JAK-STAT signalling controls cancer stem cell properties including chemotherapy resistance in myxoid liposarcoma

Abstract

Myxoid liposarcoma (MLS) shows extensive intratumoural heterogeneity with distinct subpopulations of tumour cells. Despite improved survival of MLS patients, existing therapies have shortcomings as they fail to target all tumour cells. The nature of chemotherapy-resistant cells in MLS remains unknown. Here, we show that MLS cell lines contained subpopulations of cells that can form spheres, efflux Hoechst dye and resist doxorubicin, all properties attributed to cancer stem cells (CSCs). By single-cell gene expression, western blot, phospho-kinase array, immunoprecipitation, immunohistochemistry, flow cytometry and microarray analysis we showed that a subset of MLS cells expressed JAK-STAT genes with active signalling. JAK1/2 inhibition via ruxolitinib decreased, while stimulation with LIF increased, phosphorylation of STAT3 and the number of cells with CSC properties indicating that JAK-STAT signalling controlled the number of cells with CSC features. We also show that phosphorylated STAT3 interacted with the SWI/SNF complex. We conclude that MLS contains JAK-STAT-regulated subpopulations of cells with CSC features. Combined doxorubicin and ruxolitinib treatment targeted both proliferating cells as well as cells with CSC features, providing new means to circumvent chemotherapy resistance in treatment of MLS patients.

Keywords: JAK-STAT signalling; LIF; SWI/SNF; cancer stem cells; chemotherapy resistance; myxoid liposarcoma; ruxolitinib.

Figure 1

Sarcosphere formation and side population analysis in MLS. (a) Sarcosphere formation efficiency of MLS 2645‐94, 1765‐92 and 402‐91 cells. The sarcosphere formation efficiency is calculated as the fraction of formed spheres in relation to the total number of seeded cells. Mean ± SEM is shown, n = 3. (b) Side population (SP) analysis of MLS 2645‐94, 1765‐92 and 402‐91 using Hoechst dye staining. The left panel represents untreated control cells, while the right panel represents cells treated with verapamil. The SP cells (red) are gated and their fraction is compared to the total viable cell population. (c) SP analysis of anoikis‐resistant cells compared to non‐enriched control cells. Three independent experiments (red, orange and blue) were performed. Mean ± SEM is shown, paired student t‐test *p < 0.05. [Color figure can be viewed at wileyonlinelibrary.com]

Figure 2

The effects of doxorubicin treatment in MLS. (a) Doxorubicin IC₅₀ analysis in MLS 2645‐94, 1765‐92 and 402‐91. Dose–response curve after 48 h doxorubicin treatment was generated using Alamar Blue assay. The dashed gray horizontal line indicates 50% relative proliferation. IC₅₀ was obtained by applying a symmetrical sigmoidal curve fit. Mean ± SEM is shown, n = 4. (b) Sarcosphere formation capacity of MLS cells treated for 48 h with doxorubicin relative control. The graphs show the relative number of sarcopheres formed in control (DMSO) and doxorubicin‐treated samples 3 days after cell seeding in non‐adherent sphere formation assay. Three independent experiments (green, red and blue) were performed, where each experiment consisted of the average number of spheres formed from three technical replicates. Mean ± SEM is shown, paired student t‐test *p < 0.05. (c) Effect of doxorubicin on side population (SP) cells using Hoechst dye staining. The top panel represents control (DMSO) cells, while the bottom panel represents cells treated with doxorubicin for 48 h. The SP cells (red) are gated and their fraction is compared to the total viable cell population. Note that doxorubicin arrests cells in the G2 cell cycle phase. The MLS 402‐91 control is identical to the control in Figure 1b, since both experiments were analyzed at the same time. [Color figure can be viewed at wileyonlinelibrary.com]

Figure 3

Properties of JAK–STAT signalling in MLS. (a) Schematic overview of canonical JAK–STAT signalling. (b) Heat‐map visualization of pair‐wise gene correlations using Spearman's correlation coefficient. (c) Western blot analysis of JAK1, JAK2, STAT3 and phospho‐STAT3 (p‐STAT3) in MLS 2645‐94, 1765‐92 and 402‐91. GAPDH was used as an internal protein loading control. All samples for respective antibody were analysed on the same western blot (Supporting Information Fig. S7). Blot data are cut and reorganized for visualization purposes. (d) Immunohistochemical staining of phospo‐STAT3 and Ki67 in MLS tumour tissue. Brown staining shows reactivity with the specific antibody. Cells expressing phospho‐STAT3 are concentrated at the periphery of a tumour lobule, whereas Ki67‐positive cells are scattered throughout the tumour. Scale bar is 0.1 mm. [Color figure can be viewed at wileyonlinelibrary.com]

Figure 4

Ruxolitinib inhibits JAK–STAT signalling in MLS. (a) Ruxolitinib inhibition of JAK1/2. Western blot analysis of STAT3 and phospho‐STAT3 in MLS 2645‐94, 1765‐92 and 402‐91 cells after 24 h ruxolitinib treatment compared to untreated control (DMSO) cells. GAPDH was used as an internal protein loading control. All samples for respective antibody were analysed on the same western blot (Supporting Information Fig. S8). Blot data are cut and reorganized for visualization purposes. (b) Sarcosphere formation capacity relative control of MLS cells treated 24 h with ruxolitinib. The graphs show the relative number of sarcospheres formed in control (DMSO) and ruxolitinib‐treated samples 3 days after cell seeding in non‐adherent sphere‐formation assay. Four independent experiments (green, red, orange and blue) were performed, where each experiment consisted of the average number of spheres formed from three technical replicates. Mean ± SEM is shown, paired student t‐test *p < 0.05. (c) Effect of ruxolitinib on cell proliferation. Dose response curve of ruxolitinib were generated using the Alamar Blue assay after 48 h treatment. Mean ± SEM is shown, n = 4. (d) Effect of ruxolitinib on cell migration. Cell migration was measured using a scratch assay, where the relative migration ratio between control (DMSO) and ruxolitinib‐treated cells was calculated. The red line at 100% indicates where ruxolitinib‐treated and untreated control cells display identical migration capacity. Mean ± SEM is shown, n = 3. [Color figure can be viewed at wileyonlinelibrary.com]

Figure 5

LIF acts through JAK–STAT and SWI/SNF signalling in MLS. (a) Relative protein expression level of IL6R, LIFR and GP130 in MLS 2645‐94, 1765‐92 and 402‐91 cells determined by flow cytometry. Details are shown in Supporting Information Figure S5. (b) Western blot analysis of phospho‐STAT3 (p‐STAT3) expression at different time points after LIF stimuli in MLS cells. GAPDH was used as an internal protein loading control. All samples for respective antibody were analysed on the same western blot (Supporting Information Fig. S9). Blot data are cut and reorganised for visualization purposes. (c) Western blot analysis of phospho‐STAT3 (p‐STAT3) in MLS cells after 24 h treatment with LIF, ruxolitinib or LIF and ruxolitinib in combination. GAPDH was used as an internal protein loading control. All samples for respective antibody were analysed on the same western blot (Supporting Information Fig. S10). Blot data are cut and reorganized for visualization purposes. (d) Sarcosphere formation capacity relative control of MLS cells treated 24 h with LIF. The graphs show the relative number of sarcospheres formed in control (1% FBS) and LIF‐treated samples 3 days after cell seeding in a non‐adherent sphere formation assay. Three independent experiments (orange, red and blue) were performed, where each experiment consisted of the average number of spheres formed from three technical replicates. Mean ± SEM is shown, paired student t‐test *p < 0.05 **p < 0.01. (e) Co‐immunoprecipitation of SMARCA4. Western blot analysis of phosphorylated STAT3 co‐immunoprecipitated with nuclear extracted SMARCA4 or IgG. B represents bound phosphorylated STAT3 to SMARCA4 or IgG, and NB represents non‐bound phosphorylated STAT3, I represents input material. Corresponding western blots for SMARCA4 are shown in Supporting Information Figure S6. (f) SMARCA4 and phospo‐STAT3 downregulation after SMARCA4 knock‐down using RNAi. The protein expression (in percentage) after knock‐down related to control is shown. Each protein expression was normalized against GAPDH expression before comparison. Corresponding western blots are shown in Supporting Information Figure S11. (g) Sarcosphere formation capacity of SMARCA4 knock‐downed MLS cells. The graph shows the relative number of sarcospheres formed in control and SMARCA4 knock‐downed samples 3 days after cell seeding in non‐adherent sphere‐formation assay. Three independent experiments were performed for each MLS cell line, where each experiment consisted of the average number of spheres formed from three technical replicates. Statistical analysis were performed by Student's paired t‐test *p < 0.05. The reduced sarcosphere formation capacity reached statistical significance only when all three MLS cell lines were analysed together. [Color figure can be viewed at wileyonlinelibrary.com]

Figure 6

Targeting cancer stem cell like cells and proliferating cells as a strategy to treat MLS. (a) Sarcosphere formation capacity relative control of MLS 2645‐94, 1765‐92 and 402‐91 cells when treated 48 h with doxorubicin and ruxolitinib in combination. The graphs show the relative number of sarcospheres formed in control (DMSO), doxorubicin‐ and doxorubicin/ruxolitinib‐treated samples 3 days after cell seeding in non‐adherent sphere formation assay. Three independent experiments (orange, red and blue) were performed, where each experiment consisted of the average number of spheres formed from three technical replicates. Mean ± SEM is shown, paired student t‐test *p < 0.05 **p < 0.01. (b) Phospho‐STAT3 expression. Western blot analysis of phospho‐STAT3 (p‐STAT3) in MLS cells after 48 h treatment with doxorubicin, ruxolitinib or doxorubicin and ruxolitinib in combination. GAPDH was used as an internal protein loading control. All samples for respective antibody were analysed on the same western blot (Supporting Information Fig. S12). Blot data are cut and reorganized for visualization purposes. (c) Microarray analysis. Venn diagrams showing the number of genes, after filtering, with a two‐fold upregulation or downregulation in doxorubicin‐treated MLS cells. (d) Bar chart showing, out of the 51 genes upregulated in all three cell lines, significantly enriched (FDR q‐value <0.05) gene sets among the Hallmark gene sets from the Molecular Signatures Database. [Color figure can be viewed at wileyonlinelibrary.com]