Biosynthesis of Bacterial Glycogen USE OF SITE-DIRECTED MUTAGENESIS TO PROBE THE ROLE OF TYROSINE 114 IN THE CATALYTIC MECHANISM OF ADP-GLUCOSE SYNTHETASE FROM ESCHERICHIA COLI*

Previous covalent modification studies showed that tyrosine 114 of Escherichia coli ADP-glucose synthetase is involved in substrate binding (Lee, Y. M., and Preiss, J. (1986) J. Biol. Chem. 261, 1058-1064). We have prepared, via site-directed mutagenesis, an E. coli ADP-glucose synthetase variant (Phe114) containing a Tyr114 to Phe substitution in order to test whether the phenolic hydroxyl group plays a critical role in catalysis. Kinetic characterization of Phe114 ADP-glu-cose synthetase indicates that the Tyr114 hydroxyl is not obligatory for the enzyme catalysis. However, the variant enzyme showed altered properties. It showed a decreased apparent affinity for the substrates. The variant enzyme showed less than 2-fold activation by 5 mM fructose 1,6-bisphosphate in the ADP-glucose synthesis direction. In contrast, in the pyrophosphorolysis direction, the mutant enzyme showed about a 30-fold activation by 5 mM fructose 1,6-bisphosphate. The variant enzyme is heat-labile compared to wild type enzyme. It lost about 60% enzyme activity on incubation at 65 “c for 5 min in the presence of 30 mM Pi. The wild type enzyme is stable under these condi- tions. The results indicate that tyrosine 114 is involved directly or indirectly in enzyme catalysis, but is not obligatory


USE OF SITE-DIRECTED MUTAGENESIS TO PROBE THE ROLE OF TYROSINE 114 IN THE CATALYTIC MECHANISM OF ADP-GLUCOSE SYNTHETASE FROM ESCHERICHIA
COLI* (Received for publication, March 29, 1988) Ani1 Kumar, Toshio Tanaka$, Young Moo Lee §, and Jack Preissn From the Department of Biochemistry,Michigan State University,East Lansing,Michigan 48824 Previous covalent modification studies showed that tyrosine 114 of Escherichia coli ADP-glucose synthetase is involved in substrate binding  J. Biol. Chem. 261,[1058][1059][1060][1061][1062][1063][1064]. We have prepared, via site-directed mutagenesis, an E. coli ADP-glucose synthetase variant (Phe114) containing a Tyr114 to Phe substitution in order to test whether the phenolic hydroxyl group plays a critical role in catalysis. Kinetic characterization of Phe114 ADP-glucose synthetase indicates that the Tyr114 hydroxyl is not obligatory for the enzyme catalysis. However, the variant enzyme showed altered properties. It showed a decreased apparent affinity for the substrates. The variant enzyme showed less than 2-fold activation by 5 mM fructose 1,6-bisphosphate in the ADP-glucose synthesis direction. In contrast, in the pyrophosphorolysis direction, the mutant enzyme showed about a 30-fold activation by 5 mM fructose 1,6-bisphosphate. The variant enzyme is heat-labile compared to wild type enzyme. It lost about 60% enzyme activity on incubation at 65 "c for 5 min in the presence of 30 mM Pi. The wild type enzyme is stable under these conditions. The results indicate that tyrosine 114 is involved directly or indirectly in enzyme catalysis, but is not obligatory for the enzyme catalysis.
Conversion of Tyr114 to Phe also alters the regulatory properties of the enzyme with respect to activation by fructose-1,6-P2 and inhibition by AMP.
ADP glucose synthetase (EC 2.7.7.27) catalyzes the reaction, ATP + a-glucose 1-phosphate + ADP-glucose + PPi, the rate-limiting step in bacterial glycogen syntheses (1-4). The enzyme from Escherichia coli has been purified and characterized (1-3,5-7). The structural gene for ADP-glucose synthetase has been cloned (8) and sequenced (9). Recently, covalent modification studies using azido analogues of the substrate, 8-azido-ADP-glucose (lo), and inhibitor, 8-azido-AMP (11), showed that tyrosine 114 is present at or near the substrate and/or inhibitor binding site. Here we prepared a Phe114 variant ADP-glucose synthetase using directed sitespecific mutagenesis (12), purified, and characterized the en-* This research was supported in part by United States Public Service Grant AI 22835. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
ll To whom correspondence and reprint requests should be addressed. zymatic activities of the variant enzyme to elucidate the role of Tyr114 in the enzyme catalysis. EXPERIMENTAL PROCEDURES'

RESULTS AND DISCUSSION
Directed Mutagenesis-Oligonucleotide-directed mutagenesis was carried out using the techniques as described by Zoller and Smith (12). A 1.9-kilobase pair HincII restriction fragment from the plasmid pOP12 (8) containing the complete wild type E. coli ADP-glucose synthetase gene was cloned into bacteriophage M13mp18. An oligonucleotide, 3' TTG ACC AAA GCG CCG 5', designed to produce the desired mutation was then used as primer on the single-stranded phage DNA. For rapid extension of the DNA, two other oligonucleotides, AGC ACC ATT ACC ATTAG CAAGGCC and CGT AAG AAT ACG TGG CAC AGA CA, which have complementary sequences in M13mp18 at positions 2553-2576 and 5027-5049, were also added in the reaction mixture. Following extension and ligation by the Klenow fragment of E. coli DNA polymerase I and T4 DNA ligase, the resulting covalently closed heteroduplex DNA was isolated and transfected in E. coli strain JM101. Plaques were picked and mutants were identified by dot-blot hybridization (32P-labeled oligonucleotide hybridization with nitrocellulose-bound phage DNA). Dissociation temperature difference was used to distinguish wild type from mutant phage. At this stage, a single plaque may contain a mixture of wild type and mutant phage, suspected plaques were purified to obtain homogeneous mutants. The mutant cDNA coding for Phe114 ADP-glucose synthetase was sequenced in its entirety by the chemical method of Maxam and Gilbert (13).
One mutant out of 42 plaques tested was found as tested by dot-blot hybridization. This mutant selected by dot-blot hybridization showed the desired nucleotide sequence.
glgC Gene Expression-The mutant gene, along with the wild type glgC gene (1.9-kilobase pair HincII restriction fragment containing complete glgC gene) cloned in M13mp18, was transfected in E. coli strain G6MD3, a deletion mutant with no glg genes. After growth, the cells were harvested a t 0-4 "C and a 0.5-g cell paste was suspended in 10 ml of 0.05 M glycyl-glycine buffer, pH 7.0, containing 5 mM DTT2 and 1 mM EDTA, and sonicated. The suspension was centrifuged a t 12,000 X g for 10 min in a refrigerated Sorvall RC-5B centrifuge. The supernatant which contained almost 100% activity of the homogenate was used as a source of the enzyme.
Portions of this paper (including "Experimental Procedures," Figs. 1-5, and Tables 1-111) are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are included in the microfilm edition of the Journal that is available from Waverly Press.
The abbreviation used is: DTT, dithiothreitol.
Enzyme assay was done in the pyrophosphorolysis direction as described earlier (1). In the crude extract, the mutant enzyme showed only 5.8% of the specific activity observed with the wild type enzyme. Production of the wild type enzyme as well as mutant enzyme, was also achieved by cloning a HincII fragment containing the glgC gene from M13mp18 into the plasmid pUC8. The recombinant pUC8 was first transformed in E. coli strain JM101, and screening of the recombinants was done by blue-white screening using isopropyl-1-thio-P-D-galactopyranoside and X-Gal. White recombinants were also tested by I2 staining and, after growth in Kornberg media, miniscreened (14) and restriction-digested for checking the orientation of the fragment in pUC8. In each case, recombinant plasmid DNA having the proper orientation (N-terminal codon of the gene toward the promotor of the plasmid) was isolated using the alkali lysis method (15). Crude plasmid was extracted with an equal volume of buffer-saturated phenol followed by extraction with ether. DNA was precipitated by adding an equal volume of isopropyl alcohol and storing for 5 min at room temperature. DNA was collected by centrifugation at 20 "C and resuspended in 50 mM Tris-HC1, p H 7.5, l mM EDTA and purified by cesium chloride density gradient centrifugation. The purified recombinant DNA was transformed in the deletion host E. coli G6MD3. Recombinant selection was done with ampicillin. Ampicillin-resistant E.
coli colonies were picked, grown in Kornberg media, miniscreened, restriction-digested, and ultimately tested for the enzyme activity after harvesting the cells. The cell-free extract was prepared in the same way as described above. Here, the activity of the mutant enzyme was almost similar to the wild type enzyme activity when assayed at higher concentrations of ADP-glucose, fructose-1,6-P2, and MgC12 than the wild type enzyme. At present, the reason for the difference in expression of enzyme activities of the HincII insert in pUC8 and M13mp18 is not known. DNA sequencing has been done to verify the right mutation after isolating the pUC8 recombinant DNAs from E. coli G6MD3.
Purification of the Mutant ADP-glucose Synthetase-The purified enzyme is estimated to be 90-95% pure. On sodium dodecyl sulfate-polyacrylamide gel electrophoresis, the enzyme showed only one major band corresponding to the standard ADP-glucose synthetase band and two minor bands. On the native polyacrylamide gel electrophoresis, the enzyme gave only one band corresponding to the wild type enzyme band position which showed enzyme activity with the activity stain (16).
Heat Stability of the Mutant Enzyme-The wild type ADPglucose pyrophosphorylase in the presence of 30 mM Pi is stable to heat treatment for 5 min at 65 "C. In contrast, the mutant enzyme is 60% inactivated.
Kinetic Characterization of the Mutant Enzyme: Apparent Substrate Affinity-The mutant enzyme did not show significant activity when the usual substrate concentrations of the normal enzyme were used in measuring enzyme activity due to its decreased apparent affinity for the substrates. Fig. 1 shows the saturation curves for ATP in the presence of the activator, fructose-1,6-P2. The ATP curve is sigmoidal for both the normal and mutant enzymes. However, the concentration needed for half-maximal activity, So.5, is 2.8 mM for the mutant enzyme as compared to 0.26 mM for the normal enzyme. Thus, there is about an 11-fold decrease in the apparent affinity of the Phe114 mutant enzyme for ATP. Of interest is whereas the presence of activator, fructose-1,6-P2, decreases the S0.s value 5-fold for the normal enzyme, it essentially has no effect on the So.s value obtained for ATP.
The value for MgC12 (Table 11) for the Phe114 mutant enzyme is about 13-fold higher, 19 mM, than the normal, wild type enzyme which is 1.4 mM. As observed with ATP, fructosel,6-P2 does not affect the MgCh s 0 . 5 value for the Phe mutant enzyme. In contrast, the activator decreases the MgClz s 0 . 5 value of the normal enzyme 9-fold to 1.4 mM.
The K,,, for glucose-1-P for the mutant enzyme is very similar to that found for the normal, wild type ADP-glucose synthetase (Table 11) in the presence of activator, fructosel,6-P2. However, in the absence of activator, the s 0 . 5 value of glucose-1-P (420 PM) for the Phe114 enzyme is about 7-fold higher than that observed for the wild type enzyme. Moreover, the wild type enzyme glucose-1-P saturation curve shows negative cooperativity as its interaction coefficient in the Hill plot is 0.66, a value obtained in earlier studies (17).
In the pyrophosphorolysis direction, higher SOL, values were observed for ADP-glucose, PPi, and MgC12 for the Phe114 mutant enzyme when compared to the wild type enzyme (Fig.  2 and Table 11). Values were obtained only in the presence of activator since there was insufficient activity for accurate assay in the pyrophosphorolysis direction.
Effector Kinetics-Indeed, the kinetic parameters of both the activator and inhibitor were also modified in the Phe114 enzyme. In the synthesis direction, fructose-1,6-P2 was only able to activate the mutant enzyme about 1.8-fold and much higher concentrations (3 mM) are needed to obtain maximum activation. Half-maximal activation occurs at about 1.1 mM (Fig. 3). In contrast, the normal E.
Of interest is that the mutant Phe114 enzyme is highly dependent on the presence of activator in the pyrophosphorolysis direction. High concentrations of fructose-1,6-P2 (5-7 mM) stimulate the activity 30-fold (Fig. 4). However, the A05 value for fructose-1,6-P2 is 2.3 mM, which is about 60-fold higher than the A0.5 value for fructose-1,6-P2 for the normal ADP-glucose pyrophosphorylase. Table I11 shows that the activator specificity of the Phe114 enzyme is essentially similar to that of the wild type enzyme. The superior activators are fructose-l,6-P2 and 1,6-hexanediol bisphosphate. NADPH, Penolpyruvate, and 2-P-glycerate, activators of the normal wild type enzyme (1-4), activate the Phe114 enzyme to a small extent. Fructose-6-P, glucose-6-P, and 3-P-glycerate did not activate the mutant enzyme. They were shown earlier not to be activators of the normal enzyme (1-4).
Previous studies (3,10, 11) have suggested that the binding site for the inhibitor, AMP, and the adenine nucleotide substrates share one binding domain near Tyr114. Thus, it was of interest to determine the kinetics of AMP inhibition of the Phe114 mutant enzyme. Fig. 5 and Table I1 show that the 10.5 value for the mutant and wild type enzyme is the same, 66 pM. However, Hill plots of the data show that the shape of the inhibition curves are dissimilar. The wild type enzyme has a sigmoidal inhibition curve with a Hill ri coefficient of 1.7, while the Phe114 mutant enzyme inhibition curve shows negative cooperativity with a Hill fi of 0.5. Thus, the mutant enzyme shows more sensitivity to inhibition at low AMP concentrations, but is more resistant to inhibition at AMP concentrations greater than 66 PM. The native enzyme is inhibited greater than 90% at 300 PM, while the mutant enzyme is inhibited 74%.
Thus, the single site mutation at amino acid residue 114 from tyrosine to phenylalanine not only lowers the apparent binding affinity of the substrates ADP-glucose, ATP, PPi, and the divalent cation, M P , but it also affects the apparent affinity of the activators and alters the pattern of AMP inhibition. In addition, the catalytic activity of the enzyme decreased in terms of specific activity 33% in pyrophosphorolysis and about 70% in the synthetic direction (Table 11). These observations are consistent with the suggestions of previous experiments employing 8-azido-ATP and &azido-ADP-glucose as photoaffinity analogues in chemical modification studies (10, 11, 18). These studies suggested that the adenine nucleotide substrates shared a common site with the inhibitor, AMP, and that incorporation of 8-N3-AMP into the protein was at amino acid residues close to the activator binding sites (6, 7, 11). Thus, it is believed that the activator, substrate, and inhibitor sites were juxtaposed in the threedimensional structure of the enzyme.
The major binding site for 8-N3-AMP and the 8-Na-adenine nucleotides were at Tyr114 (10,ll) and Chou-Fasman analysis (19) predicted a Rossmann fold supersecondary structure (20-23) where the Tyr114 residue is located. Thus, the in uitro mutation of Tyr114 to Phe results in the expected decrease in apparent affinity of the adenine nucleotides and the additional effects seen on the kinetics of the activator and inhibitor are also consistent with the view that inhibitor, activator, and adenine nucleotide binding sites are proximal to each other in the enzyme's tertiary structure. Alteration of the substrate adenine nucleotide binding site by the amino acid substitution could cause a change in the interaction of the activator, inhibitor, and substrate binding sites where the lower apparent affinity of the substrates causes similar alterations of the binding of the activator and inhibitor. Previous binding studies (3) have shown that binding of the substrate ATP alone or activator alone has no effect on the binding of the inhibitor. Indeed, there is a slight stimulation of AMP binding by subsaturating concentrations of ATP (3). HOWever, the presence of ATP and fructose-1,6-P2 together effectively inhibits the binding of the inhibitor, thus indicating an interplay of the three separate sites.
Oligonucleotide directed mutagenesis thus can play an important role in elucidating the structure-function relationships in various residues and domains that have been identified via previous chemical modification studies (6, 7, 10, 11) of the ADP-glucose synthetase or by isolation of mutants of the E. coli ADP-glucose synthetase affected in their regulatory properties and where the amino acid changes have been determined (24). 14.