USP11 Enhances TGFb-Induced Epithelial Mesenchymal Plasticity … · Mesenchymal Plasticity and Human Breast Cancer Metastasis Daniel A. Garcia1,2, Christina Baek1, M.Valeria Estrada3,Tiffani

Signal Transduction

USP11 Enhances TGFb-Induced Epithelial–Mesenchymal Plasticity and Human BreastCancer MetastasisDaniel A. Garcia1,2, Christina Baek1, M. Valeria Estrada3, Tiffani Tysl1,Eric J. Bennett4, Jing Yang5,6,7, and John T. Chang1

Abstract

Epithelial–mesenchymal transition (EMT) is a conservedcellular plasticity program that is reactivated in carcinomacells and drives metastasis. Although EMT is well studied itsregulatory mechanisms remain unclear. Therefore, to identifynovel regulators of EMT, a data mining approach was takenusing published microarray data and a group of deubiqui-tinases (DUB) were found to be upregulated in cells thathave undergone EMT. Here, it is demonstrated that oneDUB, ubiquitin-specific peptidase 11 (USP11), enhancesTGFb-induced EMT and self-renewal in immortalized humanmammary epithelial cells. Furthermore, modulating USP11expression in human breast cancer cells altered the migratorycapacity in vitro and metastasis in vivo. Moreover, elevated

USP11 expression in human breast cancer patient clinicalspecimens correlated with decreased survival. Mechanistical-ly, modulating USP11 expression altered the stability ofTGFb receptor type II (TGFBR2) and TGFb downstreamsignaling in human breast cancer cells. Together, these datasuggest that deubiquitination of TGFBR2 by USP11 effec-tively spares TGFBR2 from proteasomal degradation to pro-mote EMT and metastasis.

Implications: USP11 regulates TGFb-induced epithelial–mesenchymal plasticity and human breast cancer metastasisand may be a potential therapeutic target for breast cancer.Mol Cancer Res; 16(7); 1172–84. �2018 AACR.

IntroductionMetastasis is the major cause of cancer-related deaths; despite

intense investigation, it continues to be a poorly understoodaspect of cancer progression (1). Although deaths due to metas-tasis are likely to be abrogated by increased efforts in earlydetection of cancer, there is a need for more effective metasta-sis-specific therapies in the clinic (2).

Metastasis is fueled by disseminating tumor cells following acomplex cascade of events starting with translocation of primarytumor cells to a distant site and culminating with the develop-ment into a lethal outgrowth in distant organs. Two well-studieddrivers of metastasis are the epithelial–mesenchymal transition(EMT) and microenvironment-derived TGFb. The EMT is a cel-

lular developmental program that becomes reactivated duringcancer progression. Previouswork has demonstrated that the EMTendows neoplastic epithelial cells with mesenchymal and stem-like characteristics needed for subsequent steps in the metastaticcascade (3).

TGFb is a cytokine that is known to promote cancer metastasisthrough multiple mechanisms, including enhancing invasiveproperties, stemness, and permissive effects on the tumor micro-environment (4, 5). Interestingly, TGFb is a potent inducer ofEMT; thus,metastasis relies, in part, on TGFb's ability to induce anEMT program (6). The link between TGFb and EMT is mechanis-tically clear: upon TGFb type I and type II receptor (TGFBR1 andTGFBR2) heterocomplex activation, SMAD2/3 are phosphorylat-ed and shuttled to the nucleus with SMAD4 to regulate expressionof EMT transcription factors (7). However, what remain unclearare the ubiquitin-proteasome regulatory mechanisms of TGFbsignaling in the context of EMT and metastasis. Importantly, adeeper understanding of thismodeof regulationmay reveal novelenzymes that can be subject to rational and specific drug design.

There is increasing support for the importance of the ubi-quitin-proteasome system in regulating EMT and cancer pro-gression. Low proteasome activity is associated with the cancerstem cell state and EMT, and treatment of epithelial cellswith proteasome inhibitors can induce EMT (8, 9). In addition,a number of deubiquitinases (DUB) have been implicatedin cancer progression and metastasis, such as DUB3, USP4,USP13, USP15, USP28, and USP51 (10–15). Based on thiswork and the context-specific nature of EMT, we hypothesizethat there exists a set of proteins within the ubiquitin–protea-some pathway that regulate EMT specifically in normal andneoplastic human mammary epithelial cells.

1Department of Medicine, University of California San Diego, La Jolla, California.2Biomedical Sciences Graduate Program, University of California San Diego,La Jolla, California. 3Biorepository and Tissue Technology Shared Resources,Moores Cancer Center, University of California San Diego, La Jolla, California.4Division of Biological Sciences, Section of Cell and Developmental Biology,University of California, La Jolla, California. 5Department of Pharmacology,University of California SanDiego, La Jolla, California. 6Department of Pediatrics,University of California San Diego, La Jolla, California. 7Moores Cancer Center,University of California San Diego, La Jolla, California.

Note: Supplementary data for this article are available at Molecular CancerResearch Online (http://mcr.aacrjournals.org/).

Corresponding Author: John T. Chang, University of California San Diego, 9500Gilman Drive MC0063, La Jolla, CA 92093. Phone: 858-822-5795; E-mail:[email protected]

doi: 10.1158/1541-7786.MCR-17-0723

�2018 American Association for Cancer Research.

MolecularCancerResearch

Mol Cancer Res; 16(7) July 20181172

on July 11, 2020. © 2018 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from

Published OnlineFirst May 3, 2018; DOI: 10.1158/1541-7786.MCR-17-0723

http://crossmark.crossref.org/dialog/?doi=10.1158/1541-7786.MCR-17-0723&domain=pdf&date_stamp=2018-6-16

http://mcr.aacrjournals.org/

In this study, we show that specific DUBs are upregulatedduring EMT, and that ubiquitin-specific peptidase 11 (USP11)regulates TGFb-induced EMT in immortalized human mammaryepithelial cells. Further, we demonstrate that USP11 regulateshuman breast cancer cell behavior, including migration, self-renewal, and experimental metastasis. Finally, we show thatUSP11 exerts its effects on human breast cancer cell behavior bystabilizing the TGFb type II receptor (TGFBR2) and enhancingTGFb signaling. Although this relationship has been previouslyshown in lung fibroblast cells (16), we demonstrate the relevanceof this pathway in human breast cancer in vitro and in vivomodelsimplicating USP11 as a useful therapeutic target in the clinic.

Materials and MethodsAnimal studies

Animal studies were conducted using procedures approvedby the IACUC at the University of California, San Diego, La Jolla,CA (protocol #S09264). Studies were conducted in accordanceto the ARRIVE guidelines. NOD-scid IL2Rgammanull (NSG) micewere obtained from The Jackson Laboratory and UCSD AnimalCare Program. For the tail vein metastasis assay, SUM159 orMDA-MB-231 cells were resuspended in PBS and injected intothe tail vein of 6- to 8-week-old female NSG mice. A total of1�106 to1.5�106 cellswere injected in a volumeof 100mL.Micewere sacrificed after 6 to 8 weeks and lung metastases werequantified. Tumors located within vascular spaces in the lungwere excluded from analysis.

Metastasis quantificationLungs were perfused with PBS and then removed from the

thoracic cavity. The lung lobes were fixed in Bouin's solution for6hours, and tissuewas processed for sectioning andH&E staining.H&E step sections were analyzed by a pathologist (M.V. Estrada)at the Tissue Technology Shared Resource (Moores Cancer Center,University of California, San Diego, CA). Tumor burden wasassessed by whole section tumor cellularity.

Cell cultureAll cell cultures were maintained at 37�C with 5% CO2.

Human mammary epithelial cell lines (HMLE) were culturedin MEGM (Lonza). shRNA and overexpression constructs wereretrovirally transduced into HMLE cells. HEK293T, T47D, andMDA-MB-231 cells were cultured in DMEM with 10% FBS.SUM159 cells were cultured in Ham's F12 supplemented with10 mmol/L HEPES, 5% FBS, 5 mg/mL insulin, and 1 mg/mLhydrocortisone. T47D cells were obtained from Li Ma (MDAnderson Cancer Center, University of Texas, Houston, TX). Allcell lines were authenticated by short tandem repeat (STR)analysis at ATCC. Because it was not present in any STR data-base as a basis of comparison, the HMLE cell line was authen-ticated on the basis of morphology and epithelial marker expres-sion to the original cell line, as other studies have reported (17).In addition, all cell lines used in the manuscript tested negativefor mycoplasma using the service provided by the HumanEmbryonic Stem Cell Core Facility at UCSD. Cells used forexperiments were between 2 and 7 passages from thawing.

VectorsThe following retroviral vectors were gifts from Wade Harper:

Flag-HA-GFP (Addgene #22612), Flag-HA-USP13 (Addgene

#22568), Flag-HA-UCHL1 (Addgene #22563; ref. 18). Full-lengthhuman USP11 was amplified from a cDNA library and clonedinto a retroviral pDEST-Flag-HA vector. Catalytically inactiveUSP11 was generated by site-directed mutagenesis. shRNA hair-pin sequences targeting firefly luciferase or USP11 were clonedinto pINDUCER10 (miR-RUP; ref. 19). Stable expression ofDUBsand shRNAs was achieved by retroviral infection for 5 to 7 hoursand selection with 2 mg/mL puromycin 24 to 48 hours later.Retroviruses were produced with HEK293T cells as previouslydescribed (8). CAGA12-firefly luciferase reporter was a gift fromPeter ten Dijke (Leiden University Medical Center, Leiden, theNetherlands; ref. 14). pGL4.74-renilla luciferase construct wasobtained from Maryan Rizk (Guatelli Lab, University of Califor-nia, San Diego, CA). See Supplementary Data for shRNA hairpinsequences.

RNA extraction and RT-qPCRTotal RNA was extracted using Trizol (Thermo Fisher Scien-

tific) and was reverse-transcribed with High-Capacity cDNAReverse Transcription Kit (Thermo Fisher Scientific). The result-ing cDNAs were used for RT-qPCR using SsoAdvance SYBRGreen Supermix (Bio-Rad) in triplicate. RT-qPCR and datacollection were performed on CFX96 Touch Real-Time PCRDetection System (Bio-Rad). All the values were normalized toan internal control GAPDH. Relative expression for each targetgene was compared to that of cells expressing Ctrl or shCtrl.See Supplementary Data for primer oligonucleotide sequences.

Microarray and Kaplan–Meier analysisMicroarray data from GEO accession GSE24202 (20) were

analyzed with IPA (Qiagen). Briefly, 56 genes associated with theubiquitin–proteasome pathway were chosen from the list ofdifferentially expressed genes as determined by IPA. Venn diagramanalysis was performed using the list of 56 genes. Kaplan–Meiersurvival curves for overall survival were generated from The CancerGenome Atlas Breast Invasive Carcinoma gene expression dataset(n ¼ 1215). Patients were stratified based on the expression ofUSP11, USP13, or UCHL1. High and low expression were definedas the top and bottom �25% of the patient distribution. Expres-sion, overall survival, and patient tumor subtype data were down-loaded from the UCSC Cancer Genomics Browser (https://genome-cancer.ucsc.edu/). Prism (GraphPad) was used to testthe statistical significance between the overall survival curvesusing the Log-rank (Mantel–Cox) Test. Kaplan–Meier survivalcurves for overall survival and distant metastasis free survivalwere generated with the web tool KMplotter (21).

Flow cytometry analysisSuspensions of HMLE cell lines were stained with anti-CD44

antibody (BioLegend) and analyzed by flow cytometry on a BDAccuri C6 (BD Biosciences). Suspensions of MDA-MB-231 andSUM159 cell lines were fixed and permeabilized with thefixation/permeabilization concentrate kit (eBioscience), andstained with anti-TGFBR2 antibody (R&D). Data were analyzedusing FlowJo software. Cell sorting was done using a FACS Aria2 (BD Biosciences) at the UCSD Human Embryonic Stem CellCore Facility. For apoptosis analysis, suspensions of HMLE cellsfrom dissociated mammospheres were resuspended in AnnexinV Binding Buffer (BioLegend) and stained with Annexin V-APC(BioLegend) and 7-AAD (BioLegend) and analyzed on a BDAccuri C6.

TGFb-Induced EMT and Metastasis are Regulated by USP11

www.aacrjournals.org Mol Cancer Res; 16(7) July 2018 1173



https://genome-cancer.ucsc.edu/

https://genome-cancer.ucsc.edu/


Mammosphere assayHMLE mammosphere culture was performed by plating 1,000

to 5,000 cells in Mammocult medium (Stemcell Technologies)supplemented with 4 mg/mL heparin, 0.5 mg/mL hydrocortisone,and 1% methylcellulose. For serial passage into secondary mam-mosphere formation, mammospheres were dissociated into sin-gle cells by trypsinization and then plated in mammosphereculture conditions. Spheres were counted 10 to 12 days later.T47D spheres were formed by plating 1,000 to 4,000 cells inDMEM/F12 media supplemented with hEGF (20 ng/mL; Sigma),bFGF (20 ng/mL; Thermo Fisher Scientific), heparin (4 mg/mL;Sigma), 1% methylcellulose (R&D), and B27 Supplement(Thermo Fisher Scientific). All sphere assays were performed in24-well ultra-low-attachment plates (Corning).

Western blot analysisCells were lysed on ice in RIPA buffer supplemented with

Protease/Phosphatase Inhibitor Cocktail (Cell Signaling Techno-logy). Thirty micrograms of total protein from each samplewas resolved on Novex 4% to 20% Tris-Glycine Mini ProteinGels and transferred onto nitrocellulose membranes. Blots wereprobed with the appropriate antibodies: anti-USP11 (Bethyl),anti-USP13 (Bethyl), anti-phospho-SMAD2 (Cell SignalingTechnology), anti-total-SMAD2 (Cell Signaling Technology),anti-HA (eBioscience), anti-Fibronectin (eBioscience), anti-ZEB1(Cell Signaling Technology), anti-SNAI1 (Cell Signaling Techno-logy), or anti-b-actin (Sigma-Aldrich). Signals were detectedusing fluorescent secondary antibodies compatible with Odysseyinfrared imaging system (Li-Cor Biosciences). For Fig. 7C andSupplementary Fig. S6, all Western blots images are taken fromthe same membrane with the intervening irrelevant samplesremoved for clarity.

Cell migration assaysScratch wound assays were performed in six-well plates. Cells

were grown to �90% confluency and wound was created with apipette tip. Media was replenished with media containing TGFb(2.5 ng/mL). Images were taken in three separate locations alongthe scratch wound at 0 and 24 hours after wound was made.Extent of wound closure was analyzed with TScratch software(22). Transwell assays were performed in 24-well polycarbonateinserts (Falcon, 8 mm pore size). Cells were serum starved over-night, then plated in transwell inserts with serum-free media andcompletemedia in the bottom chamber. Cells in the upper part ofthe transwells were removed with a cotton swab; migrated cellswerefixed in 4%paraformaldehyde and stainedwith crystal violet0.5%. Three random fieldswere photographed and the number ofcells was counted with ImageJ and averaged. Every experimentwas repeated independently at least three times.

Cell proliferation analysisCell proliferationwasmeasured using the TetraZ Cell Counting

Kit (BioLegend) following manufacturer's instructions.

Dual luciferase reporter assaySUM159 cell lines were transfected with the TGFb signaling

reporter CAGA12-firefly luciferase and pGL4.74-renilla luciferase(Promega) constructs using FugeneHD (Promega). After 24 to 48hours, cells were treated with TGFb (0.25 ng/mL) for 6 hours.Luciferase activity was then measured using the Dual-Glo Lucif-erase Assay Kit (Promega).

Statistical analysisStatistical analyseswere performedwith an unpaired t test, one-

way ANOVA, two-way ANOVA, or log-rank (Mantel–Cox) test(for Kaplan–Meier survival curves) using GraphPad Software. Theresulting statistics are indicated in each figure as follows: ns, notsignificant (P > 0.05); �, P � 0.05; ��, P � 0.01; ��, P � 0.001.

ResultsDUB expression is associated with EMT and decreased survivalin human breast cancer patients

To identify novel regulators of EMT within the ubiquitin–proteasome pathway, we examined previously published micro-array data. Taube et al. compared the transcriptomes of immor-talized human mammary epithelial (HMLE) cells that haveundergone EMT via stable overexpression of EMT-inducing fac-tors (EMT-TFs) Twist, Snail, Gsc, or TGFb (20). We filtereddifferentially expressed genes to a curated list of genes within theubiquitin–proteasome pathway, including the 26S proteasomesubunits, ubiquitin ligases, and DUBs. Venn diagram analysisrevealed three DUBs—USP11, USP13, and UCHL1—that wereupregulated across all HMLE lines (Fig. 1A). We confirmed theupregulation ofDUBs at the protein level (Fig. 1B) inHMLE-TwistandHMLE-Snail compared to the parental line. To show thatDUBexpression was also differentially expressed without forced over-expression of EMT-TFs, parental HMLE cells were sorted based ontheir level of cell surface CD44, a previously established EMTmarker (23). In the �1% of cells that were CD44high within theparental HMLE cell line, DUB expression was markedly increasedcompared to the rest of the population (CD44low) (Fig. 1C).USP11, USP13, and UCHL1 also appear to be a part of the TGFb-induced EMT pathway, as their expression is upregulated uponTGFb treatment and EMT induction (Fig. 1D). These resultssuggest that DUB expression is associated with the EMT, whethercells are induced with TGFb or stable overexpression of EMT-TFs.

To establish a potential role for DUB expression in humandisease progression, we examined their expression in cohortsof breast cancer patients. Using TCGA Breast Cancer patientdataset, we found that patients with high expression of USP11and USP13, but not UCHL1, had decreased overall survival(OS) compared to patients with low expression (Fig. 1E). Usingthe web tool KM-plotter, we found that only patients with highexpression ofUSP11, notUSP13orUCHL1, exhibited a decreaseddistant metastasis-free survival (DMFS; Fig. 1F). These results arein line with previous work (24) and suggest that DUB expressionmay contribute to both disease progression and metastasis inhuman breast cancer patients.

To understand USP110s role in disease progression in a morespecific manner, we used TCGA and KM-plotter databases tocorrelateUSP11 expressionwith clinically relevant patient subsetsand tumor stages. In the TCGA Breast Cancer patient dataset, wefound that high USP11 expression correlated with decreased OSin ERþ patients and patients with a Luminal A tumor subtype, butnot other tumor subtypes (Supplementary Fig. S1A). Using theKM-plotter breast cancer database, we found that only patients awith Luminal B tumor subtype showed a correlation withdecreased OS and high USP11 expression (Supplementary Fig.S1B). When using DMFS as the clinical outcome, KM-plotteranalysis showed that high USP11 expression correlated withdecreased survival in patients with ERþ, Luminal A, and HER2þ

tumor subtypes (Supplementary Fig. S1B). These survival analyses

Garcia et al.

Mol Cancer Res; 16(7) July 2018 Molecular Cancer Research1174




Figure 1.

DUB expression is associated with EMT and decreased survival in human breast cancer patients. A, Venn diagram analysis of microarray data from GEO accessionGSE24202. Ubiquitin–proteasome system-associated genes were selected from the list of differentially expressed genes. B, Western blot analysis of DUBprotein levels in HMLE cells with or without retroviral overexpression of EMT transcription factors. C, RT-qPCR analysis of DUB mRNA levels in HMLE cells sortedbased on CD44 cell surface expression level. D, RT-qPCR analysis of DUB mRNA levels in HMLE cells treated or not with TGFb for 7 days. E, Kaplan–Meieranalysis of overall survival in relation to the level of DUB gene expression in TCGA breast invasive carcinoma cases. F, Kaplan–Meier analysis (KM-plotter) of distantmetastasis free survival in relation to DUB expression in patient tumor samples.






suggest that patients with tumors of epithelial nature (LuminalA/B, ERþ) aremore vulnerable to decreased survival and increasedmetastasis that is associated with high USP11 expression. Finally,we found that USP11 expression was not significantly enriched inany tumor stage, suggesting that high USP11 expression is likely arisk factor present at tumor initiation (Supplementary Fig. S1C).

USP11 and its catalytic activity are necessary for a completeacquisition of EMT characteristics in response to TGFb

To test whether DUBs play a role in regulating EMT, USP11,USP13, and UCHL1 were retrovirally overexpressed in HMLEcells (Supplementary Fig. S2A). We found that overexpression ofthese DUBs alone did not cause the cells to undergo EMT.However, when EMT was induced via treatment with TGFb for14 days, we observed that cells overexpressing USP11, but notUSP13 or UCHL1, exhibited a higher percentage of CD44high cells(Fig. 2A). Based on this result, we decided to focus on the roleUSP11 plays in regulating TGFb-induced EMT.

To determine if the effect of USP11 on EMT induction inHMLE cells was due to its deubiquitinating activity, we over-expressed eitherwild-typeUSP11 (USP11-wt) or aC318S catalyticmutant of USP11 (USP11-mut) in HMLE cells (SupplementaryFig. S2B). Unlike USP11-wt, expression of the C318S catalyticmutant failed to enhance EMT, based on CD44 cell surfaceexpression (Fig. 2B) andmRNA expression of EMTmarkersCDH1(E-cadherin),CDH2 (N-cadherin),VIM, ZEB1, TWIST1, and SNAI1(Fig. 2C). These results suggest that USP110s deubiquitinatingactivity is necessary to enhance TGFb-induced EMT.

We next stably expressed USP11 shRNA in HMLE cells (HMLE-shUSP11) to determine if USP11 is necessary for EMT induction(Supplementary Fig. S2C). After 14 days of TGFb treatment,HMLE-shUSP11 cells were unable to undergo EMT to the sameextent as cells expressing a nontargeting control shRNA (HMLE-shCtrl). HMLE-shUSP11 cells displayed a significantly lowerpercentage of CD44high events (Fig. 2D) and failed to upregulateEMT markers to the same extent at HMLE-shCtrl cells (Fig. 2E).Together, these data suggest that although USP11 alone does notinduce EMT, it is nonetheless required for TGFb-induced EMT andis dependent on its catalytic activity to do so.

Cells that have undergone EMT have been shown to possessstem-like characteristics, namely growth in suspension and self-renewal in the mammosphere formation assay (23). We inducedEMT with TGFb treatment for 14 days in the USP11 overexpres-sion and USP11 shRNA HMLE cell lines and then plated them inthe mammosphere formation assay (Supplementary Fig. S3A).We found that overexpression of USP11-wt, but not USP11-mut,increased thenumber of spheres formed compared to control cells(Supplementary Fig. S3B), although shRNAknockdownofUSP11reduced the number of spheres formed compared to shCtrl cells(Supplementary Fig. S3C). This result was expected since thepercentage of CD44high cells correlates with the extent of sphereformation, and that CD44low cells do not possessmammosphere-forming capacity (23). To determine the role of USP11 in self-renewal, we first sorted cells based on CD44 expression, and thenplated them in the sphere assay (Fig. 3A). Neither USP11 over-expression nor USP11 shRNA knockdown affected the number ofprimary spheres formed when only CD44high cells were platedsuggesting that USP11 does not alter the number of stem-like cellsin the CD44hi population (Fig. 3B and C). However, upon serialpassage, USP11-wt, but not USP11-mut cells formed more sec-ondary spheres when compared to control cells (Fig. 3B). Con-

versely, serial passage of shUSP11 spheres resulted in fewersecondary spheres compared to shCtrl (Fig. 3C). Notably, thedifferences observed in sphere formation are not due to apoptosis(Fig. 3D). These results suggest that USP11 and its deubiquitinat-ing activity are necessary for the self-renewal capacity of humanmammary epithelial cells. This is likely due to USP110s effect onthe stem-like characteristics of the daughter progenitor cellsformed in primary spheres that are then passaged on to thesecondary sphere formation experiment. It appears that an induc-tion of EMT also induces heritable self-renewal characteristics thatcontinues in subsequent passages of mammospheres, and theextent of this phenomenon is affected by USP11 expression,despite the lack of EMT stimulus (TGFb).

USP11 and its catalytic activity influence human breast cancercell migration and self-renewal in vitro

Because we found that USP11 is upregulated in a panel ofhuman breast cancer cell lines compared to normal mammaryepithelial cells (HMLE; Supplementary Fig. S4), we next sought todetermine the effects of modulating USP11 expression on thebehavior of human breast cancer cells. We chose two humanbreast cancer cell lines, MDA-MB-231 and SUM159, which exhib-ited the highest and lowest extent of USP11 expression, respec-tively (Supplementary Fig. S4).Weperformed shRNAknockdownexperiments in MDA-MB-231 cells and retroviral overexpressionexperiments in SUM159 cells and (Supplementary Fig. S2Dand S2E).

To determine the involvement ofUSP11on the in vitrobehaviorof human breast cancer cells, we performed scratch wound andtranswell cell migration assays. In a scratch wound closure assay,overexpression of USP11-wt, but not USP11-mut, in SUM159cells resulted in significantly fasterwound closurewhen comparedto control cells (Fig. 4A). USP11 depletion caused MDA-MB-231cells to close the wound significantly slower when compared tocontrol cells, as evidenced by a significantly larger open area 24hours after the scratch wound wasmade (Fig. 4B). The differencesseen in wound closure did not appear to be due to proliferativedifferences, because proliferation was not significantly differentacross cell lines (Fig. 4C). In a transwell migration assay, over-expression of USP11-wt, but not USP11-mut, caused moreSUM159 cells to migrate through the transwell membrane (Fig.4D). Conversely, shRNA knockdown of USP11 in SUM159 cellscaused fewer cells to migrate through the transwell membrane(Fig. 4D). These results suggest that deubiquitination by USP11regulates collective cell migration in response to a wound (scratchwound assay) and single cell migration in response to a chemoat-tractant (transwell assay).

In light of our observation that USP11 regulates self-renewal innormal mammary epithelial cells (Fig. 3), we next sought todetermine if USP11 also regulates self-renewal in human breastcancer cells. Neither USP11 overexpression nor shRNA knock-down affected primary sphere formation in T47D human breastcancer cells (Fig. 5A and D). However, upon serial passage,overexpression of USP11-wt, but not USP11-mut, increased thenumber of secondary spheres compared to control cells (Fig. 5A).Conversely, USP11 depletion in T47D cells decreased the numberof secondary spheres formed compared to shCtrl cells (Fig. 5D).Notably, sphere volume was significantly larger in USP11-wt cellscompared to control and USP11-mut cells at the secondarypassage, but not the primary passage (Fig. 5B and C). We alsoobserved smaller spheres at both primary and secondary sphere

Garcia et al.





passages in T47D cells lacking USP11 compared to shCtrl cells(Fig. 5E and F). Together, these results show that although USP11may not affect primary sphere formation, it does regulate sec-ondary sphere formation at subsequent passages suggesting a rolein self-renewal. Strikingly, USP11 also affects sphere size, suggest-ing a role in the proliferative capacity of progenitor cells.

The extent of metastasis of human breast cancer xenografts isdependent on USP11 expression

On the basis of the results from our in vitro experimentsdescribed above, we hypothesized that modulating USP11expression in human breast cancer cell lines would affect theirbehavior in vivo. To directly test this hypothesis, we used an

Figure 2.

USP11 and its catalytic activity are necessary for a complete TGFb-induced EMT. A–D, Representative flow cytometry analysis of CD44 expression in HMLE cellstreated or not with TGFb for 14 days. (A) Retroviral overexpression of DUBs compared to GFP expressing control (Ctrl); (B) retroviral overexpression ofwild type USP11 (USP11-wt) compared to C318S catalytic mutant USP11 (USP11-mut) and Ctrl; and (D) retroviral expression of shRNA targeting USP11 (shUSP11)compared to a nontargeting control shRNA (shCtrl). Three independent experiments are represented in the bar graphs. C and E,RT-qPCR analysis ofmRNA levels ofEMT markers in HMLE cells treated or not with TGFb for 14 days. (C) Retroviral overexpression of wild-type USP11 (USP11-wt) compared to C318S catalyticmutant USP11 (USP11-mut) and Ctrl, (E) retroviral expression of shRNA targeting USP11 compared to a nontargeting control shRNA (shCtrl). Three independentexperiments are represented in the bar graphs.






Figure 3.

USP11 regulates self-renewal in normal human epithelial cells.A,HMLE cellswere treated with TGFb for 14 days, and then stained for cell surface CD44. CD44high cellswere sorted by FACS and plated in the mammosphere assay. B and C, Primary mammosphere formation was quantified after 10 to 12 days. Primary sphereswere dissociated and replated in the secondarymammosphere assay. Secondary sphere formationwas quantified after 10 to 12 days. Three independent experimentsare represented in the bar graphs. Representative images of spheres are shown. D, Secondary mammospheres were dissociated and single cells werestained for apoptosis markers (Annexin V and 7-AAD) and analyzed by flow cytometry. Live cells (Annexin V�/7-AAD�, lower left quadrant), early apoptotic cells(Annexin Vþ/7-AAD�, upper left quadrant), and late apoptotic cells (Annexin Vþ/7-AADþ, upper right quadrant) were quantified as a percentage of total cells.Two independent experiments are represented in the bar graphs.

Garcia et al.





experimental tail vein metastasis assay. Indeed, we found thatSUM159 cells overexpressing USP11-wt, but not USP11-mut,colonized lung tissue as metastases 10-fold more than controlcells (Fig. 6A). Alternatively, when USP11 expression wasknocked down with stable shRNA expression, MDA-MB-231cells colonized lung tissue approximately 40% less than controlcells (Fig. 6B). These results suggest that the effect of USP11expression on in vitro behaviors of cancer cell lines can berecapitulated with a more stringent in vivo test of metastaticcapacity.

USP11 regulates the stability of TGFBR2 anddownstreamTGFbsignaling in human breast cancer cells

Given the known relationship between TGFb receptors andDUBs (14, 16, 25, 26), and that the USP11-dependent regulationof transwell migration we observed is TGFb-dependent (Supple-mentary Fig. S5), we hypothesized that USP11was affecting TGFbreceptor stability in human breast cancer cells. To determine themechanism of USP110s effects on cancer cell behavior, we mea-sured the level of TGFb type II receptor (TGFBR2) using flowcytometry. Overexpression of USP11-wt, but not USP11-mut,

Figure 4.

USP11 regulates human breast cancer cell migration, but not proliferation. A and B, Scratch wound away was performed with (A) SUM159 cells overexpressing GFP(Ctrl), USP11-wt, or USP11-mut and (B) MDA-MB-231 cells expressing shCtrl or shUSP11, in media containing TGFb. Percent open area was quantified bymeasuring the surface area of the scratch wound at 0 and 24 hours after scratch wound was made. Shown are representative images of the scratch wound 24 hoursafter scratch was made. Light gray area indicates area covered by cells, while dark gray area indicates the scratch wound. Three independent experimentsare represented in thebargraph.C,Cell proliferationassaywith SUM159andMDA-MB-231 cells.D,Transwellmigration assay in SUM159 cells. For eachexperiment,fiverandom microscope fields were photographed and the number of cells per field was quantified using ImageJ. Two independent experiments are representedin the bar graphs.






caused a 50% increase in TGFBR2 levels in SUM159 cells main-tained in normal culture conditions (Fig. 7A). Conversely, USP11depletion caused a 25% decrease in TGFBR2 levels in MDA-MB-231 cells maintained in normal culture conditions (Fig. 7A). Toinvestigate the stability and degradation of TGFBR2 in cancercell lines, we performed a cycloheximide (CHX, 25 mg/mL) time-course with simultaneous TGFb treatment in cells that wereserum-starved overnight prior to treatment. Although TGFBR2was degraded over time in all cell lines, TGFBR2 exhibitedincreased stability in cells overexpressing USP11-wt, but notUSP11-mut (Fig. 7B). Conversely, TGFBR2 was degraded morequickly in cells lacking USP11 (Fig. 7B). These results show thatUSP11 regulates the stability of TGFBR2during TGFb stimulation.

To examine how USP11 affects TGFb signaling, we performedWestern blots to measure the level of TGFb pathway activation.

We found that USP11 modulates TGFb signaling over shorttimescales in human breast cancer cells, as differences in phos-pho-SMAD2 levels were evident following 30 minutes of TGFbtreatment (Fig. 7C). TGFb transcriptional activity is also affectedby USP11, as the CAGA12 SMAD-dependent luciferase reporter(14) was differentially activated when USP11 expression wasaltered in SUM159 cells (Fig. 7D). Finally, we found that down-stream targets of TGFb are affected by USP11, as USP11 depletionin MDA-MB-231 cells prevented the expression of PAI-1, SNO,SKI, CTGF, and SNAI2 to the same extent as control cells (Fig. 7E).These effects are recapitulated in the induction of EMT markersby TGFb at the protein level in SUM159 and MDA-MB-231cells (Supplementary Fig. S6). Together, these results confirmthat USP11 regulates TGFb signaling and is essential for TGFb-dependent responses in human breast cancer cells.

Figure 5.

USP11 regulates human breast cancer self-renewal. A–C, Primary and secondary sphere formation of T47D cells overexpressing GFP (Ctrl), USP11-wt, orUSP11-mut. Sphere number (A) and sphere volume (B) were quantified. Representative sphere images are shown (C). D, E, F, Primary and secondary sphereformation of T47D cells expressing shCtrl or shUSP11. Sphere number (D) and sphere volume (E) were quantified. Representative sphere images are shown (F).

Garcia et al.





DiscussionAlthough EMT has been studied extensively and clearly plays a

role in metastasis, novel druggable regulators of EMT have yet tobe uncovered and introduced into the clinic. Because themajorityof cancer mortalities are due to metastasis, anti-EMT drugs are animportant deficiency in the current arsenal of cancer therapies. Inthis study, we confirm the role of USP11 in EMT using a robustEMT cell culture model. Furthermore, we establish for the firsttime USP110s importance in human breast cancer cell migration,self-renewal, and metastasis.

In search of novel regulators of EMT within the ubiquitin–proteasome pathway, we mined previously published micro-array data and found USP11, USP13, and UCHL1 to be upre-gulated during EMT. Using human mammary epithelial cells(HMLE), a robust in vitro model for EMT, we show that over-expression of these DUBs alone does not change their pheno-type. However, upon EMT induction with TGFb, only USP11overexpression enhanced the EMT phenotype. Further, modu-lation of USP11 expression affects the development of mesen-chymal characteristics in HMLE cells, namely mesenchymal

marker expression and self-renewal ability. These experimentsare in line with previous work that established a role for USP11,but not USP13 or UCHL1, in TGFb signaling and regulation ofTGFBR1 and TGFBR2 stability (16, 26). Although these studiesuse mouse or non-mammary cell lines, this study provides forthe first time a thorough phenotypic and functional character-ization using a human model of EMT in cells derived from themammary epithelium.

Our study is also the first to show the dependence of humanbreast cancer cells on USP11 for cell migration, self-renewal, andmetastasis. We showed that shRNA knockdown of USP11 slowedscratch wound closure and decreased transwell migration,although the opposite effect was shown in a previous study in786-O cells, a human renal cell adenocarcinoma cell line (27). Inthat same study, shRNA knockdown of USP11 in 786-O cellsincreased cell proliferation, which contrasts with our observationthat overexpression or shRNA knockdown of USP11 in SUM159and MDA-MB-231 cells, respectively, had no effect on cell pro-liferation in a 2D growth assay. However, we did find that USP11knockdown resulted in smaller T47D spheres at the secondary

Figure 6.

USP11 regulates experimental metastasis of human breast cancer cells in mice. A and B, Either 1.5 � 106 SUM159 cells (A) or 1 � 106 MDA-MB-231 cells (B) wereresuspended in PBS and injected via tail vein into NSG mice at a volume of 100 mL. After 6 to 8 weeks, mice were sacrificed and lungs were analyzed formetastatic colonization by fixation, paraffin embedding. H&E-stained sectionswere examined for percentage of tumor cells amongst normal cells (tumor cellularity).Each data point represents the average tumor cellularity of three step sections (100 mm apart) from one mouse. The black arrows indicate examples ofmetastatic tumors within lung tissue.






passage, raising the possibility that USP11 may regulate cellproliferation. Moreover, the tail vein metastasis assay, whichtends to assess parameters related to survival and outgrowth,demonstrated alterations in lung colonization with USP11 over-expression or knockdown. Thus, it remains unclear whetherUSP11 directly regulates cell proliferation, but it appears thatUSP110s role may be context-specific with different targets fordeubiquitination in different cell and tissue types.

In terms of self-renewal, modulating USP11 expression had asignificant effect in HMLE normal epithelial cells and T47Dhuman breast cancer cells. In general, we found that USP11 isrequired for proper sphere formation and sufficient to increasesphere formation, specifically at the secondary passage andindependent of apoptosis. These results suggest that USP11 is

responsible for the maintenance of stem cell characteristics instem-like human breast cancer cells. Unexpectedly, we seeUSP11-dependent effects on sphere size in T47D cells. Spheresize is not well studied, but it is thought that sphere sizeindicates the nature of the founder clone cell as being morestem-like in terms of being able to produce more progenitorcells (28). Our results show for the first time USP110s role inregulating stem-like behavior of human breast cancer cells invitro. Although sphere number and size are not necessarily adefinitive readout of the in vivo presence of cancer stem cells, invivo models of metastasis are rigorous tests of these character-istics. Our study shows that, in this case, the in vivo experimen-tal metastasis model does faithfully recapitulate the in vitrobehaviors of human breast cancer cells.

Figure 7.

USP11 regulates TGFBR2 stability and signaling in human breast cancer cells. A, Flow cytometry analysis of TGFBR2 median fluorescence intensity (MFI) in SUM159cells overexpressing GFP (Ctrl), USP11-wt, or USP11-mut and MDA-MB-231 cells expressing shCtrl or shUSP11. B, Flow cytometry analysis of TGFBR2 MFI overtime after CHX and TGFb treatment in SUM159 and MDA-MB-231 cells. C, Western blot analysis of phosho-SMAD2 (pSMAD2) in SUM159 cells overexpressing GFP(Ctrl), USP11-wt, or USP11-mut and MDA-MB-231 cells expressing shCtrl or shUSP11 before and after TGFb treatment for 30 minutes. All blot images are fromthe same membrane with intervening irrelevant samples removed for clarity. D, Effect of USP11-wt or USP11-mut overexpression and shRNA knockdown onCAGA12-luciferase transcriptional response induced by TGFb in SUM159 cells. E, RT-qPCR analysis of mRNA levels of TGFb target genes in MDA-MB-231 cellsexpressing shCtrl or shUSP11.


Garcia et al.




Tumor-intrinsic subtype analysis of patient survival outcomesprovided information about which patient subpopulations aremore at risk for USP11-associated disease progression andmetas-tasis. In these analyses, we found high USP11 expression inLuminal-type tumors (Luminal A/B, ERþ), but not Basal-liketumors, was associated with decreased survival and metastasis.Luminal-type tumors tend to have more epithelial characteristics,and we speculate that those tumor types are more sensitive toEMT-inducing signals, especially those enhanced by USP11.Because Basal-like tumors have already adopted a mesenchymalphenotype, USP11 may not be predictive of survival in thosepatients. By contrast, we observed an effect of USP11 in regulatingthe behavior of Basal-like human breast cancer cells in ourxenograft studies. This apparent lack of correlation represents alimitation of the study, but raises the possibility that character-istics beyond tumor subtypemay influence whether or not USP11can influence growth, survival, and metastasis.

Mechanistically, we show that USP11 enhances TGFBR2 sta-bility, thereby enhancing TGFb signaling and metastasis. USP11has been shown to target both TGFBR1 and TGFBR2 for deubi-quitination (16, 26) and our study confirms this pathway for thefirst time in human breast cancer cells. It has not escaped ournotice that USP11 likely deubiquitinates TGFBR1 in humanbreast cancer cells in order to affect EMT and metastasis. BecauseTGFb has been linked to EMT and metastasis-promoting func-tions, it is likely that USP11 regulates the metastatic cascade atmultiple points (invasion, intravasation, extravasation, coloniza-tion, any step that depends on TGFb signaling). Thus, our currentstudy suggests a role for USP11 in the complex process ofmetastasis and implicates USP11 as a potential therapeutic targetin breast cancer.

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

Authors' ContributionsConception and design: D.A. Garcia, E.J. Bennett, J.T. ChangDevelopment of methodology: D.A. Garcia, E.J. BennettAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): D.A. Garcia, C. Baek, T. TyslAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): D.A. Garcia, C. Baek, M.V. EstradaWriting, review, and/or revision of the manuscript: D.A. Garcia, J. Yang,J.T. ChangStudy supervision: J.T. Chang

AcknowledgmentsSupported by NIH grants (DK093507 and OD008469 to J.T. Chang,

and CA168689, CA174869, CA206880 to J. Yang). J.T. Chang is aHoward Hughes Medical Institute Physician-Scientist Early Career Award-ee and a V Foundation for Cancer Research V Scholar Awardee. D.A.Garcia is a Howard Hughes Medical Institute Gilliam Fellow. The TissueTechnology Shared Resource is supported by a National Cancer InstituteCancer Center Support Grant (CCSG Grant CA23100).

The authors thank Dwayne Stupack (University of California San Diego),Gen-Sheng Feng (University of California San Diego), and the members of theChang and Bui laboratory for helpful discussions.

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.

Received December 1, 2017; revised February 28, 2018; accepted April 18,2018; published first May 3, 2018.

References1. Chaffer CL, Weinberg RA. A perspective on cancer cell metastasis. Science

2011;331:1559–64.2. Weigelt B, Peterse JL, van Veer LJ. Breast cancer metastasis: markers and

models. Nat Rev Cancer 2005;5:591–602.3. Yeung KT, Yang J. Epithelial-mesenchymal transition in tumor metastasis.

Mol Oncol 2017;11:28–39.4. Padua D, Massagu�e J. Roles of TGFb in metastasis. Cell Res 2008;19:

89–102.5. BholaNE, Balko JM,Dugger TC, KubaM, S�anchezV, SandersM, et al. TGF-b

inhibition enhances chemotherapy action against triple-negative breastcancer. J Clin Invest 2013;123:1348–58.

6. Katsuno Y, Lamouille S, Derynck R. TGF-b signaling and epithelial-mesen-chymal transition in cancer progression. Curr Opin Oncol 2013;25:76–84.

7. Lamouille S, Xu J, Derynck R. Molecular mechanisms of epithelial-mesenchymal transition. Nat Rev Mol Cell Biol 2014;15:178–96.

8. Banno A, Garcia DA, van Baarsel ED, Metz PJ, Fisch K, Widjaja CE, et al.Downregulation of 26S proteasome catalytic activity promotes epithelial-mesenchymal transition. Oncotarget 2016;7:21527–41.

9. Voutsadakis IA. Proteasome expression and activity in cancer and cancerstem cells. Tumour Biol 2017;39:1–17.

10. Eichhorn PJ, Rod�on L, Gonz�alez-Junc�a A, Dirac A, Gili M, Martínez-S�aezE, et al. USP15 stabilizes TGF-b receptor I and promotes oncogenesisthrough the activation of TGF-b signaling in glioblastoma. Nat Med2012;18:429–35.

11. Wu Y, Wang Y, Lin Y, Liu Y, Wang Y, Jia J, et al. Dub3 inhibition suppressesbreast cancer invasion and metastasis by promoting Snail1 degradation.Nature Commun 2017;8:14228.

12. Wu Y,Wang Y, Yang XH, Kang T, Zhao Y,Wang C, et al. The deubiquitinaseUSP28 stabilizes LSD1 and confers stem-cell-like traits to breast cancercells. Cell Rep 2013;5:224–36.

13. Zhang J, Zhang P, Wei Y, Piao HL, Wang W, Maddika S, et al.Deubiquitylation and stabilization of PTEN by USP13. Nat Cell Biol2013;15:1486–94.

14. Zhang L, Zhou F, Drabsch Y, Gao R, Snaar-Jagalska EB, Mickanin CUSP4 isregulated by AKT phosphorylation and directly deubiquitylates TGF-b typeI receptor. Nat Cell Biol 2012;14:717–26.

15. Zhou Z, Zhang P, Hu X, Kim J, Yao F, Xiao Z, et al. USP51 promotesdeubiquitination and stabilization of ZEB1. Am J Cancer Res 2017;7:2020–31.

16. Jacko AM, Nan L, Li S, Tan J, Zhao J, Kass DJ, et al. De-ubiquitinatingenzyme, USP11, promotes transforming growth factor b-1 signalingthrough stabilization of transforming growth factor b receptor II. CellDeath Dis 2016;7:e2474.

17. ZhangH,Meng F, LiuG, Zhang B, Zhu J,Wu F, et al. Forkhead transcriptionfactor foxq1 promotes epithelial-mesenchymal transition and breast can-cer metastasis. Cancer Res 2011;71:1292–301.

18. Sowa ME, Bennett EJ, Gygi SP, Harper JW. Defining the humandeubiquitinating enzyme interaction landscape. Cell 2009;138:389–403.

19. Meerbrey KL, Hu G, Kessler JD, Roarty K, Li MZ, Fang JE, et al. ThepINDUCER lentiviral toolkit for inducible RNA interference in vitro andin vivo. PNAS 2011;108:3665–70.

20. Taube JH, Herschkowitz JI, Komurov K, Zhou AY, Gupta S, Yang J, et al.Core epithelial-to-mesenchymal transition interactome gene-expressionsignature is associated with claudin-low and metaplastic breast cancersubtypes. Proc Natl Acad Sci USA 2010;107:15449–54.

21. Gy€orffy B, Lanczky A, Eklund AC, Denkert C, Budczies J, Li Q, et al. Anonline survival analysis tool to rapidly assess the effect of 22,277 genes onbreast cancer prognosis using microarray data of 1,809 patients. BreastCancer Res Treat 2010;123:725–31.






22. Geback T, Schulz MM, Koumoutsakos P, Detmar M. TScratch: a novel andsimple software tool for automated analysis of monolayer wound healingassays. BioTechniques 2009;46:265–74.

23. Mani SA, Guo W, Liao MJ, Eaton EN, Ayyanan A, Zhou AY, et al. Theepithelial-mesenchymal transition generates cells with properties of stemcells. Cell 2008;133:704–15.

24. Bayraktar S, Gutierrez Barrera AM, Liu D, Pusztai L, Litton J, Valero V, et al.USP-11 as a predictive and prognostic factor following neoadjuvanttherapy in women with breast cancer. Cancer J 2013;19:10–7.

25. Aggarwal K,Massagu�e J. Ubiquitin removal in the TGF-b pathway. Nat CellBiol 2012;14:656–7.

26. Al-Salihi MA, Herhaus L, Macartney T, Sapkota GP. USP11 augmentsTGFb signalling by deubiquitylating ALK5. Open Biol 2012;2:120063.

27. Zhang E, Shen B, Mu X, Qin Y, Zhang F, Liu Y, et al. Ubiquitin-specificprotease 11 (USP11) functions as a tumor suppressor through deubiqui-tinating and stabilizing VGLL4 protein. Am J Cancer Res 2016;6:2901–9.

28. Pastrana E, Silva-Vargas V, Doetsch F. Eyes wide open: a critical review ofsphere-formation as an assay for stem cells. Cell StemCell 2011;8:486–98.


Garcia et al.




2018;16:1172-1184. Published OnlineFirst May 3, 2018.Mol Cancer Res Daniel A. Garcia, Christina Baek, M. Valeria Estrada, et al. and Human Breast Cancer Metastasis

Mesenchymal Plasticity−-Induced EpithelialβUSP11 Enhances TGF

Updated version

10.1158/1541-7786.MCR-17-0723doi:

Access the most recent version of this article at:

Material

Supplementary

http://mcr.aacrjournals.org/content/suppl/2018/05/03/1541-7786.MCR-17-0723.DC1

Access the most recent supplemental material at:

Cited articles

http://mcr.aacrjournals.org/content/16/7/1172.full#ref-list-1

This article cites 28 articles, 4 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://mcr.aacrjournals.org/content/16/7/1172To request permission to re-use all or part of this article, use this link



http://mcr.aacrjournals.org/lookup/doi/10.1158/1541-7786.MCR-17-0723

http://mcr.aacrjournals.org/content/suppl/2018/05/03/1541-7786.MCR-17-0723.DC1

http://mcr.aacrjournals.org/content/16/7/1172.full#ref-list-1

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

mailto:[email protected]

http://mcr.aacrjournals.org/content/16/7/1172


Documents

USP11 Enhances TGFb-Induced Epithelial Mesenchymal Plasticity … · Mesenchymal Plasticity and Human Breast Cancer Metastasis Daniel A. Garcia1,2, Christina Baek1, M.Valeria Estrada3,Tiffani