Heat shock factor 1 confers resistance to Hsp90 inhibitors through p62/SQSTM1 expression and promotion of autophagic flux

Abstract

Heat shock protein 90 (Hsp90) has an important role in many cancers. Biochemical inhibitors of Hsp90 are in advanced clinical development for the treatment of solid and hematological malignancies. At the cellular level, their efficacy is diminished by the fact that Hsp90 inhibition causes activation of heat shock factor 1 (HSF1). We report a mechanism by which HSF1 activation diminishes the effect of Hsp90 inhibitors geldanamycin and 17-allylaminogeldanamycin (17-AAG, tanespimycin). Silencing HSF1 with siRNA or inhibiting HSF1 activity with KRIBB11 lowers the threshold for apoptosis in geldanamycin and 17-AAG-treated cancer cells. Autophagy also mitigates the actions of Hsp90 inhibitors. Blocking autophagy with 3-methyladenine (3-MA), bafilomycin A1, or beclin 1 siRNA also lower the threshold for apoptosis. Exploring a potential relationship between HSF1 and autophagy, we monitored autophagosome formation and autophagic flux in control and HSF1-silenced cells. Results show HSF1 is required for autophagy in Hsp90 inhibitor-treated cells. The reduced autophagy observed in HSF1-silenced cells correlates with enhanced cell death. To investigate how HSF1 promotes autophagy, we monitored the expression of genes involved in the autophagic cascade. These data show that sequestosome 1 (p62/SQSTM1), a protein involved in the delivery of autophagic substrates and nucleation of autophagosomes, is an HSF1-regulated gene. Gene silencing was used to evaluate the significance of p62/SQSTM1 in Hsp90 inhibitor resistance. Cells where p62/SQSTM1 was silenced showed a dramatic increase in sensitivity to Hsp90 inhibitors. Results highlight the importance of HSF1 and HSF1-dependent p62/SQSTM1 expression in resistance Hsp90 inhibitors, underscoring the potential of targeting HSF1 to improve the efficacy of Hsp90 inhibitors in cancer.

Keywords: 17-AAG; 17-N-allylamino-17-demethoxygeldanamycin; ATG; Autophagy; Cancer; DMSO; HSE; HSF1; Hsp; Hsp90; LC3; SQSTM1; autophagy-related gene; dimethyl sulfoxide; heat shock element; heat shock factor 1; heat shock protein; microtubule-associated protein 1 light chain 3; p62/SQSTM1; sequestesome 1.

Fig. 1

Hsp90 inhibitors activate HSF1 and increase heat shock protein expression. a. HSF1 luciferase reporter construct (pGL4.27-HSE) and a Renilla control plasmid were co-transfected into RKO, A549 and MCF-7 then treated for 4 h with vehicle (0.1% DMSO), geldanamycin, or 17-AAG (10–500 nM). Bar graphs show mean ratios of luciferase to Renilla luminescence for each cell type, normalized to vehicle control. Error bars indicate standard deviations. b. Western blot of nuclear extracts from RKO cells treated for 1 h with geldanamycin or 17-AAG (250, 500 nM) for P-Ser-326 HSF1, total HSF1, and TFIID (loading control). c. Real-time PCR analysis of HSF1 target genes DNAJA4m DNAJB1, and HSPA1A in RKO cells at 37°C (control), 42°C (heat shock, 6 h), or treated at 37°C with geldanamycin or 17-AAG (250 nM) for 6 h. Bar graphs indicate mean starting quantities in fmol per µg of total RNA. Error bars are standard deviations for n = 4.

Fig. 2

Silencing HSF1 attenuates Hsp40 and Hsp70 expression and sensitizes cells to Hsp90 inhibitors. a. RKO cells transfected with either a negative control (NEG) or HSF1 siRNA (HSF1) were treated with geldanamycin or 17-AAG (100, 250 nM) for 8 h and total proteins extracted. Western blot was performed for HSF1, Hsp90, Hsp70, Hsp40 and actin (loading control). Blots are representative of n = 3. b. siRNA-transfected RKO cells were treated for 24 h with geldanamycin or 17-AAG (100, 250 nM) and total proteins analyzed by Western blot for PARP and caspase-3 cleavage. c. Concentration-response curves for cell viability in siRNA-transfected RKO cells treated for 48 h with geldanamycin (10–250 nM) or 17-AAG (200–1000 nM). Data points represent mean values of Calcein-AM fluorescence normalized to vehicle-treated (0.1% DMSO) control. Error bars are standard deviations for n = 8.

Fig. 3

Biochemical inhibition of HSF1 activity or inhibition autophagy both sensitize cells to geldanamycin-induced apoptotic cell death. a. RKO were pre-treated with vehicle (0.1% DMSO) or 10 nM KRIBB11 for 1 h followed by geldanamycin (50–250 nM) for 24 h. Total protein extracts were analyzed for PARP and caspase-3 cleavage. b. RKO were treated with KRIBB11 (1–50 nM) for 48 h. Data points represent mean values of Calcein-AM fluorescence normalized to vehicle-treated (0.1% DMSO) control. Error bars are standard deviations for n = 8. Label indicates % viability vs. control for 5 nM KRIBB11) c. Concentration-response curves for cell viability in RKO treated with geldanamycin (20–200 nM) + 0.1% DMSO (vehicle control, open circles) or geldanamycin (20–200 nM) + 5 nM KRIBB11. Error bars are standard deviations for n = 8.

Fig. 4

Inhibition of autophagy sensitizes cells to geldanamycin. a. RKO were pre-treated for 1 h with vehicle (0.1% DMSO), 3-MA (2 mM), or bafilomycin A1 (400 nM) followed by geldanamycin (50–250 nM) for 24 h. b. Beclin 1 expression was silenced in RKO cells using siRNA then treated with geldanamycin (100–250 nM) for 24 h. Total protein extracts were analyzed for PARP and caspase-3 cleavage. Total protein extracts were analyzed for PARP and caspase-3 cleavage. Figures are representative of n = 3.

Fig. 5

Autophagic flux in control and HSF1-silenced cells. RKO cells transfected with either a negative control (NEG) or HSF1 siRNA (HSF1) were treated with geldanamycin or 17-AAG (250 nM) for 8 h. Bafilomycin A1 (400 nM) was added for the last 4 h of treatment where indicated. LC3 and p62 flux was calculated as the difference in densitometry values in the presence (+) and absence (−) of bafilomycin A1, after normalization to actin (loading control). Flux values for LC3 and p62 are presented in bar graph, relative to vehicle-treated control (NEG) cells. Error bars represent standard deviations for n = 4 experiments and Western blots shows representative data. (*p<0.05; **p<0.0005 vs. DMSO control sample data)

Fig. 6

Autophagosome and lysosome levels in control and Hsp90 inhibitor-treated cells. a. Stable clones of RKO expressing GFP (negative control) or GFP-LC3 (NEG-siRNA or HSF1siRNA transfected) were treated with geldanamycin or 17-AAG (250 nM) for 8 h. Fluorescent images of Hoescht 33342-stained cells were collected by high content screening. Representative images were re-constructed from 3 confocal planes using PE Volocity software. b. Green fluorescent puncta (autophagosomes, white arrowheads) were counted for 4,000 nuclei using PE Columbus software. Bar graphs represent mean values and error bars represent standard deviations (*p<0.005; **p<0.001 vs. DMSO control sample data) c. siRNA-transfected RKO cells were treated with geldanamycin or 17-AAG (250 nM) for 4 h and stained with LysoTracker Red DND-99 and Hoescht 33342. d. Red fluorescent puncta (lysosomes, yellow arrowheads) were counted for 4,000 nuclei using PE Columbus software. Bar graphs represent mean values and error bars represent standard deviations, showing no statistically significant (p<0.05) differences between vehicle (DMSO)-treated and inhibitor-treated samples.

Fig. 7

Hsp70 is dispensable for autophagic flux in Hsp90 inhibitor-treated cells. RKO cells transfected with either a negative control (NEG) or Hsp70 siRNA (HSP70) were treated with geldanamycin or 17-AAG (250 nM) for 8 h. Bafilomycin A1 (400 nM) was added for the last 4 h of treatment where indicated. LC3 flux was calculated as the difference in densitometry values in the presence (+) and absence (−) of bafilomycin A1, after normalization to actin (loading control). Hsp70 immunoblots show inducible Hsp70 (Hsp70-1) as well as constitutive Hsp73 (Hsc70), which is not HSF1-dependent. Flux values are presented in bar graph, relative to vehicle-treated control (NEG) cells. Error bars represent standard deviations for n = 4 experiments and Western blots shows representative data, showing no statistically significant (p<0.05) differences between vehicle (DMSO)-treated and inhibitor-treated samples.

Fig. 8

Expression of autophagy-related (ATG) proteins and Beclin 1 in control and HSF1silenced cells. RKO were transfected with negative control (NEG) or HSF1 siRNA (HSF1) and treated with vehicle (0.1% DMSO) or geldanamycin (250 nM) for 6 h. a. Total proteins were analyzed by Western blot. Images are representative from n = 3. b. Analysis of Beclin 1 expression level by Li-Cor Odyssey, normalized to actin (loading control) and displayed relative to NEG siRNA-transfected, DMSO-treated (control) cells. Bar graph represents mean normalized values and error bars represent standard deviations (* = p<0.001).

Fig. 9

HSF1-dependence of p62/SQSTM1 expression, and effect of silencing p62/SQSTM1 on cellular sensitivity to Hsp90 inhibitors. a. Real-time PCR data for p62/SQSTM1 mRNA expression in control (NEG siRNA) and HSF1-silenced (HSF1 siRNA) cells. Values for p62/SQSTM1 amplification were normalized to GAPDH as described in Materials and Methods are expressed relative to vehicle (0.1% DMSO) treated control. Error bars represent standard deviations (n = 4). Asterisks indicates significant difference from control (* = p<0.001 and ** = p<0.005) and b. Analysis of p62/SQSTM1 expression level by Li-Cor Odyssey, normalized to actin (loading control) and displayed relative to NEG siRNA-transfected, DMSO-treated (control) cells.

Fig. 10

Model for HSF1 and autophagy in cellular response to Hsp90 inhibitors. HSF1 is activated by Hsp90 inhibitors and enhances the activity of the transcription factor HSF1, which by promoting the expression of p62/SQSTM1 maintains autophagic and mitigates cell death. Inhibiting HSF1 or autophagy increases cellular sensitivity to cell death following treatment with Hsp90 inhibitors.