- To design and synthesise potent glucose analogue inhibitors of glycogen phosphorylase
- To study the molecular interactions of these compounds with glycogen phosphorylase by X-ray crystallography and to characterise their kinetic inhibition constants
- To explore the effects of the inhibitors on glycogen metabolism under physiological conditions and to assess their potential as therapeutic agents for treatment of diabetes
- To contribute to a better understanding of enzyme-carbohydrate recognition through binding studies of oligosaccharide substrates with E. coli maltodextrin phosphorylase
In this collaborative EU project involving 6 laboratories, knowledge of the crystal structure of rabbit muscle glycogen phosphorylase in the glucose bound state has been used to design glucose-like agents that might be more potent regulators of glycogen phosphorylase than glucose itself. New C1-substituted glucopyranose analogues have been synthesised, often utilising novel chemistry (Fleet, Oxford). The essential inhibitory and binding properties of these synthetic analogues with phosphorylase have been established by enzyme kinetic studies (Oikonomakos, Athens), X-ray crystallographic studies (Johnson, Oxford and Oikonomakos, Athens) and physiological studies (Stalmans, Leuven). The work has provided a data base for assessment of quantitative predictive structure/function methods (Clementi, Perugia).
In order to further elucidate protein-oligosaccharide interactions, the crystal structure of E. coli maltodextrin phosphorylase has been determined at 2.3 Å resolution (Johnson, Oxford and Palm, Wurzburg). The structure has provided insights into the evolution of phosphorylase regulatory properties. Site directed mutagenesis studies (Wurzburg) and cocrystallisation with oligosaccharide (Oxford) have provided new information on the key interactions between protein and oligosaccharide that govern substrate specificity.
Glucose analogue inhibitors of glycogen phosphorylase
Glucose is a significant regulator of phosphorylase activity in the liver. It not only acts as a competitive inhibitor but also promotes the action of phosphorylase phosphatase to inactivate phosphorylase a by making the phosphorylase a a better substrate for the phosphatase. More potent inhibitors could have relevance to the regulation of glycogen metabolism and hence the control of hepatic output of glucose in the treatment of Type II diabetes. Examination of the crystal structure of the rabbit muscle glycogen phosphorylase molecule in complex with glucose showed that enzyme contacts in the buried site allowed little scope for modification of the glucopyranose molecule except at the anomeric C1 position. One of the early successes was the design and synthesis of N-acetyl-beta-D-glucosylpyranosylamine (Glcbeta-1-NAc), which exhibited a Ki of 32uM, approximately 60 times better than glucose (Ki = 1.7 mM for alpha-D-glucose) (Watson et al. 1995; Oikonomakos et al. 1995). The amide made a strong hydrogen bond to a carbonyl oxygen of His377, a group whose hydrogen bonding capacity was otherwise unsatisfied, and the methyl group displaced 2 water molecules. The synthesis of large quantities of this compound was readily achieved and it has been used extensively in physiological studies.
Physiological studies showed that Glcbbeta-1-NAc is indeed a potent inhibitor of liver phosphorylase in rat hepatocytes and that it was able to enhance the action of phosphorylase phosphatase on phosphorylase a, as predicted (Board et al., 1995a,1995b, 1995c) However Glcbbeta-1-NAc did not lead to activation of glycogen synthase, the rate limiting enzyme involved in glycogen synthesis. In hepatocytes, glucose, in addition to its ability to inhibit phosphorylase, also activates glycogen synthase, the mechanism for which have been tested using 2-deoxy, 2-F-alpha-D-glucopyranosyl fluoride (Masillon et al., 1995). Glucose is rapidly phosphorylated in hepatocytes by glucokinase to form glucose-6-phosphate which is an activator of glycogen synthase. Glcbeta-1-NAc is also phosphorylated to Glc-6-P-beta-1-NAc and this compound was shown to be a potent inhibitor of the phosphatase activation of glycogen synthase (Board, 1997). This has led to the synthesis of 7 carbon mimics of D-glucose (Bleriot et al. 1996). Substitution of the prochiral hydrogens by a methyl group led to compounds neither of which inhibited phosphorylase but one (the C6R-C-methyl glucose) was found to inhibit glucose-6-phosphatase, leading to increased glucose-6-P levels. Glucose-6-phosphatase is a complex membrane bound microsomal enzyme and the mechanisms for transport and release of glucose-6-P from microsomes together with studies on the protein phosphatase are under further investigation.
The lead to the best glucose analogue inhibitor of phosphorylase came from natural product chemistry. The furanose compound hydantocidin is a naturally occurring spironucleoside isolated from Streptomyces hygroscopicus that exhibits potent but non-toxic herbicide activity. Although many furanose analogues of hydantocidin have been produced, the glucopyranose analogues present considerable challenges. It was predicted that the corresponding hydantoin glucopyranose would be a potent inhibitor of phosphorylase and this was confirmed when organic chemistry studies led to the synthesis of the compound and kinetic and crystallographic binding studies (Bichard et al, 1995; Brandstetter et al., 1996a; Krulle et al. 1996; de La Fuente et al., 1997; Gregoriou et al. ms in preparation). The Ki for the glucopyranose hydantoin is 3.1 uM, more than 500 times lower than the corresponding Ki for glucose. The compound utilised the same hydrogen bond to the CO of His377 as in Glcbeta-1-NAc and made additional hydrogen bonds from its own CO group which partially mimic those of the alpha-hydroxyl group of alpha-D-glucose. The rigid nature of the substituent group means that there is little conformational energy change on binding, a further favourable factor.
Synthetic organic chemistry studies showed that a number of epimeric spirohydantoins of glucofuranose could be prepared from glucoheptononlactone. Although the furanose analogues did not inhibit phosphorylase, the synthetic routes have provided novel and efficient syntheses of new compounds of interest to other carbohydrate recognition enzymes. The spirodiketopiperazine of glucose and a number of modified hydantoin compounds have also been prepared (Krulle et al., 1995; Brandstetter et al., 1995; Brandstetter et al., 1996b & 1996c).
The binding and kinetic studies have yielded a data base of some 70 compounds for which details of both biological activity and protein interactions that govern binding are known. The data base has been used as a test data set in theoretical QSAR studies (Watson et al., 1995). The methods correlate biological activity with the known structures of the compounds but without reference to their interactions with the enzyme. The methods maps regions around the inhibitors that carry the most important information for biological activity (Cruciani et al., 1997). The predictivity of the methods has been improved when certain water molecules were added as part of the inhibitor structure (Pastor et al., 1997a & 1997b).
Colleagues at the Bayer Pharmaceutical (Wuppertal) had discovered a potent inhibitor of phosphorylase (Ki 1.6 nM with respect to AMP) whose chemical structure bore no resemblance to either glucose or other known regulators of phosphorylase activity. A cryocrystallographic binding study at 2.3 Å resolution revealed that the Bayer compound BAY W1807 bound at the AMP allosteric site of phosphorylase (Zographos et al., ms in preparation). The compound fits neatly into the site and it is remarkable that the constrained geometry of the inhibitor allows it to exploit numerous interactions. The carboxyl groups of the inhibitor make contact to 2 arginine residues Arg309 and Arg310 and partly mimic the contacts made by phosphate groups on binding compounds such as the activator AMP or the inhibitor Glc-6-P at this site. The contacts are dominated by non-polar interactions from the inhibitor to the enzyme. These non-polar contacts appear to provide the major difference which result in a nanomolar, rather than a micromolar, inhibitor. The inhibitor, a non-toxic active metabolite, is under further investigation.
E. coli maltodextrin phosphorylase:
The crystal structure of the E. coli maltodextrin phosphorylase has been solved at 2.3 Å resolution (Watson et al., 1997) The bacterial phosphorylase is controlled by induction at the level of gene expression and is constitutively active, in contrast to the mammalian enzyme which exhibits multiple regulatory mechanisms utilising 5 different control sites. Expression and purification of large amounts of recombinant E. coli maltodextrin phosphorylase were optimised for both wild type and a thermostable mutant enzyme. The bacterial and mammalian enzymes have an overall sequence identity of 46%. Structure determination has shown that the folds of the 2 proteins are similar and the catalytic sites are almost 100% conserved. Each of the regulatory sites of the mammalian phosphorylase has sequence changes in the bacterial enzyme which lead to destruction of the control sites. In particular the E. coli phosphorylase is kept locked in a permanently active state by subunit-subunit contacts which result in a loop of residues, known as the 280s loop, being hooked out of the way of the catalytic site. The structure provides insights into the creation of regulatory sites through key amino acid changes in surface loops and through changes in subunit/subunit contacts.
Oligosaccharide substrate recognition by phosphorylase is the major outstanding problem which needs to be solved to provide a better understanding of phosphorylase catalysis. Extensive site directed mutagenesis studies have been carried out with the E. coli maltodextrin phosphorylase, in order to define the roles of different amino acids in catalysis and in substrate recognition. In particular 9 residues that line the catalytic site tunnel were mutated and their contributions to oligosaccharide recognition assessed together with studies on the action pattern of binding primer molecules (Becker et al., 1995; Drueckes et al., 1996a & 1996b; Drueckes & Schinzel 1996). The structural studies with E. coli maltodextrin phosphorylase have been extended with oligosaccharide binding studies in the crystal. In conformity with its biological role E coli maltodextrin phosphorylase exhibits a 100 fold higher affinity for linear oligosaccharides than the mammalian glycogen phosphorylase. We have successfully cocrystallised maltodextrin phosphorylase with oligosaccharide and observe sugars bound in the catalytic site tunnel (O'Reilly et al., 1997). The most dramatic feature of the protein-oligosaccharide recognition is the stacking of a sugar 2 subsites removed from the catalytic site against a tyrosine residue, Tyr280. The importance of this interaction for substrate recognition was demonstrated in a site directed mutagenesis experiment in which the mutant Tyr280Ala exhibited negligible affinity to oligosaccharide substrate. These exciting observations open the way for a detailed understanding of the factors that provide oligosaccharide recognition and the conformation of alpha-1, 4-glucosyl substrates at the binding site of enzymes.
The use of cryo methods in protein crystallography has become almost routine in the last 2 years. Crystals flash frozen to liquid nitrogen temperatures (80 to 100 K) become almost immortal in the X-ray beam permitting data collection from a single crystal with minimal radiation damage. In this EU programme we have successfully utilised such methods and pioneered some aspects of instrument development with the aid of a technician supported on the grant. A device for mounting crystals under nitrogen gas (known as a 'top hat' device) and a new cryoarc modification of the goniometer head have been designed and manufactured. The latter is used to facilitate storage of crystals for cryocrystallographic work and has proved a valuable aid enabling crystals to be tested on the home source and the most promising them stored under liquid nitrogen for radiation at synchrotron sources. Crystals require some cryoprotectant. The most commonly used cryoprotectant is glycerol but this proved to bind at the phosphorylase catalytic site and to compete for glucose analogue binding (Mitchell et al., 1995). 2-Methyl-2,4-pentanediol (MPD) was found to be a useful alternative cryoprotectant. High resolution data (2 Å or better) for the glycogen phosphorylase complexes with the hydantoin (Gregoriou et al., ms in preparation) and the 2-deoxy-2-methoxycarbamido-beta-D-gluco-2-heptulopyranosonamide (Ki = 16 uM) were collected under these conditions to provide high precision description of the inhibitor interactions and as were all the data for the E. coli maltodextrin phosphorylase.
Funding SchemeCSC - Cost-sharing contracts
3000 Louvain / Leuven