LATS1 and LATS2 suppress breast cancer progression by maintaining cell identity and metabolic state

(A) Relative total tumor weight, as percentage of total body weight of 3-mo-old Lats2-CKO PyMT and WT-PyMT littermate controls (n = number of mice); mean ± SEM; * P-value < 0.05. (B) Three mammary glands (see the Materials and Methods section) from Lats2-CKO PyMT and WT-PyMT littermate control mice were histologically scored. The most advanced pathological lesion from each mouse was tallied. (C) Representative H&E-stained sections from WT-PyMT (adenoma/MIN) and Lats2-CKO PyMT (carcinoma); scale bar = 500 μm. (D) Expression levels of Lats2 mRNA in WT-PyMT tumors of different histological stages, analyzed by RT-qPCR; mean ± SEM. (E) Left panel: Heatmap representing hierarchical clustering of global expression patterns of tumors from Lats2-CKO PyMT (n = 4) and WT-PyMT littermate controls (n = 3). Each tumor was taken from a different mouse. Standardized rld values are shown for differentially expressed genes (P-value < 0.05, n = 1131); ad = adenoma/MIN, car = carcinoma. Right panel: PCA of the most differentially expressed genes between Lats2-CKO PyMT and WT-PyMT tumors (adjP-value < 0.05), deduced from RNA-seq analysis. (F) GSEA of 1,131 genes ranked by fold change (red to blue gradient) between WT-PyMT and Lats2-CKO PyMT tumors (P-value < 0.05) and compared with genes down-regulated in LATS2^L (versus LATS^H) human lumB tumors (vertical black lines).

(A) ZR75-1 cells were treated with 1 μM 5-aza-2′-deoxycytidine (5-Aza) for 4 d. Upper panel: RT-qPCR analysis of LATS2 mRNA; mean ± SD of two technical replicates. Lower panel: Western blot analysis of LATS2 protein. (B) Functional enrichment of cell viability-related terms for differentially expressed genes in luminal cancer cell lines with transient silencing (siLATS2) or stable overexpression (OE) of LATS2, compared with controls. Ingenuity Pathway Analysis (IPA); Diseases or Functions annotations. (C) ZR75-1 cells were transiently transfected with GFP-LATS1 or GFP-LATS2, stained with anti-cleaved caspase 3 (CC3) antibody after 48 h, and subjected to imaging flow cytometry (ImageStreamX). Only cells with intact nucleus and positive GFP signal were analyzed; mean ± SEM of staining intensity; ***P-value < 0.001. (D) ZR75-1 cells were transfected with GFP-LATS1 or GFP-LATS2 or GFP only (vector). 72 h later, cell viability was assessed by PI exclusion followed by flow cytometry analysis. “Cell death” represents the ratio between the percentages of dead cells (PI⁺) in the GFP-positive population versus the GFP-negative population; mean ± SEM of three independent experiments; ***P-value < 0.001. Representative FACS results are shown on the right; in each case, upper and lower panels represent the GFP-positive and GFP-negative subpopulation, respectively, of the same transfected culture. (E) ZR75-1 cells stably transduced with MYC-LATS2 plasmid or control vector were subjected to PI exclusion analysis followed by flow cytometry. Left panel: percentage of dead (PI⁺) cells. Right panel: cell count of live (PI⁻) cells; mean ± SEM of three independent experiments; *P-value < 0.05, **P-value < 0.01. (F) Histological samples from carcinomas of Lats2-CKO PyMT and WT-PyMT littermate controls were immunostained for γH2AX and cleaved caspase 3 (CC3, left panel, scale bar = 100 μm). Right panel: mean ± SEM of percentage of positive cells, based on at least nine sections of each genotype; ***P-value < 0.001.

(A) KEGG pathways enrichment analysis for genes differentially regulated in Lats2-CKO PyMT compared with WT-PyMT tumors (adjP-value < 0.05, n = 114). (B) ZR75-1 cells stably transfected with an MYC-LATS2 plasmid (LATS2) or with control vector were cultured with or without glucose for 3 d. Cell death was measured by PI exclusion followed by FACS analysis. Values from each experiment were normalized to % PI⁺ of control cells grown in glucose-containing medium; mean ± SEM of three independent experiments. (C) Extracellular acidification rate (ECAR, indicative of glycolysis) and oxygen consumption rate (OCR, indicative of respiration) of ZR75-1 cells stably expressing MYC-LATS2, relative to vector control cells, determined by Seahorse; mean ± SEM of five independent experiments. Representative tracks of Seahorse measurements are shown on the right. (D) Respiration measured by OCR in WT-PyMT and Lats2-CKO PyMT tumor-derived cell lines; mean ± SEM in log 10 scale of three independent experiments; P-value < 0.05. Bottom: representative Seahorse track.

(A) KEGG pathways significantly enriched within the list of genes commonly down-regulated in Lats2-CKO PyMT (compared with WT-PyMT) tumors and LATS2^L lumB human tumors (TCGA). (B) RT-qPCR quantification of Pparg and Plin1 expression in tumors derived from 3 mo old Lats2-CKO PyMT (n = 9) and WT-PyMT littermate controls (n = 6); mean ± SEM; *P-value < 0.05, ***P-value < 0.001. (C) Expression levels of the indicated genes (left) and proteins (right) in cultured WT PyMT and Lats2-CKO PyMT tumor-derived cells stably transduced with LATS2 or vector control; mean ± SD of 2 technical repeats. (D) ZR75-1 cells were transfected with GFP-LATS2 or control GFP plasmid, stained 48 h later with anti-PPARγ and cleaved caspase 3 (CC3) antibodies, and subjected to imaging flow cytometry (ImageStreamX). Only GFP-positive cells with intact nuclei were analyzed. Cells with nuclear localization of the transfected protein were identified by similarity between GFP and DAPI staining. Percentage of cells positively stained for PPARγ in each subpopulation is presented. Representative images are shown at the bottom. BF = bright field. (E) Left panel: scatter plot depicting PPARG and LATS2 expression in luminal tumors in the TCGA-BRCA dataset (n = 613). Right panel: GSEA of genes ranked according to expression fold change between LATS2^L tumors and LATS^H tumors in the TCGA lumB dataset. (F) Kaplan-Meier analysis of survival probability of luminal breast cancer patients (METABRIC dataset, n = 1139; Cox proportional hazards model) defined according to expression levels of LATS2 and PPARG. (G) RT-qPCR quantification of PPARG and LPL (left) and LATS2 (right) mRNA in ZR75-1 cells treated for 48 h with increasing concentrations of RGZ (0, 10, 25, 50, and 100 μM, respectively); mean ± SD of two technical replicates. (H) ZR75-1/vector (vector) and ZR75-1/LATS2 (LATS2) cells were treated with 100 μM RGZ for 48 h, followed by PI exclusion analysis; mean ± SEM of the ratio between % PI⁺ cells in each cell type, from three independent experiments. (I) Cell lines derived from WT-PyMT and Lats2-CKO PyMT tumors were treated with 100 μM RGZ for 48 h, followed by PI exclusion analysis; mean ± SEM of the ratio between % PI⁺ cells in treated versus untreated cells, from two independent experiments.

(A) Relative tumor weight, as percentage of total body weight, in three month old Lats1-CKO PyMT and WT-PyMT littermate controls (n = number of mice); mean ± SEM; *P-value < 0.05. (B) Three mammary glands from Lats1-CKO PyMT and WT-PyMT littermate controls were histologically scored and tallied. Right panel: representative H&E-stained adenosquamous carcinoma sample from a Lats1-CKO PyMT tumor (scale bar = 200 μm). (C) Immunohistochemistry analysis of ERα protein expression in tumors from 3-mo-old Lats1-CKO PyMT and WT-PyMT littermate controls. Left panel: two representative tumor sections from each genotype (scale bar = 100 μm). Right panel: mean ± SEM of % ER⁺ cells in tumors from eight Lats1-CKO PyMT and 6 WT-PyMT mice; *P-value < 0.05. (D) Immunohistochemistry analysis of CK14, performed as in (C). Right panel depicts mean ± SEM of percentage of slide area with intense membranous CK14 staining in the invasive front of the tumor; *P-value < 0.05. (E) Heatmap representing hierarchical clustering of global expression patterns of tumors from 3-mo-old Lats1-CKO PyMT and WT-PyMT littermate controls. Standardized rld values are shown for differentially expressed genes (P-value < 0.05, n = 2029). ad = adenoma/MIN, car = carcinoma, asc = adenosquamous carcinoma; (F) Left panel: GSEA assessing PPARγ transcriptional activity in Lats1-CKO PyMT versus Lats2-CKO PyMT tumors. Genes differentially expressed (adjP-value < 0.05) in either Lats1-CKO PyMT (compared with littermate controls) or Lats2-CKO PyMT (compared with littermate controls) were ranked according to fold change differences (log ratio) and compared with a gene set comprising PPARγ target genes (El Akoum, 2014). Right panel: box plot representing mean fold change (log 2) of genes comprising the PPAR pathway (KEGG database) in Lats1-CKO PyMT and Lats2-CKO PyMT, each compared with their WT-PyMT littermate controls. Only genes with P-value of comparison < 0.05 were included; ***P-value < 0.001.

(A) Genes were ranked according to fold change between Lats1-CKO PyMT and WT-PyMT tumors (P-value < 0.05), and compared with genes differentially expressed in ER-negative PyMT tumors (GSE64453), using GSEA. (B) Genes were ranked as in (A) and compared with genes down-regulated in mature luminal cells relative to mammary stem cells (GSE19446), using GSEA. (C) Genes were ranked as in (A) and compared with genes up-regulated in human basal-like tumors relative to luminal tumors (TCGA, see the Materials and Methods section) (left panel) or genes up-regulated in human basal tumors relative to lumB (GSE45827), using GSEA. (D) Box plot representing mean fold change (log2) of genes comprising the basal tumor signature (see the Materials and Methods section and Table S3), in Lats1-CKO PyMT and Lats2-CKO PyMT tumors compared with their WT-PyMT littermate controls. (E) Heatmap depicting the expression levels of the 20 most down-regulated genes in LATS1^L tumors relative to LATS^H tumors (see Table S4 and the Materials and Methods section). Tumors were sorted according to the PAM50 classification. (F) Relative expression levels of LATS1 in different breast cancer subtypes (PAM50 and TCGA-BRCA). Numbers of samples are indicated at the bottom. ANOVA coupled with Dunnett's test; ***P-value < 0.01. (G) Immunohistochemistry (IHC) analysis of LATS1 and LATS2 in luminal and triple negative (TN) human tumors. Representative tumor sections of human lumB and TN tumors are shown in the upper panel (scale bar = 100 μm). Lower panel: human tumor samples (n = number of samples in each subtype) were scored by IHC (see the Materials and Methods section) for LATS1 and LATS2; *P-value < 0.05, ***P-value < 0.001. (H) Genes were ranked according to fold change between control siRNA-transfected and LATS1 siRNA-transfected (siLATS1) MCF7 cells and compared with the indicated gene sets (Creighton et al, 2008; Massarweh et al, 2008), using GSEA. (I) WT-PyMT, Lats1-CKO PyMT, and Lats2-CKO PyMT mice were injected IP with 4-hydroxytamoxifen (Tamoxifen) or left untreated (no treatment); n = 12, 18, 7, 10, 9, and 9, respectively.