Sudden elevation of carbon dioxide concentration causes perturbation of the electron transport chain and triggers defense responses in Arabidopsis thaliana
Danial Shokouhi, Jakob Sebastian Hernandez, Dirk Walther, Gabriele Kepp, Serena Schwenkert, Dario Leister, Jürgen Gremmels, Ellen Zuther, Jessica Alpers, Thomas Nägele, Arnd G. Heyer
TL;DR
A sudden rise in CO2 causes stress in Arabidopsis by disrupting photosynthesis and triggering defense mechanisms like increased sulfur and methionine metabolism.
Contribution
The study reveals that elevated CO2 induces stress responses in wildtype Arabidopsis through electron transport chain perturbation and defense gene activation.
Findings
Wildtype Arabidopsis shows stress symptoms and photosynthetic electron transport chain disruption under sudden CO2 elevation.
Defense responses include increased sulfate assimilation, methionine cycle activity, and glucosinolate metabolism in wildtype plants.
Hexokinase1 is identified as a central regulatory hub in the wildtype's response to elevated CO2.
Abstract
Arabidopsis wildtype plants suffer symptoms of stress at a sudden increase in CO 2 concentration, resulting from perturbation of photosynthetic electron transport. Defense-related gene induction includes increased methionine cycle and glucosinolates metabolism. Elevated CO2 (eCO2) increases photosynthetic performance of plants, but also leads to decreased nitrogen-to-carbon ratio and a long-term decline in photosynthetic activity, known as photosynthetic acclimation. It is unclear whether initially increased CO2 assimilation or perturbation of the physiological homeostasis triggers acclimation. Here, we used a combination of omics analysis to investigate immediate (1 day) and delayed (7 days) responses of plants to rising atmospheric CO2, thus allowing us to discriminate regulatory from metabolic effects. Responses of wildtype Arabidopsis plants, Columbia-0, were compared to those of…
Genes, proteins, chemicals, diseases, species, mutations and cell lines named across the full text — each resolved to its canonical identifier and authoritative record.
Click any figure to enlarge with its caption.
Figure 1
Figure 2
Figure 3
Figure 4
Figure 5
Figure 6- —http://dx.doi.org/10.13039/501100001659Deutsche Forschungsgemeinschaft
- —Universität Stuttgart (1023)
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsPlant responses to elevated CO2 · Atmospheric chemistry and aerosols · Heme Oxygenase-1 and Carbon Monoxide
Introduction
Industrialization has led to an increasing atmospheric CO_2_ concentration that is predicted to reach 800 to 1000 ppm by the end of the century (IPCC 2021). It is known for a long that long-term exposure of plants to eCO_2_ causes a decline in the maximum rate of carboxylation (Vcmax) of ribulose-bisphosphate carboxylase/oxygenase (Rubisco) and thus photosynthesis (Ainsworth and Long 2005; Ainsworth and Rogers 2007). However, the extent to which this depends on physiological (e.g. stomatal conductance), morphological (canopy traits) or biochemical (e.g. enzyme activity) parameters is still under debate (Krämer et al. 2022b). One obvious effect is an increased carbon-to-nitrogen ratio that may result from (i) high CO_2_ levels stimulating carbohydrate accumulation, thus causing a dilution of N-containing compounds (Chen et al. 2005; Krämer et al. 2022a) and (ii) elevated internal CO_2_ concentrations that lead to decreased stomatal conductance, causing low transpiration rates that limit mineral uptake and transport (Ainsworth and Rogers 2007; Shi et al. 2021). An additional factor could lie in the reduction of photorespiration at eCO_2_ (Krämer et al. 2022b). The photorespiratory pathway starts from oxygenation of ribulose-1,5-bisphosphate (RuBP) and is, therefore, diminished at eCO_2_.
Oxygenation of RuBP produces phosphoglycolate, which inhibits enzymes of the Calvin-Benson-Bassham Cycle (CBBC) and must be detoxified. After dephosphorylation, glycolate is exported from plastids and converted to the amino acid glycine in peroxisomes. In the mitochondria, two molecules of glycine are converted to serine which is transported back to the peroxisomes, where it is deaminated and reduced to glycerate. Following transfer to plastids and phosphorylation, phosphoglycerate re-enters the CBBC. The photorespiratory pathway is an important source for serine and glycine in leaves of C_3_-plants and the reduction of its turnover may thus affect the whole amino acid metabolism (Friedrichs et al. 2024).
While impacts on N-metabolism are obvious, it has also been reported that photorespiration affects sulfate assimilation. Abadie and Tcherkez (2019) reported a significant enrichment of ^33^S labeling of cysteine under photorespiratory conditions and a significant increase in ^33^S-methionine and ^33^S-cysteine content at high oxygen levels in sunflowers. It is unclear whether this effect depends on the accumulation of serine, which is the substrate for cysteine synthesis. Li et al. (2008) reported that at eCO_2_ serine levels dropped in both Arabidopsis accessions, Col-0 and Cvi-0, while cysteine did not change and methionine showed accession-specific effects. In C_4_ plants (where photorespiration is suppressed) of the genus Flaveria, sulfur assimilation is preferentially localized to roots, because serine production occurs via the phosphorylated pathway mainly in the roots.
The complexity of interactions of C, N and S metabolism is further increased by the fact that plant metabolism is highly compartmentalized, which might result in metabolite levels that are not assessable in whole-cell extracts. In addition to that, isoforms of many enzymes exist in the different compartments, resulting in a redundancy of various metabolic pathways. For example, glycolysis and the oxidative pentose phosphate pathway exist in plastids and cytosol, and synthesis of various amino acids is located in the cytosol, plastids, as well as mitochondria. To disentangle overlapping reactions, mutants lacking specific isoforms of enzymes are frequently utilized that disrupt metabolite conversion in a specific compartment. While this strategy allowed insights into the complex interactions of metabolic pathways, knowledge is still limited due to the high intracellular mobility of metabolic intermediates, and also because of the so-called pleiotropic effects of mutations, which are unexpected based on current knowledge of the functions of the mutated genes (Serrano-Mislata et al. 2017).
Disruption of photorespiration by mutation of essential genes is lethal in C_3_ plants at ambient CO_2_ (Bauwe 2023) and only very few mutants exist that are severely disturbed in the photorespiratory pathway but can still grow at ambient CO_2_. One of these is the hpr1-1 mutant (Timm et al. 2008), which is deficient in peroxisomal hydroxypyruvate reductase, the second last step in photorespiration before phosphoglycerate re-enters the Calvin-Benson-Bassham Cycle (CBBC). Two other isoforms exist, located to the cytosol and plastids, respectively, and, although mutation of all three isoforms has additive effects, a triple mutant is still viable (Timm et al. 2011). This raises the question whether other pathways exist to detoxify the phosphoglycolate produced upon oxygenation of RuBP.
In the present study, we aimed at disentangling the effects of eCO_2_, on the one hand linked to increased CO_2_ fixation, and on the other hand caused by reduced photorespiration. To gain insight into early events of regulation, we exposed plants grown at ambient CO_2_ (450 ppm) to suddenly elevated (1000 ppm) CO_2_ and recorded responses of transcriptome, proteome and subcellularly resolved metabolome over a time course of seven days. This revealed that eCO_2_, which provides elevated substrate concentration for photosynthesis, caused symptoms of stress in wild-type plants by perturbing the balance of photosynthetic electron transport. In contrast, it led to a relaxation of stress symptoms in a mutant with reduced photorespiratory capacity. Comparison of the two responses yielded new insights into how plants cope with different atmospheric CO_2_ levels.
Materials and methods
Plant material and growth condition
Arabidopsis thaliana (L.) Heynh. Col-0 as wildtype, and the hpr1-1 mutant (SALK067724) were used in this study. All plants were grown in fully controlled growth chambers at Stuttgart University for six weeks in soil (mixture of black peat, white peat and coconut macaroon, Art.-Nr 455,433, Klasmann-Deilmann GmbH, Geeste, Germany) under ambient carbon dioxide (aCO_2_, 450 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 20 ppm) and short day (8-h light/16-h dark, 100 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\upmu$$\end{document} mol m^−2^ s^−1^, 60% relative humidity, temperature 22/16 °C) in order to prevent floral induction. Light was provided by a composition of six LED creating the spectrum shown in Fig. S1. Plants were fertilized with each watering (4% N, 4% P, 3% K fertilizer WUXAL, Hauert MANNA Düngerwerke GmbH, Nürnberg, Germany). After 6 weeks, three biological replicates per genotype, consisting of pools of three plants each (9 plants in total per genotype), were harvested (day 0, aCO_2,_ “TP0”), and the double number of plants exposed to elevated CO_2_ (1000 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 20 ppm over the full diurnal cycle) at otherwise identical environmental settings. Consecutive sampling was done at days 1 (“TP1”) and 7 (“TP7”), each time 4 h into the light phase at the middle of a short day. Thus, for each genotype, three biological replicates were harvested, each consisting of a pool of three full rosettes (3 replicates per genotype and time point). Aliquots of fresh plant material were prepared for Proteomics and Transcriptomics analysis, and the rest subjected to lyophilization (Heto PowerDry LL3000; Thermo Electron C., Schwerte, Germany) for subcellular fractionation.
RNA extraction and RNA sequencing
100 mg of homogenized plant material of three replicates per genotypes (Col-0, hpr1-1) and condition (aCO_2_, eCO_2_ 1 d and 7 d) was used for RNA extraction. Total RNA was isolated using a single-step Trizol-based method (Zuther et al. 2019). DNase treatment was carried out on 4 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\upmu$$\end{document} g of RNA (RapidOut DNA-removal Kit, Thermo Fisher Scientific). DNA removal was validated by qPCR using intron-specific primers (AT5G65080) (Zuther et al. 2012). RNA integrity and quality were validated with an Agilent 2100 Bioanalyzer (Agilent Technologies). Library preparation and RNA sequencing (RNA-Seq) were carried out at Genewiz-Azenta Life Sciences (Leipzig, Germany; https://www.genewiz.com/). Library preparation and sequencing were performed according to the Genewiz-Azenta workflow.
RNA-seq analysis
Paired-end RNA sequencing data for all 60 samples (35–164 million reads per sample) were checked for quality using FastQC (https://www.bioinformatics.babraham.ac.uk/projects/fastqc/, version 0.11.8), and trimmed using the CLC Genomics Workbench (https://www.qiagen.com/, version 22.0.2). Using the CLC Genomics Workbench, reads were mapped to the reference transcriptome, Col-0, as obtained from TAIR10 (https://www.arabidopsis.org/download/list?dir=Genes%2FTAIR10_genome_release/) and processed to yield quantitative expression values. On average, 97.18% of the reads mapped uniquely to transcripts, yielding quantitative information for 31,838 transcripts associated with 24,231genes.
Proteome analysis by mass spectrometry
Protein extraction and trypsin digestion were carried out following the protocol of Marino et al. (2019). Liquid chromatography-tandem mass spectrometry (LC-MS/MS) was conducted as previously described, with peptides separated over a 90-min linear gradient of 5–80% (by vol.) acetonitrile (Espinoza-Corral et al. 2023). Raw data files were processed using MaxQuant software version 2.4.14.0 (Cox et al. 2014). Peak lists were searched against the Arabidopsis reference proteome (Uniprot, https://www.uniprot.org/) using default settings, with the ‘match-between-runs’ feature enabled. Protein quantification was performed using the label-free quantification (LFQ) algorithm (Cox et al. 2014). Subsequent analysis was executed using Perseus version 2.0.11 (Tyanova et al. 2016). Potential contaminants, proteins identified only by site modification, and reverse hits were excluded from further analysis. LFQ intensities were log_2_-transformed, and missing values were imputed from a normal distribution using Perseus with standard settings.
Non-aqueous fractionation (NAF)
Subcellular fractionation of lyophilized plant samples by NAF procedures was performed as earlier described (Fürtauer et al. 2019; Hoermiller et al. 2022). Briefly, approximately 80–100 mg of lyophilized leaf homogenate were suspended in 10 ml of ice-cold mixture of tetrachloroethylene-heptane (solvent A, ρ = 1.36 g cm^−3^) and sonified on ice bath for intervals of 5 s with 15 s pauses over a total time course of 12 min (Branson Sonifier 250; Branson, Danbury, Connecticut, USA; output control 3). Under constant cooling, sonified homogenate was sieved through nylon gauze, pore size 30 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\upmu$$\end{document} m, and the filtrate was centrifuged for 10 min at 2350 g, 4 °C. The supernatant was discarded and the pellet re-suspended in 1.5 ml of fresh cold solvent A, subsequently loaded onto the ice-cold non-aqueous linear gradient combination of organic solvents initiated by solvent A, and ending up to pure tetrachloroethylene (ρ = 1.6 g cm^−3^). Gradients were subjected to ultracentrifugation (Optima^™^ L-90 K; Beckman Coulter, Krefeld, Germany) for 3 h at 121,000 g, 4 °C. Fractionation of centrifuged gradient was performed into nine 1 ml fractions and each fraction aliquoted to 5 equal sub-fractions and immediately dried under vacuum condition for subsequent metabolite and marker enzyme analysis. Alkaline pyrophosphatase served as plastidial marker, UGPase as cytosolic marker, succinyl-CoA-synthetase as mitochondrial marker, and acid phosphatase as vacuolar marker enzyme, as described earlier (Knaupp et al. 2011; Fürtauer et al. 2019; Hoermiller et al. 2022).
Metabolite profiling
For metabolite analysis, one aliquot of NAF and a respective whole cell sample was subjected to carbohydrate determination by HPLC (Dionex ICS 6000, Thermo Fisher Scientific), yielding concentrations of glucose, fructose, sucrose and raffinose as described by Friedrichs et al. (2024). Briefly, extraction was performed in 80% ethanol at 80 °C, followed by vacuum drying. Dried extracts were re-suspended in Honeywell chromatographic grade water (Honeywell Specialty Chemicals, Seelze, Germany) and subjected to HPLC analysis on Dionex CarboPac PA1 BioLC column (4 × 250 mm, Thermo Fisher Scientific). Carboxylic acids, including pyruvate, malate, fumarate and citrate, as well as minerals (nitrate, phosphate, sulfate), were quantified by anion-exchange chromatography using another aliquot of samples as described by Friedrichs et al. (2024). Briefly, extraction was done in 1 ml of 55 °C Honeywell water and incubation at 95 °C, followed by separation on Dionex IonPac AS11-HC RFIC column (4 × 250 mm, Thermo Fisher Scientific).
Amino acid measurements were performed by quantitative GC-MS/MS as described (Molnar-Perl and Katona 2000; Sobolevsky et al. 2003; Guo et al. 2015) with a few modifications using a third sample aliquot. Basically, a modified solid phase extraction of amino acids was performed by suspending NAF samples in 1 ml of 10 mM HCl and 10 min shaking at RT, followed by 2 min centrifugation at 14,000 g, RT. Hundred microliters of supernatant in addition to 10 nmol of norvaline (Acros Organics, Geel, Belgium) as internal standard were subjected to amino acid purification, using homogenized suspension of 100 mg ml^−1^ of ion-exchange resin (AmberChrom^™^, Dowex 50WX4, 200–400 mesh, Merck) in 10 mM HCl, incubated for 15 min at RT in Mobicol spin classic tubes equipped with 10 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\upmu$$\end{document} m pore-sized filters (MoBiTec GmbH, Göttingen, Germany), followed by centrifugation 5 min at 1000 g, RT. Resin was washed twice by 80% methanol and 1 min centrifugation each time at 1000 g, RT, to remove non-amino acid metabolites. Subsequently, resin containing purified amino acids was suspended in 150 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\upmu$$\end{document} l of 1:1 mixture of methanol and 8 M ammonia and centrifuged for 2 min at 5000 g, RT. Eluent, containing purified amino acids, was dried in a speed vacuum concentrator (ScanSpeed 32, Lynge, Denmark). Upon drying, derivatization was done by adding 50 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\upmu$$\end{document} l of both MTBSTFA (Sigma Aldrich) and acetonitrile, 1 h incubation at 95 °C followed by 2 h, at RT and subsequent analysis by GC-MS/MS (TQ8040, Shimadzu). One microliter of the derivatized samples was injected to device, applying helium as carrier gas at a flow of 1.12 ml min^−1^. Stationary phase was a 30 m Optima 5MS-0.25 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\upmu$$\end{document} m fused silica capillary column. Ion source, column oven and injection temperature were set up at 250 °C, 100 °C and 250 °C, respectively. A split 10 gradient program was applied with an initial column temperature of 100 °C for 1 min, followed by a 15 °C increment per minute till 290 °C, holding for 3 min, again followed by the same increment rate to reach the final temperature of 330 °C and holding for 10 min. Subsequent to 5 min solvent delay, the spectra of MS device were recorded in Q3 scanning mode.
All other metabolites including carbohydrates, sugar alcohols, carboxylic acids and polyamines were measured using the last remaining sample aliquots by GC-MS/MS as described earlier (Krämer et al. 2022b). Briefly, extraction on ice into methanol-chloroform-water mixture, 2.5:1:0.5 (by vol.) was followed by centrifugation. Polar phase was separated and along with norvaline as internal standard, dried in a speed vacuum concentrator equipped with a cold trap (ScanVac, Lillerød, Denmark). After drying, methoximation was performed using 20 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\upmu$$\end{document} l of methoxamine hydrochloride (Sigma Aldrich) dissolved in pyridine (40 mg ml^−1^) by 90 min incubation at 30 °C followed by silylation, using 80 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\upmu$$\end{document} l of MSTFA (Sigma Aldrich) and 30 min incubation at 50 °C. Detection was done by the same GC-MS/MS device as amino acids and a different split 10 gradient program as 70 °C for initial temperature and one min holding, followed by 15 °C increment per minute till reaching to final 330 °C, and holding for 10 min. Ion source, column oven and injection temperature were set up at 250 °C, 70 °C and 230 °C, respectively, and after 4.7 min solvent delay, spectra were recorded in Q3 scanning mode.
Gas exchange measurements
Diurnal photosynthesis measurements were conducted as described previously by Krämer et al. (2022b). Briefly, an infrared gas analyzer (Uras 3G; Hartmann and Braun AG, Frankfurt am Main, Germany) was utilized. The system operated at a flow rate of 40 l h^−1^, and each channel was measured sequentially for 6 min across a complete diurnal cycle at a frequency of 0.1 Hz. Spline interpolation was applied to the recorded time points to reconstruct a continuous profile of photosynthetic activity. All measurements were performed under the same growth light and CO_2_ conditions. Dark respiration (Rd) was obtained as an average over the entire dark phase, subtracted for the first five minutes of dark phase. The CO_2_ compensation point in the absence of mitochondrial respiration (Γ^*^) was determined following Krämer et al. (2022b).
Instantaneous stomatal conductance for CO_2_ (GCO2), assimilation rate (A) and intercellular CO_2_ (Ci) were measured simultaneously by GFS-3000 portable gas exchange fluorescence system (Heinz Walz GmbH, Effeltrich, Germany). Briefly, a standard measuring head, equipped with Arabidopsis chamber 3010-A, was mounted, and all environmental parameters were adjusted as growth condition, including light intensity, relative humidity, temperature and CO_2_ concentration. To avoid turbulences of real-time measurements a flow rate of 700 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\upmu$$\end{document} mol min^−1^ and an impeller speed 6 were selected. The same leaf area of individual replicates and measuring intervals of 5 s over 1 min time courses were considered. Device calibration was performed before starting measurements and at least 30 min of acclimation period were considered for the chamber and sensors to the new adjusted parameters. Collecting data started right after stabilizing parameters. Intercellular CO_2_ according to Von Caemmerer and Farquhar (1981) was calculated by the device according to Eq. 1:
\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$C_{{\mathrm{i}}} = \frac{{\left( {G_{{{\mathrm{CO}}2{ }}} - \frac{{\mathrm{E}}}{2}} \right) \times C_{{\mathrm{a}}} - A}}{{G_{{{\mathrm{CO}}2}} + \frac{{\mathrm{E}}}{2}}}$$\end{document}whereby Ci is intercellular CO_2_ mole fraction (ppm), GCO2 is stomatal conductance for CO_2_ (mmol m^−2^ s^−1^), E is transpiration rate (mmol m^−2^ s^−1^), Ca is CO_2_ mole fraction in the chamber (ppm) and A is assimilation rate ( \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\upmu$$\end{document} mol m^−2^ s^−1^).
Chlorophyll fluorescence
Chlorophyll fluorescence imaging was performed as described by Hoermiller et al. (2022) using a FluorCam FC 1000-H system (Photon Systems Instruments, Brno, Czech Republic) operated in pulse-amplitude-modulated mode (PAM). Plants were dark-adapted for at least 10 min under growth conditions prior to measurements. Right after dark adaptation, within the minimal fluorescence (F0) measurement mode, weak far-red illumination was applied by the device automatically to fully oxidize the plastoquinone pool, allowing reliable determination of F0.
Fluorescence was excited using the system’s internal low-intensity modulated red measuring light (measuring beam). Kautsky induction was initiated by continuous actinic illumination corresponding to 100 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\upmu$$\end{document} mol photons m⁻^2^ s⁻^1^, as independently verified using a quantum sensor and matching the growth irradiance. Fluorescence induction was recorded for 120 s. During actinic light exposure, five saturating light pulses for 800 ms duration were applied to determine maximal fluorescence in the light-adapted state (Fm′) and to ensure attainment of steady-state photosynthesis. Saturating pulses were empirically verified to fully close PSII reaction centers, as indicated by stable maximal fluorescence signals at steady state level. The measuring beam consisted of low-intensity, modulated red light integrated within the imaging system, fixed by the manufacturer and did not induce detectable actinic effects. After Kautsky induction, a relaxation phase of 20 s was recorded. According to the global PAM mode of FluorCam, 10 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\upmu$$\end{document} s of shutter frequency and 30% sensitivity was set. Fluorescence parameters were averaged over the whole rosette area using at least eight different regions per leaf. Five biological replicates were analyzed per genotype and CO_2_ treatment.
Accordingly, non-photochemical quenching (NPQ) and the maximum quantum efficiency of PSII (Fv/Fm) were determined from chlorophyll fluorescence measurements to assess photoinhibition. According to Eq. 2, the photochemical yield of PSII during light adaptation (ΦPSII) was calculated based on fluorescent measurements and subsequently used to estimate the linear electron transport rate ( \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$J$$\end{document} ), according to Eq. 3:
\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\Phi }_{PS{\rm I}{\rm I}}=({F}_{m}{\prime}-{F}_{s}{\prime})/{F}_{m}{\prime}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$J={\alpha \times \beta \times PPFD\times \Phi }_{PSII}$$\end{document}where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\alpha =$$\end{document} 0.84 and \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\beta =0.$$\end{document} 5, are representing the leaf absorptance and the fraction of the absorbed irradiance that reaches PSII and PPFD represents the incident photosynthetic photon flux density corresponding to the growth condition, following Pons et al. (2009).
Combining gas-exchange measurements with chlorophyll fluorescence–based estimates of linear electron transport rate, chloroplastic CO_2_ concentration (Cc) was estimated according to Eq. 4 following Harley et al. (1992). The resulting Cc values were then used to estimate mesophyll conductance (gm) based on Fick’s law (Eq. 5), as described by Kromdijk et al. (2020):
\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${C}_{\mathrm{c}}=\frac{{\Gamma }^{*}.\left[J+8\left(A+{R}_{\mathrm{d}}\right)\right]}{\left[J-4\left(A+{R}_{\mathrm{d}}\right)\right]}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\mathrm{g}}_{\mathrm{m}}=A/({C}_{\mathrm{i}}-{C}_{\mathrm{c}})$$\end{document}Biochemical modeling
The CO_2_ dependence curve of photosynthesis (A–Ci) was fitted using the Farquhar-von Caemmerer-Berry (FvCB) model as implemented in the plantecophys package in R according to Farquhar et al. (1980) and Duursma (2015), assuming net assimilation rate was given according to Eq. 6:
\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$A={V}_{\mathrm{c}} .\left(1-0.5\phi \right)-{R}_{\mathrm{d}}$$\end{document}where \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\phi$$\end{document} is representing the ratio of oxygenation (Vo) to carboxylation (Vc) of Rubisco and defined according to Eq. 7:
\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\phi =2{\Gamma }^{*}/{C}_{\mathrm{c}}$$\end{document}Rearrangement of Eq. 6, following Von Caemmerer (2000), allowed identification of Vc and Vo, and subsequently NADPH and ATP consumption rates were estimated according to Eq. 8 and 9**,** respectively:
\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\text{NADPH consumption rate}=\left(2+2\phi \right) .{V}_{\mathrm{c}}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\text{ATP consumption rate}=\left(3+3.5\phi \right) .{V}_{\mathrm{c}}$$\end{document}Average of assimilation rates obtained from seventeen biological replicates at ten given Ci set points were used as fixed input for fitting the FvCB model, which was further constrained by estimated mesophyll conductance and the \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${\Gamma }^{*}$$\end{document} values of the corresponding genotype. The Michaelis–Menten constants of Rubisco for CO_2_ (Kc) and O_2_ (Ko), at 25 °C were set to 404.9 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\upmu$$\end{document} mol mol^−1^ and 278.4 mmol mol^−1^, respectively, and were further temperature-corrected ( \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$f$$\end{document} ) for given leaf temperature (Tleaf) using Arrhenius functions according to Eq. 10:
\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$f=f\left(25^\circ{\rm C} \right).\mathrm{exp}[{E}_{\text{a }}({T}_{\mathrm{leaf}}-298)/(298 R {T}_{\mathrm{leaf}})]$$\end{document}whereby the activation energies (Ea) for Kc and Ko were 79.43 and 36.38 kJ mol^−1^, respectively. A universal gas constant (R) of 8.314 J mol^−1^ K^−1^ and a constant intercellular O_2_ concentration (O) of 210 mmol mol^−1^ were assumed (Farquhar et al. 1980; Ishikawa et al. 2007).
Following Von Caemmerer (2000), to define whether net photosynthetic CO_2_ assimilation was limited by Rubisco (Ac) or by electron transport (RuBP regeneration, Aj) according to Eq. 11, the Vcmax and maximum electron transport rate (Jmax) were estimated by fitting the FvCB model and subsequent calculation of Ac and Aj according to Eqs. 12 and 13:
\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${A}_{\mathrm{n}}=\mathrm{min} \{{A}_{\mathrm{c}}{,A}_{\mathrm{j}}\}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${A}_{\mathrm{c}}=\frac{{V}_{\mathrm{cmax}}({C}_{\mathrm{c}}-{\Gamma }^{*})}{{C}_{\mathrm{c}}+{(K}_{\mathrm{c}}(1+O/{K}_{\mathrm{o}}))}-{R}_{\mathrm{d}}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${A}_{\mathrm{j}}=\frac{J ({C}_{\mathrm{c}}-{\Gamma }^{*})}{{4C}_{\mathrm{c}}+8{\Gamma }^{*}}-{R}_{\mathrm{d}}$$\end{document}Triose phosphate utilization (TPU) limitation of the assimilation rate (Ap) was not estimated, because TPU limitation was not detected within the measured Ci range according to the model fit; therefore, Ap was not considered further as a limiting factor in this study.
Eventually, RuBP saturated carboxylation rate (Wc) under competitive inhibition by O_2_ was calculated according to Eq. 14 and compared to electron transport limited rate of RuBP regeneration (Wj) given by Eq. 15 (Farquhar et al. 1980). This comparison was used to identify the dominant biochemical limitation of photosynthesis at given Ci levels and to evaluate its accordance with physiological and omics-based data.
\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${W}_{\mathrm{c}}=\frac{{V}_{\mathrm{cmax}} {C}_{\mathrm{c}}}{{C}_{\mathrm{c}}+{(K}_{\mathrm{c}}(1+O/{K}_{\mathrm{o}}))}$$\end{document} \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$${W}_{\mathrm{j}}=\frac{{J}_{\mathrm{max}}{ C}_{\mathrm{c}}}{{4C}_{\mathrm{c}}+8{\Gamma }^{*}}$$\end{document}Phloem exudation
In order to quantify loading of carbohydrates into the phloem sap, an exudation experiment was performed according to Tetyuk et al. (2013). Five biological replicates per genotype, consisting of pools of 3 leaves from different plants, were exuded. Leaves were first treated for 1 h with 20 mM K_2_-EDTA to suppress callose formation, and then gently washed with ddH_2_O to remove remaining EDTA. Leaves were left in 1.8 ml water for exudation under the same light intensity, photoperiod and CO_2_ conditions. For eCO_2_ condition, plants which were incubated for 1 d and 7 d at eCO_2_ were used. Exudates were collected for each CO_2_ condition 4 h into the light phase at the middle of a short day. Carbohydrate quantification in phloem exudates was performed as described in the metabolite profiling section above, and by HPLC.
Data evaluation, statistics and figure illustration
Gene Ontology Term Enrichment Analysis (GO term analysis) was performed with a significance threshold of P \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$<$$\end{document} 0.01 using PlantRegMap (https://plantregmap.gao-lab.org/) following Tian et al. (2020). Protein–protein interactions were evaluated using the String database for functional protein association networks (https://string-db.org/); all active interaction sources were allowed, setting the minimum required interaction score to highest confidence, 0.9). Statistics and plots generation were performed in python (Version 3.11.5), Matlab (https://de.mathworks.com/) and R (version 4.5.1). Regarding the Col-0 metabolite profile, Tukey's HSD was applied following one-way ANOVA, as this PostHoc test is well-suited for controlling the family-wise error rate when evaluating pairwise comparisons in a single-factor design. However, for hpr1-1 vs Col-0 metabolite profiles, Bonferroni correction was employed following two-way ANOVA, because specific, planned genotype-wise comparisons at each time point were performed, and Bonferroni correction provides a more conservative adjustment appropriate for a limited number of targeted comparisons within a multifactorial framework. In order to calculate the significance of log_2_FC (fold-change) values, a one-tailed z-test was performed after calculating the geometric mean and the propagation of error as suggested by Quackenbush (2002). In addition, the P-value was adjusted for multiple testing using Bonferroni correction. All statistical tests were conducted with \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\alpha$$\end{document} = 0.05. Figures 2a, 5a and 6a were illustrated using BioRender (https://www.biorender.com/).
Results
Early responses to eCO2 at the transcriptome and proteome level
Upon shift to eCO_2_ after six weeks of growth at aCO_2_, Col-0 and hpr1-1 responded immediately, but partially contrary, at the level of transcripts, proteins and metabolites (Fig. 1). Loadings of a Principal Component Analysis (PCA) for transcripts (Fig. 1a) revealed a conserved effect after 7 days at eCO_2_ along PC2 but no convergence. A GO term enrichment of the top 1000 genes along this trajectory showed that transcripts were predominantly related to defense and stress-response mechanisms (Table S1). In contrast, the immediate response at day 1 of eCO_2_ showed a clear separation of hpr1-1 from Col-0 along PC1, which comprised enrichment of transcripts for cellular, developmental and metabolic processes (Table S1). Interestingly, the distance between the two genotypes decreased after 7 days at eCO_2_ in all omics data (Fig. 1a–c). Full lists of loadings of PCA for transcripts, proteome and metabolites and also their GO term enrichments are available as Table S1, respectively.Fig. 1. Dynamics of transcriptomes, proteomes and subcellular metabolomes under elevated CO_2_. a Principal component analysis (PCA) of mean transcript levels (n = 3). b PCA of mean protein levels (n = 3). c PCA of subcellular metabolites with individual replicates (n = 3). Colors indicate duration of eCO_2_ exposure (purple: 0; gray: 1; green: 7 days) and shapes indicate genotypes (filled circles: Col-0; filled triangles: hpr1-1). Trajectories of full-time series are indicated by solid lines, trajectories from 0 to 7 d eCO_2_ are indicated by dashed lines. Semi-transparent dashed ellipses in c represent variation of metabolite levels ( \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 1 SD), scaled for visibility. Full data sets of loadings of every PC and GO enrichment are provided in Table S1
It is known for a long that eCO_2_ causes accumulation of soluble sugars, which in turn trigger down-regulation of photosynthetic genes through sugar sensing by hexokinase 1 (HXK1) (Cheng et al. 1998). In agreement with that, upon a shift to eCO_2_ and significant accumulation of soluble sugars, we observed induction of the sugar sensor HXK1, and down-regulation of carbonic anhydrase (CA2) and light harvesting complex2 (LHCB2.4 and LHCA4) at both, transcript and proteome, level in Col-0 (Fig. 2a,b; C1vsC0_T/P). Interestingly, we also found the electron transport chain (ETC) to be affected with down-regulation of both genes encoding plastocyanin (PC) (DRT112 and PETE) and the Rieske iron-sulfur protein of Cytb6f (PETC) (Fig. 2a,b) that transfers electrons from plastoquinol to Cytb6f (Knight et al. 2002). While a trimeric complex composed of HXK1, VHA-B1 and RPT5B in response to elevated sugars is responsible for suppressing LHCB2, the details of the regulatory mechanism on PC and Cytb6f genes are not yet fully understood. This prompted us to examine primary reactions of photosynthesis and components of the redox balance. Figure 2c shows analysis of chlorophyll fluorescence, demonstrating that NPQ, i.e. dissipation of excess energy as heat, increased immediately after transfer to eCO_2_, in line with induction of the violaxanthin de-epoxidase VDE1 that converts the antenna pigment violaxanthin into protective zeaxanthin (Fig. 2a,b). Along with a significant increase in F0 and decreased Fv/Fm, photochemical quenching (qP) and fluorescence decline ratio (Rfd) were reduced (Fig. 2c). A significant decrease in ΦPSII was accompanied by a reduction of the estimated linear electron transport rate and also the NADPH and ATP consumption rate upon shift to eCO_2_ (Fig. 2c). Further indications of damage to PSII were obtained, such as evidence for plastoquinone pool over reduction, photoinhibition and oxidative stress. For example, plastidial superoxide dismutase (SOD3 and FSD1/2) and ascorbate peroxidases (SAPX and APX4) were all induced at the transcript level, clearly pointing to a disturbed redox balance as early response to eCO_2_ in Col-0 (Fig. 2b). In accordance with that, the gene for NADP-dependent malic enzyme that decarboxylates malate to pyruvate, producing NADPH, (NADP-ME4) was induced as well as several genes of the oxidative pentose-phosphate cycle (G6PD1, G6PD3 and PGD1/2/3), and isocitrate-dehydrogenase (ICDH and CICDH), which are additional sources of reducing equivalents (Fig. 2b).Fig. 2. Immediate responses of plants to eCO_2_. Colors are identical for all graphs, indicating eCO_2_ duration as purple: 0; gray: 1; green: 7 days. a Schematic representation of processes and pathways affected as immediate responses to elevated CO_2_ in wildtype. (In)direct conversions and/or induction of pathways are indicated by solid (forked) arrows and suppression is indicated by red solid lines. Electron transport chain, cyclic electron flow and proton transfer are represented by dashed arrows. The possible role of HXK1 in the regulation of Cytb6f and PC is indicated by a red dotted line. b Integrated heatmap of log_2_FC-transformed differentially expressed transcripts (_T) and proteins (P) relative to ambient conditions that relate to metabolic and signaling pathways depicted in (a). C1vsC0 and C7vsC0 compare Col-0, TP1 vs TP0 and TP7 vs TP0, respectively. H1vsH0 and H7vsH0 accordingly compare hpr1-1 (* adj. P-val \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$<$$\end{document} 0.05; n = 3; one-tailed z-test; Bonferroni correction). c Chlorophyll fluorescence measurements under ambient/elevated carbon dioxide (a/eCO_2). Asterisks (*) indicate significant difference (P \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$<$$\end{document} 0.05) between time points based on two-way ANOVA for genotype and treatment (Bonferroni multiple comparisons test, n = 5); boxes represent the interquartile range (IQR), medians and 95% confidence intervals. d Subcellular metabolite concentration (means, n = 3) as stacked bar plots. Cytosolic and plastidial metabolites are indicated by ‘C-’ and ‘P-’ prefix, respectively. Letters indicate significant differences between time points (Bonferroni multiple comparisons test following two-way ANOVA, n = 3). eCarbohydrates elevated carbohydrates, Fd ferredoxin, FNR ferredoxin-NADP^+^ reductase, ICDH isocitrate dehydrogenase, LHCB light-harvesting complex, NADP-ME NADP-malic enzyme, PC plastocyanin, PPP pentose phosphate pathway, PQ plastoquinone, PSI/II photosystem I/II
One of the pathways immediately up-regulated at the transcript level upon shift to eCO_2_ in Col-0 was sulfur assimilation, with genes for ATP-sulfurylase 1 and 4 (ATPS1/4), APS reductase 1 (APR1) and sulfite reductase (SIR) being significantly induced in Col-0 at day 1 (Fig. 2b).
Evidence from transcript data pointed to a shift of serine biosynthesis from the photorespiratory to the plastidial phosphorylated pathway (PPSB), supported by induction of the genes for plastidial glycine dehydrogenase (PGDH1/2/3), phosphoserine aminotransferase (PSAT1) and serine hydroxymethyl transferase 3 (SHM3). The PPSB is tightly coupled to sulfur assimilation, and accordingly, genes for plastidial cysteine synthesis, i.e. serine acetyltransferase (SAT1), cysteine synthase (OASB), cystathionine-γ-synthase (CGS1) and cystathionine-β-lyase (CBL) as well as methionine synthase 3 (MS3) were all induced at eCO_2_. In addition to that, genes involved in aliphatic glucosinolate metabolism (BCAT4, MAM1/3, CYP79F1, CYP83A1, SUR1 and SOT16/18) were also induced (Fig. 2b).
Genes involved in the Shikimate pathway (DHS1/2, EPSP, EMB1144/3004 and DHQS), as well as in subsequent aromatic amino acid metabolism (mainly tryptophan; ASA1/2, PAT1 and TSA/B1), were also up-regulated and play important roles in providing substrates for indolic glucosinolate metabolism, of which additional transcripts were up-regulated (CYP83B1 and CYP79B2). Besides that, genes involved in salicylic acid synthesis (ICS2 and GDG1) and jasmonic acid synthesis (LOX2/3/4/6, 4CLL5, ACX1, AIM1 and KAT2) were induced, all eventually involved in the induction of defense mechanisms that dominated PC2 (Fig. 1a).
Elevated glucose levels, sensed by HXK1, are known to induce aliphatic as well as indolic glucosinolate synthesis via transcription factors MYB28/29 and MYB 34/51/122, respectively (Eom et al. 2024). We found MYB28, 29 and 51 induced as an early response of Col-0 to eCO_2_, thus emphasizing the orchestrating role of HXK1 in early responses to eCO_2_ (Fig. 2a,b). The induction of both pathways largely relies on sufficient sulfur supply, which was reflected in a significant depletion of the free sulfate pool upon the shift to eCO_2_ (Fig. 2d).
To assess the limitation of net photosynthesis under different CO_2_ regimes, A-Ci response curves were fitted using the FvCB model. Figure 3a shows that the shift to eCO_2_ caused a marked decrease in Jmax, while Vcmax remained largely unchanged. As shown in Fig. 3c, although photorespiration was strongly suppressed under eCO_2_, as indicated by a reduced Vo/Vc ratio, the primary limitation of net assimilation shifted from Rubisco-limited assimilation (Ac) under aCO_2_ to electron transport-limited assimilation (Aj) under eCO_2,_ which was evidenced by a pronounced decrease in the parameter Aj (Fig. 3b). A similar shift was also observed from the RuBP-saturated carboxylation rate (Wc) to the electron transport-limited rate of RuBP regeneration (Wj), as shown in Fig. 3b.Fig. 3. Biochemical limitation of net photosynthesis under different CO_2_ regimes. a Net CO_2_ assimilation rate (An) as a function of intercellular CO_2_ concentration (Ci), fitted according to the FvCB model for Col-0 and hpr1-1 at different time points (TP0, TP1 and TP7). Filled black circles represent measured data points (means of n = 17 biological replicates per Ci). The fitted model of the Rubisco-limited assimilation rate (Ac) is shown in red, and the electron transport-limited assimilation rate (Aj) in blue, and the resulting limiting assimilation rate (An according to Eq. 11) in black. The Ci transition point between Ac and Aj limitation is indicated by an open black circle. Triose phosphate utilization-limited assimilation rate (Ap) was not detected and therefore not included. Estimated values of Vcmax, Jmax, and the root mean square error (RMSE) of the fitted model are shown within each panel. b Derived Rubisco-limited and electron transport-limited assimilation rates (Ac and Aj) and the corresponding RuBP-saturated carboxylation (Wc) and electron transport-limited (Wj) rates in Col-0 and hpr1-1 across time points, according to Eqs. 12–15. The Aj and Wj are shown as bar plots with diagonal black hatching, while Ac and Wc are shown as black cross-hatched bars. c Ratios of Rubisco oxygenation to carboxylation (Vo/Vc) and ratio of maximum electron transport to maximum carboxylation rates (Jmax/Vcmax) in Col-0 and hpr1-1 across time points with identical colors, illustrating changes in photorespiratory flux and the coordination between electron transport and Rubisco activity under eCO_2_. Values represent means \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} SD (n = 5). significant deviation (P \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$<$$\end{document} 0.05) is indicated by asterisks (*) and based on one-way ANOVA followed by Bonferroni multiple comparisons test across different time points (ambient/elevated carbon dioxide, a/eCO_2_) within wildtype and mutant, respectively (b and c)
The hpr1-1 mutant responded differently to eCO_2_, both, at immediate and long-term exposure (H1vsH0_T and H7vsH0_T). We found no evidence for redox imbalance in hpr1-1 plants shifted to eCO_2_, and this was in line with assessment of chlorophyll fluorescence data for NPQ, F0, Fv/Fm and qP and also with estimated ΦPSII, J, NADPH and ATP consumption rates (Fig. 2c). Neither glucosinolate metabolism nor salicylic acid or jasmonate pathways were induced in hpr1-1 at eCO_2_, the latter showing even down-regulation at day 1. However, concerted induction of genes related to pathogen responses and plant immunity underlay the shared response of hpr1-1 and Col-0 along PC2 (Fig. 1a), In addition to that, induction of serine synthesis through the PPSB pathway at eCO_2_ was a common feature of both genotypes (Fig. 2b), highlighting the impact of photorespiration on amino acid metabolism.
The situation was different at the protein level. Here, the two genotypes strongly differed initially with respect to PC1, but due to opposite responses showed some convergence after 7 days at eCO_2_ (Fig. 1b). For PC2, however, they also responded contrary and were more separated at day 7 as compared to the starting condition.
Biochemical FvCB modeling revealed a marked increase in both Vcmax and Jmax in hpr1-1 compared with Col-0 (Fig. 3a). The ratio of Rubisco oxygenation to carboxylation (Vo/Vc) was approximately 36% higher in mutant than wildtype under aCO_2_. Despite a significant decline at eCO_2_, this ratio appeared higher in the mutant (Fig. 3c). Consistent with the alleviation of photorespiration under eCO_2_, Aj and Wj became the limiting factors of net photosynthesis and carboxylation rate in hpr1-1 (Fig. 3b).
Hierarchical clustering of Euclidean distances of protein levels of Col-0 and hpr1-1 showed that proteomes of both genotypes built different clusters at day 1 at eCO_2_, while at day 7 proteomes of both genotypes were least distant (Fig. 4). This was explained by two major clusters of proteins (Fig. 4) which were either depleted in Col-0 and enriched in hpr1-1 (Fig. 4a) or vice versa (Fig. 4b). Cluster A comprised 1257 out of 1915 proteins (66%) while cluster B comprised 658 out of 1915 proteins (34%; Table S2). A GO term enrichment analysis revealed a highly significant impact on ‘metabolic processes’ in both clusters (Table S2). In cluster A, 951 out of 1257 proteins belonged to this term, in cluster B it comprised 508 out of 658 proteins. To evaluate whether enriched proteins in the metabolic process term build (potential) functional or regulatory units of metabolism, putative protein–protein interactions among them were tested using the String^®^ database (https://string-db.org/). Filtering on the highest confidence of protein association (for settings see Materials and methods), 25 disconnected graphs were found for cluster A, and 21 for cluster B. A k-means clustering comprising the 25 clusters in cluster A revealed that metabolism of cellular amino acids and purine-containing compound metabolic process were most enriched, according to the ‘−log_10_FDR’ filter, in hpr1-1 at day 1 of eCO_2_ (Fig. 4c). In cluster B, most enriched processes revealed by the k-means clustering were photosynthesis, electron transport chain, and nucleoside phosphate metabolic process (Fig. 4d). The ‘-log_10_FDR’ filter was chosen because it represents expected gene occurrences in context of its false discovery rate (Szklarczyk et al. 2025). In summary, these observations suggested a significant impact of HPR1-deficiency on the supply and metabolism of amino acids and ribonucleotides. The discrepancy between hpr1-1 and Col-0 in these metabolic processes was mitigated under eCO_2_. Second, and maybe as a consequence, proteins involved in nucleoside phosphate biosynthesis processes and photosynthesis were depleted in hpr1-1, and only after 7 days at eCO_2_ the proteome of these processes became wildtype-like.Fig. 4. Cluster analysis of proteome dynamics due to HPR1-deficiency. a and b Hierarchical clustering of mean protein levels in Col-0 and hpr1-1 based on Euclidean distances (n = 3). a Enrichment cluster of hpr1-1. b Depletion cluster of hpr1-1. c Enriched protein functions in cluster a revealed by k-means clustering of contained protein association networks (top 10 signals). d Enriched protein functions in cluster b revealed by k-means clustering of contained protein association networks (top 10 signals). FDR, false discovery rate. Size of filled circles in c and d represents the number of gene candidates that were contained in each functional category indicated on the left side. Details of gene identifiers and GO enrichments are available in Table S2
Immediate metabolic changes at eCO2
Whether the responses found at the transcriptome and/or proteome level did affect metabolic processes was analyzed by non-aqueous fractionation of subcellular compartments. This ensured that substrate concentrations were assessed at the site of reactions taking place. Besides the expected rise in sugars, especially hexoses, in cytosol and vacuole (Fig. S2), accumulation of leucine as well as aspartate, asparagine and aromatic amino acids was an immediate response to eCO_2_ in wildtype plants (Fig. S2). While glutamine accumulated transiently in the cytosol of Col-0, glutamate accumulated until day 7, and both were also enriched in plastids, indicating stimulated nitrogen assimilation. On the other hand, levels of branched-chain amino acids valine, leucine, and isoleucine sharply dropped in mitochondria on day 1 of the eCO_2_ treatment and did not reach initial levels at day 7 (Fig. S2). Carboxylic acids malate and fumarate were not affected, neither in vacuole nor mitochondria, while pyruvate accumulated in mitochondria and citrate in both mitochondria and cytosol, again pointing to enhanced amino acid formation (Fig. S2).
The increase in sugars at eCO_2_ was either similar or even stronger in the cytosol and vacuole of hpr1-1 mutant plants, indicating similar effects on primary carbon fixation (Fig. S2). However, substantial differences were observed for amino acids. While valine dropped at day 1 in plastids of Col-0, it accumulated in plastids of hpr1-1. Leucine was generally low in mitochondria of hpr1-1, but was higher in the cytosol, while isoleucine was very high in hpr1-1 at aCO_2_ in all compartments. Aspartate was generally low in hpr1-1 at aCO_2_, while asparagine accumulated in plastids at eCO_2_. Glutamate and glutamine transiently increased in mitochondria of hpr1-1, and in the case of glutamine this was observed also in the other compartments. Ornithine was very high in hpr1-1 at aCO_2_ and significantly declined at eCO_2_, and the same was observed for arginine in cytosol and plastids. Considering the extremely high levels of glycine and serine in hpr1-1 at aCO_2_, high levels of cysteine were expected. However, they did not decline in eCO_2_ in any of the compartments (Fig. S2).
Long-term effects of eCO2
To gain a better understanding of the mechanisms underlying metabolic adjustment under prolonged eCO_2_ exposure, plant responses were examined also after one week. While long term responses at day 7 showed a mild alleviation in almost all of the above-mentioned pathways that were induced at transcript levels in the wildtype, transcript levels did not align with initial levels (C7vsC0_T) (Fig. 2b). In addition, chlorophyll fluorescence data for NPQ, F0, Fv/Fm and qP at day 7 were still in support of a stress condition in wildtype plants (Fig. 2c). Accordingly, the estimated ΦPSII, J, and rates of NADPH and ATP consumption remained comparable to those observed on the first day of eCO_2_ exposure, consistent with a HXK1-dependent inhibition of electron transport (Fig. 2a,c). In accordance with that, the NADPH oxidase RBOHD that is involved in plant immunity was induced at day 7, and considering that the hydrogen peroxide sensor HPCA1 and the plasma membrane receptor kinase GHR1 were also up-regulated, this may underlie reduced stomatal conductance, which was observed in wildtype plants at day 7 (Fig. 5a) A non-significant induction of these genes was also observed in hpr1-1 (Fig. 5a).Fig. 5. Gas exchange and assimilate transport in responses to eCO_2_. a Scheme of processes affected by long-term eCO_2_ treatment. Differential transcript abundance at eCO_2_ relative to ambient is illustrated as heatmaps reflecting log_2_FC (* adj. P-val \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$<$$\end{document} 0.05 and ' adj. P-val \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$<$$\end{document} 0.07, n = 3; one-tailed z-test; Bonferroni correction). b Diurnal net CO_2_ gas exchange at different time points of eCO_2_ exposure. Abscissae reflect time per day (min) and ordinates reflect net photosynthesis (NPS) in \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\upmu$$\end{document} mol CO_2_ g^−1^ FW h^−1^; (n = 3). c Calculated internal CO_2_ concentration (Ci), stomatal conductance of CO_2_ (GCO2), and real-time assimilation rate (A) across different time points of ambient/elevated carbon dioxide (a/eCO_2_) conditions in wildtype and mutant plants. d Phloem exudation of sucrose. Ordinate reflects exuded sucrose concentration in phloem sap in \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\upmu$$\end{document} mol g^−1^ FW. Colors in all sections indicate time at eCO_2_ (purple: 0; gray:1; green: 7 days). Significant deviation (P \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$<$$\end{document} 0.05) is indicated by asterisks (*) and based on two-way ANOVA followed by Bonferroni multiple comparisons test across different time points (a/eCO_2_) within wildtype and mutant (b and d) or same timepoints among genotypes (c). CEF cyclic electron flow, eCarbohydrates elevated carbohydrates
Despite atmospheric CO_2_ enrichment, stomatal closure can impact Ci and carbon assimilation rate (A). As shown in Fig. 5c, wildtype plants exhibited a significant but transient increase of stomatal conductance for CO_2_ (GCO2) at day 1, which was lost again at day 7. Consistent with this pattern, the carbon assimilation rate was highest at day 1 and declined until day 7, when Ci was also significantly reduced. Conversely, Ci continuously increased until day 7 in hpr1-1. This elevation was not necessarily due to high CO_2_ conductance, but may instead have resulted from inefficient assimilation and a limited photosynthetic capacity, as reflected by the decrease in assimilation rate on day 7 (Fig. 5c). The CO_2_ compensation point in the absence of dark respiration (Γ^*^) was determined to be 50.6 ± 11.1 ppm in the wild type and 73.4 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 7.44 ppm in hpr1-1. Chloroplastic CO_2_ concentration (Cc) estimated using the Harley equation (Eq. 4), reached 256.6 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 13.8, 770.6 ± 68.3, and 739.4 ± 67.3 ppm at TP0, TP1, and TP7, respectively, in the wildtype, and 299.5 \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\pm$$\end{document} 5.3, 782.6 ± 25.4, and 796.2 ± 49.9 ppm in hpr1-1. Consequently, mesophyll conductance (gm, mol m^−2^ s^−1^ bar^−1^), estimated according to Fick’s law (Eq. 5), declined following the shift to eCO_2_ in both wildtype (TP0: 0.06; TP1: 0.03; TP7: 0.03) and mutant (TP0: 0.08; TP1: 0.06; TP7: 0.02).
Differential gene expression analysis in wildtype and mutant plants exposed to eCO_2_ for 7 days (C7vsC0_T and H7vsH0_T) revealed significant down-regulation of sugar transporters, SWEET11 and 12, which are involved in apoplastic sucrose phloem loading (Fig. 5a). Thus, sugar exudation from leaves was analyzed in these plants. As shown in Fig. 5d, sucrose exudation from leaves during an 8 h light phase was reduced in wildtype plants exposed to eCO_2_ for 7 days, while the reduction was less pronounced in hpr1-1. Reduced sucrose loading into the phloem may have contributed to excessive accumulation in mesophyll cells, which was observed especially for cytosol and vacuole in Col-0 at day 7 (Fig. 2d and Fig. S2).
As shown in Fig. 5b, diurnal assessment of photosynthesis revealed a significant difference in net photosynthetic rates between aCO_2_ and eCO_2_ conditions in both genotypes, while no significant difference was observed between day 1 and day 7 under eCO_2_. However, the expression of genes involved in cyclic electron flow (CEF), including PGRL1A/B, PNSB2/3/4 showed a continuous down-regulation from day 1 to day 7 of eCO_2_, pointing to the establishment of a new photosynthetic equilibrium in both genotypes (Fig. 5a).
The metabolic phenotype of hpr1-1 mutants comprises strong effects on cytosolic glycerate and pyruvate metabolism
To gain a detailed understanding of how a bottleneck in the conversion of hydroxypyruvate to glycerate in peroxisomes of the hpr1-1 mutant impacts each of the steps of the photorespiratory pathway at the different omics levels, hpr1-1 was compared to Col-0 at aCO_2_ and eCO_2_ conditions (H0vsC0_T/P, H1vsC1_T/P and H7vs C7_T/P; Fig. 6a–c). Elevated glycerate levels, especially in cytosol and plastids, differentiated hpr1-1 from Col-0 (Fig. 6c). In stark contrast, the glycolate amount was low in plastids at aCO_2_, but accumulated in the cytosol of hpr1-1 plants (Fig. S2). This pointed to an exchange of both metabolites either via the bile acid/sodium symporter BASS6 (AT4G22840) or the plastidial glycolate/glycerate antiporter PLGG1 (AT1G32080) (Fig. S3a,b). Indeed, PLGG1 had a slightly elevated transcript level in hpr1-1, which immediately declined at eCO_2_ (Fig. S3b). Although glyoxylate levels were not directly measured, the accumulation of cytosolic glycolate was consistent with the induction of cytosolic glyoxylate reductase GLYR1 (AT3G25530; log_2_FC 0.33, adj. P-val 0.008), suggesting enhanced conversion of glyoxylate to glycolate. This reaction would replenish the cytosolic glycolate pool and mitigate the accumulation of the toxic glyoxylate, as previously reported by Jiang et al. (2025). In search of a cytosolic source for glycerate fueling this exchange, the pathways depicted in Fig. 6a were scrutinized at the various omics levels.Fig. 6. Differential responses of the hpr1-1 relative to wildtype across different time points at eCO_2_. a Schematic representation of gluconeogenesis-dependent and HPR2-mediated bypasses of peroxisomal HPR1 deficiency in the hpr1-1 mutant at aCO_2_. (In)direct conversions and/or inductions of pathways are indicated by solid arrows, mutant deficiency is indicated by a red solid line, transfer of metabolites across cellular compartments is indicated by dashed lines. b Integrated heatmap of differentially expressed transcripts and proteins associated with hypothesized pathways in (a). Values represent log_2_FC of transcriptome (T) and proteome (P) in mutant relative to wildtype, thus, H0vsC0, H1vsC1, and H7vsC7 indicate mutant-versus–wildtype comparisons at TP0, TP1, and TP7, respectively. (* adj. P-val \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$<$$\end{document} 0.05; n = 3; one-tailed z-test; Bonferroni correction). c Subcellular metabolite profiles represented as ridge plots comparing mutant relative to wildtype. Colors indicate duration at eCO_2 (purple: 0; gray: 1; green: 7 days). Abscissae reflect log_2-transformed ratio of mean metabolite concentrations (hpr1-1/Col-0), cytosolic, plastidial and mitochondrial metabolites are indicated by ‘C-’, ‘P-’ and ‘M-’ prefix, respectively, and ordinate indicates time points. Asterisks (*) along with dark colors denote statistically significant differences between time points within each genotype, based on two-way ANOVA followed by Bonferroni multiple comparisons test (n = 3). Faded color ridges indicate non-significant (NS) changes. 2-PG 2-phosphoglycolate, 2-PGA 2-phosphoglycerate, 3-PGA 3-phosphoglycerate, Ala alanine, Asn asparagine, Asp aspartic acid, Cys cysteine, ETH ethylene, Gln glutamine, Glu glutamic acid, Gly glycine, Glyce glycerate, Glyco glycolate, Glx glyoxylate, HCN hydrogen cyanide, Hpyr hydroxypyruvate, KSM α-ketosuccinamate, Mal malate, Met methionine, OAA oxaloacetate, PEP phosphoenolpyruvate, Pyr pyruvate, SAMC s-adenosyl methionine cycle, Ser serine
As reported by Timm et al. (2011) and Jiang et al. (2025), hydroxypyruvate can be reduced by either peroxisomal (HPR1) or cytosolic isoform of HPR (HPR2), the latter contributing to the cytosolic glycerate pool. Here, we propose an alternative pathway branching from serine (Fig. 6a,b). This route to glycerate would involve malate, pyruvate and cysteine, thus explaining the high cysteine levels in cytosol, plastids and mitochondria of the mutant (Fig. 6c). Key enzymes in this bypass route are O-acetylserine (thiol) lyase (OASA1, OASB/C), serine acetyltransferase (SAT1/2/3) and also L-cysteine desulfhydrase 1 (DES1), which catalyzes the conversion of cysteine to pyruvate thereby releasing ammonia and sulfide. All were up-regulated at the transcript level (OASA1, OASB/C, SAT1/2/3, DES1 in Fig. 6b). Acetyl-CoA, required for acetylation of serine, could originate from either citrate via the activity of ATP-citrate lyase (ACL) or acetate, depending on the compartment of cysteine synthesis. Cytosolic citrate was found to accumulate in hpr1-1 (Fig. 6c), and transcripts for enzymes related to ATP citrate-lyase and cytosolic acetyl-CoA synthesis, (ACLA-1, ACLB-1/2, ACS and AEE7) were found to be up-regulated, too (Fig. 6b).
In accordance with the proposed pathway, the flux of cytosolic pyruvate into mitochondria should be reduced. Indeed, mitochondrial pyruvate carriers 1, 3 and 4 (MPC1/3/4; Fig. 6b) displayed lower transcript levels in hpr1-1 compared to Col-0. In line with this, the transcript of pyruvate dehydrogenase kinase (PDK), an inhibitor of the pyruvate dehydrogenase complex, was significantly up-regulated (Fig. S3c). As outlined in Fig. 6a, malate could serve to replenish glycerate via gluconeogenesis. Depending on the malic enzymes, the reaction could proceed via pyruvate and phosphoenolpyruvate (PEP), employing NADP-ME2/4 and pyruvate kinases (PK and PK9), or via oxaloacetate and its conversion to PEP, involving malate dehydrogenase (MDH1) and phosphoenolpyruvate carboxykinase (PCK1). Both pathways are consistent with the observed accumulation of cytosolic malate in hpr1-1 (Fig. 6a–c). The gluconeogenic route is supported by the significant induction of genes involved in the conversion of PEP to 2-phosphoglycerate (2-PGA; ENO1/2), followed by its transformation to 3-phosphoglycerate (3-PGA; IPGM1/2), and ultimately to glycerate via glycerate kinase (GLYK) (Fig. 6a,b).
Comparative analysis of differentially expressed genes (DEGs) between the mutant and wildtype under aCO_2_ (H0vsC0_T) revealed significant induction of genes involved in asparagine metabolism. These genes are known to be regulated in response to cellular sucrose levels via sugar starvation markers such as SNRK1.1 and SNRK1.2 (Curtis et al. 2018). In hpr1-1 under aCO_2_, sucrose starvation could be sensed by these markers, which would then lead to the observed induction of genes associated with asparagine synthesis (ASN1/2/3). Asparagine biosynthesis through this pathway relies on aspartic acid, which must be provided via transamination of oxaloacetate. Indeed, the genes for aspartate aminotransferase (ASP1/2/3/5) were found to be induced in the mutant (Fig. 6b).
In addition, induction of genes involved in the methionine cycle (MS1/2/3 and SAMS1/2/3) and ethylene biosynthesis (ACO1/2/4) was observed under aCO_2_ in hpr1-1. Ethylene biosynthesis is associated with the production of the toxic by-product hydrogen cyanide, which is predominantly detoxified in a two-step reaction starting from cysteine. The reaction is catalyzed by β-cyanoalanine synthase (CYSD1/2 and CYSC1) and results in the formation of the less toxic β-cyanoalanine. This compound is subsequently converted to asparagine or aspartate by nitrile hydratase (NIT4), providing an alternative source of asparagine. All genes involved in this pathway, particularly NIT4, were up-regulated at aCO_2_ in hpr1-1 (Fig. 6a,b).
Discussion
Response of wildtype Col-0 plants to a sudden rise in CO2 concentration: symptoms of stress
Besides immediate stimulation of primary carbon fixation, an increase in atmospheric carbon dioxide levels has profound consequences for nitrogen and sulfur assimilation, resulting in deflections in amino acid metabolism (Chen et al. 2005; Abadie and Tcherkez 2019). It has been reported that a decline in photorespiratory activity at eCO_2_ could be responsible for reduced N- and S-acquisition (Abadie and Tcherkez 2019; Krämer et al. 2022b), and thus, in the present study, we aimed at disentangling the effects of elevated C-assimilation and reduced photorespiratory activity at eCO_2_ by shifting wildtype plants and the hpr1-1 mutant, which is impaired in recycling carbon and nitrogen released in photorespiration, from ambient to elevated CO_2_ levels. It should be emphasized that a CO_2_ concentration of 1000 ppm does not completely abolish photorespiration, as stated earlier by Krämer et al. (2022b), but substantially reduces its turnover.
In both genotypes, sugars accumulated immediately at eCO_2_, while total amino acids rose in wildtype and declined in the mutant. In fact, responses were not only different among genotypes but also among individual amino acids, especially when their subcellular localization is considered. While an increase in plastidial glutamate and glutamine in both genotypes pointed to a transient stimulation of N-assimilation, possibly linked to the high sugar levels that activate nitrate reductase, methionine levels went down in plastids, but increased in the cytosol of Col-0 and remained unchanged in the cytosol of hpr1-1. Thus, methionine accumulation in the cytosol of Col-0 appears to be related to the stress symptoms, displayed by the wildtype but not by the hpr1-1 mutant.
It has earlier been reported that eCO_2_ causes increased transcript levels of disease resistance proteins, heat shock proteins and dehydration-responsive proteins (Li et al. 2008), but the underlying regulatory mechanism is unclear. High sugar levels sensed by the bifunctional enzyme/sugar sensor hexokinase 1, which is involved in various biotic and abiotic stress responses (Shen et al. 2021; Yu et al. 2022) could have triggered this response. The HXK1 gene is itself responsive to elevated glucose levels (Pourtau et al. 2006) and was up-regulated in Col-0 upon eCO_2_. The observed repression of the genes LHCB2 and CA2, both involved in photosynthesis, is in accordance with this view. It has been shown by Cho et al. (2006) that a trimeric complex of HXK1, RPT5B and VHA-B1 in response to elevated glucose binds to the promoter region of chlorophyll a-b binding proteins of PSII (LHCB) to inhibit transcription and the observed down-regulation of the RBCS gene is also in agreement with the reported regulation by HXK1 (Xu et al. 2015). Considering that sugar levels increased in both genotypes, but stress indicators were up-regulated only in the wildtype, this indicates that sugar accumulation alone is insufficient to trigger the stress response.
What discriminates the two genotypes, was the strong evidence for a disturbed linear electron transport in Col-0 upon the shift to eCO_2_. We observed repression of the PETC gene, encoding the Rieske-type FeS protein that transfers electrons from plastoquinol to Cytb6f, as well as both genes encoding plastocyanin. PETC was shown to be repressed by elevated sucrose levels (Knight et al. 2002) and was significantly down-regulated at day 1 of the eCO_2_ treatment. The same is true for the plastocyanin gene, PETE (Dijkwel et al. 1996). The observed stimulation of the violaxanthine de-epoxidase gene VDE1, involved in the violaxanthin-antheraxanthin-zeaxanthin (VAZ) cycle (Czarnocka and Karpiński 2018) and induction of genes involved in the Foyer-Halliwell-Asada cycle of reactive oxygen defense, including SOD and APX (Noctor and Foyer 1998), as well as the increase in non-photochemical quenching and F0 as an indication of oxidative damages to PSII (Baker 2008), shown in Fig. 2c, support this view. A block in photosynthetic electron flux could also have triggered repression of LHCB2 and CA2 gene expression (Oswald et al. 2001).
The contrasting light-dependent photosynthetic responses of Col-0 and hpr1-1 to eCO_2_ reveal fundamentally different strategies at the photochemical and biochemical level. In Col-0, the sustained decline in Fv/Fm following transfer to eCO_2_ that persisted even after seven days, together with increased F0 and NPQ, indicates chronic photoinhibition and PSII damage rather than a transient down-regulation (Baker 2008). This interpretation is reinforced by the strong and lasting reductions in ΦPSII and J. Similar reductions in J and ΦPSII under eCO_2_ have been reported for soybean, wheat, and woody species, where photosynthetic acclimation was involved in the repression of electron transport capacity and altered excitation energy dissipation (Zheng et al. 2019; Schweiger et al. 2022; Tang et al. 2024). Our omics data provide a mechanistic framework for this response, where repression of PETC, downregulation of LHCB2 and CA2, and induction of VDE1 and antioxidant pathways collectively point to restricted linear electron flow and enhanced photoprotective demand. Such repression is also consistent with carbohydrate-mediated signaling via HXK1. In contrast, hpr1-1 displayed increased Fv/Fm, ΦPSII, and J under eCO_2_, suggesting improved photochemical efficiency and protection of PSII. The induction of VDE1 and antioxidant enzymes in hpr1-1 supports enhanced photoprotective capacity via the xanthophyll cycle and reactive oxygen species detoxification, consistent with improved redox balance under carbon-replete conditions (Noctor and Foyer 1998; Czarnocka and Karpiński 2018).
The FvCB modeling revealed the mechanistic basis of this divergence. In Col-0, the immediate and progressive decline in Jmax without a corresponding change in Vcmax following eCO_2_ exposure indicated a shift towards electron transport limitation, with Aj and Wj becoming strongly limiting relative to Ac and Wc. Such a down-regulation has been widely observed across C_3_ species grown under elevated CO_2_ for long periods, where both Vcmax and Jmax typically decline relative to ambient conditions, reflecting reduced investment in photosynthetic machinery under high carbon availability (Zheng et al. 2019; Ancín et al. 2024). This decoupling of photochemistry from carbon metabolism is consistent with reduced NADPH and ATP consumption rates and points to sink limitation and feedback inhibition of photosynthesis.
In contrast, hpr1-1 exhibited coordinated increase in both Vcmax and Jmax upon the shift to eCO_2_, maintaining close coupling between Aj and Ac, despite its inherently higher photorespiratory burden under aCO_2_ (Fig. 3c). This suggests that enhanced photorespiration in hpr1-1 under aCO_2_ may prime electron sinks and redox buffering capacity, allowing more effective utilization of excess reducing power when CO_2_ becomes non-limiting. The stability of Vcmax and Jmax after prolonged eCO_2_ exposure, together with sustained NADPH and ATP consumption, indicates superior prearrangement of hpr1-1 to the eCO_2_ condition. This exceptional genotype response to eCO_2_ contrasts with theoretical and empirical predictions that plants under eCO_2_ tend to underinvest in electron transport relative to carboxylation (Smith and Keenan 2020). Although mesophyll conductance declined in both genotypes following eCO_2_ exposure, this limitation did not translate into photochemical or biochemical down-regulation in the mutant, highlighting that the failure of the wildtype is not based on diffusional constraints, but is regulatory. Together, these findings support a model in which HXK1-mediated sugar signaling, coupled with impaired electron transport and insufficient photoprotection, drives photoinhibition in Col-0, whereas hpr1-1 maintains photosynthetic balance through reinforced electron sinks, antioxidant capacity, and metabolic flexibility.
Elevated CO_2_ is frequently reported to alleviate photoinhibition by enhancing carbon assimilation and increasing the availability of metabolic electron sinks; however, this response is not universal and a growing body of evidence indicates that responses to eCO_2_ are highly conditional and frequently accompanied by photosynthetic downregulation. Hymus et al. (2001) demonstrated that growth under eCO_2_ can either increase or decrease photochemical efficiency and susceptibility to photoinhibition, depending on nitrogen availability and possible sink limitation. In agreement with this, our data indicate that in wildtype plants, eCO_2_ induced a sustained block in electron transport and chronic photoinhibition, resulting in decreased Fv/Fm, reduced ΦPSII and J, and enhanced photoprotective dissipation. By contrast, hpr1-1 maintained coordinated electron transport and carbon assimilation under eCO_2_, supporting the conclusion that genotype-specific regulation of sink capacity and feedback signaling, rather than CO_2_ concentration alone, determine the balance between electron utilization and photoprotection. Multiple studies have shown that exposure to eCO_2_ can reduce Vcmax and/or Jmax in diverse species, including forest trees, crops, and legumes, largely due to nitrogen dilution, sink limitation, and feedback inhibition by accumulated carbohydrates (Medlyn et al. 1999; Arora et al. 2009; Zheng et al. 2019; Sugiura et al. 2024). Consistent with these reports, our FvCB analysis revealed that in wildtype, exposure to eCO_2_ caused a rapid and sustained decline in Jmax without a concomitant increase in Vcmax, resulting in a shift towards electron transport limitation (Aj \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$<$$\end{document} Ac; Wj \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$<$$\end{document} Wc). This supports the concept that eCO_2_ can exacerbate imbalances between photochemical energy provision and downstream carbon metabolism when sink capacity or regulatory coordination is insufficient.
If, indeed, a redox imbalance would have caused the induction of defense-related genes in the wildtype, it would be difficult to separate this effect from a decline in photorespiratory activity, which would, by eliminating an important consumer of reduction equivalents, also result in a disturbance of the redox balance (Voss et al. 2013; Sunil et al. 2019). In this context, the observed stimulation of amino acid production in Col-0 immediately upon the shift to eCO_2_ likely serves as an alternative sink for excess reducing equivalents that are no longer utilized by photorespiration. While shikimic acid levels were reduced, aromatic amino acids accumulated in the cytosol of wild-type plants, and the rise in tryptophan could be related to indolic glucosinolate metabolism, being part of an immune response that was significantly up-regulated at the transcript level. In this context, positive feedback driven by elevated glucose on the induction of glucosinolate biosynthetic genes via a HXK1-dependent association with MYB transcription factors has been demonstrated by Eom et al. (2024). Induction of genes involved in glucosinolates metabolism under eCO_2_ was shown in detail by Landosky and Karowe (2014) and Wiesner-Reinhold et al. (2021). Interestingly, while methionine accumulated in the cytosol of wildtype, cysteine did not change in any of the compartments. Thus, although we observed the previously reported reduction in sulfate content at eCO_2_ (Loladze 2014), we did not find a negative impact of reduced photorespiration on sulfur-containing amino acids as reported for sunflower (Abadie and Tcherkez 2019). Note, however, that we measured steady state levels of amino acids, but not de novo synthesis, as reported by Abadie and Tcherkez (2019). The rise in methionine is consistent with the reported induction of aliphatic glucosinolate metabolism at eCO_2_ (Almuhayawi et al. 2020; AbdElgawad et al. 2023). Methionine, as the precursor for S-adenosylmethionine, is also required for ethylene biosynthesis which is up-regulated at the level of ACC-oxidase 1 (ACO1) in Arabidopsis at eCO_2_ (Smet et al. 2020; Azoulay-Shemer et al. 2023). During ethylene formation, cyanide is released by ACC-oxidase, which is then detoxified in a two-step reaction yielding either asparagine or aspartate (Machingura et al. 2016). The second step is catalyzed by a bifunctional cytosolic nitrilase/nitrile hydratase (NIT4), and both products, asparagine and aspartate, were increased at day 1 of the eCO_2_ treatment in wild-type plants.
The sustained down-regulation of genes involved in photosynthesis and the electron transport chain under long-term eCO_2_ exposure may result either from carbohydrate accumulation and reduced sucrose loading into the phloem, or from eCO_2_-dependent closure of stomata and altered internal CO_2_, which has been reviewed in detail by Foyer et al. (2025). Stomatal closure exacerbates the over-reduction of the plastoquinone pool, ultimately disrupting the electron transport chain, as demonstrated by Wang et al. (2016). In addition, sustained stomatal closure has been shown to impair cyclic electron flow and intensify the over-reduction of PSI, as reported by Li et al. (2021). Taken together, the results obtained for wildtype plants gave clear evidence for a stress response, thus extending earlier work on gene expression under eCO_2_ (Smet et al. 2020).
The hpr1-1 mutant shows superior prearrangement to eCO2
When comparing responses to sudden eCO_2_ of wildtype and hpr1-1 mutant plants, considering the diverse initial condition of both genotypes is vital. The hpr1-1 mutant is significantly retarded in growth, has very high levels of serine and glycine because of the shortage in photorespiratory turnover, but is also high in threonine, isoleucine, cysteine and arginine, and low in alanine and aspartate levels [this study and Timm et al. (2021)]. Nevertheless, sugar profiles responded similarly to eCO_2_ in both genotypes, and carboxylates showed hpr1-1 to approximate wildtype levels over the time course of 7 days. For example, malate levels that deviated strongly at aCO_2_ for cytosol and mitochondria were equal at day 7, as well as mitochondrial fumarate. However, malate in plastids of hpr1-1 was always depleted. Given its role as a redox shuttle, this indicates a strongly deviating redox situation in the mutant and, as discussed above, may be the reason for the lack of stress symptoms in hpr1-1.
While accumulation of glycerate in the cytosol of hpr1-1 could stem from the non-canonical photorespiratory route catalyzed by cytosolic HPR2 (Jiang et al. 2025), we propose an alternative pathway depending on pyruvate. This pathway would profit from high cysteine, but also malate levels in hpr1-1. Thus, the combined omics analysis applied here points to an alternative route of glycerate regeneration under photorespiratory conditions, which could make a substantial contribution in the hpr1-1 mutant. The shift to eCO_2_ caused a rapid decline in cytosolic pyruvate, malate, glycolate and glycerate, but also in threonine and isoleucine, while aspartic acid increased and reached levels similar to wildtype. It can thus be concluded that eCO_2_ caused substantial re-arrangement in the aspartate family of amino acids, which appears indirectly affected by high glycine and serine levels in the mutant at aCO_2_. The high threonine and isoleucine levels found in hpr1-1 at aCO_2_ could originate from a reduced flux from threonine to glycine via threonine aldolase (Lu et al. 2008), as well as from high serine levels (Timm et al. 2021).
We observed induction of asparagine synthesis genes in hpr1-1 in accordance with an SNRK-dependent response to sugar starvation at aCO_2_ (Baena-González et al. 2007; Curtis et al. 2018). However, asparagine could also be involved in a reaction by-passing HPR1. Although it is not possible, solely based on transcript data, to distinguish the various substrate preference of serine-glyoxylate aminotransferase (AGT1), which, besides asparagine, also includes serine and alanine, the significant up-regulation of an upstream \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\omega$$\end{document} -amidase gene suggests that asparagine could act as a key metabolite in the turnover of accumulated photorespiratory intermediates in hpr1-1. Zhang et al. (2013) showed that catalytic efficiency and Vmax of AGT1 were significantly higher for asparagine compared to the other substrates, and Modde et al. (2017) suggested that asparagine, as an additional amino group donor, is a superior substrate in metabolizing photorespiratory intermediates. A role of asparagine in glyoxylate detoxification has also been discussed by Cooper et al. (2022). It is interesting to note that, in the reaction involving \documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\omega$$\end{document} -amidase, malate and oxaloacetate pools are replenished, which are involved in the hypothesized gluconeogenic pathway of glycerate regeneration.
Taken together, our combined and subcellularly resolved omics approach revealed additional, new information on possible metabolic by-passes to peroxisomal HPR (HPR1) and assisted to elucidate non-canonical phenotypes of the hpr1-1 mutant. It clearly shows that metabolome analysis at the whole cell level reaches limits when it comes to pathways, such as photorespiration, which are shared among various cellular compartments.
Supplementary Information
Below is the link to the electronic supplementary material.Spectral composition of LED light used in all growth chambers. The X-axis shows the wave length of the growth light, while the Y-axis shows intensity as arbitrary units. The composed light has a total power of 41.47 W. Supplementary file1 (PNG 15 KB)Subcellular metabolite profiles across time points and genotypes. Concentrations are shown as the mean of biological replicates (n = 3), represented as stacked bar plots.colors indicate compartments (red: cytosol; green: mitochondrion; blue: plastid; purple: vacuole). Abscissae reflect time points and ordinates reflect metabolite concentration in μmol g^−1^ DW. Supplementary file2 (PDF 64 KB)Dynamics of transcript levels for plastidial BASS6 and PLGG1 transporters and mitochondrial pyruvate dehydrogenase kinase (PDK) in hpr1-1 over various CO_2_ conditions. a Log_2_FC of BASS6 (AT4G22840, plastidal sodium/metabolite cotransporter), betweenhpr1-1 and Col-0. b Log_2_FC of PLGG1 (AT1G32080, plastidal glycolate/glycerate translocator 1) between hpr1-1 and Col-0. c Log_2_FC of PDK (AT3G06483, mitochondrial pyruvate dehydrogenase kinase) between hpr1-1 and Col-0. (* adj. P-val < 0.05; n= 3; one-tailed z-test; Bonferroni correction). Supplementary file3 (PNG 2050 KB)Differentially expressed genes (DEG) analysis represented as dot plots and cnet plots comparing time point or genotypes.According to plot titles,C1vsC0 and C7vsC0 are comparisons for Col-0, TP1 vs TP0 and TP7 vs TP0, respectively. H1vsH0 and H7vsH0 are comparisons for hpr1-1 accordingly. H0vsC0, H1vsC1 and H7vsC7 are comparisons of hpr1-1 relative to Col-0 at respective time points. Dot plot showing enriched Gene Ontology (GO) terms (Biological Process) of DEGs under elevated CO_2_ conditions relative to ambient. The ordinate represents enriched GO categories, abscissae show gene ratio (DEGs associated with each term divided by total DEGs). Dot color reflects adj. P-val and dot size corresponds to the number of DEGs associated with each GO term. Cnetplot (category network plot) visualize the associations between DEGs and enriched GO terms. Nodes represent loading transcripts with differential expression and size represents the number of genes per GO category, gene node color indicates log_2_FC of expression under eCO_2_ versus ambient, and different colors of GO term nodes distinguish among the GO categories. The plot highlights how shared DEGs contribute to multiple biological processes. Supplementary file4 (PDF 732 KB)Differentially expressed genes (DEG) analysis illustrated as volcano plots and Venn diagrams and statistical representation of FvCB modeling. a DEG analysis of eCO_2_ conditions relative to ambient in wildtype and mutant (C1vsC0_T, H1vsH0_T, C7vsC0_T and H7vsH0_T) and mutant relative to wildtype across various time points (H0vsC0_T, H1vsC1_T and H7vsC7_T). Each point represents a gene with abscissae reflecting log_2_FC and ordinates as negative log_10_ of the adj. P-val. Thresholds for significance were set at adj. P-val < 0.05 (−log_10_ (0.05) ≈ 1.3). Red and blue colors are indicating up-and down-regulation, respectively. b The number of shared transcripts across comparisons is illustrated by Venn diagrams, shown separately for up- and down-regulated genes. The complete list of overlapping transcripts is provided in Table S5. c Measured net CO_2_ assimilation rates (Ameas) plotted against FvCB model-predicted assimilation rates (Amodel) for Col-0 and hpr1-1 under a/eCO_2_ condition. Colors indicate duration at eCO2 (purple: 0; gray: 1; green: 7 days), and points represent individual measurements. Linear regressions fitted separately for each genotype and time point, with shaded areas denoting 95 % confidence intervals. The solid-colored diagonal line represents the 1:1 relationship. Regression intercepts, slopes, coefficients of determination (R²), and their associated standard deviations (± SD) are shown for each panel, providing a quantitative assessment of model performance across CO_2_ treatments and genotypes. Supplementary file5 (PNG 10095 KB)Transcripts and proteins GO term enrichment of the top 1000 genes/proteins according to positive loading of PC1-3. PCA loadings comprising transcripts, proteins and subcellular metabolites of Fig. 1. Supplementary file6 (XLSX 1099 KB)Gene identifiers (AGI) and GO term enrichment of clustered proteins of Fig. 4. Supplementary file7 (XLSX 126 KB)Details of differentially expressed genes (DEGs) analysis and their GO ontologies of Fig. S4. Supplementary file8 (XLSX 3720 KB)Details of differentially expressed proteins (DEPs) analysis and their GO ontologies. Supplementary file9 (XLSX 1965 KB)Shared DEGs between respective comparisons in Venn diagrams of Fig. S5b (Venn3 and Venn4). Supplementary file10 (XLSX 158 KB)Raw data of individual omics analysis, including subcellular metabolomics, transcriptomics and proteomics. Raw metabolites data are representing absolute subcellular concentrations of individual metabolites in plastid, cytosol, mitochondrion and vacuole (in μmol g^−1^DW), raw transcripts data is based on log_2_TPM (transcripts per million) and raw proteomes data are based on log_2_lfq. measurements across genotypes and time points under ambient and elevated CO_2_ conditions and batch as replicates (n= 3). Supplementary file11 (XLSX 9657 KB)Input dataset of FvCB modeling (A-Ci), raw data of physiological assessments, including chlorophyll fluorescence, gas exchange, statistical analysis and phloem exudation. Data represent measurements across genotypes and time points under ambient and elevated CO_2_ conditions. Supplementary file12 (XLSX 266 KB)List of whole TAIR locus IDs versus genes name and their categories, used in manuscript. Supplementary file13 (XLSX 14 KB)
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1Smith NG, Keenan TF (2020) Mechanisms underlying leaf photosynthetic acclimation to warming and elevated CO 2 as inferred from least‐cost optimality theory. Abstr Global Change Biol 26(9):5202–5216. 10.1111/gcb.v 26.910.1111/gcb.15212