Structural and biochemical evidence of the glucose 6-phosphate-allosteric site of maize C 4 phosphoenolpyruvate carboxylase: Its importance in the overall enzyme kinetics

Activation of phosphoenolpyruvate carboxylase (PEPC) enzymes by glucose 6-phosphate (G6P) and other phospho-sugars is of major physiological relevance. Previous kinetic, site-directed mutagenesis and crystallographic results are consistent with allosteric activation, but the existence of a G6P-allosteric site was questioned and competitive activation—in which G6P would bind to the active site eliciting the same positive homotropic effect as the substrate PEP—was proposed. Here we report the crystal structure of the PEPC-C 4 isozyme from Zea mays with G6P well bound into the previously proposed allosteric site, unambiguously confirming its existence. To test its functionality, Asp239— which participates in a web of interactions of the protein with G6P—was changed to alanine. The D239A variant was not activated by G6P but, on the contrary, inhibited. Inhibition was also observed in the wild-type enzyme at concentrations of G6P higher than those producing activation, and probably arises from G6P binding to the active site in competition with PEP. The lower activity and cooperativity for the substrate PEP, lower activation by glycine and diminished response to malate of the D239A variant suggest that the heterotropic allosteric-activation effects of free-PEP are also abolished in this variant. Together, our findings are consistent with both the existence of the G6P-allosteric site and its essentiality for activation of PEPC enzymes by phosphorylated compounds. Furthermore, our findings suggest a central role of the G6P-allosteric site in the overall kinetics of these enzymes even in the absence of G6P or other phospho-sugars, because of its involvement in activation by free-PEP.


Introduction
Phosphoenolpyruvate carboxylases (PEPCs, EC 4.1.1.31) are mainly bacterial and plant enzymes that catalyze the formation of oxaloacetate and inorganic phosphate from phosphoenolpyruvate (PEP) and bicarbonate, in an essentially irreversible carboxylation reaction that requires Mg +2 ions. This reaction is the highly regulated first step in the assimilation pathway of atmospheric CO 2 in C 4 plants, such as maize. At physiological pH, the PEPC isozyme from leaves of these plants (PEPC-C 4 ) is allosterically activated by phospho-trioses or phospho-hexoses, such as D-glucose-6-phosphate (G6P), and inhibited by carboxylic acids such as L-malate or L-aspartate. Additionally, PEPC-C 4 isozymes from grasses are activated by glycine, serine or alanine. Activation by phospho-sugars constitutes a physiologically important kinetic property of PEPC enzymes, not only of those from plants, regardless of being of the C 3 , C 4 or CAM types, but also from bacteria [reviewed in 1,2]. In addition to phospho-sugars, other phosphorylated compounds were also found to behave as activators of PEPC enzymes [3][4][5]. Among these, of particular kinetic and possibly physiological relevance is the free species, i.e. non-complexed with Mg 2+ ions, of the substrate phosphoenolpyruvate (free-PEP) [6][7][8]. Thus, PEP itself behaves as both a substrate (as the complex PEP-Mg 2+ and as free-PEP) and as an activator (free-PEP).
The allosteric nature of the activation of PEPC enzymes by G6P and other phosphorylated compounds has been widely accepted on the basis of a number of kinetic [4][5][6][7][8][9][10][11][12], crystallographic [13], docking [14] and site-directed mutagenesis studies [15,16], until Schlieper et al. [17] questioned the existence of a G6P-allosteric site proposing instead competitive activation, i.e. that G6P activation of PEPC arises from its binding to the active site where it simulates the positive homotropic allosteric effects of the substrate PEP, rather than from the positive heterotropic effects resulting from its binding to an allosteric site. Their proposal was supported by the crystal structure of the Flaveria trinervia C 4 isozyme (FtPEPC-C 4 ) (PDB code 4BXC) in which G6P is bound into the active site and not into the previously proposed G6P-allosteric site, where a sulfate anion from the Downloaded from https://portlandpress.com/biochemj/article-pdf/doi/10.1042/BCJ20200304/882310/bcj-2020-0304.pdf by guest on 28 May 2020 Biochemical Journal. This is an Accepted Manuscript. You are encouraged to use the Version of Record that, when published, will replace this version. The most up-to-date-version is available at https://doi.org/10.1042/BCJ20200304 crystallization medium was found instead, leading them to propose this site as a sulfate-binding site of unknown physiological significance. Additionally, these authors reported a higher degree of activation by G6P at low PEP concentrations, which was interpreted as supporting evidence of the competitive activation suggested by them [17].
Knowing for certain the actual nature of the mechanism of PEPC activation by G6P and other phosphorylated compounds, i.e. allosteric activation or competitive activation, is of great importance because each of these two mechanisms would have different physiological effects. Indeed, it can be expected that competitive activation would be very inefficient and even could be detrimental for the plant under physiological conditions. Moreover, a model of competitive activation cannot explain the reported activation by free-PEP. Therefore, the nature of the activation by G6P and related phosphorylated compounds should be established without any doubt to fully understand the functioning of PEPC enzymes in vivo and to interpret the kinetic results obtained in vitro. Thus, with the aim of further investigating this important issue and solving the current controversy regarding the mechanism of activation of plant PEPC enzymes by G6P and other phosphorylated compounds, we obtained a crystal structure of ZmPEPC-C 4 in which, for the first time, a G6P molecule was found to be bound in a crevice in the subunit/subunit interfaces of the dimeric units of the native tetramer, thus confirming the previously proposed location of the G6P-allosteric site. In addition, we tested the functionality of this allosteric-site by generating and kinetically characterizing the ZmPEPC-C 4 D239A variant, which was not activated but inhibited by G6P. Our data also are consistent with the mutation causing the loss of activation by free-PEP, which has an important impact on the overall enzyme kinetics even in the absence of activators.

ZmPEPC-C 4 expression, purification, site-directed mutagenesis, enzymatic assay and kinetic data analyses
Downloaded from https://portlandpress.com/biochemj/article-pdf/doi/10.1042/BCJ20200304/882310/bcj-2020-0304.pdf by guest on 28 May 2020 Biochemical Journal. This is an Accepted Manuscript. You are encouraged to use the Version of Record that, when published, will replace this version. The most up-to-date-version is available at https://doi.org/10.1042/BCJ20200304

BL21-CodonPlus(DE3)-RIL E. coli cells (Novagen) transformed with a pET32a(+) (Novagen)
plasmid containing the full sequence of the gene coding for ZmPEPC-C 4 (accession number CAD60555) were used to overexpress the recombinant protein with an N-terminal His-tag, as reported [18]. The ZmPEPC-C 4 protein was purified using the Protino nickel-Tris(carboxymethyl)ethylene diamine resin (Macherey-Nagel, Germany) and then digested with enterokinase (New England Biolabs) following the procedure described in González-Segura et al. [18]. Protein concentration was determined spectrophotometrically by A 280 , using the extinction coefficient of 111,730 M -1 cm -1 per subunit predicted on the basis of the amino acid sequence by ExPASy ProtParam [19].
To generate the ZmPEPC-C 4 D239A variant, the forward and reverse primers 5' CCCCGCAGGcCGAAATGCGCTATGGGATG 3' and 5' GCATTTCGgCCTGCGGGGTGGGTTGTG 3', respectively, were used. The above-mentioned plasmid was used as the template for the site-directed mutagenesis by PCR, using the QuickChange II site-directed mutagenesis kit (Agilent Technologies) and following manufacturer's instructions. The DNA was sequenced to confirm that the desired mutation was present and that no other nucleotide changes occurred. Finally, the variant protein was overexpressed in BL21-CodonPlus(DE3)-RIL E.
coli cells (Novagen). The overexpression and purification of both the recombinant wild-type and D239A ZmPEPC-C 4 proteins was achieved as reported [18]. to give the desired concentrations of the free and Mg 2+ -complexed species were calculated as described in [7]. Enzyme activity in the presence of G6P, F6P or glycine was assayed at fixed 3 mM total-PEP; in the presence of malate at 2.5 mM total-PEP and 5 mM total-Mg 2+ . Initial velocities are expressed in units (mol of product formed per minute under our experimental conditions.). Each point shown in the figures is the average of duplicate or triplicate determinations. ORIGIN software (OriginLab Corporation) was used for kinetic data analysis by nonlinear regression and display.
When activation was observed, saturation data were fitted to equation (1). When inhibition was observed, saturation data were fitted to equation (2). Substrate-saturation data were fitted to equation where v is the experimentally determined initial velocity; v 0 is the experimentally determined initial velocity in the absence of the allosteric effector; v a max is the estimated maximum activity at saturating

Structure determination and model refinement
The data were integrated using XDS [21] and scaled and truncated with programs from the CCP4 suite [22]. Diffraction phases were determined by molecular replacement with the program Phaser [23] using the coordinates of the ZmPEPC-C 4 PDB code 5VYJ as starting model. Alternating cycles of automatic and manual refinement were carried out with the standard protocols of Phenix [24], monitoring the R work and R free split during the whole process. Phenix was also used for atomic positions and atomic displacement parameters. The simulated annealing omit map was calculated with Phenix following simulated annealing refinement at 3,000 K. The program Coot [25] was used to analyze the electron density maps (2F o -F c and F o -F c ). Structural figures and alignments were performed with Coot and PyMOL. The tetramer was generated through crystallographic symmetry operations in PyMOL.

Docking simulations
After removing the G6P bound into the G6P-allosteric site, flexible molecules of F6P, DHAP, or PEP were docked in the rigid ZmPEPC-C 4 6MGI crystal structure. Docking simulations were performed using AutoDock Vina [26] and a box of 15 Å centered on Glu369. In each simulation experiment, the top then poses were visually inspected to select the one that fulfilled the criteria of a proper position of the phosphate group. The best solution was subjected to a 1000-step energy-minimization process using the Amber force field parameters in the UCSF Chimera program [27].

Sequence analysis of PEPC enzymes
We searched PEPC orthologs from plants and bacteria, performing a blastp search on the RefSeq collection of the National Center for Biotechnology Information (http://www.ncbi.nlm.nih.gov) protein database as described [18]. The amino acid sequence of ZmPEPC-C 4 (GenBank accession CAD60555) was used as a query for PEPC-C 4; that of ZmPEPC-C 4 for PEPC-C 3 (GenBank accession PWZ12751), and that of PEPC from Escherichia coli (EcPEPC) (GenBank accession CAA29332) for bacterial sequences. All retrieved plant sequences belong to Viridiplantae. Plant PEPC-C 4 amino acid sequences were chosen based on the criteria of having serine at position equivalent to ZmPEPC-C 4 780 [28] and belonging to a C 4 plant. All sequences that in addition to fulfill these criteria included the residues analyzed were considered in our study. Representative PEPC-C 3 sequences were retrieved from C 3 plants belonging to 32 different plant families and selected with the criterion of having an alanine at position equivalent to ZmPEPC-C 4 780. Also, care was taken that their sequence identity varied between 83 and 74 %. Only two complete PEPC sequences were retrieved from two CAM plants. Since it was not clear to us whether they were of the C 3 -type or CAM-type, they were included in the analysis with the PEPC-C 3 sequences, although one of them has serine at position 780 and therefore it probably corresponds to a PEPC-CAM type isozyme. Given the great number of bacterial genomes, the analysis of bacterial PEPC sequences was performed using only sequences from ten representative organisms of each of the phyla Actinobacteria, Cyanobacteria, Firmicutes and Proteobacteria. To ensure the diversity of the sample analyzed, the degree of identity of the selected sequences with that of EcPEPC ranged from 79 to 31 %. The conservation of residues of the G6Pallosteric site was investigated by multiple amino acid residues alignments using ClustalX2 [29] and Downloaded from https://portlandpress.com/biochemj/article-pdf/doi/10.1042/BCJ20200304/882310/bcj-2020-0304.pdf by guest on 28 May 2020 the alignments were refined by hand. Sequence logos were constructed using WebLogo 3 (http://weblogo.threeplusone.com) [30,31].

Results and Discussion
The ZmPEPC-C 4 6MGI structure in complex with G6P, glycine and phosphoglycolate The crystal structure of ZmPEPC-C 4 with the activators G6P and glycine bound into their respective allosteric sites and the PEP-analog phosphoglycolate (PGA) bound into the active site was solved in space group C222 1 . This space group was the same as the previously seen for the ZmPEPC-C 4 structure with a sulfate anion bound in the putative G6P-allosteric site and no ligand bound in either the neutral amino acids-allosteric site or the active site (PDB code 1JQO) [13], but different from that of the ZmPEPC-C 4 structure with glycine bound into its allosteric site, no ligand in the G6P-allosteric site and acetate in the active site (PDB code 5VYJ) [18]. The new structure reported here was refined to a resolution of 2.98 Å. All data collection and refinement statistics are given in Table 1.
Overall, the fold of the protein polypeptide was the same as those described for the two others already reported ZmPEPC-C 4 structures mentioned above. The asymmetric unit is a dimer ( Figure   1A) from which the tetramer was constructed by crystallographic symmetry (Figure 1B). Besides the identification of the G6P-allosteric site, which will be described below, the 6MGI crystal structure permitted the observation of important properties of the active site, where an additional electron density consistent with the substrate analog PGA-added to the enzyme preparation previously to the crystallization procedure-was observed. This electron density was tentatively modeled with various components of the crystallization cocktail (G6P, glycine, ethylene glycol and glycerol), but a PGA molecule with occupancy of 0.7 gave the best fit. As shown in Figure 2A, the phosphate group of PGA interacts with the sidechains of Arg456, Glu566 and His177, which is the essential catalytic His residue. The carboxylate group is hydrogen bonded to the with the phosphate group of PGA is similar to that observed with one of the carboxylate groups of the inhibitor malonate bound into the active site of the archaeal-type PEPC of Clostridium perfringens [32]. This interaction probably takes place with PEP as well-as suggested by docking simulations using the 1JQO [13] and the 6MGI (not shown) structures-and probably is of great mechanistic relevance given the role of His177 in the PEPC reaction as the general base that stabilizes the carboxyphosphate intermediate and promotes carboxylation [33,34]. Thus, His177 appears to be correctly positioned inside the active site of 6MGI, a well as in those of 1JQO and 5VYJ, which show the same conformation of the active site (Supplementary Figure S2) despite of the fact that 1JQO has no ligand bound and that 5VYJ has acetate bound.
In the 6MGI structure, a glycine molecule was found bound into the neutral amino acidsallosteric site with occupancy of one ( Figure 2B). The conformation of this allosteric site, even in the absence of ligand as in 1JQO, is conserved [18] and similar to that observed in 6MGI (Supplementary Figure S3A); the interactions that glycine makes with protein residues in 6MGI are identical to those observed in the 5VYJ structure previously reported [18] (Supplementary Figure   S3B).
In contrast with the well formed conformations of the active site and of the allosteric sites for the two kinds of activators, which are the correct ones for carrying out the reaction or binding their respective activators, in 6MGI the inhibitory carboxylic acids-allosteric site is partially unformed and appears unable to bind the inhibitor malate, given the position of residues that participate in binding the inhibitor in other known PEPC structures [35,36] ( Figure 2C). The 1JQO and 5VYJ structures have a conformation of the carboxylic-acids allosteric-site essentially identical to that of 6MGI (Supplementary Figure S3, C and D). Interestingly, in spite of being present during crystallization, malate was not observed bound inside this allosteric-site in 1JQO [13], which is in accordance with not being properly formed. which is known to be critical for the recognition and binding of carbohydrates by proteins [37]. The length of the CH- bonds is within the range expected for these interactions [37]. The binding of G6P significantly contributes to the stabilization of the subunit/subunit interface of the dimeric units of the enzyme by bridging the two subunits through the hydrogen bonds that the activator makes with residues of opposing subunits. In addition, some of the residues of opposing subunits that form part of the G6P-allosteric site interact with other residues of the opposing subunit, thus not only participating in the allosteric regulation by G6P but also directly contributing to the stability of this interface. To explore whether other reported activators of ZmPEPC-C 4 -such as F6P, DHAP [4,38] and free-PEP [7,8] -could also be bound in the same allosteric site as G6P, as well as to learn the possible mode of their binding, we constructed in silico energy-minimized models with F6P, DHAP and free-PEP docked in this site. In the models, shown in Supplementary Figure S4, the three activators fit well into this site, without producing any steric clash, and make interactions with protein residues similar to those of the G6P molecule observed in the 6MGI structure reported here.
The results of these simulations provide additional support to the notion that this region of the protein is indeed the allosteric site for the phosphorylated compounds known to be activators of the PEPC enzymes.
The important number of interactions between G6P and residues of the G6P-allosteric site observed in the 6MGI structure reported hereand proposed by the docking simulations of F6P, DHAP and PEPmakes clear that this allosteric site has evolved to be highly specific for its likely physiological ligands: G6P, other phospho-sugars and free-PEP, even though it can accommodate other phosphorylated compounds and even small anions analogs of the phosphate group of these compounds such as sulfate anions. The involvement of several of these residues, namely Arg183, Ser185, Arg231, and Glu360 (ZmPEPC-C 4 numbering), was predicted by docking simulations [14] using the ZmPEPC-C 4 1JQO structure. Also, reported mutagenesis results indicate the importance of Arg183 and Arg231 [15,16] in the activation of ZmPEPC-C 4 by G6P, and chemical modification studies suggest the participation of a lysyl residue in the binding of G6P to its allosteric site in this enzyme [39]. Although the lysyl residue was not identified in the protein amino acid sequence, in view of the 6MGI structure, we believe that Lys363 might be this residue.
The conformation of the G6P-allosteric site with G6P bound observed in 6MGI and the one observed in 1JQO with sulfate bound are virtually identical ( Figure 4A). The sulfate anion in the 1JQO structure makes the same interactions with the sidechains of residues Arg183, Ser185 and Arg231 and the backbone NH of Arg184 (Figure 4B), as does the G6P-phosphate group in the 6MGI structure ( Figure 3A). The residues 360-363, which are the ones interacting with the pyranose ring of G6P, also have the same conformations in both structures despite not making any interaction with the sulfate anion in 1JQO. But a structural superposition of the G6P-allosteric site of 6MGI and that of 5VYJ ( Figure 4C) showed that although the conformation of the residues that interact with the phosphate group is similar in the latter structure, the conformation of the loop 350-369which comprises the residues 360-363 that interact with the pyranose ring of G6Pis very different. In The finding by Schlieper et al. [17] of G6P bound into the active site and not in the previously proposed G6P-allosteric site in the crystal structure of the FtPEPC-C 4 isozyme, led these authors to question the existence of a G6P-allosteric site and to propose instead that this site is a sulfate-binding site of unknown physiological relevance. Indeed, a sulfate anion was observed bound in this site, as has been observed in the other PEPC structures obtained in the presence of ammonium sulfate: the ZmPEPC-C 4 1JQO structure [13], three structures of FtPEPC-C 4 (PDB codes 4BXC, 4BXH, and 3ZGE) [17,36], and the PEPC-C 3 structure of Flaveria pringlei (FtPEPC-C 3 , PDB code 3ZGB) [36].
But while the sulfate anion in 1JQO was taken as an indication of the possible location of the G6Pallosteric site [13], in 4BXC the finding of sulfate instead of G6P was considered as a proof of the nonexistence of the G6P-allosteric site [17]. It has to be noted that, regardless of the important conformational differences between the G6P-allosteric sites of the ZmPEPC-C 4 and FtPEPC-C 4 reported structures (see below), a sulfate anion may bind to this site in any of these structures. Its binding to this site is consistent with the apparent desensitization of ZmPEPC-C 4 to G6P activation by increasing concentrations of ammonium sulfate [15], suggesting competition for the same activatorbinding site. Nevertheless, it has also to be noted that sulfate binds differently in ZmPEPC-C 4 than in

Conservation of the G6P-allosteric site residues in plant and bacterial PEPC enzymes
To further support our conclusion about the existence of the G6P-allosteric site in PEPC enzymes, we investigated the conservation in plant PEPC amino acid sequences of the eight residues that in 6MGI directly interact with the G6P molecule. The amino acid sequences used for this analysis were selected as described in the Materials and Methods section and are included in Supplementary Table   S1 and S2. With the exception of Lys363 (ZmPEPC-C 4 numbering)-which is conservatively changed to an arginine in a few monocot PEPC-C 4 sequences-we found an absolute conservation of Downloaded from https://portlandpress.com/biochemj/article-pdf/doi/10.1042/BCJ20200304/882310/bcj-2020-0304.pdf by guest on 28 May 2020 these residues, regardless of the PEPC isozymes being of the PEPC-C 4 , PEPC-C 3 or PEPC-CAM type (Figure 6), underlining the importance of this allosteric site in plant PEPCs. We also investigated the conservation of these residues in bacterial PEPC sequences from selected organisms (Supplementary Table S3). The two arginyl residues that the 6MGI structure shows interacting with the G6P-phosphate group (Arg183 and Arg231, ZmPEPC-C 4 numbering) are highly conserved, and the third residue (Ser185, ZmPEPC-C 4 numbering) is conservatively changed to a threonine in most of the bacterial sequences included in our analysis. Regarding the five residues that interact with the G6P pyranose ring, one of them (Asp239 ZmPEPC-C 4 numbering) is highly conserved in the bacterial PEPC sequences analyzed by us (in a few of them is conservatively changed to glutamate and in one to asparagine). The other four residues (Glu360, Phe361, Trp362 and Lys363) are not present in the bacterial enzymes because they belong to a loop that only exist in plant PEPCs. This high conservation suggests that important features of the G6P-allosteric site are also present in the bacterial PEPC enzymes, and supports the relevance of this site for the regulation of the activity of PEPC enzymes in general, with the exception of the archaeal PEPC-type where the G6P-allosteric site is absent [32].

ZmPEPC-C 4
In order to get biochemical evidence of the existence of the activating allosteric site for G6P and other phosphorylated compounds, we generated the variant ZmPEPC-C 4 enzyme D239A and tested its activation by G6P, F6P, and PEP, as well as its general kinetic properties. Of the eight residues that directly interact with G6P, we chose to change Asp239 because: (i) it was not changed in previous site-directed mutagenesis studies of the residues of the proposed G6P-allosteric site [15,16]; (ii) different from the previously changed residues, it interacts with the G6P-pyranose ring instead of the G6P-phosphate group and, as shown above, is the only one of the residues involved in binding  Figure 3A). Indeed, through these hydrogen bonds, Asp239 is the residue most connected to G6P.
Additionally, through its interactions with Trp362 and Lys363, Asp239 contributes to the correct position of Glu360 and Phe361, which are the two other protein residues hydrogen-bonded to the G6P-pyranose ring ( Figure 3A). Note that these interactions of Asp239 could not be predicted in the reported molecular modeling experiments [14] because these studies were performed using the 1JQO structure where Asp239 was modeled as alanine. Figure 7A shows that G6P and F6P not only did not activate but, on the contrary, inhibited the D239A variant enzyme, even at the lowest tested concentrations. G6P in the same concentration range exhibited a dual effect on the wild-type enzyme activity, being an activator at low and intermediate concentrations but an inhibitor at high concentrations ( Figure 7B). These activatory and inhibitory effects of G6P on the wild-type enzyme suggest its binding to both the G6P-allosteric site, producing enzyme activation, and with much lower affinity to the active site, producing inhibition by competing with the substrate PEP, as was previously suggested [38]. Then, a plausible explanation of the G6P-saturation data of the D239A variant is that the activatory effect was abolishedwhich would be consistent with the existence of a non-functional G6P-allosteric site as a consequence of the mutationwhile the inhibitory effect was preservedwhich would be consistent with an active site still able to bind G6P. Activation of the D239A variant by F6P was also lost; instead of activation, a small inhibition by this compound was observed (Figure 7A), suggesting that F6P also binds to the Downloaded from https://portlandpress.com/biochemj/article-pdf/doi/10.1042/BCJ20200304/882310/bcj-2020-0304.pdf by guest on 28 May 2020 active site but with very low affinity, in agreement with our finding that F6P apparently does not inhibit the wild-type enzyme even at very high concentrations ( Figure 7B).
The G6P-and F6P-saturation data of the D239A variant were best fitted to the equation that allows the estimation of the I 50 value of a total inhibitor (equation (1)), but the complex nature of the G6P-saturation data of the wild-type enzyme does not permit the fitting of a single saturation experiment to an equation simultaneously accounting for non-essential activation and total inhibition, so that to estimate apparent kinetic parameters of the wild-type enzyme for comparative purposes, we fitted the velocity data that do not exhibit apparent inhibition, i.e. those from zero to 15 mM G6P, to the equation corresponding to saturation by a non-essential activator (equation (2)). This equation was also used to fit the F6P-saturation data of the wild-type enzyme. The estimated kinetic parameters, given in Table 2, show that the affinity of the G6P-allosteric site for G6P is considerably higher than for F6P, consistent with the higher number of interactions made by G6P with residues of this site, as indicated by our crystallographic data and docking simulations (Figures 3 and   Supplementary S4). This also seems to be the case for the site responsible for the inhibition by these compounds. In any case, the inhibition of the enzyme by ligands of the G6P-allosteric site is probably physiologically irrelevant given the high concentrations of the activators at which it takes place.
Assuming again that the inhibitory effects of the activators are due to their binding to the active site, we cannot at present explain the apparent positive cooperativity observed in the G6P-inhibition data of the variant D239A, given the apparently loss of this cooperativity in the saturation by PEP (see below). This finding deserves further research.
The total lack of activation of the D239A variant by G6P and F6P clearly indicates that the G6P-allosteric site is the one responsible for activation of the enzyme by phospho-sugars and related phosphorylated compounds. Although in the D239A variant the web of hydrogen bonds made by this residue are lost, the non-functionality of the G6P-allosteric site is not probably due to its inability to bind these compounds since the residues involved in binding the phosphate group probably have the  Table 2). Qualitatively similar results were obtained when free-PEP was increased at constant 0.1 mM PEP-Mg 2+ : the variant D239A showed lower V max and affinity for the substrate and the positive cooperativity observed in the wild-type enzyme was abolished ( Figure 7D and Table 2 Figure 7E and Table 2), as in the other two saturation experiments described above. This finding could be explained as before, i.e. binding of the substrate is non-cooperative if free-PEP does not have its allosteric heterotropic activatory effects as a consequence of a non-functional allosteric-G6P site. In the case of the wildtype enzyme the observed decreased degree of cooperativity is likely due to the fixed concentration of free-PEP being almost saturating for the G6P-allosteric site, thus producing an important proportion of the activated non-cooperative enzyme form.
Other kinetic features of ZmPEPC-C 4 were also significantly altered in the D239A variant: the marked decrease in both the affinity for glycine and in the degree of activation achieved by this activator (Figure 7F and Table 2), and a much lower affinity for the inhibitor malate ( Figure 7G and Table 2). We believe that the diminished response to glycine observed in the D239A variant likely arise from the abolition in the variant enzyme of the allosteric activation by free-PEP, which would result in the loss of the synergy previously observed between glycine and G6P, or between glycine and other activators that bind to the G6P-allosteric site [5,20,38,40]. Therefore, our results suggest that the activation by neutral amino acids in PEPC-C 4 from grasses importantly rely on the G6Pallosteric site, even in the absence of G6P, through the ability of this site to bind free-PEP, which  Figure S6B).

Final Remarks
The widely accepted existence of an allosteric site for G6P and the allosteric nature of G6P activation were questioned on the basis of both structural and kinetic results [17]. In the present work, we provide new structural and kinetic data obtained with the ZmPEPC-C 4 isozyme that support that this site does exist and that it is responsible for the activation by G6P and other phosphorylated Downloaded from https://portlandpress.com/biochemj/article-pdf/doi/10.1042/BCJ20200304/882310/bcj-2020-0304.pdf by guest on 28 May 2020 compounds, being free-PEP among them. Knowing for certain which of the two mechanisms proposed for the activation of PEPC enzymes by G6Pand by extension other physiologically relevant phospho-sugars and free-PEP, i.e. allosteric activation or competitive activation, is of great importance because each mechanism would have different physiological effects. While in the competitive activation mechanism the activator is in fact an inhibitor and would produce inhibition if its concentration increases, particularly at low or moderate substrate concentrations [41], in the allosteric activation mechanism the activatory effect will be maintained after reaching its maximum at saturating concentrations of the activator, regardless of the substrate concentration. Because of this, the activator concentration range in which activation can be observed in the "competitive activation" mechanism is limited, and saturation of the active site by the "activator" would lead to zero enzyme activity. On the other hand, in the allosteric activation mechanism there is not such a restriction and the maximum activatory effect attained by saturation of the allosteric site by the activator persists even though the activator concentration is further increased. In the case of the PEPC enzymes, as observed in ZmPEPC-C 4 , activation by G6P is followed by inhibition at higher activator concentrations, but, different from the competitive activation mechanism, the two opposite effects are a consequence of G6P binding to different sites, which greatly differ in their affinity for the activator, being that of the activatory site, the allosteric site, much greater than that of the inhibiting site, the active site. This allows the full expression of the activatory potential of the activator in an ample range of concentration before inhibition becomes significant. Moreover, the degree of activation of the enzyme elicited by two different activators that bind to the same allosteric site and are present in the enzyme environment at the same timeas is probably the case of the photosynthetic PEPC-C 4 isozymes that respond to several physiological phospho-sugars [4,9,38,42], which concentrations increase simultaneously under light conditions [4,[43][44][45][46] and of those PEPC enzymes activated by free-PEPwill be different in the two mechanisms. While in the allosteric mechanism the activatory effects would be additive until saturation of the allosteric site by any of the activators is achieved, as it has been experimentally found [38], in the case of competitive activation the degree of the activation elicited by any of them would decrease, even at relatively low concentrations, and inhibition of the enzyme activity would be observed at lower concentration of the activators, because more active sites would be occupied by any of them and less active sites would be available to the substrate. Therefore, our conclusion that activation of PEPC enzymes by phosphorylated compounds results from their binding to the G6P-allosteric site is physiologically sound, since under physiological conditions competitive activation would be very inefficient and even could be detrimental for the plant.
It has to be noted that the kinetic results that Schlieper et al. [17] reported as a proof of competitive activationhigher degree of activation of FtPEPC-C 4 by G6P at lower PEP concentrationscould equally support allosteric activation, because in both mechanisms, activation will be greater at low substrate concentrations, as has been repeatedly observed in the case of PEPC enzymes. This is because in the case of competitive activation, competition between the activator and the substrate increases as does the substrate concentration, or in the case of allosteric mechanism because the activator increases the affinity of the enzyme for the substrate, even in those enzymes in which the allosteric activator also increases the maximum velocity [47], as seems to be the case of the ligands of the G6P-allosteric site. Neither is the greater activatory effect of high concentrations of ethylene glycol on FtPEPC-C 4 at low PEP concentrations, reported by Schlieper et al. [17] in support for the competitive activation mechanism, conclusive evidence of the existence of this mechanism, since the activation by ethylene glycol probably arises from its action as a cosolute that preferentially stabilizes and shifts the equilibrium towards the active R-state, not as a ligand that specifically binds to the active or allosteric site. This is further indicated by the high concentrations required to observe activation. Indeed, only high concentrations of ethylene glycol protected the ZmPEPC-C 4 against dilution, activate the enzyme and overcome its inhibition by malate inhibition, and these effects were attributed to the changes that it produces in the solvent properties [48]. In a similar fashion was

Competing interests
The authors declare that they have no conflict of interest with the contents of this article.          Assays were carried out at pH 7.3, 0.1 mM bicarbonate and 0.4 mM free-Mg 2+ . b Assays were conducted at fixed 3 mM total-PEP. c Mg 2+ -PEP was varied from 0.06 to 6.08 mM, free-PEP from 0.94 to 93.92 mM, and total-Mg 2+ from 0.47 to 6.49 mM. d Assays were performed at fixed 0.1 mM Mg 2+ -PEP; free-PEP varied from 1 to 80 mM and total-Mg 2+ from 0.33 to 0.107 mM. e Assays were performed at fixed 5 mM free-PEP; Mg 2+ -PEP varied from 0.05 to 5 mM, total-PEP from 5.05 to 10 mM, and total-Mg 2+ from 0.108 to 10.68 mM. f Assays were performed at fixed 2.5 mM total-PEP, 5 mM total-Mg 2+ and 1mM bicarbonate. NA, not apply.