Quantification of Peptides for the Monitoring of Protease-Catalyzed

The dynamic range of linearity was ∼1 order of magnitude. The method was applied successfully to monitor the time-dependent evolution of substrates ...
0 downloads 0 Views 221KB Size
Download PDF
Anal. Chem. 2006, 78, 291-297

Quantification of Peptides for the Monitoring of Protease-Catalyzed Reactions by Matrix-Assisted Laser Desorption/Ionization Mass Spectrometry Using Ionic Liquid Matrixes Andreas Tholey,* Masoud Zabet-Moghaddam, and Elmar Heinzle

Technische Biochemie, Universita¨t des Saarlandes, 66123 Saarbru¨cken, Germany

Ionic liquid matrixes (ILM) have been shown to allow very homogeneous sample preparations, facilitating relative quantifications using internal standards in matrix-assisted laser desorption/ionization mass spectrometry (MALDIMS). In the present work, the ability to perform quantifications of peptides without using internal standards in these matrixes was investigated. Linear correlations between peptide amount and signal intensities could be observed when increased molar matrix-to-analyte ratios were applied. The dynamic range of linearity was ∼1 order of magnitude. The method was applied successfully to monitor the time-dependent evolution of substrates and products in trypsin-catalyzed digests of single peptides and peptide mixtures. Thus, ionic liquid matrixes allow quantitative MALDI-MS without the need for internal standards, making the method a suitable tool for the fast screening of new enzymes or the search for substrates or inhibitors. Whereas the qualitative analysis of peptides and proteins by MALDI mass spectrometry is straightforward, the application for quantification of analytes is hampered by several inherent factors. In particular, ion signals depend not only on the amount of the analyte but also on its chemical composition, for example, the sequence of the amino acids in the case of peptides, and may suffer from suppression by other components present in the sample.1,2 In the case of MALDI-MS, additional problems caused by the inhomogeneous distribution of the analyte in the matrixanalyte cocrystallite resulting in poor spot-to-spot and shot-to-shot reproducibility makes quantification more difficult. Despite these problems, there still remains a potential for MALDI-MS to perform quantitative measurements.3-11 The main prerequisites for quan* To whom correspondence should be addressed. Phone: 49 (681) 302 4157. Fax: 49 (681) 302 4572. E-mail: [email protected]. (1) Cohen, S. L.; Chait, B. T. Anal. Chem. 1996, 68, 31-37. (2) Kussmann, M.; Nordhoff, E.; Rahbek-Nielsen, H.; Haebel, S.; Rossel-Larsen, M.; Jakobsen, L.; Gobom, J.; Mirgorodskaya, E.; Krol-lKristensen, A.; Palm, L.; Roepstorff, P. J. Mass Spectrom. 1997, 32, 593-601. (3) Duncan, M. W.; Matanovic, G.; Cerpa-Poljak, A. Rapid Commun. Mass Spectrom. 1993, 7, 1090-1094. (4) Gusev, A. I.; Muddiman, D. C.; Proctor, A.; Sharkey, A. G.; Hercules, D. M.; Tata, P. N.; Venkataramanan, R. Rapid Commun. Mass Spectrom. 1996, 10, 1215-1218. (5) Gobom, J.; Kraeuter, K. O.; Persson, R.; Steen, H.; Roepstorff, P.; Ekman, R. Anal. Chem. 2000, 72, 3320-3326. 10.1021/ac0514319 CCC: $33.50 Published on Web 12/02/2005

© 2006 American Chemical Society

titative MALDI-MS are (i) the application of suited measurement protocols, (ii) the use of internal standards, and (iii) the improvement of sample homogeneity. Optimized automated measurement protocols were designed to cover a representative area on the sample spot,12,13 thus reducing the errors caused by inhomogeneous sample distribution. Additionally, these protocols prevent errors caused by detector saturation. The biggest step toward quantitative use of MALDI-MS was the introduction of internal standards, thus leading to a relative quantification against a substance of known amount. The most effective internal standards are isotopomers of the analyte. Standards labeled with stable isotopes, such as 2H, 13C, 15N, or 18O, delivered best results in terms of accuracy of quantification.9-11,14-17 Another strategy is the use of non-isotope-labeled internal standards applying structurally modified compounds.3,18 For example, single amino acid exchanges19 or peptides with high molecular similiarity 20,21 were successfully used as internal standards for the relative quantification of peptides. The so-called hot-spot formation encountered in MALDI sample preparation using crystalline matrixes leads to a large variation of signal intensities and, thus, to a poor reproducibility (6) Mirgorodskaya, O. A.; Kozmin, Y. P.; Titov, M. I.; Korner, R.; Sonksen, C. P.; Roepstorff, P. Rapid Commun. Mass Spectrom. 2000, 14, 1226-1232. (7) Wittmann, C.; Heinzle, E. Biotechnol. Bioeng. 2001, 72, 642-647. (8) Horak, J.; Werther, W.; Schmid, E. R. Rapid Commun. Mass Spectrom. 2001, 15, 241-248. (9) Gusev, A. I.; Wilkinson, W. R.; Proctor, A.; Hercules, D. M. Anal. Bioanal. Chem. 1996, 354, 455-463. (10) Sleno, L.; Volmer, D. A. Anal. Chem. 2005, 77, 1509-1517. (11) Sleno, L.; Volmer, D. A. Rapid Commun. Mass Spectrom. 2005, 19, 19281936. (12) Nicola, A. J.; Gusev, A. I.; Proctor, A.; Hercules, D. M. Anal. Chem. 1998, 70, 3213-3219. (13) Kang, M. J.; Tholey, A.; Heinzle, E. Rapid Commun. Mass Spectrom. 2001, 15, 1327-1333. (14) Kang, M. J.; Tholey, A.; Heinzle, E. Rapid Commun. Mass Spectrom. 2000, 14, 1972-1978. (15) Mirgorodskaya, O. A.; Korner, R.; Novikov, A.; Roepstorff, P. Anal. Chem. 2004, 76, 3569-3575. (16) Meng, Z.; Limbach, P. A. Anal. Chem. 2005, 77, 1891-1895. (17) Zappacosta, F.; Annan, R. S. Anal. Chem. 2004, 76, 6618-6627. (18) Bungert, D.; Heinzle, E.; Tholey, A. Anal. Biochem. 2004, 326, 167-175. (19) Wang, Q.; Jakubowski, J. A.; Sweedler, J. V.; Bohn, P. W. Anal. Chem. 2004, 76, 1-8. (20) Muddiman, D. C.; Gusev, A. I.; Proctor, A.; Hercules, D. M.; Venkataramanan, R.; Diven, W. Anal. Chem. 1994, 66, 2362-2368. (21) Helmke, S. M.; Yen, C. Y.; Cios, K. J.; Nunley, K.; Bristow, M. R.; Duncan, M. W.; Perryman, M. B. Anal. Chem. 2004, 76, 1683-1689.

Analytical Chemistry, Vol. 78, No. 1, January 1, 2006 291

over a sample spot. Improvement of the homogeneity of MALDI samples can be achieved with fast evaporation,22 deposition by means of electrospray devices,23 or the addition of co-matrixes, such as fucose.9,24 Liquid matrixes generally do not exhibit hotspot formation, but up to now, there have been only a few examples of liquid matrixes combining sufficient absorption in UV together with stability under high-vacuum conditions.25 Further, these systems suffer from extensive adduct formation and low ionization efficiency.26 A breakthrough was the introduction of ionic liquid matrixes (ILM) by Armstrong, Gross, and coworkers.27 These ILM are mixtures of classically used MALDI matrixes, such as SA, CCA, or DHB, with equimolar amounts of organic bases. These ILM form highly viscous layers on the MALDI target and have been successfully applied for the measurement of peptides, proteins and polymers;27,28 oligonucleotides;29 low molecular weight compounds, such as amino acids and sugars;30-32 and phospholipids.33 Due to the high sample homogeneity achievable in these matrixes, they enable an improved relative quantification of amino acids;30 sugars;31 and peptides, small proteins, and oligonucleotides34 when applying appropriate internal standards. Enzyme-catalyzed reactions play a pivotal role in biotechnological processes as well as in medical and pharmaceutical research. For the development of new biocatalysts, the investigation of substrate specificities or the development of enzyme inhibitors, the screening process is still a bottleneck.35 Quantitative MALDI-MS has been shown to be a valuable technique for such processes.13,14,18,31,36-38 Principally, two ways of screening process are possible. First, among a pool of newly designed biocatalysts or new enzymes from natural sources, the most active catalyst has to be identified (enzyme screening). This can be effectively done by the quantification of substrates and products of the enzyme-catalyzed reactions. Alternatively, for a given enzyme, a pool of substrates or inhibitors can be tested (substrate screening). Whereas for low molecular weight substrates and products, for example, amino acids and sugars, the corresponding isotopelabeled internal standards are available in many cases, this (22) Nicola, A. J.; Gusev, A. I.; Proctor, A.; Jackson, E. K.; Hercules, D. M. Rapid Commun. Mass Spectrom. 1995, 9, 1164-1171. (23) Onnerfjord, P.; Ekstrom, S.; Bergquist, J.; Nilsson, J.; Laurell, T.; MarkoVarga, G. Rapid Commun. Mass Spectrom. 1999, 13, 315-322. (24) Distler, A. M.; Allison, J. Anal. Chem. 2001, 73, 5000-5003. (25) Turney, K.; Harrison, W. W. Rapid Commun. Mass Spectrom. 2004, 18, 629-635. (26) Cohen, L. H.; Gusev, A. I. Anal. Bioanal Chem. 2002, 373, 571-586. (27) Armstrong, D. W.; Zhang, L. K.; He, L.; Gross, M. L. Anal. Chem. 2001, 73, 3679-3686. (28) Zabet-Moghaddam, M.; Heinzle, E.; Lasaosa, M.; Tholey, A. Anal. Bioanal. Chem. 2006, in press. (29) Carda-Broch, S.; Berthod, A.; Armstrong, D. W. Rapid Commun. Mass Spectrom. 2003, 17, 553-560. (30) Zabet-Moghaddam, M.; Heinzle, E.; Tholey, A. Rapid Commun. Mass Spectrom. 2004, 18, 141-148. (31) Bungert, D.; Bastian, S.; Heckmann-Pohl, D. M.; Giffhorn, F.; Heinzle, E.; Tholey, A. Biotechnol. Lett. 2004, 26, 1025-1030. (32) Mank, M.; Stahl, B.; Boehm, G. Anal. Chem. 2004, 76, 2938-2950. (33) Li, Y. L.; Gross, M. L.; Hsu, F. F. J. Am. Soc. Mass Spectrom. 2005, 16, 679-682. (34) Li, Y. L.; Gross, M. L. J. Am. Soc. Mass Spectrom. 2004, 15, 1833-1837. (35) Tholey, A.; Heinzle, E. Adv. Biochem. Eng. Biotechnol. 2002, 74, 1-19. (36) Zabet-Moghaddam, M.; Kruger, R.; Heinzle, E.; Tholey, A. J. Mass Spectrom. 2004, 39, 1494-1505. (37) Schluter, H.; Jankowski, J.; Rykl, J.; Thiemann, J.; Belgardt, S.; Zidek, W.; Wittmann, B.; Pohl, T. Anal. Bioanal. Chem. 2003, 377, 1102-1107. (38) Liesener, A.; Karst, U. Anal. Bioanal. Chem. 2005, 382, 1451-1464.

292

Analytical Chemistry, Vol. 78, No. 1, January 1, 2006

situation changes when peptides or proteins are the substrates of the enzymes. Further, the broad peaks caused by natural isotope patterns, adduct ions, and low mass resolving power of the instrument, coupled with metastable losses of water or ammonia for higher molecular weight analytes, complicate the selection of internal standards that are well-resolved from analyte signals.9 The goal of the present work is to test further the use of ILMs for quantification without an internal standard by using MALDIMS. This would have an ideal impact on the screening for new enzyme variants, either by obtaining quantitative data for a particular conversion or by semiquantitative data for the monitoring of the change of a substrate or product amounts. Parameters such as matrix-to-analyte ratios that are necessary for the design of experiments and the dynamic range for linear correlation between the ion response for model peptides, as well as the limitations of the method, were evaluated. The applicability for the determination of enzyme activities was shown with protease-catalyzed reactions. EXPERIMENTAL SECTION Materials. The peptides, angiotensin II, substance P, neurotensin, ACTH(1-17) and ACTH(18-39); the MALDI matrixes R-cyano-4-hydroxycinnamic acid (CCA) and indoleacrylic acid (InAA); and the organic bases N,N-dimethylethylenediamine (asymmetric) (DMED) and 3-(dimethylamino)-1-propylamine (DMAPA) were obtained from Sigma-Aldrich (Taufkirchen, Germany). Trp11-neurotensin was from Bachem (Weil a.R., Germany). Sequencing grade trypsin (modified) was obtained from Promega (Madison, WI). Trifluoroacetic acid (TFA) was obtained from Fluka (Neu-Ulm, Germany). Water was purified using a Millipore water purification system (Bedford, MA). Organic solvents were all of HPLC grade. MALDI-TOF MS. Analyses were performed using a Bruker Reflex III time-of-flight mass spectrometer (Bruker Daltonics, Bremen, Germany) equipped with a SCOUT 384 probe ion source. The system was equipped with a pulsed nitrogen laser (337 nm, model VSL-337ND; Laser Science Inc., Boston, MA) with an energy of 400 µJ/ pulse. The ions were accelerated under delayed extraction conditions using an acceleration voltage of 20 kV. A reflector voltage of 22.5 kV was applied. A LeCroy 9384C, 4-GHz digital storage oscilloscope was used for data acquisition (LeCroy coop., Chestnut Ridge, NY). Data were collected using ReflexControl software (Bruker Daltonics, Bremen, Germany) and processed with FlexAnalysis software (Bruker Daltonics, Bremen, Germany). External mass calibration was achieved using standard peptides in standard crystalline matrixes. For quantitative analysis, the automated data acquisition protocol AutoXecute (Bruker Daltonics, Bremen, Germany) was used. A fixed laser power was applied throughout the measurement. The maximum allowed number of positions on each sample spot on the target was nine, and the nine positions were arranged in a cross starting from the sample center. Maximum allowed number of shots on one position on a sample spot was 30. In total, 150 shots were collected per spot. Only signals with a mass resolution higher than 1000 (based on full width of half-maximum (fwhm)) and with a signal-to-noise ratio higher than 5 were collected. All data points presented in the calibration curves or for the determination of enzyme activities represent the mean

values of five independent measurements. Mathematical outliers were removed.30 Control experiments using internal standards (Trp11-neurotensin) were performed as described earlier.18,34 Sample Preparation. Stock solutions (500 mM) of ionic liquid matrixes were prepared as described earlier30 by mixing equimolar amounts of crystalline MALDI matrix (CCA, InAA) with one of the organic bases (DMED, DMAPA) in 70% acetonitrile/0.1% TFA (CCA salts) or 80% acetonitrile/0.1% TFA (InAA salts) and characterized by MALDI-MS to obtain blank spectra.30 For sample preparation, these stock solutions were mixed with diluted peptide stock solutions (0.5 mM, in 70% acetonitrile/water) to obtain the particular matrix-to-analyte ratios and sample amounts on the target as given in the text. Premixed matrix-analyte mixtures (1 µl) were pipetted onto the stainless steel target (dried droplet preparation) and dried at ambient conditions prior to measurement. Tryptic Proteolysis of Peptides. Tryptic digestion of neurotensin (initial concentrations 100, 200, and 300 µM) and a fivepeptide mixture (angiotensin II, substance P, neurotensin, ACTH(1-17), ACTH(18-39), 100 µM each) were performed in 30 mM ammonium bicarbonate buffer (pH 8) at 37 °C. Reactions were started by addition of aqueous trypsin stock solution to obtain a substrate-to-protease ratio between 200 and 400 (g/g). Small aliquots of the reaction mixture were taken after certain time intervals and diluted 20-fold with 80% acetonitrile. MALDI samples were prepared by mixing with ionic liquid matrix stock solutions as described above without any further sample pretreatment. KM values were calculated using the Lineweaver-Burk equation. RESULTS AND DISCUSSION Qualitative Analysis of Peptides in ILM. Ionic liquid matrixes have been defined as equimolar mixtures of crystalline, acidic MALDI matrixes with organic bases. Numberless combinations of these two components are possible. In the present study, we investigated different combinations of R-cyano-4-hydroxycinnamic acid and indoleacrylic acid with the bases N,N-dimethylethylenediamine or 3-(dimethylamino)-1-propylamine for the measurement of model peptides. As known for the classical matrixes, the utilizability of a matrix for a particular analyte or a certain class of analytes has to be elucidated. Both CCA- and InAA-based matrixes were found to be suitable for the analysis of peptides. All ILM tested here formed colorless to yellowish viscous films on the MALDI target after evaporation of the solvent and were stable, showing no variation of spectrum quality, even after 12 h under high-vacuum conditions in the mass spectrometer. As described earlier for other ionic liquid matrixes,27,30,32,34 the analyte distribution in the ILM was very homogeneous, showing no hotspot formation. This results in a high shot-to-shot reproducibility of the signal intensities throughout a sample spot. Compared to the pure crystalline matrixes, the ionization of peptides in the ILM required slightly higher laser energies/fluences (e.g., 32% in CCA (arbitrary laser energy units), 40% in ILM CCA-DMAPA). The resolution of the signals did not differ in both matrixes. As described earlier for low molecular weight analytes, an increased formation of sodium and potassium adducts was observed.28,30 In addition to the signals for [M + Na]+/[M + K]+ ions, we observed small signals of ions of the form [M + 2Na - H]+and [M + Na + K - H]+ (Figure 1). The results of quantification presented below did not change significantly when the alkali adducts were

Figure 1. Mixture of the peptides angiotensin II (m/z ) 1046.5), substance P (m/z ) 1347.7), neurotensin (m/z ) 1672.9), ACTH(1-17) (m/z ) 2093.1) and ACTH(18-39) (m/z ) 2465.2) measured in CCA-DMAPA. a, [M + Na]+; b, [M + K]+; c, [M + 2Na - H]+; d, [M + Na + K - H]+; and e, DMAPA adducts. An asterisk signifies nonidentified impurity.

considered, either isolated or summed up with the protonated signals.18 Nevertheless, the more intense signals corresponding to protonated molecules were used for quantification purposes. In the case of the ILM CCA-DMED, no adducts of the peptides with components of the ILM were observed, whereas in CCADMAPA, peptides formed adducts with the basic component of the matrix (Figure 1). All results obtained in InAA-based matrixes were comparable to those in the CCA-based ILM. Influence of the Matrix-to-Analyte Ratio on the Ion Response in ILM. The application of MALDI-MS for direct quantification without use of internal standards is hampered by several factors: (i) the dependence of ionization on the laser energy/fluence;39 (ii) the hot spot formation observed frequently in crystalline matrixes, causing inhomogeneous distribution of signal intensities; and (iii) the observation that over a larger range of M/A ratios, signal intensities generally increase nonlinearly.39 The later problem is, among other complex events in the MALDI process, associated with matrix/analyte suppression effects.40 The use of ILM has been shown to diminish the problems provoked by the hot spot formation and the need for varying laser energies/ fluences caused by this phenomenon.30 This makes these matrixes a good alternative for the application in relative quantitative measurements using internal standards.30,34 The goal of the present work was to elucidate whether and in which range a linear correlation between analyte amount and signal intensities can be found in these ILM. A dilution series of the peptide neurotensin in a constant amount of the ILM CCADMAPA was prepared, and the corresponding acquired signal intensities were collected by applying an automated measurement protocol using a constant laser energy/fluence. The molar matrixto-analyte (M/A) ratios were between 25 000 (lowest amount of analyte) and 50 (highest amount of analyte). As shown in Figure 2a, the signal intensities increased nonlinearly with increasing analyte amounts up to ∼100 pmol, then reaching a region of saturation. At sample amounts above ∼150 pmol, a decrease of the signal intensities was observed. The saturation effect can (39) Dreisewerd, K. Chem. Rev. 2003, 103, 395-426. (40) Knochenmuss, R.; Zenobi, R. Chem. Rev. 2003, 103, 441-452.

Analytical Chemistry, Vol. 78, No. 1, January 1, 2006

293

Figure 2. Peak intensities of neurotensin in dependence of the sample amount and the molar matrix-to-analyte (M/A) ratio applied. (a) M/A of 25 000 (2 pmol peptide) to 50 (1000 pmol peptide); matrix, CCA-DMAPA. (b) M/A between 500 000 (500 fmol peptide) and 35 700 (7 pmol peptide); R2 ) 0.995; mean error, 8%; matrix, CCADMAPA. All measurement points represent averages of five independent measurements. A constant laser energy/fluence was applied throughout the measurement.

theoretically be caused both by saturation of the detector, which is limited to 256 counts per shot, and by suppression effects,40 but the automated measurement protocol applied excluded to the first possibility.13 It is a well-known fact that the molar matrix-toanalyte ratios have a strong influence on the detectability and intensity of analyte signals in MALDI-MS.39 Typically, molar M/A ratios between 10 and 100 were found to be optimal for low molecular weight compounds;14 for peptides and proteins, typically ratios in the order of 103-105 are applied.41 At higher sample amounts (corresponding to lower M/A values), the amount of matrix may no longer be sufficient for fulfilling its role in soft ionization and desorption of the analyte typical for MALDI-MS, leading to the decrease of signal intensities and, finally, at further decreased matrix amounts, to the complete disappearance of analyte signals (data not shown; contributions of ionization by matrix-free laser desorption/ionization (LDI) may not be relevant in this discussion). The increase at low sample amounts with the corresponding higher M/A ratios was nonlinear, but deviation from linearity was not pronounced, as shown in Figure 2a. This led to the working theory that elevated M/A ratios could favor a linear relation between intensity and sample amount. (41) Hillenkamp, F.; Karas, M.; Beavis, R. C.; Chait, B. T. Anal. Chem. 1991, 63, 1193A-1203A.

294 Analytical Chemistry, Vol. 78, No. 1, January 1, 2006

Therefore, a second set of experiments was performed in which the M/A ratios were increased to values between 25 000 and 500 000. Figure 2b shows that under these conditions, a linear relation (R2 ) 0.995, mean error ) 8%) could be observed in the range between 500 fmol and 7 pmol. Above 7 pmol, the curve reached a plateau. Other peptides investigated in this study showed similar ranges of linearity, spanning a dynamic range of ∼1 order of magnitude. This dynamic range is slightly lower than that found for relative quantifications,34 in which the relation between analyte and internal standard is the limiting factor.14 The limitation of the dynamic range is mainly caused by three factors which interfere with each other: (i) the laser energy/fluence applied, (ii) the amount of matrix which can be placed on the target, and (iii) the limit of detection and the ionization behavior of the analyte. From practical reasons, for a quantification of an unknown sample, the laser energy has to be kept constant for the calibration curve as well as for the measurement itself. On the other hand, laser energy has an influence on the detectability of low amounts of analytes and on the peak saturation at higher sample amounts. Therefore, laser energy/fluence has to be optimized and adapted to the analytical problem to be solved prior to the measurement of the calibration curve. The use of different laser energies allows for a shift of the dynamic range to either higher or lower sample amounts, but the dynamic range itself is not extended by this. To achieve the increased M/A ratios resulting in the linear relation between sample amount and signal intensities, it is possible either to increase the matrix amount on the target or to decrease sample amount. The first factor is limited by practical reasons not allowing an unlimited increase of matrix amount on the target. In our study, typically 250 nmol of the ionic liquid matrix was spotted on the target. The second factor is limited by the limit of detection of the analyte. The limits of detection for peptides in ILM were reported earlier to be slightly increased, as compared to the corresponding crystalline matrixes,32 but this effect depends strongly on both the nature of the analyte and the composition of the ionic liquid matrix. A fairly linear correlation (R2 ) 0.65, slope 16.32) between sample amount and signal intensity could also be observed in the pure crystalline matrixes, when the optimized parameters evaluated for the measurement of the calibration curve in ILM shown in Figure 2a were applied (data not shown). Compared to the results obtained in the ionic liquid matrixes, the errors and deviations are, however, tremendously increased (mean standard deviation ) 51%), which is mainly caused by the inhomogeneous distribution of the analyte in the sample preparation. Peptide Mixtures. In addition to the analysis of single peptides, the quantification of a four-peptide mixture (angiotensin II, substance P, neurotensin, ACTH(1-17)) in the matrix system CCA-DMAPA was tested. We observed an increase in peak intensities with increasing amounts of the peptides in the range between 250 fmol and 2.25 pmol, but in contrast to the situation observed for the single peptides, the increase was not perfectly linear, expressed by R2 values of 0.966 (ACTH(1-17)), 0.978 (neurotensin), 0.988 (angiotensin II), and 0.989 (substance P). This nonideal behavior may be caused by the fact that analytes in a mixture are known to influence the desorption/ionization behavior of each other. The difference in the slopes of the calibration curve for the peptide neurotensin measured in the four-peptide mixture

Table 1. Products of Tryptic Digestion of a Five Peptide Mixture Measured in CCA-DMAPA analyte/sequencea,b angiotensin II DRVYI HPF neurotensin pGlu-LYEN KPR RP YIL substance P RPKPQ QFFGL M-NH2 ACTH(1-17) SYSME HFRWG KPVGK KR ACTH(18-39) RPVKVYPNGAEDESAEAFPLEF a

theoretical products of tryptic cleavage m/z, sequence, identificationc m/z ) 290, DR, (-) m/z ) 775, VYIHPFG, (-) m/z ) 661.4, RP YIL, (+) m/z ) 1030.5, pGlu-LYEN KPR, (+) m/z ) 1056.5, SYSMEHFR, (+) m/z ) 1055.6, WGKPVGKKR, (+) m/z ) 1808.8, SYSMEHFRWGKPVGK, (+) m/z ) 771.5, WGKPVGK, (+) m/z ) 499.3, RPVK, (-) m/z ) 1984.8, VYPNGAEDESAEAFPLEF, (+)

Bold residues: cleavage site. b Underlined residues: masked cleavage site. c Signal detected: (+), not detected: (-).

(8.49) and as a single peptide (9.46) gives a clear indication of this effect. Hence, due to the complex interrelations in mixtures, single analyte measurements should be preferred when accurate quantification is the goal.40 On the other hand, the accuracy reached in this experiment is sufficient for the determination of relative enzyme activities in analyte mixtures, as outlined below. The calibration curves of the four analytes had different slopes (6.43, angiotensin II; 10.43, substance P; 8.49, neurotensin; and 5.57, ACTH(1-17)). The composition of analytes, for example, the amino acid sequence of peptides, is known to influence the behavior of an analyte in MALDI mass spectrometry.42 No obvious correlations between peptide lengths, composition of the peptides (e.g., the number of hydrophobic/hydrophilic or charged residues), and the corresponding slopes were found in the limited set of analytes investigate here, but a closer investigation of this could be a further potential application for the method presented. Monitoring of Enzyme-Catalyzed Reactions. For the screening for new enzymes or the discovery of new substrates/inhibitors for a given enzyme, it is necessary to determine the activity of the enzymes. MALDI mass spectrometry can be used to monitor enzymatic conversions quantitatively when suitable internal standards are available, but this is often not the case for peptidic substrates or inhibitors. Therefore, we tested the possibility for absolute quantification by MALDI-MS using ionic liquid matrixes to determine enzyme activities. Two sets of experiments were performed, which can be used as models for the two different screening processes: the monitoring of the enzymatic conversion of a single peptide (enzyme screening) and the conversion of a mixture of different peptides (substrate/inhibitor screening). As a model reaction for an enzyme screening, the tryptic digestion of the peptide neurotensin was investigated. Aliquots of the reaction mixture were taken after certain time intervals, and the samples were prepared with the ILM InAA-DMED without any prepurification (e.g., desalting). In addition to the substrate of the reaction, two products with m/z 661.4 and 1030.5 corresponding to the peptides RPYIL and pGlu-LYENKPR, respectively, could be observed. Neurotensin posseses three basic residues (Table 1), but two arginines are followed by proline residues, which prevent or at least prohibit a cleavage at these (42) Baumgart, S.; Lindner, Y.; Kuhne, R.; Oberemm, A.; Wenschuh, H.; Krause, E. Rapid Commun. Mass Spectrom. 2004, 18, 863-868.

residues.43 Figure 3a shows the evolution of the concentrations of the substrate and the two products. Substrate concentrations were calculated using a calibration curve (insert in Figure 3a), which was measured for on-target sample amounts between 1 and 9 pmol. In the second part of this experiment, the starting concentrations of the substrate were varied to calculate kinetic data for the enzyme reaction (Figure 3b). Using initial velocities, the KM value for the substrate neurotensin was calculated to be 520 µM, which lies in the range of comparable substrates found in the literature.44,45 Note that for accurate determination of kinetic parameters by Lineweaver-Burk plot, more sampling points in the early phase of the reaction would be beneficial to cover a wider linear range. As a further control, the signal intensities of the two products of the enzymatic cleavage were monitored. A linear correlation could be found between signal intensities and the calculated amounts of the products (insert Figure 3b). The amounts of both products can be calculated, assuming that from each molecule of the substrate peptide, two product peptides are formed in equimolar amounts. With this approach, it is possible to gain calibration curves for both products, which were used for the calculation of the product concentrations shown in Figure 3a. The small deviations between the two product concentrations in Figure 3a are a consequence of having only three product concentrations for the generation of the calibration curves in the experiment performed here. The calculated sum of substrates with the mean of the amounts of the products formed was constant over the complete reaction (Figure 3a), which serves as a further confirmation of the reliability of the results. The results of the determination of neurotensin concentration in the enzymatic conversion shown in Figure 3a were compared with those obtained by two established methods: (i) the relative quantification by MALDI-MS in ILM using a homologous peptide (Trp11-neurotensin) as internal standard and (ii) with quantification of the substrate by HPLC-UV measurements with 214-nm detection (Figure 3c). The average deviation among the three methods was 6%. The highest deviations were observed at the (43) Thiede, B.; Lamer, S.; Mattow, J.; Siejak, F.; Dimmler, C.; Rudel, T.; Jungblut, P. R. Rapid Commun. Mass Spectrom. 2000, 14, 496-502. (44) Palm, A. K.; Novotny, M. V. Rapid Commun. Mass Spectrom. 2004, 18, 1374-1382. (45) Hill, C. R.; Tomalin, G. Anal. Biochem. 1982, 120, 165-175.

Analytical Chemistry, Vol. 78, No. 1, January 1, 2006

295

Figure 4. Tryptic digest of a mixture of angiotensin II, substance P, neurotensin, ACTH(1-17), and ACTH(18-39) monitored in the matrix CCA-DMAPA. Initial on-target amount of each peptide was 2.5 pmol.

Figure 3. Tryptic digest of neurotensin monitored in the matrix InAA-DMED. (a) Time-dependent development of substrate and product concentrations. Substrate concentrations (-9-) were calculated using the calibration curve (insert, measured in InAA-DMED; molar matrix-to-analyte ratios from 250 000 to 25 000; matrix amount, 250 nmol; each data point represents the average of five independent measurements) and taking into account the dilution steps prior to sample preparation. Product concentrations (m/z 661, O; m/z 1030, right-pointing arrow) were calculated from calibration curves in the insert in (b). The dashed line (- - -1- - -) represents the calculated sum of substrate and the mean of the two product concentrations. (b) Substrate consumption vs time for three different initial substrate concentrations. Insert: Evolution of the product amounts. Product amounts were calculated assuming equimolar product formation from the substrate as described in the text. (c) Comparison of substrate concentrations determined by quantitative MALDI-MS (five independent measurements) using the ionic liquid matrix (i) using an internal standard, or (ii) without internal standard and by HPLC-UV (214 nm, single measurement). 296 Analytical Chemistry, Vol. 78, No. 1, January 1, 2006

lowest concentrations (12%, 10 and 15 min), whereas at high concentrations, deviations were below 3%. Semiquantitative Monitoring of Multisubstrate Conversions. In a second set of experiments, the conversion of an equimolar mixture of five peptides (angiotensin II, substance P, neurotensin, ACTH(1-17), and ACTH(18-39), (Table 1)) by trypsin was monitored. The signal intensities of the five peptides evolved in different manners (Figure 4). ACTH(1-17) was converted fastest among the peptides investigated, followed by neurotensin. Neutrotensin has one (see above), ACTH(1-17) three cleavage sites for trypsin (Table 1). The cleavage products described earlier for neurotensin as well as further product peptides originating from the cleavage of ACTH(1-17) could be identified in the spectra (Table 1). A direct interpretation of the kinetic parameters of a peptide with more than one cleavage site is hampered by the fact that different cleavage sites may have different kinetic parameters and by the possibility that cleavage products may also be substrates for further cleavages. We found all potentially possible signals originating from such sequential cleavages of ACTH(1-17), with the exception of the cleavage product of the C-terminal arginine residue (Table 1). Additionally, it was possible to observe the time-dependent appearance of the different peptide fragments formed by first-cleavage events, accompanied by the disappearance of the signals of these precursors (data not shown). Despite the complexicity of the cleavage pattern, a comparison of the overall efficiency of the cleavage of the peptides neurotensin and ACTH(1-17) can be clearly drawn from the experiment, showing that the peptide ACTH(1-17) is overall digested faster than the peptide with one cleavage site. The signal intensities of the peptide ACTH(18-39) were only slightly reduced over the time course. ACTH(18-39) exhibits one tryptic cleavage site, and the corresponding theoretical cleavage product with m/z ) 1984.8 could be identified in the spectra. The second product with m/z ) 499.3 was not detected. The blank spectrum measured with the analyte-free ILM did not show any signal at this m/z value; thus, signal overlap can be excluded as a reason for the nondetectability of the latter peptide. Potentially, the slower conversion of this peptide is caused by the neighboring sequence. The signal intensities of the peptides

angiotensin II and substance P showed only a very small loss of intensity of ∼20% in the first 2 min of the reaction; afterward, signal intensities were mainly constant. Angiotensin II posseses one tryptic cleavage site, but we did not observe any cleavage products in the spectrum. Potentially, the neighboring sequence with an acidic residue (Asp) N-terminal and a bulky hydrophobic residue (Val) C-terminal to the cleavage site (Table 1) inhibit the tryptic cleavage of this peptide. Substance P posseses two potential tryptic cleavage sites, both masked by proline residues, which should not be cleaved efficiently, according to the literature.43 We did not identify any cleavage products of this peptide; therefore, the reasons for the decrease of signal intensities at the beginning of the reaction still remain unclear. It has to be mentioned that the observed behaviors of the peptides in the mixture were the same as those of the isolated peptides (data not shown). As shown above, the slopes of the different peptides in the mixture differed in dependence of the molecular properties of the analytes and were additionally influenced by the presence of other analytes. In addition, caused by the formation of cleavage peptides as well as by different velocities of the reduction of the substrate amounts, the composition of the analyte mixture and, thus, the problems related to peak suppression,40 varies in an unpredictable way during enzymatic conversion. The strong fluctuation of the signal intensities of the peptides angiotensin II and substance P between 4 and 20 min shown in Figure 4 may be caused by such effects. Therefore, for the monitoring of the enzymatic conversion of the peptide mixture, signal intensities rather than concentrations calculated from calibration curves were presented here. This hampers the determination of kinetic parameters, for example, vmax or KM values, and therefore renders the method to a semiquantitative approach. Nevertheless, the differences in the evolution of the signal intensities are sufficient to identify good and bad substrates for this enzymatic conversion. To confirm our results, we additionally monitored the enzymatic conversion of the five peptide mixture using Trp11-neurotensin as internal standard.34 In accordance with the results reported above without an internal standard, the same timedependent behavior of all five peptides was observed in this experiment (data not shown). The fluctuation of the signals of the peptides angiotensin II, substance P and ACTH(18-39) was even more pronounced as compared to the situation without the internal standard. In addition, the unexplainable initial decrease of the concentrations of angiotensin II and substance P was also observed in the latter experiment. The presence of an additional analyte causes even more peak suppression, which enhances the problems described above. This also contributes to the advantages of internal standard-free quantification.

CONCLUSIONS The application of ionic liquid matrixes together with an increased matrix-to-analyte ratio allows a direct quantification of peptides by MALDI mass spectrometry without the use of internal standards. The importance of the M/A ratio complicates the application of the method for analyses, in which the concentration for the analytes is not known, for example, for quantification of peptides in a proteomics environment. For these applications, a relative quantification rather than the absolute quantification presented here is recommended. In case of enzyme reactions, the starting concentration of the substrate is a known parameter. The limited dynamic range does not cause problems, due to the fact that the initial velocities of enzyme reactions deliver the most important information for the determination of enzyme activities. A limiting factor for a multisubstrate screening is given by the number of analytes measurable in one preparation, which is caused by the effects of peak suppression. Together with the quantitative information, mass spectrometric detection also enables the simultaneous structural analysis and identification of the products. In the case of unanticipated products, for example, in cases in which unknown cleavage sites occur, the capability of mass spectrometry to elucidate structural information by means of MS/MS experiments is a potential additional tool for understanding enzymatic conversions. The method is not restricted to proteolytic reactions, but can principally be used for all kinds of enzymatic conversions that go hand-in-hand with mass changes, for example, transferase- or oxidase/reductase-catalyzed reactions as well as for nonenzymatic reactions. Together with the ability of MS for multiplexing, the possible abandonment of expensive standards, and the tolerance against salts, quantitative MALDI-MS using ionic liquid matrixes is, therefore, a suitable method for fast screening for new enzymes or for the search for substrates or inhibitors. ACKNOWLEDGMENT We thank Nathanael Delmotte for the help with LC-UV analysis and Nathalie Selevsek for helpful discussion. This work was supported by the Bundesministerium fuer Bildung und Forschung (BMBF) within the joint project Biokatalyse in ionischen Fluessigkeiten.

Received for review August 10, 2005. Accepted November 2, 2005. AC0514319

Analytical Chemistry, Vol. 78, No. 1, January 1, 2006

297