HSP90 Inhibition Synergizes with Cisplatin to Eliminate Basal-like Pancreatic Ductal Adenocarcinoma Cells

Abstract

To improve the treatment of pancreatic ductal adenocarcinoma (PDAC), a promising strategy consists of personalized chemotherapy based on gene expression profiles. Investigating a panel of PDAC-derived human cell lines, we found that their sensitivities towards cisplatin fall in two distinct classes. The platinum-sensitive class is characterized by the expression of GATA6, miRNA-200a, and miRNA-200b, which might be developable as predictive biomarkers. In the case of resistant PDAC cells, we identified a synergism of cisplatin with HSP90 inhibitors. Mechanistic explanations of this synergy include the degradation of Fanconi anemia pathway factors upon HSP90 inhibition. Treatment with the drug combination resulted in increased DNA damage and chromosome fragmentation, as we have reported previously for ovarian cancer cells. On top of this, HSP90 inhibition also enhanced the accumulation of DNA-bound platinum. We next investigated an orthotopic syngeneic animal model consisting of tumors arising from KPC cells (LSL-KrasG12D/+; LSL-Trp53R172H/+; Pdx-1-Cre, C57/BL6 genetic background). Here again, when treating established tumors, the combination of cisplatin with the HSP90 inhibitor onalespib was highly effective and almost completely prevented further tumor growth. We propose that the combination of platinum drugs and HSP90 inhibitors might be worth testing in the clinics for the treatment of cisplatin-resistant PDACs.

Keywords: HSP90 inhibition; PDAC; basal-like subtype; cisplatin; platinum-DNA adducts; synergism.

Figure A1

Chemosensitivity of PDAC cells. (A–C) Treatment of the PDAC cell panel with (A) 0.25 µM to 30 µM irinotecan, (B) 0.1 µM to 200 µM doxorubicin, and (C) 10 nM to 10 µM gemcitabine for 72 h. Viability of cells was measured by an ATP-based luminescence assay. (D,E) Immunoblot analysis of the DNA damage marker phospho-H2AX after treatment with 20 µM cisplatin for 24 h. HSP70, loading control. Replicate experiments for Figure 1C. The blots were used for quantification, cf. Figure 1D.

Figure A2

No evidence for a causative role of miRNA-200 and EMT in cisplatin resistance of PDAC cells. (A) Biological replicate to Figure 2B to confirm the immunoblot analysis of GATA6 in PDAC cell lines. GAPDH served as loading control. (B) Heat map reflecting the expression of microRNA-200 family genes [16,19]. The microRNAs were differentially expressed between resistant (MIA PaCa-2, Panc-1) and sensitive (Capan-1, AsPC-1 and BxPC-3) PDAC cells, except miRNA-200c. (C) Immunoblot analysis of Vimentin (VIM) and E-cadherin (CDH1). Biological replicate to Figure 2E. (D,E) miRNA-200a and miRNA-200b expression analysis in MIA PaCa-2 and Panc-1 cells after the transfection with Pre-miR™ miRNA Precursors ((D) 200a-3p and (E) 200b-3p) for 72 h (48 h and 24 h). snRNA U6 was used for normalization. Mean ± SD of three biological replicates. (F,G) Immunoblot analysis of MIA PaCa-2 and Panc-1 cells after transfection as described in (D,E). Staining of Zeb-1, E-cadherin, Vimentin and GATA6, GAPDH as loading control. (H) Cell viability assay of MIA PaCa-2 cells after transfection as described in (D,E) and treatment with 1 µM to 20 µM cisplatin for 72 h.

Figure A3

Sensitive PDAC cells revealed no synergism upon HSP90 inhibition and cisplatin treatment. (A) PDAC cells AsPC-1, Capan-1, BxPC-3 and Suit-028 were treated with onalespib and/or cisplatin for 72 h with concentrations as indicated. Viability was determined by quantifying the ATP concentration. (B) The combination index (CI) was calculated from (A) and plotted against the fraction affected (Fa) for the combination of onalespib and cisplatin. (C) IC50s of onalespib in sensitive and resistant human PDAC cell lines. (D) Cisplatin-resistant cells were treated with ganetespib and/or cisplatin for 72 h with concentrations as indicated. Viability was determined as in (A). (E) Cisplatin-sensitive cells were treated as described in (D). (F) Biological replicate to Figure 3C. Immunoblot detection of FANCA. Cells were treated with onalespib and cisplatin as described in Figure 3C. GAPDH served as loading control. (G) Biological replicate to Figure 3E, determining the fluorescence intensity per nucleus, depicted as scatter plot for MIA Paca-2 cells. The phospho-H2AX intensity per nucleus (arbitrary units) was calculated. The red lines indicate the mean intensity. (H,J) Representative immunofluorescence images, staining phospho-H2AX in (H) Panc-1 cells and (J) Capan-1 cells, with DAPI as counterstain. Cells were treated with onalespib and cisplatin for 24 h. Scale bar: 200 µm. (I,K,L) Scatter plot of phospho-H2AX intensity per nucleus (arbitrary units), calculated by quantification of (H) and (J). Note that two replicates of this experiment are shown for Capan-1 cells (K,L); for MIA PaCa-2 and Panc-11 cells, one replicate is shown in Figure 3C,E and another one in (G,I). (A,D,E,I–L) p values were calculated with one-way ANOVA comparing the indicated groups. ns = not significant, * p ≤ 0.05, ** p ≤ 0.01, and *** p ≤ 0.001.

Figure A3

Sensitive PDAC cells revealed no synergism upon HSP90 inhibition and cisplatin treatment. (A) PDAC cells AsPC-1, Capan-1, BxPC-3 and Suit-028 were treated with onalespib and/or cisplatin for 72 h with concentrations as indicated. Viability was determined by quantifying the ATP concentration. (B) The combination index (CI) was calculated from (A) and plotted against the fraction affected (Fa) for the combination of onalespib and cisplatin. (C) IC50s of onalespib in sensitive and resistant human PDAC cell lines. (D) Cisplatin-resistant cells were treated with ganetespib and/or cisplatin for 72 h with concentrations as indicated. Viability was determined as in (A). (E) Cisplatin-sensitive cells were treated as described in (D). (F) Biological replicate to Figure 3C. Immunoblot detection of FANCA. Cells were treated with onalespib and cisplatin as described in Figure 3C. GAPDH served as loading control. (G) Biological replicate to Figure 3E, determining the fluorescence intensity per nucleus, depicted as scatter plot for MIA Paca-2 cells. The phospho-H2AX intensity per nucleus (arbitrary units) was calculated. The red lines indicate the mean intensity. (H,J) Representative immunofluorescence images, staining phospho-H2AX in (H) Panc-1 cells and (J) Capan-1 cells, with DAPI as counterstain. Cells were treated with onalespib and cisplatin for 24 h. Scale bar: 200 µm. (I,K,L) Scatter plot of phospho-H2AX intensity per nucleus (arbitrary units), calculated by quantification of (H) and (J). Note that two replicates of this experiment are shown for Capan-1 cells (K,L); for MIA PaCa-2 and Panc-11 cells, one replicate is shown in Figure 3C,E and another one in (G,I). (A,D,E,I–L) p values were calculated with one-way ANOVA comparing the indicated groups. ns = not significant, * p ≤ 0.05, ** p ≤ 0.01, and *** p ≤ 0.001.

Figure A4

Cisplatin-sensitive PDAC cells revealed no significantly increased chromosome fragmentation upon HSP90 inhibition. (A) Representative images of metaphase spreads from Panc-1 cells, as in Figure 4E. Cells were treated with onalespib (100 nM) and cisplatin (20 µM) for 24 h, together with 20 µM zVAD to block apoptosis. Chromosomes were stained according to Giemsa. (B) Representative images of metaphase spreads. Capan-1 cells were treated as in (A). (C) Chromosome fragmentation was analyzed by counting the number of fragments shown in (B) in 40 randomly chosen cells from three independent experiments. The red lines indicate the mean. p values were calculated with one-way ANOVA comparing the indicated groups. ns = not significant, *** p ≤ 0.001.

Figure 1

Human PDAC cell lines split into two distinct groups regarding their cisplatin response. (A) PDAC cells of the indicated lines were treated with 0.25 µM to 30 µM cisplatin for 72 h. Cell viability was measured with an ATP-based luminescence assay. (B) IC50 values of cisplatin, calculated based on the results presented in (A). (C) Immunoblot analysis of the DNA damage marker phospho-H2AX in the insensitive cell lines MIA PaCa-2 and Panc-1 and in the sensitive cell lines AsPC-1, Capan-1, and BxPC-3 after treatment with 20 µM cisplatin for 24 h. HSP70 was stained as the loading control. (D) Quantification of the phospho-H2AX-derived immunoblot signal, normalized to HSP70. Mean ± SD of 3 biological replicates. p values were calculated with one-way ANOVA. (E) Representative images of Pt-(GpG) adducts in the DNA of PDAC lines BxPC-3 (sensitive; right) and Panc-1 (resistant; left) after exposure to cisplatin (40 µM for 4 h). Scale bar, 200 µm. (F) Platinum-adduct quantification in pancreatic cancer cell lines measured as in (E). (G) Platinum-adduct level quantification in PDAC cell lines, measured after 1 h pre-incubation with 40 µg/mL diphenhydramine (DIPH) followed by 5 h cisplatin treatment (40 µM) with continuous addition of DIPH. Mean ± SD of 3 biological replicates. Scale bar, 200 µm. (H) Quantification of platinum-adduct level as described in (G). (F,H) p values were calculated with two-way ANOVA. ns = not significant, * p ≤ 0.05, ** p ≤ 0.01, and *** p ≤ 0.001.

Figure 2

GATA6, miRNA200a, and 200b serve as markers for cisplatin sensitivity. (A) Expression analysis of GATA6 in PDAC cell lines by qRT-PCR, normalized to the mRNA of RPLP0. Mean ± SD of three biological replicates. (B) Immunoblot analysis of GATA6 in PDAC cell lines. GAPDH served as the loading control. (C,D) Expression analysis of (C) miRNA-200a and (D) miRNA-200b in the panel of cell lines. snRNA U6 was used for normalization. Mean ± SD of three biological replicates. (E) Immunoblot to detect epithelial marker E-cadherin (CDH1) and the mesenchymal marker vimentin (VIM). (F,G) Panc-1 cells were transfected with Pre-miRNA-200a and 200b for 48 h and re-transfected for 24 h. After treatment with different concentrations (1 µM to 20 µM) of cisplatin for 72 h, cell viability was measured. Mean ± SD of three biological replicates. (H) Immunoblot analysis of Panc-1 cells after transfection as described in (F). Staining of the miR-200 target gene product Zeb-1, E-cadherin (CDH1), and Vimentin (VIM). GAPDH was stained as the loading control. (A,C,D) p values were calculated with one-way ANOVA comparing the indicated groups. ns = not significant, * p ≤ 0.05, ** p ≤ 0.01, and *** p ≤ 0.001.

Figure 3

Platinum-resistant human PDAC cells are sensitized by combining cisplatin with a HSP90 inhibitor. (A) MIA PaCa-2, Panc-1, and PaTu8988T cells were treated with onalespib and/or cisplatin for 72 h. Viability was determined by quantifying the ATP concentration. (B) The combination index (CI) was calculated for the combinations from (A) and plotted against the fraction affected (Fa). (C) Immunoblot analysis of FANCA. Cells were treated with onalespib and cisplatin at the indicated concentrations for 24 h: MIA PaCa-2 (150 nM; 20 µM), Panc-1 (150 nM; 20 µM), Capan-1 (400 nM; 2 µM), AsPC-1 (400 nM; 2 µM), and BxPC-3 (400 nM; 5 µM). GAPDH was detected as the loading control. FANCA was eliminated by HSP90 inhibition in cisplatin-resistant cell lines, but less profoundly or not at all in sensitive cells. (D) Representative images obtained by staining of the DNA damage marker phospho-H2AX in MIA PaCa-2 cells treated with onalespib (100 nM), cisplatin (20 µM), or both for 24 h. Scale bar, 200 µm. (E,F) Fluorescence intensities per nucleus are shown in a scatter plot for (E) MIA PaCa-2 and (F) Panc-1 cells. The red lines indicate the mean intensities. The phospho-H2AX intensity per nucleus (arbitrary units) was calculated by quantification of at least 200 cells per sample from one of two independent experiments. (A,E,F) p values were calculated with one-way ANOVA comparing the indicated groups. ns = not significant, *** p ≤ 0.001.

Figure 4

Cisplatin-resistant PDAC cells reveal increased platinum-DNA adduct formation and DNA damage after combining cisplatin and HSP90 inhibitor. (A) Representative images of the staining of platinated DNA in resistant Panc-1 and sensitive BxPC-3 cells after 5 h of cisplatin treatment, with a preincubation for 24 h with 100 nM onalespib. Scale bar, 200 µm. (B) Platinum-adduct level quantification of pancreatic cancer cell lines measured from (A). Mean ± SD of 3 biological replicates. p values were calculated with two-way ANOVA comparing the indicated groups. (C) Representative images of metaphase spreads. MIA PaCa-2 cells were treated with onalespib (100 nM) and/or cisplatin (20 µM) for 24 h, along with 20 µM zVAD to block apoptosis. Chromosomes were stained with Giemsa. (D,E) Number of (D) MIA PaCa-2 and (E) Panc-1 chromosome fragments per cell, shown as scatter plot and analyzed in 40 randomly chosen cells from three independent experiments. Panc-1 cells were treated as the MIA Paca-2 cells from (C). Red lines indicate the mean. p values were calculated with 1way ANOVA comparing the indicated groups. ns = not significant, * p ≤ 0.05, and *** p ≤ 0.001.

Figure 5

Synergistic effect of HSP90 inhibitors and cisplatin in KPC cells. (A) KPC cells (LSL-KrasG12D/+; LSL-Trp53R172H/+; Pdx-1-Cre, C57/BL6 genetic background) were treated with two different concentrations of onalespib and cisplatin for 48 h, followed by the assessment of cell viability by quantification of ATP. Mean ± SD. Note that KPC cells displayed a cisplatin sensitivity comparable to human PDAC cells that we had classified as “sensitive” but were still further sensitized by onalespib. This may well be due to intrinsic differences between human and human cells regarding their response to cisplatin. (B) Combination index (CI) calculated from (A) plotted against the fraction affected (Fa) for the combination of HSP90 inhibitors onalespib and cisplatin. (C) Representative images obtained by immunofluorescence staining for phospho-H2AX, with DAPI as counterstain. Cells were treated with onalespib and/or cisplatin for 24 h. Scale bar, 200 µm. (D) Scatter plot of phospho-H2AX intensity per nucleus (arbitrary units), calculated by quantification of (C). The red lines indicate the mean nuclear phospho-H2AX staining intensity. (E) KPC cells were treated with onalespib (100 nM) and/or cisplatin (2 µM) for 24 h in the presence of 20 µM zVAD. The chromosomes were stained with Giemsa. Representative images of metaphase spreads are shown. (F) The chromosome fragmentation was analyzed by counting the number of fragments shown in (F) in 40 randomly chosen cells from three independent experiments, depicted as scatter blot. (A,D,F) p values were calculated with 1way ANOVA comparing the indicated groups. ns = not significant, *** p ≤ 0.001.

Figure 6

Anti-tumor efficiency of onalespib with cisplatin in a syngeneic mouse model of PDAC. (A) Treatment scheme for C57BL/6J mice which were orthotopically transplanted with 200,000 KPC cells. Ten days after transplantation, tumor formation was confirmed by sonography. The treatment with 25 mg/kg onalespib and 4 mg/kg cisplatin was started on day 11 and repeated on day 18. On day 14 and 16, the mice received single treatments with 25 mg/kg onalespib. 21 days after transplantation, the mice were sacrificed for removal of the tumors. (B) Weight loss of animals in response to treatment. (C) Images of three representative pancreatic tumors collected from mice treated according to (A) at the endpoint of the experiment (day 21). (D) Tumor weight was determined after necropsy on day 21. Red lines indicate the mean weight. Control, onalespib, and cisplatin single treatment groups (n = 8), and the combination treatment group (n = 7). (E,G) Tumor sections from (C) were stained for (E) phospho-H2Ax and (G) apoptosis using a TUNEL assay and subjected to fluorescence microscopy. Scale bar, 200 µm. (F,H) Calculated intensity of (F) phospho-H2AX and (H) TUNEL staining per field of view, depicted as scatter blot. Ten images per mouse were randomly chosen and analyzed. (D,F,H) p values were calculated with one-way ANOVA comparing the indicated groups. ns = not significant, * p ≤ 0.05, ** p ≤ 0.01, and *** p ≤ 0.001. The mean of all measurements within one sample is indicated by a red bar. The median values were also calculated, again resulting in significant differences when comparing the combination treatment with control or single treatments.