Leukaemia Inhibitory Factor (LIF) Inhibits Cancer Stem Cells Tumorigenic Properties through Hippo Kinases Activation in Gastric Cancer

Abstract

Cancer stem cells (CSCs) present chemo-resistance mechanisms contributing to tumour maintenance and recurrence, making their targeting of utmost importance in gastric cancer (GC) therapy. The Hippo pathway has been implicated in gastric CSC properties and was shown to be regulated by leukaemia inhibitory factor receptor (LIFR) and its ligand LIF in breast cancer. This study aimed to determine LIF's effect on CSC properties in GC cell lines and patient-derived xenograft (PDX) cells, which remains unexplored. LIF's treatment effect on CSC markers expression and tumoursphere formation was evaluated. The Hippo kinase inhibitor XMU-MP-1 and/or the JAK1 inhibitor Ruxolitinib were used to determine Hippo and canonical JAK/STAT pathway involvement in gastric CSCs' response to LIF. Results indicate that LIF decreased tumorigenic and chemo-resistant CSCs, in both GC cell lines and PDX cells. In addition, LIF increased activation of LATS1/2 Hippo kinases, thereby decreasing downstream YAP/TAZ nuclear accumulation and TEAD transcriptional activity. LIF's anti-CSC effect was reversed by XMU-MP-1 but not by Ruxolitinib treatment, highlighting the opposite effects of these two pathways downstream LIFR. In conclusion, LIF displays anti-CSC properties in GC, through Hippo kinases activation, and could in fine constitute a new CSCs-targeting strategy to help decrease relapse cases and bad prognosis in GC.

Keywords: ALDH; CD44; GP190; JAK; LATS1/2; Ruxolitinib; XMU-MP-1; YAP; gastric carcinoma.

Figure 1

The Leukaemia inhibitory factor (LIF)/LIF receptor (LIFR)/JAK/STAT pathway is functional in gastric cancer (GC) and upregulated in gastric cancer stem cells (CSCs). (A) p-STAT3^Tyr705 and STAT3 protein levels in AGS and MKN45 cell lines and GC07 patient-derived xenograft (PDX) cells after time course treatment with 50 ng/mL LIF. Values under each band represent quantification of relative tubulin-normalised protein expression according to band density (whole Western Blot available in Supplementary Figure S4) (B) JAK/STAT targets relative mRNA expressions after treatment of AGS, MKN45 and GC07 cells with (green) or without (blue) LIF. (C) Relative FACS-sorted CD44+ PDX cells versus CD44- PDX cells gene expression profiles. Three groups are represented: CSC markers, JAK/STAT positive and negative regulators. (D) Flow cytometry analysis of LIFR-GP190 protein expression in CD44+/high cells and CD44-/low cells after a 48 h LIF treatment of AGS and MKN45 cells. Mean +/− SEM is represented in each quadrant of the dot plot graphs. For MKN45 cells, P1 correspond to CD44+/high population and P2 to CD44-/low population. (E) LIFR distribution in whole cell population, CD44+/high and CD44-/low cell subpopulations obtained from cytometry analysis in Figure 1D. LIFR+ cells are represented by dark grey bars and LIFR- cells by stacked light grey bars. The cumulation of both types of bars represents either the whole population, CD44+/high or CD44- cells. LIF treatments (50 ng/mL) were carried out at either different time intervals, 0, 0.5, 2, 5, 24 h (A,B) or for 48 h (A,B,D,E), 3 < n < 4. * p < 0.05, ** p < 0.005, *** p < 0.0005 and **** p < 0.0001 versus untreated controls with ANOVA statistical analyses. $ p < 0.05, $$ p < 0.005, $$$ p < 0.0005 and $$$$ p < 0.0001 versus corresponding CD44+/high cells with 2-way ANOVA tests. LIFR+ stats are represented by dark grey $ and LIFR- cells by light grey $.

Figure 2

Leukaemia inhibitory factor presents anti-CSC effects in GC. (A) 3D tumoursphere assays carried out on GC cell lines (AGS and MKN45, on the left side of the panel) and PDX cells (GC07, GC10 and GC04 on the right side of the panel). (B) Dot plot representation (upper panel) and quantification (lower panel) of flow cytometry analysis of gastric CSC markers CD44 and ALDH’s activity. Mean +/− SEM is presented in each quadrant of the dot plot graphs. For MKN45 cells, P1 corresponds to CD44+/high population and P2 to CD44-/low population. (C) Relative mRNA levels of gastric CSC markers of PDX cells GC06 and GC cell lines AGS and MKN45. All cells were treated (green) or not (blue) with 50 ng/mL LIF for 48 h. For tumoursphere assays, LIF treatment was carried out every 48 h and sphere counting performed after 7 days. 3 < n < 5, * p < 0.05, ** p < 0.005, *** p < 0.0005 and **** p < 0.0001 versus. untreated controls with Mann–Whitney and Student t-tests.

Figure 3

LIF/LIFR signalling potentiates chemotherapy effect on gastric CSCs. (A) Representative immunofluorescence images (i) and quantification (ii) of live MKN45 7-day-old spheres stained with CD44, Aldefluor reagent and Hoechst-33342 compound. White arrows point towards chemo-resistant CD44+/ALDH+Hoechst- cells and yellow arrows point towards differentiated non-CSC CD44-/ALDH-Hoechst+ cells. (B) 3D tumoursphere assays carried out on gastric cancer cell lines and patient-derived xenograft cells, treated (green) or not (blue) with 50 ng/mL LIF for 48 h. Cells were also treated or not with chemotherapy drugs Doxorubicin (DOX) and 5-Fluorouracil (5-FU). Cells treated with chemotherapy only are represented in light blue and those with combined LIF and chemotherapy treatments are in light green. LIF treatments were carried out every 48 h and tumoursphere counting was performed after 7 days. n = 3, * p < 0.05, ** p < 0.005, *** p < 0.0005 and **** p < 0.0001 versus respective untreated controls with ANOVA statistical analyses, colours representing the matched coloured bars comparison (A(ii)). $ p < 0.05, $$ p < 0.005, $$$ p < 0.0005 and $$$$ p < 0.0001 versus conditions treated with DOX or 5-FU alone with Student t-tests.

Figure 4

Leukaemia inhibitory factor activates the Hippo pathway tumour suppressor core. (A) Hippo suppressors LATS2 and p-LATS1/2^Thr1079/1041 and Hippo effectors YAP, p-YAP^Ser127 and TAZ’s protein expression in AGS and MKN45 cell lines. Values under each band represent quantification of relative GAPDH or tubulin-normalised protein expression according to band density (whole Western Blot available in Supplementary Figure S4). (B) Representative immunofluorescence images of AGS and MKN45 cells stained with anti-YAP or anti-TAZ antibodies (green). All cells were marked with phalloidin (grey) and DAPI (blue). Scale bars 10 µm. (C) Relative quantification of cells with YAP- or TAZ-positive nucleus and respective mean grey values. (D) TEAD 8xCTIIC-luciferase reporter assay showing activity of transcription factor TEAD in AGS and MKN45 cell lines. (E) Relative Hippo target genes mRNA levels in AGS (2D) and MKN45 (3D) cell lines. All cells were either untreated (blue) or treated with 50 ng/mL LIF (green) at different time intervals (0, 0.5, 2, 5, 24, 48 h). * p < 0.05, ** p < 0.005, *** p < 0.0005 and **** p < 0.0001 versus untreated controls with ANOVA statistical analyses.

Figure 5

LIF anti-CSC effects are linked to Hippo pathway core kinases activation. (A) Hippo kinase LATS2 and p-LATS1/2^Thr1079/1041 and JAK/STAT effector STAT3 and p-STAT3^Tyr705 protein expression in AGS and MKN45 cell lines. Values under each band represent quantification of relative GAPDH or tubulin-normalised protein expression according to band density (whole Western Blot available in Figure S5). (B) TEAD 8xCTIIC-luciferase reporter assay showing activity of transcription factor TEAD in AGS and MKN45 cell lines. (C) Relative Hippo target genes mRNA levels in AGS (2D) and MKN45 (3D) cell lines. (D) 3D tumoursphere assays carried out on GC cell lines (AGS and MKN45, on the left side of the panel) and PDX cells (GC07, on the right side of the panel). All cells were treated with 50 ng/mL LIF (green) and/or 0.5 µM XMU-MP-1 (XMU) (emerald green) at different time intervals (0, 2, 5, 24 or 48 h). For tumoursphere assays, LIF treatment was carried out every 48 h, 1 µM JAK1 inhibitor Ruxolitinib (Ruxo) (hatched bars) and combination of both XMU and Ruxo (checked bars) was used and sphere counting was performed after 7 days. For each experiment, inhibitors were added 30 min before each LIF stimulation. * p < 0.05, ** p < 0.005, *** p < 0.0005 and **** p < 0.0001 versus untreated controls with ANOVA statistical analyses. $ p < 0.05, $$ p < 0.005, $$$ p < 0.0005 and $$$$ p < 0.0001 versus conditions treated with XMU-MP-1 and/or Ruxolitinib alone with Student t-tests.

Figure 6

LIF-induced chemotherapy potentiating anti-CSC effects pass through the Hippo pathway. (A-B) 3D tumoursphere assays carried out on GC cell lines (A) and patient-derived xenograft cells (B), treated (green) or not (blue) with 50 ng/mL LIF for 48 h. Cells were also treated or not with chemotherapy drugs Doxorubicin (DOX) and 5-Fluorouracil (5-FU), Hippo kinase inhibitor XMU-MP-1 (XMU) and JAK1 inhibitor Ruxolitinib (Ruxo). Cells treated with chemotherapy drugs only are represented in light blue, while cells treated with LIF in combination with chemotherapy drugs and/or inhibitors are in light green. XMU-treated conditions are represented by dotted bars and Ruxo-treated conditions by checked bars. Treatments were carried out every 48 h and tumoursphere counting was performed after 7 days. * p < 0.05, ** p < 0.005, *** p < 0.0005 and **** p < 0.0001 versus untreated controls with ANOVA statistical analyses. $ p < 0.05, $$ p < 0.005, $$$ p < 0.0005 and $$$$ p < 0.0001 versus conditions treated with DOX or 5-FU alone with Student t-tests.

Figure 7

LIF and LIFR in GC and associated prognosis. (A) Oncomine data-mining analysis showing level of LIFR mRNA in (i) gastric adenocarcinoma compared with normal gastric mucosa in the Cui (n = 90) and Cho (n = 160) gastric tissue datasets and (ii) normal tissue compared with gastric cancer Lauren classification subsets (diffuse, intestinal and mixed type GC) in the Cho dataset. (B) KMplot database analysis showing overall survival probability of diffuse type and intestinal type gastric adenocarcinoma patients (23 < n < 175) according to LIFR and LIF and combined LIFR and LIF (LIFR + LIF) expression levels. * p < 0.05, ** p < 0.005, *** p < 0.0005 and **** p < 0.0001 versus untreated controls with Student t-tests and log-rank test.