- Email: [email protected]
ionization time-of-flight mass spectrometry-based amino acid analysis
ANALYTICAL BIOCHEMISTRY Analytical Biochemistry 335 (2004) 184–191 www.elsevier.com/locate/yabio
Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry-based amino acid analysis Michail A. Alterman*, Natalya V. Gogichayeva, Boris A. Kornilayev Biochemical Research Service Laboratory and Analytical Proteomics Laboratory, University of Kansas, Lawrence, KS 66045, USA Received 12 May 2004
Abstract Amino acid analysis has been an integral part of analytical biochemistry for more than 50 years. However, its experimental design, which includes derivatization of amino acids followed by some kind of chromatographic separation, has not changed over the years. We have developed a matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF)-based method for the quantitative analysis of amino acids. This method does not require any amino acid modification, derivatization, or chromatographic separation. The data acquisition time is decreased to several seconds for a single sample. No significant ion suppression effects were observed with the developed sample deposition technique, and the method was found to be reproducible. Linear responses between the amino acid concentration and the peak intensities ratio of corresponding amino acid to internal standard were observed for all amino acids analyzed in the range of concentrations from 20 to 300 lM, and correlation coefficients were between 0.983 (for arginine) and 0.999 (for phenylalanine). Limits of quantitation were between 0.03 lM (for arginine) and 3.7 lM (for histidine and homocysteine). This method was applicable to the mixtures of free amino acids as well as to HCl hydrolysates of proteins. Furthermore, we have shown that this method can be applied to other biologically important low-molecular weight compounds such as glucose. Ó 2004 Elsevier Inc. All rights reserved. Keywords: Amino acid analysis; MALDI-TOF; Homocysteine; Glucose
Amino acid analysis (AAA)1 has been an integral part of analytical biochemistry for more than 50 years [1]. Its area of application extends from the clinical field to food analysis and structural biochemistry [2]. The main technological design of AAA has not changed during the past 50 years or so. It still includes derivatization of amino acids (precolumn or postcolumn) with a UV or fluorescent reagent followed by some kind of chromato*
Corresponding author. Fax: 1-785-864-5396. E-mail address: [email protected] (M.A. Alterman). 1 Abbreviations used: AAA, amino acid analysis; HPLC, highperformance liquid chromatography; GC, gas chromatography; CE, capillary electrophoresis; MS, mass spectrometry; MALDI-TOF, matrix-assisted laser desorption/ionization time-of-flight; CHCA, acyano-4-hydroxycinnamic acid; DHB, 2,5-dihydroxybenzoic acid; F20TTP, meso-tetra (pentafluorophenyl) porphine; BSA, bovine serum albumin; TFA, trifluoroacetic acid; RSD, relative standard deviation. 0003-2697/$ - see front matter Ó 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.ab.2004.06.031
graphic separation—ion exchange or reverse phase highperformance liquid chromatography (HPLC), or gas chromatography (GC)—or capillary electrophoresis (CE) [3–6]. Recent advances in mass spectrometry (MS) led to the application of electrospray ionization coupled with LC or CE for amino acid detection [5,7,8]. In the case of LC–MS, because of limitations in mobile phase composition, normally derivatized amino acids are analyzed, whereas use of CE or LC–MS– MS allows analysis of free underivatized amino acids [9]. On the other hand, application of GC requires modification of amino acids to produce volatile derivatives [10–12]. At its best, the analysis time for a single sample is 15–25 min in the case of GC–MS, CE–MS, or LC–MS and is 30–180 min in the case of HPLC (modification time not included) [13]. Since its introduction during the late 1980s, matrixassisted laser desorption/ionization time-of-flight
M.A. Alterman et al. / Analytical Biochemistry 335 (2004) 184–191
(MALDI-TOF) MS has developed into a major technique for analysis of biological polymers such as peptides, proteins, carbohydrates, and oligonucleotides [14]. However, examples of MALDI-TOF application to the qualitative and quantitative analysis of low-molecular weight compounds, such as amino acids, lipids, catecholamines, bile acids, and fatty acids, are limited [15–19]. To a large degree, this is related to problems associated with heterogeneity of analyte crystallization [20–22] and control of ion suppression effects [23]. In this article, we present a comprehensive MALDITOF-based method for the qualitative and quantitative analysis of underivatized normally occurring protein amino acids and two physiologically important amino acids: homocysteine and c-aminobutyric acid. This method is simple and fast, and it does not require any amino acid modification, derivatization, or chromatographic separation. The data acquisition time is decreased to 1–2 s for a single sample. We have found that this method is applicable to the mixtures of free amino acids as well as to HCl hydrolysates of proteins. Furthermore, we have shown that the technique developed can be successfully applied to other biologically important low-molecular weight compounds such as glucose.
Materials and methods Chemicals The matrices a-cyano-4-hydroxycinnamic acid (CHCA), 2,5-dihydroxybenzoic acid (DHB), and mesotetra (pentafluorophenyl) porphine (F20TTP) were obtained from Aldrich (Milwaukee, WI, USA). All amino acids used in this study were purchased from Sigma (St. Louis, MO, USA). Bovine serum albumin (BSA, 7% solution) was obtained from the National Institute of Standards. Preparation of amino acid standards Amino acid standards at concentrations of 300, 200, 100, 80, 60, and 20 lM were prepared from a 10-mM stock solution in deionized water (Tyr stock solution was 3 mM). Methyltyrosine was used as an internal standard in every experiment (concentration of stock solution was 1 mM). Concentrations of amino acids to determine limits of quantitation were 300, 100, 33.3, 11.1, 3.7, 1.2, 0.4, 0.14, and 0.03 lM. Preparation and gas phase hydrolysis of BSA The sample of BSA (500 pmol) was evaporated under vacuum to complete dryness (vacuum gauge reading 50– 55 millitor on Pico–Tag Workstation, Waters Associ-
185
ates, Milford, MA, USA). Vapor phase hydrolysis of the samples was performed with 6 N HCl containing 1% phenol for 24 h at 105 °C on Pico–Tag Workstation. The hydrolyzed samples were then dried under vacuum and reconstituted in deionized water. Acquisition of MALDI-TOF MS spectra and data analysis All spectra were acquired on an Applied Biosystems 4700 Proteomics Analyzer in a reflectron mode. Laser intensity was 4500, the instrument was operated in a positive ion mode, and the focus mass was 145. To cover as much target area as possible, a spiral search pattern was used with shots at each firing position, 75 firing positions per spot, 450 shots total. Data were collected over the mass window 50–300. Four to six spots were prepared for each sample. Experiments with F20TTP were performed as described in [17]. Dried droplet deposition of amino acid samples was used in initial qualitative experiments with DHB and CHCA (1:1 v/v mixture of sample and matrix solutions). DHB was prepared as 10 mg/ml water solution, and CHCA was prepared as saturated solution in 50% acetonitrile/0.1% trifluoroacetic acid (TFA). Conditions for sample deposition in quantitative experiments were optimized as follows. CHCA was dissolved until saturation in 85% acetonitrile/0.03% TFA, and 1 ll of this solution was spotted on a 192-well sample plate (Applied Biosystems, Foster City, CA, USA) and air-dried. Then 2 ll of internal standard solution was added to 10 ll of amino acid solution (final concentration of internal standard in all experiments was 166.7 lM), and the concentration of amino acids depended on the experimental scope (16.6–250 lM for the quantitative analysis and 0.025–250 lM for the limit of quantitation experiments). Then 1 ll of this mixture was pipetted on the target and air-dried. The peak height (intensity) values were determined by using Data Explorer version 4.5 (Applied Biosystems). The average ratio of the peak intensities was then plotted against the amino acid concentration. All calibration curves were created from the data sets obtained from mass spectra of mixtures containing all amino acids plus internal standard. The mass peak list was exported and processed in Microsoft Excel.
Results The choice of proper matrix is vital in any MALDITOF-based experiment. Correspondingly, at first we evaluated different matrices for their suitability for qualitative AAA. Three different matrices were chosen: CHCA, DHB, and F20TTP. CHCA and DHB are routinely used for acquisition of peptide mass spectra, and
186
M.A. Alterman et al. / Analytical Biochemistry 335 (2004) 184–191
two previous reports described AAA of limited numbers of amino acids in these matrices [16,19]. A substituted porphyrin (F20TTP) was selected on the basis of a previous publication showing its use for qualitative and quantitative analysis of fatty acids [17]. In the cases of DHB and F20TTP, all amino acids loaded individually were detected in positive reflector mode with no interference with the matrix ion peaks. However, when a mixture of 19 amino acids was analyzed, only 13 amino acids were observed in F20TTP and 14 were observed in DHB (Fig. 1). Additional difficulties in using F20TTP as a matrix involved the use of sodium acetate as a dopant to obtain MALDI-TOF signals of amino acids [17]. On the other hand, all 19 amino acids were ionized and detected with CHCA as a matrix (Fig.
2A). As can be seen from the inset in Fig. 2, glutamine and lysine with a difference in molecular mass of 0.036 atomic mass units are well resolved. Our initial experiments in quantitative MALDITOF-based AAA revealed that dried-droplet sample deposition did not produce consistent quantitative results. Dally and colleagues [19] recently reported quantitation of 12 amino acids using an aerospray deposition of sample/matrix (CHCA) mixture. However, this technique is relatively time-consuming (1 min/single spot deposition time) and technically not well suited for high-throughput analysis. After numerous experiments, we have found that a preformed layer technique with 85% acetonitrile/0.03% TFA provides consistent quantitative results. Fig. 3 shows some representative calibra-
Fig. 1. Positive ion MALDI-TOF MS spectrum of amino acid standard with F20TTP as a matrix (A) and with DHB as a matrix (B). Asterisks denote matrix ion peaks.
M.A. Alterman et al. / Analytical Biochemistry 335 (2004) 184–191
187
Fig. 2. Positive ion MALDI-TOF MS spectrum of 21-amino acid standard mixture with CHCA as a matrix. (A) A final concentration of amino acids of 300 lM (250 pmol of each amino acid on target). (B) A final concentration of amino acids of 20 lM (16.7 pmol of each amino acid on target). The relative peak intensities were not affected by amino acid concentration. The inset provides an expanded view showing resolution attained and separation of Lys (147.18) and Gln (147.14) peaks. Asterisks denote matrix ion peaks.
tion curves for several normally occurring amino acids and two physiologically important amino acids. The internal standard (methyltyrosine) was at a constant concentration of 166.7 lM. Each data point represents the average peak intensity of four to six experiments. Linear responses between the amino acid concentration and the peak intensities ratio of corresponding amino acid to internal standard were observed for all amino acids analyzed, and correlation coefficients were between 0.983 (arginine) and 0.999 (phenylalanine) (Table 1). Limits of quantitation were between 0.03 lM (arginine) and 3.7 lM (histidine and homocysteine) with a signal-to-noise (S/N) ratio of 10:1. Four pairs of amino acids (Ile/Leu vs. Asn, Asn vs. Asp, Gln vs. Glu, and
Lys vs. Glu) potentially have an overlap of isotopomers and might require an isotopic correction to be taken into account in calculating the peak intensities used in calibration curves. However, in our case, such correction was unnecessary because the a + 1 isotopomers of the lower mass were always resolved from a isotopomers of higher mass, with Dm > 0.03 being reliably resolved (cf. inset in Fig. 2). Remarkably, the relative amino acid peak intensities did not change in the range of concentrations that we used in our study (20–300 lM). Comparison of Figs. 2A and B shows that a 15-fold increase in the amount of amino acids loaded on the target essentially did not change the relative responses of amino acids. To determine the reproducibility of the
188
M.A. Alterman et al. / Analytical Biochemistry 335 (2004) 184–191
Fig. 3. Representative calibration curves of some amino acids. Each data point represents the mean ± standard deviation of data collected in four to six experiments. The observed linear range was from 20 to 300 lM. IS, internal standard.
method, the same mixture of amino acids was examined three times over the period of 8 days. The relative standard deviations (RSDs) for the majority of amino acids were well under 10%, with proline and arginine being exceptions with RSDs of approximately 12% (Table 2). The applicability and accuracy of the developed method for analyzing unknown samples was evaluated by examination of two samples: a solution of 10 different free amino acids and an HCl hydrolysate of BSA. Table 3 shows comparisons of measured concentrations with
actual concentrations. The data obtained were in good agreement with actual values, with percentage error values of less than 10%. Next, we analyzed an HCl hydrolysate of BSA. Table 4 presents a comparison of measured molar percentages of amino acids with calculated molar percentages based on the amino acid sequence. The measured data for some amino acids (e.g., Ala, Thr, the sum of Leu and Ile) show an excellent fit to the calculated values of molar percentage. The same refers to the Asp–Asn and
M.A. Alterman et al. / Analytical Biochemistry 335 (2004) 184–191
Glu–Gln pairs, where amides are hydrolyzed to the corresponding acids. The data for Gly, Met, and Phe exhibit the highest molar percentage errors. However, it should be emphasized that Gly and Met were the two problematic amino acids with the highest average analysis errors (> 50%) in two collaborative studies of the AAA performed by the Association of Biomolecular Resource Facilities in 1997 and 2001 (www.abrf.org/ index.cfm/group.show/aminoacidanalysis.19.htm). Overall, the quality of the data was on the same level as, or even better than, that of the established methods. Finally, the measured amino acid molar percentages were used to search SwissProt with AACompIdent, a protein amino acid composition search engine. In this case, when the species name was entered into the search engine, BSA was identified as the first-ranked protein with a score of 38, whereas the second-ranked protein had a score of 64. When no species name was given, human serum albumin was ranked first with a score of 35, whereas BSA was ranked second with a score of 38.
Table 1 Comparison of correlation coefficients and limits of quantitation Amino acid
R2
Limit of quantitation (lM)
Glycine Alanine c-Aminobutyric acid Serine Proline Valine Threonine Leucine/Isoleucine Asparagine Aspartic acid Homocysteine Glutamine Glutamic acid Lysine Methionine Histidine Phenylalanine Arginine Tyrosine Tryptophan
0.990 0.990 0.996 0.994 0.996 0.987 0.998 0.991 0.995 0.988 0.997 0.991 0.994 0.985 0.994 0.994 0.999 0.983 0.995 0.996
0.14 0.14 0.14 0.14 0.14 0.14 0.40 0.40 0.14 0.40 3.70 0.40 0.40 0.40 1.20 3.70 0.14 0.03 0.40 1.20
189
Table 2 Day-to-day reproducibility of peak intensity ratios of 100-lM solution of amino acids Amino acid
Glycine Alanine c-Aminobutyric acid Serine Proline Valine Threonine Asparagine Aspartic acid Homocysteine Glutamine Glutamic acid Methionine Histidine Phenylalanine Arginine Tyrosine Tryptophan
Peak intensity ratio (amino acid/internal standard)
Percentage RSD
Day 1
Day 5
Day 8
0.27 ± 0.05 0.38 ± 0.04 0.90 ± 0.05 0.42 ± 0.06 1.38 ± 0.07 0.43 ± 0.05 0.45 ± 0.07 0.60 ± 0.07 0.26 ± 0.03 0.20 ± 0.06 1.07 ± 0.17 0.10 ± 0.02 0.59 ± 0.05 1.33 ± 0.08 0.61 ± 0.05 2.31 ± 0.09 0.08 ± 0.01 0.66 ± 0.05
0.24 ± 0.06 0.34 ± 0.08 0.76 ± 0.16 0.33 ± 0.07 1.10 ± 0.14 0.34 ± 0.05 0.49 ± 0.07 0.63 ± 0.13 0.26 ± 0.04 0.21 ± 0.07 1.10 ± 0.12 0.08 ± 0.01 0.45 ± 0.16 1.43 ± 0.22 0.52 ± 0.10 2.59 ± 0.39 0.09 ± 0.02 0.52 ± 0.1
0.26 ± 0.05 0.40 ± 0.04 0.82 ± 0.04 0.40 ± 0.06 1.14 ± 0.07 0.43 ± 0.05 0.49 ± 0.06 0.63 ± 0.06 0.25 ± 0.04 0.23 ± 0.05 1.02 ± 0.07 — 0.56 ± 0.06 1.34 ± 0.26 0.63 ± 0.06 2.36 ± 0.73 — 0.52 ± 0.09
1.25 2.49 8.22 3.86 12.36 4.24 1.89 1.41 0.47 1.25 3.30 1.00 6.02 4.50 4.97 12.19 0.85 6.62
Table 3 Comparison of actual and measured concentrations of amino acids Amino acid
Actual concentration (lM)
Measured concentration (lM)
Error (percentage)
Alanine c-Aminobutyric acid Asparagine Glycine Homocysteine Methionine Phenylalanine Threonine Tryptophan Valine
180 180 180 260 90 260 90 180 90 180
181.0 ± 17.9 171.3 ± 10.2 187.3 ± 14.3 262.0 ± 28.1 98.0 ± 6.5 269.0 ± 20.9 98.0 ± 5.8 184.0 ± 6.7 98.5 ± 9.2 183.0 ± 14.1
0.56 5.16 4.05 0.77 8.90 3.46 8.90 2.20 9.44 1.66
190
M.A. Alterman et al. / Analytical Biochemistry 335 (2004) 184–191
Table 4 Comparison of theoretical (calculated based on amino acid sequence) and measured molar percentages of amino acids in BSA HCl hydrolysate Amino acid
Measured molar percentage
Calculated molar percentage
Glycine Alanine Serine Proline Valine Threonine Leucine + isoleucine Asparagine Aspartic acid Glutamine Glutamic acid Lysine Methionine Histidine Phenylalanine Arginine Tyrosine
1.1 8.2 3.7 5.6 8.0 6.3 12.0 1.8 6.8 0.9 13.3 8.4 1.5 3.6 7.9 5.7 4.6
2.8 8.1 4.8 4.8 6.8 6.3 12.9 2.4 6.9 3.4 10.6 9.9 0.9 3.1 5.4 4.7 3.8
Note. The values reported are the averages of data collected in five measurements.
The potential of the developed application of MALDI-TOF MS for the analysis of small biologically and physiologically important molecules was further proven by application of this technique to the analysis of glucose concentrations. Over the range of concentrations between 2 and 20 mM (i.e., physiological range of concentrations), the calibration curve was linear with a correlation coefficient of 0.988 and methyltyrosine was used as an internal standard (data not shown).
Discussion There were two previous reports describing analysis of a limited number of amino acids by MALDI-TOF MS. Wittman and Heinzle applied MALDI-TOF MS for the quantitation of lysine, alanine, and glucose in a microbial cultivation [16]. They used DHB as a matrix and stable isotope-labeled amino acids as internal standards. In spite of excellent analytical results for these three compounds, this method presents a number of analytical challenges if one would attempt to use it as a comprehensive AAA method. First, as we have shown, DHB as a matrix does not induce proper ionization of many amino acids in mixtures. Second, having an individual internal standard for each amino acid would greatly complicate the mass spectrum and could cause additional overlap of isotopomers. Further difficulty arises from the fact that all three compounds formed sodium and potassium ions in addition to the main protonated ion, and the ratio of the main protonated ion-to-sodium and potassium adduct ions depended on
sample and analysis conditions. Additional consideration in calculations required the presence of nonlabeled impurities in internal standards. In general, this method would be difficult to apply to complex mixtures of amino acids. Dally and colleagues published a study describing the MALDI-TOF MS method for AAA in cell culture media [19]. These authors developed calibration curves for 12 amino acids over the linear range of 1–100 lM. The main analytical difficulty of this study was the use of matrix ion as an internal standard. The matrix used was CHCA. It is well established that matrix ions are subject to ion suppression effects and could even be fully suppressed by analyte ions, not to mention that they are dependent on laser power, have evaporation rates different from those of amino acids, and so on [24]. Use of a protonated matrix ion as an internal standard requires relative responses of all amino acids analyzed to be constant and independent of analyte concentration. However, that was not the case in the study published by Dally and colleagues, as is evident from the MS data they reported [19]. Analysis of the mass spectra presented in that study reveals that the Pro/Val peak intensity ratio increased 1.9-fold, whereas the Val/Thr ratio decreased 1.8-fold, when the amino acid concentration increased 10-fold (from 10 to 100 lM). Relative responses of some other amino acids were changed as well. Furthermore, despite the fact that these authors analyzed an amino acid standard mixture containing 19 amino acids, they were able to develop calibration curves for only 12 amino acids and did not present any explanation as to what happened to the remaining 7 amino acids. On the other hand, the method presented in the current article allows fast and straightforward qualitative and quantitative analysis of amino acids by MALDITOF MS. The applicability of this method for the analysis of mixtures of free amino acids and HCl hydrolysates of proteins was demonstrated. No significant ion suppression effects were observed with the developed sample deposition technique, and the method was found to be reproducible. It should be recognized that this method has one inherent limitation: the analysis of isomeric amino acids (e.g., leucine, isoleucine). One way in which to get around this problem is to use MALDITOF MS–MS. Currently, experiments exploring application of MALDI-TOF MS–MS for AAA are under way in our laboratory. The most significant advantage of the presented method over conventional methods of AAA is incomparably shorter analysis time due to the absence of modification and separation steps. Possibilities for automation of the sample preparation and data processing can further advance this method. An additional area of application for this technique could be in clinical chemistry, as was shown by the data obtained with homocysteine, c-aminobutyric acid, and glucose. Normal values of plasma amino acids are well within
M.A. Alterman et al. / Analytical Biochemistry 335 (2004) 184–191
the quantitation limits and linearity intervals for corresponding amino acids determined in our study. The lowest plasma amino acid reference interval exhibits Asp (0–19 lM), whereas the highest one exhibits Gln (350–700 lM). All others demonstrate the normal range between 15 and 400 lM. Yet, for the MALDI-TOF MS AAA method to be used in clinical laboratories, a careful evaluation of the biological sample preparation technique (deproteinization by sulfosalicylic acid, use of acetonitrile, or ultrafiltration) must be performed to determine the levels of potential ion suppression compounds in the sample solution.
Acknowledgments We thank T. Williams for helpful scientific discussion and critical reading of the manuscript. Natalya Gogichayeva and Boris Kornilayev were supported by a research development grant from the University of Kansas (KUCR 00254).
References [1] S. Moore, W.H. Stein, Chromatography of amino acids on sulfonated polystyrene resins, J. Biol. Chem. 192 (1951) 663–681. [2] C. Cooper, N. Packer, K. Williams, Amino Acid Analysis Protocols, Humana Press, Totowa, NJ, 2001. [3] P. Furst, L. Pollack, T. Graser, H. Godel, P. Stehle, Appraisal of four pre-column derivatization methods for the high-performance liquid chromatographic determination of free amino acids in biological materials, J. Chromatogr. 499 (1990) 557–569. [4] V. Fierabracci, P. Masiello, M. Novelli, E. Bergamini, Application of amino acid analysis by high-performance liquid chromatography with phenyl isothiocyanate derivatization to the rapid determination of free amino acids in biological samples, J. Chromatogr. 570 (1991) 285–291. [5] M. Duncan, A. Poljak, Amino acid analysis of peptides and proteins on the femtomole scale by gas chromatography/mass spectrometry, Anal. Chem. 70 (1998) 890–896. [6] E. Zahou, H. Jornvall, T. Bergman, Amino acid analysis by capillary electrophoresis after phenylthiocarbamylation, Anal. Biochem. 281 (2000) 115–122. [7] T. Soga, D. Heiger, Amino acid analysis by capillary electrophoresis electrospray ionization mass spectrometry, Anal. Chem. 72 (2000) 1236–1241. [8] R. van Leuken, A. Duchateau, G. Kwakkenbos, Thermospray liquid chromatography/mass spectrometry study of diastereomeric isoindole derivatives of amino acids and amino acid amides, J. Pharmacol. Biomed. Anal. 13 (1995) 1459–1464.
191
[9] M. Piraud, C. Vianey-Saban, K. Petritis, C. Elfakir, J. Steghens, A. Morla, D. Bouchu, ESI–MS/MS analysis of underivatised amino acids: a new tool for the diagnosis of inherited disorders of amino acid metabolism—Fragmentation study of 79 molecules of biological interest in positive and negative ionisation mode, Rapid Commun. Mass Spectr. 17 (2003) 1297–1311. [10] A. Pons, C. Richet, C. Robbe, A. Herrmann, P. Timmerman, G. Huet, Y. Leroy, I. Carlstedt, C. Capon, J. Zanetta, Sequential GC/MS analysis of sialic acids, monosaccharides, and amino acids of glycoproteins on a single sample as heptafluorobutyrate derivatives, Biochemistry 42 (2003) 8342–8353. [11] S. Villas-Boas, D. Delicado, M. Akesson, J. Nielsen, Simultaneous analysis of amino and nonamino organic acids as methyl chloroformate derivatives using gas chromatography–mass spectrometry, Anal. Biochem. 322 (2003) 134–138. [12] P. Husek, Chloroformates in gas chromatography as general purpose derivatizing agents, J. Chromatogr. B Biomed. Sci. Appl. 717 (1998) 57–91. [13] P. Husek, P. Simek, Advances in amino acid analysis, LC–GC N. Am. 19 (2001) 986–999. [14] T. Bonk, A. Humeny, MALDI-TOF-MS analysis of protein and DNA, Neuroscientist 7 (2001) 6–12. [15] D. Mims, D. Hercules, Quantification of bile acids directly from urine by MALDI-TOF-MS, Anal. Bioanal. Chem. 375 (2003) 609–616. [16] C. Wittmann, E. Heinzle, MALDI-TOF MS for quantification of substrates and products in cultivations of Corynebacterium glutamicum, Biotechnol. Bioeng. 72 (2001) 642–647. [17] F. Ayorinde, K. Garvin, K. Saeed, Determination of the fatty acid composition of saponified vegetable oils using matrix-assisted laser desorption/ionization time-of-flight mass spectrometry, Rapid Commun. Mass Spectr. 14 (2000) 608–615. [18] M. Bucknall, K. Fung, M. Duncan, Practical quantitative biomedical applications of MALDI-TOF mass spectrometry, J. Am. Soc. Mass Spectr. 13 (2002) 1015–1027. [19] J. Dally, J. Gorniak, R. Bowie, C. Bentzley, Quantitation of underivatized free amino acids in mammalian cell culture media using matrix assisted laser desorption/ionization time-of-flight mass spectrometry, Anal. Chem. 75 (2003) 5046–5053. [20] R. Garden, J. Sweedler, Heterogeneity within MALDI samples as revealed by mass spectrometric imaging, Anal. Chem. 72 (2000) 30–36. [21] S. Cohen, B. Chait, Influence of matrix solution conditions on the MALDI-MS analysis of peptides and proteins, Anal. Chem. 68 (1996) 31–37. [22] I. Figueroa, O. Torres, D. Russell, Effects of the water content in the sample preparation for MALDI on the mass spectra, Anal. Chem. 70 (1998) 4527–4533. [23] R. Knochenmuss, A quantitative model of ultraviolet matrixassisted laser desorption/ionization including analyte ion generation, Anal. Chem. 75 (2003) 2199–2207. [24] R. Knochenmuss, F. Dubois, M. Dale, R. Zenobi, The matrix suppression effect and ionization mechanisms in matrix-assisted laser desorption/ionization, Rapid Commun. Mass Spectr. 10 (1999) 871–877.
Matrix-assisted laser desorption/ionization time-of-flight mass spectrometry-based amino acid analysis Michail A. Alterman*, Natalya V. Gogichayeva, Boris A. Kornilayev Biochemical Research Service Laboratory and Analytical Proteomics Laboratory, University of Kansas, Lawrence, KS 66045, USA Received 12 May 2004
Abstract Amino acid analysis has been an integral part of analytical biochemistry for more than 50 years. However, its experimental design, which includes derivatization of amino acids followed by some kind of chromatographic separation, has not changed over the years. We have developed a matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF)-based method for the quantitative analysis of amino acids. This method does not require any amino acid modification, derivatization, or chromatographic separation. The data acquisition time is decreased to several seconds for a single sample. No significant ion suppression effects were observed with the developed sample deposition technique, and the method was found to be reproducible. Linear responses between the amino acid concentration and the peak intensities ratio of corresponding amino acid to internal standard were observed for all amino acids analyzed in the range of concentrations from 20 to 300 lM, and correlation coefficients were between 0.983 (for arginine) and 0.999 (for phenylalanine). Limits of quantitation were between 0.03 lM (for arginine) and 3.7 lM (for histidine and homocysteine). This method was applicable to the mixtures of free amino acids as well as to HCl hydrolysates of proteins. Furthermore, we have shown that this method can be applied to other biologically important low-molecular weight compounds such as glucose. Ó 2004 Elsevier Inc. All rights reserved. Keywords: Amino acid analysis; MALDI-TOF; Homocysteine; Glucose
Amino acid analysis (AAA)1 has been an integral part of analytical biochemistry for more than 50 years [1]. Its area of application extends from the clinical field to food analysis and structural biochemistry [2]. The main technological design of AAA has not changed during the past 50 years or so. It still includes derivatization of amino acids (precolumn or postcolumn) with a UV or fluorescent reagent followed by some kind of chromato*
Corresponding author. Fax: 1-785-864-5396. E-mail address: [email protected] (M.A. Alterman). 1 Abbreviations used: AAA, amino acid analysis; HPLC, highperformance liquid chromatography; GC, gas chromatography; CE, capillary electrophoresis; MS, mass spectrometry; MALDI-TOF, matrix-assisted laser desorption/ionization time-of-flight; CHCA, acyano-4-hydroxycinnamic acid; DHB, 2,5-dihydroxybenzoic acid; F20TTP, meso-tetra (pentafluorophenyl) porphine; BSA, bovine serum albumin; TFA, trifluoroacetic acid; RSD, relative standard deviation. 0003-2697/$ - see front matter Ó 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.ab.2004.06.031
graphic separation—ion exchange or reverse phase highperformance liquid chromatography (HPLC), or gas chromatography (GC)—or capillary electrophoresis (CE) [3–6]. Recent advances in mass spectrometry (MS) led to the application of electrospray ionization coupled with LC or CE for amino acid detection [5,7,8]. In the case of LC–MS, because of limitations in mobile phase composition, normally derivatized amino acids are analyzed, whereas use of CE or LC–MS– MS allows analysis of free underivatized amino acids [9]. On the other hand, application of GC requires modification of amino acids to produce volatile derivatives [10–12]. At its best, the analysis time for a single sample is 15–25 min in the case of GC–MS, CE–MS, or LC–MS and is 30–180 min in the case of HPLC (modification time not included) [13]. Since its introduction during the late 1980s, matrixassisted laser desorption/ionization time-of-flight
M.A. Alterman et al. / Analytical Biochemistry 335 (2004) 184–191
(MALDI-TOF) MS has developed into a major technique for analysis of biological polymers such as peptides, proteins, carbohydrates, and oligonucleotides [14]. However, examples of MALDI-TOF application to the qualitative and quantitative analysis of low-molecular weight compounds, such as amino acids, lipids, catecholamines, bile acids, and fatty acids, are limited [15–19]. To a large degree, this is related to problems associated with heterogeneity of analyte crystallization [20–22] and control of ion suppression effects [23]. In this article, we present a comprehensive MALDITOF-based method for the qualitative and quantitative analysis of underivatized normally occurring protein amino acids and two physiologically important amino acids: homocysteine and c-aminobutyric acid. This method is simple and fast, and it does not require any amino acid modification, derivatization, or chromatographic separation. The data acquisition time is decreased to 1–2 s for a single sample. We have found that this method is applicable to the mixtures of free amino acids as well as to HCl hydrolysates of proteins. Furthermore, we have shown that the technique developed can be successfully applied to other biologically important low-molecular weight compounds such as glucose.
Materials and methods Chemicals The matrices a-cyano-4-hydroxycinnamic acid (CHCA), 2,5-dihydroxybenzoic acid (DHB), and mesotetra (pentafluorophenyl) porphine (F20TTP) were obtained from Aldrich (Milwaukee, WI, USA). All amino acids used in this study were purchased from Sigma (St. Louis, MO, USA). Bovine serum albumin (BSA, 7% solution) was obtained from the National Institute of Standards. Preparation of amino acid standards Amino acid standards at concentrations of 300, 200, 100, 80, 60, and 20 lM were prepared from a 10-mM stock solution in deionized water (Tyr stock solution was 3 mM). Methyltyrosine was used as an internal standard in every experiment (concentration of stock solution was 1 mM). Concentrations of amino acids to determine limits of quantitation were 300, 100, 33.3, 11.1, 3.7, 1.2, 0.4, 0.14, and 0.03 lM. Preparation and gas phase hydrolysis of BSA The sample of BSA (500 pmol) was evaporated under vacuum to complete dryness (vacuum gauge reading 50– 55 millitor on Pico–Tag Workstation, Waters Associ-
185
ates, Milford, MA, USA). Vapor phase hydrolysis of the samples was performed with 6 N HCl containing 1% phenol for 24 h at 105 °C on Pico–Tag Workstation. The hydrolyzed samples were then dried under vacuum and reconstituted in deionized water. Acquisition of MALDI-TOF MS spectra and data analysis All spectra were acquired on an Applied Biosystems 4700 Proteomics Analyzer in a reflectron mode. Laser intensity was 4500, the instrument was operated in a positive ion mode, and the focus mass was 145. To cover as much target area as possible, a spiral search pattern was used with shots at each firing position, 75 firing positions per spot, 450 shots total. Data were collected over the mass window 50–300. Four to six spots were prepared for each sample. Experiments with F20TTP were performed as described in [17]. Dried droplet deposition of amino acid samples was used in initial qualitative experiments with DHB and CHCA (1:1 v/v mixture of sample and matrix solutions). DHB was prepared as 10 mg/ml water solution, and CHCA was prepared as saturated solution in 50% acetonitrile/0.1% trifluoroacetic acid (TFA). Conditions for sample deposition in quantitative experiments were optimized as follows. CHCA was dissolved until saturation in 85% acetonitrile/0.03% TFA, and 1 ll of this solution was spotted on a 192-well sample plate (Applied Biosystems, Foster City, CA, USA) and air-dried. Then 2 ll of internal standard solution was added to 10 ll of amino acid solution (final concentration of internal standard in all experiments was 166.7 lM), and the concentration of amino acids depended on the experimental scope (16.6–250 lM for the quantitative analysis and 0.025–250 lM for the limit of quantitation experiments). Then 1 ll of this mixture was pipetted on the target and air-dried. The peak height (intensity) values were determined by using Data Explorer version 4.5 (Applied Biosystems). The average ratio of the peak intensities was then plotted against the amino acid concentration. All calibration curves were created from the data sets obtained from mass spectra of mixtures containing all amino acids plus internal standard. The mass peak list was exported and processed in Microsoft Excel.
Results The choice of proper matrix is vital in any MALDITOF-based experiment. Correspondingly, at first we evaluated different matrices for their suitability for qualitative AAA. Three different matrices were chosen: CHCA, DHB, and F20TTP. CHCA and DHB are routinely used for acquisition of peptide mass spectra, and
186
M.A. Alterman et al. / Analytical Biochemistry 335 (2004) 184–191
two previous reports described AAA of limited numbers of amino acids in these matrices [16,19]. A substituted porphyrin (F20TTP) was selected on the basis of a previous publication showing its use for qualitative and quantitative analysis of fatty acids [17]. In the cases of DHB and F20TTP, all amino acids loaded individually were detected in positive reflector mode with no interference with the matrix ion peaks. However, when a mixture of 19 amino acids was analyzed, only 13 amino acids were observed in F20TTP and 14 were observed in DHB (Fig. 1). Additional difficulties in using F20TTP as a matrix involved the use of sodium acetate as a dopant to obtain MALDI-TOF signals of amino acids [17]. On the other hand, all 19 amino acids were ionized and detected with CHCA as a matrix (Fig.
2A). As can be seen from the inset in Fig. 2, glutamine and lysine with a difference in molecular mass of 0.036 atomic mass units are well resolved. Our initial experiments in quantitative MALDITOF-based AAA revealed that dried-droplet sample deposition did not produce consistent quantitative results. Dally and colleagues [19] recently reported quantitation of 12 amino acids using an aerospray deposition of sample/matrix (CHCA) mixture. However, this technique is relatively time-consuming (1 min/single spot deposition time) and technically not well suited for high-throughput analysis. After numerous experiments, we have found that a preformed layer technique with 85% acetonitrile/0.03% TFA provides consistent quantitative results. Fig. 3 shows some representative calibra-
Fig. 1. Positive ion MALDI-TOF MS spectrum of amino acid standard with F20TTP as a matrix (A) and with DHB as a matrix (B). Asterisks denote matrix ion peaks.
M.A. Alterman et al. / Analytical Biochemistry 335 (2004) 184–191
187
Fig. 2. Positive ion MALDI-TOF MS spectrum of 21-amino acid standard mixture with CHCA as a matrix. (A) A final concentration of amino acids of 300 lM (250 pmol of each amino acid on target). (B) A final concentration of amino acids of 20 lM (16.7 pmol of each amino acid on target). The relative peak intensities were not affected by amino acid concentration. The inset provides an expanded view showing resolution attained and separation of Lys (147.18) and Gln (147.14) peaks. Asterisks denote matrix ion peaks.
tion curves for several normally occurring amino acids and two physiologically important amino acids. The internal standard (methyltyrosine) was at a constant concentration of 166.7 lM. Each data point represents the average peak intensity of four to six experiments. Linear responses between the amino acid concentration and the peak intensities ratio of corresponding amino acid to internal standard were observed for all amino acids analyzed, and correlation coefficients were between 0.983 (arginine) and 0.999 (phenylalanine) (Table 1). Limits of quantitation were between 0.03 lM (arginine) and 3.7 lM (histidine and homocysteine) with a signal-to-noise (S/N) ratio of 10:1. Four pairs of amino acids (Ile/Leu vs. Asn, Asn vs. Asp, Gln vs. Glu, and
Lys vs. Glu) potentially have an overlap of isotopomers and might require an isotopic correction to be taken into account in calculating the peak intensities used in calibration curves. However, in our case, such correction was unnecessary because the a + 1 isotopomers of the lower mass were always resolved from a isotopomers of higher mass, with Dm > 0.03 being reliably resolved (cf. inset in Fig. 2). Remarkably, the relative amino acid peak intensities did not change in the range of concentrations that we used in our study (20–300 lM). Comparison of Figs. 2A and B shows that a 15-fold increase in the amount of amino acids loaded on the target essentially did not change the relative responses of amino acids. To determine the reproducibility of the
188
M.A. Alterman et al. / Analytical Biochemistry 335 (2004) 184–191
Fig. 3. Representative calibration curves of some amino acids. Each data point represents the mean ± standard deviation of data collected in four to six experiments. The observed linear range was from 20 to 300 lM. IS, internal standard.
method, the same mixture of amino acids was examined three times over the period of 8 days. The relative standard deviations (RSDs) for the majority of amino acids were well under 10%, with proline and arginine being exceptions with RSDs of approximately 12% (Table 2). The applicability and accuracy of the developed method for analyzing unknown samples was evaluated by examination of two samples: a solution of 10 different free amino acids and an HCl hydrolysate of BSA. Table 3 shows comparisons of measured concentrations with
actual concentrations. The data obtained were in good agreement with actual values, with percentage error values of less than 10%. Next, we analyzed an HCl hydrolysate of BSA. Table 4 presents a comparison of measured molar percentages of amino acids with calculated molar percentages based on the amino acid sequence. The measured data for some amino acids (e.g., Ala, Thr, the sum of Leu and Ile) show an excellent fit to the calculated values of molar percentage. The same refers to the Asp–Asn and
M.A. Alterman et al. / Analytical Biochemistry 335 (2004) 184–191
Glu–Gln pairs, where amides are hydrolyzed to the corresponding acids. The data for Gly, Met, and Phe exhibit the highest molar percentage errors. However, it should be emphasized that Gly and Met were the two problematic amino acids with the highest average analysis errors (> 50%) in two collaborative studies of the AAA performed by the Association of Biomolecular Resource Facilities in 1997 and 2001 (www.abrf.org/ index.cfm/group.show/aminoacidanalysis.19.htm). Overall, the quality of the data was on the same level as, or even better than, that of the established methods. Finally, the measured amino acid molar percentages were used to search SwissProt with AACompIdent, a protein amino acid composition search engine. In this case, when the species name was entered into the search engine, BSA was identified as the first-ranked protein with a score of 38, whereas the second-ranked protein had a score of 64. When no species name was given, human serum albumin was ranked first with a score of 35, whereas BSA was ranked second with a score of 38.
Table 1 Comparison of correlation coefficients and limits of quantitation Amino acid
R2
Limit of quantitation (lM)
Glycine Alanine c-Aminobutyric acid Serine Proline Valine Threonine Leucine/Isoleucine Asparagine Aspartic acid Homocysteine Glutamine Glutamic acid Lysine Methionine Histidine Phenylalanine Arginine Tyrosine Tryptophan
0.990 0.990 0.996 0.994 0.996 0.987 0.998 0.991 0.995 0.988 0.997 0.991 0.994 0.985 0.994 0.994 0.999 0.983 0.995 0.996
0.14 0.14 0.14 0.14 0.14 0.14 0.40 0.40 0.14 0.40 3.70 0.40 0.40 0.40 1.20 3.70 0.14 0.03 0.40 1.20
189
Table 2 Day-to-day reproducibility of peak intensity ratios of 100-lM solution of amino acids Amino acid
Glycine Alanine c-Aminobutyric acid Serine Proline Valine Threonine Asparagine Aspartic acid Homocysteine Glutamine Glutamic acid Methionine Histidine Phenylalanine Arginine Tyrosine Tryptophan
Peak intensity ratio (amino acid/internal standard)
Percentage RSD
Day 1
Day 5
Day 8
0.27 ± 0.05 0.38 ± 0.04 0.90 ± 0.05 0.42 ± 0.06 1.38 ± 0.07 0.43 ± 0.05 0.45 ± 0.07 0.60 ± 0.07 0.26 ± 0.03 0.20 ± 0.06 1.07 ± 0.17 0.10 ± 0.02 0.59 ± 0.05 1.33 ± 0.08 0.61 ± 0.05 2.31 ± 0.09 0.08 ± 0.01 0.66 ± 0.05
0.24 ± 0.06 0.34 ± 0.08 0.76 ± 0.16 0.33 ± 0.07 1.10 ± 0.14 0.34 ± 0.05 0.49 ± 0.07 0.63 ± 0.13 0.26 ± 0.04 0.21 ± 0.07 1.10 ± 0.12 0.08 ± 0.01 0.45 ± 0.16 1.43 ± 0.22 0.52 ± 0.10 2.59 ± 0.39 0.09 ± 0.02 0.52 ± 0.1
0.26 ± 0.05 0.40 ± 0.04 0.82 ± 0.04 0.40 ± 0.06 1.14 ± 0.07 0.43 ± 0.05 0.49 ± 0.06 0.63 ± 0.06 0.25 ± 0.04 0.23 ± 0.05 1.02 ± 0.07 — 0.56 ± 0.06 1.34 ± 0.26 0.63 ± 0.06 2.36 ± 0.73 — 0.52 ± 0.09
1.25 2.49 8.22 3.86 12.36 4.24 1.89 1.41 0.47 1.25 3.30 1.00 6.02 4.50 4.97 12.19 0.85 6.62
Table 3 Comparison of actual and measured concentrations of amino acids Amino acid
Actual concentration (lM)
Measured concentration (lM)
Error (percentage)
Alanine c-Aminobutyric acid Asparagine Glycine Homocysteine Methionine Phenylalanine Threonine Tryptophan Valine
180 180 180 260 90 260 90 180 90 180
181.0 ± 17.9 171.3 ± 10.2 187.3 ± 14.3 262.0 ± 28.1 98.0 ± 6.5 269.0 ± 20.9 98.0 ± 5.8 184.0 ± 6.7 98.5 ± 9.2 183.0 ± 14.1
0.56 5.16 4.05 0.77 8.90 3.46 8.90 2.20 9.44 1.66
190
M.A. Alterman et al. / Analytical Biochemistry 335 (2004) 184–191
Table 4 Comparison of theoretical (calculated based on amino acid sequence) and measured molar percentages of amino acids in BSA HCl hydrolysate Amino acid
Measured molar percentage
Calculated molar percentage
Glycine Alanine Serine Proline Valine Threonine Leucine + isoleucine Asparagine Aspartic acid Glutamine Glutamic acid Lysine Methionine Histidine Phenylalanine Arginine Tyrosine
1.1 8.2 3.7 5.6 8.0 6.3 12.0 1.8 6.8 0.9 13.3 8.4 1.5 3.6 7.9 5.7 4.6
2.8 8.1 4.8 4.8 6.8 6.3 12.9 2.4 6.9 3.4 10.6 9.9 0.9 3.1 5.4 4.7 3.8
Note. The values reported are the averages of data collected in five measurements.
The potential of the developed application of MALDI-TOF MS for the analysis of small biologically and physiologically important molecules was further proven by application of this technique to the analysis of glucose concentrations. Over the range of concentrations between 2 and 20 mM (i.e., physiological range of concentrations), the calibration curve was linear with a correlation coefficient of 0.988 and methyltyrosine was used as an internal standard (data not shown).
Discussion There were two previous reports describing analysis of a limited number of amino acids by MALDI-TOF MS. Wittman and Heinzle applied MALDI-TOF MS for the quantitation of lysine, alanine, and glucose in a microbial cultivation [16]. They used DHB as a matrix and stable isotope-labeled amino acids as internal standards. In spite of excellent analytical results for these three compounds, this method presents a number of analytical challenges if one would attempt to use it as a comprehensive AAA method. First, as we have shown, DHB as a matrix does not induce proper ionization of many amino acids in mixtures. Second, having an individual internal standard for each amino acid would greatly complicate the mass spectrum and could cause additional overlap of isotopomers. Further difficulty arises from the fact that all three compounds formed sodium and potassium ions in addition to the main protonated ion, and the ratio of the main protonated ion-to-sodium and potassium adduct ions depended on
sample and analysis conditions. Additional consideration in calculations required the presence of nonlabeled impurities in internal standards. In general, this method would be difficult to apply to complex mixtures of amino acids. Dally and colleagues published a study describing the MALDI-TOF MS method for AAA in cell culture media [19]. These authors developed calibration curves for 12 amino acids over the linear range of 1–100 lM. The main analytical difficulty of this study was the use of matrix ion as an internal standard. The matrix used was CHCA. It is well established that matrix ions are subject to ion suppression effects and could even be fully suppressed by analyte ions, not to mention that they are dependent on laser power, have evaporation rates different from those of amino acids, and so on [24]. Use of a protonated matrix ion as an internal standard requires relative responses of all amino acids analyzed to be constant and independent of analyte concentration. However, that was not the case in the study published by Dally and colleagues, as is evident from the MS data they reported [19]. Analysis of the mass spectra presented in that study reveals that the Pro/Val peak intensity ratio increased 1.9-fold, whereas the Val/Thr ratio decreased 1.8-fold, when the amino acid concentration increased 10-fold (from 10 to 100 lM). Relative responses of some other amino acids were changed as well. Furthermore, despite the fact that these authors analyzed an amino acid standard mixture containing 19 amino acids, they were able to develop calibration curves for only 12 amino acids and did not present any explanation as to what happened to the remaining 7 amino acids. On the other hand, the method presented in the current article allows fast and straightforward qualitative and quantitative analysis of amino acids by MALDITOF MS. The applicability of this method for the analysis of mixtures of free amino acids and HCl hydrolysates of proteins was demonstrated. No significant ion suppression effects were observed with the developed sample deposition technique, and the method was found to be reproducible. It should be recognized that this method has one inherent limitation: the analysis of isomeric amino acids (e.g., leucine, isoleucine). One way in which to get around this problem is to use MALDITOF MS–MS. Currently, experiments exploring application of MALDI-TOF MS–MS for AAA are under way in our laboratory. The most significant advantage of the presented method over conventional methods of AAA is incomparably shorter analysis time due to the absence of modification and separation steps. Possibilities for automation of the sample preparation and data processing can further advance this method. An additional area of application for this technique could be in clinical chemistry, as was shown by the data obtained with homocysteine, c-aminobutyric acid, and glucose. Normal values of plasma amino acids are well within
M.A. Alterman et al. / Analytical Biochemistry 335 (2004) 184–191
the quantitation limits and linearity intervals for corresponding amino acids determined in our study. The lowest plasma amino acid reference interval exhibits Asp (0–19 lM), whereas the highest one exhibits Gln (350–700 lM). All others demonstrate the normal range between 15 and 400 lM. Yet, for the MALDI-TOF MS AAA method to be used in clinical laboratories, a careful evaluation of the biological sample preparation technique (deproteinization by sulfosalicylic acid, use of acetonitrile, or ultrafiltration) must be performed to determine the levels of potential ion suppression compounds in the sample solution.
Acknowledgments We thank T. Williams for helpful scientific discussion and critical reading of the manuscript. Natalya Gogichayeva and Boris Kornilayev were supported by a research development grant from the University of Kansas (KUCR 00254).
References [1] S. Moore, W.H. Stein, Chromatography of amino acids on sulfonated polystyrene resins, J. Biol. Chem. 192 (1951) 663–681. [2] C. Cooper, N. Packer, K. Williams, Amino Acid Analysis Protocols, Humana Press, Totowa, NJ, 2001. [3] P. Furst, L. Pollack, T. Graser, H. Godel, P. Stehle, Appraisal of four pre-column derivatization methods for the high-performance liquid chromatographic determination of free amino acids in biological materials, J. Chromatogr. 499 (1990) 557–569. [4] V. Fierabracci, P. Masiello, M. Novelli, E. Bergamini, Application of amino acid analysis by high-performance liquid chromatography with phenyl isothiocyanate derivatization to the rapid determination of free amino acids in biological samples, J. Chromatogr. 570 (1991) 285–291. [5] M. Duncan, A. Poljak, Amino acid analysis of peptides and proteins on the femtomole scale by gas chromatography/mass spectrometry, Anal. Chem. 70 (1998) 890–896. [6] E. Zahou, H. Jornvall, T. Bergman, Amino acid analysis by capillary electrophoresis after phenylthiocarbamylation, Anal. Biochem. 281 (2000) 115–122. [7] T. Soga, D. Heiger, Amino acid analysis by capillary electrophoresis electrospray ionization mass spectrometry, Anal. Chem. 72 (2000) 1236–1241. [8] R. van Leuken, A. Duchateau, G. Kwakkenbos, Thermospray liquid chromatography/mass spectrometry study of diastereomeric isoindole derivatives of amino acids and amino acid amides, J. Pharmacol. Biomed. Anal. 13 (1995) 1459–1464.
191
[9] M. Piraud, C. Vianey-Saban, K. Petritis, C. Elfakir, J. Steghens, A. Morla, D. Bouchu, ESI–MS/MS analysis of underivatised amino acids: a new tool for the diagnosis of inherited disorders of amino acid metabolism—Fragmentation study of 79 molecules of biological interest in positive and negative ionisation mode, Rapid Commun. Mass Spectr. 17 (2003) 1297–1311. [10] A. Pons, C. Richet, C. Robbe, A. Herrmann, P. Timmerman, G. Huet, Y. Leroy, I. Carlstedt, C. Capon, J. Zanetta, Sequential GC/MS analysis of sialic acids, monosaccharides, and amino acids of glycoproteins on a single sample as heptafluorobutyrate derivatives, Biochemistry 42 (2003) 8342–8353. [11] S. Villas-Boas, D. Delicado, M. Akesson, J. Nielsen, Simultaneous analysis of amino and nonamino organic acids as methyl chloroformate derivatives using gas chromatography–mass spectrometry, Anal. Biochem. 322 (2003) 134–138. [12] P. Husek, Chloroformates in gas chromatography as general purpose derivatizing agents, J. Chromatogr. B Biomed. Sci. Appl. 717 (1998) 57–91. [13] P. Husek, P. Simek, Advances in amino acid analysis, LC–GC N. Am. 19 (2001) 986–999. [14] T. Bonk, A. Humeny, MALDI-TOF-MS analysis of protein and DNA, Neuroscientist 7 (2001) 6–12. [15] D. Mims, D. Hercules, Quantification of bile acids directly from urine by MALDI-TOF-MS, Anal. Bioanal. Chem. 375 (2003) 609–616. [16] C. Wittmann, E. Heinzle, MALDI-TOF MS for quantification of substrates and products in cultivations of Corynebacterium glutamicum, Biotechnol. Bioeng. 72 (2001) 642–647. [17] F. Ayorinde, K. Garvin, K. Saeed, Determination of the fatty acid composition of saponified vegetable oils using matrix-assisted laser desorption/ionization time-of-flight mass spectrometry, Rapid Commun. Mass Spectr. 14 (2000) 608–615. [18] M. Bucknall, K. Fung, M. Duncan, Practical quantitative biomedical applications of MALDI-TOF mass spectrometry, J. Am. Soc. Mass Spectr. 13 (2002) 1015–1027. [19] J. Dally, J. Gorniak, R. Bowie, C. Bentzley, Quantitation of underivatized free amino acids in mammalian cell culture media using matrix assisted laser desorption/ionization time-of-flight mass spectrometry, Anal. Chem. 75 (2003) 5046–5053. [20] R. Garden, J. Sweedler, Heterogeneity within MALDI samples as revealed by mass spectrometric imaging, Anal. Chem. 72 (2000) 30–36. [21] S. Cohen, B. Chait, Influence of matrix solution conditions on the MALDI-MS analysis of peptides and proteins, Anal. Chem. 68 (1996) 31–37. [22] I. Figueroa, O. Torres, D. Russell, Effects of the water content in the sample preparation for MALDI on the mass spectra, Anal. Chem. 70 (1998) 4527–4533. [23] R. Knochenmuss, A quantitative model of ultraviolet matrixassisted laser desorption/ionization including analyte ion generation, Anal. Chem. 75 (2003) 2199–2207. [24] R. Knochenmuss, F. Dubois, M. Dale, R. Zenobi, The matrix suppression effect and ionization mechanisms in matrix-assisted laser desorption/ionization, Rapid Commun. Mass Spectr. 10 (1999) 871–877.