Age‐dependent increase in α‐tocopherol and phytosterols in maize leaves exposed to elevated ozone pollution

Abstract Tropospheric ozone is a major air pollutant that significantly damages crop production. Crop metabolic responses to rising chronic ozone stress have not been well studied in the field, especially in C4 crops. In this study, we investigated the metabolomic profile of leaves from two diverse maize (Zea mays) inbred lines and the hybrid cross during exposure to season‐long elevated ozone (~100 nl L−1) in the field using free air concentration enrichment (FACE) to identify key biochemical responses of maize to elevated ozone. Senescence, measured by loss of chlorophyll content, was accelerated in the hybrid line, B73 × Mo17, but not in either inbred line (B73 or Mo17). Untargeted metabolomic profiling further revealed that inbred and hybrid lines of maize differed in metabolic responses to ozone. A significant difference in the metabolite profile of hybrid leaves exposed to elevated ozone occurred as leaves aged, but no age‐dependent difference in leaf metabolite profiles between ozone conditions was measured in the inbred lines. Phytosterols and α‐tocopherol levels increased in B73 × Mo17 leaves as they aged, and to a significantly greater degree in elevated ozone stress. These metabolites are involved in membrane stabilization and chloroplast reactive oxygen species (ROS) quenching. The hybrid line also showed significant yield loss at elevated ozone, which the inbred lines did not. This suggests that the hybrid maize line was more sensitive to ozone exposure than the inbred lines, and up‐regulated metabolic pathways to stabilize membranes and quench ROS in response to chronic ozone stress.

billion per year (McGrath et al., 2015). While much is known about the metabolic and signaling responses within plant cells to elevated [O 3 ], major knowledge gaps remain with regard to: (a) the mechanisms underlying genetic variation in O 3 response, and (b) the nature of biochemical O 3 responses in the production environment of a farm field (Ainsworth, 2017). Additionally, our understanding of plant metabolic responses to O 3 stress largely comes from acute experiments in controlled environments (Carmody et al., 2016;Vainonen & Kangasjärvi, 2015), yet it is well known that the mechanisms of response to chronic O 3 exposure, generally defined as long-term exposure to concentrations of ~100 nl L −1 or less, differ from acute responses to very high [O 3 ] (Grantz & Vu, 2012;Vahala et al., 2003).
Increased emissions of precursor pollutants have primarily increased crop exposure to chronic O 3 stress, with tropospheric [O 3 ] increasing from ~10 nl L −1 in the late 1800's to ~40 to 50 nl L −1 in recent years (Brauer et al., 2016;Monks et al., 2015). In many countries, [O 3 ] continue to increase (Brauer et al., 2016), further intensifying chronic O 3 exposure.
Ozone diffuses through the stomata into the intercellular airspace where it rapidly reacts to form additional reactive oxygen species (ROS). It is also a powerful oxidizing agent capable of reacting with diverse molecules including lipids, proteins, nucleic acids, and carbohydrates. ROS formed from O 3 further react with apoplastic antioxidants and a number of proteins embedded in the plasma membrane (e.g., NADPH oxidases, aquaporins, receptor-like kinases, and G-proteins) and elicit an increase in cytosolic calcium (Short et al., 2012). Peroxidation and denaturation of membrane lipids can also occur with prolonged or acute O 3 exposure (Loreto & Velikova, 2001;Pell et al., 1997). The intracellular response to the influx of ROS depends upon the duration and intensity of O 3 exposure, with calcium, hormone signaling, and MAP kinase cascades all playing a role in regulation of transcriptional and biochemical changes (Vainonen & Kangasjärvi, 2015). Under chronic O 3 stress, transcriptional changes have been associated with decreased photosynthesis (Leitao et al., 2007;Li et al., 2019;Pell et al., 1997), increased rates of mitochondrial respiration and antioxidant production (Gillespie et al., 2012;Yendrek et al., 2015), increased hormone biosynthesis (jasmonates, ethylene, and salicylic acid) (Kangasjarvi et al., 2005;Vainonen & Kangasjärvi, 2015), and early activation of several senescence-associated genes (SAGs) (Miller et al., 1999;Fiscus et al., 2005;Kangasjarvi et al., 2005;Betzelberger et al., 2012;Gillespie et al., 2012;Yendrek et al., 2013). For example, LIGHT HARVESTING COMPLEX B6 (LHCB), encoding a minor subunit of the antennae complex responsible for transfer of light energy to photosystem II reaction centers, is an early marker of senescence (Breeze et al., 2011) and its expression is down-regulated early in plants exposed to elevated [O 3 ] (Yendrek et al., 2013). Other markers of senescence including LONG-CHAIN ACYL-COA SYNTHETASE 6 and AURORA2 were also accelerated in plants grown at elevated [O 3 ] (Yendrek et al., 2013). Shifts in metabolism from carbon assimilation to defense and detoxification in combination with early senescence are thought to be cumulative drivers of reduced plant productivity under elevated [O 3 ] (Choquette et al., 2019(Choquette et al., , 2020Dizengremel, 2001;Feng et al., 2011;Morgan et al., 2006;Yendrek, Erice, et al., 2017).
A wide range of metabolites have been reported to change in response to elevated [O 3 ] in different species, many in common with plant defense responses (Iriti & Faoro, 2009;Munne-Bosch et al., 2013). Plant steroids, including phytosterols and brassinosteroids (BRs), increase plant tolerance to a wide range of abiotic and biotic stresses as well as control plant growth, flowering time, and senescence (Vriet et al., 2012). Changes in the ratios of membrane steroids during stress is also commonly reported following abiotic and biotic stress treatments (Rogowska & Szakiel, 2020). For example, the ratio of stigmasterol to β-sitosterol increased following exposure of Arabidopsis to pathogen-associated molecular patterns and ROS, which was thought to help maintain plasma membrane fluidity and permeability during stress (Griebel & Zeier, 2010). In the chloroplasts, α-tocopherol is an important scavenger that protects photosynthetic machinery by quenching singlet oxygen or inhibiting the progression of lipid peroxidation (Havaux et al., 2005).
Exposure to many abiotic stresses that increase ROS results in increased α-tocopherol content in leaves (Munne-Bosch, 2005). As leaves age, α-tocopherol content also increases (García-Plazaola & Becerril, 2001;Hormaetxe et al., 2005). It is unknown if these metabolites play a role in O 3 response under chronic exposure and field conditions. If so, breeding or biotechnology might be used to leverage the protective roles of phytosterols and non-enzymatic antioxidants to improve tolerance to O 3 -induced oxidative stress.
More broadly, metabolomics provides a tool to explore biochemical signatures that may be predictive of environmental stress effects on primary productivity, even in field conditions. Studies have examined the relationship between leaf metabolites and physiological traits in the field under various abiotic stress conditions in maize (Obata et al., 2015;Riedelsheimer et al., 2012), rice (Melandri et al., 2020), and Guinea grass (Wedow et al., 2019). Additionally, metabolomics has been used to identify markers associated with greater yield potential in maize (Cañas et al., 2017). But, metabolomic responses of maize to elevated [O 3 ] have not yet been widely characterized despite that maize is one of the world's most widely grown crops (USDA FAS, 2020). Maize is also a model species for the C 4 plant functional group that includes many other important crops used to produce food, fuel, forage, and fiber. Therefore, we examined the metabolomic profile of maize inbred (B73 and Mo17) and hybrid  (Hong et al., 2019).
In this study, leaf metabolomic profiles were investigated at three time points from early to mid-stages of leaf senescence as characterized by chlorophyll content. Metabolite content was then linked to leaf mass per unit area and grain yield to identify potential biochemical markers for O 3 response in maize. Ozone tolerance at the leaf level has been linked to dry mass per unit leaf area (LMA) (Feng et al., 2018), which is correlated with seasonal changes in photosynthetic capacity in maize (Miner & Bauerle, 2019). Photosynthetic capacity is also highly sensitive to O 3 and correlated with yield in maize (Choquette et al., 2020). In this study, we investigated B73 and Mo17, which are two classic elite inbred maize lines (Stuber et al., 1992), and serve as the parents for widely used mapping populations to study the genetic architecture of numerous traits (Balint-Kurti et al., 2007;Mickelson et al., 2002;Pressoir et al., 2009;Sorgini et al., 2019;Wassom et al., 2008). We aimed to test the hypotheses that (  Inbred and hybrid lines were grown in separate plots to avoid the taller hybrids altering the light environment or interfering with fumigation of shorter inbred lines, which resulted in the use of 16 octagonal, 20-m-diameter plots. One replicate pair of ambient and elevated [O 3 ] plots with the hybrid line was dropped from the analysis due to water logging. Each genotype was planted in five different locations within each plot ( Figure S1). Additional site information and field conditions were described in Yendrek, Erice, et al. (2017).

| Field site and experimental conditions
Air enriched with O 3 was delivered to the experimental rings with FACE technology as described in . The target elevated [O 3 ] was 100 nl L −1 and the O 3 treatment was administered from 10:00 to 18:00 throughout the growing season when it was not raining and the wind speed was greater than 0.5 m/s. Based

| Sampling protocol and tissue handling
Leaf material for metabolomic profiling of B73, Mo17, and B73 × Mo17 was sampled on three dates corresponding to similar leaf physiological maturity and early-and mid-senescence stages between the hybrid and inbreds. Samples were taken in the inbred Leaf mass per unit area (LMA) was measured from leaf disks taken at the same time as samples for metabolomic profiling. Three leaf disks (0.02 m dia) per row were cut with cork borers, placed into coin envelopes, and dried in an oven at ~60°C for 1 week. Samples were then weighed and data from the five rows within each plot were averaged for a plot-level estimation of LMA (g/m 2 ).

| Leaf chlorophyll content
Chlorophyll content was estimated from measurements collected with a SPAD meter (Konica-Minolta SPAD-502 Chlorophyll meter).

| Seed yield
At maturity, ears were harvested from 8 plants in each of the 5 rows per genotype per plot, dried for ~1 week in an oven at ~60°C, then shelled, and weighed to estimate yield (g plant -1 ). (Sigma-Aldrich) (40 mg/ml in pyridine) for 90 min at 50°C, then with 125 µl MSTFA + 1%TMCS (Thermo) at 50°C for 120 min followed by an additional 2-hr incubation at room temperature.

| Metabolomic profiling by GC-MS
An internal standard (30 µl hentriacontanoic acid) was added to each sample prior to derivatization. Samples were analyzed on a gas chromatography/mass spectroscopy (GC/MS) system (Agilent Inc) consisting of an Agilent 7890 gas chromatograph, an Agilent 5975 mass selective detector, and a HP 7683B autosampler. Gas chromatography was performed on a ZB-5MS capillary column (Phenomenex). The inlet and MS interface temperatures were 250°C, and the ion source temperature was adjusted to 230°C.
An aliquot of 1 µl was injected with the split ratio of 10:1. The helium carrier gas constant flow rate was 2.4 ml/min. The temperature program was 5 min isothermal heating at 70°C, followed by an oven temperature increase of 50°C/min to 310°C, and a final 10 min at 310°C. The mass spectrometer was operated in a positive electron impact mode at 69.9 eV ionization energy in m/z 50-800 scan range.
Raw data files were processed with the metaMS.GC workflow hosted on the workflow4metabolomics (W4M) server (Giacomoni et al., 2015). Default settings were used except for minimum class fraction, specified at 0.6. Spectra were normalized to the internal standard and leaf dry weight was used to account for tissue and water content differences over time. Batch correction was done with the all loess sample regression model, W4M tool. Peak annotation used a custom-built database and AMDIS 2.71 (NIST, Gaithersburg, MD, USA) program. All known artificial peaks were identified and removed. The instrument variability was within the standard acceptance limit of 5%.

| Statistical analysis
The analysis of chlorophyll loss over time was tested by fitting a quadratic equation to the data where: where x equals day of year. To test for differences in chlorophyll loss over time in ambient and elevated [O 3 ], a single quadratic model was first fit to the data for each genotype (PROC NLIN, SAS 9.4, SAS Institute), and then models were fit to each genotype and treatment combination. An F statistic was used to test if the model with genotype and treatment produced a significantly better fit to the data, that is, if there were significant differences in the response of LMA and seed yield were analyzed using analysis of variance.
For the inbred experiment, the model included fixed-effect terms for inbred line and treatment, and a random term for block. The model for the hybrid experiment included treatment as a fixed effect and block as a random term in the model. Significant differences between treatments were determined by Tukey tests with a threshold of p < .05.
The inbred and hybrid metabolite data were analyzed separately because they were grown in different field plots and harvested on different dates ( Figure S1). For each dataset, metabolite data were log 10 transformed and processed with univariate analysis to identify and remove outliers (studentized residual ≥4). The total number of removed outlier points within the inbred datasets was between 3 and 8 and between 0 and 2 for the hybrid datasets. The five observations for a genotype within each FACE or control plot were averaged for analysis and time points were analyzed separately. Each metabolite was tested independently using a two-way ANOVA (Kirpich et al., 2018) for the inbred experiment with treatment and genotype as fixed effects, and a one-way analysis of variance model for the hybrid experiment ( Figure S2). Statistical differences in least squared mean estimates between ambient and elevated [O 3 ] for each time point were determined by Tukey tests with a threshold of p < .05 (Kuehl, 2000). The statistical analysis was done using SAS software (SAS, Version 9.4).
Multivariate statistics were performed using R (version 3.5.1; The R Foundation for Statistical Computing). Multivariate clustering analysis was done with the log 10 -transformed and Pareto scaling normalized data, with identified outlier observations removed.
Missing values were estimated prior to multivariate analysis using k-nearest neighbor (KNN) in the MetaboAnalystR package (Chong & Xia, 2018). The total number of missing values within the inbred datasets was between 4.2% and 6.4% of all observations and between 4.8% and 6.4% for the hybrid datasets. Principal component analysis (PCA) was performed using the prcomp function (R stat package) for the inbred and hybrid experiments independently for each time point. A singular value decomposition data matrix was applied to each normalized dataset. When the unsupervised PCA identified a clear treatment separation, a supervised partial least square -discriminant analysis (PLS-DA) was performed, with the mixOmics package (Rohart et al., 2017). The number of latent variables included in the model was selected by testing the predictability value (Q 2 ) using an increasing number of latent variables from 1 to 10. The relative importance of the metabolites in the models was summarized using PLS-DA loadings, with significance considered when the contribution was greater than 0.1.
Pearson linear correlation analysis was done to investigate correlations between metabolite content and physiological traits (chlorophyll and LMA) based on the class separation observed in the multivariate clustering. Correlations with an adjusted p value (False Discovery rate, (Benjamini & Hochberg, 1995)) of .05 or less, and a correlation coefficient of absolute value |.55| or greater were con-  (Table S1), and 51, 49, and 56 metabolites at each time point in B73 × Mo17 (Table S2).  (Table S3). In contrast, the content of only three metabolites differed between leaves grown in ambient and elevated [O 3 ] in the inbred lines (Table S3) (Table S2).
The ratios of α-tocopherol, campesterol, stigmasterol, and sitosterol to chlorophyll content are used as indices of senescence (Li et al., 2017) and were investigated in the maize inbred and hybrid lines. Alpha-tocopherol, campesterol, and stigmasterol increased over time in B73 × Mo17, especially in elevated [O 3 ] (Figure 5a-c).
These trends were due to both the decline in chlorophyll (Figure 1) and an increase in α-tocopherol, campesterol, and stigmasterol con-  (Figure 5d). The inbred lines showed an age-dependent increase in α-tocopherol:chlorophyll ratio, but no effect of elevated (Figure 6).

| LMA and yield correlations with leaf metabolite content
LMA has been linked to variation in leaf physiology, plant growth strategy, resource investment, and O 3 tolerance across diverse species (Feng et al., 2018;Shipley et al., 2006). LMA was lower in the hybrid line compared to the inbred lines (Table 1) To assess the predictive potential for metabolite concentrations to reveal changes in physiology or yield, correlations between metabolite content and LMA or yield were tested separately for inbred and hybrid maize lines. The linear correlations for B73 and Mo17 were performed independently due to the genotypic separation identified with PCA ( Figure 2). There was no treatment separation at any time point and so correlations were done across O 3 treatments. In the inbred lines, itaconic acid measured at time point A was positively correlated with yield (Table S5) (Table S5).  (Table S6). Stigmasterol, campesterol, and ethanolamine contents measured in recently mature leaves (time point A) were negatively correlated with yield, while alanine was positively correlated with yield (

| D ISCUSS I ON
Field metabolomics can be a powerful approach for profiling the metabolite changes in plants in response to environmental stress (Melandri et al., 2020;Wedow et al., 2019). The physiological response of crops to O 3 pollution has been well studied, with elevated [O 3 ] decreasing carbon assimilation, accelerating senescence and cell death, and reducing economic yield (Ainsworth, 2017). However, the metabolite profile underpinning the response to O 3 in maize has not been investigated and could provide information for targets to  (Figures 3 and 2a). However, as chlorophyll was lost more rapidly at elevated [O 3 ] during senescence in the hybrid, a clear separation in metabolite profiles between ambient and elevated [O 3 ] was apparent (Figure 2b,c). The inbred genotypes had no discernible difference in their metabolomic profiles in ambient and elevated [O 3 ] at any of the time points, although B73 was very different from Mo17 (Figure 3). These results agree with our prediction that a metabolomic signature develops with the accumulation of O 3 damage in aging leaves and reflects altered biochemistry. Ozone-induced acceleration of senescence has been previously reported to impact grain yield losses due to a decrease in photosynthetic capacity and shortened leaf lifespan . A reduction in photosynthetic carbon assimilation in maize ear leaves during grain filling was greater in hybrid maize compared to inbred lines (Yendrek, Erice, et al., 2017 (Yendrek, Erice, et al., 2017), suggesting that differential stomatal sensitivity to O 3 was not a primary driver of differences between the hybrid and inbred lines.
intermediates, and specialized metabolites such as phytosterols and fatty acids (Figure 4). In the latter two time points, citric acid and malic acid contents were greater in elevated [O 3 ]. This trend is consistent with increased flux through the TCA cycle and greater rates of mitochondrial respiration in plants exposed to O 3 stress (Betzelberger et al., 2012;Dizengremel et al., 2012;Yendrek et al., 2015). The shift in metabolism from photosynthetic carbon assimilation to the TCA cycle and mitochondrial respiration supplies energy for repair and detoxification of the plant cells against oxidative damage (Dizengremel, 2001). The additional influence of We found that α-tocopherol increased to a greater extent as leaves aged in elevated [O 3 ] in the hybrid B73 × Mo17 (Figure 5ac), but not in either inbred line ( Figure 6). The protective function of tocopherols to preserve cell membrane integrity during the final stages of leaf development has been well documented (Falk & Munne-Bosch, 2010;Fryer, 1992;Lira et al., 2017). Leaf α-tocopherol content increases in response to abiotic stresses, such  Table S3. where highly reactive singlet oxygen ( 1 O 2 ) is formed, α-tocopherol is an essential scavenger protecting the thylakoid membranes against lipid peroxidation (Piller et al., 2014). In high light conditions, the rapid turnover of α-tocopherol and plastoquinone have been correlated with the increased turnover rate of the D1 protein within PSII, thereby protecting the photosynthetic process (Krieger-Liszkay & Trebst, 2006). Experimental studies exposing plants to oxidative stresses have reported conflicting results regarding α-tocopherol levels in leaf extracts; beech leaves in full sun showed a significant increase over shade leaves, along with an acceleration in leaf senescence (García-Plazaola & Becerril, 2001).
Snap bean showed a significant increase in α-tocopherol concentrations following exposure to elevated [O 3 ]; however, the total concentration of α-tocopherol was not correlated with differences in O 3 sensitivity between cultivars (Burkey et al., 2001). In contrast, spinach leaves exposed to elevated [O 3 ] were observed to have a decrease or no change in α-tocopherol in leaf extracts (Calatayud et al., 2003(Calatayud et al., , 2004. In this study, the response of leaf α-tocopherol content to elevated [O 3 ] varied with maize genotype and leaf age (Figures 5a and 6). The conflicting results from previous studies regarding effects of elevated [O 3 ] on leaf α-tocopherol content may be attributed to the timing of sampling and the progression of leaf senescence. We would not have identified a relationship between α-tocopherol and O 3 in B73 × Mo17 had samples only been taking at the initial time point (Figure 4a), but clearly the content increased as leaves aged in elevated [O 3 ]. Elevated α-tocopherol concentrations in leaves exposed to elevated [O 3 ] are potentially quenching ROS that form during the disassembly of the photosynthetic membranes during senescence and/ or inhibiting lipid peroxidation (Rogers & Munne-Bosch, 2016).
Manipulation of α-tocopherol content in leaves has been proposed as an efficient trait to improve dynamic responses to abiotic stress (Lizarazo et al., 2010). It would be interesting to test if transgenic approaches to increase α-tocopherol in aging leaves would specifically improve O 3 tolerance.
Senescing hybrid leaves exposed to elevated [O 3 ] showed a trend toward accumulating campesterol and stigmasterol (Figure 5b,c) at the expense of sitosterol (Figure 5d). Phytosterols are proposed to modulate membrane integrity to improve abiotic stress tolerance (Dufourc, 2008;Kuczynska et al., 2019). In stress conditions stigmasterol and sitosterol interact with phospholipids to maintain the permeability and fluidity of the plasma membrane (Dalal et al., 2016;Griebel & Zeier, 2010). Heat stressed hard fescues up-regulated ethyl sterol content, including stigmasterol and sitosterol, and the more heat tolerant variety showed a significantly greater increase in stigmasterol compared to a heat sensitive variety . In the current study, we found greater concentrations of stigmasterol and campesterol in hybrid leaves exposed to elevated [O 3 ], but not in inbred lines ( Figure 5; Table S5 and Table S6). Genotypic variation in phytosterol concentrations has been documented in wheat (Nurmi et al., 2010), rice (Kumar et al., 2018), and potato (Hancock et al., 2014), and genetic variation in the concentrations of phytosterols during drought was associated with differences in stress tolerance (Kumar et al., 2015). It is interesting that in elevated [O 3 ], increased phytosterol content only occurred in the hybrid genotype that also showed accelerated senescence and significant yield loss, not in the inbred lines. This suggests that increased phytosterol content was not associated with O 3 tolerance, instead with changes to membranes in aging leaves. Furthermore, campesterol is a precursor of brassinosteroids (Fujioka & Yokota, 2003), and brassinosteroid signaling mutants display a delayed senescence phenotype (Clouse & Sasse, 1998). Thus, the accumulation of campesterol in aging leaves  the hypothesis that the balance of various phytosterols may be a key signaling response for additional cellular defenses (Griebel & Zeier, 2010;Schaller, 2003).
Previous experiments have identified metabolites associated with maize grain yield under drought and heat stress (Obata et al., 2015), and here we investigated potential metabolite markers Field experiments can be more variable than controlled environment experiments (Lovell et al., 2016), and metabolites respond rapidly to environmental changes (Caldana et al., 2011) identified novel markers of O 3 response in hybrid maize, which can be broadly tested across diverse germplasm.

CO N FLI C T O F I NTE R E S T
The authors declare no conflict of interest associated with the work described in this Manuscript.

AUTH O R CO NTR I B UTI O N S
JMW and EAA conceptualized the study. JMW performed metabolomic and statistical analysis. CHB collected and analyzed leaf chlorophyll data with EAA. LRA and ADBL collected yield data. ADBL and EAA designed the field experiment. JMW and EAA wrote the manuscript with input from all authors.