Carbon partitioning in soybean (Glycine max) leaves by combined 11C and 13C labeling


Author for correspondence:

Jacob Schaefer

Tel: +1 314 935 6844

Fax: +1 314 935 4481



  • We labeled soybean (Glycine max) leaves with 200 and 600 ppm 13CO2 spiked with 11CO2 and examined the effects of light intensity and water stress on metabolism by using a combination of direct positron imaging and solid-state 13C nuclear magnetic resonance (NMR) of the same leaf.
  • We first made 60-min movies of the transport of photosynthetically assimilated 11C labels. The positron imaging identified zones or patches within which variations in metabolism could be probed later by NMR. At the end of each movie, the labeled leaf was frozen in liquid nitrogen to stop metabolism, the leaf was lyophilized, and solid-state NMR was used either on the whole leaf or on various leaf fragments.
  • The NMR analysis determined total 13C incorporation into sugars, starch, proteins, and protein precursors.
  • The combination of 11C and 13C analytical techniques has led to three major conclusions regarding photosynthetically heterogeneous soybean leaves: transient starch deposition is not the temporary storage of sucrose excluded from a saturated sugar-transport system; peptide synthesis is reduced under high-light, high CO2 conditions; and all glycine from the photorespiratory pathway is routed to proteins within photosynthetically active zones when the leaf is water stressed and under high-light and low CO2 conditions.


Rubisco (ribulose-1,5-bisphosphate carboxylase/oxygenase) is an enzyme in plants whose original function was to fix carbon in an atmosphere of high CO2 concentration and negligible O2 concentration > 2 billion yr ago (Broda, 1975). This carboxylase activity, termed photosynthesis, converts atmospheric CO2 into carbon-rich compounds and leads to the release of O2. When plants appeared on Earth, the oxygen content of the atmosphere increased. Rubisco now operates under a CO2 concentration of c. 385 ppm (by volume) and an O2 concentration of 21% (210 000 ppm).

Atmospheric CO2 concentrations are anticipated to increase to c. 600 ppm in < 40 yr (Prentice, 2001). This will be the highest CO2 concentration on Earth in millions of years. Some analysts are convinced that a benefit of high CO2 will be increased photosynthetic activity. However, the expected theoretical gain in productivity has not been realized for soybean (Glycine max) plants grown in the open field under 550-ppm free-air CO2 enhanced (FACE) conditions (Rogers et al., 2004). These experiments used computer-controlled CO2 release from pipes surrounding 25-m-diameter rings; there were no artificial canopies, no root constraints, and no nitrogen limitations. Soybean plants appear to respond to high CO2 concentrations with fully increased photosynthetic rates and biomass accumulations (relative to those under ambient CO2 concentrations) in the absence of water stress in open-top chamber experiments (Ziska et al., 2001), but not in the presence of periodic water stress, either in open-top chamber experiments (Huber et al., 1984) or FACE experiments (Ziska & Bunce, 2007).

Because water stress appears to be a factor for plant productivity at high CO2, photorespiration may also be involved (see next paragraph). Photorespiration refers to the oxygenation of ribulose-1,5-bisphosphate (RuBP) by Rubisco (Berry et al., 1978) and the events that are stimulated by the immediate production and subsequent rescue of a two-carbon molecule that is not compatible with the Calvin cycle (Bowes et al., 1971; Ogren, 1984). Carbon salvage requires machinery distributed over three organelles to convert a two-carbon molecule (phosphoglycolate) into a useful three-carbon intermediate (3-phosphoglycerate) via the condensation of two glycines to form a serine with the concomitant release of CO2. This process is known as the photorespiratory C2 cycle. Photorespiration is generally considered a wasteful process, an unavoidable side-reaction of Rubisco. The notion that a reduction in photorespiration in C3 plants may lead to increased plant productivity has been expressed frequently over the years (for recent references, see e.g. Zhu et al., 2004; Blankenship et al., 2011).

Using solid-state nuclear magnetic resonance (NMR), we have found that a fraction of the glycine from the photorespiratory C2 cycle is not decarboxylated or returned to the Calvin cycle (Noctor et al., 1999), but is inserted instead as glycyl residues in proteins (Cegelski & Schaefer, 2005; Gullion et al., 2010). This result is consistent with earlier experiments using 14CO2 labeling and detection of label in both amino acids and proteins (Ongun & Stocking, 1965; Dickson & Larson, 1975). Glycyl-13C incorporation in leaf protein or protein precursors occurs as soon as 2 min after the start of 13CO2 labeling and is most pronounced at low external CO2 concentrations (Cegelski & Schaefer, 2006). Normally, low CO2 concentrations within a leaf result from the increased stomatal resistance and decreased gas diffusivity that accompany water stress (Bunce, 1998). These conditions deplete CO2 within the leaf, but have little effect on the much more abundant O2. Thus, the carboxylase activity of Rubisco is reduced but not the oxygenase activity (Sharkey, 1988). Under these subambient CO2 concentrations, some of the glycine from oxygenase activity has been detected by solid-state NMR in Gly-Gly peptide sequences (Yu et al., 2010), presumably part of glycine-rich proteins (GRPs) or their precursors (Cassab, 1998). In soybean leaves under 600 ppm external CO2, this routing of glycine to GRPs is less pronounced and the decarboxylation pathway is active (Yu et al., 2010), a result that suggests that a part of the soybean response to water stress may be overridden by high external CO2 concentrations.

The NMR experiments described above were performed on single intact leaves. However, leaves frequently exhibit spatial and temporal heterogeneity of their stomatal openings during active photosynthesis. Collections or ‘patches’ of stomata may close while neighbors in adjacent areas of the leaf remain open. This behavior is sometimes called ‘patchy stomatal conductance’, or ‘stomatal patchiness’ (Mott & Buckley, 2000; Mott & Peak, 2007). Patches are usually bounded by leaf veins and each patch has a uniform local conductance. Patches sometimes appear and disappear transiently on a 20-min time-scale in what appears to be a largely unpredictable way. Thus, NMR results based on whole-leaf averages may not represent accurately the metabolism in any single region of a leaf.

Patches have been observed using a variety of techniques such as iodine staining (Terashima et al., 1988), 14C autoradiography (Downton et al., 1988; Gunasekera & Berkowitz, 1992), leaf thermograms (Hashimoto et al., 1984; Jones, 1999), and leaf fluorescence (Daley et al., 1989; Genty & Meyer, 1995; Beyschlag & Eckstein, 2001). When stomata are closed, photosynthesis and electron transport from light-harvesting quinones are inhibited. The net result is that chlorophyll fluorescence as excess energy is dissipated by photon emission. Thus, the red fluorescing parts of a leaf are those for which the stomata are closed and photosynthesis has shut down.

The first two patch-detection methods mentioned above are destructive techniques, the third is most sensitive at low relative humidity, and the leaf fluorescence measurement, while the most versatile of the four, is sensitive to oxygen quenching and so works best at low concentrations of O2 (Mott & Peak, 2007). We believe that leaf patchiness and metabolic activity can be measured under high humidity and 21% O2, conditions under which photorespiration may be active, by the direct detection of 11C decay (Thorpe et al., 2007). The resulting images would provide useful space and time information about leaf metabolism and so identify regions of the leaf to be examined subsequently by NMR.

11Carbon is a positron-emitting isotope with a 20-min half life that is routinely made for use in positron (positive electron or β+ particle) emission tomography (PET). The positron emitted by 11C has an average range of c. 1 mm (in condensed matter) and then annihilates with an electron generating two 511-keV photons. The back-to-back photons serve as the signal carrier for PET imaging (Phelps et al., 1975). Coincidence detection of the annihilation photons is suitable for imaging humans, animals, and the bulk of a plant, but not for a leaf, the thickness of which can be substantially less than the 1-mm positron range. That is, the positron will often get out of the leaf, and will probably not annihilate and generate photons until it stops in some other condensed matter. Imaging the positron itself is perhaps the more appropriate tool for leaf imaging. In our work, images of 11C assimilated by the leaf are made by direct positron imaging. This is accomplished by the detection of the visible photons emitted by a plastic scintillator when the positrons generated by 11C decay are stopped within the scintillator. After imaging, the leaf is frozen in liquid nitrogen to suspend all metabolism, and lyophilized after sufficient time has been allowed for radioactive decay and safe handling of the leaf, typically a few hours. Solid-state NMR of the same intact lyophilized leaf is then possible.

In this report, we describe our efforts to answer the following two questions. Can we combine 11CO2 and 13CO2 labeling with detection of the labels by 11C positron imaging and solid-state 13C NMR to measure variations in soybean leaf metabolism for photosynthetically active and inactive regions within a single leaf? Can we use these leaf-fragment measurements to gain insight into the source(s) of the apparent shortfall in productivity (carbon assimilation) for periodically water-stressed soybean plants grown under elevated CO2 conditions?

Materials and Methods

Low-resolution imaging of leaves using 11CO2

The carbon assimilation of attached leaves of soybean (Glycine max (L.) Merr.) plants was imaged in real time following exposure to 11CO2 spiked air. One of the Washington University (St Louis, MO, USA) School of Medicine cyclotrons was used to accelerate protons to c. 16 MeV (Fig. 1a). These protons were directed at an N2 target. The reaction 14N + p → 11C + α occurs (see Fig. 1 legend for definitions) and, in the presence of O2, CO2 forms. The variable thickness of a leaf, the collimator, and the aluminized mylar optical isolator are shown in Fig. 1(b), and the dual 11C and 13C labeling experiment in schematic form in Fig. 1(c). The labeling air had controlled concentrations of 13CO2 (200 or 600 ppm) and 21% O2. The amount of 11CO2 is too small to affect the overall CO2 concentration. Radiation monitors, pairs of cesium fluoride (CsF) scintillators operating in coincidence mode, were positioned upstream and downstream of the labeling chamber (Fig. 1c) so that the 11C activity could be used to normalize one labeling experiment to another (see Supporting Information Notes S1). Once the 11CO2 entered the chamber, a uniform spatial distribution was achieved within 10 s (Figs S1, S2).

Figure 1.

Combined 11C and 13C leaf labeling. (a) Schematic of production of 11C. (b) Geometrical arrangement of leaf, collimator, and scintillator for leaf labeling. (c) Schematic of combined 11CO2 and 13CO2 labeling. (d) Labeling chamber. Approx. 3 mCi of 11CO2 is passed over an attached soybean (Glycine max) leaf as a 2-min pulse. The leaf is contained in a box, transparent on the top and viewed by the scintillating plate-CCD camera from below. Unincorporated activity is trapped downstream and held for decay. p, proton; n, neutron; α, alpha particle; β, positive electron; γ, gamma ray; X, xylem; P, phloem.

Labeling chamber

Labeling was performed inside a 180-cc chamber (142 mm inside diameter × 11.5 mm height) which was sealed by an O-ring held in place by seven thumb screws (Fig. 1d). The seal around the petiole was made with an elastic gum or vacuum grease. Leakage around this seal was removed by a negative-pressure 3-inch tube positioned just above the petiole. The gas-handling system was behind lead shielding.

Gas entered the chamber through a manifold with 12 exit ports around the periphery (Notes S1, Fig. S1). The chamber was equipped with two miniature instrument fans (Fig. 1d) to help provide a uniform distribution of the labeling gas (Fig. S2). The leaf was supported top and bottom by a mesh of nylon fishing line attached to a horseshoe-shaped ring (Fig. 1d). The carrier gas for the 11CO2 label was 13CO2-labeled air, which was circulated for 15–30 min before the 11CO2 labeling period at a flow rate of 1500 cc min−1. During the 2-min 11CO2 labeling period, the leaf was held 5 mm above the collimator to ensure uniform access of the label to the leaf. The gas flow during this time was c. 50 cc min−1. Slow-flow rates were used for the 11CO2 labeling to achieve good incorporation of label while minimizing the total radioactivity required and so the radiation hazard. At the end of the 11CO2 labeling period, the chamber was flushed with carrier gas and the fast-flow rate was re-established. The leaf was then pushed down to within 1 mm of the collimator (to optimize the imaging) using three pistons sealed with O-rings and fastened to the horseshoe ring (Fig. 1d). This operation was accomplished without opening the chamber. The fast-flow rate was maintained for times on the order of 1 h.

During the fast-flow-rate period, CO2 concentration within the chamber was essentially constant and almost equal to that of the carrier-gas cylinder. Photosynthetic assimilation by the leaf was small compared with the total flow through the chamber. The radiation monitors allowed the 11CO2 concentration in the chamber to be calculated throughout the 2-min slow-flow-rate period and the photosynthetic carbon assimilation rate of the leaf to be determined (Figs S3–S5). During this period, the average CO2 concentration was reduced by 10–50% of that of the carrier gas, depending on light intensities and photosynthetic activity of the leaf. These variations could be essentially eliminated by increasing the slow-flow rate (to 100 cc min−1) and shortening the slow-flow period (to 1 min), together with an increase in 11C activity. We prefer this approach rather than attempting to juggle the CO2 input concentrations and dealing with complicated mixing issues on a 1-min time-scale. For our initial experiments, however, we accepted the variation in CO2 concentration in favor of lower radiation levels.

The top of the leaf chamber is transparent and exposed to light. The bottom of the leaf chamber is separated from the camera by a 60-cm optics tube. The tube has a perforated G-10 circuit-board collimator (1.5-mm holes on a 2-mm square grid) above a 1-mm-thick organic scintillator. These components were separated from the leaf by a sheet of opaque aluminized mylar and a sheet of 0.027-mm aluminum foil. The scintillator emits in the green (Bicron BC-408) and was viewed using a CCD camera (Andor DU-897, Andor Technology, Belfast, UK), via a lens (f0.95) with a short depth of field. The CCD has 512 × 512 pixels and was cooled to −90°C. The aluminized mylar and the thin aluminum foil provide optical and environmental isolation of the bright and humid leaf chamber from the optics tube and the chilled camera and lens (Fig. 1b,c). The only photons visible to the camera are those generated by the scintillator. The use of the collimator ensures that the points of generation of the visible photons in the scintillator are 2D spatially correlated to the points of positron generation in the leaf. Using a thin plastic scintillator minimizes the sensitivity to annihilation photons which, unlike the positrons themselves, cannot be effectively collimated. Images started c. 2 min after the activity had swept over the leaf and the unassimilated 11CO2 had been purged. Image collection continued for 1 h.


Lighting for photosynthesis was provided by a halogen lamp. ‘High-light’ intensity (710 μmol photon m−2 s−1) was 84% of typical glasshouse growth conditions (850 μmol photon m−2 s−1), both measured using a Decagon Devices PAR sensor (Model QSO-S; Decagon Devices, Pullman, WA, USA). This device measured full sunlight (from 400-700 nm) at noon in St Louis as 2135 μmol photon m−2 s−1 (McCree, 1981). ‘Low-light’ intensity was 115 μmol photon m−2 s−1 for the experiments of Figs 3–5, and 75 μmol photon m−2 s−1 for all other low-light experiments (including those of the Supporting Information). Lighting variation from edge to edge of the imaging circle was < 10%. The temperature inside the chamber was 24°C during low-light experiments and 29°C during high-light experiments (each ± 1°C). The relative humidity was controlled near 80 ± 5% (measured using a Vaisala HMT330 humidity and temperature sensor, Helsinki, Finland), maintained by passing a portion of the carrier gas through a water bubbler (Fig. 1c). For the low-light experiments this corresponded to an air water-pressure deficit of 0.55 kPa, and for the high-light experiments, 0.69 kPa. Leaf temperature was not well defined because of photosynthetic heterogeneity.

Image analysis

Information on the transport of 11C was extracted from the images by comparing the time evolution of the count rate from collimator holes located on major veins and that from collimator holes free of any major veins. The counts from 312 camera pixels of each sort (12 collimator holes on veins and another 12 collimator holes free of major veins, with 26 pixels per each collimator hole) were averaged to obtain vein and non-vein data as a function of time. Subgroups within the set of 12 were examined to ensure that the results were not dominated by any single collimator hole and that the reported trends are generally representative of on- and off-vein behavior. To make comparisons between labeling experiments, the counts from the CCD camera were normalized to the input pulse of 11C as measured by the upstream pair of CsF annihilation radiation detectors (each with a photomultiplier tube) operated in coincidence mode. An additional normalization of a factor 3 was employed to account for the reduced fractional amount of 11C when using the 13CO2 at 600 ppm rather 200 ppm. The data were also corrected for 11C decay. The resulting relative 11C label was plotted as a function of time. The integration period for each point was 10 min, with the first period starting 2 min after flushing began (fast-flow rate), a delay chosen to guarantee that all unassimilated 11C was removed from the image area.

Solid-state NMR

Spectra were obtained using a six-frequency transmission-line probe (Schaefer & McKay, 1999), having a 12-mm-long, 6-mm inner-diameter analytical coil and a Chemagnetics/Varian/Agilent (Agilent Technologies, Santa Clara, CA, USA) magic-angle spinning ceramic stator. Lyophilized samples were contained in thin-wall Chemagnetics/Varian/Agilent 5-mm outer-diameter zirconia rotors. The rotors were spun at 7143 Hz with the speed under active control to within ± 2 Hz. The spectrometer was controlled by a Tecmag pulse programmer (Tecmag Inc., Houston, TX, USA). Radiofrequency pulses for 13C (125 MHz) and 15N (50.3 MHz) were produced by 2-kW American Microwave Technology power amplifiers (Herley Instruments, Lancaster, PA, USA). Proton (500-MHz) radiofrequency pulses were generated by a 2-kW Amplifier Systems tube amplifier driven by a 50-W American Microwave Technology power amplifier. Power-amplifier outputs were under active control to within ± 200 Hz. A 12-T static magnetic field was provided by an 89-mm bore Magnex superconducting solenoid (Magnex Scientific, Ltd., Yarton, UK). Proton-carbon cross-polarization magic-angle spinning (CPMAS) matched Hartmann-Hahn transfers were made with radiofrequency fields of 62.5 kHz. Proton dipolar decoupling was 100 kHz during data acquisition. Spectra typically involved signal averaging for one day (to improve sensitivity) for a 100-mg sample.

Spectra arising from just the 13C label were obtained by subtracting spectra arising from leaves grown under identical conditions (same pot) but not exposed to 13CO2. The subtraction took into account sample weights and total number of scans. An internal check into the quality of this procedure was available when no label was incorporated into proteins or peptides. In this situation, baselines were flat in the aliphatic-carbon region (10–50 ppm).

Plant growth

Plants for the combined 11CO2 and 13CO2 labeling experiments were grown in 15-cm-diameter pots in the glasshouse facility maintained by the Washington University Department of Biology. Plants were grown from commercial soybean seed (Stine Seed Co., Adel, IA, USA, variety 4328386, Roundup Ready®) using a formulated peat moss and coarse perlite mixture, fertilized by inorganic micronutrients and (unlabeled) ammonium nitrate and urea. Additional 13CO2 labeling, and combined 11CO2 and 13CO2 labeling experiments were also performed on seeds of soybean cultivars Williams and Loda, generously provided by Randal L. Nelson, Curator, USDA Agricultural Research Service, Urbana, IL, USA. Some plants were put under water stress by withholding water for 60 h, resulting in a leaf water deficit of c. 15% relative to a fully watered leaf (Barrs & Weatherley, 1962).


Positron images

The direct positron images are low resolution compared with a radiograph (Hanik et al., 2010) but show space–time development. Movies of 11C transport in leaves determine, in general, how the partition of carbon between mesophyll and transport elements changes with CO2 concentration and water stress. The experimental observables from 11CO2 labeling are the total label uptake, the fraction that is mobile sucrose, and the spatial variation in photosynthetic activity.

The observed variation in photosynthetic activity is presumably attributable to variations in stomatal conductance. Images from similar leaves from the same plant, under the same lighting, and with the same 11CO2 flow conditions are completely different (Fig. 2). The uptakes of 11CO2 for the central and flanking leaves of a single trifoliolate also differ (Fig. S6). If some of the observed activity were attributable to mobile sucrose then the direct positron image would indicate activity in the mesophyll decreasing faster than expected from 11C decay, and activity in the vascular system decreasing slower than the decay rate. If the activity is totally immobile (stored sucrose, starch, protein, and some protein precursors), the decrease in activity would match the 11C decay rate. The observed relative activity decreases less quickly for the veins initially than for the non-veins, causing the decay-corrected activity for the veins to increase initially, and that for the non-veins to decrease (Fig. S7, top). For longer times, there is a down-turn in activity for veins, reflecting transport out of the leaf. The transport out of the central leaf of a trifoliolate increases substantially within an hour when the flanking leaves are removed (Fig. S7, bottom), which was unexpected based on the results of conventional gas-exchange measurements (Caemmerer & Farquhar, 1984).

Figure 2.

Comparison of patchiness of two soybean (Glycine max) leaves labeled with 200 ppm 13CO2 spiked with 11CO2 and imaged on the same day (3 h apart), from the same well-watered plant, with the same low-intensity lighting (75 μmol photon m−2 s−1), and the same slow-flow gas conditions (53 cc min−1). Both images involved 10 min of signal averaging and have been corrected for radioactive decay. (a) Labeling with 5.0 mCi from 09:56 to 10:06 h. (b) Labeling with 5.5 mCi from 12:52 to 13:02 h. The two images have been normalized to equal average counts per exposed pixel.

Photosynthetic heterogeneity for well-watered leaves labeled with 600 ppm CO2 under high- and low-light conditions is presented in Fig. 3. Experiments were also performed using 200 ppm CO2 (data not shown). For both CO2 concentrations, the observed relative activity decreases less quickly for the veins initially than for the non-veins, causing the decay-corrected activity for the veins to increase initially, and that for the non-veins to decrease (Fig. 4). The relative out-of-leaf transport is faster under low-light conditions (Fig. 4b, closed orange circles) than under high-light conditions (Fig. 4a, closed red circles), even though the absolute sucrose concentration is higher under high light. The difference in carbon assimilation for the two light intensities is about a factor of 3 (Fig. 4). Only 20% of this increase is attributable to the 6°C temperature increase within the chamber (Ogren, 1984) for labeling under high-light conditions.

Figure 3.

Movement of photosynthate in a soybean (Glycine max) leaf labeled with 600 ppm 13CO2 spiked with 3 mCi 11CO2 under (a) high- and (b) low-light conditions. Images were made by direct positron detection and are corrected for radioactive decay. The two images at the far left (top and bottom) have been normalized for equal average counts per exposed pixel. Each image was accumulated over 10 min. Exposure to 11CO2 was for 2 min followed by a 2-min flush period. Time evolution began at the end of the flush period. Photosynthetic activity (red–yellow) is non-uniformly distributed over the leaf area. Much of the 11C label is immobile under high-light conditions and mobile under low-light conditions. A hole punched in the leaf was used to orient the images with respect to the chamber.

Figure 4.

Time evolution of 11C photosynthate in soybean (Glycine max) leaves exposed to (a) high- and (b) low-intensity light as a function of 13CO2 concentration. The 600 ppm 13CO2 data in (a) and (b) were taken from the images shown in Fig. 3. The closed symbols are for pixels lying along the center vein of the leaf, and the open symbols for representative pixels not along the center vein. The solid and dashed lines are linear best fits, except for the center-vein data for 600 ppm 13CO2 which are best-fit quadratics (a, red solid line; b, orange solid line). The solid and dashed lines diverge, reflecting movement of 11C label out of leaf mesophyll cells and into the central vein. The initial positive slope of the high-light condition best fit for 600 ppm 13CO2 (a, red solid line) is 2 times the slope of the high-light linear fit for 200 ppm 13CO2 (a, green solid line).

NMR analysis of 13C labeling

The experimental observables from 13CO2 labeling with solid-state NMR detection are the total label retained in the leaf, and distribution of this label in starch, sucrose, proteins and protein precursors. A comparison of four leaves labeled under different CO2 concentrations and light intensities is made in Fig. 5. Variations in signal intensities are consistent with 11C activities reported in Fig. 4. The total intensity of the 170–180 ppm carbonyl-carbon region is approximately independent of CO2 concentration but dependent on light intensity. (Note the ×2 scale factor for the low-light column in Fig. 5.) This suggests that the Rubisco oxygenase pathway is supplying most of the carbonyl carbons during the labeling period (Cegelski & Schaefer, 2006; Yu et al., 2010). Chemical-shift resolution of the 100 ppm anomeric-carbon region indicates that both sugars and starch are produced in high light at both CO2 concentrations (Fig. 5a, boxed inset, and Fig. 5c), whereas only sugars are produced in low light (Fig. 5b,d). Peptides are produced in low and high light at low CO2 concentrations but total peptide production is suppressed in high light and high CO2 (Fig. 5a, red arrow).

Figure 5.

125-MHz 13C cross-polarization magic-angle spinning (CPMAS) difference spectra of intact lyophilized soybean (Glycine max) leaves labeled in vivo with a combination of 11CO2 and 13CO2 at a concentration of 600 ppm (a, b) and 200 ppm (c, d) using high-intensity light (710 μmol photon m−2 s−1; a and c) and low-intensity light (115 μmol photon m−2 s−1; b and d). Exposure of the leaf to 11CO2 was for 2 min, and to 13CO2 for 60 min. Natural-abundance 13C peaks have been removed by subtraction so nuclear magnetic resonance (NMR) signals are attributable only to label. Images obtained from direct detection of positrons resulting from 11C decay are shown in the insets. The images were made 60 min after the 11CO2 exposure (at the end of the movie) just before freezing and lyophilization of the leaf (see Fig. 3). Low light results in the partitioning of a large fraction of photosynthate into sucrose (b, d). Under high CO2 and high-light conditions, the direct incorporation of glycine into protein from the Rubisco oxygenase pathway is suppressed (a, red arrow) and starch is a major product instead (boxed inset). The punched hole damaged a part of the vascular system of the leaf of (d). ppm, parts per million; ssb, spinning sidebands.

Similar results have been obtained using rotational-echo double resonance (REDOR) solid-state NMR (Gullion & Schaefer, 1989) for plants grown and labeled in full sunlight (Fig. S8). These spectra are analyzed in the Supporting Information (Notes S2; Figs S8–S11) where a brief tutorial on solid-state NMR may be found.

The high-light NMR results of Fig. 5(c) show that starch is deposited at 200 ppm 13CO2 even though the sucrose concentration is about half that at 600 ppm 13CO2. The starch-to-sucrose ratio is about the same in Fig. 5(a,c), based on the intensities of the anomeric-carbon peaks near 100 ppm (see Fig. 5a, boxed inset). In addition, the rate of accumulation of sucrose in the central vein at 200 ppm is also half that at 600 ppm in high light (Fig. 4a, red and green closed symbols), and one-fourth that in low light (Fig. 4a,b, orange and green closed symbols). These comparisons indicate that sucrose phloem loading in the central vein at 200 ppm in high light is not close to saturation. That is, transient starch deposition occurs even though sucrose concentration is low and the phloem transport system is at no more than half capacity.

Photosynthetic heterogeneity

We used a 11C positron image to make a template identifying variations in photosynthetic activity within a single soybean leaf (Fig. 6a, inset, top right; Fig. 6b, 70 min). The 13C label-only spectrum of a non-vein fragment of reduced activity as measured by 11C uptake (Fig. 6a, top middle) has a total 13C intensity comparable to that of the similarly labeled leaf of about the same size shown in Fig. 5(d), but with less labeled sucrose, more starch, and considerably less peptide. This is possibly an indication of the developmental stage of the two plants, with that of Fig. 6 more advanced by 4 wk. (The labeling was performed 5 wk after planting in the experiment of Fig. 5, compared with 9 wk in the experiment of Fig. 6.) The most active fragment (by 11C uptake) includes the central vein and has more 13C total label than either of the non-vein fragments, the same sucrose and starch concentrations, a reduced peptide content, and an enhanced peak near 102 ppm (Fig. 6a, top right). We assign the latter to the anomeric carbons of D-fructose conformers (Koerner et al., 1978). Apparently, the higher photosynthetic activity of this fragment has been supported both by conversion of sucrose to glucose (and subsequently to glycerate for RuBP production) and fructose, and by glycine routing back to the Calvin cycle, the latter resulting in a reduced peptide peak. For the more active non-vein fragment (by 11C uptake), starch production is down and peptide production is up significantly, relative to the other two fragments (Fig. 6a, top, horizontal dashed green internal reference lines). The same inverse correlation was observed for leaves characterized by REDOR NMR (Fig. S8).

Figure 6.

Use of a template from a 11C image to identify by nuclear magnetic resonance (NMR) the chemical composition of soybean (Glycine max) leaf patches. (a) NMR spectra of three fragments of a lyophilized leaf from a combined 11C and 13C labeling experiment using 200 ppm 13CO2 and low-light conditions (75 μmol photon m−2 s−1). Full spectra are shown at the bottom of the panel. The corresponding label-only spectra shown at the top of the panel have had contributions from the natural-abundance 13C in the leaf removed by subtraction. Dotted lines in red identify trends in intensity for carbohydrates, in green for peptides, and in blue for peptide side-chain aliphatics in the three leaf fragments. Superposition of the leaf and its positron (false-color) image after 70 min are shown in the inset (top right). Leaf fragments of varying photosynthetic activity formed a template (white lines). The template was used to guide cuts of the leaf by a razor blade. (b) Movement of photosynthate in the leaf shown in the same format as that used for Fig. 3. All of the positron images involved 10 min of signal averaging and have been corrected for radioactive decay.

Water stress and combined 11C and 13C labeling

We also used 11C imaging to identify those regions of a leaf under water stress and exposed to 600 ppm 13CO2 that were still engaged in gas exchange (Fig. 7a, top right; Fig. 7b, 70 min). Stomatal openings along the sides of the leaf and near the tip were closed, while those near the center of the leaf and the main vascular bundle remained open. Assimilation of 13C label in all parts of the leaf showed the presence of internal redistribution of the 600 ppm 13CO2 on a time-scale of 60 min (Fig. 7a). Most of the label was partitioned into sucrose and starch and was not mobile (Fig. 7b). There was some evidence of peptide formation (c. 10% of the total label assimilated) near the tip of the leaf where starch deposition was reduced (Fig. 7a, top left), the same inverse correlation noted in the description of Figs 5, 6.

Figure 7.

Use of a template from a 11C image to identify by nuclear magnetic resonance (NMR) photosynthetically active regions of a water-stressed soybean (Glycine max) leaf in a high CO2 atmosphere. (a) NMR spectra of three fragments of a lyophilized leaf from a combined 11C and 13C labeling experiment using 600 ppm 13CO2 and high-light conditions (710 μmol photon m−2 s−1). The presentation has the same format as used in Fig. 6. An absolute vertical-scale intensity bar is shown in the bottom panel, extreme right for comparison to intensities in Fig. 8. The red, green, and blue dotted lines indicate trends. (b) Movement of photosynthate in the leaf shown in the same format as that used in Fig. 3. All of the positron images involved 10 min of signal averaging and have been corrected for radioactive decay.

Partitioning of 13C label in a similar water-stressed leaf exposed to 200 ppm 13CO2 was completely different. The gas-exchange characteristics were much the same (Fig. 8a, top right), but now about half of the total label appeared in peptide and protein (Fig. 8a, top left and right, black hatch marks), and there was no starch deposition. Export of label over the 70 min of the imaging (Fig. 8b) was less than that observed under high CO2 conditions (Fig. 7b). By using the absolute vertical-scale display bars as a guide (Fig. 7a, bottom, extreme right; Fig. 8a, bottom, extreme right), the integrated intensity of the 13C label-only spectrum obtained under high CO2 conditions (Fig. 7a, top right) can be compared to that obtained under low CO2 conditions (Fig. 8a, top right). This ratio of intensities is 2.2 (Table 1). Similar protein-rich, starch-poor patterns are observed for all the fragments of the low CO2-exposed leaf (Fig. 8a).

Figure 8.

Use of a template from a 11C image to identify by nuclear magnetic resonance (NMR) photosynthetically active regions of a water-stressed soybean (Glycine max) leaf in a low CO2 atmosphere. (a) NMR spectra of three fragments of a lyophilized leaf from a combined 11C and 13C labeling experiment using 200 ppm 13CO2 and high-light conditions (710 μmol photon m−2 s−1). The presentation has the same format as used in Fig. 6. The red and black cross-hatching (top left) indicate carbon partitioning into carbohydrate and protein, respectively. The inset (top middle) shows the deconvolution of the 170–180 ppm chemical-shift region into carboxyl and peptide carbonyl carbons by using shift positions and line shapes determined by rotational-echo double resonance (Supporting Information Fig. S8). The dotted red and blue lines are for internal peak-height comparisons. An absolute vertical-scale intensity bar is shown in the bottom panel (extreme right) for comparison to intensities in Fig. 7 where the scale has been compressed by a factor of 2.3. (b) Movement of photosynthate in the leaf shown in the same format as that used in Fig. 3. All of the positron images involved 10 min of signal averaging and have been corrected for radioactive decay.

Table 1. Photosynthesis and photorespiration of water-stressed soybean (Glycine max) leaves
CO2 concentration (ppm)Carboxylase rateaOxygenase rateDecarboxylation (D)Net carbon assimilation
  1. a

    Arbitrary units.

  2. b

    Extrapolated from 400 ppm rate assuming a linear response.

  3. c

    Ratio of carbon assimilation at 600 ppm CO2 to that at 200 ppm is 2.6 (experimental value is 2.2 from Figs 7 and 8).

  4. d

    Estimates of rates from Sharkey (1988).

  5. e

    D ≈ 0 means that all glycine produced by oxygenase C2 pathway is routed to protein.



Metabolic variations between leaf patches

Stomatal patchiness was observed in our 11CO2 soybean labeling experiments on unstressed leaves (Figs 2,5,6). The red color (arbitrary choice) was associated with photosynthetic activity which was clearly non-uniform within a leaf and from leaf to leaf. Because we used pulse-chase labeling, we observed movement of photosynthate, which means that we often saw parts of patches fading in intensity with time. With the positron image as a guide, NMR easily identified sizeable differences in metabolic activity within a single leaf. The negative correlation between starch deposition and protein synthesis observed for intact whole leaves (Fig. 5) was also observed for individual leaf fragments: more starch, less protein (Figs 6–8). The systematic trends in these fragment spectra acted as replicate experiments on the same leaf. We believe such comparisons of metabolism are more insightful than comparisons of minor differences between averages taken over many different leaves, each of which can have high internal variability in metabolism.

Stomatal patchiness appears to reflect water-status heterogeneity (Beyschlag & Eckstein, 2001) and as such is under a variety of influences. In our measurements, the most pronounced patchiness was observed for water-stressed leaves (Figs 7, 8). Variations in photosynthetic activity of a factor of 4 were observed, of which at most only 10% could be attributable to variation in light intensity within the chamber. Metabolic activity monitored by solid-state 13C NMR varied from patch to patch, which rules out heterogeneities arising exclusively from slow-flow conditions and lasting only for the 2 min of 11C labeling.

Transient starch deposition, the conversion of glucose to starch catalyzed by ADP-glucose pyrophosphorylase (AGPase) in the chloroplasts (Zeeman et al., 2007), is more pronounced in photosynthetically active patches (Fig. 6). The routing of label predominantly to starch in photosynthetically active leaves exposed to high 14CO2 concentrations is well known (Zeeman et al., 2002). Transient starch deposition is sometimes considered a leaf response to sucrose ‘overflow’; that is, a storage of sugar in excess of the capacity of the phloem loading and transport system (Preiss, 1988; Stitt & Quick, 1989). However, the results of Figs 4 and 5 show that starch deposition occurs even when the central-vein transport system is far from saturation. Perhaps transport saturation is associated with some smaller veins and so occurs more locally and on a small scale.

The metabolic function of transient starch deposition has also been identified as possibly part of an acclimatory mechanism that allows plants to respond to reduced carbon availability by increasing storage and decreasing growth (Smith & Stitt, 2007). This mechanism functions on a diurnal time-scale and if operative may influence the exact partitioning between sucrose and starch in our experiments.

We observed a strong positive correlation between carbon availability and starch deposition on a 1-h time-scale (Figs 5, 6). Starch deposition in our experiments was higher in high light, consistent with light-dependent changes in the redox activation of AGPase (Hendriks et al., 2003; Kolbe et al., 2005). Starch deposition was also higher for 600 ppm CO2 than for 200 ppm CO2. Nevertheless, for well-watered leaves we observed significant starch deposition even for low carbon assimilation rates (Fig. 5c), which complicates formulating a simple correlation.

Starch deposition may have multiple metabolic functions. There is no question that it plays a role in carbon partitioning on a 1-h time-scale (Fig. 6a), and so is integrated into overall plant growth (Schulze et al., 1991). Starch deposition also responds on a longer time-scale, for example as carbon partitioning adjusts to changes in day length (Smith & Stitt, 2007). But a complete understanding of transient starch deposition has not yet been achieved, in particular an explanation of the observation that, for soybeans and other broadleaf C3 plants, the extra carbon assimilated under high CO2 conditions is not immediately deployed throughout the plant but is stuck in the leaf as starch (Poorter & Navas, 2003), at least temporarily. This delay unfortunately makes the leaf an attractive target for insect attack (Zavala et al., 2008).

Water stress and efficiency of carbon assimilation

We compared the metabolism of the most photosynthetically active fragment of a water-stressed leaf under high light and 600 ppm 13CO2 (Fig. 7a, top right) with that of a similarly photosynthetically active fragment under 200 ppm 13CO2 (Fig. 8a, top right). The partitioning between starch and protein was completely different, which suggests a major difference in the control of the Rubisco carboxylase and oxygenase pathways.

Using the absolute vertical-scale display bars, we saw that the total integrated intensity ratio for the two fragments was 2.2. That is, a little more than twice as much 13C label was assimilated at the higher CO2 concentration in the active patches. Because no peptide was observed for the photosynthetically active fragment at 600 ppm 13CO2 in high light, the photorespiratory C2 pathway was closed; that is, all glycine was returned to the Calvin cycle via serine formation with a single glycine decarboxylation and release of CO2 for each 3-phosphoglycerate formed.

A reasonable estimate of the ratio of carboxylase to oxygenase rates under near-ambient CO2 conditions (400 ppm CO2) at 25°C is c. 5 : 2 (Sharkey, 1988; Foyer et al., 2009). We will assume that, for the leaf labeled at 600 ppm 13CO2, the ratio of the carboxylase rate to the oxygenase rate increased by 50% to 7.5 : 2 (Sharkey, 1988) with a decarboxylation rate (D) of 1. Thus, the net carbon assimilation was 6.5. (Table 1, arbitrary units).

For the leaf labeled at 200 ppm 13CO2, the most photosynthetically active fragment (Fig. 8a, top right), had the carboxylase rate reduced from a relative value of 7.5 (at 600 ppm) to a relative value of 2.5 (linear response to CO2 concentration). Because oxygenase activity was unchanged at the lower CO2 concentration, the carboxylase and oxygenase rates were close to equal at 200 ppm CO2 (Table 1). If D were to remain equal to 1, the net carbon assimilation at 200 ppm would be only 1.5, and the expected ratio of total integrated intensities for the most active fragments at the two CO2 concentrations would be 6.5/1.5 = 4.6. The observed ratio was 2.2, which suggests that  0 at 200 ppm 13CO2. Zero glycine decarboxylation would increase the net carbon assimilation at 200 ppm from 1.5 to 2.5, so that the expected ratio of integrated intensities at the two CO2 concentrations would be 6.5/2.5 = 2.6, in much better agreement with experiment (Table 1).

It appears, therefore, that under water stress and low CO2 concentrations, and for times on the order of at least an hour for an active leaf fragment with ample nitrogen reserves, the photorespiratory C2 cycle is completely open, no glycine is returned to the Calvin cycle via serine and 3-phosphoglycerate, and most glycine is routed to protein (Fig. 8a, top left and right). This is the same conclusion that was reached earlier by using 17O2 labeling of the C2 pathway in a water-stressed whole soybean leaf with NMR detection of 17O label only in protein, and not in starch or sugars (Gullion et al., 2010).

In our experiments, when the net carbon assimilation rate for the whole leaf was severely reduced by water stress (e.g. to 0.5 μmol m−2 s−1; Fig. S5), the soybean leaf stopped glycine decarboxylation (the production of CO2 in the light) and inserted all C2 glycine into protein. This efficiency was lost, however, for a water-stressed leaf under high CO2 concentrations (Fig. 7). Increasing photorespiratory CO2 release by increasing external CO2 concentration is a non-intuitive result (Sharkey, 1988). This result may explain why soybean plants respond to high CO2 concentrations with fully increased photosynthetic rates and biomass accumulations (relative to those under ambient CO2 concentrations) in the absence of water stress in open-top chamber experiments (Ziska et al., 2001), but not in the presence of periodic water stress, either in open-top chamber experiments (Huber et al., 1984) or in FACE experiments (Rogers et al., 2004; Ziska & Bunce, 2007).

A completely open C2 pathway is not possible for a leaf with restricted nitrogen reserves (see Fig. S10) and therefore is not possible for extended periods of time (Igarashi et al., 2006). An open C2 pathway under water stress allows the routing of glycine into dehydrins which protect the leaf against desiccation (Layton et al., 2010) and into structural cell-wall components of protoxylem elements (Ye et al., 1991; Ringli et al., 2001).


Metabolism in leaves and the details of phloem loading, sugar transport, transient starch deposition, stomatal patchiness, and protein synthesis are obviously complicated, interdependent processes. Teasing apart their connecting relationships as a function of environmental stresses (especially water and CO2 levels) requires the sort of multivariable measurement capabilities that we have demonstrated by a combination of 11C and 13C labeling, time-dependent direct positron in vivo imaging of the leaf, and solid-state NMR analysis of the same leaf. We have shown that the 11C imaging identifies leaf fragments or patches within which systematic variations in starch deposition and protein synthesis can be unambiguously established by NMR (answering the first question posed at the end of the Introduction). We have also shown that high CO2 concentrations may reduce the carbon-assimilatory efficiency of a water-stressed soybean leaf by increasing photorespiratory release of CO2 (tentatively answering the second question posed in the Introduction). However, our initial experiments were performed on a limited number of plants and leaves and so constitute what amounts to a metabolism case study that requires confirmation and elaboration. We anticipate that these will come from combining imaging and NMR of replicate samples with liquid chromatography–mass spectrometry of the same samples. This combination will result in the detection of small-molecule metabolites extracted from specific leaf zones after the imaging and solid-state NMR experiments have been completed.


This work was supported in part by grant MCB-0613019 (expired) from the National Science Foundation. Purchase of the CCD camera was made possible by a grant from the International Center for Advanced Renewable Energy and Sustainability (I-CARES), Washington University in St. Louis. The authors thank Juan Manfedi (Washington University) for help with the initial set-up of the β imager, and William Margenau (Department of Radiology, Washington University School of Medicine) for technical help with the 11CO2 labeling. R.C.D. gratefully acknowledges salary support from NSF grant DBI-1040498 and DOE grant DE-SC0005157.