Probing the Active Site of the Hepatitis C Virus Serine Protease by Fluorescence Resonance Energy Transfer*

A serine protease domain contained within the viral NS3 protein is a key player in the maturational process-ing of the hepatitis C virus polyprotein and a prime target for the development of antiviral drugs. In the present work, we describe a dansylated hexapeptide inhibitor of this enzyme. Active site occupancy by this compound could be monitored following fluorescence resonance energy transfer between the dansyl fluorophore and protein tryptophan residues and could be used to 1) unambiguously assess active site binding of NS3 protease inhibitors, 2) directly determine equilibrium and pre-steady-state parameters of enzyme-inhib-itor complex formation, and 3) dissect, using site-di-rected mutagenesis, the contribution of single residues of NS3 to inhibitor binding in direct binding assays. The assay was also used to characterize the inhibition of the NS3 protease by its cleavage products. We show that enzyme-product inhibitor complex formation depends on the presence of an NS4A cofactor peptide. Equilibrium and pre-steady-state data support an ordered mechanism of ternary (enzyme-inhibitor-cofactor) complex formation, requiring cofactor complexation prior to inhibitor binding.

A serine protease domain contained within the viral NS3 protein is a key player in the maturational processing of the hepatitis C virus polyprotein and a prime target for the development of antiviral drugs. In the present work, we describe a dansylated hexapeptide inhibitor of this enzyme. Active site occupancy by this compound could be monitored following fluorescence resonance energy transfer between the dansyl fluorophore and protein tryptophan residues and could be used to 1) unambiguously assess active site binding of NS3 protease inhibitors, 2) directly determine equilibrium and pre-steady-state parameters of enzyme-inhibitor complex formation, and 3) dissect, using site-directed mutagenesis, the contribution of single residues of NS3 to inhibitor binding in direct binding assays. The assay was also used to characterize the inhibition of the NS3 protease by its cleavage products. We show that enzyme-product inhibitor complex formation depends on the presence of an NS4A cofactor peptide. Equilibrium and pre-steady-state data support an ordered mechanism of ternary (enzyme-inhibitor-cofactor) complex formation, requiring cofactor complexation prior to inhibitor binding.
The NS3 protein of the hepatitis C virus (HCV) 1 is a multifunctional polypeptide. Its C-terminal two-thirds encompasses an RNA helicase, thought to be involved in the separation of RNA double strand structures that transiently form during the replication of the viral genome (1,2). The N-terminal third of the protein, instead, harbors a serine protease domain responsible for proteolytic maturation of most of the nonstructural region of the viral polyprotein (reviewed in Ref. 3). Its activity is thought to be absolutely required for the assembly of the viral replication machinery. Both enzymatic functions are presently the focus of intensive research, since their inhibition is considered as a possible strategy for the development of anti-viral pharmaceuticals (4 -6).
The NS3 protein shows only very weak proteolytic activity on its own and complex formation with the viral protein NS4A is required to activate its catalytic functions (7)(8)(9)(10). In vitro, activation can be obtained by the addition of synthetic peptides encompassing residues 21-34 of the NS4A cofactor to the purified enzyme (11)(12)(13)(14)(15). The x-ray crystal structures of the uncomplexed protease domain and of the binary serine proteasecofactor peptide complex have been determined (16 -18). Based on these structures, it was proposed that the cofactor activates the enzyme by promoting a correct positioning of the residues that form its catalytic triad (19).
Both kinetic and structural data have shed light on the mechanisms by which the enzyme recognizes its substrates (13, 20 -22). The minimum length for peptide substrates was shown to be a decamer spanning from P6 to P4Ј (13). This requirement for rather large substrate molecules stems from the peculiar architecture of the substrate binding site that lacks all major loops that form the S2 or S3 binding pockets in other enzymes of this family. Specific binding is accomplished through a series of weak interactions that are distributed along an extended recognition surface. The driving forces for enzyme-substrate complex formation include, among others, 1) electrostatic interactions between the conserved negatively charged residue in the P6 position of substrates and a cluster of positively charged residues of the enzyme, 2) main chain interactions involving the substrate peptide backbone that forms an antiparallel ␤-sheet with strand E2 of the enzyme, 3) interactions between the preferred P1 cysteine residue and the S1 pocket of the enzyme that is characterized by the presence of a phenylalanine residue at its entrance, and 4) hydrophobic interactions between the P4Ј residue and a small pocket formed also by residues of the NS4A cofactor (22).
We have recently shown that the NS3 protease undergoes a remarkable inhibition by its N-terminal cleavage products (23). This has subsequently allowed us to generate hexapeptide inhibitors with affinities in the low nanomolar range, using combinatorial techniques (24). The binding mode of these molecules is of considerable interest, since they may be regarded as a starting point for the development of nonpeptidic inhibitors. Structure-activity relationships, molecular modeling, site-directed mutagenesis, and NMR spectroscopy have been used to gain knowledge about the salient features of the enzyme-inhibitor interactions (23)(24)(25)(26). Two major acidic "anchoring" points have been identified in those studies: an interaction involving an acidic function in P6/P5 and another electrostatic interaction involving the P1 ␣-carboxylic acid generated upon cleavage of the parent substrate peptide. This carboxylate group was proposed to engage in hydrogen bond interactions with the backbone amides of the oxyanion hole, the protonated ⑀-N of the catalytic histidine, and the ␥-amino group of lysine 136 (23).
In the present study, we have generated a dansyl-labeled hexapeptide acid inhibitor of the NS3 protease, termed compound P, in an attempt to probe the active site of the enzyme and to further characterize its interactions with inhibitors. Binding of this compound to the active site of the protease could be directly monitored by measuring fluorescence resonance energy transfer between tryptophan residues of the protein and the fluorescent group in the bound inhibitor. This technique allowed us to gain insight into the dynamics of enzyme-inhibitor complex formation and to study the influence of the cofactor on the binding reaction. Furthermore, we show that, using the fluorescence energy transfer signal, the contribution of single residues to formation of the enzyme-inhibitor complex can be dissected by site-directed mutagenesis even if such mutations abolish the catalytic activity of the enzyme.
(mp 123-124°C). 1  The peptide was assembled by machine-assisted solid phase synthesis using the continuous flow Fmoc-polyamide method (27). The resin used was Tentagel® (0.19 mM eq/g) derivatized with a hydroxymethylphenoxyacetic acid linker. All of the couplings were performed with a 5-fold excess of activated amino acid over the resin free amino groups, using Fmoc-amino acid/PyBOP/1-hydroxybenzotriazole/N,N-diisopropylethylamine (1:1:1:2) activation; the synthesis was interrupted after 4 cycles to allow manual incorporation of Dap(Dansyl) (PyBOP activation, 4-fold excess); coupling completion was checked by the ninhydrin test before resuming the machine assembly. At the end of the synthesis, the dry peptide-resin was treated with reagent B (28) (trifluoroacetic acid/phenol/water/triisopropylsilane (88:5:5:2)) for 1.5 h at room temperature. The resin was filtered out, and the peptide was precipitated with cold methyl-t-butyl ether. The precipitate was redissolved in 1:1 water/acetonitrile containing 0.1% trifluoroacetic acid and lyophilized. Purification to Ͼ98% homogeneity was achieved through preparative HPLC on a Waters C-18 column (250 ϫ 100 mm, 15 m) using as eluents water, 0.1% trifluoroacetic acid (A) and acetonitrile, 0.1% trifluoroacetic acid (B), and a linear gradient of 35-60% B over 30 min. The fractions corresponding to the pure peptide were pooled and lyophilized. MS (calculated) 1087.27; (found) 1087. 4.
Enzyme Preparations and Site-directed Mutagenesis-Escherichia coli BL21(DE3) cells were transformed with plasmids containing the cDNA coding for the serine protease domain of the HCV J strain NS3 protein (amino acids 1027-1206, followed by the sequence ASKKKK) under the control of the bacteriophage T7 gene 10 promoter in a pT7-7 vector. The catalytic serine to alanine mutation was introduced in the context of the NS3 protease domain spanning residues 1027-1206 of the HCV Bk strain by polymerase chain reaction-directed mutagenesis. Clones were fully sequenced on both strands to exclude the introduction of additional mutations by polymerase chain reaction. Experiments using this enzyme were performed in parallel with the wild type Bk enzyme. All enzymes were purified as described previously (29) and found to be Ͼ95% pure as judged from reversed phase HPLC performed using a 4.6 ϫ 250-mm Vydac C4 column. Enzyme preparations were routinely checked by mass spectrometry done on HPLC-purified samples using a Perkin-Elmer API 100 instrument and by N-terminal sequence analysis carried out using Edman degradation on an Applied Biosystems model 470A gas-phase sequencer. Enzyme stocks were quantified by amino acid analysis, shock-frozen in liquid nitrogen, and kept in aliquots at Ϫ80°C until use.
Enzymatic activities were determined using the substrate Ac-DE-MEECASHLPYK-NH 2 , based on the sequence of the NS4AB cleavage site of the HCV polyprotein. Assays were done using 10 nM NS3 protease in 50 mM Hepes, pH 7.5, 1 mM DTT, 15% glycerol, 1% CHAPS containing 80 M NS4A cofactor peptide (Pep4AK with the sequence KKKGSVVIVGRIILSGR-NH 2 , (15)) and analyzed by HPLC as described previously (20). Cleavage products were quantified by integration of chromatograms with respect to appropriate standards. Kinetic parameters were calculated from nonlinear least squares fit of initial rates as a function of substrate concentration with the help of Kaleidagraph software, assuming Michaelis-Menten kinetics. The K i value of compound P was determined from substrate titration experiments performed at different concentrations of probe. At least four titration curves were simultaneously fitted to Equation 1 using Sigmaplot software.
Fluorescence Measurements-Fluorescence emission spectra were recorded on a Perkin-Elmer LS50B instrument with a cuvette holder thermostatted at 23°C. 200 nM protease added to 2.5 ml of 50 mM Hepes, pH 7.5, 1 mM DTT, 15% glycerol, 1% CHAPS containing 80 M Pep4AK. Increasing concentrations of compound P were added, and emission spectra were recorded between 300 and 550 nm at a scan speed of 60 nm/min upon excitation at 280 nm. Emission and excitation slits were opened to 2.5 and 5 nm, respectively. Spectra were routinely corrected for the background signal of buffer and for dilution effects. The fluorescence intensity, F, is related to the concentration of enzymeprobe complex according to Equation 2, where F 0 indicates the fluorescence intensity in the absence of probe, f represents the intrinsic fluorescence of the enzyme-probe complex, and [EP] represents its concentration. Since enzyme and probe concentrations were similar in the titration experiments, it could no longer be assumed that the free ligand was equal to the total added ligand. Under these conditions, the concentration of EP is related to the total concentrations of enzyme and probe according to Equation 3, where K P is the equilibrium dissociation constant of the enzyme-probe complex. Substitution of [EP] in Equation 2 with Equation 3 yields an equation that directly relates the fluorescence intensity with K P . To obtain the equilibrium dissociation constant of the enzyme-probe complex, this equation was fitted to the experimental titration data with the help of Kaleidagraph software. Equilibrium dissociation constants of nonfluorescent ligands were obtained from probe displacement experiments performed using 100 nM protease in 2.5 ml of 50 mM Hepes, pH 7.5, 1 mM DTT, 15% glycerol, 1% CHAPS containing 80 M Pep4AK and 500 nM compound P. Under these conditions, depletion of probe by enzyme-probe complex formation was considered to be negligible. The equilibria involved in the competition of compound P and a nonfluorescent ligand (L) for the active site of the enzyme are outlined in Equation 4, where K P and K L are the equilibrium dissociation constants of the EP and EL complexes, respectively. It can be shown that in the presence of the competing ligand L, the concentration of EP is a function of both the concentration of ligand and its equilibrium dissociation constant K L according to Equation 5, . Among the three enzyme species, only EP is fluorescent. Hence, the measured fluorescence intensity F is only a function of the concentration of EP according to Equation 2. At an infinite concentration of compound P, the enzyme will be saturated with the probe, and the fluorescence will approach the its maximum value, F max . The relative fluorescence intensity, F r , is defined as the ratio F/F max and corresponds to the fractional occupancy of the active site by the probe. Equation 5 can therefore be rewritten as follows.
This equation directly relates fluorescence to the concentration of competing nonfluorescent ligand L and allows the determination of its equilibrium dissociation constant from nonlinear regression analysis of titration data.
Pre-steady-state Measurements-Association rate constants were determined on an SX-MV18 Applied Photophysics stopped-flow instrument equipped with a fluorescence detector and interfaced with a Risc computer. The samples and the flow cell were thermostatted at 23°C. To 70 -200 nM protease in 50 mM Hepes, pH 7.5, 1 mM DTT, 15% glycerol, 1% CHAPS containing 160 M NS4A cofactor peptide (final concentrations), increasing concentrations of compound P were added. Under these conditions, we measured a dead time of 1.8 ms with a standard reference reaction (reduction of 2,6-dichlorindophenol by ascorbic acid; Ref. 30). Protein fluorescence was excited at 280 nm, and a 400-nm cut-off filter was used to suppress both the excitation wavelength and the tryptophan emission band. Time-dependent fluorescence changes upon the addition of compound P to the NS3 protease were recorded under pseudo-first order conditions for at least 5 ϫ t1 ⁄2 and could be best fitted with a single exponential equation. k obs values derived from this fitting procedure were determined at different concentrations of compound P and used to calculate association rate constants according to Equation 7.
The dissociation rate constant k off was estimated from the y axis intercept of k obs versus [P] plots. Association rate constants of inhibitors were determined by following the kinetics of probe displacement upon addition of the inhibitor to a solution containing 100 nM protease in 50 mM Hepes, pH 7.5, 1 mM DTT, 15% glycerol, 1% CHAPS, 80 M NS4A, and 10 M compound P (final concentrations). Observed rate constants were obtained by fitting experimental data with a single exponential equation by least squares nonlinear regression analysis. Association rate constants were obtained from these data following the procedure described under "Results."

RESULTS
Competitive inhibitors containing fluorescent groups have been widely used to probe the active site of proteolytic enzymes (31)(32)(33)(34)(35)(36). Upon binding to the enzyme, the fluorescent moiety may enter into a region of lower polarity, which usually results in a blue shift of emission maxima and in an increased fluorescence yield. The active site of the NS3 protease is peculiarly flat and solvent-exposed, suggesting that fluorescent groups attached to an active site binder of this enzyme will experience only minor changes in hydrophobicity once complexed with the enzyme. In such a case, active site binding may still be monitored if the fluorescent group will be positioned in the vicinity of protein tryptophan residues, thus enabling the fluorophore to accept excitation energy from these residues. The dansyl group, attached to an active site binder has spectroscopic properties that are suitable for this purpose (32). In the light of these considerations, we attempted to develop a dansylated active site probe for the NS3 protease. We took advantage of our finding that hexamer N-terminal cleavage products of NS3 substrate peptides, spanning residues P6 -P1, bind to the enzyme's active site with higher affinity than the corresponding substrates (23). We also considered the structure-activity relationship of these product inhibitors (24). Starting from the product of the NS4A/NS4B cleavage site with the sequence DEMEEC-OH, we substituted the P2 glutamic acid with a cyclohexylalanine (Cha) to improve affinity and introduced the dansyl group, linked to diaminopropionic acid, in the P4 position. This choice was again guided by the product inhibitor structure-activity relationship, showing that aromatic residues like diphenylalanine are well accepted in this position (24). Molecular modeling has shown that in the complex of the resulting hexapeptide, having the sequence Ac-D-E-Dap(N-␤-Dansyl)-E-Cha-C-OH (compound P), and the NS3-Pep4AK protease complex, the dansyl moiety will be located at a distance Ͻ20 Å from both Trp 85 and Trp 53 (data not shown). This distance is compatible with a transfer of excitation energy between either tryptophan residue and the fluorescent group in the probe. Compound P was synthesized and characterized by steady-state kinetics, demonstrating that the compound reversibly inhibited the NS3 protease and was competitive with a peptide substrate (Fig. 1). From the substrate competition experiment, K i ϭ 200 nM could be calculated.
We next explored whether binding of P to the active site of the NS3-Pep4AK protease complex went along with changes in dansyl fluorescence. The fluorescence spectrum of the compound alone showed excitation and emission maxima at 330 and 510 nm, respectively. The addition of NS3 protease, together with its cofactor peptide Pep4AK, resulted in a weak increase in intensity, consistent with the dansyl group experiencing a slightly more hydrophobic environment upon binding to the enzyme's active site (data not shown). In addition, compound P decreased the intrinsic tryptophan fluorescence of the NS3-Pep4AK protease complex at 330 nm and concomitantly gave rise, upon excitation at 280 nm, to an emission band centered at 510 nm (Fig. 2). This additional band must result from emission of the dansyl moiety upon energy transfer from the NS3 tryptophan residue(s), which emit at 330 nm (15). Unbound compound P showed no significant fluorescence at 510 nm if excited at 280 nm in the absence of NS3 (not shown). Due to this marked difference, we decided to take advantage of the phenomenon of fluorescence resonance energy transfer to characterize ligand binding to the active site of the protease in more detail.
Characterization of the Interaction of the NS3 Protease with Compound P-We first investigated the dependence of the fluorescence emission at 510 nm on the concentration of compound P. Fig. 2 shows that the fluorescence increase was saturable at probe concentrations above 800 nM. Titration data were fitted to Equations 2 and 3 (see "Materials and Methods"), yielding an equilibrium dissociation constant of 180 nM, which is in close agreement with the K i value of 200 nM reported above. Monitoring the decrease in tryptophan fluorescence as a function of added probe ( Fig. 2A), we obtained a dissociation constant of 190 nM, which again is in good agreement with both the K i value and the equilibrium dissociation constant obtained monitoring fluorescence emission at 510 nm. This analysis is based on the assumption of a 1:1 stoichiometry of the NS3-Pep4AK-compound P complex. The validity of this assumption was experimentally verified by another set of fluorescence titrations done at an enzyme concentration 10 times higher than the equilibrium dissociation constant (Fig. 2B). Under these conditions tight binding occurs, which allowed the determination of a stoichiometry of 1:1.08 of the binding reaction.
Next, the influence of complexation of the NS3 protease with its cofactor peptide, Pep4AK, on the fluorescence resonance energy transfer with compound P was investigated. A decrease in the concentration of Pep4AK went along with a decrease in fluorescence emission by compound P at 510 nm (Fig. 3). This decrease titrated with a half-maximal saturation at 4.8 M Pep4AK. In the absence of Pep4AK, no fluorescence at 510 nm was observed up to a concentration of 1.6 M of compound P (not shown). This may be due to a different structure of the NS3-compound P complex in the absence of cofactor or, more trivially, to the fact that the affinity of the enzyme for compound P decreases substantially in the absence of Pep4AK and that therefore no detectable NS3-compound P complex is formed in the concentration range that we have explored. This was investigated by determining the capability of compound P to inhibit the activity of the NS3 protease domain in the absence of Pep4AK using a peptide substrate. No inhibition was detected up to a concentration of 80 M compound P (not shown). Since this experiment was performed at S ϭ K m this result implies that in the absence of Pep4AK the K i value of compound P is higher than 40 M.
The interaction of compound P with the NS3 protease in the presence of Pep4AK was further investigated under pre-equilibrium conditions using a stopped-flow instrument. The emission above 400 nm was monitored upon excitation at 280 nm (Fig. 4A). Association experiments were performed at different probe concentrations, and observed rate constants were plotted as a function of probe concentration (Fig. 4A, inset). From these plots, a very fast second order association rate constant of 1.5 ϫ 10 8 M Ϫ1 s Ϫ1 could be calculated, indicating that the association reaction is probably diffusion-limited. From the intercept of the plot of k obs versus probe concentration, a dissociation rate constant of 70 s Ϫ1 was estimated. The ratio of the two rate constants gave a value of of 460 nM. We notice that this value differs from the K i value of 200 nM and the dissociation constants of 190 and 180 nM determined under equilibrium conditions. The concentration range of compound P that could be explored was rather limited due to the very fast association process, which exceeded the resolution limit of the stopped-flow instrument at concentrations higher than 2 M. Therefore, it is impossible to determine whether the association process conforms to a single-step equilibrium or whether the slight discrepancy between the ratio of the two rate constants and the equilibrium dissociation constant is indicative of a more complicated mechanism.
The association rate constant of compound P did not change significantly when the Pep4AK concentration was lowered. However, the amplitude of the fluorescence change decreased with decreasing Pep4AK concentration. At Pep4AK concentrations below 160 M, the reaction became multiphasic, with a slow phase being observable on a 100-s time scale (Fig. 4B). At equilibrium, upon the addition of 4 M probe, the same level of fluorescence was observed in the presence of either 160 or 10 M Pep4AK, implying that in both cases a quantitative formation of a ternary NS3-Pep4AK-compound P complex ultimately occurred.
A possible rationale for this behavior can be derived from the analysis of the equilibria involving enzyme, cofactor, and active site ligand that are summarized in Fig. 5 assuming either a random or an ordered mechanism of compound P and Pep4AK binding to the enzyme. In a random mechanism (Fig. 5B), ternary NS3-Pep4AK-compound P complex formation can occur either via the NS3-Pep4AK complex or via the NS3-compound P complex, the preferred route depending on the [Pep4AK]/K d and [P]/K i ratios chosen for the experiment. In contrast, an ordered mechanism (Fig. 5A) implies the exclusive formation of the ternary complex via the NS3-Pep4AK complex, irrespective of the concentration of the single reactants in solution. Both fluorescence and activity data suggest that no substantial NS3-compound P complex is formed up to concentrations of compound P of 200 ϫ K i , thus strongly favoring the ordered mechanism.
Use of Probe Displacement in the Characterization of Activesite Binders of the NS3 Protease-We wanted to assess whether displacement of compound P from the active site of the NS3-Pep4AK protease complex can be accomplished by the addition of nonlabeled active-site binders, thus allowing the calculation of equilibrium dissociation constants of competitive inhibitors of the NS3-Pep4AK protease complex by probe displacement. To this purpose, increasing amounts of the product inhibitor DEMEEC-OH were added to a solution containing NS3 protease, saturating concentrations of Pep4AK cofactor, and compound P. As shown in Fig. 6, DEMEEC-OH was capable of quantitatively displacing the bound probe from the enzyme's active site in a dose-dependent way, as judged from the decrease of the intensity of the 510-nm emission band. From the displacement curve, the equilibrium dissociation constant for DEMEEC-OH can be calculated using the equations specified under "Materials and Methods." The value we obtained was 500 nM, which is in good agreement with the K i value of 600 nM reported previously (23).
We noticed that, in contrast to what has been observed with compound P, the affinity of the enzyme for the product inhibitor DEMEEC-OH was less affected by the absence of the NS4A cofactor peptide. In fact, using kinetic assays, we determined that DEMEEC-OH inhibited the uncomplexed protease with K i ϭ 6 M, a value that is only 10-fold higher than the one obtained in the presence of saturating amounts of Pep4AK (see above). Given these differences, we decided to investigate the interaction of the enzyme with a third product inhibitor having the optimized sequence DEDifEChaC-OH (24). This compound inhibited the NS3-Pep4AK complex with K i ϭ 30 nM, whereas it lost more than 200-fold in potency on the free enzyme (K i ϭ 7 M).
Characterization of Serine Trap Inhibitors of the NS3-Pep4AK Protease Complex by Probe Displacement-We observed that binding of compound P to the enzyme's active site does not require the presence of the catalytic serine residue. In fact, compound P bound with the same affinity to an NS3-Pep4AK protease complex in which the catalytic serine had been mutagenized into alanine, as detected by a dose-dependent increase in the emission at 510 nm (not shown), although the intensity of the resonance phenomenon was slightly lower. This observation can be used to characterize inhibitors of the FIG. 4. Pre-steady-state association kinetics of compound P and the NS3-Pep4AK protease. A, to 70 nM NS3 protease in 50 mM Hepes, pH 7.5, 1 mM DTT, 15% glycerol, 1% CHAPS, 160 M Pep4AK, 250 nM P were added, and the association reaction was followed using a stopped-flow instrument equipped with a fluorescence detector. Tryptophan fluorescence was excited at 280 nm, and dansyl emission was recorded Ͼ400 nm using an appropriate cut-off filter. Four runs were averaged. Data were fitted with the equation where F 0 is the initial fluorescence, A is the amplitude of the fluorescence change, and k obs is the observed pseudo-first order rate constant for the approach of equilibrium. Inset, k obs values were plotted as a function of the concentration of compound P to derive values for the second order association rate constant and for the first order dissociation rate constant as described under "Materials and Methods." B, to 100 nM NS3 protease in 50 mM Hepes, pH 7.5, 1 mM DTT, 15% glycerol, 1% CHAPS, 4 M compound P was added upon preincubation of the enzyme with 10 M Pep4AK (curve b) or 160 M Pep4AK (curve c). The fluorescence change of Ͼ400 nm was followed as described above. Curve a shows the control curve in the absence of enzyme. Four runs were averaged for each curve. protease that are targeted at the catalytic serine residue. The hexapeptide aldehyde DEDifEChaAbuF 2 -CHO was found to be a potent inhibitor of the protease with K i ϭ 0.5 nM. 2 Aldehydes were shown to interact with the active site serine residue of serine proteases via the formation of a reversible hemiacetal bond (38,39). Hemiacetal bond formation in the NS3-DEDi-fEChaAbuF 2 -CHO complex has been inferred from NMR and kinetic data. 2 The addition of this compound to a solution containing NS3 protease, Pep4AK cofactor and compound P resulted in a displacement of the probe from the active site, as judged by a decrease of the intensity of the emission band at 510 nm (Fig. 7). Since the assay was performed at an enzyme concentration of 200 nM, which is much larger than the K i value of the inhibitor, a redetermination of the potency of the aldehyde was not possible using the probe displacement procedure. In fact, the compound was found to stoichiometrically titrate the enzyme, leading to quantitative displacement of compound P at aldehyde concentrations above 400 nM. In contrast, when the titration was repeated using an active site serine-to-alanine mutant enzyme, no displacement of the probe was noticed up to an aldehyde concentration of 800 nM, indicating the involvement of the catalytic serine residue in NS3-DEDi-fEChaAbuF 2 -CHO complex formation. Displacement of compound P by added inhibitors can also be used to determine the pre-steady-state association rate constants of active site ligands. The velocity of displacement of compound P from the protease is related to the on-rate of an added inhibitor according to Equation 8.
holds only if probe dissociation does not become rate-limiting, allowing the accurate determination of pseudofirst order rate constants (k on [I]) of inhibitor association that are significantly slower than the off-rate of the probe. The first order dissociation rate constant of the probe from the NS3-Pep4AK-compound P complex has been estimated to be 70 s Ϫ1 (see above), indicating that inhibitor association is the ratelimiting event if pseudo-first order rate constants well below this value are determined. This was verified using the aldehyde inhibitor DEDifEChaAbuF 2 -CHO. Fig. 8 shows Experiments were repeated at different DEDifEChaAbuF 2 -CHO concentrations, and the observed rate constants k obs were plotted as a function of inhibitor concentration (Fig. 8, inset).
The association rate constant of the inhibitor can be calculated from these data, taking into account the fractional occupancy of the enzyme by the probe, which is related to the concentration of the probe and its equilibrium dissociation constant according to Equation 8. Having determined the K d value of compound P to be 200 nM, we calculated a second order rate constant, k on ϭ 2.7 ϫ 10 6 M Ϫ1 s Ϫ1 , for the formation of the NS3-Pep4AK-DEDifEChaAbuF 2 -CHO complex. This value compares well to the previously reported value of 2.2 ϫ 10 6 M Ϫ1 s Ϫ1 (37). DISCUSSION We herein report the design, synthesis, and characterization of a fluorescent probe of the active site of the hepatitis C virus NS3/4A protease, a primary drug development target. The possibility to directly determine binding of ligands to the active site of an enzyme in a way that does not depend on the measurement of its catalytic activity has a number of advantages that have driven this work, such as the unambiguous assessment of active site binding inhibitors and the possibility of determining true equilibrium dissociation constants in probe displacement assays. This is particularly relevant in the case of the NS3/4A protease that was observed to undergo inhibition by its cleavage products (23). In fact, the resulting inevitable accumulation of inhibitory products during kinetic assays was shown to lead to an overestimation of true K i values of competitive inhibitors (40).
A hexamer product inhibitor was used as a scaffold for the fluorophore that was introduced, by solid phase synthesis, via a dansylated diaminopropionic acid moiety, yielding the probe compound P.
The following pieces of evidence document the usefulness of compound P as an active site probe for the NS3-Pep4AK protease: The compound was shown to be a competitive inhibitor of the enzyme and bound to it with a 1:1 stoichiometry. Binding of compound P to the NS3-Pep4AK protease-activator complex also resulted in both an enhancement of fluorescence emission at 510 nm and a concomitant quenching of tryptophan fluorescence. Both phenomena titrated with equilibrium dissociation constants that were in good agreement with the K i value of compound P. Altogether the data demonstrate that, in the explored concentration range, compound P exclusively bound to the active site of the NS3-Pep4AK protease complex and that this binding event went along with a transfer of fluorescence energy from protein tryptophan residues to the dansyl group of the probe.
We have shown that compound P can be used to 1) unambiguously assess active site binding of inhibitors, 2) directly determine steady-state dissociation constants and pre-steadystate association rate constants of active site ligands of the NS3-Pep4AK protease complex, and 3) dissect the contribution of single residues of NS3 to inhibitor binding by site-directed mutagenesis even if such mutations abolish the catalytic activity of the protease. In fact, we have found that compound P binds with unaltered affinity to a catalytic serine to alanine mutant of the NS3-Pep4AK protease complex, which allowed us to use a probe displacement assay as a mechanistic tool for the rapid identification of compounds that bind to the active site of the enzyme through covalent interaction with the catalytic serine ("serine traps").
We further used compound P in an attempt to gain more insight into the mechanism of inhibition of the NS3/4A protease by its N-terminal cleavage products. The extent of this inhibition is quite unusual among serine proteases and is responsible for the nonlinearity of progress curves that has been reported in the literature (23,40,41). Binding of products to the active site of the enzyme was shown to require a dual "acid anchor": for optimum affinity, a P1 ␣-carboxylate as well as acidic functionalities in the P6/P5 positions are both required. Acidic residues are conserved around the P6 positions of all NS3/4A substrates and have been proposed to interact with a cluster of conserved basic residues in the protease. The P1 ␣-carboxylate was suggested to bind to the active site of the protease, forming hydrogen bonds with the backbone amides of Ser 138 and Gly 137 , with the ⑀-N of the catalytic histidine, and with the side chain of the conserved Lys 136 . This binding mode has been proposed in analogy to published structures of serine protease-product complexes and based on pH titration and site-directed mutagenesis experiments (23). The possibility of monitoring a spectroscopic signal arising from the formation of the enzyme-cofactor-product complex allowed us to character-ize the phenomenon of product inhibition of this enzyme in more detail, leading to the following conclusions. 1) The association of the NS3-Pep4AK protease complex with compound P was shown to be very fast with the magnitude of the second order rate constant (1.5 ϫ 10 8 M Ϫ1 s Ϫ1 ) being suggestive of a diffusion-limited process. This fast association is probably the result of both the high extent of solvent accessibility of the enzyme's active site and of electrostatic interactions between the protease and its product inhibitors. The important contribution of electrostatic interactions is highlighted by the sensitivity of K i values of hexapeptide acids to increasing ionic strength (24). As long range interactions, electrostatic forces might contribute to an acceleration of enzyme-ligand association processes.
2) The interaction of compound P with the protease was shown to be compatible with an ordered mechanism, requiring prior binding of the Pep4AK cofactor peptide. This conclusion was drawn from the lack of detectable fluorescence emission at 510 nm and the more than 200-fold loss in inhibitory potency of compound P in the absence of Pep4AK, indicating that no significant enzyme-probe complex is formed under these conditions. Furthermore, in the presence of nonsaturating concentrations of Pep4AK, the probe association kinetics became biphasic, with the slow phase being compatible with the attainment of a new equilibrium according to Fig. 5A. In fact, in the presence of a mixture of free enzyme and enzymecofactor complex, the selective binding of P to the latter species will shift the equilibrium toward the formation of the ternary enzyme-probe-activator complex. In principle, quantitative ternary complex formation could be attained at any Pep4AK concentration, provided that a high enough concentration of compound P is added. This is in accord with the fact that the overall amplitudes of the association reaction did not depend on the concentration of Pep4AK (Fig. 3B). The structural basis for this mechanism of interaction may be deduced from what is known about the effects of NS4A on the structure of the NS3 protease domain (16 -18). X-ray crystal data and recent NMR solution data (42) provide evidence for conformational changes affecting mainly the N-terminal domain of NS3 that occur in the absence of the cofactor. The NMR solution structure of the uncomplexed protease clearly shows a considerable flexibility of this portion of the molecule that also affects the catalytic histidine and aspartic acid residues (42). The former residue, on the other hand, was proposed to be involved in hydrogen bond formation with the P1 ␣-carboxylic acid function of product inhibitors (23). This interaction would be detrimentally affected by a lack of conformational stability in this region, thereby explaining the drop in potency experienced by hexapeptide acids when assayed in the absence of Pep4AK. We found that the extent of this drop differed in a sequence-specific way. The affinity of the enzyme for compound P was affected by at least 2 orders of magnitude, resulting in an absolute requirement of prior complexation of the enzyme with Pep4AK in order to allow detectable binding of the compound to the active site of the NS3-Pep4AK protease complex. In contrast, binding of DEMEEC-OH was significantly less affected (10-fold) by the absence of the cofactor, whereas the optimized hexapeptide DEDi-fEChaC-OH showed again a very large (Ͼ200-fold) drop in affinity on the uncomplexed protease. An explanation to these findings possibly resides in the different driving forces that are operative in the formation of the single enzyme-inhibitor complexes. In the context of optimized sequences (like DEDi-fEChaC-OH), the relative contribution of the P1 ␣-carboxylate in enzyme inhibitor complex formation may be energetically more incisive than in the context of wild type sequences like DEMEEC-OH, with the optimized sequence contributing to a more "productive" positioning of the carboxylate group. In the FIG. 8. Pre-steady-state association kinetics of DEDifECha-AbuF2CHO determined by probe displacement. 100 nM NS3 protease in 50 mM Hepes, pH 7.5, 1 mM DTT, 15% glycerol, 1% CHAPS, 40 M Pep4AK, and 10 M compound P were reacted with a 20 M concentration of the aldehyde inhibitor DEDifEChaAbuF2CHO. The reaction was followed by monitoring the time-dependent decrease in dansyl fluorescence Ͼ400 nm upon excitation of tryptophan fluorescence at 280 nm on a stopped-flow instrument. The result of a single run is shown. The line indicates a nonlinear least squares fit of the data to the equation F ϭ F 0 ϩ Ae Ϫkobst , where A is the amplitude of the fluorescence change, and k obs is the observed pseudo-first order rate constant for the approach of equilibrium. The inset shows a plot of k obs versus inhibitor concentration. From this plot, the second order rate constant for the formation of the enzyme-inhibitor complex was calculated as described under "Results." context of wild-type sequences, interactions with the C-terminal portion of the protease, the conformation of which is less affected by the presence of the cofactor, could be energetically more important. Selective weakening of the interaction of the enzyme with the P1 ␣-carboxylate in the absence of Pep4AK would therefore result in a higher loss in affinity of DEDi-fEChaC-OH than of DEMEEC-OH. We therefore conclude that ordered or random mechanisms of ternary enzyme-cofactorinhibitor complex formation do not result from stringent mechanistic requirements but simply reflect the magnitude of the relative loss in potency of a given ligand on the free enzyme. In agreement with this notion, Landro et al. (22) have shown that a hexapeptide aldehyde based on the sequence of the NS5AB cleavage site binds randomly to the free enzyme or to the enzyme-cofactor complex.
Irrespective of the apparent mechanism, product binding will result in a strengthening of the enzyme-cofactor interaction, since in either case the equilibrium will be shifted toward ternary complex formation. Such a strengthening was previously shown to occur also in the enzyme-substrate-cofactor complex (37). It remains to be established whether the enhancement of the enzyme-cofactor interaction through active site occupancy by substrates or cleavage products has any physiological significance.
The possibility of generating direct binding data for active site ligands of the NS3/4A protease, including pre-steady-state association rates, adds an important tool to the instruments at our disposal for the search of inhibitors of this primary drug development target.