The Arabidopsis Callose Synthase Gene GSL8 Is Required for Cytokinesis and Cell Patterning^,,

Abstract

Cytokinesis is the division of the cytoplasm and its separation into two daughter cells. Cell plate growth and cytokinesis appear to require callose, but direct functional evidence is still lacking. To determine the role of callose and its synthesis during cytokinesis, we identified and characterized mutants in many members of the GLUCAN SYNTHASE-LIKE (GSL; or CALLOSE SYNTHASE) gene family in Arabidopsis (Arabidopsis thaliana). Most gsl mutants (gsl1–gsl7, gsl9, gsl11, and gsl12) exhibited roughly normal seedling growth and development. However, mutations in GSL8, which were previously reported to be gametophytic lethal, were found to produce seedlings with pleiotropic defects during embryogenesis and early vegetative growth. We found cell wall stubs, two nuclei in one cell, and other defects in cell division in homozygous gsl8 insertional alleles. In addition, gsl8 mutants and inducible RNA interference lines of GSL8 showed reduced callose deposition at cell plates and/or new cell walls. Together, these data show that the GSL8 gene encodes a putative callose synthase required for cytokinesis and seedling maturation. In addition, gsl8 mutants disrupt cellular and tissue-level patterning, as shown by the presence of clusters of stomata in direct contact and by islands of excessive cell proliferation in the developing epidermis. Thus, GSL8 is required for patterning as well as cytokinesis during Arabidopsis development.

Cytokinesis divides the cytoplasm of a plant cell by the deposition of plasma membrane and a cell wall during late mitosis. This process requires the phragmoplast, a dynamic, plant-specific cytoskeletal and membranous array, which delivers vesicles containing lipids, proteins, and cell wall components to the division plane to construct the cell plate. Cell plate formation involves several stages: initiation through vesicle fusion, the formation of a tubular-vesicular network, a transition to a solely tubular phase, and then further fusion to form a fenestrated sheet (Samuels et al., 1995). The outward growth of the cell plate leads to its fusion with the parental cell wall (Jürgens, 2005a, 2005b; Backues et al., 2007).

Key regulators of cytokinesis include KNOLLE, KEULE, KORRIGAN, and HINKEL, which when defective induce pleiotropic phenotypes and seedling lethality (Lukowitz et al., 1996; Nicol et al., 1998; Zuo et al., 2000; Assaad et al., 2001; Strompen et al., 2002). KNOLLE, a syntaxin homolog, is required for the fusion of exocytic vesicles via a SNARE/SNAP33 complex (Lukowitz et al., 1996; Heese et al., 2001). KEULE, a homolog of yeast Sec1p, regulates syntaxin function by interacting with KNOLLE (Waizenegger et al., 2000; Assaad et al., 2001). KORRIGAN is an endo-1,4-β-glucanase required for cell wall biogenesis during cytokinesis (Zuo et al., 2000). And HINKEL is a kinesin-related protein required for the reorganization of phragmoplast microtubules during cytokinesis (Strompen et al., 2002).

Additional regulators include Formin5, TWO-IN-ONE (TIO), and Arabidopsis (Arabidopsis thaliana) dynamin-like proteins (ADLs; Kang et al., 2001, 2003; Hong et al., 2003; Collings et al., 2005; Ingouff et al., 2005; Oh et al., 2005). Formin5 localizes to the cell plate and is an actin-organizing protein involved in cytokinesis and cell polarity. TIO, a Ser/Thr protein kinase, functions in cytokinesis in plant meristems and in gametogenesis (Oh et al., 2005). Members of the Arabidopsis DRP family associate with the developing cell plate, whereas DRP1a (ADL1A) locally constricts tubular membranes, interacts with callose synthase, and may facilitate callose deposition into the lumen.

Callose, a β-1,3-glucan polymer with β-1,6-branches (Stone and Clarke, 1992), is synthesized in both sporophytic and gametophytic tissues and appears to play various roles. Callose accumulates at the cell plate during cytokinesis, in plasmodesmata, where it regulates cell-to-cell communication, and in dormant phloem, where it seals sieve plates after mechanical injury, pathogen attack, and metal toxicity (Stone and Clarke, 1992; Samuels et al., 1995; Lucas and Lee, 2004).

Twelve GLUCAN SYNYHASE-LIKE (GSL) genes (also known as CALLOSE SYNTHASE [CalS]) have been identified in the Arabidopsis genome based on sequence homology (Richmond and Somerville, 2000; Hong et al., 2001; Enns et al., 2005). A GSL that functions in callose deposition after injury and pathogen treatment is GSL5 (Jacobs et al., 2003). Five other members of the Arabidopsis GSL family are required for microgametogenesis. GSL1 and GSL5 act redundantly to produce a callosic wall that prevents microspore degeneration, and both are needed for fertilization (Enns et al., 2005). GSL2 is required for the callosic wall around pollen mother cells, for the patterning of the pollen exine (Dong et al., 2005), and for callose deposition in the wall and plugs of pollen tubes (Nishikawa et al., 2005). GSL8 and GSL10 are independently required for the asymmetric division of microspores and for the entry of microspores into mitosis (Töller et al., 2008; Huang et al., 2009).

Callose is a major component of the cell plate, especially during later plate development (Kakimoto and Shibaoka, 1992; Samuels et al., 1995; Hong et al., 2001). Callose appears to structurally reinforce the developing cell plate after the breakdown of the phragmoplast microtubule array and during plate consolidation (Samuels and Staehelin, 1996; Rensing et al., 2002). It is likely that callose is synthesized at the cell plate rather than in the endoplasmic reticulum and in the Golgi (Kakimoto and Shibaoka, 1988). GSL6 (CalS1) appears to be involved in callose synthesis at the cell plate, since a 35S∷GFP-GSL6 fusion in transgenic BY-2 tobacco (Nicotiana tabacum) cells increases callose accumulation, and GFP fluorescence was found specifically at the cell plate (Hong et al., 2001). However, functional and genetic data on the role of any GSL in Arabidopsis sporophytic cytokinesis are still lacking.

Here, we report that GSL8 (CalS10) is required for normal cytokinesis. In addition, gsl8 mutants exhibit excessive cell proliferation and abnormal cell patterning, phenotypes not previously reported for cytokinesis-defective mutants.

RESULTS

Isolation of Homozygous T-DNA Insertion Lines of GSL Family Members

We previously isolated T-DNA insertional mutants in 11 members, GSL1 to GSL9, GSL11, and GSL12, of the Arabidopsis GSL family and screened these primarily for gametophytic phenotypes (Huang et al., 2009). Homozygous T-DNA-tagged lines were identified for 11 members of the GSL family except for GSL10 (Supplemental Table S1), which was gametophytic lethal (Töller et al., 2008; Huang et al., 2009), thus explaining why gsl10 homozygous mutants could not be isolated. To identify the possible functions of different GSL genes in sporophytic callose formation and in cytokinesis, we examined gross and microscopic seedling phenotypes in homozygous gsl mutants.

We first examined gsl6 insertion lines because GSL6 has been reported to be involved in cell plate-specific callose synthesis (Hong et al., 2001). Two independent homozygous gsl6 insertion lines showed developmental phenotypes similar to those of wild-type plants (Supplemental Fig. S1, A–C). These two alleles are likely to be nulls, since no GSL6 mRNA could be detected by reverse transcription (RT)-PCR using a primer set corresponding to the catalytic domain of callose synthase (Supplemental Fig. S1D). Like Hong and Verma (2007), we were unable to detect any cytokinesis defects in gsl6 insertion lines, such as multiple nuclei in one cell and cell wall stubs (data not shown). Thus, if GSL6 regulates cytokinesis, its function might be masked by redundancy in the GSL gene family.

We then evaluated the rosette-stage phenotypes of mutations in the remaining 10 GSL family members. Mutants in nine of the remaining family members were indistinguishable from those of wild-type controls (Supplemental Fig. S1E).

gsl8 Mutations Affect Plant Development, Resulting in Seedling Lethality

Mutations in the remaining locus, GSL8, have been reported to be male gametophytic lethal, with aberrant gsl8 pollen failing to enter pollen mitosis I (Töller et al., 2008), leading these authors to speculate that the role of GSL8 during pollen development is independent of its function in callose synthesis. However, as briefly noted by Huang et al. (2009), three homozygous T-DNA insertions, gsl8-1, gsl8-2, and gsl8-3, still allow some sporophytic growth (see below). Similar phenotypes were also found in three newly isolated insertional alleles (Figs. 1 and 2). Previously reported gsl8 seedling viability problems might be attributable to poor germination and/or to the low transmission of mutations (Supplemental Table S2), since growth on 3% Suc resulted in the recovery of homozygous mutants at a rate of 9% to 15% instead of the 25% expected in the case of normal Mendelian segregation. The recovery of homozygous mutants in a medium without Suc was dramatically reduced to about 2.5%. This reduction was mainly caused by poor germination of homozygous mutants (Supplemental Table S2). Six independent homozygous mutant alleles (gsl8-1 to gsl8-6) were seedling lethal and displayed almost identical multiple phenotypes (Fig. 2).

Figure 1.

Characterization of GSL8 T-DNA insertion lines. A, The GSL8 gene contains 45 exons in the coding region. Insertion sites are numbered for T-DNA lines: Salk_111094 (gsl8-1), SAIL_679_H10 (gsl8-6), GABI_851C04 (gsl8-2), SALK_057120 (gsl8-3), SALK_109342 (gsl8-4), and SAIL_21_B02 (gsl8-5). B, RT-PCR of total RNA showing the presence of a 500-bp PCR product against the 3′ coding region from wild-type (WT) plants but none from the T-DNA lines. ACTIN2 was used as an internal control. [See online article for color version of this figure.]

Figure 2.

Phenotypes of gsl8 mutant seedlings and roots. A, Ten-day-old light-grown seedlings on MS plates containing 3% Suc. Mutants are seedling lethal and display short, wide, and serrated roots and irregular cotyledon margins. Numbers in A and C denote gsl8 mutant alleles. B, Higher magnification of gsl8 mutant cotyledons. C, gsl8 mutant seedlings grown in the dark (7 d) have short, wide, and deformed hypocotyls and roots. D and E, Higher magnifications of dark-grown cotyledons from wild-type (D) and gsl8 mutant (E) plants. Bars = 5 mm in A and C and 1 mm in B, D, and E. WT, Wild type. [See online article for color version of this figure.]

gsl8 mutants showed a dwarf phenotype, and seedlings exhibited abnormally shaped cotyledons (Fig. 2A). The surface of the gsl8 cotyledon epidermis was undulated, and hypocotyls and roots were thicker and shorter than those of the wild type (Fig. 2, A–C). Dark-grown etiolated gsl8 seedlings showed open cotyledons, fat hypocotyls, and reduced height compared with wild-type seedlings; however, hypocotyl length was significantly longer than that of light-grown gsl8 plants (Fig. 2, C–E). Because cotyledons and hypocotyls are embryonic organs, we investigated the effect of gsl8 on embryo development. Compared with wild-type controls (Fig. 3, A–E), gsl8 disrupted embryonic development, with shape defects beginning at the globular stage (Fig. 3F) and increasing in severity over time (Fig. 3, G–I). Mature gsl8 embryos showed a range of aberrant phenotypes, including asymmetrically shaped cotyledons and single or triple cotyledon-like organs (Fig. 3I). For example, the percentage (12.9%) of abnormal gsl8-3 embryos was similar to that of homozygous mutants observed at the seedling stage (Supplemental Table S2).

Figure 3.

Embryo development in wild-type and gsl8 mutant plants. Seeds were harvested from wild-type (A–D) and gsl8-4 homozygous mutant (E–I) plants at different developmental stages and then cleared and viewed using Nomarski optics. Abnormal phenotypes are detectable at the globular (F) and heart (H) stages as well as in mature embryos (I and J). The other five gsl8 mutant alleles showed similar phenotypes to those of gsl8-4. Embryos in E and I were cleared from the same gsl8-4 heterozygous siliques. Embryos in H and J were dissected away from surrounding seed tissues. Bars = 50 μm in A to D, F, and G and 100 μm in E and H to J.

gsl8 Seedlings Display Pleiotropic Phenotypes, Including Patterning Defects

The gsl8 dwarfing could be due to a disruption in cell division frequency and/or elongation. Normally, the shoot apical meristem displays well arranged layers (L1–L3) as well as cells that are small and relatively even (Fig. 4A). By contrast, gsl8 meristem layers were disrupted, shapes were irregular, and many cells were enlarged (especially in L3 and in young primordia; Fig. 4B), suggesting a loss of meristematic activity. In addition, 3-week-old callus induced from gsl8 seedlings was much smaller than that in the wild type (Supplemental Fig. S2), again indicating a growth defect.

Figure 4.

Tissue phenotypes in gsl8 mutant seedlings. A to D, Light micrographs of sections of shoot apical meristems (A and B) and cotyledons (C and D). Compared with the wild type (A and C), gsl8-4 meristem (B) and mesophyll (D) tissues are disorganized. The arrow in D shows a large intercellular space. E to H, Cross sections of hypocotyls (E and F) and longitudinal sections of roots (G and H) from the wild type (E and G) and the gsl8-4 mutant (F and H). I, TEM of columella cell layers in gsl8-4. The arrowhead indicates the quiescent center. J and K, TEM of hypocotyls in cross section of the wild type (J) and gsl8-4 (K) with endodermal cells numbered. L and M, TEM of the root tip in the wild type (L) and gsl8-4 (M). Co, Columella cells; CR, cortex; CRi, internal layer of cortex; CRo, outer layer of cortex; EN, endodermis; EP, epidermis; PE, pericycle; QC, quiescent center; S, sieve elements; T, tracheary elements. Similar patterns were observed from gsl8-1 and gsl8-2 mutants. Bars = 20 μm in A to D and I, 30 μm in E to H, and 10 μm in J to M.

Cross sections of cotyledons revealed that gsl8 mutants showed not only bulging and giant epidermal cells but also misarranged mesophyll cells with larger intercellular spaces (Fig. 4, C and D). Venation patterns were also defective in cotyledons and young leaves (Supplemental Fig. S3, A and B). Compared with the wild type, gsl8 hypocotyls showed excess divisions in the pericycle, endodermis, and phloem (Fig. 4, E–J and K). Similar phenotypes were found in roots, with extra cells appearing in the root columella (Fig. 4, G–I) and cortex (Fig. 4, L and M). Cell layering was also disrupted in gsl8 roots, especially outside the stele and the columella (Fig. 4, L and M), and many root cells were abnormally enlarged. gsl8 roots also exhibited aberrant cell division planes (Supplemental Fig. S3C). Thus, GSL8 is directly or indirectly required for cell number and spatial distribution.

GSL8 Is Required for Epidermal Patterning

The epidermis of young gsl8 cotyledons and leaves was highly disorganized (Fig. 5, B–E). Nonstomatal cells showed a much wider range in size and in aspect ratio than in the wild type (e.g. some cells were exceptionally long and narrow). In addition, cell outlines were often less sinuous (Fig. 5, B–E).

Figure 5.

GSL8 is required for stomatal and epidermal patterning. A, Developing wild-type epidermis showing distributed asymmetric divisions and relative uniformity in cell size. B to E, Developing gsl8 epidermis of leaves and cotyledons. Stars indicate clusters of developing or mature stomata or of three adjacent meristemoids in the lower part of B. Arrows in D and E indicate regions containing abnormally high numbers of small cells. Ten-day-old plants were used. All images show cell walls stained with propidium iodide and visualized by confocal microscopy. All propidium iodide images have inverted gray scales. Bars = 25 μm.

Stomata also showed extreme variations in size as well as defects in pore formation and bilateral symmetry (Fig. 5, C–E). In all six gsl8 mutant alleles, many gsl8 stomata formed in contact in clusters, whereas normally stomata were spaced at least one cell apart (Fig. 5A; Supplemental Fig. S4). Clusters in gsl8 varied in the number and arrangement of stomata as well as in overall cluster shape and outline. Strikingly, the gsl8 epidermis showed regions with an excessive number of small cells, many of which divided asymmetrically to produce stomatal precursor cells (Fig. 5, C and D). The regions with small cells were often adjacent to groups of larger and more mature epidermal cells (including stomata). By contrast, in the developing epidermis of wild-type leaves, small cells and asymmetric divisions were much more distributed and interspersed with larger cells. Thus, GSL8 appeared to be required for correctly distributing asymmetric divisions in the stomatal cell lineage throughout the epidermis.

Each asymmetric division in Arabidopsis stomatal development generates a meristemoid, a precursor cell that ultimately produces a stoma. Mutants of gsl8 appeared to produce many more meristemoids in contact than wild-type plants (Fig. 5B). Normally, meristemoids themselves divide asymmetrically, divisions that can restore a one-celled spacing pattern between stomatal precursor cells in wild-type plants (Supplemental Fig. S5). However, gsl8 meristemoids seemed to divide fewer times than in the wild type; this defect, along with the abundance of meristemoids in islands of small cells, appeared to induce pattern violations. Because stomatal patterning might involve cell-cell signaling via the apoplast, we asked whether spacing defects were only found in cells whose apoplast was abnormal. As in other gsl8 epidermal cells (see below), those in the stomatal lineage often showed gaps in their walls. However, many stomata and meristemoids in contact had apparently intact cell walls throughout their cell depth (data not shown), suggesting that local cytokinesis defects were not directly responsible for stomatal pattern violations.

GSL8 Is Widely Expressed

Data from the AtGenExpress expression atlas (Schmid et al., 2005) show that GSL8 is widely expressed in most organs throughout plant development. To determine the organ-level expression pattern of GSL8, we performed RT-PCR using RNA extracted from various organs. GSL8 transcripts were detected in all organs and regions tested (Supplemental Fig. S6). GSL8 expression was relatively higher in actively dividing cells, such as in root tips and floral buds, than in fully expanded or expanding tissues, such as rosette and cauline leaves, cotyledons, and flowers. Expression was also detected in mature organs, such as rosette leaves of 6-week-old plants.

gsl8 Mutants Exhibit Cell Wall Stubs and Incomplete Cytokinesis

GSL8 encodes a putative callose synthase, and callose is the major component of developing cell plates during plant cytokinesis. We thus examined whether gsl8 mutants disrupt cytokinesis, as evidenced by defects such as cell wall stubs and two or multiple nuclei in one cell that have previously been described for other genes (Söllner et al., 2002). We first examined the epidermal cells of gsl8 mutant cotyledons. As shown in Figure 6, A to C, cell wall stubs were observed throughout the depth of cotyledon epidermal cells stained with propidium iodide. Transmission electron microscopy (TEM) analysis confirmed the presence of binucleated cells with or without cell wall stubs in gsl8 cotyledons and roots (Fig. 6, D and E). To determine whether embryo abnormality in gsl8 mutants might result in part from defective cytokinesis, we used TEM to examine abnormal embryos at various developmental stages from siliques produced by heterozygous gsl8 plants. Perturbations similar to those found in the seedling stage were found in embryos that presumably were homozygous for gsl8. These defects included disorganized cell files, cell wall stubs, and binucleated embryo cells (Fig. 6F). Together, these results show that gsl8 mutations disrupt cell division and that GSL8 is required for Arabidopsis cytokinesis starting in the early embryo.

Figure 6.

Aberrant cell walls and two nuclei in one cell in gsl8-4 mutants. A to C, Cell wall stubs in epidermal cells visualized using propidium iodide staining and confocal microscopy. Optical sections from outer (A), middle (B), and bottom (C) parts of the same cells are shown. All propidium iodide images have inverted gray scales. The other five gsl8 mutant alleles showed similar results to those of gsl8-4. D to F, TEM of cotyledons (D) and root tips (E) from a 3-d-old gsl8-4 mutant and of a gsl8-4 embryo (F). Arrowheads and arrows indicate cell wall stubs. N, Nucleus. Bars = 10 μm in A to C and 2 μm in D to F. [See online article for color version of this figure.]

Reduced Callose Deposition at Cell Plates and New Cell Walls in gsl8

Callose localization by aniline blue staining or immunolocalization in root tips has been used frequently to localize cell plates and/or young daughter cell walls (Samuels et al., 1995; Frantzios et al., 2001). To test whether the loss of GSL8 activity in gsl8 mutants correlated with reduced callose deposition at cell plates, root tips were stained with aniline blue. In wild-type roots, we detected prominent aniline blue staining at positions that mark a cell plate (Fig. 7, A–C). Callose is also assumed to be deposited at new cell walls, since we sometimes observed callose bands on two opposite sides of cells, suggesting that at least one of them is new cell wall (Fig. 7, B and C). In gsl8 mutants, we failed to find clear callose bands at these positions, although we sometimes found ectopic callose deposition (Fig. 7, D–F). This failure to detect callose might be caused by a low frequency of cell divisions. To produce plants with a range of cytokinesis defects, we constructed inducible RNA interference (RNAi) lines for GSL8. A dexamethasone-induced reduction in GSL8 mRNA levels produced seedlings that resembled gsl8 homozygotes in their dwarfism and defective cell patterning (Supplemental Fig. S7). These lines also resembled severe gsl8 alleles in the absence of callose staining at root cell walls. Some induced plants showed relatively mild developmental phenotypes and the presence of callose at cell plates and/or the new cell walls, but their frequency and intensities were less than in noninduced lines (Fig. 7, G–I). Taken together, these observations suggest that GSL8 is involved in the deposition of callose at cell plates during cytokinesis.

Figure 7.

Callose deposition at cell plates and new cell walls. A, Comparison of images of the same roots using fluorescence optics from aniline blue staining of the wild type (bottom root) and a dexamethasone-treated GSL8 RNAi line (top root). Bar = 30 μm. B to I, Confocal images following aniline blue and FM4-64 staining from root tips of 1-week-old plants of the wild type (B and C), the gsl8 mutant (D–F), and a dexamethasone-induced GSL8 RNAi line (G–I). Merged images of aniline blue (blue) and FM4-64 fluorescence (red) are shown in B, E, and H. Gray-scale images of the blue channel are shown in C, F, and I. Bright-field images are shown in D and G. Arrowheads indicate representative aniline blue staining. Boxes indicate callose bands on two opposite sides of a cell, suggesting that at least one of them is a new cell wall that originated from the cell plate. Bars = 20 μm.

DISCUSSION

To obtain genetic data on the role of callose in cytokinesis, we identified T-DNA insertional mutants for most GSL callose synthase genes in Arabidopsis. We have shown that GSL8 is required for cytokinesis, as evidenced by the presence of incomplete cell walls and two nuclei in a single cell in seedling tissues and by reduced callose in cell plates and in new cell walls of gsl8 mutants or inducible RNAi lines of GSL8. In addition, GSL8 was found to play a novel and pervasive role in cell and tissue patterning.

GSL8 Is Critical for Cytokinesis in the GSL Gene Family

Callose deposition has been extensively studied in plants. Green algae, bryophytes, ferns, and seed plants are all capable of producing wound-induced callose, and all plant taxa investigated, from multicellular green algae to land plants, accumulate callose during cell division (Scherp et al., 2001). At least some callose deposition at the cell plate is not wound induced, since rapidly frozen cells assayed using callose-specific antibodies show an increase in staining during cell plate development (Samuels et al., 1995). Callose deposition can be indirectly inhibited by caffeine, which destabilizes the developing cell plate and generates cytokinesis defects (Samuels and Staehelin, 1996; Yasuhara, 2005). However, while many mutants have been described that affect cytokinesis, no mutant in a callose synthase gene has been shown to control cytokinesis (Hong and Verma, 2007). Our findings that gsl8 alleles and knockdown lines display severe cell division defects and reduced callose deposition at cell plates and new cell walls provide direct functional genetic evidence for the importance of callose synthesis in cytokinesis by GSL members. Although these gsl8 alleles are nulls and induce cytokinesis defects, cell divisions still continue in these mutants, raising the possibility that other GSL genes might be functionally redundant. Two candidates are GSL10 and GSL6. GSL10 encodes a callose synthase that is most similar to that of GSL8. gsl10 mutants are gametophytic lethal, and they perturb the symmetry of microspore division and induce irregular callose deposition during microgametogenesis (Töller et al., 2008; Huang et al., 2009). In addition, the silencing of GSL10 in transgenic RNAi lines resulted in a dwarfed growth habit, suggesting additional functions for GSL10 in normal plant growth (Töller et al., 2008).

Although it remains to be seen if GSL6 localizes to the cell plate in Arabidopsis, it was previously shown that a GFP-GSL6 fusion protein expressed under the control of a 35S promoter localizes to the cell plate in tobacco BY-2 cells (Hong et al., 2001). However, GSL6 does not appear to be essential for cytokinesis, since our null gsl6 mutants did not exhibit the seedling-level phenotypes found in gsl8 (Supplemental Fig. S1B) and since cytokinesis defects appear to be absent from gsl6 alleles (Hong and Verma, 2007; this study). Similarly, gsl6 does not appear to disrupt callose accumulation at the cell plate (Hong and Verma, 2007). These data suggest that GSL6 is either not required for cytokinesis or is redundant with GSL8.

Tissue Organization in gsl8

Cytokinesis mutants often display defects in tissue organization, in the range of cell sizes, and in division plane placement (Lukowitz et al., 1996; Nicol et al., 1998; Zuo et al., 2000; Assaad et al., 2001; Strompen et al., 2002). Such defects were common in gsl8 mutants (e.g. cells and intercellular spaces in the shoot apex and in cotyledons were enlarged abnormally). However, gsl8 also exhibits abnormal cell overproliferation, such as an increased number of cells in specific tissues and layers in the root cap columella, in the hypocotyl cortex and endodermis, and in the phloem. This overproliferation phenotype is especially severe in the epidermis (see below). Since an overproliferation phenotype does not appear to have been reported for any cytokinesis mutants (Lukowitz et al., 1996; Nicol et al., 1998; Zuo et al., 2000; Assaad et al., 2001; Strompen et al., 2002), these cytokinesis and division restriction functions of GSL8 might be due to different underlying processes. GSL8 was expressed in all organs tested (Supplemental Fig. S6). Its expression was not restricted to actively dividing tissues and was also detected in mature rosette leaves that were no longer dividing. These data raise the possibility that GSL8 is involved in cellular processes in addition to cytokinesis.

GSL8 in Stomatal and Epidermal Development

To our knowledge, GSL8 is the only locus identified to date that is strongly involved in both cytokinesis and stomatal patterning. Other mutants that disrupt cytokinesis show a more normal stomatal distribution (Söllner et al., 2002; Falbel et al., 2003), suggesting that patterning and cytokinesis are mostly independent processes.

Normally, stomata follow the “one-cell-spacing” rule, in which two stomata are separated by at least one intervening nonstomatal cell (Geisler et al., 2000; Bergmann and Sack, 2007). In Arabidopsis, this spacing pattern is primarily established when an asymmetric division places a new stomatal precursor (a meristemoid) away from a preexisting stoma and only secondarily when the meristemoids in contact divide away from each other (Supplemental Fig. S5; Geisler et al., 2000; Lucas et al., 2006). Several genes control these events, including those encoding likely receptors and ligands, such as TOO MANY MOUTHS, ERECTA, and EPIDERMAL PATTERNING FACTOR1. Mutations in these genes induce the formation of clusters of stomata in contact (Nadeau and Sack, 2002; Shpak et al., 2005; Bergmann and Sack, 2007; Hara et al., 2007). These clusters tend to be distributed similarly throughout the epidermis, and nonstomatal epidermal cells mostly resemble those in wild-type plants. However, unlike the above mutants, gsl8 clusters vary greatly in the arrangement and size of stomata. In addition, the gsl8 epidermis displays global disruptions in the distribution of epidermal cells, such as islands of small cells that harbor excess asymmetric divisions and meristemoids in contact. Perhaps this dense packing of small cells inhibits division and the correction of pattern violations.

Clearly, GSL8 is also required for properly organizing the distribution, planes, and number of other tissues and cell types consistent with a required and ubiquitous role in cytokinesis throughout the plant. However, it remains to be determined how likely defects in callose deposition might induce these abnormal morphologies.

In summary, GSL8 encodes a putative callose synthase required for cell plate formation during sporophytic cytokinesis. In addition to typical cytokinesis-defective phenotypes, the loss of function of GSL8 induces or results in novel developmental phenotypes, including defects in epidermal planarity and patterning, stomatal spacing, and cell type overproliferation.

MATERIALS AND METHODS

Plant Materials and Growth Conditions

All Arabidopsis (Arabidopsis thaliana) mutant lines and wild-type plants used in this study are in the Columbia background. Unless otherwise indicated, seeds were surface sterilized (20% commercial bleach and 0.1% Triton X-100) for 5 min, washed three times in sterile water, and stored at 4°C for 3 d. The seeds were sprinkled on 0.6% agar-solidified medium containing 1× Murashige and Skoog (MS) salts and 0% to 3% Suc (pH 5.8). Growth conditions were at 25°C with a 16-h-light/8-h-dark cycle, either in soil or on MS plates. gsl8-1 (SALK_111094), gsl8-2 (GABI_851C04), gsl8-3 (SALK_057120), gsl8-4 (SALK_109342), gsl8-5 (SAIL_21_B02), and gsl8-6 (SAIL_679_H10) were obtained from the Arabidopsis Biological Resource Center or the Nottingham Arabidopsis Stock Centre. To determine the site of T-DNA insertion in all of these alleles, we sequenced the PCR products amplified using the appropriate left border primer and gene-specific primer. Primer sequences for all PCR-based genotyping are listed in Supplemental Tables S3 and S4. The homozygosity of all mutant alleles of GSL8 was verified by PCR genotyping and RT-PCR. Calli were induced from cotyledons excised from 1-week-old wild-type and gsl8 seedlings and were grown on medium described previously (Encina et al., 2001).

RNA Extraction and RT-PCR

Plant tissues were harvested and total RNA was isolated using TRIzol (Invitrogen) according to the manufacturer's instructions. A total of 1 μg of DNA-free RNA was reverse transcribed using an oligo(dT) primer and reverse transcriptase (M-MLV RTase; SolGent). The PCR amplification protocol was 96°C for 3 min followed by 30 cycles of 94°C for 30 s, 55°C for 45 s, and 72°C for 30 s. The ACTIN2 transcript was used as an internal control. The primer sequences were as follows: AtGSL8-d1 (5′-CCCTCTGGATTTGAATGGCA-3′) and AtGSL8-r1 (5′-GGCTAAAGAGAGTAGAGCCC-3′) for GSL8 and ACTIN2-d1 (5′-TCAATCATGAAGTGTGATGTGG-3′) and ACTIN2-r1 (5′-TTAGAAACATTTTCTGTGAACGAT-3′) for ACTIN2.

Microscopy

For phenotypic analysis of the mutant embryos, developing seeds were collected from siliques at different developmental stages from wild-type and heterozygous gsl8 plants. They were cleared with a drop of Hoyer's solution (50 g of chloral hydrate, 3.75 g of gum arabic, and 2.5 mL of 100% glycerol in 15 mL of distilled water) for at least 30 min at room temperature. Cleared seeds were examined with an Olympus FV1000 confocal microscope equipped with Nomarski differential interference contrast optics.

For histological and TEM examination, intact tissue samples were fixed in 2% glutaraldehyde in 50 mm phosphate-buffered saline (pH 6.8) at 4°C for 12 to 24 h, followed by postfixation in 1% OsO₄ buffer at 4°C overnight. Samples were then sequentially dehydrated through an ethanol series, embedded in Epon-812 resin, and polymerized at 60°C for 48 h. Both semithin (1 μm) and ultrathin (70 nm) sections were cut using an Ultracut microtome (EM UC6; Leica). Semithin sections stained with toluidine blue O were examined with a light microscope (Axioplan 2; Zeiss), and photographs were taken with a digital camera (AxioCam HRc; Zeiss). Ultrathin sections were mounted on slots and then stained with uranyl acetate followed by lead citrate for imaging with a Philips TECNAI 12 transmission electron microscope.

Propidium iodide staining was performed as described previously (Lucas et al., 2006) using a Nikon C1 Plus confocal laser scanning microscope. Root tips were stained with callose staining buffer for 3 to 5 min, which is a mixture of 0.1% aniline blue in autoclaved triple-distilled water and 1 m Gly, pH 9.5, at a volume ratio of 2:3. Aniline blue was examined using a fluorescence microscope (Axioplan 2; Zeiss). For costaining, root tips were stained with the dye FM4-64 (Molecular Probes) according to a previous study (Fujimoto et al., 2008) before aniline blue staining. Aniline blue and FM4-64 fluorescence was examined with 405- and 523-nm lasers using a fluorescence confocal microscope (LSM5; Zeiss).

Constructs and Plant Transformation

To generate the dexamethasone-inducible GSL8 RNAi construct, the entire GSL8 RNAi cassette comprising sense and antisense arms of GSL8 interspersed by two introns (PDK and CAT) was amplified from Agrikola GSL8-sepcific hairpin RNA expression plasmids (CATMA 2a35110) using the primers dsGSL8-SalI-d1 (5′-GGGCGTCGACCGCAAGACCCTTCCTCTATAT-3′; the SalI site was introduced) and dsGSL8-SpeI-r1 (5′-CCCGACTAGTCGCATATCTCATTAAAGCAGGA-3′; the SpeI site was introduced). Sequence-confirmed PCR product was digested with SalI and SpeI and cloned into a dexamethasone-inducible binary transformation vector, pTA7002, to produce pTA7002-GSL8-RNAi. Dexamethasone treatment was carried out according to Binarovaì et al. (2006).

The pTA7002-GSL8-RNAi vector were transferred into Agrobacterium tumefaciens strain GV3101 and used to transform Arabidopsis ecotype Columbia by dipping. Seeds harvested from mature transformed plants were sterilized, stratified for 3 d, and sown on MS agar plates containing hygromycin (50 mg L⁻¹). Positive lines were selected after growing for 10 d on hygromycin plates. Putative transgenic plants were transplanted onto soil and grown in a growth chamber.

Supplemental Data

The following materials are available in the online version of this article.

• Supplemental Figure S1. Loss-of-function mutations in the GSL gene family.

• Supplemental Figure S2. Calli induction from gsl8 mutant seedlings.

• Supplemental Figure S3. Patterning phenotypes of gsl8 mutants.

• Supplemental Figure S4. Distribution of normal stomata and of clustered stomata in the epidermis of cotyledons in the wild type, tmm, and gsl8-4.

• Supplemental Figure S5. Primary and secondary stomatal patterning mechanisms in wild-type Arabidopsis.

• Supplemental Figure S6. GSL8 is expressed in all organs tested using RT-PCR.

• Supplemental Figure S7. Dexamethasone-inducible RNAi line of GSL8 phenocopies T-DNA insertional mutants.

• Supplemental Table S1. Identity of T-DNA insertion lines in GSL family genes (excluding GSL8 and GSL10).

• Supplemental Table S2. Segregation analysis of GSL8/gsl8.

• Supplemental Table S3. Primers used to isolate homozygous gsl family mutants.

• Supplemental Table S4. Primers used to isolate homozygous gsl8 mutants.