Active Site Lysine Promotes Catalytic Function of Pyridoxal 5"Phosphate in a-Glucan Phosphorylases*

Tight contact of pyridoxal 5"phosphate to the sub- strate phosphate is considered to be a crucial requirement of the phosphorolytic cleavage of polysaccharides by glycogen phosphorylases. This study demonstrates an essential role of lysine 533, the only charged residue in hydrogen bond distance to the phosphate of pyridoxal 5"phosphate in Escherichia coli maltodextrin phosphorylase. Substitution of Lys633 by Ser reduced the turnover number 600-fold. Addition of monovalent cations significantly increased activity of the LYS'~~-Ser mutant up to a factor of 10, whereas the apparent affinity for Pi was decreased up to 80-fold. Although substitution of Lys633 by Gln caused a similar reduction of k,,,, the K,,, values remained unchanged, and no response to small cations was observed. These results suggest a key role of the positive charge contributed by L Y S ' ~ ~ in catalysis, most probably in maintaining the electrostatic neutrality of the pyridoxal 5"phos-phate and aligning the close phosphate-phosphate con- tacts indispensable for the proton transfer mechanism. The activity of a-glucan phosphorylases (EC 2.4.1.1) is dependent on pyridoxal 5"phosphate (pyridoxal-P)' which, in carbohydrate phosphorolysis, acts quite differently from its role in other pyridoxal-P dependent enzymes. Numerous studies with pyridoxal-P derivatives, 31P NMR spectroscopy, one The exchanges were confirmed by concentration was deter- mined by the coupled assay described above. Heat inactivation kinetics in absence and presence of 100 mM cations at 60 "C exhibited no differences between the Lys"'"-Ser mutant and the wild-type enzyme. The modified assay allowed to determine the reaction velocity in absence of monovalent alkali cations and to measure the effects of the different monovalent cations on activity.

Tight contact of pyridoxal 5"phosphate to the substrate phosphate is considered to be a crucial requirement of the phosphorolytic cleavage of polysaccharides by glycogen phosphorylases. This study demonstrates an essential role of lysine 533, the only charged residue in hydrogen bond distance to the phosphate of pyridoxal 5"phosphate in Escherichia coli maltodextrin phosphorylase. Substitution of Lys633 by Ser reduced the turnover number 600-fold. Addition of monovalent cations significantly increased activity of the L Y S '~~-Ser mutant up to a factor of 10, whereas the apparent affinity for Pi was decreased up to 80-fold. Although substitution of Lys633 by Gln caused a similar reduction of k,,,, the K,,, values remained unchanged, and no response to small cations was observed. These results suggest a key role of the positive charge contributed by L Y S '~~ in catalysis, most probably in maintaining the electrostatic neutrality of the pyridoxal 5"phosphate and aligning the close phosphate-phosphate contacts indispensable for the proton transfer mechanism.
The activity of a-glucan phosphorylases (EC 2.4.1.1) is dependent on pyridoxal 5"phosphate (pyridoxal-P)' which, in carbohydrate phosphorolysis, acts quite differently from its role in other pyridoxal-P dependent enzymes. Numerous studies with pyridoxal-P derivatives, 31P NMR spectroscopy, and with a new class of "glycosylic" substrate analogs demonstrated that not the formation of a Schiff base intermediate, but close contacts to the pyridoxal-P 5"phosphate to the substrate phosphate is indispensable for activity (for review see Refs. 1-3). Hydrogen bond contacts enable the 5"phosphate of pyridoxal-P to act reversibly as a proton donor with the substrate phosphate as acceptor (3). However, the functional participation of individual active site residues in promoting the close phosphate-phosphate contact remained unclear.
X-ray crystallography of the phosphorylase b-heptulose 2phosphate complex revealed a cluster of positively charged amino acids around the active site (4). Among those, only Lys'"' is in hydrogen bond distance to the 5"phosphate of pyridoxal-P, but there are no direct hydrogen bonds of Lys"' to the substrate phosphate or the sugar-phosphate (Fig. 1). Since glycogen phosphorylases are highly conserved in euand prokaryotes (5) and the amino acid sequences at the catalytic site are essentially identical ( 6 ) , the role of the corresponding Lys5"" in the unregulated maltodextrin phosphorylase was probed by site-directed mutagenesis. In the present study different kinetic properties of the Lys"""-Ser and the Lys'""-Gln maltodextrin phosphorylase mutants are reported and a structure-function relationship for Lys""" is suggested.

MATERIALS AND METHODS
Strains and Plasmids-Escherichia coli strain pop2158 [malA518, F-, araD139, lacU169, rpsL, relA, thiA] was kindly provided by Dr. M. Schwartz, Institut Pasteur, Paris. Plasmid pMAPlOl was constructed from plasmid pOM13 as described before (7,8 DNA Procedures-DNA was manipulated by standard procedures (9). Restriction endonucleases, T4 DNA-ligase, and polynucleotide kinase were used as recommended by the manufacturer (Boehringer Mannheim). Transformation of E. coli was performed by the method of Hanahan (10). Oligonucleotides were synthesized on an Applied Biosystem 380A DNA synthesizer and purified by preparative electrophoresis on a denaturating 20% polyacrylamide gel.
Mutagenesis-Site-directed mutagenesis was accomplished by the gapped duplex approach (11) on the closed malP as described (12) with a 22-base pair oligonucleotide "ATTCAGATCTCACGTTTG-CACG" in which the AAA codon for Lys"":' was replzed by a TCA codon for Ser. This exchange, in addition, created a BglII recognition site to facilitate screening. The Lys"":'-Gln substitution was introduced using a 21-base pair oligonucleotide "ATTCAGATCCA-ACGTTTGCAC" changing the AAA codon to a CAA codon for c l n . Mutants were identified by screening for an additional XhoII site created by the oligonucleotide used for mutagenesis. To ensure that no secondary priming site existed, the same oligonucleotide was used as a sequencing primer. Only one priming site was found. The exchanges were confirmed by sequencing.
Enzyme Purification-E. coli pop2158 carrying plasmid pMAPlOl or mutagenized derivatives were grown in Luria broth medium supplemented with 0.4% maltose for 24 h. Washed cells were disrupted by sonication or by repeated passage through a French press. The cell-free extracts of mutant enzymes were further purified as described for mutant and wild-type enzyme (12,13). Purification of low-activity mutant proteins was followed by sodium dodecyl sulfatepolyacrylamide gel electrophoresis. Pyridoxal-P content of the protein was determined as described by Wada and Snell (14). Protein concentration was measured chemically (15) or from the absorbance at 280 nm, using EY:; = 1.36.
Kinetic Measurements-Enzyme activity was assayed at 30 "C in the direction of phosphorolysis in a coupled assay (16) in 50 mM Tris acetate or MES buffer, pH 6.9, containing 10 mM maltoheptaose (generously provided by Boehringer Mannheim) or 2% dextrin, 10 mM KH,P04, 1 mM NAD', 5 mM MgCl,, 1 mM EDTA, 5 p M glucose-1,6-bisphosphate, 2 pg/ml phosphoglucomutase, and 2 pg/ml Leuconostoc mesenteroides glucose-6-P dehydrogenase. In the direction of synthesis the enzyme was assayed in 25 mM MES buffer, pH 6.9, 10 mM Glc-1-P, and 5 mM maltotetraose. The release of P, was measured by the method of Saheki et al. concentrations of cations (520 mM) were performed a t 30 "C by varying the concentration of the substrates within a 10-fold range on a Beckmann DU64 spectrophotometer. No differences in activity was observed for several buffers used (MES, HEPES, glycerophosphate buffer). Since the enzymes of the coupled assay are sensitive to high concentrations of cations the assay described above was modified: An assay containing 10 mM maltoheptaose, varying concentrations of phosphate (as Tris phosphate, pH 6.9) and cations was adjusted to 50 mM Tris with Tris acetate, pH 6.9. Aliquots were drawn a t 30 "C at different times and heat inactivated (5 min at 70 "C). The denatured protein was removed by a short centrifugation step (3 min in an Eppendorf centrifuge), and the Glc-1-P concentration was determined by the coupled assay described above. Heat inactivation kinetics in absence and presence of 100 mM cations at 60 "C exhibited no differences between the Lys"'"-Ser mutant and the wild-type enzyme. The modified assay allowed to determine the reaction velocity in absence of monovalent alkali cations and to measure the effects of the different monovalent cations on activity.

RESULTS AND DISCUSSION
Lys""" was changed to a serine or a glutamine by the gapped duplex approach of site-directed mutagenesis (11) on a BglII-PstI fragment of the E. coli malP gene cloned into M13mp9rev. Both mutant and wild-type enzymes exhibited a single band on sodium dodecyl sulfate-polyacrylamide gel electrophoresis with an apparent molecular weight of 90,000. Binding to glycogen-Sepharose showed that substrate affinity was retained in both mutants.
The cofactor content was determined to be 1 mol pyridoxal-P/mol of subunit for the mutant enzyme, therefore, stoichiometry of pyridoxal-P is retained. The possibility that the remaining activity is due to a contamination with traces of wild-type enzyme is ruled out for the Lys""'-Ser mutant by the difference in K,,, values and the dependence of mutant activity on small cations. Furthermore, both mutants differ from the wild-type enzyme in the capability to use the glycosylic substrate analog glucal. Maltodextrin phosphorylase catalyze the formation of 2-deoxyglucosyl- (cu-1,4-glucoside), from D-glUCal and maltotetraose in Activity in the direction of degradation was measured in a coupled assay (see "Material and Methods").
* Synthesis was measured as release of phosphate from Glc-1-P.
Activity determined in absence of monovalent alkali cations. the presence of phosphate, for the reverse reaction three products were identified maltotetraose, 2'-deoxyglucose 1-P, and D-glucal. In the equivalent assay, both mutants of Lys5"{ yield 2-deoxyglucose 1-P and maltotetraose but fail to produce D-glucal from 2'-deoxyglucosyl-(a-l,4-glucoside),. Since no formation of D-glucal was observed, a contamination of more than one wild-type enzyme molecule/10,000 mutant molecules can be ruled out, this property was used to test for the absence of wild-type enzyme.' Substitution of LYS"~ by a serine reduced kc,, 600-fold in the direction of degradation and decreased the apparent affinity for Pi by a factor of about 5 in the absence of monovalent cations, whereas the K, value for maltoheptaose remained essentially unchanged (Table I). In the direction of synthesis the kc,, value was reduced by approximately the same amount, and the apparent binding of Glc-1-P was less affected.
T o probe the catalytic role of the e-amino group of Lys":','', different ligands were tested for their ability to reestablish catalytic activity of t h e L y~~~' -S e r mutant. Amines such as methyl-, ethyl-, and propylamine, ethanolamine, hydroxylamine, and hydrazine neither stimulated nor inhibited the phosphorylase reaction. Only with ammonium ions was a signficant 8-fold increase of kc,, observed. Enzymatic activity, however, was also stimulated by monovalent alkali cations. Under optimal conditions a 10-fold increase of the reaction rate was found in both directions on the addition of Li' and still a %fold increase on addition of Na' or K+. The dependence of activity on the concentration of cations followed hyperbolic saturation kinetics (Fig. 2). Although a precise determination of the Kapp values was difficult due to the high concentrations of cations required, the Kapp for Li' was estimated to be about 80 mM, the Kapp values for NH:, Na+, and K' were estimated to be 90, 120, and 180 mM, respectively. These values correlate with the ion radii of the cations, with the exception of NH:. Since the ion radii of K+ is similar to that of NH:, ammonia ions should show a K,,, similar to that of K+, which is not the case. The differential behavior o f NH: is probably due to the different electron shell of the ammonium molecule, which may allow hydrogen bonding to the protein.  In contrast to the effects on kc,, addition of 100 mM K+ caused a 4-fold decrease in the apparent affinity of Pi compared with that of the mutant in the absence of alkali cations and 20-fold compared with the wild-type enzyme. The increase of the K,,, values for Pi were even more pronounced with 100 mM Na+, NH:, or Li+ ( Table 11). The K,,, for maltoheptaose was raised about 10-fold. In contrast to the k,,, values the stimulatory action and impairment of affinity to Pi seemed to correlate inversely with the ion radii of alkali cations. Again ammonia ions deviated from this observation.
Divalent cations like Ca2+ and Mg'+ had no detectable effects on the activity of the Ly~~~'-Ser mutant at concentrations still soluble in the presence of Pi.
In contrast, substitution of by the isosteric uncharged glutamine left the apparent affinities for all substrates essentially unchanged, but the catalytic activity was reduced 430fold (Table I). No increase of activity was observed on addition of exogeneous added cations. The space taken by the Gln does not allow binding of cations at the site formerly occupied by the t-NH2 residue, therefore preventing substitution of the positive charge next to the phosphate anion. Since the K , value for phosphate is not altered, the phosphate binding site is not affected by the isosteric Lys-Gln replacement.
The failure of primary amines to stimulate activity in the Ly~~~'-Ser mutant makes a function of Lys5" as a proton donor unlikely. This is in contrast to the chemical rescue of an inactive mutant of aspartate aminotransferase by exogenously added amines, which act as proton transfer catalysts (18). Furthermore, the absence of changes in the pH dependence of the Lys5"-Ser mutant compared with wild-type supports the conclusion stated above (19).
The distinct properties of the L y~~~~-S e r and the Lys53'-Gln mutants allow to dissect the role of Lys5" into a t least two components, a conformational one, important for binding of the substrate phosphate, and a catalytic one, based on the positive charge of the e-amino group. The pronounced decrease of kcat resulting from the removal of the positive charge of L Y S~~' points to participation of this residue in two key features of phosphorylase catalysis, the requirement of at least a balanced cofactor phosphate monoanion-dianion mixture in the course of the reaction and a close contact between the cofactor phosphate and the substrate phosphate. One function of the positive charge could be to contribute to the balance of charges in the active site. Loss of one positive charge would force the cofactor phosphate to a monoanion state to preserve electrostatic neutrality. Therefore, the kcat for the mutant should be considerably reduced. Monovalent cations, which fit into the cavity created by the L y~~~~-S e r exchange, allowed for partial restoration of the electrostatic environment, but at the expense of reduced substrate affinity.
An indirect participation of t h e L Y S~~' residue in promoting the close phosphate-phosphate contact can be considered. The positively charged Lys5" might be important to align the phosphate of pyridoxal-P by electrostatic interactions. Removal of the salt bridge would allow more flexibility for the phosphate of pyridoxal-P, which in turn renders the orientation of the phosphate-phosphate contact at random and, thereby, affects catalytic efficiency. Addition of cations counteracts the flexibility caused by the mutation, thereby partially restoring activity. Extending earlier observations (12), a delicately balanced rather tight system of residues around the cofactor binding site became apparent, which constitutes one of the prerequisites for the unique function of pyridoxal-P in phosphorylase catalysis.