+ Ts. John F. EcclestonSS, Tazeen F. KanagasabaiS, and Michael A. Geevesn

Transcription

1 THE JOURNAL OF BIOLOGICAL CHEMISTRY by The American Society for Biochemistry and Molecular Biology. Ine Vol. 263, No. 10, Issue of April 5, pp , 1988 Printed in U.S.A. The Kinetic Mechanism of the Release of Nucleotide from Elongation Factor Tu Promoted by Elongation Factor Ts Determined by Pressure Relaxation Studies* (Received for publication, September 9, 1987) John F. EcclestonSS, Tazeen F. KanagasabaiS, and Michael A. Geevesn From the $National Institute for Medical Research, Mill Hill, London NW7 IAA, United Kingdom and the llbiochemistry Department, University of Bristol Medical School, University Walk, Bristol BS8 1 TO, United Kingdom The release of a chromophoric analogue ofgdp, 2-amino-6-mercaptopurine riboside 5 diphosphate (thiogdp), from its complex with elongation factor Tu (EF-Tu) is catalyzed by elongation factor Ts (EF-Ts). The mechanism of this reaction includes a ternary complex; EF-Tu. thiogdp. EF-Ts (Eccleston, J. F. (1984) J. Biol. Chem. 259, ). This mechanism has been further investigated using pressure relaxation techniques combined with spectrophotometric measurements. The equilibrium of a solution of EF- Tu, EF-TS, and thiogdp over a range of concentrations is perturbed on increasing the pressure to 150 atm. Rapid decrease of the pressure back to 1 atm results in a biphasic relaxation process, an initial fast phase which is complete within 1 ms followed by a slower phase. This is interpreted as the result of an isomerization of the EF-Tu. thiogdp.ef-ts ternary complex which occurs before the release of thiogdp. Such an isomerization process may be a general feature in the release of GDP from guanosine nucleotide-binding proteins. During the elongation cycle of protein biosynthesis of prokaryotic systems, EF-Tu is released from the ribosome as a complex with GDP (for a review, see Ref. 1). The EF-Tu. GDP complex must then be converted to EF-Tu.GTP, a process that involves a second elongation factor, EF-Ts. The EF-Tu. GTP can then bind aminoacyl-trna to continue the cycle. Since a stable EF-Tu-EF-Ts complex can be isolated (2), the conversion of EF-Tu. GDP to EF-Tu. GTP is thought to proceed via an EF-Tu. EF-Ts complex. GTP GDP Tu.GDP + Tu.Ts Ts Tu.GTP SCHEME 1 + Ts The rate of dissociation of GDP from EF-Tu is much faster in the presence of EF-Ts than in its absence. Therefore, the first step of this scheme does not involve the dissociation of *This work was supported by the Medical Research Council, United Kingdom and the Royal Society, United Kingdom. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 5 To whom correspondence should be addressed. The abbreviations used are: EF-Tu, elongation factor Tu; EF-Ts, elongation factor Ts. (These are further abbreviated to Tu and Ts in equations for the sake of clarity.) ThioGDP, 2-amino-6-mercaptopurine riboside 5 -diphosphate; Pa, pascal. GDP from EF-Tu followed by the binding of EF-Ts to form EF-Tu. EF-Ts. The simplest kinetic scheme that can be postulated for this process includes a ternary complex, EF-Tu. GDP.EF-Ts which is formed fast and from which GDP dissociates faster than from the EF-Tu-GDP complex (3). Chau et al. (4) have used rapid equilibrium isotope exchange measurements to study the formation of the EF-Tu. GDP complex after mixing EF-Tu. [3H]GDP with GDP and EF- TS. Their results were consistent with this mechanism for step 1 of Scheme 1, Kl Tu.GDP + TS K2 Tu.GDP.Ts + Tu.Ts + GDP SCHEME 2 where the values of the association equilibrium constants are Kl = 1.8 X lo5 M- and K2 = 6.4 X lo4 M-. A lower estimate for the rate-limiting step for the exchange of [3H]GDP by GDP via EF-Tu-EF-Ts (either kz or k- ) was 1270 s-. They have recently used the same technique to study the formation of EF-Tu- GTP from EF-Tu.EF-Ts, and their results are consistent with the formation of an EF-Tu GTP. EF-Ts ternary complex in this process (5). In order to define all of the rate constants involved in Scheme 2, stopped-flow spectrophotometric measurements were made using a chromophoric analog of GDP, thiogdp (2-amino-6-mercaptopurine riboside 5 -diphosphate) (6). Although this analog binds to EF-Tu approximately 100-fold more weakly than does GDP, its rate of release from EF-Tu is catalyzed by EF-Ts and its interaction with EF-Tu in the elongation cycle is similar to that of GDP (7). The results of this stopped-flow study with thiogdp were consistent with the kinetic mechanism shown in Scheme 2, and the values for the rate constants were kl 3 2 X 10 M- s-, k s-, k2 = 500 s-, and k 2 = 4 x lo5 M s-. The equilibrium constant obtained for the first step from these rate constants for thiogdp is similar to the value of Chau et al. (4) for GDP. The values of the equilibrium constants for the second step are different by a factor of 80. This is almost identical to the difference between the binding constants of GDP and thiogdp to EF-Tu. By this interpretation of the stopped-flow results EF-Ts binds to EF-Tu. thiogdp with a rate constant close to that expected for a diffusion-limited reaction whereas the binding of thiogdp to EF-Tu. EF-Ts is several orders of magnitude slower than this. To explain this result, it was speculated that this binding process actually occurs by a two-step mechanism (6). In order to investigate the process in more detail, pressure relaxation measurements have been made on an equilibrium mixture of EF-Tu, EF-Ts, and thiogdp. This method is based on the principle that increasing the pressure on an equilibrium 4668

2 Release of Nucleotide from Elongation Factor Tu 4669 system may perturb the equilibrium constant of one or more steps of the mechanism. It is known from other systems that an increase in pressure results in the dissociation of protein complexes due to the decreased solvated volume of the monomers relative to the complexes (8-10). Rapid reduction of the pressure (<200 KS) allows the approach back to the original equilibrium position to be followed. This relaxation process is controlled by the rate constants at atmospheric pressure involved in the mechanism. The results presented here show that increased pressure does perturb the equilibrium position of a solution of EF-Tu, EF-Ts, and thiogdp, and the kinetic measurements show the existence of an isomerization of the EF-Tu. thiogdp. EF-Ts ternary complex. MATERIALS AND METHODS ThioGDP and EF-Tu. EF-Ts were prepared as described previously (6, 11). Solvent conditions in all experiments were 50 mm Tris-HC1, ph 7.6, 10 mm MgC12, 0.5 mm dithiothreitol at 20 "C. The pressure relaxation equipment used was that described by Davis and Gutfreund (8). Essentially the pressure on a solution can be increased to 150 atm by means of a hydraulic oil chamber separated from the sample by a thin Teflon membrane. This pressure can be reduced back to atmospheric pressure within 200 ps. Transmission of light through a 1-cm path length of the solution can be recorded with time. The light source is a quartz halide lamp and is passed through a Farrand monochromator with a 10-nm band width. Data collection was as described by Coates et al. (12), and after averaging the data were analyzed by a nonlinear least squares fitting routine (13). Since the transmission changes were small they are directly proportional to concentration changes. Pln I ln (L c,,,,,,,,!k FIG. 1. Pressure-induced changes in the transmission of a solution of EF-Tu. EF-Ts and thiogdp at 352 nm. a, the solution contained 20 pm EF-Tu.EF-Ts, 20 p~ thiogdp, 50 mm Tris- HC1, 10 mm MgC12, 0.5 mm dithiothreitol, ph 7.6. Full scale on the transmission axis is 0.6%. The trace represents the average for 16 relaxations. The vertical arrow indicates the time of pressure release, this being a decrease from 15.2 MPa (150 atm) to 101 kpa (1 atm). b, as a except that the solution contained no EF-Tu. EF-Ts. RESULTS Effect of Pressure Changes on Solutions of EF. Tu. EF-Ts and ThioGDP-When thiogdp is added to a solution of EF- Tu. EF-TS the stopped-flow data (6) predict that the following equilibrium will be rapidly established. 1 Tu.thioGDP + Ts Tu.thi0GDp.T~ SCHEME Tu.Ts + thiogdp Increased pressure on this solution may perturb the values of one or both equilibrium constants and hence alter the equilibrium concentrations of each species. It has been shown that the difference spectrum between EF-Tu. thiogdp and thiogdp shows a positive absorption peak at 352 nm and a negative absorption peak at 328 nm and that the spectrum of EF-Tu- thiogdp.ef-ts is the same as that of EF-Tu. thiogdp (6,7). Monitoring absorption or transmission at 352 or 328 nm, therefore, gives a measurement of the change in the concentrations of EF-Tu. EF-Ts and thiogdp. Increasing the pressure on a solution of EF-Tu. EF-Ts and thiogdp from 1 to 150 atm resulted in a decrease in the transmission of the solution at 352 nm. This is consistent with an increase in the concentrations of EF-Tu.thioGDP and EF-Tu- thiogdp. EF-Ts (i.e. the equilibrium in Scheme 3 is displaced to the left). This dissociation of the dimer, EF- Tu. EF-TS, parallels the effects on the other protein-protein interactions described above. The transmission of the solution at 352 nm was recorded with time following a rapid release of the pressure ((200 Fs) back to atmospheric pressure. A biphasic relaxation process occurred (Figs. la and 2a). The first phase was an increase in transmission and was complete within 1 ms of the pressure decrease. The second phase was a slower increase in transmission back to the value before the pressure was applied. The half-time of this process was dependent upon the concen- I t,..,,,,, I OB t : * i FIG. 2. Pressure-induced change in the transmission of a solution of EF-Tu. EF-Ts and thiogdp at 352 nm. a, the solution contained 20 p~ EF-Tu.EF-Ts, 100 p~ thiogdp, and 50 mm Tris-HCI, 10 mm MgCl,, 0.5 mm dithiothreitol, ph 7.6. Full scale on the transmission axis is 0.6%. Other details are as in Fig. 1. b, as a except that the solution contained no EF-Tu. EF-Ts. trations of the EF-Tu. EF-Ts and thiogdp in the solution (see below). Similar pressure-release experiments were performed on solutions of buffer and buffer containing 20 PM EF-Tu. EF- TS. No changes in transmission at 352 nmwereobserved (data not shown). However, reduction of pressure from 150 to 1 atm on a solution of thiogdp in buffer resulted in a small fast increase in transmission at 352 nm. This was insignificant

3 4670 Release of Nucleotide from Elongation Factor at 20 p~ thiogdp (Fig. lb) but larger at 100 p~ thiogdp (Fig. 2b). This effect is accounted for by the compressibility of water. Water is compressed by 0.6% on increasing the pressure from 1 to 150 atm which results in a corresponding increase in the concentration of thiogdp and hence a decrease in the transmission. Since the absorbances of 20 and 100 p~ thiogdp at 352 nm are 0.22 and 1.09, respectively, at 1 atm, increasing the pressure to 150 atm will result in decreases in transmission of 0.3 and 1.5%. These values only give an approximation of the changes expected in the pressure relaxation experiments. A quantitative comparison is not possible because 352 nm is on a steep change in absorption, and a wide slit width is used on the pressure apparatus to obtain maximum light. The observed transmission changes (Figs. lb and 2b) are smaller than these calculated values although the relative amplitudes are in agreement. Subtraction of the signals obtained from solutions of thiogdp from those containing EF-Tu. EF-Ts and thiogdp still gave a biphasic relaxation process, although the amplitude of the fast phase was reduced (Fig. 3). Monitoring the relaxation process at 328 nm should give identical results to those obtained at 352 nm except that the direction of the transmission changes should bereversed. These measurements are more difficult to make than those at 352 nm since the decreased light output of the light source at shorter wavelengths combined with the higher absorbance of the thiogdp chromophore = 19.1 X lo3 M cm ; 6362 = 10.9 X io3 M cm ) results in a low signal to noise ratio. One experiment is shown in Fig. 4a. It can be seen that the slow phase is in the opposite direction to the signal obtained at 352 nm (Fig. la) with a similar rate, although the low signal to noise ratio of the 328-nm data precludes a quantitative analysis. The fast phase remains an increase in transmission which results from the higher absorbance at 328 nm compared to 352 nm which gives larger increases in compressibility effects. The corresponding experiment at 328 nm with thiogdp alone is shown in Fig. 4b. The low signal to noise ratio prevents the subtraction of trace 4b from 4a to show Tu l t..,,.... / FIG. 4. Pressure-induced change in the transmission of a solution of EF-Tu.EF-Ts and thiogdp at 328 nm. a, the solution contained 20 PM EF-Tu.EF-Ts, 20 PM thiogdp, 50 mm Tris- HCl, 10 mm MgC12, 0.5 mm dithiothreitol, ph 7.6. Full scale on the transmission axis is 0.6%. Other conditions are as in Fig. 1. b, as a except that the solution contained no EF-Tu.EF-Ts. TABLE I The effect of the total concentration of EF-Tu. EF-Ts and thwgdp on the observed first-order rate constant of the pressure-induced changes in transmission at 352 nm Experimental data for the slower phase of experiments shown in Figs. la and 2a were analyzed as first-order reactions. The procedure for obtaining rate constants from computer-simulated reactions is described under Discussion. lthiogdp1 [Tu. Tsl ExDerimental Simulated fim fim k, s kt. s This is an approximate value since at this low concentration of reactants, the observed signal is very small. ~ l,,,,,,,, 1 I t,..,,.,. ] FIG. 3. Pressure-induced changes in the transmission of 80- lutions of EF-Tu.EF.Ts and thiogdp at 352 nm after correcting for the effect of compressibility of water. a, trace la after subtraction of trace lb. b, trace 2a after subtraction of trace 2b. Full scale on the transmission axis is 0.6%. that the relaxation process includes a rapid phase of decrease in transmission. However, at higher concentrations of thiogdp (50 and 100 p ~ in ) the presence of EF-Tu.EF-Ts, the amplitude of the fast phase decreases. This decrease in amplitude can only be accounted for by an increase in the amplitude of the fast relaxation compensating for the compressibility effect. These results show that decreasing pressure results in a biphasic relaxation process that depends on the presence of EF-Tu, EF-Ts, and thiogdp. Effect of Concentration on the Rate of the Pressure-induced Changes in Transmission-The time course of the rapid phase described above was too fast to analyze. Although the pressure releases are made in less than 200 ps, the small amplitude of the signal combined with the very short time constant required for such a measurement makes the signal to noise ratio too low for meaningful analysis. However, the data at 352 nm for the slower phase is amenable to analysis. The slow phase of the relaxation process was analyzed by

4 a nonlinear least squares fitting routine. In all experiments, the data could be fitted to a single exponential. The firstorder rate constants obtained from this analysis are shown in Table I. It can be seen that this observed rate constant is dependent on both the concentration of EF-Tu.EF-Ts and thiogdp. DISCUSSION At all concentrations of EF-Tu. EF-Ts and thiogdp investigated, two distinct phases are seen in the relaxation following increased pressure. No processes are observed in experiments with EF-Tu EF-Ts in the absence of thiogdp. A single rapid relaxation is observed with thiogdp alone, but this is smaller than the fast relaxation observed in the presence of EF-Tu-EF-Ts at 352 nm. We, therefore, conclude that the observed reaction represents two relaxations involving a ternary complex of EF-Tu, EF-Ts, and thiogdp. The stopped-flow data were interpreted as showing that the spectral change occurred at step 2 of Scheme 3, the dissociation of thiogdp from the EF-Tu- thiogdp.ef-ts ternary complex (6). Since two relaxations are observed in the pressure experiments, in terms of this simple scheme the first relaxation must represent perturbation of step 2 (the step with an optical change) which must be faster than step 1 to give the fast relaxation. The second relaxation would then represent the slower relaxation of step 1 in response to the relaxation of step 2. However, the stopped-flow studies (6) suggested that kl 2 2 X lo8 M-' s-', k-, s-', kz = 500 s-', and 12-, = 4 X IO5 M" s-'. These values are incompatible with the observed relaxations; the relaxation of step 1 would be fast relative to that of step 2, and only a single process would be observed. If there is an additional absorbance change on step 1, then this step could be perturbed to give two relaxations, but the rates would still be incompatible with Scheme 3. We must, therefore, consider more complex schemes. The simplest scheme compatible with the data is a three-step reaction where the ternary complex undergoes an isomerization step. 1 2 Tu. thiogdp + Ts + Tu. thiogdp. Ts (A) + Release of Nucleotide from Elongation Factor Tu 4671 Tu. thiogdp. Ts (B) $ Tu. Ts + thiogdp SCHEME 4 where Tu. thiogdp. Ts (A) and Tu. thiogdp Ts (B) represent different structural states of the ternary complex. The possibility of such a scheme was discussed previously to account for the unusually low rate of association of thiogdp with EF-Tu.EF-Ts (6). In considering this scheme, we must first assign the absorbance change to one or more of the proposed steps. For simplicity we assume that the absorbance change occurs primarily on one step. The stopped-flow data suggests that the absorbance change does not occur on binding of EF-Ts to EF-Tu. thiogdp (step 1) since a limiting value of the observed rate is approached at high concentrations of EF-Ts. If the absorbance change occurs on step 3, then the stopped-flow data suggests that k-3 = 4 X lo6 M-' s-' and Kzk3 = 500 s-'. This assignment presents two difficulties. The rate of thiogdp binding to EF-Tu.EF-Ts remains unusually slow for a diffusion-controlled reaction, and perturbation of step 3 cannot give the fast relaxation complete within 1 ms as observed unless ks >> 500 s-' and then K2 << 1. If Kz is much less than 1 then the state Tu-thioGDP-Ts (B) is not significantly occupied at the experimental concentrations of thiogdp which makes this step unlikely to be 3 perturbed by increased pressure. If the absorbance change is on step 2, the proposed isomerization step, then the pressure relaxation and the stoppedflow data are consistent with the mechanism and allow us to set limits for some of the rate constants. The perturbed step must be the one with an absorbance change and be fast to give the first relaxation in less than 1 ms (i.e. kz + k z s-'). The rate of the EF-Ts-induced displacement of thiogdp has a maximum rate of approximately 500 s-' which can be assigned to k,. If thiogdp association is a diffusion-controlled step then k-, 3 lo7 M-' s-'. Thus, K3 d 5 X M. If steps 2 and 3 are considered to replace step 2 of Scheme 3, then the value of K, K3 = 1.11 X M. Therefore, K2 = 225. In order to test the argument above and to determine rate constants not yet defined, computer simulations of this mechanism were performed and the results compared to the pressure relaxation data. These simulations were performed using the experimental concentrations of EF-Tu. EF-Ts and thiogdp shown in Table I. A set of rate constants was then chosen using the criteria outlined above, and the value of Kz was perturbed by increasing the chosen value of k-, by approximately 10%. When the concentration of each species had been obtained at equilibrium, k-2 was changed back to the chosen value, and the relaxation back to the new equilibrium position was followed by monitoring the formation of EF-Tu-thioGDP.EF-Ts (B) + thiogdp. (Although a 10% perturbation of Kz is arbitrary, its value only affects the amplitude of the relaxation processes but not the relative amplitudes of the two phases or the observed rate constants.) An approximate fit to the pressure relaxation data was obtained when the values for the individual rate constants were k, = 1.25 X io7 M-' s-', k-, = 65 s-', k, = 1500 s-', k 2 = 120 s-', k3 = 375 s-', and k-3 = lo7 M" s-'. These values are also consistent with an overall equilibrium constant for the reaction being 90. A comparison of the experimental data with an analysis of the simulated data is shown in Table I. A closer fit between the experimental and simulated data could possibly be achieved, but more detailed analysis was not considered useful for the following reasons. First, the concentration range over which relaxations could be observed was narrow with only about a %fold change in the observed rate constants. Second, the signal to noise ratio in the observed relaxations is such that for the best data the observed rate constant is only defined to within 20%. Third, the model suggests that no single step is rate-limiting over the entire concentration range, and, therefore, a fit to a unique set of rate constants is unlikely to be obtained for this data set. We, therefore, tested the proposed model by simulating the stopped-flow experiments (6) using the rate constants given above. The results showed that all of the stopped-flow data could be fit to better than a factor of 2. Since the pressure relaxation and stopped-flow experiments were carried out in different laboratories with differing ambient temperatures more detailed analysis would not be meaningful. The lack of precision in the fitting of the experimental results does not detract from the basic conclusion of this work that the presence of two relaxations requires the additional step in the reaction mechanism as shown in Scheme 4. The conclusion that none of the three steps is rate-limiting for the overall process is one which is commonly found for an enzyme which acts as a rapid efficient catalyst. This model suggests that the rate constants for the two second-order processes (kl and L S ) are both approximately lo7 "' s" even though one involves a protein-nucleotide interaction and the other a protein-protein interaction. This value for the rate constant for nucleotide binding to a protein

5 4672 Release of Nucleotide from Elongation Factor Tu can be explained by application of the Smoluchowski equation to determine collision rate followed by an approximation made to allow for steric effects (14). However, current theories are inadequate to explain the interaction between electrically charged macromolecules. Experimentally determined values for the rate constants of protein-protein interactions show a wide variation, but values of greater than lo7 M" s-' have been obtained (15, 16). The mechanism of GDP release from EF-Tu.GDP catalyzed by EF-Ts is likely to be the same as that of thiogdp although the rate constants must differ since the overall equilibrium constant of Scheme 4 is approximately 100-fold different for thiogdp compared to GDP. It has previously been shown that the binding constant of EF-Ts to EF-Tu. thiogdp is similar to EF-Tu-GDP (6), and so the difference is not accounted for by kl or kl. k-3 is a diffusion-controlled step and is likely to be the same for GDP and thiogdp. If k3 accounted for the difference, the overall rate of EF-Ts-catalyzed exchange of GDP from EF-Tu. GDP would be too slow to account for the measured values (4). It is probable, therefore, that & and k-, are the rate constants which are different in the mechanisms of thiogdp and GDP release. Although the kinetic data show the occurrence of an isomerization of the EF-Tu. thiogdp. EF-Ts ternary complex, little information is gained about the nature of this structural change. However, since its equilibrium constant is perturbed by pressure, it involves a change in the exposure of ionic residues of the protein to solvent. The actomyosin- ADP ternary complex also undergoes an isomerization which is sensitive to pressure (17). It has been shown that the single tryptophan residue (Trp-184) of EF-Tu has an increased mobility and solvent accessibility in EF-Tu- EF-Ts compared to EF-Tu. GDP (18). This change probably occurs in the isomerization step although the extent of this structural change is not known. Guanosine nucleotide-binding proteins are now known to be involved in a wide range of biological control systems. The conversion of a protein-gdp complex to a protein-gtp complex is necessary in all of these systems for the occurrence of a biological response, and the initial release of GDP is either the rate-limiting step or dependent on the rate-limiting step (19, 20). The mechanism of EF-Ts-catalyzed release of GDP from EF-Tu.GDP may be typical of this type of process for other guanosine nucleotide-binding proteins. REFERENCES 1. Weissbach, H. (1980) Ribosomes; Structure, Function and Genetics, pp , University Park Press, Baltimore 2. Miller. D. L. & Weissbach., H. (1970).. Arch. Biochem. B~OD~VS. " 14i, Hwane. Y. W. & Miller. D. L. (1985).. J. Bwl. Chem Chau, V., Romero,G. & Biltonen, R. L. (1981) J. Biol. Chem. 256, Romero, G., Chau, V. & Biltonen, R. L. (1985) J. Bwl. Chern. 260, Eccleston, J. F. (1984) J. Bwl. Chem. 269, Eccleston, J. F. (1981) Biochemistry 20, Davis, J. S. & Gutfreund, H. (1976) FEBS Lett. 72, Davis, J. S. (1981) Biochern. J. 197, Engelborgs, Y., Heremans, K. A. H., De Maeyer, L. C. M. & Hoebeke, J. (1976) Nature 269, Eccleston, J. F. & Trentham, D. R. (1977) Biochern. J. 163, Coates, J. H., Criddle, A. H. & Geeves, M. A. (1985) Biochem. J. 232, Edsall, J. T. & Gutfreund, H. (1983) Biotherrnodynamics, John Wiley & Sons, New York 14. Gutfreund, H. (1987) Biophys. Chem. 26, Koren, R. & Hammes, G. G. (1976) Biochemistry 16, Pollard, T. D. (1986) J. Cell Biol. 103, Geeves, M. A. & Gutfreund, H. (1982) FEBS Lett. 140, Jameson, D. M., Gratton, E. & Eccleston, J. F. (1987) Biochernistry 26, Stryer, L. & Bourne, H. R. (1986) Annu. Rev. Cell Biol. 2, Gilman, A. G. (1987) Annu. Reu. Biochem. 66,