Girdin is a component of the lateral polarity protein network restricting cell dissemination

Abstract

Epithelial cell polarity defects support cancer progression. It is thus crucial to decipher the functional interactions within the polarity protein network. Here we show that Drosophila Girdin and its human ortholog (GIRDIN) sustain the function of crucial lateral polarity proteins by inhibiting the apical kinase aPKC. Loss of GIRDIN expression is also associated with overgrowth of disorganized cell cysts. Moreover, we observed cell dissemination from GIRDIN knockdown cysts and tumorspheres, thereby showing that GIRDIN supports the cohesion of multicellular epithelial structures. Consistent with these observations, alteration of GIRDIN expression is associated with poor overall survival in subtypes of breast and lung cancers. Overall, we discovered a core mechanism contributing to epithelial cell polarization from flies to humans. Our data also indicate that GIRDIN has the potential to impair the progression of epithelial cancers by preserving cell polarity and restricting cell dissemination.

Fig 1. Girdin cooperates with Lgl to support epithelial cell polarity.

A-O, Embryos of the indicated genotypes were fixed and co-stained for Crb and Dlg1. M/Z stands for maternal and zygotic mutant. Panels depict a portion of the ventral ectoderm or epidermis at stage (St) 11 (A, D, G, J, M), stage 13 (B, E, H, K, N), or stage 16 (C, F, I, L, O) of embryogenesis. In each case, the apical membrane is facing up. Scale bar in A represents 10 μm, and also applies to B-O. P-S, Panels depict cuticle preparations of whole mounted embryos of the indicated genotypes. The anterior part of the embryo is oriented to the left, and the dorsal side is facing up. Scale bar in P represents 100 μm, and also applies to Q-S. Panels in A-S show representative results taken from experiments that were repeated at least three times [replicate (r) ≥3], and a minimum of 25 embryos of the indicated genotypes were analyzed in each replicate.

Fig 2. Girdin mutation enhances the delocalization of the apical protein Crb in yrt mutants.

A-L, Embryos were immunostained for Crb and Dlg1, and a part of the ventral ectoderm or epidermis was imaged by confocal microscopy (the apical side of the epithelial tissue is facing up). The developmental stage of embryos (St) is indicated in the upper right corner of panels. Scale bar in A represents 10 μm, and also applies to B-L. M-P, Cuticle preparations of whole mounted embryos are shown. Scale bar in M represents 100 μm, and also applies to N-P. Panels in A-P depict representative results taken from experiments that were repeated at least three times (r ≥3), and a minimum of 55 embryos of the indicated genotypes were analyzed in each replicate.

Fig 3. Girdin antagonizes aPKC functions.

A, Stage 11–13 wild type embryos or maternal and zygotic (M/Z) Girdin mutant embryos were homogenized and processed for Western blotting. Where indicated, samples were treated with the λ Protein Phosphatase prior to electrophoresis. Arrow indicates the position of phosphorylated Yrt proteins (^PYrt). Immunoblotting of Girdin validates the genotype of embryos, whereas Gapdh1 was used as loading control. B, Girdin null embryos were incubated in a saline solution in the absence or presence of the aPKC inhibitor CRT-006-68-54 for 1h. Embryos were then homogenized and processed for Western blotting. Arrow points to phosphorylated Yrt proteins (^PYrt). Immunoblotting of Girdin confirms the genotype of embryos, and β-Tubulin was used as loading control. C-E, Cuticle preparations of aPKC knockdown (kd) embryos. Embryos were separated in three classes (I, II, or III) to account for phenotypic variability. Scale bar in C represents 100 μm, and also applies to D and E. F, Histogram showing the phenotypic distribution in percentage (%; according to the classes defined in C-E) in collections of embryos of the following genotypes: aPKC knockdown in a wild type background (aPKC kd; n = 1312), aPKC knockdown in a Girdin mutant background (aPKC kd Girdin; n = 1573). The specified number of embryos (n) represents the sum of specimens analyzed in three independent experiments. G, Phenotypic distribution of aPKC knockdown embryos expressing GFP (aPKC kd GFP; n = 603) or FLAG-Girdin (aPKC kd FLAG-Girdin; n = 697). For F and G, embryos were analyzed in three blinded experiments, and a Chi-squared test of independence was performed (****: ρ < 0.0001).

Fig 4. Human GIRDIN is essential for epithelial morphogenesis and polarity.

A-G, Caco-2 cell cysts after 7 days in culture were observed by DIC microscopy, and the proportion of 3D cellular structures with a single prominent lumen and size were assessed (shScr, n = 27; shGIR, n = 39, r = 3 independent experiments). Arrows in B and D highlight dissociated cells. Scale bar represents 25 μm. F, Histogram displaying the luminal phenotypes observed in A-E. Error bars = sd. G, Histogram displaying mean sizes (cross-sectional area) of cysts. Error bars = sd. H-M, Caco-2 cell cysts after 7 days in culture were immunostained for PAR6 and E-CAD and visualized by confocal microscopy. Arrows show an example of a structure with PAR6 localized basally. Scale bar represents 25 μm. N-O, Caco-2 cell cysts after 7 days in culture were stained with phalloidin (F-ACTIN) and Hoechst (nuclei) and visualized by confocal microscopy. Scale bar represents 25 μm. P, Histogram displaying the number of cells counted in cross-sectional images through the middle of 3D cellular aggregates. Error bars = sd. Q-T, Caco-2 cell cysts after 7 days in culture were visualized by DIC microscopy. Cells were treated with 1.5 μM of the aPKC inhibitor CRT-006-68-54 (inhib) or vehicle control (water) for the duration of the 3D culture period. Scale bar represents 100 μm. U, Histogram displaying the proportion of 3D cellular structures with a single prominent lumen (shScr, n = 278; shGIR, n = 181; shScr+Inhib, n = 321; shGIR+Inhib, n = 196; r = 3 independent experiments). Error bars = sd. Differences were determined using ANOVA with Tukey HSD (F, G, U) or Student’s t-test (P).

Fig 5. Human GIRDIN prevents cell dissemination.

A-D, Caco-2 cells were cultured for 7 days, then observed by time-lapse confocal microscopy every 20 min for 26 h. Still frames from videos display representative events observed (shScr, n = 19; shGIR, n = 28; r = 3 independent experiments). Scale bar represents 25 μm. E, Proportion of control (shScr) and GIRDIN-knockdown cysts displaying transient cellular extensions. F, Distribution of the number of transient cellular extensions per cyst. G, Proportion of control (shScr) and GIRDIN-knockdown cysts from which cells disseminated. H-M, GIRDIN knockdown and control (shScr) Caco-2 cells were cultured for 7 days, then stained with calcein AM and ethidium homodimer to visualize live (green) and dead (magenta) cells by confocal microscopy. Scale bars represent 25 μm. N, Histogram displaying the mean number of cells detached from Caco-2 cell aggregates (shScr, n = 27; shGIR, n = 30; r = 2 independent experiments) and the proportion of live and dead detached cells. O-R, GIRDIN knockdown (n = 27, r = 2) and control (shScr, n = 27, r = 2) KRAS^G12V-expressing Caco-2 cells were stained with calcein AM and ethidium homodimer to visualize live (green) and dead (magenta) cells by confocal microscopy. Scale bars represent 25 μm. S, Histogram displaying the mean number of cells detached from cyst and the proportion of live and dead detached cells. Differences were determined using Chi-squared test (E, G) or ANOVA with Tukey HSD (N, S).

Fig 6. Altered GIRDIN expression correlates with survival in epithelial cancers.

A-F, Kaplan-Meier survival plots for GIRDIN expression in breast cancer subtypes [luminal A (A), Luminal B (B), HER2-enriched (C), and basal (D)], and lung cancer subtypes [adenocarcinoma (E), squamous (F)]. p-values for differences between groups were determined using a Log-Rank test.