myc-transformed epithelial cells down-regulate clusterin ... · terin, an extracellular...

[CANCER RESEARCH 64, 3126–3136, May 1, 2004]

Myc-Transformed Epithelial Cells Down-Regulate Clusterin, Which Inhibits TheirGrowth in Vitro and Carcinogenesis in Vivo

Andrei Thomas-Tikhonenko,1 Isabelle Viard-Leveugle,3,4 Michael Dews,1 Philippe Wehrli,3 Cinzia Sevignani,1

Duonan Yu,1 Stacey Ricci,2 Wafik el-Deiry,2 Bruce Aronow,5 Gurkan Kaya,3 Jean-Hilaire Saurat,3 andLars E. French3

Departments of 1Pathobiology and 2Medicine, University of Pennsylvania, Philadelphia, Pennsylvania; 3Louis-Jeantet Skin Cancer Lab, Department of Dermatology, GenevaUniversity Medical School, Geneva, Switzerland; 4IVL BioService, Saint-Barthelemy, Le Gua, France; and 5Division of Developmental Biology, Children’s Hospital ResearchFoundation Cincinnati, Ohio

ABSTRACT

Effective treatment of malignant carcinomas requires identification ofproteins regulating epithelial cell proliferation. To this end, we comparedgene expression profiles in murine colonocytes and their c-Myc-trans-formed counterparts, which possess enhanced proliferative potential. Asurprisingly short list of deregulated genes included the cDNA for clus-terin, an extracellular glycoprotein without a firmly established function.We had previously demonstrated that in organs such as skin, clusterinexpression is restricted to differentiating but not proliferating cell layers,suggesting a possible negative role in cell division. Indeed, its transientoverexpression in Myc-transduced colonocytes decreased cell accumula-tion. Furthermore, clusterin was down-regulated in rapidly dividinghuman keratinocytes infected with a Myc-encoding adenovirus. Its knock-down via antisense RNA in neoplastic epidermoid cells enhanced prolif-eration. Finally, recombinant human clusterin suppressed, in adose-dependent manner, DNA replication in keratinocytes and other cellsof epithelial origin. Thus, clusterin appears to be an inhibitor of epithelialcell proliferation in vitro. To determine whether it also affects neoplasticgrowth in vivo, we compared wild-type and clusterin-null mice withrespect to their sensitivity to 7, 12-dimethylbenz(a)anthracene /12-O-tetradecanoylphorbol-13-acetate (DMBA/TPA)-induced skin carcinogen-esis. We observed that the mean number of papillomas/mouse was higherin clusterin-null animals. Moreover, these papillomas did not regress asreadily as in wild-type mice and persisted beyond week 35. The rate ofprogression toward squamous cell carcinoma was not altered, althoughthose developing in clusterin-null mice were on average better differenti-ated. These data suggest that clusterin not only suppresses epithelial cellproliferation in vitro but also interferes with the promotion stage of skincarcinogenesis.

INTRODUCTION

c-Myc is an oncogene that is overexpressed in many human tumorsranging from B-cell lymphoma to colon carcinoma. However, despiteaggressive research, the molecular mechanisms leading to neoplastictransformation are incompletely understood. Myc is a member of theMyc/Max/Mad network of transcription regulators (1). It interactswith Max, binds as a heterodimer to the E-box element (2), andactivates expression of genes containing this sequence (3, 4). Geneactivation by Myc/Max is thought to occur primarily via chromatinremodeling (5–7). Besides activating gene expression, Myc is alsoknown to inhibit transcription from promoters containing the initiatorelement (8), at least in part via recruitment of the Miz-1 corepressor

(9). The identities of Myc-target genes might provide crucial clues asto its normal function and the role in cancer.

Myc affects expression of many proteins whose functions rangefrom cell metabolism to ribosome biogenesis (10, 11). However, themajority of putative Myc target genes pertain to cell proliferation.Among them are ornithine decarboxylase, the enzyme involved inDNA biosynthesis (12), cyclin A (13), and cdc25A, a phosphataserequired to activate cyclin-dependent kinases (14). More recently,cyclin-dependent kinase 4 itself was shown to be up-regulated by Myc(15), and members of the Myc family were reported to activate Id2, aninhibitor of the retinoblastoma tumor/cell cycle suppressor (16). Mycalso activates the telomerase gene (17), presumably extending the lifespan of the host cell. Furthermore, several Myc-repressed genes playroles in cell cycle control: cyclin D1 (18); assorted cyclin-dependentkinase inhibitors (19–22); gadd45 (23); and gas (24). Consistent withthese observations, activation of Myc forces quiescent fibroblasts toreenter cell cycle (25). Moreover, rodent fibroblasts with targeteddisruption of Myc are severely deficient in cell proliferation (26).Consequently, in mice [but curiously not in Drosophila (27)], de-creased expression of Myc results in hypoplasia (28). A consensus hasthus emerged that Myc functions in a cell-autonomous manner viatipping the balance between intracellular pro- and antimitogenicsignals in favor of the former.

However, recent research has led to the augmentation of thisparadigm as some Myc targets encode extracellular proteins that couldpotentially affect neighboring cells in a paracrine manner. For in-stance, we have previously demonstrated that Myc down-regulatesthrombospondin-1, a large secreted glycoprotein (29, 30). Throm-bospondin-1 and thrombospondin-2 negatively affect proliferation ofepithelial (31) and endothelial cells (32, 33), suggesting that itsdown-regulation (e.g., via Myc overexpression) could benefit thetumor in two ways: by increasing proliferation of neoplastic cells andalso by stimulating the recruitment of vascular endothelium (34).Indeed, activation of Myc results in the acquisition of the angiogenicphenotype (35–38).

We were interested in determining whether down-regulation ofsecreted glycoproteins with antiproliferative activities is a commontheme underlying the transforming function of Myc. In addition, thevast majority of gene regulation studies have thus far been performedon Myc-overexpressing fibroblasts or B-lymphoid cells (39).6 Only oflate have epithelial cell systems started being used, with some inter-esting results. For example, in a recent study (40), Myc was found todown-regulate thrombospondin-1 in rat kidney epithelial RK3E cells.To identify more targets for Myc in epithelial cells, we have estab-lished a new experimental system: p53-null murine colonocytes over-expressing Myc and thus undergoing neoplastic transformation. Mi-croarray analysis of gene expression profiles has revealed that inaddition to thrombospondin-1, Myc overexpression results in down-regulation of clusterin, a heterodimeric glycoprotein of 80 kDa with a

Received 7/2/03; revised 2/10/03; accepted 2/23/04.Grant support: National Science Foundation, Geneva and Swiss Cancer Leagues, and

Ernst & Lucie Schmidheiny and Stanley Thomas Johnson Foundations (L. French),and National Cancer Institute Grant CA 071881 and National Institute of Diabetes andDigestive and Kidney Diseases Grant DK 050306 (A. Thomas-Tikhonenko).

The costs of publication of this article were defrayed in part by the payment of pagecharges. This article must therefore be hereby marked advertisement in accordance with18 U.S.C. Section 1734 solely to indicate this fact.

Note: A. Thomas-Tikhonenko and I. Viard-Leveugle contributed equally to this work.Requests for reprints: Andrei Thomas-Tikhonenko, University of Pennsylvania, 3800

Spruce Street, M/C 6051, Philadelphia, PA 19104-6051 or Lars French, Louis-JeantetSkin Cancer Lab 5.222, Geneva University Medical Center, 1, rue Michel Servet, 1211Geneve 4, Switzerland. 6 Internet address: http://www.myccancergene.org.

3126

Research. on August 17, 2020. © 2004 American Association for Cancercancerres.aacrjournals.org Downloaded from

http://cancerres.aacrjournals.org/

homology to thrombospondin-1 type 1 repeat but without a firmlyestablished function.

Clusterin is constitutively synthesized and secreted by a widevariety of cell types in many species (41). Because of its discovery indistinct settings, clusterin bears several names, including cytolysisinhibitor, sulfated glycoprotein-2, testosterone-repressed prostatemessage-2, SP-40,40, and apolipoprotein J. Clusterin has been tenta-tively implicated in several biological processes, including cell adhe-sion and cell-extracellular matrix interaction (42–45), regulation ofthe complement cascade (46–48), lipid transport (49–51), cellularprotection, and tissue remodeling (42, 52). In this context, it has beenshown that clusterin gene expression is strongly up-regulated inresponse to thermal and oxidative stress and protects cells fromapoptotic cell death caused by these phenomena (53, 54). Thus,clusterin might function as a molecular chaperon (55). However, notall forms of clusterin exhibit antiapoptotic properties. Recently, atruncated 55 kDa form of clusterin induced by irradiation in vitro hasbeen found targeted to the nucleus where it appears to act as a deathsignal (56–58). Therefore, the exact biological function of clusterinremains to be elucidated. One interesting possibility is that clusterincould influence the choice between proliferation and differentiation.Indeed, Diemer et al. (59) have shown that clusterin gene expressioncorrelates with in vitro differentiation of aortic smooth muscle cells.In our group, we have shown that clusterin is widely expressed indeveloping epithelia during murine embryogenesis. In addition, clus-terin mRNA is selectively localized within differentiating layers oftissues such as the developing skin, tooth, and duodenum, whereproliferating and differentiating compartments are readily distinguish-able (60). For instance, the epidermis is comprised of the basal layerof proliferating keratinocytes expressing keratins 5 and 14 and thesuprabasal layers of terminally differentiating postmitotic keratino-cytes expressing keratins 1 and 10. It turned out that in the skin,clusterin production is confined to suprabasal layers, and in thedeveloping hair follicles, it is localized to the inner root sheath wherecells undergo morphogenesis and differentiation (61). Thus, we hy-pothesized that clusterin synthesized by differentiating cells is in-volved in suppression of cell proliferation either directly (as a bonafide autocrine or paracrine growth factor) or indirectly (via seques-tration of other growth factors). To test this hypothesis, the effects ofits up- and down-regulation on cell accumulation were investigated. Inaddition, we studied the effects of purified recombinant clusterin onhuman epidermal keratinocytes (HEKs) in vitro as well as the effectof clusterin germ-line inactivation on skin carcinogenesis. Our datademonstrate that clusterin indeed negatively affects epithelial cellgrowth in culture and attenuates neoplastic growth in vivo.

MATERIALS AND METHODS

Neoplastic Transformation of p53-Null Colonocytes by Myc. Primarymurine colonocytes were cultured as described previously (62). They weretransfected with either LXSN or pBabePuro retroviral DNAs encoding humanc-Myc or its 4-hydroxytamoxifen (4-OHT)-dependent version (MycER), re-spectively. Transfections were performed using the Lipofectamine Plus reagent(Invitrogen, Carlsbad, CA). G418- or puromycin-resistant cells were placed insoft agar, which in case of MycER, was supplemented with 4-OHT (Sigma, St.Louis, MO). Arising colonies were photographed 10 days later and subse-quently isolated, pooled, and expanded into mass cultures. For tumorigenicityassays, Myc- and Ki-Ras-overexpressing colonocytes were transplanted intosyngeneic C57BL6/J mice (The Jackson Laboratory, Bar Harbor, ME) eithers.c. or orthotopically into the cecal wall. Orthotopic transplantation was per-formed as described by Fidler (63). Animals were sacrificed either 4 weekslater or when appearing moribund. Upon euthanasia, they were subjected togross pathological examination.

[3H]Thymidine Incorporation and Cell Accumulation Assays. Epithe-lial cells were seeded in 96-well plates at a density of 104 cells/well, with threereplicates/sample. After 48 h, cells were fed with complete medium containing250 nM 4-OHT (for MycER-expressing cells), clusterin (for primary humancells), or vehicle alone. Twenty-four to 30 h later, 1 �Ci of [3H]thymidine (2Ci/mmol; New England Nuclear, Boston, MA) was added, and cells werecultured for an additional 12–18 h. At the end of incubation, the medium wasaspirated, and cells were washed twice with 10% trichloroacetic acid andsolubilized with 0.2 M NaOH for 30 min at room temperature. The incorpo-rated radioactive thymidine was quantified by liquid scintillation counting. Toassess cell accumulation, viable cells were quantified using the WST-1 reagent(Roche Diagnostics, Indianapolis, IN). The reagent was added to the finalconcentration of 10% and incubated with cells for 2–4 h. Absorbance was thenmeasured at 450/690 nm using a plate reader.

Microarray and Real-Time Reverse Transcription-PCR Analyses.RNAs from LMycSN and LXSN (control) colonocytes were isolated using theTRI reagent (Sigma). These RNAs were converted into Cy3- and Cy5-labeledcDNAs and hybridized with the Murine Genome Array U74Av2 (Affymetrix,Santa Clara, CA) per manufacturer’s recommendations. Data were analyzedusing GeneSpring software (Silicon Genetics, Redwood City, CA). Additionalmicroarray experiments were performed using ATLAS nylon array (Clontech,Palo Alto, CA). To confirm differences in gene expression, real-time reversetranscription-PCR reactions were performed using dual-labeled Taqmanprobes and an ABI Prizm 7700 machine (Applied Biosystems, Foster City,CA). Both the clusterin and the gapdh probes were labeled with 5�-fluoresceinphosphoramidite/5�tetrachloro fluorescein phosphoramidite (FAM/TET) at the5�-end and 6-carboxytetramethylrhodamine (TAMRA) at the 3� end. Theirnucleotide compositions were as follows: agcagcctgcccttcctctggatt (clusterin)and tcccactcttccaccttcgatgcc (gapdh). Amplification primers had the followingcomposition: gacccctagagaactccac (clusterin sense), gaatcagttcttcccgag (clus-terin antisense), gctacactgaggaccaggttgtct (gapdh sense), and accaggaaatgagct-tgacaaaga (gapdh antisense). Before PCR amplification, RNAs were convertedinto cDNA using SuperScript One-Step RT-PCR System (Invitrogen).

Western Blotting. For clusterin expression analysis, either cell lysates orconditioned media were used. To prepare lysates, cells were harvested byscraping and solubilized in lysis buffer [50 mM HEPES (pH 7.5), 150 mM

NaCl, 1.5 mM MgCl2, 1 mM EGTA, 10% glycerol, 1% Triton X-100, 0.1 mM

phenylmethylsulfonyl fluoride, and 1% protease inhibitor mixture (Sigma)].Protein content was determined using the Bio-Rad DC protein assay kit(Bio-Rad, Hercules, CA). Lysates (40 �g/lane) were resolved on 4–15%PAGE, transferred to nitrocellulose (Schleicher & Schuell, Keene, NH), andprobed with an anti-clusterin antibody (C-18; Santa Cruz Biotechnology, SantaCruz, CA or A241; Quidel, San Diego, CA) diluted according to manufac-turers’ recommendations. Conditioned media were loaded on PAGE neat.Appropriate secondary antibodies were used in horseradish peroxidase-conjugated forms. Antibody binding was detected using the enhancedchemiluminescence system (Amersham Biosciences, Piscataway, NJ). Whenindicated, an antibody reactive with murine actin was used to control for equalloading.

Transient Up-Regulation of Clusterin Expression. The mouse clusterincDNA (64) was subcloned into the MigR1 retroviral vector (65). EmptyMigR1 vector was used as a negative control. p53-null MycER colonocyteswere grown to 90% confluence in 24-well plates and transfected with appro-priate plasmids using Lipofectamine 2000 (Invitrogen) according to the man-ufacturer’s instructions. Conditioned medium was collected 24 h later forWestern analysis. At the same time cell accumulation was assessed using theWST-1 reagent (see above).

Stable Down-Regulation of Clusterin Expression via Antisense RNA.The clusterin antisense construct was obtained by subcloning the 597-ntXbaI/MluI fragment of the full-length human clusterin cDNA from pGEM-4ZLI (47) into the pCIneo vector in reverse orientation. The resultant constructwas transfected into A431 cells using calcium phosphate method, and individ-ual clones with stable integration of the expression vector were obtainedfollowing neomycin selection. The two antisense clones (AS2 and AS10) inwhich clusterin expression was undetectable by Western were chosen foradditional experiments. The empty pCIneo vector was similarly transfected,and neomycin-resistant clones were pooled to produce the control culture.

Culturing of Human Epithelial Cells and Fibroblasts. HEKs (66) andouter root sheath keratinocytes (67) were maintained in Epilife medium con-

3127

Myc-TRANSFORMED CELLS DOWN-REGULATE CLUSTERIN



taining human keratinocyte growth supplement (Cascade Biologics, Portland,OR). Renal proximal tubule epithelial cells (RPTEC) were obtained fromClonetics and maintained in the RGM-defined medium (Clonetics, San Diego,CA). A431 epidermoid carcinoma cells (68) were cultured in RPMI 1640supplemented with 10% FCS as described previously (54). The 293 adenovi-rally transformed embryonic kidney cells (69) were cultured in DMEM sup-plemented with 10% FCS (Invitrogen). Primary human dermal fibroblasts werecultured in the same medium.

Overexpression of Myc in HEKs. A human c-Myc-expressing adenovirusthat also coexpresses green fluorescent protein (Ad-MYC-GFP) and an ade-novirus expressing GFP alone were generated using reagents developed in thelaboratory of Dr. Bert Vogelstein (70). Viruses were purified by CsCl gradient,and the effective titer was determined by the frequency of GFP-positive cellsafter infection. Human foreskin keratinocytes were seeded at a density of2 � 105 cells/well in Keratinocyte-SFM medium (Invitrogen). The followingday, cells were infected with adenovirus diluted in PBS for 90 min. Viralsupernatants were replaced with fresh medium, and whole cell extracts werecollected 24 and 48 h after infection in Laemmli sample buffer. Myc expres-sion was confirmed using Western blotting.

Overexpression of Recombinant Human Clusterin in 293 Cells. Forclusterin overexpression, a 1430-bp SalI-fragment of the human clusterin genecontaining the entire coding sequence (with the exception of the ATG codonand the signal peptide) was used. This fragment was inserted into the pCR3vector (Invitrogen) encoding the hemagglutinin signal peptide as well as theFLAG-tag and a histidine heptad. Then, the PvuI-linearized construct wastransfected into 293 cells using calcium phosphate transfection method, andstable transfectants were selected in G418-containing media. Individual cloneswere further analyzed using Northern and Western (with an anti-FLAG anti-body) blotting. The clone expressing highest levels of recombinant protein wasselected as a source of clusterin.

Purification of Recombinant Human Clusterin. Large scale propagationof 293 cells producing clusterin was performed at Apotech Corp. (Lausanne,Switzerland). Cells were cultured for 7 days in DMEM/F12 medium contain-ing 2% FCS. The supernatant was filtered and passed several times through acolumn containing anti-FLAG-M2 affinity gel. Retained clusterin was theneluted by 0.1 M citric acid (pH 2.5) and neutralized with 1 M Tris (pH 8.0).Eluted fractions were concentrated using a Jumbosep membrane with a cutoffsize 30 kDa (Pall Corp., East Hills, NY). After several washes with endotoxin-free PBS, the protein concentration was determined using the BCA proteinassay kit (Pierce Biotechnology, Rockford, IL). In parallel, fractions wereobtained that had been depleted of clusterin by passing the preparation over-night at 4°C through a column containing 600 �l of anti-FLAG-M2 affinity gel(Sigma). The depletion was confirmed by Western blotting of aliquots col-lected prior to (2 �l) and after (20 �l) passing through the column.

Keratinocyte Clonogenic Assay. A total of 5 � 103 HEKs was plated in35-mm Petri dishes. Next day, cells were fed with control medium or mediumcontaining 50 �g/ml recombinant clusterin. After culturing for additional 96 h,cells were washed with PBS, fixed with 4% paraformaldehyde for 30 min atroom temperature, washed again, and stained with crystal blue in 0.1% boratebuffer for 5 min. The numbers of macroscopic colonies was recorded andcorrelated with the presence or absence of clusterin.

Chemical Skin Carcinogenesis. Clusterin gene knockout (Clu-null) mice(64) were backcrossed to the FVB/N background for six generations becausethis strain is sensitive to 7,12-dimethylbenz(a)anthracene/12-O-tetradecanoyl-phorbol-13-acetate (DMBA/TPA)-induced skin carcinogenesis (71). Forty-five wild-type (wt) FVB/N mice and 42 Clu-null congenics were shaved on theback 2 days before tumor initiation and subjected to single topical applicationsof 25 �g of DMBA (Fluka, Buchs, Switzerland) in 200 �l of acetone. Tumorpromotion was carried out by weekly applications of 4 �g of TPA (Sigma) in100 �l of acetone. This procedure was continued for 20 weeks. The appearanceof papillomas and carcinomas was assessed weekly. Any mice showing obvi-ous invasive carcinomas were euthanized and excluded from additional statis-tical analysis. Tumor specimens were fixed in 4% phosphate-buffered form-aldehyde, embedded in paraffin, cut into 5-�m sections, and stained withH&E. Three sections through different levels were evaluated in a blind fashionand categorized as well, moderately, or poorly differentiated. In well-differ-entiated carcinomas, morphological characteristics of the epidermis and thecapacity to undergo keratinization were preserved. In poorly differentiatedcarcinomas, there were significant morphological differences between mostly

atypical tumor cells and the normal epidermis. Moderately differentiatedcarcinomas possessed intermediate characteristics. Differences between strainswere assessed using the nonparametrical Mann-Whitney U test for papillomasand the �2 test for carcinomas.

RESULTS

Murine p53-Null Colonocytes Are Susceptible to Transforma-tion by c-Myc. Experimental models to study neoplastic transforma-tion of epithelial cells, colonocytes in particular, remain scarce. Todetermine whether a previously established line of p53-null murinecolonocytes (62) could be transformed by the c-Myc proto-oncogene,we have used retroviruses expressing Myc in either constitutivelyactive form or as a fusion with the estrogen receptor (MycER). Thelatter is expressed continuously but requires for activity the presenceof the cognate ligand (4-OHT; Ref. 72). These two forms wereencoded by LMycSN (73) and BabePuroMycER (72) retroviruses,respectively. LMycSN and BabePuroMycER, along with the corre-sponding empty vectors (LXSN and BabePuro), were transfected intocolonocytes, and transfectants were selected in an appropriate antibi-otic (G418 or puromycin). In LMycSN cultures, there was �5-foldoverexpression of the retrovirally encoded oncoprotein compared withendogenous c-Myc (Fig. 1A, left). In MycER cultures, the fusiononcoprotein was constitutively expressed, regardless of the presenceof absence of 4-OHT (Fig. 1A, right). Selected cells were also seededin soft agar as described in “Materials and Methods.” Only colono-cytes expressing active Myc (LMycSN- or 4-OHT-treated MycER-transduced) formed large colonies (Fig. 1B and data not shown) thatcould be discerned with a naked eye. Such colonies were removedfrom soft agar, pooled, expanded into cultures, and used for additionalanalyses.

To demonstrate that their transformed phenotype depended oncontinuous presence of Myc, MycER-overexpressing cultures weredeprived of the hormone. Within two to three passages, they werenoted to revert to the normal epithelial phenotype, with characteristiccobblestone-like morphology and pronounced contact inhibition ofgrowth resulting in relatively low cell density (Fig. 1C). They alsogrew at much slower rates, as evidenced by total cell accumulation(data not shown) and tritiated thymidine incorporation. The latterassay was performed on randomly chosen single-cell clones withverified expression of Myc. In all clones tested (e.g., 12, 13, 17, and23), rates of thymidine incorporation were significantly higher in thepresence of 4-OHT (Fig. 1D).

To determine whether LMycSN-transduced colonocytes were tu-morigenic, they were engrafted, either s.c. or orthotopically, intosyngeneic or Rag1-null mice. Tumorigenic colonocytes transformedwith the Ki-Ras oncogene (62) were used as a positive control.Although Ki-Ras-overexpressing cells typically formed tumors within3 weeks, no neoplasms were apparent in mice injected with Myc-overexpressing colonocytes, even after prolonged observation (6weeks) and when large numbers of cells (107) were used (data notshown). We thus concluded that only in vitro do Myc-overexpressingcells exhibit traits of the neoplastic phenotype. Chief among them isenhanced cell proliferation, both on solid supports and in semi-solidmedia.

Thrombospondin-1 and Clusterin Are Down-Regulated inMyc-Transformed Colonocytes. To relate Myc-induced enhancedproliferation to changes in gene expression, we performed microarrayanalysis comparing cDNAs in LXSN- versus LMycSN-colonocytes.Corresponding RNAs were extracted, converted into Cy3- or Cy5-labeled cDNAs, and hybridized to an Affymetrix U74Av2 mouse chiprepresenting �12,000 genes. Additional microarray experiments wereperformed, with similar results, using ATLAS nylon arrays. Despite

3128




pronounced changes in cellular phenotypes conferred by Myc over-expression, only �50 cDNAs (including expressed sequence tags)were up- or down-regulated �3-fold (Fig. 2A). Of these, only 14 up-and 15 down-regulated cDNAs corresponded to characterized genes(Table 2). Surprisingly, none of the up-regulated genes and only onedown-regulated gene [gadd45, a well know-target for Myc (23)], havebeen implicated in cell cycle control. On the other hand, a third of

Myc-down-regulated genes encoded proteins with extracellular local-ization (shaded rows in Table 2), including thrombospondin-1, whichwe had identified as a Myc target previously (29, 30). These findingssupported the notion that an altered repertoire of extracellular proteinsmight play an important role in stimulation of cell division by Myc. Ofthese, clusterin was of particular interest to us because its function hasnot been firmly established. On the other hand, a homology has beennoticed (41) between amino acids 77–98 in clusterin and cysteine-rich

Fig. 2. Down-regulation of clusterin in Myc-overexpressing colonocytes. A, scatterplotrepresenting expression of individual cDNA in LXSN (x axis)- versus LMycSN (yaxis)-transfected colonocytes. The vast majority of dots aligned along the equal value line(inner diagonal) and fell within the 3-fold change lines (outer diagonals), indicatingconsistency in global gene expression. Dots corresponding to clusterin and throm-bospondin-1 cDNAs are circled. B, real-time reverse transcription-PCR analyses ofclusterin mRNA levels in LXSN- and LMycSN-colonocytes. Amplification curves forboth clusterin and gapdh (internal control) are shown. In the two panels, LXSN andLMycSN colonocytes are compared with respect to clusterin (�C0t�2.5) and gapdh (nodifference) mRNA expression. C, Western blotting performed on lysates from parental orMyc (LMycSN)- or Ras (LRasSN)-overexpressing colonocytes using an anti-clusterinantibody. MycER-overexpressing colonocytes were either pretreated with 4-hydroxyta-moxifen (4-OHT) for 24 h or 96 h (�OHT) or left untreated (�OHT). In all experiments,murine �-actin was used a loading control.

Fig. 1. Transformed phenotype of c-Myc-overexpressing p53-null murine colonocytes.A, detection of exogenous Myc proteins in retrovirally transfected colonocytes by Westernblotting. Parental cells were used as negative controls. Migration of c-Myc (left panel) andthe MycER fusion protein (right panel) is indicated by arrows. Murine �-actin was usedas a loading control. B, bright field microscopy of colonocyte cultures transfected witheither empty vector (LXSN) or Myc-encoding retrovirus (LMycSN) and seeded in softagar as described in “Materials and Methods.” C, bright field microscopy of colonocytesexpressing MycER protein and either treated continuously with 4-hydroxytamoxifen(4-OHT) or subsequently switched to the normal medium (without 4-OHT). Photographedcells were grown on uncoated Petri dishes and allowed to reach saturated density. D,tritiated thymidine incorporation assay. The rate of thymidine incorporation was deter-mined under two growth conditions: with and without 4-OHT. The ratio of the former tothe latter is plotted for each of the indicated cultures. MycER12, 13, 17, and 23 arerandomly chosen single-cell clones where expression of MycER has been verified usingWestern blotting.

3129




thrombospondin type 1 repeats (TSRs; Ref. 74). Provocatively, TSRsof thrombospondin-1 are thought to mediate its antineoplastic prop-erties (75, 76).

To confirm that clusterin is indeed a target of Myc, we set up areal-time reverse transcription-PCR assay that could reproduciblydetect even slight differences in expression levels of clusterin andgapdh (internal standard). This was apparent in the pilot experimentwhere serial 2-fold dilutions of cDNA were used (data not shown).When cDNAs from LXSN- and LMycSN-colonocytes were tested,�6-fold down-regulation by Myc was observed (�C0t�2.5; Fig. 2B,bottom two panels), consistent with the microarray data. No differ-ences in gapdh levels were observed. To demonstrate the correspond-ing difference in protein levels, we performed Western blotting onlysates from LMycSN- and LXSN-transduced colonocytes. As evi-denced by data in Fig. 2C, left panel, overexpression of Myc corre-lated with decreased clusterin levels. Importantly, no decrease inclusterin levels was observed in cells transformed by Ki-Ras onco-gene (Fig. 2C, parental versus LRasSN). Thus, deregulation of clus-terin should be attributed to Myc, not merely to the transformedphenotype of Myc-transduced colonocytes.

To determine whether the clusterin gene was a direct target of Myc,we analyzed MycER colonocytes that had been stimulated with4-OHT for 96 or 24 h. Down-regulation of clusterin was apparent after96 h but not after 24 h or in control parental colonocytes (Fig. 1C,three rightmost panels). This finding could be interpreted in twoways. One possibility is that clusterin is an indirect target of Myc andis repressed by intermediate Myc effectors. The other possibility isthat Myc directly interacts with the clusterin gene promoter, but itsdown-regulation is contingent upon secondary changes conferred byMyc overexpression. This second possibility is consistent with thepresence in the clusterin gene promoter (64) of a sequenceACCA�1CCCGC bearing resemblance to the pyrimidine-rich con-sensus of the initiator element YYCA�1YYYYY, a putative Myc-response element (77). Although the detection of possible Myc bind-

ing to this sequence was beyond the scope of this study, we wereinterested in determining whether down-regulation of clusterin con-tributes to enhanced cell accumulation, the most salient feature ofMyc-transformed colonocytes.

Clusterin Levels Affect Accumulation of Neoplastic Colono-cytes. As evidenced by thymidine incorporation assay (Fig. 1D),activation of Myc forces cells to enter the S-phase in greater numbers.To determine whether this results in increased cell accumulation andto correlate this increased accumulation with clusterin levels, wemeasured the number of viable MycER-transduced colonocytes withand without 4-OHT. By the time clusterin was down-regulated (96 h;Fig. 3A, Western blot in insert), 4-OHT-treated cultures contained60% more viable cells. However, slight but reproducible differencesin cell numbers were apparent as early as 24 h after 4-OHT treatment,i.e., before clusterin levels fell (compare Figs. 3A and 2C). Thus, itseemed unlikely that clusterin alone accounts for Myc-dependentincrease in cell accumulation. Still, clusterin could be one of the cellgrowth inhibitors (along with thrombospondin-1) that are down-regulated by Myc and thus contribute to enhanced accumulation.

To test whether this was indeed the case, we have generated amurine clusterin-encoding retrovirus and transfected it into 4-OHT-treated MycER colonocytes with very low levels of endogenousclusterin. We chose not to select for stable clusterin-overexpressingclones because those would arise after in vitro selection and could beenured to potential inhibitory effects of clusterin. Instead, 24 h aftertransfection, we directly measured cell accumulation and correlated itto clusterin levels. The efficiency of transient transfection was mod-erate, in the range of 15–20%, as judged by the number of GFP-expressing cells (data not shown). However, increased clusterin se-cretion was apparent when Western blotting was performed on culturesupernatants (Fig. 3B, top panel). Furthermore, because clusterin is asecreted protein, its increased levels in conditioned media couldaffect, in a paracrine fashion, all cells albeit modestly. Indeed, allcultures transfected with the clusterin construct (Clu1 through 4)

Table 1 Genes differentially expressed in Myc-transformed versus parental colonocytes

AffyID GenesFold

changeSignificance,

value

Activated by Myc98465 Interferon activated gene 204 (Ifi204) 21.8 0.00001099580 UDP-glucuronosyltransferase 1 family, member 1 (Ugt1a1) 14.0 0.00108297128 Interleukin 5 receptor, � (Il5ra) 10.6 0.000333100445 Small proline-rich protein 1B (Sprr1b) 8.8 0.00001398843 Zinc finger protein of the cerebellum 2 (Zic2) 8.6 0.000499160547 Thioredoxin interacting protein (Txnip) 6.4 0.00002193994 Synaptonemal complex protein 3 (Sycp3) 5.4 0.000007161132 Sciellin (Scel) 4.8 0.000014103226 Mannose receptor, C type 1 (Mrc1) 4.3 0.00025393929 Proliferin (Plf) 4.1 0.00000093883 Proliferin 2 (Plf2) 4.0 0.000000102906 T-cell specific GTPase (Tgtp) 3.8 0.00199192917 Matrix metalloproteinase 7 (Mmp7) 3.2 0.000068104017 Fatty acid-coenzyme A ligase, long chain 4 (Facl4) 3.1 0.000021

Repressed by Myc161045 Leucine rich repeat protein 1, neuronal (Lrrn1) 29.2 0.99999999126 Inactive X specific transcripts (Xist) 17.5 0.999987101993 Tenascin C (Tnc) 5.0 0.999995161294 Clusterin (Clu) 4.8 0.99999995286 Clusterin (Clu) 4.5 0.999999104598 Dual specificity phosphatase 1 (Dusp1) 4.2 0.99999993097 Arginase 1, liver (Arg1) 4.2 0.999999103430 Drebrin 1 (Dbn1) 4.0 0.99999997890 Serum/glucocorticoid regulated kinase (Sgk) 3.8 0.999999160469 Thrombospondin 1 (Thbs1) 3.7 0.99999994515 Sulfide quinone reductase-like (yeast) (Sqrdl) 3.6 0.99999593705 Cholinergic receptor, nicotinic, � polypeptide 1 (muscle) (Chrnb1) 3.5 0.999996160522 Growth arrest and DNA-damage-inducible 45� (Gadd45g) 3.2 0.999998102298 Nerve growth factor, � (Ngfb) 3.2 0.999999101359 Laminin, �2 (Lamb2) 3.2 0.99999793122 Acidic epididymal glycoprotein 1 (Aeg1) 3.0 0.999999

3130




contained fewer cells than their “vector only” (GFP1 through 4)counterparts (graph in Fig. 3B). On average, transient overexpressionof clusterin resulted in �12% decrease in cell accumulation (insert,light gray versus dark gray bars, P 0.033). This decrease could notaccount for the full range of Myc effects but was consistent withclusterin being one of the Myc-down-regulated inhibitors of colono-cyte proliferation. We next asked whether this “inhibit the inhibitor”strategy plays out in other types of epithelial cells, in particular inkeratinocytes that are available both as primary explants (HEKs) andas epidermoid tumor cell lines.

Clusterin Inhibits Accumulation of Transformed EpidermalCells and Is Down-Regulated in Myc-Overexpressing Keratino-cytes. To analyze the effect of clusterin on transformed keratinocytes,we have chosen A431 human epidermoid carcinoma cells (68) wherelevels of clusterin are relatively high and further increase in its levels

would be difficult to achieve. Thus, we generated an anti-sense RNAconstruct (see “Materials and Methods”) and introduced it into A431cells. The efficiency of clusterin expression inhibition was assessedusing Western blotting (Fig. 4A, insert). Clones with virtually unde-tectable clusterin levels were compared with empty vector-transfectedcells with respect to cell accumulation. Both clones with silencedclusterin grew much faster than parental cells as evidenced by in-creased numbers of viable cells (Fig. 4A, graph).

We next asked whether clusterin would be subjected to down-regulation by Myc in primary HEKs. In these cells, Myc is known toinduce hyperproliferation (78). We thus generated a recombinantadenovirus coexpressing Myc and GFP and used it to infect HEK asdescribed in “Materials and Methods.” The efficiency of infectionapproached 100%, as judged by visual examination of infected cul-tures under fluorescent light (Fig. 4B, right). When lysates wereprepared from these cultures and used for Western blotting, it becameapparent that clusterin levels in AdMycGFP-HEKs were several foldlower that in control AdGFP-HEKs (Fig. 4C). The effect of clusterindown-regulation on HEK proliferation could be studied using eithertransient expression assays (see above) or purified recombinant pro-tein. Because HEKs are rather refractory to conventional transfectiontechniques, the second approach was chosen.

Recombinant Human Clusterin Inhibits Proliferation of Kera-tinocytes and Other Epithelial Cells. To study the role of clusterinin HEK proliferation in vitro, we generated recombinant humanclusterin. We fused the clusterin coding sequence to the hemaggluti-nin signal peptide and both FLAG and histidine tags, as described in“Materials and Methods.” The recombinant construct was introducedinto 293 embryonic kidney cells, and their supernatants were used forclusterin purification. Under reducing conditions, the purified fraction(Fig. 5A “Coomassie”) revealed the 40-kDa monomeric reduced,75–80-kDa heterodimeric nonreduced form, small quantities of BSA,and also several high molecular weight species that were not reactivewith the anti-clusterin antibody (Fig. 5A, “Immunoblot”). The latterwere estimated to represent 10% of the total preparation. Thereduced 40-kDa form of clusterin was predominant and comigrated inPAGE with clusterin detected by an anti-clusterin antibody in humanserum (Fig. 5B).

To test whether clusterin inhibits keratinocyte proliferation, wetreated HEK cultures with recombinant clusterin and measured[3H]thymidine incorporation, total cell accumulation (both after 48 h),and clonal outgrowth (after 96 h) as described in “Materials andMethods.” In the first two experiments, clusterin inhibited DNAreplication and cell accumulation in a dose-dependent manner andalso negatively affected clonal outgrowth when present at a concen-tration of 50 �g/ml (Fig. 5C, left, center, and right panels, respec-tively). No cell toxicity or apoptosis was apparent in treated cultures.Importantly, inhibitory concentrations were physiologically relevantbecause clusterin’s concentration in serum is in the 50–200 �g/mlrange. To extend our observation to other epithelial cells, we alsotreated with clusterin outer root sheath keratinocytes and renal prox-imal tubule epithelial cells. With both outer root sheath keratinocytesand renal proximal tubule epithelial cells, the same inhibitory effect ofclusterin on thymidine incorporation was observed (Fig. 5D). Impor-tantly, this effect was caused by clusterin itself because the depletionof conditioned media on the affinity column efficiently removedclusterin and at the same time abolished the inhibitory effect onepithelial cell proliferation (Fig. 5E). In contrast, presence or absenceof clusterin didn’t affect proliferation of primary human dermal fi-broblasts, even at high concentrations (50 �g/ml). To rule out that cellaccumulation is negatively affected by copurifying BSA, we incu-bated HEK in media containing defined concentrations of BSA or Clu.In this experiment, BSA was found to slightly increase cell accumu-

Fig. 3. Clusterin expression and colonocyte accumulation. A, growth curves of 4-hy-droxytamoxifen-treated (�4-OHT) versus untreated (�4-OHT) MycER-expressingcolonocytes. Down-regulation of clusterin in conditioned media was confirmed usingWestern blotting (insert). B, the effect of transient clusterin overexpression on cellaccumulation. Eight cultures of MycER-transduced colonocytes were independently trans-fected with either the empty vector (GFP1 through GFP4, first four lanes) or the murineclusterin/green fluorescent protein (GFP)-encoding retrovirus (Clu1 through Clu4, lastfour lanes). Clusterin levels in conditioned media were determined using Western blotting(top panel). In the same cultures, the number of viable cells were determined using WSTassay (plotted in the bottom panel). Average cell accumulation was also plotted (insert)and analyzed using unpaired Student t test. Observed differences were statisticallysignificant (P 0.05).

3131




lation, whereas clusterin decreased it (Fig. 5F). Taken together, theseresults suggested that clusterin is a bona fide inhibitor of epithelial cellproliferation in vitro.

Clusterin Inhibits Mouse Skin Tumor Development and Per-sistence but Not Tumor Progression. To determine whether clus-terin attenuates neoplastic growth in vivo, we performed a chemicalDMBA/TPA skin carcinogenesis assay in wt and clusterin-null mice(64). In both groups tumors began to emerge after a similar latency(7–8 weeks), and their sizes and histological appearances did not varysignificantly (data not shown). However, at week 20, the mean num-ber of papillomas/mouse was significantly higher in clusterin-nullmice than in wt mice (P 0.02, Mann-Whitney U test; Fig. 6, A andB). The same was true for the entire first 18-week period (P 0.036).During weeks 20–35, mice began to develop invasive carcinomas andhad to be euthanized. Thus, comprehensive statistical analysis ofpapilloma persistence was hindered by low number of animals. Still,by week 35, the mean number of remaining papillomas in clusterin-null mice was strikingly higher than in their wt counterparts (3.3versus 0.8 papillomas/mouse). These data demonstrate that, at earlystages of carcinogenesis, the production of clusterin attenuates thedevelopment and persistence of benign tumors, probably via growthsuppression.

Both groups of mice were also monitored for the development ofinvasive squamous cell carcinomas, which in the FVB/N backgroundoccurs in �60% of mice (71). All squamous cell carcinomas werecategorized as well-, moderately, or poorly differentiated, as describedin “Materials and Methods.” There were no statistically significant

differences between clusterin-null and wt mice in the timing ofcarcinoma development and the rate of malignant conversion (19 and15%, respectively; data not shown). However, squamous cell carci-nomas developing in clusterin-null mice appeared on average betterdifferentiated and less aggressive. Indeed, at 42 weeks, the percentageof poorly or undifferentiated squamous cell carcinomas in clusterin-null animals was 20% (12 of 60), compared with over 50% (32 of 61)in wt mice (Fig. 6, C and D; �2 test, P 0.001). Therefore, once thepapilloma have developed, expression of clusterin favors the devel-opment of more poorly differentiated and more aggressive carcinomaphenotypes. This observation is consistent with the idea that tumor-attenuating properties of clusterin are realized early in tumor progres-sion when cell proliferation is likely to be a limiting factor.

DISCUSSION

Carcinogenesis in humans is a multistep process involving initia-tion and promotion mechanisms. These mechanisms rely on the in-activation of one or more tumor suppressor genes and the activation ofcertain proto-oncogenes, resulting in uncontrolled proliferation oftumor cell precursors. This enhanced proliferation could be aided bydecreased synthesis of or decreased response to endogenous growthinhibitors. The latter are exemplified by the members of the trans-forming growth factor (TGF)-� family such as TGF-�1 and TGF-�2.These are secreted proteins that are constitutively synthesized by awide variety of cell types in many species and are potentially impli-cated in many biological processes, including cell proliferation and

Fig. 4. Clusterin expression and accumulation of human keratino-cytes. A, accumulation of neoplastic epidermoid A431 cells expressingvarying levels of clusterin. Empty vector-transfected cells (Vect) werecompared with two clones (AS2 and AS10) expressing antisense clus-terin RNA (see “Materials and Methods”). Down-regulation of clusterinwas confirmed using Western blotting (insert). Recombinant humanclusterin (rClu) was used as a positive control. WT refers to parentalA431 cells. B, infection of human epidermal keratinocytes with a Myc/GFP-encoding adenovirus (Ad-MYC-GFP, bottom panels). Top panelsrefer to the empty vector-infected cells. The two left panels were ob-tained using bright field microscopy, the two right panels were obtainedusing fluorescent microscopy [for green fluorescent protein (GFP) ex-pression]. C, Western blotting performed on cells in B. For clusterin andactin detection, undiluted lysates were used; for Myc detection, theywere diluted 1:50.

3132




differentiation (79). The interplay between Myc and TGF-� is nowrecognized as an important regulator of cell proliferation (80). TGF-�is also regulated at the posttranslational level (81), and one of its keyactivators is none other than thrombospondin-1, which releases

TGF-� from a latent complex (82, 83). The binding to TGF-� ismediated by a distinct amino acid sequence flanked by two type 1repeats (TSR) within thrombospondin-1 (reviewed in Ref. 84). Inter-estingly, TSR-homologous sequences are present in dozens of other-

Fig. 5. Growth inhibitory properties of recombinant human clusterin (Clu) produced in 293 cells. A, Coomassie Blue staining (left lane) and Western blotting with an anti-Cluantibody (right lane) of the Clu preparation. Both monomeric (�40 kDa) and heterodimeric (�80 kDa) forms are present under reducing conditions. In the left lane, copurifying BSAis also detectable. B, Western blotting performed on 0.21 �g of the Clu preparation (r-hClu) electrophoresed under reducing conditions alongside a 0.2 �l-aliquot of normal humanserum (hSer). C, dose-dependent inhibition by Clu of the growth of human epithelial keratinocyte (HEK). [3H]Thymidine incorporation (left panel) and numbers of viable cells (middlepanel) were plotted against concentrations of Clu in growth medium. The right panel depicts the outgrowth of single cell clones of sparsely seeded HEK after 5 days either in the absence(two duplicate plates in top row) or in the presence (two duplicate plates in bottom row) of r-hClu (50 �g/ml). D, the effect of Clu on proliferation of outer root sheath keratinocytes(ORS) and renal proximal tubule epithelial cells (RPTECs). The experimental setup was as in C, left panel. E, proliferation of cells in Clu-containing media (50 �g/ml) before andafter depletion with an anti-FLAG antibody. The efficacy of depletion was assessed using Western blotting (top three panels; “�” and “�” refer to undepleted and depleted media,respectively. [ 3]Thymidine incorporation was assessed in HEK, RPTEC, and primary human dermal fibroblasts (FBs). All experiments were performed in triplicates. Error bars referto SE’s of two to five independent experiments. F, accumulation of HEK in control medium (“Cont”) or media supplemented with BSA (50 �g/ml) or Clu (50 �g/ml).

3133




wise unrelated proteins which comprise the TSR superfamily (74). Onthe basis of the homology with thrombospondin-1, we suggested thatclusterin might belong to the same superfamily and share some of thefunctions attributed to its members, for instance, inhibition of cellproliferation. Furthermore, we have previously demonstrated thatclusterin mRNA expression is restricted to differentiating rather thanproliferating cells in several differentiated epithelia. On the basis ofthese facts, we asked whether clusterin is a negative regulator of cellproliferation in vitro and in vivo.

In this study, we present evidence that clusterin is down-regulatedin cultured epithelial cells that accelerate their growth rate in responseto c-Myc activation. Moreover, clusterin was found to directly inhibitgrowth and accumulation of murine colonocytes, human keratino-cytes, and other epithelial cells in vitro. Inhibition of cell proliferationwas somewhat cell type specific because purified recombinant clus-terin has no effect on fibroblasts. Our results are in accordance withpublished data that proliferation of human fibroblasts geneticallymodified to overexpress clusterin is not inhibited (85), but exposure toclusterin of human prostate cancer cell line LNCaP (86) as well asSV40-immortalized prostate epithelial cells (87) resulted in slowergrowth rates.

Given the negative role that clusterin plays in epithelial cell pro-liferation, it seemed likely that it attenuates neoplastic growth in vivo.To address this possibility, we performed chemical skin carcinogen-esis in wt and clusterin-null mice. In this system, topical applicationof DMBA/TPA leads to the formation of benign papillomas. Thesesmall neoplasms arise almost exclusively due to enhanced cell pro-liferation because invasive growth, angiogenesis, and metastasis arenot required for their development (88). We demonstrated here for thefirst time that the lack of clusterin increases the susceptibility totumorigenesis after carcinogenic challenge. Indeed, the mean numberof benign papillomas/mouse is significantly increased in the absence

of clusterin. Interestingly, clusterin-null mice have not been noted todevelop any spontaneous tumors. Therefore, its absence is not suffi-cient to induce cell proliferation, but once a proliferative signal (e.g.,DMBA/TPA) is provided, neoplastic growth is less constrained. Thus,we consider clusterin to be not a tumor suppressor but a tumorattenuator acting predominantly at early stages of neoplastic growth.

This idea could explain why clusterin expression has been reportedin a variety of neoplasms. Existing data indicate that clusterin isabundantly produced in several types of human cancers such as breast(89), renal clear cell (90), and prostate (91) carcinomas. However, ithas not been demonstrated that clusterin is produced in the same cellsthat undergo proliferation. It is thus possible that clusterin is activatedto counteract oncogenic stimuli in non- or slowly proliferating tumorcompartments. Activation of growth inhibitors in tumor tissues hasbeen reported, p16 cyclin-dependent kinase-inhibitor being a primeexample (92). It would be interesting to determine, using laser mi-crodissection, whether or not expression of clusterin correlates withlower bromodeoxyuridine incorporation and lower mitotic indices.When such a detailed analysis was performed followed by serialanalysis of gene expression, strong down-regulation of clusterin wasimmediately apparent in highly malignant MD PR317 prostate ade-nocarcinoma cells (Tag Odds Normal:Cancer 5.85, The CancerGenome Anatomy Project).7

Interestingly, during late stages of skin carcinogenesis, clusterinexpression seems to favor tumor progression because in clusterin-nullmice papillomas progress into more differentiated squamous cellcarcinomas than in wt mice. In addition, some previous studies haveshown that clusterin could favor the metastatic process (93). Thus,clusterin might act in a biphasic fashion during skin carcinogenesis as

7 Internet address: http://cgap.nci.nih.gov.

Fig. 6. Chemical carcinogenesis in clusterin-null versus wild-type (WT) mice. A, gross appearance of papilloma developing in clusterin-null knockout (KO) versus WT FVB/N miceafter exposure to 7,12-dimethylbenz(a)anthracene/12-O-tetradecanoylphorbol-13-acetate (see “Materials and Methods”). B, quantitative analysis of papilloma development in 42 WT(E) and 45 KO (f) mice. For each strain, mean number of papillomas per animal was plotted against time. An apparent dip in the KO plot line is because multiple mice developedcarcinomas at weeks 23–25 and were therefore excluded from papilloma count. C, development of carcinomas in KO and WT mice. Poorly and well-differentiated carcinomas arerepresented by f and �, respectively. Total numbers of carcinomas (100%) were similar in both strains: 61 in WT and 60 in KO mice. D, representative carcinomas developing inKO and WT mice (left and right panels, respectively; original magnification, �20). A well-differentiated tumor in the left panel contains zones of keratinization and squamous cellswith moderately atypical nuclei. A poorly differentiated tumor in the right panel contains rectangular or spindle-shaped cells with significantly atypical nuclei and numerous mitoses.

3134




a tumor attenuator early on and as an enhancer of the malignantphenotype in full-fledged tumors. Similar effects have already beenattributed to TGF-� during skin carcinogenesis (94) and throm-bospondin-1 in breast and other carcinomas (95). This raises thequestion whether inactivation of clusterin should be attempted oravoided in cancer patients. Although the former approach is currentlybeing tested (96, 97), our data indicate that high levels of clusterinmight be protective against early stages of neoplastic growth. Thisdiscrepancy will have to be resolved before additional clinical inter-ventions are attempted.

ACKNOWLEDGMENTS

We thank Gertraud Radlgruber and Bernadette Mermillod (University ofGeneva) for technical assistance and for assistance with statistical analyses,respectively. We thank Dr. Jeffrey Ilardi, Gautam Rajpal, and Monica Lee(University of Pennsylvania) for their help with generation of murine clusterin-expressing retroviruses. We also thank Isaiah Fidler (M. D. Anderson CancerCenter) for his help with learning the technique of orthotopic transplantation.

REFERENCES

1. Grandori C, Cowley SM, James LP, Eisenman RN. The Myc/Max/Mad network andthe transcriptional control of cell behavior. Ann Rev Cell Dev Biol 2000;16:653–99.

2. Blackwell TK, Kretzner L, Blackwood EM, Eisenman RN, Weintraub H. Sequence-specific DNA binding by the c-myc protein. Science (Wash. DC) 1990;250:1149–51.

3. Kretzner L, Blackwood EM, Eisenman RN. Myc and Max proteins possess distincttranscriptional activities. Nature (Lond.) 1992;359:426–9.

4. Amati B, Dalton S, Brooks MW, Littlewood TD, Evan GI, Land H. Transcriptionalactivation by the human c-Myc oncoprotein in yeast requires interaction with Max.Nature (Lond.) 1992;359:423–6.

5. McMahon SB, Van BA, Dugan KA, Copeland TD, Cole MD. The novel ATM-relatedprotein TRRAP is an essential cofactor for the c-Myc and E2F oncoproteins. Cell1998;94:363–74.

6. McMahon SB, Wood MA, Cole MD. The essential cofactor TRRAP recruits thehistone acetyltransferase hGCN5 to c-Myc. Mol Cell Biol 2000;20:556–62.

7. Amati B, Frank SR, Donjerkovic D, Taubert S. Function of the c-Myc oncoprotein inchromatin remodeling and transcription. Biochim Biophys Acta 2001;1471:M135–45.

8. Li L, Nerlov C, Prendergast GC, Ziff EB. c-Myc represses transcription in vivo by anovel mechanism dependent on the initiator element and Myc box II. EMBO J1994;13:4070–9.

9. Staller P, Peukert K, Kiermaier A, et al. Repression of p15INK4b expression by Mycthrough association with Miz-1. Nat Cell Biol 2001;3:392–9.

10. Oster SK, Ho CS, Soucie EL, Penn LZ. The myc oncogene: marvelously complex.Adv Cancer Res 2002;84:81–154.

11. O’Connell BC, Cheung AF, Simkevich CP, et al. A large scale genetic analysis ofc-Myc-regulated gene expression patterns. J Biol Chem 2003;278:12563–73.

12. Bello-Fernandez C, Packham G, Cleveland JL. The ornithine decarboxylase gene isa transcriptional target of c-Myc. Proc Natl Acad Sci USA 1993;90:7804–8.

13. Jansen-Durr P, Meichle A, Steiner P, et al. Differential modulation of cyclin geneexpression by MYC. Proc Natl Acad Sci USA 1993;90:3685–9.

14. Galaktionov K, Chen X, Beach D. cdc25 cell cycle phosphatase as a target of c-myc.Nature (Lond.) 1996;382:511–7.

15. Hermeking H, Rago C, Schuhmacher M, et al. Identification of CDK4 as a target ofc-MYC. Proc Natl Acad Sci USA 2000;97:2229–34.

16. Lasorella A, Noseda M, Beyna M, Iavarone A. Id2 is a retinoblastoma protein targetand mediates signalling by Myc oncoproteins. Nature (Lond.) 2000;407:592–8.

17. Wang J, Xie LY, Allan S, Beach D, Hannon GJ. Myc activates telomerase. Genes Dev1998;12:1769–74.

18. Philipp A, Schneider A, Vasrik I, et al. Repression of cyclin D1: a novel function ofMYC. Mol Cell Biol 1994;14:4032–43.

19. Mitchell KO, El-Deiry WS. Overexpression of c-Myc inhibits p21WAF1/CIP1 ex-pression and induces S-phase entry in 12-O-tetradecanoylphorbol-13-acetate (TPA)-sensitive human cancer cells. Cell Growth Differ 1999;10:223–30.

20. Claassen GF, Hann SR. A role for transcriptional repression of p21CIP1 by c-Myc inovercoming transforming growth factor �-induced cell cycle arrest. Proc Natl AcadSci USA 2000;97:9498–503.

21. Gartel AL, Ye X, Goufman E, et al. Myc represses the p21(WAF1/CIP1) promoterand interacts with Sp1/Sp3. Proc Natl Acad Sci USA 2001;98:4510–5.

22. Seoane J, Pouponnot C, Staller P, Schader M, Eilers M, Massague J. TGF-�influences Myc, Miz-1 and Smad to control the CDK inhibitor p15INK4b. Nat CellBiol 2001;3:400–8.

23. Marhin WW, Chen S, Facchini LM, Fornace AJ, Penn LZ. Myc represses the growtharrest gene gadd45. Oncogene 1997;14:2825–34.

24. Lee TC, Li L, Philipson L, Ziff EB. Myc represses transcription of the growth arrestgene gas1. Proc Natl Acad Sci USA 1997;94:12886–91.

25. Eilers M, Picard D, Yamamoto KR, Bishop JM. Chimaeras of Myc oncoprotein andsteroid receptors cause hormone-dependent transformation of cells. Nature (Lond.)1989;340:66–8.

26. Mateyak MK, Obaya AJ, Adachi S, Sedivy JM. Phenotypes of c-Myc-deficient ratfibroblasts isolated by targeted homologous recombination. Cell Growth Differ 1997;8:1039–48.

27. Johnston LA, Prober DA, Edgar BA, Eisenman RN, Gallant P. Drosophila mycregulates cellular growth during development. Cell 1999;98:779–90.

28. Trumpp A, Refaeli Y, Oskarsson T, et al. c-Myc regulates mammalian body size bycontrolling cell number but not cell size. Nature (Lond.) 2001;414:768–73.

29. Tikhonenko AT, Black DJ, Linial ML. Viral Myc oncoproteins in infected fibroblastsdown-modulate thrombospondin-1, a possible tumor suppressor gene. J Biol Chem1996;271:30741–7.

30. Janz A, Sevignani C, Kenyon K, Ngo C, Thomas-Tikhonenko A. Activation of theMyc oncoprotein leads to increased turnover of thrombospondin-1 mRNA. NucleicAcids Res 2000;28:2268–75.

31. Guo N, Templeton NS, Al-Barazi H, et al. Thrombospondin-1 promotes alpha3beta1integrin-mediated adhesion and neurite-like outgrowth and inhibits proliferation ofsmall cell lung carcinoma cells. Cancer Res 2000;60:457–66.

32. Armstrong LC, Bjorkblom B, Hankenson KD, Siadak AW, Stiles CE, Bornstein P.Thrombospondin 2 inhibits microvascular endothelial cell proliferation by a caspase-independent mechanism. Mol Biol Cell 2002;13:1893–905.

33. Tomii Y, Kamochi J, Yamazaki H, et al. Human thrombospondin 2 inhibits prolif-eration of microvascular endothelial cells. Int J Oncol 2002;20:339–42.

34. Lawler J. Thrombospondin-1 as an endogenous inhibitor of angiogenesis and tumorgrowth. J Cell Mol Med 2002;6:1–12.

35. Ngo C, Gee MS, Akhtar N, et al. An in vivo function for the transforming mycprotein: elicitation of the angiogenic phenotype. Cell Growth Differ 2000;11:201–10.

36. Watnick RS, Cheng YN, Rangarajan A, Ince TA, Weinberg RA. Ras modulates Mycactivity to repress thrombospondin-1 expression and increase tumor angiogenesis.Cancer Cell 2003;3:219–31.

37. Brandvold KA, Neiman P, Ruddell A. Angiogenesis is an early event in the gener-ation of myc-induced lymphomas. Oncogene 2000;19:2780–5.

38. Pelengaris S, Littlewood T, Khan M, Elia G, Evan GI. Reversible activation of c-Mycin skin: induction of a complex neoplastic phenotype by a single oncogenic lesion.Mol Cell 1999;3:565–77.

39. Zeller KI, Jegga AG, Aronow B, O’Donnell KA, Dang CV. An integrated databaseof genes responsive to the Myc oncogenic transcription factor: identification of directgenomic targets. Genome Biol 2003;4:R69.1–10.

40. Louro ID, Bailey EC, Li X, et al. Comparative gene expression profile analysis of GLIand c-MYC in an epithelial model of malignant transformation. Cancer Res 2002;62:5867–73.

41. Jenne DE, Tschopp J. Clusterin: the intriguing guises of a widely expressed glyco-protein. Trends Biochem Sci 1992;17:154–9.

42. Rosenberg M, Silkensen J. Clusterin: physiologic and pathophysiologic consider-ations. Int J Biochem Cell Biol 1995;27:633–45.

43. Blaschuk O, Burdzy K, Fritz IB. Purification and characterization of a cell-aggregat-ing factor (clusterin), the major glycoprotein in ram rete testis fluid. J Biol Chem1983;258:7714–20.

44. Fritz IB, Burdzy K, Setchell B, Blaschuk O. Ram rete testis fluid contains a protein(clusterin) which influences cell-cell interactions in vitro. Biol Reprod 1983;28:1173–88.

45. Tung PS, Burdzy K, Wong K, Fritz IB. Competition between cell-substratum inter-actions and cell-cell interactions. J Cell Physiol 1992;152:410–21.

46. Choi NH, Mazda T, Tomita M. A serum protein SP40,40 modulates the formation ofmembrane attack complex of complement on erythrocytes. Mol Immunol 1989;26:835–40.

47. Jenne DE, Tschopp J. Molecular structure and functional characterization of a novelhuman cytolysis inhibitor (CLI) found in blood and seminal plasma. Identity tosulfated glycoprotein 2 (SGP-2), a constituent of rat testis fluid. Proc Natl Acad SciUSA 1989;86:7123–7.

48. Kirszbaum L, Sharpe JA, Murphy B, et al. Molecular cloning and characterization ofthe novel, human complement-associated protein, SP-40,40: a link between thecomplement and reproductive systems. EMBO J 1989;8:711–8.

49. de Silva HV, Stuart WD, Duvic CR, et al. A 70-kDa apolipoprotein designated ApoJis a marker for subclasses of human plasma high density lipoproteins. J Biol Chem1990;265:13240–7.

50. James RW, Hochstrasser AC, Borghini I, Martin B, Pometta D, Hochstrasser D.Characterization of a human high density lipoprotein-associated protein, NA1/NA2.Identity with SP-40,40, an inhibitor of complement-mediated cytolysis. ArteriosclerThromb 1991;11:645–52.

51. Jenne DE, Lowin B, Peitsch MC, Bottcher A, Schmitz G, Tschopp J. Clusterin(complement lysis inhibitor) forms a high density lipoprotein complex with apoli-poprotein A-I in human plasma. J Biol Chem 1991;266:11030–6.

52. French LE, Wohlwend A, Sappino AP, Tschopp J, Schifferli JA. Human clusteringene expression is confined to surviving cells during in vitro programmed cell death.J Clin Investig 1994;93:877–84.

53. Schwochau GB, Nath KA, Rosenberg ME. Clusterin protects against oxidative stressin vitro through aggregative and nonaggregative properties. Kidney Int 1998;53:1647–53.

54. Viard I, Wehrli P, Jornot L, et al. Clusterin gene expression mediates resistance toapoptotic cell death induced by heat shock and oxidative stress. J Investig Dermatol1999;112:290–6.

55. Wilson MR, Easterbrook-Smith SB. Clusterin is a secreted mammalian chaperone.Trends Biochem Sci 2000;25:95–8.

56. Yang CR, Leskov K, Hosley-Eberlein K, et al. Nuclear clusterin/XIP8, an X-ray-induced Ku70-binding protein that signals cell death. Proc Natl Acad Sci USA2000;97:5907–12.

3135




57. Criswell T, Klokov D, Beman M, Lavik JP, Boothman DA. Repression of IR-inducible clusterin expression by the p53 tumor suppressor protein. Cancer Biol Ther2003;2:372–80.

58. Leskov KS, Klokov DY, Li J, Kinsella TJ, Boothman DA. Synthesis and functionalanalyses of nuclear clusterin, a cell death protein. J Biol Chem 2003;278:11590–600.

59. Diemer V, Hoyle M, Baglioni C, Millis A. Expression of porcine complementinhibitor mRNA in cultured aortic smooth muscle cells: changes during differentia-tion in vitro. J Biol Chem 1992;267:5257–64.

60. French LE, Chonn A, Ducrest D, et al. Murine clusterin: molecular cloning andmRNA localization of a gene associated with epithelial differentiation processesduring embryogenesis. J Cell Biol 1993;122:1119–30.

61. Seiberg M, Marthinuss J. Clusterin expression within skin correlates with hair growth.Dev Dyn 1995;202:294–301.

62. Sevignani C, Wlodarski P, Kirillova J, et al. Tumorigenic conversion of p53-deficientcolon epithelial cells by an activated Ki-ras gene. J Clin Investig 1998;101:1572–80.

63. Fidler IJ. Orthotopic implantation of human colon carcinomas into nude mice pro-vides a valuable model for the biology and therapy of metastasis. Cancer MetastasisRev 1991;10:229–43.

64. McLaughlin L, Mistry M, Zhu G, et al. Apolipoprotein J/Clusterin limits the severityof murine autoimmune myocarditis. J Clin Investig 2000;106:1105–13.

65. Pear WS, Miller JP, Xu L, et al. Efficient and rapid induction of a chronic myelog-enous leukemia-like myeloproliferative disease in mice receiving P210 bcr/abl-transduced bone marrow. Blood 1998;92:3780–92.

66. Limat A, Hunziker T, Boillat C, Bayreuther K, Noser F. Post-mitotic human dermalfibroblasts efficiently support the growth of human follicular keratinocytes. J InvestigDermatol 1989;92:758–62.

67. Limat A, Noser F. Serial cultivation of single keratinocytes from the outer root sheathof human scalp hair follicles. J Investig Dermatol 1986;87:485–8.

68. Giard DJ, Aaronson S, Todaro GJ, et al. In vitro cultivation of human tumors:establishment of cell lines derived from a series of solid tumors. J Natl Cancer Inst(Bethesda) 2003;51:1417–23.

69. Graham FL, Smiley J, Russell WC, Nairn R. Characteristics of a human cell linetransformed by DNA from human adenovirus type 5. J Gen Virol 1977;36:59–74.

70. He TC, Zhou S, da Costa LT, Yu J, Kinzler KW, Vogelstein B. A simplified systemfor generating recombinant adenoviruses. Proc Natl Acad Sci USA 1998;95:2509–14.

71. Hennings H, Glick AB, Lowry DT, et al. FVB/N mice: an inbred strain sensitive tothe chemical induction of squamous cell carcinomas in the skin. Carcinogenesis1993;14:2353–8.

72. Littlewood TD, Hancock DC, Danielian PS, Parker MG, Evan GI. A modifiedoestrogen receptor ligand-binding domain as an improved switch for the regulation ofheterologous proteins. Nucleic Acids Res 1995;23:1686–90.

73. Blackwood EM, Lugo TG, Kretzner L, et al. Functional analysis of the AUG- andCUG-initiated forms of the c-Myc protein. Mol Biol Cell 1994;5:597–609.

74. Adams JC, Tucker RP. The thrombospondin type 1 repeat (TSR) superfamily: diverseproteins with related roles in neuronal development. Dev Dyn 2000;218:280–99.

75. Guo N, Krutzsch HC, Inman JK, Roberts DD. Thrombospondin 1 and type I repeatpeptides of thrombospondin 1 specifically induce apoptosis of endothelial cells.Cancer Res 1997;57:1735–42.

76. Iruela-Arispe ML, Lombardo M, Krutzsch HC, Lawler J, Roberts DD. Inhibition ofangiogenesis by thrombospondin-1 is mediated by 2 independent regions within thetype 1 repeats. Circulation 1999;100:1423–31.

77. Claassen GF, Hann SR. Myc-mediated transformation: the repression connection.Oncogene 1999;18:2925–33.

78. Cerezo A, Stark HJ, Moshir S, Boukamp P. Constitutive overexpression of humantelomerase reverse transcriptase but not c-myc blocks terminal differentiation inhuman HaCaT skin keratinocytes. J Investig Dermatol 2003;121:110–9.

79. Derynck R, Akhurst RJ, Balmain A. TGF-� signaling in tumor suppression andcancer progression. Nat Genet 2001;29:117–29.

80. Amati B. Integrating Myc and TGF-� signalling in cell-cycle control. Nat Cell Biol2001;3:E112–3.

81. Gumienny TL, Padgett RW. The other side of TGF-� superfamily signal regulation:thinking outside the cell. Trends Endocrinol Metabol 2002;13:295–9.

82. Crawford SE, Stellmach V, Murphy-Ullrich JE, et al. Thrombospondin-1 is a majoractivator of TGF-�1 in vivo. Cell 1998;93:1159–70.

83. Ribeiro SM, Poczatek M, Schultz-Cherry S, Villain M, Murphy-Ullrich JE. Theactivation sequence of thrombospondin-1 interacts with the latency-associated peptideto regulate activation of latent transforming growth factor �. J Biol Chem 1999;274:13586–93.

84. Chen H, Herndon ME, Lawler J. The cell biology of thrombospondin-1. Matrix Biol2000;19:597–614.

85. Petropoulo C, Trougakos I, Kolettas E, Toussaint O, Gonos E. Clusterin/apolipopro-tein J is a novel biomarker of cellular senescence that does not affect the proliferativecapacity of human diploid fibroblasts. FEBS Lett 2001;509:287–97.

86. Zhou W, Janulis L, Park II, Lee C. A novel anti-proliferative property of clusterin inprostate cancer cells. Life Sci 2002;72:11–21.

87. Bettuzzi S, Scorcioni F, Astancolle S, Davalli P, Scaltriti M, Corti A. Clusterin(SGP-2) transient overexpression decreases proliferation rate of SV40-immortalizedhuman prostate epithelial cells by slowing down cell cycle progression. Oncogene2002;21:4328–34.

88. DiGiovanni J. Multistage carcinogenesis in mouse skin. Pharmacol Ther 1992;54:63–128.

89. Redondo M, Villar E, Torres-Munoz J, Tellez T, Morell M, Petito CK. Overexpres-sion of clusterin in human breast carcinoma. Am J Pathol. 2000;157:393–9.

90. Parczyk K, Pilarsky C, Rachel U, Koch BC. Gp80 (clusterin; TRPM-2) mRNA levelis enhanced in human renal clear cell carcinomas. J Cancer Res Clin Oncol 1994;120:186–8.

91. Steinberg J, Oyasu R, Lang S, et al. Intracellular levels of SGP-2 (Clusterin) correlatewith tumor grade in prostate cancer. Clin Cancer Res 1997;3:1707–11.

92. Dai CY, Furth EE, Mick R, et al. p16(INK4a) expression begins early in human colonneoplasia and correlates inversely with markers of cell proliferation. Gastroenterology2000;119:929–42.

93. Miyake H, Gleave M, Arakawa S, Kamidono S, Hara I. Introducing the clusterin geneinto human renal cell carcinoma cells enhances their metastatic potential. J Urol2002;167:2203–8.

94. Cui W, Fowlis DJ, Bryson S, et al. TGF-�1 inhibits the formation of benign skintumors, but enhances progression to invasive spindle carcinomas in transgenic mice.Cell 1996;86:531–42.

95. Qian X, Tuszynski GP. Expression of thrombospondin-1 in cancer: a role in tumorprogression. Proc Soc Exp Biol Med 1996;212:199–207.

96. Miyake H, Hara I, Kamidono S, Gleave ME. Synergistic chemosensitization andinhibition of tumor growth and metastasis by the antisense oligodeoxynucleotidetargeting clusterin gene in a human bladder cancer model. Clin Cancer Res 2001;7:4245–52.

97. Gleave ME, Miyake H, Zellweger T, et al. Use of antisense oligonucleotides targetingthe antiapoptotic gene, clusterin/testosterone-repressed prostate message 2, to en-hance androgen sensitivity and chemosensitivity in prostate cancer. Urology 2001;58:39–49.

3136




2004;64:3126-3136. Cancer Res Andrei Thomas-Tikhonenko, Isabelle Viard-Leveugle, Michael Dews, et al. Vivo

in and Carcinogenesis in VitroWhich Inhibits Their Growth Myc-Transformed Epithelial Cells Down-Regulate Clusterin,

Updated version

http://cancerres.aacrjournals.org/content/64/9/3126

Access the most recent version of this article at:

Cited articles

http://cancerres.aacrjournals.org/content/64/9/3126.full#ref-list-1

This article cites 96 articles, 35 of which you can access for free at:

Citing articles

http://cancerres.aacrjournals.org/content/64/9/3126.full#related-urls

This article has been cited by 8 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] at

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

Permissions

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

.http://cancerres.aacrjournals.org/content/64/9/3126To request permission to re-use all or part of this article, use this link



http://cancerres.aacrjournals.org/content/64/9/3126.full#ref-list-1

http://cancerres.aacrjournals.org/content/64/9/3126.full#related-urls

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

mailto:[email protected]



myc-transformed epithelial cells down-regulate clusterin ... · terin, an extracellular...

Documents