Rubisco Catalytic Properties and Temperature Response in Crops

Variability in Rubisco kinetic parameters and their temperature dependency demonstrate differences in photosynthetic efficiency in the most important crops worldwide.

Abstract

Rubisco catalytic traits and their thermal dependence are two major factors limiting the CO₂ assimilation potential of plants. In this study, we present the profile of Rubisco kinetics for 20 crop species at three different temperatures. The results largely confirmed the existence of significant variation in the Rubisco kinetics among species. Although some of the species tended to present Rubisco with higher thermal sensitivity (e.g. Oryza sativa) than others (e.g. Lactuca sativa), interspecific differences depended on the kinetic parameter. Comparing the temperature response of the different kinetic parameters, the Rubisco K_m for CO₂ presented higher energy of activation than the maximum carboxylation rate and the CO₂ compensation point in the absence of mitochondrial respiration. The analysis of the Rubisco large subunit sequence revealed the existence of some sites under adaptive evolution in branches with specific kinetic traits. Because Rubisco kinetics and their temperature dependency were species specific, they largely affected the assimilation potential of Rubisco from the different crops, especially under those conditions (i.e. low CO₂ availability at the site of carboxylation and high temperature) inducing Rubisco-limited photosynthesis. As an example, at 25°C, Rubisco from Hordeum vulgare and Glycine max presented, respectively, the highest and lowest potential for CO₂ assimilation at both high and low chloroplastic CO₂ concentrations. In our opinion, this information is relevant to improve photosynthesis models and should be considered in future attempts to design more efficient Rubiscos.

The reported stagnation in the annual gains of cereal yields in the last decade clearly indicates that the expected demand for increased yield, at least 50% by 2050 (http://www.fao.org/economic/ess/ess-home/en/), will not be met by conventional breeding (Zhu et al., 2010). Future improvements will come from novel bioengineering approaches specifically focused on processes limiting crop productivity that have not been addressed so far (Parry and Hawkesford, 2012; Ort et al., 2015). A number of specific modifications to the primary processes of photosynthesis that could increase canopy carbon assimilation and production through step changes include the modification of the catalytic properties of Rubisco (Murchie et al., 2009; Whitney et al., 2011; Parry et al., 2013; Ort et al., 2015). First, biochemical models indicate that CO₂ fixation rates are limited by Rubisco activity under physiologically relevant conditions (Farquhar et al., 1980; von Caemmerer, 2000; Rogers, 2014). Second, Rubisco’s catalytic mechanism exhibits important inefficiencies that compromise photosynthetic productivity: it is a slow catalyst, forcing plants to accumulate large amounts of the protein, and unable to distinguish between CO₂ and oxygen, starting a wasteful side reaction with oxygen that leads to the release of previously fixed CO₂, NH₂, and energy (Roy and Andrews, 2000). These inefficiencies limit not only the rate of CO₂ fixation but also the capacity of crops for an optimal use of resources, principally water and nitrogen (Flexas et al., 2010; Parry and Hawkesford, 2012).

Rubisco kinetic parameters has been described in vitro at 25°C for about 250 species of higher plants, of which only approximately 8% are crop species (Yeoh et al., 1981; Bird et al., 1982; Sage, 2002; Ishikawa et al., 2009; Prins et al., 2016). These data revealed the existence of significant variability in the main Rubisco kinetic parameters both among C₃ species (Yeoh et al., 1980, 1981; Bird et al., 1982; Jordan and Ogren, 1983; Parry et al., 1987; Castrillo, 1995; Delgado et al., 1995; Kent and Tomany, 1995; Balaguer et al., 1996; Bota et al., 2002; Galmés et al., 2005, 2014a, 2014c; Ghannoum et al., 2005; Ishikawa et al., 2009) and between C₃ and C₄ species (Kane et al., 1994; Sage, 2002; Kubien et al., 2008; Perdomo et al., 2015b). The existence of Rubiscos with different catalytic traits implies that the success, in terms of photosynthetic improvement, of Rubisco engineering approaches in crops will depend on the specific performance of the native enzyme from each crop species. Nevertheless, our knowledge of the actual variability in Rubisco kinetics is still narrow, not only because of the limited number of species that have been examined so far but mainly because complete Rubisco kinetic characterization (including the main parameters) has been performed in very few species.

Recent modeling confirmed that Rubisco is not perfectly optimized to deliver maximum rates of photosynthesis and indicated that Rubisco optimization depends on the environmental conditions under which the enzyme operates (Galmés et al., 2014b). In particular, Rubisco catalytic parameters are highly sensitive to changes in temperature. For instance, the maximum carboxylase turnover rate (k_cat^c) increases exponentially with temperature (Sage, 2002; Galmés et al., 2015). However, at temperatures higher than the photosynthetic thermal optimum, the increases in k_cat^c are not translated into increased CO₂ assimilation because of the decreased affinity of Rubisco for CO₂ (i.e. higher Michaelis-Menten constant for CO₂ [K_c] and lower CO₂-oxygen specificity [S_c/o]) and the decreased CO₂-oxygen concentration ratio in solution (Hall and Keys, 1983; Jordan and Ogren, 1984). These changes favor the ribulose 1,5-bisphosphate (RuBP) oxygenation by Rubisco relative to carboxylation, increasing the flux through photorespiration and, ultimately, reducing the potential growth at high temperatures (Jordan and Ogren, 1984).

Beyond the discernment of the existing variability in Rubisco kinetics at a standard temperature, the knowledge of the temperature dependence of Rubisco kinetics and the existence of variability in the thermal sensitivity among higher plants are of key importance for modeling purposes. The number and diversity of plant species for which Rubisco kinetic parameters have been tested in vitro at a range of physiologically relevant temperatures are still very scarce (Laing et al., 1974; Badger and Collatz, 1977; Badger, 1980; Monson et al., 1982; Hall and Keys, 1983; Jordan and Ogren, 1984; Lehnherr et al., 1985; Uemura et al., 1997; Zhu et al., 1998; Sage et al., 2002; Galmés et al., 2005; Haslam et al., 2005; Yamori et al., 2006; Perdomo et al., 2015a; Prins et al., 2016) and mostly restricted to a few kinetic parameters; actually, there is no study examining the temperature dependencies of the main kinetic constants in the same species. The limited data reported so far suggest the existence of interspecific differences in the temperature dependence of some Rubisco kinetic parameters, like k_cat^c (Chabot et al., 1972; Weber et al., 1977; Sage, 2002) or S_c/o (Zhu et al., 1998; Galmés et al., 2005). Actually, differences in the energy of activation of k_cat^c and S_c/o seem to be ascribed to the thermal conditions typically encountered by the species in their native habitat (Galmés et al., 2005) as well as to the photosynthetic mechanism (Perdomo et al., 2015a).

The variability in the response of Rubisco kinetics to changes in temperature, if confirmed, is of paramount importance. The mechanistic models of photosynthesis at leaf, canopy, and ecosystem levels are based on the kinetic properties of Rubisco (Farquhar et al., 1980; von Caemmerer, 2000; Bernacchi et al., 2002), and the accuracy of these photosynthetic models depends on knowing the Rubisco kinetic parameters and the species-specific equations for the Rubisco-temperature dependencies (Niinemets et al., 2009; Yamori and von Caemmerer, 2009; Bermúdez et al., 2012; Díaz-Espejo, 2013; von Caemmerer, 2013; Walker et al., 2013). The need for estimations of the temperature dependencies of Rubisco kinetic parameters becomes timely as modelers try to predict the impact of increasing temperatures on global plant productivity (Sage et al., 2008; Gornall et al., 2010). Ideally, surveying variations in Rubisco kinetics and their temperature dependence should incorporate a correlative analysis with variations in the large (L)-subunit and/or small (S)-subunit amino acid sequence. Such a complementary research would permit deciphering what residue substitutions determine the observed variability in Rubisco catalysis.

In this study, we examined Rubisco catalytic properties and their temperature dependence in 20 crop species, thereby constituting the largest published data set of its kind. The aims of this work were (1) to compare the Rubisco kinetic parameters among the most economically important crops, (2) to search for differences in the temperature response of the main kinetic parameters among these species, (3) to test whether crop Rubiscos are optimally suited for the conditions encountered in plant chloroplasts, and (4) to unravel key amino acid replacements putatively responsible for differences in Rubisco kinetics in crops.

RESULTS

The Variability in Rubisco Kinetics at 25°C among the Most Relevant Crop Species

When considering exclusively the 18 C₃ crop species, at 25°C, the Rubisco Michaelis-Menten constant for CO₂ under nonoxygenic (K_c) and 21% O₂ (K_c^air) varied approximately 2-fold and 3-fold, respectively, and the k_cat^c varied approximately 2-fold (Table I). For K_c and K_c^air, Manihot esculenta presented the lowest values (K_c = 6.1 μm and K_c^air = 10.8 μm) and Spinacia oleracea presented the highest (K_c = 14.1 μm and K_c^air = 26.9 μm). Values for k_cat^c varied between 1.4 s⁻¹ (M. esculenta) and 2.5 s⁻¹ (Ipomoea batatas). The Rubisco S_c/o was the kinetic parameter with the lowest variation among the C₃ species (Supplemental Fig. S1) and ranged between 92.4 mol mol^–1 (Solanum lycopersicum) and 100.8 mol mol^–1 (Beta vulgaris and M. esculenta; Table I). Brassica oleracea and Glycine max presented the lowest value for the Rubisco carboxylase catalytic efficiency, calculated as k_cat^c/K_c (0.17 s⁻¹ μm⁻¹), and Coffea arabica presented the lowest value for the k_cat^c/K_c^air ratio (0.08 s⁻¹ μm⁻¹). With regard to the oxygenase catalytic efficiency (calculated as k_cat^o/K_o), S. oleracea displayed the lowest value (1.76 s⁻¹ nm⁻¹). Hordeum vulgare presented the highest values for the Rubisco carboxylase and oxygenase catalytic efficiencies (k_cat^c/K_c = 0.28 s⁻¹ μm⁻¹, k_cat^c/K_c^air = 0.17 s⁻¹ μm⁻¹, and k_cat^o/K_o = 3.01 s⁻¹ nm⁻¹).

Table I. Kinetic parameters of crop Rubiscos measured at 25°C.

Parameters included K_c and K_c^air, k_cat^c, S_c/o, and the carboxylation (k_cat^c/K_c and k_cat^c/K_c^air) and the oxygenation (k_cat^o/K_o) catalytic efficiencies. The k_cat^o/K_o ratio was calculated as (k_cat^c/K_c)/S_c/o × 1,000. For each species, data are means ± se (n = 3–9). Group averages were obtained from individual measurements in each species. Different letters denote statistical differences (P < 0.05) by Duncan analysis between C₃ and C₄ groups.

Species	Kc	Kcair	kcatc	Sc/o	kcatc/Kc	kcatc/Kcair	kcato/Ko
	μm	μm	s−1	mol mol−1	s−1 μm−1	s−1 μm−1	s−1 nm−1
C3 species
Avena sativa	10.8 ± 0.9	18.1 ± 2.0	2.3 ± 0.3	99.9 ± 3.0	0.21 ± 0.01	0.13 ± 0.03	2.14 ± 0.06
Beta vulgaris	10.8 ± 1.2	18.6 ± 1.1	2.0 ± 0.3	100.8 ± 2.0	0.19 ± 0.02	0.10 ± 0.01	1.94 ± 0.31
Brassica oleracea	11.8 ± 0.1	19.2 ± 0.3	2.1 ± 0.3	96.2 ± 1.3	0.17 ± 0.03	0.11 ± 0.02	1.81 ± 0.28
Capsicum annuum	9.6 ± 0.3	19.8 ± 1.5	1.9 ± 0.1	96.0 ± 4.5	0.20 ± 0.01	0.10 ± 0.01	1.98 ± 0.15
Coffea arabica	11.0 ± 0.4	22.9 ± 2.4	2.1 ± 0.2	98.7 ± 3.8	0.19 ± 0.02	0.08 ± 0.01	1.98 ± 0.18
Cucurbita maxima	9.0 ± 0.5	19.2 ± 1.0	2.2 ± 0.2	98.4 ± 0.4	0.25 ± 0.04	0.12 ± 0.01	2.45 ± 0.31
Glycine max	8.6 ± 0.2	16.2 ± 0.7	1.5 ± 0.1	97.0 ± 1.1	0.17 ± 0.02	0.09 ± 0.01	1.76 ± 0.21
Hordeum vulgare	9.0 ± 0.6	14.9 ± 1.6	2.4 ± 0.2	99.2 ± 3.8	0.28 ± 0.02	0.17 ± 0.03	3.01 ± 0.19
Ipomoea batatas	12.0 ± 0.7	21.1 ± 1.0	2.5 ± 0.1	98.5 ± 6.6	0.20 ± 0.00	0.12 ± 0.01	1.96 ± 0.08
Lactuca sativa	11.1 ± 0.3	18.2 ± 1.4	2.2 ± 0.1	94.0 ± 1.9	0.19 ± 0.00	0.12 ± 0.01	2.06 ± 0.07
Manihot esculenta	6.1 ± 0.2	10.8 ± 0.6	1.4 ± 0.1	100.8 ± 0.9	0.23 ± 0.02	0.13 ± 0.01	2.24 ± 0.17
Medicago sativa	9.7 ± 1.6	16.4 ± 1.9	1.7 ± 0.1	95.6 ± 2.2	0.20 ± 0.02	0.11 ± 0.01	2.23 ± 0.36
Oryza sativa	8.0 ± 0.4	17.3 ± 2.4	2.1 ± 0.3	93.1 ± 1.2	0.26 ± 0.04	0.14 ± 0.03	2.73 ± 0.43
Phaseolus vulgaris	7.8 ± 0.3	14.0 ± 1.0	1.7 ± 0.2	99.7 ± 2.7	0.22 ± 0.02	0.13 ± 0.02	2.11 ± 0.17
Solanum lycopersicum	9.7 ± 0.4	16.6 ± 1.4	2.3 ± 0.2	92.4 ± 2.3	0.24 ± 0.02	0.14 ± 0.01	2.48 ± 0.20
Solanum tuberosum	9.6 ± 0.2	18.0 ± 0.8	2.0 ± 0.3	95.4 ± 2.3	0.22 ± 0.05	0.12 ± 0.03	2.32 ± 0.46
Spinacia oleracea	14.1 ± 0.8	26.9 ± 0.8	2.4 ± 0.1	97.0 ± 1.2	0.18 ± 0.01	0.09 ± 0.01	1.76 ± 0.13
Triticum aestivum	11.3 ± 0.4	16.0 ± 0.6	2.2 ± 0.2	100.1 ± 1.8	0.20 ± 0.02	0.14 ± 0.01	2.08 ± 0.24
C3 average	10.0 ± 0.3a	18.0 ± 0.5a	2.1 ± 0.1a	97.5 ± 0.6a	0.21 ± 0.01a	0.12 ± 0.01a	2.17 ± 0.07a
C4 species
Saccharum × officinarum	26.3 ± 4.0	31.7 ± 2.1	3.9 ± 0.3	82.2 ± 1.8	0.15 ± 0.02	0.13 ± 0.01	1.82 ± 0.35
Zea mays	31.6 ± 1.8	42.0 ± 2.8	4.1 ± 0.6	87.3 ± 1.4	0.11 ± 0.02	0.07 ± 0.01	1.22 ± 0. 20
C4 average	27.6 ± 2.3b	36.1 ± 2.6b	4.0 ± 0.5b	84.4 ± 1.5b	0.13 ± 0.02b	0.10 ± 0.01a	1.52 ± 0.23b

When data from the two C₄ species (Saccharum × officinarum and Zea mays) were included in the comparison at 25°C, the range of variability increased for all parameters (Table I). Rubisco from the two C₄ species presented higher k_cat^c but lower affinity for CO₂ (i.e. higher K_c and K_c^air and lower S_c/o) than Rubisco from C₃ crops. On average, k_cat^c/K_c and k_cat^o/K_o of C₄ Rubiscos were 62% and 70% of those of C₃ crop Rubiscos, respectively.

The Temperature Response of Rubisco Kinetics in Crops and Tradeoffs between Catalytic Traits

Both the range of variation and the species showing the extreme values of Rubisco kinetics at 15°C and 35°C were similar to those described at 25°C, with some exceptions. As at 25°C, among the C₃ crops, Rubisco from M. esculenta presented the lowest values for K_c and K_c^air at 15°C and 35°C, while the highest values were measured on Rubisco from S. oleracea (Supplemental Table S1). The lowest and highest values for k_cat^c at 15°C were those of Rubisco from Cucurbita maxima and H. vulgare, respectively. The degree of dispersion of the data and the range of variation between the maximum and minimum values for K_c, K_c^air, and k_cat^c increased with the increment in the assay temperature (Supplemental Table S1; Supplemental Fig. S1). Regarding S_c/o, values ranged between 116.1 mol mol⁻¹ (B. oleracea) and 132.2 mol mol⁻¹ (C. maxima) at 15°C and between 74.2 mol mol⁻¹ (Oryza sativa) and 85 mol mol⁻¹ (M. esculenta) at 35°C (Supplemental Table S1). As for S_c/o, the range of variation for k_cat^c/K_c and k_cat^o/K_o also was narrowed with the increase in the assay temperature (Supplemental Table S1; Supplemental Fig. S1).

Integrating all data across three assay temperatures, k_cat^c correlated positively with K_c for both C₃ species (r² = 0.82, P < 0.001) and C₄ species (r² = 0.94, P < 0.001), with Rubisco from C₄ species showing higher K_c for a given k_cat^c than that from C₃ species (Fig. 1A). The low interspecific variability in S_c/o within each assay temperature determined a nonlinear relationship between k_cat^c and S_c/o when considering data from all temperatures together (Fig. 1B). At each temperature individually, Pearson’s correlation coefficients (PCCs) between k_cat^c and K_c and S_c/o were highly significant (Table II) when considering both C₃ and C₄ together. The results from the PICs analyses were in general more conservative compared with PCCs (Table II), and some significant correlations were lost with PICs (e.g. S_c/o versus K_c or K_c^air at 25°C). Notably, when excluding the two C₄ species, PCCs decreased in almost all correlations (Table II). Hence, the PCC between k_cat^c and K_c was no longer significant at 15°C, and the PCC between k_cat^c and S_c/o was significant only at 15°C. Furthermore, when considering only C₃ species, the unique significant PICs between k_cat^c and K_c and S_c/o were those found between k_cat^c and S_c/o at 15°C and 25°C.

Figure 1.

Relationship of k_cat^c with K_c (A) and the S_c/o (B). Black symbols correspond to C₃ species at 15°C (triangles), 25°C (circles), and 35°C (inverted triangles); white symbols correspond to C₄ species at 15°C (triangles), 25°C (circles), and 35°C (inverted triangles). Each symbol represents the average value of a single species per temperature interaction.

Table II. PICs (top part of the diagonals) and PCCs (bottom part of the diagonals) between the Rubisco kinetic parameters K_c, K_c^air, k_cat^c, and S_c/o at 15°C, 25°C, and 35°C when considering the 20 C₃ and C₄ species together and the 18 C₃ species alone.

Significant correlations are marked with asterisks: ***, P < 0.001; **, P < 0.01; and *, P < 0.05.

	15°C				25°C				35°C
	Kc	Kcair	kcatc	Sc/o	Kc	Kcair	kcatc	Sc/o	Kc	Kcair	kcatc	Sc/o
Data from C3 and C4 species analyzed together
Kc		0.826***	0.502*	−0.314		0.913***	0.819***	−0.202		0.960***	0.710***	−0.775***
Kcair	0.927***		0.036	−0.099	0.946***		0.683***	0.037	0.962***		0.707***	−0.660**
kcatc	0.810***	0.645**		−0.660**	0.941***	0.890***		−0.450*	0.894***	0.858***		−0.634**
Sc/o	−0.498*	−0.361	−0.673**		−0.772***	−0.699***	−0.749***		−0.806***	−0.737***	−0.736***
Data from C3 species alone
Kc		0.900***	0.194	0.120		0.646**	0.256	0.057		0.907***	0.363	−0.357
Kcair	0.892***		−0.118	0.394	0.829***		0.173	−0.006	0.816***		0.401	−0.155
kcatc	0.268	0.137		−0.787**	0.698***	0.587*		−0.470*	0.613**	0.476*		−0.285
Sc/o	0.025	0.145	−0.496*		−0.049	−0.162	−0.083		−0.386	−0.157	−0.259

The energy of activation (ΔH_a) for K_c varied between 38.2 kJ mol⁻¹ (Solanum tuberosum) and 83.1 kJ mol⁻¹ (O. sativa; Table III). I. batatas (40.7 kJ mol⁻¹) and M. esculenta (75.4 kJ mol⁻¹) were the species showing the lowest and highest values for ΔH_a of K_c^air. As for k_cat^c, ΔH_a varied between 27.9 kJ mol⁻¹ (H. vulgare) and 60.5 kJ mol⁻¹ (Medicago sativa). Although the range of variation across C₃ species was similar for the energies of activation of both K_c and k_cat^c (2.2-fold), nonsignificant correlation was observed between ΔH_a for K_c and ΔH_a for k_cat^c in both conventional and phylogenetically independent analyses (r² = 0.11 and 0.15, respectively, P > 0.05). The lowest and highest values for ΔH_a of the CO₂ compensation point in the absence of mitochondrial respiration (Γ*; calculated from S_c/o) were measured in B. vulgaris (19.8 kJ mol⁻¹) and G. max (26.5 kJ mol⁻¹), respectively. On average, Rubisco from C₃ crops presented significantly higher ΔH_a for K_c (60.9 ± 1.5 kJ mol⁻¹) and k_cat^c (43.7 ± 1.5 kJ mol⁻¹) than Rubisco from C₄ species (K_c = 52.4 ± 5 kJ mol⁻¹ and k_cat^c = 30.6 ± 1.6 kJ mol⁻¹). By contrast, nonsignificant differences were observed in the average ΔH_a for Γ* between C₃ species (22.9 ± 0.4 kJ mol⁻¹) and C₄ species (25 ± 0.7 kJ mol⁻¹).

Table III. ΔH_a (kJ mol^–1) and c (dimensionless) values of K_c (μmol mol^–1) and K_c^air (μmol mol^–1), k_cat^c (s^–1), and Γ* (μmol mol^–1) for the 20 crop species.

For each species, data are means ± se (n = 3–9). Group averages were calculated from individual measurements in each species. Different letters denote statistical differences (P < 0.05) by Duncan analysis between C₃ and C₄ groups. Parameter concentrations of K_c (μm) and K_c^air (μm) in liquid phase (Table I; Supplemental Table S1) were converted to gaseous phase partial pressures (K_c and/or K_c^air [μmol mol^–1] = parameter [μm] × K_h × air volume [L]/RT]. K_h is the hydrolysis constant (15°C = 22.2, 25°C = 29.4, and 35°C = 38.2). For the air volume (L): 15°C = 23.7, 25°C = 24.5, and 35°C = 25.4. The term Γ* (μmol mol ^–1) is derived from 0.5O/S_c/o, where O is the oxygen concentration.

Species	Kc		Kcair		kcatc		Γ*
	c	ΔHa	c	ΔHa	c	ΔHa	c	ΔHa
C3 species
A. sativa	31.3 ± 0.7	63.4 ± 2.0	26.0 ± 0.4	48.9 ± 1.2	17.6 ± 2.2	41.5 ± 5.5	13.3 ± 0.5	23.6 ± 1.4
B. vulgaris	28.7 ± 1.7	57.0 ± 4.4	27.2 ± 0.7	51.8 ± 1.8	21.5 ± 3.7	51.2 ± 9.6	11.7 ± 0.4	19.8 ± 1.0
B. oleracea	28.1 ± 1.1	55.3 ± 2.8	26.5 ± 0.9	50.1 ± 2.4	18.8 ± 2.6	45.7 ± 6.5	12.6 ± 0.2	21.8 ± 0.5
C. annuum	26.6 ± 1.5	51.8 ± 3.7	27.0 ± 1.6	51.2 ± 3.7	16.3 ± 2.8	39.2 ± 6.9	13.4 ± 0.7	24.1 ± 1.8
C. arabica	34.7 ± 0.3	71.5 ± 0.9	27.6 ± 1.8	52.2 ± 4.3	16.5 ± 2.6	39.0 ± 6.1	13.1 ± 0.5	23.4 ± 1.1
C. maxima	28.6 ± 0.8	57.0 ± 1.8	29.2 ± 1.1	56.8 ± 2.8	20.2 ± 1.0	48.7 ± 2.7	12.2 ± 0.9	21.1 ± 2.2
G. max	34.2 ± 0.5	71.1 ± 1.4	28.4 ± 1.2	55.3 ± 2.9	22.7 ± 2.5	55.2 ± 5.8	14.4 ± 1.7	26.5 ± 4.1
H. vulgare	31.1 ± 1.1	63.4 ± 3.0	30.7 ± 1.9	60.9 ± 5.0	12.2 ± 1.6	27.9 ± 4.0	12.3 ± 0.2	21.2 ± 0.6
I. batatas	23.0 ± 0.7	42.4 ± 1.6	22.7 ± 1.2	40.7 ± 3.1	14.3 ± 1.5	33.4 ± 3.8	13.0 ± 0.3	22.8 ± 0.8
L. sativa	28.3 ± 1.3	55.8 ± 3.2	29.0 ± 2.1	56.5 ± 5.2	14.1 ± 0.7	33.3 ± 1.7	12.3 ± 0.3	21.2 ± 0.9
M. esculenta	33.7 ± 1.4	70.8 ± 3.4	36.1 ± 1.1	75.4 ± 2.8	19.8 ± 1.6	47.4 ± 4.1	12.2 ± 0.2	21.1 ± 0.5
M. sativa	29.2 ± 1.3	58.8 ± 3.6	26.1 ± 0.4	49.5 ± 1.0	24.8 ± 1.1	60.5 ± 2.8	11.8 ± 0.2	20.1 ± 0.4
O. sativa	38.9 ± 0.8	83.1 ± 1.8	30.5 ± 1.2	60.5 ± 3.1	19.2 ± 1.8	46.4 ± 4.7	13.7 ± 0.5	24.6 ± 1.3
P. vulgaris	31.5 ± 0.8	64.6 ± 2.0	30.9 ± 2.7	61.7 ± 6.8	19.8 ± 2.1	47.7 ± 5.3	13.4 ± 0.6	24.1 ± 1.5
S. lycopersicum	30.8 ± 2.5	62.1 ± 6.3	36.0 ± 2.5	73.8 ± 6.4	14.7 ± 1.4	34.6 ± 3.6	12.5 ± 0.2	21.8 ± 0.5
S. tuberosum	21.1 ± 0.2	38.2 ± 0.5	24.4 ± 0.8	44.9 ± 1.9	19.2 ± 0.5	46.2 ± 1.1	13.7 ± 0.9	24.7 ± 2.2
S. oleracea	34.3 ± 0.8	69.9 ± 2.2	25.1 ± 0.5	45.6 ± 1.1	20.2 ± 0.7	48.0 ± 1.8	13.5 ± 0.3	25.2 ± 1.0
T. aestivum	30.1 ± 0.5	60.4 ± 2.2	34.4 ± 2.2	70.1 ± 5.4	17.4 ± 1.7	41.2 ± 4.3	13.5 ± 0.2	24.2 ± 0.4
C3 average	30.2 ± 0.6a	60.9 ± 1.5a	28.8 ± 0.6a	55.9 ± 1.5a	18.3 ± 0.6a	43.7 ± 1.5a	13.0 ± 0.2a	22.9 ± 0.4a
C4 species
S. officinarum	30.2 ± 1.9	58.3 ± 5.0	32.0 ± 1.0	62.3 ± 2.7	13.6 ± 1.5	30.2 ± 3.5	14.3 ± 0.6	25.8 ± 1.4
Z. mays	24.7 ± 3.4	44.5 ± 8.5	24.7 ± 3.4	44.5 ± 8.5	14.0 ± 0.9	31.0 ± 1.9	13.6 ± 0.1	24.3 ± 0.2
C4 average	27.9 ± 2.0a	52.4 ± 5.0b	28.9 ± 0.8a	53.7 ± 1.9a	13.7 ± 0.7b	30.6 ± 1.6b	14.0 ± 0.3a	25. 0 ± 0.7a

The CO₂ Assimilation Potential of Rubisco Kinetics in Crops

The CO₂ assimilation potential of Rubisco (A_Rubisco) was modeled at varying temperature and CO₂ availability at the catalytic site (chloroplastic CO₂ concentration [C_c]) using the species-specific kinetic data measured at each temperature (Table I; Supplemental Table S1). The simulated value of C_c = 250 μbar is representative of that encountered in the chloroplast stroma of C₃ species under well-watered conditions (Bermúdez et al., 2012; Scafaro et al., 2012; Galmés et al., 2013). Under mild to moderate water stress, when no metabolic impairment is present, the decrease in the stomatal and leaf mesophyll conductances to CO₂ provokes a decrease in the concentration of CO₂ in the chloroplast (Flexas et al., 2006). We selected a value of 150 μbar to simulate the C_c in water-stressed plants.

Differences in A_Rubisco across species were largely dependent on the temperature and the availability of CO₂ for carboxylation (Fig. 2). This fact was due to the different prevalence of RuBP-saturated (A_c) and RuBP-limited (A_j) CO₂ assimilation rates governing A_Rubisco under the contrasting temperature and C_c, assuming an invariable concentration of active Rubisco sites of 25 μmol m⁻² for all species. At 15°C, A_c limited A_Rubisco at C_c of 150 μbar in nine species (indicated by asterisks in Fig. 2). At 15°C and C_c of 250 μbar, only six species were A_c limited (Capsicum annuum, C. maxima, M. sativa, O. sativa, S. tuberosum, and S. oleracea). At 25°C and 35°C, A_Rubisco was A_c limited in all C₃ species irrespective of C_c.

Figure 2.

Simulated A_Rubisco for C₃ and C₄ species at 15°C, 25°C, and 35°C and at C_c values of 250 μbar (A) and 150 μbar (B). Equations used to calculate A_Rubisco were those described in the biochemical model of C₃ photosynthesis (Farquhar et al., 1980), as explained in “Materials and Methods.” Bars represent minimum values of A_c- and A_j-limited A_Rubisco. Asterisks beside the bars indicate A_c-limited A_Rubisco, and the absence of asterisks indicates A_j-limited A_Rubisco. The rates of electron transport were 60, 150, and 212 μmol m⁻² s⁻¹ at 15°C, 25°C, and 35°C, respectively. The concentration of active Rubisco sites was assumed to be invariable at 25 μmol m⁻² for all species and environmental conditions. The values used for the Rubisco kinetic parameters (k_cat^c, Γ*, and K_c^air) are those shown in Table I and Supplemental Table S1.

At 25°C, the best Rubisco was that from H. vulgare at both C_c values, while Rubisco from G. max yielded the lowest A_Rubisco (Fig. 2). Rubisco from B. vulgaris presented the best performance at 35°C irrespective of CO₂ availability, while C. annuum and S. officinarum Rubisco gave the lowest A_Rubisco at C_c of 250 and 150 μbar, respectively. At 15°C and C_c of 250 μbar, the highest potential for CO₂ assimilation was found in Rubisco from G. max, M. esculenta, and Triticum aestivum, while Rubisco from M. esculenta gave the highest A_Rubisco at 15°C and C_c of 150 μbar. Rubisco from C. maxima displayed the lowest A_Rubisco at 15°C, regardless of the CO₂ availability. It is interesting that Rubisco from the two C₄ species, in particular from S. officinarum, performed better than the average C₃ Rubiscos when A_Rubisco was simulated according to the photosynthesis model for C₃ leaves (Farquhar et al., 1980), at 15°C and 25°C under C_c of 250 μbar (Fig. 2A). At lower C_c (150 μbar), the C₄ Rubiscos yielded higher A_Rubisco values than the average C₃ Rubiscos at 15°C and lower values at 35°C, being similar at 25°C (Fig. 2B).

To test the performance of the different Rubiscos in the context of C₄ photosynthesis, A_c was also modeled assuming C_c of 5,000 μbar and concentration of active Rubisco sites (E) of 15 μmol m⁻². Under these conditions, the advantage of C₄-type Rubisco kinetics of S. officinarum and Z. mays, characterized by higher k_cat^c and K_c^air, became evident as providing higher A_c values at the three temperatures (data not shown). On average, at saturating CO₂ and lower concentration of Rubisco catalytic sites, C₄ Rubiscos yielded A_c of 35, 49, and 60 μmol m⁻² s⁻¹ at 15°C, 25°C, and 35°C, respectively, compared with the C₃ Rubisco averages (10, 23, and 42 μmol m⁻² s⁻¹, respectively).

Positively Selected L-Subunit Residues: Relationship with Rubisco Kinetics

The phylogeny obtained with rbcL, matK, and ndhF genes matched the currently accepted angiosperm classification (Supplemental Fig. S2; Bremer et al., 2009). When considering all species together, 10 L-subunit residues were under positive selection: 94, 262, 281, 309, 439, 446, 449, 470, 477, and amino acid insert between residues 468 and 469. Moreover, positive selection was identified in specific L-subunit residues along branches leading to species with high and low K_c, high k_cat^c, and low S_c/o at 25°C and low ΔH_a for K_c (Table IV). The residues under positive selection were located at different positions within the Rubisco tertiary structure and included functionally diverse sites participating in L-subunit intradimer and dimer-dimer interactions and interactions with S-subunits and with Rubisco activase (Table IV). No residue under positive selection was associated with ΔH_a for K_c^air, ΔH_a for k_cat^c, or ΔH_a for S_c/o.

Table IV. Amino acid replacements in the Rubisco L-subunit identified under positive selection by Bayes empirical Bayes analysis implemented in the PAML package (Yang et al., 2005; Yang, 2007) along branches of the phylogenetic tree leading to species with particular Rubisco properties.

Residuea	Amino Acid Changes	Location of Residue	Interactionb
Branches leading to species with Kc ≥ 26 μm and kcatc ≥ 3.9 s–1 at 25°C (C4 species)
94**	D, E, K → P		ID, RA
446**	R → K	C terminus
469**	Insert of G or T before residue 469	C terminus	ID
Branches leading to species with kcatc ≥ 2.5 s–1 at 25°C
281**	A → S	Helix 4	DD, SS
Branches leading to species with Kc ≥ 10.8 μm at 25°C
439***	A → T, V	Helix G
469*	Insert of G or T before residue 469	C terminus	ID
470*	A, E → K, P, Q	C terminus	ID
477**	S → E, G, P, Q	C terminus
Branches leading to species with Sc/o ≤ 94 mol mol–1 at 25°C
309**	M → I	βF-strand	ID
Branches leading to species with ΔHa for Kc ≤ 56 kJ mol–1
262**	V → A, T	Loop 3	S-subunit
439*	R → T, V	Helix G
449**	C, S, T → A	C terminus
477**	K → E, G, P, Q	C terminus

^aResidue numbering is based on the S. oleracea sequence. Values for Bayesian posterior probabilities are as follows: *, >0.90; **, >0.95; and ***, >0.99.

^bInteractions in which the selected residues and/or residues within 5 Å of them are involved: DD, dimer-dimer interactions; ID, intradimer interactions; RA, interface for interactions with Rubisco activase; SS, interactions with small subunits (Ott et al., 2000; Spreitzer and Salvucci, 2002; Du et al., 2003).

DISCUSSION

Main Crops Possess Rubiscos with Different Performance at 25°C

The kinetic data reported in this study are consistent with the range reported previously for higher plants at 25°C (Yeoh et al., 1980, 1981; Bird et al., 1982; Jordan and Ogren, 1983; Kent and Tomany, 1995; Galmés et al., 2005, 2014a, 2014c; Ishikawa et al., 2009; Prins et al., 2016; Table I) and showing the existence of significant variation among species in the carboxylase catalytic efficiency under nonoxygenic conditions (k_cat^c/K_c) and atmospheric conditions (k_cat^c/K_c^air). Recent reports related k_cat^c/K_c variation with the growth capacity in a group of closely related species with similar ecology (Galmés et al., 2014a), suggesting that improving this ratio would be an effective way to engineer a better Rubisco. Nevertheless, such an improvement becomes constrained by the tradeoffs between k_cat^c, K_c, and S_c/o (Tcherkez et al., 2006; Savir et al., 2010; Galmés et al., 2014a, 2014c). Here, we demonstrate that these tradeoffs, in particular k_cat^c versus K_c, hold when considering C₃ and C₄ species together, even after accounting for the phylogenetic signal in the data, and that they generally strengthen at increasing assay temperatures (Table II). However, most of these tradeoffs were lost when considering exclusively the C₃ species (Table II), indicative that the broad-scale patterns of covariation between the Rubisco kinetic parameters may not hold at smaller scales, as observed previously in other angiosperm species (Galmés et al., 2014c).

The k_cat^c from Z. mays and S. officinarum was 2-fold higher than that of the C₃ species, albeit at the expense of 3 times less affinity for CO₂ (Table I). This finding agrees with previously described trends between C₃ and C₄ species (Ghannoum et al., 2005; Kubien et al., 2008; Ishikawa et al., 2009) and with the fact that C₄ species present lower k_cat^c/K_c (Kubien et al., 2008; Perdomo et al., 2015a).

Unlike other reports (Sage, 2002; Ishikawa et al., 2009), the observed variation in the kinetic parameters at 25°C among C₃ species was apparently not related to the thermal climate of their respective domestication regions (data not shown). It should be noted that the origin, and hence the climatic conditions, of the selected varieties could be different from the species center of domestication and that the different crop varieties may have accumulated adaptive changes to local conditions by means of artificial selection (Meyer et al., 2012). Intraspecific variability in Rubisco catalytic traits has been reported in T. aestivum (Galmés et al., 2014c) and H. vulgare (Rinehart et al., 1983), but how this variability among genotypes is related to the adaptation of Rubisco to local environments remains elusive.

The Rubisco Kinetic Parameters of the Main Crops Present Different Thermal Sensitivities

The observed temperature response of the Rubisco kinetic parameters confirms well-described trends of increases in k_cat^c and K_c and a decrease in S_c/o with increasing assay temperature (Table I; Supplemental Table S1; Jordan and Ogren, 1984; Brooks and Farquhar, 1985; Uemura et al., 1997; Galmés et al., 2005; Prins et al., 2016).

The temperature dependency of full Rubisco catalytic constants was first provided for Nicotiana tabacum using in vivo-based leaf gas-exchange analysis (Bernacchi et al., 2001). After that report, all studies dealing with the temperature response of photosynthesis assumed the temperature dependency parameters of N. tabacum Rubisco, irrespective of the modeled species, from annual herbs to trees and from cold- to warm-adapted species (Pons et al., 2009; Keenan et al., 2010; Yamori et al., 2010; Galmés et al., 2011; Bermúdez et al., 2012; Scafaro et al., 2012). Importantly, our data set constitutes the most unequivocal confirmation that different temperature sensitivities of Rubisco kinetic parameters exist among different species and that extrapolating the temperature response of a unique model species to other plants induces errors when modeling the temperature response of photosynthesis. In this sense, the in vitro results of this study support in vivo data showing different temperature dependencies of Rubisco catalytic constants in Arabidopsis thaliana and N. tabacum (Walker et al., 2013).

In general, the Rubisco constant affinities for CO₂ (K_c and K_c^air) were more sensitive to changes in the assay temperature (i.e. presented higher ΔH_a) than k_cat^c and Γ* (Table III), in agreement with a recent study (Perdomo et al., 2015a). This fact is explained by the increase in the oxygenase catalytic efficiency (k_cat^o/K_o) at increasing temperature. However, it should be remarked that the k_cat^o/K_o ratio was calculated from the measured parameters K_c, k_cat^c, and S_c/o and that direct measurements of the oxygenase activity of Rubisco (e.g. by mass spectrometry; Cousins et al., 2010) should be undertaken to confirm this trend.

As at 25°C, the differences in the temperature dependencies of Rubisco kinetic parameters among C₃ species were not related to the thermal environment of a species’ domestication region (data not shown). This finding contrasts with previous evidence suggesting that the temperature sensitivity of Rubisco kinetic properties has evolved to improve the enzyme’s performance according to the prevailing thermal environment to which each species is adapted (Sage, 2002; Galmés et al., 2005, 2015).

Although only two C₄ species were included in this study, they presented lower ΔH_a for K_c and for k_cat^c than most of the C₃ species, in close agreement with trends observed recently by Perdomo et al. (2015a) in Flaveria spp. (Table III). A larger number of C₄ species need to be surveyed to verify the existence of differences in the temperature dependence of Rubisco kinetics between C₃ and C₄ species.

How Do the Species-Specific Properties of Rubisco Kinetics and Their Temperature Sensitivity Impact the Potential Capacity of Rubisco to Assimilate CO₂?

Modeling the effect of the species-specific Rubisco kinetics and temperature dependencies of Rubisco kinetics resulted in significant differences in A_Rubisco among the studied C₃ crops (Fig. 2). This modeling exercise highlighted which species would most benefit from the genetic replacement of their native version of Rubisco by other foreign versions with improved performance. Notably, the modeling results clearly indicate that the performance of specific Rubiscos cannot be evaluated without considering the environmental conditions during catalysis, specifically the temperature and the C_c. This fact results from the different temperature dependencies of Rubisco kinetics among crops and from the different impacts that Rubisco kinetics have on the A_c and A_j governing A_Rubisco. Hence, at 15°C and C_c of 250 μbar, A_Rubisco was limited by A_j in most C₃ species (12 out of 18), while it was limited by A_c in all C₃ species at 25°C and 35°C irrespective of the C_c value.

Detailed examination of modeled A_Rubisco suggests that future efforts to enhance Rubisco efficiency should be directed at the following C₃ species displaying the poorest performance: C. maxima and M. sativa at 15°C and both C_c values; G. max, C. annuum, and C. arabica at 25°C and 250 μbar; G. max, S. oleracea, C. annuum, and C. arabica at 25°C and 150 μbar; C. annuum, S. lycopersicum, and L. sativa at 35°C and 250 μbar; and C. annuum and S. lycopersicum at 35°C and 150 μbar.

In order to focus on the Rubisco catalytic traits, the modeling assumed invariable values for E (25 μmol m⁻² s⁻¹) and specific values for the rate of photosynthetic electron transport (J) and C_c. However, species adapt and plants acclimate to the prevailing thermal environment through changes in the concentration and/or activation of Rubisco and J (Yamasaki et al., 2002; Yamori et al., 2011). Similarly, stomatal and leaf mesophyll conductances to CO₂ also vary in response to temperature (von Caemmerer and Evans, 2015). The growth temperature effects on these parameters would have altered the equilibrium between A_c and A_j and, indirectly, the consequences of different Rubisco kinetic traits on the CO₂ assimilation potential. In the future, we aim to increase the accuracy of the present simulation by examining and including the species-specific values for stomatal and leaf mesophyll conductances, E, and J at varying environmental conditions.

The Analysis of Positive Selection in Branches Leading to Specific Rubisco Traits May Reveal Lineage-Specific Amino Acid Substitutions

We found 10 Rubisco L-subunit residues under positive selection (94, 262, 281, 309, 439, 446, 449, 469, 470, and 477; Table IV). With the exceptions of residues 469 and 477, these residues have been reported previously in other groups of plants, implying a relatively limited number of residues responsible for the Rubisco fine-tuning (Kapralov and Filatov, 2007; Christin et al., 2008; Iida et al., 2009; Kapralov et al., 2011, 2012; Galmés et al., 2014a, 2014c). However, despite widespread parallel evolution of amino acid replacements in the Rubisco sequence, solutions found in particular groups of plants may be quite different. For instance, there are only two common residues under positive selection out of 10 between this study and methodologically similar work with a different sampling design published earlier (Galmés et al., 2014c). This fact raises questions of epistatic interactions and residue coevolution within Rubisco (Wang et al., 2011) as well as residue coevolution and complementarity between Rubisco and its chaperones (Whitney et al., 2015), which both may prevent the evolution of identical amino acid replacements because of different genetic backgrounds.

We have not examined the species differences in the sequence of the Rubisco S-subunit. Some of the species included in this survey, like T. aestivum, possess a large number of S-subunit genes (rbcS) encoding different S-subunits (Galili et al., 1998). Previous reports have showed that species with identical L-subunits might have different Rubisco kinetics (Rosnow et al., 2015) and directly demonstrated that differences in the S-subunits might affect Rubisco catalytic traits (Ishikawa et al., 2011; Morita et al., 2014). Therefore, we cannot reject the idea that the observed differences in Rubisco kinetics, and their temperature dependence, among the studied crops are partially due to differences in the S-subunits.

CONCLUSION

This study confirms the significant variation in carboxylation efficiency and parameters that contribute to it among plant species and, to our knowledge, for the first time, provides full Rubisco kinetic profiles for the 20 most important crop species. Our data set could be used as an input for the next generation of species-specific models of leaf photosynthesis and its response to climate change, leading to more precise forecasts of changes in crop productivity and yield. These data could help to decide in which crops the CO₂ assimilation potential and carboxylation efficiency of Rubisco might be improved via reengineering of native enzymes or by replacement with foreign ones, as there is no a one-size-fits-all solution. The design of future attempts at Rubisco engineering in crops should be based on surveys of Rubisco catalytic and genetic diversity with a particular stress on relatives of the crops in question. Growing knowledge of the Rubisco catalytic spectrum combined with existing engineering toolkits for Rubisco (Whitney and Sharwood 2008) and its chaperones (Whitney et al., 2015) give us hope that the Rubisco efficiency, and hence the photosynthetic capacity, of crops could be improved in the near future.

MATERIALS AND METHODS

Species Selection and Growth Conditions

The following 20 crop species were selected for study: Avena sativa ‘Forridena’, Beta vulgaris ‘Detroit’, Brassica oleracea var italica ‘Calabres’, Capsicum annuum ‘Picante’, Coffea arabica ‘Catuaí Vermelho IAC 44’, Cucurbita maxima ‘Totanera’, Glycine max ‘Ransom’, Hordeum vulgare ssp. vulgare ‘Morex’, Ipomoea batatas ‘Rosa de Málaga’, Lactuca sativa ‘Cogollo de Tudela’, Manihot esculenta, Medicago sativa ‘Aragón’, Oryza sativa ‘Bomba’, Phaseolus vulgaris ‘Contender’, Saccharum × officinarum (hybrid between Saccharum officinarum and Saccharum spontaneum), Solanum lycopersicum ‘Roma VF’, Solanum tuberosum ‘Erlanger’, Spinacia oleracea ‘Butterfly’, Triticum aestivum ‘Cajeme’, and Zea mays ‘Carella’. These species represent the most important crops in terms of worldwide production (http://faostat.fao.org/). C. arabica was selected as being the most important commodity in the international agricultural trade (DaMatta, 2004). Plants were grown from seeds under natural photoperiods in a glasshouse at the University of the Balearic Islands (Spain) during 2011 and 2012. Plants were grown in soil-based compost supplemented with slow-release fertilizer and frequently watered to avoid water stress. The air temperature in the glasshouse during the growth period was maintained between 15°C and 30°C.

Determination of K_c and k_cat^c

Rubisco K_c and K_c^air were determined in crude extracts obtained as detailed by Galmés et al. (2014a). Rates of ¹⁴CO₂ fixation were measured at 15°C, 25°C, and 35°C using activated protein extracts in 7-mL septum-capped scintillation vials containing reaction buffer (100 mm Bicine-NaOH, pH 8, 20 mm MgCl₂, 0.4 mm RuBP, and about 100 W-A units of carbonic anhydrase) equilibrated previously with either N₂ or a mixture of oxygen and N₂ (21:79). Nine different concentrations of H¹⁴CO₃^– (0.1 to 9.4 mm, each with a specific radioactivity of 3.7 × 10¹⁰ Bq mol⁻¹) were prepared in the scintillation vials as described previously (Galmés et al., 2014a). Assays at 35°C using Rubisco from C₄ species required increasing H¹⁴CO₃^– up to 17.7 mm to reach saturating CO₂ concentration in the aqueous phase. Assays were started by the addition of 10 μL of protein extract and stopped after 1 min by injection of 0.1 mL of 10 m formic acid. Acid-stable ¹⁴C was determined by liquid scintillation counting (LS 6500 Multi-Purpose Scintillation Counter; Beckman Coulter) following the removal of acid-labile ¹⁴C by evaporation. K_c and K_c^air were determined from the fitted data as described elsewhere (Bird et al., 1982). Replicate measurements (n = 3–6) were made using different biological replicates for each species.

To obtain k_cat^c, the maximum rate of carboxylation was extrapolated from the Michaelis-Menten fit and divided by the number of Rubisco active sites in solution, quantified by [¹⁴C]CABP binding (Yokota and Canvin, 1985). Additional control assays undertaken as detailed by Galmés et al. (2014a) confirmed that the observed acid-stable ¹⁴C signal was uniquely the result of Rubisco catalytic activity.

Determination of S_c/o

Rubisco S_c/o was measured on purified extracts obtained as described by Gago et al. (2013). On the day of S_c/o measurement, highly concentrated Rubisco solutions were desalted by centrifugation through G-25 Sephadex columns equilibrated previously with CO₂-free 0.1 m Bicine (pH 8.2) containing 20 mm MgCl₂. The desalted solutions were made 10 mm with NaH¹⁴CO³ (1.85 × 10¹² Bq mol⁻¹) and 4 mm NaH₂PO₄ to activate Rubisco by incubation at 37.5°C for 40 min. Reaction mixtures were prepared in oxygen electrodes (Oxygraph; Hansatech Instruments) by first adding 0.95 mL of CO₂-free assay buffer (100 mm Bicine, pH 8.2, 20 mm MgCl₂, containing 0.015 mg of carbonic anhydrase). After the addition of 0.02 mL of 0.1 m NaH¹⁴CO₃ (1.85 × 10¹² Bq mol⁻¹), the plug was fitted to the oxygen electrode vessel and enough activated Rubisco (20 μL) was added. The reaction was started by the injection of 10 μL of 25 mm RuBP to be completed between 2 and 7 min depending on the assay temperature. RuBP oxygenation was calculated from the oxygen consumption, and carboxylation was calculated from the amount of ¹⁴C incorporated into PGA when all the RuBP had been consumed (Galmés et al., 2014a). Measurements were performed at 15°C, 25°C, and 35°C, with three to nine biological replicates for each species and assayed temperature.

For all Rubisco assays, pH of the assay buffers was accurately adjusted at each temperature of measurement. The concentration of CO₂ in solution in equilibrium with HCO₃⁻ was calculated assuming a pK_a for carbonic acid of 6.19, 6.11, and 6.06 at 15°C, 25°C, and 35°C, respectively. The concentration of oxygen in solution was assumed to be 305.0, 253.4, and 219.4 nmol mL⁻¹ at 15°C, 25°C, and 35°C, respectively (Truesdale and Downing, 1954).

Temperature Dependence Parameters of Rubisco Kinetics

To determine the temperature response of the Rubisco kinetic parameters from each species, values for K_c, K_c^air, and S_c/o were first converted from concentrations to partial pressures. For this, solubilities for CO₂ were considered to be 0.045, 0.034, and 0.0262 mol L⁻¹ bar⁻¹ at 15°C, 25°C, and 35°C, respectively. In turn, solubilities for oxygen of 0.0016, 0.0013, and 0.0011 mol L⁻¹ bar⁻¹ were used at 15°C, 25°C, and 35°C, respectively. The Γ* was obtained from S_c/o as described by von Caemmerer (2000) using the above solubilities for oxygen. Thereafter, values of K_c, Γ*, and k_cat^c at the three temperatures were fitted to an Arrhenius-type equation (Badger and Collatz, 1977; Harley and Tenhunen, 1991):

(1)

where c is a scaling constant, R is the molar gas constant (8.314 J K⁻¹ mol⁻¹), and T_k is the absolute assay temperature.

CO₂ Assimilation Potential of Crop Rubiscos at Varying Temperatures and CO₂ Availability

According to the biochemical model of C₃ photosynthesis (Farquhar et al., 1980), A_Rubisco is defined as the minimum of A_c and A_j:

(2)

(3)

(4)

A_Rubisco was obtained for each species at three different temperatures (15°C, 25°C, and 35°C) and two different C_c values (150 and 250 μbar), simulating situations of moderate water-stress and well-watered conditions in C₃ plants, respectively (Flexas et al., 2006). The Rubisco catalytic traits k_cat^c, Γ*, and K_c^air were taken from the species- and temperature-specific data obtained in this study. E was assumed to be invariable at 25 μmol m⁻². Values of J were assumed to be 60, 150, and 212 μmol m⁻² s⁻¹ at 15°C, 25°C, and 35°C, respectively, for all species. At 25°C, J = 150 μmol m⁻² s⁻¹ matches very well with a J/(k_cat^c × E) ratio of 1.5 (Egea et al., 2011). Values for J at 15°C and 35°C were obtained from the J temperature response described for Nicotiana tabacum by Walker et al. (2013).

Analysis of Rubisco L-Subunit Sites under Positive Selection

Full-length DNA sequences of the Rubisco L-subunit-encoding gene, rbcL (Supplemental Fig. S3), and two additional chloroplast genes (matK and ndhF) were obtained from GenBank (http://www.ncbi.nlm.nih.gov/genbank/) for the 20 studied species.

DNA sequences were translated into protein sequences for alignment using MUSCLE (Edgar, 2004). The software MODELTEST 3.7 (Posada and Crandall, 1998; Posada and Buckley, 2004) was used to check for the best model before running the phylogenetic analyses. The species phylogeny was reconstructed using concatenated alignment of all three chloroplast genes and maximum-likelihood inference conducted with RAxML version 7.2.6 (Stamatakis, 2006).

Amino acid residues under positive selection were identified using codon-based substitution models in comparative analysis of protein-coding DNA sequences within the phylogenetic framework (Yang, 1997). Given the conservative assumption of no selective pressure at synonymous sites, codon-based substitution models assume that codons with a ratio of nonsynonymous to synonymous substitution rate (d_N/d_S) < 1 evolve under purifying selection to keep protein function and properties, while codons with d_N/d_S > 1 evolve under positive Darwinian selection to modify properties of the given protein (Yang, 1997).

The codeml program in the PAML version 4.7 package (Yang, 2007) was used to perform branch site tests of positive selection along prespecified foreground branches (Yang et al., 2005; Yang, 2007). The codeml A model allows 0 ≤ d_N/d_S ≤ 1 and d_N/d_S = 1 for all branches. d_N/d_S > 1 is permitted only along prespecified foreground branches, and 0 ≤ d_N/d_S ≤ 1 and d_N/d_S = 1 is permitted on background branches. Branches leading to species with high or low K_c, k_cat^c, S_c/o, and ΔH_a were marked as foreground branches. For the purpose of these tests, high or low K_c, k_cat^c, and S_c/o ranges were taken only at 25°C, because of high correlation between values for these kinetic parameters obtained at three different temperatures. ΔH_a for these kinetic parameters also was considered. The A model was used to identify the amino acid sites under positive selection and to calculate the posterior probability of an amino acid belonging to a class with d_N/d_S > 1 using the Bayes empirical Bayes approach implemented in PAML (Yang et al., 2005).

The Rubisco L-subunit residues were numbered based on the S. oleracea sequence. The location of sites under positive selection was done using the Rubisco protein structure from S. oleracea obtained from the RCSB Protein Data Bank (http://www.rcsb.org; file 1RCX; Karkehabadi et al., 2003).

Statistical Analysis

Statistical analysis consisted of one-way ANOVA and correlation for linear regressions. For all the parameters studied, a univariate model of fixed effects was assumed. The univariate general linear model for unbalanced data (Proc. GLM) was applied, and significant differences among species and groups of species were revealed by Duncan tests using IBM SPSS Statistics for Macintosh, version 21.0. The relationships among the kinetic parameters and the temperature dependence parameters were tested with the square of the correlation coefficient observed for linear regressions using the tool implemented in R 3.1.1 (http://www.R-project.org). All statistical tests were considered significant at P < 0.05.

The PCC was calculated between pairwise combinations of the kinetic parameters K_c, K_c^air, k_cat^c, and S_c/o at the three temperatures of measurement. However, correlations arising within groups of related taxa might reflect phylogenetic signal rather than true cause-effect relationships, because closely related taxa are not necessarily independent data points and could violate the assumption of randomized sampling employed by conventional statistical methods (Felsenstein, 1985). To overcome this issue, tests were performed for the presence of phylogenetic signal in the data, and trait correlations were calculated with PICs using the AOT module of PHYLOCOM (Webb et al., 2008) using the species phylogeny based on the three chloroplast genes (see below). All these tests were considered significant at P < 0.05.

Accession Numbers

Sequence data from this article can be found in the GenBank/EMBL data libraries under accession numbers.

Supplemental Data

The following supplemental materials are available.

• Supplemental Figure S1. Box-plot depiction of Rubisco kinetic parameters (K_c, K_c^air, k_cat^c, and S_c/o) at 15°C, 25°C, and 35°C when considering the 18 C₃ species alone.

• Supplemental Figure S2. Maximum-likelihood phylogeny created using rbcL, matK, and ndhF for the selected crop species.

• Supplemental Figure S3. Rubisco L-subunit amino acid alignment for the 20 crop species used in this study.

• Supplemental Table S1. Rubisco kinetic parameters measured at 15°C and 35°C for the selected crop species.

Glossary

k_cat^c

maximum carboxylase turnover rate

K_c

Michaelis-Menten constant for CO₂

S_c/o

CO₂-oxygen specificity

K_c^air

Michaelis-Menten constant for CO₂ under 21% oxygen

PICs

phylogenetically independent contrasts

PCC

Pearson’s correlation coefficient

ΔH_a

energy of activation

Γ*

CO₂ compensation point in the absence of mitochondrial respiration

A_Rubisco

CO₂ assimilation potential of Rubisco

C_c

chloroplastic CO₂ concentration

RuBP

ribulose 1,5-bisphosphate

A_c

RuBP-saturated CO₂ assimilation rate

A_j

RuBP-limited CO₂ assimilation rate

E

concentration of active Rubisco sites

J

photosynthetic electron transport rate

d_N/d_S

ratio of nonsynonymous to synonymous substitution rate