Cell Geometry Guides the Dynamic Targeting of Apoplastic GPI-Linked Lipid Transfer Protein to Cell Wall Elements and Cell Borders in Arabidopsis thaliana

During cellular morphogenesis, changes in cell shape and cell junction topology are fundamental to normal tissue and organ development. Here we show that apoplastic Glycophosphatidylinositol (GPI)-anchored Lipid Transfer Protein (LTPG) is excluded from cell junctions and flat wall regions, and passively accumulates around their borders in the epidermal cells of Arabidopsis thaliana. Beginning with intense accumulation beneath highly curved cell junction borders, this enrichment is gradually lost as cells become more bulbous during their differentiation. In fully mature epidermal cells, YFP-LTPG often shows a fibrous cellulose microfibril-like pattern within the bulging outer faces. Physical contact between a flat glass surface and bulbous cell surface induces rapid and reversible evacuation from contact sites and accumulation to the curved wall regions surrounding the contact borders. Thus, LTPG distribution is dynamic, responding to changes in cell shape and wall curvature during cell growth and differentiation. We hypothesize that this geometry-based mechanism guides wax-carrying LTPG to functional sites, where it may act to “seal” the vulnerable border surrounding cell-cell junctions and assist in cell wall fortification and cuticular wax deposition.


Introduction
Epidermal cells demarcate the physical boundary between organism and environment. In aerial plant tissues, the epidermis serves to minimize water loss, restrict pathogen invasion, and facilitate organ growth. Starting as a uniform sheet of small boxy cells in young organs, epidermal cells expand in highly anisotropic patterns as the organ grows, creating a complex topology of epidermal cell shape and intercellular contacts. Anticlinal walls demarcate the areas of contact between epidermal cells (i.e. junctions), and are thus crucial components in epidermal functioning. Considering the epidermis as a structural composite akin to a brick wall, anticlinal wall junctions represent the mortar, and are thus structurally prone to mechanical stresses [1]. For this same reason, anticlinal wall junctions present the easiest route for an invading fungal hypha [2] and often separate and form holes in the leaf epidermises of Arabidopsis mutants with impaired cell adhesion [3]. The leaf epidermal cells of many species have wavy anticlinal walls, which are believed to enhance the structural integrity of the organ. Similar wavy boundaries have been observed in the nacre of seashells, where they assist in dissipating applied load forces over a larger area, thereby reducing the formation and propagation of cracks [4].
In contrast to epidermal anticlinal wall junctions, the inner periclinal face of an epidermal cell is adjoined to the underlying cells of the ground tissues. This face may experience varying degrees of separation from the underlying cells, as in the case of the highly branched cells of the leaf spongy mesophyll, which gradually lose contact with their epidermal neighbors as they round up and expand to form air spaces [5]. Little is known of the mechanisms mediating this separation although in general, when cell wall thickenings are opposite one another between adjacent cells, the cells will separate between the thickenings [6]. This can be seen during stomatal pore formation [7], and during mesophyll cell differentiation [8,9].
Conversely, when wall thickenings alternate, as in leaf pavement epidermal cells, the cell junctions do not separate and the junctions become wavy as they expand [10,11].The border that surrounds every cell junction is particularly crucial both structurally and biochemically. The outer borders of epidermal cell junctions typically contain thicker cell walls, and are often fortified by the presence of extra cuticle [12,13]. Similar fortification is seen around junctions with and between cells of the ground tissue, such as collenchyma. The chemistry and mechanics of junction borders plays a role in determining intercellular adhesion, however, how their specific geometries regulates the distribution of apoplastic molecules is unknown. The chemistry and mechanics of junction borders play a role in determining intercellular adhesion [14] but how their specific geometries regulate the distribution of apoplastic molecules is unknown.
The current study was brought about by the observation that a glycophosphatidylinositol (GPI)-linked lipid transfer protein (LTPG) accumulates specifically at junctional borders. This junctional accumulation depends on the geometry of the border. As borders round up during cell expansion, the junction boarder accumulation of LTPG dissipates and the protein spreads out to the free region between cells. In mature epidermal cells, we often observed LTPG distributed in a pattern reminiscent of cellulose microfibrils. Using a YFPtagged LTPG, we found that YFP-LTPG resides both in the apoplastic space between the plasma membrane and cell wall, as well as in the intercellular fluids prior to air filling. When we created artificial contact between a cell and coverslip, the YFP-LTPG fluorescence rapidly evacuated the region of contact and accumulated around its borders. Based on these observations, we propose that the varied YFP-LTPG accumulation patterns manifest from a simple mechanism, wherein YFP-LTPG distribution responds to geometrical changes within cell boundaries and becomes excluded from any sites of physical contact by a passive flow mechanism.

YFP-LTPG forms a striated pattern and accumulates over anticlinal walls
To observe the cellular distribution of LTPG, we imaged a fusion between citrine yellow fluorescent protein (YFP) and LTPG, driven by the native LTPG promoter in a complemented ltpg-1 mutant background ( pro LTPG:YFP-LTPG) [15], and examined the distribution in multiple cell types using a confocal microscope.
In mature cotyledon and leaf epidermal cells, YFP-LTPG often showed a striated distribution pattern on the outer periclinal cell faces, resembling the pattern of cellulose microfibrils found in this cell type [6] (Figure 1A, B). Incubation in 20 µM of the cellulose synthesis inhibitor 2, 6dichlorobenzene (DCB) for 2 hours was sufficient to abolish the striated YFP-LTPG pattern, suggesting that this pattern is dependent on cellulose microfibril formation ( Figure S1). In the small, boxy cells of unexpanded young leaves/cotyledons, YFP-LTPG fluorescence outlined the outer periclinal faces, accumulating directly above anticlinal wall junctions in a caulk-like pattern (hereafter referred to as the supra-anticlinal domain; See schematic) ( Figure 1C-H). On the outer periclinal face of these cells, fluorescence was typically diffuse, showing no filaments, with the exception of occasional fibers extending down the anticlinal faces underlying the supra-anticlinal domain ( Figure 1D, E). Supra-anticlinal fluorescence is clearly shown with orthogonal sections through confocal Z-series ( Figure 1F), as well as optical confocal sections through the organ midplane ( Figure 1G). Accumulation at the supra-anticlinal domain was not observed with plasma membrane control markers GFP-PIP2a, FM4-64 dye; or with cell wall marker PI ( Figure S2).
In terms of a developmental time course, newly-established post-cytokinetic anticlinal walls show relatively little YFP-LTPG fluorescence and lack supra-anticlinal YFP-LTPG accumulation ( Figure 1H, brackets; See Figure S3 for full developmental series). As the wall develops, supra-anticlinal fluorescence gradually accumulates in the middle of the new wall, forecasting the site of future lobe formation. As the new wall expands and becomes wavy, supra-anticlinal YFP-LTPG fluorescence becomes enriched at the curved regions ( Figure  1I).

Striated and Peripheral YFP-LTPG Populations Are Apoplastic
The finding that intrinsic plasma membrane proteins are not co-distributed with LTPG suggested that YFP-LTPG might be accumulating in the apoplast. To test this, we engineered a secreted apoplastic marker consisting of an ER secretion signal sequence linked to YFP (sec-YFP). Like YFP-LTPG, sec-YFP was enriched in the supra-anticlinal domain ( Figure  2A, B). Unlike YFP-LTPG, however, secYFP showed no detectable striated patterning. To confirm that the supraanticlinal LTPG domain is indeed apoplastic, YFP-LTPGexpressing leaves were plasmolyzed by treatment with mannitol (500 mM, 10-15 minutes). Upon plasmolysis, both the striations and supra-anticlinal distributions were lost, and YFP-LTPG filled in the enlarged apoplastic space between the plasma membrane and wall ( Figure 2C). In contrast, the control marker GFP-PIP2a labeled the plasma membrane and the Hechtian strands connecting to the wall during plasmolysis ( Figure 2D). Fluorescence patterns were restored for YFP-LTPG and GFP-PIP2a upon deplasmolysis in water, typically within 5-15 minutes ( Figure 2C, D). Time-lapse imaging of YFP-LTPG and FM4-64 together show the absence of YFP-LTPG in the plasma membrane, and deplasmolysis at a wall lobe shows that as the expanding protoplast fills up, YFP-LTPG fluorescence concentrates in the concave side of the lobe ( Figure 2E). Figure 2F shows that the plasmolysis-induced apoplastic YFP-LTPG filling is accompanied by loss of supra-anticlinal enrichment, suggesting that normally, the exclusion of LTPG from between adjacent epidermal cells is generated by turgor pressure. Indeed, rinsing plasmolysed cells in distilled water was sufficient to rapidly restore supra-anticlinal fluorescence ( Figure 2G). Taken together, these observations show that both the striated and supra-anticlinal YFP-LTPG fluorescence distributions are apoplastic, and require sufficient turgor pressure to be maintained.

Cell-cell junction topology affects apoplastic YFP-LTPG distribution asymmetries
We next sought to investigate the asymmetric nature of YFP-LTPG distribution. We found that the distribution of YFP-LTPG varied depending on the topology of junctions between neighboring cells. In general, YFP-LTPG fluorescence is excluded from regions where cells contact one another (i.e. cell junctions), and accumulates around the borders of these junctions. This is easily demonstrated in cases where the internal periclinal face of epidermal cells partially contacts the underlying mesophyll/cortex cells, which are rounded and only in partial contact with the inner face of the epidermis ( Figure  3A). In figure 3A, two focal planes from a confocal Z-series are pseudo-colored and overlaid, such that the relationship between the epidermal cells and the underlying mesophyll cells can be visualized. The regions that are in contact with the underlying mesophyll cells appear as darkened 'holes' with bright boarders surrounding them. In cases where the underlying cells are in full contact with the epidermis, such as those shown in Figures 1F-H, contact site fluorescence exclusion is complete on the internal periclinal faces, thereby creating a polarity in distribution to the outer cell faces.
Since the supra-anticlinal fluorescence also surrounds cellcell junctions, we hypothesized that applied contact on the outer epidermal face will also generate contact exclusion of YFP-LTPG. To test this, we created physical contact between the slide-mounted specimen and the glass coverslip by slowly wicking out the water between the slide and coverslip during imaging. Using this method, we observed distinct exclusion of fluorescence within regions where cells make contact with the coverslip (as assessed by the partial flattening of cells along a tangential plane) ( Figure 3B). As with cell-cell junctions, YFP-LTPG fluorescence often accumulated around the boundary of coverslip contact zones ( Figure 3C). In mature/fully expanded leaf epidermal cells, the striated fluorescence patterning was also excluded at sites of coverslip contact ( Figure 3D). Contact exclusion was not observed in any form with the control markers GFP-PIP2 and FM4-64; and also not for cell walls labelled with propidium iodide ( Figure S4).
To observe the dynamics of contact clearing, we used 4D imaging while altering the degree of coverslip-cell contact over time. Figure 3E shows an example of this, wherein cellcoverslip contact is initiated and gradually increased over the time course, which lasts roughly 5 minutes (See also Movie S1). Contact clearings moved from the cell center (i.e. the highest point) to the periphery over time. Clearing dynamics were remarkably rapid and reversible, showing both clearing and filling within the highest temporal resolution we were able to achieve, which was typically 30-60 seconds between acquisitions. This directional opening/closing is consistent with a malleable convex surface gradually coming into contact with a hard flat surface. In the last time point of the series, the outer faces of several cells were flattened to the plane of the anticlinal edges. Here, the fluorescence was excluded from the entirety of the outer cell face, being thus relegated to the supraanticlinal domain.
These data suggest that the absence of YFP-LTPG fluorescence at cell junctions is a manifestation of a more general contact exclusion mechanism.

Cell geometry affects apoplastic YFP-LTPG distribution asymmetries
The outer periclinal cell face of an epidermal cell is unique in that it is free from contact with other cells throughout their entire lifespan. Interestingly, we found that the degree of supraanticlinal enrichment varies with the cellular geometry of the outer periclinal surface. Epidermal cells start small and cuboid, with flat external faces and sharp edges. As they mature, this external surface gradually bulges outward, and the cell edges become round [16]. We found that as the outer cell face bulges out, YFP-LTPG fluorescence extends across its entire surface, while the supra-anticlinal fluorescence decreases ( Figure 4A). In cases of highly bulbous cells, the supra-anticlinal enrichment is strongly reduced or lost. This distributional change is exacerbated in mutants that have highly bulbous cells, such as mor1-1 [17](after 48 hours at 29° C restrictive temperature) and clasp-1 [18,19]( Figure 4B).
The correlation between cell surface flatness and supraanticlinal YFP-LTPG fluorescence lead us to hypothesize that the fluorescing borders surrounding contact-clear zones result from the flattening generated within contact zones. To test this, we asked whether complete flattening of the outer periclinal cell surface of bulgy cells can redistribute periclinal YFP-LTPG all the way to the anticlinal walls, thereby mimicking the distribution pattern in young, flat cells. Indeed, flattening the periclinal face in bulgy cells with a coverslip redistributed the periclinal fluorescence to the supra-anticlinal domain ( Figure  4C, D). Notably, contact-induced redistribution to the periphery often forced the apoplastic YFP-LTPG down into the anticlinal walls, where it localized to filamentous striations ( Figure 4E).

Filling of sub-epidermal intercellular spaces by YFP-LTPG
In addition to its apoplastic accumulation patterns, YFP-LTPG often accumulated within the extracellular spaces below the epidermis ( Figure 5A, B), in contrast to the plasma membrane marker GFP-PIP2a, which labeled only the plasma membrane ( Figure 5C). The intercellular space filling was observed typically for young leaf cells, when the spaces still contain fluids prior to air filling [5]. Intercellular fluorescence was observed in roots as well, where the expression of YFP-LTPG was restricted to elongating cells within the atrichoblast (non-hair forming) cell files ( Figure 5D). YFP-LTPG fluorescence often also appeared beneath non-expressing cells, indicating movement of the extracellular protein. Anti-YFP western blots showed the presence of full-length YFP-LTPG, excluding the possibility that the extracellular fluorescence is free YFP, cleaved from the YFP-LTPG ( Figure 5E).

Balloon in a box model
Based on our data, we propose a hypothetical model for YFP-LTPG distribution, which is nicely illustrated using the analogy of a balloon in a box ( Figure 6). When the balloon (protoplast) inflates, it gradually comes into contact with the box walls (cell wall) and fills in the air spaces (analogous to liquid-filled apoplast in planta). As the expanding balloon comes into contact with the walls of a cube-shaped box, the last spaces to be filled by the balloon are the edges and corners. If a particular box wall is allowed to selectively yield to the pressure of the balloon by bulging out, the pressure between the balloon and wall gradually spreads out more uniformly over the entire face. At the same time, the edges bordering the bulging wall become more rounded and the protoplast then uniformly fills in those spaces. This predicts that variations in the uniformity of air spaces is generated ultimately by variations in wall curvature throughout the box, with perfect uniformity appearing only in a spherical structure. Factors secondary to the fundamental curvature factor include the volume occupied by the balloon, and the mechanical strength of the box.
Applying the balloon in a box model to epidermal cells accurately predicts the observed YFP-LTPG distribution patterns. As a cell grows and changes shape (i.e. changes wall curvature), the topology of cell junctions and junction borders surrounding it change as well. Regions of the wall that are free from contact with neighboring cells (such as the outer periclinal region) are able to bend out in response to the stress induced by the turgor pressure of the protoplast. Regions of walls that are in contact with other cells (such as anticlinal wall junctions) are unable to bulge in response to turgor stress due to the counter-pressure from the neighboring cell. Similarly, differentially thickened anticlinal walls, such as those in leaf pavement cells, will yield less to turgor stress at the thickened regions. These variations in wall yield-ability generate increasingly complex patterns of wall curvatures. When we flattened a bulbous (low curvature) cell face, YFP-LTPG accumulated within the newly formed regions of higher curvature bounding the contact site. As shown in Figure 3, this mimics the accumulation around epidermal-mesophyll junction regions. Fundamentally, apoplastic YFP-LTPG distribution patterns may result from local variations in wall curvature; accumulating passively to highly curved wall regions. Biologically, passive accumulation of apoplastic YFP-LTPG beneath areas of high wall curvature may provide an efficient way to target and increase its concentration at its site of action. For example, the observed exclusion of protein from anticlinal epidermal walls passively enriches apoplastic lipid-bound LTPG along the top and bottom edges of the anticlinal walls, thus 'sealing the cracks' with lipidic compounds to prevent pathogen attack and non-stomatal water loss [20,21]. This is consistent with the accumulation of cuticle in these regions. When cell surfaces bulge out YFP-LTPG redistributes to the striated elements within this region, where it presumably would also provide lipidic compounds to those regions to fortify the wall there. This selective sealing of cell-cell junctions is reminiscent of the casparian strip in root endodermal cells, which restricts apoplastic water uptake into the root vasculature.

Balloon in a Leaky Box?
How LTPG moves through the cell wall into the intercellular spaces is unclear. We speculate that pressure from the turgid epidermal protoplasts may be sufficient to squeeze the components out through the curved junctional borders of a porous primary cell wall, analogous to a balloon in a leaky box ( Figure 6B). In this hypothetical model, the differential permeability of an anisoptropically expanding cell wall would allow for differential passage through the cell wall. For example, the wall within the supra-anticlinal region is known to be pectin-enriched and is thought to regulate wall porosity [22].

LTPG trafficking
We show here that YFP-LTPG has a prominent apoplastic and intercellular distribution in leaves and hypocotyls. In inflorescence stem epidermal cells, YFP-LTPG has previously been shown to localize to the plasma membrane [15]. This apparent discrepancy can be explained by the fact that LTPG expression is restricted to the earliest stages of development in developing leaf and hypocotyl cells, whereas it continues until quite late in development in the epidermal cells of inflorescence stems. In young leaf and hypocotyl cells, YFP-LTPG has already trafficked to its final destination in the apoplast. In contrast, the continued production of YFP-LTPG in inflorescence stem epidermal cells generates prominent localization to cytoplasmic compartments and the plasma. Interestingly, supra-anticlinal and subepidermal distribution of YFP-LTPG is apparent in Figure 5 of the DeBono et al publication although neither was mentioned [15]. Similarly, Figure 5B in our study shows intercellular YFP-LTPG in stems.

Relationship to cell wall microfibrils
When responding to turgor pressure, the outer face experiences no counter-pressure from neighboring cells, and is thus free to bulge outwardly during its expansion and development. This is reflected in the organization and composition of the cellulose microfibrils in the outer wall, which can be relatively disorganized compared to the strongly aligned ones within the inner wall [23,24]. The tendency for YFP-LTPG to accumulate non-uniformly on the outer periclinal surface in distinct striation patterns suggests a preference for wall regions in which recently synthesized cellulose microfibrils are deposited. The rapid loss of this pattern after treatment with the cellulose synthesis inhibitor DCB supports the idea that LTPG is indeed distributed along pathways delineated by cellulose microfibrils and other tension-bearing wall materials. Initially concentrated in the highly curved regions of the supra-anticlinal domain in young leaf epidermal cells, the YFP-LTPG striation pattern develops as cells become more outwardly bulging, and the curvature within the supra-anticlinal region diminishes. The apparent association of LTPG with cellulose microfibrils could either be through confinement of LTPG to sites where tensile stress is lowest or through a chemical affinity of LTPG to microfibrillar components. The low tension-seeking mechanism seems more plausible since DCB treatment will specifically prevent the formation of new tension-bearing elements but will not remove preexisting cellulose microfibrils from the wall. Interestingly, secreted YFP, which of course is foreign to plant cell walls, did not accumulate in striation patterns on the periclinal surface, which indicates that the apparent association of YFP-LTPG is a functional attribute of LTPG and not of apoplastic proteins in general.

Possible functional relationship with other LTPs
The group of LTPs with GPI anchors has been classified as a subfamily of Non-Specific LTPs (NsLTPs). These subfamilies include sporopollenin, suberin and wax-related LTPGs; with LTPG1 belonging to the wax subgroup [25]. DeBono et al, 2009 showed LTPG functions in wax transport in stems [15]. Our findings here of LTPG1 expression in additional organs suggest additional functions. For example, atrichoblastic cells have been shown to contain possible wax crystals on their surface [26], which would give an explanation for the presence of LTPG in these cells. The wider range of LTPG expression also raises the possibility that LTPG1 can work with other LTPGs. This claim is substantiated by recent work on LTPG2, which has been identified and characterized in two studies [15,27], and appears to have a partially overlapping role with LTPG1 [27].

Plant materials and growth conditions
Arabidopsis thaliana Columbia ecotype plants were grown in continuous light conditions on vertical agar plates containing either Hoagland's or ½ MS medium. Young expanding cotyledon cells were imaged at 3-4 days. Leaves were imaged prior to opening. . Plants were transferred to soil when roots became 1-2 cm long. Hypocotyls were imaged 7-10 days after radicle emergence. Images of stems were acquired from the area within 1 cm of the floral apex unless indicated otherwise.
In order to select for plants expressing sec-YFP, each bearing Basta® resistance, plants were sprayed 23 mg L-1 Basta® in dH2O (Bayer Corporation) every two days after seedling emergence for two weeks.

Molecular biology and construct design
All oligonucleotides used in this study are listed in Table 1. The constructs sec-YFP is cloned in the vector pNOS (DeBono and Rieseberg, unpublished). The vector pNOS contains the nopaline synthase transcription terminator amplified from vector pMDC32 [28] with H7 and H8 and subcloned into the XbaI and SacI restriction sites of pGreenII0229 [29].
To clone sec-YFP, the sequence corresponding to the functional LTPG promoter, amino terminal signal sequence and citrine-YFP was amplified from proLTPG:YFP-LTP [15] using P41 and P44. This amplicon contained a 5' EcoRI and 3' SpeI restriction sites and a stop codon. The P41-P44 amplicon was cloned into the EcoRI and XbaI restriction sites of pNOS.

Microscopy and image analysis
Images were acquired with three microscopes. The scanning confocal microscopes used were a Zeiss Pascal LSM5 (zeiss.com) and an Olympus Fluoview FV1000MPE (olympusfluoview.com). The third microscope was an UltraView VoX spinning disk microscope (perkinelmer.com). FM4-64 and propidium iodide stains were applied as described previously [15]. Images were processed using ImageJ software (http://

Fluorescence Quantification
For quantification of polarity along the anticlinal walls, intensity profiles were generated from lines drawn along anticlinal walls from orthogonal views of confocal Z-series. Fluorescence intensity within large 3D confocal samples is known to attenuate with increasing depth and thus often requires correction [32,33]. In the shallow depths of the epidermal cells examined here however, little axial tissue-depth fluorescence attenuation was observed using controls (GFP-PIP2a, FM4-64, and PI), as can be seen in Figure S2. Moreover, orthogonal reconstructions produced images that were indistinguishable from optical midplane slices, which have no Z-axis decay, and are routinely used for showing protein polarities [34]. The small axial decay that we did observe with GFP-PIP2a was quantified and used as a standard to normalize YFP-LTPG measurements (Fig. S2). Ratios of outer to inner GFP-PIP2a anticlinal wall fluorescence intensities were taken from line profiles and averaged, resulting in an outer:inner ratio of 1.1, termed the decay coefficient (α). Similar decay rates were observed with FM4-64 and PI (See Fig. S2). Using α = 1.1, YFP-LTPG profile measurements were divided by 1.1 to obtain normalized measurements

Plasmolysis and coverslip experiments
For plasmolysis, cotyledons from 5-8 day seedlings were mounted onto flow cells, which consisted of two thin strips of vacuum grease at the middle edges running parallel to the slide length. Cotyledons or whole seedlings were mounted in water under a cover slip and imaged prior to addition of mannitol. Mannitol (500 mM) was then pipetted into the right side of the flow cell, while simultaneously wicking with filter paper on the left side to draw in the solution. Time-lapse imaging was then performed to observe the plasmolysis process. Rinsing in water was then performed in the same manner subsequent to plasmolysis. Following deplasmolysis, cells returned to preplasmolysis appearance, showing robust cytoplasmic streaming and normal fluorescence patterns, indicating our treatment regimens did not harm cell viability after deplasmolysis. Even partial plasmolysis (500 mM mannitol, 4 minutes) was sufficient to remove anticlinal YFP-LTPG fluorescence enrichment.
For generation of coverslip contact with the specimen, the same flow cell method was used. As solution is slowly wicked out, the coverslip will gradually pull down toward the specimen. Contact with the specimen occurs well before any drying of the specimen, which remains submerged during the entire experiment. Flowing fresh solution in following flattening rapidly lifts the coverslip off the specimen.