Mapping of Cvi-0 Ozone Sensitivity Phenotypes

Our initial QTL mapping of ozone sensitivity in Cvi-0 placed the two major contributing loci on the lower ends of chromosomes 2 and 3 [8]. To identify the causative loci related to the extreme ozone sensitivity and more open stomata of Cvi-0, we created a NIL termed Col-S (for Col-0 ozone sensitive) through eight generations of backcrossing of Cvi-0 with Col-0 (Fig 1A, S1A Fig, and S1 Video). In parallel, ozone tolerance from Col-0 was introgressed to Cvi-0 by six generations of backcrossing, which generated the ozone-tolerant Cvi-T (S1 Video). Using these accessions, NILs, and recombinant inbred lines (RILs), we mapped the causative ozone QTLs to a region of 90 kb on chromosome 2 and 17.70–18.18 Mbp on chromosome 3 (S1B Fig). We have previously shown that the QTL on chromosome 2 also controls plant water loss and stomatal function [8]. We isolated both QTLs by backcrossing Col-S with Col-0 and obtained the NILs Col-S2 and Col-S3. Both of these were less sensitive to ozone than Col-S (S1A Fig), indicating that these QTLs act additively to regulate ozone sensitivity. Col-S2 (but not Col-S3) showed much higher daytime stomatal conductance than Col-0 (Fig 1B). The mapping resolution on chromosome 3 was not sufficient to identify the causative gene. Hence, we focused on Col-S2 and its role in stomatal function. Within the 90-kb mapping region on chromosome 2, one gene, At2g46070, encoding a MAP kinase MPK12, shows strong preferential guard cell expression [23]. A single point mutation was found in Cvi-0 MPK12, leading to a glycine to arginine substitution at position 53 of the protein.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Mapping Cvi-0 ozone sensitivity and Cvi-0 and cis stomatal phenotypes.

(A) Tissue damage after 6 h of O₃ exposure (350 ppb). Visual damage of plant rosettes (upper images) and cell death visualized with trypan blue staining (lower images). Scale bars 1 cm. (B) Stomatal conductance of Col-0, Cvi-0, and NILs (mean ± standard error of the mean [SEM], n = 7–12). (C) Elevated CO₂ (800 ppm) induced stomatal closure in intact whole plants (n = 9–10, except cas-2 [n = 3]). Experiment was repeated at least three times with similar results. (D) Stomatal half-response times to elevated CO₂ (800 ppm). Error bars indicate ± SEM (n = 13). Pooled data from two experimental series are shown. (E) Gene model of MPK12 (At2g46070) and BYPASS2 (At2g46080). The deletion mutant cis (renamed as mpk12-4) has a 4,772 bp deletion (end and start indicated). Col-S2 has a G to C missense mutation at position 157 of MPK12, which leads to G53R substitution in MPK12. The mpk12-3 mutant has a Syngenta Arabidopsis Insertion Library (SAIL) transfer DNA (T-DNA) insertion in the second exon of MPK12. White boxes refer to exons, grey boxes to introns, and black lines to intergenic regions. Small letters (B) and asterisks (D) denote statistically significant differences according to one-way ANOVA with Tukey honest significant difference (HSD) for unequal sample size (Spjotvoll & Stoline test) or Tukey HSD post hoc test, respectively. The raw data for panels B–D can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.2000322.g001

Stomata-Related Phenotypes of Cvi-0 and cis are Caused by Mutations in MPK12

CAS (calcium-sensing receptor) is a chloroplast-localized protein important for proper stomatal responses to external Ca²⁺ [24,25]. While testing stomatal phenotypes in cas mutants, we observed phenotypic discrepancy between different alleles of cas. Whereas the cas-2 (GABI-665G12) line had more open stomata and impaired CO₂ responses, this was neither observed in cas-1 nor in cas-3 (Fig 1C and S1C Fig). Further experiments showed that the T-DNA insert in the CAS gene was not linked to the CO₂-insensitive phenotype of cas-2. In a backcross with Col-0, the T-DNA insert in cas-2 was removed, thereby generating the mutant cis (CO₂ insensitive). Both cis and Col-S2 had impaired responses to high CO₂ (800 ppm), leading to longer half-response times, but a residual CO₂ response could still be observed (Fig 1D and S1E Fig).

In order to identify the causative mutation in cis, mapping and whole genome sequencing of cis × C24 population was performed, which revealed a complete deletion of the MPK12 gene and its neighbor BYPASS2 in cis (Fig 1E and S1D Fig). Thus, cis was renamed mpk12-4. A second mutant (gdsl3-1) from the GABI-Kat collection (GABI-492D11) contained an identical deletion of BYPASS2 and MPK12 (S2 Fig). We also identified a line with a T-DNA insert in exon 2 of MPK12 from the SAIL collection (Fig 1E), which was recently named mpk12-3 [26]. No full-length transcript was found in mpk12-3 (S3 Fig). SALK T-DNA insertion lines of MPK12 were previously described as lethal [11,23]; similarly, we were unable to retrieve homozygous plants of the same alleles, possibly indicating the presence of an additional T-DNA in an essential gene. The new mpk12 deletion, SAIL T-DNA insertion, and Col-S2 point mutation alleles allowed a detailed characterization of the role of MPK12 in stomatal regulation.

Stomatal conductance was higher throughout the day in all three lines (Col-S2, mpk12-3, and mpk12-4) (Fig 2A), suggesting that the amino acid substitution in Cvi-0 MPK12 leads to loss of function. Furthermore, Col-0 transformed with MPK12 from Cvi-0 showed stomatal conductance similar to Col-0, which excludes the option that the G53R substitution in MPK12 would lead to gain of function (S1F Fig). Moreover, the wild-type (Col-0) stomatal phenotype was observed in heterozygous F1 plants from a cross of Col-S2 and Col-gl1, in which the gl1 mutation that gives a trichome-less phenotype was used as a noninvasive method for selecting successfully crossed plants in the first generation (S1G Fig). Increased stomatal conductance may result from an increased number of stomata, larger stomata, or more open stomata. However, the stomatal index, length, and density did not differ between the lines, indicating that MPK12 regulates a function related to the stomatal aperture (S4 Fig). Because of the higher degree of stomatal opening, the instantaneous WUE was lower in mpk12-3, mpk12-4, and Col-S2 (Fig 2B). Altered WUE was previously also seen in mpk12-1 and a NIL with Cvi-0 MPK12 in Ler [10].

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Stomatal conductance of the NIL Col-S2, mpk12 mutants, and complementation lines.

(A) Diurnal pattern of stomatal conductance with 12 h/12 h light–dark periods (n = 13–16). (B) Instantaneous water use efficiency (WUE) measured as an average of daytime light period from 09:00 to 17:00 (n = 13–16). (C) Stomatal conductance of Cvi-0 transformed with Col-0 MPK12 driven by its native promoter in T2 generation (n = 9). (D) Stomatal conductance of Col-S2 complementation line in T2 generation transformed with Col-0 MPK12, driven by its native promoter (n = 5–8). (E) Stomatal conductance of T3 transformants in the mpk12-4 background transformed with either the Col-0 or Cvi-0 version of MPK12, driven by its respective native promoter (n = 5–6). All graphs present mean ± SEM. Small letters denote statistically significant differences according to one-way ANOVA with Tukey HSD post hoc test for either unequal (B, D, E) or equal sample size (C). The raw data for panels A–E can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.2000322.g002

Cvi-0 and Col-S2 were complemented by expression of MPK12 from Col-0 (Fig 2C and 2D). Similarly, mpk12-4 was complemented by expression of Col-0 MPK12 but not by Cvi-0 MPK12 (Fig 2E). We conclude that MPK12 is a crucial regulator of stomatal conductance, and a single amino acid substitution (G53R) in Cvi-0 leads to loss of function of MPK12.

MPK12 Functions in Guard Cell CO₂ Signaling

Reduction of CO₂ levels inside the leaf [27] is a signal that indicates a shortage of substrate for photosynthesis and triggers stomatal opening. The rate of stomatal opening in response to low CO₂ was severely impaired in mpk12 and Col-S2 (Fig 3A and S5A Fig). Another signal for stomatal opening is light; this response was intact in plants with impaired or absence of MPK12 (S5B and S5C Fig). The hormone ABA has dual roles in stomatal regulation; it induces stomatal closure but also inhibits light-induced stomatal opening. The latter response was impaired in mpk12 mutants and Col-S2 (Fig 3B and S5C Fig). Stomata close in response to several signals, including darkness, reduced air humidity, ozone pulse, elevated CO₂, and ABA. Of these, only the response to elevated CO₂ was impaired in mpk12 and Col-S2 (Fig 3C and 3D, S5D–S5H and S6 Figs). CO₂ signaling is impaired in the carbonic anhydrase double mutant ca1 ca4 [13], and the product of carbonic anhydrase, bicarbonate, activates S-type anion currents [15]. In Col-S2 and mpk12-4, bicarbonate-induced S-type anion currents were strongly impaired (Fig 3E). Collectively, these data indicated that MPK12 has an important role in the regulation of CO₂-induced stomatal movements in Arabidopsis.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Responsiveness of the NIL Col-S2 and mpk12 mutants to stomatal opening and closing stimuli.

(A) Stomatal opening induced by 100 ppm CO₂ in whole plants (58 min after induction; n = 12–13). (B) Light-induced stomatal opening inhibited by 2.5 μM ABA in whole plants (24 min after induction; n = 16–18). (C) Stomatal closure induced by 800 ppm CO₂ in whole plants (10 min after induction; n = 12–13). (D) Stomatal closure induced by spraying whole plants with 5 μM ABA solution (24 min after induction; n = 12–14). (E) MPK12 is required for the bicarbonate (HCO₃^-)-induced slow type anion channel activation in guard cell protoplasts. Upper panels show typical whole guard cell protoplast recordings with 11.5 mM free HCO₃^- added to the pipette solution, and lower panels show average steady-state current-voltage relationships for wild-type (Col-0), NIL Col-S2, and mpk12-4 after treatment with mock or 11.5 mM HCO₃^- (n = 4–8 per line and treatment). Small letters (A, C) and asterisks (B, D) indicate statistically significant differences according to one-way ANOVA and two-way ANOVA with Tukey HSD for unequal sample size post hoc tests (p < 0.05), respectively. Error bars mark ± SEM. The raw data for panels A–E can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.2000322.g003

MPK12 Interacts with the Protein Kinase HT1

Only a few regulators of stomatal CO₂ signaling in Arabidopsis have been identified. These include the protein kinases HT1 and OST1 [12,15,16]. To find the interaction partners of MPK12, we conducted pairwise split-ubiquitin yeast two-hybrid (Y2H) assays against several kinases and phosphatases involved in stomatal signaling (Fig 4A and 4B and S7A and S7B Fig). A strong interaction was observed between MPK12 and HT1 in yeast. The MPK12–HT1 interaction was also confirmed in Nicotiana benthamiana with bimolecular fluorescence complementation (BiFC) (Fig 4C–4E) and split luciferase complementation assays (S7C Fig). Strong interaction between MPK12 and HT1 was observed in the cell periphery (Fig 4C). Recently, HT1 was shown to be a plasma membrane–associated protein [28]. In contrast, Col-0 and Cvi-0 MPK12-YFP were located inside the cell (S8A–S8D Fig). Hence, it is likely that the interaction with HT1 brings MPK12 to the plasma membrane. HT1 interacted with both the Col-0 and Cvi-0 versions of MPK12, but the interaction with Cvi-0 MPK12 (G53R) was weaker both in quantitative BiFC and Y2H assays (Fig 4B and 4D). MPK11, an MPK from the same group as MPK12 [29], did not interact with HT1 (Fig 4C). INDOLE-3-BUTYRIC ACID RESPONSE 5 (IBR5) is a MPK phosphatase that regulates auxin signaling in roots and has been shown to interact with and regulate the activity of MPK12 [11]. We confirmed the interaction between MPK12 and IBR5 (S7A Fig). However, the ibr5-1 mutant exhibited wild-type stomatal phenotypes in response to CO₂ changes (Fig 3A and 3C, and S5A and S5E Fig), suggesting that IBR5 is not required in stomatal CO₂ signaling.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. MPK12 interacts with HT1.

(A) Split-ubiquitin yeast two-hybrid assay on the SD-LeuTrp plate (left and middle panels) indicates the presence of both bait and prey plasmids; X-gal overlay assay (middle) and growth assay on the SD-LeuTrpHisAde plate (right) show HT1 interaction with MPK12 that is similar to the positive control (pAI-Alg5). Only weak or no interaction was detected with MPK12 G53R and MPK11, similar to the negative control (pDL2-Alg5). (B) Quantitative β-galactosidase assay from pools of ten colonies each. Activities are shown as the percentage of the positive control (± SEM; n = 3). (C) High-magnification (63x objective) BiFC images from a single infiltrated N. benthamiana leaf with identical confocal microscopy acquisition settings. Scale bar = 50 μm. (D) Ratiometric BiFC shows weaker interaction of MPK12 G53R than MPK12 with HT1, while MPK11 exhibits a weak interaction with HT1. The plasma membrane–localized SLAC1-CFP was used as an internal control. Eighteen images (from three leaves) of each construct set were analyzed. (E) Western blot together with Coomassie staining of proteins extracted from BiFC samples used for confocal imaging and controls with single construct shows expression of all fusion proteins. (F) Steady-state stomatal conductance of Col-S2 ht1-2, mpk12-4 ht1-2, and Col-S2 abi1-1 (ABA insensitive 1–1) double mutants (mean ± SEM, n = 11–13). Experiments were repeated at least three times. Letters in B, D, and F denote statistically significant differences with one-way ANOVA and Tukey HSD post hoc test for equal B, D, or unequal F sample size. The raw data for panels B, D, and F can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.2000322.g004

MPK12 Inhibits HT1 Activity

The function of MPK12 in ABA and CO₂ signaling was further explored through genetic analysis. A strong loss-of-function allele, ht1-2, that has low stomatal conductance [12] was used to evaluate the relationship between mpk12 and ht1-2. The Col-S2 ht1-2 and mpk12-4 ht1-2 double mutants had a more closed stomata phenotype similar to ht1-2 (Fig 4F), suggesting that HT1 is epistatic to MPK12. The strong impairment of stomatal function in abi1-1 (ABA insensitive1-1) was additive to Col-S2 in the double mutant Col-S2 abi1-1 (Fig 4F). Hence, signaling through MPK12 seems to act—at least to some extent—independently of the core ABA signaling pathway.

Taken together, the MPK12-HT1 interaction (Fig 4A–4E) and the epistasis between ht1-2 and mpk12-4 (Fig 4F) suggest that MPK12 functions upstream of HT1 and could regulate the activity of HT1. To test this directly, we performed in vitro kinase assays with casein as the substrate for HT1 (Fig 5A). HT1 displayed strong autophosphorylation and phosphorylated casein efficiently. Addition of the Col-0 version of MPK12 and a hyperactive version (MPK12 Y122C) efficiently inhibited HT1 activity (Fig 5A and quantified in Fig 5B). A point-mutated version (MPK12 K70R) designed to remove the kinase activity of MPK12 also inhibited the autophosphorylation activity of HT1 and phosphorylation of casein by HT1, although it was less efficient than the wild-type (Fig 5B). Importantly, the Cvi-0 version of MPK12 (G53R) displayed strongly suppressed inhibition of HT1 activity (Fig 5A and 5B). MPK12 did not phosphorylate the kinase-dead version of HT1 (K113M) (Fig 5C). The kinase-dead version of HT1 (K113M) was used as a substrate, since the strong autophosphorylation activity of HT1 would otherwise have obscured the result. Wild-type MPK12 and hyperactive MPK12 (Y122C) displayed autophosphorylation, whereas MPK12 (G53R) as well as MPK12 (K70R) had lost their autophosphorylation activity, indicating that the G53R substitution in Cvi-0 MPK12 disrupts the kinase activity of the protein (Fig 5C). The inhibition of HT1 by MPK12 was specific, as MPK11, which belongs to the same group as MPK12, was not able to affect HT1 kinase activity (S9 Fig). We conclude that the stomatal phenotypes of mpk12 mutants and Cvi-0 can be explained by a lack of inhibition of HT1 activity by MPK12, which leads to more open stomata and impaired CO₂ responses (Figs 2A, 3 and 5A and 5C).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. Regulation of HT1 by MPK12.

(A) Inhibition of HT1 kinase activity in vitro by different versions of MPK12 (MPK12 G53R—Cvi-0 version of MPK12; MPK12 K70R—inactive kinase; MPK12 Y122C—hyperactive kinase). Upper panel: autoradiography of the SDS PAGE gel; lower panel: Coomassie-stained SDS PAGE. Reaction mixture was incubated for 30 min. (B) Casein phosphorylation by HT1 with different MPK12 concentrations (mean ± SEM; n = 3). The raw data can be found in S1 Data. (C) Kinase-dead HT1 K113M was not in vitro phosphorylated by different versions of MPK12, and only MPK12 and MPK12 (Y122C) display clear autophosphorylation activities.

https://doi.org/10.1371/journal.pbio.2000322.g005

Both MPK12 and MPK4 Regulate the CO₂ Signaling Pathway

MPK12 belongs to the same group of MPKs as MPK4, a crucial regulator of pathogen and stress responses [29]. In tobacco, the silencing of MPK4 impaired CO₂-induced stomatal closure [30]. Since Arabidopsis MPK4 and MPK12 are highly similar [31], it is possible that both MPK4 and MPK12 could regulate stomatal CO₂ responses. Indeed, in an Y2H screen to identify HT1 interacting proteins, one prominent interactor was MPK4 [18].

The Arabidopsis mpk4 mutant is severely dwarfed, and measurements of accurate stomatal conductance with these plants are not feasible [32]. However, the impaired stomatal response to CO₂ in mpk12-4 (Fig 3) was further enhanced by guard cell–specific silencing of MPK4 [18]; hence, in guard cells MPK4 is acting redundantly with MPK12 in stomatal CO₂ signaling. Furthermore, MPK4 could also inhibit HT1 kinase activity [18]. The G53 residue in MPK12 is conserved in all Arabidopsis MPKs [10]. Since the G53R mutation blocked MPK12 function, we tested whether a similar mutation would impair MPK4 function. This experiment showed that MPK4-induced inhibition of HT1 activity was blocked by the introduction of a G55R mutation in MPK4; this mutation corresponds to G53R in Cvi-0 MPK12 (Fig 6A). Since MPK11 did not inhibit HT1 activity (S9 Fig), the function of MPKs as kinase inhibitors in Arabidopsis may be restricted to MPK12 and its closest homologue MPK4.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 6. A conserved glycine is important for MPK4 and MPK12 function.

(A) Inhibition of HT1 kinase activity in vitro by MPK4 and MPK4 G55R. Upper panel: autoradiography of the SDS PAGE gel; lower panel: Coomassie-stained SDS PAGE. Reaction mixture was incubated for 30 min. (B) Whole protein (left) and close-up (right) view of the superposition of models for MPK12 wild-type (secondary structure and surface in white) and MPK12 G53R (secondary structure in green). There is a close structural similarity between the structures except where the arginine at position 53 protrudes from the mutant protein surface and changes the loop region for the mutant. (C) Whole protein (left) and close-up (right) view of the superposition of models for MPK4 wild-type (secondary structure and surface in white) and MPK4 G55R (secondary structure in yellow). Similar to MPK12 G53R, the arginine at position 55 in MPK4 protrudes from the mutant protein surface and changes the loop region.

https://doi.org/10.1371/journal.pbio.2000322.g006

The Arabidopsis MPK6 crystal structure [33] was used to model the structure of MPK4 and MPK12 and to address the role of the G55R and G53R mutations that were shown to be crucial for the function of these proteins (Fig 6B and 6C). The mutation of Gly to Arg in position 53 in MPK12 caused the protrusion of the arginine sidechain on the surface of the protein, which could affect its binding affinity for other proteins in addition to an altered structure of the loop region. Similarly, the Arg in position 55 of MPK4 protruded from the surface as compared to the wild-type. Thus, the MPK12 G53R and MPK4 G55R amino acid substitutions may alter protein binding affinities of these MPKs to other proteins.

Collectively, the presented experiments suggest that the CO₂ signal leading to stomatal movements is transmitted through MPK12 and MPK4, leading to inhibition of HT1, and this enables SLAC1 activation by its activators, including OST1 and GHR1. Neither MPK12 G53R from Cvi-0 nor MPK4 G55R can fully inhibit HT1 (Fig 7).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 7. Schematic model of molecular events during elevated CO₂-induced stomatal closure.

(Left guard cell) CO₂ enters guard cells through the PIP2;1 aquaporin [20] and is converted to bicarbonate by carbonic anhydrases βCA4 and βCA1. The mechanism by which bicarbonate is sensed in guard cells still needs to be resolved; nevertheless, it is likely that in elevated CO₂ conditions, activation of MPK12 and MPK4 leads to inhibition of HT1, and this enables activation of slow-type anion channel SLAC1 by OST1 [15]. Additionally, GHR1 participates in the regulation of SLAC1 activity and is involved in CO₂-induced stomatal closure [18]. Bicarbonate-induced inhibition of HT1 by RHC1 has also been shown [19]. (Right guard cell) The G53R mutation in Cvi-0 MPK12, as well as G55R mutation in MPK4, decreases the ability of these MPKs to inhibit HT1 kinase activity, which results in enhanced inhibition of SLAC1 activity by HT1 and decreased sensitivity to CO₂ in stomatal closure.

https://doi.org/10.1371/journal.pbio.2000322.g007

Plant Material and Growth Conditions

Col-0, Col-gl, Cvi-0, gdsl3-1 (GABI-492D11; CS447183), cas-1 (SALK_070416), cas-2 (GABI-665G12), and cas-3 (SAIL_1157_C10) were from the European Arabidopsis Stock Centre (www.arabidopsis.info). Seeds of ht1-2 were a gift from Dr. Koh Iba. Col-0 × Cvi-0 RILs were obtained from INRA Versailles. The abi1-1 allele used was in Col-0 accession. Double mutants and other crosses were made through standard techniques and genotyped with PCR-based markers (S1 Table).

For ozone screening, seeds were sown at high density on a 1:1 v/v mixture of vermiculite and peat (type B2, Kekkilä, Finland), and kept for 2 d at 4°C for stratification. The plants were grown in controlled growth chambers (Bio 1300, Weiss Umwelttechnik, Germany) under a 12 h photoperiod, with a 23°C/19°C day/night temperature and a 70%/90% relative humidity or in growth rooms with equivalent growth conditions. The average photosynthetic photon flux density during the light period was 200 μmol m^-2 s^-1. When seedlings were 1 wk old, they were transplanted into 8 × 8 cm pots at a density of five plants per pot.

Three-week-old plants were exposed to ozone in growth chambers under the same conditions as they were grown until the experiments. Ozone exposure was acute (300–350 ppb for 6 h) and started 2 h after light was switched on. Ozone damage was visualized with trypan blue stain or quantified as electrolyte leakage.

Mapping of Cvi-0 Ozone Sensitivity QTLs

NILs were created by crossing Col-0 with Cvi-0 and selecting the most ozone-sensitive plant in F2 and backcrossing to Col-0 for eight generations (generating Col-S) or selecting the most tolerant plant and backcrossing to Cvi-0 for six generations (generating Cvi-T). The genomes of Cvi-0 and Cvi-T were sequenced at BGI Tech Solutions (Hong Kong) with Illumina technology, and the genomes of Col-S and Cvi-T were sequenced at the DNA Sequencing and Genomics lab of the University of Helsinki with SOLiD technology. Genome sequence data is available from the NCBI BioProject database with the accession number PRJNA345097. The 90-bp-long Illumina paired end sequencing library reads were mapped onto the Col-0 reference genome (TAIR10) with using the Bowtie2 aligner (version 2.0.0-beta7; [40]) in “end-to-end” alignment mode, yielding an average genomic sequence coverage of 45-fold. Variation calling and haplotype phasing was performed with the help of samtools (tools for alignments in the SAM format, Version: 0.1.18; [41]). Based on the aligned sequences, various PCR-based markers (S1 Table) were designed to genotype Cvi-0 versus Col-0 in the NILs and informative RILs from the INRA Versailles Col-0 × Cvi-0 RIL population. The markers were also used to genotype ozone-sensitive individuals from segregating F2 populations.

Mapping of cis Mutation

Mapping population was created by crossing cis (Col-0) and C24 as an Arabidopsis accession with low stomatal conductance. High water loss from excised leaves and decreased responses to high CO₂ were used as a selective trait. Rough mapping with 22 markers using 59 F2 samples showed linkage to the bottom of chromosome 2, at the marker UPSC_2–18415 at 18.4 Mbp. Pooled genomic DNA from 66 selected F3 plants was used for sequencing. Whole genome sequencing was conducted with Illumina HiSeq 2000, and the reads were mapped against Col-0 genome (release TAIR10) by BGI Tech Solutions (Hong Kong). Genome sequence data is available from the NCBI BioProject database with the accession number PRJNA345097 and PRJNA343292. For mapping the genomic area of the mutation, the Next Generation Mapping tool was used [42], which positioned the mutation on chromosome 2 between 18,703,644 –19,136,098 bp. The deletion mutation in cis was verified by PCR to be 4,770 bp (at the position 18,945,427–18,950,196 bp).

Complementation Lines

MPK12 and its promoter were amplified from Col-0 or Cvi-0 genomic DNA using Phusion (Thermo Fisher Scientific) and Gateway (Invitrogen) cloned into entry vector pDONR-Zeo. Subsequently, the genes were cloned into pGWB13 and pMCD100. Plants were transformed with floral dipping [43].

Southern Blotting Analyses

Total DNAs from different genotyping plants were extracted by CTAB method, and 12 micrograms of total DNA was digested by HindIII or EcoRI. The DNAs were running on the gel and transformed onto Nylon membrane. Hybridization was performed with digoxigenin-labeled specific genomic DNA amplified by primers F3 and R4 for 12 h. The membrane was washed several times by washing buffer and Maleic acid buffer. The membrane was blocked by blocking solution for 1 h at room temperature and washed and incubated with anti-DIG-AP for 30 min. Detection was performed using substrate DIG CSPD.

Plant Growth and Experimental Settings for Gas Exchange Measurements

Seeds were planted on a soil mixture consisting of 2:1 (v:v) peat:vermiculite and grown through a hole in a glass plate covering the pot as described previously [44]. Plants were grown in growth chambers (MCA1600, Snijders Scientific, Drogenbos, Belgium) at 12 h/12 h day/night cycle, 23°C/20°C temperature, 100 μmol m^-2 s^-1 light, and 70% relative humidity (RH). For gas exchange experiments, 24- to 30-d-old plants were used.

Stomatal conductance of intact plants was measured using a rapid-response gas exchange measurement device consisting of eight through-flow whole-rosette cuvettes [44]. The unit of stomatal conductance mmol m^-2 s^-1 reflects the amount of H₂O moles that exits the plant through stomata per one m² of leaf area per second. Prior to the experiment, plants were acclimated in the measurement cuvettes in ambient CO₂ concentration (~400 ppm), 100 μmol m^-2 s^-1 light (if not stated otherwise), and ambient humidity (RH 65%–80%) for at least 1 h or until stomatal conductance was stable. Thereafter, the following stimuli were applied: decrease or increase in CO₂ concentration, darkness, reduced air humidity, and ozone. CO₂ concentration was decreased to 100 ppm by filtering air through a column of granular potassium hydroxide. In CO₂ enrichment experiments, CO₂ was increased by adding it to the air inlet to achieve a concentration of 800 ppm. Darkness was applied by covering the measurement cuvettes. In blue light experiments, dark-adapted plants were exposed to blue light (50 μmol m^-2 s^-1) from an LED light source (B42180, Seoul Semiconductor, Ansan, South Korea). The decreased or increased CO₂ concentration, darkness, and blue light were applied for 58 min. In the long-term elevated CO₂ experiment (Fig 1D and S1E Fig), CO₂ concentration was increased from 400 ppm to 800 ppm for 2.5 h. To calculate stomatal half-response time, the whole 2.5-h stomatal response to elevated CO₂ was scaled to a range from 0% to 100%, and the time when 50% of stomatal closure had occurred was calculated. Humidity was decreased by a thermostat system to 30%–40% RH, and stomatal conductance was monitored for another 56 min. In ozone experiments, the plants were exposed to 350–450 ppb of ozone for 3 min and stomatal conductance was measured for 60 min after the start of the exposure.

In ABA-induced stomatal closure experiments, 5 μM ABA solution was applied by spraying as described in [45]. At time point 0, plants were removed from cuvettes and sprayed with either 5 μM ABA solution (5 μM ABA, 0.012% Silwet L-77 [PhytoTechnology Laboratories], and 0.05% ethanol) or control solution (0.012% Silwet L-77 and 0.05% ethanol). Thereafter, plants were returned to the cuvettes and stomatal conductance was monitored for 56 min.

In ABA-induced inhibition of stomatal opening experiments, plants were acclimated in measurement cuvettes in darkness. At time point 0, plants were removed from cuvettes and sprayed with 2.5 μM ABA solution (2.5 μM ABA, 0.012% Silwet L-77 [PhytoTechnology Laboratories], and 0.05% ethanol) or control solution (0.012% Silwet L-77 and 0.05% ethanol). Thereafter, plants were returned to the cuvettes, dark covers were removed, and stomatal conductance was monitored in light for 56 min.

Prior to the measurement of the diurnal pattern of stomatal conductance, plants were preincubated in the measurement cuvette for at least 12 h in respective light and humidity conditions. Plants were measured in 16-min intervals. WUE was calculated based on the data of diurnal experiments as an average of daytime light period (from 9:00 to 17:00).

CO₂-induced stomatal conductance in S2 Fig was measured as following. Five-week-old healthy plants growing in a growth chamber with 70% humidity and a 16 h light/8 h dark condition were used for stomatal conductance analyses at different CO₂ concentrations by a LiCOR-6400XT, as previously described [13]. Relative stomatal conductance values were normalized relative to the last data point preceding the [CO₂] transitions (400 to 800 or 1,000 ppm).

Stomatal Aperture

The MPK12 deletion mutant mpk12-4 and wild-type plants were grown in a growth chamber at 70% humidity, 75 μmolm^-2 s^-1 light intensity, 21°C, and 16 h light/8 h dark regime. Leaf epidermal layers from 2-wk-old plants of both genotypes were preincubated in an opening buffer (10 mM MES, 10 mM KCl, and 50 mM CaCl₂ at pH 6.15) for 2 h, and stomata were individually imaged and tracked for measurement before treatment. After that, the leaf epidermal layers were incubated with buffers containing 10 μM ABA for 30 min and the individually tracked stomata were imaged. Stomatal apertures were measured by ImageJ software and genotype-blind analyses were used. The data presented are means and SEM n = 3 experiments, with 30 stomata per experiment and condition.

Stomatal Index and Density

Plants at the age of 28–30 d were used for stomatal index and density measurements. Rosette leaves of equal size were excised, and the abaxial side was covered with the dental resin (Xantopren M mucosa, Heraeus Kulzer, Germany). Transparent nail varnish was applied onto the dried impressions after the removal of the leaves. The hardened nail varnish imprints were attached onto a microscope glass slide with a transparent tape and imaged under a Zeiss SteREO Discovery.V20 stereomicroscope. For quantification, an image with the coverage of 0.12 mm² was taken from the middle of the leaf, next to the middle vein. In total, 81–84 plants per line from two independent biological repeats were analyzed—one leaf from each plant, one image from each leaf. Stomatal index was calculated with the following formula: SI = Stomatal density / (Density of other epidermal cells + Stomatal density).

Stomatal Complex Length

For the stomatal complex length measurements, plants at the age of 28–35 d were used. Whole leaves were preincubated for 4 h abaxial side down in the buffer (10 mM MES, 5 mM KCl, 50 μM CaCl₂, pH 6.15 [with TRIS]) in the light. Four to six plants per genotype and one leaf per plant were analyzed, and altogether 84–126 stomatal complexes per genotype were measured.

Y2H Interaction Tests

Interactions between MPK12 and selected protein kinases and phosphatases were tested in pairwise split-ubiquitin Y2H assays using the DUALhunter and DUALmembrane 3 kits (Dualsystems Biotech). For bait construction, the coding sequences of MPK12 were PCR-amplified from total cDNAs from Col-0 and Cvi-0. Other MPK12 variants with point mutations (K70R, Y122C, and D196G+E200A) were created by two-step overlap PCR using the Col-0 MPK12 as a template. HT1 was also PCR-amplified from Col-0 cDNA. All MPK12s and HT1 were digested with SfiI and cloned to the corresponding site in pDHB1, which contained the Cub-LexA-VP16 fusion. For prey constructs, coding sequences of each selected gene were amplified from total Col-0 cDNAs, digested with SfiI, and cloned into either pPR3-N (HT1, OST1, BLUS1, IBR5, MKP2, MPK12, MPK12G53R, MPK11) or pPR3-STE (SnRK2.2, SnRK3.11, ABI1, ABI2, HAB1, HAB2), which contained a mutated NubG. All primers used are listed in Table S1. The pAI-Alg5 with a native NubI was used as a positive prey control, whereas the pDL2-Alg5 containing NubG served as a negative control.

For pairwise Y2H assays, the yeast strain NMY51 was cotransformed with bait and prey plasmids and grown on SD-Leu-Trp plates to select for presence of both plasmids. At least ten colonies from each transformation were pooled and resuspended in water to an OD600 of 0.5, from which 100, 1,000, and 10,000x serial dilutions were prepared and spotted on SD-Leu-Trp and SD-Leu-Trp-His-Ade plates. SD-Leu-Trp plates were incubated at 30°C for 2 d, photographed, and used for β-galatosidase overlay assays. SD-Leu-Trp-His-Ade plates were incubated for 2–4 d and photographed. The quantitative β-galactosidase assay was performed with three pools of ten independent colonies from each pairwise combination using the Yeast β-galactosidase assay kit (Thermo Fisher Scientific) by the nonstop quantitative method.

Ratiometric BiFC Assay

Binary constructs containing split YFPs were designed and generated for cloning genes of interest by the ligation independent cloning (LIC) method as described in [18].

Each gene of interest was amplified by two consecutive PCR reactions: first with gene-specific primers and later with a pair of universal primers designed specifically for the LIC method. All primers used are listed in S1 Table. To prepare vectors for LIC, plasmids of 35S:YFPn and 35S:YFPc were linearized by PmlI digestion, followed by T4 DNA polymerase treatment with dGTP to create 15–16 nucleotide 5ʹ-overhangs. For insert preparation, the final PCR products of target genes were incubated with T4 DNA polymerase in the presence of dCTP to create the complementary overhangs with the vectors. Both vector and insert were mixed at room temperature and proceeded with Escherichia coli transformation after 5 min. The final constructs were sequence verified and transformed to Agrobacterium tumefaciens GV3101 for agro-infiltration experiments.

For the ratiometric BiFC assays, four different agrobacterial clones—each harboring a YFPn fusion, a YFPc fusion, the SLAC1-CFP internal control, or the gene silencing suppressor P19—were co-infiltrated to the leaves of N. benthamiana at an OD600 of 0.02 for each clone in the infiltration buffer (10mM MES, 10mM MgCl₂, 200 μM acetosyringone). Images were acquired at 3 dpi with a Zeiss LSM710 confocal microscope using a 63x objective (for high magnification images) or a 20x objective (for fluorescence quantification). The YFP signals were excited by a 514 nm laser, and emission between 518–564 nm was collected. The CFP signals were excited by a 405 nm laser, and emission at 460–530 was collected. Z-stack images of approximately 15 μm thickness were collected, and all images were acquired at the 16-bit depth for a higher dynamic range. The fluorescence intensity was measured by the ImageJ software. The leaf samples used for imaging were collected and used for protein extraction followed by western blot analysis.

Western Blot Analysis

The leaf samples (30–40 mg) were ground under liquid nitrogen and boiled for 10 min in 100 μL of 6X Laemmli buffer. 12 μL of each sample were separated on 10% SDS polyacrylamide gel. After SDS-PAGE, proteins were transferred onto nitrocellulose membrane. Immunodetection of HA-tagged proteins was performed with a monoclonal anti-HA antibody.

Split Luciferase Complementation Assay

The MPK12 cDNA was cloned into a vector containing the N-terminal half of luciferase (nLUC) and HT1 was cloned into the cLUC. The constructs in the A. tumefaciens strain GV3101 were co-infiltrated into N. benthamiana leaves with P19 at an OD600 of 0.8. The infiltrated leaves after 3 d of infiltration were harvested for bioluminescence detection. Images were captured with a CCD camera.

Measurement of S-type Anion Currents

Arabidopsis guard cell protoplasts were isolated as described previously [46]. Guard cell protoplasts were washed twice with a washing solution containing 1 mM MgCl₂, 1 mM CaCl₂, 5 mM MES, and 500 mM D-sorbitol (pH 5.6 with Tris). During patch clamp recordings of S-type anion currents, the membrane voltage started at +35 to –145 mV for 7 s with –30 mV decrements, and the holding potential was +30 mV. The bath solutions contained 30 mM CsCl, 2 mM MgCl₂, 10 mM MES (Tris, pH 5.6), and 1 mM CaCl₂, with an osmolality of 485 mmol/kg. The pipette solutions contained 5.86 mM CaCl₂, 6.7 mM EGTA, 2 mM MgCl₂, 10 mM Hepes-Tris (pH 7.1), and 150 mM CsCl, with an osmolality of 500 mmol/kg. The free calcium concentration was 2 μM. The final osmolalities in both bath and pipette solutions were adjusted with D-sorbitol. Mg-ATP (5 mM) was added to the pipette solution before use. 13.5 mM CsHCO₃ (11.5 mM free [HCO₃^-] and 2 mM free [CO₂]) was freshly dissolved in the pipette solution before patch clamp experiments. The concentrations of free bicarbonate and free CO₂ were calculated using the Henderson–Hasselbalch equation (pH = pK1 + log [HCO₃^-] / [CO₂]). pK1 = 6.352 was used for the calculation. [HCO₃^-] represents the free bicarbonate concentration and [CO₂] represents the free CO₂ concentration.

Protein Expression and Purification

For in vitro kinase assays, the respective sequences of HT1, HT1 K113M, MPK11, MPK12, MPK12 G53R, MPK12 K70R, and MPK12 Y122C were cloned into a pET28a vector (Novagen, Merck Millipore) using primers listed in S1 Table. Point mutations corresponding to K113M in HT1, K70R in MPK12, and Y122C in MPK12 were created with two-step PCR using primers listed in S1 Table. MPK4 was cloned as previously described [18].

6xHis-HT1WT, 6xHis-HT1 K113M, 6xHis-MPK12, 6xHis-MPK12 G53R, 6xHis-MPK12 K70R, 6xHis-MPK12 Y122C, 6xHis-MPK11, 6xHis-MPK4 WT, and 6xHis-MPK4 G55R were expressed in E. coli BL21(DE3) cells. A 2 mL aliquot of an overnight culture was transferred to a fresh 1 L 2xYT medium and grown at 37°C to an absorbance of ~0.6 at OD600. The cultures were chilled to 16°C and recombinant protein expression was induced by 0.3 mM isopropyl b-D-thiogalactopyranoside for 16 h. The cells were harvested by centrifugation (5,000 rpm, 10 min, 4°C) and stored at –80°C until use.

All purification procedures were carried out at 4°C. The cells were resuspended in 30 mL of lysis buffer (50 mM Tris-HCl [pH 7.4], 300 mM NaCl, 5% [v/v] glycerol, 1% [v/v] Triton X-100, 1 mM PMSF, 1 μg/ml aprotinin, 1 μg/ml pepstatin A, 1 μg/ml leupeptin) and lysed using an Emulsiflex C3 Homogenizer. Cell debris was removed by centrifugation at 20,000 rpm for 30 min. The protein-containing supernatant was mixed for 1 h at 4°C with 0.20 mL of Chelating Sepharose Fast Flow resin (GE Healthcare), charged with 200 mM NiSO₄ and pre-equilibrated in the lysis buffer. The protein–resin complex was packed into a column, and the beads were washed with 5x10 column volumes (CV) of a wash buffer I (50 mM Tris-HCl [pH 7.4], 600 mM NaCl, 5% [v/v] glycerol, 1% [v/v] Triton X-100), 5x10 CV of a wash buffer II (50 mM Tris-HCl [pH 7.4], 300 mM NaCl, 5% [v/v] glycerol, 0.1% [v/v] NP-40), and 2x10 CV of a wash buffer III (50 mM Tris-HCl [pH 7.4], 150 mM NaCl, 5% [v/v] glycerol, 0.1% [v/v] NP-40). The protein was eluted by incubating the beads for 5 min at room temperature with an imidazole-containing elution buffer (50 mM Tris-HCl, 150 mM NaCl, 5% [v/v] glycerol, 0.1% [v/v] NP-40, 300 mM imidazole). MPK12 proteins were concentrated and imidazole was removed by Millipore Amicon Ultra-0.5 Centrifugal Filter Concentrators (NMWL 3000). Glycerol was added to a final concentration of 20% (v/v), and 20 μL aliquots of the eluted protein were snap-frozen in liquid nitrogen and stored at –80°C.

In Vitro Kinase Assays

Protein concentrations were estimated on 10% SDS-polyacrylamide gel using BSA as a standard. HT1 kinase activity assay was performed by incubating a constant amount of purified recombinant HT1 and 0–30 μM MPK12, 0–20 μM MPK4, or 0–10 μM MPK11 in a reaction buffer (50 mM Tris-HCl [pH 7.4], 150 mM NaCl, 20 mM MgCl2, 60 mM imidazole, 1 mM DTT, 0.2 mg/ml insulin) at room temperature for 10 min. Then, casein (1 mg/ml), 500 μM ATP, and 100 μCi/ml 32P-γ-ATP were added and reaction aliquots were taken at the 30 min time point. Reactions were stopped by the addition of SDS loading buffer. Proteins were separated on a 10% SDS-polyacrylamide gel and visualized by Coomassie brilliant blue R-250 (Sigma) staining. HT1 activity was determined by autoradiography and quantified by ImageQuant TL Software.

Model of MPK12 and MPK4

Sequence searches and alignments were conducted with SWISS-MODEL [47]. The crystal structure with the best sequence identity and resolution was selected for building homology models. Arabidopsis MPK12 and MPK4 have sequence identity to the 3 Ångstrom resolution Arabidopsis MPK6 structure (5CI6; [33]) of 64.61% and 70.67%, respectively. This structure was then used to construct models for the wild-type and mutant structures. The RMSD from aligning the structures for MPK12 and MPK12 G53R was 0.324 Ångstroms (i.e., a close structural similarity). Structures were checked for clashes and with quality controls and were then superposed.

Statistical Analysis

Statistical analyses were performed with Statistica, version 7.1 (StatSoft Inc., Tulsa, Oklahoma, United States). All effects were considered significant at p < 0.05.