Yeast Hexokinase. A Fluorescence Temperature-Jump Study of the Kinetics of the Binding of Glucose to the Monomer Forms of Hexokinases P-I and P-I1

Transcription

1 Eur. J. Biochem. 68, (1976) Yeast Hexokinase A Fluorescence Temperature-Jump Study of the Kinetics of the Binding of Glucose to the Monomer Forms of Hexokinases P-I and P-I1 James G. HOGGETT and George L. KELLETT Department of Biology, University of York, Heslington, York (Received February IhlJune 25, 1976) The binding of glucose to the monomeric forms of hexokinases P-I and P-I1 in Tris and phosphate buffers at ph 8.0 in the presence of 1 moll- KCI has been studied using the fluorescence temperature-jump technique. For both isozymes only one relaxation time was observed ; values of z - increased linearly with increasing concentration of free reacting partners. The apparent secondorder rate constant for association was about 2 x lo6 1 mol- s-l for both isozymes; the differences in the stabilities of the complexes with P-I and P-I1 are entirely attributable to the fact that glucose dissociates more slowly from its complex with P-I than P-I1 (= 300 s-l and 1100 s- respectively). Although the kinetic data are compatible with a single-step mechanism for glucose binding, the association rate constant was much lower than that expected for a diffusion-limited rate of encounter. Other mechanisms for describing an induced-fit are discussed. It is shown that the data are incompatible with a slow prior-isomerization pathway of substrate binding, but are consistent with a substrate-guided pathway involving isomerization of the enzyme-substrate complex. The induced-fit hypothesis of enzyme action was proposed by Koshland to account for some of the shortcomings of the template theory [1,2]. In particular, the induced-fit hypothesis was able to explain the low reactivity of water in those enzymic reactions where it might be expected to compete with the acceptor molecule. In his original account, Koshland considered specifically the case of hexokinase, and proposed that the binding of sugars resulted in a conformational change from an inactive to an active form of the enzyme. He also noted that even in the absence of sugars, it must be supposed that the hexokinase would have a finite probability of adopting the reactive conformation, and predicted that hexokinase should therefore exhibit a low ATPase activity. Since then, considerable evidence has accumulated to support this view in studies on yeast hexokinase. Colowick and his co-workers detected nucleoside triphosphatase activity using ATP, ITP and GTP as substrates [3,4]. They found that upon saturation with ATP the rate of hydrolysis was fold lower than the rate of sugar phosphorylation, and that the K, for ATPase activity (4 mmol 1 ~ ) was 40 times Enzyme. Hexokinase (EC ). higher than that of ATP in the hexokinase reaction (0.1 mmol 1-l). Evidence that pentose analogues which lacked the 6-hydroxymethyl group could induce changes in conformation similar to those induced by phosphorylatable hexoses came from the observation [5] that xylose and lyxose strongly promoted the ATPase activity of the enzyme and changed the K, value to that corresponding to the hexokinase reaction. That this activation was related to the mechanism for hexokinase was supported by the fact that the activation constant for lyxose in the ATPase reaction was the same as the inhibition constant for its competitive inhibition of the hexokinase reaction. Womack and Colowick have also shown by direct binding studies that glucose and lyxose promoted the binding of nucleotide [6]. In a series of crystallographic studies [7-111 Steitz and his co-workers have obtained information about the structural changes arising on binding glucose to various crystalline forms of yeast hexokinase P-11. The first crystals isolated (form B-I) disintegrated on soaking with glucose; this fact was in itself a strong indication of a major re-arrangement having taken place. Two other crystalline forms, B-11, in which the asymmetric unit was a dimer, and B-111 in which it was a monomer (although probably

2 348 Yeast Hexokinase Relaxations resulting from enzyme which had suffered some proteolytic degradation during purification) did survive soaking experiments and were found to bind glucose and other sugars [ In a previous paper [12] we have presented evidence that there is a discrepancy between the behaviour of hexokinase P-I1 in solution and in the B-I1 crystal form: whereas in the crystalline phase in the absence of nucleotide, glucose bound only to the lower subunit of an asymmetric dimer, in solution both sites on the dimer were equivalent. Despite this difference, the crystallographic data on B-I1 showed that the binding of glucose and other sugars resulted in major conformational changes which extended over the whole of the subunit backbone to which they bound. A similar finding was obtained at higher resolution with crystal form B-111; in this case, both glucose and a-toluoylglucosamine, a competitive inhibitor, brought about extensive changes on binding, which although not identical were very similar. It is apparent that yeast hexokinase is a very flexible enzyme and that the conformational changes associated with sugar binding are mechanistically important. The kinetics of the binding of glucose to the proteolytically modified enzyme S-I have been investigated indirectly using an isotope-trapping method [13], in which the partition of radioactive label between glucose and glucose 6-phosphate was examined when pre-mixed hexokinase and ['4C]glucose were added to a solution of ATP and a large excess of unlabelled glucose. At ph 7.0 in phosphate buffer (25 mmol 1 ~ ') at 25 "C the rate constant for dissociation of the enzyme-glucose complex was determined to be 58 s-' ; the rate constant for association, calculated either from Kd obtained from rate dialysis experiments, or from values of the catalytic rate constant and K,,, for glucose on the basis that glucose was the first substrate in an ordered mechanism, was found to be about 2 x lo6 1 mol-' s-'. However, the kinetics of glucose binding to proteolytically unmodified enzyme, and the manner in which the conformational change is induced are not yet established; in particular, it is not clear whether the induced fit follows, in the terminology of Koshland [14], a 'prior-isomerization' or a 'substrate-guided' pathway. The two pathways are extreme models of possible mechanisms, the former representing initial isomerization of the enzyme followed by binding to the new conformation, and the latter involving initial binding of the ligand followed by isomerization. In order to investigate this mechanism, we have undertaken a study of the kinetics of the binding of glucose to hexokinases P-I and P-11, taking advantage of the fact that binding is accompanied by quenching of the intrinsic protein fluorescence [12]. In this paper we report the results of fluorescence temperature- jump experiments with the monomeric forms of the proteolytically unmodified P-I and P-I1 isozymes. EXPERIMENTAL PROCEDURE Reagents and Enzyme Potassium chloride, tris(hydroxymethy1)aminomethane, D( +)-glucose and potassium dihydrogen orthophosphate were of analytical grade (Merck). The details concerning all other reagents, the preparation of hexokinase isozymes, enzyme assay, measurement of enzyme concentration and fluorescence titrations are given in an earlier publication [12]. Kinetic Measurements The temperature-jump measurements were carried out on the apparatus described fully elsewhere [15, 161. A semi-micro cell (1.2 cm3) was used, and solutions were degassed briefly before being introduced into the cell. The initial temperature was "C and usually jumps of 7 "C were applied by discharging a 52.5-nF capacitor charged to the appropriate voltage. The results described in this paper relate to results carried out at an ionic strength of 1.0 mol l-', under which conditions the rise time for heating was <2 ps. Between experiments a period of 3 min was found to be sufficient to ensure equilibration to the original temperature. The light beam from a highly stabilized mercury-xenon lamp (Hanovia, either 200 W for hexokinase P-I, or 5000 W for hexokinase P-11) was passed through a high-intensity double monochromator (Schoeffel GM 250) fitted with vertical slits of 4-mm slit width; in some experiments on hexokinase P-I with a 200-W lamp only one monochromator was used. The fluorescence of the enzyme was excited at 285 nm and observed through a combination of a WG 320 and a UG 11 filter (Schott) by two photomultipliers (EM QB) set perpendicular to the incident beam. When using the 5000-W lamp, which was necessary to achieve a suitable signal-to-noise ratio for experiments with hexokinase P-11, care was taken to minimise the time of exposure of the sample to light; under favourable conditions this time could be limited to 3 s. The results were analysed by comparing the observed relaxation with a trace simulated using an an analogue computer. The multiplier signal following a temperature-jump was stored on a digital transient recorder (datalab DL 905) ; following triggering, which occurred when the capacitor began to discharge into the cell, the amplitude was held at zero electronically for a period corresponding to about 5 times the heating time of the solution in order to suppress the large unspecific decrease in fluorescence

3 J. G. Hoggett and G. L. Kellett 349 which accompanied heating. The stored signal was displayed continuously by one beam of a double-beam oscilloscope; where necessary a low-frequency-band pass between the transient recorder and the oscilloscope was used to reduce the noise on the display. The second beam of the oscilloscope displayed the time course simulated by an analogue computer (Telefunken RA 742). The computer was programmed to supply up to three exponential functions together with a baseline correction to compensate for cooling where necessary. RESULTS Equilibrium Measurements In solutions at ph 8.0 with an ionic strength of 1 mol 1-' hexokinases P-I and P-I1 [12] exist almost exclusively as monomers. Fig. 1 and 2 illustrate fluorescence titrations with glucose of P-I and P-I1 respectively in both Tris. HCl and phosphate bueer at about 10 "C under these conditions; as expected, binding could be described by a single binding constant in every case. There was no difference between the two buffers at this ionic strength, but binding to P-I (Kd = 1.2 x lop4 mol I-') was found to be stronger than to P-I1 (& = 3 x moll-'). Kinetic Meusuvernenis No relaxation effect was observed in experiments with enzyme alone. In experiments with mixtures of enzyme and glucose only one relaxation time could be resolved; an example of the oscilloscope trace is given in Fig. 3. Relaxation times were measured over I I I ~ 10~" 1 0 ~ ~ 10~2 [~ucose]/mo~ 1.' Fig. 1. Fluoresccwcc tirrarions oj lwxokitzasc P-I with ghco.scj (11 10 'C pff 8.0. (0) [Tris. HCI] = 50 mmol I ', [KCI] = I in01 1- ' ; (A) [potassium phosphate] = 50 minol I-', [KCI] = 1 n~ol 1 10" lo-* 1 0 ~ ~ 10~2 10" [Glucos~l imol c' Fig. 2. Fluorescence riiraiions of' lwxokinase P-II with glucose at 10 'CpH 8.0. Symbols as in Fig. 1 Fig. 3. Oscilloscope trace qf'the relaxation observed on a ternperature-,jump perturbation ojthe binding QJ glucose to Arxokinase P-I1 in Tris. HCI (ph 8.0, 50 nimol 1-') [KCI] = 1.0 rnol I-'. Time axis 2 ms cm-', rise time filter 50 ps; total signal 4.0 V, sensitivity 2 mv cm-'. The smooth trace gives the fitted time from an analogue computer

4 350 Yeast Hexokinase Relaxations / o I 01 I I I I I I O ([E]+[G])/mmoi 1.' Fig.4. Dependence of the reciprocal relaxation times (0) and amplitudes ( x ) for the binding of glucose to hexokinase P-I. [Potassium phosphate] = SO mmol I-' ph 8.0, [KCI] = 1.0 mol I-' 1 I I I I ([El+ [G])/mol I-' Fig. 6. Dependence ofthe reciprocal relaxation times (0) and amplitudes ( x )for the binding of glucose to hexokinase P-II. [Potassium phosphate] = 50 mmol I-' ph 8.0, [KCI] = 1.0 mol I-' 2.0 > D m -.- a " o o 2.5 ([GI +[E])/mol K' Fig.S. Dependence OJ the reciprocul reluxution times (0) and amplitudes ( x ) for the binding of glucose to hexokinase P-I. [Tris. HCI] = SO mmol I-' ph 8.0, [KCI] = 1.0 mol I-' 0 " o o 2.5 ([G]+[E])/mol r' Fig.1. Dependence of the reciprocal relaxation times (0) and amplitudes ( x ) for the binding of glucose to hexokinase P-II. [Tris. HCI] = 50 mmol I-' ph 8.0, [KCI] = 1.0 mol I-' Table 1. Rate and equilibrium constants for glucose binding to hexokinase Phosphate buffer contained 50 mmol I-' potassium phosphate, ph 8.0, 1 mol I-' KCI; Tris. HC1 buffer contained 50 mmol I-' Tris, ph 8.0, 1 mol I-' KCl P-I P-I1 Phosphate 1.8 (k0.2) 320 (k 20) 1.7 (f0.3) 1.5 Tris. HCI 1.6 (k0.2) 250 (k 30) 1.5 (k0.2) 1.5 Phosphate 2.6 (k0.3) 1100 (+loo) 4 (*I) 3 Tris. HCI 1.8 (k0.3) 1150 (+ 150) 6 (k2) 3 a range of enzyme and glucose concentrations, and the dependences of the times and amplitudes upon ([El + [GI) (where [El = enzyme concentration and [GI = glucose concentration) are illustrated in Fig. 4 and 5 for P-I, and in Fig.6 and 7 for P-11. Within experimental error the dependences of the relaxation times were linear and showed no evidence of the onset of saturation with increasing concentrations. However, because the amplitudes of the effects were small, and the relaxation times were very fast for fluorescence work, particularly in the case of hexokinase P-11, the highest concentration of free reacting partners where measurement could still be made was at most ten times the dissociation constant for glucose binding. Values of the apparent forward and reverse rate constants, taken from the slopes and

5 J. G. Hoggett and G. L. Kellett 351 intercepts respectively of Fig. 4-7 are collected in Table 1, together with the derived equilibrium constants, and for comparison, those obtained from the fluorescence titrations. DISCUSSION The kinetic results would appear to be consistent with the simplest possible mechanism in which binding occurs in a single step. + E + G + ~ E G. k The work outlined in the Introduction regarding binding in the crystal phase, and the ATPase activity of hexokinase, indicates that a conformational change does occur upon binding, therefore according to the above mechanism it would follow that the first association of glucose with the enzyme and the resulting conformational change occur simultaneously. The characteristics expected for this kinetic scheme are well known: only one relaxation time is observed which varies linearly with the total concentration of free enzyme and free glucose, increasing without limit as the concentration increases. The relaxation time is given by the following expression : T-' = Z ([El + [GI) + L. The rate constants for association (Table 1) were the same in all cases within the limits of error, and the difference in the stabilities of the complexes with P-I and P-I1 are entirely attributable to the fact that glucose dissociates more slowly from its complex with P-I than from P-11. It is also clear from Table 1 that there is little, if any, difference in the rate constants obtained in Tris and phosphate buffer. This implies that for the proteolytically unmodified P forms of the enzyme, phosphate has no specific effect on binding to the monomer. It has been claimed, since phosphate promotes the binding of glucose to the proteolytically modified S forms, that phosphate can affect the binding in a way which is not linked to the state of association of the enzyme [17]. However, this view has been questioned [12] on the grounds that the limiting value of the dissociation constant for binding to the monomeric form of P-I1 in Tris buffer at ph 7.0 was the same as that measured for S-I1 in phosphate buffer but not in glycyglycine; the alternative possibility was raised that the weaker binding to S-I1 in the latter buffer may have arisen through partial association of the enzyme at the high concentrations used. Values of the dissociation constants for glucose binding to monomeric P-I determined in the present work (1.2 x moll-' at 10 "C in Tris. HCI and phosphate buffers at ph 8.0 in the presence of 1 mol 1-' KCl) can be compared with previously reported values for S-I at ph 7.0 of 4.7 x mollp1 in glycylglycine buffer and 6.1 x mol I-' in phosphate buffer from equilibrium dialysis experiments [ 191, and 3.2 x lop5 moll-' [6] and 2.2 x moll-' [13] from rate dialysis experiments in phosphate buffer. The apparently stronger binding of glucose to S-I than to monomeric P-I in phosphate buffer may be due to the differences involved in the ionic strength, ph and temperature; we are extending our investigations of glucose binding to low concentrations of F'-I at lower ph and ionic strength in an attempt to obtain information on the birtding to proteolytically unmodified monomer under these conditions. In comparing the kinetic results obtained by the isotope-trapping method [13] (k = 2 x lo6 1 mol-' s-', = 58 s-') with the present results for P-I (Table 1), it is evident that the agreement in the values of the association rate constant is excellent, especially considering the very different methods used to obtain them. The difference in the dissociation rates is a factor of about 4-5, reflecting almost exactly the difference in the dissociation constants for glucose binding discussed above. Although the kinetic data are compatible with a single-step mechanism, the difficulty facing such a simple description is that the measured rate of association, zz 2 x lo6 1 mol-' s-', was very much lower than the expected diihsion-limited rate of encounter of a small uncharged substrate molecule with the active-site area of an enzyme, which might in the present case be expected to be about 1-5 x lo8 1 mol-' s-' [19]. This consideration suggests that it would be worth examining whether the data can be accommodated within other reasonable mechanistic models. The justification for discussing necessarily more complicated mechanisms rests on the supposition that conceptually some kind of interaction would be expected to occur upon encounter between enzyme and substrate, and om the experimental considerations that not all interactions need be associated with spectroscopic changes, and that in any case the relaxation effects observed in this work approach the present instrumental limits for fluorescence temperature-jump equipment in terms of sensitivity and time resolution ; indeed the present experiments with P-I1 were only made possible by using an extremely highintensity light source. One possible explanation of the low rate of association is that binding occurs exclusively to a form of the enzyme which is a niinor species at ph 8 as a result of an unfavourable acid-base equilibrium ; binding would then be coupled to a fast protonation reaction. However such mechanisms predict that the binding of glucose would be strongly ph-dependent, which is not the case; glucose binds to the monomeric form of hexokinase P-I1 equally strongly at ph 7 and 8 [12].

6 352 Yeast Hexokinase Relaxations The term 'induced fit' as originally used [1,14] referred to the indwtion by a ligand of a new enzyme conformation not present in significant amounts in its absence, irrespective of the kinetic pathway by which the conformational change takes place. The process may be represented by the following scheme, in which E and E* refer to the predominant conformations in the absence and presence of ligand respectively. E+G*EG.;*/pi: kl' k*,jrkrz E* + G 3 E*G Two extreme pathways are generally recognised ; substrate-guided, in which the initial binding of ligand is followed by isomerisation, E + G 2 EG 'ke*g k2i ku (1) and prior isomerization, where isomerization precedes binding. ki2: E + G$E* + GsE*G. kit Of course other mechanisms may be envisaged, involving either both pathways simultaneously or enzyme conformations other than E or E*, but our data do not warrant such extended discussion. The relaxation equations relating to mechanisms (1) and (2) are well known [20,21]; in each case there are two relaxations which can be treated as uncoupled if the times are well separated so that one relaxation time is associated with each step. Within the present discussion we consider that the binding of glucose to the two conformations of the enzyme represent direct associations occurring at the rate of encounter, and that the isomerizations are relatively slow. To assume that the isomerizations are much faster than the binding steps leads to first-order rate constants which are unrealistically large for such major conformational changes. Thus, if in mechanism (1) the binding is very much faster than isomerization (z1 < z2) then: k;i kii (2) 5;l = k21 + ([El + [GI) (1.1) where K = k21/k~2. The corresponding equations for mechanism (2) with the assumption that the prior isomerization is slow compared with binding (zl 9 52) are: in which Ki = k&/k&. The observed relaxation time was too slow for a direct encounter-controlled association in either model and must therefore be attributed to the isomerization step. Upon increasing the concentration of glucose a qualitative difference between the two mechanisms is immediately apparent: in mechanism (2) the reciprocal of the relaxation time for isomerization would be expected to decrease from ki2 + k ;~ when [GI < ([E*] + Ki) to ki2 when [GI B ([E*] + Ki), whereas in mechanism (1) an increase from k32 to a limiting value of k32 + k23 would be expected on increasing the concentration of glucose. Therefore the fact that 5-I was found to increase with increasing concentration immediately excludes slow prior isomerization as a possible mechanism in the present case. In the substrate-guided mechanism 7;' would be expected to increase from k32 when KI 9 [El + [GI to k ~3 + k32 when K1 < [El + [GI, showing a hyper- bolic dependence upon [El + [GI. There was no evidence of any curvature or saturation in the plots of z-' up to a concentration about ten times the association constant for glucose binding. However, the model is consistent with the data if it is assumed that the first interaction is very weak and that the equilibrium between EG and E*G lies overwhelmingly on the side of E*G; from the concentration dependences in Fig. 4-7, a lower limit of 10 can be set for the ratio k23/k32, although because no curvature at all was observed in the plots, the ratio is likely to be much higher than this. Provided that the first association is fast and can be treated as a pre-equilibrium, it follows from Eqn (1.2) that the limiting slope of the dependence of t;' upon [El + [GI tends to k23. klz/k21 at low concentrations. If a value of 2x lo8 1 mo1-i s-' is adopted for k12, the rate of encounter of glucose with the active-site area of the enzyme, then from the value of ft taken from Table 1 it follows that k23/k21 is of the order of (A steady-state assumption for EG would lead to the expression: k' = k12. k23/ (k21 + k23). The fact that the ratio z/k12 was w shows that k23 6 k21, and justifies treating the first association as a fast pre-equilibrium.) Since from Eqn (1.1) the lowest value of the relaxation time for the direct association is k;' at low values of [El + [GI, and k23 is greater than z 3 x lo3 s-' and 9 x lo3 sc1 for hexokinases P-I and P-11 respectively, it is clear that such a fast process would not have been detected in our experiments, even if it were associated with a change in the intrinsic protein fluorescence. A fluorescence change concomitant with weak first association might be expected to occur if there were a tryptophan residue in the close vicinity of the active site; unfortunately, although the crystal structure of proteolytically unmodified hexokinase P-I1 has been investigated at low resolution ,

7 J. G. Hoggett and G. L. Kellett 353 the amino acid sequences and detailed tertiary structures of both isozymes are unknown at present. If the concentration of EG is negligible compared with that of E*G, then the dissociation constant for glucose binding measured in equilibrium experiments is given by the following expressions : As discussed above, with the assumption of fast initial association, the measured value of is equal to k23 k12/k21 ;+he_nce the dissociation constant is given by the ratio klk. In Table 1 it is shown that there is good agreement between the values of the dissociation constants determined from equilibrium measurements and from the corresponding values of the rate constants. The experimental data presented in this paper are incompatible with a slow prior isomerization mechanism for glucose binding. Unfortunately, experimental limitations arising from the small changes in fluorescence and the fast relaxation times, preclude a decision between the substrate-guided mechanism discussed and a single-step bimolecular association, purely on the basis of the number of relaxation times observed and their dependence upon concentration. Of the two, we prefer the substrate-guided mechanism on the grounds that the observed relaxation is too slow to be attributed to an encountercontrolled association. It is possible that useful information on this point could be obtained from magnetic resonance measurements, which appear to give information about primary rapid associations rather than slower conformational changes detectable by other methods [22]. This work has been supported by the award of an SRC Research Fellowship (to J.G.H.). The kinetic work described in this paper was carried out during the course of an EMBO short-term fellowship (to J.G.H.) at the Medizinische Hochschule, Hannover. We are very grateful to Prof. G. Maass and his colleagues, particularly Drs D. Riesner and C. Urbanke, for their assistance and hospitality during this visit. REFERENCES 1. Koshland, D. E. (1959) in The Enzymes (Boyer, P. D., Lardy, H. & Myrback, K., eds) vol. I, pp , Academic Press, New York. 2. Koshland, D. E. (1958) Proc. Nut1 Acad. Sci. U.S.A. 44, Trayser, K. A. & Colowick, S. P. (1961) Arch. Biockmni. Biophys. 94, Kaji, A. & Colowick, S. P. (1965) J. Bid. ( hem. 240, DelaFuente, G., Lagunas, R. & Sols, A. (1970) Lzu. J. Biochem. 16, Colowick, S. P. & Womack, F. C. (1969) J. Biol. Cliem. 244, Steitz, T. A. (1971) J. Mol. Biol. 61, Steitz, T. A,, Fletterick, R. J. & Hwang, K. J. (1973) J. Mol. Biol. 78, Anderson, W. F., Fletterick, R. J. & Sleitz, T. A. (1974) J. Mol. Biol. 86, Anderson, W. F. & Stcitz, T. A. (1975) J. Mol. Bid. 92, Fletterick, R. J., Bates, D. J. & Steitz, T. A. (1975) Prw. Nut/ Acud. Sci. U.S.A. 72, Hoggctt, J. G. & Kellett, G. L. (1976) Eur. J. Bioclwm. 66, Rose, I. A,, O Connell, E. L., Lilwin, S. & Bar Tana, J. (1974) J. Biol. Chem. 249, Koshland, D. E. & Neet, K. E. (1968) Annu. Rev. Biodwm. 37, Coutts, S. M., Riesner, D., Romer, R., Rabl, C. R. & Maass, G. (1975) Biopkys. Chem. 3, Riglcr, R. & Ehrenberg, M. (1973) Q. Rev. Bioplzys. 6, Colowick, S. P. (1973) in The Enzymc,.s, vol. IX, (Boyer. P. D., ed.) 3rd cd, pp. 1-48, Academic Press, New York. 18. Gazith, J., Schultze, I. T., Gooding, R. H., Womack, F. C. & Colowick, S. P. (1968) Ann. N. Y. Acud. Sci. /S/, Eigen, M. & Hammes, G. G. (1968) Adv. Enzynzol. 25, I. 20. Eigen, M. (1968) Q. Rev. Biophys. 1, Halford, S. E. (1972) Biochem. J. 126, Lanir, A. & Navon, G. (1971) Biochemistry, 10, J. G. Hoggett and G. L. Kellett, Department of Biology, University of York, Heslington, York, Great Britain, YO1 5DD