Arabidopsis CUL3A and CUL3B genes are essential for normal embryogenesis

Authors

  • Alexis Thomann,

    1. Institut de Biologie Moléculaire des Plantes du CNRS, 12, rue du Général Zimmer, 67084 Strasbourg Cedex, France,
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Vladimir Brukhin,

    1. Institute of Plant Biology and Zürich-Basel Plant Science Center, University of Zürich, Zollikerstrasse 107, CH-8008 Zürich, Switzerland, and
    Search for more papers by this author
    • These authors contributed equally to this work.

  • Monika Dieterle,

    1. Institut de Biologie Moléculaire des Plantes du CNRS, 12, rue du Général Zimmer, 67084 Strasbourg Cedex, France,
    Search for more papers by this author
  • Jacqueline Gheyeselinck,

    1. Institute of Plant Biology and Zürich-Basel Plant Science Center, University of Zürich, Zollikerstrasse 107, CH-8008 Zürich, Switzerland, and
    Search for more papers by this author
  • Marylin Vantard,

    1. Laboratoire de Physiologie Cellulaire Végétale, UMR5019 CNRS/CEA/INRA/Université Joseph Fourier, DRDC-CEA Grenoble, 17 rue des martyrs, F-38054 Grenoble Cedex 9, France
    Search for more papers by this author
  • Ueli Grossniklaus,

    1. Institute of Plant Biology and Zürich-Basel Plant Science Center, University of Zürich, Zollikerstrasse 107, CH-8008 Zürich, Switzerland, and
    Search for more papers by this author
  • Pascal Genschik

    Corresponding author
    1. Institut de Biologie Moléculaire des Plantes du CNRS, 12, rue du Général Zimmer, 67084 Strasbourg Cedex, France,
      (fax +33 (0) 3 88 61 44 42; e-mail pascal.genschik@ibmp-ulp.u-strasbg.fr).
    Search for more papers by this author

(fax +33 (0) 3 88 61 44 42; e-mail pascal.genschik@ibmp-ulp.u-strasbg.fr).

Summary

Cullin (CUL)-dependent ubiquitin ligases form a class of structurally related multisubunit enzymes that control the rapid and selective degradation of important regulatory proteins involved in cell cycle progression and development, among others. The CUL3-BTB ligases belong to this class of enzymes and despite recent findings on their molecular composition, our knowledge on their functions and substrates remains still very limited. In contrast to budding and fission yeast, CUL3 is an essential gene in metazoans. The model plant Arabidopsis thaliana encodes two related CUL3 genes, called CUL3A and CUL3B. We recently reported that cul3a loss-of-function mutants are viable but exhibit a mild flowering and light sensitivity phenotype. We investigated the spatial and temporal expression of the two CUL3 genes in reproductive tissues and found that their expression patterns are largely overlapping suggesting possible functional redundancy. Thus, we investigated the consequences on plant development of combined Arabidopsis cul3a cul3b loss-of-function mutations. Homozygous cul3b mutant plants developed normally and were fully fertile. However, the disruption of both the CUL3A and CUL3B genes reduced gametophytic transmission and caused embryo lethality. The observed embryo abortion was found to be under maternal control. Arrest of embryogenesis occurred at multiple stages of embryo development, but predominantly at the heart stage. At the cytological level, CUL3 loss-of-function mutations affected both embryo pattern formation and endosperm development.

Introduction

In recent years, the ubiquitin/26S proteasome pathway was discovered to be a central player for rapid and selective degradation of key short-lived regulatory proteins that play important roles in a variety of cellular processes (Ciechanover et al., 2000). In particular, cell cycle progression and many developmental processes are tightly controlled by ubiquitin-dependent protein degradation. Ubiquitylation is achieved through an enzymatic cascade involving the sequential action of ubiquitin-activating (E1), ubiquitin-conjugating (E2) and ubiquitin-ligating (E3) enzymes. Among these enzymes, the E3s play a central role in the selectivity of ubiquitin-mediated protein degradation. To date several classes of E3s have been reported (Jackson et al., 2000). A major type of E3s are the SCF complexes (Deshaies, 1999), which in Saccharomyces cerevisiae are composed of four primary subunits: CDC53 (cullin1), RBX1, SKP1 and an F-box protein. The F-box proteins contain a degenerated protein domain of approximately 50–60 amino acid residues, identified first in the N-terminal region of cyclin F (Bai et al., 1996), and have in addition protein–protein interaction domains that confer the substrate specificity for ubiquitylation. Thus the F-box proteins are the adapter subunits that specifically recruit substrates to the core ubiquitylation complex through a physical interaction between the F-box domain and the SKP1 subunit.

In addition to CUL1, eukaryote genomes encode additional cullins (CUL2, CUL3, CUL4 and CUL5) (Gieffers et al., 2000), which are all believed to form protein complexes with ubiquitin-ligase activity. A series of recent reports has shed light on the molecular composition and function of the CUL3-based E3 ligases (reviewed in Pintard et al., 2004). Thus it was found that certain ‘Bric a brac, Tramtrack and Broad Complex/Pox virus and Zinc finger’ (BTB/POZ) domain proteins function as substrate-specific adaptors in Schizosaccharomyces pombe and Caenorhabditis elegans (Furukawa et al., 2003; Geyer et al., 2003; Pintard et al., 2003; Xu et al., 2003). These BTB domain-containing proteins bind CUL3, via the BTB domain and direct substrate specificity through an independent protein–protein interaction domain. The best-documented substrate for the CUL3-BTB ligases thus far, is the nematode MEI-1 protein, which is a regulator of meiotic progression (Furukawa et al., 2003; Pintard et al., 2003; Xu et al., 2003). In the proposed model, the MEL-26 protein binds to CUL3 through its BTB-domain and recruits the substrate MEI-1 through a MATH protein–protein interaction domain, thus promoting its ubiquitylation. Indeed the knockdown of MEL-26 or CUL3 using RNA interference in nematode embryos leads to MEI1 accumulation and, as a consequence, a failure to assemble the mitotic spindle. In Drosophila melanogaster, CUL3 may be involved in the degradation of Cubitus interruptus (Ci) in the posterior cells of the eye disc (Ou et al., 2002). Moreover, perturbation of CUL3 function has multiple effects during fly development (Mistry et al., 2004). In mammals, CUL3 function is essential and its loss-of-function in mouse produces an arrest during early embryogenesis (Singer et al., 1999). In contrast to metazoans, the function of the CUL3 orthologs is not essential in budding and fission yeasts (Geyer et al., 2003; Michel et al., 2003).

The genome of the model plant Arabidopsis thaliana contains two related CUL3 genes, called CUL3A and CUL3B, whereas single CUL3 genes are present in fungi and metazoans. Both Arabidopsis CUL3A and CUL3B proteins are able to interact with the RING-H2 finger protein RBX1 and with several members of plant BTB domain proteins (Dieterle et al., 2005; Weber et al., 2005), suggesting that they form similar CUL3-based E3 complexes. However, CUL3A loss-of-function mutants are viable and fertile and, exhibit only a mild phenotype; the cul3a mutant flowers slightly later than the control plants and presents a reduced sensitivity to far red light (Dieterle et al., 2005). The viability of the cul3a mutant plants might be attributed to functional redundancy between the two CUL3 genes in Arabidopsis. Here, we investigated the consequences on plant development of combined Arabidopsis CUL3A and CUL3B loss-of-function. Our genetic data indicate that the combined disruption of both genes reduces gametophytic transmission and causes embryo lethality. Arrest of embryogenesis occurred predominantly at heart-stages and affects both embryo pattern formation and normal endosperm development.

Results

No morphological defect was observed in a cul3b loss-of–function mutant

The Arabidopsis genome has revealed the presence of five expressed cullin-related genes, called CUL1, CUL2, CUL3A, CUL3B and CUL4 (Risseeuw et al., 2003). CUL3A (At1g26830) and CUL3B (At1g69670) proteins are 88% identical at the amino acid sequence level and map to the same phylogenetic clade as CUL3 from fission yeast, worm, fly and human. An alignment of the CUL3 protein sequences (Figure 1) indicates a high degree of conservation, which is not limited to the C-terminal half of the proteins (where the so-called cullin domain is located), but also the N-terminal region is conserved where the interaction with BTB domain proteins occurs. AtCUL3A and AtCUL3B are more closely related to each other than to the other cullins. This high degree of identity suggests that both proteins play similar, if not identical functions.

Figure 1.

Comparison of amino acid sequences of Arabidopsis CUL3A and CUL3B and other CUL3 proteins from different organisms.
The CUL3 protein sequences were aligned using ClustalW (http://www.ebi.ac.uk/clustalw/). Identical and conserved residues in AtCUL3A and AtCUL3B are shaded in black; these residues are shaded in grey in the other CUL3 proteins in case of conservation. The following CUL3 proteins were used for the alignment with ClustalW: Schizosaccharomyces pombe (pcu, SNP_592949); Saccharomyces cerevisiae (ScCUL3, AAQ91375); Arabidopsis thaliana CUL3A (AtCUL3A, CAC87120) and CUL3B (AtCUL3B, NP_177125); Mus musculus (MmCUL3, AAH27304); C. elegans (CeCUL3, NP_503151); human (HsCUL3, AAC36682); Drosophila melanogaster (DmCUL3; NP_723908).

We have previously shown that CUL3A function is not essential in Arabidopsis (Dieterle et al., 2005). To gain more insights into the function of the CUL3 genes in plants, we searched for T-DNA insertion mutants in the related CUL3B gene. The cul3b-1 mutant was identified from the GABI collection and the T-DNA interrupts the coding sequence in the first exon 110 bp downstream of the ATG start codon (Figure 2a) and thus the mutant is expected to be non-functional for CUL3B. In this mutant no CUL3B-truncated mRNA could be detected (Figure 2b) suggesting it is a null mutant. Homozygous cul3b-1 mutant plants were analysed under different growth conditions and at different developmental stages. These plants developed normally (Figure 2c) and were fully fertile. No morphological defect was observed during vegetative and floral development (not shown).

Figure 2.

CUL3B loss-of-function does not alter plant development.
(a) Schematic representation of T-DNA insertions in the CUL3B gene. LB and RB indicate the orientation of the left and right borders of the T-DNA, respectively. Exons (open boxes), introns (single line) and 5′ and 3′ UTRs (hatched boxes) are indicated.
(b) CUL3B transcript level in wild type and mutant plants. RNA was extracted from 3-week-old plants. The probe is indicated in (a).
(c) Six-week-old wild type plant, cul3b-1 homozygous and cul3a-1/CUL3A cul3b-1 mutant plants and cul3a-1 cul3b-1 double homozygous plant complemented with the genomic CUL3A fragment.

CUL3B promoter-GUS fusion (Weber et al., 2005) as well as RNA gels (data not shown) showed that CUL3B is expressed at all stages of plant development analysed, starting from young seedlings to flowering plants. Thus, at the tissue and organ level, CUL3B has a similar expression pattern as CUL3A (Dieterle et al., 2005), further suggesting that both cullins might function redundantly in many different tissues of the plant.

CUL3A and CUL3B are expressed in overlapping patterns in reproductive organs and during embryogenesis

To investigate the potentially overlapping expression profile at the cellular level, we determined the spatial and temporal expression pattern of both CUL3 genes in reproductive organs and during embryogenesis. We performed RNA in situ hybridization experiments on sections of flower buds and developing siliques using CUL3A antisense (AS) and sense control (S) probes. During early stages of flower development, transcripts were detected in the whole floral meristem with low levels in the sepals and petals (Figure 3a,b). Particularly strong signals were observed in emerging primordia. Later on, a rather strong signal was found in developing carpels, especially in emerging ovule primordia (Figure 3c,e) and then in the growing integuments and tetrads of megaspores. In mature ovules, transcripts were detected in the cells of the embryo sac and at a much lower level in the sporophytic tissue of the ovule (Figure 3f). In the stamens, CUL3A transcripts were detected at high level in sporogenous cells (Figure 3d) and then to a lesser degree in tetrads of microspores. A weaker signal was detected in tapetal tissue of the anther.

Figure 3.

Detection of CUL3 expression in developing flowers and seeds by in situ hybridization and CUL3A- and CUL3B-promoter::GUS studies.
(a) Longitudinal section of floral meristem.
(b) Flower bud.
(c) Flower bud with emerging ovules.
(d) Pollen mother cells (PMC).
(e) Ovule primordia.
(f) Mature embryo sac.
(g) Expression of CUL3A at the chalazal endosperm.
(h) Embryo at the globular stage.
(i) Embryo at the heart stage.
(j) CUL3A promoter GUS line showing GUS expression in inflorescence.
(k) CUL3B promoter GUS line showing strong GUS expression in pollen grains.
(l) CUL3B promoter GUS line showing GUS expression in seeds at the globular stage.
(m) CUL3A promoter GUS line showing GUS expression in seeds at the heart stage.
(n) CUL3B promoter GUS line showing GUS expression in seeds at the torpedo stage.
For the in situ mRNA detection, the hybridization appears as dark staining. AS, antisense probe; S, sense probe. Bars = 20 μm (a–i), 100 μm (j, k) and 50 μm (l–n).

After fertilization the CUL3A expression was observed in the embryo and endosperm (Figure 3h,i), with an especially strong signal observed in the chalazal cyst of the endosperm (Figure 3g). Signal intensity was high at the globular (Figure 3h) and heart stages (Figure 3i) of embryogenesis. The signal in sporophytic ovule tissues progressively decreased during ovule development and was not observed any more in the developing seed coat (Figure 3g–i). In the carpels no signal was detected in either the septum or the valve tissues (data not shown).

As the CUL3A antisense probe might also detect the CUL3B transcripts, we investigated more specifically the expression patterns of both genes during embryogenesis using transgenic lines carrying a CUL3A- or CUL3B-promoter-GUS fusion (Dieterle et al., 2005; Weber et al., 2005). CUL3A- and CUL3B-promoter-GUS fusion transgenes exhibited identical GUS expression patterns in mature pollen (Figure 3j,k) and at different stages during seed development (only illustrated for one construct at each developmental stage) (Figure 3l–n). Overall, this expression analysis shows that the CUL3A and CUL3B genes are expressed in largely overlapping patterns during reproductive development suggesting that they may have redundant functions in these tissues.

Disruption of both CUL3A and CUL3B reduces gametophytic transmission and causes maternal effect embryo lethality

To test whether CUL3A and CUL3B are indeed functionally redundant, the cul3b-1 mutant was used to pollinate a cul3a-1 homozygous mutant. From the progeny of this cross we identified several F2 plants, which were CUL3A/cul3a-1 cul3b-1/cul3b-1 (referred CUL3A/cul3a-1 cul3b-1) as well as cul3a-1/cul3a-1 CUL3B/cul3b-1 (referred as cul3a-1 CUL3B/cul3b-1). Although 96 plants were genotyped, no doubly homozygous cul3a-1 cul3b-1 mutant was identified. As both the cul3a-1 and cul3b-1 mutants contained a T-DNA insertion with integral selection markers (kanamycin + BASTA and sulphadiazine resistance genes, respectively), we used the kanamycin resistance marker for further genetic characterizations and segregation analysis. Thus, the cul3a-1/CUL3A cul3b-1 plants were self-pollinated and the F3 progeny was again genotyped in order to identify homozygous cul3a-1 cul3b-1 mutant plants (25% of this genotype is expected in case of unlinked genes). However, despite analysing more than 50 plants no double mutant plant was identified, suggesting that CUL3 function is essential in Arabidopsis.

A segregation analysis among the progeny of selfed cul3a-1/CUL3A cul3b-1 heterozygous plants revealed that kanamycin resistant to sensitive seedlings segregated in a ratio of 1.4:1 rather than 2:1 as expected for a zygotic embryo lethal mutation (Table 1). Similarly, the progeny of selfed cul3a-1 CUL3B/cul3b-1 plants revealed that sulphadiazine resistant to sensitive seedlings segregated in a ratio of 1.3:1 (Table 1). This finding suggests reduced gametophytic transmission. To determine to which extent transmission of the insertion was reduced through the male and the female gametophytes, reciprocal crosses to the wild type were performed with the cul3a-1/CUL3A cul3b-1 mutant plants. The transmission efficiency (TE) of the kanamycin resistance marker was reduced through both male and the female gametophytes. However, the TE reduction was much stronger through the male gametophyte (pollen) (Table 1).

Table 1.  Genetic analysis of the CUL3A/cul3a-1 cul3b-1 and cul3a-1 CUL3B/cul3b-1 mutant plants
Cross (E × Γ)KanR or SulfRKanS or SulfSTotalP-valueTEFTEM
  1. The NPTII gene conferring kanamycin resistance (KanR, kanamycin resistant seedlings; KanS, kanamycin sensitive seedlings) was used as selection maker for CUL3A/cul3a-1 cul3b-1 plants whereas the resistance to sulphadiazine (SulfR, sulphadiazine resistant seedlings; SulfS, sulphadiazine sensitive seedlings) was used for the cul3a-1 CUL3B/cul3b-1 plants. Transmission efficiencies were calculated according to Howden et al. (1998): TE = KanR/KanS × 100%; P-value, based on expected 1:2 segregation ratio as expected for a zygotic embryo lethal mutation, and 1:1 for the reciprocal crosses as expected for normal transmission; TEF, female transmission efficiency; TEM, male transmission efficiency; NA, not applicable. At P-value of <0.05 the null hypothesis is rejected.

CUL3A/cul3a-1 cul3b-1 selfed8633618514 818<0.0001NANA
cul3a-1 CUL3B/cul3b-1 selfed12419302171<0.0001NANA
CUL3A/cul3a-1 cul3b-1 × wild type4145199330.000679.8%NA
Wild type × CUL3A/cul3a-1 cul3b-12267971023<0.0001NA28.3%

We next examined mature siliques for the presence of aborted seeds (Figure 4a). Indeed, approximately 19% of the seeds aborted following self-pollination of the cul3a-1/CUL3A cul3b-1 mutant plants and a similar proportion of aborted seed was observed for the cul3a-1 CUL3B/cul3b-1 mutant combination (Table 2). The number of aborted seeds, however, was not consistent with zygotic embryo lethality where a segregation of aborted:normal seeds of 1:3 is expected. Given the low transmission of cul3a cul3b mutants through the male, only about 10% of the embryos are expected to be homozygous [(TEF/(1 + TEF) ×(TEM/1 + TEM) = 0.789/1.789) × (0.283/1.283) = 0.098; Moore, 2002]. On the other hand, female transmission is reduced by 20.2% such that approximately another 10% of the total seeds are expected to abort even if the paternally inherited allele is wild type. These observations suggest that the approximately 20% lethal embryos are in half each of the genotypes cul3a/cul3acul3b/cul3b or cul3a/CUL3Acul3b/cul3b, with the latter receiving a wild-type CUL3A allele from the father. Thus, segregation analysis, gametophytic transmission and seed phenotype suggest that a double knockout of CUL3A and CUL3B affects the maternal control of embryo development.

Figure 4.

Seed and embryo development in cul3 double mutant plants.
(a) Silique phenotype of wild-type and mutant plants. Arrowheads indicate aborted seeds.
(b–g) Toluidine blue stained sections. (b) Arrested early globular embryo, endosperm is missing in the seed. (c, d) Arrested seeds at the globular stage with defects in protoderm, procambium, suspensor and derivates of hypophyseal cell. (e) Defective coenocytic endosperm in the arrested seed. (f) Normal cellular endosperm from wild-type seed extracted from the same silique as (e). (g) Wild-type seeds at the torpedo stage from the same silique as (b–d).
(h–l) Cleared seeds. (h–j) Arrested seeds at the globular stage with defects as described for (c) and (d). (k) Wild-type globular embryo. (l) Wild-type seeds at the torpedo stage from the same silique as (h–j).
e, endosperm; h, derivates of hypophyseal cell; pc, procambial cells; p, protoderm; s, suspensor. Bar = 20 μm.

Table 2.  Analysis of mature siliques
AlleleNormal seedsAborted seedsSeeds scoredP-value
  1. Mature siliques were analysed for the presence of aborted seeds. P-value, based on an 1:3 as expected for zygotic embryo lethality. At P-value <0.05 the null hypothesis is rejected.

CUL3A/cul3a-1 cul3b-11734405 (18.9%)21390.0011
cul3a-1 CUL3B/cul3b-1868188 (17.8%)10560.0001

To further investigate at which developmental stage embryogenesis is affected, we analysed cleared seed specimens from siliques of the CUL3A/cul3a-1 cul3b-1 plants harbouring either early torpedo stage, bent-cotyledon stage or mature stage wild-type embryos (Table 3). At the early torpedo stage, we found a significant number (approximately 15%) of the seeds still at the globular-stage or even at an earlier stage. However, at the bent-cotyledon stage, most of the delayed seeds were observed at the heart-stage. When we analysed seeds at the mature-embryo stage, we observed a similar proportion of embryos at the heart stage. Our results indicate that embryos derived from cul3a-1 cul3b-1 mutant eggs arrest at multiple stages of their development, with most of them proceeding to the heart-stage, which is their prominent stage of arrest. Similar proportions of arrested embryos were found in the siliques of the other mutant combination (cul3a-1 CUL3B/cul3b-1) (Table 3). This is also consistent with the analysis of the mature siliques (see Table 2). We did not observe two distinct classes of arrested embryos that would correspond to the two genotypes of aborted embryos described above. It appears that a certain fraction of embryos derived from cul3a-1 cul3b-1 mutant eggs abort irrespective of whether the paternally inherited allele was wild-type or mutant. Thus, the cul3acul3b mutant displays gametophytic maternal effect embryo lethality at a reduced penetrance.

Table 3.  Analysis of cleared seed specimens from siliques harbouring either early torpedo stage, bent-cotyledon or mature stage wild-type embryos
AlleleDevelopmental stageEarly globular or earlierGlobularHeartTorpedoBent cotyledonTotal delayedNormal seedsN
  1. Values in parentheses are in percentage. NA, not applicable.

CUL3A/cul3a-1 cul3b-1Early torpedo 9 (2.1)56 (13.1)27 (6.3)NANA92 (21.5)336 (78.5)428 (100)
Bent cotyledon13 (4.1)35 (11.2)13 (4.1)NA61 (19.4)252 (80.6)313 (100)
Mature embryo33 (3)108 (9.7)64 (5.7)5 (0.5)210 (18.9)901 (81.1)1111 (100)
cul3a-1 CUL3B/cul3b-1Mature embryo41 (4.1)90 (8.9)51 (5.1)7 (0.7)189 (18.8)819 (81.2)1008 (100)

To verify that the embryo lethality is caused by the mutations in the CUL3 genes we transformed cul3a-1/CUL3A cul3b-1 mutant plants with the wild-type CUL3A gene. The transformed genomic fragment has already been shown to complement the weak phenotype of the cul3a-1 single mutant (Dieterle et al., 2005). T1 plants were selected on hygromycin and afterwards sprayed with BASTA to select for the cul3a-1 mutation. Plants of the T2 progeny were analysed for the presence of the transgene by the hygromycin marker and the cul3a-1 mutation by BASTA selection. Forty-eight of these plants were genotyped and 19 of them were identified as homozygous cul3a-1 cul3b-1 double mutants (as shown in Figure 2c). This result clearly shows that the loss-of-function of CUL3 causes the embryo lethality.

CUL3 loss-of-function affects both Arabidopsis embryo pattern formation and endosperm development

The progression of wild-type embryogenesis in Arabidopsis has been well documented (Jürgens and Mayer, 1994). To better characterize cul3a-1 cul3b-1 embryo arrests at the cytological level, we performed light microscopic observations on cleared seeds and sectioned material. Rarely, abnormalities were already observed at the early globular stage (Figure 4b). However, arrest of embryogenesis occurred at the globular stage (Figure 4c,d,i,j) but predominantly at the heart stage (Figure 5a,b,d,e). The features of the arrested embryos were as follows: (i) premature or abnormal division of the hypophyseal cell (h) (Figures 4c,h,i and 5b,d). Often the orientation of hypophyseal cell plates deviated from normal; (ii) underdeveloped short suspensor (S) (Figures 4c,j and 5a,b,d); (iii) abnormalities in procambial (pc) cell divisions (Figure 5a,b,d); and (iv) abnormalities in protoderm (p) formation (Figure 4c,d,j); (v) endosperm (e) in most of the arrested seeds was underdeveloped, coenocytic (nuclear) or with delayed cellularization (Figures 4b,e,i,j and 5a,b,d,e).

Figure 5.

Seeds of the cul3 double mutant arrested at the heart-shaped stage and abnormal cul3a cul3b seedlings.
(a–c) Toluidine blue stained sections. (a, b) Arrested embryos at the early heart stage. (c) Normal embryos at the torpedo stage from the same silique as (a, b).
(d–f) Cleared seeds. (d, e) Arrested embryos at the early heart stage. (f) Normal embryos at the torpedo stage from the same silique as (d, e).
(g) Defective cul3a cul3b seedling that failed to grow after germination.
(h) Toluidine blue stained section of defective cul3a cul3b seedling similar to (g).
(i) Defective cul3a cul3b seedling showing three unexpanded cotyledons.
(j) Immunodetection with a katanin antibody performed on 20 μg of soluble protein extracts separated by 10% SDS-PAGE. Lane 1, 5-day-old germinated Col-0 wild-type seedlings; lane 2, 12-day-old germinated CUL3A/cul3a-1 cul3b-1 seedlings; lane 3, 12-day-old germinated cul3a-1 cul3b-1 seedlings. The same blot was subsequently striped and immunoblotted with the polyclonal anti-PSTAIRE antibody used as a loading control.
e, endosperm; h, derivates of hypophyseal cell; pc, procambial cells; p, protoderm; s, suspensor. Bar = 20 μm.

Taken together, the defects observed in cul3a-1 cul3b-1 mutant embryos indicate that CUL3 is involved in the regulation of normal embryo and endosperm development. Moreover, CUL3 also seems to restrict cell division, especially in the hypophysis, where we saw extra divisions in mutant embryos and abnormal planes of division.

A small proportion of cul3a-1 cul3b-1 double mutants completes embryogenesis but arrests soon after germination

Although most of the double mutants arrest at the heart-stage during embryogenesis, we investigated whether some might still be able to complete the embryonic development. When germinated on plates the large majority of the F3 seed grew into normal plants, however a small proportion (0.7%) failed to develop after germination. The abnormal seedlings had unexpanded cotyledons, but no real root (Figure 5g,h). This is consistent with the defects earlier in development in the hypophysis, which plays a crucial role in root formation. Some of these seedlings also exhibited single (not shown) or even triple cotyledons (Figure 5i). In addition, the epidermis was abnormal (Figure 5h) as was the protoderm of arrested embryos at earlier stages of development. When kept for longer time on plate these seedlings never developed leaves and died after about 3 weeks following germination. PCR analysis indicated that most of these arrested seedlings (44 of 54 tested) corresponded to cul3a-1 cul3b-1 homozygous mutants.

The MEI-1 katanin from Caenorhabditis elegans is one of the few known CUL3 substrates (see Introduction). MEI-1 corresponds to the catalytic AAA ATPase subunit, which forms a katanin-like heterodimer together with the regulatory subunit MEI-2 (Srayko et al., 2000). As the Arabidopsis genome encodes a homologue of the catalytic subunit of animal katanin (Burk et al., 2001; Stoppin-Mellet et al., 2002; Webb et al., 2002), we tested whether this protein accumulates in the cul3a cul3b double mutant. However, no significant increase of the plant katanin-like protein was observed in the double mutant seedlings (Figure 5j).

Discussion

The cul3a-1 cul3b-1 mutant arrests at multiple stages of Arabidopsis embryo development

Recent work has shed new light on the molecular composition of the CUL3-based E3 ligases (reviewed in Pintard et al., 2004). It was found that several BTB-domain proteins interact directly with CUL3 and might thus function as substrate-specific adaptors. Whereas fission yeast only possesses three BTB-domain proteins, 105, 141 and 208 such proteins have been predicted in C. elegans, D. melanogaster and human, respectively (Geyer et al., 2003). Also in plants, BTB-domain proteins form a large family, as illustrated by the 76 BTB-proteins identified in Arabidopsis thaliana (Dieterle et al., 2005). This suggests that the CUL3-based E3s must have multiple functions and impinges likely on the activity of many different signalling pathways in all higher eukaryotes. Accordingly, it was demonstrated that CUL3 is essential in metazoan (see below), in contrast to budding yeast (Michel et al., 2003) and fission yeast (Geyer et al., 2003). In C. elegans, CUL3 forms a complex with the BTB-protein MEL-26 to destroy the microtubule-severing protein MEI-1 in early embryos (reviewed in Pintard et al., 2004). Thus loss of CUL3 leads to a failure of cytokinesis in single-cell embryos (Kurz et al., 2002). In mouse, CUL3 knockout caused an embryonic lethal phenotype (Singer et al., 1999), with abnormal patterning both in the embryonic and extra-embryonic tissues.

Here we have explored the function of CUL3 in the model plant Arabidopsis thaliana. In contrast to metazoans for which genomic sequences are available, the Arabidopsis genome encodes two CUL3-related proteins (Shen et al., 2002). Several observations suggest that both proteins share largely redundant functions: (i) the two plant CUL3-type proteins exhibit 88% of identity at the amino acid sequence level; (ii) both CUL3A and CUL3B genes show overlapping expression patterns at different developmental stages of the plant (Dieterle et al., 2005; Weber et al., 2005, and this work); (iii) and strikingly, both CUL3A and CUL3B proteins interact with the same Arabidopsis BTB-domain proteins (Dieterle et al., 2005). As individual cul3a (Dieterle et al., 2005) and cul3b (this work) T-DNA insertion mutants only exhibited a subtle or no phenotype, respectively, we engineered the cul3a cul3b double mutant. Similar to metazoan, CUL3 function is essential in plants, as its loss-of-function severely affected embryo development and totally impaired post-embryonic development. The embryos arrested at different stages of Arabidopsis seed development. Our results are consistent with two recent reports that appeared while our manuscript was under revision and that also demonstrate that CUL3 is essential for embryo development (Figueroa et al., 2005; Gingerich et al., 2005). In addition, we found that although most of the cul3a cul3b embryos arrested at the heart-stage, alterations in embryo pattern formation, suspensor development and normal endosperm formation were already detectable at the globular-stage.

As the Arabidopsis genome encodes proteins sequence-related to the worm MEL-26 and MEI-1 (Dieterle et al., 2005; Weber et al., 2005), we speculated that the cul3a-1 cul3b-1 mutant might exhibit major cytoskeletal defects. However, we did not observe extreme embryo abnormalities as seen in certain pilz and titan groups of mutants (Liu and Meinke, 1998; Mayer et al., 1999; McElver et al., 2000), where cytoskeletal defects strongly interfere with mitosis and cytokinesis. Moreover, we did not observe a higher accumulation of the plant katanin-like protein in extracts from seedlings of the double homozygous cul3a cul3b mutant, compared with wild-type seedlings taken at a similar developmental stage. In addition, we were also unable to demonstrate an interaction between the Arabidopsis katanin and the closest Arabidopsis homologue to MEL-26 by using a yeast two-hybrid assay (data not shown). Nevertheless, the endosperm of the cul3a cul3b double mutant was impaired or delayed in cellularization and certain globular and early heart stage embryo cells were enlarged and exhibited an abnormal cell shape. For these reasons, we still do not exclude a function of CUL3 in microtubule dynamics.

Does CUL3 play a function in the control of the cell cycle?

Recessive mutations that disrupt essential cell cycle regulator genes in plants should result in lethality in gametophyte and/or in early embryo development. Indeed this has been described for the two main cullin-based E3s, playing a key function in the control of the cell cycle: the SCF, which controls the G1/S transition (at least in fungi and metazoan), and the Anaphase Promoting Complex/Cyclosome (APC/C), which controls mitotic progression and exit (for review see Vodermaier, 2004). Thus, mutation of the Arabidopsis CUL1, which encodes the scaffold protein of the SCF, reduced gametophyte inheritance and produced early embryonic lethality at the zygote stage (Hellmann et al., 2003; Shen et al., 2002), whereas loss-of-function of the cullin-related subunit APC2, a subunit of the APC/C, impaired female megagametogenesis (Capron et al., 2003). In contrast to these mutants, the disruption of CUL3 produces arrests at later stages of embryo development; some were even able to complete embryogenesis. This result does not exclude the possibility that cell cycle regulatory proteins are targeted by the plant CUL3-based E3, but indicates that their accumulation does not produce a specific cell cycle arrest. Interestingly in mouse, CUL3 knockout arrested embryos accumulated cyclin E and showed abnormal cell cycle regulation (Singer et al., 1999). However, it is noteworthy that the typical class of E-type cyclin does not exist in plants (Vandepoele et al., 2002).

Do all plant CUL-based E3s regulate phytohormone biosynthesis and/or signalling?

The ubiquitin/26S proteasome pathway seems to play a key function in the action of most plant hormones (reviewed in Smalle and Vierstra, 2004). In particular it was found that SCF-like E3 complexes are involved in auxin (Gray et al., 2001), jasmonate (Xu et al., 2002), gibberellin (McGinnis et al., 2003; Sasaki et al., 2003) and ethylene (Guo and Ecker, 2003; Potuschak et al., 2003) signalling. Moreover mutations in the HOBBIT gene, encoding a putative APC/C component, also suggest a function of this E3 in auxin-mediated cell division and differentiation responses (Blilou et al., 2002).

Recent studies revealed that ETO1, an Arabidopsis BTB-domain protein controls the stability of ACS5, a member of the 1-aminocyclo-propane-1-carboxylic acid synthases that catalyses a rate-limiting step in ethylene biosynthesis (Wang et al., 2004). ETO1 physically interacts with CUL3A and might thus form the first reported plant CUL3-based E3 for which a substrate has been identified. However, most cul3a-1 cul3b-1 embryo phenotypes are unlikely to be the consequence of ethylene overproduction.

Strikingly, the cul3a-1 cul3b-1 double mutant exhibited abnormal division of the hypophysis, the founder cell of the future root meristem, and failed to develop a normal suspensor. Consequently, the few mutants that finally accomplished embryogenesis were unable to produce a normal primary root (Figure 5g,h). It has been shown that at the globular and heart stages, auxin accumulates in the hypophyseal cells (Friml et al., 2003). Interestingly, a failure in the establishment of normal hypophyseal and suspensor cells has been previously described in several auxin-insensitive mutants, such as monopteros (mp) (Berleth and Jürgens, 1993), bodenlos (bdl) (Hamann et al., 1999) and auxin resistant 6 (axr6) (Hobbie et al., 2000). Interestingly, axr6 mutations affect the ability of Arabidopsis CUL1 to assemble into a stable SCF complex with TIR1 (Hellmann et al., 2003). A model has been proposed in which the SCFTIR1 complex degrades BDL protein, thus activating MP (Hamann et al., 2002; Hellmann et al., 2003). Further work will determine whether Arabidopsis CUL3 proteins also participate in auxin signal transduction.

Experimental procedures

Plant material and transgenic plants

The cul3a-1 mutant (ecotype WS) has been described previously (Dieterle et al., 2005). The cul3b-1 mutant (ecotype Col-0, line 003D02) was identified searching the flanking sequence tag database of the GABI-KAT lines (Rosso et al., 2003). The insertion site was confirmed by sequencing the T-DNA flanking sequences. The promoter::GUS reporter lines for CUL3A and CUL3B have been described (Dieterle et al., 2005; Weber et al., 2005).

A 4.75-kb genomic CUL3A fragment including 2.2 kbp upstream from the ATG to 220 bp downstream of the stop codon and flanked by a KpnI and a XhoI restriction site was amplified from genomic DNA with oligo Cul3a5 (5′-GGGGTACCGAGAGACGTGACGTGTGATTGGTA-3′) and Cul3a3 (5′-ATGGGCCTTAAAAGGCCTCGAGC-3′) and cloned in a modified pBIN vector carrying the hygromycin marker for selection of transformed plants (Dieterle et al., 2005). Transformation of cu3a-1/CUL3A cul3b-1 mutant plants was carried out by floral dip (Clough and Bent, 1998). T2 plants were analysed for complementation.

For segregation analysis plants were surface-sterilized, incubated for 3 days at 4°C and then grown on MS medium including vitamins and MES buffer (M 0255; Duchefa, Haarlem, The Netherlands) supplemented with 1% sucrose and the appropriate antibiotics under a 16h light/8 h dark photoperiod. Soil-grown plants were kept under a 12 h light/12 h dark photoperiod.

Determination of genotypes by PCR

The forward primer MD35 (5′-AGACTTCAGAGGAGACAATGCGT-3′) and the reverse primer MD139 (5′-GTTTCCTCCATATGTCGGAGATATCCAT-3′) were used to amplify a 1280 bp fragment containing the CUL3A wild-type allele. The cul3a-1 mutant allele was detected using the T-DNA specific primer Tag5 (5′-CTACAAATTGCCTTTTCTTATTCGA-3′) and primer MD139. The primer MD139 is located downstream of the genomic fragment used for complementation and thus enables us to distinguish between the endogenous CUL3A gene and the genomic fragment used for complementation.

The primers MD105 (5′-TTCTGATTCTACGATTGATCTAAGG-3′) and MD21 (5′-GCCTAGTCTGAATCTTACTCGAATAC-3′) amplify a 672 bp CUL3B wild-type fragment. The cul3b-1 mutant allele was detected by using primer MD105 and primer GABI 8760 (5′-GGGCTACACTGAATTGGTAGCTC-3′).

Antibodies and immunoblotting

The polyclonal rabbit katanin antibody (laboratory of M Vantard) was used at a dilution of 1:1000. The Cdc2 (PSTAIRE) affinity purified polyclonal rabbit antibody (Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA) was used at a dilution of 1:4500. For immunodetection, a denaturing buffer was used to extract proteins from ground plant material (Büche et al., 2000). Twenty micrograms of total protein extracts was separated by SDS gels and blotted onto Immobilon-P membrane (Millipore, Bedford, MA, USA). The immunoreactive proteins were detected using peroxidase-conjugated goat anti-rabbit antibodies (Dianova, Chalfont St Giles, Bucks, UK) and ECL Western blot analysis system from Amersham.

Northern blot analysis

RNA was extracted using TRIzol reagent (Gibco BRL, Invitrogen Corporation, Carlsbad, CA, USA). RNA gels were performed with 15 μg of total RNA per lane. The RNA gel-blotting procedure is described elsewhere (Genschik et al., 1998). The integrity and the amount of RNA applied to each lane were verified by ethidium bromide staining. A 464-bp α-32p dCTP-labelled cDNA fragment corresponding to the AtCul3B N-terminal coding region prepared with the Prime-a-Gene random prime labelling kit (Promega Corporation, Madison, WI, USA) was used as probe.

In situ hybridization

Arabidopsis flower buds and siliques were fixed in 4% paraformaldehyde in 1x PBS with 0.1% Tween 20 overnight, dehydrated through the alcohol series followed by the Histoclear (National Diagnostics, Atlanta, GA, USA) series and embedded in Paraplast Plus (Sigma, St Louis, MO, USA). Hybridization was performed on the 10 μm sections. Sections were cleared with Histoclear, washed in alcohol series, digested with proteinase K. For post-fixation slides were treated in PBS (pH 7), then in 0.1 m triethanolamine and acetic anhydride. Sections were hybridized with DIG-labelled CUL3A-specific riboprobe followed by washing with SSC of different stringencies. Then sections were blocked in blocking buffer, equilibrated in 1% BSA solution. The hybridization signal was detected by antidigoxigenin-alkaline-phosphotase-coupled antibody binding and developed by fresh colour-substrate solution (NBT). When the colour reaction was complete, slides were washed twice in TE buffer to stop the reaction.

To synthesizing the CUL3A (At1g26830) probe the full-length cDNA of the Cul3a sequence was used. The latter was hydrolysed in 150 bp. The DNA fragments from the gene were inserted into the plasmid pGEM-T (Promega). Sense and antisense digoxygenin-UTP-labelled riboprobe was generated by run-off transcription using T7 and Sp6 RNA polymerases according to the manufacturer's protocol (Roche Diagnostics, Rotkreuz, Switzerland).

Histological analysis

For phenotypic characterization, seeds were cleared with chloral hydrate following the protocol of Yadegari et al. (1994). Specimens were observed using a Leica DMR microscope (Leica Microsystems, Bensheim, Germany) under DIC optics. For preparation of semi-thin sections, plant siliques were fixed overnight in 3.7% paraformaldehyde in PHEM buffer (60 mm PIPES, 25 mm HEPES 10 mm EGTA, 2 mm MgCl2, pH 6.9) on ice. Specimens were dehydrated in an ethanol series (30, 50, 70, 80, 90, 95, 3× 100%) and transferred into Technovit 7100 (Heraeus Kulzer, Wehrheim, Germany) according to the manufacturer's instructions. The tissue was sectioned at 8 μm thickness on a Leica RM 2145 microtome. After staining with 0.05% Toluidine Blue, sections were observed under bright-field optics using a Zeiss Axioplan microscope.

GUS staining

The CUL3A- and CUL3B-promoter GUS transgenic lines are described in Dieterle et al. (2005) and Weber et al. (2005). The method for histochemical assays for GUS activity is described in Capron et al. (2003).

Acknowledgements

The cul3b-1 T-DNA mutant was generated in the context of the GABI-KAT program and provided by Bernd Weisshaar (MPI for plant breeding research, Cologne, Germany). We thank Michi Federer (University of Zürich) and Mathieu Erhardt M. (IBMP-CNRS) for producing semi-thin sections and the gardeners of the IBMP for excellent plant care. A.T. was supported by PhD fellowship of the French Government. V.B. is thankful for financial support by the ‘Kredit zur Förderung des akademischen Nachwuchses der Stiefel-Zangger-Stiftung’ of the University of Zürich. This project was supported, in part, by grants of the Novartis Foundation and the Swiss National Science Foundation to U.G.

Ancillary