Dual control of MAPK activities by AP2C1 and MKP1 MAPK phosphatases regulates defence responses in Arabidopsis

Abstract Mitogen-activated protein kinase (MAPK) cascades transmit environmental signals and induce stress and defence responses in plants. These signalling cascades are negatively controlled by specific Ser/Thr protein phosphatases of the type 2C (PP2C) and dual-specificity phosphatase (DSP) families that inactivate stress-induced MAPKs; however, the interplay between phosphatases of these different types has remained unknown. This work reveals that different Arabidopsis MAPK phosphatases, the PP2C-type AP2C1 and the DSP-type MKP1, exhibit both specific and overlapping functions in plant stress responses. Each single mutant, ap2c1 and mkp1, and the ap2c1 mkp1 double mutant displayed enhanced stress-induced activation of the MAPKs MPK3, MPK4, and MPK6, as well as induction of a set of transcription factors. Moreover, ap2c1 mkp1 double mutants showed an autoimmune-like response, associated with increased levels of the stress hormones salicylic acid and ethylene, and of the phytoalexin camalexin. This phenotype was reduced in the ap2c1 mkp1 mpk3 and ap2c1 mkp1 mpk6 triple mutants, suggesting that the autoimmune-like response is due to MAPK misregulation. We conclude that the evolutionarily distant MAPK phosphatases AP2C1 and MKP1 contribute crucially to the tight control of MAPK activities, ensuring appropriately balanced stress signalling and suppression of autoimmune-like responses during plant growth and development.


Introduction
Reversible protein phosphorylation is one of the most commonly used mechanisms for the molecular transmission of stress signals and developmental cues. This mechanism is based on the opposing actions of protein kinases and protein phosphatases. Mitogen-activated protein kinases (MAPKs) are highly conserved major components of developmental and stress signalling cascades in eukaryotes. MAPKs are activated by upstream MAPK kinases via phosphorylation of Thr and Tyr residues within their activation loop. This activation eventually leads to the reprogramming of cellular activities, including the modulation of gene expression, to generate appropriate responses. The activation of MAPKs does not represent a simple on/off switch, as both the magnitude and duration of activation are crucial for determining the signalling outcome (Marshall, 1995). Prolonged or constant activation of a MAPK cascade can have detrimental effects, as illustrated by the hypersensitive response-induced cell death in plants expressing a constitutively active version of MAPK kinase (Ren et al., 2002;. Thus, negative regulation and inactivation mechanisms are important for the correct cellular response. Specific protein phosphatases can dephosphorylate and thereby inactivate MAPKs. As dual phosphorylation of the Thr-X-Tyr motif in the activation loop is required for MAPK activation (Caunt and Keyse, 2013), dephosphorylation of either phospho-amino acid residue inactivates the MAPK and inhibits downstream signalling. Interestingly, this inactivation can be accomplished by evolutionarily distant protein phosphatases, including Ser/ Thr protein phosphatases of the type 2C (PP2C) MAPK phosphatases (Schweighofer et al., 2004;Schweighofer et al., 2007;Fuchs et al., 2013) and PTP-type dual specificity (Tyr and Ser/Thr) phosphatases (DSPs) (Bartels et al., 2010;Jiang et al., 2018). However, their interplay is presently unknown.
A group of PP2C-type phosphatases, including AP2C1, interacts with MAPKs and controls their activities (Schweighofer et al., 2007;Umbrasaite et al., 2010). AP2C1 is induced by wounding and biotic stress, and functions as a negative regulator of MPK3, MPK4, and MPK6, controlling levels of wound-induced jasmonate and ethylene (ET) as well as plant immunity (Schweighofer et al., 2007;Galletti et al., 2011;Sidonskaya et al., 2016;Shubchynskyy et al., 2017). ap2c1 plants do not display obvious developmental phenotypes under standard growth conditions (Schweighofer et al., 2007), implying a specific function of AP2C1 under stress conditions and the contribution of other, presently unknown, MAPK phosphatases for MAPK control in the absence of AP2C1.
Activation of transcription factors (TFs) and changes in gene expression are part of the cellular response to a perceived signal in order to reprogram cellular processes (Rauf et al., 2013). A number of TFs, including WRKY and AP2-domain/ethyleneresponsive factor (AP2/ERF) family members, have been suggested or demonstrated to act downstream of MAPKs in plants (Asai et al., 2002;Kim and Zhang, 2004;Menke et al., 2005;Nakano et al., 2006;Popescu et al., 2009;Bethke et al., 2009;Mao et al., 2011;Li et al., 2012;Meng and Zhang, 2013;Guan et al., 2014). Subsequently, these proteins may constitute an important link between pathogen-or wound-induced MAPK signalling and downstream transcriptional reprogramming.
Considering the broad spectrum of signals transmitted by the same MAPKs (Rodriguez et al., 2010;Meng and Zhang, 2013), such as MPK6, it is puzzling how the specificity of the responses for perceived stimuli is generated. The phylogenetic diversity and distinct enzymatic mechanisms of protein phosphatases that are able to inactivate MAPKs support the idea of a contribution of MAPK phosphatases to the versatility and specificity of MAPK networks. Here, we investigate the roles of the phylogenetically distant Arabidopsis MAPK phosphatases AP2C1 and MKP1 and, in particular, their functional redundancies. We show that AP2C1 and MKP1 together repress plant autoimmune-like responses, including salicylic acid (SA) and ET accumulation, and early senescence. These observations in the ap2c1 mkp1 double mutant are underlined by the mis-expression of specific TF genes, including members of the WRKY, AP2/ERF, and Arabidopsis NAM, ATAF, and CUC (ANAC) families whose expression is at least partially mediated by MPK6 and MPK3.
For experiments at the seedling stage, seeds were surface sterilized and spread on plates containing half-strength Murashige and Skoog (MS) medium (Duchefa), pH 5.7, 1% (w/v) sucrose, and 0.7% plant agar (w/v; Duchefa). Seedlings were grown under long-day conditions (16 h light/8 h dark) at 22 °C. For suppression of growth phenotypes, plants were kept under short-day conditions in a tray with closed lid at 24 °C with saturating relative humidity.

Ex vivo kinase activity assay and MAPK immunoblotting
Plant protein extraction and the ex vivo kinase assay were performed as described previously (Schweighofer et al., 2007;Schweighofer et al., 2009) using polyclonal antibodies for immunoprecipitation and myelin basic protein (MBP) as the in vitro substrate of immunoprecipitated MAPKs. A master mix containing MBP and γ-ATP was used for each series of kinase assays to ensure equal distribution of the substrate to all reactions. Treatment with the pathogen-associated molecular pattern (PAMP) flg22 was performed as described previously (Shubchynskyy et al., 2017). Signal intensities of phosphorylated MBP were quantified with ImageQuant software (version 5.1; Amersham). MAPK protein amounts were visualized with the antibodies Anti-AtMPK3 (M8318), Anti-MPK4 (A6979), and Anti-AtMPK6 (A7104) (all Sigma).

RNA extraction and quantitative reverse transcription-PCR
Total RNA from leaves was isolated with the RNeasy Plant Mini Kit (Qiagen) and treated with TURBO DNA-free DNase I (Ambion) according to the manufacturers' instructions. RNA integrity was checked on 1% (w/v) agarose gels and the concentration was measured before and after DNAse I digestion. The absence of genomic DNA was verified by PCR using primers targeting an intron of the control gene At5g65080. cDNA synthesis was performed using the First Strand cDNA Synthesis Kit (Thermo Scientific). The efficiency of cDNA synthesis was estimated by quantitative reverse transcription-PCR (RT-qPCR) analysis using a primer pair amplifying the 3ʹ part of the control gene encoding GAPDH and a primer pair amplifying the 5ʹ part of the same gene. RT-qPCR reactions were performed as described previously (Balazadeh et al., 2008). ACTIN2 was selected as a reference gene for which four replicates were measured in each PCR run, and their average cycle threshold (CT) was used for relative expression analyses. TF expression data were normalized by subtracting the mean ACTIN2 gene CT value from the CT value (ΔCT) of each gene of interest. The expression value in the comparison between different genotypes was calculated using the expression 2 -ΔΔCT , where ΔΔCT represents the ΔCT of the mutant of interest minus the ΔCT of the control (WT). For TF expression profiling, an advanced version of an expression profiling platform (Balazadeh et al., 2008) that was originally described by Czechowski et al. (2004) was used, covering 1880 Arabidopsis TF genes. Statistical analysis was performed with JASP software (https://jasp-stats.org; version 0.14.1).

ET measurements and quantification of total SA and camalexin
Measurements of ET were performed by gas chromatography (Hewlett Packard 5890 Series II) with an Al 2 O 3 column (Agilent Technologies). Whole rosettes of 4-week-old plants grown under long-day conditions were collected, and leaves were wounded and transferred into 20 ml vials containing 4 ml half-strength MS medium with 0.8% (w/v) plant agar, to reduce the volume of the head space, and the vials were sealed so they were airtight. After 24 h, 100 µl of the gas phase was taken from the vials and analysed by gas chromatography-flame ionization detection. ET production was calculated per milligram of fresh tissue per hour.
Total SA was quantified as described previously (Rozhon et al., 2005) except that 20 µM EDTA was added to the HPLC eluent. Camalexin levels were determined as described previously (Shubchynskyy et al., 2017).

ap2c1 mkp1 double mutant plants show growth and developmental defects that are at least partially mediated by MPK6 and MPK3
To investigate the specific and/or overlapping roles of the MAPK phosphatases AP2C1 and MKP1, we took advantage of the Arabidopsis T-DNA insertion knockout mutants ap2c1 and mkp1, respectively (Schweighofer et al., 2007;Bartels et al., 2009). Phenotypically, ap2c1 and mkp1 mutant plants did not show any difference from WT plants when grown for up to 5 weeks under short-day conditions (Fig. 1A). However, mkp1 plants grown under long-day conditions showed altered morphology, such as aberrant leaf development and early senescence, which appeared approximately 3 weeks after germination ( Supplementary Fig. S1A), as described previously (Bartels et al., 2009).
To further analyse the functions of AP2C1 and MKP1 in plants, we generated a double mutant by genetic crossing. ap2c1 mkp1 plants showed phenotypic differences from both the WT and the single mutants. These appeared about 2 weeks after germination under standard growth conditions in soil in all ap2c1 mkp1 double mutants; first their sizes started differing, and during further growth ap2c1 mkp1 plants developed more pronounced multiple defects, including severe dwarfism and aberrant leaf development (Fig. 1A, B; Supplementary Fig. S1B-G). Four weeks after germination, phenotypic abnormalities became even more evident, including early senescence, spontaneous macroscopic lesions, and abnormal leaf morphology. These developmental defects were suppressed when plants were grown under conditions of elevated humidity and increased temperature ( Supplementary Fig. S2), indicating a dependency on environmental cues. However, during flowering, misshapen inflorescences and strongly reduced fertility were always observed ( Supplementary Fig. S1E). These phenotypes were specific for ap2c1 mkp1 plants, as crossing ap2c1 with ptp1 did not lead to phenotypic alterations ( Fig. 1C) compared with mkp1 ptp1 (Bartels et al., 2009). Crossing mkp1 with either of two other clade B AP2C mutants, ap2c2 or ap2c3 (Umbrasaite et al., 2010), led to only mild phenotypes compared with the strong defects of ap2c1 mkp1 plants (Fig. 1B).
Since MPK6 and MPK3 are targets of AP2C1 and MKP1 (Ulm et al., 2002;Schweighofer et al., 2007;Galletti et al., 2011), we addressed the impact of MPK6 and MPK3 on the phenotypic aberrations detected in ap2c1 mkp1 plants. To this end, ap2c1 mkp1 mpk6 and ap2c1 mkp1 mpk3 triple mutant plants were created and their phenotypes were compared with those of ap2c1 mkp1. The phenotype of ap2c1 mkp1 mpk6 plants was more similar to that of the WT than that of ap2c1 mkp1. The loss of MPK6 suppressed most phenotypic defects observed in ap2c1 mkp1, such as extreme dwarfism, aberrant leaf shapes, premature leaf senescence, and impaired fertility (Fig. 1A, Supplementary Fig. S3). However, at later developmental stages, ap2c1 mkp1 mpk6 plants still appeared overall smaller than WT and displayed senescence in the older leaves ( Supplementary  Fig. S3E, M). Similarly, ap2c1 mkp1 mpk3 plants suppressed the observed ap2c1 mkp1 phenotypes in early development ( Fig.  1A) but to a lesser extent at later developmental stages compared with ap2c1 mkp1 mpk6 ( Supplementary Fig. S3F, N). Both ap2c1 mkp1 mpk6 and ap2c1 mkp1 mpk3 developmental phenotypes were suppressed during growth in conditions of elevated temperature and increased humidity, as observed with ap2c1 mkp1 (Supplementary Fig. S2).
Overall, these results suggest that AP2C1 and MKP1 protein phosphatases act partially redundantly and that the presence of at least either gene is necessary for normal plant development. The phenotypes observed in ap2c1 mkp1 plants are predominantly MPK6-and MPK3-dependent.

Dual control of stress-induced MAPK activities by AP2C1 and MKP1
Our previous work revealed the involvement of AP2C1 in the regulation of MAPK activities induced by wounding, nematode feeding, and PAMPs (Schweighofer et al., 2007;Sidonskaya et al., 2016;Shubchynskyy et al., 2017). To check for a potential overlapping role of MKP1, we first analysed MPK3, MPK4, and MPK6 activities after wounding the leaves of WT plants and of the single mutant plants ap2c1 and mkp1. Kinase activities were assayed after immunoprecipitation from total protein extracts using specific antibodies. In agreement with our previous findings (Schweighofer et al., 2007), ap2c1 plants showed higher and more sustained wound-induced activities of MPK3, MPK4, and MPK6 compared with the WT (Fig. 2). Interestingly, MPK4 activity was more intense and sustained in ap2c1 compared with mkp1 plants, indicating a specific role of AP2C1 in the regulation of MPK4 during wounding. MPK3 activity in ap2c1 plants was slightly higher and more sustained in comparison to WT and mkp1. In mkp1 plants, the peak activity of MPK6 was shifted to an earlier time point compared with the WT. In ap2c1 mkp1 plants, we detected strongly and moderately enhanced basal activity of MPK4 and MPK6, respectively, whereas the basal activity of MPK3 was not affected in comparison to the WT or single mutant lines. The stronger and more sustained wound-induced activation of MAPKs observed in single-mutant plants was additionally enhanced in the double mutant ap2c1 mkp1 (Fig.  2, Supplementary Fig. S4A). The MPK3, MPK4, and MPK6 protein amounts were comparable in both the single and double mutants and the WT (Fig. 2). Treatment of WT, ap2c1 and mkp1 single and double mutants, and ap2c1 mkp1 mpk3 and ap2c1 mkp1 mpk6 triple mutant plants with the PAMP flg22 led to similar MAPK activation patterns as those observed after wounding, with increased and prolonged MAPK activities in the ap2c1 mkp1 plants compared with the single mutants and the WT ( Supplementary Fig. S4B, C).
MKP1 has been reported to be constitutively expressed (Ulm et al., 2002), whereas AP2C1 is transcriptionally responsive to stress (Schweighofer et al., 2007;Sidonskaya et al., 2016). We tested whether reciprocal compensational expression may occur under long-day conditions and thus analysed AP2C1 and MKP1 mRNA levels in the mkp1 and ap2c1 mutants, respectively. RT-qPCR analyses showed only very slightly increased expression of MKP1 in ap2c1 plants, whereas the expression of AP2C1 was approximately 160% of the WT level in the mkp1 plants ( Supplementary Fig. S5), suggesting a compensatory transcriptional activation of AP2C1 in the absence of MKP1. Our results suggest cooperative action and partial redundancy in the regulation of MAPKs by these two evolutionarily distant and unrelated MAPK phosphatases.

AP2C1 and MKP1 play partially redundant roles in the control of wound-induced ET synthesis
Enhanced ET production is an early response of plants subjected to biotic or abiotic stresses Ju and Chang, 2012). We have previously shown that ectopic expression of AP2C1 suppresses MPK6 activation and woundinduced ET production in plant leaves (Schweighofer et al., 2007). Since both AP2C1 and MKP1 control MPK6 activity, a major determinant in the regulation of ET biosynthesis (Liu and Zhang, 2004;Li et al., 2012), we analysed the amounts of wound-induced ET in leaves of WT, ap2c1, mkp1, ap2c1 mkp1, and ap2c1 mkp1 mpk6 plants. As reported previously (Schweighofer et al., 2007), the wound-induced ET amounts were similar in ap2c1 and WT plants (Fig. 3A). However, significantly greater (P<0.05) amounts of ET accumulated in wounded mkp1 plants and even higher amounts in the ap2c1 mkp1 double mutant (Fig. 3A). Our data suggest a primary role of MKP1 in the control of wound-triggered ET production and that although disruption of AP2C1 alone is not sufficient to alter ET production upon wounding, it contributes significantly to the regulation of ET amounts in the absence of MKP1. Interestingly, and in agreement with the overall milder phenotype, wound-induced ET accumulation in ap2c1 mkp1 mpk6 plants was similar that detected in the WT (Fig. 3A).
The transcriptional regulation of 1-aminocyclopropane-1-carboxylic synthase (ACS) enzymes contributes to the control of ET production (Li et al., 2012). Therefore, we quantified the transcripts of ACS6, the expression of which is significantly induced after pathogen attack (Li et al., 2012) and wounding (Li et al., 2018). Compared with the WT, no changes in ACS6 transcript levels were detected in ap2c1, slightly higher levels in mkp1, and a 9-fold increase in ap2c1 mkp1, whereas in ap2c1 mkp1 mpk6 plants levels were not significantly different from those in WT plants (Fig. 3B). Thus, our data show that ACS6 is more strongly expressed in ap2c1 mkp1 plants, which likely contributes to the increased amounts of ET upon wounding, and that both effects are mediated by MPK6. Taken together, the wound-induced MAPK activities, expression patterns, and effects on ET production suggest that AP2C1 and MKP1 have both distinct and overlapping functions in wounded leaves.

Transcription factor gene expression is deregulated in ap2c1 mkp1 plants
To investigate if and how AP2C1 and MKP1 influence the regulation of gene expression under standard growth conditions, we used an RT-qPCR platform for high-throughput expression profiling of 1880 Arabidopsis TF-encoding genes (Balazadeh et al., 2008). We selected genes showing an at least 3-fold mean difference of expression levels in ap2c1, mkp1, or ap2c1 mkp1 plants compared with the WT. We identified three genes encoding TFs that were deregulated in ap2c1 but not in mkp1 (Supplementary Table S1), 25 genes that were deregulated in mkp1 but not in ap2c1 (Supplementary Table  S2), and four genes concomitantly regulated by AP2C1 and MKP1 (Supplementary Table S3). Figure 4 shows the number of genes whose expression levels were changed in ap2c1, mkp1, or ap2c1 mkp1 plants, compared with the WT. The TF genes dysregulated in the double mutant, and their expression values relative to the WT, are presented in Supplementary  Table S4. The deregulation of 76 TF-encoding genes (58 up-regulated, 18 down-regulated) was found reproducibly in at least three different experiments in ap2c1 mkp1 double mutant plants. Among them, genes encoding members of the WRKY family were most abundant: 15 WRKY genes were up-regulated (Fig. 5, Supplementary Table S4) and one was down-regulated (Supplementary Table S4). A further prevalent group of TF-encoding genes affected in ap2c1 mkp1 plants includes AP2/ERF genes, described for their involvement in development, including RAP2.6L  and WIND3 (Smit et al., 2020) (Fig. 6). ANAC TF family members are implicated in senescence and stress-related processes (Bu et al., 2008;Jensen et al., 2010;Wu et al., 2012;Saga et al., 2012). Our results show that several ANAC TF-encoding genes were up-regulated in ap2c1 mkp1 plants (Fig. 7). Thus, our data suggest a cooperative function of AP2C1 and MKP1 in the transcriptional regulation of a set of WRKY, AP2/ERF, and ANAC genes in the WT.
Our observation that ap2c1 mkp1 mpk6 and ap2c1 mkp1 mpk3 plants are phenotypically much less affected than the ap2c1 mkp1 double mutant suggested that severe phenotypic aberrations in the double mutant are mediated mainly by MPK6 and probably to a lesser extent by MPK3. This prompted us to investigate ap2c1 mkp1 mpk6 and ap2c1 mkp1 mpk3 plants for the expression of TFs misregulated in ap2c1 mkp1. Indeed, most of the TF genes strongly affected in ap2c1 mkp1 plants (Supplementary Table S4) were not significantly altered in their expression in ap2c1 mkp1 mpk6 and ap2c1 mkp1 mpk3, linking MAPK over-activation to the TF misexpression in the double mutant (Figs 5-7). Among these, the expression of WRKY75, WRKY71, WRKY64, and WRKY40 (Fig. 5) and of At2g33710 (Fig. 6) in the double mutant are more MPK3-independent, while the expression of other identified WRKYs (Fig. 5), the AP2/ERF family members RAP2.6L, WIND3, DREB19, ERF14, ERF17, ERF71, and ERF98 (Fig.  6), and several ANAC TF-encoding genes, such as ANAC005, ANAC042, ANAC003, ANAC047, and ANAC055 (Fig. 7), are mainly dependent on the presence of both MPK6 and MPK3.
Defence responses, camalexin, SA, and the senescence marker gene SAG12 are up-regulated in ap2c1 mkp1 plants It has been shown previously that mkp1 plants accumulate higher levels of the phytoalexin camalexin (Bartels et al., 2009). To investigate whether the expression of genes encoding camalexin biosynthesis enzymes was affected in ap2c1 mkp1 plants, we studied the expression of a key gene in the pathway, CYP71B15/PAD3. A strong up-regulation (>300-fold) was detected in ap2c1 mkp1 plants compared with the WT (Supplementary Fig. S6). Moreover, ap2c1 mkp1 mpk6 plants still had remarkably high transcript levels (up-regulation >10-fold) of CYP71B15/PAD3. In addition, mkp1 single mutant plants showed a ~10-fold up-regulation of the gene (Supplementary Fig. S6).
To investigate whether the increased CYP71B15/PAD3 expression level correlates with camalexin accumulation, total camalexin was quantified in WT and mutant plants. Indeed, we found increased camalexin concentrations in mkp1 plants, in agreement with previous findings (Bartels et al., 2009), and very high camalexin accumulation in the ap2c1 mkp1 mutant, which was not solely dependent on MPK6 and MPK3 (Fig. 8).
The up-regulation of MAPK activities and macroscopic lesion formation in leaves of ap2c1 mkp1 indicated the possible activation of a hypersensitive-like response in these plants. Since this response is associated with the accumulation of the stress hormone SA, we measured SA in leaves of ap2c1 mkp1 and ap2c1 mkp1 mpk6, as well as in WT and the single mutants. We found a 35-fold increase of SA in ap2c1 mkp1 plants  compared with the WT (Fig. 9), whereas ap2c1, ap2c1 mkp1 mpk6, and mpk6 plants showed SA concentrations similar to the WT. In agreement with previous data (Bartels et al., 2009), we also detected enhanced total SA concentrations (>2-fold) in mkp1 plants compared with the WT (Fig. 9).
The leaf necrosis observed in ap2c1 mkp1 leaves ( Fig. 1A; Supplementary Fig. S1G) and the up-regulation of WRKY6 (Fig. 5), which is a senescence-related marker gene (Rushton et al., 2010), suggested that early senescence was induced in these plants. Thus, we investigated the expression of the senescencespecific marker gene SENESCENCE-ASSOCIATED GENE12 (SAG12) (Noh and Amasino, 1999) and found that it was strongly up-regulated in ap2c1 mkp1 plants, dependent on MPK6 (Fig. 10). These data, along with the up-regulation of WRKY6 (Fig. 5), indicate aberrant early induction of senescence-related processes in the double phosphatase mutant.

Coordinated control of MAPK activities by AP2C1 and MKP1
Acclimation for survival is a fundamental principle that relies on intracellular signalling in every organism. Different signals converge at the level of MAPK cascades and from there diverge into a range of different downstream pathways and responses (Andreasson and Ellis, 2010;Rodriguez et al., 2010;Rasmussen et al., 2012). Considering the broad spectrum of signals transduced by overlapping players of MAPK pathways, it is puzzling how response specificity is attained (Lampard et al., 2009;Rodriguez et al., 2010;Meng and Zhang, 2013). Several signalling scenarios have been investigated that could help explain pathway specificity, including activity-dependent kinase distribution and localization, protein complex formation (e.g.  interaction with scaffolding proteins), and dephosphorylation by protein phosphatases (Krysan and Colcombet, 2018). Over the past decades, mainly the functions of MPK3/MPK4/ MPK6 in diverse pathways have been described, indicating them as both points of divergence and integration hubs in cellular signalling (Peng et al., 2018;Bigeard and Hirt, 2018).
Here, we provide evidence that two evolutionarily distinct MAPK phosphatases control stress-related signalling in Arabidopsis by inactivating an overlapping set of target MAPKs that mediate stress and defence responses. The Ser/Thr PP2C phosphatase AP2C1 and the DSP MKP1 contribute to ensure the appropriate inactivation of MAPKs during stress. Both AP2C1 and MKP1 target MPK3, MPK4, and MPK6 (Ulm et al., 2002;Schweighofer et al., 2007;Bartels et al., 2009;Anderson et al., 2011;Galletti et al., 2011;Sidonskaya et al., 2016;Shubchynskyy et al., 2017). The enhanced activation of MAPKs by wounding and by PAMP (flg22), and the constitutive stress signalling in the absence of stress in ap2c1 mkp1 plants, indicate that the lack of both MAPK phosphatases creates a shortfall downstream of MAPKs, exemplified by the deregulated expression of TF-encoding genes.
The increased kinase activity in mkp1 plants versus WT at the earlier time points after wounding and flg22 application compared with ap2c1 versus WT suggests that the contribution of MKP1 to inactivating MAPKs is already set before stress treatment, or during a very early stage of signalling. In agreement with this suggested function of MKP1 during early signalling, previous work has also shown that mkp1 and mkp1 ptp1 mutant plants demonstrate equally elevated MAPK activities without stress treatment, underlining the major role of MKP1 in MAPK regulation in ambient conditions (Bartels et al., 2009). On the contrary, AP2C1 adds to MAPK inactivation at later time points. It is possible that AP2C1 is primarily responsible for keeping the stress-induced activation below a certain threshold and controlling the duration of kinase activation during acute stress, acting as an 'emergency brake', while MKP1 is predominantly responsible for suppressing kinase activities under normal conditions, providing a 'constitutive brake'. This hypothesis is supported by the induction of AP2C1 expression by a plethora of stresses, while MKP1 shows comparatively marginal changes in expression (https:// www.genevestigator.com). These observations are also consistent with a recent comprehensive analysis of the Arabidopsis proteome, which covers more than 14 000 proteins and where in ambient conditions the overall MKP1 abundance exceeds by far that of AP2C1 (http://athena.proteomics.wzw.tum.de/) (Mergner et al., 2020), underlining the rather specific role of AP2C1 under stress conditions. The AP2C1 paralogues AP2C2 and AP2C3 (Umbrasaite et al., 2010;Umbrasaite et al., 2011;Schweighofer et al., 2014) as well as MKP1 and PTP1 interact with the same MAPKs and dephosphorylate them to various extents (Bartels et al., 2009). However, the rather mild phenotypes of ap2c2 mkp1 and ap2c3 mkp1 plants, and the WT-like appearance of ap2c1 ptp1 (Fig. 1C) compared with ap2c1 mkp1 plants, clearly indicate specific genetic interactions and redundant functions of the evolutionarily distant AP2C1 and MKP1 phosphatases in the regulation of signalling pathways.

Manifestation of cell death in ap2c1 mkp1 plants
The lesions in leaves of ap2c1 mkp1 plants suggest autoimmune-like responses, most likely caused by misregulation of MAPKs and/or failed control of guarding resistance (R) proteins (Rodriguez et al., 2016). AP2C1 and MKP1 share the target MAPKs MPK3, MPK4, and MPK6, where MPK4 and some of its upstream MAPK cascade members, for example, MEKK1 and MKK1/2, were originally described as negative regulators of plant immunity based on their mutant plant phenotypes (Petersen et al., 2000;Rasmussen et al., 2012). The improper activation of the R-gene SUMM2 is mainly responsible for the phenotypic defects of the mpk4 mutant and other mutant plants in the pathway, identifying the MEKK1-MKK1/2-MPK4 module as a positive regulator of stress responses . Similar observations connecting phosphatase-targeted MAPKs with autoimmune-like phenotypes have been made by ectopically expressing constitutively active MPK3 (Genot et al., 2017) or by inducibly expressing MKK5, which activates MPK3 and MPK6 (Lassowskat et al., 2014). Both approaches led to a plethora of phenotypic and molecular changes, including dwarfism, lesion formation, de-repression of defence gene expression, and the accumulation of stress hormones, similar to the ap2c1 mkp1-related phenotypes described in this work (Figs 1-10; Supplementary Figs S1, S3, S4, S6; Supplementary Table S4).
The single mkp1 and double mkp1 ptp1 mutants show constitutive defence responses including increased levels of SA and camalexin, suggesting partially overlapping functions of MKP1 and PTP1 in repressing SA biosynthesis (Bartels et al., 2009). Similarly, the strong accumulation of SA and camalexin in ap2c1 mkp1 compared with mkp1 plants suggests a collaborative action of both AP2C1 and MKP1 as negative regulators of SA and camalexin production (Figs 8,9; Supplementary Fig. S6). The camalexin accumulation in ap2c1 mkp1 is largely reduced in ap2c1 mkp1 mpk6 and ap2c1 mkp1 mpk3 triple mutants, indicating the dependency of camalexin biosynthesis on partially redundant actions of MPK6 and MPK3, which is in agreement with previous findings (Mao et al., 2011). The SA accumulation in ap2c1 mkp1 plants is probably mainly MPK6-dependent, as the introduction of the mpk6 mutation in ap2c1 mkp1 mpk6 plants restores SA concentrations similar to those of WT and mkp1 ptp1 mpk6 mutant plants (Bartels et al., 2009). Notably, rescue of the severe ap2c1 mkp1 growth phenotypes by elevated temperature is in accordance with the observed temperature dependency of SA-related phenotypes (Ichimura et al., 2006;Suarez-Rodriguez et al., 2007;, as well as with the suppression of SNC1 expression and reduction of SNC1 activity by high temperature (Yang and Hua, 2004;Zhu et al., 2010). The resistance protein SNC1 is a modifier of mkp1 in the Col-0 accession, where partial rescue of mkp1 and mkp1 ptp1 growth phenotypes by a loss-of-function snc1 mutation indicates a sensitized SNC1 signalling pathway in the absence of MKP1 (Bartels et al., 2009).
Previous findings that SA acts together with ET to regulate cell death (Rao et al., 2002), the requirement of ET biosynthesis for H 2 O 2 accumulation and subsequent cell death (Overmyer et al., 2003), and the induction of cell death in Arabidopsis leaves by persistent activation of MAPKs with gain-of-function MKK4 and MKK5 (Ren et al., 2002) all correlate with the cell-death phenotype observed in the ap2c1 mkp1 mutant, where MAPKs-and other stress-related factors-may be (hyper)-activated. Therefore, we conclude that the majority of the phenotypes observed in ap2c1 mkp1 plants, both visible and molecular, are due to the misregulation of MAPK pathways, even in the absence of stress.

AP2C1 and MKP1 affect MAPK-regulated ET biosynthesis
Activated MPK6 controls ET levels by inducing the transcription of ACS family genes and by phosphorylating ACS proteins, the rate-limiting enzymes in ET biosynthesis. Phosphorylated ACSs become more stable and, thus, ET synthesis is increased by elevated MPK6 activity (Kim et al., 2003;Liu and Zhang, 2004;Xu et al., 2008;Li et al., 2012). In ap2c1 mkp1, the increased ET production is likely due to, at least in part, the highly increased expression of ACS6 compared with the WT. A considerable additive effect on ET overproduction in the double ap2c1 mkp1 mutant compared with the mkp1 single mutant suggests that even though MKP1 is a determining MAPK phosphatase affecting ET production, there are overlapping and non-redundant functions of AP2C1 and MKP1 in the regulation of stress-induced ET biosynthesis. The detection of increased and MPK6-dependent expression of WRKY33, which encodes a TF that binds to the promoter of ACS genes and is a substrate of MPK3 and MPK6, suggests an involvement of WKRY33 itself in ACS overexpression in ap2c1 mkp1 plants (Figs 3, 5;Li et al., 2012). The identification of genes encoding TFs of the AP2/ERF family (ET-responsive element-binding proteins) among the most strongly induced ones in ap2c1 mkp1 plants suggests a path to increased ET amounts in these plants.

AP2C1 and MKP1 control the expression of stressresponsive TF-encoding genes, predominantly via partially redundant actions of MPK6 and MPK3
Transcriptional reprogramming in response to activated MAPK signalling suggests an involvement of TFs. Our results indicate that the concomitant lack of the MAPK regulators AP2C1 and MKP1 results in elevated basal MAPK activities and leads to highly increased expression of WRKY TF genes, in some cases by more than 100-fold compared with the WT. The ap2c1 mkp1 mutant phenotypes and the described functions of some up-regulated WRKYs indicate that stress responses are constitutively active in these plants. This correlates with reports demonstrating an involvement of WRKYs in oxidative stress responses, in the induction of ET and camalexin biosynthesis (WRKY30, WRKY33), the response to pathogens (WRKY71, WRKY40), basal defence (WRKY38), and defence-and senescence-related processes (WRKY6) (Rushton et al., 2010).
Direct feedback mechanisms among WRKYs themselves have been demonstrated (Mao et al., 2011) and are generally proposed, where WRKYs positively auto-regulate their own gene expression and/or cross-regulate the expression of other WRKY genes (Pandey and Somssich, 2009;Mao et al., 2011;Birkenbihl et al., 2017). Thus, it could be that the increased activation of MAPKs in ap2c1 mkp1 plants leads to phosphorylation and thus activation of MAPK target WRKY proteins, which serve as activated TFs for a further series of WRKY genes. In any case, MPK6 and MPK3 seem to be major players responsible for mediating the up-regulation of several WRKYs, AP2/ERFs, ANACs, and other TF-encoding genes. Both MAPKs control the expression of several WRKYs to different extents, as shown in ap2c1 mkp1 plants compared with ap2c1 mkp1 mpk6 and ap2c1 mkp1 mpk3 plants (Fig. 5). These data also demonstrate that not only MAPKs but also other factor(s) affect WRKY gene expression. We confirmed MPK6-and MPK3dependent WRKY33 expression (Mao et al., 2011); however, the higher MPK4 activities in ap2c1 mkp1 may also lead to higher amounts of active WRKY33 protein (Qiu et al., 2008;Birkenbihl et al., 2017). Our data suggest that AP2C1 and MKP1 may play a dual role in regulating camalexin biosynthesis, on the one hand by controlling MPK6 and MPK3 activities, which positively regulate WKRY33 expression, and on the other hand by controlling MPK4 activity, which in turn stimulates WRKY33, leading to the transactivation of CYP71B15/PAD3.
Remarkably, the concomitant absence of AP2C1 and MKP1 in non-stress conditions causes a distinct transcriptional activation of TFs compared with single mutant ap2c1 and mkp1 plants after stress treatments: challenging ap2c1 plants with Pto for 4 h led to 88 differentially regulated TFs (Shubchynskyy et al., 2017), whereas only four TFs from this set were up-regulated in untreated ap2c1 mkp1 plants (Supplementary Table S4).
Comparing transcriptional changes of PAMP-treated mkp1 plants (Jiang et al., 2017) with untreated ap2c1 mkp1 double mutant plants (Supplementary Table S4) revealed a more MKP1-specific TF induction: a next-generation sequencing (RNA-seq) transcriptome analysis of mkp1 plants 90 min after treatment with the PAMP elf26 revealed that from 67 identified TFs among the 1102 MKP1-dependent transcripts (Jiang et al., 2017) 21 TFs were also changed in untreated ap2c1 mkp1 plants (Supplementary Table S4). In accordance with the severe phenotype of mkp1 ptp1 mutant plants and its suppression by elevated growth temperature or crossing with mpk3 and mpk6 mutants (Bartels et al., 2009), the proposed predominant role of MKP1 under non-stress conditions and AP2C1 during stress conditions to regulate (MAPK) signalling are emphasized.

Senescence is repressed by AP2C1 and MKP1 phosphatases in an MPK6-dependent way
Several lines of evidence indicate that the ap2c1 mkp1 mutant undergoes precocious senescence. Leaf senescence is a highly regulated process that finally leads to cell death and tissue disintegration, at the same time contributing to the fitness of the whole plant. Senescence is controlled by endogenous and environmental cues, and can be triggered prematurely by different abiotic/biotic stresses such as pathogen attack, wounding, UV light irradiation, or high ozone levels (Hanfrey et al., 1996;Miller et al., 1999;John et al., 2001;He et al., 2001;Lim et al., 2007). The MKK9-MPK6 cascade has been shown to positively regulate leaf senescence in Arabidopsis (Zhou et al., 2009). Hyperactivation of MPK6 and other MAPKs, in addition to autoimmune-like responses, also promotes senescence, which is very evident in older leaves of ap2c1 mkp1 plants and correlates with significant up-regulation of the senescence-specific marker gene SAG12 (Noh and Amasino, 1999;Guo and Gan, 2005). Partial suppression of SAG12 overexpression in ap2c1 mkp1 mpk6 suggests an MPK6-dependent regulation (possibly involving other MAPKs) in promoting plant senescence.
We conclude that the induction of senescence processes as well as hypersensitive response-like cell death results in the premature death of leaves in ap2c1 mkp1 plants. The crosstalk between senescence and abiotic stress or pathogen responses is accentuated in ap2c1 mkp1 plants, where up-regulation of TFs involved in these processes occurs.
Taken together, our results show that two evolutionarily unrelated MAPK phosphatases, AP2C1 and MKP1, perform both distinct and overlapping functions in the regulation of stress-induced MPK3, MPK4, and MPK6 activities. Our genetic dissection indicates that the known roles of MPK6 and MPK3 in mediating cell death and ET-, SA-and senescencerelated phenotypes are counter-balanced by both AP2C1 and MKP1. It also demonstrates that the expression of specific TF-encoding genes is affected by hyperactivation of MAPK(s) due to the lack of these two MAPK phosphatases in planta, revealing potential new signalling target genes downstream of MPK6 and MPK3. In the future, the study of individual and combinatorial mutants will allow us to genetically disentangle the contribution of specific protein kinases and phosphatases to complex signalling networks and downstream cell responses.

Supplementary data
The following supplementary data are available at JXB online. Fig. S1. Loss of both AP2C1 and MKP1 leads to severe phenotypes in growth and development. Fig. S2. Plant phenotypes grown in conditions with increased humidity and elevated temperature. Fig. S3. Phenotypes of Arabidopsis single, double and triple mutant plants. Fig. S4. AP2C1 and MKP1 control stress-induced MAPK activities. Fig. S5. Detection of MKP1 and AP2C1 expression levels in ap2c1 and mkp1 mutants. Fig. S6. Up-regulation of the camalexin biosynthetic gene CYP71B15/PAD3 is mainly mediated by MPK6. Table S1. Expression of TF-encoding genes modulated by the absence of AP2C1 but not MKP1. Table S2. Expression of TF-encoding genes modulated by the absence of MKP1 but not of AP2C1. Table S3. Expression of TF-encoding genes modulated by the absence of both MKP1 and AP2C1. Table S4. TF-encoding genes deregulated in ap2c1 mkp1 plants. Table S5. The Arabidopsis Genome Initiative (AGI) codes of the genes analysed in this report.