Jul 25, 2006 - M" s-'), the dissociation rate is also rapid (kfg. = 15. s-I). Furthermore, there is no isomerization of the ter- nary complex of trimethoprim with ...
THEJOURNAL OF BIOLOGICAL CHEMISTRY 0 1988 by The American Society for Biochemistry and Molecular Biology, Inc.
Vol. 263,No. 21,Iasue of July 25, pp. 10304-10313,1988 Printed in U.S.A.
Kinetics of the Formation and Isomerizationof Methotrexate Complexes of Recombinant Human Dihydrofolate Reductase* (Received for publication, October 14, 1987)
James R. Appleman$, Neal PrendergastP, Tavner J. Delcampi, James H. Freisheimi, and Raymond L. BlakleySq From the $Department of Biochemical and Clinical Pharmacology,St. Jude Children's Research Hospital, Memphis, Tennessee 38101, the TDepartment of Pharmacology, University of Tennessee College of Medicine-Memphis, Memphis, Tennessee 38163, and the $Department of Biochemistry, MedicalCollege of Ohio, Toledo, Ohio 43699
Thekinetics of inhibitorbindingtohighlypuriused in cancer chemotherapy. The interaction of MTX with fied recombinant human dihydrofolate reductase itstarget, dihydrofolate reductase (DHFR),has been the (rHDHFR) have been examined. Methotrexate (MTX) subject of numerous and intensive investigations, but most of binds rapidly (CZ,, = 1.0 X 10' M" s-') and tightly(kff/ these have been with bacterial DHFR. The structure of the k,,,= 210 PM) to the preformed complex of rHDHFR MTX complexes with enzyme from Escherichia coli and Lacwith NADPH. The initialassociation reaction between tobacillus casei has been determined to 1.7-A resolution (Bolin rHDHFR. NADPH and MTX is followed by an isomer- et al., 1982). Binding kinetics of MTX to form binary or ization of the resultingcomplex (ki, = 0.4 s-') leading ternary complexes of enzyme from E . coli, L. casei, Streptoto a new conformerin which MTX is bound even more coccus faecium, and chicken liver have been studied by tightly ( K i = 3.4 PM). Similar results have been obtained with a major metabolite of MTX having four stopped-flow techniques (Dunn and King, 1980; Cayleyet al., additional glutamate residues for which K;= 1.4 PM. 1981; Blakley and Cocco, 1985a,198513; Appleman et al., 1988) 7-HydroxyMTX, another major metaboliteof MTX, is and by progress of inhibition methods (Williams et al., 1979; a weak inhibitor of rHDHFR (Ki = 8.9 nM), and a Williams et al., 1980; Stone et al., 1984; Blakley and Cocco, polyglutamate form of this metabolite is an equally 1985a; Stone and Morrison, 1986). These studies indicated weak inhibitor (Ki = 9.9 nM), so that the addition of that thereis synergism in binding between NADPH and MTX glutamate residues toMTX or 7-hydroxyMTX has lit- and that the complexes initially formed undergo some kind tle effect on their binding. It follows that the signifi- of conformational change, the effect of which is to increase cance of MTXpolyglutamate formationrelates to other the tightness of binding, in some cases by a substantialfactor. roles such as increasing the cytotoxicity of MTX by Nuclear magnetic resonance studies (Cocco et al., 1981a, prolonging intracellular retentionof the drug.Another 1981b, 1983), ultraviolet difference spectroscopy (Erickson antifolate, trimethoprim, binds tightly to dihydrofolate and Mathews, 1972; Poe et al., 1974; Gupta et al., 1977; Hood reductasesfrombacterialsources,butweaklyto and Roberts, 1978; Subramanian and Kaufman, 1978; Cocco rHDHFR in the ternary complex ( K D= 0.5 MM). Al- et al., 1981a, Stone and Morrison, 1983), resonance Raman though the association step is rapid (ken = 0.4 x 10' spectroscopy (Saperstein et al., 1978; Ozaki et al., 1981), M" s-'), the dissociation rate is also rapid (kfg= 15 potentiometric titration (Subramanian and Kaufman, 1978), s-I). Furthermore, thereis no isomerization of the ter- and crystallographic studies(Matthews et al., 1977,1978; nary complex of trimethoprim with rHDHFR, in con- Bolin et al., 1982) all indicate that MTX bound in the active trast to the known isomerization of complexes of tri- site of DHFR is protonated at N1 with ionic and hydrogen methoprim with bacterial dihydrofolatereductases. bond interaction between N1 and the carboxylate group of the active site aspartic acid or glutamic acid. Once again, the majority of these studies has been done with bacterial DHFR, Methotrexate (MTX)' is an antifolate that has long been and few have been done with mammalian enzyme. So far as clinical relevance is concerned, however, the * This work was supported inpart by United States Public Health interaction of MTX with human DHFR is of primary significance. A major problem in studying the human enzyme is the Service Research Grants R01 CA31922 (to R. L. B.) and RO1CA 41461 (to J. H. F.) and Cancer Center Core Grant P30 CA 21765 (to low level of DHFR in human tissues (Kamen et al., 1985). R. L. B.) from the National Cancer Institute, National Institutes of Even when human cell lines are obtainedwith elevated levels Health, by Biomedical Research Support Grant SO7 RR 05584 (to J. of DHFR by selection in vitro with MTX (Domin et al., 1982; R.A.) from the National Institutes of Health, and by American Lebanese Syrian Associated Charities ( t o R. L. B. and J. R. A.). The Delcamp et al., 1983) the amount of highly purified enzyme costs of publication of this article were defrayed in part by the that can be obtained from them is relatively small. This payment of page charges. This articlemust therefore be hereby problem has been alleviated by the expression of the recommarked "advertisement" in accordance with 18 U.S.C. Section 1734 binant human DHFR gene in a suitable vector (Prendergast solely to indicate this fact. et al., 1987, 1988). We present here the results of a study of
The abbreviations used are: MTX, methotrexate, 4-amino-4deoxy-10-methylfolic acid; MTX+Glu,, 4-amino-4-deoxy-10-meth10-methylpteroyl - L - glutamyl - y - L - glutamyl - y - L - glutamyl ylpteroyl-~-glutamyl-y-~-glutamyl-y-~-glutamyl-y-~-glutamyl-y-~glutamate; MTX+Glu3, 4-amino-4-deoxy-l0-methylpteroyl-~-gluy-L-glutamate; DHFR, dihydrofolate reductase (EC 1.5.1.3); HDHFR, tamyl-y-L-glutamyl-y-L-glutamyl-y-L-glutamate; 7-hydroxyMTX, 7- human dihydrofolate reductase; rHDHFR, recombinant human hydroxymethotrexate, 4-amino-4-deoxy-7-hydroxy-lO-methylfolicDHFR expressed in anE. coli host; Trmp, trimethoprim (2,4-diaminoacid 7-hydroxyMTX+Glur, 4-amino-4-deoxy-7-hydroxy-lO-methyl-5-(3,4,5-trimethoxybenzyl)pyrimidine);DAMP, 2,4-diamino-5-(1adamantyl)-6-methylpyrimidine;MES, 4-morpholineethanesulfonic pteroyl-~-glutamyl-y-~-glutamyl-y-~-glutamyl-y-~-glutamyl-y-~glutamate; 'I-hydroxyMTX+Glus, 4-amino-4-deoxy-7-hydroxy- acid Hpfolate, dihydrofolate; Rfolate, tetrahydrofolate.
10304
Methotrexate Complexes of Human Dihydrofolate
Reductase
10305
with an Interactive Structures A113 analog input system on an Apple IIe microcomputer. Optical transmittance andfluorescence data were stored for subsequent analysis. Intervals were timed using The Clock from Mountain Computers, Inc. Data were collected a t rates of up to 5000 points/s. Typically, 4096 data points were obtained per kinetic trace. Measurement of enzymic activity on the stopped-flow spectrophotometers was primarily of transient-state conditions. Such measurements included development of inhibition by MTX and MTX+Glu4, and determination of K,,, for Hzfolate through both progress curve and initial velocity analysis. Additional experimental details may be found in the text andfigure legends. Analysis of Progress of Inhibition Curves a t High Concentrations of rHDHFR and MTX or MTX+Glu,-As described under “Results,” inhibition of enzymic activity occurs in four distinct phases when a buffered solution containing rHDHFR and NADPH is mixed with a solution containing Hzfolate and MTXin a stopped-flow experiment and the formation of product is followed through changes in absorMATERIALS ANDMETHODS bance at 340 nm. The onset of inhibition during the second and third In general, the experimental procedures and materials were as phases was analyzed as described previously for DHFR from E. coli previously used (Appleman et al., 1988). MTX+Glu, (4-amino-4- (Appleman et al., 1988), according to the following kinetic scheme, deoxy-l0-methylpteroyl-~-glutamyl-y-~-glutamyl-y-~-glutamyl-y-~A glutamyl-y-L-glutamate) was purchased from American Radiolabeled E-EA+E+P Chemicals, St. Louis, MO. HPLC indicated that this material contained one contaminant (5-10%) with elution time corresponding konI ‘Oiirn approximately to MTX+Glu3. 7-HydroxyMTX and 7-hydroxyEI EI’ kr.iao MTX+Glu, were prepared enzymically by the method of McGuire et al. (1984). The rabbit liver aldehyde oxidase used was prepared slow according to the method of Rajagopalan et al. (1962) up to the third purification step. HPLC of the final products indicated absence of where E represents rHDHFR, A represents Hzfolate, I represents the starting material and 89% purity, the single contaminant corre- MTX or MTX+GlQ, P represents products, and the addition of sponding in elution timeto 7-hydroxyMTX+Gl&. Methotrexate and NADPH (which is at saturating concentrations) is not explicitly NADPH were from Sigma. Dihydrofolate was prepared from folic shown. This is the simplest model capable of describing the onset of acid by the method of Blakley (1960). rHDHFR was prepared as inhibition. described previously (Prendergast et al., 1987). Trimethoprim and Stopped-flowFluorescence Measurements-Reaction-rate measDAMP were generous gifts from Dr. J. J. Burchall of Burroughs- urements were made on either of the stopped-flow instruments deWellcome, and from Dr. Vivian Cody, Medical Foundation of Buffalo, scribed above in the fluorescence mode. The excitation wavelength Inc., respectively. was 280 nm. Fluorescence was monitored through filters that elimiExperimental Conditions-All measurements were made at 20 “C nated light scattering but allowed broad-band transmittance of fluoin a buffer mixture containing 50 mM Tris, 25 mM acetate, 25 mM rescence. Reaction time courses were analyzed as described previously MES, 100 mM NaC1, and 0.02% sodium azide, pH 7.65. (Appleman et al., 1988). FluorescenceEmission Spectra-The emission spectra of enzymes, Determination of Steady-state Rates of Hfolate Formation by ligands, and theircomplexes were measured on a Perkin-Elmer LS-5 rHDHFR in the Presence of the Tight-binding Inhibitors MTX and fluorescence spectrophotometer interfaced to anApple IIe microcom- MTX+Glu4-When Hzfolate is added to a reaction mixture already puter. Spectrophotometer output is automatically corrected for vari- containing rHDHFR, NADPH, and either MTX or MTX+Glu,, the ations in lamp intensity and detector sensitivity as a function of time course of H4folate (P, product) formation is described well by, wavelength. P , = a(e-” - 1) + V,t Equilibrium Dissociation Constants-Dissociation constants for the interaction of ligands with enzymes were determined by titrating the change in enzyme or ligand fluorescence that accompanies complex where the exponential term represents the lag time required to reach formation. Titrations were typically carried out by serial addition of the steady-state rate ( V8J and is aconsequence of the slow release of concentrated ligands to a buffered enzyme solution contained in a inhibitor from rHDHFR. NADPH. MTX formed prior to addition of quartz cuvette. After addition of an aliquot to thecuvette, the solution Hzfolate. Similarly, when Hzfolate andMTXare simultaneously was mixed on a magnetic stirrer using a microbar and subsequently added to a reaction mixture containing rHDHFR and NADPH, the allowed to standin the fluorimeter cell holder for at least 5 min. The time course of product formation is described by excitation shutter was then opened, and a fluorescence emission scan P , = a K k t + v,t was performed. Corrections for ligand fluorescence and inner filter effects were calculated from parallel titrations in which enzyme was unless the concentrations of enzyme and inhibitor are very high (see absent or replaced with a quantity of tryptophan with a fluorescence “Progress Curve Analysis” under “Results”). The time course deintensity equal to that of the enzyme, respectively. The dependence scribed by this equation is a consequence of the slow rate of formation of fluorescence on ligand and protein concentration was fitted as of ternary complex with inhibitor. Inthe two cases discussed here the described by Dunn et al. (1978) in order to determine the value of KD. time course of product formation was fitted to the corresponding Enzyme Concentration-The concentration of rHDHFR was de- equation using nonlinear least squares techniques, and the best-fit termined by monitoring the decrease of fluorescencewhen the enzyme value of V, determined in thismanner was subsequently used in the was titrated with MTX in the presence or absence of NADPH, or by calculation of Ki (Equation 1below). Under conditions involving high titration of activity with MTX. Identical resultswere obtained by the enzyme and inhibitor concentrations, steady-state rates were calcutwo methods. lated as described previously (Appleman et al., 1988). Enzyme Assays-The activity of rHDHFR was determined by Determination of Ki for Enzyme Inhibitors-The values of Ki for following the decrease of NADPH and Hzfolate by absorbance meas- all inhibitors were calculated from the dependence of steady-state urements at 340 nm. Enzyme was preincubated with NADPH. Slow enzyme activity on inhibitor and Hzfolate concentration. Velocities reactions were monitored on the Cary 219 spectrometer interfaced to were corrected for blank rates (i.e. the rate of absorbance change a t a Northstar Horizon microcomputer using software purchased from 340 nm measured in the absence of enzyme). In thecase of MTX and On Line Instrument Systems. For example, the steady-state rates MTX+GlQ steady-state velocities (V) were fitted by nonlinear least used to calculate the K, values for trimethoprim and MTX and its squares techniques to theequation for tight-binding inhibitors, derivatives at low concentrations of rHDHFR were measured using this system. Fast reactions were monitored on either a Dionex D-110 or Hi-Tech Scientific PQ/SF 53 stopped-flow spectrometer having cells with path lengths of 1 and 2 cm, respectively, and dead times of 6 and 1.6 ms, respectively. The photomultiplier signal was recorded where A = [& + Ki,app- [Elmt, [Atotis the total concentration of
inhibitor binding to the recombinant human DHFR (rHDHFR). Our results differ from thoseobtained with DHFR from human cells in a number of regards. In particular, the Ki for a methotrexate polyglutamate is only 3-fold lower than thatfor MTX, and theincrease in k,,for MTX binding to the polyglutamate is only %fold greater than for MTX. Furthermore, we found that a polyglutamate of 7-hydroxymethotrexate is just as poor an inhibitor as 7-hydroxymethotrexate. Like complexes of MTX with DHFR from other sources the rHDHFR. MTX.NADPH complex undergoes a slow isomerization to a thermodynamically favored conformation, and thisisomerization contributes significantly to the overall binding.
1
-
Methotrexate Complexes of Human Dihydrofolate Reductase
10306
inhibitor, [Eltot is the total enzyme concentration, K;,spp= Ki (1 + [HZfolate]/Km),and V" is the velocity in the absence of inhibitor. For inhibition by Trmp, 7-hydroxyMTX, or 7-hydroxyMTX+Gl~velocities were similarly fitted to the classical equation for competitive inhibitors.
marized in Table I. Representative fluorescence emission spectra for rHDHFR and some of these enzyme-ligand complexes are shown in Fig. 1. Values of the dissociation constants (KDs)for the binding of NADPH to apoenzyme and for Trmp binding to the rHDHFR. NADPH complex were calculated V,,[H~folate] from the dependence of fluorescence intensity upon ligand V= Km(l+ [II/Ki) + [H~folate] concentration (Table I). KO for either the binary or ternary Determination of Dissociation Rate Constants by Competition Meth- MTX complex wastoo small to be accurately measured under ods-Preformed rHDHFR. NADPH.inhibitor complexes were mixed the experimental conditions by this technique. The value of with a second inhibitor so as to displace the first inhibitor from the the KO for Trmp binding (about 0.5 PM) is much larger than original complex, and the change in absorbance a t 340 nm or in for most bacterial DHFRs. Tight binding of MTX but weak fluorescencewas monitored. These changes reflect release of inhibitor binding of Trmp isa general property of DHFRs from vertefrom the original complex and formation of a new complex. brate sources. Consider the reaction sequence when the pre-existing ligand comInhibition of rHDHFR by Methotrexate and Methotrexate plex (EL,) is mixed with a second ligand, Lz, that competes for the Polyglutamates-Since titration with fluorescence detection same binding site. for the determination of the K D for MTX proved impractical we chose instead to measure the Ki for MTX andother related inhibitors of rHDHFR as a means of quantitating the tightness of ligand binding (Table 11). These inhibitors compete for binding solely with the substrate Hzfolate. This follows from linear competitive inhibition with respect to Hzfolate If (a) kz [L2]>> k-,, (b) kz [Lz] >> kl [LI], and (c) kz [L]/k-2 >> k , [L,]/k-, then the net reaction may be treated as ELI
k-1
-L1, + L
EL.
In thisfashion the rateof release of MTX, MTX+Gl%, or Trmp (L,) from ternary complexes with rHDHFR was assessed by competition with DAMP or MTX (Lz).Reversal of Trmp binding by MTX in the ternary complex wasmonitored in stopped-flow experiments through a decrease in energy transfer fluorescence. Reversal of either MTX or MTX+Glu4 binding by DAMP was monitored by the decrease in absorbance at 340 nm accompanying deprotonation of MTX at N1 as it dissociates from the enzyme complex. The time course of either fluorescence or absorbance change was analyzed as described previously (Appleman et al., 1988). Determination of the K,,, for Dihydrofolate with rHDHFR-The value of K,,, was determined by two methods. The first method consisted of fitting the initial rate of product formation as a function of dihydrofolate concentration using the Michaelis-Menten equation. The second method involved fitting full progress curves by integrating the Michaelis-Menten equation, P, =
TABLE I Fluorescence quenchingproperties and dissociation constants for various rHDHFR .lkand comDlenes Relative fluorescence" Ligands
330 nm
nM
1.00 0.07 None 0.19 NADPH 1.00 40 f 10
