Autophagy-dependent production of secreted factors facilitates oncogenic RAS-driven invasion

Abstract

The tumor-promoting functions of autophagy are primarily attributed to its ability to promote cancer cell survival. However, emerging evidence suggests that autophagy plays other roles during tumorigenesis. Here, we uncover that autophagy promotes oncogenic RAS-driven invasion. In epithelial cells transformed with oncogenic RAS, depletion of autophagy-related genes suppresses invasion in three-dimensional culture, decreases cell motility, and reduces pulmonary metastases in vivo. Treatment with conditioned media from autophagy-competent cells rescues the invasive capacity of autophagy-deficient cells, indicating that these cells fail to secrete factors required for RAS-driven invasion. Reduced autophagy diminishes the secretion of the promigratory cytokine interleukin-6 (IL-6), which is necessary to restore invasion of autophagy-deficient cells. Moreover, autophagy-deficient cells exhibit reduced levels of matrix metalloproteinase 2 and WNT5A. These results support a previously unrecognized function for autophagy in promoting cancer cell invasion via the coordinate production of multiple secreted factors.

Significance: Our results delineate a previously unrecognized function for autophagy in facilitating oncogenic RAS-driven invasion. We demonstrate that an intact autophagy pathway is required for the elaboration of multiple secreted factors favoring invasion, including IL-6.

Figure 1. Autophagy is required for the formation of invasive protrusions mediated by HRAS^V12 in 3D culture

(A-B) HRAS^V12 MCF10A cells stably expressing non-targeting control shRNA (shCNT) or shRNAs against autophagy genes (shATGs) were 3D cultured on Matrigel for the indicated number of days. Representative phase contrast images at the indicated magnifications are shown. Bar, 100μm.

Figure 2. Autophagy inhibition in HRAS^V12 cells restores basement membrane integrity and restricts ECM proteolysis in 3D culture

(A) HRAS^V12 cells expressing shCNT or shATGs were 3D cultured on Matrigel for 8 days. Structures were fixed and immunostained with antibodies against the basement membrane protein LAMA5 (human specific), counterstained with DAPI to detect nuclei, and imaged by confocal microscopy. Two representative images of each condition are shown. Bar, 50μm. (B) HRAS^V12 MCF10A cells were 3D cultured on Matrigel containing 25μg/mL fluorescein DQ-collagen IV (DQ-COL4) for 5 days. Structures were fixed, counterstained with phalloidin (to visualize F-actin) and DAPI, and imaged by confocal microscopy. Green fluorescence represents areas of proteolytic cleavage of the DQ-COL4 present in the ECM. Bar, 50μm.

Figure 3. Autophagy inhibition in HRAS^V12 MCF10A structures does not promote apoptosis or proliferation arrest

(A) Left: Two representative images of day 8 3D cultures of BABE and HRAS^V12 MCF10A cells expressing shCNT or shATGs immunostained with antibody against cleaved CASP3 and counterstained with DAPI to detect nuclei. Bar, 50μm. Right: Quantification of cleaved CASP3 positive cells present within 3D cultures of each indicated cell type (mean +/− s.d., Student’s t-test). (B) Representative phase (top) and corresponding wide-field fluorescence (bottom) images of BABE and HRAS^V12 cells expressing shCNT or shATGs stained with the intravital dye ethidium bromide (EtBr). Bar, 100μm. (C) Left: Two representative images of day 8 3D cultures of BABE and HRAS^V12 cells expressing shCNT or shATGs immunostained with antibody against Ki67 and DAPI counterstained. Bar, 50μm. Right: Quantification of Ki67 positive nuclei present within 3D cultures of each indicated cell type (mean +/− s.d., Student’s t-test).

Figure 4. ATG knockdown suppresses the motility and reduces the metastatic potential of cells expressing oncogenic RAS

(A) Representative images (left) and quantification (right) of wounding assay on HRAS^V12 MCF10A cells expressing shCNT or shATGs. Confluent monolayers were scratched and wound width was measured at 0 and 6h after initial wounding to quantify the decrease in scratch width. (mean +/− s.d., Student’s t-test, shCNT n=16, shATG7-2 n=8, shATG12 n=14). Bar, 100μm. (B) Transwell migration of HRAS^V12 MCF10A cells expressing shCNT or shATGs. 24h after plating, cells that migrated to the bottom of the filter were stained with crystal violet. Results are expressed as the mean crystal violet extracted from stained cells (mean +/− s.d., Student’s t-test, n=9). (C) Wounding assays of MDA-MB-231 cells expressing siATGs or in presence of 10nM bafilomycin A (BafA). Graphs represent the decrease in scratch width at 10h and 9h after initial wounding, respectively (mean +/− s.d., Student’s t-test, siCNT n=16, siATG7 n=16, siATG12 n=10, DMSO n=6, BafA n=6). (D) Representative images (left) and quantification (right) of ZsGreen positive metastatic foci following tail vein injection of ZsGreen expressing HRAS^V12 shCNT, shATG7-1 or shATG12 cells (mean +/− s.e.m., shCNT n=7, shATG7-1 n=7, shATG12 n=8).

Figure 5. ATG knockdown in HRAS^V12 cells inhibits the production of pro-invasive secreted factors in 3D culture

(A) 3D co-culture of HRAS^V12 shATG7-1 with HRAS^V12 shCNT cells rescues invasion of HRAS^V12 shATG7-1 cells. HRAS^V12 shATG7-1 cells expressing GFP were cultured for 8 days in 3D either alone (left) or together with HRAS^V12 shCNT cells expressing an empty vector (BABE). Structures were imaged by phase contrast and wide-field fluorescence microscopy or fixed, counterstained with phalloidin (to visualize F-actin) and DAPI and imaged by confocal microscopy. Phase: Bar, 100μm. Confocal: Bar, 50 μm. (B) HRAS^V12 MCF10A cells expressing shATGs were cultured in 3D for 3d and subsequently treated with BABE or HRAS^V12 shCNT conditioned media (CM). Representative phase contrast images at 24h and 72h following the addition of CM. Bar, 100μm. (C) 3D cultures of HRAS^V12 MCF10A cells expressing shATGs were treated with BABE or HRAS^V12 shCNT CM for 72h; thereafter, cultures were fixed and immunostained with an antibody against LAMA5 (human specific) to detect basement membrane and DAPI counterstained. Two representative images per condition are shown. Bar, 50μm.

Figure 6. Autophagy supports IL6 secretion necessary for oncogenic RAS-driven invasion in 3D culture

(A) Levels of IL6 in conditioned media collected on day 6 from 3D cultures of the indicated cell types. (mean +/− s.d., ANOVA, BABE n=3, HRAS^V12 n=5). (B) IL6 expression levels normalized to GAPDH in cells collected from day 8 3D cultures. (mean relative to BABE +/− s.d., Student’s t-test, n=3). (C) IL6 protein levels in day 8 3D cultures from the indicated cell types. (D) Representative phase contrast images of HRAS^V12 shATG 3D cultures treated for 48h with BABE CM (top) or with HRAS^V12 shCNT CM containing an IL6 function-blocking antibody (bottom) or IgG control antibody (middle). Bar, 100μm. (E) Representative phase contrast images of HRAS^V12 shATG 3D cultures grown in the presence or absence of 200ng/mL recombinant human IL6 for 7d. Bar, 100μm.

Figure 7. WNT5A and MMP2 are reduced following autophagy inhibition in 3D culture

(A-B) RNA was isolated from BABE, HRAS^V12 shCNT, and HRAS^V12 shATG cells cultured in 3D for 8 days. Expression levels of MMP2 and WNT5A were determined by qPCR and normalized to an internal control GAPDH. Results represent the mean relative to BABE +/− s.d. (MMP2, n=4; WNT5A, n=3; Student’s t-test). (C) Conditioned media was collected from BABE, HRAS^V12 shCNT and HRAS^V12 shATG cells grown in 3D culture. Activity levels of MMP9 and MMP2 in the conditioned media were determined by zymography. (D) HRAS^V12 shCNT cells were grown in the absence (top) or presence (bottom) of 25μm Arp-100. Left: Structures were imaged on day 8 by phase contrast microscopy. Right: Representative confocal images of structures immunostained with anti-phospho-ERM (P-ERM) to detect cell borders and counterstained with DAPI. Bars, 100μm. (E) BABE, HRAS^V12 shCNT, and HRASV12 shATG cells were collected from 3D culture on day 8, lysed, and protein levels of WNT5A were determined by immunoblot analysis. (F) HRAS^V12 shATG7-1 cells were grown in 3D for 8 days in the absence (top) or presence (bottom) of 500ng/mL WNT5A. Left: Representative phase contrast images. Right: Representative confocal images of structures immunostained with anti-phospho-ERM to detect cell borders and counterstained with DAPI. Bars, 100μm.