Expression of the Potato Tuber ADP-glucose Py~ophosp~ory~ase in Escherichia coli*

cDNA clones encoding the putative mature forms of the large and small subunits of the potato tuber ADP-glucose pyrophosphorylase have been expressed separately and together in an Escherichia coli B mutant deficient in ADP-glucose pyrop~osphorylase activity. Expression of both subunits from compatible vectors resulted in restoration of ADP-glucose pyrophosphorylase activity. Maximal enzyme activity required both subunits. The expressed ADP-glucose pyrophosphorylase was purified and characterized. The recombinant enzyme exhibited catalytic and allosteric kinetic prop- erties very similar to the enzyme purified from potato tuber. The expressed enzyme activity was neutralized by incubation with antibodies raised against potato tuber and spinach leaf ADP-glucose pyrophosphorylases but not with anti-Escherichia coli enzyme serum. 3-Phosphoglycerate was the most efficient activator and its effect was increased by dithiothreitol. In the ADP-glucose synthesis direction, 3-phosphoglycerate

regulation by small effector molecules, the levels of which vary during normal carbon metabolism in these organisms (1, 2). ADP-Glc pyrophosphorylase from enteric bacteria is activated by fructose-l,6-bisphosphate and inhibited by AMP (1). The enzyme found in the leaves of higher plants (2,3), green algae (3,4), cyanobacteria ( 5 ) , and several non-chlorophyllous plant tissues (2, 3, 6-9) is allosterically activated by 3PGA and inhibited by Pi. ~u b s t . a n t i a~ evidence now exists indicating that, in the leaf as well as in algae, the ratio of 3PGA to Pi within the chloroplast regulates starch synthesis by affecting ADP-Glc pyrophosphorylase activity (4, 10-12).
ADP-Glc pyrophosphorylase from a11 sources is found to be a tetrameric protein. However, the enzyme derived from enteric bacteria is homotetrameric (1, 21, whereas the enzyme from angiosperm plants is more complex in structure, being composed of two subunits that give rise to an a& heterotetrameric native enzyme ( 2 , 13,14). Since the bacterial and plant enzymes catalyze the same reactions, their structural dissimilarities may reflect their differences in allosteric specificity. A high structural homology between small subunits of higher plant ADP-Glc pyrophosphorylase has been reported (2,15, Is), with analysis of cDNA clones corroborating these observations (16). The large subunits of plant enzyme were the most divergent and shared less sequence identity (14,16). On the basis of this homology between them (15, 16), it has been speculated that the two plant subunits were originally derived from the same gene. During evolution there was gene duplication of the pyrophosphorylase gene followed by divergence of the genes to produce two polypeptides, both of which may be required for optimal activity ( 2 , 3,15,16). This idea is reinforced by recent evidence showing that the cyanobacterial enzyme, which has the higher plant allosteric specificity, is a homotetramer like the bacterial enzyme ( 5 , 17).
ADP-Glc pyrophosphorylase from potato tuber has been purified to homogeneity (14). The enzyme is composed of two different subunits of molecular mass 50 and 51 kDa (14). Kinetic studies showed that the potato tuber enzyme is regulated as a typical plant ADP-glucose pyrophosphorylase; thus, it is activated by 3PGA and inhibited by Pi (7,8). A major problem found for a complete characterization of this protein is that purification procedures results in low recoveries and the enzyme is quite unstable when isolated from potato (8,14). Recently, cDNA clones to the small (18-20) and the large (20) subunits of the potato tuber ADP-Glc pyrophosphorylase have been isolated and may provide new tools for a complete characterization of the potato tuber enzyme. In this work, we report on the expression of the cDNA clones in an Escherichia coli B mutant deficient in the pyrophosphorylase activit,y. The purified recombinant protein showed properties similar to the potato tuber enzyme.

EXPERIMENTAL PROCEDURES
Construction of pMON17335 and pMONl7336 The compatible E. coli vectors for the expression of the mature forms of the potato tuber ADP-glucose pyrophosphorylase subunits were completed as follows. A complete copy of the small subunit cDNA was first assembled by combining a portion of the genomic clone and the almost complete cDNA clone (20). A methionine (and alanine) residue were then engineered before the leucine at residue 83 (20) by polymerase chain reaction-based mutagenesis using the following oligonucleotide: GAATTCACAGGGCCATGGCACTT-GACCCAGATGC. The C terminus of the coding region was engineered with the following oligonucleotide to introduce a Sac1 restriction site sequence: CCAAGTTAAAACGGAGCTCATCAGATGAT-GATTCC. This putative mature small subunit coding sequence was then cloned as an NcoI-Sac1 fragment into a pACYC-based (21) vector to form pMON17335 (Fig. 1). The following oligonucleotide was used to introduce a methionine in the putative mature region of the large subunit: AAGATCAAACCTGCCATGGCTTACTCTGT-GATCACTACTG. This coding sequence was then cloned as an NcoI-Hind111 fragment (the latter site is located 105 base pairs after stop and expresses the mature form of the small subunit from the tac promoter (40) and the phage T7 gene10 leader (GIOL) translation enhancer (41). B, the plasmid pMON17336 is a derivative of pBR327 (22) containing aadA spectinomycin/streptomycin (SpclStr) resistant gene from Tn7 (42) and expresses the mature form of the large subunit from the PrecA and GlOL expression cassette described previously (41). ADPGPP, ADP-Glc pyrophosphorylase.
codon in the cDNA clone)' into a pBR327-based (22) vector to form pMON17336 (Fig. 1). The mutagenized cDNAs were resequenced to verify that no changes other than the expected had occurred.
Assay of ADP-Glc Pyrophosphoryluse The activity of ADP-Glc pyrophosphorylase was determined in pyrophosphorolysis (assay A) or synthesis (assay B) direction.
Assay A-Pyrophosphorolysis of ADP-Glc was followed by the formation of ATP from 3'PPi. The reaction mixture contained 20 pmol of glycylglycine buffer (pH 8.0), 1.25 pmol of MgCl', 0.75 pmol of DTT, 2.5 pmol of NaF, 0.5 pmol of ADP-Glc, 0.38 pmol of 32PPi (1.0 to 6.0 X lo6 cpm/pmol), 50 pg of crystalline bovine serum albumin, 1 pmol of BPGA, and enzyme in a final volume of 0.25 ml. The reaction was started by the addition of 3'PPi, and after 10 min incubation at 37 "C it was terminated by the addition of 3 ml of cold 5% trichloroacetic acid. The [32P]ATP formed was measured as described previously (13). A unit of ADP-Glc pyrophosphorylase activity is defined as that amount of enzyme catalyzing synthesis of 1 pmol of ATP/min under the reaction conditions described.  Ao.s,and 10.6 values, corresponding to the concentrations giving 50% maximal activity, activation, and inhibition, respectively, and Hill coefficients (nH) were calculated from Hill plots (24). All kinetic parameters are the mean of at least two determinations and are reproducible to within at least f 10%.

Antibody Neutralization
Neutralization of the ADP-Glc pyrophosphorylase activity was performed basically as previously described (9). About 0.05 unit of the partially purified recombinant potato enzyme was mixed with 3 pmol of Hepes-NaOH, pH 7.0, containing 10 pg of bovine serum albumin, 1.25 pmol of Pi, 0.1 pmol of DTT, 5 mg of sucrose, and 50 p1 of serum containing varying amounts of either anti-potato, antispinach leaf, or anti-E. coli ADP-Glc pyrophosphorylase immune serum diluted into the corresponding preimmune serum in a total volume of 0.1 ml. The mixture was incubated for 30 min at 30 "C and then for 2 h on ice prior to centrifugation for 5 min in Eppendorf microcentrifuge. Enzyme activity in the supernatant was measured by using assay A.

Protein Determination
Protein concentration was measured after Smith et al. (25) using Pierce prepared bicinchoninic acid (BCA) reagent and bovine serum albumin as the standard.

PAGE and Western Blotting
PAGE for the native enzyme was performed in the Ornstein-Davis system (26) using 7.5% and 3% acrylamide gels for resolving and stacking, respectively. SDS-PAGE was performed in 9% gels according to Laemmli (27). After electrophoresis, proteins on the gel were transferred onto nitrocellulose membranes as described by Burnette (28). Following electroblotting, nitrocellulose membranes were treated with mouse anti-potato tuber or affinity-purified rabbit antispinach leaf ADP-Glc pyrophosphorylase immunoglobulin G, and the antigen-antibody complex was visualized as previously described (5).
Expression of ADP-Glc Pyrophosphoryluse in E. coli E. coli strain AC70R1-504 containing plasmids pMON17335 and pMON17336 were inoculated from 15% stocks stored at -70 "C onto Luria broth (LB) plates containing 15 pg/ml kanamycin and 50 pg/ ml spectinomycin and cultured overnight at 37 "C. Single colonies 2T. W. Okita, P. A. Nakata, T. Greene, and M. J. Laughlin, unpublished observations. were inoculated into 4-ml LB cultures containing kanamycin and spectinomycin as above for 12 h before subculturing 750 p1 into 750 ml of fresh LB plus kanamycin and spectinomycin in 2.8-liter Fernbach flasks for overnight incubation at 37 "C. Late log phase cultures were used to inoculate MSCA medium (42 mM Na2HP04, 22 mM KH2P0, 8.6 mM NaC1,lS mM NH4Cl, 100 p M CaCh 2 mM MgSO,, 0.2% glucose, 0.2% casamino acids) at a rate of 1.5 liter of culture to 100 liters of MSCA in a Braun model 120 fermentor. Cells were grown for 7 h to ODsw = 0.738 prior to adding 100 p M isopropyl-0-Dthiogalactopyranoside and 5 gg/ml nalidixic acid to induce expression of the ADP-Glc pyrophosphorylase subunits. After 3 h of expression, cells were chilled to 10 "C by adding ice and harvested in a refrigerated Sharples centrifuge. Cell paste was kept overnight at -20 "C before use.

Purification of the Recombinant ADP-Glc Pyrophosphorylase
All steps were carried out at 0-4 "C. Assay A was used to monitor enzyme activity throughout the purification. Cell paste was resuspended in extraction buffer (about 5 ml of buffer/g of cells) containing 50 mM glycylglycine (pH 7.5), 5 mM MgC12, 1 mM EDTA, 5 mM DTT, and 20% sucrose. The suspension was disrupted by sonic oscillation in a Heat Systems Ultrasonic sonicator model W-220F and then centrifuged at 12,000 X g for 15 min. The pellet was washed once in additional buffer (about half the volume of the original homogenate) and centrifuged. The supernatants were combined (crude extract) and absorbed onto a DEAE-Sepharose fast-flow column (2.25 X 37 cm) that had been equilibrated with the extraction buffer supplemented with 5 mM potassium phosphate (pH 7.5). After washing with 5-bed volumes of the above buffer, the enzyme was eluted with a linear gradient consisiting of 4-bed volumes of the extraction buffer in the mixing chamber and 4-bed volumes of 50 mM potassium phosphate (pH 6.0), containing 2 mM DTT and 0.4 M KC1 in the reservoir chamber. The fractions containing ADP-Glc pyrophosphorylase activity were pooled and the protein precipitated by addition of crystalline ammonium sulfate up to 60% saturation.
The precipitate was collected by centrifugation, dissolved in a small volume of medium (buffer A) containing 20 mM Bis-Tris-propane buffer (pH 7.0), 5 mM potassium phosphate, 5 mM MgC12, 1 mM EDTA, 10% (w/v) sucrose, and 2 mM DTT, and then desalted by passing through Econo-Pac lODG columns (Bio-Gel P-6 desalting gel from Bio-Rad) equilibrated with buffer A. This sample (28 ml) was applied to a Mono Q HR10/10 column equilibrated with buffer A.
The column was washed with 40 ml of buffer A and eluted with a linear KC1 gradient (100 ml, 0-0.5 M) in buffer A. Fractions of 5 ml were collected, and those containing activity were pooled and then concentrated to 10 ml in an Amicon concentrator fitted with a PM 30 membrane. The concentrated enzyme fraction was diluted with an equal volume of 2 M potassium phosphate (pH 7.0) and then applied to an aminopropyl-agarose column (C, column, 1 X 12 cm) previously equilibrated with 1 M potassium phosphate buffer (pH 7.0) containing 2 mM DTT. The enzyme was absorbed, and the column was successively washed with 10-bed volumes of each of the following Pi buffers (pH 7.0): 1, 0.75, 0.5, 0.20, and 0.05 M. The enzyme was eluted with 0.75 M Pi buffer. Fractions containing ADP-Glc pyrophosphorylase activity were pooled and concentrated by Amicon ultrafiltration with a PM 30 membrane and then dialyzed for 12 h against 1 liter of the buffer used for Mono Q chromatography (buffer A). This fraction was stored at -80 "C, and the enzyme, under these conditions, was fully active for at least 3 months.

RESULTS
Plastid-targeted proteins are typically processed during import into the plastid by removal of a N-terminal transit peptide (29). T h e N terminus of the plastid-localized subunits has only been determined for the spinach leaf ADP-glucose pyrophosphorylase (13). By comparison of the aligned protein sequences of the ADP-glucose pyrophosphorylases from plant and bacterial sources (20), a putative mature N-terminal region has been determined for the subunits of the potato tuber enzyme. This assignment is also in agreement with our current understanding of the structural features of transit peptides. Following the addition of an initiator methionine, the putative mature subunit coding regions have been expressed separately and together (from compatible expression vectors) in E. coli. Crude extracts of E. coli AC70R1-504 cells transformed with the different plasmids were analyzed for ADP-Glc pyrophosphorylase activity. As shown in Table I, expression of the small (plasmid pMON17335) or large (plasmid pMON17336) subunit separately resulted in extracts containing low ADP-Glc pyrophosphorylase activity. Enzyme activity could be partially reconstituted by mixing equal amounts of each of the above extracts (Table   I). E. coli cells transformed with both pMON17335+17336 yielded extracts containing the highest ADP-Glc pyrophosphorylase activity (Table I).
The recombinant enzyme was found to exhibit a heat stability different from that of the native potato ADP-Glc pyrophosphorylase. The enzyme purified from potato remains fully active after 5 min incubation at 70 "C, and this treatment was utilized as a purification step in potato crude extracts (8,14). Under the same conditions, the recombinant potato ADP-Glc pyrophosphorylase is almost completely inactivated.
In order to characterize the recombinant ADP-Glc pyrophosphorylase in greater detail, it was purified from E. coli AC70R1-504 cells transformed with pMON17335+17336. In crude extracts the recombinant enzyme accounted for approximately 0.13% of the total soluble protein of the cells, based on the specific activity of the pure native potato enzyme determined previously (14). Table   I1 summarizes a typical purification of ADP-Glc pyrophosphorylase from 132 g (100liter culture) of E. coli transformed cells. The purification procedure resulted in a 952-fold purified enzyme with a specific activity of 63.8 units/mg and 50% recovery.
Native PAGE of the purified recombinant enzyme showed a major protein band (data not shown). In SDS-PAGE, two major protein bands of molecular masses about 50 and 80 kDa were observed. Immunoblot analysis of SDS-PAGE showed that only the broad band of molecular mass 50-52 kDa was recognized by anti-potato tuber and anti-spinach leaf ADP-Glc pyrophosphorylase immune sera, thus suggesting that it corresponds to the two subunits of the potato recombinant enzyme. Results suggest that the preparation is about 50% pure and it contains a major contaminant, which co-migrates with the recombinant enzyme in native PAGE. Altough the recombinant enzyme was partially purified, its specific activity (63.8 units/mg) was slightly higher than that reported for the enzyme purified from potato tuber (56.9 units/mg for an enzyme nearly 80% pure; see Ref. 14).
Kinetic and regulatory properties of the purified recombinant ADP-Glc pyrophosphorylase were determined in both pyrophosphorolysis and ADP-Glc synthesis directions. The results were compared with those previously reported for the potato tuber enzyme (7,8). Table I11 shows the effect of different metabolites on the physiological synthetic activity of the enzyme. 3PGA behaved as the most potent activator by increasing the enzyme activity over 30-fold. Fructose-1,6-    bisphosphate and phosph~nolpyruvate also activated the recombinant enzyme slightly but to a much lesser extent than 3PGA (Table 111). Neither 2-phosphoglycerate nor 2,3-bisphosphoglycerate were effective as activators. These results are in good agreement with those reported for the ADP-Glc pyrophosphorylase purified from potato tuber (7, 8). The recombinant enzyme was found less sensitive to Pi inhibition and, as shown in Table 111, the enzyme expressed in E. coli was slightly inhibited at 2 mM Pi, whereas the potato tuber pyrophosphorylase was reported to be 93% inhibited by 1 mM Antibody neutralization studies were carried out using antipotato tuber, anti-spinach leaf, and anti-E. coli ADP-Glc pyrophosphorylase immune sera. Fig. 2 shows that only the antibodies raised against the plant protein effectively neu- tralized the recombinant enzyme activity. The amounts of antiserum causing 50% inhibition were about 300 and 225 pi/ unit of enzyme for anti-po~to tuber and anti-spina^^ leaf serum, respectively. Interestingly, antibody for the E. coli ADP-Glc pyrophosphorylase (up to 1700 pllunit) caused no inhibition of the recombinant enzyme activity (Fig. 2).
The activation of the recombinant ADP-Glc pyrophosphorylase by 3PGA was studied in more detail. In the pyrophosphorolysis direction, 3PGA exhibited a hyperbolic pattern with a calculated A,, value of 45 p~ and a %fold maxima1 activation. DTT (4 mM) slightly decreased the Ao.s for 3PGA to 34 p~ without significantly affecting the degree of the activation. Fig. 3 shows that the recombinant enzyme was activated &fold by 3PGA in the ADP-Glc synthesis direction. Activation by different 3PGA concentrations followed hyperbolic kinetics with an Ao, value of 160 pM. The presence of 4 mM DTT increased the affinity of the recombinant enzyme toward the activator = 57 p~) and enhanced the maximal activation by nearly ?,-fold (Fig. 3) as observed with the potato tuber enzyme (8).
Effective inhibition by Pi of the ADP-Glc synthesis activity of the recombinant enzyme required of the presence of 3PGA in the assay medium and maximal inhibition values of only 20% were obtained with 4 mM Pi in the absence of 3PGA. The effect of Pi on the rate of ADP-Glc synthesis at three different 3PGA concentrations is shown in Fig. 4. Fifty percent inhibition occurred with Pi concentrations of 83, 320, and 680 PM in the presence of 10, 125, and 250 p~ SPGA, respectively. In all cases, the inhibition curve was s i~o i d a l , with nH values of 2.6 (10 p~ 3PGA), 1.6 (125 p~ 3PGA), and 3.0 (250 p~ 3PGA). When assayed in the pyrophosphorolysis direction, the enzyme was not inhibited by Pi, even in the presence of 3PGA.
To characterize the recombinant ADP-Glc pyrophosphorylase in greater detail, the kinetic parameters of the enzyme in the absence and in the presence of 2.5 mM 3PGA were determined. The results are compared with the values previously reported for the potato tuber enzyme in Table IV. As shown, 3PGA decreased the So,, value for ADP-Glc, PPi, and Glc-1-P of the recombinant enzyme, with a higher increase in affinity observed for Glc-1-P. The activator also changed the sigmoidal pattern exhibited by ADP-Glc to a hyperbolic saturation curve for this substrate. These results are, as a whole, in good agreement with the kinetic parameters previously reported for the potato tuber ADP-Glc pyrophosphorylase (7, 8). The main difference found for the recombinant enzyme is that it exhibited a higher apparent affinity toward Glc-1-P (Table IV). So.s values for Mg'+ were also determined for the potato enzyme expressed in E. coli. Under all the experimental conditions, the divalent cation saturation curve was sigmoidal. Table IV shows that in the ADP-Glc synthesis direction, 3PGA increased the affinity of the recombinant enzyme for Mg2+ by almost 2-fold.

DISCUSSION
We have constructed different plasmids containing cDNA encoding the putative mature primary sequence of the large and/or small subunit of the potato tuber ADP-Glc pyrophosphorylase. These plasmids were used to transform cells of E. coli B, strain AC70R1-504, a glycogen-less mutant deficient in ADP-Glc pyrophosphorylase activity (30). Only low levels of enzyme activity were obtained when either subunit was expressed alone. High enzyme activity levels were achieved when both expression vectors were simultaneously used to transform the E. coli cells. To the best of our knowledge, this is the first time that two dissimilar eukaryotic genes have been expressed together in E. coli rendering a functional enzyme.
The recombinant enzyme was purified and characterized and its properties compared with the potato tuber ADP-Glc pyrophosphorylase. The recombinant enzyme expressed in E. coli possesses immunologic, kinetic, and allosteric regulatory properties similar to the potato tuber enzyme. Thus, (i) the recombinant enzyme was recognized by polyclonal antibodies raised against plant ADP-Glc pyrophosphorylases but not by the anti-E. coli enzyme serum; (ii) kinetic parameters for Glc-1-P, ATP, ADP-Glc and PPi determined in this study were very similar to those reported for the enzyme purified from potato tuber (7, 8); and (iii) SPGA was the most important activator of the recombinant enzyme, the activation kinetics being in agreement with those found in the potato tuber pyrophosphorylase (7,8).
It is worth considering two main differences found between the recombinant enzyme and the native ADP-Glc pyrophosphorylases. One difference results from the low heat stability exhibited by the recombinant protein. The second is the lower inhibitory effect of Pi on the enzyme expressed in E. coli. It is tempting to speculate that both differences could be due to the same cause. Heat treatment at 60-70 "C has been successfully used as a purification step for ADP-Glc pyrophosphorylase from bacteria and plant ( 5 , 8,13,31). In many cases, it is necessary to add Pi (20-30 mM) to the medium prior to the heat treatment in order to adequately protect the enzyme from heat denaturation. Thus, the lower affinity of the recombinant ADP-Glc pyrophosphorylase for Pi could result in a reduced binding of this compound to the protein, unabling the anion to effectively protect the enzyme against thermal inactivation. Interestingly, Pi alone gave very low inhibition of the recombinant enzyme activity, suggesting that it binds poorly to the protein. The inhibitory effect of Pi observed when 3PGA was also present, and the fact that higher 3PGA concentration resulted in a higher for Pi, suggest some kind of interaction between both compounds in their binding to the allosteric sites of the enzyme. Thus, higher concentrations of SPGA can desensitize the enzyme to Pi inhibition. Taking into account that the small subunit cDNA in pMON17335 lacks sequences coding for about 10 residues of the N terminus or the possible presence of mutations produced during the cloning and isolation of the large and small subunit cDNA plasmids, it is quite possible that the expressed protein is lacking part of the Pi-binding site.
In the pyrophosphorolysis direction, the recombinant ADP-Glc pyrophosphorylase was activated 3-fold by 3PGA with an AOS value nearly 40 PM. This Ao, value is higher than that reported for the potato tuber enzyme (Ao.5= 5 p~; see Ref. 7). On the contrary, the kinetic parameters for 3PGA activation in the ADP-Glc synthesis direction showed a higher affinity of the recombinant enzyme toward the activator (Ao.s = 160 and 57 PM in the absence and in the presence of DTT, respectively) as compared with the potato tuber enzyme (Ao.,=