the nuclear pool of tetraspanin cd9 contributes to mitotic ... · monly, tetraspanin web (9)....

Oncogenes and Tumor Suppressors

The Nuclear Pool of Tetraspanin CD9 Contributes to MitoticProcesses in Human Breast Carcinoma

Germana Rappa, Toni M. Green, and Aurelio Lorico

AbstractTetraspanin-29 (CD9) is an integral membrane protein involved in several fundamental cell processes and in

cancermetastasis. Here, characterization of a panel of breast cancer cells revealed a nuclear pool of CD9, not presentin normal human mammary epithelial cells. Antibody binding to surface CD9 of breast cancer cells resulted inincreased nuclear CD9 fluorescence. CD9 was also found, along with a plasma membrane–associated pool, in thenuclei of all primary ductal breast carcinoma patient specimens analyzed. In all patients, about 40%of the total CD9cellular fluorescence was nuclear. CD9 colocalized at the nuclear level with CEP97, a protein implicated incentrosome function, and with the IGSF8, an established CD9 partner in the plasma membrane. Co-immuno-precipitation of CEP97 and IGSF8 with CD9 was shown in nuclear extracts from breast cancer cells expressing aCD9–GFP fusion protein. However, by fluorescence resonance energy transfer (FRET) analysis, no direct bindingof CD9 with either protein was observed, suggesting that CD9 is part of a larger nuclear protein complex. CD9depletion or exposure of parental breast cancer cells to anti-CD9 mAb resulted in polynucleation and multipolarmitoses. These data indicate that the nuclear CD9 pool has an important role in the mitotic process.

Implications:The discovery of a nuclear pool of CD9 has prognostic and/or therapeutic potential for patients withductal carcinoma of the breast. Mol Cancer Res; 12(12); 1840–50. �2014 AACR.

IntroductionMorbidity from breast cancer largely arises from the

dissemination and growth of tumor cells at metastatic sites.Therefore, metastatic spread is the biggest concern in thetreatment of human breast cancer, as it is for most cancers.The pleiotropic molecule CD9, reportedly involved in cellfusion, adhesion, motility, proliferation, and signaling (1),has been implicated in the metastatic process, as inhibitor ofcell invasion and metastasis (2), or as prometastatic, depend-ing on the context (3). CD9 belongs to the tetraspaninfamily, which in humans contains 33 distinct proteins, allcharacterized by 4 transmembrane domains and conservedmotifs, in particular CCG and PXSC in the large extracel-lular loop (4). It is expressed in multiple cell types, includinghematopoietic, epithelial, and many cancer cells (5). Itinteracts with other tetraspanins and with other proteins,including 2 members of the immunoglobulin superfamily,IgSF8 (also called EWI-2) and EWI-F (6–9), participating to

the formation of structural and functional units calledtetraspanin-enriched microdomains (TEM) or, more com-monly, tetraspanin web (9). Within TEMs, CD9 canregulate many physiologic and pathologic processes, suchas fertilization and metastasis (3, 10, 11).Here, we report for the first time that breast cancer cells

(BCC) contain a nuclear CD9 pool and that the abro-gation of CD9 expression results in multipolar mitosesand polynucleation. The presence of nuclear CD9 inpatient-derived breast cancer material warrants futurestudies to assess the prognostic and therapeutic relevanceof nuclear CD9.

Materials and MethodsCell linesHuman MDA-MB-231 (MDA) and MCF-7 BCC were

obtained from the ATCC. The human MA-11 BCC line,established from bone marrow micrometastases of a patientwith breast cancer (12, 13), was obtained by Dr. Fodstad,Norwegian Radium Hospital (Oslo, Norway). All cell lineswere stored in aliquots in liquid nitrogen and kept in culturefor less than 3months after resuscitation. All cells were testedfor mycoplasma contamination every 6 months byPCR and DAPI staining and authenticated by morphologycheck every 2 weeks. Complete culture medium consisted ofRPMI-1640 (Gibco), 10% FBS (Atlanta Biologicals, Inc.),100 units/mL penicillin, 100 mg/mL streptomycin, and2 mmol/L L-glutamine.

Cancer Research Center, Roseman University of Health Sciences, LasVegas, Nevada.

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

Corresponding Author: Aurelio Lorico, Cancer Research Center, Rose-man University of Health Sciences, Las Vegas, NV 89135. Phone: 702-8225395; Fax: 702-9685949; E-mail [email protected]

doi: 10.1158/1541-7786.MCR-14-0242

�2014 American Association for Cancer Research.

MolecularCancer

Research

Mol Cancer Res; 12(12) December 20141840

on October 12, 2020. © 2014 American Association for Cancer Research. mcr.aacrjournals.org Downloaded from

Published OnlineFirst August 7, 2014; DOI: 10.1158/1541-7786.MCR-14-0242

http://mcr.aacrjournals.org/

AntibodiesThe following antibodies (Ab) were used throughout this

study: anti-CD9 clone P1/33/2 (Santa Cruz; www.scbt.com),anti-CD9 clone MEM-61 (Abcam; www.abcam.com), anti-CD9 clone M-L13 (BD Biosciences; www.bdbiosciences.com); anti-CD9 clone HI9a (Biolegend; www.biolegend.com); anti-CD9 Ab clone 72F6 (ThermoScientific);anti-CD81 cat. # 349502 (Biolegend); anti–ADAM-10cat. # sc-25578 (Santa Cruz); anti–b-actin (cat. # 47778(Santa Cruz); anti-Cep97 cat. # sc-100028 (Santa Cruz);anti-IgSF8 cat. # sc-103561 (Santa Cruz); and anti–pan-cytokeratin Ab cat # sc-15367 (Santa Cruz). For inhibi-tion experiments, sodium azide was removed by desaltingthrough Sephadex G-25 spin columns.

Western blottingCells were lysed with 50 mmol/L Tris-HCl, pH 7.5,

þ 100 mmol/L NaCl þ 1% Triton X-100 and incubatedon ice for 30minutes. The lysates were centrifuged at 200� gfor 5 minutes to remove unlysed cells and nucleic acids, thenthe supernatants were loaded onto a 4% to 12% Bis–Trisprecast gel (Invitrogen), along with a prestained proteinmolecular weight ladder (Genetex). The gel was transferredto a nitrocellulose membrane and afterward incubated in ablocking buffer consisting of PBS þ 1% BSA overnight at4�C. The membrane was probed with a primary Ab over-night at 4�C. The membrane was washed 3 times for 10minutes each using PBS þ 0.1% Tween 20 and afterwardincubated with a secondary Ab (Licor) for 30 minutes atroom temperature (RT). Themembrane was washed again 3times for 10 minutes each in PBS þ 0.1% Tween 20 andthen rinsed in ddH2O before viewing by an Odyssey CLxsystem (Licor).

Retroviral and lentiviral vectorsTo inhibit the expression of CD9, CD9 shRNA lenti-

viral particles (Santa Cruz, sc-35032-V, www.scbt.com)were used. The particles consisted of a pool of transduc-tion-ready lentiviral vectors containing 3 target-specificconstructs encoding 19 to 25 nt (plus hairpin) shRNAdesigned to knockdown gene expression. shRNA lentiviralparticles (sc-108080), containing an shRNA constructencoding a scrambled sequence that will not lead to thespecific degradation of any known cellular mRNA, wereused as control. For transduction, retroviral supernatantsor transduction-ready lentiviral particles were preloadedonto recombinant fibronectin (Retronectin)-coated platesand centrifuged at 950 � g for 30 minutes at 4�C. Theoperation was repeated a second time with fresh super-natant. The supernatant was then removed and the plateswashedwith PBS before addition of cells. After transduction,stable cell lines were isolated via selection with 2 mg/mLpuromycin. A few days later, cells were cloned by limitingdilution.

Transfection of CD9-GFP plasmidMDA, MA-11, and MCF-7 cells were electroporated

with 10 mg of CD9 cDNA PS100010 plasmid (ORF with

C-terminal GFP tag; OriGene Tech.), using a Gene PulserX-cell electroporator (Bio-Rad). CD9-GFP–positive cellswere selected by 400 mg/mL G-418. G-418 was removedfrom the medium at least 1 week before the experiments.

Confocal microscopyIndicated cell lines (100,000) were plated into 35-mm

microscopy dishes with 0.17-mm coverslips on the bot-tom (MatTek) and incubated overnight at 37�C. Forexperiments involving translocation of CD9 to the nucle-us, dishes were incubated with 25 mg/mL anti-CD9 Ab(clone HI9a; Biolegend), a 1:10 dilution of phycoerythrin(PE)-conjugated anti-CD44 Ab (clone G44–26; BD Phar-mingen) in serum-free media, or serum-free media alonefor 30 to 45 minutes at 4�C. Where required, dishes werewashed and stained with a 1:100 dilution of PE-conju-gated anti-mouse IgG secondary Ab (Rockland Immuno-chemicals) 1:50 dilution of Cy5-conjugated anti-mouseIgG secondary Ab (Jackson ImmunoResearch) in serum-free media for 30 minutes at 4�C. Afterwards, the cellswere washed again, then suspended in complete RPMImedium supplemented with 10% FBS, and incubated at37�C for 1.5 hours. The dishes were washed twice withPBS, then fixed in 4% paraformaldehyde for 20 minutes at4�C and, after further washings, stained with 1 mg/mL40,6-diamidino-2-phenylindole (DAPI; Sigma) in PBS for2 minutes at room temperature. The excess DAPI wasremoved and the cells were washed twice more with PBSand then resuspended in PBS before viewing. For experi-ments in which CD9 did not translocate to the nucleus,cells were washed with PBS and then fixed in 4% para-formaldehyde (PFA) as described. Cells were then per-meabilized in 0.2% Tween 20 for 15 minutes at roomtemperature, washed in permeabilization buffer, and incu-bated with 25 mg/mL anti-CD9 Ab (Biolegend) in per-meabilization buffer for 1 hour at room temperature. Cellswere then incubated with a 1:100 dilution of PE-conju-gated anti-mouse IgG secondary Ab (Rockland Immuno-chemicals) or a 1:50 dilution of Cy5-conjugated anti-mouse IgG secondary Ab (Jackson ImmunoResearch) for30 minutes at room temperature. Cells were stained withDAPI as described and washed in PBS before viewing. Inall experiments, cells were also stained with secondary Abalone as a control. Cells were imaged using confocal laserscanning microscopy on a Nikon A1R or A1Rþ using agalvano scanner and a 40� or 60� Apo-TIRF oil immer-sion objective with a numerical aperture of 1.3 or 1.49,respectively, at 512 � 512 or 1,024 � 1,024 pixelresolution. To excite DAPI, GFP/FITC, TRITC/PE, andCy5, 405 nm, 488 nm, a 561-nm solid-state, and 638 nmlasers were used, respectively. DAPI, FITC, PE, and Cy5fluorescence emissions were collected using 425–475,500–550, 570–620, and 662–737 nm long-pass filters,respectively, and recorded using NIS Elements software(Nikon). GFP/FITC and TRITC/PE emissions werecollected using higher sensitivity GaAsP detectors on theA1Rþ microscope. Raw images were processed usingImageJ or Fiji.

Nuclear CD9 in Breast Carcinoma

www.aacrjournals.org Mol Cancer Res; 12(12) December 2014 1841




Live cell staining for CD9 partnersA total of 10 mg/mL anti-Cep97 Ab (Santa Cruz Bio-

technologies) or 5 mg/mL anti-IgSF8 Ab (Santa Cruz Bio-technologies) in 20 mmol/L HEPES buffer (Gibco) wascombined with a 1:20 dilution of PE-conjugated anti-goatIgG secondary Ab (Rockland Immunochemicals) in thepresence of 40 mg/mL Lipofectamine (Invitrogen) to favorAb penetration (14) and incubated at room temperature for15 minutes. Cells were washed in serum-free media, theantibodies/Lipofectamine complex was diluted in serum-free media, and the cells incubated at 37�C for 4 hours, thenthe media were exchanged for complete growth media andimmediately viewed using confocal laser scanning micros-copy. At least 5 different x–y positions were collected for eachdish.

Fluorescence resonance energy transfer andcolocalizationMDA-CD9-GFP cells plated in 35-mm dishes with

0.17-mm coverslips on the bottom were fixed with 4%PFA for 20 minutes at 4�C. The cells were permeabilizedwith 0.2% Tween 20 in PBS for 15 to 30 minutes atroom temperature and then stained with 4 mg/mL anti-Cep97 or anti-IgSF8 Abs for 1 hour. The cells werewashed and stained with a 1:100 dilution of TRITC-conjugated anti-goat secondary Ab (Sigma) for 30 min-utes. Cells stained with TRITC-conjugated anti-goatsecondary Ab alone were used as a control. The cellswere washed with PBS before viewing on a Nikon A1Rþconfocal microscope. GFP donor and TRITC acceptorchannels were collected using 500–550 and 570–620 nmlong-pass filters, respectively. Z-stacks were acquiredbefore FRET imaging for colocalization analysis. TheColoc 2 plugin of Fiji was used to determine Pearsoncoefficients of regions of interest (ROI) within back-ground-subtracted images. A 2-tailed Student t testassuming equal variance was used to compare Pearsoncoefficients between samples. Values are reported asaverages � SEM. Fluorescence resonance energy transfer(FRET) was performed using the acceptor photo-bleach-ing method (15). Donor and acceptor channels wererecorded every 11 seconds for 576 seconds total. Betweenimage acquisition points, the acceptor was bleached for 8seconds within specific ROIs using the 561 nm laser at80% power and 12.1 pixel dwell settings.

ImmunoprecipitationNuclear and cytoplasmic fractions of MDA-CD9-GFP

and MA-11-CD9-GFP cells, with and without anti-CD9monoclonal antibody (mAb) treatment, were isolatedusing NE-PER nuclear and cytoplasmic extractionreagents (ThermoScientific) in the presence of 1� haltprotease inhibitor cocktail (ThermoScientific) to preventprotein degradation. Isolated cytoplasmic and nuclearfractions were added to 20 mL pre-equilibrated GFP-Trapmagnetic particles bound to a GFP-binding protein(ref. 16; Chromotek), incubated overnight at 4�C withconstant rotation, and subsequently eluted with TruSep

SDS sample buffer (NuSep) supplemented with 5%b-mercaptoethanol.

ImmunohistochemistryAbout 4-mmbreast cancer tissue slices from 5 patients with

ductal carcinoma of the breast were deparaffinized and rehy-drated. Antigen retrieval was performed using 10 mmol/Lsodium citrate, pH 6.0, 0.05% Tween 20 for 20 minutes ina 95�C water bath. The slides were washed with PBSsupplemented with 0.1%Tween 20 (PBST) before blockingfor 2 hours using PBS supplemented with 10% FBS. Theslides were incubated with primary antibodies [20 mg/mLmouse anti-CD9 Ab, clone MEM-61 (Abcam), and 1:50rabbit anti–pan-cytokeratin Ab] overnight at 4�C.The slideswere washed with PBST, then incubated with 1:50 dilutionsof Cy5-conjugated anti-mouse (Jackson ImmunoResearch)and 1:250 dilution FITC-conjugated or 1:200 dilutionTRITC-conjugated anti-rabbit secondary antibodies (Sou-thernBiotech or Abcam, respectively) for 1.5 hours at roomtemperature. Slides were washed with PBS and then stainedwith 1 mg/mL DAPI for 15 min at room temperature. Anti-fade (Electron Microscopy Sciences) was added to slides,whichwere thenmountedwith coverslips and viewed using a60� oil-immersion Apo-TIRF objective (NA, 1.39) on anA1Rþ confocal microscope.

Figure 1. CD9 is present in BCC nuclei. BCC, fixed and permeabilized,were sequentially incubated with anti-CD9 Ab (Biolegend, clone H19a)and Cy5-conjugated anti-mouse IgG secondary Ab. CD9 positivity wasobserved in the nuclei of all BCC lines (arrows). Cells were imaged usingconfocal laser scanning microscopy on a Nikon A1Rþ. Images arerepresentative slices taken from z-stacks using a 0.5 to 0.7 mm step size.Scale bars for images on left represent 50 mm. Scale bars for images onright represent 25 mm. Magenta, CD9. Blue, DAPI.

Rappa et al.

Mol Cancer Res; 12(12) December 2014 Molecular Cancer Research1842




Statistical analysisStatistical significance was determined using a 2-tailed

Student t test, and P < 0.05 was considered statisticallysignificant by ANOVA.

ResultsNuclear localization of CD9 in BCCWe investigated by confocal laser microscopy the cellular

localization of CD9 in fixed and permeabilized MDA,MA-11, and MCF-7 cells. In all 3 BCC lines, in additionto the main plasma membrane/cytoplasmic pool, a nuclearpool of CD9 was observed (Fig. 1). Exposure of MDA toanti-CD9 mAb (clone H19a, Biolegend) for 90 minutes at37�C resulted in a 2-fold increase in nuclear CD9 fluores-cence (Figs. 2A and 3B). No increase was observed for MA-11 andMCF-7 cells (Fig. 2A). The results were confirmed byusing the P1/33/2 anti-CD9mAb (Santa Cruz), targeted at adifferent epitope (not shown). To confirm the presence of a

nuclear CD9 pool, we stably transfectedMDA,MA-11, andMCF-7 cells with a CD9–GFP fusion plasmid. As forendogenous CD9, CD9–GFP was detected in the nucleiof MDA (Fig. 3), MA-11 and MCF-7 cells (SupplementaryFig. S1), along with a plasma membrane/cytoplasmic pool.Exposure of MDA/CD9-GFP to anti-CD9 mAb for 90minutes at 37�C resulted in a 2.8-fold increase in nuclearlevels of CD9–GFP (Fig. 3). CD9 was not detectable innuclei of human mammary epithelial cells under baselineconditions, with minimal positivity upon exposure to anti-CD9 mAb (Supplementary Fig. S2). A phycoerythrin-con-jugated secondary Ab alone, or a mAb directed against anexternal epitope of CD44, as negative controls, did not shownuclear positivity in any of the 3 BCC lines (SupplementaryFig. S3). To pinpoint the nuclear localization of CD9, wecalculated for each experimental condition the mean fluo-rescence of CD9 or DAPI within the nuclear region for eachZ-slice for at least 3 x–y positions. The means were then

MDA-MB-231

MA-11

MCF-7

MDA-MB-231 MDA-CD9-GFP

MA-11

MCF-7

Z slice number Z slice number

Z slice number

Z slice number

Rel

ativ

e fl

uo

resc

ence

Rel

ativ

e fl

uo

resc

ence

Rel

ativ

e fl

uo

resc

ence

Rel

ativ

e fl

uo

resc

ence

1

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0

1

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0

1

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0

1

0.9

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0

0 5 10 15 20 25 30 350 5 10 15 20 25 30 35

0 5 10 15 20 25 30 35

0 10 20 30 40

DAPI

CD9

CD9 (37°C)

DAPI (37°C)

DAPI

CD9

CD9 (37°C)

DAPI (37°C)

DAPI

CD9

CD9 (37°C)

DAPI (37°C)

DAPI

CD9

CD9 (37°C)

DAPI (37°C)

A B

Figure 2. Treatment of BCCwith anti-CD9mAb results in nuclear CD9 staining. A, non-fixed BCCwere sequentially incubated for 90minutes at 37�Cwith theH19a anti-CD9 Ab and thenwith Cy5-conjugated anti-mouse IgG secondary Ab, followed by fixation andDAPI staining. CD9 immunopositivity was observedin the nuclei of all BCC lines. Cells were imaged using confocal laser scanning microscopy on a Nikon A1Rþ. Images are representative slices takenfrom z-stacks using a 0.5 to 0.7 mm step size. Scale bars for images on left represent 50 mm. Scale bars for images on right represent 25 mm. Magenta, CD9.Blue, DAPI. B, exposure of BCC to anti-CD9 mAb causes a shift in nuclear localization of CD9.






divided by the maximum mean value for each Z-stack,plotted against Z-slice number and the curves baselinecorrected (Fig. 2B). All BCC lines showed overlapping CD9and DAPI fluorescence in several slices. Regardless of incu-bation with anti-CD9mAb at 37�C, all BCC lines showed 2distinct CD9 peaks, corresponding to the bottom and thetop of the nucleus. Upon mAb incubation at 37�C for 90minutes,MDA andMA-11 cells showed a shift of CD9 fromthe bottom nuclear pool toward the center of the nucleus(arrows), whereas MDA-CD9-GFP cells did not showmovement of CD9. However, in contrast to non–mAb-incubated cells, MDA-CD9-GFP cells showed a brighterpool of CD9 toward the top than toward the bottom of thenucleus (Fig. 2B, arrow). This could be due to the over-expression of CD9 in MDA-CD9-GFP cells; it is possiblethat as CD9moves up through the nucleus, the bottom poolis replenished, leading to an accumulation of more CD9 at

the top of the nucleus compared with the bottom. MCF-7cells did not present movement of CD9 fluorescence withinthe nucleus upon mAb exposure at 37�C. While these datacollectively indicate an accumulation of CD9 toward thecenter of the nucleus upon incubation with anti-CD9 mAb,the reason for differences in nuclear CD9 fluorescence shiftamong the different BCC is not clear.Presence of CD9 in BCC nuclei was confirmed by

immunoreactivity of CD9-GFP in nuclear extracts fromMDA-CD9-GFP and MA-11-CD9-GFP cells (Fig. 4A).Also, incubation with anti-CD9 Ab for 90 minutes resultedin an increase in nuclear levels of CD9 in MDA andMA-11cells. The purity of the nuclear fraction was established bythe almost exclusive association of ADAM10 with thecytoplasmic fraction of both MDA and MA-11 cells; inter-estingly, in bothMDA andMA-11 cells. Cep-97 was mainlyassociated with the nuclear fraction (Fig. 4A).

MDA-CD9-GFP

Notreatment

37°C Abtreatment

Nuclear CD9 ratios of BCC lines1*

Cell line SEM

MDA (No Ab treatment)

MDA (Ab treatment)

MDA CD9 GFP (No Ab treatment)

MDA CD9 GFP (Ab treatment)

12.4

24.1*

6.8

18.8*

1.3

3.0

2.4

1.3

A

B

Nuclear CD9/totalCD9 ratio

Figure 3. MDA cells expressingCD9-GFP present a nuclear CD9pool that increases upon exposureto anti-CD9mAb. A, treatment withanti-CD9 Ab before fixation andDAPI staining shows greaternuclear localization of CD9,compared with untreated cells.Arrows represent areas of CD9nuclear positivity. Cells wereimaged using confocal laserscanning microscopy on a NikonA1Rþ. Images are representativeslices taken from z-stacks using a0.5 to 0.7 mm step size. Scale barson left represent 50 mm. Scale barson right represent 25mm.B, nuclearCD9 ratios of BCC lines.

Rappa et al.





CD9 associates with Cep97 and IgSF8 in the nucleusTo identify binding partners of CD9 at the nuclear level,

MDA-CD9-GFP and MA-11-CD9-GFP cells (both GFP-sorted) were treated with 25 mg/mL anti-CD9 Ab (Biole-gend) in serum-free media or mock-incubated for 30 min-utes at 4�C. The cells were then washed and resuspended incomplete media and incubated at 37�C for 90 minutes.

Isolated cytoplasmic and nuclear fractions of MDA-CD9-GFP and MA-11-CD9-GFP cells, with and without anti-CD9 mAb treatment, were incubated with GFP-Trapmagnetic particles bound to a GFP-binding protein(16). As shown in Fig. 4B, both Cep97, a protein requiredfor correct centrosomal function (17), and IgSF8, a CD9-binding partner (6) were pulled down.

MDA

MDA

MA-11

MA-11

-Ab -Ab+Ab +AbC C C CN N N

C N C N

N

CD9-GFP

CD9-GFP

CEP-97

ADAM10

b-Actin

Anti-CEP-97 Merge

No Abtreatment

No Abtreatment

treatmentCD9 Ab

treatmentCD9 Ab

CD9-GFP Anti-lgSF8 Merge

Co-IP

CEP-97

lgSF8~70 kDa

A B

C

D

E

F

Figure 4. CD9 is associated withCep97 and IgSF8 inMDA nuclei. A,immunoblotting of nuclear (N) andcytoplasmic (C) lysates fromMDA-CD9-GFP and MA-11-CD9-GFP before and after exposure ofthe cells to anti-CD9 mAb for90 minutes. Single protein bandsof about 50 and 100 kDa wereobserved for CD9-GFP andCep97, respectively; 2 bands ofabout 55 and 70 kDa for IgSF8,and one band of about65 kDa for ADAM10. B,co-immunoprecipitation of Cep97and IgSF8 with CD9-GFP at thenuclear (N), but not at thecytoplasmic (C) level. C–F,colocalization experiments wereperformed by combining 10 mg/mLanti-Cep97 Ab or 5 mg/mL anti-IgSF8 Ab with a 1:20 dilution ofPE-conjugated anti-mouse IgGsecondary Ab in 20 mmol/LHEPES buffer in the presence of40 mg/mL Lipofectamine 2000.After incubating for 4 hours at37�C, cells were imaged usingconfocal laser scanningmicroscopy. D and F, anti-Cep97and anti-IgSF8 were preceded by90-minute incubation with anti-CD9 Ab to enhance nucleartranslocation of CD9.White arrowsindicate areas of colocalization.Scale bars represent 50 mm.






To confirm the nuclear association of CD9 with Cep97and IgSF8, we analyzed by confocal microscopy MDA-CD9-GFP cells stained with PE-conjugated secondary Abbound to anti-Cep97 and anti-IgSF8 mAbs. For bothproteins, a punctate pattern of distribution was observed,Cep97 being mainly localized to the nucleus (Fig. 4C) and

IgSF8 to membrane and cytoplasm (Fig. 4E). Upon pre-incubation with CD9 mAb, both antigens were localizedmainly in the nucleus and colocalized with CD9 (Fig. 4Dand F). Pearson colocalization coefficient, calculated fromthe images shown in Fig. 5, was significantly higher (P ¼0.01) for Cep97 and CD9-GFP (0.38 � 0.03) than for

CD9-GFP Anti-CEP-97 Merge

MergeAnti-lgSF8CD9-GFP

Time (s) Time (s)0 100 200 300 400 500 600 0 100 200 300 400 500 600

1

0.8

0.6

0.4

0.2

0

1

0.8

0.6

0.4

0.2

0

CE

P-9

7- T

RIT

C F

F/F

0

lgS

F8-

TR

ITC

FF

/F0

1.5

1.4

1.3

1.2

1.1

1

1.5

1.4

1.3

1.2

1.1

1

CD

9-G

FP

-F/F

0

CD

9-G

FP

-F/F

0

A

B

C D

Figure 5. Colocalization and FRET analysis of CD9 and Cep-97 or CD9 and IgSF8. A and B, colocalization of CD9-GFP and Cep97 or CD9-GFP and IgSF8.MDA-CD9-GFP cells were fixed, permeabilized, and stained with anti-Cep97 Ab (A) or anti-IgSF8 (B) followed by TRITC-conjugated anti-goat IgG secondaryAb. Images are representative confocal slices from z-stacks collected before FRET analysis. From left to right: GFP channel, TRITC channel, and bothchannels merged onto DIC image. Colocalization is observed in and around the nucleus for Cep97 and CD9 and on the plasmamembrane for IgSF8 (arrows).C andD, average relative fluorescence values for GFP and TRITC over time during sequential acceptor photobleaching FRET inmultiple ROIs for the samplesdescribed in A and B. The donor (TRITC) was sequentially bleached for 8 seconds, followed by donor (GFP) and acceptor acquisition for 576 secondstotal. TRITC fluorescence decreased over time, but there was no increase in GFP fluorescence in either Cep97- or IgSF8-stained dishes.

Rappa et al.





IgSF8 and CD9-GFP (0.29 � 0.01). To investigatewhether the binding between CD9 and Cep97 and CD9and IgSF8 was direct, we performed FRET analysis usingthe acceptor photobleach method (Fig. 5). MDA-CD9-GFP cells were stained with anti-Cep97 or anti-IgSF8antibodies, followed by a TRITC-conjugated secondaryAb. The acceptor (TRITC) was sequentially bleached,followed by donor (GFP) and acceptor acquisition for 576seconds total. Relative fluorescence values for GFP andTRITC over time are shown in Fig. 5C and D. As TRITCfluorescence decreased over time, there was no observableincrease in donor fluorescence in either Cep97- or IgSF8-stained dishes. This indicates no direct association (within10 nm) between Cep97 and CD9, nor IgSF8 and CD9.

CD9 knockdown induces multipolar mitoses andpolynucleationTo investigate whether binding of CD9 to Cep97 affected

centrosome function, we knocked down CD9 in MDA andMCF-7 cells by a pool of 3 specific shRNAs. Completedepletion of CD9 was evidenced by complete lack of mAbreactivity (Supplementary Fig. S4). A large increase inmultipolar mitoses, potential cause of heterogeneity andaneuploidy, was observed in CD9-knockdown BCC versustheir mock-transduced counterparts (Fig. 6A and B). Thus,while 2.8% and none of mitotic figures observed in parentalMDA and MCF-7, respectively, were multipolar, the per-centage increased to 15% and 4.6% for CD9-knockdowncell lines (Fig. 6D). No multipolar mitoses were observed in

25

20

15

10

5

0

16141210

86420

MDA MCF-7MDA MCF-7

Mul

tipol

ar m

itose

s (%

)

Pol

ynuc

leat

ed c

ells

(%

)

ParentalCD9 shRNAParental + anti-CD9 Ab

ParentalCD9 shRNAParental + anti-CD9 Ab

A

B C

D E

Figure 6. Disruption ofCD9 causespolynucleation andmultipolarmitoses. A andB, representative tripolarmitoses (arrows) and (C) polynucleated cells inCD9-knockdown MDA cells 2 weeks after stable transduction with a pool of transduction-ready CD9 shRNA lentiviral vectors containing 3 target-specificconstructs. shRNA lentiviral particles, containing an shRNA construct encoding a scrambled sequence that will not lead to the specific degradation of anyknown cellular mRNA, were used as control. Cells were stained with DAPI. D, percentage of multipolar mitoses/total mitoses for mock-transfected MDAand MCF-7 cells compared with cells transfected with CD9shRNA plasmid, selected for 1 week with 400 mg/mL G-418, and grown for 1 more week in theabsence of G-418. Cells were fixed and stained with DAPI and 300 mitoses per experiment were counted. Columns, means of 3 experiments; bars, SD.��, P < 0.01. E, percentage of polynucleated cells for mock-transfected MDA and MCF-7 cells compared with cells transfected with CD9shRNAplasmid, selected for 1 week with 400 mg/mL G-418, and grown for 1 more week in the absence of G-418. Columns, means of 3 experiments; bars, SD.�, P < 0.05; ��, P < 0.01. Statistical significance was determined using a 2-tailed Student t test, and P < 0.05 was considered significant by ANOVA.






MSC. Also, higher numbers of polynucleated cells wereobserved in BCC after CD9 knockdown (Fig. 6C and E).We then investigated whether the addition of anti-CD9antibodies increased CD9-dependent nuclear-related func-tions. Exposure of MDA cells to 25 mg/mL anti-CD9 mAbat 37�C resulted in a 4.2 � 0.5-fold and a 2.1 � 0.3-foldincrease in polynucleated cells and atypical mitoses, respec-tively (Fig. 6D and E).

Nuclear localization of CD9 in primary tumors of breastcarcinoma patientsTo establish the clinical relevance of our observations, we

then investigated by confocal laser microscopy the intracel-lular localization of CD9 in formalin-fixed, paraffin-embed-ded sections from primary ductal carcinomas of the breast(Supplementary Table S1). In all cases, in addition to plasmamembrane–associated positivity, a nuclear pool of CD9was observed with MEM-61 (Abcam; Fig. 7) and 72F6(Thermo) mAbs (Supplementary Fig. S5), directed againstdifferent CD9 epitopes. In agreement with previous findings(18), CD9 staining at the plasmamembrane/cytoplasm levelwas usually homogeneous, with all cells showing a similarintensity. Also, nuclear positivity was homogeneous amongpatients, as assessed by the ratio of nuclear over totalfluorescence (NF/TF; Supplementary Table S1).

DiscussionWe have herein identified by confocal microscopy and

immunoblotting a nuclear pool of the tetraspanin CD9 inMDA,MA-11, andMCF-7 cells under na€�ve conditions andupon forced expression of a CD9–GFP fusion protein.Nuclear CD9was also found in tumor sections from patientswith ductal breast carcinoma. To the best of our knowledge,this is the first nonanecdotal report of a nuclear pool of atetraspanin protein. Nuclear CD9 staining was previouslydescribed for WM9 melanoma cells, but its possible impli-cations were not discussed in that study (19).We found by confocal microscopy and co-immunopre-

cipitation that CD9 colocalizes with IgSF8 and Cep-97 atthe nuclear level. Because both CD9 and IgSF8 are consid-ered non-nuclear proteins, their nuclear association wassurprising. However, the 2 proteins reportedly (6) have astoichiometric interaction, suggesting that in MDA theymight translocate as a complex from plasma membrane andmembrane protrusions to the nucleus. The structural sim-ilarity between IgFS8 and anti-CD9 mAb, belonging to thesame Ig superfamily, suggests that, as observed for anti-CD9,binding of IgFS8 to CD9 triggers the translocation of theCD9–IgFS8 complex to the nucleus, where it binds Cep97,regulating the mitotic process. However, as FRET experi-ments showed no direct binding ofCD9 toCep-97 or IgSF8,it is conceivable that CD9 translocates to the nucleus as alarger protein complex. Both CD9 knockdown and additionof anti-CD9 mAb to parental cells resulted in multipolarmitoses and polynucleation, suggesting that CD9 is involvedin centrosomal function. This is supported by the followingobservations: (i) Cep97was presentmainly in BCCnuclei by

Patient 1

Patient 2

Patient 3

Patient 4

Patient 5

Figure 7. Nuclear expression of CD9 in patient with breast cancer. Upondeparaffination, rehydration, and antigen retrieval, 4-mm breast cancertissue slices from 5 patients with ductal carcinoma of the breast wereincubated with anti-CD9 Ab (Abcam) and anti–pan-cytokeratin Ab (SantaCruz Biotech), followed by Cy5-conjugated anti-mouse and FITC-conjugated anti-rabbit secondary antibodies and then stainedwith DAPI.Images are horizontal maximum intensity projections of z-stacks. Scalebars represent 50 mm for images on left, 10 mm for images on the right.Green, pan-cytokeratin; red, CD9; blue, DAPI.

Rappa et al.





both Western blotting and confocal laser microscopy;(ii) Cep97 is known to be required for the recruitment ofCP110, a protein that regulates centrosomal duplication(17); (iii) in BCC, CD9 knockdown resulted in multipolarmitoses and polyploidy, phenocopying the effects describedby others in BCC upon Cep97 or CP110 knockdown (17);(iv) depletion of clathrin, a protein involved in receptor-mediated endocytosis, with a biologically important nuclearpool (20), reportedly impaired mitotic spindle stability andcytokinesis (21); (v) Cp190 and CP60, other importantcentrosomal proteins, reportedly translocated from thenucleus to the centrosome during different phases of thecell cycle (22, 23). CD9 is reportedly involved in cell fusion(1), which may lead to formation of multipolar spindles andpolynucleation. However, cell fusion is not the cause of themitotic abnormalities observed upon abrogation of CD9expression because in CD9-knockdown MDA cells and inthe presence of CD9-blocking antibodies, both processes ofcell invasion and fusion are decreased (G. Rappa andcolleagues, manuscript submitted for publication).The prognostic value of CD9 expression is currently

unclear; whereas some studies found an association betweendecreased CD9 expression and poor prognosis (24–26),others found no significant correlation (18, 27). Our dis-covery of a nuclear CD9 pool associated with Cep97,together with our finding that CD9 is a determinant ofinvasiveness and cell fusion (G. Rappa and colleagues,manuscript in preparation), shed some light on the "pushand pull" role of tetraspanins in metastases (3): loss of CD9expression might predispose to tumor development throughthe formation of heterogeneous aneuploid clones induced bymitotic abnormalities, whereas re-expression of CD9 mightsubsequently favor invasiveness, heterologous fusion, andthe development of metastasis. Consistent with this hypoth-esis, while melanoma cells reportedly (19) had reduced CD9expression relative to normal melanocytes, forced increase inCD9 expression stimulated their invasiveness. Similarly, in

cervical cancers, CD9 was downregulated in primary sites,but re-expressed at sites of transendothelial invasion topromote the expansion of malignant cells (28). Our findingswarrant further studies (i) to evaluate in large cohorts ofpatients with ductal breast carcinoma whether nuclear CD9has prognostic value or can predict therapeutic response and(ii) to develop novel CD9-targeted therapeutics to specifi-cally interfere with the function of nuclear CD9 in BCC.

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

DisclaimerThe content is solely the responsibility of the authors and does not necessarily

represent the official views of the National Cancer Institute or the NIH.

Authors' ContributionsConception and design: G. Rappa, T.M. Green, A. LoricoDevelopment of methodology: G. Rappa, T.M. Green, A. LoricoAcquisition of data (provided animals, acquired and managed patients, providedfacilities, etc.): G. Rappa, T.M. Green, A. LoricoAnalysis and interpretation of data (e.g., statistical analysis, biostatistics, compu-tational analysis): G. Rappa, T.M. Green, A. LoricoWriting, review, and/or revision of the manuscript: G. Rappa, T.M. Green,A. LoricoAdministrative, technical, or material support (i.e., reporting or organizing data,constructing databases): G. Rappa, T.M. Green, A. LoricoStudy supervision: A. Lorico

AcknowledgmentsThe authors thank their Roseman University colleagues, and in particular Harry

Rosenberg, Renee Coffman, and Ronald R. Fiscus for their constant support andencouragement and Fabio Anzanello for invaluable technical assistance. The pathologyslideswere kindly provided byClaraMagyar,Department of Pathology andLaboratoryMedicine at UCLA (Los Angeles, California).

Grant SupportThis work was supported by U.S. NIH R01CA133797 (G. Rappa).The costs of publication of this article were defrayed in part by the payment of page

charges. This article must therefore be herebymarked advertisement in accordance with18 U.S.C. Section 1734 solely to indicate this fact.

Received May 5, 2014; revised June 11, 2014; accepted July 15, 2014;published OnlineFirst August 7, 2014.

References1. Hemler ME. Tetraspanin proteins mediate cellular penetration, inva-

sion, and fusion events and define a novel type of membrane micro-domain. Annu Rev Cell Dev Biol 2003;19:397–422.

2. Ikeyama S, Koyama M, Yamaoko M, Sasada R, Miyake M. Sup-pression of cell motility and metastasis by transfection with humanmotility-related protein (MRP-1/CD9) DNA. J Exp Med 1993;177:1231–7.

3. Zoller M. Tetraspanins: push and pull in suppressing and promotingmetastasis. Nat Rev Cancer 2009;9:40–55.

4. Seigneuret M, Delaguillaumie A, Lagaudriere-Gesbert C, Conjeaud H.Structure of the tetraspanin main extracellular domain. A partiallyconserved fold with a structurally variable domain insertion. J BiolChem 2001;276:40055–64.

5. Boucheix C, Duc GH, Jasmin C, Rubinstein E. Tetraspanins andmalignancy. Expert Rev Mol Med 2001;2001:1–17.

6. StippCS,Kolesnikova TV, HemlerME. EWI-2 is amajor CD9 andCD81partner and member of a novel Ig protein subfamily. J Biol Chem2001;276:40545–54.

7. Charrin S, Le Naour F, Oualid M, Billard M, Faure G, Hanash SM,et al. The major CD9 and CD81 molecular partner. Identification

and characterization of the complexes. J Biol Chem 2001;276:14329–37.

8. Berditchevski F. Complexes of tetraspanins with integrins: more thanmeets the eye. J Cell Sci 2001;114:4143–51.

9. Levy S, Shoham T. The tetraspanin webmodulates immune-signallingcomplexes. Nat Rev Immunol 2005;5:136–48.

10. Le Naour F, Rubinstein E, Jasmin C, Prenant M, Boucheix C. Severelyreduced female fertility in CD9-deficient mice. Science 2000;287:319–21.

11. Kaji K, Oda S, Shikano T, Ohnuki T, Uematsu Y, Sakagami J, et al. Thegamete fusion process is defective in eggs of CD9-deficient mice. NatGenet 2000;24:279–82.

12. Rye PD, Norum L, Olsen DR, Garman-Vik S, Kaul S, Fodstad O. Brainmetastasismodel in athymic nudemice using a novelMUC1-secretinghuman breast-cancer cell line, MA11. Int J Cancer 1996;68:682–7.

13. Rappa G, Lorico A. Phenotypic characterization of mammosphere-forming cells from the human MA-11 breast carcinoma cell line. ExpCell Res 2010;316:1576–86.

14. Weill CO, Biri S, Erbacher P. Cationic lipid-mediated intracellulardelivery of antibodies into live cells. Biotechniques 2008;44:Pvii-xi.






15. Van Munster EB, Kremers GJ, Adjobo-Hermans MJ, Gadella TW Jr.Fluorescence resonance energy transfer (FRET) measurement bygradual acceptor photobleaching. J Microsc 2005;218:253–62.

16. Rothbauer U, Zolghadr K, Muyldermans S, Schepers A, Cardoso MC,Leonhardt H. A versatile nanotrap for biochemical and functionalstudies with fluorescent fusion proteins. Mol Cell Proteomics 2008;7:282–9.

17. Spektor A, Tsang WY, Khoo D, Dynlacht BD. Cep97 and CP110suppress a cilia assembly program. Cell 2007;130:678–90.

18. Jamil F, Peston D, Shousha S. CD9 immunohistochemical staining ofbreast carcinoma: unlikely to provide useful prognostic information forroutine use. Histopathology 2001;39:572–7.

19. Fan J, ZhuGZ, Niles RM. Expression and function of CD9 inmelanomacells. Mol Carcinog 2010;49:85–93.

20. Ohmori K, Endo Y, Yoshida Y, Ohata H, Taya Y, Enari M. Monomericbut not trimeric clathrin heavy chain regulates p53-mediated transcrip-tion. Oncogene 2008;27:2215–27.

21. Foraker AB, Camus SM, Evans TM, Majeed SR, Chen CY, Taner SB,et al. Clathrin promotes centrosome integrity in early mitosisthrough stabilization of centrosomal ch-TOG. J Cell Biol 2012;198:591–605.

22. Oegema K, Whitfield WG, Alberts B. The cell cycle-dependentlocalization of the CP190 centrosomal protein is determined by the

coordinate action of two separable domains. J Cell Biol 1995;131:1261–73.

23. Oegema K, Marshall WF, Sedat JW, Alberts BM. Two proteinsthat cycle asynchronously between centrosomes and nuclearstructures: Drosophila CP60 and CP190. J Cell Sci 1997;110:1573–83.

24. Miyake M, Nakano K, Ieki Y, Adachi M, Huang CL, Itoi S, et al. Motilityrelated protein 1 (MRP-1/CD9) expression: inverse correlation withmetastases in breast cancer. Cancer Res 1995;55:4127–31.

25. Miyake M, Nakano K, Itoi SI, Koh T, Taki T. Motility-related protein-1(MRP-1/CD9) reduction as a factor of poor prognosis in breast cancer.Cancer Res 1996;56:1244–9.

26. Mimori K, Kataoka A, Yoshinaga K, Ohta M, Sagara Y, Yoshikawa Y,et al. Identification of molecular markers for metastasis-relatedgenes in primary breast cancer cells. Clin Exp Metastasis 2005;22:59–67.

27. Arihiro K, Kaneko M, Fujii S, Inai K. Loss of CD9 with expression ofCD31 and VEGF in breast carcinoma, as predictive factors of lymphnode metastasis. Breast Cancer 1998;5:131–8.

28. Sauer G, Windisch J, Kurzeder C, Heilmann V, Kreienberg R, DeisslerH. Progression of cervical carcinomas is associated with down-reg-ulation of CD9 but strong local re-expression at sites of transendothe-lial invasion. Clin Cancer Res 2003;9:6426–31.


Rappa et al.




2014;12:1840-1850. Published OnlineFirst August 7, 2014.Mol Cancer Res Germana Rappa, Toni M. Green and Aurelio Lorico Processes in Human Breast CarcinomaThe Nuclear Pool of Tetraspanin CD9 Contributes to Mitotic

Updated version

10.1158/1541-7786.MCR-14-0242doi:

Access the most recent version of this article at:

Material

Supplementary

http://mcr.aacrjournals.org/content/suppl/2014/08/16/1541-7786.MCR-14-0242.DC1

Access the most recent supplemental material at:

Cited articles

http://mcr.aacrjournals.org/content/12/12/1840.full#ref-list-1

This article cites 28 articles, 13 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/12/12/1840To 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-14-0242

http://mcr.aacrjournals.org/content/suppl/2014/08/16/1541-7786.MCR-14-0242.DC1

http://mcr.aacrjournals.org/content/12/12/1840.full#ref-list-1

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

mailto:[email protected]

http://mcr.aacrjournals.org/content/12/12/1840


the nuclear pool of tetraspanin cd9 contributes to mitotic ... · monly, tetraspanin web (9)....

Documents