OsSFR6 is a functional rice orthologue of SENSITIVE TO FREEZING-6 and can act as a regulator of COR gene expression, osmotic stress and freezing tolerance in Arabidopsis


Author for correspondence:
Heather Knight
Tel: +44 (0)191 334 3215
Email: p.h.knight@durham.ac.uk


  • The Arabidopsis protein SENSITIVE TO FREEZING-6 (AtSFR6) is required for cold- and drought-inducible expression of COLD-ON REGULATED (COR) genes and, as a consequence, AtSFR6 is essential for osmotic stress and freezing tolerance in Arabidopsis. Therefore, orthologues of AtSFR6 in crop species represent important candidate targets for future manipulation of stress tolerance. We identified and cloned a homologue of AtSFR6 from rice (Oryza sativa), OsSFR6, and confirmed its orthology in Arabidopsis.
  • OsSFR6 was identified by homology searches, and a full-length coding region isolated using reverse transcription polymerase chain reaction (RT-PCR) from Oryza sativa cDNA. To test for orthology, OsSFR6 was expressed in an Arabidopsis sfr6 loss-of-function mutant background, and restoration of wild-type phenotypes was assessed.
  • Searching the rice genome revealed a single homologue of AtSFR6. Cloning and sequencing the OsSFR6 coding region showed OsSFR6 to have 61.7% identity and 71.1% similarity to AtSFR6 at the predicted protein sequence level. Expression of OsSFR6 in the atsfr6 mutant background restored the wild-type visible phenotype, as well as restoring wild-type levels of COR gene expression and tolerance of osmotic and freezing stresses.
  • OsSFR6 is an orthologue of AtSFR6, and thus a target for future manipulation to improve tolerance to osmotic and other abiotic stresses.


Freezing of plants in the field can cause significant damage, a major part of which is due to cellular dehydration as a result of water loss from the cell protoplast when extracellular ice forms (Levitt, 1960; Thomashow, 1999). It is perhaps not surprising, therefore, that of the numerous genes whose expression increases in response to low temperature, many are also inducible by drought (Hughes & Dunn, 1996; Thomashow, 1999). In Arabidopsis, the COLD ON-REGULATED (COR) genes represent a major cold-inducible gene regulon (Fowler & Thomashow, 2002); their expression is activated via the C-repeat (CRT) promoter motif or drought-inducible element (DRE) (Yamaguchi-Shinozaki & Shinozaki, 1994). Two distinct families of transcription factors activate COR gene expression via the CRT/DRE in Arabidopsis; the C-box binding factors (CBFs) 1–3 (Gilmour et al., 1998), also known as DRE-binding proteins 1A–C (DREB1A–C; Shinwari et al., 1998), in response to cold and DREB2A and 2B in response to drought (Liu et al., 1998). A further, less closely related member of the CBF family, CBF4, is also involved in drought-, but not cold-inducible COR gene expression (Haake et al., 2002). Overexpression of active forms of both families of transcription factor in Arabidopsis leads to tolerance of both drought and frost (Jaglo-Ottosen et al., 1998; Liu et al., 1998; Sakuma et al., 2006).

The CRT/DRE motif is utilized in the control of gene expression in response to cold and drought in several crop species, including rice (Dubouzet et al., 2003; Ito et al., 2006). Overexpression of CBF/DREB1 transcription factors, both native and heterologous, has been shown to induce native crop COR genes, and lead to osmotic stress tolerance in these species (Jaglo et al., 2001; Dubouzet et al., 2003; Gao et al., 2009). Interestingly, CBF transcription factors have been identified in chilling-sensitive species such as tomato, which are not able to achieve freezing tolerance (Jaglo et al., 2001; Hsieh et al., 2002a,b; Zhang et al., 2004). In these cases it appears that CBF transcription factors and the CRT/DRE motif are involved in inducing genes required for both drought and chilling tolerance (Jaglo et al., 2001; Hsieh et al., 2002a,b; Zhang et al., 2004). Manipulating the expression and function of these transcription factors, therefore, has led to the possibility of engineering altered tolerance not only to desiccation stresses, such as freezing and drought, but also to chilling.

We have recently described the cloning of SENSITIVE TO FREEZING-6 (AtSFR6); a protein that regulates CBF/DREB-dependent COR gene expression in Arabidopsis (Knight et al., 2009). Our previous work has shown that AtSFR6 is needed for induction of COR genes in response to both cold and osmotic stresses and that it is required for tolerance to osmotic stress and the acquisition of freezing tolerance (Knight et al., 1999, 2009; Boyce et al., 2003). In the case of cold at least, SFR6 acts post-translationally of the transcription factors that activate COR genes via the CRT/DRE motif (Knight et al., 2009). Orthologues of AtSFR6 in crop species are therefore obvious candidate targets for manipulation of osmotic stress tolerance. The first step towards such a long-term goal is to demonstrate that functional orthologues of AtSFR6 exist in crop plants. Here we describe the identification of a homologue of AtSFR6 in rice, its cloning and sequencing, and demonstrate orthology through genetic complementation.

Materials and Methods

Plant materials and growth conditions

Arabidopsis thaliana (L.) Heynh. ecotype Columbia (Col-0) was obtained from Lehle Seeds (Round Rock, TX, USA). The Arabidopsis mutant, sfr6-1, also in Col-0 background, has been described previously (Knight et al., 1999, 2009; Boyce et al., 2003). Rice (Oryza sativa L.) seedlings of cv Japonica var. Lemont (Herbiseed, Twyford, UK) were used for extraction of mRNA for cloning OsSFR6. Plants were grown in a SANYO MLR351 growth chamber (Sanyo E&E Europe BV, Loughborough, UK) under a 16 : 8 h, light : dark cycle at 150 μmol m−2 s−1 at 20 ± 1°C unless stated otherwise. The cold treatments used in gene expression experiments were carried out in the same growth chambers set to 4°C. Osmotic stress-induced gene expression was measured in plants floated on 350 mM mannitol solutions in transparent plastic cell culture dishes in the same growth chambers set to 20°C. All samples were harvested after 6 h of treatment.

Cloning OsSFR6 and production of overexpression construct

Total RNA was extracted from rice leaf tissue using RNeasy Plant Total RNA Kit (Qiagen, Crawley, UK), following the manufacturer’s instructions. Total plant RNA (5 μg) was annealed to 0.5 μg oligo-dT primer (Fermentas, York, UK) and reverse-transcribed at 42°C for 60 min using 200 units of H minus M-MuLV Reverse Transcriptase (Fermentas) according to the manufacturer’s instructions. The full-length OsSFR6 coding sequence (3510 bp) was PCR-amplified from the cDNA produced using the following primers: 5′-CCGGTACCCCCGGGGATGCGCGTGCCCGAGCTCTGCAGGAACTT-3′ (forward) and 5′-GGGCGGGGGCGGCCGATCCCGTCAAATTCAAACGACTTTCAC-3′ (reverse). Amplification was performed with Phusion DNA polymerase (Finnzymes, Keilaranta, Finland) according to the manufacturer’s instructions. The OsSFR6 coding sequence was cloned into the pENTR1A gateway entry vector (Invitrogen) using the Kpn1 and Not1 sites and sequenced. The full-length OsSFR6 coding sequence was transferred by LR recombination from pENTR1A into the pB7WG2 gateway binary destination vector (Karimi et al., 2002), which contains the cauliflower mosaic virus (CaMV) 35S promoter upstream. For comparison, the full-length AtSFR6 genomic coding sequence (Knight et al., 2009) was cloned into the same binary vector.

Plant transformation

Binary vectors containing 35S::AtSFR6 and 35S::OsSFR6 were introduced into Agrobacterium tumefaciens C58C1 and transformed into Col-0 and sfr6-1 mutant using the floral dip method (Clough & Bent, 1998). Primary T1 transformants were identified by glufosinate ammonium (Basta; 250 mg l−1) selection (Bayer Crop Science, Cambridge, UK) on soil. Subsequent analyses were performed on the T2 generation.

Quantitative real-time PCR

A high capacity cDNA reverse transcription kit (Applied Biosystems, Foster City, CA, USA) was used to reverse-transcribe cDNA from 1.5 μg total RNA extracted using the Qiagen RNeasy plant mini kit in conjunction with RNAse-free DNase (Qiagen) to remove any genomic DNA contamination. Quantitative real-time PCR (qRT-PCR) was performed on 10 μl of 1 : 50 diluted cDNA reaction in a 25 μl reaction using an Applied Biosystems 7300 system. Relative transcript abundances were measured using gene-specific TaqMan probes from Applied Biosystems for AtSFR6 (At4g04920; probe identifier At02209654_g1), KIN2 (At5g15970; At02354775_s1) and LTI78 (At5g52310; At02320470_g1), and expression levels were normalized to the expression of PEX4 (At5g25760; At02304594_g1), an endogenous control gene. A custom-made TaqMan probe was prepared for OsSFR6 by Applied Biosystems to the following specifications: forward primer, CGGTGGTGACTAAGTGGTTGTC; reverse primer, GTACTAGAGTTTGCAGGAAGCCAT; FAM™-labelled probe, CTATACCGGAGAAATTC. Reactions were performed in an optical 96-well plate (Applied Biosystems) with three technical replicates for each sample. In all cases, relative quantitation (RQ) was performed by the ΔΔCT (comparative CT) method (Livak & Schmittgen, 2001) and RQ values and estimates of statistical variation (SV) for each sample were calculated as described previously (Knight et al., 2009). The algorithm used is described in Relative Quantitation (RQ) algorithms, Applied Biosystems Real-Time PCR Systems Software, July 2007.

Freezing assays

To test complementation with AtSFR6 7-d-old seedlings (grown as described earlier) were transferred to peat plugs and maintained for 5 wk in a growth chamber (Arctic plant growth chamber A3655, Weiss Gallenkamp Ltd, Loughborough, UK) programmed for short-day conditions (8 : 16 h, light : dark cycle), 20 ± 0.5°C, 60% relative humidity and 150 μmol m−2 s−1 light intensity. Experiments to test complementation with OsSFR6 were performed on plants grown under comparable conditions using a SANYO MLR351 chamber. Cold acclimation in both cases was performed under the same day length and light intensities at 4°C for 11 d. The temperature was subsequently reduced to below freezing (−6.5, −7.5 and −8.5°C) for 24 h, then returned to ambient values. The temperature increases and decreases were achieved by ramping over 3 h.

Osmotic stress tolerance

Osmotic stress tolerance was assessed in seedlings as we have described previously (Knight et al., 1998; Boyce et al., 2003). Eight-d-old seedlings grown under the conditions described earlier were floated on 2 ml of water, with 330, 440 or 550 mM mannitol (BDH, Poole, UK), in a transparent 24-well culture plate. Five seedlings were added to each well. The plate was sealed with micropore tape and returned to the growth chamber for 72 h before photographing.

Sensitivity of germination to osmotic stress was assessed as described previously (Boyce et al., 2003). Seeds were sown on solid Murashige and Skoog (MS) medium containing different concentrations of osmoticum (0, 200, 300 or 400 mM mannitol) and 0.8% agar at pH 5.8. Seeds of each line to be tested were sown at a density of c. 30–80 seeds per 55-cm-diameter Petri dish, with six replicate Petri dishes for each line/treatment. Seeds were stratified on the agar plates at 4°C for 4 d and transferred to standard growth chamber conditions for 7 d. Germination was scored on the basis of radicle emergence.

Statistical inference

For each osmoticum treatment and plant line pairing, we estimated the probability of seed germination using maximum likelihood. To account correctly for potential unknown variation among plates (e.g. subtle variations in the dryness of agar) and differing numbers of seeds per plate, we assumed that the variation in our data between plates for each treatment could be well described by a beta-binomial distribution. The log-likelihood equation we maximized when estimating each probability and details regarding fitting can be found in Richards (2008). Uncertainty in these probabilities was estimated using the profile-likelihood approach (Venzon & Moolgavkar, 1988).

To explain any potential patterns in our wild type Col-0 and sfr6-1 data (i.e. variation in germination among osmoticum treatments), we proposed that the relationship between the probability of seed germination (p) and the osmoticum concentration (x) could be described by Eqn 1:

image(Eqn 1)

This relationship is a modified form of the commonly adopted logistic equation; however, the x-axis has also been scaled by the positive parameter β1. The parameter α1 describes the strength at which osmoticum concentration affects germination success; here a negative value indicates that germination declines with increasing osmoticum strength. We proposed that germination success for sfr6-1 was potentially affected by osmoticum concentration in a similar manner, but also allowed germination to be affected by the abundance of OsSFR6 transcript in each complemented line. In this case, if the OsSFR6 transcript abundance was y, then germination success of the mutant was predicted to be:

image(Eqn 2)

For this model, a positive value of β2 indicates that an increase in OsSFR6 transcript abundance increased germination success. Model parameters (i.e. the αi and the βi) were fitted to the data using maximum likelihood, and we again assumed that our data were consistent with a beta-binomial distribution. In all cases, when we checked our fits we found that our residuals were as expected.

To look for evidence that sfr6-1 and Col-0 wild-type differed in their response to elevated concentrations of osmoticum, we performed a likelihood ratio test (LRT; Sokal & Rohlf, 1995). In this case, the null model assumed that model parameters β0, β1 and α1 were identical for both lines; whereas, the alternative model assumed that the β0, β1 and α1 had to be estimated separately for each line. Evidence that OsSFR6 transcript abundance affected the complemented mutant’s response to osmoticum was also investigated using a LRT. Specifically, the null model assumed that transcript abundance did not affect germination success (by setting β2 = 0 and fitting β0, β1 and α1); whereas, the alternative model also allowed β2 and α2 to vary). Finally, we used a LRT to look for evidence that the highest OsSFR6-expressing line differed in its response to osmoticum with respect to the wild-type. This last test was identical to the first mentioned LRT test, except that we replaced the noncomplemented mutant with the highest OsSFR6-expressing line (= 6.85).


We have recently cloned the AtSFR6 gene (At4g04920) from Arabidopsis (Knight et al., 2009). This gene controls freezing and osmotic stress tolerance in Arabidopsis (Knight et al., 1999, 2009; Boyce et al., 2003). We sought, therefore, to identify orthologues of AtSFR6 from crop species, as potential targets for future manipulation of crop stress tolerance. Using homology searches, we found a single gene in the rice genome (Os10g35560) that showed strong homology to AtSFR6. We named this gene OsSFR6. Having identified the gene, we cloned and sequenced the full-length coding region from cDNA derived from rice mRNA. Fig. 1 shows a line-up of the predicted protein sequence of OsSFR6 with AtSFR6. When comparing the whole sequences, there is 61.7% protein identity between AtSFR6 and OsSFR6. OsSFR6 encodes a predicted protein of 1170 amino acids (the length of AtSFR6 protein is 1268; Knight et al., 2009).

Figure 1.

Alignment of predicted Arabidopsis thaliana Arabidopsis and rice SFR6 protein sequences. Alignment was produced using Vector NTI software http://www.invitrogen.com. Identity and similarity of amino acid sequences are indicated in dark and pale grey boxes respectively. Gaps in the amino acid sequences are indicated by ‘-’.

To establish whether OsSFR6 is an orthologue of AtSFR6, we tested complementation of the Arabidopsis sfr6-1 mutant (Knight et al., 2009). Previously, we used three mutant alleles of AtSFR6 to prove linkage of AtSFR6 to the phenotypes of freezing sensitivity, pale cotyledons and leaves and large cotyledons but complementation had not been attempted. Therefore, before testing the effect of OsSFR6 expression in an sfr6 mutant background, we tested whether AtSFR6 itself, expressed under the control of the 35S promoter, was capable of complementing the visible sfr6-1 mutant phenotype. Fig. 2(a) shows four independent 35S::AtSFR6 lines in the sfr6-1 background (lower row). These all showed complementation of the visible pale leaf and cotyledon phenotype. This complementation was not apparent in 35S::GUS controls in the sfr6-1 background (Fig. 2a,b). Similarly, expression of 35S::OsSFR6 in the sfr6-1 background resulted in complementation of the visible phenotype (Fig. 3). However, in contrast to complementation with AtSFR6, OsSFR6 complemented to different extents in different lines. Fig. 3(b) shows one line, #8, with relatively weak complementation compared with another line, #10, which showed strong complementation.

Figure 2.

Visible phenotype of the Arabidopsis thaliana sfr6-1 mutant is restored by expression of 35S::AtSFR6. Columbia (Col-0), sfr6-1 and T2 progeny of sfr6-1 plants transformed to express either 35S::AtSFR6 (four independent transformed lines shown) or 35S::GUS (two independent transformed lines shown) were grown on full-strength Murashige and Skoog (MS) agar plates. (a) Seedlings 10 d after germination. (b) Close-up of Col-0, sfr6-1, sfr6-1 expressing 35S::GUS and sfr6-1 expressing 35S::AtSFR6. Bars, 10 mm.

Figure 3.

Visible phenotype of the Arabidopsis thaliana sfr6-1 mutant is restored by expression of 35S::OsSFR6. T2 progeny of sfr6-1 plants expressing 35S::OsSFR6 (six independent transformed lines shown) alongside Columbia (Col-0) and sfr6-1 mutant for comparison. (a) Seedlings 10 d after germination. (b) Close-up of Col-0, sfr6-1 and sfr6-1 expressing 35S::OsSFR6. Bars, 10 mm.

We have previously shown that sfr6 mutants of Arabidopsis are unable to acclimate to freezing, as a result of reduced cold-induced COR gene expression (Knight et al., 1999, 2009; Boyce et al., 2003). Therefore, to test if the reduced COR gene expression phenotype could also be complemented with AtSFR6, we tested expression of AtKIN2, a typical COR gene, which shows reduced expression in sfr6 mutants following cold treatment (Knight et al., 1999, 2009; Boyce et al., 2003). As can be seen in Fig. 4, whilst the sfr6-1 mutant showed low levels of AtKIN2 expression in the cold compared with wild-type Columbia (as reported previously; Knight et al., 1999, 2009; Boyce et al., 2003), three lines complemented with AtSFR6 showed levels of KIN2 expression comparable to the wild-type (Fig. 4a). These three lines, #1, #2 and #6, were chosen as they showed medium, low and high levels of AtSFR6 expression, respectively (Fig. 4b). Interestingly, differences in AtSFR6 expression did not result in different levels of AtKIN2 expression (Fig. 4a).

Figure 4.

AtKIN2 expression is restored in the Arabidopsis thaliana sfr6-1 mutant by complementing with AtSFR6. Seven-d-old seedlings were subjected to cold treatments at 4°C for 6 h. (a) Relative quantitation (RQ) values for KIN2 expression in three complemented lines relative to Columbia (Col-0). (b) RQ values for AtSFR6 expression in three complemented lines relative to Col-0. Expression of AtKIN2 and SFR6 was normalized to expression values for β-TUBULIN4 (endogenous control). Each value is the mean of three technical replicates. Error bars indicate RQmin and RQmax and constitute the acceptable error for a 95% confidence limit according to Student’s t-test.

To test whether OsSFR6 also was capable of complementing the low AtKIN2 expression phenotype, we tested six sfr6-1 lines complemented with OsSFR6. As can be seen in Fig. 5, these six lines showed a range of OsSFR6 expression levels: there was an approx. sixfold difference between the lowest (line #8) and the highest level (line #10). We therefore tested both of these lines, and a third line expressing OsSFR6 to intermediate levels (line #19) for COR gene expression in the cold. Fig. 6 shows the expression of COR genes AtKIN2 and AtLTI78 in these three lines. AtKIN2 and AtLTI78 expression was significantly lower in line #8 than in line #10. Line #19 showed slightly reduced COR gene expression, but not significantly, when compared with line #10 (Fig. 6).

Figure 5.

Expression levels of OsSFR6 in Arabidopsis thaliana sfr6-1 transformed with 35S::OsSFR6. Relative quantitation (RQ) values of OsSFR6 expression in 7-d-old Arabidopsis seedlings from six independently transformed lines are presented relative to the level of OsSFR6 expression in line #20. OsSFR6 expression was normalized to expression values for PEX4 (endogenous control). Each value is the mean of three technical replicates. Error bars indicate RQmin and RQmax and constitute the acceptable error for a 95% confidence limit according to Student’s t-test.

Figure 6.

Cold-inducible AtKIN2 (a) and AtLTI78 (b) expression is restored in the Arabidopsis thaliana sfr6-1 mutant by expressing 35S::OsSFR6. Seven-d-old seedlings from three independently transformed lines (low, medium and high levels of OsSFR6 expression) were subjected to cold treatments at 4°C for 6 h alongside Columbia (Col-0) and untransformed sfr6-1. Data represented here are relative quantitation (RQ) values of gene expression in the transformed lines relative to cold-treated Col-0. OsSFR6 expression was normalized to expression values for PEX4 (endogenous control). Each value is the mean of three technical replicates. Error bars indicate RQmin and RQmax and constitute the acceptable error for a 95% confidence limit according to Student’s t-test.

Given the complementation of the COR gene expression phenotype, we tested whether this would also lead to restoration of freezing tolerance. Fig. 7 shows that the three lines of sfr6-1 complemented with AtSFR6 that were tested for COR gene expression all showed freezing tolerance comparable to that of the wild-type (Fig. 7a). In a separate experiment, the three lines of sfr6-1 complemented with OsSFR6 showed visible symptoms consistent with variable degrees of freezing tolerance: line #8 appearing indistinguishable from the original sfr6-1 mutant, and lines #10 and #19 showing tolerance comparable to the wild-type (Fig. 7b).

Figure 7.

Freezing tolerance is restored in the Arabidopsis thaliana sfr6-1 mutant by expression of either 35S::AtSFR6 or 35S::OsSFR6. Five-wk-old plants were cold-acclimated at 4°C for 11 d under short-day conditions before exposure to either −6.5, −7.5 or −8.5°C for 24 h and then returned to 20°C. Photographs were taken 5 d after returning to 20°C. (a) Freezing tolerance of Columbia (Col-0), sfr6-1, sfr6-1 expressing 35S::GUS and sfr6-1 expressing 35S::AtSFR6 (Lines #1, #2 and #6). (b) Freezing tolerance of Col-0, sfr6-1 and sfr6-1 expressing 35S::Os SFR6 (lines #8, #10 and #19). Panels (a) and (b) depict two separate experiments and should not be compared. Bars, 50 mm.

We have shown previously that AtSFR6 is a regulator of both osmotic stress and low temperature responses (Knight et al., 1999; Boyce et al., 2003). To assess whether OsSFR6 is a potential regulator of osmotic stress responses too, we examined the ability of OsSFR6 to restore osmotic stress responsiveness and tolerance in the three complemented lines. Transcript abundances of the COR genes AtKIN2 and AtLTI78 were measured in response to a 6-h treatment with 350 mM mannitol. As expected, the treatment strongly induced both genes in Col-0 wild-type plants, with a reduced response seen in sfr6-1 (Fig. 8). Varying degrees of restoration of the response were seen in the three complemented lines; little or no effect was observed with the lowest expresser, line #8, whilst AtLTI78 and AtKIN2 transcript abundances in lines #10 and #19 were restored almost to wild-type values (Fig. 8).

Figure 8.

Osmotic stress-inducible AtKIN2 (a) and AtLTI78 (b) expression is restored in the Arabidopsis thaliana sfr6-1 mutant by expression of 35S::OsSFR6. Seven-d-old seedlings from three independently transformed lines (low, medium and high levels of OsSFR6 expression) were floated on water (open bars) or 350 mM mannitol (closed bars) at 20°C for 6 h alongside Columbia (Col-0) and untransformed sfr6-1. Data represented here are relative quantitation (RQ) values of gene expression in the transformed lines relative to water-treated Col-0. OsSFR6 expression was normalized to expression values for PEX4 (endogenous control). Each value is the mean of three technical replicates. Error bars indicate RQmin and RQmax and constitute the acceptable error for a 95% confidence limit according to Student’s t-test.

To examine whether this restoration of osmotically induced COR gene expression was accompanied by a return to wild-type degrees of osmotic stress tolerance, we performed two assessments. We showed previously that sfr6-1 is sensitive to osmotic stress at both the germination and seedling stages (Boyce et al., 2003). Therefore we tested the ability of the three OsSFR6 complemented lines to tolerate a range of mannitol concentrations. Seedlings were floated on mannitol (0, 330, 440 and 550 mM) for 72 h in a standard 16 : 8 h, light : dark cycle and examined for signs of osmotic stress-induced chlorosis after this time. Seedlings of each line maintained in water showed no signs of damage (Fig. 9). Wild-type plants showed slight signs of chlorosis with the 440 mM treatment, becoming more severe at 550 mM, whilst sfr6-1 was clearly more susceptible, showing some signs of chlorosis even at 330 mM and becoming severe at 440 mM. In complemented line #8, only very minor improvements in osmotic stress tolerance were observed; in lines #10 and #19, tolerance was restored to levels similar to the wild-type (Fig. 9).

Figure 9.

Osmotic stress tolerance is restored in Arabidopsis thaliana sfr6-1 mutant seedlings by expression of 35S::OsSFR6. Five 8-d-old seedlings from three independently transformed lines (low, medium and high levels of OsSFR6 expression) were transferred to a multi-well culture dish into wells containing water, 330, 440 or 550 mM mannitol and maintained at 20°C for 72 h alongside Columbia (Col-0) and sfr6-1. Bar, 10 mm.

Seeds sown on agar plates containing 0, 200, 300 or 400 mM mannitol were used to assess the effects of elevated concentrations of osmoticum on germination success. This assay allowed us to make a quantitative assessment of the effects of expressing OsSFR6 to different levels in sfr6-1. Small reductions in the percentage of wild-type Col-0 seeds germinating were observed with each increase in mannitol concentration; germination rate fell from close to 100% to c. 70% in wild-type plants when mannitol concentration was raised from 0 to 400 mM. As reported previously, sfr6-1 seed germination was more sensitive to the high osmoticum concentrations; germination fell to only 38% at 300 mM and to below 20% at 400 mM (Fig. 10a). Our analysis confirmed that germination success on elevated concentrations of osmoticum differed significantly between Col-0 wild-type and sfr6-1 (LRT; G4 = 73.2, < 0.001). For both lines, germination success was reduced as osmoticum concentration increased; however, for any given concentration of osmoticum, germination frequency was always higher for the wild-type (Fig. 10a).

Figure 10.

Expression of 35s::OsSFR6 in Arabidopsis thaliana sfr6-1 restores the ability to germinate on high concentrations of osmoticum (a). Observed and best-fit model predictions of seed germination over a range of mannitol (osmoticum) concentrations for wild-type Columbia (Col-0) (squares) and sfr6-1 (circles). (b) Observed and predicted effect of OsSFR6 transcript expression level, using Eqn 2, on germination success of OsSFR6 complemented sfr6-1 lines #8 (1), #19 (3.67) and #10 (6.85). Numbers in parentheses refer to relative OsSFR6 expression levels in each line, and increasing amounts of shading indicate increased transcript abundances. sfr6-1 is shown for comparison. In both panels, the lines represent maximum-likelihood model fits to the data, and the error bars represent estimated 95% confidence intervals for the probability of seed germination, based on the data in each line–treatment pairing.

When comparing the behaviour of the three complemented lines with noncomplemented sfr6-1, we also found significant evidence that the abundance of OsSFR6 transcripts (see Fig. 5) affected germination success in sfr6-1 lines transformed with 35S::OsSFR6 (LRT; G2 = 53.7, < 0.001). Specifically, an increase in OsSFR6 transcript abundance increased germination success across all concentrations of osmoticum investigated (Fig. 10b). The complemented line associated with the highest OsSFR6 transcript abundance (line #10; y = 6.85) exhibited a significantly higher germination success rate compared with wild-type Col-0 (LRT; G4 = 40.9, < 0.001). In fact, for all four concentrations of osmoticum, this line showed higher germination success than the wild-type (c.f. Fig. 10a,b). Interestingly, the fits suggest that the rate of reduction in germination success with increased osmoticum may be less for the wild-type (Fig. 10).


Identification of plant genes that contribute to environmental stress tolerance is vital for crop breeding if food security is to be maintained for a rapidly growing human population in an increasingly unpredictable climate. Previous work has identified a number of genes that contribute to these traits in plants, but arguably the most significant discoveries have been key regulators, for instance, transcription factors. Such genes encode master regulators that control the expression of many other genes involved in a particular trait, and thus their effect individually is profound. Good examples of these are the CBF/DREB1 (Stockinger et al., 1997; Jaglo-Ottosen et al., 1998) and DREB2 (Liu et al., 1998; Sakuma et al., 2006) transcription factors, originally identified in Arabidopsis but which also exist in rice (Dubouzet et al., 2003; Ito et al., 2006; Matsukura et al., 2010). The CBF/DREB1 and DREB2 transcription factors regulate the expression of so-called COR genes via a single promoter motif, the DRE/CRT (Yamaguchi-Shinozaki & Shinozaki, 1994), in response to low temperature and osmotic stress, respectively.

Our previous work showed that CBF/DREB1- and DREB2-dependent stress gene expression in Arabidopsis requires AtSFR6 (Knight et al., 1999, 2009; Boyce et al., 2003). Loss-of-function sfr6 mutants of Arabidopsis show reduced expression of genes controlled by the DREB transcription factors in response to either osmotic stress or cold. As a result, atsfr6 mutants are unable to mount the correct defence against these conditions and are sensitive to both dehydration and freezing. Thus, SFR6 is a hub regulating at least two transcription factor systems in Arabidopsis, affecting two overlapping gene regulons leading to freezing and osmotic stress tolerance. Orthologues of AtSFR6 in crop species therefore represent good targets for future breeding or manipulation. With this in mind, we identified a rice homologue of AtSFR6 and, through testing its function, confirmed it as an orthologue.

Examination of the rice genome revealed a sole gene (Os10g35560) showing any significant homology to AtSFR6. We named this gene OsSFR6. AtSFR6 also exists as a single copy gene in Arabidopsis. Empirical determination of the coding region of OsSFR6 showed that the predicted coding region had high homology to AtSFR6 (61.7% identity and 71.1% similarity at the predicted protein sequence level: Fig. 1). Interestingly, the N-terminal half of the predicted OsSFR6 protein sequence was more highly conserved than the C-terminal half. In Arabidopsis, three mutations in the N-terminal third of AtSFR6 strongly affect phenotype (Knight et al., 2009). Thus it seems likely that the N-terminal parts of AtSFR6 and OsSFR6 are important for their function.

Having identified a potential orthologue of AtSFR6, we sought to test for orthology by functional complementation of an Arabidopsis atsfr6 mutant. First, it was necessary to demonstrate that this was a viable approach, therefore we tested complementation with Arabidopsis AtSFR6 itself. Expressing AtSFR6 using a 35S CaMV constitutive promoter in an atsfr6 background fully restored the ability to induce COR gene expression in response to cold, and also to allow cold acclimation and acquisition of freezing tolerance (Figs 4, 7). It is most likely that the restoration of cold acclimation is as a direct consequence of the restoration of full levels of COR gene expression: up-regulation of COR gene expression by overexpression of CBF/DREB1 transcription factors at ambient temperature is sufficient to induce freezing tolerance (Jaglo-Ottosen et al., 1998).

Having established a system for functional testing of SFR6 orthologues by complementation, we used this approach with the coding region of OsSFR6. OsSFR6, like AtSFR6, was able to restore both cold-induced COR gene expression and acquisition of freezing tolerance (Figs 6, 7). However, wild-type levels of COR gene expression and freezing tolerance were only achieved in the highest OsSFR6 expressing lines (lines #10 and #19); poor levels of complementation were observed in the low (line #8) level expresser (Figs 6, 7).

These experiments demonstrated that OsSFR6 (from rice, a species incapable of freezing tolerance) can act as a functional orthologue of AtSFR6 in the acquisition of freezing tolerance in Arabidopsis. Osmotic stress is a major component of freezing stress, and in accordance with this, the targets of CBF/DREB1 and DREB2 transcription factors overlap substantially. Overexpression of both CBF/DREB1 (Jaglo-Ottosen et al., 1998) and constitutively active forms of DREB2 (Sakuma et al., 2006) led to elevated levels of COR gene expression and to both freezing and osmotic stress tolerance. It appears likely, therefore, that the role of OsSFR6 in rice is to facilitate tolerance to osmotic rather than freezing stress. To test this possibility, we examined COR gene expression and sensitivity to osmotic stress conditions in sfr6-1 lines overexpressing OsSFR6. Osmotic stress-inducible COR gene expression and tolerance of elevated osmoticum concentrations at seedling and germination stages were all complemented in sfr6-1 lines expressing 35S::OsSFR6 (Figs 8–10).

When we modelled our quantitative germination data, our best-fitting model demonstrated a significant increase in germination success with increasing transcript abundances in sfr6-1 lines expressing 35S::OsSFR6 (Fig. 10b). OsSFR6 transcript abundances are likely to be a predictor of protein levels (although the relationship between the two cannot be assumed to be linear). Therefore our data strongly suggest that the degree of restoration of wild-type phenotype in sfr6-1 is positively correlated with the level of OsSFR6 protein expression. This is similar to the trend we saw in the qualitative assessments of freezing and osmotic stress tolerance (Figs 7, 9), and our measurements of COR gene expression (Figs 6, 8). Interestingly, only in the case of germination did we observe indications that expressing OsSFR6 to higher levels can actually supersede wild-type degrees of tolerance (Fig. 10). This result might suggest a significant role for SFR6 in osmotic stress tolerance in the germinating seed.

The fact that OsSFR6 appears to fully complement Arabidopsis sfr6 loss-of-function mutants only when expressed at relatively high levels, whilst all levels of AtSFR6 overexpression resulted in complementation, could be interpreted as differences in protein sequence between the two orthologues producing proteins with different efficiencies. However, our quantitation of SFR6 transcripts was relative; comparison of absolute levels of OsSFR6 with AtSFR6 cannot be made from our data. Furthermore, irrespective of whether or not OsSFR6 and AtSFR6 transcripts were expressed to similar levels, we cannot rule out the possibility of substantial differences in the levels of expressed OsSFR6 and AtSFR6 proteins in our complemented lines and that these differences account for the dose-dependent effect we see with OsSFR6 complementation. In either scenario, it can still be concluded that OsSFR6 is a functional equivalent (orthologue) of AtSFR6. As OsSFR6 affects osmotic stress-responsive COR gene expression and tolerance in Arabidopsis, it is very likely that OsSFR6 plays a role in tolerance of rice to osmotic stress during periods of low water availability.

Our data demonstrate that OsSFR6 is a potential target for breeding or manipulation to achieve increased abiotic stress tolerance in rice. Given that OsSFR6 is functionally equivalent to AtSFR6, it is most likely that homologues from other crops will be orthologues as well, and thus be equally valuable targets. The most obvious avenue to explore in the exploitation of SFR6 would be to increase its production in crop species; however, we have observed that overexpression of AtSFR6 in wild-type Arabidopsis does not lead to enhanced expression of COR genes in response to cold (Supporting Information Fig. S1), or enhanced freezing tolerance (data not shown). In addition to the implications this has on the use of SFR6 in future crop protection strategies, this result gives an insight into the possible mode of action of the protein. Because increasing the titre of AtSFR6 protein has no effect in vivo, we surmised that SFR6 is likely to work in conjunction with other proteins in stoichiometric proportions, as part of a complex. If this were the case, elevating SFR6 levels in the absence of increases in the amounts of these other proteins would not be expected to enhance COR gene expression.

This hypothesis has now been proved correct, with the identification of At4g04920 (AtSFR6) as the gene that encodes the Arabidopsis homologue of yeast MED16, part of the mediator complex (Bäckström et al., 2007). Mediator is a multi-subunit transcriptional coactivator complex that acts as a bridge between DNA-bound transcriptional regulators and the general RNA polymerase II transcriptional machinery. MED16 is one of the so-called ‘tail’ subunits of mediator, whose functions are considered to be directly involved with transcription factor recruitment (Casamassimi & Napoli, 2007). Yeast MED16 (SIN4) (Li et al., 1995) and Drosophila MED16 orthologues (Kim et al., 2004) have demonstrated roles in facilitating transcriptional activation by transcription factors. If the stoichiometry of mediator subunits remains constant, simple overexpression of OsSFR6 in rice or orthologues in other crop species is unlikely to result in enhanced stress tolerance. However, the ability of OsSFR6 to elevate atsfr6-1 germination rates on high concentrations of osmoticum to above wild-type values does raise the possibility that orthologues from different species may have differing effectiveness in some cases. In the main, however, future exploitation of SFR6 in rice or other crop species is likely to necessitate engineering the protein sequence to improve efficiency. Identification of transcription factor binding sites in SFR6 and tailoring these to optimize transcription factor binding may be one approach that could be adopted. This will be the focus of our future work in this area.


We thank Project Sri Lanka for the PhD studentship awarded to D.L.W. We are grateful to Lesley Edwards for the gift of rice seeds (Oryza sativa) cv Japonica var. Lemont.