Maize inbred lines (B73, W22, Mo17, PH207) and expired Plant Variety Protection lines (PHB47, PHZ51) were from the North Central Regional Plant Introduction Station (USDA-ARS, Ames, IA, United States). Maize UniformMu transposon mutant stocks for Zmcyp710a8-1 (mu1034077, UFMu-03461) and Zmcyp710a8-2 (mu1019883, UFMu-00809) alleles were acquired from Maize Genetics Cooperation Stock Center (USDA-ARS, Urbana-Champaign, IL, United States). Transposon insertions were genotyped as previously described (Liu et al., 2016). For gene expression and sterol analysis of maize at V1 stage (Ritchie et al., 2005), seedlings were grown for 10 days in Sunshine LC1 mix (SunGro Horticulture, Agawam, MA, United States) in a growth chamber with light intensity 200 μmol m^–2 s^–1, 16:8 h day/night cycle and 25°C (day)/22°C (night). For mutant characterization and non-targeted metabolite analysis, maize seedlings were grown in cigar rolls as previously described (Kumar et al., 2012) for 11 days without fungicide. Tissue samples were frozen immediately in liquid nitrogen and stored in –80°C.

Arabidopsis thaliana var. Columbia (Col-0, stock CS70000) seeds were from Arabidopsis Biological Resource Center (Columbus, OH) and the Atcyp710a1 mutant (GK-325E09, stock N421631; Wang et al., 2012) seeds were from Nottingham Arabidopsis Stock Center (Loughborough, United Kingdom). Arabidopsis was grown in the following conditions: light intensity was 150 μmol m^–2 s^–1, 16:8 h day/night cycle and 23°C. Seeds were grown in soil for plant transformation, transformants identification, and sterol analysis. For non-targeted metabolite analysis, surface-sterilized Arabidopsis seeds were grown in Magenta boxes as described by Suza and Staswick (2008) in liquid Murashige and Skoog (MS) medium (Murashige and Skoog, 1962). Tissues from 3-week-old seedlings were harvested, frozen in liquid nitrogen and stored in –80°C.

Stigmasterol, ribitol, nonadecanoic acid, methoxyamine hydrochloride, pyridine, and 5α-cholestane were obtained from Sigma-Aldrich (St. Louis, MO, United States), sitosterol from Avanti Polar Lipids (Alabaster, AL, United States), campesterol and brassicasterol were from Cayman chemicals (Ann Arbor, MI, United States). Glufosinate ammonium was from Crescent Chemicals (Islandia, NY, United States). Chloroform, methanol, diethyl ether, hexane, and HPLC water were obtained from Fisher Scientific (Fair Lawn, NJ, United States).

Phylogenetic analysis using the Maximum Likelihood method and JTT matrix-based model (Jones et al., 1992) was carried out to evaluate the evolutionary relationship of Zm00001d039384 with other CYP710A-like homologs. Full length protein sequences were aligned using MUSCLE algorithm (Edgar, 2004). The tree construction was carried out with default parameters: 1,000 bootstrap replicates and, uniform sites and use all sites in MEGA X (Kumar et al., 2018). Saccharomyces cerevisiae CYP61 was used to root the tree. The protein sequences were from previous reports or downloaded from the Phytozome portal¹ and are as described previously (Aboobucker and Suza, 2019).

ZmCYP710A8 ORF was amplified from maize B73 genomic DNA using primer pairs, ZmCYP710A8_F and ZmCYP710A8_R (Supplementary Table 1). PrimeSTAR GXL DNA Polymerase (Clontech, Mountain View, CA, United States) was used with the following conditions: 98°C for 10 sec initial denaturation followed by 35 cycles of 98°C for 10 s, 55°C for 15 s and 68°C for 90 s and a final 68°C for 5 min. After purification using PCR purification kit (Qiagen, Germantown, MD, United States), the purified PCR product and pTF101.1-35S vector were digested with KpnI and SacI (New England Biolabs, Ipswich, MA). After ligation for overnight at 4°C, they were transformed into NEB10-b cells. Successful clones were transformed into Agrobacterium tumefaciens strain C58C1 using the freeze-thaw method (Holsters et al., 1978).

Arabidopsis transformation was using the floral dip method (Clough and Bent, 1998). Putative transformants were identified by spraying glufosinate (60 mg/L) at stage 1.02 (Boyes et al., 2001) and PCR confirmation. Single insertion and homozygous lines based on Mendelian segregation were identified by growing seedlings in vitro with 10 mg/L glufosinate.

Hormone treatment method was adapted from Ortiz-Masia et al. (2007) and Louis et al. (2015). Hormone stock solutions (1000×) of methyl jasmonate (50 mM), abscisic acid (100 mM), indole acetic acid (100 mM) or salicylic acid (50 mM) were prepared fresh every time by dissolving in ethanol. To prepare 10 mM 6-benzyladenine stock solution, it was first dissolved in 2.5 mL (2.5 M HCl) followed by adding 2.5 mL ethanol to make a total of 5 mL. Final hormone spraying solutions (400 mL) were prepared by diluting freshly made stock solutions (or ethanol for mock) in double-distilled water + 0.1% Tween-20. Ten days old (V1 stage; Ritchie et al., 2005) maize B73 seedlings (3 per pot) were sprayed with a fine mist of the hormone solutions until drip around mid-day (11 AM). Second leaf samples (Figure 3A; Yan et al., 2015) from three seedlings were pooled and frozen immediately in liquid nitrogen for one treatment per time point. All samples were stored in –80°C until further processing.

Sterol stock solutions were prepared by dissolving sterol powders in ethanol and stored in –20°C. For gene expression experiments, feeding solutions of sterols (10 μM in ddH₂O) were prepared by diluting stock solutions one day prior and equilibrated in the growth chamber. New germination papers were soaked by spraying with the feeding solution or ddH₂O (control). 4- or 11-days old maize seedlings were placed on the papers and rolled in such a way that roots are completely covered inside the roll, while shoots protrude from the rolls. Feeding solutions were also poured over the rolls at the shoot-root junction to ensure the rolls were sufficiently drenched. The rolls were then placed in beakers containing feeding solutions, while ensuring that only the bottom ∼5 cm of the rolls (of the 30 cm long) was immersed in the solutions. For non-targeted metabolite analysis, the feeding procedure was the same except for two modifications. First, stigmasterol (5 μM) was used and second, maize seedlings were 4 days old when the feeding was performed and incubated for another 7 days. After the indicated times, root or shoot tissues were harvested, frozen in liquid nitrogen, and stored at –80°C.

Total RNA from maize or Arabidopsis was extracted using RNeasy Plant Mini Kit (Qiagen, Germantown, MD, United States) and treated with rDNaseI (Thermo Fisher, Waltham, MA, United States) per manufacturers’ instructions. First strand cDNA was synthesized from 1 μg of total RNA and oligo (dT)₂₀ primers using Superscript III First Strand cDNA synthesis system (Invitrogen, Carlsbad, CA, United States). For qRT-PCR analysis, SsoAdvanced Universal SYBR Green Supermix (Bio-Rad, Hercules, CA, United States) was used on an Applied Biosystems Mx3000P thermocycler with 10 ng (for maize) or 5 ng (for Arabidopsis) of cDNA as template in a 10 μL reaction. Primer sequences are described in Supplementary Table 1. Relative expression levels compared to internal reference ZmACTIN (Louis et al., 2015) or AtEF1α-A (Aboobucker et al., 2017) were calculated using the 2^–ΔΔCt method (Pfaffl, 2001).

Sterol analysis was performed as previously described (Suza and Chappell, 2016) with following modifications. Lyophilized tissue powder (∼20 mg) and 8 μg 5α-cholestane (internal standard) were added in a clean glass tub together with chloroform (5 mL) and sonicated at room temperature (RT) for 10 min followed by a 45 min incubation at 50°C. After equilibrating to RT, water (5 mL) was added, vortexed and incubated at 50°C for 45 min. The organic layer was transferred to a new tube and dried under dry nitrogen. The samples were re-suspended in chloroform (2 mL), hydrolyzed with 1 mL of 1.25 M HCl (in methanol) and incubated at 50°C for 2 h. The samples were dried under nitrogen followed by methanol (1 mL) addition and dried under nitrogen. Organic extracts were re-suspended in 1.2 mL of diethyl ether:hexane (9:1, v/v) and transferred to a new vial followed by drying under nitrogen. The samples were derivatized in 60 μL pyridine and 40 μL BFTSA + 1% TMCS (Sigma) and incubating at 50°C for 1 h (Koek et al., 2006).

Trimethylsilyl derivatized samples were analyzed using an Agilent Technologies Gas Chromatograph (Model 7890C) equipped with a DB-5 column (Agilent, 30 m × 0.25 mm, 0.25 mm phase thickness) and coupled to a mass spectrometer (Agilent, Model 5975C). The samples were injected in split less mode with an inlet and transfer line temperature of 280°C. Helium at a flow rate of 1 mL/min was the carrier gas. The initial oven temperature was 110°C and subsequently increased to 275°C at a gradient of 20°C/min followed by a gradient of 2°C/min up to 320°C held for 6 min. The mass spectrometer was held at standard settings with an auto tune method used for calibrating and tuning. Quantitative sterol measurements were by integrating peak areas to each sterol species and comparing to the internal standard peak area. Peak detection and deconvolution were performed using AMDIS (Automated Mass Spectral Deconvolution and Identification System, National Institute of Standards and Technology) software. Peaks were identified by comparing to available spectra in NIST17/Wiley11 libraries and authentic standards and a previous report for steryl glucosides (Phillips et al., 2005; Supplementary Figure 1).

Roots of 3-day-old maize seedlings were embedded in cryo-molds containing 2% medium viscosity carboxymethylcellulose (CMC; Sigma-Aldrich, St. Louis, MO, United States). The embedded roots were immediately placed in a prechilled cryostat (Leica Biosystems, Buffalo Grove, IL) to equilibrate prior to being cryosectioned. 30 μm thick longitudinal sections of CMC embedded roots were taken and applied directly onto conductive indium tin oxide (ITO) coated slides (part no.: 8237001, Bruker Daltonik, Bremen, Germany). Sample sections were dried and brought to room temperature in a vacuum desiccator. Prior to matrix deposition, bright field images of the root sections were captured using a Macro Zoom imaging system (ZEISS, Oberkochen, German). A matrix consisting of sputter coated silver was applied to the root sections for 30 s using a Denton Desk II sputter coater (Denton Vacuum, Moorestown, NJ, United States) equipped with a silver target (99.99% Ag, Ø60 mm × 0.1 mm) (Ted Pella, Inc., Redding, CA, United States).

Mass spectrometry imaging (MSI) was performed using a Bruker SolariX fourier-transform ion cyclotron resonance mass spectrometer (FT-ICR MS) equipped with a 7.0 tesla superconducting magnet (Bruker Daltonik, Bremen, Germany). Sterols, as silver cationic adducts, were detected using matrix-assisted laser desorption/ionization mass spectrometry (MALDI-MS), as described previously (Jun et al., 2010; Dufresne et al., 2013; Sekula et al., 2015; Xu et al., 2015). MSI data was acquired in positive mode with a mass range from m/z 200.90 to 2000 while collecting one Megaword of data points per scan. A quadrupole isolation ranging from m/z 500-530 was used to isolate the sterol adducts and reduce silver aggregate background interference. The laser raster was set to acquire 50 μm spots. The FT-ICR MS instrument was operated using ftmsControl software (version 2.1 version 4.1, Bruker Daltonik, Bremen, Germany) while flexImaging software (version 4.1, Bruker Daltonik, Bremen, Germany) was used to collect and analyze the imaging data.

Statistical analyses were conducted as described in the figure or table legends. P values were calculated by two-tailed Student’s t-test using Microsoft 365 Excel or by one-way ANOVA multiple comparisons with Tukey test in R.

Identification of putative sterol C-22 desaturases in maize involved querying the MaizeGDB database² with AtCYP710A1 amino acid sequence (Morikawa et al., 2006). The query resulted in a single 1,551 bp intron-less gene model (Zm00001d039384) predicted to encode a 515 aa protein with 63% similarity to AtCYP710A1 and we renamed it to ZmCYP710A8 as per P450 nomenclature (Nelson, 2009). Multiple sequence alignment showed that the amino acid sequence of ZmCYP710A8 is highly similar to CYP61 from yeast, and previously characterized CYP710A proteins from plants (Supplementary Figure 2). Further, conserved sites and characteristic motifs for CYP710A are also present in ZmCYP710A8 (Supplementary Figure 2).

Phylogenetic analysis suggested that other grass species may also encode single sterol C-22 desaturases (Figure 1C). The CYP710A-like homologs cluster into two monophyletic clades, monocots, and eudicots, with ZmCYP710A8 placed in the monocot clade. Moreover, no CYP710A-like gene from either clade clustered together, suggesting that the plant CYP710A-like genes are from a single ancestor and developed into different branches after the lineages diverged. Within the clades, some species have more than one representative, suggesting these paralogs might have resulted from recent genome duplication events. These findings suggested that ZmCYP710A8 may encode the maize sterol C-22 desaturase.

To validate the function of ZmCYP710A8 in stigmasterol biosynthesis, the full-length coding sequence was transformed into Arabidopsis (Figure 2A) and positive transformants were confirmed by PCR. Sterols were analyzed by GC-MS from rosette leaves of transgenic Arabidopsis overexpressing ZmCYP710A8 with single insertion in the T3 generation (Supplementary Data 1). Wild-type (Col-0) produced trace amounts of stigmasterol, while its concentration in transgenic lines ranged from 166.3 to 1415.2 μg g^–1 DW (Figure 2B). The stigmasterol increase in transgenic plants was concomitant with a reduction in sitosterol, suggesting a direct substrate-product relationship. Further, stigmasterol content strongly correlated with ZmCYP710A8 mRNA levels (Figure 2B), suggesting transcriptional regulation of stigmasterol biosynthesis. The concentration of campesterol, however, remained relatively constant in transgenic lines. The apparent morphology of ZmCYP710A8 overexpressing Arabidopsis plants did not differ from that of wild type plants (Figure 2C). Taken together, ZmCYP710A8 encodes a sterol C-22 desaturase in maize.

To investigate the developmental expression of ZmCYP710A8, we measured its mRNA in various tissues of maize seedlings (Figure 3A), along with mRNAs for several key plant sterol biosynthesis genes including HMGS, HMGR, SQS, SMT1, SMT2 (Suza and Chappell, 2016). Except for DWF1 (Best et al., 2016) and ZmCYP710A8, maize encodes multiple copies for HMGS, HMGR, SQS, SMT1, and SMT2 (Supplementary Table 2). Therefore, PCR primers were designed to capture all molecular transcripts from a given gene family as previously described in N. benthamiana (Suza and Chappell, 2016).

Sterol profile of maize seedlings was also measured in addition to mRNA expression (Figure 3 and Table 1). Stigmasterol content was highest in roots compared to aerial tissues in B73 (Figure 3D) and a similar trend was observed in roots and shoots of additional maize genotypes (Figure 3C and Supplementary Figure 3). In contrast, sitosterol concentration displayed an inverse pattern than that of stigmasterol. Like stigmasterol, total sterol content was highest in roots followed by aerial tissues (Table 1). The mRNA levels for ZmCYP710A8 and other sterol biosynthesis genes were highest in roots compared to aerial tissues (Figure 3B), correlating with stigmasterol content (Figure 3D). In addition, ZmCYP710A8 mRNA levels strongly correlated (R² = 0.94) with stigmasterol but showed an inverse correlation (R² = –0.92) with sitosterol in B73 and other maize genotypes (Figure 3C, Supplementary Figure 3, Supplementary Table 3). The results suggest that the control of stigmasterol biosynthesis is at the transcriptional level (Suza and Chappell, 2016), and as suggested (Figure 2B), a single sterol C-22 desaturase may exist in the maize genome.

The expression pattern of ZmCYP710A8 in response to phytohormone treatments was examined to study stigmasterol regulation by hormone treatment. All hormone treatments used except methyl jasmonate suppressed the expression of ZmCYP710A8 at indicated time points and concentrations (Figure 4).

To generate a stigmasterol free system for studying its biological function in maize, we sought to identify plants compromised in stigmasterol production from the maize UniformMu population (McCarty et al., 2005). PCR analysis led to the identification of transposon insertions in two independent alleles (Zmcyp710a8-1 and Zmcyp710a8-2) within and near the ZmCYP710A8 open-reading frame (Figures 5A,B). Zmcyp710a8 plants did not show any observable phenotypic defects compared to wild type and they were able to grow and produce seeds under controlled conditions and in the field (Figure 5C and Supplementary Figure 4).

Sterol analysis by GC-MS showed the Zmcyp710a8-1 allele did not produce stigmasterol, but Zmcyp710a8-2 produced 0.3- to 0.5-fold stigmasterol as the W22 wild type (Figure 5D and Supplementary Table 4). Compared to W22, the F₁ progeny from a cross between W22 and Zmcyp710a8-1 (W22/Zmcyp710a8-1) produced 0.8- and 0.6-fold less stigmasterol in roots and shoots, respectively (Supplementary Figure 5). These results support our phylogenetic analysis and heterologous expression data (Figures 1, 2) allowing us to conclude that ZmCYP710A8 is the sole sterol C-22 desaturase in maize.

The composition of free sterols and their intermediates were altered in the Zmcyp710a8-1 and Zmcyp710a8-2 mutants. For instance, intermediate sterols including cycloartenol and isofucosterol were reduced in roots and shoots of Zmcyp710a8-2 relative to W22 (Supplementary Table 4). In roots, sitosterol was 25% of the total sterols in W22, 54% in Zmcyp710a8-2, and 62% in Zmcyp710a8-1, while in shoots, it was 56% in W22, 70% in Zmcyp710a8-2, and 80% in Zmcyp710a8-1 (Figure 3C and Supplementary Table 4).

Further, campesterol composition was significantly different between W22 and Zmcyp710a8-1; the difference in roots being 28% of total sterols in W22 and 35% in shoots, while in shoots W22 had 14% and Zmcyp710a8-1 had 18%. There was no appreciable difference, however, in campesterol proportions between W22 and Zmcyp710a8-2 (Figure 5E and Supplementary Table 4). Nonetheless, it is noteworthy that the composition of steryl glucosides was proportional to those of free sterols in roots and shoots (Supplementary Figure 6). Mass spectrometry imaging of wild type and Zmcyp710a8-1 roots indicate that sitosterol is localized to the root tip region while stigmasterol is more broadly distributed (Figure 5F). Taken together, the disruption of ZmCYP710A8 eliminates stigmasterol production and affects the composition of free sterols and steryl glucosides, and the distribution of sitosterol in roots.

The positive correlation between stigmasterol content and mRNA for key sterol biosynthesis genes (Figure 3) suggests stigmasterol may have a stimulatory effect on gene expression. Identification of a stigmasterol-free background in Zmcyp710a8-1 (Figure 5) provided a system to test whether genes such as HMGR and SMT whose expression in various tissues correlates with stigmasterol content (Figure 3) might be responsive to exogenously supplied stigmasterol.

Treatment of W22 roots with exogenous stigmasterol resulted in the induction of ZmHMGR and ZmSMT2 mRNA after 4 and 8 h (Figure 6A). Although ZmHMGR mRNA remained elevated, ZmSMT2 mRNA slightly decreased after 8 h (Figure 6A). Exogenous sitosterol, in contrast, did not have a significant impact on ZmHMGR and ZmSMT2 mRNA after 4 h (Figure 6B) and 8 h (not shown).

Similar to the observation in W22, ZmHMGR and ZmSMT2 mRNA levels were higher in roots and shoots of Zmcyp710a8-1 treated with stigmasterol (Figures 6C,D). Sitosterol, however, had a modest effect on ZmHMGR and ZmSMT2 mRNA in roots of Zmcyp710a8-1 (Figures 6C,D). The impact of stigmasterol on mRNA for ZmHMGR and ZmSMT2 suggests that stigmasterol might modulate transcripts for plant sterol biosynthesis genes.

To identify cellular metabolites influenced by stigmasterol, we used non-targeted GC/MS metabolite analysis of roots and shoots of maize Zmcyp710a8 and wild type. Molecular features with significant changes among maize genotypes were identified based on the filtering criteria as described in section “Materials and Methods.” A total of 29 molecular features (13 known and 16 unknown) were identified in roots and 31 features (17 known and 14 unknown) in shoots (Figure 7 and Supplementary Data 2). The concentration of molecular features correlated with stigmasterol content among genotypes in maize and the features belonged to similar classes including sterol, amino acid, primary metabolism, and unknown features in roots and shoots (Figure 7 and Supplementary Figure 7).

Metabolites such as glutamine and an unknown feature with retention index (RI) 1796.8 in Zmcyp710a8-1 roots were 1.6- and 1.9-fold higher compared to W22, respectively (Figure 7A and Supplementary Figure 7). Threonine and tyrosine in shoots were 1.6- and 1.7-fold higher in Zmcyp710a8-1 compared to W22 (Figure 7A), respectively. Among genotypes, changes in metabolites correlated with stigmasterol content (Figure 7A). For instance, threonine and myo-inositol showed a negative correlation with stigmasterol content in roots and shoots. Other metabolites that negatively correlated with stigmasterol were asparagine, 5-oxoproline and glutamine in roots, and tyrosine and triacontanoate in shoots (Figure 7A). Furthermore, unknown features with RI 1286.4 in roots and shoots; 1475.7, 1796.8, 1178.2, and 1600.6 in roots, and 2097.6 and 3074.5 in shoots were among the metabolites inversely correlated with stigmasterol (Supplementary Figure 7). Similarly, sitosteryl and campesteryl glucosides in roots and shoots, respectively, correlated negatively with stigmasterol (Figure 7A). Compounds that correlated positively with stigmasterol were stigmasteryl glucoside in roots and shoots and histidine, asparagine and non-acos-1-ene in shoots (Figure 7A). Unknown features with RIs of 1969.8 in roots and shoots; 2062.7, 1621.8, and 1720.0 in roots; 1545.3 and 1483.0 in shoots were also directly correlated with stigmasterol (Supplementary Figure 7).

Since changes in metabolites among the genotypes correlated with stigmasterol content, we reasoned that exogenous stigmasterol may reverse the metabolite levels in Zmcyp710a8-1 mutant to wild type levels. Treatment of the Zmcyp710a8-1 mutant with exogenous stigmasterol shifted the levels of sitosterol, campesterol, and myo-inositol toward W22 levels (Figure 7B), suggesting a complementation effect.

To test whether stigmasterol impacts cellular metabolism in a dicot species, we extended the study to Arabidopsis. The Atcyp710a1 mutant (Wang et al., 2012) produces trace amounts of stigmasterol in roots (44.74 μg g^–1 DW) and shoots (12.29 μg g^–1 DW) providing a good system to test the impact of restoring or increasing stigmasterol production on cellular metabolism (Table 2). The full-length ZmCYP710A8 ORF with the 35S viral promoter (Figure 2A) was transformed into Atcyp710a1 allowing us to obtain several lines (data not shown) expressing ZmCYP710A8. Two lines (L12 and L13) with single insertion of the transgene in T4 generation (Supplementary Data 1) were selected for further analysis. There was a direct correlation between stigmasterol content and ZmCYP710A8 mRNA in each of the transgenic lines (Figures 8A,B). In line L13, ZmCYP710A8 mRNA and stigmasterol were high in roots (3417.28 μg g^–1 DW) and shoots (2766.02 μg g^–1 DW), while their sitosterol levels were reduced consistent with the presence of AtCYP710A1 in Col-0 and ZmCYP710A8 in L13. In contrast, ZmCYP710A8 mRNA and stigmasterol concentration of line L12 in roots (182.76 μg g^–1 DW) and shoots (1824.12 μg g^–1 DW) was reciprocal to that of the wild type (Figures 8A,B and Table 2). Line L12 had a higher sitosterol content in roots as in Atcyp710a1 consistent with low ZmCYP710A8 mRNA in L12 and reduced AtCYP710A1 activity in Atcyp710a1. In roots and shoots, sitosterol and campesterol followed an inverse trend with stigmasterol while brassicasterol positively correlated with stigmasterol (Table 2). The campesterol content in shoots was also similar between Atcyp710a1 and Col-0 but lower in L12 and L13. Brassicasterol levels were comparable among Atcyp710a1, Col-0 and L12 (69.44–95.96 μg g^–1 DW) but were considerably higher in L13 (136.07 μg g^–1 DW) (Table 2).

We also used non-targeted GC/MS analysis to study cellular metabolites in the Atcyp710a1 mutant, Col-0, L12 and L13. Molecular features with significant changes among Arabidopsis genotypes were identified like in maize as described in section “Materials and Methods.” A correlation between Arabidopsis genotypes and the concentration of molecular features was observed. Metabolite concentrations among the Arabidopsis genotypes correlated with their stigmasterol content like maize (Figure 8C). A total of 39 molecular features (15 known and 24 unknown molecular features) were identified in roots and 28 features (17 known and 11 unknown molecular features) in shoots (Figure 8C and Supplementary Data 3). Sitosteryl glucoside levels in roots and shoots were negatively correlated with stigmasterol, whereas stigmasteryl glucoside was positively correlated with stigmasterol (Figure 8C). Methionine was 0.7-fold and RI 2288.9 was 2.4-fold in shoots of Atcyp710a1 as in Col-0 (Figure 8C), and myo-inositol and threonine were affected in both Arabidopsis and maize shoots. It is noteworthy that, Zmcyp710a8-1 and Zmcyp710a8-2 have increased levels of several amino acids in roots and shoots. Conversely, the Atcyp710a1 mutant produces comparable levels of several amino acids in roots and shoots, but there is an upward shift in amino acid levels when stigmasterol is increased as in L13 (Figure 8C).

Metabolites such as tetracosanoate and molecular feature RI 2996.5 were 1.4- and 1.6-fold higher, respectively, in Atcyp710a1 roots compared to Col-0 and L13 (Figure 8C and Supplementary Figure 8). Several other metabolites were also affected in Arabidopsis correlating with stigmasterol, fructose and aspartic acid contents having negative correlations with stigmasterol in roots and shoots (Figure 8C). Other compounds that correlated negatively with stigmasterol in roots were 24-methylenecycloartanol, tetracosanoate, and several unknown features with retention indices (RI) 1743.4, 1853.8, 2430.4, 2686.9, 2815.5, 2996.5, 3313.5, 3436.2, 3574.1, and 3616.1 (Figure 8C and Supplementary Figure 8). Metabolites correlated positively with stigmasterol content in roots were cycloartenol, hydrocarbon-like3 (an unknown molecular feature with spectral similarity to known normal-chain hydrocarbons), isoleucine and 3-hydroxytyrosine, and unknown features with RIs 1496.8, 1578.5, and 2382.9 (Figure 8C and Supplementary Figure 8).

In shoots, isofucosterol, ornithine, α-ketoglutaric acid, malic acid, glyceric acid and a few unknown features with RIs 1965.6, 2288.9, and 2430.4 were inversely correlated with stigmasterol (Figure 8C and Supplementary Figure 8). On the other hand, there was a direct correlation between stigmasterol and several metabolites such as 11-eicosenoic acid, methionine, glucose, and unknown features with RI 1525.6, and 1945.7 in shoots (Figure 8C and Supplementary Figure 8).