IMB1, a bromodomain protein induced during seed imbibition, regulates ABA- and phyA-mediated responses of germination in Arabidopsis

Imbibition-inducible 1 (IMB1) is a BET protein

To investigate the role of BET proteins in plants, we identified and isolated a gene putatively encoding a bromodomain-containing protein, At2g34900, annotated in the Arabidopsis Genome Initiative database as ‘RING3-like protein’. The cDNA was obtained by RT-PCR, cloned into a binary vector, and sequenced. Comparison with the GenBank sequence revealed that the gene, which we later named IMB1, contains nine exons and eight introns, with the 8th exon being 42 bp shorter than predicted by the conceptual translation in the database.

The IMB1 gene encodes a protein of 386 amino acids with one 89-amino-acid bromodomain and an N-terminal, 64-amino-acid NET domain (Figure 1a) that is unique to the bromodomain subfamily of BET proteins. The amino acid sequences of the two domains are closely related to those of other putative BET transcription factors identified in the Arabidopsis genome (Florence and Faller, 2001). Moreover, both the bromodomain (Figure 1b) and the NET domain (Figure 1c) of IMB1 also show homology to their counterparts in other previously characterized BET proteins, such as human RING3 (Denis et al., 2000), murine MCAP (Dey et al., 2000), Drosophila Fsh (Haynes et al., 1989), and yeast Bdf1 (Lygerou et al., 1994).

Nuclear localization of IMB1

The IMB1 amino acid sequence contains a C-terminal putative bipartite nuclear localization signal. In addition, bromodomain-containing proteins have been found to be associated with chromatin. Therefore, the prediction was that IMB1 would function in the nucleus. To verify the subcellular localization of IMB1, onion epidermal cells were transiently transformed with an IMB1–YFP fusion protein under the control of the cauliflower mosaic virus (CaMV) 35S promoter. In contrast to the control YFP protein, which was distributed throughout the cytoplasm and the nucleus (Figure 2a), the IMB1–YFP fusion protein was localized exclusively in the nuclei of onion cells (Figure 2b). These results strongly suggest that IMB1 is indeed a nuclear protein.

Expression pattern of IMB1

Northern blot analysis was used to investigate the expression of IMB1 in different tissues and at different stages of the life cycle in Arabidopsis thaliana. The IMB1 transcript was barely detectable in stems, leaves, siliques, and dry seeds, but was present at considerable levels in roots and flowers (Figure 3a). Furthermore, we assessed steady-state levels of the IMB1 transcript at different time points during seed imbibition, the process of water uptake by dry seeds, which is considered to trigger germination (Figure 3b,d). We also investigated the expression pattern of IMB1 at different stages of seed germination (Figure 3b,c).

The IMB1 transcript was barely detectable in dry seeds, but its levels increased markedly during seed imbibition at 4°C in darkness (stratification). Consequently, IMB1 was expressed at high levels after 72 h of stratification (Figure 3b). When the seeds were transferred to 22°C in light, germination (radicle emergence from the seed coat) occurred rapidly and was accompanied by a steep decline in the levels of IMB1 transcript. Indeed, the transcript was barely detectable 72 h after stratification (Figure 3b), at which time cotyledon expansion and greening had already occurred (data not shown). If the seeds were transferred to 22°C but kept in the dark, germinated seedlings exhibited as expected elongated hypocotyls and etiolated cotyledons (data not shown) and the IMB1 transcript was also barely detectable after 72 h (Figure 3c); expression levels were therefore comparable to those observed when the seeds were allowed to germinate in light for the same period of time (Figure 3b). This indicates that the down regulation of IMB1 expression during germination is light-independent. In addition, if seed imbibition occurred in darkness but at 25°C, IMB1 was similarly induced except that the rise in transcript levels occurred more rapidly than at 4°C, with the gene being highly expressed after 24 h (Figure 3d). This indicates that the observed up regulation of IMB1 transcript levels during seed imbibition is temperature independent.

Molecular characterization of an IMB1 knockout

To gain insight into the function of IMB1, we screened a Ds-transposant collection (Parinov et al., 1999; Sundaresan et al., 1995) for transposon insertions within the IMB1 gene. To this end, the genomic sequence was used to screen a database containing the sequences adjacent to the insertion sites within the transposant collection (Parinov et al., 1999). One line (SET6311) with an insertion in IMB1, hereafter referred to as imb1, was identified. Sequencing of PCR fragments obtained using primers flanking the insertion site and a 5′-end Ds primer confirmed that the transposon was inserted in the first intron of IMB1, 239 bp downstream of the ATG start codon, in the orientation indicated in Figure 4(a). Moreover, comparison of PCR results using the same primers allowed the identification of imb1 lines homozygous for the transposon insertion (data not shown).

Southern blot analysis of XmnI-digested Landsberg erecta (Ler) and imb1 genomic DNA using a hybridization probe for the Ds insertion revealed, in the imb1 lane, a single band of the expected size (the Ds element is approximately 6.7 kb), which was absent in DNA from Ler (Figure 4b, transposon probe). This indicated that the mutant line contained a single transposon insertion. Hybridization of the same membrane with a probe derived from IMB1 identified the same band (approximately 8.5 kb) observed with the transposon probe in the imb1 lane and, as predicted from the XmnI restriction map, an approximately 1.8-kb band in the Ler lane (Figure 4b, IMB1 probe). This confirmed that the gene was indeed interrupted in the imb1 mutant, and that the lines isolated using the PCR strategy were, in fact, homozygous for the insertion.

Northern blot analysis using an IMB1 probe detected a single band of approximately 1.2 kb in total RNA from Ler seeds imbibed for 72 h at 4°C (Figure 4c). This corresponds approximately to the size of the isolated IMB1 cDNA. No band could be detected in RNA from homozygous imb1 imbibed seeds (Figure 4c), suggesting that imb1 is a true null mutant.

Phenotypic analysis of the imb1 mutant

Adult imb1 knockout plants appeared normal, and their general morphology was indistinguishable from wild-type Ler plants (data not shown). As the IMB1 transcript was detected at considerable levels in roots and flowers (see Figure 3a), particular attention was paid to these organs. Root morphology, flower morphology, and flowering time of mutant and wild-type plants did not show appreciable differences (data not shown). On the other hand, the characteristic seed expression pattern of IMB1 suggested that the gene might be involved in the regulation of seed dormancy and/or germination. We therefore compared several seed germination responses in Ler and imb1 plants.

imb1 seeds are hypersensitive to ABA-mediated inhibition of cotyledon greening

Disruption of the IMB1 gene did not affect seed dormancy or the timing of germination, and neither the cold nor the light requirement for seed germination was affected (data not shown). Likewise, when germination rates were scored under conditions of osmotic stress (imposed by 0.45 m mannitol), no difference was apparent between wild-type and mutant seeds (data not shown). In addition, IMB1 does not appear to control gibberellin-regulated germination responses, as sensitivity to neither GA₃ nor paclobutrazol (an inhibitor of gibberellin biosynthesis) was affected in the knockout mutant (data not shown). By contrast, although no appreciable differences were observed when radicle emergence from the seed coat in different ABA concentrations was scored (Figure 5a), imb1 seeds showed an ABA-hypersensitive response to the expansion and greening of cotyledons (Figure 5b,c). After stratification and 14 days in 1 µm ABA, around 15% of the Ler seedlings exhibited expanded green cotyledons, whereas only approximately 3% of imb1 seedlings had turned green (Figure 5b,c). At lower ABA concentrations (0.5 µm), rates of cotyledon greening were greater than 80% and indistinguishable between the two genotypes, whereas both Ler and imb1 seedlings were incapable of greening in the presence of 1.5 µm ABA (data not shown).

Although imb1 seedlings show ABA hypersensitivity, the IMB1 seed expression pattern was unchanged by ABA; no differences were observed when IMB1 transcript levels of wild-type seeds imbibed and germinated for 0, 1, 2, and 4 days in 5 µm ABA were analyzed by Northern blotting (data not shown).

The ABI5 protein is required for an ABA-dependent developmental arrest, which occurs after radicle emergence and prior to the onset of autotrophic growth (Lopez-Molina et al., 2001). This knowledge prompted us to investigate whether the ABI5 protein would be present at higher levels in imb1 seeds, which are hypersensitive to the ABA inhibition of cotyledon expansion and greening during the germination process (Figure 5b,c). Western blot analysis revealed that indeed ABI5 was up regulated in imb1 seedlings grown in 1 µm ABA; as expected, the protein was not detectable in control seedlings grown in the absence of ABA (Figure 5d). Matching results were obtained by Northern blotting, in which the ABI5 transcript was detected at nearly threefold higher levels in imb1 seeds germinated in ABA when compared to the wild-type counterpart (Figure 5e). Similar analysis of ABI3, which acts upstream of ABI5 to execute the growth arrest of embryos germinating in ABA (Lopez-Molina et al., 2002), revealed no appreciable differences in the levels of the protein between wild-type and mutant seedlings, with the ABI3 transcript being undetectable in Northern blots (data not shown).

imb1 seeds show deficient phyA-mediated induction of germination

We next investigated whether IMB1 would be involved in the photoregulation of seed germination. In Arabidopsis, light-induced germination is controlled primarily by phyA and phyB (Shinomura et al., 1996). Promotion of germination by FR pulses constitutes a VLFR and is mediated by phyA (Botto et al., 1996), which is synthesized during seed imbibition (Sharrock and Clack, 2002). By contrast, phyB is present in freshly imbibed seeds (Sharrock and Clack, 2002) and mediates the R/FR-reversible LFR (Casal et al., 1998). In the present study, seed germination rates in darkness after exposure to pulses of FR and/or R were scored. Seeds were irradiated with FR for 30 min immediately after imbibition to minimize the levels of active phyB and then stratified for 3 days before any light treatment.

Exposure of Ler and imb1 seeds to a pulse of R induced high germination rates (approximately 90–95%), with no significant difference being observed between the two genotypes (Figure 6). By contrast, an FR pulse, which is incapable of triggering germination of the phyA null mutant (Shinomura et al., 1996), led to decreased germination rates. Importantly, imb1 seeds germinated at considerably lower rates (approximately 35%) than the wild type (approximately 65%), indicating reduced phyA function in the imb1 mutant (Figure 6). Irradiation with R followed by FR resulted in germination rates comparable to those observed after FR alone (data not shown), whereas the lowest germination frequencies (approximately 20%) were recorded after no light treatment (Figure 6). In addition, no differences between Ler and imb1 germination rates or hypocotyls lengths were registered after exposure to continuous FR for 3 days (data not shown). In summary, light experiment results demonstrate that the imb1 mutant shows deficient phyA-mediated VLFR of germination.

Complementation of the imb1 phenotypes

To test for complementation, the imb1 mutant was transformed with the full-length wild-type IMB1 cDNA expressed under the control of a 35S promoter. The empty vector was used as a negative control, and three independent transgenic lines were characterized. After an FR pulse, all three transgenic lines showed almost complete (approximately 85%) restoration of Ler germination levels (Figure 7a). Furthermore, unlike the control transgenic line, a fraction of the seedlings (approximately 20%) from the complemented mutant lines was able to green in 1 µm ABA (Figure 7b). The finding that IMB1 was expressed at high levels transiently in imbibed seeds had already pointed to a germination-specific phenotype later observed in the knockout mutant. Taken together, results from complementation analysis show that the loss of IMB1 function is indeed responsible for the imb1 mutant phenotypes.

Microarray analysis

In an attempt to gain clues on the molecular mechanisms by which the putative transcription factor IMB1 might be regulating germination responses, genomic-scale expression-profiling experiments were performed. Using commercially available Arabidopsis Affymetrix gene chips, transcript levels in Ler and imb1 seeds imbibed for 72 h at 4°C in darkness were compared to determine which genes showed the greatest change in expression in the imb1 mutant relative to the wild type. Two independent extractions for each genotype were performed, generating four comparisons. Among approximately 24 000 different Arabidopsis genes represented in the arrays, 29 were found to be repressed at least twofold in the imb1 mutant in all four comparisons (Table 1), whereas applying the same criteria only five genes were found to be induced (Table 2). Down regulation of five of the 29 genes (At2g34900, AJ278354, At4g30270, At2g05790, and At2g37640) was confirmed by Northern blot analysis (data not shown). When less stringent criteria were applied and genes induced or repressed at least twofold in three out of the four comparisons were selected (see Tables S1 and S2), 139 genes were found to be down regulated whereas only six were found to be up regulated in the imb1 mutant. Results from the microarray analysis point to a role of IMB1 in modulating transcript levels of genes with diverse functions.

Among the 29 genes showing the most reduced levels in imb1 versus the wild type (Table 1), approximately 25% encode proteins with unidentified function. Not surprisingly, IMB1 was one of the most down regulated genes, supporting the reliability of the data. A considerable fraction (about 20%) of the genes repressed in the imb1 mutant, such as genes encoding a putative expansin or β-glucanases, is likely to be implicated in cell-wall metabolism. Interestingly, five of the 29 down-regulated genes are encoded by the plastid genome. One encodes a component of photosystem II, whereas all the other four appear to be ndh genes, which encode different subunits of the thylakoid NADH dehydrogenase complex. Consistent with this, a nuclear-encoded putative subunit of a plastid RNA polymerase was also significantly less expressed in imb1. Other genes down regulated in the mutant include two genes involved in lipid metabolism and a putative ribose 5-phosphate isomerase.

Using less stringent criteria, an additional set of 110 genes was found to be repressed in imb1 (see Tables S1 and S2). Of these, 10 are also plastid-encoded genes (mostly components of photosystem II and another subunit of the thylakoid NADH dehydrogenase complex). Furthermore, many other genes involved in cell-wall metabolism were identified, such as two additional putative expansins and one xyloglucan endotransglycosylase, a β-xylosidase, an endo-β-1,4-glucanase, and a pectate lyase.

Of the five genes found to be induced (Table 2), the one showing by far the most increased levels in the mutant encodes an Arabidopsis protein similar to the murine BET protein MCAP (Dey et al., 2000). Moreover, of the remaining four genes selected, two encode proteins putatively involved in the regulation of transposable elements, whereas the function of the other two remains unidentified. When the less stringent criteria were considered, only one additional gene was selected as being up regulated in the mutant (see Tables S1 and S2), indicating that IMB1 acts predominantly as a transcriptional activator.

Plant materials and growth conditions

Seeds from A. thaliana (L.) Heyhn, ecotype Ler, were surface-sterilized, sown in Petri dishes containing Murashige and Skoog (MS) salts (JRH Biosciences, Lenexa, KS, USA), 0.8% agar, and 1% sucrose. Seeds were stratified for 3 days at 4°C in the dark (to break dormancy), before being placed in a growth chamber (22°C, 16-h photoperiod). After 2–3 weeks, plants were transferred to soil in individual pots.

The imb1 Ds-insertion line (derived from the Ler ecotype) was obtained from a previously described Ds-tagging population (Parinov et al., 1999; Sundaresan et al., 1995) and grown as described above. To confirm the exact location of the Ds insertion and determine its orientation, DNA fragments were amplified from genomic DNA (isolated using the Plant DNA Mini Kit; Qiagen, Chatsworth, CA, USA) with primers flanking the Ds insertion, SET-F (5′-AGCAAAGCGTTAGCTGC-3′) and SET-R (5′-AAGAAAGTAGCAGTGCC-3′), and primer DS5-3 (5′-CGGTCGGTACGGGATTTTCC-3′), which anneals at the 5′ end of Ds. The PCR product obtained using SET-F and DS5-3 was then sequenced. The same primers were used to identify homozygous lines for the transposon insertion; genomic DNA from individual plants was used for two PCR reactions – one using SET-F/SET-R and the other SET-F/DS5-3 – and the results were compared. Phenotypic analysis was performed on imb1 homozygous lines backcrossed once with the Ler wild type.

Germination assays

Seed germination is a complex process influenced by multiple factors, such as storage conditions and exposure to light or cold. Germination experiments must therefore be carefully controlled. In this study, Ler and imb1 plants were sown and grown simultaneously under identical conditions. Mature seeds were harvested at the same time from dehydrated siliques and stored in the dark at 4°C. All germination assays were performed with seeds from comparable lots stored for 2–6 months. Batches of 80–160 seeds were sown in triplicate in Petri dishes containing MS salts and 0.8% agar. Average germination percentages were calculated with SEs of the triplicates.

For ABA sensitivity tests, seeds were surface-sterilized and sown on MS salts and 0.8% agar in the presence of the ABA (mixed isomers, A1049; Sigma, St Louis, MO, USA) concentrations indicated in Figure 5. Dishes were incubated in darkness at 4°C prior to transfer to a growth chamber (22°C, 16-h photoperiod). Germination, defined as the obvious protrusion of the radicle through the seed coat, was scored 7 days after transfer to the growth chamber. Greening of expanded cotyledons was scored 14 days after transfer to 22°C.

For light experiments, seeds were surface-sterilized and sown on MS salts and 0.8% agar in a green-safelight chamber and immediately irradiated for 30 min with FR. A 3-day dark treatment at 4°C was followed by exposure to FR or R for 10 min. FR and R were supplied by LED light sources (Quantum Devices, Barnveld, WI, USA) filtered through Plexiglass (WestLake Plastics, Lenni, PA, USA). The fluence rate for both FR and R was 20 µmol photons m⁻² sec⁻¹. After the light treatment, seeds were incubated in darkness for 5 days and germination (defined as protrusion of the radicle from the seed coat) was scored. Germination percentages were calculated from the total number of seeds that germinated after a further 5-day incubation in a growth chamber (22°C, 16-h photoperiod).

Northern, Southern, and Western blot analyses

Total RNA was extracted from different tissues of 35-day-old plants using the RNeasy Plant Mini Kit (Qiagen, Chatsworth, CA, USA). RNA from seeds was extracted according to Vicient and Delseny (1999). For Northern blot analysis, 7-µg samples of total RNA were size-fractionated on a formaldehyde-agarose gel and subsequently transferred to a nylon membrane (Duralon, Stratagene, La Jolla, CA, USA) according to standard procedures (Ausubel et al., 1994). After hybridization in 50% formamide, 5% dextran sulfate, 5× SSC, and 0.5% SDS at 42°C with random prime-labeled fragments (Megaprime DNA Labeling System RPN 1604, Amersham Biosciences, Piscataway, NJ, USA), membranes were washed three times with 0.2× SSC and 0.1% SDS at 55°C. As a probe for IMB1, the DNA fragment obtained by PCR amplification of genomic DNA using primers SET-F and SET-R was used. As a probe for ABI5, the DNA fragment obtained by PCR amplification of genomic DNA using primers 5′-CTAGAGAAACGAAGTTGACG-3′ and 5′-CAGCCTTCACCAAGAAATCC-3′ was used.

For Southern blot analysis, plant genomic DNA was extracted as described by Reiter et al. (1992) and digested with XmnI, before being size-fractionated through an agarose gel and transferred to a nylon membrane according to standard procedures (Ausubel et al., 1994). After hybridization in 5% dextran sulfate, 5× SSC, and 0.5% SDS at 65°C with random prime-labeled fragments, membranes were washed as described for Northern blot analysis. As a probe for the transposon insertion, a PCR fragment amplified from the β-glucuronidase (GUS) gene of the Ds insertion with primers 5′-CCATTTGAAGCCGATGTCCACGCCG-3′ and 5′-CCAGTTGCAACCACCTGTTGATCCGC-3′ was used. The same fragment used in Northern blot analyses was used as an IMB1 probe.

Western blot analysis of ABI5 and ABI3 was performed as described by Lopez-Molina et al. (2001).

Isolation of IMB1 cDNA and complementation analysis

First-strand IMB1 cDNA was obtained from total RNA of flower tissue using the ProStar First-Strand RT-PCR Kit (Stratagene, La Jolla, CA, USA) and primers 5′-CCGCTCGAGATGTCTGTACATGTCAA-3′ and 5′-GGTTAATTAATCAAGCTTTCTTAGCTC3′, which added a XhoI and a PacI restriction site, respectively. The same primers were used for PCR amplification of the first-strand cDNA, performed with Tgo DNA polymerase (Roche, Indianapolis, IN, USA). The resulting DNA fragment was digested with XhoI and PacI, ligated into XhoI–PacI-digested binary vector pBA002 (Kost et al., 1998), and sequenced. Comparison with the hypothetical mRNA sequence (At2g34900) in GenBank revealed that the 8th exon was 42 bp shorter than predicted in the annotation.

The empty pBA002 vector and the complementation construct containing the XhoI–PacI fragment described above were introduced into Agrobacterium strain GV3101 and used to transform homozygous imb1 plants by vacuum infiltration (Clough and Bent, 1998). T1 transformants were selected on BASTA-containing medium, grown to maturity, and selfed. Phenotypic analysis of T2 seeds stored for 30–50 days was performed by scoring germination after a FR pulse and after 28 days in 1 µm ABA as described above.

Subcellular localization of a YFP–IMB1 fusion protein in onion cells

The YFP coding sequence (Miyawaki et al., 1997) was cloned into the MluI and SpeI sites of pBA002 to generate pBA–YFP, and the IMB1 cDNA was fused in-frame to the 5′ end of the YFP sequence. To this end, the IMB1-coding region was amplified using primers 5′-GGTTAATTAAATGTCTGTACATGTCAA-3′ and 5′-TCCCCCGGGAAGCTTTCTTAGCTCTTTT-3′, which added a PacI restriction site and a XmaI site while deleting the stop codon, respectively. The resulting fragment was introduced into pBA-YFP, which alone was used as a control. Onion epidermal cells were bombarded with 1.6-mm-diameter gold particles coated with the constructs using a helium-driven biolistic particle delivery system (PDS-1000/He, BIO-RAD, Hercules, CA, USA). Epidermal cells were kept at room temperature in darkness for 16–20 h, before YFP fluorescence was visualized using an Axioskop microscope (Carl Zeiss, Thornwood, NY, USA).

Microarray experiments

For microarray analysis, total RNA from Ler and imb1 seeds imbibed for 72 h at 4°C in darkness was extracted following the method of Vicient and Delseny (1999) and further purified using the RNeasy Plant Mini Kit (Qiagen, Chatsworth, CA, USA). Two independent extractions for each genotype were performed. From each sample, 5–6 µg of total RNA were used for cDNA synthesis, labeling, hybridization, and scanning of the DNA chips carried out at the Gene Array Resource Center of The Rockefeller University according to the manufacturer's instructions (Affymetrix, Santa Clara, USA). Processing of expression levels (raw data) and comparative analysis between experiments were performed using Affymetrix MicroarraySuite 5.0 and Microsoft Excel software. Genes that were up or down regulated at least twofold in relation to the wild type in all four comparisons between the chip experiments (two for imb1, two for Ler) were selected as being misregulated. The average of the four values of fold change was then calculated. In addition, genes induced or repressed at least twofold in the imb1 mutant in three out of the four comparisons (provided that the gene was also found to be up- or down regulated, respectively, in the fourth comparison) are presented as Tables S1 and S2. This microarray analysis conforms to the MIAME recommendations.

Upon request, all biological and chemical materials not commercially available described in this publication will be provided for non-profit research purposes.