Cellular requirements for PIN polar cargo clustering in Arabidopsis thaliana


Auxin is a fundamental plant hormone that plays crucial roles in a plethora of developmental processes (Mockaitis & Estelle, 2008; Grones & Friml, 2015). The key mechanism in auxin action is its directional (polar) transport between cells for its differential distribution in plant tissues. This process underlies a plethora of developmental processes, such as embryonic axis establishment, root and shoot tropisma, meristem activities, and root and shoot organogenesis (Adamowski & Friml, 2015). Polarly localized members of the plant‐specific family of PIN‐FORMED (PIN) auxin transporters regulate both the rate and the directionality of this auxin transport that is essential to connect polarities at the individual cell level and the tissue and organ levels (Wiśniewska et al., 2006; Sauer et al., 2006a; Glanc et al., 2018; Skokan et al., 2019; Mazur et al., 2020; Zhang et al., 2020). In animal, cell polarity is regulated through several conserved factors (Crumbs, Scribble and PAR) that are absent in known plant genomes (Geldner, 2009; Kania et al., 2014), and tight junctions separating the polar domains between neighboring epithelial cells (Nelson & Beitel, 2009). Instead of tight junctions, plants possess a cell wall, a crucial cellular component that provides structural integrity to plant tissues and controls cellular growth and architecture (Wolf et al., 2012a). Mechanical strains exerted on the cell wall are transmitted toward the plasma membrane (PM) through still undiscovered connections. Both PIN polarity and endocytic trafficking can be affected by mechanical or osmotic stresses (Nakayama et al., 2012; Zwiewka et al., 2015) and PIN internalization is also regulated by the cytoskeleton‐linked Rho‐like GTPase (Craddock & Yang, 2012).

How can polarity of the polar cargoes be maintained within the rather fluid PM in plant cells in the absence of diffusion barriers? An experimental and computational simulation revealed that within the PM polar domains, PIN proteins are recruited into nonmobile signal aggregates, called clusters, suggesting that PIN clusters may play a role in polar auxin transport. This phenomenon might be critical for polarity maintenance together with endosome‐guided cargo recycling, super‐polar targeting of PIN proteins to the center of polar domains, and PIN protein retrieval at the lateral cell side by clathrin‐dependent endocytosis (Kleine‐Vehn et al., 2008, Kleine‐Vehn et al., 2011; Glanc et al., 2019; Narasimhan et al., 2020). Clustering, which has been reported to regulate polarity in Schizosaccharomyces pombe (fission yeast), Saccharomyces cerevisiae (baker’s yeast) and Caenorhabditis elegans cells (Dodgson et al., 2013), might be a universal way to regulate protein polarity, and its mechanism may well be conserved. However, the characteristics of PIN clusters and the cellular and genetic factors that regulate clustering of polar cargoes remain unclear.

Genetic and pharmacological interference with the deposition of cellulose, which is the major component of the cell wall, and mechanical disruption of the cell wall increased lateral diffusion and defects in the polar distribution of PIN proteins (Feraru et al., 2011). A connection between microtubule arrangements, the cell wall and PIN1 localization has also been suggested in the shoot apical meristem (Heisler et al., 2010). In addition, the cell wall has been shown to play a crucial role in immobilizing PM proteins and the cellulose deposition pattern in the cell wall to affect the trajectory and speed of PM protein diffusion (Martinière et al., 2012). Although both cell wall and polar cargo clustering are important for lateral diffusion and polarity, their exact roles and mutual relationship remain unclear. Here, we dissected the cellular characteristics of PIN clusters by using different microscopy methods and identified the cellular factors involved in polar cargo clustering, such as lipid kinase pathways, the cytoskeleton, cell wall and cell wall–PM connection.

Materials and Methods

Plant growth conditions

Arabidopsis thaliana (L.) Heyhn. seeds were sterilized by chlorine gas or 75% (v/v) ethanol and sown on plates containing half‐strength Murashige and Skoog (½MS) medium (pH 5.7) with 1% (w/v) sucrose and 0.8% (w/v) agar. After stratification at 4°C for 2 d, plates with seeds were transferred to a growth room at 22°C under a 16 h : 8 h, light : dark photoperiod. The seedlings were grown vertically for 3 d. The transgenic reporter lines were described previously: PIN2::PIN2‐GFP in eir1‐4 (Baster et al., 2012). PIN2::PIN2‐Venus (Leitner et al., 2012), PIN1::PIN1‐GFP (Benková et al., 2003), 35S::GFP‐PIP2a (Cutler et al., 2000), ndr1‐1 mutant, ndr1‐1/NDR1::T7‐NDR1 and ndr1‐1/35S:HA‐NDR1 (Coppinger et al., 2004; Day et al., 2006), XVE>>YFP‐PIP5K1 (Tejos et al., 2014), pip5k1 pip5k2 (Tejos et al., 2014), pROP6>>GFP‐rop6 CA (Poraty‐Gavra et al., 2013), 35S::P30‐GFP (Kim et al., 2005), cesa3je5 (Desprez et al., 2007) and cesa6prc1‐1 (Fagard et al., 2000; Desprez et al., 2007). The cesa3; PIN2‐GFP line was cesa3je5 crossed with PIN2::PIN2‐GFP and the cesa6; PIN2‐GFP line was cesa6prc1‐1 crossed with PIN2::PIN2‐GFP. For the generation of pREM1.2::GFP‐REM1.2 constructs, a 1.5 kb promoter and full‐length REM1.2 was cloned into the Gateway pDNORP4P1R and pDONR221 vectors, respectively, and subsequently cloned into pB7m24GW2 vector using Gateway cloning technology (www.invitrogen.com). The resultant constructs were introduced in Columbia (Col‐0) or individual rem mutants by Agrobacterium‐mediated genetic transformation.

Confocal microscopy

Three days after germination, the Arabidopsis seedlings were mounted in liquid ½MS under a cover glass or under a small piece of growth medium agar in a chamber with a cover glass bottom. In most cases, the meristematic root zone was imaged, but other regions were imaged in some experiments as indicated (Johnson et al., 2020). Confocal images were obtained with a Zeiss LSM 700 using a ×100/NA 1.46 oil objective lens. Green fluorescent protein (GFP) fluorescence was excited at 488 nm (laser power 10%), and emission was collected 493 nm. Data for Supporting Information Videos S1 and S2 were taken with a spinning disk microscope (Andor Spinning Disc System; Andor Technology, Belfast, UK) with an inverted observer (Zeiss) and a ×100/NA 1.4 oil objective. The reconstruction of confocal 3D images and videos were processed in Imaris (v.7.7.4).

Vibratome root cutting

The 3‐d‐old seedlings were imbedded into 5% (w/v) low‐melting analytical agarose (Promega). The agar with roots was cut into cubes and glued onto the blade of Vibratome VT 1200S (Leica Microsystems, Wetzlar, Germany). Sections of 200–400 µm thickness were cut and placed on slides mounted with water for immediate observation under the confocal microscope.

Postembedding immunogold transmission electron microscopy

Root tips of 3‐d‐old seedlings of wild type or chemically treated Arabidopsis seedlings were excised, immersed in 20% (w/v) BSA, and frozen immediately in a high‐pressure freezer (EM PACT; Leica Microsystems). Freeze substitution was carried out in an EM AFS2 (Leica Microsystems). Cells were freeze‐substituted in dry acetone with 0.1% (v/v) glutaraldehyde for 4 d as follows: −90°C for 24 h, 2°C h–1 increase for 15 h, −60°C for 16 h, 2°C h–1 increase for 15 h, and −30°C for 8 h. At −30°C, the carriers were rinsed three times with acetone for 20 min each time. Samples were then slowly warmed to 4°C, stepwise infiltrated over 3 d at 4°C in hard‐grade LR‐white resin (London Resin) and embedded in capsules. Polymerization was done in an EM AFS (Leica Microsystems) with UV illumination over 6 d, starting at 0°C and ending at 37°C. Ultrathin sections of gold interference color were cut with an ultramicrotome (EM UC6; Leica Microsystems) and collected on Formvar‐coated copper mesh grids. All immunolabeling steps were done in a humid chamber at room temperature. Grids were floated upside down on 25 µl aliquots of blocking solution (5% (w/v) BSA, 1% (w/v) fish skin gelatin (FSG) in PBS) for 20 min, followed by washing five times for 5 min each wash (1% (w/v) BSA in PBS). Grids were incubated in a 1 : 300 dilution (1% (w/v) BSA in PBS) of biotin‐conjugated goat anti‐GFP primary antibodies (Rockland 600‐106‐215) for 120 min, followed by washing five times for 5 min each wash (0.1% (w/v) BSA in PBS). The grids were then incubated with a 1 : 10 000 dilution of unconjugated rabbit anti‐biotin (Rockland 100‐4198) bridging antibodies for 30 min, followed by washing five times for 5 min (0.1% (w/v) BSA in PBS). After a final incubation with protein A/10‐nm gold (PAG10nm; Cell Biology, Utrecht University, the Netherlands), grids were sequentially washed twice for 5 min each time with 0.1% (w/v) BSA in PBS, PBS and double‐distilled water. Control experiments consisted of treating sections with bridging antibodies and/or PAG10nm alone. Sections were post‐stained in an automatic EM AC20 contrasting system (Leica Microsystems) for 30 min in uranyl acetate at 20°C and for 7 min in the lead stain at 20°C. Grids were viewed with a transmission electron microscope (JEM1010; Jeol, Tokyo, Japan) operating at 80 kV with the Image Plate Technology from Ditabis (Pforzheim, Germany). For each sample, cluster numbers and distribution were calculated from the analysis of at least 19 images, at least five seedlings and 15 cells that had been analyzed.

SDS‐digested freeze‐fracture replica labeling

Sodium dodecyl sulfate‐digested freeze‐fracture replica labeling (SDS‐FRL) was applied with some modifications from the method described for mammalian tissue samples (Kaufmann et al., 2010; Möbius et al., 2010). Details of the modified method are provided in Methods S1.

Sampling and analysis of SDS‐FRL data

Four to seven replicas were used for quantification of immunolabeling per area of interest that were apical (shoot apex‐facing), lateral and basal (root apex‐facing) domains of PIN2‐Venus and PIP2a‐GFP epidermal cells. Within these areas, profiles were selected at random and electron micrographs were taken at a magnification of ×39 000 to ×93 000. The magnification was verified by a calibration grid. Quantification was done either manually with the item ce software (Olympus Soft Imaging Solutions, Münster, Germany) or semi‐automatically using Fiji and Matlab. The semi‐automatic method was done as follows, in Fiji: black shadows were manually overdrawn with white, and then a threshold was applied manually to highlight gold particles only. The plugin ‘Analyze Particles’ was applied. The obtained list of coordinates was imported into Matlab and particles were assigned to groups based on their maximal distance of 55 nm. Data were expressed as mean ± SD. To compare the density of immunoparticles in different domains and genetic lines, a Mann–Whitney U‐test (P = 0.05) test was applied. Statistical analysis was carried out in Prism (GraphPad, La Jolla, CA, USA). At least 25 cells were analyzed.

Quantification of clusters from confocal microscopy

The clusters are distributed in cell membranes, generally perpendicular to the optical axis of the microscope. Hence, the quality of the images is limited by the axial (Z) resolution of the objective lens used. For confocal microscopy, a ×100/1.46 oil objective lens was used with a Zeiss LSM700 confocal microscope. The obtained pixel sizes were 0.089/0.089/0.313 µm (x/y/z). The axial asymmetry was ignored, assuming that elongations in Z are artificial. As the very small volumes made segmentation problematic, a blob detection algorithm was implemented to find the clusters. For computational convenience, the Maximum Intensity Projections (MIPs) of the axial image sections were used to identify the local maxima and to classify them according to their size. First, the brightest pixels (local maxima) were localized. Each of these local maxima is the absolute maximum of its local 3 × 3 group of pixels. In other words, the eight nearest neighbors around it receive dimmer intensities. The next step is to estimate the corresponding cluster size for each maximum. This is done by dilating the local domains around them successively, and testing if the pixels in these larger domains start to become brighter again. The considered domains have initially 3 × 3 pixels (the considered pixel and its nearest neighbors), then 5 × 5 pixels to incorporate the second neighbors, then 7 × 7 pixels, etc. The area across which no brighter pixels are found defines the estimated size of the clusters. For a radius of n pixels, a 2n + 1 × 2n + 1 pixel domain has to be considered. To analyze large sets of cell membranes, we developed a software program in Matlab (MathWorks, Natick, MA, USA). A batch series involving the steps described above was executed on a series of regions of interest (ROIs) selected interactively by the user. These ROIs might have different sizes, but the surface concentrations are given. We expect to find some correlation of the surface cluster densities for cells of the same type, and some significant difference from one cell type to another.

At small sizes, the cluster detection was not reliable. A simple explanation might be that the numbers were not consistent because the observed structures were not the clusters of interest, the size of which we did not know in advance. At the large extreme, other structures were found instead, such as overlapped sections of the same membrane. Between these boundaries, clusters ranging from 0.44 µm (5 pixels) to 0.80 µm (9 pixels) at the confocal microscope were selected. We considered that any thresholding would bias the analyses, and for this reason, we excluded the use of the actual intensity values for blob detection. The way to do it was to compare intensity differences between neighbors, but not particular intensity values. In this way, the brightest spot of a 9 × 9‐pixel domain will be classified as an 8‐pixel‐diameter cluster, regardless of whether the whole group is brighter or dimmer. The only condition is that the intensities decrease consistently from the maximum away. This was very robust in general, but some false positives might arise if a bright pixel appears in a very dim area, such as the border of another cell in a dark area. For this reason, the projections were checked manually and some cases were discarded.


Three‐ to 4‐d‐old seedlings were immunolocalized using an in situ pro robot (Intavis, Cologne, Germany) according to the described protocol (Sauer et al., 2006b). The primary antibodies were rabbit anti‐PIN2 (Abas et al., 2006) 1 : 1000 and mouse anti‐GFP (Sigma‐Aldrich) 1 : 600, and the secondary antibodies were Cy3 anti‐rabbit (Sigma‐Aldrich) 1 : 600 and Alexa488 anti‐mouse (Invitrogen) 1 : 600.

Lipid‐protein blot overlay assay

The lipid‐protein binding assay was performed as previously described (Tan et al., 2020b). In brief, the recombinant His‐PIN2HL was expressed and purified from Escherichia coli. PIP strips (P‐6001, Echelon Bioscience, Salt Lake City, UT, USA) membrane was blocked in the blocking buffer with 3% BSA in 1× TBST for 1 h. Purified His‐PIN2HL (20 µl in 10 ml 1× TBST) was incubated with the membrane for 2 h. The membrane was washed three times for 5 min with 1× TBST, and then incubated with anti‐His HRP‐conjugated antibody (dilution 1 : 4000; Agrisera, Vännäs, Sweden) for 2 h at room temperature. After washing three times for 5 min with 1× TBST, the bound protein was detected using SuperSignal western detection reagents (Thermo Fisher Scientific, Waltham, MA, USA) in an Amersham 600RGB molecular imaging system (GE Healthcare, Little Chalfont, UK).

Auxin transport assay

The auxin transport assay was performed in etiolated hypocotyls as previously described (Lewis & Muday, 2009; Tan et al., 2020a). First, 6‐d‐old etiolated seedlings were transferred to the MS medium plates. For 1‐N‐naphthylphthalamic acid (NPA) treatment, 6‐d‐old etiolated Col‐0 seedlings were transferred to the MS medium plates supplemented with 5 µM NPA. MS medium/1.25% agar droplets with 500 µM 3H‐IAA were prepared and were placed on the upper part of decapitated hypocotyls of 6‐d‐old seedlings, one droplet per hypocotyl. Fifteen hypocotyls were regarded as one replicate, with three replicates per genotype. After incubation in the dark for 6 h, upperparts with 3H‐IAA droplet and roots of hypocotyls were cut off, and the rest of the hypocotyls were collected and frozen in liquid nitrogen, and then homogenized in the 1 ml scintillation solution. After incubation in the scintillation solution overnight, the samples were evaluated with a scintillation counter (Hidex 300XL). The sample with only 1 ml scintillation solution was also measured as a background control.

Chemical treatments

PIN2‐Venus seedlings were treated with 30 µM wortmannin (WM) for 2 h, 0.8% (v/v) 1‐butanol or 2‐butanol for 3 h (Li & Xue, 2007), 10 µM U‐73343 or U‐73122 for 2 h (Chu et al., 2016), 40 µM oryzalin for 1–3 h (with similar results), 20 µM latrunculin B for 2 h, 0.1% (w/v) macerozyme for 5 min, 5% (w/v) cellulase for 5 min, 5 nM isoxaben for 3 h, 0.1% (v/v) driselase for 15 min, and 0.5 M mannitol (Feraru et al., 2011) and 100 µM AlCl3 for 1–3 h (Yang et al., 2014). At least 10 roots, five cells per root were analyzed for each treatment. For the observation of the localization of the estrogen receptor‐based chemical‐inducible XVE>PIP5K1‐YFP expression, 3‐d‐old seedlings were treated with 2.5 µM estradiol in liquid plant growth medium overnight before observation. For the epigallocatechin gallate (EGCG) treatment plants were sprayed with 50 µM EGCG or H2O as control. SDS‐FRL was performed 24 h after treatment.

Accession numbers

Sequence data from this article can be found in the Arabidopsis Genome Initiative or GenBank/EMBL databases under the following accession numbers: PIN1, At1g73590; PIN2, At5g57090; PIP2a, At3G53420; PIP5K1, At1g21980; PIP5K2, At1g77740; KTN1, At1g80350; TOR1/SPR2, At4g27060; TOR2/TUA4, At1g04820; CESA3, At5g05170; CESA6, At5g64740; and NDR1, At3g20600.


Clustering of polar cargos observed by confocal microscopy

In A. thaliana, the auxin transporters PIN1 and PIN2 are polarly localized proteins that form clusters at the PM (Kleine‐Vehn et al., 2011). We examined PIN2‐Venus clusters (Leitner et al., 2012) in three dimensions and compared them to the nonpolar membrane marker PIP2a‐GFP (Plasma Membrane Intrinsic Protein 2a) (Cutler et al., 2000) as a control (Fig. 1a,b,e). We selected the nonpolar PM aquaporin PIP2a based on its localization and functional differences from PINs: polarly localized PINs are located in different membrane microdomains from PIP2a and PIN directionally transports auxin, while PIP2a is internalized upon exposure to salt stress conditions (Chevalier & Chaumont, 2014). Video S1 shows the distribution of PIN2‐GFP in a 3D rotation of a typical image stack. Live‐cell imaging showed that the PIN2‐GFP clustering appears to be stable over 13 min while the root tip grows through the field of view (Video S2). Due to the limited axial (Z) resolution of the objective lens in light microscopes, clusters appear longer in the z‐direction. To show that this is a purely optical artefact and not caused by the shape of clusters, we visualized clusters of PIN2 and PIP2a in vibratome‐cut cross‐sections of the meristematic zone of the root tip (Fig. 1b). However, all quantifications of clusters were done in living samples and showed pronounced clusters in PIN2‐Venus when compared to the nonpolar marker PIP2a‐GFP or ROP6‐GFP (Fig. 1e). To quantify the clusters from Z‐stack confocal images, we developed software in the Matlab environment (Fig. 1c,d). Using this software, we quantified cluster densities in different zones and at various developmental stages and found similar PIN2 cluster densities throughout the root cap, transition and meristem zones (Fig. S1). Clustering of the polar‐localized PIN2 appeared to be specific to these polar PM proteins because the nonpolar PIP2a marker does not form comparable clusters.

Visualization and quantification of Arabidopsis PIN clusters by confocal microscopy. (a) Three‐dimensional view of clustering of the nonpolar marker PIP2a‐GFP (left) and the polarly localized PIN2‐Venus (right). Lower panel shows enlargements of the yellow boxes from upper panel. Bars, 10 µm. (b) Vibratome‐cut cross‐section of the root tip meristematic zone of PIP2a‐GFP (left) and PIN2‐Venus (left). Bars, 10 µm. Lower panel shows enlargements of the yellow boxes from upper panel. Bars, 10 µm. (c) Cluster quantification in the Matlab program. Cell membranes are marked and numbered. (d) Enlargement of cell membrane number 5 (white dashed box). Each cropped substack was projected (maximum intensity) along the x/z direction. To identify clusters local maxima were found and grouped using Matlab (see detailed description in the Materials and Methods section). Brightness thresholding reveals the area to calculate the actual density of clusters per area. (e) Quantitative analysis of the number of clusters with diameter sizes from 0.45 to 0.8 µm per µm2 in GFP‐ROP6 (0.03 ± 0.02), PIP2a‐GFP (0.06 ± 0.04) and PIN2‐Venus (0.19 ± 0.11). Values are means ± SD from at least 10 cells. Mann–Whitney U‐test: ***, P 
Read More

Show More

Related Articles

Leave a Reply

Your email address will not be published.

Back to top button