The standards for IAA and indole-¹³C₆-labeled IAA were purchased from Merck (Germany). Standards for IAA-Asp, IAA-Glu, IAA-Leu, IAA-Phe, oxIAA, and ¹³C₆-[benzene ring]-IAA-Asp, [¹³C₆]IAA-Glu and [¹³C₆]oxIAA were purchased from OlChemIm (Czech Republic). IAA-glc, oxIAA-glc, [¹³C₆]IAA-glc, [¹³C₆]oxIAA-glc, oxIAA-Asp, oxIAA-Glu, oxIAA-Leu, oxIAA-Phe, oxIAA-[¹³C₄,¹⁵N]Asp, oxIAA-[¹³C₅,¹⁵N]Glu were synthesised in accordance to (Kai et al., 2007a; Kai et al., 2007c) with minor modifications. Plant agar and Murashige & Skoog media were purchased from Duchefa; methanol for HPLC analysis and all other chemicals were purchased from Merck (Germany) and Lach-Ner (Czech Republic).

Arabidopsis seeds (ecotype Col-0) were sterilized with 70% ethanol with the addition of 0.1% Tween 20 for 5 minutes, sown on Murashige & Skoog square agar plates (10 g sucrose, 4.4 g MS medium, 10 g agar per liter, pH 5.7) and cold treated for 4 days prior to germination. The plates were then transferred to a cultivation chamber and vertically placed under long-day conditions (16 h light/8 h dark) at 22 ± 1°C under cool white fluorescent light (maximum irradiance 550 μmol m⁻² s⁻¹). The seedlings were harvested at different growth stages according to the BBCH scale ( Table S1 ; Figure S1 ). Shoots and roots were separated and weighed in five replicates, homogenized and immediately frozen in liquid nitrogen, and stored at -80°C until extraction.

Pea (Pisum sativum arvense L.), wheat (Triticum aestivum L.), and maize (Zea mays L.) seeds were left germinating in the dark for two, three and four days, respectively, selected according to their uniformity from a large population, transferred to hydroponic boxes, watered with Hoagland’s solution and left growing in the cultivation room (16 h light/8 h dark) at 22 ± 2°C. Three to five plants per growth stage were harvested ( Table S1 ; Figure S1 ). Shoots, roots, and cotyledons were separated and weighed in five replicates, homogenized and immediately frozen in liquid nitrogen, and stored at -80°C until extraction.

Frozen samples containing 10 mg of fresh weight (pea, maize, wheat, Arabidopsis shoot) or 3 mg (Arabidopsis root) were extracted in 1 ml (500 µl for Arabidopsis root samples) of ice-cold Na-phosphate buffer (50 mM, pH 7.0, 4°C) containing 0.1% diethyldithiocarbamic acid sodium salt. A mixture of stable isotope-labeled internal standards ([¹³C₆]IAA, [¹³C₆]IAA-Asp, [¹³C₆]IAA-glc, [¹³C₆]IAA-Glu, [¹³C₆]oxIAA, oxIAA-[¹³C₄,¹⁵N]Asp, [¹³C₆]oxIAA-glc, oxIAA-[¹³C₅,¹⁵N]Glu) was added to each sample in amount of 5 pmol each (2.5 pmol for Arabidopsis roots). Due to the high concentration of oxIAA-Asp in pea seeds, five replicates were diluted in an ice-cold Na-phosphate buffer by one hundred (10 µl of sample to 990 µl of Na-phosphate buffer) and then 5 pmol of internal standards were added as described above. All samples were then homogenized using a MM400 bead mill (Retsch GmbH, Germany) at a frequency of 27 Hz for 10 minutes after adding three zirconium oxide beads. Extracted samples were incubated at 4°C with continuous shaking for 15 minutes and then centrifuged (15 minutes, 206 642 g, 4°C). From each sample, 200 µl of Na-phosphate buffer was transferred to clean microtubes, pH adjusted to 2.7 by 0.1 M hydrochloric acid and purified using in-tip μSPE (Pěnčík et al., 2018) with two types of extraction sorbents (three layers of each): HLB AttractSPE™ (Affinisep, France) and SDB-XC Empore™ (3M, USA). Briefly, multi-StageTip microcolumns were activated by 50 µl of acetone (centrifugation 10 minutes, 3 846 g, 8°C), 50 µl of methanol (10 minutes, 3 846 g, 8°C), and 50 µl of water (15 minutes, 4 654 g, 8°C). The acidified samples were then applied to activated microcolumns (centrifugation 30 minutes, 16 961 g, 8°C), washed by 50 µl of 0.1% acetic acid (20 minutes, 9 846 g, 8°C) and eluted with 50 µl of 80% methanol (20 minutes, 8 653 g, 8°C). After elution, the samples were evaporated to dryness in vacuo, and stored at -20°C until HPLC-MS/MS analysis.

For optimization of in-tip μSPE method, three types of extraction sorbents (SDB-XC Empore™, HLB AttractSPE™, and C18 Empore™) and their combinations were tested. For the single sorbent, five layers of HLB or SDB-XC were applied to the StageTip microcolumn. For multiple sorbent combinations, three layers of each sorbent were used. To compare extraction recovery of each microcolumn, 200 µl of acidified Na-phosphate buffer containing 2 pmol of unlabeled standards (IAA, IAA-Asp, IAA-glc, IAA-Glu, oxIAA, oxIAA-Asp, oxIAA-glc, oxIAA-Glu, oxIAA-Leu, oxIAA-Phe) was applied to activated microcolumns, washed and eluted as mentioned above. The samples were then evaporated to dryness in vacuo and stored at -20°C until HPLC-MS/MS analysis. Finally, process efficiency of each sorbent was expressed as a percentage recovery and calculated as the ratio of the peak area of an unlabeled analyte spiked before μSPE to the peak area of the neat solution (Matuszewski et al., 2003).

The evaporated samples were dissolved in 30 μl of 10% methanol prior to HPLC-MS/MS analysis using a 1290 Infinity LC system and a 6490 Triple Quadrupole LC/MS system equipped with Jet Stream and Dual Ion Funnel systems (Agilent Technologies, CA, USA). The chromatographic reverse-phase column (Kinetex C18 100A, length 50 mm, diameter 2.1 mm, particle size 1.7 μm; Phenomenex, CA, USA) was used for the separation of individual analytes by HPLC. The mobile phase consisted of deionized water (A) and methanol (B) with the addition of 0.1% acetic acid. The time of each analysis was 18 min, flow rate 0.3 ml/min and IAA metabolites were eluted using gradient elution as follows: 0 min – 10% B, 11.5 min – 60% B, 11.75 min – 100% B, 14.75 min – 100% B, 15 min – 10% B. For IAA-glc quantification in Arabidopsis, different type of column (Kinetex C18 100A, length 150 mm, diameter 2.1 mm, particle size 1.7 μm; Phenomenex, CA, USA) was used with the identical chromatographic conditions as described above. During analysis, the samples were stored in an autosampler at 4°C, a column tempered at 40°C and 10 μl of each sample was injected.

Individual analytes were detected by the MS instrument combining positive and negative ESI mode (+/-) with optimised conditions as follows: nebulizer pressure, 25 psi; drying gas flow and temperature, 14 l/min and 130°C; sheath gas flow and temperature, 12 l/min and 400°C; capillary voltage, 2.8 kV in positive mode and 3.0 kV in negative mode; nozzle voltage, 0 V. Analytes were detected and quantified using diagnostic multiple reaction monitoring (MRM) transitions of precursor and appropriate product ions using optimal collision energies and 50 ms dwell time ( Table S2 ). For data interpretation, Mass Hunter software (Agilent Technologies, CA, USA) was used.

Seven-day-old pea and wheat seedlings were harvested, weighed, and immediately frozen in liquid nitrogen, and stored at -80°C until extraction. Samples of 10 mg of FW were extracted in 1 ml of Na-phosphate buffer (50 mM, pH 7.0, 4°C, 0.1% diethyldithiocarbamic acid sodium salt) as previously described. After centrifugation, the samples were diluted to 5 ml with Na-phosphate buffer and each sample divided into doses of 200 µl (15 samples per plant). Subsequently, in five replicates, samples were supplemented with 1 pmol, 10 pmol or without unlabeled standards (IAA, IAA-Asp, IAA-glc, IAA-Glu, oxIAA, oxIAA-Asp, oxIAA-glc, oxIAA-Glu, oxIAA-Leu, oxIAA-Phe) and 5 pmol of internal stable isotope-labeled standards ([¹³C₆]IAA, [¹³C₆]IAA-Asp, [¹³C₆]IAA-Glu, [¹³C₆]IAA-glc, [¹³C₆]oxIAA, oxIAA-[¹³C₄,¹⁵N]Asp, [¹³C₆]oxIAA-glc, oxIAA-[¹³C₅,¹⁵N]Glu). All samples were then acidified and purified by in-tip μSPE method as described above. Finally, concentrations of all analytes were quantified by HPLC-MS/MS using a standard isotope dilution method. To calculate the accuracy of the method, the deviation of the determined analyte concentration (after subtraction of endogenous level) from its nominal level (known amount of analyte standard added to sample) was divided by the nominal level and expressed as percentage bias. Method precision was calculated as the relative standard deviation of determined levels.

Multivariate data analysis was performed using SIMCA 17 software (Sartorius Stedim Data Analytics, Umeå, Sweden). Dataset containing endogenous concentrations (pmol/g FW) was log transformed, centered and scaled using unit-variance scaling method. Values under limit of detection (LOD) were replaced with 0.95-fold of respective LOD value. Principal component analysis (PCA) was calculated for description of data structure. All PCA models were cross-validated to assess model reliability. Variables with no or neglectable effect in score plot were excluded.

Solid phase extraction is becoming the most widely used method for the purification of phytohormones. In the last decade, there has been a trend towards minimal use of solvents, which has led to the development of new solid phase microextraction such as in-tip µSPE, which was first used for cytokinin analysis (Svačinová et al., 2012) and later for auxin purification (Liu et al., 2012; Pěnčík et al., 2018). We adopted the in-tip µSPE purification protocol described by Pěnčík et al. (2018) to achieve the highest possible extraction efficiency for all IAA metabolites including novel oxIAA-AA conjugates. The recovery rate of analytes was tested using three types of extraction sorbents and their combination: SDB-XC Empore™, HLB AttractSPE™, and C18 Empore™. The recoveries of added auxin standards in acidified Na-phosphate buffer applied onto microcolumns are summarized in Figure S2 . The single-StageTips filled with HLB or SDB-XC sorbents recovered 49 ± 9% and 59 ± 11% of auxin compounds, respectively. In good agreement with previously published data (Pěnčík et al., 2018), multi-StageTips C18/SDB-XC also showed high extraction yields (mean recovery 60 ± 11%). Under our experimental conditions, the HLB/SDB-XC microcolumns combined hydrophilic-lipophilic balance/polymer-based sorbents provided the highest recoveries ranging from 47% to 89% across all analytes ( Figure S2 ). Importantly, oxIAA-AA conjugates were efficiently retained on HLB/SDB-XC sorbents with recovery rates ranging from 47 ± 14% (oxIAA-Asp) to 76 ± 5% (oxIAA-Leu). According to our findings, a combination of HLB/SDB-XC sorbents was used to isolate oxIAA-AA conjugates from the studied plant material.

All auxin metabolites were analysed by LC-MS/MS method using electrospray ionization in positive and negative modes. MRM transitions and collision energies were optimised for all analytes and their corresponding stable isotope labelled internal standards ( Table S2 ). Confirmation of the newly identified metabolites in pea cotyledons was performed based on comparisons to the retention times of a synthetic oxIAA-AA reference standard ( Figure 2 ). Importantly, two chromatographic peaks were detected due to the optical activity of investigated compounds at position three in the oxindole molecule, 8.9/9.5 min for oxIAA-Leu and 9.3/9.8 min for oxIAA-Phe. Moreover, each sample extract (Arabidopsis, wheat and pea) was spiked with a mixture of oxIAA-Leu and oxIAA-Phe at known levels (0, 1.0 and 10.0 pmol). These results also confirmed the detection of two chromatographic peaks of both new amino acid conjugates. Representative MRM chromatograms of the spiked Arabidopsis samples are shown in Figure 2 .

For method validation, 18-point calibration curves ranging from 45 amol to 90 pmol were constructed and the LOD (S/N ratio > 3) as well as the dynamic linear range were calculated ( Table S2 ). For newly identified metabolites, the linear range stretched from 9 fmol to 90 pmol, and for the other compounds, the linear range started even at 90 amol. Finally, a spiking experiment with authentic standards of auxin metabolites was performed. Wheat and pea extracts were used for method validation as representatives of monocots and dicots, respectively. A standard mixture containing 1 or 10 pmol of individual authentic standards was added to extracts containing 2 mg of plant material (homogenized whole plants) and their recovery was calculated after subtraction of endogenous levels (Matuszewski et al., 2003). The method accuracy (percentage deviation from the accepted reference value, % bias) and method precision (relative standard deviation of determined levels, RSD) were then calculated ( Table S3 ). Mean method accuracy was 15 ± 12% bias for pea samples and 19 ± 15% bias for extracts prepared from wheat seedlings. Method precisions of all analytes were below 14% RSD for both tested matrices ( Table S3 ). Overall, our results showed that the method has good linearity, high sensitivity and sufficient precision and accuracy, and thus the levels of auxin metabolites were finally determined.

Endogenous levels of oxIAA-AAs were determined in roots, cotyledons and shoots of three crop species (maize, pea, and wheat), and in shoots and roots of Arabidopsis ( Figure 3 ; Table S4 ). In order to obtain comparable data, all species were harvested at the same growth stage 1.0 ( Table S1 ; Figure S1 ).

Conjugates of oxIAA with Asp and Glu were determined in all four analysed species, but the levels varied greatly depending on the species and particular part of the plant. In general, oxIAA-Asp was found to be the most abundant oxIAA conjugate in pea tissues, for example concentrations close to 300 nmol·g^-1 FW in cotyledons ( Table S4 ). In contrast, oxIAA-Glu was the major oxIAA-AA conjugate in Arabidopsis. The determined levels of oxIAA-Glu were 81/92 pmol·g^-1 FW in Arabidopsis shoots/roots, while oxIAA-Asp was detected in lower quantities (16/11 pmol·g^-1 FW). In both members of the Poaceae family, wheat and maize, oxIAA conjugates with Asp and Glu were partially determined. No oxIAA-Glu was detected in maize shoot samples ( Figures 3B ). Moreover, both amino acid conjugates were below the limit of detection in wheat shoots and cotyledons. Interestingly, endogenous levels of oxIAA-Asp showed similar pattern in maize tissues compared to pea samples, but at picomolar levels ( Figure 3A ).

The newly identified metabolites, oxIAA-Leu and oxIAA-Phe, were detected only in cotyledon samples, oxIAA-Phe in maize and both conjugates in pea ( Figures 3C, D ). Importantly, these Leu and Phe conjugates were much less abundant compared to oxIAA-Asp or oxIAA-Glu. Pea cotyledons contained 7 pmol·g^-1 of oxIAA-Phe and less than 1 pmol·g^-1 of oxIAA-Leu, similar to oxIAA-Phe level in maize cotyledons ( Table S4 ).

Full auxin metabolite profiles including free IAA, oxIAA and their low-molecular-weight conjugates with amino acids and glucose were determined in the same four plant models at growth stage 1.0. The levels of IAA and oxIAA conjugates with individual amino acids (Asp, Glu, Leu and Phe) were summed up into two corresponding groups IAA-AA and oxIAA-AA, respectively. The contributions of different conjugate classes as well as glucosyl esters were calculated as their relative abundances to the total pool of IAA metabolites ( Figure 4 ; Table S5 ). Overall, the relative distribution of auxin metabolites differed notably within studied species as well as within the individual plant parts.

The dominant conjugate in Arabidopsis was oxIAA-glc taking 81% (shoots) and 68% (roots) of all measured metabolites, while IAA-glc and amino acid conjugates of both IAA and oxIAA, as well as free IAA, were found to have a minor representation in Arabidopsis. The second most abundant metabolite was oxIAA, which took 15% of the IAA metabolite pool in the shoots and 23% in the roots ( Figure 4A ). In contrast, the major auxin metabolites in pea were oxIAA-AA conjugates, accounting for 53% in shoots and 58% in roots. IAA-AAs were the second most abundant conjugates in pea shoots (22%) and roots (15%). In cotyledons, oxIAA-AAs were fully dominant, taking 99% of all auxin metabolites ( Figure 4B ). Similar profiles of auxin metabolites were observed in maize and wheat. In both species, oxIAA was major metabolite in shoots (79% in maize; 88% in wheat). In other parts, oxIAA was also the most abundant inactive auxin metabolite; only free IAA had a greater proportion in maize roots and wheat cotyledons. Amide conjugates of oxIAA were also present in maize and wheat, but occupied a relatively minor part of the total auxin metabolite profile ( Figures 4C, D ).

We further investigated the distribution of IAA and its metabolites during the early development of Arabidopsis, pea, maize and wheat. Therefore, different plant parts such as roots, shoots (all species) and cotyledons (only crop species) were sampled at different growth stages. The first analysed stage was 1.0, in which the first leaf emerges from the coleoptile in grains, or cotyledons are fully unfolded in pea and Arabidopsis. The following stages were 1.1 and 1.2, in which the first and second leaf are fully developed and unfolded ( Figure S1 ). However, the stage 1.1 is not differentiated in Arabidopsis as first two leaves develop simultaneously.

Principal component analysis (PCA) was used to observe the variability in auxin metabolite content. As already indicated in Figure 4 , the significant differences of the relative distribution of auxin metabolites between plant species was observed. Therefore, each species was evaluated in individual PCA model to describe the distribution of metabolites in roots and shoots and growth stages ( Figure 5 , complete dataset provided as Supplementary Table 6 ). The most complex patterns found in this study were later analysed using additional PCA to describe changes in the development of individual plant parts ( Figure 6 ).

In Arabidopsis, only stages 1.0 and 1.2 were analysed. At stage 1.0 the cotyledons are completely opened, while stage 1.2 is defined by fully developed first two true leaves (Boyes et al., 2001). A more pronounced stage-dependent alteration of auxin metabolome was observed in shoots, where auxin metabolite content was higher at the early stage (1.0) and dropped during seedling development up to stage 1.2 ( Figure 5A ). This trend was observed not only in auxin metabolites, but also in free active IAA. Interestingly, there were no significant changes in auxin metabolome in the roots during development. Regardless of the stages, oxIAA-glc levels were much higher in shoots, compared to roots. Pea roots contained a higher concentration of auxin metabolites than shoots at all growth stages ( Figure 5B ). The roots at stage 1.0 were characterised by the highest content of oxIAA-Asp and oxIAA-Glu, the concentrations of which gradually decreased until stage 1.2. A similar trend was observed in the shoots, where the highest levels of all IAA and oxIAA conjugates were determined at stage 1.0. The levels of conjugates dropped towards later stages, while oxIAA showed a maximum at middle stage 1.1 ( Figure 6A ). In maize and wheat, higher concentrations were also observed in roots compared to shoots throughout the development period. In maize, the two earlier stages were associated with higher levels of oxIAA, while the last stage was characterized by the accumulation of oxIAA-glc in roots ( Figure 5C ). With the increasing stage of development, the concentration of amino acid conjugates of both IAA and oxIAA decreased in wheat roots, while oxIAA and oxIAA-glc levels have changed in the opposite manner ( Figures 5D , 6B ).