Influence of drought on the heading date is weaker in rcn1 mutants than in WT plants
To test if RCN1 loss of function affects rice flowering time, we obtained two different insert mutations in the coding region of RCN1 using the clustered regularly interspaced short palindromic repeats-associated protein-9 (CRISPR-Cas9) technology (Figure S1), the frame-shift mutations caused premature termination of translation or amino acid substitutions. In both mutant lines RCN1 was not functional. The experiment was carried out during the growing season. Rice plants were subject to severe drought stress, as the soil relative water capacity (SRWC) was maintained between 30%–50% (Figure S2). In rcn1 mutants, heading occurred approximately 7 days earlier than that in the WT plants under normal water conditions (Table 1), indicating a negative role of RCN1 in controlling flowering transition. Both WT and rcn1 mutant plants showed a significantly late-heading phenotype under severe drought conditions (Table 1). It is noteworthy that the heading date of the WT plants was delayed for approximately 14 days, and the two rcn1 mutant plants showed delays of 9.2 and 10.9 days, respectively. The results indicated that RCN1 loss of function reduced the effect of drought on rice heading date (Table 1).
The expression of RCN1 in the rcn1 mutant plants was not significantly different from that of the wild-type plants under normal watering conditions (Figure S3a). In contrast, in rcn1 mutant plants, RCN1 showed a significantly decreased responsiveness to drought. The rice florigens Hd3a and RFT1 promote flowering. Under normal watering conditions, there was no obvious difference in Hd3a/RFT1 expression levels between rcn1 mutants and wild-type plants. Under drought stress conditions, both the wild-type and the rcn1 mutant plants showed a significant down-regulation of Hd3a and RFT1. However, Hd3a and RFT1 were not regulated by RCN1; they were down-regulated under drought conditions in an RCN1-independent manner (Figure S3b, c). Drought stress can delay flowering by inhibiting florigen expression, and there was no effect of RCN1 on florigens transcription.
Tissue-specific expression and localization of RCN1
To examine the spatio–temporal expression of RCN1, a quantitative reverse transcription- PCR (qRT-PCR) analysis was performed. The expression of RCN1 was detected in the leaves, roots, and stems throughout the growth period, and the highest expression was observed in the roots (Fig. 1a). To further examine the expression pattern of RCN1, the 5′ distal region (4 kb) of the RCN1 locus was fused with the beta-glucuronidase (GUS) reporter and used to conduct histochemical staining of the GUS gene. The expression of GUS was detected in the vascular tissues of the leaves and roots (Fig. 1b, c). In the cross-section of stems, a high GUS expression was detected in the nodes and basal internodes (Fig. 1d, e), whereas no expression was detected in young panicles (Fig. 1f). These results were consistent with the qRT-PCR results. Furthermore, GUS expression was observed in the glumes, and this has never been reported (Fig. 1g).
To examine the localization of the RCN1 protein in the plant, we constructed an expression vector with GREEN FLUORESCENT PROTEIN (GFP) as the reporter gene, by fusing the RCN1-coding region with the GFP-coding region (RCN1-GFP), and used the same promoter as that in the GUS analysis. The distribution of RCN1-GFP was consistent with the results of the qRT-PCR and GUS staining, suggesting the localization of RCN1 to the vasculature of the leaves and roots (Fig. 1h, i). We also observed the subcellular localization of RCN1-GFP in lateral root cells, where it was mainly distributed in the cell nucleus, cytoplasm, and cellular membrane (Fig. 1j); the same was observed in rice protoplasts (Figure S4).
Drought stress activates RCN1 transcription in an ABA-dependent manner
For the osmotic stress treatment, plants were transferred to pots containing three different concentrations of polyethylene glycol (PEG6000), while those grown without PEG6000 were used as the controls. The expression of RCN1 was detected by qRT-PCR. The results indicated that 30% PEG6000 strongly increased the expression of RCN1 in the shoots and roots (Fig. 2a, b), and therefore this concentration was used in the subsequent experiments. It is widely known that ABA is accumulated in response to drought; thus, experiments were designed to verify whether RCN1 would respond to ABA. We added 50 μM ABA to the rice culture solution and detected the expression level of RCN1 at 0, 3, 6, 24, 30, 48, 54, 96, and 102 h of incubation, by qRT-PCR. Overall, ABA strongly increased the expression of RCN1 in the shoots and roots, particularly in the latter (Fig. 2c, d). Because the upregulation of RCN1 in the shoots was only detected 24 h post-ABA treatment, samples collected at this time point were used to evaluate the effects of ABA treatment. To evaluate the relationship between ABA and RCN1, we further examined the effect of different concentrations of exogenous ABA on RCN1 expression. The results showed that RCN1 expression was proportional to the exogenous ABA concentration applied (Fig. 2e, f). Compared with the control, the expression of RCN1 in the shoots increased by up to 10-fold (Fig. 2e), far below the 39-fold observed in the roots (Fig. 2f). To identify if PEG6000 induced RCN1 expression in an ABA-dependent manner, we treated rice plants with 40 μΜ fluridone for 30 min, and then moved them to 30% PEG6000 solution with fluridone. Fluridone, an inhibitor of carotenoids, which are the main precursors of ABA in plants (Raikhel et al. 1986; Kowalczyk-Schroder and Sandmann 1992), also inhibits the biosynthesis of ABA. Fluridone treatment was effective inhibit the increase in ABA content (Hsu et al. 2006; KAO et al. 2003; Perales et al. 2005; Shi et al. 2012).The results revealed that, after blocking ABA synthesis, RCN1 did not respond to PEG6000 (Fig. 2g, h), suggesting that the PEG6000-mediated induction of RCN1 was dependent on ABA. The ABA-mediated induction of RCN1 was completely inhibited by cycloheximide (CHX), which is an inhibitor of protein synthesis (Fig. 2i, j). This indicated that protein synthesis is necessary to ABA regulate RCN1 expression, one reason for the requirement may have been that transcription factors play an important role in regulation. PEG6000- and ABA-induced RCN1 expression was partially restrained by K252a, which is a phosphorylation blocker (Fig. 2g, h, k, i). These results revealed an ABA-mediated induction of RCN1 that was dependent on phosphorylation, and provided further evidence for ABA-mediated induction of RCN1 under drought.
ABA induces RCN1 through transcription factors
The basic leucine-zipper (bZIP) transcription factors are bound to the ABA-responsive cis-element ABRE in the promoters of ABA-inducible genes to regulate plant stress responses. The conserved sequence of ABRE is ACGT, and the C/G/A nucleotides flanking the ACGT core enhance bZIP protein binding specificity and affinity (Hattori et al. 2002). We screened 26 ABA response transcription factors from authoritative databases and selected five transcription factors that accumulated in response to ABA faster than RCN1 (Figure S5a). A new ABA treatment experiment was then carried out to determine the expression levels of these transcription factors. Finally, OsAREB1, OSBZ8, and TRAB1 showed a rapid and strong response to ABA in both leaves and roots (Figure S5b, c). Because the induction of OsAREB1, OSBZ8, and TRAB1 expression by ABA was not blocked by CHX (Figure S5d, e), they appear to be primary ABA response genes.
Electrophoretic mobility shift assays (EMSA) were used to examine the binding activities of the OsAREB1, OSBZ8, and TRAB1 proteins with the ABRE element in the RCN1 promoter. The 14 sequence specific fragments around the RCN1 locus contained 19 ACGT elements that were synthesized and labeled with biotin (Fig. 3a). OsAREB1, OSBZ8, and TRAB1 were expressed and extracted from the BL21 competent E. coli strain. Among the six different fragments tested, OsAREB1 and OSBZ8 exhibited the strongest interaction with Probe 13 (containing three ACGT elements), a slightly weaker interaction with Probes 9, 10, and 12 (each containing two ACGT elements), and the weakest interactions with Probes 11 and 14 (each containing one ACGT element) (Fig. 3b, c). To examine the specificity of these interactions, a 50- or 200-fold unlabeled probe was applied to compete with the labeled probe. The interactions of OsAREB1 and OSBZ8 and labeled probes were inhibited by excess unlabeled probes. No interaction was detected with mutant probes that lacked the ABRE element sequence (Fig. 3b, c). There were no interactions between OsAREB1 and OSBZ8 and Probes 1–8 (Figure S6a, b), and between the TRAB1 protein and any of the 14 probes (Figure S6c). Therefore, OsAREB1 and OSBZ8 seemed to bind only to the ABRE element, which is located in the 3’ distal region of the RCN1 locus, and their binding ability is likely be related to the number of ABRE elements included in the DNA fragments.
The EMSA experiments revealed that OsAREB1 and OSBZ8 can bind to the 3′ distal region of the RCN1 locus in vitro. To determine whether they can regulate RCN1 expression in vivo, glucocorticoid-inducible transgenic lines carrying plasmid Gos2::OsAREB1 or Gos2::OSBZ8 in Nipponbare background were generated. The expression of OsAREB1 and OSBZ8 in the Gos2::OsAREB1 and Gos2::OSBZ8 transgenic plant leaves, respectively, was induced by a low concentration of dexamethasone (DEX) (Fig. 4a, b). As expected, the expression of RCN1 was upregulated with the increase in OsAREB1 and OSBZ8 expression (Fig. 4c, d). Taken together, these findings suggest that OsAREB1 and OSBZ8 function as transcriptional regulators that modulate the drought response of RCN1 via an ABA-dependent pathway.
To further examine whether the 5′ proximal region of the RCN1 locus response to ABA, we treated pRCN1::RCN1-GFP:NOS and pRCN1::GUS:NOS transgenic plants with ABA, and detected the expression levels of GUS and GFP and the endogenous expression level of RCN1 (Fig. 5). The expression of GFP and GUS was not induced by ABA, whereas the endogenous expression of RCN1 was significantly induced (Fig. 4b, c). The results of the qRT-PCR indicated that the ABA response element was not located on the 5′ proximal region of RCN1, but may be the ABRE elements on 3' distal region of RCN1.