usp28 deﬁciency promotes breast and liver carcinogenesis as … · oncogenes and tumor...

Oncogenes and Tumor Suppressors

USP28 Deficiency Promotes Breast and LiverCarcinogenesis as well as Tumor Angiogenesisin a HIF-independent MannerKati Richter1, Teija Paakkola2,3,4, Daniela Mennerich1, Kateryna Kubaichuk1,Anja Konzack1, Heidi Ali-Kippari1,2, Nina Kozlova1, Peppi Koivunen1,2,5,Kirsi-Maria Haapasaari6, Arja Jukkola-Vuorinen7, Hanna-Riikka Teppo1,6, Elitsa Y. Dimova1,Risto Bloigu8, Zoltan Szabo9, Risto Kerkel€a2,3,9, and Thomas Kietzmann1,2

Abstract

Recent studies suggest that the ubiquitin-specific proteaseUSP28 plays an important role in cellular repair and tissueremodeling, which implies that it has a direct role in carcinogen-esis. The carcinogenic potential of USP28 was investigated in acomprehensive manner using patients, animal models, and cellculture. The findings demonstrate that overexpression of USP28correlates with a better survival in patients with invasive ductalbreast carcinoma. Mouse xenograft experiments with USP28-deficient breast cancer cells also support this view. Furthermore,lack of USP28 promotes a more malignant state of breast cancercells, indicated by an epithelial-to-mesenchymal (EMT) transi-tion, elevatedproliferation,migration, and angiogenesis aswell as

a decreased adhesion. In addition to breast cancer, lack of USP28in mice promoted an earlier onset and a more severe tumorformation in a chemical-induced liver cancer model. Mechanis-tically, the angio- and carcinogenic processes driven by the lack ofUSP28 appeared to be independent of HIF-1a, p53, and 53BP1.

Implications: The findings of this study are not limited toone particular type of cancer but are rather applicable for carci-nogenesis in a more general manner. The obtained data supportthe view that USP28 is involved in tumor suppression and has thepotential to be a prognostic marker. Mol Cancer Res; 16(6); 1000–12.�2018 AACR.

IntroductionCarcinogenesis is associated with excessive protein synthesis

and degradation. One major pathway involved in protein degra-dation is the ubiquitin-mediated proteasomal degradation. Thisprocess is reversible and enzymes, termed deubiquitinases(DUB), remove ubiquitin from protein substrates. The humangenome encodes for up to 100 DUBs, which can be divided intosix subgroups; the ubiquitin-specific proteases (USP), the ubiqui-tin carboxyl-terminal hydrolases (UCH), the Otubain/Ovariantumor-domain containing proteins (OTU), the Machado–Josephdisease domain superfamily (MJD), the JAB1/MPN/MOV34 pro-

teases (JAMMs), and the monocyte chemotactic protein-inducedproteins (MCPIP; ref. 1). All DUBs are proteases and belongmainly to the group of cysteine proteases; only a small fraction,the JAMMs, belong to the zinc metalloproteases (2). The USPscomprise the largest group among the DUBs, and although somemembers have been implicated to have important roles in pro-cesses like inflammation, immunity, or cell cycle, not manymechanistic details of their proper involvement in cancer areknown (3–5).

Another integral part of the carcinogenic process in solidtumors is a restricted supply with oxygen (commonly calledhypoxia). Importantly, the hypoxic areas in various tumorentities display an increased amount of hypoxia-induciblefactor-1a (HIF-1a) and are associated with a poor prognosis(6, 7). Our recent findings from studies on HIF-1a have pointedout that a DUB called ubiquitin-specific protease 28 (USP28) isa positive regulator of HIF-1a protein stability which wouldimply that USP28 has a procarcinogenic role (8). This issupported by findings showing that USP28 opposes the deg-radation of oncogenes such as c-Myc, c-Jun, NICD1, and LSD1(9–13).

In contrast, recent reports employing genome-wide CRISPR/Cas9 knockout screens highlight a role of USP28 for p53 stabi-lization andG1 cell-cycle arrest (14, 15). These data, together withearlier findings showing that USP28 stabilizes the checkpointarrest protein claspin, and the tumor suppressor Chk2 (16) led toa model in which USP28 deficiency would predispose to path-ologic outcomes such as increased tumor formation. Thus, a clearview about the causative role of USP28 in carcinogenesis has notbeen reached.

1Faculty of Biochemistry and Molecular Medicine, University of Oulu, Oulu,Finland. 2Biocenter Oulu, University of Oulu, Oulu, Finland. 3Medical ResearchCenter Oulu, Oulu University Hospital, University of Oulu, Oulu, Finland. 4PED-EGO Research Unit, University of Oulu, Oulu, Finland. 5Centre of Excellence inCell-Extracellular Matrix (ECM) Research, University of Oulu, Oulu, Finland.6Department of Pathology, University of Oulu, Oulu, Finland. 7Department ofOncology and Radiotherapy, Oulu University Hospital, University of Oulu, Oulu,Finland. 8Medical Informatics and Statistics Research Group, University of Oulu,Oulu, Finland. 9Department of Pharmacology and Toxicology, Research Unit ofBiomedicine, University of Oulu, Oulu, Finland.

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

Corrected online September 21, 2018.

Corresponding Author: Thomas Kietzmann, Faculty of Biochemistry andMolecular Medicine, Biocenter Oulu, University of Oulu, Aapistie 7B, OuluFI-90230, Finland. Phone: 3582-9448-7714; E-mail: [email protected]

doi: 10.1158/1541-7786.MCR-17-0452

�2018 American Association for Cancer Research.

MolecularCancerResearch

Mol Cancer Res; 16(6) June 20181000

on August 19, 2020. © 2018 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from

Published OnlineFirst March 15, 2018; DOI: 10.1158/1541-7786.MCR-17-0452





http://crossmark.crossref.org/dialog/?doi=10.1158/1541-7786.MCR-17-0452&domain=pdf&date_stamp=2018-9-20

http://mcr.aacrjournals.org/



On the basis of these conflicting observations, we took theunbiased approach and investigated the role of USP28 in acomprehensive manner in cancer patients, cell cultures, andanimal models.

Materials and MethodsPatient and tumor material

Seventy-two formalin-fixed, paraffin-embedded breast cancersamples from female patients and 7 formalin-fixed, paraffin-embedded samples from patients with hepatocellular carcinoma(Supplementary Fig. S1A) were provided by the Department ofPathology,OuluUniversityHospital (Oulu, Finland). All patientsgave written consent and the studies were conducted in accor-dancewith the recognized ethical guidelines from theDeclarationof Helsinki and were approved by the local Ethical Committeeand the Finnish National Supervisory Authority for Welfare andHealth. Clinical and pathologic records gave information aboutthe patient's characteristics, while scoring was performed accord-ing to the WHO classification.

ScoringAll samples were evaluated according to the percentage of

stained nuclei and intensity. The classification of estrogenreceptor (ER), progesterone receptor (PR), HER2, Ki-67, HIF-1a, HIF-2a, and PHD1-3 can be found in ref. 17. Percentages ofUSP28-stained nuclei were scored as followed: 0 ¼ negative, 1¼ 1%–25%, 2 ¼ 26%–50%, 3 ¼ 51%–75%, 4 ¼ 76%–100%;intensity: 1 ¼ very weak, 2 ¼ weak, 3 ¼moderate, 4 ¼ strong, 5¼ very strong. The categorization was conducted by a pathol-ogist who was blinded and the samples were considered to beUSP28 positive if the percentage of stained nuclei was >25%with an intensity >2.

Cell culture and maintenanceThe human breast cancer cell lines MDA-MB-231 (#HTB-26)

and BT-549 (#HTB-122) cells were purchased from ATCC andcultured under normoxia [16%O2, 5%CO2 and 79%N2 (v/v)] inDMEM or RPMI1640 supplemented with 10% FBS, respectively.MCF7 (#ACC 115) cells were purchased fromDSMZ and culturedunder normoxia in DMEM with 10% FBS. Stable transfectants ofthe MDA-MB-231 cell line were generated using two differentUSP28 shRNA–expressing vectors (USP28-kd-1 andUSP28-kd-3)and a respective scrambled control vector (USP28-sc; ref. 8). Allcell lines were tested for Mycoplasma and authenticated by STRprofiling.

ImmunofluoresenceCells were grown in low density on glass coverslips within 6-

well plates. After washing, cells were fixed with 4% paraformal-dehyde for 10minutes, followed by blocking in 1%BSA and0.1%saponin in 1� PBS for 1 hour. The primary antibody againsta-tubulin (Sigma-Aldrich, T5168) was used in a 1:1,000 dilutionin 1%BSA followed by incubationwith the secondary anti-mouseAlexa Fluor 488–conjugated antibody (Thermo Fisher Scientific,A-11001) in a 1:500 dilution. After two washing steps, cells werestained with 1 mg/mL Hoechst (bisBenzimide H 33258, Sigma-Aldrich). Stainings were visualized with the Zeiss LSM700 con-focal microscope using the 63� PlanApo oil immersion objectiveand the Zen2009 software.

Cell proliferationA total of 4� 105 cells were seededonto 100-mmculture dishes

and cell number was determined by counting with a hemocy-tometer after 5 days. Incorporation of bromodeoxyuridine(BrdUrd) into newly synthesized DNA was detected by immu-nolabeling according to the manufacturer's instructions (MerckMillipore) using 2� 104 cells per well in a 96-well plate. A total of5 � 103 cells were seeded onto 96-well plates and images weretaken every 2 hours with a 10� objective from 4 optical fields/well. Images were analyzed as cell/background ratios in a live-cellimager (IncuCyte).

Western blot analysisWestern blot analysis was carried out as described previously

(18). In brief, 100 mg of protein were loaded onto SDS–poly-acrylamide gels and after electrophoresis blotted onto a nitrocel-lulose membrane. The primary antibodies against HIF-1a (BDBiosciences, 610959), c-Jun (Cell Signaling Technology, 9162),Notch1 (Abcam, 65297), a-SMA (Sigma-Aldrich, A2547),a-tubulin (Sigma-Aldrich, T5168), b-catenin (Santa Cruz Bio-technology, sc-7963), c-Myc (Santa Cruz Biotechnology, sc-764),E-cadherin (BD Biosciences, 610181), p53 (Cell Signaling Tech-nology, 2524), USP28 (Sigma-Aldrich, HPA006778), and 53BP1(Novus Biologicals, NB100-304) were used in a 1:1,000 dilution.The secondary antibody, anti-mouse or anti-rabbit immunoglob-ulinGhorseradish peroxidase (Bio-Rad) respectively, was appliedin a 1:5,000 dilution. The enhanced chemiluminescence (ECL)system (Amersham) was used for detection.

Adhesion assay and crystal violet stainingA total of 2 � 106 cells/well were seeded onto 6-well plates.

After 24 hours, one plate was transferred to an orbital shaker for 2hours at 250 rpm and 37�C while the control plate remained inthe incubator. After washing, cells were fixed with 4% parafor-maldehyde followed by staining in 0.0075% crystal violet solu-tion (Applichem/VWR) for 15 minutes.

Anchorage-independent growth in soft agarAnchorage-independent growth was analyzed in a three-layer

soft agar assay, containing a basal agar layer followed by a celland a top agar layer. The agar layers were prepared by mixingprewarmed (37�C) 2 � DMEM (containing 20% FBS) with anequal volume of agarose (DNA grade). For the basal and toplayer, 1.2% agar was used, while only 0.8% agar was used for thecell-containing layer with 2 � 103 cells/well. Fifty-microlitersuspension was used for each layer per well of a 96-well plate.The plates were incubated in a humidified incubator at 37�Cwith 5% atmospheric CO2 for 2 weeks. Cell growth was mea-sured using the Alamar blue assay (Invitrogen) and fluorescencewas recorded in a Fluroskan Ascent FL type 374 (ThermoScientific) reader with an excitation wavelength of 530 nm andemission at 590 nm.

Wound healing assayMigrationwas analyzed in awoundhealing assay using the live-

cell imager (IncuCyte). A total of 2 � 105 cells/well in 100 mL ofserum-free DMEM were seeded onto 96-well ImageLock Micro-plates (#4379, Essen BioScience) and a 96-pin wound-makingtool was used to create homogeneous, 700–800 mmwide scratchwounds in cell monolayers. Detached cells were removed andreplaced by 100mL of fresh serum-freeDMEM. Automated images

Carcinogenic Activity of USP28 in Breast and Liver

www.aacrjournals.org Mol Cancer Res; 16(6) June 2018 1001




were taken every hour and cellmigrationwas quantified as relativewound densities (%).

qRT-PCRThe GenElute mammalian total RNA miniprep kit (Sigma-

Aldrich) was used to extract total RNA from cells, followed byreverse transcription (cDNA Synthesis Kit, Quanta Bioscience, GEHealthcare). qRT-PCR was performed with the Universal SYBRGreen Supermix reaction kit (Bio-Rad) with cDNA in a 1:25dilution. Forward and reverse primers used in this reaction aresummarized in Supplementary Table S1. The relative mRNAexpression was determined using the DDCt analysis method(19) using b-actin as reference gene.

Chorioallantoic membrane assayTo analyze angiogenesis, MDA-MB-231 USP28-sc and USP28-

kd cell spheroids were grown in hanging drop culture and seededonto a transwellmembranewhichwas transferred to a 7-day ex ovochicken CAM. Vessel growth and formation was followed by dailymonitoring.

Mouse modelsAll animalswere housed in the Laboratory Animal Center of the

University ofOulu in specific pathogen-free facilities on a 12-hourlight/dark cycle, at a constant temperature of 22�C and receiveddiet and water ad libitum. All protocols for animal use andexperiments were approved by theNational Animal ExperimentalBoard of Finland (ELLA) aswell as the AnimalWelfare Body of theLaboratory Animal Center and conducted according to the EUdirective 2010/63/EU.

Xenograft mouse modelA total of 5�105 cellswere injected in the thoracic and inguinal

mammary fat pad of female athymic nudemice (Envigo), housedin IVC cages for up to five weeks. For tumor collection, animalswere euthanized by CO2 inhalation.

Usp28 knockout miceUsp28 knockout mice were generated by insertion of a gene

trap vector (pGT01xr) into the first intron of the Usp28 gene of amouse E14TG2a.4 embryonic stem (ES) cell line (SupplementaryFig. S1B). A stock of the heterozygous ES cells (SIGTR ES cellline AE0197; #023027-UCD) was obtained from MutantMouse Resource and Research Centers (University of CaliforniaDavis, Davis, CA). The cells were injected into the blastocystsfrom C57BL/6J mice to produce chimeras. Resulting chimericmale animals were backcrossed to wild-type C57BL/6J miceand the heterozygous offspring was used for further mating togenerate Usp28 knockout (Usp28�/�) and the associated control(Usp28þ/þ) mice. To verify the presence of the entire gene trapvector, we performed PCR analysis on genomic DNA from therespective offspring (Supplementary Fig. S1C). In addition, theknockout of Usp28 was determined by Western blot analysis(Supplementary Figs. S1D and S2A–S2C).

Diethylnitrosamine-induced liver carcinomaTo induce hepatocellular carcinogenesis, male mice were

injected intraperitoneally (i.p.) with diethylnitrosamine (DEN,20 mg/g body weight) two weeks after birth. After different timepoints (3, 6, 9, and 12 months) the animals were euthanized byCO2 inhalation and subjected to necropsy. The tumor develop-

ment in alive animals was followed by ultrasound using the Vevo2100 system (VisualSonics).

Mouse necropsy and sample preparationFor necropsy, the animal was sacrificed by CO2 inhalation and

additional cervical dislocation. After weighing, external examina-tion anddesinfectionwith 70%EtOH, the abdomen and the chestwere opened. Blood samples for serum analysis were obtainedfrom the inferior vena cava. The liver was fixed overnight in 4%buffered formaldehyde, rinsed in water, and incubated in 70%EtOH. After dehydration using the tissue processor (Tissue-TekVIP Jr.), the samples were embedded in paraffin for histologicexamination. Determination of alanine aminotransferase (ALT)in serum was carried out by a standard procedure (20).

IHCIHC staining was performed on formalin-fixed, paraffin-

embedded specimens. In general, 4-mm tissue sections weredeparaffinized and rehydrated in a descending ethanol row. Forthe hematoxylin/eosin (HE) staining the samples were stained for1minute inHarris–hematoxylin solution (Sigma-Aldrich), rinsedinH2O, anddipped into a 1%acid alcohol solution, followedby acounterstain in Eosin-Y-solution (Sigma-Aldrich). After awashingstep inH2O, the samplesweredehydrated in an ascending ethanolrow and mounted with Pertex (Histo Lab, Espoo). For antibodystaining, samples were treated with Tris/EDTA for antigen retriev-al, followed by blocking of nonspecific binding in blockingsolution (En Vision Detection Sysyem) and incubation with theprimary antibody against endomucin (Santa Cruz Biotechnology,sc-65495), glutamine synthetase (GS; Novus Biologicals, NB110-41404), gluthathione-S-transferase pi (GST-P; Novus Biologicals,NBP1-42011), p53 (Santa Cruz Biotechnology, sc-126), USP28(Sigma-Aldrich, HPA006778), and 53BP1 (Novus Biologicals,NB100-304) in a 1:500 dilution. After incubation with the sec-ondary peroxidase-conjugated antibody, goat anti-rabbit IgG(Bio-Rad, 170-6515) or goat anti-rat IgG (Abnova, PAB29749),respectively, in a 1:200 dilution, DAB was used according to themanufacturer's protocol (En Vision Detection System, Dako) forchromogenic detection followedbyhematoxylin counterstaining,dehydration, and mounting. Samples were analyzed using theOlympus BX51 microscope (Olympus Corporation).

Statistical analysisStatistical analysis of the patient cohort samples were carried

out using the SPSS software (Released 2008. SPSS Statistics forWindows, Version 17.0; SPSS Inc.). All clinical characteristics wereexpressed as percentages. To evaluate the associations and corre-lations the c2 or Fisher exact test was used. The breast cancer–specific and disease-free survival rates were analyzed using theKaplan–Meier method. Within the whole study, all values areexpressed as mean � SE, and statistical significance was deter-mined using the two-tailed Student t test.

ResultsUSP28 expression is associated with a better survival of humanbreast cancer patients

To gain insight whether USP28 may be important in breastcancer, we investigated USP28 expression (Fig. 1A) in a breastcancer patient cohort and correlated it with major clinical prog-nostic factors. The patient cohort comprised 72 women with a

Richter et al.

Mol Cancer Res; 16(6) June 2018 Molecular Cancer Research1002




mean age of 59 years (range from 28 to 87 years) with invasiveductal breast carcinoma. The major clinical characteristics of thepatients, with respect to the TNM classification, associated withUSP28 expression are given in Fig. 1B. Themajority of the tumorswere classified as T1 or T2, nodal negative (N0) or N1; further-more, two patients displayedmetastasis, while tumor grades I–IIIwere almost equally distributed. The status of the hormonereceptors revealed that a positive USP28 expression was accom-panied by a 70% positive expression of the estrogen receptor (P¼0.016), while no correlations between USP28 expression and

progesterone receptor or human epidermal growth factor receptor2 were found. Furthermore, analysis of breast cancer–specificsurvival and disease-free survival by the Kaplan–Meier methodrevealed that a positive USP28 expression is associated with alonger disease-free survival of the patients (P¼ 0.047; Fig. 1C). Inaddition, a similar strong positive survival trend was seen bycomparing the USP28 expression with the breast cancer–specificsurvival (P¼ 0.056; Fig. 1D). In addition, we queried six publiclyavailable datasets (www.cbioportal.org) from patients with inva-sive breast carcinoma for genomic changes in USP28. Depending

Figure 1.

USP28 expression correlateswith a better survival in ductal breast carcinomas.A,Representative USP28 stained IHC images for normal breast tissue (left) and breastcancer tissue (right; scale bar, 50 mm). B, Distribution of USP28 compared to clinicopathologic variables. Data are presented as percentages; � , significantdifference (P < 0.05). Disease-free (C) and D breast cancer–specific survival (D) of patients according to the USP28 expression. Disease-free survival and breastcancer–specific survival of 72 patients were analyzed by the Kaplan–Meier method: P ¼ 0.047 (C) and P ¼ 0.056 (D).





www.cbioportal.org


Richter et al.





on the study, we found that about 1%–2%of the cases hadUSP28alterations in formofmutations anddeletions, the latter being themost prominent with values up to about 1.5%.No amplificationsof USP28 in breast cancer were found. Survival data from theTCGA2015Breast InvasiveCarcinoma study showed that patientswith alterations in USP28 had a median survival of 72.5 months,whereas patients without alterations in USP28 had a medianoverall survival of 130.06months. Together, these data show thatUSP28 expression in breast cancer correlates with a better disease-free survival, implying that USP28 expressionmay have a defensefunction promoting patient survival.

USP28 deficiency promotes a more malignant state of breastcancer cells

To further investigate the role of USP28 in breast cancer, wegenerated two pools of human USP28-knockdown (USP28-kd-1 and USP28-kd-3) breast cancer cell lines, in which USP28 wasdepleted by 80% and 50% when compared with the scrambled(USP28-sc) cells, respectively. Crystal violet staining (Fig. 2Aand B) as well as Hoechst and a-tubulin double staining (Fig.2C) revealed that both USP28-kd cells (kd-1 and kd-3) resem-bled a fibroblast-like morphology with an elongated shape andprotrusions while the USP28-sc cells were of rather sphericalshape, which is typical for a polar epithelial cell. The analysis ofa-tubulin, which together with b-tubulin forms the microtu-bules of the cytoskeleton, by immunofluorescence microscopysupported the detected morphologic differences between theUSP28-sc and USP28-kd cells (Fig. 2C). Thus, the changes incell morphology seen in the two USP28-kd cell lines mayresemble the process of epithelial-to-mesenchymal transition(EMT), a feature typically seen in cancer cells. As the process ofEMT is associated with a decrease in E-cadherin and an increasein alpha-smooth muscle actin (a-SMA) expression, we ana-lyzed the expression of both markers in the breast cancer celllines MDA-MB-231, and BT549 upon knockdown of USP28.The Western blot analyses show that knockdown of USP28reduced E-cadherin, whereas it induced a-SMA expression(Fig. 2D and E; Supplementary Fig. S3A and S3B). Reexpressionof USP28 in the MDA-MB-231 USP28-kds cell lines restoredthe E-cadherin and a-SMA expression (Fig. 2E and F).

Next, we determined whether the knockdown of USP28 affectsproliferation, adhesion, migration, and colony formation of thecells. The knockdown of USP28 in MDA-MB-231, BT-549, and

MCF-7 cells increased proliferation as indicated by cell count,BrdUrd incorporation, and live-cell confluence measurements(Fig. 2G; Supplementary Fig. S3C–S3E and S4A–S4C). The induc-tion of proliferation in the USP28-kd cells was antagonized byreintroducingUSP28 (Fig. 2H). Furthermore, deficiency ofUSP28decreased cell adhesion in USP28-kd cells; this was rescued byreexpression of USP28 in the USP28-kd cells (Fig. 2I–L; Supple-mentary Fig. S4D–S4E).Wealso investigatedwhether knockdownof USP28 affects anchorage-independent cell growth and foundthat lack of USP28 increased cell colony formation in soft agar(Fig. 2M and N). Migration assays revealed that the USP28-deficient cells had a higher migration capacity than the controlcells (Fig. 2O and P).

As USP28 was previously shown to be involved in the regula-tion ofHIF-1a, c-Myc, c-Jun, Notch1, and b-catenin, we examinedtheir protein levels under normoxia and hypoxia in MDA-MB-231, BT549, and MCF-7 USP28-kd cells and their respectiveUSP28-sc counterparts. While HIF-1a levels where barely detect-able under normoxia, the knockdown of USP28 reduced thehypoxia-dependent induction of HIF-1a as in previous studies(Fig. 3A–C; Supplementary Figs. S3C, S3D, S4A and S4B). In linewith expectations, knockdown of USP28 also decreased the levelsof c-Myc, c-Jun, and Notch1 under both normoxia and hypoxia(Fig. 3A–C; Supplementary Figs. S3C and S3D and S4A and S4B).Again, reexpression of USP28 in the three different USP28-kd celllines abrogated the decrease in HIF-1a, c-Myc, c-Jun, and Notch1(Fig. 3B and C). In contrast, the expression of b-catenin was notaffected by the knockdown ofUSP28 nor by the overexpression ofUSP28 (Fig. 3B and C).

Together, these data indicate that lack of USP28 contributes toEMT, and to development of a more proliferative and aggressivecellular phenotype.

Lack of USP28 promotes tumor growth and angiogenesis in amouse xenograft model and on chicken CAM

As the abovementioned data pointed out that lack of USP28would promote carcinogenesis, we next conducted in vivo tests. Todo this, we transplanted USP28-sc and USP28-kd cells into themammary fat pads of female nude mice and followed the tumorgrowth forfiveweeks. A higher number of tumors developed fromthe transplanted USP28-kd cells (Fig. 4A). In particular, duringweek four, from day 27 to day 34, these USP28-kd cell–derivedtumors increased by about 10-fold in volume compared with the

Figure 2.USP28 knockdown promotes EMT and conversion into a more malignant phenotype. To determine the morphologic change from a spherical shape of the USP28-positive control cells (USP28-sc) toward an elongated, fibroblast-like phenotype in the USP28 knockdown cells (USP28-kd-1 and USP28-kd-3), cells wereassayed by crystal violet staining. A, Quantification of crystal violet staining. >200 cells were analyzed per experiment. Cell surface area was quantifiedmicroscopically by using a CellCountmodule and ImageJ software. The average cell surface of sc-cells was set to 100%; Values aremeans� SE of three independentexperiments. � , Significant difference (P < 0.05). B, Representative images of crystal violet staining (scale bar, 10 mm). C, Immunofluorescence stainingwith an antibody against a-tubulin (scale bar, 10 mm). D and E, Quantification of Western blot analysis from 100 mg protein of USP28-kd (kd-1 and kd-3) andUSP28-sc (sc) cells probed with antibodies against the EMT markers a-SMA and E-cadherin as well as USP28. The values are presented as mean � SE of 4independent experiments. � , Significant difference (P < 0.05). F, Representative immunoblots. G and H, The proliferative ability of the cells was measured by cellcount, BrdUrd incorporation and the Incucyte live-cell imager. The data are displayed as means � SE of 3 independent experiments and normalized to USP28-sccells and USP28-kd, respectively. �, Significant difference (P < 0.05). V, vector; U, USP28. I and K, Cell adherence was measured by an adhesion assay wherecells were under continuous movement for 2.5 hours. The values are presented as mean � SE of three different experiments. The cell number of USP28-sc (sc) orUSP28-k (kd) cells transfected with an empty control vector (V) without motion were set to 100%. � , Significant difference (P < 0.05). U, USP28 transfected.J and L, Representative images from an adhesion experiment.M,USP28-sc (sc) and USP28 knockdown cells (kd-1 and kd-3) were allowed to grow in soft agar for 10days. The number of colonies from USP28-sc (sc) cells was set to 100%. The experiment was done in triplicates. � , significant difference (P < 0.05). N,Photographs from a representative soft-agar experiment. O, Migration of USP28-sc (sc) and USP28-kd (kd-1 and kd-3) cells was analyzed for 72 hours using theIncucyte live-cell imager. The data are shown as mean � SE of 4 different experiments and normalized to USP28-sc cells; � , significant difference(P < 0.05). P, Representative images of migrated cells.






tumors grown from the control cells (Fig. 4A and B). Furthermore,the USP28-deficient cells gave rise to an approximately 2-foldhigher number of tumors (Fig. 4C), which had also an about 2-fold higher weight (Fig. 4D). IHC revealed that the angiogenic

marker endomucin was enriched in USP28-kd cell–derivedtumors (Fig. 4E), indicating that USP28 deficiency promotestumor angiogenesis. Indeed, USP28-kd–mediated angiogenesiswas not only enhanced in themouse xenograft model, but also in

Figure 3.

Knockdown of USP28 downregulates HIF-1a, c-Myc, c-Jun, and Notch1 protein levels. A, Scrambled control cells (USP28-sc) and USP28 knockdown (kd-1 and kd-3)cells were exposed to normoxia (16% O2) and hypoxia (5% O2) for 12 hours. The protein levels in the USP28-sc (sc) cells under normoxic conditions were setto 1. Values are means� SE of 3 different experiments. � , Significant difference (P > 0.05) between USP28-sc and USP28-kd cells. B, For rescue experiments USP28knockdown (kd-1 and kd-3) cells were transfected with an expression vector for USP28 and further cultured under normoxia (16% O2) and hypoxia (5% O2)for 12 hours. The respective protein levels measured byWestern blot analysis in USP28-sc cells under normoxia were set to 1. Data are means� SE of 3 independentexperiments. � , Significant difference (P < 0.05) between USP28-sc and USP28-kd cells. C, Representative Western blot analysis. One-hundred microgramsof total protein lysates were analyzed with antibodies against USP28, HIF-1a, c-Myc, c-Jun, Notch1, b-catenin, a-SMA, E-cadherin, and a-tubulin.

Richter et al.





the chicken chorioallantoic membrane assay (CAM) were anincreased growth of blood vessels toward the USP28-kd cellscould be observed (Fig. 4F). To further substantiate this, weexamined the expression of pro- and antiangiogenic markers byqRT-PCR. We could observe an increase in the proangiogenicmarkers fibroblast growth factor 2 (FGF2), VEGF-A, and plasmin-ogen activator inhibitor-1 (PAI-1) in the USP28-kd cells, whereasexpression of the antiangiogenic markers thrombospondin 1(TSP1) and thrombospondin 2 (TSP2) remained unaltered whencompared with the control USP28-sc cells (Fig. 4G).

USP28 deficiency promotes the formation of liver tumors inresponse to diethylnitrosamine

To examinewhether the tumor suppressing effects ofUSP28 arealso applicable to other cancer types, we extended our investiga-tions towards liver cancer. First, we analyzed a small cohort of 7patients with hepatocellular carcinoma for USP28 expression andfound that USP28 expression is reduced within the carcinoma

when compared with normally looking liver tissue of the samepatient (Fig. 5A and B; Supplementary Fig. S1A). Furthermore, thesurvival data from the 373 cases in the Liver HepatocellularCarcinoma (TCGA) study showed that, similar to breast cancer,about 1.2%of all cases had genetic alterations ofUSP28 in formofmutations and deletions. Again, deletions made the most prom-inent part with about 1.5%; no amplifications were indicated.Although not significant, there was a clear trend showing that theoverall survival was reduced to 45.07months in patient caseswithUSP28 alterations versus 58.84 months in cases without altera-tions. The mean median months' disease-free survival was 16.13versus 20.93 months in cases with alterations versus withoutalterations, respectively. Thus, these data suggest a tumor-sup-pressing role of USP28 for liver cancer.

To further investigate this in vivo, we created Usp28 knockoutmice (Supplementary Fig. S1B–S1D) Altogether, more than 50male and female animals were followed up to the age of one year.Usp28 knockout mice were fertile and did not display any

Figure 4.

USP28 deficiency promotes tumorgrowth and angiogenesis. A, Tumordevelopment in a mouse xenograftmodel (12 mice per group).Representative image of tumors derivedfrom USP28-sc and USP28-kd cells. B,Tumor growth was recorded over time,the volume determined, and therespective volume increase of theUSP28-kd tumors was compared withthe USP28-sc tumors. C, Increase intumor number of the USP28-kd tumorscompared with the USP28-sc tumors.The total number of tumors derivedfrom the USP28-sc cells was set to 1. D,Increase in tumor weight of the USP28-kd tumors compared with the USP28-sctumors. The average weight of tumorsderived from the USP28-sc cells was setto 1. The data are presented as mean �SE. � , Significant difference (P < 0.05),USP28-sc versus USP28-kd cell–derivedtumors. E, Representative histologicimages of HE- and endomucin-stained4-mm sections of paraffin-embeddedtumor samples grown from USP28-sc(left) and USP28-kd (right) cells (scalebar, 100 mm). F, Representative imagesof a chorioallantoic membrane (CAM)assay where spheroids from USP28-scand USP28-kd cells were seeded onchicken CAM; vessel growth after 7 daysis indicated by arrows. G, QuantificationofmRNA levels ofUSP28, proangiogenicmarkers (FGF2, VEGF-A and PAI-1), andantiangiogenicmarkers (TSP1 and TSP2)in USP28-sc and USP28-kd cells. Foreach mRNA, the levels in USP28-sc cellswere set to 1. The values are presentedas mean � SE of 3 independentexperiments. � , Significant difference(P < 0.05).






abnormalities in phenotype, behavior nor any spontaneoustumor development. The weaning behavior, the litter size as wellas the survival of the littermates, was comparable with that ofwild-type mice. In general, nursing, ingestion, behavior, and lifespanofUsp28�/�micewere not different comparedwith thewild-type mice. Organ morphology did not show abnormalities in

particular not in breast tissue, and, in this model, in intestinaltissues. To assess the role of USP28 in carcinogenesis we decidedto boost the carcinogenic process by appling diethylnitrosamine(DEN) that causes liver cancer (21–23).

To follow the development of hepatic tumors over a prolongedtime period, DEN-treated Usp28þ/þ and Usp28�/� mice were

Figure 5.

Lack of USP28 is associated withhepatocellular carcinoma in humansand promotes liver damage andcarcinogenesis in thediethylnitrosamine (DEN)-induced livercancermousemodel.A,RepresentativeIHC images for USP28-positive normalliver (left) and USP28-negativehepatocellular carcinoma (right)expression (scale bar, 50 mm). B,Intensity and distribution of USP28nuclear staining in seven patients withhepatocellular carcinoma. Percentagesof USP28-stained nuclei were scored as:� ¼ negative, 1 ¼ 1%–25%, 2 ¼ 26%–50%, 3 ¼ 51%–75%, 4 ¼ 76%–100%.C, Usp28þ/þ and Usp28�/� mice wereinjectedwith 20 mg/g bodyweight DENtwo weeks after birth. Timeline of DENtreatment with representativeultrasound images of livers of anUsp28þ/þ mouse (top) and anUsp28�/� mouse (bottom) followedover several months (tumorigenic areasare circled in white, scale bar: 2 mm). D,Alanine aminotransferase (ALT) levelsin serum samples of 9-month-olduntreated and DEN-injected Usp28þ/þ

and Usp28�/� mice. The data arepresented as mean� SE of 4 animals. � ,significant differences (P < 0.05) DEN-Usp28�/� versus DEN-Usp28þ/þ mice.E, Representative images ofmacroscopic changes in livermorphology of 9- and 12-month-oldDEN-treated Usp28þ/þ andUsp28�/� mice. The appearance ofmacroscopically visible tumor nodesare indicated by arrows (scalebar: 1 cm).

Richter et al.





subjected to noninvasive ultrasound examination (Fig. 5C).Usp28�/� animals displayed an increased and earlier appearanceof hepatic neoplastic lesions as well as formation of tumors whencompared with the Usp28þ/þ animals. Lesions, indicative forpreneoplastic changes could be observed at an average age of 7months in Usp28�/� and at an age of 9 months in the Usp28þ/þ

mice. Furthermore, the progress of tumor formation up to 12months is much more advanced in Usp28�/� animals comparedwithUsp28þ/þmice (Fig. 5C). These observations were supportedby elevated serum alanine aminotransferase (ALT) levels inUsp28�/� mice, an indicator for hepatocellular damage (Fig.5D). Macroscopic analysis of livers from 9- and 12-month-oldmice displayed more severe changes in liver morphology andhigher appearance of tumors inDEN-treatedUsp28�/�micewhencompared with DEN-injected Usp28þ/þ mice (Fig. 5E). Whilemost of the DEN-injected Usp28þ/þ mice showed a liver tumorburden of around20%, the range extendedup to 80% in theDEN-treated Usp28�/� animals. The tumor-promoting effect of USP28deficiency was even more evident 12 months after DEN injection(Supplementary Table S2; Fig. 5C and D).

Microscopic analysis of the liver tissue supported the macro-scopic observations. While hematoxylin–eosin–stained liver tis-sue sections from untreated mice were morphologically normalanddisplayedno signs of liver damageor failure, livers fromDEN-treatedmice showed neoductular proliferation in the portal triadsas well as periportal and intra-acinar inflammatory infiltrates oflymphocytes, leucocytes, and macrophages that were even moresevere in Usp28�/� compared with Usp28þ/þ mice (Fig. 6A).

Glutamine synthetase (GS) is a well-known indicator for anintact liver architecture as it is expressed only in the last one or twocell layers around the central vein. The analysis of GS expressionrevealed that the normal perivenous zonation was not affected bythe lack of USP28. However, DEN injection caused a loss of thisspecific GS pattern and cells spotted throughout the entire acinusbecameGS-positive; this scattering was enhanced in theUsp28�/�

mice (Fig. 6A). Furthermore, glutathione-S-transferase pi (GST-P)is known to be a tumormarker (24) and comparedwith untreatedanimals we could observe GST-P expression in DEN-injectedUsp28þ/þ mice. This expression was much more pronounced inUsp28�/� animals (Fig. 6A and B). Again, we tested whether thetumor vascularization in Usp28�/� animals is stronger than inUsp28þ/þ mice and stained liver tissue for endomucin, which isspecifically expressed in vascular endothelial cells. In livers ofUsp28�/� animals, the expression and distribution of endomucinwas very strong compared withUsp28þ/þmice (Fig. 6A–C). Thesedata are in linewith the xenograft experiments andCAMassay andagain support that lack of USP28 contributes to angiogenesis.Furthermore, recent evidence indicated that USP28 contributes tothe regulationof p53via 53BP1,which are important regulators ofthe cell cycle (14, 15, 25). To obtain more mechanistically insightwe checked the levels of p53 and 53BP1 in theUsp28�/�mice andUSP28-deficient cells. Importantly, we could not detect a differ-ence in the levels of both, p53 and 53BP1 in mice and cells,respectively (Supplementary Fig. S2A–S2D). Altogether, the datafrom these experiments show that the absence ofUSP28promotesliver carcinogenesis.

DiscussionIn this study, we investigated the carcinogenic potential of

USP28 in a comprehensive manner in patients, animal models,

and cell cultures. Thereby, the data reveal three entirely newaspects with respect to the role of USP28 in cancer. First, thefindings show a positive correlation between USP28 expressionand survival inpatientswithductal breast carcinomas. Second, thedeficiency of USP28 in breast cancer cells enhances conversiontoward a more aggressive phenotype by promoting EMT, prolif-eration, migration, angiogenesis, and decreased adhesion. Third,lack of USP28 results in advanced tumor development in amousexenograft model as well as in a chemically induced liver cancermouse model.

Although few studies have pointed to changes in USP28expression in human tumors due to genetic alterations includingpoint mutations, deletions, and amplifications, no data for breastcancerwere available yet. Thus, the present investigation is thefirstshowing a correlation of USP28 expression and survival in breastcancer. Other examples where USP28 is overexpressed are colo-rectal, ovarian and bladder carcinomas, whereas USP28 proteinexpression is downregulated in prostate carcinomas and mela-nomas due to deletions at the USP28 locus, suggesting that loss ofUSP28 may promote tumorigenesis (26–28).

The association of USP28 with a better survival fits also withdata from several other studies where the USP28 interactionpartners 53BP1 and Chk2 were analyzed in breast cancer. Thesestudies showed that 53BP1 is underexpressed in most triple-negative breast cancers (29). In addition, Chk2 loss of functionwas found to be associated with breast cancer as well as withvarious other cancers (30, 31). Interestingly, and in line with thecurrent study, the closest relative of USP28, USP25, was alsofound to be overexpressed in breast cancer tissue, but no correla-tions with other parameters such as survival were investigated(32).

The findings from the patient cohort suggesting that USP28expression in breast cancer favors a more positive prognosis andoutcome from the disease is further supported by the experimentswith USP28-kd cells as well as the xenograft and knockout mouseexperiments. A feature having a central role in cancer progressionandmetastasis is EMT. The analysis of the USP28-kd cells showedthat these cells appear to be more prone to EMT compared withthe USP28-sc cells. The described morphologic alterations andreduced adherent ability are accompanied with reorganization ofthe cytoskeleton, connections to neighboring cells or the base-ment membrane as well as other motility properties. Thereby thehighly conserved trans-membrane protein E-cadherin is essentialfor cell adhesion and maintenance of cell–cell contacts (33).During EMT E-cadherin is cleaved at the plasma membrane andsubsequently degraded resulting in disruption of the cell–cellconnection (34, 35). This process goes in linewith theobservationof a reduced E-cadherin level in the USP28 knockdown cells. Theobservation that USP28-kd cells undergo EMT is further sup-ported by an increased expression of the myofibroblast markera-SMA, asa-SMA is known to be expressed infibroblasts that arisefrom endothelial cells by EMT (36).

Importantly, hypoxia and activation of HIF-1a as well asenhanced expression of Myc are important triggers and modula-tors of EMT and known to benefit tumorigenesis (37, 38). Fur-thermore, HIF-1a is a bona fide driver of the angiogenic process.Interestingly, our data from the xenograftmodel revealed a highertumor number, volume, and weight as well as expression of theangiogenic marker endomucin in tumors originating fromUSP28-kd cells. The CAM assay also supports the enhancedangiogenic properties of the USP28-kd cells. Importantly, and in






Figure 6.

USP28 deficiency enhances carcinogenesis and angiogenesis in theDEN-induced liver cancermousemodel.A,Histologic images of HE, USP28, glutamine synthetase(GS), gluthathione-S transferase pi (GST-P), and endomucin-stained 4-mm sections of paraffin-embedded livers of untreated and DEN-injected 12-month-oldUsp28þ/þ and Usp28�/� mice (scale bar, 50 mm). Quantification B, of GST-P and C, endomucin levels in immunohistologic liver samples from untreated andDEN-injected Usp28þ/þ and Usp28�/� mice (n ¼ 5). Values from untreated Usp28þ/þ mice were set to 1. The data are presented as mean � SE; �, significantdifference (P < 0.05).


Richter et al.




line with earlier studies (8, 39), the USP28-kd cells showed thatthe corresponding HIF-1a and Myc protein levels were decreasedin USP28-kd cells under both normoxia and hypoxia whencompared with cells expressing USP28. These findings, togetherwith EMT, point toward a HIF-1a or c-Myc–independent tumor-promoting effect of the USP28 knockdown.

To extend our investigations beyondbreast cancer, we also useda mouse model for hepatocellular carcinoma (HCC). HCC is thefifth most common cancer in men and the ninth common inwomen (American Cancer Society, 2015) and in addition tochronic hepatitis B and C virus infections and nonalcoholic fattyliver disease (NAFLD) causatively linkedwith chemically induceddamage (e.g., amines, ethanol, aflatoxin). The latter process isrepresented in thewell characterizedDEN-liver cancermodel usedhere. Similar to breast cancer, deficiency of Usp28 confirmed ahigher ability for tumor growth, tumor vascularization, andtumor progression in the DEN model.

The tumor suppressor p53pathway is ultimately linked toDNAdamage regulation and in addition to p53, USP28 was shown toact as a stabilizer of Chk2 and 53BP1 in response to DNA doublestrand breaks, thus influencing the Chk2–p53–PUMA pathway(16). Further studies underline the role for USP28 in the p53surveillance pathway, where it initiates together with 53BP1 p53-mediated growth arrest (14, 15, 25). These data, together with theearlier findings showing that USP28 stabilizes the checkpointarrest protein claspin, and the tumor suppressor Chk2 (16)support the data of our study in which USP28 deficiency con-tributed to an increased tumor formation. However, in this and inanother study (10), no changes in p53 and p53BP1were observedupondeficiency ofUsp28.While the reason is currently unknown,these data already indicate another level of complexity which isbeyond the scope of this study. However, it is tempting tospeculate that either p53 independent, tissue-specific effects orthe proposed mutual interplay with Fbw7 and its substrates (40)may contribute to the regulation of several key players amongthem HIFs, p53, and c-Myc in the one or other direction. Furtheranalysis, for example identification ofUSP28-modifying proteins,may shed light on this. Even more, these analyses could help tounravel also pathways unrelated to cancer but involved in otherphysiologic adaptation mechanisms. Indeed, a study appearingduring revision of thismanuscript describes a dual role for USP28in coordinating the p53 and GATA4 branches of the senescenceprogram (41). The data of the authors also support the tumorsuppressive role of USP28. Similar to this study, they analyzed theCancer Genome Atlas and found that the P value for USP28deletion significance is 1.1� 10�88 with a q-value of 2.6� 10�86.

Analysis of mutational patterns in the somatic mutation databaserevealed that USP28 has only 27% benign mutations, 47% high-functional-impact missensemutations, and 22%nonsensemuta-tions, which is higher than average. Their further analysis with atumor suppressor prediction algorithm (TUSON Explorer),ranked USP28 as the 83rd strongest tumor suppressor candidateout of 18,680 genes, with a P value of 9.6� 10�06 and a q-value of2.1 � 10�03 (41).

Altogether, the obtained data from "mice and men" currentlysupport the view that USP28 can be involved in tumor suppres-sion and has the potential to be used as a prognostic marker.

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

Authors' ContributionsConception and design: T. KietzmannDevelopment of methodology: K. Richter, E.Y. Dimova, T. KietzmannAcquisition of data (provided animals, acquired and managed patients,provided facilities, etc.): K. Richter, T. Paakkola, A. Konzack, H. Ali-Kippari,N. Kozlova, P. Koivunen, K.-M. Haapasaari, Z. Szabo, R. Kerkel€a, T. KietzmannAnalysis and interpretation of data (e.g., statistical analysis, biostatistics,computational analysis): K. Richter, T. Paakkola, K. Kubaichuk, A. Konzack,H. Ali-Kippari, K.-M. Haapasaari, H.-R. Teppo, R. Bloigu, T. KietzmannWriting, review, and/or revision of the manuscript: K. Richter, T. Paakkola,D. Mennerich, A. Konzack, P. Koivunen, K.-M. Haapasaari, R. Kerkel€a,T. KietzmannAdministrative, technical, or material support (i.e., reporting or organizingdata, constructingdatabases):H.Ali-Kippari, A. Jukkola-Vuorinen,H.-R.Teppo,E.Y. Dimova, T. KietzmannStudy supervision: R. Kerkel€a, T. Kietzmann

AcknowledgmentsThis work was supported by the Academy of Finland SA296027, the Jane

and Aatos Erkko Foundation, the University of Oulu, and Biocenter Oulu (toT. Kietzmann).

The authors are grateful to the transgenic animal core facility (BiocenterOulu) for the generation of USP28-deficient animals. Further, we would like tothank Dr. Ilya Skovorodkin for the introduction to the CAM assay. In addition,we want to express our gratitude for skillful technical assistance to Lea Boten,Juliana Peters, Jonas B€ohm, and Lea Cleve. Funding was provided by theAcademy of Finland, the Jane and Aatos Erkko Foundation, the University ofOulu, and Biocenter Oulu.

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 August 21, 2017; revisedNovember 14, 2017; accepted February 21,2018; published first March 15, 2018.

References1. Nijman SM, Luna-VargasMP, Velds A, Brummelkamp TR, Dirac AM, Sixma

TK, et al. A genomic and functional inventory of deubiquitinating enzymes.Cell 2005;123:773–86.

2. Ambroggio XI, Rees DC, Deshaies RJ. JAMM: A metalloprotease-like zinc site in the proteasome and signalosome. PLoS Biol 2004;2:E2.

3. Kim J, Kim WJ, Liu Z, Loda MF, Freeman MR. The ubiquitin-specificprotease USP2a enhances tumor progression by targeting cyclin A1 inbladder cancer. Cell Cycle 2012;11:1123–30.

4. Zhong B, LiuX,WangX,Chang SH, LiuX,WangA, et al.Negative regulationof IL-17-mediated signaling and inflammation by the ubiquitin-specificprotease USP25. Nat Immunol 2012;13:1110–7.

5. Zhong B, Liu X, Wang X, Liu X, Li H, Darnay BG, et al. Ubiquitin-specificprotease 25 regulates TLR4-dependent innate immune responses throughdeubiquitination of the adaptor protein TRAF3. Sci Signal 2013;6:ra35.

6. Bos R, van der Groep P, Greijer AE, Shvarts A, Meijer S, Pinedo HM, et al.Levels of hypoxia-inducible factor-1alpha independently predict prognosisin patients with lymph node negative breast carcinoma. Cancer 2003;97:1573–81.

7. Liu ZJ, Semenza GL, Zhang HF. Hypoxia-inducible factor 1 and breastcancer metastasis. J Zhejiang Univ Sci B 2015;16:32–43.

8. Fl€ugelD,G€orlach A, Kietzmann T.GSK-3ß regulates cell growth,migration,and angiogenesis via Fbw7 and USP28-dependent degradation of HIF-1a.Blood 2012;119:1292–301.






9. Popov N, Wanzel M, Madiredjo M, Zhang D, Beijersbergen R, Bernards R,et al. The ubiquitin-specific protease USP28 is required for MYC stability.Nat Cell Biol 2007;9:765–74.

10. Diefenbacher ME, Popov N, Blake SM, Sch€ulein-V€olk C, Nye E,Spencer-Dene B, et al. The deubiquitinase USP28 controls intestinalhomeostasis and promotes colorectal cancer. J Clin Invest 2014;124:3407–18.

11. Diefenbacher ME, Chakraborty A, Blake SM, Mitter R, Popov N, Eilers M,et al. Usp28 counteracts Fbw7 in intestinal homeostasis and cancer. CancerRes 2015;75:1181–6.

12. GuoG, Xu Y, GongM, Cao Y, An R. USP28 is a potential prognostic markerfor bladder cancer. Tumour Biol 2014;35:4017–22.

13. 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.

14. FongCS,MazoG,Das T,Goodman J, KimM,O'Rourke BP, et al. 53BP1 andUSP28mediate p53- dependent cell cycle arrest in response to centrosomeloss and prolonged mitosis. eLife 2016;5:e16270.

15. Meitinger F, Anzola JV, Kaulich M, Richardson A, Stender JD, Benner C,et al. 53BP1 and USP28 mediate p53 activation and G1 arrest aftercentrosome loss or extended mitotic duration. J Cell Biol 2016;214:155–66.

16. Zhang D, Zaugg K, Mak TW, Elledge SJ. A role for the deubiquitinatingenzyme USP28 in control of the DNA-damage response. Cell 2006;126:529–42.

17. Peurala E, Koivunen P, Haapasaari KM, Bloigu R, Jukkola-Vuorinen A. Theprognostic significance and value of cyclin D1, CDK4 and p16 in humanbreast cancer. Breast Cancer Res 2013;15:R5.

18. Kietzmann T, Roth U, Jungermann K. Induction of the plasminogenactivator inhibitor-1 gene expression by mild hypoxia via a hypoxiaresponse element binding the hypoxia inducible factor-1 in rat hepato-cytes. Blood 1999;94:4177–85.

19. Schmittgen TD, Livak KJ. Analyzing real-time PCR data by the comparativeC(T) method. Nat Protoc 2008;3:1101–8.

20. Lenart J, Dombrowski F, Gorlach A, Kietzmann T. Deficiency ofmanganesesuperoxide dismutase in hepatocytes disrupts zonated gene expression inmouse liver. Arch Biochem Biophys 2007;462:238–44.

21. Leenders MWH, Nijkamp MW, Borel Rinkes IHM. Mouse models in livercancer research: a review of current literature. World J Gastroenterol2008;14:6915–23.

22. Newell P, Villanueva A, Friedman SL, Koike K, Llovet JM. Experimentalmodels of hepatocellular carcinoma. J Hepatol 2008;48:858–79.

23. Bakiri L, Wagner EF. Mouse models for liver cancer. Mol Oncol 2013;7:206–23.

24. Simic T, Savic-Radojevic A, Pljesa-Ercegovac M, Matic M, Mimic-Oka J.Glutathione S-transferases in kidney and urinary bladder tumors. Nat RevUrol 2009;6:281–9.

25. Lambrus BG,Daggubati V, Uetake Y, Scott PM,Clutario KM, Sluder G, et al.A USP28-53BP1-p53-p21 signaling axis arrests growth after centrosomeloss or prolonged mitosis. J Cell Biol 2016;214:143–53.

26. Uhlen M, Oksvold P, Fagerberg L, Lundberg E, Jonasson K, Forsberg M,et al. Towards a knowledge-based human protein atlas. Nat Biotechnol2010;28:1248–50.

27. Mermel CH, Schumacher SE, Hill B, Meyerson ML, Beroukhim R, Getz G.GISTIC2.0 facilitates sensitive and confident localization of the targets offocal somatic copy-number alteration in human cancers. Genome Biol2011;12:R41.

28. Forbes SA, Beare D, Gunasekaran P, Leung K, Bindal N, Boutselakis H, et al.COSMIC: Exploring the world's knowledge of somatic mutations inhuman cancer. Nucleic Acids Res 2015;43:D805–11.

29. BouwmanP, Aly A, Escandell JM, PieterseM, Bartkova J, VanDerGuldenH,et al. 53BP1 loss rescues BRCA1 deficiency and is associated with triple-negative and BRCA-mutated breast cancers. Nat Struct Mol Biol 2010;17:688–95.

30. Meijers-Heijboer H, Van denOuweland A, Klijn J,WasielewskiM, De ShooA, Oldenburg R, et al. Low-penetrance susceptibility to breast cancer due toCHEK2�1100delC in noncarriers of BRCA1 or BRCA2 mutations: TheCHEK2-breast cancer consortium. Nat Genet 2002;31:55–9.

31. Offit K, Garber J. Time to checkCHEK2 in families with breast cancer? J ClinOncol 2008;26:519–20.

32. Deng S, Zhou H, Xiong R, Lu Y, Yan D, Xing T, et al. Over-expression ofgenes and proteins of ubiquitin specific peptidases (USPs) and proteasomesubunits (PSs) in breast cancer tissue observed by the methods of RFDD-PCR and proteomics. Breast Cancer Res Treat 2007;104:21–30.

33. Meng W, Takeichi M. Adherens junction: molecular architecture andregulation. Cold Spring Harb Perspect Biol 2009;1:a002899.

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

35. Niehrs C. The complex world ofWNT receptor signalling. Nat RevMol CellBiol 2012;13:767–79.

36. Kim KK, Kugler MC, Wolters PJ, Robillard L, Galvez MG, Brumwell AM,et al. Alveolar epithelial cell mesenchymal transition develops in vivoduring pulmonaryfibrosis and is regulated by the extracellularmatrix. ProcNatl Acad Sci U S A 2006;103:13180–5.

37. Cho KB, Cho MK, Lee WY, Kang KW. Overexpression of c-myc inducesepithelialmesenchymal transition inmammary epithelial cells. Cancer Lett2010;293:230–9.

38. Zhang W, Shi X, Peng Y, Wu M, Zhang P, Xie R, et al. HIF-1a promotesepithelial-mesenchymal transition and metastasis through direct regula-tion of ZEB1 in colorectal cancer. PLoS One 2015;10:e0129603.

39. Popov N, Herold S, Llamazares M, Schulein C, Eilers M. Fbw7 and Usp28regulate myc protein stability in response to DNA damage. Cell Cycle2007;6:2327–31.

40. Sch€ulein-V€olk C, Wolf E, Zhu J, Xu W, Taranets L, Hellmann A, et al. Dualregulation of Fbw7 function and oncogenic transformation by Usp28. CellRep 2014;9:1099–109.

41. Mazzucco AE, Smogorzewska A, Kang C, Luo J, Schlabach MR, Xu Q, et al.Genetic interrogation of replicative senescence uncovers a dual role forUSP28 in coordinating the p53 and GATA4 branches of the senescenceprogram. Genes Dev 2017;31:1933–8.


Richter et al.




Correction

Correction: USP28 Deficiency PromotesBreast and Liver Carcinogenesis as well asTumor Angiogenesis in a HIF-independentManner

In the original version of this article (1), the name of the sixth author, Heidi Ali-Kippari, is incorrect. The name has been corrected in the latest online HTML and PDFversions of the article. The authors regret this error.

Reference1. Richter K, Paakkola T, Mennerich D, Kubaichuk K, Konzack A, Ali-Kippari H, et al. USP28

deficiency promotes breast and liver carcinogenesis as well as tumor angiogenesis in a HIF-independent manner. Mol Cancer Res 2018;16:1000–12.

Published first November 1, 2018.doi: 10.1158/1541-7786.MCR-18-0980�2018 American Association for Cancer Research.

MolecularCancerResearch

www.aacrjournals.org 1815

http://crossmark.crossref.org/dialog/?doi=10.1158/1541-7786.MCR-18-0980&domain=pdf&date_stamp=2018-10-16

2018;16:1000-1012. Published OnlineFirst March 15, 2018.Mol Cancer Res Kati Richter, Teija Paakkola, Daniela Mennerich, et al. well as Tumor Angiogenesis in a HIF-independent MannerUSP28 Deficiency Promotes Breast and Liver Carcinogenesis as

Updated version

10.1158/1541-7786.MCR-17-0452doi:

Access the most recent version of this article at:

Material

Supplementary

http://mcr.aacrjournals.org/content/suppl/2018/03/15/1541-7786.MCR-17-0452.DC1

Access the most recent supplemental material at:

Cited articles

http://mcr.aacrjournals.org/content/16/6/1000.full#ref-list-1

This article cites 41 articles, 10 of which you can access for free at:

Citing articles

http://mcr.aacrjournals.org/content/16/6/1000.full#related-urls

This article has been cited by 2 HighWire-hosted articles. Access the articles 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/16/6/1000To 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-17-0452

http://mcr.aacrjournals.org/content/suppl/2018/03/15/1541-7786.MCR-17-0452.DC1

http://mcr.aacrjournals.org/content/16/6/1000.full#ref-list-1

http://mcr.aacrjournals.org/content/16/6/1000.full#related-urls

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

mailto:[email protected]

http://mcr.aacrjournals.org/content/16/6/1000


usp28 deﬁciency promotes breast and liver carcinogenesis as … · oncogenes and tumor...

Documents