The FUS3 transcription factor functions through the epidermal regulator TTG1 during embryogenesis in Arabidopsis


  • Yuichiro Tsuchiya,

    1. Department of Botany, University of Toronto, 25 Willcocks Street, Toronto, Ont., Canada,
    2. Division of Applied Biosciences, Graduate School of Agriculture, Hokkaido University, Sapporo 060-8589, Japan, and
    Search for more papers by this author
  • Eiji Nambara,

    1. Laboratory for Reproductive Growth Regulation, Plant Science Center, RIKEN, 2-1 Hirosawa, Wako, Saitama 351-0198, Japan
    Search for more papers by this author
  • Satoshi Naito,

    1. Division of Applied Biosciences, Graduate School of Agriculture, Hokkaido University, Sapporo 060-8589, Japan, and
    Search for more papers by this author
  • Peter McCourt

    Corresponding author
    1. Department of Botany, University of Toronto, 25 Willcocks Street, Toronto, Ont., Canada,
      For correspondence (fax +1 416 978 5878; e-mail
    Search for more papers by this author

For correspondence (fax +1 416 978 5878; e-mail


Loss-of-function mutations in the FUSCA3 (FUS3) gene of Arabidopsis result in alterations in cotyledon identity, inability to complete late seed maturation processes, and the premature activation of apical and root embryonic meristems, which indicates that this transcription factor is an essential regulator of embryogenesis. Although FUS3 shows a complex pattern of expression in the embryo, this gene is only required in the protoderm to carry out its functions. Moreover, the epidermal morphogenesis regulator TRANSPARENT TESTA GLABRA1 (TTG1) is negatively regulated by FUS3 in the embryo. When a loss-of-function ttg1 mutation is introduced into a fus3 mutant, a number of fus3-related phenotypes are rescued, indicating a functional TTG1 gene is required to manifest the fus3 mutant phenotype. It therefore appears that one of the functions of FUS3 is to restrict the domain of expression of TTG1 during embryogenesis. The FUS3–TTG1 interaction is both maternal and zygotic, suggesting a complex relationship is required between these gene products to allow correct seed development.


An essential feature of embryogenesis is the formation of distinct tissue and cell types during development. In Arabidopsis as with most higher plants, the information that specifies the positioning of the cotyledon, hypocotyl, and embryonic root as well as epidermal, ground, and vascular cell fates is established relatively early during embryogenesis (Laux and Jurgens, 1997). After these early patterning events, cellular differentiation continues as the embryo begins a stage of maturation during which seed storage reserve accumulation occurs followed by the establishment of desiccation tolerance, and seed dormancy (Goldberg et al., 1989).

Mutational analysis has been extensively applied to the Arabidopsis embryo to define the genes that specify many of these stages of embryogenesis. One set of genes collectively designated LEAFY COTYLEDON (LEC) genes appear to be major regulators of a variety of embryonic stages. Loss-of-function mutations in any one of these genes result in the partial replacement of cotyledons with leaf primordia, precocious growth of the apical and root meristem and a reduction in the establishment of seed desiccation tolerance, and dormancy (see Harada, 2001 for review). To date, three LEC genes have been molecularly characterized LEC1, LEC2, and FUSCA3 (FUS3; Baumlein et al., 1994; Keith et al., 1994; Meinke et al., 1994; West et al., 1994). The LEC1 gene encodes a protein related to a transcription factor subunit of the HAP3 gene family in mammals, while LEC2 and FUS3 encode proteins with close amino acid similarity to the B3 domain of the ABI3/VP1 transcription factor family (Lotan et al., 1998; Luerssen et al., 1998; Stone et al., 2001). Although genetic studies suggest LEC1, LEC2, and FUS3 all contribute to similar processes in the embryo, the distinct phenotypes of single loss-of-function alleles indicate that they do not encode completely redundant functions (Meinke et al., 1994; Raz et al., 2001). The most detailed information on the embryonic expression patterns and developmental programs regulated by LEC genes has involved LEC1. These studies support the idea that this gene plays multiple roles during various stages of embryogenesis (Lotan et al., 1998; West et al., 1994). Furthermore, a comprehensive histological analysis of lec mutants in relation to embryo growth and misexpression studies using LEC1 and LEC2 in adult tissues suggest that LEC genes function to suppress vegetative development and encourage embryo arrest (Lotan et al., 1998; Raz et al., 2001; Stone et al., 2001).

The ectopic production of trichomes and anthocyanin in lec mutant cotyledons suggests that LEC genes regulate epidermal differentiation in the embryo. One downstream candidate of LEC gene regulation is the epidermal morphogenesis regulator gene TRANSPARENT TESTA GLABRA1 (TTG1; Galway et al., 1994; Koornneef, 1981; Schiefelbein, 2003; Walker et al., 1999). Two adult phenotypes observed in loss-of-function ttg1 mutants such as the loss-of-anthocyanin production and reduction in leaf trichome number are notable opposite phenotypes to those observed in fus3 loss-of-function mutant embryos. Formally, this could mean that LEC genes negatively regulate TTG1 expression in the embryo. Although the function of TTG1 has been extensively studied post embryonically, less information is available on its role during embryogenesis. Loss-of-function mutations in TTG1 result in reduced anthocyanin synthesis in the seed testa, loss of mucilage in the seed coat, and reduced dormancy in the seed (Debeaujon et al., 2000; Leon-Kloosterziel et al., 1994; Shirley et al., 1995; Western et al., 2001). The ttg1 mutation also significantly reduces the level of the epidermal cell specific transcription factor, GLABRA2 (GL2), throughout embryonic and post-embryonic development, suggesting that TTG1 may act in a similar fashion in embryo and adult development (Lin and Schiefelbein, 2001).

In order to investigate the relationship of LEC genes to TTG1, we undertook a detailed characterization of FUS3 in relationship to TTG1 function. Using both RNA in situ hybridization and a sensitive FUS3::GUS reporter, we show that FUS3 is preferentially expressed in the protoderm of young embryos. Furthermore, targeted FUS3 embryonic expression to protodermal tissue is sufficient to rescue all fus3 mutant embryonic phenotypes, indicating that expression of FUS3 in this tissue is sufficient for FUS3 function during embryogenesis. We also show that embryonic FUS3 is required to negatively regulate TTG1 during embryogenesis. Many fus3-related phenotypes such as ectopic trichome production, anthocyanin accumulation, desiccation intolerance, and seed storage protein accumulation are partially rescued when seeds contain a ttg1 loss-of-function mutation. The interplay between FUS3 and TTG1 revealed in double mutant studies also suggests the protoderm and seed testa developmentally interact in the establishment of cotyledon identity, seed storage reserve accumulation, and desiccation tolerance in the seed.


FUS3 expression during embryo development

The most detailed information on the embryonic expression patterns by LEC genes has mostly involved LEC1 studies (Lotan et al., 1998; West et al., 1994). This gene is expressed early in embryogenesis in both the suspensor and embryo proper but by the heart stage, transcripts are preferentially localized in the protoderm (Lotan et al., 1998). After this, point expression becomes more ubiquitous and dissipates until little transcript is detected in the mature embryo. To determine if FUS3 transcript accumulation is related to LEC1 expression patterns throughout embryogenesis, mRNA in situ hybridization was performed. In addition, a sensitive GUS transcriptional reporter (FUS3::GUS) was introduced into wild-type Arabidopsis plants, and siliques from a number of independent lines were sectioned and stained for GUS activity. At the globular stage, all cells in the embryo proper showed FUS3 expression with the protodermal and suspensor cells displaying the highest activity (Figure 1a,d). By the heart stage, embryos showed preferential FUS3 signal in the protodermal tissue with weaker expression in the internal cells (Figure 1b,e). In mature embryos, FUS3 mRNA accumulation was difficult to detect by RNA in situ hybridization with weak transcript accumulation been localized to the provascular tissue in the hypocotyl and to root epidermal cells (Figure 1c). By contrast, however, transgenic GUS reporter plants yielded GUS activity in the provascular tissue, the root cap, the mature epidermis, and the aleurone layer in mature seeds (Figure 1f). This difference between the mRNA in situ and FUS3::GUS patterns may simply be the result of the persistence of the GUS protein signal or may reflect the higher sensitivity of the GUS assay system, which involves an enzymatic amplification. The detection of a GUS signal in the aleurone layer of mature embryos is, however, consistent with the observation that loss-of-function fus3 mutants show abnormal development of this cell type (Keith et al., 1994).

Figure 1.

FUS3 expression profiles at various stages of embryogenesis.

(a–c) RNA in situ hybridizations.

(d–f) GUS activity staining.

FUS3 mRNA in situ hybridization and GUS histochemical staining of transgenic embryos containing a FUS3::GUS promoter fusion.

(a,d) Globular embryo, bar = 20 µm; (b,e) heart stage embryo, bar = 30 µm; and (c,f) mature embryo, bar = 100 µm. a, aleurone; p, protoderm; and s, suspensor.

Protodermal expression of FUS3 is sufficient for its function during embryogenesis

The early expression of FUS3 in the protodermal tissue of heart stage embryos is similar to LEC1 mRNA accumulation patterns suggesting that these genes may act in concert in this tissue to regulate various aspects of embryogenesis. Furthermore, it suggests that expression of LEC genes in this tissue at early stages is important to their function in embryogenesis. To determine directly the role of protodermally derived FUS3, the coding region of the FUS3 gene was fused to the Arabidopsis thaliana meristem layer 1 (ATML1) promoter of Arabidopsis and introduced into the fus3-3 loss-of-function mutant. During embryogenesis in Arabidopsis, the ATML1 promoter shows restricted expression to the protodemal tissue by the globular stage, and this pattern is maintained throughout the heart stage (Lu et al., 1996). After the heart stage of embryogenesis, AtML1-driven expression decreases and becomes restricted to the L1 layer of the apical meristem. Seeds were allowed to desiccate and 10 transgenic lines were identified. Introduction of the ATML1::FUS3 gene into fus3-3 plants resulted in restoration of wild-type seed development in all the lines. All obvious fus3 phenotypes, including seed coloration, dormancy, desiccation intolerance, and ectopic trichome development, in cotyledons were restored to the wild type (Figure 2a, data not shown). Furthermore, SDS–PAGE analysis of seed storage protein accumulation in a variety of independent transgenic lines showed partial to full restoration of seed storage protein accumulation (Figure 2b). The restriction of FUS3 expression to the protodermal tissue during early embryogenesis is sufficient to rescue fus3-3 phenotypes, indicating that this tissue plays a major role in the function of this gene during embryogenesis in Arabidopsis.

Figure 2.

Genetic rescue of fus3-3 mutant seed by the AtML1::FUS3 transgene.

(a) Dry mature seeds of the wild type, fus3-3, and fus3-3 containing AtML1::FUS3 transgene.

(b) SDS–PAGE of wild-type, fus3-3, and three transgenic fus3-3 seeds transformed with AtML1::FUS3. The same volume of protein extract from 20 seeds of each genotype was loaded onto the gel.

TTG1 is a downstream target of FUS3

The importance of FUS3 expression in the protoderm during embryogenesis could mean that a number of FUS3-dependent functions act through this tissue to regulate development in the embryo. The TTG1 gene is a known regulator of both anthocyanin and trichome production in post-embryonic development and the absence of these traits in wild-type embryos (Koornneef, 1981). This suggests FUS3 may negatively regulate this epidermal fate regulator in the cotyledon. To determine if FUS3 can influence embryonic TTG1 expression, we examined the effects of the fus3 mutation on TTG1 transcript accumulation. In young wild-type embryos, TTG1 transcripts are detected in all cells in both mutant and wild-type globular stage embryos (Figure 3a,b). By the heart stage, TTG1 transcript accumulation is reduced in the protodermal tissue of wild-type embryos (Figure 3c). In contrast, TTG1 transcripts appeared slightly reduced throughout the fus3 heart-stage embryo (Figure 3d). More notably unlike wild-type embryos TTG1 expression was not excluded from the protodermal layer of heart stage embryos.

Figure 3.

Transcript accumulation of TTG1 wild-type and fus3-3 embryos.

(a, d) Globular stage embryo TTG1 antisense.

(b, e) Heart stage embryo TTG1 antisense.

(c, f) Walking Stick embryos TTG1 antisense.

(g–i) Wild-type globular, heart, and walking stick stage embryos probed with TTG1 sense.

Bars = 20 µm for (a,d,g); bars = 30 µm for (b,e,h); and bars = 100 µm for (c,f,i).

At later stages of embryogenesis, the influence of the fus3 mutation on TTG1 transcript accumulation is more complex. After the heart-stage in wild-type embryogenesis, TTG1 transcript localization become more restricted and is eventually limited to the procambium of mature walking stick embryos and the seed testa (Figure 3e). Interestingly, this pattern partially resembles the transcript accumulation pattern of FUS3 in wild-type embryos. By contrast, TTG1 expression in fus3 mutants was quite variable (Figure 3f). Overall, the transcript accumulation of FUS3 was more diffuse across the embryo proper except at the embryonic root where TTG1 transcripts could be observed in the epidermal layers particularly around the embryonic root and the tips of the cotyledons. Interestingly, although TTG1 is thought to specify late events during cellular morphogenesis in epidermal tissues, it is clearly expressed in internal tissues of wild-type embryos. Internal embryonic phenotypes, however, such as altered provascular development in ttg1 loss-of-function mutations have not been reported (Koornneef, 1981; Lin and Schiefelbein, 2001). By contrast, mature fus3 embryos do show precocious provascular development (Keith et al., 1994).

fus3 anthocyanin and trichome production requires TTG1

Following the discovery that the fus3 mutation affects TTG1 transcript accumulation during embryogenesis, we decided to determine if a functional relationship existed between these two gene products by constructing double mutants. Because of the known effects of the ttg1 mutation on maternal traits, such as seed coat development, double mutant embryos were produced in a variety of maternal tissue genotypes. To generate fus3 ttg1 double mutant embryos with a maternal ttg1 background, F2 seed from an F1 plant heterozygous for both mutations were germinated and screened for the glabrous leaf phenotype expected of a ttg1 homozygous plant. A number of these seedlings were screened for plants heterozygous for the fus3-3 mutation using a molecular polymorphism that identified this allele (see Experimental procedures). F3 seeds from a single FUS3/fus3 ttg1/ttg1 F2 silique segregated a 33 light yellow to 15 dark green seed as expected for the segregation of a simple recessive mutation (3 : 1, χ2 = 1.00; P > 0.90, Figure 4a). The dark green seeds were highly viviparous, but if removed at an earlier developmental stage, they were viable and produced seedlings that could be grown to maturity. The F4 seeds from these plants only produced the dark green viviparous seeds, indicating that this phenotype requires both a maternal and zygotic fus3 ttg1 genotype (Figure 4b). The increased vivipary of the double mutant suggests the ttg1 mutation enhances the non-dormant phenotype of the fus3 embryos, which is consistent with previous reports that ttg1 mutants show reduced seed dormancy (Debeaujon et al., 2000; Leon-Kloosterziel et al., 1994).

Figure 4.

Maternal affects of TTG1 function on FUS3-dependent embryogenesis.

(a) Genotypes and seed phenotypes of a F2FUS3/fus3 ttg1/ttg1 silique segregating the fus3-3 mutation in F3 embryos.

(b) Phenotype of F4 seeds generated an F3 dark green seed (fus3/fus3 ttg1/ttg1) shown in panel a.

(c) Genotypes and seed phenotypes of an F2fus3/fus3 TTG1/ttg1 silique segregating the ttg1 mutation in F3 embryos.

(d) Representative leafy cotyledon phenotype of a germinated dark red seed (fus3/fus3 TTG1/TTG1) taken from a silique (fus3/fus3 TTG1/ttg1) segregating the ttg1 mutation as shown in panel c.

(e) Representative glabrous cotyledon phenotype of a germinated light green seed (fus3/fus3 ttg1/ttg1) taken from a silique (fus3/fus3 TTG1/ttg1) segregating the ttg1 mutation as shown in panel c.

To generate fus3 ttg1 embryos in which the maternal tissue was fus3/fus3 TTG1/ttg1, F1 seeds heterozygous for both mutations were germinated, but this time F2 seeds were screened for trichomes on the cotyledons to identify fus3 homozygous lines. A number of these lines segregated dark red, light red, and light green F3 seeds (Figure 4c). For example, a single F2 silique yielded 17 dark red, 33 light red, and 14 light green F3 seeds, indicating the segregation of a simple additive Mendelian trait (1 : 2 : 1, χ2 = 0.15; P > 0.95). Germination of immature dark red F3 seeds produced cotyledons with trichomes as expected for a seedling that is fus3/fus3 TTG1/TTG1 (Figure 4d). Light green seeds only produced glabrous cotyledons as expected of a fus3/fus3 ttg1/ttg1 seedling (Figure 4e). The ratio of light red F3 seeds suggested that these embryos are genotypically fus3/fus3 TTG1/ttg1. Together, these results suggest that the zygotic dosage of wild-type TTG1 determines the expressivity of the fus3 red embryo phenotype.

The difference in seed phenotypes of the fus3 ttg1 double mutant embryos in these two experiments suggests that the TTG1 genotype of the maternal tissue can dramatically influence the phenotype of the double mutant embryos. To confirm this, light green fus3 ttg1 double mutant seed generated in a fus3/fus3 TTG1/ttg1 maternal background were germinated and grown to maturity. Seeds from these plants, which are now double mutant both maternally and zygotically, only produced dark green, highly viviparous seeds similar to those seen in Figure 4(b; data not shown). Therefore, a functional copy of the TTG1 in the maternal tissue can dramatically influence the development of a fus3 ttg1 double mutant embryo.

Maternal and zygotic TTG1 dosage influences fus3 late embryo defects

When the maternal tissue of the seed is ttg1, the fus3 ttg1 double mutant phenotypes are enhanced. By contrast, if the maternal tissue is TTG1, the overall seed morphology of double mutant embryos appear to be more normal. We closely analyzed how TTG1 can influence fus3 ttg1 embryos by testing the ability of these seeds to establish dormancy, become desiccation tolerant and accumulate seed storage proteins. Seeds from 12 to 14 day post-pollinated fus3-3/fus3-3 TTG1/ttg1 plants were harvested and first tested for their ability to germinate after different times of after ripening. Good germination was observed from freshly harvested seed, and germinated plantlets showed both glabrous and non-glabrous cotyledons at an expected 3 : 1 ratio (Figure 5, 0 days). The results indicate that seeds were viable but non-dormant irrespective of their zygotic TTG1 genotype. By contrast, if seeds were allowed to dry for 10 days, approximately 75% did not germinate indicating fus3-dependent desiccation intolerance was occurring (Figure 5, 10 days). The majority of the seed that did germinate, however, were glabrous, suggesting that they were homozygous for the ttg1 mutation. The overall seed germination continued to decrease 20–30 days after drying, but the seed that did germinate was predominantly ttg1 glabrous (Figure 5; 20 and 30 days). These results suggest that zygotic ttg1 can partially suppress fus3-dependent desiccation intolerance, but that this suppression requires that the maternal tissue is TTG1.

Figure 5.

Suppression of desiccation intolerance of the fus3 mutation by the ttg1 mutation.

Times represent days after harvesting when seeds were put on Petri plates. Black bar, percentage of seedlings that showed a glabrous cotyledon phenotype after germination; and white bar, percentage of seedlings that showed a leafy cotyledon phenotype after germination.

Because the appropriate TTG1 dosage in the maternal tissue and zygote can partially suppressed the desiccation intolerant phenotype of fus3 embryos, we analyzed other late embryo phenotypes of this genetic background to determine the extent of suppression. When dark red seeds (fus3/fus3 TTG1/TTG1) were separated from light green seeds (fus3/fus3 ttg1/ttg1) and analyzed by SDS–PAGE analysis, seed storage protein levels were substantially higher in the light green seed versus the dark red seed sample (Figure 6). Therefore, it appears that suppression of other fus3-dependent late embryo processes such as seed storage accumulation is also dependent on the TTG1 genotype of the embryo and the testa.

Figure 6.

SDS–PAGE of proteins isolated from mature seeds of wild-type (WT) and mutant lines.

The same volume of protein extract from 50 seeds of each genotype was loaded onto the gel. In the fus3 mutant lanes, the maternal tissue was fus3/fus3 ttg1/ttg1. The locations of the 12S and 2S seed storage proteins are indicated on the left. Lane 1, WT; lane 2, fus3/fus3 TTG1/-; and lane 3, fus3/fus3 ttg1/ttg1.


Early roles for FUS3 in embryogenesis

After the establishment of its basic body plan, the Arabidopsis embryo enters a maturation stage of seed storage accumulation followed by a quiescent period in which desiccation tolerance and dormancy are established (Laux and Jurgens, 1997). In fus3 mutants, these three stages of development are all perturbed to some degree (Baumlein et al., 1994; Keith et al., 1994; Meinke et al., 1994). Based solely on phenotypes, it is difficult to determine if early defects in fus3 embryos results in later aberrant embryonic development or if this gene has specific functions at different stages of embryogenesis. The early patterns of FUS3 expression share similarities with LEC1 transcript accumulation, suggesting that these two transcription factors respond to the same positional information that establishes specific cell types in the early embryo. In this study, it appears that FUS3 expression in the protoderm of globular and heart stage embryo is essential and sufficient to regulate a variety of events during embryogenesis. This is further supported by the observation that FUS3 is required early to negatively control the epidermal regulator TTG1 in the protoderm of heart-staged embryos. The FUS3-dependent negative regulation of TTG1 in the protoderm offers an explanation as to why wild-type cotyledons are normally glabrous and lack anthocyanin. Loss of this function results in misexpression of TTG1 and possibly other epidermal regulators in the protoderm, resulting in ectopic anthocyanin and trichome production (Figure 7).

Figure 7.

A working hypothesis for the role of the FUS3-TTG1 gene products during embryogenesis.

The model shows FUS3 in a repressive role on TTG1 action in the protodermal layer (dark grey) of the young embryo. FUS3 also has a positive role on TTG1 transcript accumulation in the provasculature of mature embryos (light grey). TTG1 has a repressive role on seed germination through its expression in the maternal seed coat. This role may or may not be mediated through ABA (?).

In Arabidopsis, the protoderm can be visually distinguished in young dermatogen embryos (Mansfield and Briarty, 1992). After this stage, cell divisions in protodermal layer that surround the cotyledon primordial, increase in frequency to give the stereotypically heart-stage embryo. The suggestion that FUS3 may function as an embryonic cell arrest signal could mean protodermally derived FUS3 modulates the rate of cell divisions in this layer. If true, cell division rates in this tissue may in turn be required for correct specification of future cotyledon parenchyma cells.

Roles for FUS3 in late embryogenesis

One mechanism to bypass the effects of the fus3 mutation is by introducing the ttg1 mutation into the embryo. Loss of TTG1 in the embryo can also partially restore desiccation tolerance and seed storage protein accumulation to mutant seeds, indicating that other late embryogenesis programs have been restored to some degree. The suppression of fus3 by ttg1 suggests a functional TTG1 gene is required for the fus3 mutation to fully manifest its late embryo phenotypes. Transcript accumulation of TTG1 is altered in fus3 late embryos versus the wild type in that TTG1 mRNA is less localized and shows variable expression throughout mutant embryos. These patterns suggest that either a reduction of TTG1 localization to the provascular tissue or the weak ectopic expression of TTG1 throughout mutant embryos contributes to fus3 late embryonic phenotypes. Loss of provascular TTG1 transcripts is, however, inconsistent with the ability of a loss-of-function ttg1 mutation to partially suppress fus3. By contrast, fus3 phenotypes dependent on weak misexpression of TTG1 would be suppressed by loss of TTG1 function (Figure 7)

Double mutant studies between the ttg1 and fus3 in late embryogenesis also uncover a maternal influence to this genetic interaction. If the seed testa is genotypically ttg1, then suppression of fus3 late embryo phenotypes does not occur. On the contrary, ttg1 mutant testa further decreases fus3 seed dormancy. The imposition of dormancy by the testa is well established in higher plants and a number of mutations that affect the testa including ttg1 have been shown to reduce Arabidopsis seed dormancy (Debeaujon et al., 2000). In many cases, the cause of reduced dormancy relates to reduced pigmentation in the seed endothelium and a thinner parenchymatic cell layer. Although fus3 embryos retain the capacity to grow when excised at the torpedo stage of embryogenesis, mutant seed is seldom viviparous, indicating that the barrier of the seed coat plays some role in inhibiting fus3 germination (Raz et al., 2001). A wild-type seed testa may retard fus3 germination enough at the heart/torpedo stage to allow them to enter later stages of embryogenesis where the zygotic ttg1 mutation can suppress fus3 defects such as desiccation intolerance (Figure 7). By contrast, a weakened seed coat because of a maternal ttg1 mutation may encourage fus3 vivipary thereby not allowing zygotic ttg1 the opportunity to suppress fus3 in the late embryo.

An alternative possibility to a mechanical model is that TTG1 controls a component in the testa that acts non-cell autonomously on the embryo to regulate seed maturation. Interactions between maternal and embryonic tissues in the establishment of late embryo functions have been previously reported in Arabidopsis (Gualberti et al., 2002; Koornneef et al., 1989). Embryonic phenotypes of weak abi3 mutants, for example, can be enhanced if the maternal tissue is made deficient for abscisic acid (ABA) biosynthesis. Related to this, mutations in fus3 decrease seed ABA levels and fus3 seed containing a mutation in ABA biosynthesis result in a dark green highly viviparous embryo that is reminiscent of fus3 seeds deficient in both maternal and zygotic TTG1 (Nambara et al., 2000). If TTG1 acted as a positive regulator of ABA synthesis in the testa, the loss of this function would reduce maternally derived ABA, which in turn could enhance fus3 seed vivipary. By analogy, a wild-type TTG1 testa producing ABA could partially inhibit early germination of fus3 embryos by allowing them to enter late embryogenesis (Figure 7). The observation that immature fus3 seed will not germinate on exogenous ABA is consistent with this model (Keith et al., 1994). These genetic studies demonstrate the complex interactions between the maternal and embryonic tissues in plants that would be difficult to study in a wild-type background. They also suggest that M3 mutant screens performed in appropriate genetic background in which the maternal tissue can be homozygous recessive for mutations may uncover new maternal–embryo interactions. Such screens may further unravel how these two tissues co-ordinate to regulate late seed development.

Experimental procedures

Plant material and propagation

The fus3-3 and ttg1 mutations used in this study were both derived in a Columbia genetic background (Keith et al., 1994). Sterile and non-sterile conditions have been described previously by Haughn and Somerville (1986). Homozygous fus3-3 seeds were grown as described previously by Keith et al. (1994). For the desiccation tolerance assay, seeds from lines segregating the ttg1 mutation in a fus3/fus3 TTG1/ttg1 parent were isolated 12–14 days after pollination and were directly plated to minimal media. Plates were grown under constant light at room temperature. Seeds were scored for germination after 5–7 days post-plating.

In situ hybridization

For experiments involving DIG-RNA in situ hybridization, the expressed sequence tag (EST) clone TAP0257 (GenBank Accession #F20055) for TTG1 was obtained from the Arabidopsis Biological Resource Center (ABRC). To avoid non-specific hybridization, the 5′ end of the insert was removed by digestion with BglII and BamHI, followed by self-ligation and a PCR fragment was amplified from this clone using M13 forward and M13 reverse universal primers. RNA probes were synthesized with digoxigenin detection (DIG) RNA labeling mix (Roche, Laval, Canada) following the manufacturer's instructions. T3 RNA polymerase and T7 RNA polymerase were used for sense and antisense probes, respectively. Whole siliques were fixed in fluoroacetic acid (FAA) overnight at 4°C, dehydrated by dilution series of ethanol, infiltrated by Citril Solv. (Fischer, Nepean, Canada) and then infiltrated by paraffin. Seven-micrometer thick sections were prepared and hybridization, washing and immunological detection were carried out as described elsewhere by Vielle-Calzada et al. (1999). DIG nucleic acid detection kit (Roche) was used for immunological detection.

GUS histochemical assays

To construct transgenic plants containing the GUS gene fused to the FUS3 promoter approximately 3 kbp on promoter sequence was used. Plants were transformed using the floral dip method (Clough and Bent, 1998). GUS staining was performed as described previously by Jefferson et al. (1987), followed by fixation and paraffin infiltration as described above. The tissue was sectioned at 5 µm thicknesses and paraffin was removed by dipping slide glasses into Citril Solv. two times for 10 min prior to observations of the sections.

Construction of the AtML1::FUS3 fusion protein

The AtML1::FUS3 fusion was constructed using a KpnI fragment of pTRXFusFUS3 (Luerssen et al., 1998). This fragment was cloned into pGEM7Zf(+) (p7ZFUS). The XbaI and SacI fragments of p7ZFUS were subsequently cloned into pBI101 (p101-FUS). A 3-kbp fragment of AtML1 promoter was amplified by PCR from the Columbia genome with following primers: pML1-F: 5′-CATTTACACATCCTGTCG-3′; and pML1-R; 5′-ATAACTAGTGGATTCAGGGAG-3′. The pML1-R primer contained a mutation to introduce a SpeI site into the amplified sequence. The PCR product was then cloned into the PCR2.1 (Invitrogen, Burlington, Canada) using SpeI and the AtML1 promoter was introduced into XbaI site of p101-FUS.

Identification of the fus3-3 mutation using a molecular polymorphism

A PCR fragment including fus3-3 mutation was amplified with FUSPM-F (CTAGCCAATTTTGAAATTCTTGCTAACC) and FUSPM-R (GCCTCCTATTTCCCACGTGCCAACTCC) primers. RsaI digestion of the PCR product produced a 500- and a 100-bp fragment in wild-type Columbia, and in the fus3-3 mutant only produced a 600-bp band.

Extraction of seed protein

Seeds were removed from a fus3/fus3 TTG1/ttg1 plant and separated into dark red (fus3/fus3 ttg1/ttg1) and light green (fus3/fus3 TTG1/ttg1) seeds. Seed proteins were extracted by grinding 50 seeds of each genotype in an Eppendorf tube with 20 µl µg−1 seed extraction buffer (100 mm Tris–HCl, pH 8.0, 0.5% SDS, 10% glycerol, and 2% mercaptoethanol, all w/v) Extracts were boiled for 3 min and centrifuged. Proteins were resolved by SDS–PAGE using a 15% gel. Proteins were visualized by staining with Coomassie Brilliant Blue R250.


We are grateful to George Haughn for supplying a ttg1 allele in a Columbia background. We also acknowledge the use of the Ohio State Arabidopsis stock center (ABRC) for access to a TTG1 EST sample. We would also like to thank members of the McCourt research group for helpful discussion, in particular S. Lumba for reading and editing drafts of the manuscript. We also thank S. Lumba for creating a CAPS marker for fus3-3 identification. Upon request, all novel materials described in this publication will be made available in a timely manner for non-commercial research purposes. This work was funded by an NSERC grant to P.M. and a Japanese fellowship to Y.T.