Mechanistic Studies of Phosphoserine Phosphatase, an Enzyme Related to P-type ATPases*

Phosphoserine phosphatase belongs to a new class of phosphotransferases forming an acylphosphate during catalysis and sharing three motifs with P-type ATPases and haloacid dehalogenases. The phosphorylated residue was identified as the first aspartate in the first motif (DXDXT) by mass spectrometry analysis of peptides derived from the phosphorylated enzyme treated with NaBH4 or alkaline [18O]H2O. Incubation of native phosphoserine phosphatase with phosphoserine in [18O]H2O did not result in 18O incorporation in residue Asp-20, indicating that the phosphoaspartate is hydrolyzed, as in P-type ATPases, by attack of the phosphorus atom. Mutagenesis studies bearing on conserved residues indicated that four conservative changes either did not affect (S109T) or caused a moderate decrease in activity (G178A, D179E, and D183E). Other mutations inactivated the enzyme by >80% (S109A and G180A) or even by ≥99% (D179N, D183N, K158A, and K158R). Mutations G178A and D179N decreased the affinity for phosphoserine, suggesting that these residues participate in the binding of the substrate. Mutations of Asp-179 decreased the affinity for Mg2+, indicating that this residue interacts with the cation. Thus, investigated residues appear to play an important role in the reaction mechanism of phosphoserine phosphatase, as is known for equivalent residues in P-type ATPases and haloacid dehalogenases.

aspartate that is phosphorylated during the catalytic cycle. Lack of precise structural information (the resolution of the best structural models is 8 Å (9, 10)), due to the fact that ATPases, as all membrane proteins, are difficult to crystallize, prevents the full understanding of the detailed mechanism by which the control of phosphoenzyme formation and hydrolysis is exerted by binding of ions and conformational changes.
The newly identified class of phosphotransferases is characterized by a conserved DXDX(T/V) motif close to the N terminus (5). It comprises at least 10 different enzymes that are typically soluble proteins acting either as monophosphate phosphatases (e.g. phosphoserine phosphatase and phosphoglycolate phosphatase) or as phosphomutases (␤-phosphoglucomutase and eukaryotic phosphomannomutase) (4,5,11,12). Formation of an acylphosphate intermediate was shown for these four enzymes (5,(13)(14)(15) and, in the case of phosphomannomutase, the phosphorylated residue was shown to be the first aspartate in the DXDX(T/V) motif (5). Results of sitedirected mutagenesis of the two aspartates in phosphoserine phosphatase were compatible with this conclusion (5).
Iterated sequence comparisons and position-specific iterated BLAST (PSI-BLAST) searches (16) starting from haloacid dehalogenase have shown that this enzyme shares three statistically significant motifs ( Fig. 1) with the new class of phosphomonoesterases/phosphomutases and P-type ATPases (4,11). The first of these motifs (DXDXT in phosphomonoesterases/phosphomutases, DKTGT in ATPases, and DXYGT in dehalogenases) contains an absolutely conserved aspartate, which covalently binds phosphate in P-type ATPases and in phosphomannomutases and an ␣-hydroxy acid in haloacid dehalogenases (17). The second motif contains a strictly conserved serine or threonine, and the third motif contains a strictly conserved lysine residue followed, at some distance, by lesser conserved residues and a strictly conserved aspartate. Some of these residues have been mutated in P-type ATPases (reviewed in Refs. 7 and 18) and in Pseudomonas sp. haloacid dehalogenase (19) and shown to play an important role in catalysis. In addition, the three-dimensional structure of haloacid dehalogenase indicates that the three motifs line the catalytic pocket (20 -22). This group of enzymes appears to be also related to phosphonoacetaldehyde phosphatases, with which they share the first motif (DWAGT), and possibly also part of the third motif (23).
We previously noted that the third motif was particularly conserved between phosphoserine phosphatases and P-type ATPases (15), all these enzymes showing the same GDGXXD sequence (Fig. 1), whereas the corresponding sequence in haloacid dehalogenase is SSXXXD (11). Another difference between ATPases and haloacid dehalogenases is that in dehalogenases, the covalent intermediate is hydrolyzed by attack of the ␥-carbon of the aspartate (17), whereas in ATPases the attack occurs on the phosphorus atom (24). Nothing was known in this respect for phosphoserine phosphatase.
As phosphoserine phosphatase may turn out to be an interesting model for the study of ATPases, the purpose of the present work was to explore further the reaction mechanism of this enzyme by 1) providing definitive identification of the phosphorylated residue, 2) determining the site of attack of the phosphoryl enzyme intermediate by water, and 3) investigating the effect of mutations of conserved residues of the second and third motifs, which have not yet been studied until now. Identification of the Phosphorylated Residue-Phosphorylated phosphoserine phosphatase was prepared in the following manner: 500 g of the human recombinant enzyme was incubated at 0°C in a mixture (500 l) containing 25 mM Hepes, pH 7.5, 20 mM serine, 10 mM phosphoserine, 1 mM dithiothreitol, and 5 mM MgCl 2 ; a control was incubated in the absence of phosphoserine. The reaction was stopped after 5 s by the addition of 1 ml of 10% ice-cold trichloroacetic acid. After 10 min on ice, the mixture was heated 10 min at 30°C to prevent renaturation of the enzyme (this treatment does not significantly hydrolyze the phosphoenzyme), placed on ice for 5 min, and then centrifuged for 20 min at 10,000 ϫ g and 4°C. The pellet was washed in 200 l of 10 mM ice-cold HCl, centrifuged for 10 min at 4°C, dried under vacuum, and used for reduction with borohydride or for alkaline hydrolysis. Previous results have shown that such conditions result in the incorporation of about 0.07 mol of phosphate/mol of enzyme subunit (15).

Materials-Glucose
For the reduction with borohydride, the protein pellets corresponding to phosphorylated and control enzymes were resuspended in 67 l of Me 2 SO and mixed with 2.7 mol of NaBH 4 in 33 l of Me 2 SO. After 10 min of incubation at 30°C, 1 ml of 0.44 M ice-cold HClO 4 was added; the samples were placed on ice for 30 min and then centrifuged for 20 min at 10,000 ϫ g and 4°C. The resulting pellets were washed with 200 l of ice-cold 10 mM HCl and dried under vacuum. For the alkaline hydrolysis, the protein pellets were resuspended in 25 l of 1 M NaOH prepared in [ 18 O]H 2 O and incubated at 37°C. After 15 min, 500 l of 10% ice-cold trichloroacetic acid was added and the precipitated protein processed as above. In both cases, the protein was resuspended in 50 l of 50 mM Tris/HCl, pH 8.5, containing 1% octyl glucoside and 3 g of trypsin, and the mixture was incubated overnight at 37°C. The reaction was stopped by the addition of 2 l of 50% trifluoroacetic acid, and the mixture was analyzed by mass spectrometry. All mass spectral analyses were performed on a Finnigan LCQ ion-trap equipped with an electrospray source. The samples were introduced directly into the source at a flow rate of 2 l/min. The spray was obtained by applying a potential difference of 5 kV and with the help of a sheath gas (N 2 ). The temperature of the heated capillary was 210°C. The LCQ was operated under manual control in the Tune Plus view with default parameters and active Automatic Gain Control.
For SDS-PAGE analysis, 32 P-labeled phosphoenzyme was prepared by incubating 50 g of phosphoserine phosphatase during 10 s at 0°C in a mixture (100 l) containing 10 M [ 32 P]phosphoserine (500,000 cpm), 20 mM serine, 5 mM MgCl 2 , 1 mM dithiothreitol, 50 mM Hepes, pH 7.5, and 0.1 mg/ml albumin. The protein pellet was prepared as explained above and digested at 25°C with endoproteinase Glu-C (3 g) in 25 l of a mixture containing 1% octyl glucoside and 50 mM ammonium acetate, pH 4. After 0, 15, 60, or 120 min, the samples were loaded onto 20% SDS-polyacrylamide gels that were run for 3 h at 0°C and 300 V. The gels were dried and autoradiographed.
Site-directed Mutagenesis-Site-directed mutagenesis was performed by using Pwo DNA polymerase and "back-to-back" mutated primers as described (26). The polymerase chain reaction was directly performed on a pET3a plasmid (27) containing the sequence of human phosphoserine phosphatase (15). The plasmids were re-circularized, amplified in Escherichia coli JM109, and checked by sequencing using T7 Thermosequenase, fluorescent primers, and the LI-COR automated DNA sequencer 4000L.
Expression of the Recombinant Proteins-Bacteria BL21(DE3)pLysS (27) harboring the expression plasmids were grown aerobically in 0.5 liters of M9 medium at 37°C until A 600 reached 0.5-0.6. The culture was then maintained on ice for 60 min before addition of isopropylthiogalactoside to a final concentration of 0.4 mM. Phosphoserine phosphatase was expressed during 13 h at 18°C. Bacterial extracts were prepared as described in (15) and centrifuged for 40 min at 20,000 ϫ g and 4°C. The resulting supernatant (25 ml) was diluted 3-fold with buffer A (25 mM glycine, pH 9, 1 mM dithiothreitol, 1 g/ml leupeptin, 1 g/ml antipain) and applied onto a Q-Sepharose column (1.6 ϫ 10 cm). The column was washed with 100 ml of buffer A, and protein was eluted with a NaCl gradient (0 -400 mM in 400 ml of buffer A). Phosphoserine phosphatase came out at a Na ϩ concentration of approximately 300 mM.
Enzyme and Protein Assays-Except if indicated otherwise, the hydrolytic activity of phosphoserine phosphatase was measured by the release of [ 14 C]serine from [ 14 C]phosphoserine as in (5) but with 1 mM phosphoserine. To study the inhibition by serine, the enzyme was measured by the release of 32 P i (28) from 1 mM [ 32 P]phosphoserine (29). The exchange reaction (incorporation of [ 14 C]serine into phosphoserine) and the phosphoenzyme formation were determined as described (15). Protein was measured according to Bradford (30) with bovine gamma globulin as a standard. The NIH Image program (developed at the National Institutes of Health) was used to analyze SDS-PAGE gels stained with Coomassie Blue. (5) to identify the phosphorylated residue by reduction of the phosphorylated aspartate with tritiated borohydride, followed by digestion of the protein with trypsin and separation of the digested peptides was unsuccessful, presumably because the low degree of phosphorylation of the protein (about 7%) prevented the identification of a specific radioactive peak in the high pressure liquid chromatography elution profile. Knowing that the phosphorylated residue was most likely the first aspartate in the DXDXT motif, we decided to omit separation of peptides by high pressure liquid chromatography and to analyze directly by tandem mass spectrometry the protein digests derived from the preparations of both non-phosphorylated/reduced and phosphorylated/reduced phosphoserine phosphatase. Prediction of the digestion pattern indicated that the positive ion corresponding to the phosphorylatable peptide (DVDSTVIR) would have an m/z of 904 if not reduced and of 890 after reduction of phosphoaspartate to homoserine. An m/z 904 ion was indeed present in the electrospray ionization spectrum of the two digests, as well as a small amount of the m/z 890 ion (not shown). Upon fragmentation, this m/z 904 ion yielded two major ions with m/z 789 and 575 corresponding to fragments y7 (VDSTVIR) and y5 (STVIR) (not shown). We fragmented also the minor m/z 890 ion. In the case of the phosphorylated/reduced enzyme preparation, two major fragments of m/z 789 and 575 were also observed, showing that this peptide is identical to the m/z 904 ion, except for reduction of the first aspartate to homoserine 1 The abbreviations used are: Mes, 4-morpholineethanesulfonic acid; PAGE, polyacrylamide gel electrophoresis.

Identification of the Phosphorylated Aspartate by Reduction with Borohydride-A previous attempt
FIG. 1. Sequences surrounding the three motifs of the dehalogenase superfamily in human phosphoserine phosphatase (PSPase), rabbit sarcoplasmic Ca 2؉ ATPase, and Pseudomonas sp. haloacid dehalogenase (Dehal.). The conserved residues in phosphoserine phosphatases, in P-type ATPases, and in haloacid dehalogenases are indicated in bold. The alignment is based on those of Koonin and Tatusov (4) and Aravind et al. (11), which were obtained with iterative approaches using haloacid dehalogenase as the starting sequence. (Fig. 2B). By contrast, in the fragmentation spectrum of the m/z 890 ion originating from the control enzyme, the peak at m/z 789 was much smaller, and no m/z 575 ion was present ( Fig. 2A), indicating that the reduced peptide was not present in the tryptic/chymotryptic digest.
The approach used above showed that Asp-20 is phosphorylated by its substrate in phosphoserine phosphatase. To ensure that no other aspartate is phosphorylated, we used a different approach, in which the enzyme is labeled with [ 32 P]phosphoserine, denatured, digested with endoproteinase Glu-C, and analyzed by SDS-PAGE at low temperature, in order to preserve the acylphosphate bond (14). Autoradiography of the gel disclosed the presence of only one radiolabeled peptide of Ϸ2 kDa in the samples, indicating that only one aspartate had been phosphorylated (not shown).
Incorporation of 18 18 O results in a 2-dalton increase of the molecular mass that should also allow one to localize the modified residue.
Phosphoserine phosphatase was incubated with (phosphorylated) or without (control) phosphoserine. The protein was denatured in trichloroacetic acid, dried, and resuspended in 1 M NaOH prepared in [ 18 O]H 2 O. After 15 min, the samples were digested with trypsin and chymotrypsin and analyzed by mass spectrometry. As for the experiment with NaBH 4 (see above), the m/z 904 ion was found both for the control and for the partially phosphorylated protein. Its fragmentation yielded two ions (m/z 789 and 575) whose masses correspond to the y7 and y5 fragments of the DVDSTVIR peptide (not shown). In both preparations, fragmentation of the m/z 906 (904 ϩ 2) ion yielded ions of m/z 789, 790, and 791 as well as of m/z 575, 576, and 577 (Fig. 3). However, the ratios of the abundance of the m/z 789/791 and m/z 575/577 were significantly higher in the case of the phosphorylated preparation than in the case of the control, indicating the presence of an excess of 18 O in the aspartate residue that had been removed upon fragmentation. These results confirmed therefore that Asp-20 is the phosphorylated residue and showed that it was possible to measure 18 O incorporation by mass spectrometric analysis of a crude digest.
Lack of Paracatalytic Incorporation of 18 4) was detectable in the ESI spectrum, whereas the peak of 904 was clearly apparent (not shown). Furthermore, fragmentation of ions with m/z 908 did not yield the characteristic fragments of 575 and 789 m/z, whereas fragmentation of the 906 and 904 ions yielded the same results as observed with the control enzyme (see above). It was checked that lyophilization did not inactivate phosphoserine phosphatase.
Site-directed Mutagenesis-To study the role of the residues conserved in ATPases and phosphoserine phosphatase, we have constructed and expressed 10 mutated proteins (Table I). Extracts have been prepared 13 h after the addition of isopropylthiogalactoside, and SDS-PAGE analysis showed that the 10 mutants were expressed as soluble proteins of Ϸ25-kDa subunit (except for D183N, which appeared as a 27-kDa band, but was found by matrix-assisted laser desorption ionization mass spectrometry to have the expected subunit mass).
These proteins were chromatographed on Q-Sepharose at pH 9. Analysis of the fractions by SDS-PAGE and assay of phosphoserine phosphatase activity indicated that this chromatographic step clearly separated the mutants of human phosphoserine phosphatase from bacterial phosphoserine phosphatase (5). In all cases, the recombinant phosphoserine phosphatases were similarly pure, representing in the most purified fractions Ϸ20% of the material stained with Coomassie Blue.
The purified proteins were used to investigate their kinetic properties and their ability to form a phosphoenzyme (Table I). Mutant S109T had the same phosphatase activity as the wild type enzyme, and four other mutants (G178A, D179E, G180A, and D183E) still had between 15 and 78% of this activity. An intermediate value (6%) was found with mutant S109A, whereas the other mutants had only about 1% of the control activity (K158R and D179N) or no detectable activity (K158A and D183N). Significant increases in the K m for phosphoserine were observed with mutants G178A (80-fold) and D179N (10fold). The apparent affinity for Mg 2ϩ was increased for mutants S109A and K158R and decreased in the case of mutants D179E and D179N. Remarkably, the modification of the exchange reaction did not always parallel the change in phosphatase activity. The most striking example was mutant G180A, which had a slightly increased exchange reaction, whereas its phosphatase activity was decreased by 6-fold. In the case of mutant G178A, the decrease in the ratio of the two reactions is partly due to the fact that the phosphoserine concentration used in the exchange reaction (1 mM) was not saturating. DISCUSSION Identification of the Phosphorylation Site and Mechanism of the Phosphoenzyme Hydrolysis-Our finding that the positive ion corresponding to the phosphorylatable peptide (DVDSTVIR) had an m/z of 904 when not reduced and of 890 after reduction of phosphoaspartate to homoserine, as well as the fact that fragmentation of these two ions yielded identical y7 (VDSTVIR) and y5 (STVIR) ions, allows us to conclude that the phosphorylated aspartate in phosphoserine phosphatase is Asp-20.

TABLE I Effect of mutations on the kinetic properties of human recombinant phosphoserine phosphatase
The phosphatase activity was measured through the release of [ 14 C]serine from [ 14 C]phosphoserine. The exchange reaction was determined at 0.5 mM serine and 1 mM phosphoserine. The phosphoenzyme was measured at 1 M phosphoserine; for the calculation of the stoichiometry of incorporation, all proteins were assumed to be Ϸ20% pure. Results shown are representative of at least two independent determinations. Abbreviations used are: WT, wild type; ND, not determined because the activity was too low. distribution of these isotopes is random in the control peptide but not in the peptide where 18 O has been incorporated during alkaline hydrolysis of the phosphoenzyme. Thus, elimination of the first aspartate removes a larger proportion of the ϩ2 atomic mass unit excess, explaining the higher proportion of the m/z 575 and 789 ions. This Asp-20 residue of phosphoserine phosphatase is in equivalent position to the phosphorylated aspartate in eukaryotic phosphomannomutase and ATPases and to the esterified residue in haloacid dehalogenases (5,11). That there is no second phosphorylatable site is indicated by the fact that only one radiolabeled peptide was observed when the enzyme phosphorylated with [ 32 P]phosphoserine was digested with endoproteinase Glu-C. This result agrees also with the finding that of the four conserved aspartate residues in phosphoserine phosphatase, three can be replaced by glutamate with conservation of substantial (Ն50%) activity, whereas the D20E mutation abolishes the activity (5).
The fact that no 18 O was incorporated from [ 18 O]H 2 O during hydrolysis of phosphoserine by phosphoserine phosphatase indicates that the phosphoenzyme is attacked on the phosphorus, not on C-␥ of the aspartylphosphate. The same conclusion, based on the retention of configuration of the ␥-phosphate of ATP in inorganic phosphate, was previously reached for AT-Pases (24). This contrasts with the situation observed with haloacid dehalogenases, in which the covalent ester intermediate is hydrolyzed by attack of C-␥ of the modified aspartate (17). This indicates that the attacking water molecule must have a different localization in phosphoserine phosphatase and ATPases as compared with haloacid dehalogenases.
Interpretation of the Effect of Mutations-As expected, almost all mutations affecting a well conserved amino acid decreased the activity of the enzyme, although to different degrees. The only exception was mutant S109T, which had essentially the same kinetic properties as the control enzyme. The fact that the mutated proteins had conserved the same behavior upon chromatography indicated that the loss of activity was not due to a gross folding problem (31). Some of the mutations affected the exchange reaction slightly more than the phosphatase activity (D179N and D183E) indicating that they acted primarily by inhibiting the formation of the phos-phoenzyme. This was presumably also the case for mutations that abolished the activity (K158A and D183N) since in these cases no phosphoenzyme could be detected. Three mutants (S109A, K158R, and G180A) displayed lowered phosphatase activities (1-15% of the wild type) but lesser decreased exchange reactions, indicating that the hydrolysis of the phosphoenzyme was also affected.
The 80-fold increase in the K m for phosphoserine observed when Gly-178 was mutated to alanine suggests that this residue participates in substrate binding. Such is also the case for Asp-179 since mutations of this residue to asparagine (though not to glutamate) resulted in a marked increase in the K m . Both D179N and D179E mutations decreased the apparent affinity for Mg 2ϩ suggesting that Asp-179 also participates in binding of this indispensable cation, as recently proposed (32) on the basis of a structural homology between haloacid dehalogenase and CheY, a Mg 2ϩ -dependent enzyme that also forms a phosphoaspartate. Considering that the assays were performed in the presence of 1 mM Mg 2ϩ , the low activities of mutant D179N and D179E are partly explained by the use of a subsaturating concentration of the divalent cation. The effect of the G180A mutation to decrease the affinity for serine suggests that the Gly-180 residue participates in the binding of this product of the reaction.
Comparison with ATPases and Dehalogenases- Table II compares the effect of mutations to those of residues in equivalent position (Fig. 1) in Ca 2ϩ ATPase (33)(34)(35) and Pseudomonas haloacid dehalogenase (19). Site-directed mutagenesis of the first motif had indicated previously that Asp-20 could not be substituted by glutamate or asparagine without complete loss of activity, whereas the replacement of Asp-22 by glutamate (but not by asparagine) allowed the retention of 50% of the activity (5). As mentioned above, this agrees with the critical role played by Asp-20, which forms the phosphoenzyme intermediate. Interestingly, mutations of the residues in equivalent positions in the dehalogenases (Asp-10 and Tyr-12) and in Ca 2ϩ ATPase (Asp-351 and Thr-353) gave similar results, with a 100% loss of activity in the D10E and D351E mutants and 60 -80% decreases in the case of the Y12F and T353S mutants, respectively (19,34,35).
With respect to the second motif, Ser-109 is conserved in the superfamily as a serine or a threonine, and it is therefore not surprising that the conservative change S109T does not result in a loss of activity. The presence of the hydroxyl group on this residue seems particularly important since the S109A mutation resulted in a Ϸ17-fold decrease in activity, an effect that is much more marked than the Ϸ3-fold decrease observed for the equivalent mutation in haloacid dehalogenases. In the latter enzyme, Ser-118 was shown to form a hydrogen bond with the substrate carboxylate (22). By analogy, Ser-109 could bind the phosphate group of the substrate in phosphoserine phosphatase, as suggested by Ridder and Dijskstra (32). However, this interpretation does not agree with the absence of effect of the mutations affecting this residue on the K m of the enzyme for phosphoserine (Table I).
The almost complete loss of activity of phosphoserine phosphatase observed when Lys-158, the first conserved residue in the third motif, is replaced by arginine agrees with similar observations made with haloacid dehalogenase and with the fact that a lysine residue is almost always observed in this position. The only exception in the alignment presented by Aravind et al. (11) is an open reading frame homologous to sucrose-phosphate synthase in the Synechocystis genome and which has not been shown to encode an active protein. In haloacid dehalogenase, Lys-151 binds to the conserved Asp-180 in the free enzyme and to the carboxylate group of Asp-10 in the enzyme-substrate complex (22). It is likely that it plays a similar role by interacting with Asp-20 and Asp-183 in phosphoserine phosphatase and with the equivalent residues in ATPases. To the best of our knowledge, no report has, however, appeared on the effect of such a mutation on this last class of enzymes.
The intermediate residues of the third motif (Gly-178, Asp-179, and Gly-180 in human phosphoserine phosphatase) are strictly conserved in phosphoserine phosphatases and ATPases and are replaced in dehalogenases by a different stretch of conserved residues (Ser-175, Ser-176, and Asn-177). In all cases, mutation analysis (Table II) indicates that these residues are important for catalysis although they must obviously play different roles in ATPases and phosphoserine phosphatases on the one hand and in dehalogenases on the other hand. We speculated above that Gly-178 and Asp-179 participate in substrate and Mg 2ϩ binding in phosphoserine phosphatase. By analogy, Gly-803 and Asp-804 could play a similar role in ATPases; unfortunately, the K m and sensitivity for Mg 2ϩ have not been determined in the relevant mutants of the Ca 2ϩ ATPase. Since haloacid dehalogenases are not dependent on Mg 2ϩ for their activity, the fact that Asp-179 in phosphoserine phosphatase is replaced by serine in haloacid dehalogenases is consistent with the hypothesis that Asp-179 binds Mg 2ϩ .
Replacement of the highly conserved Asp-183 at the end of the third motif by residues other than glutamate results in a near complete loss of activity in phosphoserine phosphatase, haloacid dehalogenase, and Ca 2ϩ ATPase and even its replacement by glutamate results in important losses of activity, although more in the case of the dehalogenase and the ATPase than in the case of phosphoserine phosphatase. In the case of haloacid dehalogenase, Asp-180 was proposed to play a role in activating a water molecule that would be used in hydrolysis of the covalent intermediate (22). However, one should remember that the atom attacked by the activated water molecule is phosphorus and not C-␥ of aspartate in phosphoserine phosphatase and ATPases (see above). Furthermore, the fact that mutation of Asp-183 in phosphoserine phosphatase or the equivalent residue in Ca 2ϩ ATPase by asparagine abolishes the formation of the phosphoenzyme indicates that this aspartate plays a major role in this process (33).
In conclusion, the results of these and of our previous mutagenesis experiments indicate that the three conserved motifs play an important role in phosphoserine phosphatase, as it had been previously shown for haloacid dehalogenase and, to some extent also, for ATPases. This is in keeping with the fact that these residues are known to form the catalytic site in the case of haloacid dehalogenase, the only enzyme of the superfamily for which the detailed three-dimensional structure is available (20 -22). The stringent requirement for these residues is indicated by the fact that in most cases even what can be considered as a conservative change caused a reduction in activity. On the basis of the finding of low but significant homologies between dehalogenases and ATPases, it has been previously proposed that the active site of ATPases can be modeled on the dehalogenase fold (11). Our results indicate that phosphoserine phosphatase may be a better model than dehalogenase both with respect to the type of reaction catalyzed and to the residues involved in catalysis.