Anthranilic acid regulates subcellular localization of auxin transporters during root gravitropism

The distribution of the phytohormone auxin within plant tissues is of great importance for developmental plasticity, including root gravitropic growth. Auxin flow is directed by the subcellular polar distribution and dynamic relocalization of plasma membrane-localized auxin transporters such as the PIN-FORMED (PIN) efflux carriers, which are in turn regulated by complex endomembrane trafficking pathways. Anthranilic acid (AA) is an important early precursor of the main natural plant auxin indole-3-acetic acid (IAA). We took advantage of an AA-deficient, and consequently IAA-deficient, mutant displaying agravitropic root growth to show that AA rescues root gravitropic growth at concentrations that do not rescue IAA levels. Treatments with, or deficiency in, AA result in defects in PIN polarity and gravistimulus-induced PIN relocalization within root cells. Taken together, our results reveal a previously unknown role for AA in the regulation of PIN subcellular localization and dynamics involved in root gravitropism, which is independent of its better-known role in IAA biosynthesis.


Introduction
The distribution of the phytohormone auxin in controlled concentration gradients within certain tissues and organs plays an important role in regulating the dynamically plastic growth and development of plants (Vanneste and Friml, 2009). Over the past couple of decades, an intense research effort has revealed many of the complex mechanisms by which plasma membrane-localized auxin carrier proteins are polarly distributed in order to direct the flow of auxin in plant tissues and maintain these gradients (reviewed by Luschnig and Vert, 2014;and Naramoto, 2017). These proteins, including the wellstudied PIN-FORMED (PIN) auxin efflux carriers, are remarkably dynamic in that they rapidly relocalize within the cell in response to signals, becoming more or less polar or shifting the direction of their polarity entirely. This dynamic responsiveness, which is facilitated by vesicular cycling and complex endomembrane trafficking pathways, is essential for altering the direction and strength of cell-to-cell auxin flow and redistributing auxin in response to external cues, thereby regulating cell and tissue growth and plasticity.
Root development in Arabidopsis thaliana has received particular attention as a model system demonstrating the importance of auxin gradients for plant growth and development (Clark et al., 2014). Mutations affecting auxin transporters often disturb root gravitropism and specific PIN proteins within the root tip have been shown to relocalize in response to changes in the gravity vector, leading to changes in auxin flow and consequently, organ growth adjustment (reviewed by Geisler et al., 2014). PIN2 in the root tip epidermis is particularly important for this response; being apically (shootward) polarized within epidermal cells (Müller et al., 1998), PIN2 transports auxin upwards in this tissue, contributing in a concerted manner together with other PIN proteins to the maintenance of the auxin maximum required in the root apical meristem for proper root development (Adamowski and Friml, 2015). However, in the case of a reorientation of the root to a horizontal position, PIN2 is rapidly redistributed from the plasma membrane to the vacuole within epidermal cells at the upper organ side (Kleine-Vehn et al., 2008;. This results in accumulation of auxin and consequent inhibition of cell elongation at the lower root side, leading to the root tip bending downwards towards the gravity vector. In the root columella, the cellular relocalization of PIN3 and PIN7 has also been shown to play an important role in root gravitropic growth responses. While these PIN proteins are generally apolar in columella cells, they redistribute towards the basal (rootward) plasma membranes upon reorientation of the root (Friml et al., 2002b;Kleine-Vehn et al., 2010), which is presumed to redirect the flow of auxin within the columella, thus further contributing to auxin accumulation at the lower root side.
Despite our knowledge on the mechanisms of PIN polarity and redistribution, a lot of information is still lacking regarding the regulation of PIN dynamics. In our previous work, we employed a chemical biology approach, whereby we isolated and characterized small synthetic molecules selectively altering the polarity of specific PIN proteins, to dissect the trafficking pathways involved in regulating their localization (Doyle et al., 2015a). This approach led us to identify a potential role for the endogenous compound anthranilic acid (AA) in PIN polarity regulation, which we investigate in the current study. AA is an important early precursor of the main natural plant auxin indole-3-acetic acid (IAA) (Maeda and Dudareva, 2012) and as auxin itself has been shown to regulate PIN polarity in a feedback mechanism to control its own flow (Paciorek et al., 2005), we hypothesized that AA may play a similar regulatory role. Herein, using Arabidopsis root gravitropism as a model system for auxin-regulated plastic growth, we provide strong evidence in favor of this hypothesis. Ultimately, we reveal a previously unknown role for the IAA precursor AA in the regulation of PIN polarity and relocalization required for root gravitropic responses and furthermore, we show that this role of AA is distinct from its well-known role in IAA biosynthesis.

AA rescues root gravitropic growth and length differently in an AA-deficient mutant
Using a chemical biology approach, we previously isolated the small synthetic molecule Endosidin 8 (ES8), which disturbs the polarity of selective PIN proteins in the root of Arabidopsis thaliana, leading to an altered root auxin distribution pattern and defective root growth (Doyle et al., 2015a). Intriguingly, the chemical structure of ES8 reveals that this molecule is an analog of the endogenous plant compound AA ( Fig. 1 A), a precursor of tryptophan (Trp), the main precursor of the predominant plant auxin IAA (Ljung, 2013;Zhao, 2014). This prompted us to question whether endogenous AA might play a role in growth and development of the root. We therefore investigated the root phenotype of a loss-of-function Arabidopsis mutant in both the ANTHRANILATE SYNTHASE SUBUNIT ALPHA1 gene (ASA1, also known as the WEAK ETHYLENE INSENSITIVE2 gene, WEI2) and the ANTHRANILATE SYNTHASE SUBUNIT BETA1 gene (ASB1, also known as WEI7). In the double mutant, wei2wei7 , the AA level is presumed to be reduced, which is supported by the rescue of an ethylene sensitivity phenotype in the single wei2-1 and wei7-1 mutants by treatment with AA . To confirm this, we analyzed the levels of several IAA precursors/catabolites in the mutant and the wild type (WT) Columbia-0 (Col-0), revealing that AA content was indeed significantly reduced in wei2wei7 compared to Col-0, as were the levels of the IAA precursors Trp, indole-3-acetonitrile (IAN) and indole-3-acetimide (IAM) (Fig. S1), most likely due to the decreased AA content. However, neither the IAA precursor tryptamine (Tra) nor catabolite 2oxoindole-3-acetic acid (oxIAA) showed altered content in the mutant compared to the WT (Fig. S1).
We were interested in the strong agravitropic and short phenotypes of wei2wei7 roots compared to WT Col-0 seedlings (Fig. 1 B and C), considering that ES8 treatment reduces both gravitropic root growth and root length in Col-0 (Doyle et al., 2015a). To investigate AA-mediated rescue of these root phenotypes in wei2wei7, we performed long-term AA treatments by growing WT and mutant seedlings on medium supplemented with a range of AA concentrations. In Col-0, none of the tested concentrations affected gravitropic root growth, while concentrations of 10 µM or more decreased root length in a dose-dependent manner (Fig. S2 A). As expected, in wei2wei7 both root gravitropic growth and length were rescued by AA ( Fig. 1 C and D), however we observed a striking difference between the AA rescue patterns of these two root phenotypes. While root gravitropic growth in the mutant was fully rescued to WT levels at all AA concentrations applied, root length rescue was concentration-dependent, with maximal rescue at 5 µM ( Fig. 1 D). Furthermore, root length in the mutant was only partially rescued at all AA concentrations applied ( Fig. 1 D), compared to WT root length ( Fig. S2 A). We hypothesized that the different rescue patterns of root gravitropic growth and length phenotypes by AA in wei2wei7 might reflect two different roles of AA, one known role in auxin biosynthesis and a distinct, as yet unknown role in regulating auxin distribution, considering that ES8 disturbs auxin distribution patterns in the root (Doyle et al., 2015a).
We next investigated whether ES8, as an analog of AA, could rescue either root gravitropic growth or length in wei2wei7. While long-term treatments with increasing concentrations of ES8 decreased both root gravitropic growth and length in a dosedependent manner in Col-0 ( Fig. S2 B), only the highest ES8 concentrations (15 and 20 µM) decreased root gravitropic growth and length in wei2wei7 ( Fig. 1 E). Moreover, while root length was not rescued in the mutant at any ES8 concentration ( Fig. 1 E), 5 µM ES8 partially rescued the root gravitropic phenotype of the mutant ( Fig. 1 C and E).
The partial root gravitropic rescue of wei2wei7 by ES8 without any effect on root length supported our hypothesis that the gravitropic rescue of wei2wei7 by AA may be due to a role of AA other than that in auxin biosynthesis.
To further test our hypothesis, we attempted to replicate these results using another analog of both ES8 and AA -ES8 analog no. 7 (ES8.7; Fig. S3 A; ES8 analogs ES8.1 to 6 were previously described by Doyle et al. (2015a)). We also tested the control compound ES8.7-Trp, in which the AA was exchanged for a Trp (Fig. S3 B). In Col-0, long-term ES8.7 treatment at a range of concentrations revealed a similar but weaker effect than ES8 on reduction of root gravitropic growth and length (Fig. S3 C).
Strikingly, ES8.7 had a stronger effect than ES8 in rescuing root gravitropic growth in wei2wei7, significantly rescuing this phenotype at a range of concentrations from 1 to 15 µM, with almost no effects on root length ( Fig. S3 A and D). Moreover, ES8.7-Trp had almost no effect on root gravitropic growth or length at any concentration in both Col-0 ( Fig. S3 E) and wei2wei7 (Fig. S3 B and F). These results strongly suggest that it is the AA part of the ES8 compounds, and not any other part of the molecules, that rescues gravitropic growth of wei2wei7 roots.
To investigate the possibility that the ES8 compounds might be degraded or metabolized during our experiments to release AA or Trp, we performed both short-term and long-term treatments of Col-0 and wei2wei7 seedlings with the ES8 compounds, followed by compound analysis (Fig. S4). We measured the concentrations of the relevant ES8 compound, AA or Trp and the non-AA or non-Trp part of the ES8 compound in planta as well as in ES8 compound-supplemented medium to which no seedlings were added. Our results showed that after short-term treatment (5 hours incubation in liquid treatment medium) with 5 µM ES8, high levels of ES8 were detectable in the seedlings and the seedling-free treatment medium remained at about 5 µM ES8 (Fig. S4 A). After long-term treatment (9 days of growth on solid treatment medium), the concentrations of ES8 in the seedlings and the seedling-free treatment medium had lowered considerably, suggesting degradation of ES8 over time. Compared to ES8, much lower levels of ES8.7 and ES8.7-Trp were present in the seedlings after short-term treatment ( Fig. S4 B and C), suggesting that ES8 may be more efficiently taken up into seedling tissues or ES8.7 and ES8.7-Trp may be degraded or metabolized during the short-term treatment. Degradation of ES8.7-Trp was supported by our measurements of its concentration in the seedling-free treatment medium, which had already lowered to 3.3 µM after short-term incubation and to 0.5 µM after long-term incubation (Fig. S4 C). Moreover, the levels of ES8.7-Trp in the seedlings were considerably lower after long-term compared to short-trem treatment. While these results suggest that ES8 and ES8.7-Trp are likely degraded over time, the levels of AA and Trp in the seedlings after ES8 compound treatment were not different to the levels after mock treatment and neither AA nor Trp were detected in the treatment medium samples (Fig. S4 D and E). Furthermore, we did not detect non-AA or non-Trp parts of the ES8 compounds at any time point, neither in the seedlings nor in the seedling-free treatment medium (Fig. S4 F). Therefore, the observed activities of the ES8 compounds are not due to the release of AA or Trp and are therefore unlikely to be due to increased IAA biosynthesis.

AA and ES8 can rescue root gravitropic growth in wei2wei7 without rescuing IAA level
As AA is a precursor of IAA, an essential regulator of root growth (Goh et al., 2014;Clark et al., 2014), we investigated the possibility that the rescue of root gravitropic growth by ES8 and AA might indirectly result from increased IAA biosynthesis. We therefore investigated the effects of these compounds on levels of IAA. First, we measured IAA concentrations in Col-0 and wei2wei7 seedlings after long-term treatments with different concentrations of AA. We found a significant reduction of IAA content in wei2wei7 compared to Col-0 in control conditions (Fig. 2 A). In Col-0, only treatment with 10 µM AA significantly increased the IAA level ( Fig. 2 A). Raised IAA concentrations are therefore the most likely reason why treatments of Col-0 with 10 µM and higher concentrations of AA resulted in significantly shorter roots (Fig. S2 A). While treatment of wei2wei7 with 1 or 10 µM AA rescued the IAA level to that of mock-treated Col-0, treatment with 0.5 µM AA had no effect on IAA content ( Fig. 2 A), despite this concentration having fully rescued root gravitropic growth and partially rescued root length in wei2wei7 seedlings ( Fig. 1 D). Next, we measured IAA content in Col-0 and wei2wei7 seedlings treated long-term with 5 µM ES8, ES8.7 or ES8.7-Trp.
While the IAA level was slightly but significantly reduced in mock-treated wei2wei7 compared to Col-0, none of the ES8 compounds significantly affected IAA content compared to mock treatment in either genotype (Fig. 2 B). Taken together, these results suggest that AA might play a role in the regulation of root gravitropic growth independently from its function in IAA biosynthesis.

Root gravitropic response is impaired by AA when its conversion to downstream IAA precursors is repressed
To test our hypothesis that AA may regulate root gravitropic growth via a role independent of IAA biosynthesis, we aimed to generate transformed Arabidopsis lines in which the gene encoding ASA1 (Niyogi and Fink, 1992) is constitutively overexpressed and that encoding PHOSPHORIBOSYLANTHRANILATE TRANSFERASE 1 (PAT1), which converts AA to the next downstream IAA precursor (Rose et al., 1992), is subject to estradiol-induced silencing. Of several homozygously  To dissect the effects of ASA1 overexpression and simultaneous silencing of PAT1 on the IAA biosynthetic pathway, we analyzed the levels of IAA and several IAA precursors/conjugates/catabolites in WT and AxP lines treated long-term with estradiol (grown on supplemented medium). We found that while AA levels tended to be higher in both AxP lines than in the WT, the levels of the precursors Trp, IAN and IAM were more variable between the two AxP lines (Fig. S6 A). Importantly, the IAA content was not affected in the AxP lines compared to the WT, while the conjugates IAA-aspartate (IAAsp) and IAA-glutamate (IAGlu) and the catabolite oxIAA showed reduced levels in the transformed lines compared to the WT (Fig. S6 A). These results suggest that simultaneous overexpression of ASA1 and silencing of PAT1 result in increased AA levels, but do not alter IAA levels, perhaps due to a reduced conversion of IAA to IAA conjugates and catabolites.
We next investigated the root phenotypes of the AxP lines. In control conditions without estradiol treatment, both lines displayed similar root gravitropic growth to that of the WT (Fig. S6 B), while having slightly shorter roots than the WT (Fig. S6 C).
After long-term estradiol treatment, the gravitropic growth of WT and AxP roots was slightly reduced, to a similar extent (Fig. S6 B). While long-term estradiol treatment significantly reduced the root length of all genotypes, the treatment affected the AxP lines more severely than the WT (Fig. S6 C). To analyze root gravitropic responses in the AxP lines, we turned the seedlings 90° and subsequently measured the gravistimulated root bending angles (Fig. S6 D). In control conditions, Col-0 and both AxP lines responded to the gravistimulus with a very similar range of root bending angles, with the majority of roots bending 75-105° (Fig. 3). Estradiol treatment inhibited the gravitropic response of Col-0 roots, reducing their bending angles, resulting in a reduction in the proportion of roots bending 75-105° and an increase in the proportion bending <75° (Fig. 3 A and B). The AxP lines, however, responded somewhat differently to estradiol than the WT. As for the WT, estradiol treatment resulted in both a reduction in the proportion of AxP roots bending 75-105° and an increase in the proportion bending <75° but additionally resulted in an increase in the proportion bending >105° (Fig. 3 C-F). Therefore, while estradiol treatment specifically reduces root bending in response to a gravitropic stimulus in the WT, the same treatment results in both under-and over-bending of roots in response to a gravistimulus in both AxP lines, strongly suggesting that increased AA levels in these lines interferes with root gravitropic responses.

PIN polarity in the stele is altered in wei2wei7 and partially rescued by ES8
ES8 has been shown to disturb auxin distribution patterns in the root by altering PIN polarity (Doyle et al., 2015a). Considering that IAA itself can influence its own transport by regulating PIN abundance at the plasma membrane (Paciorek et al., 2005;Robert et al., 2010), we reasoned that AA, as a precursor of IAA, might also play such a role. To investigate this possibility, we first studied the effects of long-term ES8 and AA treatments on the expression pattern of the auxin-responsive promoter DR5 in the root. To observe the effects of ES8 more easily, we used treatment at a high concentration of 15 µM, which led to a strong decrease in GFP signal in the stele of DR5::GFP WT roots ( Fig. S7 A), in agreement with previously published work (Doyle et al., 2015a). Furthermore, DR5::GFP crossed into the wei2wei7 background showed a similarly low GFP signal in the stele in control conditions, which was reduced even further by 15 µM ES8 treatment (Fig. S7 A). While 10 µM AA treatment did not noticeably affect the GFP signal in the stele of the WT, the signal in the wei2wei7 stele was rescued by this treatment (Fig. S7 A). These results suggest that AA may play a role in auxin distribution in the stele.
Next, we focused on the GFP signal in the root tip, particularly around the quiescent center (QC) and in the columella (Fig. S7 B). This time, we treated with ES8 at both 5 and 15 µM concentrations. In agreement with previously published work (Doyle et al., 2015a), ES8 treatment of DR5::GFP WT led to an accumulation of GFP signal in cell file initials surrounding the QC, which were not labeled in control conditions ( Fig. S7 B). A rather striking accumulation of DR5::GFP signal in the WT, extending into lateral columella and root cap cells, was particularly apparent at the higher ES8 treatment concentration of 15 µM. As found for the stele, DR5::GFP crossed into the wei2wei7 background showed a similar GFP signal pattern in the root tip in control conditions as that induced by ES8 in the WT, with an accumulation of signal in the file initials surrounding the QC (Fig. S7 B). This signal pattern was also apparent in the wei2wei7 background after 5 µM ES8 treatment and was enhanced after 15 µM ES8 treatment. While 0.5 µM and 10 µM AA treatment did not noticeably affect the GFP signal in the root tip of the WT, the signal pattern in the wei2wei7 root tip was slightly increased by 0.5 µM AA treatment and rescued to that of the control WT by 10 µM AA treatment ( Fig. S7 B). Therefore, in our experiments we observed a negative correlation between the DR5::GFP signal strength in the root stele and tip. Together, these results indicate that AA may play a role in auxin distribution in the root stele and tip, which likely affects gravitropic growth.
Our observations of DR5::GFP signal in the stele prompted us to investigate the basal to lateral plasma membrane fluorescence ratio (hereafter referred to as basal polarity) of PIN1, PIN3 and PIN7 in the provascular cells of Col-0 and wei2wei7 root tips. We treated seedlings short-term (2 hours) with 15 µM ES8 or 10 µM AA, performed immunolabeling to observe endogenous PIN1 and PIN7 and used the . Although PIN3-GFP basal polarity was not affected by ES8 or AA treatments in either the Col-0 or wei2wei7 backgrounds, it was increased by over 20% in the mutant compared to the WT (Fig. 4 E). Finally, although PIN7 basal polarity was not affected by ES8 or AA treatment in Col-0, it was strongly increased in the mutant compared to the WT, and like PIN1, was rescued in the mutant back to the level of the WT by ES8 treatment (Fig. 4 F). These results suggest that AA may play a role in maintenance of PIN polarity in root provascular cells. One possible speculation on why treatment with AA, in contrast to ES8, did not rescue PIN1 or PIN7 polarity in the mutant may be a rapid conversion of AA to downstream IAA precursors within the seedlings.
As AA is a precursor of auxin, which is known to affect transcription of PIN genes (Vieten et al., 2005;Paponov et al., 2008), we were interested in the expression levels of these genes in Col-0 and wei2wei7. We first investigated gene expression levels for all the plasma membrane-localized PIN proteins (PIN1, PIN2, PIN3, PIN4 and PIN7) in WT and mutant seedlings at nine days old, the age at which we performed our root gravitropic growth and length studies. The expression levels of PIN1, PIN2 and PIN4 were strongly decreased in wei2wei7 compared to Col-0, while PIN3 and PIN7 expression were somewhat decreased, but not significantly ( Fig. S8 A). We next investigated the expression levels of PIN1, PIN3 and PIN7 in the same conditions used for our PIN polarity studies in root provascular cells (five-day-old seedlings treated with ES8 and AA for 2 hours). At this stage, expression of PIN1, PIN3 and PIN7 were somewhat decreased in the mutant compared to the WT, but not significantly ( Fig. S8 B). Furthermore, treatment with ES8 and AA did not significantly affect the expression of these genes ( Fig. S8 B). These results imply that while transcription of PIN genes is decreased in wei2wei7, the effects of ES8 and AA on PIN polarity are not due to PIN gene transcriptional changes. Overall, our data suggest that endogenous AA may play a role in regulating the polarity of PIN1, PIN3 and PIN7 in root provascular cells via a mechanism unrelated to PIN gene expression levels.

AA regulates root gravitropism via repolarization of PIN3 and PIN7 in the columella
Our observations of DR5::GFP signal in the columella (Fig. S7 B) indicate that AA may also play a role in auxin distribution specifically in this particular root tissue.
Additionally, previous studies of the expression patterns of ASA1 and ASB1 promoter-GUS fusions in dark-grown Arabidopsis roots revealed strong expression in the root meristem and columella . Plasma membrane-localized PIN3 and PIN7 in the columella are thought to act in redistribution of auxin in response to gravistimulus (Friml et al., 2002b;Kleine-Vehn et al., 2010), potentially redundantly with PIN4, which is also localized in columella cells (Friml et al., 2002a;Vieten et al., 2005). We therefore reasoned that high expression of anthranilate synthase genes in the columella may reflect a role of AA in regulating gravity-responsive polarity of these PIN proteins. First, to confirm the expression patterns of the ASA1 and ASB1 promoters in light-grown roots, we performed GUS staining of ASA1::GUS and ASB1::GUS seedlings. We observed strong expression of the ASB1 promoter, but not the ASA1 previous studies using PIN3 and PIN4 antibodies (Friml et al., 2002b;Friml et al., 2002a). We therefore continued our studies of columella PIN proteins using PIN3::PIN3-GFP and PIN7::PIN7-GFP lines crossed into the wei2wei7 background appears to correlate with the partial rescue of root gravitropic growth by the same treatment ( Fig. 1 E). These results imply that endogenous AA may play a role in regulating relocalization of PIN proteins in the columella, especially PIN3, in response to gravity.
To further investigate a potential role for PIN proteins in AA-regulated root gravitropism, we analyzed root gravitropic growth in a range of pin mutants and their crosses with wei2wei7. Interestingly, while the eir1-4 (pin2) mutant showed intermediate root gravitropic growth between wei2wei7 and Col-0, crossing these mutants caused an additive effect, with wei2wei7eir1-4 being more severely agravitropic than wei2wei7 (Fig. S10 A). Of the tested pin3, pin4 and pin7 alleles, none of the single mutants were affected in root gravitropic growth compared to the WT and introduction of the pin3-4 or pin7-2 mutations into the wei2wei7 background did not affect the gravitropic growth. Interestingly, in contrast to the eir1-4 mutation, introduction of the pin3-5 or pin4-3 mutations into wei2wei7 partially rescued the root gravitropic growth compared to wei2wei7 (Fig. S10 A). While both tested pin1 alleles showed increased root gravitropic growth compared to the WT, the pin1-501 mutation also partially rescued the root gravitropic growth of wei2wei7 but the pin1-201 mutation had no effect (Fig. S10 A). The double and triple pin mutants tested showed little or no differences in gravitropic growth compared to the WT.
We next tested the effects of long-term treatments with high concentrations of ES8 and AA on root gravitropic growth in the mutants. Most of the tested pin mutants showed a similar sensitivity to ES8 as the WT in terms of reduction in root gravitropic growth, except for eir1-4, which was more sensitive to ES8 than the WT (Fig. S10 B).
Interestingly, although statistically significant differences were not often determined due to large variation in gravitropic index values, introduction of eir1-4 or either of the pin1 alleles to the wei2wei7 background tended to enhance the sensitivity of wei2wei7 to ES8 (Fig. S10 B). Large variation of gravitropic index values occurred mainly for the most agravitropic samples, due to random directions of root growth. In contrast to eir1-4 and pin1, addition of pin3, pin4 or pin7 alleles to wei2wei7 had no effect on its ES8 sensitivity ( Fig. S10 B). In the case of AA treatment, similarly to the WT, none of the pin mutants tested showed any sensitivity in terms of changes in root gravitropic growth ( Fig. S10 C). While the introduction of pin3-4 or pin7-2 to wei2wei7 did not alter its sensitivity to AA in terms of increase in gravitropic index, crossing eir1-4 or pin1-201 into wei2wei7 significantly increased its sensitivity to AA (Fig. S10 C). In contrast, introduction of pin3-5, pin4-3 or pin1-501 to wei2wei7 reduced its sensitivity to AA, resulting in decreased rescue of root gravitropic growth (Fig. S10 C).
In particular, the rescue of, as well as the reduction in AA-induced recovery of, wei2wei7 root gravitropic growth by introducing pin1, pin3 or pin4 mutations, provide further evidence for the involvement of PIN1 and PIN3, as well as implicating involvement of PIN4, in AA-regulated root gravitropism. The well-known important role of PIN2 in root gravitropism Kleine-Vehn et al., 2008), however, is most likely not related to AA-regulated root gravitropism, considering the strong additive effect of eir1-4 and wei2wei7 mutations in reducing root gravitropic growth and increasing sensitivity to both ES8 and AA.
Taken together, our results strongly support a new role for endogenous AA in root gravitropism via regulation of selective PIN protein polarity and thereby auxin distribution in both the stele and columella and that this role of AA is independent of its well-known function in IAA biosynthesis.

Discussion
Our work provides strong evidence in favor of a role for AA in root gravitropic growth through regulation of the subcellular localization of auxin transporter proteins, which likely influences the flow of auxin within the organ. The cellular distributions of these proteins are subject to regulation by dynamic and complex endomembrane trafficking.
Following their synthesis at the ER, most plasma membrane-targeted proteins are sorted and packaged into selective secretory trafficking routes (Gendre et al., 2014). It has been shown, for instance, that the auxin importer AUXIN-RESISTANT1 (AUX1) and exporter PIN1, when targeted to apical or basal plasma membranes of root tip cells respectively, are transported in distinct endosomes, subject to control by different regulatory proteins (Kleine-Vehn et al., 2006). The trafficking routes of such proteins may be distinct even if targeted to the same plasma membrane. For example, plasma membrane-targeted trafficking pathways for AUX1 and PIN3 in epidermal hypocotyl cells of the apical hook are distinct and subject to different regulatory proteins (Boutté et al., 2013). Such a remarkably complex system of endomembrane trafficking pathways is thought to allow for a high level of control, suggesting the likely existence of an array of selective endogenous compounds and/or signals regulating these trafficking routes.
Once polar plasma membrane-targeted auxin carriers have reached their destination, they remain remarkably dynamic, being subject to constant vesicular cycling (Geldner et al., 2001) to either maintain their localization or rapidly retarget them in response to external stimuli (reviewed by Luschnig and Vert, 2014;and Naramoto, 2017). Auxin itself has been shown to promote its own flow by inhibiting clathrin-mediated endocytosis of PIN transporters, thus enhancing their presence at the plasma membrane (Paciorek et al., 2005). Our results suggest that AA, an important early precursor in the IAA biosynthetic pathways, may also act in a feedback mechanism on PIN plasma membrane localization to regulate the flow of auxin, through currently unknown mechanisms.
The use of pharmacological inhibitors, identified through chemical biology approaches, has proven a powerful strategy that has greatly assisted in unravelling the details of auxin transporter trafficking routes and mechanisms (reviewed by Doyle et al., 2015b;and Hayashi and Overvoorde, 2014). In our previous study, we employed such a chemical biology strategy, revealing that the AA analog ES8 selectively inhibits an early ER-to-Golgi secretory pathway involved in basal targeting of PIN1 without affecting the polarity of apical plasma membrane proteins (Doyle et al., 2015a). We suggest that AA itself may act endogenously on trafficking regulation in a similar way to ES8 and it will be interesting to investigate this possibility in follow-up studies. As ES8 appears to mimic the effects of AA on PIN localization and root gravitropic growth, without releasing AA through degradation and without affecting IAA levels, this synthetic compound has proven extremely useful for dissecting this newly discovered role of AA from its better-known role in auxin biosynthesis.
Interestingly, the amino acid para-aminobenzoic acid, which has a similar structure to AA and is produced from the same precursor, has also recently been shown to play a role in root gravitropism, distinct from its better known role in folate biosynthesis (Nziengui et al., 2018). However, unlike AA, para-aminobenzoic acid promotes gravitropic root growth in wild type plants as well as promoting gravistimulated root bending by enhancing the asymmetric auxin response between the two root sides (Nziengui et al., 2018). AA is an important endogenous compound in plants, functioning as an early precursor in the chloroplast-localized biosynthetic pathway producing the amino acid tryptophan (Maeda and Dudareva, 2012).
Tryptophan is itself an essential amino acid, acting as a precursor for several indolecontaining plant compounds, including IAA, the predominant auxin in plant tissues (Mano and Nemoto, 2012 Korasick et al., 2013). A family of GRETCHEN HAGEN 3 (GH3) IAA-amido synthetases conjugate IAA to several amino acids (Staswick et al., 2005), while two DEOXYGENASE FOR AUXIN OXIDATION (DAO) enzymes have recently been identified in Arabidopsis (Porco et al., 2016), with DAO1 specifically demonstrated to catalyze oxIAA (Zhang et al., 2016). These enzymes are thought to be extremely important in regulating auxin homeostasis by converting IAA to inactive and storage forms (reviewed by Zhang and Peer, 2017).
The agravitropic growth of wei2wei7 roots may be due to a combination of decreased auxin content caused by reduced AA levels and the AA deficiency itself, as both auxin and AA affect the localization and polarity of PIN proteins. As was shown previously for ES8 (Doyle et al., 2015a), AA appears to act selectively depending on the PIN protein and the root tissue. PIN1, PIN3 and PIN7 all display increased basal polarity in provascular cells of wei2wei7 compared to Col-0, suggesting increased flow of auxin towards the root tip in the mutant. Correspondingly, we found decreased expression of the auxin-responsive promoter DR5 in the root stele and increased expression around the root tip QC in the mutant, a pattern that was also observed in WT roots upon ES8 treatment. PIN7, but not PIN3, is also abnormally polarized in columella cells of wei2wei7, while both these proteins appear to be completely unresponsive to gravitropic stimulus in the mutant columella. Furthermore, the high expression of ASA1 and ASB1 in the root columella of the WT suggests the importance of AA in this tissue in particular, which our results suggest is due to a role for this compound in gravity-regulated PIN distribution on the plasma membranes. The particular importance of PIN1 and PIN3 in AA-regulated root gravitropism was further supported by the rescue, as well as the reduction in AA-induced rescue, of wei2wei7 root gravitropic growth by introduction of pin1 or pin3 mutations.
Taken together, our results strongly suggest that the endogenous compound AA plays a role in root gravitropism by regulating the polarity and gravity-induced relocalization of specific PIN proteins in the provascular and columella cells.
Furthermore, this role of AA is distinct from its well-known function in auxin biosynthesis, which we suggest is more important for root elongation than gravitropic growth.

Plant material and growth conditions
For surface sterilization of Arabidopsis thaliana seeds, one tablet of Bayrochlor (Bayrol) was dissolved in 40 ml distilled water before diluting 1:10 in 95% ethanol.
Seeds were then incubated in the Bayrochlor solution (active ingredient sodium dichloroisocyanurate) for 6 min followed by two rinses with 95% ethanol. Seeds were then allowed to dry in a sterile environment before sowing on plates of growth medium containing ½ strength Murashige and Skoog (MS) medium at pH 5.6 with 1% sucrose, 0.05% MES and 0.7% plant agar (Duchefa Biochemie). Two d after stratification at 4°C to synchronize germination, the plates were positioned vertically and the seedlings were grown for 5 or 9 d at 22 °C on a 16 h : 8 h light : dark photoperiod. The Columbia-0 (Col-0) accession was used as the WT except for growth of pin3-5pin4-3pin7-1 seedlings, for which the Landsberg erecta (Ler) accession was also used, being the background for the pin7-1 allele. See Table S1 for the previously published Arabidopsis lines used. All mutants and marker lines in the wei2wei7 background were generated in this study by crossing. For selection of homozygous lines at the F2 generation, seedlings displaying a wei2wei7 phenotype were initially selected and then genotyped for the third mutation for triple mutants (see Table S2 for primers used), or observed on a fluorescence stereomicroscope for GFP marker line crosses. Finally, selected lines were confirmed for homozygous wei2-1 and wei7-1 mutations by genotyping (see Table S2 for primers used). The heterozygous lines pin1-201 and pin1-501 (we added 01 to the name of this mutant to distinguish from another pin1-5 allele) and their crosses with wei2wei7 were transferred to soil after imaging on treatment plates and only those that later formed pin-like inflorescences were included in root measurements in the initial images. All root length and gravitropic index measurements were performed on 9-d-old seedlings, while microscopy studies were performed on 5-d-old seedlings due to cell collapse in a proportion of the tips of older wei2wei7 roots. ImageJ software (https://imagej.nih.gov/ij/) was used to measure root length and vertical gravitropic index, which was calculated for each root as a ratio of L y : L, where L y is the vertical distance from root base to tip, or the real depth of root tip penetration, and L is the root length, as described in Grabov et al. (2005). For gravistimulated root bending experiments, seedlings were grown vertically for 5 d in treatment-free conditions, then transferred to mock or estradiol-supplemented medium for 24 h, and then gravistimulated by turning the plates 90° for 24 h before measuring the root bending angles. The angles were measured in the direction of root bending between two lines intersecting at the former root tip position before gravistimulus, one being horizontal and the other originating from the current root tip position (Fig. S6 D).
Chemicals were diluted in liquid medium for short-term treatments (2 h) or growth medium for long-term treatments (5 or 9 d), in which case seeds were directly sown and germinated on chemical-supplemented medium. Equal volumes of solvent were used as mock treatments for controls. For live imaging, seedlings were mounted in their treatment medium for microscopic observations.

IAA metabolite analysis
For quantification of endogenous IAA and its metabolites, 20-30 whole seedlings per sample were flash-frozen in liquid nitrogen and ground with plastic microtube pestles.
Approximately 20 mg of ground tissue was collected per sample and stored at -80 o C.
Extraction and analysis were performed according to Novák et al. (2012). Briefly, frozen samples were homogenized using a MixerMill bead mill (Retsch GmbH) and extracted in 1 ml of 50 mM sodium phosphate buffer containing 1% sodium diethyldithiocarbamate and a mixture of 13 C 6 -or deuterium-labeled internal standards.
After pH adjustment to 2.7 by 1 M HCl, a solid-phase extraction was performed using Oasis HLB columns (30 mg 1 cc; Waters). Mass spectrometry analysis and quantification were performed by a liquid chromatography-tandem mass spectrometry

Chemical synthesis
The following procedure was used to synthesize (4-(N-(4-chlorobenzyl) methylsulfonamido)benzoyl)tryptophan (ES8.7-Trp). Methyl sulfonyl chloride (3.1 ml, 45.5 mmol, 1.2 equivalent) and pyridine (3 ml, 37.1 mmol, 1.1 equivalent) were added to a solution of 4-aminoethylbenzoate (5.1 g, 33.7 mmol) in acetonitrile (100 ml) at 0°C and then stirred overnight at room temperature. The reaction mixture was concentrated and the resulting residue was dissolved in ethyl acetate. The organic layer was washed with HCl (2 N, 100 ml), saturated NaHCO 3 , H 2 O and brine, dried over Na 2 SO 4 , and concentrated to give an off-white solid (7.1 g, 94%), ethyl 4-(methylsulfonamido) benzoic acid, which was used without further purification. Then 4-chlorobenzyl bromide (1.43 ml, 11 mmol, 1.2 equivalent) and K 2 CO 3 (3.8 g, 27.4 mmol, 3 equivalent) were added to a solution of ethyl 4-(methylsulfonamido)benzoate (2.1 g, 9.15 mmol) in dimethylformamide (DMF) (14 ml). The reaction mixture was stirred overnight after which liquid chromatography-mass spectrometry indicated the reaction was complete, yielding 4-(N-(4-chlorobenzyl)methylsulfonamido)ethylbenzoate (2.8 g, 88%). Next, NaOH (2 N, 15 ml) and H 2 O (10 ml) were added to 2 g of 4-(N-(4chlorobenzyl)methylsulfonamido)ethylbenzoate, followed by overnight stirring. The reaction mixture was acidified to pH 5 with concentrated HCl and partitioned between ethyl acetate and H 2 O. The organic layer was washed with brine, dried over Na 2 SO 4 , and concentrated to give 4-(N-(4-chlorobenzyl)methylsulfonamido)ethylbenzoate (1.7 g, 92%), which was used without further purification. Then 10 ml thionyl chloride was added to a solution of 1 mmol 4-(N-(4-chlorobenzyl)methylsulfonamido)ethylbenzoate and refluxed for 12 h under a nitrogen atmosphere. The thionyl chloride was removed under reduced pressure to give acid chloride. Then 1.2 mmol tryptophan was dissolved in 5 ml DMF and cooled to 0°C. At this temperature the acid chloride was added slowly by dissolving in 3 ml DMF. The reaction mixture was stirred at room temperature for 24 h. Ice cold water was added to the reaction mixture, which was then extracted with chloroform and washed with brine solution three times to afford the crude mixture.

Compound degradation analysis
For short-term treatments with 5 µM ES8 compounds, 5-day-old Col-0 and wei2wei7 seedlings were incubated for 5 h in ES8, ES8.7 and ES8.7-Trp-supplemented liquid medium before harvesting in liquid nitrogen along with samples of liquid treatment medium incubated for 5 h without any seedlings. For long-term treatments with 5 µM ES8 compounds, Col-0 and wei2wei7 seedlings were grown for 9 days on ES8, ES8.7 and ES8.7-Trp-supplemented solid medium before harvesting in liquid nitrogen along with samples of solid treatment medium incubated for 9 days without any seedlings.
Samples from two biological replicates were harvested and divided into two technical replicates each. The medium samples were diluted 1:100 with 30% v/v acetonitrile and  (http://wmd3.weigelworld.org). The amiRNA was obtained using the pRS300 vector as a PCR template, as described previously (Ossowski et al., 2008). The fragments were introduced into the Gateway pENTR/D-TOPO cloning vector (Invitrogen) and verified by sequencing. The ASA1 coding sequence was then cloned into the DLphosphinothricin-resistant vector pFAST-R02 (Shimada et al., 2010), while the amiRNA was cloned into the hygromycin-resistant vector pMDC7b containing the estradiol-inducible XVE system (Curtis and Grossniklaus, 2003), using the LR reaction (Invitrogen). Agrobacterium-mediated transformation of the constructs into Arabidopsis thaliana Col-0 was achieved by floral dipping (Clough and Bent, 1998). Transformed plants were selected via antibiotic resistance on agar plates supplied with the respective antibiotics, 50 µg ml -1 hygromycin B or 25 µg ml -1 DL-phosphinothricin. Four and six independent homozygous 35S::ASA1 and XVE::amiRNA-PAT1 lines were analyzed, respectively. Two lines per construct were then selected that displayed strong and reproducible constitutive ASA1 induction or induced PAT1 silencing, as determined by performing qPCR (see Quantitative PCR) using RNA extracted from one-week-old whole seedlings (Fig. S6 A and B). To induce silencing in the XVE::amiRNA-PAT1 seedlings, the seedlings were germinated and grown on agar plates supplemented with 20 µM estradiol, with DMSO used as a mock treatment control. Each selected 35S::ASA1 line was crossed with each selected XVE::amiRNA-PAT1 line and homozygous F2 generation offspring, named AxP (ASA1 x PAT1) lines, were selected as before via antibiotic resistance. The homozygous lines were then tested for gene expression as before via qPCR (12 independent lines were analyzed) (Fig. S6 C and D).
Finally, two lines displaying reproducible simultaneous constitutive ASA1 induction and induced PAT1 silencing were selected for use in further experiments.

Quantitative PCR
Total RNA was extracted from whole seedlings using the RNeasy Plant Mini Kit AT4G34270 and AT1G13320 were analyzed alongside the target genes (see Table S2 for primers used). Expression levels of the target genes were normalized against the two most stably expressed reference genes, as determined using GeNorm (Biogazelle) (Vandesompele et al., 2002), using the formula below, where E = efficiency, R = reference gene, Cq = quantification cycle mean, T = target gene. For each target gene, the normalized expression values were scaled relative to that of the WT control.

Generation of PIN7 antibody
For the generation of anti-PIN7, a region of 882 bp corresponding to the hydrophilic loop of PIN7 was amplified and attB1 and attB2 recombination sites were incorporated (see Table S2 for primer sequences). The amplicon was recombined into the

Immunolocalization and confocal microscopy
Immunolocalization in Arabidopsis roots was performed as described previously, using an Intavis InsituPro Vsi (Doyle et al., 2015a). Primary antibodies used were anti-PIN1 at 1:500 (NASC), anti-PIN3 at 1:150 (NASC), anti-PIN4 at 1:400 (NASC) and anti-PIN7 at 1:600. Secondary antibodies used were Cy3-conjugated anti-rabbit and antisheep at 1:400 and 1:250, respectively (Jackson ImmunoResearch). Confocal laser scanning microscopy on seedling root tips was performed using a Zeiss LSM 780 confocal microscope with 40X water-immersion objective lens and images were acquired with Zeiss ZEN software using identical acquisition parameters between mock and chemical treatments and between the WT and mutant in each experiment. For PIN basal polarity index quantification in provascular cells, the 'mean gray area' tool in ImageJ was used to measure plasma membrane fluorescence intensity in confocal images and a basal (lower) to lateral fluorescence ratio was calculated for each cell measured. To monitor gravitropically induced PIN relocalization in root columella cells, confocal Z-scans were acquired before and 30 min after gravistimulation, during which the seedlings were rotated 90° to the horizontal position. Fluorescence intensity at the apical (upper), basal (lower) and lateral plasma membranes of the cells was measured on maximal intensity projections of the Z-scans using ImageJ. For PIN apicl-basal polarity index measurements, the apical-plus-basal to lateral-plus-lateral fluorescence ratio was calculated. For PIN relocalization measurements, the signal intensity ratio between the outermost lateral plasma membranes on the periphery of the columella (left and right sides) was measured before and after gravistimulation and these ratios were compared.

GUS staining
Seedlings were fixed in 80% acetone at -20°C for 20 min, washed three times with distilled water and then incubated in 2 mM X-GlcA in GUS buffer (0.1% triton X100, 10 mM EDTA, 0.5 mM potassium ferrocyanide and 0.5 mM potassium ferricyanide in 0.1 M phosphate buffer (Na2HPO4 / NaH2PO4) at pH 7). The samples were then infiltrated for 10 min in a vacuum desiccator before incubation in the dark at 37°C. The GUS reaction was stopped by replacing the GUS buffer with 70% ethanol. Samples were then mounted in 50% glycerol and observed on a Zeiss Axioplan microscope.

Statistical analyses
For all experiments, at least 3 biological replicates were performed and always on different days. When more than 3 or 4 biological replicates were performed, this was due to poor growth of wei2wei7 that occasionally resulted in a low number of seedlings or quantifiable roots in some of the replicates. Wilcoxon rank sum (Mann-Whitney U) tests or Student's t-tests were performed on full, raw datasets of nonparametric or parametric data, respectively, to determine statistically significant differences. On all charts, the means and standard errors of the biological replicates are displayed.

Supplemental material
Ten supplementary figures and two supplementary tables accompany this article.              Supplemental Tables   Table S1. Arabidopsis mutants and transformed lines used in this study.