Dynamic analysis of Arabidopsis AP2 σ subunit reveals a key role in clathrin-mediated endocytosis and plant development

Endocytosis is the inward budding of the plasma membrane to form membrane vesicles, which are then transported to various subcellular compartments, and is the means by which cells take up extracellular materials, plasma membrane proteins and lipids (Conner and Schmid, 2003). In animal cells, endocytosis plays an essential role in signal transduction by specifically internalizing cell surface receptors to deactivate the response of cells to an extracellular stimulus (Sorkin and von Zastrow, 2009; McMahon and Boucrot, 2011). Clathrin-mediated endocytosis is the principal route by which cells internalize the transmembrane receptors (Doherty and McMahon, 2009; Taylor et al., 2011). Increasing evidence has demonstrated that clathrin-mediated endocytosis regulates cellular homeostasis, synaptic transmission and cell differentiation by controlling constitutive and stimulated internalization of many receptors (McMahon and Boucrot, 2011). Bacterial and viral toxins also enter cells through this pathway (Chen and Zhuang, 2008; Cureton et al., 2009; Abrami et al., 2010).

During clathrin-mediated endocytosis, clathrin-coated vesicle (CCV) formation and cargo selection require endocytic adaptor proteins. Two classes of adaptors have been identified based on different functions in clathrin-mediated endocytosis: the monomeric adaptor proteins (such as AP180, epsin and β-arrestin) and the heterotetrameric adaptor protein complexes, such as the AP2 (or AP-2) adaptor complex (Owen et al., 2004; Robinson, 2004). The AP2 complex is the main component among endocytic vesicle coat proteins, second in abundance only to clathrin itself (Schmid and McMahon, 2007; Kelly and Owen, 2011), and can link membrane cargo containing YXXØ-type sorting signals to the endocytic vesicle via binding of the μ2 subunit (Kelly et al., 2008; Traub, 2009). Recent experiments have shown that the α and σ2 subunits form a hemicomplex that interacts with the [DE]XXXL[LI] sorting signal and is also involved in the selective internalization of cargo proteins (Doray et al., 2007; Diril et al., 2009). Moreover, crystallography has identified the σ2 subunit as the major interaction site (Jackson et al., 2010).

The plant genome encodes proteins homologous to those that participate in clathrin-mediated endocytosis in mammalian cells. However, the molecular machinery of clathrin-mediated endocytosis in the plant cell is still uncharacterized (Samaj et al., 2005; Bassham et al., 2008). Recent studies indicate that clathrin-dependent endocytosis exists in plant cells and participates in many fundamental cellular processes (Pérez-Gómez and Moore, 2007; Kaiserli et al., 2009; Chen et al., 2011). Sequence analysis has suggested that Arabidopsis possesses four AP complexes (AP1, AP2, AP3 and AP4), which might be involved in different intracellular trafficking processes (Happel et al., 2004; Bar et al., 2009; Feraru et al., 2010; Zwiewka et al., 2011). Recently, a fifth adaptor protein (AP5) was identified. AP5 is localized on late endosomes and may be involved in endosomal dynamics (Hirst et al., 2011; Hirst et al., 2013). Although the AP2 complex is not essential for clathrin-mediated endocytosis in yeast (Yeung et al., 1999), it is involved in the internalization of membrane proteins in mammalian cells (Motley et al., 2003; Huang et al., 2004; Boucrot et al., 2010). The role of the homologous complex in plants has yet to be investigated.

In this study, we identified and characterized the putative σ subunit of the AP2 complex, a novel adaptor of the clathrin-mediated endocytic machinery in Arabidopsis. We found that Arabidopsis AP2 σ was ubiquitously expressed, with a protein distribution at both the plasma membrane and in the cytoplasm. A loss-of-function ap2 σ mutant was impaired in endocytosis, showing defects in multiple aspects of plant development. Co-immunoprecipitation and fluorescence cross-correlation spectroscopy (FCCS) analyses showed that AP2 σ interacts with clathrin light chain (CLC) in vivo. To investigate the molecular dynamics at high spatiotemporal resolution on the plasma membrane, we made used of variable-angle total internal reflection fluorescence microscopy (VA-TIRFM) combined with a single-particle tracking (SPT) assay (Li et al., 2011; Wan et al., 2011) to reveal that AP2 σ displays similar dynamic behavior to CLC and accumulates at the endocytic clathrin-coated pits (CCPs). These findings provide new insight into the machinery and function of clathrin-mediated endocytosis in plant cells.

Arabidopsis thaliana plants of ecotype Colombia-0 (Col-0) were used in all experiments. Seeds were surface sterilized in 75% ethanol for 10 minutes, washed five times with 95% ethanol, then plated on half-strength Murashige and Skoog (MS) medium (Murashige and Skoog, 1962) containing 1% (w/v) sucrose and 0.9% (w/v) agar (Duchefa, The Netherlands) and grown vertically at 22°C under a light regime of 16 hours light and 8 hours dark. An ap2 σ T-DNA insertion mutant was purchased from ABRC. The genotype of the ap2 σ insertion line was determined by PCR and confirmed by RT-PCR. Primers for T-DNA insertion mutant identification and RT-PCR are listed in supplementary material Table S3.

The CLC-EGFP and AP2 σ-EGFP plant expression vectors were constructed as follows. The CLC and AP2 σ sequences, which included the promoter elements and coding regions, were PCR amplified from Arabidopsis genomic DNA with primers CLC_for/CLC_rev and AP2 σ_for/AP2 σ_rev and then subcloned as PstI-SpeI and PstI-KpnI fragments into PM999-EGFP (kindly provided by Jian Xu, National University of Singapore) to produce PM999-CLC-EGFP and PM999-AP2 σ-EGFP, respectively. The CLC_rev and AP2 σ_rev primers contained linker sequences encoding Gly-Gly-Gly-Gly-Ala that appended the CLC and AP2 σ genomic fragment onto the N-terminus of EGFP. After digestion with PstI and SmaI, the fragments of PM999-CLC-EGFP and PM999-AP2 σ-EGFP that included the genomic fragments of CLC/AP2 σ, EGFP and NOS were subcloned into the binary vector pCambia 1301 (CAMBIA, Canberra, Australia) to obtain pCambia 1301-CLC-EGFP and pCambia 1301-AP2 σ-EGFP.

To create the AP2 σ-mCherry expression vector, the coding sequence for mCherry was amplified using mCherry_for/mCherry_rev and then subcloned as a KpnI/BamHI fragment into pCambia 1301-AP2 σ-EGFP to replace EGFP, producing pCambia 1301-AP2 σ-mCherry.

Primers used for plasmid construction are listed in supplementary material Table S3.

Cotyledon veins of 7-day-old seedlings were observed. The cotyledons were cleared in 100% ethanol overnight to remove chlorophyll and then mounted on slides containing 70% lactic acid. For analysis of developmental defects of embryos, developing seeds were cleared in an 8:3:1 (w:v:v) mixture of chloral hydrate:water:glycerol and observed on a Zeiss LSM 510 META microscope under differential interference contrast optics. For transmission electron microscopy (TEM), the samples were prepared according to a standard procedure (Luft, 1961) and examined under an FEI Tecnai Spirit microscope.

FRAP experiments were performed on an FV1000MPE multiphoton laser-scanning microscope (Olympus). A rectangular region of interest was drawn around the area for bleaching by 488-nm laser lines operating at 100% laser power. The fluorescence recovery was monitored at 2-second intervals. The fluorescence recovery data obtained were corrected for bleaching during imaging as described (Konopka et al., 2008).

To evaluate the rate of endocytosis, the fluorescent styryl membrane probe FM4-64 (Invitrogen) was used. FM4-64 was kept as a 5 mM stock solution and diluted with half-strength liquid MS medium to produce a 5 μM working solution. Wild-type and ap2 σ seedlings were grown on half-strength MS plates for 3 days and then incubated in FM4-64 working solution for 5 minutes at room temperature. The seedlings were washed three times in half-strength liquid MS medium and transferred to glass slides. The internalization of FM4-64 was monitored at 5-minute or 30-minute intervals at room temperature using a Leica TCS SP5 II laser-scanning confocal microscope.

To monitor the expression of the AP2 σ-EGFP fusion protein in transgenic plants, total proteins were extracted from AP2 σ-EGFP and 35S::EGFP transgenic seedlings grown on half-strength MS plates. The seedlings were ground into a fine powder in liquid nitrogen, mixed in a 1:1 (v:v) ratio with SDS-PAGE sample buffer, heated at 95°C for 15 minutes and then centrifuged at 12,000 rpm (10,625 g) for 5 minutes to remove cellular debris. The supernatants were subjected to 10% SDS-PAGE. Immunoblot analysis was performed with monoclonal anti-GFP antibody (Sigma-Aldrich). Horseradish peroxidase (HRP)-conjugated anti-mouse secondary antibody (Sangon Biotech, Shanghai, China) was used to detect the primary antibody.

For co-immunoprecipitation, 1 g of 10-day-old transgenic plants expressing CLC-EGFP, AP2 σ-mCherry and CLC-EGFP/AP2 σ-mCherry were harvested and ground in liquid nitrogen. The co-immunoprecipitation assay was performed as previously described (Serino and Deng, 2007).

TyrA23, TyrA51 and BFA (Sigma-Aldrich) were dissolved in DMSO to yield a 50 mM stock solution. For drug treatment, vertically grown seedlings were incubated in half-strength liquid MS medium containing 50 μM TyrA23 or 50 μM BFA at room temperature for the indicated time before VA-TIRFM or laser-scanning confocal microscopy analysis. Control cells were treated with equivalent amounts of DMSO.

Plasmolysis was carried out by incubating 4-day-old seedlings in 1 M NaCl for 5 minutes before observation.

FCCS measurements were carried out on the Leica TCS SP5 microscope. To avoid overestimation of cross-correlation, a low-power excited laser was used to prevent bleed-through between channels. The emitted signal was detected by avalanche photodiodes (APD) and transferred to a second computer where it was analyzed with ISS VISTA software. The fluorescence autocorrelation functions of the red G_r(τ) and green G_g(τ) channels, as well as the cross-correlation function G_c(τ), were calculated using the integrated function within ISS VISTA, where τ denotes the time delay. The acquired G_r(τ), G_g(τ) and G_c(τ) data were fitted to a 2D-Gaussian, two-particle type model. The relative cross-correlation ([G_c(0)]/[G_g(0)] and [G_c(0)]/[G_r(0)]) was obtained by normalizing the amplitude of the cross-correlation function to the amplitude of the autocorrelation function of CLC-EGFP or AP2 σ-mCherry.

Seedlings were grown on half-strength MS agar plates and cultured vertically at 22°C for 3 days, then mounted on a glass slide with half-strength liquid MS medium and observed under a home-built VA-TIRF microscope, which was based on an Olympus IX-71 microscope equipped with a ×100 oil-immersion objective (numerical aperture of 1.45, Olympus) and makes use of subcritical incident angles of the laser to image the dynamics of fluorescently labeled proteins in plant epidermal cells. A diode laser (Changchun New Industries Optoelectronics Tech. Co., Changchun, China) was used for illumination through a total internal reflection illuminator. EGFP- and mCherry-labeled proteins were excited with 473-nm and 561-nm laser lines and the emission fluorescence signals were passed through two band-pass filters (525/45 and 609/54 nm) before being detected by a back-illuminated EM-CCD camera (ANDOR iXon DV8897D-CS0-VP; Andor Technology, Belfast, UK). Images were acquired using a 200-msecond exposure time at 1-second intervals and were processed using ImageJ software (NIH). To analyze the dynamic properties, AP2 σ-EGFP and CLC-EGFP spots were tracked using spatial and temporal global particle assignment as described (Jaqaman et al., 2008).

Sequence data used in this study are available from the Arabidopsis Information Resource (TAIR) database under accession numbers At2g40060 (CLC) and At1g47380 (AP2 σ).

To characterize the function of AP2 σ in Arabidopsis, we first performed phylogenetic analysis using Molecular Evolutionary Genetics Analysis (MEGA) 4. Our analysis showed that the Arabidopsis genome contains a single gene for AP2 σ, which is closely related to animal AP2 σ (supplementary material Fig. S1A). In addition, AP2 σ is highly conserved across the fungal, animal and plant kingdoms (supplementary material Fig. S1B). Sequence alignment analysis further revealed that AP2 σ shares high similarity (61% identity) with human AP2 σ (AP2S1) (supplementary material Fig. S1C).

We obtained the AP2 σ heterozygous T-DNA insertion mutant (SALK_141555) from the Arabidopsis Biological Resource Center (ABRC) and screened for the homozygous line (Fig. 1A). We could not detect AP2 σ transcripts in the homozygous mutant by RT-PCR using primers flanking the T-DNA insertion site (Fig. 1B). Phenotypic analysis showed that multiple stages of plant growth and developmental processes were disturbed in the ap2 σ mutant. Most strikingly, the cotyledons exhibited defective organogenesis, and 19.6% of the seedlings had single, triple or fused cotyledons (Fig. 1C-G; supplementary material Table S1). Moreover, the abnormal cotyledon was also associated with an altered vascular pattern. In the wild type, the vascular tissues of the cotyledons formed closed networks (supplementary material Fig. S2A). By contrast, the cotyledons of the ap2 σ mutant exhibited an open vascular network, which was characterized by an increase in free-ending veins (supplementary material Fig. S2B-D). When grown in soil, the ap2 σ mutant seedlings exhibited dwarf phenotypes, with reduced leaf size and stem height (Fig. 1H,I). In addition, the dwarf seedlings also produced shorter siliques than wild-type plants (Fig. 1J,K): the average silique lengths were 12.49±1.60 mm for the wild type and 3.77±0.54 mm for the ap2 σ mutant (Fig. 1L).

To confirm that the developmental defects in the T-DNA mutant were caused by the loss of AP2 σ function, we transformed an AP2 σ-EGFP translational fusion construct into the ap2 σ mutant. Immunoblot analysis showed that AP2 σ was expressed as a fusion protein with EGFP (supplementary material Fig. S3A). Importantly, the transformation of the ap2 σ mutant with AP2 σ-EGFP fully rescued the developmental defects (supplementary material Fig. S3B-D). These results indicate that AP2 σ-EGFP is functional and that AP2 σ is required for normal plant growth and development.

Given the prominent defect in cotyledon formation, we investigated embryo development in the ap2 σ mutant. In contrast to the wild type (Fig. 2A-C), we observed abnormal embryo shapes at the globular stage in the ap2 σ mutant (Fig. 2D). In the heart and torpedo stages, single cotyledon primordium was visible in ap2 σ mutant embryos (Fig. 2E,F). Moreover, the distribution of DR5::GFP (a synthetic auxin reporter) was altered in the tip of incipient cotyledons of ap2 σ mutant embryos (Fig. 2G-L), indicating that the distribution of auxin is disrupted.

As the establishment of correct auxin maxima and formation of a normal bilateral axis depend on the polar localization of the auxin efflux transporter PIN1 in the embryo, we next compared the localization of a functional PIN1-GFP fusion protein in the wild type and ap2 σ mutant. PIN1-GFP was expressed throughout the wild-type embryo, with higher expression in the apical and central region of the globular embryo. At the heart stage, PIN1-GFP was restricted to the vascular precursor cells, with higher levels at the basal boundary towards the site of the future root tip. The opposite polarity was observed in the protodermal cell layer of cotyledon primordia (Fig. 3A-D). In the ap2 σ mutant, PIN1-GFP was mainly localized in the region of the cotyledon and vasculature initiation sites at the globular and heart stages, but the subcellular localization of PIN1-GFP was significantly disrupted. The polar localization of PIN1-GFP was lost, accompanied by large aggregations in the membrane-associated regions (Fig. 3E-H).

Considering that PIN1 undergoes constitutive clathrin-dependent endocytosis and recycling between plasma membrane and endosomes (Dhonukshe et al., 2007; Dhonukshe et al., 2008), we investigated the effects of the ap2 σ mutation on its internalization. To visualize PIN1-GFP endocytosis, we inhibited endocytic recycling using the fungal toxin brefeldin A (BFA), which causes the heterogeneous aggregation of internalized PIN1 in intracellular compartments (known as ‘BFA bodies’) (Pan et al., 2009; Naramoto et al., 2010). As expected, the aggregation of PIN1-GFP in BFA bodies was clearly observed in the wild type when treated with BFA (Fig. 3I). Although BFA treatment also caused the formation of PIN1-GFP aggregates, the formation of BFA bodies was remarkably reduced in the ap2 σ mutant (Fig. 3J). Further statistical analysis revealed that the fluorescence intensity and size of the PIN1-GFP-labeled BFA bodies decreased by 25% and 45%, respectively, in the ap2 σ mutant as compared with the wild type (Fig. 3K,L). Colocalization and FCCS analysis further revealed that PIN1-GFP was associated with AP2 σ-mCherry (supplementary material Fig. S4A,B). Immunoblot analysis demonstrated that there were no differences in PIN1-GFP protein levels between the wild type and the ap2 σ mutant, ruling this out as having any effect on PIN1 internalization and localization (supplementary material Fig. S5A). These results indicate that AP2 σ is required for the proper internalization of PIN1.

Microarray data (publicly available via Genevestigator, https://www.genevestigator.com) show that AP2 σ is ubiquitously expressed in all tissues and throughout all developmental stages and that its expression level is barely affected by the factors analyzed in the microarray data (Zimmermann et al., 2004). Live cell imaging revealed the AP2 σ-EGFP fluorescence signal as a thin layer at the cell periphery in all cell types analyzed, although some fluorescence was also observed in the cortical region of the cells (supplementary material Fig. S6A-D). To further examine whether the AP2 σ-EGFP fluorescent signals are localized to the plasma membrane, periplasmic space or cell wall, we performed a plasmolysis experiment with 1 M NaCl treatment before confocal imaging. A substantial fraction of the AP2 σ-EGFP fluorescent signal remained associated with the shrunken plasma membrane, with little residual signal in the cell walls of the plasmolyzed root epidermal cells (supplementary material Fig. S7). These results indicate that AP2 σ-EGFP is localized both at the plasma membrane and in the cytoplasm.

To study how AP2 σ-EGFP is recruited to the plasma membrane, we examined the dynamics of AP2 σ-EGFP in the subapical lateral plasma membrane of growing root hairs using FRAP. The initial bleaching region was divided into three parts designated as the top, middle and bottom sectors (Fig. 4A). Our FRAP experiment showed that the three regions recovered at equal rates (Fig. 4A,B), with a similar average half-life and percentage recovery (supplementary material Table S2), suggesting that AP2 σ-EGFP is primarily recruited from the cytoplasm to the plasma membrane.

To test whether the AP2 complex plays a role in endocytosis in Arabidopsis, we examined the uptake of the commonly used endocytic tracer FM4-64 in the wild type and ap2 σ mutant (Fig. 5A,B). In the wild type, FM4-64-labeled fluorescent puncta could be detected in the cytoplasm 5 minutes after labeling, whereas in the ap2 σ mutant little FM4-64 uptake could be detected even 25 minutes after labeling (Fig. 5C,D).

To determine the endocytic rate, we monitored the internalization of FM4-64 on a long time scale. Thirty minutes after labeling, the internalization of the dye was obvious in wild-type roots; by contrast, no dye accumulation was apparent in the ap2 σ mutant. One hour after labeling, a large number of FM4-64-labeled vesicles were detected in the cytoplasm of wild-type cells, whereas in the ap2 σ mutant only limited cytoplasmic fluorescence was detected even 90 minutes after labeling (supplementary material Fig. S8). The effect of AP2 σ on endocytosis was further confirmed by rescue of the FM4-64 internalization rate in the complementation lines (supplementary material Fig. S9A,B).

To gain further insight into the molecular nature of the endocytic defect of the ap2 σ mutant, we investigated the colocalization of AP2 σ and CLC. We generated transgenic plants co-expressing CLC-EGFP and AP2 σ-mCherry driven by their native promoters. Confocal imaging analysis showed a high level of CLC-EGFP and AP2 σ-mCherry overlap at the plasma membrane (Fig. 6A-C), as confirmed by quantitative analysis [the protein proximity index (PPI)] (Zinchuk et al., 2011). Fig. 6D is a 3D plot of the cross-correlation of CLC-EGFP and AP2 σ-mCherry as a function of pixel shift. The peak at the center of the 3D surface plot decreased significantly upon shifting the image a few pixels (Fig. 6D), indicating specific colocalization of CLC and AP2 σ. The mean PPI values were 0.69±0.15 for the proximity of CLC to AP2 σ and 0.62±0.13 for the proximity of AP2 σ to CLC (Fig. 6E).

The possible interaction between CLC and AP2 σ was then investigated using co-immunoprecipitation and FCCS. AP2 σ-mCherry was co-immunoprecipitated with CLC-EGFP in transgenic plants co-expressing the two proteins, indicating that AP2 σ interacts with CLC in vivo (supplementary material Fig. S10). The interaction between AP2 σ and CLC was further analyzed using FCCS. For each dataset, the amplitude of cross-correlation was compared with the green (CLC-EGFP) or red (AP2 σ-mCherry) autocorrelation amplitude, which represented the percentage of molecules associated with each other. A strong association was indeed observed between CLC-EGFP and AP2 σ-mCherry in living cells, and the cross-correlation values were 0.61±0.13 for CLC to AP2 σ and 0.63±0.14 for AP2 σ to CLC (Fig. 6F,G). Moreover, we found that both tyrphostin A23 (TyrA23), an inhibitor of clathrin-dependent endocytosis, and its inactive analog tyrphostin A51 (TyrA51) had no significant effect on the association of the two proteins (supplementary material Fig. S11A-C).

We then applied VA-TIRFM to analyze the dynamics of AP2 σ-EGFP at the plasma membrane. When transgenic plants expressing AP2 σ-EGFP were viewed under VA-TIRFM, the fluorophores appeared as small diffraction-limited fluorescent spots (Fig. 7A) and displayed similar dynamic behavior to CLC-EGFP, including the appearance into, and disappearance from, the focal plane (Fig. 7B-F). Interestingly, AP2 σ-EGFP exhibited some unique characteristics. Most of the AP2 σ-EGFP signals were diffraction-limited fluorescence spots, and no large puncta were observed in the AP2 σ-EGFP transgenic plants, whereas the cortical trans-Golgi network could be clearly detected in CLC-EGFP transgenic plants (Fig. 7A,C; supplementary material Movies 1, 2).

We further analyzed the dynamic behavior of AP2 σ-EGFP and CLC-EGFP by measuring their diffusion coefficients and mean velocities. The diffusion coefficients were calculated by performing a linear fit of the mean-squared displacement (MSD) versus time. Interestingly, the diffusion coefficients of AP2 σ-EGFP and CLC-EGFP were similar, both ranging from 10^-3 μm²/second to 1 μm²/second. The Ĝ values (characteristic of diffusion coefficients) were 2.67×10^-2 μm²/second for AP2 σ-EGFP and 2.43×10^-2 μm²/second for CLC-EGFP (Fig. 7G,H). The two proteins also moved with similar velocities: 6.17±0.13×10^-1 μm/second for AP2 σ-EGFP and 5.94±0.065×10^-1 μm/second for CLC-EGFP (Fig. 7I,J).

To verify the relationship between AP2 σ and clathrin, we investigated the effects of TyrA23 on the dynamics of AP2 σ. When AP2 σ-EGFP transgenic Arabidopsis seedlings were treated with TyrA23, the number of AP2 σ-EGFP spots at the plasma membrane was significantly reduced. When the seedlings were treated with TyrA51, this effect was not observed (supplementary material Fig. S12A,B), indicating that AP2 σ-EGFP-labeled fluorescent spots were specifically sensitive to TyrA23, which is similar to its effect on the dynamin-related protein DRP1A during clathrin-mediated endocytosis in Arabidopsis (Fujimoto et al., 2010).

Given that AP2 σ had a high cross-correlation with CLC and displayed very similar dynamics to CLC-EGFP at the plasma membrane, we next examined their dynamics at the plasma membrane by imaging plants co-expressing AP2 σ-mCherry and CLC-EGFP using dual-color VA-TIRFM. In root epidermal cells, the CLC-EGFP and AP2 σ-mCherry fusion proteins appeared as well-dispersed diffraction-limited fluorescent spots at the plasma membrane and the AP2 σ-mCherry fluorescent spots overlapped with those of CLC-EGFP. In addition, some fluorescent spots containing only GFP or mCherry signal were observed (Fig. 8A-C). Fig. 8D shows time-lapse imaging analysis of a single fluorescent spot exhibiting both green and red fluorescence. At t=5 seconds, the AP2 σ-mCherry fluorescence appears and increases rapidly, and then the CLC-EGFP fluorescence appears at the same site and colocalizes with the red signal (t=11.25 seconds). At t=35 seconds, the AP2 σ-mCherry fluorescence has disappeared and only the CLC-EGFP spot is still visible, until it disappears 2 seconds later. From the kymograph and a comparison of the fluorescence intensity profiles representing the CCP labeled by CLC-EGFP and AP2 σ-mCherry (Fig. 8E,F), it is obvious that the AP2 σ-mCherry fluorescence appeared and disappeared ahead of the CLC-EGFP fluorescence.

To provide further insight into the requirement of AP2 σ in clathrin-mediated endocytosis, we examined the localization and dynamics of CLC-EGFP in the ap2 σ mutant. Immunoblot analysis showed that the CLC-EGFP protein level in the wild type and ap2 σ mutant were similar (supplementary material Fig. S5B). Confocal analysis showed that CLC-EGFP was localized to the plasma membrane, intracellular structures (presumably the trans-Golgi network) and cytoplasm. However, the plasma membrane-associated CLC-EGFP signal was significantly lower in the ap2 σ mutant than in wild-type seedlings (supplementary material Fig. S13A-C). Under VA-TIRFM, CLC-EGFP in the wild type was distributed in small spots at the plasma membrane and underwent constant internalization (Fig. 9A). In the ap2 σ mutant CLC-EGFP spots still formed, but the CCP density was significantly lower than in the wild type (0.65±0.13 μm² in the wild type versus 0.38±0.05 μm² in the ap2 σ mutant) (Fig. 9B,E; supplementary material Movie 3).

Because differences in CCP lifetime can affect their density, we further measured the turnover rate of CCPs. As shown in the kymograph, CLC-EGFP in the wild type was efficiently recruited to the plasma membrane, whereas the turnover rate was greatly reduced in the ap2 σ mutant (Fig. 9C,D). The appearance and disappearance rate of CLC-EGFP in the ap2 σ mutant decreased significantly to 70% of that in the wild type (Fig. 9F). TEM showed that 0.30±0.03% of the cell surface was occupied by CCVs in the wild type, whereas CCVs were less abundant in the ap2 σ mutant, occupying only 0.14±0.01% of the cell surface (supplementary material Fig. S14A-C).

Clathrin-mediated endocytosis is a highly dynamic process during which cell surface proteins and plasma membrane components are internalized into the cytoplasm. This process can be tracked in living cells using FM4-64, a membrane-selective fluorescent dye that can stain membrane in an activity-dependent manner (Leborgne-Castel et al., 2008; Men et al., 2008). Our results showed that depletion of AP2 σ function significantly inhibited the uptake of the dye, indicating that AP2 σ is required for general endocytic events in Arabidopsis. We also found that AP2 σ was ubiquitously expressed and mainly localized at the plasma membrane. Our FRAP analysis further showed that AP2 σ was directly recruited from the cytoplasm to the plasma membrane. Given the subcellular localization of AP2 σ-EGFP and the prominent interference of loss of function of AP2 σ in FM4-64 internalization in plant cells, we can safely conclude that AP2 σ represents one of the major adaptor proteins in Arabidopsis.

During clathrin-mediated endocytosis, more than 60 functionally distinct adaptor and accessory proteins are required, with functions ranging from clathrin recruitment and cargo sorting to membrane bending and vesicle scission (Maldonado-Báez and Wendland, 2006; Mettlen et al., 2009). Among these proteins, the AP2 complex was the first CCV adaptor to be characterized (Robinson, 1987; Collins et al., 2002). Arabidopsis AP180 has been identified as a clathrin assembly protein that binds to plant αC-adaptin (Barth and Holstein, 2004). To determine whether AP2 participates in the formation of CCVs in plants, we first analyzed the colocalization of fluorescently labeled AP2 with clathrin. Fluorescence microscopy images, although informative, cannot be used to precisely quantify protein colocalization owing to the possibility of false values resulting from the image background. In the present study, we applied a newly developed PPI to evaluate colocalization quantitatively in living cells (Li et al., 2010; Zinchuk et al., 2011). The PPI analysis revealed that AP2 σ has a high degree of colocalization with clathrin in living plant cells. Using the co-immunoprecipitation assay, we further demonstrated a direct physical interaction between AP2 σ and clathrin. This was supported by FCCS analysis, which yielded detailed quantitative information about the transient and dynamic interaction between AP2 σ-mCherry and CLC-EGFP (Bacia et al., 2006; Erdel et al., 2010; Fitzpatrick and Lillemeier, 2011). In cells expressing AP2 σ-mCherry and CLC-EGFP, we observed a significant cross-correlation between AP2 σ and CLC in accordance with the colocalization data calculated using PPI. These comprehensive analyses provided strong evidence that AP2 σ and CLC exhibit a high-affinity interaction in living cells, indicating that AP2 σ serves as a key binding partner for clathrin.

Clathrin-mediated endocytosis is complex and dissecting how the multiple components in this process are spatially and temporally organized poses a challenge. VA-TIRFM is a type of microscopy that takes advantage of the unique property of the evanescent wave produced at the interface between the specimen and a glass coverslip and confines the excitation of the fluorophores to a depth of 100-400 nm in the specimen (Mattheyses et al., 2010; Wan et al., 2011). This provides a powerful approach to clarify spatially the details of complex kinetics at the single-molecule level in living plant cells (Ehrhardt and Frommer, 2012). From VA-TIRFM observations, we found that AP2 σ-EGFP formed discrete foci at the plasma membrane. SPT analysis showed that these foci displayed similar dynamics to those formed by CLC. Moreover, the localization of AP2 σ-EGFP at the plasma membrane was strongly disturbed by treatment with TyrA23. We could not rule out the possibility that TyrA23 acts as an inhibitor of tyrosine kinase to influence other cellular processes or that it inhibits the interaction between the tyrosine-based motif and other adaptor complexes, leading to the disruption of other membrane trafficking events (Banbury et al., 2003). Therefore, we examined the effect of TyrA51, an analog of TyrA23 with no influence on clathrin-mediated endocytosis, and found that it had no significant influence on the dynamic behavior of AP2 σ-EGFP. These observations further support the notion that AP2 σ and CLC are parts of the same endocytic machinery.

Clathrin-mediated endocytosis can be divided into five distinct stages in animal cells: initiation, assembly, maturation, vesicle scission and uncoating (McMahon and Boucrot, 2011). In the current investigation, we found that most of the AP2 σ-mCherry fluorescent foci visualized in the VA-TIRFM images also contained CLC-EGFP signal, confirming that AP2 σ colocalizes with CLC at the plasma membrane. In ap2 σ mutant cells, both the number and turnover rate of CLC-EGFP spots were significantly attenuated, indicating that AP2 σ promotes the formation of CLC-EGFP spots at the plasma membrane. Time-lapse imaging and kymograph analysis further revealed that the AP2 σ-mCherry fluorescence spots appeared and disappeared before the CLC-EGFP spots. Assuming that the above five stages of clathrin-mediated endocytosis exist in plant cells, we tentatively conclude that AP2 is involved in the CCV initiation, assembly and maturation stages, but is not necessary for the detachment of endocytic vesicles from the plasma membrane in Arabidopsis.

Some previous studies have focused on the role of AP2 in the regulation of embryonic development (Mitsunari et al., 2005; Yu et al., 2007). However, unlike its counterparts in animals, AP2 was considered a non-central player in clathrin-mediated endocytosis in yeast (Boettner et al., 2012). The experiments described here show that the loss-of-function ap2 σ mutant was impaired in many developmental processes, including embryo patterning, cotyledon organogenesis, vegetative growth and silique length. Among ap2 σ mutant seedlings 19.6% had single, triple or fused cotyledons with associated defective vascular patterns, resembling the phenotypes of auxin-related mutants (Mattsson et al., 2003; Furutani et al., 2004; Péret et al., 2012). More importantly, we found that the internalization of the auxin efflux carrier PIN1 was strongly inhibited and the polar localization of PIN1 on the plasma membrane was disrupted in the absence of AP2 σ. These observations suggest that AP2 might serve as a core adaptor linking clathrin and PIN1, through which PIN1 is internalized via the clathrin-dependent endocytic pathway, and that AP2 σ regulates cotyledon organogenesis by maintaining the proper localization of the auxin transporter.

The molecular and cellular studies presented here demonstrate that the AP2 complex is crucial for endocytosis and plant growth. These findings might help address the potential role of clathrin-dependent endocytosis in the regulation of plasma membrane dynamics and signal transduction in plant cells.

We thank Qihua He (Peking University) for technical assistance with FCCS; Shufeng Sun (the Institute of Biophysics, CAS) for help with TEM; and Jian Xu (National University of Singapore) for kindly providing PIN1-GFP plasmid and DR5::GFP seeds.