08_Hortin.qxd 11/21/06 3:20 PM Page 103 Clinical Proteomics Copyright © 2006 Humana Press Inc. All rights of any nature whatsoever are reserved. ISSN 1542-6416/06/02:103–116/$30.00 (Online) Original Article Mass Determination of Major Plasma Proteins by Matrix-Assisted Laser Desorption/Ionization Time-of-Flight Mass Spectrometry Glen L. Hortin1,* and Alan T. Remaley1,2 1Department of Laboratory Medicine,National Institutes of Health,Bethesda,MD;and2National Heart Lung and Blood Institute,National Institutes of Health,Bethesda,MD and doubly charged monomer, and provided Abstract measured masses for these proteins. A Matrix-assisted laser desorption/ionization number of proteins, including albumin, trans- time-of-flight mass spectrometry (MALDI-TOF/ ferrin, apolipoproteins A-I, A-II, C-I, C-II, and MS) serves as a rapid and accurate means C-III, and prealbumin, could be analyzed todetermine masses of proteins independent directly in serum with appropriate dilution. of their shapes or interactions with other Measured values for masses of major plasma molecules. It provides one of the most funda- proteins will assist in analysis of serum and mental characterizations of major plasma pro- plasma. It is possible to analyze a number of teins. Purified proteins in saline or serum components by MALDI-TOF/MS directly in specimens were prepared for analysis by dilu- diluted serum. Extremely simple sample tion, mixing with a solution of sinapinic acid, preparation techniques may be useful in ana- and drying on a target plate. Specimens were lyzing structural variation of several major analyzed in a linear TOF mode with external plasma proteins, particularly those with calibration. Analyses of 24 purified plasma masses <30 kDa, including a number of proteins showed predominance of singly apolipoproteins and markers of nutritional charged ions with lesser amounts of dimer status or acute phase responses. *Author to whom all correspondence and reprint requests should be addressed: Glen L. Hortin, Department of Laboratory Medicine, National Institutes of Health, Building 10, Room 2C-407, Bethesda, MD 20892. E-mail: [email protected] 103 08_Hortin.qxd 11/21/06 3:20 PM Page 104 104 ________________________________________________________________________Hortin and Remaley Introduction might be applicable for routine in the clinical laboratory and that would avoid modification Matrix-assisted laser desorption/ionization or selective losses of components during time-of-flight mass spectrometry (MALDI- preparation. TOF/MS), invented in the late 1980s (1,2), serves as a technique for analyzing peptides Materials and Methods and proteins in relatively complex samples. It has even been employed for the direct analysis Peptides and proteins used as calibrators and of tissue specimens (3). In this technique, speci- sinapinic acid used as the specimen matrix were mens containing peptides and protein are dried purchased from Bruker Daltonics (Billerica, on a target plate together with a light-absorbing MA). Masses of calibrators were: (bovine matrix molecule. Vaporization of the mixture of insulin)+, 5734.61; (equine cytochrome c)+, protein and matrix by a laser releases ionized 12,361.20; (equine myoglobin)+, 16,952.62; protein molecules, often with a single charge, (bovine serum albumin)2+, 33,216; (bovine which can be analyzed in a TOF mass spec- serum albumin)+, 66,431; (bovine serum albu- trometer. It serves as a simple and rapid analyt- min dimer)+, 132,859. Purified proteins were ical technique for determining the molecular obtained from the following sources: human weights and heterogeneity of small amounts of holo-transferrin, prealbumin, immunoglobulin α α protein. In general, intact molecular ions are G (reagent grade), 1-antitrypsin, and 1-acid formed from individual peptide chains, allow- glycoprotein, from Sigma Chemical (St. Louis, ing accurate determination of polypeptide mass MO); human haptoglobin (mixed types), α α (4–6). However, there may be some technique- 2HS-glycoprotein, Gc globulin, 2-macroglob- dependent variation in measured mass related ulin, immunoglobulin A, neutrophil lysozyme, to variable loss of labile groups, such as sialic and C-reactive protein from Calbiochem (La β acids or association of sodium or matrix with Jolla, CA); 2-glycoprotein I, Glu-plasminogen, protein molecular ions (4–10). and antithrombin III from Haematologic Tech- Sample preparation is a key step in the nologies, Inc. (Essex Junction, VT). Lipopro- analysis of complex specimens by MALDI- teins including apolipoproteins A-I, A-II, C-I, TOF/MS. Variables, such as protein con- C-II, C-III, and E were purified by standard centration, salt concentration, or specimen published procedures (17). Purified proteins fractionation, may strongly influence the were generally in stock solutions from 1 to observed spectra. Usual approaches for sample 10g/L in concentration. Serum specimens preparation for MALDI-TOF/MS involve were remnants of human specimens submitted desalting and cleanup of specimens by tech- to the clinical laboratory used in accordance niques such as dialysis, solid-phase extrac- with a protocol approved by an Institutional tions, or fractionation through absorption to Review Board. a specialized target surface (in a variant of Calculated masses of proteins were based MALDI-TOF/MS that has been termed sur- on amino acid sequences in well-validated face-enhanced laser desorption/ionization or databases, usually the Swiss-Prot database SELDI) or to magnetic beads (5–16). These (www.ebi.ac.uk/swissprot). Calculations of preparation steps involve significant time, polypeptide masses were performed with expense, and potential losses of selected com- software offered by the National Institute ponents. In the present study, we examined of Diabetes and Digestive and Kidney Dis- direct analysis of serum and purified serum eases mass spectrometry group (sx.102a. components as a very simple approach that niddk.nih.gov/peptide). Calculated peptide Clinical Proteomics________________________________________________________________Volume 2, 2006 08_Hortin.qxd 11/21/06 3:20 PM Page 105 Direct Analysis of Serum Proteins _____________________________________________________________105 masses did not include the masses of A–C, respectively, or for apolipoprotein A-I oligosaccharide components. and transthyretin in Fig. 2. These analyses Analyses were performed with an Ultraflex served as a relatively rapid method to deter- TOF mass spectrometer (Bruker Daltonics) in a mine the mass of specific components and to linear positive ion mode similar to previously evaluate the tendency to form dimers and described analyses (12). Specimens were multiply charged ions. analyzed with two programs of separation Table 1 lists the measured masses of 23 parameters. To examine an m/z range of plasma proteins that were available as puri- 5000 to 30,000, pulsed extraction had a delay fied components, plus serum amyloid A, of 300 ns. To examine m/z > 30,000, there was which was observed in serum from a patient a delay of 550 ns. Specimens were prepared by with an acute phase response as characterized dilution with 10 mmol/L ammonium acetate in other studies (18). The broader peaks of and then by mixing 1 vol of diluted specimen high molecular weight proteins, probably with 2 vol of matrix solution containing 10 g/L reflecting both lower resolution and increased sinapinic acid in 40% acetonitrile/10% microheterogeneity from variable glycosyla- ethanol/50% water containing 0.1% trifluo- tion, led to greater imprecision in the mea- roacetic acid. Specimens were applied twice to surement of masses for these components. a 384-position sample plate as 1-µL aliquots Comparison of measured values in Table 1 vs and allowed to air-dry. Specimens were ana- older published values based on techniques lyzed with manual laser positioning and such as compositional analysis, electrophoresis, adjustment of laser power. All figures shown and ultracentrifugation generally showed represent data collected from 10 positions, fairly good agreement for proteins lacking summing a total of 300 laser pulses. Measure- oligosaccharides. MALDI-TOF/MS provided ments of m/z were performed using external lower estimates of masses for most glycopro- calibration. Masses of proteins were measured teins than the classical textbook values. Mea- at the peak apex. sured values for several proteins with high α carbohydrate content such as -acid glyco- Results 1 α protein, antithrombin III, -HS-glycoprotein, 2 α Most serum proteins are globulins that and -antichymotrypsin were more than 10% 1 require the presence of some salt to maintain lower than older published values. Measured solubility, but salts tend to suppress ionization values for glycoprotein I and plasminogen of proteins in MALDI-TOF/MS. Therefore, were about 10% higher than published values. there is a compromise between physiological MALDI-TOF/MS should provide highly salt conditions that maintain protein solubility accurate of proteins without posttransla- and the efficiency of ionization. We examined tional modifications, and, in most cases, the ability to analyze purified proteins in there is good agreement for these proteins saline solutions by MALDI-TOF/MS, using with calculated masses based on amino acid simple dilution of the proteins with matrix sequence databases. For proteins bearing com- solution. Generally, these analyses yielded a plex oligosaccharides and other substantial pattern in which the singly charged molecular posttranslational modifications, the sequence ion predominated, but lesser amounts of databases are not very useful for estimating dimer, trimer, or doubly charged monomer masses. Values measured by MALDI-TOF/MS often were observed, as shown for the exam- should represent accurate estimates with the α ples of -acid glycoprotein, transferrin, and caveat that there may be some underestima- 1 immunoglobulin G as shown in Fig. 1 panels tion of mass as a result of loss of some groups Volume 2, 2006________________________________________________________________Clinical Proteomics 08_Hortin.qxd 11/21/06 3:20 PM Page 106 106 ________________________________________________________________________Hortin and Remaley Fig.1.Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry of α -acid glycoprotein (A), 1 transferrin (B),and immunoglobulin G (C). Clinical Proteomics________________________________________________________________Volume 2, 2006 08_Hortin.qxd 11/21/06 3:20 PM Page 107 Direct Analysis of Serum Proteins _____________________________________________________________107 any detectable spectra. The high protein con- tent of serum probably did not allow a high enough ratio of matrix to protein to provide for efficient ionization of undiluted specimens. Matrix usually needs to be in more than 1000-fold molar excess relative to protein to yield good spectra (6). Several diluents were examined such as 10 mM ammonium acetate, 10 mM ammonium bicarbonate, normal saline, and other solutions with varying ionic strength. Diluents with an ionic strength above that of normal saline suppressed ionization. The solutions of 10 mM ammonium acetate and 10 mM ammonium bicarbonate were sub- jectively considered to yield similar results with greatest peak intensities (not shown). A solution of 10 mM ammonium acetate was used as diluent for subsequent experiments described here. Dilution of serum with 19 vol of 10 mM ammonium acetate plus 40 vol of matrix solution yielded spectra as shown in Fig. 3. In the top panel, the m/z range from Fig. 2. Matrix-assisted laser desorption/ionization 5000 to 30,000 is shown and, in the bottom time-of-flight mass spectrometry of apolipoprotein panel, m/z from 30,000 to 100,000 is shown. A-I (Top) and transthyretin also known as pre- For m/z > 30,000, singly and doubly charged albumin (Bottom). albumin is the dominant component. A peak corresponding to expected position of transfer- such as sialic acids during desorption and ion- rin generally was visible and a number of ization (6–10). Losses of sialic acid may vary other minor unidentified peaks were observed. depending on technique of analysis, leading If the analysis was extended beyond m/z of to differences in measured mass; as an exam- 150,000, peaks were observed in the expected ple, measurements of the mass of α -acid positions for albumin dimer and IgG. In the 1 glycoprotein by MALDI-TOF/MS vary from m/z range from 5000 to 30,000, there were a about 32,000 to 36,000 in different studies number of sharper peaks, corresponding to the (5,9,21). In general, the ionization efficiency expected positions of apolipoproteins A-I, A-II, appeared to be poorer for high molecular C-I, C-II, C-III, and transthyretin. Although weight and highly glycosylated components. some of these proteins, such as apolipoprotein Haptoglobin and α -macroglobulin were not C-I, C-II, and C-III, are often considered 2 included in Table 1 as they yielded very weak relatively minor serum components, they are and diffuse peaks at m/z of about 96,000 and among the 20 most abundant components on a 124,000 for haptoglobin (mixed types) and molar basis (20). In general, ion intensities 178,000 for α -macroglobulin. forindividual proteins in MALDI-TOF/MS 2 Serum required an initial dilution by a factor should relate to the molar concentration of a of about fivefold with aqueous diluent before protein, and there is a strong bias for detection further threefold dilution with matrix to obtain of low molecular weight proteins (20,21). Volume 2, 2006________________________________________________________________Clinical Proteomics 08_Hortin.qxd 11/21/06 3:20 PM Page 108 asey 6802 47028702520209023400 6602 6502 5602 5402 br 7/ 6/7/6/0/6/ 7/ 7/ 6/ 6/ ataent P02/15 P02/15P02/15P19/15P01/15001/31 P02/15 P02/15 P02/15 P02/15 D 6 6 6 6 6P0 6 6 6 6 N1 al r, n ported Values aMass Peptide modifications.ercentage carbohydrate(CHO) None or cysteinylated,0% CHO3% CHO see None (text for possible)6% CHO 45% CHO 12% CHO N-terminal pyroGlumonomer ordisulfide-linked dime0% CHO None or cysteinylated, 0% CHO Phosphorylation X 2, 13% CHO No CHO orGalNAcGal* orGalNAcGalSAorGalNAcGalSA28% CHO None or loss of N-termiThrPro, 0% CHO ee P Rd vs pti ble 1d by MALDI-TOF/MS ques and Calculated Pe Calculated peptide mass/length in amino acids 6,438/585 aa or 66,557/586 aaNot calculated variable sequence 8,079/243 aa 5,144/679 aa 1,631/183 aa 4,325/394 aa 690.9 (monomer)/77 aa 17,379.8 (dimer)/2 X 77 aa 3,761.5 (subunit)/127 aa 2,944/A-chain 282 aa +B-chain 27 aa764.7 or 9130.0 or9421.3 or 9712.6/each 79 aa Not calculated, variable sequence630.6 or 6432.4/57or 55 aa Tareni 6 2 7 2 4 8 1 3 8 6 uh sc ae Masses Menalytical T Reportedmass (Ref) (38)6,000 (38)0,000 (39)8,016 (38)7,000 (38)0,000 (38)5,000 (39)7,414 4,400 4 X(38)3,600 (40)0,000 (39)8800 (38)0,000 (39)6630 A 6 6 2 7 4 5 1 51 5 7 ein al 1 1 tn oo Prti of adi D 00 80 8 34 50 30 312 64 30 33 40 11 n Tr S 2 1 1 3 1 1 13 risofor pa d Com Measure(m/z) 66,400 148,420 28,077 79,110 35,290 50,300 869617,393 13,82613,862 42,420 91329427 160,820 66306431 I I n II A- )ei C- st Protein Albumin IgG Apolipoprotein Transferrin α-Acid 1glycoproteinα-Proteinase 1inhibitorApolipoprotein A-II Transthyretin (prealbumin)(four subunitαHS-Glyco-pro2 Apolipoprotein IgA Apolipoprotein C-I 108 08_Hortin.qxd 11/21/06 3:20 PM Page 109 s 4212925282729252124003 e 7010405000404030408030 c 7/0/7/6/0/7/6/7/7/8/2/ n P026/15P016/15P026/15P0212/8P016/15P006/15P026/15P026/15P026/15P019/15NP001/17 refere e h t m, ns, om 4% CHO 27% CHO 19% CHO None, 0% CHO 15% CHO Phosphorylation X 1, 3% CHOSequence polymorphis0% CHONone or truncated chai0% CHON-terminal pyroGlu,0% CHONone, 0% CHO None, 0% CHO hydrate contents were fr o b r a 6 (E2)9 aa 6 aa ated cntries. 89 0 me 6/458 aa 6/400 aa 3/326 aa 0/79 aa 3/432 aa 6/791 aa 7 (E3) or 34,134,288 (E4)/22.8/104 aa or526.6/103 aa9 (subunit)/2 9.2/99 aa 2.7/130 aa ngle run. Estiated database alic acid. 51,21 45,26 36,23 8915. 49,03 88,38 34,23or 11,6811,23,02 11,72 14,69 n a siindicnd si withid on se, a (41)51,000 (41)68,000 (41)40,000 (39)8900 (41)65,000 (41)81,000 (39)34,145 (37)11,685 118,000 5 X (38)23,600(38)11,800 (38)14,500 asurements es were basemine, galacto meassosa 160 510 160 23270 500 170 1115 4 7 d by 10 ptide mylgalact eet npe mid ac 250 330 940 916342150 100 420 683528006 727 690 deterculateate N- 51, 52, 44, 810,57, 90, 34, 11,11,23, 11, 14, D) was ses. CalAindic Gc globulin α-Antichymo-1trypsinβ-Glyco-protein I2 Apolipoprotein C-II Antithrombin III Plasminogen Apolipoprotein E Serum amyloid A C-reactive protein (X 5 subunits)β-microglobulin2 Lysozyme aThe imprecision (Scited for reported masGalNAc, Gal, and S 109 08_Hortin.qxd 11/21/06 3:20 PM Page 110 110 ________________________________________________________________________Hortin and Remaley Fig. 3. Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry of diluted serum. (Top) m/z range from 5000 to 30,000 with tentative identification of major peaks. (Bottom) m/z range from 30,000 to 100,000. Fig. 4. Peak intensities (peak heights) with varying Apeak corresponding to the expected position dilution of serum.Listed dilutions reflect dilution with = (m/z 6630) of apolipoprotein C-I was one of 10 mM ammonium acetate prior to further three- the strongest peaks, suggesting unusually effi- fold dilution with matrix solution. cient ionization of this protein, whereas the ionization of apolipoprotein A-II appeared owing to their much lower abundance relative relatively inefficient considering its higher to proteins. abundance relative to apolipoproteins C-I, C-II, Most peaks in the spectra appeared to and C-III. Different peptides and proteins are decrease in signal intensity with increasing recognized to have highly variable ionization dilution with 10 mM ammonium acetate as in MALDI-TOF/MS with content of arginine shown in Fig. 4 (prior to further threefold residues a significant factor in ionization of dilution with matrix solution). Most compo- peptides (22). Although there are reported to nents had decreasing intensities for dilutions be a large number of small peptide compo- from about 10- to 300-fold; however, peaks nents in serum (23–26), components with related to albumin increased in intensity over m/z < 5000 were not observed clearly under this range of dilutions and began to decrease the conditions of analysis used here and this in intensity only at dilutions more than 100- region of the spectrum is not shown. Detection fold. Apparently the concentrations of albumin of small peptides probably is suppressed were saturating at low dilutions. Absolute Clinical Proteomics________________________________________________________________Volume 2, 2006 08_Hortin.qxd 11/21/06 3:20 PM Page 111 Direct Analysis of Serum Proteins _____________________________________________________________111 Table 2 Major Peaks in MALDI-TOF Mass Spectrometric Analysis of Two Serum Specimens and Tentative Assignmentsa Measured m/z Assignment Expected mass High m/z region (>30,000) 147,860 ±490/147,170 ±700 [IgG +H]+ ~148,000 133,320 ±410/132,960 ±270 [Albumin +H]+ 132,855 2 79,180 ±240/79,360 ±310 [Transferrin +H]+ 79,110 66,540 ±90/66,550 ±140 [Albumin +H]+ 66,428 33,290 ±130/33,240 ±170 [Albumin +2H]+2 33,216 Low m/z region (5000–30,000) 28,015 ±15/28,038 ±12 [Apo A-I +H]+ 28,078 13,876 ±2/13,868 ±1 [Prealbumin +Cys +H] 13,946 13,756 ±2/13,747 ±2 [Prealbumin +H]+ 13,827 9424 ±1/9417 ±3 [Apo C-III +H]+ 9422.3 1 8914 ±1/8911 ±1 [Apo C-II +H]+ 8917 8809 ±1/8806 ±1 [Apo A-II +Cys +H]+ 8811 8686 ±3/8680 ±1 [Apo A-II (monomer) +H]+ 8692 6630 ±1/6631 ±1 [Apo C-I +H]+ 6631 6433 ±1/6432 ±1 [Apo C-I (–ThrPro-) +H]+ 6433 aValues for m/z are means ±SD of six measurements. intensities of signals in MALDI-TOF/MS usu- components. Data for repeated analysis of two ally are subject to substantial variation from specimens are shown in Table 2. Most peaks one laser shot to the next. The fairly dense corresponded to expected values within the application of matrix with serum in this study, precision of measurement. An exception was however, yielded fairly consistent signals, and transthyretin, which is known to be polymor- variation was reduced somewhat by summing phic and to undergo a variety of modifications signals from 300 laser shots for each analysis. that may affect it mass (27,28). The mass mea- Because peaks were compared within the surements indicated that it was possible to same analysis, other peaks within an analysis obtain accurate mass measurements of several served as controls for consistency of the results. of the most abundant components in the com- Comparison of peak ratios between compo- plex mixture of serum proteins without exten- nents might be considered a more valid sive fractionation or removal of salts. For the approach for this analytical technique, but it smaller proteins, the pairing of sodium ions or leads to similar results. These dilution studies matrix molecules with the protein would have led to a conclusion that there was an optimal been apparent in mass measurements. dilution for analysis of most components other Discussion than albumin that was about a 10- or 20-fold dilution with ammonium acetate prior to Significant information about molecular further dilution with matrix. masses and heterogeneity of concentrated The measured m/z for a number of peaks solutions of purified proteins or even the in serum samples appeared to correspond to complex mixture represented by serum was the expected values for several major serum obtained by MALDI-TOF/MS simply by Volume 2, 2006________________________________________________________________Clinical Proteomics 08_Hortin.qxd 11/21/06 3:20 PM Page 112 112 ________________________________________________________________________Hortin and Remaley dilution of specimens together with matrix helpful in examining other protein compo- and drying on a target plate. Measured mass nents in this region. This may be a practical values in Table 1 serve as a database for the procedure considering that less than 1 µL of expected positions of peaks for a number serum is required per analysis. For ions with of serum components when analyzed by m/z < 30,000, only several of the most abun- MALDI-TOF/MS. These data may be of value dant components are observed. Specimen frac- in helping to identify which proteins are form- tionation would be required to see lower ing peaks and in identifying peaks that may abundance components. Consequences of the be useful for internal calibration of masses general suppression of ionization by other during an analyses. Of course, masses of some ions and proteins are relatively weak signals, proteins may vary owing to sequence poly- detection only of high abundance compoents, morphism, variable posttranslational modifi- and the need to spend more time with data cation, or degradation during processing or acquisition to sum data from a large number storage. Measured masses of highly sialylated of laser shots. Use of relatively high laser proteins by MALDI-TOF/MS also may vary power may be necessary to obtain measurable depending on several variables, as previously peaks, and this may lower resolution and pro- noted. Also, it must be considered that there mote fragmentation of proteins. Low peak is structural variation in glycosylation of pro- intensities and resolution also can be a limit- teins in different physiological states that may ing factor in obtaining precise and accurate lead to changes in glycoprotein mass (27,28). mass measurements. Direct analysis of diluted serum specimens The method of sample preparation by MALDI-TOF/MS may offer a rapid and described here offers a simple approach to inexpensive approach to analyze several major analyze a number of major serum compo- serum components with minimal opportunities nents. In particular, it appears best suited to for modification or selective loss of compo- analyze components with a mass less than nents. For proteins lacking labile posttransla- 30,000 where there is better sensitivity, resolu- tional modifications, the mass measurements tion, and less interference from albumin. This determined by MALDI-TOF/MS should be technique would be useful for identifying accurate measures of protein mass. For several polymorphisms or modifications of compo- of the most abundant proteins with masses less nents, such as by cysteinylation in the exam- than 30,000, mass measurements can be made ple of transthyretin (29,30). A significant with sufficient precision and accuracy to allow portion of apolipoprotein A-I, appeared as a detection of most chemical modifications of hump following the major expected peak, sug- proteins or sequence polymorphisms. gesting a modification increasing its mass by Two drawbacks of the analysis of a complex about 150 Da, which merits further investiga- mixture, such as serum, without fractionation tion; this protein may undergo fatty acylation are the detection of only a few of the most and is susceptible to nitration and chlorination abundant components and the general sup- (31,32). Alternatively, this may represent adduct pression of signals by the high protein and salt formation with a matrix molecule (6). Variants concentrations. Dominance of the albumin of a number of other apolipoproteins have peaks for monomer, doubly charged monomer, been described including a truncated form of and singly charged monomer interfere with apolipoprotein C-I, which has lost two amino the ability to detect other peaks with m/z > acids from its N-terminus (33,34). The latter 30,000. Selective immunodepletion of albumin truncated form of apolipoprotein C-I was or albumin plus immunoglobulin G may be observed consistently in our analyses. Clinical Proteomics________________________________________________________________Volume 2, 2006
Description: