Site-directed Mutagenesis of the Phosphorylatable Serine iSers) in C4 Phosphoenolpyruvate Carboxylase from Sorghum THE EFFECT OF NEGATIVE CHARGE AT POSITION 8*

The properties of the dephospho and in vitro phosphorylated forms of recombinant sorghum phosphoenolpyruvate carboxylase have been compared with those of the authentic dark (dephospho) and light (phospho) leaf enzyme forms and two mutant enzymes in which the phosphorylatable serine residue (Sera) has been changed by site-directed mutagenesis to Cys (SSC) or Asp (SSD). Kinetic analysis of the purified recombinant, mutant, and leaf enzyme forms at pH 8.0 indicated virtually identical V,,,, apparent K,(phos-phoenolpyruvate), and half-maximal activa-tion(g1ucose 6-P) values of about 44 units/mg, 1.1 mM, and 0.23 mM, respectively. In contrast, the Sera, SSC, and dark leaf enzymes were about %fold more sensi-tive to inhibition by L-malate at

The properties of the dephospho and in vitro phosphorylated forms of recombinant sorghum phosphoenolpyruvate carboxylase have been compared with those of the authentic dark (dephospho) and light (phospho) leaf enzyme forms and two mutant enzymes in which the phosphorylatable serine residue (Sera) has been changed by site-directed mutagenesis to Cys (SSC) or Asp (SSD). Kinetic analysis of the purified recombinant, mutant, and leaf enzyme forms at pH 8.0 indicated virtually identical V,,,, apparent K,(phosphoenolpyruvate), and half-maximal activation(g1ucose 6-P) values of about 44 units/mg, 1.1 mM, and 0.23 mM, respectively. In contrast, the Sera, SSC, and dark leaf enzymes were about %fold more sensitive to inhibition by L-malate at pH 7.3 than the Sera-P, SSD, and light leaf enzyme forms. These comparative results indicate that: (i) Sera is an important determinant in the regulation of sorghum phosphoenolpyruvate carboxylase activity by negative (L-malate), but not positive (glucose 6-phosphate) metabolite effectors, (ii) phosphorylation of this target residue can be functionally mimicked by Asp, but not Cys, and (iii) negative charge contributes to the effect of regulatory phosphorylation on this C4-photosynthesis enzyme. Phosphoenolpyruvate carboxylase (PEPC,' EC 4.1.1.31) catalyzes the irreversible @-carboxylation of PEP by HCOT in the presence of Me2+ to yield oxalacetate and Pi, a reaction that serves a variety of physiological functions in higher plants. The cytoplasmic leaf enzyme is a homotetramer of -110-kDa subunits. During photosynthesis by C4 plants (e.g. sorghum and maize), PEPC is the initial carboxylating enzyme that fixes atmospheric C02 into C4-dicarboxylic acids (oxalacetate, malate, and aspartate), from which COz is subsequently released internally by various decarboxylating enzymes and photosynthetically reassimilated by the Calvin cycle (1, 2). While light-induced changes in the cytoplasmic levels of positive (glucose 6-P, triosephosphate) and negative (L-malate) metabolite effectors likely contribute to the overall regulation of C4 PEPC activity in vivo (1-3), recent attention has focused on the light/dark regulation of this enzyme's activity by posttranslational modification (4). I n vitro and in vivo studies have established that the light-induced phosphorylation of C4-leaf PEPC at a single serine residue (Sera in sorghum and Ser" in maize) results in a severalfold increase in the enzyme's apparent Ki for L-malate at suboptimal, but physiological levels of PEP and pH, without affecting K,,,(PEP) at (sub)optimal pH (reviewed in 4). These lightdependent changes in the regulatory properties and phosphorylation status of this cytoplasmic target enzyme are governed largely by a protein-serine kinase that is reversibly light-activated in vivo by an increasingly complex signaltransduction chain originating in the chloroplast (5)(6)(7)(8).
Protein phosphorylation has been long recognized as an ubiquitous mechanism by which metabolism is controlled by extracellular and intracellular signals. In contrast, only recently have crystallographic and site-directed mutagenesis approaches been exploited to provide insight into the molecular events by which regulatory phosphorylation alters the activity of an individual target enzyme or protein (9-15). For example, the effect of serine phosphorylation on enzyme activity may be due, at least in part, to the introduction of negative charge, since it has been mimicked by using in vitro mutagenesis to replace the regulatory Ser with Asp or Glu in several microbial and mammalian target enzymes (9, 10, 15). However, such functional mimicry is not universal (11, 14), e.g. the replacement of the phosphorylatable serine (Ser'33) in the CAMP-dependent response element binding protein with an acidic residue does not duplicate the effects of phosphorylation of Ser'33 by protein kinase A (11). In the present work, we have investigated the effects on PEPC properties resulting from the replacement of Sera with either Cys (S8C) or Asp (S8D) in the recently described (16) recombinant sorghum enzyme. The results support the hypothesis that negative charge contributes to the effect of phosphorylation on the regulatory properties of this photosynthetic target enzyme.

16759
Pharmacia LKB Biotechnology Inc., unless noted otherwise. Enzymes for the recombinant DNA techniques were obtained from Appligene, Pharmacia, or BRL and used according to the manufacturer's instructions. All radiochemicals were from Amersham Corp.
Leaf PEPC was purified to apparent electrophoretic homogeneity from light (3 h at -600 pmol of photons m-' s-') and dark (12 h) adapted plants of greenhouse-grown sorghum (Sorghum uulgare, cv. Tamaran) as previously described for the maize enzyme (17,18). The PEPC protein-serine kinase preparation was partially purified from light-adapted (3 h) maize (Zea mays L.) leaves by a modification' of our previous procedure (18).
Bacterial Strains, Plasmids, and Media-Escherichia coli strain DH5aF' was used to produce high yields of plasmids and M13 singlestranded and replicative forms. A PEPC-deficient mutant of E. coli (strain XH11; see Ref. 16) was used for the production of the wildtype and mutant PEPCs encoded by recombinant pKK233-2 vectors (16) under the control of a trc promoter. E. coli strain RZ1032 (19) was used to produce dU-substituted DNA template for oligonucleotide-directed mutagenesis (19,20). Bacteria were grown at 37 "C on LB medium (21); ampicillin (50 pg/ml) was included when the bacteria carried plasmids conferring drug resistance.
Cloning of the Wild-type and Mutant PEPC cDNA-All methods involving preparations of DNA and transformation of bacterial cells were as described (21). The full-length cDNA coding for the sorghum leaf PEPC (PEPC cDNA) was isolated from the pCP310 plasmid (16) using SmaI-Hind111 restriction enzymes. This -3.2-kilobase pair DNA fragment was recovered by agarose gel electrophoresis (22). The fragment was first transferred into M13 phage vector (M13 310) to remove an internal NcoI site (see Ref. 16) in a conservative manner by mutagenesis with the L1 oligonucleotide 5"CAGGACCTA-CATGGCCC-3'. The resulting M13 311 construct was used to generate single-stranded DNA for the site-directed mutagenesis experiments. The mutagenic oligonucleotides L2 (M13 312, GCAC-CACTGCATCGACG) and L3 (M13 313, GCACCACMCATCG-ACG) were used to substitute Sera with either Cys (S8C) or Asp (S8D), respectively. Mutations were introduced into dU-substituted DNA template by the method of Kunkel (20). Uracil-containing single-stranded M13 DNA (200 ng) was annealed with 5 pmol of phosphorylated oligonucleotide designed to replace the original codon for Ser (TCC) with that for Cys (TGC) or Asp ( M C ) . After in vitro polymerization, phagemids were used to transform the DH5aF' E. coli strain. Mutagenesis frequencies were sufficiently high (50-60%) to allow identification of mutants by sequence analysis. Singlestranded templates from a few plaque isolates were prepared and sequenced through the mutagenesis site by the dideoxynucleotide chain-termination method (23) using a T7 sequencing kit (Pharmacia). The corresponding cDNAs were then subcloned into the NcoI/ Hind111 restriction sites of the pKK233-2 vector (pKK 311, Sera PEPC; pKK 312, mutation S8C; pKK 313, mutation S8D) for expression of recombinant PEPC in the PEPC-deficient X H l l strain of E. coli (see Ref. 16 for details). To confirm that no other nucleotides within the PEPC cDNA were changed during mutagenesis, the entire sequence of the PEPC cDNA fragments was verified by double-strand sequencing.
Growth and Haruesting of Cells-Complemented cells producing the wild-type and mutant forms of the sorghum enzyme were grown a t 37 "C with orbital shaking in LB medium (21), pH 7.4, supplemented with 50 pg/ml ampicillin for 10-12 h ( A~w = 1.5). The cells from the 1-liter cultures were collected by centrifugation, resuspended, and washed twice in 0.1 M Mops-KOH, pH 7.3,lO mM MgC12, 5 mM L-malate (pH adjusted), and the pellets (-4 g) were stored at -80 "C until used.
Purification of the Wild-type and Mutant PEPCs-AI1 procedures were performed at 4 "C, and all chromatographic separations were done by using a Pharmacia FPLC system. The washed cells (3-6 g wet weight) were resuspended in 10 volumes of 0.1 M Mops-KOH, pH 7.3, 10 mM MgClz, 5 mM DTT, 5 mM L-malate (pH adjusted), 1 mM fresh phenylmethylsulfonyl fluoride, 100 pg/ml chymostatin, 10 pg/ml leupeptin, 10 pg/ml pepstatin A, 10 pg/ml E-64, and disrupted by three passes through a chilled French pressure cell at -20,000 psi. The lysate was clarified by differential centrifugation at 27,000 X g for 10 min and 50,000 X g for 25 min. The second supernatant fluid was diluted to 100 ml with extraction buffer and fractionated with (Buffer A), and the clarified solution applied to a column (2.5 X 6 cm) of Bio-Gel HTP hydroxylapatite (Bio-Rad) equilibrated with Buffer A. After thorough washing, the column was eluted with a 150ml linear gradient of 0-0.4 M KHzP04/K,HP04, pH 7.5, 0.1 mM EDTA, 1 mM DTT, 5 mM L-malate, 10% (v/v) glycerol at a flow rate of 1.5 ml/min. The fractions eluting at -200 mM Pi contained the majority of the PEPC activity-units. Alternatively, the PEG pellets from the wild-type and Cys' mutant cell extracts were redissolved in Buffer A, pH 8.4, containing 5 mM DTT, applied to a Q-Sepharose Fast Flow column, and eluted with a linear gradient of 0-0.5 M NaCI. The fractions centered around 220 mM NaCl were pooled, brought to 30% saturation with (NH&S04 (BRL), and applied to a phenyl-Sepharose CL-4B column (24). As with the authentic sorghum leaf enzyme (24), PEPC activity peaked around 0 M (NH4)ZSOI.
The PEPC-containing fractions from the hydroxylapatite or hydrophobic columns were pooled and precipitated by 70% saturation (NH&S04. The centrifuged suspension was dissolved in 20 mM Tris-HCl, pH 8.0, 0.1 mM EDTA, 5 mM DTT, 5% (v/v) glycerol, 2 mM Lmalate (Buffer B), and desalted on a Sephadex G-25 column equilibrated with Buffer B. The protein eluant was then applied at a flow rate of 0.5 ml/min to a Mono Q HR 5/5 column equilibrated with Buffer B, and PEPC eluted by a 20-ml linear gradient of 0-0.35 M NaCl in Buffer B at a flow rate of 1 ml/min. The peak PEPC activityfractions, eluting at -180 mM NaCl, were pooled, diluted with an equal volume of Buffer B, and rechromatographed on the Mono Q column. For any PEPC preparation that was judged to be <95% pure at this stage by SDS-PAGE, the peak activity-fractions were concentrated and applied to a Superose 12 HR 10/30 column equilibrated with Buffer A, pH 8.4, containing 5 mM DTT and 50 mM NaCl.
Protein concentrations were determined as described previously N-terminal Sequence Analysis-The purified wild-type and mutant PEPCs were desalted thoroughly against distilled water, lyophilized, and then covalently attached by their free amino groups to 8-mm diameter Sequelon-DITC membrane disks (Millipore Corp.) according to Ref. 27. Following incubation for about 10 h at 25 "C in 1 ml of methanol containing 6 pl of HCI, the immobilized samples were subjected to covalent protein microsequence analysis (27) on a Millipore 6600 ProSequencer system.

RESULTS AND DISCUSSION
Purification and Properties of Wild-type and M u t a n t PEPCs-The three recombinant enzymes were purified >95% (Fig. lA) by procedures adapted from our previous work on sorghum and maize leaf PEPC (17, 18,24, 25). In the case of the mutant S8C enzyme, final purification was often only achieved by inclusion of a size-exclusion step on Superose 12 in the routine protocol. The final yield of electrophoretically pure enzyme (Fig. lA) was about 150 and 500 pg of PEPC/3 g (wet weight) of S8C and S8D or wild-type (Sers) E. coli cells, respectively.
Consistent with our recent observations on the immunopurified Se? recombinant sorghum PEPC (16), the wild-type and mutant enzymes used in the present study had an iden- When the three purified recombinant enzymes were subjected directly to covalent protein microsequence analysis, the Nterminals were blocked, presumably due to the presence of m e t a t position 1. However, following deblocking by overnight incubation of the immobilized wild-type protein in acidic methanol, the sequence of ASERHHSIDAQLR.. . was obtained, which corresponds exactly to positions 2-14 in the authentic leaf enzyme (17, 29). Similar treatment of the immobilized mutant PEPCs yielded an identical N-terminal sequence except for the expected Asp and Cys substitutions of the phosphorylatable Ser in the S8D and S8C enzymes, respectively. Taken together, these results not only directly establish the nature of the Sera replacements, but also indicate that the general properties of the three recombinant PEPCs closely approximate those of the authentic sorghum and maize leaf enzymes described previously (17, 24, 25, 28).
In Vitro Phosphorylation-Incubation of the purified recombinant Sera enzyme with Mg-ATP and the C4-leaf PEPC protein-serine kinase resulted in the incorporation of "P from y-labeled ATP into the -110-kDa subunit (Fig. lB, lane 2) and, as expected (6,8,16,18), a concomitant increase in the enzyme's Io.5 value for L-malate (Fig. 2). In contrast, neither the S8C nor S8D enzyme was phosphorylated in vitro by the C4-leaf protein-serine kinase (Fig. lB, lanes 4 and 5 ) . These findings with the recombinant sorghum PEPCs further document that Sera (or its structural homolog in maize PEPC, S e P ) is the only phosphorylatable residue in the target enzyme, both in uitro and in vivo (17, 30, 31).
Enzymatic and Regulatory Properties-Kinetic analysis of the purified Sera, S8D, and S8C enzymes at pH 8.0 indicated    (Table I). Notably, these values closely approximated those obtained with the in vitro phosphorylated wild-type enzyme and the authentic light (phospho) and dark (dephospho) forms of PEPC purified from sorghum leaves and assayed under identical conditions (Table I and Ref. 24). These findings confirm previous indications that the nature of the specific N-terminal residue at position 8 (sorghum PEPC) or 15 (maize PEPC) in the C4-leaf enzyme (e.g. Ser, Ser-P, Asp, Cys) has little influence on its activation by glucose 6-P (32) or its apparent affinity for PEP at pH 7.3 or 8.0 (24, 25). Similarly, the specific activity of a proteolyzed form of maize PEPC purified by conventional procedures is still high (-25 units/mg at pH 8.0 and 25 "C), yet the isolated protein starts at LeuZ3 (33).
In contrast to the virtually identical kinetic constants summarized in Table I, there was marked variation in the 10. 5 value for L-malate depending on the nature of the specific residue at position 8 in sorghum PEPC (Fig. 2). Whereas the recombinant Sera and S8C enzymes had IOa values of about 0.14 mM at pH 7.3 and 2.5 mM total PEP, the S8D mutant closely approximated the wild-type PEPC phosphorylated in vitro by the C4-leaf protein-serine kinase (about 0.45 and 0.65 mM, respectively). Similarly, the purified dark (dephospho) and light (phospho) in vivo forms of the sorghum leaf enzyme had Io&malate) values of about 0.15 and 0.40 mM, respectively, under identical assay conditions (data not shown). These latter values are comparable with those reported previously for the purified dark and light maize enzyme 'forms (17, 25). Thus, these collective findings indicate that the recombinant Sera and S8C PEPCs closely approximate the dark (dephospho) form of the authentic C,-leaf protein, while the S8D and Ser8-P enzymes are similar to the in vivo light (phospho) enzyme form.
Concluding Remarks-Modification of the N-terminal region of sorghum (or maize) PEPC by in vivo or in vitro phosphorylation of Ser8 (or Ser") has a significant effect on the enzyme's apparent Ki for L-malate, without markedly affecting K,,,(PEP), V,,,,,, or its activation by glucose 6-P (Table I, Fig. 2, and Refs. 4, 6, 8, 16-18, 24, 25, 32). The finding that substitution by Asp at position 8, which has a negative charge but not the shape of phosphoserine, causes a similarly specific increase in apparent Ki(L-malate), whereas Cysa does not (Fig. 2), argues that the introduction of negative charge at this position in sorghum PEPC, either by phosphorylation or mutagenesis, is somehow involved in the decreased sensitivity of the enzyme toward its negative effector. Similar cases of functional mimicry by substitution of an acidic residue for a phosphorylatable serine have been reported with some target proteins (9, 10, 15), but not others (11,14). In the case of the recombinant sorghum PEPCs, the slight, but reproducible difference in values between the S8D and in uitro phosphorylated Ser' enzymes ( Fig. 2) may simply reflect the density and spatial distribution of the dianionic charge of a phosphoserine group at position 8, as contrasted to the monoanionic aspartic acid side chain (15).
One possible explanation for the decreased sensitivity of the light (phospho) form of sorghum or maize PEPC to inhibition by L-malate is that the introduction of negative charge by reversible phosphorylation of the target serine residue (or the conversion of Se9 to Asp') hinders this Nterminal domain's interaction with the rest of the enzyme monomer. Presumably, this less "efficient" interaction alters the conformation of PEPC in a specific manner such that the putative binding site for L-malate now possesses decreased affinity for this negative effector, while the catalytic (PEP) and activator (glucose 6-P) sites are unaffected. Consistent with this speculative view are our findings with the truncated Leuz3 form of the purified maize leaf enzyme which possesses an abnormally high 10.5 value for L-malate (-1.9 mM) under our standard assay condition^,^ but a typical specific activity (33). Alternatively, the N-terminal region of C4 PEPC, including the target serine, could form part of the binding site for L-malate, which is perturbed by the introduction of negative charge or N-terminal proteolysis.