GA Modulates the Balance Between Vacuolar Trafficking and Exocytosis.

To dissect the cellular GA-targeted processes that regulate PIN levels at the PM, we used the photoconvertible pPIN2::PIN2-Dendra line. The irreversible conversion of the Dendra fluorochrome from green to red allowed us to follow the subcellular fate of the already present PIN2 in GA-deficient and GA-treated conditions. Conceptually, GA can enhance PIN levels at the PM by (i) inhibiting PIN endocytosis, (ii) promoting its exocytosis, or (iii) inhibiting PIN vacuolar delivery and degradation.

In the nonphotoconverted situation, a GA synthesis inhibitor, paclobutrazol (PAC) (14), strongly decreased the total PIN2-Dendra signal, whereas GA treatment rapidly reverted this effect (Fig. 1 A–D). Therefore, PIN2-Dendra reacted identically to PAC and GA treatments as PIN2-GFP, the regulation of which by GA has been verified both by microscopy (Fig. S1 A–D) and Western blot analysis (Fig. S1 E–H). However, when PIN2-Dendra was photoconverted into its red form and its depletion from the PM was followed, the decrease in the red signal over time was not affected by PAC pretreatment (Fig. 1 E–H and O), suggesting that PIN2 internalization from the PM is not affected by GA availability and, thus, that the observed PIN2 depletion from the PM in GA-deficient conditions (8) is not due to enhanced PIN2 endocytosis. To assess the alternative scenario of GA promoting exocytosis, we photoconverted PIN2-Dendra and monitored the green signal recovery at the PM, which reflects the delivery of de novo synthesized PIN2-Dendra proteins (Fig. 1Q). Notably, in GA optimal conditions, the green signal intensity increased up to 40% within 5 h (Fig. 1 I, J, and O). On the contrary, in PAC-treated roots, the recovery was severely inhibited, reaching only ∼5% (Fig. 1 K, L, and O). GA treatment was sufficient to revert the PAC effect and to recover the green PIN2-Dendra at the PM (Fig. 1 M, N, and P). Treatment with Brefeldin A (BFA), an established inhibitor for PIN delivery to the PM (15, 16), had no obvious effect on PAC- and GA-modulated PIN2 recovery at the PM (Fig. S2 A–D and G). This illustrates a BFA-insensitive delivery of newly synthesized PIN2 to the apical membranes of root epidermis cells, consistent with previous studies (17, 18).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 1.

GA effects on PIN2 vacuolar trafficking and exocytosis. (A–C) Decrease of total PIN2-Dendra signal at PM in seedlings pretreated with 1 µM paclobutrazol (PAC) for 1 d (B) that was rescued by 50 µM GA treatment for 4 h (C) compared with nontreated seedlings (A). (D) PIN2-Dendra signal quantification. A rectangular region of interest (ROI) at the PM was used for PM signal intensity quantification, and the PM signal was measured in 16 epidermal cells, including trichoblasts and atrichoblasts in eight individual roots. (E–H) Photoconversion of PIN2-Dendra into its red form induced by illuminating the region of interest and the depletion of the red signal over time followed in nontreated seedlings (E and F) versus seedlings pretreated with 1 µM PAC (G and H). Note the increased PIN2-Dendra vacuolar delivery in PAC-treated samples (G). (I–N) Green signal recovery at the PM after photoconversion reflects the delivery of de novo synthesized protein in nontreated seedlings (I and J), pretreated with 1 µM PAC (K and L), and treated with 50 µM GA for 5 h after 1 µM PAC pretreatment (M and N). Note the faster accumulation of de novo PIN2 at the PM in seedlings untreated versus PAC treatment (I and J) and treated with GA (M and N). (O and P) Decrease in PIN2-Dendra red signal and increase in the green signal 5 h after photoconversion, reflecting the rate of endocytosis and exocytosis, respectively. The percentage of the increase in the green signal intensity was measured after 5 h, revealing the exocytosis rate in untreated epidermal cells versus 1 µM PAC treated (O) and roots pretreated with 1 µM PAC and followed by 50 µM GA treatment (P). Note the increase in the exocytosis rate of PIN2 in roots upon GA application. (Q) Schematic representation of PIN2-Dendra red internalization reflects the rate of endocytosis and delivery of a newly synthesized green PIN2-Dendra to the PM linked to the exocytosis rate. (Scale bar: 10 µm.)

Next, we confirmed the PIN2-Dendra observations by the fluorescence recovery after photobleaching (FRAP) and live imaging of PIN2-mCherry. The PIN2-mCherry fluorescence was almost completely bleached in the whole cellular region, so that the signal recovery indicated predominantly the delivery of de novo synthesized PIN2. As shown for PIN2-Dendra, PAC treatment decreased the PM PIN2-mCherry recovery with a concomitantly increased vacuolar delivery (Fig. S2 E, F, and H–K).

In summary, the photoconversion and FRAP experiments revealed that GA-deficient conditions have little effect on PIN2 endocytosis but lead to a strong inhibition of PIN2 delivery to the PM and trafficking redirection to the vacuole. GA treatment is sufficient to completely revert this effect by inhibiting vacuolar targeting and recovering PIN2 at the PM. This suggests that GA signaling targets processes balancing PIN2 vacuolar trafficking and its recycling back to the PM.

Nontranscriptional GA Mechanism Regulates Trafficking of Multiple Cargoes.

We performed several experiments to confirm that GA increases PIN2 delivery to the PM but not through the regulation of PIN2 transcription or protein synthesis (Fig. S3). Quantitative reverse transcription PCR (qRT-PCR) showed that PIN2 transcription was not influenced by PAC or GA treatments (Fig. S3 A and B). Moreover, real-time recording of PIN2 promoter activity (in pPIN2::NLS-GFP roots) showed no differences between mock, PAC, and PAC+GA treatments (Fig. S3 C–E). PIN2 promoter activity-independent action of GA was confirmed by the observation that PIN2 and PIN1 distributions in estradiol-inducible and 35S-driven lines were regulated by GA in a similar way as protein fusions under their native promoters (Fig. S3 H–K). Furthermore, PIN2-GFP levels were monitored microscopically following GA and PAC cotreatments with cycloheximide (CHX), an inhibitor of protein synthesis. Whereas CHX significantly decreased the level of PIN2 at the PM, PAC treatment led to an additional drop in PIN2 abundance, which was reversed by GA even in the presence of CHX (Fig. S3 G and F). These observations confirm that not only GA regulates PIN abundance independently of transcription but also that the GA effect itself utilizes a nontranscriptional mechanism.

To test the specificity of these GA effects on PIN proteins, we examined vacuolar targeting of different cargo proteins in GA-deficient conditions. The vacuolar degradation increased not only for PIN2 but also for several other PM proteins, such as Boron transporter 1 (BOR1), Brassinosteroid insensitive 1 (BRI1), P-glycoprotein 19 (PGP19), and PM intrinsic protein 1;4 (PIP1;4) (Fig. S4 A–I). These findings are not consistent with previous observations that GA did not visibly affect PM stability of AUX1, PGP19, and PIP2 (9). This discrepancy might be explained by a different evaluation method: GA effect on the PM proteins abundance (9) versus the dark-visualized protein vacuolar delivery, which seems to be more sensitive (12). Using the dark conditions or concanamycin A treatment, we detected a pronounced PIN2 vacuolar degradation following PAC treatment, which was reversed by GA (Fig. S4 J–O). Similar differences in vacuolar accumulation of cargo proteins like BOR1 or PGP19 were observed despite the absence of obvious differences in abundance at the PM (Fig. S4 A, B, F, and G). qRT-PCR for these proteins in identical conditions showed no significant difference in mRNA levels following PAC or GA treatment, with the exception of PIP1;4 (Fig. S5).

These results imply a nontranscriptional mechanism of GA regulation of the balance between vacuolar trafficking and recycling back to the PM of multiple PM cargoes.

Gibberellin Targets the Retromer Complex for Trafficking Modulation.

The evolutionarily conserved retromer complex that functions at the interface of exocytosis/recycling and vacuolar trafficking mediates the recovery of proteins from the vacuolar degradation pathway back to the recycling pathway (19). As shown previously (20, 21), SNX1 and other components of the retromer complex are involved in PIN protein sorting from the prevacuolar compartment to make them again available for recycling to the PM (12). To test whether SNX1-dependent protein retrieval is required for GA action on PIN2 trafficking, we analyzed the GA sensitivity of snx1 mutants. In PAC-pretreated snx1-1 seedling roots, GA application was largely ineffective to restore PIN2-GFP levels at the PM (Fig. 2 A–D, Q, and R). Moreover, snx1-1 seedlings were less sensitive to the application of PAC because we observed a less pronounced decrease of the total PM PIN2 in the snx1-1 mutant in comparison with the control (Fig. S6 A–D, G, and H). Similar observations were made in the protein affected trafficking 3 (pat3) mutant defective in the VACUOLAR PROTEIN SORTING35A (VPS35A), another subunit of the retromer complex (21). Also in pat3-3 roots, GA was unable to restore PIN2-GFP incidence at the PM (Fig. S7 A, B, and E), confirming the importance of the retromer function for GA effect on PIN trafficking. However, in mutants generally defective in vacuolar trafficking and function, such as pat4 (22), the sensitivity to GA was not affected (Fig. S7 C–E). This observation together with previously shown GA-insensitive PIN2 accumulation in Wortmannin bodies (9) indicates that the GA regulation of the PM protein accumulation is not due to decreased transit into prevacuolar compartments of PIN2 but is regulated more upstream and specifically requires the retromer complex components.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 2.

GA action on PIN2 requires SNX1 and its associated protein CLASP1. (A–F) PIN2-GFP (A and B), snx1-1×PIN2-GFP (C and D), and clasp1×PIN2-GFP (E and F) seedlings pretreated with 1 µM PAC and subsequently treated with 50 µM GA. Reduced sensitivity to GA was observed in snx1-1 (C and D) and clasp1 (E and F) mutant seedlings in comparison with wild type (A and B). (G and H) SNX1-GFP-labeled endomembranes in the root epidermis of untreated seedlings (G) and treated with 1 µM PAC (H). Note the abnormally enlarged SNX1-GFP–labeled bodies in the PAC-treated seedlings (H) in comparison with the punctate bodies in untreated seedlings (G). (I) Decrease in microsomal (membrane) SNX1 protein upon PAC treatment and increase upon GA in Arabidopsis root tips by Western blot analysis. (J and K) GFP-CLASP1 association with the cell edges at 0 min (J) and after 45 min (K) of GA application in epidermal cells. Note the gradual enrichment of the GFP-CLASP1 at the cell edges starting 45 min after GA application (K). (L–N) Cellular localization of GFP-CLASP1 in untreated seedlings (L), treated with 1 µM PAC (M) and treated with 40 µM oryzalin (N). Note that PAC-treated roots have less cell edge-localized GFP-CLASP1 but more GFP-CLASP1 accumulated in the cytoplasm (M) compared with mock (L), whereas oryzalin treatment entirely removed GFP-CLASP1 from the cell edges, and most of it localized in cytoplasm and vesicles (N). (O and P) Quantification of the size of SNX1-GFP–labeled bodies (O) and total amount of endosomes (P). (Q–S) Quantified relative increase of PIN2-GFP signals after GA treatment in the root epidermis of wild type (Q), snx1-1 (R), and clasp1 (S). Measurements were performed as described in Fig. 1. Color code was used for PIN2-GFP intensity visualization. (Scale bar: 10 μm.)

In addition, we observed a pronounced GA effect on SNX1-positive endosomal compartments. PAC treatments resulted in the appearance of abnormally enlarged SNX1-GFP bodies, distinct from the punctuated SNX1-GFP–labeled endosomes (Fig. 2 G–O); however, the total number of SNX1-GFP endosomes was not significantly changed (Fig. 2P). Prolonged PAC treatments led to an increased cytosolic SNX1-GFP signal (Fig. S7 F and G). GA application resulted in reduced numbers of these abnormal endosomes (Fig. S7 H and I). Overall, it has been common to observe in one root treated with PAC both patterns—cells with enlarged SNX1-GFP endosomes and the cells with SNX1-GFP in the cytosol. Western blots of root samples confirmed a decrease of SNX1-GFP protein in the membrane (microsomal) fraction following PAC treatment (Fig. 2I). When other markers for the endosomal compartments, such as VHA-GFP or FYVE-GFP, were studied, no such effects occurred (Fig. S7 J–M), which hinted at a more specific GA effect on the retromer-related compartments.

The observations that mutants defective specifically in retromer components are insensitive to GA in terms of regulation of trafficking together with GA effects on the SNX1-labeled endomembranes suggest that GA targets the retromer or related components to balance vacuolar trafficking and PM delivery.

Gibberellin Effect on Trafficking Requires CLASP1 Activity.

The MT-associated protein CLASP1 has been shown to interact directly with SNX1 and this interaction is important for retromer-mediated trafficking regulation (13). Therefore, we assessed the potential contribution of CLASP1 in GA-mediated regulation of PIN2 trafficking. The clasp1 mutant was less sensitive to PAC treatment in terms of PIN2 depletion from the PM (Fig. S6 E, F, and I), and more importantly, GA treatment was ineffective to recover PIN2 at the PM (Fig. 2 E, F, and S), suggesting that both retromer complex and CLASP1 activity are necessary for the GA effect on trafficking.

Next, we used GFP-CLASP1 fusion protein driven by its endogenous promoter to visualize the subcellular CLASP1 distribution. PAC-treated roots showed slightly but significantly fewer cells with cell edge-localized GFP-CLASP1 compared with mock (Fig. 2 L and M). PAC treatment also led to an increase of the GFP-CLASP1 cytosolic signal, similarly to our observations when MTs were depolymerized by oryzalin. The complete MT depolymerization with oryzalin removed GFP-CLASP1 from cell edges entirely (Fig. 2N). By using time lapse imaging following GA treatment, we perceived a gradual enrichment of GFP-CLASP1 association with the cell edges, starting as early as 45 min after GA application (Fig. 2 J and K). These observations suggest that GA targets both the retromer complex and its associated protein CLASP1, further strengthening the notion that these regulators are part of the mechanism by which GA regulates PIN2 trafficking.

The Gibberellin Effect on Trafficking Needs Intact and Dynamic Microtubules.

Because CLASP1 is known to be associated with MTs in different eukaryotes (23⇓–25), we tested the role of MTs in the GA effect on PIN2 trafficking. The effect of GA on MT orientation is well described, and in the majority of the studies, GA promotes the transversal orientation of MTs (26, 27). Taking into account that SNX1 and CLASP1 might be involved in the GA effect on trafficking, we studied MT behavior after PAC treatment. We used live imaging of different MT reporter lines (such as MAP4-GFP, VENUS-TUA6, and EB1b-GFP) and found that PAC treatment resulted in the transition of the MT array from a more organized pattern with a distinctive bundled network to a more randomized pattern with fewer MTs per area unit (Fig. S8 A and B). Using an Eb1b-GFP marker line, we recorded the enrichment in longitudinally or obliquely oriented MTs in the roots treated with PAC (Fig. S8 E–H). Similarly, as observed for SNX1-GFP and GFP-CLASP1, we found that prolonged PAC treatment led to an overall decrease in the MT network labeled with VENUS-TUA6 and an increase in fluorescent signal in the cytoplasm (Fig. S8 C and D). Overall, we observed that at the early stages of GA deficiency, MT network stayed intact although it reoriented randomly or longitudinally coinciding with mostly cell-edge localized CLASP1, and SNX1 present in enlarged endosomes. At later stages, polymerization of MTs was visibly affected correlating with an increase in the cytosolic CLASP1 and SNX1.

To identify a possible requirement for MT cytoskeleton in GA-modulated PIN2 stability at the PM, we used oryzalin to disrupt the MTs (28, 29) and tested whether GA was still able to restore PIN2 abundance in PAC-grown roots. Whereas GA readily restored PIN2 at the PM in the mock control (Fig. 3 A and B), this effect was abolished when MTs were depolymerized by oryzalin (Fig. 3 C–E). Also, when MTs were stabilized by taxol (30), the GA effect on the PM restoration of PIN2 was abolished (Fig. 3 F–J). Oryzalin and taxol treatments themselves resulted in a decrease of PIN2 at the PM compared with control plants (Fig. S8 I–N). To test if genetic interference with MTs also changes PIN2 responses to GA, we used the katanin 1 (ktn1) mutant known to be compromised in severing MTs and proper formation of cortical MT arrays (31, 32). Notably, in ktn1 roots pretreated with PAC, GA application was not able to restore PIN2 at PM (Fig. 3 K–O) or to decrease vacuolar PIN2 degradation (Fig. S9 A–E). In general, ktn1 mutants contained less abundant PIN2 at PM, which was not further affected by PAC application (Fig. S8 O–Q). Furthermore, tortifolia 1 (tor1), another mutant defective in MTs (33), showed an unusual response to PAC, which rather increased instead of decreasing PIN2 incidence at the PM (Fig. S9 F–M).

• Download figure
• Open in new tab
• Download powerpoint

Fig. 3.

Regulation of GA-dependent PIN2 trafficking by MTs. (A–E) Depolymerization of MTs with 20 µM oryzalin prevented GA (50 µM) effect on PIN2 restoration at the PM in PAC-treated seedlings (C and D) compared with the oryzalin-free mock (A and B). Quantifications (E). (F–J) Ability of GA to increase PIN2 at the PM of PAC-treated seedlings was abolished by 10 µM of MT-stabilizing agent taxol (H and I) compared with the taxol-free mock (F and G). Quantifications (J). (K–O) Efficiency of GA to increase PAC-compromised PIN2 amount (K and L) reduced in the ktn1 mutant (M and N). Quantifications (O). Color code was used for PIN2-GFP intensity visualization. (Scale bar: 10 µm.)

The pharmacological and genetic interferences with MTs revealed that an intact, dynamic MT cytoskeleton is essential for GA effect on PIN2 trafficking. Thus, GA regulation of trafficking requires retromer activity, MT cytoskeleton, and CLASP1 protein, which directly links these cellular structures.

The DELLA Components of GA Signaling Mediate GA Effect on Trafficking.

Next, we assessed which GA signaling mechanism acts upstream of MTs and retromer to regulate PIN trafficking. DELLA proteins are key components of GA signaling, and they are typically viewed as nucleus-based transcriptional modulators of downstream genes (34⇓–36). Additionally, a supplementary mode of DELLA action has recently been proposed based on the DELLA interaction with Prefoldins (37), which are important factors of tubulin folding in the cytoplasm (38).

To test the involvement of DELLA proteins in trafficking regulation by GA, we used lines with deficient or modified DELLA activity. First, we studied the GA sensitivity of PIN2 trafficking in the quintuple della knockout mutant. Notably, in the quintuple della roots, PAC-based interference with GA biosynthesis did not result in a substantial decrease in PIN2 at the PM, and more importantly, GA treatment was ineffective in up-regulating PIN2 levels at the PM (Fig. 4 A–H). Furthermore, the dominant-negative DELLA mutant gaiΔ17 roots showed less PIN2 abundance in comparison with wild type or della knockout mutant (Fig. S10 A–C and E), similarly to the previous observations for PIN1 (9). Also, the gaiΔ17 mutant roots were insensitive to GA treatment such that it failed to restore the PIN2 levels (Fig. S10 D and E). Finally, we introduced PIN2-GFP into the GAI:gai-1-GR line, which after induction by the synthetic glucocorticoid dexamethasone (DEX) targets a dominant DELLA mutant protein from the cytosol to the nucleus (37). In this line, the DEX treatment resulted in a significant down-regulation of PIN2 levels at the PM (Fig. 4 I–L). Concurrent GA application and DEX induction did not recover PIN2 levels at the PM (Fig. 4K). Collectively, these experiments demonstrated that GA effect on PIN2 trafficking strictly requires the DELLA function.

• Download figure
• Open in new tab
• Download powerpoint

Fig. 4.

DELLA proteins and their PFD interactors mediating GA effects on PIN2. (A–H) Immunolocalization of PIN2 at the PM in nontreated wild-type (A–C) and della knockout mutant (E–G) seedlings. Note reduced sensitivity of della knockout seedlings to 1 µM PAC and 50 µM GA (F and G) in comparison with wild type (B and C). Quantifications (D and H). (I–L) GAI::gai-1-GR induced in seedlings by treatment with 10 µM DEX. After induction of dominant-negative DELLA, PIN2 amount at the PM was significantly reduced (J), and GA treatment did not restore PIN2 levels at the PM (K). Quantifications (L). (M–Q) Reduced efficiency of 50 µM GA to increase PIN2 level at the PM in 1 µM PAC- treated seedlings of pfd5 pfd6 (O and P) mutant seedlings in comparison with wild type (M and N). Color code was used for PIN2-GFP intensity visualization. The 3D Z stacks are presented. Quantifications (Q). (Scale bar: 10 μm.)

A possible mechanism for DELLA proteins to regulate MTs and trafficking is through the tubulin-folding factors Prefoldins (PFDs) that are known to control MT folding and dynamics in nonplant organisms (39, 40) and have been shown to interact with DELLAs in plants (37). Indeed, the pfd6 mutant had clearly reduced levels of PIN2 at the PM (Fig. S10 F–H). To further confirm the role of PFDs in GA action on PIN2 trafficking, we analyzed GA sensitivity of the pfd5 pfd6 double mutant. In PAC-pretreated pfd5 pfd6 seedling roots, GA application was ineffective in restoring PIN2-GFP levels at the PM (Fig. 4 M–Q). Moreover, pfd5 pfd6 double mutants were less sensitive to PAC treatment in terms of PIN2 decrease in the PM (Fig. S10 I–L). This observation supports the hypothesis that the DELLA interacting partner PFD mediates GA-dependent regulation of PIN2 trafficking.