•Arabidopsis SENSITIVE TO FREEZING6 (SFR6) controls cold- and drought-inducible gene expression and freezing- and osmotic-stress tolerance. Its identification as a component of the MEDIATOR transcriptional co-activator complex led us to address its involvement in other transcriptional responses.
•Gene expression responses to Pseudomonas syringae, ultraviolet-C (UV-C) irradiation, salicylic acid (SA) and jasmonic acid (JA) were investigated in three sfr6 mutant alleles by quantitative real-time PCR and susceptibility to UV-C irradiation and Pseudomonas infection were assessed.
•sfr6 mutants were more susceptible to both Pseudomonas syringae infection and UV-C irradiation. They exhibited correspondingly weaker PR (pathogenesis-related) gene expression than wild-type Arabidopsis following these treatments or after direct application of SA, involved in response to both UV-C and Pseudomonas infection. Other genes, however, were induced normally in the mutants by these treatments. sfr6 mutants were severely defective in expression of plant defensin genes in response to JA; ectopic expression of defensin genes was provoked in wild-type but not sfr6 by overexpression of ERF5.
•SFR6/MED16 controls both SA- and JA-mediated defence gene expression and is necessary for tolerance of Pseudomonas syringae infection and UV-C irradiation. It is not, however, a universal regulator of stress gene transcription and is likely to mediate transcriptional activation of specific regulons only.
The sfr6 mutant of Arabidopsis was originally identified as being unable to acclimate to freezing temperatures (McKown et al., 1996; Warren et al., 1996). Cold acclimation is the process whereby some temperate species become, upon exposure to low nonfreezing temperatures, able to increase their tolerance of subsequent freezing conditions (Thomashow, 1990). During acclimation a number of physiological and biochemical changes occur in the plant, as well as the increased expression of a battery of cold-inducible genes (Thomashow, 1999). sfr6 is unable to cold acclimate due to a failure to activate full expression of the COR (cold onregulated) genes (Knight et al., 1999) via the CRT/DRE (c-repeat or drought-responsive promoter element (Yamaguchi-Shinozaki & Shinozaki, 1994)) (Boyce et al., 2003). The c-repeat binding factor, CBF (DRE-binding, DREB1) transcription factors (Gilmour et al., 1998; Liu et al., 1998) activate COR gene expression via the CRT/DRE in response to low temperature, whilst DREB2 and CBF4 transcription factors activate expression via the same element in response to osmotic stress (Liu et al., 1998; Haake et al., 2002). As in the case of low temperature, osmotic-stress-activated COR gene expression is reduced also in sfr6 mutants (Knight et al., 1999; Boyce et al., 2003). We have shown that CBF transcripts and CBF1 protein are expressed to wild-type levels in sfr6-1 in response to cold (Knight et al., 1999, 2009), and that overexpression of CBF1 or CBF2 leads to ectopic COR gene expression in wild-type but not sfr6-1 plants (Knight et al., 2009). Together, these observations indicate that SFR6 acts downstream of the CBF transcription factors (TFs) and is likely to be involved in facilitating specific TF activity at promoter cis elements. However, it is clear that, in the case of cold at least, not all TFs are subject to regulation by SFR6; the targets of some TFs show no misexpression, indicating that SFR6 is not a general regulator of transcription and its action is specific to certain gene regulons (Knight et al., 1999).
The hypothesis that SFR6 could facilitate TF activity at promoter elements was supported by the identification of SFR6 as MED16, a subunit of the multiprotein transcriptional co-activator complex Mediator (Bäckström et al., 2007). Mediator exists in all eukaryotes and acts as a bridge between TFs and RNA polymerase II, conveying stimulus-specific information between the two to achieve positive or negative regulation of gene expression (Conaway & Conaway, 2011). Plant Mediator consists of c. 30–35 subunits (Bäckström et al., 2007; Bourbon, 2008). The discovery that SFR6 encodes a component of Mediator explains the lack of effect of overexpressing SFR6 (Wathugala et al., 2011). SFR6 is most likely to be present in stoichiometric proportions with other Mediator subunits and therefore overexpressing SFR6 without increasing the abundance of the other subunits would be unlikely to alter the efficacy of the whole complex. The identification of SFR6 as a Mediator subunit also offered an explanation as to why SFR6 affects other gene regulons apparently unrelated to CBF/DREB action, specifically the control of the circadian clock-related and flowering time gene expression (Knight et al., 2008).
A number of previously described Arabidopsis mutants are now known to be Mediator subunit loss of function mutants; the specific phenotypes of these can now be ascribed to alterations in the composition of the Mediator complex. Developmental and biochemical aberrations are associated with these mutations although none have been attributed with failure to respond to low temperature. swp (struwwelpeter) was originally reported to affect leaf development and cell numbers (Jonak et al., 2002) and seth10 results in reduced pollen tube growth (Lalanne et al., 2004). SWP and SETH10 have now been identified as MED14 and MED8, respectively (Bäckström et al., 2007). REF4 was originally identified through a mutant screen for components involved in phenylpropanoid metabolism (Ruegger & Chapple, 2001; Stout et al., 2008) but has now been identified as a MED5 subunit (Bäckström et al., 2007; Bourbon, 2008). Recent work has revealed PFT1 (PHYTOCHROME FLOWERING TIME 1), a regulator of flowering time, to be the Mediator subunit MED25 (Bäckström et al., 2007) and an additional role for MED25 along with MED8 has been demonstrated in defence against necrotrophic pathogens (Kidd et al., 2009).
Therefore, we considered it possible that as a Mediator subunit, SFR6 may control additional transcription events dependent upon other combinations of cis- and trans-acting factors. We sought to discover whether SFR6 plays a role in other transcriptional response pathways, particularly those utilized in plant defence against biotic stresses.
Materials and Methods
Plant materials and growth
Arabidopsis thaliana (L.) Heynh ecotype Columbia (Col-0), eds1 (Parker et al., 1996), NahG overexpressing line (Delaney et al., 1994) and sfr6-1, sfr6-2 and sfr6-3 mutant allele plants (Knight et al., 2009) were grown on 1× Murashige and Skoog 0.8% (w/v) agar plates as described previously (Knight et al., 1999). Lighting was maintained at 150–200 μmol m−2 s−1 with a 16 h photoperiod with temperature at 20 (± 1)°C. Soil-grown plants were raised in growth chambers under comparable temperature and light conditions. ERF5 overexpressing lines were generated by cloning the full length ERF5 (At5g47230) coding sequence into the Gateway entry vector pENTR/D-TOPO (Invitrogen, Paisley, UK), before cloning into the pK2GW7 vector (Karimi et al., 2002), which contains the cauliflower mosaic virus 35S promoter. Col-0 and sfr6-1 plants were transformed using Agrobacterium-mediated floral dip with Agrobacterium tumefaciens strain C58C1 as described previously (Clough & Bent, 1998). Transformants were selected on the basis of their growth on MS medium containing 50 μg ml−1 kanamycin and four lines each of transformed Col-0 and sfr6-1 confirmed as having similar levels of ERF5 expression were chosen for further study.
Measurement of gene expression
Gene expression levels were analysed by quantitative real-time PCR using an Applied Biosystems 7300 system. 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 RNeasy Plant Total RNA Kit (Qiagen) in conjunction with Qiagen RNase-free DNase to remove any genomic DNA contamination. For quantitative real time PCR reactions, 10 μl of a 1 : 50 dilution of cDNA (with the exception of CASPASE8, for which a 1 : 20 dilution was used) was amplified in a 25-μl reaction in an optical 96-well plate with three technical replicates for each sample on the same plate. Gene-specific TAQMAN PROBES (Applied Biosystems) were used for CASPASE8 (At1G16420; AB probe identifier At02209654_g1), PR1 (At2G14610; At02170748_s1), OXI1 (At3g25250; At02280071_g1) and EDS5 (At4G39030; At02170432_g1) in conjunction with Taqman universal PCR mix (Applied Biosystems). Expression levels were normalized to the expression of either β-TUBULIN4 (At5G44340; At02337699_g1) or PEX4 (At5g25760; At02304594_g1), (endogenous control genes we have used previously; Knight et al., 2009; Wathugala et al., 2011) with the exception of data from experiments using UV irradiation. We discovered that expression of β-TUBULIN4 is strongly induced by UV irradiation (data not shown); therefore, in the case of these experiments, we normalized to the expression of At4g24410 (probe identifier At02239002_g1), a gene whose expression does not alter under such conditions (Genevestigator; https://www.genevestigator.com). All other transcripts studied were detected using gene specific primers, which we had first validated, using Fast Start SYBR green mastermix with ROX (Roche Diagnostic). In these experiments, 5 μl cDNA (1 : 50 dilution) was used in a 15-μl reaction. All primer sequences are shown in Supporting Information Table S1. These included ERF5 (At5g47230), PDF1.2A (At5g44420), PDF1.1 (At1g75830), PR2 (At3g57260), PR5 (At1g75040), WRKY18 (At4g31800), WRKY60 (At1g80840) and TCH3 (At2g41100). In the case of the closely related genes PDF1.1 and PDF1.2A, primers were designed to discriminate between the two transcripts. In these experiments expression levels were normalized to expression of PEX4 (At5g25760) except in the case of UV treatments, where normalization to At4g24410 was used. Relative quantification was performed by the ΔΔCT (comparative CT) method (Livak & Schmittgen, 2001) and Relative Quantification (RQ) values and estimates of statistical variation (SV) for each sample calculated as previously (Knight et al., 2009). The algorithm used is described in Relative Quantitation (RQ) algorithms (Applied Biosystems, 2007). Error bars represent RQMIN and RQMAX and constitute the acceptable error level for a 95% confidence level according to student’s t-test. Results shown in each case are representative of three separate biological repeat experiments.
Seeds were sown individually and evenly on horizontal 1× MS agar plates. At 7 d the seedlings were irradiated with 5, 10, 15 and 20 kJ m−2 of UV-C, (wavelength 254 nm) by removing the plate lids and placing in a UV cross linker (Uvitec Ltd, Cambridge, UK) set to deliver the designated level of energy. Lids were removed from the control plates for 10 min (the maximum time taken to administer the 20 kJ m−2 treatments). Immediately after irradiation all plates including control plates were resealed with micropore tape and wrapped with aluminium foil individually (to avoid initiation of the blue light induced repair pathway) and returned to the growth chamber. After 24 h they were unwrapped and maintained in the same chamber. The number of seedlings surviving was recorded 10 d post-treatment, and scored on the basis of a green coloured meristem indicating survival.
For analysis of UV-C-induced gene expression, UV irradiation was performed as described above and then plates were resealed and returned to the growth room. Samples were collected 1, 6 and 24 h after treatment. Fifteen to twenty seedlings were collected in a 1.5-ml microfuge tube from each sample and frozen immediately in liquid nitrogen for RNA extraction.
Inoculation with Pseudomonas syringae DC3000
Five-week-old plants grown in relatively short days (12 h light : 12 h dark cycles in 150–200 μmol m−2 s−1 light) to promote rosette leaf growth were inoculated using the syringe injection method (Katagiri et al., 2002). Responses to pathogens may vary depending on at what point in the circadian cycle plants are challenged (Roden & Ingle, 2009); therefore, inoculations were always performed 3 h into the light phase. A single colony of Pseudomonas syringae (Pst DC3000) grown on solid KB medium (King’s medium B: 10 g l−1 glycerol, 10 g l−1 tryptone, 10 g l−1 peptone, 1.5 g l−1 MgSO4.7H2O, 1.5 g l−1 K2HPO4, 15 g l−1 agar) with rifampicin (50 μg ml−1) was used to inoculate an overnight culture in 50 ml liquid KB medium 1 d before plant inoculation.
On the day of inoculation bacteria were harvested from the 50 ml overnight culture by centrifugation for 5 min at 13 400 g and cells resuspended in 10 ml of sterile water. Cells were centrifuged again for 1 min at 13 000 rpm and resuspended in 5 ml of sterile water. The sample densities were adjusted to OD600nm = 0.2 (this is equivalent to 108 CFU ml−1) with sterile water. The culture was diluted 1 : 100 and thus infections were carried out with OD600nm = 0.002 = 106 CFU ml−1. Three to five leaves of the same age from each plant were selected before inoculation. Three replicate plants per line were infected for each time point assayed using a 1-ml syringe to apply gentle pressure to the abaxial side of the leaf. Control plants were infiltrated with sterile water. The infected plants were covered with cling-film in order to maintain high humidity and returned to the 24°C chamber.
The first sample (day 0) was harvested 3 h after inoculation and the subsequent samples at 24 h intervals thereafter. Samples were taken from three independent plants for each time point. One leaf was harvested from each plant and leaves were laid in a stack and circles punched out from them using a cork borer (tissue area 0.283 cm2). Leaf discs collected together in this way were pooled as one sample. Discs were homogenized immediately using a plastic micropestle in a 1.5-ml microfuge tube containing 1 ml of sterile water. The following serial dilutions were made for samples from each time point: day 0, undiluted and 10−1; day 1, 10−1 to 10−3; day 2, 10−1 to 10−6; day 3, 10−1 to 10−8. Ten microlitres of each dilution for each plant line was plated onto a single KB agar plate containing rifampicin, and plates inverted after spots had dried and incubated at 28°C for 2 d. After 2 d of incubation the number of colonies was counted for each dilution. Colonies were counted on the plate corresponding to the highest dilution at which their numbers could be easily and accurately counted and this was used to calculate CFU per unit area. For RNA extraction before gene expression analysis, leaf samples were collected from inoculated leaves, transferred to 1.5-ml microfuge tubes and frozen immediately in liquid nitrogen.
Salicylic acid and jasmonic acid treatments
To administer salicylic acid (SA) treatment soil-grown 3-wk-old plants were dipped in 1 mM sodium salicylate solution with 0.01% Silwet L-77 (Lehle Seeds, Round Rock, TX, USA) then returned to the growth chamber for 24 h before sample collection. Control plants were dipped in Silwet L-77 only. For jasmonic acid (JA) experiments, 12-d-old seedlings were floated on 4 ml water in 6-well plastic culture dishes overnight in the growth chamber. The next day, JA was added to a final concentration of 100 μM and mixed by gentle swirling. Water containing 0.1% ethanol (the solvent for the JA stock) was added to control plants. Samples were returned again to the growth chamber, dishes sealed with parafilm to prevent transfer of the volatile JA between samples, and harvested after 24 h.
SFR6 was identified in a screen for Arabidopsis mutants that fail to cold acclimate to freezing temperatures (Warren et al., 1996). We have shown previously that SFR6 is essential for the full expression of COR (cold onregulated) genes in response to either low temperature or drought. As a result of this deficiency, sfr6 mutants show reduced tolerance to osmotic stress and freezing conditions (Knight et al., 1999, 2009; Boyce et al., 2003). In addition to its effect on COR genes, mutation of SFR6 results in reduced expression of circadian clock-controlled genes whose promoter contains an evening element (EE) (Knight et al., 2008). sfr6 mutants exhibit altered responses to the circadian clock and delayed flowering, likely as a consequence of reduced expression of clock-related and flowering time genes (Knight et al., 2008).
Whilst these data suggest that SFR6 may have a wider role as regulator of gene transcription, our earlier studies into its role in low temperature-induced gene regulation indicate that SFR6 regulates specific gene regulons only (Knight et al., 1999). We sought, therefore, to test whether SFR6 was likely to be involved in the activation of other stress-responsive gene regulons, leading to tolerance of these conditions. Following our identification of SFR6 as At4g04920 (Knight et al., 2009) we learned that the protein encoded had been identified as MED16, a subunit of Mediator (Bäckström et al., 2007), a eukaryotic transcriptional co-activator complex required for communication between transcription factors (TFs) and RNA polymerase II (Conaway & Conaway, 2011). Other plant Mediator subunits have been identified recently and roles in development, biotic and abiotic stress tolerance ascribed to them (Bäckström et al., 2007; Kidd et al., 2009; Elfving et al., 2011). In this study, we sought to discover whether SFR6 has additional roles as a transcriptional regulator of other gene regulons, focusing on the pathways leading to defence gene expression.
Response to a virulent pathogen
Pseudomonas syringae pv. tomato (Pst) is a hemibiotrophic bacterial pathogen that is virulent on a number of plant species including certain accessions of Arabidopsis. The Pseudomonas strain DC3000 is established as a virulent pathogen of Arabidopsis accession Col-0 (Whalen et al., 1991), provoking a compatible response typified by the production of grey-brown lesions with chlorosis spreading out from the site of infection.
We examined the response of rosette leaves of 5-wk-old sfr6-1 mutant and wild-type plants to infection by virulent P. syringae DC3000. Additionally, plants of the pathogen-susceptible mutant eds1 (Parker et al., 1996) were inoculated as a positive control. eds1 and sfr6-1 showed similar levels of chlorosis when observed 5 d postinoculation, whilst Col-0 wild-type plants were less affected (Fig. 1a). In order to verify that the increased susceptibility to pathogen attack was attributable to loss of SFR6 function, two additional loss of function mutant sfr6 alleles (Knight et al., 2009) were studied and similar levels of chlorosis were observed as in sfr6-1 (Fig. 1b). The success of the pathogen strain on mutant and wild-type Arabidopsis was measured quantitatively by counting colony formation in plate assays using the leaf disc extraction method (Katagiri et al., 2002). Leaf discs were harvested from three replicate plants on day 0, and 1, 2 or 3 d postinoculation. The susceptible mutant eds1 exhibited far higher CFU values than wild-type, indicating that the assay was capable of discriminating between plants with varying susceptibility to this hemibiotrophic pathogen. Although less severely affected than eds1, all three sfr6 mutants showed significantly higher than wild-type CFU values 1, 2 or 3 d postinoculation (Fig. 1c), indicating increased susceptibility in sfr6 (med16 ) mutants. Each datum is the average of three leaf disc samples, each taken from a different plant and error bars (which may be too small to be visible) represent standard error of the mean (SEM).
Expression of stress-inducible genes in response to virulent pathogen infection
As SFR6/MED16 is a subunit of Mediator, it was possible that the compromised defence against a virulent pathogen observed in sfr6 mutant alleles would be due to a failure to upregulate defence gene expression. Expression of three pathogenesis-related genes PR1, PR2 and PR5, was assayed using quantitative real-time PCR (qRT-PCR). Samples were collected at 48 h after inoculation, a time point at which previous studies have shown PR1 to be highly expressed (Glazebrook et al., 1996). As expected, expression of all three PR genes was strongly induced after infection; but in all three cases this was induced to a significantly lesser degree in all three sfr6 alleles than in wild-type (Fig. 2a–c). Expression of PR5 was particularly severely reduced in sfr6 mutants (Fig. 2c). This indicated that SFR6/MED16 is required for the transcriptional response of these PR genes to pathogen infection and suggests that a failure to upregulate PR gene expression may be a cause of increased susceptibility in sfr6 mutants. To test whether sfr6 mutants exhibit a generic failure to respond transcriptionally to P. syringae DC3000, we monitored the expression of other genes. Genes encoding the transcription factors WRKY18 and WRKY60 are inducible by virulent P. syringae (Xu et al., 2006). OXI1, a gene that we have shown to be inducible by a number of signals involving oxidative burst (Rentel et al., 2004), encodes a kinase necessary for defence against P. syringae (Petersen et al., 2009). Samples were harvested at 24 h from the same sets of plants as treated in the experiments above, this being closer to the peak of expression for each gene. Transcripts of the two WRKY transcription factors were expressed normally or even to slightly higher levels in sfr6 mutants compared with wild-type (Fig. 3a,b). Similarly, OXI1 was expressed to at least wild-type levels in the three sfr6 mutant alleles (Fig. 3c). These data indicated that SFR6 is not required for all transcriptional responses to virulent P. syringae.
A complex array of defence genes is upregulated in plants in response to pathogen infection, using a number of different signalling pathways in which various plant hormones play a role (Bari & Jones, 2009). Salicylic acid (SA) is a key component in the pathway leading to defence gene expression in response to biotrophic and hemibiotrophic pathogens such as P. syringae. In order to test whether SFR6 is required specifically for mediating the induction of pathogenesis-related genes such as PR1 via SA, we measured the effect of direct addition of SA to wild-type and mutant plants. In all three sfr6 mutant alleles SA-induced levels of PR1 and PR2 gene expression were significantly reduced when measured 24 h after treatment (Fig. 4a,b). These results indicate that MED16/SFR6 is required for activation of PR gene expression via the SA-mediated pathway.
Expression of stress-responsive genes after UV treatment
Many defence-related genes are inducible by UV-C irradiation as well as by hemibiotrophic or biotrophic pathogens. Therefore, as we had observed reduced PR gene expression following infection with virulent P. syringae, we tested the transcriptional response of sfr6 mutants to UV-C treatment. Seven-day-old seedlings grown on 1× MS agar plates were treated with 5 kJ m−2 UV-C using a UV cross-linker. The expression of PR1 and EDS5 and CASPASE8, two other genes shown previously to be inducible by UV irradiation, was assayed at time points post-treatment that were chosen on the basis of peak expression occurrence in wild-type (Nawrath et al., 2002; He et al., 2008). Expression was measured using qRT-PCR and normalized to expression levels of At4g24410, a gene with stable expression that is not altered by UV treatment (https://www.genevestigator.com). Our data showed that expression of all three genes increased significantly after UV-C treatment. PR1 was highly upregulated in wild-type seedlings 24 h after treatment (Fig. 5) but levels were markedly reduced in all three mutant alleles (Fig. 5a). Similarly, expression of EDS5 6 h post-treatment was significantly reduced in mutant plants compared with wild-type (Fig. 5b). A third gene, CASPASE8, measured 1 h after treatment, was expressed to significantly lower levels in all three sfr6 mutant alleles than in the wild-type (Fig. 5c). To test whether sfr6 mutants exhibit a generic failure to mount a transcriptional response to UV-C stimuli, we assessed the expression of two other genes; OXI1, which we have observed to be UV-C inducible, and TCH3, which has been shown previously to be inducible by UV-C (Narusaka et al., 2003). Both genes were inducible by UV in our samples and expressed to normal wild-type levels in the three sfr6 mutant alleles (Fig. 6a,b). This indicated that SFR6 is required only for the induction of specific genes in response to UV-C.
Tolerance of UV-C radiation
In order to assess whether or not the necessity for SFR6 for UV-C induction of genes such as PR1, EDS5 and CASPASE8 could influence tolerance of UV-induced damage, we tested the ability of sfr6 mutants to survive a range of doses of UV-C. Seven-day-old seedlings grown on 1× MS agar plates were exposed to doses of UV-C irradiation between 5 and 20 kJ m−2. We have reported previously that nonstressed sfr6 mutant plants are distinctly yellow and less green than their wild-type counterparts (Knight et al., 2009). Therefore, we avoided the use of chlorophyll measurement to assess susceptibility to UV-C, and instead monitored whole seedling survival. The proportion of seedlings surviving after 10 d was monitored for wild-type and all three mutant alleles. Survival rates of wild-type seedlings were reduced after exposure to doses of 10 kJ m−2 UV-C irradiation or higher. Survival rates were significantly lower in the sfr6 mutants, which were sensitive even to the lowest dose of UV-C, 5 kJ m−2, showing 20–40% survival (Fig. 7b). Figure 7(a) shows the appearance of wild-type and sfr6-1 seedlings 10 d after treatment with 5 or 10 kJ m−2. sfr6-1 seedlings exhibited far higher levels of damage than wild-type plants; recovery was evident in the emerging true leaves at the apical meristem of wild-type seedlings but not the mutants. These data show that SFR6 is required for normal levels of PR1 and other defence-related gene expression in response to UV-C treatment, suggesting that reduced levels of expression seen in sfr6 mutants result in reduced UV-C tolerance.
In addition to SA-mediated defence, plants employ JA-mediated pathways also (Bari & Jones, 2009). As the Mediator subunits MED8 and MED25 have been shown to play a role in the transcriptional response to JA signalling (Kidd et al., 2009), we sought to test whether SFR6/MED16 functions in this response also. We began by adding JA directly to wild-type and sfr6 mutant Arabidopsis plants and testing the expression of known JA-responsive genes. Plants floated on water and treated with 100 μM JA showed a high degree of transcriptional upregulation of the plant defensin genes PDF1.2A and PDF1.1 in wild-type (Fig. 8). Very poor induction of the genes was seen in all three mutant alleles (Fig. 8a,b). These data showed that SFR6/MED16 is a regulator of JA-mediated defence gene expression as well as SA-mediated pathways, and that some JA-mediated defensin gene expression may be severely suppressed by loss of SFR6.
In order to investigate further the effect of SFR6/MED16 on JA-dependent gene expression, we overexpressed the ERF5 transcription factor in sfr6-1 mutants. The ERF transcription factors comprise a large gene family (Nakano et al., 2006), many of whose members are involved in defence responses (Ohme-Takagi et al., 2000; Berrocal-Lobo & Molina, 2004; McGrath et al., 2005; Onate-Sanchez et al., 2007). Four transgenic Col-0 wild-type lines and four sfr6-1 mutant lines expressing a 35S::ERF5 construct were chosen on the basis of their high, and similar, levels of ERF5 expression compared with untransformed Col-0 and sfr6-1 plants (Fig. 9a). We observed that PDF1.2A expression was highly induced in all four Col-0 overexpressing lines when compared with untransformed plants; however, sfr6-1 plants expressing ERF5 to similar levels showed little or no increase in PDF1.2A or PDF1 expression (Fig. 9b,c). This result indicates that SFR6/MED16 is required for plant defensin gene expression regulated by one or more TFs that include ERF5, and that SFR6/MED16 acts downstream of the ERF5 transcription factor.
The influence of SFR6 upon stress gene expression extends beyond the CBF-controlled regulon
We have demonstrated previously that SFR6 is an essential regulator of low-temperature gene expression leading to the control of freezing tolerance in Arabidopsis (Knight et al., 1999, 2009). Drought-responsive expression of the same (COR or cold-onregulated) genes is similarly affected (Knight et al., 1999; Boyce et al., 2003). These genes are controlled by the CBF (DREB1) and DREB2 families of transcription factors (TFs) (Liu et al., 1998; Shinwari et al., 1998; Haake et al., 2002; Gilmour et al., 2004). Expression of the CBF/DREB1 transcripts themselves are not affected by mutation of SFR6 (Knight et al., 1999; Boyce et al., 2003). In addition to dramatic effects upon the COR regulon, we observed a pleiotropic phenotype in sfr6-1, with physical appearance, flowering time and circadian clock behaviour all affected (Knight et al., 2008, 2009). We reasoned that it was unlikely that these additional phenotypes could be explained entirely through effects upon CBF/DREB function, indicating that SFR6 might play a role in the regulation of further gene regulons whose transcription is controlled by other TFs in addition to the CBF/DREB2 families.
Evidence to support this hypothesis was gained with the identification of SFR6 as At4g04920 (Knight et al., 2009), and discovery that this gene encodes a subunit of the Mediator complex (Bäckström et al., 2007). Mediator is a eukaryotic transcriptional co-activator complex consisting of c. 25–35 protein subunits; it takes information from positive and negative regulatory factors alighting on DNA elements and transmits it to RNA polymerase II (Conaway & Conaway, 2011). Mediator has been well described in yeast and higher eukaryotes but until the publication of data by Bäckström et al. (2007), had not been identified in plants, most probably due to the very low protein sequence homology between Arabidopsis and other eukaryotic Mediator subunits (Bäckström et al., 2007). At4g04920/SFR6 shows homology to yeast Med16/Sin4 (Bäckström et al., 2007; Bourbon, 2008), a component of the Mediator ‘tail’, the part of the complex considered most likely to interact with trans-acting factors (Bourbon, 2008). Phylogenetic studies support the suggestion At4g04920/SFR6 and Med16 are equivalents (Loncle et al., 2007; Bourbon, 2008).
As a component of Mediator, it might be expected that AtSFR6/MED16 could control other gene regulons in addition to those we have already described. PFT1, originally described as a regulator of flowering time (Cerdan & Chory, 2003), was identified recently as the MED25 subunit of Mediator (Bäckström et al., 2007) and has been shown to control transcriptional responses to plant pathogens (Bäckström et al., 2007; Kidd et al., 2009). We sought to discover whether SFR6 plays a similar role in regulating plant defence responses.
SFR6 is involved in the control of defence-related gene expression
We tested susceptibility of sfr6 mutant alleles to the virulent pathogen P. syringae DC3000 and discovered that three mutant alleles were all significantly more susceptible to infection than wild-type (Fig. 1). Given the identification of SFR6 as a subunit of Mediator, a likely reason for this was the failure of mutants to sufficiently upregulate genes required for the defence response against virulent biotrophs such as Pseudomonas. Whilst the expression of pathogenesis-related (PR) genes has not been proven as causal in tolerating virulent biotrophs, the PR genes are commonly used markers of SA-mediated immunity and their expression levels correlate with the level of tolerance plants exhibit. We tested, therefore, the transcriptional response of wild-type and sfr6 mutant plants to Pseudomonas infection, and saw that expression of the pathogenesis-related genes PR1, PR2 and PR5 was reduced by varying degrees compared with the levels seen in wild-type (Fig. 2). The requirement for SFR6 appears to be greatest in the case of PR5 induction and least in the case of PR1. It is possible that these differences could be attributed to the choice of time point at which expression was assayed; however, there are other potential explanations. One possible reason for the varying level of requirement for SFR6 could be the following: genes entirely dependent on activation by TFs that interact with Mediator via SFR6 solely may be those most affected in sfr6 mutants, whilst genes that can be activated via other tail subunits in addition to SFR6 are only slightly affected. Demonstration of direct interactions between TFs and MED subunits would be required to discover whether this is the case.
In response to the same pathogen challenge, OXI1 expression and the induction of two WRKY TF genes showed no requirement at all for SFR6 (Fig. 3). This suggests OXI1 expression in response to virulent P. syringae is likely to employ TFs that can activate RNA polymerase II-mediated transcription independently of the presence of SFR6/MED16 in the Mediator complex, most likely via (an)other Mediator subunit(s). Pathogenesis-related genes including PR1 are the targets of the WRKY18 TF (Chen & Chen, 2002); our data indicate that SFR6 is not required for the expression of WRKY18 itself but that it may act downstream of WRKY18 to effect activation of PR gene expression. A similar role for SFR6 acting downstream of the cold-responsive CBF TFs was revealed in our earlier study (Knight et al., 2009). Together, our results indicate that rather than being responsible for all transcriptional responses to a particular stress stimulus, SFR6/MED16 is required for activation of some regulons but not for others.
Expression of genes such as PR1 in response to biotrophic and hemibiotrophic pathogens is regulated via a salicylic acid (SA)-mediated pathway (Bari & Jones, 2009). As virulent Pseudomonas-inducible expression of PR1, PR2 and PR5 was reduced in sfr6 mutants, it appeared likely that SFR6 is required for activation of gene targets dependent on the SA-mediated pathway. We tested whether induction of pathogenesis-related gene expression in direct response to SA was compromised in sfr6 mutants and discovered that PR1 and PR2 expression was severely reduced (Fig. 4). These data support the suggestion that SFR6/MED16 may mediate the control of PR gene expression via the action of SA-dependent TFs. Interestingly, in the case of PR2 where basal expression levels were easily detectable it was apparent that these were influenced by the presence or absence of SFR6. This indicates that as in the case of cold-inducible gene expression (Knight et al., 2009), SFR6 is required for all transcriptional activity via this promoter and not only for effecting increases in transcription in response to stress/hormone treatments.
Having demonstrated that SA-mediated PR gene expression in response to virulent P. syringae requires SFR6, we examined the role of SFR6 in the transcriptional response to UV-C irradiation, which includes upregulation of many of the same genes (Nawrath et al., 2002). Whether or not transcriptional responses to UV-C irradiation employ the SA-mediated pathway has not previously been demonstrated; however, SA has been shown to accumulate in response to UV-C (Yalpani et al., 1994) and these increases in SA levels are necessary for some responses to UV irradiation including the promotion of flowering (Martinez et al., 2004). It has also been shown that SA promotes tolerance to UV; NahG-expressing Arabidopsis (with low SA levels) show increased sensitivity to UV (Besteiro et al., 2011) and application of SA to some plants has been shown to increase UV-induced antioxidant capacity (Mahdavian et al., 2008). We reasoned, therefore, that SA was likely to play a role in the activation of PR gene expression by UV-C and we tested this possibility. We observed that NahG plants overexpressing the bacterial enzyme salicylate hydroxylase (Delaney et al., 1994) did indeed show severe reductions in UV-C-induced PR1 expression (Supporting Information Fig. S1). As these plants cannot accumulate SA, our result supports the view that SA is involved in mediating this transcriptional response.
Like PR1, EDS5 is inducible by UV-C and requires SA (Nawrath et al., 2002). We discovered that UV-C-induced levels of PR1, EDS5 and CASPASE8 expression were significantly lower in three sfr6 mutant alleles than in Col-0 wild-type controls (Fig. 5). This demonstrated that SFR6/MED16 is required for full expression of pathogenesis-responsive genes in response to UV, and indicates that as in the case of induction by bacterial pathogens, this is likely to be a consequence of failure to fully respond to the SA-mediated pathway. The inability of sfr6 mutants to fully express the UV-responsive defence genes described above was reflected in poor tolerance of UV-C irradiation and limited recovery after exposure to it (Fig. 7). It is possible that these effects are mediated through SA-dependent gene expression. Our data show that MED16 is required for both PR gene expression and tolerance in response to UV-C. As we observed in response to Pseudomonas infection, not all UV-responsive gene expression required SFR6 and other genes were expressed to wild-type levels in sfr6 mutants (Fig. 6). The proteins encoded by these genes may be less likely, or insufficient, to promote tolerance to UV-C.
SFR6/MED16 also controls JA-responsive gene expression
Two major pathways, SA- and JA-mediated, lead to defence gene transcription (Bari & Jones, 2009). Whilst the SA-mediated pathway controls expression of genes such as PR1, which are induced in response to biotrophic and hemibiotrophic pathogens, the JA-mediated pathway is utilized by plants responding to necrotrophs such as Botrytis. The JA pathway controls the expression of other defence genes, notably those of the plant defensin (PDF) group (Manners et al., 1998; Brown et al., 2003). The SA- and JA-mediated pathways frequently act antagonistically to each other, meaning that increased susceptibility to necrotrophic pathogens is often accompanied by reduced susceptibility to biotrophs (Pieterse et al., 2009; Moore et al., 2011). MED25, originally described as PFT1 (PHYTOCHROME FLOWERING TIME 1), has been shown to play a role in JA-responsive signalling, with pft1 mutants deficient in JA- and Botrytis-induced defensin gene expression (Kidd et al., 2009). Interestingly, pft1 mutants are also deficient in their response to SA and show defective expression of PR1 and PR5, as we saw in sfr6 mutants (Kidd et al., 2009).
We sought to discover whether SFR6/MED16 might play a role in JA-responsive expression in addition to its role in the SA-mediated pathway. Our results revealed the potential for JA-inducible transcription to be severely compromised in mutants lacking SFR6. Expression of PDF1.2A and PDF1.1 was strongly induced by JA in the wild-type but no induction at all was detectable in any of the sfr6 mutant alleles at the time point used (Fig. 8). It has been shown recently that contrary to the prevailing view, JA- and SA-mediated defence pathways do not always act antagonistically to one another and each can have positive effects on immunity to both biotrophic and necrotrophic pathogens (Tsuda et al., 2009). Therefore, it is possible that the reduced flux through the JA signalling pathway that we observe in sfr6 mutants may be one of the contributors to reduced tolerance of P. syringae. The dramatic loss of JA-induced PDF1.2A and PDF1.1 expression we observed in sfr6 mutants indicated that expression of these plant defensin genes is completely dependent upon SFR6 for expression and suggests that under the conditions we used, the TF(s) employed to activate JA-inducible defensin gene expression rely solely upon SFR6/MED16. Previous work has shown that full JA-responsive expression is dependent upon MED25/PFT1 (Kidd et al., 2009). Our data may suggest the possibility that the presence of SFR6/MED16 is required for MED25 to be fully functional. Future research into the structure of plant Mediator will reveal whether MED25 is likely to require interaction with MED16 in order to make contact with the Mediator complex as a whole.
Given the strong dependence of JA-responsive plant defensin gene expression upon SFR6/MED16, we focused on this aspect of defence gene regulation for further investigation. The ERFs comprise a large gene family of AP2-type TFs known to control a variety of processes, including the response to plant pathogens (Onate-Sanchez et al., 2007). A number of ERFs have demonstrated roles in the regulation of JA-responsive defence gene expression (Lorenzo et al., 2003; Berrocal-Lobo & Molina, 2004; McGrath et al., 2005; Pre et al., 2008) and JA-responsive activation of the defensin gene PDF1.2A has been shown to occur via the CCG box, the cognate cis element for ERF TFs (Brown et al., 2003). ERF5 plays a positive role in wild-type Arabidopsis in the induction of JA-responsive genes such as PDF1.2A (Fig. 9). Given this property, ERF5 constituted an ideal candidate for testing the ability of sfr6 mutants to activate JA-responsive gene expression in response to constitutive expression of a transcription factor.
In an alternative study published whilst we were preparing this manuscript, we learned that other workers overexpressed ERF5 in Arabidopsis and recorded observations that conflict with our data; PDF1.2A expression was not amongst the target genes they found to be expressed in response to ERF5 overexpression (Son et al., 2011). We conclude that unknown differences in our experimental conditions or the age of plants used, may have contributed to our different observations. We overexpressed ERF5 in wild-type and sfr6-1 mutant plants and showed that whilst ERF5 is able to ectopically induce the expression of JA-responsive genes in the absence of a stimulus in wild-type, overexpression failed to have any effect in the sfr6-1 mutant background (Fig. 9). This result shows that the SFR6/MED16 subunit of Mediator is essential for ERF5-activated expression of PDF1.2A and PDF1.1 and indicates that, consistent with its role as a Mediator subunit, SFR6/MED16 acts downstream of ERF5. An alternative possibility is that ERF5 action requires the presence of a protein encoded by another gene whose expression depends upon SFR6. We are not aware of any such component that is required specifically by ERF5, but this prospect should be considered. It is quite possible that SFR6/MED16 effects transcriptional activation of defence genes in response to other (ERF) transcription factors in addition to ERF5. Furthermore, it is possible that SFR6 affects transcription of defence genes indirectly, through influencing the expression of the TFs themselves.
The role of SFR6 and other Mediator subunits in defence gene expression
Our results demonstrate that SFR6/MED16 plays an important role in regulating both SA-dependent and JA-controlled defence responses. Together with previous published work, our study indicates that specific aspects of this role are shared with other Mediator subunits including MED21 (Dhawan et al., 2009), MED8 and MED25 (Kidd et al., 2009), but that each is likely to exert different levels of influence on transcriptional regulation. A similar situation exists in yeast, where a number of tail subunits contribute to similar roles in transcriptional regulation and likely act in concert (Jiang et al., 1995; Li et al., 1995).
Its key role in cold- and drought-inducible gene transcription as well as in defence-related expression places SFR6/MED16 in a pivotal position which may facilitate cross-talk between abiotic and biotic stress-responsive pathways, and may indicate a fundamental role for this protein in Mediator function. Our results show that SFR6/MED16 is required for the expression of a subset of JA- and SA-responsive defence genes, but that SFR6/MED16 is not a general regulator of transcription in plants. One explanation for why the influence of SFR6 is limited to the control of particular defence genes may be that it is required for transcriptional activation by some, but not all, TFs. Future research will be required to test this possibility and may also reveal roles for alternative Mediator subunits in the expression of additional defence genes.
We thank Durham University’s Project Sri Lanka for funding D.L.W. (University of Ruhuna) and the Biotechnology and Biological Sciences Research Council for funding (grants BBD0091621 and BB/F01984X/1) and for a PhD studentship awarded to C.S.M. P.C. is grateful for the award of an Erasmus visiting studentship. We would like to thank Rebecca Lamb for excellent technical assistance, Dr Steve Chivasa for advice and help with Pseudomonas infections and Lindsay Petersen (University of Cape Town) for donating to us P.st. DC3000. We thank the Arabidopsis seed stock centre (NASC) for supplying seeds of sfr6-2 and sfr6-3 insertional mutants. We thank Johan Kroon (Durham University), Luis Mur (Aberystwyth University, UK), Anne Rehmany and Katherine Denby (Warwick HRI, UK) for gifts of seeds throughout the project. We thank Rob Ingle (University of Cape Town) for useful discussions and for sharing preliminary data.