Identification of Protein Ligands in Complex Biological Samples Using

Hutchens, T. W.; Yip, T. T. Rapid Comun. Mass Spectrom. 1993, 7, 576−580. [Crossref], [CAS] . New desorption strategies for the mass-spectrometric a...
0 downloads 0 Views 351KB Size
Anal. Chem. 2003, 75, 3385-3395

Identification of Protein Ligands in Complex Biological Samples Using Intensity-Fading MALDI-TOF Mass Spectrometry Josep Villanueva,‡,† Oscar Yanes,‡,† Enrique Querol,‡ Luis Serrano,*,§ and Francesc X. Aviles*,‡

Institut de Biotecnologia i de Biomedicina, and Departament de Bioquı´mica, Universitat Auto` noma de Barcelona, 08193 Bellaterra (Barcelona), Spain, European Molecular Biology Laboratory (EMBL) Meyerhofstrasse 1, D-69117 Heidelberg, Germany

The easy detection of biomolecular interactions in complex mixtures using a minimum amount of material is of prime interest in molecular and cellular biology research. In this work, a mass spectrometry MALDI-TOF based approach, which we call intensity-fading (IF MALDITOFMS), and which was designed for just such a purpose, is reported. This methodology is based on the use of the MALDI ion intensities to detect quickly the formation of complexes between nonimmobilized biomolecules in which a protein is one of the partners (protein-protein, proteinpeptide, protein-organic molecule, and protein-nucleic acid complexes). The complex is detected through the decrease (fading) of the molecular ion intensities of the partners as directly compared to the MALDI mass spectrum of the mixture (problem and control molecules) following the addition of the target molecule. The potential of the approach is examined in several examples of model interactions, mainly involving small nonprotein and protein inhibitors of proteases, at both the qualitative and semiquantitative levels. Using this method, different protein ligands of proteolytic enzymes in total extracts of invertebrate organisms have been identified in a simple way. The proposed procedure should be easily applied to the high-throughput screening of biomolecules, opening a new experimental strategy in functional proteomics. Cell biology involves a complex network of biomolecular interactions1 which, in many cases, is the target for the development of new drugs in the pharmaceutical industry.2,3 Therefore, multiple methodologies have been or are in the process of being developed to help discover ligands or partners in complex mixtures.4,5 * Corresponding authors. (Aviles) Phone: +34-93-581-1315. E-mail: fxaviles@ einstein.uab.es. (Serrano) Phone: +49-6221-38-7320., E-mail: [email protected]. † These authors contributed equally to this work. ‡ Universitat Auto ` noma de Barcelona. § EMBL. (1) Scott, J. D.; Pawson, T. Sci. Am. 2000, 282, 72-79. (2) Weng, Z.; DeLisi, C. TRENDS Biotechnol. 2002, 20, 29-35. (3) Archer, R. Nat. Biotechnol. 1999, 17, 834. (4) Wahler, D.; Reymond, J. L. Curr. Opin. Chem. Biol. 2001, 5, 152-158. (5) Gavin, A. C. et al. Nature 2002, 10, 415 (6868), 141-147. 10.1021/ac020644k CCC: $25.00 Published on Web 05/31/2003

© 2003 American Chemical Society

The study of noncovalent complexes by mass spectrometry (MS) has seen a great deal of activity over the last 10 years. Much of this work has been performed with electrospray ionization mass spectrometry (ESI-MS), particularly since the introduction of the nanospray variant.6 However, matrix-assisted laser-desorption ionization mass spectrometry (MALDI-MS), generally complemented by time-of-flight analysis (MALDI-TOFMS), has more recently become an outstanding technique for complex biological samples, which are difficult to analyze by ESI.7 The recently introduced instruments ESI-TOF, which allows for the measurement of high masses by ESI mass spectrometry, and MALDI-qTOF, which permits both the simultaneous MALDI peptide mapping and the de novo peptide sequencing inside the mass spectrometer, will play a major role in biomolecular interaction studies.8,9 Up to now, some approaches have been reported to study biomolecule interactions by MALDI-TOFMS. The affinity-based approach takes advantage of an affinity-chromatography selection step prior to the identification of the interacting molecules by mass spectrometry.10-12 Chemical cross-linking has also been used to determine the composition and stoichiometry of protein oligomers.13-15 Proteolysis has been applied to study DNA-protein and protein-protein complexes, allowing for the identification of regions protected against degradation.16,17 Surface plasmon resonance-mass spectrometry has also been applied to detect com(6) Wilm, M.; Mann, M. Anal. Chem. 1996, 68 (8),1-8. (7) Mann, M.; Hendrickson, R. C.; Pandey, A. Annu. Rev. Biochem 2001, 70, 437-473. (8) Krutchinsky, A. N.; Zhang, W.; Chait, B. T. J. Am. Soc. Mass Spectrom. 2000, 6, 493-504. (9) Sobott, F.; Hernandez, H.; McCammon, M. G.; Tito, M. A.; Robinson, C. V. Anal. Chem. 2002, 74, 1402-1407. (10) Hutchens, T. W.; Yip, T. T. Rapid Comun. Mass Spectrom. 1993, 7, 576580. (11) Zhao, Y.; Muir, T. W.; Kent, S. B. H.; Tischer, E.; Scardina, J. M.; Chait, B. Proc. Natl. Acad. Sci. U.S.A. 1996, 93, 4020-4024. (12) Ru ¨ diger, A. H.; Ru ¨ diger, M.; Carl, U. D.; Chakraborty, T.; Roepstorff, P.; Wehland, J. Anal. Biochem. 1999, 275, 162-170. (13) Rappsilber, J.; Siniossoglou, S.; Hurt, E. C.; Mann, M. Anal. Chem. 2000 72, 267-275. (14) Bennett, K. L.; Kussmann, M.; Bjork, P.; Godzwon, M.; Mikkelsen, M.; Sorensen, P.; Roepstorff, P. Protein Sci. 2000, 9, 1503-18. (15) Farmer, T. B.; Caprioli, R. M. Biol. Mass Spectrom. 1991, 20 (12), 796800. (16) Cohen, S. L.; Ferre-D’Amare, A. R.; Burley, S. K.; Chait, B. T. Protein Sci. 1995, 4, 1088-1099. (17) Kriwacki, R. W.; Siuzdak, G. Methods Mol. Biol. 2000, 146, 223-238.

Analytical Chemistry, Vol. 75, No. 14, July 15, 2003 3385

plexes and determine affinity constants between biomolecules.18 Detection of entire noncovalent complexes by MALDI-TOFMS has also been reported.19-23 Finally, in the field of structural immunology, new immunoaffinity assays based on mass spectrometry have successfully been applied to epitope mapping.24-25 In this work, we report a method to detect the formation of complexes involving any kind of biomolecules, but particularly proteins in solution, by MALDI-TOFMS. The formation of a complex is detected through the decrease (fading) of molecularion intensities, as directly compared to the MALDI-TOF mass spectrum of the problem and control mixture following the addition of the target molecule. To show the capability and feasibility of this method, it has been applied to several examples of model interactions, mainly involving small protein inhibitors of proteases (serine proteases, cysteine proteases, and carboxypeptidases) and their corresponding partner enzymes, which were studied at both the qualitative and semiquantitative levels. In addition, it has been used to detect the formation of proteinsmall organic molecule and protein-nucleic acid complexes. Subsequently, the approach was tested on highly complex samples consisting of extracts from invertebrate organisms. Following this, several high-affinity protein ligands for trypsin-like serine proteases (SPs) and metallocarboxypeptidases (CPs) from a sea anemone and from a leech have been identified. The CP-binding molecule detected in sea anemone has been purified and partially characterized, being the first reported from marine organisms. EXPERIMENTAL PROCEDURES Materials. Sinapinic acid (SA) was obtained from Fluka. R-Cyano-4-hydroxycinnamic acid (HCCA) was purchased from Aldrich Chemical Co. Bovine pancreatic trypsin inhibitor (BPTI or aprotinin), trans-epoxysuccinyl-L-leucylamido-(4-guanidino)butane (E-64), antipain, somatostatin, insulin, bovine carboxypeptidase A, and bovine trypsin were purchased from Sigma. Papain and bovine pancreatic ribonuclease A were obtained from Roche. Total RNA was obtained from Escherichia coli by the Ultraspec RNA Kit, Biotecx Laboratory Inc. 3BP2 was obtained as previously described.26 ADA2h (activation domain of human procarboxypeptidase A2) was obtained as previously described.27 Stefin A and the Hirudo medicinalis extract were supplied by the group of Profs. H. Fritz and C. Sommerhoff (Chirurgischen Klinik Innenstadt, Ludwig-Maximilians-Universitat, Munich, Germany). Leech carboxypeptidase inhibitor (LCI), potato carboxypeptidase inhibitor (PCI), and mutant ∆3PCI were obtained as previously described.28,29 Stichodactyla helianthus extract was produced as (18) Nelson, R. W.; Krone, J. R. J. Mol. Recognit. 1999, 12, 77-93. (19) Woods, A. S.; Buchsbaum, J. C.; Worrall, T. A.; Berg, J. M.; Cotter, R. J. Anal. Chem. 1995, 67, 4462-4465. (20) Rosinke, B.; Strupat, K.; Hillenkamp, F.; Rosenbusch, J.; Dencher, N.; Kruger, U.; Galla, H. J. J. J. Mass Spectrom. 1995, 30, 1462-1468. (21) Lecchi, P.; Pannell, L. K. J. Am. Soc. Mass Spectrom. 1995, 6, 972-975. (22) Glocker, M. O.; Bauer, S. H. J.; Kast, J.; Volz, J.; Przybylski, M. J. Mass Spectrom. 1996, 31, 1221-1227. (23) Farmer, T. B.; Caprioli, R. M. J. Mass Spectrom. 1998, 33, 697-704. (24) Papac, D. I.; Hoyes, J.; Tomer, K. B. Protein Sci. 1994, 3, 1485-1492. (25) Kiselar, J. G.; Downard, K. M. Anal. Chem. 1999, 71, 1792-1801. (26) Viguera, A. R.; Arrondo, J. L.; Musacchio, A.; Saraste, M.; Serrano, L. Biochemistry 1994, 33, 10925-33. (27) Villegas, V.; Martinez, J. C.; Aviles, F. X.; Serrano, L. J. Mol. Biol. 1998, 283, 1027-1036. (28) Reverter, D.; Vendrell, J.; Canals, F.; Horstmann, J.; Avile´s, F. X.; Fritz, H.; Sommerhoff, C. P. J. Biol. Chem. 1998, 49, 32927-32933.

3386

Analytical Chemistry, Vol. 75, No. 14, July 15, 2003

previously reported30 and supplied by the group of Prof. M. Angeles Chavez (Department of Biochemistry, Universidad de La Habana, Cuba). Biomolecule Interaction Experiments. (a) General Sample Preparation. Lyophilized samples of LCI, PCI, BPTI, and stefin A were dissolved at molar concentrations of 10, 12, 12, and 15 µM, respectively. Lyophilized samples of RNase A were dissolved in 20 mM Tris-HCl (pH 8.0) to a molar concentration of 10 µM. Antipain was dissolved in water to a molar concentration of 10 µM. E-64 was dissolved in DMSO to 1 mM and later diluted in deionized water to a concentration of 10 µM. Nonbinding controls were dissolved in deionized water at a concentration that gave MALDI-MS ion signals intensities similar to those from the protease inhibitors assayed in parallel. Initially, proteases were diluted in deonized water at 10-15 molar excess concentration related to the counterpart inhibitor concentration. However, in certain cases (i.e., in repeated experiments) smaller ratios (such as 2-5 molar excess) were also analyzed. Both the S. helianthus and the H. medicinalis extracts were dissolved in deionized water at a concentration of 20 mg/mL. These solutions were centrifuged at 8000g for 10 min, and the supernatants were processed by a reversed-phase resin-based protocol to clean and concentrate peptides and small proteins.31 The reversed-phase resin-bound molecules were eluted with 50% 2-propanol, lyophilized, and dissolved in deionized water. (b) Protease-Inhibitor Interaction. A mixture of 0.5-1 µL of protease inhibitor, 0.5 µL of the nonbinding control, and 0.5 µL of deionized water (control sample), or 0.5 µL of protease, was incubated for 3 min at room temperature. In the case of the S. helianthus and H. medicinalis samples, 1 µL of the dissolved final lyophilized solutions (see above) was mixed with 0.5 µL of deionized water (control sample) or with 0.5 µL of carboxypeptidase A or trypsin and incubated for 3 min at room temperature. (c) RNase-RNA Interaction. A mixture of 0.5-1 µL of RNase A, 0.5 µL of the nonbinding control and 0.5 µL of deionized water (control sample), or 0.5 µL of total RNA, was incubated for 3 min at room temperature. MALDI-TOF Mass Spectrometry. All mass spectra were acquired on a Bruker Biflex TOF mass spectrometer equipped with a nitrogen laser with an emission wavelength of 337 nm. Spectra were obtained in the linear mode at an accelerating voltage of 19 KV. Deflection of the low-mass ions was used to enhance the target protein signal. Spectra were acquired by averaging 50100 shots. Our approach relies, in part, on the proper selection of the nonbinding (control) molecules added in the assay and used for measuring the relative decrease in intensity of the molecular ion of the problem molecules. Such control molecules must not interact with the rest of the biomolecules included in the final sample mixture; that is, its intensity signal should not be affected by the addition of the target. This should be checked beforehand, particularly when complex mixtures (i.e., biological extracts) are (29) Marino-Buslje, C.; Venhudova, G.; Molina, M. A.; Oliva, B.; Jorba, X.; Canals, F.; Avile´s, F. X.; Querol, E. Eur. J. Biochem. 2000, 267, 1502-1509. (30) Delfin, J.; Martinez, I.; Antuch, W.; Morera, V.; Gonzalez, Y.; Rodriguez, R.; Marquez, M.; Saroyan, A.; Larionova, N.; Diaz, J.; Padron, G.; Chavez, M. Toxicon 1996, 34, 1367-76. (31) Villanueva, J.; Canals, F.; Querol, E.; Avile´s, F. X. Enzymol. Microb. Technol. 2001, 29, 99-103.

Figure 1. (A) MALDI-TOFMS of a mixture of BPTI and a nonbinding control (PCI) before (upper) and after (lower) mixing with trypsin. The additional peaks in the lower panel correspond to the autolysis of trypsin. (B) MALDI-TOFMS of a mixture of stefin A and a nonbinding control (LCI) before (upper) and after (lower) mixing with papain. The additional peaks in both panels correspond to the heterogeneity of stefin A. (C) MALDI-TOFMS of a mixture of LCI and a nonbinding control (insulin) before (upper) and after (lower) mixing with CPA. Insulin was added as an internal nonbinding control. Sinapinic acid was used as a matrix for the experiments showed in this figure. Table 1. Relative Intensities for Protease Inhibitors and Their Respective Nonbinding Controls before (-) and after (+) Reaction with the Proteasea - protease

+ protease

protease-protease inhibitor pair

nonbinding control

protease inhibitor

nonbinding control

protease inhibitor

LCI-CPA BPTI-trypsin stefin A-papain antipain-papain E-64-papain CPA binding molecule-CPA

0.98 ( 0.10 0.72 ( 0.09 0.70 ( 0.16 0.65 ( 0.19 0.89 ( 0.15 0.90 ( 0.12

0.99 ( 0.12 1.00 1.00 1.00 0.99 ( 0.09 0.99 ( 0.11

1.00 1.00 1.00 1.00 1.00 1.00

0.03 ( 0.00 0.02 ( 0.00 0.26 ( 0.05 0.12 ( 0.02 0.06 ( 0.00 0.00 ( 0.00

a

The number of repetitions for each experiment is 10.

analyzed. However, when a judicious selection of the control molecules is made, they are also useful as internal mass standards. Preparation of Samples for MALDI-TOF Mass Spectrometry. Sample preparation is critical for the success of the methodology. An excess of one component in a mixture may suppress the ion signals of the other components. In addition, impurities in a complex mixture may interfere with analyte detection and reduce sensitivity. Similarly, preparation of the matrix is essential; a MALDI matrix preparation close to neutral pH is frequently very important to obtain satisfactory results. The MALDI matrixes used in this work had a pH of between 5 and 6 (see below). Experiments performed using a matrix containing 0.1% trifluoroacetic acid in many cases perturbed the complexes that we studied here. We also noticed that keeping a low concentration of organic solvent in the matrix solution is essential

to maintain the integrity of our complexes. The MALDI matrixes used in this work contained 30-50% of acetonitrile (which diluted to 22.5-25% after mixing with the aliquots to be analyzed). (a) MALDI-TOF Matrix Preparation. After a series of trials, a 50%/50% (v/v) acetonitrile/deionized water solution at a pH of 5.5 was the initial solvent composition for the matrixes selected to perform most of the experiments reported in this work. They contained 10 mg/mL of sinapic acid or 10 mg/mL of HCCA in the above matrix solvent. The spectral signals of entire complexes were detected by the addition of 1 M amonium citrate to the matrix preparation (50%/50%, v/v, acetonitrile/1 M ammonium citrate solution, at pH 6).19 To analyze samples from biological extracts with a high number of compounds (such as those from S. helianthus and H. medicinalis), a matrix containing only 30% of acetonitrile was used (30%/70%, v/v, acetonitrile/deionized water solution, at a pH of 5.5), to keep the organic solvent concentration low enough when mixed with the biological sample, at a 1/3 sample/matrix (v/v) ratio (see below). (b) Sample Preparation Methods. Aliquots from the proteinligand interaction model assays and from the assays to detect the entire complexes were successfully studied by using the standard dried-droplet MALDI sample preparation method.32 On the other hand, aliquots from the biological extracts (S. helianthus and H. medicinalis) were analyzed using a modified version of a reported sample preparation method.32 According to this method, 0.5 µL of the matrix is deposited onto the target and allowed to dry. After this, 0.5 µL of the sample/matrix (1/3, v/v) mixture is added to the preformed matrix layer. Shortly after “crystallization” begins (32) Hillenkamp, F.; Karas, M. Anal. Chem. 1988, 60, 2299-2301.

Analytical Chemistry, Vol. 75, No. 14, July 15, 2003

3387

Figure 2. (A) MALDI-TOFMS of a mixture of the inhibitors PCI, ∆3PCI, and a nonbinding control (insulin) before (top of the panel) and after the addition of CPA from enzyme stock solutions at decreasing dilutions of 1/1, 1/20, 1/80, and 1/160. At the highest concentration, the enzyme was at 10-15 molar ratio in excess over the inhibitors. (B) Plot representing the RI of both wt PCI and ∆3PCI molecules (corrected with the RI of the control) after the addition of CPA at different dilutions. Sinapinic acid was used as a matrix for the experiments showed in this figure.

(1-2 min at room temperature), 2 µL of deionized water is added to the sample/matrix mixture, giving rise to a droplet. After 5-10 s, the droplet is pipetted off, and 0.5 µL of matrix is added and allowed to dry. In this work, the mixture sample/matrix 1/3 (v/ v) gave the best statistical results to analyze the complex mixtures described herein. RESULTS Basis for the Detection of Noncovalent Complexes by MALDI-TOFMS. The method described here is based on the analysis of MALDI-TOFMS ion intensities to detect the occurrence of complexes between proteins and biomolecules. One of the 3388

Analytical Chemistry, Vol. 75, No. 14, July 15, 2003

binding counterparts (the problem molecule) is mixed with a nonbinding molecule (the control) in such a way that both give a MALDI-MS spectrum of similar intensity and mass. The complex is detected by a direct comparison of the mass spectra of the mixture of problem and control molecules with and without the addition of a target molecule (a protein). Differences between the two spectra are distinguished by a visual comparison of ion intensities. A relative decrease (fading) of the ion intensity of the problem (binding) molecule, or its disappearance, with respect to the control (nonbinding) molecule can be attributed to complex formation between the former and the added target. The remain-

Figure 3. (A) MALDI-TOFMS of a mixture of antipain and a nonbinding control (somatostatin) before (upper) and after (lower) addition of papain and (B) MALDI-TOFMS of a mixture of E-64 and a nonbinding control (3BP2) before (upper) and after (lower) mixing with papain. The additional peaks correspond to adducts produced by the matrix. R-Cyano-4-hydroxycinnamic acid was used as a matrix for the experiments showed in this figure.

ing ion signals observed are relatively unaffected by the presence or absence of such a target molecule if this is properly selected. Differences can be quantified by measuring the relative intensities (RIs) before and after the mixture (Table 1). When using the current MALDI-TOF matrix and sample preparations that preserve complex formation19-23 (such as the ones described in the Experimental Procedures section of this paper), frequently, the MS signal corresponding to the entire complex is hardly detected unless salt is added to the preparation and assayed at different concentrations (data not shown). However, the peaks corresponding to the entire complexes observed under such conditions are small, and their RIs show large standard deviations among independent experiments. Conversely, the decrease in the RI of one of the molecules when its partner is present is consistent and reliably measured, showing small standard deviations (see Table 1). Therefore, we decided to monitor complex formation along this work by using the intensity fade in the RI of one of the molecules forming the noncovalent complex instead of measuring the RI of the complex by itself. Detection of Protein-Protein Interactions in Model Mixtures. To optimize the methodology, a series of trials was initially performed. Protease-protein inhibitor interactions were chosen as models to set up this method. Three protease inhibitors (BPTI, stefin A, and LCI) targeting three different classes of proteases, serine proteases, cysteine proteases, and metallocarboxypeptidases, respectively, were chosen as models. The three ligands are tight-binding inhibitors with inhibition constants (Ki) between 10-9 and 10-13 M.28,34 One nonbinding control (a protein of similar molecular mass) was added to each of the analyzed samples as

an ion intensity control (also being used as an internal standard). Figure 1 shows the MALDI-TOF mass spectra corresponding to three different analyses for complex formation. The figure shows each of the problem molecules (a protease inhibitor) before and after the addition of the target molecule (a protease). In the three examples, the molecular-ion signal of the binding molecule is greatly decreased after the addition of the target molecule (see also Table 1). Detection of Protein-Organic Molecule Interaction. To show that our approach is capable of detecting complex formation between proteins and nonpeptidic organic molecules, two small organic inhibitors for cysteine proteases, antipain and E-64, were assayed against papain, a target cysteine protease. Figure 3A shows the MALDI-TOF mass spectrum of antipain before and after the addition of papain. The signal intensity of the inhibitor in the mass spectrum is greatly diminished by the presence of the protease (see Table 1). It should be mentioned that the signal fading of this organic inhibitor is reversed by addition (in excess) of the protein inhibitor stefin A, a competitive inhibitor. A fading effect is also observed after the reaction of E-64 with the same enzyme, as shown in Figure 3B and Table 1. However, in this case, the interpretation should be different given the known irreversible binding of this inhibitor to the enzyme. Detection of Protein-Nucleic Acid Interaction. Proteins that bind specific sites on nucleic acids have been a challenging and (33) Kussmann, M.; Nordhoff, E.; Rahbek-Nielsen, H.; Haebel, S.; Rossel-Larsen, M.; Jakobsen, L. J. Mass Spectrom. 1997, 32, 593-601. (34) Aviles, F. X. Innovations in Proteases and their Inhibitors; Walter de Gruyter: Berlin, 1993.

Analytical Chemistry, Vol. 75, No. 14, July 15, 2003

3389

Figure 4. MALDI-TOFMS of a mixture of RNase A and a nonbinding control (ADA2h) before (upper) and after (lower) addition of total RNA. Sinapinic acid was used as a matrix for the experiments showed in this figure.

interesting problem since the earliest days of molecular biology. The detection of these complexes could allow for the rapid identification of proteins involved in many cellular processes. We have selected the ribonuclease A-RNA interaction as a proteinnucleic acid complex model. Figure 4 shows the MALDI-TOF mass spectra of bovine pancreatic ribonuclease A (RNase A) before and after reaction with total E. coli RNA. The RNase A signal intensity in the mass spectrum is greatly diminished by the complex formation with RNA. The fact that the RNA used was very heterogeneous did not affect either the detection of the control molecule signal or the fading of the RNase A signal. These results might open a new way to study nucleic acid interactions as a result of its ease and speed and that it only requires a very small amount of sample. Determination of the Affinities between the Binding Molecules. Experiments were performed to analyze whether our approach could provide information about the relative affinities between different variants of the binding molecules and their target molecule and whether it could give rise to a quantitative procedure later on. With this aim, the wild-type form of the carboxypeptidase A inhibitor from potatoes (PCI) (Ki ) 1.5 × 10-9 M) and a site-directed mutant of this inhibitor (∆3PCI) (Ki ) 40 × 10-6 M)29 were mixed simultaneosly and at the same concentra3390

Analytical Chemistry, Vol. 75, No. 14, July 15, 2003

tion (50 µg/mL) with the target molecule (carboxypeptidase A, CPA). The reaction conditions were identical to those described for the experiments of the previous section (see the Experimental Procedures section), except that different relative concentrations of CPA were assayed, in a decreasing trend. That is, increasing dilutions on the enzyme stock solution were used. Figure 2A shows the MALDI-TOF mass spectra of such mixtures. Ion intensity differences between the inhibitors at decreasing concentrations of CPA can be correlated with their different binding affinities experimentally obtained by classic enzymologic methods.29 In this case, it is clear that PCI is a better ligand than ∆3PCI. When using a high CPA relative concentration, such as the one from a nondiluted stock, corresponding to a 10-15 molar excess of enzyme with respect to the inhibitors, both wt PCI and ∆3PCI molecule signals nearly disappear from the spectrum of the final reaction mixture (Figure 2A). When the concentration of CPA is gradually decreased, the signal of wt PCI remains faded, while the signal of ∆3PCI, with lower affinity for CPA, is gradually increased (Figure 2A). Figure 2B shows a plot representing the RI of both wt PCI and ∆3PCI molecules (corrected with the RI of the control) versus decreasing concentrations of CPA. This plot clearly shows how the RIs of the two molecules are affected differently by the addition of CPA at different dilutions. Detection of Protein Ligands in Complex Samples. Once the methodology was set up with model molecules, the next step was to apply it to complex samples. With this aim, highly heterogeneous extracts from invertebrate organisms were analyzed by MALDI-TOFMS to detect protease inhibitors, assaying the effect of the addition of different proteases on their mass spectra. Two kinds of invertebrate organisms that have been discovered to be an important source of bioactive molecules were used for such a purpose: the sea anemone S. helianthus and the aquatic leech H. medicinalis. Protease inhibitors, cytolysins, cardiotoxins, and neurotoxins have recently been isolated from these organisms.30,35-36 Figure 5 shows the MALDI-TOF mass spectra corresponding to the whole body extract of the sea anemone S. helianthus before and after the addition of bovine trypsin. After reaction with trypsin, one signal of the MALDI-TOF mass spectrum, with a molecular mass of 6110 m/z, has been assigned as the SphI-1 Kunitz protease inhibitor, previously reported by Delfin et al.30 Isolation of this species by HPLC, followed by sequencing of its nine N-terminal residues by Edman degradation, confirmed this identification. Figure 5B shows a plot of the RIs of some of the molecular ions present in the MALDI-TOF mass spectrum of the extract before and after reaction with trypsin (data of 10 fully independent experiments were used to draw the plot). It is seen that only the molecular ion with a m/z of 6110 is statistically affected by the addition of trypsin. On the other hand, the addition of bovine carboxypeptidase A instead of trypsin to the extract gave rise to the fading of another spectral signal with m/z of 3484, whose identification and characterization as a CPA ligand is described in the next subsection. Figure 6 shows the MALDI-TOF mass spectra corresponding to an extract from the digestive system of the aquatic leech H. (35) Lanio, M. E.; Morera, V.; Alvarez, C.; Tejuca, M.; Gomez, T.; Pazos, F.; Besada, V.; Martinez, D.; Huerta, V.; Padron, G.; de los Angeles Chavez, M. Toxicon 2001, 39, 187-94. (36) Schoofs, L.; Salzet, M. Curr. Pharm. Des. 2000, 8, 125-133.

Figure 5. (A) MALDI-TOFMS of a S. helianthus extract before (upper) and after (lower) mixing with trypsin. An additional peak at 4546 m/z appears after mixing as a result of the autolysis of trypsin. (B) Plot of the RI of the most prominent molecular ions present in the mass spectrum of the S. helianthus extract before and after reaction with trypsin. Ten independent experiments were used to draw the plot. Sinapinic acid was used as a matrix for the experiments shown in this figure.

medicinalis before and after the addition of bovine CPA. After the reaction with CPA, one peak of the MALDI-TOF mass spectrum with a m/z of 7326 clearly fades. It has been identified as the leech carboxypeptidase inhibitor previously reported by our group.28,37 Figure 6B shows a plot of the changes in RIs of some of the molecular ions present in the mass spectrum of the same extract before and after reaction with CPA (data of 10 independent experiments were used to draw the plot). The results demonstrate that only the molecular ion with a m/z of 7326 is statistically affected by the addition of CPA and is detected by our procedure. It is worth mentioning that when bovine trypsin is added to the H. medicinalis extract instead of CPA, 11 peaks of the MALDITOF mass spectra clearly fade. This is indicative, in agreement with other reports, of the existence of a large number of molecules with capability to bind to trypsin in the leech.36 Work is in progress now to isolate and characterize such molecules. To eliminate the possibility that the disappearance of spectral signals is due to the proteolytic action of the added proteases, competition experiments were performed in all the above cases. The final addition of either BPTI or PCI, for the trypsin- and (37) Reverter, D.; Fernandez-Catalan, C.; Baumgartner, R.; Pfander, R.; Huber, R.; Bode, W.; Vendrell, J.; Holak, T. A.; Aviles, F. X. Nat. Struct. Biol. 2000, 7, 322-328.

carboxypeptidase-based experiments, respectively, gave rise to the recovery of the faded signals. Preliminary Characterization of a New CPA-Binding Molecule. To assay the simplicity and consistency of our method for deep characterization of protein ligands detected by the IF approach and because of its potential novelty, we concentrated our attention on the CPA ligand found in the S. helianthus extract (see the previous subsection). Figure 7 displays the mass spectrum of the same extract as in Figure 5, but showing now the effects of the addition of bovine CPA. In this case, a molecular ion with a mass of 3484 Da is absent from the mass spectrum of the sample after the addition of the enzyme. The recovery of its spectral signal in a competition assay with PCI added in excess evidenced that such a signal corresponds to a true CPA-binding molecule (Figure 7A, lower), probably interacting at a similar site of the enzyme as PCI (the active site). Figure 7B represents the results of repeated assays, showing that only the molecular ion with a m/z of 3484 is statistically affected by the addition of CPA. Only a few proteic inhibitors for metallocarboxypeptidases in living organisms have been reported up to now, and none from marine organisms.38 The homologies among the few inhibitors (38) Vendrell, J.; Querol, E.; Aviles, F. X. Biochim. Biophys. Acta 2000, 1477, 284-298.

Analytical Chemistry, Vol. 75, No. 14, July 15, 2003

3391

Figure 6. (A) MALDI-TOFMS of a H. medicinalis extract before (upper) and after (lower) mixing with carboxypeptidase A. (B) Plot of the RI of some of the molecular ions present in the mass spectrum of the H. medicinalis extract before and after reaction with CPA. Ten independent experiments were used to draw the plot. Sinapinic acid was used as a matrix for the experiments showed in this figure.

that have been reported are low, but they are fairly unspecific against different pancreatic-like metallocarboxypeptidases and share homology at their C terminus, probably because this part of the molecule is generally involved in interactions with the enzyme.37,38 To isolate and characterize the above-mentioned CPA binding molecule identified in the S. helianthus sea anemone, a lyophilized extract of this species was submitted to two fractionation steps by reversed-phase (RP) HPLC. The molecule showing a molecular mass of 3484 Da was isolated. Analysis of this molecule (after reductive carboxymethylation of cysteine residues) at both its C-terminal sequence (5 residues, CY-I/L-SF), by time-controlled digestion with carboxypeptidase Y followed by MALDI-TOFMS,39 and its N-terminal sequence (16 residues, WVGNGGRCSSSLDCCK) by automatic Edman degradation indicated that it shares no homology either with any of the reported metallo-carboxypeptidase inhibitors38 or with any other sequence in the databases. Furthermore, it seems to contain four cysteines involved in disulfide bridges, as deduced from nondegradative reduction-alkylation experiments monitored by MALDI-TOF mass spectrometry. The presence of disulfide bridges is a common pattern in protease inhibitors of the carboxypeptidase family (PCI has three, and LCI, four). To verify the capability of such a molecule to give rise to the detected interaction with CPA in the extract of the whole organism, binding (39) Patterson, D. H.; Tarr, G. E.; Regnier, F. E.; Martin, S. A. Anal. Chem. 1995, 67, 3971-3978.

3392 Analytical Chemistry, Vol. 75, No. 14, July 15, 2003

experiments were repeated with the purified molecule. The MALDI-TOF mass spectra, shown in Figure 8, indicate that a clear depletion of the ion peak of the molecule is produced when the target enzyme is added; the peak reappears after addition of an excess of PCI. Therefore, the isolated molecule forms a stable complex with CPA both in a whole extract from the sea anemone and in the purified state. Enzymological analysis of the purified molecule using small chromogenic substrates28 failed to show a capability to inhibit bovine CPA. However, additional experiments using large substrates showed positive results, indicating that this molecule might be a nonclassical carboxypeptidase inhibitor (results not shown). Work is in progress to further characterize the new molecule. DISCUSSION Biomolecular interactions in general, and those involving proteins in particular, generate great interest both in basic and applied research. Different approaches have been described to increase throughput in the search for protein-protein and protein-molecule interactions, of which a recent example is the development of a protein chip containing the yeast proteome.40 In the present work, a method based on the use of MALDITOFMS ion intensities to detect the formation of complexes (40) Zhu, H.; Bilgin, M.; Bangham, R.; Hall, D.; Casamayor, A.; Bertone, P.; Lan, N.; Jansen, R.; Bidlingmaier, S.; Houfek, T.; Mitchell, T.; Miller, P.; Dean, R. A.; Gerstein, M.; Snyder, M. Science 2001, 293, 2101-2105.

Figure 7. (A) MALDI-TOFMS of a S. helianthus extract before (upper) and after (central) the addition of carboxypeptidase A (CPA). The effect of competition with the potato inhibitor of carboxypeptidase A (PCI), added in excess, is shown (lower panel). (B) Plot of the RI of some of the molecular ions present in the mass spectrum of the S. helianthus extract before and after reaction with CPA. Ten independent experiments were used to draw the plot. Sinapinic acid was used as a matrix for the experiments showed in this figure.

between proteins and other ligands is reported. The simplicity and rapidity of this approach, together with its high operational sensitivity, should prove a valuable tool not only in molecular and cellular biology fields, but also in drug research. We think that the approach presented here can be of general use if complementary steps for sample preparation and analysis are taken (see the Experimental Procedures section for details). Although here we only show that the method is able to quickly and easily find protein ligands of proteic, organic or nucleotidic nature in simple and in complex biological mixtures, it might have a much wider applicability. In principle, it should be applicable to studying interactions established between any kind of “problem” biomolecules, provided that their complexes are stable in the conditions used for MALDI-TOF sample preparation. Two further requirements for a successful mass spectrometry analysis by our procedure are that the assayed molecules behave well in a MALDITOF mass spectrometer (i.e., high efficiency of desorption and transfer to the gas phase) and that the intensities of their molecular ions are affected by the addition of the binding counterpart. Given that the observed effect is a decrease/fading in the intensity of the signal of one of the molecules that forms the complex after the addition of the other binding molecule, we

have named this approach intensity-fading MALDI-TOFMS (IFMALDI-TOFMS). An important issue in this screening approach is that some peaks not related to the ligand may suffer changes in their intensities after the addition of the target molecule. For instance, combining large amounts of protease (the target) with a sample with many proteic compounds may alter some of the RIs of some of them by proteolytic degradation. In these cases, controls are required to differentiate a substrate from an inhibitor, as well as to identify protease degradation effects. This is shown in the peak with m/z 4546 in Figure 5A, which appeared after the addition of trypsin: by performing a control digestion without sample, it was identified as an autolysis product of the same protease. Sample artifacts, derived from the addition of the target, can be discarded by performing parallel reactions. Similarly, in case of doubt, a series of spectra at different target concentrations or assaying different pHs should be recorded to determine whether a peak is statistically affected by the presence of the target. Nevertheless, a key experiment to test the specificity of an interaction is by observing a reversal of fading within a competition assay with distinct known partners, a procedure that also helps to discard other perturbation effects. Analytical Chemistry, Vol. 75, No. 14, July 15, 2003

3393

measuring spectral signals at different ratios between the partners) once the rigorous quantification of the observed effects is well controlled. More research should be done on this issue to explore the best conditions for such a quantification. By now, we think that our approach is more reliable at the qualititative and semiquantitative levels. On the other hand, the continuous improvements in spectral resolution in MALDI-TOF mass spectrometers in the mass region required to visualize proteins is making it increasingly feasible to focus the analysis simultaneously on the small- and medium-sized ligands (i.e., m/z below 10 000, such as in this work) and on the target proteins of medium or large size. In the present work, the detected ligands (i.e., protease inhibitors from whole organism extracts) have been characterized through their isolation in parallel HPLC experiments because a simple MALDI-TOF mass spectrometer is used. However, nowadays it should be feasible to perform such a characterization in a continuous way using additional stages of fragmentation of the selected molecular ion of the ligand followed by the analysis of the fragmentation pattern (i.e., MS/MS or MSn variants). In addition, it should be feasible that differential screening of very complex biological samples of various sources (tissues or fluids, along the development, normal and pathogenic, from different organisms,) can be performed in a high-throughput way, following the current approaches of Proteomics. Work is in progress in our laboratory to attain such goals.

Figure 8. MALDI-TOFMS of a mixture of the CPA-binding molecule isolated from S. helianthus and a nonbinding control (insulin) (A) before and (B) after addition of carboxypeptidase A (CPA). Sinapinic acid was used as a matrix for the experiments showed in this figure.

An alternative approach also checked in this work involves the use of irreversibly inactivated proteolytic enzymes, which still maintain the capability to bind specific inhibitors, as interacting (fading) partners: i.e., anhydrotrypsin, carboxymethyl papain, and active-site mutants of CPA (such as Glu270Gln) (Villanueva et al., unpublished results). In this way, the danger of severe proteolytic effects on the proteinaceous species of the analyzed sample is minimized, widening the applicability of the approach to labile samples. Work is in progress to further develop this variant. Some interesting studies on noncovalent interactions by direct MALDI-TOF mass spectrometry have been reported.19-23 Most of them are based on the detection of the entire complexes using modifications of the MALDI-TOF standard method, such as adding salt to the matrix preparation.19 However, even with such modifications, the ion signal corresponding to the complex frequently is small, probably because the complex does not ionize properly. We have noticed such problems in our experiments, particularly when working with samples that have many components (i.e., whole extracts from organisms) in which the decrease of resolution caused by the addition of salt is not desirable. In general, we have found it advantageous to concentrate the analysis more on the fading of one of the partners when the other one (the target) is added than on the complex by itself. Our procedure should also facilitate the derivation of binding affinities (by 3394 Analytical Chemistry, Vol. 75, No. 14, July 15, 2003

CONCLUSIONS A novel MALDI-TOFMS method to perform quick and easy screening for the detection of protein ligands in simple and complex biological mixtures has been shown. The main advantages of this approach are (a) the use of minimal amounts of target protein to perform the screening; (b) the simultaneous obtainment of the masses of the binding compounds; (c) the wide mass range allowed for the binding molecules and the unlimited mass for the target molecule; (d) the capability to detect either the target or the ligand molecules, depending on the biological problem; (e) the detection of the complex in solution, discarding artifacts derived from immobilization; (f) the capability to screen complex samples; (g) the low cost and rapid application of the approach; and finally, (h) its potential to be integrated into high-throughput approaches. Thus, this method would add a tool of great power to the current approaches of Proteomics. Abbreviations: ESI-MS, Electrospray ionization mass spectrometry; MALDI-TOFMS, Matrix assisted laser-desorption ionization time-of-flight mass spectrometry; HCCA, a-cyano-4-hydroxycinnamic acid; RI, relative intensity; SP, serine protease; CP, carboxypeptidase; CPA, carboxypeptidase A; ADA2h, activation domain of human procarboxypeptidase A2; BPTI, bovine pancreatic trypsin inhibitor or aprotinin; PCI, carboxypeptidase inhibitor from potato; LCI, carboxypeptidase inhibitor from leech. ACKNOWLEDGMENT We are indebted to Prof. M. Angeles Chavez (University de La Habana, Cuba) for useful critical discussions about and for providing us with the sea anemone extract, and to Profs. H. Fritz and C. Sommerhoff (Chirurgical Clinic, Munich, Germany) for providing us with the leech extract. We thank F. Canals for helpful

discussions. This work has been supported by Grant BIO20012046 from MCYT (Ministerio de Ciencia y Tecnologı´a, Spain) and by the Centre de Refere`ncia en Biotecnologia de la Generalitat de Catalunya.

Received for review October 16, 2002. Accepted April 10, 2003. AC020644K

Analytical Chemistry, Vol. 75, No. 14, July 15, 2003

3395