (m/z), mean peak area values, and CVs for 10 replicate analyses across five experiments are shown. 578 Molecular & Cellular Proteomics 5.4 MS Assays for Plasma Proteins TABLE II—continued Molecular & Cellular Proteomics 5.4 579 MS Assays for Plasma Proteins TABLE II—continued spin column). In comparing average peak areas obtained in compared with the lower of the two fragment CVs for each analyses of digests of whole and depleted samples, we found MRM, the average reduction in CV in experiments A–E ranges substantial reductions in albumin (1.3E08 reduced to from 0.7% to 0.1%. These small improvements in CV come 1E04), transferrin (1.5E05 reduced to 5E03), and at the cost of doubling the measurement time (or halving the haptoglobin (4.6E06 reduced to 1E05).  -Antitrypsin number of peptides monitored) and thus are unlikely to be was not detected reliably in any of the runs (presumably a bad useful in routine operation for improving reproducibility. peptide choice for MRM assay), and so its removal could not The relationship between CV and peak area, shown in Fig. be confirmed. We did not incorporate MRMs to measure the 3 for experiments A–E, indicates that peak areas below two immunoglobulins subtracted by the MARS column. 1E04 are unlikely to yield acceptable CVs (i.e. below 10%). Multiple measurements of an MRM would be expected to A cutoff of 1E04 corresponds to a signal-to-noise ratio of improve CVs, and we thus examined whether the sum of the 10, which is consistent with the quantitative requirement of two fragment MRMs measured separately for 59 of the pep- a S/N ratio of 10 for a reported lower limit of quantitation. The tides exhibited a smaller CV across 10 replicate runs than the highest peak areas measured (albumin peptides in whole individual MRMs. As shown in Table III, the average CV for the plasma digest samples) are above 1E08, demonstrating a summed MRMs across 59 peptides is 1–3% lower than the maximal working dynamic range of 4 orders of magnitude averages of either individual MRM. If the summed CV is above this cutoff. 580 Molecular & Cellular Proteomics 5.4 MS Assays for Plasma Proteins Calibration with Internal Standard Peptides—A set of 13 SIS slightly higher than CVs of the component measurements (80 peptides having the same sequences as 13 of the MRM- of 130 comparisons). measured tryptic peptides were used to assess the reproduc- Sensitivity—Two proteins with relatively low normal con- ibility of the quantitative measurements (five of the 18 SISs centrations in plasma were clearly detected among the MRMs measured by MRM yielded unusable quantitative information tested: L-selectin and fibronectin. The soluble form of L- because the peak area signal of either the standard or its selectin is a 33-kDa protein present in plasma at a normal cognate sample digest peptide fell below 1E04). The SIS concentration of 0.67 g/ml (29) or 20.3 pmol/ml. Fibronec- 13 15 tin is a 260-kDa protein present in plasma at a normal con- peptides were labeled with [U- C ,U- N ]lysine during syn- 6 2 centration of 300 g/ml (30) or 1200 pmol/ml. Given that an thesis from a synthetic DNA template in an in vitro transcrip- amount of digest corresponding to 0.01 l of plasma was tion/translation system and were cleaved from the polySIS loaded on column in experiment D, these peptides would be protein by trypsin. Samples spiked with SIS peptides were expected to be present on column at 200 and 12,000 amol, analyzed in experiments D and E (Table I) at relative peptide loadings of 1 and 3.3, respectively (SIS peptides at relative respectively. In the case of L-selectin we had spiked a SIS loadings of 1 and 3). Measured peak areas of MRM pep- peptide at 2.0 fmol and thus could determine that the natural tides showed an average 2.7 difference between these ex- (sample-derived) peptide was present at 0.1 times the amount of SIS (single point quantitation), yielding a measured 200 periments. Ratios between the natural sample-derived (Nat) amol and implied plasma concentration of 0.6–0.67 g/ml in and labeled (SIS) versions of these peptides were extremely good agreement with expectation. CVs for fibronectin in ex- similar in the two experiments (Fig. 4) and maintained a linear periments D and E were 4 and 4%, respectively, and for relationship (R  0.998) over almost 3 orders of magnitude in L-selectin were 22 and 11%, respectively, presumably reflect- concentration. Because the Nat:SIS ratios combine the vari- ing the fact that L-selectin was near the lower limit (1E04) ance of two measurements, the CVs of the ratios are typically for high quality detection. Six of the 53 selected target proteins were not reliably TABLE III observed, however. Three are probably explained by low nor- Improvement of CV obtained by combining peak areas of two mal abundances: we did not obtain a reproducible signal for fragment MRMs versus keeping them separate the selected peptides from coagulation factor V, vitamin K- Avg, average; frag(s), fragment(s). dependent protein C, or C4b-binding protein. There were also Avg CV Experiment instances in which peptides from more abundant proteins Sum of frags Frag 1 Frag 2 were not reliably detected: the inter- trypsin inhibitor light chain (despite the fact that a peptide from the heavy chain of A 10.5 11.8 14.8 this protein gave a good quality MRM), apolipoprotein C-II, B 16.2 20.0 19.4 C 11.0 13.0 14.4 and  -antitrypsin. In these latter cases, the problem is most D 8.0 9.4 12.3 likely poor choice of peptides: numerous alternative peptides E 8.5 9.4 11.9 exist for both the inter- trypsin inhibitor light chain and FIG.3. Plot of peak area versus CV for all MRMs in experiments A–E. Molecular & Cellular Proteomics 5.4 581 MS Assays for Plasma Proteins FIG.4. Comparison of computed amounts (based on ratios between Nat and SIS peptides) for 13 peptides in experiments D and E. Data are in- cluded for those ratios where both nu- merator and denominator peak areas were 1E04 in experiment D (13 peptides). -antitrypsin, but for small proteins, such as apolipoprotein The Density of Signals in MRM Space: Behavior of Random C-II, there may be no better alternative, and additional enrich- MRMs—Most of the MRMs we designed appeared to detect ment of these peptides will be necessary. only a single peak during the LC run of a complex digest such In general, depletion of the most abundant proteins using as depleted plasma (e.g. Fig. 1): whereas 73% had a robust the Agilent MARS column improved the performance of peak (defined as area 32,000) corresponding to the target MRMs for non-subtracted proteins substantially. This effect peptide analyte, only about 8% had a second peak meeting appeared to be due to improved detection sensitivity and the same peak area criterion (histograms in Fig. 7). We there- improved chromatographic peak shape (achieved by de- fore attempted to confirm that the density of peptide peaks in creasing the total peptide loaded by 6-fold), both of which “MRM space” was indeed low (equivalent to high MS/MS contribute to improved MRM peak signal-to-noise and lower detector specificity relative to sample complexity) by exam- CVs. Fig. 5 illustrates the benefit of depletion in removing the ining two types of randomized MRMs in the same depleted albumin peptide (major peak in Fig. 5A) and thus boosting the plasma digest sample. In a first set, 100 MRMs were gener- minor peaks in the depleted sample (Fig. 5B). At very high ated with “parent” masses randomly distributed over the loading of undepleted plasma digest, we noticed large shifts mass range of real peptides used in the 137 designed MRMs in peak retention times, but at loadings in the region of our and “fragment” masses randomly distributed between the nominal load the effect of high abundance peptides on MRM parent mass and the maximum fragment mass among the retention times was minor. designed MRMs (“random MRMs”). Of the 100 random Reproducibility of Immunodepletion and Tryptic Digestion— MRMs, only six showed a peak with area 32,000, and none In a sixth experiment (F series), we performed MARS deple- of these MRM peaks produced MS/MS spectra that led to a tion on two aliquots of the same plasma sample and then protein identification when searched with Mascot against separately digested two aliquots of each depleted sample Swiss-Prot. A second set of 131 random MRMs was gener- (total of four samples; e.g. F1_2 refers to the second digest of ated by randomly pairing parent and fragment masses from the first depletion). Four replicate runs of the 137 MRMs were the set of designed MRMs detectable in plasma, excluding carried out for each sample in randomized order to avoid any those cases where the fragment mass was lower than the sequence effect. Fig. 6 compares the mean peak areas of two parent (“random combination MRMs”). By using real peptide digests of a single depleted sample (Fig. 6A, F1_1 versus and fragment masses, these MRMs avoided any potential F1_2) and two depletions (Fig. 6B, F1_1 versus F2_1). The bias arising from the tendency of real peptide masses to peak area values span 4 orders of magnitude of which 2.5 cluster around integral masses (the mass defect). In this sec- orders is usable (i.e. beginning at peak area 1E04, the ond set, about 12% of the MRMs exhibited a peak with peak approximate cutoff below which CVs become large). Dupli- area 32,000, and once again none of these peaks gave cate digests show excellent comparability (R  0.995 and MS/MS spectra yielding a protein identification. All the peaks 0.998 for F1_1 versus F1_2 and F2_1 versus F2_2, respec- observed in the random MRM sets occurred late in the LC tively). Duplicate depletions (which necessarily include the gradient (after 100 min) after the elution of a large majority of effects of different digests as well) are only slightly worse (e.g. the designed plasma protein MRMs. These results suggest R  0.989 and 0.991 for F2_1 versus F1_1 or F2_2 versus that the density of quantifiable features in MRM space at our F1_2, respectively). current sensitivity, even for a very complex peptide sample 582 Molecular & Cellular Proteomics 5.4 MS Assays for Plasma Proteins FIG.5. Total ion current profiles across the chromatographic peptide separation for digests of undepleted (A, whole) and depleted (B) plasma from experiments C and D, respec- tively. Pie charts represent the protein composition of the samples: whole plasma contains 50% albumin (stip- pled region), whereas the proteins re- maining after MARS depletion include fibrinogen,  -macroglobulin, comple- ment C3, and all other lower abundance proteins (four remaining pie slices or- dered clockwise from 12 o’clock), show- ing the proportion of protein removed by depletion (white segments in B). The ma- jor peptide in A at 20.3 min is derived from albumin (whose abundance is shown by the black pie segment with white dots). cps, counts per second. and using unit resolution in both mass analyzers, is only eight were dropped before testing because of expected in- 6–12% of which a minority may be canonical tryptic peptides. sufficient abundance. Thus 13 in silico selections survived, The distribution of peak areas observed for random MRMs whereas 10 were replaced in testing. closely matches the distribution for second (non-target pep- Of the six failures (see above), we expect three proteins to tide) peaks in our designed MRMs, indicating that these ad- be within the concentration range that should be measurable ditional peaks represent a random background. and are selecting substitute peptides for these. The distribu- Selection of Best MRMs for Plasma Proteins—From the 119 tion of CVs for the 47 best MRMs is shown in Fig. 2B:in tested MRMs for sample peptides, we selected the best frag- experiment D, 40 of these had CVs below 10%, and 19 had ment ions (from 59 cases of two tested fragment ions) and the CVs below 5%. best peptide (from seven cases where two peptides were DISCUSSION tested per protein). A usable MRM resulted for 47 of the 53 initial proteins (indicated by an X in the Best MRM column in We designed and characterized mass spectrometric MRM Table II). Each of the 47 peptide sequences was verified as assays for tryptic peptides representing 53 high-to-medium unique in the human proteome (represented by the Ensembl abundance proteins of human plasma. The resulting panel of peptides) and occurred only once in the target protein. Three 47 successful MRMs, having within-run CVs ranging from 2 to of the peptides (representing antithrombin III, apolipoprotein 22% in depleted plasma, should be useful in the exploration E, and vitamin K-dependent protein C) occur in the mouse of protein-disease relationships and expression control in homologs as well, and seven (apolipoprotein E, vitamin K-de- clinical serum and plasma specimens. The fact that MRM pendent protein C, complement C4  and , fibronectin, assays, once developed, can be implemented on triple qua- haptoglobin , and inter- trypsin inhibitor heavy chain) occur drupole (QqQ)-MS instruments, widely used for precise in rat homologs (all the other human sequences did not occur quantitation of small molecules at high throughput, offers in the other species’ Ensembl peptides). the prospect of data collection across very large clinical Of the final 47 MRMs, 12 were contributed by the in silico sample sets. Our approach exemplifies the development of approach leading to the 30 polySIS peptides, and one (he- targeted specific assays needed to enable the statistical val- mopexin) was contributed by an earlier in silico effort (19): idation of proposed biomarkers in plasma (“candidate-based these are indicated in Table II by an X in the SIS column for the proteomics” (4)). associated Lys-labeled internal standard. A total of eight in Our initial attempts to select usable tryptic peptides by silico selections were replaced by better performing peptides purely in silico means were reasonably successful as approx- from the same target protein as a result of experimental imately half of the peptides chosen (13 of 23) produced ac- testing (four before and four after selection of the 137 MRMs), ceptable MS signals in plasma. Although the prediction of two subsequently failed and have not yet been replaced, and ionization properties of tryptic peptides can be expected to Molecular & Cellular Proteomics 5.4 583 MS Assays for Plasma Proteins general process for MRM design using these sources of in- formation is shown in Fig. 8 in which bold arrows indicate the preferred approach. The final step in the process, selection among alternative candidate MRMs for a given protein based on measured CVs in replicate runs, is recommended because of the significant differences in CV observed among MRMs with similar signal strength: CV appears to be a characteristic of a peptide to some extent separate from ion current. Despite the complexity of plasma digests (particularly those of depleted plasma where a small number of superabundant peptides have been removed), most MRMs exhibited only a single peak across the peptide LC chromatogram. This ob- servation is consistent with the low density of peaks in two sets of randomly distributed MRMs measured in depleted plasma digests and demonstrates the exquisite specificity of the two-stage QqQ-MS selection process used as the detec- tor. Nevertheless the existence of secondary peaks (whether or not they are actually tryptic peptides) in a small subset (10%) of MRMs indicates that chromatographic elution time is an important additional factor in providing the absolute analyte specificity desired in these assays. Following more extensive experience with specific MRMs and any variation that may occur in them due to shifts in peak elution times, it may be possible to establish a set of plasma MRMs that are truly free of secondary peaks and that could therefore poten- tially be measured using short LC separations (constrained only by increases in ion suppression effects as peptides crowd together). The unambiguous measurement of a peptide derived from FIG.6. Comparison of mean peak areas values for all MRMs in L-selectin (normal concentration of 0.67) suggests that pro- two replicate digests (A) and two depletion runs (and subsequent digests) (B). Error bars show 1 standard deviation computed from teins present in normal plasma at 1 g/ml are measurable replicate runs. by MRM in depleted and undepleted samples with minimal upfront sample fractionation. Different tryptic peptides from the same protein can produce ion currents differing by factors improve substantially in the future, we used experimental of at least 1E03 in LC-MS/MS experiments, excluding the MS/MS data, in combination with computational methods, to peptides that are not detected at all. This variation is due to select more successful target peptides. Two experimental multiple factors, including propensity to ionize in the electro- methods proved particularly useful. High abundance peptides spray source, coincidence in elution time with other easily were detected in conventional LC-MS/MS data-dependent ionizing peptides, efficiency of release during tryptic diges- full scan MS experiments in which a subset of high signal tion, and presence of unrecognized post-translational modi- peptides seen in MS1 are subjected to MS/MS. Lower abun- fications arising due to biology or during sample preparation. dance peptides were detected by constructing lists of candi- Although we do not yet know the frequency of high efficiency date MRMs to all appropriately sized predicted tryptic pep- peptides among the tryptic products of each protein, we tides from a target protein and then characterizing any detected MRM peaks by MS/MS (the MIDAS workflow in believe that numerous additional plasma proteins can be which MRM methods are designed using a specifically de- measured down to 1 g/ml using MRMs and estimate that signed script within the Analyst software). Because MRMs are 50–100 proteins may be added to our list as we extend this typically more sensitive than full scan survey MS for detection effort to additional candidates from the plasma component of very low abundance components, the MIDAS approach database (2, 20). allowed us to find successful MRMs for more lower abun- Given that albumin, the most abundant protein in plasma, is dance peptides and represents the preferred strategy for measurable by MRM in undepleted plasma in the same ex- MRM design going forward. This process was facilitated by periment that detects L-selectin, the dynamic range of quan- the combination of high sensitivity triple quadrupole MRM and titatively measurable proteins appears to be 1 g/ml to 55 ion trap MS/MS scan capabilities on the hybrid triple quadru- mg/ml or 5E04. This dynamic range of 4–5 orders of pole linear ion trap 4000 Q TRAP mass spectrometer. A magnitude is consistent with typical quantitation experiments 584 Molecular & Cellular Proteomics 5.4 MS Assays for Plasma Proteins FIG.7. Histograms of peak areas obtained with four sets of MRMs in a digest of depleted plasma. it is easily possible to select lower signal peptides from high abundance proteins (and high signal peptides from low abun- dance proteins) to diminish the dynamic range requirement at the peptide level. As long as stable isotope-labeled internal standards are used, the quantitative information contained in Nat:SIS ratios should be unaffected by peptide choice. The reproducibility of peptide MRM measurements was impressive in many cases. For the three experiments analyz- ing depleted plasma digests, an average of 78% of the 47 best MRMs had within-run CVs of 10% or less, and 36% had CVs less than 5%. A number of MRMs gave low CVs across five experiments (i.e. in depleted and whole plasma digests at widely varying total loads) including peptides from hemopexin (CVs ranging between 2 and 6%), vitronectin (4–7%), kinino- gen (5–6%), complement factor B (3–6%), complement C4 (3–5%), apolipoprotein B100 (5–7%), antithrombin III (5–7%), and  -antichymotrypsin (2–7%). These precision values are comparable to many small molecule MRM measurements and approach the results of conventional immunoassays used in clinical diagnostics (32). They demonstrate clearly that pep- tide MRM measurements can be used in high precision quan- FIG.8. Schematic diagram of a process for selecting MRMs. titative assays. Based on reproducibility at this level, it ap- Thicker arrows represent the preferred process. Square-cornered pears likely that useful comparative data across samples boxes represent informatics components, and rounded-corner boxes might be obtained by comparing simple peak area measure- are experimental steps. ments (i.e. without SIS standardization) and that relatively small quantitative differences could be detected (differences performed using MRM on this instrument and covers almost of 20% would be 3 standard deviations from the mean for half the logwise dynamic range of currently known and meas- many of the best MRMs and thus very easily detected). The ured proteins in human plasma (1). The dynamic range of peak observation that individual peptides showed consistently area measurements appears to be similar, extending from higher or lower CVs suggests that quantitative precision de- 1.3E08  5% for undepleted albumin to less than 1E04 pends on characteristics of the peptide and perhaps its envi- (the approximate cutoff below which CVs become 10%). In ronment of co-eluting peptides. Therefore it will be useful in this context it is important to note that measurement of pro- future work to select MRM peptides based on CV across teins over this dynamic range does not require measuring peptides at such a broad range of signal intensities because replicate runs in addition to the characteristics normally con- Molecular & Cellular Proteomics 5.4 585 MS Assays for Plasma Proteins sidered (e.g. ion current and chromatographic peak shape). set. It thus appears that both depletion and digestion can be MRMs designed for absolute quantitation using a set of 13 carried out with sufficient reproducibility to provide useful SIS peptides (with sequences identical to 13 of the selected measurements and that MRM assays provide an ideal method tryptic peptides) performed well. When spiked at known load- for assessing sample preparation reproducibility. ing these provided internal standards for quantitation and The multiplexing capability of LC-QqQ-MS platforms for resulted in a measured value for L-selectin in almost exact measuring peptides in complex digests is substantial, provid- agreement with the literature value for normal subjects (0.67 ing an opportunity to measure large panels of proteins accu- g/ml). This result is encouraging but not definitive as we did rately in each run. Based on the performance of the present not measure target protein concentrations in our sample by set of 137 MRMs, which were all monitored continuously independent methods, and the individual SIS peptides (de- across the entire LC gradient as 18-ms sequential measure- rived by digestion of a quantitated precursor protein) were not ments, it is clear that 100–200 MRMs might be used routinely individually quantitated. Thus not all of our endogenous: to measure peptides in long LC gradients. However, given standard ratio measurements agreed so well with literature reproducible chromatographic elution times, it is possible with values. In addition, single point concentration curves as used existing systems to measure each MRM only during a short here do not provide the same accuracy in quantitation as time window when the peak is expected to occur (e.g. a multiple point curves. Nevertheless the extreme reproducibil- window of 10% of total run length given an average 2–2.5% ity (R  0.991) of the measured ratios (and implied absolute CV in peak elution time measured in our experiments D and peptide concentrations) between experiments D and E (Fig. 4) E). Once this approach is implemented, based on extensive across almost 3 orders of magnitude in peak area and 3-fold knowledge of elution time and column reproducibility, and difference in total peptide load are striking. To effectively provided that MRMs are selected that do not cluster too much standardize peptides (and proteins) over wider ranges, a more in elution time, substantially more MRMs could be used in a sophisticated strategy will be utilized using smaller equimolar single LC MRM experiment. groups of peptides selected to represent each decade of An additional important consideration for throughput of peptide concentration (i.e. one polySIS product or a set of MRM measurements is the duration of the chromatography equimolar SIS peptides for each decade of the abundance run. In our replicate experiments D and E, a 30-min gradient scale). In addition, the completeness of polySIS digestion, was used that led to a total cycle time (including intersample and thus the relative stoichiometry of the resulting SIS pep- wash) of 75 min. It would clearly be advantageous if this could tides, must be rigorously characterized. be reduced, allowing more samples to be run per day. The Depletion of the highest abundance plasma proteins using extreme analyte specificity indicated by the low density of an immobilized antibody column proved to be very useful in peaks in MRM space suggests that most of our MRMs should measuring our MRMs. We expected to be able to load a perform well with less benefit from chromatographic separa- digest of 6 times as much depleted plasma as undepleted tion, and the ability to focus the MRM measurements in dis- plasma because these two samples would contain approxi- crete time windows may allow more MRMs to be brought mately equal amounts of total peptides, and this was con- closer together in elution time without sacrificing the required firmed. We also confirmed that the reference loading chosen multiple measurements across each peak. Hence we expect (110 ng of peptides derived from digestion of 10 nl of that substantial improvements in run time should be possible, depleted plasma) was near but not above the limit for good probably in conjunction with a shift to higher flow rate (e.g. quality LC peak shape and reproducible chromatography in a capillary flow) systems providing increased robustness in rou- nanoflow system. However, we had few if any MRMs that tine operation. Higher flow rate LC systems have significantly were detectable in the depleted sample but not the unde- lower (10) absolute sensitivity than nanoflow, but in the pleted sample. The major difference emerged instead in the current application it would be easily possible to load 10 lower CVs of peaks measured in the depleted sample due more sample (i.e. 100 nl) as sample at these loadings is rarely presumably to the decreased level of competing peptides and limiting. consequent higher signal to noise. In future extensions of this Considered in a broad context, MRM assays appear to method to lower abundance proteins, it is likely that depletion have several advantages in addition to those described will expose numerous targets not otherwise detectable. De- above. 1) The instrumentation used to measure peptide pletion significantly reduced the levels of albumin, transferrin, MRMs is very similar to that used in existing robust platforms and haptoglobin as expected. for high throughput quantitative measurement of drug metab- Highly reproducible depletion and tryptic digestion are a olites in plasma, for the detection of inborn errors of metab- necessity as steps in routine sample preparation for MRM olism in newborns and for pesticide analysis. A large base of analysis. In a small study (experiment F) designed to look at compatible instrumentation and expertise thus already exists. these effects, R values 0.995 were obtained for peak areas 2) Because the target proteins are detected via tryptic peptide in duplicate digests, and R values 0.989 were obtained for surrogates after sample denaturation and digestion, peptide duplicate depletion and digestion in the same experimental MRM results should be insensitive to alterations in protein 586 Molecular & Cellular Proteomics 5.4 MS Assays for Plasma Proteins cation of C-reactive protein in the serum of patients with rheumatoid folding and intersubunit associations (factors that can nega- arthritis using multiple reaction monitoring mass spectrometry and C- tively affect immunoassays). 3) MRM results are obtained by labeled peptide standards. Proteomics 4, 1175–1186 integrating discrete, anticipated peaks in simple ion chro- 4. Anderson, N. L. (2005) The roles of multiple proteomics platforms in a pipeline for new diagnostics. Mol. Cell. Proteomics 4, 1441–1444 matograms using well established and widely used commer- 5. Joos, T. O., Stoll, D., and Templin, M. F. (2002) Miniaturised multiplexed cial software, and thus the method is not dependent on the immunoassays. Curr. Opin. Chem. Biol. 6, 76–80 much more complex data processing infrastructure required 6. Haab, B. B. (2005) Antibody arrays in cancer research. Mol. Cell. Proteom- to handle the peak matching and spectral identification chal- ics 4, 377–383 7. Kostiainen, R., Kotiaho, T., Kuuranne, T., and Auriola, S. (2003) Liquid lenges of a discovery proteomic pipeline. 4) It is probable, chromatography/atmospheric pressure ionization-mass spectrometry in although not yet demonstrated, that the cost per high quality drug metabolism studies. J. Mass. Spectrom. 38, 357–372 measurement will be lower than for immunoassays (both to 8. Tai, S. S., Bunk, D. M., White, E. T., and Welch, M. J. (2004) Development and evaluation of a reference measurement procedure for the determi- create the assay to begin with and to apply it to large sample nation of total 3,3 ,5-triiodothyronine in human serum using isotope- sets). dilution liquid chromatography-tandem mass spectrometry. Anal. Chem. Important limitations of peptide MRMs must also be recog- 76, 5092–5096 9. Ahmed, N., and Thornalley, P. J. (2003) Quantitative screening of protein nized. 1) Although the method is applicable to post-transla- biomarkers of early glycation, advanced glycation, oxidation and nitro- tionally modified peptides, the post-translational modification sation in cellular and extracellular proteins by tandem mass spectrometry must be known or specifically hypothesized, and its proper- multiple reaction monitoring. Biochem. Soc. Trans. 31, 1417–1422 10. Sannino, A., Bolzoni, L., and Bandini, M. (2004) Application of liquid chro- ties must be designed into the assay up front (25) (and nor- matography with electrospray tandem mass spectrometry to the deter- mally cannot be discovered de novo by our approach). 2) mination of a new generation of pesticides in processed fruits and Some proteins may not readily produce usable peptides. The vegetables. J. Chromatogr. A 1036, 161–169 11. Barr, D. B., Barr, J. R., Maggio, V. L., Whitehead, R. D., Jr., Sadowski, M. A., protein may be too short and hence produce few candidates Whyatt, R. M., and Needham, L. L. (2002) A multi-analyte method for the to begin with or may, like many immunoglobulin sequences, quantification of contemporary pesticides in human serum and plasma be too variable. These cases will require further exploration using high-resolution mass spectrometry. J. Chromatogr. B. Anal. Tech- nol. Biomed. Life Sci. 778, 99–111 (alternative proteolytic enzymes or sample derivatization pro- 12. Roschinger, W., Olgemoller, B., Fingerhut, R., Liebl, B., and Roscher, A. A. cedures). 3) Genetic variants altering a single amino acid in (2003) Advances in analytical mass spectrometry to improve screening the selected peptide will prevent its determination by the for inherited metabolic diseases. Eur. J. Pediatr. 162, Suppl. 1, S67–S76 13. Streit, F., Armstrong, V. W., and Oellerich, M. (2002) Rapid liquid chroma- wild-type MRM; hence these must be recognized in advance tography-tandem mass spectrometry routine method for simultaneous and designed specifically. Provided these issues are taken determination of sirolimus, everolimus, tacrolimus, and cyclosporin a in into account, none appears to prevent productive use of whole blood. Clin. Chem. 48, 955–958 peptide MRMs for measuring abundances of plasma proteins. 14. Gerber, S. A., Rush, J., Stemman, O., Kirschner, M. W., and Gygi, S. P. (2003) Absolute quantification of proteins and phosphoproteins from cell Based on the results presented here, numerous clinically lysates by tandem MS. Proc. Natl. Acad. Sci. U. S. A. 100, 6940–6945 important plasma proteins can be measured by peptide 15. Barr, J. R., Maggio, V. L., Patterson, D. G., Jr., Cooper, G. R., Henderson, MRMs with precision comparable to current clinical immuno- L. O., Turner, W. E., Smith, S. J., Hannon, W. H., Needham, L. L., and Sampson, E. J. (1996) Isotope dilution-mass spectrometric quantification assays. The panel of assays presented should have general of specific proteins: model application with apolipoprotein A-I. Clin. use as the nucleus of candidate-based biomarker validation Chem. 42, 1676–1682 approach and should lead ultimately toward a comprehensive 16. Wu, S. L., Amato, H., Biringer, R., Choudhary, G., Shieh, P., and Hancock, W. S. (2002) Targeted proteomics of low-level proteins in human plasma assay platform for all human proteins in plasma. by LC/MSn: using human growth hormone as a model system. J. Pro- teome Res. 1, 459–465 Acknowledgment—We thank Dr. Tina Settineri (Applied Biosys- 17. Barnidge, D. R., Goodmanson, M. K., Klee, G. G., and Muddiman, D. C. tems) for advice and support. (2004) Absolute quantification of the model biomarker prostate-specific * The costs of publication of this article were defrayed in part by the antigen in serum by LC-MS/MS using protein cleavage and isotope payment of page charges. This article must therefore be hereby dilution mass spectrometry. J. Proteome Res. 3, 644–652 marked “advertisement” in accordance with 18 U.S.C. Section 1734 18. Liao, H., Wu, J., Kuhn, E., Chin, W., Chang, B., Jones, M. D., O’Neil, S., solely to indicate this fact. Clauser, K. R., Karl, J., Hasler, F., Roubenoff, R., Zolg, W., and Guild, § To whom correspondence should be addressed: The Plasma B. C. (2004) Use of mass spectrometry to identify protein biomarkers of Proteome Inst., P. O. Box 53450, Washington, D. C. 20009-3450. disease severity in the synovial fluid and serum of patients with rheuma- toid arthritis. Arthritis Rheum. 50, 3792–3803 Tel.: 301-728-1451; Fax: 202-234-9175; E-mail: leighanderson@ 19. Anderson, N. L., Anderson, N. G., Haines, L. R., Hardie, D. B., Olafson, plasmaproteome.org. R. W., and Pearson, T. W. (2004) Mass spectrometric quantitation of peptides and proteins using stable isotope standards and capture by REFERENCES anti-peptide antibodies (SISCAPA). J. Proteome Res. 3, 235–244 20. Anderson, N. L. (2005) Candidate-based proteomics in the search for 1. Anderson, N. L., and Anderson, N. G. (2002) The human plasma proteome: biomarkers of cardiovascular disease. J. Physiol. 563, 23–60 history, character, and diagnostic prospects. Mol. Cell. Proteomics 1, 21. Hopp, T. P., and Woods, K. R. (1981) Prediction of protein antigenic 845–867 determinants from amino acid sequences. Proc. Natl. Acad. Sci. U. S. A. 2. Anderson, N. L., Polanski, M., Pieper, R., Gatlin, T., Tirumalai, R. S., 78, 3824–3828 Conrads, T. P., Veenstra, T. D., Adkins, J. N., Pounds, J. G., Fagan, R., 22. Krokhin, O. V., Craig, R., Spicer, V., Ens, W., Standing, K. G., Beavis, R. C., and Lobley, A. (2004) The human plasma proteome: a non-redundant list and Wilkins, J. A. (2004) An improved model for prediction of retention developed by combination of four separate sources. Mol. Cell. Proteom- times of tryptic peptides in ion pair reversed-phase HPLC: its application ics 3, 311–326 3. Kuhn, E., Wu, J., Karl, J., Liao, H., Zolg, W., and Guild, B. (2004) Quantifi- to protein peptide mapping by off-line HPLC-MALDI MS. Mol. Cell. Molecular & Cellular Proteomics 5.4 587 MS Assays for Plasma Proteins Proteomics 3, 908–919 Guillet, P. (1986) Differences between young and elderly subjects in 23. Adkins, J. N., Varnum, S. M., Auberry, K. J., Moore, R. J., Angell, N. H., seasonal and circadian variations of total plasma proteins and blood Smith, R. D., Springer, D. L., and Pounds, J. G. (2002) Toward a human volume as reflected by hemoglobin, hematocrit, and erythrocyte counts. blood serum proteome: analysis by multidimensional separation coupled Clin. Chem. 32, 801–804 with mass spectrometry. Mol. Cell. Proteomics 1, 947–955 29. Atalar, E., Ozmen, F., Haznedaroglu, I., Acil, T., Ozer, N., Ovunc, K., 24. Craig, R., Cortens, J. P., and Beavis, R. C. (2004) Open source system for Aksoyek, S., and Kes, S. (2002) Effects of short-term atorvastatin treat- analyzing, validating, and storing protein identification data. J. Proteome ment on global fibrinolytic capacity, and sL-selectin and sFas levels in Res. 3, 1234–1242 hyperlipidemic patients with coronary artery disease. Int. J. Cardiol. 84, 25. Unwin, R. D., Griffiths, J. R., Leverentz, M. K., Grallert, A., Hagan, I. M., and 227–231 Whetton, A. D. (2005) Multiple reaction monitoring to identify sites of 30. van Helden, W. C., Kok-Verspuy, A., Harff, G. A., and van Kamp, G. J. protein phosphorylation with high sensitivity. Mol. Cell. Proteomics 4, (1985) Rate-nephelometric determination of fibronectin in plasma. Clin. 1134–1144 Chem. 31, 1182–1184 26. Cox, D. M., Zhong, F., Du, M., Duchoslav, E., Sakuma, T., and McDermott, 31. Deleted in proof J. C. (2005) Multiple reaction monitoring as a method for identifying 32. Lestin, M., Hergert, M., Lestin, H. G., Brinker, C., Storm, H., Kuhrt, E., Kuhrt, protein posttranslational modifications. J. Biomol. Tech. 16, 83–90 B., Lambrecht, H. G., and Kuhnel, W. (2002) Evaluation of the chemilu- 27. Beynon, R. J., Doherty, M. K., Pratt, J. M., and Gaskell, S. J. (2005) minescence immunoassays for the measurement of troponin I, myoglo- Multiplexed absolute quantification in proteomics using artificial QCAT bin and CK-MB using the IMMULITE system in comparison to other proteins of concatenated signature peptides. Nat. Methods 2, 587–589 measuring systems. Clin. Lab. 48, 211–221 28. Touitou, Y., Touitou, C., Bogdan, A., Reinberg, A., Auzeby, A., Beck, H., and 588 Molecular & Cellular Proteomics 5.4

Journal

Molecular & Cellular Proteomics – American Society for Biochemistry and Molecular Biology

Published: Apr 1, 2006

There are no references for this article.