AMP and IMP Binding to Glycogen Phosphorylase b A CALORIMETRIC AND EQUILIBRIUM DIALYSIS STUDY*

Reaction microcalorimetry and equilibrium dialysis have been used to study the binding of AMP and IMP to glycogen phosphorylase b (EC 2.4.1.1) at 25 "C and pH 6.9. The combination of both techniques has ena-bled us to obtain some of the thermodynamic parame- ters for these binding processes. Four binding sites were found to be present in the dimeric active enzyme for both and IMP. The binding to two high-affinity sites, which, in our opin- ion, correspond to the activator sites, seems to be co-operative. The two low-affinity sites, which would then correspond to the inhibitor sites, appear to be independent when the nucleotides bind to the enzyme. The negative AGO of bindinglsite at 25 "C is the result in all cases of a balance between negative enthalpy and entropy changes. The large differences in A H and AS" for the binding of AMP to the activator sites (-27 and -70 kJ mol-'; -22 and -150 J.K-' mol-') the existence of rather extensive conformational changes taking place in phosphorylase b on binding with the allosteric activator. Whereas the of AMP for the activator sites is about 1 order of magnitude higher than of IMP, the of in- cluding their AH and to be the same for the inhibitor sites.

Glycogen phosphorylase b (EC 2.4.1.1) is a key enzyme in glycogen metabolism which undergoes a distinctive allosteric activation by AMP. Initial studies (1,2) showed the existence of one AMP site/enzyme protomer. Recent evidence for two AMP-binding sites/monomeric unit of the enzyme has, nevertheless, been provided at 4 "C by equilibrium dialysis (3) and x-ray diffraction studies (4). Support for the presence of these two sites at 25 "C was also supplied by the biphasic thermal titration curves of the enzyme with AMP (5)(6)(7)(8). We have shown, however, in a recent communication (9) that these biphasic calorimetric profiles were due, in our case, to the presence of an impurity, AMP aminohydrolase (EC 3.5.4.6), and that monophasic thermal curves are obtained when phosphorylase b is freed from this impurity (9,10). This result cast doubts on previous experimental evidence for a second AMP site/monomer of phosphorylase b in solution at 25 "C.
In order to perform quantitative thermodynamic analysis of ligand binding to multisubunit proteins displaying site cooperativity, additional equilibrium techniques, besides cal-*This research was supported by a grant from the Comision Asesora from the Spanish government. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$Supported by a fellowship from the Formaci6n de Personal Investigator. orimetry, are required. Even with several techniques, this energetic characterization can often become complicated due to the complexity of the process itself, and not much information of this kind is found in the literature (11)(12)(13)(14)(15).
In this context, we have undertaken the study of the binding of AMP, and also IMP, to phosphorylase b by equilibrium dialysis and microcalorimetry at 25 "C and pH 6.9. Under these conditions, four sites have been found per active dimer of the enzyme for both AMP and IMP, which have been assigned to the two nucleotide or activator sites, N, and to the two nucleoside or inhibitor sites, I (4,16,17). The binding is, in all cases, enthalpy-driven at 25 "C, overcoming a negative entropy barrier.

MATERIALS AND METHODS
Glycogen phosphorylase b was prepared from rabbit skeletal muscle by the method of Fisher et al. (18,19) with the modifications described by Krebs et al. (20). The catalytic activity of the enzyme was determined by the assay of Hedrick and Fisher (21). The preparations used had specific activities of 80-90 units/mg. Protein concentration was determined from absorbance measurements at 280 nm using the absorbance coefficient E:?m = 13.2 (22). The molecular weight of the monomer was taken as 97,400 (23). The enzyme was crystallized at least three times and used within 1 week of the final crystallization. Phosphorylase b preparations were freed from AMP by passing them through a Sephadex G-25 column equilibrated with 50 mM KC1, 0.1 mM (3-mercaptoethanol, 0.1 mM EDTA, 50 mM buffer solution (glycylglycine, glycerophosphate, or Tris), adjusted to pH 6.9. The A2,&m ratio for the AMP-free phosphorylase b solutions was always less than 0.53. Traces of AMP aminohydrolase were eliminated by incubation with alumina Cy, as has been described elsewhere (10).
[5'-"C]AMP and [5'-"C]IMP were obtained from the Radiochemical Center, Amersham, England. AMP, IMP, glycylglycine, alumina Cy, Tris, and (3-mercaptoethanol were purchased from Sigma; sodium glycerophosphate was from Merck, and EDTA was from Fluka. All chemicals used were of the highest available purity. Distilled, deionized water was used throughout. An LKB flow microcalorimeter with a water bath at 25 'C was used for the calorimetric measurements. The temperature in the water bath was controlled by a proportional heater with adjustable precision based on a combination threshold detector and zero-crossing trigger (24). The control of the bath temperature was better than 0.01 "C. Electrical and chemical calibrations were made in the same range as that which we obtained in the calorimetric experiments themselves. The chemical calibration was accomplished by the neutralization of Tris with HCl (25). Enzyme and AMP (or IMP) solutions were allowed to flow into the calorimeter at rates of 7 ml h-l in most experiments, with occasional changes in order to check the completeness of the reaction. The dilution gradient method of Mountcastle et al. (26) was also used in the phosphorylase b/IMP calorimetric titration (see Fig. 5). All appropriate corrections for heats of dilution and mixing were applied. The enzyme activity was routinely checked prior to and after the calorimetric and dialysis experiments. The pH values of the buffer, AMP, IMP, and enzyme solutions were controlled at 25 "C before initiating the binding reaction. The equilibrium dialysis experiments were carried out at 25 "C as described by Helmreich et al. (27).

RESULTS
Binding of AMP to Phosphorylase b-The binding of AMP to phosphorylase b was observed as a function of the activator concentration by equilibrium dialysis at 25°C and pH 6.9 (Fig.  1). The result of this binding is presented in a Scatchard plot in Fig. 2, where Y stands for the saturation fraction and Y = nY for n (number of sites) equal to 4. The shape of the curve which shows a maximum is of the form expected for a system exhibiting positive cooperativity. Extrapolation of the lower part of the plot leads to a value clearly higher than 2 mol of AMP bound per mol of enzyme at saturation. Morange et al. (3), using the same technique, found at 4 "C that there were two classes of AMP-binding siteslsubunit of phosphorylase b, a low affinity and a high affinity site, i.e. four sitesldimer. Similar results were also obtained at 4 "C by Johnson et al.
The shape of the Scatchard plot suggests that AMP starts to bind to the lower-affinity sites when the high-affinity sites are practically saturated. Hence, most of the experimental points at low AMP concentration in Fig. 1 would mainly correspond to the binding of AMP to the high-affinity sites, these being for the most part responsible for the cooperativity shown in the Scatchard plot. Assuming then that the values with v < 2 correspond only to the binding to the two highaffinity sites, a straight Hill plot is obtained for these values with a Hill coefficient of 1.4 _+ 0.1, which agrees well with those obtained by other authors (28)(29)(30)(31) for these sites at similar AMP concentrations.
The reaction of AMP with phosphorylase b can be considered as the binding of AMP to two independent sets of sites. The high-affinity sites show positive cooperativity according to their Hill coefficient, while the low affinity sites can themselves be considered as independent of each other, an assumption that is justified below. On this basis, the saturation fraction, Y , as a function of the free AMP concentration is where K,, stands for the microscopic binding constants at the ith site, and K,, = K,,. K,, and K,, values can be initially calculated from the Hill coefficient, nH, and the concentration of the free ligand at 50% saturation of the high-affinity sites, So.s, obtained from the previous Hill plot, according to the following relations (32).  Table I. Curves in Figs. 1 and 2 are the theoretical ones using the calculated binding constants ( Table I).
The results of the calorimetric titration of phosphorylase b with AMP in three different buffer solutions at 25 "C and pH 6.9 are shown in Fig. 3. The binding of AMP was exothermic in all cases, giving rise to a well-defined monophasic curve. This curve is the same regardless of the buffer system used, and since the heats of ionization of the three buffers are different (34), no proton uptake or release seems to occur in the activator binding to the enzyme, particularly to the highaffinity sites. Fig. 3 shows the heat evolved per mol of phosphorylase b dimer as a function of free AMP concentration in equilibrium with the enzyme. This concentration was calculated by using the binding constants previously obtained in the dialysis experiments ( Table I) Table  I. The dashed line is the theoretical one corresponding to the binding of AMP to the high-affinity sites (see text). Fig. 1. The theoretical curve was calculated using the values listed in Table I. using the values shown in Table I.  The uncertainties are standard errors in the fitting of the curves. where Kc and Km2 are the macroscopic and microscopic binding constants, respectively, AHi stands for the enthalpy values/ mol of AMP bound to each site, and K,, = K,, and AH, = AH, since the two low-affinity sites are considered as being independent.

FIG. 2 (center). Scatchard plot of the data included in
From the above Equation 3, the enthalpy change/mol of enzyme (dimer), AH, is given by + which can be expressed as the following.

+ K,,[S]
The experimental data in Fig. 3 were fitted by trial and error to Equation 5 giving the values for AHH, that minimize the deviations of the observed values of AH from the calculated ones. These AH, values were later optimized by the Newton-Gauss method (see Ref. 33) using the K,, values previously obtained. The curve in Fig. 3 is that corresponding to Equation 5 for the calculated K,, and AH, values. Thermodynamic parameters for the binding of AMP to phosphorylase b at 25 "C and pH 6.9 obtained from the AH, values and the microscopic binding constants are included in Table  I.
Binding of IMP to Phosphorylase b-The binding of IMP to phosphorylase b was also achieved by equilibrium dialysis and calorimetry at 25 "C and pH 6.9 in the same way as described above for the AMP binding. The enzyme concentration for the equilibrium dialysis experiments was 15 mg/ ml, higher than that used for the AMP binding, since phosphorylase b displays a lower affinity for IMP than for AMP. The experimental results for Y and u uersus free IMP concentration are shown in Fig. 4. A Scatchard plot of these values clearly extrapolates to u = 4 (data not shown).
Thermal titration data of the protein by IMP are displayed Fig. 5 for two different buffer systems, Tris and glycylglycine. Experiments using the latter buffer were carried out following the exponential gradient method described by Mountcastle et al. (26). There is no apparent difference between the results for the two buffers and the methods used for the thermal titrations.
The dialysis data of the IMP binding give rise to a practically straight Scatchard plot, which might be taken as evidence for the existence of four identical and independent binding sites. It has been established from x-ray diffraction studies (4), however, that there are two types of IMP sites, namely, the two N and two I sites/dimer. This being the case, the simplest analysis of the data would be to consider that the IMP is binding to two independent sets of sites without any cooperativity. Thus, the saturation fraction is given as On the other hand, the fact of having different sites should give rise to a Scatchard plot showing convexity towards the Xaxis (47), while we have indicated that our plot is practically straight. This means that some factor is counteracting the curvature towards the axis, and this could be explained if there were some positive cooperativity during the binding process (47). In this respect, the binding of IMP to the N sites has been reported by several authors (35-37) as cooperative, as is the case for the AMP binding to those sites. In addition, the binding of several ligands (nucleosides, nitrogen bases, FMN) to the I sites has been shown to be non-cooperative (3, 38), which, as we saw before, seems also to be the case for AMP.
Therefore, the nature of the IMP binding to phosphorylase b seems to be qualitatively comparable to that of AMP. In other words, the overall IMP binding process could be described similarly to that of AMP and, consequently, with the same equation for the saturation fraction as Equation  4 (left). Binding of IMP to phosphorylase b at pH 6.9 and 25 "C as measured by equilibrium dialysis. The enzyme concentration was 75 PM, and the buffer used was the same as in Fig. 1 Table I. FIG . 5 (right). Thermal titration of phosphorylase b with IMP at pH 6.9 and 25 OC. The heat evolved per mol of enzyme is plotted uersus the total IMP concentration. The enzyme concentration was 6.3 mg/ml in 50 mM KCI, 0.1 mM EDTA, and 0.1 mM P-mercaptoethanol. A, 50 mM glycylglycine; .,50 mM Tris and following the exponential gradient method (see text). The solid curue gives values of enthalpy calculated according to an equation similar to Equation 5 using the data listed in Table I. fact that the binding of AMP to the N sites shows higher cooperativity than that of IMP to those sites and also because of the very close affinity of IMP for both the N and I sites, while the AMP affinity for the N sites is more than 1 order of magnitude higher than the AMP affinity for the I sites (Table I).
It should be stressed at this point that it is not possible, when working within a limited range of accuracy, to obtain a unique valid fit of experimental binding data when several different binding sites are involved. It is only through given certain assumptions and accepting certain restrictions that a given set of binding parameters can be arrived at. Furthermore, their physical meaning is not only limited by their standard uncertainties (particularly high in our case for the IMP binding to phosphorylase b) but also by the assumptions and, to some extent, the behavior that these parameters are forced to follow.  (7,8). The results of these latter groups showed clear divergences concerning both the AMP saturation range for the second site, I, and the AH value for this site. We initially obtained biphasic curves for this thermal titration, but, as was shown, it was due to the presence of an enzymic impurity in our phosphorylase preparations, which, once eliminated (lo), gave rise to monophasic calorimetric curves (9). It is not possible, however, to obtain information about the number of sites from calorimetric experiments alone when dealing with allosteric enzymes, as it is in the case of phosphorylase b. The combination of equilibrium dialysis and calorimetric studies enables us now to actually obtain the number of binding sites for AMP and IMP in solution at 25 "C, as well as to characterize the thermodynamic parameters associated on binding both nucleotides. The main difference between the binding of these two activators appears to be in the much closer affinity of IMP for the two types of sites, N and I, in the dimer than in the case of AMP.
Ho and Wang (6) gave an average A H value, -13.2 kcal (mol site)-', for AMP binding to the two N sites of the phosphorylase b dimer a t 25 "C. Twice this value compares well with AH, and AH2 obtained in our study for this site (Table I). These authors detected an association of the enzyme to the tetrameric state upon AMP binding a t 18 "C, while no association seemed to occur a t 25 "C (6). We did calorimetric experiments a t different phosphorylase concentrations (Fig.  6), and there was no detectable change in the heat evolved for the concentration range investigated. Thus, no protein association effects seemed to occur in our experiments.
To our knowledge, only Steiner et al. (39) have tried to identify the contribution of each N site to the enthalpy change on the binding of AMP, although they did not investigate the other binding sites because they only used low AMP concentrations (dl mM). While their values at 23 "C are somewhat different from ours, the total AH for the N sites coincides with ours within experimental uncertainty. This difference also correlates with the comparison between our K,, and K,, values and theirs, in the sense that we obtain a higher homotropic cooperativity as indicated by the Hill coefficient (1.4 in our case, 1.2 in their case). This variation could be attributed to small differences in temperature, pH, and buffer used, although Dreyfus et al. (29) also obtained a Hill coefficient of 1.4 at 20 "C for this binding. An equal value of 1.4 had previously been obtained by Avramovic and Madsen (28) and other authors (30,31).
The binding of AMP and IMP to the N sites shows positive homotropic cooperativity. This effect is more pronounced in the AMP binding ( n H = 1.4) than that of IMP (nH = 1.2). The difference in cooperativity is also seen in the AH values. Thus, in the case of AMP, the AH values are very different, -27 and -70 kJ mol-', favoring the entrance of a second AMP molecule into the dimer. The corresponding ASo values go in an opposite direction, balancing to some extent the enthalpic influence. The binding at 25 "C is clearly of an enthalpic character. The same qualitative situation applies to the IMP binding, although here the A H and ASo values are comparatively lower than in the AMP binding. The differences in AH and ASo values between the two N sites for both nucleotides make for a higher affinity and cooperativity in the binding of AMP than in IMP binding to those sites.
A structural interpretation of these thermodynamic parameters is not possible at this point since these values are the joint product of those of the so-called intrinsic binding (40) and those of the conformational change produced in the macromolecule, which is structurally responsible for the cooperative effect. We are currently working out a method to evaluate both contributions to the overall binding process (41).
The corresponding thermodynamic parameters for the binding of AMP and IMP to the I sites are very close for the two nucleotides within experimental uncertainty (Table I).
This implies that the I site (nucleoside or inhibitor site) does not apparently discriminate between the two nucleotides. This site seems to be more specific for the inhibitors of the enzyme, such as adenine or adenosine, although their binding to the I site is also non-cooperative (3,38). It has been suggested that any ligand bound to this site would produce some enzyme inhibition (17), and we have detected in this case a decrease in the enzymic activity at high AMP concentrations (results not shown). In this context, it has also been suggested that the low activation produced by IMP in comparison to that produced by AMP could be assigned to the close affinity of IMP for both the N (activator) and I (inhibitor) sites (17). Besides this possibility, however, we have seen that the thermodynamic parameters for binding AMP and IMP to the activator sites are different enough to expect different confor-mational changes and, therefore, different final conformations. The differences in these AMP-and IMP-induced conformational changes have been reported by several researchers using various techniques, such as ESR (42), fluorescence (43), and SH reactivity (3). The binding of these nucleotides to the I site is enthalpydriven at 25 "C, overcoming the negative entropy barrier ( Table I). It seems that the enthalpic contribution would mainly come from the interaction between Tyr-612 and/or Phe-285 with the nitrogen bases in the nucleotides (44). The ASo values cannot be easily accounted for since the site structure is not well defined at present. The binding of FMN to this site in phosphorylase a shows a negative AC, value which is itself a function of temperature (38). If it were also the case for our nucleotides, one would expect a change in the sign of A S o at temperatures somewhat lower than 25 "C. This FMN binding shows the same enthalpy value as AMP and IMP, -37 kJ mol-', but a positive change in the entropy is responsible for the higher affinity of this compound for the I site at 25 "C (38).
Finally, the binding of AMP to the I site (which, it has been suggested, may play a possible role in vivo in controlling the enzyme activity ( 7 ) ) does not seem to have much physiological role in muscle given the levels of free AMP in either resting muscle or under extreme fatigue (45, 46) and the affinity of AMP for this site, although our affinity values were obtained in vitro, in the absence of any other physiological effectors and/or substrates.