C₄ Photosynthesis at Low Temperature. A Study Using Transgenic Plants with Reduced Amounts of Rubisco

Abstract

C₄ plants are rare in the cool climates characteristic of high latitudes and elevations, but the reasons for this are unclear. We tested the hypothesis that CO₂ fixation by Rubisco is the rate-limiting step during C₄ photosynthesis at cool temperatures. We measured photosynthesis and chlorophyll fluorescence from 6°C to 40°C, and in vitro Rubisco and phosphoenolpyruvate carboxylase activity from 0°C to 42°C, in Flaveria bidentis modified by an antisense construct (targeted to the nuclear-encoded small subunit of Rubisco, anti-RbcS) to have 49% and 32% of the wild-type Rubisco content. Photosynthesis was reduced at all temperatures in the anti-Rbcs plants, but the thermal optimum for photosynthesis (35°C) did not differ. The in vitro turnover rate (kcat) of fully carbamylated Rubisco was 3.8 mol mol^–1 s^–1 at 24°C, regardless of genotype. The in vitro kcat (Rubisco Vcmax per catalytic site) and in vivo kcat (gross photosynthesis per Rubisco catalytic site) were the same below 20°C, but at warmer temperatures, the in vitro capacity of the enzyme exceeded the realized rate of photosynthesis. The quantum requirement of CO₂ assimilation increased below 25°C in all genotypes, suggesting greater leakage of CO₂ from the bundle sheath. The Rubisco flux control coefficient was 0.68 at the thermal optimum and increased to 0.99 at 6°C. Our results thus demonstrate that Rubisco capacity is a principle control over the rate of C₄ photosynthesis at low temperatures. On the basis of these results, we propose that the lack of C₄ success in cool climates reflects a constraint imposed by having less Rubisco than their C₃ competitors.

C₄ plants often dominate the warm climate regions of the earth when they have access to at least moderate light intensities (Sage et al., 1999). Conversely, C₄ species are relatively rare in the cool climates characteristic of high latitudes or high elevations (Teeri and Stowe, 1976; Tieszen et al., 1979; Rundel, 1980; Long, 1983; Sage et al., 1999). In North America, temperature is the best predictor of the success of C₄ grasses, which rarely occur when the minimal temperature of the warmest month of the growing season is below 8°C (Teeri and Stowe, 1976). Globally, C₄ plants are rare at latitudes and elevations where the average growing season temperatures are less than approximately 16°C (Sage et al., 1999). The transition from C₄- to C₃-dominated landscapes generally occurs between 30°N and 40°N and between 1,500 and 3,000 m elevation (Sage et al., 1999).

The reason for the relative lack of C₄ plants in cool regions remains unclear. The lower quantum yield of photosynthesis (ϕ_CO2, the initial slope of the light-response curve) in C₄ versus C₃ species at low temperatures has been proposed to account for the differences in the biogeography of the two pathways (Ehleringer and Björkman, 1977; Ehleringer, 1978). However, differences in quantum yield only effect carbon uptake when light is limiting, and in plant canopies CO₂ assimilation is a function of both light-saturated and light-limited photosynthetic rates (Ehleringer, 1978; Long, 1999). Alternatively, one or more of the carbon-concentrating reactions of the mesophyll might be inherently chilling sensitive. Pyruvate orthophosphate dikinase (PPDK, EC 2.7.9.1) and phosphoenolpyruvate (PEP) carboxylase (PEPCase, EC 4.1.1.31) can both dissociate below 8°C to 12°C in vitro (Sugiyama and Boku, 1976; Edwards et al., 1985; Krall and Edwards, 1993). However, considerable species and ecotypic variability has been noted, and these steps are not fundamentally prone to failure at low temperatures in vivo (Leegood and Edwards, 1996; Matsuba et al., 1997).

The possibility that the bundle sheath reactions may limit C₄ photosynthesis at suboptimal temperatures has received less attention. Björkman and Pearcy (1971) found that the activation energy (Ea) of photosynthesis was similar to that of Rubisco (EC 4.1.1.39) in several warm-climate C₃ and C₄ species. Variation in Rubisco activity was correlated with differences in the photosynthetic capacity of Atriplex lentiformis (Torr.) Wats. grown at different temperatures (Pearcy, 1977). In Bouteloua gracilis Lag. and Muhlenbergia montanum (Nutt.) A.S. Hitch., the light-saturated net photosynthetic rate and the maximal in vitro Rubisco activity are equivalent below 17°C and 22°C, respectively (Pittermann and Sage, 2000, 2001). These findings indicate that Rubisco capacity can limit the rate of C₄ photosynthesis below approximately 20°C.

A Rubisco limitation can be quantified if the amount of the enzyme is changed without affecting the activities of other enzymes (Stitt and Schulze, 1994; Stitt, 1995). This enables determination of a flux control coefficient for Rubisco (C_ra) within the context of the carbon assimilation pathway (Kacser and Burns, 1973; Stitt, 1995). A C_ra of 1 indicates complete control of photosynthetic flux by Rubisco; a 10% change in the amount of enzyme will result in a 10% change in the flux through the pathway. A control coefficient between zero and one would indicate that the control of photosynthesis is shared between Rubisco and other processes. To determine C_ra, a series of plants with differences in the content of Rubisco is needed (Stitt and Schulze, 1994). Antisense molecular techniques provide a means of meeting this requirement.

In the C₄ dicot Flaveria bidentis L. Kuntze, Rubisco content has been reduced with antisense-RNA constructs targeting the nuclear-encoded small-subunit of Rubisco (RbcS; Chitty et al., 1994; Furbank et al., 1996, 1997). The amounts of other photosynthetic enzymes are not effected by the transformation (Furbank et al., 1996). Reducing Rubisco by antisense in F. bidentis results in plants that have reduced steady-state photosynthetic rates under a range of conditions (Furbank et al., 1996; Siebke et al., 1997; von Caemmerer et al., 1997). Anti-RbcS F. bidentis has been used to show that Rubisco is an important control over C₄ photosynthesis at temperatures near the thermal optimum, where C_ra was as high as 0.7 (Furbank et al., 1997). The effect of temperature variation on this control is unknown.

In the present study, we tested the hypothesis that Rubisco is the primary rate-limiting step during C₄ photosynthesis at low temperatures. We used three F. bidentis genotypes with a 3-fold difference in Rubisco content. To examine the nature of the rate limitation during the temperature response of C₄ photosynthesis, we measured gas exchange and chlorophyll a fluorescence across a range of temperatures from 6°C to 40°C and the in vitro activities of Rubisco and PEPCase from 0°C to 42°C. To quantify the control of CO₂ assimilation by Rubisco and whether this control is affected by temperature, we determined the C_ra at temperatures ranging from 6°C to 40°C.

RESULTS

The antisense constructs led to significant reductions in the amount of Rubisco present in F. bidentis leaves (Table I). The 136-13 and 141-1 anti-RbcS lines had 49% and 32% of the wild-type Rubisco content, respectively. The Rubisco turnover rate (kcat) did not differ between any of the lines, indicating that the antisense constructs had no effect on the kinetics of the enzyme. There was no difference in the activity of PEPCase between the genotypes (Table I). In each genotype, the Arrhenius plots for Rubisco and PEPCase indicated enzyme dissociation at low temperatures in vitro (Table I). The Ea of Rubisco increased from approximately 57 to 100 kJ mol^–1 between 12°C and 18°C, whereas the Ea of PEPCase increased from 71 to 180 kJ mol^–1 at similar temperatures. The wild-type plants had higher chlorophyll content than the anti-RbcS lines, but there were no changes in the ratio of chlorophyll a/b (Table I).

Table I.

Biochemical characteristics of Flaveria bidentis wild type and anti-RbcS plants

The concentration of Rubisco active sites was determined by the ¹⁴CABP-labeling assay described in “Materials and Methods.” The kcat and enzyme activity data are values at 24°C. The Rubisco kcat and PEP carboxylase activity values were determined by the incorporation of ¹⁴C into acid-stable products. Activation energies were calculated as described by Berry and Raison (1981) across the temperature ranges indicated by superscript letters. Chlorophyll determinations follow Porra et al. (1989). Each value represents the mean (± se) of four or five measurements. Values with different capital letters are statistically different from each other (P < 0.05, Tukey).

	Rubisco			PEP Carboxylase		Gross CO2 Assimilation	Chlorophyll
Plant	Catalytic Sites	kcat	Ea	Activity	Ea	Eaa	Content	a/b
	μmol m-2	mol mol-1 s-1	kj mol-1	μmol m-2 s-1	kj mol-1	kj mol-1	μmol m-2
Wild type	13.2 ± 0.9A	3.9 ± 0.3	100.6 ± 2.0b	159.9 ± 6.8	175.2 ± 3.8b	72.7 ± 1.4A	592 ± 59A	3.7 ± 0.1
			56.1 ± 1.4c		71.6 ± 1.0c
anti-RbcS 136-13	6.5 ± 0.9B	3.8 ± 0.2	97.4 ± 6.1b	186.3 ± 13.5	193.6 ± 3.8b	87.6 ± 1.6B	453 ± 37B	3.7 ± 0.2
			57.0 ± 2.1c		69.4 ± 3.4c
anti-RbcS 141-1	4.2 ± 0.5B	3.7 ± 0.1	100.9 ± 2.4b	146.9 ± 13.3	183.4 ± 6.6b	96.2 ± 2.6C	469 ± 41AB	3.8 ± 0.2
			59.3 ± 1.1c		74.0 ± 1.9c

^aMeasured between 5°C and 30°C. ^b Measured between 0°C and 12°C. ^c Measured between 18°C and 42°C.

Reducing the amount of Rubisco by antisense led to a large reduction in the net CO₂ assimilation rate (A) relative to the wild type at all measurement temperatures (Fig. 1a). Furthermore, the ratio of wild type to antisense photosynthesis increased as leaf temperature was reduced. F. bidentis had a photosynthetic thermal optimum of about 35°C when grown under these conditions, regardless of genotype (Fig. 1a). Dark respiration (Rd) did not vary with genotype (P = 0.67, ANOVA; Fig. 1a). The Ea of net photosynthesis was about 20% higher in the anti-RbcS lines (Table I). The antisense lines had lower stomatal conductance than the wild type below 20°C (Fig. 1b) but maintained a higher intercellular CO₂ (Ci) across the range of temperatures measured, due to their greatly reduced photosynthetic rates (Fig. 1c). The Ci corresponding to an ambient CO₂ of 370 μbar was sufficient to saturate photosynthesis at each measurement temperature in all genotypes (data not shown). Stomatal conductance was relatively stable in each genotype below 20°C, and Ci rose markedly as temperature declined below this point.

Figure 1.

The temperature responses of the rates of net CO₂ assimilation and Rd (a), stomatal conductance (b), and the partial pressure of Ci (c) in F. bidentis wild type (•) and anti-RbcS (○ and ). Photosynthesis was measured at a temperature- and genotype-dependent PPFD that was just sufficient to saturate photosynthesis at 370 μbar CO₂ and 200 mbar O₂. Each point represents the mean (± se) of measurements on five different leaves. Rd was determined as the y intercept of the light response of photosynthesis in the wild-type and 141-1 lines. Respiration did not differ between genotypes, and pooled values are shown here. The relationship between Rd and temperature is described by Rd = 3.76e^–3 T² + 5.74e^–4 T + 8.56e^–9 (where T is leaf temperature [°C], R² = 0.927). This was used to correct the net assimilation for respiration in all ensuing calculations.

At 30°C, the ratios of variable to maximal fluroescence (F_v/F_m) were 0.80 ± 0.01 and 0.82 ± 0.01 (+ se; n = 8) for the wild type and 141-1 lines, respectively. At a light intensity of 1,500 μmol m^–2 s^–1, the wild-type leaves had a higher quantum yield of photosystem II (PSII; Φ_PSII) than the anti-RbcS leaves at temperatures below the thermal optimum (Fig. 2a). The same pattern was detected when the temperature curves were measured under illumination that was just saturating, although in that case Φ_PSII at low temperatures (<15°C) was about 15% higher than the data obtained at a constant PPFD of 1,500 μmol m^–2 s^–1 (data not shown). The wild-type leaves maintained a greater proportion of open PSII than the antisense plants below 25°C, as indicated by the higher photochemical quenching (qP) values (Fig. 2b); at higher temperatures there was no difference between the two groups. There were no statistical differences in non-photochemical quenching between the two genotypes (Fig. 2c).

Figure 2.

The temperature responses of the quantum yield of PSII (a; Φ_PSII), photochemical quenching (b; qP), and non-photochemical quenching (c; qN) in F. bidentis wild type (•) and anti-RbcS (141-1, ○). Measurements were made at a constant PPFD of 1,500 μmol m^–2 s^–1, 370 μbar CO₂, and 200 mbar O₂. Each point represents the mean (± se) of measurements on three different leaves.

The instantaneous quantum requirement of PSII per CO₂ (Φ_PSII/Φ_CO2*) was constant in the wild-type leaves above 25°C and increased at lower temperatures (Fig. 3a). The anti-RbcS leaves had a higher Φ_PSII/Φ_CO2* than the wild type at all temperatures. The quantum requirement of photosynthesis in F. bidentis was sensitive to temperature, increasing more than 3-fold as temperature was reduced from 40°C to 10°C (Fig. 3a). The increase in the quantum requirement for CO₂ fixation with declining leaf temperature is consistent with increased leakage (ϕ) of CO₂ from the bundle sheath (Fig. 3b).

Figure 3.

a, The ratio of the quantum yields of PSII (Φ_PSII) and gross CO₂ assimilation (Φ_CO2*) in wild-type (•) and anti-RbcS (141-1, ○) F. bidentis as a function of temperature. b, Modeled leakiness (ϕ) of CO₂ from the bundle sheath, relativized to the estimated value for the wild type at the thermal optimum (35°C). Chlorophyll a fluorescence was measured in the 660- to 710-nm waveband, thereby reducing the contribution of PSI; Φ_PSII was assessed using the technique of Genty et al. (1989). Gross assimilation was determined using the temperature correction for respiration described in Figure 1. Each value is the mean (± se) of measurements on three different leaves. Leakiness (ϕ) was determined from the relationship Φ_PSII/Φ_CO2* = 4 + 2.66m/(1 – ϕ), assuming equal contributions of linear and cyclic electron transport (e.g. m = 0.5) and that 43% of quanta are absorbed by PSII (Siebke et al., 1997). Leakiness values are shown relative to the wild-type value at the thermal optimum (35°C) because of the potential uncertainty surrounding the calculation of ϕ from fluorescence data. At 35°C, ϕ in the wild type was determined to be 0.39 ± 0.09 (n = 3).

The in vitro kcat increased with increasing assay temperature, but there were no differences between the genotypes (Fig. 4a). Dividing the gross CO₂ assimilation rate by the concentration of Rubisco catalytic sites yielded the in vivo kcat (Fig. 4b). At 15°C or lower, in vitro and in vivo kcat were the same in each genotype. Above 20°C, in vitro kcat exceeded the in vivo value in each genotype. The in vivo kcat in the anti-RbcS lines was greater than the wild-type value above 30°C (Fig. 4b).

Figure 4.

The temperature dependence of the in vitro (a), and in vivo (b) kcat for Rubisco in wild-type (▪) and anti-RbcS (□, ) F. bidentis. Note the different scales on the two panels. The in vitro data reflect the activity of the fully carbamylated enzyme; in vivo kcat is estimated as gross photosynthesis divided by the number of Rubisco catalytic sites. Each value represents the mean (± se) of four measurements.

The Rubisco control coefficient (C_ra) was determined from the relationship between gross photosynthesis and the concentration of Rubisco catalytic sites (Fig. 5a). At low-measurement temperatures, this relationship is linear, and a higher amount of Rubisco increased photosynthesis. At warmer temperatures, the relationship between photosynthesis and Rubisco content reached a plateau as other limitations became important. The control coefficient was inversely related to temperature, being about 0.68 at the thermal optimum and rising to 0.99 at the lowest measurement temperature; at 6°C, C_ra was statistically equivalent to one (Fig. 5b).

Figure 5.

a, The relationship between gross photosynthesis and Rubisco catalytic site concentration at three temperatures; b, Rubisco C_ra in F. bidentis as a function of leaf temperature. The error bars in b were determined from the se of the regression of photosynthesis on catalytic sites.

DISCUSSION

C₄ plants are largely excluded from cool climates, probably because of poor photosynthetic performance at low temperatures relative to C₃ species (Osmond et al., 1982). This poor performance may reflect an inherent biochemical limitation, and different steps of C₄ photosynthesis have been suggested to be the rate-limiting factor at low temperatures. Using F. bidentis modified with antisense constructs to reduce Rubisco content, we determined the pattern of control exerted by Rubisco on C₄ photosynthesis over the 6°C to 40°C temperature range. Although control is shared near the thermal optimum, Rubisco becomes a principle control over C₄ photosynthesis at low temperatures. Several lines of evidence support this assertion. The in vitro activity of Rubisco matches the in vivo rate of gross photosynthesis at low temperatures, which indicates that the enzyme is the rate-limiting step. The apparent leakage of CO₂ from the bundle sheath increases at low temperature in the wild type to an extent similar to the increase accompanying the reduction of Rubisco by antisense. Finally, the Rubisco C_ra increases to near unity at low temperatures, which indicates almost complete control of C₄ photosynthesis by Rubisco.

Rubisco Kinetics in Vitro and in Vivo

Between 6°C and 15°C, the kcat of Rubisco in vitro and the rate of gross photosynthesis in vivo are equivalent in wild-type F. bidentis. This indicates a strong Rubisco limitation of C₄ photosynthesis at low temperatures. Reducing Rubisco content extends its control of C₄ photosynthesis to higher temperatures, as shown by the wider thermal range across which the in vivo and the in vitro kcat values are equivalent in the anti-RbcS lines. The in vivo kcat was less than the in vitro kcat of Rubisco above 15°C in wild-type plants, and above 25°C in the anti-RbcS lines. A similar finding has been previously reported in the C₄ grass B. gracilis (Pittermann and Sage, 2000). In B. gracilis populations from high elevation, the maximal in vitro Rubisco activity and net photosynthesis were equivalent below 22°C, whereas in plants from lower elevations, they were equivalent below 17°C. The high-elevation B. gracilis plants had 13% less Rubisco than low-elevation plants, which increased the temperature at which Rubisco activity became non-limiting.

At temperatures where Rubisco exerts high control, the activation energy (Ea) of gross assimilation (A*) should reflect the Ea of the enzyme. In F. bidentis, the situation is complicated by the increase in the Ea of Rubisco observed below 15°C. The Ea of A* between 5°C and 30°C was between the values for Rubisco above and below the break in the thermal response of the enzyme, as would be expected if the thermal response of the enzyme controls the thermal response of CO₂ assimilation. The Ea of Rubisco from F. bidentis determined between 18°C to 42°C is similar to the 50 to 60 kJ mol^–1 reported for a range of C₄ species (Sage, 2002). A break in the thermal response of Rubisco has been previously noted in rice (Oryza sativa; Sage, 2002), and in both C₃ and C₄ species of Atriplex (Björkman and Pearcy, 1971). In wild-type F. bidentis, the Ea of PEPCase in vitro above 18°C is equivalent to that of photosynthesis. However, during the temperature response measurements, photosynthesis was operating on the plateau of the A/Ci curve (data not shown), above the low-CO₂ region where PEPCase is postulated to control C₄ photosynthesis (von Caemmerer and Furbank, 1999).

Leakage of CO₂ from the Bundle Sheath

Reducing Rubisco capacity by antisense results in greater CO₂ leakage (ϕ) than in wild-type F. bidentis across the 6°C to 40°C range, as indicated by the increase in Φ_PSII/Φ_CO2*. In C₄ plants, the degree of overcycling, and hence CO₂ ϕ, should increase if the ratio of the mesophyll to bundle sheath reaction rates increase (Henderson et al., 1992). As indicated by the similarity of the PEPCase activities in the three genotypes, the mesophyll reactions in the anti-RbcS leaves proceed at rates similar to those of the wild type. In this case, the CO₂ concentration in the bundle sheath will increase, and ϕ will increase. Consistent with this, anti-RbcS F. bidentis shows increased carbon isotope discrimination at 25°C, relative to the wild type (von Caemmerer et al., 1997); in C₄ plants, increased leakage of CO₂ from the bundle sheath increases the discrimination against ¹³C (Farquhar, 1983). It is thus evident that the antisense plants are unable to reduce the rates of the mesophyll reactions to compensate for the reduced flow of carbon through Rubisco. Other explanations for the increase in Φ_PSII/Φ_CO2* at low temperatures cannot be completely excluded, but appear unlikely. Increases in the strength of alternative electron sinks, such as the Mehler reaction, should lead to increased Φ_PSII/Φ_CO2* (Krall and Edwards, 1991). However, the direct reduction of O₂ likely accounts for less than 10% of total electron flux in C₃ species, and there is no evidence of substantial rates in C₄ plants (Badger et al., 2000). A mesophyll limitation of C₄ photosynthesis, such as by PPDK, would reduce bundle sheath CO₂ levels and hence reduce both ϕ and Φ_PSII/Φ_CO2*, at least until photorespiration begins to increase. Thus, although alternative explanations cannot be completely excluded, they are not consistent with the results presented here and in previous studies.

The quantum requirement of CO₂ assimilation is constant above 15°C in a range of C₄ dicots and monocots, indicating that the stoichiometry of the C₄ and C₃ cycles is unaltered at intermediate and warm temperatures (Oberhuber and Edwards, 1993). Similar results are reported here; Φ_PSII/Φ_CO2* is relatively constant in wild-type F. bidentis at temperatures where the control of photosynthesis is shared between Rubisco and other processes. In wild-type plants, the values of Φ_PSII/Φ_CO2* at warm temperatures (>25°C) shown here are similar to those reported by Siebke et al. (1997). As Rubisco becomes the principle control over the rate of photosynthesis at cooler temperatures (<20°C), bundle sheath leakiness increases, indicating overcycling of the CO₂-concentrating reactions. In plants with reduced Rubisco, ϕ begins to increase at warmer temperatures than in wild-type leaves, as would be predicted from a higher Rubisco control of photosynthesis.

The Control of C₄ Photosynthesis

The C_ra defines the control exerted by Rubisco over the flow of carbon through the entire C₄ photosynthetic pathway (Stitt, 1995). The high C_ra below 20°C determined here shows that Rubisco dominates control at suboptimal temperatures, whereas the intermediate value at the thermal optimum shows that the control of photosynthetic carbon flux is shared between Rubisco and other processes. If several enzymes are colimiting, then reducing any one of them by antisense should result in a control coefficient close to one for that enzyme. This cannot be excluded here, although the agreement between the in vitro and in vivo kcat values of Rubisco would support the strong control exerted by this enzyme at lower temperatures. Theoretical models of C₄ photosynthesis at the leaf level indicate conditions under which other biochemical processes may dominate the control of A, such as at low light or low CO₂ (von Caemmerer and Furbank, 1999). At low PPFD, Rubisco control is not indicated, because the net CO₂ assimilation rate of anti-RbcS F. bidentis is the same as that of a null-transformant control (Furbank et al., 1996; Siebke et al., 1997). Using Amaranthus edulis mutants with reduced PEPCase, Dever et al. (1997) have shown that PEPCase has a high C_ra for C₄ photosynthesis at low CO₂, but this control is reduced when CO₂ is saturating (Dever et al., 1997). For PPDK and NADP malate dehydrogenase, the control coefficients are 0.3 and 0, respectively, at 25°C (Furbank et al., 1997). Model predictions indicate that at or above the thermal optimum in air, C₄ photosynthesis is limited by the regeneration of PEP or ribulose 1,5-bisphosphate (von Caemmerer and Furbank, 1999), but our results indicate that Rubisco continues to exert significant control at or above the thermal optimum in F. bidentis.

Consequences for C₄ Photosynthesis in Cool Climates

A limitation by Rubisco capacity on C₄ photosynthesis at low temperature represents a mechanism to explain the relative rarity of the C₄ syndrome in cool climate habitats. The carboxylation efficiency of Rubisco improves at low temperatures, because both the relative availability of CO₂ versus O₂ and the specificity of Rubisco for CO₂ increase as temperature declines (Badger and Collatz, 1977; Berry and Raison, 1981). This improved efficiency is more advantageous to C₃ species, because they typically have three to four times as much Rubisco as C₄ plants (Ku et al., 1979; Long, 1999). Furthermore, the high partial pressure of CO₂ in the bundle sheath ensures that little oxygenation occurs in C₄ plants, so the improved efficiency of Rubisco at low temperatures will not directly affect C₄ photosynthesis.

The minimal amount of Rubisco theoretically required for C₄ plants to match C₃ photosynthetic rates increases at lower temperatures (Long, 1999). Consistent with this, Rubisco accounts for only 4% to 8% of soluble leaf protein in the summer-active Death Valley native Tidestromia sp. but about 20% in cool-coastal Atriplex sp. (Osmond et al., 1982). Conversely, high-elevation ecotypes of B. gracilis contain less Rubisco than plants from low-elevation populations (Pittermann and Sage, 2000). Whereas C₃ species frequently acclimate to low temperatures by increasing the amount of Rubisco (Treharne and Eagles, 1970; Holaday et al., 1992; Hurry et al., 1995), in C₄ plants, the responses are more variable. In A. lentiformis, the Rubisco activity of plants grown at 23°C/18°C (day/night) is 60% greater than that of plants grown at 43°C/30°C (Pearcy, 1977), whereas Rubisco activity is insensitive to growth temperatures between 19°C and 31°C in maize (Zea mays; Ward, 1987) and between 14°C and 26°C in Muhlenbergia glomerata, a C₄ grass native to boreal Canada (Kubien, 2003).

The Rubisco content of C₄ species may be limited by the compartmentalization of the enzyme, because it is restricted to a reduced fraction of the leaf volume relative to C₃ species (Dengler and Nelson, 1999). Increasing the content of Rubisco would likely require changes in the proportion and arrangement of mesophyll and bundle sheath tissues within the leaves of C₄ plants or in the positioning of chloroplasts within the bundle sheath tissue. As a response to low temperatures this seems unlikely, because the spatial arrangement of these compartments influences the intercellular communication required for the efficient operation of the CO₂-concentrating mechanism (Dengler and Nelson, 1999). In addition, the Rubisco to chlorophyll ratio is the same in C₃ and C₄ species if the amount of the enzyme in the C₄ species is expressed on the basis of chlorophyll extracted from isolated bundle sheath cells (Ku et al., 1979; O. Ghannoum, personal communication). If this ratio is fixed, then C₄ species could not match the Rubisco content of C₃ plants, and the potential to increase the enzyme may be limited. Even if the bundle sheath cells could accommodate additional Rubisco, increasing the amount of the enzyme would have a negative effect on the nitrogen economy of C₄ species, thus mitigating one of the ecological advantages maintained over C₃ vegetation (Long, 1999; Pittermann and Sage, 2000). If this constraint exists, a potential solution would be to increase the kcat of the enzyme. Both kcat and K_m for CO₂ are higher in Rubisco from C₄ species than from C₃ plants (Seemann et al., 1984; Sage and Seemann, 1993; Sage, 2002). This increased turnover has not enabled C₄ species to become common in cool climates.

In summary, we propose that C₄ plants cannot contain sufficient Rubisco to match the photosynthetic rates of ecologically similar C₃ species at low temperatures. A reduction in the amount of Rubisco by C₄ species is possible because of the high CO₂ concentration in the bundle sheath and is one of the fundamental advantages C₄ plants have over their C₃ competitors (Osmond et al., 1982). High control of C₄ photosynthesis by Rubisco is a disadvantage at low temperatures and may be an inherent feature of the C₄ pathway that precludes such species from becoming common in cool climates.

MATERIALS AND METHODS

Plant Growth

Wild-type and anti-RbcS transgenic Flaveria bidentis were germinated in sand in a naturally lit greenhouse. The transgenic plants were T₂ progeny of the 141-1 (one insert) and 136-13 (four inserts) primary transformants (Furbank et al., 1996). Four (wild type) or 7 (transgenic) weeks after germination, the seedlings were transplanted to 12-L pots containing 69% (v/v) Promix (Plant Products, Brampton, Canada), 17% (v/v) sand, and 17% (v/v) plant-compost. Plants were subsequently moved to a controlled environment chamber (GC-20, Enconair, Winnipeg, Canada) and maintained under a 16-h photoperiod with a maximal PPFD of 750 μmol m^–2 s^–1. The day/night temperature and relative humidity were 28°/20°C and 50%/75%, respectively. Plants were watered daily and fertilized weekly with 0.5× Hoagland solution supplemented with 3 mm NH₄NO₃.

Gas-Exchange Measurements

The photosynthetic responses to temperature and CO₂ were measured with an open type leaf gas-exchange system using an infrared gas analyzer (Li-6262, Li-Cor, Lincoln, NE) to detect both CO₂ and water vapor. In this system, mass flow controllers (model 840, Sierra Instruments, Monterey, CA) were used to supply N₂, O₂, and CO₂ at the desired levels. All temperature and CO₂ responses were measured at 200 ± 5 mbar O₂. The air stream was humidified by passing the mixture through a water-filled flask which was set to a specific temperature in a water bath. For measurements at lower temperatures, the flask was placed on ice and filled with either water or a 70% (w/v) Suc solution. After humidification, CO₂ was injected, and the flow of air was measured by a mass-flow transducer (831, Edwards, Wilmington, MA) before being passed through the temperature-controlled leaf cuvette and an infrared gas analyzer. Leaf temperature was measured by placing three fine wire (36-gauge) thermocouples in contact with the abaxial surface of the leaf. Illumination was provided by a cool-light source (KL-2500, Schott, Mainz, Germany). All gas exchange measurements were made on the youngest fully expanded leaf and were calculated according to von Caemmerer and Farquhar (1981).

Photosynthetic temperature responses were measured either at a constant PPFD of 1,500 μmol m^–2 s^–1 or at a temperature-dependent PPFD that was sufficient to saturate photosynthesis. These points were determined by evaluating the photosynthetic responses to light at 12°C, 22°C, and 32°C, using a portable photosynthesis system (Li-6400, Li-Cor). The y intercept of light response curve was taken as an estimate of Rd at each temperature. This approach was used to mitigate the potential for photoinhibition, particularly at the lower temperatures. Light intensity in the cuvette was measured using a photodiode (G1738, Hamamatsu, Bridgewater, NJ) calibrated against a quantum sensor (Li-190s, Li-Cor). The temperature responses were measured at an ambient CO₂ of 370 ± 2 μbar. The leaf to air vapor pressure deficit was maintained at 12 ± 2 mbar at temperatures greater than 10°C; at cooler temperatures, vapor pressure deficit was reduced. All temperature response measurements were initiated at 30°C; leaf temperature was subsequently increased in 5°C intervals to 40°C and then decreased to the lower temperatures. At each temperature, the leaf was allowed to equilibrate for a minimum of 15 min before measurement. After the last measurement was completed, the leaf was warmed to about 15°C, and two leaf discs (1.55 cm² each) were rapidly removed and frozen in liquid N₂. Leaf samples were stored at –80°C until enzymes were assayed.

Chlorophyll a Fluorescence Measurements

Chlorophyll a fluorescence was determined simultaneously with gas exchange during the temperature response measurements of the wild-type and 141-1 lines. We used a PAM-101 (Walz, Effeltrich, Germany) equipped with an emitter-detector unit (ED-101BL, Walz) that provides excitation light at 470 nm and detection in the 660- to 710-nm waveband. This enabled us to isolate the fluorescence signal originating from PSII (Pfündel, 1998). Each leaf was allowed to dark-adapt at 30°C for 30 min before the ratio of variable to maximal fluorescence (F_v/F_m) was assessed. Reaction center closure was achieved by applying a 0.8-s pulse of saturating light (approximately 4,000 μmol m^–2 s^–1). Once a leaf had reached steady state at a given temperature, the quantum yield of PSII (Φ_PSII) was measured (Genty et al., 1989). Saturating pulses were applied at 90-s intervals; at each temperature, the average of three measurements was taken. There was no reduction in F_m′ with successive pulses. Thirty seconds after the last Φ_PSII estimate was obtained, F_o′ was assessed by rapidly darkening the leaf in the presence of far-red light. Leaf absorbance was determined from the chlorophyll concentration (Siebke et al., 1997). Fluorescence nomenclature and calculations follow van Kooten and Snel (1990).

Enzyme and Chlorophyll Assays

The in vitro activities of Rubisco and PEPCase were assayed from 0°C to 42°C using leaf discs harvested from the leaves used for gas exchange analysis. Leaf samples (3.1 cm²) were rapidly ground (<90 s) at 0°C using a ten-broek glass-in-glass homogenizer containing 7 mL of extraction buffer (100 mm HEPES, pH 7.6, 2 mm Na-EDTA, 5 mm MgCl₂, 5 mm dithiothreitol [DTT], 9 mg mL^–1 polyvinyl polypyrrolidone, 2 mg mL^–1 bovine serum albumin, 2 mg mL^–1 polyethylene glycol, 2.8% (v/v) Tween-80, 2 mm NaH₂PO₄, 11 mm amino-n-caproic acid, and 2.2 mm benzamide). Chlorophyll content was determined spectrophotometrically in N,N-dimethylformamide, using two aliquots of the crude extract (Porra et al., 1989).

Rubisco was quantified in aliquots of the crude extract, using a [¹⁴C]carboxy-arabinitol bisphosphate (CABP)-binding assay and assuming 6.5 binding sites per Rubisco (Butz and Sharkey, 1989). The CABP assay buffer consisted of 100 mm Bicine, 20 mm MgCl₂, and 10 mm NaHCO₃ at pH 8.2. The leaf extract (40 μL) was incubated in 40 μm ¹⁴CABP (specific activity, 27 Bq nmol^–1) at room temperature for 15 min, followed by a 2.5-h incubation at 37°C in the presence of rabbit anti-Rubisco serum. The Rubisco-¹⁴CABP complexes were filtered with 0.45-μm Supor filters (Gelman, Ann Arbor, MI) and thoroughly washed with a 10 mm sodium phosphate buffer (pH 7.6, containing 10 mm MgCl₂ and 150 mm NaCl). The radioactivity bound to the filters was measured by liquid scintillation spectroscopy.

A 3.33-mL aliquot of the crude leaf extract was added to 370 μL of a Rubisco activating solution (100 mm Bicine, pH 8.2) containing 280 mm MgCl₂ and 200 mm NaHCO₃, giving final concentrations of 28 mm MgCl₂ and 20 mm NaHCO₃, respectively (Sage and Seemann, 1993). This mixture was incubated at room temperature for 20 to 25 min to fully carbamylate Rubisco. The carbamylated extract was then kept on ice until being assayed. The remaining crude extract was kept on ice for subsequent determination of PEPCase activity.

Rubisco activity was assayed in a buffer containing 100 mm Bicine (pH 8.2), 1 mm Na-EDTA, 20 mm MgCl₂, 5 mm DTT, 1 unit mL^–1 ribulose-5-P kinase, 1.7 unit mL^–1 phospho-ribulo-isomerase, 2 mm ATP, 2 mm Rib-5-P, and 12 mm NaH¹⁴CO₃ (specific activity, 27 Bq nmol^–1, ICN Pharmaceuticals, Costa Mesa, CA; Pittermann and Sage, 2000). The assay buffer (400 μL) was allowed to equilibrate at a given temperature for 90 s, after which the assay commenced with the addition of 100 μL of the carbamylated extract. Assays ran for 30 to 60 s and were terminated by the addition of 500 μL of 2 n HCl. The radioactivity of acid-stable products was measured by liquid scintillation spectroscopy. Rubisco kcat (mol CO₂ fixed [mol active sites]^–1 s^–1) was determined from the in vitro activity and the concentration of catalytic sites.

PEPCase activity was assayed in a buffer containing 50 mm Bicine (pH 8.2), 1 mm Na-EDTA, 5 mm MgCl₂, 5 mm DTT, 4.6 mm PEP, and 4.6 mM G-6-P, 0.2 mm NADH, 0.3 unit mL^–1 MDH, and 5.4 mm NaH¹⁴CO₃ (Pittermann and Sage, 2000). Assays were initiated by the addition of 20 μL of the crude extract to 480 μL of the assay buffer, and terminated by the addition of 500 μL of 2 n HCl. Acid stable radioactivity was determined by liquid scintillation spectroscopy.

C_ra

A flux control coefficient (C_ra) was calculated to determine the extent to which Rubisco controls C₄ photosynthesis across the range of measurement temperatures (Kacser and Burns, 1973; Stitt, 1995). The coefficient was defined as:

(1)

where A* is the gross CO₂ assimilation rate (A + Rd, where Rd is respiration) and [E] is the concentration of Rubisco catalytic sites. To obtain the relationship between photosynthesis and Rubisco content, we regressed gross CO₂ assimilation against catalytic site concentration, using the data from the temperature response measurements. We made no underlying assumptions regarding the shape of this relationship and simply used the curve that gave the best fit at each temperature. The derivatives of each curve were then taken at each wild-type catalytic site concentration to determine δA*/δ[E] between 6°C and 40°C. A second-order polynomial was fit to the respiration rates, determined during the light response measurements, to provide an estimate of respiration at each temperature.