https://orcid.org/0000-0001-6554-8446
Abstract
The cell wall is essential for plant survival. Determining the relationship between cell wall structure and function using mutant analysis or overexpressing cell wall–modifying enzymes has been challenging due to the complexity of the cell wall and the appearance of secondary, compensatory effects when individual polymers are modified. In addition, viability of the plants can be severely impacted by wall modification. A useful model system for studying structure–function relationships among extracellular matrix components is the seed coat epidermal cells of Arabidopsis thaliana. These cells synthesize relatively simple, easily accessible, pectin-rich mucilage that is not essential for plant viability. In this study, we expressed enzymes predicted to modify polysaccharide components of mucilage in the apoplast of seed coat epidermal cells and explored their impacts on mucilage. The seed coat epidermal-specific promoter TESTA ABUNDANT2 (TBA2) was used to drive expression of these enzymes to avoid adverse effects in other parts of the plant. Mature transgenic seeds expressing Rhamnogalacturonate lyase A (RglA) or Rhamnogalacturonate lyase B (RglB) that degrade the pectin rhamnogalacturonan-I (RG-I), a major component of mucilage, had greatly reduced mucilage capsules surrounding the seeds and concomitant decreases in the monosaccharides that comprise the RG-I backbone. Degradation of the minor mucilage component homogalacturonan (HG) using the HG-degrading enzymes Pectin lyase A (PLA) or ARABIDOPSIS DEHISCENCE ZONE POLYGALACTURONASE2 (ADPG2) resulted in developing seed coat epidermal cells with disrupted cell–cell adhesion and signs of early cell death. These results demonstrate the feasibility of manipulating the seed coat epidermal cell extracellular matrix using a targeted genetic engineering approach.
Introduction
In Arabidopsis thaliana (Arabidopsis), fertilization occurs at the same time as anthesis and initiates the development of the seed coat from the ovule integuments (Beeckman et al. 2000, Western et al. 2000, Windsor et al. 2000). By 5-day post-anthesis (DPA), the epidermal cells of the outer integument start to secrete pectinaceous mucilage into the apoplast at the junction between the outer and radial primary cell walls. Secretion peaks at 7 DPA when the mucilage forms a donut-shaped ring known as the mucilage pocket that is adjacent to the outer primary cell wall and surrounds a column of cytoplasm. By 9 DPA, mucilage secretion is complete and the cells begin to synthesize a secondary cell wall that replaces the cytoplasm to form the columella. Once columellae formation is complete, the cells undergo programmed cell death. On hydration of mature dry seeds, the outer primary cell wall of the epidermis ruptures and mucilage is released to form a mucilage halo with two distinct layers. The outer non-adherent mucilage layer is easily removed by gentle shaking, while the inner adherent layer is tightly attached to the seed coat surface and can only be removed with harsh physical, chemical or enzymatic treatments (Macquet et al. 2007).
The major polysaccharides in Arabidopsis mucilage are pectins (∼90%) plus cellulose and the hemicelluloses xylan and galactoglucomannan (reviewed in Voiniciuc et al. 2015). Seed coat mucilage has unique properties that allow rapid expansion on contact with water and strong adherence of the inner mucilage layer to the seed surface. These unique properties are mediated by interactions between polysaccharides in the mucilage matrix that are controlled by their structures (reviewed in Šola et al. 2019).
Seed mucilage pectin is ∼90% rhamnogalacturonan-I (RG-I) and 10% homogalacturonan (HG; Huang et al. 2011; reviewed in Voiniciuc et al. 2015). RG-I has a backbone of alternating d-galactopyranosyluronic acid (GalA) and l-rhamnopyranose (Rha) residues in a [→2)-α-Rha-(1→4)-α-GalA-(1→] pattern (McNeil et al. 1980) that is branched at the Rha residues with side chains consisting of d-galactopyranose (Gal), l-arabinofuranose (Ara) or both (arabinogalactans; Lau et al. 1987, Lerouge et al. 1993). Mutation of the MUCILAGE-MODIFIED4 (MUM4) gene, which encodes an NDP-l-Rha synthase, leads to a major reduction in the quantity of RG-I synthesized and a concurrent decrease in mucilage quantity (Western et al. 2001, 2004, Usadel et al. 2004, Oka et al. 2007). Mutation of the RG-I:rhamnosyltransferase 1 (RRT1) gene involved in RG-I synthesis also leads to a decrease in mucilage RG-I levels and a 42% decrease in the volume of extruded mucilage (Takenaka et al. 2018). HG is a linear polymer of α-(1→4)-linked GalA residues that can be methylesterified to alter the gelling properties of the HG matrix (Harholt et al. 2010). Decreased methylesterification promotes increased calcium cross-linking between adjacent HG chains that consequently increases the rigidity of the mucilage. Conversely, increased methylesterification or the use of calcium chelators can decrease mucilage cohesion (reviewed in Šola et al. 2019).
Seed coat mucilage comprises the same major polysaccharide groups as the primary cell wall (pectins, hemicelluloses and cellulose) but is more accessible, can be disrupted with no adverse effects on the rest of the plant and is not required for viable seeds under laboratory conditions. In addition, seed mucilage has been relatively well-characterized and is comparatively simple compared to primary cell walls. Therefore, seed coat mucilage is an appealing model system for investigating the structure–function relationships between different polysaccharides and their impact on mucilage properties.
Many previous studies have focused on genetic screens involving analysis of mutants with impaired seed coat development. This approach has been productive and has identified a number of key genes in mucilage biogenesis (reviewed in Šola et al. 2019). However, with the development of new tools and resources, there are opportunities to make targeted changes to mucilage polysaccharides and examine the effects on mucilage properties by expressing carbohydrate-active enzymes specifically in the developing seed coat epidermis. As well as providing insights into the structure–function relationships between different polymers, there are also potential applications including modification of the seed coat for specific purposes and production of recombinant proteins that are deposited into the mucilage pocket.
One important set of tools required for this are seed coat-specific promoters that can be used to express enzymes at the appropriate developmental stage and only in seed coat epidermal cells. Several seed coat-specific promoters have been identified including TESTA ABUNDANT2 (TBA2; Tsai et al. 2017), PEROXIDASE36 (PER36; Kunieda et al. 2013) and a seed coat-specific promoter fragment from MUM4 fused with the Cauliflower Mosaic Virus (CaMV) 35S core promoter (MUM40.3Pro_35S; Dean et al. 2017). Comparison of the TBA2, PER36 and MUM40.3Pro_35S promoters showed that the TBA2 promoter (TBA2p) directs the strongest expression of chimeric genes in the seed coat epidermis (McGee et al. 2019).
Enzymes that are active against linkages found in seed mucilage can be used to modify mucilage polysaccharides. As the major component of mucilage is RG-I, modification of this component should cause changes that are easily discernible and have obvious effects on mucilage. Rhamnogalacturonate lyases (RGLs; CAZy Polysaccharide Lyase 4 (PL4) family, EC 4.2.2.23) degrade the RG-I backbone to release GalA→Rha disaccharides (reviewed in Bonnin et al. 2014). Rhamno lyase A (RglA) and Rhamnogalacturonan lyase B (RglB) from the saprophytic fungus Aspergillus nidulans have previously been shown to degrade linseed RG-I (Bauer et al. 2006), making these two enzymes suitable candidates for expression in Arabidopsis seed coat epidermal cells.
A second interesting target for modification is HG. Although the degree of methylesterification can influence the gelling properties of mucilage as described above, it is not clear if HG plays additional roles in determining mucilage structure because pectin biosynthetic mutants appear to cause alterations in both RG-I and HG (Caffall and Mohnen 2009; reviewed in Šola et al. 2019). This may occur because HG and RG-I can be part of a single polymer, which is thought to be assembled during pectin biosynthesis either as separate domains that are then joined together or assembled concurrently as a large single polymer (Atmodjo et al., 2013). The Pectin lyase A (PelA) gene from A. nidulans encodes Pectin lyase A (PLA) from the CAZy Polysaccharide Lyase 1 (PL1) family (EC 4.2.2.10), which degrades citrus pectin to methyl-esterified GalA oligosaccharides (Bauer et al. 2006). The ARABIDOPSIS DEHISCENCE ZONE POLYGALACTURONASE2 (ADPG2) Arabidopsis gene encodes an endo-polygalacturonase (PG) from the CAZy Glycosyl Hydrolase 28 (GH28) family (EC 3.2.1.15) that facilitates cell separation in the dehiscence and abscission zones during floral development (Ogawa et al. 2009).
In this study, we generate transgenic plants that express RglA, RglB, PelA or ADPG2 under the control of the TBA2p and examined the ability of these chimeric genes to specifically modify seed mucilage. We provide evidence of targeted degradation of mucilage RG-I in seed coat epidermal cells, which results in an almost total loss of the adherent mucilage layer as well as a dramatic change in mucilage pocket shape. Targeted degradation of mucilage HG led to a number of unexpected cellular phenotypes that serve to highlight the central role HG plays within the cell.
Results
Expressing secreted RG-I-degrading enzymes in developing seed coat epidermal cells eliminated extruded seed mucilage
The relationship between RG-I structure and mucilage properties was investigated by expressing the RG-I degrading enzymes RglA and RglB in seed coat epidermal cells during mucilage synthesis. The sequences predicted to encode the secretion signal peptides of these fungal-derived enzymes were identified and replaced with the plant-derived MUM2 signal peptide (MUM2sp), which directs secretion to the mucilage pocket and radial cell walls (Lee 2018). Signal peptides were identified using the Phobius (Käll et al. 2007), SignalP (Nielsen 2017), Signal-BLAST (Frank and Sippl 2008) and UniProt databases (Bateman et al. 2017). Versions of TBA2p:MUM2sp-RglA and TBA2p:MUM2sp-RglB without a tag (referred to as RglA and RglB) and with a C-terminal Citrine yellow fluorescent protein tag (referred to as RglA-Citrine and RglB-Citrine) were generated, as well as MUM2sp-Citrine as a negative control. Citrine tags were added to the C-terminus of RglA and RglB because both enzymes have been shown to be active when the C-terminus was Myc + His-tagged (Bauer et al. 2006). All five chimeric genes were placed under the control of the seed coat-specific TBA2p, which is active in seed coat epidermal cells during mucilage synthesis and secretion (6–8 DPA; Tsai et al. 2017, McGee et al. 2019). Constructs were introduced into wild-type Arabidopsis, and the resulting transgenic seeds were screened for their ability to extrude mucilage.
Seed from nine independent transformants expressing MUM2sp-Citrine were examined, and all had wild-type-like mucilage halos when hydrated in water (Fig. 1A). In contrast, 24 independent transformants expressing RglA and 24 independent transformants expressing RglA-Citrine produced seeds that had little to no mucilage halo when hydrated in water (Fig. 1A, Supplementary Fig. S1A). Similarly, almost no extruded mucilage was observed in hydrated seeds from 18 independent transformants expressing RglB or 15 independent transformants expressing RglB-Citrine (Fig. 1A, Supplementary Fig. S1A). Next, seeds were treated with sodium carbonate (Na2CO3), which promotes mucilage release by extracting pectins that are covalently linked within the extracellular matrix (Selvendran 1985, Redgwell and Selvendran 1986, Dean et al. 2007). While Na2CO3 slightly increased mucilage extrusion in transgenic seeds compared with water, much less mucilage was extruded than from wild-type seeds (Supplementary Fig. S1B). These data suggest that the activities of RglA and RglB result in a near absence of extruded mucilage, and that these enzymes are active when fused to Citrine.
Phenotypes of RGL-expressing seeds. (A) Mature seeds from wild type, wild type expressing MUM2sp-Citrine, RglA-Citrine, RglB-Citrine or RglB(K215A + H277A)-Citrine and secCitrine in mum4 lines hydrated in water and then stained with ruthenium red to visualize extruded mucilage. Scale bar = 200 μm. Localization of Citrine fluorescence in longitudinal (upper panel) and transverse (lower panel) optical sections of 7 DPA developing seed coat epidermal cells lines using fluorescence microscopy. Scale bars = 20 μm. Insets (lower panel) show the embryo removed from these seeds to confirm developmental stage. Scale bars = 50 μm. (B) Scanning electron micrographs of mature seeds from wild type, mum4, RglA-Citrine and RglB-Citrine. Scale bars = 20 μm. (C) Monosaccharide composition of extruded mucilage and whole seed AIR prepared from wild type, three independent RglA-Citrine lines and three independent RglB-Citrine lines. Values are the mean of four replicates ± standard deviation and are expressed as nmole monosaccharide per milligram of seeds used to extract extruded mucilage or milligram of AIR recovered from whole seeds. Data presented are one of two comparable replicates. WT = wild type; M = mucilage pocket; C = cytosol; R = radial cell wall; Co = columella.
To test whether the catalytic activity of RglB was responsible for these mucilage phenotypes, the predicted catalytic amino acids within the active site were mutated and the effect on extruded mucilage was re-examined. In an RGL from Aspergillus aculeatus (RGL4), lysine (K) 150 and histidine (H) 210 are important for the catalytic activity (Jensen et al. 2010). The corresponding amino acids in RglB were identified as K215 and H277 by aligning the protein sequence of RglB with RGL4 (Q00019) and two other characterized RGLs (P78710 and Q8NJK5; McDonough et al. 2004) using MUSCLE (Madeira et al. 2019). The sequences encoding these predicted catalytic residues were mutated to yield the TBA2p:MUM2sp-RglB(K215A + H277A)-Citrine construct (RglB(K215A + H277A)-Citrine). This construct was introduced into wild-type Arabidopsis and resulted in 11 independent transformants that produced seeds with wild-type-like mucilage halos (Fig. 1A). Taken together, these data suggest that aberrant mucilage halos are indeed caused by the expression of active RG-I-degrading enzymes in wild-type plants.
RG-I-degrading enzymes were successfully secreted to the mucilage pocket
Changes to the seed mucilage phenotype observed in lines expressing tagged or untagged RglA and RglB (Fig. 1A, Supplementary Fig. S1) suggested that the encoded enzymes were secreted to the mucilage pocket and degraded RG-I. To verify this, localization of the Citrine-tagged enzymes in developing seed coat epidermal cells was examined using fluorescence microscopy. Developing seeds were examined at 7 DPA when mucilage synthesis and secretion are at their peak and when the level of transcription driven by TBA2p is also at its highest (Tsai et al. 2017). No fluorescence was detected in wild-type epidermal cells (Fig. 1A). The distribution of MUM2sp-Citrine was examined and Citrine fluorescence localized to the apoplast, predominately within the mucilage pocket and weakly within the radial cell walls (Fig. 1A, labeled as M and R, respectively). Fluorescence was not detected in the cytosol (Fig. 1A). Similarly, Citrine fluorescence was detected in the mucilage pocket and radial cell walls and not in the cytosol of 7 DPA epidermal cells expressing RglA-Citrine or RglB-Citrine (Fig. 1A), confirming that Citrine-tagged RG-I-degrading enzymes were localized to the mucilage pocket during the peak of mucilage synthesis and secretion. In lines expressing the inactivated variant RglB(K215A + H277A)-Citrine, fluorescence was also only observed in the mucilage pocket and radial cell walls (Fig. 1A), indicating that alteration of the protein sequence does not alter its localization.
Mucilage pockets were smaller in lines expressing RglA or RglB
The mucilage pockets in RglA-Citrine and RglB-Citrine lines appeared noticeably smaller than the wild-type-like mucilage pockets of both MUM2sp-Citrine and RglB(K215A + H277A)-Citrine lines (Fig. 1A). To ensure this difference was not due to incorrect staging of the seeds used for imaging, developmental stage was confirmed by examining the embryos (Fig. 1A, insets).
The small mucilage pockets in RglA-Citrine and RglB-Citrine lines were reminiscent of those observed in the mum4 mutant (Western et al. 2001, 2004, Usadel et al. 2004, Oka et al. 2007). MUM4 encodes an enzyme involved in RG-I biosynthesis, and loss-of-function mum4 mutants produce reduced amounts of mucilage and consequently have smaller mucilage pockets than those of wild type. To visualize these small mucilage pockets in mum4, we examined plants expressing a secreted form of Citrine (secCitrine) in the mum4 background (Lee 2018). Citrine fluorescence was observed in the apoplast of mum4 seed coat epidermal cells. The mucilage pockets were smaller than those of wild type and similar in size to those of RglA-Citrine and RglB-Citrine lines (Fig. 1A), suggesting that the small mucilage pockets in RglA-Citrine and RglB-Citrine lines contained less mucilage.
In wild type, the prominent volcano-shaped columellae of mature dry seed coats can be visualized using a scanning electron microscope (SEM), whereas in mum4 mutants, the small mucilage pockets give the columella a distinctive flattened appearance (Western et al. 2004). Examination of mature RglA-Citrine and RglB-Citrine seeds indicated that RglA-Citrine and RglB-Citrine columellae resemble those of mum4 (Fig. 1A). Sections of 7 DPA developing seeds from wild type, mum4 and two independent RglB-Citrine lines were prepared and examined to confirm that wild-type seed coat epidermal cells had larger mucilage pockets than both mum4 and RglB-Citrine epidermal cells (Supplementary Fig. S2).
The fact that the localization of RglA-Citrine and RglB-Citrine fluorescence in the mucilage pockets was correlated with a reduction in mucilage pocket size, and lines expressing inactivated RglB [RglB(K215A + H277A)-Citrine] had wild-type-like mucilage pockets, implies that the enzymatic activity of these RG-I-degrading enzymes are directly responsible for these phenotypes.
Mucilage pockets in seeds expressing RglA or RglB had an unusual fluorescence pattern
When viewed in transverse cross-section parallel with the seed surface, mucilage pockets in 7 DPA seeds appear donut-shaped. In seeds expressing secreted forms of Citrine, fluorescence was distributed uniformly throughout the mucilage pocket in both wild type (large pockets) and mum4 (small pockets; Fig. 1A). Similar uniform distributions have been observed in other mucilage-localized proteins fused to Citrine including TBA1, TBA2 and TBA3 (Tsai et al. 2017); MUM2/BGAL6, BGAL11, BGAL16 and BGAL17 (McGee et al. 2019); SOS5 (Griffiths et al. 2016) and when PRX50 was fused to tagRFP (Francoz et al. 2019). In contrast, the fluorescence distribution of RglB-Citrine and occasionally RglA-Citrine exhibited an unusual, striped fluorescence pattern that was arranged like spokes on a wheel (Fig. 1A). Like other aspects of the observed transgenic phenotype, this unusual fluorescence pattern was dependent on the RglB enzyme activity since lines carrying inactivated RglB (RglB(K215A + H277A)-Citrine) had wild-type-sized mucilage pockets with fluorescence evenly distributed throughout (Fig. 1A).
To determine whether this striped pattern was unique to these Citrine-tagged RGL proteins in RGL-induced small mucilage pockets, we expressed a different Citrine-tagged protein in lines expressing untagged RglB. Similar to the RglB-Citrine lines, lines expressing untagged RglB had smaller mucilage pockets than wild type, as confirmed by introducing the plasma membrane marker mRFP-VAMP721 (Ichikawa et al. 2014) into the RglB background and examining the cells using fluorescence microscopy (Supplementary Fig. S3A). When MUM2sp-Citrine was introduced into plants carrying untagged RglB, the observed MUM2sp-Citrine fluorescence pattern was also striped in the same way as observed in plants expressing RglB-Citrine (Fig. 1A, Supplementary Fig. S3B). This was in contrast to the uniform distribution of MUM2sp-Citrine in the absence of RglB (Fig. 1A), suggesting that this striped localization of mucilage pocket proteins is not specific to RglB-Citrine but is observed when other proteins were localized to the RGL-induced small mucilage pockets.
To examine the shape of the mucilage pocket in developing seeds, the plasma membrane dye FM 4-64 was employed. Developing seeds carrying functional full-length MUM2 fused with Citrine (MUM2-Citrine) that complements the mum2 mutant (McGee et al. 2019) were stained with FM 4-64 to examine the plasma membrane in seeds with wild-type-like mucilage pockets. Transverse sections of 7 DPA MUM2-Citrine seeds showed two rings of FM 4-64-stained plasma membrane surrounding the mucilage pocket where MUM2-Citrine fluorescence is localized (Fig. 2A, arrows, see transverse diagram in Supplementary Fig. S4A). Longitudinal sections showed the plasma membrane forms a smooth curved surface underlying the uniform MUM2-Citrine fluorescence within the mucilage pocket (Fig. 2A, arrows; see longitudinal diagram in Supplementary Fig. S4A). The three-dimensional (3D) modeling of Z-stacks allowed visualization of the plasma membrane delineating the volcano-shaped cytoplasmic column at the center of the cell that is surrounded by the donut-shaped mucilage pocket made visible by the fluorescence of MUM2-Citrine (Fig. 2A; see 3D wireframe model in Supplementary Fig. S4A). To verify the shape of the plasma membrane, we examined wild-type plants carrying the plasma membrane marker mRFP-VAMP721 (Ichikawa et al. 2014). In wild type, the mRFP-VAMP721 signal delineated the cytoplasmic column (Supplementary Fig. S5A) in a pattern similar to the FM 4-64 signal in MUM2-Citrine lines (Fig. 2A).
Visualization of developing mucilage pockets. 7 DPA developing seed coat epidermal cells expressing (A) MUM2-Citrine in mum2 and (B) RglB-Citrine in wild type were imaged using fluorescence microscopy. Mucilage pockets were visualized using Citrine fluorescence and the plasma membrane was stained using FM 4-64. For each genotype, transverse (upper panel) and longitudinal (middle panel) optical sections were collected. The lower panel shows a 3D model of Z-stacks. Arrows denote plasma membrane surrounding the mucilage pocket in (A). WT = wild type; M = mucilage pocket; C = cytosol, PM = plasma membrane; R = radial cell wall. Scale bar = 20 μm.
Next, developing seeds expressing RglB-Citrine were stained with FM 4-64, and the region around the mucilage pocket was examined. In contrast to the MUM2-Citrine lines FM 4-64 fluorescence in transverse sections of RglB-Citrine lines showed a striped pattern that appeared to follow the contours of the similarly shaped RglB-Citrine fluorescence (Fig. 2B), suggesting that the plasma membrane surrounding the mucilage pocket had accordion-like folds that divide the pocket into radial spoke-like domains. Similarly, folded plasma membrane and radial spoke-like patterns in the mucilage pockets were observed in RglB-Citrine plants carrying the plasma membrane marker mRFP-VAMP721 (Supplementary Fig. S5B, see transverse diagram in Supplementary Fig. S4B). In addition, longitudinal sections of the FM 4-64 fluorescence overlaid with the apoplastic RglB-Citrine fluorescence showed that the mucilage pockets were very shallow in RglB-Citrine lines when compared to the wild-type-sized pockets observed in MUM2-Citrine (Fig. 2; compare longitudinal sections in 2A vs. 2B). Shallow mucilage pockets were also apparent in RglB-Citrine lines carrying mRFP-VAMP721 when compared to wild type carrying mRFP-VAMP721 (Supplementary Fig. S5; compare longitudinal sections in S5A vs. S5B). The 3D modeling of Z stacks clearly indicated that unlike the smooth volcano-shaped plasma membrane in MUM2-Citrine lines (Fig. 2A) the plasma membrane underlying the mucilage pocket of cells from the RglB-Citrine lines was folded, forming ridges that produced an alternating pattern of a relatively shallow and narrow groove followed by a relatively deep and wide groove (Fig. 2B; see longitudinal diagrams and 3D wireframe model in Supplementary Fig. S4B). In addition, there was also a ring of deeper mucilage pocket immediately surrounding the cytoplasmic column (Fig. 2B, Supplementary Figs. S4B, S5B). In summary, these data suggest that a folded plasma membrane underlies the small mucilage pockets in transgenic lines where functional RG-I-degrading RglB was synthesized and that the mucilage pockets vary in depth to create a radial spoke-like pattern.
RG-I monosaccharides were greatly reduced in RglA-Citrine and RglB-Citrine seeds
As described above, the presence of active RG-I-degrading enzymes in the mucilage pocket results in small mucilage pockets that released little to no mucilage when mature seeds were hydrated in water (Fig. 1A). These data suggest that RglA and RglB degrade mucilage RG-I, the primary polysaccharide in the mucilage pocket, resulting in a reduced amount of mucilage and a smaller pocket size. To confirm this, we determined the monosaccharide composition of both extracted mucilage and whole seed alcohol insoluble residue (AIR).
To release mucilage for analysis, mucilage was extracted with Na2CO3 from wild type, three independent lines carrying RglA-Citrine and three independent lines carrying RglB-Citrine. In wild-type seeds, the major monosaccharides detected were rhamnose (Rha) and galacturonic acid (GalA) derived from the RG-I backbone (Fig. 1C). Relatively small amounts of galactose (Gal), arabinose (Ara), glucose (Glc), xylose (Xyl) and mannose (Man) derived from RG-I side chains and other mucilage polysaccharides were also observed (Fig. 1C). In contrast, an 80–90% reduction in Rha and GalA was detected in mucilage extracted from seeds expressing RglA-Citrine or RglB-Citrine, while most other monosaccharides remained comparable (Fig. 1C). The amount of extracted Gal was increased in RglA-Citrine but not in RglB-Citrine lines, suggesting that RglA and RglB may have different substrate preferences. These data suggest that the amount of RG-I in extractable mucilage was much lower in these transgenic lines than in wild type.
To ensure that these data were not a consequence of mucilage that is less easily extractable due to changes in polysaccharide structure, monosaccharide analysis was performed on AIR prepared from whole mature seeds. Similar decreases in Rha and GalA were observed in seeds expressing RglA-Citrine and RglB-Citrine when compared to wild type (Fig. 1C). Taken together, these data suggest that secretion of RglA-Citrine or RglB-Citrine fusion proteins to the mucilage pocket results in the loss of the majority of RG-I in seed mucilage.
Transgenic seeds expressing HG-degrading enzymes targeted in the mucilage pocket extruded little to no mucilage
The pectin HG is the second most abundant polysaccharide in mucilage and like RG-I is thought to be interconnected to other polymers in a highly ordered network that establishes mucilage properties (reviewed in Šola et al. 2019). Since mutants with major reductions exclusively in seed mucilage HG have not been reported, the structural role that HG plays within this network and how it impacts mucilage properties remains unclear (reviewed in Šola et al. 2019). To address these questions, we introduced two HG-degrading enzymes, ARABIDOPSIS DEHISCENCE ZONE POLYGALACTURONASE2 (ADPG2) from Arabidopsis (Ogawa et al. 2009) and PLA (encoded by the PelA gene) from A. nidulans (Bauer et al. 2006), into the mucilage pocket during seed development and examined the subsequent changes on mucilage. The same approach as described above for RglA and RglB was used to identify the sequences predicted to encode the native ADPG2 and PLA signal peptides, which were then replaced with MUM2sp. Transgene expression was again driven by the seed coat epidermal cell-specific promoter TBA2p. Chimeric constructs were generated that encoded untagged ADPG2 (ADPG2) and PLA (PelA) as well as N-terminal Citrine-tagged ADPG2 (Citrine-ADPG2) and C-terminal Citrine-tagged PLA (PelA-Citrine). These termini were selected for tagging because His-ADPG2 and PLA-Myc + His had previously been shown to degrade HG in vitro (Bauer et al. 2006, Ogawa et al. 2009). The constructs were transformed into wild-type Arabidopsis plants, and the resulting transgenic seeds were screened for changes in mucilage extrusion.
When transgenic seeds were hydrated in water, very little mucilage was observed on the seed surface of 20 independent transformants expressing ADPG2 and 17 independent transformants expressing Citrine-ADPG2 that were screened, in contrast to the large mucilage halos observed in wild-type seeds (Fig. 3A, Supplementary Fig. S1A). Similarly, only small patches of mucilage were extruded in seeds from 13 independent transformants expressing PelA and 13 independent transformants expressing PelA-Citrine (Fig. 3A; Supplementary Fig. S1A). Since similar mucilage phenotypes were observed with both tagged and untagged versions of both enzymes, the Citrine tag does not appear to have a measurable impact on their activity.
Phenotypes of seeds expressing HG-degrading enzymes. (A) Mature seeds from MUM2sp-Citrine, Citrine-ADPG2, PelA-Citrine and PelA(R256A)-Citrine lines in wild type hydrated in water and then stained with ruthenium red to visualize extruded mucilage. Scale bar = 200 μm. Localization of Citrine fluorescence in longitudinal (upper panel) and transverse (lower panel) optical sections of 7 DPA developing seed coat epidermal cells using fluorescence microscopy. Scale bar = 20 μm. (B) Transverse optical sections of 7 DPA developing seed coat epidermal cells of MUM2sp-Citrine, Citrine-ADPG2 and PelA-Citrine in wild type using fluorescence microscopy. Mucilage pockets were visualized using Citrine fluorescence and the plasma membrane was stained using FM 4-64. Arrows denote plasma membrane surrounding the mucilage pocket in MUM2-Citrine. Arrowheads highlight intercellular spaces in Citrine-ADPG2 and PelA-Citrine. Scale bar = 20 μm. (C) Scanning electron micrographs of mature seeds from wild type, Citrine-ADPG2 and PelA-Citrine. WT = wild type; M = mucilage pocket; C = cytosol; R = radial cell wall; PM = plasma membrane; Co = columella. Scale bar = 20 μm.
As HG comprises 10% of the extruded mucilage in wild type (reviewed in Šola et al. 2019), it was not expected that degrading HG would lead to a large decrease in mucilage halo size. To determine whether the reduced halos were caused by mucilage that adheres poorly to the seed surface and is removed during the staining procedure, seeds were videoed while being hydrated directly in ruthenium red. While wild-type seeds formed a mucilage halo within ∼10 s, almost no mucilage was released from Citrine-ADPG2 or PelA-Citrine seeds after 50 s of staining (Supplementary Fig. S6), indicating that extruded mucilage is not less adherent in these lines. Conversely, it is possible that reduced HG in seed coat mucilage could make it less easily extractable and result in a reduced mucilage halo. However, extruded mucilage was still markedly lower in ADPG2, Citrine-ADPG2, PelA and PelA-Citrine compared to wild type when seeds were treated with Na2CO3 (Supplementary Fig. S1B).
To establish if the mucilage phenotypes in lines expressing PelA were the result of PLA activity, the predicted catalytic arginine (R) was identified by aligning the full-length amino acid sequence of PLA (Q5BAU9) with three well-characterized pectin lyases (Q01172, Q00205 and P11073; Sánchez-Torres et al. 2003) using MUSCLE (Madeira et al. 2019). R256 was identified and mutated before the resulting chimeric construct (TBA2p:MUM2sp-PelA(R256A)-Citrine) was transformed into wild-type Arabidopsis. Mature seeds from 18 independent PelA(R256A)-Citrine lines tested had wild-type-like mucilage halos when hydrated in water (Fig. 3A), suggesting that PLA activity was responsible for the mucilage extrusion phenotypes in the PelA and PelA-Citrine lines (Fig. 3A, Supplementary Fig. S1A).
Seed coat epidermal cells expressing PelA or ADGP2 had abnormal morphology
To determine whether Citrine-ADPG2 and PLA-Citrine were localized to the mucilage pocket, we examined the sub-cellular localization of these fusion proteins using fluorescence microscopy during mucilage secretion at 7 DPA. As observed previously, MUM2sp-Citrine fluorescence localized exclusively to the mucilage pocket and radial cell walls but not the cytosol (Fig. 3A). In Citrine-ADPG2 and PelA-Citrine lines, fluorescence was observed in the radial wall at the position where the mucilage pocket normally forms but mucilage pockets were typically absent or greatly reduced in size (Fig. 3A). In addition, fluorescence from Citrine-ADPG2 and PLA-Citrine frequently appeared in the cytosol (Fig. 3A). The reduction in size and number of mucilage pockets suggests that secretion is disrupted early in the differentiation of seed coat epidermal cells in Citrine-ADPG2 and PelA-Citrine lines such that mucilage is not deposited. In contrast, seed coat cells expressing an inactivated version of PLA [PelA(R256A)-Citrine] showed wild-type-sized mucilage pockets, wild-type-like amounts of extruded mucilage and a fluorescence pattern comparable to MUM2sp-Citrine (Fig. 1A), indicating that the altered fluorescence in PelA-Citrine lines was likely to be due to PLA activity (Fig. 3A).
To investigate the cell structure in detail and to determine more clearly whether mucilage pockets were present in these transgenic lines, the plasma membrane of developing seeds at 7 DPA was visualized using the dye FM 4-64. In the wild-type-like epidermal cells of MUM2-Citrine lines, two distinct rings of plasma membrane that delineate the mucilage pocket can be seen in transverse sections (Fig. 3B, arrows). In contrast, this pattern was not observed in epidermal cells of Citrine-ADPG2 or PelA-Citrine lines, suggesting that typical mucilage pockets were not present. In addition, what appeared to be intracellular FM 4-64 signal was observed in many cells (Fig. 3B).
As described previously, MUM2-Citrine fluorescence was evident in the radial cell walls of adjoining cells in the wild-type-like cells of MUM2-Citrine lines (Fig. 3B, labeled R). The FM 4-64-stained plasma membranes of neighboring cells were also clearly visible, and at this resolution appeared as a single line (Fig. 3B, labeled PM). In contrast, regions without either Citrine or FM 4-64 fluorescence were often observed between adjacent cells in Citrine-ADPG2 and PelA-Citrine lines (Fig. 3A, B, arrowheads), suggesting that the radial cell walls were not interconnected and that plasma membranes from adjacent cells were further apart than in MUM2-Citrine lines. Taken together, these data suggest the presence of large intercellular spaces between developing epidermal cells in Citrine-ADPG2 and PelA-Citrine lines, which typically arise when cell–cell adhesion is disrupted (Bouton et al. 2002, Mouille et al. 2007).
Finally, we used SEM to image the seed surface of mature seeds. Unlike mature wild-type epidermal cells, which are hexagonal with raised radial walls and have a prominent columella, those of Citrine-ADPG2 appeared flattened, lacked a columella and were often not hexagonal, and those of PelA-Citrine were even more irregular, with a jagged surface that lacked discernible columellae or radial walls (Fig. 3C).
Taken together, these data indicate that expressing Citrine-ADPG2 or PelA-Citrine in seed coat epidermal cells led to abnormal morphology of seed coat epidermal cells, which included a lack of mucilage pocket formation, and apparent cell–cell adhesion defects.
Plasma membrane integrity was compromised in seed coat epidermal cells expressing ADPG2 or PelA
The plasma membrane dye FM 4-64 dye is internalized by endocytosis and transported by vesicles to the tonoplast if the staining time is extended (Aniento and Robinson 2005). Under our experimental conditions, FM 4-64 fluorescence was observed exclusively in the plasma membrane in the wild-type-like epidermal cells of MUM2-Citrine lines, but in Citrine-ADPG2 and PelA-Citrine lines, it was observed in the plasma membrane and internally in the majority of surveyed cells (Fig. 3B). This increased internalization could be due to increased endocytosis (Rigal et al. 2015) or loss of plasma membrane integrity (Evans and Cousin 2007). To differentiate between these two possibilities, developing seeds before TBA2p is active (4 DPA) and when TBA2p activity is strong (7 DPA; Tsai et al. 2017) were stained with propidium iodide (PI) and trypan blue, two dyes that are only internalized if the plasma membrane is compromised (Tran et al. 2011, Jones et al. 2016). At 4 DPA, PI stained the cell wall but not the cytosol of wild-type, Citrine-ADPG2 and PelA-Citrine seed coat epidermal cells, suggesting an intact plasma membrane (Fig. 4A). At 7 DPA, PI staining of Citrine-ADPG2 and PelA-Citrine developing seeds revealed extensive internalization of the dye, in contrast to wild-type seeds where the cell wall and mucilage pockets were stained, but no signal was visible in the cytosol (Fig. 4A). Similarly, developing seeds of wild type, Citrine-ADPG2 and PelA-Citrine were impermeable to trypan blue at 4 DPA (Supplementary Fig. S7), but by 7 DPA the plasma membrane was more permeable in Citrine-ADPG2 and PelA-Citrine seed coat cells than in wild type (Supplementary Fig. S7).
Plasma membrane integrity and evidence of early cell death in seed coats expressing HG-degrading enzymes. (A) Propidium iodide staining of developing seeds at 4 and 7 DPA in wild type, Citrine-ADPG2 and PelA-Citrine. (B) DAPI staining of nuclei in developing seeds at 7 DPA in wild type, Citrine-ADPG2 and PelA-Citrine. WT = wild type; M = mucilage pocket; C = cytosol; R = radial cell wall; N = nucleus. Scale bar = 20 μm.
Seed coat epidermal cells expressing ADPG2 or PelA exhibited signs of early cell death
Compromised plasma membrane integrity is characteristic of dying cells (Zhang et al. 2018). During cell death, chromatin in the nucleus undergoes condensation and the nuclear envelope is broken down, changes that result in the nucleus losing its regular shape and more intense staining of nuclear DNA by 4′,6-diamidino-2-phenylindole (DAPI) (Darzynkiewicz et al. 1992, Toné et al. 2007). In wild-type seed coat epidermal cells at 7 DPA, the nuclei appeared round and were not stained intensely with DAPI (Fig. 4B). In contrast, cells expressing Citrine-ADPG2 or PelA-Citrine had irregularly shaped nuclei that typically stained brightly with DAPI (Fig. 4B).
The data showing compromised plasma membrane integrity and DAPI staining of nuclei suggest that epidermal cells expressing Citrine-ADPG2 or PelA-Citrine were undergoing cell death at 7 DPA rather than continuing to develop to maturity.
Increased expression of plant defense genes was not detected in Citrine-ADPG2 lines
When PG or PLs degrade HG within the cell wall, short fragments of HG known as oligogalacturonides (OGAs) are released (Ferrari et al. 2013). Invading pathogens secrete HG-degrading enzymes that release OGAs and elicit plant defense responses (Ferrari et al. 2013). The defense genes PATHOGENESIS-RELATED1 (PR1), PATHOGENESIS-RELATED2 (PR2) and PHYTOALEXIN DEFICIENT3 (PAD3) are upregulated by exogenous application of OGAs (Ferrari et al. 2007, 2008, Galletti et al. 2008, Van Aubel et al. 2016, Gallego-Giraldo et al. 2020), and PR1 is also upregulated in plants overexpressing a PG from Aspergillus niger (Ferrari et al. 2008). ARABIDOPSIS DEHISCENCE ZONE POLYGALACTURONASE 1 (ADPG1), which has 70% amino acid sequence similarity with ADPG2, is required for the release of cell wall–derived elicitors that induce PR1 expression (Gallego-Giraldo et al. 2020). Since the expression of Citrine-ADPG2 caused cellular changes that are reminiscent of those triggered by plant defense, we examined whether PR1, PR2 and PAD3 transcripts were upregulated in Citrine-ADPG2 lines using quantitative reverse transcription polymerase chain reaction (RT-qPCR).
RNA from 5 DPA (as TBA2p activity begins) and 7 DPA (when TBA2p activity is high) siliques from wild type, the snc1 autoimmune mutant that constitutively overexpresses PR1 and PR2 (Zhang et al. 2003, Zhu et al. 2010) and three independent transformants carrying Citrine-ADPG2 were analyzed. No consistent and obvious differences in PR1, PR2 or PAD3 transcript abundance between wild type and Citrine-ADPG2 lines were detected, whereas the expression of PR1 and PR2 was upregulated in snc1 as expected (data not shown). To determine whether the upregulation of the target genes in seed coat epidermal cells was masked by inclusion of silique tissues, we analyzed RNA from developing seeds removed from siliques. The time points were adjusted to 3 DPA (prior to TBA2p activity) and 6 DPA (just prior to peak TBA2p activity; Tsai et al. 2017). Again, PR1 and PR2 expressions were increased in snc1, but consistent and convincing upregulation of PR1, PR2 or PAD3 was not detected in Citrine-ADPG2 lines (Supplementary Fig. S8).
As the RT-qPCR data do not provide clear evidence of PR1, PR2 or PAD3 defense gene upregulation, it is not clear whether the phenotypes of plants expressing Citrine-ADPG2 are mediated by OGA-induced expression of other gene targets or by other independent mechanisms.
Discussion
Arabidopsis seed mucilage is a specialized extracellular matrix that has unique properties due to its distinctive composition of cellulose, hemicellulose and pectin (reviewed in Šola et al. 2019). In this study, we explored the potential of using genetic engineering to manipulate the composition of mucilage polysaccharides to determine structure–function relationships within the extracellular matrix. Well-characterized cell wall–degrading enzymes were expressed and targeted to the apoplast of developing seed coat epidermal cells, and the effect on mucilage was examined. RglA and RglB are catabolic enzymes that degrade RG-I, the primary component of seed mucilage. Expression of these enzymes reduced seed mucilage by 80–90%, demonstrating that mucilage polysaccharides can be modified in a targeted manner. Expression of the ADPG2 and PLA enzymes that degrade HG disrupted cell–cell adhesion and resulted in premature cell death, highlighting that modifying even minor components of the extracellular matrix can give rise to unexpected and dramatic effects.
Targeting of Rhamnogalcturonan lyases to the mucilage pocket resulted in the degradation of mucilage RG-I
The expression and secretion of either of the two RGLs, RglA and RglB, resulted in similar phenotypes, characterized by the near absence of extruded mucilage in mature seeds (Fig. 1A, Supplementary Fig. S1), smaller mucilage pockets (Fig. 1A, Supplementary Fig. S2) and an 80–90% reduction in the RG-I backbone monosaccharides Rha and GalA when compared to wild type (Fig. 1C). These phenotypes are comparable to that of mum4 (Fig. 1A), an Rha biosynthetic mutant that synthesizes little RG-I in seed coat epidermal cells (Western et al. 2001, 2004, Usadel et al. 2004, Oka et al. 2007). The mucilage pocket shape and amount of extruded mucilage in transgenic lines expressing a catalytically inactive RglB appeared wild-type-like (Fig. 1A), suggesting that the observed phenotypes were specifically due to RglB catalytic activity. One hypothesis consistent with these data is that RglA and RglB are secreted to the mucilage pocket and consequently degrade mucilage RG-I, resulting in decreased mucilage and a concomitant decrease in the size of mucilage pockets. It is important to note that RglA and RglB are secreted at the same time as RG-I. Therefore, it is possible that RG-I degradation occurs anywhere from the site of synthesis in the Golgi (Young et al. 2008) to the site of deposition in the apoplast.
The low levels of Rha and GalA monosaccharides within the extruded mucilage of RglA-Citrine and RglB-Citrine seeds (Fig. 1C) suggest that the GalA→Rha disaccharides and other oligosaccharides generated from the hydrolysis of RG-I by RGLs (Bonnin et al. 2014) must be reabsorbed into the cytoplasm through the sugar salvaging pathway (Barnes and Anderson 2018). Reabsorption may occur through facilitated diffusion driven by membrane-bound sugar transporter proteins, such as those used to transport the disaccharide sucrose (Julius et al. 2017), or through endocytosis, as documented for di- and trisaccharides consisting of Ara residues (Baluška et al. 2005). In future, this could be investigated by introducing click chemistry-compatible Rha and GalA sugar analogs that, once incorporated into RG-I (Anderson et al. 2012), could be used to follow its fate.
Rgl transgenic seed coat epidermal cells have unusually shaped mucilage pockets
The mucilage pockets of seed coat epidermal cells of RglB and RglB-Citrine transgenic plants are unusually complex in shape, when compared to the larger pockets of wild-type cells (Fig. 1A, Fig. 2, Supplementary Figs. S3, S4, S5). In these transgenic lines, the plasma membrane lining the mucilage pocket is highly folded, resulting in a pocket with an alternating spoke-like pattern of shallow and narrow grooves and deep and wide grooves (Fig. 2B, Supplementary Figs. S4B, S5B). In addition, a ring of deeper mucilage pocket was observed immediately surrounding the cytoplasmic column (Fig. 2B, Supplementary Figs. S4B, S5B). Such complexity was not observed in the smaller mucilage pockets of seed coat epidermal cells of the mum4 mutant, which, like wild-type cells, had contents that were distributed uniformly (Fig. 1A) even though the mum4 mutant, like RglA-Citrine and RglB-Citrine transgenic plants, produces mucilage with low levels of RG-I. A possible difference between the mum4 mutant and the RglA and RglB transgenic plants that could account for the difference in mucilage pocket shape is the fact that the mum4 mutant does not synthesize or secrete RG-I to the mucilage pocket (Western et al. 2001, 2004, Usadel et al. 2004, Oka et al. 2007), whereas in the transgenic cells, RG-I must have been made and may have been secreted before it was degraded by RglA or RglB. Secretion would be expected to increase the surface area of the plasma membrane underlying the pocket through vesicular fusion. However, enzymatic degradation of the RG-I would prevent expansion of the pocket and force the membrane to fold as the surface area increased.
The exact cause of the radial pattern is unclear. It is possible that the pattern is simply the result of the most efficient way for an extended membrane to be folded into a small pocket as the RGI is degraded. Alternatively, the pattern may be predetermined by a non-uniform distribution of mucilage components in the pocket. For example, during secretion, vesicles carrying specific cargo such as RG-I may be targeted to a domain of the plasma membrane distinct from that of vesicles carrying other polysaccharides such as HG, resulting in a non-uniform (spokes on a wheel) distribution of mucilage components. In this case, degradation of RG-I could result in the observed shape of the mucilage pocket. However, to date, there is no evidence to suggest that any component of mucilage is arranged in this way, and how the pattern could be elaborated is also unclear.
One polysaccharide that does have a non-uniform distribution in seed mucilage is cellulose. Cellulose microfibrils are deposited along the outer surface of the cytoplasmic column plasma membrane such that they wrap around the cytoplasmic column parallel with the seed surface (Griffiths et al. 2015; reviewed in Šola et al. 2019). The presence of cellulose could exclude other components of mucilage from the region immediately adjacent to the cytoplasmic column and may explain the deep, low-fluorescence pockets immediately surrounding the columellae in RglB-Citrine lines (Fig. 2B, Supplementary Figs. S4B, S5B). Localized deposition of cell wall components is also known to occur in secondary cell walls, where lignin is deposited in specific bands in protoxylem tracheary elements (Zhong and Ye 2012). Further research is required to elucidate the cause of these unusually complex mucilage pockets.
Targeting of secreted HG-degrading enzymes to the mucilage pocket resulted in disrupted cell–cell adhesion and promotion of early cell death in seed coat epidermal cells
The expression and secretion of two distinct HG-degrading enzymes, ADPG2 and PLA, in seed coat epidermal cells led to complex cellular phenotypes that included a large reduction in mucilage deposition, decrease in the number and size of mucilage pockets, apparent loss of cell–cell adhesion, loss of plasma membrane integrity and apparent degradation of the nucleus (Fig. 3, Fig. 4, Supplementary Figs. S1, S6, S7). It is likely that all these changes derive directly or indirectly from the HG-degrading activity of ADPG2 or PLA because expression of an inactivated form of PLA [PelA(R256A)-Citrine] resulted in the loss of all aspects of the phenotype (Fig. 3A). How HG-degrading activity could lead to such cellular disruption is unclear. Since the middle lamella is HG-rich and plays an essential role in maintaining cell–cell adhesion (Bouton et al. 2002, Jarvis et al. 2003, Mouille et al. 2007), secretion of ADPG2 or PLA into the apoplast at the site of mucilage secretion could have resulted in the large intercellular spaces observed between adjacent seed coat epidermal cells (Fig. 3A, B). Similar cell–cell adhesion defects were found in mature dry seed coats in the gatl5 pectin biosynthesis mutant (Kong et al. 2013). Additionally, the occurrence of intercellular spaces also aligns with the role of ADPG2 in cell separation within the dehiscence and abscission zones during floral development (Ogawa et al. 2009).
The loss of integrity of the cell wall induced by HG-degrading enzymes could have been detected by mechanosensors that detect changes in membrane tension such as MscS-Like (MSL) 10 (Veley et al. 2014). Alternatively the loss of integrity may have caused impaired intercellular communication, which is necessary for coordinating cellular development (Long et al. 2015). Another possibility is that the detection of OGAs generated from the enzymatic hydrolysis of HG can trigger defenses against plant pathogens (Van Breusegem and Dat 2006, Galletti et al. 2008, Gallego-Giraldo et al. 2020).
Regardless of how the loss of cell wall integrity is detected, it is conceivable that the cell could have reacted by abandoning differentiation and undergoing premature cell death leading to the disruption of mucilage secretion, degradation of the nucleus and decrease in membrane integrity observed in this study. Further research will be needed to test this hypothesis and elucidate the undoubtedly complex chain of events that led to the phenotype observed in these transgenic lines. If induction of cell death through expression of HG-degrading enzymes is generally applicable, it could be used to remove specific cells or tissues for both basic and applied research. For example, seed mucilage-producing cells from crops such as Linum usitatissimum (flax), Sinapis alba (white mustard) and Salvia hispanica (chia) could be targeted since mucilage compromises seed oil quality and inhibits seed oil extraction (Balke and Diosady 2000, Fabre et al. 2015, Castejon et al. 2017). Indeed, it may be possible to target the ablation of any cell type using such an approach.
Conclusion and perspectives
Overall, this study has demonstrated the feasibility of using genetic engineering to modify specific components of the extracellular matrix in a non-essential single cell type. In the future, this approach could be used with both anabolic and catabolic enzymes to investigate the effects of modifying both the amount and composition of different polysaccharides in the extracellular matrix. It might also be possible to engineer seed mucilage of valuable crop species such as flax or mustards for agronomic purposes (Balke and Diosady 2000, Fabre et al. 2015).
Materials and Methods
Plant materials and growth conditions
The mum4-1 and mum2-1 mutants were isolated previously from an ethyl methanesulfonate-mutagenized population of Col-2 seeds (Western et al. 2001). Col-2 was the wild type employed throughout this study. mum4-1 plants transformed with MUM41.5Pro:secCitrine, a chimeric construct in the pAD binary vector (DeBono 2011) containing 1.5 kb of the MUM4 promoter (MUM41.5Pro; Dean et al. 2017) fused upstream of the signal sequence from an Arabidopsis chitinase gene (Batoko et al. 2000) in-frame with Citrine, a dimeric acidic stable variant of yellow fluorescent protein (Griesbeck et al. 2001), which was a gift from Yi-Chen Lee (2018). mum2-1 mutant lines complemented with TBA2p:MUM2-Citrine are described in McGee et al. (2019). Col-0 seeds containing VAMP721p:mRFP-VAMP721 were provided by Professor Masa H. Sato (Ichikawa et al. 2014).
Seeds were germinated on AT minimal medium plates (Haughn and Somerville 1986) at 20°C under continuous light (90–120 μmol m−2s−1 PAR). Ten-day-old seedlings were transplanted to soil (Sunshine 4 mix, Sun Gro Horticulture), watered once with liquid AT medium and grown under the same conditions.
To obtain seeds of the required developmental stage, pedicels of open flowers (0 DPA) were marked with non-toxic, water-soluble paint to allow the selection of developing siliques at the required age.
Plasmid construction and plant transformation
Constructs were generated using standard molecular cloning techniques as well as the Gateway cloning system (Thermo Fisher Scientific; Nakagawa et al. 2008, Oshima et al. 2011). Site-directed mutagenesis was performed according to the protocol detailed in Zheng et al. (2004). Details of construct assembly, all primer sequences and a complete list of constructs are given in the Supplementary Methods, and Supplementary Tables S1 and S2. Key vectors are available at Addgene (addgene.org).
Sanger sequencing was used to verify constructs before they were introduced into Agrobacterium tumefaciens GV3101 (pMP90) and used to transform Col-2 plants using the floral spray method (Chung et al. 2000). T1 seeds were selected on AT plates containing the appropriate antibiotics, and transgenic plants were genotyped for transgenes.
Microscopy
Extruded mucilage was visualized by hydrating mature seeds in distilled water or 20 mM Na2CO3 and shaking them at 130 rpm for 1 h using a platform shaker (Innova 2000, New Brunswick Scientific). Seeds hydrated with water or 20 mM Na2CO3 were washed twice with distilled water and then stained in 0.02% (w/v) ruthenium red (Cat. No. 00541, Sigma-Aldrich) by shaking at 130 rpm for 15 min. After staining, ruthenium red was replaced with distilled water before bright-field images were captured using a Zeiss Axioskop 2 microscope (Carl Zeiss) equipped with a CCD camera (DFC450 C, Leica) and LAS software (v4.2.0, Leica).
Trypan blue staining was performed using a staining solution containing a 1:1 (v/v) solution of water and 0.4% trypan blue (Cat. No. T10282, Gibco, Fisher Scientific). One valve was removed from each silique to expose the seeds before the silique was immersed in staining solution for 30 min, rinsed twice in 100% ethanol and then destained for 72 h in fresh 100% ethanol. The tissue was imaged using a digital camera (EOS Rebel T5, Canon) mounted to a dissecting microscope (Stemi 2000 C, Carl Zeiss) with an external light source (Ace EKE, Schott-Fostec).
Fluorescence microscopy was performed on developing seeds mounted in water and viewed through a 63× objective lens with 100% glycerol as an immersion medium using a spinning disk confocal microscope (UltraView VoX, PerkinElmer) equipped with an EMCCD camera (C9100-02, Hamamatsu) and Volocity software (Improvision). Citrine was excited at 514 nm and emission detected using a 540/30-nm filter. Developing seeds were stained for 5–10 min at room temperature (RT) with 5 μg ml−1 FM 4-64 (Thermo Fisher Scientific) or 2 mg/mL propidium iodide and for 5–10 min at RT in the dark with 10 μg ml−1 DAPI. FM 4-64 was excited at 514 nm, propidium iodide at 561 nm and DAPI at 405 nm. Emission was detected at 650/75 nm for both FM 4-64 and propidium iodide and at 460/50 nm for DAPI.
Bright field microscopy on embryos isolated from developing seeds were mounted in water and viewed through a 20× objective lens using the same confocal microscope described above. Overlapping images were stitched together using the MosaicJ plugin (Thévenaz and Unser 2007) for ImageJ (Schindelin et al. 2012). ImageJ was also used to add scale bars, and figures were produced using Inkscape (https://inkscape.org).
To visualize the seed surface, mature dry seeds were fixed to stubs using double-sided tape, sputter coated with 10 nm of gold-palladium using a 208 HR High-Resolution Sputter Coater (Cressington) and examined using an S-2600N SEM (Hitachi High-Tech) at an acceleration voltage of 8 or 20 kV.
High-pressure freezing and sectioning
Developing 7 DPA seeds were punctured using an insect pin (size 00) and transferred into Type B copper hats (Cat. No. 39201, Ted Pella, Redding, CA, USA) filled with 1-hexadecene. High-pressure freezing was performed immediately (HPM-100, Leica). Samples were then transferred under liquid nitrogen to cryovials containing the freeze-substitution media (2% (w/v) osmium tetroxide and 8% (v/v) 2-dimethoxypropane in anhydrous acetone). The cryovials were closed and placed in a dry ice and acetone slurry in a glass beaker and then sealed in a freeze-substitution box and stored at −80°C. After 5 d, the lid of the box was opened after which the box was strored at −20°C for 1 d. The following day samples were placed at 4°C for 2 h and then at RT, after which the freeze-substitution medium was removed and the samples were removed from their hats using an insect pin and washed five times with fresh anhydrous acetone. Samples were then infiltrated with increasing concentrations of Spurr’s resin (Spurr 1969) in acetone over the course of 4 d, as detailed in Quilichini et al. (2014). Samples were embedded in 100% Spurr’s resin in Beem capsules (Cat. No. 130, Ted Pella) and polymerized at 60°C for 2 d. Sections (0.5 μm) were prepared using a UC7 Ultramicrotome (Leica), stained with 1% toluidine blue in 1% sodium borate, mounted in Per mount (Thermo Fisher Scientific) and imaged using a VHX-5000 digital microscope (Keyence Corp.).
Monosaccharide analysis by high-performance anion-exchange chromatography
For each construct, three independent transformants were selected for analysis. For each independent transformant, seeds were harvested and pooled from eight plants. All lines to be compared were grown together in the same growth chamber. Additional biological replicates were grown in the same way in a different growth chamber.
For each genotype, four 20-mg aliquots of seed were taken from seed pools grown at the same time for each genotype and used to prepare and analyze extruded mucilage extracted with 20 mM Na2CO3, as described in Dean et al. (2019). The experiment was repeated with a second set of pooled seed grown independently. One representative dataset is shown.
To analyze alcohol insoluble residue (AIR), one 20-mg aliquot of seed was taken from each pool of independently grown seeds for each genotype, giving a total of three samples per genotype. Four 2-mm zirconium oxide beads (Next Advance, Inc.) were added to tubes containing the seeds before they were homogenized for 20 s at 6,000 rpm (Precellys 24, Bertin Technologies). All subsequent steps were performed as described in Dean et al. (2019). The experiment was repeated with seeds sampled in the same way from the same populations. One representative dataset is shown.
Monosaccharides were quantified by converting peak area to molar amount as described in Dean et al. (2007) using Excel 2007 (Microsoft). One of the four technical replicates of mucilage extracted from RglA-Citrine #13 was identified as an outlier using the Grubbs outlier test (Grubbs 1950), alpha = 0.05 (graphpad.com/quickcalcs/grubbs1), and all monosaccharide data generated from that replicate were subsequently removed from the analysis.
RNA extraction, RT-PCR and RT-qPCR
Transcript abundance of PR1, PR2 and PAD3 relative to GAPC were measured by RT-qPCR in three independent lines expressing Citrine-ADPG2. Silique RNA was extracted as detailed in Meisel et al. (2005), and data were analyzed using the ΔCT method (Livak and Schmittgen 2001). For full details, see Supplementary Methods.
Supplementary Data
Supplementary data are available at PCP online.
Data Availability
Sequence data from this article can be found in the GenBank/European Molecular Biology Laboratory databases under the following accession numbers: MUM2 (At5g63800, Q9FFN4), TBA2 (At1g62060, O04573), MUM4 (At1g53500, Q9LPG6), ADPG2 (At2g41850.1, Q8RY29), RglA (AN7135.2, Q5AX45), RglB (AN6395.2, Q5AZ85), PelA (AN2331.2, Q5BAU9), PR1 (At2g14610, P33154), PR2 (At3g57260, P33157), PAD3 (At3g26830, Q9LW27) and GAPC (At3g04120, P25858). Key vectors are available at Addgene ((addgene.org).
Funding
Natural Sciences and Engineering Research Council of Canada (NSERC) Discovery grants (to G.W.H., S.D.M. and Y.Z.); NSERC Collaborative Research and Training Experience (CREATE) Plant Responses to Eliminate Critical Threats (PRoTECT) program (to G.W.H. and Y.Z.); University of British Columbia Four-Year Fellowship (4YF) Tuition Award (to R.M.).
Acknowledgements
We thank Dr. Faride Unda for invaluable technical assistance with high-performance anion-exchange chromatography. We are grateful to the University of British Columbia BioImaging Facility for access to microscopes and assistance with imaging, and to Professor Masa H. Sato for providing VAMP721p:mRFP-VAMP721 seed.
Disclosures
The authors have no conflicts of interest to declare.
