Differential regulation of rho GTPases during lung adenocarcinoma migration and invasion reveals a novel role of the tumor suppressor StarD13 in invadopodia regulation

Abstract

Background: Lung cancer is the second most commonly occurring cancer. The ability to metastasize and spread to distant locations renders the tumor more aggressive. Members of the Rho subfamily of small GTP-binding proteins (GTPases) play a central role in the regulation of the actin cytoskeleton and in cancer cell migration and metastasis. In this study we investigated the role of the RhoA/Cdc42 GAP, StarD13, a previously described tumor suppressor, in malignancy, migration and invasion of the lung cancer cells A549.

Methods: We knocked down StarD13 expression in A549 lung cancer cells and tested the effect on cell migration and invadopodia formation using time lapse imaging and invasion assays. We also performed rescue experiments to determine the signaling pathways downstream of StarD13 and transfected the cells with FRET biosensors for RhoGTPases to identify the proteins involved in invadopodia formation.

Results: We observed a decrease in the level of expression of StarD13 in lung tumor tissues compared to normal lung tissues through immunohistochemistry. StarD13 also showed a lower expression in the lung adenocarcinoma cell line A549 compared to normal lung cells, WI38. In addition, the depletion of StarD13 increased cell proliferation and viability in WI38 and A549 cells, suggesting that StarD13 might potentially be a tumor suppressor in lung cancer. The depletion of StarD13, however, inhibited cell motility, conversely demonstrating a positive regulatory role in cell migration. This was potentially due to the constitutive activation of RhoA detected by pull down and FRET assays. Surprisingly, StarD13 suppressed cell invasion by inhibiting Cdc42-mediated invadopodia formation. Indeed, TKS4 staining and invadopodia assay revealed that StarD13 depletion increased Cdc42 activation as well as invadopodia formation and matrix degradation. Normal lung cells depleted of StarD13 also produced invadopodia, otherwise a unique hallmark of invasive cancer cells. Cdc42 knock down mimicked the effects of StarD13, while overexpression of a constitutively active Cdc42 mimicked the effects of its depletion. Finally, immunostaining and FRET analysis revealed the absence of StarD13 in invadopodia as compared to Cdc42, which was activated in invadopodia at the sites of matrix degradation.

Conclusion: In conclusion, StarD13 plays distinct roles in lung cancer cell migration and invasion through its differential regulation of Rho GTPases. Video abstract.

Keywords: Invadopodia; Invasion; Migration; Rho GTPases; StarD13.

Fig. 1

StarD13 is a potential tumor suppressor in lung adenocarcinoma. a Representative fluorescent micrographs of formalin-fixed normal lung tissue from biopsy (upper) or lung adenocarcinoma tissue (lower), paraffin embedded, sectioned and then stained with DAPI (left panels) and immunostained with anti-StarD13 antibody (middle panels). The right hand panels show the merged channels. The graph is a quantitation of the immunohistochemistry. The mean fluorescent intensity/pixel was measured using ImageJ software and expressed to the corresponding tissues. Data are the mean −/+ SEM from 3 different experiments (with 4 tissues each) and *p < 0.05. Scale bar is 100 μm. b Data analyzed from Oncomine website. mRNA of the indicated number of samples (indicated in the legend) were quantified for expression levels of StarD13 in Landi Lung. c A549 and WI38 cells were lysed and immunoblotted by western blot analysis for StarD13 (upper gel) or for β-actin (lower gel) for loading control. The graph is a quantitation (using the ImageJ software) of the bands from the StarD13 gel normalized to the corresponding bands in the actin gel. Data are the mean −/+ SEM from 3 different experiments and *p < 0.05. d A549 cells were transfected with luciferase control siRNA or with 2 oligos of StarD13 siRNA for 72 h. The cells were then lysed and immunoblotted by western blot analysis for StarD13 (upper gel) or for β-actin (lower gel) for loading control. The graph is a quantitation (using the ImageJ software) of the bands from the StarD13 gel normalized to the corresponding bands in the actin gel. Data are the mean −/+ SEM from 3 different experiments and *p < 0.05. e Cell proliferation was determined using WST-1 reagent. Cell viability of siRNA-transfected A549 cells was expressed as fold increase from control (luciferase-transfected). Data are the mean −/+ SEM from 3 different experiments and *p < 0.05. d The micrographs are phase contrast representative images of luciferase siRNA- or StarD13 siRNA-transfected WI38 cells grown in culture for 72 h. Cell proliferation for siRNA-transfected WI38 cells was determined using WST-1 reagent and cell viability expressed as fold increase from control (luciferase-transfected). Data are the mean −/+ SEM from 3 different experiments and *p < 0.05. g A549 cells were transfected with luciferase or StarD13 siRNA permeabilized and stained with 30 μL propidium iodide for 10 min. Cells were analyzed using a C6 flow cytometer, which indicated the distribution of the cells into their respective cell cycle phases based on their DNA content. G0/G1 cells were 2n; S-phase cells were > 2n but <4n while G2/M were 4n. Cell DNA content was determined by CellQuest software. The graph is a quantitation expressed as percentage of cells in G1 and combined percentage of cells in S + G2. Data is the mean +/− SEM from 3 different experiments. h A549 cells were transfected with either control luciferase siRNA or with StarD13 siRNA. The cells were then trypsinized and stained with 5 μL of Annexin V FITC and 10 μL of Propidium Iodide. The fluorescence of the cells was determined immediately with a flow cytometer. Cells, which are early in the apoptotic process, will stain with the Annexin V FITC Conjugate alone. Live cells will show no staining by either the Propidium Iodide solution or Annexin V FITC conjugate. Necrotic cells will be stained by both the Propidium Iodide solution and Annexin V FITC conjugate. The graph is a quantitation expressed as percentage of cells. Data is the mean +/− SEM from 3 different experiments and *p < 0.05. i A549 transfected with Luciferase or StarD13 siRNA were grown in suspension and mixed with Matrigel to form spheres, as described in the Materials and Methods. The graphs are a quantitation of sphere diameter (μm) and area (μm²) of the spheres. Data is the mean +/− SEM from 50 spheres/condition and *p < 0.05

Fig. 2

StarD13 is required for cell motility in lung adenocarcinoma. a A549 cells were transfected with luciferase control siRNA or with StarD13 siRNA (2 oligos) for 72 h. Cells were grown in a monolayer then wounded and left to recover the wound then imaged at the same frame after 48 h (lower micrographs). The graph is a quantitation where wound widths were measured at 11 different points for each wound, and the average rate of wound closure for the luciferase and the StarD13siRNA-transfected cells was calculated in μm/hr. Data are the mean −/+ SEM from 3 independent experiments and *p < 0.05. Scale bar is 100 μm. b The net paths of projected 120 frames from 2 h long time lapse movies of cells (SF268) transfected with luciferase control siRNA or with StarD13 siRNA undergoing random motility in serum (each trace is a cell). The right hand panels for each condition are a close-up trace of a cell area and a net path projected over 2 h. Graph is a quantitation of the cell speed for the time lapse movies in μm/min. Data are the mean −/+ SEM from 15 cells (from 3 independent experiments/10 movies/condition/experiment) and *p < 0.05. Scale bar is 20 μm

Fig. 3

StarD13 silencing inhibits lung adenocarcinoma motility through the constitutive activation of RhoA and inhibition of Rac1. a A549 cells were transfected with either luciferase as control or StarD13 siRNA. After 72 h, cells were lysed and incubated with GST-RBD (Rhotekin binding domain) to pull down active RhoA. Samples were then blotted with RhoA antibody. The lower gel is a western blot for the total cell lysate, used as a loading control. The graph is a quantitation of the active RhoA bands using the ImageJ software. The bands were normalized to the amount of total proteins and the data presented as fold change to luciferase control. Data are the mean from 3 experiments −/+ SEM and *p < 0.05. b Representative micrographs of A549 cells transfected with luciferase or StarD13 siRNA and with the RhoA FRET biosensor and fixed. The cells were then imaged in CFP, YFP and FRET channels and image analysis performed as described in materials and methods to obtain the FRET ratios (RhoA activation). The graph is a quantitation of the FRET signal (Ratio FRET/CFP signal) in the total cell area presented as fold difference to luciferase control. Data are the mean −/+ SEM from 15 cells (from 3 independent experiments) and *p < 0.05. Scale bar is 10 μm. c A549 cells were transfected with luciferase control siRNA or with RhoA siRNA for 72 h. The cells were then lysed and immunoblotted for RhoA or β-actin (lower gel) for loading control. The graph is a quantitation of the bands from the RhoA gel, using the imageJ software, normalized to the corresponding bands in the actin gel. Data are the mean −/+ SEM from 3 different experiments and *p < 0.05. d A549 cells were transfected with either luciferase, StarD13, RhoA, or StarD13 + RhoA siRNA. After 72 h, cells were lysed and incubated with GST-CRIB (Cdc42 and Rac interactive binding protein) to pull down active Rac1. Samples were then blotted with Rac1 antibody. The lower gel is a western blot for the total cell lysate, used as a loading control. The graph is a quantitation of the active Rac1 bands using the ImageJ software. The bands were normalized to the amount of total proteins and the data presented as fold change to luciferase control. Data are the mean from 3 experiments −/+ SEM and *p < 0.05. e A549 cells were transfected with luciferase or Rac1 siRNA for 72 h. The cells were then lysed and immunoblotted for Rac1 or β-actin (lower gel) for loading control. The graph is a quantitation of the bands from the Rac1 gel, using the imageJ software, normalized to the corresponding bands in the actin gel. Data are the mean −/+ SEM from 3 different experiments and *p < 0.05. f A549 cells were transfected with luciferase siRNA + pcDNA3.1 vector, RhoA-CA (constitutively active), Rac1 siRNA, StarD13 siRNA + Rac1-CA (constitutively active), RhoA siRNA, and StarD13 + RhoA siRNA. Cells were then grown in a monolayer, wounded and left to recover the wound then imaged at the same frame after 48 h (lower micrographs). The graph is a quantitation of the wounds in F. Wound widths were measured at 11 different points for each condition and the average rate of wound closure was calculated in μm/hr. Data are the mean −/+ SEM from 3 wound closure assays and *p < 0.05. Scale bar is 100 μm

Fig. 4

StarD13 silencing increases the size of adhesion structures in lung adenocarcinoma cells and inhibits their tail detachment. a Representative micrographs of A549 cells transfected with luciferase control, Rac1 siRNA, RhoA siRNA, StarD13 siRNA or RhoA-CA (constitutively active) that were fixed and immunostained with anti-Vinculin. Scale bar is 10 μm. The graph is a quantitation (as described in the materials and methods) represented as average adhesion size in cells of each condition presented as fold difference to luciferase control. Data are the mean −/+ SEM from 15 cells/condition from 3 different experiment and *p < 0.05. b Montage of time-lapse movie (showing frames that are 4 min apart) of Luciferase siRNA- or StarD13 siRNA-transfected A549 cells undergoing random motility in serum (for 2 h). The black arrows indicate the direction of movement. Scale bar is 10 μm

Fig. 5

StarD13 silencing increases lung adenocarcinoma cell invasion through an increase in Cdc42 activation. a A549 cells were transfected with luciferase siRNA or StarD13 siRNA (2 oligos) for 72 h. Cells were then allowed to invade towards 10% FBS for 24 h. 1 × 106 cells/ml were used in each assay. The micrographs are representatives of invaded cells on the bottom side of the membrane stained with cell stain according to assay instructions. Cell stain was then extracted and colorimetric measurements were taken at 560 μm. The graph shows the measurements. Data are the mean −/+ SEM from 3 experiments and *p < 0.05. b A549 cells were transfected with vector alone or with RhoA-CA. The micrographs are representatives of invaded cells on the bottom side of the membrane stained with cell stain according to assay instructions. Cell stain was then extracted and colorimetric measurements were taken. Data are the mean −/+ SEM from 3 experiments and *p < 0.05. c A549 cells were transfected with either luciferase or StarD13 siRNA. After 72 h, cells were lysed and incubated with GST-CRIB (Cdc42 and Rac interactive binding protein) to pull down active Cdc42. Samples were then blotted with Cdc42 antibody. The lower gel is a western blot for the total cell lysate, used as a loading control. The graph is a quantitation of the active Cdc42 bands using the ImageJ software. The bands were normalized to the amount of total proteins and the data presented as fold change to luciferase control. Data are the mean from 3 experiments −/+ SEM and *p < 0.05. d A549 cells were transfected with luciferase or StarD13 siRNA and with the Cdc42 FRET biosensor. The cells were then imaged in CFP, YFP and FRET channels and image analysis performed as described in materials and methods to obtain the FRET ratios (Cdc42 activation). The graph is a quantitation of the FRET signal (Ratio FRET/CFP signal) in the total cell area presented as fold difference to luciferase control. Data are the mean −/+ SEM from 15 cells (from 3 independent experiments) and *p < 0.05. Scale bar is 10 μm. e A549 cells were transfected with either vector alone control or with Cdc42-CA (constitutively active). Cells were then grown in a monolayer, wounded and left to recover the wound then imaged at the same frame after 48 h. The graph is a quantitation of the wounds obtained. Wound widths were measured at 11 different points for each condition and the average rate of wound closure was calculated in μm/hr. Data are the mean −/+ SEM from 3 wound closure assays and *p < 0.05. f A549 cells were transfected with luciferase control siRNA or Cdc42 siRNA for 72 h. The cells were then lysed and immunoblotted by western blot analysis for Cdc42 (upper gel) or for β-actin (lower gel) for loading control. The graph is a quantitation of the bands from the Cdc42 gel, using the imageJ software, normalized to the corresponding bands in the actin gel. Data are the mean −/+ SEM from 3 different experiments and *p < 0.05. g A549 cells were transfected with luciferase siRNA + vector alone, StarD13 siRNA Cdc42-DA, or StarD13 siRNA +Cdc42 siRNA for 72 h. Cells were then allowed to invade towards 10% FBS for 24 h. Cells are then stained, stains extracted and colorimetric measurements taken at 560 μm. The graph shows the measurements. Data are the mean −/+ SEM from 3 experiments and *p < 0.05

Fig. 6

StarD13 silencing increases Cdc42-mediated invadopodia formation in lung adenocarcinoma cells. a Representative micrographs of A549 cells cells transfected with luciferase siRNA + vector alone, StarD13 siRNA, Cdc42-CA (constitutively active Cdc42), or StarD13 siRNA +Cdc42 siRNA for 72 h, fixed and immunostained for TKS4. Scale bar is 10 μm. The graph is a quantitation of the number of invadopodia per cell (described in materials and methods) expressed as values in every condition (15 cells/condition/experiment). Data are the mean of +/− SEM of 3 independent experiments and *p < 0.05. b Representative micrographs of WI38 cells transfected with luciferase or starD13 siRNA, fixed and immunostained for TKS4 and stained with Rhodamin-Phalloidin. Scale bar is 10 μm. The graph is a quantitation of the number of invadopodia per cell expressed as values in every condition. Data are the mean of +/− SEM of 3 independent experiments and *p < 0.05. c Representative micrographs of control A549 cells fixed and immunostained for TKS4 and StarD13. Scale bar is 10 μm. d Representative histogram of fluorescence intensity across the indicated lines within the cell (across invadopodia) stained for TKS4 and StarD13. The intensity is plotted as a function of distance (in pixels). e Representative micrographs of control A549 cells plated on Alexa568-labeled matrix for 8 h. Cells were then fixed and stained for TKS4 and Cdc42 as well as the Rhodamin channel for matrix degradation. Scale bar is 10 μm. f Representative micrographs of control A549 cells untreated and transfected with the Cdc42 FRET biosensor for 24 h then plated on Alexa568-labeled matrix for 8 h. Cells were then fixed and imaged in CFP, YFP and FRET channels to obtain the Cdc42 FRET signal as well as the Rhodamin channel for matrix degradation. Scale bar is 10 μm. g Representative micrographs of cells transfected with luciferase or StarD13 or StarD13+ Cdc42 siRNA along with a GFP vector, or with GFP-Cdc42-Q61L, or with StarD13 siRNA + GFP-Cdc42-Q61L. siRNA + GFP vector alone, StarD13 siRNA Cdc42-DA, or StarD13 siRNA +Cdc42 siRNA and plated on Alexa568-labeled matrix for 48 h. Cells were then fixed and immunostained for TKS4. Scale bar is 10 μm. The graph is a quantitation matrix degradation index expressed as ratio of mean fluorescent intensity in an ROI in the background to mean fluorescent intensity in the cell trace (as described in materials and methods). Data are the mean −/+ SEM from 3 experiments and *p < 0.05

Fig. 7

Model. The role of StarD13 in 2D and 3D migration in lung adenocarcinoma cells. The model depicts the cyclical inhibition of StarD13 of RhoA (temporal regulation) at the leading edge of migrating cells as well as the tail. This allows the cell to detach at the tail through the dissolution of focal adhesions once RhoA inactivates. At the leading edge, StarD13 inactivates RhoA which allows Rac1 to activate leading to the formation of new point contacts. RhoA then cycles back to its active form, attenuating Rac1 activity and leading to the maturating of point contacts into focal adhesions. This allows the protrusion to constructively pull the cell forward in its 2D direction. In 3D, StarD13 potentially plays a spatial inhibitory role for Cdc42 keeping its activity concentrated in invadopodia