Conversion of Non-allosteric Pyruvate Kinase Isozyme into an Allosteric Enzyme by a Single Amino Acid Substitution*

Pyruvate kinase M 1 , a non-allosteric isozyme, was con- verted into an allosteric enzyme by replacement of an amino acid in the intersubunit contact. The substitution of Ala-398 with Arg resulted in the pronounced allosteric enzyme. The Hill coefficient and the substrate con- centration giving one-half of V max for the mutant with respect to phosphoenolpyruvate were 2.7 and 0.41 m M , respectively, whereas those values for the wild type were 1.0 and 0.049 m M . This mutation, however, gave rise to only minor effects on the apparent values of K m for ADP and on V max . Furthermore, in contrast to the wild- type enzyme, the mutant was activated by fructose 1,6-bisphosphate. The Hill coefficient of the mutant was no longer increased by the allosteric inhibitor, L -phenylala-nine, indicating that the equilibrium for the unligated enzyme is largely shifted toward the T-state. These results suggest that Ala-398 is one of the most critical residues allowing the enzyme to prefer the R-state and that allosteric regulation of pyruvate kinase involves amino acid residues in the intersubunit contact. Pyruvate kinase (PK, 1 ATP:pyruvate 2- O -phosphotrans-ferase, EC 2.7.1.40) catalyzes the transfer of a phosphoryl group of phosphoenolpyruvate (PEP) to ADP and plays a key role as a regulatory enzyme in the glycolytic pathway ting data for varied concentrations of PEP to the Hill equation. The Michaelis-Menten equation was used to determine parameters for ADP. These calculations were carried out using non-linear regression analy- sis based on the Marquardt algorithm. When effects of FBP and L phenylalanine on kinetic parameters were investigated, kinetic exper- iments were performed similarly in the presence of these allosteric effectors. In some kinetic analyses, the concentration ranges of PEP were varied around the S 0.5 (the substrate concentration giving one-half of V max ), depending on the properties of enzymes. Protein Determination— Protein contents were determined according to the method of Bradford (31) using bovine serum albumin as a standard.

The M 1 and M 2 isozymes are produced from a single gene locus by mutually exclusive alternative splicing (8 -10); the M 2 isozyme is expressed predominantly in the kidney, and M 1 is predominant in skeletal muscle, heart, and brain (1). In addition, it is also known that expression of these isozymes shifts from M 2 to M 1 during development of some fetal tissues (2). In contrast to the other PK isozymes, M 1 isozyme usually exhibits neither homotropic nor heterotropic allosteric effects (1,5,6). This isozyme, however, displays cooperative behavior under certain experimental conditions such as the presence of Lphenylalanine, an allosteric inhibitor (11)(12)(13). Furthermore, heterotropic activation of the M 1 isozyme by FBP is observed only as a reversal of such an inhibited enzyme (14). Nevertheless, the biological significance of the lack of allosteric effects on the M 1 isozyme is not fully understood.
In the rat PK-M 1 and -M 2 isozymes, the exon that is exchanged due to the alternative splicing encodes 56 amino acids, in which a total of 21 amino acid residues differ within a length of 35 residues (8,9). Thus, it was proposed that the distinguishable kinetic properties of M 1 and M 2 isozymes could be attributed to these amino acid substitutions. Moreover, it has been shown by x-ray crystallographic analyses and computer modeling that the corresponding regions of their polypeptides participate directly in the intersubunit contact, which may be primarily responsible for the intersubunit communication required for allosteric cooperativity (15,16). Therefore, it has been proposed that the amino acid residue(s) in the intersubunit contact of the allosteric isozymes, particularly in the two ␣ helices (Fig. 1, C␣1 and C␣2), play a central role in the allosteric interaction of the subunits (17,18).
The M 1 isozymes of vertebrates seem to result from amino acid substitutions in ancestral allosteric isozymes since nonallosteric PKs such as the M 1 isozyme have been found in vertebrates but never in microorganisms (19 -22). The M 1 isozyme is thought to be a specialized isozyme that arose in vertebrates for energy metabolism in particular tissues such as muscle and brain. The structure of the intersubunit contact of the PK-M 1 isozyme appears to prefer the active conformation, as the isozyme has been regarded as a model of the R-state (14). Many studies have focused on the structures prerequisite to allosteric properties in PKs (17,18,23), while the structural basis of the maintenance of the R-state in the M 1 isozyme still remains to be elucidated.
Since the primary structures of the M 1 and M 2 isozymes are identical except the region forming the intersubunit contact, the structure of the corresponding region of the M 2 isozyme permits allosteric interactions whereas that of the M 1 isozyme appears to prevent the enzyme from changing from its active conformation. To identify an amino acid residue required to maintain the active conformation as the R-state in the PK-M 1 isozyme, our approach was to substitute the amino acid residues specific to the allosteric isozymes of PK so that we would convert the rat M 1 isozyme into an allosteric enzyme. In the current study, we prepared a series of PK-M 1 mutants in which the residues in the intersubunit contact were replaced and characterized them to examine whether the substitutions induce allosteric properties. This functional analysis provides insight into the contribution of a single amino acid residue to the maintenance of the R-state. In addition, since PK-M 1 is the most stable isozyme in mammalian PKs, this allosteric mutant of PK-M 1 might be available as a T-state model for further structural analyses such as x-ray crystallography.

EXPERIMENTAL PROCEDURES
Materials-Restriction endonuclease and DNA-modifying enzymes were purchased from Takara. Fructose 1,6-bisphosphate trisodium salt was from Sigma. Phosphoenolpyruvate and ADP were obtained from Wako Pure Chemicals and Oriental Yeast, respectively. Lactate dehydrogenase and NADH were products of Boehringer Mannheim. Oligonucleotide primers were synthesized by Greiner Japan. Other common chemicals were from Wako Pure Chemicals or Nacalai Tesque.
Construction of the Transfer Plasmid-A 5Ј SalI-PstI 0.65-kb fragment of rat PK-M 1 cDNA, which we cloned previously (8), was subcloned into a pSVK3 vector (Pharmacia Biotech Inc.). Subsequently, a BamHI-PstI 0.67-kb fragment was excised from the resulting plasmid and then ligated into a pBluescript SKϩ. A 3Ј BglII-EcoRI 1.3-kb cDNA fragment was ligated downstream of the 5Ј cDNA fragment contained in that plasmid, following digestion by these restriction endonucleases. The BamHI-EcoRI 1.9-kb fragment, which contains the entire coding sequence, was excised from the resultant plasmid and inserted into a transfer vector, pVL1393 (Invitrogen).
Site-directed Mutagenesis-Site-directed mutagenesis was carried out according to Kunkel (24), as described previously (25). Prior to mutagenesis, the SphI-KpnI 0.2-kb fragment of rat PK-M 1 cDNA, which contains the amino acid sequences different between the M 1 and M 2 isozymes, was subcloned into a pTV119N vector (Takara). Subsequently, a HindIII-KpnI fragment excised from the resultant plasmid was then ligated to pBluescript SKϩ. The uracil-substituted singlestranded template was prepared from Escherichia coli CJ236 transformed by the plasmid. The uracil-template was used with synthetic oligonucleotide primers to replace residues of the M 1 isozyme with those of mammalian allosteric PKs. The oligonucleotide primers used in this study are follows: 5Ј-GCAGCCGTGTACCACCGCCTG-3Ј for replacement of Phe-389 by Tyr (designated as F389Y), 5Ј-GAAGAGCTTCGC-CGAGCCTCC-3Ј for Ala-398 by Arg (A398R), 5Ј-GAGCTTGCGCG-AGCGGCCCCCCAATCCACAGACCCC-3Ј for a double mutation of Ser-401 by Ala and Ser-402 by Pro (S401A/S402P), 5Ј-TCCACAGATCCC-ACCGAGGCCATG-3Ј for Leu-408 by Thr (L408T), 5Ј-TATAAATGTT-GCGCAGCAGC-3Ј for Leu-423 by Cys (L423C), and 5Ј-GCAGCAGC-CATCATAGTTCTG-3Ј for Leu-427 by Ile (L427I). The resulting mutations were verified by dideoxy sequencing using a DNA sequencer (Applied Biosystems, model 373A), as were the entire sequences subjected to mutagenesis. The corresponding region of the wild-type PK-M 1 cDNA was replaced by each mutant sequence. The transfer plasmids for the mutant enzymes were constructed similarly to those of the wildtype enzyme and used for transfection.
Preparation of Recombinant Viruses-The purified transfer plasmids containing the wild-type or mutant PK-M 1 cDNA (1 g) were co-transfected into 5 ϫ 10 5 Sf21 cells with 10 ng of BaculoGold DNA (PharMingen), which was used as Autographa californica nuclear polyhedrosis viral genome. Transfection experiments were carried out by the Lipofectin (Life Technologies, Inc.) method (27), as described previously (28,29). Media containing the recombinant viruses generated by homologous recombinations were collected 6 days after transfection. The recombinant viruses were further amplified to more than 5 ϫ 10 7 plaqueforming units/ml prior to use.
Expression of Recombinant Rat PK-M 1 in Insect Cells-2 ϫ 10 8 Sf21 cells were infected with the recombinant viruses carrying either wildtype or mutant PK-M 1 at multiplicity of infection of more than 8. The infected cells were harvested about 90 h postinfection to purify the expressed proteins.
Purification of the Recombinant Enzymes-Expressed recombinant PKs were purified basically according to Imamura and Tanaka (1). Sf21 cells producing the recombinant enzymes were pelleted by centrifugation at 2,500 ϫ g for 10 min. The cells were homogenized in 20 mM Tris-HCl, 5 mM MgSO 4 , 1 mM EDTA, and 10 mM 2-mercaptoethanol (pH 7.5) with a Dounce homogenizer and were centrifuged at 10,000 ϫ g to obtain clarified extracts. Subsequently, the supernatants were treated with polyethyleneimine to remove contaminated nucleic acids and then subjected to fractionation by 45-80% saturation of ammonium sulfate. After dialysis against 10 mM sodium phosphate, 2 mM MgSO 4 , and 10 mM 2-mercaptoethanol (pH 6.0), the enzymes were subjected to a phosphocellulose column (Whatman P-11) pre-equilibrated with the same buffer used in dialysis. After washing intensively with the buffer containing 80 mM KCl, the bound enzymes were eluted with an addition of 0.5 mM PEP to the wash buffer.
Electrophoresis-The purified enzymes were subjected to SDS-PAGE analysis on 10% gels, according to Laemmli (30). The proteins were visualized by Coomassie Brilliant Blue R-250.
Enzyme Activity Assay-Standard assay for PK activity was performed at 37°C using 2 mM both PEP and ADP in 50 mM Tris-HCl buffer, 0.1 M KCl, 5 mM MgSO 4 , and 0.5 mM FBP (pH 7.5) as described (1). This substrate mixture (1 ml) also contained 17 units of lactate dehydrogenase and 0.17 mM NADH to monitor release of pyruvate as a change of absorbance at 340 nm. One unit of activity was defined as the quantity of the enzyme that released 1 mol of pyruvate per min.
Kinetic Analyses-Enzymatic activity was assayed at 37°C using various concentrations of PEP or ADP. The condition used for kinetics was the same as above except the substrates and effector. In assessment of parameters for one substrate, the concentration of the other was fixed at 2 mM. Ten different concentrations of PEP between 10 M and 2.0 mM were used to obtain kinetic parameters for the substrate. When parameters for ADP were determined, six concentrations of the substrate from 62.5 M to 2.0 mM were employed with the constant concentration of MgSO 4 (5 mM). Therefore, the parameters obtained for ADP are apparent ones. Release of pyruvate was monitored using a Beckman DU-640 spectrophotometer by coupling with the lactate dehydrogenase-NADH system. Kinetic parameters were obtained by fit- ting data for varied concentrations of PEP to the Hill equation. The Michaelis-Menten equation was used to determine parameters for ADP. These calculations were carried out using non-linear regression analysis based on the Marquardt algorithm. When effects of FBP and Lphenylalanine on kinetic parameters were investigated, kinetic experiments were performed similarly in the presence of these allosteric effectors. In some kinetic analyses, the concentration ranges of PEP were varied around the S 0.5 (the substrate concentration giving one-half of V max ), depending on the properties of enzymes.
Protein Determination-Protein contents were determined according to the method of Bradford (31) using bovine serum albumin as a standard.

RESULTS
We used the baculovirus-insect cell system, which is known as one of the most powerful expression systems (32), to produce a sufficient amount of active recombinant rat PK-M 1 . When insect cells were infected with the recombinant virus carrying the PK-M 1 cDNA, the PK activity increased to about 100 units/mg of cell protein 90 h after infection. The activity was more than 100 times greater than that in uninfected cells. As shown in Fig. 2, the expression level of the PK-M 1 protein was apparently as high as 20% of the total soluble proteins in the infected cells. Thus, a large amount of the active enzyme of rat PK-M 1 can be produced by the baculovirus system.
We obtained more than 10 mg of the purified enzyme from 2 ϫ 10 8 cells infected with the recombinant virus. SDS-PAGE analysis demonstrates a single band of about 57 kDa, which corresponds to a single subunit of the M 1 isozyme (Fig. 2). When the purified enzyme was subjected to a gel permeation chromatography on a fast protein liquid chromatography system equipped with Superose 12 (Pharmacia), the elution time of the recombinant enzyme was identical to that of the authentic M 1 isozyme purified from rat muscle (data not shown), suggesting that the recombinant enzyme is a tetramer similar to the authentic isozyme (1,2). In addition, kinetic properties of the recombinant M 1 given in Table I were indistinguishable from those of the authentic rat M 1 reported (1).
Amino acid sequences of the regions forming the intersubunit contacts of various PKs (7-10, 15, 37) were aligned to find candidate amino acid residues to be replaced for the conversion of PK-M 1 into an allosteric enzyme (Fig. 3). Mutagenesis was used to introduce changes into the corresponding positions where residues are conserved in allosteric PKs but not in M 1 isozymes. The residues indicated by arrowheads in rat PK-M 1 were substituted by the corresponding amino acids that are conserved in allosteric PKs. The mutant M 1 isozymes were expressed in insect cells and were purified as described under "Experimental Procedures." These purified mutants also exhibited a single protein band with apparent molecular mass of 57 kDa on SDS-PAGE (Fig. 4).
Kinetic behavior of the purified mutant M 1 enzymes with respect to PEP are shown in Fig. 5, and Table I (1, 38). Thus, a single substitution by arginine at position 398 is sufficient to convert the non-allosteric PK-M 1 into a typical allosteric enzyme comparable to the PK-L, -R, and -M 2 isozymes. To examine whether amino acid replacements caused concomitant effects on other kinetic parameters, apparent parameters of V max and K m for ADP were assessed by varying the concentrations of ADP in the presence of the fixed concentrations of PEP (2 mM) and MgSO 4 (5 mM). Hyperbolic responses were observed in all of the mutants, similar to the wild-type PK-M 1 (data not shown). Furthermore, none of the amino acid substitutions including the A398R caused any substantial changes of the apparent parameters for ADP (Table I). These results indicate that the substitutions affect the binding of PEP without any effect on the binding of ADP or on the catalysis.
Effects of FBP and L-phenylalanine on the A398R mutant were evaluated to characterize the nature of the allosteric properties of this mutant. It is known that these effectors act on the activity of PK through the shift of equilibrium of the allosteric transition between the active R-state and the inactive T-state (13,14,33). While FBP had little effect on the properties of the wild type (Fig. 6), the profile of the kinetic behavior of A398R changed remarkably from a sigmoidal curve to a nearly hyperbolic one as the concentration of FBP increased (Fig. 6B). As shown in Fig. 6, C and D, the S 0.5 and Hill coefficient for A398R decreased in a simple saturating manner with the increase of FBP concentration. FBP caused little change in V max of the mutant. The FBP activation profiles observed for A398R seem to be similar to those reported by Consler et al. (14), in which rabbit muscle PK was activated by FBP in the presence of L-phenylalanine. Activation of the A398R mutant by FBP allowed this enzyme to behave similar to the wild type, indicating that the effects of the substitution of Ala-398 by Arg on the kinetic properties are reversal by FBP. These results suggest that the consequence of the substitution is a reversible change that shifts the allosteric transition from the R-state to the T-state.
The other allosteric effector, L-phenylalanine, exhibits inhibitory effects by shifting the allosteric transition toward the T-state (14). S 0.5 values increase with L-phenylalanine concentration, revealing that the T-state became predominant relative to the R-state in both the wild type and A398R mutant (Fig. 7, A, B, and C). In the wild type, the Hill coefficient increased from 1.0 up to about 2.5 in the range of the Lphenylalanine concentrations examined (Fig. 7, A and D). In contrast, the Hill coefficient of the mutant was not increased further by L-phenylalanine, but rather decreased at higher concentrations of the effector (Fig. 7D). Thus, these results show that the homotropic cooperativity of this mutant for PEP is nearly at its maximum and that the state of the allosteric transition in the unligated enzyme is largely shifted toward the T-state. DISCUSSION In the present study, we examined the effects of amino acid substitutions on the kinetic properties of the rat PK-M 1 isozyme to identify the residues responsible for the enzyme to favor the R-state. Our results suggest that Ala-398 is a critical residue for the stabilization of the R-state in the PK-M 1 isozyme. The requirement of Ala-398 for maintenance of the R-state was revealed by a major disturbance of the allosteric properties of the A398R mutant (Fig. 5). The other residues examined in this study appear to participate, at least in part, in exhibition of the unique kinetic properties of the PK-M 1 isozyme distinct from allosteric isozymes since replacements of these residues also affected the properties of the PK-M 1 isozyme, albeit to a lesser extent. It seems that the subtle interactions of residues in the intersubunit contact allow the PK-M 1 isozyme to barely maintain its active conformation regarded as the R-state.
Although Ala-398 has been shown to play an important role for the rat PK-M 1 isozyme to exhibit non-allosteric properties, the alanine residue is not conserved among M 1 isozymes from other species. Amino acids with aliphatic side chains of various lengths are common to position 398 or its equivalent position for all of the M 1 isozymes (Fig. 3). Therefore, it seems likely that allosteric effects are induced in the rat PK-M 1 by introduction of a positive charge to position 398 rather than by that of a bulky side chain. Furthermore, Ala-398 is located at a position that is able to interact with Arg-391 of the symmetrically associated subunit (16), a residue that is also conserved in L and R isozymes but not in M 2 isozyme. This could account for a more pronounced cooperativity found in the A398R mutant than for the native M 2 isozyme. It is possible that an interaction of the arginine residue introduced into position 398 with Arg-391 of the facing subunit could contribute significantly to the occurrence of more cooperative behavior.
The reaction mechanism of PK is believed to involve the ternary complex that consists of PEP, ADP, and the enzyme (15). In this view, the transfer of a phosphoryl group occurs directly from PEP to ADP, and does not proceed via a phosphoenzyme species. The subsites for binding of these substrates must be very close to each other to carry out the direct transfer. Nonetheless, operations of the residues in the intersubunit contact induced both homotropic and heterotropic allosteric effects in the PK-M 1 isozyme without substantial changes of the apparent K m for ADP and of the catalytic constant, indicating that the effects of the amino acid replacements are specific to the binding of PEP and do not affect arrangement of catalytic residues or the subsite for ADP binding. Therefore, it is unlikely that gross structural effects occurred in the mutants resulting in a distortion of the PEPbinding site. The mutants of the PK-M 1 isozymes seem to mimic the allosteric transition displayed by the native allosteric isozymes. These are consistent with the suggestion that the amino acid residues in the intersubunit contact region regulate the state of the allosteric transition of PKs.
In the FBP activation profile of A398R, data fitting by a simple saturation model reveals that the apparent activation constant for FBP is about 1 mM (Fig. 6, C and D). This value is much higher than micromolar or submicromolar levels for native allosteric PKs such as M 2 and L isozymes (1). In contrast to these allosteric isozymes, the low binding affinity for FBP is intrinsic to mammalian M 1 isozymes, as indicated by the apparent activation constant of millimolar range (14). Therefore, the amino acid substitution has no significant effect on the binding of FBP since it is likely that the replacement leads to no alteration in the intrinsic properties of the M 1 isozyme. This suggests that the mutation introduced into the intersubunit contact does not change the local environment at the FBPbinding site.
Vertebrate PK-M 1 isozymes are predominantly expressed in muscle, heart, and brain, organs that are always more energyconsuming than other organs and tend to demand more glucose. Those organs seem to prefer smooth consistent catabolism of glucose rather than a demand-regulated metabolism. The reaction catalyzed by the PK-M 1 isozymes might not be a ratelimiting step in the course of glycolysis because of lack of allosteric regulation. These circumstances could lead to the specialized uncontrollable properties of the vertebrate PK-M 1 isozymes during evolution. In fact, a non-allosteric PK such as M 1 isozyme has not been found in microorganisms including bacteria and yeast (19 -22). Hence, the PK-M 1 isozyme is likely to be a highly specialized descendant of allosteric PKs. This specialization for PK might have been accomplished as a result of a shift in the allosteric state by a relatively small number of amino acid substitutions.
In the current study, we showed that a single amino acid substitution in the intersubunit contact converts the non-allosteric PK-M 1 isozyme into a more typical allosteric form of PK. For mammalian PKs, only the crystal structures of M 1 isozymes have been solved as an R-state conformation. On the other hand, it would be difficult to prepare crystals of the T-state form available for structural analysis in mammalian allosteric PKs such as M 2 and L isozymes probably because they are unstable without FBP. The stability of A398R, the most pronounced allosteric mutant, was found to be comparable to that of the wild-type PK-M 1 isozyme, as it could be purified and stored in the absence of a stabilizing agent such as FBP. Hence, this mutant is available as a stable model of PK in the T-state and might enable us to prepare a crystal of the T-state of mammalian PKs.
Recently, the unligated PK of E. coli was crystallized as the T-state conformation, and its crystal structure has been solved. The comparison between the structures of the E. coli PK (as the T-state) and mammalian M 1 isozyme (as the R-state) showed that rotatory movements of domains give rise to the allosteric transition (34 -36). However, similar structural analysis with the mutant and the wild-type PK-M 1 isozyme would allow a direct comparison of the R-state and the T-state in the identical enzyme with the exception of only a single residue and thereby might provide a more detailed molecular basis of the transition.