The Transcriptional Cascade in the Heat Stress Response of Arabidopsis Is Strictly Regulated at the Level of Transcription Factor Expression

Abstract

Group A1 heat shock transcription factors (HsfA1s) are the master regulators of the heat stress response (HSR) in plants. Upon heat shock, HsfA1s trigger a transcriptional cascade that is composed of many transcription factors. Despite the importance of HsfA1s and their downstream transcriptional cascade in the acquisition of thermotolerance in plants, the molecular basis of their activation remains poorly understood. Here, domain analysis of HsfA1d, one of several HsfA1s in Arabidopsis thaliana, demonstrated that the central region of HsfA1d is a key regulatory domain that represses HsfA1d transactivation activity through interaction with HEAT SHOCK PROTEIN70 (HSP70) and HSP90. We designated this region as the temperature-dependent repression (TDR) domain. We found that HSP70 dissociates from HsfA1d in response to heat shock and that the dissociation is likely regulated by an as yet unknown activation mechanism, such as HsfA1d phosphorylation. Overexpression of constitutively active HsfA1d that lacked the TDR domain induced expression of heat shock proteins in the absence of heat stress, thereby conferring potent thermotolerance on the overexpressors. However, transcriptome analysis of the overexpressors demonstrated that the constitutively active HsfA1d could not trigger the complete transcriptional cascade under normal conditions, thereby indicating that other factors are necessary to fully induce the HSR. These complex regulatory mechanisms related to the transcriptional cascade may enable plants to respond resiliently to various heat stress conditions.

Figure 1.

Domain Analysis of the Putative Regulatory Domain of HsfA1d in Arabidopsis Protoplasts. (A) Schematic representation of the HsfA1d structure. The analyzed regions of the putative regulatory domain are indicated by the numbers in open boxes. The numbers on dashed lines indicate the positions of amino acid residues. DBD, DNA binding domain; HR-A/B, oligomerization domain; NLS, nuclear localization signal; AHA, transactivation domain; NES, nuclear export signal. (B) Effect of deletion of the putative regulatory domain on the transactivation activity and DNA binding activity of HsfA1d. The left panel shows schematic diagrams of the deletion mutants. The middle and right panels show the transactivation activity and DNA binding activity, respectively. Effectors were expressed under the control of the CaMV 35S promoter. The transactivation activity and DNA binding activity were evaluated with the HSP18.2_pro:GUS and 35S_pro:HSE9-GUS reporters, respectively. Note that DNA binding of HSFs was detected as repression of GUS expression from 35S_pro:HSE9-GUS. To normalize the transfection efficiency, pBI35SΩ-ELUC was cotransfected in each experiment. The reporter activities obtained with full-length HsfA1d (dFL) and empty vector (Vec) were set to 1 for the assays of the transactivation and DNA binding activities, respectively. The error bars indicate the sd from three replicated samples. Statistically significant differences between effectors are indicated by different lowercase letters (Tukey’s test, P < 0.01). (C) Protein accumulation levels of the deletion mutants. The proteins of the deletion mutants were expressed as sGFP-fused proteins under the control of the CaMV 35S promoter. A plasmid constitutively expressing a sGFP fused to a 3×FLAG tag was cotransfected in each experiment as an internal control. The levels of the fusion proteins were analyzed via immunoblot analysis with an antibody against GFP. (D) Repressive effects of regions 1 and 2 on the activity of HsfA1d. The transactivation activity and DNA binding activity of the deletion mutants were analyzed as described in (B).

Figure 2.

Activation of the HsfA1d Deletion Mutants by Heat Stress in the Protoplasts Derived from the hsfa1a/b/d Triple Mutant. The transactivation activity of the deletion mutants was analyzed with the HSP18.2_pro:GUS reporter under normal (22°C) and heat stress (37°C) conditions. The heat-stressed samples were harvested after 1 h of heat shock treatment followed by 1 h of recovery at 22°C. The reporter activity obtained with dFL under normal conditions was set to 1. The error bars indicate the sd from three replicated samples. Asterisks indicate statistically significant differences between the reporter activities (Student’s t test with Bonferroni correction, *P < 0.05, **P < 0.01, and ***P < 0.001). The effectors used in each panel are as follows: (A) single-deletion mutant; (B) multiple-deletion mutants; and (C) multiple-deletion mutants retaining region 1 or regions 1 and 2.

Figure 3.

Mutation of the Tyrosine Residue in the Conserved Motif Disrupts the Repressive Function of Region 1. (A) Effect of the mutation on Tyr-271 on the transactivation activity (left panel) and DNA binding activity (right panel) of HsfA1d. Tyr-271 was replaced with phenylalanine (dY271F) or aspartic acid (dY271D). The reporter activities obtained with dFL and Vec were set to 1 for the assays of the transactivation and DNA binding activities, respectively. The error bars indicate the sd from three replicated samples. Asterisks indicate statistically significant differences between the reporter activities (Student’s t test with Bonferroni correction, **P < 0.01 and ***P < 0.001). (B) Schematic representation of HsfA1d/VP16 chimera TFs. The VP16AD was fused to the 260 (260/VP) or 296 (296/VP) N-terminal amino acids or the 296 N-terminal amino acids with the Y271D mutation (296YD/VP) of HsfA1d. (C) Transactivation activity (left panel) and DNA binding activity (right panel) of HsfA1d/VP16 chimeric TFs. The reporter activities obtained with 296/VP and Vec were set to 1 for the assays of the transactivation and DNA binding activities, respectively. The error bars indicate the sd from three replicated samples. Asterisks indicate statistically significant differences between the reporter activities (Student’s t test with Bonferroni correction, *P < 0.05 and **P < 0.01). (D) Activation of the HsfA1d/VP16 chimeric TFs by heat stress in the protoplasts derived from the hsfa1a/b/d triple mutant. The reporter activity obtained with FL under normal conditions was set to 1. The error bars indicate the sd from three replicated samples. Asterisks indicate statistically significant differences between the reporter activities (Student’s t test with Bonferroni correction, **P < 0.01 and ***P < 0.001).

Figure 4.

Region 1 Is Required for the Interaction with HSP70s. (A) and (B) Coimmunoprecipitation of HsfA1d interactors. sGFP and sGFP-HsfA1d were immunoprecipitated from non-stressed or heat stress-treated 35S_pro:sGFP/abd (sGFP) and 35S_pro:sGFP-HsfA1d/abd (A1d) plants by an anti-GFP antibody. The purification of the proteins was confirmed via immunoblot analysis with anti-GFP antibody (A) and silver staining (B). The open and closed arrowheads indicate sGFP and sGFP-HsfA1d, respectively. (C) Y2H analysis using HSC70-1 and various truncated forms of HsfA1d. HSC70-1 and the truncated forms of HsfA1d were fused to the GAL4 DNA binding domain (BD) and GAL4 activation domain (AD), respectively. The left panel shows a schematic diagram of the truncated HsfA1d derivatives. The right panel shows the growth of yeast strains on nonselective medium (SD-LW) or selective medium (SD-LWH). (D) In vitro pull-down assays of GST-HsfA1d with 6×His-HSC70-1. Purified proteins from E. coli cells expressing only 6×His tag (6×His) or 6×His-HSC70-1 (HSC70-1) were incubated with GST or GST-HsfA1d (dFL) bound to glutathione Sepharose beads. The bound proteins were analyzed by immunoblot analysis with anti-His or anti-GST antibodies. The lower panel shows a longer exposure image of the anti-His immunoblot image in the upper panel. (E) Effect of HSC70-1 on the transactivation activity of HsfA1d/VP16 chimeric TFs. The reporter activity obtained with 296/VP cotransfected with Vec was set to 1. The error bars indicate the sd from three replicated samples. Asterisks indicate statistically significant differences between the reporter activities (Student’s t test, ***P < 0.001).

Figure 5.

Generation of Overexpressors of HsfA1d Derivatives. (A) RNA levels of sGFP-HsfA1d derivatives in the transgenic plants (dFL_OX, FL; dmNES_OX, mNES; dΔ1_OX, dΔ1; dΔ1-3_OX, dΔ1-3). HsfA1d derivatives were expressed under the control of the CaMV 35S promoter as sGFP fusions. Lowercase letters indicate two independent transgenic lines that express the same transgene. Ethidium bromide-stained images of rRNA are shown as a loading control. (B) Protein levels of sGFP-HsfA1d derivatives in transgenic plants. The total protein was analyzed by immunoblot analysis with anti-GFP antibody. The Rubisco large subunit (rbcL) stained with Ponceau S is shown as a loading control. (C) and (D) Growth of transgenic plants that expressed HsfA1d derivatives. Representative images (C) and the maximum rosette radius (D) of 21-d-old plants grown on agar plates under normal conditions are shown. The experiment was performed twice, and a representative result is shown. Asterisks indicate statistically significant differences compared with the VC (Student’s t test with the Bonferroni correction, **P < 0.01; n = 15). Bars = 2 cm. (E) Subcellular localization of HsfA1d derivatives under normal conditions or after heat stress treatment. Images of the differential interference contrast (DIC) and GFP fluorescence and merged images of the DIC and GFP are shown. Bars = 20 μm.

Figure 6.

Effect of Overexpression of HsfA1d Derivatives on the HSR. (A) and (B) Coimmunoprecipitation of endogenous HSP70 and HSP90 with HsfA1d derivatives. Extracts from transgenic plants that expressed HsfA1d derivatives were immunoprecipitated using an anti-GFP antibody. The input and immunoprecipitated proteins were analyzed via immunoblot analysis with anti-GFP, anti-HSP70, and anti-HSP90 antibodies. The lower panel of HSP90 shows a longer exposure image of the image in the upper panel. The signal intensities of the bands of coimmunoprecipitated HSP70 and HSP90 normalized by the intensity of the corresponding HsfA1d derivatives are shown in (B). The error bars indicate the sd from triplicate experiments. (C) and (D) Coimmunoprecipitation of endogenous HSP70 and HSP90 with HsfA1d. A complementation line (HsfA1d_pro:sGFP-HsfA1d/abd, Comp) and an overexpressor (35S_pro:sGFP-HsfA1d/abd, OX) of HsfA1d in the hsfa1a/b/d background were used. These plants were harvested under normal conditions or after heat stress treatment (37°C for 0.5 h). Extracts from these samples were immunoprecipitated using an anti-GFP antibody. The input and immunoprecipitated proteins were analyzed via immunoblot analysis with anti-GFP, anti-HSP70, and anti-HSP90 antibodies. The lower panels of HSP70 and HSP90 show longer exposure images of the images in the upper panels. The signal intensities of the bands of coimmunoprecipitated HSP70 normalized by the intensity of HsfA1d are shown in (D). The signal intensities for the non-stressed samples were set to 1 for each transgenic plant. The error bars indicate the sd calculated from triplicate experiments. (E) Expression analysis of HS-inducible genes in transgenic plants. Ethidium bromide-stained images of rRNA are shown as a loading control. (F) and (G) Thermotolerance test of transgenic plants. Seedlings were treated at 43°C for 50 or 80 min. After recovery for 7 d, photographs of the seedlings (F) were taken, and their survival rates (G) were determined. The viable plants are defined as those that generated new rosette leaves during the recovery period. The control plants were grown at 22°C throughout the experiments. The error bars indicate the sd from three (control) or five (heat-stressed) replicates (n = 10 each). Statistically significant differences are indicated by different lowercase letters (Tukey’s test, P < 0.01).

Figure 7.

Transcriptome Analysis of the dΔ1 Overexpressor. (A) Stress inducibility of the upregulated genes in dΔ1_OX. 206 genes was significantly upregulated by more than 4-fold in dΔ1_OX compared with VC under normal conditions. The responses of the top 100 upregulated genes in response to abiotic stress or hormone treatments are shown as a heat map. (B) Venn diagram comparing the upregulated genes in dΔ1_OX with the HsfA1-downstream HS-inducible genes. The total numbers of the genes in each group are shown in parentheses. The HsfA1-downstream HS-inducible genes that were upregulated or not upregulated in dΔ1 were classified as “Δ1-responding genes” or “non-Δ1-responding genes,” respectively. (C) Number of HSEs in the promoters of all genes registered in TAIR9 (27,684 genes), Δ1-responding genes, or non-Δ1-responding genes. The HSE sequences (nGAAnnTTCn or nTTCnnGAAn) in 1-kb promoter regions were counted. (D) Functional categorization of Δ1-responding or non-Δ1-responding genes. (E) qRT-PCR analysis of several Δ1-responding chaperones and non-Δ1-responding TFs in the overexpressors of HsfA1d derivatives. The expression levels of each gene in VC were measured under both normal and heat stress conditions. For the overexpressors of HsfA1d derivatives, the expression levels were measured under normal conditions. The expression of each gene in the heat-stressed VC (37°C, 1 h) was set to 100. The error bars indicate sd (n = 3). nd, not detected. (F) and (G) Venn diagram comparing the Δ1-responding genes and non-Δ1-responding genes with the upregulated genes in the HsfA2 overexpressor and HsfA3 overexpressor (F) or with the hsfb1 hsfb2b double mutant (G). The total numbers of the genes in each group are shown in parentheses.

Figure 8.

Region 1 Is Not Required for Attenuation of HsfA1d Activity after Prolonged Heat Stress or during Recovery. Expression patterns of HS-inducible genes in dFL_OX and dΔ1_OX during the heat stress treatment and recovery. The plants were treated with heat stress at 37°C for 0 to 6 h (H0 to H6) or allowed to recover for 2 or 5 h after 1 h of heat stress (H1R2 or H1R5). Ethidium bromide-stained images of rRNA are shown as a loading control.

Figure 9.

Working Model of the HSR in Arabidopsis. The conventional (A) and newly proposed (B) models for activation of the HSR. The important factors and regulatory mechanisms discussed in this article are summarized. (A) Conventional activation mechanism of the HSR. HSP70/90 interact with HsfA1d and negatively regulate HsfA1d activity under normal conditions. The HSP70/90 interaction sites of HsfA1d were unclear. Under heat stress conditions, increased amounts of unfolded proteins competitively interact with HSP70/90 such that HsfA1d is released from HSP70/90 and becomes active. The activation of HsfA1d enables HsfA1d to induce HS-inducible genes, including HSPs and TFs. TFs trigger the transcriptional cascade and amplify the HSR. (B) Our proposed activation mechanism of the HSR. HSP70/90 interact with HsfA1d and negatively regulate HsfA1d activity under normal conditions. HSP70 interacts with DBD and region 1 of HsfA1d. By contrast, region 1 and the NES are involved in the interaction between HsfA1d and HSP90. Region 1 is conserved among HsfA1s as the TDR domain (indicated by a light-blue area in region 1). Under heat stress conditions, the interaction between HsfA1d and HSP70 is disrupted by unknown mechanisms. Protein kinases (PKs) may be involved in the activation mechanisms. Although it is unclear whether HSP90 dissociates from HsfA1d upon heat shock, the repressor function of HSP90 seems to be inactivated by heat stress. HSP70-free HsfA1d becomes active and upregulates the expression of HS-inducible genes. However, the induction of non-Δ1-responding genes requires coregulators. TF genes are primarily included among non-Δ1-responding genes; thus, the activity of the transcriptional cascade is under strict control. The solid and dashed arrows between HSP70/90 and HsfA1d indicate strong or weak interactions, respectively. The other arrows and bar-heads indicate positive and negative regulation, respectively. The question marks denote links or factors to be confirmed.