Diverse Allyl Glucosinolate Catabolites Independently Influence Root Growth and Development^,

Allyl-glucosinolate and its catabolites use multiple mechanisms to affect plant growth and development through specific responses that are optimal to any given environment.

Abstract

Glucosinolates (GSLs) are sulfur-containing defense metabolites produced in the Brassicales, including the model plant Arabidopsis (Arabidopsis thaliana). Previous work suggests that specific GSLs may function as signals to provide direct feedback regulation within the plant to calibrate defense and growth. These GSLs include allyl-GSL, a defense metabolite that is one of the most widespread GSLs in Brassicaceae and has also been associated with growth inhibition. Here we show that at least three separate potential catabolic products of allyl-GSL or closely related compounds affect growth and development by altering different mechanisms influencing plant development. Two of the catabolites, raphanusamic acid and 3-butenoic acid, differentially affect processes downstream of the auxin signaling cascade. Another catabolite, acrylic acid, affects meristem development by influencing the progression of the cell cycle. These independent signaling events propagated by the different catabolites enable the plant to execute a specific response that is optimal to any given environment.

Pathogens and herbivore attacks are critical life-long threats to any plant. To survive these attackers, plants have developed a variety of defense mechanisms and resistance strategies, including the production of defensive chemicals (Chen, 2008). However, indiscriminate use of these defense chemicals by the plant can have detrimental effects on growth or can introduce ecological costs by attracting specialized attackers (Hartmann, 2004). Therefore, maximizing the effectiveness while limiting the detriments of these defense chemicals, and the plant immune system in general, requires that they be produced in the proper tissue, cell, and developmental stage. This requires a central coordination with the developmental programming of the plant, though the nature of this coordination is yet to be fully understood (Agrawal et al., 1999; Strauss and Agrawal 1999; Campos et al., 2016; Kliebenstein, 2016; Guo et al., 2018). While previous models assumed that this coordination was a simple tradeoff between growth and defense, it is now clear that growth, development, and defense are in a more complex relationship with the potential for synergism (Coley et al., 1985; Kliebenstein, 2016; Wasternack 2017). For example, any cost of defense metabolism is likely at most temporary, to allow the plant to deal with the immediate threat, and increases the long-term fitness potential.

To fully understand the connection between defense and growth requires understanding how the two sides of this equation mutually interact. Development is well known to influence defense, as there are dramatic changes in a plant’s defense arsenal during its life cycle to adjust to the different biotic attackers prevalent at each stage. For example, plants alter physical defenses like trichomes and spines across their development. Further, plants dramatically shift their chemical defense throughout development (Barton and Koricheva, 2010). A well-studied example of how chemical defenses change across the plant life-cycle is displayed by the glucosinolates (GSLs), key insect and pathogen defense metabolites in cruciferous plants including the model plant Arabidopsis (Arabidopsis thaliana; Beekwilder et al., 2008). GSL accumulation and composition dramatically changes throughout the life of a plant and can alter the plant’s sensitivity to pathogens or insects (Brown et al., 2003; Korves and Bergelson, 2004; Wentzell et al., 2008; Wentzell and Kliebenstein, 2008; Clay et al., 2009; Hopkins et al., 2009; Wittstock and Burow, 2010). In addition to influencing chemical defenses across time, development also influences defense patterning within a specific tissue by creating nonuniform defense distributions, which makes it more difficult for the pathogen to establish an efficient and effective attack (Shelton, 2004, 2005; Shroff et al., 2008; Kliebenstein, 2013). Several recent works identify mechanisms by which development influences defense (Li et al., 2004; Shelton, 2004, 2005; Melotto et al., 2008; De Bruyne et al., 2014).

In addition to development modulating defense, there is a growing understanding that defense outputs, like defense metabolites, can modulate development. For example, GSL studies in Arabidopsis and other Brassicaceae have started to highlight the potential for defense metabolites to directly influence growth and development. In the Raphanus genus, seedling phototropism can be regulated by a local gradient of catabolism of a specific GSL by influencing the TRANSPORT INHIBITOR RESPONSE (TIR1) auxin receptor (Hasegawa et al., 2000; Yamada et al., 2003). Similarly, the indole GSL catabolism product indole-3-carbinol, produced during pathogen attacks, inhibits auxin signaling by competing with auxin for binding the TIR1 auxin receptor, leading to growth arrest (Katz et al., 2015b; Katz and Chamovitz, 2017). Further highlighting the potential for defense metabolites to influence growth-related signaling pathways, 3-hydroxypropylglucosinolate (3OHPGSL) interacts with the Target of Rapamycin pathway (a central developmental regulator), which enables it to inhibit growth in both plants and fungi (Malinovsky et al., 2017). While these studies indicate the potential of defense metabolites to influence growth and development signaling in plants, the mechanisms behind the coordination of this influence are unknown. One question that has not been assessed is whether a single defense metabolite creates a single mechanistic response or signal, or whether single defense metabolites might branch into multiple signals or responses.

To gain a better understanding of how plant defense metabolites may integrate growth and defense, we studied how allyl-GSL influences growth in Arabidopsis. Allyl-GSL has been linked to resistance to numerous insects and pathogens and is an attractant to specialist insects that are adapted to the Brassicaceae (Lankau, 2007). Although allyl-GSL is one of the most widespread aliphatic GSLs within the Brassicaceae, the Brassicaceae also display extensive variation for the presence or absence of allyl-GSL due to independent loss of the 2-oxoglutarate-dependent dioxygenases2 (AOP2) enzyme required for its biosynthesis. This diversity suggests that the accumulation of allyl-GSL may be detrimental in some environments. This is further supported by the observation that Brassica genotypes accumulating high levels of allyl-GSL are less competitive in comparison to low-accumulating genotypes (Lankau, 2007). In addition, exogenous or endogenous introduction of allyl-GSL into Arabidopsis accessions without allyl-GSL leads to a reduction in growth (Wentzell and Kliebenstein, 2008; Burow et al., 2015; Francisco et al., 2016a; Urbancsok et al., 2017). Allyl-GSL’s effects on growth are not limited to Brassicaceae; this metabolite is also linked to allelopathic effects against diverse plants in other plant families (Bialy et al., 1990; Vaughn and Berhow, 1999; Vaughn et al., 2006; Lankau, 2007; Uremıs et al., 2009). However, the mechanism by which allyl-GSL functions, and which bioactive compound(s) the most likely responsible, are unknown. One possible mechanism for these effects is that allyl-GSL induces localized stomatal closure in response to a wound and alters the circadian periodicity of Arabidopsis, likely via isothiocyanate (Zhao et al., 2008; Khokon et al., 2011). An alternative possibility is suggested by a genome-wide association study that found an association between natural variation in auxin signaling genes and altered growth in response to allyl-GSL in Arabidopsis (Francisco et al., 2016a, 2016b).

To develop a deeper mechanistic understanding of how allyl-GSL modulates plant growth, we focused on the effect of allyl-GSL and its associated catabolites on Arabidopsis root growth and root morphology. Allyl-GSL effects were dependent on the specific catabolite, with three different allyl-GSL-derived catabolites activating at least two distinct signaling processes. One pathway led to altered cell-cycle progression, while the other pathway functioned downstream of auxin perception to modulate the PIN-FORMED (PIN) protein distribution. With each catabolite influencing a different component of root growth and development, the plant may be able to respond to the specific biotic events that are influencing defense metabolism. Thus, this could allow the plant to create a specific response that is optimal to any given environment. These results extend our understanding of how plants integrate growth and defense and thrive under changing biotic environments.

RESULTS

Allyl-GSL and Auxin Coordinately Influence Growth

Arabidopsis accessions present extensive variation in the concentration of endogenous allyl-GSL, which ranges between hundreds and thousands of micromoles in fresh weight during different growth stages (Table 1; Kliebenstein et al., 2001c; Chan et al., 2011). About one-third of natural Arabidopsis accessions, such as ecotype Columbia (Col-0), do not produce endogenous allyl-GSL due to a natural knockout in the AOP2 enzyme (Kliebenstein et al., 2001c; Chan et al., 2011), but when the AOP2 enzyme is reintroduced into these genotypes they produce ∼150 μm of endogenous allyl-GSL and have reduced growth (Table 1; Francisco et al., 2016a). In addition to their response to endogenous allyl-GSL, Col-0 plants growing on exogenous allyl-GSL will accumulate allyl-GSL and show a typical responsiveness to allyl-GSL in comparison to other accessions (Table 1; Francisco et al., 2016a; Jeschke et al., 2019). In our experiments, we used the Arabidopsis Col-0 accession, as it provides a clean background, with no endogenous compound to confound our experiments. Based on the multiple lines of evidence presented in Table 1, we decided to work with a concentration of 50 μm, which is lower than the endogenous allyl-GSL concentration produced by other accessions and also lower than the endogenous concentration produced by AOP2::Col-0 plants, and hence can be considered as physiologically relevant. As a control, we first tested whether allyl-GSL was taken up and accumulated in the plants under our experiments (Francisco et al., 2016a). We grew Col-0 seedlings on a medium with or without allyl-GSL, and after 14 d we measured the GSL content in the seedlings. As expected, accumulation of allyl-GSL was detected only in plants that grew on a medium supplemented with this molecule (Supplemental Fig. S1A).

Table 1. Allyl-GSL and raphanusamic acid concentrations.

Concentrations were determined by grams per fresh weight. DAG, Days after germination.

Tissue	Species/Genotype or Ecotype	Concentration	References
Allyl-GSL
Seeds	Arabidopsis/Radk-1	8,200 μm	E. Katz and D.J. Kliebenstein, unpublished data
Seeds	Arabidopsis/Hodja	380 μm	Kliebenstein et al., 2001b
Seedling, 7 DAG	Arabidopsis/CIBC-17	108 μm	Chan et al., 2011
Leaves, 3 weeks old	Arabidopsis/Kondara	519.8 μm	Kliebenstein et al., 2001b
Seedling, 15 DAG	Arabidopsis/AOP2	150 μm	Francisco et al., 2016a
Seedling, 9 DAG, grown on 50 μm allyl-GSL	Arabidopsis/Col-0	60 μm	Jeschke et al., 2019
Raphanusamic Acid
Seedling, 9 DAG	Arabidopsis/Col-0	2 μm	Jeschke et al., 2019
Leaves, 3 to 4 weeks old	Arabidopsis/Col-0	4 μm	Bednarek et al., 2009
Leaves, 3 weeks old	Arabidopsis/Col-0	27 μm	Sanchez-Vallet et al., 2010
Seedling, 9 DAG, grown on 50 μm allyl-GSL	Arabidopsis/Col-0	130 μm	Jeschke et al., 2019
Leaves, 3 weeks old	Arabidopsis/irx1-6: cell wall mutant with resistance to pathogens	85 μm	Sanchez-Vallet et al., 2010
Seedlings, 3 DAG	Broccoli (Brassica oleracea var italica) sprouts	150 μm	Palliyaguru et al., 2019
Leaves, 4 to 5 weeks old, inoculated with Botrytis cinerea	Crucihimalaya lasiocarpa	50 μm	Bednarek et al., 2011
Leaves, 4 to 5 weeks old, inoculated with Botrytis cinerea	Arabis alpina	85 μm	Bednarek et al., 2011

Previous genome-wide association mapping of genes that might influence the effect of allyl-GSL on plant growth showed enrichment in putative auxin signaling genes, suggesting a link between allyl-GSL responses and auxin (Francisco et al., 2016b). To investigate this potential connection, we analyzed the effect of allyl-GSL on two auxin-mediated responses, inhibition of root elongation and root curliness. To test whether allyl-GSL modulates the ability of auxin to inhibit root elongation, Arabidopsis Col-0 seedlings were grown vertically on Murashige and Skoog (MS) medium supplemented with different concentrations of allyl-GSL and auxin (indole-3-acetic acid [IAA]).

After 14 d we measured the length of the primary root. IAA and allyl-GSL individually inhibited root growth in a dose-dependent manner. Critically, at the higher allyl-GSL concentration, allyl-GSL and IAA had a synergistic interaction leading to higher root inhibition than expected from combining the effects of the individual treatments (P < 1.55E−08; Fig. 1, A and B). Using the same experimental design, we tested for an interactive effect of auxin and allyl-GSL on root curliness, an auxin-dependent phenotype (Dolan, 1998). Arabidopsis Col-0 seedlings were grown vertically on clean MS medium for 4 d, then transferred to medium supplemented with different concentrations of IAA, allyl-GSL, or both. After one additional day, the gravity stimulus was applied by tilting the plates to a 45° angle against the gravity vector (Okada and Shimura, 1990). After 3 d of growth in a tilted position, we counted the number of root curls within 1 cm of the root tip (Mochizuki et al., 2005). Again, independent addition of allyl-GSL or auxin decreased the number of curls in the roots (Fig. 1, C and D; Supplemental Table S1). This experiment shows that the addition of allyl-GSL causes the plant to behave as if it had been exposed to a higher dose of auxin. The inhibitory effect of auxin on root curls was amplified by the addition of allyl-GSL only at lower auxin concentrations, with no effect at higher auxin concentrations (Fig. 1, C and D). Statistical ANOVA showed a highly significant interaction between auxin and allyl-GSL, supporting the idea that they have a synergistic relationship (Supplemental Table S1). These results suggest that allyl-GSL inhibits auxin-linked phenotypes and that the two compounds have a synergistic interaction, indicating that they function via a similar pathway.

Figure 1.

Effect of allyl-GSL on root growth and auxin responses. A, Effect of allyl-GSL and auxin (IAA) on root length of Arabidopsis seedlings. Seedlings were grown vertically on MS medium supplemented with IAA (0.1–0.5 μm) or IAA and allyl-GSL (50 and 100 μm). Root length was measured at 14 d. Data are least-squared means over two independent experimental replicates with 15 to 18 seedlings per condition per experiment. Significance was tested via two-way ANOVA (for detailed statistics, see Supplemental Table S1). B, Phenotypes of 14-d-old plants grown with or without IAA and allyl-GSL. C, Effect of allyl-GSL and IAA on number of root curls within 1 cm of the root tip. Seedlings were grown vertically on MS medium for 4 d, then transferred to medium with IAA (0.1–0.5 μm) or IAA and allyl-GSL (50 μm). After another day the plates were tilted back 45° and photographed 3 d later. Results are least-squared means over two independent experimental replicates, with 15 to 18 seedlings per condition per experiment. Significance was tested via two-way ANOVA (***P < 0.0001 relative to control seedlings; quadratic curves were fitted to the model; for detailed statistics, see Supplemental Table S1). D, Phenotypes of seedlings grown with or without allyl-GSL.

Allyl-GSL Catabolites Inhibit Root Growth

While the above activities are linked to the application of allyl-GSL, it is possible that the allyl-GSL is converted into other compound(s) that mediate the identified activities. GSLs are broken down by myrosinase and various modifying enzymes into a diverse set of catabolites that are known to have different activities against biotic attackers (Xue et al., 1992; Halkier and Gershenzon, 2006). Thus, we proceeded to examine whether any common allyl-GSL catabolism products might be responsible for the observed root inhibition.

In roots of Arabidopsis, catabolism of allyl-GSL by myrosinase and nitrile specifier proteins results in formation of allyl-nitrile and allyl-isothiocyanate (Fig. 2A; Burow et al., 2008, 2009; Wentzell and Kliebenstein, 2008; Urbancsok et al., 2017). To test whether the effect of these catabolites is similar to allyl-GSL, seedlings were grown on MS medium supplemented with each of these molecules and IAA.

Figure 2.

Effect of allyl-GSL breakdown products and auxin on root length. A, Allyl-GSL biosynthesis and breakdown pathway. GSOH, GLUCOSINOLATE HYDROXYLASE; MSB, methylsulfinylbutyl; MTB, methylthiobutyl; NSP, nitrile specifier protein. B and C, Effect of allyl-GSL catabolites and IAA on root length. Arabidopsis seedlings were grown vertically on MS medium supplemented with IAA (0.1–0.5 μm) or IAA and the different allyl-GSL catabolites: 50 μm of allyl-GSL, allyl-nitrile, or allyl-isothiocyanate (allyl ITC; B) and 50 μm allyl-GSL, butenoic acid, acrylic acid, or raphanusamic acid (C). At 14 d old, their root lengths were measured. Results are least-squared means over at least two independent experimental replicates, with 15 to 18 seedlings per condition per experiment. Significance was tested via two-way ANOVA (***P < 0.0001 relative to control seedlings. Quadratic curves were fitted to the model; for detailed statistics, see Supplemental Table S1). D, Arabidopsis seedlings were grown vertically on clean MS medium or MS media supplemented with 0.5 μm IAA and 50 μm allyl-GSL, as indicated. After 7 d, the seedlings were transferred to MS medium with treatments as indicated. At day 14 they were photographed, and the root lengths were measured. Results are least-squared means over two independent experimental replicates, with 15 to 18 seedlings per condition per experiment. Significance was tested by Student’s t-test; ***P < 0.0001; error bars represent the means ± se. E, Seedlings were grown on MS medium with or without 50 μm allyl-GSL. Seven-day-old seedlings were stained with trypan blue and photographed. Two experiments were conducted, with 5 to 10 replicates in every experiment for each treatment.

Allyl-isothiocyanate had no detectable effect on root length with or without IAA (Fig. 2B). One potential explanation for this is that isothiocyanate is considered to be highly reactive and the molecule may have reacted with the media or the external cell walls, decreasing the potential concentration. In contrast, allyl-nitrile affected root length similarly to allyl-GSL (Fig. 2B). Roots of seedlings grown on medium supplemented with allyl-nitrile were shorter than roots of seedlings grown on the control medium. Further, the allyl-nitrile effect was dependent on the level of IAA in the medium (Fig. 2B; Supplemental Table S1). Higher concentrations of all allyl-GSL catabolites (100 μm) were tested and showed the same trends (Supplemental Fig. S2).

Allyl-isothiocyanate can be further converted, by separate catabolic pathways, to acrylic acid and raphanusamic acid (Fig. 2A; Bednarek et al., 2009; Wittstock et al., 2016; Piślewska-Bednarek et al., 2018). Correspondingly, allyl-nitrile can be converted by nitrilase to 3-butenoic acid, which has activity against aliphatic nitriles in Arabidopsis (Bartling et al., 1992, 1994; Vorwerk et al., 2001; Wajant and Effenberger, 2002; Burow et al., 2008, 2009; Piotrowski, 2008; Kissen and Bones, 2009; Kuchernig et al., 2012; Piślewska-Bednarek et al., 2018; Urbancsok et al., 2018).

Raphanusamic acid can be synthesized from other GSL-derived isothiocyanate (Bednarek et al., 2009); thus, Col-0 plants produce low concentrations of endogenous raphanusamic acid during different growth stages that correlate to total GSL amounts (Table 1; Bednarek et al., 2009; Sanchez-Vallet et al., 2010; Jeschke et al., 2019). When the plants are grown on exogenous allyl-GSL, higher concentrations of raphanusamic acid accumulate in the plant (Table 1; Jeschke et al., 2019). The Arabidopsis cell wall irx1-6 mutant shows high resistance to pathogens and presents high endogenous concentrations of raphanusamic acid (Table 1; Sanchez-Vallet et al., 2010). Different cruciferous plants produce concentrations of endogenous raphanusamic acid (50–150 μm) before and after pathogen attack (Table 1; Bednarek et al., 2011; Palliyaguru et al., 2019). These data suggest that working with a concentration of 50 μm (as for allyl-GSL) can be considered physiologically relevant for raphanusamic acid, especially as the Arabidopsis Col-0 accession is among the lower GSL-accumulating Arabidopsis accessions. As a control, we tested whether raphanusamic acid was taken up and accumulated in the plants under our experiments (Francisco et al., 2016a). As expected, accumulation of raphanusamic acid was detected in plants grown on a medium supplemented with this molecule (Supplemental Fig. S1B).

Separation and quantification of acrylic acid and butenoic acid is difficult due to the generic structure and low mass of these molecules, and data on endogenous concentrations of these catabolites do not exist. Given the accepted biosynthetic pathway for raphanusamic acid production from an isothiocyanate, it is predicted that there will be a 1:1 equivalency in the production of raphanusamic acid and carboxy acid precursor from every isothiocyanate (Bednarek et al., 2009; Wittstock and Burow, 2010; Wittstock et al., 2016; Piślewska-Bednarek et al., 2018). Therefore, we compromised on a uniform concentration of 50 μm for all of the catabolites.

To test the effect of these catabolites on root length, we grew seedlings on MS media with different concentrations of IAA and each compound. In contrast to allyl-isothiocyanate and allyl-nitrile, all three compounds strongly inhibited root length, and their effect was greater than with allyl-GSL, nitrile, or isothiocyanate (Fig. 2C). Further, there were two distinct inhibition trends. Raphanusamic acid showed an IAA interaction similar to allyl-GSL albeit with stronger effects. In contrast, butenoic acid and acrylic acid had strong effects on root growth across all IAA concentrations. These two molecules have similar structure; butenoic acid has a four-carbon backbone, while acrylic acid has a three-carbon backbone (Fig. 2C; Supplemental Fig. S2; Supplemental Table S1). Based on the root growth assays, we conclude that these three catabolites of allyl-GSL are all bioactive, and because of the stronger activity are probably closer to the actual active compounds responsible for the effects we observed when adding allyl-GSL. The presence of two different activities suggests that these molecules function via at least two mechanisms: butenoic acid and acrylic acid may affect root growth through a similar pathway, while raphanusamic acid works through a different one.

To determine whether allyl-GSL or one of its catabolites produce toxic compounds that lead to root inhibition by causing cell death, seedlings were grown on allyl-GSL or each catabolite for 14 d. All were still green and viable, arguing against toxicity effects. To further check the viability of the seedlings after allyl-GSL treatment, we tested whether the effect of allyl-GSL on root growth was reversible. Seedlings were grown vertically on a clean medium or a medium supplemented with allyl-GSL and IAA. After 7 d, the seedlings were transferred to the opposite medium, or to a new medium with the same supplementation. After an additional 7 d, the primary root length of the seedlings was measured. Roots of seedlings grown on a medium with allyl-GSL and IAA and then transferred to a clean medium were significantly longer than roots of seedlings grown on allyl-GSL and IAA for the entire experiment (Fig. 2D). This indicates that the seedlings were able to recover from the allyl-GSL treatment. We then used a more direct method to test whether allyl-GSL or any of its catabolites cause cell death. Seedlings were grown on each of the catabolites and stained with trypan blue, which selectively stains dead cells. Surveying roots from 10 seedlings grown in each of the treatments showed no evidence of cell death (Fig. 2E). As a positive control for the staining method, seedlings were treated with 500 mm of NaCl for 24 h, then imaged. These seedlings showed a strong blue coloration, indicating massive cell death (Supplemental Fig. S3). These results show that the application of allyl-GSL or its catabolites did not lead to cell death.

Allyl-GSL Catabolites Affect GSL Content

To further characterize how allyl-GSL and each of its catabolites affect roots, we checked whether they affected endogenous GSL accumulation in the plants. Endogenous GSL levels were measured from seedlings grown for 14 d on media supplemented with each catabolite. We found that allyl-GSL and butenoic acid affected the aliphatic 4-carbon GSL pathway (Fig. 3). This phenotype was defined by calculating the ratio of 4-methylsulfinylbutyl to 4-methylsulfinylbutyl + 4-methylthiobutyl ( Fig. 3). The amount of indolic GSL, synthesized from Trp was also impacted, hence presenting an effect on a parallel GSL pathway. Interestingly, only acrylic acid affected this phenotype, as seedlings that were grown on a medium with acrylic acid had a higher amount of indolic GSL (Fig. 3; Supplemental Table S1).

Figure 3.

Effect of allyl-GSL catabolites on GSL accumulation. Arabidopsis seedlings were grown on MS medium supplemented with 50 μm of allyl-GSL breakdown products. After 14 d, GSL content was measured using HPLC. Results are least-squared means over at least two independent experimental replicates, with two to three replicates per condition per experiment. Significance was tested via two-way ANOVA, followed by Dunnett’s test (*P < 0.05, **P < 0.01, and ***P <0.0001, relative to untreated control seedlings; for detailed statistics, see Supplemental Table S1). Error bars represent means ± se.

Allyl-GSL Catabolites Have Different Effects on the Root Meristem

To dissect the developmental process by which the allyl-GSL catabolites influence root growth, we measured their effects on root morphology. Inhibition of primary root growth can arise either from a reduction in the number of cells (as a result of an inhibition of cell division) or because each cell is smaller (Beemster and Baskin, 1998). As the root growth assays suggest that the catabolites are closer to the actual active compounds, we focused on testing their effects on root morphology. Seedlings were grown on MS medium supplemented with each compound, and the meristem size and the distance from the tips to the elongation zone of the seedlings were measured. The distance from the tip to the elongation zone was defined as the distance from the root tip to the first root hair. Seedlings grown on raphanusamic acid, butenoic acid, and acrylic acid had shorter elongation zones in comparison to seedlings grown on control media (Fig. 4A; Supplemental Table S1). We then measured the meristem size as the number of cells between the quiescent center and the first elongated cell. Only acrylic acid caused a significant reduction in the number of cells in the meristem compared to control roots (Fig. 4B; Supplemental Table S1). To test how these catabolites may interact with IAA signaling processes, we assayed how these catabolites affect auxin signaling in the meristem. For this purpose, we used plants expressing interaction domain II of Aux/IAA attached to a VENUS marker (DII-VENUS) that is sensitive to the presence of auxin in a dose-dependent manner (Brunoud et al., 2012). DII-VENUS seedlings were grown on MS medium and treated with one of the allyl-GSL catabolites. The seedlings were imaged using confocal microscopy, and the mean density of the VENUS fluorescence in the root was measured. The root meristem requires a specific concentration of auxin for proper development (Sabatini et al., 1999); hence, we quantified the mean fluorescence intensity only in the meristem area, as indicated in Figure 4C. Two hours following treatment with raphanusamic acid or butenoic acid, the DII-VENUS fluorescence intensity in the meristems of treated plants was significantly lower in comparison to the control meristem (Fig. 4D; Supplemental Table S1). In contrast, acrylic acid did not have a significant effect on the DII-VENUS fluorescence intensity. This indicates that the short treatment with raphanusamic acid or butenoic acid increased auxin-related signaling in the root meristem. In opposition to our hypothesis, these results indicate that even though acrylic acid and butenoic acid have similar root inhibition phenotypes, this likely happens via different mechanisms. This opens the door for a potential third mechanism that affects root length, demonstrating that each allyl-GSL catabolite may affect root inhibition by a different mechanism. We then continued to dissect the involvement of the catabolites in each one of these mechanisms.

Figure 4.

Effect of allyl-GSL breakdown products on meristem development. A, Arabidopsis seedlings were grown on MS medium with allyl-GSL breakdown products. At 5 d old, the seedlings were photographed, and the length from the root tips to the first root hair was measured. B, At 7 d, cell walls were stained using propidium iodide and imaged using confocal microscopy, and the numbers of cells from the quiescent center to the first elongated cell were counted. C, Confocal images of 7-d-old DII-VENUS seedlings stained with propidium iodide. The meristematic area as measured is defined with a deltoid shape (one row under the quiescent center + eight rows above the quiescent center − two rows on each side). D, Seven-day-old DII-VENUS seedlings grown on MS medium and treated for 2 h with allyl-GSL breakdown products in liquid MS medium. The seedlings were imaged using confocal microscopy. The relative integrated density of the VENUS fluorescence in the meristem area was quantified. E, A cell expressing stronger cyclin B1-1::GFP expression in the nucleus then in the cytoplasm. F, Percentage of G2 cells. Seven-day-old seedlings expressing cyclin B1-1::GFP were grown on MS medium, treated for 2 h with allyl-GSL catabolites in liquid MS medium, and imaged using confocal microscopy. Cells expressing stronger nuclear GFP intensity compared with the cytoplasm were counted, and their percentage with respect to the total number of cells expressing GFP was calculated. Results are least-squared means over two independent experimental replicates, with 5 to 10 seedlings per condition per experiment. Error bars represent means ± se. Significance was tested via two-way ANOVA (*P < 0.05 and ***P < 0.0001, relative to untreated control seedlings; for detailed statistics, see Supplemental Table S1).

Acrylic Acid Affects Meristem Development

The presence of fewer cells in the root meristem after acrylic acid catabolite exposure suggested that this compound may affect cell cycle dynamics. To test this hypothesis, we used plants expressing cyclin B1-1::GFP (Bush et al., 2015), a cell cycle reporter. Cyclin B1-1 is expressed in G2 and early M phases (Kobayashi et al., 2015), and the expression of this reporter gene indicates the progress of cells through mitosis. Stronger expression in the nucleus than in the cytoplasm is an indication that the cell is at a late G2 phase (Fig. 4E; Bush et al., 2015). These seedlings were grown on MS medium with or without acrylic acid, butenoic acid, or raphanusamic acid for 2 h, and then imaged using confocal microscopy. We counted the cells in the late G2 phase in each root, and calculated the number as a percentage of all cells expressing this reporter gene. The number of cells expressing this reporter gene was not significantly different following treatment with the different catabolites. However, the percentage of cells in the late G2 phase was significantly reduced in cells of plants treated with acrylic acid compared to the control cells (Fig. 4F). Butenoic acid treatment resulted in a minor decrease in this percentage, while raphanusamic acid treatment resulted in a slight increase in this percentage. However, neither was significant, in agreement with their lack of effect on the meristem cell number. This indicates that only acrylic acid, and not butenoic acid or raphanusamic acid, influences cell cycle.

Raphanusamic Acid Affects the Auxin Machinery

Since raphanusamic acid had a rapid effect on auxin signaling in the meristem and a synergistic interaction with auxin, we tested whether this catabolite influences different parts of the auxin signaling machinery. To do so, we used several auxin marker lines that represent different parts of the auxin machinery and analyzed the effect of raphanusamic acid on those markers at different time points. The different marker lines were grown on MS medium, treated with raphanusamic acid, and imaged every 30 min using confocal microscopy. At each time point, we quantified the fluorescence signal in the root tips (not only in the meristem area, as above) and calculated the signal relative to that in the untreated seedlings by dividing the fluorescence of the treated seedlings by the mean fluorescence of the untreated seedlings. Treatment for 60 min with raphanusamic acid resulted in a significant reduction in the DII-VENUS intensity compared to that in the untreated seedlings, indicating a quick increase in auxin signaling (Fig. 5, A and C). We then tested whether raphanusamic acid affects auxin transporters. We first used seedlings expressing the auxin transporter PIN1-GFP marker (Benková et al., 2003) and found a significant reduction in the intensity of this transporter within 60 min of exposure to raphanusamic acid (Fig. 5A; Supplemental Fig. S4B). Using seedlings expressing the pPIN7::PIN7-GFP auxin transporter marker, we found that the localization and amount of PIN7 were not affected significantly by raphanusamic acid treatment (Fig. 5A; Supplemental Fig. S4C; Blilou et al., 2005). We then checked how raphanusamic acid treatment might directly affect the TIR1 auxin receptor using seedlings expressing TIR1-VENUS (Wang et al., 2016). Raphanusamic acid significantly reduced the activity of this reporter 60 min following the treatment (Fig. 5A; Supplemental Fig. S4A). These experiments show that raphanusamic acid affects auxin signaling and transporters at ∼60 min following the treatment.

Figure 5.

Effect of raphanusamic acid and butenoic acid on auxin responses. A and B, Arabidopsis seedlings expressing the different auxin markers (indicated in the inset key) were grown on MS medium. At 7 d old, they were either left in clear medium or treated with 50 μm raphanusamic acid (A) or butenoic acid (B) and imaged using confocal microscopy. The integrated density of the fluorescence markers in the root tips was measured for each treatment at every time point, and the intensity was calculated relative to that of control seedlings (untreated). Results are least-squared means over two independent experimental replicates, with 5 to 10 seedlings per condition per experiment. Error bars represent means ± se. Significance was tested via two-way ANOVA (*P < 0.05, **P < 0.01, and ***P < 0.0001, relative to control untreated seedlings). C, Images of 7-d-old DII-VENUS seedlings treated with 50 μm of raphanusamic acid for 60 min or 50 μm of butenoic acid for 30 min or not treated (control). D, Yeasts expressing TIR1/AFBs and IAA protein were grown on a medium containing auxin (20 μm), or auxin and raphanusamic acid or butenoic acid (150 μm). The interaction between the proteins was analyzed after overnight incubation at 30°C. E, Arabidopsis seedlings were grown vertically on clear MS medium or MS medium supplemented with 50 μm of raphanusamic acid and butenoic acid. At 14 d old, the seedlings were photographed and their roots were measured. Results are least-squared means over two independent experimental replicates, with 15 to 18 seedlings per condition per experiment. Error bars represent means ± se.

Butenoic Acid Affects the Auxin Machinery

As butenoic acid also had a rapid effect on auxin signaling in the meristem, we tested how this molecule affects each of the auxin marker and receptor lines. Seedlings treated for 30 min with butenoic acid showed a significant reduction in the DII-VENUS intensity compared to untreated seedlings (Fig. 5, B and C). We also found a significant reduction in the intensity of the PIN1 transporter 60 min following the treatment (Fig. 5B; Supplemental Fig. S4B) and a significant reduction in the intensity of the PIN7 transporter in the root cap 90 min following the treatment (Fig. 5B; Supplemental Fig. S4C). Finally, we found that treatment with butenoic acid significantly reduced the activity of the TIR1-VENUS reporter 90 min following the treatment (Fig. 5B; Supplemental Fig. S4A). Using these marker lines, we conclude that butenoic acid affects the auxin machinery in a specific order, first affecting auxin signaling, then auxin transporters PIN1 and PIN7, and finally auxin receptor TIR1.

Our results clearly show that both raphanusamic acid and butenoic acid affect the auxin machinery, but the timing of the effects of each one of the catabolites on the different components of the auxin machinery is different. In combination with their different chemical structure, this suggests that raphanusamic and butenoic acid may have different molecular targets. To test this, we analyzed whether they interact to modulate root growth. In pharmacological assays, an interaction between two compounds suggests that they work through the same target (Jia et al., 2009). Seedlings were grown on medium either untreated or supplemented with butenoic acid, raphanusamic acid, or both, and root lengths were measured. ANOVA was used to test the effect of each defense catabolite individually, as well as their combined effect, on root length (Fig. 5E). Within these conditions, there was no detectable interaction between the two catabolites. In agreement with the auxin-related marker genes, this suggests that the two catabolites function through different targets. Hence, we conclude that although butenoic acid and raphanusamic acid both affect the auxin signaling machinery, they do so via independent mechanisms.

Raphanusamic Acid and Butenoic Acid Do Not Interact with the Auxin Receptors

As both butenoic acid and raphanusamic acid have a synergistic interaction with auxin with respect to root length, and both have a very quick effect on auxin signaling, an obvious hypothesis was that they interact with any of the TIR1/Auxin signaling F‐Box (AFB) auxin receptors. This hypothesis was supported by previous work suggesting that a specific GSL catabolite, indole-3-carbinol, interacts specifically with TIR1 at the same binding site as auxin (Katz et al., 2015b). Because the interaction between TIR1/AFBs and the Aux/IAA (IAA) family proteins is facilitated by auxin, we used a yeast two-hybrid (Y2H) measuring system to determine whether raphanusamic acid or butenoic acid alters this interaction (Calderón Villalobos et al., 2012). We tested the interaction between the receptors (TIR1 and AFB1-3) expressed in yeast (Saccharomyces cerevisiae; EGY48) and 10 different IAA proteins grown on media supplemented with different concentrations of auxin, and with or without raphanusamic acid or butenoic acid. Raphanusamic acid or butenoic acid by itself did not facilitate the TIR1/AFB and Aux/IAA interaction. Furthermore, adding these catabolites with auxin to the growth media did not change the interaction between the receptors in comparison to auxin alone (Fig. 5D; for additional concentrations and receptors, see Supplemental Fig. S5). We then performed in vitro pull-down assays with Myc:TIR1 and GST:IAA14 protein in the presence of auxin or auxin and either raphanusamic acid or butenoic acid (Supplemental Fig. S5). Again, raphanusamic acid and butenoic acid did not affect the auxin-facilitated interaction between TIR1 and Aux/IAA. These experiments suggest that neither raphanusamic acid nor butenoic acid interacts directly with the auxin receptors and that the observed effects are via different mechanism(s).

To genetically test the in planta interaction of the catabolites with the auxin machinery and its receptors, we measured their effect on root growth of the tir1-1 afb2-1 afb3-1 triple mutant. This mutant controls for the partial redundancy between TIR1 and the AFBs (Dharmasiri et al., 2005). We grew Col-0 and the triple mutant seedlings on medium with or without allyl-GSL, raphanusamic acid, and acrylic acid, measured their root length at day 7, then calculated the percentage of elongation compared to roots of untreated seedlings. The auxin triple mutant affected the sensitivity to allyl-GSL, but in contrast, it had no effect on the sensitivity to raphanusamic acid and acrylic acid (Supplemental Fig. S6). This supports the observation that acrylic acid does not work through the auxin pathway, and it suggests that allyl-GSL might have some additional auxin-related pathways that are yet to be identified.

Allyl-GSL Catabolites Affect Root Growth in Diverse Species

The previous suggestions that allyl-GSL may influence other plants led us to test whether this extended to the allyl-GSL-related catabolites. For this, we focused on the effect of allyl-GSL catabolites on the root growth of non-Brassicaceae plant species that do not produce GSL. We used four different species from different families (basil [Ocimum basilicum], Lamiales; dill (Anethum graveolens), Apiales; lettuce[Lactuca sativa], Asterales; and tomato [Solanum lycopersicum], Solanales) and grew them on a medium with or without raphanusamic acid, acrylic acid, or butenoic acid, and measured their root length after 5 d. All three allyl-GSL catabolites had an effect on the tested species, but there was specificity to the effects for each catabolite. Acrylic acid inhibited the root growth of dill, lettuce, and basil, with no effect on tomato. In contrast, butenoic acid inhibited the root growth of dill and lettuce, while stimulating growth in basil and having no influence on tomato. Finally, raphanusamic acid also inhibited root growth in dill and lettuce while dramatically inducing growth in tomato root (Fig. 6). These results show that the allyl-GSL catabolites can influence growth across a wide range of species (probably since it works through conserved mechanisms), and that, similar to the case for Arabidopsis, it is likely that there are three different mechanisms being targeted.

Figure 6.

Effect of allyl-GSL catabolites on root length of different species. Basil, dill, lettuce and tomato seedlings were grown vertically on MS medium with or without 50 μm of the different allyl-GSL catabolites. At 5 d old their root lengths were measured. Results are least-squared means over two independent experimental replicates, with 12 to 24 seedlings per species per treatment. Error bars represent means ± se. Lowercase letters represent statistically different values within species according to ANOVA followed by Tukey’s honestly significant (HSD) mean-separation test (P < 0.05).

DISCUSSION

Allyl-GSL Catabolites Work through Different Mechanisms

In this work we found that the defense metabolite allyl-GSL affects Arabidopsis root growth and development by three different catabolic products, with each compound having a unique regulatory effect on root development (summarized in Fig. 7). Of the catabolites tested, acrylic acid has the most unique mechanism. Acrylic acid was the only molecule that affected meristem development by influencing the cell cycle in the root tips. In contrast to acrylic acid, both butenoic acid and raphanusamic acid manipulated several steps of the auxin machinery. However, each molecule affected each of the tested parameters in a different order, suggesting that they work through different mechanisms. Supporting the idea that butenoic acid and raphanusamic acid are working through different pathways was the fact that they showed no pharmacological interaction. However, future work is required to identify the immediate targets of these compounds. A few connections between Arabidopsis GSLs and auxin were suggested in the past. One is the antagonistic character of an indolic GSL and auxin (Katz et al., 2015a, 2015b), and another shows that allyl isothiocyanate treatment induces the expression of 19 auxin-related genes (Kissen et al., 2016). Our results suggest that there are additional connections between Arabidopsis GSL and auxin. This raises the question of how many other links may exist given the diversity of GSLs across the Brassicaceae.

Figure 7.

Allyl-GSL catabolites inhibit root growth through different pathways. Raphanusamic acid and butanoic acid manipulate several steps of the auxin machinery, while acrylic acid influences the cell cycle in the root tips.

Do Different Environments Favor Different Molecules?

Following tissue rupture, potentially as a result of herbivory attack, GSLs will mix with the myrosinase enzyme and a variety of breakdown products will be generated. The generation of the different catabolic products is regulated by specific factors, including the expression of specific enzymes that differ across tissues and accessions (Brown et al., 2003; Halkier and Gershenzon, 2006; Wentzell and Kliebenstein, 2008; Fu et al., 2016). In Arabidopsis Col-0, the nitrile-specifier protein needed for production of nitriles from exogenously fed allyl-GSL is not expressed in the leaves but is present in the roots. This leads to allyl-isothiocyanate, the precursor of acrylic acid and raphanusamic acid, being the prevalent form in Arabidopsis Col-0 leaves when plants were fed exogenous allyl-GSL. In contrast, expression of the nitrile-specifier protein, and hence allyl-nitrile, the precursor of butenoic acid, is more common in Col-0 roots (Wentzell et al., 2008; Wentzell and Kliebenstein, 2008; Kissen and Bones, 2009; Urbancsok et al., 2017). Further, growing plants in crowded or uncrowded conditions can plastically alter the production of nitriles versus isothiocyanates in older plants (Wentzell and Kliebenstein, 2008). The fact that different GSL catabolites are favored in specific tissues and in specific conditions implies that the effect of allyl-GSL on development will be conditional on the tissue or environment in which the study occurs. Thus, different environments and situations in the plant’s life will favor the breakdown of allyl-GSL to different catabolites, creating different signals and consequently generating different developmental effects. This creates the opportunity for the plant to potentially use this to execute a specific response that is optimal to any given environment. As the complexity of the environment is greater than the number of metabolites in a plant, this suggests that individual metabolites within a plant should have multiple roles to maximize the efficiency and fitness of the plant (Kliebenstein 2018a). Furthermore, complex and less predictable defense mechanisms presented by the plant will probably decrease the rate of counteradaptation of the attacker (Hopper, 1999; Veening et al., 2008; Kliebenstein, 2018b). This can be a key aspect of allyl-GSL potential allelopathic effects, as each of the catabolites also has different effects on other plant species.

Future Perspectives

The role of GSLs as signaling molecules and their effects on developmental processes in the plant are now starting to be revealed, and the accumulation of specific GSLs can provide direct feedback regulation within the plant to calibrate defense and growth. Recent work shows that specific GSLs affect plant growth through different mechanisms, among them an interaction with the auxin machinery and the Target of Rapamycin pathway (Katz et al., 2015b; Malinovsky et al., 2017). Here we expand the list of potential mechanisms influenced by GSL and related catabolites. We show that one GSL catabolite affects cell cycle regulation, and the other two affect the auxin machinery, probably by different mechanisms than previously shown. Moreover, we showed that even one GSL can have multiple mechanisms by which it affects plant development. Thus, rather than a single link between defense metabolism and growth, there is a web of signals from defense metabolism that can influence growth. Future studies are needed to assess how this may occur in plants that do not contain GSLs, as other defense metabolites are known to affect plant growth and development (Hartmann, 2004).

The fact that the effect of allyl-GSL is dependent upon environmental conditions suggests that future studies aiming to dissect GSL mechanisms will have to consider a precise description of the behavior of each molecule under different environments and different conditions. Adding to this complexity is the fact that some of the catabolites can be synthesized by more than one precursor, rather than only from allyl-GSL. This is the case for butenoic acid, which can also be synthesized from but-3-enyl GSL (which is not produced in Col-0 plants), and raphanusamic acid, which can be synthesized from any GSL-derived isothiocyanate (Bednarek et al., 2009). Furthermore, raphanusamic acid has been linked to other plant processes, including chlorophyll content, across a wide range of plant species, including those with and without GSL production (Inamori et al., 1992). This implies that even though GSLs are young compounds, their catabolites can play a wider range of signaling roles than was previously considered. This raises the possibility that there is an interconnected network within plants that may measure multiple components of their metabolism to fine-tune plant development and defense.

The complexity of the role of defense metabolites in coordinating defense and growth is known. In this work, we present another layer to this complexity by showing that one metabolite can have multiple mechanisms by which it affects development under different environments. We show that allyl-GSL, which can regulate its catabolic products according to the changing biotic conditions, has multiple ways to affect plant growth and development. We propose that this allows the plant to sense the specific processes influencing defense metabolism and then enable a specific response that is optimal to any given environment. This may apply to other plant species and different defense metabolites also known to have different developmental effects.

MATERIALS AND METHODS

Plant Material and Growth Conditions

Arabidopsis (Arabidopsis thaliana) strains used in this work were all in the Col‐0 background. Transgenic lines used in this work were the following: DII-VENUS (Brunoud et al., 2012), PIN1:GFP (Benková et al., 2003), PIN3:GFP (Xu and Scheres, 2005; Zádníková et al., 2010), cyclin B1-1::GFP (Culligan et al., 2006), TIR1-VENUS (Wang et al., 2016), PIN7:GFP (Blilou et al., 2005), and tir1-1 afb2-1 afb3-1 (Dharmasiri et al., 2005).

Seeds were surface sterilized with 4.125% (v/v) sodium hypochlorite and 0.01% (w/v) Tween 20 for 15 min, then rinsed five times with distilled water. The seeds were next cultivated on petri dishes containing MS basal salt mixture, 1% (w/v) Suc (pH 5.8), and 0.8% (w/v) agar.

For each experiment, six seeds per row were sown ∼13 mm apart in six rows across a 36-grid-square 100 × 100 × 15 mm petri dish (Fisherbrand), with one seed per grid square. Each treatment was replicated across three independent plates within an experiment. Each experiment was conducted at least two independent times and the data were combined for analysis (Francisco et al., 2016a, 2016b). After planting, plates were stratified for two nights in the dark at 4°C to break dormancy.

Allyl-GSL [(−)-Sinigrin hydrate (S1647)], allyl-nitrile (122793), allyl-isothiocyanate (377430), 3-butenoic acid (134716), acrylic acid (147230), raphanusamic acid ([4r)-(−)-2-thioxo-4-thiazolidinecaboxylic acid (273449)], and IAA (I5148) were purchased from Sigma-Aldrich (catalog numbers in parentheses). To test the effect of the GSL catabolites the different chemicals were dissolved in distilled water and were added to the autoclaved MS medium to a final concentration of 50 μm (unless stated otherwise). All compounds were soluble under these conditions. The plates were placed vertically at 22°C under light/dark conditions of 16 h white light and 8 h darkness, with light at 100 to 120 μEi. For short treatments, GSL catabolites were added to liquid MS media with 7-d-old seedlings.

To test root curliness, seedlings were grown vertically on clean MS medium for 4 d, then transferred to medium supplemented with different concentrations of IAA, allyl-GSL, or both. After one additional day, the plates were tilted to a 45° angle against the gravity vector, and after 3 d of growth, the root curls within 1 cm of the tip were counted (Okada and Shimura, 1990; Mochizuki et al., 2005).

Root length was measured using ImageJ software (https://imagej.nih.gov/ij/).

Cell Death Detection by Trypan Blue Staining

Seedlings were harvested in acetic acid:ethanol solution (1:3 [v/v]) for 1 h under constant shaking, then in acetic acid:ethanol:glycerol (1:5:1 [v/v/v]) for an additional hour, then stained with trypan blue for 1 h (0.05% [w/v] trypan blue and 0.01% [w/v] aniline blue in lactic acid:phenol:distilled water:glycerol, 1:1:1:0.75 [v/v/v/v]; Chung et al., 2010). The staining solution was removed, and the seedlings were rinsed with distilled water before placing on microscope slides. For the positive control, 6-d-old seedlings were treated with 500 mm NaCl and stained with trypan blue after 24 h.

Confocal Microscopy

Seedlings were submerged in 0.005 mg mL⁻¹ propidium iodide in distilled water, placed on microscope slides, and imaged using a Zeiss LSM700 laser scanning microscope with 20×/NA 0.8. All pictures were taken with the exact same settings. Rainbow spectrum was applied for PIN1:GFP pictures. YFP/GFP fluorescence in the root tips was quantified using ImageJ software (https://imagej.nih.gov/ij/). The mean fluorescence was compared to the mean fluorescence of the untreated plants (control). The meristematic area (Fig. 4C) was defined as having a deltoid shape, with one row under the quiescent center plus eight rows above the quiescent center minus two rows on each side.

Measurement of GSL Content

GSLs were measured as previously described (Kliebenstein et al., 2001a, 2001b, 2001c). Briefly, four to six 14-d-old seedlings from the same petri plate were pooled, weighed and harvested in 400 μL of 90% [w/v] methanol. Tissues were homogenized for 3 min in a paint shaker and centrifuged, and the supernatants were transferred to a 96-well filter plate with DEAE sephadex. The filter plate with DEAE sephadex was washed once with water, once with 90% [w/v] methanol, and once more with water. The sephadex-bound GSLs were eluted after overnight incubation with 110 μL of sulfatase. Individual desulfo-GSLs within each sample were separated and detected by HPLC-DAD, identified, quantified by comparison to standard curves from purified compounds, and further normalized to the fresh weight.

Uptake of Raphanusamic Acid

Col-0 seeds were germinated on MS medium with or without 5 µm raphanusamic acid (MS medium contained 3 mm nitrogen and 1.68 mm sulfur). Single seedlings were harvested 5 d after germination and analyzed for raphanusamic acid as described (Jeschke et al., 2019).

Yeast Two-Hybrid Assays

Yeast two-hybrid experiments were performed as described previously (Prigge et al., 2010). The TIR1/AFB1-5 bait vector pGILDA and the IAA (IAA1, IAA3, IAA5–IAA8, IAA12, IAA14, IAA18, IAA19, IAA28, and IAA31) prey vector pB42AD (Prigge et al., 2010) were cotransformed into yeast (Saccharomyces cerevisiae; EGY48). The negative control comprised yeast expressing pGILDA or pB42AD empty vectors.

In Vitro Pull-Down Assays

In vitro pull-down assays were performed as described previously (Parry et al., 2009). The TIR1:Myc fusion protein was incubated with GST:IAA14 protein in the presence or absence of 50 μm IAA or 50 μm IAA plus 150 μm of GSL catabolite. Proteins were purified with glutathione agarose beads (Sigma-Aldrich; G4510), pulled down using monoclonal anti-c-Myc antibodies (Sigma-Aldrich; 11814150001), separated by SDS-PAGE, and detected using anti-GST (Sigma-Aldrich; G7781).

Statistical Analyses

Statistical analyses were conducted using R software (https://www.R-project.org/) with the RStudio interface (http://www.rstudio.com/). Significance was tested via two-way ANOVA using the “stat” package. Specific models are listed in Supplemental Table S1, but they followed the following general format: in each experiment, Treatment (application of allyl-GSL or associated catabolites) and Auxin (auxin concentrations ranging from 0 to 0.5 μm) were considered as fixed effects. Experiment and Plate were treated as random effects in the linear model.

Accession Numbers

Sequence data from this article can be found in the GenBank/EMBL data libraries under accession number AT4G03060 (AOP2).

Supplemental Data

The following materials are available in the online version of this article.

• Supplemental Figure S1. Uptake of allyl-GSL and raphanusamic acid by Col-0 seedlings.

• Supplemental Figure S2. Effect of allyl-GSL catabolites and auxin on root length.

• Supplemental Figure S3. Trypan blue staining.

• Supplemental Figure S4. Effect of raphanusamic acid and butenoic acid on auxin responses.

• Supplemental Figure S5. Effect of raphanusamic acid and butenoic acid on the interaction between TIR1/AFBs and Aux/IAA.

• Supplemental Figure S6. Effect of allyl-GSL catabolites on root length of the auxin mutant.

• Supplemental Table S1. Statistics for Figures 1 to 4.