deregulation of cofactor of brca1 expression in breast cancer cells

10
Journal of Cellular Biochemistry 103:1798–1807 (2008) Deregulation of Cofactor of BRCA1 Expression in Breast Cancer Cells Jianlong Sun, 1 Gareth Watkins, 2 Ashley L. Blair, 3 Christopher Moskaluk, 4 Sagar Ghosh, 1 Wen G. Jiang, 2 and Rong Li 1 * 1 Department of Molecular Medicine, Institute of Biotechnology, 15355 Lambda Drive, University of Texas Health Science Center at San Antonio, San Antonio, Texas 78245 2 Metastasis and Angiogenesis Research Group, University Department of Surgery, School of Medicine Cardiff University, Cardiff, UK 3 Department of Biochemistry and Molecular Genetics, School of Medicine, P.O. Box 800733, University of Virginia, Charlottesville, Virginia 22908-0733 4 Department of Pathology, School of Medicine, P.O. Box 800733, University of Virginia, Charlottesville, Virginia 22908-0733 Abstract Cofactor of BRCA1 (COBRA1) is an integral component of the human negative elongation factor (NELF), a four-subunit protein complex that inhibits transcription elongation. Previous in vivo work indicates that COBRA1 and the rest of the NELF complex repress estrogen-dependent transcription and the growth of breast cancer cells. In light of the COBRA1 function in breast cancer-related gene expression, we sought to examine regulation of COBRA1 expression in both established breast cancer cell lines and breast carcinoma tissues. We found that COBRA1 expression was inversely correlated with breast cancer progression, as tumor samples of patients who had distant metastasis and local recurrence expressed very low levels of COBRA1 mRNA when compared to those who were disease free for over 10 years (P ¼ 0.0065 and 0.0081, respectively). Using both breast and prostate cancer cell lines, we also explored the possible mechanisms by which COBRA1 expression is regulated. Our results indicate that the protein abundance of COBRA1 and the other NELF subunits are mutually influenced in a tightly coordinated fashion. Small interfering RNA (siRNA) that targeted at one NELF subunit dampened the protein levels of all four subunits. Conversely, ectopic expression of COBRA1 in the knockdown cells partially rescues the co-depletion of the NELF subunits. In addition, our study suggests that a post-transcriptional, proteasome-independent mechanism is involved in the interdependent regulation of the NELF abundance. Furthermore, a lack of COBRA1 expression in breast carcinoma may serve as a useful indicator for poor prognosis. J. Cell. Biochem. 103: 1798–1807, 2008. ß 2007 Wiley-Liss, Inc. Key words: COBRA1; NELF; protein stability; transcriptional repression; metastatic breast cancer Cofactor of BRCA1 (COBRA1) was first identified based on its ability to interact with the breast/ovarian cancer susceptibility gene product BRCA1 [Ye et al., 2001]. The physical interaction between the two proteins prompted the initial characterization of COBRA1 in breast epithelial/carcinoma cells. Immunohistochem- istry (IHC) using normal human mammary ducts indicates that COBRA1 is preferentially expressed in the nuclei of luminal breast epithelial cells rather than myoepithelial cells, which coincides with the expression pattern of estrogen receptor a (ERa) in mammary gland [Aiyar et al., 2004]. Functional studies demon- strate that COBRA1 physically binds to the ligand-binding domain (LBD) of ERa and inhibits ligand-dependent transcription activa- tion by ERa at a subset of ERa-responsive genes in breast carcinoma cells [Aiyar et al., 2004]. Consistent with its role as a transcriptional ß 2007 Wiley-Liss, Inc. This article contains supplementary material, which may be viewed at the Journal of Cellular Biochemistry website at http://www.interscience.wiley.com/jpages/0730-2312/ suppmat/index.html. Grant sponsor: National Institutes of Health; Grant number: DK064604. *Correspondence to: Rong Li, Department of Molecular Medicine, Institute of Biotechnology, 15355 Lambda Drive, University of Texas Health Science Center at San Antonio, San Antonio, TX 78245. E-mail: [email protected] Received 24 May 2007; Accepted 14 August 2007 DOI 10.1002/jcb.21568

Upload: jianlong-sun

Post on 11-Jun-2016

220 views

Category:

Documents


7 download

TRANSCRIPT

Page 1: Deregulation of cofactor of BRCA1 expression in breast cancer cells

Journal of Cellular Biochemistry 103:1798–1807 (2008)

Deregulation of Cofactor of BRCA1 Expression inBreast Cancer Cells

Jianlong Sun,1 Gareth Watkins,2 Ashley L. Blair,3 Christopher Moskaluk,4 Sagar Ghosh,1

Wen G. Jiang,2 and Rong Li1*1Department of Molecular Medicine, Institute of Biotechnology, 15355 Lambda Drive,University of Texas Health Science Center at San Antonio, San Antonio, Texas 782452Metastasis and Angiogenesis Research Group, University Department of Surgery,School of Medicine Cardiff University, Cardiff, UK3Department of Biochemistry and Molecular Genetics, School of Medicine, P.O. Box 800733,University of Virginia, Charlottesville, Virginia 22908-07334Department of Pathology, School of Medicine, P.O. Box 800733, University of Virginia,Charlottesville, Virginia 22908-0733

Abstract Cofactor of BRCA1 (COBRA1) is an integral component of the human negative elongation factor (NELF), afour-subunit protein complex that inhibits transcription elongation. Previous in vivo work indicates that COBRA1 and therest of the NELF complex repress estrogen-dependent transcription and the growth of breast cancer cells. In light of theCOBRA1 function in breast cancer-related gene expression, we sought to examine regulation of COBRA1 expression inboth established breast cancer cell lines and breast carcinoma tissues. We found that COBRA1 expression was inverselycorrelated with breast cancer progression, as tumor samples of patients who had distant metastasis and local recurrenceexpressed very low levels of COBRA1 mRNA when compared to those who were disease free for over 10 years (P¼ 0.0065and 0.0081, respectively). Using both breast and prostate cancer cell lines, we also explored the possible mechanisms bywhich COBRA1 expression is regulated. Our results indicate that the protein abundance of COBRA1 and the other NELFsubunits are mutually influenced in a tightly coordinated fashion. Small interfering RNA (siRNA) that targeted at one NELFsubunit dampened the protein levels of all four subunits. Conversely, ectopic expression of COBRA1 in the knockdowncells partially rescues the co-depletion of the NELF subunits. In addition, our study suggests that a post-transcriptional,proteasome-independent mechanism is involved in the interdependent regulation of the NELF abundance. Furthermore, alack of COBRA1 expression in breast carcinoma may serve as a useful indicator for poor prognosis. J. Cell. Biochem. 103:1798–1807, 2008. � 2007 Wiley-Liss, Inc.

Key words: COBRA1; NELF; protein stability; transcriptional repression; metastatic breast cancer

Cofactor of BRCA1 (COBRA1) was firstidentified based on its ability to interact withthe breast/ovarian cancer susceptibility gene

product BRCA1 [Ye et al., 2001]. The physicalinteraction between the two proteins promptedthe initial characterization of COBRA1 in breastepithelial/carcinoma cells. Immunohistochem-istry (IHC) using normal human mammaryducts indicates that COBRA1 is preferentiallyexpressed in the nuclei of luminal breastepithelial cells rather than myoepithelial cells,which coincides with the expression pattern ofestrogen receptor a (ERa) in mammary gland[Aiyar et al., 2004]. Functional studies demon-strate that COBRA1 physically binds to theligand-binding domain (LBD) of ERa andinhibits ligand-dependent transcription activa-tion by ERa at a subset of ERa-responsive genesin breast carcinoma cells [Aiyar et al., 2004].Consistent with its role as a transcriptional

� 2007 Wiley-Liss, Inc.

This article contains supplementary material, which maybe viewed at the Journal of Cellular Biochemistry websiteat http://www.interscience.wiley.com/jpages/0730-2312/suppmat/index.html.

Grant sponsor: National Institutes of Health; Grantnumber: DK064604.

*Correspondence to: Rong Li, Department of MolecularMedicine, Institute of Biotechnology, 15355 Lambda Drive,University of Texas Health Science Center at San Antonio,San Antonio, TX 78245. E-mail: [email protected]

Received 24 May 2007; Accepted 14 August 2007

DOI 10.1002/jcb.21568

Page 2: Deregulation of cofactor of BRCA1 expression in breast cancer cells

corepressor in estrogen-mediated gene expres-sion, ectopic expression of COBRA1 inhibits,whereas its partial depletion accelerates, theproliferation of an ERa-positive breast cancercell line [Aiyar et al., 2004].

COBRA1 has also been identified through anindependent study as an integral subunit of thehuman negative elongation factor (NELF) com-plex, which functions to repress transcriptionelongation by stalling RNA Polymerase II(RNAPII) [Yamaguchi et al., 1999; Wada et al.,2000]. The NELF complex is composed of fourfunctionally important subunits (NELF-A, -B,-C, and -E), with NELF-B identical to COBRA1[Yamaguchi et al., 1999, 2002; Narita et al.,2003]. In support of the in vitro biochemicalfindings, COBRA1 and the other NELF sub-units are simultaneously recruited to multipleERa-regulated genes in an estrogen-dependentmanner [Aiyar et al., 2004]. Depletion of eitherCOBRA1 or NELF-E gives rise to a similareffect on estrogen-dependent transcription andcell growth [Aiyar et al., 2004], suggesting anintimate functional interdependence of theNELF subunits.

In light of the implication of COBRA1 in thecontrol of estrogen-dependent transcription andbreast cancer cell growth, we examined regu-lation of COBRA1 expression in breast cancercells. Our results reveal an interesting corre-lation between reduced COBRA1 expressionand breast cancer progression. In addition,our study also shows that the protein abun-dance of COBRA1 and the other NELF subunitsare tightly coordinated in an interdependentmanner.

MATERIALS AND METHODS

Cell Lines and Culture

T47D, HeLa, ES2, MCF10A, SKBR3, H118,SKOV3, LNCaP, and HEK293T cells lineswere obtained from the American Type CultureCollection (Manassas, VA). The KGN cell linewas a generous gift from Dr. Hajime Nawata.The growth conditions for various cell lines areas follows: T47D, HeLa, MCF7, and HEK293Tin DMEM supplemented with 10% fetal bovineserum (FBS); ES2 cells in McCoy’s 5A supple-mented with 1 mM sodium pyruvate and 10%FBS; SKBR3 and SKOV3 cells in McCoy’s 5Asupplemented with 10% FBS; KGN cells inDMEM-F12 supplemented with 10% FBS;MCF10A cells in DMEM-F12 supplemented

with 5% horse serum, 20 ng/ml EGF, 500 ng/mlhydrocortisone, 100 ng/ml cholera toxin, and10 mg/ml insulin; H118 cells in RPMI 1640supplemented with 0.5 mM sodium carbonate,1 mM sodium pyruvate, and 10% FBS. LNCaPwas maintained in phenol red-free RPMI1640 supplemented with 5% FBS. Penicillin(100 units/ml) and streptomycin (100 mg/mL)were supplemented in all above-mentionedculture media.

Antibodies and Plasmids

The generation of anti-COBRA1 and anti-NELF-E antibodies was previously described[Aiyar et al., 2004]. To generate anti-NELF-Aand -C polyclonal antibodies, His-tagged full-length NELF-A and NELF-C proteins wereexpressed and purified from Escherichia coliand used for immunization in rabbits. Thefollowing antibodies were commercially avail-able: anti-a-tubulin antibody (EMD Chemicals,San Diego, CA), anti-c-Jun (Santa Cruz Bio-technology, Santa Cruz, CA), and anti-Ranantibody (BD Biosciences, San Jose, CA). Therecombinant adenovirus expression vector forCOBRA1 was cloned by inserting the COBRA1cDNA into an adenoviral expression vectorpAdTrack-CMV (a gift from Dr. B. Vogelstein),and the resulting construct was verified bysequencing.

Immunoblotting

Cells were lysed in a modified protein samplebuffer (50mM Tris pH6.8, 2% SDS, and 10%glycerol), supplemented with a cocktail ofprotease inhibitors (Sigma–Aldrich, St. Louis,MO). Protein concentration was determined bythe BCA assay (Pierce Biotechnology, Rockford,IL) and an equal amount of whole cell lysate wasresolved by 10% SDS–polyacrylamide gel elec-trophoresis (SDS–PAGE). Western blot wasperformed according to standard protocolsand the proteins of interest were visualizedwith SuperSignal West Pico kit (Pierce Bio-technology).

RT-PCR

Total RNA was isolated from cultured cellswith Trizol reagent (Invitrogen, Carlsbad, CA)according to the manufacture’s instruction. Onemicrogram of RNA was used to synthesizecDNA with ImPromp II kit (Promega, Madison,WI). The steady state mRNA level of NELFsubunits was determined by real-time PCR

Stability of NELF Subunits 1799

Page 3: Deregulation of cofactor of BRCA1 expression in breast cancer cells

analysis using the 7900HT Real-Time PCRSystem (Applied Biosystems, Foster City, CA).The level of GAPDH mRNA was measured asthe internal control. The real-time PCR primersfor the NELF subunits and GAPDH will beprovided upon request.

The real-time PCR condition used for analyz-ing COBRA1 mRNA levels in primary tumortissues was the same as previously described[Cai et al., 2006]. The sequences for the real-time PCR primers are: COBRA1F2: CTGGAC-GTGGGTGAAATC, COBRA2ZR2: ACTGAAC-CTGACCGTACACTCTTGCTGTCCACGAAC.A Taqman detection kit for b-actin was pur-chased from Perkin-Elmer (Surrey, UK). Thecellularity of the tissues was assessed usingCK19. The CK19 forward and reverse primerswere 50-CAGGTCCGAGGTTACTGAC-30 and50-ACTGAACCTGACCGTACACACTTTCTGC-CAGTGTGTCTTC-30, respectively.

siRNA Transfection

siRNAs for NELF-A (M-012156-00), COBRA1(M-015839-00), NELF-C (M-020811-00), NELF-E (M-011761-00), and the negative controlluciferase GL3 duplex (D-001400-01-20) werepurchased from Dharmacon (Chicago, IL) andresuspended according to the manufacture’sinstruction. Transfection of the siRNA duplexin T47D or LNCaP cells was carried out byusing the transfection reagent LipofectamineRNAiMAX (Invitrogen). Briefly, approximately2.5� 105 cells were transfected with 50 nM ofsiRNA using 5 ml of RNAiMAX reagent. Cellswere subsequently cultured for 72 h and thenharvested for protein or RNA analysis.

Adenovirus Infection

COBRA1-expressing adenovirus was pro-duced and purified according to manufacture’sinstruction (Q-Biogene, BD Biosciences). T47Dcells were first transiently transfected withsiRNA and cultured for 48 h. The transfectedcells were then infected with 50 MOI of LacZ(control) or COBRA1 expression recombinantadenovirus for 2 h. Fresh growth medium wassupplemented and the culture was incubatedfor an additional 48 h before harvesting.

Human Specimens

A total of 87 ductal carcinoma tissue sampleswere collected immediately after mastectomy

and stored in liquid nitrogen until use. Theprocedure for tissue procurement was approvedby the local ethical committee at Cardiff Uni-versity, Cardiff, UK. Patients were followedclinically for an average of 120 months aftersurgery (June 2004). Detailed clinical informa-tion on patients in this study is given inSupplemental Table I. The normal humantissue array was obtained from NCI Coopera-tive Human Tissue Network (CHTN).

Immunohistochemistry (IHC)

IHC for COBRA1 was performed as previ-ously described [Aiyar et al., 2004]. Frozensections of breast tumor and background tissuewere cut at a thickness of 6 mm using a cryostat.The sections were mounted on super frost plusmicroscope slides, air-dried, and then fixed in amixture of 50% acetone and 50% methanol.The sections were then placed in ‘‘Optimax’’wash buffer for 5–10 min to rehydrate. Sectionswere incubated for 20 min in a 0.6% BSAblocking solution and probed with the polyclo-nal COBRA1 antibody, and without primaryantibody as a negative control. Followingextensive washings, sections were incubatedfor 30 min in the secondary biotinylated anti-body (Multilink Swine anti-goat/mouse/rabbitimmunoglobulin, Dako, Inc.). Following wash-ings, the Avidin Biotin Complex (Vector Labo-ratories, Peterborough) was then applied tothe sections, followed by extensive washingsteps. Diamino benzidine chromogen (VectorLabs) was then added to the sections, andincubated in the dark for 5 min. Sections werethen counter stained in Gill’s Haematoxylin anddehydrated in ascending grades of methanolbefore clearing in xylene and mounting under acover slip. Immunohistochemical analysis ofCOBRA was conducted on a portion of the breasttumors (n¼ 33) and background normal tissues(n¼ 32), due to the availability of tissuesections.

Statistical Analysis

The results of the COBRA1 mRNA analysisfrom the clinical samples were analyzed usingnon-paired Student’s t-test. COBRA1 transcriptvalues generated in the study are shown asmean copy number þSD. A P-value less than0.05 was defined as statistically significant.Survival analysis (overall and disease free) wasconducted using SPSS program (version 12.2).

1800 Sun et al.

Page 4: Deregulation of cofactor of BRCA1 expression in breast cancer cells

RESULTS

Expression of COBRA1 in Established CancerCell Lines and Clinical Samples

Given the well-documented functional andphysical association among the NELF subunits,we sought to simultaneously examine by immu-noblotting the expression pattern of theseproteins in multiple established human celllines representing different tissue origins. Poly-clonal antibodies that were raised againstrecombinant NELF subunits recognized thecorresponding proteins ectopically expressedin HEK293T cells (lane 10; Fig. 1A), as well asthe endogenous proteins with the expectedmolecular weights in most of the cell linesexamined (lanes 1–9; Fig. 1A). Intriguingly,both NELF-A and -B (COBRA1) were expressedat low levels in SKBR3 (lane 6), an ERa-negative breast cancer cell line with aggressivegrowth and metastatic phenotypes. This is incontrast to the expression of the NELF subunits

in a non-cancerous breast epithelial cell lineMCF10A (lane 4) and a less invasive, ERa-positive breast cancer cell line T47D (lane 5).

Next, we used clinical samples to examineNELF expression in normal and tumor tissues.Of all available NELF-specific antibodies raisedin our lab, the polyclonal antibody againstCOBRA1 is most suitable for IHC. We firstperformed IHC on a normal tissue array thatincluded most tissue and cell types present inthe human body. Consistent with the cell-linestudy (Fig. 1A), COBRA1 protein was detectedin most primary tissues (Fig. 1B, and Supple-mental Table II). Stronger expression ofCOBRA1 is more evident in the epithelium thanstroma of most organs. The intensity andprevalence of COBRA1 expression vary amongdifferent organs. For example, several organs ofthe gastrointestinal tract, such as stomach,small intestine, and colon, express COBRA1 invirtually 100% of their epithelial cells, whereasonly a portion of these cells in prostate and

Fig. 1. NELF expression in established cell lines and clinicaltissues. A: Western blot analysis of NELF expression in multiplecell lines. The blots were probed with anti-NELF-A, anti-COBRA1 (NELF-B), anti-NELF-C, and anti-NELF-E antibodies,respectively. Ran was used as a loading control. B: Normal tissuearray was immunostained with a COBRA1-specific polyclonalantibody. Several representative images of selective tissues areshown on the right. The images were taken at �100 magnifica-tion. Abbreviations are as following: GFEN, endocervix; GFES,endometrium, secretory; GIGA, gastric mucosa, antral; GISI,

small intestine, mucosa; GFFT, fallopian tube; GIC, colon,mucosa; BE, breast epithelium; and HPG, gallbladder.C: Representative staining pattern of COBRA in normalmammary (panels A,B) and in tumor tissue (panels C,D).Mammary epithelial cells showed a strong nuclear pattern ofstaining of COBRA, whereas breast cancer cells showed amarkedly reduced staining compared with normal epithelialcells. The panels are shown with �100 magnification, and theinsets with �400 magnification. The arrows indicate the areaswhere the magnified insets come from.

Stability of NELF Subunits 1801

Page 5: Deregulation of cofactor of BRCA1 expression in breast cancer cells

seminal vesicle were positive for COBRA1staining. In the COBRA1-expressing cells, theimmunostaining signal was predominantlyfound in the nucleus, which was in line withits known function in gene regulation.Figure 1C shows the representative stainingpattern of COBRA in normal mammary (A,B)and in tumor tissue (C,D). Mammary epithelialcells showed a strong nuclear pattern of stain-ing of COBRA. In contrast, breast cancer cellsshowed a markedly reduced staining comparedwith normal epithelial cells.

Low Level of COBRA1mRNA is AssociatedWith Metastatic Breast Cancer

Prompted by the finding of the relatively lowlevels of COBRA1 protein in the aggressivebreast cancer cell line and breast tumor tissues(Fig. 1), we examined COBRA1 expression ina cohort of ductal carcinoma samples (n¼ 87)from patients with known clinical outcomes. Ina 120-month follow-up study, the patients fromwhom the tumor samples were collected weredivided into the following four groups: thosewho remained disease-free, had metastasis orlocal recurrence, and those who died of breastcancer (those who died of conditions unrelatedto breast cancer were excluded in the analysis).As shown in Figure 2A, patients with meta-stasis or local recurrence had significantly lowerlevels of COBRA1 mRNA expression than thedisease-free group (P¼ 0.0065 and 0.0081).

Those who had poor prognosis also displayedlower levels of COBRA1 than the disease-freegroup (Fig. 2B). Kaplan–Meier survival analy-sis revealed an interesting trend that low levelsof COBRA1 transcript were associated with ashorter survival (116 (96–137, 95%CI) months)compared with high levels of COBRA1 (128(115–142, 95%CI) months)(Fig. 2C). This differ-ence nonetheless was not statistically signifi-cant (P¼ 0.101), perhaps due to the relativelysmall size of the study cohort.

Interdependent Protein Stabilityof the NELF Subunits

During the analyses of COBRA1 expressionin cell lines and clinical samples, we noticed thatthe fluctuation in COBRA1 protein level wasnot always correlated with the changes at themRNA level, For example, the SKBR3 cellsexpressed extremely low levels of the COBRA1and NELF-A proteins (Fig. 1A), yet the corre-sponding mRNA levels were similar to those inother breast cancer cell lines (data not shown).This suggests that the deregulation of COBRA1expression can occur at a post-transcriptionalstep. In addition, the concurrent reduction ofNELF-A and COBRA1 protein in SKBR3 cellsalso raised the possibility of mutual influence ofthe protein abundance by the NELF subunits.To test this possibility, we individually knockeddown each NELF subunit by siRNA andexamined the steady-state levels of the NELF

Fig. 2. Quantitative RT-PCR analysis of COBRA1 mRNA expression in human ductal carcinoma tissues.A: Relationship between COBRA1 expression and clinical outcome over a 10-year follow-up period. There isa strong correlation between low COBRA1 levels and patients with metastasis or local recurrence(*P¼0.0065 and 0.0081, respectively). B: Comparison between those patients who remained disease freeand those who developed further disease progression (metastasis, recurrence, and motality combined).C: Kaplan–Meier survival curve for a disease free survival. [Color figure can be viewed in the online issue,which is available at www.interscience.wiley.com.]

1802 Sun et al.

Page 6: Deregulation of cofactor of BRCA1 expression in breast cancer cells

proteins. As shown in the second panel inFigure 3A, siRNA oligos that targeted NELF-A, -C, or -E all reduced the levels of COBRA1(NELF-B) as did the COBRA1-specific oligos,albeit to different degrees (compare lanes 1 and2 with 3–6). Interestingly, this co-depletionphenomenon was not restricted to the COBRA1protein. In fact, reduction of any of the fourNELF subunits led to reduction of expressionfor all four proteins (lanes 3–6). A similar co-depletion of the NELF subunits by siRNA wasalso observed in a prostate cancer cell lineLNCaP (Fig. 3B) and breast cancer cell lineMCF7 (data not shown), suggesting the exis-tence of a cell type-independent mechanism

that governs the interdependency of the NELFsubunits.

To obtain more insight into the mechanismunderlying the co-depletion, we performed atime course experiment to examine the kineticsof reduction of the individual NELF subunitsfollowing the COBRA1-specific siRNA trans-fection. As shown in Figure 4A, the proteinlevel of COBRA1 began to decrease 24 h aftersiRNA transfection and its reduction continuedthroughout the entire 4-day time course. Intri-guingly, the other three NELF subunits weredepleted at similar rates as that for COBRA1,suggesting a rapid response to any compro-mised integrity of the NELF complex.

Fig. 3. Protein levels of NELF subunits in control and NELF knockdown cells. T47D (A) and LNCaP (B) cellswere transfected with siRNA oligos for the control or the individual NELF subunits. Whole cell extracts wereprepared 72 h after transfection and an equal amount of the extract was resolved by 10% SDS–PAGE. Proteinlevels of each NELF subunits were determined by Western blot analysis with the corresponding antibodies.a-tubulin was used as the loading control.

Fig. 4. Further characterization of the interdependency of theNELF subunits. A: Time course study of the NELF subunitsfollowing COBRA1 siRNA knockdown. T47D cells were trans-fected with the control or COBRA1 siRNA duplex and harvestedfor whole cell extract preparation at 24, 48, 72, and 96 h aftertransfection. Protein levels of the individual NELF subunits were

determined as described in Figure 3. B: Partial rescue of the NELFco-depletion with ectopic expression of COBRA1. T47D cellswere first transfected with siRNA duplex for 48 h. Transfectedcells were then infected with the recombinant adenovirus forthe control protein LacZ or COBRA1. Whole cell extract wasanalyzed for the expression of the NELF subunits.

Stability of NELF Subunits 1803

Page 7: Deregulation of cofactor of BRCA1 expression in breast cancer cells

To examine whether the reduction of NELFsubunits was reversible, T47D cells were firsttransfected with the COBRA1-specific siRNA.48 h after the siRNA transfection, cells wereinfected with either control (Ade-LacZ) orCOBRA1-expressing recombinant adenovirus(Ade-COBRA1). When compared with the mockor Ad-LacZ-infected COBRA1-knockdown cells(lanes 4 and 5; Fig. 4B), infection with theCOBRA1-expressing adenovirus resulted inpartial rescue of the expression of COBRA1 aswell as the other three subunits (lane 6). Thislends further support to the notion that theabundance of the four NELF subunits is highlyinterdependent.

In order to determine whether the coordi-nated expression of the individual NELF sub-units was due to changes in the levels of thecorresponding mRNA, we examined the mRNAlevels of the four NELF subunits in each siRNA-mediated knockdown set by quantitative RT-PCR analysis. As shown in Figure 5, when cellswere transfected with siRNA that targeted oneNELF subunit, only the mRNA of the corre-sponding subunit was substantially reduced.The mRNA levels of the other three subunitsremained largely unaffected. This result indi-cates that post-transcriptional mechanismsplay an important role in the coordinatedexpression of the NELF subunits.

To test whether protein degradation wasresponsible for this co-depletion phenomenon,we first used the cycloheximide treatment tocompare the half-lives of NELF-A, -C, and -Ein cells transfected with either control orCOBRA1-specific siRNA. Intriguingly, thehalf-lives of the NELF-A and -E subunits inCOBRA1 knockdown cells were comparable tothose in the control cells, whereas the proteinstability of NELF-C in the absence of COBRA1was significantly decreased (Fig. 6A). Thissuggests that the interdependent coordinationof different NELF subunits may be regulated bydistinct post-transcriptional mechanisms. Toexplore the possible protein degradation path-ways responsible for the NELF-C degradation,we treated the control and COBRA1-depletedcells with inhibitors for the proteasome-(MG132) and lysosome-mediated (ALLN) pro-tein degradation machineries. As expected,treatment with neither drug in COBRA1-depleted cells could significantly stabilizeNELF-A or -E subunits (compare lane 1 withlanes 4–6; Fig. 6B). However, it is somewhat

surprising that the drug treatment did not fullyrestore the protein levels of NELF-C to those inthe control siRNA-transfected cells either. As apositive control, c-Jun protein level was drasti-cally elevated upon the treatment of eitherdrug. Thus, this result suggests that neitherproteasome- nor lysosome-mediated proteindegradation pathway is likely to serve as themajor mechanism for the regulation of NELFprotein stability.

DISCUSSION

COBRA1, first identified as a BRCA1-inter-acting protein [Ye et al., 2001], has been shownin subsequent tissue culture studies to be

Fig. 5. Examination of the mRNA levels of the NELF subunits incontrol and NELF knockdown cells. T47D cells were transfectedwith control or siRNA specific for the individual NELF subunits.Total RNA was extracted and analyzed for mRNA expression ofeach subunit by quantitative RT-PCR. Relative expression levelwas determined by normalization against the levels of GAPDHmRNA.

1804 Sun et al.

Page 8: Deregulation of cofactor of BRCA1 expression in breast cancer cells

involved in estrogen-dependent and independenttranscription in breast cancer cells [Aiyaret al., 2004, 2007]. Furthermore, COBRA1inhibits the growth of ER-positive breast cancercells in vitro [Aiyar et al., 2004], which isconsistent with the finding that COBRA1expression is reduced in several establishedbreast cancer cell lines [Zhu et al., 2004]. Inthe current study, we extended the previousstudies by establishing a link between reducedCOBRA1 expression and breast cancer progres-sion. In particular, our data suggest that lowCOBRA1 expression is strongly associated withmetastatic breast cancer. Taken together, thesedata are in line with the notion that COBRA1may play a role as a tumor/metastasis suppres-sor in breast cancer development. Aberrantexpression of other steroid receptor coregula-tors has been linked with breast cancer develop-ment as well. For example, overexpression ofER coactivators such as SRC-3/AIB1 in breastcancer has been documented [Anzick et al.,

1997]. On the other hand, low expression of thecorepressor N-CoR has been shown to bevaluable in predicting resistance to endocrinetherapies [Lavinsky et al., 1998; Girault et al.,2003]. It will be of interest, therefore, toevaluate in future the COBRA1 expression inpatients with various degrees of resistance toendocrine therapies.

Our tissue array data indicate that COBRA1is predominantly expressed in epithelia ofmultiple tissues. Therefore, its involvement intumorigenesis may not be restricted to breasttissue. In fact, recent studies indicate thathigher expression of COBRA1 and NELF-E isassociated with tumorigenesis in upper gastro-intestinal tract [Midorikawa et al., 2002;McChesney et al., 2006], which contrasts withits implicated role as a potential tumor sup-pressor in breast cancer. The molecular basisfor this potential tissue-specific function ofCOBRA1 remains to be elucidated. However,it is possible that, through its interactions

Fig. 6. Mechanistic study of the interdependency of the NELFsubunits. A: Evaluation of the protein half-lives of NELF-A, -C,and -E in the control and COBRA1 knockdown T47D cells. T47Dcells were first transfected with control or COBRA1 siRNA.Seventy-two hours post-transfection, cells were treated withcycloheximide (100 mg/ml) for 0, 2, 4, 6, 8, and 12 h. Whole cellextract was prepared and analyzed for the levels of the individualNELF subunits by Western blot analysis. Protein bands werequantified with ImageQuant V5.2 and relative intensity was used

for plotting the graphs. B: Effect of proteasome and lysosomeinhibitors on the stability of the NELF subunits. T47D cells werefirst transfected with control or COBRA1 siRNA, and then treatedwith proteasome inhibitor MG132 (5 mM) or lysosome inhibitorALLN (200 mM) for 8 h. Whole cell extract was prepared andprotein level of each NELF subunits was determined by Westernblot analysis. c-Jun was used as a positive control for the drugeffects. a-tubulin was used as a loading control.

Stability of NELF Subunits 1805

Page 9: Deregulation of cofactor of BRCA1 expression in breast cancer cells

with different site-specific transcription factors[Zhong et al., 2004; McChesney et al., 2006],COBRA1 may modulate distinct transcriptionprograms in different cell types. Alternatively,the same COBRA1-regulated genes may playdifferent physiological roles under different celland tissue contexts.

The long-term follow-up study has demon-strated that COBRA1 is correlated with meta-stasis and local recurrence, as patients withboth conditions had a markedly reduced levelof COBRA1 transcripts. In this regard, it isimportant to point out that many of theCOBRA1-regulated genes identified in ourprevious gene expression profiling study areassociated with advanced breast cancer [Aiyaret al., 2004, 2007]. For example, as a tran-scription corepressor, COBRA1 is physicallyassociated with, and regulates the activity of,the promoters of the trefoil factor (TFF) genefamily (TFF1–3) [Aiyar et al., 2007]. Notably,both TFF1 and TFF3 are associated with breastcancer metastatic to bone [Smid et al., 2006] andadvanced prostate cancer [Vestergaard et al.,2006]. Thus, reduction of COBRA1 expressionmay lead to over-expression of its cancer-associated target genes, which may in turncontribute to the metastasis and/or local recur-rence of the disease. The intriguing correlationobserved in the current study merits furthervalidation with larger sample pools.

Our study clearly indicates an interdepend-ent regulation of the protein abundance of theNELF subunits. This is consistent with thecontrol of other known homomeric or hetero-meric protein complexes [Buchler et al., 2005].While the current manuscript is under prepa-ration, the same co-depletion phenomenon wasalso reported in NELF-E knockdown HeLa cells[Narita et al., 2007]. It is reasonable to postulatethat the lack of one NELF subunit hampersthe proper formation of the remaining NELFcomplex, which would subsequently exposethe other subunits for protein degradation.Although this scenario could account for thedecreased half-life of NELF-C observed inCOBRA1-knockdown cells, the minimal changeof the protein half-lives for NELF-A and -E inthe same knockdown cells argues that addi-tional mechanisms must be involved in theregulation of the latter two NELF subunits. It isconceivable, for example, protein translationmay serve as a regulatory step for the expres-sion of some NELF subunits.

While ectopic expression of COBRA1 inCOBRA1-depleted cells partially restored theprotein expression of the other subunits, thesame adenoviral infection of the parental T47Dcells significantly increased the protein level ofCOBRA1, but not the other NELF subunits(Fig. 4). This may reflect a tight control of theNELF abundance under normal cellular con-texts. It is also worth noting that the stability ofvarious subunits is affected to different degreesby the knockdown of a particular subunit. Thiscould be explained by the relative configurationof the individual subunits in the NELF complex.For example, previous biochemical study sug-gests that NELF-A in the complex directlyinteracts with NELF-C but not NELF-E [Naritaet al., 2003]. Consistent with this scenario,NELF-A knockdown displays a greater impacton the level of NELF-C than that of NELF-E(Fig. 3A). In light of the observation thatmultiple cell lines show similar interdepend-ency of the NELF subunits, it is somewhatsurprising that only the protein levels ofCOBRA1 and NELF-A, but not those of NELF-C and -E, are significantly reduced in theSKBR3 cells. Given the aggressive nature ofSKBR3 in tumorigenesis, it is possible that themechanism that dictates the co-regulation ofindividual NELF subunits may be disrupted atadvanced stages of the breast cancer. It is alsoconceivable that the remaining subunits/com-plex could act as a gain-of-function mutant topromote cancer progression.

In conclusion, the current work providescompelling evidence for a tight interdependentregulation of the abundance of the NELFcomplex, loss of which may contribute to thereduced COBRA1 expression in advancedbreast cancer. The finding of reduced expressionof COBRA1 in breast cancer patients withmetastasis and local recurrence points to thepotential of COBRA1 as a useful clinical markerin predicting clinical outcome. Together withfindings from the earlier biochemical and tissueculture work, the current study supports thenotion that COBRA1 may play an importantrole in suppression of breast tumor progression.

ACKNOWLEDGMENTS

We thank Dr. Asma Amleh for critical readingof the manuscript, and Dr. Bert Vogelstein forthe adenoviral construct. The work was sup-ported by a grant to R.L. from the National

1806 Sun et al.

Page 10: Deregulation of cofactor of BRCA1 expression in breast cancer cells

Institutes of Health (DK064604) and a pre-doctoral fellowship to J.S. from the Departmentof Defense (W81XWH-06-1-0302). W.G.J. ack-nowledges Cancer Research Wales (CRW) andCancer Research UK (CR-UK) for support.

REFERENCES

Aiyar SE, Blair AL, Hopkinson DA, Bekiranov S, Li R.2007. Regulation of clustered gene expression by cofactorof BRCA1 (COBRA1) in breast cancer cells. Oncogene26:2543–2553.

Aiyar SE, Sun J-L, Blair AL, Moskaluk CA, Lv Y, Ye Q-N,Yamaguchi Y, Mukherjee A, Ren D-M, Handa H, Li R.2004. Attenuation of estrogen receptor alpha-mediatedtranscription through estrogen-stimulated recruitmentof a negative elongation factor. Genes Dev 18:2134–2146.

Anzick SL, Kononen J, Walker RL, Azorsa DO, TannerMM, Guan XY, Sauter G, Kallioniemi OP, J.M T, MeltzerPS. 1997. AIB1, a steroid receptor coactivator amplifiedin breast and ovarian cancer. Science 277:965–968.

Buchler NE, Gerland U, Hwa T. 2005. Nonlinear proteindegradation and the function of genetic circuits. ProcNatl Acad Sci USA 102:9559–9564.

Cai J, Parr C, Jiang WG, Boulton ME. 2006. Expression ofpigment epithelial derived factor (PEDF) in humanbreast cancer and the impact on angiogenesis. ClinicalCancer Res 12:3510–3517.

Girault I, Lerebours F, Amarir S, Tozlu S, Tubiana-HulinM, Lidereau R, Bieche I. 2003. Expression analysis ofestrogen receptor alpha coregulators in breast carci-noma: Evidence that NCOR1 expression is predictive ofthe response to tamoxifen. Clin Cancer Res 9:1259–1266.

Lavinsky RM, Jepsen K, Heinzel T, Torchia J, Mullen JTM,Schiff R, Del-Rio AL, Ricote M, Ngo S, Gemsch J,Hilsenbeck SG, Osborne CK, Glass CK, Rosenfeld MG,Rose DW. 1998. Diverse signaling pathways modulatenuclear receptor recruitment of N-CoR and SMRTcomplexes. Proc Natl Acad Sci USA 95:2920–2925.

McChesney PA, Aiyar SE, Lee OJ, Zaika A, Moskaluk C, LiR, El-Rifai W. 2006. Cofactor of BRC A1: A noveltranscription factor regulator in upper gastrointestinaladenocarcinomas. Cancer Res 66:1346–1353.

Midorikawa Y, Tsutsumi S, Taniguchi H, Ishii M, KobuneY, Kodama T, Makuuchi M. 2002. Identification of genesassociated with dedifferentiation of hepatocellular carci-

noma with expression profiling analysis. Jpn J CancerRes 93:636–643.

Narita T, Yamaguchi Y, Yano K, Sugimoto S, Chanarat S,Wada T, Kim D-K, Hasegawa J, Omori M, Inukai N,Endoh M, Yamada T, Handa H. 2003. Human tran-scription elongation factor NELF: Identification of novelsubunits and reconstitution of the functionally activecomplex. Mol. Cell. Biol 23:1863–1873.

Narita T, Yung TM, Yamamoto J, Tsuboi Y, Tanabe H,Tanaka K, Yamaguchi Y, Handa H. 2007. NELFinteracts with CBC and participates in 3’ end processingof replication-dependent histone mRNAs. Mol Cell 26:349–365.

Smid M, Wang Y, Klijn JG, Sieuwerts AM, Zhang Y, AtkinsD, Martens JW, Foekens JA. 2006. Genes associated withbreast cancer metastatic to bone. J Clin Oncol 24:2261–2267.

Vestergaard EM, Borre M, Poulsen SS, Nexo E, Torring N.2006. Plasma levels of trefoil factors are increased inpatients with advanced prostate cancer. Clin Cancer Res12:807–812.

Wada T, Orphanides G, Hasegawa J, Kim D-K, Shima D,Yamaguchi Y, Fukuda A, Hisatake K, Handa H. 2000.FACT relieves DSIF/NELF-mediated inhibition oftranscriptional elongation and reveals functional differn-ces between P-TEFb and TFIIH. Mol Cell 5:1067–1072.

Yamaguchi Y, Inukai N, Narita T, Wada T, Handa H. 2002.Evidence that negative elongation factor represses tran-scription elongation through binding to a DRB sensitiv-ity-inducing factor/RNA polymerase II complex andRNA. Mol Cell Biol 22:2918–2927.

Yamaguchi Y, Takagi T, Wada T, Yano K, Furuya A,Sugimoto S, Hasegawa J, Handa H. 1999. NELF, amultisubunit complex containing RD, cooperates withDSIF to repress RNA polymerase II elongation. Cell97:41–51.

Ye Q, Hu Y-F, Zhong H, Nye AC, Belmont AS, Li R. 2001.BRCA1-induced large-scale chromatin unfolding andallele-specific effects of cancer-predisposing mutations.J Cell Biol 155:911–921.

Zhong H, Zhu J, Zhang H, Ding L, Sun Y, Huang C, Ye Q.2004. COBRA1 inhibits AP-1 transcriptional activity intransfected cells. Biochem Biophys Res Commun 325:568–573.

Zhu J, Song S, Jiang Z, Yan J, Lu Q, Huang C, Ye Q. 2004.Characterization of COBRA1 in human breast cancer celllines using a new polyclonal antibody against COBRA1.IUBMB Life 56:161–166.

Stability of NELF Subunits 1807