Isotope ratio-based quantification of carbon assimilation highlights the role of plastidial isoprenoid precursor availability in photosynthesis

We report a method to estimate carbon assimilation based on isotope ratio-mass spectrometry (IRMS) of 13CO2 labeled plant tissue. Photosynthetic carbon assimilation is the principal experimental observable which integrates important aspects of primary plant metabolism. It is traditionally measured through gas exchange. Despite its centrality in plant research, gas exchange performs poorly with rosette growth habits typical of Arabidopsis thaliana, mutant lines with limited biomass, and accounts poorly for leaf shading. IRMS-based carbon assimilation values from plants labeled at different light intensities were compared to those obtained by gas exchange, and the two methods yielded similar values. Using this method, we observed a strong correlation between 13C content and labeling time (R2 = 0.999) for 158 wild-type plants labeled for 6 to 42 min. Plants cultivated under different light regimes showed a linear response with respect to carbon assimilation, varying from 7.38 nmol 13C mg−1 leaf tissue min−1 at 80 PAR to 19.27 nmol 13C mg−1 leaf tissue min−1 at 500 PAR. We applied this method to examine the link between inhibition of the 2C-methyl-d-erythritol-4-phosphate (MEP) pathway and suppression of photosynthesis. A significant decrease in carbon assimilation was observed when metabolic activity in the MEP pathway was compromised by mutation or herbicides targeting the MEP pathway. Mutants affected in MEP pathway genes 1-DEOXY-d-XYLULOSE 5-PHOSPHATE SYNTHASE (DXS) or 1-HYDROXY-2-METHYL-2-(E)-BUTENYL 4-DIPHOSPHATE SYNTHASE (HDS) showed assimilation rates 36% and 61% lower than wild type. Similarly, wild type plants treated with the MEP pathway inhibitors clomazone or fosmidomycin showed reductions of 52% and 43%, respectively, while inhibition of the analogous mevalonic acid pathway, which supplies the same isoprenoid intermediates in the cytosol, did not, suggesting inhibition of photosynthesis was specific to disruption of the MEP pathway. This method provides an alternative to gas exchange that offers several advantages: resilience to differences in leaf overlap, measurements based on tissue mass rather than leaf surface area, and compatibility with mutant Arabidopsis lines which are not amenable to gas exchange measurements due to low biomass and limited leaf surface area. It is suitable for screening large numbers of replicates simultaneously as well as post-hoc analysis of previously labeled plant tissue and is complementary to downstream detection of isotopic label in targeted metabolite pools.


Introduction
Photosynthetic carbon assimilation integrates many aspects of plant metabolism and environmental response into a single response variable. Carbon assimilation is typically determined through real-time measurements of gas exchange, which provide information on stomatal conductance, transpiration, intercellular CO 2 concentration (Ci), and net carbon assimilation (A) via water vapor and CO 2 gas analyzers [51]. These sensors are often combined with additional detector systems, such as online mass spectrometers, to provide information on the relative contributions of photorespiratory and non-photorespiratory sources of CO 2 loss, which in turn inform efforts at improving crop productivity [15,39,52,53]. Gas exchange measurements are the primary input into biochemical models of photosynthesis, which have yielded significant insights into the coupling of photosynthetic electron flow to chemical energy production through the carbon reduction cycle [4,26,87]. The accurate estimation of carbon assimilation is therefore central to understanding the photosynthetic response of plants to changes in CO 2 concentration (i.e. A/C i response curves) [75], water deficit [48], and salinity [27]. Accurate measurement of carbon assimilation is also a prerequisite for investigating different metabolic modes of the Calvin-Benson cycle, such as Rubisco-limited photosynthesis versus RuBP-regeneration-limited photosynthesis [75].
Measurements of carbon assimilation have been instrumental in resolving key aspects of central metabolism such as photorespiration [78][79][80][81]. Such measurements have also helped resolve the impact of biotic and abiotic stress on photosynthesis including heat [36,71,74], cold [2,5,10,65], drought [48], and light stress [21]. Measurements of assimilation can also help reveal coordinated signaling transduction networks. Herbivory stress, for instance, generally inhibits photosynthesis rapidly and substantially, even in excess of what is predicted through loss of photosynthetically active leaf surface area [82,86] (reviewed in [58]). The metabolic down regulation of photosynthesis in response to herbivory is thought to be accompanied by a shift to defensive metabolism, a process mediated by jasmonate and phytochrome B signaling [35]. Indeed, these growth-defense trade-offs can be uncoupled by relieving transcriptional suppression imposed by this regulatory pathway [17]. These findings highlight the importance of leaf carbon assimilation as a tool to understand fundamental questions of plant metabolism.
Despite the broad utility of gas exchange in measuring net carbon assimilation, limitations include low throughput and the difficulty of measuring plants with small leaves. Carbon assimilation is usually measured in tandem with water vapor fluxes in commercially available gas exchange systems that measure gas concentrations using infra-red gas analyzers (IRGAs) [14,88]. These systems are well suited to measuring carbon assimilation on single leaves but require individual leaves (or plants) to be enclosed within a measurement cuvette or leaf clamp for extended periods of time to gather accurate measurements. Arrayed systems constructed to measure net carbon assimilation in multiple plants have been reported but are limited in throughput to the number of chambers and gas switching channels [30]. Their increased complexity also hampers more widespread adoption among researchers. While higher-throughput tools probing the light energy use efficiency of photosynthesis using chlorophyll fluorescence are available, direct measurements of gas exchange still offer the most direct quantification of carbon assimilation [6,22,72]. The limitations noted above restrict the high throughput use of assimilation measurements on a large number of plants with small leaf areas, which would be required to screen for or characterize mutants of model plants like Arabidopsis thaliana with impaired growth phenotypes. Such phenotypes are expected to occur when investigating mutations that disrupt metabolic networks that interact with photosynthesis.
One metabolic domain closely linked to photosynthesis is that of chloroplastidic terpenoid (or isoprenoid) metabolism. All known terpenoids are synthesized from the universal precursors isopentenyl and dimethylallyl diphosphate (IDP and DMADP). IDP and DMADP are produced by two independent, compartmentally separated pathways in plants cells: the cytosolic mevalonic acid (MVA) pathway and the plastid localized 2C-methyld-erythritol-4-phosphate (MEP) pathway [63]. Efforts to dissect the regulatory mechanisms controlling these two pathways have relied heavily on specific inhibitors targeting enzymes of each pathway. Mevastatin (MEV) targets the rate determining step of the MVA pathway, 3-hydroxy-3-methyl-glutaryl-coenzyme A reductase [43], while clomazone (CLZ) and fosmidomycin (FSM) selectively block 1-deoxy-d-xylulose 5-phosphate synthase (DXS) [54] and 1-deoxy-d-xylulose 5-phosphate reductoisomerase (DXR) [46], respectively, the first and second enzymes of the MEP pathway (Fig. 1). Norflurazon (NFZ) has been widely utilized as a specific inhibitor of the downstream enzyme phytoene desaturase [11] due to its ability to isolate regulation of the MEP pathway from that of carotenoid biosynthesis, a principal sink for MEP pathway derived IDP and DMADP in chloroplasts.
While the MVA pathway yields IDP and DMADP primarily for the production of sesquiterpenoids, phytosterols, brassinosteroids, polyprenols, and ubiquinone [84], most if not all terpenoid-derived co-factors required for photosynthesis utilize IDP and DMADP provided by the MEP pathway. For instance, the MEP pathway contributes building blocks for the biosynthesis of photosynthetic pigments including carotenoids and chlorophyll [70] and the electron transport co-factors phylloquinone [89] and plastoquinone [49] (Fig. 1). However, the link between IDP and DMADP availability, flux through the MEP pathway, and photosynthetic carbon assimilation has not been well established. Part of this deficiency in understanding relates to the limited information regarding the proportion of total fixed carbon dedicated to isoprenoid metabolism for the synthesis of photosynthetic pigments in the chloroplast. Metabolic flux through the MEP pathway has recently been reported in model [90] and non-model plant systems [32,61]. Yet such studies have so far not calculated the measured flux through this pathway as a function of the total assimilated carbon budget. Such information would be highly valuable for clarifying the role of isoprenoid precursor availability in supporting photosynthesis as well as the impact of photosynthesis on flux through the MEP pathway.
Here we describe a method for estimating carbon assimilation in plants that does not require real time gas exchange measurements, is flexible enough to be used on small plants, and could be implemented on large numbers of plants simultaneously. It is based on elemental analysis-isotope ratio mass spectrometry (EA-IRMS) analysis of 13 C content in isotopically labeled plant tissue. IRMS is a technique involving a magnetic sector mass spectrometer that measures the 13 C/ 12 C ratio and total C content in a gas or tissue sample with high precision. It may either be fed by an EA combustion oven, which combusts plant tissue to CO 2 [57] or by the gas flow from a plant cuvette exhaust, which supplies CO 2 directly to an inline mass spectrometer [15]. Its high sensitivity has made it instrumental in identifying adulterants in agricultural products such as honey [16] as well as detecting performance enhancing drugs in sports [7]. When used in conjunction with gas chromatography (GC), it can provide compound specific isotope analysis [55]. It has been used in the plant sciences to investigate the biosynthetic origins of metabolites such as leaf waxes [67].
Herein we describe a new application of IRMS which takes advantage of the rise in popularity of whole plant 13 CO 2 labeling. We demonstrate its similarity to the results obtained from gas exchange measurements and provide proof of principle using Arabidopsis mutants and herbicide treatments which target the MEP pathway. Given the role of the MEP pathway in supplying the biosynthesis of photosynthetic pigments and redox co-factors ( Fig. 1), we initially postulated that long term deficiencies in the availability of IDP and DMADP might result in reduced photosynthetic efficiency. Using this method, we have instead determined that the dependence of photosynthetic carbon assimilation on flux through the MEP pathway is much more immediate, highlighting the rapid turnover of terpenoidderived photosynthetic co-factors.

Plant lines and cultivation
Arabidopsis thaliana lines designated 'wild type' were ecotype Columbia 0 seeds. Mutant lines hds3, dxs3, and prl1 have been described previously [28,63] , or NFZ (10 or 50 μM) in 0.5% methanol in water, or 0.5% methanol in water only (control) and returned to the growth chamber for 60 min. They were then acclimatized in the flow cuvette for 30 min and labeled for 15 min in an atmosphere containing 400 μL L −1 13 CO 2 prior to flash freezing, as described below under "Whole plant labeling experiments". DMADP labeling in lyophilized, ground leaf tissue was analyzed by gas chromatography-mass spectrometry following phosphoric acid conversion to isoprene as described previously [12]. Methane positive chemical ionization and exact label incorporation calculations were performed as described in [90]. Carbon assimilation trends for herbicide treatments and mutant lines were calculated by linear regression of the data points. Statistical significance was determined by Student's t-tests. P values and confidence intervals were calculated in Microsoft Excel (version 2016) using the Analysis ToolPak.

Whole plant labeling experiments
Whole plant, short term 13 CO 2 labeling assays were based on previously reported protocols [34,50] and as described in the schematic workflow (Fig. 2). To establish the linearity of the technique, 14-19 plants were analyzed per time point from 6 to 42 min (Fig. 3a) for a total of 158 individual whole plant labeling experiments. For mutant, herbicide treatment, and light intensity time courses, between 5 and 11 plants were used in each group, as indicated in the corresponding figure legends. Environmental variables (temperature, light intensity, humidity) for labeling in the dynamic flow cuvette were adjusted to be identical to a given plant's cultivation conditions in growth chambers, based on readings obtained with the same Li-Cor 250A light sensor used to measure light intensity in growth chambers and a TC01  13 C label by EA-IRMS. Solid arrows signify the order of steps. Hollow arrows represent gas flow. Both air sources (normal and labeled air) were supplied by compressed air tanks containing 400 μL L −1 CO 2 or 13 CO 2 (99% enrichment) in a mixture of nitrogen:oxygen (80:20). Air was (de)humidified by passing through a chilled wash bottle containing water. CO 2 and H 2 O vapor were quantified before entering the cuvette (reference) and in the cuvette exhaust (sample). After processing labeled plant tissue, 2 mg aliquots were analyzed by EA-IRMS, which consisted of combustion of the sample to carbon dioxide, separation in a magnetic sector mass analyzer, and data acquisition USB thermocouple (National Instruments), which was installed in the flow cuvette and placed in continuous contact with the abaxial leaf surface (Additional file 1: Figure S1). Prior to initiating labeling, all plants were acclimated in a normal atmosphere for a minimum of 30 min at a flow rate of 1.0 L min −1 until photosynthesis had stabilized, as judged by gas exchange measurements performed with either a Licor 840a CO 2 /H 2 O analyzer or Licor 6400 photosynthesis measurement system as described previously [90]. Cuvette temperature was maintained at 21 °C and confirmed with thermocouple readings. CO 2 concentration was maintained at 400 μL L −1 based on IRGA sensor readings during acclimation. Once gas exchange measurements in the normal atmosphere had stabilized (< 1% variation over 5 min), labeling was initiated by a single step change to an atmosphere identical to the previous one but with 400 μL L −1 13 CO 2 (99% enrichment; Linde Gas). Atmospheric switching was controlled manually with a 3-way switch valve just upstream of the plant cuvette. Harvesting and metabolic quenching were accomplished through freezing in liquid nitrogen, and the uncorrected labeling time was recorded as the time from admitting the labeling atmosphere to the cuvette until plant freezing. Upon switching to the labeling atmosphere containing 400 μL L −1 13 CO 2 , the atmospheric half-life was determined from the decay of the 12 CO 2 signal by calculating the time to reach the midpoint between the 12 CO 2 signal prior to initiating labeling and the minimum signal observed in the 13 CO 2 containing atmosphere (the IRGA, while tuned to Assimilation of 13 C during time course labeling assays of wild-type Arabidopsis as determined by elemental analysis-isotope ratio mass spectrometry (EA-IRMS). a plants were equilibrated in a dynamic flow cuvette under standard conditions (see "Methods and materials" for details) in a natural atmosphere until a photosynthetic steady state was attained, then labeled with 400 μL L −1 13 CO 2 before flash freezing in liquid nitrogen. Aliquots of ground, lyophilized tissue were analyzed by EA-IRMS. Naturally occurring isotope abundance was subtracted using the y-intercept of the raw data, which was identical to the natural abundance detected in unlabeled controls. Data points indicate net 13  c Alternative representation of data in a as 13 C/ 12 C isotope ratios where δ 13 C = (R sample /R PDB − 1) × 1000, R sample is the 13 C/ 12 C ratio of the sample, and RPDB is the same ratio of the PeeDee Belemnite reference material (0.0112372). The y-intercept value (− 36.16 ‰, see inset) can be used to infer the 13 C discrimination in Arabidopsis and closely matches values obtained for unlabeled control plants, which is within the normal range for a C3 plant ▸ maximize sensitivity to 12 CO 2 , nonetheless detects 13 CO 2 with reduced sensitivity, providing a means of estimating atmospheric half-life under these flow conditions). This value, representing the time to reach a 50:50 mixture of the two atmospheres, was subtracted from the uncorrected labeling times. The corrected labeling times were therefore calculated from the halfway point between atmospheric changeovers until harvest. All plants were thoroughly ground to a fine powder while frozen in liquid nitrogen and lyophilized to dryness prior to analysis.
To measure background isotope abundances, nine wild-type negative control plants were individually acclimated in the chamber and harvested without exposure to the isotopically enriched atmosphere. The background % atom 13 C values of these unlabeled control plants, as judged by IRMS, were compared to the y-intercept of time course labeling regression lines obtained from raw data and found to be essentially identical. This naturally occurring 13 C background was subtracted from the raw IRMS data to calculate 13 C assimilated during the labeling experiment.

EA-IRMS analysis
Aliquots of approximately 2 mg lyophilized plant tissue were weighed into tin capsules (Elementar Microanalysis) using a Mettler Toledo XP2U microbalance accurate to 10 -7 g and their exact weights recorded. EA-IRMS was performed by the Utah State University Stable Isotope Laboratory (Logan, UT). The samples were analyzed for total C and 13 C abundance using continuous flow directcombustion and IRMS with a PDZ Europa Scientific ANCA 20-20 system (Sercon Ltd., Cheshire, U.K.). The flow rate of high purity He carrier gas was 90 mL min −1 and an 18 s pulse of ultra-high purity O 2 was used for sample combustion. Temperature settings were as follows: combustion furnace, 969 ºC; reduction furnace, 595 ºC; GC oven, 65 ºC. Electronic settings included the following: electron current 200 μA, ion repeller 4.8 V, ionization energy 99 eV, ion focus 83 V, and high tension (HT) at 2469 V. During tuning, HT was adjusted to the center of the 2/1 beam ratio plateau following injection of pure CO 2 gas (800 μg C) at natural abundance 13 C. With these settings, integration of sample N 2 peaks (masses 28,29,30) occurred between 91 and 190 s and integration of sample CO 2 (representing all C) peaks (masses 44,45,46) occurred between 230 and 440 s. Samples were run versus a glucose standard containing 950 μg C at 1.10068 atom % 13 C. Precision, as judged by deviation from expected 13 C values observed in the glucose standard, was better than 0.1‰ 13 C. To test the reproducibility and precision of the complete analytical pipeline, an Arabidopsis thaliana wild-type control plant labeled under a 13 CO 2 atmosphere for 1 h was weighed out as described above and analyzed ten times.

Isotope based estimates of carbon assimilation closely parallel gas exchange values
Rosette stage Arabidopsis plants were subjected to physiological 13 CO 2 labeling in a dynamic flow cuvette following adaptation in a normal atmosphere, during which time gas exchange measurements were taken. At the end of a pre-determined labeling period lasting from 6 to 42 min, the plants were flash frozen in liquid nitrogen. When aliquots of the powdered, lyophilized tissue were subjected to EA-IRMS analysis, a linear accumulation of 13 C was observed over time (Fig. 3a), indicating that the rate of 13 C assimilation could be obtained from the resulting slope (9.89 nmol 13 CO 2 mg −1 D.W. min −1 ). When plants were grown and labeled at a range of light intensities, the slope increased with light intensity as expected, indicating that 13 C content was a reliable indicator of differential assimilation rates (Fig. 3b). For plants exposed only to a normal atmosphere, where the 13 C abundance is approximately 1.1%, the 13 C content is typically depleted in plant tissue to an extent which reflects a combination of discrimination processes, including the discrimination of Rubisco against 13 CO 2 and a decreased diffusion constant of the heavier isotope in the gas through the stomata and the cytosol as it enters the chloroplast. This depletion is often represented as δ 13 C (expressed in ‰ and calculated as 1000 × (R sample -R standard )/R standard ) and usually falls in the range of − 22 to − 35‰ for a C3 plant such as Arabidopsis [57]. We calculated δ 13 C values for time course labeled plants shown in Fig. 3a; however, as expected, after less than 1 min in an atmosphere containing 400 μL L −1 13 CO 2 , the δ 13 C levels surpassed the natural isotopic abundance and increase to upwards of 1400‰ after nearly 42 min of continuous labeling (Fig. 3c). The calculated y-intercept for the computed linear regression of these data was − 36.16‰, which closely matched δ 13 C values measured in unlabeled control plants.
We estimated intra-replicate variability of this approach by carrying out multiple IRMS analyses of % 13 C and total C content on a control sample which had been 13 CO 2 labeled for ~ 1 h. The % 13 C in this sample was measured at 4.70009 ± 0.01105% with a relative standard deviation of 0.7% (n = 10). Instrumental precision based on labeled plant standards was measured at 0.07%. Most of the error, while minimal, was likely incurred during the weighing step.
Although this IRMS-based approach to estimating carbon assimilation calculates assimilated 13 C on a "per dry weight" basis, we used additional metrics collected for the samples in Fig. 3b (total rosette dry weight and photosynthetically active surface area) to convert these values to those used in gas exchange, i.e. μmol CO 2 m −2 s −1 (Fig. 4a). This conversion was done on an individual basis since the correlation between rosette dry mass and photosynthetically active surface area we observed across ~ 90 samples was too weak to apply a general rule for interconverting mass with surface area (R 2 = 0.443) (Fig. 4b). After calculating the total amount of 13 C label in a given plant and dividing this by the calculated leaf surface area and labeling time, we converted the initial units obtained by IRMS analysis (μg 13 C g −1 D.W.) to those used in gas exchange (μmol CO 2 m −2 s −1 ). In this fashion, we were able to compare carbon assimilation rates between the two methods using identical units for plants labeled over a range of light intensities (80, 140, or 500 PAR) for which we expected a range of carbon assimilation rates. When paired assimilation data derived from gas exchange and IRMS were plotted for each sample, the resulting slope was close to 1 (0.903; 95% confidence interval (CI) [0.8652, 0.9408], n = 89) (Fig. 4a), demonstrating that the IRMS-based method yielded highly similar results in terms of net carbon assimilation when compared to the gas exchange values for the same samples. The slope of the relationship is expected to be slightly less than 1, representing the loss of unlabeled carbon measured during the adaptation phase of the experiment that is immediately released as CO 2 and not maintained in the biomass.

Mutants affected in the MEP pathway or supply of its substrates have suppressed photosynthetic rates
Having established the suitability of this IRMS-based approach to estimate carbon assimilation, we next applied this technique to a collection of Arabidopsis wildtype and mutant lines affected either in structural genes of the MEP pathway or genes related to the transport and supply of substrates for this pathway. Compared to wildtype plants, plants defective in the XYLULOSE-5-PHOS-PHATE TRANSPORTER2 gene (xpt-2) displayed significantly diminished carbon assimilation rates (24% lower than wild-type; 95% CI [18%, 29%], n = 5) (Fig. 5). The prl1 mutant, defective in the PLEIOTROPIC REGU-LATORY LOCUS gene, is indirectly involved in regulating supply of substrate entering the MEP pathway [28]. IRMS analysis showed a decrease in 13 C assimilation of 44% in this mutant line (95% CI [36%, 52%], n = 9). However, the inhibition of photosynthesis was even more evident in partial loss of function mutants affected in MEP pathway structural genes such as dxs-3 or 1-hydroxy-2-methyl-2-(E)-butenyl 4-diphosphate synthase 3 (hds-3), where the IRMS-based carbon assimilation rate was 36% (95% CI [33%, 39%], n = 8) and 61% (95% CI [55%, 66%], n = 9) reduced, respectively, compared to wild-type plants The correlation between surface area and dry mass of Arabidopsis rosettes is weak and insufficient to establish a general rule used in gas exchange and 13 C labeling experiments. Surface area was estimated by comparison to a calibrated size standard as described in methods. Mass was determined following lyophilization of the intact rosette. Each point represents a single rosette stage Arabidopsis plant 50-70 days old. Outliers were removed according to the interquartile range rule (n = 10) (Fig. 5) as determined by a Student's t-test of the linear regressions from 13 C assimilation rates [40]. Taken together, mutants affected in the MEP pathway demonstrated larger declines in assimilation rates than mutants affected in substrate transport and supply. However, we could not rule out that long term, secondary developmental effects of these mutations could also be partly responsible for the observed impact on photosynthesis.

Herbicides inhibiting chloroplast terpenoid metabolism phenocopy MEP pathway mutants
We next explored the role of the MEP pathway in supporting the biosynthesis of photosynthetic co-factors in the short term by treating wild-type plants with CLZ or FSM, which selectively block the first (DXS) and second (DXR) step of the MEP pathway, respectively. Plants were subjected to the same time course labeling experiments 24 h after treatment, and the resulting lyophilized tissue was similarly analyzed by EA-IRMS. As controls, we also treated plants with MEV (which blocks HMG-CoA reductase in the cytosolic mevalonate pathway and exerts a similar effect on the cytosolic pool of IDP and DMADP), NFZ (which blocks the phytoene desaturase step of carotenoid biosynthesis), CAM, which blocks plastidic protein synthesis, or water (control). We hypothesized that CLZ and FSM treatment prior to labeling would simulate the reduced availability of IDP and DMADP in the chloroplast imposed by mutation in structural genes but would not have the same developmental effects resulting from these mutations or long term effects related to reduced levels of photosynthetic pigments or co-factors available to participate in photosynthesis. Compared to wild-type plants treated only with water (Fig. 5a), CAM treated plants displayed a 34%, reduction in carbon assimilation rates (95% CI [22%, 46%], n = 9) (Fig. 6a). Plants treated with CLZ or FSM had reductions of 52% (95% CI [46%, 57%], n = 7) and 43%, (95% CI [32%, 54%], n = 11), respectively (Fig. 6b, c). However, when the cytosolic MVA pathway was blocked with MEV (n = 10), no significant reduction in assimilation rate was observed (Fig. 6d). This suggested the inhibition of photosynthesis observed in mutant plants and wild-type plants treated with MEP pathway inhibitors was specifically related to IDP and DMADP supply in the chloroplast.
NFZ treatment also provoked a significant decline in carbon assimilation (32%; 95% CI [28%, 37%], n = 6) (Fig. 6e), although not as severe as herbicides targeting the MEP pathway. Therefore, we next carried out additional short term herbicide treatments (90 min) with CLZ, FSM, and NFZ to confirm that CLZ and FSM (but not NFZ) impacted flux in the MEP pathway under our experimental conditions. Label incorporation into DMADP was measured in the resulting tissue via acid hydrolysis to isoprene gas as described previously [12]. We observed significant reductions in flux towards DMADP in CLZ and FSM treated plants (particularly at the highest concentrations employed) but no significant decrease in NFZ treated plants (Fig. 7).

IRMS-based measures of carbon assimilation offer complementary features compared to gas exchange
Despite the centrality of gas exchange measurements in elucidating the mechanisms of photosynthetic carbon assimilation, there are limitations to this technique when working with the model plant Arabidopsis. For instance, many mutants of Arabidopsis affected in some aspect of the carbon reduction cycle may present a stunted growth phenotype which assimilate too little carbon to quantify gas exchange parameters reliably. Applying leaf clamps may not be physically feasible in this case, and Arabidopsis leaf surface area is instead estimated graphically. However, the partially overlapping leaves of Arabidopsis rosettes, whose extent increases as the plant develops, do not lend themselves well to accurate calculations of leaf surface area. In addition, gas exchange in general is a laborious process poorly suited to screening large numbers of plant lines.
For these reasons, we evaluated IRMS-based estimates of carbon assimilation in a collection of time course 13 CO 2 labeled plants to facilitate measurement of gas exchange in Arabidopsis and found that this approach provides the following advantages. First, it is based on tissue mass rather than leaf surface area, which avoids the uncertainty of partial leaf overlap and differences in individual leaf assimilation rates based on partial or total shading. Partially shaded leaves carry out photosynthesis at a lower rate which is otherwise difficult to account for. Gas exchange as a technique originated in combination with leaf clamps on large leaf species such as maize but performs poorly for small rosette leaves, where leaf surface area must be estimated graphically. These problems were largely overcome by basing carbon assimilation on mass of dry tissue, which provided a highly linear (R 2 > 0.999) measure of assimilation (Fig. 3a) whose δ 13 C curve coincided with the expected depletion of 13 C isotopes in unlabeled plants (Fig. 3c). The agreement between the δ 13 C measured in unlabeled Arabidopsis tissue and the y-intercept value obtained by linear regression of time course labeled plants supported the notion that IRMS-based estimates of carbon assimilation in 13 C labeled plants provide physiologically meaningful measures of photosynthetic rates. Second, the analysis described here is post hoc, e.g. it can be performed after the fact with previously labeled  Fig. 5a. Each point represents a single, intact plant. P-values are based on t-test of regression slopes compared to the control: ns, P ≥ 0.05; *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001 samples if stored properly and requires no online detection system. Even 13 C content in older or partially degraded samples can be measured accurately since EA-IRMS uses combustion to convert organic matter to CO 2 prior to entering the mass analyzer. Third, if the 13 C IRMS data are to be correlated to targeted analysis of label in metabolite pools, it is convenient that assimilation rates and labeling of target metabolites be based on incubation in the same 13 CO 2 atmosphere. Previously reported labeling techniques involve measuring gas exchange in a normal atmosphere and then switching to a new atmosphere for the labeling portion of the experiment [90]. This may inadvertently introduce undesirable changes to the plant's metabolism. The approach described herein permits carbon assimilation estimates and post-harvest targeted analysis to be correlated without the assumption that the change between an unlabeled and labeling atmosphere do not induce metabolic changes.
Finally, small Arabidopsis mutants or young plants which cannot physically undergo leaf clamping or enclose enough area to produce a sufficient CO 2 drawdown (as measured on an IRGA) can nonetheless be labeled and analyzed by EA-IRMS to estimate carbon assimilation using the technique described here. Plate grown seedlings as young as 10-14 days could theoretically be labeled and analyzed with this technique if grown at a sufficient density on sterile plates to yield enough tissue for EA-IRMS analysis. Moreover, large biological replicate pools can be labeled simultaneously in a sufficiently large cuvette, increasing the throughput of labeling experiments and improving the statistical power of analysis. In this fashion, this technique can be used to screen multiple lines or treatments in parallel based on comparative carbon fixation rates under a given condition. It should be noted that, by design, the data presented here only address leaf level carbon fixation and do not account for carbohydrates transported to root tissue. However, the transport of 13 C into root tissue may be monitored with exactly the same approach by harvesting root tissue separately and subjecting aliquots to EA-IRMS. Indeed, this technique may provide a powerful method for analyzing the partitioning of carbon resources by comparison of 13 C detected in leaf and root tissue separately.
The use of IRMS to investigate photosynthesis has several precedents. IRMS analysis of carbon isotopes in leaf tissue has previously been used to determine the influence of CO 2 partial pressure across stomata on isotopic discrimination during carbon assimilation [25]. In this study, which relied on natural 13 C abundances rather than labeling, changes in carbon isotope composition of CO 2 passing over the surface of a leaf were correlated to isotopic discrimination in intact leaves as the CO 2 partial pressure was experimentally varied. It confirmed that 13 C discrimination increased as the CO 2 partial pressure gradient across stomata decreased in C 3 plants but not in C 4 plants. IRMS analysis has also been used to probe the source of carbon supplying isoprene emissions in myrtle, buckthorn, and velvet bean [1], where δ 13 C differences in CO 2 and isoprene emissions implicated DXS as the main isotopic discrimination step in isoprene formation. IRMS analysis was further used to identify the presence of carbon pools contributing to dark respiration not derived from recent photosynthate [59]. To our knowledge, our method is the first report using 13 CO 2 and IRMS to estimate carbon assimilation directly.

IRMS-based estimates of carbon assimilation closely resemble those of gas exchange measurements
To compare assimilation rates estimated by 13 C content to traditional gas exchange measurements, we converted 13 C content on a per mg basis to the units used in gas exchange (μmol CO 2 m −2 s −1 ). The observed slope, slightly less than 1, reflects unlabeled carbon loss released during the labeling segment. This release is comprised of carbon within the photorespiratory pool, additional pools lost from the C2 cycle, and related processes termed non-photorespiratory CO 2 release [76,77,83]. In a normal atmosphere, this CO 2 release is included in A. However, once the atmosphere is switched to a 13 CO 2 containing atmosphere, the slow loss of residual, unlabeled CO 2 is no longer subtracted from the apparent net assimilation of 13 C (A 13 ) since we include only 13 C quantified by EA-IRMS in our analysis. Therefore, the same time points yield slightly higher values for A 13 than for A, and consequently, the slope remains below 1. This slope remained consistent over all time point assayed, suggesting the slow release of unlabeled CO 2 from non-photorespiratory sources continued throughout the ~ 45 min labeling period. In contrast, photorespiratory intermediates are expected to become fully labeled within minutes, and their loss as 13 CO 2 would then be factored into the 13 C detected in lyophilized plant tissues. Our results support previous observations that non-photorespiratory CO 2 loss is a significant and distinct form of respiration than photorespiratory CO 2 loss [59].

A block in the MEP pathway disrupts photosynthesis through multiple mechanisms
We observed a strict decline in carbon assimilation when the MEP pathway was blocked with CLZ or FSM for 24 h (Fig. 6), and the rapid inhibition of the MEP pathway reported elsewhere for these herbicides was confirmed here by monitoring flux into DMADP shortly after treatment (1.5 h) (Fig. 7). While FSM acts as a competitive inhibitor of DXR, CLZ is converted in planta to ketoclomazone, which acts as an uncompetitive DXS inhibitor with respect to pyruvate but a mixed inhibitor with respect to d-glyceraldehyde-3-phosphate [54]. These herbicide inhibitor results were similar to the assimilation rates observed in partial loss of function mutant hds-3 [33] (Fig. 5c) as well as in the temperature sensitive mutant of DXS originally named chill sensitive 5 (chs-5, [3]) (later renamed dxs-3 [63]) (Fig. 5b), confirming that both a pharmacological and genetic block in the MEP pathway produce similar results. CAM treatment serves as positive control for inhibiting photosynthesis as the gene encoding the large subunit of Rubisco, rbcL, is encoded in the plastid genome of land plants and algae [13]. The dependence of photosynthesis on the MEP pathway is well established due to its roles in supplying the precursors for chlorophyll and carotenoid biosynthesis [70], but the exact cause of photoinhibition and lethality under conditions of MEP pathway inhibition are less clear. A reduction in photosystem II chlorophyll fluorescence, Rubisco carboxylase activity, photosynthetic electron transport, and carbon assimilation were previously reported in plants one hour after treatment with FSM [64]. This is consistent with the essential role of DXR in supplying IDP and DMADP for pigment synthesis as demonstrated in transgenic Arabidopsis lines up and down regulating this gene [18].
Carotenoids and chlorophylls are continuously synthesized and degraded during the day [19], and a block in their synthesis is expected to noticeably impact carbon assimilation rates within hours to days if the degradation pathway remains active. 14 CO 2 pulse-labeling experiments resulted in detection of radiolabel in β-carotene and chlorophyll a in as little as 30 min [8], underscoring the short half lives of these pigments exposed to high doses of radiation and the dependency on MEP pathway flux to maintain carbon assimilation rates. FSM inhibition of photosynthesis could be relieved by insertion of the complete MVA pathway into the tobacco chloroplast genome, providing an alternative source of IDP and DMADP for GGPP synthesis that was not susceptible to FSM inhibition [45]. However, FSM photoinhibition likely results from a combination of depleted pigment reserves as well as an accumulation of phototoxic intermediates of chlorophyll biosynthesis, which relies on equal contributions from the tetrapyrrole and phytyl diphosphate pathways. When the MEP pathway is blocked by FSM, a shortage of available GGPP for phytyl diphosphate production causes an accumulation of free tetrapyrrolic intermediates in the chlorophyll pathway which induces the formation of singlet oxygen and causes photooxidative stress [44]. Kim et al. further demonstrated that this FSM toxicity could be reversed by phytol supplementation or a chemical or genetic block in tetrapyrrole biosynthesis, suggesting the surplus of photoreactive tetrapyrroles in phytol-deficient cells is the source of toxicity in the short term. Indeed, Arabidopsis mutants which accumulate protochlorophyllide display a FSM-poisoning phenotype related to photooxidative stress when transitioned to the light [56], underscoring the requirement for careful coordination of these two pathways during chlorophyll biosynthesis. NFZ, in contrast, is expected to impact carotenoid levels without directly disrupting tetrapyrrole/phytol ratios because this herbicide targets phytoene desaturase downstream of the MEP pathway and should not, in principle, affect chloroplast pools of IDP and DMADP. This should therefore result in a lesser degree of photoinhibition compared to a block in the MEP pathway as seen in CLZ and FSM treated plants. Consistent with this prediction, these NFZ treatment results suggested that a disruption in supply of carotenoid precursors downstream of IDP and DMADP, while essential to a functional photosynthetic apparatus, was not as detrimental as blocking the MEP pathway, which supplies precursors for multiple photosynthetic components, including chlorophylls, and did not impair flux through the MEP pathway (Fig. 7). We therefore considered that a disruption in MEP pathway flux provoked inhibition of photosynthesis through multiple mechanisms. The phototoxic potential of excess tetrapyrrolic intermediates may explain their role in negative feedback downregulation of the tetrapyrrole biosynthetic pathway [73] and implication in chloroplast-to-nucleus retrograde signaling to coordinate the metabolic status of chloroplasts with the synthesis of nuclear encoded chloroplastlocalized proteins [20]. In summary, our results in FSM, CLZ, and NFZ treated plants likely reflect the combined effects of short term (~ 1 h) imbalances in tetrapyrrole/ phytol ratios for chlorophyll biosynthesis and/or longer term (> 24 h) impacts on carotenoid and chlorophyll steady state levels.

Exchange of common isoprenoid intermediates is insufficient to rescue photoinhibition caused by a block in the MEP pathway
Our results may be interpreted in light of the well studied exchange of common intermediates between the MEP and MVA pathways. Since the elucidation of the MEP pathway [68], attempts to establish exchange of IDP and DMADP between the cytosolic and plastidic compartments have relied on a combination of inhibitors and isotopically labeled advanced precursors for each pathway [31,41,47,62]. While many examples of terpenoid secondary metabolites containing isoprenoid units of mixed origin have been documented [23,37,38,66,85], this exchange appears limited to specialized structures and metabolic contexts and is generally unable to reverse the lethality of a complete genetic or pharmacological block [69]. However, some cytosol-to-plastid exchange of common intermediates for phytohormone or carotenoid synthesis may occur in etiolated seedlings during the transition from skotomorphogenic to photomorphogenic development [43,60]. While in vitro evidence of a proton symporter capable of facilitating the plastid-to-cytosol unidirectional transport of IDP and GDP was described in spinach, kale, and mustard [9], to date no gene for such a transporter has been isolated from any plant species. Our IRMS-based carbon assimilation results on plants treated with FSM or CLZ are consistent with the notion that any contribution from the MVA pathway to carotenoid or chlorophyll biosynthesis in adult plants is, at most, minor and insufficient to meet the demand for precursors needed to form light harvesting pigments in the chloroplast. Moreover, due to the marked declines in carbon assimilation in FSM or CLZ treated plants we report here and the decrease in fixed carbon resources this implies, it is unsurprising that a block in the MEP pathway should result in a decline in cytosolic terpenoid biosynthesis, though this is not necessarily indicative of exchange of common intermediates.
A more significant form of exchange may occur upstream of the MEP and MVA precursor pathways through the oxidative steps of the cytosolic pentose phosphate pathway [76,77]. The xylulose 5-phosphate (Xu5P) transporter (XPT) imports Xu5P and ribulose 5-phosphate from the cytosol into chloroplasts [24], both of which may join the Calvin-Benson cycle. Incomplete labeling of isoprene emissions has been attributed to the activity of XPT [76], and this transporter also reportedly recognizes DXP [29], although the physiological significance of translocating DXP between the plastid and cytosol is unclear. We observed a statistically significant reduction in carbon assimilation in the xpt2 mutant (Fig. 5e), suggesting that it indeed plays a role in maximizing plant performance but is not essential for photosynthetic metabolism.
PLEIOTROPIC REGULATORY LOCUS1 (PRL1) is a WD-40 RNA binding protein with multiple roles in immunity and stress tolerance. The prl1 mutant accumulates higher than normal levels of chlorophyll and carotenoids and is highly resistant to FSM and CLZ despite no detectable changes to MEP pathway transcript or protein levels [28]. It has also been implicated in root stem cell niche activity [42] and in regulation of miRNAs [91]. Due to its apparently higher than normal flux through the MEP pathway and elevated photosynthetic pigment levels, we compared its carbon assimilation rates to partial loss of function mutants of the MEP pathway (dxs3 and hds3). However, all three mutants displayed significant declines in carbon assimilation rate (Fig. 5), indicating that any increase in flux through the MEP pathway in the prl1 mutant comes at the expense of developmental abnormalities in stem cell organization or disruptions to miRNA processing.
The IRMS-based carbon assimilation measuring technique presented here represents a novel alternative to gas exchange measurements that is well suited to the investigation of central metabolism in Arabidopsis, particularly mutant lines with reduced photosynthetic capacities and stunted growth phenotypes or in combination with herbicide treatments intended to dissect regulatory mechanisms controlling metabolism. Unambiguous assignment of carbon assimilation rates in greening seedlings and other early developmental stages, or in leaves vs. roots to address carbon partitioning between organs, can be implemented with minimal modification of this method. When combined with targeted analysis of isotopic tracer by more conventional metabolomics approaches, this method presents many avenues for furthering insights into the basic mechanisms of photosynthesis.