Selective Reversible Inhibition of ... - Cancer Research · Orthotopic xenograft mouse model of breast cancer using MDA-MB-231 cells All animal studies were performed according to

Microenvironment and Immunology

Selective Reversible Inhibition of Autophagy inHypoxic Breast Cancer Cells Promotes PulmonaryMetastasisChristopher M. Dower, Neema Bhat, Edward W.Wang, and Hong-Gang Wang

Abstract

Autophagy influences how cancer cells respond to nutrientdeprivation and hypoxic stress, two hallmarks of the tumormicroenvironment (TME). In this study, we explored the impactof autophagy on the pathophysiology of breast cancer cells usinga novel hypoxia-dependent, reversible dominant-negative strat-egy to regulate autophagy at the cellular level within the TME.Suppression of autophagy via hypoxia-induced expression of thekinase-dead unc-51-like autophagy-activating kinase (ULK1)mutant K46N increased lung metastases in MDA-MB-231 xeno-graft mouse models. Consistent with this effect, expressing a

dominant-negative mutant of ULK1 or ATG4b or a ULK1-target-ing shRNA facilitated cell migration in vitro. Functional proteomicand transcriptome analysis revealed that loss of hypoxia-regulatedautophagy promotes metastasis via induction of the fibronectinintegrin signaling axis. Indeed, loss of ULK1 function increasedfibronectin deposition in the hypoxic TME. Together, our resultsindicated that hypoxia-regulated autophagy suppresses metasta-sis in breast cancer by preventing tumor fibrosis. These results alsosuggest cautions in the development of autophagy-based strate-gies for cancer treatment. Cancer Res; 77(3); 646–57. �2016 AACR.

IntroductionThe tumor microenvironment (TME) is frequently subject to

fluctuating oxygen concentrations due to varying access to vas-culature. Tumor cells within the hypoxic TME respond to oxygendeficiency through global gene expression alterations, mediatedprimarily by the transcription factor, hypoxia-inducible factor-1(HIF1). The effects of hypoxia, and the corresponding increase inHIF1a expression, on cancer progression have been well studied(1). In a clinical context, hypoxia within the TME is associatedwith poor prognosis due to increased angiogenesis, chemoresis-tance, and metastatic potential (2, 3).

Similarly, it has become clear that autophagy plays paradoxicalroles in tumorigenesis. In general, autophagy provides a survivalmechanism that gives cancer cells metabolic flexibility, allowingfor their survival in nutrient and oxygen poor TMEs (4, 5). Here,autophagy catabolizes malfunctioning proteins and organelles,yielding nutrients that can be recycled and reallocated to essentialcellular pathways, allowing for tumor progression. Conversely,autophagy also removes reactive oxygen species, which preventsgenomic instability and cancer-driving mutations (6). Indeed,autophagy has been shown to be activated in response to hypoxiaand ischemia, and may be a driving mechanism behind HIF1-associated survival and chemoresistance (7, 8).

A growing body of research has begun exploring the effects ofhypoxia-regulated autophagy on tumor progression in efforts tobetter understand the biological dynamics associated with thehypoxic TME. Several of these studies have suggested hypoxia-induced autophagy increases tumor cell resistance to chemother-apy and promotes cell survival (9–12). However, these investiga-tions are dependent on in vitro model systems, in which cellsundergo an "all-or-none" hypoxia treatment that does not accu-rately represent the dynamic and heterogeneous nature of thehypoxic TME. Similarly, due to the context-dependent nature ofautophagy, traditional autophagy–inhibitory strategies that donot incorporate TME dynamics may not be ideal for examiningthe role of autophagy within the TME. Taken together, a morephysiologically accurate assessment, which incorporates the con-textual nature of hypoxia-regulated autophagy on tumor biology,has remained elusive.

Accordingly, we investigated the role of autophagy withinthe hypoxic TME of a triple-negative breast cancer model usinga novel hypoxia-dependent reversible dominant-negativeULK1 (dnULK1) gene expression strategy to physiologicallymodulate autophagy at single-cell levels. ULK1 (Unc-51 likekinase 1) is the only autophagy gene to encode a serine/threonine kinase that forms a complex with multiple regula-tory subunits, comprising of ATG13 and FIP200. Importantly,ULK1 has been well studied in the context of nutrient andoxygen deprivation, in which the mTORC1 negatively regu-lates ULK1 through hyperphosphorylation in nutrient-richenvironments; while AMP-activated protein kinase (AMPK)binds and phosphorylates ULK1 to activate autophagy underhypoxic and nutrient deprivation (13, 14). Thus, ULK1 plays acritical upstream role in regulating autophagy in response tohypoxic conditions, and provides an excellent target for inves-tigating hypoxia-regulated autophagy. Here, we demonstratethat the inhibition of autophagy selectively in hypoxictumor cells through HIF1a-dependent expression of dnULK1

Department of Pediatrics, Milton Hershey Medical Center, The PennsylvaniaState University College of Medicine, Hershey, Pennsylvania.

Note: Supplementary data for this article are available at Cancer ResearchOnline (http://cancerres.aacrjournals.org/).

Corresponding Author: Hong-Gang Wang, The Pennsylvania State UniversityCollege of Medicine, 500 University Drive, Hershey, PA 17033. Phone: 717-531-4574; Fax: 717-531-4789; E-mail: [email protected]

doi: 10.1158/0008-5472.CAN-15-3458

�2016 American Association for Cancer Research.

CancerResearch

Cancer Res; 77(3) February 1, 2017646

on November 24, 2020. © 2017 American Association for Cancer Research. cancerres.aacrjournals.org Downloaded from

Published OnlineFirst November 15, 2016; DOI: 10.1158/0008-5472.CAN-15-3458

http://cancerres.aacrjournals.org/

increases breast cancer metastasis to the lung, and provide acomprehensive overview of the underlying mechanisms driv-ing metastasis within the hypoxic TME.

Materials and MethodsCell lines

Human cancer cell lines MDA-MB-231 and HT-1080 werepurchased from ATCC, and ULK1�/� mouse embryonic fibro-blasts (MEF) were a kind gift from Dr. Mondira Kundu (St. JudeChildren's Research Hospital, Memphis, TN). All cell lines werepassaged in our laboratory for fewer than 6months before use andperiodically authenticated by morphologic inspection, growthcurve analysis, and mycoplasma testing.

Orthotopic xenograft mouse model of breast cancer usingMDA-MB-231 cells

All animal studies were performed according to the guidelinesestablished by the Institutional Animal Care and Use Committeeat the Penn State College of Medicine. An orthotopic xenograftbreast cancer model was generated by injecting 2.0 � 106 MDA-MB-231 cells into the inguinal mammary fat pad of femaleNOD SCID Gamma (NSG) mice, aged 6–8 weeks. Cells wereinjected in a 50:50 mixture of PBS and Matrigel basement mem-brane matrix (Fisher Scientific; catalog # CB-40234). Mice wereimaged for luciferase expression on a weekly basis for 7 weeksvia a Xenogen IVIS bioluminescent imager. Mice were inject-ed with 5 mL/gram body weight of 30 mg/mL D-Luciferin(Gold Biotechnology, catalog # LUCK-1G) in PBS, 5 minutesprior to imaging. Photon flux was calculated using region ofinterest (ROI) measurements of either the primary tumor site,or the ventral thoracic area for lung metastasis. Tumor volumewas alsomeasured using calipers and calculated as pLW2/6. At theexperiment endpoint,micewere euthanized and tumor, lung, andliver tissue were harvested for ex vivo analysis and subsequenthistology.

In vivo tail vein metastasis assayLung seeding assays were performed by injecting 5.0 � 105

MDA-MB-231 cells into the tail vein of 6- to 8-week-old femaleNSG mice. Mice were imaged on the ventral side for luciferaseexpression on a weekly basis for 7 weeks via Xenogen IVISimaging, the same way described above. At the experiment end-point, mice were euthanized and tumor, lung, and liver tissuewere harvested for ex vivo analysis and subsequent histologicanalysis.

ImmunoblottingCells were cultured in normal culturemediumunder normoxia

(21%O2) or hypoxia (1%O2) conditions for the indicated times.Cells were subsequently lysed in RIPA lysis buffer containingprotease inhibitors and subjected to immunoblotting with pri-mary antibodies: Flag (Sigma, catalog # F1804_200UG); p62(American Research Products, catalog # 03-GP62-C); HIF1 (CellSignaling Technology, catalog # 3716S); b-actin (Sigma-Aldrich,catalog # A5441); mCherry (Abcam, catalog # ab125096); Fibro-nectin (Abcam, catalog # ab2413); Tom20 (Santa Cruz Biotech-nology, catalog # sc-11415); LOX (Abcam, catalog # ab174316);HSP60 (Cell Signaling Technology, catalog # 12165P), and detec-tion with a LI-COR Odyssey CLx Imager.

ImmunohistochemistryTissues from NSG mice were subjected to immunohistochem-

ical analysis as described previously (15). For detection of hypoxicareas, mice were treated with an intraperitoneal injection ofhypoxyprobe-1 (60 mg/kg Hypoxyprobe, catalog # HP1-100),one hour prior to euthanasia. Hypoxyprobe was subsequentlydetected using the antibody specific to pimonidazole adducts(Hypoxyprobe, catalog # HP1-100).

qPCR quantificationPeripheral blood fromorthotopic xenograftmousemodels was

takenbiweekly. CirculatingDNAwas isolated from100mLof eachplasma sample using the QIAamp Circulating Nucleic Acid Kit(Qiagen, catalog # 55114.). The amount of human-derived DNAin each sample was quantified via real-time PCR (QuantStudio12K Flex) using a human-specific FOXP2 gene primer/probe set(Life Technologies, catalog # 4400291). ULK1 gene expressionwas assessed in MDA-MB-231 cells using ULK1 gene primer/probe set (Life Technologies, catalog # 4331182).

Fluorescent microscopy and phalloidin-mediated actinlabeling

Afterfixing in 4%paraformaldehyde for 1minutes, followed bypermeabilizationwith 0.1%TritonX-100 for 5minutes, cellswerestainedwith primary antibody or Alexa Flour 488 phalloidin (LifeTechnologies, catalog # A12379) in PBS and 1% BSA for 20minutes at room temperature. Cells were mounted with ProLongGold antifade reagent with DAPI (Invitrogen, catalog # P36936)and visualized on an Olympus IX81 microscope. Lamellipodiawere quantified by counting the number of cells displayinglamellipodia within a field at�200 magnification, and averaging15 randomly selected fields.

Wound-healing assayConfluent cells were scratched with a sterile P1000 pipette tip to

generate a defined scratchwound. Detached cells were removed bywashingwithPBSprior to incubation in competemediumineithernormoxia or hypoxia. Images were taken using a digital cameramounted on an Olympus CKX41 microscope every 24 hours atfixed locations on the plate and compared with the startingposition. The total distance was quantified using ImageJ software.

Boyden chamber invasion assayBoyden chambers (Corning BioCoat Matrigel Invasion Cham-

bers, catalog # 354480) were prepared according to the manu-facturer's instructions. MDA-MB-231 cells were placed in eithernormoxia (21% O2) or hypoxia (1% O2) for 48 hours andsubsequently fixed in 10% formaldehyde and stained with0.05% crystal violet. The amount of invaded cells was quantifiedusing ImageJ software.

Functional proteomic analysis via reverse-phase protein arrayThe expression levels of 298 proteins were assessed by

reverse-phase protein array (RPPA) through the MD Andersoncancer center RPPA core facility. Proteins significantly (P < 0.05)enriched in the HRE-dnULK1 tumors compared with controltumors were submitted to Ingenuity Pathway Analysis (IPA)software (QIAGEN; https://www.qiagenbioinformatics.com/products/ingenuity-pathway-analysis/) to generate network-

Hypoxia-Regulated Autophagy in Cancer Metastasis

www.aacrjournals.org Cancer Res; 77(3) February 1, 2017 647




module gene-set analysis to visualize previously described rela-tionships among the proteins.

RNA isolation, library preparation, and sequencingTotal RNA from cells was isolated with mirVana miRNA iso-

lation lysis buffer (Ambion, catalog # 8540G21). RNA integrationnumber (RIN) was measured using an Agilent 2100 BioanalyzerRNA 6000 Nano Kit to confirm RIN above 7. cDNA libraries wereprepared using the NEXTflex Rapid RNA Sequencing kit (BioOScientific) as per the manufacturer's instructions. The denaturedlibraries were diluted to 10 pmol/L by prechilled hybridizationbuffer and loaded onto a TruSEQ Rapid flow cell on an IlluminaHiSeq 2500 and run for 50 cycles using a single-read recipeaccording to the manufacturer's instructions.

mRNA sequencing data processing and pathway analysisDemultiplexed sequencing reads passing the default purify

filtering of Illumina CASAVA pipeline (released version 1.8) werequality trimmed/filtered using the FASTX-Toolkit (http://han-nonlab.cshl.edu/fastx_toolkit) by having a quality score cutoffof 20. Filtered reads were aligned to human reference genome(hg38) using Tophate (v2.0.9) by allowing up to 2 mismatches(16). FPKM values were calculated using Cufflinks v2.0.2 pro-vided with the Ensemble gene annotation (release 78; ref. 17).IPA (QIAGEN; www.qiagen.com/ingenuity) was used to iden-tify functionally relevant signaling pathways in differentiallyexpressed genes and proteins.

Statistical analysisData were analyzed using GraphPad Prism and statistical

software SASversion9.4 (SAS Institute). Student t test (two-tailed)was used for single comparisons. Group differences were evalu-ated using ANOVA or repeated-measure ANOVA models. Datawere considered statistically significant when P < 0.05.

ResultsHRE-dnULK1 provides hypoxia-dependent suppression ofautophagy

To investigate the role of hypoxia-regulated autophagy withinthe TME, we generated a plasmid vector utilizing the hypoxiaresponse element (HRE) promoter to regulate expression of adominant-negative ULK1(K46N) gene (dnULK1), to allow forhypoxia-dependent inhibition of ULK1 (Fig. 1A). A constructcontaining the HRE promoter, but lacking dnULK1, was alsoincluded as a negative control. With this, naturally occurringhypoxic regions generated by tumor growth will activate theHRE-dnULK1 construct, allowing for physiologic inhibition ofthe essential autophagy kinase, ULK1. Likewise, if a hypoxictumor cell regains access to oxygen, the construct will be turnedoff, allowing endogenousULK1 to reactivate to restore autophagicfunction (Supplementary Fig. S1A). To validate this, upon trans-ductionof theHRE-dnULK1 construct into a triple-negative breastcancer cell line, MDA-MB-231; the expression of dnULK1 andmCherry were verified to be dependent upon hypoxic exposure,and correlate with HIF1a expression (Fig. 1A and B; Supplemen-tary Fig. S1B). Hypoxia-induced autophagy was monitoredthrough several established autophagy markers, namely LC3lipidation and p62 accumulation (18). The HRE-dnULK1 con-struct was verified to inhibit autophagy under hypoxic conditions,as evident by a significant level of p62 protein accumulation

under hypoxic conditions, as compared with control cells (P <0.05; Fig. 1C–E). While LC3 lipidation was not blocked by HRE-dnULK1, an increase in both LC3-I and LC3-II suggest reducedautophagy flux. Furthermore, degradation of two mitochondrialmarkers, HSP60 and Tom20, were found to be blocked in HRE-dnULK1–expressing cells in hypoxic conditions, indicating asuppression of mitophagy (Fig. 1C). Together, these results indi-cate that HRE-dnULK1 provides a system that allows for hypoxia-inducible suppression of autophagy.

Suppression of ULK1 in the hypoxic TME increases metastasisof MDA-MB-231 cells to the lungs in xenograft mouse models

To assess the effects of hypoxia-regulated autophagy in aphysiologic setting, MDA-MB-231 breast cancer cells harboringeither the HRE-Empty or HRE-dnULK1 construct were injectedinto the inguinal mammary fat pad of NSG mice, generating anorthotopic xenograft mouse model of triple-negative breast can-cer. A luciferase reporter gene was also introduced into each celltype to allow for noninvasive monitoring of tumor growth andmetastasis. Measurements of growth of the primary tumor siterevealed no significant difference between the HRE-dnULK1 andHRE-Empty groups (Fig. 2A–C). Ex vivo analysis of xenografttissues confirmed that the primary tumor size remained similarbetween experimental groups (Fig. 2D). Importantly, primarytumors expressing the HRE-dnULK1 construct contained higherlevels of p62 accumulation as well as an increase in LC3-I,indicating inhibition of autophagy in vivo (Fig. 2E). Furthermore,the expression of dnULK1 within tissues strongly correlated tohypoxic areas within the primary tumors, demonstrating dnULK1expression is indeed hypoxia-dependent in vivo (Fig. 2F).

Although primary tumor growth remained similar between theHRE-Empty and HRE-dnULK1 groups, unexpectedly, metastasisfrom the primary tumor site to lung tissue was significantly higherinmice containing HRE-dnULK1 expressing tumors at the week 7time point (P < 0.05; Fig. 3A and B). A marked increase incirculating tumor DNA (ctDNA) levels was also detected at thispoint in HRE-dnULK1 mice (P < 0.0001; Fig. 3C), indicatinghigher levels of intravasation of tumor cells and/or circulatingtumor cells in theHRE-dnULK1 group.Gross tissue andhistologicanalysis confirmed an increase in metastasis to the lungs in theHRE-dnULK1 group (Fig. 3D and E). Interestingly, tumor growthwithin the lung generated a hypoxic TME,which, in turn, activatedthe HRE-dnULK1 construct within the lung tissue (Supplemen-tary Fig. S2A). A trend of increased metastasis of HRE-dnULK1tumors to the liver was also observed; however, quantification ofliver metastasis through histologic analysis did not reach statis-tical significance between the groups (Supplementary Fig. S2B).

Metastasis is a multistep process that involves: (i) intravasationof tumor cells into the blood stream or lymphatic circulation, (ii)survival of tumor cells in circulation, (iii) extravasation out of theblood stream, and (iv) engraftment and growth at the secondarysite. To further assess how hypoxia-regulated autophagy influ-ences metastasis, we performed an in vivo lung seeding assaythat allows the intravasation step of metastasis to be bypassed.Thus, the effects of hypoxia-regulated autophagy on survival inthe blood stream and engraftment into the lungs can be assessed.Here, HRE-Empty and HRE-dnULK1MDA-MB-231 cells culturedin normoxic conditions prior to tail vein injection yielded similarlevels of lung metastasis, as detected by luciferase expression inthe thoracic cavity (Fig. 4A and B). However, preincubation of thetwo respective cell lines in hypoxic conditions for 48 hours prior

Dower et al.

Cancer Res; 77(3) February 1, 2017 Cancer Research648




to injection resulted in a significant increase in lung engraftmentin the HRE-dnULK1–expressing group (P < 0.01; Fig. 4A and B).This verifies that the hypoxia-activated dnULK1 gene is indeedcausing the metastatic phenotype, and that the loss of autophagyunder hypoxic conditions increases engraftment into the lungs.Histologic analysis of lung tissue from these groups verifiedthis finding, as the hypoxia-incubated HRE-dnULK1 groupexhibited higher levels of tumor engraftment at significant levels(P < 0.001; Fig. 4C and D).

Inhibition of hypoxia-mediated autophagy increasesmigrationand invasion in vitro

To corroborate the increased metastasis exhibited by MDA-MB-231 cells expressing the HRE-dnULK1 construct in vivo, weaimed to replicate this phenotype in vitro. Accordingly, migra-tion and invasion of MDA-MB-231 cells were assessed using

wound-healing and invasion assays, respectively. In both cases,incubation of HRE-dnULK1–expressing MDA-MB-231 cells inhypoxia increased migration and invasion compared with theHRE-Empty counterpart (Fig. 5A and B). This indicates that theloss of autophagy can increase the motility of MDA-MB-231cells, as well as their ability to invade, which may facilitateboth intravasation and extravasation. Furthermore, there wasa significant increase (P < 0.01) in lamellipodia formationobserved in HRE-dnULK1–expressing MDA-MB-231 cells inhypoxia, indicating these cells exhibit a morphology conduciveto migration and characteristic of mesenchymal migration andmetastasis (Fig. 5C; ref.19). As an alternative approach, shRNAtargeting ULK1 (shULK1) was used to confirm that the loss ofULK1 function indeed results in an increase in migration inboth hypoxic and normoxic conditions (Fig. 5D–F). Further-more, we targeted an additional autophagy gene, ATG4b, to

Figure 1.

Hypoxia-induced expression of dnULK1 suppresses autophagy in MDA-MB-231 cells in vitro. A, Schematic representation of plasmid vectors expressing thehypoxia-inducible dominant-negative ULK1 gene (HRE-dnULK1) and the hypoxia-inducible blank control (HRE-Empty). Each vector contains hypoxia-dependentmCherry expression via an internal ribosomal entry site (IRES). B, Immunoblot assessing hypoxia-inducible expression of dnULK1 and mCherry. C, Immunoblotof autophagy markers in HRE-dnULK1 MDA-MB-231 cells compared with HRE-Empty control at normoxia (N), hypoxia (1% O2) for 48 hours (H), andreoxygenation for 24 hours after hypoxia (H/R). D, Quantification of p62 accumulation from three independent experiments using Image Studio Digitssoftware (� , P < 0.05, t test). E, Immunofluorescent staining of p62 in MDA-MB-231 cells expressing HRE-Empty or HRE-dnULK1 in normoxia or hypoxia (1% O2)for 48 hours.






determine whether the increase in migration is a ULK1-depe-dent phenotype. The protease ATG4b lies downstream of ULK1and is required for processing pro-LC3 prior to lipidation, a keystep in autophagosome biogenesis. Here, expression of a dom-inant-negative ATG4b gene [mStrawberry-ATG4b(C74A)], wasverified to suppress autophagic flux as evident by increased p62levels (Supplementary Fig. S3A). Indeed, expression ofdnATG4b increased migration of MDA-MB-231 cells comparedwith empty control cells (P < 0.05; Supplementary Fig. S3B).Together, these results indicate the loss of autophagy in hypoxiapromotes migration and invasion of MDA-MB-231 cellsthrough mesenchymal-associated mobility.

To expand these findings into an additional cell line, weexamined the effects of suppressing hypoxia-induced autophagybyHRE-dnULK1on cellmigration inHT-1080 cells. Expression ofdnULK1 was verified to be hypoxia- and HIF1a-dependent, andprovided suppression of autophagy in hypoxia as evident by p62

accumulation compared with HRE-Empty control cells (Supple-mentary Fig. S3C). Migration of HRE-dnULK1–expressing HT-1080 cells was subsequently assessed in normoxia and hypoxiausing a wound-healing assay. As observed in the MDA-MB-231cell line, therewas no significant difference betweenHRE-dnULK1and HRE-Empty HT-1080 migration when incubated in nor-moxia. However, HRE-dnULK1–expressing cells migrated signif-icantly further than HRE-Empty cells over 24 hours in hypoxia(P < 0.05; Supplementary Fig. S3D). Thus, the finding thatsuppression of hypoxia-induced autophagy by HRE-dnULK1promotes migration is not limited to MDA-MB-231 cells.

Loss of ULK1 function in hypoxia increases fibronectindeposition and reprograms metastatic signaling networks

To identify the mechanisms underlying the metastatic pheno-type exhibited by HRE-dnULK1–expressing cells, we performedboth functional proteomic and transcriptome analysis to yield a

Figure 2.

Loss of ULK1 function in the hypoxic tumor cells does not affect primary tumor growth. A, Tumor volume of primary xenograft tumors over 7 weeks (mean � SD,n ¼ 15 mice per group). B, Quantification of primary tumor luciferase expression photon flux (photons/sec) in HRE-dnULK1 and HRE-Empty xenografts(mean� SEM, n¼ 15mice per group).C,Representative images of luciferase expression fromprimary xenograft tumors over a 7-week time period.D,Representativeimages of tumors from HRE-dnULK1 and HRE-Empty orthotopic xenograft mice and quantification of final weight of excised primary tumors (mean � SD,n ¼ 15; ns, no significance, t test). E, Immunoblot analysis of dnULK1 and p62 expression from three primary tumors from HRE-empty and HRE-dnULK1xenografts (SE, short exposure; LE, long exposure; � , p62 aggregate). F, Immunohistochemical staining of primary tumors from HRE-Empty and HRE-dnULK1xenografts at week 7. Hypoxyprobe detects hypoxic areas that correlate to dnULK1 expression.

Dower et al.





comprehensive overview of signaling pathways involved in can-cer metastasis. First, the expression of 298 proteins was assessedin primary tumors from MDA-MB-231 orthotopic xenograftsusing RPPA analysis (Fig. 6A). A total of 41 proteins displayedsignificant enrichment or suppression (P < 0.05) in HRE-dnULK1expressing tumors compared with the HRE-Empty control (Sup-plementary Fig. S4A–S4E). Of these, 21 proteins reached highersignificance thresholds (P < 0.01), signifying a high degree ofconfidence for their involvement in metastasis from suppres-sion of hypoxia-regulated autophagy (Fig. 6B). In particular,

fibronectin displayed the greatest enrichment with an over 2-foldincrease in protein expression across 6 HRE-dnULK1 tumorscompared with 6 HRE-empty tumors. Enrichment in fibronectinwas verified in primary tumor tissues, and was also found to beenriched in the metastatic lung tissue of HRE-dnULK1 tumors,along with p62 (Fig. 6C and D). Importantly, a pronouncedincrease in fibronectin staining was found in the extracellularmatrix of HRE-dnULK1 xenograft tumors in areas correspondingto hypoxia-induced expression of dnULK1 (Fig. 6E). This in-crease in fibronectin in hypoxic areas was absent in HRE-Empty

Figure 3.

Loss of ULK1 function in the hypoxic TME increases metastases of MDA-MB-231 xenografts to the lungs. A, Quantification of metastasis of MDA-MB-231 xenograftsto the lungs via luciferase expression (photons/sec) in the thoracic cavity (mean � SEM; n ¼ 15; � , P < 0.05, t test). B, Representative images of luciferaseexpression in MDA-MB-231 xenograft mice within the thoracic cavity in weeks 5 through 7 of a 7-week time course. Primary tumor signal was shieldedto allow luciferase signal in the lung cavity tobemeasured.C,QuantitativePCRanalysis of circulatingMDA-MB-231 xenograft DNA (mean�SD;n¼3; �� ,P<0.0001,t test). D, Representative gross lung tissue and hematoxylin and eosin (H&E) stains from HRE-Empty and HRE-dnULK1 MDA-MB-231 xenografts at week 7illustrating the degree of metastasis to the lung. F, Quantification of lung metastasis as measured by the percent of tumor tissue within a lung section [mean� SD;n ¼ 5 (HRE-Empty); 10 (HRE-dnULK1); �� , P < 0.001, t test].






tumors, indicating hypoxia-induced dnULK1 promotes fibronec-tin deposition in vivo. A similar effect was found in vitro, wherehypoxic ULK1 knockdown MDA-MB-231 cells exhibitedincreased fibronectin deposition (Fig. 6F). Moreover, this effectwas very pronounced in fibroblast cells, as MEF cells lackingULK1 displayed dense networks of fibronectin between cellscompared with wild-type control (Fig. 6G). This increase infibronectin was less dramatic in both MDA-MB-231 and MEFcells incubated at normoxia, indicating hypoxia-regulated autop-hagy, in particular, is critical in preventing tumor fibrosis (Sup-plementary Fig. S3E and S3F).

Subsequently, input of significantly modified proteins (P <0.05) from RPPA analysis into pathway analysis revealed enrich-ment in integrin signaling and metastasis related pathways (i.e.,EMT signaling, FAK signaling, Actin Cytoskeleton signaling, TGFbsignaling, Rac signaling; Fig. 6H). A reduction in mTOR activityand protein synthesis was also observed, as well as an increase inDNAdamage and cell-cycle arrest pathways (p53 signaling, G2–MDNA damage checkpoint signaling, ATM signaling). To gainfurther insight, RNAseq was performed to evaluate global geneexpression changes in HRE-dnULK1 MDA-MB-231 cells in hyp-oxia compared with HRE-Empty MDA-MB-231 cells in hypoxia.Genes that were significantly upregulated anddownregulated (P <

0.05) were used to analyze pathway enrichment in HRE-Emptyand HRE-dnULK1 groups under hypoxia. Pathway analysisrevealed HRE-dnULK1–expressing cells exhibited enrichment inmany pathways known to be involved in cytoskeletal remodeling,migration, and metastasis. Specifically, we observed enrichmentin integrin signaling, paxillin signaling, regulation of EMT path-ways, Tec kinase signaling, IL8 signaling, VEGF signaling, ARP-WASP complex signaling, FGF signaling, FAK signaling, TGFbsignaling, PAK signaling, and CXCR4 signaling (Fig. 6I). Inaddition, we also observed strong enrichment in cholesterolbiosynthesis, which is known to promote metastasis in breastcancer, among others (20, 21). Genes showing significant enrich-ment (P < 0.05) that were major contributors to these pathwaysare presented (Supplementary Fig. S4F–S4H).

Finally, compounding candidate molecules from both prote-omic and transcriptomic analyses into a single network analysisindicated fibronectin is acting as a central node connecting manyof the genes found to have significant expression differences,suggesting fibronectin may be acting as a driver of metastasisthrough the activation of integrin signaling pathways (Supple-mentary Fig. S4I). In addition, previous reports have found thesecretion of LOX/LOXL4 is strongly implicated in breast cancermetastasis and premetastatic niche formation (22–24). Indeed,

Figure 4.

Loss of hypoxia-regulated autophagy increases engraftment of MDA-MB-231 cells into the lung. MDA-MB-231 cells expressing HRE-Empty or HRE-dnULK1were preincubated in normoxia (N) or hypoxia (H; 1% O2) for 48 hours and subjected to tail-vein injection into NSG mice (10 mice per group). A, Quantification ofmetastasis of MDA-MB-231 xenografts to the lungs via luciferase expression (photons/sec; mean � SEM; n ¼ 10; �� , P < 0.01; �� , P < 0.001, repeated-measureANOVA). B, Representative images of luciferase expression from MDA-MB-231 HRE-dnULK1 and HRE-Empty cell lines within the lungs of NSG mice overa 6-week time period. C, Quantification of lung metastasis as measured by the percent of tumor tissue within a lung section [mean � SD; n ¼ 8 per group(4 for HRE-dnULK1 at 1% O2)], �� , P < 0.001, t test. D, Representative images of hematoxylin and eosin (H&E)-stained lung tissues.

Dower et al.





these molecules were found to be significantly upregulated inboth the RNA expression in cell culture and the protein expres-sion in metastatic lung tissue in the HRE-dnULK1 group, and maybe key contributors to extravasation and engraftment into thepulmonary tissue (Supplementary Fig. S4G and Fig. 6D). Altogeth-er, these data indicate that the loss of autophagy within the hypoxicTME promotes fibronectin deposition in the TME and subsequentstimulation of integrin signaling and migratory pathways.

Reduced autophagy gene expression predicts poor prognosis inbreast cancer

As our results indicate that the loss of autophagy in the TMEmay promote breast cancer progression by increasing metastasis,we investigated whether autophagy gene expression correlateswith prognosis in humanbreast cancer patients using the Kaplan–Meier Plotter database (http://kmplot.com), which has the abilityto assess the effect of over 54,000 genes on the survival and

relapse-free survival (RFS) of 4,142 breast cancers (25). Weevaluated the potential predictive value of 5 autophagy genes(ULK1, ULK2, BECN1, ATG4b, and LC3C) involved at distinctsteps of autophagosomes biogenesis in the RFS of patients withbreast cancer. As shown in Fig. 7A–E, the patient group with lowexpression of the 5 genes had significantly poorer RFS. Similarly,thepatient groupwith high expressionof the autophagy substrate,p62, also showed a significantly reduced RFS (Fig. 7F). Theseresults indicate that the reduced autophagic activity in humanbreast cancer predicts worse prognosis, and thus supports ourfindings that the loss of autophagy within the hypoxic TMEpromotes breast cancer progression via metastasis.

DiscussionThe influence of hypoxia on gene expression and cancer pro-

gression has long been appreciated. Likewise, autophagy has

Figure 5.

Loss of autophagic function in hypoxia increases invasion andmigration in vitro.A,Wound-healing assay assessingmigration of MDA-MB-231 cells expressing eitherHRE-Empty or HRE-dnULK1 over 48 hours under hypoxic (1% O2) and normoxic conditions. Quantification was performed by measuring the distance betweenthe two leading edges using ImageJ software (mean � SD; n ¼ 3; ��, P < 0.01, t test). B, Transwell migration assay assessing invasion of MDA-MB-231 cellsexpressing either HRE-Empty or HRE-dnULK1 through a Matrigel matrix over 48 hours under hypoxic (1% O2) and normoxic conditions. Quantification wasperformed using ImageJ software (mean � SD; n ¼ 4; �� , P < 0.01, t test). C, Assessment of actin cytoskeletal dynamics by Alexa Fluor 488–conjugatedphalloidin staining of MDA-MB-231 cells expressing either HRE-Empty or HRE-dnULK1 incubated in either normoxia and hypoxia (1% O2; magnification,�200; blue,DAPI; red, mCherry; green, Alexa Fluor 488). Observed lamellipodia are indicated by white arrows. Lamellipodia were quantified by counting the numberof cells displaying lamellipodia within a field and averaging 15 randomly selected fields (mean � SD; �� , P < 0.01, t test). D, qPCR assessment of ULK1 geneknockdown by shULK1 in MDA-MB-231 cells (mean � SD; n ¼ 3; �� , P < 0.001, t test). E, Immunoblot assessing autophagy markers in shRNA MDA-MB-231cells versus HRE-Empty control. F, Wound-healing assay assessing migration of MDA-MB-231 cells expressing either HRE-Empty or shULK1 over 24 hours undernormoxic and hypoxic (1% O2) conditions. Quantification was performed by measuring the distance between the two leading edges using ImageJ software[mean � SD; n ¼ 3 (normoxia) and 6 (hypoxia); � , P < 0.05; ns, no significance, t test].






Figure 6.

Functional proteomic and transcriptomic assessment of hypoxia-regulated autophagy. A, Functional proteomic analysis by RPPA comparing expression levelsof 298 proteins in 6 HRE-dnULK1 primary xenograft tumors and 6 HRE-Empty primary xenograft tumors. B, Proteins displaying highly significant enrichment orsuppression (P < 0.01) in HRE-dnULk1 primary xenograft tumors compared with HRE-Empty primary xenograft tumors. C, Immunoblot analysis of primarytumor protein lysate fromHRE-dnULK1 andHRE-Empty orthotopic xenografts.D, Immunoblot analysis ofmetastatic lung tissue protein lysate fromHRE-dnULK1 andHRE-Empty orthotopic xenografts. E, Immunohistochemical staining of primary tumors from HRE-Empty and HRE-dnULK1 xenografts at week 7 withhematoxylin and eosin (i), hypoxyprobe (ii), Flag (Flag-dnULK1; iii), and fibronectin (iv). All images were taken at �40 magnification. F, Immunofluorescentstaining of fibronectin in the indicated MDA-MB-231 cells incubated in hypoxia (1% O2) for 48 hours. G, Immunofluorescent staining of fibronectin in MEFwild-type cells (MEF;ULK1þ/þ) andMEFULK1-knockout cells (MEF;ULK1�/�) incubated in hypoxia (1%O2) for 48 hours.H,Canonical pathway analysis of significantlyenriched or suppressed proteins (P < 0.05) in RPPA analysis. I, Canonical pathway analysis of significantly enriched or suppressed (P < 0.05) genes inHRE-dnULK1 MDA-MB-231 cells in hypoxia (1% O2) compared with HRE-Empty MDA-MB-231 cells in hypoxia (1% O2) as assessed through RNAseq analysis.

Dower et al.





emerged as a powerful cellular process that can greatly influencehow cancer cells respond to stress and the microenvironment as awhole. Here, we assessed the role of autophagy in the hypoxicTME. Specifically, we found that the suppression of hypoxia-induced autophagy increased the metastasis of MDA-MB-231cancer cells to the lung using two separate mouse modelapproaches. Interestingly, the primary tumor sizes remainedsimilar between HRE-Empty and HRE-dnULK1 xenografts, indi-cating that overall growth rate was not affected by the loss ofhypoxia-regulated autophagy. Although autophagy has been pre-viously found to effect cellular migration and invasion, thedefinitive role of autophagy in cell movement remains conten-tious, as autophagy can act as both a suppressor (26–30) and apromoter (31–34) of migration and metastasis. This is in agree-ment with a growing body of evidence displaying a context-dependent nature of autophagy, and suggests the influence ofautophagy on metastasis is dependent on multiple factors. Ourapproach that allows for physiologic regulation of autophagywithin the hypoxic TME provides evidence that hypoxia-induced

autophagy suppresses metastasis, and thus gives unique andvaluable insight into the physiologic role of autophagy withinthe TME.

To further understand the mechanisms of how inhibition ofhypoxia-mediated autophagy promotes metastasis, an assessmentof both protein and RNA expression was performed to provide acomprehensive overview of alterations in signaling pathways.Transcriptome analysis indicated that the loss of autophagy underhypoxia stimulates cellular movement pathways through expres-sion of many cytoskeleton remodeling proteins, and activation ofintegrin signaling and pathways directly associated with cellmigration. In vitro assessment of HRE-dnULK1 expressingMDA-MB-231 cells supported this finding, as an increase inmigration and invasion was observed, as well as an increase inlamellipodia formation, illustrating cytoskeleton remodelingpathways are indeed being activated. More importantly, ex vivoRPPA analysis of xenograft tumors yielded significant enrichmentin many of the same migratory pathways, indicating these path-ways are activated in both in vitro and in vivo systems.

Figure 7.

Kaplan–Meier (KM) survival curves relativeto autophagy gene expression for breastcancer. Kaplan–Meier RFS curves ofbreast cancer patients with highand low expression of autophagy-relatedgenes ULK1 (A), ULK2 (B), BECN1 (C),ATG4b (D), LC3C (E), and p62 (F) wereanalyzed with KM-plotter software(http://kmplot.com/analysis/).






Several proteins in particular displayed marked expressionchanges within the primary tumors, and may be acting as driversof metastasis. Fibronectin displayed significant enrichment with-in HRE-dnULK1 MDA-MB-231 primary tumors as well as in themetastatic lung tissue. Moreover, amarked increase in fibronectindeposition was found in the hypoxic TME of xenograft tumorswhen autophagy is suppressed. Notably, fibronectin expressionwas not found to be increased through RNAseq analysis, indicat-ing the observed increase in fibronectin levels within the TME isunlikely due to increased gene expression. Thus, the source offibronectin may be due to a reduction in autophagic degradationor an increase in secretion of fibronectin protein. Fibronectin is akey component of the extracellular matrix that is expressed byboth cancer-associated fibroblasts and cancer cells, and drivesmetastasis viamechanotransduction (35–37). Importantly, fibro-nectin engages integrin receptors, leading to the activation ofintracellular GTPases for cytoskeletal remodeling and cellmotility(35, 38). Supporting this, transcriptome and functional proteo-mic analysis indicated both integrin and several downstreampathway constituents were enriched in HRE-dnULK1 tumors andcell lines. These findings are in agreement with recent studiesreporting autophagy to be a regulator of fibrogenesis and fibro-nectin-mediated migration (26, 39–41).Furthermore, autophagyhas been shown to modulate cell migration through b1-integrinmembrane recycling, in which the loss of autophagy preventsintegrin sequestration within autophagosomes and subsequentdegradation in lysosomes (26). Thus, we hypothesize that the lossof hypoxia-inducible autophagy promotes metastasis throughextracellular remodeling and afibronectin–integrin signaling axis.

Importantly, the use of physiologic regulation of autophagyprovides an inherent induction of heterogeneity within the TME.That is, modulation of autophagy can only take place in areas ofsufficient hypoxia to activate the suppressing gene (i.e., dnULK1),while leaving autophagy in normoxic cells intact. Thus, thisgenerates a heterogeneous population of autophagy-competentand autophagy-deficient cells within a single tumor, which maycreate the opportunity for cooperation between the two popula-tions. This may allow for the highly stressed autophagy-deficienttumor cells to be supported by autophagy-competent tumor cellsduring the process of migration out of a nutrient poor environ-ment. Indeed, previous studies have shown tumor–stromal cellinteractions promote tumor progression through different levelsof autophagic activity (42, 43). However, autophagy-mediatedcell cooperativity has yet to be fully investigated between tumor

cells, and may elucidate key intercellular networks within theTME. Thus, we provide a promising physiologic model systemthat will allow for future research of tumor cell–tumor cellinteractions within the hypoxic TME.

In summary, we used a novel hypoxia-inducible dominant-negative strategy to demonstrate that the suppression of autop-hagy in the hypoxic TME promotes metastasis both in vitro andin vivo. Gene and protein expression analysis indicate fibronectinand downstream integrin signaling are key mediators of theobserved metastatic phenotype. This study provides key insightinto themechanistic interplay of hypoxia-induced autophagy andmetastasis within the TME.

Disclosure of Potential Conflicts of InterestNo potential conflicts of interest were disclosed.

Authors' ContributionsConception and design: C. Dower, H.-G. WangDevelopment of methodology: C. Dower, E.W. Wang, H.-G. WangAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): C. Dower, N. BhatAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): C. Dower, E.W. Wang, H.-G. WangWriting, review, and/or revision of the manuscript: C. Dower, H.-G. WangAdministrative, technical, or material support (i.e., reporting or organizingdata, constructing databases): C. DowerStudy supervision: H.-G. Wang

AcknowledgmentsWe thank Yuka Imamura (Penn State Hershey Genome Science Facilities) for

her technical help with the RNAseq analysis, as well as Sriranga (TJ) Iyyanki andJay Zhu for their help with large dataset analysis and statistics. We are gratefulto the Wang lab members who gave insight and technical help. We would liketo thank J.M. Brown (Stanford University, Stanford, CA) and Mondira Kundu(St. Jude Children's Research Hospital) for providing us the 5�HRE promoterand ULK1�/� MEFs, respectively.

Grant SupportThis work was supported by the NIH grant CA171501, and by the Lois High

Berstler Endowment Fund and the Four Diamonds Fund of the PennsylvaniaState University College of Medicine.

The costs of publication of this articlewere defrayed inpart by the payment ofpage charges. This article must therefore be hereby marked advertisement inaccordance with 18 U.S.C. Section 1734 solely to indicate this fact.

ReceivedDecember 23, 2015; revisedNovember 1, 2016; acceptedNovember2, 2016; published OnlineFirst November 15, 2016.

References1. Semenza GL. The hypoxic tumor microenvironment: a driving force

for breast cancer progression. Biochim Biophys Acta 2016;1863:382–91.2. Liu Z-J, Semenza GL, Zhang H-F. Hypoxia-inducible factor 1 and breast

cancer metastasis. J Zhejiang Univ Sci B 2015;16:32–43.3. Cosse J-P, Michiels C. Tumour hypoxia affects the responsiveness of cancer

cells to chemotherapy and promotes cancer progression. Anticancer AgentsMed Chem 2008;8:790–7.

4. Rosenfeldt MT, O'Prey J, Morton JP, Nixon C, MacKay G, Mrowinska A,et al. p53 status determines the role of autophagy in pancreatic tumourdevelopment. Nature 2013;504:296–300.

5. Sato K, Tsuchihara K, Fujii S, Sugiyama M, Goya T, Atomi Y, et al.Autophagy is activated in colorectal cancer cells and contributes to thetolerance to nutrient deprivation. Cancer Res 2007;67:9677–84.

6. Kongara S, Karantza V. The interplay between autophagy and ROS intumorigenesis. Front Oncol 2012;2:171.

7. Zhang H, Bosch-Marce M, Shimoda LA, Tan YS, Baek JH, Wesley JB, et al.Mitochondrial autophagy is an HIF-1-dependent adaptive metabolicresponse to hypoxia. J Biol Chem 2008;283:10892–903.

8. Mazure NM, Pouyss�egur J. Hypoxia-induced autophagy: cell death or cellsurvival? Curr Opin Cell Biol 2010;22:177–80.

9. He W-S, Dai X-F, Jin M, Liu C-W, Rent J-H. Hypoxia-induced autophagyconfers resistance of breast cancer cells to ionizing radiation. Oncol Res2012;20:251–8.

10. Hu Y-L, DeLay M, Jahangiri A, Molinaro AM, Rose SD, Carbonell WS, et al.Hypoxia-induced autophagy promotes tumor cell survival and adaptationto antiangiogenic treatment in glioblastoma. Cancer Res 2012;72:1773–83.

11. NomanMZ, Janji B, Kaminska B, VanMoerK, PiersonS, Przanowski P, et al.Blocking hypoxia-induced autophagy in tumors restores cytotoxic T-cellactivity and promotes regression. Cancer Res 2011;71:5976–86.


Dower et al.




12. Peng X, Gong F, Chen Y, Jiang Y, Liu J, Yu M, et al. Autophagy promotespaclitaxel resistance of cervical cancer cells: involvement of Warburg effectactivated hypoxia-induced factor 1-a-mediated signaling. Cell Death Dis2014;5:e1367.

13. Egan DF, Kim J, Shaw RJ, Guan K-L. The autophagy initiating kinase ULK1is regulated via opposing phosphorylation by AMPK and mTOR. Autop-hagy 2011;7:645–6.

14. Lee JW, Park S, Takahashi Y, Wang H-G. The association of AMPK withULK1 regulates autophagy. PLoS One 2010;5:e15394.

15. Takahashi Y, Hori T, Cooper TK, Liao J, Desai N, Serfass JM, et al. Bif-1haploinsufficiency promotes chromosomal instability and acceleratesMyc-driven lymphomagenesis via suppression of mitophagy. Blood2013;121:1622–32.

16. Trapnell C, Pachter L, Salzberg SL. TopHat: discovering splice junctionswith RNA-Seq. Bioinformatics 2009;25:1105–11.

17. Trapnell C, Williams BA, Pertea G, Mortazavi A, Kwan G, van Baren MJ,et al. Transcript assembly and quantification by RNA-Seq reveals unan-notated transcripts and isoform switching during cell differentiation. NatBiotechnol 2010;28:511–5.

18. Klionsky DJ, Abdelmohsen K, Abe A, Abedin MJ, Abeliovich H, AcevedoArozena A, et al. Guidelines for the use and interpretation of assays formonitoring autophagy (3rd edition). Autophagy 2016;12:1–222.

19. Yilmaz M, Christofori G. EMT, the cytoskeleton, and cancer cell invasion.Cancer Metastasis Rev 2009;28:15–33.

20. Antalis CJ, Uchida A, BuhmanKK, Siddiqui RA.Migration ofMDA-MB-231breast cancer cells depends on the availability of exogenous lipids andcholesterol esterification. Clin Exp Metastasis 2011;28:733–41.

21. Nelson ER, Chang C, McDonnell DP. Cholesterol and breast cancerpathophysiology. Trends Endocrinol Metab 2014;25:649–55.

22. Erler JT, BennewithKL,NicolauM,Dornh€oferN, KongC, LeQ-T, et al. Lysyloxidase is essential for hypoxia-induced metastasis. Nature 2006;440:1222–6.

23. Erler JT, Bennewith KL, Cox TR, Lang G, Bird D, Koong A, et al. Hypoxia-induced lysyl oxidase is a criticalmediator of bonemarrow cell recruitmentto form the premetastatic niche. Cancer Cell 2009;15:35–44.

24. Cox TR, Bird D, Baker A-M, Barker HE, Ho MW-Y, Lang G, et al. LOX-mediated collagen crosslinking is responsible for fibrosis-enhancedmetas-tasis. Cancer Res 2013;73:1721–32.

25. Gyo��rffy B, Surowiak P, Budczies J, L�anczky A. Online survivalanalysis software to assess the prognostic value of biomarkers usingtranscriptomic data in non-small-cell lung cancer. PLoS One 2013;8:e82241.

26. Tuloup-Minguez V, Hama€� A, Greffard A, Nicolas V, Codogno P, Botti J.Autophagy modulates cell migration and b1 integrin membrane recycling.Cell Cycle 2013;12:3317–28.

27. CatalanoM,D'AlessandroG, Lepore F,CorazzariM,Caldarola S, ValaccaC,et al. Autophagy induction impairs migration and invasion by reversingEMT in glioblastoma cells. Mol Oncol 2015;9:1612–25.

28. Chourasia AH, Tracy K, Frankenberger C, BolandML, SharifiMN,Drake LE,et al.Mitophagy defects arising fromBNip3 loss promotemammary tumorprogression to metastasis. EMBO Rep 2015;16:1145–63.

29. Indelicato M, Pucci B, Schito L, Reali V, Aventaggiato M, Mazzarino MC,et al. Role of hypoxia and autophagy in MDA-MB-231 invasiveness. J CellPhysiol 2010;223:359–68.

30. Yoshida T, Tsujioka M, Honda S, Tanaka M, Shimizu S. Autophagysuppresses cell migration by degrading GEF-H1, a RhoA GEF. Oncotarget2016;7:34420–9.

31. Sharifi MN, Mowers EE, Drake LE, Collier C, Chen H, Zamora M, et al.Autophagy promotes focal adhesion disassembly and cell motility ofmetastatic tumor cells through the direct interaction of paxillin with LC3.Cell Rep 2016;15:1660–72.

32. Nitta T, Sato Y, Ren XS, Harada K, Sasaki M,Hirano S, et al. Autophagymaypromote carcinoma cell invasion and correlate with poor prognosis incholangiocarcinoma. Int J Clin Exp Pathol 2014;7:4913–21.

33. Macintosh RL, Timpson P, Thorburn J, Anderson KI, Thorburn A, Ryan KM.Inhibition of autophagy impairs tumor cell invasion in an organotypicmodel. Cell Cycle 2012;11:2022–9.

34. Lock R, Kenific CM, Leidal AM, Salas E, Debnath J. Autophagy-dependentproduction of secreted factors facilitates oncogenic RAS-driven invasion.Cancer Discov 2014;4:466–79.

35. Topalovski M, Brekken RA. Matrix control of pancreatic cancer: Newinsights into fibronectin signaling. Cancer Lett 2016;381:252–8.

36. Park J, Schwarzbauer JE. Mammary epithelial cell interactions with fibro-nectin stimulate epithelial-mesenchymal transition. Oncogene 2014;33:1649–57.

37. von Au A, Vasel M, Kraft S, Sens C, Hackl N, Marx A, et al. Circulatingfibronectin controls tumor growth. Neoplasia 2013;15:925–38.

38. Levental KR, Yu H, Kass L, Lakins JN, Egeblad M, Erler JT, et al. Matrixcrosslinking forces tumor progression by enhancing integrin signaling. Cell2009;139:891–906.

39. Singh KK, Lovren F, Pan Y, Quan A, Ramadan A, Matkar PN, et al. Theessential autophagy gene ATG7 modulates organ fibrosis via regulation ofendothelial-to-mesenchymal transition. J Biol Chem 2015;290:2547–59.

40. Ghavami S, Cunnington RH, Gupta S, Yeganeh B, Filomeno KL, Freed DH,et al. Autophagy is a regulator of TGF-b1-induced fibrogenesis in primaryhuman atrial myofibroblasts. Cell Death Dis 2015;6:e1696.

41. LiW, Zou J, Yue F, Song K, ChenQ,McKeehanWL, et al. Defects inMAP1S-mediated autophagy cause reduction in mouse lifespans especially whenfibronectin is overexpressed. Aging Cell 2016;15:370–9.

42. Martinez-Outschoorn UE, Pavlides S, Howell A, Pestell RG, Tanowitz HB,Sotgia F, et al. Stromal–epithelial metabolic coupling in cancer: Integratingautophagy andmetabolism in the tumormicroenvironment. Int J BiochemCell Biol 2011;43:1045–51.

43. Martinez-Outschoorn U, Sotgia F, Lisanti MP. Tumor microenvironmentand metabolic synergy in breast cancers: critical importance of mitochon-drial fuels and function. Semin Oncol 2014;41:195–216.






2017;77:646-657. Published OnlineFirst November 15, 2016.Cancer Res Christopher M. Dower, Neema Bhat, Edward W. Wang, et al. Cancer Cells Promotes Pulmonary MetastasisSelective Reversible Inhibition of Autophagy in Hypoxic Breast

Updated version

10.1158/0008-5472.CAN-15-3458doi:

Access the most recent version of this article at:

Material

Supplementary

http://cancerres.aacrjournals.org/content/suppl/2016/11/15/0008-5472.CAN-15-3458.DC1

Access the most recent supplemental material at:

Cited articles

http://cancerres.aacrjournals.org/content/77/3/646.full#ref-list-1

This article cites 43 articles, 9 of which you can access for free at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected]

To order reprints of this article or to subscribe to the journal, contact the AACR Publications Department at

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/77/3/646To request permission to re-use all or part of this article, use this link



http://cancerres.aacrjournals.org/lookup/doi/10.1158/0008-5472.CAN-15-3458

http://cancerres.aacrjournals.org/content/suppl/2016/11/15/0008-5472.CAN-15-3458.DC1

http://cancerres.aacrjournals.org/content/77/3/646.full#ref-list-1

http://cancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]

http://cancerres.aacrjournals.org/content/77/3/646


Documents

Selective Reversible Inhibition of ... - Cancer Research · Orthotopic xenograft mouse model of breast cancer using MDA-MB-231 cells All animal studies were performed according to