Isolation of a human liver fructose-1,6-bisphosphatase cDNA and expression of the protein in Escherichia coli. Role of ASP-118 and ASP-121 in catalysis.

A cDNA encoding human liver fructose-1,6-bisphosphatase was isolated from a lambda gt11 library by screening with a rat liver fructose-1,6-bisphosphatase cDNA. The cDNA (1421 base pairs) contains an open reading frame encoding 337 amino acids, corresponding to a protein with an estimated molecular weight of 36,697. Its primary sequence is highly homologous to that of the pig kidney and rat liver enzymes. The human liver cDNA was used to construct a T7 RNA polymerase-transcribed expression vector, and the enzyme was expressed in Escherichia coli BL21 (DE3). Approximately 50% of the expressed human fructose-1,6-bisphosphatase was soluble and enzymatically active, and the enzyme was purified to homogeneity by heat treatment, ammonium sulfate fractionation, and substrate/AMP elution from carboxymethyl-Sephadex. Expressed human liver fructose-1,6-bisphosphatase had a specific activity (9.8 mumol/min/mg of protein) that was half that of the rat liver enzyme, but had an identical Km for substrate. However, the human enzyme was more sensitive to inhibition by fructose-2,6-bisphosphate (Ki = 0.3 microM) and AMP (Ki = 12 microM) than the rat liver form (fructose 2,6-P2, Ki = 4 microM; AMP, Ki = 40 microM). Crystallographic analyses have suggested that Asp-118 and Asp-121 are catalytic residues located in a negatively charged pocket that binds divalent metal cations. These residues were mutated to alanine, and the E. coli-expressed mutant enzymes were purified to homogeneity. The Asp-118-->Ala and Asp-121-->Ala mutants had 1/5000 and 1/20,000 lower Kcat values than the wild-type enzyme, respectively, consistent with their critical role in fructose-1,6-bisphosphatase catalysis.

Isolation of a Human Liver Fructose-1,6-bisphosphatase cDNA and Expression of the Protein in Escherichia coli ROLE OF ASP-118 AND ASP-121 IN CATALYSIS* (Received for publication, November 30,1992, andin revised form, January 25, 1993) M. Raafat El-Maghrabi, Madhavi Gidh-Jain, LeeAnn R. Austin A cDNA encoding human liver fructose-1,6-bisphosphatase was isolated from a Xgtl 1 library by screening with a rat liver fructose-l,6-bisphosphatase cDNA. The cDNA (1421 base pairs) contains an open reading frame encoding 337 amino acids, corresponding to a protein with an estimated molecular weight of 36,697. Its primary sequence is highly homologous to that of the pig kidney and rat liver enzymes. The human liver cDNA was used to construct a T7 RNA polymerasetranscribed expression vector, and the enzyme was expressed in Escherichia coli BL2 1 (DE3). Approximately 50% of the expressed human fructose-1,6-bisphosphatase was soluble and enzymatically active, and the enzyme was purified to homogeneity by heat treatment, ammonium sulfate fractionation, and substrate/ AMP elution from carboxymethyl-Sephadex. Expressed human liver fructose-1,6-bisphosphatase had a specific activity (9.8 pmollminlmg of protein) that was half that of the rat liver enzyme, but had an identical K,,, for substrate. However, the human enzyme was more sensitive to inhibition by fructose-2,6bisphosphate (Ki = 0.3 p~) and AMP (Ki = 1 2 p~) than the rat liver form (fructose 2,6-P2, Ki = 4 KM; AMP, Ki = 40 p~) .
Crystallographic analyses have suggested that Asp-1 18 and Asp-121 are catalytic residues located in a negatively charged pocket that binds divalent metal cations. These residues were mutated to alanine, and the E. coli-expressed mutant enzymes were purified to homogeneity. The Asp-1 18 + Ala and Asp-121 4 Ala mutants had 1/5000 and 1/20,000 lower Kc, values than the wild-type enzyme, respectively, consistent with their critical role in fructose-1,6-bisphosphatase catalysis.
The gluconeogenic enzyme fructose-1,6-bisphosphatase, Fru-1,6-P2ase' (~-fructose-1,6-bisphosphate l-phosphohydrolase, EC 3.1.3.11), is regulated at two levels. Acute modulation * This work was supported by an endowment from Gensia Pharmaceuticals and by National Institutes of Health Grant DK 38354 and National Science Foundation Grant DMB 8608989. 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.
The nucleotide sequence(s) reported in thispaper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) L10320.
Fru-1,6-Pzase from pig kidney (12), sheep liver (13), and rat liver (11) have been purified, and their complete amino acid sequences have been determined. Their primary sequence is highly homologous (approximately go%), and there is complete identity of putative active and allosteric site residues as well as in the flanking sequences (14). This homology extends to enzyme forms from plants (15,16), yeast (17), and Escherichia coli (18), which suggests that Fru-1,6-P2ase has undergone minimal changes throughout evolution, even though regulation of the enzyme's activity differs in different cell types.
The human enzyme has, for obvious reasons, been studied the least (19,20). Since increased rates of gluconeogenesis contribute significantly to hyperglycemia in type I and type I1 diabetes (21), a better understanding of the residues that contact substrate and inhibitors such as Fru-2,6-P2 and AMP may provide a means for targeting down-regulation of human liver Fru-1,B-Pzase for drug therapy. For example, mutation of the conserved active site residue Lys-274 in rat liver Fru-1,6-P2ase revealed that it is an important substrate/Fru-2,6-P2 binding residue but is not involved in catalysis (22). Here, as a first step in elucidating the mechanism of catalysis of Fru-1,6-Pzase, we describe the isolation of the cDNA for the human liver enzyme, the construction of a plasmid for its expression in E. coli, and purification and characterization of the recombinant enzyme. In addition, the catalytic role of 2 aspartate residues (Asp-118 and Asp-121), conserved in all known Fru-1,6-P2ase sequences and predicted to participate in the binding of divalent metal cations (14), was studied by mutating each residue to alanine and by analyzing the kinetic properties of the mutants.

EXPERIMENTAL PROCEDURES
Materials-The pET3a expression vector containing the 610 promoter for T7 RNA polymerase and the bacterial host cell strains BL21(DE3) f pLysS were kind gifts of Dr. William Studier (Brookhaven National Laboratories). Restriction enzymes were from New England Biolabs, and T4 DNA ligase and polynucleotide kinase were from Boehringer Mannheim as were yeast glucose-6-phosphate dehydrogenase and phosphoglucose isomerase. Isopropyl-P-D-thiogalactopyranoside was from New Jersey Glove and ~-[~'S]methionine (1061 Ci/mmol) was from Du Pont-New England Nuclear. Oligodeoxynucleotides were synthesized on an Applied Biosystems Model 380A synthesizer and purified on OPCTM cartridges. The human liver cDNA/Xgtll library was from Clontech, and the plasmid vector Bluescript was from Stratagene. Hot Tub DNA polymerase and lox buffer were from Amersham, and the thermal cycler was a Model 50 Tempcycler from Coy Laboratory Products Inc.
Isolation of a Human Liver cDNA and Construction of the Expression Plasmid-Four independent cDNA clones of human liver Fru-1,6-Pzase were isolated by screening a Xgtll library with a randomprime 32P-labeled cDNA of the rat liver enzyme (11). The cDNA was ligated into the EcoRI site of a Bluescript plasmid vector to give pHLFBP-BS. Since there were no convenient restriction sites with which to subclone the cDNA into the pET3a expression vector, an oligonucleotide-directed polymerase chain reaction (PCR) was used to introduce an NdeI recognition site at the protein initiation codon: 1 pg of oligonucleotide primer, 5"GCATGGCTGCATATGGCT-GACCAGG-3' encoding the NdeI site, the initiation Met and flanking sequences, and 1 pg of the M13 reverse primer were denatured at 100 "C for 5 min with 100 ng of the pHLFBP-BS. The PCR was started by the addition of a 0.2 mM concentration of each deoxynucleotide triphosphate in I X Amersham polymerase buffer and 2 units of Hot Tub polymerase. The thermal cycler was set for 25 cycles, each cycle consisting of a rapid ramp to 94 "C, a 30-s soak, a rapid ramp to 50 "C, a 30-s soak, a rapid ramp to 72 "C, and a 2-min soak.
The PCR product was ethanol-precipitated, dissolved in restriction enzyme buffer, and digested with NdeI and EcoRI restriction enzymes, and the NdeI-EcoRI fragment was ligated into an NdeIIEcoRIdigested pET3a vector to give pHLFBP-ET1. A second expression vector, lacking the 3"untranslated region, was constructed by substituting the primer, 5'-GCAGGGCAGGAATTCACTGGGC-3', for the M13 reverse primer in the PCR. This primer introduces an EcoRI recognition site immediately following the termination codon, and the resulting PCR product was also ligated into the pET3a vector to give pHLFBP-ET2. Both constructs were verified by double-stranded DNA sequencing of both strands by dideoxy-chain termination as previously described (8).
Expression of Fru-l,6-Pzase-Competent E. coli strains BL21(DE3) f pLysS were transformed with pHLFBP-ET1 and -2 and grown in LB medium at 37 "C containing 100 pg/ml ampicillin, and, in the case of the pLysS-containing strain, chloramphenicol (30 pg/ml), as described for the expression of rat liver Fru-1,6-Pzase (25). Typically, cultures were grown at 37 "C until an Am nm of -1, the inductions were started by the addition of solid IPTG to a final concentration of 0.4 mM, and the incubations were continued for 6 h. Aliquots of the cells were removed every hour during the induction period, for analysis of total Fru-1,6-Pzase accumulation by SDSpolyacrylamide gel electrophoresis, and for quantitation of enzyme activity.
Isolation of Fru-l,6-Pzase-At the end of the induction, the cells were harvested by centrifugation, suspended in 1/20 of the volume of lysing buffer that contained 20 mM KPi, pH 7.5, 5 mM EDTA, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, leupeptin (2.5 pg/ml), and lysozyme (1 mg/ml), and lysed by three cycles of freezing in liquid NZ and thawing. Bacterial DNA was then digested by treatment with DNase I (2 mg/liter of culture) and 10 mM MgSO, for 1 h at 4 "C. The lysate was clarified by centrifugation at 30,000 x g for 30 min, and the supernatant was adjusted to 10 mM EDTA and heated at 58 "C for 3 min. The denatured protein was removed by another centrifugation, and Fru-l,6-Pzase was precipitated between 45 and 75% saturation of (NH4)&3O,. The precipitate was collected by centrifugation and dissolved in a minimum volume of buffer (5 mM sodium malonate, pH 7.2, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 0.1 mM EDTA, and leupeptin (2.5 pg/ml). Residual (NH4)zS04 was removed by gel filtration on a Sephadex G-50 column, and the desalted sample was applied to a carboxymethyl-Sephadex C-50 column equilibrated with the same buffer. The column was washed until the AZm nm of the eluate was 50.05, and the enzyme was eluted with 2 mM Fru-1,6-P~, 1 mM AMP. The enzyme was then precipitated at 75% saturation of (NH4)zSO4, and the precipitate was harvested, dissolved in buffer (20 mM K P , pH 7.5, 20% glycerol, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 0.1 mM EDTA, and leupeptin (2.5 pg/ml), dialyzed extensively against the same buffer, and stored at -70 "C.
Assay of Fru-l,6-Pzase Actiuity-Fru-1,6-Pzase activity was measured spectrophotometrically in a reaction that coupled the production of Fru-6-P to the reduction of NADP+, catalyzed by phosphoglucose isomerase and glucose-6-phosphate dehydrogenase (25). Cells from induction time courses (1 ml) were harvested by centrifugation, suspended in 100 pl of lysing buffer, and lysed as described above, and, after DNase treatment, the clear supernatant was assayed for enzyme activity.
Other Methods-Total protein was determined by the method of Lowry et al. (26). Total Fru-l,6-Pzase induction was monitored by suspending the cell pellets (100 pl) from induction time courses in SDS sample loading buffer and subjecting them to electrophoresis in SDS-polyacrylamide gels according to Laemmli (27). Protein synthetic rates were monitored by preincubating 100 p1 of the cells with 1.5 pCi of [35S]Met at 37 "C for 3 min. The cells were then harvested and dissolved in SDS-PAGE sample buffer.

RESULTS AND DISCUSSION
Isolation of Human Liver Fru-l,6-Pzase cDNA-Four independent clones were isolated from a human liver cDNA library by screening with the cDNA to the rat liver Fru-1,6-Ppase. Three of the clones were too short to contain the entire protein coding sequence. The fourth and longest clone (1421 base pairs) contained a 385-amino acid open reading frame and 260 base pairs of the 3"untranslated region including a polyadenylation signal sequence and polyadenylation tail (Fig.  1). The deduced amino acid sequence, starting after the first Met, contains 336 residues and is highly homologous (about 90%) with those of the pig kidney (12), rat liver (11), and that of chicken liver Fru-1,6-P2ase, deduced from a recently isolated cDNA2 (Fig. 1). The conserved residues include those predicted to interact with the 6-phosphate group of Fru-1,6-Pp , the 2phosphate group of Fru-2,6-Pz (Gly-122, Ser-124, Lys-274, Asn-1251, the furanose ring (Lys-274, Met-248), those predicted to bind metal ions (Glu-97, Glu-98, and Asp-118), and those predicted to interact with the allosteric inhibitor, AMP (Leu-30, Ala-24, Met-177, Thr-31, Tyr-113, and Arg-140) (14). These findings attest to the identity of the isolated human cDNA, as well as provide additional support for a very high conservation in the primary sequence of all known eukaryotic Fru-l,6-Pzases. Based on the sequence, the estimated subunit molecular weight (36,697) of human liver Fru-1,6-Pzase is very similar to those of the pig kidney and sheep liver enzymes (Fig. 1). Although the rat liver Fru-1,6-Pzase sequence ends in a unique carboxyl-terminal 25-amino acid extension, which yields a subunit of higher molecular weight (-40,000) (8), the remainder of the sequence is homologous to other eukaryotic sequences.
The nucleotide sequence of the human liver cDNA, with the exception of a single conservative mutation (Ala-216 for Gly-216), two other mutations in the noncoding regions (an additional T a t position 1188 and a G for A substitution a t M. R. El-Maghrabi and S. J. Pilkis, manuscript in preparation.    position 1109), and the presence of a polyadenylation signal sequence and tail is otherwise the same as a cDNA cloned from human leukemia cells (lo), although no Fru-l,6-P2ase activity could be detected in these cells. This is consistent with the presence of a single Fru-1,B-P2ase gene in the human genome (8), but does not completely rule out the presence of another gene(s) encoding Fru-1,6-P2ase in tissues such as skeletal muscle and brain (11,12).
Expression of Human Liver Fru-1,6-P2ase in E. coli-The human liver Fru-l,6-P2ase cDNA was subcloned into a T7 RNA polymerase-transcribed expression vector, pET3a (23, 28), by mutation of the translation initiation region to create a unique NdeI recognition site, as described under "Experimental Procedures." Fig. 2 shows a Coomassie Blue-stained polyacrylamide gel of the time course of expression of total (Fig. 2, lanes 2-8) and soluble (Fig. 2, lanes 9-15) protein in E. coli BL21(DE3) at 37 "C in LB medium. There was no expression of the human form in the absence of IPTG. After the addition of IPTG, a band corresponding to the predicted size of the human Fru-l,6-P2ase subunit accumulated, reaching a maximum level after 4-6 h (Fig. 2, lanes 6-8). Analysis of protein synthesis by [35S]methionine incorporation showed that 70-80% of newly synthesized protein during the first 2 h was Fru-1,6-P2ase (data not shown). Approximately half of this protein was in the soluble fraction (Fig. 2, lanes [13][14][15] and was coincident with the appearance of Fru-l,6-Pzase activity (Fig. 3). The induction of Fru-1,6-P2ase activity also reached a maximal value after 6 h, remaining unchanged for up to 22 h (Fig. 3). Based on the amount of enzyme in the soluble fraction (Fig. 2) and the specific activity of the purified rat liver enzyme (20 units/mg), the yield of the expressed human enzyme was calculated to be approximately 20 mg/ liter of cells. A second construct, lacking the 3'-untranslated region and polyadenylation tail, showed essentially identical induction profiles, indicating that this region of the gene had no effect on Fru-1,6-P2ase expression in E. coli (data not shown). On 10 11 12 13 14 15 the other hand, expression of human Fru-l,6-P2ase in a BL21(DE3) strain carrying the pLysS plasmid, whose inclusion has been reported to result in slower rates of T7 RNA polymerase activity (28), yielded consistently lower total, as well as enzymatically active, Fru-l,6-P2ase protein (data not shown). These findings are similar to those reported for the expression of the rat liver enzyme in E. coli (25). Substituting an even richer medium, 2 X YT containing 0.4% glycerol, for the LB medium doubled the yield of Fru-1,6-Pzase and was, therefore, used for the large-scale purification of the enzyme.
Purification of Recombinant Human Liver Fru-1,6-Pme-The protocol used to purify human liver Fru-l,6-P2ase from E. coli extracts was a modification of that used to purify the rat liver enzyme (25), which involved heat treatment, ammonium sulfate fractionation, and a substrate/effector elution of the homogeneous enzyme from a cation exchange resin (see "Experimental Procedures"). Human liver Fru-l,6-P2ase was stable to heat treatment, but precipitated a t a higher saturation of (NH4)2S04 (50 to 70%) than that which precipitated the rat liver enzyme (25). It was also necessary to raise the pH of the cation exchange chromatographic step from 5.8 to 7.2, in order to bind and subsequently to elute the human liver Fru-l,6-P2ase. This may be a consequence of a higher calculated PI value (6.96) for the human form, in comparison to the rat form (PI 5.96), due to the presence of 2 additional basic residues and 5 fewer acidic residues in the human liver enzyme (Fig. 1). Fig. 4 shows a Coomassie Blue-stained SDS-polyacrylamide gel of human liver Fru-1,B-P2ase of fractions at various stages of purification. The final preparation appeared homogeneous (Fig. 4, lane 5 ) , migrating as a 37,000-dalton peptide, slightly faster than the rat liver enzyme subunit (Mr = 40,000, Fig. 4, lane 6). Table I  was 32 mg of purified enzyme/4 liters of cells ( Table I).
Characterization of Expressed Human Liver Fru-1,G-Pzase-In order to confirm that the purified protein was Fru-

TABLE I Purification of E. coli-expressed human liver Fru-l,6-Pzase
The human liver Fru-1,6-P2ase was purified after extraction from 4 liters of E. coli culture by sequential chromatography on Sephadex G-50 and CM-Sephadex, and the activity was monitored spectrophotometrically as described under "Experimental Procedures." Units represent pmollmin at 30 "C.

Fraction
Volume

TABLE I1
Comparison of the kinetic properties of purified human liver Fru-I,6-P2ase with those of rat liver Fru-I,6-Pzase Inhibition of Fru-l,6-P2ase by Fru-2,6-P* and AMP were done at 1 pM Fru-1,6-P2. The enzyme was assayed spectrophotometrically as described under "Experimental Procedures." The results represent the average f S.E. of the mean of three to five determinations. 1,6-Pzase, it was subjected to 10 cycles of automated sequencing by Edman degradation, and the following amino acid sequence was obtained Ala-Asp-Gln-Ala-Pro-Phe-Asp-Thr-Asp-Val-. This sequence is consistent with that deduced from the cDNA and predicted from the expression construct. The kinetic properties of the expressed human liver enzyme are summarized in Table I1 and compared to those of rat liver Fru-l,6-Pzase. The specific activity of the homogeneous enzyme was 9.8 units/mg of protein or approximately half that of the enzyme purified from rat liver (18 units/mg)(25), although both had identical K,,, values for Fru-l,6-Pz (2 p M ) .
The human liver enzyme appeared to be more sensitive to inhibition by Fru-2,6-Pz and AMP, having Ki values of 0.3 p~ and 12 p~, respectively. Under the same assay conditions, rat liver Fru-1,6-P2ase had a Ki for Fru-2,6-P2 of 3 PM and a Ki for AMP of 40 PM (22). The explanation for the lower specific activity and greater sensitivity to Fru-2,6-P2 and AMP inhibition of the human enzyme is not clear. All residues that are postulated to interact with Fru-2,6-P2 and AMP (13) are conserved in both forms. For example, Lys-274, implicated in binding the 2-phospho group of Fru-2,6-P2 to rat liver Frul, 6-P2ase (22), is conserved in the human sequence. One possible explanation for the differences in specific activity and sensitivity between human and rat liver enzymes is the presence of the 25-amino acid carboxyl-terminal extension in rat liver Fru-l,6-P2ase (Fig. l ) , which may have a positive effect on catalysis, or decrease binding of Fru-2,6-P2, but not Fru l,6-P2, to the active site. Based on this argument, the specific activity and Fru-2,6-P2 inhibition of the pig kidney and sheep liver enzymes should be similar to that of the human liver Fru-l,6-P2ase and to a form of the rat liver enzyme from which the carboxyl-terminal extension has been deleted. Because of differences in isolation procedures and assay conditions, a comparison of the published kinetic parameters of pig, sheep, and rat enzymes is not possible. Analysis of the effect of deleting the carboxyl-terminal extension of the rat liver enzyme, however, is currently under investigation.

Mutation of Human Liver Fru-l,6-P2ase Asp-118 + Ala and Asp-121 + Ala and Expression and Characterization of the Mutant Enzyme
Forms-The initial crystallographic analysis of pig kidney Fru-l,6-P2ase suggested that a group of residues including Glu-97, Asp-118, and Asp-121 form a negatively charged pocket in the active site of the enzyme, a t which Mg2+ or Mn2+ is predicted to bind (14). Zhang et al.
(29) recently reported refined structures of pig kidney Frul,6-P2ase complexed with Fru-l,6-P2 and of various ternary complexes of Fru-1,6-P2ase with divalent metal ions and nonhydrolyzable substrate analogs. Two metal ion binding sites were located at the enzyme active site complexed with substrate analog. Metal site 1 is coordinated by the carboxylate groups of Glu-97, Asp-118, Glu-280, and the 1-phosphate group of the substrate analog, while metal site 2 is coordinated by the carboxylate groups of Glu-97, Asp-118, the phosphate group of the substrate analog, and the carboxyl oxygen of Leu-120. Based on these findings, Zhang et al. (29) postulate a reaction mechanism in which a metal-activated water molecule attacks the l-phosphate of the a-analog of the substrate, while a proton is transferred from the carboxyl group of Asp-121 to the substrate ester oxygen either directly or with the help of the C2 hydroxyl group of the substrate. This hypothesis predicts that mutation of either Asp-118 or Asp-121 would greatly decrease Fru-l,6-P2ase catalysis.
In order to determine whether these residues play a role in catalysis, they were mutated to alanine, the mutant enzyme forms expressed as described under "Experimental Procedures," and purified as described above for the wild-type enzyme. Fig. 5 shows a polyacrylamide gel of the two purified mutant enzymes, which were homogeneous and migrated as 37,000-dalton peptides. Both the Asp-118 + Ala and the Asp-121 + Ala mutants bound to CM-Sephadex and were eluted by substrate in exactly the same manner as the wild-type enzyme, suggesting that it is unlikely that either mutation altered the gross secondary structure of the enzyme. In addition, the circular dichroism spectra of the mutant enzymes were indistinguishable from the wild-type enzyme (data not shown). As shown in Table 111, the Asp-118 + Ala mutant had a specific activity of 2 milliunits/mg, while that of the Asp-121 + Ala mutant was 0.1 milliunit/mg, or about 1/5000 and 1/20000, respectively, of the activity of the wild-type The Asp-118 + Ala and Asp-121 + Ala mutants of human liver Fru-1,6-Ppase expression plasmids were prepared, and the enzyme forms were expressed and purified as described under "Experimental Procedures." Lane 1, 2 pg of Asp-118 + Ala Fru-l,6-P2ase; lane 2, 2 pg of Asp-121 + Ala Fru-l,6-Pzase; lane 3, marker standards with sizes shown in kilodaltons.

Kinetic properties of the wild-type and Asp-1 18 +Ala and Asp-121 +Ala mutants
The wild-type and mutant enzymes were assayed spectrophotometrically as described under "Experimental Procedures." In order to detect reaction rates it was necessary to use 50 to 100 times more mutant Drotein than wild-twe enzyme. for Fru-l,6-P2 was observed, which suggests that these residues are not involved in substrate binding per se, but rather mediate catalysis. The specific activities of the mutants did not increase when they were assayed in the presence of 10-to 50-fold higher concentrations of Mg+ than that routinely used in the assay (2 mM). These results suggest that Asp-118 may play a more important role in catalysis than in M e binding, and that other residues such as Glu-97 and Glu-280 are more important for the latter. The results are consistent, however, with the proposed essential role of Asp-121 in catalysis.
The mechanism postulated by Zhang et al. (29) implicates Asp-118 in metal ion binding and activation of water. In the crystal structure, only one Mg2+ ion is found near the metal site 1 of the active site in the complex of the a-analog of substrate and enzyme even with concentrations of 50 mM Mg2+ in the crystallization buffer. It has been postulated that metal ions act to position the 1-phosphate group of the substrate in the correct conformation so that the ester oxygen can receive hydrogen bonds from the C-2 hydroxyl group of the substrate, from the main chain NH group of Gly-122, and perhaps from the carboxyl group of . Mutation of Asp-118 to alanine probably results in looser metal binding and decreased activation of water. However, some water activation still occurs in the Asp-118 + Ala mutant, judging from a reduction in catalytic rate of only 3 orders of magnitude. Other residues (Glu-97, Glu-280) can coordinate metal ion, and, therefore, Mg2+ binding and water activation would not be expected to be completely abolished by mutation of Asp-118. The greater decrease in catalytic rate of the Asp-121 -Ala mutant of more than 4 orders of magnitude is consistent with the loss of the proton donor function of the carboxyl group of Asp-121 and, therefore, supports its postulated role in catalysis. However, in the a-substrate analog complex with Mg"', the side chain carboxyl group of Asp-121 is also coordinated to M$+ (29), which suggests that part of the reduction in the catalytic rate of the Asp-121 + Ala mutant may be the result of decreased metal binding as well. Under the assay conditions employed in this study, i.e. pH 7.4 at 2 mM M e and even at 20 mM Mg"', only 1 M$+ ion is present in the active site, and it is unclear whether a second Mg"' ion is required to activate the enzyme and/or whether M$+ ion may activate the enzyme by a mechanism different from other divalent cations, i.e. Mn2+ and Zn2+ (29). Additional characterization of the metal ion binding properties of these mutants and their crystallization are in progress. Crystals of the recombinant wild-type human liver Fru-1,6-P2ase~ave already been obtained, and the structure solved a t 2.1 A r e s~l u t i o n .~ Examination of this structure revealed that the same active site constellation of residues in the pig kidney enzyme is present in the human liver enzyme, including the negatively charged pocket containing Glu-97, Glu-98, Asp-118, and Asp-121. Therefore, conclusions with regard to reaction mechanism based on the pig kidney x-ray structure can be confidently applied to the human liver enzyme. Also in progress are sitedirected mutagenesis experiments of Glu-97 and Glu-98, additional residues predicted to be important in catalysis (29). The results of this report are consistent with a Fru-1,6-Pzase reaction mechanism involving metal binding and water activation and proton donation to the oxygen ester bond which weakens that bond. There is no evidence for a covalent intermediate in Fru-1,6-Pzase catalysis, and the reaction has been shown to proceed via inversion of stereochemical configuration of the transferred phosphate group (30). The catalytic mechanism of Fru-1,6-P2ase is, therefore, different from the catalytic mechanisms of other phosphatases including the alkaline phosphatase/cofactor-independent phosphoglycerate mutase enzyme family (31, 32) and the Fru-2,6-P2ase/cofactor-dependent phosphoglycerate mutase/acid phosphatase family (33). Catalysis in the former family utilizes two zinc atoms and a phosphoenzyme intermediate involving a serine residue and results in retention of the configuration around the transferred phosphate. Thus, in both the Fru-1,6-P~ase and alkaline phosphatase reaction mechanisms, the nucleophiles are metal-bound water molecules or hydroxide ions. Interestingly, catalysis of the Fru-2,6-Pzase/cofactor-dependent phosphoglycerate mutase/acid phosphatase enzyme family also involves a phosphoenzyme intermediate and retention of configuration of the transferred phosphate group, but, in this case, a histidine is utilized in phospho group transfer and metal ions play no role in catalysis (33).
In summary, the E. coli expression system and the rapid Y. Zhang, J.-Y. Liang, S. Huang, H. Ke, M. R. El-Maghrabi, S. J. Pilkis, and W. N. Lipscomb, manuscript in preparation. purification procedure described in this report provide a simple, efficient way to produce high yields of human liver Fru-1,6-Pzase and of mutant forms that allow for crystallization and structure/function studies of this key regulatory gluconeogenic enzyme. The availability of the human liver Fru-1,6-Pzase cDNA will also allow the isolation of the human gene, and thus will be invaluable in determining the molecular basis for several inherited Fru-1,6-P2ase deficiencies reported in human patients (34).