ATR3 encodes a diflavin reductase essential for Arabidopsis embryo development


  • Janani Varadarajan,

    1. Department of Botany and Plant Pathology, Purdue University, West Lafayette, IN 47907-1155, USA
    Search for more papers by this author
  • Jocelyne Guilleminot,

    1. Laboratoire Génome et Développement des Plantes, UMR-CNRS-IRD 5096, Université de Perpignan Via Domitia 58 Avenue Paul Alduy, 66860 Perpignan-Cedex, France
    Search for more papers by this author
  • Claude Saint-Jore-Dupas,

    1. Centre National de la Recherche Scientifique, Unité Mixte de Recherche 6037, IFRMP 23, UFR des Sciences, Université de Rouen, 76821 Mont-Saint-Aignan Cedex, France
    Search for more papers by this author
  • Benoît Piégu,

    1. Laboratoire Génome et Développement des Plantes, UMR-CNRS-IRD 5096, Université de Perpignan Via Domitia 58 Avenue Paul Alduy, 66860 Perpignan-Cedex, France
    Search for more papers by this author
  • Marie-Edith Chabouté,

    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
  • Véronique Gomord,

    1. Centre National de la Recherche Scientifique, Unité Mixte de Recherche 6037, IFRMP 23, UFR des Sciences, Université de Rouen, 76821 Mont-Saint-Aignan Cedex, France
    Search for more papers by this author
  • Ronald C. Coolbaugh,

    1. Department of Botany and Plant Pathology, Purdue University, West Lafayette, IN 47907-1155, USA
    Search for more papers by this author
  • Martine Devic,

    1. Laboratoire Génome et Développement des Plantes, UMR-CNRS-IRD 5096, Université de Perpignan Via Domitia 58 Avenue Paul Alduy, 66860 Perpignan-Cedex, France
    Search for more papers by this author
  • Valérie Delorme

    1. Laboratoire Génome et Développement des Plantes, UMR-CNRS-IRD 5096, Université de Perpignan Via Domitia 58 Avenue Paul Alduy, 66860 Perpignan-Cedex, France
    Search for more papers by this author

Author for correspondence:
Valérie Delorme
Tel: +33 4 6866 2242


  • The Arabidopsis genome possesses two confirmed Cytochrome P450 Reductase (CPR) genes, ATR1 and ATR2, together with a third putative homologue, ATR3, which annotation is questionable.
  • Phylogenetic analysis classified ATR3 as a CPR-like protein sharing homologies with the animal cytosolic dual flavin reductases, NR1 and Fre-1, distinct from the microsomal CPRs, ATR1 and ATR2. Like NR1 and Fre-1, ATR3 lacks the N-terminal endoplasmic reticulum (ER) anchor domain of CPRs and is localized in the cytoplasm. Recombinant ATR3 in plant soluble extracts was able to reduce cytochrome c but failed to reduce the human P450 CYP1A2.
  • Loss of ATR3 function resulted in early embryo lethality indicating that this reductase activity is essential. A yeast 2-hybrid screen identified a unique interaction of ATR3 with the homologue of the human anti-apoptotic CIAPIN1 and the yeast Dre2 protein.
  • This interaction suggests two possible roles for ATR3 in the control of cell death and in chromosome segregation at mitosis. Consistent with these results, the promoter of ATR3 is activated during cell cycle progression. Together these results demonstrated that ATR3 belongs to the NR1 subfamily of diflavin reductases whose characterized members are involved in essential cellular functions.


Diflavin reductases are enzymes which emerged as a gene fusion of ferredoxin reductase and flavodoxin. The enzymes of this family tightly bind two flavin cofactors, FAD (Flavin adenine dinucleotide) and FMN (Flavin mononucleotide), and catalyse transfer of the reducing equivalents from the two-electron donor NADPH to a variety of one-electron acceptors (Murataliev et al., 2004). Cytochrome P450 reductase (CPR) was the first enzyme of this family to be isolated (Dignam & Strobel, 1975; Yasukochi & Masters, 1976), followed by several other dual flavin enzymes, sulfite reductase in bacteria (Ostrowski et al., 1989) as well as three proteins identified in humans: nitric oxide synthase (Bredt et al., 1991; Schmidt et al., 1992), methionine synthase reductase (Leclerc et al., 1998; Olteanu & Banerjee, 2001) and a cytoplasmic protein NR1 with yet unknown functions that is expressed in cancer cells (Paine et al., 2000). Cytochrome P450 reductase is the most extensively characterized member of this family and the overall structural resemblance is the highest between CPR and NR1 or its homologue, Fre-1 (flavin reductase-1), reported in Caenorhabditis elegans (Kwasnicka et al., 2003).

In contrast to animals, where a single CPR is responsible for the electron transport to diverse P450s (57 genes encoding P450s annotated in the human genome; Huang et al., 2005), higher plants possess multiple CPRs to meet the high reductive demand for P450-mediated reactions (272 genes encoding P450s annotated in the Arabidopsis genome, (Werck-Reichhart et al., 2002)). In plants, polyploidy contributes to gene duplication and may result in functional redundancy or divergence. The Arabidopsis genome contains two authentic and one putative CPR genes named ATR (Arabidopsis Thaliana P450 Reductase). ATR1 and ATR2 are functionally active P450 reductases, as demonstrated by reconstitution of the CYP73A5 activity (Urban et al., 1997; Mizutani & Ohta, 1998). By contrast, no P450 reductase activity has been reported for the putative CPR encoded by ATR3. ATR1 and ATR2 are more closely related to each other than they are to the third potential CPR. ATR3 deviates in the strictly conserved FMN, FAD, NADPH-binding domains from all other CPRs, making its annotation as a third CPR questionable.

The endoplasmic reticulum (ER) membrane is generally accepted as the primary subcellular site for eukaryotic P450s and CPRs. Little amino acid sequence identity is found between the first 100 residues of ATR1 and ATR2 except for a hydrophobic stretch that is critical for the binding of CPRs to microsomal membrane. ATR1 encodes a single protein while ATR2 encodes either ATR2-1 or ATR2-2 depending on the choice of the initiation codon. In ATR2-1, the ER membrane anchoring signal is preceded by a predicted chloroplast targeting signal (Urban et al., 1997). ATR2-2 possesses only the ER signal peptide. To date, there is no published data confirming the subcellular localization of ATR1 and ATR2. This leaves open the hypothesis that ATR1 and ATR2-2 may be ER anchored and ATR2-1 targeted to the chloroplast also housing P450-mediated reactions. A similar organization of the N-termini of CPRs exists in poplar. However, green fluorescent protein (GFP) fusion experiments demonstrated that the homologues of ATR1 and ATR2 in hybrid poplar are localized to the ER and not in the chloroplast in transgenic Arabidopsis (Ro et al., 2002). As ATR3 lacks this N-terminal extension, ATR3 is probably not resident in the ER.

Here we report an integrative approach showing that ATR3 is the homologue of NR1 in Arabidopsis. ATR3 possesses cytochrome c reductase activity but no detectable P450 reductase activity. Genetic analysis of ATR3 loss of function revealed that ATR3 is an essential gene for early embryo development. An ATR3-interacting partner homologous to the human CIAPIN1 was identified by a yeast two-hybrid screen. Together, these results provide new perspectives on the elucidation of the functions associated with the NR1 subfamily of diflavin reductases.

Materials and Methods

Plant materials

The Arabidopsis thaliana atr3-1 mutant (EYL4, INRA collection of T-DNA lines) was of the Wassilewskija ecotype (Bechtold et al., 1993). The atr3-2 mutant (SALK _123688) was of Columbia 0 ecotype (Alonso et al., 2003). Arabidopsis plants were grown as previously described (Ronceret et al., 2005). Tobacco plants were grown under 18 h : 6 h light : dark period at 22–25°C. Tobacco BY-2 cell line was grown as previously described (Nagata et al., 1992).

Phylogenetic analyses

blastp (Altschul et al., 1997) searches were performed to screen the protein database of NCBI and find proteins homologous to our targets (ATR3/CPR or CIAPIN). ATR3/CPR sequences were aligned using clustalw and the alignment was modified manually using seaview (Galtier et al., 1996). Consensus sequences were obtained by using the program cons from the emboss suite (Rice et al., 2000). The accession numbers are provided in the legend of Supporting Information Fig. S1. CIAPIN1 protein sequences were aligned using m-coffee (Wallace et al., 2006), and their accession numbers are listed in Table S3. Phylogenetic tree construction was performed with Phylogenies by Maximum Likelihood (PhyML; Guindon & Gascuel, 2003)). The invoked options in the maximum likelihood analysis with PhyML program were 100 bootstrap replications, Whelan & Goldman (WAG) substitution model (Whelan & Goldman, 2001), shape parameter of gamma distribution estimate from data and BIONJ tree as a starting point for the iterative ML searches.

Genotyping and reverse-transcription polymerase chain reaction (RT-PCR) expression studies

For atr3-1, the T-DNA insertion site was identified by PCR walking (Devic et al., 1997). For reverse genetics with the SALK line, two primers, 5′-gctcagtctagcccggttgc-3′ and 5′-accaagcttcaacattgtacctc-3′, were designed from the genomic sequences flanking the T-DNA insertion and used in combination with T-DNA primers to verify the insertion and the hemizygous or homozygous status of the plants. RNA extraction and RT-PCR were performed as previously described (Ronceret et al., 2005) using the two primers, 5′-acctcatcatgaggaggctg-3′ and 5′-cctgatgggtgttgatcatctc-3′ for expression of the ATR3 gene.

Complementation of the atr3-1 mutation

A 5.7-kb genomic fragment of ATR3 including a 1.6-kb promoter was amplified using the Phusion polymerase (Finnzymes, Espoo, Finland) with the gateway primers 5′-attB1/cgcgttgagcgcgtgtcgtcgatttctg-3′ and 5′-acatcatcttc-acttttaattaatttg/attB2-3′. The PCR product was recombined into pDONR221 entry vector (Invitrogen), sequenced and shuttled into pHGWFS7 destination vector (Karimi et al., 2002). Heterozygous atr3-1 plants were transformed by the floral dip protocol (Clough & Bent, 1998). Siliques of the T2 progeny of nine independent primary transformants were opened and seeds counted.

Analysis of promoter expression in synchronized BY-2 cells

A 1.6-kb promoter of ATR3 was amplified using the Phusion polymerase and the primers 5′-cgggtacccgttgagcgcgtgtcgtc-3′ and 5′-cgccatggcgccgccactgccaccac-3′ designed to introduce, respectively, KpnI and NcoI restriction sites. The digested PCR product was introduced into pLuk07 to replace the original 35S CaMV promoter (Mankin et al., 1997). The promoter–luciferase fusion was subcloned in the KpnI–XbaI sites of the pCGN1549 binary vector (Calgene, Davis, CA, USA). Sequenced construct was introduced into Agrobacterium tumefaciens LBA 4404 and used to transform tobacco BY-2 cells (Chaboute et al., 2000). Luciferase activity, DNA synthesis and mitotic index were determined as previously described (Chaboute et al., 2002).

Glucuronidase (GUS) histochemistry and embryo phenotype

A 1.6 kb promoter of ATR3 was amplified from Col0 genomic DNA using the primers 5′-attB1/cgcgttgagcgcgtgtc-gtcgatttctg-3′ and 5′-tcgccgccactgccaccacccgtccg/attB2-3′. The PCR product was recombined into pDONR221 entry vector, sequenced and shuttled into pKGWFS7 destination vector (Karimi et al., 2002). Columbia 0 plants were transformed and five independent homozygous ATR3 promoter::GUS transgenic lines were obtained. Drug treatment of ATR3 promoter::GUS transgenic plants using 5 mg ml−1 colchicine (Sigma-Aldrich) or 5 μg ml−1 aphidicolin (Sigma-Aldrich) were carried out as previously described (Wang & Liu, 2006) except that seed imbibition was done in the presence of the drug for 24 h or 48 h and that the GUS staining protocol differed. In one set of experiments, the GUS staining solution differed from the recipe described in Blazquez et al. (1997) by the absence of methanol (no fixative) while in the second set, the GUS staining solution contained methanol (with fixative), as described in Blazquez et al. (1997) cited by Wang & Liu (2006). Dimethyl sulfoxide (DMSO) (10 μl ml−1 final) was used as a control as both drugs were soluble in this solvent. Study of the embryo phenotype and GUS histochemistry of the ATR3::promoter expression in the tapetum at floral stage 10 was performed as described previously (Ronceret et al., 2005).

Expression in tobacco BY-2 and epidermal cells

Two sequential runs of PCR (25 cycles) were necessary to amplify the ATR3 CDS minus stop from cDNAs from 10-wk-old seedlings using the primers 5′-attB1/cgatgggagaaaa-acaaaggaagctgcttg-3′ and 5′-aagaccaagcttcaacattgtacctcc/attB2-3′. The ATR3 CDS was fused to EGFP into pK7FWG2 destination vector (Karimi et al., 2002). Stable and transient expression of the ATR3::GFP and mRFP-HDEL respectively in tobacco BY-2 and epidermal cells were performed as described earlier (Chaboute et al., 2000; Saint-Jore et al., 2002).

Enzyme activity assays

The 35S CaMV promoter::ATR3::EGFP was engineered independently at Purdue University and ligated to the Not1 site of the binary vector pMCB302 (a gift from Dr Robert Pruitt, Purdue University). Agroinfiltration of Nicotiana benthamiana leaves was performed (Wroblewski et al., 2005). Infiltrated leaves were detached 5 d later for protein extraction and immunochemical analysis. Cytoplasmic and microsomal fractions were isolated (Ro et al., 2002) and analysed by sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE) and immunoblot analyses. The cytochrome c reductase enzymatic reaction, consisting of 50 mM TES buffer (N-Tris(hydroxymethyl)methyl-2-aminoethanesulfonic acid), 1 mM KCN, 50 μm cytochrome c, and 0.1 mM NADPH in a final volume of 1 ml, was initiated by the addition of 20–30 μg of microsomal or soluble protein fractions. The rate of reduction of cytochrome c over 1 min at 550 nm was measured with a Shimadzu UV2100 spectrophotometer. The reference sample contained all the reaction components except the protein. The specific activity was calculated as μmol of cytochrome c reduced min−1 mg−1 microsomal or soluble protein, using an extinction coefficient of 21 mM−1 cm−1 at 550 nm for cytochrome c. In vitro reconstitution of the O-deethylase activity of CYP1A2 was performed as described by Yamazaki and Shimada (2006), using human CPR or soluble and microsomal ATR3::GFP expressed in N. benthamiana.

SDS-PAGE and immunoblot analysis

Five micrograms of microsomal and cytoplasmic protein samples from N. benthamiana were resolved by 10% SDS-PAGE and transferred to polyvinyl difluoride membranes (Millipore). The blot was probed with GFP antibodies (BD Biosciences, San Jose, CA, USA) at 1 : 1000 v : v dilution. After treatment with appropriate secondary antibodies, the reactive bands were detected with the alkaline phosphatase substrate BCIP/NBT (Bio-Rad).

Yeast two-hybrid (Y2H) assays

For the Y2H assays, the open reading frame (ORF) of ATR3 was amplified using Pfu polymerase and primers 5′-cagtggatccgcatgggagaaaaacaaag-3′ and pER3-Rev designed to introduce BamHI and Xho1 restriction sites. The digested PCR product was ligated to the corresponding sites of pSos plasmid (Stratagene Agilent Technologies, La Jolla, CA, USA) in-frame with the N-terminus hSOS gene to yield the pSos-ATR3 translational fusion construct. The Arabidopsis cDNA library (SRS.Y2H library), prepared with the Cytotrap XR system from whole Arabidopsis seedlings (Wassilewskija), was a gift from Dr Zhixiang Chen (Department of Botany and Plant Pathology, Purdue University, West Lafayette, IN, USA). To identify ATR3-interacting Arabidopsis proteins, cdc25H cells were transformed with pSos-ATR3 and the SRS.Y2H library and the positive yeast colonies harboring the candidate interactors were subsequently screened. DNA sequencing of the putative interacting candidate was performed with the primers 5′-cgtcaaggagaaaaaaccccg-3′ and 5′-cgtacacgcgtctgtacagaa-3′. To further verify the interaction, yeast cotransformants harboring appropriate bait plasmids and SRS.Y2H library cDNAs isolated from the putative positives were tested for galactose-dependent growth at 37°C in retransformation assays.

Coimmunoprecipitation assay

For translating the AtCIAPIN1 homologue in vitro, the ORF of At5g18400 was amplified from pMyr-UP with Taq polymerase and primers 5′-atggattcgatgatgaatcag-3′ and 5′-tatgtcagcttcaaggaagttttgag-3′. The PCR product was ligated to the pGEMT-Easy vector (Promega). 35S-labelled AtCIAPIN1 homologous protein was synthesized in vitro with the TNT-T7 Coupled Reticulocyte System (Promega). The 35S-labeled AtCIAPIN1 homologous protein was incubated with recombinant ATR3 from yeast in a final volume of 400 μl made up with 1 ×  phosphate-buffered saline (PBS) supplemented with protease-inhibitor cocktail (Complete mini, EDTA-free; Roche) shaking for 2 h at room temperature (RT). The reactions were then incubated with 1 : 100 (v : v) dilution of ATR3 antibodies at RT for 2 h and the immune complexes were captured with Protein A agarose beads (1 : 8 v :v) shaking at RT for 2 h. The bead-bound immune complexes were pelleted by centrifugation at 12 000 g for 30–45 s, the pellet was washed five times with 1× PBS buffer, and the bound protein complexes were eluted from the beads with 1× Laemmli sample buffer. The eluted proteins were resolved by 10% SDS-PAGE and the 35S-AtCIAPIN1 homologous protein was detected by autoradiography.


ATR3 belongs to the same clade as NR1 and Fre-1

Three CPR genes, ATR1-3, are annotated in the Arabidopsis genome (Paquette et al., 2009) and registered in TAIR ( ATR3 (At3g02280) encodes a protein of 623 amino acids showing 25.7% and 24% identity with ATR1 and ATR2, respectively, whereas these last two share 60.5% identity. ATR3 shares 42.1% and 35.4% identity with NR1 and Fre-1, respectively, while in humans, NR1 shares 32.7% identity with CPR. These data raise the question as to whether ATR3 is an authentic CPR or a homologue of NR1 and Fre-1. To address this question, a Clustalw alignment was used to compare plant and animal CPRs together with plant ATR3 and animal NR1 homologues available at the time of the analysis (Altschul et al., 1997). A phylogenetic tree was derived and showed that ATR3, NR1 and their homologues form a group distinct from the CPRs (Fig. 1a).

Figure 1.

ATR3 belongs to the clade of NR1/Fre1 diflavin reductases. (a) Phylogenetic tree with per cent bootstrap value (1000 bootstrap) of 30 cytochrome P450 reductase (CPR) sequences and 12 ATR3-like sequences from 25 plant (in bold) and animal species. Aa, Artemisia annua; At, Arabidopsis thaliana; Ce, Caenorhabditis elegans; Cery, Centaurium erythraea; Chlamy, Chlamydomonas reinhardtii; Cr, Catharanthus roseus; Dm, Drosophila melanogaster; Gm, Glycine max; Hs, Homo sapiens; Ht, Helianthus tuberosus; Mm, Mus musculus; Mt, Medicago truncatula; Os, Oryza sativa; Ostta, Ostreococcus tauri; Pc, Petroselinum crispum; Pm, Pseudotsuga menziesii; Ps, Pisum sativum; Pso, Papaver somniferum; Ptd, Populus balsamifera subsp. trichocarpa × Populus deltoids; Rn, Rattus norvegicus; Sc, Saccharomyces cerevisiae; Ta, Triticum aestivum; Tc, Taxus chinensis; Vr, Vigna radiata; Vs, Vicia sativa. (b) Sequence alignment for FMN, FAD and NADPH cofactor-binding regions of consensus of plant/animal CPRs and plant/animal ATR3s. The protein consensus was obtained retaining only sequences from genome-sequencing projects ensuring that both CPR and ATR3 were available. Comprehensive alignment for all sequences is presented as in the Supporting Information, Fig. S1. Identical and similar residues are respectively black and grey box-shaded. The percentage of identity was calculated for each binding region and × indicates in the table which consensus sequences were compared. Asterisks indicate Ser457, Asp675 and Cys630, residues of the catalytic triad in CPRs.

The CPR and ATR3/NR1 groups belong to a family of enzymes that shuttles two electrons from NADPH to its physiological redox partner, via two cofactors, FAD and FMN. We compared the FMN, FAD and NADPH-binding domains between CPR and ATR3/NR1 sequences (Figs 1b, S1). A similar domain arrangement was found in CPR and ATR3 sequences. However, a greater degree of conservation exists among plant/animal ATR3/NR1 sequences than plant/animal CPRs, in particular for the FMN-binding regions.

The rate of hybrid transfer from NADPH to the FAD/NADPH domain of NR1 is c. 200-fold slower than with CPR as measured in rat (Finn et al., 2003). The authors noted the substitution of a cysteine residue (Cys630 in CPRs) by an alanine in NR1. In CPRs, Cys630 forms part of the catalytic triad with Ser457 and Asp675. Substitution of Cys630 by an alanine residue in CPR decreases the enzymatic activity 49-fold (Shen et al., 1999). Interestingly, this alanine variant is found in all ATR3/NR1 sequences with the exception of poplar ATR3 where a serine replaces the cysteine and yeast where the cysteine remains (Figs 1b, S1).

The major difference in domain organization between CPR and ATR3/NR1 sequences is associated with the N-terminal region. CPRs contain a hydrophobic N-terminal anchor domain which tethers the protein to the endoplasmic reticulum. This domain is absent in ATR3/NR1 proteins.

ATR3 is localized in the cytoplasm and the nucleus

To determine ATR3 localization in vivo, GFP was used as a tag and ATR3::GFP was transiently expressed together with the mRFP-HDEL ER marker in tobacco leaf epidermal cells (Fig. 2a–d) and stably expressed in tobacco BY-2 cells (Fig. 2e–h). In both expression systems, mRFP-HDEL highlighted the cortical ER (Figs 2b, S2) while ATR3::GFP was found in the cytoplasm and in the nucleus but excluded from the nucleolus (Fig. 2a,d,e). As the ER is localized in the cytosol, a coincident yellow signal results from the overlap of red ER and green cytoplasm emissions (Fig. 2c,g,k). However, there is no evidence of a fluorescent yellow labelled network typical and diagnostic of the cortical ER (Fig. S2). As a control, cytosolic GFP (cGFP) was coexpressed with the mRFP-HDEL and the micrographs (Fig. 2i–l) were comparable to that obtained for ATR3::GFP and mRFPHDEL (Fig. 2e–h). These results clearly show that ATR3::GFP resides in the cytoplasm and the nucleus.

Figure 2.

 ATR3 is localized in the cytoplasm and the nucleus, has a cytochrome c reductase activity but fails to reconstitute a P450-mediated reaction. (a–d) Localization of ATR3::GFP and mRFP-HDEL in tobacco leaf epidermal cells using confocal microscopy. (a–c) cortical view showing that the majority of ATR3::GFP is in the cytoplasm (a) and does not colocalize with the endoplasmic reticulum (ER) network labeled with mRFP-HDEL (b,c). (d) Merged nuclear view showing that ATR3::GFP is cytoplasmic and nuclear but excluded from the nucleolus. (e–l) Nucleus views of tobacco BY-2 cells stably coexpressing ATR3::GFP and mRFP-HDEL (e–h) or cGFP and mRFP-HDEL (i–l). ATR3::GFP and cGFP highlight the cytoplasm (e–i) whereas mRFP-HDEL resides in the ER (f–j). Merged images show that the proteins are located to distinct compartments (g–k). Panels h and l are transmitted light images corresponding to panels e–g and i–k, respectively. (a,e,i) ATR3::GFP (488 nm argon ion laser line; λ emission: 493–538 nm); (b,f,j) mRFP-HDEL (543 nm HeNe laser line; λ emission: 580–620 nm); (c,d,g,k) merged images. Bars, 8 μm; N, nucleus; n, nucleolus. (m) Soluble (S) and microsomal (M) fractions from Nicotiana benthamiana expressing GFP or ATR3::GFP were resolved by a 10% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS–PAGE). After transfer, the blot was probed with GFP antibody. Protein band(s) corresponding to ATR3::GFP fusion protein and to GFP are indicated by arrow and asterisk, respectively. (n) Cytochrome c reduction by GFP and ATR3::GFP from microsomal and soluble fractions expressed in tobacco. The control indicated as GFP corresponds to the vector, which includes the GFP lacking a fusion and represents the background activity. (o) O-deethylation of 7-ethoxycoumarin by CYP1A2 to form 7-hydroxycoumarin in the presence of human cytochrome P450 reductase (CPR) or ATR3.

ATR3 reduces cytochrome c but not the human P450 CYP1A2

To establish the catalytic properties of ATR3, recombinant ATR3::GFP was transiently expressed in N. benthamiana by agroinfiltration. The ATR3::GFP fluorescence pattern observed in the epidermal cells of N. benthamiana showed a strong accumulation in the cytosol similar to that observed in Fig. 2 (data not shown). Immunoblot analysis of the soluble and microsomal fractions isolated from agroinfiltrated leaves was performed using GFP antibodies (Fig. 2m). In leaf samples expressing GFP alone, a band of c. 27 kDa was observed in the soluble fraction. No free GFP was detected in leaves over-expressing ATR3::GFP. The recombinant ATR3::GFP protein was detected predominantly in the soluble fraction and to a lesser extent in the microsomal fraction where contamination with soluble ATR3::GFP cannot be excluded. Cytochrome c, an artificial acceptor of electrons from CPRs, was used as a substrate for assaying the reductase activity of CPRs (Murataliev et al., 2004). Cytochrome c reduction by ATR3 or ATR3::GFP was first assayed in yeast in order to confirm that the ATR3::GFP fusion was functional. The activities measured were equivalent with and without the GFP tag (ATR3::GFP 4.78 ± 0.66 μmol min−1 mg−1 while vector was only 1.32 ± 0.4 μmol min−1 mg−1; ATR3 6.3 ± 0.14 μmol min−1 mg−1 while vector was only 1.86 ± 0.09 μμmol min−1 mg−1; each value is the mean ± SE from three independent protein preparations). Subsequently, the rate of reduction of cytochrome c by recombinant ATR3::GFP in the soluble and microsomal fractions of agroinfiltrated leaves was determined in the presence of NADPH. Both fractions were functional, although a higher activity was measured in the soluble fraction (Fig. 2n). As the recombinant ATR3::GFP expressed in N. benthamiana could be appropriately modified and suited for enzymatic activity, the plant cytosolic and microsomal ATR3::GFP were tested for cytochrome P450 reductase activity by in vitro reconstitution experiments with CYP1A2. In contrast to the human CPR, neither soluble nor microsomal ATR3::GFP was able to reconstitute the O-deethylase activity of CYP1A2 in vitro (Fig. 2o). These results indicate that ATR3, like NR1, has a cytochrome c reductase activity but fails to reconstitute a P450-mediated reaction.

Genetic analysis of ATR3 loss of function

We have characterized an embryo-defective (emb) mutant (EYL4) by screening the INRA collection of T-DNA insertion lines and determined that the T-DNA has inserted into the fourth exon of ATR3 (Fig. 3a). The line was renamed atr3-1. Heterozygous atr3-1/+ plants produced a 2 : 1 ratio of KanR : KanS plants, which is consistent with a single locus of T-DNA integration linked to an homozygous lethal emb mutation (1157 KanR : 561 KanS seedlings, χ2 = 0.36). To confirm that the emb phenotype resulted from the disruption of ATR3, heterozygous atr3-1/+ plants were transformed with a T-DNA carrying a wild-type ATR3 gene and the hygromycin resistance marker. The segregation analysis of the progeny of nine independent T1 lines is provided in the Supporting Information, Table S1. Four T1 plants were homozygous for the atr3 mutation (lines 1, 2, 5 and 7) and would not exist unless there is complementation. This experiment demonstrated that ATR3 plays an essential role in embryogenesis for which ATR1 and ATR2 are unable to compensate. An independent atr3 allele (SALK_123688) was identified in the SALK Institute collection (Alonso et al., 2003). The T-DNA has inserted into the last exon of ATR3 (Fig. 3a). Siliques of SALK_123688/+ plants contained 50% aborted seeds and further analyses highlighted that the SALK_123688 line has two distinct emb loci located on the same chromosome in genetic opposition (emb1 + : + emb2). Back-crosses allowed us to isolate the atr3-2 allele. Crosses between atr3-1 and atr3-2 heterozygous plants demonstrated that the two mutations were allelic (655 wild type seeds, 200 emb seeds, χ= 1.11).

Figure 3.

 Genetic analysis of ATR3 loss of function. (a) Schematic representation of intron–exon structure and T-DNA integration in the ATR3 gene. Black boxes, exon; black line, intron. LB, T-DNA left border. (b) Distribution of embryo phenotypes for atr3-1 and atr3-2 alleles. (c–k) Phenotypes of atr3-1 and atr3-2 mutant embryos. (c) 2-cell atr3-1 embryo and (f), 8-cell atr3-2 embryo when wild type is at globular stage (i). (d) 2–4 cell atr3-1 embryo and (g), atr3-2 embryo when wild type is at triangular stage (j). (e) Four-cell atr3-1 embryo and (h), atr3-2 embryo when wild type is at heart stage (k). (g,h) atr3-2 embryos with more than eight cells. Bar, 20 μm. Asterisks point to individual endosperm nucleoli. Black arrows mark the limit between embryo and suspensor. E, embryo proper; S, suspensor.

The phenotype of atr3/atr3 seeds was studied in detail. The percentage of emb seeds in the siliques is, on average, 27% for the atr3-1 allele (total seed number, 1481; emb seeds, 400; χ= 3.19) and 24.2% for the atr3-2 allele (total seed number, 1032; emb seeds, 250; χ= 0.33), values not significantly different from the 25% expected for a recessive lethal sporophytic mutation. However, the transmission of the T-DNA was determined in reciprocal crosses in order to test the possibility of additional gametophytic lethality (Table S2). No bias of transmission was observed and this confirmed the sporophytic nature of the atr3 mutation. The major stage of arrest of embryo development was two- to four-cell stages in atr3-1 and eight cells in atr3-2 (Fig. 3b). This difference in the severity of the phenotype may be caused by the position of the T-DNA insertion in the ATR3 gene (Fig. 3a). In atr3-2, the production of a nearly complete ATR3 protein, although missing a functional catalytic triad, may allow the embryo to develop further compared with atr3-1 embryos. Fig. 3(c–k) presents typical mutant embryos observed for atr3-1 (c,d,e) and atr3-2 (f,g,h) alleles when the wild-type seeds of the same silique were at globular (i), triangular (j) and heart (k) stages. The endosperm nuclei were able to divide and the number of nucleoli was consistent with the number usually present in a wild-type seed containing a two- to eight-cell stage embryo. This suggests that the development of the embryo and the endosperm are arrested at the same time. Moreover, in cases when the embryo eventually produces more than eight cells (Fig. 3g,h), it fails to perform periclinal divisions, which is considered as the first step of the radial patterning of the embryo (Mayer & Jurgens, 1998).

ATR3 is regulated during cell cycle progression

Genes that play essential functions in early embryogenesis may be involved in cell division. For example, TITAN and PILZ genes act at late phases of the cell cycle and are involved in the mechanisms of chromosome separation (Steinborn et al., 2002; Tzafrir et al., 2002). DNA polymerase epsilon (Jenik et al., 2005; Ronceret et al., 2005) and thymidylate kinase genes (Ronceret et al., 2008) act at early phases of the cell cycle and are transcribed at the G1/S-phase transition of the cell cycle. A detailed examination of the promoter region of ATR3 revealed the presence of conserved cis-elements, CHR (cell cycle homology region) and CDE (cell cycle-dependent expression), known to be involved in transcriptional cell cycle regulation of tobacco RNR2 gene (Chaboute et al., 2000). A reverse CHR/CDE at −1100 bp and a forward CDE in the 5′ leader sequence are found upstream from the initiation codon of ATR3 (Fig. S3). A 1.6 kb promoter region containing the potential CHR and CDE cis-elements was fused to luciferase (Fig. 4a) and used to stably transform tobacco BY-2 cells. Cells were synchronized by aphidicolin and luciferase activity was measured at different time-points of the cell cycle. A peak of luciferase activity was visible 4–5 h after drug washout occurring after DNA synthesis (S phase) and preceding the maximal mitotic index reached at 8–9 h (M phase; Fig. 4b). This induction peak was reproducible in three independent transformation experiments (Fig. 4c). These results demonstrated that the ATR3 promoter is most probably activated at the G2 and G2/M-phase transition of the cell cycle.

Figure 4.

 Cell cycle activation of the ATR3 promoter. (a) Schematic representation of ATR3 promoter::luciferase construct. The 1.6 kb ATR3 promoter corresponds to the sequence upstream of the ATG. Gray line, promoter region upstream of the +1 transcription; open triangle, cell cycle dependent expression (CDE) sequence; tinted triangle, cell cycle homology region (CHR) sequence. Orientation of the triangle mimics the orientation of the cis-element. (b) Measurement of the ATR3 promoter activity in synchronized stably transformed BY-2 cells. Cells were synchronized by aphidicolin. After removal of the drug, DNA synthesis (upper panel, triangles) was determined by 3H thymidine pulse experiments on the basis of total proteins (counts per minute, CPM. Mitotic index (upper panel, squares) was determined by microscopic observation of 4,6-diamidino-2-phenylindole (DAPI) stained cells. Luciferase activity (lower panel, diamonds) was measured on cells harvested at different time points and expressed in RLU (relative light unit) per total protein content. The position of cell cycle phases is indicated between graphs. (c) Luciferase activity (RLU) measured for three independent transformation experiments (T1, diamonds; T2, squares; T3, triangles) and their corresponding mitotic index (MI 1, diamonds; MI 2, squares; MI 3, triangles, dotted lines). The position of cell cycle phases is indicated above the graph. The G2 induction peak was reproducible in the three independent transformation experiments. The T2 experiment is the one presented in more detail in (b).

Developmental expression of ATR3

The level of expression of the ATR3 gene is very low during the Arabidopsis plant life cycle, as recorded in microarray databases and confirmed by RT-PCR experiments where 45 cycles are required to detect its expression compared with 23 cycles for the constitutive control EF1α (Fig. 5a). Despite its low level of expression, the ATR3 band of 267 bp was detected in every sample tested. The spatial and temporal expression of ATR3 during Arabidopsis development was analysed in five independent transgenic lines containing the GUS gene under the control of the 1.6 kb ATR3 promoter. Overall, GUS expression was not detected throughout plant and embryo development with the exception of dry seeds 24–48 h after imbibition and the tapetum of the anthers (Fig. 5b) at the floral stage 10 according to (Bowman, 1994). mRNA in situ hybridizations were performed but no signal was detected throughout embryo development, suggesting that the level of expression of the ATR3 gene is below the threshold of detection.

Figure 5.

 Cell cycle regulation of the ATR3 promoter::GUS fusion after seed imbibition. (a) Expression of the ATR3 gene in Arabidopsis. Reverse-transcription polymerase chain reaction (RT-PCR) was performed with RNA from 10-d-old seedlings (Se), roots (R), rosette leaves (L), stem (St), cauline leaves (Cl), flower buds (FB), flowers (Fl), siliques after fertilization up to torpedo stage (immature stage, IM) and siliques at mature cotyledon stage (MS). Lane C shows the control with no DNA and lane G, the control with genomic DNA. Expression of EF1α was used as a positive and constitutive control. (b) illustrates ATR3 promoter::GUS expression in the tapetum at floral stage 10 according to (Bowman, 1994). Glucuronidase (GUS) staining was carried out for 48 h. Bar = 20 μm (c) Graphic representation of cell cycle progression during seed imbibition and germination based on (Barroco et al., 2005) (lane 1) and seed imbibition experiments carried out using aphidicolin and colchicine (lanes 2–5). Time is given in hours after seed imbibition (HAI). Lanes 2–3, 24 h block and release experiment. Drugs were removed at 24 HAI; GUS staining was carried out for 24 h in the absence of fixative allowing cells to proceed through the cell cycle after drug washout; cytological observations were performed at 48 HAI. Lanes 4–5, 48 h block experiment. Drugs were removed at 48 HAI; GUS staining was carried out for 24 h in the presence of fixative preventing cells to proceed through the cell cycle after drug washout; cytological observations were done at 72 HAI. Black arrows (downward) indicate when drug was added and red arrows (upward) when drug was washed out. Blue arrows indicate when GUS staining was performed for 24 h and whether fixative was added (+Fix) or not (−Fix). Green arrows indicate when seeds were cleared in Hoyer’s and observed. Phases of the cell cycle are indicated. Arrowhead indicates radicle protrusion. (d) ATR3 promoter::GUS expression in Arabidopsis roots after 24 HAI in the presence of dimethyl sulfoxide (DMSO) as a control or 5 μg ml−1 aphidicolin or 5 μg ml−1 colchicine. (e) ATR3 promoter::GUS expression in Arabidopsis roots after 48 HAI in the presence of DMSO as a control or 5 μg ml−1 aphidicolin or 5 μg ml−1 colchicine. (f–h) Illustrate ATR3 promoter::GUS expression in Arabidopsis roots with DMSO (f), 5 μg ml−1 aphidicolin for the 24 h block and release experiment (g) and 5 μg ml−1 colchicine for the 48 h block experiment (h), Bar = 20 μm.

As the ATR3 promoter is activated during cell cycle progression and GUS expression is occurring in imbibed seeds, experiments similar to those described in Wang & Liu (2006) were performed. Barroco et al. (2005) have demonstrated that most cells of dry seeds are arrested in the G1 phase. During imbibition, DNA synthesis is initiated in the radicle. A major transition through S phase toward G2 was detected when the radicle starts to protrude 42 h after seed imbibition (HAI), while entry into M phase occurred after radicle protrusion (Fig. 5c, lane 1). Dry seeds of two homozygous ATR3 promoter::GUS lines showing the highest GUS activity were imbibed in 5 μg ml−1 aphidicolin or 5 mg ml−1 colchicine, which arrests cells at S phase or M phase, respectively (Fig. 5c, lines 2–5). A DMSO treatment was used as a control as both drugs were solubilized in this solvent. The two lines exhibited similar responses to drug treatments. In the first experiment (Fig. 5c), seed imbibition was carried out in sterile water for 24 h in the presence of DMSO as a control or in the presence of aphidicolin (lane 2) or in the presence of colchicine (lane 3). The seeds were then washed with sterile water to remove the drug and incubated in GUS staining buffer without methanol (− Fix) for 24 h. Cytological observations were performed at 48 HAI (including the period of GUS staining). Under such conditions, the absence of fixative allows cells synchronized by aphidicolin to proceed through the G2/M phase during GUS staining. The number of roots expressing GUS was c. twofold higher (Fig. 5d) and at a higher level than the DMSO control (Fig. 5f,g). By contrast, colchicine treatment had no effect on GUS expression, consistent with the removal of the drug before cells advanced through mitosis.

The second experiment (Fig. 5c) was carried out in a similar manner, with the exception of the time for seed imbibition (48 h) and the addition of methanol (+ Fix) in the GUS staining solution. Cytological observations were performed at 72 HAI including 24 h of GUS staining. In this experiment the presence of fixative does not allow cells to proceed through the cell cycle after drug washout. The number of GUS-positive radicles observed for the aphidicolin treatment was not significantly different from the DMSO control indicating that cells arrested at S phase did not express the reporter gene (Fig. 5e). By contrast, the number of GUS-positive radicles observed for the colchicine treatment was threefold higher (Fig. 5e) and at higher level than the DMSO control (Fig. 5f,h), indicating that cells went through the G2/M-phase transition and were arrested at M phase before fixation. Together, these experiments confirmed that the ATR3 promoter is activated during cell cycle progression, most probably at G2 and G2/M phases.

ATR3 interacts with a protein similar to human CIAPIN1

The SRS.Y2H system (Cytotrap; Stratagene) which detects interactions in the cytoplasm (Aronheim et al., 1997) was used to identify proteins that interact with ATR3 in order to provide insight to its biological function. The system is based on the ability of protein interactions to rescue the galactose-dependent growth of the cdc25H mutant yeast strain at the restrictive temperature of 37°C. Screening of an Arabidopsis whole-seedling cDNA library with ATR3 (pSos-ATR3) as bait yielded a single cDNA clone (At5g18400) encoding the homologue of the human cytokine-induced inhibitor of apoptosis protein 1 (CIAPIN1). Sequence alignment of animal CIAPIN1 and plant CIAPIN1 homologues available for genomes entirely sequenced showed that the average 35% amino acid identities and 50% similarities are concentrated in the two functional domains predicted for the human CIAPIN1 (Hao et al., 2008): a generic methyltransferase motif in the N-terminal region and Zn-ribbon-like motifs including several conserved cysteines in the C-terminal region (Fig. 6a). A phylogenetic tree was derived from an m-coffee alignment of 67 CIAPIN1 sequences representative of different phylum (Fig. 6b). This tree indicated that there is clearly a common ancestor for animal and plant lineages. The closest sequence of At5g18400 is the Drosophila simulans sequence (49% identity, 64% similarities) in the Metazoa phylum and the Picea sitchensis sequence (50% identity, 68% similarities) in the Streptophyta phylum. We did not find any other paralogous gene in any of the genomes entirely sequenced to date. Therefore, we can consider that At5g18400 (hereafter designed AtCIAPIN1) is a unique gene and corresponds to the homologue of human CIAPIN1. In retransformation experiments, the growth of cdc25H cells cotransformed with pSos-ATR3 and pMyrAtCIAPIN1 was rescued on SD/galactose but not on SD/glucose at 37°C, thereby confirming the specificity of the protein interaction (Fig. 6c). The negative controls did not grow on galactose at 37°C, whereas the positive control (pSos-MAFB and pMyr-MAFB) grew well on SD/galactose. To confirm the interaction between ATR3 and AtCIAPIN1 in vitro, coimmunoprecipitation assays were performed with ATR3 antibodies. 35S-AtCIAPIN1 coimmunoprecipitated with the soluble recombinant ATR3, but not with the control yeast fraction (Fig. 6d). ATR3 antibodies did not immunoprecipitate the 35S-AtCIAPIN1 in the absence of recombinant ATR3. These results suggest that AtCIAPIN1 may be an acceptor/donor protein requiring ATR3 reductase activity for its function.

Figure 6.

 ATR3 interacts with the Arabidopsis homologue of human CIAPIN1. (a) Sequence alignment of animal and plant CIAPIN1 proteins displaying DUF689 domain (Pfam PF05093) which contains several uncharacterized proteins displaying an N-terminal generic methyltransferase motif (dotted line) and a C-terminal zinc ribbon-like motif (plain line) including several conserved cysteines (highlighted by asterisks). The CIAPIN1 proteins retained for the alignment here are from entirely sequenced genomes and their accession numbers are listed in Fig. S3. Arth, Arabidopsis thaliana (At5g18400); Orsa, Oryza sativa japonica; Mumu, Mus musculus; Hosa, Homo sapiens; Drme, Drosophila melanogaster; Dare, Danio rerio; Sace, Saccharomyces cerevisiae. (b) Phylogenetic tree of 67 CIAPIN1 sequences of the following phyla: Apicomplexa (API: 6), Bacillariophyta (BAC: 2), Chlorophyta (CHL: 1), Choanoflagellida (CHO: 1), Ciliophora (CIL: 2), Fungi (FUN: 16), Metazoa (MET: 30), Mycetozoa (MYC: 1), Streptophyta (STR: 8). Accession numbers are listed in Table S3. The number of replicate trees in which the associated taxa clustered together in the bootstrap test (1000 replicates) is shown next to the branches. Arabidopsis and human sequences are indicated in bold type. (c) SRS-Y2H retransformation assays. Yeast cotransformed with pSos-ATR3 and pMyr-AtCIAPIN1 plasmids were selected for galactose-dependent growth at 37°C. Positive control, pSos-MAFB + pMyr-MAFB; negative controls: pSos + pMyr-AtCIAPIN1, pSos-ATR3 + pMyr, pSos + pMyr. MAFB (v-Maf musculo-aponeurotic fibrosarcoma oncogene homolog B) belongs to the leucine zipper family of transcription factors and forms homodimers (Cytotrap two-hybrid system, Stratagene). (d) Co-immunoprecipitation of AtCIAPIN1 with ATR3. 35S-AtCIAPIN1 translated in vitro was tested for interaction with the yeast cytosolic soluble ATR3 (ATR3-S) in co-immunoprecipitation assays with ATR3 antibodies. Controls included soluble fraction (Vector-S) from yeast transformed with empty vector. Arrow indicates the 35S-AtCIAPIN1.


The Arabidopsis ATR3 gene encodes a diflavin reductase, the homologue of NR1

Several lines of evidence have led us to question the identity of ATR3 as an authentic CPR. The phylogenetic tree clusters ATR3 with the animal oxidoreductases NR1 and Fre-1 distinct from the clade containing the plant and animal CPRs. The FMN, FAD, NADPH-binding domains of plant ATR3 proteins display a higher percentage of identity with their animal counterparts (67%, 45% and 55%, respectively) than with plant CPRs (32%, 40%, and 45% respectively) and the cysteine residue essential for the catalytic activity of CPRs is often replaced by an alanine. All plant and animal ATR3/NR1 proteins lack an ER membrane anchor domain at their N-terminal end, a characteristic of authentic CPRs (Wang et al., 1997). Plant and animal CPRs are membrane bound enzymes, whereas ATR3 was found in the cytosol and the nucleus, a localization similar to that of NR1 (Paine et al., 2000).

Although CYP1A2 is a human P450 enzyme and unlikely to constitute an optimal partner for ATR3, ATR1 and ATR2 have been shown to support the catalytic activity of the human P450 enzyme with differing efficiencies (Louerat-Oriou et al., 1998). Like ATR3, NR1 also failed to reconstitute the O-deethylase activity of the CYP1A2 in vitro (Paine et al., 2000). In conclusion, we showed that ATR3 possesses a cytochrome c reductase activity in the cytosol but we were unable to demonstrate a P450 reductase activity in agreement with its subcellular localization. These results suggest that reduction of the P450 enzymes attached to the endoplasmic reticulum is an unlikely physiological role for both ATR3 and NR1. The observation that loss of ATR3 resulted in an emb phenotype consolidated our hypothesis that ATR3 plays an essential role, distinct from that of ATR1 and ATR2. Knocking out the unique CPR gene results in defects in establishing the polarity in early C. elegans embryo (Rappleye et al., 2003) and embryonic lethality in mouse (Shen et al., 2002). Therefore, neither Fre-1 nor NR1 were able to compensate the loss of CPR function. Taken together, the sequence homologies, subcellular localization, enzymatic activities and loss of function phenotype, demonstrated that ATR3 is the homologue of human NR1 and C. elegans Fre-1 and belongs to the NR1 subfamily of diflavin reductases.

Is ATR3 a novel regulator of cell death in plants?

As ATR3 has physiological partners other than P450s, the next step towards the understanding of ATR3 function was to identify ATR3-interacting proteins. Recent work on NR1 suggested a role in cell death and indicated a functional cooperation between NR1 and DcpS, a scavenger decapping enzyme that plays an important role in the 3′–5′ mRNA decay pathway after the degradation by the exosome. Overexpression of NR1 in human embryonic kidney cells enhances the cytotoxic action of the artificial electron acceptor menadione and DcpS physically interacts with NR1 and modulates its activity promoting cell survival (Kwasnicka et al., 2003). This information prompted us to search for a homologue of DcpS in Arabidopsis. Structural and biochemical studies have revealed that a central histidine in a histidine triad (HIT) domain in DcpS is responsible for hydrolysis of the cap structure (Liu et al., 2002). The best Arabidopsis DcpS candidates displayed identities only in the histidine triad domain and this limited similarity may question their functional homologies. These candidates when tested in yeast two-hybrid assays, failed to interact with ATR3 (data not shown).

The yeast two-hybrid assay strategy was pursued by screening an Arabidopsis seedling cDNA library. We identified an Arabidopsis protein homologous to CIAPIN1 as an ATR3 interacting partner. Human CIAPIN1 was shown to be a mediator of the RAS signalling pathway and to play a vital role in fetal liver hematopoiesis (Shibayama et al., 2004). CIAPIN1 is described as a novel anti-apoptotic molecule that does not show any homology to known apoptosis regulatory molecules such as Bcl-2 family members or caspases (Shibayama et al., 2004). Microarray analysis revealed reduced expression of Bcl-xL and Jak2 in fetal liver of ciapin1 null mice (Shibayama et al., 2004) and depletion of CIAPIN1 was shown to trigger apoptosis in HCC cells downregulating Bcl-2 and Bcl-xL and upregulating Bax (Li et al., 2008). This suggests that CIAPIN1 might exert its function by modulating the expression of apoptosis regulatory genes. Although no interaction has been reported between the human NR1 and CIAPIN1, such an interaction was observed in yeast. The genome of yeast encodes both a CPR (ScCPR) and an ATR3/NR1 (TAH18) diflavin reductase (Figs 1a, S1). The interaction between TAH18 and Dre2, the yeast homologue of CIAPIN1 was found to control oxidative stress-induced cell death (Vernis et al., 2009). Dre2-Tah18 interaction takes place in the cytoplasm in the absence of oxidative stress. The tah18 and dre2 mutated alleles are synthetic lethal in yeast. Moreover, human CIAPIN1 is able to replace the yeast Dre2 in vivo and physically interacts with TAH18. The involvement of the Dre2–TAH18 complex in apoptosis, together with the role of human CIAPIN1, indicates a possible role for ATR3 in programmed cell death via an interaction with AtCIAPIN1.

In plants, the cell death is critical for the establishment of polarity at early stages of plant embryogenesis, when the differentiation of the temporary basal organ, the suspensor, is followed by its programmed elimination (Bozhkov et al., 2005). However, the observation of mutant embryos for atr3-1 and atr3-2 alleles did not reveal any abnormalities in the suspensor development as the embryo development arrests much earlier than the torpedo stage coincident with suspensor degeneration. Interestingly, we observed a transient expression of ATR3 in the tapetum during anther development at floral stage 10 according to (Bowman, 1994). All the programmed cell death (PCD) hallmarks have been reported for tapetum degeneration and the PCD signal commences at the tetrad stage (Kawanabe et al., 2006) corresponding to floral stage 9 according to (Bowman, 1994). Therefore, PCD signalling precedes ATR3 expression (as monitored by GUS activity) at stage 10, which occurs before cells degenerate at stage 11–12. This chronology is consistent with an involvement of ATR3/AtCIAPIN1 complex in PCD. However, the embryonic lethality of ATR3 prevented us from further addressing its role during the later stages of the plant life cycle. By example, a putative role of ATR3 in the PCD of tapetum cells cannot be functionally tested in the heterozygous atr3/+ plants. Similarly, a role of ATR3 in germination and in plant growth could not be investigated in the knockout lines.

Alternative function of ATR3 in DNA replication and cell division

Is the function of ATR3 uniquely restricted to the control of cell death? Although the Dre2–TAH18 complex controls cell death in yeast, this process is conditioned by stress. This raises the question of the nature of the essential function of Dre2–TAH18 interaction in absence of stress.

TAH18 function was identified as a gene involved in impaired DNA replication and possibly DNA repair/DNA integrity check (Vernis et al., 2009). In the same way, thermosensitive dre2 mutant exhibits a cin phenotype (cin for increased chromosome instability) and this phenotype was scored as severe in chromosome transmission fidelity (CTF) assays (Ben-Aroya et al., 2008). In the same way, expression of the human CIAPIN1 influences the rate of cell division. Ectopic expression of CIAPIN1 by cDNA transfection resulted in suppression of cell proliferation and inhibition of cell cycle progression while knockdown of CIAPIN1 with siRNA accelerated cell proliferation and promoted cell cycle progression in these cells (Hao et al., 2009). In this context, the confirmation of ATR3 as a gene essential for early embryogenesis brings a novel perspective to the role of ATR3-AtCIAPIN1 interaction in cell division. We (Ronceret et al., 2005, 2008), and others, have demonstrated the importance of cell cycle related genes at the gametophytic and sporophytic transition and at early embryo development. Furthermore, the transcription of ATR3 itself is regulated during the cell cycle at the G2 and G2/M-phase transition. Similarly loss of cdc5, expressed at the G2/M transition, also arrested Arabidopsis embryo development at an early stage (Lin et al., 2007). The control of ATR3 transcription could constitute a mechanism to regulate AtCIAPIN1 activity during the cell cycle.

Based on our results and the comparison with the functions of their human and yeast homologues, we propose that ATR3-AtCIAPIN1 interaction in Arabidopsis may constitute an important yet previously unrecognized component in the general mechanisms of cell division and cell death.


We thank Nicole Bechtold and Roger Voisin for their work on the INRA T-DNA insertion collection. We thank Danièle Werck-Reichhardt, Loïc Faye, Thomas Roscoe, Jean-Philippe Reichheld and Christophe Riondet for critical reading of the manuscript. The work of VD, JG and MD was supported by the GENOPLANTE AF015 grant. CSJD and VG are supported by the Centre National de la Recherche Scientifique (CNRS), France. Plant Gateway vectors were provided by Functional Genomics Division of the Department of Plant Systems Biology (VIB-Ghent University, Belgium).