Characterization of a Monomeric Escherichia coli Alkaline Phosphatase Formed upon a Single Amino Acid Substitution*

Alkaline phosphatase (AP) from Escherichia coli as well as APs from many other organisms exist in a dimeric quaternary structure. Each monomer contains an active site located 32 Å away from the active site in the second subunit. Indirect evidence has previously suggested that the monomeric form of AP is inactive. Molecular modeling studies indicated that destabilization of the dimeric interface should occur if Thr-59, located near the 2-fold axis of symmetry, were replaced by a sterically large and charged residue such as arginine. The T59R enzyme was constructed and characterized by sucrose-density gradient sedimentation, size-exclusion chromatography, and circular dichroism (CD) and compared with the previously constructed T59A enzyme. The T59A enzyme was found to exist as a dimer, whereas the T59R enzyme was found to exist as a monomer. The T59A, T59R, and wild-type APs exhibited almost identical secondary structures as judged by CD. The T59R monomeric AP has a melting temperature (Tm) of 43 °C, whereas the wild-type AP dimer has a Tm of 97 °C. The catalytic activity of the T59R enzyme was reduced by 104-fold, whereas the T59A enzyme exhibited an activity similar to that of the wild-type enzyme. The T59A and wild-type enzymes contained similar levels of zinc and magnesium, whereas the T59R enzyme has almost undetectable amounts of tightly bound metals. These results suggest that a significant conformational change occurs upon dimerization, which enhances thermal stability, metal binding, and catalysis.

In some oligomeric proteins, the isolated monomeric subunits must be assembled before the protein becomes functional. Such behavior is exemplified by studies on homodimeric chorismate mutase (1) and the trimeric catalytic subunit of aspartate transcarbamoylase (2). In both of these enzymes, the active site contains residues donated from neighboring chains, and each active site is located at the interface between subunits. The reasons for obligatory subunit association are less obvious for enzymes that do not have a shared active site or exhibit allosteric control and regulation. For example, triosephosphate isomerase (3,4) and dihydroorotate dehydrogenase A (5) are both dimeric enzymes that have active sites in which all the catalytically important residues are derived from a single polypeptide chain. Nevertheless, monomeric mutants of triose-phosphate isomerase and dihydroorotate dehydrogenase A were found to lack structural stability and to exhibit the absence or reduction of activity. A monomeric form of a dimeric enzyme that contains an unshared active site in each subunit, Escherichia coli alkaline phosphatase (phosphomonoester hydrolase (EC 3.1.3.1)), is characterized in this study.
Alkaline phosphatase is a homodimeric metalloenzyme, which hydrolyzes phosphomonoesters into inorganic phosphate and the corresponding alcohol. The dimer contains two active sites, located 32 Å apart, each of which accommodates a magnesium and two zinc ions (6). It has been determined that all three metals play a role in the catalytic mechanism of the enzyme (7). Dimerization of the AP 1 monomer occurs spontaneously upon the addition of metals, whereas AP will not dimerize in the absence of metals (8,9). The two monomers of AP are held together by an extensive network of hydrophobic and hydrogen-bonding interactions along the interface. The interface of each subunit is composed of 86 amino acids, encompasses 20% of the protein, 2900 Å 2 of surface area, and is comprised of a secondary structure of four ␣-helices and three ␤-strands (7). Analysis of the interface using the most recent high resolution (1.75 Å) structure of AP as the basis (7) reveals that there are 26 hydrogen-bonding interactions occurring among the residues of the interface. It was suggested in 1965 that these interface interactions of the AP dimer alter the tertiary structure of each subunit upon dimer formation in a manner that allows the metals to bind, creating the functional active site (10).
Monomers of the wild-type enzyme have been obtained through a variety of techniques such as thiol reduction in the presence of urea (11), acid treatment (12), metal chelation, followed by incubation in 5% formamide (9), or by isolation of matrix bound monomers (13). However, the characterization of these monomeric forms of E. coli AP have been conducted under very specialized conditions. Therefore, to understand better the functional role of the dimerization process in E. coli AP, a monomeric version of the enzyme was rationally designed by altering a single amino acid side chain in the monomermonomer interface. A stable monomeric version of the protein allowed a more direct comparison between the properties of the monomer and those of the wild-type dimeric form under benign buffer conditions.

EXPERIMENTAL PROCEDURES
Materials-Agar, agarose, ampicillin, p-nitrophenyl phosphate, magnesium chloride, and zinc sulfate were purchased from Sigma. Tris, sucrose, and enzyme-grade ammonium sulfate were supplied by ICN Biomedicals (Costa Mesa, CA). Tryptone and yeast extract were obtained from Difco Laboratories (Detroit, MI). DNA sequencing was carried out by Beth Israel Deaconess Medical Center, Molecular Medicine Unit. The oligonucleotides required for site-specific mutagenesis and sequencing were obtained from Operon Technologies (Alameda, * This work was supported by Grant GM42833 from the National Institutes of Health. 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. ‡ To whom correspondence should be addressed. E-mail: evan. kantrowitz@bc.edu. CA). Plasmids were isolated and purified using the QIAprep Spin Miniprep kit purchased from Qiagen (Valencia, CA).
Construction of a Plasmid for the Expression of the T59R Alkaline Phosphatase-The T59R mutation was induced in the phoA gene contained in the plasmid pEK154 using the procedures outlined by Stratagene in the QuikChange mutagenesis kit protocol. The entire gene was sequenced to ensure that the correct mutation was present and to confirm that no other mutations had been introduced. The final plasmid containing the T59R mutation in the phoA gene was designated pEK618.
Modeling Studies-All molecular modeling studies were carried out using the program QUANTA, and energy minimization was accomplished using X-PLOR (16). The 1.75-Å wild-type AP Protein Data Bank structure 1ED8 (7) was used as the base structure.
Protein Expression-SM547, an E. coli strain with the wild-type phoA gene deleted from the chromosome and a mutation in the phoR regulatory gene, was used as the host strain for expression of the wild-type and mutant AP. The deletion of the phoA gene ensures that no alkaline phosphatase can be produced in the absence of a phoA-containing plasmid present in the host strain.
Protein Purification-The wild-type, T59R, and T59A enzymes were isolated from the periplasmic space by osmotic shock and ammonium sulfate precipitation by the methods previously described (14). Final purification was carried out on a Bio-Rad BioLogic HR system using a Q-Sepharose Fast Flow column (1.6 ϫ 10 cm). A 200-ml gradient from 0 to 0.1 M NaCl in TMZP buffer (0.01 M Tris-HCl, 0.001 M MgCl 2 , 10 Ϫ5 M ZnSO 4 , 10 Ϫ4 M NaH 2 PO 4 , 0.31 ϫ 10 Ϫ2 M NaN 3 , pH 7.4) was used to elute the protein. The purity of the enzyme was determined by SDS-PAGE (17).
Determination of the Protein Concentration-The concentration of the wild-type enzyme was determined by absorbance measurements at 278 nm with an extinction coefficient of 0.71 cm 2 /mg (18). The concentration of the mutant enzymes was determined by the Bio-Rad version of Bradford's dye binding assay (19) using wild-type alkaline phosphatase as the standard.
Determination of the Enzymatic Activity-Alkaline phosphatase activity was measured spectrophotometrically utilizing p-nitrophenyl phosphate as the substrate at 25°C by monitoring the release of pnitrophenolate at 410 nm (20) in 1 M Tris buffer, pH 8.0.
Sucrose-density Gradient Centrifugation-Sedimentation coefficients of the mutant enzymes were determined by sucrose-density gradient centrifugation. A 4.6-ml gradient of 6 -25% sucrose in 50 mM NaH 2 PO 4 , pH 8.0, was used. After preparation, 200 l of ϳ5 mg/ml protein solution was carefully layered on top of the gradient. The tubes were spun at 170,000 ϫ g for 18 h using a Beckman SW 55Ti rotor in a Beckman L-70 centrifuge.
The gradients were fractionated using a Brandel BR-9620 fractionator. A 50% sucrose, 2 mg/ml bovine serum albumin solution was slowly pumped from the fractionator to a UV detector (Gilson Model 112) in order to monitor changes in absorbance. The increase in absorbance, due to the bovine serum albumin, was used to determine the end of the gradient collection. Calibration was accomplished using carbonic anhydrase, bovine serum albumin, and wild-type AP as standards.
Size-exclusion Chromatography-Gel filtration chromatography was carried out on a Bio-Rad BioLogic HR system using a Bio-Rad Bio-Prep SE 100/17 column. The column was eluted with 40 mM NaH 2 PO 4 buffer, pH 8.0, at a flow rate of 0.5 ml/min and calibrated using as standards bovine serum albumin, ovalbumin, chymotrypsin A, cytochrome c, and rabbit muscle aldolase, which have known Stokes radii of 3.55, 3.05, 2.09, 1.70, and 4.81 nm, respectively.
Thermal Stability-An Aviv circular dichroism spectrometer model 202 was used to obtain CD spectra from 195 to 300 nm and determine T m curves for the wild-type and mutant alkaline phosphatases. All experiments were carried out using 2 ml of ϳ0.04 mg/ml enzyme in 10 mM KH 2 PO 4 buffer, pH 8.0. For the T m experiments, the CD signal at 222 nm was monitored as the temperature increased in 1°C intervals from 25 to 100°C. The sample was allowed to equilibrate at each temperature for 1 min before the CD signal was measured.
Determination of Metal Content-A PerkinElmer 3100 atomic absorption spectrometer was used to determine the metal content of the enzymes. Before analysis, the enzyme samples were dialyzed against metal-free 10 mM Tris buffer, pH 8.0, at 4°C for 6 h. Enzyme solutions of ϳ5 mg/ml were aspirated into the flame at a rate of 1 ml/min and compared with magnesium chloride and zinc sulfate standards.

RESULTS
Design of a Monomeric Alkaline Phosphatase-The most appropriate mutation for destabilization of dimeric AP was determined through molecular modeling studies. When a Thr to Arg mutation was introduced at position 59 in the AP structure (Protein Data Bank number 1ED8) using the program QUANTA and subsequently minimized using X-PLOR, large movements of the protein backbone around the location of the amino acid substitution were required to accommodate the mutation, suggesting that the physical mutation may interfere with dimer formation.
Quaternary Structure of the T59A and T59R Alkaline Phosphatases-The quaternary structure of the T59A and T59R enzymes was determined by both sucrose-density gradient sedimentation and size-exclusion chromatography. The sedimentation patterns of the wild-type and T59A enzymes are virtually identical (Fig. 1) indicating that they have the same dimeric quaternary structure. The sedimentation pattern of the T59R enzyme exhibited no species migrating as a dimer; rather a slower migrating species was observed. The sedimentation coefficients of the T59A and T59R enzymes were 6.1 and 3.6 S, respectively. Based upon these sedimentation coefficients, it can be inferred that the T59A enzyme migrates as a dimer and the T59R enzyme migrates as a monomer.
Size-exclusion chromatography was used to determine the Stokes radius of the T59R, T59A, and wild-type enzymes. The elution profiles for the T59A and wild-type enzymes were very similar with nearly identical retention times. However, the T59R enzyme eluted with a longer retention time. These results further support the proposal of a dimeric structure for the T59A enzyme and a monomeric structure for the T59R enzyme. Based upon these experiments, the Stokes radii of the wildtype, the T59A, and the T59R were determined to be 3.41, 3.35, and 2.93 nm, respectively.
An estimate of molecular mass can be determined from the sedimentation coefficient and Stokes radius (21). The calculated molecular mass for both the wild-type and T59A enzymes was 87,000 daltons, whereas the T59R had a molecular mass of 44,000 daltons, again indicating that the T59R enzyme is monomeric.
Kinetic Characterization of the T59R Alkaline Phosphatase-The T59R AP had substantially reduced activity compared with the wild-type enzyme. Although the activity of the T59R AP was 10 4 -fold lower than that of the wild-type enzyme, it was still considerably higher than that observed for the noncatalyzed reaction. To rule out the possibility that the observed activity was due to the formation of trace amounts of the more active dimeric form, the effects of protein concentration on the specific activity were measured. Over more than a 200-fold range of protein concentration (0.0019 to 0.39 mg/ml), the specific activity was found to be virtually unaltered. The lack of a concentration dependence of the specific activity is indicative of the absence of any monomer-dimer equilibrium occurring for the T59R enzyme. As seen in Table I, the kinetic parameters for the wild-type and T59A enzymes are similar. However, the k cat of the T59R enzyme is reduced by more than 35,000-fold as compared with the wild-type enzyme.
Influence of Metals on the Activity of T59R Alkaline Phosphatase-Two different sets of measurements were performed to determine whether increased amounts of Zn 2ϩ and/or Mg 2ϩ altered the kinetic properties of the T59R AP. First, the activity of the T59R AP was determined in the presence of Zn 2ϩ and/or Mg 2ϩ at up to 100-fold higher concentrations than normally used for the wild-type enzyme. In the second set of experiments, the T59R AP was incubated with 10 mM Zn 2ϩ (1000-fold higher concentration than normally used) at 25°C and 37°C for 4 h before the activity measurements were performed. The T59R AP in TMZP and in metal-free 0.01 M Tris, pH 8.0, buffer was incubated with Zn 2ϩ followed by activity measurements in 1 M Tris, pH 8.0, with 1000 M p-nitrophenyl phosphate as the substrate. The activity of the T59R AP carried out in the presence of Zn 2ϩ and Mg 2ϩ and after incubation with 10 mM Zn 2ϩ showed very little change in activity when compared with the activity of the untreated T59R AP. Fig. 2, the CD spectra of the T59R, T59A, and wild-type enzymes are virtually identical. Deconvolution of these CD spectra (22) indicates that the three enzymes have almost identical secondary structures. Each enzyme contained almost identical percentages of ␣-helix, ␤-sheet, ␤-turn, and random coil.

Analysis of the Mutant and Wild-type APs by Circular Dichroism-As seen in
CD was also used to determine the thermal stability of the wild-type and mutant enzymes by monitoring the temperaturedependent change in ellipticity at 222 nm. The T m curves for the T59A, T59R, and wild-type enzymes are shown in Fig. 3. The T59A enzyme exhibits a T m of ϳ92°C, about 5°C lower than that of the wild-type enzyme. However, the T59R mutation induced a dramatic effect on thermal stability. The T m for the T59R enzyme was 43°C, significantly lower than the 97°C observed for the wild-type enzyme. These data clearly indicate that the T59R mutation has a dramatic effect on the thermal stability of the enzyme, whereas the T59A mutation has almost no influence on the thermal stability.
Metal Content of the Wild-type and Mutant Enzymes-Atomic absorption was used to determine the metal content of the wild-type and mutant enzymes. This experiment provides data only on tightly bound metals, because the enzyme is dialyzed in metal-free buffer in preparation for the experimental determination. The wild-type and T59A enzymes were found to have ϳ3.4 mol of zinc and 2.0 and 1.5 mol of magnesium per mol of enzyme, respectively. The atomic absorption analysis indicated that there were only trace levels of both magnesium and zinc in the T59R enzyme. These results indicate that either the T59R enzyme does not bind zinc or magnesium or that it binds these metals so weakly that the dialysis step used in the preparation of the enzyme removes them. DISCUSSION The two active sites in the dimeric E. coli alkaline phosphatase are ϳ32 Å apart. Because no active site residues are contributed from the adjacent subunit, no obvious requirement exists for obligatory dimerization. Yet mutants of the E. coli enzyme have been isolated that exhibit intergenic complementation. A heterodimeric form of the enzyme, containing different mutations in the two subunits, exhibits significantly higher activity than would be expected based upon the activity of the homodimeric species used to create the hybrid enzymes (12,(23)(24)(25)(26). This observed intergenic complementation suggests that the formation of the dimer interface may result in structural alterations that influence activity.
Although most APs have quaternary structures that are dimeric or higher order aggregrates of dimers, there are a few APs reported to be monomeric, such as ones from Pyrococcus abyssi (27), Bombyx mori (28), Vibrio sp. (29), Bacillus inter- a The k cat values are calculated from the V max using a dimer molecular weight of 94,000 (35) for wild type and T59A and a monomer molecular weight of 47,000 for T59R.
b The values for the T59A enzyme were reported previously (15).
FIG. 2. CD spectra for wild-type, T59A, and T59R enzymes. All spectra were obtained using 2 ml of ϳ0.04 mg/ml-enzyme in 10 mM KH 2 PO 4 buffer, pH 8.0, at 25°C. The three spectra have been staggered for clarity. WTAP, wild-type AP.
medius (30), and one of the two APs from Bacillus subtilis (31). In some cases at least, the evidence for the monomeric nature of these APs is weak. In fact, it has recently been established that the P. abyssi AP does not exist as an active monomer although stable monomers can be isolated under certain conditions. 2 Martin et al. (15) have attempted to disturb the dimer interface of AP using site-directed mutations that eliminated hydrogen-bonding interactions across subunits. Although some of the AP mutants were found in a detectable monomer-dimer equilibrium, none of the mutations weakened the interface sufficiently to allow the isolation of a monomer form of the enzyme. In this work, an amino acid with a bulky charged side chain was introduced into a confined area of the interface so that steric hindrance would completely destabilize the dimeric form of the enzyme. The study of a monomeric form of AP will allow us to understand better the role of the dimer interface for the function of E. coli AP.
Creation of a Monomeric Alkaline Phosphatase-Modeling studies of the interface region of AP suggested a number of locations that were potential candidates for the introduction of a large, bulky, and charged side chain that would disturb the formation of the dimeric structure. One such location was Thr-59, which is near the 2-fold axis of symmetry. The replacement of Thr-59 with Arg would, therefore, place two Arg residues in the interface exactly juxtaposed to each other adding charge repulsion in addition to steric interference to destabilize the interface.
Analysis of the T59R AP indicated that the mutation indeed disrupted the quaternary structure of the enzyme. The T59R enzyme was found by sucrose-density gradient sedimentation to sediment as expected for a monomer. The molecular mass of the T59R enzyme was determined to be approximately half that of the wild-type enzyme by a combination of sucrosedensity gradient sedimentation and size-exclusion chromatography. Because no intermediate migrating species were observed between the positions expected for the monomer and dimer in the sucrose gradient profile of the T59R enzyme, subunit association must be completely prevented in the T59R enzyme.
Unlike the T59R enzyme, the T59A AP exists as a dimer, indicating that the absence of hydrogen bonding when methyl replaced a hydroxy group was not enough to impair dimerization. In contrast, the bulky charged side chain of Arg at position 59 is sufficient to destabilize the interface and prevent dimer formation. To determine whether the T59R mutation caused monomer formation by significantly altering the secondary structure of the enzyme, circular dichroism was used to compare the T59A, T59R, and wild-type enzymes. The monomer form of the T59R enzyme retains the same overall secondary structural fold as the wild-type enzyme. The CD spectra of these three APs were almost identical, indicating that the mutations at position 59 do not significantly alter the secondary structure of the enzyme. Therefore, the loss of the ability of the T59R enzyme to dimerize is not the result of alterations in the secondary structure of the enzyme due to the mutation.
Altered Properties of the Monomeric Alkaline Phosphatase-The monomer form of alkaline phosphatase generated in this work has significantly different catalytic activity and structural stability compared with the wild-type enzyme. The thermal melting profiles for the monomer, as determined by CD, indicate an enormous change in thermal stability. Considering that E. coli is not a thermophilic organism, E. coli AP is an extremely thermally stable protein for its size with a T m of 97°C. The more than 50°C reduction in T m of the monomeric T59R AP (T m ϭ 43°C) indicates that the presence of the dimer interface adds tremendous structural stabilization to the native enzyme. Thus, one of the roles of the interface in E. coli AP is to provide thermal stability for the enzyme. It is interesting to note that one of the alkaline phosphatases reported to exist as a monomer was isolated from Vibrio sp. (29), a psychrophile. The formation of higher order quaternary structures may well provide a general mechanism for thermal stability for AP in particular and multimeric enzymes in general.
The monomeric T59R AP has severely impaired catalytic ability. In fact, the k cat of the T59R AP is 4-fold lower than that shown by mutant versions of AP in which the critical nucleophilic Ser-102 residue has been replaced by Ala or Gly (32). Even at Zn 2ϩ concentrations 1000-fold higher than required for the wild-type enzyme, no significant increase in activity was detected. These results indicate that the low activity of the T59R AP is not simply the result of weakened binding of the metals to the active site of the enzyme. Because the T59A enzyme has almost the same k cat as the wild-type enzyme, the reduction in activity induced by the T59R mutation must be a direct result of the different quaternary structures of the T59A and T59R enzymes. The CD data, which showed that the secondary structures of the T59A, T59R, and wild-type APs were indistinguishable, support this conclusion.
Two points have been made evident by these experiments. First, an intact subunit interface is absolutely required for normal catalytic function of E. coli AP. Second, stability of the subunit is greatly enhanced upon association and formation of the dimer interface. As found in earlier studies, maintenance of the structural integrity of the interface has an apparent role in dimer stabilization. In fact, the first 40 residues of the amino terminus of AP have more intersubunit contacts than intrasubunit contacts. This structural feature seems to indicate that the amino-terminal segment of the protein is important for maintaining the dimeric structure. For example, truncation of either the first 10 or 35 residues from the amino terminus greatly destabilizes the dimeric form of the E. coli enzyme and reduces the activity by 20% (8,33).
Both a monomeric and dimeric form of AP exists in Bacillus subtilis (31,34). The amino acid identities found between the B. subtilis AP III (dimeric) and AP IV (monomeric) and the 2 S. Zappa, personal communication. All experiments were carried out using 2 ml of ϳ0.04 mg/ml enzyme in 10 mM KH 2 PO 4 buffer, pH 8.0. The ellipticity was monitored at 222 nm as the temperature was increased from 25 to 100°C. The three plots were staggered for clarity. WTAP, wild-type AP.
E. coli enzyme are 33% and 34% respectively. The core and active site residues of the B. subtilis enzymes are well conserved when compared with E. coli AP. However, in contrast, the interface residues of the B. subtilis APs are far less conserved when compared with E. coli AP. Another notable difference in the sequence alignment between the two B. subtilis and E. coli APs is residue 59 (E. coli numbering) (34). In the E. coli enzyme, residue 59 is a Thr; and in the dimeric B. subtilis AP, the corresponding residue is a Ser; whereas in the monomeric B. subtilis AP, the residue is Arg. The monomeric T59R E. coli enzyme, therefore, can be compared with that of the monomeric B. subtilis enzyme where the sequence-aligned residue at position 59 is Arg. These comparisons taken together suggest that dimerization of the monomeric form of B. subtilis AP and of E. coli T59R AP may not occur because the Arg side chain sterically hinders intersubunit interactions.
This work has elucidated aspects of the functional and structural role of the AP interface, such as the requirement for interfacial interactions, which are critical for the formation of a catalytically functional active site pocket. The monomeric T59R AP bound insignificant amounts of magnesium or zinc, which are required in the proposed catalytic mechanism of wild-type AP (7). Because interfacial interactions are absent in the T59R monomer, the active site pocket remains impaired, and this effect is reflected in the exceedingly low enzymatic activity. However, the loss of the interface interactions and incomplete formation of an active site pocket do not affect the secondary structure, yet they influence the overall folding stability. These findings suggest that the interfacial interactions of AP not only strengthen the protein fold of AP but also align the residues in the active sites that are required for optimum metal-binding and phosphatase activity.