A Fluorescence Resonance Energy Transfer Method for Measuring the Binding of Inhibitors to Stromelysin

Analytical Biochemistry 275, 141–147 (1999) Article ID abio.1999.4306, available online at http://www.idealibrary.com on

A Fluorescence Resonance Energy Transfer Method for Measuring the Binding of Inhibitors to Stromelysin Dennis E. Epps, 1 Mark A. Mitchell, Gary L. Petzold, John H. VanDrie, and Roger A. Poorman Pharmacia and Upjohn Company, 7000 Portage Road, Kalamazoo, Michigan 49001

Received February 12, 1999

A sensitive fluorescence resonance energy transfer method was developed for the direct measurement of the dissociation constants of stromelysin inhibitors. The method is applied to the thiadiazole class of stromelysin inhibitors and it takes advantage of the fact that, upon binding to the active site of enzyme, the thiadiazole ring, with its absorbance centered at 320 nm, is able to quench the fluorescence of the tryptophan residues surrounding the catalytic site. The changes in fluorescence are proportional to the occupancy of the active site: Analysis of the fluorescence versus inhibitor concentration data yields dissociation constants that are in agreement with the corresponding competitive inhibitory constants measured by a catalytic rate assay. The affinity of nonthiadiazole inhibitors of stromelysin—such as hydroxamic acids and others— can be determined from the concentration-dependent displacement of a thiadiazole of known affinity. Using this displacement method, we determined the affinities of a number of structurally diverse inhibitors toward stromelysin. Since the three tryptophan residues located in the vicinity of the active site of stromelysin are conserved in gelatinase and collagenase, the method should also be applicable to inhibitors of other matrix metalloproteinases. © 1999 Academic Press

Key Words: stromelysin; fluorescence; inhibitor binding; ligand displacement; energy transfer; matrix metalloproteinase; fluorescence resonance energy transfer.

Inhibitors of stromelysin (MMP-3) and of other matrix metalloproteinases (MMPs) 2 are potential thera-

peutic agents against cancer, arthritis, restenosis, and other diseases that are caused by or result in the degradation of connective tissue matrices (1). The mainstay of the intensive search for novel and specific highaffinity inhibitors of MMPs is an array of rate assays, using a variety of substrates, in which a loss of enzymatic activity is indicative of competitive inhibition. Once a promising compound is identified, detailed kinetic analysis of the enzyme, substrate, and inhibitor dependency of the rate is required for ascertaining the type of inhibition and the efficacy of the inhibitor. Even with such a battery of kinetic experiments it is not always easy to discern whether the inhibition is due to a direct noncovalent interaction with the enzyme or to some other, less desirable mode of action, such as interaction of the inhibitor with the substrate instead of the enzyme, or to covalent modification of the enzyme. For this reason we felt that a facile method of measuring directly the binding of inhibitors to MMPs would be a worthwhile addition to the existing tools of search for stromelysin inhibitors. The structure of the catalytic fragment of stromelysin determined by NMR (2) shows that all three tryptophan residues of the enzyme are located within 15 Å of the catalytic Zn site. For enzymes that have one or more tryptophans located near their active site, the binding of inhibitors of appropriate spectral properties is accompanied by a decrease in fluorescence emission of the tryptophan(s) by fluorescence resonance energy transfer (FRET). For example, the acidic protease renin has three tryptophans near its active site (3) and the binding of a dansylated specific inhibitory peptide does decrease the enzyme’s tryptophan emission (4). In this report, we describe the results of our efforts to develop and validate a FRET assay for measuring the

1

To whom correspondence should be addressed. Abbreviations used: MMP, matrix metalloproteinase; FRET, fluorescence resonance energy transfer; SAR, structure activity relationships; CHAPS, 3-[(3-cholamidopropyl)-dimethylammonio]-1-pro2

0003-2697/99 $30.00 Copyright © 1999 by Academic Press All rights of reproduction in any form reserved.

panesulfonate; D-PBS, calcium- and magnesium-free Dulbecco’s phosphate-buffered saline. 141

142

EPPS ET AL.

FIG. 1. Chemical structures of inhibitors.

affinity of stromelysin inhibitors, using derivatives of a zinc-chelating thiadiazole. MATERIALS AND METHODS

Reagents. Recombinant truncated stromelysin was purified as described previously (5). The purified enzyme was activated by incubation at 55°C for 1.5 h, and the propeptide was removed by dialysis against 20 mM Tris, 10 mM CaCl 2, 10 mM ZnCl 2, pH 7.60. Stromelysin inhibitors to be tested were reconstituted as concentrated DMSO stock solutions. The dansylated derivatives of PNU-107859, PNU-140171, and the other inhibitors were synthesized by our medicinal chemists. Representative structures of the inhibitors tested are shown in Fig. 1. Absorption was measured using a Cary 2200 spectrophotometer and fluorescence measurements were made with an ISS K2 spectrofluorometer. The spectral data were processed using the software supplied by the instrument manufacturers. Particle concentration fluorescence assay. Avidincoated polystyrene particles and Fluoricon-CA assay plates were purchased from IDEXX (Westbrook, ME). Polystyrene U-bottom 96-well plates were from Corning-Costar (Cambridge, MA). Calcium- and magne-

sium-free Dulbecco’s phosphate-buffered saline (DPBS) was from GIBCO-BRL (Grand Island, NY). 3-[(3Cholamidopropyl)-dimethylammonio]-1-propanesulfonate (CHAPS) was purchased from Sigma Chemical Co. (St. Louis, MO). The stromelysin substrate, biotinPro-Leu-Ala-Leu-Trp-Ala-Arg-Lys(Fluorescein)-OH, was synthesized using solid-phase peptide synthesis techniques and is similar to those used by others (9 – 11). Stock solutions of stromelysin inhibitors in DMSO were serially diluted into a buffer containing 150 mM NaCl, 50 mM Tris-HCl, pH 7.50, 10 mM CaCl 2, 5 mM CHAPS, and 0.02% (w/v) NaN 3. Stromelysin and its substrate were made in the same buffer. Each well of a polystyrene U-bottom 96-well plate contained the inhibitor, 60 nM stromelysin, and 100 nM substrate in a total volume of 100 ml. The final DMSO concentration was adjusted, when necessary, to 2% (v/v). The inhibitor and stromelysin were coincubated for 15 min at 23°C prior to substrate addition. To obtain 90 –95% substrate hydrolysis the plate was incubated in the dark for 90 min at 23°C. The reaction was terminated by transferring 50-ml aliquots from the 96-well plate to a Fluoricon-CA assay plate containing an equal volume of 0.1% avidin-coated polystyrene particles suspended in wash buffer (D-PBS; 50 mM EDTA, 50 mM EGTA).

STROMELYSIN BINDING BY FLUORESCENCE ENERGY TRANSFER

Following a 15-min incubation in the dark at 23°C, the plate was loaded into an IDEXX Screen Machine 2000. The cleaved fluorescent fragment was removed by vacuum filtration and washed with buffer. Sample epifluorescence was measured using an Ex 485 nm/Em 535 nm filter. Continuous fluorescence assay. The substrate (Dnp)-Pro-Leu-Ala-Leu-Trp-Ala-Arg-NH 2, U-94761E, was synthesized on a 0.5-mmol scale using an Applied Biosystems Inc. (ABI) 430A instrument with doublecouple cycles and standard t-Boc chemistry. The t-BocL-amino acids and benzhydrylamine resin were purchased from ABI. The 2,4-dinitrophenyl group was added to the N-terminus of the peptide/resin by reaction for 3 h with 5.0 mmol of 2,4-dinitrofluorobenzene and 0.75 mmol of N-ethylmorpholine in dry DMF. The peptide was removed from the supporting resin, with simultaneous removal of the side-chain protecting groups, by treatment with HF containing 20% anisole: butanedithiol (2:1). The peptide was purified by preparative reversed-phase chromatography using a Vydac C-18 column: The eluent consisted of the initial eluent of water containing 0.1% trifluoroacetic acid, followed by a 2-h linear gradient to water/acetonitrile (%B 5 30) mixture (0.1% trifluoroacetic acid containing sufficient acetonitrile to effect elution and purification of the product). The purified peptide was lyophilized from water/acetic acid to yield a yellow powder. The peptide gave a single symmetrical peak on analytical HPLC and gave, by FAB/MS, the expected molecular weight. Peptide stock solutions were prepared by dissolving about 1 mg of peptide in 1 ml DMSO. Water was distilled in a Corning Mega-Pure MP-3A system. The assay buffer, containing 20 mM Tris-HCl, pH 7.50, 5 mM CaCl 2, and 10 mM ZnCl 2, was prepared daily, degassed, and stored under argon as a precautionary measure to prevent photooxidation of the peptides. The assay buffer, 870 ml, was mixed with an amount of the peptide stock solution containing about 1 mg of substrate and DMSO to yield a final concentration of 3.4% DMSO. Five hundred microliters of this mixture was transferred to a dual-path-length cell, 10 3 2 mm, with a capacity of 0.8 ml, and placed in a Perkin Elmer Luminescence-Spectrometer LS 50. The excitation was 293 nm, emission 350 nm, slit width 2.5 mm, response 0.5 s, and temperature 25°C. After the baseline became stable, activated stromelysin was added to a final concentration of about 120 nM and the mixture was briefly stirred with a glass rod. The fluorescence was recorded until completion of the reaction. The data were digitized and analyzed in terms of a first-order reaction according to f 5 f 0 1 Df 3 ~1 2 e 2k expt !,

[1]

143

where f is the time-dependent fluorescence, f 0 is the background fluorescence, Df is the fluorescence change associated with the full hydrolysis of the substrate, t is time, and k exp is the experimental pseudo-first-order rate constant. About three hundred f, t data pairs were collected per experiment and analyzed using Eq. [1] and a nonlinear least-squares program. All our results were fully consistent with Eq. [1], thereby confirming that the substrate concentration used was well below K m. Binding assay. The binding of PNU-140171 to stromelysin was measured as follows. Activated stromelysin, 0.15 mM, was added to 10 mM Tris-HCl, 10 mM ZnCl 2, 20 mM CaCl 2, pH 7.60, with or without 5 mM CHAPS, in a final volume of 2 ml. The temperature of the buffer was maintained at 23.4°C by means of a Lauda thermostated circulating bath. In the FRET assay, small aliquots of thiadiazole inhibitors in DMSO were added sequentially to the cuvette, and the ratioed fluorescence emission at 325 nm was recorded after each addition with excitation at 293 nm to excite only tryptophan residues. The total amount of DMSO added never exceeded 0.75% and had no effect on the fluorescence readings. For the displacement assay, a fixed amount of thiadiazole inhibitor of known K t 3 was added to a given amount of stromelysin in the assay mixture described above. The fluorescence of stromelysin in the presence of the thiadiazole was recorded, and then aliquots of nonthiadiazole inhibitors were added, and the new fluorescence was recorded. In both assays, fluorescence readings and inhibitor concentrations were corrected for the small dilution effect. Data analysis. In order to obtain the best signal to noise with the fluorometer, the enzyme concentrations used in measuring the binding of high-affinity FRET inhibitors could not be made much lower than the dissociation constant of the inhibitor, K t. In those cases, the dependency of the fluorescence emission on the inhibitor concentration was analyzed using

f 5 aE0 1

b2a ~T 0 1 E 0 1 K t 2 2 Î~T 0 1 E 0 1 K t! 2 2 4T 0 E 0 ,

[2]

where f is the fluorescence intensity, T 0 is the analytical concentration of the thiadiazole inhibitor, E 0 is the 3

The symbol K i designates the dissociation constant of the stromelysin–inhibitor complex as measured by a catalytic rate assay, the symbol K t designates the dissociation constant of the stromelysin– thiadiazole inhibitor complex as measured by FRET, and the symbol K u designates the dissociation constant of the stromelysin–nonthiadiazole inhibitor complex as measured by the displacement of a FRET inhibitor.

144

EPPS ET AL.

analytical stromelysin concentration, and a and b are proportionality constants (7). For lower affinity FRET inhibitors, where the enzyme concentration could be made much lower than K t, data were analyzed by nonlinear least squares using the simple Langmuir model (7): f 5 aE0 1 ~b 2 a!

T 0E 0 . Kt 1 T0

[3]

f 5 E0

Kt

Ku

and

E1U | 9 = X,

[4]

where E represents the concentration of free stromelysin and T that of the thiadiazole inhibitor, U is the concentration of the nonthiadiazole inhibitor, and C and X are the concentrations of complexes. For experiments where E 0 ! T 0 and E 0 ! U 0 , and hence T 0 ' T, the dissociation constants for the two inhibitors are defined as Kt 5

T0 z E ; C

Ku 5

U0 z E . X

[5]

Solving for C and X yields C5E

T0 Kt

and

X5E

U0 . Ku

[6]

Since fluorescence intensities are additive, the fluorescence, f, of the reaction mixture is given by f 5 a ~E 1 X! 1 b C,

[7]

where a and b are molar emissivities. In our system a . b, since tryptophans are unquenched in E and X and quenched in C. Combining Eqs. [6] and [7] yields

FS

f5E a 11

D

G

T0 U0 1b . Ku Kt

[8]

By combining Eqs. [6] and [7] with the stoichiometric equation E 0 5 E 1 C 1 X one obtains

D

U0 bT0 1 Ku Kt . T0 U0 11 1 Kt Ku

[9]

The fluorescence in the presence of the thiadiazole inhibitor and in the absence of a nonthiadiazole inhibitor is defined as

bT0 Kt . T0 11 Kt

a1

The K u of simple competitive inhibitors not containing an intrinsic quenching group can be determined by measuring the fluorescence increase occurring by displacement of a thiadiazole inhibitor of known K t. In this case, the system is represented by the equilibria,

E1T | 9 = C

S

a 11

f0 5 E0

[10]

Similarly, one may define the fluorescence of the solution with the thiadiazole inhibitor completely displaced as f ` 5 a E 0 . Substituting this expression into Eq. [10] yields

bE0

S

D

T0 T0 5 f0 1 1 2 f `. Kt Kt

[11]

Substituting the above expression into Eq. [9] yields f 5 f0 1

~f ` 2 f 0 !U 0 . T0 I0 1 Ku 1 1 Kt

S

D

[12]

The experimental f versus U 0 data may be analyzed in terms of Eq. [12] using a nonlinear least-squares method where f 0 , f ` , and K u are the parameters to be optimized. It is advantageous first to estimate the value of K u and then choose T 0 ' 1 mM 3 K u/K t. RESULTS AND DISCUSSION

As a first approach in our search for a FRET inhibitor of stromelysin, we selected as templates the zincchelating thiadiazole-containing inhibitors that had already been shown by rate assays to be competitive inhibitors of stromelysin. These inhibitors appeared to be specific toward stromelysin since they showed much lower affinities toward collagenase and gelatinase. Docking models indicated that several of these inhibitors and in particular PNU-107859, with its K i of about 1 mM, may be able to bring a covalently attached dansyl group into the neighborhood of the active site without losing much of their affinity toward the enzyme. The dansyl group was chosen by virtue of its spectral properties, which are suitable for quenching the tryptophan fluorescence via FRET. When the UV absorption spectra of PNU-107859 and of its dansylated derivative, PNU-140171, were examined, it be-

STROMELYSIN BINDING BY FLUORESCENCE ENERGY TRANSFER

FIG. 2. Absorption spectrum of U-107859 and fluorescence emission spectrum of truncated stromelysin. The absorption spectrum of U-107859 (E) was obtained in the assay buffer using a Cary 2200 Spectrophotometer. The fluorescence emission spectrum (F) was recorded as described under Materials and Methods.

came apparent that not only the spectrum of the dansyl derivative but also the spectrum of the parent compound significantly overlapped the emission spectrum of the tryptophans of the enzyme. In fact, the absorption spectrum of PNU-107859 (l max 5 320 nm, e 5 18,000 M z cm 21), shown in Fig. 2, overlapped more efficiently the Trp emission of stromelysin (l em 5 317 nm) than the absorption spectrum of the dansyl derivative itself (l max 5 350 nm). As a consequence, the titration of stromelysin with the dansylated inhibitor, PNU-140171, resulted in a significant progressive quenching of the Trp fluorescence with minimal reemission of the absorbed light as dansyl fluorescence at 425 nm (data not shown), but a more robust signal with greater dynamic range was obtained when the emission of stromelysin was quenched upon binding of PNU-107859 to the enzyme (Fig. 3). Thus, the thiadiazole group itself is an efficient FRET acceptor and, consequently, the affinities of all the thiadiazole class of stromelysin inhibitors could be determined directly by measuring the fluorescence decrease of the enzyme as a function of the inhibitor concentration. Progressive increase in the concentration of PNU140171 is accompanied by a concentration-dependent decrease in the intrinsic tryptophan fluorescence of the protein (Fig. 4). The competitive nature of this interaction was confirmed by the fact that addition of an excess of Batimistat—a nonthiadiazole-type high-affinity competitive inhibitor of stromelysin (6)— completely restored the fluorescence of the enzyme. Since in the case of PNU-140171 the enzyme concentrations needed to produce a good emission signal are comparable to the inhibitor K t the concentration dependency

145

FIG. 3. Effect of the addition of PNU-107859 on the fluorescence emission spectrum of stromelysin. Stromelysin, 0.15 mM, was titrated with increasing concentrations of PNU-107859 (0 to 5 mM in 0.5 mM increments) as described under Materials and Methods. All spectra were corrected for the dilution effect.

of the quenching was analyzed in terms of Eq. [2], using a nonlinear least-squares program. As shown in Fig. 4, the fluorescence changes associated with the addition of PNU-140171 to a solution containing an analytical stromelysin concentration of 0.15 mM are fully consistent with a mechanism of competitive inhibition, as evidenced by the agreement between the experimental data and the theoretical curve calculated from Eq. [2] with the best-fit parameters K t 5 0.10 6 0.04 mM and E 0 5 0.17 6 0.05 mM. The agreement of both these parameters with the expected values further confirmed the validity of the experimental

FIG. 4. Titration of stromelysin with U-140171. The binding of U-140171 to stromelysin was monitored by quenching of the intrinsic tryptophan fluorescence of the protein as described under Materials and Methods. The solid line represents the theoretical fit to the data points using Eq. [2] in the text.

146

EPPS ET AL. TABLE 1

Comparison of the Dissociation Constants of Stromelysin– Thiadiazole Complexes as Measured by Fluorescence Quenching (K t) and by Kinetic Assay (K i)

FIG. 5. Titration of stromelysin with U-109648. The binding of U-109648 to 0.15 mM stromelysin was monitored by quenching of the intrinsic tryptophan fluorescence of the protein as described under Materials and Methods. The solid line represents the theoretical fit to the data points using Eq. [3].

method. In a separate experiment, titration of 0.30 mM stromelysin with PNU-140171 yielded K t 5 0.11 6 0.05 mM and E 0 5 0.35 6 0.06 mM. Thus, both the K t and the stoichiometry of the reaction can be determined with a high degree of accuracy by this method. In the case of this inhibitor, and all others tested, the K t values did not change significantly upon addition of CHAPS to the reaction mixture, indicating that adsorption of the enzyme to the reaction vessel is negligible. The concentration dependency of the quenching curves of thiadiazole inhibitors with K t @ E 0 was analyzed by nonlinear least-squares analysis using Eq. [3]. As an example of this type of experiment, Fig. 5 shows the quenching of the tryptophan fluorescence of 0.15 mM stromelysin by increasing concentrations of PNU-109648. The analysis of the data yielded K t 5 2.00 6 0.08 mM and the data were fully consistent with Eq. [3] as shown by the agreement between the data points and the theoretical curve calculated using the best-fit parameters. The FRET method was ultimately validated by comparing the K t’s obtained by the FRET method and K i’s obtained from the inhibition of a catalytic rate assay. As reported in Table 1, the K t’s agree, in general, reasonably well with the K i values. Thus, because of the agreement of these values, the rate assay and the FRET method measure the binding of the inhibitor to the active site. All compounds listed in Table 1 produced around 50% quenching of the stromelysin fluorescence, thereby indicating that they all occupy approximately the same location at the active site.

PNU number

K i (mM)

K t (mM)

107859 109303 109304 109396 109648 109651 140171 140008 140146

0.38 6 0.05 0.52 6 0.03 0.30 6 0.02 0.27 6 0.03 1.10 6 0.10 ND ND 6.10 6 0.5 ND

0.50 6 0.10 0.25 6 0.06 0.32 6 0.02 0.19 6 0.04 2.00 6 0.08 0.20 6 0.03 0.10 6 0.04 4.3 6 0.3 30.0 6 0.8

Note. K i determinations were made using either the Fluoricon fluorescent plate assay or the continuous recording methods using the DNP substrate as described under Materials and Methods.

For competitive inhibitors not containing an intrinsic quenching group, determination of their K u’s can be accomplished by measuring the fluorescence increase occurring by displacement of a thiadiazole inhibitor of known K t. When increasing amounts of the hydroxamic inhibitor CAS108383-58-0 (8) were added to 0.15 mM stromelysin and 7.14 mM PNU-107859 (K t 5 0.5 mM), there was a gradual relief of the quenching (Fig. 6), indicating competitive displacement of the FRET inhibitor. The results were analyzed in terms of Eq. [12] using a nonlinear least-squares fitting program. The data were fully consistent with this model as evi-

FIG. 6. Displacement of U-107859 from stromelysin by CAS108383-58-0. The binding of U-99533 to stromelysin was monitored by displacement of 7.28 mM U-107859 from the protein (0.15 mM) which resulted in a return of the native tryptophan fluorescence. The data were analyzed as described under Materials and Methods using Eq. [12], and the solid line represents the theoretical fit to the data.

STROMELYSIN BINDING BY FLUORESCENCE ENERGY TRANSFER TABLE 2

Comparison of the Dissociation Constants of Stromelysin– Hydroxamic Acid Inhibitor Complexes as Measured by Fluorescence Quenching (K u) and by Kinetic Assay (K i) Compound

K u (mM)

K i (mM)

CAS108383-58-0 Ro31-9790

0.10 6 0.02 0.21 6 0.02

0.078 6 0.002 0.13 6 0.02

denced by the agreement between the experimental data and the theoretical curve. The analysis yielded K u 5 0.10 6 0.02 mM, comparable to K i 5 0.078 6 0.002 mM. The K u’s of hydroxamic acid inhibitors measured by the FRET method are compared to the corresponding K i values in Table 2. Again, the agreement is reasonable. In summary, we have developed a titration/displacement assay based on fluorescence resonance energy transfer to thiadiazole inhibitors from the fortuitously well-situated tryptophan residues of stromelysin. Inhibitors lacking the thiadiazole group can be analyzed by displacement of a thiadiazole inhibitor with known K t. The K t’s obtained by the use of either of these two assays were in excellent agreement with the K i values for the same compounds determined by inhibition of enzyme catalysis. The major limitation of the assay in its present form is for thiadiazole inhibitors that would contain a chemical group emitting in the same spectral region as tryptophan. Also, the calculation of K t’s of inhibitors containing strong UV absorbing fluorophores may necessitate a correction for inner filter effects (7). Such problems could be ultimately circumvented by conjugating a long-wavelength probe to a competitive inhibitor of known K i which would also allow for high-throughput screening of inhibitors in a 96-well plate reader format. The method should be

147

readily adaptable to any protein/enzyme–ligand system where tryptophans reside in or near to the binding site. ACKNOWLEDGMENTS The technical assistance of Martha Warpehoski in the assays is gratefully acknowledged as well as the synthesis of several of the inhibitors by John Jacobsen and Don Harper.

REFERENCES 1. Stetler-Stevenson, W. G., Aznavoorian, S., and Liotta, L. A. (1993) Annu. Rev. Cell Biol. 9, 541–573. 2. Stockman, B. J., Waldon, D. J., Yuan, P., Scahill, T. A., Belonga, K. L., Jacobsen, E. J., Treiberg, J. A., Schostarez, H. J., Mitchell, M. A., Liggett, WS-F., Marshall, V. P., Petzold, G. L., and Poorman, R. A. (1996) Abstract, TO XVIITH ICMBRS Conference, Keystone, CO. 3. Epps, D. E., Poorman, R. A., Hsi, J., and Heinrikson, R. L. (1987) J. Biol. Chem. 262, 10570 –10573. 4. Epps, D. E., Schostarez, H., Argoudelis, C. V., Poorman, R. A., Hinzmann, J., Sawyer, T. K., and Mandel, F. (1989) Anal. Biochem. 181, 172–181. 5. Becker, J. W., Marcy, A. I., Rokosz, L. L., Axel, M. G., Burbaum, J. J., Fitzgerald, P. M. D., Cameron, E., Hagmann, W. K., Hermes, J. D., and Springer, J. P. (1995) Protein Sci. 4, 1966 – 1976. 6. Grams, F., Crimmin, M., Hinnes, L., Huxley, P., Pieper, M., Tschesche, H., and Bode, W. (1995) Biochemistry 34, 14012– 14020. 7. Epps, D. E., Raub, T. J., and Ke´zdy, F. J. (1995) Anal. Biochem. 227, 342–350. 8. Beckett, R. P., Davidson, A. H., Drummond, A. H., Huxley, P., and Whittaker, M. (1996) Drugs Under Dev. 1, 16 –26. 9. Knight, C. G., Willenbrock, F., and Murphy, G. (1992) FEBS Lett. 296, 263–266. 10. Netzel-Arnett, S., Mallya, S. K., Nagase, H., Birkendal-Hansen, H., and Van Wart, H. E. (1991) Anal. Biochem. 195, 86 –92. 11. Stack, M. S., and Gray, R. D. (1989) J. Biol. Chem. 264, 4277– 4281.