An Arabidopsis RNA lariat debranching enzyme is essential for embryogenesis.

An embryo-defective mutant of Arabidopsis thaliana was isolated that arrests development at a variety of stages, from as early as the globular stage of embryogenesis to as late as formation of an abnormal bent cotyledon stage embryo. Defects in the suspensor, a normally transient structure derived from the fertilized egg, were often associated with the arrested embryo. The lesion was within a gene encoding a protein with domains characteristic of lariat debranching enzymes, which has been named AtDBR1 (for Arabidopsis thaliana Debranching enzyme 1). Cleavage of the 2'-5'-phosphodiester bond found in excised intron lariats ("debranching") is essential for turnover of intronic sequences as well as generation of some small nucleolar RNAs. The mutation within AtDBR1 was confirmed by complementation as being responsible for the embryo-lethal phenotype, and the activity of the encoded protein in cleavage of 2'-5'-phosphodiester bonds was verified using an in vitro debranching assay.

Isolation of embryo-defective mutants has been a valuable approach for identification of genes that are essential for plant growth and development and that are not protected by redundancy. Although some embryo-defective mutants, such as lec1 and lec2, represent lesions in genes specific to the embryonic phase (1-3), more often the defective genes would normally be expressed not only during embryo development but throughout the life cycle of the plant. In these cases, the embryo lethality is due to the fact that a recessive mutation in homozygous state cannot survive past this early point in development. Thus, in a broader sense, many embryo-defective mutants (often referred to as emb mutants) define a set of non-redundant genes essential for life (4,5). Of 110 EMB genes recently characterized, 42% encode proteins of unknown function, whereas the remainder are involved in a diverse range of processes including metabolism, cell growth, transcription, and translation (5). Many of these genes are likely to be expressed throughout the plant life cycle. For example, the ttn5 and gnom/emb30 mutants show defects very early in embryo development, but the wild-type genes encode proteins involved in vesicular traffick-ing at all stages of development (6 -8). Research programs, such as SeedGenes, are oriented toward coordinating identification of all genes that produce embryo-lethal phenotypes when disrupted in order to better understand not only embryo development but also the minimal set of genes necessary for cellular processes (see Ref. 9 and www.seedgenes.org/). There are postulated to be 500 -750 genes encoding non-redundant functions in this minimal set in Arabidopsis (5,9).
A subclass of embryo-defective mutants shows abnormal development of suspensor cells. In a wild-type Arabidopsis embryo, the suspensor, which is derived from the basal cell of the first zygotic division, forms a single file of seven to nine cells (reviewed in Ref. 10). The suspensor provides a supportive role for the embryo proper and normally degenerates during later embryo development. Although part of the basal cell lineage persists in the embryo proper in the form of the root cortex initials and central part of the root cap, the remainder of the embryo proper is derived from the apical cell of the two-cell embryo (reviewed in Ref. 11). The suspensor (sus) and raspberry mutants show defects in morphogenesis of the embryo proper, with the most common arrest occurring at the globular stage of development. In these mutants, once defects become apparent in the embryo proper, cells in the suspensor began to divide abnormally (12,13). Abnormal development of the suspensor ranges from a few extra divisions, which may be accompanied by expression of programs normally limited to the embryo proper (12,13), to entire viable embryos in the case of the twin (twn) mutants (14,15). It has been proposed that some yet unidentified signal from the embryo proper may repress the developmental potential of the suspensor. In the sus, raspberry, and twn mutants, this signal is not produced, and cells in the suspensor acquire different degrees of embryo identity (12).
Here we report the isolation of an embryo-lethal mutant and identification of the genetic lesion. Embryos homozygous for this recessive allele arrest at a variety of stages and show a range of associated suspensor defects. The lesion is within a gene that encodes a putative lariat debranching-like enzyme. Lariat debranching enzymes (DBRs) 1 are involved in cleavage of a 2Ј-5Ј-phosphodiester bond that is generated during excision of introns (16,17). This cleavage converts the forked or branched lariat intron to linear form that can be subsequently degraded (18,19). In some cases, the excised intron encodes one or several small nucleolar RNAs (snoRNAs) that are necessary for processing of ribosomal RNAs and small nuclear RNAs (reviewed in Refs. 20 -22). Generation of at least some functional snoRNAs relies on debranching of the intron (23). Our results indicate that some aspect of debranching is essential for normal embryo development.

Plant Materials and Growth Conditions-Arabidopsis thaliana
ecotype Wassilewskija (Ws) seeds and seeds from dbr1 heterozygous mutant plants with or without complementation constructs were sown on germination media (supplemented with 10 g/liter sucrose, 0.5 g/liter MES, and 7 g/liter agar, pH 5.6 -5.7, with 50 g/ml kanamycin for transgenic seeds, and for complementation an additional 20 g/ml hygromycin was added) (see Ref. 24), chilled for 2 days at 4°C, and transferred to a growth room at 23-24°C for a 23-h light/1-h dark regime. After 7-10 days, seedlings were transplanted to potting mix (ProMix BX, Premier Brands, Inc., Canada). Plants were grown at 20/18°C under a 16-h light/8-h dark regime in a growth chamber.
Isolation of the dbr1 Mutant and Characterization of the Phenotype-dbr1 was isolated during generation of plants carrying a ␤-glucuronidase (GUS) reporter gene that was expressed by the 5Ј-regulatory region of DTA1 (25). One line carrying the DTA1:GUS construct was noted to produce ϳ25% aborted seed. After verifying the presence of a single transferred DNA (T-DNA) insert, by backcrossing to wild-type and by kanamycin resistance segregation, further analysis, TAIL-PCR, and complementation studies were performed as described below.
To examine the defect in the dbr1 mutant, developing seeds were made transparent by incubation in Hoyer's solution overnight (12), followed by examination on a Zeiss Axioplan2 equipped with a 35-mm camera and using DIC optics (Carl Zeiss, Inc., Thornwood, NY). Slides were scanned using a Nikon LS-2000 scanner (Nikon Corp., Japan) and assembled using Photoshop 5.0 and Illustrator 7.0 (Adobe Systems, Mountain View, CA).
Thermal Asymmetric Interlaced (TAIL)-PCR-TAIL-PCR was performed as described previously by Liu et al. (26), with minor modifications. Genomic DNA was isolated from plants heterozygous for the T-DNA insertion causing the embryo lethality according to a protocol obtained from the Wisconsin Arabidopsis Knockout facility (www.biotech.wisc.edu/Arabidopsis/). Three T-DNA-specific primers (TR1, TA-ATGGTTTCTGACGTATGTGCTT; TR2, TGTGCTTAGCTCATTA-AACTCCAG; and TR3, TTCTGTCAGTTCCAAACGTAAAAC) and an arbitrary degenerate primer (AD2) (26) were used for the primary, secondary, and tertiary reactions using KlenTaq1 TM (Ab Peptides, St. Louis, MO) and a PTC-100 TM thermocycler (MJ Research, Inc., Waltham, MA), as in Liu et al. (26), except slightly more stringent annealing temperatures were used in some reactions. A single band from the tertiary PCR was gel-purified and sequenced using TR3 as a primer.
After bolting, individual lines corresponding to the T2 generation of DBR1/dbr1 plants carrying the dexamethasone-inducible complementation construct pTA-cAtDBR1 were sprayed daily with 30 M dexamethasone (Sigma) containing 0.02% Silwet L-77 (Lehle Seeds, Round Rock, TX) for 10 consecutive days. After 10 days, dexamethasone treatment was ceased on one set of plants but continued on another set of plants. Staged siliques were opened 3-4 days later and cleared in Hoyer's solution as described above to assess embryo development. Siliques were also examined 9 days later, and the number of aborted seeds was counted.
Expression Analysis-Total RNA was extracted from a variety of tissues using the TRIzol Reagent (Invitrogen). The hot borate method was used to extract total RNA from siliques (28). Reverse transcription and semi-quantitative PCR were performed as described previously (25) by using hot start at 94°C and 40 cycles for AtDBR1 and 24 cycles for TUB2 under the following conditions: 94°C for 30 s, 58°C for 30 s, and 72°C for 45 s, followed by a final extension at 72°C for 7 min. The primers were as follows: for AtDBR1, 5Ј primer and 3Ј primer were 5Ј-GCAAAAACCATACTTTCGTC-3Ј and 5Ј-ATCGATTCATGCATCGT-CTCTTGTATGA-3Ј. For TUB2, primers were 5Ј-CTCAAGAGGTTCTC-AGCAGTA-3Ј and 5Ј-TCACCTTCTTCATCCGCAGTT-3Ј. The PCR products were visualized on 1.2% agarose gels and captured using a ChemiImager (Alpha Innotech Corp., San Leandro, CA).
AtDBR1 Enzymatic Activity-To test whether AtDBR1 was able to cleave 2Ј-5Ј-phosphodiester bonds, the protein was produced in Escherichia coli and used in in vitro debranching assays (17,29,30). Briefly, the sequences encoding amino acid residues 1-418 were placed under control of an inducible promoter in pET28a (Novagen, Madison, WI) and introduced into the expression host BL21-CodonPlus(DE3)-RIL (Stratagene, La Jolla, CA). Cells were grown to an A 600 of 0.4, and expression of the construct was induced by addition of 0.1 mM of isopropyl-1-thio-␤-D-galactopyranoside and incubation at 18°C for 10 h, followed by an additional 3 h at 4°C. Cells carrying the pET28a empty vector were also grown and induced as controls. S30 cell extracts were prepared as described by Kim et al. (30). Briefly, ϳ0.15 g of cell pellet was resuspended in 600 l of buffer (0.1 M Tris-HCl, pH 7.5, 2 mM EDTA, 1% Triton X-100) containing 100 g/ml lysozyme (Sigma) and lysed by incubation at 37°C for 30 min with gentle shaking. The lysate was centrifuged at 33,000 ϫ g for 1 h, and the presence of the induced protein in the supernatant was confirmed by SDS-PAGE and protein gel blot analysis using monoclonal antibody specific for a T7 tag (Novagen, Madison, WI) at the amino-terminal end of AtDBR1.
The substrate for the debranching assay was produced from E. coli harboring plasmid pDB808 (generous gift of Dr. J. Boeke, The Johns Hopkins University School of Medicine, Baltimore, MD). pDB808 causes accumulation of msDNA Ec86 that was isolated and radiolabeled with [␥-32 P]ATP and gel-purified as described previously (17). The debranching assay was performed as described in Ref. 17 by using 30 l of S30 extract from cells expressing AtDBR1 or control cells, and 2-5 fmol (ϳ2000 cpm) of the radiolabeled msDNA in a 50-l reaction for 1 h at 30°C (reaction buffer consisted of 20 mM HEPES, pH 7.6, 40 mM KCl, 3 mM MgCl 2 , 1 mM dithiothreitol, and 10% glycerol). The reactions were extracted with phenol/chloroform/isoamyl alcohol (25:24:1), ethanolprecipitated, resuspended in 50 l of formamide containing 0.05% xylene cyanol and 0.05% bromphenol blue, and an aliquot (15 l) was analyzed on a 20% polyacrylamide gel containing 8 M urea.

dbr1 Embryos Arrest at a Variety of Stages and Exhibit
Suspensor Defects-In the course of generation of transgenic plants expressing a reporter construct DTA1:GUS (25), one line was obtained that produced approximately one-quarter aborted seeds. Initially, aborted seeds appeared white or transparent in color at a stage when wild-type appearing siblings were green (data not shown). Later in development, the aborted seeds were dark brown and withered in appearance (Fig. 1A). The lethality was not due to expression of the transgene because this individual line was the only one out of 72 analyzed that exhibited a seed-lethal phenotype. 2 It was more likely that the T-DNA inserted into a gene essential for seed development.
Seeds were cleared in Hoyer's solution at different stages of development to examine the progression of embryogenesis in defective seeds and in wild-type appearing siblings from the same silique. When wild-type appearing (i.e. greening) seeds were at the torpedo stage of development (Fig. 1B), aborted seeds from the same silique (i.e. white in color) remained at the globular stage (Fig. 1F) or successfully transited to bilateral development, but had a stunted appearance (Fig. 1J). Occasionally, the aborted seeds appeared to contain an embryo arrested as early as a two-cell embryo (data not shown). The suspensor, which normally develops into a single file of cells, often exhibited abnormal divisions (Fig. 1F, arrowhead). When wild-type embryos were at an early bent cotyledon stage of development (Fig. 1C), embryo development in defective seeds ranged from globular (Fig. 1G) to a somewhat abnormal early torpedo (Fig. 1K). When wild-type siblings were at later stages of development (Fig. 1, D and E), the mutant siblings were delayed and showed a variety of abnormalities. In some seeds, the cotyledons did not develop normally, and the embryo had a triangular shape (Fig. 1H). In other cases one cotyledon was somewhat developed, whereas the other was more severely stunted (Fig. 1I). In other instances, both cotyledons developed but were stunted in appearance, as was the hypocotyl (Fig. 1, L and M). Abnormal suspensor development could be observed in some aborted seeds (Fig. 1, F-I, arrowheads). In other cases, the suspensor appeared to develop (Fig. 1, J and K) and subsequently degenerate (Fig. 1, L and M) normally. In yet other seeds, the suspensor showed dramatic development and appeared to form a secondary, albeit abnormal, embryo (data not shown).
Identification of the Disrupted Gene-Genetic tests were performed to assess whether a T-DNA insertion was likely to be associated with the disrupted gene responsible for the embryo lethality. The transgene included the NPTII gene that confers kanamycin resistance, allowing the ratio of kanamycin resistant to sensitive plants to be assessed to determine the likely number of T-DNA inserts. Heterozygous DBR1/dbr1 plants produced viable seeds that showed a kanamycin resistant to sensitive ratio of 2.0 to 1 (679 seedlings screened) indicative of a single T-DNA insert. In addition, seeds obtained from selfpollination of individual kanamycin-resistant plants always showed ϳ25% aborted seeds, further indicating that the lesion causing the embryo lethality was likely due to the T-DNA insertion. Plants producing all kanamycin-resistant seeds were never obtained, indicating that the mutant allele could not be transmitted in the homozygous state and therefore that the penetrance of the embryo lethal phenotype was 100%.
To identify the gene potentially disrupted by the T-DNA, TAIL-PCR was performed. After tertiary TAIL-PCR using T-DNA-specific and degenerate primers, an ϳ500-bp product was obtained, sequenced, and used to search the Arabidopsis database. The sequence matched a region of chromosome 4 containing a gene (At4g31770) encoding a protein predicted to contain a carboxyl-terminal lariat debranching enzyme domain (amino acid residues 234 -352; Pfam profile PF05011). In other organisms, this domain is always found in association with a calcineurin-like phosphoesterase domain. At4g31770 contains such a domain at the amino-terminal end (amino acid residues 1 to 229; Pfam profile PF00149). Therefore, the product encoded by At4g31770 is predicted to function in cleavage of the 2Ј-5Ј-phosphodiester branch point linkage generated during excision of introns. The gene was named AtDBR1 (for Arabidopsis thaliana Debranching enzyme 1) in accordance with the nomenclature used for other organisms (30,31). The predicted protein is relatively conserved among eukaryotes with the Arabidopsis protein having 37-47% identity and 57-65% similarity to DBR proteins from mouse (32), human (30), yeast (31), and Caenorhabditis elegans (18). AtDBR1 appears to be unique within the Arabidopsis genome with no other highly homologous protein product encoded by any other gene. The gene structure of AtDBR1 consists of nine exons encoding a protein of 418 amino acid residues with a predicted molecular mass of 48.2 kDa. The T-DNA insertion in dbr1 was within the first intron.
Complementation of dbr1-To confirm further that the embryo lethality was due to T-DNA insertion into At4g31770, plants heterozygous for dbr1 (T0, DBR1/dbr1) were transformed with a construct consisting of the ϳ1.25-kb 5Ј region encompassing the native promoter, the coding region including introns, and ϳ0.57-kb 3Ј region of the stop codon. This complementation construct, pCAMBIA-gAtDBR1, included a hygromycin resistance gene. T1 plants carrying both the insertion into AtDBR1 and the complementation construct were selected on germination media containing kanamycin and hygromycin. Resistant seedlings were transferred to potting mix, and subsequent generations were screened for embryo lethality. A transgenic line with a single complementation insert showed reduced embryo lethality, with an average of 6.6% aborted seeds. This was very close to that expected by Mendelian segregation for one insert at an unlinked position (i.e. 1:16). In   FIG. 2. Complementation of dbr1. A, siliques obtained by selfpollination of a DBR1/dbr1 plant (top) and of a DBR1/dbr1 plant containing multiple copies of a complementation construct consisting of DBR1:DBR1 (bottom). B-D, complementation of DBR1/dbr1 with a dexamethasone (dex)-inducible construct. Siliques obtained from DBR1/dbr1 plants carrying the inducible construct that were not treated with dexamethasone (B), that were treated daily with dexamethasone (C), and that were initially treated with dexamethasone, but treatment was stopped and siliques were allowed to develop for an additional 9 days before observation (D). addition, very little lethality was observed in an independent complementation line that contained multiple complementation inserts as shown in Fig. 2A (bottom). Twenty five independent transgenic lines showed reduced seed abortion, indicating partial to complete rescue of the embryo-lethal phenotype by pCAMBIA-gAtDBR1.
Plants heterozygous for dbr1 were also transformed with a dexamethasone-inducible complementation construct (27). After flowering, plants heterozygous for dbr1 and carrying the inducible complementation construct were sprayed with dexamethasone. Whereas untreated plants of the same genotype showed the expected ϳ25% seed abortion (Fig. 2B), siliques from dexamethasone-treated plants contained few to no aborted seeds (Fig. 2C).
Continuous treatment with dexamethasone was necessary for continued growth and development. As shown in Fig. 2D, when dexamethasone treatment was ceased on pTA-cAtDBR1 and DBR1/dbr1 plants, aborted seeds were observed at later time points. Embryo defects were observed as early as 3-4 days after cessation of dexamethasone treatment. Whereas wildtype appearing siblings were at a late torpedo to early bent cotyledon stage of development, other seeds from the same silique exhibited a variety of defects including abnormal suspensor development (Fig. 3, A and B, respectively). The embryos were at an approximately globular stage upon cessation of dexamethasone treatment (data not shown). Defects were also observed when dexamethasone treatment was stopped somewhat later in development (treatment stopped when embryos were at approximately late-heart to early-torpedo stage of development). Wild-type appearing siblings were at late-bent cotyledon to maturation stage of development ( Fig. 3D and data not shown), whereas aborted seeds most often appeared to have a somewhat stunted torpedo shape or triangular shape (Fig. 3E and data not shown). When dexamethasone treatment was ceased later in development, no obvious morphological defects were observed 3 days later (data not shown). Embryos from staged siliques of plants with continued dexamethasone treatment appeared normal and did not show any suspensor defects (Fig. 3, C and F).
Expression Pattern of AtDBR1-To determine the expression pattern of AtDBR1 during normal growth and development, reverse transcription (RT)-PCR analysis was performed using gene-specific oligonucleotide primers, and RNA extracted from a variety of tissues. As shown in Fig. 4, mRNA from AtDBR1 was detected in all tissue types tested, including siliques containing developing seeds. A high number of PCR cycles was necessary to detect AtDBR1 mRNA (40 cycles for AtDBR1 compared with 24 for TUB2) indicating that the level of At-DBR1 mRNA accumulation was relatively low. RNA gel blot analysis confirmed ubiquitous low level accumulation of At-DBR1 mRNA (data not shown).
Functional Analysis of AtDBR1-To determine whether the gene product encoded by AtDBR1 was able to cleave 2Ј-5Јphosphodiester bonds, an in vitro assay was performed by using E. coli-produced AtDBR1 and a reagent derived from a multicopy single-stranded DNA (msDNA). msDNAs are found in some prokaryotes and consist of a DNA molecule linked to an RNA molecule via a 2Ј-5Ј-phosphodiester bond (33). An E. coli strain, containing a plasmid (pDB808) that expresses a msDNA (Ec86) (33), has been used extensively to produce substrate to test activity of DBR enzymes (17,29,30,32) and was used to assess activity of E. coli-produced AtDBR1 (construct kindly provided by Dr. Jef Boeke). The isolated msDNA Ec86 was treated with RNase A to produce a branched nucleic acid consisting of an RNase A-resistant triribonucleotide linked by a 2Ј-5Ј bond to the 5Ј end of an 86-nucleotide DNA molecule and was subsequently radiolabeled for the DBR assay (17).
As shown in lane 2 of Fig. 5, when RNase A-resistant radiolabeled msDNA Ec86 (lane 1) was incubated with E. coli cell extract containing AtDBR1, a product with mobility consistent with released trinucleotide RNA was observed in a manner consistent with results published previously (17,29,30,32). Extracts produced from E. coli cells containing the vector construct alone did not produce the triribonucleotide product (Fig.  5, lane 3). The higher mobility product observed in lane 3 was due to DNases present in the cell extract as determined in previous publications using this system (29). Therefore, At-DBR1 was able to catalyze hydrolysis of the 2Ј-5Ј bond, demonstrating function as a debranching enzyme. DISCUSSION Seed development in the model plant A. thaliana occurs within 2-3 weeks and involves the expression of several thousand genes (34,35). Isolation of mutants has been an invaluable approach to understanding the function of expressed genes and defining their role in developmental processes, but interpretation of embryonic mutants requires some caution. Although some embryo-defective mutants represent lesions in genes that are specific for embryogenesis, the developing embryo, as the initiation of the new diploid sporophytic generation, also represents the first context where a recessive mutant allele is present in a homozygous state. In this sense, embryodefective (emb) mutants have been of primary importance in defining a set of gene products essential for general growth and development. EMB gene products identified thus far are involved in a broad range of functions, including cell growth, metabolism, transcription, and translation, and of the 110 emb genes characterized in detail, 42% are of unknown function (5). It is important to note that many other genes expressed during

FIG. 4. Accumulation of AtDBR1 transcripts in various tissues.
RT-PCR products using gene-specific oligonucleotide primers to assess expression levels of AtDBR1 and TUB2 in tissues from Arabidopsis were analyzed on an agarose gel. embryogenesis encode products that are involved in essential functions. These genes, when mutated, may not produce any obvious phenotypes because many processes are protected by redundancy.
Redundant functions are often due to different members of a protein family having overlapping developmental roles. In Arabidopsis, and in several other organisms including yeast, the RNA lariat debranching enzyme activity appears to be encoded by a single gene, greatly reducing the probability of a redundant function. However, uniqueness cannot be assumed to guarantee lack of redundant functions. Non-homologous gene products may have partially redundant functions or there may be sufficient feedback and cross-talk in regulatory networks to accommodate loss of an individual component (reviewed in Ref. 36). For example, even though there is accumulation of most intron lariats in the Saccharomyces cerevisiae dbr1 mutant (19), other introns are unexpectedly degraded. This was found to be due to a partially redundant function corresponding to the yeast ortholog of RNase III (Rnt1p) that mediates degradation of some lariat introns by recognition and cleavage of doublestranded RNA regions present in these introns (37). It is interesting to note that sus1 (allelic to carpel factory (caf) and short integuments1 (sin1)), another Arabidopsis mutant that shows embryo defects accompanied by abnormal suspensor development, has a lesion in a Dicer-like protein that includes an RNase III domain (38,39).
Regardless of whether any functional redundancy exists for the Arabidopsis DBR1 gene product, the dbr1 mutant itself exhibited an obvious and dramatic phenotype. A T-DNA insertion into the first intron resulted in an embryo-defective phenotype with complete penetrance. The phenotype of dbr1 was particularly interesting in that arrest of embryo development occurred at a variety of stages, from relatively early globular stage arrests to arrest after the basic body plan was laid out. Over 60% of the emb mutants reported to date arrest at or before the globular stage of development with only ϳ8% showing a variable arrest phenotype (5). The furthest the dbr1 embryos progressed was into the bent cotyledon stage, but the cotyledons were small and deformed (data not shown). Often suspensor defects were observed (Fig. 1, F-I), and sometimes development of suspensor cells progressed to a point where a secondary embryo appeared to be present within the seed (data not shown). Unlike the twin mutants, which also produce supernumerary embryos from suspensor cells (14,40), dbr1 embryos were not viable.
The mutant phenotype was complemented by a transgene consisting of the native gene regulatory regions driving expression of AtDBR1 ( Fig. 2A) and by a dexamethasone-inducible AtDBR1 transgene (Fig. 2, B-D), confirming that loss of At-DBR1 caused the embryo-defective phenotype. Interestingly, 3-4 days after cessation of dexamethasone treatment to young siliques, defects were apparent (Fig. 3, B and E). This could indicate that DBR1 is a relatively short lived protein or that, with cessation of expression, the level of DBR1 was not high enough to maintain function in the context of the rapid cell divisions occurring in the embryo. Although the dexamethasone system has been reported as supporting reporter gene expression even 4 days after removal of dexamethasone, the level of accumulation of reporter gene mRNA is greatly decreased by 2 days and is barely detectable by 3 days after removal from dexamethasone-containing media (27). In these experiments, the hormone was supplied continuously by growing the plants on agar media containing dexamethasone. For the experiments presented in this report, the glucocorticoid was applied by spraying on the inflorescences once daily. The hormone must penetrate several layers of maternal tissues to reach the developing embryo. Therefore, it seems likely that the level of inducer would not be as high as when supplied continuously in media as indicated by Aoyama and Chua (27).
The predicted gene product of AtDBR1 contained an aminoterminal calcineurin-like phosphoesterase domain from amino acid residues 1-233. Proteins that contain this type of domain are relatively diverse (41) but, when combined with the lariat debranching domain, function in cleavage of 2Ј-5Ј-phosphodiester bonds such as those found in excised introns. The predicted protein is very conserved among eukaryotes with the Arabidopsis protein having 37-48% identity and 57-66% similarity to proven or predicted DBR proteins or unknown proteins from mouse, human, budding and fission yeast, zebrafish, C. elegans, Drosophila, and Xenopus laevis. In higher plants, the genomic sequences encoding RNA debranching activities have yet to be identified, but a number of sequences are present in EST data bases that likely encode DBR activities. ESTs are present in the data base from barley, wheat, and maize that are over 70% identical to regions of AtDBR1. A cDNA from rice encodes a predicted protein with 64% identity (76% similarity) through the phosphoesterase and lariat debranching domains. ESTs encoding putative lariat debranching activities were obtained from a variety of tissues, supporting the finding that AtDBR1 was constitutively expressed (Fig. 4).
Perturbation of expression of the C. elegans DBR1 gene by RNAi has been shown to cause partial embryo lethality with a slow growth phenotype (see the Supplemental Material in Ref. 42). In yeast, the extent of the defect appears to be correlated with the amount of intronic sequence present in the genome. In S. cerevisiae, ScDBR1 is non-essential. Although excised introns accumulate, no growth defects were observed. Only 2.5% of the genes in S. cerevisiae contain introns (18,30). In the fission yeast Schizosaccharomyces pombe, which has introns in ϳ40% of its genes, a severe growth defect results when Sp-DBR1 is non-functional (18,30). A majority (79%) of Arabidopsis genes contains introns (43), and, as in S. pombe and C. elegans, dbr1 Arabidopsis embryos have a slow growth phenotype. Progression of mutant embryos through morphogenesis was substantially delayed and abnormal compared with phenotypically wild-type siblings (Fig. 1), but ultimately, none of the Atdbr1 homozygous embryos were viable. In addition to morphogenesis, plant embryos must first prepare for and then undergo desiccation during the later stages of embryogenesis in order to produce a mature dry seed. Arabidopsis dbr1 embryos may be unable to survive desiccation because they have not prepared adequately for this stage of development. However, embryo-rescue experiments to culture and maintain homozygous Arabidopsis dbr1 tissue have been unsuccessful. 3 Why would the inability to remove spliced introns produce a growth defect? Cleavage of the 2Ј-5Ј bond is essential for subsequent degradation of the intron (18,19), and one explanation for the growth defects seen in dbr1 mutants is that ribonucleotides remain unavailable for reuse when intron lariats persist. In addition, some snoRNAs have been found to be intronically encoded (reviewed in Refs. 20 -22), and correct processing of at least some snoRNAs requires debranching (23). In vertebrates, all guide snoRNAs are encoded within introns of protein-coding genes (reviewed in Ref. 22). In plants studied thus far, the degree of intronic snoRNAs varies from relatively rare in Arabidopsis (22,44) to more than half-present as intronic gene clusters in rice (22,45). snoRNAs primarily function in modification of rRNAs and small nuclear RNAs by 2Ј-O-ribose methylation and conversion of uridine to pseudouridine. In animals, the targets of several snoRNAs preferentially or specifically expressed in the central nervous system remain unknown, but intriguingly, some have been suggested to be involved in modification of mRNAs, thus potentially providing important regulatory functions (reviewed in Refs. 20 and 21).
AtDBR1 has been demonstrated to have debranching activity as shown in Fig. 5. The msDNA triribonucleotide release assay developed (17,29) and used to demonstrate DBR activity in yeast (29), humans (30), and mouse (32) was utilized here because it is relatively easy to obtain large amounts of the msDNA from E. coli. AtDBR1 was able to catalyze the hydrolysis of the 2Ј-5Ј bond between the RNase-resistant triribonucleotide and the 86-nucleotide DNA, demonstrating debranching function. Debranching activity is necessary for Arabidopsis embryogenesis, and, given the constitutive expression pattern of AtDBR1, is likely needed throughout the plant life cycle. Future work will focus on attempts to rescue homozygous dbr1 mutants by using the dexamethasone-inducible system and by assessing the accumulation of intron lariats in response to loss of AtDBR1.