Influence of starch deficiency on photosynthetic and post-photosynthetic carbon isotope fractionations

Carbon isotope composition of starch-deficient mutants is primarily explained by photosynthetic 13C fractionations, and thus plant physiological responses to changes in carbon storage, rather than by post-photosynthetic 13C fractionations.


Introduction
Short-term variation in the carbon isotope composition (δ 13 C) in plant-respired CO 2 and in the related respiratory substrates is of wide interest for plant ecophysiological studies investigating carbon allocation (Brueggemann et al., 2011;Hagedorn et al., 2016;Galiano Pérez et al., 2017), for reconstructions of plant functional responses to climatic conditions (Ehleringer et al., 1997;Gessler et al., 2014;Ehlers et al., 2015), and for the plant biochemical community measuring and modeling metabolic carbon fluxes Szecowka et al., 2013;Sweetlove et al., 2014). The enzymatic and diffusional carbon isotope fractionations during photosynthesis are well known and can be mathematically modeled (Farquhar et al., 1989), with the ratio of leaf internal (c i or, more precisely, chloroplastic c c ) to atmospheric (c a ) CO 2 concentrations as a key parameter for the 13 C depletion of organic matter relative to atmospheric CO 2 . The c i /c a ratio is regulated by plant physiological parameters such as the photosynthetic assimilation rate (A n ) and stomatal conductance (g s ) in response to environmental conditions (Brugnoli et al., 1988;Scheidegger et al., 2000). Subsequently, the photosynthetic assimilates, such as leaf (transitory) starch and sugars, and downstream metabolites undergo additional, so-called post-photosynthetic (synonym: post-carboxylation) 13 C fractionation processes during the maintenance of plant metabolism, respiratory processes, carbon allocation, and/ or biosynthesis of various primary or secondary metabolites . However, given that post-photosynthetic 13 C fractionation processes can overlap with each other or with the c i /c a -driven photosynthetic 13 C fractionations (Brandes et al., 2006), their magnitude and influence on short-term δ 13 C variation in leaf dark-respired CO 2 (δ 13 C R ) and respiratory substrates is still under debate Gessler et al., 2014).
One of the most important post-photosynthetic 13 C fractionations is probably related to an equilibrium isotope effect on the plastidic fructose-1,6-bisphosphate aldolase (pFBA) reaction (Brugnoli et al., 1988;Gleixner et al., 1993;Gleixner and Schmidt, 1997). This isotope effect potentially explains the heterogeneous 13 C distribution in carbohydrates that has been observed in glucose derived from C 3 beet sucrose and C 4 corn starch (Rossmann et al., 1991) or from transitory starch in potato and beet leaves (Gleixner et al., 1993;Gilbert et al., 2012). The pFBA reaction reflects a metabolic branching point favoring 13 C in fructose-1,6-bisphosphate under equilibrium conditions (Fig. 1). This 13 C enrichment is passed to transitory starch (or ribulose-1,5-bisphosphate), whereas the 13 C depleted triose phosphates are exported into the cytosol for sugar biosynthesis (e.g. sucrose) and glycolysis. Such a mechanism probably explains why transitory starch has been found to be more 13 C enriched than sugars in beet leaves (Gleixner et al., 1993). The pFBA reaction has been suggested to cause diel δ 13 C variation in sugars and starch (Brugnoli et al., 1988;Gleixner et al., 1993;Goettlicher et al., 2006;Gessler et al., 2007;Lehmann et al., 2015), thereby also influencing δ 13 C R and the apparent respiratory 13 C fractionation (e; Ghashghaie et al., 2001Ghashghaie et al., , 2003Werner et al., 2009). In fact, diel δ 13 C variations in starch and in the organic acid malate have been identified as main drivers of diel changes in δ 13 C R across different soil moisture and temperature conditions in potato plants (Lehmann et al., 2015). Diel δ 13 C variation in leaf starch and sugars related to the pFBA reaction has also been found to partially influence the δ 13 C of phloem-exported sugars and source to sink 13 C fractionations (Goettlicher et al., 2006;Gessler et al., 2008;Barthel et al., 2011), and thus ultimately the δ 13 C of tree-rings (Jäggi et al., 2002;Gessler et al., 2014;Rinne et al., 2015).
Both photosynthetic and post-photosynthetic 13 C fractionations (e.g. via pFBA or respiration) are likely to be closely connected to leaf starch metabolism. Biosynthesis and storage of leaf starch occur in the chloroplast and are under strong control by the circadian clock, leading to significant diel variation in starch concentrations (Zeeman et al., 2007;Stitt and Zeeman, 2012;Kolling et al., 2015). Starch functions as a highly flexible buffer molecule, balancing carbon supply and demand, and was identified as an important factor maintaining plant performance and growth (Huber and Hanson, 1992;Sulpice et al., 2009). The leaf starch content also shows strong seasonal (Jäggi et al., 2002) and species-specific variation (Ivanova et al., 2008), and is additionally influenced by abiotic stresses such as drought (Lehmann et al., 2015;Galiano Pérez et al., 2017;Thalmann and Santelia, 2017). The regulation and pathways of starch biosynthesis are still under debate (Streb et al., 2009;Geigenberger, 2011).
To study starch biosynthesis and degradation, knockout mutants have proven to be invaluable (Stitt and Zeeman, 2012). However, mutant plants have only rarely been combined with stable isotope analysis. Pulses of 13 CO 2 have been applied to mutant plants to study biosynthetic pathways (Figueroa et al., 2016;Baslam et al., 2017), but very few studies have included carbon isotope analysis at natural isotope abundances to investigate δ 13 C differences among starch fractions or explore respiratory 13 C fractionations (Scott et al., 1999;Duranceau et al., 2001). Thus, it is not yet clear whether plant mutants are helpful for elucidating 13 C fractionations and the resulting 13 C isotope signature of organic compounds.
Mutant plants lacking the expression of a functional plastidial phosphoglucomutase (PGM) have negligible starch concentrations. The enzyme catalyzes the reversible interconversion of glucose-6-phosphate to glucose-1-phosphate (EC 2.7.5.1) and is the main route responsible for starch production (Periappuram et al., 2000;Streb et al., 2009). The remaining starch residue in the pgm mutant has been suggested to be produced via cytosolic bypass reactions (Geigenberger, 2011). A number of pgm mutants have been isolated from species such as Arabidopsis thaliana (Caspar et al., 1985), Nicotiana sylvestris (Hanson and McHale, 1988), Pisum sativum (Harrison et al., 2000), and Mesembryanthemum crystallinum (Cushman et al., 2008). The mutants feature a phenotype with a reduced growth rate (Caspar et al., 1985;Huber and Hanson, 1992;Geiger et al., 1995) but increased sugar concentrations during the day, higher leaf to phloem export, and enhanced root respiration (Brauner et al., 2014).
Findings from some studies have indicated a reduction of assimilation rates in pgm mutants compared with wild-type plants (Caspar et al., 1985;Huber and Hanson, 1992;Geiger et al., 1995;Sun et al., 1999). This may cause differences in photosynthetic 13 C fractionations and thus changes in δ 13 C values of plant material, but the magnitude of such an effect has yet to be determined. On the other hand, the starch deficiency in pgm mutants probably influences post-photosynthetic 13 C fractionations via pFBA. The reduced need for hexose-phosphates in the chloroplasts of pgm mutants causes an increased flux, in the form of triose-phosphates, to the cytosol, leading to soluble sugar biosynthesis. We therefore expect that the equilibrium isotope effect on the pFBA reaction is expressed to a lesser extent or not at all in the direction of starch, causing a 13 C enrichment of cytosolic sugars and all downstream metabolites in pgm mutants compared with wild-type plants. Moreover, the PGM-knockout might also affect apparent respiratory 13 C fractionations, given that the absence of starch metabolism may lead to changes in the δ 13 C of potential respiratory substrates (i.e. sugar, starch, and organic acids) and their pool sizes, thereby causing variation in δ 13 C R . The study of pgm mutants may therefore help improve our mechanistic understanding of 13 C fractionations in plants.
Here we performed several experiments with pgm mutants and wild-type plants of different species. We hypothesized that the starch deficiency induced by the PGM-knockout leads to changes in (i) photosynthetic (via leaf gas exchange) and (ii) post-photosynthetic 13 C fractionations (via pFBA and respiration) and thus to δ 13 C variation in organic matter, dark-respired CO 2 , and potential respiratory substrates. To test our hypothesis, we first screened pgm mutants and wild-type plants of four species for average differences in assimilate concentrations and δ 13 C values. We then investigated short-term 13 C fractionation mechanisms in N. sylvestris pgm mutants and wild-type plants by analyzing the δ 13 C OM , δ 13 C R , and δ 13 C of individual compounds (e.g. sugars, starch, malate, and citrate) and their concentrations during a diel cycle. Finally, we measured the leaf gas exchange in both N. sylvestris genotypes to determine if potential δ 13 C differences caused by the PGM-knockout are influenced more by photosynthetic or post-photosynthetic 13 C fractionations.

Experiment 1
We first screened wild-type and pgm genotypes of the four species A. thaliana, M. crystallinum, N. sylvestris, and P. sativum. All four species, including the facultative Crassulacean acid metabolism (CAM) plant M. crystallinum, use the C 3 photosynthetic pathway (δ 13 C values less than -30‰, Fig.  2C). Plants were grown from seeds in a climate-controlled growth chamber. Light was provided during a 12 h photoperiod with a photosynthetic photon flux density (PPFD) of ~160 µmol m -2 s -1 . Relative humidity was constantly at 70% during the day/night cycle, while temperature was 22 °C during the day and 18 °C during the night. Two weeks after germination, plantlets were transplanted into 160 ml pots filled with potting soil. Timing of sampling was adapted to the different growth habits of the four species but occurred during the exponential growth phase of each species, prior to flowering or pot limitation. A.thaliana and M. crystallinum plants were sampled 28 d after germination, while P. sativum and N. sylvestris were sampled after 17 d and 42 d, respectively. At the end of each sampling day, leaf discs (20 mm 2 ) of three individuals were punched and transferred to reaction vials, frozen in liquid nitrogen to inactivate metabolism, and immediately stored at -80 °C.

Experiment 2
N.sylvestris wild-type and pgm mutant plants were grown from seeds under identical growth conditions to those in experiment 1. Two weeks after germination, plantlets were transplanted into 1 liter pots filled with potting soil. After 8 weeks, samples were taken over a 24 h period: fully expanded leaves from individual plants (n=3) were sampled after 0, 4, 8, 12, 14, 20, and 24 h, transferred to paper bags, and frozen in liquid nitrogen. In addition, aliquots of leaf dark-respired CO 2 and climate chamber air were taken at each point in time (see methods below).

Experiment 3
For leaf gas exchange measurements, an additional batch of N. sylvestris wild-type and pgm mutant plants were grown under conditions identical to those described above. Measurements were performed at three points in time on individual pgm mutants and wild-type plants (n=5). Leaf material was sampled on the next day after 4 h of light, as described above.

Isotope ratio analyses of leaf dark-respired CO 2
Leaf dark-respired CO 2 was sampled at the above-listed time points using the in-tube incubation method originally described by Werner et al. (2007) and modified by Lehmann et al. (2015). In short, leaf material was placed in 12 ml gas-tight glass vials ('Exetainer', Labco Ltd, Lampeter, Ceredigion, UK) and flushed with CO 2 -free synthetic air (Pangas, Fig. 1. Reaction scheme for potential post-photosynthetic 13 C fractionations via pFBA in pgm mutant plants. Photosynthetic triose phosphates from the CBB cycle are linked via the pFBA reaction to starch biosynthesis in the chloroplast of wild-type plants. An equilibrium isotope effect (EIE) on the pFBA causes the triose phosphates used for starch biosynthesis to be more 13 C enriched than those exported to the cytosol for sugar (e.g. sucrose) biosynthesis (Gleixner and Schmidt, 1997). In contrast, the main route for starch biosynthesis is blocked in pgm mutants. The surplus of 13 C enriched triose phosphates in the chloroplast might be transported to the cytosol, potentially causing a 13 C enrichment in sugars. Additional bypass reactions for the starch residue in pgm mutants are indicated. CBB cycle, Calvin-Benson-Bassham cycle; pFBA and cFBA, plastidic and cytosolic fructose-1,6-bisphosphate aldolase; PGM, phosphoglucomutase; TPT, triosephosphate translocator; Glc-6-PT, Glc-6-phosphate translocator; DHAP, dihydroxyacetone phosphate; GAP, glyceraldehyde 3-phosphate.
Dagmersellen, Switzerland). The exetainer was immediately darkened for 4 min, and subsequently an aliquot of the sample air (now with respired CO 2 ) was transferred with a gas-tight syringe into a second exetainer pre-filled with dry N 2 (N 2 5.0, Pangas). In addition, climate chamber air samples were taken and transferred to N 2 -filled exetainers. The gaseous samples in the exetainers were then analyzed using a modified Gasbench II (Zeeman et al., 2008) connected to a Delta Plus XP isotope ratio mass spectrometer (Thermo-Fisher, Bremen, Germany) with a precision of ~0.1‰ (SD) for a quality control standard (400 ppm CO 2 in artificial air).

Isotope ratio analysis of organic compounds Extraction and purification of organic compounds
Leaf samples from all experiments were freeze-dried (Beta 2-8 LD plus, Martin Christ, Osterode am Harz, Germany) and subsequently milled to a fine powder with a steel ball-mill (MM 200, Retsch, Haan, Germany) for further chemical analyses. Owing to biomass limitations, some of the leaf material from experiment 1 was pooled for δ 13 C analysis of sugars (Table 1).
For isotope analysis, 100 mg of dry leaf material was transferred into a 2 ml reaction vial and the water-soluble compounds (WSCs) were extracted in 1.5 ml of deionized water at 85 °C for 30 min in a water bath according to Lehmann et al. (2015). After centrifugation (10 000 g, 2 min), the supernatant holding the WSC fraction was transferred into a new reaction vial and stored at -20 °C for further purification, while the pellet was used for isolation of starch (see below). Soluble carbohydrates ('sugars') and the bulk organic acid fractions were isolated from the WSC fraction by ion-exchange chromatography, as described by Lehmann et al. (2015). In brief, Dowex 50WX8 in hydrogen form and Dowex 1X8 in formate form (both 100-200 mesh, Sigma-Aldrich, Buchs, Switzerland) were packed in 5 ml syringes (B. Braun, Melsungen, Germany) arranged in a custom-built rack so that the outlet of the cation exchanger (Dowex 50WX8) syringe was connected to the inlet of the anion exchanger (Dowex 1X8) syringe. After extensive flushing of the Dowex material with deionized water, the WSC fraction (~1 ml) was added to the upper ion exchanger and the neutral sugar fraction was eluted with 30 ml of deionized water. Subsequently, the organic acid fraction was eluted from the anion exchanger with 35 ml of 1 M HCl. All aqueous and acidic samples were frozen at -20 °C and freeze-dried (Hetosicc CD 52-1, Heto, Birkerød, Denmark). The remaining pellet was dissolved in 1 ml of deionized water. All samples were stored at -20 °C until isotope analysis.
Leaf starch was isolated from the remainder of the hot water extraction by enzymatic hydrolysis (Richter et al., 2009;Lehmann et al., 2015). In short, the pellet was washed several times with 1.5 ml of a methanol/chloroform/water (MCW) solution and subsequently with deionized water, and then bench-dried overnight. On the next day, the pellet was additionally oven-dried at 60 °C for 1 h to fully remove chloroform  residues, suspended in deionized water, and boiled at 100 °C for 15 min in a water bath to gelatinize starch. The starch was hydrolyzed with a heatstable α-amylase (EC 3.2.1.1, Sigma-Aldrich) at 85 °C for 2 h and then centrifuged (10 000 g, 2 min). The supernatant was freed from enzymatic residues by using centrifugation filters (Vivaspin 500, Sartorius, Göttingen, Germany) and stored in 2 ml reaction vials at -20 °C until isotope analysis.
δ 13 C values of bulk leaf organic matter, sugars, and starch δ 13 C analysis of bulk organic matter, sugars, and starch was performed using an elemental analyzer (Flash, ThermoFisher) coupled to a Delta Plus XP isotope ratio mass spectrometer (Werner et al., 1999;Brooks et al., 2003). Leaf material was weighed into Sn capsules (5×9 mm, Säntis, Teufen, Switzerland), and aliquots of solubilized sugars and starch were pipetted into the Sn capsules and oven-dried at 60 °C. Positioning of the samples, blanks, and laboratory standards, as well as the referencing of the measurement, was done as suggested by Werner and Brand (2001). Measurement precision of a long-term quality control standard was typically better than 0.2‰ (SD). The applied chemical isolation methods were shown to be free of isotope fractionation by analysis of commercial standard materials with respect to the measurement precision.

δ 13 C values and concentrations of individual organic acids
The analysis of δ 13 C values and concentrations of individual organic acids (citrate and malate) was performed by coupling an Isolink HPLC device to a DeltaV isotope ratio mass spectrometer (all Thermo-Fisher; Lehmann et al., 2015Lehmann et al., , 2016b. Before analysis, the aqueous bulk organic acid fractions have to be passed through a 0.45 µm syringe filter (Infochroma, Zug, Switzerland). Organic acids were separated on a 4.6×300 mm Allure Organic Acids column (Restek, Bellefonte, PA, USA) at 8 °C using 100 mM KH 2 PO 4 (pH 3) as a mobile phase with a flow rate of 500 µl min -1 . All organic compounds were oxidized to CO 2 at 99 °C using Na 2 S 2 O 8 under acidic conditions. The CO 2 was subsequently separated from the mobile phase and measured for its isotope ratio and concentration in the IRMS. Every 10 samples, a set of malate and citrate laboratory standards of different concentrations (10-180 ng C μl -1 ) was analyzed. Offset corrections to EA-IRMS δ 13 C values and determination of concentrations were performed according to Rinne et al. (2012). The applied method was shown to be free of isotope fractionation by analysis of standard materials with respect to the measurement precision, which was typically better than 0.4‰ (SD).

Sugar and starch concentration measurements
Soluble sugars were directly extracted from leaf discs (experiment 1) or 10 mg of leaf powder (experiment 2) by two sequential additions of 80% ethanol and one addition of 50% ethanol, where the supernatant was recovered and pooled. During each extraction, the sample was heated to 80 °C for 30 min. Sugars were photometrically measured according to Mueller-Roeber et al. (1992). Starch in the remaining pellet was extracted and photometrically measured as described in Hostettler et al. (2011). Sugar and starch concentrations are presented in μmol glucose equivalents per mg dry mass calculated using a pre-determined conversion factor.

Leaf gas exchange measurements
A cuvette with a measurement area of 6 cm 2 coupled to a Licor 6400 was used for leaf gas exchange measurements (all supplied by LI-COR, Lincoln, NE, USA). Temperature and relative humidity conditions in the cuvette were 21.9±0.5 °C and 63.9±7.3%, respectively, throughout all diurnal measurements. A flow rate of 250 µmol s -1 , CO 2 concentration of 400 µmol mol -1 CO 2 , PPFD of 160 µmol m -2 s -1 , and leaf temperature of 22.0 °C were constantly maintained. Climatic and light conditions in the cuvette were thus nearly identical to growth conditions.

Data analysis and calculations
Leaf gas exchange data were used to model δ 13 C values of recent assimilates (δ 13 C M ). The calculation was based on standard 13 C discrimination models (Farquhar et al., 1982): where δ 13 C Air reflects the diel average of climate chamber air (-13.7±1.2‰, mean ±1 SD), a stands for the diffusional isotope fractionation (4.4‰), b for the enzymatic isotope fractionation (27‰, mainly due to Rubisco-catalyzed carboxylation reactions), and c i /c a is the ratio of leaf internal and atmospheric CO 2 concentrations (c a was held constant at 400 µmol mol -1 ). The diel apparent respiratory 13 C fractionation was calculated as: where δ 13 C Substrate is the average δ 13 C value of a respiratory substrate, and δ 13 C R is the average δ 13 C value of leaf dark-respired CO 2 during a 24 h diel cycle. Linear mixed effects models were used to test the individual effects of species and genotype (experiment 1) or genotype and time (experiment 2) on gas exchange, δ 13 C values, and concentrations, as well as interactions between these effects. If no significant interaction between individual effects was found, the model was repeated without the interaction term. The individual plant ID was included as a random effect. Effects with P-values <0.05 were considered to be statistically significant. Where not indicated otherwise, all given errors denote standard errors. All statistical analyses were performed in R version 3.4.4. (R Core Team, 2018).

Experiment 1: screening of pgm mutants and wild-type plants
Functioning of the PGM-knockout in all screened species was demonstrated by clearly lower starch concentrations in pgm mutants than in wild-type plants ( Fig. 2A). Sugar concentrations were generally higher in pgm mutants than in wild-type plants, with the greatest difference in N. sylvestris plants, while only M. crystallinum plants showed little difference in sugar concentrations (Fig. 2B). Isotope analysis of leaf bulk organic matter showed a clear δ 13 C difference of 1.6‰ between pgm and the wild type for N. sylvestris plants (Fig. 2C) but not for other species. A similar tendency was found for δ 13 C values Modeled δ 13 C values of assimilates (δ 13 C M , ‰, Equation 1) derived from diel average c i values (µmol mol -1 ) and observed δ 13 C values of organic matter (δ 13 C OM ) are shown (all from experiment 3). The diel averages for δ 13 C OM , δ 13 C of dark-respired CO 2 (δ 13 C R ), and δ 13 C of different substrates are also given (all from experiment 2). For all parameters, the genotype difference (pgm−WT) is indicated. Mean values ±1 SD are shown.
of sugars (Fig. 2D), with the largest differences being up to 2.0‰ for sugars of A. thaliana and M. crystallinum. Whether the results were statistically significant for each species could not be determined because all three replicates were pooled for the isotope analysis of sugars. In summary, the effect of the PGMknockout was greatest for δ 13 C OM and δ 13 C values of sugars in N. sylvestris plants, which also showed the largest difference in sugar and starch concentrations between the two genotypes.

Experiment 2: diel cycle of N. sylvestris pgm mutants versus wild-type plants
Given the above-mentioned results, we further investigated the mechanisms of 13 C fractionation in N. sylvestris pgm mutants and wild-type plants by measuring δ 13 C values of leaf bulk organic matter (δ 13 C OM ), assimilates, organic acids, and darkrespired CO 2 (δ 13 C R ) during a diel cycle ( Fig. 3; Table 1). In general, we observed clear effects of time and genotype for δ 13 C OM and δ 13 C R , and for δ 13 C values of all substrates (P<0.001 for both time and genotype). δ 13 C OM values, ranging from -32.2 to -33.8 for both genotypes and all points in time, were on a diel average 0.8‰ more negative in pgm mutants than in wild-type plants. δ 13 C values of sugars and starch, ranging from -31.3‰ to -35.2‰ for both genotypes and all points in time, were on a diel average 2.1‰ and 1.3‰ more negative in pgm mutants than in wild-type plants. On a diel average, starch was 0.4‰ and 1.2‰ 13 C enriched compared with sugars for wild-type plants and pgm mutants, respectively. In comparison with the δ 13 C values of plant assimilates, the δ 13 C R and δ 13 C values of malate and citrate were less negative (except for some δ 13 C values at 0, 20, and 24 h) and showed wider variation during the diel cycle for both genotypes ( Fig. 3; Table  1). δ 13 C R values ranging from -25.7‰ to -33.6‰ increased by 4.3‰ during the day and decreased by 5.9‰ during the night for both genotypes. δ 13 C R values were on a diel average ~1.4‰ more negative in pgm mutants than in wild-type plants, with the exception that after 8 h of illumination δ 13 C R values were 3.6‰ more negative. δ 13 C values of malate and citrate, ranging from -22.5‰ to -31.9‰ for both genotypes and all points in time, were on a diel average 1.9‰ and 2.6‰ more negative for pgm mutants than for wild-type plants. δ 13 C values of malate showed an increase of 1‰ for pgm mutants and 2‰ for wild-type plants during the day, while a similar decrease of 4.1‰ during the night was observed for both genotypes. Also, δ 13 C values of citrate for pgm mutants showed an increase of 1.9‰ during the day and a decrease of 5.9‰ during the night, while the values for wild-type plants steadily decreased by 4.1‰ during the diel cycle. Thus, δ 13 C R and δ 13 C values of all substrates were clearly more negative for pgm mutants than for wild-type plants.
We also measured assimilate and organic acid concentrations during a diel cycle in both pgm mutants and wild-type N. sylvestris plants (Fig. 4). As expected, the pgm mutant showed only very low diel starch concentrations, while wild-type plants showed a clear diel cycle, with the lowest and highest concentrations occurring at the beginning and end of the day (P<0.001 for the interaction between time and genotype). In contrast, sugar concentrations showed a clear diel cycle in pgm mutants, with values up to 76% higher during the day and up to 53% lower during the night compared with values in wild-type plants, which showed no clear diel cycle (P<0.05 for the interaction between time and genotype). Moreover, malate and citrate concentrations were both on average 28.5% lower in pgm mutants than in wild-type plants during the diel cycle (P<0.01). Malate concentrations were highest at the end of the day (nearly twice higher at 8 h in wild-type plants compared with concentrations in pgm mutants) and lowest at the end of the night in both genotypes (P<0.001), while citrate concentrations showed no distinct diel cycle in either genotype (P>0.05). Thus, both assimilate and organic acid metabolism were strongly influenced by the PGM-knockout.

Experiment 3: leaf gas exchange measurements and modeling of δ 13 C M
The PGM-knockout affected the assimilation rate (A n ; P<0.001) and stomatal conductance to water vapor (g s ; P<0.05), causing up to 2.0 μmol m -2 s -1 and 14.5 mmol m -2 s -1 lower values, respectively, in pgm mutants than in wild-type plants of N. sylvestris (Fig. 5). Although the genotype differences in A n and g s tended to increase during the diurnal cycle, no clear significant temporal variations during the light period were observed (P>0.05). The changes in A n and g s caused up to 54.0 µmol mol -1 higher intercellular CO 2 concentrations (c i ) in pgm mutants than in wild-type plants (P<0.001), but again without clear temporal variation (P>0.05).
The c i values derived from leaf gas exchange measurements were used to model the δ 13 C of assimilates (δ 13 C M ; Table 1, Equation 1). δ 13 C M values of both genotypes were on a diel average consistent with the measured δ 13 C OM values of wildtype plants from the same experiment. δ 13 C OM differences between experiment 2 and 3 were probably caused by temporal δ 13 C differences in the climate chamber air. In particular, fossil fuel CO 2 emissions in winter have most probably caused lower δ 13 C Air and thus lower δ 13 C OM for plants of experiment 2 than for those of experiment 3, which was conducted in summer. The genotype difference in δ 13 C M of 2.3‰ was similar to that for sugars (2.1‰), malate (1.9‰), and citrate (2.6‰), but tended to be greater than the difference observed for δ 13 C OM (0.8-1.1‰) and δ 13 C R (1.4‰). Thus, the genotype difference in both modeled and observed δ 13 C values showed a consistent 13 C depletion of pgm mutants compared with wildtype plants across all experiments.

Apparent respiratory 13 C fractionation and correlations with δ 13 C of substrates
We calculated the diel average apparent respiratory 13 C fractionation using Equation 2. The e values differed among the potential respiratory substrates for both genotypes (Table 2), with the highest values observed for malate and citrate (up to 4.5‰) and the lowest values observed for sugars, starch, and bulk organic matter (up to -3.9‰). No clear significant differences were observed between the genotypes for any substrate (P>0.05, t-test). In addition, we investigated the correlation between individual δ 13 C R values and the δ 13 C values of different substrates for the two genotypes and all points in time (Fig.  6). We found a strong correlation for malate (R=0.81, P<0.001) and citrate (R=0.63, P<0.001) but a weaker correlation for sugars (R=0.31, P<0.05) and no correlation for starch (R=0.09, P>0.05). An analysis of covariance (ANCOVA) showed that the relationship between δ 13 C R and each δ 13 C Substrate was not influenced by the PGM-knockout (P>0.05). In addition, genotype differences across the four species in the δ 13 C of sugars were found to be negatively related to a genotype difference in sugar concentrations (r 2 =0.61, P=0.118) and positively related to those in starch concentrations (r 2 =0.86, P=0.024; Fig. 7).

Discussion
Our screening (experiment 1) confirmed that pgm mutants had lower starch concentrations but higher sugar concentrations at the end of the day compared with wild-type plants for most species ( Fig. 2A, B), clearly demonstrating that the primary assimilate metabolism was affected by the PGM-knockout across all species. As shown for N. sylvestris plants (experiment 2), the sugar pool in pgm mutants was used up during the night, similar to the starch pool in wild-type plants for various biosynthetic and respiratory processes (Fig. 4). This demonstrates that the PGM-knockout leads to a re-routing of freshly assimilated triose phosphates towards cytosolic sugar biosynthesis, as plastidic starch biosynthesis is prevented in mutant plants (Streb and Zeeman, 2012). Thus, the lack of starch as an important temporary carbon sink and chemical energy buffer in pgm mutants is at least partially counterbalanced by higher sugar biosynthesis. The pgm mutant therefore has great potential as a tool to study 13 C fractionation responses related to changes in the diel starch metabolism.
On the post-photosynthetic 13 C fractionation via pFBA in response to pgm-induced starch deficiency We hypothesized that post-photosynthetic 13 C fractionation via pFBA plays an important role in the δ 13 C variation in plant material. We expected that, owing to the blocked starch biosynthesis in pgm mutants, the equilibrium isotope effect on the pFBA reaction would be only slightly or not at all expressed (Fig. 1). Given that more triose phosphates are exported to the cytosol under such conditions, more 13 C enriched sugars and, by extension, more 13 C enriched organic matter should be produced in pgm mutants than in wild-type plants. However, carbon isotope analysis of leaf organic matter and sugars of the four species revealed, in contrast to our hypothesis, either no clear δ 13 C differences or a 13 C depletion in pgm mutants compared with wild-type plants of N. sylvestris (Fig. 2C, D). This shows that the pFBA-related post-photosynthetic 13 C fractionation is probably not the main reason for the observed δ 13 C differences between the genotypes.
Theoretically, only one out of six triose phosphate molecules after CO 2 fixation is used for sugar/starch synthesis, while the other five molecules enter the regeneration part of the Calvin-Benson-Bassham cycle. The PGM-knockout might therefore affect the kinetic operation conditions of the pFBA reaction much less than expected. For example, the 13 C enrichment of starch compared with sugars was 0.8‰ greater on a diel average in pgm mutants than in wild-type plants (Fig. 3). This is a surprising result given that the main route of starch biosynthesis is blocked by the PGM-knockout in the mutant plant. The 13 C enrichment of the starch residue compared with that of sugars in pgm mutants must therefore be caused by a different biosynthetic pathway such as a cytosolic bypass reaction. Higher glucose-6-phosphate concentrations in chloroplasts of pgm mutants than in those of wild-type plants have been observed (Kofler et al., 2000). Assuming that pFBA is operating normally in pgm mutants, glucose-6-phosphate molecules could still show the typical 13 C enrichment induced by pFBA. Transport of glucose-6-phosphate to the cytosol via the glucose-6-phosphate translocator in connection with the glucose-1-phosphate transport from the cytosol to the chloroplast (Fettke et al., 2011) would mean that the plastidic PGM could be bypassed by the cytosolic isoenzyme. Such a mechanism might explain why starch is more 13 C enriched compared with sugars in the pgm mutant compared with wild-type plants.
The low temporal variation in the δ 13 C of starch and sugars is in line with observations that there is no clear circadian rhythm in δ 13 C of leaf bulk sugars and starch of other plant species (Sun et al., 2009;Lehmann et al., 2016b). However, this result contradicts previous findings that starch-related postphotosynthetic 13 C fractionations influence (e.g. via pFBA) short-term δ 13 C variation in leaf and phloem assimilates . A n , net assimilation rate (µmol m -2 s -1 ), g s , stomatal conductance (mmol m -2 s -1 ), and C i , intercellular CO 2 concentration (µmol mol -1 ). Mean values ±1 SE are given (n=5). (Gessler et al., 2008;Kodama et al., 2011;Lehmann et al., 2015). These opposing observations stress that species-specific differences need to be considered when modeling temporal δ 13 C variation in assimilates, including the extent of isotope effects on enzymatic reactions, compartmentalization of sugar/starch pools, and their pool sizes and turnover rates. In summary, potential post-photosynthetic 13 C fractionation via pFBA cannot explain the observed δ 13 C difference between pgm mutants and wild-type plants.

pgm-induced starch deficiency affects daytime organic acid metabolism but not apparent respiratory 13 C fractionations
The organic acid metabolism (i.e. δ 13 C values and concentrations of malate and citrate) and δ 13 C R were clearly affected by the PGM-knockout-induced starch deficiency (Figs 3, 4). Both organic acids were generally 13 C enriched compared with starch and sugars of both genotypes. This pattern can be  explained by the activity of the phosphoenolpyruvate carboxylase (PEPC) reaction that catalyzes the conversion of phosphoenolpyruvate and hydrogen carbonate to oxaloacetate with a net isotope discrimination against 13 C of -5.7‰ relative to CO 2 in equilibrium with hydrogen carbonate (Farquhar et al., 1989). The 13 C enriched oxaloacetate functions as a precursor for malate and citrate. The PEPC reaction is therefore assumed to reflect an anaplerotic flux in C 3 plants that replenishes withdrawn carbon skeletons from the tricarboxylic acid (TCA) cycle Lehmann et al., 2015). However, it is generally accepted that the TCA cycle is not fully functional during the day due to light inhibition of key enzymes (Hanning and Heldt, 1993;Tcherkez et al., 2005;Sweetlove et al., 2010). Given the light-induced limitation of the TCA cycle, the light-activated PEPC reaction (together with the non-inhibited malate dehydrogenase reaction) must be responsible for the often-observed accumulation of malate (via oxaloacetate) during the day (Scheible et al., 2000;Gessler et al., 2009;Igamberdiev and Bykova, 2018). In fact, we observed a simultaneous increase in δ 13 C values and concentration of malate for both N. sylvestris genotypes, although these increases were lower in pgm mutants than in wild-type plants (Figs 3, 4). This strongly suggests that the anaplerotic PEPC flux is downregulated in pgm mutants during the day, potentially owing to the increase in sugar concentration which supplies glycolysis and the TCA cycle with additional carbon skeletons (Figs 2, 4). Moreover, we observed that diel δ 13 C R values were strongly related to δ 13 C values of malate and citrate for both genotypes but were weakly or not at all correlated with δ 13 C values of sugars and starch, respectively (Fig. 6). This result is in line with findings from previous studies that malate is a key substrate for leaf dark-respired CO 2 (Gessler et al., 2009;Lehmann et al., 2015Lehmann et al., , 2016b, particularly shortly after darkening, as shown in an experiment with position-specific 13 C-labeled malate (Lehmann et al., 2016a). However, despite the lower daytime δ 13 C values and concentrations of malate in N. sylvestris pgm compared with wild-type plants (Figs 3, 4), we generally observed no genotype differences in the average diel apparent respiratory 13 C fractionation for various potential substrates (Table 2) or in the δ 13 C relationships between dark-respired CO 2 and substrates (Fig. 6). Thus, starch deficiency induced by the PGM-knockout had no clear influence on apparent respiratory 13 C fractionations.

pgm-induced starch deficiency causes photosynthetic 13 C fractionations
Interestingly, the differences in sugar and starch concentrations between genotypes were related to the δ 13 C difference in sugars across all species (Fig. 7), with the largest δ 13 C difference corresponding to the largest differences in sugar and starch concentrations between N. sylvestris genotypes. This suggests that the changes in the assimilate pool cause 13 C fractionations. It has been demonstrated in several studies that an increase in sugar concentrations (as observed in pgm mutants) decreases the photosynthetic activity of a plant (Krapp and Stitt, 1995;Paul and Driscoll, 1997;Burkle et al., 1998). This hypothesis is supported by our gas exchange measurements, which showed lower A n and g s values and higher c i values in pgm mutants than in wild-type N. sylvestris plants during the day (Fig. 5). In particular, the lower assimilation rates in pgm mutants are widely supported by findings in A. thaliana and N. sylvestris plants (Caspar et al., 1985;Huber and Hanson, 1992;Geiger et al., 1995;Sun et al., 1999). Thus, the changes in starch and sugar pool sizes in response to the PGM-knockout (Fig. 7) have influenced leaf gas exchange and thus caused differences in photosynthetic 13 C fractionations, explaining the observed 13 C depletion in the organic matter, sugars, organic acids, and dark-respired CO 2 in pgm mutants (Figs 2, 3).
To determine whether photosynthetic 13 C fractionations are actually the main driver of the δ 13 C differences observed between the N. sylvestris genotypes, we modeled δ 13 C values of assimilates from leaf gas exchange measurements and found that δ 13 C M and δ 13 C OM values from the same experiment were in good agreement (Table 1). Most importantly, the δ 13 C M difference between the two genotypes of 2.3‰ was similar to the differences observed for sugars and organic acids. This again indicates that δ 13 C differences in fresh assimilates caused by the pgm-induced starch deficiency are primarily driven by photosynthetic 13 C fractionations. Further, the smaller genotype difference for δ 13 C OM of 0.8-1.1‰ compared with the difference for δ 13 C M might be explained by structural components in organic matter. For example, cellulose reflects and integrates all leaf gas exchange variation that occurs during the period of growth or leaf expansion. Given that pgm mutants grow differently from wild-type plants (Huber and Hanson, 1992), the leaf gas exchange and thus the photosynthetic 13 C fractionations that have shaped the δ 13 C of cellulose and organic matter might differ from those that have shaped the δ 13 C of assimilates measured at the end of the growth period. In addition, potential isotope fractionations related to sugar and starch biosynthesis may cancel each other out during the diel buildup of organic matter, reducing the genotype differences in δ 13 C OM compared with δ 13 C M . Overall, we conclude that the majority of δ 13 C genotype differences can be primarily explained by leaf gas exchange adaptations and thus by c i /c a -driven photosynthetic 13 C fractionations, while potential post-photosynthetic 13 C fractionations in response to the PGM-knockout are of minor importance.

Conclusions and implications
Through this investigation, we showed that the δ 13 C values of organic matter, substrates, and dark-respired CO 2 are strongly influenced by differences in leaf gas exchange and corresponding photosynthetic 13 C fractionations. Changes in assimilate pool sizes due to the PGM-knockout may have induced the photosynthetic 13 C fractionations by suppressing the photosynthetic activity of the mutant plants. δ 13 C variability in plant material might therefore be indicative of increases in sugar concentrations or changes in the sugar/starch ratio. It is likely that the observed shift in δ 13 C due to starch deficiency might also occur in non-genetically modified plants in response to changes in environmental conditions (e.g. drought) or nutrient availability; this topic should be investigated in future studies.
Moreover, we demonstrated that post-photosynthetic 13 C fractionation (e.g. via pFBA or respiration) is not or only slightly affected by pgm-induced starch deficiency. Instead the isotopic signature of both plant mutants and wild-type plants is dominated by the photosynthetic 13 C fractionation processes. This might have important implications for the reconstruction of leaf gas exchange responses to climatic conditions of the past using δ 13 C values of plant compounds or other biomarkers (Ehleringer et al., 1997;Gessler et al., 2014).
Finally, we conclude that δ 13 C changes in plant material may not only contain information on physiological and biochemical processes but might also be helpful for inferring and reconstructing genetic responses. Our findings might therefore be interesting for retrospective studies on tree decline and mortality using tree-ring growth patterns in combination with stable isotope analysis of carbon, oxygen, and hydrogen (Scheidegger et al., 2000;Cormier et al., 2018;Gessler et al., 2018) and with genome analysis (Heer et al., 2018). Our mutant approach thus paves the way for future studies exploring the biochemical and genetic background of isotope fractionations in plants.