Endosidin2 targets conserved exocyst complex subunit EXO70 to inhibit exocytosis

ES2 is a previously identified plant endomembrane trafficking disruptor (Fig. 1A) that inhibits polarized growth of pollen tubes in a dose-dependent manner (Fig. S1 A and B) (16). Arabidopsis seedlings grown on media containing ES2 have shorter roots and fewer and shorter root hairs and are less sensitive to gravity stimulation (Fig. S1 C–G). ES2 disrupted the trafficking of proteins that are actively recycled between the plasma membrane and endosomes, such as the brassinosteroid receptor (BRI1) and the auxin transporters PINFORMED1 (PIN1) and PIN2 after short time treatment (2 h) (Fig. S2A) (16, 17). Although ES2 was originally identified from the same phenotype cluster as bioactive compounds ES1 and ES3, it did not target the same proteins as ES1 and ES3 because it did not induce aggregation of trans-Golgi network marker SYP61 compared with ES1 and did not affect ROP6 localization compared with ES3, respectively (Fig. S2A) (16, 17). ES2 also did not affect the localization of cellular markers such as HDEL:GFP [endoplasmic reticulum (ER)], GOT1p:YFP (Golgi), SYP22:YFP [tonoplast and prevacuolar compartment (PVC)], PGP4:GFP (plasma membrane), or PIP2a:GFP (plasma membrane) (Fig. S2B).

We further explored ES2 effects at the cellular level using GFP-tagged PIN2 protein because it is known to traffic to the plasma membrane, endosomes, and vacuoles (18–21). Short term ES2 treatment reduced the amount of the plasma membrane-localized PIN2 compared with control seedlings, as shown by a fluorescence intensity plot profile in time-lapse images of PIN2:GFP seedlings treated with DMSO or ES2 (Fig. 1 B–E). The mean fluorescence intensity of total plasma membrane-localized PIN2:GFP in seedlings treated with ES2 (90.6 ± 13.7, mean ± SD, n = 30) was significantly lower than that in seedlings treated with DMSO (118.3 ± 17.9, mean ± SD, n = 30) (P < 0.05). When we performed ES2 treatment of PIN2::PIN2:GFP-expressing seedlings in the dark to inhibit vacuolar-localized GFP fusion protein degradation (22), we found an increased amount of GFP fluorescence in the vacuoles compared with the control (Fig. 1F). The results indicated that ES2 treatment inhibited trafficking to the plasma membrane, and, as a consequence, the trafficking to the vacuole for degradation is enhanced.

The feret diameter of PIN2-localized compartments observed from fluorescence confocal microscope images upon ES2 treatment under light conditions was 1.18 ± 0.47 μm (mean ± SD, n = 391, from 107 cells of 11 seedlings), with a maximum feret diameter of 2.9 μm and a minimal feret diameter of 0.4 μm (Fig. 1G). Size distribution of ES2-induced PIN2 agglomerations indicates that they are very different from known brefeldin A (BFA)-induced agglomerations resulting from abnormal trafficking at the Golgi. PIN2 vacuolar trafficking involves the retromer complex, including its component Sorting Nexin1 that colocalizes with ARA7/RabF2b endosomes (19, 20). We found that, in seedlings expressing both PIN2::PIN2:GFP and ARA7/RabF2b:mRFP, the ES2 treatment induced PIN2 accumulation in the ARA7/RabF2b endosomal compartments (Fig. 1H). We manually examined 338 PIN2 agglomerations from roots of 12 seedlings treated with ES2 and found that all of these agglomerations had partial or complete colocalization with ARA7/RabF2b-labeled late endosomes/PVC. Accumulation of PIN2 in ARA7/RabF2b-positive compartments was consistent with our observation that ES2 reduced PIN2 plasma membrane localization and that PIN2 trafficking to the vacuole is increased.

We next examined whether a reduced quantity of PIN2 at the plasma membrane was a result of reduced PIN2 recycling. The PIN2::PIN2:GFP seedlings grown on normal media were treated with BFA for 2 h and then permitted to recover in liquid media containing either ES2 or DMSO for another 1.5 h before imaging. We found that the disappearance of BFA bodies in ES2-treated seedlings showed a significant delay in comparison with the DMSO containing media (Fig. 1 I and J), indicating that ES2 treatment reduced PIN2 recycling through endosomes. Similar delayed PIN2 recycling has been observed in mutants that are defective in exocytosis (23).

The ES2 molecule contains an N-acyl hydrazone group at its core and could have the propensity for hydrolysis to 3-fluorobenzohydrazide and 4-hydroxy-3-iodo-5-methoxybenzaldehyde in aqueous solution (Fig. S3A). To confirm whether ES2 is stable in our system, we tested the stability of ES2 in a water solution using ¹H NMR analysis. We collected ¹H NMR spectra of ES2 at different time points over a course of 1 wk and observed no hydrolysis products under these conditions (Fig. S3B). In addition, we tested the activity of the two possible ES2 hydrolysis products in inducing PIN2 localization in late endosomal compartments. We found PIN2 localization at the plasma membrane after we treated PIN2::PIN2:GFP seedlings with 40 μM 3-fluorobenzohydrazide or 4-hydroxy-3-iodo-5-methoxybenzaldehyde for 2 h, similar to what we observed in the DMSO control, but significantly different from the intracellular localization pattern in ES2 treated samples (Fig. S3 C–F). These data showed that ES2 is a stable compound under aqueous solution and that the induced trafficking phenotypes are not due to any in situ hydrolysis byproducts.

Overall, we concluded that ES2 reduced trafficking to the plasma membrane and that protein trafficking to the vacuole is increased as a consequence. This observation suggests a possible mechanism to regulate the dynamics of vesicle trafficking.

Structure-activity relationship (SAR) analysis was performed to identify moieties in ES2 that were dispensable for its activity based on the induction of PIN2 localization in agglomerations (Fig. 2A and Fig. S4). We found that the iodine in the molecule was necessary for its activity whereas the benzoic ring with the fluorine could accommodate different atoms while maintaining activity. To generate analogs with biotin to facilitate target identification, we synthesized new active and inactive analogs with an amine group in the benzoic ring with the fluorine named analog-688 (Ana-688) and analog-680 (Ana-680), as active and inactive analogs, respectively (Fig. 2 B and D). These two analogs were further modified to produce biotinylated molecules using the amine group and named Bio-688 and Bio-680, respectively (Fig. 2 C and E) (see SI Materials and Methods for schemes and Dataset S1 for characterization of synthesized compounds). Ana-688 and Bio-688 induced PIN2 agglomerations after short-term treatment whereas Ana-680 and Bio-680 did not, indicating they could be used as active analogs and inactive analogs, respectively.

Bio-688 and Bio-680 were coupled to streptavidin agarose, resulting in active and inactive matrices, respectively, which were incubated with Arabidopsis cell extracts. Proteins bound to the active and inactive matrices were eluted by ES2, and the eluted fractions were analyzed using mass spectrometry (MS). Although the peptide abundance in the elution fractions was low (Dataset S2), we detected a peptide from Arabidopsis EXO70G2, which belongs to the EXO70 family in Arabidopsis that is involved in exocytosis, from the active matrix but not the inactive matrix elution. EXO70G2 belongs to the EXO70 family that has 23 members in Arabidopsis divided into subclasses A to H (7, 24, 25). EXO70A1, which shares 24% amino acid sequence identity with that of EXO70G2, is a member of the EXO70 family that has been well-studied, and there are resources available for us to do further investigation. We then took other approaches to test for possible interaction between ES2 and EXO70 proteins in Arabidopsis. Using an available EXO70 antibody, we further tested the presence of a close paralog EXO70A1 on the matrix by Western blot (Fig. 2F) (7–9, 23, 26). The intensity of the Western blot bands indicated that the Bio-688 matrix was more potent in pulling down EXO70A1, in comparison with the Bio-680 matrix and biotin controls, indicating that EXO70A1 interacted more strongly with the active ES2 analog compared with the inactive analog.

We took a relatively new approach for chemical target identification called drug affinity responsive target stability (DARTS) to test the interaction between ES2 and EXO70A1 (27). The DARTS approach was developed based on the observation that some proteins are protected from degradation by proteases when bound to the ligand (27). We incubated an Arabidopsis protein extract with ES2 or DMSO and then digested with different concentrations of proteases. After normalizing EXO70A1 protein Western blot band intensity against that of the actin internal control, we found that the degradation of EXO70A1 was significantly protected by ES2 compared with actin, which was detected on the same blotting membrane at protease dilutions of 1:3,000 and 1:10,000 (Fig. 2G). We further expressed and purified the EXO70A1 protein from Escherichia coli and tested for its interaction with ES2, using saturation-transfer difference NMR (STD-NMR) (28). EXO70A1 amino acid residues 75–638 were used for STD-NMR due to the instability of the full-length protein (Fig. 2H). The ¹H assignments of ES2 in DMSO-d₆ were determined from gCOSY (gradient correlation spectroscopy) and gNOESY (gradient nuclear overhauser effect spectroscopy) spectral analysis (Fig. S5 A and B and Table 1). Assignments in D₂O were made by comparison with the 1D spectrum recorded in DMSO-d₆. To determine the optimum saturation time, an STD build-up curve was generated using 400 μM ES2 and 20 μM EXO70A1 samples (Fig. S6C). The build-up curve (Fig. 2I) clearly indicated that there was direct interaction between ES2 and EXO70A1. The ¹H spectral peak from an unrelated molecule showed no interaction with EXO70A1 (Fig. 2I, arrow). Based on the results from the build-up curve, all further STD-NMR measurements were taken using a 2-s saturation time. We titrated EXO70A1 with different concentrations of ES2 in an STD-NMR experiment, and the STD amplification factor was calculated as a function of ES2 concentration (Fig. 2J). The dissociation constant (K_d) of the EXO70A1 and ES2 interaction using STD-NMR was 400 ± 170 μM and the B_max was 12.9 ± 2.74 (Table 2). We also performed the ¹H assignments of ES2 analog8 (Table S1 and Fig. S5D), an inactive analog, and studied the interaction between analog8 and EXO70A1 using STD-NMR. The calculated STD amplification factors did not show a significant increase, with elevated concentrations of analog8 (Fig. 2J). This result confirmed that the detected interaction between ES2 and EXO70A1 using STD-NMR is not due to random interaction of any small molecule with EXO70A1 protein.

To further confirm results from STD-NMR, we used the technique of microscale thermophoresis (MST) to quantify the dissociation constant for the complex of ES2 (titrant) with EXO70A1 (target molecule). This method observes the motion of molecules in response to a temperature gradient (29–31). Thermophoresis is characterized by monitoring the time-dependent fluorescence, referred to as a time trace, of a labeled target molecule in a small zone subject to localized heating by an infrared laser (30). Multiple time traces were acquired for serial dilutions of a binding partner. Because binding of the titrant with the target molecule results in a change of mass, charge, or hydration entropy, the complex exhibits different thermophoretic behavior from the target molecule alone (29, 31). Plotting the thermophoretic effect as a function of titrant concentration presents dose-dependent behavior. From the dose-responsive curve, we calculated a K_d of 253 ± 63.5 μM for the interaction of ES2 with EXO70A1. Although the mean value from MST is lower than the K_d from STD-NMR, the results are not significantly different with a 95% level of confidence. This result suggests a micromolar affinity for the binding of ES2 to EXO70A1 consistent between binding assays based on different physical principles (Table 2). Moreover, when MST is performed for the interaction of a negative control, analog8, with EXO70A1, we observed no change in thermophoretic behavior as expected (Fig. 2K).

We concluded from these different assays that ES2 interacted with EXO70A1 in vitro, suggesting that it could be a target in vivo.

To investigate the relationship between ES2 and the EXO70 gene family at the genetic level, we tested root growth phenotypes of available exo70 mutants as listed in Table S2, in the presence of ES2. None of the 24 mutants that we tested displayed significant differences in response to ES2 compared with WT, except one. Heterozygous seedlings of T-DNA (transfer DNA) insertion allele exo70A1-3 (SALK_026036C) showed resistance to ES2 in root growth (Fig. 3A). To study the response of T-DNA insertion mutant plants to ES2 at the cellular level, we crossed PIN2::PIN2:GFP with exo70A1-3 heterozygous plants. The F3 population from an F2 seedling that was homozygous for PIN2::PIN2:GFP and heterozygous for T-DNA insertion was used to study the difference between WT plants and exo70A1-3 heterozygous plants. Upon ES2 treatment, heterozygous T-DNA insertion mutant seedlings showed smaller and fewer PIN2 agglomerations compared with WT seedlings from the same segregating population (Fig. 3 B–E). Although heterozygous exo70A1-3 seedlings had normal growth at the seedling stage, the plants had retarded growth after bolting, more shoot stems, reduced seed yield, and abnormal flower development (Fig. S6A). Homozygous exo70A1-3 seedlings had severe growth phenotypes that included arrested root growth and abnormal root meristem organization and were seedling lethal (Fig. S6B), similar to previously characterized exo70A1-1 and exo70A1-2 mutant alleles, but were more severe (7). The more severe phenotypes in exo70A1-3 homozygous plants were reported previously, and it was not understood why this T-DNA insertion line caused the severe phenotypes (32). We decided to further characterize this allele to find out the linkage between T-DNA insertion in EXO70A1 and ES2 resistance. We analyzed the homozygous mutant at the transcription level and found that it accumulated the 5′ end of the EXO70A1 mRNA upstream of the T-DNA insertion site whereas the 3′ end of mRNA downstream of the T-DNA insertion was not detected (Fig. 3 F and G). We suspected that this mutation resulted in dominant resistance due to the truncated mRNA encoding a stably accumulated N-terminal peptide that caused the severe phenotypic defect in exo70A1-3. In contrast, exo70A1-1 and -2 were reported as null and knockdown alleles that exhibit a milder phenotype (7). This observation led us to characterize the mutant allele at the protein level.

The expected mass of the truncated peptide was ∼25 kDa, with 231 amino acids. Because the anti-EXO70A1 antibodies did not recognize the N-terminal region of EXO70A1, we took advantage of mass spectrometry analysis. We first surveyed the E. coli-expressed EXO70A1 protein with nano-LC/MS/MS (liquid chromatography/mass spectrometry/mass spectrometry) and identified a peptide ion from its N-terminal region (amino acids 90–109) with strong signal intensity (Fig. S7). Using this peptide ion with known m/z, charge state, and retention time as the fingerprint for EXO70A1, we then analyzed plant proteins derived from WT and the T-DNA line. We analyzed SDS gel bands with approximate masses of 70 kDa and 25 kDa, respectively, from both the WT and exo70A1-3 mutant using nano-LC/MS (the system was confirmed with no background after blank injection). It was found that the amino acid 90–109 fingerprint was detected in WT 70-kDa and exo70A1-3 25-kDa samples (Fig. 3H), indicating that the exo70A1-3 allele accumulates a truncated peptide in planta that might be responsible for the ES2 resistance. To test the contribution of the truncated peptide to the growth resistance of the exo70A1-3 mutant, we expressed EXO70A1 amino acids 1–231 in WT plants under the Cauliflower Mosaic Virus 35S promoter and then analyzed the effect of ES2 on the transgenic lines. Homozygous transgenic lines expressing the N terminus of EXO70A1 showed partial resistance to ES2 in root growth when grown on media containing lower concentrations of ES2 (Fig. 3I). We also found resistance to ES2 when exo70A1-3 heterozygous seedlings or EXO70A1 N terminus expression seedlings were transferred from normal media to ES2-containing media (Fig. 3J). Reproduction of ES2 resistance in transgenic lines confirmed the linkage between EXO70A1 N terminus expression and ES2 resistance. Similar to exo70A1-3 heterozygous plants, plants expressing the N-terminal truncated peptide displayed normal growth at the seedling stage in the absence of ES2, but retarded growth, small stature, reduced seed yield, and flower development defects after bolting on soil (Fig. S8). Thus, we concluded that the EXO70A1 N-terminal peptide was sufficient to induce dominant ES2 resistance in plants, and this observation confirmed in planta that EXO70A1 was a target of ES2. This observation also suggested that the exocyst could be an important site for controlling the dynamics between recycling to the plasma membrane and vacuole targeting and that the N-terminal domain probably served distinct roles in exocyst regulation. Heterozygous exo70A1-3 plants show developmental phenotypes in later stages of plant development, which is similar to the EXO70A1 N-terminal peptide expression line. This observation may also reflect the regulatory roles of the EXO70A1 N-terminal domain and the contribution of EXO70A1 to different stages of plant development. Although further investigation is needed to understand the underlying cellular mechanisms for the roles of the EXO70A1 N terminus region in regulating exocytosis and in plant development, the genetic resistance in the mutant and transgenic lines confirms the linkage between ES2 and exocyst regulation at the genetic level.

To discover whether ES2 inhibited EXO70 cellular dynamics directly, we examined the cellular localization of GFP-tagged EXO70A1 (GFP:EXO70A1) in Arabidopsis root cells upon ES2 treatment. GFP:EXO70A1 showed plasma membrane localization with distinct polarized maximum at the outer lateral side of root epidermal cells in the root tip (Fig. 4A, white arrows). We found that, upon ES2 treatment, the polarized localization pattern of EXO70A1 was lost (Fig. 4 A–C). Plot of fluorescence intensity across the root transition zone in DMSO-treated control seedlings revealed high fluorescence intensity in the outside layer of the root epidermal cells (Fig. 4B). After 2 h of 40 μM ES2 treatment, the lateral polarity of EXO70A1 was changed, as reflected by reduced fluorescence intensity at the outside layer of root epidermal cells (Fig. 4C). The mean fluorescence intensity of lateral GFP:EXO70A1 in ES2-treated seedlings (32.0 ± 12.2, mean ± SD, n = 10) was significantly lower than DMSO control seedlings (88.1 ± 18.6, mean ± SD, n = 10) (P < 0.05). However, the lateral polarity pattern of two other exocyst components, EXO84 and SEC8, was not affected (Fig. S9A, white arrows). In root hair cells, where active exocytosis is required for polarized growth (33, 34), it was found that, upon ES2 treatment, the fluorescence recovery of GFP:EXO70A1 after photobleaching was significantly slower compared with control cells (Fig. 4D). The maximum fluorescence recovery was more than 80% of the prebleached intensity within 3 min in control cells whereas the fluorescence recovery was limited to 20% of the prebleached level at the same time in cells treated with ES2 (Fig. 4E). This result indicated that ES2 significantly inhibited the cellular dynamics of EXO70A1 in root hair cells as well. The fact that ES2 strongly interfered with EXO70A1 localization without perturbing two other exocyst subunits further supported that, in planta, EXO70A1 was a target of ES2. The specificity of mislocalization suggests that the mechanism of EXO70A1 lateral plasma membrane targeting is distinct from that of other exocyst components. Furthermore, EXO70A1 may dissociate from the complex independently from other subunits, indicating that the maintenance of its lateral polarity may be distinct (assuming that EXO84, SEC8, and EXO70A1 are part of the exocyst while at the lateral plasma membrane). Furthermore only lateral polarity was affected (not isolateral localization), indicating that EXO70A1 lateral polarity is distinct.

Due to the evolutionary conservation of the composition and function of the exocyst complex, we were interested in investigating whether ES2 can target EXO70 in other systems. We examined whether ES2 would affect exocytosis in human cells using the transferrin (Tfn) recycling assay, which measures the recycling of endocytosed transferrin to the plasma membrane. The assay has been commonly used to study protein trafficking to the plasma membrane. After treatment with DMSO as a control or ES2 for 1 h, the cells were pulsed with Tfn-AlexaFluor488 on ice for 5 min and chased with complete media to track Tfn trafficking over time. Most of the Tfn was exocytosed after a 90-min chase in cells treated with DMSO (Fig. 5A). However, Tfn accumulated markedly at the protrusion sites of cells treated with ES2, indicating that exocytosis was reduced. Comparison of the ratio of Tfn retention at 30 min and 90 min after chasing with that of 5 min after chasing in DMSO- and ES2-treated cells indicated reduced Tfn trafficking in ES2-treated cells (Fig. 5B). To test whether the reduced exocytosis in mammalian cells was related to EXO70 as in plants, we tested the localization of GFP-tagged rat EXO70 (rEXO70) protein after ES2 treatment. We found that ES2 treatment induced the accumulation of EXO70 vesicles near the plasma membrane (Fig. 5C). A similar effect was observed for human EXO70 isoform2 and isoform5 proteins upon ES2 treatment (Fig. S9B). These vesicles may suggest a block of vesicle tethering and fusion with plasma membrane. To test direct interaction between ES2 and rEXO70, we purified E. coli-expressed rEXO70 (Fig. 5D) and performed an STD-NMR experiment. The interaction between ES2 and rEXO70 was confirmed by the presence of saturation energy transfer (Fig. 5E). These data confirm that ES2 directly targets EXO70 in vivo to inhibit exocytosis in mammalian cells as well as in plants. It further indicates that, in human cells, ES2 can target multiple isoforms of EXO70, resulting in misregulation of exocytosis.

The fact that plant and mammalian EXO70 proteins are targets of ES2 suggests structural similarity. To better understand the structural basis for the conservation of the altered regulation of EXO70A1 by ES2, we crystallized Arabidopsis EXO70A1 and determined its structure at 3.1-Å resolution. The crystal structure of EXO70A1 revealed that it adopted an elongated architecture resembling that previously observed for yeast (35, 36) and mouse EXO70 (mEXO70) (37). We were able to trace 17 α-helices in the structure, which were further divided into three domains based on interdomain hinge points and the overall arrangement of helices: N-terminal (75-379), C-terminal (511-629), and middle (380-510) connecting the N- and C-terminal domains (Fig. 6A) (36). The relative conformations of the three domains in EXO70A1 and mEXO70 were similar, except for a slight difference in the orientation of the N-terminal domain. Superposition of the middle C-terminal domains of EXO70A1 and mEXO70 gave a root mean square deviation (rmsd) of 1.59 Å on 147 Cα atoms (Fig. 6B). We were able to trace all of the helices in the final structure, except that a number of loops connecting the helices were missing due to poor electron density. The statistics for the X-ray diffraction data and structure are summarized in Table 3. Such a high structural similarity suggested that, despite low sequence identity (32% in middle C-terminal domains), the biochemical functions of plant and mammalian EXO70 proteins were most likely conserved. This result supported our conclusion that ES2 targets EXO70 in plants and three mammals (rats, humans, and, probably, mice).

With an available crystal structure of EXO70A1, a molecular docking tool, Autodock, was applied to predict possible ES2 binding sites of the EXO70A1, by fixing the protein and allowing ES2 to freely bind to several potential pockets of the EXO70A1. We decided to use molecular docking to predict possible ES2 binding sites on EXO70A1. Using Autodock (38), we found one possible binding pocket located at the C terminus of EXO70A1. The binding cavity was principally composed of the hydrophobic amino acids Y592, L596, K597, P601, R598, I613, and T616 (Fig. 6C). The conformation of ES2 can fit well in the C-terminal pocket of the WT EXO70A1 protein. The fluorine-containing aromatic ring of ES2 locates in the cavity created by P601, R598, and K597 main chains. The iodine-containing aromatic ring is surrounded by hydrophobic sidechains of several residues, including I613, L596, Y592, and T616. When we performed the docking with mutations of L596 and I613 to Ala, the binding pocket became larger because of the missing of I613 and L596 sidechains. However, in this case, the ES2 conformation couldn’t fit the pocket properly (Fig. 6D). Although the iodine-containing aromatic ring remained deeply in the pocket and formed interactions with A613, fewer interactions between the fluoride-containing aromatic ring and Y592/A596 were shown. In addition, in contrast to WT EXO70A1, the L596A and I613A mutations resulted in missing attractions between the ES2 and P601, R598, and T616, which would generate higher binding free energy and a weaker ligand-binding mode. To confirm that the predicted pocket plays a role in ES2 binding, we mutated amino acids L596 and I613 to Alanine (L596A;I613A) and tested for the binding activity of mutant protein to ES2 using STD-NMR. We found that, under the same protein and ES2 concentrations, L596A;I613A had less binding to ES2, which was reflected by reduced STD-NMR integral integrity (Fig. 6 E and F and Fig. S10 A and B). This observation was consistent with the prediction that L596 and I613 participate in interaction with ES2 although the mutations did not completely abolish the interaction between EXO70A1 and ES2. We also compared the thermophoretic mobility of EXO70A1 and EXO70A1-L596A;I613A when titrated with ES2 (Fig. 6G). Although the maximal thermophoretic mobility was lower for the mutated protein than for EXO70A1, the K_d of 252 ± 50.6 for the interaction of ES2 with EXO70A1-L596A;I613A is not significantly different at a 95% level of confidence than the K_d for ES2 with EXO70A1 (Table 2). Despite this result, the nonlinear fit for the interaction of ES2 with EXO70A1 is more robust with an R-squared of 81.7% than for the interaction of ES2 with EXO70A1-L596A;I613A, with an R-squared of 46.5%. This observation suggested that amino acids L596 and I613 were not essential for binding of ES2 but may interact with binding site residues and indirectly affect the local binding microenvironment. Amino acids L596 and I613 are conserved between Arabidopsis EXO70A1 and mammalian EXO70 proteins (Fig. S10C), explaining why ES2 could interact with both Arabidopsis EXO70A1 and rat EXO70.

T-DNA insertion mutant SALK_026036C (exo70A1-3) was obtained from the Arabidopsis Biological Resource Center. PIN2::PIN2:GFP;Ara7:mRFP line was a gift from Jiri Friml (Institute of Science and Technology Austria, Klosterneuburg, Austria). Transgenic seeds lines of 36S::GFP:EXO70A1 (41), GOT1p:YFP (42), mRFP:Ara7 (43), PIN2::PIN2:GFP (44), PIN2::PIN1:GFP (45), ROP6:GFP (46), BRI1:GFP (47), SYP61:CFP (17), HDEL:GFP (48), SYP22:YFP (17), PGP4:GFP (49), and PIP2:GFP (50) have been published previously.

Half-strength Murashige and Skoog medium (0.5× MS) [0.5× Murashige and Skoog salts, 1% (wt/vol) sucrose (pH 5.7)] was used as growth media. For plants that were grown on solid 0.5× MS media, 0.8% (wt/vol) phytoagar (Research Products International) was included. To test a chemical's growth effect, the chemicals were added to 0.5× MS solid media at 55 °C before pouring the plate. For short-term chemical treatment, the seedlings were grown on 0.5× MS solid media for 5–7 d and then transferred to 24-well plates containing 0.5× MS liquid media and chemicals. For dark treatment of PIN2::PIN2:GFP seedlings in DMSO or ES2, the seedlings were grown on 0.5× MS solid media for 5 d, and the seedlings were treated with ES2 or DMSO in the dark for 4 h before imaging with confocal microscopy. All plants, including soil-grown plants, were grown in an environmental chamber at long-day lighting conditions (16 h light/8 h dark) and a temperature of 22 °C. The root hair growth phenotypes were documented using a SPOT camera (SPOT Imaging Solutions) connected to a dissecting microscope.

Total RNA isolation, reverse transcription, and PCR analysis of the EXO70A1 transcript in exo70A1-3 followed the published protocol (51). The sequences of primers used in RT-PCR were as follows: EXO70A1 F1, ccATGGCTGTTGATAGCAGA; EXO70A1 R1, CGGAGGAGATCGAATTGAGA; EXO70A1 R2, TGGCTATAGCATCCCCAAAG; EXO70A1 F3, GATGGAACTGTCCACCCACT; EXO70A1 R3, ACCCAATCATCGCCTAACAA; Actin 2 F, GTTTTGCGTTTTAGTCCCATTGT; Actin 2 R, ACAAAAGGAATAAAGAGGCATCAATT.

Pollens from soil-grown Arabidopsis plants were dusted on a solid medium [18% (wt/vol) sucrose, 0.01% (wt/vol) boric acid, 1 mM CaCl₂, 1 mM Ca(NO₃)₂, 1 mM MgSO₄, pH 6.4, and 0.5% (wt/vol) agar] and incubated at 28 °C for 3 h before observation under an inverted microscope (Nikon Eclipse TE300). Images were taken by a cooled CCD camera (Hamamatsu CA4742-95) attached to the microscope. To measure the length of pollen tubes treated by ES2, an 8-mM ES2 stock solution was added to the pollen medium to a final concentration of 2 μM, 4 μM, 8 μM, or 16 μM before dusting the pollens on the medium. ImageJ software (rsb.info.nih.gov/ij) was used for measuring the length of pollen tubes.

Fluorescence imaging was performed using a Leica TCS SP5 confocal microscope (Leica Microsystems). The manufacturer’s default settings were used for imaging GFP-, RFP-, and YFP-tagged proteins. To image FM4-64–stained cells, a laser line of 543 nm was used for excitation, and an emission light with a wavelength of 600–700 nm was collected. For BFA washout experiments, 5-d-old seedlings were treated with 40 µM BFA for 2 h and quickly washed three times with the normal media. The treated seedlings were transferred to normal media containing 0.5% DMSO or 40 µM ES2 and recovered for 90 min before imaging with confocal microscopy.

To quantify the size of PIN2 agglomerations induced by ES2 treatment, 5-d-old PIN2::PIN2:GFP seedlings were treated with 40 µM ES2 for 2 h, and the root epidermal cells in the meristem zone were imaged with confocal microscopy. Z-stack images that cover the entire volume of the epidermal cells were collected. From each Z-stack image, a few adjacent image slices that do not have overlapped agglomerations in the XY directions or the Z direction were selected, and a maximum Z-projection image was generated by ImageJ software (imagej.nih.gov/ij/). The maximum Z-projection images were thresholded to get rid of the diffusive fluorescence, and the agglomerations within a single cell were manually selected and then measured by the “Analyze Particles” function of ImageJ. The agglomerations with a maximum diameter of fewer than 2 pixels were discarded during statistic analysis.

To measure the colocalization between ES2-induced PIN2 agglomerations with RabF2b/Ara7, 5-d-old seedlings of PIN2::PIN2:GFP;RabF2b/Ara7:RFP were treated with 40 µM ES2 for 2 h, and the root epidermal cells in the meristem zone were imaged by confocal microscopy under the line sequential scanning mode for GFP and RFP in XYZ directions. The collected two-channel Z-stack images were thresholded in both channels to get rid of the background fluorescence and were then analyzed by the “Colocalization” plugin in ImageJ. The resultant Z-stack images were examined manually to find whether each PIN2 agglomeration was associated with a punctate RabF2b/Ara7 structure. A total of 120 cells from 12 individual seedlings were examined.

To analyze the effect of ES2 on EXO70A1 dynamics in root hair cells, the FRAP module in a SP5 confocal microscope was used. Seven-day-old GFP:EXO70A1 seedlings were treated with 0.05% DMSO or 4 µM ES2 for 1 h. Root hairs that had a horizontal orientation and were not twisted by the glass slide and coverslip were selected, and the image plane was focused on the region where the root hair width was at a maximum. The region of interest (ROI) for photobleaching was selected by freehand selection tool to include the plasma membrane and the cytosolic pool of GFP:EXO70A1.

Wild-type (WT) seeds were plated on normal media containing DMSO or ES2, stratified for 2 d, and then light-treated for 8 h. Plates were then placed vertically in the dark. After 3 d of growth, plates were rotated 90°. After another 2 d of growth, seedlings were documented using an Epson scanner (Model 2450), and the angles of root curvature were quantified using ImageJ. The gravitropic root response was also observed using high temporal resolution imaging. Five-day-old seedlings of WT were reoriented by rotating the plates 90°. High temporal resolution images captured root curvature every 2 min for 8 h using an AVT Marlin camera. Images were exported, and root curvature after gravity stimulation was measured by MATLAB-based custom image analysis software (53). Root curvature was then graphed as a function of time.

To verify that the mRNA corresponding to the N terminus of EXO70A1 was truly translated into a polypeptide in exo70A1-3 cells, we isolated total proteins from 10-d-old WT and exo70A1-3 homozygous seedlings using the same procedure as was used in the pull-down assay. The total proteins were separated on SDS/PAGE and then stained with Coomassie blue. Proteins with the molecular weight of around 70 kDa (corresponding to full-length EXO70A1) and 25 kDa (corresponding to the N-terminal portion of EXO70A1) were excised from the stained gel using a razor blade. The gel bands were processed and treated with trypsin as described (54). Because EXO70A1 has a relatively low abundance in cells, a general proteomics profiling of the gel bands is not a suitable method to be able to determine the presence of EXO70A1 in the samples. With a targeted analysis method (55), an Escherichia coli-expressed EXO70A1 was used to determine that a peptide ion corresponding to amino acids 90–109 showed the strongest signal intensity in a nano-LC/MS spectrum. After long hours of extensive washing until there were not any detectable signals due to column carry over, samples from gel bands were injected and analyzed with nano-LC/MS for the detection of this signature peptide ion, which served as evidence of the presence of either full-length EXO70A1 or its N terminus. Subsequently, further nano-LC/MS/MS was also performed only for this ion to confirm its amino acid sequence.

To pull down proteins bound to biotinylated ES2 analogs, protein extracts from 16-d-old seedlings grown from normal 0.5× MS agar media with the plates in a vertical orientation were used. Then, a 4-mL extraction buffer (1× PBS, 0.5% Triton X-100, 2 mM DTT, 1× protease inhibitor mixture) was added to the tissue powder (ground in liquid nitrogen) that resulted from 2 g of seedlings. The cell extracts were passed through two layers of miracloth, and the flow-through was first spun at 1,000 × g for 30 min. The supernatant from 1,000 × g was collected and spun at 16,000 × g for 15 min. The resulting 16,000 × g supernatant fraction was used as protein input during pull down. High capacity streptavidin agarose resins (Thermo Scientific) were equilibrated with protein extraction buffer and then incubated with 100 µM biotin-tagged ES2 analogs at room temperature with gentle end-to-end inverting for 1 h. The streptavidin resins were collected and incubated with 2 mL of protein extract for 2 h at room temperature with end-to-end inverting. The streptavidin resins were then washed with extraction buffer three times, and the putative ES2 binding proteins were eluted with 100 µL of extraction buffer containing 100 µM ES2 by end-to-end inverting for 1 h at room temperature. The eluted proteins were digested with 1 µg of trypsin. Tryptic peptides were analyzed with a five-fraction MudPIT method described in a previous study (56). To detect EXO70A1 protein on the streptavidin resins after pull down, 1× SDS loading buffer was added to the resins and boiled for 5 min. The entire resin fractions were loaded to SDS/PAGE for Western blotting using an anti-EXO70A1 antibody.

We followed the published protocol for a DARTS assay (39). The protein extract used for the DARTS assay was obtained using the same protocol as the pull-down assay. In brief, 300 μL of protein extracts were incubated with either 400 μM ES2 or 1% DMSO for 1 h at room temperature. This mixture was divided into six aliquots of 50 μL, to which different concentrations of pronase (Sigma) were added, and digested for 30 min at room temperature. We then added SDS-loading buffer and boiled the samples to stop the reaction. The denatured samples were loaded to SDS/PAGE, and the same membrane was probed with anti-EXO70A1 and anti-actin antibodies. The resulting X-ray films were scanned and quantified using Image J. The signal intensity of each lane was calculated and subtracted from the background signal, and the ratio between the ES2-treated sample and the DMSO-treated sample at each of the pronase concentrations was calculated.

The cDNA sequence encoding EXO70A1 (residues 75–638) was amplified by PCR and subcloned into a modified pRSFDuet-1 vector, in which it was separated from a preceding hexa-histidine-SUMO tag by a ubiquitin-like protease 1 (ULP1) cleavage site. The plasmid containing the fusion protein was transformed into BL21(DE3) RIL cell strain (Novagen) for overexpression. The cells were grown at 37 °C and induced by 0.05 mM isopropyl β-d-1-thiogalactopyranoside (IPTG) when the OD₆₀₀ reached 0.6. After induction, the cells continued to grow at 20 °C overnight. The fusion protein was purified using a Ni-NTA column. Subsequently, the His6-SUMO tag was removed by ULP1 cleavage, followed by a second Ni-NTA run. The EXO70A1 protein was finally purified through size exclusion chromatography on a Superdex 200 16/60 column. The protein sample was concentrated to ∼17 mg/mL and stored in a buffer containing 50 mM Tris⋅HCl, pH 8.0, 300 mM NaCl, and 5% (vol/vol) glycerol. For crystallization, selenium methionine (SeMet)-labeled EXO70A1 was expressed in M9 minimum medium and purified in the same way as described above.

SeMet-labeled EXO70A1 was crystallized using the sitting-drop vapor diffusion method by mixing 1.5 μL of protein and 1.5 μL of reservoir solution containing 200 mM di-ammonium tartrate, pH 7.0, and 21.5% (wt/vol) PEG3350 at 16 °C. Crystals that grew into full size in a week were equilibrated in a reservoir solution supplemented with 25% (vol/vol) glycerol before being quick-frozen in liquid nitrogen. The X-ray diffraction data were collected at beamline 5.0.1. at the Lawrence Berkeley National Lab Center for Structural Biology (BCSB) and integrated and scaled with the HKL2000 package. The structure was solved by the SAD (single-wavelength anomalous dispersion) method, with 23 out of the 26 selenium atoms found in the two protein molecules per asymmetric unit. The initial model of EXO70A1 was built in Coot (57), followed by iterative cycles of model rebuilding and refinement using COOT and PHENIX (58). Translation-Libration-Screw-rotation model refinement was applied to improve the electron density map.

NMR spectra were collected at 25 °C using a Bruker Avance spectrometer operating at a 600-MHz proton frequency with a TXI 5-mm probehead with a z-gradient. The standard Bruker pulse program stddiffesgp.3 was used for data collection using a 2-s STD saturation time. Spectral acquisition and processing parameters were similar to those used in ref. 40. A sample prepared in 500 μL of buffered D₂O, containing 20 μM EXO70A1 protein and 400 μM ES2, was used for the initial STD-NMR experiment. To prevent precipitation of ES2 in the D₂O solutions, a 5-mM ES2 stock solution was made with DMSO-d₆ as solvent and added to the protein solution. The D₂O buffer solution contained 50 mM Tris•HCl (pH 8.0) and 150 mM NaCl. When comparing EXO70A1 and EXO70A1-L596A;I613A interaction with ES2, 200 μM ES2 and 10 μM protein concentrations were used.

MST experiments were carried out using a Monolith NT.115 (NanoTemper Technologies GmbH). Purified E. coli-expressed EXO70A1 and EXO70A1-L596A;I613A were fluorescently labeled with NT-647 (available from NanoTemper Technologies GmbH) via amine conjugation. Increasing concentrations of titrant (either ES2 or ES2 analog 8) were titrated against constant concentrations (50 nM) of the labeled target protein (either EXO701A or EXO701A-L596A;I613A) in a standard MST buffer (50 mM Tris, pH 7.5, 150 mM NaCl, 10 mM MgCl₂, 0.05% Tween 20). The small molecules were dissolved in DMSO for a final concentration of 7.5% (vol/vol) when added to an equal volume solution of target protein. MST premium-coated capillaries (Monolith NT.115 MO-K005) were used to load the samples into the MST instrument. Triplicate time traces were acquired for each test system. For each titration, two controls were measured to observe the thermophoretic response of the target protein when titrant was not present. The mean thermophoretic response of the controls was subtracted from the thermophoretic response of the target protein in the presence of the titrant. The resulting data were processed with Graphpad Prism software (Graphpad). Data were fit to the Hill equation using least-squares, nonlinear regression to calculate the max binding (B_max), Hill coefficient (h), and dissociation constant (K_d). K_d values were further compared between all test systems using one-way ANOVA within Graphpad Prism.

For pulse–chase experiments, HeLa cells were washed with PBS and incubated with serum-free MEM for 30 min. Fluorescence-labeled human transferrin (Life Technologies) was added to cells for 5 min using 50 µg/mL at 4 °C. After pulse, cells were washed three times in DMEM and incubated with fresh full media for indicated times and immediately fixed with 4% (wt/vol) paraformaldehyde (PFA). Fluorescent intensity was measured using ImageJ (NIH). The ratio of transferrin fluorescence intensity at 30 min or 90 min chasing against that of 5 min after chasing in ES2- and DMSO-treated cells was calculated in each experiment by dividing fluorescence intensities at 30 or 90 min by fluorescence intensities at 5 min. The SD was calculated from three independent experiments with around 100 labeled cells in each experiment. To study the effect of ES2 on EXO70 in mammalian cells, HeLa cells were cultured in DMEM with 10% (wt/vol) FBS. rExo70 in the pEGFP-C1 vector was transformed into cells with Lipofectamine 2000. Then, 24 h after transfection, the cells were treated with ES2 or DMSO for 3 h. Cells were fixed with 4% (wt/vol) PFA and observed using a Leica DMI 6000B inverted microscope equipped with a DFC350 FX camera and a 63× objective.

To better understand ligand-bound conformation and the effect of I613A and L596A mutation in an EXO70-ES2 system, we used an Autodock program (38) with a Lamarckian genetic algorithm to execute ligand docking by fixing a protein and allowing an ES2 to move around I613 and L596 at the C-terminal of EXO70A1. The 3D experimental coordinate of the EXO70A1 crystal structure was obtained in this study. We created a 3D structure of ES2 by using the VegaZZ program (59). The Autodock scoring function is a subset of the Assisted Model Building with Energy Refinement force field that treats molecules using the United Atom model; and Gasteiger charges (60) were assigned to the molecules by applying Autodock tools 1.5.4 (61). The ES2 docking simulations were performed on the two types of EXO70A1: One is the protein with the WT sequence and the other protein contains I613A and L596A mutations. Autogrid version 4.0 was used to create affinity grids with 0.375-Å spacing. The cubic grid box with a dimension of 2.25 nm was centered near the EXO70A1 C terminus. In each molecular docking, we trailed 10 docking simulations; and one million energy evaluations for each trail were performed.