Cell Wall Polysaccharides are Mislocalized to the Vacuole in echidna Mutants

During cell wall biosynthesis, the Golgi apparatus is the plat-form for cell wall matrix biosynthesis and the site of packa-ging, of both matrix polysaccharides and proteins, into secretory vesicles with the correct targeting information. The objective of this study was to dissect the post-Golgi trafﬁcking of cell wall polysaccharides using echidna as a vesicle trafﬁc mutant of Arabidopsis thaliana and the pectin-secreting cells of the seed coat as a model system. ECHIDNA encodes a trans -Golgi network (TGN)-localized protein, which was previously shown to be required for proper structure and function of the secretory pathway. In echidna mutants, some cell wall matrix polysaccharides accumulate inside cells, rather than being secreted to the apoplast. In this study, live cell imaging of ﬂuorescent protein markers as well as transmission electron microscopy (TEM)/immunoTEM of cryoﬁxed seed coat cells were used to examine the consequences of TGN disorganization in echidna mutants under conditions of high polysaccharide production and secretion. While in wild-type seed coat cells, pectin is secreted to the apical surface, in echidna , polysaccharides accumulate in post-Golgi vesicles, the central lytic vacuole and endoplasmic reticulum-derived bodies. In contrast, proteins were partially mistargeted to internal multilamellar membranes in echidna . These results suggest that while secretion of both cell wall polysaccharides and proteins at the TGN requires ECHIDNA, different vesicle trafﬁcking components may mediate downstream events in their secretion from the TGN. NSF ( N -ethylmaleimide-sensitive factor) attachment protein receptor; Tris-buffered trans -Golgi network; tonoplast intrinsic protein; TVP23, Tlg2p-vesicle 23; YFP, yellow ﬂuorescent protein; YIP, YPT/RAB-interacting protein. as well as mucilage accumulation in the endoplasmic reticulum (ER) and the vacuole. Membrane proteins targeted to the PM and the tonoplast (including ABCG11, PIP2, g -TIP and SYP-61) are partially mislocalized to multilamellar bodies (MLBs) that do not contain any mislocalized polysaccharides.


Introduction
Cell wall secretion is essential for plant cell expansion and growth. Primary cell walls consist of about 20-30% cellulose embedded in a matrix of soluble hemicellulosic and pectic polysaccharides with small amounts of structural and enzymatic proteins (Sandhu et al. 2009, Carpita 2011. While cellulose microfibrils are synthesized by plasma membrane-localized cellulose synthase enzymes, pectins and hemicelluloses are synthesized by Golgi-localized enzymes and secreted to the plasma membrane via secretory vesicles. However, very little is known about the secretion of the soluble polysaccharide matrix of the cell wall (Worden et al. 2012). One of the key limitations in studying soluble cell wall polysaccharides is that unlike proteins, polysaccharides cannot be fluorescently tagged in live cells to follow their biosynthesis and trafficking. Therefore, the analysis of polysaccharide trafficking requires the use of high quality cellular preservation and antibodies specific for polysaccharide epitopes of interest , or click chemistry (Anderson et al. 2012).
The importance of efficient and targeted post-Golgi vesicle traffic in cell wall secretion has been documented sporadically, since most studies of post-Golgi trafficking have been focused on protein secretion. However, several lines of evidence suggest that chemical or genetic perturbation of the Golgi apparatus and post-Golgi compartments has dramatic effects on cell wall polysaccharide export. Both monensin (a proton ionophore; Boss et al. 1984) and concanamycin A (a V-type H + -ATPase inhibitor; Huss et al. 2002) upset the proton gradient of the secretory pathway, leading to morphological defects in endomembrane compartments and a decrease in protein secretion , Robinson et al. 2004, Dettmer et al. 2006. Interestingly, although monensin causes intracellular accumulation of secreted proteins and hemicelluloses, which are modified in the trans-Golgi/trans-Golgi network (TGN), it did not affect secretion of pectins (Zhang et al. 1996), implying that several routes of post-Golgi trafficking to the plasma membrane and cell wall may exist.
More recently, molecular genetic studies have corroborated these inhibitor studies and identified several specific targets of these inhibitors. A screen employing secreted green fluorescent protein (secGFP), which does not significantly fluoresce in the apoplast but accumulates in the endomembrane system of secretion mutants, has revealed several Arabidopsis genes required for protein secretion and cell wall synthesis (Zheng et al. 2004). ECHIDNA (ECH) is a protein of unknown function that, based on homology and conserved function between yeast and plants, could act in conjunction with the SYP61-SYP41-VTI12-SNARE [soluble NSF (N-ethylmaleimide-sensitive factor) attachment protein receptor] complex to achieve proper targeting of post-Golgi vesicles (Stein et al. 2009, Gendre et al. 2011. Recently, two YPT/RAB-interacting proteins (YIPs), YIP4a and YIP4b, have been identified as ECH interactors, implying that ECH/YIP4a/YIP4b may recruit RABs for post-Golgi vesicle targeting (Gendre et al. 2013). In echidna and yip4a yip4b mutants, Golgi function is relatively normal, but TGN morphology and post-Golgi trafficking are defective (Gendre et al. 2011, Gendre et al. 2013. echidna mutants also accumulate secGFP and display growth defects (Gendre et al. 2011). The most striking defect of the echidna mutant phenotype was disruption of cell wall polysaccharide secretion, resulting in accumulation of pectins in the center of cells that were actively secreting cell wall material, such as hypocotyls and seed coat epidermal cells (Gendre et al. 2013).
Here, a high-resolution study of the echidna mutant phenotype was undertaken to determine the location of these polysaccharide accumulations, to characterize the endomembrane defects contributing to this phenotype and to compare polysaccharide and protein traffic. Loss of ECH in seed coat mucilage secretory cells, which are highly specialized for soluble polysaccharide secretion, led to a large accumulation of pectins in the lytic vacuole as marked by g-tonoplast intrinsic protein (TIP)-GFP. Alterations in the endomembrane system in echidna included swellings of the endoplasmic reticulum (ER) that contained both pectin and ER markers. In contrast, protein markers for the TGN and plasma membrane-localized proteins were partially mistargeted to internal multilamellar membranes in echidna mutants. These data indicate that ECH is required for proper targeting of pectins to the cell wall, and that in the absence of ECH, pectins accumulate in the vacuole and the ER. Furthermore, the vesicular traffic of polysaccharides and proteins is affected differently in echidna mutants, suggesting that different macromolecules are packaged into vesicles leaving the TGN with diverse mechanisms that all require ECH.

Results
Pectins accumulate within echidna seed coat mucilage secretory cells The mucilage secretory cells of the Arabidopsis thaliana seed coat epidermis provide a model secretory system in which large amounts of cell wall polysaccharides are secreted as pectic mucilage (Western et al. 2000). In order to determine whether the post-Golgi trafficking mutant echidna is defective in seed coat polysaccharide secretion, we employed the ruthenium red assay (Western et al. 2000). When wild-type seeds are incubated in ruthenium red stain, mucilage is released from epidermal cells and a halo of red-stained mucilage surrounds the seed ( Fig. 1A; Western et al. 2000). This halo was greatly reduced in echidna seed coats, indicating a putative defect in mucilage biosynthesis, secretion or release ( Fig. 1A; Gendre et al. 2013). Scanning electron microscopy (SEM) examination of the seed coat epidermal surface showed that mature wild-type seed coat cells were characterized by raised central columellae, surrounded by a mucilage pocket (Fig. 1B). In contrast to the wild type, seed coat cells of echidna are severely altered, including flattened or sunken columellae, reduced apoplastic mucilage pockets and abnormal radial cell walls (Fig. 1B). The lack of a raised columella and altered mucilage pockets indicate that mucilage production or secretion is defective in echidna seeds.
The surface defects observed by SEM were consistent with the phenotype observed in sections of high-pressure frozen, freeze-substituted echidna and wild-type seeds examined using light microscopy ( Fig. 1C; Gendre et al. 2013). During mucilage secretion, wild-type seed coat cells change from typical parenchymous cells to being dominated by a large deposit of polysaccharide on the apical face of the epidermal cells (arrowheads, Fig. 1C; Young et al. 2008). Although early epidermal development was similar to that of the wild type, echidna mucilage secretory cells had few or no mucilage pockets during mid-and post-secretory stages and contain large accumulations that stained the same color and intensity as extracellular mucilage, inside the center of the cell (Fig. 1C). When the inclusions were viewed at high resolution with transmission electron microscopy (TEM), they appeared to be inside the vacuole (Fig. 2). Additionally, echidna mucilage secretory cells at midand post-secretory stages did not display the same polarization of cytoplasmic contents as wild-type cells. In wild-type cells, the large vacuole is positioned at the basal region of the cell, leaving most of the cytoplasm, including amyloplasts, concentrated in the apical cytoplasmic column. In contrast, the vacuole is roughly centered in echidna cells, with cytoplasm evenly distributed around it (Fig. 1C).
In addition, TEM revealed that in wild-type mucilage secretory cells at the peak of mucilage secretion, Golgi stacks are abundant and are associated with a highly complex TGN, which Ruthenium red staining (A) of mature wild-type and echidna seeds reveals a mucilage defect upon echidna seed coat hydration. Scanning electron micrographs (B) of mature seeds revealed morphological defects in echidna seed coats including aberrant mucilage pockets (arrowheads) and a lack of a defined columella that is present in the wild type (arrow). Toluidine blue-stained sections (C) of developing seed coat cells. At the early secretory phase (top), there are no distinguishable differences between the wild type and echidna. At mid-secretory phase (middle), mucilage is secreted to the apoplast in the developing mucilage pocket (arrowheads), and the developing columella (arrow) is visible in the wild type. In echidna mid-secretory phase cells, only a small amount of mucilage accumulates in the apoplast (arrowheads), and a large accumulation develops in the center of the cell (*). This phenotype persists into postsecretory phase cells (bottom). Scale bars represent 200 mm in A, 100 mm in B (top) and 10 mm in B (bottom) and C. develops in correlation with the developmental increase in mucilage synthesis and secretion ( Fig. 2D; Young et al. 2008). In echidna mucilage secretory cells, while the Golgi stacks were intact, the Golgi cisternae were shorter and the TGN contained fewer darkly staining secretory vesicle clusters and was less tightly associated with the Golgi stacks ( Fig. 2E; Gendre et al, 2011). In addition, there was also an accumulation of darkly staining vesicles in the cytoplasm of echidna mucilage secretory cells (Fig. 2C, F). In the wild type, dense vesicles contain pectin and are evenly distributed throughout the cytoplasm in mucilage secretory cells . In contrast, dense vesicles accumulate in a polar fashion in echidna mutants, with the highest concentration nearest the domain of the plasma membrane adjacent to the mucilage pocket (Fig. 2F).

Pectins accumulate within the vacuole of echidna cells
To confirm that the accumulations in echidna seed coat cells are composed of pectin, sections of seeds at the mid-secretory stage were probed with the anti-seed coat mucilage antibody, CCRC-M36 , which binds to unbranched rhamnogalacturonan I (Pattathil et al. 2010). In the wild type, CCRC-M36 strongly labels the apoplastic mucilage pockets of the mucilage secretory cells, as well as some cytoplasmic puncta, which correspond to the mucilage-producing Golgi stacks, but it does not label any other cell wall material, or anything outside of the mucilage secretory cells ( Fig. 3A; Supplementary Fig. S1). In echidna mutants, CCRC-M36 signal was concentrated at the small apoplastic mucilage pockets, but also strongly labeled the center of the cell and cytoplasmic puncta (Fig. 3A). This confirmed that the accumulation in the center of echidna mutant seed coat cells is mucilage.
To determine the distribution of the pectin in the endomembrane organelles of the seed coat cells in echidna, immunoTEM was performed using the CCRC-M36 anti-pectic mucilage antibody. Gold particles conjugated to secondary antibody detected CCRC-M36 epitopes in wild-type mucilage secretory cells in the apoplastic mucilage pocket, the Golgi apparatus, the TGN and secretory vesicles ( Fig. 3B; see Supplementary Fig. S2 for controls). In echidna mutants, gold particles indicated antibody binding to the Golgi apparatus, the TGN, the small mucilage pockets, as well as the central  Supplementary Fig. S2). Additionally, CCRC-M36 signal was detected in the vesicles that accumulate near the apical plasma membrane in echidna seed coat cells (Fig. 3E), implying that secretory vesicles containing pectin can form properly in echidna but they cannot properly fuse with their target membrane. The anti-pectic mucilage antibody did not significantly label the pre-vacuolar compartment/multivesicular bodies. ImmunoTEM using the CCRC-M36 antibody reveals that mucilage is also localized in the Golgi and TGN in the wild type (B). In echidna, mucilage is localized to the apoplast (C), the Golgi and TGN (D), as well as in accumulations of vesicles near the apoplast (E, arrows) and the center of the cell (F). Squares in A and B indicate areas within the cells where TEM images were taken for B-F. Scale bars represent 10 mm in A and B, and 200 nm in B-F; representative gold particles in B-F are highlighted by circles.
Although mucilage was clearly mistargeted in echidna mutants, the nature of the compartment containing mucilage was not known. Since this mucilage appeared to accumulate in the central lytic vacuole, the tonoplast fluorescent protein marker g-TIP-GFP (Boursiac et al. 2005) was crossed into echidna mutants and visualized with confocal microscopy. In both wildtype and echidna seed coat cells at the mid-secretory stage, g-TIP-GFP labeled the tonoplast, which outlined the lytic vacuole around the periphery of the cell, including where the vacuole was pushed inwards by amyloplasts, which are visible in brightfield (Fig. 4A). ImmunoTEM confirmed that anti-g-TIP (Jauh et al. 1999) labeled the tonoplast and the periphery of the mucilage accumulation in echidna mutants (Fig. 4B). These data verify that mucilage, which is synthesized but not secreted in echidna mutants, accumulates in the central lytic vacuole.
Pectins also accumulate within ER bodies of echidna cells TEM of echidna mutants also revealed unusual endomembrane compartments, which resembled ER dilations. Therefore, the ER marker GFP-HDEL (Batoko et al. 2000) was used to determine whether there were ER morphology changes in echidna mutants. In contrast to the wild-type cell, where ER was confined to an even reticulum around the cell cortex, echidna mutant ER displayed many dilations that correspond in size, shape and relative frequency to the ER bodies observed in TEM (Fig. 5A). In TEM preparations that were optimized for morphology, the ER structures of wild-type and echidna (Fig. 5B) were compared. Compartments of 300-1,500 nm diameter that were surrounded by ribosomes were observed exclusively in echidna. In some planes of section, direct connections to the ER were seen (Fig. 5B), indicating that they are ER-derived bodies. These ER bodies are morphologically distinct from oil or protein storage bodies found in embryos, or ER bodies typically found in leaf and root epidermal cells (Hayashi et al. 2001, Herman 2008. ImmunoTEM using CCRC-M36 confirmed that pectins accumulate in these ER bodies (Fig. 6A), and anti-calreticulin (an ER marker; Coughlan et al. 1997) confirmed the ER origin of these structures (Fig. 6B). Together, these results demonstrate that loss of normal ECH-mediated vesicle traffic has broad impacts on anterograde cell wall polysaccharide cargo secretion, with pectin cargo inappropriately accumulating, directly or indirectly, in both the vacuole and ER bodies.

Pectins and membrane proteins accumulate in distinct compartments in echidna mutants.
Our data demonstrated severe consequences of loss of function of ECHIDNA for polysaccharide secretory vesicle traffic. This raised the question of whether similar consequences would be seen for TGN resident proteins and for cargo proteins destined for the plasma membrane. Given that distinctive morphological changes occur to the TGN in seed coat cells during mucilage secretion , the localization of the TGN marker SYP61-cyan fluorescent protein (CFP) (Robert et al. 2008) was examined in live cells of wild type and echidna mutants. In wild-type secretory phase seed coat cells, SYP61-CFP was localized to abundant puncta throughout the cytoplasm (Fig. 7A). In echidna secretory phase seed coat cells, this TGN marker was localized to a few puncta, and also strongly labeled circular membrane inclusions (arrowheads, Fig. 7A). In TEM, multilamellar membrane structures of 2-5 mm in diameter were observed and resembled pre-autophagy structures or autophagosomes (Fig. 7C). Similar fluorescent membrane inclusions were observed in the seed coat cells expressing g-TIP-GFP (Fig. 4A), and immunoTEM with anti-g-TIP confirmed that this vacuolar marker is localized to the multilamellar membrane structures (Fig. 7E) in addition to the signal detected at the tonoplast (Fig. 4B). Interestingly, these multilamellar structures in which proteins accumulated did not label positively for pectin when tested with the CCRC-M36 antibody (Fig. 7F).
Previous characterization of the TGN and secretion defects in echidna root cells has demonstrated that the TGN-localized H + -ATPase subunit, VHA-a1, is mislocalized to circular puncta (Gendre et al. 2011), which correspond in size and morphology to the multilamellar membrane structures observed in echidna seed coat cells. In order to test whether mucilage mislocalization in echidna mutants could be due to VHA-a1 mislocalization, VHA-a1-GFP (Dettmer et al. 2006) was examined in echidna mutants. In wild-type secretory phase seed coat cells, VHA-a1-GFP was localized to abundant puncta throughout the cytoplasm, corresponding to the TGN. In echidna, VHA-a1-GFP signal was primarily localized to these puncta, with some mislocalization to the multilamellar membrane aggregations ( Supplementary Fig. S3A, B). To test whether this mild VHA-a1-GFP mislocalization affects mucilage secretion, VHA-a1 levels were down-regulated using ethanol-inducible RNA interference (RNAi)-VHA-a1 lines, and this was confirmed by reverse transcription-PCR (RT-PCR) (Supplementary Fig.  S3C; Brüx et al. 2008). Seeds induced during the pre-to postsecretion stages of seed development were screened using the ruthenium red assay (Western et al. 2000), and showed no discernible defects in mucilage release, compared with the wild type ( Supplementary Fig. S3). These data suggest that the mislocalization of VHA-a1 in echidna is not responsible for the mislocalization of pectin, and that echidna plays a direct role in the proper trafficking of pectin cargo.
To test the traffic of a plasma membrane-localized protein, yellow fluorescent protein (YFP)-ABCG11, an ATP-binding cassette transporter that is required for wax and cutin secretion (Bird et al. 2007), was observed in mid-secretory phase seed coat cells. In the wild type, YFP-ABCG11 signal was confined to the plasma membrane of seed coat secretory cells (Fig. 7B).
In echidna mutants, YFP-ABCG11 was primarily detected at the plasma membrane; however, there was also some YFP-ABCG11 accumulation in the circular membrane accumulations, as observed with g-TIP-GFP and SYP61-CFP (arrowheads, Fig. 7B). Since, in this system, YFP-ABG11 was ectopically expressed with the 35S promoter, specific antibodies against another plasma membrane protein, aquaporin PIP2 (plasma membrane intrinsic protein 2) (Bots et al. 2005), were employed. Anti-PIP2 specifically labeled the plasma membrane in wild-type cells, but in echidna mutants, anti-PIP2 was found primarily in the large, circular membrane accumulations, as well as at low levels in the plasma membrane and in the tonoplast (Fig. 7D). These data indicate that the circular multilamellar membrane inclusions contain TGN, tonoplast and plasma membrane proteins, but not mucilage. These data . g-TIP-GFP is localized to the vacuole in echidna mutants, but also strongly labels a circular structure (arrowheads) that does not colocalize with any of the amyloplasts. ImmunoTEM using anti-g-TIP (B) confirms g-TIP localization to the tonoplast in the wild type, and demonstrated that the mucilage accumulation in the center of echidna cells is surrounded by the tonoplast. Scale bars represent 10 mm in A and 200 nm in B; gold particles in B are highlighted by circles. also suggest that post-Golgi trafficking of soluble cell wall polysaccharides is distinct from post-Golgi trafficking of these membrane proteins in echidna mutants, which suggests that different post-Golgi mechanisms exist that are affected to varying degrees in echidna mutants.

Discussion
In this study, the seed coat phenotype of the echidna mutant provided a rare opportunity to observe the perturbation of cell wall vesicle trafficking during a developmental period of . GFP-HDEL is localized to the ER network in echidna mutants, but also strongly labels puncta throughout the cytoplasm. In TEM (B), these punctae appear to be ER dilations (arrowheads) which are not present in the wild type and appear to be directly connected to the ER (arrow). Scale bars represent 10 mm in A and 500 nm in B. abundant pectin secretion. The loss of function of ECH had dramatic effects on efficient cell wall polysaccharide export (Fig. 8). Rather than pectin being secreted to the apical domain of the plasma membrane surrounding the cytoplasmic column, as is the case in the wild type (Fig. 8A), pectin was found in the central vacuole and, to a lesser extent, in ER bodies in echidna (Fig. 8B). Accumulations of dense vesicles that labeled with the anti-pectin CCRC-M36 antibody at the cell cortex, adjacent to the usual site of secretion, indicate that vesicles did not dock or fuse efficiently at their usual target membrane in echidna, although the small apoplastic mucilage pockets present do suggest that a small number of vesicles were still able to target, dock and fuse with their target membrane. These successful vesicles could represent a population that contained the appropriate post-Golgi secretory vesicle identity, including the correct suite of vesicle traffic machinery, e.g. RABs and SNAREs. Thus, it appears that the primary defect in the echidna mutant is that the majority of secretory vesicles no longer efficiently traffic to or fuse with the plasma membrane, suggesting a role for ECH in targeting dense secretory vesicles carrying cell wall polysaccharides to the correct destination, probably by recruiting the appropriate vesicle trafficking proteins such as RABs and SNAREs at the TGN.
ECH is homologous to the yeast TGN-localized protein, TVP23 (Tlg2p-vesicle protein 23), which genetically interacts with a trans-Golgi SNARE complex and acts in retrograde trafficking (Stein et al. 2009). Yeast mutants that are defective in TVP23 display accumulation of a recycled protein in early endosomes, which is indicative of a defect in early endosome to TGN trafficking (Stein et al. 2009). Interestingly, expression of Arabidopsis ECH cDNA in Átvp23Áypt6 yeast was able to rescue fully the growth phenotype (Gendre et al. 2011), and immunological isolation and proteomics of SYP61-positive TGN/vesicles from Arabidopsis detected ECH in the same vesicles as the trans-Golgi SNARE SYP61 , suggesting conservation of function between yeast and plants. Fig. 6 ImmunoTEM reveals that mucilage is mislocalized to ER dilations in echidna. ImmunoTEM using CCRC-M36 (A) reveals that while mucilage does not accumulate in the ER of the wild type (arrows), mucilage is present in the ER dilations in echidna (arrowheads). ImmunoTEM using anti-calreticulin (B) that labels the ER in the wild type confirms that these mucilage-containing structures in echidna are ER derived (arrowheads). Scale bars represent 200 nm; representative gold particle are highlighted by circles.
Furthermore, the recent characterization of YIP4 proteins (RAB-interacting proteins) as interacting partners with TVP23 in yeast (Inadome et al. 2007) and ECH in plants (Gendre et al. 2013) strengthens the conservation of function. By analogy to yeast, it is possible that breakdown in post-Golgi trafficking in echidna mutants could result from altered retrograde traffic between the endosome and the TGN. However, the endomembrane phenotypes of yeast and Arabidopsis echidna mutants are distinct, suggesting that ECH may have evolved alternative or additional roles in cell wall secretion in plants.
One of the major concerns of working with a protein of unknown function is that mutant phenotypes include both primary and secondary defects. For example, the TGN-localized H + -ATPase, VHA-a1, is mislocalized in root cells of echidna (Gendre et al. 2011), which may disrupt the proton gradients that help to maintain the identity of endomembrane compartments (Dettmer et al. 2006, Brüx et al. 2008). However, mislocalization of GFP-VHA-a1 was minimal in echidna seed coat cells, and seeds in which RNAi-VHA-a1 (Brüx et al. 2008) had been induced showed no defect in mucilage release. This suggests that the loss of ECH, and not the mislocalization of VHA-a1, may cause the defects in cell wall secretion.
Whatever the underlying primary defect in secretory vesicle function in echidna mutants, the unexpected result was the steady-state accumulation of pectin in the central vacuole. This could be due to either a stress response to vesicle accumulation or the problems with post-Golgi vesicle identity leading to direct fusion of vesicles with the tonoplast. It is possible Fig. 7 TGN and plasma membrane proteins accumulate in compartments distinct from mucilage accumulation in echidna mutants. TGNlocalized SYP61-CFP (A) is partially mislocalized to circular structures in echidna that are distinct from amyloplasts. This is also true for the plasma membrane-localized YFP-ABCG11 (B). TEM reveals that these circular structures are multilamellar (C). ImmunoTEM using anti-PIP2 (D) and anti-g-TIP (E) confirms the presence of mislocalized plasma membrane and vacuole proteins, respectively, in these structures. However, immunoTEM using CCRC-M36 (F) reveals that mucilage is not mislocalized to these compartments. Scale bars represent 10 mm in A and B, 500 nm in C and 200 nm in D-F; representative gold particles in D-F are highlighted by circles. that mucilage may reach the vacuole by autophagy, which would serve to sequester the mucilage and reclaim the vesicle membranes and targeting proteins. Unfortunately, due to the difficulties in imaging established markers for autophagy such as GFP-ATG8e or monodanslycadaverine (MDC) (Contento et al. 2005) in the seed coat, we were unable to determine whether autophagic compartments are up-regulated in echidna. Pectin also accumulated in the ER bodies in less dense aggregates than in the vacuole in echidna mutants. Pectin mislocalization to the ER bodies may be a secondary defect of the post-Golgi trafficking issues in echidna mutants (Gendre et al. 2011). Although the Golgi apparatus is still morphologically distinct from the ER in echidna mutants, it is possible that the endomembrane disorganization observed in echidna mutants is also causing leakage of Golgi cargo (i.e. mucilage) or Golgi-resident enzymes (i.e. glycosyl transferases and associated nucleotide sugar transporters) to the ER, resulting in mucilage accumulation in ER dilations. The pectin in ER bodies may also contribute to the vacuolar accumulations of pectin, and direct delivery to the vacuole from the ER cannot be excluded.
Interestingly, secreted plasma membrane proteins and secreted polysaccharide cargo within echidna cells accumulate in distinct places within the cell. While pectic polysaccharides accumulated in the vacuole and the ER bodies, plasma membrane proteins ABCG11 and PIP2 accumulated in large multilamellar membrane accumulations that also contained the TGN/post-Golgi SNARE, SYP61, and the tonoplast marker g-TIP. These structures could represent precursors to autophagosomes or protein aggregates similar to those seen in nutrientstarved tobacco BY-2 cells (Toyooka et al. 2009, Takatsuka et al. 2011. Because cell wall polysaccharides and transmembrane plasma membrane proteins are mislocalized to distinct compartments in echidna mutants, these data suggest that, although both cargo types are secreted to the plasma membrane, they may travel in vesicles with distinct trafficking machinery, or that the plasma membrane proteins are recycled through the TGN/early endosome in an ECH-dependent manner. These data corroborate previous results that found that the dominant-negative version of the SYP121 SNARE caused a decrease in secretion of soluble protein markers b--glucuronidase (GUS) and secGFP, but had no effect on either a secreted glycosylphosphatidylinositol (GPI)-anchored protein (De Caroli et al. 2011) or secreted polysaccharides (Leucci et al. 2007). These results highlight the complexity of the plant endomembrane trafficking network since different cargoes destined to the same target may be transported in different pathways, using different sets of vesicular trafficking machinery and/or targeting motifs.
In summary, we have shown here that secretion of noncellulosic cell wall components to the plasma membrane Fig. 8 Overview of the phenotypes observed in echidna seed coat cells. At the height of mucilage secretion in wild-type seed coat cells (A), the trans-Golgi network (TGN)-localized ECHIDNA is required for secretion of large amounts of Golgi-synthesized pectic mucilage (purple) via secretory vesicles. The TGN also acts as an early endosome (EE) by receiving endocytic material, which may be recycled to the plasma membrane (PM) or transported to the vacuole via a multivesicular body/pre-vacuolar compartment (MVB/PVC). In the echidna mutant (B), vacuolar transport and endocytosis remain largely unaffected, but the TGN loses its ability to generate competent vesicles for wild-type levels of mucilage secretion. This results in the polar accumulation of dense, polysaccharide-containing vesicles near the PM, as well as mucilage accumulation in the endoplasmic reticulum (ER) and the vacuole. Membrane proteins targeted to the PM and the tonoplast (including ABCG11, PIP2, g-TIP and SYP-61) are partially mislocalized to multilamellar bodies (MLBs) that do not contain any mislocalized polysaccharides. requires ECH-dependent machinery for efficient trafficking. In the absence of this proper machinery, these soluble cell wall cargoes are mislocalized to the vacuole and the ER, either directly, through vesicle fusion, or indirectly, through autophagy or improper retrograde trafficking. The mislocalization of membrane-localized proteins and soluble cargo in echidna emphasizes the importance of studying polysaccharide traffic, in addition to protein trafficking, in order to form a complete understanding of cell wall synthesis and secretion.

Plant material
Arabidopsis thaliana seeds were sown on AT media plates (Haughn and Somerville 1986) and grown at 21 C, 70-80% humidity and constant light ($100 mE m À2 sÀ 1 ) in an environmental growth chamber (Conviron). After 10-14 d, seedlings were transferred to Sunshine mix 5 soil and returned to the same growth conditions. To obtain seeds at different stages in development, siliques were marked with non-toxic paint at the day of anthesis, according to Western et al. (2000). Because of general developmental defects in echidna mutants, it was difficult to obtain accurately developmentally staged seeds according to established convention of days post-anthesis (dpa; Western et al. 2000). Therefore, echidna mutant seeds were grouped into three stages: pre-secretion (similar to the wild type 4 dpa, before mucilage accumulation has begun), mid-secretion (similar to the wild type 7 dpa, at which time mucilage synthesis and secretion is at its peak) and post-mucilage secretion (similar to the wild type 10 dpa, at which time mucilage synthesis and secretion has finished and secondary cell wall synthesis has begun).

Gene expression analysis
Seed coats from mid-secretory phase seeds were isolated from the embryo and quickly frozen with liquid nitrogen. RNA was isolated using the RNAqueous-Micro kit (Ambion). cDNA was synthesized using 1 mg of RNA using a mixed oligo dT15, dT17 and dT20 primer set and M-MLV reverse transcriptase (Promega). RT-PCR was performed for 29 cycles using 5 ml of cDNA with intron-flanking, gene-specific primers for VHA-a1 (see Supplementary Table S1). The UBC10 ubiquitin-conjugating enzyme was used as the loading control. PCR products were visualized using SYBR-Safe (Invitrogen).

Ruthenium red staining
Mature seeds were incubated in 0.01% (w/v) aqueous solution of ruthenium red for 30 min at room temperature under gentle shaking. Seeds were then mounted in water and viewed using a Leica DMR microscope equipped with a QICAM digital camera (QIMAGING).

Scanning electron microscopy
Mature seeds were dry mounted on stubs and coated with 10 nm of palladium-gold using a Cressington 208HR sputter coater. Samples were viewed on a Hitachi S4700 FE-SEM with an accelerating voltage of 5 kV, beam current of 10 mA and a working distance of 12 mm, using a mix of both the upper and lower detectors.

Confocal microscopy
Mid-secretory phase seeds were mounted in water and viewed under a Leica DMI6000 B inverted microscope with a Perkin Elmer spinning-disk scan head. YFP-ABCG11 was detected using the 514 nm laser diode and 540/30 nm emission filters. GFP-HDEL was detected using the 488 nm laser diode and 525/ 36 nm emission filters. SYP61-CFP, VHA-a1-GFP and g-TIP-GFP were viewed under an axioplan inverted microscope with a Zeiss LSM 780 spectral CLSM system. g-TIP-GFP and VHA-a1-GFP were detected using the 488 nm laser and 493-598 nm emission filters. SYP61-CFP was detected with the 405 nm laser and 454-581 nm emission filters.

High-pressure freezing, TEM and immunogold labeling
Staged seeds were high-pressure frozen in 1-hexadecene in Btype sample holders (Ted Pella) using a Leica HPM-100. For morphology, freeze substitution in 2% osmium tetroxide and 8% dimethoxypropane, resin infiltration with Spurr's resin, sectioning, and post-staining with 2% uranyl acetate and lead citrate were performed as described in McFarlane et al (2008). For immunolabeling, high-pressure frozen samples were freeze-substituted in 0.1% uranyl acetate, 0.25% glutaraldehyde and 8% dimethoxypropane, then infiltrated with LR White resin (London Resin Company), as described in McFarlane et al. (2008). Immunolabeling was performed according to McFarlane et al. (2008). For immunoTEM, primary antibodies were 1/10 M36 (generated in mouse against seed coat mucilage from the Complex Carbohydrate Research Center, University of Georgia; Pattathil et al. 2010), 1/20 polyclonal anti-calreticulin [generated in rabbit against Castor bean (Ricinus communis