Expression of Rat Phosphoribosylpyrophosphate Synthetase Subunits I and I1 in Escherichia coli ISOLATION AND CHARACTERIZATION OF THE RECOMBINANT ISOFORMS*

The 34-kDa subunit of rat liver phosphoribosylpyrophosphate synthetase is a mixture of the two highly homologous isoforms, PRS I and PRS 11. Heretofore, it was not possible to separate the two. We now describe isolation and characterization of the recombinant iso- forms, named rPRS I and rPRS 11. The respective rat cDNAs were inserted into vectors constructed from pKK233-2 by replacing its replication origin with that of pGEM-1 and expressed in Escherichia coli. The rPRS I and rPRS I1 were purified to apparent homo- geneity with specific activities of 33,400 and 46,200 milliunits/mg, respectively; these values were at least 2.5-fold higher than the highest value for the mammalian enzyme so far reported. Both isoforms showed a similar dependency on Pi as an absolute activator. Sulfate partially substituted for Pi. The maximal ac- tivities of rPRS I and rPRS I1 with sulfate were 43 and 7%, respectively, of those seen with Pi. The two isoforms differed in sensitivity to inhibition by ADP and GDP. Inhibition of rPRS I and rPRS I1 by 0.3 mM ADP was 87 and 54%, respectively, and inhibition by 1 mM GDP was 93 and 24%, respectively. rPRS I1 was 180-fold more sensitive than rPRS I to heat inactivation at

aggregates of 3 4 , 38-, and 40-kDa components, with the 34-kDa species being the catalytic subunit (15). Our cloning of rat cDNA (16) and amino acid sequencing of the enzyme purified from rat liver (15) revealed that the 34-kDa component is actually a mixture of two highly homologous isoforms, designated as PRS I and PRS 11. The deduced amino acid sequence of both sets of 317 residues differ only by 13 residues (96% homology) (16). These two isoforms are encoded by different X-linked genes (PRPSl and PRPSB, respectively) (17). The amino acid sequences of both PRS I and PRS I1 are highly conserved. The deduced amino acid sequences of human (18,19) and rat (16) PRS I1 differ only by 3 residues (99% homology), and those of human (19,20) and rat (16) PRS I are completely conserved (100% homology). Rat mRNAs of PRPSl and PRPSB genes are expressed in almost all tissues, but the amounts differ with the tissue (21). All these observations suggest functional differences between catalytic and/or regulatory properties of PRS I and PRS 11. The kinetic and physical characteristics of the individual isoforms have remained unknown due to difficulty in isolating the respective species.
We now report the expression of respective rat PRS I and PRS I1 in E. coli. The strategy we used produced the unfused proteins, possessing catalytic activity. We isolated and characterized the recombinant isoforms.

EXPERIMENTAL PROCEDURES
Materials-Synthetic oligonucleotides were prepared using an Applied Biosystems DNA synthesizer, model 380B. Restriction and DNA-modifying enzymes were obtained from Takara Shuzo, Toyobo, and New England Biolabs. Vectors pGEM-1 and pKK233-2 were from Stratagene and Pharmacia LKB Biotechnology Inc, respectively. TSK DEAE-5PW and G4000SW columns were purchased from Tosoh Manufacturing. Sources of all other reagents were as described (15).
Construction of Expression Vectors pGlKHB and pG1KHM"The expression vectors were constructed by two modifications of pKK233-2. (a) The replication origin of pKK233-2, contained in the 3.4 kb of the EcoRI-Puul fragment (22), was replaced with that of pGEM-1.
(b) Multiple cloning sites, contained in the 45 bp of the HindIII-EcoRI fragment of pGEM-1, were inserted into the HindIII site located immediately downstream from the ATG initiation codon (22). The 31 bp of the EcoRI-Hind111 fragment of pBR322 was used as an adaptor. The resulting plasmid is designated pG1KHB. Another vector, pGlKHM, was constructed from pGlKHB by inserting the MZuI linker (GACGCGTC) into the SmaI site in the multiple cloning sites.
Construction of Expression Plasmids for Rat PRS I and PRS II-For expression of PRS I, the rat PRS I cDNA fragment spanning bases 28-1193 (16) was cloned in pGlKHB between the NcoI and the SmaI sites. The complementary synthetic oligonucleotides encoding the NHZ-terminal 9 amino acids were used as an adaptor. The constructed pKrI contained the complete coding region of 954 bp and 15693 I and II 0.24 kb of the 3'-noncoding region to the DraI site of rat PRS I cDNA (23).
For expression of PRS 11, the rat PRS I1 cDNA fragment spanning bases 14-2086 (16) was cloned in pGlKHM between the NcoI and the BarnHI sites. The synthetic oligonucleotides encoding the NH2terminal 5 amino acids were used as an adaptor. The resulting plasmid, pKrII, contained the complete coding region of 954 bp and 1.1 kb of the 3'-noncoding region to the BglII site of rat PRS I1 cDNA The sequences of the resulting expression plasmids were confirmed by restriction mapping and sequence analysis. The plasmids were introduced into the E. coli strain MV1304, which contains a lacP repressor. The transformed strains were designated MV1304/pKrI and MV1304/pKrII, respectively. As a negative control, pGlKHM alone was also transformed into MV1304. The cloning steps described above were performed essentially as described by Maniatis et al. (25).
Cell Growth-The transformed MV1304 cells were grown at 30 "C for 15 h in 10 liters of M9 medium (25) containing 1 mM thiamin and ampicillin (20 pg/ml). After adding 1 mM IPTG, the cells were grown for an additional 3 h at 30 "C and then harvested by centrifugation.
Purification of Recombinant PRS I and PRS ZI-All procedures for enzyme purification were carried out at 4 "C, and all buffers used contained enzyme-stabilizing agents (13,14), 0.3 mM ATP, 6 mM MgCl2, 0.1 mM EDTA, and 2.5 mM 2-mercaptoethanol, unless otherwise stated. The pH values indicated were measured at 4 "C.
The E. coli cells were suspended in a minimum volume of a solution containing 50 mM potassium phosphate (pH 7.4) and 0.1 mM phenylmethylsulfonyl fluoride and lysed by treatment with egg white lysozyme (1 mg/ml) for 20 min. The mixture was made up to 6 volumes with the same buffer and sonicated. The cell debris was sedimented at 10,000 X g for 15 min.
The next two steps, polyethylene glycol precipitation and acid precipitation, were performed according to the protocol described previously (15), with the following modifications. (a) rPRS I was precipitated with 3% (w/v) polyethylene glycol 6000. The precipitate was dissolved in 50 mM potassium phosphate (pH 7.4) containing 0.1 mM phenylmethylsulfonyl fluoride. The protein concentration was adjusted to 3.3 mg/ml. The enzyme was then precipitated at pH 6.15. (b) rPRS I1 was precipitated with 5% (w/v) polyethylene glycol 6000 and then precipitated at pH 5.85.
The acid precipitates of rPRS I and rPRS I1 were dissolved in 50 mM potassium phosphate buffer (pH 7.4) containing 0.1 mM phenylmethylsulfonyl fluoride. The protein concentrations were adjusted to 5 mg/ml and the pH to 7.4. The purification up to this step was performed within 1 day, and the preparations were stored at -80 "C.
A portion of the above preparations (5 mg of protein for each) was applied to a DEAE-5PW HPLC column (0.75 X 7.5 cm) equilibrated with 30 mM potassium phosphate buffer (pH 7.4) containing 6 mM MgC1, and 0.1 mM EDTA instead of the enzyme-stabilizing agents. Elution was performed first with 100 mM potassium phosphate (pH 6.8) containing the stabilizers and then with a linear gradient of 0-0.3 M KC1 in the same buffer. The eluted enzymes were pooled, dialyzed against 50 mM potassium phosphate (pH 7.41, and stored at -80 "C. These preparations were used for all kinetic studies. Enzyme Assay-The enzyme activity was assayed using the two methods described elsewhere (15). Briefly, method 1, used for crude enzyme preparations, involved use of the reaction mixture (0.5 ml) containing 50 mM potassium HEPES (pH 7.4) (adjusted at 37 " C ) , 10 mM potassium phosphate, 4 mM MgC12, 1 mM EDTA, 0.8 mM ATP, 0.2 mM ribose 5-phosphate, 1 mM dithiothreitol, 1.2 mM phosphoenolpyruvate, pyruvate kinase (9 units), and the enzyme, unless otherwise stated. The amount of PRPP synthesized was determined enzymatically.
Method 2, used for preparations after the acid precipitation step, measured ribose 5-phosphate-dependent [I4C]AMP production from [I4C]ATP. The composition of the assay mixture (0.1 ml) was identical with the above mixture, except that the unlabeled ATP was replaced with 0.4 mM [8-I4C]ATP, and phosphoenolpyruvate and pyruvate kinase were omitted.
One unit of enzyme activity was defined as the amount catalyzing formation of 1 pmol of PRPP/min, under standard conditions. Protein Concentration-Protein concentration was determined by the method of Bradford (26), using bovine serum albumin as a standard. The concentrations of the purified rPRS I and rPRS I1 were also determined from dry weight measurements. The method of Bradford slightly overestimated the concentrations of the two isoforms by a factor of 1.3, relative to the determination from dry weight measurements. (24). Molecular Weight Determination by Gel Filtration-The molecular weight of rPRS I and rPRS I1 was determined by gel filtration using a TSK G4000SW HPLC column (0.75 X 60 cm). The running buffer was 50 mM potassium phosphate (pH 7.4) containing the enzymestabilizing agents.
Other Methods-Rat liver PRPP synthetase was prepared as described previously (15). Polyacrylamide (10%) slab gel electrophoresis in SDS was performed using the buffer system of Laemmli (27). The NH2-terminal amino acid sequence analysis was performed using an Applied Biosystems 477A sequencer with a model 120A on-line HPLC system. Isoelectric focusing was carried out using a linear gradient of sucrose containing ampholine and 1 mM dithiothreitol, as described (28).

RESULTS
Construction of Expression Plasmids-The cDNA fragments of rat PRS I and PRS I1 were cloned into expression vectors derived from pKK233-2 (Fig. l ) , as described under "Experimental Procedures." For a high level expression (29), the cDNA fragments that did not contain A-T tails were inserted into the expression plasmids. The constructed pKrI and pKrII consisted of a highly inducible trc promoter, a ribosome-binding site, the PRS I and PRS 11 cDNAs, respectively, and a strong rrnB transcription terminator.
Expression of Rat PRS I and PRS 1 1 in E. coli Cells-The induction of rPRS I and rPRS I1 by IPTG in the bacterial strains MV1304/pKrI and MV1304/pKrII was examined by the enzyme assay on soluble fractions of the crude cellular lysates. The enzyme activity increased within 2 h after addition of IPTG and reached a maximum in 3-4 h. The specific enzyme activities of the extracts of the cells containing pKrI, pKrII, and pGlKHM as the negative control were 720, 400, and 20 milliunits/mg, respectively, at 4 h. A low level of enzyme activity detected in the control cells may represent the constitutive PRPP synthetase of the E. coli itself (11).
Purification of rPRS I and rPRS 11-Since homology of the deduced amino acid sequences of the E. coli and recombinant enzymes is high (47%), their separation may be difficult. Therefore, it is important to repress the synthesis of the bacterial PRPP synthetase, and this synthesis is regulated (30,31). We examined the enzyme levels in MV1304/pKrI and MV1304/pGlKHM cells grown at 30 or 37 "C in M9 minimal or LB medium. As shown in Table I, when the cells were grown at 30 "C in M9 medium, the cells containing the vector alone showed the lowest level of the enzyme activity, and the level in the cells containing pKrI was higher. Thus, the cells containing pKrI or pKrI1 were grown at 30 "C in M9 medium, and the rPRS I and rPRS I1 were purified from the soluble fractions of the extracts as described under "Experimental Procedures." The results are summarized in Table 11. The apparent increase in total activity after the first purification step of rPRS I and rPRS I1 is probably due to elimination of endogenous inhibitors. The bulk of the E. coli proteins was removed at the step of polyethylene glycol pre-

TABLE I1
Purification of rPRS I and rPRS II rPRS I and rPRS I1 were purified from transformed E. coli MV1304 cells as described under "Experimental Procedures." The cells were cultured in 10 liters of M 9 medium at 30 "C and harvested after a 3h induction with 1 mM IPTG.  cipitation. Samples from various steps of purification were analyzed by SDS-PAGE (Fig. 2). The purified rPRS I and rPRS I1 preparations gave a single band with a mobility identical with that of the 34-kDa subunit of the rat liver enzyme. rPRS I had a final specific activity of 33,400 milli-units/mg, and rPRS I1 had a slightly higher value of 46,200 milliunits/mg.
For assay of the activity of the purified enzymes, we used method 2, measuring [l4C]AMP production from [14C]ATP, as described under "Experimental Procedures." This approach can assay an exchange reaction between AMP and ATP, independent of PRPP synthesis (32). We examined the ["C] AMP-ATP exchange reaction under the assay conditions of method 2, except that 0.4 mM [I4C]ATP was replaced with unlabeled ATP and 0.05 mM [I4C]AMP was added. The purified rPRS I and rPRS I1 produced 0.055 and 0.013 nmol of [I4C]ATP, respectively, for 5 min, and these values were 1.9 and 0.4%, respectively, of [I4C]AMP production from ["C] ATP measured in parallel. Thus, the activities assayed here by method 2 were not a measure of the AMP-ATP exchange reaction.
NH2-terminal Sequence Analysis-The E. coli host cells possessed their own PRPP synthetase, as described above. The amount of the E. coli enzyme included in the rPRS I and rPRS I1 preparations was estimated by NH*-terminal sequence analysis. Analyses of the first 9 amino acids indicated that those preparations did not contain any detectable E. coli PRPP synthetase and that the initiator Met was not completely removed from the recombinant proteins. For example, at the third cycle of Edman degradation of rPRS I, we detected 114 pmol of Ile, 77 pmol of Asn, but no Met (less than 0.1 pmol), which should be derived from the E. coli enzyme (11).
For rPRS I, the major sequence found was Met-Pro-Asn-Ile-Lys-Ile-Phe-Ser-Gly, in agreement with the sequence deduced from the cDNA sequence, with a secondary sequence beginning with Pro. The relative yields of the two sequences indicated that the initiator Met was removed from 40% of the rPRS I protein. Sequencing of the rPRS I1 showed that the initiator Met was removed from 65% of the rPRS I1 protein.
Physical and Kinetic Properties of rPRS I and rPRS II-Some physical and kinetic properties of rPRS I and rPRS I1 were compared (Table 111).
The mammalian PRPP synthetases exist in multiple, highly aggregated states (33), and the purified rat liver enzyme has a molecular mass of over 1,000 kDa (15). The purified rPRS I and rPRS I1 were analyzed by HPLC gel filtration on TSK G4000SW. rPRS I was eluted as a broad single peak near the void volume, showing that the molecular mass was as high as that of rat liver PRPP synthetase. rPRS I1 was eluted as a sharp peak at 550 kDa.
The apparent Michaelis constants for ATP and ribose 5phosphate were determined. There was no marked difference between the apparent K,,, values for rPRS I and rPRS 11.
A remarkable difference was found between the heat stabilities of rPRS I and rPRS 11. rPRS I1 was much more labile than rPRS I; at 49 "C, the half-life of inactivation of rPRS I1 was 0.5 min, whereas the half-life of rPRS I was 90 min, 180fold greater.
Actiuation by Pi-PRPP synthetase is activated by P;, which has been accepted to play a primary role in the control of intracellular PRPP synthesis (33). Activation by Pi of rPRS I and rPRS I1 were compared. The two isoforms had no activity when assayed at Pi concentrations lower than 0.4 mM; therefore the Pi concentration-activity relationship showed a slight sigmoidicity. At higher Pi concentrations, activation of the two isoforms followed a hyperbolic curve. Double-reciprocal plots were linear at Pi concentrations from 1 to 50 mM. K, values for Pi of rPRS I and rPRS I1 were 1.8 and 2.4 mM, respectively. Thus, the two isoforms have a similar Pi dependency.

TABLE I11 Comparison of properties of rPRSI and rPRS II
The A:?,,, values were obtained by absorbance and dry weight measurements. For determination of the apparent K,,, values, assays were conducted at 37 "C in 10 mM potassium phosphate, 50 mM potassium HEPES buffer, pH 7.4. For the K , of ATP, the ribose 5-phosphate concentration was 0.2 mM, the MgC1, concentration was 4 mM excess over the concentration of ATP, and the ATP concentration was varied. For the K, of ribose 5-phosphate, the MgCl, concentration was 4 mM, the ATP concentration was 0.4 mM, and the ribose 5-phosphate concentration was varied. The concentrations of other components were fixed as described under "Experimental Procedures." The values were determined using a Lineweaver-Burk plot. The half-life of heat inactivation was determined at a protein concentration of 12 pg/ml in 50 mM potassium phosphate buffer (pH 7.4) containing the enzyme-stabilizing aaents. Sulfate ion can partially substitute for Pi, whereas at 50 and 100 mM, monovalent anions C1-, HCO;, and CH3COOwith K' as the cation had no activating effects. The effects of sulfate ion on rPRS I and rPRS I1 are considerably different. rPRS I gave a maximum activity at 60 mM K2S04 as the activator (Fig. 3), and the value was 43% of the activity with 50 mM Pi. In contrast, the rPRS I1 activity reached a maximum at 20-40 mM sulfate, and higher concentrations were inhibitory. The maximal activity was only 7% of that seen with Pi. Sulfate inhibited Pi activation of rPRS I1 in a competitive manner (with a I C i value of 5.7 mM). On the other hand, Pi activation of rPRS I was inhibited by sulfate but not in a simple competitive manner; increasing Pi concentrations could not completely overcome the suppressive effects of sulfate (data not shown).
Inhibition by Nucleotides-PRPP synthetases from various sources are inhibited by nucleotides, of which ADP is the most potent (12,(34)(35)(36). When we compared the inhibition of rPRS I and rPRS I1 by nucleotides (Table IV), ADP, GDP, and the reaction product AMP were seen to be the most effective. Other nucleotides only weakly inhibited both isoforms (0-19%). Noteworthy is the different sensitivity of rPRS I and rPRS I1 to inhibition by ADP and GDP; the inhibition of rPRS I and rPRS I1 by 0.3 mM ADP was 87 and Thus, rPRS I and rPRS I1 possess differential properties.
Since the 34-kDa component of the native rat liver PRPP synthetase is a mixture of PRS I and PRS 11, and the enzyme contains other polypeptides (15), we are now examining the combination of rPRS I, rPRS 11, and other components for catalytic and/or regulatory properties of the enzyme.

DISCUSSION
The catalytically active 34-kDa species of rat liver PRPP synthetase was partly isolated from 38-and 40-kDa components by gel filtration in the presence of 1 M MgClz (15). The 34-kDa component was a mixture of PRS I and PRS 11, and no means were available to separate the two. We expressed the cDNAs of rat PRS I and PRS I1 in E. coli, isolated each of the 34-kDa catalytic subunits, and studied some properties of the respective isoforms.
The recombinant isoforms were purified to apparent homogeneity by a three-step procedure (Table I1 and Fig. 21, free from the endogenous E. coli PRPP synthetase. The yield of purified rPRS I from a 10-liter culture was about 50 mg and that of purified rPRS I1 was 15 mg, in contrast to a low yield of the enzyme from rat liver (about 2 mg from 600 g of liver) (15). The specific activities of the purified rPRS I and rPRS I1 were 2.5-and 3.3-fold higher, respectively, than that of the most purified rat liver enzyme, in terms of the 34-kDa component (15). The lower activity of the rat liver enzyme might be due to inhibitory effects of the 38-and/or 40-kDa components in the aggregate. The NH2-terminal sequence analysis showed a partial removal of the initiator Met from rPRS I and rPRS 11, unlike the complete removal of the residue in the native enzymes of rat liver (15) and E. coli (11). Incompleteness of the removal of an extra Met has presented problems in case of other foreign proteins overexpressed in E. coli (37). Attempts to remove the extra Met using aminopeptidase M were unsuccessful, possibly due to obstruction by folded and aggregated forms of rPRS I and rPRS 11. The manner in which the extra Met affects the enzyme activity remains to be elucidated.
PRPP synthetases from mammals and bacteria show an absolute requirement for Pi for catalytic activity (33). The two isoforms also showed a similar Pi dependency. The new findings are that sulfate also acts as an activator and that the effects on rPRS I and rPRS I1 are considerably different (Fig.   3); the extent of activation of rPRS I by sulfate is much greater than that of rPRS 11. Furthermore, sulfate inhibited Pi activation in a different manner with rPRS I than with rPRS 11. Although the effects of sulfate may be of little physiological significance, the different responses of the two isoforms to the ion may be helpful in the approximate determination of the isoform composition of tissue PRPP synthetases. Differences between the two isoforms were also found in the sensitivity to inhibition by ADP and GDP. rPRS I was more sensitive to the two inhibitors than rPRS 11.
The two isoforms differ only by 13 deduced amino acid residues, 7 out of which are conservative substitutions (16). Regarding the structure-function relationships, the substitutions of Lys for Val and Lys for Gln at positions 4 and 152 are notable. These two substitutions give net additional positive charges to PRS I. The different sensitivity to nucleotide inhibition and/or the different responses to sulfate may be due to either or both of the substitutions.
Existence of two isoforms with different regulatory properties and their relative amounts that can vary with tissues in rats (21) suggest that PRPP synthetase has tissue-specific regulatory properties. Furthermore, attention should be given to the above possibility in studies on PRPP synthetase superactivity (38-40) as the molecular basis of human X-linked disease. Alterations in their relative amounts may produce various phenotypes of PRPP synthetase.
The important contribution of this work is the development of a simple and efficient system for overexpressing cDNAs. Since the expression vectors contain nine restriction cloning sites placed immediately downstream from the ATG initiation codon, a cDNA fragment can be conveniently inserted using an appropriate restriction site. Furthermore, an increased copy number of the vectors in E. coli cells was ensured by using the replication origin of pGEM-1 derived from pUC12 instead of pKK233-2 derived from pBR322 (41).