Anal. Chem. 2007, 79, 1251-1255
Technical Notes
Mass Spectrometry-Assisted Protease Substrate Screening Hartmut Schlu 1 ter,* Jana Rykl, Joachim Thiemann, Sandra Kurzawski, Johan Gobom, Martin Tepel, Walter Zidek, and Michael Linscheid
Med. Clinic IV, Nephrology, Campus Benjamin Franklin, Charite´-University Medicine Berlin, Hindenburgdamm 30, 12200 Berlin, Germany
Since sequencing of the human genome was completed, more than 500 genes have been annotated as proteases. Exploring the physiological role of each protease requires the identification of their natural substrates. However, the endogenous substrates of many of the human proteases are as yet unknown. Here we describe a new assay that addresses this problem. The assay, which easily can be automated, is based on the incubation of immobilized protein fractions, which may contain the natural substrate, with a defined protease. After concentrating the proteolytically released peptides by reversed-phase chromatography they are analyzed by tandem mass spectrometry and the substrates identified by database searching. The proof of principle in this study is demonstrated by incubating immobilized human plasma proteins with thrombin and by identifying by tandem mass spectrometry the fibrinopeptides, released by the action of thrombin from their natural substrate fibrinogen, in the reaction mixture. Proteases are enzymes that are pivotal regulators of biological processes during the life span of all organisms, including conception, birth, growth, maturation, aging, disease, and death. Proteases control activation, synthesis, and turnover of proteins. They regulate physiological processes such as differentiation and intraand extracellular signaling. By combining data from the MEROPS, InterPro, and Ensembl databases as well as own experimental data, Puente et al.1 annotated a total of 553 genes that encode proteases or protease homologues in the human genome. Approximately 14% of the documented human proteases are reported to be under investigation as drug targets. Identification of new protease substrates is important for better understanding the biological role of proteases. Hence, substrate discovery may provide the next generation of drug targets for multiple therapeutic areas.2 Screening for natural substrates is usually performed by using either potential surrogate substrates, which can be hydrolyzed by the enzyme in vitro, or endogenous proteins, shown to be cleaved in * To whom correspondence should be addressed. Tel: 0049-(0)30-450-514891. Fax: 0049-(0)30-450-514-991.
[email protected]. (1) Puente, X. S.; Sanchez, L. M.; Overall, C. M.; Lopez-Otin, C. Nat. Rev. Genet. 2003, 4, 544-558. (2) Southan, C. Drug Discovery Today 2001, 6, 681-688. 10.1021/ac061482l CCC: $37.00 Published on Web 01/03/2007
© 2007 American Chemical Society
vivo. Combinatorial peptide library screening3 or phage display screening4,5 are utilized for determining surrogate substrates. The number of candidate substrates might be narrowed down in silico, if cleavage sites are identified. However, these cleavage site sequences cannot be searched with any precision. In addition, identification of an endogenous substrate by screening orphan enzymes without further information about the substrate targets is difficult for the following reasons: (a) many proteases show broad specificities; (b) substrate specificity can be derived from the colocalization of active protease and its substrate in particular compartments or developmental stages; (c) activation pathways might be involved, in which a post-translational modification of the substrate is necessary for being recognizable by the protease. Therefore, to identify substrates of orphan proteases detailed biochemical experiments are mandatory. Mass spectrometry (MS) based on soft ionization techniques is sensitive, rapid, and semiquantitative for the analysis of peptides and proteins; it is well suited for automation and high-throughput investigations. MALDI-MS is particularly attractive for the analysis of peptidase and protease reactions.6 We developed an automated MS-based strategy (MES) to screen protein fractions for defined enzymatic activities.6,7 The method utilizes the conversion of proteins into protein libraries by covalently immobilizing the proteins to chromatography materials. By incubating individual immobilized proteins from the library, enzymatic activities are detected with reaction-specific probes. Aliquots of the reaction mixtures, defined by their incubation times, are analyzed with MALDI-MS. The mass signals in the spectra from the expected reaction products indicate the presence of the target enzyme in the investigated fraction of the protein library. (3) Zhu, Q.; Uttamchandani, M.; Li, D.; Lesaicherre, M. L.; Yao, S. Q. Org. Lett. 2003, 5, 1257-1260. (4) Turk, B. E.; Huang, L. L.; Piro, E. T.; Cantley, L. C. Nat. Biotechnol. 2001, 19, 661-667. (5) Rosse, G.; Kueng, E.; Page, M. G.; Schauer-Vukasinovic, V.; Giller, T.; Lahm, H. W.; Hunziker, P.; Schlatter, D. J. Comb. Chem. 2000, 2, 461-466. (6) Schluter, H.; Jankowski, J.; Rykl, J.; Thiemann, J.; Belgardt, S.; Zidek, W.; Wittmann, B.; Pohl, T. Anal. Bioanal. Chem. 2003, 377, 1102-1107. (7) Jankowski, J.; Stephan, N.; Knobloch, M.; Fischer, S.; Schmaltz, D.; Zidek, W.; Schluter, H. Anal. Biochem. 2001, 290, 324-329.
Analytical Chemistry, Vol. 79, No. 3, February 1, 2007 1251
Figure 1. Principle of the MAPS system. Proteins from any protein fraction (A) are covalently immobilized to cyanogen bromide-activated Sepharose beads (A f B). The immobilized proteins (B) are washed several times with pure water to remove any molecules (B f C), which are not immobilized (F). A defined protease is added to the mixture of the immobilized proteins and pure water (C f D). After defined incubation times (D f E), an aliquot of the reaction mixture is desalted by reversed-phase material filled in tips and eluted to a MALDI target (E f F). The reaction products are analyzed by MALDI mass spectrometry (F). The target substrate is present, if a signal of the cleaved protein fragment (here A-B-C) appears in the mass spectrum.
In this study, we demonstrate that mass spectrometry-based detection of enzymatic reactions is also suitable for screening for natural substrates of defined proteases. EXPERIMENTAL SECTION Preparation of Plasma. Peripheral blood (4 mL) was obtained from healthy volunteers by catheterization of the cubital vein and was collected in tubes containing EDTA (7.2 mg). The blood samples were centrifuged at 2100g for 10 min at 4 °C for isolation of plasma (2.1 mL). Mass Spectrometry-Assisted Protease Substrate Screening (MAPS) Procedure. The steps of the MAPS procedure are summarized in Figure 1 and consists of the following steps: Immobilization of Protein Fractions. Human fibrinogen in excess (5 mg) (Sigma, Deisenhofen, Germany) and proteins present in 20 µL of human plasma (∼1 mg) were covalently immobilized to 30 µL of CNBr-activated Sepharose 6 MB (GE Healthcare, Freiburg, Germany) according to the instruction of the manufacturer. As negative control, the reaction buffer in the absence of any protein passed the immobilization procedure. Incubation of the Immobilized Proteins with the Protease. Five units of human protease thrombin (Sigma) in 150 µL of 20 mM MES, 0.9% w/v NaCl were added to each immobilized protein fraction, and the reaction solution was incubated on a shaker at 37 °C for 2 h. MALDI-MS Analysis of the Proteolysis Products. The 10µL aliquots were removed from the incubation mixture and 1 µL of 1% v/v trifluoroacetic acid (Sigma) was added to each aliquot. The reaction products were concentrated with reversed-phase 1252 Analytical Chemistry, Vol. 79, No. 3, February 1, 2007
Figure 2. TOF-MALDI-MS spectrum of peptides derived from purified fibrinogen (A) and immobilized plasma proteins (B) after 2 h of incubation with thrombin.
chromatography material containing tips (OMIX C18, Varian, Darmstadt, Germany) according to the instruction of the manufacturer. Each eluate (5 µL) from the reversedphase tips was mixed with 5 µL of the MALDI matrix (40 mg /mL R-cyano-4-hydroxy-cinnamic acid in 50% acetonitrile/0.1% trifluoroacetic acid) and applied to a MALDI target (AnchorChip, Bruker Daltonics, Bremen, Germany). The samples were analyzed on a MALDI-TOF mass spectrometer (Reflex III, Bruker Daltonics, Bremen, Germany) in positive ion reflex mode. One hundred single spectra of each sample were summed.
Figure 3. TOF-TOF-MALDI-MS of peptides derived from immobilized plasma proteins after 2 h of incubation with thrombin. (A) Spectrum of the parent ion with m/z 1350 (Figure 2). The amino acid sequence SGEGDFLAEGGGVR is part (22-35) of the fibrinogen R/R-E chain precursor (P02671). (B) Spectrum of the parent ion with m/z 1384 (Figure 2). The amino acid sequence VNDNEEGFFSAR is part (33-44) of the fibrinogen β-chain precursor (P02675). (C) Spectrum of the parent ion with m/z 1466 (Figure 2). The amino acid sequence DSGEGDFLAEGGGVR (des-alanine-(1)-fibrinopeptide-A) is part (21-35) of the fibrinogen R/R-E chain precursor (P02671). (D) Spectrum of the parent ion with m/z 1553 (Figure 2). The amino acid sequence QGVNDNEEGFFSAR (fibrinopeptide-B) is part (31-44) of the fibrinogen β-chain precursor (P02675). (E) Spectrum of the parent ion with m/z 1537 (Figure 2). The amino acid sequence ADSGEGDFLAEGGGVR (fibrinopeptide-A) is part (20-35) of the fibrinogen R/R-E chain precursor (P02671). (F) Spectrum of the parent ion with m/z 1617 (Figure 2). The amino acid sequence ADSpGEGDFLAEGGGVR (fibrinopeptide-A phosphorylated at serine-(3)) is part (20-35) of the fibrinogen R/R-E chain precursor (P02671).
MALDI-MS/MS Analysis of Selected Peptides. Three microliters of each reaction mixture was acidified by addition of 2.5 µL of trifluoroacetic acid and prepared for MALDI using microcolumns packed with reversed-phase media (Poros R2, Applied Biosystems, Weiterstadt, Germany) as previously described.8 Samples were prepared both using cyano-4-hydroxycinnamic acid and 2,5-dihydroxbenzoic acid as MALDI matrixes. Mass spectra of positively charged ions were recorded on an Ultraflex LIFT MALDI TOF/TOF mass spectrometer (Bruker Daltonics), equipped with a Nd:YAG laser. Calibration constants were determined using peptide standards prepared (8) Gobom, J.; Nordhoff, E.; Mirgorodskaya, E.; Ekman, R.; Roepstorff, P. J. Mass Spectrom. 1999, 34, 105-116.
on adjacent target positions. Recorded mass spectra were the sum of 1000 single-shot spectra. MS/MS analysis of selected peptides was performed on an Ultraflex LIFT MALDI-TOF/TOF instrument (Bruker Daltonics) operated in the LIFT mode. Recorded MS/MS spectra were the sum of 3000-4000 single-shot spectra. Protein identification was performed using the Mascot search engine (Matrixscience), searching the SwissProt database with the taxonomy selection restricted to Homo sapiens. Posttranslational modified peptides were assigned using the software BioTools 3.0 (Bruker Daltonics). RESULTS AND DISCUSSION The procedure for MAPS in complex protein fractions comprises six steps (Figure 1). Proteins present in complex protein Analytical Chemistry, Vol. 79, No. 3, February 1, 2007
1253
fractions such as raw extracts are covalently immobilized to solid supports such as cyanogen bromide-activated sepharose beads (step 1, A f B in Figure 1). The immobilization and subsequent washing step reduces the chemical background in the spectra by removing all nonbinding biomolecules (step 2, B f C in Figure 1), thus increasing the detection sensitivity for the proteolytic reaction products and simplifying the interpretation of the mass spectra. Furthermore, immobilization stops autoproteolytic activities present in complex protein fractions. In the third step, the defined protease is added (step 3, C f D in Figure 1) and immobilized proteins are incubated with a defined protease (step 4, D f E in Figure 1). In this study, unfractionated human plasma proteins were immobilized to beads, incubated with thrombin, and analyzed according to the described procedure. In addition, fibrinogen was immobilized, incubated with thrombin, and included as a positive control, and beads without any immobilized proteins, incubated with thrombin, were included as a negative control. The peptides generated by proteolysis were concentrated by adsorbing them to reversed-phase material and eluting them onto a MALDI target (step 5, E f F in Figure 1). This step removes buffer salts and the protease from the peptides. The reaction products of the incubation of the immobilized proteins with the protease were subsequently analyzed by MALDI-MS (step 6, F in Figure 1). If the natural substrate of the protease is present, one or more peptides are released by the activity of the protease from the substrate. The identity of the peptides is established by MALDI MS/MS analysis and database searching. Figure 2 shows the mass spectra of the reaction products obtained by incubating immobilized purified fibrinogen as control (Figure 2A) and immobilized unfractionated human plasma proteins (Figure 2B) with thrombin. The signals in both spectra A and B of Figure 2 are nearly identical. This finding shows that signals from a complex mixture of plasma proteins (Figure 2) did not interfere with those from the reaction products of thrombin. Fibrinogen is a soluble dimeric plasma protein, each half of which is composed of disulfide-bonded polypeptide chains termed AR, Bβ, and γ. The N-terminal glutamine of the Bβ chain, which yields fibrinopeptide-B (FPB), is completely cyclized to pyroglutamic acid.9 As a clotting factor, fibrinogen is an essential component of the blood coagulation system, being the precursor of fibrin. Thrombin, a trypsin-like serine endopeptidase, cleaves fibrinogen at AR-chain between Arg16 and Gly17 and at the Bβ-chain between Arg14 and Gly15, releasing the fibrinopeptides fibrinopeptide-A (FPA) and FPB and the reactive species, monomeric fibrin, which rapidly polymerizes forming a network of fibers known as the fibrin clot.10 We identified the peptides underlying the signals in Figure 2 by a tandem mass spectrometric experiment performed with a MALDI-TOF/TOF mass spectrometer (Figure 3). The mass (9) Blomback, B. M.; Blomback, P.; Edman, P.; Hessel, B. Biochim. Biophys. Acta 1966, 115, 371-396 (10) Profumo, A.; Turci, M.; Damonte, G.; Ferri, F.; Magatti, D.; Cardinali, B.; Cuniberti, C.; Rocco, M. Biochemistry 2003, 42, 12335-12348. (11) Henschen-Edman, A. H. Ann. N. Y. Acad. Sci. 2001, 936, 580-593. (12) Serrano, S. M.; Shannon, J. D.; Wang, D.; Camargo, A. C.; Fox, J. W. Proteomics 2005, 5, 501-510 (13) Saghatelian, A.; Trauger, S. A.; Want, E. J.; Hawkins, E. G.; Siuzdak, G.; Cravatt, B. F. Biochemistry 2004, 43, 14332-14339.
1254
Analytical Chemistry, Vol. 79, No. 3, February 1, 2007
spectra (Figure 3) confirm the existence of des-alanine-(1)-des(2)-aspartic acid-fibrinopeptide-A (dAdD-FPA), a peptide fragment 22-35 (SGEGDFLAEGGGVR, Figure 3A) of the fibrinogen R/R-E chain precursor fibrinopeptide-A (P02671); of des-(1)-glutaminedes-(2)-glycine-fibrinopeptide-B (dQdG-FPB), a peptide fragment 33-44 (VNDNEEGFFSAR, Figure 3B) of the fibrinogen β-chain precursor (P02675); of des-alanine-(1)-fibrinopeptide-A (dA-FPA), a peptide fragment 21-35 (DSGEGDFLAEGGGVR, Figure 3C) of the fibrinogen R/R-E chain precursor (P02671); of FPB, a peptide fragment 31-44 (QGVNDNEEGFFSAR, Figure 3D) of the fibrinogen β-chain precursor (P02675); of FPA, a peptide fragment 20-35 (ADSGEGDFLAEGGGVR, Fig, 3E) of the fibrinogen R/R-E chain precursor (P02671); and of fibrino-peptide-A phosphorylated at serine-(3) (pFPA), a peptide fragment 20-35 (ADSpGEGDFLAEGGGVR, Figure 3F) of the fibrinogen R/R-E chain precursor (P02671). Fibrinopeptide-A phosphorylated at serine-(3) and des-alanine(1)-fibrinopeptide-A are known fibrinogen protein species. Blomback et al. reported that ∼10% of fibrinogen had lost its alanine(1) at the AR, probably due to the action of a protease.9 Incubation of the protein species des-alanine-(1)-fibrinogen with thrombin yielded dA-FPA. P-FPA is a hydrolysis product of phosphorylated fibrinogen, released by thrombin. In a healthy person, 25% of fibrinogen is phosphorylated.11 CONCLUSIONS This study demonstrates the utility of MAPS for several applications. MAPS is suited for helping to deorphanize genes annotated as proteases, detecting their endogenous substrate even in complex protein fractions. After establishing a MAPS assay, the unknown substrate of a defined protease can be identified after a MAPS-guided purification. MAPS also has the potential for being a diagnostic tool and can be used for the molecular phenotyping of individuals: It may be used to characterize the heterogeneity of target proteins such as fibrinogen by detecting heterogeneitydisplaying peptides, which are released from the target protein by a defined protease. Furthermore MAPS may be useful for characterizing enzymes present in venoms and identifying their substrates. MAPS may also overcome some problems of zymography, which is a classical method for characterizing the function of snake venom proteases. Zymography is restricted to a small number of substrates, which can be screened, shows a reduced resolution compared to 2D-electrophoresis, and may fail to detect enzymatic activities of venom proteases after electrophoresis.12 A promising strategy for the assignment of endogenous substrates to enzymes was published by the group of Cravatt.13 Endogenous substrates of enzymes are identified by untargeted liquid chromatography-mass spectrometry analysis of tissue metabolomes from wild-type and enzyme-inactivated (knockout) organisms. Although this method was not applied for the identification of protease substrates yet, it can be assumed that the method of Cravatt may also help in this issue. However, focusing on regulatory proteases, which activate other abundant proteases, false positive protease substrates may be identified, which are the substrates of the abundant proteases. In this case, the method described here should be more helpful, since the method focuses directly on the reaction products of the orphan protease.
MAPS can also be used for the characterization of inhibitors as already shown in the description of the MES method,6 because the readout system for detecting the reaction products, MALDIMS, is identical. The MES method utilizes the immobilization of the enzyme, the substrate being present in the reaction solution. In the MAPS approach, the substrate is immobilized and incubated with the enzyme dissolved in a buffer. MAPS can be easily automated and hence used for highthroughput screening since all liquid-handling and incubation steps described above can be performed in parallel in 96-well
plates, and automated MALDI sample preparation and MALDI mass spectrometry can be used. ACKNOWLEDGMENT This work was supported by grants (31P3072, 0313694A) of the BMBF (Bundesministerium fu¨r Bildung and Forschung). Received for review August 9, 2006. Accepted November 29, 2006. AC061482L
Analytical Chemistry, Vol. 79, No. 3, February 1, 2007
1255