Influence of temperature on measurements of the CO2 compensation point: differences between the Laisk and O2-exchange methods.

The CO2 compensation point in the absence of day respiration (Γ*) is a key parameter for modelling leaf CO2 exchange. Γ* links the kinetics of ribulose-1,5-bisphosphate carboxylase-oxygenase (Rubisco) with the stoichiometry of CO2 released per Rubisco oxygenation from photorespiration (α), two essential components of biochemical models of photosynthesis. There are two main gas-exchange methods for measuring Γ*: (i) the Laisk method, which requires estimates of mesophyll conductance to CO2 (g m) and (ii) measurements of O2 isotope exchange, which assume constant values of α and a fixed stoichiometry between O2 uptake and Rubisco oxygenation. In this study, the temperature response of Γ* measured using the Laisk and O2-exchange methods was compared under ambient (25 °C) and elevated (35 °C) temperatures to determine whether both methods yielded similar results. Previously published temperature responses of Γ* estimated with the Laisk and O2-exchange methods in Nicotiana tabacum demonstrated that the Laisk-derived model of Γ* was more sensitive to temperature compared with the O2-exchange model. Measurements in Arabidopsis thaliana indicated that the Laisk and O2-exchange methods produced similar Γ* at 25 °C; however, Γ* values from O2 exchange were lower at 35 °C compared with the Laisk method. Compared with a photorespiratory mutant (pmdh1pmdh2hpr) with increased α, wild-type (WT) plants had lower Laisk values of Γ* at 25 °C but were not significantly different at 35 °C. These differences between Laisk and O2 exchange values of Γ* at 35 °C could be explained by temperature sensitivity of α in WT and/or errors in the assumptions of O2 exchange. The differences between Γ* measured using the Laisk and O2-exchange method with temperature demonstrate that assumptions used to measure Γ*, and possibly the species-specific validity of these assumptions, need to be considered when modelling the temperature response of photosynthesis.

Introduction

Models of photosynthesis are important tools for predicting the response of plants to climate change. The Farquhar, von Caemmerer, and Berry biochemical model of C3 photosynthesis was first parameterized to predict photosynthetic rates at 25 °C using the kinetic parameters of ribulose-1,5-bisphosphate carboxylase-oxygenase (Rubisco), the enzyme responsible for initiating carbon fixation (Farquhar et al., 1980). This model has proven to be robust in predicting the effects of CO2 availability on photosynthesis at 25 ºC but has been expanded to account for the temperature response of Rubisco kinetics with mixed success (Bernacchi et al., 2001, 2002, 2003; Sage et al., 2008). For example, the temperature response of the initial slope of the photosynthetic CO2 response (A-C i) curve in Picea mariana is less than that predicted by the Nicotiana tabacum Rubisco kinetic’s temperature model (Sage et al., 2008; Way and Sage, 2008). The authors attributed this deviation to a greater deactivation of Rubisco with temperature in P. mariana compared with N. tabacum or to differences in the temperature response of Rubisco kinetics between these species.

For accurate modelling, it is important to have correct Rubisco kinetics and assumptions concerning the major fluxes of CO2 and O2 during photosynthesis. The biochemical model of photosynthesis predicts net leaf CO2 exchange from the balance of carbon gain through Rubisco carboxylation with carbon loss through day respiration (R d) and photorespiration. Photorespiration releases CO2 at a given stoichiometry of CO2 per oxygenation (α), which is assumed to remain constant at 0.5 based on current understanding of photorespiratory biochemistry (Reumann and Weber, 2006). In C3 plants, photorespiration releases carbon at approximately 25% the rate of gross CO2 fixation, reducing the quantum efficiency of photosynthesis (von Caemmerer and Farquhar, 1981; Sharkey, 1988). Therefore, the CO2 compensation point in the absence of day respiration (Γ*), which quantifies photorespiratory loss of CO2 and the kinetic properties of Rubisco, is an essential term in models of photosynthesis (see Equations 2 and 4 below).

Γ* can be measured either biochemically through in vitro assays or in vivo using gas-exchange methods. Generally, in vivo Γ* is measured with the so-called ‘Laisk method’ as the intersection of A-C i curves measured at multiple subsaturating light intensities (Γ* L) (Laisk, 1977). The original method described by Laisk did not take into account mesophyll conductance of CO2 (g m) (Equation 3) to adjust the intercellular CO2 partial pressure (C i) to the CO2 partial pressure at the site of Rubisco (C c); however, several recent publications have reviewed the importance of including g m in estimates of Γ* L and gas-exchange generally (Warren, 2008; Furbank et al., 2009). Alternatively, mass spectrometer measurements of leaf O2 isotope exchange can also be used as an in vivo estimate of Γ* (Γ* O). This method does not require estimates of g m but does require assumptions related to leaf O2 exchange and α (see Equations 4, 7, and 8) (Ruuska et al., 2000; Bernacchi et al., 2002). O2 exchange is typically measured by placing a leaf disk in a sealed cuvette in an 18O2 atmosphere attached to a mass spectrometer via a membrane inlet (Canvin et al., 1980; Beckmann et al., 2009). The exchange of O2 in and out of the leaf is measured by following the uptake of 18O2 and evolution of the natural abundance of 16O2 from water splitting during photosynthesis (see Equations 7 and 8). The Γ* O calculations assume that α is constant at 0.5, O2 consumption from day respiration is the same as in the dark, and rates of photoreduction of O2 to water (the Mehler reaction) are negligible (Canvin et al., 1980; Badger, 1985). These assumptions appeared valid at 25 ºC when compared with independent measurements of gas exchange and Rubisco kinetics (Ruuska et al., 2000), but their accuracy as temperature increases has not been widely characterized (Badger et al., 2000).

Unfortunately, measurements of α are inherently difficult because they require determining the rate of CO2 release from photorespiration and the rate of Rubisco oxygenation (v o) while Rubisco carboxylation (v c) and CO2 release from R d continue in the light. However, at 25 °C the post-illumination burst (PIB) and 12CO2 release following a saturating 13CO2 injection both scale with photorespiratory CO2 release, providing an estimate of the CO2 component of α (Doehlert et al., 1979; Delfine et al., 1999; Loreto et al., 2001; Cousins et al., 2008, 2011). Additionally, v o can be estimated using isotopic exchange of 18O2 and 16O2, but these measurements are subject to the assumptions of O2 exchange outlined previously and discussed in the theory section below (Canvin et al., 1980; Badger, 1985; Cousins et al., 2008, 2011). Recently, measurements of 12CO2 release and 18O2 and 16O2 exchange indicated an increase in α in Arabidopsis thaliana lacking both isoforms of peroxisomal malate dehydrogenase (pmdh1pmdh2) and peroxisomal hydroxypyruvate reductase (hpr) (Cousins et al., 2008, 2011).

Despite the importance of Γ* to gas-exchange models and the value of understanding the temperature response of photosynthesis, to our knowledge there are no published comparisons of Γ* L and Γ* O at ambient and elevated temperatures. Such a comparison would help determine whether the two methods give consistent results and identify which assumptions may need re-evaluating at elevated temperatures. Therefore, this study examined the temperature response of Γ* L and Γ* O measured in N. tabacum (Bernacchi et al., 2001, 2002). Additionally, the temperature and O2 response of Γ* L and Γ* O were measured in A. thaliana wild-type (WT) and pmdh1pmdh2hpr plants. These data were used to explore the potential physiological explanations for differences between the two measurements of Γ*, including increases in α and changes in O2 exchange with temperature.

Theory

The rate of net assimilation of CO2 (A) can be modelled by subtracting CO2 released by photorespiration and mitochondrial respiration from Rubisco carboxylation rates:

where R d is the rate of day respiration (Farquhar et al., 1980). Additionally, the Farquhar, von Caemmerer, and Berry biochemical model describes Rubisco-limited photosynthesis as:

where V cmax, K c, and K o represent the maximum rate of v and Michalis–Menten constants for reactions with CO2 and O2, respectively (von Caemmerer, 2000). C c can be calculated from intercellular CO2 partial pressure (C i) using g m according to:

Γ*, the CO2 compensation point in the absence of day respiration is described by the Rubisco specificity for CO2 over O2 (S c/o), partial pressure of O2 (O) and α as:

Changes in Γ* affect estimates of net assimilation and, as indicated in Equation 4, are directly proportional to O and α. The CO2 compensation point in the presence of R d (Γ) is expressed as:

and is measured as the CO2 partial pressure where A is zero.

Photosynthesis at higher CO2 partial pressures is not Rubisco limited (Equation 2) but is usually limited by the ability of the Calvin—Benson cycle to regenerate intermediates for carbon fixation due to insufficient production of NADPH. Under these conditions, photosynthesis is dependent on the maximum rate of electron transport (J max) and energy demand of photosynthesis and photorespiration according to:

The Laisk method of measuring Γ* (Γ* L) is limited by the ability to measure g m accurately to convert measured values of C i to C c (Equation 3), whereas Γ* measured on the mass spectrometer (Γ* O) relies on estimates of v o and v c and assumes α=0.5 (Equation 4) (Ruuska et al., 2000; Bernacchi et al., 2002). In this method, v o is determined assuming that rates of O2 uptake in the dark are equal to uptake by mitochondrial respiration in the light. Additionally, the total O2 uptake by Rubisco is determined assuming 1 mole of O2 is consumed from the atmosphere during oxidation of glycolate in the peroxisome for every two oxygenations of Rubisco (Badger, 1985). Therefore, the rate of v o is:

Assuming that all electrons from water splitting reduce NADPH for v c and v o (Badger, 1985; Ruuska et al., 2000), v c can be determined by:

Subsequently, Γ* O can then be calculated from v c and v o (Equation 4). Both the Laisk and the O2-exchange method rely on measurements of the net exchange of CO2 and O2, respectively, to determine Γ* assuming that CO2 and O2 are exchanged primarily through reactions of photosynthesis, photorespiration, and R d. However, there are several other carboxylases and decarboxylations within plant cells, including phosphoenolpyruvate carboxylase and carbamoyl phosphate synthetase, that could mask the true Γ* with unaccounted fluxes (Raven and Farquhar, 1990). Whilst these additional fluxes are important physiologically, their rates are typically a tenth to a one-thousandth the rate of CO2 flux through Rubisco and have a negligible impact on calculations of Γ*

Materials and methods

Growth conditions

WT A. thaliana Columbia accession and mutant pmdh1pmdh2hpr (Pracharoenwattana et al., 2007) were grown in a climate-controlled cabinet (Econair Ecological Chambers, Winnipeg, Canada) under a photosynthetic flux density of 300 μmol m–2 s–1 and 2000 μbar CO2 to minimize the phenotype of pmdh1pmdh2hpr1 (Pracharoenwattana et al., 2007). Day/night cycles were 11/13h and 23/18 °C. Seeds were cold stratified for 3 d and germinated on sterile agar plates supplemented with MS medium (Plant Media, Dublin, OH, USA) and 1% sucrose. Following cold stratification, plates were placed in the growth chamber for 1 week and the seedlings were then transferred to soil for an additional 3 weeks and fertilized weekly with Peters 20-20-20 (J.R. Peters, Allentown, PA, USA). The youngest fully expanded leaves of 31–40-d-old plants were used for gas-exchange measurements.

Laisk CO2 compensation points

The Laisk method (Laisk, 1977) was used to measure the apparent compensation point (C *) in WT and pmdh1pmdh2hpr plants under different O at 25 and 35 °C. Different O2 partial pressures (92, 184, and 368 mbar O2) were generated using O2 and N2 mixed with calibrated mass flow controllers (model GFC17; Aalborg, Orangeburg, NY, USA). A-C i curves were measured on a leaf fully enclosed in a 2cm2 measuring head (6400–40 Leaf Chamber Fluorometer; Li-Cor Biosciences, Lincoln, NE, USA) at subsaturating light intensities using a Li-Cor 6400 XT (Li-Cor Biosciences) and the x and y coordinates of these points were used to determine R d (y coordinate) and C * (x coordinate). CO2 diffusion through the gasket was corrected according to the manufacturer’s instructions (Li-cor 6400XT manual version 6). The CO2 compensation point in the absence of day respiration (Γ*) was subsequently calculated from C * by accounting for mesophyll conductance (g m) and R d according to Γ*=C *+R d/g m, with g m equal to 0.2 and 0.35mol CO2 m–2 bar–1 at 25 and 35 °C, respectively. The value of 0.2mol CO2 m–2 bar–1 at 25 °C was the average of several A. thaliana ecotypes measured under various conditions (Tazoe et al., 2011) and this value becomes 0.35mol CO2 m–2 bar–1 at 35 °C according to the temperature-response model of (Bernacchi et al., 2002).

Mass spectrometric measurements

Rates of v c and v o were determined from measurements of 18O2 consumption and 16O2 evolution according to Equations 7 and 8. 18O2 consumption in the light and dark and 16O2 evolution in the light was measured as described previously (Canvin et al., 1980; Ruuska et al., 2000; Cousins et al., 2008). Briefly, a leaf disc was placed in a temperature-controlled closed cuvette system connected to a mass spectrometer (Delta V; Thermo Scientific) via a temperature-controlled membrane inlet. The cuvette was flushed with N2 gas, injected with 18O2 gas to obtain a given O2 partial pressure, and sealed. 18O2 consumption and 16O2 evolution rates were monitored during dark and light periods. The PIB was determined from the maximum transient rate of CO2 release in the dark following a 10min period of illumination. The 12CO2 release was determined in the light from the maximum rate of 12CO2 released following a saturating injection of 13CO2 (Delfine et al., 1999; Loreto et al., 2001; Cousins et al., 2008, 2011). 13CO2 was generated from acid-released 13CO2 from 98% 13CO2/sodium hydrogen carbonate (Sigma Aldrich, St Louis MO, USA). Γ* was calculated according to Equation 4.

Parameterization and temperature-response modelling

A-C c curves were measured using a Li-Cor 6400 assuming a g m of 0.2 and 0.35mol CO2 m–2 bar–1 at 25 and 35 °C, respectively (Bernacchi et al., 2002; Tazoe et al., 2011). Measurements were made under saturating illumination (photosynthetic flux density of 1200 μmol m–2 s–1) and vapour pressure deficits below 15 mbar at 25 °C and 25 mbar at 35 °C. The A-C c curves were fitted to the leaf photosynthesis model (Farquhar et al., 1980) using in vivo N. tabacum Rubisco constants at 25 and 35 °C to determine the effects of changes in Γ* on V cmax and J max (Bernacchi et al., 2001, 2002). The Γ* temperature and O2 response was modelled by first changing the scaling constant of Bernacchi et al. (2001, 2002) so that the function gave values of Γ* that matched WT A. thaliana at 25 °C. The heat of activation determined previously in N. tabacum Bernacchi et al. (2001, 2002) was used to model values at 35 °C.

Statistics

A four-way analysis of variance (ANOVA; Table 1) was used to determine the influence of genotype, temperature, measurement method, and O2 levels on Γ* using Statistix 9 (Analytical Software, Tallahassee, FL, USA). A three-way ANOVA (Table 2) was used to determine the influence of genotype, temperature, and O2 levels on measurements of v o, 12CO2 release per v o and PIB per v o using Statistix 9. A two-way ANOVA was used to determine the significance in measured and modelled V cmax and J max values (Table 3) using R (R Foundation for Statistical Computing, Vienna Austria, http://www.R-project.org). Significance was assumed to be P <0.05.

Table 1.

Effects of using the Laisk or O2-exchange methods to estimate Γ* on modelling CO2 assimilation curves under elevated temperature. Maximum rate of Rubisco carboxylation (V cmax) and electron transport (J max) at 25 and 35 °C calculated using standard biochemical models of leaf photosynthesis (von Caemmerer, 2000) with Γ*L or Γ*O from Bernacchi et al. (2001) or Bernacchi et al. (2002). The modelled V cmax and J max at 35 °C were scaled from 25 °C measurements using the temperature-response functions of Bernacchi et al. (2001, 2002). Results are shown as means ±SE of five leaves from separate plants. Statistical analysis was conducted using a one-way ANOVA; different superscript letters indicate significant differences between α assumptions and temperatures at P <0.05.

Temperature	Assumed Γ*	Vcmax (μmol m–2 s–1)		J max (μmol m–2 s–1)
		Measured	Modelled	Measured	Modelled
25°C	Γ*L	41.3±1.5a	–	87.6±3.1a	–
	Γ*O	38.1±0.5a	–	84.6±2.7a	–
35°C	Γ*L	85.9±5.1c	96.5±3.4c	107.9±3.1b	163.2±6.2c
	Γ*O	70.8±2.1b	89.0±1.2c	97.0±3.0ab	148.9±4.8c

Table 2.

Results of ANOVA comparing the Laisk and O2-exchange methods of measuring Γ* at various oxygen partial pressures and temperatures in WT and pmdh1pmdh2hpr A. thaliana. Method refers to the measurement technique of Γ* with measurements of O2 exchange on the mass spectrometer indicated as Mass to avoid confusion with the O effect and the mutant pmdh1pmdh2hpr referred to as 3X for convenience. O is indicated as the partial pressure in mbar (92O2, 184O2, and 368O2) and temperature is in °C (25T, 35T). Asterisks indicate a significant interaction according to ANOVA (P <0.05) and different superscript letters denote significant differences according to a Tukey’s post-hoc test (P <0.05). Results are shown as the means ±SE of three to six leaves from separate plants.

Parameter	Factor	F ndf, ddf	Interactions
Γ*	Genotype	72.21,87*
	O2	397.32, 87*
	Temp	84.71, 87*
	Method	0.61, 87
	Genotype, O2	6.12, 87*
	Temp, Method	4.51, 87*
	Genotype, Method	4.91, 87*
	Genotype, O2, Temp	1.65, 87
	Genotype, Temp, Method	15.41, 87*	3X35TLaiskab, 3X35TMassa, 3X25TLaiskbc, 3X25TMassbc, WT35TLaiskab, WT35TMasscd, WT25TLaiskde, WT25TMasse
	O2, Temp, Method	3.84, 87*	368O235TLaiska, 368O235TMassa, 368O225TLaiskb, 368O225TMassb, 184O235TLaiskb, 184O235TLaiskbc, 184O225TLaiskcd, 184O225TLaiskde, 92O235TLaiske, 92O235TLaiskf, 92O235TLaiskf, 92O235TLaiskf
	Genotype, O2, Temp, Method	2.04, 87	–

Table 3.

Results of two-way ANOVA on mass spectrometric measures of CO2 release during photorespiration in WT and pmdh1pmdh2hpr A. thaliana. ANOVA analysis between rates of Rubisco oxygenation (v o), PIB:v o, and 12CO2 release following a saturating injection of 13CO2 (12CO2):v o as measured on leaf punches with a membrane inlet mass spectrometer. The mutant pmdh1pmdh2hpr is referred to as 3X for convenience, O is indicated by the partial pressure in mbar (92O2, 184O2, and 368O2) and temperature is in °C (25T, 35T). Asterisks indicate a significant interaction according to ANOVA (P <0.05) and different superscript letters denote significant differences according to a Tukey’s post-hoc test (P <0.05). Results are shown as the means ±SE of three to six leaves from separate plants.

Parameter	Factor	F ndf, ddf	Interactions
v o	Genotype	37.71,72*
	Temp	22.41,72*
	O2	148.12,72*
	Genotype, O2	14.92,72*	WT92O2 c, WT184O2 b, WT368O2 a, 3X92O2 c, 3X184O2 b, 3X368O2 b
	Genotype, Temp	37.71,72*	WT25T a, WT35T a, 3X25T b, 3X35T a
	Temp, O2	4.51,72*	25T92O2 d, 25T184O2 c, 25T368O2 b, 35T92O2 d, 35T184O2 b, 35T368O2 a
	Genotype, O2, Temp	2.42,72	–
PIB/v o	Genotype	54.31,72*
	Temp	2.21,72
	O2	7.82,72*
	Genotype, O2	5.92,72*	WT92O2 a, WT184O2 b, WT368O2 b, 3X92O2 a, 3X184O2 a, 3X368O2 a
	Genotype, Temp	21.81,72*	WT25T c, WT35T b, 3X25T a, 3X35T ab
	Temp, O2	17.21,72*	25T92O2 b, 25T184O2 b, 25 T368O2 ab, 35T92O2 a, 35T184O2 b, 35T368O2 b
	Genotype, O2, Temp	2.62,72	–
12CO2/v o	Genotype	4.81,72*
	Temp	0.11,72
	O2	8.92,72*
	Genotype, O2	2.42,72
	Genotype, Temp	15.92,72*	WT25T b, WT35T ab, 3X25T a, 3X35T b
	Temp, O2	0.32,72	–
	Genotype, O2, Temp	2.42,72	–

Results

Comparison of Γ*L and Γ*O from Bernacchi et al. (2001, 2002)

The response of Γ* L to temperature was modelled as described by Bernacchi et al. (2001) using the Laisk method to estimate the temperature response assuming an infinite mesophyll conductance (g m) (Fig. 1, dotted line). This response was then corrected for g m using the values reported by Bernacchi et al. (2002), the temperature response of R d from Bernacchi et al. (2001), and Equation 3 (Fig. 1, solid line). The dashed line in Fig. 1 represents the temperature response of Γ* O according to Bernacchi et al. (2002), which was determined from O2 exchange using a membrane inlet mass spectrometer assuming a constant stoichiometric release of CO2 per oxygenation of 0.5. The 25 ºC Γ* L values were greater than Γ* O regardless of g m and were more responsive to temperature.

Fig. 1.

Modelled response of the CO2 compensation point in the absence of day respiration (Γ*) to temperature and O2. The graph shows the temperature response of Γ* from Bernacchi et al. (2001) with a correction for mesophyll conductance (g m) (solid line) and with an infinite g m (dotted line). The dashed line represents the response according to Bernacchi et al. (2002), which was determined assuming a stiochiometric release of CO2 per oxygenation of 0.5.

Effects of Γ*L and Γ*O on modelling of photosynthetic CO2-response curves under elevated temperatures

In WT A. thaliana, the photosynthetic parameters V cmax and J max were calculated (Farquhar et al., 1980; von Caemmerer, 2000) from the net CO2 assimilation rates as a function of CO2 partial pressures at 25 and 35 °C assuming Γ* L, Γ* O, and Rubisco kinetics from Bernacchi et al. (2001, 2002, 2003) (Fig. 2 and Table 1). V cmax was not significantly different when Γ* L or Γ* O was used at 25 °C (41.3±1.5 μmol m–2 s–1 for Γ* L and 38.1±0.5 μmol m–2 s–1 for Γ* O). However, at 35 °C, V cmax was significantly different depending on whether Γ* L (85.9±5.1 μmol m–2 s–1) or Γ* O (70.8±2.1μmol m–2 s–1) was used for the calculation. Additionally, at 35 °C, the modelled temperature response of V cmax was significantly different from the measured values using Γ* O (Bernacchi et al., 2002) but not Γ* L (Bernacchi et al., 2001). However, the calculated values of J max were not significantly different when using either Γ* O or Γ* L. Additional key gas-exchange parameters, including net CO2 assimilation, intercellular CO2 concentration, stomatal conductance to H2O, and H2O transpiration rates, are presented in Supplementary Table S1 (at JXB online).

Fig. 2.

The CO2 response of photosynthesis at 25 °C (closed circles) and 35 °C (open circles) as measured by a Li-Cor 6400. Chloroplastic CO2 partial pressure (C c) was determined from previously published mesophyll conductance values (with a mesophyll conductance (g m) of 0.2 and 0.35 mmol m–2 s–1 bar–1 at 25 °C (Tazoe et al., 2011) and 35 °C, respectively. Results are shown as means ±standard error (SE) of five leaves from separate plants.

CO2 compensation point in the absence of day respiration under elevated temperature

Γ* L and Γ* O were measured and modelled in A. thaliana WT and pmdh1pmdh2hpr plants at 25 and 35 °C in response to various O (Fig. 3). WT values of Γ* L and Γ* O increased linearly with O at both 25 °C (r 2=1.0000 for Γ* L and r 2=0.9997 for Γ* O) and 35 °C (r 2=0.9982 for Γ* L and r 2=0.9854 for Γ* O) with a significantly higher Γ* L compared with Γ* O at 35 °C regardless of O (Fig. 3A, C and Table 2). There was also a linear response of Γ* L and Γ* O in the pmdh1pmdh2hpr plants to O at 25 °C (r 2=0.9999 for Γ* L and r 2=0.9423 for Γ* O) and 35 °C (r 2=0.9979 for Γ* L and r 2=0.9773 for Γ* O) (Fig. 3B, D); however, there was no significant difference between Γ* L and Γ* O at either temperature (Table 2). The WT Γ* L and Γ* O was lower than pmdh1pmdh2hpr at 25 °C regardless of O (Table 2). However, at 35 °C, there was no difference in Γ* L between the two genotypes, but Γ* O was significantly higher in pmdh1pmdh2hpr compared with WT at all O (Table 2).

Fig. 3.

The CO2 compensation point in the absence of day respiration (Γ*) in WT (A, C) and pmdh1pmdh2hpr (B, D) A. thaliana plants under various O2 partial pressures at 25 °C (closed symbols; A, B) and 35 °C (open symbols; C, D) measured using the Laisk method for WT (circles) and pmdh1pmdh2hpr (upward triangles) plants. Measurements of Γ* using O2 exchange are also shown for WT (squares) and pmdh1pmdh2hpr (downward triangles) plants. Solid lines represent predicted Γ* values from Bernacchi et al. (2001) with a mesophyll conductance (g m) of 0.2 and 0.35 mmol m–2 s–1 bar–1 at 25 and 35 °C, respectively. Dotted lines show the results from Bernacchi et al. (2002) with α=0.8 for the pmdh1pmdh2hpr plants. Results are shown as means ±SE of three to six leaves from separate plants. Laisk data (25 °C) from WT and pmdh1pmdh2hpr plants were also presented in Cousins et al. (2011).

As the model was fitted to WT values of Γ* L and Γ* O at 25 °C for Bernacchi et al. (2001, 2002), there was good agreement between the measured and modelled values at each O for the WT. However, at 35 °C, Γ* L was slightly underestimated and Γ* O was slightly overestimated by Bernacchi et al. (2001). The modelled Γ* for the pmdh1pmdh2hpr plants was adjusted to a higher α of 0.8 (Equation 4). At 25 °C in the pmdh1pmdh2hpr plants, the modelled values fitted Γ* L at all O and Γ* O at 92 and 184 mbar O (Fig. 3B). However, at 35 °C, Γ* L was underestimated by the model of Bernacchi et al. (2001) and Γ* O was overestimated by the model of Bernacchi et al. (2002) (Fig. 3D).

Rates of Rubisco oxygenation and carboxylation from measurements of O0 and of CO2 isotope exchange

A membrane inlet mass spectrometer was used to measure rates of CO2 and O2 isotope exchange in WT and pmdh1-pmdh2hpr plants in response to temperature and O. From these measurements, the PIB, the release of 12CO2 in a saturating 13CO2 background, and the rate of Rubisco oxygenation (v o) were determined. At 25 and 35 °C, there was a significant response of v o to O for both genotypes (Fig. 4 and Table 3). However, v o did not respond to temperature in WT plants but was significantly different between 25 and 35°C in the pmdh1pmdh2hpr plants. In WT plants, the PIB:v o ratio was significantly lower at 25 °C compared with that at 35 °C, regardless of O2 level, and was significantly lower in WT compared with pmdh1pmdh2hpr plants at 25 but not at 35°C across all O2 levels (Fig. 5 and Table 3). In both genotypes, there was no significant response of PIB:v o to O2 at 25 °C, but at 35 °C, the PIB:v o ratio was higher at 92 mbar compared with at 184 and 368 mbar O2. The 12CO2:v o ratio responded significantly to O2, regardless of temperature and genotype, but decreased with temperature in the pmdh1pmdh2hpr plants but not in the WT plants. At 25 °C, the 12CO2:v o ratio was greater in the pmdh1pmdh2hpr plants compared with the WT plants, but there was no difference between genotypes at 35 °C. In summary, at 25 °C the PIB:v o and 12CO2:v o ratios were higher in the pmdh1pmdh2hpr plants compared with the WT plants, but at 35°C, they were not significantly different between genotypes, regardless of O. Additionally, PIB:v o was significantly higher in the WT plants at 35 °C compared with 25 °C but did not significantly respond to temperature in the pmdh1pmdh2hpr plants. In the WT plants, PIB:v o was different at 92 mbar compared with at 184 and 368 mbar O2 but not in the pmdh1pmdh2hpr plants.

Fig. 4.

Rates of Rubisco oxygenation (v ) at the CO2 compensation point estimated from measurements of 18O2 and 16O2 exchange as described in Materials and methods. Measurements were made at 92, 184, and 368 mbar O2 at 25 and 35°C for both WT (A) and pmdh1pmdh2hpr (B) A. thaliana plants. Results are shown as means ±SE of three to six leaves from separate plants.

Fig. 5.

PIB per v (A, B) and 12CO2 release per v (C, D) at various O2 partial pressures in WT (A, C) and pmdh1pmdh2hpr (B, D) A. thaliana at 25 and 35 °C measured from CO2 and isotopic O2 with the membrane inlet mass spectrometer. Results are shown as means ±SE of three to six leaves from separate plants.

Discussion

Effects of Rd and gm on measurements of Γ* using the Laisk and O2-exchange methods

Bernacchi et al. (2001) measured the temperature response of Γ* in N. tabacum using the Laisk method (Γ* L) (Laisk, 1977) to develop a temperature response model of Γ*. Measurements of Γ* L require no assumptions about the photorespiratory stoichiometry of CO2 released per oxygenation (α) or leaf O2 exchange but must be corrected for the difference between C i and C c with values of mesophyll conductance to CO2 (g m) (Equation 3). Additionally, the temperature response of Γ* was measured using O2 exchange at Γ (Γ* O) (Bernacchi et al., 2002), which does not require values of g m (see below), but assumes that: (i) α is equal to 0.5 (Equation 4), (ii) O2 is consumed only by photorespiration and Rubisco oxygenation, (iii) rates of O2 consumption by R d are the same as respiration in the dark, and (iv) all electrons passed to NADPH drive either photosynthesis or photorespiration (Equations 4, 7, and 8) (Badger, 1985). Given that all these assumptions are correct, then Γ* L should equal Γ* O. However, a direct comparison of these two methods of estimating Γ* has not been conducted, particularly in response to temperature.

As noted in several publications, it is important to account for g m to measure Γ* L accurately (von Caemmerer et al., 1994; von Caemmerer, 2000; Ethier and Livingston, 2004; Furbank et al., 2009). This is because Γ* L is determined from the intercept of several A-C i curves measured under subsaturating light conditions. The x value of this intercept represents C i at Γ*, and the y value of A is negative and represents rates of R d (Laisk, 1977). To estimate Γ* L, the values of C i must be corrected for g m to obtain an accurate C c (Equation 3) (Ethier and Livingston, 2004; Furbank et al., 2009). As Γ* L is determined when A is negative, Γ* L before accounting for g m is lower than after correcting for g m. The Γ* L values reported by Bernacchi et al. (2001) were uncorrected for g m, meaning that they are lower than the g m-corrected value would be. Therefore, to accurately describe Γ* the model of Bernacchi et al. (2001) must be corrected for the temperature response of g m. Alternatively, at Γ, there is no net photosynthesis and the ratio of A:g m approaches zero regardless of g m value (Equation 3). Therefore, because A is zero at Γ, measurements of C i are equal to C c. Under these conditions, stomatal conductance is similarly negligible and C c can be determined from measured CO2 partial pressure inside the sealed cuvette without correcting for g m. Consequently, measurements of Γ* O are not sensitive to errors in g m (Equation 4). This latter approach was used by Bernacchi et al. (2002) to measure Γ* independently of assumptions of g m. To compare these two models of Γ* at 25 °C, the values of Bernacchi et al. (2001) for Γ* L were corrected using g m according to Bernacchi et al. (2002). At 25 °C, the g m-corrected modelled values of Γ* L were higher than the modelled Γ* O (Fig. 1) and differences between Γ* L and Γ* O increased with temperature, highlighting the greater temperature sensitivity of Γ* L compared with Γ* O. The difference between Γ* L and Γ* O is not explained by potential errors in assumptions of g m because the difference is significant when g m is assumed to be infinite (no restriction to CO2 diffusion and C i=C c) and the discrepancy increases as g m decreases (Fig. 1). Similarly, any assumed value of R d (from zero to infinity) also increases the discrepancy between the values of Γ* L from Bernacchi et al. (2001) and the values of Γ* O from Bernacchi et al. (2002) (Equation 3). In summary, the temperature response and absolute values of Γ* L are higher than Γ* O even when corrected for g m and regardless of R d. Both of these estimates of Γ* are used to determine the maximum rate of Rubisco carboxylation (V cmax) and the maximum rate of electron transport (J max) from gas-exchange measurements of A-C i curves (von Caemmerer, 2000). Additionally, the temperature response of Γ* is essential for modelling the response of these parameters and photosynthesis to changes in leaf temperatures. Therefore, it is important to determine how the difference in temperature response of Γ* between Bernacchi et al. (2001) and Bernacchi et al. (2002) influences estimates of V cmax and J max derived from A-C c measurements. To test this, V cmax and J max were determined from A-C c curves measured at 25 and 35 °C in A. thaliana with Γ* from the two Bernacchi et al. (2001, 2002) temperature-response models.

Sensitivity of Vcmax and Jmax to Γ*

At temperatures above 25 °C, previous publications have attributed lower V cmax estimated from leaf gas-exchange measurements compared with modelled values as changes in the Rubisco activation state (Sage et al., 2008). However, some of this difference could also be explained by errors in Γ* and its modelled temperature response. For example, using the g m-corrected Γ* L from Bernacchi et al, (2001) to compare measured and modelled V cmax values from A. thaliana A-C c curves, there was no significant difference at 35 °C (Table 1). However, if Γ* O from Bernacchi et al. (2002) was used to calculate V cmax, then the measured values were significantly lower than the modelled V cmax (Table I). This could be interpreted as deactivation of Rubisco using Γ* O but not with Γ* L. This difference between V cmax calculated using Γ* L versus Γ* O highlights the importance of determining which method is most appropriate for modelling photosynthesis at different temperatures, as well as understanding which assumptions within the two models are valid in response to changing temperatures.

It is possible that the differences in Γ* L (Bernacchi et al., 2001) versus Γ* O (Bernacchi et al., 2002) are dependent on the differences in Rubisco content between genotypes used for each study. For example, Rubisco antisense plants were used by Bernacchi et al. (2001) to measure the temperature response of Γ* L. These plants have lowered photosynthetic rates compared with WT plants, which may have introduced errors into measuring the intercept of A-C i curves at low CO2 partial pressures (Hudson et al., 1992). Indeed, the 25 °C value of Γ* L of 41.9 μbar CO2 from antisense plants measured by Bernacchi et al. (2001) is higher than other reports of Γ* L measured in both WT N. tabacum and other C3 plants. For example, Γ* L values at 25 °C typically range between 36.7 and 40.8 μbar CO2 (Brooks and Farquhar, 1985; von Caemmerer et al., 1994; Laisk and Loreto, 1996).

Additionally, differences between Γ* L and Γ* O could be driven by errors in the assumptions used to parameterize each method. As previously discussed, Γ* L must be corrected for g m; however, including corrections for g m increases the difference between Γ* L and Γ* O. Alternatively, there are several assumptions used in determining Γ* O with unknown temperature responses. For example, measurements of Γ* O assume that α is constant at 0.5 (Equation 4). Additionally, Γ* O relies on measurements of v and v , which require assumptions relating O2 exchange to Rubisco reactions (discussed below). Therefore, to test these assumptions at 25 and 35°C, measurements of Γ* L and Γ* O were made in A. thaliana WT and the photorespiratory mutant (pmdh1pmdh2hpr), previously characterized as having an increased α, to determine which parameters contribute to the discrepancies between Γ* L and Γ* O.

Differences in Γ*L and Γ*O in WT A. thaliana

Measurements of Γ* L and Γ* O in A. thaliana were used to confirm and characterize the differences between the two methods of measuring Γ* presented by Bernacchi et al. (2001, 2002) (Fig. 3). In A. thaliana, there was no significant difference between Γ* L and Γ* O in WT plants at 25 °C (Table 2). This is different from what was observed previously in N. tabacum where Γ* L determined by Bernacchi et al. (2001) was higher than Γ* O from Bernacchi et al. (2002) at 25 °C (Fig. 1). The different response of Γ* L and Γ* O at 25 °C between A. thaliana and N. tabacum could be the result of the different genotypes used in each study. As mentioned before, Bernacchi et al. (2001) used Rubisco small-subunit antisense plants, whilst Bernacchi et al. (2002) measured WT N. tabacum; however, in the current study, WT A. thaliana plants were used for both estimates of Γ*.

The close agreement of Γ* L and Γ* O in WT A. thaliana at 25 °C at a variety of O values provides strong support that the independent assumptions of both methods are valid at 25 °C in this species. However, at 35 °C, Γ* L and Γ* O in A. thaliana were significantly different across all O (Table 2 and Fig. 3), confirming a similar increased temperature response of Γ* L over Γ* O as suggested by the comparison of data from Bernacchi et al. (2001, 2002). As discussed previously, larger values of g m increase Γ* L estimated from measured C i values (Equation 3); however, regardless of the g m values used (0.10 to infinity), Γ* L was always larger than Γ* O (data not shown). Therefore, errors in g m do not explain the differences between Γ* L and Γ* O at 35 °C. These findings in A. thaliana confirm differences in Γ* L and Γ* O above 25 °C, although the difference is less than reported in N. tabacum (Bernacchi et al., 2001, 2002). To determine whether the different temperature responses of Γ* L and Γ* O could be explain by changes in α the photorespiratory mutant pmdh1pmdh2hpr was compared with WT at both 25 and 35°C.

Response of α to temperature

Biochemical models of photosynthesis and measurements of Γ* O (Equation 4) typically assume α=0.5 under all conditions. However, there are several recent publications demonstrating changes in α when the traditional photorespiratory pathway is disrupted through genetic manipulation. For example, the photorespiratory mutants pmdh1pmdh2, hpr, and pmdh1- pmdh2hpr had lower net photosynthetic rates under photorespiratory conditions, higher Γ and higher Γ* L than WT plants (Cousins et al., 2008, 2011). Additionally, measurements of CO2 and O2 isotope gas exchange in the pmdh1pmdh2 and hpr plants at 25 °C confirmed that Γ and Γ* L were higher due to an increase in α (Cousins et al., 2008, 2011). Similar to previously published work on hpr and pmdh1pmdh2 plants, the photorespiratory mutant pmdh1pmdh2hpr in this study had higher Γ* L and CO2 release per v o compared with WT plants, indicating an increase in α in the pmdh1pmdh2hpr plants at 25 °C (Fig. 5 and Table 3, discussed below). The Γ* O value in the pmdh1pmdh2hpr plants was modelled with α=0.8 instead of α=0.5, a stoichiometry that also modelled Γ and Γ* L in this and previous studies with photorespiratory mutants with increased α (Equation 4 and Fig. 3) (Cousins et al., 2008, 2011). This suggests that misestimates of α can lead to inaccurate calculations of Γ* O.

Measurements of Γ* L, which do not require assumptions of α, in WT plants were significantly lower than in pmdh1-pmdh2hpr plants at 25 °C; however, at 35 °C, the values were not significantly different between genotypes across all O (Table 2). It is expected that the S c/o of Rubisco is conserved between WT and pmdh1pmdh2hpr plants at a given temperature (Jordan and Ogren, 1984); therefore, differences in Γ* L at 25 °C could be attributed to α (Equation 4). However, at 35 °C, the values of Γ* L were the same between WT and pmdh1pmdh2hpr plants, suggesting that α may increase with temperature in WT A. thaliana to a stoichiometry similar to that in pmdh1pmdh2hpr plants. In WT plants, an increase in α could also explain why Γ* L and Γ* O were the same at 25 °C but Γ* O was lower than Γ* L at 35 °C when assuming α=0.5. The linear response of Γ* L to O2 at 35 °C indicated that the increase in α would be constant at a given temperature, regardless of O (Equation 4 and Fig. 3).

Two putative reactions of photorespiratory intermediates within the peroxisome could explain increases in α. Specifically, excess glyoxylate and hydroxypyruvate could react with H2O2 releasing CO2, formate, and glycolate with or without an enzyme catalyst (Elstner and Heupel, 1973; Halliwell, 1974). Indeed, this reaction is hypothesized to be a major source of formate in leaves (Igamberdiev et al., 1999). Formate can be further decarboxylated in the peroxisome (Halliwell and Butt, 1974) or oxidized to CO2 by formate dehydrogenase in the mitochondria (Hourton-Cabassa et al., 1998), whilst glycolate could re-enter the photorespiratory pathway. These reactions would result in additional CO2 release per Rubisco oxygenation and divert carbon from the Calvin–Benson cycle and the regeneration of ribulose-1,5-bisphosphate.

It has also been hypothesized that, in WT plants, similar increases in α occur under elevated temperatures due to an increase in glycolate oxidase activity relative to catalase within the peroxisomes (Grodzinski and Butt, 1977). Additionally, in isolated peroxisomes and mitochondria, an increase in H2O2 can react with glyoxylate and hydroxypyruvate leading to an increase release of CO2 (Grodzinski and Butt, 1977; Grodzinski, 1978; Hanson and Peterson, 1985). Furthermore, overexpression of catalase in N. tabacum reduced the levels of H2O2 and lowered Γ as temperature increased compared with WT plants (Brisson et al., 1998). These data suggest that α could increase from non-catalysed decarboxylation reactions with H2O2, decreasing the efficiency of phosphoglycolate recycling but not completely disrupting the photorespiratory pathway. Therefore, measurements of labelled CO2 and O2 isotope exchange as described by Cousins et al. (2008, 2011) were used to determine the influence of temperature on α and to probe some of the assumptions of O2 exchange used to measure Γ O at 35 °C.

Rates of Rubisco oxygenation in A. thaliana WT and pmdh1pmdh2hpr plants

Rates of v and v are determined by the relative availability of O2 and CO2, Rubisco kinetics, and the activation state of Rubisco (Salvucci and Crafts-Brandner, 2004; von Caemmerer et al., 2004; Sage et al., 2008). It has been shown that Rubisco deactivates under high temperatures, decreasing both v c and v (Kobza and Edwards, 1987; Feller et al., 1998). Deactivation of Rubisco at 35 °C could explain the insensitivity of v to temperature across all O2 treatments in WT A. thaliana. However, in contrast to the WT, v increased with temperature in pmdh1pmdh2hpr plants but was constant at 184 and 368 mbar O2 at 35 °C. This is paradoxical given the apparent decrease in 12CO2 release per v o in pmdh1pmdh2hpr plants under higher Rubisco oxygenation conditions when perturbations to photorespiration would be more severe (Table 3 and Fig. 5). However, this could be explained by errors in measuring v o and/or photorespiratory CO2 release (discussed below).

Alternatively, the increase in v with temperature in pmdh1pmdh2hpr but not in WT plants could be attributed to changes in O2 exchange by alternative oxidations of photorespiratory intermediates traditionally not described as part of the photorespiratory pathway. For example, measurements of v would decrease if the 2:3 ratio of v to net O2 uptake used in Equation 7 decreased due to additional oxygenation of photorespiratory intermediates. This would subsequently increase the ratios of PIB and 12CO2 release per v o. These reactions could also explain the apparent discrepancies seen in v o and CO2 release per v o at 35 °C in the pmdh1pmdh2hpr plants (Fig. 5). If similar increases in α occur in WT plants under elevated temperature due to non-enzymatic or enzymatic reactions, measurements of v o and CO2 release per v o would also be affected.

CO2 release per Rubisco oxygenation reaction

To measure α accurately, both the flux of CO2 from photorespiration and the corresponding rates of v o must be determined. The CO2 released from photorespiration cannot be measured directly; however, the combined flux from photorespiration and R d can be estimated from the PIB and by the rate of 12CO2 evolution following a saturating injection of 13CO2 on an illuminated leaf (Cousins et al., 2008, 2011). As discussed previously, the photorespiratory mutant pmdh1pmdh2hpr had a higher PIB and 12CO2 release per v o at 25 °C compared with WT plants across O, suggesting an increased α in the pmdh1pmdh2hpr plants (Fig. 5 and Table 3). However, at 35°C, the PIB per v o was not significantly different between WT and pmdh1pmdh2hpr plants (Fig. 5 and Table 3). Curiously, PIB per v o was significantly higher at the lowest O compared with the other O levels in both WT and pmdh1pmdh2hpr plants. This increase in CO2 release per v o at the lowest O was not expected based on the linear relationship between Γ* L and O (Fig. 3).

The discrepancy between Γ* L and Γ* O and the downward trend in PIB and 12CO2 release per v o in response to O seen in the WT plants might also be explained by errors in two major assumptions of O2 exchange: (i) the O2 uptake from day respiration is equal to rates of dark respiration and (ii) the rates of Mehler reaction are negligible. It is generally accepted that rates of respiration in the light (R d) are less than rates of respiration in the dark (Villar et al., 1994; Lambers and Ribas-Carbo, 2005) and that R d may respond to changes in rates of photorespiration (Tcherkez et al., 2008). The Laisk measurements of Γ* L can estimate the CO2 release from R d, which could be used in place of uptake of 18O2 in the dark in estimates of v assuming a stoichiometry of CO2 evolution to O2 uptake during respiration (Equation 7). However, when R d was used to calculate v o instead of the measured dark rates of O2 consumption, there was no change in the trends of PIB and 12CO2 release per v o, as presented in Fig. 5, when the stoichiometry of CO2 evolution to O2 uptake was held constant regardless of the value (comparison not shown). Therefore, there would have to be changes in the stoichiometry of CO2 evolution to O2 uptake to explain the changes in Fig. 5.

In addition to differences in dark versus light respiration, higher rates of the Mehler reaction at 35 °C could introduce errors in the calculated rates of v o due to the consumption of O2 independent of photosynthesis and photorespiration (Ort and Baker, 2002). This would lead to overestimations of v o (Equation 7) and underestimations of PIB and 12CO2 release per v o. The rates of Mehler would have to range from 10% of O2 evolution at 92 mbar to 60% at 368 mbar to maintain a constant PIB and 12CO2 release per v o with O (calculations not shown). However, at 25 °C, the rates of Mehler in C3 plants are reported to range from 0 to 30% of photosynthetic electron transport at 25 °C (Asada, 1999; Badger et al., 2000; Ruuska et al., 2000; Driever and Baker, 2011), but the temperature dependence and O2 response of these reactions are not well known for A. thaliana. Measurements of O2 exchange under various conditions in N. tabacum found that v explained O2 consumption under low and elevated temperatures, suggesting that the Mehler rate does not increase with temperature (Badger et al., 2000). Therefore, the temperature response of the Mehler reactions in A. thaliana would have to be significantly different compared with N. tabacum to explain the downward trend in Fig. 5.

Finally, the downward trend in PIB and 12CO2 release per v o in response to O seen in the WT plants at 35 °C could be explained if the CO2 released from photorespiration does not scale with PIB and 12CO2 release at elevated temperatures across O. In this situation, PIB and 12CO2 release would no longer be proportional to the CO2 released from photorespiration at 35 °C in response to O. This would lead to a decrease in the ratio PIB and 12CO2 release per v o as O increases that does not correspond to changes in α. The observation that PIB saturates with increasing O at 25 °C and with temperature supports this suggestion (Doehlert et al., 1979). Therefore, at 35 °C, the discrepancy between a constant α described by the linear response of Γ* L and the decreasing trend in PIB and 12CO2 release per v o as O increases could be explained by a saturating response of PIB and 12CO2 release to photorespiratory rates. Unfortunately, there is insufficient evidence to determine whether the downward trend in PIB and 12CO2 release per v o with O is the result of unaccounted rates of the Mehler reaction, changes in differences between dark and light respiration rates, or saturation of CO2 released from photorespiration as measured by PIB and 12CO2 release. Each of these could individually or collectively affect estimates of PIB and 12CO2 release per v o in response to O.

Conclusion

The data presented here demonstrate differences in temperature-response models of Γ* from N. tabacum between the Laisk and O2-exchange methods. These differences were large enough to impact both measured and modelled values of V cmax and J max. Differences in Γ* determined from the Laisk and O2-exchange method were also seen in A. thaliana at 35 °C. The difference in estimates of Γ* were probably due to errors in assumptions used in O2-exchange calculations at elevated temperature. The extent of these errors and the species-specific differences in these assumptions should be considered when modelling the temperature response of photosynthesis with Γ* values derived from O2 exchange.

Supplementary data

Supplementary data are available at JXB online.

Supplementary Table S1. Gas-exchange parameters from CO2-response curves measured at 25 and 35 °C.