Regulation of the heat stress response in Arabidopsis by MPK6-targeted phosphorylation of the heat stress factor HsfA2

Plant materials and growth conditions

The Arabidopsis thaliana ecotype Columbia (Col) was used as wild-type for developmental experiments as indicated. The mpk6 mutant was described earlier (Nakagami et al., 2006). Plants were grown on 0.5X MS medium (Sigma) under long day condition (16 h light/8 h dark) with a humidity level of 50% and 50 µE cool white light. Stress treatments were performed according to Charng et al. (2007).

Transient expression assays

Transient expression assays were performed using Arabidopsis suspension cell culture and tobacco mesophyll protoplasts, respectively, as described (Forreiter, Kirschner & Nover, 1997; Nakagami, Kiegerl & Hirt, 2004; Lyck et al., 1997; Scharf et al., 1998) with slight modifications. Protoplasts were isolated and transformed by PEG (polyethylene glycol) mediated transformation at room temperature (25 °C) under dark conditions. Plant expression vectors used are based on the pRT series of vectors (Töpfer, Schell & Steinbiss, 1988; Döring et al., 2000). The expression vector for 3HA-HsfA2 and the reporter construct P_Hsp18−CI::GUS was described by Schramm et al. (2006).

Inhibitors

To investigate the effect of phosphorylation and dephosphorylation on heat stress response, several broad-spectrum inhibitors were tested. All inhibitors were provided from Calbiochem, dissolved in DMSO as recommended and used at the following final concentrations: Staurosporin (Cat No. 569396) 10 µM, Cantharidin (Cat. No. 210155) 10 µM, Calyculin (Cat. No. 208851) 0.25 µM and Okadaic acid (Cat. No. 459618) 2 µM.

In-gel kinase assays

C-terminal domains (CTD) of several Hsfs were cloned into the pDEST-15 vector (Invitrogen) and expressed in E. coli strain Rosetta (Novagen). Sequences of CTD-Hsfs proteins are described in Supplemental Table S1. GST-tagged Hsf proteins were purified using Glutathione sepharose 4B resin (Amersham) according the manufacturer recommendations. Cell extracts were prepared at different times after heat stress in extraction buffer (25 mM Tris ± HCl pH 7.8, 15 mM EGTA, 75 mM NaCl, 1 mM dithiothreitol (DTT), 10 mM MgCl₂, 1 mM NaF, 0.5 mM NaVO₃, 15 mM β-glycerophosphate, 15 mM 4-nitrophenylphosphate, 0.1% Tween-20, 0.5 mM phenylmethylsulfonyl fluoride (PMSF), 5 mg/ml leupeptin, 5 mg/ml aprotinin). After centrifugation at 20000xg for 45 min, the cleared supernatant was used. For in-gel protein kinase reactions, cell extracts containing 20 µg of total protein per lane were separated by SDS-PAGE. Myelin basic protein (MBP; 0.5 mg/ml) was used as a substrate polymerized in the polyacrylamide gel. After protein renaturation, the kinase reactions were performed in the gel as described (Usami et al., 1995). The gels were dried and analyzed by autoradiography.

Immunocomplex kinase assays

Protoplast extracts containing equal protein amounts (100 µg) were subjected to a 2-h preincubation in the presence of 20 µl of mixed protein A- and G-Sepharose beads (1:1). The supernatant was then immunoprecipitated with 5 µl of anti-HA monoclonal antibody and 20 µl of protein G-Sepharose beads overnight at 4 °C. The beads were washed three times with wash buffer (50 mM Tris, pH 7.4, 250 mM NaCl, 5 mM EGTA, 5 mM EDTA, 0.1% Tween 20) and once with kinase buffer (20 mM HEPES, pH 7.4, 10 mM MgCl₂, 5 mM EGTA, and 1 mM dithiothreitol). Kinase reactions of the immunoprecipitated proteins were performed in 15 µl of kinase buffer containing 5 µg of MBP, 0.1 mM ATP, and 2 µCi of (γ-³² P) ATP. The protein kinase reactions were performed at room temperature for 30 min, and the reactions were stopped by adding 4x SDS loading buffer. The phosphorylation of MBP was analyzed by autoradiography after separation on 15% SDS-PAGE.

Immunoblotting

Immunoblotting was performed with equal amounts of protein separated by 10% SDS-PAGE and transferred to PVDF membranes (Millipore) by electroblotting. Membranes were probed with A-purified M23 (raised against MPK6), or with either anti-HA monoclonal antibody or with anti-Myc polyclonal antibody (A-14, Santa Cruz Biotechnology) or with anti-GFP antibody (sigma). Membranes were developed by enhanced chemiluminescence as recommended by the manufacturer (Gene Image, Amersham Biosciences).

Confocal imaging using Arabidopsis mesophyll protoplasts

HsfA2 WT and mutated sequences were PCR amplified using the pRTHsfA2 vector (Döring et al., 2000) as a template. Primers used for PCRs are listed in Supplemental Table S2. For site directed mutagenesis a first PCR using the following pairs were generated: Hsfa2NS-GW2-CterF/Hsfa2-249A-R (PCR1); Hsfa2NS-GW2-CterF/Hsfa2-T249D-R (PCR2); Hsfa2NS-GW2-R/Hsfa2-T249A-F (PCR3) and Hsfa2NS-GW2-R/Hsfa2-T249D-F (PCR4). PCR1 and PCR3 or PCR2 and PCR4 products were diluted and mixed together and then re-amplified with the Hsfa2NS-GW2-CterF/Hsfa2NS-GW2-R primer pairs to generate HsfA2-T249A and HsfA2-T249D mutants. pRTHsfA2 was also amplified with Hsfa2NS-GW2-CterF/Hsfa2NS-GW2-R to generate the HsfA2-WT version. The three PCR products were then introduced in pDNR207 (Invitrogen) trough a classical Gateway-BP reaction. After sequence verification HsfA2 constructs were introduced in p2GWF7 (Karimi, Depicker & Hilson, 2007) by a Gateway-LR reaction (Invitrogen) to give HsfA2-WT-GFP, HsfA2-T249A-GFP, and HsfA2-T249D-GFP plasmids. MPK7 construct was amplified with MPK7NS-GW2-CterF/MPK7NS-GW2-R to generate MPK7-GFP plasmid and was used as a control for immunoblotting (see below). Transient expression assays were performed using Arabidopsis mesophyll protoplasts isolated and transformed as described (Yoo, Cho & Sheen, 2007). Confocal imaging were performed with a Leica TCS-SP2 confocal microscope and analysed with the Leica confocal software (LCS). Autofluorescence and overlay pictures are described in Supplemental Figure S1.

Mass spectrometry

Coomassie blue stained phosphorylated and non-phosphorylated HsfA2 protein bands were excised from the SDS-PAGE. Ion trap mass spectrometry was performed as according to the user manual and as described in de la Fuente van Bentem et al. (2006).

MPK6 is activated by heat stress

To get insights into the protein kinases regulated in response to heat stress, we performed in-gel kinase assays of protein extracts from Arabidopsis leaves before and after exposure to heat stress (Fig. 1). Detached leaves were incubated in pre-heated (37 °C) liquid medium and then placed at 37 °C for 1 h. As shown by in-gel kinase assays, several MBP-phosphorylating protein kinases in the range between 40–72 kDa become activated by heat stress (Fig. 1A). The strongest activity corresponds to a protein kinase with an approximative molecular mass of 48 kDa that displayed a maximum activation within 2 min of heat stress before returning to lower levels by 20 min. The transient activation of the 48 kDa protein kinase that is able to phosphorylate MBP in Arabidopsis resembled to the class of mitogen-activated protein kinases that was previously reported in tomato and alfafa (Link et al., 2002; Sangwan & Dhindsa, 2002). We confirmed this hypothesis by immunocomplex kinase assays with a specific antibody against Arabidopsis MPK6 (Fig. 1B). The corresponding immunoblotting experiments indicated that the observed heat-induced kinase activity was not due to changes in MPK6 protein amounts (Fig. 1B). These results indicate that in Arabidopsis, MPK6 might be the major heat stress transducer in plant cells and the activation of this MAPK is a highly conserved process in plants. These results confirm also the HS induced MPK6 activity detected in Arabidopsis leaves obtain by Li et al. (2005).

With in-gel kinase assays we could observe in samples from heat-stressed leaves one dominant heat-activated protein kinase of approximately 48 kDa (Fig. 1A) similar to that reported by Link et al. (2002) for tomato. MBP phosphorylation has been also detected to a lower extent for proteins in the range around 60 kDa proteins (Fig. 1A) and these kinases might also participate in the heat stress response. However, by immunocomplex assays with specific antibodies, we identified MPK6 as the most prominent heat stress-induced protein kinase in whole cell extracts (Fig. 1B). These results show that heat stress activation of MAPKs is also a highly conserved process in plants and add another factor to the list of activating conditions of MPK6, which was so far reported to be activated upon cold, salt, oxidative, hypoosmotic, ozone and genotoxic stress (Droillard et al., 2004; Lee & Ellis, 2007; Teige et al., 2004; Ulm et al., 2002; Yuasa et al., 2001), plant defence to biotic stress (Brader et al., 2007; Menke et al., 2004; Schweighofer et al., 2007) as well as jasmonate and ethylene signalling (Ouaked et al., 2003; Takahashi et al., 2007; Schweighofer et al., 2007).

Phosphorylation of heat stress transcription factors by MPK6

In order to address the possibility that MPK6 could affect the heat stress response by direct phosphorylation of Hsfs as the regulators of the heat stress response, we performed in vitro kinase assays using active MPK6 on 16 recombinant GST-Hsf fusion proteins, representing all Hsf protein subgroups. To clarify the specificity of the enzymatic interactions, the GST-Hsf fusions carried only the divergent C-terminal regulatory domains but not the highly conserved N-terminal DNA binding domains (see Kotak et al., 2004). As shown in Fig. 2A, MPK6 was able to phosphorylate several GST-Hsfs in vitro, but GST-HsfA2 appeared to be most strongly phosphorylated. In order to validate that the phosphorylation reactions towards the Hsf domains were specific, in vitro kinase assays were also performed with GST and MBP as negative and positive controls, respectively. Whereas MPK6 strongly phosphorylated MBP and GST-HsfA2, no phosphorylation was detected for GST (Fig. 2B).

Because transcription factors are common targets of MAPKs, we undertook a systematic in vitro screen for MPK6 targets among the 21 Arabidopsis Hsfs. Out of 16 candidates, representing all Arabidopsis Hsf subfamilies, MPK6 phosphorylated several members of different Hsf subclasses, but showed strong phosphorylation of HsfA2 (Fig. 2A). MPK6 was also shown to phosphorylate full length HsfA2 (Fig. 2B). Recently, putative targets of MPK3 and 6 have been described by Sörensson et al. (2012). HsfA4e appear to contain the PPSPR motif and might be another target of MPK6.

In vivo complex formation between MPK6 and HsfA2

The in vitro kinase assays (Fig. 2B) indicated that MPK6 is able to phosphorylate HsfA2. To corroborate that HsfA2 is a specific target of MPK6, we also tested whether MPK6 can directly interact with HsfA2 in vivo. For this purpose, we transiently expressed MPK6 and full-length wild type HsfA2 in Arabidopsis cells under the constitutive 35S Cauliflower Mosaic virus (CaMV) promoter (Fig. 3). MPK6 and HsfA2 were expressed as tagged proteins with either the Myc-epitopes or the Hemagglutinin-(HA), respectively, to allow their immunoprecipitation and immunodetection. Protoplasts co-expressing MPK6-Myc and 3HA-HsfA2 were then incubated at 37 °C for 5 min before protein extraction. Using an anti-Myc antibody, MPK6-Myc was immunoprecipitated, and the presence of 3HA-HsfA2 in the complex was assayed by protein gel blot analysis using anti-HA antibody (Fig. 3A). HsfA2 was found to interact strongly with MPK6 at 5 min after heat stress treatment.

Although both 3HA-HsfA2 and MPK6-Myc were produced under the 35S CaMV promoter, we also tested whether the protein amounts changed during the course of the experiment. Anti-Myc immunoblotting showed that the protein amounts of MPK6-Myc and 3HA-HsfA2 did not change during the heat treatment (Fig. 3B). Overall, these experiments indicate that HsfA2 is a target of MPK6.

MPK6 phosphorylates HsfA2 on threonine 249

To determine the MPK6 phosphorylation sites in HsfA2, we performed an in vitro kinase assay by incubating the recombinant GST-HsfA2 protein in the presence of cold ATP with or without active MPK6. The reaction products were separated by SDS-PAGE and excised after Coomassie blue staining. Upon in-gel alkylation-tryptic digestion, each digested sample was subjected to mass spectrometric analysis. No phosphorylated residue was detected in the nonphosphorylated sample, whereas a single phosphopeptide with the sequence 245RLTSTPSLGTMEENLLHDQEFDR268 was found for HsfA2 that was incubated with active MPK6 (Fig. 4A). A single phosphorylation site corresponded to the 247TST249 motif, but could not be identified to a specific residue within this sequence with certainty. We next mutated the 247TST249 in HsfA2 to 247AAA249 and expressed the mutant protein as recombinant GST fusion protein. When tested with active MPK6, the mutant HsfA2 247AAA249 protein could not be phosphorylated any more (unpublished results). Since MAP kinases absolutely require a proline residue in the + 1 position following the serine/threonine phosphorylation site, we then only mutated T249 to alanine (T249A). As shown in Fig. 4B, MPK6 phosphorylated wild type GST-HsfA2 but not GST-HsfA2 T249A, indicating that T249 is the amino acid residue that is phosphorylated by MPK6. However, because of the presence of additional putative phosphorylation sites within the HsfA2 protein sequence we cannot rule out that other sites might be targeted by MPK6 or other kinases and play a role in its regulation.

T249 phosphorylation regulates heat stress-induced nuclear accumulation of HsfA2

T249 lies in close proximity to the nuclear localization sequence (NLS) of HsfA2 (Fig. 4A), suggesting that phosphorylation of T249 could affect the localization of HsfA2. To analyse the possible consequences of phosphorylation of T249, we mutated T249 not only to alanine T249A to disable phosphorylation but also to glutamic acid T249D to mimic a phosphorylated status. Wild type HsfA2 (HsfA2-WT) and mutant T249A (HsfA2-T249A) or T249D (HsfA2-T249D) were then fused with GFP and expressed in Col-0 protoplasts under the constitutive 35S CaMV promoter. At ambient room temperature of 22 °C (RT), wild type HsfA2-WT and mutant HsfA2-T249A and HsfA2-T249D proteins showed similar cytoplasmic and nuclear localization (Fig. 5A, Col-0, RT). Upon heat stress, wild type and HsfA2-T249D accumulated in the nuclear compartment, whereas HsfA2-T249A was still distributed between cytoplasm and nucleus as in untreated cells (Fig. 5A, Col-0, HS). Threonine 249 that lies close to the nuclear localization signal of HsfA2, suggests that phosphorylation of this site might affect the intracellular localization of HsfA2. Localization studies performed in this work confirmed this hypothesis proving that T249 phosphorylation regulates the nuclear localization of HsfA2 upon heat stress. However we cannot discriminate whether nuclear localization of HsfA2 upon heat stress is a result of a cytosolic instability or a basic shuttling between the cytosol and the nucleus.

MPK6 regulates heat stress-induced nuclear accumulation of HsfA2

To clarify whether MPK6 is responsible for the heat stress-induced nuclear accumulation, we then tested the behaviour of the HsfA2-GFP construct in protoplasts derived from mpk6 knock out plants (Fig. 5B). Therefore, we expressed the wild type HsfA2-GFP construct in protoplasts prepared from mpk6 knock out plants at ambient room temperature of 22 °C (RT) or 37 °C (HS). Under ambient conditions or heat stress conditions, HsfA2-GFP showed a similar distribution between nucleus and cytoplasm (Fig. 5B). Under heat stress conditions, HsfA2-GFP showed nuclear accumulation only in Col-0 derivative protoplasts, indicating that MPK6 is the responsible factor that induces nuclear shuttling of HsfA2. S320 Phosphorylation of the human Hsf1 by a protein kinase plays a significant role in it nuclear localization (Murshid et al., 2010). Results presented in our work confirm that phosphorylation events are conserved mechanisms among organisms and kingdoms and regulate important biological responses to stress.

Phosphorylation-dependent regulation of HsfA2 protein levels upon heat stress

Because HsfA2 phosphorylation can affect HsfA2 function we used protein kinase and phosphatase inhibitors in Arabidopsis mesophyll protoplasts. Upon application of the broad-spectrum protein kinase inhibitor staurosporine, or the protein phosphatase inhibitors cantharidin, calyculin and okadaic acid, the expression levels of endogenous HsfA2 or one of its targets, the heat stress-induced protein Hsp18.1-CI (Schramm et al., 2006) was determined in Arabidopsis cells (Fig. 6A). After application of the inhibitors, the cells were split into two aliquots and one aliquot was heat stressed for 3 h at 37 °C whereas the other aliquot was kept at 22 °C. Applying cantharidin had no marked effects on HsfA2 protein level in heat stressed cells (Fig. 6A). However, application of the kinase inhibitor staurosporin as well as the phosphatase inhibitor okadaic acid resulted in dramatically reduced levels of HsfA2 protein and consequently blocked almost completely the expression of sHsp-CI (Fig. 6A). These experiments indicated that phosphorylation and dephosphorylation events can influence the heat stress response of cells. Yet, it was not clear if HsfA2 and its downstream target Hsp18-CI were affected at the gene expression or protein levels, or if the observed effects were due to indirect effects of the inhibitors. Therefore, the inhibitors were also investigated in cells that constitutively express 3HA-HsfA2 under the 35S CaMV promoter but in the absence of heat stress. Protein extracts of the differently treated cells were probed with HA antibody in Western blots. DMSO as the solvent of the inhibitors alone had no marked effect on 3HA-HsfA2 levels, but both staurosporine as well as the three phosphatase inhibitors resulted in strongly reduced 3HA-HsfA2 protein levels and GUS accumulation mediated by the Hsp18.1-CI promoter (Figs. 6B and 6C). The changes in 3HA-HsfA2 protein levels were not due to a general toxic effect on the cells as protein amounts of Rubisco did not change significantly (Fig. 6B). These results show that phosphorylation-dependent mechanisms regulate HsfA2 protein amounts in vivo.

Using broad-spectrum protein kinase and phosphatase inhibitors resulted in a strong reduction in HsfA2 protein levels in Arabidopsis cells upon heat stress activation (Fig. 6). Since HsfA2 is regulated at both, the transcriptional and post-translational level (Busch, Wunderlich & Schoffl, 2005; Schramm et al., 2008; Cohen-Peer et al., 2010; Ikeda, Mitsuda & Ohme-Takagi, 2011), we also tried to uncouple these two levels of regulation by testing whether these effects are also observed when HsfA2 was permanently expressed under the constitutive 35S CaMV promoter. Under these conditions, we observed that inhibition of phosphorylation still reduced HsfA2 protein abundance upon heat stress except for Cantharidin for which no marked effect has been observed in a native promoter condition (Fig. 6). These data suggest that these drugs might target different protein phosphatase complexes associated with plant heat stress signalling.

Phosphorylation-dependent HsfA2 protein abundance regulation depends on other protein kinases than MPK6

Our work so far revealed that HsfA2 is a specific substrate of MPK6 (Fig. 2) that interacts with MPK6 in a heat stress-dependent manner (Fig. 3A). We also obtained evidence that HsfA2 protein abundance is regulated by phosphorylation (Fig. 5). To test this hypothesis, we performed Western blots of 3HA-HsfA2 upon co-expression of MPK6-Myc alone, or together with a constitutively active form of an MPK-activating MAPKK MKK2 (MKK2^GOF), in Arabidopsis cells under the constitutive 35S CaMV promoter (Fig. 7). Under these conditions, MPK6 is directly activated by MKK2^GOF and no heat stress is necessary for its activation. To differentiate between activated and non-activated MPK6, HsfA2 and MPK6 were also co-expressed with the wild type form of MKK2. When HsfA2 was co-expressed with MPK6 or together with MKK2 or MKK2^GOF, HsfA2 levels were not affected when compared to expression of HsfA2 alone (Fig. 7). As a control, we also co-expressed HsfA2 with the MKK2 wild type and GOF version without MPK6. Under both conditions, HsfA2 protein levels did not show any reduction when compared to HsfA2 expression alone. To confirm that MPK6 was not involved in HsfA2 stability, we compared the level of HsfA2-GFP protein in Col-0 versus mpk6 protoplasts under heat stress conditions (Supplemental Figure S2). Under these conditions, degradation of HsfA2 wasn’t affected in the absence of MPK6. These data indicate that activated MPK6 does not regulate HsfA2 protein abundance. We conclude that other yet unidentified sites and most likely other protein kinases regulate HsfA2 protein stability.