Nucleotide binding to tubulin-investigations by nuclear magnetic resonance spectroscopy

Biochi~ic~a

ELSEVIER

et Biophysica A~ta Biochimica et Biophysica Acta 1292 (1996) 77-88

,,,

Nucleotide binding to tubulin-investigations by nuclear magnetic resonance spectroscopy Sadananda S. Rai 1, Kavita Kuchroo, Sitapati R. Kasturi * Tata Institute of Fundamental Research, Homi Bhabha Road, Colaba, Bombay-400 005, India Received 29 March 1995; revi~ed 9 August 1995; accepted 11 September 1995

Abstract In an attempt to distinguish between the interaction of GTP and ATP with tubulin dimer, high-resolution ~H- and 31P-NMR experiments have been carried out on the nucleotides in the presence of tubulin. The location of the ATP binding sites on the protein in relation to the GTP sites is still not clear. Using NMR spectroscopy, we have tried to address this question. Evidence for the existence of a site labelled as X-site and another site (labelled as L-site) for both the nucleotides on tubulin has been obtained. It is suggested that this X-site is possibly the putative E-site. In order to gain further insight into the nature of these sites, the Mg(II) at the N-site has been replaced by Mn(II) and the paramagnetic effect of Mn(II) on the linewidth of the proton resonances of tubulin-bound ATP and GTP has been studied. The results show that the L-site nucleotide is closer to the N-site metal ion compared to the X-site nucleotide. On the basis of these results, it is suggested that the L-site of ATP is distinct from the L-site of GTP while the X-site of both the nucleotides seems to be same. By using the paramagnetic effect of the metal ion, Mn(II), at the N-site on the relaxation rates of tubulin-bound ATP at L-site, distances of the protons of the base, sugar and phosphorous nuclei of the phosphorous moiety of ATP, from the N-site metal ion have been mapped. The base protons are = 0.7-1 nm distant from the N-site metal ion, while the protons of the sugar are = 0.8-1 nm from this metal ion site. On the other hand, the phosphorous nuclei of the phosphate groups are somewhat nearer ( = 0.4-0.5 nm) from the N-site metal ion. Keywords: ATP; GTP; Binding site; Distance mapping; Microtubule; Tubulin; NMR

1. Introduction Microtubules are assembled from the dimeric protein tubulin, an a - / 3 heteroclimer of 40% sequence similar polypeptides of molecular mass of 50 kDa each. The tubulin dimer is known 1:o bind two guanine nucleotides, one non-exchangeable (N-site) and the other exchangeable (E-site) GTP [1]. The N-site binds GTP so tightly that it

Abbreviations: N-site, non-exchangeable nucleotide-binding site of tubulin; E-site, exchangeable ~ucleotide-binding site of tubulin; L-site, exchangeable low-affinity nucleotide binding site of tubulin; Pipes, 1,4piperazinediethanesulfonic acid; EGTA, ethylene glycol bis( fl-aminoethyl ether)-N,N,N',N'-tetraacetic acid; SDS, sodium dodecyl sulfate; NMR, nuclear magnetic resonance. * Corresponding author. Fax: +91 22 2152110; e-mail: [email protected]. 1 Present address: Laboratov/ of Biochemical Pharmacology Bldg. 8, Room No. 2A-23, National In:;titutes of Health, Bethesda, MD 20892, USA. 0167-4838/96/$15.00 © 1996 Elsevier Science B.V. All rights reserved SSD10167-4838(95)0019~;-0

can be released only upon denaturation [2], while GTP or GDP at the E-site can be exchanged with the free nucleotide in the medium [2]. During the polymerisation of the protein, GTP at the exchangeable site is hydrolyzed [2,3], while the N-site GTP remains unaffected [2,3]. It is also known that GTP binding at E-site to tubulin is strongly dependent on Mg 2+, whereas G D P binding is independent of Mg 2+ [4]. The divalent cations also are known to play an important role in the in vitro assembly of tubulin [5,6]. Tubulin dimer binds one Mg 2÷ tightly and can be removed only by denaturing the protein [5,7]. It has been found that this high-affinity site is most likely the GTP-occupied N-site and this site exchanges divalent cations such as Mn 2+ very slowly [8,9]. Further, tubulin which has GTP at the E-site possesses an additional high-affinity site for the divalent metal ion at the E-site, whereas tubulin with G D P at the E-site possesses only low-affinity site for the metal ion [9]. In addition, there are other weak binding sites for the metal ion most likely associated with the acidic carboxy terminal

78

s.s. Rai et al./ Biochimica et BiophysicaActa 1292 (1996) 77-88

region of both subunits and are required for efficient assembly of tubulin into microtubules [5,7]. Osei et al. [10] have provided further evidence that the tightly bound divalent metal ion (Mg 2+ or Mn 2+) is in fact associated with the N-site GTP. It is well established that GTP bound at the E-site is hydrolyzed to GDP during the assembly of microtubules [3,11]. However, the role of ATP in promoting microtubule assembly is not clearly established. ATP is known to induce the formation of aggregates of tubulin tings [12] which in turn serve as nucleation centers in the polymerisation of tubulin into microtubules [12]. The effect of ATP on microtubule assembly has been examined by a number of investigators [12-17]. It is not clear whether ATP interacts with tubulin by binding at the GTP binding sites or by binding at a site(s) distinct from these sites [ 13,14, l 7]. Zabrecky and Cole [13,14] proposed that ATP interacts with tubulin at a binding site on c~-tubulin and is neither the N-site nor the E-site at which GTP binds. On the contrary, Duanmu et al. [17] suggested that tubulin polymerisation in the presence of ATP is mediated through a feeble interaction at the E-site. Farr et al. [18] showed that ATP binds to free /3 subunit of tubulin. Our recent experiments on the binding of Ant-ATP, a fluorescent analogue of ATP, to tubulin using steady state fluorescence techniques have also suggested that the E-site of ATP and GTP is the same [19]. In this work, we have investigated the interaction of ATP and GTP with tubulin using one-dimensional Fourier transform Nuclear Magnetic Resonance (NMR) spectroscopic methods and our NMR results provide evidence for the existence of a site labelled by us as X-site and another site labelled as L-site (to imply low affinity site) for ATP and GTP. It is found that X-site for ATP and GTP is the same and it is suggested that this X-site could possibly be the putative E-site. Using nuclear magnetic resonance relaxation methodology [20-23] in the presence of the paramagnetic divalent metal ion, Mn 2+, located at the N-site, the distances between the N-site metal ion (Mn 2+) and the nuclei (1H and 31p) of the ATP bound at the L-site have been mapped.

2.2. Preparation of tubulin

Microtubule protein was purified from goat brain by three assembly/disassembly cycles according to the procedure of Shelanski et al. [24] in a buffer containing 0.1 M Pipes (pH 6.9), 1 mM MgC12, 2 mM EGTA, 8 M glycerol. Tubulin was purified from microtubule associated protein by chromatography on phosphocellulose (Whatman P-11) column equilibrated with buffer (0.1 M Pipes, pH 6.9, 1 mM MgC12, 2 mM EGTA, 0.1 mM GTP). Tubulin fractions were collected and concentrated using Amicon membrane cones (CF50A). Purity was checked by SDS gel electrophoresis on heavily overloaded gels. Protein concentration was estimated by the method of Lowry et al. [25]. Protein was stored at - 8 0 ° C until used but most of the experiments were carried out with freshly prepared protein. 2.3. Exchange of magnesium by manganese

In order to replace magnesium by manganese at the high-affinity site (N-site), concentrated protein was incubated with excess GDP ( > 5 mM) and the free nucleotide was removed by gel filtration through Sephadex G-25 column (1 × 10 cm) equilibrated with the experimental buffer. The concentrated protein was divided into two aliquots. To one aliquot, 0.5 mM MnCI 2 was added and to the other aliquot 0.5 mM MgCI 2 was added. These aliquots were incubated at 4°C for 30 minutes following the procedure of Correia et al. [9]. The low-affinity metal ions and free metal ions were removed by passing through a Chelex-100 column. The metal ion under these conditions will be located at the N-site only and will be occupied by either Mg 2÷ or Mn 2+ in the paramagnetic protein solution and by Mg 2÷ in the diamagnetic protein solution. Protein and the nucleotide solutions were lyophilised separately and the nucleotide at appropriate concentration was added to the NMR tube during the experiment. For distance mapping measurements, protein and ATP in the appropriate buffer were lyophilised together and were dissolved in 0.5 ml D20 (99.96 atom% D). 2.4. Atomic emission spectroscopy

2. Materials and methods 2.1. Reagents

Pipes, EGTA, GDP(Type I) and GTP (Type I and II), glycerol, ATP (Type I), D20 (99.96 atom% D) were obtained from Sigma (St. Louis, MO, USA). Chelex-100 resin was supplied by Bio-Rad (Richmond, CA, USA). MnCI 2 and MgCI 2 solutions used as standard in atomic emission spectroscopic measurements were prepared from chemicals supplied by Sigma. All other chemicals used were of analytical grade. Alkaline phosphatase was obtained from Promega (USA).

Magnesium and manganese concentrations were determined on an inductively coupled Plasma Atomic emission spectrometer (ICP-AES Plasma Lab 8440 from GBC, Australia). 2.5. Nucleotide estimation

To estimate the total nucleotide content of tubulin, gel filtered protein sample was precipitated with 5% perchloric acid and the precipitate was removed by centrifugation. The supernatant was mixed with 0.5 M Pipes (pH 7.0) and the pH was adjusted to 7.0 by adding NaOH. The optical density of the supernatant was measured at 252 nm where

79

S.S. Rai et aL / Biochimica et Biophysica Acta 1292 (1996) 77-88

0.5

E 0.4 0.5 8

8 o.z

.O

0.1 0.O

0

5

I

I

I0

15

20

Time (rain) Fig. 1. Polymerisation of tubulin (1.7 mg/ml) induced by GTP (0.5 mM) in the assembly buffer (0.1 M Pipes, 0.5 mM MgCI2, 10% DMSO, 1 mM EGTA, pH 6.9). (a) Sample passed through Sephadex and lyophilized for NMR experiments (b) lyophilized sample kept at 25°C after typical NMR experiments ( ~ 2 h). Maximura O.D. at 350 nm for a freshly prepared protein of the same concentrations (data not shown) before lyophilisation is also same.

guanosine absorbs. This method gives the total content of GTP and GDP.

2.6. Polymerisation study Polymerisation experiments in the presence of GTP were carried out in assembly buffer (0.1 M Pipes, 0.5 mM MgCI 2, 10% DMSO, 1 mM EGTA, pH 6.9). Polymerisation was initiated by adding the appropriate nucleotide and then warming to 37°C, in a thermostatically controlled cuvette chamber of a Milton Roy spectrophotometer (Model 1201). The increase in turbidity was normally monitored at 350 nm unless otherwise mentioned. Tubulin is known to be a labile protein and hence it is necessary to monitor that tubulin retains its property of self-assembly after passing through various steps such as gel-filtration, chelexing and lyophilisation. Fig. 1 shows the data of tubulin polymerisation after lyophilisation (curve a) and after two hours of N M R experiments (curve b). It can be seen from the figure that tubulin retains most of its assembly competence. It was also observed that the polymerisation is reversible. Some denaturation during the course of N M R experiments cannot be ruled out as can be seen from curve b.

2.7. NMR Measurements

whereas 31p_NMR spectra were recorded in 50 mM Pipes buffer (pH 6.9). The temperature of the sample was regulated to _+0. I°C and all measurements were carried out at 25°C unless otherwise mentioned. For I H-NMR experiments, TSP was used as internal reference and for 31P-NMR experiments, 85% H3PO 4 placed in a capillary was used as an external reference for measurements of chemical shifts. Spin-lattice relaxation time (T 1) measurements were made using the standard inversion recovery sequence (180°-~--90°), where ~- is the variable delay between the 180 ° and 90 ° pulses. In the case of ~H-NMR experiments, solvent resonance was suppressed using gated decoupling technique to avoid dynamic range problem. The errors quoted in Table 2 for (PTlp) -1 and (pT2p) -1 values are standard deviations between measurements on different samples. Relaxation measurements on a particular sample were finished within two hours or less in order to avoid possible denaturation of the sample. In view of the high sensitivity of the high field N M R spectrometer, only 32 scans for IH and 64 scans for 3~p-NMR spectra were required for obtaining good N M R spectra. However, in order to observe the peaks assigned to the X-site, larger number of scans were required.

2.8. Analysis of NMR relaxation data The effect of a paramagnetic ion on the spin-lattice relaxation rate (1~Tip) of a ligand nucleus in a macromolecular complex is given by [26-31] 1

q

PTle

TIM + ~'M

where TI~ is the relaxation rate of the observed nucleus of the ligand in the macromolecule-cation-ligand complex,

gL A.

Dp, ~H-NMR experiments were carded out at 500 MHz on an AMX-500 Bruker Fourier Transform N M R spectrometer. 3~p-NMR experimems were carded out on the same spectrometer at 202.4 MHz with a 5 mm inverse probe. The sample size was 0.5 ml and 99.96% D 2 0 was used as internal lock. D 2 0 used for this purpose in the protein sample was passed through a Chelex column. In order to avoid dynamic range prohlem, proton N M R spectra of the solutions were recorded in 10 mM Pipes buffer (pH 6.9),

(1)

•

812

I

.

.

.

.

I

Fig. 2. 500 MHz IH-NMR spectra of (A) GTP (5 mM) in Pipes buffer in D20, pH 6.9 at 25°C and no protein. (B) Tubulin proton resonances recorded in Pipes buffer 10 mM, [tubulin]= 400 p.M (C) alkaline-phosphatase treated-tubulin in Pipes buffer, 10 mM (pH 6.9), [tubulin]= 400 /zM. No external GTP or ATP is added to these protein samples. Number of scans for A is 128 and for B and C it is 1000 scans. Incubation with alkaline phosphatase treatment was done at 25°C for 30 rain. Free nuc|eotide in tubulin and alkaline phosphatase treated samples has been removed by gel filtration. All spectra shown are fully relaxed spectra (10 s relaxation delay).

S.S. Rai et al. / Biochimica et Biophysica Acta 1292 (1996) 77-88

80

where r c is the correlation time for the dipolar interaction between the unpaired electron on the paramagnetic ion and the relaxing nucleus in the complex and tot is the Larmor frequency of the relaxing nucleus. Structural information can be obtained using Eq. (2) provided Tl~ can be determined from Eq. (1). T~-d can be obtained from the experimentally d e t e r m i n e d (pTLp) I values using Eq. (1) provided fast exchange conditions prevail (i.e., TIM >> rM). Further, rc can be determined either by using a frequency dependence study of T(M~ or from the methods described below. The theory and the assumptions involved in arriving at Eq. (2) for obtaining structural information have been well tested for Mn(II) ion-macromolecular complexes [22,23,30-32]. r c can be obtained using the relation [30,32]

OI

c

TiM 2 7 T2---~ - 3 to~z~2+ g o'

8

~

2_

A _2 I

00m

'0'.2

00m

6'.0

Fig. 3. Proton NMR spectra of GTP free (A) and in the presence of tubulin (B-G) for various concentrations of GTP added externally. [tubulin] = 400 /xM. GTP concentrations used in (A) is 5 mM. Concentration of GTP added externally is (B) 0 mM (C) 0.3 mM (D) 0.7 mM (E) 1.7 mM (F) 3.2 mM (G) 5 mM. All solutions are contained in Pipes buffer 10 raM, pH 6.9 and 25°C. Number of scans for (A) is 128 and for B-G, it is 256. All spectra shown are fully relaxed spectra (10 s relaxation delay).

where T2r~ is the transverse relaxation rate of the observed nucleus of the ligand in the macromolecule-cation-ligand complex. Under fast exchange conditions, the contribution of r ~ can be neglected giving T I M / T 2 M = T I p / T 2 p . T~ol is the paramagnetic contribution to the spin-spin relaxation rate of a iigand nucleus in the macromoleculecation-ligand complex and is determined from the full width of the resonance line at half-height (Avj/2) according to the relationship 1 / T 2 p = II(A/)I/2). However, the possibility of intermediate exchange where there could be a small contribution from r M in Eq. (1) has to be considered. It is possible to obtain r c and r M in the case of intermediate exchange as suggested by Lanir and Navon [32,33] Ti M

r M is the residence time of the ligand in the complex, p is the mole fraction of the ligand in the complex and q is the relative stoichiometry of the ligand and the paramagnetic ion in the macromolecule-cation-ligand complex. ( p T lp)- l values are determined by subtracting the diamagnetic contribution from the observed paramagnetic relaxation rates for the protein-cation-ligand complexes. T~-r~ is related to the distance, r, between the paramagnetic ion on the protein and the ligand nucleus as follows.

(4)

[(r2

+

-

+ rM =

TIp

+ rM)

(5)

From the experimentally determined Tlp/T2p values for two protons in the same ligand, TIM/T2M and r M can be determined using Eq. (5). Using these values, in Eq.(4) r e can be determined. In the present work, % has been determined using both the methods.

3. Results

in which C is a constant which depends on the nature of the paramagnetic ion and the relaxing nucleus. For Mn (II)-lH and Mn(II)- 31p, the values of C are 81.2 and 60.1 nm s -1/3 respectively [31]. For the Mn(II) ion in the macromolecular complex and at the NMR frequencies that are used in this work, f(%) can be simplified [30,31] to give 3% f ( r c ) = 1 + w~r/

(3)

In order to investigate the interaction of ATP and GTP with tubulin, proton NMR spectra have been recorded under various conditions.

3.1. Interaction of GTP with tubulin Fig. 2 shows a portion of the proton NMR spectra of tubulin in Pipes buffer (10 mM). Also the spectrum of free GTP (A) in the same buffer is shown for comparison. Peak a at 8.14 ppm in Fig. 2A is due to the H8 resonances of the base while the doublet at 5.93 ppm (peak b) is due to the

s.s. Rai et al. / Biochimica et Biophysica Acta 1292 (1996) 77-88

HI' sugar protons of GTP. Fig. 2(B) shows a portion of the proton N M R spectrum of tubulin with prominent peaks a' and b' at 8.19 ppm and 5.94 ppm to be noted. Peak a' at 8.19 ppm is assigned to the H8 protons of tubulin-bound GTP at a site labelled as X-site by us, and peak b' to sugar protons HI' of tubulin-bound GTP at this site. A downfield shift of about 0.05 ppm has been observed for the peak a' compared to that of the free GTP (peak a). On the other hand, a downfield shift of only 0.01 ppm has been observed for the HI' protons of the sugar ring of GTP at this site (peak b') compared to the free GTP (peak b). Incubation of tubulin with alk~dine phosphatase is generally done to remove the nucleotide at E-site [34] and Fig. 2C shows that the peaks a' and b' assigned to the guanine nucleotide bound at the X-site are not observed suggesting that the peaks a' and b' are probably due to the protons of the guanine nucleotide at E-site. The nucleotide content of gel filtered tubulin has been estimated and the ratio between the tubulin and guanine nucleotide concentrations, [protein]/[nucleotide], is of the order 1 : 1 . 5 - 1 . 6 in our sample. This suggests fitrther that the protein bound peaks (a', b') arising from the X-site of GTP of tubulin could be due to GTP at E-site. The peaks in (B) are broader compared to those of the free nucleotides due to the interaction with the protein. It is, however, interesting to note that the peaks assigned to this X-site are still narrow compared to that one observes for protein-bound resonances as well as protein resonances. The narrow resonances observed for the nucleotide at this site, however, go

81

against the broad linewidths that would be expected for the E-site GTP consistent with the tight binding reported for MgGTP at E-site. However, it must be noted that in our preparations, the free metal ion as well as the metal ion at E-site has been removed by chelexing. In the absence of the metal ion, GTP is known to bind weakly at E-site [4]. Hence, the nature of binding in the absence of the metal ion could be such as to give narrow resonance. The narrow resonance may also be because the X-site region where the GTP binds may be in a flexible region of the protein. Evidence that the peak a' in Fig. 2(B) is due to the X-site guanine nucleotide and is not due to the N-site nucleotide comes from the proton N M R spectra recorded in the presence of the exogenous nucleotide added externally to the tubulin solution as shown in Fig. 3. Fig. 3A shows the H8 (peak a) and H 1' (peak b) resonances of free GTP. Fig. 3B is the spectrum of tubulin (bound peaks are not seen due to the less number of scans and reduced scale). Fig. 3 C-G show the tubulin N M R spectra with different concentrations of GTP added to tubulin solutions. Two observations can be made from these NMR spectra. First, the peaks a' (H8 protons) and b' (HI' protons) due to X-site GTP increase in intensity due to the additional binding of GTP at this site. The fact that externally added GTP can bind at this site rules out the possibility that these peaks could be due to the N-site GTP. Further, the peak a' and b' could be observed even in the absence of externally added GTP (Fig. 2B) confirming that these peaks a' and b' (doublet) are due to GTP resonances bound at a site on the

h

d!

g

h

?

h

D-

~

-

C.-,

a~

.

f

h J ~ .

.

,

d

-~"

A

•

~l~m

!

8.4

'

~]I

I

.2

'

pam

i

6.2

Fig. 4• Proton NMR spectra of ATP in the presence and absence of tubulin. (A) In the absence of tubulin; [ATP] = 5 raM; number of scans = 128 (B) In

the absence of ATP but in the presence of tubulin; [tubulin] = 400 p.M. Other spectra are of ATP at various concentrations in tubulin solutions containing [tubulin] = 400 k~M (C) 0.5 rraMATP (D) 1.5 mM ATP (E) 3.5 mM ATP (F) 5 mM ATP. The number of scans for each spectra is 256 scans These are fully relaxed spectra. All spectra are recorded in Pipes buffer 10 mM (pH 6.9) at 25°C.

82

S.S. Rai et al. / Biochimica et Biophysica Acta 1292 (1996) 77-88

protein different from N-site. Second, additional peak c starts appearing when GTP is added externally. This peak grows in intensity, as is expected, when more of GTP is added due to the additional binding of GTP at this site on tubulin. This peak is also broader than the peak a' and also that of free GTP. Hence, the peak c arises due to the exchange of GTP between the GTP bound on tubulin at this site and free GTP. As this peak c appears only when external GTP is added to tubulin, this peak must be due to GTP exchanging between a third site which we call as low affinity site or L-site and free GTP. The peak c is therefore due to H8 protons of GTP bound at this site (L-site) and free GTP. It cannot be due to free GTP alone because of the large broadening that is observed. Hence, the signal c contains information about the L-site. The maximum difference in the chemical shift value of the H8 resonances of free GTP and the peak c (assigned to H8 resonance of GTP at L-site in exchange with free GTP) is, however, very small ( = 0.007 ppm) but is discernible on this high frequency instrument. It may be further noted that the peak a' corresponding to the X-site becomes narrower as the peak c corresponding to the nucleotide bound at L-site starts appearing. This suggests an exchange between the nucleotide bound at the X-site and L-site. Evidence for this exchange between the X-site and L-site has been obtained by temperature dependence studies (data not shown). Further, it may be noted that the resonance c becomes narrower as the nucleotide concentration is increased due to the exchange between the nucleotide bound at the L-site and the free nucleotide. This in turn has an effect on the resonance a' making it narrower. It may also be noted that the peaks a' and c move closer to each other as the concentration of the nucleotide is increased suggesting an exchange between the two sites. In order to check whether the two resonances (a' and c) observed for GTP interacting with tubulin are due to a specific interaction or due to artefacts such as viscosity or non-specific interactions, the following control experiments have been done. GTP spectra recorded in the presence of bovine serum albumin of similar concentration did not exhibit any difference in chemical shifts for H8 and HI' resonances from that of free GTP. Similarly, GTP spectra have been recorded in 5% glycerol solution in order to check the possible effects of the viscosity of the solution on chemical shift measurements although viscosity will only broaden the lines but not affect the chemical shifts values. GTP resonances H8 and HI' recorded in 5% glycerol solutions have exhibited chemical shifts identical to that of free GTP. In the case of tubulin heated above 60°C, these peaks (a' and c) seem to merge into one but detailed experiments as a function of the denaturant are planned to probe the fate of these signals (a' and c) as the protein denatures. The possibility that the signal a' could be due to free G D P resulting from limited hydrolysis of GTP has been explored. The H8 proton resonances of free G D P resonate

at 8.12 ppm slightly upfield from that of free GTP. Similarly, GDP-tubulin prepared by incubating tubulin with G D P exhibits the resonances for the H8 protons, one at 8.17 ppm and the other at 8.13 ppm corresponding to the X- and L-sites referred to in the case of GTP (data not shown). 3.2. I n t e r a c t i o n o f A T P w i t h tubulin

In order to probe the nature of binding of ATP to tubulin, we have recorded spectra of ATP in the presence of tubulin similar to those shown for GTP in Fig. 3. Fig. 4 shows a portion of such spectra for ATP. The peaks, d at 8.57 ppm and e at 8.27 ppm in Fig. 4A are due to H8 and H2 protons respectively in the base whereas the peak f at 6.15 p p m is due to H I ' sugar protons of ATP in the absence of tubulin. Fig. 4B shows the spectrum of tubulin in this region (the intensity appears smaller due to the reduced scale used in this figure). Fig. 4C shows the spectra of ATP in the presence of tubulin. Peak d' at 8.59 ppm arises due to the H8 ATP protons binding at a site which we call it as X-site of ATP and peak g at 8.54 ppm arises due to the H8 of A T P protons binding at a site we call it as L-site and free ATP in exchange, quite analogous to the GTP binding. Similarly, peak h at 8.27 ppm is due

B

I

Dpm

I

8.4

'

'

I

8.2

Fig. 5. Proton NMR spectra of ATP in the presence and absence of tubulin showing competition between ATP and GTP for the X-site (A) free ATP (5 mM) (B) free GTP (5 mM) (C) proton NMR spectra of tubulin alone; [tubulin]= 400 /xM (D) Proton NMR spectrum of ATP (0.3 raM) in the presence of tubulin (400/xM). Notice the decrease in the intensity of the peak a' due to H8 protons of GTP at X-site and the appearance of the peak a at position corresponding to free GTP H8 resonance position, thus demonstrating that GTP is displaced by ATP at the X-site. All NMR spectra were recorded in Pipes buffer 10 mM (pH 6.9) at 25°C. Number of scans is 1000 except for (A) and (B) which is 128. All these spectra are fully relaxed spectra.

S.S. Rai et al. / Biochimica et Biophysica Acta 1292 (1996) 77-88

to H2 protons of ATP ~inding at L-site and free ATP in exchange. The peak due'. to H2 protons of ATP binding at X-site arises at 8.23 ppm (peak e') on the upfield side of peak h as ATP concentration is increased. HI' sugar protons of ATP binding at the X- and L- sites cannot be resolved and they occur as a single slightly broadened resonance g' at 6.16 ppm like in the case of GTP binding. However, evidence for some structure in the peak c of Fig. 3F and 3G as well as in peaks g of Fig. 4 can be noticed if one examines these peaks closely at higher nucleotide concentrations. This arises due to the exchange of ATP binding at this site being in exchange with the exogenous free nucleotide. This is in quite contrast to the ATP or GTP binding at X-site which exchanges slowly on NMR time scale resulting in a separate peak due to the nucleotide bound at X-site. However, our preliminary temperature dependence studies (not shown) indicate that some exchange between the nucleotides at X-site and L-site exists resulting in narrowing of the resonance at X-site as nucleotide concentration is increased (Fig. 4E and F) like in the case of GTP. Control experiments similar to that of GTP experiments were carried out in this case also. It was found that ATP in BSA solutions of concentrations similar to that of tubulin did not exhibit chemical shifts different from that of free ATP H8, H2 and HI' protons. Similarly, ATP spectra recorded in 5% glycerol solutions to check the effect of viscosity did not show any difference in the chemical shifts of the H2 and H 1' while a small upfield shift of 0.015 ppm has been observed for the H8 peak. 3.3. Is the X-site of ATP and GTP the same?

In order to ascertain whether the X-site of ATP and GTP is the same or they are two distinct sites, a competi-

Table 1 Effect of the paramagnetic ion, at N-site on the linewidth of the H8 resonances of the nucleotides at X- and L-site in tubulin-Mn-nucleotide complex System

Free ATP Mg-ATP-tubulin Mn-ATP-tubulin Free GTP Mg-GTP-tubulin Mn-GTP-tubulin

Avl/2 (Hz) a Free

X-site

L-site

2.6 __+0.2 2.4 ± 0.2 -

3.2 + 9.2 ± 3.1 ± 6.1 ±

0.2 1.2 0.1 0.2

4.5 20.5 3.7 9.2

± 0.2 + 2.2 ± 0.2 ± 1.0

All solutions are in Pipes buffer 10 mM (pH 6.9) at 25°C. The concentrations of tubulin, M g 2+ (or Mn 2+ in the paramagnetic complex) and the nucleotide are the same in all these experiments. [Tubulin]= 213 /,M; [ATP or GTP] = 1.4 mM; [Mg 2+ ] = 2.8 /xM; [Mn 2+ ] = 210 /xM in the paramagnetic complex. In the diamagnetic complex all the N-site metal ion sites are occupied by M g 2+. The divalent metal ion concentrations reported are those bound to tubulin and not added concentrations. a Avl/2 is the full line width at half-height determined by Lorentzian line shape fitting.

83

tion experiment was carried out. The results of such an experiment are as shown in Fig. 5. Fig. 5A and B show the proton NMR spectra of free ATP and GTP respectively and peak a is due to the H8 protons of GTP and peaks d and e are due to H8 and H2 protons of ATP. Fig. 5C shows the spectrum of tubulin with no ATP added to the solution. Peak a' is due to the H8 protons of GTP at X-site. Fig. 5D shows the NMR spectrum of ATP recorded in the presence of tubulin. After the addition of ATP, it can be noted that peak a' due to H8 protons of X-site GTP goes down in intensity due to the displacement of GTP at X-site by ATP. Also can be seen the peak a corresponding to the H8 protons of free GTP probably resulting from displacement by ATP. Other resonances e, g and h are the same as explained in Fig. 4. In the case of HI' protons of the sugar of ATP, since there is no appreciable chemical shift between the HI' protons of the X-site GTP and the free nucleotide, it is not possible to demonstrate this effect for these protons. 3.4. Is the X-site or the L-site of the nucleotide nearer to the metal ion at N-site ?

The question concerning the relative location of the X-site (peak a' of Fig. 3) and L-site of GTP (peak c of Fig. 3) and ATP (peak d' and g of Fig. 4) with reference to the metal ion at N-site was sought to be answered qualitatively by replacing Mg 2+ at N-site by the paramagnetic metal ion, Mn z ÷ and studying its effect on the line widths of the relevant resonances. The results of such an experiment are summarised in Table 1. As can be seen from the table, the L-site peak is broader compared to the X-site peak in both diamagnetic and paramagnetic complexes. This suggests that the X-site and L-site are distinct sites and that the GTP at L-site is more immobilized compared to GTP at X-site implying that the environment for GTP at L-site is much more restricted compared to the X-site for its motion. The table also shows the data for ATP in the diamagnetic complex, tubulin-Mg-ATP and in the paramagnetic complex, tubulin-Mn-ATP. Like in the case of GTP, differential broadening of the peaks corresponding to the Xand L-site can be noticed leading to the same conclusion regarding the X- and L-sites. A comparison of the linewidths of the H8 resonances of the ATP and GTP X-site and L-site (Table 1) shows that the effect of the paramagnetic metal ion, Mn 2+, at N-site is greater on the L-site nucleotide compared to that at X-site. This implies that the L-site is nearer to the N-site metal ion compared to the X-site. This effect is much more predominant on the L-site ATP compared to the GTP at L-site. A quantitative estimate of these distances is not possible due to the weak signal from X-site, although attempts are being made in our laboratory in this direction. However, distances of the ATP nuclei at L-site from the paramagnetic metal ion at N-site have been mapped and are described below.

84

S.S. Rai et al. / Biochimica et BiophysicaActa 1292 (1996) 77-88

3.5. Mapping o f the L-site o f A T P from the paramagnetic metal ion at N-site

Fig. 6(A) shows the complete H - N M R spectrum of ATP in the diamagnetic and paramagnetic complexes. Assignment of the resonances corresponding to the various protons have been made earlier [35]. In view of the high concentration of ATP used in the relaxation measurements,

the X-site peak as well as protein resonances are not visible in this spectrum due to condensed scale and the relaxation measurements have been made on the ATP resonances arising from the L-site ATP. As can be seen from the Fig. 6 there is a broadening of the ATP resonances in the diamagnetic complex compared to the free ATP and somewhat larger broadening in the paramagnetic complex. In these experiments, the metal ion is at the

A

...... _ k

j

lh

~

C

Hv

[

I

r,

T T ,

~t~

it

rt

i

r

i,

,-i

,;'TTm'm'r-r~r'rtn'r~r~'~rL~TrrT~r~rrt-, - 10

I

I

'

- ~5

'

,1,

,

,

, ,

,

,

,

,

,

,

,

,

;

,

,

. . . .

I

. . . . . . . . . . . . . . . . . . . . . . . .

-20

Fig. 6. (A) Proton NMR spectrum of ATP at 500 MHz, 25°C in the presence and absence of tubulin-metal ion complex. (a) ATP (5 mM), no tubulin (b) tubulin-Mg-ATP complex, [ATP] = 2 raM, [tubulin]= 228 /xM (c) tubulin-Mn-ATP complex, [tubulin] = 280/xM, [ATP] = 3.7 raM, [Mn2÷ ] = 210 /zM. All solutions were contained in Pipes buffer 10 mM, pH 6.9, in 99.96% D20. (B) Phosphorous NMR spectrum of ATP at 202.4 MHz, 25°C in the presence and absence of tubulin-metal ion complex. (a) ATP (5 mM), no tubulin (b) tubulin-Mg-ATP complex, [tubulin] = 348 /zM, [ATP] = 5 mM (c) tubulin-Mn-ATP complex, [tubulin] = 371 /zM, [ATP] = 5 mM, [Mn2+ = 36 /xM. All solutions were contained in Pipes buffer 50 mM (pH 6.9), D20 is 30%.

85

S.S. Rai et al. / Biochimica et Biophysica Acta 1292 (1996) 77-88

Table 2 Effect of Mn 2+ on IH and 31p relaxation rates of ATP nuclei in tubulin-ATP-Mn complex ATP nuclei

Chemical shtft (ppm)

NMR frequency (MHz)

H8 H2 HI' H3' H4' H5' ct-P fl-P T-P

8.55 8.27 6.16 4.6 4.4 4.3 - 10.8 -22.2 - 8.0

500

202.4

( p T I t') - i ( s - t)

( p T 2p)- 1 ( s - I)

700 _+ 12 128 + 19 200 + 44 1112 + 445 398 + 20 329 + 47 2363 +_ 469 1940 + 154 2147 + 367

AE a (kJ/mol)

1574 504 _ 69

3.4 12.6 10.7

7.4 13.8

Samples for proton NMR measurements were made in 10 mM Pipes buffer (pH 6.9) and for 3] P-NMR in 50 mM Pipes (pH 6.9) at 25 + I°C. or-P, /3-P and y-P are the ol-, /3- and y-pho,;phate groups of ATP. In the sample containing protein and ATP, all the protein has been assumed to be present as tubulin-ATP complex. Errors shown in the ( p T 1p ) - I and ( p T 2 p ) - I values are outer limits arrived at by measurements on a few samples (typically 4 - 5 different samples). Typical concentrations of protein in the samples varied between 160-320 p~M, [ATP] = 5 mM, [Mn 2+ ] = 35-100 /xM. a Activation energies zlE were determined from the temperature dependence of (pT~ p ) - t shown in Fig. 7.

N-site and there is no exchange of this into the solution unless denatured. Also, the fact that there is no leakage of the N-site metal ion into solution was checked by monitoring the water proton resonance line width. ATP at L-site, however, exchanges into the bulk as mentioned earlier. Fig. 6 (B) shows the phosphorous NMR spectrum showing the a-, /3- and y-phosphorous resonances in the diamagnetic and paramagnetic complexes. Relaxation measurements have been made for the a-, /3- and y-phosphorous nuclei of ATP in the diamagnetic and paramagnetic complexes of tubulin. Broadening of the phosphorous resonances of ATP in the diamagnetic and paramagnetic complexes can be noted due to the binding of ATP on tubulin. There is an enhancemem of this broadening in the paramagnetic complexes. Also some structure can be noticed in b and c more clearly in the a- and y-peaks probably due to exchange of the ATP between the bulk and the tubulinmetal-ATP complex on a scale which is intermediate compared to the NMR time scale, resulting in two barely resolved sets of c~-, /3- ~Lnd y-resonances of phosphorous nuclei. Also, an upfield shift (0.1 ppm for c~, 0.5 ppm for /3 and 1.2 ppm for y-phosphorous nuclei) has been observed for the a-, /3- and y-phosphorous nuclei of tubulin-bound ATP showing that the phosphate groups of ATP interact directly with the protein. Similar results were observed for the tubulin-bound GTP (not shown). Spin-lattice relaxation rates (T~- l ) of the protons of the base and sugar of ATP have been measured in the presence and absence of the protein and in the diamagnetic and paramagnetic complexes of tubulin-ATP-metal ion. The metal ion, Mn 2+, at the N-site is the paramagnetic reference point for distance measurements. The only ligand of interest which exchanges into the bulk is ATP apart from water molecules. Because of the high concentrations of ATP that are used in the relaxation experiments, (i.e., [Protein] << [ATP] = 5 mM >> K d for tubulin-ATP complex), protein will be saturated with ATP even if one takes K d = 200 /xM for ATP binding to tubulin [14]. However,

part of the protein molecules will be in the form of diamagnetic complex and part will be in the form of paramagnetic complex. Normalised paramagnetic contributions to the relaxation rates (pTlp) -1 were evaluated by estimating the metal ion concentrations (Mg 2÷ and Mn 2+) in the protein solution and subtracting the diamagnetic contributions from the total measured relaxation rates. The normalized paramagnetic contribution to the relaxation rates (pT~ p ) - l and (pT 2p)-I so obtained from the T 1 and line width measurements for the different protons and phosphorous nuclei of ATP are summarised in Table 2. In order to ascertain whether the (pT~p) -1 values so determined can be used to evaluate the distances of the ligand nuclei from the metal ion, the contribution of the residence time, %4, of the ligand in the complex has to be evaluated. A temperature dependence study of the ( pT 1p)- l for the tubulin-Mn-ATP complex was carried out in the range of 10-30°C for H2, H8 and HI' protons at 500 MHz I0

0

i-~ 6 ,m, C

4

2

I

3.2

3.:3.

I

3.4

I

3.5

3.6

IO~T (K -I) Fig. 7. Temperature dependence of the normalized paramagnetic contribution to the relaxation rates of ~H and 31p of tubulin-bound ATP. Data for H2 (O), HI' (A), H8 (C)), ct-P ( D ) and 7-P ( A ) are shown. Activation energies (AE) calculated from these plots are shown in Table 2. Sample conditions are same as in Table 1.

S.S. Rai et al. / Biochimica et Biophysica Acta 1292 (1996) 77-88

86

Table 3 Distances between the Mn 2+ at N-site and the nuclei of bound ATP (at L-site) on tubulin ATP nuclei

Chemical shift (ppm)

N M R frequency (MHz)

f(rc) a

H8 H2 HI' H3' H4' H5' a-P fl-P y-P

8.55 8.27 6.16 4.6 4.4 4.3 - 10.80 - 22.2 - 8.0

500

0.33 - 0 . 4 6

(ns)

Distance ( r ) b (nm)

=

202.4

0.66-1.18

= = =

0.68-0.72 0.92-1.03 0.83-1.0 0.55-0.75 0.76-0.82 0.78-0.86 0.34-0.47 0.43-0.47 0.42-0.46

a The range of f(~'c) values reported is based on the range determined for zc on the basis of the experimental uncertainties for ( p T I p)- l and ( p T 2 p)- l values. b The distances reported have been calculated after subtracting the contribution of ( r M ) - i to the normalized paramagnetic contributions to the relaxation rates. (See text for details.)

and a- and r-phosphorous of ATP at 202.4 MHz. The results are shown in Fig. 7 in the form of Arrhenius plot and the activation energies evaluated for the protons and phosphorous nuclei are summarised in Table 2. The activation energies for the protons are in the range of 3.4-10.7 k J / m o l and for the a- and y-phosphorous groups are in the range of 7.4-13.8 kJ/mol. These energies are in the range expected for Tf~ associated process [36,37] and are significantly lower than the value expected for the exchange-dominated processes [37] ( A E for the rM-dominated processes are in the range 6-35 kJ/mol). The positive slope of the plot of ln(pTjp) -1 vs. 103/T shown in Fig. 7, also clearly indicates that ( p T l p)-~ is dominated by (T~-~) associated processes and not by 1/'c M, the exchange rate of the ligand, ATP, from tubulin-metal ion-ATP complex. This implies that the (pTxp)-l values can be used to elicit structural information in using Eqs. (1) and (2). Nevertheless, a small contribution to the normalised relaxation rates (pT~p)-1 from ~.~1 in addition to T~-1 cannot be ruled out. This contribution has been evaluated using the approach of Lanir and Navon [32,33]. ~'M, the residence time of the ligand in the tubulin-Mn-ATP complex was evaluated by using Eqs. (5) and (4) and the data are shown in Table 3 for the protons. We obtain ~'M to be in the range 0.34-0.41 ms for the tubulin-ATP-Mn 2÷ complex, r c calculated using Eq. (5) and after taking the contribution of r ~ 1 to the measured (pTlp)-1 values into account, turns out to be in the range 0.62-0.71 ns. ~'c also has been evaluated from the ratio of ( p T l p ) -~ and (pT2p)- 1 determined for H8 and H2 protons and assuming fast exchange conditions as shown in Fig. 7. The range of rc values is in the range 0.45-0.65 ns. The f(%) values calculated using this range of r c values are also shown in Table 3 for 1H- and 31p_nuclei of ATP in the protein-MnATP complex. The31zc value for the dipolar interaction between Mn 2+ and P nucleus of the phosphate groups is taken to be the same as that of the 1H-Mn(II) interaction. The distances calculated between Mn 2+ at the N-site and

the various protons and phosphorous nuclei of ATP in this complex are summarised in Table 3. The range of distances is obtained by taking into account experimental uncertainties and the contribution of ~-~ to the measured values of (pTlp) -I . It is to be noted, however, that the contribution of ~-~l is quite significant to the normalised paramagnetic relaxation rates, (pTlp) -~ for HY protons and a-, r - and y-phosphorous nuclei. As a result, the distances reported for these nuclei should be considered as approximate and are an upper limit for the distances of these nuclei. The base protons are about 0.71-1.0 nm and the sugar protons are about 0.78-0.97 nm away from the N-site metal ion. But the a-, /3- and y-phosphorous groups of ATP are nearer and are within 0.45-0.55 nm distance from the N-site metal ion. It is to be noted that a more accurate measurement of r, involves a measurement of frequency dependence of the (pTlp) -1 values [20-23]. However, the zc values determined by us for the ATP nuclei in tubulin-Mn-ATP complex are in the same range as reported for other enzymes [22,23,30,31] and are of the same order as those for the colchicine-tubulin-Mn-GTP(yF) complex [38]. However, distances can be measured accurately (within _ 10%) due to the ~th power dependence of r on these parameters, even if large errors are made in the determination of parameters such as %. In this work, only a range has been reported for the distances.

4. Discussion

We have tried to address the question regarding the nature of interaction of ATP and GTP with tubulin and the relative location of the binding sites of ATP and GTP on this protein using NMR spectroscopy. It is quite wellestablished that guanine nucleotide has two binding sites, namely, the N-site and E-site on this protein [1]. However, it is not known whether ATP binds at the GTP E-site or it

S.S. Rai et a l . / Biochimica et Biophysica Acta 1292 (1996) 77-88

binds at a site distinct from the E-site of GTP [17] or at the same site as GTP E-site. Zabrecky and Cole [13,14] suggested that ATP interacts with tubulin at a site which is distinct from that of the N-site or E-site where GTP binds. On the other hand, the suggestion that ATP interacts weakly at the E-site was given by Duanmu et al. [17] as well as by other workers [39,40]. In this work, we have identified an NMR resonance due to the H8 protons of GTP occupying a site we have labelled as X-site. Similar NMR peak due to GTP at N-site could not be detected possibly due to extreme broadening experienced by GTP located at N-site. In addition, an additional peak due to H8 protons of GTP appears which has been identified to be due to GTP in fast (or intermediate) chemical exchange between the GTP bound at this site and free nucleotide in solution. This site has been labelled as L-site or low-affinity site in order to distinguish it from the X-site since this site is occupied only after the addition of the nucleotide in the concentration range > 0.5 mM. It must be noted, however, that our experiments do not provide a quantitative comparison between the relative affinities of the nucleotide binding to these two sites. Our suggestion for the existence of another site for GTP is in agreement with that by earlier workers [13,14,41,42] who suggested the existence of a third nucleotide-binding site on tubulin and that this low-affinity site may accommodate either ATP or GTP and may exert an inhibiting influence on the microtubule assembly [13,41,42]. Similar NMR experiments on the interaction of ATP with tubulin have shown two separate NMR sets of peaks for the H8 and H2 protons of ATP, and these peaks have been identified as due to X-site and L-site of ATP on tubulin (Fig. 4). The question whether the X-site of ATP and GTP is the same or different has been addressed by doing competition NMR experiment in which it was found that GTP bound at X-site could be displaced by ATP. The affinity of GTP is in the nM range in the presence of Mg 2÷ ion [18] while it is in the order of /zM in the absence of externally added Mg 2+. For our NMR experiments, the free as well as the metal ion at X-site has been removed by chelexing. Hence GTP and ATP affinities aJ~e of similar order in the absence of the metal ion [19]. So ATP can replace GTP even though they are of similar concentrations. These results have been interpreted to imply that the X-site of ATP and GTP is the same. This is a~lso supported by the fact that the line width of the H8 protons of the nucleotide binding at X-site is approximately the same (Table 1). From the differential broadening of the H8 resonances of the nucleotide at X- and L-sites that has been observed for both ATP and GTP in tubulin-Mn-nucleotide complexes it appears that the two sites (the X- and L- sites) are distinct. Also, the paramagnetic effect of the N-site metal ion, Mn(II), on the L-site nucleotide is larger than that on X-site nucleotide (Table 1) suggesting the proximity of the L-site to the N-site metal ion in comparison to the X-site.

87

The question arises whether the L-site of ATP and GTP is the same or different. According to our results, the y-phosphate of ATP is located within 0.42-0.46 nm from the N-site metal ion, (Table 3) thus considerably closer than that of GTP from the N-site metal ion [38]. These results seem to imply that the L-sites of the two nucleotides are distinct sites and are spatially separated from each other. However, a definite comparison cannot be made at this stage in the absence of our GTP distance measurements which are still in progress. This is, however, supported by the larger effect of the Mn(II) ion at N-site on the line width of the H8 resonances of ATP at L-site compared to that on the H8 resonances of GTP at L-site (Table 1), suggesting that the L-site of ATP is nearer than that of GTP from the N-site. Considering the distances reported for both protons and phosphate groups in the tubulin-Mn(II)-ATP complex, it is clear that the nucleotides do not possibly bind in the first coordination sphere of the N-site metal ion as the distances between the metal and the phosphate are typically in the range 0.28-0.3 nm [43] for first coordination sphere complexes. The distances that have been obtained are consistent with the fact that the N-site is non-exchangeable and non-catalytic [44] and is not necessary that it should form first coordination complexwith the E-site or L-site GTP or ATP. It appears that the X-site detected by us in NMR experiments is the same as the putative E-site. The observation of a signal for GTP bound to tubulin in the absence of the exogenous nucleotide and the fact that the externally added GTP binds at concentrations as low as 0.3 mM at this site suggests that this is a high-affinity site and possibly the E-site. Further evidence that this X-site could be the same as E-site comes from the fact that the peaks due to X-site were not observed in alkaline phosphatase treated tubulin. Also, the nucleotide content estimation of the gel-filtered tubulin shows that [Protein]/[nucleotide] is of the order 1:1.5-1.6 suggesting that the X-site detected by us could be E-site. The line widths of the X-site are narrow compared to the line widths of the L-site nucleotide resonances or the line widths of the protein resonances. This implies a high degree of flexibility for the E-site GTP if X-site is the E-site which is not consistent with the fact that GTP binds tightly at the E-site. However, as mentioned earlier, the peak a' becomes narrower as the nucleotide concentration is increased due to exchange between the X-site nucleotide and the L-site nucleotide and also exchange between the L-site nucleotide and the free nucleotide. The observed competition between ATP and GTP at the X-site (Fig. 5) is consistent with our finding that the E-site of ATP and GTP is the same from our fluorescence experiments on the binding of Ant-ATP, a fluorescent analogue of ATP, on tubulin [19]. However, the fact that the peak a' increases in intensity much more than what is expected for GTP binding at E-site needs to be explained if X-site is the E-site. It appears from our

88

S.S. Rai et al. / Biochimica et Biophysica Acta 1292 (1996) 77-88

NMR experiments that there is an interaction between the X-site and L-site. It is conceivable that binding of GTP or ATP at the L-site perturbs the binding of the nucleotide at the E-site thus altering the affinity of the nucleotide to the E-site. It is, however, not possible at this stage to determine the spatial separation between the X-site and L-site. It appears, therefore, that the X-site detected by us in the NMR experiments is the same as the E-site. Finally, the role of the L-site reported by us in tubulin polymerisation is not clear at this stage. Under the conditions which are normally used for polymerisation studies in vitro in the presence of GTP or ATP, the concentrations of the nucleotide are such that the nucleotides bind at both the E- and L-sites. It appears from our results, therefore, that the nucleotide bound at the L-site may also have a physiological role.

Acknowledgements The authors gratefully acknowledge helpful advice provided by Dr. B. Bhattacharyya of the Department of Biochemistry, Bose Institute, Calcutta, regarding the purification and isolation of the protein. The authors also thank Mr. H.P. Maity of this group for help in some experiments. The authors also acknowledge the help provided by the staff of the 500 MHz FT-NMR National Facility at Tata Institute of Fundamental Research, Bombay, India in the NMR experiments and the staff of Regional Sophisticated Instrumentation Centre, I.I.T., Powai, Bombay, India for the metal ion estimation in our samples.

References [1] Weisenberg, R.C., Borisy, G.G. and Taylor, E. (1968) Biochemistry 7, 4466-4479. [2] Weisenberg, R.C., Deery, W.J. and Dickinson, P.J. (1976) Biochemistry 15, 4248-4254. [3] Kobayashi, T. (1975) J. Biochem. (Tokyo) 77, 1193-1197. [4] Correia, J.J., Baty, L.T. and Williams, R.C.J. (1987) J. Biol. Chem. 262, 17278-17284. [5] Olmsted, J.B. and Borisy, G.G. (1975) Biochemistry 14, 2996-3005. [6] Lee, J.C. and Timasheff, S.N. (1977) Biochemistry 16, 1754-1764. [7] Himes, R.H., Burton, P.R. and Gaito, J.M. (1977) J. Biol. Chem. 252, 6222-6228. [8] Buttlaire, D.H., Czuba, B.A., Stevens, T.H., Lee, Y.C. and Himes, R.H. (1980) J. Biol. Chern 255, 2164-2168. [9] Correia, J.J., Beth, A.H. and Williams, R.C.J. (1988) J. Biol. Chem. 263, 10681-10686. [I0] Osei, A.A., Everett, G.W. and Himes, R.H. (1990) FEBS Lett. 276, 85-87.

[11] David-Pfeuty, T., Erickson, H.P. and Pantaloni, D. (1977) Proc. Natl. Acad. Sci. USA 74, 5372-5376. [12] Zabrecky, J.R. and Cole, R.D. (1980) J. Biol. Chem. 255, 1198111985. [13] Zabrecky, J.R. and Cole, R.D. (1982) J. Biol. Chem. 257, 46334638. [14] Zabrecky, J.R. and Cole, R.D. (1982) Nature 296, 775-776. [15] Islam, K. and Burns, R.G. (1984) FEBS Lett. 178, 264-270. [16] Zabrecky, J.R. and Cole, R.D. (1983) Arch. Biochem. Biophys. 225, 475-481. [17] Duanmu, C., Lin, C.M. and Hamel, E. (1986) Biochim. Biophys. Acta 881, 113-123. [18] Farr, G.W., Yaffe, M.B. and Sternlicht, H. (1990) Proc. Natl. Acad. Sci. USA 87, 5041-5045. [19] Rai, S.S. and Kasturi, S.R. (1993) Arch. Biochem. Biophys. 306, 133-138. [20] Mildvan, A.S., Leigh, J.S. and Cohn, M. (1967) Biochemistry 6, 1805-1818. [21] Mildvan, A.S. and Cohn, M. (1970) Adv. Enzymol. Relat. Areas Mol. Biol. 33, 1-70. [22] Jarori, G.K., Ray, B.D. and Nageswara Rao, B.D. (1985) Biochemistry 24, 3487-3494. [23] Jarori, G.K., Iyer, S.B., Kasturi, S.R. and Kenkare, U.W. (1990) Eur. J. Biochem. 188, 9-14. [24] Shelanski, M.L., Gaskin, F. and Cantor, C.R. (1973) Proc. Natl. Acad. Sci. USA 70, 765-768. [25] Lowry, O.H., Rosebrough, N.J., Farr, A.L. and Randall, R.J. (1951) J. Biol. Chem. 193, 265-275. [26] Solomon, I. (1955) Phys. Rev. 99, 559-565. [27] Solomon, I. and Bloembergen, N. (1956) J. Chem. Phys. 25, 261266. [28] Bloembergen, N. (1957a) J. Chem. Phys. 27, 572-579. [29] Bloembergen, N. (1957b) J. Chem. Phys. 27, 595-596. [30] Mildvan, A.S. and Gupta, R.K. (1978) Methods Enzymol. 49, 322-359. [31] Mildvan, A.S., Granot, J., Smith, G.M. and Liebman, M.N. (1980) Adv. Inorg. Biochem. 2, 211-236. [32] Navon, G. and Valensin, G. (1987) Metal Ions in Biological Systems (Helmut Sigel, ed.), Vol. 21, pp. 1-45, Marcel Dekker, New York. [33] Lauir, A. and Navon, G. (1972)Biochemistry 11, 3536-3544. [34] Purich, D.L. and MacNeal, R.K. (1978) FEBS Lett. 96, 83-86. [35] Tran-Dinh, S. and Chachaty, C. (1977) Biochim. Biophys. Acta 500, 405 -418. [36] Dwek, R.A. (1973) NMR in Biochemistry, Ch 9-10, Clarendon Press, Oxford. [37] Burton, D.R, Forsen, S., Karlstrom, G. and Dwek, R.A. (1979) Prog. Magn. Reson. Spectrosc. 13, 1-45. [38] Monasterio, O. (1987) Biochemistry 26, 6099-6106. [39] Mejillano, M.R., Barton, J.S., Nath, J.P. and Himes, R.H. (1990) Biochemistry 29, 1208-1216. [40] Seckler, R., Wu, G.M. and Timasheff, S.N. (1990) J. Biol. Chem. 25, 7655-7661. [41] Jameson, L. and Caplow, M. (1980) J. Biol. Chem. 255, 2284-2292. [42] Maccioni, R.B. and Seeds, N.W. (1982) J. Biol. Chem. 257, 33343338. [43] Mildvan, A.S. and Grisham, C.M. (1974) Struct. Bonding (Berlin) 20, 1-21. [44] Spiegelman, B.M., Penningroth, S.M. and Kirschner, M.W. (1977) Cell 12, 587-600.