View Article Online / Journal Homepage / Table of Contents for this issue
Published on 01 January 1992. Downloaded by Christian Albrechts Universitat zu Kiel on 25/10/2014 12:09:13.
Faraday Discuss., 1992, 93, 13 1- 142
Catalytic Mechanism of Glycogen Phosphorylase L. N. Johnson,* S-H. Hut and D. Barford$ Laboratory of Molecular Biophysics, South Parks Road, Oxford OX 1 3QU, UK
Proposals for the catalytic mechanism of glycogen phosphorylase based on crystallographic studies with the T-state form of the enzyme are reviewed in the light of new structural data from studies with the R-state enzyme. The observed position for a sulfate ion at the catalytic site and the crystallographic binding studies of glucose-1-P to the R-state enzyme support the previous proposals in which the 5'-phosphate group of the essential cofactor pyridoxal phosphate functions as an acid-base to promote attack by the substrate phosphate on the polysaccharide substrate. The sulfate (phosphate) recognition site, which is fully formed only in the R state, comprises interactions from the side chains of Arg-569 and Lys-574 and the main chain nitrogen of Gly-135 at the start of an a-helix. The interactions of the cofactor 5'-phosphate do not change between the T and the R state. Other groups on the protein play important roles in binding the substrate but are not involved in the catalytic reaction. The presumed reactive conformation of bound substrate has been observed with heptulose-2-P in the T state and in this conformation stereoelectronic arguments suggest the C ( 1)-O( 1) bond is weakened. For the natural substrate glucose-1-P it is proposed that the reactive conformation is achieved only in the presence of the oligosaccharide component in the reactive ternary enzyme-substrate complex. The phosphate recognition sites are discussed.
Glycogen phosphorylase (E.C. 2.4.1.1) catalyses the intracellular degradation of glycogen in muscle to provide energy to sustain muscle contraction and in liver to provide fuel for other tissues. The enzyme-catalysed phosphorolytic cleavage of the a-1,4 glycosidic bond at the non-reducing end of glycogen yields glucose-1-phosphate. (a-1,4-glucoside), +Pi c* (a-l,4-glu~oside),-~ 3-Glc-1-P The reaction proceeds with retention of configuration and therefore following Koshland' a double displacement mechanism with the formation of either a p-glucosyl enzyme intermediate or a carbonium ion stabilised by electrostatic groups has been invoked. In the absence of the second substrate phosphorylase will catalyse neither molecular exchange between Glc-1-P and phosphate nor positional isotope exchange between the peripheral and ester oxygens of Glc-1-P. Only with the potato enzyme, where it was possible to use cyclodextrin as a pseudo second substrate, was positional isotope exchange observed with rates similar to those observed for catalysk2 These experiments provide the major evidence for the glucosyl or carbonium intermediate and also emphasise the importance of formation of the ternary enzyme substrate complex for ?Present address: Laboratory of Molecular Biophysics, The Rockfeller University, 1230 York Avenue, New York, NY 10021, USA. $' Present address: Cold Spring Harbor Laboratory, P.O. Box 100, Cold Spring Harbor, New York, NY 11724, USA. 131
View Article Online
Published on 01 January 1992. Downloaded by Christian Albrechts Universitat zu Kiel on 25/10/2014 12:09:13.
132
Catalytic Mechanism of Glycogen Phosphorylase
initiation of the reaction. The kinetic analysis indicates that the reaction proceeds through a mechanism in which the rate-limiting step is the interconversion of the ternary enzyme-substrate ~ o m p l e x The . ~ binding of glucosyl polymer, phosphate and glucose-1P are considerably stronger to the appropriate enzyme-binary complexes than to the free enzyme (reviewed in ref. 4). This discussion paper reviews our current understanding of the mechanism with emphasis on the mechanism by which attack by phosphate is promoted and hydrolysis by water excluded. Phosphorylase is an archetypal control enzyme. It exhibits regulation both by reversible phosphorylation and by allosteric effectors and integrates diverse signals associated with ligand binding at five spatially distinct sites. To a first approximation these effects can be understood in terms of an equilibrium between several conformational states ranging from a low-affinity T state to a high-affinity R state according to the model of Monod, Wyman and Changeux. In resting muscle, phosphorylase b (GPb) is in the T state and requires AMP for activation to the R state. In response to nervous or hormonal signals the enzyme is converted to phosphorylase a (GPa) through phosphorylation of Ser-14 catalysed by phosphorylase kinase. The major effect of the allosteric transition is to increase the affinity for phosphate or Glc-1-P. GPb shows a 15-fold increase in affinity for phosphate as the concentration of AMP is increased from 0.015 to 0.5 mmol dm-3.5 The KD for phosphate has been estimated to be 93 mmol dm-3 for the free enzyme, 15 mmol dm-3 for the enzyme-AMP complex and 2.2 mmol dm-3 for the enzyme-AMP-glycogen complex (reviewed in ref. 6). Glycogen, in addition to its role as a substrate, is also an effector of phosphorylase. Glycogen binds to a separate site on the surface of the enzyme that is distinct from the catalytic site. In experiments with oligosaccharide substrates this site ( K D= 1 mmol dmP3)has a 20-fold higher affinity than the catalytic site.7 I n vivo this glycogen storage site serves to locate phosphorylase on the glycogen particle. The storage site also plays a role in the activation mechanism. Phosphorylase covalently attached to oligosaccharide at this site’ exhibits an eight-fold increase in activity at zero time, which arises from the ability of glycogen or oligosacchardies to dissociate less active R-state tetramers to more active R-state dimers, but the modified enzyme also exhibits a 2.4-fold decrease in K , for glucose 1-P, characteristic of the binary complexed enzyme with glycogen, indicating communication between the catalytic site and the storage site. Phosphorylase contains an essential cofactor, pyridoxal phosphate (PLP) linked through a Schiff base to Lys-680. Under extreme conditions the Schiff base can be reduced with no loss of activity, showing that the cofactor plays a different role in phosphorylase to the conventional vitamin B6 dependent enzymes? Removal of the cofactor results in inactivation, but the apoenzyme can be fully reactivated by reconstitution with PLP. Reconsitution experiments with a number of modified cofactors (see ref. 6, 10, 11 for reviews) have demonstrated that the 5‘-phosphate plays an obligatory role in catalysis. Only PLP analogues with a dianion that can be protonated such as -0PO;or -CH,PO:give significant activity. The enzyme reconstituted with pyridoxal is inactive, but addition of dianions (such as phosphite) can confer activity. The activity is inhibited by pyrophosphate which is competitive with both the dianion and the substrate phosphate and this observation provided the first direct evidence for the proximity of the substrate and cofactor phosphates.’* 3’P-NMR studies showed that the state of ionisation of the cofactor is dependent on the state of activation of the enzyme: the 5‘-phosphate is a monoanion in the T-state conformation and a dianion in the R-state conformation. In the presence of substrates or inhibitors there is a change in the 3’P chemical shift which has been interpreted either as a partially protonated state or as a distorted d i a n i ~ n . ” ” ~ There have been two main proposals for the role of the 5’-phosphate in the catalytic mechanism based on crystallographic, biochemical and NMR studies (reviewed in 6, 10, 11). Both mechanisms invoke close interactions between the cofactor 5’-phosphate
View Article Online
Published on 01 January 1992. Downloaded by Christian Albrechts Universitat zu Kiel on 25/10/2014 12:09:13.
L. N.Johnson, S-H. Hu and D. Barford
133
and the substrate phosphate. In one proposal it is suggested that the 5’-phosphate group functions in an acid-base mechanism to promote attack by the substrate phosphate on the polysaccharide substrate. In the other proposal a closer association of the cofactor and substrate phosphates is envisaged in which the 5’-phosphate group acts as an electrophile to promote the attack by the substrate phosphate. Both proposals depend on the correct orientation of the substrate phosphate with respect to the cofactor phosphate. The structure of the four key conformations of glycogen phosphorylase have been determined by X-ray crystallography. High-resolution structures are available for T-state GPb crystallized with the weak activator IMP’4 and T-state GPa crystallized in the presence of the inhibitor glucose.’’ In the crystals both T-state forms exist as dimers and comparison of the structures showed the key changes in the conformation of the N-terminal 20 residues, from disorder to order, resulting from phosphorylation on Ser-14.I6 The structures of R-state GPb” and GPal* crystallized in the presence of 1.0 mol dm-3 ammonium sulfate have been determined to a resoltution of 2.9 A. The two R-state structures are similar as the high sulfate concentration induces changes in the N-terminal tail of GPb similar to those observed in GPa on phosphorylation. A comparison of the T- and R-state structures has provided an explanation for the enzyme’s cooperative behaviour on ligand binding and its regulation by allosteric effectors and have determined the R-state strucreversible phosphorylation. Recently Sprang et ture of GPb in the presence of AMP and a modified cofactor, in which the enzyme was crystallised from polyethylene glycol.
Crystallographic Studies at the Catalytic Site with T-state GPb The catalytic site is situated at the centre of the subunit removed from the subunitlsubunit interface of the functional dimer but connected to it by the 280s loop of chain. The site is buried some 15 A from the bulk solvent at the base of a narrow tunnel formed by the interface of the two domains of the subunit and close to the essential cofactor PLP. Early kinetic studies had shown that the enzyme is active in the crystalline state and exhibited a ca. 30-fold decrease in rate compared with the enzyme in solution and with similar K , values for substrates oligosaccharide and glucose-l-P.*O The studies showed a very large value for the K , of oligosaccharide (ca. 175 mmol dm-3). X-Ray experiments on catalysis in the crystal showed that the reaction could be followed either in the direction of oligosaccharide breakdown or synthesis, but oligosaccharide binding at the catalytic site was never observed in the time-averaged difference maps, although oligosaccharide must have visited the catalytic site in the crystal in order to achieve catalysis. Binding of inorganic phosphate was never observed at the catalytic site, presumably because of the low affinity of the free T-state enzyme for phosphate. In the T-state access to the catalytic site is restricted by a loop of chain termed the 280s loop (residues 281-287) and Asp-283 points into the catalytic site linked via two water molecules to the 5’-phosphate of PLP [Fig. 1 (a)]. These observations provide an explanation for the failure to bind oligosaccharides and for the shielded environment of the cofactor phosphate that results in monoanion being favoured. On conversion to the R state the 280s loop is displaced, His-571 breaks its hydrogen bond with Asp-283 and forms a new interaction with Tyr-613, Arg-569 replaces the acidic group Asp-283 and access to the catalytic site is available [Fig. 1 ( b ) ] . These observations provides an explanation for the high affinity of the R state for phosphate and the change in state of ionisation of the 5’-phosphate to a dianion. The most informative studies in the T state were performed with the small substrate heptenitol. The use of glycosylic substrates to probe carbohydrase enzyme mechanisms have been pioneered by Hehre and Lehmann and their colleagues?’ These compounds of non-glycosidic structure have the potential anomeric carbon atom linked via an
View Article Online
Published on 01 January 1992. Downloaded by Christian Albrechts Universitat zu Kiel on 25/10/2014 12:09:13.
134
Catalytic Mechanism of Glycogen Phosphorylase (a)
-281
165
%
Lys-680
Lys-574
J
Ile-165
-281
access to catalytic site
ly-134 Arg-569 \ Q
,
J
qTis-571
4
Lys-574
"/o-o
LYS-680
(a,
Fig. 1 The catalytic site of glycogen phosphorylase b. (a) T-state GPb, showing the cofactor pyridoxal phosphate and access blocked by the 280s loop, residues 282-285. Asp-283 has its side chain directed towards the cofactor phosphate and hydrogen bonded to it by two water molecules (not shown). Arg-569 is buried. ( b ) R-state GPb, showing the displacement of the 280s loop and the movement of Arg-569 into the catalytic site. A sulfate ion is bound with contacts to Arg-569, Lys-574, main-chain N Gly-135 and the 5'-phosphate of pyridoxal phosphate
electron-rich bond. In the presence of phosphate, phosphorylase catalyses the nonreversible phosphorylation of heptenitol to the product heptulose 2-phosphate (p- 1-Cmethyl-a-D-glucose l-phosphate)22 [see Fig. 4( a ) , later]. Heptulose-2-P is a potent hhibitor with K i= 14 pmol dm-3 in solution. The product is bound with considerably higher affinity than the closely related substrate (or product) Glc-1-P where KD== 3 mmol dm-3. In a series of time-resolved experiments that exploited the brilliance of the synchrotron radiation source at Daresbury the conversion of heptenitol to heptulose2-P was followed in the crystal in which the earliest time shot was recorded ca. 1 h after
View Article Online
Published on 01 January 1992. Downloaded by Christian Albrechts Universitat zu Kiel on 25/10/2014 12:09:13.
L. N. Johnson, S-H.Hu and D. Barford
135
1 Gly-677
Fig. 2 Schematic diagram of the contacts between heptulose-2-P and GPb in the T state. Arg-569 has shifted from its buried T-state native position to a position in which it can stabilise the phosphate-phosphate interaction between the product and cofactor. The diagram also shows the principal polar contacts of the cofactor 5'-phosphate. Reproduced with permission from L. N. Johnson, K. R. Archarya, M. D. Jordan and P. J. McLaughlin, J. Mol. BioL, 1990, 29, 509
initiation of the reaction.23 This gave a tantalising glimpse of what might be phosphate in the attacking position but this has still to be confirmed by Laue experiments that allow a finer time slice. The detailed analysis of the end product heptulose-2-P provided that basis for the crystallographic proposals for the catalytic mechanism. The crystal structure of the GPb-heptulose-2-P complex has been refined at 2.9 A r e s ~ l u t i o n .The ~ ~ product is firmly bound at the catalytic site and exhibits temperature factors that are comparable with the most well ordered regions of the enzyme. The major conformational change involves the movement of an arginine residue, Arg-569, from a position buried in the protein to a new position in which the guanidinium group contacts the product phosphate. The importance of this residue for phosphorylase catalysis and control had been anticipated from chemical modification experiment^.^' The arginine displaces an acidic group, Asp-283, from the catalytic site and this replacement of an acidic group by a basic group is a key feature of the creation of a high-affinity phosphate recognition site. Three water molecules are displaced, two by the phosphate group of heptulose-2-P and one by the O(3) hydroxyl of the glucose-like moiety. The interactions of the PLP with the enzyme are essentially unchanged from those of the native enzyme apart from an additional contact to the heptulose-2-P. The interactions between the product and the enzyme are summarised in Fig. 2. All the polar groups on the sugar form hydrogen bonds with the enzyme with the exception of the O(5) ring oxygen atom, whose separation from the main-chain N of Leu-136 (3.5 A) is too long for a strong hydrogen bond. Most of the protein groups involved in hydrogen bonds are from planar groups (main-chain N of Gly-675, Asn-484, His-377, Glu-672 and Arg-569). The product phosphate is hydrogen bonded to the cofactor phosphate. The position is stabilised by ionic interactions and hydrogen bonds to Arg-569, Lys-574, Tyr-573 and the main-chain N of Gly-135. Contacts between the two phosphates are also mediated by a water molecule that bridges the phosphate oxygens and links to the O(4) hydroxy group of the sugar.
View Article Online
Published on 01 January 1992. Downloaded by Christian Albrechts Universitat zu Kiel on 25/10/2014 12:09:13.
136
Catalytic Mechanism of Glycogen Phosphorylase
The conformation of heptulose-2-P bound to the enzyme differs from that observed for glucose-1-P in the torsion angle about the glycosidic bond. In the heptulose-2-P structure the torsion angle 0(5)-C(l)-O(l)-P is 224" and the bond angle at 0(1) is 135". The corresponding angles for glucose-1- P when complexed with phosphorylase are 152" and 129°.26 The change in torsion angle decreases the PLP-phosphorus to product-phosphorus distance from 6.1 A in the glucose-1-P complex to 4.8 A in the heptulose-2-P complex and allows a direct hydrogen bond between the two phosphate groups in the latter complex. The P-methyl group of heptulose-2-P makes satisfactory van der Waals contacts but does not appear to contribute significantly to the bindin energy. Indeed, substituents at the p position generally lead to a decrease in affinity.*$ Heptulose-2-P and glucose-1-P differ only in the addition of the P-methyl group, and the major role of this group in enhancing the binding of heptulose-2-P to phosphorylase appears to be conformational. A detailed comparison26of the binding of a series of a-D-glucosylglucosyl phosphates (glucose-1-P, 2-deoxy-2-fluoro-a-~-glucose-l-~, methylphosphonate and heptulose-2-P) to T-state GPb has shown variability in phosphate positions ranging from P-P separation to the cofactor phosphate of 6.9 A for the 2-fluoro compound, to 6.1 A for glucose-1-P and 5.5 A for the methylene compound (see Fig. 4 of ref. 26). The conformations and their stabilities have been analysed by computational energy calculations to provide a rationalisation for the observed conformations.
Crystallographic Studies at the Catalytic Site with R-state GPb The allosteric transition involves localised changes in tertiary structure associated with larger changes in quaternary struct~re.'~'''On the transition from the T state to the R state there are dramatic changes at the catalytic site. In the R state crystals of both GPa and GPb obtained in the presence of 1 mol dm-3 ammonium sulfate there is a sulfate ion bound at the catalytic site whose position is presumed close to the phosphate recognition site. The dianion site was created by the mutual shifts of Arg-569 and Asp-283 and the disorder of the 280s loop [Fig. 1(b)]. These concerted movements can be linked to changes at the subunit/subunit interface, especially the tower/tower interface, and hence are sensitive to the activation state of the enzyme. In the T state, the-280s loop packs against a loop formed by residues 377-384 (the 380s loop). On transition to the R state, discordering of the 280s loop is correlated with a conformational change of the 380s loop. Residue Glu-382 is displaced by 4.5 A and the T-state salt bridge between Glu-382 and Arg-770 is broken2' This link restrains two subdomains, the glycogen-storage subdomain and the C-terminal subdomain in the T state and on the transition to the R state these two subdomains move apart and open up access to the catalytic site.*' In terms of the overall structure of the subunit there is little relative movement between the two major domains (residues 10-484 and residues 485-842). The relative dispositions of the residues that form the glucosyl recognition site for glucose-1-P in the T state are similar the R state. The major changes are at the phosphate-recognition site. The contacts for the sulfate ion involve the side chains of Arg-569, Lys-574 and the main-chain nitrogen of Gly-135 [Fig. l(b)]. The site is 5 A from the cofactor 5'-phosphate and there is a hydrogen bond between the two ions with the hydrogen presumably being donated by the PLP phosphate. The contacts to the PLP 5'-phosphate are essentially the same as in the T state and involve hydrogen bonds to the main-chain nitrogen of Thr-675 and Gly-677 and to the side chain of Lys-568. Superposition of the T-state structure of the heptulose-2-P complex on the R-state structure with a matrix derived from a least-squares algorithm shows that the sulfate and the heptulose-2-P sites are close but not identical (e.g. Fig. 4 of ref. 29). The sulfate position is suitable for the attacking phosphate in the enzyme-catalysed phosphorolysis of heptenitol.
View Article Online
Published on 01 January 1992. Downloaded by Christian Albrechts Universitat zu Kiel on 25/10/2014 12:09:13.
L. N.Johnson, S-H. Hu and D. Barford
137
V
Fig. 3 Comparison of the native R-state GPb (open lines) and R-state glucose-1-P complex (solid lines) showing glucose-1-P bound at the catalytic site. The phosphate group of glucose-1-P makes contacts to Arg-569 and to the main-chain N of Gly-135 and Leu-136 but not to Lys-574, which shifts towards the cofactor 5'-phosphate (from ref. 30)
Recently the binding of glucose-1-P has been studied in the R-state crystal form?' In these experiments the sulfate concentration had to be reduced since, although sulfate activates by binding at the Ser-14 phosphate recognition site, it also competes with substrate for the catalytic The crystals were transferred from 1 mol dm-3 sodium tartrate, 50 mmol dmA3ammonium sulfate and ammonium sulfate to 1mol 200 mmol dm-3 glucose-1-P, 10 mmol dm-3 @-glycerolphosphate, pH 7.5. Some sulfate was retained in the soak solution in order to keep the Ser-14 phosphate-recognition site saturated. The experiment had an additional complication in that the unit-cell edge in the c direction was doubled, giving rise to eight subunits (molecular weight 97 400) per asymmetric unit. Although a data collection and refinement (by XPLOR) were formidable, the analysis proceeded fairly routinely and resulted in a refined structure (R = 0.188 to 2.9 A resolution) that revealed the details of glucose-1-P binding. The conformation of glucose-l-P and the torsion angle about C( 1)-O( l ) were similar to those observed for the T-state glucose-1-P complex. There is no direct contact between the phosphate and the cofactor 5'-phosphate group (the closest approach between the phosphorus atoms is 6.4A). In the R state the glucose-1-P makes contact through the phosphate group to the side chain of Arg-569 in addition to contacts to the main-chain N of Gly-135 and Leu-136. Surprisingly Lys-574 does not make a direct contact; the distance (averaged from the eight subunits) from the NZ atom to the phosphate oxygen is 4.5A. There are small conformational changes between the glucose-1-P complex and the native R state that include a shift of Arg-569 by 1.8A, shifts in Lys-568 so as to weaken the contact between Lys-568 and the cofactor 5'-phosphate and to make a hydrogen bond with Glu-672, and a shift in Lys-574 away from Glu-672 and towards the cofactor 5'-phosphate(Fig. 3). The phosphate position of glucose-1-P is 3 from the position observed for the sulfate ion (which is expelled in the presence of tartrate). A third phosphate-recognition site has been identified in a crystallo raphic study of GPb reconstituted with a modified cofactor pyridoxal pyrophosphate.zgJ3 In the R-state ammonium sulfate crystals, the second phosphate covalently linked to the cofactor phosphate makes contact to Lys-574 and the main-chain N of Arg-569 but occupies a distinct site that is 3 A and 4.8 A from the sulfate and glucose-1-P sites, respectively.
View Article Online
Catalytic Mechanism of Glycogen Phosphorylase
Published on 01 January 1992. Downloaded by Christian Albrechts Universitat zu Kiel on 25/10/2014 12:09:13.
138
0
\/4-
-O'l
" dO-n-*o ; PXL'
'0
\o
eWOR
0
'"R
\"0'
PXL
" /" -i
/\do\p/o"
/ \o P X L P
(c)
Fig. 4 Proposed mechanism of phosphorylase catalysis for ( a) phosphorolysis of heptenitol, (b) phosphorolysis of oligosaccharide and (c) oligosaccharide synthesis. The S'-phosphate group of the pyridoxal (PXL) phosphate and the substrate phosphate are stabilised by the interactions shown in Fig. l(b) and Fig. 2. Further details are given in the text. Reproduced with permission from L. N. Johnson, K. R. Archaya, M. D. Jordan and P. J. McLaughlin, J. Mol. Biol., 1990, 29, 509.
Catalytic Mechanism The evidence leading to the proposals for the catalytic mechanism and the role of the ' The proposals put forward on cofactor 5'-phosphate group have been the basis of the crystallographic resultsz4with T-state GPb and now supported by studies with R-state GPb and R-state GPa complement those deduced from the results of kinetic and NMR experiments and demonstrate the additional contribution of stereoelectronic effects. The structures observed favour a mechanism [Fig. 4( a)] in which phosphorylysis of heptenitol is catalysed by general acid attack of the substrate phosphate promoted by the cofactor phosphate. The proton is donated by the substrate phosphate in a concerted reaction in which it immediately gains a proton from the cofactor phosphate. After protonation of the methyene carbon, the glucosyl carbonium ion is stabilised by the negatively charged substrate phosphate group. No covalent intermediate has been detected in phosphorylase catalysis and the structure of the protein provides no suitable
View Article Online
Published on 01 January 1992. Downloaded by Christian Albrechts Universitat zu Kiel on 25/10/2014 12:09:13.
L. N. Johnson, S-H.Hu and D. Barford
139
candidates for a covalent glucosyl intermediate, hence it is assumed that the reaction proceeds through a carbonium ion intermediate. The reaction is completed by nucleophilic attack of the substrate phosphate on the carbonium ion to give the product heptulose-2-P. The mechanism can be readily extended to the natural reaction as shown in Fig. 4( 6 ) . In the first step the substrate phosphate protonates the glycosidic oxygen resulting in the cleavage of the C(1)-O(1) bond. The glycosyl carbonium ion is stabilised by the phosphate dianion as before and the reaction completed by attack of the phosphate on the carbonium ion to give glucose-1-I? The crystallographic results show that there are no ionisable groups in the vicinity of the C(l) atom that could contribute to the stabilisation of the transition-state carbonium ion nor are there any protein groups in the vicinity of the O( 1) oxygen that could protonate the substrate (Fig. 2 and 3). Indeed the proximity of the main-chain N of Leu-136 to the ring oxygen O(5) is likely to discourage charge delocalisation to an oxonium-carbonium ion in the transition state and provides a rationalisation as to why classical inhibitors of many glycosidases such as norjirimycin are poor inhibitors of phosphorylase. The mechanism of phosphorylase is different from glycosidases such as lysozyme in the disposition of groups that promote general acid catalysis and which stabilise the transition state. In lysozyme these functions are performed by two acid residues (Glu-35 and Asp-52) whose properties are governed by their environment in the protein. In phosphorylase these functions are performed by the substrate phosphate promoted by the cofactor phosphate and the catalysis depends on the direct interaction of the substrate phosphate and the cofactor phosphate. The transition-state stabilisation by the substrate dianion with the confined and shielded environment of the catalytic site means that the reaction can proceed only in the direction of phosphorolysis and not hydrolysis. In the reverse reaction the cofactor phosphate acts as an acid to protonate the phosphate of glucose-1-P and the C(1)-0(1) bond has to be cleaved without direct protonation of the glycosidic oxygen [Fig. 4( c)]. The conformation observed for heptulose-2-P is similar to that predicted from stereoelectronic arguments to weaken the exo-anomeric effect and promote cleavage of the a-glycosidic bond.34 The theory predicts that an 0-C bond in the grouping 0-C-0 is strongest (ie. charge delocalisation is greatest) when the lone-pair orbital of the oxygen atom is antiperiplanar (180") to the C-0 bond of the other oxygen. For a-glycosides the unshared pair of electrons on the ring oxygen are properly disposed (ie. antiperiplanar to the glycosidic bond) to strengthen the C(1)-0(5) bond (endo-anomeric effect).and to contribute to charge delocalisation so as to assist the formation of the oxycarbonium ion without the need for distortion of the sugar ring (Fig. 5 ) . However, in the preferred conformation for a-glycosides, such as that observed in the single-crystal structure of glucose-1-P where the torsion angle 0(5)-C(l)-O(l)-P is 90" and both bond lengths C(1)-O(1) and C( 1)-0(1) are short, the torsion angle about C(1)-0(1) is such that the unshared pair of electrons on the glycosidic oxygen is antiperiplanar to the C(1)-O(5) bond. This will tend to strengthen the glycosidic bond (exo-anomeric effect) thus making cleavage of this bond more difficult. Rotation about the C( 1)-O( 1) bond to the position observed in heptulose-2-it> complex weakens the exo-anomeric effect and results in increasing polarisation of the C(1)-0(1) bond (Fig. 5). Thus heptulose-2-P formed in the crystal exhibits a conformation that is anticipated to promote cleavage of the glycosidic bond. The presence of the P-methyl group prevents access of oligosaccharide and probably explains why the reaction in the direction of glycogen synthesis is not observed with this compound despite its reactive conformation. For glucose-1-P it is anticipated that the presence of oligosaccharide in the catalytic site favours a conformation in which the phosphate is turned away from the conformation observed in the binary complex (as observed in the structural studies with both the T and R states) towards that observed in the heptulose-2-P complex so that the cofactor
View Article Online
Published on 01 January 1992. Downloaded by Christian Albrechts Universitat zu Kiel on 25/10/2014 12:09:13.
140
Catalytic Mechanism of Glycogen Phosphorylase
I
t I
Qoo
i/ p -0-
0
Fig. 5 A simplified diagram showing stereoelectronic effects for a-glycosides: ( a ) in the preferred conformation for a-glycosides both C(1)-0(5) (endo effect) and C(1)-0(1) (ex0 effect) are stabilised; ( b )in the conformationobserved for hsptulose-2-P the em-anomeric effect is weakened. Reproduced with permission from L. N. Johnson, K. R. Archaya, M. D. Jordan and P. J. McLaughlin, J. Mol. BioL, 1990, 29, 509
phosphate can function as a catalytic group with the assistance of the activated conformation predicted from stereoelectronic principles. Some support for this proposal has been given by the result for the a-D-glucosylmethylphosphonate complex obtained in the presence of oligosaccharide where the phosphate was directed significantly closer to the cofactor phosphate than in the Glc-1-Pcomplex.26 The crystallographic results suggest that when the reaction proceeds in the direction of glycogen synthesis the glycosidic bond is weakened by steric factors that facilitate the development of the carbonium ion and ensure that the phosphate is able to act as a base to abstract a hydrogen from the O(4) hydroxyl of the oligosaccharide. When the reaction proceeds in the direction of glycogen degradation the glycosidic bond is weakened by direct protonation. In both directions the substrate phosphate, stabilised by its interactions with the enzyme, plays a crucial role in the transition-state stabilisation. In an alternative proposal it has been suggested that the cofactor phosphate acts as an electrophile to withdraw electrons from the substrate phosphate and destabilise the glycosidic bond13 (reviewed in ref. 10). A closer association between the cofactor and substrate phosphates is required than has been observed in the crystallographic results to date. Such a close approach is mimicked by the analogue pyridoxal pyrophosphate but the structural show no evidence for additional interactions with the cofactor 5’-phosphate that could stabilise the constrained dianion which is an essential feature of this mechanism. However, as is generally true with X-ray crystallographic
View Article Online
Published on 01 January 1992. Downloaded by Christian Albrechts Universitat zu Kiel on 25/10/2014 12:09:13.
t.N.Johnson, S-H. Hu and D. Barford
141
studies, only thermodynamically stable structures can be examined and the existence of an unstable intermediate can never be definitely excluded. A survey of phosphate-recognition sites in proteins carried out in 1984;’ at a time when the structures of some 3 1 phosphate-recognition sites distributed among some 16 proteins were known, indicated differences between binding sites and catalytic sites. Binding sites, where the phosphate was solely involved in localisation of the ligand, were well located in the crystallographic studies and phosphate groups made contacts with one or more arginine side chains, the main-chain N at the amino termini of helices (helix dipole interactions) and less frequently, with lysine or neutral polar residues. At the catalytic sites the phosphate position was often less well defined and in some proteins appeared to make only indirect interactions with the protein. Phosphorylase contains four phosphate-recognition sites and it is now possible to extend some of these generalisations with comments on the consequences on phosphate recognition. (i) The Ser-14 phosphate site involves two arginine residues (Arg-69 from its own subunit and Arg-43’ from the other subunit) and this site is a tight binding site formed only in the R state. Although the major driving force for the interaction is electrostatic the biological effects for allosteric activation also involve the interactions and interdigitisation of the surrounding residues of the N-terminal tail with the rest of the (ii) The AMP allosteric phosphate-recognition site also involves two arginine residues (Arg-309 and Arg-310 on an a-helix) and is a tight-binding site recognised in both the T and the R states. There is scope for some mobility at this site created by the nearby presence of Arg-242 and in the R state the phosphate of AMP is shifted 1 A towards this residue compared with the T state while in the T state both inorganic phosphate and glucose-6-P utilise this third arginine in their binding modes.16*18 (iii) The 5’-phosphate recognition site for pyridoxal phosphate has the same interactions in both T and R state. The site involves only one basic group (Lys-568) and the helix dipole arising from a helix that commences with residues Thr-676 and Gly-677 (Fig. 2). Some fluctuations have been observed as in the glucose-l-Pcomplex described above. (iv) The catalytic site exhibits the greatest variability in phosphate positions. In the T state differentpositions are observed for the phosphates of glucose-l-P and heptulose2-P. In the R state despite the constellation of two basic side chains (Arg-569 and Lys-574) and the potential helix dipole ( N of Gly-135) different phosphate positions are observed that depend on whether the group is free as in the sulfate ion or bound to other groups as in glucose-l-P complex or pyridoxal pyrophosphate. Such mobility may be anticipated for the enzyme that both promotes attack by phosphate and, in the reverse direction, reacts with the covalent phosphorylated product. The position observed for glucose-l-P is consistent with the notion discussed earlier that the phosphate only adopts the ‘productive’ mode of binding in the presence of the oligosaccharide complex. The major outstanding problem for a complete understanding of the phosphorylase mechanism is to define the position of the oligosaccharide at the catalytic site. This is not easy since the affinity is low and no binding has been observed to date in diffusion experiments with both T- and R-state phosphorylases (unpublished results) or in a cocrystaflisation experiment with R-state phosphorylase. l 9 Model building experiments with R-state phosphorylase show that even with the open access to the catalytic site there is a requirement for some distortion from the preferred conformation of a-1,4-linked oligosaccharides. Attempts to crystallise active and inactive ternary complexes and to determine the structures of other phosphorylases are in progress. References 1 D. E. Koshland, Biol. Rev., 1953, 28, 416. 2 F. C. Kokesh and Y . Kakuda, Biochemistry, 1977, 16, 2467.
View Article Online
Published on 01 January 1992. Downloaded by Christian Albrechts Universitat zu Kiel on 25/10/2014 12:09:13.
142
Catalytic Mechanism of Glycogen Phosphorylase
3 H. D. Engers, W, A. Bridger and N. B. Madsen, Can. J. Biochem., 1970,48, 746. 4 D. J. Graves and J. H. Wang, in The Enzymes, ed. P. D. Boyer, Academic Press, New York, 3rd edn. 1972, VOI. 7, pp. 435-482. 5 E. Helmreich and C. F. Cori, Proc. Natl. Acad. Sci. USA, 1964, 51 131. 6 L. N. Johnson, J. Hajdu, K. R. Acharya, D. I. Stuart, P. J. McLauchlin, N. G. Oikonomakos and D. Barford, in Allosteric Enzymes, ed. G. Herve, CRC Pres, Boca Raton, Florida, 1989, pp. 81-127. 7 P. J. Kasvinsky, N. B. Madsen, R. J. Fletterick and J. Sygusch, J. Biol. Chem., 1978, 253, 1290. 8 G. Philip, G. Gringel and D. Palm, Biochemistry, 1987, 21,3042. 9 E. H. Fischer, A. B. Kent, E. R. Snyder and E. G. Krebs, J. Am. Chem. Soc., 1958,80, 2906. 10 N. B. Madsen and S. G. Withers, in Coenzymes and Cofactors; Pyridoxal Phosphate and Derivatives, ed. D. Dolphin, R. Paulson and 0. Avramovic, Wiley, New York, 1986, pp. 1-29. 11 D. Palm, H. W. Klein, R. Schinzel, M. Buehner and E. J. M. Helmreich, Biochemistry, 1990,29, 1099. 12 R. F. Parrish, R. J. Uhing and D. J. Graves, Biochemistry, 1977, 16, 139. 13 S. G. Withers, N. B. Madsen, B. D. Sykes, M. Takagi, S. Shimomura and T. Fukui, J. Biol. Chem., 1981, 256, 10759. 14 K. R. Acharya, D. I. Stuart, K. M. Varvill and L. N. Johnson, Glycogen Phosphorylase b: Description of the Protein Structure, World Scientific Publishers, Singapore, 1991. 15 S. R. Sprang and R. J. Fletterick, J. Mol. Biol., 1979, 131, 523. 16 S. R. Sprang, K. R. Acharya, E. J. Goldsmith, D. I. Stuart, K. M. Varvill, R. J. Fletterick, N. B. Madsen and L. N. Johnson, Nature (London), 1988,336, 215. 17 D. Barford and L. N. Johnson, Nature (London), 1989, 340,609. 18 D. Barford, S-H. Hu and L. N. Johnson, J. Mol. Biol., 1991, 218, 233. 19 S. R. Sprang, S. G. Withers, E. J. Goldsmith, R. J. Fletterick and N. B. Madsen, Science, 1991,254,1367. 20 P. J. Kasvinsky and N. B. Madsen, J. Biol. Chem., 1976, 251, 6852. 21 E. J. Hehre, C. F. Brewer, T. Uchiyama, P. Schlesselman and J. Lehmann, Biochemistry, 1980,19,3557. 22 H. W. Klein, M. J. Im and D. Palm, Eur. J. Biochem., 1986, 157, 107. 23 J. Hajdu, K. R. Acharya, D. I. Stuart, P. J. McLauchlin, D. Barford, N. G. Oikonomakos, H. W. Klein & L. N. Johnson, EMBO J., 1987,6, 539. 24 L. N. Johnson, K. R. Acharya, M. D. Jordan and P. J. McLaughlin, J. Mol. Biol., 1990, 211, 645, 25 B. Vandenbunder and H. Buc, Eur. J. Biochem., 1983, 133, 509. 26 J. L. Martin, L. N. Johnson and S. G. Withers, Biochemistry, 1990, 29, 10745. 27 J. L. Martin, K. Veluraja, K. Ross, L. N. Johnson, G. W. J. Fleet, N. G. Ramsden, I. Bruce, M. G. Orchard, N. G. Oikonomakos, A. C. Papageorgiou, D. D. Leonidas and H. S. Tsitoura, Biochemistry, 1991, 31, 10101. 28 D. Barford and L. N. Johnson, Protein Science, in the press. 29 D. D. Leonidas, N. G. Oikonomakos, A. C. Papageorgiou, K. R. Acharya, D. Barford and L. N.
Johnson, Protein Science, in the press. 30 S-H. Hu, D. Phil. Thesis, University of Oxford, 1991. 31 D. D. Leonidas, N. G. Oikonomakos, A. C. Papageorgiou, A. Xenakis, C. T. Cazianis and F. Bem, FEBS Lett., 1990, 261, 23. 32 D. D. Leonidas, N. G. Oikonomakos and A. C. Papageorgiou, Biochim. Biophys. Acta, 1991,1076,305. 33 S . R. Sprang, N. B. Madsen and S. G. Withers, Protein Science, in the press. 34 J. P. Praly and R. U. Lemieux, Can. J. Chem., 1987, 65, 213. 35 L. N. Johnson, in Inclusion Compounds, ed. J. L. Atwood, J. E. D. Davies and D. D. Macnichol, Academic Press, New York, 1984, vol. 3 pp. 509-569.
Paper 2f00131D; Received 10th January, 1992
Published on 01 January 1992. Downloaded by Christian Albrechts Universitat zu Kiel on 25/10/2014 12:09:13.
Faraday Discuss., 1992, 93, 13 1- 142
Catalytic Mechanism of Glycogen Phosphorylase L. N. Johnson,* S-H. Hut and D. Barford$ Laboratory of Molecular Biophysics, South Parks Road, Oxford OX 1 3QU, UK
Proposals for the catalytic mechanism of glycogen phosphorylase based on crystallographic studies with the T-state form of the enzyme are reviewed in the light of new structural data from studies with the R-state enzyme. The observed position for a sulfate ion at the catalytic site and the crystallographic binding studies of glucose-1-P to the R-state enzyme support the previous proposals in which the 5'-phosphate group of the essential cofactor pyridoxal phosphate functions as an acid-base to promote attack by the substrate phosphate on the polysaccharide substrate. The sulfate (phosphate) recognition site, which is fully formed only in the R state, comprises interactions from the side chains of Arg-569 and Lys-574 and the main chain nitrogen of Gly-135 at the start of an a-helix. The interactions of the cofactor 5'-phosphate do not change between the T and the R state. Other groups on the protein play important roles in binding the substrate but are not involved in the catalytic reaction. The presumed reactive conformation of bound substrate has been observed with heptulose-2-P in the T state and in this conformation stereoelectronic arguments suggest the C ( 1)-O( 1) bond is weakened. For the natural substrate glucose-1-P it is proposed that the reactive conformation is achieved only in the presence of the oligosaccharide component in the reactive ternary enzyme-substrate complex. The phosphate recognition sites are discussed.
Glycogen phosphorylase (E.C. 2.4.1.1) catalyses the intracellular degradation of glycogen in muscle to provide energy to sustain muscle contraction and in liver to provide fuel for other tissues. The enzyme-catalysed phosphorolytic cleavage of the a-1,4 glycosidic bond at the non-reducing end of glycogen yields glucose-1-phosphate. (a-1,4-glucoside), +Pi c* (a-l,4-glu~oside),-~ 3-Glc-1-P The reaction proceeds with retention of configuration and therefore following Koshland' a double displacement mechanism with the formation of either a p-glucosyl enzyme intermediate or a carbonium ion stabilised by electrostatic groups has been invoked. In the absence of the second substrate phosphorylase will catalyse neither molecular exchange between Glc-1-P and phosphate nor positional isotope exchange between the peripheral and ester oxygens of Glc-1-P. Only with the potato enzyme, where it was possible to use cyclodextrin as a pseudo second substrate, was positional isotope exchange observed with rates similar to those observed for catalysk2 These experiments provide the major evidence for the glucosyl or carbonium intermediate and also emphasise the importance of formation of the ternary enzyme substrate complex for ?Present address: Laboratory of Molecular Biophysics, The Rockfeller University, 1230 York Avenue, New York, NY 10021, USA. $' Present address: Cold Spring Harbor Laboratory, P.O. Box 100, Cold Spring Harbor, New York, NY 11724, USA. 131
View Article Online
Published on 01 January 1992. Downloaded by Christian Albrechts Universitat zu Kiel on 25/10/2014 12:09:13.
132
Catalytic Mechanism of Glycogen Phosphorylase
initiation of the reaction. The kinetic analysis indicates that the reaction proceeds through a mechanism in which the rate-limiting step is the interconversion of the ternary enzyme-substrate ~ o m p l e x The . ~ binding of glucosyl polymer, phosphate and glucose-1P are considerably stronger to the appropriate enzyme-binary complexes than to the free enzyme (reviewed in ref. 4). This discussion paper reviews our current understanding of the mechanism with emphasis on the mechanism by which attack by phosphate is promoted and hydrolysis by water excluded. Phosphorylase is an archetypal control enzyme. It exhibits regulation both by reversible phosphorylation and by allosteric effectors and integrates diverse signals associated with ligand binding at five spatially distinct sites. To a first approximation these effects can be understood in terms of an equilibrium between several conformational states ranging from a low-affinity T state to a high-affinity R state according to the model of Monod, Wyman and Changeux. In resting muscle, phosphorylase b (GPb) is in the T state and requires AMP for activation to the R state. In response to nervous or hormonal signals the enzyme is converted to phosphorylase a (GPa) through phosphorylation of Ser-14 catalysed by phosphorylase kinase. The major effect of the allosteric transition is to increase the affinity for phosphate or Glc-1-P. GPb shows a 15-fold increase in affinity for phosphate as the concentration of AMP is increased from 0.015 to 0.5 mmol dm-3.5 The KD for phosphate has been estimated to be 93 mmol dm-3 for the free enzyme, 15 mmol dm-3 for the enzyme-AMP complex and 2.2 mmol dm-3 for the enzyme-AMP-glycogen complex (reviewed in ref. 6). Glycogen, in addition to its role as a substrate, is also an effector of phosphorylase. Glycogen binds to a separate site on the surface of the enzyme that is distinct from the catalytic site. In experiments with oligosaccharide substrates this site ( K D= 1 mmol dmP3)has a 20-fold higher affinity than the catalytic site.7 I n vivo this glycogen storage site serves to locate phosphorylase on the glycogen particle. The storage site also plays a role in the activation mechanism. Phosphorylase covalently attached to oligosaccharide at this site’ exhibits an eight-fold increase in activity at zero time, which arises from the ability of glycogen or oligosacchardies to dissociate less active R-state tetramers to more active R-state dimers, but the modified enzyme also exhibits a 2.4-fold decrease in K , for glucose 1-P, characteristic of the binary complexed enzyme with glycogen, indicating communication between the catalytic site and the storage site. Phosphorylase contains an essential cofactor, pyridoxal phosphate (PLP) linked through a Schiff base to Lys-680. Under extreme conditions the Schiff base can be reduced with no loss of activity, showing that the cofactor plays a different role in phosphorylase to the conventional vitamin B6 dependent enzymes? Removal of the cofactor results in inactivation, but the apoenzyme can be fully reactivated by reconstitution with PLP. Reconsitution experiments with a number of modified cofactors (see ref. 6, 10, 11 for reviews) have demonstrated that the 5‘-phosphate plays an obligatory role in catalysis. Only PLP analogues with a dianion that can be protonated such as -0PO;or -CH,PO:give significant activity. The enzyme reconstituted with pyridoxal is inactive, but addition of dianions (such as phosphite) can confer activity. The activity is inhibited by pyrophosphate which is competitive with both the dianion and the substrate phosphate and this observation provided the first direct evidence for the proximity of the substrate and cofactor phosphates.’* 3’P-NMR studies showed that the state of ionisation of the cofactor is dependent on the state of activation of the enzyme: the 5‘-phosphate is a monoanion in the T-state conformation and a dianion in the R-state conformation. In the presence of substrates or inhibitors there is a change in the 3’P chemical shift which has been interpreted either as a partially protonated state or as a distorted d i a n i ~ n . ” ” ~ There have been two main proposals for the role of the 5’-phosphate in the catalytic mechanism based on crystallographic, biochemical and NMR studies (reviewed in 6, 10, 11). Both mechanisms invoke close interactions between the cofactor 5’-phosphate
View Article Online
Published on 01 January 1992. Downloaded by Christian Albrechts Universitat zu Kiel on 25/10/2014 12:09:13.
L. N.Johnson, S-H. Hu and D. Barford
133
and the substrate phosphate. In one proposal it is suggested that the 5’-phosphate group functions in an acid-base mechanism to promote attack by the substrate phosphate on the polysaccharide substrate. In the other proposal a closer association of the cofactor and substrate phosphates is envisaged in which the 5’-phosphate group acts as an electrophile to promote the attack by the substrate phosphate. Both proposals depend on the correct orientation of the substrate phosphate with respect to the cofactor phosphate. The structure of the four key conformations of glycogen phosphorylase have been determined by X-ray crystallography. High-resolution structures are available for T-state GPb crystallized with the weak activator IMP’4 and T-state GPa crystallized in the presence of the inhibitor glucose.’’ In the crystals both T-state forms exist as dimers and comparison of the structures showed the key changes in the conformation of the N-terminal 20 residues, from disorder to order, resulting from phosphorylation on Ser-14.I6 The structures of R-state GPb” and GPal* crystallized in the presence of 1.0 mol dm-3 ammonium sulfate have been determined to a resoltution of 2.9 A. The two R-state structures are similar as the high sulfate concentration induces changes in the N-terminal tail of GPb similar to those observed in GPa on phosphorylation. A comparison of the T- and R-state structures has provided an explanation for the enzyme’s cooperative behaviour on ligand binding and its regulation by allosteric effectors and have determined the R-state strucreversible phosphorylation. Recently Sprang et ture of GPb in the presence of AMP and a modified cofactor, in which the enzyme was crystallised from polyethylene glycol.
Crystallographic Studies at the Catalytic Site with T-state GPb The catalytic site is situated at the centre of the subunit removed from the subunitlsubunit interface of the functional dimer but connected to it by the 280s loop of chain. The site is buried some 15 A from the bulk solvent at the base of a narrow tunnel formed by the interface of the two domains of the subunit and close to the essential cofactor PLP. Early kinetic studies had shown that the enzyme is active in the crystalline state and exhibited a ca. 30-fold decrease in rate compared with the enzyme in solution and with similar K , values for substrates oligosaccharide and glucose-l-P.*O The studies showed a very large value for the K , of oligosaccharide (ca. 175 mmol dm-3). X-Ray experiments on catalysis in the crystal showed that the reaction could be followed either in the direction of oligosaccharide breakdown or synthesis, but oligosaccharide binding at the catalytic site was never observed in the time-averaged difference maps, although oligosaccharide must have visited the catalytic site in the crystal in order to achieve catalysis. Binding of inorganic phosphate was never observed at the catalytic site, presumably because of the low affinity of the free T-state enzyme for phosphate. In the T-state access to the catalytic site is restricted by a loop of chain termed the 280s loop (residues 281-287) and Asp-283 points into the catalytic site linked via two water molecules to the 5’-phosphate of PLP [Fig. 1 (a)]. These observations provide an explanation for the failure to bind oligosaccharides and for the shielded environment of the cofactor phosphate that results in monoanion being favoured. On conversion to the R state the 280s loop is displaced, His-571 breaks its hydrogen bond with Asp-283 and forms a new interaction with Tyr-613, Arg-569 replaces the acidic group Asp-283 and access to the catalytic site is available [Fig. 1 ( b ) ] . These observations provides an explanation for the high affinity of the R state for phosphate and the change in state of ionisation of the 5’-phosphate to a dianion. The most informative studies in the T state were performed with the small substrate heptenitol. The use of glycosylic substrates to probe carbohydrase enzyme mechanisms have been pioneered by Hehre and Lehmann and their colleagues?’ These compounds of non-glycosidic structure have the potential anomeric carbon atom linked via an
View Article Online
Published on 01 January 1992. Downloaded by Christian Albrechts Universitat zu Kiel on 25/10/2014 12:09:13.
134
Catalytic Mechanism of Glycogen Phosphorylase (a)
-281
165
%
Lys-680
Lys-574
J
Ile-165
-281
access to catalytic site
ly-134 Arg-569 \ Q
,
J
qTis-571
4
Lys-574
"/o-o
LYS-680
(a,
Fig. 1 The catalytic site of glycogen phosphorylase b. (a) T-state GPb, showing the cofactor pyridoxal phosphate and access blocked by the 280s loop, residues 282-285. Asp-283 has its side chain directed towards the cofactor phosphate and hydrogen bonded to it by two water molecules (not shown). Arg-569 is buried. ( b ) R-state GPb, showing the displacement of the 280s loop and the movement of Arg-569 into the catalytic site. A sulfate ion is bound with contacts to Arg-569, Lys-574, main-chain N Gly-135 and the 5'-phosphate of pyridoxal phosphate
electron-rich bond. In the presence of phosphate, phosphorylase catalyses the nonreversible phosphorylation of heptenitol to the product heptulose 2-phosphate (p- 1-Cmethyl-a-D-glucose l-phosphate)22 [see Fig. 4( a ) , later]. Heptulose-2-P is a potent hhibitor with K i= 14 pmol dm-3 in solution. The product is bound with considerably higher affinity than the closely related substrate (or product) Glc-1-P where KD== 3 mmol dm-3. In a series of time-resolved experiments that exploited the brilliance of the synchrotron radiation source at Daresbury the conversion of heptenitol to heptulose2-P was followed in the crystal in which the earliest time shot was recorded ca. 1 h after
View Article Online
Published on 01 January 1992. Downloaded by Christian Albrechts Universitat zu Kiel on 25/10/2014 12:09:13.
L. N. Johnson, S-H.Hu and D. Barford
135
1 Gly-677
Fig. 2 Schematic diagram of the contacts between heptulose-2-P and GPb in the T state. Arg-569 has shifted from its buried T-state native position to a position in which it can stabilise the phosphate-phosphate interaction between the product and cofactor. The diagram also shows the principal polar contacts of the cofactor 5'-phosphate. Reproduced with permission from L. N. Johnson, K. R. Archarya, M. D. Jordan and P. J. McLaughlin, J. Mol. BioL, 1990, 29, 509
initiation of the reaction.23 This gave a tantalising glimpse of what might be phosphate in the attacking position but this has still to be confirmed by Laue experiments that allow a finer time slice. The detailed analysis of the end product heptulose-2-P provided that basis for the crystallographic proposals for the catalytic mechanism. The crystal structure of the GPb-heptulose-2-P complex has been refined at 2.9 A r e s ~ l u t i o n .The ~ ~ product is firmly bound at the catalytic site and exhibits temperature factors that are comparable with the most well ordered regions of the enzyme. The major conformational change involves the movement of an arginine residue, Arg-569, from a position buried in the protein to a new position in which the guanidinium group contacts the product phosphate. The importance of this residue for phosphorylase catalysis and control had been anticipated from chemical modification experiment^.^' The arginine displaces an acidic group, Asp-283, from the catalytic site and this replacement of an acidic group by a basic group is a key feature of the creation of a high-affinity phosphate recognition site. Three water molecules are displaced, two by the phosphate group of heptulose-2-P and one by the O(3) hydroxyl of the glucose-like moiety. The interactions of the PLP with the enzyme are essentially unchanged from those of the native enzyme apart from an additional contact to the heptulose-2-P. The interactions between the product and the enzyme are summarised in Fig. 2. All the polar groups on the sugar form hydrogen bonds with the enzyme with the exception of the O(5) ring oxygen atom, whose separation from the main-chain N of Leu-136 (3.5 A) is too long for a strong hydrogen bond. Most of the protein groups involved in hydrogen bonds are from planar groups (main-chain N of Gly-675, Asn-484, His-377, Glu-672 and Arg-569). The product phosphate is hydrogen bonded to the cofactor phosphate. The position is stabilised by ionic interactions and hydrogen bonds to Arg-569, Lys-574, Tyr-573 and the main-chain N of Gly-135. Contacts between the two phosphates are also mediated by a water molecule that bridges the phosphate oxygens and links to the O(4) hydroxy group of the sugar.
View Article Online
Published on 01 January 1992. Downloaded by Christian Albrechts Universitat zu Kiel on 25/10/2014 12:09:13.
136
Catalytic Mechanism of Glycogen Phosphorylase
The conformation of heptulose-2-P bound to the enzyme differs from that observed for glucose-1-P in the torsion angle about the glycosidic bond. In the heptulose-2-P structure the torsion angle 0(5)-C(l)-O(l)-P is 224" and the bond angle at 0(1) is 135". The corresponding angles for glucose-1- P when complexed with phosphorylase are 152" and 129°.26 The change in torsion angle decreases the PLP-phosphorus to product-phosphorus distance from 6.1 A in the glucose-1-P complex to 4.8 A in the heptulose-2-P complex and allows a direct hydrogen bond between the two phosphate groups in the latter complex. The P-methyl group of heptulose-2-P makes satisfactory van der Waals contacts but does not appear to contribute significantly to the bindin energy. Indeed, substituents at the p position generally lead to a decrease in affinity.*$ Heptulose-2-P and glucose-1-P differ only in the addition of the P-methyl group, and the major role of this group in enhancing the binding of heptulose-2-P to phosphorylase appears to be conformational. A detailed comparison26of the binding of a series of a-D-glucosylglucosyl phosphates (glucose-1-P, 2-deoxy-2-fluoro-a-~-glucose-l-~, methylphosphonate and heptulose-2-P) to T-state GPb has shown variability in phosphate positions ranging from P-P separation to the cofactor phosphate of 6.9 A for the 2-fluoro compound, to 6.1 A for glucose-1-P and 5.5 A for the methylene compound (see Fig. 4 of ref. 26). The conformations and their stabilities have been analysed by computational energy calculations to provide a rationalisation for the observed conformations.
Crystallographic Studies at the Catalytic Site with R-state GPb The allosteric transition involves localised changes in tertiary structure associated with larger changes in quaternary struct~re.'~'''On the transition from the T state to the R state there are dramatic changes at the catalytic site. In the R state crystals of both GPa and GPb obtained in the presence of 1 mol dm-3 ammonium sulfate there is a sulfate ion bound at the catalytic site whose position is presumed close to the phosphate recognition site. The dianion site was created by the mutual shifts of Arg-569 and Asp-283 and the disorder of the 280s loop [Fig. 1(b)]. These concerted movements can be linked to changes at the subunit/subunit interface, especially the tower/tower interface, and hence are sensitive to the activation state of the enzyme. In the T state, the-280s loop packs against a loop formed by residues 377-384 (the 380s loop). On transition to the R state, discordering of the 280s loop is correlated with a conformational change of the 380s loop. Residue Glu-382 is displaced by 4.5 A and the T-state salt bridge between Glu-382 and Arg-770 is broken2' This link restrains two subdomains, the glycogen-storage subdomain and the C-terminal subdomain in the T state and on the transition to the R state these two subdomains move apart and open up access to the catalytic site.*' In terms of the overall structure of the subunit there is little relative movement between the two major domains (residues 10-484 and residues 485-842). The relative dispositions of the residues that form the glucosyl recognition site for glucose-1-P in the T state are similar the R state. The major changes are at the phosphate-recognition site. The contacts for the sulfate ion involve the side chains of Arg-569, Lys-574 and the main-chain nitrogen of Gly-135 [Fig. l(b)]. The site is 5 A from the cofactor 5'-phosphate and there is a hydrogen bond between the two ions with the hydrogen presumably being donated by the PLP phosphate. The contacts to the PLP 5'-phosphate are essentially the same as in the T state and involve hydrogen bonds to the main-chain nitrogen of Thr-675 and Gly-677 and to the side chain of Lys-568. Superposition of the T-state structure of the heptulose-2-P complex on the R-state structure with a matrix derived from a least-squares algorithm shows that the sulfate and the heptulose-2-P sites are close but not identical (e.g. Fig. 4 of ref. 29). The sulfate position is suitable for the attacking phosphate in the enzyme-catalysed phosphorolysis of heptenitol.
View Article Online
Published on 01 January 1992. Downloaded by Christian Albrechts Universitat zu Kiel on 25/10/2014 12:09:13.
L. N.Johnson, S-H. Hu and D. Barford
137
V
Fig. 3 Comparison of the native R-state GPb (open lines) and R-state glucose-1-P complex (solid lines) showing glucose-1-P bound at the catalytic site. The phosphate group of glucose-1-P makes contacts to Arg-569 and to the main-chain N of Gly-135 and Leu-136 but not to Lys-574, which shifts towards the cofactor 5'-phosphate (from ref. 30)
Recently the binding of glucose-1-P has been studied in the R-state crystal form?' In these experiments the sulfate concentration had to be reduced since, although sulfate activates by binding at the Ser-14 phosphate recognition site, it also competes with substrate for the catalytic The crystals were transferred from 1 mol dm-3 sodium tartrate, 50 mmol dmA3ammonium sulfate and ammonium sulfate to 1mol 200 mmol dm-3 glucose-1-P, 10 mmol dm-3 @-glycerolphosphate, pH 7.5. Some sulfate was retained in the soak solution in order to keep the Ser-14 phosphate-recognition site saturated. The experiment had an additional complication in that the unit-cell edge in the c direction was doubled, giving rise to eight subunits (molecular weight 97 400) per asymmetric unit. Although a data collection and refinement (by XPLOR) were formidable, the analysis proceeded fairly routinely and resulted in a refined structure (R = 0.188 to 2.9 A resolution) that revealed the details of glucose-1-P binding. The conformation of glucose-l-P and the torsion angle about C( 1)-O( l ) were similar to those observed for the T-state glucose-1-P complex. There is no direct contact between the phosphate and the cofactor 5'-phosphate group (the closest approach between the phosphorus atoms is 6.4A). In the R state the glucose-1-P makes contact through the phosphate group to the side chain of Arg-569 in addition to contacts to the main-chain N of Gly-135 and Leu-136. Surprisingly Lys-574 does not make a direct contact; the distance (averaged from the eight subunits) from the NZ atom to the phosphate oxygen is 4.5A. There are small conformational changes between the glucose-1-P complex and the native R state that include a shift of Arg-569 by 1.8A, shifts in Lys-568 so as to weaken the contact between Lys-568 and the cofactor 5'-phosphate and to make a hydrogen bond with Glu-672, and a shift in Lys-574 away from Glu-672 and towards the cofactor 5'-phosphate(Fig. 3). The phosphate position of glucose-1-P is 3 from the position observed for the sulfate ion (which is expelled in the presence of tartrate). A third phosphate-recognition site has been identified in a crystallo raphic study of GPb reconstituted with a modified cofactor pyridoxal pyrophosphate.zgJ3 In the R-state ammonium sulfate crystals, the second phosphate covalently linked to the cofactor phosphate makes contact to Lys-574 and the main-chain N of Arg-569 but occupies a distinct site that is 3 A and 4.8 A from the sulfate and glucose-1-P sites, respectively.
View Article Online
Catalytic Mechanism of Glycogen Phosphorylase
Published on 01 January 1992. Downloaded by Christian Albrechts Universitat zu Kiel on 25/10/2014 12:09:13.
138
0
\/4-
-O'l
" dO-n-*o ; PXL'
'0
\o
eWOR
0
'"R
\"0'
PXL
" /" -i
/\do\p/o"
/ \o P X L P
(c)
Fig. 4 Proposed mechanism of phosphorylase catalysis for ( a) phosphorolysis of heptenitol, (b) phosphorolysis of oligosaccharide and (c) oligosaccharide synthesis. The S'-phosphate group of the pyridoxal (PXL) phosphate and the substrate phosphate are stabilised by the interactions shown in Fig. l(b) and Fig. 2. Further details are given in the text. Reproduced with permission from L. N. Johnson, K. R. Archaya, M. D. Jordan and P. J. McLaughlin, J. Mol. Biol., 1990, 29, 509.
Catalytic Mechanism The evidence leading to the proposals for the catalytic mechanism and the role of the ' The proposals put forward on cofactor 5'-phosphate group have been the basis of the crystallographic resultsz4with T-state GPb and now supported by studies with R-state GPb and R-state GPa complement those deduced from the results of kinetic and NMR experiments and demonstrate the additional contribution of stereoelectronic effects. The structures observed favour a mechanism [Fig. 4( a)] in which phosphorylysis of heptenitol is catalysed by general acid attack of the substrate phosphate promoted by the cofactor phosphate. The proton is donated by the substrate phosphate in a concerted reaction in which it immediately gains a proton from the cofactor phosphate. After protonation of the methyene carbon, the glucosyl carbonium ion is stabilised by the negatively charged substrate phosphate group. No covalent intermediate has been detected in phosphorylase catalysis and the structure of the protein provides no suitable
View Article Online
Published on 01 January 1992. Downloaded by Christian Albrechts Universitat zu Kiel on 25/10/2014 12:09:13.
L. N. Johnson, S-H.Hu and D. Barford
139
candidates for a covalent glucosyl intermediate, hence it is assumed that the reaction proceeds through a carbonium ion intermediate. The reaction is completed by nucleophilic attack of the substrate phosphate on the carbonium ion to give the product heptulose-2-P. The mechanism can be readily extended to the natural reaction as shown in Fig. 4( 6 ) . In the first step the substrate phosphate protonates the glycosidic oxygen resulting in the cleavage of the C(1)-O(1) bond. The glycosyl carbonium ion is stabilised by the phosphate dianion as before and the reaction completed by attack of the phosphate on the carbonium ion to give glucose-1-I? The crystallographic results show that there are no ionisable groups in the vicinity of the C(l) atom that could contribute to the stabilisation of the transition-state carbonium ion nor are there any protein groups in the vicinity of the O( 1) oxygen that could protonate the substrate (Fig. 2 and 3). Indeed the proximity of the main-chain N of Leu-136 to the ring oxygen O(5) is likely to discourage charge delocalisation to an oxonium-carbonium ion in the transition state and provides a rationalisation as to why classical inhibitors of many glycosidases such as norjirimycin are poor inhibitors of phosphorylase. The mechanism of phosphorylase is different from glycosidases such as lysozyme in the disposition of groups that promote general acid catalysis and which stabilise the transition state. In lysozyme these functions are performed by two acid residues (Glu-35 and Asp-52) whose properties are governed by their environment in the protein. In phosphorylase these functions are performed by the substrate phosphate promoted by the cofactor phosphate and the catalysis depends on the direct interaction of the substrate phosphate and the cofactor phosphate. The transition-state stabilisation by the substrate dianion with the confined and shielded environment of the catalytic site means that the reaction can proceed only in the direction of phosphorolysis and not hydrolysis. In the reverse reaction the cofactor phosphate acts as an acid to protonate the phosphate of glucose-1-P and the C(1)-0(1) bond has to be cleaved without direct protonation of the glycosidic oxygen [Fig. 4( c)]. The conformation observed for heptulose-2-P is similar to that predicted from stereoelectronic arguments to weaken the exo-anomeric effect and promote cleavage of the a-glycosidic bond.34 The theory predicts that an 0-C bond in the grouping 0-C-0 is strongest (ie. charge delocalisation is greatest) when the lone-pair orbital of the oxygen atom is antiperiplanar (180") to the C-0 bond of the other oxygen. For a-glycosides the unshared pair of electrons on the ring oxygen are properly disposed (ie. antiperiplanar to the glycosidic bond) to strengthen the C(1)-0(5) bond (endo-anomeric effect).and to contribute to charge delocalisation so as to assist the formation of the oxycarbonium ion without the need for distortion of the sugar ring (Fig. 5 ) . However, in the preferred conformation for a-glycosides, such as that observed in the single-crystal structure of glucose-1-P where the torsion angle 0(5)-C(l)-O(l)-P is 90" and both bond lengths C(1)-O(1) and C( 1)-0(1) are short, the torsion angle about C(1)-0(1) is such that the unshared pair of electrons on the glycosidic oxygen is antiperiplanar to the C(1)-O(5) bond. This will tend to strengthen the glycosidic bond (exo-anomeric effect) thus making cleavage of this bond more difficult. Rotation about the C( 1)-O( 1) bond to the position observed in heptulose-2-it> complex weakens the exo-anomeric effect and results in increasing polarisation of the C(1)-0(1) bond (Fig. 5). Thus heptulose-2-P formed in the crystal exhibits a conformation that is anticipated to promote cleavage of the glycosidic bond. The presence of the P-methyl group prevents access of oligosaccharide and probably explains why the reaction in the direction of glycogen synthesis is not observed with this compound despite its reactive conformation. For glucose-1-P it is anticipated that the presence of oligosaccharide in the catalytic site favours a conformation in which the phosphate is turned away from the conformation observed in the binary complex (as observed in the structural studies with both the T and R states) towards that observed in the heptulose-2-P complex so that the cofactor
View Article Online
Published on 01 January 1992. Downloaded by Christian Albrechts Universitat zu Kiel on 25/10/2014 12:09:13.
140
Catalytic Mechanism of Glycogen Phosphorylase
I
t I
Qoo
i/ p -0-
0
Fig. 5 A simplified diagram showing stereoelectronic effects for a-glycosides: ( a ) in the preferred conformation for a-glycosides both C(1)-0(5) (endo effect) and C(1)-0(1) (ex0 effect) are stabilised; ( b )in the conformationobserved for hsptulose-2-P the em-anomeric effect is weakened. Reproduced with permission from L. N. Johnson, K. R. Archaya, M. D. Jordan and P. J. McLaughlin, J. Mol. BioL, 1990, 29, 509
phosphate can function as a catalytic group with the assistance of the activated conformation predicted from stereoelectronic principles. Some support for this proposal has been given by the result for the a-D-glucosylmethylphosphonate complex obtained in the presence of oligosaccharide where the phosphate was directed significantly closer to the cofactor phosphate than in the Glc-1-Pcomplex.26 The crystallographic results suggest that when the reaction proceeds in the direction of glycogen synthesis the glycosidic bond is weakened by steric factors that facilitate the development of the carbonium ion and ensure that the phosphate is able to act as a base to abstract a hydrogen from the O(4) hydroxyl of the oligosaccharide. When the reaction proceeds in the direction of glycogen degradation the glycosidic bond is weakened by direct protonation. In both directions the substrate phosphate, stabilised by its interactions with the enzyme, plays a crucial role in the transition-state stabilisation. In an alternative proposal it has been suggested that the cofactor phosphate acts as an electrophile to withdraw electrons from the substrate phosphate and destabilise the glycosidic bond13 (reviewed in ref. 10). A closer association between the cofactor and substrate phosphates is required than has been observed in the crystallographic results to date. Such a close approach is mimicked by the analogue pyridoxal pyrophosphate but the structural show no evidence for additional interactions with the cofactor 5’-phosphate that could stabilise the constrained dianion which is an essential feature of this mechanism. However, as is generally true with X-ray crystallographic
View Article Online
Published on 01 January 1992. Downloaded by Christian Albrechts Universitat zu Kiel on 25/10/2014 12:09:13.
t.N.Johnson, S-H. Hu and D. Barford
141
studies, only thermodynamically stable structures can be examined and the existence of an unstable intermediate can never be definitely excluded. A survey of phosphate-recognition sites in proteins carried out in 1984;’ at a time when the structures of some 3 1 phosphate-recognition sites distributed among some 16 proteins were known, indicated differences between binding sites and catalytic sites. Binding sites, where the phosphate was solely involved in localisation of the ligand, were well located in the crystallographic studies and phosphate groups made contacts with one or more arginine side chains, the main-chain N at the amino termini of helices (helix dipole interactions) and less frequently, with lysine or neutral polar residues. At the catalytic sites the phosphate position was often less well defined and in some proteins appeared to make only indirect interactions with the protein. Phosphorylase contains four phosphate-recognition sites and it is now possible to extend some of these generalisations with comments on the consequences on phosphate recognition. (i) The Ser-14 phosphate site involves two arginine residues (Arg-69 from its own subunit and Arg-43’ from the other subunit) and this site is a tight binding site formed only in the R state. Although the major driving force for the interaction is electrostatic the biological effects for allosteric activation also involve the interactions and interdigitisation of the surrounding residues of the N-terminal tail with the rest of the (ii) The AMP allosteric phosphate-recognition site also involves two arginine residues (Arg-309 and Arg-310 on an a-helix) and is a tight-binding site recognised in both the T and the R states. There is scope for some mobility at this site created by the nearby presence of Arg-242 and in the R state the phosphate of AMP is shifted 1 A towards this residue compared with the T state while in the T state both inorganic phosphate and glucose-6-P utilise this third arginine in their binding modes.16*18 (iii) The 5’-phosphate recognition site for pyridoxal phosphate has the same interactions in both T and R state. The site involves only one basic group (Lys-568) and the helix dipole arising from a helix that commences with residues Thr-676 and Gly-677 (Fig. 2). Some fluctuations have been observed as in the glucose-l-Pcomplex described above. (iv) The catalytic site exhibits the greatest variability in phosphate positions. In the T state differentpositions are observed for the phosphates of glucose-l-P and heptulose2-P. In the R state despite the constellation of two basic side chains (Arg-569 and Lys-574) and the potential helix dipole ( N of Gly-135) different phosphate positions are observed that depend on whether the group is free as in the sulfate ion or bound to other groups as in glucose-l-P complex or pyridoxal pyrophosphate. Such mobility may be anticipated for the enzyme that both promotes attack by phosphate and, in the reverse direction, reacts with the covalent phosphorylated product. The position observed for glucose-l-P is consistent with the notion discussed earlier that the phosphate only adopts the ‘productive’ mode of binding in the presence of the oligosaccharide complex. The major outstanding problem for a complete understanding of the phosphorylase mechanism is to define the position of the oligosaccharide at the catalytic site. This is not easy since the affinity is low and no binding has been observed to date in diffusion experiments with both T- and R-state phosphorylases (unpublished results) or in a cocrystaflisation experiment with R-state phosphorylase. l 9 Model building experiments with R-state phosphorylase show that even with the open access to the catalytic site there is a requirement for some distortion from the preferred conformation of a-1,4-linked oligosaccharides. Attempts to crystallise active and inactive ternary complexes and to determine the structures of other phosphorylases are in progress. References 1 D. E. Koshland, Biol. Rev., 1953, 28, 416. 2 F. C. Kokesh and Y . Kakuda, Biochemistry, 1977, 16, 2467.
View Article Online
Published on 01 January 1992. Downloaded by Christian Albrechts Universitat zu Kiel on 25/10/2014 12:09:13.
142
Catalytic Mechanism of Glycogen Phosphorylase
3 H. D. Engers, W, A. Bridger and N. B. Madsen, Can. J. Biochem., 1970,48, 746. 4 D. J. Graves and J. H. Wang, in The Enzymes, ed. P. D. Boyer, Academic Press, New York, 3rd edn. 1972, VOI. 7, pp. 435-482. 5 E. Helmreich and C. F. Cori, Proc. Natl. Acad. Sci. USA, 1964, 51 131. 6 L. N. Johnson, J. Hajdu, K. R. Acharya, D. I. Stuart, P. J. McLauchlin, N. G. Oikonomakos and D. Barford, in Allosteric Enzymes, ed. G. Herve, CRC Pres, Boca Raton, Florida, 1989, pp. 81-127. 7 P. J. Kasvinsky, N. B. Madsen, R. J. Fletterick and J. Sygusch, J. Biol. Chem., 1978, 253, 1290. 8 G. Philip, G. Gringel and D. Palm, Biochemistry, 1987, 21,3042. 9 E. H. Fischer, A. B. Kent, E. R. Snyder and E. G. Krebs, J. Am. Chem. Soc., 1958,80, 2906. 10 N. B. Madsen and S. G. Withers, in Coenzymes and Cofactors; Pyridoxal Phosphate and Derivatives, ed. D. Dolphin, R. Paulson and 0. Avramovic, Wiley, New York, 1986, pp. 1-29. 11 D. Palm, H. W. Klein, R. Schinzel, M. Buehner and E. J. M. Helmreich, Biochemistry, 1990,29, 1099. 12 R. F. Parrish, R. J. Uhing and D. J. Graves, Biochemistry, 1977, 16, 139. 13 S. G. Withers, N. B. Madsen, B. D. Sykes, M. Takagi, S. Shimomura and T. Fukui, J. Biol. Chem., 1981, 256, 10759. 14 K. R. Acharya, D. I. Stuart, K. M. Varvill and L. N. Johnson, Glycogen Phosphorylase b: Description of the Protein Structure, World Scientific Publishers, Singapore, 1991. 15 S. R. Sprang and R. J. Fletterick, J. Mol. Biol., 1979, 131, 523. 16 S. R. Sprang, K. R. Acharya, E. J. Goldsmith, D. I. Stuart, K. M. Varvill, R. J. Fletterick, N. B. Madsen and L. N. Johnson, Nature (London), 1988,336, 215. 17 D. Barford and L. N. Johnson, Nature (London), 1989, 340,609. 18 D. Barford, S-H. Hu and L. N. Johnson, J. Mol. Biol., 1991, 218, 233. 19 S. R. Sprang, S. G. Withers, E. J. Goldsmith, R. J. Fletterick and N. B. Madsen, Science, 1991,254,1367. 20 P. J. Kasvinsky and N. B. Madsen, J. Biol. Chem., 1976, 251, 6852. 21 E. J. Hehre, C. F. Brewer, T. Uchiyama, P. Schlesselman and J. Lehmann, Biochemistry, 1980,19,3557. 22 H. W. Klein, M. J. Im and D. Palm, Eur. J. Biochem., 1986, 157, 107. 23 J. Hajdu, K. R. Acharya, D. I. Stuart, P. J. McLauchlin, D. Barford, N. G. Oikonomakos, H. W. Klein & L. N. Johnson, EMBO J., 1987,6, 539. 24 L. N. Johnson, K. R. Acharya, M. D. Jordan and P. J. McLaughlin, J. Mol. Biol., 1990, 211, 645, 25 B. Vandenbunder and H. Buc, Eur. J. Biochem., 1983, 133, 509. 26 J. L. Martin, L. N. Johnson and S. G. Withers, Biochemistry, 1990, 29, 10745. 27 J. L. Martin, K. Veluraja, K. Ross, L. N. Johnson, G. W. J. Fleet, N. G. Ramsden, I. Bruce, M. G. Orchard, N. G. Oikonomakos, A. C. Papageorgiou, D. D. Leonidas and H. S. Tsitoura, Biochemistry, 1991, 31, 10101. 28 D. Barford and L. N. Johnson, Protein Science, in the press. 29 D. D. Leonidas, N. G. Oikonomakos, A. C. Papageorgiou, K. R. Acharya, D. Barford and L. N.
Johnson, Protein Science, in the press. 30 S-H. Hu, D. Phil. Thesis, University of Oxford, 1991. 31 D. D. Leonidas, N. G. Oikonomakos, A. C. Papageorgiou, A. Xenakis, C. T. Cazianis and F. Bem, FEBS Lett., 1990, 261, 23. 32 D. D. Leonidas, N. G. Oikonomakos and A. C. Papageorgiou, Biochim. Biophys. Acta, 1991,1076,305. 33 S . R. Sprang, N. B. Madsen and S. G. Withers, Protein Science, in the press. 34 J. P. Praly and R. U. Lemieux, Can. J. Chem., 1987, 65, 213. 35 L. N. Johnson, in Inclusion Compounds, ed. J. L. Atwood, J. E. D. Davies and D. D. Macnichol, Academic Press, New York, 1984, vol. 3 pp. 509-569.
Paper 2f00131D; Received 10th January, 1992
