The Regulation of Glycogen Phosphorylase a by Nucleotide Derivatives KINETIC AND X-RAY CRYSTALLOGRAPHIC STUDIES*

a function of variable concentrations of nucleotide activators and derivatives. These kinetic data are correlated with the results of crystallographic binding experiments. Two separate binding sites for nucleotide are identified in the phosphorylase monomer. These sites are progressively saturated as the nucleotide concentration is raised and may be distinguished both functionally and physically. The site a stronger for nucleotide corre-sponds that the allosteric activator It also the phosphate, and other The second weaker for and experiments

Initial reaction rate studies of rabbit muscle phosphorylase a (1,4-ru-o Redemann and Niemann (23). Since the product is heat-labile (24), the reaction was kept at room temperature and the reaction time was extended to 72 h. The pentaacetate was isolated from an ice/water mixture as described by Wolfram and Thompson (25). Kinetic Studies-Initial reaction rates were determined in the direction of saccharide synthesis. Two assay methods were used. Method I was a modification of the method of Hu and Gold (26) as described previously (18,27 Enzyme and glycogen were preincubated for 15 min at 30°C before initiating the enzymatic reaction with glucose-l-P in both methods. These assays were used interchangeably without noticeable differences in results. Method I, however, is considerably more sensitive than Method II. X-ray Crystallography-Single crystals of phosphorylase a, suitable for diffraction studies, were prepared as previously described (29). Glutaraldehyde cross-linking of the crystals, when desired, was according to Kasvinsky and Madsen (18). They observed no effect on the crystal structure of the enzyme. Standard buffer for ligand binding studies contained 10 rn~ Bes, 10 rn~ magnesium acetate, and 1 rn~ EDTA, pH 6.8. In most cases, the inhibitor glucose, which is required for the formation of tetragonal crystals, was removed prior to initiating the binding studies. Difference electron density maps at 3.0-A resolution indicate that this removal does not result in any alteration of the protein structure. Crystals of phosphorylase a were soaked in solutions of the various ligands for 3 h. Diffraction data were measured with a Syntex P2,, diffractometer using data acquisition techniques as described before (27,29). The basis for calculation of the 4.5-and 3.0. A resolution difference Fouriers utilizing about one-half the observable data has been reported previously (27). All 6-and 4.5-A resolution analyses by difference Fouriers re uired a single crystal for each ligand.
The 3.0-A AMP and 3.0-1 5.thio-D-glucose-l-P experiments required six and nine crystals, respectively. The Fourier map revealing AMP bound at the nucleotide site was sectioned along the crystallographic Z axis and placed in a Richards box (30) for fitting a Labquip model (1 cm/A) of AMP. The coordinates of the carbon, oxygen, and nitrogen atoms were measured from this model.

Kinetics
-Heterotropic cooperativity between AMP and glucose-1-P has been reported previously for phosphorylase a (6, 28, 31, 32). These papers show that phosphorylase a is a "K" Primary kinetic plots showed no homotropic cooperativity for IMP, AMP, or glycogen at 55 mM glucose-l-P. The apparent dissociation constant for glycogen from the phosphorylase b. glucose-1-P complex varies from 3.9 to 0.5 rn~ as AMP increases from zero to "infinity." However, when IMP is used, the apparent K,, for glycogen is invariant at 2.6 rn~ even though V,,, increases nearly 3-fold. We observed a dissociation constant of 0.15 IIIM for AMP in the absence of glycogen.
This may be compared to the value 3.7 mu for IMP under similar conditions. Previous studies (7)  competitive inhibitor of glucose-l-P binding to phosphorylase 6, we examined the effects of this compound on phosphorylase a. We found that this glucose-l-P derivative was competitive with that substrate (Fig. 6)    0, control, K, apparent = 1.4 mM glucose-l-P, 0, 0.6 mM 5thio-D-glucose-l-P, K,,t apparent = 2.6 rnM glucose-l-P.
K, apparent = 0.3 mM. Assays were carried out by Method I (see "Materials and Methods"). occupied with 5 mM inosine (Fig.  76). The division of the nucleotide binding site into three subsites for the base, sugar, and phosphate rests on the 3-A resolution difference Fouriers for AMP and 5-thio-n-glucose-l-l' which clearly define these subsites. Fig. 8 (Table  I) with the occupancies observed for the two types of compounds (Table  III) suggests that binding of nucleoside derivatives at the nucleoside site defined in this paper is responsible for the competitive inhibition with respect to glucose-l-P. Evidence for this latter conclusion is presented in Fig. 7, c and d. The virtual substrate &thio-D-glucose-l-P, as pointed out earlier, is a competitive inhibitor with respect to glucose-1-P. The difference Fourier map obtained with this inhibitor at 10 mM concentration indicated that 5-thio-n-glucose-l-P in the absence of glucose, unlike glucose-l-P at 50 mM concentration (131, binds at two sites 30 A apart in the phosphorylase monomer. These are the AMP and glucose sites reported earlier show no competition between nucleotide and glucose-l-P (Fig. 1). However, the existence of a second glucose-l-P site at the glucose binding locus (recall that glucose is a competitive inhibitor with respect to glucose-l-P (32)) is consistent with the latter site being part of the active site of phosphorylase a. Binding of glucose-l-P, Pi, and other anionic substrate analogs at elevated concentration (300 mM) in the crystal has been shown at both the AMP and glucose-l-P (glucose) sites. Difference Fourier maps for some of these binding studies have been published (36). It is of great interest that glucose-l-P bound in the active site is adjacent to the pyridoxal 5'-phosphate coenzyme (36). The most striking demonstration that the nucleoside site blocks the active site and a unique demonstration of the location of the active site itself is shown in Fig. 7d where, in the presence of inosine, 5-thio-D-glucose-l-P no longer binds to the glucose-l-P (glucose) binding site. This result is consistent with the kinetic results (Figs. 3, 5, and 6) which indicate that inosine and the nucleoside site to which it binds allosterically exclude glucose-l-P from the active site. Note that 5-thio-Dglucose-l-P still remains firmly bound at the AMP activator site as shown by the positive density peak at the phosphate subsite (Fig. 7d, site P). However, no inosine is bound at the adenine subsite (compare Fig. 7, d with a). Phosphorylase a crystals, obtained in the presence of glucose (291, contain glucose in the active site and are most likely in the inhibited "T" conformation as described by Helmreich et al. (32) 4.5-A resolution than in a 6-A resolution map). In the presence of both substrates maltoheptaose and glucose-l-P (or 5-thio-nglucose-l-P), the entire o( helix (amino acids -50 to 75) in the NH,-terminal region moves inward toward the body of the monomer (data to be presented elsewhere). We consider this movement to be a prelude to activation of the enzyme and formation of the "R" state.

DISCUSSION
The presence of two classes of AMP binding sites per monomer of phosphorylase b has been suggested previously by Wang and collaborators (8,37). Similar conclusions were made more recently by Morange et al. (11). The present study verifies the interpretation of these authors and by the use of kinetic and x-ray crystallographic methods assigns a functional role and physical location to each of the nucleotide sites in the phosphorylase a monomer.
The high affinity AMP binding site (8,11,37) is probably equivalent to our AMP activator site (13). As pointed out earlier, the kinetic and crystallographic results are consistent with two sites per monomer of phosphorylase a. The site having the highest affinity for IMP (or AMP), which is responsible for enzyme activation (Fig. 11, should be fully occupied by nucleotide at relatively low concentrations. We find, at 0.5 mM AMP, that one of the two binding sites (the AMP site previously reported (13)) is essentially fully occupied by ligand (Table III). The second, low affinity nucleoside site (13) is, however, only partially occupied. Kinetic and crystallographic experiments indicate that binding of ligand to the nucleoside site inhibits glucose-l-P binding. The apparent K, values for AMP and IMP binding to the nucleoside site are, respectively, 6.0 and 1.5 mM (Table I). Wang et al. (37) have reported similar values (K,,,, = 3 mM) for the calorimetrically determined binding of AMP to the low affinity site of phosphorylase b.
Two classes of binding sites have also been suggested for adenine and adenosine binding to phosphorylase b (11). Again the kinetic and crystallographic results presented here agree with this finding. It is interesting and not unexpected that the sites which bind the nucleoside or purine bases (Table III) are the same as those for the nucleotides. The occupancy data for the purine bases at these sites, however, is reversed with respect to those obtained with the nucleotides.
Thus, the inhibitory site is preferred by the nucleosides or free bases. The K, obtained by Morange et al. (11) for adenine binding (0.2 mM) and the Kos for adenosine binding (1.1 mM) to phosphorylase b is similar to our K, values for these and analogous compounds listed in Table I. It is interesting that But and collaborators (11) found adenine and adenosine inhibitory with respect to the residual activity of phosphorylase b, which is present in the absence of AMP. Such a result is consistent with the conclusion that the inhibitory site is one other than that for AMP allosteric activation, especially since these compounds do not release radioactive AMP or glucose-6-P bound at the allosteric activator site until the low affinity nucleoside site is saturated (11).
The important unanswered question remains the physiological role for these effector sites and what ligands function in uivo. Effects of purine derivatives on muscle phosphorylase The competition with respect to glucose-l-P, however, is of a direct nature in the case of glucose but is allosteric with the nucleosides or purine bases. The involvement of various metabolites with the regulation of glycogen metabolism is well documented (42). The specific effects of AMP, glucose, and glucose-6-P on phosphorylase phosphatase has been thoroughly investigated by Martensen et al. (43,44). Methylated oxypurines were reported by Wosilait and Sutherland (45) to have an activating effect on phosphorylase phosphatase. This effect, which is exerted via the substrate, has been examined with muscle phosphorylase phosphatase (46,47) and more recently with purified liver phosphatase (48). The similarity between the activation of phosphorylase phosphatase by glucose and theophylline (48) and inhibition by AMP (43)(44)(45)(46)(47)(48) suggests that the nucleoside site also functions as a regulatory site of phosphatase activity in muscle.
This function might be analogous to the glucose regulation of phosphatase in liver tissue (42). The synergistic relationship between the glucose and nucleotide inhibition of phosphorylase (Table II) might support this hypothesis. Structural changes in the NH,-terminaliserine 14 PO, region of the phosphorylase a monomer, which are present in the 5-thio-nglucose-l-P (glucose-free) crystal (Fig. 7c, NT) are blocked in the presence of nucleoside (compare Fig. 7, c and d). These structural changes will be the subject of another manuscript;J however, it will suffice to say here that they are consistent with a role similar to that which has been proposed for glucose.
As intriguing as this suggestion of a physiological function for the nucleoside site may be, we can provide no information as to the identity of the ligand(s) which physiologically binds to this site. Investigations in this area are now underway.
In The location of these sites is shown on a stereo diagram of the high resolution structure. Finally, the third site is 10 A away from the glucose-l-P site, is in the nucleotide-binding fold, and was