Uncoupling Stress-Inducible Phosphorylation of Heat Shock Factor 1 from Its Activation

Abstract

In mammals the stress-inducible expression of genes encoding heat shock proteins is under the control of the heat shock transcription factor 1 (HSF1). Activation of HSF1 is a multistep process, involving trimerization, acquisition of DNA-binding and transcriptional activities, which coincide with several posttranslational modifications. Stress-inducible phosphorylation of HSF1, or hyperphosphorylation, which occurs mainly within the regulatory domain (RD), has been proposed as a requirement for HSF-driven transcription and is widely used for assessing HSF1 activation. Nonetheless, the contribution of hyperphosphorylation to the activity of HSF1 remains unknown. In this study, we generated a phosphorylation-deficient HSF1 mutant (HSF1Δ∼PRD), where the 15 known phosphorylation sites within the RD were disrupted. Our results show that the phosphorylation status of the RD does not affect the subcellular localization and DNA-binding activity of HSF1. Surprisingly, under stress conditions, HSF1Δ∼PRD is a potent transactivator of both endogenous targets and a reporter gene, and HSF1Δ∼PRD has a reduced activation threshold. Our results provide the first direct evidence for uncoupling stress-inducible phosphorylation of HSF1 from its activation, and we propose that the phosphorylation signature alone is not an appropriate marker for HSF1 activity.

FIG 1

Characterization of the HSF1 mutant that is phosphorylation-deficient within the RD, HSF1Δ∼PRD. (A) Schematic illustration of the HSF1 functional domains with the known phosphorylation sites. In HSF1Δ∼PRD, 15 phosphorylation sites in the regulatory domain (RD) were mutated from serine (S) and threonine (T) residues to alanines (A) as indicated. Additional HSF1 domains include the DNA-binding domain (DBD), heptad repeat domains (HR-A/B and HR-C), and transactivation domain (TAD). Note that the figure is not drawn to scale. (B) hsf1^−/− MEFs were transfected with Mock plasmid [pcDNA3.1/myc-His(−)A], Myc-His-HSF1 WT, or Myc-His-HSF1Δ∼PRD. hsf1^+/+ represents the endogenous levels of HSF1 in MEFs. Cells were either left untreated (−) or exposed to heat shock (+). HSF1 protein levels from cell lysates were detected by Western blotting with anti-HSF1 antibody. Hsc70 is shown as a loading control. An asterisk indicates an HSF1 protein that migrates slower on SDS-PAGE due to hyperphosphorylation (53). The difference in size between the endogenous HSF1 from hsf1^+/+ MEFs and exogenous HSF1 WT is caused by the Myc-His tag on the human HSF1 WT construct. (C) hsf1^−/− MEFs were transfected as in panel B. Cells were either left untreated (−) or heat shocked (+). Cell lysates were treated with lambda protein phosphatase (+λPP) or left untreated. Samples were analyzed by using Western blotting. α-Actin is shown as a loading control. An asterisk indicates the HSF1 protein that migrates slower on SDS-PAGE due to hyperphosphorylation (53). (D) hsf1^−/− MEFs were transfected as in panel B and treated with cycloheximide (CHX) at 37°C for the indicated times. Cell lysates were analyzed with anti-HSF1 and anti-HSF2 antibodies. α-Actin is shown as a loading control.

FIG 2

HSF1Δ∼PRD localizes to the same subcellular compartments as HSF1 WT under normal and stress conditions. HeLa cells were transfected with Mock plasmid [pcDNA3.1/myc-His(−)A], Myc-His-HSF1 WT, or Myc-His-HSF1Δ∼PRD, left untreated (C) or exposed to heat stress (HS; 1 h at 42°C), and analyzed by immunofluorescence microscopy. A monoclonal antibody against myc was used to detect exogenously expressed HSF1 protein, whereas an anti-HSF1 antibody was used to detect both endo- and exogenously expressed HSF1. DNA was stained with DAPI. The merge figure is an overlay of myc, HSF1, and DAPI signals. Scale bars, 25 μm.

FIG 3

HSF1Δ∼PRD binds to DNA in a stress-inducible manner. hsf1^−/− MEFs were transfected with Mock plasmid [pcDNA3.1/myc-His(−)A], Myc-His-HSF1 WT, or Myc-His-HSF1Δ∼PRD, and left either untreated (C) or exposed to a 30-min heat shock at 43°C (A) or heavy metal stress (B). (A) The occupancy of HSF1 at the HSPA1A/B (Hsp70) and HSPB1 (Hsp25) promoters was analyzed by ChIP, followed by qPCR. The qPCR values of the immunoprecipitations were normalized to the input values and related to the HSF1 WT control sample, which was set to value 1. The data are presented as mean values from three independent experiments plus the standard errors of the mean (SEM). The values obtained for the nonspecific antibody (normal rabbit serum) are 1.07 for HSPA1A/B and 2.92 for HSPB1. (B) For assessing HSF1Δ∼PRD DNA-binding activity during prolonged stress, the cells were treated with 60 μM CdSO₄ for the indicated times (3+R: 3 h CdSO₄, followed by a 3-h recovery in fresh culture medium). The HSE-HSF complex (HSF-HSE) was analyzed by EMSA. Expression of HSF1 constructs was detected by Western blotting with anti-HSF1 antibody. α-Actin was used as a loading control. The pound sign indicates nonspecific HSE interactions, and the asterisk indicates HSF1 protein that migrates more slowly on SDS-PAGE due to hyperphosphorylation (53).

FIG 4

Phosphorylation in the regulatory domain suppresses HSF1 transactivating capacity. (A) hsf1^−/− MEFs were transfected with Mock plasmid [pcDNA3.1/myc-His(−)A], Myc-His-HSF1 WT, or Myc-His-HSF1Δ∼PRD and left either untreated (C) or exposed to heat stress at 43°C up to 60 min. The mRNA levels of HSPA1A/B (Hsp70) and HSPB1 (Hsp25) were quantified with qRT-PCR and normalized against RNA18S5. The values are shown relative to the respective mRNA levels in the Mock-transfected cells in control conditions (C), which was arbitrarily set to value 1. (B) hsf1^−/− MEFs were transfected as in panel A and either left untreated (C), treated with 60 μM CdSO₄ for 3 h (3 h), or treated for 3 h and left to recover in fresh culture medium for 3 h (3 h + R3h). mRNA quantification and data analysis were performed as in panel A. The data are presented as mean values from at least three independent experiments plus the SEM. *, P ≤ 0.05; ***, P ≤ 0.001; ****, P ≤ 0.0001.

FIG 5

HSF1Δ∼PRD requires a lower threshold for activation than HSF1 WT. hsf1^−/− MEFs were transfected with Mock plasmid [pcDNA3.1/myc-His(−)A], Myc-His-HSF1 WT, or Myc-His-HSF1Δ∼PRD and left either untreated (C) or exposed to heat stress (39, 40, and 41°C for 30 and 60 min) (A) or heavy metal stress (40 and 60 μM CdSO₄ for 60, 120, and 180 min) (B). The mRNA levels of HSPA1A/B (Hsp70) were quantified using qRT-PCR and normalized against RNA18S5. The values are shown relative to the respective mRNA levels in the Mock-transfected cells in control conditions (C), which was arbitrarily set to value 1. The data are presented as mean values from at least three independent experiments plus the SEM. *, P ≤ 0.05; **, P ≤ 0.01; ****, P ≤ 0.0001.

FIG 6

Phosphorylation defines capacity of the RD to control transactivation. (A) Schematic illustration of the chimeric proteins consisting of Gal4 DNA-binding domain (Gal4-DBD, aa 1 to 147), HSF1 WT or HSF1Δ∼PRD regulatory domain (HSF1 WT RD and HSF1Δ∼PRD RD, aa 220 to 389), and herpes simplex virus protein VP16 activation domain (VP16-AD, aa 413 to 490). Note that the figure is not drawn to scale. (B and C) HeLa cells were transfected with plasmids encoding Gal4-driven luciferase and β-galactosidase, together with indicated plasmids encoding chimeric proteins described in panel A or with an empty plasmid (Mock). Cells were left untreated (C) or heat shocked for 30 min at 42°C and let to recover for 5 h at 37°C (HS + R5h). In panel B, the relative luciferase activity was calculated against the activity in the Gal4-VP16 samples under control conditions, which was set to value 100. The data are presented as mean values from four independent experiments plus the SEM. ns, nonsignificant; **, P ≤ 0.01; ****, P ≤ 0.0001. In panel C, the protein levels of the Gal4-VP16 chimeras, under control conditions, were analyzed using Western blotting with anti-VP16 antibody. Hsc70 is shown as a loading control. The arrowheads indicate Gal4-VP16 (white), Gal4-VP16-HSF1 WT (black), and Gal4-VP16-HSF1Δ∼PRD (gray). The difference in the migration pattern between Gal4-VP16-HSF1 WT and Gal4-VP16-HSF1Δ∼PRD is likely due to phosphorylation.