Phosphorylation of the CAMTA3 Transcription Factor Triggers Its Destabilization and Nuclear Export^,

Phosphorylation of the CAMTA3 transcription factor triggers its destabilization and nuclear export.

Abstract

The Arabidopsis (Arabidopsis thaliana) calmodulin-binding transcription activator3 (CAMTA3) is a repressor of immunity-related genes but an activator of cold-induced or general stress-responsive genes in plants. Post-transcriptional or posttranslational mechanisms have been proposed to control CAMTA3 functions in different stress responses. Here, we show that treatment with the bacterial flg22 elicitor induces CAMTA3 phosphorylation, which is accompanied by its destabilization and nuclear export. Two flg22-responsive mitogen-activated protein kinases (MAPKs), MPK3 and MPK6, directly phosphorylate CAMTA3, with the phospho-sites contributing to CAMTA3 degradation and suppression of downstream target gene expression. However, the flg22-induced nuclear export and phospho-mobility shift can still be observed for the CAMTA3 phospho-null variant of the MAPK-modified sites, suggesting additional flg22-responsive kinases might be involved. Taken together, we propose that flg22-induced CAMTA3 depletion facilitates de-repression of downstream defense target genes, which involves phosphorylation, increased protein turnover, and nucleo-cytoplasmic trafficking.

Unlike animals, plants do not possess a circulatory adaptive immune system. Instead, they rely on an efficient cellular innate immunity system to defend against various pathogens. The first layer of this immunity system is initiated by perception of pathogen-associated molecular patterns (PAMPs) or microbe-associated molecular patterns by pattern recognition receptors. A well-studied system is the recognition of the bacterial flagellin-derived flg22 peptide PAMP by the FLS2 receptor (Gómez-Gómez and Boller, 2000; Chinchilla et al., 2007). Ligand recognition and binding trigger receptor activation (Zipfel et al., 2004; Chinchilla et al., 2007; Roux et al., 2011) and transduce this into intracellular signaling events, including a rise in the cytosolic Ca²⁺ level (Ranf et al., 2011), production of extracellular reactive oxygen species (Kadota et al., 2014; Li et al., 2014), and activation of mitogen-activated protein kinase (MAPK) cascades (Asai et al., 2002) and calcium-dependent protein kinases (CDPKs; Boudsocq et al., 2010). This signaling network induces transcriptional and metabolic reprogramming to establish pattern-triggered immunity (PTI). A second layer of immunity is the so-called effector-triggered immunity (ETI) that is activated after direct or indirect recognition of pathogen effectors or their modifications of host proteins, respectively. However, a distinction between PTI and ETI is not always clear (Thomma et al., 2011).

In Arabidopsis (Arabidopsis thaliana), the best studied stress-responsive MAPKs are MPK3, MPK4, and MPK6, which are known to be activated during pathogen infection or elicitation with PAMPs, such as flg22 and elf18 (Asai et al., 2002). A putative MPK4 paralog, MPK11, was subsequently identified as a fourth PAMP-activated MAPK (Eschen-Lippold et al., 2012). The activated MAPKs phosphorylate a variety of substrates to transduce upstream signals to trigger further defense responses. These substrates could be transcriptional/translational regulators, structural components, or enzymes. Phosphorylation may affect the function of the substrates through altered protein stability, cellular localization, enzyme activity, or the ability to bind to other partners (Lee et al., 2015; Zhang et al., 2016). Therefore, identifying the MAPK targets responsible for regulating the downstream immunity activation is essential for engineering pathogen resistance (Hoehenwarter et al., 2013; Lassowskat et al., 2014; Rayapuram et al., 2014).

Ca²⁺ influx is one of the earliest signaling events after PAMP perception and plays a crucial role in PTI (Ranf et al., 2011). The stimulus-specific spatio-temporal features of Ca²⁺ signals are referred to as Ca²⁺ signatures (Cheval et al., 2013). Ca²⁺-sensing proteins, such as calmodulin (CaM), CaM-like proteins, CDPKs, and calcineurin B-like proteins (DeFalco et al., 2010), read and transduce these Ca²⁺ signatures into downstream cellular responses (Seybold et al., 2014). While calcineurin B-like proteins need to recruit additional partners for further signal transduction, CDPKs are both sensor and signal relay in one entity. CDPKs are calcium-regulated Ser/Thr kinases that are unique to plants (Boudsocq and Sheen, 2013). Four Arabidopsis CDPKs, i.e. CPK4, CPK5, CPK6, and CPK11, are transiently activated upon flg22 perception and act as positive regulators in PTI (Boudsocq et al., 2010). CaM is highly conserved in eukaryotes, and seven CaM isoforms are encoded in the Arabidopsis genome (Ranty et al., 2006). A conformational change of CaM induced by Ca²⁺ binding can promote interaction with downstream targets that contain CaM-binding domains (CaMBDs), such as transcription factors, kinases, or metabolic enzymes (DeFalco et al., 2010; Cheval et al., 2013).

The Arabidopsis CaM-binding transcription activator3 (CAMTA3, AT2G22300, also abbreviated as SR1) belongs to a six-membered gene family (Finkler et al., 2007). Key conserved domains include an N-terminal CG-1 domain that binds a conserved 6-bp motif (with the consensus sequence [A/C/G]CGCG[G/T/C]) in the promoter of target genes (Yang and Poovaiah, 2002) and two isoleucine-glutamine (IQ) motifs and a CaMBD near the C terminus, which are thought to mediate Ca²⁺-independent and -dependent CaM binding, respectively (Finkler et al., 2007; Du et al., 2009). Transcriptomics analysis of two transfer DNA insertion knockout mutants (camta3-1 and camta3-2) revealed upregulation of many defense genes (e.g. PRs, non-race-specific disease resistance1 [NDR1], PAD4, ZAT10, and various WRKY transcription factors; Galon et al., 2008). Under normal growth conditions (at 19°C to 21°C), both camta3 mutants display reduced growth, chlorotic lesions, constitutive expression of PR genes (systemic acquired resistance [SAR]-associated marker genes) and enhanced resistance to bacterial and fungal pathogens. Hypersensitive response and SAR-related features such as accumulation of salicylic acid (SA), reactive oxygen species, and autofluorescent compounds are also enhanced (Galon et al., 2008; Du et al., 2009; Nie et al., 2012). Thus, CAMTA3 is thought to be a suppressor of defense responses in Arabidopsis, by directly binding to the promoter of target genes at specific “CGCG”-containing cis-elements. In agreement, chromatin–immunoprecipitation assay showed CAMTA3 binding to the promoters of defense regulators such as Enhanced Disease Susceptibility1 (EDS1), a positive regulator of SA biosynthesis (Du et al., 2009), NDR1, and ethylene insensitive3 (Nie et al., 2012). During pathogen infection, repressor functions of CAMTA3 are alleviated through its degradation. This proteasome-dependent degradation of CAMTA3 is promoted via interaction with SR1 Interaction Protein1 (SR1IP1), a substrate-adaptor for cullin3-based E3 ubiquitin ligase (Zhang et al., 2014). Contrary to the above-described understanding of CAMTA3 function as a negative regulator, the enhanced defense phenotype of camta3 is recently reported to be a form of ETI-like autoimmunity triggered by the activation of two nucleotide-binding domain Leu-rich repeat-containing proteins (Lolle et al., 2017). A recent transcriptome analysis further pinpoints CAMTA3 as an early convergence point between PTI and ETI (Jacob et al., 2018). Thus, despite ambiguity about its negative regulator function, CAMTA3 has been independently isolated in a number of disease resistance genetic screens and is clearly important for plant immunity (Du et al., 2009; Jing et al., 2011; Nie et al., 2012).

In contrast to its role as a repressor, CAMTA3 was initially reported to act as transcriptional activator in cold stress response. For instance, expression of the cold-responsive gene C-repeat-Binding factor2 is positively regulated by CAMTA3 in response to cold stress (Doherty et al., 2009). It was also found to be a transcriptional activator in general stress response and regulation of glucosinolate metabolism (Laluk et al., 2012; Benn et al., 2014). Therefore, CAMTA3 appears to function as a transcriptional repressor for immunity-related genes but acts as a transcriptional activator when bound to promoters of genes involved in cold acclimation or other general stresses. How CAMTA3 switches between activator and repressor functions remains unclear. One possibility is through molecular interactions, either with other proteins or intramolecularly. A recent study showed that in unstressed plants, CAMTA3 represses transcription of SA pathway genes through an N-terminal repression module (NRM) of CAMTA3; cold stress promotes CaM binding to IQ and/or CaMB domains, resulting in a conformation change that interferes with the repressor activity of the NRM (Kim et al., 2017). Additionally, post-translational modifications such as phosphorylation may also be involved (Zhang et al., 2014). Proteomics analysis identified CAMTA3 as a putative MPK3/6 substrate (Hoehenwarter et al., 2013). In addition, in the PhosPhAt database (an Arabidopsis database for experimentally determined phosphorylation sites; Heazlewood et al., 2008), five phosphopeptides are annotated, which contain typical MAPK phosphorylation sites.

In this work, we explored if CAMTA3 is phosphorylated during PTI and how this might affect its function. We report that CAMTA3 is phosphorylated by flg22-responsive MAPKs, which promotes its destabilization. We further investigated whether SR1IP1 (Zhang et al., 2014) is involved in regulating this phospho-dependent destabilization but could not confirm physical interaction between CAMTA3 and SR1IP1, showing instead that they have distinct cellular localization and therefore cannot physically interact before PTI activation. Flg22-induced CAMTA3 destabilization is also not compromised in the sr1ip1 mutant, but appears to be stabilized by overexpression of SR1IP1. Finally, we propose that there may be additional flg22-responsive kinases that phosphorylate CAMTA3, which may include CDPKs or other downstream kinases.

RESULTS

flg22 Induces CAMTA3 Phosphorylation and Destabilization In Vivo

The function of CAMTA3 is ultimately determined by its protein expression within cells, which is controlled by transcriptional, post-transcriptional, and post-translational processes. For instance, it was shown recently that a 3′region within the open reading frame controls mRNA stability during salt stress (Abdel-Hameed et al., 2020). Therefore, we first investigated how PAMP treatment affects CAMTA3 expression at the mRNA level. In agreement with PAMP-inducible accumulation of CAMTA3 transcripts (Hruz et al., 2008), reverse-transcription quantitative PCR (RT-qPCR) showed a rapid (30–120 min) but transient induction of CAMTA3 expression after flg22 treatment in Arabidopsis Col-0 seedlings (Supplemental Fig. S1). Next, we investigated how flg22 treatment might affect CAMTA3 protein levels. Due to the lack of specific anti-CAMTA3 antibodies, western-blot analysis was performed with protein extracted from Arabidopsis protoplasts that transiently express hemagglutinin (HA)-epitope–tagged CAMTA3 under the control of a cauliflower mosaic virus 35S promoter. CAMTA3 levels appeared to be unchanged or only slightly reduced by flg22 treatment (see 3-h time point; Fig. 1A). Note that this may be due to the 35S promoter, which although typically used as a constitutive promoter, is also weakly flg22-responsive (J.L., own observations; see also the discussion in the legend to Supplemental Fig. S1C). Furthermore, expression driven by such a strong promoter can mask subtle changes in protein levels. Hence, to improve visualization of protein turnover, cycloheximide (CHX) was included during elicitation to stop translation. Compared to the water-treated control samples, the CAMTA3 protein level was reduced upon flg22 treatment (in the presence of CHX; Fig. 1A, top), suggesting that CAMTA3 has a shorter half-life after flg22 treatment. Proteasome-mediated degradation is involved for this destabilization because it could be blocked by MG115 cotreatment (Fig. 1A, bottom). Hence, in agreement with the repressor functions of CAMTA3, the findings support a model where CAMTA3 destabilization upon PAMP treatment may lead to its removal from promoters and the de-repression of defense gene expression. On the transcriptional level, flg22-inducible CAMTA3 expression in the native context (Supplemental Fig. S1) may reflect the replacement of the degraded CAMTA3 proteins and hence a negative feedback regulation to restore repression.

Figure 1.

Flg22 induces CAMTA3 phosphorylation and destabilization in vivo. A, Flg22-induced CAMTA3 protein destabilization is dependent on the proteasome pathway. Top, Protoplasts transfected with plasmids expressing HA-epitope–tagged CAMTA3 were treated with 100 nm of flg22 or water as a control, and proteins were extracted for standard SDS-PAGE and western-blot analysis (anti-HA antibody). Where indicated, CHX (1 μm) cotreatment was included to block protein translation and facilitate visualization of differential protein turnover. Bottom, Protoplasts were pretreated with the proteasome inhibitor MG115 (50 µmfor 30 min) before elicitation. B, Flg22-induced phospho-mobility shift of CAMTA3. Top, Extracted proteins were separated by Manganese(II)–Phos-Tag–based SDS-PAGE and analyzed by western blot. Bottom, Protein extracts were incubated with λ-phosphatase (37°C, 10 min) before Phos-Tag–based separation and western blotting. Arrowhead marks the “unmodified” CAMTA3 and asterisk marks the various phosphorylated CAMTA3 forms. Amido-black staining of the large subunit of Rubisco was used as loading a control for all western blots. C, Analysis of CAMTA3 protein stability after cold treatment. Transfected protoplasts were resuspended in media at 4°C or room temperature (RT), harvested at the indicated time points, and processed for standard western blotting. D, Cold treatment was performed as described in C and the proteins analyzed for phospho-mobility shift of CAMTA3 after separation by Phos-Tag–based SDS-PAGE. Bottom, Western blot with phosphorylation specific α-pTEpY antibody was used to monitor MAPK activation. Flg22 treatment was used as a positive control for displaying the three PAMP-responsive MAPKs. Similar results were obtained four times for A and twice for all other experiments.

Phosphorylation has been speculated to be involved in the destabilization of CAMTA3 after bacterial infection (Zhang et al., 2014). We therefore investigated phosphorylation status upon flg22 elicitation. Here, we employed Manganese (II)-Phos-Tag–based western-blot analysis, where phosphorylated forms show retarded mobility in the gels compared to the nonphosphorylated proteins (Kinoshita et al., 2009). Unlike the well-defined single band in standard SDS-PAGE, CAMTA3 from unstressed or water-treated protoplasts appears as smeary multiple bands on Phos–Tag-based western blot, suggesting that it exists as several partially phosphorylated forms. By contrast, a mobility shift was seen for CAMTA3 from flg22-treated protoplasts within 10 min of treatment (Fig. 1B, top). Upon λ-phosphatase treatment of the protoplast extracts (Fig. 1B, bottom), the mobility shift is abrogated and CAMTA3 migrated as a well-defined band, suggesting that the smeary bands or bands with reduced mobility are due to phosphorylation. Taken together, the Phos-Tag and the phosphatase analyses indicate low levels of basal and a rapid flg22-induced CAMTA3 phosphorylation in vivo.

By contrast, cold stress (transfer to ice-cold media and placement in a 4°C cold room) did not induce any CAMTA3 degradation within the tested time period and in fact, there appeared to be enhanced protein stability (or less protein degradation) despite termination of protein translation by CHX treatment (Fig. 1C). This is in line with CAMTA3 functioning as a transcriptional activator for cold stress (Doherty et al., 2009). Cold treatment also did not lead to phospho-mobility shift of CAMTA3 in the Phos-Tag analysis (Fig. 1D, top), where MAPKs were correspondingly not activated within the tested time points (Fig. 1D, bottom). Hence, unlike PAMP (flg22) elicitation, cold stress does not trigger CAMTA3 phosphorylation and degradation, suggesting a differential mechanism exists to determine positive or negative functions of CAMTA3 in these distinct stress contexts.

CAMTA3 Interacts with and Is Phosphorylated by PAMP-Responsive MAPKs

MAPKs are rapidly activated after flg22 treatment and are thus candidate kinases responsible for the CAMTA3 phosphorylation. The best-studied MAPKs are MPK3, MPK4, and MPK6, which are known to be activated by flg22 (Asai et al., 2002). Reconstitution of yellow fluorescent protein (YFP) signals in the bimolecular fluorescence complementation (BiFC) assays suggest that CAMTA3 can interact directly with MPK3, MPK4, and MPK6 (Fig. 2A). By contrast, no YFP signal was seen with an unrelated MAPK (MPK8), suggesting some specificity of the BiFC assay. To investigate whether CAMTA3 is phosphorylated by flg22-responsive MAPKs, an in vitro kinase assay was performed with recombinant His-tagged CAMTA3 as a substrate and GST-tagged MPK3, MPK4, or MPK6. While nearly similar levels of in vitro autophosphorylation of the kinases were seen, CAMTA3 was phosphorylated by MPK3 and MPK6, but barely by MPK4 (Fig. 2B).

Figure 2.

CAMTA3 interacts with and is phosphorylated by PAMP-responsive MAPKs. A, CAMTA3 interaction with stress-activated MAPKs (MPK3, MPK4, and MPK6). BiFC assay was performed in protoplasts coexpressing CAMTA3 (fused with a C-terminal fragment of YFP, cYFP) and MPK3 or MPK4 or MPK6 (fused with an N-terminal fragment of YFP, nYFP). MPK8, a non-PAMP–responsive MAPK, was used as a negative control. Interaction is visualized as reconstituted YFP signals (which is representative of six observed protoplasts). Scale bars = 10 µm. Western-blot validation detecting intact fusion proteins is shown in Supplemental Figure S7A. B, In vitro kinase assay of recombinant CAMTA3 by PAMP-responsive MAPKs. Recombinant His-tagged CAMTA3 (as a substrate) was incubated with the indicated GST-tagged MAPKs in the presence of ³²P-labeled ATP (for 30 min at 37°C). After SDS-PAGE separation, phosphorylated proteins were visualized by autoradiography. Bottom is the corresponding CBB staining. This result has been reproduced in three independent experiments. C and D, In vivo phosphorylation of CAMTA3 through MPK3 and MPK6. C, Increased phosphorylation and destabilization of CAMTA3 after coexpression of constitutively-active (CA) MPK3 and MPK6. Protoplasts were transfected with the indicated constructs and the extracted proteins were analyzed by western blotting after separation on Phos-Tag gels or by conventional SDS-PAGE to monitor phospho-mobility shifts (top) or protein stability (middle), respectively. In the bottom, CA-MAPKs and wild-type–MAPKs transiently expressed in protoplasts were immunoprecipitated using anti-HA antibody and their autophosphorylation (in the presence of ³²P-labeled ATP) used to compare their kinase activities. Western blot was used to estimate the expression levels of the MAPKs. Note that the MPK6 immunological signals (see arrows) partially overlap with the IgG H chain signals. The result shown is representative of three independent experiments. D, Phospho-mobility shift and reduced protein stability of CAMTA3 upon coexpression of a constitutively active MKK5 (DD) or a kinase-deficient version (KR). CAMTA3 was transiently expressed in protoplasts together with MKK5^DD, which activates MPK3/6. The kinase-inactive variant MKK5^KR was used as a control. After an overnight incubation for protein expression, Phos-Tag–based or conventional SDS-PAGE–based western blots with the indicated antibodies were performed as described above. Expression of CAMTA3 was monitored with anti-HA antibody, and specific MPK3 and MPK6 phosphorylation by MKK5^DD with antipTEpY antibody, and MKK5 (DD and KR) with anticMyc antibody. Arrowhead idicates “Unmodified” CAMTA3; Asterisks indicate phosphorylated CAMTA3 forms. The result shown is representative of five independent experiments.

To verify in vivo phosphorylation, constitutively active MAPKs (CA-MAPKs; Berriri et al., 2012; Genot et al., 2017) were transiently co-expressed with CAMTA3 in protoplasts. Increased kinase activities of these CA-MAPKs compared to the native kinases were demonstrated by the higher autophosphorylation of the immunoprecipitated MAPKs. In vivo CAMTA3 phosphorylation and any effect on the protein levels were then monitored by western blot after Phos-Tag–based or standard SDS-PAGE separation, respectively. CAMTA3 was phospho-shifted when coexpressed with CA-MPK6 and accumulated to lower levels. For CA-MPK3 coexpression, some phosphorylation (albeit weaker when compared to MPK6) was observed but a strong CAMTA3 destabilization was seen. Consistent with the in vitro kinase assay, CA-MPK4 coexpression did not induce CAMTA3 phospho-shift (Fig. 2C). However, it should be noted that the CA-MPK4 displayed lower autophosphorylation activity than the other two kinases, and the introduced mutations near the activation loop might change the substrate specificities as compared to MAPKs that are natively activated through upstream MAPK kinases. Nevertheless, while bearing these two caveats in mind, MPK4 is probably not involved in inducing in vivo phosphorylation and destabilization of CAMTA3.

To validate our results and simulate flg22-activation of MAPKs, we transiently expressed in protoplasts a constitutively active MKK5 (MKK5^DD) to specifically activate endogenous MPK3 and MPK6 but not MPK4 (Lee et al., 2004; Lassowskat et al., 2014). In this case, MPK3/6 are activated “naturally” through phosphorylation of its kinase activation loop and not through mutation. Similar to observations above for CA-MPK3 and CA-MPK6, CAMTA3 protein level was reduced through MKK5^DD coexpression. The phosphorylated CAMTA3 proteins are difficult to visualize in the Phos-Tag gels because the strongly reduced protein levels are further separated into smeary, barely detectable, bands. However, the slower mobility of these fuzzy bands is in agreement with MPK3/6-mediated phosphorylation. Control transfection without any MKKs or cotransfection with a kinase-inactive MKK5 (MKK5^KR) did not show these effects (Fig. 2D). Hence, the results above suggest that two out of three known flg22-responsive MAPKs phosphorylate CAMTA3, and in vivo flg22-induced CAMTA3 destabilization may be triggered by phosphorylation via MPK3 and MPK6.

CAMTA3 Is Phosphorylated by PAMP-Responsive MPK3/6 at Multiple Sites, Which Contribute to CAMTA3 Protein Destabilization

After confirming in vitro and in vivo CAMTA3 phosphorylation by MPK3 and MPK6, we proceeded to identify the phospho-sites in CAMTA3. CAMTA3 contains 11 potential MAPKs phosphorylation sites, which are Ser or Thr followed by a Pro (S/TP; Fig. 3A). Both in vitro kinase assay (with nonradioactive ATP) and infiltrating flg22 to CAMTA3-overexpression lines were used for collecting phosphorylated CAMTA3 protein samples. MPK3- or MPK6-phosphorylated CAMTA3 bands, or flg22-induced phosphorylated CAMTA3 proteins immunoprecipitated from overexpression plant extracts, were excised after SDS-PAGE separation. After tryptic digestion and phospho-peptide enrichment, samples were subjected to mass spectrometry (MS) analysis (MS/MS spectra mapping the respective phospho-sites are shown in Supplemental Fig. S2). Three phospho-peptides with phosphorylated S272, S454, or S780 were detected (Fig. 3A, marked in red). The PhosPhAt 4.0 database (http://phosphat.uni-hohenheim.de/), an Arabidopsis phosphorylation site database incorporating experimental phosphoproteomic data (Heazlewood et al., 2008), annotates two further phosphorylation sites at S8 (Hoehenwarter et al., 2013) and S587 in CAMTA3 (Fig. 3A, marked in blue). Based on these data, we generated phospho-mutants: CAMTA3-mutP1 and -mutP2, with respectively three or five sites mutated to nonphosphorylatable residues (Fig. 3A). However, in vitro MPK3- or MPK6-mediated phosphorylation of the CAMTA3-mutP1 and -mutP2 variants was not reduced compared to CAMTA3–wild-type (Fig. 3B), suggesting the presence of additional MAPK-targeted sites.

Figure 3.

CAMTA3 is phosphorylated by MPK3/6 at multiple sites, which contribute to CAMTA3 destabilization. A, Schematic depiction of conserved domains within CAMTA3. CG-1, CG-box binding domain; TIG, immunoglobulin-fold domain (putative unspecific DNA binding); ANK, Ankyrin repeat. The 11 possible MAPK phospho-sites are indicated (color code: red, phospho-site identified by MS after tryptic digest; blue, phospho-site annotated in PhosPHAT; green, novel phospho-site identified after optimizing digestion). The embedded table summarizes the different phospho-site mutations generated in this study. Note that substitution to Ala was used to generate nonphosphorytable residues in all cases except at S8, S198, S272, T243, S469, and T736, where Gly was used due to the convenience of the cloning procedure. See “Materials and Methods” for details. B, In vitro kinase assay with wild-type (WT)– and phospho-site mutants of CAMTA3. The indicated CAMTA3 variants were incubated with active MPK3 or MPK6 and ³²P-labeled ATP. Shown are the autoradiographs of phosphorylated CAMTA3 bands and the corresponding CBB staining. C, MPK3/6-mediated phosphorylation of CAMTA3 promotes degradation. CAMTA3 variants were transiently coexpressed in protoplasts with constitutively active MKK5 (DD) or a kinase-deficient version (KR) as a negative control. Extracted proteins were analyzed by western blotting to detect the abundance of CAMTA3 (α-HA), MKK5 (α-cMyc), and the activated MPK3/6 (α-pTEpY). Amido-black staining of the Rubisco band served as a loading control. This experiment was repeated eight times and a representative blot shown. D, Quantitative western-blot estimation of CAMTA3 abundance. After transient expression and western blotting as described in C, CAMTA3 abundance was quantified as density of protein bands of five replicates (each consisting of independent protoplast samples). The absolute values from individual experiment were normalized by the median of each dataset. To take into account the different basal levels of the different protein variants, the percentage decrease comparing KR and DD was additionally calculated (within each experiment), using the formula (KR-DD)/KR × 100. Error bars = SEM. Statistically distinct groups after one-way ANOVA (P < 0.05) are marked with different lowercase letters.

Some regions of the CAMTA3 sequence cannot be detected with tryptic-digested material. To increase MS sequence coverage of CAMTA3, we repeated the experiment but digested the protein with both trypsin and endoproteinase Glu-C before MS analysis. This resulted in >60% sequence coverage that encompassed most of the SP/TP motifs, leading to the identification of phosphorylation at T243, S587, and S780 (Supplemental Fig. S2). Notably, T243 (Fig. 3A, marked in green) is a novel phospho-site, never detected in any previous reports. When the phospho-mutant CAMTA3-mutP2+ (containing phospho-site mutations at S8, T243, S272, S454, S587, and S780) was tested, in vitro MPK3- or MPK6-mediated phosphorylation was strongly reduced compared to CAMTA3–wild-type (Fig. 3B), so that these six sites are likely the major MPK3/6-targeted sites. We also generated CAMTA3-mutP3 (as a phospho-null for all 11 potential phospho-sites) and the corresponding phospho-mimetic mutant (CAMTA3-mimic), in which all 11 potential sites (S/T) were substituted by aspartic acids (D; Fig. 3A). Phosphorylation of these variants was completely lost (Fig. 3B), which also proved that MPK3/6 did not unspecifically phosphorylate other residues in CAMTA3.

To investigate the role of phosphorylation in protein destabilization, we compared the stability of the different CAMTA3 phospho-site variants. Coexpression of MKK5^DD (or the kinase-inactive MKK5^KR variant as a control) was employed to investigate the actions of MPK3/6. A quantitative western-blot system (Li-COR Odyssey CLx multiplex imaging system) was used to determine the protein levels in five replicates (consisting of independent protoplast samples). Figure 3C shows a representative western blot, and the bar chart in Figure 3D (top) shows the quantification of the protein levels. Reduction of CAMTA3–wild-type protein level was seen when MPK3/6 activation was induced by MKK5^DD coexpression (Fig. 3C); statistical analysis shows that this was also true for all the tested CAMTA3 variants (Fig. 3D). Because sustained activation through MKK5^DD may potentially activate the proteasome pathway (Lee et al., 2015), all CAMTA3 variants will eventually be degraded and subtle differences in protein stability may be partially masked. However, note that basal expressions (i.e. without MPK3/6 activation) of all the phospho-null mutants are higher levels than the CAMTA3–wild-type while the opposite is true for the CAMTA3-mimic (Fig. 3D). Thus, the MAPK-targeted phospho-sites may be involved in controlling CAMTA3 stability.

As seen in the representative western blot (Fig. 3C), the mutP2+ and mutP3 phospho-mutants appear slightly more stable than the CAMTA3–wild-type protein after MPK3/6 activation, which is substantiated by statistical analysis for the mutP3 variant. Although interpretation for mutP2+ lacked statistical robustness, this may be a type-II (false-negative) statistical error because we do see this qualitative difference in most of the independent replicates. As seen in the recalculation of the percent decrease (Fig. 3D, bottom), one datapoint (out of five) may be an outlier and the average percent decrease of mutP2+ is intermediate between wild type and mutP3. Taken together, these results indicate that CAMTA3 destabilization requires phosphorylation at multiple MPK3/6-targeted sites, including the six major phospho-sites analyzed in the mutP2+ variant.

Strength of Downstream Target Gene Suppression Correlates with Enhanced Stability of CAMTA3 Phospho-Variants

To investigate the impact on downstream target genes, we created overexpressing lines of the phospho-site variants. Where possible, two independent lines per construct were chosen (except the phospho-mimetic version, where only one transgenic line was recovered, and the mutP2+variant, which was not included due to poor transformation). Under control conditions, all of the CAMTA3-overexpression lines suppressed basal EDS1 expression to a similar extent when compared to the Col-0 background (Fig. 4A). While the phospho-mimic CAMTA3 is thought to be less stable, it behaved like the wild-type–CAMTA3, suggesting that sufficient proteins accumulate in this line for its repressor function. These data also imply that: the phosphorylation sites are not essential for repressor properties of CAMTA3, and even multisite mutations of all 11 phospho-sites (in the mut-P3 or phospho-mimic constructs) most likely did not compromise its tertiary protein structure substantially to impede DNA binding.

Figure 4.

Overexpression of the wild-type (WT)– or phospho-site–mutated CAMTA3 suppresses expression of a direct downstream target. Expression of EDS1 was analyzed by RT-qPCR in transgenic plants with the indicated genotype (OE = transgenic overexpression lines of the phospho-mutant variants described in Fig. 3). Four-week–old plants were infiltrated with water (A) or 1 µm of flg22 (B) for 1 h and leaf material collected for RT-qPCR gene expression or western-blot analysis (bottoms). For the gene expression, statistically distinct groups after one-way ANOVA (P < 0.05) are marked with different lowercase letters. Error bars = SEM; n = 4.

After flg22 treatment, EDS1 transcripts are generally 2- to 3-fold higher than in the water-treated controls. Similar to the water-treated controls (Fig. 4A), CAMTA3-overexpression suppressed the flg22-inducible expression when compared to the Col-0 (Fig. 4B). However, a differential effect between mutants is seen, with the repressive effect increasing progressively with the number of phospho-null mutations (i.e. the strongest repression was with the mut-P3 variant; Fig. 4B). Note that a similar trend can be seen for a second CAMTA3 direct target, CBP60g (Supplemental Fig. S3). A caveat here is that the mutP3 and also one of the mutP2 lines accumulate higher levels of the CAMTA3 proteins, which may be due to transgene copy numbers. However, all other transgenic lines are single-copy transfer DNA insertion and accumulate similar protein levels. Although a strong reduction of CAMTA3 levels is not seen at this short timepoint of flg22 treatment, our earlier analysis using CHX demonstrates that the phospho-mutants have longer half-lives. Hence, the data are consistent with a model where enhanced stability of the phospho-null CAMTA3 variants (e.g. mut-P2+ mut-P3) increases cellular CAMTA3 concentrations, shifting the DNA-binding equilibrium toward continued occupancy of cis-elements in target gene promoters. This would explain the stronger repression of EDS1 expression in the CAMTA3 overexpression lines both before and after flg22 treatment. Nevertheless, as in all mutational analysis, misfolding of the mutated protein and other indirect effects cannot be completely excluded.

The Role of SR1IP1 in CAMTA3 Stability

The mechanism behind CAMTA3 destabilization through phosphorylation is still unknown. A previous study proposed that the CAMTA3-interacting protein SR1IP1 recruits CAMTA3 for ubiquitination and degradation by the 26S proteasome (Zhang et al., 2014). To test the involvement of SR1IP1, we monitored flg22-induced destabilization of CAMTA3 in the sr1ip1-1 mutant background by transient protoplast transfection but could not discern any significant difference compared to the Col-0 background. However, when SR1IP1 was additionally overexpressed in the protoplasts, besides basal CAMTA3 proteins accumulating to higher levels, the flg22-induced degradation was also compromised (Fig. 5A). This means that while SR1IP1 is not required for CAMTA3 degradation per se, it does contribute to regulation of CAMTA3 protein levels during flg22 treatment. Thus, we next checked if phosphorylation affects the SR1IP1–CAMTA3 physical interaction. We performed several assays including yeast two-hybrid, BiFC, and split-luciferase assays in transfected Arabidopsis protoplasts (Supplemental Fig. S4), but none of these assays supported interaction between SR1IP1 and CAMTA3.

Figure 5.

Flg22 and MPK3/6-mediated phosphorylation induce nuclear export of CAMTA3. A, Flg22-induced CAMTA3 destabilization in an sr1ip1 mutant (SALK_064178) background. Flg22-induced changes in CAMTA3 turnover (i.e. with CHX cotreatment) was performed as described in Figure 1 and compared between protoplasts prepared from Col-0, sr1ip1 mutant, and additionally, transient overexpression of myc-tagged SR1IP1 in the sr1ip1 background. An exemplary immunoblot (top) and a quantitative analysis of CAMTA3 levels (bottom, mean ± se, n = 9) are displayed; the latter is pooled from three separate experiments (each consisting of three independent transfection subsets). All values are normalized to the average of the triplicate Col-0 values (set as 1) within each experiment. Statistically distinct groups after one-way ANOVA (P < 0.05) are marked with different lowercase letters. B, Distinct localization of CAMTA3 and SR1IP1. Top, CAMTA3-YFP was coexpressed with ERF104-CFP (used as a nuclear marker) in protoplasts. Bottom, SR1IP1-CFP was transiently expressed in protoplasts. After an overnight incubation period to enable protein expression, the protoplasts were observed by confocal microscopy. Position of chloroplasts are visualized through the chlorophyll autofluorescence. Scale bars = 10 µm. C, Relocalization of CAMTA3-YFP signals after flg22 treatment or coexpression with constitutively-active MKK5^DD. Protoplasts coexpressing CAMTA3-YFP and SR1IP1-CFP were treated with flg22 (100 nm, for 2 h) or water as control (-flg22) and observed by confocal microscopy. Bottom row shows additional cotransfection with MKK5^DD (to activate MPK3 and MPK6). The inactive MKK5^KR serves as a negative control. Protoplasts were observed by confocal microscopy after an overnight incubation. The right shows the same experimental setup except that the phospho-null CAMTA3-mutP3 was used in place of the wild-type (WT)–CAMTA3. Scale bars = 10 µm. Western-blot detection of intact proteins is shown in Supplemental Figure S7B. The red digits represent the number of protoplasts showing the depicted localization out of the total number observed (black digits).

MAPK-Induced Phosphorylation Triggers CAMTA3 Subcellular Relocalization

To clarify the discrepancy to the literature, we looked at the cellular distribution of the two proteins. CAMTA3 is a transcription factor and is mainly localized in the nucleus (Yang and Poovaiah, 2002), which we could also confirm in the unstressed protoplasts using a CAMTA3-YFP fusion construct (Fig. 5B, top). However, SR1IP1-cyan fluorecent protein (CFP) signals are in the periphery (presumably plasma membrane) of the protoplasts (Fig. 5B, bottom). Hence, the different localization might explain why we could not detect any interaction between SR1IP1 and CAMTA3 in any of the plant cell-based assays. For yeast two-hybrid assay, we employed a GAL4-based system while the reported SR1IP1-CAMTA3 interaction is based on the pSOS system (Zhang et al., 2014), which relies on myristylation of the bait proteins. Thus, the recruitment of the reporter to the plasma membrane may permit interaction. Nevertheless, the distinct localization of CAMTA3 and SR1IP1 suggests that they cannot physically interact unless there is a relocalization of either proteins from the nucleus or the plasma membrane, respectively.

Interestingly, most of the protoplasts showed a relocalization of the nuclear CAMTA3-YFP signal to the cytoplasm after flg22 elicitation, and multiple cytoplasmic aggregates were often observed (Fig. 5C; see also the time-course experiment in Supplemental Fig. S5A). The aggregates are not cleaved YFP fragments because western-blot analysis revealed intact CAMTA3-YFP protein bands (Supplemental Figs. S5B and S7B). By contrast, SR1IP1 localization was not visibly affected by flg22 treatment. Although CAMTA3 relocalization into the cytoplasm may potentially allow interaction with SR1IP1 and there seems to be partial colocalization between CAMTA3-YFP and SR1IP1-CFP after flg22 treatment (Fig. 5C), we never obtained any BiFC or split-LUC signals between CAMTA3 and SR1IP1 even upon flg22 elicitation (Supplemental Fig. S4). Nonetheless, we cannot exclude the possibility that CAMTA3 and SR1IP1 interacted upon flg22 elicitation, but was degraded too rapidly for detection of the complex.

Notably, activation of MPK3/6 (through MKK5^DD coexpression) can also induce CAMTA3 relocalization to the cytoplasm (Fig. 5C, lower left), thus phenocopying the effects of flg22 elicitation. The nonphosphorylatable CAMTA3-mutP3 remained nuclear-localized with MKK5^DD coexpression (Fig. 5C, lower right), demonstrating that the mutated phospho-sites are crucial for relocalization and that CAMTA3 nuclear export can be triggered by its MPK3/6-mediated phosphorylation. This enhanced retention in the nucleus could contribute to the enhanced repression of EDS1 expression by the phospho-site mutants observed in Figure 4. To our surprise, the phospho-sites also appear to be unnecessary for the flg22 response because the CAMTA3-mutP3 mutant can still relocalize to the cytoplasm upon flg22 treatment (Fig. 5C, top right). In summary, the MAPK-targeted phospho-sites are dispensable for the subcellular relocalization of CAMTA3 induced by flg22 treatment but are essential for the MPK3/6-induced pathway. Thus, during flg22 elicitation, alternative (non-MPK3/6) pathways that induce CAMTA3 relocalization exist.

Additional Kinases May Be Involved in flg22-Mediated Phosphorylation of CAMTA3

During the Phos-Tag–based analysis of CAMTA3vphospho-status, we noticed that flg22 treatment can induce additional mobility shift to the phospho-shift already triggered by MKK5^DD coexpression (i.e. in vivo MPK3/6 activation; Fig. 6A, left). This strong mobility shift was also seen for the mutP3 variant, which showed no difference in the mobility shift between coexpressing MKK5^KR and MKK5^DD (Fig. 6A, right). Taken with the observations of CAMTA3 relocalization above, there are likely other flg22-responsive kinases (besides MPK3/6) that phosphorylate CAMTA3. The analysis with the mutP3 variant that lacks all MAPK-targeted sites also excludes MPK4 or other MAPKs known to be weakly activated by flg22 (e.g. MPK11; Bethke et al., 2012; Nitta et al., 2014).

Figure 6.

Additional kinases may be involved in flg22-induced phosphorylation of CAMTA3. A, Additional kinases induce phosphorylation of CAMTA3 in a MAPK-independent manner. Arabidopsis protoplasts were transfected to coexpress CAMTA3–wild-type (WT; or CAMTA3-mutP3) with constitutively active MKK5^DD or its kinase-dead variant (KR), respectively. One batch of the protoplasts transfected with MKK5^DD was additionally treated with 100 nm of flg22 for 10 min. The extracted proteins were analyzed by western blot after separation on a Phos-Tag gel to visualize phospho-mobility shift (top) or by conventional SDS-PAGE for MKK5 expression and MAPK activation (bottom). B, Comparison of CAMTA3 phosphorylation and destabilization after coexpression with CDPKs from different subfamilies. CAMTA3 was coexpressed with the indicated Arabidopsis CDPKs, in its constitutively-active variants (abbreviated as CPK-VK). Arrowhead indicates “unmodified” CAMTA3; asterisks inidcarte phosphorylated CAMTA3 forms. Middle, Red asterisks mark the expected positions of the CDPKs in the anti-Flag western blot. Note that some CDPKs are not detectable, despite promoting CAMTA3 destabilization. Bottom is the quantitative western-blot analysis of CAMTA3 levels after coexpression with the indicated CDPKs (n = 3 independent replicates; error bars = SEM). Statistical distinct groups (one-way ANOVA, P < 0.05) are marked with distinct lowercase letters. C, Role of the MAPK-targeted phospho-sites for the CPK5-induced CAMTA3 destabilization. Protoplasts were transfected as described as in A above but with DEX-inducible expression of a constitutively active CPK5-VK. After an overnight incubation period, the protoplasts were treated with 20 μm of DEX for 5 h (+) and analyzed for a phospho-mobility shift (top) and reduced protein stability (bottom) by immunoblotting. Lane 3 of each representation represents protein dephosphorylation after λ-phosphatase treatment (as described in Fig 1B).

Besides MAPK cascades, several CDPKs are rapidly activated after exposure to PAMPs or pathogens (Ludwig et al., 2004). In Arabidopsis, CDPKs are encoded by a large gene family of 34 members and classified into four subfamilies (Boudsocq et al., 2010). Selected members from all four CDPK subfamilies were tested if they are involved in phosphorylation of CAMTA3. Deleting the C-terminal Ca²⁺ regulatory and autoinhibitory domains of CDPKs creates active kinases. When such constitutively-active CDPKs (designated as VK variants) were transiently coexpressed with CAMTA3 in the protoplasts, most of them induced a phospho-shift of CAMTA3 and, in part, a change in CAMTA3 protein levels (Fig. 6B). According to the quantitative western-blot and statistical analysis, CAMTA3 protein stability was significantly reduced by coexpression of CPK1, CPK2, or CPK5, while the tested members from subfamilies III and IV did not reduce, but instead increased, CAMTA3 levels significantly (Fig. 6B, bottom). Note that, although in this experimental setup many of the constitutively-active CDPKs can biochemically modulate CAMTA3 stability, most of them are not known to be PAMP-activated. Among the four Arabidopsis flg22-responsive CPKs (CPK4, CPK5, CPK6, and CPK11; Boudsocq et al., 2010), only CPK5 highly phosphorylated and destabilized CAMTA3. This is in line with CPK5 being a key calcium-dependent regulator of innate immune responses in plants (Dubiella et al., 2013; Liu et al., 2017; Guerra et al., 2020). Thus, we tested the role of CPK5 by transiently expressing CAMTA3–wild-type (or mutP3), together with a construct for inducible expression of CPK5-VK in protoplasts. Dexamethasone (DEX)-induced CPK5-VK expression led to a strong destabilization and mobility shift of CAMTA3 or mutP3 variant. The mobility shift is due to phosphorylation because it can be partially reverted by phosphatase treatment. Thus, CPK5 is most likely the relevant flg22-responsive kinase that induces MAPK-independent CAMTA3 phosphorylation. However, so far, no direct phosphorylation of CAMTA3 by recombinant CPK5 was obtained using in vitro kinase assays (Supplemental Fig. S6). Taken together, despite an in vivo CPK5-induced phospho-shift, CAMTA3 is either not phosphorylated by CPK5 directly, or additional components/conditions are needed to recapitulate this reaction in vitro.

DISCUSSION

CAMTA3 has been reported to act both as a transcriptional activator and repressor. The results presented here support its role as a negative regulator of defense genes, as seen in CAMTA3 overexpression suppressing both the basal and flg22-induced expression of downstream target genes of CAMTA3. Furthermore, flg22-activated MPK3 and MPK6 phosphorylate CAMTA3 directly, which is accompanied by flg22-induced CAMTA3 degradation via a proteasome-dependent pathway and the relocalization of nuclear CAMTA3 to the cytoplasm. It is important to note that while a subtle difference in protein turnover (i.e. requiring translation inhibition with CHX for visualization) is seen with transient MAPK activation during flg22 treatment, sustained MAPK activation based on the constitutively active kinase systems leads to rapid CAMTA3 degradation. Hence, during PAMP exposure in natural infection, the PAMP-activated kinases may “mark” the proteins for degradation but additional signaling thresholds or processes need to be attained to further propel targeting of CAMTA3 to the proteasome systems (e.g. activation of the proteasome). Nonetheless, when taken together, the rapid nuclear exclusion and heightened turnover of CAMTA3 protein can lead to de-repression of downstream target genes. Our findings are analogous to MPK3/6 activation during cold stress, leading to MPK3/6-mediated degradation of the transcription activator, ICE1 (Li et al., 2017; Zhao et al., 2017). A slight conceptual difference is that MPK3 and MPK6 are acting here as positive regulators of immunity by removing repressors (e.g. CAMTA3), while in cold/freezing tolerance, MPK3 and MPK6 function as negative regulators by depleting levels of transcriptional activators (i.e. ICE1). One puzzling point is that we did not detect MAPK activation by short-term cold treatment and consequently no phospho-mobility shift of CAMTA3. This might be due to differences in the experimental systems: e.g. protoplasts versus seedlings or initial prestress temperatures of 18°C versus 23°C. This should be clarified in the future to determine if cold-activated MPK3 and MPK6 also leads to CAMTA3 degradation. Such depletion would explain the crosstalk between cold acclimation and pathogen response, where CAMTA3-mediated repression of immunity is overcome by exposing plants to 4°C for a period of at least a week (Kim et al., 2017). Alternatively, both temperature and pathogen stresses may trigger distinct signaling pathways in parallel, including additional post-translational modifications that could differentially modify CAMTA3 functions.

Against this background, we also propose that besides MAPKs, alternative flg22-responsive kinase(s) can induce CAMTA3 phosphorylation that strongly promotes its degradation. Based on our in vitro kinase assays, we currently do not know if this is mediated by CPK5 or possibly an unknown kinase downstream of CPK5. If true, CAMTA3 may be a converging point of two major flg22-activated phosphorylation pathways, namely MAPK cascades and CDPKs. It has been shown that activation of CDPKs and MAPK cascades can act independently (Boudsocq et al., 2010) or synergistically to regulate the expression of some PTI genes (Boudsocq and Sheen, 2013). The typical phosphorylation motifs in MAPK and CDPK substrates are different, with the minimal MAPK targeted site being S/T-P (Pitzschke, 2015), while the predicted CDPK target motif is φ₋₁-[ST]₀-φ₊₁-X-B₊₃-B₊₄ (φ = hydrophobic residue, B = basic residue, X = any residue; Sebastià et al., 2004). While distinct substrates are implicit from their specificities, common target proteins at different sites have been identified. For instance, tomato (Solanum lycopersicum) ACS2, a key enzyme of ethylene biosynthesis, is phosphorylated by tomato CPK2 and MAPK in response to wound signaling (Kamiyoshihara et al., 2010) or recently, WRKY33 was found to be phosphorylated by both MPK3/6 and CPK5/6 to regulate camalexin biosynthesis (Zhou et al., 2020). To understand the possible crosstalk between MAPK- and CDPK-mediated phosphorylation on CAMTA3, it would be crucial to first clarify if a putative “intermediate” kinase acts downstream of CPK5 or the conditions needed for CPK5 to directly phosphorylate CAMTA3. For example, some form of CAMTA3 post-translational modifications may be required before direct phosphorylation by CPK5. However, this is unlikely to be through MAPK phosphorylation because CPK5 or flg22 treatment can both still induce a phospho-shift of the CAMTA3-P3 mutant (where all 11 putative MAPK phosphorylation sites are mutated).

In agreement with pathogen-induced CAMTA3 degradation, we showed that flg22-induced CAMTA3 phosphorylation is related to its in vivo destabilization and propose this may de-repress expression of downstream target genes. Equally possible would be a direct phospho-dependent reduction in the DNA binding properties of CAMTA3, but only allosteric effects can be expected because the key MAPK-targeted phospho-sites are not directly within the DNA-binding domain (Fig. 3A). Unfortunately, we failed to obtain convincing electrophoretic mobility shift (EMSA) data proving specific CAMTA3 binding to “CGCG”-core containing DNA probes (Supplemental Fig. S8). All published EMSA results employed truncated CAMTA3 comprising only the DNA-binding domain (Du et al., 2009; Nie et al., 2012; Sun et al., 2020) while we used full-length CAMTA3, suggesting the presence of domains that regulate DNA binding. Future analysis of DNA binding may consider employing CAMTA3 fragments harboring most of the phospho-sites but lacking these negatively-acting domains. Additionally, one needs to investigate the requirement of Ca²⁺, CaM, or other interacting proteins that may promote proper conformation of CAMTA3.

In this context, there is already indirect evidence for the relevance of CaM association affecting DNA binding by CAMTA3. This is seen in the phenotype of the loss-of-function CAMTA3^K907E mutation, which cannot bind CaM and does not repress expression of SA pathway genes (Du et al., 2009), and the gain-of-function CAMTA3^A855V mutation, which is in a putative IQ motif (Nie et al., 2012). Although no loss in CaM binding could be shown in CAMTA3^A855V, the in vitro pull-down binding assays employed may have overlooked quantitative differences due to the presence of both IQ and CaMBD motifs. Alternatively, instead of changing binding affinity, the A855V mutation may affect regulation by CaM allosterically. Recently, it was shown that CAMTA3-mediated repression of defense genes involves an NRM that can act independently of CaM binding to the IQ or CaMBD (Kim et al., 2017). While this challenges the prevalent concept of regulatory roles by CaM of CAMTA3, a possible explanation is that the CAMTA3 DNA-binding domain within the NRM fragment suffices as a repressor when directly bound to target promoters. Their subsequent analysis using mutated full-length protein also showed that the suppressor function of the A855V mutation was dominant over the K907E mutation (i.e. CaM association at A855 can override loss of CaM association to the CaMBD around K907). This indicates that there is differential regulation of NRM’s function through the IQ and CaMBDs. One may propose that the repressor function of the NRM domain alone is due to the uncoupling of CaM regulation, thereby allowing it to bind DNA. Furthermore, Kim et al. (2017) also reported phosphorylation at S454 and/or S964 to partially contribute to suppressor function of full-length CAMTA3. S454 is one of the major MAPK-targeted sites identified in our report here. This strengthens the idea of regulation of CAMTA3 repressor function through phosphorylation. Unfortunately, we did not detect non-MAPK phospho-sites here, possibly reflecting different dynamics or sequential phosphorylation of CAMTA3, which were not addressed in our current study. Analogously, differential phospho-states at individual sites of the Elk-1 transcription factor have been shown to have opposing effects on transcriptional activation regulation after sequential modification by MAPKs (Mylona et al., 2016). Such multistep phosphorylation can act as tunable signaling circuits, and therefore deserves attention in the future.

How CAMTA3 functions as a transcriptional repressor (Du et al., 2009; Nie et al., 2012) or an activator (Doherty et al., 2009) in different contexts is still unknown. A recent study questioned the repressor function of CAMTA3 in defense-related genes (e.g. EDS1/NDR1). The authors proposed that the constitutive immune activation in camta3 mutants is caused by activation of two nucleotide-binding domain Leu-rich repeat-containing proteins, DSC1 and DSC2, rather than the loss of CAMTA3 function as a transcriptional repressor (Lolle et al., 2017). While this idea is plausible, the transcriptional repression activity of CAMTA3 involved in plant immunity in several other studies cannot be ignored. Foremost, the effects on immunity regulation are not restricted to the camta3 knock-out mutants. CAMTA3 transgenic overexpression lines display compromised SAR and basal defense resistance (enhanced susceptibility to Pst DC3000; Jing et al., 2011), but also reduced basal and flg22-induced expression levels of defense-related genes (EDS1 and CBP60g; Fig. 4; Supplemental Fig. S3). Furthermore, phenotypes of the CAMTA3^A855V gain-of-function mutant (Jing et al., 2011; Nie et al., 2012) or the loss-of-function, non-CaM-binding CAMTA3^K907E (Du et al., 2009) cannot be explained by guarding through DSC1/DSC2 because CAMTA3 proteins are present. An alternative scenario would be that DSC1/DSC2 are not guarding the absence of CAMTA3 but rather may be monitoring a molecular process or a component involved in the CAMTA3-CaM interaction via the IQ or CaM-binding motifs.

Our data also questions the previously reported role of SR1IP1 in CAMTA3 degradation (Zhang et al., 2014), because we saw no difference in flg22-triggered CAMTA3 degradation in the sr1ip1 mutant and also failed to validate direct SR1IP1-CAMTA3 interaction. This might be due to the differences in the experimental setup used to analyze CAMTA stability: We used a well-defined synthetic peptide elicitor while pathogen infection was employed in the published report. Thus, additional factors (e.g. secreted pathogen effector proteins, toxins, or hormone-like compounds) as well as the timing for the pathogen infection process may contribute to the difference in outcome. On the contrary, overexpressing SR1IP1 seems to increase CAMTA3 stability in our hands (Fig. 5A). An explanation for this might be the chelation of components relevant for protein degradation through SR1IP1overexpression. Hence, SR1IP1 may indeed be involved in regulating CAMTA3 degradation, but its role as an E3 ligase adaptor needs to be re-evaluated.

CONCLUSION

In conclusion, the recurrent isolation of CAMTA3 mutants in numerous independent genetic screens for pathogen/stress responses coupled with its post-translational control through (direct/indirect) phosphorylation by possibly two stress-responsive kinase pathways positions it as a key node in (a)biotic stress signaling. Recent discovery of its negative regulation of genes involved in SA and N-hydroxypipecolic acid biosynthesis galvanizes its function in regulating immunity genes in SAR (Kim et al., 2020; Sun et al., 2020). Many gaps remain in the understanding of CAMTA3 regulation, but phosphorylation by multiple kinases, protein stability, and nucleo-cytoplasmic trafficking may be involved.

MATERIALS AND METHODS

Plant Growth Conditions, Protoplast Assays, and Immunoblot Analysis

Arabidopsis (Arabidopsis thaliana) plants were grown on soil in climate chambers for 5 to 6 weeks (22°C; 8 h of light, 16 h of darkness; 140 μE). Protoplast isolation and transfection were performed as described in Yoo et al. (2007). The protoplasts were incubated overnight (∼16 h) at 18°C to allow protein expression. Proteins were extracted by directly adding SDS-loading buffer to the pelleted protoplasts and processed for immunoblotting as described in Lee et al. (2004).

General Molecular Cloning

Coding sequences of genes used in transient expression experiments were amplified from complementary DNA (cDNA) and subcloned in either pDONR221 or pENTR-d-Topo vectors (Thermo Fisher Scientific). Gateway LR Clonase II Enzyme mix (Thermo Fisher Scientific) was used to generate constructs in destination vectors. Primers and description of transient expression constructs used in this study are listed in Supplemental Tables S1 and S2. Site-directed mutagenesis was performed using the primers listed in Supplemental Table S3 as described in Palm-Forster et al. (2012) and Eschen-Lippold et al. (2014).

RT-qPCR Analysis

Total RNA was extracted from plant tissues using TRIzol reagent (Roth). Two micrograms of RNA were used for cDNA synthesis (RevertAid First Strand cDNA synthesis kit; Thermo Fisher Scientific). Diluted cDNA (1:10) was used for RT-qPCR following the manufacturer’s protocol from 5×QPCR Mix EvaGreen (ROX; Bio & Sell). Primers and probes are listed in Supplemental Table S4. The RT-qPCR was performed in MX3005P cyclers (Agilent), and the program consisted of an initial activation step (95°C, 15 min) followed by 40 cycles (15 s at 95°C and 60 s at 60°C).

Protein Stability Assay

After transfection and overnight incubation for protein expression, 300 μL of the protoplasts were elicited with 100 nm of flg22 for the indicated time (water was used as a mock control). Where indicated, 1 μm of CHX was included to block translation. For the proteasome inhibitor, 50 μm of MG115 (Serva) was used to pretreat protoplasts for 30 min before flg22 and CHX elicitation. Cold treatment involves replacement with ice-cold media and immediate transfer to an ice-bath and cold room (4°C). To evaluate the effect of specific kinases on protein stability, CAMTA3 was coexpressed with 35S-promoter-driven expression of MKK5^DD, CPK5-VK, or other indicated CDPKs. Protoplasts were pelleted and boiled in Laemmli loading buffer, separated on SDS-PAGE, and processed for western blotting.

Protein Phosphorylation-Dependent Mobility Shift Assay in Phos-Tag Gels

Protoplasts (300 μL) transfected with pUGW14-CAMTA3 (to express HA-tagged CAMTA3) were elicited with 100 nm of flg22 and harvested at the indicated time points. To dephosphorylate the proteins, λ-phosphatase was used to treat extracted proteins according to the manufacturer’s protocol of Lambda Protein Phosphatase (New England Biolabs). Samples were boiled in Laemmli loading buffer and a Phos-Tag–based western blot was performed to analyze protein mobility shift. For separation of CAMTA3, Phos-Tag SDS-PAGE gel was prepared to contain 6% (w/v) acrylamide mix, 100 μm of MnCl₂, and 50 μm of Phos-Tag AAL-107 (Fujifilm Wako Chemicals) reagent. After separating protein samples on the Phos-Tag gel, Mn²⁺ from the gel was eliminated to improve blotting transfer efficiency. The gel was washed three times for 30 min with transfer buffer containing 10 mm of EDTA and one time with 1 mm of EDTA. Subsequently, proteins were transferred in a wet blot apparatus onto a nitrocellulose membrane (Macherey-Nagel) for 135 min at 100 V.

Quantitative Western Blot

An Odyssey CLx multiplex imaging system (Li-COR) was used for quantitative western blot. After blotting, nitrocellulose membranes were stained with REVERT Total Protein Stain (Li-COR), imaged in the Odyssey imaging system (700 nm channel) and quantified with the software Image Studio (Li-COR). Membranes were blocked with Odyssey Blocking buffer (Li-COR) for 1 h and subsequently incubated with primary antibody diluted in Odyssey Blocking buffer (0.2% [v/v] TWEEN 20) for 1 h. After a brief rinse, membranes were incubated with secondary antibody IRDye 800CW Goat anti-Mouse or anti-Rabbit IgG (H+L; Li-COR) diluted in Odyssey Blocking buffer (0.2% [v/v] TWEEN 20) for 1 h. After wash steps, target proteins were imaged (800-nm channel) as described above. To compensate for loading differences, the target protein signals (800 nm) were normalized against the total protein quantification (700 nm) according to the Odyssey CLx Application Protocol Manual (Li-COR). Typically, three (or more) biological replicate experiments were performed. The normalized values from individual experiments were further subjected to median-polishing before statistical analysis.

Recombinant Protein Expression and Purification

CAMTA3–wild-type and phospho-mutant variants were expressed as His₁₀-tagged fusion proteins via pDEST-N110 in KRX (Promega) Escherichia coli cells. Protein expression was induced with 0.1% (w/v) Rhamnose at 20°C for 10 h. MPK3, MPK4, and MPK6 were expressed as GST-tagged fusion proteins via pGEX4T-1 in BL21 cells, and expression induced with 1 mm of IPTG at 16°C overnight. Standard Ni-NTA or GSH-sepharose purification procedures were followed.

Protein Immunoprecipitation from Protoplasts

To obtain CPK5 or constitutively-active MPKs for kinase assays, protoplasts were transfected with the corresponding constructs and proteins were immunoprecipitated after an overnight expression period. The following plasmids were used: pEXSG encoding CPK5-FL-strepII or CPK5m-FL-strepII (Dubiella et al., 2013); p2HAGW7 encoding HA-tagged MAPKs (Berriri et al., 2012; Genot et al., 2017). For CPK5, protoplasts were lysed in a modified extraction buffer (100 mm of Tris at pH 8.0, 5 mm of EGTA, 5 mm of EDTA, 100 mm of NaCl, 20 mm of DTT, 0.5 mm of 4-(2-aminoethyl)benzenesulfonyl fluoride hydrochloride, 2 μg mL⁻¹ of Aprotinin, 2 μg mL⁻¹ of Leupeptin, Protease-Inhibitor Mix P [Serva], and 0.5%[ v/v] TritonX-100), and immunoprecipitated with MagStrep “Type3” XT beads (IBA Lifesciences). For MAPKs, MAPK extraction buffer (25 mm of Tris at pH 7.8, 75 mm of NaCl, 15 mm of EGTA, 15 mm of β-glycerophosphate, 15 mm of 4-nitrophenylphosphate, 10 mm of MgCl₂, 1 mm of NaF, 0.5 mm of Na₃VO₄, 1 mm of DTT, 0.1% [v/v] TWEEN 20, 1 mm of PMSF, 10 μg mL⁻¹ of Leupeptin, and 10 μg mL⁻¹ of Aprotinin) was used and immunoprecipitated with HA antibody coupled to Protein G Sepharose 4 Fast Flow (GE Healthcare).

In Vitro Kinase Assay

In vitro phosphorylation assay of CAMTA3 by MAPKs was performed in kinase substrate buffer 1 (20 mm of HEPES at pH 7.5, 15 mm of MgCl₂, 5 mm of EGTA, 1 mm of DTT, Protease-Inhibitor Mix HP, and 2 μCi [gamma-³²P]ATP) using recombinant His-CAMTA3 variants and GST-MPK3, GST-MPK4, and GST-MPK6. Samples were incubated for 30 min at 30°C, and the reactions were stopped by adding 5× loading buffer and 5 min of boiling. Samples were subsequently separated on SDS-PAGE, stained with Coomassie brilliant blue (CBB) and analyzed by autoradiography. Similarly, an in vitro phosphorylation assay of CAMTA3 by strepII-CPK5-FL or His-tagged CPK5-VK was performed as above, except that EGTA was excluded and replaced with 0.5 μg of CaM/0.1 mm of CaCl₂. Reciprocally, to check for calcium dependency (negative control), CaCl₂ was replaced with 2.4 mm of EGTA.

Phosphorylation Site Mapping

Phosphorylated CAMTA3 was obtained by in vitro phosphorylation with MPK3 or MPK6, or immunoprecipitated from leaves of CAMTA3-overexpressing plants that were infiltrated with flg22 (1 μm, 1 h). Amino acid residue-specific phosphorylation of CAMTA3 was mapped by liquid chromatography on-line with high-resolution accurate MS. Proteins separated by SDS-PAGE were subjected to in-gel tryptic digestion or with a combination of Glu-C. The resulting peptides were separated using C18 reverse phase chemistry employing a precolumn (EASY column SC001, length: 2 cm; inside diameter: 100 μm; particle size: 5 μm) in line with an EASY column SC200 with a length of 10 cm, an inside diameter of 75 μm and a particle size of 3 μm on an EASY-nLC II (all from Thermo Fisher Scientific). Peptides were eluted into a Nanospray Flex ion source (Thermo Fisher Scientific) with a 60- or 120-min gradient increasing from 5% to 40% (v/v) acetonitrile in double-distilled water with a flow rate of 300 nL/min and electrosprayed into an Orbitrap Velos Pro mass spectrometer (Thermo Fisher Scientific). The source voltage was set to 1.9 kV, the S Lens RF level to 50%. The delta multipole offset was −7.00. The automatic gain control target value was set to 1e06 and the maximum injection time (max IT) was set to 500 ms in the Orbitrap. The parameters were set to 1e04 and 100 ms in the linear trap quadropole with an isolation width of 2 D for precursor isolation and MS/MS scanning. Peptides were analyzed by a targeted data acquisition scan strategy with an inclusion list to specifically select and isolate CAMTA3-phosphorylated peptides for MS/MS peptide sequencing. Multistage activation was applied to further fragment ion peaks resulting from neutral loss of the phosphate moiety by dissociation of the high-energy phosphate bond to generate b- and y-fragment ion series rich in peptide sequence information.

MS/MS spectra were used to search The Arabidopsis Information Resource (TAIR10 database; ftp://ftp.arabidopsis.org; 35,394 sequences, 14,486,974 residues) with the software Mascot v2.5 (http://www.matrixscience.com/mascot_support_v2_5.html) linked to the tool Proteome Discoverer v1.4 (https://www.thermofisher.com/store/products/OPTON-30945#/OPTON-30945). The enzyme specificity was set to trypsin and two missed cleavages were tolerated. Carbamidomethylation of Cys was set as a fixed modification and oxidation of Met and phosphorylation of Ser and Thr as variable modifications. The precursor tolerance was set to 7 ppm and the product ion mass tolerance was set to 0.8 D. A decoy database search was performed to determine the peptide spectral match and peptide identification false-discovery rates. Phosphorylated peptides with a score surpassing the false-discovery–rate threshold of 0.01 (Q value < 0.01) were considered positive identifications. The phosphoRS module was used to specifically map phosphorylation to amino acid residues within the primary structure of phosphopeptides.

Microscopy

For localization studies, pUBC-based plasmids (Grefen et al., 2010) expressing CAMTA3-YFP or SR1IP1-CFP fusions were transfected into protoplasts. Where indicated, pEXSG-ERF104-CFP (Bethke et al., 2009) was included as a nuclear marker for colocalization studies or pRT100myc-MKK5^DD (or -MKK5^KR; Lee et al., 2004) was cotransfected for in vivo activation of MPK3/6. For BiFC assay, protoplasts were cotransfected with constructs of pE-SPYNE/pUC-SPYNE or pE-SPYCE/pUC-SPYCE (Walter et al., 2004) encoding the indicated pairs to be analyzed (see Supplemental Table S2 for details). After an overnight incubation, protoplasts were directly analyzed by confocal laser scanning microscopy with an LSM780 (Zeiss). Fluorescence image settings are: YFP (Ex: 514 nm, Em: 500–570 nm) or CFP (Ex: 458 nm, Em: 480–520 nm). For time-series–based observations, protoplasts were allowed to settle to the bottom of a depression slide before elicitation. This minimizes movement and the same protoplast (or groups of protoplasts) can be tracked using the accurate position tracking function through the motorized XY scanning stage.

Statistical Analyses

Statistical significance was analyzed with the software Prism 5 (GraphPad).

Accession Numbers

Sequence data from this article can be found in the TAIR database (https://www.arabidopsis.org/) under accession numbers: CAMTA3 (At2g22300), SR1IP1 (At5g67385), MPK3 (At3g45640), MPK6 (At2g43790), and CPK5 (AT4G35310). The sr1ip1-1 mutant (Zhang et al., 2014; SALK_064178) was obtained from the Nottingham Arabidopsis Stock Centre (http://arabidopsis.info/) and the camta3 mutants were provided by Hillel Fromm and Yuelin Zhang.

Supplemental Data

The following supplemental materials are available.

• Supplemental Figure S1. Analysis of flg22-induced changes in CAMTA3 transcript levels.

• Supplemental Figure S2. MS2 spectra of peptides with phosphorylated site found in CAMTA3.

• Supplemental Figure S3. Expression of the CAMTA3 target, CBP60g, in lines overexpressing CAMTA3phospho-variants.

• Supplemental Figure S4. Yeast- or plant-based interaction assays between CAMTA3 and SR1IP1.

• Supplemental Figure S5. Time course of flg22-induced CAMTA3 subcellular relocalization.

• Supplemental Figure S6. Radioactive in vitro kinase assays with constitutively active (truncated, VK) or full-length CPK5 showing no-to-little phosphorylation of CAMTA3.

• Supplemental Figure S7. Compilation of western blots validating the expression of intact fusion proteins for the BiFC or colocalization experiments.

• Supplemental Figure S8. EMSA for in vitro binding of full-length CAMTA3 to a “CGCG-containing” promoter fragment of EDS1.

• Supplemental Table S1. Primers for cloning into entry vector pENTR/D.

• Supplemental Table S2. List of constructs used in this study.

• Supplemental Table S3. Primers used for site-directed mutagenesis of CAMTA3 in entry vectors.

• Supplemental Table S4. Primers and probes for RT-qPCR.