Lysine⁶³-linked ubiquitylation of PIN2 auxin carrier protein governs hormonally controlled adaptation of Arabidopsis root growth

Abstract

Cross-talk between plant cells and their surroundings requires tight regulation of information exchange at the plasma membrane (PM), which involves dynamic adjustments of PM protein localization and turnover to modulate signal perception and solute transport at the interface between cells and their surroundings. In animals and fungi, turnover of PM proteins is controlled by reversible ubiquitylation, which signals endocytosis and delivery to the cell’s lytic compartment, and there is emerging evidence for related mechanisms in plants. Here, we describe the fate of Arabidopsis PIN2 protein, required for directional cellular efflux of the phytohormone auxin, and identify cis- and trans-acting mediators of PIN2 ubiquitylation. We demonstrate that ubiquitin acts as a principal signal for PM protein endocytosis in plants and reveal dynamic adjustments in PIN2 ubiquitylation coinciding with variations in vacuolar targeting and proteolytic turnover. We show that control of PIN2 proteolytic turnover via its ubiquitylation status is of significant importance for auxin distribution in root meristems and for environmentally controlled adaptations of root growth. Moreover, we provide experimental evidence indicating that PIN2 vacuolar sorting depends on modification specifically by lysine⁶³-linked ubiquitin chains. Collectively, our results establish lysine⁶³-linked PM cargo ubiquitylation as a regulator of polar auxin transport and adaptive growth responses in higher plants.

Plants have evolved a repertoire of mechanisms for continuously adapting vital parameters in response to fluctuating environmental conditions. Sensing and responding to such variations depend to a large extent on the activity of plasma membrane (PM)-associated proteins that function in stimulus perception and solute transport. Specifically, adjustments in subcellular distribution of PM proteins are an efficient means to modulate their activity and involve continuous protein cycling between PM and endosomes as well as irreversible targeting for degradation in the lytic vacuole/lysosome (1).

An evolutionary conserved machinery controls PM protein sorting for degradation, and, specifically in animals and fungi, it was demonstrated that PM protein fate is decisively influenced by their reversible ubiquitylation, triggering cargo endocytosis and delivery to the lytic compartment (2–4). Related mechanisms appear to be operative in plants (5–9) because, recently, ubiquitylation has been described for some plant PM proteins, linking nutrient transport and stimulus perception to endocytic protein turnover (6–9). Strikingly, different patterns of protein ubiquitylation have been observed, with mono-, di-, as well as polyubiquitylation implicated in regulating endocytic trafficking and degradation of distinct proteins (5). This resembles the situation in nonplant organisms, in which mono- and polyubiquitylation have been associated with cargo endocytosis (10).

Vacuolar sorting was also demonstrated for PIN1-type auxin carrier proteins, which are instrumental for cellular efflux of the phytohormone auxin (11, 12). PINs control numerous aspects of plant development and adaptive growth responses via readjustments in auxin distribution that are mediated by dynamic changes in PIN localization (11, 12). However, whereas intracellular cycling and phosphorylation-controlled transcytosis of PIN proteins were demonstrated to be intimately involved in control of auxin transport (13, 14), no clear biological role for PIN vacuolar targeting and degradation has been established so far.

Here we present a systematic analysis of PIN2 sorting and degradation in root meristem cells, in which we identified ubiquitylation as principal cis-acting signal sufficient for PIN2 endocytosis from the PM. Efficient targeting to the vacuole appears to require K63-linked polyubiquitin chain formation, whereas diminished K63-linked polyubiquitylation causes a stabilization of PIN2. PIN2 polyubiquitylation is stimulus-dependent, which feeds back on PIN2 stability by means of determining the rate of PIN2 vacuolar targeting. In addition, we provide evidence that variations in PIN2 ubiquitylation and vacuolar targeting affect auxin distribution in root meristems and demonstrate a function in the control of gravitropic root growth.

Altogether, our results identify K63-linked polyubiquitin chain formation as a rate-limiting and specific signal for vacuolar sorting of plant PM proteins and establish its role in hormonally controlled development of plants.

Results and Discussion

PIN2 Is Modified by K63-Linked Polyubiquitin Chains.

Earlier work demonstrated ubiquitylation of overexpressed PIN2, but no clear link to PIN protein fate has been established so far (15). We therefore focused on analysis of endogenous PIN2 and probed PIN2-immunoprecipitation (IP) from solubilized wild type (WT) membrane protein extracts with nondiscriminating ubiquitin antibody (P4D1). We observed diffuse signals extending into the high molecular weight range but no signals in IPs made with eir1-4 [a pin2 null allele (15)] protein extracts, indicating ubiquitylation of endogenous PIN2 (Fig. 1A). Similar results were obtained with eir1-4 expressing PIN2 cDNA under control of the PIN2 promoter, further establishing that nonectopically expressed PIN2 is subject to ubiquitylation (Fig. S1A).

Fig. 1.

Analysis of PIN2 ubiquitylation. (A) PIN2 IPs performed with rabbit anti-PIN2 from eir1-4 and WT probed with nondiscriminating ubiquitin antibody (α-UBQ) and an antibody, specifically recognizing ubK63-linked chains (α-K63-UBQ). (B) PIN2-specific IPs from eir1-4, Col-0, and rglg1 rglg2 probed with α-UBQ (Top) and α-K63-UBQ (Middle). (Bottom) IP probed with α-PIN2 used as loading control. Normalized signal intensities are indicated below blots. (C Upper) VENUS signals in root meristem cells of eir1-4 PIN2p::PIN2:VEN and rglg1 rglg2 PIN2p::PIN2:VEN (4 DAG, 12 roots were analyzed for each genotype). (Scale bar: 20 μm.) (Lower) Col-0 and rglg1 rglg2 membrane protein extracts probed with α-PIN2. Anti–α-tubulin (α-TUB) was used for normalization. Normalized signal intensities are indicated below blots. (D) PIN2 IPs from eir1-4, eir1-4 PIN2, eir1-4 PIN2p::pin2^12K-R, and eir1-4 PIN2p::pin2^17K-R probed for ubK63 chains (α-K63-UBQ) and for PIN2 (α-PIN2). A distinct band corresponding to IgG is detectable in all IP Western blots (asterisk). (E) Orientation of root primary root tips of eir1-4 PIN2p::PIN2, eir1-4, eir1-4 PIN2p::pin2^12K-R, and eir1-4 PIN2p::pin2^17K-R at 6 DAG incubated on vertically oriented nutrient plates. Root angles are indicated as the deviation from 0°. Number of individuals is indicated.

Different types of ubiquitylation are involved in the control of protein endocytosis (10), and, in nonplant organisms, formation of K63-linked ubiquitin (ubK63) chains has been implicated in PM protein endocytosis and vacuolar/lysosomal targeting (16–18). We therefore probed PIN2 IPs with an antibody specific for ubK63 chains [HWA4C4 (19)] and obtained signals comparable to those observed with nondiscriminating ubiquitin antibody (Fig. 1A). This result indicates a function for ubK63 chain formation in controlling PIN2 fate.

Next, we analyzed PIN2 ubiquitylation in the Arabidopsis rglg1 rglg2 double mutant, which is deficient in RING-finger E3 ligases demonstrated to catalyze ubK63 chain formation in vitro (20). We found a reduction in the amounts of ubK63 chain-specific signals in PIN2-IPs, suggesting that RGLG proteins make an important contribution to PIN2 ubiquitylation in planta (Fig. 1B). However, we did not observe a complete loss of ubiquitin-specific signals, indicative of additional regulators involved (Fig. 1B). When viewing subcellular localization of a functional PIN2–VENUS fusion protein (PIN2p::PIN2:VEN) in rglg1 rglg2, no differences to WT could be detected (Fig. 1C and Fig. S1 D and E), implying that PIN2 targeting to the PM is not substantially altered. However, PIN2 protein levels in rglg1 rglg2 were increased, whereas transcript levels remained unaltered (Fig. 1C and Fig. S1F), suggesting that diminished ubK63 chain formation could stabilize PIN2.

Loss of PIN2 Polyubiquitylation Affects Root Gravitropism and Auxin Distribution.

To directly assess whether PIN2 ubiquitylation affects its stability, we generated pin2 alleles with diminished ubiquitylation and mutagenized the majority of 28 lysines found in the PIN2 ORF, each representing a potential ubiquitylation site. In total, a set of 21 mutant pin2 alleles, with variable numbers of lysines replaced by arginines, was tested for rescue of eir1-4 root gravitropism defects. Single-point mutations and combinations of a few K-to-R exchanges did not interfere with complementation (Table S1). However, combining six K-to-R point mutations, all affecting lysines in the PIN2 central hydrophilic loop, failed to fully complement eir1-4, becoming even more apparent the more lysines in the loop were replaced. Specifically, PIN2p::pin2^{K158, 201, 318, 321, 322, 361, 362, 363, 381, 497, 556, 614R} (pin2^12K-R) and PIN2p::pin2^{K221, 273, 303, 318, 321, 322, 361, 362, 363, 381, 401, 417, 429, 443, 457, 463, 464R} (pin2^17K-R), in which either 12 or 17 loop-residing lysines were mutagenized, failed to rescue eir1-4 (Fig. 1E). When analyzing ubiquitylation of pin2^K-R alleles, a prominent reduction was observed in pin2^12K-R and pin2^17K-R when probed with either nondiscriminating or K63 chain-specific ubiquitin antibody, whereas a pin2^K-R allele that rescued eir1-4 still exhibited ubiquitin-specific signals comparable to those of WT (Fig. 1D and Fig. S1 B and C). These findings imply that multiple lysines situated in the central hydrophilic loop redundantly control PIN2 K63-linked polyubiquitylation and might be essential for PIN2 function in root gravitropism.

Mutant pin2^K-R alleles that no longer complement the eir1-4 mutant phenotype could be affected in their functionality in auxin transport. We therefore determined auxin transport in tobacco BY-2 cells (21). Lines conditionally expressing PIN2 alleles showed reduced accumulation of [³H]1-naphthaleneacetic acid (NAA) over time, with pin2^12K-R:VEN exhibiting less tracer accumulation than WT PIN2:VEN did (Fig. 2A). These observations indicate functionality of pin2^12K-R in auxin efflux. PIN2:VEN and pin2^12K-R:VEN localized predominantly to the PM and exhibited a somewhat polar distribution at the junctions between neighboring BY-2 cells (Fig. 2 B and C). Remarkably, expression of pin2^12K-R:VEN in BY-2 cells reproducibly resulted in stronger signals at the lateral PM (Fig. 2D), which might give rise to reduced [³H]NAA accumulation as a result of elevated efflux (Fig. 2A).

Fig. 2.

Expression and activity of pin2^K-R alleles. (A) Relative [³H]NAA retention in BY-2 cells expressing TA::PIN2:VEN or TA::pin2^12K-R:VEN after 48 h of induction with 5 μM dexamethasone (DEX). Accumulation was determined in induced and noninduced cells and scored after 15 min of [³H]NAA incubation. Error bars (SEM) are indicated (n = 4). (B and C) Localization of PIN2:VEN (B) and pin2^12K-R:VEN (C) in dexamethasone-treated BY-2 cells. White arrowheads indicate “polar” reporter signals, and red arrowheads highlight prominent pin2^12K-R:VEN signals at the PM. (D) Normalized mean gray levels (8 bit) at the lateral PM domain of BY-2 cell aggregates expressing TA::PIN2:VEN or TA::pin2^12K-R:VEN. Signals from 32 TA::PIN2:VEN and 43 TA::pin2^12K-R:VEN cells were determined. SDs are indicated as bars. *P < 0.05, two-tailed t test, significant difference in signal intensities. (E and F) Expression of DR5::mRFP in eir1-4 PIN2p::PIN2:VEN (E) and eir1-4 PIN2p::pin2^12K-R:VEN (F) gravistimulated for 90 min. White arrowheads indicate the DR5 expression gradient in gravistimulated eir1-4 PIN2p::PIN2:VEN. Yellow arrowheads show the direction of the gravity vector. (G) DR5::mRFP signal intensities determined in lateral root cap cells at the upper (area indicated by blue rectangles in E and F) and the lower (area indicated by yellow rectangles in E and F) sides of horizontally positioned roots after 90 min of gravistimulation. Roots exhibiting at least twofold higher signal intensities at the lower versus the upper side were scored positive for establishment of a DR5 expression gradient. Total number of DR5::mRFP eir1-4 PIN2p::PIN2:VEN and DR5::mRFP eir1-4 PIN2p::pin2^12K-R:VEN seedlings analyzed is indicated on top of bars. (H–J) Comparison of PIN2p::PIN2:VEN (H), PIN2p::pin2^12K-R:VEN (I), and PIN2p::pin2^17K-R:VEN (J) expression in eir1-4 root meristems at 6 DAG. Roots were stained with propidium iodide (red); white arrowheads indicate ectopic signals. (K and L) Expression of PIN2p::PIN2:VEN (K) and PIN2p::pin2^12K-R:VEN (L) in the root tip (yellow). White arrowheads indicate ectopic signals in columella root cap cells. (M–O) PIN2p::PIN2:VEN (M) PIN2p::pin2^12K-R:VEN (N), and PIN2p::pin2^17K-R:VEN (O) signals in eir1-4 root epidermis cells at 3–4 DAG. (P and Q) Representative eir1-4 PIN2p::PIN2:VEN (P) and eir1-4 PIN2p::pin2^17K-R:VEN (Q) seedling roots at 3 DAG (n = 30). (Scale bars: B and C, 25 μm; E, F, and K–O, 20 μm; H–J, 50 μm; P and Q, 250 μm.)

Given the activity of pin2^12K-R in BY-2 cells, we analyzed consequences of pin2^12K-R expression on auxin responses in Arabidopsis. Expression of auxin-responsive DR5::mRFP (22) was reduced in eir1-4 pin2^12K-R root meristems (Fig. S2 A and B), and root elongation was less inhibited than in controls, when germinated on low auxin concentrations, potentially reflecting alterations in auxin distribution or responses (Fig. S2C). Moreover, defects in dynamic changes in DR5::mRFP expression became apparent in gravistimulated pin2^12K-R seedlings. eir1-4 expressing WT PIN2 had a pronounced tendency to establish a DR5 expression gradient, with more intense reporter signals at the lower side of gravity-responding roots (Fig. 2 E and G). This expression pattern was suggested to reflect asymmetric auxin flow that would result in differential cell elongation in bending roots (23). In contrast, the majority of gravistimulated eir1-4 pin2^12K-R roots failed to establish a clear DR5 expression gradient (Fig. 2 F and G). This observation links diminished PIN2 ubiquitylation to altered auxin responses upon gravistimulation and supports a scenario in which PIN2 ubiquitylation is essential for correct auxin distribution.

Loss of PIN2 Ubiquitylation Interferes with Vacuolar Targeting.

Our data indicate that defects in PIN2 ubiquitylation do not interfere with auxin carrier function per se but affect auxin distribution. In an attempt to reveal the causes for these deficiencies, we analyzed PIN2 localization in Arabidopsis roots. WT PIN2p::PIN2:VEN is expressed in lateral root cap, epidermis, and cortex cells of root meristems and exhibits a polar localization that determines directionality of auxin transport (24) (Fig. 2 H, K, and M). PIN2p::pin2^12K-R:VEN and PIN2p::pin2^17K-R:VEN showed a similar reporter localization in these cell files, demonstrating targeting of both alleles to polar PM domains (Fig. 2 I, J, N, and O and Fig. S2D). However, we detected ectopic accumulation of reporter proteins, possibly reflecting deficiencies in protein trafficking and/or degradation. Unlike PIN2:VEN, pin2^12K-R:VEN and pin2^17K-R:VEN are visible in columella root cap cells (compare Fig. 2 H–L and Fig. S3 A and B). Moreover, soon after primary root emergence, eir1-4 PIN2p::pin2^17K-R:VEN roots frequently exhibited ectopic signals of variable size and shape that were no longer detectable in older seedlings, indicative of deficiencies in pin2^17K-R:VEN sorting and/or proteolytic turnover during early seedling development (compare Fig. 2 M and O–Q and Fig. S3C). In contrast, PIN2p::pin2^K-R:VEN alleles that still complemented eir1-4 exhibited an expression pattern indistinguishable from PIN2p::PIN2:VEN (Fig. S3 D–G).

We then analyzed pin2^12K-R:VEN intracellular sorting. When cells are cotreated with brefeldin A (BFA) and cycloheximide (to block de novo protein synthesis), they form “BFA compartments,” which give an estimate of PIN sorting from the PM to endosomes (13). Mutant pin2^12K-R:VEN exhibited accumulation in BFA compartments comparable to that in WT PIN2:VEN (Fig. 3 A–C), indicating that PIN2 sorting via a BFA-sensitive pathway is not substantially affected by lysine substitutions in pin2^12K-R. In addition, targeting to and lateral diffusion at the PM of pin2^12K-R:VEN seems comparable to results in WT PIN2:VEN, as indicated by fluorescence recovery after photobleaching (FRAP) experiments. No significant differences in signal recovery were observed when comparing bleached PM sections of root epidermis cells in eir1-4 PIN2p::PIN2:VEN and eir1-4 PIN2p::pin2^12K-R:VEN (Fig. 3 D and E).

Fig. 3.

Intracellular sorting and degradation of PIN2 and pin2^12K-R:VEN. (A and B) PIN2:VEN (A) and pin2^12K-R:VEN (B) localization in root epidermis cells of 4-DAG seedlings after treatment with 25 μM BFA and 50 μM cycloheximide for 90 min. (C) Relative number (%) of seedlings exhibiting formation of BFA compartments upon treatment with 25 μM BFA and 50 μM cycloheximide for 90 min. Number of seedlings analyzed is indicated on top of bars. (D) FRAP analysis of PIN2:VEN (Left) and pin2^12K-R:VEN (Right) in 4-DAG seedlings. Bleached areas are highlighted by turquoise rectangles in PM portions immediately before bleaching (p). Signal recovery after bleaching is shown at the indicated time points (0, 10, and 30 min). (E) Signal recovery (percentage of intensity before bleaching) in bleached PM sections of PIN2:VEN and pin2^12K-R:VEN. SDs are indicated as bars (n = 20). (F–K) PIN2:VEN (F, H, and J) and pin2^12K-R:VEN (G, I , and K) signals in eir1-4 root epidermis cells at 4 DAG: Light-grown seedlings (E and G), seedlings after dark incubation for 5 h (H and I; arrowheads indicate vacuolar compartment), and after treatment with 20 μM NAA for 2.5 h and shifted to the dark for 1.5 h (J and K; arrowheads indicate vacuolar compartment). A total of 20–25 seedlings was analyzed for each genotype and condition. Seedlings were treated with FM 4-64 (red) to visualize vacuolar compartments. (L) Relative signal intensity in the vacuolar compartment of eir1-4 PIN2p::PIN2:VEN and eir1-4 PIN2p::pin2^12K-R:VEN root epidermis cells expressed as percentage of PIN2 signals at the PM. Signals were determined from seedlings incubated in the light (light), shifted into the dark for 5 h (darkness), or treated with 20 μM NAA for 2.5 h (NAA). SDs are indicated as bars (n = 18). *P < 0.05, two-tailed t test, significant difference in signal intensities. (M and N) Localization of PIN2:VEN (M) and pin2^12K-R:VEN (N) in root meristems gravistimulated for 60 min. Elevated abundance of PIN2:VEN at the lower side (K), is indicated by arrowheads. (O and P) PIN2:VEN (O) and pin2^12K-R:VEN (P) root meristem epidermis cells at a root’s upper side gravistimulated for 90 min and incubated in darkness for 4 h (arrowheads indicate vacuolar signals). A total of 16–26 seedlings was analyzed for each condition and genotype. (Q) PIN2 steady-state levels in 4-DAG seedlings treated with 20 μM NAA in eir1-4 PIN2p::PIN2 (Upper) and in eir1-4 PIN2p::pin2^12K-R (Lower) root membrane protein fractions. Control (con.): time course in the absence of NAA. Tubulin (α-TUB) was used for normalization. (R) K63-linked ubiquitylation of PIN2 and pin2^12K-R after auxin treatment. Seedlings (4 DAG) were incubated with 20 μM NAA for 2 h (+) and subjected to IP with α-PIN2. As control, untreated material (−) was used. PIN2 signals were normalized (Lower) and then probed for ubK63-linked chains (α-K63-UBQ; Upper). A distinct signal corresponding to IgG is detectable in all IP samples (asterisk). Normalized signal intensities are indicated below blots. (Scale bars: A, B, F–K, O, and P, 10 μm; M and N, 50 μm.)

Alterations in pin2^12K-R:VEN protein fate were observed when testing its vacuolar targeting after dark incubation for 5 h. Under these conditions, which allow for visualization of vacuole-localized PIN2 more accurately because of stabilization of fluorescent protein-tagged reporters in the lytic compartment (25, 26), we observed diminished accumulation of pin2^12K-R:VEN compared with PIN2:VEN (Fig. 3 F–I and L-I). This result could reflect reduced pin2^12K-R vacuolar targeting/accumulation as a result of its diminished ubiquitylation.

Differential vacuolar PIN2 sorting has been described during gravitropic root bending (15, 26), and we therefore analyzed pin2^12K-R:VEN fate upon gravistimulation. WT PIN2:VEN appears endocytosed preferentially in epidermis cells at the upper side of horizontally positioned roots, coinciding with establishment of an auxin gradient in gravity-responding roots (15, 26) (Fig. 3 M and O). This asymmetric protein distribution was abolished in PIN2p::pin2^12K-R:VEN, highlighting defects in pin2^12K-R endocytosis in response to gravity (Fig. 3 N and P and Fig. S2E) and which could affect auxin signaling or distribution as indicated by altered DR5::mRFP expression in pin2^12K-R lines (Fig. 2 E–G).

Besides gravistimulation, extended auxin treatment induces PIN2 degradation (15, 27). To determine the relationship between PIN2 ubiquitylation and its proteolytic turnover, we treated seedlings with high auxin concentrations that cause a robust decrease of total PIN2 levels. After 60 min of hormone treatment, PIN2 protein levels started to decrease in membrane protein fractions, which coincided with increased vacuolar signals in auxin-treated eir1-4 PIN2p::PIN2:VEN (Fig. 3 J and Q), indicative of vacuolar targeting and degradation in response to the hormone. No comparable auxin effects were observed in pin2^12K-R, reflected in constant protein levels in eir1-4 PIN2p::pin2^12K-R and diminished vacuolar accumulation in eir1-4 PIN2p::pin2^12K-R:VEN (Fig. 3 K and Q). Analysis of IPs normalized for PIN2 revealed that auxin-induced protein degradation coincides with increased PIN2 K63-linked polyubiquitylation, whereas we were unable to detect increased ubiquitylation of stable pin2^12K-R (Fig. 3R).

Collectively, these results suggest that K63-linked polyubiquitylation acts as rate-limiting signal for endocytic sorting and vacuolar degradation of PIN2 in response to stimuli such as gravity and auxin.

Ubiquitylation Signals Endocytosis and Vacuolar Sorting of PIN2.

Our data indicate that a loss of PIN2 K63-linked ubiquitylation interferes with endocytic targeting to the lytic vacuole. To test whether PIN2 ubiquitylation could function as a signal for such sorting, we generated PIN2 alleles that mimic constitutive ubiquitylation and fused ubiquitin, with its C-terminal two glycines replaced by alanines [to prevent processing by ubiquitin proteases (28)], into the central hydrophilic loop of PIN2. PIN2p::PIN2:ubq:VEN failed to rescue eir1-4 (Fig. S4A) and exhibited only faint signals at the PM, together with dispersed intracellular signals, a distribution not found for PIN2:VEN (Fig. 4 A, B, H, and I). Moreover, when analyzing PIN2:ubq levels, we detected only weak signals in membrane protein fractions of all tested lines, which is suggestive of altered PIN2 stability when fused to ubiquitin (Fig. 4C and Fig. S4B).

Fig. 4.

Analysis of PIN2:ubq alleles. (A and B) Expression of PIN2p::PIN2:VEN (A) and PIN2p::PIN2:ubq:VEN (B) in eir1-4 root meristems at 4 DAG. (C) Membrane (mem.) and soluble (sol.) protein extracts from eir1-4 PIN2p::PIN2 (PIN2) and eir1-4 PIN2p::PIN2:ubq (PIN2:ubq) probed with anti-PIN2. Filled arrowhead, PIN2; open arrowhead, PIN2:ubq. Tubulin (α-TUB) was used for normalization. (D and E) Treatment of PIN2p::PIN2:VEN (D) and PIN2p::PIN2:ubq:VEN (E) with 25 μM BFA for 90 min. White arrowheads indicate intracellular aggregates of reporter signals. Approximately 20 seedlings were analyzed for each genotype. (F and G) FM 4-64 staining (30 min, red signals) of eir1-4 PIN2p::PIN2:VEN (F) and eir1-4 PIN2p::PIN2:ubq:VEN (G) at 4 DAG (PIN2 reporter signals are pseudocolored in green). Arrowheads highlight colocalization of dye and PIN2 signals in endocytic compartments. A total of 15 seedlings was analyzed for each genotype; representative images are shown. (H–L) Comparison of reporter signals in root meristem epidermis cells of eir1-4 PIN2p::PIN2:ubq:VEN (H), eir1-4 PIN2p::PIN2:ubq:VEN (I), eir1-4 PIN2p::PIN2:ubq^I44A:VEN (J), eir1-4 PIN2p::pin2^17K-R:VEN (K), and eir1-4 PIN2p::pin2^17K-R:ubq:VEN (L) at 6 DAG. Identical laser settings were used for comparison of abundance/localization by confocal laser scanning microscopy. Representative images from 20 seedlings analyzed for each genotype are shown. (M–P) Reporter signals in root meristem epidermis cells of 6-DAG eir1-4 PIN2p::PIN2:VEN (M), eir1-4 PIN2p::PIN2:ubq:VEN (N), eir1-4 PIN2p::pin2^17K-R:VEN (O), and eir1-4 PIN2p::pin2^17K-R:ubq:VEN (P) incubated in the dark for 5 h and treated with FM 4-64 (4 μM) for 60 min. (Left) VENUS signals (green). (Right) Merged images (FM 4-64 in red). White arrowheads indicate vacuolar compartments. Representative images from 15–20 seedlings analyzed for each genotype are shown. (Q and R) Relative signal intensities in cytoplasm and vacuoles of eir1-4 PIN2p::PIN2:ubq:VEN (Q) and eir1-4 PIN2p::pin2^17K-R:ubq:VEN (R) root meristem epidermis cells incubated in the dark for 5 h. SDs are indicated as bars (n = 9). (Scale bars: A and B, 50 μm; D–G and M –P, 10 μm; H–L, 20 μm.)

Next, we treated seedlings with BFA and FM 4-64 and found PIN2:ubq:VEN signals in BFA compartments and colocalization with FM 4-64–labeled endosomes, respectively (Fig. 4 D–G and Fig. S5A). We then asked whether altered localization and abundance could result from deregulated endocytosis of PIN2:ubq, and so we treated with proteasome inhibitor MG132, which was demonstrated to interfere with PIN2 internalization and vacuolar targeting (15, 25). The experiments revealed an increase of PIN2:ubq:VEN signals predominantly at the PM (Fig. S5 B–E) and elevated PIN2:ubq steady-state levels upon MG132 treatment (Fig. S5F). This finding indicates that, although PIN2:ubq is still targeted to the PM, constitutive ubiquitylation appears to accelerate its endocytosis and turnover, which is antagonized by proteasome inhibition.

To address mechanistic aspects of PIN2:ubq endocytic sorting, we mutagenized K48 and/or K63 in the ubiquitin moiety of PIN2:ubq. PIN2p::PIN2:ubq^K48R:VEN, PIN2p::PIN2:ubq^K63R:VEN, and PIN2p::PIN2:ubq^{K48, 63R}:VEN all gave signals similar to PIN2:ubq:VEN, suggesting that neither ubK48- nor ubK63-linked chain formation at the ubiquitin moiety is required for PIN2:ubq internalization (Fig. S6 A–E). In contrast, mutagenesis of ubiquitin-isoleucine 44, which is essential for cargo recognition by adaptor proteins via their ubiquitin-binding domains (29, 30), efficiently interfered with PIN2 internalization. Unlike PIN2p::PIN2:ubq:VEN, PIN2p::PIN2:ubq^I44A:VEN exhibited predominantly polar signals at the PM and complemented eir1-4 (compare Fig. 4 I and J, Fig. S4A, and Fig. S6F), supporting models in which PIN2:ubq internalization involves recognition of its ubiquitin moiety by the plant’s endocytic machinery.

We then used the strong polyubiquitylation-deficient pin2^17K-R allele, with most of its loop-resident lysines mutagenized (Fig. 1D and Table S1), and introduced ubiquitin at the same position as in PIN2:ubq, resulting in PIN2p::pin2^17K-R:ubq:VEN. We found that PIN2p::pin2^17K-R:ubq:VEN exhibits diffuse signals within epidermis cells, which is similar to PIN2:ubq:VEN distribution and indicative of pin2^17K-R internalization as a consequence of its fusion to ubiquitin (compare Fig. 4 K and L). However, unlike PIN2p::PIN2:ubq:VEN, PIN2p::pin2^17K-R:ubq:VEN failed to accumulate strong vacuolar reporter signals upon extended dark incubation (Fig. 4 M–R). Thus, although pin2^17K-R:ubq:VEN appears to be constitutively endocytosed, as are PIN2:ubq alleles mimicking constitutive ubiquitylation, it is no longer efficiently sorted into the vacuolar compartment, hence resembling pin2 alleles deficient in K63-linked polyubiquitylation, such as PIN2p::pin2^17K-R:VEN (Fig. 4P). This finding strongly suggests a requirement of K63-linked polyubiquitylation for efficient vacuolar targeting of PIN2.

Conclusions

Ubiquitylation of PM proteins has been demonstrated in higher plants (6–9, 15), and, analogous to the situation in animals and fungi (30–32), covalent attachment of a single ubiquitin seems a sufficient signal for endocytosis and vacuolar targeting of selected PM proteins (7, 8). This modification has been implicated in modulating abundance of nutrient carrier and light receptor proteins at the PM, thereby controlling intracellular solute homeostasis as well as stimulus-mediated growth responses (6–8). Furthermore, diubiquitylation and proteasome-dependent degradation of polyubiquitylated protein were presented as potential determinants of PM protein stability (6, 7, 15), suggesting overall an involvement of distinct ubiquitylation modes in the regulation of endocytic cargo sorting and turnover in plants.

In baker’s yeast and mammalian cells, K63-linked cargo polyubiquitylation was demonstrated to mediate efficient PM protein targeting into the lytic pathway (16–18). Vacuolar targeting of yeast Gap1, for example, strictly depends on K63-linked polyubiquitylation, whereas its monoubiquitylation appears to signal internalization from the PM (18). Our findings provide evidence for a similar pathway in plants with PIN2 alleles, mimicking constitutive ubiquitylation, being endocytosed, whereas vacuolar targeting coincides with formation of K63-linked polyubiquitin chains. In plants, mono- or multimonoubiquitylation is suggested to primarily trigger internalization from the PM, possibly in conjunction with clathrin-mediated endocytosis (6). K63-linked polyubiquitylation could designate PM proteins for an alternate sorting pathway, mediating responses that require fast and/or locally restricted adjustments in protein turnover. As for PIN2, this process appears essential for the control of differential PIN2-mediated auxin transport upon gravitropic root bending (15, 26).

Dynamic variations in auxin availability within plant tissues depend on alterations in localization and amounts of PIN proteins (11, 12). This process involves constitutive protein cycling and transcytosis that affect auxin flow by altering distribution of PM-localized and intracellular PIN protein pools (13, 14), whereas PIN ubiquitylation could act as a rate-limiting determinant for spatiotemporal control of PIN degradation. A number of recent reports connected light and growth regulator signaling to vacuolar sorting of PIN proteins (25, 33, 34), substantiating a scenario in which multiple pathways impinge on PIN protein degradation to orchestrate auxin distribution in a multifaceted developmental context. Variations in PIN polyubiquitylation might function as a common transmitter of such a diverse range of signals, but versatility and molecular constituents of this pathway remain to be determined (35, 36).

Materials and Methods

Plant Material and Constructs.

eir1-4 (SALK_091142), DR5::mRFP, and BY-2 tobacco cell lines have been described previously (15, 21, 22). PIN2p::PIN2:VEN was obtained by replacing GFP from PIN2p::PIN2:GFP (15) with VENUS (37). For PIN2p::PIN2, the tag was omitted. Further details on constructs are provided in SI Materials and Methods. Oligonucleotides used for PIN2 mutagenesis are listed in Table S2.

Membrane Protein Extraction, IP, and Western Blot Analysis.

For Western blotting and IP, 5–20 mg or 150–200 mg, respectively, of root material [3–6 d after germination (DAG)] was homogenized and processed as described in ref. 38. Drug and growth regulator treatments were performed as indicated. Further details are provided in SI Materials and Methods.

Microscopy.

Plant samples were mounted in tap water and viewed on Leica TCS SP2 and SP5 confocal laser scanning microscopes. FM 4-64 (4 μM) treatment for 30–60 min in tap water was performed before microscopy to visualize endocytic compartments. Propidium iodide (1 mg/mL) staining was used to visualize cell walls. For imaging, the 488-nm laser line was used. Laser intensity settings were kept constant in individual sets of experiments to allow for a comparison of expression and localization of reporter proteins. BY-2 cells were observed at 2 d after the induction of gene expression with 5 μM dexamethasone (added at the beginning of the subculture interval) with a Zeiss LSM 5 DUO confocal microscope. Images were cropped and equally processed for brightness and contrast with Adobe Photoshop.