PIBF (PROGESTERONE INDUCED BLOCKING FACTOR) IS OVEREXPRESSEDIN HIGHLY PROLIFERATING CELLS AND ASSOCIATED WITH THECENTROSOMEMargit LACHMANN
1, Dieter GELBMANN1, Endre KALMAN
3, Beata POLGAR2, Michael BUSCHLE
1, Alexander VON GABAIN1
Julia SZEKERES-BARTHO2,4 and Eszter NAGY
1*1Intercell AG, Vienna, Austria2Department of Medical Microbiology and Immunology, University of Pecs, Pecs, Hungary3Department of Pathology, University of Pecs, Pecs, Hungary4Reproductive and Tumor Immunology Research Group of the Hungarian Academy of Sciences, Pecs, Hungary
Key words: PIBF; alternative splicing; centrosomal localization
PIBF (Progesterone Induced Blocking Factor) was discovered inthe context of reproductive immunology as a secreted product ofhuman pregnancy lymphocytes and described as an immunomodu-latory molecule relevant for the maintenance of pregnancy.1 PIBFexerts immunosuppressive effects by inhibiting NK cell activity,polarizing immune responses towards TH2 dominance and down-regulating effector cytokine responses.1–4 Recently, evidence hasemerged that PIBF is expressed by malignant cells as it wasdetected in the culture supernatant of the human mammary carci-noma cell line MCF-7 upon induction with insulin.5 This suggeststhat PIBF produced by a tumor cell line might be secreted similarlyto pregnancy or activated lymphocytes in culture. In addition to thesecreted form of PIBF, a higher molecular weight intracellularisoform was detected in MCF-7 cells independently of progester-one or insulin stimulation. The size of the intracellular PIBFprotein is 90 kDa and corresponds to that predicted by the openreading frame of the PIBF mRNA identified during the cloning ofthe gene.5 Recent genetic studies indicated pibf among the candi-date genes for breast cancer predisposition and cancer progressionin the 13q21–q22 chromosomal region.6 Considering its immuno-suppressive effects, PIBF secreted by undifferentiated, malignantcells may contribute to tumor development. There are alreadyknown examples of proteins like the regeneration and tolerance
factor (RTF) or the oncofoetal protein 5T4 that are expressed inpregnancy tissues as well as in malignantly transformed cells.7–10
RTF, similarly to secreted PIBF, exerts an immunosuppressiveactivity and is expressed during pregnancy by the placenta.7
Here we present evidence that tumor cells overexpress PIBFrelative to normal tissues. We also demonstrate that different formsof the molecule are generated by alternative mRNA splicing bothin normal and malignant cells. The full-length 90 kDa protein,which seems to be the most abundant PIBF form, is intracellularand associated with the centrosome.
MATERIAL AND METHODS
Cells and cell culturePeripheral blood mononuclear cells (PBMCs) were isolated
from human blood by density separation over Ficoll-Hypaque.HeLa, MCF-7, NK-92 and K562 cell lines were purchased fromATCC. MDA-MB-468, T47D, SK.BR-3 and OVCAR-3 were akind gift from T. Grunt (University of Vienna, Austria). HEK 293cells were kindly provided by P. Kovarik (University of Vienna,Austria). All cell lines were cultured in MEM medium supple-mented with 10% heat inactivated fetal calf serum, 100 U/mlpenicillin and 100 �g/ml streptomycin at 37°C in a humidifiedatmosphere of 5% CO2.
Human tumor samplesTissues from breast tumors were obtained from the Department
of Surgery II at the University of Pecs, Hungary with a procedurethat complied with ethical standards required at this institute. Thetissues were collected from women with confirmed histopatholog-ical diagnosis of breast cancer who were undergoing lumpectomyor mastectomy. A section of the tissue obtained as part of thatrequired for diagnostic purposes, and that would have otherwisebeen discarded, was provided to the Department of Pathology,University of Pecs. Tumors were frozen within 5 min after removalfrom the patients. Progesterone- and estrogen-receptor (PR, ER)status was determined by standard immunohistochemical stainingof parallel formalin-fixed paraffin embedded tissue blocks withanti-PR and anti-ER antibodies.
Grant sponsor: Hungarian National Research Fund; Grant number:OTKA T031737; Grant sponsor: Hungarian Ministry of Health; Grantnumber: ETT 045/2003; Grant sponsor: Hungarian Academy of Sciences
Publication of the International Union Against Cancer
AntibodiesAnti-PIBF rabbit immune serum was generated by immuniza-
tion with a 54 amino acid peptide corresponding to exon 17 of thePIBF mRNA. Anti-PIBF-exon17 antibodies were affinity purifiedwith the 54 amino acid long peptide cross-linked to AminoLink�Plus Coupling Gel (Pierce, Rockford, IL). IgG from rabbit preim-mune serum was isolated with UltraLink� immobilized protein G(Pierce, Rockford, IL) and used as isotype control. Anti-PIBF-exon17 and rabbit preimmune IgGs (both 1 mg/ml) were diluted1:200 for immunostaining (IS) and 1:1,000 for Western blotting(WB). The mouse monoclonal anti-human Golgin-97 CDF4 anti-body (Molecular Probes, Leiden, Netherlands) was used at 1:200dilution for IS and at 1:1,000 for WB. The mouse monoclonalanti-FLAG� M2 antibody (Sigma Chemical Co., Vienna, Austria)was diluted 1:1,000 for IS and 1:2,000 for WB. The mousemonoclonal anti-�-tubulin antibody (GTU-88) (Sigma ChemicalCo., Vienna, Austria) was used at 1:1,000 and 1:10,000 dilutionsfor IS and WB, respectively. The mouse monoclonal anti-�-tubu-lin antibody (B-5-1-2) (Sigma Chemical Co., Vienna, Austria) wasdiluted 1:6,000 for IS. The goat polyclonal anti-lamin B1 antibody(M-20) (Santa-Cruz, Heidelberg, Germany) was diluted 1:250 forWB. Secondary antibodies for WB were donkey anti-rabbit IgHRP linked (Amersham Biosciences, Uppsala, Sweden), rabbitanti-mouse Ig HRP linked (DAKO) and rabbit anti-goat IgG-HRP(Sigma Chemical Co., Vienna, Austria), used at 1:5,000 dilution.Secondary antibodies for IS were Alexa Fluor� 488-labeled goatanti-rabbit IgG (H�L) highly cross-adsorbed and Alexa Fluor�594-labeled goat anti-mouse IgG (H�L) (both Molecular Probes,Leiden, Netherlands) used at 1:750 dilution. Antibodies detectingPR and ER by immunohistochemisty (1A6 and 1D5, respectively)were purchased from DAKO (Vienna, Austria) and used accordingto the manufacturer’s instructions. The specificity of the anti-PIBFantibodies were tested by Western blotting using recombinantPIBFs, as well PIBF negative (or low) and PIBF positive tissues.
RNA isolationTotal RNA was isolated from approximately 107 cells (PBMCs,
cell lines) or 50–200 mg tissue (breast tumors, human and mouseplacenta, mouse embryo) using TRIzol reagent (Invitrogen, Lofer,Germany), following the manufacturer’s instructions. Total RNAsamples from tissues or cell lines were treated with 5 units of RQ1RNase-free DNase (Promega, Mannheim, Germany) in the pres-ence of 40 mM Tris-HCl (pH 8.0), 10 mM MgSO4, 1 mM CaCl2and 40 units of rRNasin ribonuclease inhibitor (Promega, Mann-heim, Germany) for 30 min at 37°C followed by heat inactivationat 65°C for 15 min.
RT-PCRTen micrograms of DNA-free total RNA was reverse tran-
scribed using oligo (dT)15 primers and SuperScript II RNase H�
reverse transcriptase (Invitrogen, Lofer, Germany) following themanufacturer’s instructions. The reaction was incubated at 42°Cfor 50 min before heat inactivation at 70°C for 15 min. The wholereaction (20 �l) was incubated with 2 units RNaseH at 37°C for 20min. One tenth of the synthesized cDNA or 15 ng of humanstomach and uterus matched cDNA pairs (HP104S/HP104U; BDBiosciences Clontech, Heidelberg, Germany) were amplified byPCR using primers listed in Table I. The cycling parameters thatused the primer pair Exon1/Exon18 were: 94°C for 1 min, then94°C for 20 sec and 68°C for 3 min for 20–45 cycles, followed byanother extension at 68°C for 10 min. For all other primer com-binations, the parameters were as follows: 94°C for 1 min, then94°C for 20 sec, 55°C for 30 sec and 72°C for 1–3 min for 20–30cycles, followed by another extension at 72°C for 10 min. Therelative abundance of the PIBF mRNA in cell lines and tissues wasassessed by semiquantitative RT-PCR, by comparing band inten-sities generated using 3 different amounts of input RNA for thereverse transcription reaction, as well as 5 different cycle numbersduring PCR. Ribosomal protein S9 or �-actin mRNAs served asloading controls.
Identification of different PIBF mRNA isoformsRT-PCR products were generated from total cellular RNA iso-
lated from different cell lines, human PBMCs, human and mouseplacenta and mouse embryo. A mouse testis cDNA library (num-ber 7455-1, BD Biosciences Clontech, Heidelberg, Germany) wasused for the cloning of the murine homologue of the full-lengthPIBF mRNA by 5� RACE, as well as for identification of alterna-tively spliced mRNAs in this tissue. PCR products generated withPIBF gene specific primers (listed in Table I) were directly clonedinto the pCR 2.1-TOPO vector using a TOPO TA Cloning kit(Invitrogen, Lofer, Germany). Clones carrying plasmids with in-serts of different length were selected for sequencing with standardM13 forward and reverse primers; mRNA structures were deter-mined by aligning RNA sequences to the pibf gene (http://www.ncbi.nlm.nih.gov/genome/guide/human/).
Cell fractionationFrozen breast cancer tissues (100–200 mg) were cut into small
pieces and mixed with 2 ml of lysis buffer (10 mM HEPES/NaOH,pH 7.4, 10 mM NaCl, 3 mM MgCl2, 0.05% NP-40, 1 mM EGTA,1 mM phenylmethylsulfonyl fluoride and Complete™ EDTA freetablets). The samples were pressed through a metal sieve and the
TABLE I – PRIMERS FOR PCR ANALYSIS
Name Orientation Sequence
hu Exon 1 S GGTTGTCTTGGTTACGGGTCCTAACGGTCChu Exon 1 nested S CTGCCTTGAAATCCCTTGTTGAGGhu Exon 2 S ATGTCTCGAAAAATTTCAAAGGAGTChu Exon 2-BamHI S AAGGGATTCATGTCTCGAAAAATTTCAAAGGAGTChu Exon 9 S CCAACCAAGAAATTGATCAACTTCGAAATGCCTCTAGGGhu Exon 13 AS GGCTGTTGTGGGAACATTAGChu Exon 15 S GACAGAGCCAATTCGCTATTAAACCAGACTCAACAGChu Exon 18 nested AS CCATCAGAAAAAGAACTTCAGGhu Exon 18 AS GCTGAGTACACGATTAAGCTGAATTTTGTTTTCCATCAGmu Exon 1 S GGTTGTCTTGGTTACGGGTCCTAACGmu Exon 1 nested S CCTGAGCTGCTGACTGAAGACGmu Exon 2 S ATGTCTCGCAAAATTGCCAAGGAACCmu Exon 18 nested AS CCAACAGACATTGCACTTCAGGmu Exon 18 AS CAAGTACATGATTGAAGTGAGCCTTTCTCCChu Exon 18-FLAG-BamH AS TCTAAGGGATCCCTACTTGTCATCGTCGTCCTTGTAGTCGGTCTTCATCTTTTGTTTCTTAGACCRibosomal protein S9 5� S GATGAGAAGGACCCACGGCGTCTGTTCGRibosomal protein S9 3� AS GAGACAATCCAGCAGCCCAGGAGGGACA�-actin 5� S TGAAGTCTGACGTGGACATC�-actin 3� AS ACTCGTCATACTCCTGCTTGGAPDH 5� S GACCCCTTCATTGACCTCAACTACAGAPDH 3� AS GGTCTTACTCCTTGGAGGCCATGT
52 LACHMANN ET AL.
flow-through was filtered through a cell strainer. The nuclei wereseparated from the cytosolic fraction by density gradient centri-fugation at 400g for 10 min. The pellet containing nuclei andinsoluble material was washed once with lysis buffer before it wasresuspended in 100 �l urea containing IEF sample buffer (8 Murea, 4% CHAPS, 0.5% SDS and 100 mM DTT). The supernatantfractions containing cytosolic and extracellular material were pre-cipitated with 3-fold vol of abs ethanol over night at �20°C andcentrifuged. The dried pellet was dissolved in 100–300 �l IEFsample buffer.
Western blot analysisProtein samples were separated by 10% SDS-PAGE, transferred
to a Hybond ECL membrane (Amersham Biosciences, Uppsala,Sweden) and blocked with 5% nonfat dry milk in PBS containing0.1% Tween 20 (PBST) for 1 hr. The primary and secondaryantibodies were diluted in blocking buffer and incubated with themembranes for 1 hr at room temperature. The blots were devel-oped with the ECL Plus Western blotting detection system (Am-ersham Biosciences, Uppsala, Sweden) and exposed to HyperfilmECL (Amersham Biosciences, Uppsala, Sweden).
ImmunohistochemistryParaffin-embedded breast tumor and corresponding normal tis-
sue sections (Asterand, Detroit, MI) were used to study PIBFexpression. Samples were deparaffinized and rehydrated by treat-ing twice with xylene for 5 min followed by addition of absethanol 2� for 3 min and washing twice with PBS. Antigens wereretrieved by using the NaCitrate antigen retrieval from DAKOfollowing the manufacturer’s user manual. Staining was doneusing the DAKO Envison™ IHC Kit. Peroxidase Blocking Re-agent was added for 5 min and rinsed with PBS. The primaryantibodies used were affinity purified rabbit polyclonal antibodiesdirected against the exon 17 of PIBF and preimmune rabbit IgG asnegative control, both diluted 1:200 in PBS and incubated for 20min before addition of substrate-chromagen for 15 min. Sectionswere counterstained with hematoxylin.
Localization studies and immunofluorescence microscopyCells were grown in chamber slides (Nalgene, Rochester, NY).
In case of analyzing the effect of nocodazole and Brefledin A onthe localization of endogenous PIBF, HeLa cells were treated with5 �g/ml nocodazole (Sigma Chemical Co., Vienna, Austria) for 1or 4 hr. Cells were washed once with PBS and fixed with a fixingsolution consisting of 95% ethanol and 5% acetone for 8 min at�20°C. Cells were washed twice with PBS and blocked with 0.2%BSA in PBS for 30 min.
The appropriate primary antibody was diluted in blocking bufferand incubated for 40 min at room temperature. Cells were washed3 times with PBS, incubated with the appropriate fluorescentsecondary antibody diluted in blocking buffer for 30 min in darkfollowed by washing 3 times with PBS and counterstained with 0.1�g/ml DAPI in PBS. Slides were examined with a fluorescence-equipped microscope (Zeiss, Jena, Germany). Images for illustra-tion purposes were obtained using a digital camera (SPOT Diag-nostic Instruments, Sterling Heights, MI).
Expression of exogenous PIBF in human cells in vitroHEK 293 (human embryonic kidney) cells stable transfected
with the regulatory plasmid pVpRXR of an ecdysone-induciblemammalian expression system (Invitrogen, Lofer, Germany) ex-pressing the ecdysone receptor were kindly provided by PavelKovarik (University of Vienna). Cells were cultured until 90–95%confluence was reached before transiently transfected with thepIND/Hygro PIBF constructs. To create these constructs, PCR wasperformed using the primers hu Exon 2-BamHI and hu Exon18-FLAG-BamHI (Table I), the latter primer containing DNAsequences corresponding to the amino acids of the FLAG-tag.Vectors carrying the genes for the 90 and 35 kDa PIBF forms wereused as templates. The PCR products were directly cloned into the
pCR 2.1-TOPO vector using a TOPO TA Cloning kit (Invitrogen,Lofer, Germany). After selection of positive clones, plasmid DNAwas prepared, digested with BamHI and the pibf gene inserts weresubsequently ligated into the pIND/Hygro vector. Twenty micro-grmas of DNA was used for transfection of cells (in 75 ml cultureflasks) with 20 �l of lipofectamin (LF2000 Reagent, Invitrogen,Lofer, Germany) in serum free medium (OPTI-MEM I ReducedSerum Medium, Invitrogen, Lofer, Germany) for 24 hr. The tran-siently transfected cells were induced with 10 �M Ponasterone A,a synthetic analog of ecdysone. Expression of the exogenous PIBFwas monitored by Western blotting using anti-FLAG tag antibod-ies.
Centrosome isolationIsolation of the centrosomes was done as described by Hsu and
White.11 Exponentially grown HEK 293 cells were incubated withculture medium containing 0.6 �g/ml nocodazole and 1 �g/mlcytochalasin D for 1 hr at 37°C, to depolymerize microtubulefilaments and actin, respectively. The cells were harvested andresuspended in 1ml lysis buffer (1 mM HEPES pH�7.2, 0.5%NP-40, 0.5 mM MgCl2, 0.1% 2-mercaptoethanol, 1 mM phenyl-methylsulfonyl fluoride and Complete™ EDTA free tablets). Theswollen nuclei and chromatin aggregates were removed by cen-
FIGURE 1 – PIBF mRNA is expressed by malignant cells both invitro and in vivo. (a) RT-PCR analyses of in vitro cultured malignantlytransformed human cell lines (indicated above) were performed withPIBF gene specific primers as described in the Material and Methods.(b) Semiquantitative PCR analysis of cDNA samples from stomachtumor (ST) and corresponding normal tissue (SN), uterus tumor (UT)and corresponding normal tissue (UN) were performed with PIBFprimers annealing in exon 1 and 18 (upper panel) and exon 2 and 18(middle panel), respectively. Ribosomal protein S9-specific primerswere used for loading control (lower panels).
53EXPRESSION AND CENTROSOMAL LOCALIZATION OF PIBF
trifugation at 2,500g for 10 min. The nuclear fraction was washedonce with lysis buffer before it was dissolved in IEF sample buffer.The supernatant was filtered through a 50 �m nylon mesh. HEPESwas adjusted to 10 mM followed by a DNase I (2 units/ml)treatment for 30 min on ice. The lysate was underlaid with 1 ml60% sucrose solution (60% wt/wt sucrose in 10 mM PIPES, pH7.2, 0.1 % Triton X-100, 0.1 % 2-mercaptoethanol). Centrosomeswere sedimented into the sucrose cushion by centrifugation at10,000g for 1 hr. The 200 �l fraction from the bottom of thesucrose gradient was diluted with 1 ml 10 mM PIPES buffer (pH7.2). Centrosomes were recovered by centrifugation at 16,000g for10 min in a microfuge. The whole cell lysate was prepared bydissolving the cells in IEF-sample buffer followed by brief soni-cation. The proteins were analyzed by Western blotting.
RESULTS
Overexpression of PIBF mRNA in malignant tumorsPIBF was identified as an immunomodulatory molecule secreted
by pregnancy lymphocytes or activated PBMCs upon progesteronetreatment.1 However, previous protein expression studies revealedthat PIBF was also expressed by the progesterone receptor positivehuman mammary carcinoma cell line, MCF-7.5 An extensiveBLAST search analysis of the cloned PIBF mRNA sequence usingthe human dEST library database (http://www.ncbi.nlm.nih.gov/BLAST, est_human database) suggested a more universal expres-sion and revealed that PIBF mRNA is expressed not only inlymphocytes but also in many other cell types, mainly in embry-onic tissues, pregnant uterus, placenta and testis ( 30% of allentries). PIBF mRNA was most frequently (50% of all entries)
FIGURE 2 – Exon structure and predicted molecular signatures of PIBF mRNA. Schematic representation of the full-length PIBF mRNA (a)and its splicing variants found in stomach (b) and uterus (c) tumors. Predicted domains within the protein are indicated: NLS, nuclear localizationsignal (http://psort.nibb.ac.jp/form2.html), LeuZip, leucine zipper (http://psort.nibb.ac.jp/form2.html, http://us.expasy.org/prosite), bZIP, basicleucine zipper (http://www.sanger.ac.uk/Software/Pfam/search.shtml), PEST signal (http://www.at.embnet.org/embnet/tools/bio/PESTfind/),coiled-coil domain (http://smart.embl-heidelberg.de/).
TABLE II – PIBF mRNA VARIANTS FOUND BY RT-PCR IN DIFFERENT HUMAN AND MURINE TISSUES
Length ofmRNA1 Exons present in mRNA Predicted MW of
identified in cDNA libraries prepared from human malignant tu-mors with different tissue origin (only entries containing at least 2exons of the PIBF mRNA sequence or the last exon with polyA tailwere included). This analysis indicated that PIBF mRNA might bepreferentially expressed in tissues with high rate of proliferation,such as embryonic cells and pregnancy tissues, testis as well as inmalignant primary tumors and cell lines. Northern blot analysis ofnormal human tissues ascertained the presence of an approxi-mately 3,200 nt long mRNA (data not shown), which was themajor RNA band and was compatible with the size predicted fromthe cloned cDNA (GenBank Account number NM006346).5 Eachof the 23 normal adult tissues tested (using commercially availablepremade Northern blots purchased from Clontech; 7760-1, 7766-1and7767-1) showed expression of the PIBF mRNA at variable butdetectable levels. The highest signal was detected with RNAsamples isolated from testis, followed by thyroid tissue (data notshown). Strong expression of the PIBF mRNA in testis relative to7 other normal tissues was also reported previously by an inde-pendent research group.6
These experimental and in silico findings prompted us to initiatefurther studies on the expression of PIBF in tumor cells. In orderto determine whether PIBF mRNA is expressed in malignantlytransformed cells, cDNA samples generated from total RNA ofdifferent human tumor cell lines, such as HeLa, OVCAR-3, SK-BR3, HEK 293, K562, T47D, MCF-7 and NK-92, were subjected
to RT-PCR analysis. All cell lines exhibited expression of PIBFmRNA using internal gene specific primers (Fig. 1a).
To measure PIBF expression in tumors compared to normal tissuecounterparts, semiquantitative PCR was performed on tumor/normalcDNA pairs from a gastric and an endometrial adenocarcinoma, aswell as from normal stomach and uterus tissues of the same tumorbearing patients. The results of the PCR analysis indicated overex-pression of the PIBF mRNA in the tumor tissues (Fig. 1b). UsingPIBF gene specific primers annealing to the 5�UTR (exon 1) and3�UTR (exon 18) of the mRNA, a significantly higher level offull-length PIBF mRNA (2,560 nt) was found in the stomach tumorrelative to its normal tissue counterpart. A markedly higher level ofPIBF mRNA was detected in the uterus tumor as well, yet representedby a smaller mRNA with approximately 1,000 nt length (Fig. 1bupper panel). This finding prompted us to look more carefully foralternative mRNA forms. Using a different primer pair annealing tothe START codon region of the ORF (exon 2) and to the 3�UTR(exon 18), we identified an approximate 950 nt long mRNA presentonly in the stomach tumor sample (Fig. 1b middle panel). This primerpair did not amplify the short PIBF mRNA found in the uterus tumor,suggesting that exon 2 sequences were missing from this PIBF tran-script.
Alternative splicing of PIBF mRNATo confirm the specificity of the amplified cDNAs and to
determine the structure of the corresponding mRNAs, the PCRfragments were cloned and sequenced. Sequence analysis revealedthat the 2 shorter mRNA forms were the result of alternativesplicing, generated by a perfect exon skipping mechanism, exclud-ing exons 2-12 or exons 6-16 in the case of the uterus and stomach
FIGURE 3 – PIBF mRNA and its different forms are overexpressed inhuman primary breast tumors independent of the progesterone receptorstatus. (a) Semiquantitative RT-PCR of cDNAs from a malignantbreast tumor (T1) and pooled normal breast tissues (N) was performedusing different amounts of input RNA for reverse transcription. ForPCR analysis PIBF primers (annealing to exon 15 and 18) and �-actinprimers were used. (b) Semiquantitative PCR analyses of cDNAsamples derived from different breast tumors (T1–T5) and poolednormal breast tissues (N) were performed using PIBF primers (anneal-ing to exon 9 and 13), �-actin and GAPDH primers.
FIGURE 4 – PIBF protein expression in breast tumors. Malignantbreast tumor tissues (T6–T8) with different progesterone and estrogenreceptor status were fractionated as described in Material and Meth-ods. Proteins of the nuclear/cytoskeletal (NUC/CS) and cytoplasmic/extracellular (CP/EC) fractions were separated by 10% SDS-PAGEand subjected to Western blot analysis using peptide affinity purifiedanti-PIBF-exon17 antibodies.
55EXPRESSION AND CENTROSOMAL LOCALIZATION OF PIBF
tumor forms, respectively (Fig. 2). The exon 1�13-18 short formfound in the uterus tumor was predicted to encode for the C-terminal 87 amino acids of PIBF, resulting in an approximately 10kDa protein. The exon 1-5�17-18 mRNA form isolated from thegastric adenocarcinoma encoded for a protein with a predictedmolecular weight of 35 kDa, containing the N-terminal and C-terminal amino acid sequences of the full-length PIBF protein.Most of the predicted molecular signatures, such as the 2 NLSs(nuclear localization signals), the Leucine zipper(s) and bZIPdomain were missing from this splice variant mRNA.
To obtain a better molecular characterization of PIBF expres-sion and to learn whether alternative splicing of the PIBF mRNAwas characteristic of malignant cells, a series of RT-PCR analyseswere performed. The mRNA corresponding to the full-length ORFpredicted from the genome data of the human chromosome 13 wasthe most abundant transcript, especially in normal tissues. Analyz-ing human PBMCs, placenta and a mammary carcinoma cell line(MCF-7), we found a number of different forms all generated byperfect exon skipping mechanism (Table II). This complex patternof PIBF mRNA splicing was also confirmed with murine tissues,such as placenta, whole embryo and testis. The full-length mousePIBF mRNA (also deduced from the genome data of the mousechromosome 14) that we cloned by 5�RACE shared 88% identitywith the human counterpart. The most frequently identified splicevariant encoding for the 35 kDa PIBF protein was detected notonly in stomach tumor, but in many other tissues such as PBMCs(male, female and pregnancy), human and murine placenta andtotal murine embryo (Table II). Almost all identified mRNAs(except one mouse form found in embryo) contained exon 17(besides exon 1 and 18 where the primers used for amplificationwere annealing), while other exons were alternatively included(Table II).
Multiple forms of PIBF are expressed in breast tumorsindependently from the progesterone receptor status
It has been well established that breast adenocarcinomas areheterogeneous with regard to progesterone receptor expression(PRs). PIBF was identified as an immunomodulatory moleculeinduced by progesterone in PR expressing lymphocytes1 and inaddition was recently associated with breast cancer susceptibilityin genetic studies.6 Therefore further experiments were designed tocarefully address the progesterone-dependence of PIBF expressionby analyzing mRNA levels, the pattern of alternative splicing, aswell as the protein expression in breast tissues. Total RNA isolatedfrom several human primary mammary adenocarcinoma tissuesrepresenting progesterone/estrogen receptor negative (PR�/ER�)and positive (PR�/ER�) tumors, as well as pooled normal breasttissue was analyzed by RT-PCR using different PIBF gene specificprimers under experimental conditions that allowed semiquantita-tive analysis (Fig. 3a). �-actin mRNA was used as a loadingcontrol, and GAPDH mRNA served as a positive control since itsexpression in human primary breast tumors has been shown to bedependent on the PR/ER status.12 During these studies, four im-portant observations were made. First, PIBF mRNA levels wereelevated in breast tumors relative to normal breast tissue whenmeasured with internal gene specific primers (Fig. 3b). Second,similarly to other tumor cells and highly proliferating tissues,mammary adenocarcinoma cells also expressed several variants ofPIBF mRNAs generated by a perfect exon skipping mechanism.The identified forms are listed in Table II. Third, progesteronesensitivity (presence of the PR) of the tumors influenced neitherthe level of PIBF mRNA nor the pattern of alternative splicing.Fourth, not only malignant breast tissues but also RNA isolatedfrom a breast fibroadenoma showed expression and alternativesplicing of PIBF.
We also investigated the pattern of expression of PIBF proteinand its variants in breast tumors. Tumor tissue blocks were dis-rupted in the presence of Triton X-100, and the soluble (containingmainly extracellular and cytoplasmic proteins) and insoluble(mainly nuclear and cytoskeleton-associated proteins) fractions
were analyzed by Western blotting. In order to detect all possiblePIBF forms, we generated a hyperimmune serum against a 54amino acid synthetic peptide corresponding to exon 17 of PIBF,which was present in almost all splice variant mRNAs identified.Peptide affinity purified anti-PIBF-exon17 antibodies from theresulting antiserum recognized recombinant PIBF in Western blots
FIGURE 5 – Specific staining of breast tumor tissues with anti-PIBFantibody. Immunohistochemical stainings of infiltrating ductal adeon-carcinoma (a–c) and corresponding normal (d) tissues are shown, asrepresentative samples from the TMA analysis. Tumor and normaltissues were stained with peptide affinity purified anti-PIBF-exon17antibodies (a,b,d) or with rabbit IgGs, as isotype control (c). The slideswere counterstained with hematoxylin.
FIGURE 6 – Subcellular localization of endogenous PIBF. Humancell lines were analyzed by immunofluorescent microscopy usingpeptide affinity purified anti-PIBF-exon17 antibodies. HeLa cells werestained with anti-PIBF-exon17 (a) and counter-stained with DAPI (b),the merged image of a and b (c) and that of HeLa cells stained withisotype control (d) are shown. (e,f) Merged images of K-562 andOVCAR-3 cell lines, respectively, stained with anti-PIBF-exon17 andcounter-stained with DAPI.
56 LACHMANN ET AL.
(data not shown). Using these antibodies, we detected severalprotein bands with different electrophoretic mobility. The mostabundant PIBF form corresponded to the full-length protein with apredicted molecular weight of 90 kDa and was found mainly in theinsoluble nuclear/cytoskeletal fractions of the breast tumor tissues(Fig. 4). A band with the same molecular weight was identified inthe nuclear/cytoskeletal fractions of many different cell lines,human lymphocytes, human and mouse placenta, embryo andtestis (data not shown). In addition, several other anti-PIBF-exon17 reactive bands were found both in the nuclear/cytoskeletaland cytoplasmic/extracellular fractions. No characteristic patternor form of PIBF that correlated with progesterone receptor expres-sion was found using the exon 17 specific anti-PIBF antibodies.
Tissue microarray analysis (TMA) of breast tumor/normaltissues
An immunohistochemical analysis with tissue microarray wasperformed to get a global picture of PIBF expression in breasttumors vs. normal tissue counterparts. The slide (provided byAsterand, Detroit, MI) contained 90 good quality cores fromhuman breast tissues, 57 derived from adenocarcinoma and 33from normal tissues (paired with tumor tissue counterparts). Im-
munohistochemical analysis using the anti-PIBF-exon17 antibodyrevealed that 40% of the tumor-derived and none of the normalsamples showed specific PIBF staining. As a representative exam-ple of this analysis, a tumor/normal tissue pair is shown in Figure5. The breast adenocarcinoma-derived sample (Fig. 5a,b) dis-played positive staining, while corresponding normal breast tissue(from the same patient) was negative (Fig. 5d), confirming atumor-specific PIBF positivity. The specific PIBF staining was nothomogenous and mainly associated with tumor cells, while stromaand connective tissues were negative. The positive reaction wasobserved within the cytoplasm with perinuclear and cytoplasmicspots (Fig. 5b).
Subcellular localization of PIBFThe appearance of the PIBF positivity in breast cancer tissues
prompted us to analyze the subcellular localization in more detail.Immunofluorescence microscopic analyses were performed onseveral different transformed human cell lines (also tested for theexpression of PIBF mRNA), among those were HeLa, K-562 andOVCAR-3. All cell lines stained with the anti-PIBF-exon17 anti-body showed a similar staining pattern. We consistently found aspecific, highly fluorescent perinuclear spot signal surrounded by a
FIGURE 7 – PIBF is sensitive to nocodazole and colocalizes with the centrosomal protein -tubulin. (a) HeLa cells were immunostained withanti-PIBF-exon17 and anti Golgin-97 antibodies and counter stained with DAPI to visualize nuclei. Images of untreated cells (upper panel), andcells treated with 5�g/ml nocodazole for 4 hr (lower panel) are shown. (b) Mitotic HEK 293 cells were double immunostained withanti-PIBF-exon17 and anti--tubulin antibodies. Upper panels show prometaphase cells, and lower panels show cells in metaphase. Cells werecounterstained with DAPI to visualize nuclei.
57EXPRESSION AND CENTROSOMAL LOCALIZATION OF PIBF
diffuse or fine granular cytoplasmic staining with variable intensity(Fig. 6). In dividing cells, we consistently detected two PIBF dots(Fig. a,c, indicated by arrows).
This characteristic localization suggested a potential interactionof PIBF with a perinuclear organelle like the Golgi apparatusand/or the microtubule organizing center/centrosome. Double-im-munostaining of HeLa cells with anti-PIBF-exon17 and anti-Gol-gin-97 antibodies suggested that PIBF resided in the cytoplasmicregion also occupied by the Golgi apparatus (Fig. 7a upper panel).However, the PIBF dots were not completely overlapping with themore diffuse and irregularly shaped Golgin-97 signal and in somecells a clear juxtaposition was seen. Treatment of HeLa cells withBrefeldin A, a fungal metabolite that disrupts the structure andfunction of the Golgi apparatus, did not affect PIBF localization,although the altered appearance of the Golgi complex was obviousin these cells (data not shown). In contrast, treatment with nocoda-zole (at 5 �g/ml concentration for 4 hr), a microtubule-disruptingagent led to the time-dependent disappearance of the perinuclearPIBF dots, which paralleled with the disintegration of the Golgiapparatus (7a lower panel).
Involvement of the microtubular network in the localization ofPIBF was directly addressed by colocalization studies in HEK 293cells. The disappearance of the PIBF signal upon nocodazole treat-ment and the appearance of two PIBF signals on opposite sides of thenuclei in dividing cells suggested that PIBF might associate with themicrotubule organizing center (MTOC) or centrosome. This organelleis present in one copy in nondividing cells and replicates prior to celldivision to form the two poles of the mitotic spindle. Double immu-nostaining of mitotic (prometaphase and metaphase) HEK 293 cellsusing anti-�-tubulin and anti-PIBF-exon17 antibodies suggested thatPIBF localized to the poles of the mitotic spindle (data not shown).For a more detailed analysis of the localization, a coimmunostainingof PIBF with -tubulin, which is an essential component of thecentrosome, was performed. Closely spaced PIBF double spots inearly mitosis and two separated signals in late mitosis (Fig. 7b)overlapped with the -tubulin staining, indicating that PIBF colocal-ized with the centrosomes.
In order to further confirm the association of PIBF with thecentrosomes, these organelles were purified from HEK 293 cellsaccording to a published method.11 The fractionation procedureinvolved treatment of cells with nocodazole at a reduced concen-tration (0.6 g/ml) for 1 hr that allowed the separation of or-ganelles but retained proteins loosely associated with centrosomes.Western blot analysis demonstrated that PIBF was among theproteins copurified with centrosomes, similarly to -tubulin (Fig.8). The absence of lamin B1 (a strictly nuclear protein) andGolgin-97 (marker for the Golgi apparatus) proteins in this fractionexcluded nuclear and cytoplasmic contamination, respectively.
The abundance and predominance of the 90 kDa PIBF proteinon Western blots of cell lines5 and tumor tissues (Fig. 4) as well asits copurification with centrosomes suggested that it was mainlythe full-length PIBF that was detected in the perinuclear dots bythe anti-PIBF-exon17 antibody in the immunofluorescent studies.To verify this notion, a FLAG-tagged version of the 90 kDa PIBFand the most frequently isolated alternatively spliced species hav-ing exons 1-5�15-18 encoding for a 35 kDa protein were over-expressed in transiently transfected HEK 293 cells that constitu-tively expressed the heterodimeric ecdysone receptor (RXR andVgEcR) from Drosophila (from the regulatory plasmid pVgRXR).Western blot analysis of transfected HEK 293 cells using anti-FLAG or anti-PIBF Exon17 antibodies detected the intact 90 and 35kDa proteins upon induction with Ponasterone A (a synthetic ecdy-sone analog) (Fig. 9a). Coimmunostainings of these transfected cellswith anti-Golgin-97 and anti-PIBF-exon17 antibodies revealed thatthe exogenous full-length PIBF also appeared in a subcellular com-partment shared with the Golgi apparatus as it was seen before forendogenous PIBF in HeLa cells (Fig. 9b). In addition, coimmuno-staining with anti-PIBF-exon17 and anti-FLAG antibodies confirmeda similar localization of exogenous and endogenous full-length PIBF,whereas the 35 kDa variant was dispersed in the cytoplasm (Fig. 9c).These data suggest that the amino acid sequences required for theperinuclear localization are missing from the shorter PIBF protein andare located within exons 6–16.
DISCUSSION
In our work, we addressed the question of whether PIBF isoverexpressed by tumor cells and by highly proliferating normalcells. PCR analysis on numerous tumor cell lines, matched humanprimary tumor/normal cDNA samples, various breast tumors andpooled normal breast tissue have demonstrated that PIBF mRNA isexpressed in many normal tissues and its level is significantlyhigher in tumors relative to normal tissue counterparts. Further-more, we uncovered a complex pattern of alternative splicinggenerating different PIBF transcripts. We also observed a differentsplicing pattern in malignant cells that have higher level of thealternative PIBF transcripts, besides elevated levels of the predom-inant full-length mRNA, compared to their normal counterpart.This observation raises the possibility that the dysregulation ofalternative splicing of the pibf gene could be involved in tumorformation. Importantly, we found no correlation between sex hor-mone responsiveness and PIBF levels neither at mRNA nor atprotein levels when human primary breast tumor tissues with orwithout progesterone and estrogen receptor expression were ana-lyzed.
It has been widely recognized that alternative mRNA splicing isa hallmark of generating protein isoform diversity.13 PIBF wasoriginally described as a 34 kDa protein.1 Expression cloning withantibodies generated against this secreted form identified a tran-script corresponding to a 90 kDa protein encoded by 18 exons5 thatseems to be the predominant PIBF mRNA form. In Western blotanalyses using PIBF-specific antibodies, we consistently detectedan approximately 90 kDa protein, which represented the dominantPIBF protein form in all tissues and cell lines tested. Identificationof the exon 1-5�17-18 transcript encoding for a 35 kDa PIBF mayclarify the discrepancy between the sizes of the first identifiedsecreted (34 kDa) and the predominant cellular form (90 kDa). The
FIGURE 8 – PIBF copurifies with centrosomes. HEK 293 cells wereharvested in exponential growth phase and proteins of the centrosomalfraction (CEN) and total whole cell lysate (WCL) were isolated asdescribed in Material and Methods. The fractions were separated by10% SDS-PAGE and subjected to Western blot analysis using anti-bodies specific for PIBF, -tubulin, lamin B1 or Golgin-97 proteins, asindicated.
58 LACHMANN ET AL.
deletion observed in this transcript preserves the open readingframe of the full-length PIBF protein, and translation of the tran-script results in a PIBF isoform predicted to be 35 kDa andcontaining the N-terminal 223 and C-terminal 75 amino acids.
The mechanism of altered splice site selection in cancer cells isnot known, although it has been shown that mutations or sequencepolymorphisms in splice regulatory sequences, changes in splicingfactors and activation of particular signal transduction pathwaysmay modulate the use of alternative splice sites.14 Alterations inthe splicing pattern or splicing efficiency of several genes havebeen implicated in tumor progression (e.g., CD44, WT-1 andC-CAM1) and susceptibility to cancer (e.g., BRCA1).15–18
In this study, we show a higher level of expression of the PIBFprotein in breast tumors relative to normal breast tissue by immu-nohistochemistry. The TMA analysis suggested that at least 40%of breast tumors express a higher then normal level of PIBF. Inaddition, Polgar et al.19 have recently presented evidence thatmany different human primary tumors are highly positive withdifferent polyclonal and monoclonal anti-PIBF antibodies, includ-ing melanoma, epithelial carcinoma, lung tumors, colon adenocar-cinoma and bladder carcinoma among others. Moreover, abnor-mally high levels of PIBF can be detected in the urine of tumorpatients, which is dramatically reduced postsurgery or after che-motherapy.19 These data suggest that PIBF might serve as a tumormarker to monitor the progression of malignant diseases.
Although there are already possible roles assigned to thesecreted PIBF,1–4 the function of the intracellular full-length PIBFbeing the dominant protein form in proliferating cells has not yetbeen addressed. Association of PIBF with the centrosomes isclearly demonstrated in the present study. The nocodazole-sensi-tivity of the PIBF-centrosome interaction suggests that PIBF is notan integral component of the centrosome but rather a microtubule-associated protein. This is supported by preliminary copurificationstudies with -tubulin, where no direct interaction of PIBF with-tubulin was detected (data not shown).
The centrosome is the primary microtubule organizing-center(MTOC) in animal cells. Molecules that regulate centrosome du-plication have been identified and the list of proteins with centro-some-anchored activities, the functions of which have yet to bedetermined, is steadily expanding.20 It has been long recognizedthat disturbed centrosome duplication causes unequal segregationof chromosomes and, ultimately, tumorigenesis. Overexpressionand mutation of the centrosome-associated proteins observed intumors is correlated with centrosome amplification and aneuplo-idy.21 We have consistently detected in several cell lines thatconstitutive overexpression of the full-length PIBF led to theappearance of cells with abnormal nuclear morphology and ulti-mately to cell death (data not shown).
Interestingly, a number of proteins shown to be involved in tumor-igenesis are associated with the centrosome. The most prominent
FIGURE 9 – Full-length (90 kDa) PIBF colocalizes with centrosomes, while an alternative smaller (35 kDa) form has diffuse cytoplasmicdistribution. Subcellular localization of exogenous PIBF was examined in HEK 293 cells transiently transfected with plasmid vectors induciblewith ecdysone and coding for FLAG-tagged PIBF proteins. (a) Whole cell lysates of HEK 293 transfected for the expression of FLAG-PIBF90-kDa or FLAG-PIBF35kDa and induced for 0, 8 and 24 hr were separated on a 10% SDS-PAGE and subjected to Western blot analysis withanti-FLAG (upper panel) and anti-PIBF Exon17 (lower panel) antibodies. (b) Cells induced for 24 hr to express FLAG-PIBF90-kDa weredouble-stained with anti-PIBF-exon17 antibody and anti-Golgin-97 antibodies. (c) Cells induced for expression of the FLAG-PIBF90kDa (upperpanel) and that of the FLAG-PIBF35kDa (lower panel) were stained with anti-PIBF-exon17 and anti-FLAG antibodies and counterstained withDAPI.
59EXPRESSION AND CENTROSOMAL LOCALIZATION OF PIBF
example is the best-characterized breast cancer susceptibility geneBRCA111,22 that is linked to the development of breast and ovariancancers. Importantly, recent genetic studies indicated PIBF among thecandidate genes for breast cancer predisposition and cancer progres-sion.6 BRCA1 interacts with a variety of proteins regulating centro-some duplication, including BRCA2, CDK2-Cyclin A, CDK2-CyclinE, Gadd45, p21, p53 and retinoblastoma protein (RB).23,24 The moregeneral tumor suppressor gene p53 has also been shown to localize tocentrosomes and regulate centrosome duplication in a manner inde-pendent of its function as a transcription factor.25
Another important group of proteins associated with the centro-somes are kinases like Aurora and Polo-like kinases (Plk) that areimportant for the integrity and correct separation of the mitoticspindle and therefore also for accurate segregation of chromo-somes and correct cytokinesis (reviewed in references 26–29).Overexpression of these kinases can be observed in tumors andcorrelated with centrosome amplification and aneuploidy.30,31
Members of the transforming acidic coiled-coil (TACC) family ofproteins are implicated in cancer and have been shown to be concen-trated at centrosomes, where they regulate microtubule stability.32 Theonly shared structural feature among these proteins is their C-terminal200 amino acid long coiled coil region that is involved in theinteraction with microtubules. It has been suggested that coiled coilsare common features of many proteins in the pericentriolar material
(reviewed in reference 33). Importantly, a coiled coil structure is alsopredicted for the full-length PIBF protein.
Similarly to PIBF demonstrated in this study, the TACC1 andBRCA1 mRNAs also exist in alternatively spliced forms.17,18 Oneof the BRCA1 variants, deltaExon 11 missing two predicted nu-clear localization signals was found to be localized in the cyto-plasm, while the full-length protein was associated with the cen-trosome during mitosis and with the nucleus during interphase.34,35
This is comparable to the predominant 35 kDa variant of PIBF thatis also localized in the cytoplasm when it is overexpressed in HEK293 cells and lacks the two predicted nuclear localization signals.
At this point it is unclear whether the predominant 35 kDasplice-variant is identical to the previously found secreted 34 kDaPIBF, whether it also exerts immunomodulatory activity and uponwhich induction signal it is expressed or secreted. Future studiesneed to address the question whether overexpression of full-lengthPIBF and unbalanced expression of its alternative isoforms disturbcentrosomal function and thereby contribute to tumorigenesis.
ACKNOWLEDGEMENTS
We thank C. Ruckenbauer for helpful discussions and for pro-viding antibodies, C. Paar for technical help as well as T. Grundtand P. Kovarik for providing cell lines.
REFERENCES
1. Szekeres-Bartho J, Kilar F, Falkay G, Csernus V, Torok A, Pacsa AS.The mechanism of the inhibitory effect of progesterone on lympho-cyte cytotoxicity. I. progesterone-treated lymphocytes release a sub-stance inhibiting cytotoxicity and prostaglandin synthesis. Am J Re-prod Immunol Microbiol 1985;9:15–8.
2. Szekeres-Bartho J, Par G, Dombay Gy, Smart YC, Volgyi Z. Theanti-abortive effect of PIBF in mice is manifested by modulating NKactivity. Cell Immunol 1997;177:194–9.
3. Szekeres-Bartho J, Wegmann, TG. A progesterone-dependent im-muno-modulatory protein alters the Th1/Th2 balance. J Reprod Im-munol 1996;31:81–95.
4. Szekeres-Bartho J, Par G, Szereday L, Smart CY, Achatz I. Proges-terone and non-specific immunological mechanisms in pregnancy.Am J Reprod Immunol 1997;38:176–82.
5. Polgar P, Kispal G, Lachmann M, Paar C, Nagy E, Csere P, Miko E,Szereday L, Varga JP, Szekeres-Bartho J. Molecular cloning andimmunological characterization of a novel cDNA coding for PIBF.J Immunol 2003;171:5956–63.
6. Rozenblum E, Vahteristo P, Sandberg T, Bergthorsson JT. A genomicmap of a 6-Mb region at 13q21–q22 implicated in cancer develop-ment: identification and characterization of candidate genes. HumGenet 2002;110:111–21.
7. Lee GW, Boomer JS, Gilman-Sachs A, Chedid A, Gudelj L, RukavinaD, Beaman KD. Regeneration and tolerance factor of the humanplacenta induces IL-10 production. Eur J Immunol 2001;31:687–91.
8. Aslakson CJ, Lee G, Boomer JS, Gilman-Sachs A, Kucuk O, BeamanKD. Expression of regeneration and tolerance factor on B cell chroniclymphocytic leukemias: a possible mechanism for escaping immunesurveillance. Am J Hematol 1999;61:46–52.
9. Naganuma H, Kono K, Mori Y, Takayoshi S, Stern PL, Tasaka K,Matsumoto Y. Oncofetal antigen 5T4 expression as a prognostic factor inpatients with gastric cancer. Anticancer Res 2002;22:1033–8.
10. Kopreski MS, Benko FA, Gocke CD. Circulating RNA as a tumormarker: detection of 5T4 mRNA in breast and lung cancer patientserum. Ann N Y Acad Sci 2001;945:172–78.
11. Hsu LC, White RL. BRCA1 is associated with the centrosome duringmitosis. Proc Natl Acad Sci U S A 1998;95:12983–8.
12. Revillieon F, Pawlowski V, Hornez L, Peyrat JP. Glecerladehyde-3-phosphate dehydrogenase gene expression in human breast cancer.Eur J Can 2000;36:1038–42.
13. Maniatis T, Tasic B. Alternative pre-mRNA splicing and proteomeexpansion in metazoans. Nature 2002;418:236–43.
14. Stamm S. Signals and their transduction pathways regulating alterna-tive splicing: a new dimension of the human genome. Hum Mol Genet2002;11:2409–16.
15. Caballero OL, de Souza SJ, Brentani RR, Simpson AJ. Alternativespliced transcripts as cancer markers. Dis Markers 2001;17:67–75.
16. Wang L, Lin SH, Wu WG, Kemp BL, Walsh GL, Hong WK, Mao L.C-CAM1, a candidate tumor suppressor gene, is abnormally ex-pressed in primary lung cancers. Clin Cancer Res 2000;6:2988–3.
17. Line A, Slucka Z, Stengrevics A, Li G, Rees RC. Altered splicing
pattern of TACC1 mRNA in gastric cancer. Cancer Genet Cytogenet2002;139:78–83.
18. Orban TI, Olah E. Emerging roles of BRCA1 alternative splicing.J Clin Pathol Mol Pathol 2003;56:191–7.
19. Polgar P, Miko E, Monstad I, Vona I, Baliko Z, Schmelzer M, NagyE, Szekeres-Berto J. PIBF (Progesterone Induced Blocking Factor) isa tumor marker and has immunomodulatory activity relevant to cancertherapy. Submitted.
21. Fisk HA, Mattison CP, Winey M. Centrosomes and tumour suppres-sors. Curr Opin Cell Biol 2002;14:700–5.
22. Xu X, Weaver Z, Linke SP, Li C, Gotay J, Wang XW, Harris CC,Ried T, Deng CX. Centrosome amplification and a defective G2-Mcell cycle checkpoint induce genetic instability in BRCA1 exon 11isoform-deficient cells. Mol Cell 1999;3:389–95.
23. Deng CX. Roles of BRCA1 in centrosome duplication. Oncogene2002;21:6222–7.
24. Tutt A, Gabriel A, Bertwistle D, Connor F, Paterson H, Peacock J,Ross G, Ashworth A. Absence of Brca2 causes genome instability bychromosome breakage and loss associated with centrosome amplifi-cation. Current Biol 1999;9:1107–10.
25. Fukasawa K, Choi T, Kuriyama R, Rulong S, Vande Woude GF.Abnormal centrosome amplification in the absence of p53. Science1996;271:1744–7.
26. Dutertre S, Descamps S, Prigent C. On the role of aurora-A incentrosome function. Oncogene 2002;21:6175–83.
27. Andrews PD, Knatko E, Moore WJ, Swedlow JR. Mitotic mechanics:the auroras come into view. Curr Opin Cell Biol 2003;15:672–83.
28. Dai W, Wang Q, Traganos F. Polo-like kinases and centrosomeregulation. Oncogene 2002;21:6195–200.
29. Ohkura H. Phosphorylation: polo kinase joins an elite club. Curr Biol2003;13:R912–4.
30. Gritsko TM, Coppola D, Paciga JE, Yang L, Sun M, Shelley SA,Fiorica JV, Nicosia SV, 30. Cheng JQ. Activation and overexpressionof centrosome kinase BTAK/Aurora-A in human ovarian cancer. ClinCancer Res 2003;9:1420–6.
31. Takahashi T, Sano B, Nagata T, Kato H, Sugiyama Y, Kunieda K,Kimura M, Okano Y, Saji S. Polo-like kinase 1 (PLK1) is overex-pressed in primary colorectal cancers. Cancer Sci 2003;94:148–52.
32. Gergely F, Karlsson C, Still I, Cowell J, Kilmaritn J, Raff JW. The TACCdomain identifies a family of centrosomal proteins that can interact withmicrotubules. Proc Natl Acad Sci U S A 2000;19:14352–7.
33. Gergely F. Centrosomal TACCtics. Bioessays 2002;24:915–25.34. Thakur S, Zhang HB, Peng Y, Le H, Carroll B, Ward T, Yao J, Farid
LM, Couch FJ, Wilson RB, Weber BL. Localization of BRCA1 anda splice variant identifies the nuclear localization signal. Mol Cell Biol1997;17:444–52.
35. Bachelier R, Dalla Venezia N, Mazoyer S, Frappart L, Lenoir GM,Vincent A. Differential expression and subcellular localization ofmurine BRCA1 and BRCA1-delta 11 isoforms in murine and humancell lines. Int J Cancer 2000;88:519–24.
