Arabidopsis MAP kinase 4 regulates gene expression through transcription factor release in the nucleus

Jin‐Long Qiu, Berthe Katrine Fiil, Klaus Petersen, Henrik Bjørn Nielsen, Christopher J Botanga, Stephan Thorgrimsen, Kristoffer Palma, Maria Cristina Suarez‐Rodriguez, Signe Sandbech‐Clausen, Jacek Lichota, Peter Brodersen, Klaus D Grasser, Ole Mattsson, Jane Glazebrook, John Mundy, Morten Petersen

Author Affiliations

  1. Jin‐Long Qiu1,,
  2. Berthe Katrine Fiil1,,
  3. Klaus Petersen1,
  4. Henrik Bjørn Nielsen2,
  5. Christopher J Botanga3,
  6. Stephan Thorgrimsen1,
  7. Kristoffer Palma1,
  8. Maria Cristina Suarez‐Rodriguez1,
  9. Signe Sandbech‐Clausen1,
  10. Jacek Lichota4,
  11. Peter Brodersen5,
  12. Klaus D Grasser4,
  13. Ole Mattsson1,
  14. Jane Glazebrook3,
  15. John Mundy1 and
  16. Morten Petersen*,1
  1. 1 Department of Biology, University of Copenhagen, Copenhagen, Denmark
  2. 2 Center for Biological Sequence Analysis, Department of Systems Biology, Technical University of Denmark, Kongens Lyngby, Denmark
  3. 3 Department of Plant Biology and Center for Microbial and Plant Genomics, University of Minnesota, St Paul, MN, USA
  4. 4 Department of Life Sciences, Aalborg University, Aalborg, Denmark
  5. 5 Institut de Biologie Moléculaire des Plantes du CNRS, Unité Propre de Recherche 2357, 12 rue du Général Zimmer, Strasbourg Cedex, France
  1. *Corresponding author. Department of Biology, University of Copenhagen, Ole Maaloees Vej 5, Copenhagen 2200, Denmark. Tel.: +45 35322127; Fax: +45 35322128; E-mail: shutko{at}
  1. These authors contributed equally to this work


Plant and animal perception of microbes through pathogen surveillance proteins leads to MAP kinase signalling and the expression of defence genes. However, little is known about how plant MAP kinases regulate specific gene expression. We report that, in the absence of pathogens, Arabidopsis MAP kinase 4 (MPK4) exists in nuclear complexes with the WRKY33 transcription factor. This complex depends on the MPK4 substrate MKS1. Challenge with Pseudomonas syringae or flagellin leads to the activation of MPK4 and phosphorylation of MKS1. Subsequently, complexes with MKS1 and WRKY33 are released from MPK4, and WRKY33 targets the promoter of PHYTOALEXIN DEFICIENT3 (PAD3) encoding an enzyme required for the synthesis of antimicrobial camalexin. Hence, wrky33 mutants are impaired in the accumulation of PAD3 mRNA and camalexin production upon infection. That WRKY33 is an effector of MPK4 is further supported by the suppression of PAD3 expression in mpk4wrky33 double mutant backgrounds. Our data establish direct links between MPK4 and innate immunity and provide an example of how a plant MAP kinase can regulate gene expression by releasing transcription factors in the nucleus upon activation.


Plants have evolved a multi‐layered system of defence responses that can be activated upon recognition of invading pathogens. One layer includes transmembrane receptors that recognize evolutionarily conserved pathogen‐associated molecular patterns (PAMPs) and when activated trigger an immune response. Successful pathogens can deliver effectors that suppress the immune response and contribute to pathogen virulence (Jones and Dangl, 2006). Another layer involves recognition of pathogen effector molecules through host resistance (R) genes, triggering a rapid defence response that often includes a localized programmed cell death reaction known as the hypersensitive response (Nimchuk et al, 2003).

Recognition by animal and plant innate immune systems activates defence responses mediated by protein kinase signalling pathways (DeYoung and Innes, 2006). These pathways regulate the expression of numerous genes, including genes involved in the production of antimicrobial compounds (Fehlbaum et al, 1994; Schonwetter et al, 1995; Zhou et al, 1999; Couillault et al, 2004). For example, the expression of some antimicrobial genes in mammals and Drosophila requires members of the NF‐kappaB class of transcription factors (Meng et al, 1999; Diamond et al, 2000). Studies on Arabidopsis and other plants implicate MAP kinases and WRKY transcription factors in the regulation of genes required for pathogen resistance (Eulgem et al, 2000; Asai et al, 2002; Andreasson et al, 2005; Journot‐Catalino et al, 2006; Xu et al, 2006). The genome of the model plant Arabidopsis encodes more than 20 MAP kinases, including MPK3, MPK4 and MPK6 implicated in innate immunity responses (Petersen et al, 2000; Asai et al, 2002; Menke et al, 2004). WRKY proteins constitute a large family of transcription factors in plants. They bind W‐box sequences in the promoters of pathogen‐induced genes, including WRKY genes themselves (Rushton and Somssich, 1998; Eulgem et al, 2000). Direct transcriptional targets have been suggested for several WRKY factors (Robatzek and Somssich, 2002), but demonstrated only for parsley PcWRKY1 (Turck et al, 2004). WRKY proteins have also been linked to MAP kinase (MAPK) cascades in Arabidopsis: WRKY22 and WRKY29 are thought to function downstream of the flagellin receptor FLS2 in a pathway that includes the MAP kinase components MEKK1, MKK4/MKK5 and MPK3/MPK6 (Asai et al, 2002). However, molecular evidence directly linking an MAP kinase to a WRKY factor and its target gene(s) has not been reported.

MPK4 has been proposed to function in a cascade(s) that includes the MAP kinase kinases MKK1 and MKK2 and the MAPK triple kinase MEKK1 (Ichimura et al, 1998). Abiotic stresses and the bacterial elicitors flagellin and harpin activate MPK4 (Teige et al, 2004; Suarez‐Rodriguez et al, 2007). These results appear in contrast to other reports indicating that MPK4 functions as a negative regulator of defence responses. For example, loss‐of‐function mpk4 mutants have elevated levels of the hormone salicylic acid (SA), accumulate pathogenesis‐related transcripts, and have increased resistance towards biotrophic pathogens, including Pseudomonas syringae and Hyaloperonospora parasitica, but increased susceptibility towards the necrotrophic pathogen Alternaria brassicicola (Petersen et al, 2000; Brodersen et al, 2006). Attempts to complement mpk4 mutants with kinase‐inactive forms have led to the conclusion that MPK4 activity is needed to suppress SA‐dependent defence responses and supports the model that MPK4 is a negative regulator (Brodersen et al, 2006). Recently, MKS1 was identified as a MPK4 substrate, and analyses of transgenic plants and transcript profiling indicated that MKS1 is required for full resistance in mpk4 mutants. Two transcription factors WRKY25 and WRKY33 interact with MKS1 in yeast, suggesting that these two WRKYs regulate gene expression downstream of MPK4 (Andreasson et al, 2005).

wrky33 loss‐of‐function mutants support normal growth of virulent P. syringae, but transgenic plants that overexpress WRKY33 support enhanced growth of P. syringae (Zheng et al, 2006, 2007). In addition, wrky33 mutants exhibit enhanced susceptibility to Botrytis cinerea and A. brassicicola, whereas overexpression of WRKY33 increases resistance to these necrotrophs (Zheng et al, 2006). This suggests that WRKY33 activates genes encoding proteins required to efficiently combat necrotrophic pathogens. An example of such a gene is PHYTOALEXIN DEFICIENT3 (PAD3), which encodes cytochrome P450 monooxygenase 71B15 (Zhou et al, 1999). PAD3 is required in the last step of the synthesis of the antimicrobial compound camalexin, and pad3 mutants exhibit enhanced susceptibility to A. brassicicola (Thomma et al, 1999; Schuhegger et al, 2006).

We report here that MPK4 and MKS1 associate with WRKY33 in vivo. Infection leads to the activation of MPK4 and phosphorylation of MKS1. Subsequently, MKS1 and WRKY33 are released from MPK4, and WRKY33 is recruited to the promoter of PAD3. We show that PAD3 mRNA accumulation in response to infection is greatly reduced in wrky33 mutants, and that WRKY33 is an effector of expression from the PAD3 promoter in reporter gene assays. We propose how MPK4 regulates PAD3 expression through WRKY33 upon pathogen‐induced activation.

Results and discussion

Putative target genes of WRKY33

To understand how MPK4 regulates gene expression, we exploited our previous findings that MPK4 and WRKY33 may function together to regulate specific immune responses. Expression profiling was performed using ATH1 GeneChips (Affymetrix) to screen for putative target genes of WRKY33. To this end, triplicate mRNA samples of wrky33 mutants and wild‐type (Col‐0) plants harvested before and 24 h after treatment with the SA analogue benzothiadiazole (BTH) were compared.

To enrich for putative WRKY33 target genes, transcripts accumulating in response to BTH treatment in wild type, but not in wrky33, were identified (Supplementary Table S1). Only a few genes failed to accumulate properly in BTH‐treated wrky33. Of 29 transcripts initially identified, only 4 showed dramatic differences between wild type and wrky33. These four included PAD3 and CYP71A13, which are strongly co‐regulated and both are required for the synthesis of the phytoalexin camalexin (Zhou et al, 1999; Schuhegger et al, 2006; Nafisi et al, 2007). In addition, NUDT6 and a peptidylprolyl isomerase (PPIase) failed to accumulate in wrky33. The expression of NUDT6 has been found to be dependent on the disease regulators EDS1 and PAD4 in RPM1‐conditioned responses (Bartsch et al, 2006). PPIase, a member of the cyclophilin family, has not been linked to resistance responses previously.

To validate these WRKY33 targets, real‐time PCR was performed on BTH‐treated Col‐0 and wrky33. The mRNAs of all four genes failed to accumulate normally in wrky33 upon BTH treatment (Figure 1A). As loss of MPK4 function leads to increased levels of SA (Petersen et al, 2000) but WRKY33 could function downstream of MPK4, yet upstream of SA. Therefore, we decided to broaden this analysis, and confirm the expression of the co‐regulated PAD3 and CYP71A13 genes in a more biologically relevant context. This analysis revealed that the two genes failed to accumulate to wild‐type levels in wrky33 treated with flagellin or locally infected with Pst DC3000 or Pst DC3000 expressing the avrRpm1 effector that triggers resistance through the host resistance gene RPM1 (Figure 1B). PAD3 and CYP71A13 mRNA levels peaked 2‐ to 4‐h after flagellin treatment and in plants infected with virulent Pst DC3000. In wild‐type plants challenged with Pst DC3000 avrRpm1, the initial increase in PAD3 and CYP71A13 mRNA levels was also observed, but 4 h post‐infection PAD3 and CYP71A13 mRNA levels were five‐fold higher and again, this increase was not observed for wrky33 (Figure 1B). These results indicate that WRKY33 is required for the accumulation of PAD3 and CYP71A13 mRNAs in response to flagellin and bacterial pathogens in the very early stages of infection. In addition, WRKY33 also seems to be required for further enhancement of the expression of these two genes upon R‐gene activation.

Figure 1.

Real‐time PCR quantitation of PAD3, CYP71A13, NUDT6 and PPIase mRNAs. (A) Expression in Col‐0 and wrky33 before or 24 h after treatment with BTH. (B) Expression in Col‐0 and wrky33 before or after treatment with flg22, Pst DC3000 or Pst DC3000 (avrRpm1) for 30 min, 1, 2 and 4 h. The samples were tested in triplicate and normalized to ACTIN2. Means±s.d. are shown. d.p.i.: days post‐infection. h.p.i.: hours post‐infection.

WRKY33 binds to and activates transcription from the PAD3 promoter

Genes whose expression is not induced in wrky33 upon BTH treatment, flagellin or infection could represent direct and indirect WRKY33 targets. To examine whether PAD3 or CYP71A13 could be directly regulated by WRKY33, we assayed the association of WRKY33 with putative promoter regions of these genes by chromatin immunoprecipitation (ChIP). The PAD3 promoter contains four sequences corresponding to the core and extended WRKY‐binding site or W‐box (TTGAC and TTGACC/T; Eulgem et al, 2000) at positions −1109, −1015, −555 and −388 upstream of the PAD3 transcription start site. A primer combination that amplified an 88‐bp fragment spanning the W‐box at −555 could repeatedly amplify genomic DNA (P1; Supplementary Table S2). As the sonication procedure sheared the ChIP DNA to ∼500‐bp fragments, this primer combination would amplify regions of the proximal promoter containing the W‐boxes at −388, −555 and possibly one or both of the more 5′ upstream boxes. We therefore assayed whether genomic DNA fragments immunoprecipitated with WRKY33 could be amplified by this primer pair by real‐time PCR. Interestingly, DNA sequences from the promoters of PAD3 and CYP71A13 were not recovered from extracts of untreated tissue (CYP71A13 data not shown). However, the amount of PAD3 promoter DNA recovered from wild‐type Col‐0 treated with flagellin for 1.5 and 2 h was higher than PAD3 promoter DNA recovered from mock and wrky33 immunoprecipitates (Figure 2A; data not shown). Next, we performed ChIP with anti‐WRKY33 on wild‐type Col‐0 4 h post‐infection with Pst DC3000 expressing avrRpm1 or untreated control plants. PAD3 promoter DNA was not amplified from DNA isolated from untreated plants, but was readily amplified from pathogen‐infected plants (Figure 2B). DNA from the PAD3 promoter was not amplified from wrky33‐infected tissue (Figure 2B), again demonstrating the specificity of the anti‐WRKY33 antibody. These results provide direct evidence that WRKY33 is recruited to the promoter of PAD3 in vivo in response to flagellin treatment or when plants are infected with a pathogen. In addition, it demonstrates that the presence of WRKY33 on the PAD3 promoter correlates with the abundance of PAD3 mRNA (Figure 1B).

Figure 2.

WRKY33 binding to and activation from the PAD3 promoter. (A, B) Enrichment of P1 amplicons from chromatin immunoprecipitates with anti‐WRKY33 antibodies from regions of the PAD3 promoter over background (measured with primers to a promoter region of the ROC5 control gene) in Col‐0 and wrky33 treated with flg22 for 1.5 h (A) and after treatment with Pst DC3000 (avrRpm1) for 4 h (B). The samples were tested in triplicate and normalized to ROC5. Means±s.d. are shown. (C) Transient expression assays using a 2.4‐kb fragment of the PAD3 promoter fused to the GUS reporter gene (PAD3:GUS) inoculated with Pst DC3000 (avrRpm1). Bombardments of detached leaves of Col‐0 and wrky33 mutant were done with PAD3:GUS in combination with an empty vector or an effector of the CaMV 35S promoter driving WRKY33 or WRKY22 full‐length cDNA. A 35S:LUC construct was used as a control for the efficiency of transient transformation. GUS activity in each sample was expressed relative to LUC activity to normalize data for variation in transformation efficiency.

To confirm the significance of WRKY33 binding to the PAD3 promoter, we used a transient gene expression assay in leaves to monitor WRKY33‐dependent gene expression from the PAD3 promoter. Co‐bombardments of a PAD3 promoter fusion to the GUS reporter gene (PAD3:GUS) together with a CaMV 35S:WRKY33 effector plasmid in wrky33 mutants resulted in strong GUS activity after Pst DC3000 (avrRpm1) infection, whereas the PAD3:GUS construct with empty vector only resulted in very weak GUS activity (Figure 2C). As mentioned in the introduction, WRKY22 functions downstream of the flagellin receptor, FLS2, to activate early defence genes (Asai et al, 2002). Unlike WRKY33, ectopically expressed WRKY22 did not activate PAD3 promoter‐driven GUS expression (Figure 2C), suggesting that PAD3 is not a general target of pathogen‐responsive WRKYs. This result demonstrates that WRKY33, and not the related WRKY22, can activate gene expression from the PAD3 promoter upon infection.

WRKY33 is likely to regulate a number of genes other than PAD3. As shown above, CYP71A13 may also be subject to regulation by WRKY33. Although initial ChIP experiments have not detected WRKY33 binding to its upstream regions, the promoter of CYP71A13, similar to that of PAD3, may well be a WRKY33 target. This is because CYP71A13 and PAD3 are tightly co‐regulated, both are required for the synthesis of camalexin (Nafisi et al, 2007) and the CYP71A13 promoter also contains a WRKY‐binding site at position −286 upstream of the CYP71A13 transcription start site. Other genes from our initial screen, such as NUDT6, could also represent WRKY33 target candidates downstream of MPK4. NUDT6 expression requires WRKY33 upon infection with strains of P. syringae (data not shown) as well as EDS1 and PAD4, whose loss of function suppresses the mpk4 phenotype (Bartsch et al, 2006; Brodersen et al, 2006). It is likely that our transcript profiling detected only a subset of WRKY33 target genes as PAD3 and CYP71A13 were induced much more strongly by flagellin and pathogen infections than by BTH. Therefore, it may be possible to identify additional WRKY33 targets by comparing transcript profiles of flagellin‐treated and untreated wild‐type and wrky33 plants. Similarly, it may be useful to compare wrky33 and mpk4 single mutants with the double wrky33/mpk4 mutant (below) to identify targets directly regulated by WRKY33 through the MPK4 pathway.

WRKY33 is required for camalexin synthesis after pathogen attack

To study the role of WRKY33 in the control of camalexin production, camalexin levels were determined in Col‐0 and wrky33 mutant plants 24 and 48 h after infection with Pst DC3000 or Pst DC3000 expressing avrRpm1. This showed that these pathogens induce the production of camalexin in Col‐0 but not in wrky33 mutants (Figure 3A). However, the camalexin‐deficient mutant pad3 does not exhibit altered susceptibility to these pathogens (Glazebrook and Ausubel, 1994), but is markedly more susceptible to infection by the necrotrophic fungus A. brassicicola (Thomma et al, 1999) that elicits synthesis of camalexin in Arabidopsis (Nafisi et al, 2007). We therefore used this pathogen to examine the role of WRKY33 in disease resistance. First, PAD3 mRNA levels were determined in Col‐0 and wrky33 mutant plants 24 h after infection with A. brassicicola (Figure 3B); PAD3 expression was strongly induced in wild‐type but not wrky33 plants. Similar results were obtained for CYP71A13 (Supplementary Figure S1). The effect of reduced PAD3 and CYP71A13 expression on camalexin synthesis after infection was assessed by camalexin measurements. After infection with A. brassicicola, camalexin levels in wrky33 plants were much lower than in wild type (Figure 3C). Thus, the disease phenotypes of wrky33 mutants after A. brassicicola infection were similar to those of pad3 (Supplementary Figure S2). Taken together, these results indicate that WRKY33 controls the production of camalexin in response to infection by activating expression of genes encoding camalexin biosynthetic enzymes.

Figure 3.

Levels of PAD3 mRNA and camalexin in Col‐0 and wrky33 plants after infection with pathogens. (A) Camalexin levels after inoculation with Pst DC3000 or Pst DC3000 (avrRpm1). (B) Real‐time PCR detection of PAD3 mRNA after inoculation with A. brassicicola. The samples were tested in triplicate and normalized to ACTIN2. (C) Camalexin levels after inoculation with A. brassicicola. Means±s.d. are shown. d.p.i.: days post‐infection. h.p.i.: hours post‐infection.

WRKY33 is released from complexes with MPK4 upon infection

We previously showed that WRKY33 interacts in yeast and in vitro with the MPK4 substrate MKS1 (Andreasson et al, 2005). To extend these findings in vivo, the presence of WRKY33 was demonstrated in immunoprecipitates from nuclear extracts using an anti‐MKS1 antibody (Figure 4A, top). The specificity of the MKS1 pull‐down was confirmed by the absence of WRKY33 in immunoprecipitates from an mks1 loss‐of‐function mutant. This mks1 transposon insertion line (GT.108403) fails to accumulate MKS1 mRNA or protein (data not shown). Nuclear extracts were used for co‐immunoprecipitation because WRKY33 and MKS1 protein levels are very low in total cellular protein extracts and because both proteins are nuclear‐localized (Andreasson et al, 2005; Zheng et al, 2006). In a reciprocal experiment, MKS1 was detected in immunoprecipitates prepared using an anti‐WRKY33 antibody (Figure 4A, bottom). The specificity of the WRKY33 pull‐down was confirmed by the absence of MKS1 in immunoprecipitates from the wrky33 loss‐of‐function mutant. These results demonstrate that WRKY33 and MKS1 interact in vivo, and are consistent with our earlier demonstration that MKS1 and MPK4 also interact in vivo (Andreasson et al, 2005).

Figure 4.

Complexes between WRKY33, MKS1 and MPK4. (A) Co‐immunoprecipitation of WRKY33 and MKS1 from nuclear extracts from: Bottom, Col‐0 and wrky33 mutant immunoprecipitated with anti‐WRKY33 and stained (WB) with anti‐MKS1; Top, Ler and mks1 mutant immunoprecipitated with anti‐MKS1 and stained with anti‐WRKY33. One‐tenth of inputs was stained with the same antibodies as a loading control. IgH: Ig heavy chain. (B) Top, co‐immunoprecipitation of WRKY33 and MPK4 before and after treatment of Pst DC3000 (avrRpm1). Nuclear extracts of transgenic plants expressing MPK4‐HA in the mpk4 background were immunoprecipitated with anti‐WRKY33 and stained with anti‐HA. Wild‐type Ler plants were used as a negative control. One‐tenth of input before and after the immunoprecipitation was stained with anti‐HA as a control. Infected samples were harvested after 4 h. Bottom, co‐immunoprecipitation performed as above for WRKY33 and MPK4 in Ler, mks1 and wrky33.

To functionally link MPK4 to WRKY33, we examined whether MPK4 is found in complexes with WRKY33 in vivo by assaying for the presence of MPK4 in anti‐WRKY33 immunoprecipitates from nuclear extracts. Interestingly, MPK4 was readily detected in anti‐WRKY33 immunoprecipitates from uninfected leaves (Figure 4B). In contrast, MPK4 was barely detectable in anti‐WRKY33 immunoprecipitates from leaves treated with flagellin or infected with Pst DC3000 expressing avrRpm1 (Figure 4B, top). This was apparently not due to a reduction in the levels of MPK4 protein in nuclei, as MPK4 was readily detected in the supernatant of these induced extracts following immunoprecipitation of WRKY33 (Figure 4B, second panel). MPK4 and WRKY33 do not interact in yeast (Andreasson et al, 2005). This could imply that MPK4 and WRKY33 exist in a complex that depends on MKS1. Thus, we looked for the presence of an MPK4–WRKY33 complex in the mks1 mutant. This showed that MKS1 is indeed required for such a complex, as we were repeatedly unable to pull down MPK4 with anti‐WRKY33 in mks1 plants (Figure 4B, bottom). Therefore, it is most likely that MKS1 forms a ternary complex with MPK4 and WRKY33 and that such complexes might sequester or regulate WRKY33.

MKS1 is needed to fine‐tune PAD3 expression

The release of WRKY33 from MPK4 and/or MKS1 could be mediated by changes in the activity of MPK4 followed by phosphorylation of MKS1. PAMPs such as flagellin and virulent Pst DC3000 activate MPK4 (Brader et al, 2007; Suarez‐Rodriguez et al, 2007). Infections with both Pst DC3000 and Pst DC3000 expressing the avrRpm1 effector dramatically increased MPK4 activity (Figure 5A). In addition, infection with type III secretion‐defective and coronatine‐deficient Pst DC3118 COR hrpS, which cannot deliver effectors into the host, lead to strong and sustained MPK4 activity (Figure 5A). Collectively, these data support a model in which MPK4 is activated by PAMPs. In addition, effectors delivered by the pathogens do not significantly contribute to MPK4 activation, indicating that MPK4 is primarily engaged in PAMP‐induced defence responses.

Figure 5.

Effects of infection on MPK4 activity and MKS1 protein, and of mks1 mutation on PAD3 mRNA levels. (A) Kinase activity of HA‐tagged MPK4 immunoprecipitated from total protein extracts was assayed using 32P‐labelled ATP and myelin basic protein (MBP) as substrate. h.p.i., hours post‐infection. (B) Plants were treated with Pst DC3000 (avrRpm1) for 4 h. Total nuclear extracts in the presence of 10 μM MG‐132 were resolved on 12% SDS–PAGE, blotted and stained with anti‐MKS1. Nonspecific band was used as a loading control. (C) Co‐immunoprecipitation of MKS1 and WRKY33 4 h after P. syringae treatments. Nuclear extracts from untreated Ler and plants treated with Pst DC3000 (avrRpm1) were immunoprecipitated with anti‐WRKY33 and stained with anti‐MKS1. (D) Real‐time PCR quantitation of PAD3 mRNA in Ler and mks1 mutant before or after treatment with Pst DC3000 (avrRpm1) for 4 and 21 h. (E) Camalexin levels after inoculation with Pst DC3000 or Pst DC3000 (avrRpm1).

Next, we examined the phosphorylation status of MKS1. Two MKS1 forms are detectable by SDS–PAGE; the unphosphorylated form with a higher mobility predominates in uninduced plants (Figure 5B). Infections with Pst DC3000 expressing avrRpm1 or treatments with flagellin converted a portion of the faster migrating form of MKS1 into a lower mobility form caused by phosphorylation (Figure 5B; Supplementary Figure S3A). This suggests that activation of MPK4 followed by phosphorylation of MKS1 leads to the release of WRKY33 from complexes with MPK4 upon infection. To examine whether MKS1 remains in a complex with WRKY33 after its release from MPK4, we co‐precipitated MKS1 with anti‐WRKY33 after infection. Interestingly, MKS1 was detected in the immunoprecipitates both before and after infection (Figure 5C). Furthermore, MKS1 was not detectable in the supernatants of these extracts following immunoprecipitation, suggesting that MKS1 mostly exists in complexes with WRKY33 (Figure 5C). This assumption is supported by our inability to detect MPK4–MKS1 complexes after flagellin treatment (Supplementary Figure S3B). That MKS1 and WRKY33 are associated in plants both before and after infections suggest the MKS1–WRKY33 complex exists independently of MKS1 phosphorylation. This conclusion is supported by a pull‐down experiment demonstrating the existence of MKS1–WRKY33 complexes both before and after phosphatase treatments (Supplementary Figure S3C), and that phospho‐mimics, non‐phosphorylatable and wild‐type forms of MKS1 bind WRKY33 equally well in yeast (see Supplementary data, and data not shown). Interestingly, MPK4 also interacts equally well with these forms of MKS1 in yeast (data not shown). One model to explain this observation could be that other, as‐yet unidentified factor(s) bind the MKS1–WRKY33 complex when MKS1 becomes phosphorylated facilitating MPK4 release. Nevertheless, to test the relevance of a MKS1–WRKY33 complex for WRKY33 function, we examined whether MKS1 is required for the full pathogen‐induced expression of PAD3. As seen in Figure 5D, the accumulation of PAD3 mRNA is lower in the early stages and higher at later stages of infection in mks1 mutants compared with wild type. This suggests that although WRKY33 continues to activate the expression of PAD3, a binary complex between MKS1 and WRKY33 may optimize this WRKY33 function. In this context, we note that the altered expression of PAD3 in mks1 mutants is insufficient to affect camalexin production (Figure 5E).

PAD3 expression in mpk4 requires WRKY33 but not SA

A large number of defence‐related genes, including PAD3, are constitutively expressed in mpk4 loss‐of‐function mutants. If WRKY33 functions directly downstream of MPK4, PAD3 should be expressed to comparable levels in mpk4 and in mpk4 SA‐deficient backgrounds. To this end, we compared the levels of PAD3 in mpk4 and mpk4 transgenic plants expressing the SA hydroxylase NahG. This analysis revealed that PAD3 is de‐repressed in mpk4 independently of SA (Figure 6A). Nonetheless, the elevation of PAD3 mRNA in mpk4 and mpk4–NahG is insufficient to result in a detectable increase in camalexin levels (data not shown), indicating that MPK4‐independent factor(s) are also required to produce camalexin (Ren et al, 2008). To extend this analysis, we introduced the loss‐of‐function wrky33 allele into the mpk4 mutant, and confirmed double mutant progeny using PCR (data not shown). In contrast to mpk4 single mutants that accumulate PAD3 to higher levels than wild type, double homozygous mpk4–wrky33 plants did not (Figure 6B). On the other hand, the wrky33 mutation did not significantly change the accumulation of PR1 messenger in mpk4 (data not shown). Furthermore, wrky33 partially suppressed the mpk4 phenotype (Figure 6C). These results provide genetic evidence that PAD3 is a direct target regulated by WRKY33 downstream of the MPK4 pathway independently of the accumulation of SA. Because the wrky33 mutation only partially rescued mpk4 phenotypes, WRKY33 is presumably not the only effector downstream of MPK4.

Figure 6.

WRKY33 functions downstream of MPK4. (A) Expression of PAD3 in Ler, mpk4 and mpk4–nahG. (B) Effect of wrky33 mutation on constitutive PAD3 expression in mpk4. Samples were tested in triplicate and normalized to ACTIN2. Means±s.d. are shown. (C) Partial rescue of the mpk4 dwarf phenotype by wrky33 mutation.

In summary, we show here that MPK4 functions as a regulator of PAMP‐induced defence responses in Arabidopsis. MPK4 is activated in response to PAMPs and, similar to other MAP kinases, presumably phosphorylates effector proteins that directly or indirectly regulate a spectrum of responses. More specifically, changes in MPK4 activity and phosphorylation of MKS1 on multiple sites (Caspersen et al, 2007), somehow triggers the release of WRKY33 from nuclear complexes with MPK4 to permit WRKY33 to activate the expression of target genes. These targets include PAD3 whose expression leads to the production of antimicrobial camalexin. These findings may also explain the genetic evidence that MPK4 functions as a negative regulator of basal resistance. When MPK4 is absent or dysfunctional in mpk4 mutants, WRKY33 and presumably other transcription factors are not sequestered in nuclear complexes with the kinase in the absence of pathogens. Instead, they inappropriately activate gene expression. This mode of action is different from that reported for many other MAP kinases and their transcription factor substrates in which the latter translocate to the nucleus following phosphorylation by the kinase in the cytoplasm (Wilkinson and Millar, 1998; Djamei et al, 2007).

Materials and methods

Plant growth conditions and treatments

Plants were grown in chambers with 8 h light/16 h darkness at 75% RH and 22°C. The wrky33 mutant was wrky33‐2 (GABI_324B11; Zheng et al, 2006). BTH (100 μM; Bion 50WG, 50% active ingredient) was sprayed onto leaves. Bacterial infections and growth assays with Pst DC3000 and related strains were described previously (Mackey et al, 2003). The plants for camalexin assay with Pst DC3000, Pst DC3000 avrRpm1 and A. brassiciola were grown in chambers with 12 h light/12 h darkness at 75% RH and 22°C. The plants were syringe‐inoculated with Pst DC3000 or Pst DC3000 avrRpm1 at a dose of OD600=0.01 as described by Parisy et al (2007). For ChIP infections, seedlings were vacuum infiltrated with Pst DC3000 avrRpm1. A. brassicicola infections of 3‐week‐old plants with strain ATCC 96866 were described previously (Nafisi et al, 2007). Camalexin was determined according to Glazebrook and Ausubel (1994). Arabidopsis plants were treated with a 10 μM flg22/0.001% Silwett solution. For ChIP, plants were grown on MS plates, transferred to liquid MS media for 24 h and flg22 was added to a final concentration of 10 μM.

RNA isolation and quantitative RT–PCR

Total RNA was extracted with Tri‐Reagent RiboPure kit (Ambion, Austin, TX, USA). For quantitative PCR, RNAs were treated with RQ1 DNase (Promega, Madison, WI, USA). Quantitative RT–PCR used Superscript III Platinum SYBR Green One‐Step qRT–PCR kit (Invitrogen, Carlsbad, CA, USA) with 10 pmol of each primer and 100 ng total RNA in 20 μl. Except for experiments in Figure 3 and Supplementary Figures S1 and S2, reactions were run on an icycler IQ (Bio‐Rad, Hercules, CA, USA). Quantitative PCR reactions were performed in triplicate for each individual line, and quantification of threshold cycle (CT) values was achieved by calculating means of normalized expressions using Q‐gene software (Muller et al, 2002). For Figure 3 and Supplementary Figures S1 and S2, reactions were run on an Applied Biosystems 7500 Real Time PCR. Primers for real‐time PCR are in Supplementary Table S2. In all figures except Supplementary Figure S2B, qPCR result values displayed are relative to wild‐type untreated plants, which are set to a relative value of 1.


Mouse monoclonal antibodies were against the HA epitope (clone 12CA5; Roche) and against MKS1 (HYB 330‐01; Statens Serum Institut, Denmark). Rabbit polyclonal antibodies against WRKY33 peptides (CEPEAGKRWKGDNETNG and CQEQQKKNQSEQWSQT) were made by Eurogentec (Belgium). MPK4 was immunoprecipitated using anti‐MPK4 (Sigma).

Western blot and kinase assays

Samples were heated for 10 min at 100°C and subjected to 12 or 15% SDS–PAGE. Electrophoresed proteins were transferred by semi‐dry electrophoresis to cellulose membranes after standard protocols. Membranes were incubated with primary antibodies (1:3000) and then with horse radish peroxidase‐conjugated secondary antibodies (Dako A/S, Denmark, 1:20 000) for the detection of immunoreactive bands with SuperSignal West Pico Chemiluminescent Substrate (Pierce Biotechnology, Rockford, USA). Kinase assays were performed as described previously (Brodersen et al, 2006).

Nuclear protein extraction and immunoprecipitation

Nuclei were isolated according to Gendrel et al (2002) and Nelson et al (2006). Nuclei were resuspended in co‐immunoprecipitation buffer (100 mM Tris–Cl, 75 mM NaCl, 1 mM EDTA, 0.1% Triton X‐100, 0.05% SDS, 10% glycerol, 2.5 mM DTT, proteinase and phosphatase inhibitors), and sonicated. Benzonase (10 U) was added and samples were then incubated on ice for 1 h, and centrifuged at 16 000 g for 30 min. The supernatants were incubated with antibodies overnight at 4°C followed by precipitation with protein A–Sepharose CL‐4B or protein G‐agarose beads (GE Healthcare) for 2 h. After washing four times with co‐immunoprecipitation buffer, proteins were eluted by boiling in 40 μl Laemmli loading buffer for 10 min. The samples were then processed by western blotting.


Leaves of 3‐ to 4‐week‐old wild‐type (Col‐0) and wrky33 plants grown in soil, or on MS plates were vacuum‐infiltrated with formaldehyde crosslinking solution and ChIP performed as described (Gendrel et al, 2002; Johnson and Bresnick, 2002). Samples were sonicated (Virsonic 600; Virtis) 6 × 10 s pulses at level 3, with 10‐s breaks. Real‐time PCR was used to quantify the immunoprecipitated DNA, and the quantity of PAD3 promoter normalized to nonspecific DNA levels (ROC5 promoter; Chou and Gasser, 1997). Primers for real‐time PCR are in Supplementary Table S2.

Transient gene expression assay

The 2424 bp 5′ upstream region of PAD3 was amplified by PCR and introduced in the BamHI and NcoI sites of a GUS reporter construct (Raventos et al, 1998). For the CaMV 35S‐driven WRKY33 or WRKY22 constructs, WRKY33 or WRKY22 was PCR amplified and introduced into the BamHI and StuI sites of p35SE9 (Andreasson et al, 2005).

Leaves from 3‐ to 4‐week‐old, short‐day grown wild‐type or wrky33 plants were bombarded 1–2 h after detachment as previously described (Leah et al, 1994). Each bombardment used ∼1 mg of 1 μm gold particles coated with 1 μg of PAD3‐1:GUS reporter, 1 μg of empty vector or WRKY33 or WRKY22 effector, and 1 μg of 35S:LUC reference plasmids. LUC expression was used to normalize transfection efficiency. Bombardments were at 2 bar with a distance of 10 cm using a Particle Inflow Gun. Bombarded leaves were incubated for about 23 h under long‐day conditions, vacuum infiltrated with Pst DC3000 (avrRpm1) at 1 × 105 c.f.u./ml and incubated for another hour. The GUS activity in each sample was expressed relative to LUC activity to normalize data for variation in transformation efficiency according to Leah et al (1994) using a Wallac Victor II fluorometer.

Supplementary data

Supplementary data are available at The EMBO Journal Online (

Supplementary Information

Supplementary Information [emboj2008147-sup-0001.pdf]


We thank Suksawad Vongvisuttikun for technical assistance. This study was supported by the Danish Research Councils to MP (23020101) and JM, (23030076, 272050367 and 272060049), and by the US Department of Energy to JG (DE‐FG02‐05ER15670).