Covalent Modification of Matrix Metalloproteinases by a Photoaffinity

Jan 12, 2009 - CEA, iBiTecS, Service d'Ingénierie Moléculaire des Protéines (SIMOPRO), CE-Saclay Gif/Yvette, F-91191, France, ... Imperial College ...
0 downloads 0 Views 1MB Size
Bioconjugate Chem. 2009, 20, 367–375

367

Covalent Modification of Matrix Metalloproteinases by a Photoaffinity Probe: Influence of Nucleophilicity and Flexibility of the Residue in Position 241 Anne-Sophie Dabert-Gay,† B. Czarny,† E. Lajeunesse,† R. Thai,† H. Nagase,‡ and V. Dive*,† CEA, iBiTecS, Service d‘Inge´nierie Mole´culaire des Prote´ines (SIMOPRO), CE-Saclay Gif/Yvette, F-91191, France, and Matrix Biology Department, Kennedy Institute of Rheumatology Division, Faculty of Medicine, Imperial College London, London, United Kingdom. Received November 4, 2008; Revised Manuscript Received December 12, 2008

A photoaffinity probe, developed for the specific labeling of matrix metalloproteinase (MMP) active sites, was recently shown to covalently modify a single residue in human MMP-12, namely, Lys241, by reacting selectively with the side chain ε-amino group of that residue. The residue in position 241 of MMPs is not conserved; thus, variability in this position may be responsible for the dispersion in cross-linking yield observed between MMPs when labeled by this photoaffinity probe. By studying the pH dependence of the labeling properties of this probe toward different MMPs (MMP-12, MMP-3, MMP-9, and various mutants of human MMP-12) and identifying the site of covalent modification of MMP-3 by this probe, our new data demonstrated that the nucleophilicity of the residue in position 241 plays a key role in determining the cross-linking yield of MMP modification by the probe. However, these studies also reveal that subtle additional structural parameters, including local conformation and flexibility, of the residue in position 241 should also be taken into consideration, a property adding a further degree of complexity in our understanding of the photolabeling probe reactivity and in designing optimal photoaffinity probes for performing functional proteomic studies of zinc proteinases like MMPs.

INTRODUCTION The development of photoaffinity probes able to covalently label specifically the active site of zinc proteinases (activitybase probe, ABP) may lead to important applications in functional proteomics and diagnosis (1-4). For serine and cysteine proteinases, successful development of activity-based probes (ABPs) has been achieved by exploiting the presence of highly conserved nucleophiles in the active sites of these enzymes and selecting appropriate reactive groups to perform covalent modification of the targeted enzymes (5, 6). For zinc metalloproteinases, like matrix metalloproteinases (MMPs), the lack of corresponding active site nucleophiles has required the use of a photochemical group that must be incorporated into the structure of synthetic specific MMP inhibitors (1, 2). Following this strategy, we recently developed a selective MMP photoaffinity probe that consists of phosphinic peptide core, able to block potently and selectively a large set of MMPs, to which has been grafted an azido group on the P1′ side chain of the inhibitor (probe 1, Scheme 1) (4). While probe 1 was shown to selectively covalently modify only the active site of MMPs, high variation in the cross-linking yield was observed between different MMPs. In fact, whereas a ≈ 45% of cross-linking yield was determined for the covalent modification of human MMP12 by probe 1, for human MMP-8, this yield decreased to few percent (4). As the molecular determinants controlling the crosslinking yield of photoaffinity probes to their targets remain poorly understood, we initiated a research project in order to identify in MMPs the site of their covalent modifications by probe 1. We recently reported results of a first study of human MMP-12 (hMMP-12), showing that probe 1 covalently modifies only a single residue in this MMP, by reacting with the ε NH3+ group of Lys241 (7). This result was unexpected, as Lys241 is * Corresponding author. Dr. V. Dive: [email protected]. † Service d‘Inge´nierie Mole´culaire des Prote´ines (SIMOPRO). ‡ Imperial College London.

located on a loop segment of hMMP-12, which, based on both X-ray and NMR studies, displays high flexibility with the lysine side chain exhibiting high mobility (8-10). Another remarkable feature of this study is the fact that, at pH 7, the conditions of hMMP-12 covalent modification, photoactivated azides are expected to react with good nucleophiles, but less so with the protonated positively charged ε amino group of Lys241 (11). This remark has led us in the present study to compare the labeling of hMMP-12 by probe 1 at pH 7 and 12. On the basis of the results obtained, human MMP-3 (hMMP-3) was studied in detail, as this MMP possesses in position 241 a histidine residue, whose nucleophilicity is also pH-dependent. Two other MMPs containing an arginine residue in position 241, mouse MMP12 (mMMP-12) and hMMP-9, were also examined. Finally, to better understand the variability in cross-linking yield observed between MMPs with probe 1, several mutants of hMMP-12 have been produced and studied. In these mutants, position 241 has been replaced by residues observed in different MMPs in this position.

EXPERIMENTAL METHODS Commercial reagents were used without additional purification. Buffers and salts were from Sigma. Synthesis and tritium radiolabeling of probe 1 were carried out as previously reported (4). Mca-Pro-Leu-Gly-Leu-Dpa-Ala-Arg-NH2 (Mca-Mat) was from Novabiochem, and TIMP-1 inhibitor was from R&D systems. Proteins. Synthetic genes encoding the catalytic domain of MMP-12 (Met98-Lys266) and murine MMP-12 (Gln109-Gly267) were from Geneart (Geneart-AG, Germany). Genes were inserted into the pET24a vector, between the NdeI and BamHI sites, for expression under the PT7 promoter. Lys241 was substituted by His (primer 5‘-ccgtaatgttccccac-ctacCaCtatgttgacatcaacaca-3′) or Arg (primer 5‘-ccgtaatgttccccacctacCGatatgttgacatcaacaca-3′) or Gln (primer 5‘-ccgtaatgttccccacctacCaG-

10.1021/bc800478b CCC: $40.75  2009 American Chemical Society Published on Web 01/12/2009

368 Bioconjugate Chem., Vol. 20, No. 2, 2009

Dabert-Gay et al.

Scheme 1

tatgttgacatcaacaca-3′) or Thr (primer 5‘-ccgtaatgttccccacctacaCGtat-gttga catcaacaca-3′). The corresponding hMMP-12 mutants were produced using the Site-Directed Mutagenesis Kit (Stratagene, Heidelberg, Germany). All plasmids were propagated in the Escherichia coli strain XL1-Blue at 37 °C, and all constructions were verified by DNA sequencing using the ABI PRISM 310 Genetic analyzer (Applied Biosystems). Recombinant proteins were expressed in E. coli BL21 (DE3 star) cells carrying the MMP-12 catalytic domain-encoding plasmids. Production and purification of MMP-12 mutants were as described in ref 7. Protein molecular weights were determined by ESI-MS, and the presence of mutations was verified by MaltiTOF, after tryptic digestion. Human catalytic domain of MMP-3 was obtained as previously described (12). Enzyme Assays. Enzyme assays and inhibition and titration experiments were carried out in 50 mM Tris/HCl buffer, pH 7, 10 mM CaCl2, at 25 °C, using the Mca-Pro-Leu-Gly-Leu-DpaAla-Arg-NH2 (Mca-Mat) substrate, as described in ref 7. Titration experiments using a calibrated TIMP-1 solution were performed to determine the exact enzyme concentration for each hMMP-12 mutant and wild-type hMMP-12. Values for kcat/Km were determined from first-order full-time course reaction curves obtained at [S] , Km ([S] ) 0.5 µM), at 5 nM final enzyme concentration. The Ki of probe 1 was determined using the method proposed by Horovitz and Leviski (13). Mass Spectrometry. nanoESI-MS experiments were performed in an ion trap mass spectrometer (Bruker, Esquire-HCT). Ion trap parameters were set as follows: nanoelectrospray potential, 1000 V; skimmer voltage, 40 V; capillary exit, 226 V; and a source temperature of 150 °C. The MS survey scan was m/z 400-3000 Da with a target mass fixed at m/z 1800 Da and a 5 spectrum scan average. MS/MS Sequencing of Covalently Modified Peptides. Radioactive fractions containing covalent adduct peptides were concentrated using a C18 ZipTip. Off-line nanoESI-MS/MS experiments were performed in an ion trap mass spectrometer.

Ion trap parameters for MS were set as above. The scan number was increased to 20 spectra over an m/z of 250-2000 Da, the isolation width was set to 1 Da, and the collision energy was tested using a range of 0.2-0.9 V in the MS/MS mode. MS/ MS spectra were recorded for double-charged molecular peptide ions. µ-RP-HPLC. Peptide separation was carried out in an Agilent 1100 series apparatus using an X-Bridge (Waters) C18 5 µm, 300 Å column (2.1 × 150 mm, 5 µm) at a flow rate of 200 µL/min and with detection at 214 nm. Solvent systems were as follows: (A) 100% water, 0.1% FA, and (B) 100% acetonitrile, 0.09% FA; with the following gradient, t ) 0 min, 10% B; t ) 5 min, 10% B; and t ) 45 min, 100% B. Photoaffinity Labeling. MMPs (1 µM) were incubated with 2 µM 3H-probe 1 in 50 mM Tris-HCl, pH 7, 10 mM CaCl2, 0.01% Brij35, for 10 min in the dark at room temperature, as described in ref 7. For pH dependence studies, prior probe addition, the pH was adjusted by adding ammoniac or chloride acid in order to keep the same buffer composition. After photoactivation, the reaction was quenched by the addition of Laemmli loading buffer followed by boiling (5 min, 95 °C). These samples were immediately processed for subsequent SDS gel analysis. Electrophoresis. Radiolabeled proteins were diluted in Laemmli loading buffer (final concentration: 0.1% (w/v) bromophenol blue, 2% (w/v) SDS, 10% (v/v) glycerol, 50 mM TrisHCL pH 6.8, and 100 mM DTT) and were resolved by SDSPAGE electrophoresis in a 12% 1-mm-thick SDS gel, using a mini-protean III apparatus (BioRad). Silver staining was performed as previously described (7). Blotting. Transfer of proteins onto PVDF (polyvinylidene fluoride) membrane was achieved using a semidry transfer blot apparatus (BioRad). Gels were rinsed in a 50 mM Tris-HCl, pH 8.5, 20% methanol, 40 mM glycine, 0.0375% SDS in distilled water (transfer buffer). The PVDF membrane was activated in a bath of methanol and was then equilibrated in

Covalent Modification of MMPs by a Photoaffinity Probe

Bioconjugate Chem., Vol. 20, No. 2, 2009 369

Figure 1. pH dependence for the covalent modification of wild-type h-MMP-12 and Ala-hMMP-12 mutant by probe 1. hMMP12 or Ala mutant (1 µM) was incubated with probe 1 (2 µM) for 10 min, before UV irradiation (2 min). The corresponding complexes (5 pmol) were resolved by 1D-SDS PAGE electrophoresis and visualized on the gel by silver staining, or the proteins were transferred onto a PVDF membrane that was analyzed with a radioimager.

transfer buffer. We formed a membrane sandwich for protein transfer between the cathode and the anode: this included, from anode to cathode, a sheet of extra-thick blotting paper (BioRad), soaked in transfer buffer, the activated PVDF membrane, gels, and finally 2 sheets of extra-thick blotting paper soaked in transfer buffer. After transfer, membranes were dried before radioactivity analysis or were stained using Coomassie Blue (0.1% R250, 50/49/1 water/ethanol/acetic acid). Radioimaging. We carried out radioactivity imaging and counting on the PVDF membranes, using the beta-Imager 2000 from Biospace (Paris, France). This apparatus allows absolute counting of the tritium beta particles, with a detection threshold of 0.007 cpm/mm2 for tritium. PVDF-Tryptic Digestion. Pieces of PVDF membranes containing labeled or unlabeled h-MMP-12 were excised and destained using a 50/50 water/ethanol solution to remove excess R250. PVDF pieces were incubated in a solution of 50 mM NH4 HCO3 pH 8/ acetonitrile (50/50) with trypsin (12.5 ng/µL) for 18-20 h at 50 °C. PVDF pieces were then rinsed with a solution of 50% acetonitrile with 0.1% TFA. Edman Sequencing. N-Terminal amino acid sequence analysis was performed by automated Edman degradation using an ABI model 477A/120A Protein-Peptide Sequencing/ Analysis System and Analysis Software System, model 920A (Applied Biosystems Inc., Foster City, CA). Radioactive fractions eluted from the µ-HPLC column were pooled and loaded onto a TFAtreated cartridge filter, previously conditioned with BioBrene Plus. Prior to each sequence analysis, the filters were calibrated using PTH-amino acids standard solution. Molecular Modeling. Models of probe 1 interacting with hMMP-12 and hMMP-3 were obtained as previously described. The NMR structures in Figure 8 were selected from an ensemble of 20 NMR structures of the hMMP-12 and hMMP-3 catalytic domain deposited in RCSB PDB under access code 2POJ (10) and 2JNP (14), respectively.

RESULTS pH Dependence of hMMP-12 Photolabeling. As shown previously, silver staining of a gel loaded with UV-irradiated hMMP-12 in complex with probe 1 in buffer pH 7 revealed two bands of approximately equal intensity (Figure 1, left panel) (4). The lower band was shown to represent the unlabeled form of hMMP-12, while the higher one corresponds to the covalent hMMP-12/probe 1 complex. Relative intensity of these two bands suggested that ∼45% of hMMP-12 has been cross-linked by probe 1. As expected, a radioactive signal was only observed

for the higher band (Figure 1, middle panel; probe 1 incorporates a radioactive tag which allows covalent h-MMP12/probe 1 complexes to be detected by radioimaging; Scheme 1). Under the same experimental conditions, but with photoactivation performed in a buffer pH 12, either silver staining or radioactivity counting indicated a higher cross-linking yield of hMMP12 by probe 1 (Figure 1, from 45% to 85% based on silver staining). In contrast, when the Ala-MMP-12 mutant (Lys241 has been replaced by Ala) is submitted to similar probe 1 labeling, the cross-linking yield remains extremely low, a result similar to that observed at pH 7 (Figure 1, right panel). Thus, this latter observation demonstrated that the pH dependence of the cross-linking reported for the wild-type hMMP-12 relies exclusively on the presence of Lys241 in this MMP. When hMMP-12 was first incubated with a potent active site MMP inhibitor (compound 2; Scheme 1), followed by probe 1 addition and UV irradiation, results reported in Figure 1 (middle panel) indicated that at pH 12 only the active site of hMMP-12 is modified by the probe, as observed previously when these experiments were performed at pH 7 (4). In combination with our previous study showing that probe 1 covalently modifies the ε amino group of Lys241 (7), observations reported in Figure 1 reveal that the Lys241 nucleophilicity determines the crosslinking yield of probe 1 toward hMMP-12. pH Dependence and Site of Covalent Modification of hMMP-3 by Probe 1. hMMP-3 contains in position 241 a histidine residue; thus, the pH dependence of its covalent modification by probe 1 was examined. Incubation of hMMP3 with probe 1 in buffer pH 7, followed by UV irradiation and protein resolution by electrophoresis, led to the observation of two bands in a gel loaded with hMMP-3/probe 1 complex and stained with silver (Figure 2a, left panel, first lane). Radioimaging analysis indicated that only the higher band incorporated radioactivity (Figure 2a, right panel, first lane). Under the same experimental conditions, lowering the pH of the sample from 7 to 5 resulted in a strong decrease of the cross-linking yield (from 65% to less than 1%), as visualized either by silver staining or radioactivity counting (Figure 2a, second and third lanes in left and right panels). It should be mentioned that an optimum pH of 5.5 has been reported for hMMP-3 in cleaving synthetic substrates (15), a property explaining the low labeling of hMMP-3 by probe 1 observed previously at pH 5.5 (4). When hMMP-3 was first incubated at pH 7 with an excess of compound 2 (Scheme 1), and this was followed by probe 1 addition and UV irradiation, only a single band was observed by silver staining and no radioactivity was detected (Figure 2b,

370 Bioconjugate Chem., Vol. 20, No. 2, 2009

Dabert-Gay et al.

Figure 2. (a) pH dependence for the covalent modification of h-MMP-3 by probe 1. hMMP-3 (1 µM) was incubated with probe 1 (2 µM) for 10 min, before UV irradiation (2 min). (b) Same experiments, but performed in the presence of compound 2 (10 µM) before the irradiation. The corresponding complexes (5 pmol) were resolved by 1D-SDS PAGE electrophoresis and visualized on the gel by silver staining or the proteins were transferred onto a PVDF membrane that was analyzed with a radioimager.

left and right panels). This result demonstrates that the covalent labeling of hMMP-3 involved exclusively residues of its active site. The marked pH dependence of hMMP-3 labeling by probe 1, in a range of 5-7 that encompasses the His pKa value (∼6), and the higher labeling observed at pH 7 suggested that probe 1 probably modifies His241 in hMMP-3, as His is a better nucleophile at pH 7 than at pH 5. To confirm the site of hMMP-3 modification, microsequencing and mass spectrometry experiments were carried out. Briefly, the protocol used relied on isolation of covalent hMMP-3/probe 1 complex by 1D-SDS gel, transfer of the corresponding protein species on PVDF membrane, and digestion first by trypsin and then by V-8 proteinase. Unmodified form of hMMP-3 was treated similarly for comparison. The µ-HPLC profiles of the peptide mixtures resulting from the double digestion of unlabeled and labeled hMMP-3 are reported in Figure 3. Comparison of these µ-HPLC profiles indicated that two peaks observed for unmodified hMMP-3 (labeled 1 and 2 in Figure 3, top panel) cannot be detected in the HPLC profile of modified hMMP-3 sample (Figure 3, lower panel). Mass analysis indicated that these two peaks correspond to two S1′ loop fragments of hMMP-3: the first one comprises residues from Ala234 to Arg248 and the second residues Ala234 to Asp245. Lack of these two peaks in the µ-HPLC profile corresponding to modified hMMP-3 suggests that probe 1 probably modifies residues of the S1′ loop, an event affecting their HPLC retention time. Indeed, in the modified sample, radioactivity analysis of the µ-HPLC flowthrough indicated the presence of a radioactive signal in fractions eluting at 25-26 min (Figure 3, lower panel). These radioactive fractions were collected and analyzed by Edman sequencing. N-terminal sequencing started at Ala234 and went to Tyr240; next, no amino acid could be assigned (His241 position, Figure 4). Afterward, the next sequencing step indicated a Ser242; then, the sequencing identified the expected S1′ loop sequence up to Arg248. Nano-ESI offline mass spec-

trometry analysis of the above purified radioactive fractions allowed the detection of three peaks (Figure 5), two doublecharged ions, [M1 + 2H]2+ ) 1094.6 and [M2 + 2H]2+ ) 1279.6 Da, respectively, and one triple-charged ion, [M1 + 3H]3+ ) 853.4 Da. These ions represent peptide fragments of respective mass: Mw (M1) ) 2187.2 and Mw (M2) ) 2557.2 Da. The mass at 2187.2 Da corresponds to the expected mass of the S1′ loop of hMMP-3 (residues Ala234 to Asp245, Mw(1) ) 1422.9 Da), after trypsin and V8 protease digestion, covalently modified by probe 1 (Mw ) 776.27 Da, minus 28 Da due to loss of N2 after irradiation of the azide group). The difference of 16 Da between the expected theoretical mass of the S1′ loop of hMMP-3 covalently modified by probe 1 and the observed mass is explained by Met236 oxidation, a residue present in the S1′ loop sequence of hMMP-3. A similar methionine oxidation was also observed in related covalently modified S1′ loop fragments of hMMP-12 (7). The second mass at 2557.2 Da is explained by similar arguments, except that the S1′ loop fragment possesses three additional residues (Leu246-Thr247-Arg248. This fragment, which was produced by trypsin cleavage, was not cleaved by the V8 protease. MS/MS experiments were performed on either double-charged or triple-charged ions described above in order to confirm by mass the site of covalent modification in hMMP3. However, despite several attempts using different values of collision energy for the peptide fragmentation, only peaks of extremely low intensities were observed in the corresponding MS/MS spectra, preventing the sequencing of the covalently modified peptide fragments. Covalent Labeling of MMP-12 Mutants. Probe 1 was previously shown to modify specifically the lysine residue located in position 241 of hMMP-12 (7), a position exhibiting high variability in MMPs (Ala(MMP8), Gln(MMP14), Thr(MMP2,13,11), Arg(MMP9), His(MMP3) (16). In order to see how the chemical structure of the side chain in position 241 influences the yield of covalent labeling by probe 1 in hMMP-12, independent of

Covalent Modification of MMPs by a Photoaffinity Probe

Bioconjugate Chem., Vol. 20, No. 2, 2009 371

Figure 3. µ-HPLC profile of a peptide mixture obtained by trypsin and V8-protease digestion of unmodified or covalently hMMP-3 by probe 1. UV trace (214 nm) of a PVDF digest (100 pmol) of unmodified (top panel) and modified h-MMP-3 (lower panel) by probe 1. Peaks labeled 1 and 2 in top panel correspond to fragments reported in this figure generated by trypsin and V8 protease cleavage of the S1′ loop (2 was not cleaved by the V8 protease). In the lower panel, radioactive fractions corresponding to fragments covalently modified by probe 1 are identified by a radioactive symbol.

Figure 4. Edman degradation of the purified fragment peptide of h-MMP-3 (25 pmol) covalently modified by probe 1.

the S1′ loop size which varies between MMPs, a series of mutants of hMMP-12 were produced by substituting position 241 in hMMP-12 with residues present in MMPs. Comparison of the catalytic efficiency of these mutants in cleaving a fluorogenic synthetic substrate specific for MMPs, as well as Ki values of probe 1 toward mutants, as compared with wildtype hMMP-12, indicated similar catalytic and recognition properties between mutants and wild-type enzyme (Table 1). Photolabeling of these mutants by probe 1, analyzed either by

silver staining or radioactivity counting, revealed a much lower cross-linking yield of the mutants (∼3-6% based on radioactivity), as compared with a cross-linking yield value of ∼45% observed in wild-type hMMP-12 (Figure 6). As previously reported, the presence of an alanine in position 241 results in lower cross-linking yield (1%) (7). Covalent Labeling of mMMP-12 and hMMP-9. mMMP12 and hMMP-9 both contain an arginine residue in position 241. Thus, the labeling of these two MMPs by probe 1 was

372 Bioconjugate Chem., Vol. 20, No. 2, 2009

Dabert-Gay et al.

Figure 5. Νano-ESI offline mass spectra of the radioactive fractions collected after µ-HPLC purification (shown in Figure 3). Molecular weights corresponding to the double- and triple-charged ion species observed in the mass spectrum are indicated. Table 1. Characterization of hMMP-12 Mutants (Ala, Arg, Gln, Thr, His)a -1 -1

kcat/Km (M .s ) Ki (nM) a

hMMP-12

Ala mutant

Arg mutant

Gln mutant

Thr mutant

His mutant

1.53 × 10 ( 0.03 × 104 0.17 ( 0.01

1.32 × 10 ( 0.05 × 104 0.35 ( 0.04

2.17 × 10 ( 0.04 × 104 0.15 ( 0.01

1.94 × 10 ( 0.01 × 104 0.27 ( 0.02

1.44 × 10 ( 0.02 × 104 0.16 ( 0.01

1.17 × 104 ( 0.03 × 104 0.30 ( 0.02

4

4

4

4

4

kcat/Km and Ki were determined in Tris/Hcl buffer 50 mM, pH 6.8, CaCl2 10 mM.

Figure 6. Labeling of hMMP-12 mutants by probe 1, visualized by silver staining (left panel) or radioimaging (right panel), as compared with hMMP-12 and hMMP-3. Conditions as described in Figure 1.

examined at pH 7 and 13. The pH dependence of probe 1 labeling was also determined toward the Arg-hMMP-12 mutant, for comparison. As shown in Figure 7, for the three proteins stronger labeling was observed by increasing the pH from 7 to 13. Differences in cross-linking yield were observed among the three proteins, at pH 13 (11% for Arg-MMP-12, 16% for mMMP-12, and 19% for hMMP-9). As compared with hMMP12, the cross-linking yield observed for these MMPs are smaller than that observed for hMMP-12 at pH 12 (85%).

DISCUSSION Previous studies of the identification of the site of hMMP12 covalent modification by photoactivated probe 1 have shown that this probe specifically modifies the ε amino group of Lys241 in this MMP (7). Taking into account the well-known efficiency of the nitrene intermediate in reacting with nucleophiles (11, 17), this result led us to determine the reactivity of probe 1 at pH 12, a pH at which the ε amino group of Lys241 should be in its

nonprotonated form (NH2 instead NH3+) and thus represents a better nucleophile. The 2-fold increase in cross-linking yield observed for the labeling of hMMP-12 by probe 1 between pH 7 and 12 supports the proposal that the nucleophilicity of Lys241 is in fact a key parameter that determines the cross-linking yield of probe 1. This observation suggests that other MMPs, possessing in position 241 a good nucleophile, like a lysine or any other amino acid bearing on its side chain a nucleophilic moiety, should also be modified by probe 1 with a high crosslinking yield. This led us to examine hMMP-3, the only MMP possessing in position 241 a good nucleophile too, a histidine residue. Indeed, a very high yield of cross-linking (65%) was observed for the covalent modification of this MMP by probe 1 at pH 7. In contrast, at pH 5, a pH at which the nitrogen atoms of the histidine side chain are weaker nucleophiles, a strong decrease of hMMP-3 labeling by probe 1 was observed. Microsequencing and mass spectrometry analyses identified the His241 as the unique site of modification of hMMP-3 by probe

Covalent Modification of MMPs by a Photoaffinity Probe

Bioconjugate Chem., Vol. 20, No. 2, 2009 373

Figure 7. pH dependence for the covalent modification of, from the left to the right, Arg241-hMMP-12 mutant, mMMP-12, and hMMP-9, as analyzed by radioimaging. Conditions as described in Figure 1.

1, a result explaining the observed pH cross-linking dependence. Altogether, these results again highlight the key role played by the nucleophilicity of the residue in position 241 of hMMP-3 and hMMP-12 for high labeling by probe 1. As mentioned in the results sections, the MS/MS experiments performed to identify the site of covalent modification of hMMP-3 by probe 1 have all failed. These negative results are in contrast with those reported for hMMP-12 (7). In this case, it was possible to sequence by MS/MS experiments the whole S1′ loop fragment covalently modified by probe 1 and identify formally the occurrence of the probe modification on the lysine residue. Thus, even with peptides of similar size and sequence, sophisticated techniques like mass spectroscopy may not provide the expected analytical results, rendering classical Edman sequencing a still powerful technique. According to X-ray and NMR studies, His241 in hMMP-3 and Lys241 in hMMP-12 are located on a loop segment (8-10) (10, 14), the S1′ loop, which exhibits high flexibility with these two residues displaying high mobility. Thus, the high yield of crosslinking observed for the modification of these two MMPs by probe 1 would be hard to predict taking into consideration this structural and dynamic information. The high cross-linking efficiency reported in this study implies the occurrence in enzyme/probe 1 complexes of conformations in which both Lys241 and His241 residues point toward the activated form of the probe 1 azide group, as shown in Figure 8. Accordingly, probe 1 can be viewed as a conformational probe providing clues about the structure and dynamics of MMPs (9). One important objective indicated by this study would be to determine whether the binding of probe 1, in its nonactivated or activated form, stabilizes a particular conformation of the hMMP-3 and hMMP-12 S1′ loops; as such, a conformation selection process may have important implications for the design of highly selective inhibitors of MMPs, a challenge only partially achieved today (18-20). The pH dependence observed for the efficient labeling of hMMP-3 and hMMP-12 led us to examine the importance of this parameter for the covalent modification of mMMP-12 and hMMP-9, two MMPs harboring an arginine in position 241. Increasing the pH form 7 to 13 resulted in a significant increase of the cross-linking yield (from 6% to 16-19% yield), a result suggesting that, in these two MMPs, like in hMMP-3 and hMMP-12, the amino acid side chain in position 241 is the target of probe 1 covalent modification. It should be noted that Arg241 is the only residue in the mMMP-12 and hMMP-9 active sites that can be affected by a pH variation from 7 to 13. The much

Figure 8. Superimposed model of probe 1 in complex with hMMP-12 and hMMP-3 showing the proximity between the azide group of probe 1 and the Lys241 and His241. Ribbon plot representation of the catalytic domain of hMMP-12 (black) and hMMP-3 (gray). Lys241 from hMMP12 is colored in yellow, His241 from hMMP-3 is blue, probe 1 in green, and catalytic zinc ion in purple.

lower yield of cross-linking observed for mMMP-12 and hMMP-9 highlights again the importance of the nucleophilicity of the residue in position 241, as the nitrogen atoms borne by the guanidinium moiety of arginine are weaker nucleophiles, as compared with the nitrogen present in the histidine and lysine residues. We previously reported that probe 1 covalently modifies MMPs, but with a very high variation in the crosslinking yield (4). The strict identification of the residue targeted by probe 1 in hMMP-3 and hMMP-12, plus the indirect proofs about this residue from the pH dependence observed for m-MMP-12 and hMMP-9, led us to suggest that this crosslinking yield dispersion reflects the variability of position 241 in MMPs (MMP-8(Ala241), MMP-2, 11,13(Thr241), MMP-10(Asn241), MMP-14(Gln241)). The presence in these MMPs at position 241 of residues harboring weaker nucleophile atoms on their side chain, as compared with hMMP-12 and hMMP-3, explains the

374 Bioconjugate Chem., Vol. 20, No. 2, 2009

lower cross-linking yield observed for these MMPs. Thus, one possibility to derive a probe displaying similar cross-linking yields toward all MMP members will be to design a new photoaffinity probe able to target conserved residues of the MMP S1′ cavity. Analysis of hMMP-12 mutants exemplifies that the presence of a nucleophile in the side chain of residue in position 241 is mandatory, as the presence of an alanine instead a of lysine results in less than 1% cross-linking. The site of covalent modification in these mutants has not been determined directly, but the pH dependence observed for the labeling of the Arg mutant is in agreement with the proposal that probe 1 targets the Arg241 in this protein. Mutants with a side chain in position 241 containing weaker nucleophiles (Arg, Gln, Thr), as compared with lysine, are labeled with a lower cross-linking yield. In contrast, the observation reported for the His mutant is quiet unexpected. This result illustrates that parameters other than the soil nucleophilicity of the residue in position 241 can determine the cross-linking yield. As discussed above, the conformational properties of the residue in position 241 also play a critical role. The same arguments probably explain the differences in cross-linking yield observed among Arg241hMMP-12, mMMP-12, and hMMP-9 at pH 13, despite the presence of the same arginine residue in these MMPs. Altogether, these results illustrate the difficulties that must be overcome in developing optimized activity-based probes incorporating photolabile groups, a strict requirement for the covalent labeling of zinc proteinases. Probe 1 should be useful for the detection of hMMP-12 and hMMP-3 in biological fluids. At neutral pH and in buffer, the detection limit of hMMP-12 and hMMP-3 with this radioactive probe is around 2.5 fmol, a limit that can be improved by a factor two for hMMP-12 by fixing the pH at 12 before the photoactivation step. Operating at basic pH seems to be a requirement for detecting mMMP-12 and hMMP-9. At basic pH, the detection limit with probe 1 should be around 150 fmol for these MMPs, a rather good sensitivity. However, even though the results of gelatin zymography should be interpreted with great caution before drawing conclusions about the presence of active forms of MMP-9 in biological fluids (21), it should be kept in mind that this technique is able to detect MMP-9 with a threshold of detection of 100 amol (22). Actual failure in detecting active forms of MMPs by using activitybased probes has been explained by arguing that these proteases are mostly expressed as inactive zymogen forms and that most of the activated MMP fractions are blocked by endogenous inhibitors (i.e., the TIMPs) (3). Thus, active forms of MMPs may exist at levels that are well below the current limit of activity-based probe detection, even when using the exceptional resolution and sensitivity of multidimensional liquid chromatography coupled with mass spectroscopy. To replace gelatin zymography or to be used in proteomic profiling by mass spectroscopy, activity-based probes for detection of the MMP active form need further improvement. Our results give pointers to the development of a new generation of those probes that should be more efficient for the equivalent detection of all MMPs.

ACKNOWLEDGMENT This work was supported by funds from the Commissariat a` l’Energie Atomique and from the European Commission (FP6RT, Cancer Degradome project, LSHC-CT-2003-503297).

LITERATURE CITED (1) Saghatelian, A., Jessani, N., Joseph, A., Humphrey, M., and Cravatt, B. F. (2004) Activity-based probes for the proteomic

Dabert-Gay et al. profiling of metalloproteases. Proc. Natl. Acad. Sci. U.S.A. 101, 10000–5. (2) Chan, E. W., Chattopadhaya, S., Panicker, R. C., Huang, X., and Yao, S. Q. (2004) Developing photoactive affinity probes for proteomic profiling: hydroxamate-based probes for metalloproteases. J. Am. Chem. Soc. 126, 14435–46. (3) Sieber, S. A., Niessen, S., Hoover, H. S., and Cravatt, B. F. (2006) Proteomic profiling of metalloprotease activities with cocktails of active-site probes. Nat. Chem. Biol. 2, 274–81. (4) David, A., Steer, D., Bregant, S., Devel, L., Makaritis, A., Beau, F., Yiotakis, A., and Dive, V. (2007) Cross-linking yield variation of a potent matrix metalloproteinase photoaffinity probe and consequences for functional proteomics. Angew. Chem., Int. Ed. Engl. 46, 3275–7. (5) Cravatt, B. F., Wright, A. T., and Kozarich, J. W. (2008) Activity-based protein profiling: from enzyme chemistry to proteomic chemistry. Annu. ReV. Biochem. 77, 383–414. (6) Kato, D., Boatright, K. M., Berger, A. B., Nazif, T., Blum, G., Ryan, C., Chehade, K. A., Salvesen, G. S., and Bogyo, M. (2005) Activity-based probes that target diverse cysteine protease families. Nat. Chem. Biol. 1, 33–8. (7) Dabert-Gay, A. S., Czarny, B., Devel, L., Beau, F., Lajeunesse, E., Bregant, S., Thai, R., Yiotakis, A., and Dive, V. (2008) Molecular determinants of matrix metalloproteinase-12 covalent modification by a photoaffinity probe: Insights into activity-based probe development and conformational variability of matrix metalloproteinases. J. Biol. Chem. 283, 31058–67. (8) Nar, H., Werle, K., Bauer, M. M., Dollinger, H., and Jung, B. (2001) Crystal structure of human macrophage elastase (MMP12) in complex with a hydroxamic acid inhibitor. J. Mol. Biol. 312, 743–51. (9) Bertini, I., Calderone, V., Cosenza, M., Fragai, M., Lee, Y. M., Luchinat, C., Mangani, S., Terni, B., and Turano, P. (2005) Conformational variability of matrix metalloproteinases: beyond a single 3D structure. Proc. Natl. Acad. Sci. U.S.A. 102, 5334– 9. (10) Bhaskaran, R., Palmier, M. O., Bagegni, N. A., Liang, X., and Van Doren, S. R. (2007) Solution structure of inhibitor-free human metalloelastase (MMP-12) indicates an internal conformational adjustment. J. Mol. Biol. 374, 1333–44. (11) Kotzyba-Hibert, F., and Goeldner, M. (1995) Recents trends in photoaffinity labeling. Angew. Chem., Int. Ed. Engl. 34, 1296– 1312. (12) Suzuki, K., Kan, C. C., Hung, W., Gehring, M. R., Brew, K., and Nagase, H. (1998) Expression of human pro-matrix metalloproteinase 3 that lacks the N-terminal 34 residues in Escherichia coli: autoactivation and interaction with tissue inhibitor of metalloproteinase 1 (TIMP-1). Biol. Chem. 379, 185–91. (13) Horovitz, A., and Levitzki, A. (1987) An accurate method for determination of receptor-ligand and enzyme-inhibitor dissociation constants from displacement curves. Proc. Natl. Acad. Sci. U.S.A. 84, 6654–8. (14) Alcaraz, L. A., Banci, L., Bertini, I., Cantini, F., Donaire, A., and Gonnelli, L. (2007) Matrix metalloproteinase-inhibitor interaction: the solution structure of the catalytic domain of human matrix metalloproteinase-3 with different inhibitors. J. Biol. Inorg. Chem. 12, 1197–206. (15) Harrison, R. K., Chang, B., Niedzwiecki, L., and Stein, R. L. (1992) Mechanistic studies on the human matrix metalloproteinase stromelysin. Biochemistry 31, 10757–62. (16) Cuniasse, P., Devel, L., Makaritis, A., Beau, F., Georgiadis, D., Matziari, M., Yiotakis, A., and Dive, V. (2005) Future challenges facing the development of specific active-site-directed synthetic inhibitors of MMPs. Biochimie 87, 393–402. (17) Fleming, S. A. (1995) Chemical reagents in phtoaffinity labeling. Tetrahedron 51, 12479–12520. (18) Overall, C. M., and Kleifeld, O. (2006) Validating matrix metalloproteinases as drug targets and anti-targets for cancer therapy. Nat. ReV. Cancer 6, 227–239.

Covalent Modification of MMPs by a Photoaffinity Probe (19) Fingleton, B. (2007) Matrix metalloproteinases as valid clinical targets. Curr. Pharm. Des. 13, 333–46. (20) Yiotakis, A., and Dive, V. (2008) Synthetic active site-directed inhibitors of metzincins achievement and perspectives. Mol. Aspects Med. 29, 329–38. (21) Hesek, D., Toth, M., Meroueh, S. O., Brown, S., Zhao, H., Sakr, W., Fridman, R., and Mobashery, S. (2006) Design and characterization of a metalloproteinase inhibitor-tethered resin

Bioconjugate Chem., Vol. 20, No. 2, 2009 375 for the detection of active MMPs in biological samples. Chem. Biol. 13, 379–86. (22) Masure, S., Proost, P., Van Damme, J., and Opdenakker, G. (1991) Purification and identification of 91-kDa neutrophil gelatinase. Release by the activating peptide interleukin-8. Eur. J. Biochem. 198, 391–8. . BC800478B