- Email: [email protected]
The state of water associated with the phosphoenzyme of the Ca-ATPase of sarcoplasmic reticulum
473
Bioelectrochemistry and Bioenergetics, 17 (1987) 473-487 A section of J. Electroanal. Chem., and constituting Vol. 231 (1987) Elsevier Sequoia S.A., Lausanne - Printed in The Netherlands
983 - THE STATE OF WATER ASSOCIATED WITH THE PHOSPHOENZYME OF THE Ca-ATPase OF SARCOPLASMIC RETICULUM A SPIN PROBE STUDY
PHILIPPA
M. WIGGINS
Department
of Medicine,
GRAHAM
A. BOWMAKER
Department
of Chemistry
(Manuscript
University of Auckland
School of Medicine, Private Bag, Auckland
University of Auckland,
received August
Private Bag, Auckland
(New Zealand)
(New Zealand)
20th 1986)
SUMMARY Electron spin resonance spectra of nitroxide free radicals in the aqueous phase associated with suspensions of sarcoplasmic reticulum vesicles have been measured. While the Ca-ATPase was cycling there was a loss of signal which returned to normal as ATP was consumed. The fractional loss of signal increased with protein concentration. but was independent of concentration of spin probe, suggesting that it was unlikely to be due to binding of the probe to one of the conformations assumed by the enzyme during a catalytic cycle. The results are consistent with an increase in viscosity of water associated with the phosphoenzyme intermediate of the ATPase, similar to the increase of viscosity of water which has been observed in hydrophobic pores of synthetic polymers. It is suggested that probes in this relatively viscous microaqueous phase tumble more slowly than probes in the rest of the water, so that with formation of the phosphoenzyme a fraction of the signal broadens and loses peak height.
INTRODUCTION
The Ca-ATPase of sarcoplasmic reticulum is one example of many enzymes which have their active sites buried in hydrophobic pockets. DuPont and Pourgeois [l] and Nakamoto and Inesi [2] using the fluorescence of TNP nucleotides found that an environment of low polarity before phosphorylation of the enzyme decreased drastically in polarity and increased in viscosity following phosphorylation. They suggested that the ATP binding site is in a hydrophobic cleft which closes upon phosphorylation. Because water in such a small hydrophobic domain may be structurally perturbed, we have investigated the state and properties of water in the 0302-4598/87/$03.50
0 1987 Elsevier Sequoia
S.A.
Fig. 1. Diagrammatic illustrations of three different structures of water. (a) The continuous random defective network of a normal liquid phase. (b) Stretched water, contained in hydrophobic pores, and in equilibrium with a normal aqueous phase or with solutions of Na+, Li+, Ca*+ or Mg*+ salts. (c) Vapour-like water, contained in small hydrophobic pores and in equilibrium with a solution of KC1 or CsCl. For discussion see text.
small hydrophobic pores of cellulose acetate membranes, in order to determine whether or not it is likely that changes in solvent structure contribute to the coupling mechanism of the Ca-ATPase [3]. Water in small hydrophobic pores of regular copolyoxamide and cellulose acetate membranes was found to exist in two states, both of which were profoundly different from the normal liquid state. They are illustrated schematically in Fig. 1, and contrasted with the continuous random defective network of normal liquid water (Fig. la). In the absence of solutes the water was stretched (Fig. lb). The infrared spectrum was characteristic of water monomers adsorbed to the hydrophobic surface, and of large strongly hydrogen-bonded clusters. Absorbance in the region of weaker hydrogen bonds was relatively slight. When the membranes were equilibrated with chlorides of Na+, Li’, Ca2+ or Mg’+ the strongly-bonded clusters increased still more at the expense of weakly-bonded clusters. Partition coefficients of the salts were less than one and decreased in the order Na+ > Li+ z=- Ca2+ > When the membranes were equilibrated Mg2+> and with increasing concentration. with KC1 or CsCl at concentrations of 1 mol rnp3 or higher, water in the pores assumed a vapour-like state illustrated in Fig. lc. The infrared spectrum showed no absorbance in the regions of strong hydrogen-bonding but only in the regions where monomers and dimers absorb. This water was a good solvent for salts of the small cations. KC1 and CsCl were accumulated into the stretched water with partition coefficients as high as 30, provided that the external concentration was low enough (< 0.2 mol mp3). It was suggested that KC1 prevents stretching of the pore water by accumulating into it, and generating an osmotic pressure gradient which oppose stretching.
415
The perturbation of water structure in small hydrophobic pores illustrated in Fig. 1, was found to increase as the size of the domain decreased. The hydrophobic cleft of the Ca-ATPase is likely to be smaller than the cellulose acetate pores of diameter approximately 3 nm. Therefore water in that cleft must be similarly perturbed. In the presence physiological concentrations of KC1 the cleft should contain water in the vapour-like state. When, however, the cleft closes upon the phosphorylation, some solution is squeezed out, and the remaining water, which now occupies a much smaller hydrophobic domain, will stretch out of the cleft, and assume the kind of structure illustrated in Fig. lb. This will persist until KC1 diffuses in and water is forced to revert to the vapour state. The self-diffusion coefficient of water in cellulose acetate membranes was found to be of the order of lo-’ cm2 s-l [4] which is much lower than that of liquid water (2.2 x lop5 cm2 s-‘) [5], suggesting that, as would be expected from its highly hydrogen-bonded structure, stretched water is very viscous. This high viscosity will delay influx of KC1 and allow the stretched state to persist for an appreciable time. A mechanism for coupling of ATP hydrolysis to cation transport by means of transient changes in water structure has already been outlined [6-81, and will be described in more detail elsewhere. This paper describes experiments which have been designed to test the hypothesis that the phosphoenzyme intermediate of the Ca-ATPase of sarcoplasmic reticulum contains stretched water. The phosphoenzyme was formed from ATP in the absence could be formed of Mg2+, so that when ATP was used up no more phosphoenzyme from inorganic phosphate (Pi). The electron spin resonance (e.s.r.) spectra of nitroxide free radicals give information about the microenvironment of the spin probe. Those probes relatively immobilised in the cleft when it closes and the water stretches, should resonate with a broader signal than probes in the rest of the water of the system. Under the conditions of the experiments reported here line broadening was most sensitively detected as a loss in peak height. It is shown that there was a loss of signal height from a spin probe in solution in a suspension of sarcoplasmic reticulum vesicles when the phosphoenzyme was formed; the loss in peak height returned as the concentration of phosphoenzyme declined with consumption of ATP. As expected the lost signal height amounted to only l-2% of the total. MATERIALS
AND
METHODS
Sarcoplasmic reticulum vesicles were prepared from rabbit back and leg muscles as previously described [9], and finally suspended at a protein concentration of approximately 100 mg cme3 in a standard buffer solution containing KC1 100 mol rne3, Tris (hydroxymethyl) aminomethane (Tris) maleate (5 mol m-3, pH 7). Stock solutions of ATP (1 mol dmp3), CaCl, (0.5 mol dmp3), ethyleneglycol-bis-(Paminoethylether) N,N’-tetraacetic acid (EGTA), (0.5 mol dm-3) and 4-hydroxy2,2,6,6-tetramethylpiperidine-1-oxyl (TEMPOL) (5 X lo-’ mol dme3) were made up in the same buffer solution with pH adjusted to 7: 2,2,6,6_tetramethylpiperidine1-0~~1 (TEMPO) (lop2 mol dmw3) and the ionophore A23187 (0.5 mg cm-‘),
generously supplied by Lilley Laboratories, were made up in ethanol. Phosphate production was measured as previously described [9]. Protein was determined using a biuret method. For experiments in which the reaction was to be controlled by the presence or absence of Ca*+, EGTA (0.5 mol dme3) and A23187 (5 X lop4 mg extraction. cmp3) were added to the 0.6 mol dm-3 KC1 used for the overnight Residual Ca2+ was then. negligibly low, and the contaminating Ca*+ added with ATP was completely complexed by an additional 1 mol rnp3 EGTA added with the ATP. e.s.r. spectra were recorded at room temperature on a Varian E-4 spectrometer operating at about 9.1 GHz with a modulation frequency of 100 kHz. Samples were contained in the B248 aqueous solution sample cell. EXPERIMENTAL
PROCEDURE
The concentrated suspensions used were difficult to pipette accurately and to mix efficiently. All additions had therefore to be made before the sample was transferred to the flat aqueous cell of the spectrometer. It was found that the signals from both TEMPO and TEMPOL in the presence of vesicles decayed with time (see Fig. 4 below), so that comparisons of absolute amplitudes from experiment to experiment could not be made. The time-courses of the decay in signal were therefore followed in paired experiments in which the upfield resonance of a spin probe was measured: (i) during and after ATPase activity, and (ii) in the absence of ATPase activity, other conditions being the same. In a typical pair of experiments 1 g aliquots of vesicles at a protein concentration of 100 mg cmp3 and containing 3 pmole of Ca*+ and 10 pg of ionophore A23187 were weighed into each of two small plastic tubes. At time zero 20 mm3 ATP and 10 mm3 TEMPOL were added to the first tube, the contents mixed thoroughly with a Pasteur pipette, taken up in a syringe and transferred to the flat aqueous cell which was then adjusted in the cavity of the spectrometer. Spectral measurements were made at minute intervals from 4 to 10 min, and at 2 to 5 min intervals thereafter. To the second tube 10 mm3 TEMPOL and 20 mm3 buffer were added at zero time, and spectral readings taken after the same time intervals. RESULTS
It was first established that neither TEMPO nor TEMPOL in the concentration used impaired ATPase activity, which was 1.2 pmol P, mgg’ protein min-’ in the absence of either spin probe, when the probe was added at the same time as the ATP which started the reaction, and when the vesicles were preincubated with the probe for 12 h before activation. In concentrated suspensions of vesicles both nitroxide radicals (Fig. 2) gave characteristically sharp e.s.r. spectra with three hyperfine splittings (Fig. 3). TEMPO, which is more hydrophobic than TEMPOL, distributed between lipid and aqueous phases, giving a further splitting of the high-field resonance [9], due mainly to the
471
G< N-O HO
N-O
TEMPOL
TEMPO
Fig. 2. Structural
formulae
of TEMPO
and TEMPOL.
difference of dielectric constant which is 78 in water and 6 in lipids. The signal height was followed by measuring h in Figs. 3a and 3b. Fig. 3c shows the spectrum of TEMPOL in a vesicle suspension frozen at - 50 o C, illustrating the broadening of the signal and loss of amplitude which resulted from immobilisation of the probe in random orientations. Loss of amplitude from the upfield resonance was especially marked. Spontaneous
decay of the resonance
signal
In the absence of ATPase activity, the signal gave characteristic decay curves illustrated in Fig. 4. The absolute values of h at t = 0 are not comparable because it was impossible to prevent formation of air bubbles in the viscous solution. The decay was not first order, since log h uersus time curves were not linear. The rate of decay and its curvature increased with protein concentration and changed with composition of the supporting medium. The signal from TEMPO decayed more rapidly than that from TEMPOL and with greater curvature. Decay of the signals
_ ”
20 gauss H
/
‘J
( ,’ / / ,
Fig. 3. First derivative e.s.r. spectra of nitroxide free radicals in suspensions of sarcoplasmic reticulum vesicles at a protein concentration of 101 mg cmm3. (a) TEMPOL (10m5 mol dmm3) at room temperature; its concentration in the aqueous phase is assumed to be proportional to h; (b) TEMPO (10V4 mol dme3) at room temperature; its concentration in the aqueous phase is assumed to be proportional to h; (c) TEMPOL in suspensions frozen at - 50 DC.
478
-m---=-=-m -=--m-m-m
23A_-.
--L-L_. -A-A-A
_ ***-o--.-
.-.-
c 0" 2.1-J
I
0
I
I
20
t
.-.-_.
-A-A-A -•-•--0-0
IllIn) I
I
40
1
1
60
Fig. 4. Spontaneous decay of the resonances of TEMPO and of TEMPOL in the aqueous phase of concentrated suspensions of sarcoplasmic reticulum vesicles. (m, A, 0) TEMPOL; protein concentration: (0) 80 (m) 80 mg cme3; (A) 120 mg cmm3; (0) 105 mg cm -3. (0, 0, A) TEMPO; protein concentration: mg cmm3; (0) 120 mg cm-3; (a) 150 mg cme3.
from TEMPO and TEMPOL in buffer solution was extremely slow and first order. The higher rate of decay in the presence of vesicles was not due to chemical reaction between the nitroxide and a membrane component, because, when a suspension containing TEMPO was left in the dark for two days, the lost signal could be restored by extraction with an equal volume of ethanol. It seemed probable that the decay in signal was due to a relatively slow diffusion of the probe into more rigid lipid components of the membrane, with immobilisation and loss of signal. This mechanism accounts for the faster decay of TEMPO which is more soluble than TEMPOL in lipids, and its increase in decay rate with protein concentration. Whatever its mechanism the spontaneous loss of signal necessitated readings at closely-spaced intervals of time, so that the width could not be expanded sufficiently to allow accurate measurements. Therefore loss of peak height was measured rather than increase in line width. The effect of A TPase activity Figure 5 shows examples of decay CUNeS for TEMPO in concentrated suspensions of vesicles to which ATP was added at zero time. When a low concentration of ATP was added to a highly concentrated suspension the early points of the decay curve appear to lie below any reasonable extrapolation of the rest of the curve to zero time. The marked curvature of the plots precludes more than the qualitative conclusion that while there is still an excess of ATP present and one might expect the concentration of phosphoenzyme to be high, some signal height appears to have
479
Fig. 5. The effect of ATPase activity upon the resonance of TEMPO from the aqueous phase of a suspension of vesicles. ATP and TEMPO were added at zero time to a suspension of vesicles already containing Ca*+ and ionophore A23187, as described in Methods. (0) Protein, 150 mg cm-3; ATP 20 mol md3; (m) protein, 100 mg cmm3; ATP, 200 mol m-3; (A) protein, 100 mg cmm3; ATP, 20 mol rne3; (A) protein, 160 mg cm-3; ATP, 30 mol me3.
been lost from the aqueous phase. At a lower concentration of protein (100 mg cm-3) the initial points are not obviously below the extrapolation of the rest of the curve; while in the presence of excess ATP the initial points appear to be normal but the loss of signal at later times appears to be slowed down. Before an estimate of the magnitude of the lost signal could be made, conditions under which the decay curve could be reliably extrapolated back to zero time were explored. The use of difference plots Even when plotted on a logarithmic scale the decay curves had significant curvature, the magnitude of which was reflected in the value of the coefficient c in the equation log h=a+bt+ct*
(1) Curves of this kind were fitted to the data by parabolic regression in order to find conditions under which coefficient c was a minimum and therefore extrapolation to zero time most reliable. Table 1 shows values of the three coefficients of equation 1 for decay curves of either TEMPO or TEMPOL in ATP-free suspensions of vesicles. In all cases the coefficient c was small and positive; it increased with protein concentration and was always higher for TEMPO than for TEMPOL. When one such curve is subtracted from another, the difference plot A log h = (aI -al)
+ (b, - b2)t + (cl - cz)t2
is more nearly linear than either of the parent curves, provided two coefficients of the quadratic terms are of similar magnitude.
(2) that ci and c2, the
480 TABLE
I
Coefficients of the equation log h = a + bt + ct2 fitted by parabolic regression to decay curves for the upfield resonance from the aqueous phase of either TEMPO or TEMPOL in ATP-free suspensions of vesicles protein
Spin probe
(mgcm TEMPO TEMPO TEMPO TEMPOL TEMPOL TEMPOL
-3
a
-bx103
CX106
1.84 2.08 1.77 2.28 2.17 2.25
3.55 3.8 4.3 1.05 1.69 1.94
18.7 21.4 25.6 2.33 5.46 6.59
1
80 110 120 80 105 120
Figure 6 shows representative difference plots for pairs of spontaneous decay curves. All difference plots for TEMPO had significant residual curvature, so that linear extrapolation of spectral readings obtained after 20 min would lead to considerable error. Similarly difference plots for TEMPOL in a pair of experiments with different protein concentrations. and, therefore, different c-values, were significantly curved. When, however, the parent curves had similar and low values of c, the difference curve was linear within experimental error. Therefore when one of a
010 --•---•_.___ •~~~~~~~---~
038
.-.-
i 004
c 0”
002
i
4 ----iT&A-*
._~_._.__._._._._._
0 221 --~~__._.-m-.-.-._C.-._
0200
I 20
(mln) 40
60
Fig. 6. Representative difference plots for spontaneous decay of resonance. In a pair of experiments TEMPO or TEMPOL was added at zero time to a suspension of vesicles in which the ATPase was inhibited, and the two spectral parameters h, and h, measured at time intervals. A log h = log h, log h,; c is the coefficient of the quadratic term A log h = a + br + ct2; cl and c2 the corresponding coefficients of the parent curves; r is the correlation coefficient for linear regression. (0) TEMPOL, c=2x10-6 cr = 7.2 x 10m6, c2 = 5.2 x 10e6; r = 0.97. Protein concentrations were both 105 mg crne3. (A) TEMPOL; c = - 4.5 X lo@, c, = 2.8 x lo@, c2 = 7.2 X 10-6, r = 0.83. Protein concentrations were 80 and 105 mg cme3. (W) TEMPO; c = 4.1 x 10m6, c, = 2.9 X lo-‘, c2 = 4.2 X lo-’ r = 0.87. Protein concentrations were both 108 mg cm- 3.
481
I 0
I
20
I
40
60
Fig. 7. Difference plots (experimental-control) for TEMPO in equal aliquots of a single preparation of vesicles; protein concentration 120 mg cm-3. (a) The parent curves contained 20 mol mm3 ATP and no ATP; (b) the parent curves contained 23 mol m -3 ATP and no ATP; (c) the parent curves contained 25 mol mm3 ATP and no ATP.
pair of curves is obtained from a suspension to which ATP is added at zero time, and the second curve is from a suspension of the same concentration without ATP, any departure from linearity in the difference plot can be assumed to reflect ATPase activity. The residual curvature in the difference plots for all experiments using TEMPO prevented this method of quantitative analysis. Figure 7 shows three difference plots for TEMPO in a concentrated suspension of vesicles to which 20, 23 and 25 nmol ATP was added at zero time. The common control was an experiment in which buffer was added instead of ATP. The c-value of the decay curve without ATP (4.2 X 10e5) increased in these difference curves to 2 X 10m4, 9 X 10e5 and 1 x lop4 respectively. The increase in curvature is again consistent with a loss of signal height associated with the initial high concentration of phosphoprotein, and its return as ATP was used up. Since, however, it is very probable that the difference between the two spontaneous components of the decay curves was not linear, a reliable extrapolation cannot be made. Spectral measurements of TEMPOL were analysed using difference plots of paired experiments from a single preparation, so that the concentration of protein, which largely determined the curvature of the signal decay, was constant. Under these conditions difference plots of control experiments in which the ATPase was inhibited in a variety of ways were all linear within experimental error. Table 2 shows that the c-values of these difference plots were all considerably lower than those of the parent curves (Table l), that the correlation coefficients of the linear regressions were all greater than 0.97, and, most significantly the value of A log h at zero time obtained by extrapolation of the equation for linear regression using only readings taken after 20 mins, differed from its value obtained from the parabolic
482 TABLE
2
Coefficients of parabolic regression (A log h = a + bt + ct’) from t = 0 to t = 60 min, and of linear regression (A log h = a’ + b’t; r = correlation coefficients) from t = 20 to t = 60 min for difference plots derived from experiments in which TEMPOL was added at zero time to each of a pair of samples from a single preparation. In all samples the ATPase was inhibited. ax102
a’XlO*
cx106
r
80 80
7.09
7.2
1.72
0.977
110 110
9.52
9.41
1.9
0.98
ATP (10 mol mm3); EGTA(1 mol mm3) AMP-PNP (10 mol md3); Ca*+(l mol me3)
105 105
1.32
1.41
1.8
0.99
ATP(10 mol mm3); EGTA(l ATP(50 mol mm3); EGTA(1
105 105
1.49
1.65
1.7
0.98
Conditions
Protein (mg cm-3)
of experiment
No ATP; TEMPOL No ATP; TEMPOL
(1O-4 mol dmm3) (S x 10m4 mol dmm3)
ATP (20 mol mA3); EGTA(l No ATP
mol m-3)
mol me3) mol m-3)
curve, using all the readings, by only 0.001. It was concluded that provided in a single set of experiments difference plots for control experiments were linear, it was legitimate to use linear extrapolation of a difference plot for a pair of experiments in which one was a control, and one had an active ATPase, to determine the magnitude of the effect of formation of phosphoenzyme upon the signal from TEMPOL. Table 2 also shows that ATP itself did not cause departure from linearity in the difference plots. When excess EGTA was added to complex residual Cazf, ATP could be added to one of a pair of experiments and the difference plot was linear. Similarly the non-hydrolysable analogue of ATP, adenyl imidodiphosphate (AMP-PNP), could be added with Ca2+ without causing the difference plot to deviate from a straight line. The method of analysis by internal comparison is illustrated for the set of four decay curves shown in Fig. 8A. Two of the curves (a and d) represent spontaneous decay in the absence of ATP. To the other two samples (b and c) 20 mol rnp3 ATP was added with TEMPOL at zero time. The concentration of TEMPOL was 5 x 10e5 mol dme3 (a and b) and lop5 mol dme3 (c and d). The difference plot derived from a and d was linear, justifying the application of the empirical analytical method to the other pairs of curves. The four difference plots (b - d), (c - d), (b - a) and (c - a) all show clear departure from linearity apparent before 30-35 min. The extrapolation of the line through the points from 35-60 min lies above the earliest experimental points, which then rise to join the line. Production of phosphate under experimental conditions identical to those of curve b stopped after 35 min. It can be assumed, therefore, that the phosphoenzyme, which reached its maximal concentration before the first spectral reading was taken at 4 min, declined to zero in 35 min and that the loss of signal height, which followed the same time course, was associated with the phosphoenzyme. The curve (b - c) in Fig. 8b is the
483
004
0
20
40
60
Fig. 8. An example of intcmal comparison by difference plots for TEMPOL in a single preparation of protein concentration 120 mg cme3. (A) The parent decay curves: (a) TEMPOL, 5 X lo-’ M, no ATP; (b) TEMPOL, 5 x lo-’ M, ATP, 20 mol me3; (c) TEMPOL, lo-’ M, ATP 20 mol me3; (d) TEMPOL 10m5 M, no ATP. (B) The difference curves derived from combinations of the parent curves. The figures on the right hand ordinate are for a-d only.
difference between two experiments, each containing 20 mol me3 ATP, but different concentrations of TEMPOL. The linearity of that curve will be discussed later. Figure 9 shows a series of difference plots obtained from different preparations, but always using an internal control. In the presence of 5, 10, 15 and 20 mol rnp3 ATP the anomalously low points met the straight line after approximately 10, 20, 30 and 40 min. With 50 and 100 mol m- 3 ATP, which should last approximately 100 and 200 min respectively, the readings for 60 min are not enough to define the curves. For subsequent experiments 10 mol m -3 ATP was usually used. More than 10 mol rnp3 ATP required too long an extrapolation, and with less ATP the concentration of phosphoenzyme had probably declined significantly before the first spectral measurement could be made. Table 3 summarises the results. The loss of signal due to the presence of the phosphoenzyme was taken as lOO(10” - 1)Yo where x was the difference between extrapolated and experimental values of A log h at 4 min. For different combina-
484
tions of concentrations of protein, TEMPOL and ATP, the percentage signal lost per mg protein was fairly constant, with a mean and standard deviation of 1.29 & 0.02% for 22 determinations from 8 preparations of vesicles. The range of protein concentrations was only from 80 to 120 mg cme3; below 80 mg cmp3 a loss of a signal was not detectable, and intact vesicular preparations more concentrated than 120 mg cme3 were difficult to manipulate. The useful range of ATP concentrations was limited to 5-20 mol rnd3 (see Fig. 8) and that of TEMPOL from lop5 mol dme3 (below which the signal to noise ratio was too low) to 10e4 mol dm-3 (above which h did not increase linearly with concentration, presumably because of probe-probe interactions). Under the conditions of these experiments the signal to noise ratio was very high. Repetitive values of h measured on the same sample differed by only 0.1-0.2 scale divisions. Experimentally h was always 200-220 scale divisions, so that a loss of l-2% of signal height amounting to 2-4 scale divisions
TABLE
3
Estimates of the percentages of the signal lost at 4 min during ATPase activity, taken as lOO(10’ - 1) where x was the difference between the extrapolated and experimental values of A log h at 4 min. Mean standard deviation 1.29 + 0.02 (n = 26) Protein (mg cmm3) 80 80 80 120 120 120 120 105 105 108 108 108 110 110 110 110 108 108 105 105 105 109 97 97 97 97
TEMPOL (mol dmm3 x 105)
ATP mol mm3
% lost signal at 4 min
% lost signal per mg protein X 10 2
5 5
12 12 12 20 20 20 20 10 20 10 10 10 20 20 20 20 10 10 10
5 10 75 75 10 10
15 10 10 10 10 10
1.0 0.9 1.0 1.4 2.1 1.5 1.6 1.2 1.2 1.4 1.5 1.3 1.2 1.3 1.4 1.6 1.4 1.3 1.6 1.15 1.5 1.65 1.3 1.4 1.2 1.2
1.25 1.13 1.25 1.17 1.75 1.25 1.33 1.14 1.14 1.30 1.39 1.20 1.09 1.18 1.27 1.45 1.30 1.20 1.52 1.10 1.43 1.51 1.34 1.44 1.24 1.24
2.5 3.5 3.5 3.5 3.5 2.5 2.5 2.5 5 5 5
5 5
485
-002-
-----,-;;.A
I_._.-.~_I_~ ._
A-L.-.-.-.-
-003_
AAuAr
I.._._.
Fig. 9. The effect of ATP concentration on difference plots for the same protein concentration (108 mg cme3). The concentrations of ATP added at zero time with lo-’ mol dmm3 TEMPOL were: (0) 100 mol mm3; (A) 50 mol m-l; (W) 20 mol mm3; (0) 15 mol mm3; (A) 10 mol mm3; (0) 5 mol me3.
was detectable reproducibility
with a 5-10% error. was of that order.
The last column
of Table
3 shows
that
the
DISCUSSION
It has been shown that during the limited cycling of the Ca-ATPase in the *+ ions there were losses in peak height in the upfield e.s.r. spectra of absence of Mg both TEMPO and TEMPOL; the measured loss was greatest at the earliest time of measurement after addition of ATP, and then decreased to zero as ATP was consumed. The return of the lost signal height precludes the possibility that the probes were chemically transformed during the ATPase cycle. The loss of peak height must therefore have been due to spectral broadening which might result from immobilisation of probes which were bound to the phosphoenzyme, EP, interactions between spin probes and the magnetic fields of free radical intermediates of the ATPase reaction, or to an increase in microscopic viscosity of the environment of some probes. If TEMPOL was immobilised because it bound to one or more of the conformations assumed by the ATPase during a catalytic cycle, or if its resonance was broadened by the presence of free radical intermediates, then the fraction of signal lost at a constant protein concentration should decrease with increasing concentra-
486
tions of TEMPOL. Table 3 shows, however, that the percentage signal lost was constant in the TEMPOL concentration range 10-5-10-4 mol dmp4. Further evidence that the fraction of signal lost was independent of TEMPOL concentration can be deduced in the following way: if the decay curve for the resonance signal in an ATP-free suspension at a TEMPOL concentration m, is given by log h, =a+bt+ct*
(3)
and the fraction of TEMPOL immobilised in a paired experiment was (pi then the component of spontaneous decay of the signal taining experiment was log(1 - c~i)h, = a, + b,t + c,t2
containing ATP in the ATP-con-
(4)
so that log hi = a, + b,t + c,t* + log(1 - a,) For another
concentration
of TEMPOL,
m2,
log h, = a2 + b,t + c,t* + log(1 - 0~~)
(5)
The two fractions (pi and I_X*are functions of time which are determined by the rate of disappearance of the conformer which immobilised the spin probe and on the are the same values of q and a2 at t = 0. If protein, Ca2+ and ATP concentrations in the two experiments, and if, at all times (Ye= (Y*, then log hi - log h, = (a, - a2) + (b, - b2)t + (cl - c2)t2
(6)
i.e. the difference plot for samples of the same preparation of vesicles with the same concentration of ATP but different concentrations of TEMPOL should be linear only if the fraction of signal lost is independent of TEMPOL concentration. This is illustrated in curve (b - c) of Fig. 8B. Six such plots over a tenfold concentration range of TEMPOL were indistinguishable from straight lines within the limits of experimental error (c < 2.5 X 10p6; r > 0.97). It therefore seems unlikely that the loss of resonance signal height was due to binding of the probe or to interactions with free radical intermediates. The findings are consistent, however, with a change in the microenvironment of probes present in the cleft of the ATPase when it is phosphorylated: following closure of the cleft, water stretches outward, and probes remaining in the more viscous stretched water resonate with a broader signal. ACKNOWLEDGEMENTS
This work was carried out during the tenure of a Medical Research Council of New Zealand Career Fellowship (P.M.W.). We would like to thank the New Zealand University Grants Committee and the Medical Research Council of New Zealand for financial assistance. We are grateful for the skilled technical assistance of Mr. R.T. van Ryn.
487 REFERENCES 1 2 3 4 5 6 7 8 9
Y. DuPont and R. Pougeois, FEBS Lett., 156 (1983) 93. R.K. Nakamoto and G. Inesi, J. Biol. Chem., 259 (1984) 2961. P.M. Wiggins and R.T. van Ryn, J. Macromol. Sci. Chem., A23 (1986) 875. P. Meares, Phil. Trans. B, 278 (1977) 113. D. Eisenberg and W. Kauzmann, The Structure and Properties of Water, Clarendon 1969, pp. 121 and 218. P.M. Wiggins, J. Theor. Biol., 99 (1982) 645. P.M. Wiggins, J. Theor. Biol., 99 (1982) 665. P.M. Wiggins, Bioelectrochem. Bioenerg., 14 (1985) 313, 327, 339. P.M. Wiggins, J. Biol. Chem., 255 (1980) 11365.
Press, Oxford,
Bioelectrochemistry and Bioenergetics, 17 (1987) 473-487 A section of J. Electroanal. Chem., and constituting Vol. 231 (1987) Elsevier Sequoia S.A., Lausanne - Printed in The Netherlands
983 - THE STATE OF WATER ASSOCIATED WITH THE PHOSPHOENZYME OF THE Ca-ATPase OF SARCOPLASMIC RETICULUM A SPIN PROBE STUDY
PHILIPPA
M. WIGGINS
Department
of Medicine,
GRAHAM
A. BOWMAKER
Department
of Chemistry
(Manuscript
University of Auckland
School of Medicine, Private Bag, Auckland
University of Auckland,
received August
Private Bag, Auckland
(New Zealand)
(New Zealand)
20th 1986)
SUMMARY Electron spin resonance spectra of nitroxide free radicals in the aqueous phase associated with suspensions of sarcoplasmic reticulum vesicles have been measured. While the Ca-ATPase was cycling there was a loss of signal which returned to normal as ATP was consumed. The fractional loss of signal increased with protein concentration. but was independent of concentration of spin probe, suggesting that it was unlikely to be due to binding of the probe to one of the conformations assumed by the enzyme during a catalytic cycle. The results are consistent with an increase in viscosity of water associated with the phosphoenzyme intermediate of the ATPase, similar to the increase of viscosity of water which has been observed in hydrophobic pores of synthetic polymers. It is suggested that probes in this relatively viscous microaqueous phase tumble more slowly than probes in the rest of the water, so that with formation of the phosphoenzyme a fraction of the signal broadens and loses peak height.
INTRODUCTION
The Ca-ATPase of sarcoplasmic reticulum is one example of many enzymes which have their active sites buried in hydrophobic pockets. DuPont and Pourgeois [l] and Nakamoto and Inesi [2] using the fluorescence of TNP nucleotides found that an environment of low polarity before phosphorylation of the enzyme decreased drastically in polarity and increased in viscosity following phosphorylation. They suggested that the ATP binding site is in a hydrophobic cleft which closes upon phosphorylation. Because water in such a small hydrophobic domain may be structurally perturbed, we have investigated the state and properties of water in the 0302-4598/87/$03.50
0 1987 Elsevier Sequoia
S.A.
Fig. 1. Diagrammatic illustrations of three different structures of water. (a) The continuous random defective network of a normal liquid phase. (b) Stretched water, contained in hydrophobic pores, and in equilibrium with a normal aqueous phase or with solutions of Na+, Li+, Ca*+ or Mg*+ salts. (c) Vapour-like water, contained in small hydrophobic pores and in equilibrium with a solution of KC1 or CsCl. For discussion see text.
small hydrophobic pores of cellulose acetate membranes, in order to determine whether or not it is likely that changes in solvent structure contribute to the coupling mechanism of the Ca-ATPase [3]. Water in small hydrophobic pores of regular copolyoxamide and cellulose acetate membranes was found to exist in two states, both of which were profoundly different from the normal liquid state. They are illustrated schematically in Fig. 1, and contrasted with the continuous random defective network of normal liquid water (Fig. la). In the absence of solutes the water was stretched (Fig. lb). The infrared spectrum was characteristic of water monomers adsorbed to the hydrophobic surface, and of large strongly hydrogen-bonded clusters. Absorbance in the region of weaker hydrogen bonds was relatively slight. When the membranes were equilibrated with chlorides of Na+, Li’, Ca2+ or Mg’+ the strongly-bonded clusters increased still more at the expense of weakly-bonded clusters. Partition coefficients of the salts were less than one and decreased in the order Na+ > Li+ z=- Ca2+ > When the membranes were equilibrated Mg2+> and with increasing concentration. with KC1 or CsCl at concentrations of 1 mol rnp3 or higher, water in the pores assumed a vapour-like state illustrated in Fig. lc. The infrared spectrum showed no absorbance in the regions of strong hydrogen-bonding but only in the regions where monomers and dimers absorb. This water was a good solvent for salts of the small cations. KC1 and CsCl were accumulated into the stretched water with partition coefficients as high as 30, provided that the external concentration was low enough (< 0.2 mol mp3). It was suggested that KC1 prevents stretching of the pore water by accumulating into it, and generating an osmotic pressure gradient which oppose stretching.
415
The perturbation of water structure in small hydrophobic pores illustrated in Fig. 1, was found to increase as the size of the domain decreased. The hydrophobic cleft of the Ca-ATPase is likely to be smaller than the cellulose acetate pores of diameter approximately 3 nm. Therefore water in that cleft must be similarly perturbed. In the presence physiological concentrations of KC1 the cleft should contain water in the vapour-like state. When, however, the cleft closes upon the phosphorylation, some solution is squeezed out, and the remaining water, which now occupies a much smaller hydrophobic domain, will stretch out of the cleft, and assume the kind of structure illustrated in Fig. lb. This will persist until KC1 diffuses in and water is forced to revert to the vapour state. The self-diffusion coefficient of water in cellulose acetate membranes was found to be of the order of lo-’ cm2 s-l [4] which is much lower than that of liquid water (2.2 x lop5 cm2 s-‘) [5], suggesting that, as would be expected from its highly hydrogen-bonded structure, stretched water is very viscous. This high viscosity will delay influx of KC1 and allow the stretched state to persist for an appreciable time. A mechanism for coupling of ATP hydrolysis to cation transport by means of transient changes in water structure has already been outlined [6-81, and will be described in more detail elsewhere. This paper describes experiments which have been designed to test the hypothesis that the phosphoenzyme intermediate of the Ca-ATPase of sarcoplasmic reticulum contains stretched water. The phosphoenzyme was formed from ATP in the absence could be formed of Mg2+, so that when ATP was used up no more phosphoenzyme from inorganic phosphate (Pi). The electron spin resonance (e.s.r.) spectra of nitroxide free radicals give information about the microenvironment of the spin probe. Those probes relatively immobilised in the cleft when it closes and the water stretches, should resonate with a broader signal than probes in the rest of the water of the system. Under the conditions of the experiments reported here line broadening was most sensitively detected as a loss in peak height. It is shown that there was a loss of signal height from a spin probe in solution in a suspension of sarcoplasmic reticulum vesicles when the phosphoenzyme was formed; the loss in peak height returned as the concentration of phosphoenzyme declined with consumption of ATP. As expected the lost signal height amounted to only l-2% of the total. MATERIALS
AND
METHODS
Sarcoplasmic reticulum vesicles were prepared from rabbit back and leg muscles as previously described [9], and finally suspended at a protein concentration of approximately 100 mg cme3 in a standard buffer solution containing KC1 100 mol rne3, Tris (hydroxymethyl) aminomethane (Tris) maleate (5 mol m-3, pH 7). Stock solutions of ATP (1 mol dmp3), CaCl, (0.5 mol dmp3), ethyleneglycol-bis-(Paminoethylether) N,N’-tetraacetic acid (EGTA), (0.5 mol dm-3) and 4-hydroxy2,2,6,6-tetramethylpiperidine-1-oxyl (TEMPOL) (5 X lo-’ mol dme3) were made up in the same buffer solution with pH adjusted to 7: 2,2,6,6_tetramethylpiperidine1-0~~1 (TEMPO) (lop2 mol dmw3) and the ionophore A23187 (0.5 mg cm-‘),
generously supplied by Lilley Laboratories, were made up in ethanol. Phosphate production was measured as previously described [9]. Protein was determined using a biuret method. For experiments in which the reaction was to be controlled by the presence or absence of Ca*+, EGTA (0.5 mol dme3) and A23187 (5 X lop4 mg extraction. cmp3) were added to the 0.6 mol dm-3 KC1 used for the overnight Residual Ca2+ was then. negligibly low, and the contaminating Ca*+ added with ATP was completely complexed by an additional 1 mol rnp3 EGTA added with the ATP. e.s.r. spectra were recorded at room temperature on a Varian E-4 spectrometer operating at about 9.1 GHz with a modulation frequency of 100 kHz. Samples were contained in the B248 aqueous solution sample cell. EXPERIMENTAL
PROCEDURE
The concentrated suspensions used were difficult to pipette accurately and to mix efficiently. All additions had therefore to be made before the sample was transferred to the flat aqueous cell of the spectrometer. It was found that the signals from both TEMPO and TEMPOL in the presence of vesicles decayed with time (see Fig. 4 below), so that comparisons of absolute amplitudes from experiment to experiment could not be made. The time-courses of the decay in signal were therefore followed in paired experiments in which the upfield resonance of a spin probe was measured: (i) during and after ATPase activity, and (ii) in the absence of ATPase activity, other conditions being the same. In a typical pair of experiments 1 g aliquots of vesicles at a protein concentration of 100 mg cmp3 and containing 3 pmole of Ca*+ and 10 pg of ionophore A23187 were weighed into each of two small plastic tubes. At time zero 20 mm3 ATP and 10 mm3 TEMPOL were added to the first tube, the contents mixed thoroughly with a Pasteur pipette, taken up in a syringe and transferred to the flat aqueous cell which was then adjusted in the cavity of the spectrometer. Spectral measurements were made at minute intervals from 4 to 10 min, and at 2 to 5 min intervals thereafter. To the second tube 10 mm3 TEMPOL and 20 mm3 buffer were added at zero time, and spectral readings taken after the same time intervals. RESULTS
It was first established that neither TEMPO nor TEMPOL in the concentration used impaired ATPase activity, which was 1.2 pmol P, mgg’ protein min-’ in the absence of either spin probe, when the probe was added at the same time as the ATP which started the reaction, and when the vesicles were preincubated with the probe for 12 h before activation. In concentrated suspensions of vesicles both nitroxide radicals (Fig. 2) gave characteristically sharp e.s.r. spectra with three hyperfine splittings (Fig. 3). TEMPO, which is more hydrophobic than TEMPOL, distributed between lipid and aqueous phases, giving a further splitting of the high-field resonance [9], due mainly to the
471
G< N-O HO
N-O
TEMPOL
TEMPO
Fig. 2. Structural
formulae
of TEMPO
and TEMPOL.
difference of dielectric constant which is 78 in water and 6 in lipids. The signal height was followed by measuring h in Figs. 3a and 3b. Fig. 3c shows the spectrum of TEMPOL in a vesicle suspension frozen at - 50 o C, illustrating the broadening of the signal and loss of amplitude which resulted from immobilisation of the probe in random orientations. Loss of amplitude from the upfield resonance was especially marked. Spontaneous
decay of the resonance
signal
In the absence of ATPase activity, the signal gave characteristic decay curves illustrated in Fig. 4. The absolute values of h at t = 0 are not comparable because it was impossible to prevent formation of air bubbles in the viscous solution. The decay was not first order, since log h uersus time curves were not linear. The rate of decay and its curvature increased with protein concentration and changed with composition of the supporting medium. The signal from TEMPO decayed more rapidly than that from TEMPOL and with greater curvature. Decay of the signals
_ ”
20 gauss H
/
‘J
( ,’ / / ,
Fig. 3. First derivative e.s.r. spectra of nitroxide free radicals in suspensions of sarcoplasmic reticulum vesicles at a protein concentration of 101 mg cmm3. (a) TEMPOL (10m5 mol dmm3) at room temperature; its concentration in the aqueous phase is assumed to be proportional to h; (b) TEMPO (10V4 mol dme3) at room temperature; its concentration in the aqueous phase is assumed to be proportional to h; (c) TEMPOL in suspensions frozen at - 50 DC.
478
-m---=-=-m -=--m-m-m
23A_-.
--L-L_. -A-A-A
_ ***-o--.-
.-.-
c 0" 2.1-J
I
0
I
I
20
t
.-.-_.
-A-A-A -•-•--0-0
IllIn) I
I
40
1
1
60
Fig. 4. Spontaneous decay of the resonances of TEMPO and of TEMPOL in the aqueous phase of concentrated suspensions of sarcoplasmic reticulum vesicles. (m, A, 0) TEMPOL; protein concentration: (0) 80 (m) 80 mg cme3; (A) 120 mg cmm3; (0) 105 mg cm -3. (0, 0, A) TEMPO; protein concentration: mg cmm3; (0) 120 mg cm-3; (a) 150 mg cme3.
from TEMPO and TEMPOL in buffer solution was extremely slow and first order. The higher rate of decay in the presence of vesicles was not due to chemical reaction between the nitroxide and a membrane component, because, when a suspension containing TEMPO was left in the dark for two days, the lost signal could be restored by extraction with an equal volume of ethanol. It seemed probable that the decay in signal was due to a relatively slow diffusion of the probe into more rigid lipid components of the membrane, with immobilisation and loss of signal. This mechanism accounts for the faster decay of TEMPO which is more soluble than TEMPOL in lipids, and its increase in decay rate with protein concentration. Whatever its mechanism the spontaneous loss of signal necessitated readings at closely-spaced intervals of time, so that the width could not be expanded sufficiently to allow accurate measurements. Therefore loss of peak height was measured rather than increase in line width. The effect of A TPase activity Figure 5 shows examples of decay CUNeS for TEMPO in concentrated suspensions of vesicles to which ATP was added at zero time. When a low concentration of ATP was added to a highly concentrated suspension the early points of the decay curve appear to lie below any reasonable extrapolation of the rest of the curve to zero time. The marked curvature of the plots precludes more than the qualitative conclusion that while there is still an excess of ATP present and one might expect the concentration of phosphoenzyme to be high, some signal height appears to have
479
Fig. 5. The effect of ATPase activity upon the resonance of TEMPO from the aqueous phase of a suspension of vesicles. ATP and TEMPO were added at zero time to a suspension of vesicles already containing Ca*+ and ionophore A23187, as described in Methods. (0) Protein, 150 mg cm-3; ATP 20 mol md3; (m) protein, 100 mg cmm3; ATP, 200 mol m-3; (A) protein, 100 mg cmm3; ATP, 20 mol rne3; (A) protein, 160 mg cm-3; ATP, 30 mol me3.
been lost from the aqueous phase. At a lower concentration of protein (100 mg cm-3) the initial points are not obviously below the extrapolation of the rest of the curve; while in the presence of excess ATP the initial points appear to be normal but the loss of signal at later times appears to be slowed down. Before an estimate of the magnitude of the lost signal could be made, conditions under which the decay curve could be reliably extrapolated back to zero time were explored. The use of difference plots Even when plotted on a logarithmic scale the decay curves had significant curvature, the magnitude of which was reflected in the value of the coefficient c in the equation log h=a+bt+ct*
(1) Curves of this kind were fitted to the data by parabolic regression in order to find conditions under which coefficient c was a minimum and therefore extrapolation to zero time most reliable. Table 1 shows values of the three coefficients of equation 1 for decay curves of either TEMPO or TEMPOL in ATP-free suspensions of vesicles. In all cases the coefficient c was small and positive; it increased with protein concentration and was always higher for TEMPO than for TEMPOL. When one such curve is subtracted from another, the difference plot A log h = (aI -al)
+ (b, - b2)t + (cl - cz)t2
is more nearly linear than either of the parent curves, provided two coefficients of the quadratic terms are of similar magnitude.
(2) that ci and c2, the
480 TABLE
I
Coefficients of the equation log h = a + bt + ct2 fitted by parabolic regression to decay curves for the upfield resonance from the aqueous phase of either TEMPO or TEMPOL in ATP-free suspensions of vesicles protein
Spin probe
(mgcm TEMPO TEMPO TEMPO TEMPOL TEMPOL TEMPOL
-3
a
-bx103
CX106
1.84 2.08 1.77 2.28 2.17 2.25
3.55 3.8 4.3 1.05 1.69 1.94
18.7 21.4 25.6 2.33 5.46 6.59
1
80 110 120 80 105 120
Figure 6 shows representative difference plots for pairs of spontaneous decay curves. All difference plots for TEMPO had significant residual curvature, so that linear extrapolation of spectral readings obtained after 20 min would lead to considerable error. Similarly difference plots for TEMPOL in a pair of experiments with different protein concentrations. and, therefore, different c-values, were significantly curved. When, however, the parent curves had similar and low values of c, the difference curve was linear within experimental error. Therefore when one of a
010 --•---•_.___ •~~~~~~~---~
038
.-.-
i 004
c 0”
002
i
4 ----iT&A-*
._~_._.__._._._._._
0 221 --~~__._.-m-.-.-._C.-._
0200
I 20
(mln) 40
60
Fig. 6. Representative difference plots for spontaneous decay of resonance. In a pair of experiments TEMPO or TEMPOL was added at zero time to a suspension of vesicles in which the ATPase was inhibited, and the two spectral parameters h, and h, measured at time intervals. A log h = log h, log h,; c is the coefficient of the quadratic term A log h = a + br + ct2; cl and c2 the corresponding coefficients of the parent curves; r is the correlation coefficient for linear regression. (0) TEMPOL, c=2x10-6 cr = 7.2 x 10m6, c2 = 5.2 x 10e6; r = 0.97. Protein concentrations were both 105 mg crne3. (A) TEMPOL; c = - 4.5 X lo@, c, = 2.8 x lo@, c2 = 7.2 X 10-6, r = 0.83. Protein concentrations were 80 and 105 mg cme3. (W) TEMPO; c = 4.1 x 10m6, c, = 2.9 X lo-‘, c2 = 4.2 X lo-’ r = 0.87. Protein concentrations were both 108 mg cm- 3.
481
I 0
I
20
I
40
60
Fig. 7. Difference plots (experimental-control) for TEMPO in equal aliquots of a single preparation of vesicles; protein concentration 120 mg cm-3. (a) The parent curves contained 20 mol mm3 ATP and no ATP; (b) the parent curves contained 23 mol m -3 ATP and no ATP; (c) the parent curves contained 25 mol mm3 ATP and no ATP.
pair of curves is obtained from a suspension to which ATP is added at zero time, and the second curve is from a suspension of the same concentration without ATP, any departure from linearity in the difference plot can be assumed to reflect ATPase activity. The residual curvature in the difference plots for all experiments using TEMPO prevented this method of quantitative analysis. Figure 7 shows three difference plots for TEMPO in a concentrated suspension of vesicles to which 20, 23 and 25 nmol ATP was added at zero time. The common control was an experiment in which buffer was added instead of ATP. The c-value of the decay curve without ATP (4.2 X 10e5) increased in these difference curves to 2 X 10m4, 9 X 10e5 and 1 x lop4 respectively. The increase in curvature is again consistent with a loss of signal height associated with the initial high concentration of phosphoprotein, and its return as ATP was used up. Since, however, it is very probable that the difference between the two spontaneous components of the decay curves was not linear, a reliable extrapolation cannot be made. Spectral measurements of TEMPOL were analysed using difference plots of paired experiments from a single preparation, so that the concentration of protein, which largely determined the curvature of the signal decay, was constant. Under these conditions difference plots of control experiments in which the ATPase was inhibited in a variety of ways were all linear within experimental error. Table 2 shows that the c-values of these difference plots were all considerably lower than those of the parent curves (Table l), that the correlation coefficients of the linear regressions were all greater than 0.97, and, most significantly the value of A log h at zero time obtained by extrapolation of the equation for linear regression using only readings taken after 20 mins, differed from its value obtained from the parabolic
482 TABLE
2
Coefficients of parabolic regression (A log h = a + bt + ct’) from t = 0 to t = 60 min, and of linear regression (A log h = a’ + b’t; r = correlation coefficients) from t = 20 to t = 60 min for difference plots derived from experiments in which TEMPOL was added at zero time to each of a pair of samples from a single preparation. In all samples the ATPase was inhibited. ax102
a’XlO*
cx106
r
80 80
7.09
7.2
1.72
0.977
110 110
9.52
9.41
1.9
0.98
ATP (10 mol mm3); EGTA(1 mol mm3) AMP-PNP (10 mol md3); Ca*+(l mol me3)
105 105
1.32
1.41
1.8
0.99
ATP(10 mol mm3); EGTA(l ATP(50 mol mm3); EGTA(1
105 105
1.49
1.65
1.7
0.98
Conditions
Protein (mg cm-3)
of experiment
No ATP; TEMPOL No ATP; TEMPOL
(1O-4 mol dmm3) (S x 10m4 mol dmm3)
ATP (20 mol mA3); EGTA(l No ATP
mol m-3)
mol me3) mol m-3)
curve, using all the readings, by only 0.001. It was concluded that provided in a single set of experiments difference plots for control experiments were linear, it was legitimate to use linear extrapolation of a difference plot for a pair of experiments in which one was a control, and one had an active ATPase, to determine the magnitude of the effect of formation of phosphoenzyme upon the signal from TEMPOL. Table 2 also shows that ATP itself did not cause departure from linearity in the difference plots. When excess EGTA was added to complex residual Cazf, ATP could be added to one of a pair of experiments and the difference plot was linear. Similarly the non-hydrolysable analogue of ATP, adenyl imidodiphosphate (AMP-PNP), could be added with Ca2+ without causing the difference plot to deviate from a straight line. The method of analysis by internal comparison is illustrated for the set of four decay curves shown in Fig. 8A. Two of the curves (a and d) represent spontaneous decay in the absence of ATP. To the other two samples (b and c) 20 mol rnp3 ATP was added with TEMPOL at zero time. The concentration of TEMPOL was 5 x 10e5 mol dme3 (a and b) and lop5 mol dme3 (c and d). The difference plot derived from a and d was linear, justifying the application of the empirical analytical method to the other pairs of curves. The four difference plots (b - d), (c - d), (b - a) and (c - a) all show clear departure from linearity apparent before 30-35 min. The extrapolation of the line through the points from 35-60 min lies above the earliest experimental points, which then rise to join the line. Production of phosphate under experimental conditions identical to those of curve b stopped after 35 min. It can be assumed, therefore, that the phosphoenzyme, which reached its maximal concentration before the first spectral reading was taken at 4 min, declined to zero in 35 min and that the loss of signal height, which followed the same time course, was associated with the phosphoenzyme. The curve (b - c) in Fig. 8b is the
483
004
0
20
40
60
Fig. 8. An example of intcmal comparison by difference plots for TEMPOL in a single preparation of protein concentration 120 mg cme3. (A) The parent decay curves: (a) TEMPOL, 5 X lo-’ M, no ATP; (b) TEMPOL, 5 x lo-’ M, ATP, 20 mol me3; (c) TEMPOL, lo-’ M, ATP 20 mol me3; (d) TEMPOL 10m5 M, no ATP. (B) The difference curves derived from combinations of the parent curves. The figures on the right hand ordinate are for a-d only.
difference between two experiments, each containing 20 mol me3 ATP, but different concentrations of TEMPOL. The linearity of that curve will be discussed later. Figure 9 shows a series of difference plots obtained from different preparations, but always using an internal control. In the presence of 5, 10, 15 and 20 mol rnp3 ATP the anomalously low points met the straight line after approximately 10, 20, 30 and 40 min. With 50 and 100 mol m- 3 ATP, which should last approximately 100 and 200 min respectively, the readings for 60 min are not enough to define the curves. For subsequent experiments 10 mol m -3 ATP was usually used. More than 10 mol rnp3 ATP required too long an extrapolation, and with less ATP the concentration of phosphoenzyme had probably declined significantly before the first spectral measurement could be made. Table 3 summarises the results. The loss of signal due to the presence of the phosphoenzyme was taken as lOO(10” - 1)Yo where x was the difference between extrapolated and experimental values of A log h at 4 min. For different combina-
484
tions of concentrations of protein, TEMPOL and ATP, the percentage signal lost per mg protein was fairly constant, with a mean and standard deviation of 1.29 & 0.02% for 22 determinations from 8 preparations of vesicles. The range of protein concentrations was only from 80 to 120 mg cme3; below 80 mg cmp3 a loss of a signal was not detectable, and intact vesicular preparations more concentrated than 120 mg cme3 were difficult to manipulate. The useful range of ATP concentrations was limited to 5-20 mol rnd3 (see Fig. 8) and that of TEMPOL from lop5 mol dme3 (below which the signal to noise ratio was too low) to 10e4 mol dm-3 (above which h did not increase linearly with concentration, presumably because of probe-probe interactions). Under the conditions of these experiments the signal to noise ratio was very high. Repetitive values of h measured on the same sample differed by only 0.1-0.2 scale divisions. Experimentally h was always 200-220 scale divisions, so that a loss of l-2% of signal height amounting to 2-4 scale divisions
TABLE
3
Estimates of the percentages of the signal lost at 4 min during ATPase activity, taken as lOO(10’ - 1) where x was the difference between the extrapolated and experimental values of A log h at 4 min. Mean standard deviation 1.29 + 0.02 (n = 26) Protein (mg cmm3) 80 80 80 120 120 120 120 105 105 108 108 108 110 110 110 110 108 108 105 105 105 109 97 97 97 97
TEMPOL (mol dmm3 x 105)
ATP mol mm3
% lost signal at 4 min
% lost signal per mg protein X 10 2
5 5
12 12 12 20 20 20 20 10 20 10 10 10 20 20 20 20 10 10 10
5 10 75 75 10 10
15 10 10 10 10 10
1.0 0.9 1.0 1.4 2.1 1.5 1.6 1.2 1.2 1.4 1.5 1.3 1.2 1.3 1.4 1.6 1.4 1.3 1.6 1.15 1.5 1.65 1.3 1.4 1.2 1.2
1.25 1.13 1.25 1.17 1.75 1.25 1.33 1.14 1.14 1.30 1.39 1.20 1.09 1.18 1.27 1.45 1.30 1.20 1.52 1.10 1.43 1.51 1.34 1.44 1.24 1.24
2.5 3.5 3.5 3.5 3.5 2.5 2.5 2.5 5 5 5
5 5
485
-002-
-----,-;;.A
I_._.-.~_I_~ ._
A-L.-.-.-.-
-003_
AAuAr
I.._._.
Fig. 9. The effect of ATP concentration on difference plots for the same protein concentration (108 mg cme3). The concentrations of ATP added at zero time with lo-’ mol dmm3 TEMPOL were: (0) 100 mol mm3; (A) 50 mol m-l; (W) 20 mol mm3; (0) 15 mol mm3; (A) 10 mol mm3; (0) 5 mol me3.
was detectable reproducibility
with a 5-10% error. was of that order.
The last column
of Table
3 shows
that
the
DISCUSSION
It has been shown that during the limited cycling of the Ca-ATPase in the *+ ions there were losses in peak height in the upfield e.s.r. spectra of absence of Mg both TEMPO and TEMPOL; the measured loss was greatest at the earliest time of measurement after addition of ATP, and then decreased to zero as ATP was consumed. The return of the lost signal height precludes the possibility that the probes were chemically transformed during the ATPase cycle. The loss of peak height must therefore have been due to spectral broadening which might result from immobilisation of probes which were bound to the phosphoenzyme, EP, interactions between spin probes and the magnetic fields of free radical intermediates of the ATPase reaction, or to an increase in microscopic viscosity of the environment of some probes. If TEMPOL was immobilised because it bound to one or more of the conformations assumed by the ATPase during a catalytic cycle, or if its resonance was broadened by the presence of free radical intermediates, then the fraction of signal lost at a constant protein concentration should decrease with increasing concentra-
486
tions of TEMPOL. Table 3 shows, however, that the percentage signal lost was constant in the TEMPOL concentration range 10-5-10-4 mol dmp4. Further evidence that the fraction of signal lost was independent of TEMPOL concentration can be deduced in the following way: if the decay curve for the resonance signal in an ATP-free suspension at a TEMPOL concentration m, is given by log h, =a+bt+ct*
(3)
and the fraction of TEMPOL immobilised in a paired experiment was (pi then the component of spontaneous decay of the signal taining experiment was log(1 - c~i)h, = a, + b,t + c,t2
containing ATP in the ATP-con-
(4)
so that log hi = a, + b,t + c,t* + log(1 - a,) For another
concentration
of TEMPOL,
m2,
log h, = a2 + b,t + c,t* + log(1 - 0~~)
(5)
The two fractions (pi and I_X*are functions of time which are determined by the rate of disappearance of the conformer which immobilised the spin probe and on the are the same values of q and a2 at t = 0. If protein, Ca2+ and ATP concentrations in the two experiments, and if, at all times (Ye= (Y*, then log hi - log h, = (a, - a2) + (b, - b2)t + (cl - c2)t2
(6)
i.e. the difference plot for samples of the same preparation of vesicles with the same concentration of ATP but different concentrations of TEMPOL should be linear only if the fraction of signal lost is independent of TEMPOL concentration. This is illustrated in curve (b - c) of Fig. 8B. Six such plots over a tenfold concentration range of TEMPOL were indistinguishable from straight lines within the limits of experimental error (c < 2.5 X 10p6; r > 0.97). It therefore seems unlikely that the loss of resonance signal height was due to binding of the probe or to interactions with free radical intermediates. The findings are consistent, however, with a change in the microenvironment of probes present in the cleft of the ATPase when it is phosphorylated: following closure of the cleft, water stretches outward, and probes remaining in the more viscous stretched water resonate with a broader signal. ACKNOWLEDGEMENTS
This work was carried out during the tenure of a Medical Research Council of New Zealand Career Fellowship (P.M.W.). We would like to thank the New Zealand University Grants Committee and the Medical Research Council of New Zealand for financial assistance. We are grateful for the skilled technical assistance of Mr. R.T. van Ryn.
487 REFERENCES 1 2 3 4 5 6 7 8 9
Y. DuPont and R. Pougeois, FEBS Lett., 156 (1983) 93. R.K. Nakamoto and G. Inesi, J. Biol. Chem., 259 (1984) 2961. P.M. Wiggins and R.T. van Ryn, J. Macromol. Sci. Chem., A23 (1986) 875. P. Meares, Phil. Trans. B, 278 (1977) 113. D. Eisenberg and W. Kauzmann, The Structure and Properties of Water, Clarendon 1969, pp. 121 and 218. P.M. Wiggins, J. Theor. Biol., 99 (1982) 645. P.M. Wiggins, J. Theor. Biol., 99 (1982) 665. P.M. Wiggins, Bioelectrochem. Bioenerg., 14 (1985) 313, 327, 339. P.M. Wiggins, J. Biol. Chem., 255 (1980) 11365.
Press, Oxford,