Conformational change induced by coenzyme binding to bovine liver dihydrofolate reductase: a spectrofluorimetric study

435

Biochimica et Biophysica Acta, 1076 (1991) 435-438 © 1991 Elsevier Science Publishers B.V. 0167-4838/91/$03.50 ADONIS 016748389100118S

BBAPRO 33847

Conformational change induced by coenzyme binding to bovine liver dihydrofolate reductase: a spectrofluorimetrie study O. Kimet, M. C h a n v e t , M. Sarrazin a n d M. B o u r d e a u x Ddpartement de Physique, UFR de Pharmacie, Marseille (France)

(Received 19 June 1990)

Key words: Dihydrofolare reductase; Protein conformation; NADPH binding; Speclrofluorimetry; (Bovine liver)

When NADPH was added in excess to a bovine liver DltFR solution, a fluorescence peak due to an euer~ transfer mechanism was apparentat 450 rim. It did not vary over time. The intrinsic fluorescence peak of DHFR at 320 nm was quenched and this phenomenon increased over the time-course alter NADPH addition. This result was ascribed to a slow DEIFR conformational change induced by NADPH binding, which has never been previously described in such a long time scale (more than 30 rain). A kinetic scheme accounting for this mechanism has been proposed. Fmthermore, this interconversion between two protein conformers led to an increase in the initial apparent rate of the enzymatic reaction catalyzed by D I t F R .

Introduction Dihydrofolate reductase (DHFR) (5,6,7,8-tetrahydrofolate: NADP + oxidoreductase, EC 1.5.1.3.) catalyzes the NADPH-dependent reduction of 7,8-dihydrofolate (H2-folate) to 5,6,7,8-tetrahydrofolate (H4-folate). This enzyme is the molecular target for a number of clinically important antifolate drugs such as the antineoplastic agent, methotrexate, the antibacterial agent, trimethoprim and the antimalarial agent, pyrimethamine. Whatever the !igand, biological or exogenous, its binding to DHFR often results in an enzyme conformational change [i,2]. Studies on NADPH-DHFF, binary complex formation are rather scarce. The conformational change induced by NADPH binding is usually deduced from the modified properties of the binary complex, when compared to those of the enzyme alone, and has not yet been precisely defined. Gronenborn et al. [3], then Ratnam et al. [4] pointed out that the antigenic property of DHFR differed with or without bound NADPH. The difference was discussed in terms of a conformational change brought about by cocnzyme

Abbreviation: DHFR, dihydrofolate reductase. CorrespondeTtce: M. Bourdeaux, D£tpartement de Physique, UFR de Pharmacie, 27. Boulevard Jean Monlin. 13385 Marseille Cedex 5, France.

binding. Kitchell and Henkeus [5], in their study using circular dichroism and fluorescence polarization techniques, concluded to a stabilization of the protein towards heat denaturation due to NADPH binding, which they also explained by the enzyme conformational change. Subramanian and Kaufman [6], in their microcalorimetric experiments, attributed the large positive entropy change associated with the NADPH binding to the enzyme conformational change. Sometimes it was assumed that NADPH binding, because it influences other subsequent mechanisms, in particular antifolate binding, led to DHFR conformational change [7]. Furthermore, in vitro kinetic studies of the reaction catalyzed by DHFR also pointed to an NADPH-induced conformational change [8-10]. Baccanari et al. [8] described the transients observed in the E. coli DHFR activity assay in terms of the hysteresis concept. When enzyme was incubated with NADPH, the hysteretic complex was formed and then, upon addition of H2-folate, there was an apparent slow conversion from this inactive complex to an active enzyme species. Appleman et al. [9] recently compared the hysteretic behavior of DHFR from various sources. They reported that human and bovine DHFR exhibited hysteretic behavior in a time scale almost 1000-times shorter than that of the bacterial enzymes, i.e., in the 50-150 ms range. Furthermore, their model asmraed two conformers in equilibrium, and this equilibrium swings towards a less active conformer under the influence of substrates.

436 Stopped-flow fluorescence experiments have been performed to study the reaction of NADPH with bacterial enzymes [11,12]. They all pointed to the existence of at least two slow interconverting enzyme conformers, only one of which was able to bind NADPH. The intrinsic fluorescence quenching recorded over less than 1 rain was a biexponential process. No paper has dealt with a direct kinetic study of the enzyme conforraational change induced by NADPH binding to a mammalian DHFR. Our experiments evidence a fluorescence change caused by NADPH binding to D H F R from bovine liver. Furthermore, this quenching occurred in a time scale of up to 320 min. A kinetic scheme that accounts for this process is proposed. Materials and Methods DHFR was prepared according to the method of Kaufman and Kemerer [13]. The enz)ane purity was checked on SDS polyacrylamide gel electrophoresis; its specific activity was about 20 UI per mg. Fluorescence spectra of solutions containing purified enzyme saturated by coenzyme [14], were recorded on a Kontron SFM25 spectrofluorimeter equipped with thermostated cells, under gentle magnetic stirring (Hellma Cur-o-stir). Experiments were performed at 10, 15 and 28°C. The excitation wavelength was 280 ± 2.5 nm. At this wavelength, the absorbance was less than 0.100 to avoid any inner f'dter effect. Uncorrected emission spectra were recorded in a 300-500 nm wavelength range as soon as NADPH was added and several times after, up to 310 min later. All experiments were performed at least in triplicate. Spectra of enzyme solutions without coenzyme were also recorded under similar experimental conditions to check the stability of D H F R over time. The Raman contribution to emission was subtracted from all spectra. The light emitted at 320 nm and normalized with regard to the intensity emitted at 450 nm was plotted vs. time. Experimental data were fitted to theoretical curves by using a Basic program on IBM PC-AT based on non-linear regression. The influence of NADPH incubation on enzymatic activity was assessed using a spectrofluorimetric technique elaborated in the laboratory [15]. Assays without NADPH incubation were also performed for comparison.

Re~lts and Discussion Experiments performed at 28°C (Fig. 1) showed that adding NADPH to a D H F R solution led to a fluorescence peak at 450 nm [16]; tiffs peak did not vary within 30 min. It was attributed to an energy transfer from D H F R to bound NADPH [17]. Its intensity was proportional to the amount of NADPH bound to DHFR.

FN

2.0

1.0

!J

300

\\\ 350

400

450

500

Mnm)

Fig. l . N A D P H - D H F R complex fluorescence spectrum at 2 8 ° C . 3.,~ = 2 8 0 + 2 . 5 rim, 0.1 M phosphate buffer ( p H 6.9). [ N A D P H ] = 11 ~ M , [ D H F R | = 3.7/~M. - - , N A D P H addition; - - - - - - , 15 rain later; a n d . . . . . , 3 0 rain later.

The intrinsic protein fluorescence peak at 320 nm was classically quenched as soon as NADPH was added, but this process increased along the time-course (Fig. 1); this has never been described, at least in such a long time scale (30 rain). Then, the question was: did this process actually reflect a conformational change subsequent to NADPH binding or wag it an artefact, the origin of which remained to be determined? Because H2-folate was used during the affinity chromatography D H F R purification step, our first hypothesis was that the fluorescence change observed was due to the enzymatic reaction. However, NADPH and NADP + do not emit fluorescence light at 320 nm when excited at 280 nm; under these spectroscopic conditions, H4-folate is to a slight extent more fluorescent than H2-folate. Thus the enzymatic reaction was not able to provoke a fluorescence quenching due to substrate and coenzyme transformations. Another possibility was that spontaneous NADPH oxidation into NADP + occurred, as no mercaptoethanol was added in the solutions studied. The consequences of such an NADP + production were evaluated. Firstly, this reaction did not result in an increased absorbance, either at the excitation wavelength or at the emission one. Therefore the quenching increase over time was not due to an inner filter effect increasing over time. Secondly, NADP + is known to bind to D H F R in competition with NADPH and this reaction has been shown to produce a fluorescence quenching [12]. Nevertheless, such a competition was not likely in our case, since it would have required a high NADP + concentration with regard to the NADPH one; actually, the affmity constant of the oxidized species is more than 10-times lower than that of the reduced one [12]. Finally, after checking the stability of a D H F R solution maintained at 28°C during 30 min, we concluded that the fluorescence quenching change observed in the time-course was likely to be due to a slow conforma-

437

tional change of the NADPH-DHFR binary complex. This interconversion did not modify the fluorescence peak intensity at 450 nm. Thus the bound NADPH ratio was not altered by this process and the energy transfer mechanism was still able to occur. Furthermore, this peak was used to normalize the peak intensity at 320 nm so that the value obtained corresponded to a mole of NADPH-DHFR complex whatever the experimental DHFR concentrations used. It seemed of interest to study the kinetics of this conformational change, which were evaluated by using the normalized peak intensity at 320 nm. However, experimental data obtained at 28°C were not used for this purpose because, over 30 min, the NADPH peak diminished, as did the intrinsic fluorescence of a D H F R solution maintained at this temperature. These decreases indicated that a denaturation process probably occurred. Kinetic studies were therefore performed at 10 and 15°C. Experimental data are plotted in Fig. 2. The fluorescence intensity at 320 nm decreased vs. time and finally reached a plateau. To account for this phenomenon, we estabiished the foiiowing assumptions: (i) if two D H F R conformers, Et and E 2, existed initially in equilibrium, they identically bound the coenzyme, since the NADPH peak at 450 nm did not vary over time; and (ii) rapid NADPH binding induced a shift of the equilibrium concentrations of both binary complex conformers, BI and B2, and this process was slow.

The supposed scheme was the following: El

Bi

k~ ,ff k "..z + N A D P H ~ kI J~k _ I E2 !]2

where k are rate constants. Then, the equilibrium constants are:

K~ = [%|

"-

[~]

If [ET] is the total D H F R concentration, the Bz concentration vs. time is:

- K~ X,~) [ET] cxp(-- kat ) + ~ [ E T ] [Bz) = (l + K~ K~)(I-~ where k a is the apparent rate of transformation of B~ into 132. The normalized fluorescence intensity corresponding to 1 mol of N A D P H - D H F R complex is:

K~q- K~

a, +

a~.,¢,~

F=(az-a2)(l+K~)(l+Keq)exp(-kat)4 l + K e q

(11

where a I and a 2 represent the relative fluorescence quantum yields of B1 and 132, taking into account the NADPH fluorescence quantum yield at 450 nm. It must be noted that this equation is still valid when there is initially only one enzyme species. Experimental plots were fitted zo Eqn. 1, which represents a sum of an exponential function and a constant term. Theoretical curves caleulatec from three different reagent concentrations are shown in Fig. 2. The apparent rate constant, ka, of the conformational change was 2.36-10 -2 rain - I at 10°C and 4.63-10 -2 rain - I at 15°C, The corresponding flnorescenc¢ changes t l / 2 were 29.4 and 15.0 rain. Thus, a temperature increase sped up the conformational change process.

F. 2.5~

2.0

1.5'

1.0'

0.5"

2"...

i "

........

i ........

t ........

i ........

0

o .m,d 50 100 140 200 2~) 3'00 Fig. 2. Kinetics of the DHFR confonnational change induced by NADPH binding. At 10°C: o, [NADPH]=10.8 FM, [DHFR]=6.5 FM; [3, [NADPH]= 10.2 FM. [DHFR]= 7.5 FM; z~,[NADPH]= 11.9/zM, [DHFR]= 6.7/~M; , fitted curve(r ~ 0.99); and FN = 1.70eXlR-2.36. 10-2t)+0.64. At 15°C: A, [NADPH]=ll.6 FM, [DHFR]=5.I /AM; t. [NADPH]=10.3 FM, [DHFR]=7.I ItM; O, [NADPH]=ll.8 FM, [DHFR]= 4.1/~M; . . . . . . , fitted curve(r = 0.99); and F~4~ 1.65 exp(-4.63. l0-2t ) + 0.46.

438 The initial theoretical normalized fluorescence intensity was 2.34 at 10°C and 2.11 at 15°C, i.e., more than 3-times lower than that of a corresponding D H F R solution. Thus, NADPH binding provoked a strong quenching which we considered to be instantaneous in our time scale. It cannot be excluded that a conformational change occurred during this step, as reported by Appleman et al. [9], but ::ts progress could not be recorded by our technique. The fluorescence quenching may also be explained by the fact that the D H F R tryptophan 24 residue is known to be implicated in the NADPH site; therefore its emission was probably inhibited by coenzyme binding. The plateau value at 10°C (0.64) was higher than that at 15°C (0.46). This result agrees with the classical fluorescence variation vs. temperature; it might also be explained by a Keq increase with temperature. Moreover, the plateau values were markedly lower than the corresponding theoretical initial values. Such a difference might be explained by the involvement of several fluorophores per D H F R mole in the quenching process. In fact, three tryptophan and six tyrosin residues are distributed along the protein chain. The microenvironment of some of the residues might be changed. Albeit, there was not a gross conformational change, since the energy transfer to NADPH was not modified. The incubation of D H F R with NADPH for 90 min caused an increase in the initial rate of the enzymatic reaction: the apparent activity was multiplied by 1.2 when incubation was done at 10°C and by 1.4 at 15°C incubation. At the former temperature, the conformational change was not completely achieved, but it was finished at the latter one. Hence, there was a hysteretic behavior as defined by Frieden [17]. In conclusion, it would be of interest to study the influence of this slow interconversion of NADPHD H F R complexes on the binding of biological sub-

strates, H2-folat¢ and H4-folate, as well as of antifolate drugs. This work is in progress. Aclmowledgments The authors thank H. Bouteille for his graphical work and M. Vidalin for typing the manuscript. Ref¢~enees 1 Matthews"D.A.,Alden, R.A., Freer,S.T.,Xuong,N.H. and KrauL J. (1979)J. Biol. Chem. 254, 4144-4151. 2 Appleman,J IL, Prendergast,NJ., Delcamp,TJ., Freisheim,J.H. and Blakley, R.L. (1988) J. Biol. Chem.263, 10304-313. 3 Gronenborn,A.M., Papadopoulos, P. and Clore, G.M. (1984) J. Biol. Chem.259, 1082-1085. 4 Ramam,M., Tan, X., Prendergast,N.J., Smith,P.L. and Freisheim, 3.H. (1988) Biochemistry27, 4800-4804. 5 Kitcbell, B.B. and Henkens, ILW. (1978) Biochim. Biophys.Aeta 534, 89-98. 6 Subramanian,S. and Kaufman,B.T. (19"/8)Proc. Natl. Acad. ScL USA 75, 3201-3205. 7 Roberts,G.C.K. (1987) Biochem.Soc. Trans. 15, 762-766. 8 Baccanari, D.P. and Joyner, S.S. (1981) Biochemistry20, 17101716. 9 Appleman,J.IL, Beard, W.A., Delcamp, T.J., Prendergast, N.J., Freisheim, J.H. and Blakley, ILL. (1989) J. Biol. Chem. 264, 2625-2633. 10 Penner, M.H. and File.den, C. (1985) J. Biol. Chem. 260, 53665369. 11 Dunn, S.M., Batchelor, J.G. and King, R.W. (1978) Biochemistry 17, 2356-2364. 12 Andrews, J., Fierke, C.A., Birdsall, B., Ostler, G., Feeney, J., Roberts, G.C. and Benkovic, SJ. (1989) Biochemistry28, 57435750. 13 Kanfman,B.T.and Kemerer,V.F. (1976)Arch. Biochem.Biophys. 172, 289-300. 14 Blakley, ILL. and Cocco, L. (1984) Biochemistry23, 2377-2383. 15 Rimer, O., Chauvet, M., Bourdeanx M. and Briand, C. (1987) J. Biocbem. Biophys.Methods 14, 335-342. 16 Perkins,J.P. and Bertino, J.IL (1966) Biochemistry5, 1005-1012. 17 Velick, S.F. (1958)J. Biol. Chem. 233, 1455-1459. 18 Frciden, C. (1979) Annu. Rev. Biochem.48, 471-489.