Detection of Matrix Metalloproteinase Active Forms in Complex

Detection of Matrix Metalloproteinase Active Forms in Complex Proteomes: Evaluation of Affinity versus Photoaffinity Capture Sarah Bregant,† Ce´line Huillet,† Laurent Devel,† Anne-Sophie Dabert-Gay,† Fabrice Beau,† Robert Thai,† Bertrand Czarny,† Athanasios Yiotakis,‡ and Vincent Dive*,† CEA, iBiTecS, Service d’Inge´nierie Mole´culaire des Prote´ines (SIMOPRO), Bat 152, CE-Saclay Gif/Yvette, F-91191, France, and Laboratory of Organic Chemistry, University of Athens, Panepistimiopolis, Zografou, 15771 Athens, Greece Received December 15, 2008

Various attempts to detect matrix metalloproteinase (MMP) active forms from complex proteomes, based on the use of specific photoactivatable affinity probes, have up to now failed. To overcome this failure, an affinity approach has been evaluated as an alternative to the photoaffinity one. For this purpose, two probes were synthesized to interact specifically with the active site of MMPs and allow isolation of MMP/probe complexes on magnetic beads through a biotin linker. Using phosphinic peptide chemistry, we prepared an affinity probe displaying picomolar potency toward several MMPs, and a related photoaffinity probe incorporating a photoactivatable azido group exhibiting subnanomolar affinity toward these targets. By a combination of silver-staining detection and MALDI peptide mass fingerprints, a systematic comparison was made of both strategies in terms of hMMP-12 and hMMP-8 recovery and identification when present in mixtures of different complexity. The results obtained show that the affinity protocol is superior to the photoaffinity strategy in terms of quantity of captured MMPs and number of MMP tryptic fragments detected in MALDI-MS. The specificity and efficiency of the affinity capture protocol developed in this study allowed easy, fast, and unambiguous detection by MALDI-MS of three hMMPs (2, 8, and 12), from a single affinity capture experiment, when added (10-36 ng of MMPs) to a tumor extract (10 µg). Thus, the tools and approaches reported should enable us to progress in the detection of endogenous active forms of MMPs in complex proteomes, an important objective with many diagnostic applications. Keywords: activity-based probe • MMP profiling • phosphinic inhibitor • MMPs • functional proteomics

Introduction Activity-based proteomics is a relatively new field in which distinct sets of enzymes under active forms within complex proteomes are monitored. This proteomic approach relies on the development of small probes, termed activity-based probes (ABPs), which are able to modify covalently conserved residues present in enzyme active sites.1-3 A fluorescent or affinity tag in the activity-based probe structure is used to detect and distinguish modified enzymes from other unmodified proteins. For serine and cysteine proteinases, ABPs have been developed by exploiting the presence of conserved nucleophiles in the active site of these enzymes and selecting appropriate reactive groups to perform covalent modification of the targeted enzymes.4,5 The use of these APBs has led to the successful detection of active forms of these proteinases in complex proteomes.6,7 For zinc metalloproteinases, like the matrix metalloproteinases (MMPs), the lack of corresponding active site nucleophile has required the incorporation of photochemi* To whom correspondence should be addressed. Tel.: 33-1-6908-2603. Fax: 33-1-6908-9071. E-mail: [email protected]. † CEA. ‡ University of Athens.

2484 Journal of Proteome Research 2009, 8, 2484–2494 Published on Web 03/09/2009

cal groups into the structure of synthetic MMP inhibitors to achieve covalent modification of MMP active sites. Despite the development of different dedicated MMP ABPs incorporating various photolabile groups, all attempts to detect active forms of MMPs in complex proteomes with these probes have failed up to now.8-10 To explain this failure, it has been argued that these proteases are mostly expressed as inactive zymogen forms and that most of the activated MMP fractions are blocked by endogenous inhibitors. Thus, active forms of MMPs may exist at levels that are well below the current limit of ABP detection.10 Among the various factors that might limit the sensitivity of the photoaffinity probes developed for detecting MMPs, we suggested that the variability in the active site amino acid composition between MMPs may affect the cross-linking yield of those probes. In the specific case of an MMP probe developed in our laboratory, this hypothesis was verified. Whereas a 45% yield of human MMP-12 (hMMP-12) covalent modification was determined for this probe, for human MMP-8 (hMMP-8) the cross-linking yield was only a few percent (1-2%).11 These results led us to ask whether a specific capture of the targeted enzymes by an affinity probe, instead of a photoaffinity one, may be a better way to detect MMPs when present in complex proteomes. To address this issue, a pho10.1021/pr801069c CCC: $40.75

 2009 American Chemical Society

Evaluation of Affinity versus Photoaffinity Capture toaffinity probe and an affinity probe directed toward MMPs were developed and compared side by side in their ability to detect either a single MMP (hMMP-12 and hMMP-8) or several MMPs (hMMP-2, hMMP-8 and hMMP-12), respectively, in samples of different complexity. Efficiencies of both approaches were evaluated using either classical protein detection by silver staining, after protein separation by electrophoresis, or by MALDI peptide mass fingerprints.

Materials and Methods Human hMMPs were from R&D systems, with the exception of hMMP-12, which was produced with a mutation in position 171 (F171D).26 Tumor extracts were prepared from tumors grown in mice with a BALB/c genetic background, after subcutaneous injection of C26 cells. Enzymatic assays were performed as previously described.15 Capture experiments used magnetic beads functionalized with Streptavidin (Dynabeads M-280 Streptavidin (DYNAL Biotech)). Beads were handled using an MPC-E (Magnetic Particle Concentrator for Microtubes of Eppendorf type) or an MPC-S magnet (DYNAL Biotech). A 4800 spectrometer MALDI-TOF/TOF Proteomics Analyzer (Applied Biosystems, Foster City, CA) was used for mass spectrometry analysis of samples. The BioRad MiniProtean Electrophoresis System was used with mini-gels (thickness 1 mm). Polyvinylidene difluoride membranes (PVDF, Immobilon Transfer Membranes, pore size: 0.45 µm) and blotting paper (Extra thick Blot paper, PROTEAN XL size) used for protein gel-transfer were from Millipore and BioRad, respectively. Protein was transferred using Trans-Blot SD SemiDry Transfer Cell from BioRad. Buffered samples were concentrated using a Speed vac (Savant). Enzymatic digestions were performed at 50 °C using an Eppendorf tube heater (Grant QBT). Glycerol, Tris base (2-amino-2-(hydroxymethyl)-1,3propanediol), CaCl2 (calcium chloride), NH4(HCO3) (ammonium bicarbonate), SDS (sodium dodecyl sulfate), TEMED (N,N,N′,N′-tetramethylethylenediamine), APS (ammonium persulfate), AgNO3 (silver nitrate), formaldehyde 37%, R-250 Coomassie blue, glycine, Na2CO3 (sodium carbonate), bromophenol blue, β-mercaptoethanol and Na2S2O3 (sodium thiosulfate), mammalian protease inhibitor cocktail P8340 were from SIGMA Aldrich. Methanol, ethanol and acetic acid were from de VWR Prolabo. Acrylamide 40%/bis solution, 37.5:1(2.6% C) and molecular markers (Precision Plus Protein Standards, All Blue) were from BioRad. CH3CN was from Riedel-de Hae¨n. Porcine trypsin (seq. grade modified trypsin, porcine, 20 µg) for enzymatic digestion was from Promega. R-Cyano-4-hydroxycinnamic acid (HCCA) from Fluka was used as organic matrix for mass spectrometry analysis. Commercial reagents were used without additional purification. Streptavidin-HRP and enhanced chemiluminescence assay ECL were respectively from Sigma and Pierce. Inhibition Studies. Enzyme inhibition assays were carried out in 200 µL of 50 mM Tris/HCl buffer, pH 6.8, 10 mM CaCl2, at 25 C°, using the Mca-Pro-Leu-Gly-Leu-Dpa-Ala-Arg-NH2 peptide as substrate (Mca: 7-methoxycoumarin-4 acetic acid; Dpa: N3-(2,4-dinitrophenyl)-L-2,3 diaminopropionic acid). Highly potent inhibitors such as probe 1 and 2 impose long time of enzyme/probe complex formation, so that the equilibrium cannot be observed before complete depletion of the substrate. In this case, Ki determination is involved long enzyme/probe incubation time (overnight at 4 °C), choosing enzyme concentration well below the expected Ki value of the probe. At room temperature, reaction was initiated by adding to these mixtures

research articles the fluorogenic substrate to a final concentration of 4.5 µM. In these experiments, enzyme concentrations were as follow: [hMMP-2], 0.001 nM; [hMMP-8], 0.049 nM; [hMMP-9], 0.003 nM; [hMMP-12], 0.03 nM; [hMMP-13], 0.003 nM; [hMMP-14], 0.085 nM. To compensate for the slow conversion of the substrate imposed by the low enzyme concentration chosen, progress reaction curves at different probe concentrations were monitored for 4 h. Ki was determined from these progress curves as previously described.15 Enzyme Titration. Concentrations of hMPP-2, hMMP-8 and hMMP-12 samples used for capture experiments were determined by titration experiments using a calibrated concentration of a highly potent MMP inhibitor (structure supplied in Supporting Information). Tumor Extract Preparation. Whole harvested tumors were suspended in TCNB buffer (800 mg/mL) in presence of commercial mammalian protease inhibitor cocktail (10 µL/mL) P8340 targeting all proteases with the exception of metalloproteinases. The suspension was manually crushed at 4 °C within a potter to provide an extract that was centrifuged for 30 min at 4000 r/min. Supernatant was then isolated from the pellet and the homogeneous tumor extract solution was kept and stored at -80 °C after Bradford’s dosage of proteins. Photolabeling Experiments. All irradiation experiments (λ, 310 nm; 100 µW.cm-2) were performed under inactinic light (Na light) on samples at 10 °C. Photolabeling Experiments for SDS-PAGE and PVDF Blotting. hMMP-8 or hMMP-12 (50 nM) was incubated with compound 1 [100 nM] in 40 µL of 50 mM Tris-HCl, pH 6.8, 10 mM CaCl2, 0.01% Brij35, for 10 min (10 °C). Eppendorf tubes were irradiated for 10 min (10 °C) prior to addition of 10 µL of 5× SDS loading buffer to the reaction mixture for electrophoretic migration. Photolabeling Experimental Procedure A Prior to Capture Experiments and SDS-PAGE Analysis. hMMP-8 or hMMP-12 (100 nM) was incubated with compound 1 [100 nM] in 50 µL of 50 mM Tris-HCl, pH 6.8, 10 mM CaCl2, 0.01% Brij35, for 10 min (10 °C). Eppendorf tubes were then irradiated for 10 min (10 °C). Photolabeling Experimental Procedure B Prior to Capture Experiments and MALDI-MS Analysis. hMMP-8 or hMMP-12 (10 nM) was incubated with compound 1 [50 nM] in 50 µL of 50 mM Tris-HCl, pH 6.8, 10 mM CaCl2, 0.01% Brij35, for 10 min (10 °C). Eppendorf tubes were then irradiated for 10 min (10 °C). Procedure A was applied to mixtures of hMMP [100 nM] and 10 µg of tumor extract and procedure B to hMMP [10 nM] plus 10 µg of tumor extract. Capture Experiments. Magnetic Beads “Pre-Wash”. Before capture experiments, magnetic beads were subjected to different washes to remove the BSA and sodium azide present in the conditioning solution. Typically, beads (50 µL of commercial stock solution) were washed using a magnet, twice with deionized water (2 mL), twice with 50 mM Tris buffer, 10 mM CaCl2, pH 6.8 (2 mL), and then suspended in 50 µL of 50 mM Tris buffer, 10 mM CaCl2, pH 6.8. Volumes of Beads Used for Capture Experiments. In experiments characterizing captured hMMP by gel silver staining, 5 µL of beads solution was used. For MALDI-MS analysis, 0.5 and 2.5 µL were added for capture of a single hMMP with probe 2 and probe 1, respectively. Capture using probe 2 of a mixture of 3 hMMP was performed using 1.5 µL of beads solution. Calibration experiments indicated that a volume of 1 Journal of Proteome Research • Vol. 8, No. 5, 2009 2485

research articles µL of commercial Streptavidin beads solution has the binding capacity allowing the capture of 1 picomole of biotinylated probe whether the probe was studied alone or bound to hMMP-2, hMMP-8 or hMMP-12. Capture and Subsequent Bead Treatment. Captures were performed in 50 µL of buffered solutions (50 mM Tris-HCl, pH 6.8, 10 mM CaCl2, 0.01% Brij35 buffer), in the absence or presence of 10 µg of tumor extract. In sample analysis by silver staining, the concentration of MMPs was set at 100 nM, whereas for MALDI-MS analysis the concentration was 10 nM. Samples were first incubated for 45 min with either probe 1 or probe 2. Probe concentrations were adjusted to MMP concentrations, except for experiments involving probe 1 in the presence of 10 nM MMPs, where the probe concentration was set to 50 nM. Samples containing probe 1 were photoirradiated prior to capture experiments. Five µL of washed beads solution was added to samples with MMPs (100 nM) and probe 1 or 2 (100 nM); 0.5 µL of bead solution was added to samples with MMPs (10 nM) and probe 2; and 2.5 µL of bead solution was added to samples with MMPs (10 nM) with probe 1. In capture experiments involving a mixture of hMMP (2, 8 and 12), each at 10 nM final concentration in tumor extract, probe 2 was set at 30 nM and 1.5 µL of bead solution was used. In all cases, magnetic beads were incubated for 1 h, with time to time gentle vortexing, before subsequent analyses. After 1 h, supernatants were separated from beads with the help of an appropriate magnet. For capture characterization purpose using silverstained gels, isolated supernatants were concentrated under high vacuum prior to addition of 20 µL of loading SDS-PAGE buffer. Isolated beads from captures involving probe 2 (and related control experiments without probe) were then consecutively washed with 500 and 50 µL of 1 M NaCl; 400 µL, 50 and 20 µL of 50 mM Tris-HCl buffer, pH 6.8, 10 mM CaCl2; and finally 20 µL of deionized water. Isolated beads from captures involving probe 1 (and related control experiments without probe) were mixed with 20 µL of loading buffer and heated at 37 °C for ten minutes. Blue supernatants were removed with the help of a suitable magnet and resulting beads were then washed with 500 and 50 µL of 1 M NaCl, followed by 400 µL, 50 and 20 µL of 50 mM Tris-HCl buffer, pH 6.8, 10 mM CaCl2 and finally 20 µL of deionized water. Treated beads were then transferred into new Eppendorf tubes, from which final aqueous supernatants were discarded. Isolated washed beads were then either mixed with 20 µL of loading buffer 1× for SDS-PAGE analysis or suspended in 9 µL of a mixture of 50 mM sodium bicarbonate/ acetonitrile (9:1), prior to addition of trypsin for enzymatic digestion and MALDI-MS analysis. Electrophoresis. SDS-PAGE was performed according to Laemmli25 using Mini-Protean III apparatus (BioRad). Beads or supernatant samples in loading buffer were boiled at 95 °C for 5 min and subsequently loaded into wells of a 12% SDS gel (1 mm thickness). Silver Staining. Gels were fixed for 2 h in 30% ethanol, 5% acetic acid in deionized water before washing for 4 × 10 min in deionized water. Gels were sensitized by a 1-min incubation in 0.02% sodium thiosulfate solution and rinsed twice with deionized water for 1 min. Gels were then immersed in 0.01% formaldehyde, 0.2% silver nitrate solution and kept in the dark. After 45 min, gels were rinsed in deionized water and immersed in aqueous solution containing 0.01% formaldehyde, 2.4% sodium carbonate, and 0.001% of sodium thiosulfate, with gentle shaking. When the desired intensity of staining was 2486

Journal of Proteome Research • Vol. 8, No. 5, 2009

Bregant et al. achieved, gels were treated with a solution at 330 mM Tris, 2% acetic acid in deionized water to stop the development. Blotting. Following electrophoresis, the proteins were transferred from gels onto PVDF membranes using a semidry transfer blot apparatus (BioRad). Gels were rinsed in 50 mM Tris-HCl, pH 8.5, 20% methanol, 40 mM glycine, 0.0375% SDS in deionized water (transfer buffer). The PVDF membrane was first activated with ethanol, rinsed with water and equilibrated in the transfer buffer. The transfer was realized as a sandwich between the cathode and anode. A sheet of extra-thick blotting paper, wetted with transfer buffer, was placed on the anode, followed by the activated PVDF membrane. Gels were laid on membranes, covered by 2 sheets of extra-thick blotting paper wetted with transfer buffer, and finally by the cathode. After transfer, membranes were blocked overnight at 4 °C with a solution of 5% BSA in TBS/0.1% Tween. Membranes were washed three times for 10 min each with TBS/0.1% Tween before incubation with a solution of 1% BSA and Streptavidin HRP (1/1000) for 45 min. Resulting membranes were then rinsed three times with TBS/0.1% Tween before being placed in contact with a 1:1 mixture of chemiluminescent substrates for film development. Enzymatic Digestion. Isolated beads resulting from capture experiments were suspended in 9 µL of 50 mM sodium bicarbonate/acetonitrile solution to which was added 1 µL of freshly prepared porcine trypsin solution (100 ng.µL-1 in a mixture of 50 mM sodium bicarbonate/acetonitrile (9:1)). Samples were heated at 50 °C for 45 min. MALDI-TOF MS Analysis. The resulting digest (0.5 µL) was manually spotted on MALDI plate with an equal volume of HCCA (R-cyano-4-hydroxycinnamic acid) matrix solution prepared at 10 mg/mL in H2O/CH3CN/TFA (50/50/0.4). MS spectra were recorded from crystallized samples. Database Search. The DataExplorer software (Ver 4.9) from ABI was used to generate ASCII peak lists from peptide mass fingerprinting MS analyses. Each peak list was manually applied for searches using MASCOT software (www.matrixscience.com) in the NCBInr database updated on Dec 2008. The parameters used for the search were as follows: a taxonomy restriction was placed to Homo sapiens (human), one missed trypsin cleavage was allowed, a maximum mass tolerance was set at 20 ppm because of an internal calibration made on fragments resulting from trypsin autolysis for all MS analyses and methionine oxidation was set as a variable modification. MASCOT proteins hit Mowse score greater than 66 (assuming p < 0.05) were considered significant. All analyses were performed at least in triplicate.

Results Probe Development and Characterization. To derive a probe endowed with high affinity for hMMP-12 and hMMP-8, phosphinic peptide chemistry was used as it leads to compounds that behave as good transition-state analogs of zinc metalloproteinases12 and thus to very potent active sitedirected inhibitors of these enzymes.13,14 Based on structurefunction relationships previously used to develop potent MMP inhibitors,11 the phosphinic peptide core of compound 1 was selected (1, Scheme 1). In this photoaffinity probe, an azido group was introduced at a position previously shown to provide good cross-linking yield, at least for hMMP-12. To isolate on magnetic beads the MMPs cross-linked by the photoaffinity probe, a biotin moiety was introduced into compound 1 through a peg spacer. The design of the affinity probe (com-

research articles

Evaluation of Affinity versus Photoaffinity Capture Scheme 1. Chemical Structures of Phosphinic Photoaffinity Probe 1 and Phosphinic Affinity Probe 2

Table 1. Inhibitory Potencies of Photoaffinity Probe 1 and Affinity Probe 2 for Human MMPs MMP

2h

8h

9h

12 h

13 h

14 h

1 - Ki (pM) 2 - Ki (pM)

180 7

70 23

140 21

600 11

400 8

5500 50

pound 2, Scheme 1) followed the same principles, except that the probe was further optimized by N-terminal elongation of the phosphinic part in order to increase the probe’s affinity for MMPs. As shown in Table 1, this design strategy led to probes with subnanomolar (compound 1) and picomolar affinity (compound 2) for hMMP-8 and hMMP-12, but also for other MMPs, a result confirming the value of these two probes in the study of MMPs. In all experiments described below, the concentrations of either probe 1 or 2, when incubated with MMPs, were chosen in order to observe full enzyme inhibition. Thus, any differences in capture efficiency that could be observed between probes would not be related to their respective affinity toward their targets. Covalent Modification of hMMP-12 and hMMP-8 by Probe 1. Previous work has demonstrated that the covalent modification of hMMP-12 by a photoaffinity phosphinic probe results in two protein species that can be resolved by 1D SDSPAGE and observed distinctly by silver staining, the upper band corresponding to the covalently labeled hMMP-12 and the lower band to unmodified hMMP-12.11,15 As shown in Figure 1 (lane a (no photoexcitation) and lane b (after photoexcita-

Figure 1. Silver stained SDS-PAGE 12% polyacrylamide gel and film recorded from PVDF membrane revealed using HRP-Streptavidin peroxidase in experiments with hMMP-12 (lanes a, b, c, a′, b′, c′) and hMMP-8 (d, e, f, d′, e′, f′). hMMP-12 or hMMP-8 (50 nM) was incubated with compound 1 (100 nM) for 10 min, before UV irradiation (hν, 10 min, lanes b/b′ (hMMP-12) and lanes e/e′ (hMMP-8)) or not (lanes a/a′ (hMMP-12) and lanes d/d′ (hMMP8)). Competition experiments involved the mixture of MMP and competitor prior to addition of probe 1 and irradiation (lanes c/c′ (hMMP-12) and lanes f/f′ (hMMP-8)).

tion)), similar results were observed for the covalent modification of hMMP-12 by probe 1. The absence of the upper band in lane a demonstrates that only a covalent modification of hMMP-12 accounts for this band. When a potent active sitedirected inhibitor (chemical structure and Ki supplied in Supporting Information) was first incubated with hMMP-12, prior to addition of probe 1 and photoexcitation, the upper band was not observed (lane c), indicating that the covalent modification of hMMP-12 by probe 1 specifically targets the hMMP-12 active site. In addition to band splitting, covalent modification of hMMP-12 by the biotinylated probe 1 was confirmed by biotin detection (Streptavidin HRP). Biotinylation of hMMP-12 was only detected after light excitation of the hMMP-12:probe 1 complex and in the absence of competitor (lane b′ versus lanes a′ and c′). In contrast to hMMP-12, photomodification of hMMP-8 by probe 1 did not lead to protein species that could be resolved by electrophoresis. Analysis of the irradiated hMMP-8:probe 1 complex revealed only a single band migrating at the same level as unmodified hMMP-8 (50 kDa). Thus, only positive biotin detection for the irradiated hMMP-8 sample demonstrated the occurrence of a covalent modification of this MMP by probe 1, observed at 50 kDa (lane e′). For the band detected by silver staining at 25 kDa, no corresponding biotin signal was observed suggesting that the corresponding protein is devoid of activity. No biotin signal could be observed in the absence of photoirradiation or in the presence of a competitor (d′ and f′ versus e′). Difference in signal intensities related to biotinylated protein species may suggest a higher yield of hMMP-12 covalent modification by probe 1 compared with hMMP-8 (lane b′ versus lane e′). Capture of hMMP-12 and hMMP-8 by Probe 1 or Probe 2. hMMP-12 or hMMP-8, either in buffer or in tumor extract, was incubated with probe 1 or probe 2. To mimic physiological conditions, in experiments in the presence of tumor extract, hMMP-12 or hMMP-8 was first mixed with tumor extract solution (containing 10 µg of protein), followed by probe addition. After photoexcitation of solutions containing probe 1, magnetic beads functionalized with Streptavidin were added to each MMP mixture, which was gently shaken for 1 h. Volumes of Streptavidin beads were adjusted to the quantity of added probes. Supernatants were then separated from beads with the help of a suitable magnet (Scheme 2) before subsequent treatment of beads to SDS-PAGE or mass spectrometry analysis. Protocols were designed to limit the number of steps and time to treat complex mixtures until SDS-PAGE or mass spectrometry detection of specifically captured proteins. Assessment by SDS-PAGE Analysis of hMMP-12 and hMMP-8 Capture Efficacy by Probe 1 and Probe 2. Concentrations of hMMP-12 or hMMP-8, either in buffer or in tumor extract, and of probes 1 or 2, were set at 100 nM. After 45 min of incubation at room temperature, the degree of hMMP-12 Journal of Proteome Research • Vol. 8, No. 5, 2009 2487

research articles

Bregant et al.

Scheme 2. Isolation of Targeted Metalloproteinases Based on Covalent Linkage to Probe 1 or Affinity Interactions with Probe 2

or hMMP-8 inhibition by the probes in buffer was determined by enzymatic tests, based on the cleavage of a specific MMP fluorescent substrate. As expected from the probe inhibitory potencies reported in Table 1 and selected concentrations of probes and MMPs for these experiments, no substrate cleavage was observed under these conditions, thus in each case the enzyme is saturated by the probe. Isolated beads resulting from affinity capture with probe 2 were submitted to washing steps that preserved interactions within the MMP:probe 2:Streptavidin complex. Beads isolated after the covalent capture of MMPs by probe 1 were treated under more stringent denaturing conditions, prior to several washes. In both cases, washes were discarded. Supernatants and beads resulting from both approaches were submitted to SDS-PAGE analysis to compare protein capture efficiency of each approach through the detection of relative quantities of enzyme present on beads and in supernatants, an information more difficult to retrieve from MALDI-TOF MS. hMMP-12 and hMMP-8 Capture in Buffer. According to the results reported in Figure 2, hMMP-12 was quantitatively captured on beads with the protocol involving the affinity probe 2 (lane I versus lane annotated C, Figure 2). In all bead eluates, whether probe was involved or not, an additional protein band was detected (lanes I, III, V, and VII, Figure 2) corresponding to Streptavidin monomers released from the magnetic beads due to the reductive property of the loading buffer. Quantitative recovery of hMMP-12 from beads (line I) was in agreement with the quasi-absence of hMMP-12 in the corresponding supernatant (lane II). Capture of hMMP12 depended on the presence of probe 2, as when probe 2 was omitted all hMMP-12 was recovered in the supernatant (lane IV) and none from beads (lane III). When capture of hMMP-

Figure 2. Silver stained SDS-PAGE 12% polyacrylamide gel analysis of bead elution (lanes I, III, V, VII) and supernatants (lanes II, IV, VI, VIII) of capture experiments with 100 nM hMMP-12 using beads and probe 2 (100 nM) [lanes I, II], no probe [lanes III, IV], probe 1 (100 nM) and irradiation [lanes V, VI] or irradiation only [lanes VII, VIII]. Line C refers to reference hMMP-12 and line Chν refers to control mixture resulting from irradiation of hMMP-12 in presence of probe 1. 2488

Journal of Proteome Research • Vol. 8, No. 5, 2009

Figure 3. Silver stained SDS-PAGE 12% polyacrylamide gel analysis of bead elution (lanes I, III, V, VII) and supernatants (lanes II, IV, VI, VIII) of capture experiments with 100 nM hMMP-8 using beads and probe 2 (100 nM) [lanes I, II], no probe [lanes III, IV], probe 1 (100 nM) and irradiation [lanes V, VI] or irradiation only [lanes VII, VIII]. Line C refers to reference hMMP-8 and line Chν refers to control mixture resulting from irradiation of hMMP-8 in presence of probe 1.

12 was performed with the photoaffinity probe 1, the covalent adduct between hMMP-12 and probe 1 was released from beads (lane V), but a fraction corresponding to unmodified hMMP-12 remained in the supernatant (lane VI). When probe 1 was not used in the capture protocol, all the hMMP-12 was detected in the supernatant (lane VIII) and none on the beads (lane VII). The observation of unmodified hMMP-12 in the supernatant after probe 1 photoirradiation was unexpected, since prior to irradiation no hMMP-12 activity was detected, suggesting a full blockade of hMMP-12 by probe 1. This observation suggests that the fraction of photoirradiated probe 1, not cross-linked to hMMP-12, corresponds to a novel chemical structure that has a lower affinity for hMMP-12, explaining the presence of uncaptured hMMP-12 in the supernatant. This hypothesis is supported by the observation of a partial enzyme activity recovery in the sample after the photoactivation step. Based on the band intensity corresponding to captured hMMP-12 in lanes I and V, it can be concluded that the affinity capture allowed isolation on beads of a greater quantity of hMMP-12, compared with photoaffinity capture. The quantity of hMMP-12 captured through the affinity route and observed in elution from beads matches the quantity used in the experiment (lane C). In the corresponding supernatant, only faint traces of hMMP-12 could be distinguished (lane II). Experiments on hMMP-8 capture (Figure 3) performed with probe 2 gave results similar to those for hMMP-12. Active hMMP-8 was efficiently captured by probe 2 (lane I), although some hMMP-8 was still detected in the corresponding supernatant (lane II). This is not due to lack of probe 2 capture efficiency because enzymatic tests performed with this supernatant revealed no hMMP-8 proteolytic activity suggesting a incorrect folding for this uncaptured fraction of hMMP-8. Affinity capture of hMMP-8 was more efficient than photoaffinity capture performed with probe 1, as analysis of the beads corresponding to the latter suggested an absence of hMMP-8 capture (lane V). Thus, the cross-linking yield of hMMP-8 by probe 1 should be rather low, in agreement with the biotin labeling experiments reported in Figure 1. Presence of hMMP-8 in the supernatant of photoaffinity capture (lane VI) is probably related, as suggested above, to the chemical structure adopted by probe 1 after photoirradiation, which should have a lower affinity for hMMP-8. hMMP-12 and hMMP-8 Capture in Tumor Extract. When hMMP-12 was present in tumor extract solutions, the use of probe 2 allowed isolation in very pure form of most of the

Evaluation of Affinity versus Photoaffinity Capture

Figure 4. Silver stained SDS-PAGE 12% polyacrylamide gel analysis of bead elution (lanes I, III, V, VI) and supernatants (lanes II, IV) of capture experiments with 100 nM hMMP-12 in 10 µg of tumor extract using beads and probe 2 (100 nM) [lanes I, II], probe 1 (100 nM) and irradiation, [lanes III, IV], no probe with or without irradiation [lanes V, VI]. Line C refers to reference hMMP-12 and line Chν refers to control mixture resulting from irradiation of hMMP-12 in presence of probe 1.

Figure 5. Silver stained SDS-PAGE 12% polyacrylamide gel analysis of bead elution (lanes I, III, V, VII) and supernatants (lanes II, IV, VI, VIII) of capture experiments with 100 nM hMMP-8 in 10 µg of tumor extract using beads and probe 2 (100 nM) [lanes I, II], probe 1 (100 nM) and irradiation, [lanes III, IV], no probe with or without irradiation [lanes V, VI]. Line C refers to reference hMMP-8 and line Chν refers to control mixture resulting from irradiation of hMMP-8 in presence of probe 1.

hMMP-12 added to this complex model proteome (lane I versus II, Figure 4). hMMP-12 was also partially extracted from this model proteome using probe 1 (lane III, Figure 4), in a form of hMMP-12 covalently modified by probe 1. In this case, less hMMP-12 was isolated on beads than with probe 2 (lane III versus I). In both cases, effective hMMP-12 capture was dependent on the presence of the probe (lanes V and VI). For the capture of hMMP-8, results reported in Figure 5 clearly indicate that affinity capture (lane I) is more efficient than photoaffinity capture (lane III). Characterization by MALDI-TOF Spectrometry of the Efficiency of Capture of hMMP-12 and hMMP-8 by Probes 1 and 2. Both strategies were then evaluated using MALDI mass spectrometry to identify the captured proteins by mass fingerprinting. In these capture experiments, MMP concentrations were set at 10 nM. Thus, probe 2 concentration was also set at 10 nM for affinity strategy, whereas in the covalent strategy a 50 nM concentration of probe 1 was selected. This probe 1 concentration was selected to ensure 100% of hMMP-12:probe 1 complex formation prior to photoirradiation. For comparison purposes, similar concentrations of probe and enzyme were used in experiments with hMMP-8. After MMP capture, isolated washed beads were subjected to trypsin digestion. The corresponding digests were submitted to mass spectrometry, without further purification or separation. In MALDI-MS analysis, only 1/20 of tryptic digests were loaded on the MALDI plate, thus representing a maximum of 25 femtomoles for each tryptic peptide fragment. hMMP-12 and hMMP-8 Capture in Buffer. When probe 2 was used for the capture of hMMP-12, MALDI-MS analysis of

research articles the tryptic digest led to the detection of 9 tryptic fragments (star-labeled peaks, panel A, Figure 6) out of a theoretical 13 deduced from the hMMP-12 sequence (Figure 7). Masses corresponding to observed peaks and their corresponding relative intensities were in good agreement with the distribution of peaks and their relative areas observed in the analysis of a digest obtained by direct trypsin digestion of hMMP-12 in solution (spectra of hMMP-12 digest, 250 femtomoles loaded on the MALDI plate, is supplied in Supporting Information section). From MALDI-MS data obtained from capture experiment using probe 2, successful identification of captured hMMP-12 (mascot protein score: 99) was achieved by submitting these MS data (mass tolerance: 20 ppm) to Mascot search using NCBInr database restricted to human taxinomy. Internal calibration was set on fragments resulting from trypsin autolysis (peaks pentagon-labeled in Figure 6) whose detection was due to the trypsin quantity used to cleave hMMP (equimolar amount of trypsin and MMP). When the capture of hMMP-12 was performed with probe 1, only 7 hMMP-12 peptides were detected in the MALDI-MS spectrum (star-labeled peaks, panel B, Figure 6) and allowed the identification of hMMP-12, but through a lower score as compared to the affinity capture (mascot protein score: 68). The quantity of detected hMMP-12 fragments, related to the quantity of captured protein, was thus lower when probe 1 was used for capture, as compared with probe 2. This observation agreed with previous from SDS-PAGE capture analyses (lane I vs lane V, Figure 2). Interestingly, the peak detected with the highest intensity (m/z ) 1027.5203) and a less intense peak (m/z ) 956.4890) in experiments with probe 2 were never observed in the MALDIMS spectrum of samples prepared with probe 1. In the latter case, additional signals related to hMMP-12 tryptic fragments cross-linked by probe 1 were not detected. This suggested that enzyme covalent modification by a photoaffinity probe may prevent either the production or the detection of tryptic enzyme fragments that can be detected with high sensitivity and that might be useful for target identification. Capture experiments performed with probe 2 led to the detection of 17 tryptic fragments of hMMP-8 (round-labeled peaks, panel C, Figure 8), whereas only 9 fragments were detected when probe 1 was used in the covalent capture protocol (round-labeled peaks, panel D, Figure 8). The latter 9 fragments correspond to the most intense ones observed in the experiments involving affinity capture. Valid Mascot identification of hMMP-8 was obtained only for capture using probe 2 (mascot protein score: 125) whereas a protein score of only 24 was attributed to hMMP-8 from MS data resulting from the capture using probe 1. hMMP-12 or hMMP-8 tryptic fragments were not observed in digests obtained from captures without probes, confirming that the capture depended on specific interactions between MMPs and probes (control spectrum provided in Supporting Information, Figure 2). However, in capture and control experiments, tryptic fragments from the Streptavidin linked to beads and residual BSA from conditioned bead media were detected. The presence of these unspecific tryptic fragments did not prevent the unambiguous identification of hMMP, as in this case no overlap between masses of tryptic fragments of targeted MMP and contaminant proteins occurs. Comparable peak intensity between tryptic fragments of trypsin, Streptavidin and BSA indicates that Streptavidin on-beads is poorly digested by Journal of Proteome Research • Vol. 8, No. 5, 2009 2489

research articles

Bregant et al.

Figure 6. MALDI-MS spectra of digests when hMMP-12 was captured with probe 2 (A) or via covalent linkage through probe 1 (B). Ticked mass and star-labeled peaks correspond to tryptic fragments of hMMP-12. Pentagon-labeled peaks refer to fragments resulting from trypsin autolysis.

Figure 7. Sequence of hMMP-12 used in capture experiments and theoretical masses of tryptic fragments. Underlined sequence corresponds to S1′ loop of hMMP-12.

Figure 8. MALDI-MS spectra of digests when hMMP-8 was captured with probe 2 (C) or probe 1 (D). Ticked mass and round-labeled peaks correspond to tryptic fragments of hMMP-8. Pentagon-labeled peaks refer to fragments resulting from trypsin autolysis.

trypsin in the nonreductive conditions used and that BSA is present as a minor contaminant. hMMP-12 and hMMP-8 Capture in Tumor Extract. Through the affinity approach using probe 2, capture experiments performed on tumor extract containing hMMP-12 led to the identification of this MMP through the detection of 8 specific 2490

Journal of Proteome Research • Vol. 8, No. 5, 2009

tryptic fragments (protein mascot score 99), whereas only 6 of these fragments were detected through the covalent approach based on probe 1 (protein mascot score of 68). Once again, only peaks detected with the highest intensity in digests from capture involving probe 2 were observed in the digest obtained with probe 1 (Figure 9).

Evaluation of Affinity versus Photoaffinity Capture

research articles

Figure 9. MALDI-MS spectra of digests when hMMP-12 was captured probe 2 (E) or probe 1 (F) in the presence of 10 µg of tumor extract. Ticked mass and star-labeled peaks correspond to tryptic fragments of hMMP-12. Pentagon-labeled peaks refer to fragments resulting from trypsin autolysis.

Figure 10. MALDI-MS spectra of digests when hMMP-8 was captured with probe 2 (G) or probe 1 (H) in the presence of 10 µg of tumor extract. Ticked mass and round-labeled peaks correspond to tryptic fragments of hMMP-8. Pentagon-labeled peaks refer to fragments resulting from trypsin autolysis.

In the case of hMMP-8 capture in the presence of tumor extract, 16 tryptic fragments of hMMP-8 were observed when the affinity approach was applied leading to the identification of hMMP-8 with a mascot protein score of 115, whereas the only 3 weak hMMP-8 specific signals detected by the covalent approach (Figure 10) did not allow hMMP-8 identification (score < 10). In control experiments without probe 2, the main peaks observed correspond to tryptic fragments of trypsin, Streptavidin and BSA (Supporting Information, Figure S3). This suggests that the washing steps used in the affinity capture protocol remove all proteins interacting nonspecifically with the beads. Capture of Several hMMPs in Tumor Extract. Since the aim was to detect active forms of MMPs in biological samples,

whose members can simultaneously be present in the same sample, the efficiency of the affinity strategy for capturing three MMPs (hMMP-2, hMMP-8 and hMMP-12) in the presence of tumor extract was evaluated. Concentrations of each hMMP were set to 10 nM and that of probe 2 at 30 nM. In the MALDIMS spectra resulting from this experiment, hMMP-2, hMMP-8 and hMMP-12 were detected through respectively 13, 12, and 6 related specific fragments (Figure 11). Submission of these MALDI MS data to NCBInr database restricted to human taxinomy resulted in the successful identification of hMMP-2, hMMP-8 and hMMP-12 with respective mascot protein scores of 79, 107, and 69. These results clearly indicated that signal suppression did not prevent observation of the expected peaks of each hMMP. Thus the strategy is compatible with a straightforward MALDI Journal of Proteome Research • Vol. 8, No. 5, 2009 2491

research articles

Bregant et al.

Figure 11. MALDI-MS spectra of digests of a mixture of hMMP-2, hMMP-8 and hMMP-12 captured with probe 2, in the presence of 10 µg of tumor extract. Table 2. Summary of Protein Scores for hMMP Resulting from Submission of MALDI-MS Data of Each Capture Experiment to Database Search (NCBInr Database; Human Taxonomy; Mass Tolerance 20 ppm) MMP

spiked MMP

spectrum

strategy

Identification of MMP

MMP score

In buffer

hMMP-12

A B C D E F G H I

AFF/probe 2 hν/probe 1 AFF/probe 2 hν/probe 1 AFF/probe 2 hν/probe 1 AFF/probe 2 hν/probe 1 AFF/probe 2

hMMP-12 hMMP-12 hMMP-8 no valid identification of hMMP-8 hMMP-12 hMMP-12 hMMP-8 no identification of hMMP-8 hMMP-2 hMMP-8 hMMP-12

99 68 125 24 99 68 115