In a previous study, we found that CCE was completely suppressed in ech2–1 fugu5–1 and ech2–2 fugu5–1 mature cotyledons, suggesting that IBA, a substrate of ECH2, is likely involved in fugu5 CCE [22]. However, Li et al. [31] found that the accumulation of the precursor compound of the metabolic reaction catalyzed by ECH2 caused ech2–1 developmental defects. Simultaneously, this metabolic disorder in ech2–1 also indirectly affected the conversion of IBA to produce IAA. These findings indicate that the loss of ECH2 affects many metabolic reactions, including IBA-to-IAA conversion. Moreover, the triple mutant ibr1,3,10 exhibits a typical low-auxin phenotype reminiscent of ech2–1 [33]. Based on these studies, we conducted genetic analyses using ibr1,3,10 fugu5–1 to corroborate the importance of IBA-to-IAA conversion in CCE.

Quantitative analyses revealed that cell numbers, cell sizes and cotyledon sizes in ibr1-2 fugu5-1, ibr3-1 fugu5-1, and ibr10-1 fugu5-1 were comparable to the WT (S1 Fig). However, the cotyledon of the ibr1,3,10 fugu5-1 quadruple mutant was smaller than that of the WT and fugu5–1 (Fig 1A and 1B). Subsequent quantification at the cellular level revealed a reduced cell number in ibr1,3,10 fugu5–1 to the same extent as in fugu5–1 and ibr1,3,10 (Fig 1B). However, cells in the quadruple mutant were comparable in size to those of the WT (Fig 1B). Note that fugu5 cotyledon area was not different from the WT (Figs 1B and S1B).

Next we tested whether the reduced cell numbers in fugu5–1 is due to the lack of Suc, acting as the trigger of CCE [22–25,27]. To do so, we assessed the phenotypic effects of exogenous Suc supply on the ibr1,3,10 fugu5–1 quadruple mutant phenotype. While cell numbers in the quadruple mutant cotyledons were reset to the WT levels, the cell size remained unchanged and was comparable to the fugu5–1 single mutant (S2 Fig). In addition, we noticed smaller rosette leaves, shorter flowering stems in fugu5–1 and the quadruple mutant compared to the WT and ibr1,3,10 (S3 Fig), suggesting a growth delay in ibr1,3,10 fugu5–1, as is seen in the fugu5–1 single mutant [23,47].

These results indicate that Class II CCE in fugu5–1 is suppressed when IBA-to-IAA conversion is genetically impaired, mimicking ech2–1 fugu5–1 [22,25]. Surprisingly, although cell numbers in ibr1,3,10 cotyledons were decreased to the same level as in fugu5–1 (Fig 1B), this phenotype significantly but only partially recovered following exogenous supply of Suc (S2 Fig).

The small cotyledon phenotype in ibr1,3,10 has been attributed to the failure of IBA-to-IAA conversion [34]. However, the cellular phenotypes of ibr1–2, ibr3–1, and ibr10–1 single mutant cotyledons have not been reported. Although the single mutant gross morphology, and cotyledon aspect-ratios were comparable to that of the WT (S4A and S4B Fig), quantification of their cotyledon cellular phenotypes revealed that while ibr1–2 and ibr3–1 display normal cell numbers and cell sizes, ibr10–1 exhibited significantly fewer cells and thus CCE (S4C Fig).

As the ech2–1 mutation completely suppresses CCE in fugu5–1 [22], ech2–1 was introgressed into ibr10–1, and the cellular phenotypes in cotyledons from the double mutant were analyzed. Our results revealed that CCE did not occur in ibr10–1 ech2–1 despite having a significantly reduced cell number (Fig 2A and 2B). Moreover, exogenous Suc supply cancelled compensation in the ibr10–1 mutant background (Fig 2A and 2C). Analyses of another allele, ibr10–2 (SALK_201893C), confirmed our findings (S5A and S5B Fig).

Because cell number can be rescued by the application of exogenous Suc as in the fugu5 single mutants (Fig 2C; [23]), we next focused on key metabolites in the ibr10 mutants during seedling establishment. First, quantification of TAG content in dry seeds and etiolated seedlings revealed that although TAG breakdown in fugu5–1, ibr1–2, and ibr3–1 single mutants was relatively normal, it was significantly delayed in ech2–1, ibr10–1, and ibr10–2 (Fig 2D). Compared to the WT, fugu5–1 mutants produced less UDP-Glc, and thus less de novo Suc due to high cytosolic PPi, but TAG breakdown was almost unaffected [23,27]. However, in the ibr10 and ech2–1 mutants, both of which are defective in peroxisomal enzymes, TAG breakdown was markedly impaired (Fig 2D). To gain insight into the effects of the ibr10 mutations on gluconeogenesis, we quantified the Suc levels. Our measurements revealed that the Suc levels in the ibr10 mutants were decreased to nearly 60% of the WT (Fig 2E), suggesting that fatty acid β-oxidation was impaired in the ibr10 mutants. Surprisingly, Suc levels were either slightly increased or unchanged in ibr1-2 and ibr3-1, respectively (S6 Fig). Based on these findings, the ibr10 mutants could qualify as Class II CCE mutants, with similar properties to fugu5. ibr10–1 was used as a representative allele for the following analyses. Note that CCE in ibr10–1 was also suppressed in the ibr1,3,10–1 triple mutant background (Fig 1B), indicating the importance of IBA-derived IAA for ibr10–1 cell size increase.

IBA is transported over long distances by several carriers, including ABCG36/PEN3/PDR8 [42] and ABCG37/PDR9/PIS1 [42–44], both of which mediate IBA efflux through the plasma membrane (reviewed in [48,49]). These IBA transporters play important regulatory roles in cellular IBA levels (reviewed in [50]). The pen3–4 mutant displayed high-auxin phenotypes in several organs, including the cotyledons, indicating that cellular IAA levels within these organs are increased due to disrupted IBA outflow [42]. Therefore, we investigated the contribution of pen3–4 and pdr9–2 mutations on CCE.

To examine the effects of IBA accumulation on CCE, we generated pen3–4 fugu5–1, pdr9–2 fugu5–1, and pen3–4 pdr9–2 fugu5–1 mutant combinations and compared their cotyledon cellular phenotypes to fugu5–1. All double and triple mutants in the fugu5–1 background exhibited slightly bigger oblong cotyledons, reminiscent of fugu5–1, as indicated by the cotyledon size (Fig 3A) and cotyledon aspect-ratios (Fig 3B). On the other hand, while cell numbers in pen3–4, pdr9–2, and pen3–4 pdr9–2 were almost unaffected (Fig 3C), cell sizes were significantly larger than the WT (Fig 3D). Notably, cells in pen3–4 fugu5–1, pdr9–2 fugu5–1 were significantly larger than those in fugu5–1 single mutants (Fig 3D), indicating that CCE was enhanced due to the high IBA levels. In other words, the pen3–4 and pdr9–2 mutations promoted CCE. Hereafter, this phenotype will be referred to as overcompensated cell enlargement (OCE).

Based on the genetic evidence, CCE in fugu5 can either be enhanced or suppressed in an IBA-derived IAA-level dependent manner. In general, IAA signaling is mediated by AUX/IAA proteins and ARF transcription factors (TFs) [51–55]. In this study, we focused on ARF7 and ARF19 because they are strongly expressed in cotyledons from 1-week-old seedlings [56]. To address the role of IAA signaling in fugu5, we generated arf7–1 fugu5–1, arf19–1 fugu5–1 and arf7–1 arf19–1 fugu5–1 mutants and analyzed their cellular phenotypes.

Our results indicated that the cotyledon size in arf7–1 arf19–1 and arf7–1 arf19–1 fugu5–1 mutants was notably smaller when compared to WT or fugu5–1 (Fig 4A and 4B). Next, quantification of cotyledon cellular phenotypes revealed that all of the mutants in the fugu5–1 background displayed significantly decreased cotyledon cell numbers (Fig 4B). In addition, the cotyledonary palisade tissue cells in the arf7–1 fugu5–1 double mutant were slightly smaller than those in fugu5–1 (Fig 4B). Importantly, cell size in arf7–1 arf19–1 fugu5–1 was indistinguishable from the WT (Fig 4B). These findings indicate that an auxin signaling pathway involving both ARF7 and ARF19 might play a major role in driving CCE in fugu5–1.

Our results strongly suggest that IAA was the key molecule driving Class II CCE in fugu5. In addition, previous kinematic analyses had revealed that cell expansion peaks in fugu5–1 cotyledons around 15 days after sowing (DAS) [3]. To relate both findings, IAA concentration was quantified by a UPLC-Q-TOF-MS system using cotyledons dissected from seedlings at 10, 15 and 20 DAS. We found that the IAA concentration was 1.7-fold higher in fugu5–1 than in the WT at 10 DAS (Fig 5A). At 15 and 20 DAS, IAA concentration was not significantly different among fugu5–1 and the WT anymore (Fig 5A). Next, we quantified IAA in a time-course manner in dry seeds, or in young seedling shoots collected at 2 days after induction of seed germination (DAI), and at 4, 6, 8, and 10 DAS. There was no significant difference in IAA concentration between the WT and fugu5–1 in the dry seeds and seedlings collected at 2 DAI, 4 and 6 DAS (Fig 5B). Surprisingly, at 8 DAS, fugu5–1 seedlings contained 2.2-fold more IAA that the WT (Fig 5B).

To validate the above results, IAA concentration was further quantified in higher-order mutant combinations, namely ibr1,3,10–1, ibr1,3,10–1 fugu5–1, and ibr1,3,10–1 pen3-4 fugu5–1, in which CCE is suppressed (Figs 1, S7A and S7B). We found that IAA levels in all the above higher-order mutant lines were comparable to the WT (S8 Fig). This suggests that the accumulation of IAA in cotyledons which occurs in fugu5–1 at 8–10 DAS (Fig 5A and 5B) is a key transition point to drive CCE in fugu5 cotyledons.

Consistent with its suppressive effect on fugu5 CCE [22], the ech2–1 mutation completely suppressed CCE in icl–2, mls–2, pck1–2 [25] and ibr10 (Figs 2 and S5).

Next, to validate the role of IBA in Class II CCE, we constructed double mutants between the above mutants and pen3–4. Our results indicate that while the pen3–4 mutation triggered OCE in icl–2, mls–2, and pck1–2, in ibr10–1 CCE remained unaffected in the pen3–4 background (Fig 6). Taken together, these results further confirm that Class II CCE is promoted by IBA-derived IAA.

Vacuoles are not only the largest organelles of mature plant cells, they also play pivotal roles in plant growth and development, notably through the regulation of turgor pressure. This might involve vacuole acidification through the V-ATPase, downstream of auxin signaling [45,46]. To investigate the putative contribution of V-ATPase-dependent vacuole acidification on fugu5 CCE, we analyzed the vha-a2 vha-a3 fugu5–1 triple mutant. Note that the vha-a2 vha-a3 fugu5–1 triple mutant has no H⁺-pumping activity across the tonoplast, resulting in nearly null ΔpH [57]. Strikingly, the vha-a2 vha-a3 [57,58], which harbors mutations in two a-subunits of the V-ATPase, completely suppressed CCE in the fugu5–1 background (Fig 7). This suggests that the H⁺ pumping activity via the V-ATPase complex is essential for CCE.

A longstanding question in biology is how organ size is determined [4,10]. How do organs know when to stop growing? Is size merely a function of the proliferative growth of individual cells, or is it rather determined by a global control system that acts at the level of the whole organ?

In the simplest scenario, leaf size would be a function of cell number and size. However accumulating evidence suggests that a severe decrease in cell number usually triggers excessive cell enlargement post-mitotically. The inverse relationship that exists between the two major determinants of leaf size, namely cell number and size, was first described by the so-called “compensation” [9]. In other words, compensation highlights the coupling between cell division and expansion at the level of the entire organ. Over the last few years, the number of compensation-exhibiting mutants has increased, suggesting that compensation might reflect a general size regulatory mechanism in plants. Here we focus on the fugu5 and report how cell expansion can compensate for lower cell division during morphogenesis, whereby IBA-to-IAA conversion plays a key role, shedding light on the mechanisms coupling these processes during organ development.

Compensation consists of two interdependent stages: the induction stage and the response stage [16]. While the induction stage in fugu5 (i.e., the mechanism leading to the reduction in cell number) has been extensively investigated, the response stage mechanism that mediates CCE remains largely unknown. However, two studies have provided some insights into this stage. In the first study, genetic screening and phenotypic analyses revealed that CCE in fugu5 cotyledons was specifically suppressed by ech2–1 mutations. Therefore, we proposed that CCE in fugu5 is driven by IBA-derived IAA [22]. The second study validated the above hypothesis and further demonstrated that for Class II CCE to occur, the peroxisomal IBA-to-IAA conversion was a prerequisite not only in fugu5 but also in icl–2, mls–2, and pck1–2 [25]. In this study, we analyzed the contribution of the phytohormone auxin behind CCE by generating higher-order mutant lines where endogenous level of IBA was specifically up- or downregulated. We also quantified the endogenous IAA levels in fugu5, analyzed the role of auxin signaling in this process by mutating key TFs (ARF7 and ARF19), and opening to the potential implication of the vacuole in CCE.

Collectively, our results revealed that while CCE was completely suppressed in ibr1,3,10–1 fugu5–1, it was significantly enhanced in pen3–4 fugu5–1 and pdr9–2 fugu5–1, resulting in OCE (Figs 1 and 3). Thus, it became evident that IBA-to-IAA conversion is a prerequisite for CCE to occur in fugu5–1.

Interestingly, the average cell size in ibr1,3,10–1 pen3–4 fugu5–1 was similar to that of the WT (S7 Fig). The ibr1,3,10–1 pen3–4 also displayed the low auxin phenotype, similar to that of ibr1,3,10–1. This occurred because additional IBA accumulation due to the pen3–4 mutation was not properly converted into IAA due to the ibr1,3,10–1 mutations [34]. The OCE observed in pen3–4 fugu5–1 was completely suppressed in the ibr1,3,10–1 pen3–4 fugu5–1 quintuple mutant, in which the cell size was reset to the WT levels (S7B Fig). Consistently, the cell size of the different ibr pen3–4 fugu5–1 triple mutant combinations were comparable to that of fugu5–1 (S9 Fig). Therefore, CCE in fugu5–1 is either suppressed or enhanced in an IAA concentration-dependent manner. Although IBA is not recognized by the auxin TIR1/AFBs receptor [59,60], it represents a precursor for IAA biosynthesis. Therefore, by serving as a source of auxin, IBA plays a major role in IAA supply sustaining CCE during fugu5 cotyledon development.

Auxin signaling regulates various developmental processes, such as cell division, cell growth, or cell differentiation via AUX/IAA proteins and ARF TFs. Previous studies have shown that ARF7 and ARF19 promote lateral and adventitious root formation [55,56], and redundantly promote leaf cell expansion [56].

To clarify the relationship between ARF7, ARF19, IAA and CCE, we used the arf7–1 arf19–1 fugu5–1 triple mutant because, as mentioned above, ARF7 and ARF19 are strongly expressed in cotyledons of 1-week-old seedlings and are involved in cell expansion control [56]. Cellular phenotypic analyses of the cotyledons of the above triple mutants revealed that only cell size was decreased, but cell numbers were unchanged (Fig 4B), indicating that CCE in fugu5–1 was completely suppressed by the arf7–1 arf19–1 mutations (Fig 4). Altogether, these findings indicate that the CCE in fugu5–1 cotyledons was triggered by IAA via the ARF7 and ARF19 TFs related signaling pathways.

The above findings were further supported by the increase in endogenous IAA concentration in fugu5–1, which was significantly higher than the WT, particularly at 8–10 DAS (Fig 5), the stage during which post-mitotic expansion is exponentially increased (see kinematic analyses in [3]). While IAA concentrations in fugu5 were trending upward (Figs 5A and 5B and S8), the differences in IAA concentration between the WT, fugu5-1 and higher order mutant lines with suppressed CCE were not statistically significant (S8 Fig). This discrepancy in IAA concentrations could be due to slight differences in plant growth between independent experiments, and/or to the homeostasis. In fact, as mentioned above, besides its highly mobile nature, endogenous IAA levels are maintained through complex, yet specific metabolic pathways that can release/store IAA from/into its inactive forms, or synthesize IAA de novo from several independent pathways [35–37,61]. Hence, although measuring such low hormone levels has been challenging, the variability inherent among independent experiments and different genetic backgrounds should be considered carefully.

Finally, vacuolar acidification mediated by V-ATPase complex has been shown to play a crucial role in fugu5 CCE (Fig 7), which corresponds with the role of turgor pressure in distended vacuoles during post-mitotic cell expansion [62–64]. According to the so-called “acid growth theory”, it is widely accepted that IAA activates the plasma membrane H⁺-ATPase, acidifies the apoplast and activates a range of enzymes involved in cell wall loosening ([65]; for a review see [66]). Although the above theory links IAA-mediated extracellular acidification and subsequent cell wall loosening to turgor-dependent cell expansion [67], little is known about the role of the tonoplast in the IAA-mediated growth of plant cells. Importantly, increased vacuolar occupancy has been recently shown to allow cell expansion through a mechanism that requires LRXs and FER module, which senses and conveys extracellular signals to the cell to ultimately coordinate the onset of cell wall acidification and loosening with the increase in vacuolar size [68]. More recently, it has been reported that plant vacuoles significantly increase their volume (ca. two-fold) upon incubation with 1 μM IAA [45,46], linking auxin and vacuolar acidification, and providing insights into vacuole volume regulation during post-mitotic plant cell expansion.

Based on the above results, the major events involved in the compensation process in fugu5 can be summarized as shown in our working model (Fig 8). First, during germination, excess cytosolic PPi inhibits Suc synthesis de novo from TAG, causing a significant decrease in the number of cells in fugu5 cotyledons [23]. Second, the metabolic reaction of IBA-to-IAA conversion is somehow promoted, resulting in increased IAA concentration at 8–10 DAS (Fig 5A and 5B), which seems to be a crucial transition point to drive CCE in fugu5 cotyledons. Third, one can assume that high endogenous IAA triggers the TIR/AFB-dependent auxin signaling pathway through ARF7 and ARF19, and subsequently activates the vacuolar type V-ATPase leading to an increase in turgor pressure. This ultimately triggers cell size increase and CCE (Fig 8). Although the above scenario is plausible, its robustness still needs to be challenged experimentally in the future.

Our work links CCE, auxin and metabolism. Indeed, mobilization of TAG is essential for Arabidopsis seed germination and seedling establishment [26,69]. Because ibr10 etiolated seedlings were as long as the WT, IBR10 was thought not to be involved in the degradation of seed stored lipids [34,39]. However, our analyses focusing on cotyledons revealed that both ibr10 mutant alleles displayed slow FAs degradation (Fig 2D), produced less Suc de novo from TAG (Fig 2E), and exhibited CCE (Fig 2B). On the contrary, the fact that ibr1–2 and ibr3–1 have no defects in TAG degradation and Suc synthesis, may suggest that they are functionally different from ibr10 with respect to TAG mobilization (Figs 2D and S6). On the other hand, our results indicated that while CCE was not suppressed in ibr1-2 ibr3-1 fugu5-1, ibr1-2 ibr10-1 fugu5-1, and ibr3-1 ibr10-1 fugu5-1, it was totally suppressed only in ibr1,3,10 fugu5-1 quadruple mutants (Figs 1B and S7B). Altogether, the above results suggest that the loss of function of all possible ibr double mutant combinations in fugu5 background are not sufficient to suppress CCE, and point to their redundant enzymatic role in IBA-to-IAA conversion. Although it remains unclear how IBR10 contributes to FAs degradation, our findings suggest that IBR10 plays a pivotal role during stored lipid mobilization in Arabidopsis.

The icl–2, mls–2, and pck1–2 mutants, which have a metabolic disorder in the TAG to Suc pathway, displayed short etiolated seedling phenotypes [25]. Although pck1–2 has the lowest Suc contents, we previously reported that hypocotyl elongation of pck1–2 etiolated seedlings was only mildly affected compared to icl–2 and mls–2 [25]. Classically, the elongation defects in etiolated seedlings of Arabidopsis have been ascribed to a deficit in endogenous Suc during seedling establishment [26]. However, our findings confirmed that the length of etiolated seedling does not necessarily reflect Suc availability, and that hypocotyl elongation in the dark is a more complex trait than was expected.

As described above, the ech2–1 mutation completely suppressed CCE in icl–2 mls–2, pck1–2, and ibr10–1 (Fig 2B; [25]). However, CCE in Class I mutants was not suppressed by the ech2–1 mutation [22]. These findings suggest that IBA-to-IAA conversion is involved explicitly in driving Class II CCE. Each of the Class II CCE mutants has a metabolic disorder in the TAG-to-Suc pathway during the early stages of post-germinative development. Therefore, the low-sugar-state in seedlings somehow resulted in metabolic changes in the IBA-related metabolism. In brief, Class II compensation is fully controlled by metabolic networks. Indeed, Suc produced by photosynthesis is converted into phosphoenolpyruvic acid (PEP), which in Arabidopsis, is subsequently converted into tyrosine, phenylalanine, and tryptophan by the shikimic acid pathway [70]. On the other hand, IAA is a metabolite synthesized from the aromatic amino acid tryptophan [71]. This suggests that the gluconeogenesis pathway and the IAA biosynthesis pathway are tightly related. To this end, it would be interesting to revisit the shade avoidance response mechanism based on the above findings because the low-sugar-state under a low-light regime might trigger IBA-to-IAA conversion and concomitant cell elongation.

Living organisms produce thousands of metabolites at varying concentrations and distributions, which underlies the complex nature of metabolic networks. In this study, several key metabolites were identified to play a role in cell/cotyledon size control. Our findings provide the basis for further research to elucidate the mechanisms of organ-size regulation in multicellular organisms. More specifically, while organ-size regulation has been interpreted by specific sets of key TFs over the last decades, applying metabolomics might be a useful approach in efforts aiming to identify metabolites playing key roles in organ-size control in other kingdoms.

The WT plant used in this study was Columbia-0 (Col-0), and all of the other mutants were based on the Col-0 background. Seeds of ibr1–2, ibr3–1, ibr10–1, pen3–4, and ech2–1 were a gift from Professor Bonnie Bartel (Rice University). icl–2, mls–2, and pck1–2 mutants seeds were a gift from Professor Ian Graham (The University of York).

Seeds were sown on rockwool (Nippon Rockwool Corporation), watered daily with 0.5 g L^-1 Hyponex solution and grown under a 16/8 h light/dark cycle with white light from fluorescent lamps at approximately 50 μmol m^-2 s^-1 at 22°C.

Sterilized seeds were sown on MS medium (Wako Pure Chemical) or on MS medium with 2% (w/v) Suc where indicated, and solidified using 0.2–0.4% (w/v) gellan gum to determine the effects of medium composition on plant phenotype. After sowing the seeds, the MS plates were stored at 4°C in the dark for 3 d. After cold treatment, the seedlings were grown either in the light (for the cellular phenotype analyses) or in the dark (for Suc or TAG quantification) for the designated periods of time.

fugu5–1 was characterized as the loss-of-function mutant of the vacuolar type H⁺-PPase [23]. fugu5–1, icl–2, mls–2, pck1–2, ech2–1, ibr1–2, ibr3–1, ibr10–1, and pen3–4 were genotyped as described previously (S1 Table; [25,28–30,33]). ibr10–2 and pdr9–2 were obtained from the Arabidopsis Biological Resource Center (ABRC/The Ohio State University) and genotyped using specific primer sets (S1 Table). fugu5–1 plants were crossed with other mutants to obtain double, triple, quadruple or quintuple mutants, and the genotypes of the higher-order mutants were checked in the F₂ plants using a combination of PCR-based markers.

Photographs of the gross plant phenotypes at 10 DAS were taken with a stereoscopic microscope (M165FC; Leica Microsystems) connected to a CCD camera (DFC300FX; Leica Microsystems), and those at 25 DAS were taken with a digital camera (D5000 Nikkor lens AF-S Micro Nikkor 60 mm; Nikon).

Cotyledons were fixed in formalin/acetic acid/alcohol and cleared with chloral solution (200 g chloral hydrate, 20 g glycerol, and 50 mL deionized water) to measure cotyledon areas and cell numbers, as described previously [72]. Whole cotyledons were observed using a stereoscopic microscope equipped with a CCD camera. Cotyledon palisade tissue cells were observed and photographed under a light microscope (DM-2500; Leica Microsystems) equipped with Nomarski differential interference contrast optics and a CCD camera. Cell size was determined as the mean palisade cell area, detected from a paravermal view, as described previously [23]. The cotyledon aspect ratio was calculated as the ratio of the cotyledon blade length to width.

The quantities of seed lipid reserves in dry seeds and in 1-, 2-, 3-, and 4-day-old etiolated seedlings were measured by determining the total TAG using the Triglyceride E-Test assay kit (Wako Pure Chemicals). Either 20 dry seeds or 20 seedlings were homogenized with a mortar and pestle in 100 μL sterile distilled water. The homogenates were mixed with 0.75 mL reaction buffer provided in the kit, as described previously [23,73]. The sample TAG concentration was determined according to the manufacturer’s protocol. The length of the etiolated seedlings was determined as described previously [23].

Etiolated seedlings at 3 DAI were collected in one tube in liquid nitrogen and were freeze-dried. Then the samples were extracted using a bead shocker in a 2 mL tube with 5 mm zirconia beads and 80% MeOH for 2 min at 1,000 rpm (Shake Master NEO, Biomedical Sciences). The extracted solutions were centrifuged at 10⁴ g for 1 min, and 100 μL centrifuged solution and 10 μL 2 mg/L [UL-¹³C₆^glc]-Suc (omicron Biochemicals, USA) were dispensed in a 1.5 mL tube. After drying the solution using a centrifuge evaporator (Speed vac, Thermo), 100 μL Mox regenet (2% methoxyamine in pyridine, Thermo) was added to the 1.5 mL tube, and the metabolites were methoxylated at 30°C and 1,200 rpm for approximately 6 h using a thermo shaker (BSR-MSC100, Biomedical Sciences). After methoxylation, 50 μL 1% v/v of trimetylchlorosilane (TMS, Thermo) was added to the 1.5 mL tube. For TMS derivatization, the mixture was incubated for 30 min at 1,200 rpm at 37°C as mentioned above. Finally, 50 μL the derivatized samples were dispensed in vials for GC-QqQ-MS analyses (AOC-5000 Plus with GCMS-TQ8040, Shimadzu Corporation). Suc and [UL-¹³C₆^glc]-Suc were detected in the multiple reaction monitoring (MRM) mode. MRM transitions were Suc-8TMS, 361.0 > 73.0; [UL-¹³C₆^glc]-Suc-8TMS, 367.0 >174.0 (parent > daughter). The GC-QqQ-MS analyses were carried out as follows: GC column, BPX-5 0.25 mm I.D. df = 0.25 μm × 30 m (SGE); insert, split insert with wool (RESTEK); temperature of GC vaporization chamber, 250°C; gradient condition of column oven, 60°C for 2 min at the start and 15°C/min to 330 for 3 min of hold; injection mode, split (1:30); carrier gas control, 39 cm²/s; interface temperature of MS, 280°C; ion source temperature of MS, 200°C; loop time of data collection, 0.25 second. Raw data collection and calculation of the GC-MS peak area values were carried out using GCMS software solution (Shimadzu Corp., Kyoto, Japan). Suc contents were quantified per etiolated seedling using the internal standard method.

Endogenous IAA was extracted with 80% (v/v) acetonitrile containing 1% (v/v) acetic acid from whole WT and mutant seedlings after freeze-drying. IAA was purified using a solid-phase extraction column (Oasis WAX, Waters Corporation, Milford, MA, USA) and the IAA levels were determined using a quadrupole/time-of-flight tandem mass spectrometer (Triple TOF 5600, SCIEX, Concord, Canada) coupled with the Nexera UPLC system (Shimadzu Corp., Kyoto, Japan). Conditions for purification, LC and MS/MS analyses were previously described [74].

In this study, the statistical analyses included the Student’s t-test, Dennett’s test, or Tukey’s honestly significant difference (HSD) test (R ver. 3.5.1; [75]). Multiple comparisons were performed using the multcomp package [76]. A bee swarm plot was created using the beeswarm package [77].