Arabidopsis exocyst subunits SEC8 and EXO70A1 and exocyst interactor ROH1 are involved in the localized deposition of seed coat pectin

Authors

  • Ivan Kulich,

    1. Department of Experimental Plant Biology, Faculty of Science, Charles University, Vinicna 5, 128 44 Prague 2, Czech Republic
    Search for more papers by this author
  • Rex Cole,

    1. Department of Botany and Plant Pathology and Center for Genome Research and Biocomputing, Oregon State University, Corvallis, OR 97331, USA
    Search for more papers by this author
  • Edita Drdová,

    1. Institute of Experimental Botany, Academy of Sciences of the Czech Republic, Rozvojova 263, 165 02 Prague 6, Czech Republic
    Search for more papers by this author
  • Fatima Cvrčková,

    1. Department of Experimental Plant Biology, Faculty of Science, Charles University, Vinicna 5, 128 44 Prague 2, Czech Republic
    Search for more papers by this author
  • Aleš Soukup,

    1. Department of Experimental Plant Biology, Faculty of Science, Charles University, Vinicna 5, 128 44 Prague 2, Czech Republic
    Search for more papers by this author
  • John Fowler,

    1. Department of Botany and Plant Pathology and Center for Genome Research and Biocomputing, Oregon State University, Corvallis, OR 97331, USA
    Search for more papers by this author
  • Viktor Žárský

    1. Department of Experimental Plant Biology, Faculty of Science, Charles University, Vinicna 5, 128 44 Prague 2, Czech Republic
    2. Institute of Experimental Botany, Academy of Sciences of the Czech Republic, Rozvojova 263, 165 02 Prague 6, Czech Republic
    Search for more papers by this author

Author for correspondence:
Viktor Žárský
Tel: +420 221951685
Email: zarsky@ueb.cas.cz

Summary

  • Polarized deposition of cell wall pectins is a key process in Arabidopsis thaliana myxospermous seed coat development. The exocyst, an octameric secretory vesicle tethering complex, has recently been shown to be involved in the regulation of cell polarity in plants. Here, we used the Arabidopsis seed coat to study the participation of the exocyst complex in polarized pectin delivery.
  • We characterized the amount of pectinaceous mucilage and seed coat structure in sec8 and exo70A1 exocyst mutants. Using a yeast two-hybrid screen, we identified a new interactor of the exocyst subunit Exo70A1, termed Roh1, a member of the DUF793 protein family.
  • T-DNA insertions in SEC8, EXO70A1 caused considerable deviations from normal seed coat development, in particular reduced pectin deposition and defects in the formation of the central columella of seed epidermal cells. A gain-of-function mutation of ROH1 also caused reduced pectin deposition. Interestingly, we observed a systematic difference in seed coat development between primary and secondary inflorescences in wild-type plants: siliques from secondary branches produced seeds with thicker seed coats.
  • The participation of exocyst subunits in mucilage deposition provides direct evidence for the role of the exocyst in polarized cell wall morphogenesis.

Introduction

In some plant families (Solanaceae, Linaceae, Planta-ginaceae and Brassicaceae, which includes Arabidopsis), the outer layer of the seed coat consists of highly specialized cells producing pectinaceous mucilage, which is stored between primary and secondary cell walls. On seed hydration, this material absorbs massive amounts of water and extrudes, erupting through the cracked primary (outer) cell wall, resulting in a seed completely surrounded by a thick layer of mucilage. This property is known as myxospermy (Boesewinkel & Bowman, 1995).

The outer layer of the Arabidopsis thaliana seed coat consists of highly specialized ‘volcano cells’ (named according to their characteristic shape). This layer represents an interesting model for the study of localized and vectorial secretion processes, as the pectinaceous mucilage is exported from the epidermal cells via exocytosis (Ridley et al., 2001). Moreover, the availability of seed coat mutants has advanced this system as a popular model for developmental geneticists (Haughn & Chaudhury, 2005). The mucilage-modified 1 and 2 (mum1 and mum2) mutants produce mucilage, but are defective in seed coat release after hydration. Recently, it has been shown that mum2 encodes a putative β-galactosidase, important for the degradation of β-(1–4)-galactan side-chains of pectin, which increases the hydrophilic potential of rhamnogalacturonan I (Dean et al., 2007; Macquet et al., 2007b). Mutants mum3 and mum5 produce mucilage with altered composition and properties (Macquet et al., 2007a). Other mutations resulting in seed coat phenotypes alter transcription factors (AP2 – APETALA2, TTG1 – TRANSPARENT TESTA GLABRA1, TTG2/MYB61 and GL2 – GLABRA2), most of which (AP2, TTG1 and GL2) regulate the transcription of a putative pectin biosynthetic enzyme encoded by MUM4 (Western et al., 2004). Given that the mucilage secretory domain is delimited by a dense network of cortical microtubules, cytoskeletal mutants are also expected to show a mucilage production/deposition phenotype. One of them, mor1-1, has been shown to deposit a decreased amount of mucilage at its restrictive temperature (McFarlane et al., 2008).

Pectin deposition in the apoplast requires not only the correct synthesis of the mucilage precursor, but also appropriately asymmetrically localized secretion. Surprisingly, no secretory pathway mutants with a seed coat development-related phenotype have been described so far. However, involvement of the trans-Golgi apparatus has been implicated, because, although Golgi stacks, often containing mucilage, are randomly distributed in the cytoplasm of volcano cells, their proliferation is correlated with the initiation of mucilage secretion (Young et al., 2008).

The delivery of post-Golgi vesicles to the plasma membrane requires the cooperation of multiple components, such as SNARE (SNAP receptor) proteins, Rab, Rho and Arf guanosine triphosphatases (GTPases), and vesicle coat complexes. Among these core components, tethering complexes that mediate the docking of secretory vesicles at the target membrane appear to play a central role (Whyte & Munro, 2002). The exocyst complex, consisting of eight subunits – Sec3, Sec5, Sec6, Sec8, Sec10, Sec15 Exo70 and Exo84 – is such a tethering complex. In nonplant eukaryotes, it is known to facilitate the final stages of exocytosis and membrane recycling at the plasma membrane (Terbush et al., 1996). To accomplish this task, the exocyst acts as a downstream effector of small GTPases, crucial for the tethering of post-Golgi vesicles to the plasma membrane, and thus supporting polarized secretion (Lipschutz & Mostov, 2002; Novick & Guo, 2002).

Angiosperms have homologues of all exocyst subunits; the Arabidopsis genome encodes single copies of Sec6 and Sec8, two copies of Sec3, Sec5, Sec10 and Sec15, three copies of Exo84 and 23 copies of Exo70 (Eliášet al., 2003; Hála et al., 2008). Some of these subunits are essential for the correct growth of plant cells, for strongly polarized cell types such as pollen tubes or root hairs, which grow by means of tip growth. For example, sec8 mutations cause pollen germination and pollen tube growth defects, resulting in absolute pollen sterility in the case of strong alleles. The male-specific transmission defect can be rescued by the expression of Sec8 from a pollen-specific promoter (Cole et al., 2005). Mutants lacking EXO70A1, the most widely expressed Exo70 isoform, exhibit significantly reduced root hair growth, aberrant pollen with a sporophytically determined germination defect and other developmental defects, including decreased elongation of the stigmatic papillae (Synek et al., 2006). In both Arabidopsis thaliana and Brassica napus, Exo70A1 is crucial for pollen acceptance, and is specifically degraded during the self-incompatibility response of B. napus (Samuel et al., 2009). The roothairless1 (rth1) mutant of maize, which encodes a homologue of the SEC3 exocyst subunit, has a root hair phenotype similar to that of exo70A1 (Wen et al., 2005).

The plant exocyst, like the exocyst in other eukaryotes, is expected to be regulated by small GTPases, acting as effectors of both RAB and ROP (Rho of plants) GTPase families. The identification of a protein, ICR1, which interacts with both active ROP GTPases and the exocyst subunit SEC3, has provided the first insight into plant exocyst regulation by ROP GTPases (Lavy et al., 2007).

Here, we present evidence that two exocyst subunits (SEC8 and EXO70A1), together with a newly identified interactor of one of them (ROH1), participate in the polarized development of the mucilage-producing volcano cells of the seed coat. ROH1, which is a member of a previously uncharacterized, plant-specific gene family (characterized by the conserved domain designated DUF793), exhibits significant similarity to BYPASS1, a gene previously implied in root-to-shoot signal transmission (Van Norman et al., 2004). Our observations thus also provide a first clue to the possible cellular role of DUF793 family proteins.

Materials and Methods

Plants and culture conditions

Arabidopsis thaliana ecotype Columbia-0 (Col0) lines with T-DNA insertions were obtained from the SALK Institute (Alonso et al., 2003). The following SALK lines were investigated: roh1-p, SALK_082841; roh1-e, SALK_075815; roh1-d, SALK_133970. sec8-1c (sec8-1 mutant complemented by LAT52::SEC8) and sec8-4 have been described previously (Cole et al., 2005); exo70A1-2 was described in Synek et al. (2006). Plants were grown at 22°C under long-day conditions (LD 16 : 8 h) in soil or in peat pellets (Jiffy).

Genotype analysis

T-DNA insertions in ROH1, EXO70 and SEC8 genes were followed by PCR genotyping of individual plants, as described previously (Cole et al., 2005; Synek et al., 2006). roh1 mutants were genotyped using the following primers: LP_ROH1_E (GGAGCTGACTTGTCTTGTTGC) and RP_ROH1_E (TCCATGAGAAAATCGGAGATG) for roh1-e; LP_ROH1_P (CCCCATCGATTGTTGTCTA-TG) and RP_ROH1_P (CTTTGAATTCAGCTTCG-CAAC) for roh1-p; LP_ROH1-D (CTTCGGAGCC-GTAACTAAAGG) and RP_ROH1-D (TTAGGTGACG-GTAACGTGAGG) for roh1-d. The left border of the T-DNA sequence was detected using SALK primer LBb1.3.

Seed coat visualization and measurements

Seeds were hydrated in deionized water, inverting the tube several times. After 20 min of imbibition, a water solution of ruthenium red was added to a final concentration of 0.25 mg ml−1. Seeds were incubated at room temperature for 15–20 min, washed with deionized water and evaluated using light microscopy. Because of the high variability of seed coat size, we used at least two, in most cases three, generations of each mutant. Acido-alkali treatment was performed according to Macquet et al. (2007b). Seed coat size and starch grain size were measured in blind experiments; the results were evaluated by t-test in Microsoft Excel. All error bars shown in the figures represent standard errors.

Seed surface structural analysis

Siliques with developing seeds were fixed in 3% (w/v) glutaraldehyde, dehydrated using an ethanol–butanol gradient, embedded in paraffin and cut into 11 μm sections which were stained in paraffin with toluidine blue according to Sakai (1973) and observed under transmitted light on an Olympus BX51 (Olympus Europa GmbH, Hamburg, Germany) with an Apogee Alta U4000 (Apogee Instruments Inc., Roseville, California, USA) camera. To section adult seeds, these were methylated for 12 h using 0.1 M HCl in methanol at 60°C and microtomed at 12 μm using Shadon Cryotome® (77200226, Shandon Sci., Astmoor Runcorn, UK), mounted in glycerol and observed as paraffin sections.

Scanning electron microscopy (SEM) of mature seeds was performed on a JEOL JSM-6380 scanning electron microscope (JEOL Ltd. Tokyo, Japan) with a pressure of 25 Pa and voltage of 10 kV; coating was not necessary.

Yeast two-hybrid screening and assays

As a host strain, Saccharomyces cerevisiae Y190 was used. EXO70A1 cDNA was subcloned into the pGBT9 vector and used as bait to screen a Nicotiana tabacum pollen library (provided by Benedikt Kost, Warwick University, Wellesbourne, Warwickshire, UK). Transformation was performed according to the Clontech (Takara Bio Europe/Clontech, Saint-Germain-en-Laye, France) MATCHMAKER GAL4 Two-Hybrid User Manual (PT3061-1), with selection on media lacking leucine (Leu), tryptophan (Trp) and histidine (His) and containing 30 mM 3-amino-1,2,4-triazole (3-AT). Clones positive with both histidine selection and β-galactosidase activity assay were isolated, restriction mapped, and a representative of each structure was sequenced.

To demonstrate the interaction of Exo70A1 with the Arabidopsis homolog of NtRoh1, the single-exon AtROH1 gene was cloned after PCR amplification from Arabidopsis genomic DNA using the following primers: forward, AGAGGATCCCCAAACAAAATCATGAGACCTG; reverse, TTCGAAAGGATCCCCCAA TAATTCAAAACT. PCR was performed at an annealing temperature of 58°C with Sigma AccuTaq™ DNA polymerase.

Pull-down assay

Escherichia coli BL25 Codon+ was transformed with 6 × HIS:Exo70A1 (pCAT), GST:ntROH (pGEX5x-3) and empty vector containing glutathione S-transferase (GST) (pGEX5x-3). For recombinant protein expression, overnight inocula were diluted 1 : 10 with Luria–Bertani (LB) medium, grown for 3 h and induced overnight with 0.1 mM isopropyl-β-d-thiogalactopyranoside (IPTG) at 4°C. Cells were then centrifuged, resuspended in Lysis Buffer (50 mM NaH2PO4; 300 mM NaCl; 10 mM imidazole; pH 8.0) and sonicated using a Bandelin UW 2070 sonicator (BANDELIN electronic, Berlin, Germany) at 40% capacity. Extracts (0.5 ml) were centrifuged to remove cell debris and incubated for 2 h at 4°C with Ni2+-nitrilotriacetate (NiNTA) beads (Qiagen NiNTA Agarose, lot No. 11555873) and glutathione beads (Glutathione Sepharose™ 4B, Amersham Biosciences, Uppsala, Sweden, lot No.297134). The beads were then washed four times with Washing Buffer (50 mM NaH2PO4; 300 mM NaCl; 20 mM imidazole; pH 8.0). To elute bound proteins, Elution Buffer (50 mM NaH2PO4; 300 mM NaCl; 250 mM imidazole; pH 8.0) was used. Samples were then heat denatured and resolved by standard sodium dodecylsulfate-polyacrylamide gel electrophoresis (SDS-PAGE) on a 10% polyacrylamide gel.

Bioinformatics

Searches of the GenBank (http://www.ncbi.nlm.nih.gov) and Joint Genome Institute (http://www.jgi.doe.gov/) databases were performed as described previously (Dvořákováet al., 2007), with the ROH1 protein sequence as a query. Alignments were constructed and phylogenetic trees computed from the protein sequence data using the neighbor-joining (NJ) and maximum likelihood (ML) methods, as described in Dvorakova et al. (2007).

Semiquantitative reverse transcription-polymerase chain reaction (RT-PCR)

Total RNA was isolated from freshly harvested inflorescences using the Qiagen RNEasy Plant Mini kit. Reverse transcription and subsequent PCR were performed as described previously (Dvorakova et al., 2007) with the following modifications. Primers ROH1_RTfor (CGCAAGTAATC-AAGGCTCG) and ROH1_RTrev2 (GAAACTTTCCC-GGCTTGCTC) were used for the detection of the ROH1 transcript; primers GAPC_F2 (CACTTGAAGGGTG-GTGCCAAG) and GAPC_R (CCTGTTGTCGCCAAC-GAAGTC) (Kerschen et al., 2004) were employed for the detection of the internal standard – glyceraldehyde-3-phosphate carboxylase mRNA; 0.5 μl template per 10 μl reaction was used, and PCR was performed in a Biometra T-personal cycler (Biometra GmbH, Goettingen, Germany) with the following temperature profile: 2 min initial denaturation at 96°C; 30 cycles of 30 s at 92°C, 90 s at 58.6°C, 90 s at 72°C; final extension of 5 min at 72°C. Reaction products were separated by agarose electrophoresis with visualization using the SYBR Green stain, and photographed on blue light irradiation on a DarkReader transilluminator (Clare Chemical Research, Dolores, Colorado, USA).

Results

Mutants in two exocyst subunits, sec8 and exo70A1, exhibit reduced seed coat mucilage thickness

To investigate the putative role of the exocyst complex in seed coat mucilage secretion, we selected T-DNA insertion mutants in two exocyst subunits that produce viable seeds that germinate and develop into healthy plantlets even in the homozygous state. Six previously described mutants of the SEC8 subunit (Cole et al., 2005) have T-DNA insertions at different positions within the coding region, with sec8-1 being most N-terminal. This mutant is probably a null allele, and is able to produce homozygotes only after complementing its pollen development (and subsequent male transmission) defects using a LAT52::SEC8 transgene, active predominantly in pollen (Cole et al., 2005). We refer to this line as sec8-1c (complemented in pollen). Two other alleles, sec8-4 and sec8-6, are C-terminal insertions, and each causes only a partial male transmission defect (Cole et al., 2005). The second exocyst subunit mutant investigated was exo70A1-2. This mutant has a reduced ability to produce viable seeds in the genetic background of the Col0 ecotype; however, after crossing exo70A1-2 into the Landsberg erecta (Ler) ecotype background, we obtained homozygous mutant plants that produced viable seeds in amounts similar to the sec8-1c line.

As shown in Figs 1 and 2, the extruded mucilage layers of all sec8 mutants are thinner than those of the wild-type. This phenotype ranges from only marginally reduced thickness in the sec8-6 mutants to a severe phenotype deviation in the sec8-1c mutant line, where ruthenium red staining revealed that no mucilage eruption had occurred. We hypothesized that there may still be some pectinaceous mucilage present in the sec8-1c seed coat, but its quantity may be insufficient to crack the outer cell wall open on imbibition. To test this hypothesis, we performed acido-alkali treatment of the seeds, which weakens the cell wall to facilitate mucilage release (Macquet et al., 2007b), and revealed that a very small amount of mucilage was still present in the mutant (Fig. 3). To verify that this phenotype is caused by the mutation of SEC8, we complemented sec8-1 mutants with the SEC8 cDNA under its native promoter. In the complemented mutant, the extrusion of mucilage and the development of volcano cells was restored to levels similar to those of the wild-type (see Supporting Information Fig. S1).

Figure 1.

 Quantitative analysis of seed coat mucilage thickness after imbibition, based on staining with ruthenium red. In the mutants sec8-6, sec8-4, sec8-1c, exo70A1-2, roh1-p and roh1-e, thicknesses are significantly smaller than the control (P for sec8-6 and sec8-4 are 0.0022 and 2.6 × 10−6, respectively; all other mutant values are smaller than 1 × 10−10). Col0/Ler primary – seeds from primary inflorescence; Col0/Ler secondary – seeds from secondary inflorescence; Col0/Ler combined – seeds from whole plant. For mutants and Ler, only data from the pooling of seed from primary and secondary inflorescences are shown. Error bars represent standard errors. By t-test, seed coat mucilage is significantly thinner in wild-type primary than wild-type secondary in both Col0 and Ler (P = 5.12 × 10−14 and 3.6 × 10−5, respectively).

Figure 2.

 Seed coat mucilage after imbibition of wild-type, sec8, exo70A1-2 and roh1-e mutants, stained by ruthenium red. Wild-type Arabidopsis ecotypes Landsberg erecta (Ler) and Columbia-0 (Col0); roh1-e, sec8-1c and sec8-4 are in Col0; exo70A1-2 is in the Ler background. Scale bar in Ler is valid for all panels.

Figure 3.

 Seed coat after acido-alkali treatment, stained with ruthenium red. Scale bar in Col0 is valid for all panels.

The exo70A1-2 seeds also produced significantly less mucilage (Figs 1 and 2), but the phenotype was less dramatic than that of sec8 mutants. This might be a result of partial redundancy with other Exo70 paralogs. Sectioning of seeds revealed obvious differences between wild-type and exo70A1-2 mutants at the heart stage of embryo seed development (Fig. 4). At this stage, mutants exhibit much smaller pectinaceous pockets, together with larger starch grains (average grain diameter of 4.3 μm in the mutant and 3.4 μm in the wild-type, t-test P value of 2 × 10−5); moreover, starch grains are localized outside the central part of future columella, that is in an area which is devoid of starch in developing wild-type volcano cells.

Figure 4.

 Sections of exo70A1-2 mutant seeds in heart stage (5 d after anthesis) of seed development with wild-type Ler. Mutant seeds have less pectinaceous extracellular matrix (em) on their outer surface. Arrows point at starch grains.

Exocyst mutants show altered seed surface pattern

A defect in the generation of seed coat mucilage is not necessarily linked with a change in seed appearance (Dean et al., 2007). Yet, the reduction of mucilage deposition could be accompanied by alterations in the structure of the seed epidermis. Sectioning (Fig. 5) and SEM (Fig. 6) of the seed surface revealed that mutations in both SEC8 and EXO70A1 loci cause obvious deviations from the typical, more or less hexagonal shape of volcano cells with a mucilage deposition domain surrounding the central columella. Both exo70A1-2 and all three sec8 mutants have more rectangular cells, and the seed surface is generally flattened. In sec8 mutants, the columella is mostly absent and develops only rarely and in an aberrant form, whereas, in exo70A1-2 seeds, the columella is present but appears flat and extended. Cross-sections of mature wild-type and mutant seeds confirmed that both the sec8-1c and sec8-4 lines form a flat epidermal layer with very small pectin compartments, The columella either does not develop, or is present in only a rudimentary form (Fig. 5).

Figure 5.

 Sections of mature seeds of wild-type (WT) and mutants. Mucilage (m) surrounds the columella (c). Brown layer consists of seed coat layers underneath volcano cells. Scale bar is 20 μm long and valid for all panels.

Figure 6.

 Scanning electron microscopy images of seed coat surfaces. Scale bars are valid for entire columns.

AtEXO70A1 interacts with ROH1, a paralog of BYPASS 1

To identify possible regulators and effectors of exocyst, we screened yeast two-hybrid libraries for Exo0A1 interactors, and identified several candidates in an N. tabacum pollen library. As T-DNA insertion mutants for Arabidopsis homologues of these interactors were available, we used seed coat mucilage in these mutants as a marker for the evaluation of the biological significance of our interactors. Here, we focus on one such interactor (GenBank Accession EU596604), found as two independent clones in the yeast two-hybrid screen. Its A. thaliana homologue, At1g63930, also interacts with Exo70A1 and, importantly, Arabidopsis plants that are mutant for this gene exhibit a seed coat developmental defect. We propose naming this gene ROH1 (from the Czech word ‘roh’, which means a corner), referring to the characteristic pattern of pectin deposition to the corners of the cross-sectioned volcano cells (Fig. 5, marked as ‘m’). To verify interaction of ROH1 with EXO70A1, we used both yeast two-hybrid and in vitro pull-down assays (Fig. 7). We also observed an interaction between Roh1 and the paralog Exo70C1 in the yeast two-hybrid assay (data not shown), suggesting that ROH1 may be a promiscuous interactor with more than a single protein of the Exo70 family.

Figure 7.

 Interaction between NtRoh1 and Exo70A1. Top – a yeast two-hybrid experiment, growth in restrictive conditions (–His plates; 3-aminotriazole; AD, activation domain construct; BD, binding domain construct). Bottom – pull-down assay using 6 × HIS:Exo70A1 (black arrows), GST:ntROH1 and GST (white arrows) with Ni2+-nitrilotriacetate (NiNTA) beads [demonstrating that 6 × HIS:Exo70A1 protein does not bind glutathione S-transferase (GST)] and glutathione (GS) beads (demonstrating binding of 6 × HIS:Exo70A1 with GST:ntRoh1, but not free GST).

ROH1 (At1g63930) is a single-exon gene encoding a 415-amino-acid protein containing a conserved domain of unknown function, DUF793 (InterPro: IPR008511). Database searches revealed at least 10 additional proteins containing DUF793 in Arabidopsis (At1g01550, At1g18740, At1g22030, At1g43630, At1g74450, At1g77855, At2g46080, At4g01360, At4g11300, At4g23530), numerous relatives in other angiosperms, two cDNA sequences from pine and six members of the family in the moss Physcomitrella patens (see Figs 8, S2), but no relatives in algae (including Prasinophyta) or nonplant species. None of the Arabidopsis members of the family have been functionally characterized so far, with the exception of At1g01550, which is identical to BYPASS1 (BPS1), a gene identified on the basis of a mutation affecting root–shoot signaling. The molecular mechanisms underlying the developmental phenotype of bps1 mutants remain unknown, although evidence suggests that BPS1 is involved in the production and/or transmission of a carotenoid-derived signal that inhibits shoot growth from the root (Van Norman et al., 2004; Van Norman & Sieburth, 2007). However, a phylogenetic tree constructed using both NJ and ML methods places ROH1 on a branch of the DUF793 family distinct from that harboring BPS1 (Fig. 8).

Figure 8.

 An unrooted neighbor-joining (NJ) tree of DUF793-containing proteins from several plant species (At, Arabidopsis thaliana; Ipni Ipomoea nil; Nb, Nicotiana benthamiana; Nt, Nicotiana tabacum; Pp, Physcomitrella patens; Ps, Picea sitchensis; Pt, Populus tremula; Os, Oryza sativa; Vv, Vitis vinifera). With the exception of the ROH and BPS genes, specifically discussed in this paper, Arabidopsis and rice sequences are denoted by standard locus identifiers, moss sequences by JGI database Protein IDs, and the remaining sequences by GenBank or Uniprot accessions preceded by species abbreviations. Nicotiana benthamiana NbBPS1 and NbBPS2 correspond to GenBank accessions ABX26124 and ABX26123, respectively. Bootstrap values for the tree shown (out of 500 replicates) are indicated above each branch; bootstrap values of a maximum likelihood (ML) tree from the same data are shown below the branches. The ML version of four significantly different branches that disagreed with the NJ tree is shown by dashed lines, with bootstrap values in italics. The presumed position of the only partially sequenced NtROH1 gene isolated in our primary two-hybrid screen (GenBank EU596604) is shown by dotted brackets; the ROH1 and BPS1 subfamilies are shaded.

Expression of ROH1 affects seed coat formation

We obtained three mutants with T-DNA insertions in the ROH1 gene, roh1-p (insertion in ROH1 promoter sequence), roh1-d (insertion in the coding sequence) and roh1-e (insertion in the 3′-untranslated region immediately downstream of the stop codon, erroneously annotated as residing within the CDS; see Fig. 9). Intriguingly, plants carrying either of the two mutant alleles with insertions outside of the coding sequence produce significantly less mucilage than wild-type seeds (P < 1 × 10−15) (see Figs 1 and 2). As in the case of stronger sec8 mutant alleles, the mucilage amount is often insufficient to crack fully the outer cell wall after imbibition, but can be released using acido-alkali treatment (Fig. 3). However, unlike the exocyst subunit mutants, roh1 mutants do not exhibit alterations in seed surface structure detectable by SEM or light microscopy (Figs 5 and 6). The effects of both mutations were semidominant, with heterozygous plants exhibiting an intermediate phenotype (see Fig. S3).

Figure 9.

 Top: map of the ROH1 gene with sites of T-DNA insertions (CDS marked by a box, transcript by a black arrow). Open arrows denote T-DNA insertion sites (p corresponds to the roh1-p allele, e to roh1-e and d to roh1-d). Bottom: expression of ROH1 in wild-type and mutant inflorescences, as detected by reverse transcription-polymerase chain reaction (RT-PCR) [with glyceraldehyde-3-phosphate dehydrogenase subunit C (GAPC) as internal standard]. For each allele, the azygous sister control (WT) is shown. Note that the position of the T-DNA insertion in the roh1-d allele excluded its detection in the RT-PCR experiment.

As roh1-p and roh1-e T-DNA insertions reside outside of the ROH1 open reading frame (ORF), the phenotype might be due to alterations in the expression level of the ROH1 gene. According to publicly available microarray data (Zimmermann et al., 2005; Winter et al., 2007), ROH1 is co-expressed with several members of the Exo70 family (see Fig. S4); it is mainly expressed in cells expanding in a polar manner (pollen, root hair, etc.), further suggesting its involvement in a polarized secretory pathway. In the microarray data, ROH1 exhibits a very low level of expression in most above-ground organs, with relatively higher levels in the inflorescences (possibly because of the presence of pollen in these samples). Nevertheless, we were unable to detect the transcript in wild-type inflorescences via standard semiquantitative RT-PCR. However, we readily detected the ROH1 signal in inflorescences of both roh1-p and roh1-e mutants under identical conditions, indicating that the T-DNA insertions indeed cause the upregulation of the transcript (and, possibly, an accompanying overexpression of the Roh1 protein), rather than the loss of function of the ROH1 gene (Fig. 9), consistent with their semidominant phenotype.

To compare a possible knockout mutation with these two semidominant alleles, we assessed plants carrying the third allele (roh1-d), as the insertion disrupts the reading frame. Unlike roh1-e or roh1-p, roh1-d mutants did not exhibit any significant differences from the wild-type in either seed coat structure or mucilage layer thickness (data not shown). This further supports the idea that roh1-e and roh1-p phenotypes are a result of misregulation rather than loss of function of the gene.

Primary and secondary inflorescences produce seeds with different seed coat size

During the course of our seed coat measurements, we noticed that seeds from secondary Arabidopsis inflorescences and inflorescence branches exhibited a consistently thicker seed coat than those from primary inflorescences. This phenomenon was consistently observed in both Col0 and Ler wild-types (Fig. 1) and in mutant plants (not shown). However, in all cases, differences between wild-types and mutants were sufficiently prominent to allow the pooling of seeds from whole plants, albeit at the expense of somewhat increasing the observed variation (Fig. 1).

Discussion

As already proposed by others (Western et al., 2000), the development of the mucilage coat of myxosperm seeds (such as those of Arabidopsis) provides a good model for studying mechanisms of polarized secretion. In the present study, we showed that two subunits of the exocyst, a major regulator of eukaryotic exocytosis, are crucial for correct seed coat development. Although we have focused so far only on exocyst subunits Sec8 and Exo70, it is likely that other subunits of this complex will also be important for seed coat development, as the exocyst complex is a functional biochemical entity in angiosperms (Hála et al., 2008). Although EXO70 is encoded by multiple genes, the EXO70A1 gene included in this study presents a highly expressed, ‘housekeeping’ member of the EXO70 gene family, whose loss cannot be fully compensated by other paralogs (Synek et al., 2006). The observed differences in seed development in Col0 and Ler backgrounds might be attributed to varying expression levels of other paralogs – for example, EXO70A2 is more strongly expressed in Ler than in Col0 according to publicly available microarray data (Winter et al., 2007). In addition, we have shown that a newly identified interactor for Exo70A1, Roh1, can also influence the same process (accumulation of seed coat mucilage), apparently when it is misregulated, and possibly overexpressed, in the seed.

The observed defects in the seed coat of exocyst and roh1 mutants can be interpreted as a result of defects in the process of exocytosis. Lack of mucilage in the exocyst mutants correlates with other exocyst phenotypes, suggesting a defect of secretion-dependent cell expansion, such as decreased root hair length, pollen tube growth rate or hypocotyl expansion in the dark (Cole et al., 2005; Wen et al., 2005; Synek et al., 2006; Hála et al., 2008). Mutant seeds either produce less pectin, or are delayed in transferring pectins to the cell wall, secondarily slowing down seed coat morphogenesis. If the exocyst functions as a docking/tethering complex in seed coat pectin deposition, it is likely to be specifically targeted to the ‘corner’ domain of the seed epidermal cell, which develops prominent and very special cortical assembly of microtubules (McFarlane et al., 2008). The aberrant shape and underdevelopment of the columella raise the possibility that correct secretion of mucilage in seed coat cells might be crucial for the rise of the columella, as the columella is formed by a thick secondary cell wall, synthesized subsequent to mucilage secretion. This possibility is also implied by Western et al. (2000). Alternatively, the lack of noticeable changes in columella formation in the two roh1 mutants, which have altered mucilage accumulation, indicates that columella development is not completely dependent on the presence of mucilage.

Interestingly, we noticed significant differences between the amount of mucilage in seed coats formed on the primary and secondary inflorescences, even in wild-type plants. Apart from an obvious methodical aspect (i.e. differences in seed coat size have to be taken into account when evaluating phenotypes), these differences might be evolutionarily meaningful. In agreement with the general survival strategy of A. thaliana as an opportunistic ruderal plant, it may be advantageous to produce seeds at first as quickly as possible (and with a smaller energy investment) in order to generate at least some progeny. However, later, after the development of more source biomass, higher quality seeds surrounded by more mucilage can be produced, perhaps resulting in increased tolerance to transient water deficit during germination, which may contribute to the broadening of the available niche.

In mammals and yeast, the exocyst is controlled by multiple regulators, including, in particular, small GTPases. In plants, the situation is likely to be similar, at least in part. In order to identify possible regulators of the plant exocyst, we performed a yeast two-hybrid screen using the Exo70A1 subunit as bait, and identified an interacting protein, Roh1, whose mutational misregulation (probably overexpression) results in a seed coat developmental defect reminiscent of that seen in the exocyst subunit mutants. These results, together with expression data that suggest co-regulation between ROH1 and several EXO70 family genes (Fig. S1), support a functional link between ROH1 and the exocyst. However, mutational disruption of the ROH1 ORF (roh1-d) causes no readily observable phenotype, and, in particular, no alterations in the seed coat. This raises the possibility that ROH1 may not have any function in the seed coat in the wild-type, but that its misexpression in these cells interferes with exocyst function, thus inducing the observed phenotype. Alternatively, ROH1 function in the seed coat may be obscured in the knockout mutant by redundancy, as ROH1 is a member of a family of at least 11 Arabidopsis genes, including two closely related paralogs. According to public microarray data, one of these (At4g11300) is expressed in all organs except ripening siliques (Winter et al., 2007), and thus could easily replace a mutant ROH1. However, given the wide array of phenotypes associated with exocyst mutants, we cannot explain why the semidominant roh1 alleles produce viable progeny with no drastic growth, developmental or other readily observable phenotypes apart from seed coat reduction. It should be noted that certain aspects of the roh1 mutants have yet to be evaluated in detail, in particular, pollen transmission rates and the quantification of root hair growth.

ROH1 and its relatives contain a conserved, plant-specific domain previously annotated as DUF (domain of unknown function) 793. Whilst searching for additional members of the family, we noticed that a somewhat divergent version of this domain can also be found in the Arabidopsis BPS1 (BYPASS1) gene of unknown molecular role which has been identified on the basis of a mutation that caused aberrant development of leaf blades and their vasculature. The primary defect of bps1 loss-of-function mutants is thought to be caused (at least in part) by the overproduction of a carotenoid-based, graft-transmissible signal molecule in the root, which is taken up in the shoot to affect its development (Van Norman et al., 2004; Van Norman & Sieburth, 2007). It is tempting to speculate about a possible role for Bps1 as an inhibitor of a secretory event that is crucial for the production, transmission or perception of this signal. In this view, interaction of Bps1 with the exocyst would allow the control of exocytosis for this process. Indeed, our observations are consistent with another DUF793-containing protein, Roh1, which can act as a negative regulator of secretion, as its presumed overexpression results in seed coat reduction similar to that seen in exocyst loss-of-function mutants. However, as ROH1 and BPS1 represent different branches of the extensive family of DUF793-containing proteins, it is possible that each may regulate different targets, possibly even in a different manner. Nevertheless, the present report provides evidence of a biologically relevant molecular interaction with a protein of the DUF793 family and suggests that it may have the potential to inhibit secretion.

A recently published detailed analysis of the mucilage secretion domain of the plasma membrane in Arabidopsis seed coat cells has defined a specific docking domain for secretory vesicles with pectinaceous cargo (McFarlane et al., 2008). As Exo70 exocyst subunits are known to function as a landmark for vesicle docking in yeast and mammals (Boyd et al., 2004; Wu et al., 2009), its interactor Roh1 might influence this possible landmark function of the plant Exo70A1 protein.

Acknowledgements

The authors would like to thank Z. Vejlupkova and H. Pham (Oregon State University, Corvallis, OR, USA) for assistance with Arabidopsis growth and genotyping, Olga Votrubova, Drahomira Bartakova and Martin Mazuch (Charles University, Prague, Czech Republic) for the introduction into seed anatomy techniques and SEM, and Marta Cadyova (Charles University) for technical assistance. We thank the Salk Institute Genomic Analysis Laboratory for providing the sequence-indexed Arabidopsis T-DNA insertion mutants, and an anonymous reviewer for valuable suggestions. The work in the laboratory of VZ was supported by the Ministry of Education of the Czech Republic grants (MSM Kontakt ME841 – collaboration with US partner, MSM0021620858 and MSM LC06034 ‘REMOROST’). The work in the laboratory of JF was supported by the US National Science Foundation (#IBN-0420226 and IOS-0920747), which also supported this international collaboration.

Ancillary

Advertisement