Function Assignment to Conserved Residues in Mammalian Alkaline Phosphatases*

We have probed the structural/functional relationship of key residues in human placental alkaline phosphatase (PLAP) and compared their properties with those of the corresponding residues in Escherichia coli alkaline phosphatase (ECAP). Mutations were introduced in wild-type PLAP, i.e. [E429]PLAP, and in some instances also in [G429]PLAP, which displays properties charac-teristic of the human germ cell alkaline phosphatase isozyme. All active site metal ligands, as well as residues in their vicinity, were substituted to alanines or to the homologous residues present in ECAP. We found that mutations at Zn2 or Mg sites had similar effects in PLAP and ECAP but that the environment of the Zn1 ion in PLAP is less affected by substitutions than that in ECAP. Substitutions of the Mg and Zn1 neighboring residues His-317 and His-153 increased k cat and increased K m when compared with wild-type PLAP, contrary to what was predicted by the reciprocal substitutions in ECAP. All mammalian alkaline phosphatases (APs) have five cysteine residues (Cys-101, Cys-121, Cys-183, Cys-467, and Cys-474) per subunit, not homologous to any of the four cysteines fit

Alkaline phosphatases (APs 1 ; EC 3.1.3.1) are a family of dimeric metalloenzymes catalyzing the hydrolysis of monoesters of phosphoric acid (1). APs also catalyze a transphosphorylation reaction in the presence of large concentrations of phosphate acceptors. APs occur widely in nature and are found in many organisms from Escherichia coli to man. APs are homodimeric enzymes, and each catalytic site contains three metal ions (two Zn ions and one Mg ion) that are necessary for enzymatic activity (2). Whereas the main features of the catalytic mechanism are conserved between mammalian APs and the E. coli enzyme, there are important differences. Mammalian APs (a) have higher specific activity and K m values, (b) have a more alkaline pH optimum, (c) are inhibited by L-amino acids and peptides through an uncompetitive mechanism, and (d) display lower heat stability. These properties, however, differ noticeably within the group of mammalian AP isozymes. Many of the isozyme-specific properties have been attributed to a flexible loop region or "crown" domain of the molecule (3,4). Some mammalian APs, such as the human intestinal isozyme, are activated by magnesium ions, whereas the human placental AP is more similar to the E. coli enzyme in that its activity is not enhanced by the addition of magnesium (1).
For many years, E. coli AP (ECAP) was the only source of structural information on APs (2), but now the three-dimensional structure of the first mammalian AP, i.e. human placental alkaline phosphatase (PLAP), has been solved (4). As had been predicted from sequence comparisons (5), the central core of PLAP, consisting of an extended ␤-sheet and flanking ␣-helices, is very similar to that of ECAP. The same is true in the immediate vicinity of the three catalytic ions. However, a number of distinctive features, including a different positioning of the amino-terminal segment of the molecule and the expanded top loop or "crown" domain, are now apparent (4). An additional noncatalytic metal-binding site not present in ECAP was uncovered, which appears to be occupied by calcium (4,6). ECAP has four cysteine residues that are all involved in disulfide bond formation (7), whereas PLAP has five nonhomologous residues. Furthermore, whereas ECAP is located in the periplasmic space of the bacterium, PLAP is an ectoenzyme bound to the plasma membrane via a glycosyl-phosphatidylinositol anchor (8,9).
In the present paper, we embarked on a wide-range mutagenesis study on structure-function relationships in mammalian APs, using the PLAP structure as a paradigm. The aim was to pinpoint the features of PLAP responsible for the spe-cific properties of this AP isozyme, as well as the properties of mammalian APs in general. The active site metal ligand residues were mutated to alanines or to the analogous residues in the structure of ECAP, and cysteines were mutated to serines. Furthermore, Tyr-367 is conserved in all mammalian APs and occupies a unique position in the PLAP structure, serving as a bridge from one subunit in the dimer to the active site of another subunit. We mutagenized the Tyr-367 residue to Ala and Phe. Mutations were studied individually as well as in combination with the E429G substitution because this change has been shown to significantly affect many of the enzymatic properties of PLAP by conferring germ cell alkaline phosphatase characteristics to the resulting mutant enzyme (10 -13).
The present analysis of structurally and catalytically important residues in PLAP has identified critical positions serving a different structural role in mammalian and bacterial alkaline phosphatases. We have also defined the location of the hydrophobic pocket that participates in stabilizing the side chains of uncompetitive inhibitors in the immediate vicinity of the active site of mammalian alkaline phosphatases.
Mutations Involving Residue 367-The following PCR primers were used: Y367A(R), CGA AGA TGG AGC TCC CTC GCA GGG GGG CGC CTC C; Y367F(R), CGA AGA TGG AGC TCC CTC GCA GGG GGA AGC CTC C; in PCR reactions. For PCR reaction 367A/ we used primers H153A and Y367A(R), whereas for reaction 367F/ the primers H153A and Y367F(R) were used. Restriction fragments were then cut and combined with fragments from PLAP-FLAG/pCDNA3 to make the final constructs (

Expression and Purification of FLAG-tagged Enzymes
PLAP-FLAG constructs were transfected into COS-1 cells for transient expression by the DEAE-dextran-or calcium phosphate-mediated method. Three and 6 days after transfection, conditioned media were collected and concentrated about 10 times by ultrafiltration on 50 Centricon columns (Amicon Inc., Beverly, MA). PLAP-FLAG proteins were purified by affinity chromatography with anti-FLAG M2 antibody gel (Sigma) according to the manufacturer's instructions.

Characterization of Recombinant APs
To measure relative specific activities, microtiter plates were coated with 2 g/ml M2 anti-FLAG monoclonal antibody (Sigma). After the addition of recombinant APs, the activity of the bound enzymes was measured as the change in absorbance at 405 nm over time at 37°C upon the addition of 20 mM p-nitrophenylphosphate as substrate in 1.0 M diethanolamine buffer (pH 9.8), 1 mM MgCl 2 , and 20 M ZnCl 2 . PLAP-FLAG served as a reference for each microtiter plate. The pnitrophenol concentration formed was calculated using an extinction coefficient of 10,080 liter ϫ mole Ϫ1 ϫ cm Ϫ1 . To calculate K m , substrate concentration was varied between 0.2 and 1.6 mM p-nitrophenylphosphate, and the initial reaction rate was measured at 37°C over a time interval of 5 min. Results were fit by nonlinear regression to the Michaelis-Menten equation using GraphPad Prism version 3.02 (GraphPad Software, San Diego, CA).
AP pH/activity profiles were done using 20 mM p-nitrophenylphosphate as substrate in 50 mM buffer with 1 mM MgCl 2 and 20 M ZnCl 2 ; bis-tris propane for pH range 6.5-9.5, and 2-amino-2-methyl-1-propanol for pH range 9.5-11. The pH was checked after dissolving the substrate. The 50 mM 3-(cyclohexylaminol)-1-propanesulfonic acid buffer proved to be inhibitory to PLAP but not ECAP activity. The carboxyl-terminal FLAG fusion protein of ECAP (Eastman Kodak Co.) was used for comparison. The heat stability of PLAP mutants was studied by incubating the enzyme samples in 1 M diethanolamine buffer, pH 9.8, containing 1 mM MgCl 2 and 20 M ZnCl 2 . After a 10-min incubation at a given temperature, the samples were placed on ice. Residual activities were then measured in duplicate with 10 mM p-nitrophenylphosphate at pH 9.8 in the same buffer.

Fluorescence Labeling of PLAP
The cysteine-specific probe 4-(aminosulfonyl)-7-fluoro-2,1,3-benzoxadiazole (ABD-F; Molecular Probes Inc., Eugene, OR) was used to label wild-type PLAP (not modified with FLAG peptide). Free ABD-F shows no fluorescence but can form fluorescent adducts with protein thiol groups with an excitation maximum at 390 nm and an emission maximum at about 500 -520 nm. In a modification reaction, 20-l aliquots of 50 mM ABD-F solution in Me 2 SO were added to 1-ml samples of PLAP solution (0.4 M/liter) in 0.1 M Tris-HCl buffer, pH 7.6, either at nondenaturing conditions or in the presence of denaturant (4 M guanidine hydrochloride), reducing agent (1 mM tris-(2-carboxylethyl)phosphine; Molecular Probes Inc), or both. The samples were then incubated for 3 h at 37°C while protected from light. Fluorescence spectra were measured with a FluoroMax-2 fluorometer with excitation at 390 nm.

Computer Models and Docking Strategy
The binding of inhibitors L-Leu and L-Phe to the PLAP molecule was modeled using the flexible ligand docking program FlexX (14). The active site of PLAP was defined as a sphere of 9 Å around the Zn1 atom. The phosphate ion present in the enzyme structure was included in the receptor model. Ligands were prepared in the SYBYL mol2 format, in energy-minimized form, with all hydrogens added and formal charges assigned. The amino groups of L-Leu and L-Phe were left unprotonated.
The manual "base fragment" selection (14) option was used, and either the carboxylic, amino, or hydrophobic groups of the ligand were chosen as the base fragment.

RESULTS AND DISCUSSION
To pinpoint the residues important for catalysis and stability of human PLAP, we constructed and characterized 23 individual site-directed mutants of PLAP (  (7), and we mutagenized each of these 5 cysteines of PLAP to Ser and also studied the [C467S, C474S] double mutant. In addition, mutagenesis of the Tyr-367 residue, a signature feature of mammalian APs, was also performed.
To simplify the recovery and purification of the recombinant PLAP variants, the glycosylphosphatidylinositol anchoring sequence of PLAP was replaced by the FLAG octapeptide, and all mutants were expressed as secreted, epitope-tagged, enzymes. We had previously determined that the addition of the FLAG sequence does not interfere with the kinetic properties of the molecule (15). This strategy facilitated the production of large amounts of recombinant protein and enabled the fast and efficient isolation of each mutant AP using anti-FLAG affinity purification. Two substitutions resulted in mutant enzymes that did not have any AP activity over background levels in the culture media, i.e. C121S and C183S. We could not detect any FLAG-tagged protein in the media for these mutants by Western blot analysis, indicating that these mutations had severe adverse consequences on protein structure, proper folding, and/or secretion. Fig. 2 shows a detailed comparison of the structure of the active site region of PLAP and ECAP including all the metal ligands that were mutagenized in this study. When substituting the Zn1 ligands in PLAP, i.e. Asp-316, His-320, and His-432, two of the three mutants, i.e.

Mutagenesis of Metal Ligands in PLAP-
[A316]PLAP and [A432]PLAP, retained significant activity ( Table I). The k cat and K m of [A316]PLAP showed a 2.8-and 2.25-fold decrease relative to wt PLAP. Thus, the catalytic efficiency (k cat /K m ) of the [A316]PLAP mutant remains comparable to that of wt PLAP. The k cat of [A432]PLAP was also reduced 2.7-fold, but its K m increased 3.7-fold for a resulting 5.8-fold reduction in catalytic efficiency. In contrast, the introduction of the H320A mutation reduced the specific activity of PLAP by Ͼ200-fold. Saturation of each of the mutants with Zn 2ϩ concentrations of up to 10 mM did not result in any increase in activity. It should be noted that analogous mutations in ECAP were reported to have very different consequences. Notably, the D327A substitution in ECAP (analogous to Asp-316 in PLAP) resulted in a 3000-fold decrease in k cat and a 2000-fold increase in K m for a 10 7 -fold decrease in catalytic efficiency that was not reversible by the addition of Zn 2ϩ (16). In contrast, the activity of the H412A mutant in ECAP (analogous to H432A in PLAP) was responsive to 0.2 mM Zn 2ϩ , reaching k cat and K m values only 2-fold lower that those of wt ECAP (17). The H331A mutation (analogous to H320A in PLAP) has not been studied in ECAP. These results indicate that there are significant differences in the environment of Zn1 in the PLAP structure compared with the ECAP structure and that substitutions of the Zn1 ligands are better tolerated in PLAP than in ECAP. This may reflect the fact that the top flexible loop, or crown domain, that harbors E429 in PLAP appears to provide additional stabilization to the active site environment, so that Zn 2ϩ cannot easily diffuse in or out of the PLAP molecule, as we have previously observed (18). Thus, even though the state of coordination of Zn1 is affected by the mutations, the Zn 2ϩ ion remains in place and is able to function in catalysis.
Alanine substitutions of the Zn2 ligands in PLAP, i.e. Asp-42, Asp-357, and His-358, resulted in significant decreases in specific activity, ranging from Ͼ25-fold (D357A) to undetectable levels (H358A  (20). Interestingly, the S155T substitution hardly affects the activity of the resulting mutant and even doubles its catalytic efficiency (Table I).
Mutagenesis of Nonconserved Active Site Residues in PLAP-Whereas most of the AP active site residues are perfectly conserved throughout evolution, some important differences exist in the neighborhood of the Mg ion (Fig. 2). His-153 and His-317 in PLAP are homologous to Asp-153 and Lys-328, respectively, in ECAP. The substitution of D153H and K328H in ECAP produced enzymes with kinetic properties similar to  those of mammalian APs. For example, the D153H/K328H double ECAP mutant displayed a 5.6-fold higher k cat and a 30-fold higher K m , a decrease in heat stability, and a shift in pH optimum to alkaline pH values (21). We constructed the reciprocal mutations, i.e. H153D and H317K, in PLAP as well as the double mutation (H153D/H317K). We also introduced a H153A and H317A mutation in both PLAP and [G429]PLAP. The expectation was that by reverting to the residues found in ECAP, one would confer ECAP-like properties to PLAP. Surprisingly, however, no decrease in K m was observed in any of the mutants. Instead, the K m values consistently increased for all the mutants. The effects on k cat were variable. There were no significant changes in the pH dependence or heat stability of the mutants compared with the wt PLAP (data not shown).
In the ECAP active site, the environment of the Mg ion is one of octahedral coordination, including three amino acids and three water molecules (2). These water molecules are further coordinated and stabilized by other amino acid residues including Asp-153. The D153H mutation in ECAP was shown to destabilize the octahedral Mg coordination in favor of a tetrahedral one and resulted in an enzyme that had reduced Mg 2ϩ affinity and increased Zn 2ϩ affinity and was significantly activated by Mg 2ϩ (21). This is strongly reminiscent of the behavior of the IAP isozyme (1) but not of human PLAP, which binds Mg 2ϩ tightly and in which the further addition of Mg 2ϩ does not increase activity. A possible explanation for our results comes from the analysis of the structure of PLAP around the Mg ion (see Fig. 2). In PLAP, His-153 and His-317 are positioned so that they can serve the same purpose as the corresponding Asp and Lys in ECAP, i.e. they are direct ligands to active site water molecules and indirect ligands to the Mg ion and the noncovalently bound phosphate group. We propose that His-153 and His-317 in PLAP are already well positioned to stabilize these water-mediated interactions and that introducing different residues at these positions would result in a decrease in affinity for phosphate, thus increasing K m and k cat rather than decreasing these parameters. Because the IAP isozyme is more dependent on Mg 2ϩ activation, one can speculate that the structure of IAP in the immediate environment of the Mg ion is more similar to and can be better modeled by the structure of ECAP, rather than the structure of PLAP. Upon mutagenizing H153A or H317A in PLAP, both mutant enzymes displayed about a 2-fold increase in k cat and about a 3-fold increase in K m compared with wt PLAP. This can be explained by the disruption of their water-mediated interactions with the phosphate group (see Fig. 2) via the same water molecule. Disruption may lead to an enzyme with smaller affinity for both substrate and product, thus having higher k cat and K m values. Interestingly, by combining the H153A and H317A mutations with the G429 mutation, we were able to engineer mutant enzymes with increased catalytic efficiency when compared with wt PLAP but not when compared with [G429]PLAP. The [A153, G429]PLAP and [A317, E429G]PLAP mutants had K m values that were restored to those of wt PLAP while largely preserving the increase in k cat . This is especially true of [A317, G429]PLAP, in which both specific activity and catalytic efficiency (k cat and k cat /K m values) were increased 2-fold compared with wt PLAP and 5-fold when compared with [G429]PLAP.
Other Conserved Residues in the Active Site Area-Two other active site residues, His-360 and His-319, that are perfectly conserved in mammalian APs were also investigated. In ECAP, His-372 is 3.8 Å away from Zn1 in the active site, and correspondingly, His-360 is located within 4.4 Å from the Zn1 ion in PLAP. The side chain of His-360 forms a hydrogen bond to the side chain of Asp-316, which is a direct ligand to Zn1. The [A372]ECAP showed changes in catalytic behavior, such as a 20% reduction in k cat and a 4-fold reduction in K m in the presence of a phosphate acceptor (22). In contrast, [A360]PLAP displayed an increase in specific activity 20% over that of wt PLAP and a 3.8-fold increase in K m .
Judging from its three-dimensional positioning, residue His-319 should mainly have a structural role, being involved in hydrogen bonds with Thr-48 and Tyr-393. The H319A and H319A/E429G mutations, however, displayed a 70-fold and 37-fold decrease in specific activity, respectively, indicating that the interactions formed by His-319 are important for maintaining an optimal conformation in the active site.
Hierarchical Significance of the Five PLAP Cysteines-PLAP, as well as all mammalian APs, has 5 cysteine residues/ subunit in positions 101, 121, 183, 467, and 474 (Fig. 3A). They form two disulfide bonds, Cys-121-Cys-183 and Cys-467-Cys-474, whereas the Cys-101 residue remains in free form. We found that the cysteine recombinant PLAP mutants C101S, C467S, C474S, and C467S/C474S were appropriately expressed by transfected COS-1 cells and retained residual activity, whereas the C121S and C183S mutants were degraded inside the cell. The PLAP structure reveals that the sequence contained within the Cys-121-Cys-183 disulfide bond (Fig. 3B) harbors three important elements of secondary structure: (a) a stretch of the central ␤-strand; (b) the Arg-166 residue known to be crucial for catalysis (13); and (c) the ligands stabilizing the fourth Ca ion. Thus, interfering with Cys-121-Cys-183 disulfide bond formation appears to be incompatible with proper enzyme folding.
The k cat and K m values for the active mutants are summarized in Table I. No significant changes were observed in the substrate affinity of these mutant enzymes as compared with wt PLAP. In the case of [S101]PLAP, even the catalytic efficiency was similar to that of wt PLAP. We examined the possibility that Cys-101 would be available for covalent modification by the fluorescent probe ABD-F. Only when wt PLAP was incubated with guanidinium chloride did the Cys-101 side chain become available for covalent modification (Fig. 4A). This confirms that Cys-101 is not likely to play a role in regulating activity or stability of the enzyme. In contrast, [S467]PLAP, [S474]PLAP, and [S467, S474]PLAP enzymes all displayed k cat values that were 2-fold lower than the wt PLAP value of 460 s Ϫ1 . The double [S467, S474]PLAP mutant was marginally more active than the [S467]PLAP or [S474]PLAP enzymes. The stability of these PLAP mutations toward heat inactivation was tested at 68°C in the same buffer used for the activity measurements (Fig. 4B). All four cysteine mutants displayed stability similar to that of wt PLAP, with [S467]PLAP and [S474]PLAP being slightly less stable than the other enzymes. When compared with other PLAP mutations known to affect the heat stability, such as the E429G mutation, one can conclude that no significant stability changes occurred with any of these cysteine mutations.
Significance of the Conserved Y367-An interesting structural feature of PLAP, with no counterpart in the ECAP structure, is Tyr-367. This residue is part of the subunit interface in the PLAP dimer, where it protrudes from one subunit and is positioned within 5.6 Å of the catalytic Zn1 ion in the active site of the other subunit (Fig. 5A). The HO Ϫ group of Tyr-367 also forms a hydrogen bond to the peptide group nitrogen of His-432, a direct ligand to the Zn1 ion (4). The location of Tyr-367 in the structure and the fact that this residue is perfectly conserved in mammalian APs (Fig. 1) implicate this residue in an important structural/functional role.
One of the specific properties of APs of higher organisms is their ability to be inhibited stereo-specifically by L-amino acids and peptides through an uncompetitive mechanism (23). Our previous studies have revealed that the nature of residue 429 has a profound effect on the inhibition. The E429G, E429S, and E429H substitutions, all naturally occurring substitutions ( Fig. 1), had the same effect of facilitating the access of the inhibitor to the active site (10). Other residues located near 429 include Tyr-367 and His-317, and their participation in the inhibition mechanism has not been investigated previously. Most of the specific inhibitors of mammalian APs contain a hydrophobic side chain, such as in L-Phe or L-Leu. The nature of this side group affects the inhibition properties, and it has been postulated that a hydrophobic pocket exists to accommodate this part of the inhibitor molecule (24), although its exact location remains unclear. We postulated that Tyr-367 might play a role in stabilizing the inhibitor.
We  (Table I). The k cat values were in the range of 39 -45% of the wt PLAP value. The K m values were not significantly changed from the value of wt PLAP (0.35 mM). However, the Y367A and Y367F substitutions significantly compromised the heat stability of the mutant PLAP enzymes (Fig. 6). In addition, whereas [F367]PLAP, like wt PLAP, displayed a monophasic inactivation curve, [A367]PLAP instead displayed a biphasic inactivation mechanism. Interestingly, the [A367]PLAP mutant was more stable than the [F367]PLAP enzyme after a prolonged incubation, despite a higher initial rate of inactivation.
We then examined the inhibition properties of the mutants toward L-Phe, L-Leu, and L-(2-phenyl)-glycine (Fig. 7). Because all these enzyme variants had comparable kinetic parameters, any observed changes in inhibition properties should mainly reflect the changes in true binding constants for the inhibitors studied (13). The Y367A mutation in PLAP was found to have a very profound destabilizing effect on the inhibition by L-Phe and L-Leu increasing their K i values by 17.5-and 12.7-fold,  (Table II). The Y367F substitution had similar but less pronounced consequences, increasing K i values by 3.2-and 4-fold, respectively. These results show that the side chain of Tyr-367 is crucial for the binding of inhibitors to PLAP, most likely by providing a local binding area for the hydrophobic side chains of Leu and Phe. In the double mutants, [A367, G429]PLAP and [F367, G429]PLAP, the pattern of inhibition by L-Leu and L-Phe was intermediate. The effect of the Y367A substitution overweighed that of E429G, whereas the effect of Y367F was completely rescued by the E429G mutation (Fig. 7). Inhibition of [F367]PLAP and [A367]PLAP by D-Phe was even less pronounced than the inhibition by L-Phe and was so low that accurate determination of K i was difficult (K i Ͼ 90 mM) (data not shown). Whereas the effect of the Y367A mutation was the most significant, the H317A substitution also had a mild influence on inhibition by both L-Phe and L-Leu, decreasing K i by 2.6-and 2.5-fold, respectively. The [A317, G429]PLAP double mutant displayed a similar decrease in K i compared with [G429]PLAP. We also studied the inhibition of PLAP mutants by L-(2-phenyl)-glycine, which, from a structural viewpoint, can be considered a "truncated" form of L-Phe. The results (Table II) show that L-(2-phenyl)-glycine is as potent an inhibitor of PLAP and PLAP mutants as L-Phe and L-Leu. However, L-(2-phenyl)-glycine does not discriminate between the Y367A and Y367F mutations, suggesting that its shorter hydrophobic side chain makes less contact with Tyr-367. Our results indicate that substitutions at positions 367 and 429 can act independently in determining inhibition properties of PLAP. It is clear that removal of Glu-429 facilitates access of the inhibitor to the active site, whereas removal of Tyr-367 eliminates stabilization of the inhibitor's positioning at the active site.
To better explain the inhibition results, we performed a docking simulation of the binding of L-Phe and L-Leu to wt PLAP. For that purpose, we used the program FlexX (14), which performs flexible ligand docking by an incremental construction mechanism. Ligands with nonprotonated amino groups were used, in agreement with the conclusions from previous experimental studies on the pH dependence of the inhibition (13). The enzyme active site was defined as a spheric area 9 Å around the Zn1 atom. In agreement with the discussion in Ref. 14, we found the results of docking simulation to be sensitive to the selection of the "base fragment" of the ligand, i.e. the first group for which the program tries to find the putative positions in the active site. Automatic selection of the base fragment led to a cluster of solutions located far from the active site atoms, which was not relevant for the inhibition modeling. Hence we used manual selection of the base fragment and tried all three main fragments of ligand: carboxylic, amino, and hydrophobic group (the phenyl ring for phenylalanine and a methyl group for leucine). After manual selection of the carboxylic group as the base fragment, results were similar to the automatic mode. When the amino group or the hydrophobic moiety was chosen as the base fragment, the ligands docked close to the active site Zn1, and a similar orientation of the hydrophobic chain was obtained for both inhibitors. Furthermore, using the hydrophobic group as the base fragment also positioned the inhibitor at coordination distance to the active site Zn1, a feature that we previously found to be important in the inhibition mechanism (13). The results for L-Phe and L-Leu are presented in Fig. 5, B and C, respectively, where the conformations with the highest energy score are shown. As can be seen, the predicted location of the hydrophobic group is similar for L-Leu and L-Phe. The pocket accommodating this hydrophobic group is formed by the side chains of Tyr-367, Phe-107, Gln-108, and Asp-91. The polar groups of the ligands are found to interact with Zn1 ion and with the side chains of Glu-429 and Tyr-367. This docking model is consistent with our mutagenesis studies and provides an explanation for the important role of Tyr-367 in the inhibition of PLAP by amino acids.
Whereas the hydrophobic phenyl ring of Tyr-367 is important for the formation of the hydrophobic pocket in PLAP, the HO Ϫ group of Tyr-367 might help maintain the correct orientation of phenyl ring via a hydrogen bond to the backbone nitrogen of His-432. This HO Ϫ group may additionally be involved in a hydrogen bond with the polar groups of the ligand, as was found in some of the docking solutions.
In conclusion, our analysis of the structure-function relationship of residues conserved throughout evolution from the E. coli to the mammalian enzymes has revealed largely conserved functions for those residues that stabilize the active site Zn and Mg metal ions. The nonhomologous disulfide bonds differ in their structural significance, and the free cysteine in mammalian APs can be substituted without major consequences to enzyme function. Significantly, we found that nonconserved residues that contribute active site stabilization and cross-talk between enzyme monomers also determine the heat stability and uncompetitive inhibition properties of mammalian alkaline phosphatases.