The plant hormone abscisic acid (ABA) plays important roles in plant growth and development (Finkelstein, 2013). Plants are exposed to various adverse environmental conditions such as drought, high salinity, and extreme temperatures during their life cycle. Under these conditions, ABA level increases, and ABA mediates protective responses to the abiotic stresses (Nambara and Marion-Poll, 2005). For instance, ABA minimizes water loss through transpiration by promoting stomatal closure and, at the same time, inhibiting stomatal opening under water deficit condintion (Kim et al., 2010). It also controls the synthesis of compatible osmolytes and coordinates other responses to protect cells from the damage caused by various stresses (Xiong et al., 2002). Additionally, ABA functions, in general, as a negative regulator of plant growth under the stress conditions (Finkelstein, 2013). It inhibits seed germinaton, seedling establishment and the onset of postgermination growth, and subsequent seedling growth. Cellular ABA level is high also in developing seeds, and it plays a key role in the synthesis of storage components and establishing seed dormancy during seed maturation (Holdsworth et al., 2008). Most of the cellular processes controlled by ABA entail changes in gene expression, and numerous genes are known to be regulated by ABA (Fujita et al., 2011).

The molecular components comprising the ABA response pathway have been identified by various genetic and biochemical studies, and the core signaling pathway leading to the ABA-dependent gene expression has been unraveled (Cutler et al., 2010; Yoshida et al., 2015). ABA is perceived by PYR/PYL/RCAR family receptor proteins (Ma et al., 2009; Park et al., 2009). The ABA-bound receptor proteins interact with the members of the clade A subfamily of PP2Cs (e.g, ABI1, ABI2, and HAB1). The PP2Cs are negative regulators of ABA signaling and, in the absence of ABA, inactivate a subset of SnRK2s (e.g., OST1/SnRK2.6/SnRK2E, and SnRK2.2/SnRK2D, and SnRK2.3/SnRK2I) (Fujii and Zhu, 2009; Fujita et al., 2009). The receptor binding to the PP2Cs inhibits the phosphatase activity of PP2Cs, resulting in the activation of the SnRK2s. The activated SnRK2s phosphorylate the ABF/AREB/ABI5 family of bZIP class transcription factors that regulate ABA-responsive gene expression. Thus, the core ABA signaling pathway that leads to the ABA-regulated gene expression consists of PYR/PYL/RCAR-PP2Cs-SnRK2s-ABFs/AREBs/ABI5 (Fujii et al., 2009).

Mitogen-activated protein kinase (MAPK) signaling cascades play diverse roles in eukaryotic cell signaling (Colcombet and Hirt, 2008; Rodriguez et al., 2010). A typical MAPK cascade consists of three MAP kinases: MAP kinase kinase kinase (MAPKKK/MAP3K), MAPkinase kinase (MAPKK/MAP2K), and MAP kinase (MAPK). The signaling module, which is universal among eukaryotic kingdom, is activated by various internal or external stimuli, such as mitogen, hormones, developmental signal, pathogen, and abiotic stresses. The kinases comprising the module is sequentially activated by a relay of phosphorylation. A MAP3K, which is activated by a specific stimulus or stimuli, phosphorylate MAP2K, which, in turn, phosphorylate MAPK. The activated MAPKs phosphorylate downstream targets to modify their activities.

In the Arabidopsis genome, approximately 80 MAP3Ks, 10 MAP2Ks, and 20 MAPKs have been identified (Colcombet and Hirt, 2008). Numerous studies have been carried out to determine the functions of individual MAPKs, especially those of MAP2Ks and MAPKs. The studies show that they are involved in various cellular processes, such as plant development, hormone responses, disease resistance, and abiotic stress responses (Colcombet and Hirt 2008; Rodriguez et al., 2010). In several cases, the entire signaling cascades comprised of MAP3K-MAP2K-MAPK have been delineated. For instance, the MEKK1-MKK2-MPK4/6 modules are involved in salt and cold response (Teige et al., 2004), the MEKK1-MKK4/5-MPK3/6 modules function in innnate immunity (Asai et al., 2002), and the YODA-MKK4/5-MPK3/6 modules regulate the stomatal development (Lampard et al., 2009).

Some MAPKs are involved in ABA response (Liu, 2012). The MKK1-MPK6 module mediates ABA response during seed germination by regulating CAT1 expression and glucose-induced ABA level (Xing et al., 2008; 2009). MPK9 and MPK12 positively regulate both ABA- and MeJA-dependent guard cell signaling (Jammes et al., 2009; Khokon et al., 2015). Additionally, a large number of MAPKs in Arabidopsis and other plant species are known to be induced or activated by ABA and abiotic stresses (Danquah et al., 2014). Although the role of individual MAPKs and their signaling module in ABA signaling have been established for a number of cases, specific functions of MAPKs remain largely unknown. This is especially so in the case of MAP3Ks. MAP3Ks constitute the largest MAPK family (i.e., approximately 80 members), but only a few of them have been charaterized functionally (Rodriguez et al., 2010). MEKK1 regulates innnate immunity and abiotic stress responses, as mentioned above, and YODA controls stomatal patterning (Lampard et al., 2009; Teige et al., 2004). ANP1 and its homologs, on the other hand, positively regulate cytokinesis (Takahashi et al., 2010). Most recently, MAP3K17/18 have been reported to mediate ABA response via the MKK3-MPK1/2/7/14 pathway (Danquah et al., 2015; Matsuoka et al., 2015). MAP3K17/18 is activated by ABA, and their mutants are ABA-hypersensitive. Mitula et al. (2015) further reported that MAP3K18 activity is regulated by ABI1.

We are interested in the roles of MAP3Ks in ABA and/or abiotic stress signaling and set out to investigate the ABA-associated functions of several MAP3Ks. We chose MAP3K16 and other related MAP3Ks for our study. MAP3K16 is one of the several salt-inducible MAP3Ks, together with MAP3K14, MAP3K15, MAP3K17, and MAP3K18 (i.e., MAP3K14/15/17/18). MAP3K17 and MAP3K18 are also ABA-inducible. As a first step toward their functional analysis, we investigated whether MAP3K16 and the related MAP3Ks, MAP3K14/15/17/18, are involved in ABA response. In this paper, we show that MAP3K16 regulates a subset of ABA response and present the data suggesting its possible substrates. Our results further indicate that MAP3K14/15/17/18 play similar roles in ABA response.

Plant growth and RNA isolation

Plants (Arabidopsis thaliana, ecotype Ler and Col-0) were grown at 22°C under long day condition (16 h light/8 h dark cycle). For aseptic growth, seeds were placed on MS medium supplemented with 1% sucrose and solidified with 0.8% agar. If necessary, the MS medium was supplemented with ABA, salt or other supplements as indicated in the Figure legends. Seeds were placed at 4°C for 3–5 days in the dark to break residual dormancy before being transferred to normal growth temperature. For soil growth, seeds were sown on 1:1:1 mixture of vermiculite, perlite, and peat moss, and, after stratification, placed at normal growth condition. Plants were watered once a week by immersing pots in 0.1% Hyponex (Hyponex Co., USA) solution.

RNA isolation, semi-quantitative RT-PCR, and real-time RT-PCR were performed described previously (Lee et al., 2015). Briefly, RNA was isolated employing the RNeasy plant mini kit (Qiagen) and treated with DNase I to remove contaminating DNA. For RT-PCR, the first strand cDNA was synthesized using Superscript III (Invitrogen), and qRT-PCR was carried out in a Bio-Rad CFX96 real-time PCR system using the primers shown in the Supplementary Table S2. UBC9 (At4g27960) (Czechowski et al., 2005) was used as a reference gene.

Yeast two-hybrid assay and other interaction study

Yeast two-hybrid assays were performed as described before (Choi et al., 2005; Lee et al., 2009). To prepare bait constructs, insert fragments were amplified using the primers shown in the Supplementary Table S2 and cloned into the pPC62LexA vector (Lee et al., 2009). The prey constructs were prepared by cloning the individual amplified fragments into pYESTrp2 (Invitrogen). Bait constructs were individually introduced into the reporter yeast L40 (MATα, his3Δ200, trp1-901, leu2-3112, ade2, LYS2::[LexAop(x4)-HIS3], URA3:: LexAop[x8]-LacZ, GAL4) (Invitrogen; Carlsbad, Calif) by transformation, and relevant prey constructs were subsequently introduced. To test interactions, yeast transformants were grown on SC-His medium, and the LacZ activity was assayed by X-gal overlay assay (Duttweiler, 1996) or by the liquid assay using O-nitrophenyl-β-D-galactopyranoside (ONPG) as a substrate.

Pulldown assays were performed as described before (Choi et al., 2005) with minor modification. For ABR1, maltose–binding protein (MBP) was used as a negative control, and glutathione S-transferase (GST) was used as a negative control for MKKs (see below). Approximately 5 μg of fusion proteins, 30 μl of MBP (ABR1) or GST (MKKs) resin (50% slurry), and 20 μl of in vitro translation products were used in each binding reaction. For in vitro translation, the MAP3K16 coding region was cloned into the Eco RI/Xho I sites of pCITE-4a (Novagen; USA). In vitro translation was carried out employing the TNT in vitro translation kit (Promega; Madison, WI) according to the supplier’s instruction. BiFC was performed according to Walter et al. (2004). MAP3K16, ABR1, and MKK3 were individually cloned into pSPYNE-35S and pSPYCE-35S, respectively. Pairwise combinations of the constructs (i.e., two pairs for ABR1 and two pairs for MKK3) were then used to infiltrate tobacco leaves (N. benthamiana). Because epifluorescence signals were very weak, protoplasts were prepared (Yoo et al., 2007) from the tobacco leaves 3–5 days after infiltration and observed under the microscope (Olympus BX51). Fluorescence signals were detected from the pairs shown in Figs. 6C and 7C.

Preparation of recombinant proteins and in vitro kinase assay

MAP3K16 recombinant proteins were prepared employing the pMAL system (NEB) with the modification described previously (Lee et al., 2015). The entire coding region (amino acids 1-443) or the kinase domain (amino acids 1-263) were amplified and cloned into the Sal I/Eco RI sites of pMAL-c5X with a 6X His-tag. Protein induction and protein purification were carried out according to the supplier’s instruction, and the proteins were further purified employing the Ni-NTA resin (Qiagen). ABR1 recombinant protein was prepared in a similar way. The entire ABR1 coding region (amino acids 1-391) was amplified and cloned in to the Nde I/Sal I sites of pMAL-c5X (NEB), and protein was purified by employing amylose resin followed by Ni-NTA resin. Primer sequences are shown in the Supplementary Table S2. MKK3 and other MKK (Supplementary Figs. S4 and S6) recombinant protein were prepared using a GST-fusion vector, pColdIII-GST-His. The vector was constructed by cloning the GST coding region into Nde I/Kpn I sites of the pColdIII vector (Takara) and then by inserting a 6X His-tag next to the Xba I site. The entire coding region of each MKK was amplified using the primer set in the Supplementary Table S2 and cloned into pColdIII-GST-His. Protein induction and purification was carried out according to the GST Gene Fusion System manual (GE Healthcare), and then further purified on Ni-NTA resin.

Kinase assays were carried out as described (Choi et al., 2005; Lee et al., 2015). Briefly, approximately 0.5–1 μg of MAP3K16 recombinant proteins were incubated with similar amounts of substrates in a buffer (25 mM Tris–HCl, pH7.5, 10mM MgCl₂, 10 μM ATP) containing 2 μCi of γ-³²P for 30 min at 30°C. After the reaction, the reaction mixtures were separated by SDS-PAGE, and gels were stained with Coo-massie Brilliant Blue R, dried and autoradiographed.

Expression patterns of Arabidopsis MAP3K14-18

To examine the stress-induction pattern of MAP3K16 expression, we conducted real-time RT-PCR using RNA isolated from seedlings. MAP3K16 transcript level was increased 12-fold by salt and 3-fold by mannitol (Table 1). Similar induction levels were observed with MAP3K15. The salt induction levels of MAP3K17 and MAP3K18 were much higher, 62-fold and 333-fold, respectively. In addition, the expression of the two MAP3K genes was highly induced by mannitol (33-fold and 413-fold, respectively) and ABA (11-fold and 69-fold, respectively). The induction levels of MAP3K14 expression was relatively lower (i.e., 5–6 fold by high salt and high osmolarity) (Table 1). However, the basal level of MAP3K14 expression (i.e., the expression level under normal growth condition) is higher than other MAP3Ks, and thus, its expression level after induction is comparable to those of other MAP3Ks.

We investigated the expression pattern of MAP3K16 further by determining its tissue-specific expression pattern. Semi-quantitative RT-PCR (Fig. 1A) showed that MAP3K16 is expressed in leaves, roots, flowers, and siliques. The transcript level in leaves was lower than in other tissues. To determine temporal and spatial expression patterns, transgenic plants harboring a 2.2.kb promoter-GUS reporter construct were prepared, and the promoter activity was assayed by histochemical GUS staining. Figure 1B shows that GUS activity was observed in young, emerging leaves (panel a) and lateral roots of seedlings (panel b). In mature plants, GUS activity was observed in lateral roots (Fig. 1B, b), anthers, and stigma (Fig. 1B, c). Embryos also exhibited GUS activity (Fig. 1B, d).

MAP3K16 OX lines are partially ABA-insensitive

To investigate the in vivo function of MAP3K16, we generated its overexpression (OX) lines. The coding region of MAP3K16 was fused to the constitutive 35S promoter in pBI121 (Jefferson et al., 1987), and transgenic plants were prepared using the construct. Thirteen homozygous lines were recovered, and, after preliminary analysis, three representative lines were chosen for phenotype analysis (Fig. 2A).

Plants overexpressing MAP3K16 grew normally, except that they grew faster than wild type plants and, therefore, have a tendency to reach flowering stage slightly earlier (Supplementary Fig. S1). To assess the effect of MAP3K16 overexpression on ABA response, we examined the ABA- and stress-associated phenotypes of the OX lines. The most notable phenotype was the partial insensitivity to ABA during early seedling growth stage (Figs. 2B and 2C). When wild type seeds were germinated and grown on the media containing various concentrations of ABA, inhibitory effect of ABA was evident, and the postgermination growth (i.e., shoot and root growth) of wild type seedlings was severely inhibited when ABA concentration was greater than 0.75μM. ABA inhibition of seedling growth was also observed with the MAP3K16 OX lines. However, the transgenic plants grew much more efficiently than the wild type plants, suggesting that they were less sensitive to ABA inhibition. To assess the effects of MAP3K16 overexpression in more detail, we determined the ABA sensitivity during three different stages of early seedling growth: Germination, cotyledon greening/expansion, and primary root elongation. Figure 2D shows that the relative germination rates of transgenic seeds in the presence of ABA were higher than the wild type rates, although the differences were not great. Cotyledon greening/expansion efficiencies of the transgenic seedlings were much higher compared with those of the wild type plants (Fig. 2E). On the other hand, primary root elongation of the transgenic seedlings was more severely affected by ABA than wild type seedlings, although the degree of difference was very low (Fig. 2F). Collectively, the data indicate that ABA sensitivity of the MAP3K16 OX lines was dependent on developmental stage and that the most prominent growth-related phenotype was the ABA insensitivity during the post-germination seedling establishment (i.e., cotyledon greening) stage.

Stress responses of MAP3K16 OX lines

Because MAP3K16 expression is induced by high salt, we also investigated the salt sensitivity of the MAP3K16 OX lines. The result (Figs. 3A–3C) showed that both germination and cotyledon greening of the transgenic plants were less sensitive to salt than wild type plants. The partial insensitivity to high salt was observed only at the germination/cotyledon greening stage, and no significant phenotypic changes were observed at later growth stages in salt-containing media (data not shown). The root elongation rates in the presence of salt were similar to the wild type rates, suggesting that the salt sensitivity of primary root elongation was not significantly affected in the MAP3K16 OX lines.

ABA plays a pivotal role in mediating water-deficit response. Thus, the reduced ABA sensitivity during early seedling growth suggested that the MAP3K16 OX lines might exhibit altered water-deficit response, presumably reduced tolerance. To test the possibility, we determined the survival rates of the transgenic plants under water-deficit condition. Figures 3D and 3E show that the survival rates of the transgenic lines in soil were lower than the wild type rates. We also determined the transpiration rates of the plants to gain information about the possible mechanism of the reduced tolerance to water-deficit. Figure 3F shows that detached leaves of the MAP3K16 OX lines lost water faster than the wild type leaves. Thus, our result indicates that the water loss rates of the transgenic leaves were higher.

Phenotypes of a MAP3K16 knockout mutant

To further investigate the function of MAP3K16, we acquired its knockout (KO) line, map3k16, (Figs. 4A and 4B) and examined its ABA sensitivity (Figs. 4C–4E). Figure 4D shows that the germination rates of the map3k16 seeds were more severely affected than the wild type seeds in the presence of various concentrations of ABA. Similarly, the efficiency of cotyledon greening was also more severely affected by ABA than wild type seedlings (Fig. 4E). Thus, the map3k16 mutant exhibited ABA-hypersensitive phenotypes during germination and cotyledon greening stage, although the degree of hypersensitivity was relatively low. In addition, the map3k16 mutant was hypersensitive to high salt during germination and cotyledon greening (Figs. 4F–4H). Overall, our results indicate that the map3k16 mutant displayed weak ABA-hypersensitive phenotypes, whereas its OX lines exhibited partial ABA insensitivity during early seedling growth (i.e., germination and cotyledon greening/expansion stage).

Phenotypes of other MAP3K OX and KO lines

As mentioned earlier, MAP3K14/15/17/18 are similar to MAP3K16 in their ABA- and stress-induced expression patterns (Table 1) and are homologous to MAP3K16 to various degrees (i.e., 35 to 64% amino acid sequence identity, Supplementary Fig. S2A). To address whether they also function in ABA and stress responses, we prepared their OX lines and analyzed their phenotypes (Fig. 5).

The plants overexpressing each of MAP3K14/15/17/18 grew normally except that, like the MAP3K16 OX line plants, they reached bolting stage slightly earlier than wild type plants (Supplementary Fig. S2B). We first determined the ABA sensitivity of the MAP3K14 OX lines. Figures 5A–5C show that the transgenic plants were partially insensitive to ABA during early seedling growth stage. Similarly, the OX lines of MAP3K15/17/18 were partially insensitive to ABA during the same growth stage (Figs. 5D–5L). We next investigated the drought tolerance of the OX lines by scoring the survival rates of seedlings under water-deficit condition. The result showed that the OX lines were more susceptible to water stress than wild type seedlings (Supplementary Fig. S3). Additionally, we acquired and analyzed MAP3K15/17/18 KO mutants (see Materials and Methods) and found that they displayed similar phenotypes to those of the map3k16 mutant (summarized in Supplementary Table S1). Collectively, our data indicate that the OX and KO phenotypes of MAP3K14/15/17/18 are similar to those of MAP3K16.

MAP3K16-interacting proteins

MAP3K16 and MAP3K14/15/17/18 belong to the MAP3K family proteins (Ichimura et al., 2002). Because MAP3Ks are components of the MAP kinase module consisted of MAP3K-MAP2K-MAPK, we addressed whether the MAP3Ks could interact with MAP2Ks by carrying out yeast two-hybrid assays. Bait constructs containing each of the MAP3Ks were prepared, and their interactions with MKKs were examined. The result (Fig. 6) showed that MAP3K16 interacted with MKK3 (Fig. 6A). The interaction was relatively weak (Fig. 6B), but it was consistently detected not only in the two-hybrid assay but also in GST pulldown assay and in bimolecular fluorescence complementation (BiFC) assay. In the pulldown assay (Supplementary Fig. S4), in vitro translated MAP3K16 was preferentially retained by MKK3. In the BiFC assay (Fig. 6C), fluorescence signals, which were distributed throughout the cell, were detected in tobacco cells infiltrated with Agrobacteria containing the YFP^N-MAP3K16 and the YFP^C-MKK3 construct. Interactions with MKK2/4/5 were also observed occasionally. However, the interactions with these MKKs were very weak (Fig. 6B) and inconsistent, i.e., the results of our interaction assays varied from experiment to experiment (e.g., compare Fig. 6A and Supplementary Fig. S4). Among other MAP3K-MKK pairs, we detected a strong interaction between MAP3K17 and MKK3 (Supplementary Fig. S5).

It is possible that MAP3K16 may interact with proteins other than MKKs. To address the possibility, we performed yeast two-hybrid screening employing the full-length MAP3K16 as bait. We screened approximately 4.5 million yeast transformants obtained with a cDNA expression library prepared from salt- and ABA-treated seedling RNAs (Choi et al., 2000). Several putative positive clones, including APUM10 (At1g35750), CAD7 (At4g37980), ABR1 (At5g64750), and GDA1 (At4g19180), were isolated from the screen. Among the isolates, we chose ABR1 for further analysis, because it is known to be a negative regulator of ABA response (Pandey et al., 2005). As shown in Fig. 7A, ABR1 interacted with the full-length MAP3K16, and it also interacted with the kinase domain alone albeit with lower intensity (Fig. 7A). The interaction between the full-length MAP3K16 and ABR1 was confirmed by pulldown assays (Fig. 7B), in which in vitro translated MAP3K16 was retained by MBP (maltose-binding protein)-tagged ABR1 (Fig. 7B, lane 2) but not by MBP alone (Fig. 7B, lane 1). In addition, BiFC assays were performed to further confirm the MAP3K16-ABR1 interaction (Fig. 7C). In the assay, fluorescence signals, which appeared as a single broad spot in the cell, were observed in tobacco cells infiltrated with Agrobacteria harboring the YFP^N-MAP3K16 and the YFP^C-ABR1 constructs.

Once confirmed the interaction between the full-length MAP3K16 and ABR1, we performed a deletion study to map the interaction domain of ABR1. Various ABR1 deletion constructs shown in Figs. 7D and 7E were prepared, and their interactions with the full-length MAP3K16 were investigated by two-hybrid assay (Fig. 7F). Deletion of the N-terminal portion (amino acids 1-180) did not abolish the interaction, indicating that the remaining portion, i.e., fragment AP2C consisted of the AP2 domain (amino acids 181-260) and the C-terminal portion (amino acids 261-391) is sufficient for the interaction. On the other hand, the C-terminal deletion (amino acids 261-391) abolished the interaction, suggesting that it is essential for the interaction and that the remaining portion (i.e., fragment NAP2, amino acids 1-260) is not sufficient for the interaction. Although necessary, C-terminal portion alone could not interact with MAP3K16. In summary, the deletion analysis showed that both AP2 domain and the C-terminal portion of ABR1 are necessary for the interaction.

In vitro kinase activity of MAP3K16

To investigate the enzymatic activity of MAP3K16, we carried out in vitro kinase assays. We first determined whether MAP3K16 possessed kinase activity. Recombinant MAP3K16 proteins, full-length or a N-terminal partial fragment containing the kinase domain only, were prepared as described in Materials and Methods. Their phosphorylation activity was then determined employing myelin basic protein (MyBP) as a substrate. Figure 8A shows that the full-length MAP3K16 could phosphorylate MyBP (lanes 1, 2). The partial fragment consisted of the MAP3K16 kinase domain (amino acids 1-263) also could phosphorylate MyBP (lanes 3, 4). Compared with the full-length protein, the partial fragment, which is considered to be a constitutively active form (Rodriguez et al., 2010), exhibited much higher phosphorylation activity.

Next, we investigated whether MAP3K16 could phosphorylate ABR1. In the above assay to examine MyBP phosphorylation, recombinant ABR1 was also phosphorylated by the constitutive active form (i.e., the kinase domain-containing fragment) of MAP3K16 (Fig. 8A, lane 4), when it was added to the reaction mixture with MyBP. In a separate assay employing ABR1 as an only substrate, ABR1 was phosphorylated by MAP3K16 (Fig. 8B, lanes 3, 4). Thus, our results indicate that the constitutive active form of MAP3K16 could phosphorylate ABR1 in vitro. Similarly, we also examined whether MAP3K16 could phosphorylate MKK3. Recombinant MKK3 protein was prepared as described in Materials and Method, and kinase assays were carried out. The result showed that MAP3K16 could phosphorylate MKK3 (Fig. 8C, lane 3). The constitutively active form of MAP3K16 may be lacking the putative regulatory domain (Rodriguez et al., 2010) and, consequently, may have lost its substrate specificity. To address the question, we examined the phosphorylation of several other MKKs, and the result showed that MKK3 was preferentially phosphorylated by the active form of MAP3K16 (Supplementary Fig. S6).

The primary goal of our current study is to find out MAP3Ks that are involved in ABA and stress responses. Toward the end, we searched database and chose MAP3K16 and several other ABA- or salt-inducible MAP3Ks for functional analyses.

In our transgenic analyses, overexpression of MAP3K16 conferred partial ABA insensitivity during early seedling growth stage and the susceptibility to water-deficit condition. On the other hand, the MAP3K16 KO line displayed opposite phenotypes, notably, the ABA hypersensitivity during early seedling growth. The degree of hypersensitivity, however, was low probably because of the functional redundancy discussed below.

ABA affects several aspects of plant growth and development (Finkelstein, 2013), such as seed germination, seedling establishment (i.e., cotyledon greening/expansion and onset of vegetative growth) and subsequent postgermination growth, stress tolerance, and seed dormancy. Although ABA generally plays an inhibitory role during germination and postgermination growth, it inhibits postgermination seedling growth more efficiently than the germination process (Lopez-Molina et al., 2001). The ABA insensitivity of the MAP3K16 OX lines was observed at this stage, and seed germination was not affected (Supplementary Fig. S1). In contrast, primary root elongation of transgenic seedlings was slightly hypersensitive to ABA inhibition (Fig. 2F), suggesting that MAP3K16 OX phenotypes are developmental stage-dependent. Another noticeable ABA-associated phenotype of the MAP3K16 OX lines was the susceptibility to water deficit, which may result from higher transpiration rates (Figs. 3D–3F). The reduced water-deficit tolerance is consistent with the ABA insensitivity observed at the seedling establishment stage.

Our transgenic analyses of other salt-inducible MAP3Ks (summarized in the Supplementary Table S1), MAP3K14/15/17/18, indicate that they are likely to play similar roles to MAP3K16 in ABA and stress responses. The OX lines of the genes were ABA-insensitive during seedling establishment (i.e., cotyledon greening/expansion) stage and susceptible to water deficit (Fig. 5 and Supplementary Fig. S3). Like MAP3K16 OX lines, the OX lines exhibited weak ABA-insensitivity during germination and weak ABA-hypersensitivity during root growth. Additionally, their KO lines exhibited similar phenotypes to those of the MA3K16 KO line (Supplementary Table S1). High degree of functional redundancy is, therefore, expected among the salt-inducible Arabidopsis MAP3Ks, and the weak KO phenotypes of MAP3K16 and MAP3K14/15/17/18 may be due to this functional redundancy.

It is noteworthy that the results of our transgenic analyses suggest the negative regulatory role of MAP3K14/15/16/17/18 in ABA response. The conclusion is based on our observation of OX phenotypes and the phenotypes of single KO lines without complementation analyses. Thus, the results must be interpreted with caution. Nonetheless, most distinct OX phenotypes (i.e., ABA insensitivity during seedling establishment stage and drought susceptibility) were observed with all OX lines (Supplementary Table S1), and opposite and complementary phenotypes were observed with their respective KO lines. Collectively, the results strongly support the conclusion that MAP3K14/15/16/17/18 play mainly negative regulatory roles in ABA response.

Recombinant MAP3K16 protein possesses kinase activity (Fig. 8), and, in an effort to identify putative MKK substrates, we found out that it interacts with MKK3 (Fig. 6, Supplementary Fig. S4). Consistent with this observation, recombinant MKK3 was phosphorylated by MAP3K16 in vitro (Fig. 8C). It remains to be determined whether MKK3 is phosphorylated by MAP3K16 in vivo. Nonetheless, the result suggests that MAP3K16 may function via the MAP kinase pathway reported by Danquah et al. (2015) (see below). They showed that MAP3K17/18 mediate ABA response by the MAP3K17/18-MKK3-MPK1/2/7/14 pathway. We also observed the MAP317 interaction with MKK3 (Supplementary Fig. S5A), although the MAP3K18 interaction with MKK3 was not observed presumably because of the lower sensitivity of our two-hybrid assay system.

Our results suggest that ABR1 may also be a substrate of MAP3K16. ABR1 was isolated as one of the positive clones in the two-hybrid screen to isolate MAP3K16-interacting proteins. Additionally, recombinant ABR1 was phosphorylated by MAP3K16 in vitro (Figs. 8A and 8B). As in the above case of MKK3, it remains to be determined whether ABR1 is a MAP3K16 substrate in planta. However, it is noteworthy that ABR1 is one of the few negative regulators of ABA response (Pandey et al., 2005) and that MAP3K16 plays mainly a negative regulatory role. In line with the notion that ABR1 may be an in vivo substrate of MAP3K16, several genes (COR15A, COR47, RAB18, RD29A, RD22, and RD29A) that are up-regulated in the abr1 mutant were found to be down-regulated in the MAP3K16 OX lines in our limited analysis of gene expression changes (Supplementary Fig. S7). There is a precedent in which a MAP3K phosphorylates a substrate other than MKKs. For instance, Arabidopsis MAP3K CTR1 phosphorylates EIN2 (Ju et al., 2012; Qiao et al., 2012) as well as MKK9 (Yoo et al., 2008).

While our work is in progress, the function of MAP3K17 and MAP3K18 has been reported. According to Danquah et al. (2015), MAP3K17/18 activate MKK3, which in turn activates MPK1/2/7/14. The MAPK cascade is activated by the core ABA signaling module, and the mkk3 and the map3k17map3k18 mutants are ABA-hypersensitive. These mutant phenotypes are consistent with the map3k16 phenotypes we observed. Matsuoka et al. (2015) reported similar results (i.e., MAP3K18-MKK3-MPK1/2/7 pathway) and showed that MAP3K18 is involved in leaf senescence process. Mitula et al. (2015), on the other hand, reported that MAP3K18 affects ABA sensitivity and stomatal index. Additionally, they showed that ABI1 regulates the MAP3K18 activity by promoting proteasomal degradation of MAP3K18.

In summary, our study demonstrated that MAP3K16 functions mainly as a negative regulator of ABA response during early seedling growth and in water-deficit response, although it plays a positive role in the regulation of root growth. MAP3K16 possesses a kinase activity and could phosphorylate ABR1, a negative regulator of ABA response. MAP3K16 also interacts with MKK3 and phosphorylate it in vitro, suggesting that, like MAP3K17/18, it may also function in the MKK3-MPK1/2/7/14 pathway. Other salt-inducible MAP3Ks, MAP3K14/15/17/18, exhibited OX and KO phenotypes similar to those of MAP3K16, suggesting that they may play similar roles in ABA response.