Disruption of epithelial architecture caused by loss of PTEN or by oncogenic mutant p110α/PIK3CA but not by HER2 or mutant AKT1

Abstract

Genetic changes in HER2, PTEN, PIK3CA and AKT1 are all common in breast cancer and lead to the elevated phosphorylation of downstream targets of the PI3K/AKT signalling pathway. Changes in HER2, PTEN, PIK3CA and AKT have all been reported to lead to both enhanced proliferation and failures in hollow lumen formation in three dimensional epithelial culture models, but it is not clear whether these failures in lumen formation are caused by any failure in the spatial coordination of lumen formation (hollowing) or purely a failure in the apoptosis and clearance of luminal cells (cavitation). Here, we use normal murine mammary gland (NMuMG) epithelial cells, which form a hollow lumen without significant apoptosis, to compare the transformation by these four genetic changes. We find that either mutant PIK3CA expression or PTEN loss, but not mutant AKT1 E17K, cause disrupted epithelial architecture, whereas HER2 overexpression drives strong proliferation without affecting lumen formation in these cells. We also show that PTEN requires both lipid and protein phosphatase activity, its extreme C-terminal PDZ binding sequence and probably Myosin 5A to control lumen formation through a mechanism that does not correlate with its ability to control AKT, but which is selectively lost through mutation in some tumours. These findings correlate AKT-independent signalling activated by mutant PIK3CA or PTEN loss, but not strongly by HER2, with disrupted epithelial architecture and tumour formation.

Figure 1

Knockdown of PTEN disrupts lumen formation in MDCK and NMuMG epithelial cells. A and B, MDCK cells and MDCK cells with stably reduced PTEN expression (MDCK PTEN KD) were analysed. PTEN expression and AKT phosphorylation were analysed by immunoblotting (a) of cell lysates from adherent cultures, treated with either DMSO vehicle or the PI3K inhibitor LY294002 (10μM for 1h). Parallel cultures were seeded into Matrigel (b) for 10 days and fixed. Immunofluorescent detection of β-catenin localisation and DNA is shown. (c) MDCK PTEN KD cells were transduced with lentiviruses encoding wild-type PTEN or a vector control before seeding into matrigel for 10 days and staining for F-actin (red) and DNA (blue). (d and e) NMuMG cells were treated with lentiviruses expressing 3 different shRNA targeting Pten or a scrambled control shRNA. 48 hours after transduction, parallel cell samples were analysed by immunoblotting with antibodies against PTEN, P-S473 AKT or total AKT or were seeded into matrigel. After 7 days of culture in matrigel, colonies were stained for DNA (blue) and F-actin (red) and photographed. Numbers for control and PTENsh2 show the mean percentage of colonies with a single hollow central lumen +/− SD from four experiments (n>100 colonies in each experiment). (f). MDCK cells and MDCK PTEN KD cells were grown for 5 days in matrigel before colonies were fixed and stained for cleaved (active) caspase 3 (red) and DNA (blue). Quantitation shows the number of caspase 3 positive cells per cell colony as the range of data from ten colonies and the 95% confidence interval. (g) NMuMG cells were transduced with lentiviruses expressing either PTEN targeting or control scrambled (SCR) shRNA. 48 hours later, cells were then seeded into matrigel and samples fixed after 1, 2, 3 and 5 days and stained for F-actin (red) and DNA (blue).

Figure 2

PI3K inhibitors can rescue lumen formation in PTEN knockdown NMuMG cells. (a). NMuMG cells were transduced with either scrambled control viruses or those expressing PTEN targeting shRNA. Transduced cells were seeded into matrigel for 7 days in the presence of DMSO vehicle or 10μM LY294002 and stained for F-actin (red) and DNA (blue). (b). Adherent NMuMG cultures were treated with the indicated doses of the PI3K inhibitor LY294002 for 30 minutes to determine effects on AKT phosphorylation by immunoblotting. (c) Rescue of lumen formation by LY294002 in cells transduced with PTEN shRNA was quantified by counting (n>100 in each case). Data are shown as the mean percentage of colonies with a single hollow central lumen +/− the range of these means from two experiments. This experiment has been performed twice using LY294002 and twice with similar results using the alternative PI3K inhibitor GDC0941 (See Figure S4).

Figure 3

PTEN is required for the formation, but not maintenance of NMuMG cell tight junctions. NMuMG cells were transduced with scrambled (SCR) control lentiviruses or those expressing PTEN targeting shRNA. (a) 24 hours after transduction, cells were trypsinised and re-seeded at high confluence. Cell-cell junction formation was allowed to proceed for 48 hours before cells were fixed and stained for occludin, ZO-1 and F-actin. Similar data were obtained when junctions were disrupted by calcium depletion and allowed to reform after calcium re-addition (Figure S6). (b) Confluent NMuMG cells were transduced in situ and 72 hours later, PTEN expression and tight junction integrity was analysed by staining for occludin, ZO-1 and F-actin.

Figure 4

Transformation of NMuMG cells by PTEN loss, PIK3CA, AKT and HER2. NMuMG cells were transduced with lentiviruses to knockdown PTEN expression (a-c), express oncogenic mutant p110α/PIK3CA (d-f), express mutant AKT1 E17K (g-i) or over-express wild-type HER2 (j-l). (a, d, g and k) 24 hours after transduction, cells were trypsinised, re-seeded and cell-cell junction formation was allowed to proceed for 48 hours before cells were fixed and ZO-1 localised by immunofluorescence (green). (b, e, h and k) Transduced cells were seeded into matrigel for 7 days and colonies fixed and stained for DNA (blue) and F-actin (red). Quantitation of these data are shown as the mean length of ZO-1 staining per nucleus +/− SEM and mean percentage of colonies with a single central hollow lumen +/− SEM. Significant differences (T-test) are shown by p values. NS indicates not significant. Immunoblotting data to show the altered expression of PTEN, p110α, AKT or HER2 and the effects on AKT phosphorylation is shown (c, f, i and l) from parallel transduced adherent cells. (m) Proliferation of cells overexpressing GFP or HER2.

Figure 5

PTEN requires both lipid and protein phosphatase activities together to support lumen formation. To knockdown expression of the endogenous PTEN protein and replace this with PTEN mutants, NMuMG cells were simultaneously co-transduced with lentiviruses expressing shRNA and encoding either GFP or the indicated PTEN proteins. Transcribed shRNA was either a scrambled control or targeting the endogenous PTEN sequence. Virally expressed wild-type or mutant PTEN cDNAs contained a silent mutation causing resistance to the PTEN targeting shRNA used. (a and b) Transduced cells were seeded into matrigel for 7 days before fixation and staining for F-actin (red) and DNA (blue). Each image panel axis represents 615μM. Quantitation shows the mean percentage of colonies with a single central hollow lumen +/− SD from between four and six independent experiments although some (control, PTEN RNAi, PTEN wt and PTEN C124S rescue) have been performed at least eleven times each with similar results (see Figure 6). Due to variability between experiments in the hollow lumen formation in control NMuMG cells (between 71% and 46%), data are displayed relative to this control. PTEN expression and effects on AKT phosphorylation were analysed by immunoblotting of lysates from parallel adherent cell lysates. (c) To confirm the effects of PTEN Y138L expression on AKT phosphorylation in epithelial cells, PTEN null MDA-MB-468 cells were transduced with the indicated PTEN proteins and AKT phosphorylation analysed by immunoblotting.

Figure 6

Tumour derived mutants of PTEN selectively correlate tumour suppression with lumen formation. (a and b) NMuMG cells were transduced with viruses expressing either a control scrambled (SCR) shRNA or a PTEN targeting shRNA. Cells with PTEN shRNA were also transduced to express either GFP or the indicated PTEN proteins using shRNA resistant expression vectors. (a) Colonies grown in matrigel for 7 days are shown stained for DNA (blue) and F-actin (red). Each panel axis represents 615μM. (b). Parallel adherent cultures were analysed by immunoblotting. Quantification shows the mean percentage of colonies with a single central hollow lumen +/− SD from between three and six independent experiments although some (control, PTEN shRNA, and PTEN wt rescue) have been performed at least twelve times each with similar results (see Figure 5). (c) To confirm the effects of these PTEN mutants on AKT phosphorylation in PTEN-null epithelial cells, MDA-MB-468 cells were transduced as indicated and analysed by immunoblotting for PTEN and phospho-AKT. One control sample of NMuMG lysate was included to relate the level of PTEN expression to cells expressing endogenous PTEN. (d) NMuMG cells were transduced with shRNA encoding lentiviruses targeting PTEN or the indicated known PTEN binding proteins. Quantification of mean hollow lumen formation from a single experiment is shown. This experiment has been performed twice with similar results. (e) NMuMG cells were transduced with lentiviruses either to knockdown PTEN expression or express the dominant negative globular domain of MyoVa. Parallel adherent cultures were analysed by western blotting and hollow lumen formation was quantified and represented as the mean from 4 experiments +/− SD. Data were analysed by ANOVA/Tukey.