Different HSP90 Inhibitors Exert Divergent Effect on Myxoid Liposarcoma In Vitro and In Vivo

Abstract

The therapeutic options for patients with relapsed or metastatic myxoid liposarcoma (MLS) remain scarce and there is currently no targeted therapy available. Inhibition of the HSP90 family of chaperones has been suggested as a possible therapeutic option for patients with MLS. However, the clinical effect of different HSP90 inhibitors vary considerably and no comparative study in MLS has been performed. Here, we evaluated the effects of the HSP90 inhibitors 17-DMAG, AUY922 and STA-9090 on MLS cell lines and in an MLS patient-derived xenograft (PDX) model. Albeit all drugs inhibited in vitro growth of MLS cell lines, the in vivo responses were discrepant. Whereas 17-DMAG inhibited tumor growth, AUY922 surprisingly led to increased tumor growth and a more aggressive morphological phenotype. In vitro, 17-DMAG and STA-9090 reduced the activity of the MAPK and PI3K/AKT signaling pathways, whereas AUY922 led to a compensatory upregulation of downstream ERK. Furthermore, all three tested HSP90 inhibitors displayed a synergistic combination effect with trabectidin, but not with doxorubicin. In conclusion, our results indicate that different HSP90 inhibitors, albeit having the same target, can vary significantly in downstream effects and treatment outcomes. These results should be considered before proceeding into clinical trials against MLS or other malignancies.

Keywords: HSP90 inhibition; combination therapy; drug treatment; myxoid liposarcoma; receptor tyrosine kinase signaling.

Figure 1

Effect of HSP90 inhibitors in vitro. (A–D) Cell viability assays of the HSP90 inhibitors 17-DMAG, AUY922 and STA-9090 and the chemotherapeutic agent doxorubicin on MLS cell lines 402-91 (A), 2645-94 (B) and 1765-92 (C) and human fibroblasts (F470) (D). Data is normalized to untreated controls. Mean ± SD is shown. (E) Table of IC50 values of tested compounds on MLS cell lines and human fibroblasts. N/A signifies IC50 value above maximum drug concentration (>1 µM). (F) Cell cycle analysis using flow cytometry on MLS cell line 402-91 treated with 100 nM 17-DMAG, 50 nM AUY922 or 80 nM STA-9090 for 24 h. (G) Cell cycle analysis using flow cytometry on MLS cell lines 402-91, 2645-94 and 1765-92, treated with 30 nM 17-DMAG for 24 or 48 h. (H) Western blot of whole-cell extracts from MLS cell lines 402-91, 1765-92 and 2645-94 treated with 100 nM 17-DMAG, 50 nM AUY922 or 80 nM STA-9090 for 24 h with antibodies against DDIT3 (targeting FUS-DDIT3), HSP90, HSP70, caspase 3, cleaved caspase 3 and VCL (vinculin). GAPDH was used as an internal protein loading control. The band intensities of target proteins were quantified by densitometric analysis and normalized to untreated controls. Normalized expression is shown in numbers below each blot. For double bands, both bands were included in the densitometric analysis.

Figure 2

Effect on p-RTK activity upon HSP90 inhibition. Scatter plots of global pRTK expression upon HSP90 inhibition with 100 nM 17-DMAG, 50 nM AUY922 or 80 nM STA-9090 on MLS cell lines 402-91 (A) and 2645-94 (B). Each dot represents the normalized expression of a single pRTK. Mean expression ± SD is indicated for each treatment. (C) Venn diagram indicating pRTKs downregulated by all three HSP90 inhibitors in MLS 402-91 (red), MLS 2645-94 (blue) or in both cell lines (purple). No pRTKs were commonly upregulated by all three HSP90 inhibitors in both MLS 402-91 and 2645-94. (D) Visualization of the pRTKs from the pRTK membranes that were commonly downregulated by all three HSP90 inhibitors in both MLS 402-91 and MLS 2645-94.

Figure 3

Downstream analysis on MAPK and PI3K/AKT signaling pathways. (A) Western blot analysis of ERBB3, EGFR and downstream signaling through MAPK and PI3K/AKT pathways on MLS cell lines 402-91, 1765-92 and 2645-94 treated with 100 nM 17-DMAG, 50 nM AUY922 or 80 nM STA-9090 for 24 h. VCL (vinculin) was used as an internal protein loading control. The band intensities of target proteins were quantified by densitometric analysis and normalized to untreated controls. Normalized expression is shown in numbers below each blot. For double bands, both bands were included in the densitometric analysis. (B) Illustration of affected signaling pathways after HSP90 inhibition with the effect of respective HSP90 inhibitor indicated by arrows.

Figure 4

In vivo treatment of HSP90 inhibitors on an MLS PDX model. (A) Tumor growth normalized to baseline during a 5-week treatment course with HSP90 inhibitor 17-DMAG, AUY922, STA-9090 or vehicle control. All drugs were administered by intraperitoneal injection. Each treatment group consisted of at least 8 tumors. Data is presented as mean ± SEM. Statistical analysis was performed using Kruskal–Willis test with multiple comparison correction using Dunn’s test. Significance was calculated using p-value, where p < 0.05 was considered significant (** = p < 0.01, no marker = n.s). (B) Normalized weight of mice during treatment. Data is shown as mean ± SEM. (C) Tumor growth normalized to baseline upon rechallenge with 17-DMAG in mice pretreated with AUY922. Control group consisted of mice previously receiving vehicle control, treated with vehicle control upon rechallenge of drug treatment. Initial treatment was given for 5 weeks, followed by 8 weeks of treatment intermission, followed by 3 weeks of rechallenge treatment. Each treatment group consisted of at least 4 tumors. Data is shown as mean ± SEM. (D–F) H&E staining of representative tumors, treated with indicated drugs. The lower image is an enlargement of the area indicated by the frame in the upper image.

Figure 5

Combination cell viability assays. (A–F) Dose-response curves and synergy matrixes with mean ZIP synergy score of MLS 402-91 treated with HSP90 inhibitors and doxorubicin (A,C,E) or trabectedin (B,D,F) at indicated doses. Dose-response curves display inhibition of cell viability upon indicated concentration of each drug. IC50 values for each individual drug are indicated in red. The synergy matrixes plot ZIP synergy scores for each dose combination of the drugs. A mean ZIP synergy score is displayed above each matrix. Negative values of ZIP score indicate antagonism, 0 indicates additive drug effect and positive values indicate synergistic effect.