A single serine to alanine substitution decreases bicarbonate affinity of phosphoenolpyruvate carboxylase in C4Flaveria trinervia

Flaveria trinervia C4 phosphoenolpyruvate carboxylase (PEPc) has higher affinity for HCO3− than its closely related Flaveria pringlei C3 PEPc, which drives higher modeled rates of C4 photosynthesis under low CO2 partial pressures.

Higher plants contain multiple PEPc-encoding (ppc) genes comprising a multigene family where most of the genes encode a non-photosynthetic C 3 PEPc (Christin and Besnard, 2009). C 4 plants obtained a modified PEPc isoform to power C 4 photosynthesis through mutations to a native ppc coding region (Christin et al., 2007;Rosnow et al., 2014) and upstream promoter region (Schaffner and Sheen, 1992;Gowik et al., 2004). Changes to the C 4 ppc promoter region led to strong, mesophyll-specific expression, resulting in high PEPc activity to drive the CO 2 -concentrating mechanism of C 4 photosynthesis . Work on PEPc peptide sequences from members of the Poaceae, Amaranthaceae, Asteraceae, and Cyperaceae families (Christin et al., 2007), as well as the Chenopodaceae family (Rosnow et al., 2014), identified amino acid residues predicted to be under positive selection in these C 4 lineages. Comparing PEPc sequences of species within and between families shows that C 4 PEPc isoforms from different species possess different combinations of amino acid residues under positive selection (Christin et al., 2007;Rosnow et al., 2014). These findings suggest that there are multiple ways the C 4 PEPc kinetic properties can arise in different C 4 origins or that there is diversity in the PEPc kinetics between species.
An increase in PEPc activity in the leaf mesophyll cytosol in the intermediate C 3 /C 4 species would be likely to lead to selection for changes in kinetic properties ). This was previously tested in a variety of C 3 / C 4 species that displayed a progression in altered K PEP (K m for PEP) and decreased malate sensitivity . C 4 plants contain high levels of malate in the mesophyll cytosol, so there would be selection for amino acid substitutions that transition the malate-sensitive C 3 PEPc to a less sensitive C 4 PEPc Paulus et al., 2013). This is supported by the Gly884 substitution in the Flaveria trinervia C 4 PEPc to the Flaveria pringlei C 3 PEPc arginine that caused the C 4 PEPc to lose its resistance to malate (Paulus et al., 2013).
As PEPc transitioned from C 3 to C 4 function, it has been suggested that certain amino acid substitutions were under positive selection to alter K PEP and K HCO3 (K m for HCO 3 − ). It is hypothesized that certain mutations in the C 4 ppc coding region resulted from strong selective pressures to obtain a lower K HCO3 than that of the C 3 PEPc (Jacobs et al., 2008). The lower K HCO3 of the C 4 PEPc may enhance the efficiency of C 4 photosynthesis, especially when HCO 3 − availably is low due to reduced stomatal conductance. Alternatively, the C 4 PEPc has been shown to have a higher K PEP , with values typically reported between 100 µM and 590 µM Dong et al., 1998;Westhoff and Gowik, 2004;Lara et al., 2006;Rosnow et al., 2015), as compared with C 3 non-photosynthetic PEPc K PEP values which range from 13 µM to 60 µM Bläsing et al., 2002;Gowik et al., 2006;Lara et al., 2006;Rosnow et al., 2015). It was hypothesized that this increase in C 4 K PEP was an unavoidable consequence of the reduction of K HCO3 since the two kinetic traits may be linked by certain amino acids (Jacobs et al., 2008;Gowik and Westhoff, 2011). Alternatively, since the PEP pools in a C 4 leaf are higher than in a C 3 leaf, the high K PEP of the C 4 PEPc may ensure stronger diurnal regulation of PEPc (Budde and Chollet, 1986;Hatch, 1987).
Residue S774 in F. trinervia (S780 in maize) was shown to be under positive selection by Poetsch et al. (1991) and Hermans and Westhoff (1992), and substituting the conserved C 4 serine for the conserved C 3 alanine in F. trinervia (S774A) significantly decreased the K PEP of the C 4 PEPc (Bläsing et al., 2000;Engelmann et al., 2002). Since the S774A substitution affects the K PEP of PEPc, it is possible that it may also affect the K HCO3 , making S774 one of the residues potentially linking K PEP and K HCO3 . However, this serine residue was unimportant for the high K PEP in the C 4 Chenopodaceae (Rosnow et al., 2014).
The only study to publish K HCO3 of both a C 3 and C 4 PEPc showed that the K HCO3 of five C 4 species representing the Poaceae and Amaranthaceae families was ~26 µM compared with preliminary evidence suggesting that the K HCO3 of the C 3 PEPc from Flaveria cronquistii (Asteraceae family) was 80 µM (Bauwe, 1986). Other studies reported C 4 PEPc K HCO3 values ranging from 14 µM to 180 µM (Janc et al., 1992;Gao and Woo, 1995;Parvathi et al., 2000;Boyd et al., 2015), where the C 3 K HCO3 of 80 µM falls within this reported range of C 4 K HCO3 values. However, comparing K HCO3 for closely related C 3 and C 4 PEPc isoforms can provide a more accurate analysis of the change in C 3 to C 4 K HCO3 and whether there was a strong selective force on PEPc K HCO3 , but to date this has not been performed. Additionally, the previously reported K HCO3 values were obtained by coupling PEPc activity to spectrophotometrically measured NADH oxidation rates. It is difficult to obtain accurate K HCO3 values using this method because it does not directly measure changes in HCO 3 − concentration in the assay. This is problematic because measurements of K HCO3 require accurate determinations of PEPc activity and HCO 3 − concentrations below the K HCO3 , which is in the micromolar range. To overcome this problem, membrane-inlet mass spectrometry (MIMS) can be used to measure HCO 3 − consumption by PEPc accurately and directly in real-time over a wide range of inorganic carbon (C i ) concentrations, including concentrations well below the K HCO3 , without the complication of coupling PEPc activity to the NADH dehydrogenase reaction (Boyd et al., 2015).
In this study, we use MIMS to obtain K HCO3 values for the photosynthetic PEPc from the C 4 plant F. trinervia and the non-photosynthetic PEPc from the C 3 plant F. pringlei that were overexpressed and purified from the PEPc-less PCR1 Escherichia coli strain (Sabe et al., 1984;Svensson et al., 1997). We found that the S774A substitution increases the C 4 K HCO3 , whereas the A774S substitution did not affect the C 3 K HCO3 , suggesting that additional amino acids besides S774 are involved in the C 4 K HCO3 trait. Since previous studies reported PEPc K PEP changing in the presence of the allosteric activator glucose 6-phosphate (G6-P) and the inhibitor malate (Huber and Edwards, 1975;Wedding et al., 1990;Gupta et al., 1994;Bläsing et al., 2002), we tested whether these allosteric regulators also affected K HCO3 . We report that G6-P and malate have a minimal effect on the K HCO3 of PEPc. We address how differences in calibration methods, assay conditions, and enzyme extractions can produce different in vitro kinetic values, and we report an improvement to the MIMS PEPc assay. Lastly, we demonstrate how the decrease in K HCO3 between the C 3 and C 4 PEPc isoforms increases the modeled rates of C 4 photosynthesis at low CO 2 concentrations.

Generating PCR1 PEPc-overexpressing lines
The PEPc-less E. coli strain, PCR1 (Sabe et al., 1984), and PEPc cDNA constructs used in Svensson et al. (1997) and Bläsing et al. (2000) were generously provided by Professor Peter Westhoff 's lab. The plasmid, pTrc99A, carrying the cDNA coding for either the C 4 F. trinervia PEPc, the C 3 F. pringlei PEPc, or Flaveria PEPc with either an alanine or serine substitution at residue 774, C 4 -S774A or C 3 -A774S, respectively, was transformed into the PCR1 E. coli strain. PCR1 transformants producing plant PEPc were selected following the method of Svensson et al. (1997).

Growth of PCR1 PEPc-overexpressing lines for PEPc extraction
A 4 ml growth culture (Luria-Bertani broth; 0.1% w/v dextrose; 100 µg ml −1 ampicillin) was inoculated with a glycerol stock of the PCR1 strain carrying a Flaveria PEPc construct and was incubated at 28 °C with shaking at 160 rpm overnight. The following morning, the 4 ml culture was centrifuged at 1538 g for 10 min at room temperature. The supernatant was discarded, and the bacterial pellets were resuspended and transferred to a large 500 ml growth culture which was incubated at 28 °C and shaken at 160 rpm. After 6 h of incubation, isopropyl-β-d-1-thiogalactopyranoside (IPTG) was added to a final concentration of 100 µM to the 500 ml growth culture to induce PEPc production overnight.

PEPc extraction and purification from E. coli
The 500 ml growth culture was centrifuged at 2602 g for 10 min at room temperature and the bacterial pellets were resuspended in a total volume of 20 ml of ice-cold lysis buffer [50 mM Tris-HCl, pH 8.0; 0.5 M NaCl; 10 mM DTT; 1 mM EDTA, pH 8.0; 20 µl ml −1 E. coli protease inhibitor (Sigma); 1 mg ml −1 lysozyme (Bioworld); 10% (v/v) glycerol; 20% (w/v) sucrose]. The resuspended cells were placed in ice for 30 min and then lysed via sonication (BioLogics Ultrasonic Homogenizer 300 V/T). The sonicated cells were transferred to centrifuge tubes and were spun at 30 597 g for 30 min at 4 °C. The supernatant was collected and MgCl 2 was added to the supernatant to a final concentration of 10 mM. Polyethylene glycol (50% PEG 8000) was added to the supernatant to a final concentration of 6% (v/v) before placing the supernatant on ice for 15 min with gentle mixing. The supernatant was again spun at 30 597 g for 20 min at 4 °C. The protein pellets were discarded and 50% PEG 8000 was added to the final concentration of 12% (v/v). The supernatant was slowly stirred on ice for 15 min before centrifugation at 30 597 g for 20 min at 4 °C.
The protein pellet was collected and resuspended in 6 ml of Buffer A [0.5 M (NH 4 ) 2 SO 4 ; 20 mM Tris-HCl, pH 7.5; 0.1 mM DTT; 1 mM EDTA, pH 8.0; 5% (v/v) glycerol] supplemented with 1 mM phenylmethylsulfonyl fluoride (PMSF) . The protein sample was loaded onto a HIC phenyl-Sepharose column (1 cm×5.5 cm) pre-incubated with Buffer A at a flow rate of 1.5 ml min −1 . The hydrophobic properties of PEPc were used for the partial purification of PEPc by following the protocol of Svensson et al. (1997).
Fractions from the phenyl-Sepharose column were analyzed for PEPc activity by coupling the PEPc and NADH dehydrogenase reactions following the protocol of Boyd et al. (2015). Fractions displaying the highest PEPc activity were pooled and desalted in Buffer B [100 mM HEPES-KOH, pH 7.6; 1 mM DTT; 1 mM EDTA, pH 8.0] and concentrated using Corning Spin-X UF columns (6 ml volume, 100 K molecular weight cutoff) according to the Corning procedure. Glycerol was added to a final concentration of 20% (v/v) before storage at -80 °C. Total protein content of the PEPc samples was measured by a modified Bradford assay (Bio-Rad Protein Assay Kit II) (Bradford, 1976) using the Bio-Rad procedure.
Extracting PEPc from F. trinervia and Setaria viridis PEPc samples were extracted and desalted from leaves of F. trinervia and S. viridis following the procedure of Boyd et al. (2015). Once the desalted PEPc extracts were collected and moved through a Millex-GP 0.22 µm syringe filter (Millipore), the extracts were concentrated by spinning the samples at 2880 g for 20 min at 4 °C in an Amicon Ultra-4 Ultracel-100K centrifugal filter (Millipore). Glycerol was added to the concentrated PEPc samples to a final concentration of 20% (v/v) and stored at -80 °C.

Obtaining V Pmax , K HCO3 , and Hill values for the different PEPc isoforms
The HCO 3 − -dependent PEPc assays were run in a 600 µl cuvette attached to the inlet of a mass spectrometer as described by Cousins et al. (2010). A CO 2 calibration was conducted before each HCO 3 − response curve as reported by Boyd et al. (2015). The calibration consisted of three 2 µl injections of 10 mM NaHCO 3 into 0.1 N HCl and three 6 µl injections of 100 mM NaHCO 3 into the PEPc reaction mixture [100 mM HEPES-KOH, pH 7.6; 10 mM MgCl 2 ; 1 mM DTT; 50 µg ml −1 carbonic anhydrase (CA); 5 mM G6-P; 5 mM PEP]. The second calibration step differs from Boyd et al. (2015) as their 100 mM NaHCO 3 injections went into buffer lacking DTT, G6-P, and PEP. DTT, G6-P, and PEP were added to the second calibration to take into account the pH changes these compounds have on the assay buffer. As the CO 2 calibrations take into account total C i and total CO 2 in the cuvette, the HCO 3 − concentration could be deduced by subtracting the amount of CO 2 in the cuvette from the total C i measured in the cuvette (Boyd et al., 2015).
To measure PEPc HCO 3 − kinetics, seven NaHCO 3 concentrations (50, 100, 200, 350, 500, 750, and 1000 µM) were used for the assays. The CO 2 was removed from the assay buffer containing 100 mM HEPES-KOH, pH 7.6 and 10 mM MgCl 2 by continuously bubbling the buffer with humidified N 2 gas starting at least 1 h prior to initiating the assays. The assay buffer (600 µl) followed by 1 mM DTT, 50 µg ml −1 CA, 5 mM G6-P, 5 mM PEP, and various NaHCO 3 concentrations were added to the reaction cuvette and held at a constant 25 °C with a temperaturecontrolled water bath.
A blank rate was obtained by measuring the change in the mass 44 ( 12 C 16 O 16 O) signal during a 30 s period before initiating the reaction with the addition of 10-15 µg of total protein of the PEPc extract. The PEPc reaction was run for 5 min but the first 20 s of the PEPc reaction were discarded to allow for enzyme mixing and rate stabilization. The MIMS reports a mass 44 signal every 0.8 s, so a robust 10 s running average of the change in mass 44 was used as a single data point for PEPc activity (V P ) at an averaged [HCO 3 − ]. Data points from 10 s running averages were taken immediately following the 20 s mixing phase. For the larger NaHCO 3 injections (350-1000 µM NaHCO 3 ), 10 s running averages were taken until a drop in PEPc activity was observed to avoid data points where there might be end-product inhibition of the reaction. For the lower NaHCO 3 concentrations (50-200 µM), 10 s running averages were taken until the reaction was depleted of C i . Once the C i was depleted from the 50, 100, and 200 µM NaHCO 3 injections, as indicated by a zero slope for the mass 44 signal, a 30 s running average of the zero slope was taken to obtain a mass 44 zero. These mass 44 zeroes accounted for mechanical drift in the MIMS as the different zeroes were taken at various times throughout the HCO 3 − response curve. The kinetic parameters V Pmax , K HCO3 , and Hill value (h) were obtained by using the Hill equation: (1) where the Hill equation was fit to the HCO 3 − response curve using Excel's Solver function to produce the kinetic parameters listed above.

Measuring the impact of G6-P and malate on K HCO3
To measure the effect of G6-P on K HCO3 , G6-P was omitted from the assay described above to compare K HCO3 values in the presence or absence of 5 mM G6-P. Alternatively, 2.5 mM malate (pH 7.6) was added to the assay described above to determine if malate affects PEPc K HCO3 values in the presence or absence of G6-P.

Measuring PEP effects on malate inhibition
The MIMS assay described above was used to determine if PEP concentration affects malate inhibition of PEPc, in the absence of G6-P. Five malate concentrations were used (0, 1, 2, 3, and 4 mM) to determine the percentage change in enzyme activity of the C 3 and C 4 PEPc isoforms in the presence of saturating NaHCO 3 (1000 µM) and when PEP was saturating (5 mM) or non-saturating (150 µM and 750 µM for the C 3 and C 4 isoforms, respectively). The non-saturating PEP concentrations were determined to be twice the reported K PEP values (2×K PEP ) in the absence of G6-P .

Extraction source, pH, and calibration effects on PEPc K HCO3
Previously, MIMS measurements of desalted PEPc extracts from S. viridis reported a K HCO3 of 62.8 ± 5.0 µM (Boyd et al., 2015). Therefore, we tested whether differences in PEPc source (plant versus E. coli), pH, or MIMS calibrations caused the K HCO3 reported here to differ from those of Boyd et al. (2015). The MIMS assay described above containing 5 mM G6-P and no malate was used to test whether PEPc samples extracted from leaves of F. trinervia and S. viridis produced different K HCO3 values from the C 4 PEPc extracted from E. coli. For each assay, 5-10 µl of plant extract was added to initiate the PEPc reaction. PEPc extracts from S. viridis were used to compare K HCO3 values obtained at pH 7.6, the pH used in this study, with K HCO3 values obtained at pH 7.8 used by Boyd et al. (2015). These extracts were also used to compare K HCO3 values obtained at pH 7.8 using either the current calibration method outlined above or the calibration method of Boyd et al. (2015).

Statistical analysis of experimental data
Statistical analyses of the kinetic data were performed using RStudio version 1.1.447 (RStudio Team, 2016). Homogeneity of variance was checked using Levene tests, and normality was checked using residual quantile plots and residual versus fitted value plots. Non-normal data were log transformed but they reported the same statistical outcomes as non-transformed data, so for simplicity only non-transformed data analyses are presented. One-way ANOVA and Tukey HSD post-hoc tests were used to determine statistical significance (P<0.05) of K HCO3 between PEPc isoforms. Two-way ANOVA (P<0.05) and Tukey HSD post-hoc tests were used to analyze statistically significant differences in K HCO3 between isoforms and the impact of potential allosteric effectors. A twoway repeated measures ANOVA (P<0.05) was used to test if the change in PEPc activity in response to malate significantly differed between C 3 and C 4 PEPc isoforms at various PEP concentrations. One-way ANOVA and Tukey HSD post-hoc tests were used to determine significant differences between C 4 PEPc extracted from E. coli and PEPc extracted from F. trinervia and S. viridis leaves. Student's t-tests (P<0.05) were separately used to determine significant differences in S. viridis PEPc K HCO3 assayed at pH 7.6 and 7.8 and for S. viridis PEPc K HCO3 assayed at pH 7.8 using the two calibration methods.

Modeling the effect of K HCO3 on C 4 photosynthesis
The modeled effect of K HCO3 on the response of C 4 enzyme-limited photosynthetic CO 2 assimilation (A c ) to changing mesophyll CO 2 concentrations (C m ) was determined by solving the quadratic formula using the set of equations as described by von Caemmerer (2000). The equations and input variables were taken from von , von Caemmerer (2000), Tholen and Zhu (2011), and Ubierna et al. (2013), and are presented in Supplementary Table S1 at JXB online.

Kinetics of the C 3 , C 4 , and chimeric PEPc isoforms
The Hill equation was used to determine the maximum rate of PEPc carboxylation (V Pmax ), the K m for bicarbonate (K HCO3 ), and the co-operativity of the PEPc active sites (h) from 25 °C MIMS measurements of PEPc activity (V p ) in response to changes in HCO 3 − concentrations ( Supplementary  Fig. S1). Measurements were made on C 4 , C 3 , and chimeric Flaveria PEPc isoforms expressed and partially purified from E. coli. The C 4 PEPc had a significantly lower K HCO3 than the C 3 PEPc, 26.6 ± 1.7 µM and 64.0 ± 2.4 µM, respectively (Fig. 1). Additionally, the C 3 PEPc had a lower V Pmax compared with the C 4 PEPc, 5.1 ± 0.7 µmol mg protein −1 min −1 and 8.1 ± 0.7 µmol mg protein −1 min −1 , respectively (Supplementary Table S2). Neither isoform displayed co-operativity towards HCO 3 − binding, with Hill values close to 1.0 under all assay conditions (Supplementary Table S2).
The substitution of the conserved C 4 serine at residue 774 (780 in maize) with the conserved C 3 alanine (C 4 -S774A) significantly increased the K HCO3 by 45% from 26.6 ± 1.7 µM to 38.6 ± 5.5 µM (Fig. 1). However, the C 4 -S774A substitution had no effect on V Pmax (Supplementary Table S2). The reverse substitution, C 3 -A774S made in the C 3 PEPc, did not significantly change the K HCO3 (from 64.0 ± 2.4 µM to 61.5 ± 9.1 µM; Fig. 1) nor did it affect V Pmax (Supplementary  Table S2). Fig. 1. The K HCO3 of the C 3 , C 4 , and chimeric PEPc isoforms. The K HCO3 values were obtained from the MIMS assayed in 100 mM HEPES-KOH buffer (pH 7.6) that contained 10 mM MgCl 2 , 5 mM PEP, 50 µg ml −1 CA, 1 mM DTT, and 5 mM G6-P. Error bars represent the mean ±SD of four independent extractions from E. coli for each PEPc isoform. Significance was determined by one-way ANOVA and Tukey HSD post-hoc tests. Bars with different letters are significantly different (P<0.05).

The impact of G6-P and malate on K HCO3
The V Pmax did not change by omitting G6-P from the assay, regardless of the PEPc isoform (Supplementary Table S2). Additionally, the K HCO3 values of the PEPc isoforms were not significantly altered by the presence or absence of G6-P in the assay (Fig. 2).
Under the current measurement conditions of pH 7.6, 5 mM PEP, and the absence of G6-P, the addition of 2.5 mM malate decreased the V Pmax in the C 4 , C 4 -S774A chimeric, and the C 3 PEPc by 57.5, 24.4, and 6.8%, respectively (Supplementary  Tables S2, S3). However, the K HCO3 values of the PEPc isoforms were not significantly altered by the presence of malate in the assay (Fig. 3A). HCO 3 − response curves in the absence of G6-P were not obtained for the C 3 -A774S PEPc due to severe inhibition of the chimeric PEPc by malate ( Supplementary  Fig. S2). When 5 mM G6-P and 2.5 mM malate were both present in the PEPc assay, V Pmax decreased in the C 4 , C 4 -S774A chimeric, C 3 , and C 3 -A774S chimeric PEPc by 44.4, 28.6, 2, and 14%, respectively (Supplementary Tables S2, S3). However, the K HCO3 values of the PEPc isoforms were not significantly changed with both G6-P and malate in the assay (Fig. 3B). Interestingly, decreasing the total PEP concentration in the assay from 5 mM to twice the reported K PEP of the C 3 and C 4 isoforms Paulus et al., 2013), 150 µM and 750 µM PEP, respectively, caused the C 3 PEPc to lose activity dramatically in the presence of malate, whereas the change in PEP concentration had a smaller effect on malate inhibition of C 4 PEPc activity (Fig. 4).
Although not statistically significant, changing the pH of the assay buffer from 7.6 to 7.8 increased the K HCO3 of S. viridis PEPc by 21.3% from 30.0 ± 3.0 µM to 36.4 ± 5.3 µM (Fig. 5A, 5B). When using the Boyd et al. (2015) MIMS calibration method at pH 7.8, the S. viridis K HCO3 increased to 62.9 ± 8.7 µM (Fig. 5B). The K HCO3 values of the different PEPc isoforms were obtained from MIMS PEPc assays where 5 mM G6-P was present (white bars) or absent (gray bars) in the assay buffer. White bars are data represented from Fig. 1. Error bars represent the mean ±SD of four independent extractions from E. coli for each PEPc isoform. A two-way ANOVA determined that G6-P had a non-significant effect on K HCO3 but there was a significant isoform effect, and a Tukey HSD post-hoc test was used to determined significance. PEPc isoforms with different letters are significantly different (P<0.005).

Modeling the effect of different K HCO3 values on C 4 photosynthesis
The C 3 and C 4 PEPc K HCO3 values from Fig. 1 were input into the C 4 photosynthesis model from von Caemmerer (2000) to determine how differences in K HCO3 would impact modeled rates of C 4 photosynthesis. Varying K HCO3 with a constant V Pmax significantly changed the modeled rates of C 4 photosynthesis under CO 2 conditions below 20 Pa. The lower K HCO3 of the C 4 PEPc resulted in higher modeled rates of C 4 photosynthesis at these low mesophyll CO 2 concentrations (C m ; Fig. 6). However, the difference between the C 3 and C 4 K HCO3 modeled no differences in net CO 2 assimilation above ~20 Pa C m (Fig. 6).

Kinetic changes during the evolution of the C 4 PEPc
We and others (Jacobs et al., 2008;Gowik and Westhoff, 2011) have hypothesized that there was a strong selective pressure to reduce the K HCO3 of the C 4 PEPc isoform. Additionally, previous studies have reported that changing the PEPc amino acid residue 774 in Flaveria spp. (780 in maize) influences K PEP and its allosteric regulation (Engelmann et al., 2002;Endo et al., 2008). Therefore, the aim of this research was to test the hypotheses that the K HCO3 of the C 4 PEPc isoform from F. trinervia would be lower than the K HCO3 of the C 3 PEPc isoform from F. pringlei and that changes to residue 774 will impact K HCO3 and its allosteric regulation. Residue 774 was chosen because others have shown the C 4 F. trinervia S774A substitution reduces K PEP (Bläsing et al., 2000;Endo et al., 2008). Furthermore, residue 774 is near both the PEP-and HCO 3 − -binding sites, and may also influence K HCO3 . We have analyzed the influence of this residue on K HCO3 and showed that the C 4 -S774A chimeric PEPc had a significantly higher K HCO3 compared with the C 4 PEPc (Fig. 1). This fits with previous data that suggest that K PEP and K HCO3 are inversely linked through specific amino acid residues near the two binding sites. For example, the K829G substitution in the F. trinervia C 4 PEPc resulted in a small decrease in K HCO3 and a simultaneous increase to K PEP (Gao and Woo, 1996). Alternatively, swapping Lys600 with either an arginine or threonine in F. trinervia led and desalted PEPc extracts from F. trinervia and S. viridis (gray bars) were obtained from the MIMS assayed in 100 mM HEPES-KOH buffer (pH 7.6) with 10 mM MgCl 2 , 5 mM PEP, 50 µg ml −1 CA, 1 mM DTT, and 5 mM G6-P. Bars represent the mean ±SD of four independent PEPc extractions. Significance was determined by one-way ANOVA and Tukey HSD tests. Bars with different letters are significantly different (P<0.05). (B) S. viridis PEPc K HCO3 values measured from assays at pH 7.8 using either the current MIMS calibration or the Boyd et al. (2015) calibration. Significance between the K HCO3 of S. viridis PEPc assayed at (A) pH 7.6 and (B) 7.8 was determined by a Student's t-test (P=0.08). Significance between the K HCO3 of S. viridis PEPc assayed at pH 7.8 obtained by either the current calibration method or the Boyd et al. (2015) calibration method was determined by a Student's t-test (P<0.05). Fig. 4. Malate resistance of the C 3 PEPc is affected more by changing PEP concentrations than that of the C 4 PEPc. At 5 mM PEP (filled circles and squares), the C 3 PEPc (solid line) is more resistant to malate than the C 4 PEPc (dashed line). When the PEP concentration was dropped to twice the reported K PEP for the C 3 and C 4 PEPc isoforms (open circles and squares), malate resistance of the C 3 PEPc dropped drastically compared with the malate resistance for the C 4 PEPc. Malate activity assays were performed in 100 mM HEPES-KOH (pH 7.6), 10 mM MgCl 2 , 1 mM DTT, 50 µg ml −1 CA, 2.5 mM malate (pH 7.6), 1 mM NaHCO 3 , and various PEP concentrations. Shapes and error bars represent the mean ±SD of four independent extractions from E. coli for both the C 3 and C 4 PEPc isoforms. A two-way repeated measures ANOVA (P<0.05) determined that there was a significant difference between the increased sensitivity to malate of the C 3 PEPc versus the C 4 PEPc when PEP concentration was decreased to 2×K PEP for each isoform.
to increases in both K PEP and K HCO3 , but this residue is one of the four conserved amino acids comprising the HCO 3 − binding site (Gao and Woo, 1995;Kai et al., 2003).
The chimeric C 3 PEPc of F. pringlei, C 3 -A774S, had a minimal effect on K HCO3 (Fig. 1). This same amino acid substitution was also shown not to influence the K PEP of the C 3 PEPc with G6-P present in the assay. However, in the absence of G6-P, the same A774S substitution did increase K PEP (Bläsing et al., 2000). Taken together, these results support the analysis that multiple amino acid residues, in addition to S774 (S780 in maize), influence PEPc kinetics. Indeed, swapping amino acid residues 296-437 from the C 4 enzyme into the C 3 -A774S chimeric PEPc increased the C 3 K PEP even closer to the established C 4 K PEP value (Engelmann et al., 2002). However, further research is needed to understand how changes in PEPc amino acid composition influence K HCO3 , particularly how specific amino acid changes influence the impact that allosteric regulators such as G6-P and malate have on K HCO3 .

The influence of G6-P and malate on K HCO3
PEPc is regulated by post-translational modifications (PTMs) and interactions with allosteric effectors. Previous studies showed that glycine and G6-P activate PEPc Westhoff et al., 1997;Endo et al., 2008) while the endproducts, aspartic acid and malate, inhibit PEPc (Huber and Edwards, 1975). Additionally, previous studies have shown that the PEP-binding sites of the C 4 PEPc tetramer display positive co-operativity for K PEP , which can be altered by the binding of G6-P and malate to PEPc (Wedding et al., 1990;Bläsing et al., 2002;Rosnow et al., 2015). We show that neither G6-P nor malate appears to impact K HCO3 and the co-operativity of HCO 3 − binding to the extent that they influence K PEP and co-operative PEP binding (Figs 2, 3; Supplementary  Tables S2, S3). Using the crystal structure (4BXC) deposited by Schlieper et al. (2014), the HCO 3 − -binding site is further from the G6-P and aspartic acid/malate allosteric binding sites than the PEP-binding site is from the allosteric sites, so any structural changes to PEPc caused by allosteric binding may affect the PEP-binding site more than the HCO 3 − -binding site. Additionally, the S774A and A774S substitutions did not influence the allosteric regulation of PEPc to the extent that the R884G and G884R substitutions affected malate sensitivity of the F. pringlei and F. trinervia PEPc isoforms, respectively (Paulus et al., 2013). This discrepancy may be due to residue 884 being closer to the residues of the aspartate/malate-binding sites compared with residue 774, whereas residue 774 is closer to the PEP-and HCO 3 − -binding sites (Kai et al., 2003;Paulus et al., 2013).
Another possibility is that PEP binds before HCO 3 − (Janc et al., 1992), potentially conferring the primary allosteric regulation of PEPc to the binding of PEP. Alternatively, under our assay conditions, the high PEP concentration may have reduced the impact G6-P and malate had on K HCO3 . For example, G6-P has a greater activating effect on PEPc under limiting PEP concentrations at 0.5 mM (Gupta et al., 1994). In addition, multiple studies suggest that there is a regulatory PEP-binding site different from the PEPc active site (Rustin et al., 1988;Rodríguez-Sotres and Muñoz-Clares, 1990;Mújica-Jiménez et al., 1998;Yuan et al., 2006), and it is possible this regulatory PEP site may not be saturated under low PEP concentrations. Saturating this regulatory PEP-binding site might supersede G6-P activation and overcome malate inhibition of PEPc (Huber and Edwards, 1975). The assay conditions used in this study contained saturating (5 mM) levels of PEP, which were well above the K PEP of both PEPc isoforms, since limiting PEP would complicate the response to changes in HCO 3 − concentrations. The C 3 PEPc was more resistant to malate inhibition than the C 4 PEPc under these saturating PEP conditions, which is in contrast to previous reports Paulus et al., 2013). However, we found that the C 3 PEPc was more sensitive to malate than the C 4 PEPc when the PEP concentration in the assay was reduced to 2×K PEP (Fig. 4). This suggests that the PEP regulatory site for the C 4 PEPc may be less sensitive than the C 3 PEPc to changes in free PEP availability. Alternatively, the C 4 PEP regulatory site may not have as much influence on malate tolerance as the C 3 PEP regulatory site under the current assay conditions.
We were unable to obtain kinetic data for the C 3 -A774S chimeric PEPc due to drastic inhibition of the enzyme by malate when G6-P was absent from the assay ( Supplementary  Fig. S2). This result was unexpected since the addition of 2.5 mM malate had a small effect on the activity of the C 3 PEPc (Supplementary Tables S2, S3). However, since PEP and Fig. 6. Modeled rates of C 4 photosynthesis with C 3 and C 4 K HCO3 . The K HCO3 values for the C 3 PEPc (filled circles) and C 4 PEPc (open circles) were input into the C 4 photosynthesis model from von Caemmerer (2000) to determine the modeled rate of CO 2 assimilation (A net ) at various mesophyll CO 2 concentrations (C m ). Symbols represent means ±SD of four independent K HCO3 values presented in Fig. 1 input into the model, where all other variables in the C 4 model were held constant. The maximal rates of PEP regeneration (V pr ), Rubisco carboxylation (V Cmax ), and maximum PEPc carboxylation per unit leaf area [V Pmax(plant) ] were set to 80, 60, and 120 µmol m −2 s −1 , respectively (von Caemmerer, 2000), and all other values are presented in Supplementgary Table S1. A pK a of 6.12 and assumed a mesophyll cytosol pH of 7.2 were used to convert µM HCO 3 − to µM CO 2 . Pa CO 2 was obtained by using Henry's constant for CO 2 (0.034 mol l −1 atm −1 ) and assumed standard pressure (101 325 Pa atm −1 ). malate interact differently with the C 3 and C 4 PEPc isoforms, it is possible that the A774S substitution in the C 3 PEPc modified these interactions to allow potent inhibition of the C 3 -A774S chimeric PEPc. Further analysis is needed to test the extent of malate inhibition on the C 3 -A774S PEPc and other chimeric PEPc isoforms under various assay conditions. It is worth noting that malate has a stronger inhibitory effect on PEPc at pH 7.0 than at pH 8.0 (Huber and Edwards, 1975;Gupta et al., 1994). As discussed below, pH and other assay conditions used to measure PEPc activity can influence the absolute values of the kinetic parameters.
Assay conditions, extraction method, and source can affect PEPc kinetics MIMS can directly measure dissolved CO 2 even at very low C i concentrations below the K HCO3 of PEPc (Beckmann et al., 2009;Cousins et al., 2010). Previously, Boyd et al. (2015) reported a MIMS-measured K HCO3 value of 62.8 µM for the S. viridis C 4 PEPc which is higher than our MIMS-measured K HCO3 value of 26.6 µM for the F. trinervia C 4 PEPc extracted from E. coli. This difference in K HCO3 between the S. viridis and F. trinervia C 4 PEPc may be due to any combination of species differences in enzyme kinetics, enzyme purity, pH of the assay, and MIMS calibrations. Plant PEPc extracted from F. trinervia, a dicot in the Asteraceae family, and S. viridis, a monocot in the Poaceae family, had similar K HCO3 values at pH 7.6 ( Fig. 5A). Bauwe (1986) also reported similar K HCO3 values for different C 4 PEPc isoforms extracted from multiple grasses and Glomphrena globosa, a dicot from the Amaranthaceae family.
The F. trinervia C 4 PEPc partially purified from E. coli was reported to be unphosphorylated at the N-terminal serine residue  and had a significantly lower K HCO3 than the desalted plant PEPc extracts taken from F. trinervia leaves during the day (Fig. 5). This suggests that potential differences in post-translational modifications might influence the kinetic properties of PEPc. Parvathi et al. (2000) observed a decrease in K HCO3 as PEPc changed from the unphosphorylated to the phosphorylated state, and that PEPc extracts from illuminated leaves had lower K HCO3 values than PEPc extracted in the dark. In the current study, the phosphorylation status of the PEPc extracts was not tested, so it cannot be confirmed that the difference in K HCO3 between the plant and E. coli extracts is due to changes in PTMs. Alternatively, Bauwe (1986) observed that unpurified C 4 PEPc had higher K HCO3 values than purified C 4 PEPc extracts. It is possible that the impurity of our desalted plant PEPc extracts from F. trinervia contributed to the increased K HCO3 relative to the C 4 PEPc purified from E. coli. The potential differences in PTMs and enzyme purity do not completely explain why the K HCO3 for the S. viridis PEPc reported here and by Boyd et al. (2015) differ; however, this discrepancy in K HCO3 can be explained by differences in assay conditions and MIMS calibrations.
Raising the pH of the PEPc assay from 7.6 to 7.8, the pH used by Boyd et al. (2015), increased the S. viridis PEPc K HCO3 by ~21% (Fig. 5A, B). In addition to the pH of the assay buffer, the MIMS calibration method can alter the measured K HCO3 . This is because two MIMS calibrations are required to convert a voltage signal of mass 44 to a micromolar concentration of CO 2 and to determine the HCO 3 − concentration in the reaction cuvette. Since the development of a novel MIMS technique to measure K HCO3 of PEPc (Boyd et al., 2015), we have improved the MIMS calibration method to obtain more accurate K HCO3 values to analyze kinetic differences between PEPc isoforms. The calibration method presented here differed from that of Boyd et al. (2015) because all reaction components except the enzyme extract were included in the calibration. This would account for slight pH changes to the assay when adding DTT, G6-P, or PEP, which is important for determining the CO 2 :HCO 3 − ratio. If there is a slight reduction in pH from adding assay components that is not accounted for during the calibrations, then the CO 2 :HCO 3 − ratio can be slightly overestimated, leading to higher estimations of K HCO3 . Using the calibration method and pH of 7.8 from Boyd et al. (2015), the K HCO3 values reported here (62.9 ± 8.7 µM) and by Boyd et al. (2015) (62.8 ± 5.0 µM) were nearly identical (Fig. 5B), suggesting that assay conditions such as pH and differences in MIMS calibration methods can affect the estimated K HCO3 . These findings also highlight the important consideration of how well in vitro assay conditions reflect the in vivo conditions where PEPc operates. So far, in vivo PEPc kinetics can only be obtained by models using gas exchange (von Caemmerer, 2000). Further research is needed to compare in vitro and in vivo PEPc kinetics, since accurate PEPc kinetics are needed to model C 4 photosynthesis.

K HCO3 affects modeled rates of C 4 photosynthesis
The C 4 photosynthesis model developed by von Caemmerer (2000) was used to test if differences in K HCO3 between the C 3 and C 4 PEPc isoforms were enough to influence rates of net CO 2 assimilation during C 4 photosynthesis. The C 4 model predicts that a lower K HCO3 may not affect photosynthetic rates under high CO 2 partial pressures (Fig. 6). This is expected since C 4 photosynthesis rates are typically not limited by PEPc under these conditions (von Caemmerer, 2000). However, a large K HCO3 may limit rates of C 4 photosynthesis under low CO 2 partial pressures (Fig. 6), for example when reduced stomatal conductance limits CO 2 movement into the leaf. Flux control analysis found that PEPc has substantial control of C 4 photosynthesis under low CO 2 partial pressures (Dever et al., 1997;Bailey et al., 2000). Therefore, there is likely to be strong selective pressure to increase PEPc affinity for HCO 3 − in C 4 plants to increase the amount of C i entering C 4 photosynthesis. Our results, combined with those of others (Bläsing et al., 2000;Engelmann et al., 2002;Endo et al., 2008), show that as C 4 K HCO3 dropped, there was a concurrent increase in C 4 K PEP . Due to the higher PEP levels observed in leaves of C 4 plants compared with C 3 plants (Leegood and von Caemmerer, 1994), it can be argued that there was stronger selective pressure to increase PEPc affinity for HCO 3 − than to maintain high affinities for PEP since an increase in K PEP may not negatively impact C 4 photosynthesis rates to the extent that changes in K HCO3 can under lower CO 2 partial pressures (Fig. 6).

Conclusion
The direct comparison of closely related C 3 and C 4 PEPc isoforms from Flaveria demonstrates that the photosynthetic C 4 PEPc isoform has a significantly higher affinity for HCO 3 − than its closely related C 3 PEPc isoform. This reduced K HCO3 impacts net CO 2 assimilation rates, particularly at low CO 2 availability, suggesting selective pressure to reduce the C 4 K HCO3 to optimize inorganic carbon flux through C 4 photosynthesis. Alternatively, the increase in K PEP can be seen as strengthening the diurnal regulation of C 4 PEPc but residue S774 appears to link K HCO3 and K PEP , indicating that the increase in K PEP could be a negative consequence of reducing K HCO3 . Testing different plant species will provide new insights into which amino acids control K HCO3 and provide a better understanding of the structure and function relationship of the enzyme. Obtaining a better understanding of what controls K HCO3 will also lead to enhancing C 4 photosynthesis, particularly at low CO 2 partial pressures when stomata are partially closed. This raises interesting questions of whether there is a range in K HCO3 across the diverse lineages of C 4 plants and finding relationships between certain amino acid residues and ranges of K HCO3 values can be beneficial for promoting strategies to optimize C 4 photosynthesis in crop species for drought conditions.

Supplementary data
Supplementary data are available at JXB online. Table S1. Variable descriptions and values used to model C 4 photosynthesis. Table S2. Kinetic properties of PEPc isoforms from F. trinervia and F. pringlei. Table S3. Kinetic properties of PEPc isoforms in the presence of 2.5 mM malate. Fig. S1. Representative MIMS responses of the C 3 and C 4 PEPc activities with changing HCO 3 − concentrations. Fig. S2. HCO 3 − response curves for the C 3 , C 3 -A774S, C 4 , and C 4 -S774A PEPc isoforms in the presence of 2.5 mM malate.