Fig 2: Llgl1 loss promotes cell piling and increased survival in soft agar and mammary fat pads and its expression correlates with increased survival in metastatic breast cancer patientsMCF12A shControl, shLlgl1, shLlgl1 CD44hi/49flo, and shLlgl1CD44lo/49fhi cells were grown (A) on filters, (B and C) in soft agar, or (D) in the mammary fat pads of NOD-SCID mice. A. Cells were grown on transwell filters in normal growth media and probed with anti-GM130 antibody and DAPI. Cells were then imaged on the confocal microscope in the Z plane, arrows indicate polyp formation. B. and C. Cells were plated in soft agar, allowed to grow for 13 days, visualized and enumerated with ImageJ. Representative images (C) are shown beneath quantifications (B). 3 replicates per treatment group, experiment repeated twice. D. 500,000 MCF12A shControl, shLlgl1, shLlgl1 CD44hi/CD49flo, and shLlgl1 CD44lo/CD49fhi cells were suspended in matrigel and injected into the mammary fat pad of NOD-SCID mice. Cell masses were palpated and measured every 3-5 days, shControl n = 3, n = 4 for all other groups. Kaplan-Meier analysis of GSE3494 dataset, (E) N = 236 and (F). N = 234, were analyzed for Llgl1 expression and divided into two groups, top 25% expression levels of Llgl1 and the remaining 75% of lower expression. E. Survival curves show that loss of Llgl1 in breast cancer tumors correlates with lower survival probability. F. Tumor samples were further grouped by elston grades (G1 = grade 1 tumor, G3 = grade 3 tumor) and survival curves for these subgroups are shown (Bottom 75% Llgl1 expression in Grade 3 tumors = dotted green line, top 25% Llgl1 expression in Grade 3 tumors = solid pink line).

Fig 3: Hugl1 and Hugl2 regulate acinar size, nuclear size and lumen formation in MCF10As.Cells with stable expression of Hugl1-1 shRNA (MH11), Hugl2-C shRNA (MH2C) and control shRNA (MC) were grown in Matrigel for 21 days, fixed with 2% PFA, and incubated with 4',6-diamidino-2-phenylindole (DAPI). (A–D) Images were taken at 10X (scale bar = 300 microns) and acini were measured for length on longest axis in pixels. Boxplots (E, F) reflect quantification of acinar size. For each experiment, boxplots reflect data from one of three trials producing similarly significant results. Arrows point to acini reflective of mean. (G–I, L–Q) Cross-sectional confocal images were taken of each acinus at 630X on a Leica confocal microscope. (G–I) A ratio of luminal nuclei/peripheral nuclei was calculated for each acinus to quantify infilling of the lumen. Arrows indicate luminal nuclei. (J–K) Boxplots reflect quantification of luminal infilling. (L–Q) Nuclear size was measured via length on longest axis in microns. White boxes outline magnified area (L, N, P). Nuclear measurements (M, O, Q) were recorded for each acinus and the largest value for each acinus was used to generate boxplots (R, S). Two-tailed Student’s t tests were performed to calculate p values. Scale bars indicate size in microns.

Fig 4: Loss of Llgl1 induces EGF-dependent mammosphere formationA.-F. Mammosphere assays were performed using cells generated as described in Figure 1 and Figure 2. A. Mammosphere formation with EGF treatment (20ng/mL) was quantified by counting mammospheres greater than 60µm (3 replicates per treatment group, each experiment was performed 3 times). B. Due to coalescing mammospheres in shLlgl1 CD44hi/CD49flo mammopshere formation was determined by counting viable cells at each time point (3 replicates per treatment group, each experiment was performed 3 times). C. MCF12A shLlgl1 CD44hi/CD49flo cells from primary mammospheres were labeled with the lipophilic tracer dyes Di-O, Di-I, or Di-D. D.-E. Mammospheres from shLlgl1 CD44lo/CD49fhi and CD44hi/CD49flo populations were incubated with Alexa Fluor 488 phalloidin (green) and DAPI (blue). The arrow indicates smooth cortical actin (D') and the arrowhead indicates invasive cortical actin (E'). F. Mammospheres were treated with either 2µM EJ1 or 1µm AG1478. All primary passages are shown in black, secondary passages in dark gray, and tertiary passages in light gray. Error bars show ± standard deviation. *P < 0.05, **P < 0.01, ****P < 0.0001.

Fig 5: Llgl1 loss drives EGFR mislocalization and novel signal transductionA. and B. Protein lysates were collected from MCF12A parental, shControl, and shLlgl1 cells and analyzed by immunoblot using the antibodies: anti-TAZ, anti-Slug, anti-pAkt, anti-Akt, anti-EGFR, anti-E-cadherin, anti-dpERK1/2, anti-ERK1/2, anti-pStat3, anti-Stat3, anti-Llgl1, and anti-ßactin. Immunoblots against anti-ßactin are shown for each set of lysates. C.-J. MCF12A shControl and shLlgl1 cells were grown on plastic and (C-D) serum starved overnight or (E-J) in normal growth conditions then evaluated for localization of the indicated proteins. Cell were incubated with either anti-EGFR 1005, anti-TAZ, anti-SLUG, or anti-YAP antibodies and mounted (C-D) with DAPI or (E-J) without DAPI. Arrows indicate membrane localization and arrowheads indicate nuclear localization. C', D', E', F', G', H', I', and J' panels represent increased magnifications of C, D, E, F, G, H, I, and J panel insets.