Structural Basis of the Synergistic Inhibition of Glycogen Phosphorylase a by Caffeine and a Potential Antidiabetic Drug

Archives of Biochemistry and Biophysics Vol. 384, No. 2, December 15, pp. 245–254, 2000 doi:10.1006/abbi.2000.2121, available online at http://www.idealibrary.com on

Structural Basis of the Synergistic Inhibition of Glycogen Phosphorylase a by Caffeine and a Potential Antidiabetic Drug Katerina E. Tsitsanou, Vicky T. Skamnaki, and Nikos G. Oikonomakos 1 Institute of Biological Research and Biotechnology, The National Hellenic Research Foundation, 48 Vas. Constantinou Avenue, Athens 11635, Greece

Received June 9, 2000, and in revised form September 20, 2000

Caffeine, an allosteric inhibitor of glycogen phosphorylase a (GPa), has been shown to act synergistically with the potential antidiabetic drug (ⴚ)(S)3-isopropyl 4-(2-chlorophenyl)-1,4-dihydro-1-ethyl-2methyl-pyridine-3,5,6-tricarboxylate (W1807). The structure of GPa complexed with caffeine and W1807 has been determined at 100K to 2.3 Å resolution, and refined to a crystallographic R value of 0.210 (R free ⴝ 0.257). The complex structure provides a rationale to understand the structural basis of the synergistic inhibition between W1807 and caffeine. W1807 binds tightly at the allosteric site, and induces substantial conformational changes both in the vicinity of the allosteric site and the subunit interface which transform GPa to the Tⴕ-like state conformation already observed with GPa– glucose–W1807 complex. A disordering of the N-terminal tail occurs, while the loop of polypeptide chain containing residues 192–196 and residues 43ⴕ– 49ⴕ, from the symmetry related subunit, shift to accommodate W1807. Caffeine binds at the purine inhibitor site by intercalating between the two aromatic rings of Phe285 and Tyr613 and stabilises the location of the 280s loop in the T state conformation. © 2000 Academic Press Key Words: glycogen metabolism; diabetes; glycogen phosphorylase; synergistic inhibition; inhibitor site; caffeine; crystal structure.

muscle, glucose-1-P is utilized via glycolysis to generate metabolic energy, and in the liver it is converted to glucose. GP is an allosteric enzyme whose activity is primarily controlled by reversible phosphorylation of Ser14 of the dephosphorylated enzyme (GPb) to form the phosphorylated enzyme (GPa). Allosteric effectors, such as AMP, ATP, glucose-6-P, glucose, and caffeine can promote the equilibrium between a less active T state and a more active R state, the structures of which have been characterized (1–3). Because of its central role in the regulation of glycogen metabolism, GP has been exploited as a molecular target for structure-assisted design of compounds that might prevent unwanted glycogenolysis under high glucose conditions that may be relevant to the control of diabetes (4 –12). GP contains at least six potential regulatory sites: the catalytic site that binds the substrates glycogen and glucose-1-P, glucose and glucose analogues, the inhibitor site, which binds caffeine and related compounds, the Ser14-P recognition site, the allosteric site that binds AMP, IMP, ATP, and glucose6-P, the glycogen storage site, and a novel allosteric site situated at the dimer interface (Fig. 1). Positive homotropic, positive heterotropic, and negative heterotropic effects [in the terminology of Monod et al. (13)] have been observed between the various binding sites (1). The catalytic site has been probed with glucose and glucose analogues inhibitors (4 – 8, 10, 14) designed on

Glycogen phosphorylase (GP) 2 catalyzes the degradative phosphorolysis of glycogen to glucose-1-P. In 1

To whom correspondence and reprint requests should be addressed. Fax: ⫹301-7273758. E-mail: [email protected]. 2 Abbreviations used: GP, glycogen phosphorylase, 1,4-a-D-glucan: orthophosphate a-glucosyltransferase (EC 2.4.1.1); GPb, glycogen phosphorylase b; GPa, glycogen phosphorylase a; PLP, pyridoxal 0003-9861/00 $35.00 Copyright © 2000 by Academic Press All rights of reproduction in any form reserved.

5⬘-phosphate; glucose, a-D-glucose; Glucose-1-P, a-D-glucose 1-phosphate; Glucose-6-P, Glucose-6-phosphate; W1807, (⫺)(S)-3-isopropyl 4-(2-chlorophenyl)-1,4-dihydro-1-ethyl-2-methyl-pyridine-3,5,6tricarboxylate; CP320626, 5-chloro-1H-indole-2-carboxylic acid [1-(4fluorobenzyl)-2-(4-hydroxypiperidin-1-yl)-2-oxoethyl]amide; BES, N,N-bis(2-hydroxy-ethyl)-2-aminomethane sulphonic acid; DTT, dithiothreitol; RMSD, r.m.s. deviation, root-mean-square deviation. 245

246

TSITSANOU, SKAMNAKI, AND OIKONOMAKOS

FIG. 1. A schematic diagram of the GPa dimeric molecule viewed down the twofold axis for residues 5 to 838. The positions are shown for the catalytic, inhibitor, allosteric activator, and the new allosteric inhibitor site. The catalytic site, marked by glucose (shown in balland-stick representation), is buried at the center of the subunit accessible to the bulk solvent through a 15 Å long channel. Glucose, a competitive inhibitor of the enzyme that also promotes the less active T state through stabilisation of the closed position of the 280s loop (shown in white), binds at this site. The allosteric site, which binds the allosteric activator AMP and the allosteric inhibitor W1807 (shown), is situated at the subunit–subunit some 30 Å from the catalytic site. The phosphorylated site, contained in the N-terminal tail, is also located at the subunit–subunit interface some 12 Å from the allosteric site. The inhibitor site, which binds purines, such as caffeine (shown), is located on the surface of the enzyme some 10 Å from the catalytic site and, in the T state, obstructs the entrance to the catalytic site tunnel. Two aromatic residues, Phe285 and Tyr613 (shown in white), define this binding site. Ligands that bind to this site are stabilized by intercalating between the side chains of Phe 285 and Tyr 613. Occupation of this site stabilises the T state conformation of the enzyme and blocks access to the catalytic site, thereby inhibiting the enzyme in synergism with glucose at the catalytic site. The new allosteric inhibitor site (12), located inside the central cavity formed on association of the two subunits, binds CP320626 molecule (shown) and is some 15 Å from the allosteric effector site, 33 Å from the catalytic site, and 37 Å from the inhibitor site.

the basis of information derived from the crystal structure of T state GPb– glucose complex. The inhibitor site shows great diversity: purines such as adenine and caffeine, nucleosides such as adenosine and inosine, nucleotides, e.g., AMP, IMP, and ATP, NADH and certain related heterocyclic compounds such as FMN (flavin mononucleotide) have been shown to bind at this site (15–17) in muscle GPb and GPa, but liver GPa shows a more stringent selectivity for inhibitors (15); the compounds can inhibit enzyme activity synergistically with glucose. The allosteric site has been shown

FIG. 2. (A) Synergistic inhibition of GPa by W1807 and caffeine. Kinetic data obtained with 5 ␮g/ml GPa in the absence of AMP, at constant concentration of glycogen (1% w/v) and substrate glucose-1-P (10 mM) and various concentrations of inhibitor W1807 (10, 20, 40, 60, 80, 120, and 250 nM) in the absence of caffeine (E) (IC50 ⫽ 76.1 ⫾ 2.1 nM) or with caffeine 0.1 mM (F) (IC50 ⫽ 41.0 ⫾ 0.6 nM), 0.2 mM (‚) (IC50 ⫽ 28.9 ⫾ 1.7 nM), 0.4 mM (Œ) (IC50 ⫽ 18.0 ⫾ 0.9 nM) and 1 mM (䊐) (IC50 ⫽ 6.9 ⫾ 2.7 nM). (■) Kinetic data obtained with various concentrations of caffeine (0.05, 0.1, 0.2, 0.4, 1, and 2 mM) in the absence of W1807 (IC50 ⫽ 0.45⫾ 0.01 mM). The data are plotted as percent inhibition vs concentration of compound. (B) The effect of caffeine on the potency of W1807. IC50 values for GPa inhibition were determined in the direction of glycogen synthesis as described in Experimental Procedures with 5 ␮g/ml enzyme at constant concentrations of glucose-1-P (10 mM) and glycogen (1%), and varied W1807 concentrations (10–250 nM) in the absence or presence of various concentrations of caffeine. The IC50 values were as in (A). The normalised values (obtained by dividing these values by the IC50 value obtained in the absence of caffeine) are plotted as a function of caffeine concentration.

FIG. 3. Stereo diagram of the electron density of the final weighted 2F o–F c. A close up showing the bound caffeine molecule (orientation I) at the inhibitor site. The contour levels correspond to 1 r.m.s. deviation of the map.

CAFFEINE AND W1807 BINDING TO PHOSPHORYLASE a

to bind the compound W1807; W1807 exhibits blood glucose lowering effects in rats (18), is the most potent inhibitor known of GP (K i ⫽ 1.6 nM for GPb) (9) acts in synergy with glucose (11). CP320626 has been shown to produce marked glucose lowering in diabetic ob/ob mice, to be a potent inhibitor of liver GPa (with an IC 50 in the nM range) and to act synergistically with both glucose and caffeine (19, 20); the compound binds to a new allosteric site (Fig. 1), located at the subunit interface in the region of the central cavity of the dimeric structure (12). In addition to these inhibitors, another potent inhibitor of muscle and liver GPa, 1,4-dideoxy1,4-imino-D-arabinitol (DAB), has been recently identified (21, 22); DAB is a very potent inhibitor of liver cell glycogenolysis, it displays antihypergycemic effect in obese mice, and, despite its structural analogy to glucose, it is not synergistic with caffeine (21, 22). Given the potential importance of the detailed interactions of various ligands specific for each binding site and the fascinating allosteric properties of GP in the structure-assisted and development of drugs that can improve existing therapies of type II diabetes mellitus, we present a detailed description of the regulation of GPa activity by caffeine and the compound W1807. We demonstrate through kinetic studies a strong synergistic inhibition of GPa by the pair W1807/caffeine. To investigate this, we cocrystallized the GPa– caffeine– W1807 complex and determined the structure at 2.3 Å resolution. The structural details of the ligand interactions with their binding sites are presented. The results show that both caffeine and W1807 promote the less active T state conformations, thereby explaining their synergistic inhibition.

247

Refinement. The structure of the 100K GPa– caffeine–W1807 complex cryoprotected with 30% glycerol was solved by difference Fourier methods, starting with the published 100K GPa– glucose– W1807 structure (11), and refined using bulk solvent corrections and standard building and refinement protocols (27–29) to give R-factor ⫽ 21.0% (R free ⫽ 25.8%). The final model contained residues 5–250, 260 –314, 325– 835, and 704 water molecules. A Luzatti plot (30) suggested an average positional error of approximately 0.26 Å. The average B-factors for main chain, side-chain atoms, PLP, W1807, caffeine, glycerol, Ser14-P, and water molecules were 28.1, 29.6, 21.4, 26.3, 32.1, 23.8, ⬎100.0, and 35.5 Å 2, respectively. Residues where overall 具B典-factors values (given in parentheses) exceed 60Å 2 include 5–26 (93.5 Å 2), 209 –211 (67.8 Å 2), and 831– 835 (74.5 Å 2). The Ramachandran plot (31) shows 90.0% of residues in the most favored regions. Structure comparisons. The structures were analyzed with the graphics program O (27). Solvent-accessible areas have been calculated for caffeine in isolation and when bound to provide an estimate of the molecular surface area involved in binding to the enzyme, by using the program NACCESS (32). GP structures were superimposed over well-defined residues using LSQKAB (33) and subunit rotations were determined following the method of Sprang et al. (34). Superposition of the 100 K GPa– glucose and GPa– caffeine–W1807 structures over regions 5–250, 261–313, and 325– 835 gave RMSDs of 0.378, 0.390, and 0.775 Å for C␣, main chain, and side chain atoms, respectively, indicating no large conformational change associated with ligand binding. The major shifts for C␣ atoms are for residues 38⬘ to 46⬘ (0.4 – 0.6 Å) from the cap⬘ region, residues 47⬘ to 75⬘ (0.5– 0.8 Å) from the ␣2 helix, and residues 184⬘ to 196⬘ (0.4 – 0.6 Å) from the loop following the ␤6 strand and from the short strand ␤7. These shifts affect the subunit-subunit interface in the region between the cap⬘, ␣2 helix and the loop between ␤7 and ␤8 strands of the symmetry-related subunit. Coordinate sets for comparison were: 100 K GPa– glucose complex [PDB code 2GPA, (11)] and 100 K GPa– glucose–W1807 [PDB code 3AMV, (11)]. Coordinates for T state 100K GPa– caffeine–W1807 complex have been deposited with the RCSB Protein Data Bank (http://www. rcsb.org/) (PDB code 1C8L).

RESULTS AND DISCUSSION

Synergistic Inhibition EXPERIMENTAL PROCEDURES Kinetics. GPb was isolated from rabbit skeletal muscle according to Fischer and Krebs (23) using 2-mercaptoethanol instead of Lcysteine and recrystallized at least four times. GPa was prepared, recrystallized, and assayed as described (11). Protein concentration was determined from absorbance measurements at 280 nm using an absorbance index A 1% 1cm ⫽ 13.2 (24). Glucose-1-P (dipotassium salt), AMP, (oyster) glycogen, and other chemicals were obtained from Sigma Chemical Co. Glycogen was freed of AMP by the method of Helmreich and Cori (25). Crystallization and data collection. Crystals of GPa– caffeine– W1807 complex were grown in a medium consisting of 29.1 mg/ml enzyme, 0.9 mM W1807, 4.9 mM caffeine, 3 mM DTT, 10 mM BES, 0.1 mM EDTA, 0.02 % sodium azide, pH 6.7. Needle-like rods in space group P4 32 12 with unit cell dimensions a ⫽ b ⫽ 126.7 Å, c ⫽ 115.3 Å appeared within a few days. Just before data collection, the crystals were transferred to a fresh buffer solution (3 mM DTT, 10 mM BES, 0.1 mM EDTA, 0.02 % sodium azide, pH 6.7) containing 30% (v/v) glycerol for 30 – 60 s prior to mounting in a loop, and flash frozen with nitrogen gas at 100K. Data were collected from a single crystal using an image plate on the beamline X11 at Hamburg (␭ ⫽ 0.928 Å) at a maximum resolution of 2.3 Å. Data frames of 0.8° rotation angle were collected over a total angular range of 37.6°. Data were processed and reduced using DENZO and SCALEPACK (26).

Kinetic experiments with GPa showed that W1807 is a potent inhibitor of the enzyme with an IC 50 ⫽ 76.1 nM in the direction of glycogen synthesis (Fig. 2A). The IC 50 value for caffeine was calculated to be 0.45 mM. GPa inhibition is synergistic with caffeine, resulting in a reduction of IC 50 for W1807 by more than 10-fold (IC 50 ⫽ 6.9 nM), at a concentration of 1.0 mM caffeine. The effect of varying caffeine concentration on the relative IC 50 value for W1807 is shown in Fig. 2B. Caffeine, a T state inhibitor of the enzyme, with a K i value of 0.1– 0.2 mM, is known to function with glucose in a synergistic mode (15), with each compound promoting the binding of the other (with an interaction constant (35) ␣ ⫽ 0.3) (36). Previous experiments showed that GPa inhibition by W1807 is also synergistic with glucose (11). Crystallization Crystals of the T state GPa can only be obtained in the presence of glucose (37). Similarly, crystals of the T state GPa– glucose–W1807 complex can be prepared

248

TSITSANOU, SKAMNAKI, AND OIKONOMAKOS TABLE 1

Diffraction Data and Refinement Statistics for T State GPa–Caffeine–W1807 Complex Space group No. of images (degrees/image) Unit cell dimensions Resolution range No. of observations No. of unique reflections I/␴(I) (outermost shell) a Completeness (outermost shell) R merge (outermost shell) b Multiplicity Outermost shell Mosaicity Refinement (resolution) No of reflections used (free) Residues included No of protein atoms No of water molecules No of ligand atoms

Final R (R free) c r.m.s.d. in bond lengths (Å) r.m.s.d. in bond angles (°) r.m.s.d. in dihedral angles (°) r.m.s.d. in improper angles (°) Average B (Å 2) for protein residues Overall CA,C,N.O Side chain Average B (Å 2) for ligands PLP W1807 Caffeine Glycerol Water Ser14-P

P4 32 12 47 (0.8°) a ⫽ b ⫽ 126.7 Å, c ⫽ 115.3 Å 26.5–2.3 Å 183,590 41,698 10.5 (2.6) 98.4% (98.5%) 0.078 (0.375) 4.4 2.34–2.30 Å 0.27 26.5–2.3 Å 39,461 (2099 free) 5–250,260–314,325–835 6606 704 15 (PLP) 4 (Ser14-P) 28 (W1807) 14 (caffeine) 6 (glycerol) 21.0% (25.7%) 0.007 1.38 24.4 0.73 5–250,260–314,325–835 28.8 28.1 29.6

ing that W1807 alone might not promote dissociation of the GPa tetrameric species. X-Ray Crystallography The overall architecture of the native T state GPb with the location of the cofactor pyridoxal 5⬘-phosphate (PLP), the catalytic site, the allosteric (or nucleotide) site, the inhibitor (or nucleoside) site, and the new allosteric inhibitor site is presented in Fig. 1. Crystallographic data collection at 100K, processing, and refinement statistics for the cocrystallized GPa– caffeine–W1807 complex are shown in Table I. The refined 2F o–F c electron density map indicated that caffeine (Scheme I) bound at the inhibitor site of GPa. Additional density at the allosteric site indicated tight binding of W1807 in a position and conformation similar with that observed for the refined T state GPa– glucose–W1807 complex structure (11). There was also binding of a glycerol molecule at the catalytic site; glycerol, used as a cryoprotectant at a concentration of 30% (v/v) in the 100K crystallographic experiment, was found to bind at the catalytic site. The electron density for the caffeine molecule as bound to the inhibitor site of T state GPa are shown in Fig. 3. We describe briefly the GPa/ligand interactions at the catalytic and allosteric sites and in more detail the interactions of caffeine with the inhibitor site.

21.4 26.3 32.1 23.8 35.5 ⬎100.0

␴(I) is the standard deviation of I. R merge ⫽ ⌺ i ⌺ h/具I h典 ⫺ I ih兩/⌺ i⌺ hI ih, where 具I h典 and I ih are the mean and ith measurement of intensity for reflection h, respectively. c Crystallographic R ⫽ ⌺兩兩F o兩 ⫺ 兩F c兩兩/⌺兩F o兩, where 兩F c} are the observed and calculated structure factor amplitudes, respectively. R free is the corresponding R value for a randomly chosen 5% of the reflections that were not included in the refinement. a b

with 1 mM W1807 in the presence of 50 mM glucose (11). Crystals of the GPa– caffeine–W1807 complex were grown, under similar conditions, with 0.9 mM W1807 and 4.9 mM caffeine in the absence of glucose. This is, to our knowledge, the first report of muscle T state GPa crystallization without glucose. GPa exists in a tetrameric state even at relatively low concentrations (38), but may be dissociated by T state inhibitors, such as glucose and caffeine, to give inactive T state dimers (38, 39). The inclusion of W1807 (1.0 mM) rather than glucose or caffeine in the crystallization medium did not result in any crystallization, suggest-

SCHEME I. W1807, glycerol, and caffeine showing the numbering system used.

CAFFEINE AND W1807 BINDING TO PHOSPHORYLASE a

Catalytic Site The mode of binding and the interactions that glycerol makes with GPa are similar with those for GPb (14). Glycerol, on binding to GPa, binds in such a way that its three oxygen atoms are located in similar positions as O-3, O-4, and O-6 of the glucopyranose ring in the GPa– glucose complex. O-1 makes three direct contacts to Glu672 (OE2), Ser674 (N), and Gly675 (N) and two indirect contacts to Ala673 (N) and His377 (O), and to Glu672 (OE2) via water molecules Wat136 and Wat6, respectively. O-10 interacts directly with Asn484 (OD1) and Gly675 (N) and an indirect contact to Thr676 (N and OG1) via water molecule Wat109. O-7 makes hydrogen bonds to Asn484 (OD1) and His377 (ND1). The structural results indicate that, in the presence of caffeine and W1807, glycerol can be accommodated at the catalytic site of GPa with essentially no disturbance of the structure; there are no significant differences in amino acid side-chain positions to those observed in the GPa– glucose complex. Allosteric Site W1807 on binding at the allosteric site of GPa makes charge/charge and polar/polar interactions and exploits numerous van der Waals contacts that are responsible for the high affinity of W1807 for the enzyme. These comprise (1) ion pairing with three arginines, Arg242, Arg309, and Arg310, (2) hydrogen bonding with Gln71, Gln72, and Asp42⬘ through water molecules, and (3) substantial nonpolar contacts to Val45⬘, Trp67, Ile68, Tyr75, and Phe196 (superscript prime refers to residues from the symmetry-related subunit). The nonpolar contact comprise mainly aromatic/aromatic interactions (chlorophenyl group/side-Phe196 side-chain), CH/␲ electron interactions (methyl group/ Tyr75 side-chain, ethyl group/Tyr75 side-chain, and Val45⬘ side-chain/chlorophenyl group), and nonpolar/ nonpolar interactions (isopropyl group/CD1 of Trp67 and CA and CG1 of Ile68 and aliphatic part of Gln71). In addition, at least seven water molecules are displaced on binding W1807. Inhibitor Site The inhibitor (or nucleoside) site is a hydrophobic binding pocket of relatively low specificity. It is located at the entrance to the catalytic site, near the domain interface, and comprises residues from both domains 1 (residues 13– 484) and 2 (residues 485– 842) (Fig. 1). In the T state, Phe285, from the 280s loop (residues 282 to 286), is stacked close to Tyr613, from the start of the ␣19 helix (residues 613– 631), and together these two hydrophobic aromatic residues form the inhibitor site (1). The physiological significance of this site has yet to be established but it may be used by an unidentified

249

compound to enhance the effects of the control of liver GP by glucose, possibly in response to insulin (40, 41). Difference electron density maps derived from previous medium-resolution x-ray crystallographic analyses (16) were interpreted with the caffeine in either of two orientations, I and II, related by a rotation of ca. 180° about the long axis of the caffeine molecule. These authors have chosen orientation I because it placed the oxygen subtituents (O11 and O13) in higher electron density than the methyl groups (C10, C12, C14). The resolution of the present structure allows us to describe the binding site for caffeine in some detail. The structural data suggest two possible orientations, orientation I and orientation II related by a rotation of ca. 180° about the C2–C5 axis of the six-membered ring and not about the long axis of the caffeine molecule. The fitting of each model in the density map was investigated, and the two orientations were explored in the refinement. Both gave similar crystallographic R values and similar temperature factors for the relevant atoms, but the best fitting in the electron density map was achieved with the orientation I (Fig. 3). In both orientations, the O11, is positioned over the Tyr613 hydroxyl group, but in orientation II O13 occupies a position close to that occupied by N9 in orientation I. It is possible, however, that the binding site can accommodate either orientation I or orientation II. The separation between the six-membered ring center of purine and the centers of Phe285 and Tyr613 is 3.3 Å and 3.4 Å, respectively, and these two aromatic residues are also in similar positions in the T state GPa– glucose complex structure. Caffeine on binding to GPa makes a total of 2 hydrogen bonds and 87 van der Waals interactions (44 nonpolar/nonpolar, 4 polar/polar and 39 polar/nonpolar) (Table II). Carbonyl oxygen O11 makes an indirect contact to Glu382 (OE2) via a water molecule (Wat230), and nitrogen N9 is hydrogen bonded to another water molecule (Wat256), which itself interacts with Asp283 (N and O) (Fig. 4). The previous structural study (16), determined at a nominal resolution of 2.5 Å, did not identify these two water-mediated links between caffeine O11 and N9 atoms with Glu382 and Asp283. The most characteristic feature for caffeine binding to GPa is its stacking interactions with two aromatic residues, Phe285, from the 280s loop, and Tyr613, from the start of the ␣19 helix (residues 613– 631). On forming the complex with GPa the inhibitor becomes buried. The solvent accessibilities of the free and bound caffeine molecules are 355 and 81 Å 2, indicating that a surface area of 274 Å 2 becomes inaccessible to water or that caffeine becomes 77% buried in the enzyme complex. Both polar and nonpolar groups are buried, but the greatest contribution comes from the nonpolar groups which contribute 237 Å 2 (78%) of the surface that becomes inaccessible. On the protein

250

TSITSANOU, SKAMNAKI, AND OIKONOMAKOS TABLE II

Hydrogen Bonds and van der Waals Contacts Between Caffeine and GPa A. Hydrogen bonds a Protein atom Caffeine atom O11

1st b Wat230 (2.8)

2nd c

Glu382 OE2 (2.8) Wat29 (2.6)

Wat212 (3.3)

N9

Wat256 (2.8)

Asp283 O (2.7) Asp283 N (3.3) Wat277 (2.8)

3rd

Glu572 N (2.9) Tyr573 N (2.9) Tyr613 OH (3.3) Asn284 O (2.8) Glu382 OE2 (3.0) Wat230 (3.3)

Ile570 O (2.7) Wat256 (2.8)

B. Van der Waals contacts d Caffeine atom

Protein atom

No. of contacts

N1 C2 C10 C6 N3 O11 C12 C4 C5 N9 O13 N7 C8 C14 Total

Phe285 CD2,CE1; Tyr613 CD1,CE1 Phe285 CD2; Tyr613 OH,CE2,CZ,CD1,CE1; Wat230 Tyr613 CE1 Phe285 CD2,CE2,CZ; Tyr613 CG,CD1 Phe285 CB,CG,CD2; Tyr613 CG,CD1,CD2,CE1,CE2,CZ Tyr613 OH,CE1,CZ Asp283 O; Asn284 O; Phe285 CB; His571 ND1,NE2,CE1; Tyr613 CD2,CE2,CZ; Wat212; Wat256 Phe285 CB,CG,CD1,CD2; Tyr613 CB,CG,CD2; Wat256 Phe285 CG,CD1,CD2,CE1,CE2,CZ; Tyr613 CA,CB,CG,CD1 Phe285 CB,CG,CD1; Ala610 CB; Tyr613 CD2 Phe285 CE2,CZ; Gly612 O; Wat446 Phe285 CG,CD1,CE1; Tyr613 N,CA,CB Asn282 ND2; Phe285 CB,CG,CD1; Ala610 CB; Tyr613 CB; Wat256 Phe285 CD1,CE1; Gly612 C,O,CA; Tyr613 N,CA

4 7 1 5 9 3 11 8 10 5 4 6 7 7 87

a

The distance between O11 and OH of Tyr613 is just 3.4 Å. 1st, 2nd, and 3rd represent the 1st, 2nd, and 3rd coordination protein atoms, respectively. c Values in parentheses represent hydrogen bond distances in Å. d Hydrogen-bonds were assigned if the distance between the electronegative atoms was less than 3.3 Å and if both angles between these atoms and the preceding atoms were greater than 90°. Van der Waals interactions were assigned for non-hydrogen atoms separated by less than 4 Å. b

surface, a total of 181 Å 2 solvent-accessible surface area becomes inaccessible on binding caffeine, of which 140 Å 2 (74%) is contributed by nonpolar groups. Sprang et al. (16) from binding and thermodynamic analysis of the interactions of several purines, purine nucleosides, nucleotides, and related compounds with GP have concluded that the major source of binding energy derives from the three-ring stacking interaction itself. In support of this, Soman and Philip (42) have shown that the ability of aromatic compounds to inhibit GP is related to their hydrophobicity. In total seven water molecules, that are present in the “native,” 2.0 Å resolution, structure (T state GPa– glucose, PDB code 2GPA), are displaced on binding caffeine, Wat253 (27.9 Å 2), Wat349 (36.4 Å 2), Wat412 (27.1 Å 2), Wat447

(57.9 Å 2), Wat571 (57.1 Å 2), Wat682 (44.8 Å 2), and Wat688 (47.5 Å 2), where values in parentheses represent the B factor values of these water molecules. The increase in entropy from the release of these waters, together with the aromatic stacking interactions and the polar contacts appear to be the major source of binding energy of caffeine. Ser14-P Site In the T state structure of GPb (43), N-terminal residues 10 –22 (residues 1–10 are not located) are in an extended ␤-stranded conformation, are poorly ordered, and make intrasubunit contacts. On phosphorylation, there is a dramatic conformational change so

CAFFEINE AND W1807 BINDING TO PHOSPHORYLASE a

FIG. 4.

251

The contacts between caffeine and GPa, in the vicinity of the inhibitor site. The view is similar to that shown in Fig. 3.

FIG. 5. Superposition of residues from the inhibitor binding site in the T-state GPa. GPa– glucose complex (A), and GPa– glucose–W1807 complex (B) superimposed onto the GPa– caffeine–W1807 complex are shown in stereo. Green, GPa– caffeine–W1807 complex; white, GPa– glucose complex; yellow, GPa– glucose–W1807 complex.

252

TSITSANOU, SKAMNAKI, AND OIKONOMAKOS

that the N-terminal residues 10 –22, swing through 120° with respect to their position in GPb and change their structure to a distorted 3 10 helix and their contacts to intersubunit. The change in position of the tail is accompanied by a change in the torsion angles at Asn23 from ⫺45°, ⫺46° in GPb to ⫺90°, ⫺176° in GPa. In the GPa– glucose structure, the Ser14-P makes ionic contacts with two arginine residues, Arg69 from its own subunit and Arg43⬘ from the other subunit. In the GPa– caffeine–W1807 complex structure, the N-terminal tail, residues 5–22, is difficult to locate; examination of the electron density map indicates that there is poor density in this region. Residues 5–26 are the most disordered residues of the GPa– caffeine–W1807 complex structure; when modeled and included in the refinement, showed high overall 具B典-factors (93.5 Å 2 with a value of ⬎100.0 Å 2 for Ser14-P). The alternative binding mode of residues 13–23, modeled in the same position to that of GPb, was investigated in the refinement, but it gave poor density for residues Ile13, Val15, Arg16, Gly17, Leu18, Ala19, Val21, Glu22, and Asn23. Partial occupancy, however, of the location of the Nterminal tail in the position found in GPb cannot be ruled out. Conformational Changes The binding of W1807 in the T state GPa crystals, in the presence of glucose, is accompanied by substantial conformational changes (11). The shifts in the region of 43⬘ to 49⬘ and 192 to 196 affect residues remote from the allosteric site (such as Pro194), which in turn affect the subunit/subunit interface packing. The enzyme shows a small rotation (1.2°), this rotation bringing the two subunits of the dimer closer together, residues 43⬘ to 49⬘ move closer to the other subunit and tighten the W1807 site, while residues 192 to 196 shift to contribute (through Phe196) van der Waals interactions (11). These shifts appear important in stabilising a modified T state, denoted T⬘, that is “more T state than the T state.” The same shifts are observed on binding W1807 to the allosteric site of GPa, in the presence of caffeine (see Experimental Procedures), and this indicates that binding of W1807 to the allosteric site of GPa– caffeine complex promotes a similar conformational state. Indeed, the comparison of the GPa– caffeine–W1807 and GPa– glucose–W1807 complex structures showed that the positions of the C␣, main chain, and side chain atoms for residues 5–250, 261– 313, and 325– 835 deviate from their mean positions by 0.076, 0.091, and 0.193 Å, respectively, indicating that the complexes are similar in their overall conformation to within the limits of the resolution of the study.

In the T state GPa– glucose complex structure, there are also hydrogen bonds across the domain interface in the vicinity of the caffeine binding site that are important in stabilising the T state conformation. Arg770 NH1 and NH2 groups hydrogen bond to OE2 of Glu382, Arg569 NH2 hydrogen bonds to OD1 of Asn133, and NE2 of His571 hydrogen bonds to OD2 of Asp283. Figure 5 shows superpositions of the T state GPa– glucose–W1807 and GPa– glucose complexes, respectively, onto the T state GPa– caffeine–W1807 complex in the vicinity of the inhibitor site. The superposition shows small shifts in the side chain conformations of Asn284, Glu382, His571, and Arg770. In particular, there is a change of the Asn284 ␹1 angle (from ⫺117° to ⫺79°), resulting in a movement of the ND2 atom away from the inhibitor site and toward the catalytic site by 1.5 Å. This atom is hydrogen bonded to the O-2 hydroxyl of glucose in the T state GPa complexes. Apart from these small shifts, the interdomain hydrogen bonding pattern is also retained in the GPa– caffeine–W1807 complex. Mechanism of Synergistic Inhibition The GPa– caffeine–W1807 complex structure shows that W1807 acts as an allosteric inhibitor binding at a site distant from the catalytic site, consistent with the Monod–Wyman–Changeux model for allosteric effects (13); it induces conformational changes characteristic of a T⬘ state conformation, e.g., a disordering of the N-terminal tail and shifts of the loop of chain containing residues 192–196 and of residues 43⬘– 49⬘. Caffeine binds more tightly in this state by making contacts to the 280s loop in the closed conformation at the catalytic site. This conformation prevents the crucial conformational changes that are critical for catalytic activity, e.g., movement of Arg569 into the catalytic site to create the phosphate recognition site (2) and also prevents the binding of the glycogen substrate. Therefore by promoting the T⬘ state, W1807 will also promote synergistic binding of caffeine, and our kinetic experiments on the separate and combined effects of caffeine and W1807 confirm these observations. The W1807 site is remote from the caffeine binding site but the two sites make contacts to residues of two key regions on the protein, the ␣2 helix and the cap⬘ region of the other subunit, and the 280s loop. By tuning these contacts, effectors may promote either the T state or the R state. Crystallographic evidence (33, 44, 45) suggests that the inhibitor site does not exist in the R state, indicating that the T state favors binding of caffeine while the R state does not. It has been shown (46) that the inhibitory effect of caffeine to GPa is largely overcome by the presence of the nucleotide AMP, indicating that caffeine exhibits poor binding to AMP-activated

CAFFEINE AND W1807 BINDING TO PHOSPHORYLASE a

GPa. Conversely binding of caffeine will shift the equilibrium T 7 R in the favor of the T state. In conclusion, the GPa– caffeine–W1807 complex structure provides a more detailed description of the binding site for caffeine, the improvement mostly being due to the application of cryotechniques not available before (16), and a rationale for W1807 synergism with caffeine. Knowing the details of caffeine interactions with the enzyme will be important in the design and optimization of inhibitors. Furthermore, undestanding of the interactions of various ligands specific for each binding site and of protein conformational flexibility may lead to novel strategies to use a GP inhibitor in combination with other compounds in the treatment of type II diabetes mellitus. ACKNOWLEDGMENTS This work was supported by the Greek GSRT (PENED1999, 99ED237) and EMBL Hamburg through the HCMP Access to LIP (CHGE CT93 0040). W1807 was kindly provided by Bayer AG (Wuppertal, Germany). We acknowledge the assistance of the staff at EMBL, Hamburg, in X-ray data collection. We are grateful to Dr. Spyros E. Zographos for help in the production of figures. Figures 1 and 3–5 were produced using XOBJECTS, a molecular illustration programme (M.E.M. Noble, unpublished results).

REFERENCES 1. Johnson, L. N., Hajdu, J., Acharya, K. R., Stuart, D. I., McLaughlin, P. J., Oikonomakos, N. G., and Barford, D. (1989) in Allosteric Enzymes (Herve, G., Ed.), pp. 81–127, CRC Press, Boca Raton, FL. 2. Johnson, L. N. (1992) FASEB J. 6, 2274 –2282. 3. Oikonomakos, N. G., Acharya, K. R., and Johnson, L. N. (1992) in Posttranslational Modification of Proteins (Harding, J. J., and Crabbe, M. J. C., Eds.), pp. 81–151, CRC, Boca Raton, FL. 4. Martin, J. L., Veluraja, K., Johnson, L. N., Fleet, G. W. J., Ramsden, N. G., Bruce, I., Oikonomakos, N. G., Papageorgiou, A. C., Leonidas, D. D., and Tsitoura, H. S. (1991) Biochemistry 30, 10101–10116. 5. Watson, K. A., Mitchell, E. P., Johnson, L. N., Son, J. C., Bichard, C. J. F., Orchard, M. G., Fleet, G. W. J., Oikonomakos, N. G., Leonidas, D. D., Kontou, M., and Papageorgiou, A. C. (1994) Biochemistry 33, 5745–5758. 6. Watson, K. A., Mitchell, E. P., Johnson, L. N., Cruciani, G., Son, J. C., Bichard, C. J. F., Fleet, G. W. J., Oikonomakos, N. G., Kontou, M., and Zographos, S. E. (1995) Acta Crystallogr. D51, 458 – 472. 7. Bichard, C. J. F., Mitchell, E. P., Wormald, M. R., Watson, K. A., Johnson, L. N., Zographos, S. E., Koutra, D. D., Oikonomakos, N. G., and Fleet, G. W. J. (1995) Tetrahedron Lett. 36, 2145– 2148. 8. Oikonomakos, N. G., Kontou, M., Zographos, S. E., Watson, K. A., Johnson, L. N., Bichard, C. J. F., Fleet, G. W. J., and Acharya, K. R. (1995) Protein Sci. 4, 2469 –2477. 9. Zographos, S. E., Oikonomakos, N. G., Tsitsanou, K. E., Leonidas, D. D., Chrysina, E. D., Skamnaki, V. T., Bischoff, H., Goldman, S., Schram, M., Watson, K. A., and Johnson, L. N. (1997) Structure 5, 1413–1425. 10. Gregoriou, M., Noble, M. E. M., Watson, K. A., Garman, E. F., Krulle, T. M., Fuente, C., Fleet, G. W. J., Oikonomakos, N. G., and Johnson, L. N. (1998) Protein Sci. 7, 915–927.

253

11. Oikonomakos, N. G., Tsitsanou, K. E., Zographos, S. E., Skamnaki, V. T., Goldmann, S., and Bischoff, H. (1999) Protein Sci. 8, 1930 –1945. 12. Oikonomakos, N. G., Skamnaki, V. T., Tsitsanou, K. E., Gavalas, N. G., and Johnson, L. N. (2000) Structure 8, 575– 484. 13. Monod, J., Changeux, J. P., and Jacob, F. (1965) J. Mol. Biol. 12, 88 –118. 14. Tsitsanou, K. E., Oikonomakos, N. G., Zographos, S. E., Skamnaki, V. T., Gregoriou, M., Watson, K. A., Johnson, L. N., and Fleet, G. W. J. (1999) Protein Sci. 8, 741–749. 15. Kasvinsky, P. J., Shechosky, S., and Fletterick, R. J. (1978) J. Biol. Chem. 253, 9102–9106. 16. Sprang, S. R., Fletterick, R. J., Stern, M. J., Yang, D., Madsen, N. B., and Sturtevant, J. S. (1982) Biochemistry 21, 2036 –2048. 17. Stura, E. A., Zanotti, G., Babu, Y. S., Sansom, M. S. P., Stuart, D. I., Wilson, K. S., Johnson, L. N., and Van de Werve, G. (1983) J. Mol. Biol. 170, 529 –565. 18. Goldmann, S., Ahr, H.-J., Puls, W., Bischoff, H., Petzinna, D., Schlossmann, K., and Bender, J. (1988). U.S. Patent 4,786,641, Bayer Ag. 19. Martin, W. H., Hoover, D. J., Armento, S. J., Stock, I. A., McPherson, R. K., Danley, D. E., Stevenson, R. W., Barrett, E. J., and Treadway, J. L. (1998) Proc. Natl. Acad. Sci. USA 95, 1776 – 1781. 20. Hoover, D. J., Lefkowitz-Snow, S., Burgess-Henry, J. L., Martin, W. H., Armento, S. J., Stock, I. A., McPherson, R. K., Genereux, P. E., Gibbs, E. M., and Treadway, J. L. (1998) J. Med. Chem. 41, 2934 –2938. 21. Andersen, B., Rassov, A., Westergaard, N., and Lundgren, K. (1999) Biochem. J. 342, 545–550. 22. Fosgerau, K., Westergaard, N., Quistorff, B., Grunnet, N., Kristiansen, M., and Lungren, K. (2000) Arch. Biochem. Biophys. 15, 274 –284. 23. Fischer, E. H., and Krebs, E. G. (1962) Methods Enzymol. 5, 369 –373. 24. Kastenschmidt, L. L., Kastenschmidt, J., and Helmreich, E. J. M. (1968) Biochemistry 7, 3590 –3608. 25. Helmreich, E. J. M., and Cori, C. F. (1964) Proc. Natl. Acad. Sci. USA 51, 131–138. 26. Otwinowski, Z. (1993) DENZO. Data collection and Processing, DL/SC1/R34, SERC Laboratory, Daresbury, Warrington, UK. 27. Brunger, A. T. (1996) X-PLOR Version 3.8. Dept. of Molecular Biophysics and Biochemistry, Yale University, New Haven, CT. 28. Jones, T. A., Zou, J. Y., Cowan, S. W., and Kjeldgaard, M. (1991) Acta Crystallogr. A47, 110 –119. 29. Read, R. J. (1986) Acta Crystallogr. A42, 140 –149. 30. Luzatti, V. (1952) Acta Crystallogr. 5, 802– 810. 31. Laskowski, R. A., MacArthur, M. W., Moss, D. S., and Thorton, J. M. (1993) J. Appl. Crystallogr. 26, 283–291. 32. Hubbard, S. J., and Thornton, J. M. (1993) NACCESS, computer program, Department of Biochemistry and Molecular Biology, University College London. 33. Collaborative Computational Project, Number 4. (1994) The CCP4 Suite: Programs for Protein Crystallography. Acta Crystallogr. D50, 760 –763. 34. Sprang, S. R., Withers, S. G., Goldsmith, E. J., Fletterick, R. J., and Madsen, N. B. (1991). Science 254, 1367–1371. 35. Segel, I. H. (1975) Enzyme Kinetics, pp. 465–504, Wiley Interscience, New York. 36. Madsen, N. B., Shechosky, S., and Fletterick, R. J. (1983) Biochemistry 22, 4460 – 4465.

254

TSITSANOU, SKAMNAKI, AND OIKONOMAKOS

37. Fletterick, R. J., Sygusch, J., Murray, N., Madsen, N. B., and Johnson, L. N. (1976). J. Mol. Biol. 103, 1–13. 38. Huang, C. Y., and Graves, D. J. (1970) Biochemistry 21, 5364 – 5371. 39. Withers, S. G., Sykes, B. D., Madsen, N. B., and Kasvinsky, P. J. (1979) Biochemistry 18, 5342–5348. 40. Kasvinsky, P. J., Fletterick, R. J., and Madsen, N. B. (1981) Can. J. Biochem. 59, 387–395. 41. Ercan-Fang, N., and Nuttall, F. Q. (1997) J. Pharmacol. Exp. Ther. 280, 1312–1318.

42. Soman, G., and Philip, G. (1975) Biochem. J. 147, 369 –371. 43. Acharya, K. R., Stuart, D. I., Varvill, K. M., and Johnson, L. N. (1991) Glycogen Phosphorylase: Description of the Protein Structure. World Scientific, Singapore/New Jersey/London/Hong Kong. 44. Barford, D., and Johnson, L. N. (1989) Nature 340, 609 – 616. 45. Barford, D., Hu, S. H., and Johnson, L. N. (1991) J. Mol. Biol. 218, 233–260. 46. Leonidas, D. D., Oikonomakos, N. G., Papageorgiou, A. C., Xenakis, A., Cazianis, C. T., and Bem, F. (1990) FEBS Lett. 261, 23–27.