Expression of rice SUB1A and SUB1C transcription factors in Arabidopsis uncovers flowering inhibition as a submergence tolerance mechanism


  • Julián M. Peña-Castro,

    1. Center for Plant Cell Biology and Department of Botany and Plant Sciences, University of California, Riverside, CA 92521, USA
    Search for more papers by this author
    • Present address: Instituto de Biotecnología, Universidad del Papaloapan, Tuxtepec, Oaxaca 68301, México.

  • Martijn van Zanten,

    1. Plant Ecophysiology, Institute of Environmental Biology, Utrecht University, Padualaan 8, 3584 CH Utrecht, the Netherlands
    Search for more papers by this author
    • Present address: Max Planck Institute for Plant Breeding Research, Carl von Linné Weg 10, 50829, Cologne, Germany.

  • Seung C. Lee,

    1. Center for Plant Cell Biology and Department of Botany and Plant Sciences, University of California, Riverside, CA 92521, USA
    Search for more papers by this author
  • Mohini R. Patel,

    1. Center for Plant Cell Biology and Department of Botany and Plant Sciences, University of California, Riverside, CA 92521, USA
    Search for more papers by this author
  • Laurentius A. J. C. Voesenek,

    1. Plant Ecophysiology, Institute of Environmental Biology, Utrecht University, Padualaan 8, 3584 CH Utrecht, the Netherlands
    Search for more papers by this author
  • Takeshi Fukao,

    1. Center for Plant Cell Biology and Department of Botany and Plant Sciences, University of California, Riverside, CA 92521, USA
    Search for more papers by this author
  • Julia Bailey-Serres

    Corresponding author
    1. Center for Plant Cell Biology and Department of Botany and Plant Sciences, University of California, Riverside, CA 92521, USA
    Search for more papers by this author

(fax (951) 827-4437; e-mail


Submergence of plant organs perturbs homeostasis by limiting diffusion of oxygen, carbon dioxide and ethylene. In rice (Oryza sativa L.), the haplotype at the multigenic SUBMERGENCE1 (SUB1) locus determines whether plants survive prolonged submergence. SUB1 encodes two or three transcription factors of the group VII ethylene response factor family: SUB1A, SUB1B and SUB1C. The presence of SUB1A-1 and its strong submergence-triggered ethylene-mediated induction confers submergence tolerance through a quiescence survival strategy that inhibits gibberellin (GA)-induced carbohydrate consumption and elongation growth. SUB1C is invariably present and acts downstream of the enhancement of GA responsiveness during submergence. In this study, heterologous ectopic expression of rice SUB1A and SUB1C in Arabidopsis thaliana was used to explore conserved mechanisms of action associated with these genes using developmental, physiological and molecular metrics. As in rice transgenic plants that ectopically express SUB1A-1, Arabidopsis transgenic plants that constitutively express SUB1A displayed GA insensitivity and abscisic acid hypersensitivity. Ectopic SUB1C expression had more limited effects on development, stress responses and the transcriptome. Observation of a delayed flowering phenotype in lines over-expressing SUB1A led to the finding that inhibition of floral initiation is a component of the quiescence survival strategy in rice. Together, these analyses demonstrate conserved as well as specific roles for group VII ethylene response factors in integration of abiotic responses with development.


Crops and wild plants are frequently devastated by partial or complete submergence because of the decreased availability of oxygen and carbon dioxide, frequently accompanied by increases in ethylene. Within minutes to hours, waterlogging of roots, and, in some cases, submergence of aerial organs, leads to a switch from aerobic to anaerobic metabolism, and, in days to weeks, to complete depletion of carbohydrate reserves resulting in death (Bailey-Serres and Voesenek, 2008, 2010). It is well established that plant cells deprived of oxygen limit protein synthesis to conserve energy, and show enhanced production of enzymes associated with sucrose catabolism, glycolysis and fermentation (Bailey-Serres and Voesenek, 2008). With the advent of genomic techniques, the transcripts that are induced and selectively translated under low-oxygen stress have been widely described for Arabidopsis thaliana (reviewed by Bailey-Serres and Voesenek, 2010). The cadre of mRNAs translated under oxygen deprivation extends from metabolic enzymes to transcription factors, signaling pathway components, generators and regulators of reactive oxygen species, cell-wall modification enzymes and heat shock proteins (Liu et al., 2005; Loreti et al., 2005; Branco-Price et al., 2008; Mustroph et al., 2009). There is a strong overlap between low oxygen- and submergence-induced mRNAs, and some strongly up-regulated genes encoding proteins of unknown function modulate both low-oxygen and submergence tolerance (Mustroph et al., 2010; Lee et al., 2011).

Understanding of the physiological and developmental response mechanisms associated with low-oxygen stress and submergence has benefited from cross-species studies including Arabidopsis thaliana, Rumex palustris and Oryza sativa (Bailey-Serres and Voesenek, 2008, 2010; Mustroph et al., 2010). Two acclimation responses have been identified: the low-oxygen escape strategy (LOES) and the low-oxygen quiescence strategy (LOQS; Bailey-Serres and Voesenek, 2008). LOES involves a suite of traits including a high rate of carbohydrate consumption, allowing rapid elongation of aerial organs to keep the leaves above the water level. In deepwater rice, whose internodal regions rapidly elongate when partially submerged, LOES is mediated by at least three hormones (Bailey-Serres and Voesenek, 2008), and by SNORKEL1 and SNORKEL2 (SK1/SK2), which are tandem ethylene responsive factor (ERF) genes (Hattori et al., 2009). On the other hand, quiescence involves stress-induced repression of carbohydrate consumption, concomitant with inhibition of plant growth. This tolerance strategy is reversed once the stress is relieved. In lowland rice, LOQS is determined by the SUB1A gene of SUBMERGENCE1 (SUB1), a multigenic locus encoding two (SUB1B and SUB1C) or three (SUB1A, SUB1B and SUB1C) ERFs (Fukao et al., 2006; Xu et al., 2006). Interestingly, despite their antithetical action, the SK and SUB1A genes are members of the same subclade of ERFs, designated group VII by Nakano et al. (2006).

Ethylene-mediated induction of SUB1A initiates LOQS. SUB1C is also induced by ethylene, but accumulation of its mRNA during submergence is most likely driven by increased responsiveness to GA, which follows the ethylene-promoted reduction in abscisic acid (ABA; Fukao and Bailey-Serres, 2008). When SUB1A-1 is present in the SUB1 haplotype, there is a less rapid consumption of carbohydrate reserves for anaerobic metabolism in aerial tissue (Fukao et al., 2006) and increased accumulation of SLENDER RICE 1 (SLR1) and SLENDER RICE 1-LIKE 1 (SLRL1) proteins, which inhibit elongation growth by repressing GA signaling (Fukao and Bailey-Serres, 2008). Because the abundance of SUB1C mRNA is up-regulated by GA and down-regulated by SUB1A (Xu et al., 2006), it was hypothesized that SUB1C is important for GA-mediated carbohydrate catabolism and elongation growth (Fukao et al., 2006).

The rice (O. sativa ssp. japonica cv. Nipponbare) and Arabidopsis (Columbia-0) genomes encode 15 and five group VII ERFs, respectively (Nakano et al., 2006). In Arabidopsis, low-oxygen stress and submergence up-regulate the mRNAs of two group VII ERFs [HYPOXIA RESPONSIVE ERF1 (HRE1)/AtERF73 (At1g72360) and HRE2/AtERF71 (At2g47520)] (Licausi et al., 2010; Lee et al., 2011). In seedlings, over-expression of HRE1 constitutively elevates anaerobic response mRNAs, including ALCOHOL DEHYDROGENASE1 (ADH1), whereas the hre1 hre2 double mutant shows limited induction of these mRNAs in response to low-oxygen stress (Licausi et al., 2010). Other group VII ERFs also contribute to regulation of ADH1 in Arabidopsis, including the dark- and ethylene-inducible genes RAP2.2 (Hinz et al., 2010) and RAP2.12 (Papdi et al., 2008). Despite its role in restriction of carbohydrate catabolism, SUB1A positively regulates ADH1 and other anaerobic response mRNAs in rice during submergence (Fukao et al., 2006; Jung et al., 2010; Mustroph et al., 2010).

To examine evolutionary conservation in the role of group VII ERFs and to obtain insight into the function of individual SUB1s, we heterologously over-expressed rice SUB1A and SUB1C in Arabidopsis. Although ectopic expression of these genes did not improve submergence survival, expression of SUB1A inhibited GA responsiveness and stimulated ABA responsiveness, as observed in rice (Fukao and Bailey-Serres, 2008; Fukao et al., 2011). Moreover, we found that expression of SUB1A strongly delayed flowering of both Arabidopsis and rice, concomitant with reductions in mRNAs associated with flowering induction. Our results suggest conservation of pathways regulated by group VII ERFs across plant species.


OxSUB1A and OxSUB1C plants display pleiotropic phenotypes

Transgenic Arabidopsis (Col-0) plants were produced that ectopically express N-terminally FLAG-tagged SUB1A or SUB1C under the control of the CaMV 35S promoter (OxSUB1A and OxSUB1C). OxSUB1A homozygous plants had compact, broad and round rosette leaves with shortened petioles compared with Col-0 (Figure 1a,b and Figure S1a). The rosettes of OxSUB1C homozygotes had less dramatic abnormalities that included shortened petioles and downward curled leaves. These gross morphological changes were accompanied by alterations in the size and shape of abaxial epidermal cells of the leaf and stem (Figure S1b).

Figure 1.

 Pleiotropic phenotypes of Arabidopsis plants expressing rice SUB1 genes.
Rosette leaf, petiole and anther phenotypes of Col-0, OxSUB1A and OxSUB1C plants grown under a long-day photoperiod.
(a) OxSUB1A-L5 develops short and rounded leaves and OxSUB1C-L10 develops short and curled leaves (30-day-old plants). The black vertical bar represents 5 cm.
(b) Petiole lengths of 23-day-old plants (n = 12). Values are means ± SD. Letters indicate significant differences between genotypes at the same leaf stage (< 0.05, Student’s t test).
(c–h) Anther filament and dehiscence phenotypes. (c) Col-0 flower. (d) OxSUB1A-L5 flower with short filaments. (e) OxSUB1A-L12 flower with short filaments and showing inhibition of anther dehiscence. (f) OxSUB1C-L10 flower showing inhibition of anther dehiscence. (g, h) Flowers from Col-0 plants treated with 1 or 5 μm PAC, respectively. Front sepals were removed to visualize anthers and stamen. The black vertical bar represents 2 mm. (i) Accumulation of SUB1A and SUB1C protein in 12-day-old seedlings of over-expression lines. Seedlings were immersed in 100 μm MG132 (+) or mock-treated (−) for 4 h before harvesting. FLAG-tagged SUB1 proteins were immunoprecipitated, fractionated by SDS–PAGE, and detected using anti-FLAG monoclonal antiserum. The apparent molecular mass is indicated.

Because of the severe reduction in seed production in most of the individual OxSUB1A and OxSUB1C transgenic plants, only partially fertile lines could be established. The line OxSUB1A-L5 only produced seed after the first 20 flowers, and OxSUB1A-L12 yielded seed in lateral inflorescences late in reproductive development. Two SUB1C over-expressing lines, OxSUB1C-L6 and OxSUB1C-L10, produced some fertile siliques from early and late flowers (16 ± 7%). Examination of floral organs revealed that OxSUB1A lines developed anther filaments that were too short to self-pollinate the stigma and failed to dehisce (Figure 1c–e), whereas OxSUB1C anthers failed to dehisce (Figure 1f). Shortened anther filaments and reduced dehiscence were promoted in Col-0 by treatment with paclobutrazol (PAC), an inhibitor of GA biosynthesis (Figure 1g,h). However, neither the anther nor fertility defects of OxSUB1A and OxSUB1C flowers were reversed by GA application during bolting (Figure S2).

The proteosome regulates SUB1A and SUB1C abundance in Arabidopsis

To assess whether the differences in OxSUB1A and OxSUB1C phenotypes correlated with differences in accumulation of FLAG-tagged SUB1 protein among individual transformants, we performed immunoprecipitation followed by an immunoblot analysis with seedling extracts (Figure 1i). Pre-treatment of seedlings with the 26S proteosome inhibitor MG132 significantly enhanced levels of SUB1A and SUB1C. Consistent with our hypothesis that differences in phenotypes correlate with levels in protein accumulation, SUB1A abundance was lowest in OxSUB1A-L5, which was the most fertile of the OxSUB1A lines (Figure S2b). The levels of FLAG-tagged SUB1C levels were lower in OxSUB1C-L6 than OxSUB1C-L10 seedlings, although both lines produced few seed. These results demonstrate that accumulation of both SUB1s is regulated by proteosome-dependent proteolysis in Arabidopsis. Similar studies indicated that SUB1A levels are proteosome-regulated in rice (Takeshi Fukao, unpublished results).

Over-expression of SUB1A in Arabidopsis reduces survival of submergence and prolonged darkness

Strong submergence induction or constitutive over-expression of SUB1A-1 confers tolerance to submergence for 14 days or longer in rice (Xu et al., 2006; Fukao et al., 2006; Singh et al., 2009). Recently, Vashisht et al. (2011) determined the submergence survival amongst 86 Arabidopsis accessions and identified significant natural variation. To compare the submergence tolerance of OxSUB1A, OxSUB1C and Col-0, we used identical treatment and analysis methods as Vashisht et al. (2011) to evaluate the median lethal time (LT50) of submergence (Figure 2a). We found that the LT50 in OxSUB1C plants was almost identical (7.00–7.06 days) to that of Col-0, whereas that of both OxSUB1A transgenic lines was 4.8 ± 0.2 days (Figure 2b). In marked contrast to the submergence tolerance conferred by SUB1A-1 in rice, ectopic expression of this gene in Arabidopsis decreased submergence survival. The highest and lowest LT50 values for the Arabidopsis accessions evaluated were 11 and 4 days, respectively (Vashisht et al., 2011), indicating that OxSUB1A transgenic plants were extremely sensitive to submergence. Because the submergence treatment was performed in darkness to increase severity, we also monitored the response of these genotypes to prolonged darkness under non-flooded conditions (air). This revealed that OxSUB1A is hypersensitive to darkness, as indicated by the rapid yellowing of rosette leaves in the dark (Figure 2a). Detached OxSUB1A leaves also yellowed more rapidly than the other two genotypes (Figure S3). We also calculated a hazard ratio that integrates the submergence and dark LT50 data (Vashisht et al., 2011). OxSUB1A lines had a hazard ratio ≤1, indicating that their survival was greater under submergence than in darkness in air (Figure 2c). Together, these data indicate that ectopic expression of rice SUB1A-1 in Arabidopsis neither enhances submergence tolerance nor survival in darkness, most likely because of accelerated senescence.

Figure 2.

 Submergence survival of OxSUB1A and OxSUB1C genotypes.
(a) Col-0 and two OxSUB1A and two OxSUB1C transgenic lines were grown to the ten-leaf stage and submerged under darkness for up to 10 days. Control plants were kept in air under darkness. Photographs were taken immediately after de-submergence or release from darkness.
(b) Median lethal time (LT50) ±SE for each genotype, obtained from the number of survivors after 9 days of recovery for each duration of submergence (n = 10 plants) (Vashisht et al., 2011).
(c) Hazard ratio, determined from the survival in submergence and darkness as compared to the survival in darkness alone, after 9 days of recovery. Genotypes with a hazard ratio ≤1 are better able to survive darkness than submergence.

OxSUB1A and OxSUB1C seeds are hypersensitive to ABA and PAC

Over-expression of SUB1A-1 in rice decreases GA responsiveness and increases ABA sensitivity during seed germination (Fukao and Bailey-Serres, 2008; Fukao et al., 2011). To evaluate whether the response to these hormones is similarly altered in OxSUB1A and OxSUB1C transgenic plants of Arabidopsis, germination tests were performed. We found that relative to Col-0, germination of OxSUB1A and OxSUB1C seeds was delayed 2 days, and was hypersensitive to inhibition by PAC and ABA (Figure 3a–c). Germination of seeds in the presence of PAC with various doses of GA confirmed that both OxSUB1A and OxSUB1C plants have significantly reduced responsiveness to GA during germination (Figure 3d). The observed altered sensitivity of these transgenic seeds to ABA and GA is similar to the effect of ectopic SUB1A expression in rice (Fukao and Bailey-Serres, 2008; Fukao et al., 2011).

Figure 3.

 Seed germination phenotypes of OxSUB1A and OxSUB1C genotypes.
(a) Time course of germination of stratified seeds on MS agar containing 1% w/v sucrose grown under a long-day photoperiod.
(b, c) Percentage of germinated seeds after 4 days in the presence of various concentrations of ABA or PAC. Values are means ± SD.
(d) Percentage of germinated seeds after 4 days in the presence of various concentrations of GA and 10 μm PAC. Values are means ± SD.
Asterisks indicate a significant difference from Col-0 (< 0.05, Student’s t test). Each data point represents three independent experiments (n = 30). The experiment was performed twice with similar results.

OxSUB1A and OxSUB1C transgenic plants display reduced petiole hyponasty in response to environmental stimuli

To further examine the effects of SUB1 expression on GA-regulated processes, we compared hyponastic growth responses of OxSUB1A, OxSUB1C and Col-0 leaves. First, leaf positioning was measured. Rosette leaves 7–10 of Col-0 had an initial leaf angle of approximately 18° relative to horizontal, whereas the maximum leaf angle observed for developmentally matched leaves of OxSUB1A transgenic plants was approximately 5° (Figure 4a). OxSUB1C rosette leaves also showed a significantly reduced petiole angle (>60%) relative to Col-0. Application of GA led to hyponastic petiole growth that increased the leaf angle in Col-0 to 33° and restored the leaf angles of OxSUB1A and OxSUB1C to values statistically indistinguishable from GA-treated Col-0 plants (Figure 4a).

Figure 4.

 Leaf angles and hyponastic growth kinetics of Col-0, OxSUB1A and OxSUB1C plants.
(a) Absolute petiole angle of leaves 7–10, relative to the horizontal, of approximately 25-day-old plants grown under a short-day photoperiod (> 10; mean ± SE). Letters indicate significant differences (< 0.05, anova).
(b, c) Ethylene treatment (5 μl L−1).
(d, e) Ethylene treatment combined with pre-treatment using 50 μm GA.
(f, g) Low-light treatment (reduction from 200 to 15 μE m−2 sec−1 photosynthetically active radiation).
(h, i) Low-light treatment combined with pre-treatment using GA.
(j, k) High-temperature treatment (38°C).
(l, m) High-temperature treatment combined with pre-treatment using GA.
Transgenic line data are indicated by circles and Col-0 data are indicated by a line. Data in (b–m) are means ± SE (> 10). The angles shown were pairwise-subtracted and normalized to 0° (Benschop et al., 2007). Figure S2 shows similar results for two additional transgenic lines.

Next, we evaluated the dynamics of hyponastic petiole growth in these genotypes promoted by gaseous ethylene, low light intensity or high temperature (38°C) (Millenaar et al., 2009; van Zanten et al., 2009, 2010). OxSUB1A lines had reduced responses to all three treatments relative to Col-0 (Figure 4b,f,j and Figure S4), unless pre-treated with GA (Figure 4d,h,l and Figure S4). The dynamics of petiole hyponasty in OxSUB1C lines was similar to that in OxSUB1A transgenic plants, other than the OxSUB1C-L6 response to heat, which was similar to that of Col-0 (Figure S4). GA pre-treatment restored the response of OxSUB1C transgenic plants to ethylene and low light, but not to heat (Figure 4e,i,m and Figure S4). Together, these data indicate that SUB1A and SUB1C over-expression limits hyponastic leaf movement in response to environmental cues in a manner consistent with reduced GA sensitivity.

Altered flowering phenotypes of OxSUB1A and OxSUB1C lines are GA-regulated

The pleiotropic morphological alterations of the SUB1 over-expression lines included altered inflorescence stem architecture and flowering time. OxSUB1A plants had shorter inflorescence stems than Col-0 (Figure 5a) and elongated more slowly (0.5 cm per day for OxSUB1A versus 2.6 cm per day for Col-0). GA application partially reversed the reduction in inflorescence stem length in OxSUB1A plants (Figure 5a). Although OxSUB1A inflorescence stems were shorter, they developed more flowers/siliques than Col-0 (55 ± 10 versus 40 ± 5, respectively), with shorter internodes, and the first fruit positioned closer to the stem base (Figure 5b). By contrast, OxSUB1C inflorescence lengths were similar to those of Col-0, but these transgenic plants developed more flowers/siliques (71 ± 7) than either Col-0 or OxSUB1A transgenic plants. Strikingly, OxSUB1A transgenic plants showed a pronounced delay in flowering (Figure 5c). These lines had twice the number of rosette leaves at bolting when grown under a short-day (SD) photoperiod (Figure 5e), and retained a late-flowering phenotype even under a flowering-inductive long-day (LD) photoperiod (Figure 5f). However, the number of leaves at flowering time was restored to that of Col-0 under SD or LD conditions by exogenous application of GA. By contrast, OxSUB1C transgenic plants showed normal flowering time behavior under both photoperiods (Figure 5d–f).

Figure 5.

 Flowering stem and flowering time phenotypes of Col-0, OxSUB1A-L5 and OxSUB1C-L10 plants.
(a) Developmental time course of inflorescence stem elongation with (open symbols) and without GA application (closed symbols) until flower production has finished under long-day conditions (n = 14–18; means ± SD).
(b) Graphical representation of main inflorescence stem architecture: distance from the rosette base of each silique and internode length under long-day conditions (n = 5).
(c, d) Representative 66-day-old plants grown under short-day conditions. (c) Representative OxSUB1A-L12 plant showing aberrant flowering phenotype. (d) Representative OxSUB1C-L10 plant showing normal flowering time. The black vertical bars represent 5 cm.
(e, f) Number of rosette leaves at flowering time of plants grown under short-day (e) or long-day (f) conditions with mock or 100 μm GA application every 3 days starting at 7 days old. Two independent lines were evaluated for each transgene construct. Data are expressed as the number of leaves when floral buds were first visible (n = 18; means ± SD). Letters indicate significant differences (< 0.05, Student’s t test).

SUB1A represses flowering induction genes in Arabidopsis and rice

To further investigate the delay in flowering observed under floral inductive LD conditions in OxSUB1A transgenic plants, we quantified the temporal accumulation of key transcripts associated with flowering time regulation. As flowering-inducing CONSTANS (CO) and florigen protein FLOWERING LOCUS T (FT) mRNAs are induced 6–10 days after germination under LD conditions in Arabidopsis (Yoo et al., 2005), experiments were performed on 7-day-old seedlings. These analyses confirmed that the levels of CO and FT mRNAs were significantly lower in OxSUB1A transgenic plants compared to Col-0 (Figure 6a,b). It is likely that SUB1A similarly influences floral induction in rice, because rice transgenic plants that constitutively over-express SUB1A-1 (UBI:SUB1A) flower 1 month later than the non-transgenic control cultivar Liaogeng (LG, SUB1 haplotype SUB1B-2, SUB1C-2) (Fukao and Bailey-Serres, 2008). To address this possibility, the aforementioned rice lines and the near-isogenic lines M202(Sub1) (haplotype SUB1A-1, SUB1B-1, SUB1C-1) and M202 (haplotype SUB1B-2, SUB1C-2) were examined for accumulation of transcripts encoding HEADING DATE 1 (Hd1) and HEADING DATE 3a (Hd3a), which are rice orthologs of CO and FT, respectively, under a flowering-inductive SD photoperiod as well as during submergence. These experiments were performed on 33-day-old plants, as this is within the period that Hd3a mRNA levels reportedly peaked in japonica cv. Nipponbare plants grown under SD conditions (Kojima et al., 2002). As our previous studies on submergence responses of these genotypes were performed on 14-day-old plants (Fukao and Bailey-Serres, 2008), we also evaluated the effect of submergence on SUB1A transcript accumulation at the 33-day-old developmental stage (Figure 6c). These experiments revealed a peak of SUB1A mRNA at ZT3 (Zeitgeber Time, ZT0 equals start of daylight), followed by a decline during the day and a second transient increase at approximately ZT14 (midnight). We found that under SD and non-submerged conditions, the abundance of Hd1 and Hd3a mRNAs peaked during the night, similar to CO and FT in Arabidopsis (Figure 6d,e). Submergence further increased Hd1 mRNA abundance during the night, but this increase was lower in M202(Sub1) plants. Conversely, the level of Hd3a transcripts decreased by 50% in submerged M202 plants, and these were completely repressed in M202(Sub1) plants over the 24 h period.

Figure 6.

 SUB1A regulates the flowering time control loci CONSTANS (CO) and FLOWERING LOCUS T (FT) in Arabidopsis, and their orthologs HEADING DATE 1 (Hd1) and HEADING DATE 3a (Hd3a) in rice.
(a, b) Transcript accumulation in 7-day-old Arabidopsis seedlings grown under long-day conditions in air. (a) CO mRNA. (b) FT mRNA.
(c–e) Gene expression under short-day conditions in developmentally matched 33-day-old M202 and M202(Sub1) rice plants under normal growth in air or after 2 days of complete submergence. (c) SUB1A mRNA. (d) Hd1 mRNA. (e) Hd3a mRNA.
(f, g) Transcript accumulation under short-day conditions in 47-day-old rice plants over-expressing SUB1A-1 (UBI:SUB1A-3) and 33-day-old LG plants (developmentally matched non-transgenic control). (f) Hd1 mRNA. (g) Hd3a mRNA.
Transcript abundance was determined by quantitative RT-PCR and normalized to the highest abundance in the control genotype (means ± SE). The experiments were performed twice, with independent replicate samples showing similar results. Asterisks and letters represent a significant difference between genotypes and treatments (< 0.05, Student’s test).

To address the disparity between CO/Hd1 mRNA oscillations in Arabidopsis and rice, Hd1 mRNA expression was followed over 24 h in UBI:SUB1A-1 transgenic plants and developmentally matched non-transgenic LG plants (47 and 33 days old, respectively). As for CO and FT mRNAs in OxSUB1A transgenic plants, over-expression of SUB1A in rice significantly dampened Hd1 and Hd3a mRNA accumulation during the night (Figure 6f,g). Moreover, Hd3a mRNAs were undetectable in UBI:SUB1A plants. Most likely, the differences in regulation of Hd1 and Hd3a mRNAs in the two japonica cultivars LG and M202 reflect genotypic differences (Figure 6d–g). These data indicate that ectopic or submergence-induced expression of SUB1A affects the abundance of mRNAs encoding two conserved regulators of floral induction in rice.

Ectopic SUB1A and SUB1C expression alters the seedling transcriptome

To assess the transcriptional network mobilized by heterologous constitutive expression of SUB1A and SUB1C in Arabidopsis, Affymetrix ATH1 arrays were hybridized with RNA isolated from 7-day-old seedlings grown under LD conditions, yielding highly reproducible datasets (r2 ≥ 0.99) (Table S1a). We found that 182 mRNAs were up-regulated in OxSUB1A seedlings, and 97 mRNAs down-regulated (Signal Log2 Ratio (SLR) ≥ |1|, adjusted P value ≤ 0.05; Table S1b), whereas 35 mRNAs were up-regulated in OxSUB1C seedlings and 11 mRNAs were down-regulated. A gene ontology (GO) analysis revealed that the SUB1A up-regulated genes were associated with responses to biotic stress, cell wall-associated, lipid storage and ABA responsiveness (adjusted P values 9.10E-15, 3.29E-6, 3.01E-5 and 1.60E-3, respectively; Table S1c). Differential expression of genes known to control GA biosynthesis, GA signaling or flowering was not evident, probably because the transcriptome captured differences that were large or reflective of many cell types. Fourteen mRNAs were up-regulated in both OxSUB1A and OxSUB1C plants, representing almost half of the mRNAs elevated in OxSUB1C seedlings. These included low oxygen-induced ADH1 (At1g77120), genes associated with pathogen infection (BGL2, At3g57260; PCC1, At3g22231; DEFENSIN-LIKE, At3g59930), lipid storage (OLE1, At4g25140; OLE2 At5g40420) and protein storage (CRA1, At5g44120). Other transcripts associated with protein and lipid storage were also up-regulated in both OxSUB1A and OxSUB1C plants, although the overlap was limited between the two genotypes. In the case of down-regulated genes, no overlap was found between the transgenic plants. Uncharacterized transcription factors and kinases, as well as proteins of unknown function, accounted for 30% of the up-regulated mRNAs in OxSUB1A.

To validate the microarray analysis, we confirmed by quantitative RT-PCR that both OLE1 and OLE2 mRNAs were up-regulated in 7-day-old seedlings of two independent OxSUB1A and OxSUB1C transgenic plants (Figure S5a,b). Oleosins form the outer surface of oil bodies, restricting coalescence of small oil bodies, and can influence cellular triacylglycerol lipid content (Shimada et al., 2008). To assess the effect of enhanced OLE1 and OLE2 mRNA levels on oil body size, seedlings were stained with Nile Red and examined by confocal microscopy. Although lipid bodies were evident in Col-0 hypocotyls (Figure S5c,d), they were not detectable in OxSUB1A or OxSUB1C hypocotyls (Figure S5e–h). A PROTEOLYSIS6 mutant (prt6-1) that retains large oil bodies in hypocotyls (Holman et al., 2009) was used as a control (Figure S5i,j). These data reveal that ectopic expression of SUB1A and SUB1C affects the storage or mobilization of lipid reserves in seedlings.

We also compared the differentially expressed genes in OxSUB1A and OxSUB1C to those of seedlings over-expressing the Arabidopsis group VII ERFs HRE1 or HRE2 (Licausi et al., 2010) and germinating seeds of the ABA-hypersensitive germination mutants agh1-1 and agh3-1, both are protein phosphatase 2C mutants (Nishimura et al., 2007; Table S1d). ADH1 was the only gene that was differentially regulated in OxSUB1A, OxSUB1C and 35S::HRE1 transgenic seedlings. Intriguingly, ADH1, which encodes a key anaerobic protein, was significantly down-regulated in 35S::HRE2 seedlings. These data indicate that over-expression of individual rice and Arabidopsis group VII ERFs in Arabidopsis distinctly changes the transcriptome in seedlings grown under aerobic conditions. Comparison with ABA-hypersensitive mutants identified 40 differentially expressed genes in OxSUB1A whose levels were also increased in the agh1-1 mutant, and 18 genes that were down-regulated in both genotypes. A very similar overlap in differentially expressed genes was observed between OxSUB1A and agh3-1. This strongly supports the conclusion that the OxSUB1A seedling transcriptome is reconfigured in a manner consistent with ABA hypersensitivity.


This evaluation of heterologous ectopic expression in Arabidopsis of the group VII ERFs SUB1A and SUB1C from rice demonstrated that the broad action of SUB1A partially overlaps with that of SUB1C. Although ectopic expression of SUB1A in Arabidopsis reduced rather than enhanced tolerance to submergence in complete darkness (Figure 2), we found commonalities in developmental processes and environmental responses regulated by SUB1A in rice and Arabidopsis.

SUB1A and SUB1C over-expression alters GA signaling during development in Arabidopsis and rice

Gibberellin promotes seed germination, cell elongation, hyponastic growth and the transition to reproductive development (Djakovic-Petrovic et al., 2007; Achard and Genschik, 2009). In Arabidopsis, this is achieved via DELLA transcription factors, which are responsible for regulation of transcription of GA inactivation enzymes, positive ABA signaling components and GA homeostasis genes (Zentella et al., 2007). Binding of GA to its receptor triggers proteosome-mediated turnover of the DELLA transcription factors, thereby promoting GA-mediated responses. GA responsiveness is enhanced in rice during submergence due to degradation of ABA in a SUB1A-independent manner (Fukao and Bailey-Serres, 2008). In genotypes that possess SUB1A-1, the submergence-induced decrease in ABA is accompanied by limited turnover of the DELLA ortholog SLR1, as well as hyper-accumulation of SLRL1 (Fukao and Bailey-Serres, 2008), which lacks a DELLA domain (Itoh et al., 2005). Consistent with the reduction of GA responsiveness by SUB1A in rice, the phenotypes of Arabidopsis OxSUB1A and OxSUB1C transgenic plants suggest defects in GA-mediated processes. Both displayed limitations in cell, petiole and stem elongation (Figures 1b and 5, and Figure S1a), inhibition of anther dehiscence (Figure 1c), reduced hyponastic growth of petioles (Figure 4), delayed germination (Figure 3) and late flowering (Figure 5). These GA-associated defects were generally less pronounced in OxSUB1C transgenic plants. Additionally, we found that exogenous GA application partially rescued germination, hyponastic growth and floral initiation, but not the inhibition of anther dehiscence. Importantly, the finding that OxSUB1A and OxSUB1C seeds were hypersensitive to PAC and ABA and less sensitive to GA during germination (Figure 3) is consistent with the function of SUB1A in rice (Fukao and Bailey-Serres, 2008; Fukao et al., 2011).

We previously demonstrated that SUB1A acts as a negative feedback inhibitor of ethylene signaling during submergence in rice (Fukao et al., 2006; Fukao and Bailey-Serres, 2008). The limited petiole hyponasty response of OxSUB1A and OxSUB1C transgenic plants indicates that disruption of petiole elongation growth operates when these genes are expressed in Arabidopsis. OxSUB1A and OxSUB1C plants had a dampened hyponastic growth response to ethylene (Figure 4b,c and Figure S4). The increase in the hyponastic angle in response to low light and high temperature was also affected (Figure 4f,g,j,k and Figure S4), although the response to heat was distinct in the two OxSUB1C transgenic plants. All three stimuli (low-light, high temperature and ethylene) rely on GA signaling/biosynthesis to effect the change in leaf angle in Rumex palustris (Cox et al., 2004; van Zanten et al., 2010). Here, application of exogenous GA largely restored the hyponastic angle response, providing further evidence of altered GA responsiveness due to SUB1 gene over-expression in Arabidopsis. Additionally, these results highlight the key role of GA in the control of both leaf positioning and hyponastic leaf movement induced by ethylene, heat and low light in Arabidopsis.

SUB1A inhibits flowering in Arabidopsis and rice

Over-expression of SUB1A in Arabidopsis led to the finding that floral induction is regulated during submergence in rice. We observed a late-flowering phenotype of OxSUB1A plants, even under a flowering-inductive LD photoperiod. This was correlated with down-regulation of transcripts encoding two key positive regulators of flowering, CO and FT (Figure 6a,b). Evaluation of the dynamic diurnal accumulation of the orthologous transcripts in rice revealed that submergence increased the accumulation of Hd1 mRNA during the night and continuously suppressed levels of Hd3a mRNA. There were differences in accumulation of both transcripts at night in M202(Sub1) compared to M202. We reported previously that UBI:SUB1A rice plants showed significantly delayed flowering (Fukao and Bailey-Serres, 2008), and show here that ectopic SUB1A expression limits accumulation of both Hd1 and Hd3a mRNAs, closely resembling the effect of SUB1A over-expression on CO and FT mRNA in Arabidopsis. Taken together, these data provide strong evidence that ectopic expression of SUB1A in rice and Arabidopsis limits the expression of key genes associated with floral induction.

Transition to the reproductive stage and flowering are energetically costly (Obeso, 2002; Early et al., 2009). In the light of the transcript accumulation data presented here, it can be hypothesized that repression of flowering is a means to limit the energetic output. Thus, the SUB1A-mediated repression of flowering could be considered part of the LOQS, adding another component to this suite of traits. Consistent with this, Singh et al. (2009) reported that flowering in 30-day-old SUB1A-1-containing (tolerant) plants that were fully submerged for 12 and 17 days in the field was delayed by 17 and 27 days, respectively. The parental lines lacking SUB1A-1 (intolerant) that survived inundation for 12 and 17 days showed a considerably longer delay in flowering (24 and 37 days, respectively). Thus, SUB1A is responsible for a hiatus in floral induction that only slightly exceeds the duration of the actual flooding event (Singh et al., 2009). We propose that transient repression of floral induction during submergence in SUB1A-1 lines is followed by effective recovery of the ability to induce flowering upon de-submergence.

SUB1A mRNA levels oscillate in submergence-stressed rice

As Arabidopsis and rice lines expressing SUB1A displayed the ability to alter expression of the circadian cycle-regulated CO and FT mRNAs, the possible oscillatory accumulation of SUB1A mRNA was explored in rice. An oscillation approximately every 12 h was found under submerged and non-submerged conditions (Figure 6c). Such an ultradian cycle is uncommon in plants (Baskin, 2007), but could reflect SUB1A regulation in response to different stimuli. Ethylene produced under circadian cycle control could drive transcript accumulation for the midday peak of SUB1A, as shown for sorghum and two Poa species (Finlayson et al., 1998; Fiorani et al., 2002), as ethylene positively regulates SUB1A mRNA level (Fukao et al., 2006). The increase in SUB1A mRNA at the midnight peak could be due to the limited levels of carbohydrates at this point in the circadian cycle (Graf et al., 2010). Interestingly, petiole elongation and expansin mRNA accumulation driven by ethylene in submerged Rumex palustris are primarily a daytime response (Vreeburg et al., 2005). We intend to examine whether the decrease in SUB1A mRNA after the daytime peak is a result of negative regulation of ethylene evolution by SUB1A, and whether the decrease during the night reflects SUB1A-mediated management of energy homeostasis.

SUB1A and SUB1C promote overlapping alterations in the seedling transcriptome in Arabidopsis

Ethylene responsive factors recognize a GCCGCC core motif in target promoters, and are known to modulate developmentally and environmentally regulated processes (Ohta et al., 2001; Nakano et al., 2006). Several studies have used heterologous ERF expression to enhance stress tolerance in model and crop species (Karaba et al., 2007; Quan et al., 2010; Zhang et al., 2010), but tolerance was often accompanied by developmental abnormalities. Here, we found that the effect of heterologous SUB1A and SUB1C expression on steady-state mRNA accumulation in seedlings of Arabidopsis partially overlapped (Table S1d) with more extensive transcriptomic adjustments in OxSUB1A seedlings accompanied by more severe pleiotropic phenotypes. Of the differentially expressed genes in both OsSUB1A and OxSUB1C seedlings, only ADH1 was also up-regulated in Arabidopsis seedlings constitutively expressing the related group VII ERF gene HRE1. Several factors could contribute to the transcriptomic adjustments characteristic of individual group VII ERFs. For example, ERF accumulation may be distinctly regulated in a conditional or developmental-specific manner, interactions may occur between the ERF and specific co-factors, or individual ERFs may bind distinct DNA binding motifs.

We observed extensive changes in the seedling transcriptome in OxSUB1A plants. These included a number of mRNAs associated with biotic stress responses in Arabidopsis, possibly reflecting the role of several ERFs as transducers of ethylene signaling in response to pathogens (Chakravarthy et al., 2003; Lorenzo et al., 2003). Consistently, submergence up-regulated transcripts associated with pathogen responses in M202(Sub1) rice (Jung et al., 2010). OxSUB1A transgenic plants also showed increased levels of transcripts associated with lipid and protein reserves (Figure S5), suggesting that nutrient reserve storage or management is modulated by heterologous SUB1A expression. Despite phenotypes indicative of altered GA responsiveness, the seedling transcriptomes did not reveal differences associated with known GA biosynthesis or signaling components. Quantitative RT-PCR measurement of mRNAs involved in GA biosynthesis and perception (GA20Ox1, At4g25420; GA20Ox2, At5g51810; GA2Ox1, At1g78440; GID1, At3g05120) revealed no differences between genotypes (data not shown). As genes modulating GA biosynthesis and response are highly redundant, and their mRNAs are regulated in a spatial and temporal manner (Rieu et al., 2008), the whole-seedling transcriptome analysis may not have captured the subtle regulation of these or other genes responsible for the GA-associated phenotypes. However, OxSUB1A seedlings had higher levels of a number of mRNAs that are increased in the ABA-hypersensitive mutants ahg1-1 and ahg3-1 (Table S1d), indicating a conserved effect of SUB1A on ABA responsiveness in Arabidopsis, as in rice (Fukao et al., 2011).

Although the heterologous near-constitutive expression of the rice group VII ERF transcription factors SUB1A and SUB1C in Arabidopsis did not improve submergence survival, it enabled the discovery that flowering inhibition is an integral part of the quiescence response strategy mediated by submergence-induced expression of SUB1A-1 in rice. Interestingly, we observed a significant overlap in the pleiotropic effects of SUB1A and SUB1C over-expression on development, leaf positioning, environmentally regulated leaf movement, responses to GA and ABA, and the seedling in Arabidopsis. However, their function in rice is distinct (Fukao et al., 2006; Fukao and Bailey-Serres, 2008), presumably due to differences in regulation of expression, interaction with specific co-factors, and/or distinctions in target genes. Altogether, the results of this study demonstrate that group VII ERFs act in evolutionarily conserved networks that integrate abiotic stress responses with development.

Experimental procedures

Plant genetic material

Arabidopsis thaliana genotypes included Col-0 ecotype and prt6-1 (Holman et al., 2009). Oryza sativa L. ssp. japonica genotypes included M202 and M202(Sub1), which are near-isogenic lines differing at the SUB1 locus (Fukao et al., 2006), the homozygous transgenic line UBI:SUB1A-3 cv. Liaogeng (LG), which over-expresses SUB1A-1 (Xu et al., 2006; Fukao and Bailey-Serres, 2008), and LG as the control. SUB1A-1 and SUB1C-1 coding regions were amplified from M202(Sub1) mRNA (Fukao et al., 2006), cloned into the Gateway-compatible vector p35S:HF-GATA, which includes the CaMV 35S promoter and an N-terminal His6-FLAG tag, and transformed into Col-0, as described in Appendix S1. T4 homozygotes were further characterized.

Plant growth conditions

Unless stated otherwise, for growth on soil, Arabidopsis seeds were sterilized and stratified in water for 3 days at 4°C, germinated on 1× Murashige & Skoog (MS) agar medium (0.43% w/v MS salts, pH 5.75) additionally containing 1% w/v sucrose, in vertically oriented plates in a growth chamber (23°C, 16 h light/8 h dark, 125 μE m−2 sec−1 photosynthetically active radiation), and after 7 days of growth transferred to autoclaved soil (Metro Mix 200, 2% w/w Osmocote, 1% w/w Marathon; SunGro, in a growth chamber (21 ± 1°C) under short-day (SD) conditions (9 h light/15 h dark, 175 μE m−2 sec−1) or long-day (LD) conditions (16 h light/8 h dark; 175 μE m−2 sec−1). Rice was grown in soil as described by Fukao et al. (2006) but in a growth chamber under SD conditions (10 h light, 28°C/14 h dark, 24°C, 500 μE m−2 sec). Genotypes were arranged to minimize experimental noise due to position in the growth chamber.

Hormone treatment, stress conditions and phenotypic characterization

Flowering time and GA treatments.  Leaves (n = 18) were sprayed twice with GA solution (100 μm in 0.5% v/v ethanol) every 3 days after transferring the plants to soil when they were 7 days old. Mock-treated plants were sprayed with 0.5% v/v ethanol. Leaf number was counted when the floral bud was visible (flowering time). Plants or siliques were scored as fertile when one or more seeds were produced per silique.

Petiole, stem and inflorescence internode length.  Plants were grown in soil under LD conditions until 23 days old (n = 12). Petiole length was measured from the base to the petiole/lamina junction. Stem elongation was measured from the rosette base to the floral meristem from the time the first bud appeared until flower production stopped (n = 14–18). Vegetative stem length was measured as the distance between the rosette base and the first flower. Inflorescence internode length was measured as the distance between siliques. The latter two phenotypes were measured when flower production stopped (n = 5).

Seed germination.  Seeds were harvested and dried at the same time for all genotypes. Seeds (n = 30 for each treatment) were grown on MS agar with 1% w/v sucrose as described, above except that ABA, GA or PAC (Sigma, in 0.1% v/v solvent were added to the warm medium after autoclaving. Mock-treated plates contained 0.1% v/v ethanol (for comparison with ABA and GA) or 0.1% v/v methanol (for comparison with PAC). Germination was recorded as the time of protrusion of the radical.

PAC treatment on soil.  Col-0 plants grown under SD conditions were watered by soil immersion every 5 days using a PAC solution at designated concentrations.

Leaf angle and hyponastic growth measurements.  Plants were grown in a soil/perlite mixture (2:1) in a growth chamber at 20°C, 70% relative humidity, 9 h SD photoperiod (200 μE m−2 sec−1; Millenaar et al., 2005). Spectral neutral low light (15 μE m−2 sec−1 photosynthetically active radiation) and ethylene treatments (5 μl L−1) were performed as described by Millenaar et al. (2009), and heat treatment (38°C) was performed as described by van Zanten et al. (2009). For GA treatment, plants were sprayed with 50 μm GA at 17, 31 and 55 h before measurements. Leaf angles and hyponastic growth were measured on plants at approximately stage 3.9 (Boyes et al., 2001) using two petioles per plant (length 0.7–1 cm, n = 10–32). Measurements were performed using an automated time-lapse camera as described by Millenaar et al. (2005, 2009) and in Appendix S1.

Submergence stress and survival measurements

Submergence stress treatments and determination of the median lethal time (LT50) and hazard ratio were performed as described previously (Vashisht et al., 2011) and in Appendix S1.

Submergence treatment and diurnal cycle sampling for rice

Rice plants were grown from seed in soil as described above, for 33 days, except for the genotype UBI:SUB1A-3 which was grown for 47 days to developmentally match the other plants. At this time, LG and UBI:SUB1A-3 shoot tissue was harvested at ZT3, and then every 3 h for 24 h. M202 and M202(Sub1) were fully submerged in tanks as described by Fukao et al. (2006) for 2 days, and then shoot tissue was harvested at ZT3 on day 3, and then every 3 h for 24 h. Tissue from non-stressed control plants was collected at the same time points.

Transcriptome analyses

Seven-day-old Arabidopsis seedlings were grown on 1× MS agar medium as described above, and harvested at midday. RNA extraction was performed as described by Mustroph et al. (2010). Hybridization to ATH1 GeneChips was performed according to the manufacturer’s method (Affymetrix,, with experimentally independent duplicate samples. Data were analyzed using Robin (RMA model; Lohse et al., 2010). Differentially regulated genes (SLR ≥ |1|, adjusted P value < 0.05) were classified into functional groups using GO (Horan et al., 2008; Mustroph et al., 2010). CEL files of published data were obtained from the Gene Expression Omnibus [GES17099 (Licausi et al., 2010) and GSE6638 (Nishimura et al., 2007)] ( and processed in the same manner. The microarray data generated in this study are accessible from the Gene Expression Omnibus database under accession number GSE27669.

Protein immunopurification and detection

Thirteen-day-old seedlings (3 g) grown on solid MS as described above were harvested at midday and immersed in 12 ml of a solution containing 100 μm of MG132 (Peptides International, in 1% v/v DMSO. After 4 h of rocking, seedlings were washed three times for 1 minute. Immunoprecipitation and immunodetection were performed as described previously (Zanetti et al., 2005).

Measurement of transcript abundance by quantitative RT-PCR

For rice and Arabidopsis tissue, RNA was extracted as previously described (Mustroph et al., 2010), and cDNA was synthesized from 2.5 μg total RNA using SuperScript II reverse transcriptase (Invitrogen, according to the manufacturer’s protocol. Quantitative PCR was performed using 1 μl cDNA, 10 μmol each primer and 10 μl of IQ SYBR Green Supermix (Bio-Rad, using a IQ5 real-time PCR machine. Primers used to detect transcripts of CO (At5g15840), FT (At1g65480), OLE1 (At4g25140), OLE2 (At5g40420), TUB2 (At1g65480), Hd1 (LOC_Os06g16370.1), HD3a (LOC_Os06g06320.1) and UBQ (LOC_Os03g13170) are provided in Appendix S1. Data were analyzed using the comparative Ct method (Schmittgen and Livak, 2008).


We thank A. Mustroph, H. Yang, B. Walter, C.J.H. Jang, G. Hicks, D. Carter, M. Perales, A. Rosado, C.Y. Huang, B.E. Barrera (University of California, Riverside) and A.J.M. Peeters (Utrecht University) for valuable technical assistance and discussions. This work was supported by postdoctoral fellowships to J.M.P.-C. (University of California Institute for Mexico and the United States and Consejo Nacional de Ciencia y Tecnología de México), and US National Science Foundation grants (IBN-0420152 and IOS-0750811) and a National Institute of Food and Agriculture grant (2008-35100-04528) to J.B.-S.