C3 plants enhance rates of photosynthesis by reassimilating photorespired and respired CO2


F. A. Busch. E-mail: florian.busch@anu.edu.au


Photosynthetic carbon gain in plants using the C3 photosynthetic pathway is substantially inhibited by photorespiration in warm environments, particularly in atmospheres with low CO2 concentrations. Unlike C4 plants, C3 plants are thought to lack any mechanism to compensate for the loss of photosynthetic productivity caused by photorespiration. Here, for the first time, we demonstrate that the C3 plants rice and wheat employ a specific mechanism to trap and reassimilate photorespired CO2. A continuous layer of chloroplasts covering the portion of the mesophyll cell periphery that is exposed to the intercellular air space creates a diffusion barrier for CO2 exiting the cell. This facilitates the capture and reassimilation of photorespired CO2 in the chloroplast stroma. In both species, 24–38% of photorespired and respired CO2 were reassimilated within the cell, thereby boosting photosynthesis by 8–11% at ambient atmospheric CO2 concentration and 17–33% at a CO2 concentration of 200 µmol mol−1. Widespread use of this mechanism in tropical and subtropical C3 plants could explain why the diversity of the world's C3 flora, and dominance of terrestrial net primary productivity, was maintained during the Pleistocene, when atmospheric CO2 concentrations fell below 200 µmol mol−1.


The vast majority of Earth's plants utilize the C3 photosynthetic pathway, in which the enzyme ribulose 1,5-bisphosphate carboxylase/oxygenase (Rubisco) catalyzes the first step in CO2 fixation and production of carbohydrates (Sage, Sage & Kocacinar 2012). Rubisco is a dual functional enzyme that catalyzes the carboxylation of its substrate ribulose 1,5-bisphosphate (RuBP), but also oxygenates RuBP in photorespiration, an inhibitory process where O2 is assimilated and CO2 is released by the plant (Bauwe, Hagemann & Fernie 2010). Rubisco oxygenase activity and photorespiration become substantial above 25 °C and atmospheric CO2 concentrations below the current value of 390 µmol mol−1. To compensate for high rates of photorespiration, certain plants have evolved mechanisms that suppress photorespiration by raising the CO2 concentration around Rubisco. Of these, the C4 photosynthetic pathway is the most productive, with twice the photosynthetic efficiency of C3 plants in hot environments (Zhu, Long & Ort 2008). C3 plants are not known to exhibit specific mechanisms to suppress photorespiration, and in hot, low latitude climates, they potentially face large yield reductions and severe competition from C4 plants (Gerhart & Ward 2010). One possibility of how C3 plants counteract these shortcomings is that they also express a mechanism to diminish the inhibitory effects of photorespiration. Here we examine the hypothesis that C3 plants, specifically rice (Oryza sativa) and wheat (Triticum aestivum), are able to offset high rates of photorespiration by positioning chloroplasts to block the efflux of photorespired CO2 from the cytosol to the intercellular air space (IAS), such that the CO2 must re-enter the chloroplast, thereby increasing photosynthetic carbon assimilation (Sage & Sage 2009).

The leaf anatomy of some C3 plants suggests that there may be high potential to trap photorespired CO2 before it escapes the cell. Photorespired CO2 is released in the mitochondria during the glycine decarboxylation step of photorespiratory metabolism (Bauwe et al. 2010). Once formed, it can exit the cell by either diffusing through the chloroplast into the IAS, or by directly leaving the cytosol, with the path of least resistance for CO2 efflux being gaps between the chloroplasts (Evans et al. 2009). In most C3 plants, chloroplasts are arrayed around the cell periphery; mitochondria, by contrast, are often located towards the interior of a mesophyll cell, juxtaposed to chloroplasts (Sage & Sage 2009). This organelle configuration is particularly pronounced in rice, where chloroplasts cover 95% of the periphery of the mesophyll cells (Sage & Sage 2009). To enhance the coverage of the cell periphery, chloroplasts in rice produce stromules, which are chloroplast extensions that contain Rubisco (Bourett, Czymmek & Howard 1999). This extensive sheath of chloroplasts and stromules is proposed to block the diffusion of photorespired CO2 out of the cell via the cytosol, thereby forcing it to exit the cell through chloroplasts. In essence, the sheath of chloroplasts and stromules may form an intracellular trap for photorespired CO2, which can then supplement the CO2 supply from the IAS. Under high rates of photorespiration, entrapped CO2 from mitochondria could significantly increase the CO2 concentration around Rubisco, thus enhancing its efficiency.

Reassimilation of photorespired and respired CO2 has been measured directly for only the CO2 fraction returning from the IAS (Haupt-Herting, Klug & Fock 2001; Haupt-Herting & Fock 2002). Intracellular reassimilation has been assumed to be negligible (Gerbaud & Andre 1987), or measured as part of intrafoliaceous (intracellular plus via the IAS) reassimilation (Bauwe et al. 1987; Pärnik & Keerberg 1995, 2007; Loreto, Delfine & Di Marco 1999; Pärnik, Ivanova & Keerberg 2007). Intrafoliaceous reassimilation estimates in C3 plants range from about 15% in rye and sunflower (Pärnik & Keerberg 2007) to close to 50% in wheat (Pärnik & Keerberg 2007) and Flaveria cronquistii (Bauwe et al. 1987); some estimates in C3 dicots are as high as 80% (Loreto et al. 1999). To evaluate whether chloroplast sheaths could trap photorespired CO2, it is necessary to measure refixation of CO2 directly within the cell. Here we present a novel method to estimate the rate of CO2 release by photorespiration and mitochondrial respiration in the light as well as the rates of Rubisco carboxylation (Vc) and oxygenation (Vo). Separate quantification of the CO2 fluxes into and out of the leaf using 13CO2 provides procedures to estimate the reassimilation of photorespired CO2 within the mesophyll cells of rice and wheat. Concurrently, we measure the relative chloroplast coverage of the mesophyll cell periphery exposed to the IAS, allowing us to evaluate whether high chloroplast coverage promotes trapping and reassimilation of photorespired CO2. We discuss our results and the influence on the performance of plants with different photosynthetic pathways in view of the low atmospheric CO2 levels of the recent geological past, when concentrations were well below current levels, averaging 270 µmol mol−1 during the past 10 000 years, and falling to 180 µmol mol−1 during the late-Pleistocene epoch (Lüthi et al. 2008).


Plant material and growth conditions

Plants of winter wheat (Triticum aestivum‘Daws’) and indica rice (Oryza sativa‘IR64’) were grown in 4 L pots in growth chambers for 5 weeks before measurements were taken. Wheat plants were grown under a 12 h photoperiod at 25 °C in the light and 18 °C in the dark at constant relative air humidity of 70%. The plants were fertilized once a week with a Scotts-Peters Professional 20-20-20 fertilizer (6.0 g L−1; The Scotts Co., Marysville, OH, USA). Rice plants were grown under a 12 h photoperiod at 28 °C in the light and 22 °C in the dark at constant relative air humidity of 70%. The plants were fertilized twice a week with a nutrient solution containing Sprint 330 iron chelate (1.3 g L−1), magnesium sulfate (0.6 g L−1), Scotts-Peters Professional 10–30-20 compound (2.8 g L−1) and Scott-Peters Soluble Trace Element Mix, 8.0 mg L−1 (The Scotts Co., OH).

The CO2 compensation point measurements on sunflower (Helianthus annuus‘Mammoth Russian’), rye (Secale cereale), tobacco (Nicotiana tabacum‘Petit Havana’) and F. cronquistii were performed on plants grown in the greenhouse. Greenhouse-grown unfertilized wheat plants were measured at the age of 10 weeks to obtain CO2 compensation points of aging leaves.

Electron microscopy

Leaves were taken out of the leaf chamber directly after the reassimilation or CO2 compensation point measurement and sampled for transmission electron microscopy, using fixation procedures described elsewhere (Sage & Williams 1995). The percentage of the cell periphery adjacent to the IAS that was covered by chloroplasts and stromules (Sc/Sm) was calculated as the length of cell periphery facing the IAS covered by chloroplasts (Sc) divided by the total length of cell periphery facing the IAS (Sm). Electron micrographs of a minimum of 10 cells from each of three plants per species were quantified using Image J (National Institutes of Health, Bethesda, MD, USA).

CO2 isotope fluxes

Net and gross fluxes of CO2 were measured using 12CO2 and 13CO2 in combination with membrane inlet mass spectrometry. A two-step process allowed us to determine gross rates of CO2 assimilation, photorespiration, mitochondrial respiration in the light and rates of reassimilation of (photo)respired CO2, as shown in Fig. 1. Firstly, the air surrounding the leaf was switched from ambient air, containing 12CO2, to air with the same concentration of 13CO2 (200 or 350 µmol mol−1, Fig. 1a). Plants were grown in ambient air containing approximately 99% 12CO2; therefore, we can assume, without introducing a large error, that the carbon in the leaf is fully labelled as 12C. Hence, the CO2 that is evolved from photorespiration and mitochondrial respiration at the time of the switch from 12CO2 to 13CO2 is released from the leaf as 12CO2 (R12C). The gross CO2 assimilation rate can then be calculated as the uptake of 13CO2 (A13C); however, some of the CO2 evolved is reassimilated from within the cell (AI) or from the IAS (AR). Therefore, in a second step, the air surrounding the leaf, containing ambient concentration of 12CO2, was rapidly switched to a gas stream containing 10 000 µmol mol−1 of 13CO2 (Fig. 1b). The high concentration of 13CO2 prevents the reassimilation of 12CO2 evolved from photorespiration (RPR) and mitochondrial respiration in the light (Rd); thus, the 12CO2 coming off the leaf represents the total rate of CO2 being evolved (Rtotal), equalling the sum of RPR and Rd. At 29‰ for C3 plants (McNevin et al. 2006), the discrimination of Rubisco against 13CO2 is negligible for the purpose of our measurements.

Figure 1.

Schematic representation of the CO2 fluxes measured to estimate reassimilation of (photo)respired CO2 within the cell and via the intercellular air space (IAS) during illumination. (a) Instantaneous replacement of the 12CO2 in the gas mix with the same concentration of 13CO2 (either 200 or 350 µmol mol−1). The experimental setup then allows for separate measurements of the exchange of 13CO2 and 12CO2 between the leaf and ambient air (A13C: gross CO2 uptake, R12C: release of 12CO2). The reassimilation from the IAS (AR) and the gross release of CO2 into the IAS (RC) can be calculated from the CO2 flux and stomatal conductance. (b) In order to measure the CO2 that is reassimilated within the cell (AI), the leaf is exposed to 10 000 µmol mol−1 of 13CO2, effectively preventing any reassimilation of evolved 12CO2. Then RI can be calculated as the difference between Rtotal and RC. As the high concentration of 13CO2 reduces the evolution of 12CO2 from photorespiration to zero after about 2 min, Rtotal can be separated into RPR (photorespiratory flux) and Rd (CO2 flux from mitochondrial respiration). For more details, see Materials and Methods.

Experimental setup

Expanding on the method described by Haupt-Herting et al. (2001), the fluxes depicted in Fig. 1 were analysed in an open gas exchange system (LI-6400XT; Li-Cor Environmental, Lincoln, NE, USA) coupled to an isotope ratio mass spectrometer (Delta V Plus, Thermo Scientific, West Palm Beach, FL, USA) through a gas permeable elastic polycarbonate-silicone membrane (MEM 213, General Electric, Waterford, NY, USA) followed by an ethanol/dry ice trap to freeze out any remaining humidity (Cousins, Badger & von Caemmerer 2006a). An air stream of 21% O2 in N2 was mixed from tank air by mass flow controllers (Aalborg, Orangeburg, NY, USA) and directed to the air inlet of a LI-6400XT gas exchange system (Fig. 2). After 12CO2 was added with the LI-6400XT internal mixer, the air was passed through a humidifying flask and a condenser set to 25 °C before entering the LI-6400XT leaf chamber. In separate tanks of 21% O2 in N2, mixtures of 350 (or 200) and 10 000 µmol mol−113CO2, respectively, were prepared and the gas of each tank was separately passed through a humidifying flask and a condenser set to 25 °C. A three-way valve was used to select the concentration of 13CO2. During the experiment, the air entering the chamber from the LI-6400XT console could be replaced within 30 s by one of the two concentrations of 13CO2 air by a four-way valve located before the LI-6400XT leaf chamber (Fig. 2). All lines were kept short to minimize the time to change between gas mixtures. The efflux of the leaf chamber was connected to the membrane inlet of the mass spectrometer (Cousins et al. 2006a; Cousins, Badger & von Caemmerer 2006b). The LI-6400XT was used to measure the water mole fraction in the air and to control the leaf temperature; the concentrations of 12CO2 and 13CO2 in the air were measured with the mass spectrometer. To account for drift in the signal, the mass spectrometer was calibrated twice daily with known concentrations of CO2.

Figure 2.

Schematic diagram of the gas exchange system used for the reassimilation measurements. FC, mass flow controller; MS, isotope ratio mass spectrometer.

Gas exchange measurements

The gas exchange system was used to determine the rates of intercellular and intracellular CO2 reassimilation in rice and wheat. For rice, the third leaf, and for wheat, the youngest fully expanded leaf was used for the measurements. The leaves were placed in the LI-6400XT chamber at a leaf temperature of 30 °C and a light intensity of 1500 µmol photons m−2 s−1 until photosynthetic steady state was reached, typically within 1 h. The LI-6400XT was used to control the flow rate (1 L min−1) and chamber conditions. After taking a measurement at steady state, a four-way valve was used to switch the air stream containing 12CO2 (350 or 200 µmol mol−1) to one containing the same concentration of 13CO2 for 100 s, while the concentrations of both 12CO2 and 13CO2 were continuously monitored with the mass spectrometer. After 15 min, the air stream was switched to 10 000 µmol mol−113CO2 and the concentration of 12CO2 was monitored over the following 150 s (Fig. 3).

Figure 3.

Time courses for 12CO2 and 13CO2 following a rapid switch from an air stream containing 350 µmol mol−112CO2 only, to one containing 13CO2. The traces show CO2 concentrations for a wheat measurement following the switch to 13CO2 air and the preceding and succeeding measurements of an empty chamber. Inserts show the difference between the measurement of the leaf and the average of the two empty chamber measurements, indicating the rate of CO2 uptake or release. The grey shaded areas denote the time it took to fully flush the chamber with the new gas mix. Traces were fitted with a first-order exponential decay function for the time after the chamber was fully flushed. The functions were extrapolated back to the time of the switch to estimate fluxes under steady-state conditions. (a) Uptake of 13CO2 (A13C) and (b) release of 12CO2 evolved from the mitochondria (R12C) during the short-term exposure to 350 µmol mol−113CO2. (c) The release of all 12CO2 evolved during inhibition of reassimilation as a response to the short-term exposure to 10 000 µmol mol−113CO2 (Rtotal), separated into CO2 evolved from photorespiration (RPR) and mitochondrial respiration in the light (Rd). Measurements were conducted at an irradiance of 1500 µmol photons m−2 s−1 and 30 °C. [Correction added on 27 July 2012, after first online publication: The y-axis label of panels (b) and (c) was amended from 13CO2 to 12CO2.]

Each leaf measurement was flanked by two empty chamber measurements, which were used as a reference to estimate the CO2 fluxes. Measuring both, the leaf sample and the reference, in the same chamber requires a high temporal stability of the gas supply. We confirmed that no drifts in the CO2 signal occurred during the measurements by checking that the two empty chamber measurements before and after the leaf measurement are tightly overlapping. This setup then ensures – in combination with a high flow rate and potential gasket leaks around the leaf veins sealed with vacuum grease – that diffusion of CO2 or H2O is not affecting the measurements.

The CO2 fluxes cannot be calculated from steady-state measurements directly, as the plants are exposed to changing CO2 concentrations during the switch to air containing 13CO2. Therefore, the time series of 12CO2 and 13CO2 concentrations measured during the exposure to air containing 13CO2 was used to obtain the differential between the absolute CO2 concentrations of the leaf measurement and the empty chamber reference. This was achieved by fitting the time course of the measured CO2 differential with a first-order exponential decay function between the time the chamber was fully flushed with air containing 13CO2 (approx. 30 s) and the end of the exposure to 13CO2 (100 or 150 s, inserts of Fig. 3). Because an increasing fraction of (photo)respired 12CO2 was replaced by 13CO2 over time following the exposure to 13CO2, the fitted curve was extrapolated back to the time of the switch to estimate CO2 exchange under steady-state conditions. The extrapolation is particularly important for the flux calculation during the exposure to 10 000 µmol mol−113CO2, as the high concentration of CO2 is also inhibiting photorespiration.

Calculation of CO2 fluxes

The CO2 fluxes in the leaf were calculated as described previously (von Caemmerer & Farquhar 1981), using the 13CO2 flux into the leaf and the 12CO2 flux out of the leaf as determined by the mass spectrometer. The rate of 13CO2 assimilation (µmol CO2 m−2 s−1) is given by:


where F is the gas flow rate (µmol s−1), 13CR and 13CS are the mole fractions of 13CO2 in the chamber without and with a leaf; WR and WS are the corresponding water mole fractions; and a is the illuminated leaf area in the chamber (m2). The rate of 12CO2 release R12C (µmol CO2 m−2 s−1) is calculated as:


The intercellular CO2 concentrations can be derived from the measured variables as:


using the total conductance for CO2 (gtc) and the transpiration rate (E), which are calculated as previously described (von Caemmerer & Farquhar 1981).

We can deduce the rate of intercellular CO2 reassimilation (AR) as


neglecting the small effect of Rubisco discrimination and assuming an equal probability of 12CO2 and 13CO2 being assimilated from the IAS (Haupt-Herting et al. 2001).

The rate of CO2 released from the mitochondria into the IAS (RC) from both photorespiration and respiration is the sum of intercellular 12CO2 reassimilation (AR) and the 12CO2 flux out of the leaf (R12C):


To quantify the rate of CO2 that is reassimilated within the cell, we measured the 12CO2 evolution during exposure of the leaf to 10 000 µmol mol−113CO2. The reassimilation of 12CO2 is then effectively inhibited by an over-abundance of 13CO2 in the IAS. The total amount of CO2 respired in the mitochondria (Rtotal) is released into the chamber air and can therefore be estimated using Eq 2 with the 12CO2 input values obtained from the switch to 10 000 µmol mol−113CO2. The rate of intracellular reassimilation (AI) is then calculated as:


The carboxylation rate of Rubisco is calculated as the sum of 13CO2 assimilation (A13C) and the 12CO2 that is reassimilated intracellularly (AI) and via the IAS (AR):


To estimate the contributions of photorespiration (RPR) and mitochondrial respiration in the light (Rd) to the total amount of CO2 released during exposure to 10 000 µmol mol−113CO2 (Rtotal), the exponential decay fit of the measurement curve was separated into its two parts (Fig. 3c). The offset of the fit represents Rd, which uses unlabelled photosynthates as the source of respired CO2 and is therefore constant over the first 2 min of exposure to 10 000 µmol mol−1 CO2. We assume that Rd is not affected by the high CO2 concentration during this short exposure, unlike the variable part that is therefore attributed to photorespiration (RPR). As RPR is the carbon flux associated only with photorespiration, it is equal to 0.5 Vo, half of the oxygenation rate of Rubisco (von Caemmerer 2000). Hence, our method is suitable to directly estimate both Vc and Voin vivo.

The stimulation of the assimilation rate due to intracellular reassimilation was estimated comparing Anet with a modelled value inline image that only takes into account reassimilation from the IAS. Assuming that the entire amount of (photo)respired CO2 diffuses into the IAS, the CO2 flux into the IAS increases by a factor of inline image. The rate of reassimilation from the IAS is therefore modelled as:


Then inline image equals:


The knowledge of rates of Rubisco oxygenation and carboxylation allows us to calculate the chloroplastic CO2 concentration (Cc). The ratio Vo/Vc only depends on Rubisco kinetic constants comprised in Γ* and Cc, and can therefore be written as (von Caemmerer 2000):


Solving for Cc results in an estimate of the chloroplastic CO2 concentration when intracellular reassimilation is included. Substituting this into the equation for the RuBP consumption limited carboxylation rate (von Caemmerer 2000)


yields an estimate for Vcmax, the maximal Rubisco carboxylation rate, for each individual plant. Hereby, O denotes the chloroplastic oxygen concentration; Kc and Ko are the Michaelis−Menten constants of Rubisco carboxylation and oxygenation, respectively, calculated for 30 °C using the temperature dependencies described by von Caemmerer (2000). The chloroplastic CO2 concentration assuming no intracellular reassimilation can then be calculated as:


derived from the equation for the RuBP consumption limited rate of CO2 assimilation (von Caemmerer 2000), using the modelled net CO2 assimilation rate inline image and Vcmax as determined by equation (11).

CO2 compensation point measurements

The apparent intercellular CO2 partial pressure at the CO2 compensation point in the absence of mitochondrial respiration (C*) was calculated as the intersection of the initial slopes of A/Ci curves measured at different light intensities, using an LI-6400XT following the method proposed by Laisk (1977) as described by Brooks & Farquhar (1985). The in vitro values for the CO2 compensation point in the absence of mitochondrial respiration (Γ*) at 30 °C were calculated from published Rubisco CO2/O2 specificity values for rice and wheat (Kane et al. 1994), F. cronquistii (Kubien et al. 2008), tobacco (Whitney et al. 1999) and sunflower (Sharwood et al. 2008) using the Arrhenius function described by von Caemmerer (2000). To convert the Rubisco specificity values from concentration to partial pressures, solubilities of 0.0334 mol L−1 bar−1 for CO2 and 0.00126 mol L−1 bar−1 for O2 were used.


We measured the mesophyll cell periphery facing the IAS that is covered by chloroplasts (Sc) and the total periphery of the mesophyll cell exposed to the IAS (Sm) to quantify chloroplast coverage. In wheat, most of the cell periphery adjacent to the IAS is covered by chloroplasts and stromules (Sc/Sm = 0.925 ± 0.013; Fig. 4a,b). The cell periphery not covered by chloroplasts occurs where the cell contacts neighbouring cells (Fig. 4a). In rice, chloroplasts cover almost the entire periphery of highly lobed cells and stromules extend along the cell periphery not covered by chloroplasts (Sc/Sm = 0.989 ± 0.003; Fig. 4c,d). In rice and wheat, mitochondria occur towards the inside of the mesophyll cells, in close association with chloroplasts and stromules (Fig. 4b,d).

Figure 4.

Transmission electron micrographs of transverse sections of wheat (Triticum aestivum, a–b) and rice (Oryza sativa, c–d). Stromules, indicated by arrows, were only found in wheat and rice – filling the gaps between chloroplast (C) to the intercellular air space (IAS). This cellular arrangement provides an additional resistance for CO2 evolved from the mitochondria diffusing into the IAS. Peroxisomes (p) and mitochondria (m) are located on the interior of the cell; (n) nucleus. Bars = 2 µm.

A rapid replacement of the ambient 12CO2 in the gas mix with the same concentration of 13CO2, the fluxes of 13CO2 into the leaf and 12CO2 out of the leaf can be separately determined with the mass spectrometer (Fig. 1a). The origin of the evolved 12CO2 is determined with a pulse of 10 000 µmol mol−113CO2, as demonstrated by a typical time course of a measurement on a wheat plant shown in Fig. 3. Photorespiration is suppressed by the high concentration of 13CO2, allowing Rtotal to be separated into Rd and RPR. After 2 min of exposure to 10 000 µmol mol−1 of 13CO2, the photorespiratory metabolite pools are exhausted and the 12CO2 detected originates only from Rd, with rice showing rates of 0.5–0.7 µmol m−2 s−1 and wheat 0.9–1.1 µmol m−2 s−1 (Table 1, Fig. 3c). RPR can then be calculated as the difference between Rtotal and Rd. Consequently, Vc and Vo can directly be quantified in vivo (Table 1). The CO2 concentrations at the site of Rubisco (Cc) calculated from Vc and Vo at ambient CO2 concentrations of 350 and 200 µmol mol−1 were 199.8 and 126.1 µmol mol−1 for rice, and 224.3 and 139.5 µmol mol−1 for wheat, respectively.

Table 1.  Rates of assimilation and respiration components in leaves of Oryza sativa and Triticum aestivumThumbnail image of

The reassimilation of photorespired 12CO2 from the IAS (AR) and the reassimilation within the cell (AI) are quantified from the measured flux rates of both isotopes and the respective CO2 concentration in the IAS (Ci). In both rice and wheat exposed to CO2 concentrations of 350 or 200 µmol mol−1, the gross CO2 flux into the leaf (A13C) was considerably higher than the steady-state, net CO2 assimilation rate (Anet) (Table 1). CO2 reassimilation via the IAS (AR) was 20–23% of all CO2 evolved from the mitochondria. At current levels of atmospheric CO2, intracellular reassimilation (AI) was substantial in both species, with 29% of the photorespired CO2 being recovered in rice and 26% in wheat (Table 1). In rice, intracellular reassimilation increased to 38% of the evolved CO2 following a decrease in ambient CO2 concentration from 350 to 200 µmol mol−1; this increase was not observed in wheat (Table 1). Overall, intrafoliaceous reassimilation at a CO2 concentration of 350 µmol mol−1 was 51% for rice and 46% for wheat (Table 1), equivalent to values observed previously (Pärnik & Keerberg 2007).

The impact of CO2 trapping on photosynthesis was estimated by modelling the CO2 assimilation rates that would be present in the absence of intracellular reassimilation of CO2. In rice, intracellular reassimilation of photorespired CO2 stimulates Anet by 11% at 350 µmol mol−1 CO2 and by 33% at 200 µmol mol−1 CO2; in wheat, Anet is enhanced by 9% at 350 µmol mol−1 and 17% at 200 µmol mol−1 (Table 1). These increases in Anet correspond to an increased chloroplastic CO2 concentration of 14 µmol mol−1 at an ambient CO2 of 200 µmol mol−1, and 12–16 µmol mol−1 at an ambient CO2 of 350 µmol mol−1 (Table 1).

In six C3 plant species, there was a strong negative relationship between Sc/Sm and the apparent CO2 compensation point of Rubisco in the absence of mitochondrial respiration, C* (Fig. 5a). Rice has the highest Sc/Sm of the species examined, followed by wheat (Fig. 5a). Sc/Sm in tobacco, rye, sunflower and F. cronquistii ranged between 0.5 and 0.8; none of these species were observed to produce stromules. Of particular importance, the predicted C* value at Sc/Sm = 0 was statistically equivalent to the mean of the true CO2 compensation points of Rubisco (Γ*) previously determined in vitro for rice, wheat, sunflower, tobacco and F. cronquistii. We observed a similar result in aging wheat leaves with a wide variation in chloroplast coverage (Fig. 5b). Close to 80% of the variation in C* was explained by Sc/Sm, indicating that Sc/Sm is the main determinant of C* (Fig. 5).

Figure 5.

The relationship between the fraction of the cell periphery facing the intercellular air space that is covered by chloroplasts (Sc/Sm) and the intercellular CO2 partial pressure at the CO2 compensation point (C*) for (a) six different C3 species (rice, Oryza sativa; wheat, Triticum aestivum; Flaveria cronquistii; rye, Secale cereale; sunflower, Helianthus annuus and tobacco, Nicotiana tabacum); and (b) senescing wheat leaves that cover a range of Sc/Sm values. Both panels show the in vitro CO2 compensation point of Rubisco (Γ*), calculated as the average from published Rubisco specificity values, with an assumed Sc/Sm value of zero (open square). The regression lines shown are: (a) y = −12.79x + 58.19; (b) y = −10.68x + 55.98. n = 3 to 5 ± SE.


The presented technique using 12CO2/13CO2 isotopes in connection with membrane inlet mass spectrometry allows for measurements of the net CO2 fluxes inside the leaf. In vivo rates of photorespiration and respiration as well as reassimilation can be determined accurately, which also provide an estimate for Rubisco carboxylation and oxygenation. Hereby, the use of 12CO2/13CO2 has the advantage over pulse chase methods using 14CO2 in that the plant, grown in normal air, is fully ‘pre-labelled’ with 12C. Briefly exposed to a gas stream containing 13CO2, the 12CO2 evolution from all source processes and any kind of substrate is accounted for during the measurement, and the gradual replacement of emitted 12CO2 with 13CO2 can be traced and accounted for during the measurement (Fig. 3a,b, inserts). Previous studies using 12CO2/13CO2 isotope techniques do not take into account intracellular reassimilation, leading to an underestimation of rates of CO2 assimilation and evolution as well as reassimilation (Haupt-Herting et al. 2001; Haupt-Herting & Fock 2002). Our results show that intracellular reassimilation is substantial and should not be neglected.

Enhancing photosynthesis by refixation of photorespired CO2

C3 plants utilize a mechanism that compensates for high rates of photorespiration by trapping and reassimilating photorespired CO2 within single mesophyll cells. By doing so, they create a supplemental CO2 supply that enhances carbon gain in warm, low CO2 environments. As the rate of photorespiration increases, the significance of the trapping mechanism also increases, such that where photorespiration would have been most severe – at high temperature and low CO2 concentrations – the stimulation of Anet is substantial, exceeding 33% in rice.

Photorespiratory CO2 trapping should result in an inverse relationship between C* and Sc/Sm. Higher Sc/Sm would reduce pathways of low-resistance out of the cell and thus proportionally more photorespired CO2 would pass into chloroplasts in addition to the CO2 entering from the IAS, raising the chloroplast CO2 concentration (Cc). Higher Cc would in turn decrease Vo/Vc, the ratio of oxygenase to carboxylase activity, and thus C* (von Caemmerer 2000). Consistently, a linear, inverse relation was observed between C* and Sc/Sm, regardless of whether variation in Sc/Sm was generated by comparing different species or wheat leaves of different age. Chloroplasts and stromules in young wheat and rice leaves formed a near complete sheath around the cell periphery facing the IAS, and correspondingly, they had the highest observed reduction in C* of 10 µmol mol−1 below Γ*. In cells with less coverage, such as tobacco, where Sc/Sm was about 0.6, the reduction in C* below Γ* was less, only 5 µmol mol−1. Older leaves of wheat also exhibited reduced Sc/Sm, and correspondingly, had a less effective trapping mechanism as indicated by a rise in C* towards Γ*. These results demonstrate that C* is an effective index of photorespiratory CO2 trapping in C3 leaves, thereby allowing for comparisons between species, developmental stages, environmental conditions and different mechanisms of reducing photorespiration.

C4 plants reduce C* to near 0 µmol mol−1, allowing for high Anet at low CO2 but at the cost of additional ATP to operate the C4 metabolic cycle. At 30 °C, rice and wheat reduce C* by 8–10 µmol mol−1, or a fifth as much as C4 plants, but without the added ATP cost. Although not as effective as C4 photosynthesis in warm conditions, photorespiratory CO2 trapping partially offsets the benefits of the C4 pathway while maintaining the advantage of the C3 pathway during cooler periods of the day and growing season. Plants termed C3-C4 intermediates also employ a mechanism to trap and reassimilate photorespired CO2, although it operates between leaf mesophyll and bundle sheath cells rather than within a single cell (Sage et al. 2012). In C3-C4 intermediates, C* is reduced below Γ* by 20–30 µmol mol−1 (Vogan, Frohlich & Sage 2007), indicating that the two-tissue mechanism of C3-C4 intermediates is 2–3 times more effective than the single-celled system of rice and wheat. The thick cell walls and large vacuoles of C3-C4 intermediates slow the diffusive efflux of photorespired CO2 more than is possible with a single layer of chloroplasts and stromules, thus allowing for greater trapping efficiency and more effective refixation. The C3-C4 mechanism comes at a cost, however, because these plants are relatively few in number and are restricted to very hot, often disturbed soils, where photorespiration is very high (Sage, Christin & Edwards 2011; Sage et al. 2012). Numerous efforts have been made to engineer plants with reduced photorespiration. The most successful of these has been the introduction of a bacterial pathway for glycolate metabolism into the chloroplasts of Arabidopsis thaliana, thereby creating a metabolic bypass that releases photorespired CO2 inside the chloroplast and increases Cc (Kebeish et al. 2007). In Arabidopsis, the photorespiratory bypass reduced C* by 6 µmol mol−1 below the wild type, or about 8–10 µmol mol−1 below the Γ* corresponding to the measurement temperature of 26 °C. The similarity between the reduction in C* in rice and Arabidopsis indicates that the photorespiratory CO2 trapping confers similar benefits as the photorespiratory bypass mechanism.

Stromules and CO2 refixation

The highest level of CO2 refixation was present in the two species that formed stromules. Wheat and rice showed extensive stromule formation that allowed Sc/Sm to approach 1.0; the other species examined did not form stromules under our growth conditions and Sc/Sm never exceeded 0.8 in young leaves. These observations support the hypothesis that stromules in photosynthetic cells function to seal breaches between adjacent chloroplasts, thus improving the trapping mechanism (Sage & Sage 2009). Stromules are known to form in mesophyll cells of plants exposed to warmer temperatures (Holzinger et al. 2007). These observations, combined with our results, indicate that stromules represent a plastic response to improve the trapping ability of mesophyll cells under conditions favourable to photorespiration.

Assessment of the method

In addition to providing estimates for reassimilation, our method allows for a direct measurement of RPR and Rd. Four methods are commonly used for the non-invasive measurement of photorespiration: (1) the burst of CO2 following sudden darkness; (2) inhibition of photorespiration by low O2 concentrations; (3) CO2 efflux into CO2 free air; and (4) uptake of 14CO2 labelled air. All four methods have significant deficiencies, as outlined by Sharkey (1988). A more recent method using 14CO2 circumvents these deficiencies, but is still based on the assumption that photorespiration is linearly dependent on the oxygen concentration (Pärnik & Keerberg 2007). Another method was introduced recently to measure photorespiration more accurately using stable carbon isotopes (Haupt-Herting et al. 2001). However, the omission of the measurement of intracellular reassimilation results in an underestimation of photorespiration rates. Here we describe a method that avoids the mentioned limitations and could therefore prove as a powerful technique to measure photorespiration in vivo.

Similarly, measuring mitochondrial respiration (Rd) is problematic because of interference from photosynthetic and photorespiratory fluxes, yet accurate Rd estimates are essential to properly understand and measure vegetation carbon flux (Yin et al. 2011). Rd is measured indirectly using leaf gas exchange (Kok 1948; Laisk 1977; Yin et al. 2009), or directly by carbon isotopes transients (Haupt-Herting et al. 2001; Loreto, Velikova & Di Marco 2001; Pärnik & Keerberg 2007); however, not all these methods account for reassimilation and they require changes in CO2 or O2 concentration, or light intensity, all of which may affect Rd. Here we provide a new means of directly measuring Rd that eliminates the need to vary O2 (Pärnik et al. 2007) and light intensity (Kok 1948; Yin et al. 2009), or conduct the measurements at low CO2 levels (Laisk 1977). There is contrasting evidence for the effect of high CO2 concentrations on Rd (Kirschbaum & Farquhar 1987; Amthor, Koch & Bloom 1992; Pärnik et al. 2007); however, this effect might arise, at least in part, from not considering reassimilation of the released CO2. Our measurements do require elevated CO2 transients, but these are much faster than the ones used for current methods conducted at elevated CO2 (Haupt-Herting et al. 2001) and therefore may be more accurate. Further experiments could elucidate to what degree the observed difference between mitochondrial respiration in the dark and in the light is methodological and in fact the result of reassimilation of released CO2 in the light. When we use an Arrhenius function to adjust for the effect of temperature (von Caemmerer 2000), Rd at 25 °C equals 0.48 ± 0.04 µmol m−2 s−1 for rice and 0.58 ± 0.07 µmol m−2 s−1 for wheat. These values fall within the range of reported values for these species determined with other methods (Pärnik et al. 2007; Yin et al. 2011). Our measurements observed no consistent CO2 effect on Rd in wheat or rice, but at both CO2 concentrations wheat had a higher Rd than rice, yet equivalent CO2 assimilation rates. Rice has thinner leaves than wheat, and half the mesophyll cell volume, which may explain the lower Rd (Sage & Sage 2009). Lower Rd in rice than wheat indicates lower maintenance costs which would offset the greater photorespiration cost in its warm habitat.

Consequences of photorespiratory CO2 refixation for the modern flora

Approximately 75% of terrestrial net primary productivity arises from C3 photosynthesis (Still et al. 2003), so a well-developed ability to reassimilate photorespiratory CO2 has important implications for vegetation-atmosphere dynamics and the global carbon cycle (Heimann & Reichstein 2008). The leading models of C3 photosynthesis use Γ* as an index of Vo/Vc (von Caemmerer 2000). In most cases, C* is used to parameterize Γ*, which could be imprecise if photorespiratory CO2 trapping is pronounced, as it is in rice and wheat. Current models often use C* estimates from tobacco (Bernacchi et al. 2001), a species with relatively low trapping potential compared with rice. We recommend that future modelling efforts either use direct Γ* measurements determined in vitro, or correct for deviations between C* and Γ*. The C* versus Sc/Sm relationship can provide correction factors, and may also be used to estimate Γ* in a given species. Corrections would be particularly important when modelling C3 photosynthesis through time, as the changing atmospheric CO2 and temperature will affect photorespiratory potential. Trapping of photorespired CO2 would increase net primary productivity in low CO2 atmospheres such as those that predominated during the ice ages (Pagani et al. 2009), and would increase the temperature required for C4 photosynthesis to be more productive than C3 photosynthesis (Ehleringer, Cerling & Helliker 1997).

During the low CO2 episodes in Earth's recent history, C3 plants experienced intercellular CO2 concentrations approaching the estimated minimum of 100–150 µmol mol−1 predicted for successful growth and reproduction in warm environments (Lovelock & Whitfield 1982; Campbell et al. 2005). In warm environments, terrestrial net primary production may collapse below CO2 concentrations of 180 µmol mol−1 (Pagani et al. 2009), in large part because the inhibition of photosynthesis by photorespiration can exceed 50% above 30 °C (Sage & Coleman 2001; Sage 2004). C4 plants should be favoured at the low CO2 concentrations of the late Pleistocene at growing-season temperatures above 10 °C (Ehleringer et al. 1997). There is, however, little evidence for widespread suppression of the C3 flora during that time (Coltrain et al. 2004; Bush, Gosling & Colinvaux 2011; Cowling 2011), and to this day, C3 plants dominate the earth's flora – C3 photosynthesis occurs in over 90% of terrestrial plant species (Sage, Li & Monson 1999) and accounts for about 77% of terrestrial primary productivity (Still et al. 2003). Even low latitude landscapes, where photorespiration is significant, are dominated by C3 plants. Approximately 72% of the world's 270 000 plant species occur in tropical regions (Antonelli & Sanmartin 2011), and of these, 170 000 use the C3 photosynthetic pathway (about 15 000 tropical species are CAM and 5000 are C4 plants; Sage et al. 2011 and unpublished results). Rice and wheat were both domesticated when atmospheric CO2 levels were near 270 µmol mol−1 (Sage & Zhu 2011). Rice in particular stands out as a plant of warm to hot environments, where photorespiration can be severe, raising the question of how it, and so many C3 plants in general, was able to persist in warm, low CO2 environments of recent geological time.

Reassimilation of photorespired CO2 could explain why so many C3 species persisted in the tropics and subtropics during low CO2 episodes, even in conditions when competition from C4 plants was severe. This would be particularly true for rice, a species growing in hot, high-light environments, where C4 plants do particularly well. Rice had the highest Sc/Sm values and lowest C* values of the species observed here, and wheat was a close second. The ability of rice and wheat to effectively trap and reassimilate photorespired CO2 could explain why these species have become so important to global agriculture. Greater productivity in the 270 µmol mol−1 atmospheres of the early Holocene could have attracted the attention of proto-farmers and led to the domestication of these species, and consistently better performance throughout the Holocene could have led to widespread use of these crops. Today, however, the adaptive significance of the photorespiratory CO2 trap appears to be declining, as indicated by the 8 to 11% enhancement of Anet observed in wheat and rice at current levels of CO2.

In conclusion, C3 plants are more resourceful with their use of photorespired CO2 than previously thought, a finding that has widespread implications when estimating C3 photosynthesis. Subsequent research is needed to clarify the overall significance of photorespiratory CO2 trapping, particularly in hot, tropical settings, where the bulk of the world's species occur. Variation in Sc/Sm and C* indicates that photorespiratory CO2 trapping is a dynamic trait, and may be induced by enhancing stromule formation. If so, enhancing the trapping mechanism through either breeding or molecular engineering may be a means to improve crop production in warm environments.


We thank R. Giuliani, J. King, K. Sault and N. Ubierna for technical assistance. This research was supported by NSERC Discovery grants to R.F.S and T.L.S. by the Division of Chemical Sciences, Geosciences, and Biosciences, Office of Basic Energy Sciences of the US Department of Energy through Grant DE- FG02_09ER16062 for funding the mass spec reassimilation measurements and instrumentation obtained through an NSF-MRI grant (#0923562) to A.B.C., and the IRRI C4 Rice programme.