Major Changes in the Kinetic Mechanism of AMP Inhibition and AMP Cooperativity Attend the Mutation of Arg49 in Fructose-1,6-bisphosphatase*

The significance of subunit interface residues Arg49 and Lys50 in the function of porcine liver fructose-1,6-bisphosphatase was explored by site-directed mutagenesis, initial rate kinetics, and circular dichroism spectroscopy. The Lys50 → Met mutant had kinetic properties similar to the wild-type enzyme but was more thermostable. Mutants Arg49 → Leu, Arg49 → Asp, Arg49 → Cys were less thermostable than the wild-type enzyme yet exhibited wild-type values fork cat and K m . TheK i for the competitive inhibitor fructose 2,6-bisphosphate increased 3- and 5-fold in Arg49 → Leu and Arg49 → Asp, respectively. The K a for Mg2+ increased 4–8-fold for the Arg49mutants, with no alteration in the cooperativity of Mg2+binding. Position 49 mutants had 4–10-fold lower AMP affinity. Most significantly, the mechanism of AMP inhibition with respect to fructose 1,6-bisphosphate changed from noncompetitive (wild-type enzyme) to competitive (Arg49 → Leu and Arg49 → Asp mutants) and to uncompetitive (Arg49 → Cys mutant). In addition, AMP cooperativity was absent in the Arg49mutants. The R and T-state circular dichroism spectra of the position 49 mutants were identical and superimposable on only the R-state spectrum of the wild-type enzyme. Changes from noncompetitive to competitive inhibition by AMP can be accommodated within the framework of a steady-state Random Bi Bi mechanism. The appearance of uncompetitive inhibition, however, suggests that a more complex mechanism may be necessary to account for the kinetic properties of the enzyme.

In the context of FBPase kinetics, the binding of AMP results in a diverse and complex set of phenomena. The inhibition of FBPase by AMP and Fru-2,6-P 2 is synergistic (13,14), and the Fru-2,6-P 2 -induced enhancement of AMP binding is attributed to a decrease in the k off for the nucleotide (4). AMP and Mg 2ϩ are mutually exclusive in their binding to FBPase (11,15). Crystallographic studies have shown that AMP indirectly perturbs metal binding sites (16). AMP inhibition is cooperative, with a Hill coefficient of 2 (7)(8)(9). The first two molecules of AMP putatively bind with positive cooperativity, whereas the last two molecules bind with negative cooperativity (17). The mechanism for AMP inhibition is nonlinear and noncompetitive, with respect to Fru-1,6-P 2 and nonlinear and competitive with respect to Mg 2ϩ (11).
The four identical subunits (C1, C2, C3, and C4) of FBPase each consist of single AMP (residues 1-200) and FBP (residues 200 -335) binding domains (10). The tetramer is roughly a square with the upper left vertex occupied by subunit C1 followed by C2, C3, and C4 in a clockwise sense. Structures of AMP complexes of FBPase define the T-state, whereas structures of FBPase in complexes with substrates or substrate analogs without AMP define the R-state. To a first approximation, the transition from the T-to R-state is a 17°rotation of the lower subunit pair (C3-C4) relative to upper subunit pair (C1-C2) coupled with a 1.9°rotation of the AMP domain relative to the FBP domain within each subunit. The R-to T-state transition results in conformational changes at interfaces between C1 and C2, C1 and C4, and the AMP and FBP domains within a subunit (16). The C-terminal residues of helix H2, particularly residues 49 and 50, are strategically positioned at the C1-C2 interface, near the molecular center of the tetramer, approximately 20 Å from both the active site and the AMP binding site. The significance of Arg 49 and Lys 50 to the function of porcine liver FBPase is examined here by site-directed mutagenesis, circular dichroism (CD) spectroscopy, kinetics, and structural modeling. Our results clearly demonstrate the importance of position 49 in AMP cooperativity and inhibition. In addition, both residues are important determinants in the thermostability of the FBPase tetramer. duction of specific base changes into a double-stranded plasmid (18). Four mutagenic primers, 5Ј-ACCGCAGTCGGCAAGGC-3Ј, 5Ј-ACCG-CAGTCGACAAGGC-3Ј, 5Ј-ACCGCAGTCTGCAAGGC-3Ј, and 5Ј-GCAGTCCGCATGGCGGGCA-3Ј (mismatched bases in boldface), were used to mutate Arg 49 3 Leu, Arg 49 3 Asp, Arg 49 3 Cys, and Lys 50 3 Met, respectively. A selective primer, 5Ј-CAGCCTCGCCTC-GAGAACGCCA-3Ј, which exchanged the original NruI site for a unique XhoI site on the pET-11a vector was used. The double-stranded FBPase expression plasmid (pET-FBP) (19) and mutagenic and selective primers were denatured, annealed, and polymerized as described by Deng and Nickoloff (18).
Mutations were confirmed by NruI/XhoI digestion and by fluorescent dideoxy chain termination sequencing at the Nucleic Acid Facility at Iowa State University. The sequencing primer 5Ј-GTCATTGGAGAG-GACATC-3Ј was used to confirm the mutations. The mutagenesis plasmid was finally transformed into Escherichia coli DF 657, a strain deficient in the FBPase gene.
Protein concentration was assayed as described by Bradford (21) with bovine serum albumin (Sigma) as the standard. The protein purity was determined by using 12% SDS-polyacrylamide gel electrophoresis according to Laemmli (22).
Kinetic Studies-Specific activity during purification was determined by the phosphoglucoisomerase and glucose-6-phosphate dehydrogenase-coupled spectrometric assay at pH 7.5 and pH 9.6 (11,23). All other kinetic experiments were done at pH 7.5 (Hepes buffer) and 25°C by using a coupled spectrofluorometric assay (11). Initial rate data were analyzed by using a computer program written in MINITAB with an ␣ value of 2.0 (11,24). Cooperativity was evaluated by using either the ENZFITTER program (25) or the MINITAB program.
Circular Dichroism Spectrometry-CD studies on the wild-type and mutant forms of FBPase in the presence or absence of ligands employed a Jasco J710 CD spectrometer and a 1-mm cell at room temperature. Spectra were collected from 200 to 260 nm in 1.3-nm increments, and each spectrum was blank-corrected and smoothed by using the software package provided with the instrument.
Modeling Studies-Models for the Arg 49 3 Leu, Arg 49 3 Cys, Arg 49 3 Asp, and Lys 50 3 Met mutants were built from the x-ray structures of the R-and T-states of the wild-type pig kidney enzyme (26,27), whose amino acid sequence is identical to that of porcine liver FBPase, using the program XtalView (28). The Arg 49 3 Asp T-state model was taken from the partially refined x-ray crystal structure determined at Iowa State University. 2 With the exception of the Arg 49 3 Asp T-state model, the best conformations of the side chains of Leu 49 , Cys 49 , Asp 49 , and Met 50 were obtained by a systematic search and then subjected to 20 steps of conjugate gradient energy minimization using XPLOR (29). The coordinates of all atoms except those within a 10 Å radius around the residue of interest were fixed at their crystallographic positions. Solvent-accessible surface areas were calculated using GRASP (30) for residues 168 -171 (turn T2) for the R-and T-state models after removal of all water molecules.

Purification of the Wild-type and Mutant Forms of FBPase-
The elution profiles from the CM-Sephadex column were similar for mutant and wild-type enzymes, except that Arg 49 3 Leu does not bind to the column. An additional Cibacron blue column was used to obtain homogeneous Arg 49 3 Leu enzyme. The wild-type and mutant forms of FBPases exhibited identical mobility and are Ͼ96% homogeneous using electrophoresis as a criterion (Fig. 1). The pH 7.5/pH 9.6 activity ratios (Table I), along with the SDS-polyacrylamide gel electrophoresis results, suggest that the enzymes had not undergone proteolysis (23).
Temperature Sensitivity of Wild-type and Mutant FBPases-The thermosensitivity of wild-type and mutant FBPases were investigated to further evaluate the effect of mutations. Arg 49 3 Leu and Arg 49 3 Cys activity decreased to approximately 10 and 40%, respectively, compared with the wild-type enzyme after a 10-min incubation at 57°C (Fig. 2). In contrast, the activity of Arg 49 3 Asp was Ͻ10% at 47°C. The activity of the wild-type and Lys 50 3 Met enzymes did not change at 57°C; however, a 70 and 20% decrease in activity was observed at 67°C for wild type and Lys 50 3 Met mutations, respectively.
Kinetic Properties of FBPase Mutants-Mutations of Arg 49 and Lys 50 do not alter the specific activity of FBPase or the K m for Fru-1,6-P 2 ( Table I). The K a and Hill coefficient for Mg 2ϩ of the wild-type and mutant forms of FBPase were determined at saturating Fru-1,6-P 2 concentrations (12-20 M). The wildtype as well as the mutant enzymes showed sigmoidal kinetics for Mg 2ϩ ; however, 4 -8-fold increases in the K a for Mg 2ϩ was found in the mutant enzymes relative to the wild-type enzyme (Table I). Slight decreases in affinity (3-5-fold) for Fru-2,6-P 2 were found in Arg 49 3 Leu and Arg 49 3 Asp, but no change was found for either Arg 49 3 Cys or Lys 50 3 Met relative to wild-type FBPase. Thus, Arg 49 and Lys 50 are involved neither in Fru-2,6-P 2 inhibition nor in the discrimination between Fru-1,6-P 2 and Fru-2,6-P 2 .
AMP inhibition is nonlinear and noncompetitive with respect to Fru-1,6-P 2 but nonlinear and competitive relative to Mg 2ϩ for the wild-type enzyme (11). The expected pattern for noncompetitive inhibition of AMP relative to Fru-1,6-P 2 was found in wild-type and Lys 50 3 Met FBPases; however, the mechanism was changed to competitive inhibition in Arg 49 3 Leu and Arg 49 3 Asp and to uncompetitive inhibition in the Arg 49 3 Cys mutant. In addition, the cooperativity for AMP inhibition was lost in the three Arg 49 mutants. A double-reciprocal plot of 1/velocity versus [1/Fru-1,6-P 2 ] at various concentrations of AMP for Lys 50 3 Met gave an excellent fit to Equation 1 when n ϭ 2 (data not shown). The "goodness of fit" was 4%. AMP inhibition, then, is noncompetitive and cooperative as for the wild-type enzyme (data not shown). Equation 1 is, where v, V m , A, I, K a , K i , and K ii represent initial velocity, maximal velocity, the concentration of Fru-1,6-P 2 , the concentration of AMP, the Michaelis constants for Fru-1,6-P 2 , the dissociation constants for AMP from the enzyme AMP, and the enzyme⅐Fru-1,6-P 2 -AMP complexes, respectively. tant (data not presented) gave a family of lines intersecting on the 1/v axis. The Arg 49 3 Asp mutant gave a similar result (data not shown). Thus, AMP inhibition of Arg 49 3 Asp and Arg 49 3 Leu is competitive with respect to Fru-1,6-P 2 . The data fit best to Equation 2 with n ϭ 1.
where v, V m , A, I, n, K a , and K i are defined as above. The goodness of fit is Ͻ4% for both mutants when n ϭ 1 and Ͼ10% when n ϭ 2. AMP inhibition relative to Fru-1,6-P 2 of Arg 49 3 Cys gave a family of parallel lines (Fig. 3), indicating a mechanism of uncompetitive inhibition. Equation 3 with n ϭ 1 best accounts for the data.
where v, V m , A, I, n, K a , and K i are as above. The goodness of fit was Ͻ4%. Thus, cooperativity for AMP inhibition is lost in addition to the change in the kinetic mechanism. Residue 49 plays an important role in AMP cooperativity and the mechanism of AMP inhibition. All four mutants as well as wild-type FBPase exhibited competitive inhibition by AMP relative to Mg 2ϩ (data not shown). The K i (I 50 ) for AMP increased 4 -10-fold for Arg 49 mutants.
The Lys 50 3 Met mutation did not influence AMP affinity.
CD Spectrometry of FBPase-In the absence of ligands, the CD spectra from 200 to 260 nm of wild-type and the four mutant FBPases were superimposable (data not shown), indicating no major differences in secondary structure. In the presence or absence of saturating concentrations of Fru-1,6-P 2 and Mg 2ϩ (R-state), the CD spectra for the wild-type enzyme from 200 to 260 nm were superimposable (Fig. 4). The CD spectra of the wild-type enzyme in the presence of AMP and substrate or in the presence of AMP alone (data not shown) were identical to each other but differed in the region of 210 nm from the spectrum of the wild-type enzyme in the R-state (Fig. 4). CD spectroscopy is thus sensitive to differences in the R-and T-states of the wild-type enzyme.
The CD spectra of Lys 50 3 Met in the R-and T-state are identical to the CD spectra of wild-type FBPase in the R-and T-states, respectively (data not shown). On the other hand, the CD spectra of the R-and T-states of Arg 49 3 Leu, Arg 49 3 Asp, and Arg 49 3 Cys are identical (data not shown). Furthermore, the R-and T-state spectra of Arg 49 3 Leu are similar to the R-state spectrum of wild-type enzyme but dissimilar at 210 nm to the T-state spectrum of the wild-type enzyme (Fig. 5). These data indicate the existence of a CD-sensitive, R-like feature in AMP complexes of Arg 49 mutants. DISCUSSION Lipscomb and co-workers (16, 26) have proposed that FB-Pase exists in two different conformational states, an active Rand an inactive or less active T-state. Differences between the  2. Temperature sensitivity of wild-type and mutant forms of FBPases. The purified wild-type (f), Lys 50 3 Met (ϫ), Arg 49 3 Cys (Ⅺ), Arg 49 3 Leu (ϩ), and Arg 49 3 Asp (OE) at enzyme concentrations of 67 g/ml were incubated 10 min at 30, 37, 47, 57, and 67°C, respectively, in 50 mM malonate buffer at pH 6.0. Enzyme activity was assayed spectrophotometrically at 25°C (as described under "Experimental Procedures") immediately after the incubation and is expressed as a percentage of the relative activity at 30°C. The protein concentration was 0.2 mg/ml in each assay. Each reaction was either in duplicate or triplicate. two states are most apparent at the C1-C4 and C1-C2 interfaces (and interfaces related to these by molecular symmetry). Residues in the AMP binding domain contribute to both interfaces (16). In particular, C1 Arg 49 hydrogen bonds with C2 Gly 168 and C2 Ser 169 of turn T2, which itself is in contact with loop L6 (residues 128 -130 of C2) located at the C-terminal end of helix H4 (residues 123-127 of C2). Helix H4 is near the active site and may have an influence on the binding of metal ions and/or the 1-phosphate group of the substrate (16). AMP initiates the transition from the active R-to the inactive or less active T-state or stabilizes the T-state relative to the R-state. The R-to T-state transition involves significant intra-and intersubunit rearrangements that are coupled in the wild-type enzyme (31).
Preliminary crystal structures of Arg 49 3 Asp in the presence of AMP reveal a T-state tetramer, which has R-like subunits. 2 The FBP binding domains of the C1-C2 dimer (the asymmetric unit of the crystal) can be superimposed on the corresponding domain of the canonical T-and R-states of FBPase. The orientation of the AMP binding domain relative to the FBP binding domain of the Arg 49 3 Asp mutant approximates those of the R-state subunit. Thus, the T-state tetramer of Arg 49 3 Asp has R-like subunits in the context of the canonical T-state quaternary arrangement of those subunits.
The CD spectra of position 49 mutants in the presence and absence of AMP are identical to each other and to the spectrum of the R-state of the wild-type enzyme. In contrast, the CD spectra of wild-type R-and T-states are different. Invariant CD spectra (with or without AMP) of Arg 49 mutants may reflect the absence of conformational change in the mutant or be the result of conformational changes that are CD-insensitive. The Arg 49 mutants are not locked into an R-state quaternary arrangement of subunits, as the crystal structure of Arg 49 3 Asp clearly reveals a T-state subunit arrangement. CD spectroscopy may be sensitive then to changes in relative orientation of the AMP and FBP domains within a subunit. Thus, an R-like subunit may produce an R-state CD spectrum regardless of the quaternary subunit arrangement.
The most significant findings reported here are the loss of AMP cooperatively and the change in mechanism of AMP in-hibition in the Arg 49 mutants. Loss of AMP cooperatively has been observed by Shyur et al. (32) in an Arg 22 3 Met mutant (C1-C4 interface) and more recently by Lu et al. (33) in mutations of Glu 192 to alanine and glutamine (C1-C4 interface). The Arg 49 mutants are the first instances of mutations at the C1-C2 interface that abolish AMP cooperativity. Presumably the stabilization by AMP of a T-like conformation of one subunit (say C1) can be communicated through the C1-C2 interface to subunit C2 or the C1-C4 interface to subunit C4. If subunit C1 cannot adopt a T-like conformation and remains R-like in the presence of AMP, subunit C2 and/or C4 may not be influenced by the binding of AMP to subunit C1, hence the loss of positive cooperativity in AMP association. The modest decrease in AMP affinity observed in position 49 mutants may represent the loss in synergy between a pair of AMP molecules bound to C1/C2 or C1/C4. As mutations at the C1-C2 interface (Arg 49 ) and the C1-C4 interface (Arg 22 and Glu 192 ) independently abolish AMP cooperativity, we cannot however unambiguously assign the phenomenon of cooperativity to a specific pair of subunits. Nonetheless, the crystal structure of the AMP complex of the Arg 22 3 Met mutant reveals a canonical T-state, 3 implying that the loss in cooperativity is due to an effect (perhaps electrostatic in origin) localized to the C1-C4 interface (31). In contrast, the Arg 49 3 Asp mutant has an impact on both C1-C2 and C1-C4 interfaces. As only the C1-C4 interface is perturbed by both mutations, the subunit pair most likely associated with positive cooperativity is C1-C4. Lu et al. (33) also ascribed the origin of AMP cooperativity to the C1-C4 interface. The involvement of the C1-C2 interface in AMP cooperativity, however, was not considered by Lu et al., perhaps because their work predates the discovery of a C1-C2 interface mutant that abolishes AMP cooperativity.
It is possible to rationalize the data for noncompetitive inhibition and competitive inhibition within the context of the steady-state Random Bi Bi mechanism proposed for FBPase (34) and the suggested relationship of interactions of FBPase with AMP and Mg 2ϩ (15). If it is assumed that AMP can bind to the free enzyme and the FBPase⅐FBP complex, it is possible to reduce the steady-state Random Bi Bi initial rate equation to the competitive and noncompetitive models by simply making assumptions regarding the relationships that must exist among rate constants. Similar manipulations were made by Rudolph and Fromm (35) for hexokinase by computer simulations. On the other hand, the steady-state Random Bi Bi mechanism cannot readily account for the results of Fig. 3 (uncompetitive inhibition). The kinetic data infer that AMP inhibits wild-type FBPase by several related (and coupled) mechanisms and that the kind of side chain at position 49 determines the dominant mechanism of inhibition.
Although other explanations are possible, the decrease in thermal stability of the Arg 49 mutant can be explained by structural perturbations localized to the C1-C2 interface. In the R-state, C2 Arg 49 hydrogen bonds with C1 Gly 168 and is in van der Waals contact with C1 Arg 49 (16). In the T-state, C2 Arg 49 hydrogen bonds with C2 Gly 168 , C2 Ser 169 , and C2 Thr 171 (16). The position-49 mutations eliminate the C1 Arg 49 -C2 Arg 49 van der Waals contacts and the hydrogen bonds in both the Rand T-states. Another consequence of the mutation may be an increased exposure of residues 168 -171 (turn T2) of the C1-C2 interface to water (Table II). In fact, for the Arg 49 3 Asp mutation, nonbonded contacts are not only lost but replaced with repulsive electrostatic interactions between C1 Asp 49 and C2 Asp 49 , 2 possibly accounting for the increased thermosensitivity of the Arg 49 3 Asp mutant (Fig. 2). Both the loss of intersubunit hydrogen bonds and the increase in solvent-acces-sible surface area of the C1-C2 interface may contribute significantly to the decreased thermostability of the other Arg 49 mutations.
The enhanced thermostability of the Lys 50 3 Met mutant, however, is not readily apparent from an examination of the Rand T-states. The energy-minimized model for the Lys 50 3 Met mutant in the R-state retains van der Waals contacts but replaces the salt-link of C1 Lys 50 to C2 Asp 187 with a nonbonded contact. The T-state model for Lys 50 3 Met retains similar nonbonded contacts to those of the wild-type enzyme. The enhanced thermostability of the Lys 50 3 Met mutant then must originate from either a kinetic phenomenon (an increase in the energy of activation of a fundamental step in the unfolding process) or a difference in the free energy between the unfolded states of the mutant and wild-type enzymes.