The Arabidopsis receptor kinase STRUBBELIG undergoes clathrin-dependent endocytosis.

Signaling mediated by cell surface receptor kinases is central to the coordination of growth patterns during organogenesis. Receptor kinase signaling is in part controlled through endocytosis and subcellular distribution of the respective receptor kinase. For the majority of plant cell surface receptors, the underlying trafficking mechanisms are not characterized. In Arabidopsis, tissue morphogenesis requires the atypical receptor kinase STRUBBELIG (SUB). Here, we studied the endocytic mechanism of SUB. Our data revealed that a functional SUB-enhanced green fluorescent protein (EGFP) fusion is ubiquitinated in vivo. We further showed that plasma membrane-bound SUB:EGFP becomes internalized in a clathrin-dependent fashion. We also found that SUB:EGFP associates with the trans-Golgi network and accumulates in multivesicular bodies and the vacuole. Co-immunoprecipitation experiments revealed that SUB:EGFP and clathrin are present within the same protein complex. Our genetic analysis showed that SUB and CLATHRIN HEAVY CHAIN (CHC) 2 regulate root hair patterning. By contrast, genetic reduction of CHC activity ameliorates the floral defects of sub mutants. Taken together, the data indicate that SUB undergoes clathrin-mediated endocytosis, that this process does not rely on stimulation of SUB signaling by an exogenous agent, and that SUB genetically interacts with clathrin-dependent pathways in a tissue-specific manner.

Introduction

Intercellular communication is a central requirement for tissue morphogenesis as cells have to coordinate their relative behaviors to allow proper organ development. Cell surface receptor kinases play crucial roles in this process. Control of tissue morphogenesis in Arabidopsis involves the leucine-rich repeat receptor kinase (RK) STRUBBELIG (SUB). SUB, also known as SCRAMBLED (SCM), controls several developmental processes, including floral morphogenesis, integument outgrowth, leaf development, and root hair patterning (Chevalier ; Kwak ; Lin ). SUB represents an atypical receptor kinase as enzymatic activity of its kinase domain is not required for its function in vivo (Chevalier ; Vaddepalli ). SUB is glycosylated in the endoplasmic reticulum (ER) (Hüttner ), subject to ER-associated protein degradation (Vaddepalli ; Hüttner ), and found at the plasma membrane (PM) (Yadav ; Vaddepalli ). SUB not only localizes to the PM but is also present at plasmodesmata (PD), channels interconnecting most plant cells (Otero ; Sager and Lee, 2018), where it physically interacts with the PD-specific protein QUIRKY (QKY) (Vaddepalli ). In line with described models of RK-mediated control of PD-based intercellular communication, SUB and QKY function in a non-cell-autonomous manner (Yadav ; Vaddepalli ) indicating that SUB signaling involves a yet unknown factor that moves between cells. More recently, a genetic link between SUB signaling and cell wall biology has been proposed. The cell wall-localized β-1,3-glucanase ZERZAUST (ZET) participates in SUB signal transduction, and sub, qky, and zet mutants share overlapping defects in cell wall biochemistry (Fulton ; Vaddepalli ).

Maintenance of the PM composition is in part achieved through exocytosis/secretion and endocytosis (Paez Valencia ; Reynolds ). In general, plant cells internalize PM-bound material or cargo via membrane transport into the trans-Golgi network (TGN), an organelle that also functions as an early endosome (EE) and that serves as a sorting hub for subsequent trafficking pathways. The cargo may get recycled back to the PM via secretory vesicles. Cargo may also be targeted for degradation via endosomal transport to multivesicular bodies (MVBs), also known as late endosomes (LEs), containing intra-luminal vesicles. MVBs eventually fuse with the tonoplast, discharging their content into the vacuolar lumen where degradation takes place.

Internalization of PM proteins is mediated by clathrin-dependent and clathrin-independent endocytosis (Geldner and Robatzek, 2008; Robinson ; Irani and Russinova, 2009; Paez Valencia ; Reynolds ). Clathrin-mediated endocytosis (CME) is a central mechanism for the internalization of PM-localized material or cargo (Dhonukshe ; Paez Valencia ; Reynolds ). CME involves the budding of cargo-containing clathrin-coated vesicles (CCVs) from the PM. CCVs consist of vesicles surrounded by a polyhedral lattice of clathrin triskelia comprising three clathrin heavy chains (CHCs), each bound by a clathrin light chain (CLC) (Fotin ). In Arabidopsis, three genes encode CLC chains while the likely redundantly acting CHC1 and CHC2 encode CHCs (Scheele and Holstein, 2002). Clathrin is also present at the TGN/EE, at a subpopulation of MVB/LEs, and at the cell plate indicating that it functions in multiple vesicular trafficking steps as well as cytokinesis in the plant cell (Samuels ; Staehelin and Moore, 1995; Konopka ; Fujimoto ; Stierhof and El Kasmi, 2010; Kang ; Van Damme ; Ito ).

Fine-tuning the spatio-temporal dynamics of receptor-mediated endocytosis and endosomal trafficking is a central element in the regulation of RK-dependent signal transduction. Such a mechanism can, for example, maintain the steady-state level of RKs at the PM through recycling internalized RKs back to the PM, promote signaling by activated RK complexes localized on endosomes, or attenuate RK signaling by controlled removal of activated receptors from the PM followed by sorting into MVBs and finally degradation in the vacuole (Geldner and Robatzek, 2008; Irani and Russinova, 2009; Di Rubbo and Russinova, 2012; Bakker ; Critchley ).

Following RK internalization and subsequent trafficking upon RK stimulation with exogenous application of ligand has been instrumental in analysing the endocytic pathways of several plant RKs, including BRASSINOSTEROID INSENSITIVE 1 (BRI1) (Russinova ; Geldner ; Irani ; Di Rubbo ), FLAGELLIN SENSING 2 (FLS2) (Robatzek ; Beck ; Du ; Mbengue ), and PEP1 RECEPTOR 1 (PEPR1) (Ortiz-Morea ). A ligand for SUB has yet to be described, rendering such an experimental approach not possible. However, some RKs undergo endocytosis independently of exogenous application of ligand, including BRI1 (Russinova ; Geldner ; Jaillais ), SOMATIC EMBRYOGENESIS RECEPTOR-LIKE KINASE 1 (SERK1) (Shah , 2002), BRI1-ASSOCIATED RECEPTOR KINASE 1 (BAK1)/SERK3 (Russinova ), and ARABIDOPSIS CRINKLY4 (ACR4) (Gifford ).

SUB can be found in internal compartments as well (Kwak and Schiefelbein, 2008; Yadav ; Vaddepalli ; Wang ) and it was recently shown that ovules of plants homozygous for a hypomorphic allele of HAPLESS13 (HAP13) preferentially accumulate signal from a functional SUB–enhanced green fluorescent protein (EGFP) reporter in the cytoplasm, rather than the PM (Wang ). HAP13/AP1M2 encodes the µ1 subunit of the adaptor protein (AP) complex AP1 that is present at the TGN/EE network and is involved in post-Golgi vesicular trafficking to the PM, vacuole, and cell-division plane (Park ; Teh ; Wang ). Interestingly, ovules of plants with reduced HAP13/AP1M2 activity show sub-like integuments (Wang ). These results indicate that the AP1 complex is involved in subcellular distribution of SUB in a functionally relevant manner.

Here, we have further assessed the internalization and subsequent endocytic trafficking behavior of SUB. We show that the intracellular domain of a functional SUB:EGFP fusion protein becomes ubiquitinated. Upon endocytosis SUB:EGFP is sorted to MVBs and the vacuole. Our data further indicate that CME contributes to internalization of SUB:EGFP. Finally, we provide genetic data suggesting that CHC is part of the SUB-dependent signaling mechanism that mediates root hair patterning while conversely a reduction in CHC activity alleviates the floral defects of sub mutants.

Materials and methods

Plant work, plant genetics and plant transformation

Arabidopsis (L.) Heynh. var. Columbia (Col-0) and var. Landsberg (erecta mutant) (Ler) were used as wild-type strains. Plants were grown as described earlier (Fulton ). The sub-1 (Ler) was described previously (Chevalier ). The sub-9 mutant (Col), carrying a T-DNA insertion (SAIL_1158_D09), was described in (Vaddepalli ). The chc1-1 (SALK_112213), chc1-2 (SALK_103252), chc2-1 (SALK_028826), and chc2-2 (SALK_042321) alleles (all Col) (Alonso ) were described in (Kitakura ). Wild-type and mutant plants were transformed with different constructs using Agrobacterium strain GV3101/pMP90 (Koncz and Schell, 1986) and the floral dip method (Clough and Bent, 1998). Transgenic T1 plants were selected on kanamycin (50 μg ml−1), hygromycin (20 μg ml−1) or glufosinate (Basta) (10 μg ml−1) plates and transferred to soil for further inspection. The hydroxytamoxifen-inducible line INTAM>>RFP-HUB/Col line (HUB) was described previously (Robert ; Kitakura ). Seedlings were grown on half-strength Murashige and Skook (1/2 MS) agar plates (Murashige and Skoog, 1962).

Recombinant DNA work

For DNA and RNA work, standard molecular biology techniques were used. PCR fragments used for cloning were obtained using Q5 high-fidelity DNA polymerase (New England Biolabs, Frankfurt, Germany). All PCR-based constructs were sequenced. Primer sequences used in this work are listed in Supplementary Table S1 at JXB online.

Reporter constructs

The pCAMBIA2300-based pSUB::SUB:EGFP construct was described previously (Vaddepalli ). To obtain pUBQ10::SUB:EGFP, a 2 kb promoter fragment of UBQ10 (At4g05320) was amplified from Ler genomic DNA using primers pUBQ(KpnI)_F and pUBQ(AscI)_R. The resulting PCR product was digested using KpnI/AscI and used to replace the pSUB fragment in pSUB::SUB:EGFP. The pGL2::GUS:EGFP construct was assembled using the GreenGate system (Lampropoulos ). The promoter region of GL2 (AT1G79840) was amplified with primer pGL2_F1 and pGL2_R1 from genomic Col-0 DNA. The internal BsaI site was removed during the procedure as described in Lampropoulos . The β-glucuronidase (GUS) coding sequence was amplified from plasmid pBI121 (Jefferson ) with primer GUS_F and GUS_R, digested with BsaI, and used for further cloning.

Genotyping PCR

PCR-based genotyping was performed with the following primer combinations: sub-9, SUB_LP158, SUB_RP158, and SAIL_LB2; chc2 salk-042321, CHC2-LP321, CHC2-RP321, and SALK_LBb1.3; chc2 salk-028826, CHC2_LP826, CHC2_RP826, and SALK_LBb1.3; chc1 salk-112213, CHC1_LP213, CHC1_RP213, and SALK_LBb1.3; chc1 salk-103252, CHC1_LP252, CHC1_RP252, and SALK_LBb1.3.

Chemical treatments

Brefeldin A (BFA), cycloheximide, tyrphostin A23 (TyrA23), wortmannin, and concanamycin A (ConcA) were obtained from Sigma-Aldrich and used from stock solutions in DMSO (50 mM BFA, cycloheximide, TyrA23; 30 mM wortmannin, 2 mM ConcA). FM4-64 was purchased from Molecular Probes (2 mM stock solution in water). Five-day-old seedlings were incubated for the indicated times in liquid 1/2 MS medium containing 50 µM BFA, 50 µM cycloheximide, 75 µM TyrA23, 33 µM wortmannin, and 2 µM ConcA. For FM4-64 staining seedlings were incubated in 4 µM FM4-64 in liquid 1/2 MS medium for 5 min prior to imaging. 4-Hydroxytamoxifen was obtained from Sigma-Aldrich (10 mM stock solution in ethanol). Seedlings were grown for 3 d on 1/2 MS plates, transferred onto 1/2 MS plates containing 2 µM 4-hydroxytamoxifen (or ethanol as mock treatment) for 4 d and then imaged using confocal microscopy.

Immunoprecipitation and western blot analysis

Five hundred milligrams of 7-day-old wild-type or transgenic seedlings were lysed using a TissueLyser II (Qiagen) and homogenized in 1 ml lysis buffer A [50 mM Tris–HCl pH 7.5, 100 mM NaCl, 0.1 mM phenylmethylsulfonyl fluoride (PMSF), 0.5% Triton X-100, protease inhibitor mixture (Roche)]. Cell lysate was mildly agitated for 15 min on ice and centrifuged for 15 min at 13 000 g. For lines carrying green fluorescent protein (GFP)-tagged proteins, supernatant was incubated with GFP–Trap magnetic agarose (MA) beads (ChromoTek) for 2 h at 4 °C. Beads were concentrated using a magnetic separation rack. Samples were washed four times in buffer B [50 mM Tris–HCl pH 7.5, 100 mM NaCl, 0.1 mM PMSF, 0.2% Triton X-100, protease inhibitor mixture (Roche)]. Bound proteins were eluted from beads by heating the samples in 30 µl 2× Laemmli buffer for 5 min. Samples were separated by SDS-PAGE and analysed by immunoblotting according to standard protocols. Primary antibodies included mouse monoclonal anti-GFP antibody 3E6 (Invitrogen/Thermo Fisher Scientific), mouse monoclonal anti-ubiquitin antibody P4D1 (Santa Cruz Biotechnology), and polyclonal anti-CHC antibody AS10 690-ALP (Agrisera). Secondary antibodies were obtained from Pierce/Thermo Fisher Scientific: goat anti-rabbit IgG antibody (1858415) and goat anti-mouse IgG antibody (1858413).

Microscopy and art work

To assess the cellular structure of floral meristems, samples were stained with modified pseudo-Schiff propidium iodide (Truernit ). Confocal laser scanning microscopy was performed with an Olympus FV1000 set-up using an inverted IX81 stand and FluoView software (FV10-ASW version 01.04.00.09) (Olympus Europa GmbH, Hamburg, Germany) equipped with a water-corrected ×40 objective (NA 0.9) at ×3 digital zoom. For SUB:EGFP subcellular localization upon drug treatments or colocalization with endosomal markers, confocal laser scanning microscopy was performed on epidermal cells of root meristems located about 8–12 cells above the quiescent center using a Leica TCS SP8 X microscope equipped with GaAsP (HyD) detectors. The following objectives were used: a water-corrected ×63 objective (NA 1.2), a ×40 objective (NA 1.1), and a ×20 immersion objective (NA 0.75). Scan speed was set at 400 Hz, line average at between 2 and 4, and the digital zoom at 4.5 (colocalization with FM4-64), 3 (drug treatments), or 1 (root hair patterning). EGFP fluorescence excitation was performed at 488 nm using a multi-line argon laser (3% intensity) and detected at 502 to 536 nm. FM4-64 fluorescence was excited using a 561 nm laser (1% intensity) and detected at 610–672 nm. For the direct comparisons of fluorescence intensities, laser, pinhole, and gain settings of the confocal microscope were kept identical when capturing the images from the seedlings of different treatments. The intensities of fluorescence signals at the PM were quantified using Leica LAS X software (version 3.3.0.16799). For the measurement of the fluorescence levels at the PM, optimal optical sections of root cells were used for measurements. On the captured images the fluorescent circumference of an individual cell (region of interest; ROI) was selected with the polygon tool. The mean pixel intensity readings for the selected ROIs were recorded and the average values were calculated. For determination of colocalization, the distance from the center of each EGFP spot to the center of the nearest FM4-64, monomeric Kusabira Orange (mKO) or monomeric red fluorescent protein (mRFP) signal was measured by hand on single optical sections using ImageJ/Fiji software (Schindelin ). If the distance between two puncta was below the resolution limit of the objectives lens (0.24 µm) the signals were considered to colocalize (Ito ). Arabidopsis seedlings were covered with a 22×22 mm glass coverslip of 0.17 mm thickness (no. 1.5H, Paul Marienfeld GmbH & Co. KG, Lauda-Königshofen, Germany). Scanning electron microscopy was performed essentially as reported previously (Schneitz ; Sieburth and Meyerowitz, 1997). Images were adjusted for color and contrast using ImageJ/Fiji software. Statistical analysis was done in Prism 8 (GraphPad Software, San Diego, CA, USA).

Results

The endocytic route of SUB:EGFP

To investigate the endocytic pathway followed by SUB we made use of a previously well-characterized line carrying the sub-1 null allele and a transgene encoding a SUB:EGFP translational fusion driven by its endogenous promoter (pSUB::SUB:EGFP). The line exhibits a wild-type phenotype demonstrating the presence of a functional reporter (Vaddepalli ; Vaddepalli ). We studied the subcellular distribution of the pSUB::SUB:EGFP reporter signal in epidermal cells of the root meristem using confocal laser scanning microscopy. These cells serve as an ideal model as SUB controls the early patterning of root hairs, cells that are generated by the epidermis (Dolan ). In the absence of any obvious exogeneous stimulation of SUB signaling we observed SUB:EGFP signal at the PM and in cytoplasmic foci (Fig. 1A). Moreover, we noticed that the SUB:EGFP signal labelled structures resembling vesicles as well as the vacuole. These observations raise the possibility that SUB:EGFP undergoes internalization from the PM and is shuttled to the vacuole for degradation.

Fig. 1.

Sub-cellular localization of SUB:EGFP. Fluorescence micrographs show optical sections of epidermal cells of root meristems of 5- to 6-day-old seedlings. (A) Partial colocalization of SUB:EGFP and FM4-64 foci upon treating cells with FM4-64 for 5 min. (B) Partial colocalization of SUB:EGFP and mRFP:SYP43 puncta. (C) Partial colocalization of SUB:EGFP and mKO:CLC signals. (D) Box-and-whiskers plot depicting the results of a quantitative colocalization analysis of SUB:EGFP-positive foci and reporter signals shown in (A–C). A total of 344 (FM4-64), 278 (mRFP:SYP43), and 365 (mKO:CLC) puncta were analysed. Data points indicate percentage of colocalization per analysed cell; 8–14 SUB:EGFP foci were analysed per cell. FM4-64: n=30 cells across four roots; mRFP:SYP43 n=23/3; mKO:CLC n=31/4. Scale bars: 5 μm. The experiments were independently repeated twice with similar results.

To assess the early process of SUB:EGFP endocytosis we imaged cells upon a 5-min treatment with the endocytic tracer dye FM4-64 (Fig. 1A, D). Using a previously described criterion for colocalization (Ito ), the internal SUB:EGFP and FM4-64 signals were considered colocalized when the distance between the centers of the two types of signals was below the limit of resolution of the objective, in our case 0.24 µm. We observed that 70% (n=344) of all cytoplasmic SUB:EGFP foci were also marked by FM4-64 supporting endocytosis of SUB:EGFP.

To explore if endosomal trafficking of SUB:EGFP involves the TGN/EE we investigated colocalization of SUB:EGFP with the TGN marker mRFP:SYP43 (Fig. 1B, D; Ebine ; Ito ; Uemura ). We observed a frequency of 44% colocalization (n=278) between internal SUB:EGFP and mRFP:SYP43 puncta. To further assess colocalization of SUB:EGFP with the TGN we made use of a previously characterized translational fusion between CLC2 and mKO under the control of the cauliflower mosaic virus 35S promoter (mKO:CLC) (Fujimoto ). CLC2 fused to fluorescent tags also localizes to the TGN in live cell imaging experiments (Ito ). We observed a frequency of 33% colocalization (n=365) between internal SUB:EGFP and mKO:CLC puncta (Fig. 1C, D).

To further investigate internalization of SUB:EGFP, we treated 5-day-old seedlings with wortmannin. Wortmannin is a phosphatidylphosphate-3-kinase inhibitor that amongst other things interferes with vesicle formation from the PM (Tse ; Wang ; Ito ; Cui ). We analysed the number of internal SUB:EGFP-labelled puncta in cells upon treatment with 33 µM wortmannin for 60 min in the light (Fig. 2A). We found a substantial reduction in the number of such puncta in drug-treated cells when compared with mock-treated cells (Fig. 2A, B). Moreover, we noted a significant increase in SUB:EGFP signal intensity at the PM in wortmannin-treated cells (Fig. 2A, C).

Fig. 2.

Subcellular localization of SUB:EGFP upon drug treatments. Fluorescence micrographs show optical sections of epidermal cells of root meristems of 5- to 6-day-old seedlings. (A) Subcellular localization of SUB:EGFP signal in the presence of wortmannin (Wm) and DMSO (mock control). (B) Box-and-whiskers plot depicting the number of SUB:EGFP-positive endosomes per cell after incubation (n=30 cells across six roots). Asterisks represent statistical significance (P<0.0001, Student’s t test). (C) Box-and-whiskers plot of the quantification of the EGFP fluorescence intensity at plasma membrane after incubation (n=50–60 cells across six roots). Asterisks represent statistical significance (P<0.0001, Student’s t test). (D) SUB:EGFP signal is detected in BFA bodies upon BFA treatment. (E) SUB:EGFP signal is detected in BFA compartments in the presence of cycloheximide (CHX). (F) Box-and-whiskers plot of the quantification of SUB:EGFP fluorescence intensity of PM in (E). Graph represents quantification of the EGFP fluorescence intensity at plasma membrane after incubation (n=50 cells across six (BFA) or seven (BFA+CHX) roots). Asterisks represent statistical significance (P<0.0001, Student’s t test). (G) SUB:EGFP- or FM4-64-derived signal in meristematic root epidermal cells of 5-day-old seedlings treated with FM4-64 for 10 min followed by an incubation in DMSO (mock) or wortmannin for 120 min in the dark. Note the SUB:EGFP- and FM4-64-derived signals on the vacuoles in DMSO-treated cells and on the ring-like structures in wortmannin-treated cells. (H) Typical result of a similar experiment to that in (G) but using the MVB marker ARA7:YFP (wave line 2; W2Y). Note the intracellular ring-like structures labelled by ARA7:YFP and FM4-64 upon wortmannin treatment. (I) Colocalization of SUB:EGFP and ARA7:mRFP on wortmannin-induced ring-like structures. We scored 161 SUB:EGFP-labelled ring-like structures (one to four ring-like SUB:EGFP structures per cell, 8–17 cells per root, eight roots total); 159 ring-like structures also exhibited an ARA7:mRFP signal. (J) SUB:EGFP signal is observed in lytic vacuoles after ConcA treatment. Scale bars: 5 μm. The experiments were independently repeated twice with similar results.

To corroborate the presence of SUB:EGFP at the TGN/EE we exposed sub-1 pSUB::SUB:EGFP seedlings to the fungal toxin BFA. Treatment with BFA results in the formation of so-called BFA compartments or bodies that contain secretory and endocytic vesicles (Robinson ; Paez Valencia ). We observed prominent SUB:EGFP signal in BFA compartments in root epidermal cells of seedlings treated with DMSO for 30 min followed by a DMSO/BFA (50 µM) treatment for 60 min, confirming previous data (Fig. 2D) (Kwak and Schiefelbein, 2008; Yadav ; Vaddepalli ; Wang ).

Next, we explored the relative contribution of signal at the TGN/EE originating from the secretion of newly translated SUB:EGFP versus endocytic SUB:EGFP-derived signal. To this end we first treated seedlings with the protein synthesis inhibitor cycloheximide (50 µM) for 60 min followed by co-incubation with 50 µM BFA for 30 min. In those seedlings SUB:EGFP still prominently localized to BFA bodies (Fig. 2E) as noted earlier (Wang ). However, we also observed a reduction in SUB:EGFP signal intensity at the PM in cells co-treated with cycloheximide and BFA (Fig. 2E, F) indicating that the BFA bodies included both endocytic and exocytic SUB:EGFP-containing vesicles.

We then investigated if internalized SUB:EGFP is sorted into MVBs. Apart from affecting vesicle formation at the PM, wortmannin also interferes with the maturation of LEs and causes formation of enlarged MVB/LEs (Tse ; Wang ; Cui ). Treating seedlings for 60 min with 33 µM wortmannin results in the formation of large globular structures labelled by SUB:EGFP signal (Fig. 2A). Such structures are typical for enlarged MVBs (Jia ). To corroborate our findings, we incubated 5-day-old seedlings with 4 µM FM4-64 for 10 min (pulse) followed by an incubation in a mock or 33 µM wortmannin solution lacking FM4-64 for 120 min in the dark (chase). In the mock treated seedlings we observed an accumulation of FM4-64- and SUB:EGFP-derived signal in the vacuole of root epidermis cells (Fig. 2G). In contrast, wortmannin-treated seedlings showed FM4-64 and SUB:EGFP signal could be seen in ring-like structures typical of enlarged MVBs (Fig. 2G). We obtained a similar result when analysing wave line 2 (W2Y) expressing a fusion of ARA7/RABF2b to yellow fluorescent protein (YFP) (Fig. 2H; Geldner ). The Rab small GTPase ARA7/RABF2b has been previously shown to localize to MVBs (Kotzer ; Lee ; Haas ). To further substantiate these results we generated a line hemizygous for the pSUB::SUB:EGFP and pUBQ10::ARA7:mRFP transgenes. Upon treating 5-day-old seedlings of this line with wortmannin for 120 min we observed 99% colocalization of SUB:EGFP- and ARA7:mRFP-labelled ring-like structures in root epidermal cells (Fig. 2I) (159/161). Finally, and in accordance with these results, SUB:EGFP was detected at MVBs in immunogold electron microscopy experiments (see Fig. 2W in Vaddepalli ). Taken together, the results indicate that SUB:EGFP is sorted into MVBs.

ConcA inhibits vacuolar ATPase activity at the TGN/EE and in the tonoplast thereby interfering with the trafficking of newly synthesized materials to the PM, the transport of cargo from the TGN/EE to the vacuole, and the vacuolar degradation of cargo (Dettmer ; Robinson ; Viotti ; Scheuring ). Upon treatment with 2 µM ConcA for 1 h, seedlings showed large roundish structures labelled by a diffuse SUB:EGFP signal (Fig. 2J) indicating that SUB:EGFP was not degraded efficiently and thus accumulated in the vacuole.

In summary, the results are consistent with the notion that the endocytic route of SUB:EGFP involves the TGN/EE, the MVB/LEs, and the vacuole where it becomes degraded. A noticeable portion of SUB:EGFP puncta colocalizes with the TGN/EE supporting passage of SUB:EGFP through the TGN/EE. However, we cannot exclude that a fraction of SUB:EGFP also traffics via a TGN/EE-independent route, as does, for example, the AtPep1–PEPR1 signaling complex (Ortiz-Morea ).

SUB:EGFP is ubiquitinated in vivo

Ubiquitination plays an important role in endocytosis and endosomal sorting of PM proteins (MacGurn ; Paez Valencia ; Isono and Kalinowska, 2017), such as the brassinosteroid receptor BRI1 (Martins ) or the auxin efflux facilitator PINFORMED 2 (PIN2) (Leitner ). To test if SUB:EGFP is ubiquitinated in vivo, we made use of our sub-1 pSUB::SUB:EGFP reporter line as well as a previously described line carrying the SUB:EGFP translation fusion driven by the UBIQUITIN10 (UBQ) promoter (pUBQ::SUB:EGFP) (Vaddepalli ). We immunoprecipitated SUB:EGFP from 7-day-old, plate-grown seedlings using an anti-GFP antibody. Immunoprecipitates were subsequently probed with the commonly used P4D1 anti-ubiquitin antibody recognizing mono- and polyubiquitinated proteins. P4D1-dependent signal could not be reproducibly detected when testing immunoprecipitates from lines expressing the pSUB::SUB:EGFP reporter due to low abundance of SUB:EGFP in the immunoprecipitate. By contrast, we clearly observed a high-molecular mass smear in immunoprecipitates obtained from pUBQ::SUB:EGFP lines (Fig. 3). This smear is typical for ubiquitinated proteins. We did not detect signals in immunoprecipitates obtained from wild-type seedlings. These results indicate that a fraction of SUB proteins becomes ubiquitinated.

Fig. 3.

In vivo ubiquitination of SUB. Western blot analysis of immunoprecipitates obtained from wild-type (Ler) and sub-1 pUBQ::gSUB:EGFP lines. Immunoprecipitation was performed using an anti-GFP antibody. Immunoblots were probed with the P4D1 anti-Ub antibody (top panel) and an anti-GFP antibody (bottom panel). B, bound fraction; IN, input. The experiment was independently repeated three times with similar results.

SUB:EGFP internalization involves clathrin-mediated endocytosis

So far, the obtained results indicate that SUB:EGFP is continuously internalized and eventually targeted to the vacuole for degradation. Next, we wanted to assess if SUB:EGFP relates to a clathrin-dependent process. We first tested if SUB:EGFP and endogenous CHC occur in the same complex in vivo. To this end, SUB:EGFP was immunoprecipitated from 7-day-old plate-grown pUBQ::SUB:EGFP sub-1 seedlings using an anti-GFP antibody. Immunoprecipitates were subsequently probed using an anti-CHC antibody. We could detect a CHC signal in immunoprecipitates derived from SUB:EGFP plants but not from wild-type (Fig. 4) indicating that SUB:EGFP and CHC are present in the same protein complex in vivo.

Fig. 4.

Co-immunoprecipitation of CHC with SUB:EGFP. (A) Total extracts of 7-day-old SUB:EGFP-expressing seedlings (upper lanes) or wild-type seedlings (lower lanes) were immunoprecipitated using GFP–Trap MA beads. Immunoblots were probed with anti-CHC (left panel) or anti-GFP antibodies (right panel). (B) Signal intensity quantification of anti-CHC-probed immunoblots (as shown in (A)) based on three independent experiments (means ±SD). Asterisks represent statistical significance (P<0.005, Student’s t test). B, bound fraction; IN, input.

We next assessed the contribution of clathrin to the internalization and subcellular distribution of SUB:EGFP. To this end, we investigated the effects of a transient but robust impairment of clathrin activity on the internalization and subcellular distribution of SUB:EGFP. Ectopic expression of the C-terminal part of CHC1 (HUB1) results in a dominant-negative effect due to the HUB1 fragment binding to and out-titrating clathrin light chains (Liu ). To assess the effect of the presence of the HUB fragment on the subcellular distribution of SUB:EGFP, the previously characterized 4-hydroxytamoxifen-inducible INTAM>>RFP-CHC1 (HUB) line (Robert ; Kitakura ) was crossed into a Col-0 wild-type line carrying the pUBQ::SUB:EGFP reporter. Epidermal cells of the root meristem of HUB/pUBQ::SUB:EGFP plants, hemizygous for each transgene, were then analysed upon induction.

The length of induction period was first determined, which enabled us to detect via confocal microscopy a defect in endocytosis, as indicated by a reduction of internal FM4-64 foci following a 5–10 min exposure to the stain. Under our growth conditions a significant reduction of internal FM4-64 puncta was observed after 3 d of continuous growth on induction medium while near complete absence of internal FM4-64 foci was detected after 4 d (Fig. 5B, C). If SUB:EGFP is subject to CME, a block in HUB-sensitive endocytosis should result in fewer internal SUB:EGFP-labelled foci and higher SUB:EGFP signal at the PM when compared with the SUB:EGFP-derived signal of a control line. We found a significant reduction in cytoplasmic SUB:EGFP puncta in the HUB/pUBQ::SUB:EGFP line after 3 d of growth on induction medium in comparison to the control (Fig. 5B). Upon 4 d of induction we detected an increase in SUB:EGFP signal at the PM (Fig. 5C). Taken together, our results suggest that CME contributes to the internalization of SUB:EGFP.

Fig. 5.

Requirement of clathrin function for SUB endocytosis. Fluorescence micrographs show optical sections of epidermal cells of root meristems. (A–C) Internalization of SUB:EGFP and uptake of endocytic tracer dye FM4-64 in epidermal meristems cells of 3-day-old INTAM>>RFP-CHC1 (HUB1)/pUBQ::SUB:EGFP seedlings that were placed on 2 μM 4-hydroxytamoxifen-containing induction medium for 2, 3, or 4 d, respectively. Ethanol served as mock. Box-and-whiskers plots of the quantification of the number of SUB:EGFP-positive spots per cell (A, B) and of the EGFP intensity at plasma membrane (C) after incubation. Asterisks represent statistical significance (P<0.0001, Student’s t test) DOI, days on induction medium. Scale bars: 5 μm. The experiments were independently repeated twice with similar results.

SUB genetically interacts with CLATHRIN HEAVY CHAIN

To further assess the role of clathrin in the SUB signaling mechanism, we tested a possible genetic interaction between SUB and CHC. To this end, we made use of several previously characterized T-DNA insertion lines carrying knock-out alleles of CHC1 and CHC2 (Kitakura ). Plants lacking both CHC1 and CHC2 function appear to be lethal (Kitakura ). Mutations in individual CHC genes, however, result in endocytosis defects and affect key processes such as polar distribution of PIN proteins, internalization of ATRBOHD, stomatal movement, and resistance to powdery mildew (Kitakura ; Hao ; Wu ; Larson ).

To test if clathrin is involved in SUB-controlled processes, we first investigated if chc mutants show a defect in root hair patterning. We compared the number of hair and non-hair cells in the non-hair (N) and hair (H) positions of the root epidermis (Dolan ), respectively, in wild-type, sub-9, and two different chc1 and chc2 alleles. In Col wild-type plants we found that 99% of cells at the H position were hair cells while only 1% of cells at the N position were hair cells (Table 1). In contrast, sub-9 mutants exhibited 75% hair cells in the H position and 29% hair cells in the N position, in line with previous results (Table 1; Kwak ). Next, we monitored root hair patterning in chc1-1, chc1-2, chc2-1, and chc2-2 mutants. We observed the strongest effect for chc2-2 as in this mutant 85% of cells at the H position were hair cells while 13% of cells at the N position were hair cells as well (Table 1). Comparable though slightly reduced effects were observed for chc2-1, while in chc1-1 and chc1-2 mutants the alterations in root hair patterning were noticeably weaker than in chc2 mutants. These results suggest that CHC2, and to a lesser extent CHC1, affect root hair patterning. We then assessed root hair patterning in sub-9 chc1 and sub-9 chc2 double mutants. We did not observe an exacerbated phenotype; rather, all the double mutants showed a sub-9-like phenotype indicating that sub-9 is epistatic to chc1 and chc2.

Table 1.

Distribution of root hair and non-hair cells in the root epidermis

		H position		N position
Genotype	n (roots)	Hair (%)	Non-hair (%)	Hair (%)	Non-hair (%)
Col-0	16	98.6±3.8	1.4±3.8	0.6±2.5	99.4±2.5
sub-9	14	75±10.4	25±10.4	28.9±13.6	71.1±13.6
chc1-1	15	93.4±8.6	6.6±8.6	7.8±9.3	92.2±9.3
chc1-2	16	93.5±8	6.5±8	7.4±9.1	92.6±9.1
chc2-1	10	89.8±11	10.2±11	11.5±10.6	88.5±10.6
chc2-2	23	84.4±13.7	15.6±13.7	13.2±12.3	86.8±12.3
chc1-1 sub-9	11	76.9±13.2	23.1±13.2	19.5±13.6	80.5±13.6
chc1-2 sub-9	10	76.6±15.4	23.4±15.4	24.5±15.9	75.5±15.9
chc2-1 sub-9	12	74.2±13.7	25.8±13.7	21.7±16.4	78.3±16.4
chc2-2 sub-9	12	71.9±12.6	28.1±12.6	28.3±7.1	71.7±7.1

Values are mean ±standard deviation. The experiment was repeated twice with similar results.

To corroborate these findings we generated homozygous chc mutants carrying a translational fusion of bacterial GUS to EGFP (GUS:EGFP) under the control of the Arabidopsis GLABRA2 (GL2) promoter (pGL2::GUS:EGFP). The GL2 promoter drives expression specifically in non-root hair cells and is commonly used to monitor root hair patterning (Masucci ; Kwak ). Accordingly, we found that all chc alleles tested showed an altered pattern of reporter signal in the root epidermis similar to sub-9, with chc2 alleles causing more prominent aberrations compared with chc1 mutations (Fig. 6; Table 2; Supplementary Fig. S1). Moreover, chc1 sub-9 or chc2 sub-9 double mutants did not show an exacerbated phenotype indicating that CHC1, CHC2, and SUB do not act in an additive fashion. The combined results indicate that SUB, CHC1, and CHC2 act in the same genetic pathway regulating root hair patterning.

Fig. 6.

Expression pattern of the pGL2::GUS:EGFP reporter in chc2-2 and chc2-2 sub-9 mutants. Fluorescence micrographs show optical sections of epidermal cells of root meristems of 7-day-old seedlings. FM4-64 was used to label cell outlines. (A) Col-0. (B) sub-9. (C) chc2-2. (D) chc2-2 sub-9. Note the similarly altered pattern in (B–D). Scale bars: 25 μm.

Table 2.

Distribution of pGL2::GUS:EGFP-expressing cells in the root epidermis

		H position		N position
Genotype	n (roots)	NECa (%)	ECb (%)	NECa (%)	ECb (%)
Col-0	8	96.6±3.4	3.4±3.4	1.6±3.4	98.4±3.4
sub-9	11	70.9±7.5	29.1±7.5	25.1±12.6	74.9±12.6
chc1-1	10	91.9±3.1	8.1±3.1	6.1±3.8	93.9±3.8
chc1-2	15	90.5±5.9	9.5±5.9	6.1±5.2	93.9±5.2
chc2-1	12	84.6±5.8	15.4±5.8	4.3±3.5	95.7±3.5
chc2-2	17	80.8±8.9	19.2±8.9	10.6±5.6	89.4±5.6
chc1-1 sub-9	6	65.9±16.7	34.1±16.7	32.3±9.3	67.7±9.3
chc1-2 sub-9	7	65.7±10.3	34.3±10.3	18.6±5.7	81.4±5.7
chc2-1 sub-9	11	67.4±8.6	32.6±8.6	19.4±7.1	80.6±7.1
chc2-2 sub-9	9	67.5±11.8	32.5±11.8	24.6±6.4	75.4±6.4

Values are mean ±standard deviation. The experiment was repeated twice with similar results.

Cells not displaying a pGL2::GUS:EGFP reporter signal.

Cells displaying a pGL2::GUS:EGFP reporter signal.

Next, we assessed if CHC1 and CHC2 participate in SUB-dependent floral development. In the Col-0 background null alleles of SUB cause a weaker floral phenotype when compared with similar alleles in the Ler background (Vaddepalli ). The Col-0 sub-9 allele causes mild silique twisting, mis-orientated cell division planes in the L2 layer of floral meristems, and ovule defects (Fig. 7; Tables 3, 4; Vaddepalli ). When analysed, we did not detect any obvious defects in floral meristems, flowers, and ovules of plants homozygous for the tested chc1 or chc2 alleles (Fig. 7; Tables 3, 4; Supplementary Fig. S2). We then investigated the phenotype of chc1 sub-9 and chc2 sub-9 double mutants. Interestingly, the cell division defects in the L2 layer of the FM were reduced in chc1 sub-9 and chc2 sub-9 double mutants in comparison to sub-9 single mutants (Fig. 7; Table 3; Supplementary Fig. S2). Suppression of the sub-9 phenotype in chc1 sub-9 or chc2- sub-9 double mutants was also observed for silique twisting and ovule development (Fig. 7; Table 4; Supplementary Fig. S2).

Fig. 7.

Phenotype comparison between Col-0, sub-9, chc2-2, and chc2-2 sub-9. (A–D) Scanning electron micrographs of stage 4 ovules (stages according to Schneitz ). In (B) note the aberrant outer integument (arrow). (E–H) Morphology of mature stage 13 or 14 flowers (stages according to Smyth ). (I–L) Morphology of siliques. (M–P) Central region of stage 3 floral meristems stained with modified pseudo-Schiff propidium iodide. (N) Arrowheads indicate aberrant cell division planes. In (P) note that the defects of the sub-9 phenotype were partially rescued in chc sub-9 double mutants. Scale bars: (A–D) 50 μm, (E–H) 0.5 mm, (I–L) 1 mm, (M–P) 10 μm.

Table 3.

Number of periclinal cell divisions in the L2 layer of stage 3 floral meristems

Genotype	NPCDa	Percentage	NFMb
Col-0	12	17.6	68
sub-9	17	36.2	47
chc1-1	6	23.1	26
chc1-2	7	20.0	35
chc2-1	7	22	31
chc2-2	5	22.5	25
chc1-1 sub-9	6	20.0	30
chc1-2 sub-9	7	19.4	36
chc2-1 sub-9	7	25.9	27
chc2-2 sub-9	7	18.9	37

Number of periclinal cell divisions observed.

Number of floral meristems observed.

Table 4.

Comparison of integument defects between sub-9, chc, and chc sub-9 mutants

Genotype	n total	n with defects	Percentage
Col-0	274	0	0
sub-9	291	82	28.2
chc1-1	130	0	0
chc1-2	121	0	0
chc2-1	126	0	0
chc2-2	230	0	0
chc1-1 sub-9	235	14	6
chc1-2 sub-9	185	11	6
chc2-1 sub-9	211	14	6.6
chc2-2 sub-9	237	13	5.5

Discussion

An impressive body of published work has elucidated many of the intricacies of receptor-mediated endocytosis of plant RKs. Much is known about the internalization and endocytic trafficking of plant RKs with functional kinase domains. The atypical RK SUB carries an inconspicuous kinase domain, but enzymatic kinase activity could not be demonstrated in in vitro biochemical experiments and is not required for its function in vivo (Chevalier ; Vaddepalli ; Kwak ). Using SUB as a model we have explored the endocytic route of an atypical RK (Figs 1, 2). We investigated this process by examining the subcellular distribution of a functional SUB:EGFP reporter in epidermal cells of the root meristem. Our data are compatible with the notion that PM-localized SUB becomes internalized and traffics from the TGN/EE to MVB/LEs and eventually the vacuole where it is destined for degradation. SUB:EGFP was observed to enter this endocytic route in the apparent absence of activation of SUB signaling by artificial stimulation or application of exogenous ligand. A similar observation was, for example, made for ACR4 (Gifford ). One interpretation of this finding could be that endogenous SUB ligand is always present at sufficient levels to promote SUB endocytosis. In another possible scenario, the rate of SUB internalization may be independent of ligand availability, as was shown for BRI1 (Russinova ; Geldner ). In any case, our data indicate that the endocytic route of the atypical RK SUB for the most part seems to adhere to the established pattern of plant receptor-mediated endocytosis.

Apart from being a central signal for proteasome-mediated degradation ubiquitination is a major endocytosis determinant of PM proteins (Haglund and Dikic, 2012; Isono and Kalinowska, 2017). Several plant RKs are known to be ubiquitinated, including FLS2 (Lu ), BRI1 (Martins ; Zhou ), and LYK5 (Liao ). The observed in vivo ubiquitination of SUB:EGFP (Fig. 3) is compatible with the notion of SUB being internalized and transported to the vacuole for degradation. However, it remains to be determined which E3 ubiquitin ligase promotes ubiquitination of SUB and how SUB endocytosis relates to the control of its signaling. Internalization can be linked with downstream responses, as was demonstrated for FLS2 or the AtPep1–PEPR complex (Mbengue ; Ortiz-Morea ), or contribute to signal downregulation, as it is the case for BRI1 (Irani ; Zhou ) or LYK5 (Liao ).

Several lines of evidence support the notion of SUB:EGFP undergoing CME. First, CHC in vivo co-immunoprecipitated with SUB:EGFP (Fig. 4). Second, we observed a reduction in intra-cellular SUB:EGFP puncta accompanied with a stronger SUB:EGFP signal at the PM in the HUB-line upon induction (Fig. 5). Third, our genetic analysis revealed a connection of SUB with a clathrin-dependent process. Plants with a defect in CHC2 show a significantly reduced endocytosis rate of FM4-64 and aberrant polar localization of the polar auxin transporter PINFORMED 1 (PIN1) (Kitakura ), as well as reduced internalization of, for example, PEP1 (Ortiz-Morea ), FLS2 (Mbengue ), and BRI1 (Wang ). Accordingly, chc2 mutants show multiple defects, including patterning defects in the embryo (Kitakura ), impaired mitogen-activated protein kinase activation (Ortiz-Morea ), and defective stomatal closure as well as callose deposition upon bacterial infection (Mbengue ). Our genetic analysis revealed that CHC2, and to a lesser effect CHC1, also affects root hair patterning (Fig. 6). Importantly, it provides evidence for a biologically relevant interaction between SUB and a CHC-dependent process.

Interestingly, the genetics suggests that the type of genetic interaction between SUB and CHC depends on the tissue context. In the root, SUB and CHC2 act in the same genetic pathway regulating root hair patterning. Several hypotheses are conceivable that could explain the result. As our data support the notion of SUB:EGFP undergoing CME, one model states that CME of SUB is required for root hair patterning. Therefore, SUB internalization in single chc mutants would be reduced resulting in a hyperaccumulation of SUB at the PM. Two alternative further scenarios are compatible with this notion. In the first scenario hyperaccumulation of SUB at the PM interferes with root hair patterning. This view is supported by the observation that not just a reduction of SUB activity but also ectopic expression of SUB in p35S::SUB plants results in a weak defect in root hair patterning (Kwak and Schiefelbein, 2007), similar to what we observed for chc2 mutants. In the second scenario, a reduction of SUB internalization leads to fewer SUB-labelled endosomes, which in turn impairs root hair patterning. This scenario implies that SUB signals while being present on endosomes. In another model, a reduction of CHC activity could influence clathrin-dependent secretion of newly translated and/or recycled SUB to the PM thereby reducing the level of active SUB at the PM below a certain threshold. Finally, given the pleiotropic phenotype of chc mutants the genetic data do not rule out a more indirect interaction between SUB and CHC. Further work remains to be done to discriminate between the different possibilities. However, we currently favor the notion that CME of SUB is critical for root hair patterning as a block of CME of SUB:EGFP in the HUB line results in a reduction of internalized SUB:EGFP vesicles and elevated levels of SUB:EGFP at the PM.

A role of CHC in floral development is not revealed by chc single mutants, as they show apparently wild-type flowers (Fig. 7). However, involvement of CHC could be masked by tissue-specific redundancy between CHC1 and CHC2. Indeed, the wild-type appearance of floral organs of sub chc double mutants indicates that CHC contributes to floral morphogenesis. The respective role of CHC and the molecular basis of the genetic interaction between SUB and CHC in this process remain to be investigated. It will be an exciting challenge to unravel the molecular details of how SUB and clathrin interact to allow tissue morphogenesis in future studies.

Supplementary data

Supplementary data are available at JXB online.

Fig. S1. Comparison of the root hair patterning phenotype of wild-type, sub-9, chc1-1, and chc2-1 mutants.

Fig. S2. Comparison of the floral phenotype of wild-type, sub-9, chc1-1, and chc2-1 mutants.

Table S1. List of all primers used in this study.

Click here for additional data file.