Structural basis for the catalytic activity of human serine/threonine protein phosphatase-5.

Serine/threonine protein phosphatase-5 (PP5) affects many signaling networks that regulate cell growth and cellular responses to stress. Here we report the crystal structure of the PP5 catalytic domain (PP5c) at a resolution of 1.6 A. From this structure we propose a mechanism for PP5-mediated hydrolysis of phosphoprotein substrates, which requires the precise positioning of two metal ions within a conserved Asp271-M1:M2-W1-His427-His304-Asp274 catalytic motif (where M1 and M2 are metals and W1 is a water molecule). The structure of PP5c provides a structural basis for explaining the exceptional catalytic proficiency of protein phosphatases, which are among the most powerful known catalysts. Resolution of the entire C terminus revealed a novel subdomain, and the structure of the PP5c should also aid development of type-specific inhibitors.

Serine/threonine protein phosphatase-5 (PP5) 1 is a member of the PPP gene family of protein phosphatases that is widely expressed in mammalian tissues and is highly conserved among eukaryotes. In human breast carcinoma cells (MCF-7) the expression of PP5 is responsive to 17␤-estradiol, aiding estrogen-induced cell growth without affecting the expression of c-Myc or cyclin D1 (1). When overexpressed, PP5 inactivates apoptosis signal-regulating kinase-1 (2) and converts MCF-7 cells from an estrogen-dependent into an estrogen-independent phenotype (1). How PP5 affects cell growth is not entirely clear. PP5 associates with several proteins that affect signal transduction networks, including the glucocorticoid receptor (GR)heat shock protein 90 heterocomplex (3), the CDC16 and CDC27 subunits of the anaphase-promoting complex (4), cryptochrome 2 (5), eukaryotic initiation factor 2␣ kinase (6), the A subunit of serine/threonine phosphatase type 2A (7), the G12-␣/G13-␣ subunits of heterotrimeric G proteins (8), and DNAdependent protein kinase (9). The binding of PP5 to a GR-heat shock protein 90 complex occurs in the cytoplasm in a competitive manner with three immunophilin proteins: Cyp40, FKB51, and FKB52 (9). When PP5 expression is suppressed, ligand-independent nuclear translocation of GRs occurs (10), and dexamethasone-induced phosphorylation of p53 at Ser 15 is augmented (11), thereby enhancing nuclear retention of p53 (12), potentiating the p53-dependent transcription of p21 Waf1/cip1 (13), and facilitating GR-mediated growth suppression (11). PP5 has also recently been reported to play a role in the repair of DNA double-stranded breaks in association with DNA-dependent protein kinase (14). To date, the physiological/ pathological roles of PP5 in other signaling networks are not fully understood.
The catalytic domain of PP5 (PP5c) shares 35-45% sequence identity with the catalytic domains of other PPP phosphatases, including protein phosphatase 1 (PP1), 2A (PP2A), 2B/calcineurin (PP2B), 4 (PP4), 6 (PP6), and 7 (PP7). Like PP1, PP2A, and PP4, PP5 is also sensitive to inhibition by okadaic acid, microcystin, cantharidin, tautomycin, and calyculin A (15). PP5 differs from PP1-PP7 in that it contains an extended N-terminal domain with three tetratricopeptide repeat (TPR) motifs. The TPR domain acts as an interface for protein-protein interactions and likely provides a means of targeting PP5 to particular regions within the cell or facilitating interactions with substrates (16). In vitro, proteolytic removal of the TPR domain produces a marked increase in catalytic activity suggesting that there is also an autoinhibitory function for the N-terminal region (17,18). Mutagenesis studies suggest that amino acids within the 13 C-terminal residues of PP5 also participate in the negative regulation of activity, possibly in a coordinated manner with the TPR domain (18,19).
To aid the analysis of PP5 function, we expressed and purified PP5c, encompassing residues 169 -499, and determined its three-dimensional structure by x-ray crystallography at a resolution of 1.6 Å. From this structure we inferred a mechanism for PP5-mediated hydrolysis of phosphoprotein substrates. In addition, resolution of the entire C terminus reveals the structure of the C-terminal subdomain of PP5 and its interactions with the rest of the catalytic domain.

MATERIALS AND METHODS
Expression and Purification of PP5c-PP5c (residues 169 -499) was expressed as a maltose-binding fusion protein with a linker sequence containing a hexahistidine tag and a tobacco etch virus protease cleavage site, using a modified pMal2cE vector. The MBP-PP5c fusion protein was expressed in BL21 (DE3) Tuner pLacI cells and purified using a nickel-iminodiacetate-Sepharose resin. The purified fusion protein was then digested with tobacco etch virus protease leaving two vectorderived amino acid residues (Gly-Ala) at the N terminus of PP5c, and free PP5c was purified via anion exchange chromatography using Q-Sepharose resin. The active fractions, identified by activity against p-nitrophenyl phosphate, were dialyzed against the crystallization buffer containing 40 mM Tris, pH 7.5, 1 mM tris-carboxyethyl phosphine, 1 mM sodium azide. The final PP5c preparation was determined to be 20.2 mg/ml by Bradford assay and was ϳ98% pure as judged by SDS-PAGE.
Crystallization, X-ray Diffraction, and Fluorescence-Hanging droplets containing a 1:1 mixture of the protein and reservoir solution were equilibrated at 22°C against reservoir solutions containing 45-60% 2-methyl-2,4-pentanediol. For data collection, the crystals were flash frozen directly in liquid nitrogen and transferred to the x-ray beam for data collection. The crystals contained ϳ30% solvent (V m ϭ 1.8) and belonged to the P2 1 space group with unit cell parameters of a ϭ 40.8 Å, b ϭ 80.3 Å, c ϭ 92.2 Å, ␣ ϭ 90°, ␤ ϭ 94.3°, and ␥ ϭ 90°. Two PP5c molecules were present in the asymmetric unit. The diffraction data extending to a resolution of 1.47 Å were integrated and reduced with HKL2000 (Table I) (20). Fluorescence and anomalous dispersion spectra from a crystal of PP5c revealed distinct signals at energies of 6553.88, 7134.39, and 9635.95 eV, indicating the presence of Mn 2ϩ , Fe 2ϩ , and Zn 2ϩ ions at a ratio of 1:0.35:0.15, respectively, that compete for two metal-binding sites in the catalytic domain. In light of this ambiguity, metal ions were refined with the number of electrons corresponding to Mn 2ϩ ions.
Structure Determination and Refinement-The structure of PP5c was determined by molecular replacement using a homology model constructed with SWISS-MODEL (21) from fragments of PP1 (Protein Data Bank (22) accession numbers 1FJM and 1JK7), PP2B (Protein Data Bank (22) accession numbers 1AUI and 1TCO), and bacteriophage -phosphatase (PP) (Protein Data Bank (22) accession number 1G5B). The homology model contained approximately three-quarters of the PP5c sequence (residues 226 -479). The molecular replacement search resulted in solutions of the rotation function with a correlation value of 36.5 for the highest peak and 32.7 for the second peak and an initial R factor of 0.43, as defined in MOLREP (23,24). Correctly oriented partial models of both PP5c molecules were then subjected to a de novo model building and refinement procedure implemented in the programs wARP (23,25) and REFMAC (23,26). Subsequent rounds of refinement alternating with manual model building using XtalView (27) yielded the positions of metal and phosphate ions and all of the residues at the C-terminal end of the protein. The final cycles of refinement were performed using REFMAC combined with the translation libration screw method (23,26,28). The final content of the asymmetric unit consists of two PP5c molecules, each with two metal ions and one phosphate ion, with residues 176 -499 in monomer A and residues 175-499 in monomer B. There are also two 2-methyl-2,4-pentanediol and 491 water molecules in the asymmetric unit. Side chains of 14 amino acids were refined in two alternative conformations. The final model was analyzed with PROCHECK (29) showing 99.2% of the residues in the allowable regions of the Ramachandran plot except for Asp 274 and Arg 275 from both PP5c molecules. In each case, those resi-dues have well defined electron density. The final refinement statistics are shown in Table I.

RESULTS AND DISCUSSION
PP5c Fold-The structure of PP5c encompasses the entire catalytic domain (residues 169 -499), which includes 18 residues (residues 169 -187) that link the catalytic domain to the TPR domain, and a C-terminal subdomain (residues 477-499). PP5c has a compact ␣/␤ fold comprised of 11 ␣-helices and 14 ␤-strands (Fig. 1). Eleven of 14 strands (␤1-␤6 and ␤10 -␤14) form two ␤-sheets that are arranged in a ␤-sandwich structure in the center of the molecule. One side of this ␤-sandwich is occupied by three short helices (␣G-␣I) and a three-stranded ␤-sheet (␤7-␤9) that nearly forms a separate subdomain. On the other side of this ␤-sandwich there is an ␣-helical bundle that includes six helices (␣A-␣F) plus the AЈ-helix that is immediately C-terminal to the TPR linker.
The C-terminal subdomain is composed of the J-helix tethered by a 14-amino acid linker extending from the terminal residue Phe 476 of the ␤14-strand. His 481 of the linker forms hydrogen bonds with the ␤-carboxylate oxygen of Asp 453 located on the ␤12-␤13 loop near the catalytic site. From that position, the linker continues around the residues of the ␤3-␣C loop and the C-helix carrying three important catalytic residues, thus reinforcing its interactions with the catalytic domain through the contact between the amido nitrogen of Met 487 and the carbonyl oxygen of Arg 275 (2.9 Å).
Catalytic Site-The catalytic site of PP5, identified from the positions of two metal ions and a bound phosphate ion, is located at the base of a shallow depression on the surface formed by the residues of four loops: ␤4-␣D, ␣G-␣H, ␤10-␤11, and ␤12-␤13 (Fig. 1). X-ray fluorescence and anomalous dispersion from PP5c crystals revealed the presence of Mn 2ϩ , Fe 2ϩ , and Zn 2ϩ ions within two binding sites. In light of this ambiguity, we refer to the metals as M 1 and M 2 , based upon the coordinating residues. Each metal ion is coordinated by six ligands in a slightly distorted octahedral geometry, with the bridging carboxylate oxygen of Asp 271 and a water molecule (W 1 ) each coordinating both M 1 and M 2 (Fig. 2 (Fig. 2). The arginine residues along with His 304 when protonated and the metal ions create the only large pocket of positive electrostatic potential on the surface of PP5c, and mutation of either of the equivalent residues in PP1 or PP2B perturbs substrate binding and catalysis. Further mutational studies indicate that Asp 274 and His 304 in the catalytic motif are also critical for catalytic function (30,31). In the structure of PP5c, O 4 of the phosphate ion forms a hydrogen bond (2.6 Å) with the nearby His 304 residue, which also forms a hydrogen bond (2.7 Å) with the ␤-carboxylate oxygen of Asp 274 (Fig. 2B).
Catalytic Mechanism-Over the years, several mechanisms have been proposed for the PPP family phosphatases (32). In 1985, Martin et al. (33) put forward three initial mechanisms for PP2B. The first involved nucleophilic attack by an active  site residue with the formation of a phosphoenzyme intermediate. In the second putative mechanism, direct transfer of the phosphoryl group to water proceeds via nucleophilic attack by a metal-ligated water molecule that is activated to hydroxide. The third mechanism involved unimolecular breakdown of the metal-ligated substrate after protonation of the substrate by the enzyme to form a monoanion. Subsequent work (32)(33)(34) indicated that PPP family phosphatases do not form phosphoenzyme intermediates, thus ruling out the first putative mechanism. Also, kinetic isotope effect data indicated that PPP gene family phosphatases act upon the dianionic forms of their substrates (35), making the third hypothesis highly unlikely. This leaves nucleophilic attack by metal-ligated water as the most viable catalytic mechanism for PPP family phosphatases. However, there are still several candidates for the metal-activated nucleophile. For example, in the case of PP5c, the bridging water (W 1 ) and the nonbridging water (W 2 ) would both be possibilities. In addition, the native substrates of PP5 are phosphorylated polypeptides that will likely make numerous contacts with PP5 upon binding and may impose restraints upon the positioning of the phosphoryl group within the active site. Therefore, because the phosphoryl group may bind to the enzyme in a fashion that differs from the binding mode of phosphate shown in the PP5c structure, water molecules occupying the metal-coordinating positions taken up by the phosphate oxygen ions O 1 and O 2 must also be considered.
Several mechanisms have proposed binding modes for the substrate phosphoryl group that differ from the bidentate bind-ing of phosphate observed in the PP5c structure. One such mechanism (36), proposed for PP2B, models the phosphoryl group with a nonbridging oxygen displacing the bridging water and the leaving group oxygen coordinated to one of the metal ions, which would function to neutralize the negative charge on the leaving group in the transition state. The nucleophile in this case would be a hydroxide ion terminally ligated to the other metal ion. Nucleophilic attack by this hydroxide would produce the phosphate product bound to the dimetal center in a tridentate fashion, similar to that observed for sulfate in a complex with PP (37). Although the mechanism is certainly feasible, this binding mode for the phosphoryl group, when modeled into the active site of PP5c, would require significant perturbation of active site residues near the metal ions to accommodate the seryl/threonyl residue of the polypeptide substrate to which the leaving group oxygen is attached. Such steric interactions could be especially problematic (potentially involving metal-coordinating residues) if the leaving group is coordinated to M 2 . Also, in the absence of significant perturbation of the geometry and coordination environment of the dimetal center, this binding mode leads to a rather acute angle between the nucleophile, phosphorus, and leaving group oxygen that is not optimal for in-line attack. On the other hand, modeling the leaving group at one of the noncoordinating positions (i.e. pointing away from the plane formed by the carboxylate oxygen of Asp 271 , M 1 , M 2 , and W 1 ) avoids the necessity of rearranging active site residues to accommodate the leaving group and allows for an ϳ180°angle between the nucleophile, phosphorus, and leaving groups. Charge neutralization in this scheme could be accomplished through interaction of the leaving group with positively charged residues in the active site (Arg 275 , Arg 400 , and His 304 ) and/or general acid catalysis by His 304 .
Another possible mechanism, proposed originally for PP2B, has a monodentate binding mode for the phosphoryl group with the substrate leaving group oxygen coordinated to M 2 and nonbridging phosphoryl oxygens hydrogen-bonded to the active site arginines (38). Again, the nucleophile is a hydroxide terminally ligated to the other metal, and again, M 2 would help to neutralize negative charge on the leaving group in the transition state. Additional charge neutralization is provided in this mechanism via protonation of the leaving group oxygen by His 304 . As above, this binding mode would require rearranging active site residues to accommodate the leaving group. The nucleophile-phosphorus-leaving group angle would also be less than optimal.
All of the proposed mechanisms that rely upon a terminally ligated hydroxide nucleophile require some means of deprotonating the terminally ligated water molecule. The presence of a trivalent metal ion, such as Fe 3ϩ , in the active site could accomplish this nicely by dramatically lowering the pK a of a terminally coordinated water, thus facilitating the loss of a proton through equilibration with solvent/buffer such that hydroxide is the predominate species present at physiological pH. Ferric ion, when coordinated by neutral ligands (as, for example, in the hexaaquo complex), can lower the pK a of a ligated water from 15.7 to Ͻ3 (39); negatively charged ligands in the coordination sphere would tend to increase the pK a . If a trivalent metal ion is unavailable (active site metals are divalent Fe 2ϩ , Mn 2ϩ , or Zn 2ϩ ), then the pK a of a terminally ligated water could only be lowered to ϳ9 -11 (39), and an active site general base would be required to form the hydroxide by abstraction of a proton.
The identities of the physiologically relevant metal ions in most PPP family phosphatases are unknown. However, for PP2B, they have been identified as Fe 2ϩ and Zn 2ϩ ions (40). Because the catalytic domains of the PPP family phosphatases are so highly conserved, PP2B has been used as a model system for studying the PPP family (32), and so by analogy, the physiological metal ions in other PPP family phosphatases are typically assumed to be Fe 2ϩ and Zn 2ϩ as well. The oxidation state of the active site Fe 2ϩ in PP2B has been debated for years (41)(42)(43). However, recent work by Namgaladze et al. (44) and Ullrich et al. (45) indicates that the physiologically active en- zyme has ferrous ion and divalent Zn 2ϩ in the active site. This casts doubt upon previous studies that used the results of dithionite titration to argue for the presence of ferric ion in the active form of PP2B (41). Thus, the argument that PPP family phosphatases have a trivalent active site metal is not on a firm foundation at this time. In addition, the physiological active site metals in PP have recently been identified as Mn 2ϩ (46). Therefore, if there is a conserved catalytic mechanism within the PPP family, divalent active site metals should be sufficient.
The acidic limbs of the pH rate profiles of PPP family phosphatases have generally been attributed to the ionization of a metal-ligated water (35, 47) but could possibly be interpreted instead as evidence of a residue that needs to be deprotonated for activity, such as would be the case for a general base that is needed to deprotonate a metal-ligated water to hydroxide. However, the fact that the acidic limb of the pH-k cat profile is still present for the PP mutant in which His 76 (equivalent to His 304 in PP5) is changed to an asparagine residue strongly militates against a general base function for the active site histidine (48). There are no other active site residues in the PPP family phosphatases that would be good candidates for a general base function. Alternatively, if protonated His 76 serves to further reduce the pK a of a metal-ligated hydroxide by additional stabilization of its negative charge and the acidic limb of the pH-k cat profile is due to ionization of a metal-ligated water, then the pK a of the acidic limb (and the pH optimum) for the H76N mutant should increase significantly relative to the wild-type enzyme. This is not observed (48). In addition, incoming substrate acting as a general base to deprotonate the incipient nucleophile in a concerted fashion has been ruled out by the 18(V/K) nonbridge kinetic isotope effects measured for p-nitrophenyl phosphate hydrolysis catalyzed by PP2B and PP (35,48).
Therefore, although it is important to emphasize that one cannot definitively rule out any of the aforementioned mechanisms that utilize a terminally ligated nucleophile, the concerns alluded to above lead us to propose a mechanism for PP5 in which the nucleophile is a bridging hydroxide ion ligated to both metal ions in the active site and the leaving group oxygen does not coordinate an active site metal. The dianionic phosphate ion mimics the dianionic phosphoryl group of a substrate, thus providing a means of modeling substrate interactions with the enzyme. It is reasonable to assume that the substrate phosphoryl group might bind to PP5 in a bidentate mode similar to that of phosphate ion shown in Fig. 2. In this model, the phosphoester bond oxygen of a substrate would correspond to the phosphate oxygen (O 4 ) that forms a hydrogen bond to His 304 , which would be perfectly positioned to act as a general acid in the phosphoester hydrolysis reaction by protonating the leaving group alcohol/alkoxide. Even if general acid catalysis does not occur, protonated His 304 , by virtue of its positive charge, may still contribute (in conjunction with Arg 275 and Arg 400 ) to the neutralization of negative charge on the leaving group in the transition state. Modeling the leaving group oxygen O 4 in this position also allows for a reasonable binding mode for phosphopeptide/phosphoprotein substrates.
As mentioned above, PPP gene family phosphatases do not form detectable phosphoenzyme intermediates, thus ruling out active site residues as nucleophiles (32,34,49). Rather the nucleophile in the catalyzed reaction is likely one of the two metal-ligated water molecules, W 1 or W 2 , that become deprotonated to hydroxide prior to attack (32,34,49). The order of nucleophilicity of water-derived species is generally as follows: (50). Although a bridging interaction of W 1 with the Lewis acidic metal centers might be expected to reduce the nucleophilicity of a water molecule below that of the corresponding terminally ligated W 2 , two metal ions can better stabilize the charge of a coordinated ligand. This means that W 1 in the bridging position should have its pK a lowered to a greater extent than would occur in a terminally ligated position (35,47), thus, at the approximately neutral pH optima of these enzymes, producing a much greater population of reactive hydroxide nucleophiles at the bridging position. The nonbridging water W 2 is also significantly farther away (4.0 Å) from the phosphorus atom than is W 1 (2.9 Å) and is not positioned well for an in-line attack on the phosphorus atom (W 2 -P-O 4 angle of ϳ130°). On the other hand, W 1 not only makes a close contact with the phosphorus atom (less than the van der Waals' contact distance) but also forms a W 1 -P-O 4 bond angle of nearly 180°. In addition, the main chain carbonyl oxygen of His 427 forms a short hydrogen bond with W 1 (2.6 Å) with a bond angle between the carbonyl O, W 1 , and P of ϳ109°. Therefore, the interaction of W 1 with His 427 and the dinuclear metal center help fix the orientation of a lone pair of electrons from the hydroxide W 1 toward the phosphorus atom, thereby helping to reduce the entropic barrier for nucleophilic attack (Figs. 2 and 3). The interaction between the carbonyl oxygen of a metal-coordinating His 427 and W 1 is highly conserved. From an examination of atomic coordinates for PP1, PP2B, PP, a purple acid phosphatase (uteroferrin), and mre11 nuclease, the interaction is seen to be present not only within the PPP family phosphatases (37,(51)(52)(53)(54)(55) but is also present in several distantly related phosphoesterases. This evolutionary conservation of what appears to be a mechanism for orienting a bridging nucleophile bolsters the case for W 1 being the hydroxide nucleophile in the reaction. Further support for the bridging hydroxide model comes from ENDOR spectroscopic data of free uteroferrin and its complexes with phosphate ions (56). The use of the uteroferrin-PO 4 complex as a model for the catalytically competent enzymesubstrate complex is supported by the observation that uteroferrin catalyzes the exchange of 18 O from phosphate to H 2 18 O (57).
These considerations and the structure of the catalytic site of PP5c lead us to propose the mechanism of metal-ion mediated hydrolysis of phosphorylated substrates shown in Fig. 3 where the bridging hydroxide W 1 (Fig. 3, Step 1), serves as the nucleophile that attacks the phosphorus atom of the substrate phosphoryl moiety. Based on the structure of PP5c, this hydroxide is precisely positioned for in-line nucleophilic attack by the metal ions and the carbonyl oxygen of His 427 . Kinetic isotope effects measured for PP2B (35) and PP (58) indicate a highly dissociative transition state for the reaction catalyzed by the PPP gene family phosphatases, with little bond formation to the nucleophile and advanced bond dissociation to the leaving group oxygen. Because the entering nucleophile W 1 and leaving group are both held in place through multiple interactions to the enzyme, neither of them is likely to move much during the formation of the transition state. Instead, as the reaction progresses from Step 2 to Step 3 (Fig. 3), the phosphorus atom moves along the reaction coordinate toward W 1 , and the bond angles between the equatorial P-O bonds increase such that a planar metaphosphate-like dissociative transition state is formed (Fig. 3, Step 3) with the attacking W 1 and leaving group oxygen O 4 occupying axial positions with low bond order (47). Kinetic isotope effect studies of phosphoryl transfer reactions with dissociative transition states (58) and ab initio studies of metaphosphate (59) have challenged the validity of the classical representation of metaphosphate with five bonds to phosphorus, indicating instead that the equatorial P-O bonds are highly polarized with little double-bond charac-ter. These polarized bonds in the transition state are stabilized by interactions with the metal ions and the active site residues Arg 275 , Arg 400 , and Asn 303 . The movement of the equatorial P-O bonds into a planar metaphosphate-like moiety also improves coordination geometry for the two metal-bound O 1 and O 2 from the phosphoryl group in the ground state, whereas the negative charge development on the leaving O 4 strengthens its interactions with positively charged Arg 275 and His 304 . Thus, transition state stabilization is provided through strong interactions with the dinuclear metal center and four conserved active site residues, i.e. Arg 275 , Arg 400 , Asn 303 , and His 304 (30, 60 -62).
The strong electrostatic effects conveyed by the dinuclear metal center underlie a major part of its catalytic role in the phosphoester hydrolysis. The metal ions, by stabilizing negative charges, can not only activate the bridging water but can also make the phosphate moiety more susceptible to nucleophilic attack by stabilizing the polarized P-O bonds that develop during the attack. Similarly, the active site arginine residues are not only necessary for binding and orienting the substrate, they also provide stabilizing interactions for the transition state. As the reaction continues past Step 3 toward Step 4 (Fig. 3), the bond order between the phosphorous atom and W 1 increases, and bond dissociation to the leaving group O 4 nears completion. At some point during dissociation of the bond, the pK a of the leaving group oxygen becomes high enough to accept a proton from His 304 , which acts as a general acid. As the reaction reaches completion (Step 4), the hydroxide nucleophile W 1 becomes a hydroxyl group of the phosphate dianion product, which would be bound to M 1 and M 2 in a tridentatelike fashion similar to one binding mode reported for sulfate ion complexed with PP (37). Ejection of the phosphate product may be facilitated by this tridentate binding, which would be less stable than the bidentate binding because of the strained coordination geometry and less favorable electrostatic interactions with the metal ions.
In the structure of PP5c, the distance between W 1 and the phosphate phosphorous atom is 2.9 Å (Fig. 2B). If we take this to be the reaction coordinate distance in the enzyme-substrate complex just prior to nucleophilic attack, the Pauling bond order relationship can be used to estimate the degree of associativity of the reaction mechanism under the simplifying as- sumption that only the phosphorus atom moves along the reaction coordinate to form a symmetric transition state in which the PO 3 moiety is equidistant from the entering and leaving atoms (63). If a value of 1.6 Å for the length of a single phosphorous-oxygen bond is used, then the bond order to the entering atom in the transition state for the reaction catalyzed by PP5 would be ϳ0.076. An associativity of 7.6% is consistent with the kinetic isotope effects measured for the related PPP gene family members, PP2B and PP phosphatases, which clearly indicate a highly dissociative mechanism for these enzymes (35,58).
This mechanism requires the regeneration of the hydroxide nucleophile W 1 and protonated His 304 to begin another catalytic cycle. In view of the relatively low turnover (k cat ) of phosphoprotein/phosphopeptide substrates by the PPP phosphatases (64), it is likely that each catalytic cycle allows enough time for the regeneration of the active ionization state of the enzyme (i.e. bridging hydroxide nucleophile and protonated His 304 general acid) to occur through the exchange of protons with intracellular solvent and/or buffer molecules (e.g. organic acids, phosphates, and free amino acids), which, for relatively solvent-accessible groups, can occur at rates far greater than the observed k cat values of PPP gene family phosphatases (65). Although it is tempting to propose, in analogy to the His-Asp-Ser catalytic triad of the serine proteases, that His 304 acts as a general base to deprotonate the newly accrued bridging water, no firm kinetic evidence in support of general base catalysis in PPP family phosphatases exists (63). Still, such a function for His 304 cannot be ruled out.
Implications of PP5c Structure for Other PPP Gene Family Phosphatases-In the absence of enzyme, the half-time for the hydrolysis of alkyl phosphate dianions at 25°C is over 1 trillion years; k non ϭ ϳ2 ϫ 10 Ϫ20 s Ϫ1 (64). Because typical substrate turnover rates (k cat ) for PPP gene family phosphatases range from 1 to 100 s Ϫ1 , these enzymes enhance the rate of hydrolysis by a factor of ϳ10 21 , placing them among the most powerful known catalysts, with catalytic proficiencies ([k cat /k M ]/k non ) of ϳ10 25 -10 26 M Ϫ1 (30,64,66). Such a remarkable rate enhancement would require not only the activation of a water molecule to a more nucleophilic hydroxide but also the exquisite alignment of the attacking nucleophile with the substrate and profound stabilization of the altered substrate in the transition state. The high resolution structure of PP5c reveals an arrangement of the catalytic site suggestive of the near attack configuration for the mechanism proposed above, with distances and angles optimal for phosphomonoester hydrolysis through an in-line nucleophilic attack. The structure shows that two conserved residues, Asp 274 and His 304 , together with Asp 271 , His 427 , and W 1 coordinating to two metal ions form a catalytic motif, Asp 271 -M 1 :M 2 -His 427 -W 1 -His 304 -Asp 274 . This catalytic motif is common to all members of the PPP gene family phosphatases, as is the hydrogen bond network that ensures this catalytic activity. Catalysis requires the precise alignment of W 1 and the substrate phosphoryl group, represented in the PP5c structure by the bound phosphate ion, for in-line nucleophilic attack. This alignment, facilitated by substrate contacts with M 1 , M 2 , Arg 275 , Asn 303 , His 304 , and Arg 400 and interactions of W 1 with M 1 , M 2 , and His 427 , significantly enhances the rate of catalysis by separating the entropic costs of properly positioning and orienting the reactants from the actual chemical steps of the reaction.
Superposition of the structure of PP5c onto the structures of other eukaryotic PPP gene family phosphatases such as PP1 (␥-isoform) and PP2B gives root mean square deviations of 2.0 Å for both structures within the 269-residue region used in calculations (Fig. 4). Within that superposition, comparison of the loop regions of PP5, PP1, and PP2B reveals conformational differences, notably in the ␤12-␤13 loop that plays an important role in toxin-mediated inhibition of catalytic activity (67)(68)(69), suggesting the feasibility of developing type-specific inhibitors. Structural similarities among the eukaryotic PPP gene family phosphatases also include the orientation and packing of the AЈ-helix against the C-helix adjacent to the ␤3-␣C loop, potentially contributing to the function of the catalytic residues. This structural homology is also evident from the comparison shown in Fig. 4. Thus, despite the lack of significant sequence homology within the AЈ-helix, the structure argues for the recognition of the AЈ-helix as a common structural motif to the eukaryotic PPP gene family phosphatases.
Model for the Regulation of PP5 Activity-Deletion of the TPR domain and/or the C-terminal region from Asn 491 to Met 499 results in a 10 -20-fold increase in basal activity (17)(18)(19), indicating that both the TPR domain and the C-terminal subdomain of PP5 contribute to the regulation of catalytic activity. The structure of PP5c reveals that the C-terminal subdomain forms a number of contacts with the main body of the catalytic domain, including a hydrogen bond between Ala 490 and Ile 312 (3.1 Å) that involves only main chain atoms. Asn 491 also forms an important contact with the ␥-amidyl nitrogen of Gln 495 , contained in the C-terminal J-helix, which helps stabilize the short coil-helix-like structure of the entire C-terminal subdomain. The importance of the interaction between Asn 491 and Gln 495 is supported by biochemical studies where deletion of residues from Gln 495 to Met 499 resulted in a 10-fold elevation of basal activity (19). The structure of PP5 also reveals that the AЈ-helix, just C-terminal to the linker connecting the TPR domain with the catalytic domain forms close contacts with the C-helix, which is immediately adjacent to four amino acids Asp 271 , Asp 274 , Arg 275 , and His 304 that are critical for catalysis. This suggests that interactions of the TPR domain with other proteins could produce a conformational change that would allow the substrate to access the active site. Collectively, the structural and biochemical studies are consistent with a model in which contacts between the TPR domain and the residues of the C-terminal subdomain regulate the basal activity of PP5. In this model, the C-terminal subdomain is sandwiched between the catalytic domain and the TPR domain, forming an inactive state. An active state would be obtained when interactions between the TPR domain and the C-terminal subdomain are disrupted. Such a cooperative involvement of the TPR and the C-terminal subdomain may function to keep free PP5 predominately in an inactive state, allowing activation of PP5 only when the TPR domain interacts with appropriate regulatory molecules.