Aberrant Expression of X-Linked RbAp46, Rsk4, and Cldn2 in Breast Cancer

Archana Thakur,1 KM Wahidur Rahman,1 Jack Wu,1 Aliccia Bollig,1 Hector Biliran,1 Xiukun Lin,2 Hind Nassar,1 David J. Grignon,1 Fazlul H. Sarkar,1 and Joshua D. Liao1

1Department of Pathology, Wayne State University, Karmanos Cancer Institute, Detroit, Michigan and 2Institute of Oceanology, Chinese Academy of Science, Qing Dao, Shan Dong Province, P.R. China

Abstract Introduction The consequence of activation status or gain/loss of In female mammalian cells, one of the two X- an X- in terms of the expression of tumor is inactivated during a short window of embryonic life through suppressor genes or oncogenes in breast cancer heritable epigenetic modifications, thus achieving dosage has not been clearly addressed. In this study, we compensation with males who have one active X-chromosome investigated the activation status of the X-chromosomes and a sex-determining Y-chromosome (1, 2).The epigenetic in a panel of human breast cancer cell lines, human modifications render females mosaic for two cell types, either breast carcinoma, and adjacent mammary tissues and a carrying an active maternal or paternal X-chromosome.These panel of murine mammary epithelial sublines ranging cells are distributed in a 50:50 ratio, and deviation from from low to high invasive potentials. Results show that this distribution can lead to skewed X-chromosome inactiva- most human breast cancer cell lines were homozygous, tion (Xi). but both benign cell lines were heterozygous for highly Because female mammalian cells only have one active polymorphic X-loci (IDS and G6PD). On the other X-chromosome (aX), either loss of heterozygosity at the aX or hand, 60% of human breast carcinoma cases were skewed Xi may result in the loss of the function of an X-linked heterozygous for either IDS or G6PD markers. tumor suppressor in a single step and may lead to cancer Investigation of the activation status of heterozygous predisposition.Buller et al.have proposed a concept that Xi in cell lines revealed the presence of only one active females is equivalent to a functional loss of heterozygosity for X-chromosome, whereas most heterozygous human X-linked genes (3).This hypothesis is relevant to cancer breast carcinoma cases had two active X-chromosomes. formation given that X-linked tumor suppressor genes are also Furthermore, we determined whether or not an subjected to Xi, like most other X-linked genes.In this case, a additional active X-chromosome affects expression preexisting functional loss of heterozygosity only requires levels of tumor suppressor genes and oncogenes. single step for complete loss of function of a tumor suppressor Reverse transcription-PCR data show high expression of gene either by a true loss of heterozygosity or a loss-of-function putative tumor suppressor genes Rsk4 and RbAp46 in type mutation. This unique situation (i.e., physiologic loss of 47% and 79% of breast carcinoma cases, respectively, heterozygosity due to Xi) may render X-linked tumor whereas Cldn2 was down-regulated in 52% of breast suppressor genes more liable to lose their function than tumor cancer cases compared with normal adjacent tissues. suppressor genes on autosomal chromosomes (4).Loss of Consistent with mRNA expression, immunostaining function of those genes that function to suppress tumor for these also showed a similar pattern. In formation or cell proliferation, or to promote apoptosis, may conclusion, our data suggest that high expression of predispose an individual to cancer formation. RbAp46 is likely to have a role in the development or Although nonrandom losses and gains of whole or progression of human breast cancer. The activation chromosomal segments in tumors are often evidenced as a status of the X-chromosome may influence the consequence of mutational selection for abnormal cell growth expression levels of X-linked oncogenes or tumor (5, 6), the consequence of gain of an additional aX or loss of an suppressor genes. (Mol Cancer Res 2007;5(2):171–81) inactive X-chromosome (iX) in breast carcinogenesis is not clear.Earlier studies have reported loss of iX (bar body) and gain of an aX in breast and ovarian cancers (7-10), showing that loss of iX and subsequent gain of an additional aX is not an uncommon phenomenon.More recent studies showed that Received 3/14/06; revised 11/22/06; accepted 12/27/06. activation or inactivation of certain X-linked genes is a Grant support: Elsa U.Pardee Cancer Foundation (KM Wahidur Rahman). predisposing factor for breast cancer (11).Furthermore, skewed The costs of publication of this article were defrayed in part by the payment of page charges.This article must therefore be hereby marked advertisement in Xi was shown to be a predisposing factor for the development accordance with 18 U.S.C. Section 1734 solely to indicate this fact. of invasive ovarian cancer and early onset of breast cancer Note: KM Wahidur Rahman and A.Thakur equally contributed to this work. Requests for reprints: KM Wahidur Rahman and Joshua D.Liao, Department of (12, 13).The role of X-chromosome dosage in cancer is further Pathology, Karmanos Cancer Institute, Wayne State University, 110 East Warren supported with the observation that 47 XXY males have Avenue, Detroit, MI 48201.Phone: 313-576-8273; Fax: 313-576-8389. increased predisposition for breast cancer compared with 46 E-mail: [email protected] or [email protected] Copyright D 2007 American Association for Cancer Research. XY males (14, 15).These studies, thereby, underscore the doi:10.1158/1541-7786.MCR-06-0071 significance of the X-chromosome or X-linked genes in cancer.

Mol Cancer Res 2007;5(2). February 2007 171 Downloaded from mcr.aacrjournals.org on October 2, 2021. © 2007 American Association for Cancer Research. 172 Thakur et al.

To date, there are very few studies that have addressed the human breast cancer cell lines, only one marker (i.e., IDS) was issue of whether known X-linked tumor suppressor genes are informative in 42.8% of cell lines (3 of 7 cell lines), whereas the subjected to Xi or if they escape from Xi.Investigation of G6PD locus was noninformative for breast cancer cell lines. genetic changes that occur in breast tumors and their role in Interestingly, we found that two normal cell lines (i.e., WSU15 breast carcinogenesis will elucidate novel approaches for cancer and MCF10) as well as the malignant MCF10CA1a, which was treatment.In this study, we investigated the activation status of derived from MCF10 by ectopic expression of Ras construct X-chromosome by allelotyping and reverse transcription-PCR (16), were heterozygous for IDS.Four other malignant cell analysis for two highly polymorphic X-linked loci [iduronate- lines (i.e., MDA-MB-231, MCF7, BT20, and T47D) were 2-sulfatase (IDS) and glucose-6 phosphate dehydrogenase homozygous for both IDS and G6PD.Genotyping data for cell (G6PD)] in human breast cancer cell lines and human breast lines are shown in Fig.1A. cancer biopsies.Furthermore, we assessed human breast cancer The heterozygosity of the IDS and G6PD genes was also cell lines, human breast cancer biopsies, and isogenic murine determined in 20 breast carcinoma biopsies and their adjacent cell lines with different oncogenic potentials for the expression normal/benign (bulk) breast tissues.In human breast carcino- of X-linked tumor suppressor or oncogenes by reverse ma cases, IDS was informative in 45% of cases (9 of 20 transcription-PCR and Western blot to examine whether the cases), and G6PD was informative in 27.3% of 11 remaining activation/inactivation status of X-chromosomes affects the cases (3 of 11 remaining cases).Thus, at least one marker was expression levels of these genes. informative in 60% of cases.The heterozygosity rate of IDS was similar to what has been reported for the normal healthy Results Caucasian population (17); however, no data have been Heterozygosity of Marker Genes in Breast Cancer Cell reported for G6PD in a normal healthy population.A Lines, Breast Cancer Biopsies, and Microdissected representative data for IDS genotyping in 18 cases are shown Breast Cancer and Normal Samples in Fig.1B. IDS and G6PD are two highly polymorphic genes and thus In microdissected breast cancer cells, IDS was informative in were used to determine X-chromosome activation status.In 56% of cases (14 of 25 cases), and G6PD was informative in

FIGURE 1. Genotying of human breast cancer cell lines and breast carcinoma samples. A. Genotyping for two highly polymorphic genes (IDS and G6PD)in human breast cancer cell lines. Amplified PCR product was digested with HpaII for IDS gene and PvuIforG6PD gene. IDS marker was infor- mative in 42.8% of cell lines (3 of 7 cell lines), whereas G6PD was noninformative in all cell lines. B. Representa- tive data showing IDS gene polymorphism in 18 human breast cancer (T) and adja- cent mammary tissues (N). Nine of 20 cases were het- erozygous for the IDS gene (45%). C. Representative data showing IDS gene poly- morphism in 14 microdis- sected human breast cancer and adjacent normal sam- ples. Fourteen of 25 cases were heterozygous for the IDS gene (56%).

Mol Cancer Res 2007;5(2). February 2007 Downloaded from mcr.aacrjournals.org on October 2, 2021. © 2007 American Association for Cancer Research. X-Linked Genes and Breast Cancer 173

FIGURE 2. Activation sta- tus of X-chromosome. A. RNA from three cell lines, heterozy- gous for the IDS locus, was reverse transcribed and ampli- fied by reverse transcription- PCR to detect the inactivation/ activation status of the X-chro- mosome. Amplified product was digested with HpaII and separated on a 6% nondenatur- ing polyacrylamide gel along with the undigested amplified product. All three cell lines showed only one band, indica- tive of the presence of only one active X-chromosome. Repre- sentative data for IDS gene expression. Reverse transcrip- tion-PCR analysis of two highly polymorphic genes in bulk breast cancer and adjacent normal tissues (B) and micro- dissected breast cancer and adjacent normal samples (C) were used to detect the inacti- vation/activation status of the X-chromosome in heterozy- gous patients. Representative data for IDS gene expression. Heterozygous samples showed two bands after HpaII diges- tion, suggesting the presence of two active/partially active X-chromosomes.

36% of cases (4 of 11 remaining cases).A representative data samples but activation of two X-chromosomes in adjacent breast for IDS in 14 cases are shown in Fig.1C. tissues.These results suggest that reactivation of the inactive X- chromosome is a common phenomenon in breast cancer, and that X-chromosome Status in Breast Cancer Cell Lines, this might occur at an early stage of tumor development because Breast Cancer Biopsies, and Microdissected Breast the adjacent tissue, which usually contains benign lesions, also Cancer and Normal Samples showed reactivation of the inactive X-chromosome.Results We used the transcribed polymorphisms of X-linked genes from the investigation of X-chromosome activation status in (IDS and G6PD) to characterize the activation/inactivation heterozygous cases for IDS marker are shown in Fig.2B. status of the X-chromosome in a panel of human breast cancer We compared these data (bulk tumor samples) with micro- cell lines and human breast cancer biopsy samples.Our results dissected samples.Of 14 IDS informative cases, 6 (43%) show that all three heterozygous cell lines, including two cases (cases 1, 2, 4, 7, 12, and 13) showed only one active benign cancer cell lines (WSU15 and MCF10NT) and one X-chromosome in adjacent normal tissue but two active X- breast cancer cell line (MCF10CA1a), had only one amplified chromosome in tumor cells.In two cases (cases 3 and 10), IDS cDNA band, which is indicative of one active X- tumor cells showed gain of an additional active X-chromosome chromosome.Data for X-chromosome activation status in (note an amplified product of lower band) along with the heterozygous cell lines are shown in Fig.2A. reactivation of the inactive X-chromosome.In general, these In contrast to human breast cancer cell lines, nine heterozy- data are in agreement with the data obtained from bulk tumor gous human breast cancer cases showed two bands for amplified samples.Representative data from 14 IDS informative cases are IDS cDNA.Intriguingly, of these nine cases, 5 (55%) cases shown in Fig.2C and Table 1. (cases 1, 4, 7, 13, and 14 in Fig.2B) had one active, or in some case a second partially active X-chromosome in the adjacent Expression of Tumor Suppressor Genes and Oncogenes tissue, but two equally active X-chromosomes in the cancer mRNA in Human Breast Cancer Cell Lines biopsies.In case 11, both cancer and adjacent tissues showed two Next, we tested whether the gain of an additional active X- active X-chromosomes.Three cases (cases 2, 3, and 10) showed chromosome was reflected in the expression levels of tumor a partial activation of inactive X-chromosome in cancer biopsy suppressor genes or oncogenes.For this purpose, we selected six

Mol Cancer Res 2007;5(2). February 2007 Downloaded from mcr.aacrjournals.org on October 2, 2021. © 2007 American Association for Cancer Research. 174 Thakur et al.

Table 1. Activation Status of the X-chromosome (Xa) Using IDS and G6PD Markers in Microdissected Cases

Markers Cases Informative cases Two Xa in N + T One Xa in N + T One Xa in N and two Xa in T Two Xa in N and one Xa in T

IDS 25 14 6 2 6 0 G6PD 11 (remaining) 4 2 0 0 2 Total 25 18 8 2 6 2

putative oncogenes (Gpc4, IL13R1a, Pdk3, Rsk2, Smc111, and samples but only one active X-chromosome in normal samples, Tm4sf6) and six putative tumor suppressor genes (Rsk4, Dkc1, showed increased expression of RbAp46 and Rsk4 genes Apx1, cldn2, Fgf13, and RbAp46) based on our previous report (Table 2).Similarly, tumor samples from cases 3 and 10, which of the microarray analysis of X-linked genes in mammary showed reactivation of the inactive X-chromosome as well as tumors from MMTV-c-myc transgenic mice (18).The expres- duplication of an active X-chromosome, also showed increased sion patterns of the putative X-linked tumor suppressor genes expression of RbAp46 and Rsk4 transcripts, suggesting that Rsk4, Dkc1, Apx1, Cldn2, Fgf13, and RbAp46 as well as copy number may influence the expression levels of X-linked oncogenes Gpc4, IL13R1a, Pdk3, Rsk2, Smc111, and Tm4sf6 oncogenes or tumor suppressor genes. did not reveal any specific pattern representing activation of one or more X-chromosomes.These results suggest a cell line– Expression of Tumor Suppressor Genes and Oncogenes specific regulation of the gene activation/inactivation mecha- in Murine Mammary Epithelial Cell Lines nism.Of interest was the expression of a putative tumor For an additional study of tumor suppressor and oncogenes suppressor gene Apx1, which was only found in the malignant expression in cells with different oncogenic potentials, we cell lines and not in the WSU15 and MCF10NT benign cell lines used murine sublines generated from the benign mouse and the MCF10-derived malignant clone MCF10CA1a.Results mammary epithelial cell line NMuMG.These sublines showed are shown in Fig.3A and B. variable oncogenic potential ranging from low to highly invasive cells in a soft agar assay and in a severe combined Expression of Tumor Suppressor Genes and Oncogenes immunodeficient mouse model.Interestingly, two highly mRNA in Human Breast Cancer Samples invasive sublines, transfected with either cyclin D1 (ND5) Similar to cell lines, the expression profiles of tumor or cyclin D1/c-Myc (ND5/C23), showed a differential expres- suppressor genes and oncogenes were also tested in human sion pattern compared with a low-oncogenic c-myc (NC2)– breast cancer biopsies and adjacent mammary tissues.Surpris- transfected cell line and vector control cells (NN2 or NH2). ingly, we found that a large percentage of breast cancer cases Both ND5 and ND5/C23 cell lines showed very low levels, or had high expression of two putative tumor suppressor genes no expression of tumor suppressor genes and high expression (Rsk4 and RbAp46), which have been shown to be growth of most of oncogenes. inhibitory genes in in vitro studies (19-23).The RbAp46 gene The expression of Cldn2, a putative tumor suppressor gene, showed increased expression in 79% of cases (15 of 19) was almost undetectable in the highly malignant ND5/C23 cell compared with adjacent mammary tissues.Similarly, 47% of clone, whereas it was weakly expressed in ND5 cells that are cases (9 of 19) showed higher expression of Rsk4 in breast less invasive than ND5/C23 cells.These data are in line with cancer tissues compared with adjacent mammary tissues. our finding in human breast cancer biopsies and suggest that Conversely, Cldn2 expression was reduced in 52.6% of cases decreased expression of Cldn2 may occur in more advanced (10 of 19) compared with adjacent normal tissues (Fig.4A). stages of breast cancer. The expression of putative oncogenes, such as Gpc4, Pdk3, The expression of RbAp46 was higher in ND5 and ND5/ Smc111, and Rsk2, was also analyzed in the human cancer C23 malignant cells as well as in c-myc –expressing NC2 biopsies.All these genes showed increased expression in breast cells than in the vector-expressing cells, thus suggesting that cancer biopsies compared with their adjacent normal tissue; the RbAp46 may be an oncogene in malignant cells.Interest- frequencies of aberrant expression for Gpc4, Pdk3, Smc111, and ingly, Rsk4 expression seems to be regulated by c-Myc as Rsk2 were 42% (5 of 12), 58% (7 of 12), 58% (7 of 12), and the NC2 clone showed high expression of Rsk4.It is likely 25% (3 of 12), respectively.Results are shown in Fig.4B. that Rsk4 may mediate the dual functions of c-Myc under Next, we determined whether activation status of X- certain cellular context.However, it is yet to be determined chromosome influences the expression levels of some of the whether or not this gene also has dual roles like c-Myc in key X-linked genes, such as RbAp46, Rsk4, and Cldn2,in growth inhibition and proliferation.The expression of Apx1 microdissected samples.Both RbAp46 and Rsk4 showed was lost in the ND5 and ND5/C23 malignant cells and increased expression in 77% and 50% tumor cases, respectively, highly expressed in the benign cell clones; yet, in the human compared with adjacent normal tissues (Table 2).Although cell lines, it was highly expressed in malignant cell lines.On these data are surprisingly similar to the percentage of positive the other hand, Apxl expression could not be detected in any cases from bulk tumor and normal adjacent tissue samples but of the human breast carcinoma or adjacent mammary tissues. showed different expression profile with regard to X-chromo- The reason for this controversy is unclear and remains to be some activation status.It is noteworthy that six cases, which addressed.Expression levels of putative oncogenes ( Gpc4, showed two active X-chromosome in microdissected tumor IL13R1a, and Pdk3) were increased in ND5 and ND5/C23

Mol Cancer Res 2007;5(2). February 2007 Downloaded from mcr.aacrjournals.org on October 2, 2021. © 2007 American Association for Cancer Research. X-Linked Genes and Breast Cancer 175

manner as mRNA expression levels in human breast cancer samples.The immunostaining for Rsk4, RbAp46, and Cldn2 proteins was carried in a series of 20 paraffin-embedded sections of breast cancer biopsies (Table 3).Some of the human breast cancer sections contained areas of ductal carcinoma in situ and invasive ductal carcinoma as well as nonneoplastic or benign tissue, therefore allowing us to assess these proteins during breast cancer progression.Consistent with mRNA expression data, Rsk4 was positively stained in both nucleus and cytoplasm of 35% of invasive breast carcinoma cases.Interestingly, normal mammary ducts also showed strong expression of Rsk4, which was completely or partially lost in benign breast lesions (papilloma) on the same section (Fig.6A, e and f).Our data show that RbAp46 was strongly expressed in the nucleus and cytoplasm in 40% of invasive ductal carcinoma lesions, whereas the remaining 20% of cases showed moderate to weak expression of the RbAp46 protein (Fig.6 B).Serial sections stained with anti-RbAp46 and anti-Rsk4 antibodies showed that f50% of Rsk4-positive cases were also positively stained for the RbAp46 protein in human breast cancers.The Cldn2 protein was weak to moderately expressed in 20% of the cases and localized in both nucleus and cytoplasm of invasive ductal cancer cells (Fig.6 C).

Discussion FIGURE 3. X-linked gene expression in human breast cancer cell lines. Reverse transcription-PCR analysis for the expression of selected putative This study shows two important findings: (a) RbAp46 and tumor suppressor (A) and oncogenes (B) showed variable expression Rsk4 genes may play roles in the development or progression of levels among breast cancer cell lines with more than one active X- human breast cancer.( b) The expression of X-linked tumor chromosomes, suggesting cell line – specific activation or inactivation of tumor suppressor genes or oncogenes. suppressor genes and oncogenes seems to be regulated at the individual gene level, and overexpression may be associated with the X-chromosome copy numbers. cell clones compared with benign cell clones NC2, NN2, and We found that 60% human breast cancer samples showed NH2.Results are shown in Fig.5A and B. heterozygosity for two highly polymorphic markers tested in this study, whereas most breast cancer cell lines (4 of 5) were Protein Expression of Tumor Suppressor Genes and homozygous for both these markers.These data are consistent Oncogenes in Human Breast Cancer Samples with the recently published report by Sirchia et al.(24), which We performed the immunostaining to confirm whether also showed a high percentage of heterozygosity in human protein expression is also increased or decreased in a similar breast carcinoma cases and homozygosity in breast cancer cell

FIGURE 4. X-linked gene expression pattern of putative tumor suppressor genes and oncogenes in human breast cancer and adjacent mammary tissue biopsies with one or more active X-chromosomes. Reverse transcription-PCR analysis for the expression of selected putative tumor suppressor genes (A) and oncogenes (B) showed variable expression levels among different breast cancer samples and their adjacent benign mammary tissues.

Mol Cancer Res 2007;5(2). February 2007 Downloaded from mcr.aacrjournals.org on October 2, 2021. © 2007 American Association for Cancer Research. 176 Thakur et al.

Table 2. Activation Status of the X-chromosome and Expression Intensities of RbAp46, Rsk4, and Cldn2 mRNA in 18 Informative Microdissected Cases

Cases Expression intensity

RbAp46 Rsk4 Cldn2

Adjacent normal tissue Tumor tissue Adjacent normal tissue Tumor tissue Adjacent normal tissue Tumor tissue

Two Xa (N+T) 3 ++ ++++ +++ ++++ +++ +/À 5 À ++++++/À + 6 + ++ ++ ++ ++ ++ 8+/À ++À +/À +/À 10 ++ +++++ À +++ ++ +++ 14 ++ +++ ++ + ÀÀ 15 ÀÀ+ ++++++ 16 À +/ÀÀ ++ ++ À One Xa (N+T) 9+ ++ À ++ ÀÀ 11 + +++ À ++++ One Xa in N and two Xa in T 1 + ++ +/À ++++ 2+/À +++ ++ +++ + + 4 À ++++++/À + 7 ++ ++++ +++ ++++ +++ ++ 12 + +++ ++ +++ ++ +++ 13 ++ +++ ++ + + ++ Two Xa in N and one Xa in T 17 +++ + +++ ++ ++++ ++ 18 ++ ++ + À +/ÀÀ

lines.However, the number of human breast carcinoma cases reported in cases of hepatoblastomas (26).Because inactivation used in the investigations by Sirchia et al.were small; in of X-chromosomes only occurs during a short window of addition, the authors used corresponding peripheral blood embryonic life, additional gains of X-chromosome through the lymphocytes as a normal control.In the present study, we used duplication of aX are not likely to be inactivated.Two separate 20 human breast cancer biopsy samples along with the adjacent studies by Van Dyke et al.(27) and Wang et al.(11) have non-tumor mammary tissues and 25 microdissected human shown that postembryonic nondisjunction of the active X- breast cancer samples.In all human heterozygous cases, we chromosome is not followed by inactivation in tumors.Similar found that the polymorphic transcripts of IDS and G6PD were results were shown by Sirchia et al.(24) such that in spite of expressed from two alleles, indicating the presence of two the presence of two or more X-chromosomes, none of them active X-chromosomes.The higher frequency of heterozygosity was functionally inactivated, irrespective of BRCA1 status and in breast cancer cases may suggest two possible mechanisms: Xist expression in a panel of wild-type BRCA1 and BRCA1- either the reactivation of whole inactive X-chromosome or deficient breast cancer cell lines and non-cultured primary selective activation of some regions or genes on the inactive tumors. X-chromosome.Higher frequency of heterozygosity for at least The expression profiles of tumor suppressor genes or two informative markers indicates that the reactivation of an iX oncogenes in human breast cancer samples showed surprisingly is equally as common as the duplication of an aX in human higher expression levels of two putative tumor suppressor genes breast carcinoma cases.On the other hand, similar to previous (RbAp46 and Rsk4).A higher expression of these two putative findings, human breast cancer cell lines showed a high tumor suppressor genes in breast cancer biopsy samples frequency of homozygosity (24).This showed that in spite compared with adjacent normal mammary tissues conflicts the presence of more than one active X-chromosome, MCF7, with previous in vitro studies that suggest that these two genes T47D, and MB231 cell lines were homozygous, thereby have growth inhibitory activity (19-23).However, this is the suggesting duplication of aX.These results are not surprising first study to screen the mRNA and protein expression of these as continuous culturing of breast cancer cell lines can lead to two genes (RbAp46 and Rsk4) and other putative tumor further mutations, gain/loss of chromosomes, and deletion and suppressor gene (Cldn2) in human breast cancer and adjacent translocation of chromosomal segments compared with non- normal tissues.The Rsk4 protein was strongly expressed in cultured breast carcinoma samples. normal mammary ducts but was completely or partially lost in Duplication of an active X-chromosome is also reported in benign mammary tissue and reappeared in ductal carcinoma other cancer types.McDonald et al.(25) reported duplication of in situ and invasive ductal carcinomas.It is likely that sup- the active X-chromosome in at least 20% of non-Hodgkin’s pression of Rsk4 may provide a growth advantage and thus may lymphomas, whereas duplication of the inactive chromosome play a role in the early stages, such as initiation of tumori- was not associated with any specific genetic abnormality. genesis.The expression pattern of RbAp46 in cancer versus Similarly, a high frequency of X-chromosome gains was adjacent tissue suggests that RbAp46 might act as an oncogene

Mol Cancer Res 2007;5(2). February 2007 Downloaded from mcr.aacrjournals.org on October 2, 2021. © 2007 American Association for Cancer Research. X-Linked Genes and Breast Cancer 177

culturing conditions.Although isogenic human cell lines with different oncogenic potentials, such as the MCF10 series, are widely used, these cell lines have been established for decades and cultured under various conditions and thus are likely to have accumulated many more genetic defects than breast cancer in patients.Our results in these sublines clearly show that the expression of transcripts for most of the tumor suppressor genes is either lost or decreased, as there is a progression from less malignant to more malignant phenotype. Intriguingly, the expression pattern of RAp46 and Cldn2 in NMuMG clones corroborate our results in human carcinoma cases; for example, RbAp46 expression was higher in more malignant cell clones, whereas Cldn2 expression was lost in a more malignant cell clone, similar to our findings in human breast cancer cases. Surprisingly, in microdissected tumor samples, we ob- served difference between samples with reactivation or partial reactivation of inactive X-chromosomes (heterozygous for two markers) or duplication of an active X-chromosome (homo- zygous for two markers) in the expression pattern of tumor suppressor genes/oncogenes.However, it is not clear ( a) whether gain of an additional aX is a consequence rather than a cause of breast carcinogenesis, or (b) whether this consequence contributes to the disease progression.Alterna- tively, it is likely that the heterogeneous nature of breast FIGURE 5. The gene expression profile of X-linked putative tumor cancer and its microenvironment dictates the selective up- suppressor or oncogenes in a series of murine mammary epithelial cell lines ranging from benign (NN2 and NH2), premalignant (NC2), invasive regulation or down-regulation of certain X-linked genes at the (ND5), to highly invasive (ND5/C23) cell clones. These cell lines with gene level regardless of chromosomal copy number or different oncogenic potential showed low levels to complete loss of some activation status.Gilbert et al.(28) also suggested that the tumor suppressor genes (A) with the progression of their oncogenic potential (ND5 and ND5/C23 clones) and higher expression of putative promoter-specific hypoacetylation may be a key component oncogenes (B). of the X-inactivation machinery that operates at the level of individual genes.Nevertheless, we do not exclude the possibility of aneuploidy-associated aberrant gene expression. in breast cancer, although we cannot rule out that it may have To this end, Sirchia et al.(24) suggested that overexpression a growth-suppressive activity or a cell-specific role in other of X-linked genes may be a key point of breast tumorigenesis tissues or cell line models.Interestingly, we found that the in the absence of gene silencing on all of the X- expression of Cldn2, a putative tumor suppressor protein, was chromosomes, which is based on the finding that most weak in most of the cases, and only 20% of breast cancer human tumor cell lines or human breast carcinoma cases have cases showed positive staining, suggesting that Cldn2 may be more than one aX. a tumor suppressor protein.Strikingly, the expression of Cldn2 In summary, for the first time, our data show higher was also lost in the more malignant ND5/C23 cell clone of expression of RbAp46 in breast carcinoma cases compared with NMuMG, suggesting that loss or reduced expression of Cldn2 adjacent normal tissues, suggesting that this gene may play a may provide a growth advantage.Further studies in more breast role in the progression of breast cancer.On the other hand, cancer biopsies with regard to histologic patterns and tumor expression pattern of Rsk4 and decreased expression of Cldn2 stages are needed to confirm the involvement of Rsk4, RbAp46, in most of the breast cancer cases suggest that these genes may and Cldn2 in breast carcinogenesis.The expression levels of have potential growth inhibitory effects in breast cancer, tumor suppressor genes and oncogenes also varied among although it is likely that cellular microenvironment may breast cancer cell lines, suggesting that there may be several influence the activity of these genes.Exploration of these hierarchical levels of control that regulate X-linked gene expression. Furthermore, we dissected the expression pattern of tumor Table 3. Immunostaining in 20 Breast Cancer and Adjacent Benign or Normal Mammary Tissues suppressor genes and oncogenes in murine NMuMG sublines. We used these sublines for two important reasons.First, this Total positive (%) series of cell lines, which were generated from a benign mouse mammary epithelial cell line, have different oncogenic Cancer cells Adjacent normal potentials ranging from nonmalignant to highly malignant cell RbAp46 12 (60) 2 (10) clones.Second, and most importantly, these cell lines have RSK4 7 (35) 5 (30) recently been generated in our lab and thus may not have Cldn2 4 (20) 5 (30) accumulated additional genetic defects due to continuous

Mol Cancer Res 2007;5(2). February 2007 Downloaded from mcr.aacrjournals.org on October 2, 2021. © 2007 American Association for Cancer Research. 178 Thakur et al.

FIGURE 6. A. Immunostaining for Rsk4 protein in breast cancer biopsies. a, weak expression of cytoplasmic Rsk4 protein in infiltrative invasive ductal can- cer. b, moderate cytoplasmic and nuclear staining in a representative example of invasive ductal carcinoma. c, strong Rsk4 immunoreactivity in invasive ductal carci- noma-colloid type. d, representative case showing a strong positive staining for Rsk4 in ductal carcinoma in situ. e, no staining was detected in a benign breast lesion (papilloma). f, strong nuclear (inset) and cytoplasmic staining in normal mam- mary ducts. B. Immunostaining for RbAp46 protein in breast cancer biopsies. a and b, strong expression of both nuclear and cytoplasmic RbAp46 protein in invasive ductal cancer. c and d, strong expression of nuclear and cytoplasmic RbAp46 ex- pression in invasive ductal cancer-colloid type. e, representative example showing no detectable RbAp46 protein in invasive ductal cancer. f, representative example showing no detectable RbAp46 protein in benign tissue. C. Immunostaining for Cldn2 protein in breast cancer biopsies. a and b, strong nuclear and cytoplasmic (inset) staining in invasive ductal carcinoma- colloid type. b and c, representative examples showing weak cytoplasmic and nuclear staining (inset) in invasive ductal carcinomas. d and e, negative staining in invasive ductal carcinoma. f, benign tissue showing a weak immunostaining for Cldn2 protein. High-power magnifica- tion, Â400.

Mol Cancer Res 2007;5(2). February 2007 Downloaded from mcr.aacrjournals.org on October 2, 2021. © 2007 American Association for Cancer Research. X-Linked Genes and Breast Cancer 179

oncogenes and tumor suppressor genes may elucidate novel Genomic DNA was extracted by proteinase K digestion, approaches for cancer treatment. phenol/chloroform purification, and dehydration.Purified DNA from cancer and adjacent normal/benign tissues was Materials and Methods PCR amplified for IDS and G6PD, two X-linked genes known Human and Mouse Cell Lines and Culture Conditions to be highly polymorphic (29).Total RNA was extracted by Breast cancer cell lines MCF10CA1a, MCF7, BT20, T47D, using Trizol reagent (Invitrogen) according to the manufac- and MDA-MB-231 (MB-231) and benign breast cancer cell turer’s instruction and immediately converted to cDNA using a lines MCF10NT and MCF12 were maintained in DMEM, Taqman reverse-transcription kit (Applied Systems, Foster City, À j DMEM/F12, or RPMI 1640 (Invitrogen, San Diego, CA) with CA) and stored at 20 C until further use. 10% FCS.For the determination of inactivation status, DNA and RNA were extracted from f80% confluent 10-cm2 plates. Genotyping of IDS and G6PD Polymorphic X-Linked Loci The mouse mammary epithelial cell line NMuMG obtained Genomic DNA extracted from human breast cancer cell lines from the American Type Culture Collection (Manassas, MD) and human breast cancer and adjacent tissues was used for was used to generate the stably transfected cell clones over- genotyping of IDS and G6PD polymorphisms as described expressing c-myc (NC2), cyclin D1 (ND5), or both cyclin D1/c- previously (8, 29). IDS and G6PD alleles were differentiated by myc (ND5/C23) along with vector control clones [NN2 (vector PCR amplification with allele-specific primers.The primers control for G418) and NH2 (vector control for hygromycin)]. used for the detection of polymorphic regions of these genes are NMuMG is a benign cell line, suitable to investigate the described previously (29).Primers and PCR conditions are mechanism of carcinogenesis of breast cancer as a normal shown in Table 4.Briefly, a mismatch was introduced in the mammary epithelial cell line.Cells were cultured in DMEM sequence of one primer to create a restriction site specifically in (Invitrogen) with 10% fetal bovine serum either in the presence PCR products from one of the two alleles (Table 4A). of G418 or hygromycin (Invitrogen) at 37jC and 5% CO2. Restriction fragments were separated by nondenaturing acryl- amide gel electrophoresis after digestion with the specific Tissue Samples and Laser-Assisted Microdissection enzymes and visualized by UV illumination.In heterozygous Forty-five histologically confirmed frozen or paraffin- females, two bands were observed (151 and 126 bp for IDS and embedded tissues of breast cancer cases and corresponding 183 and 162 bp for G6PD), whereas only one band was found normal/benign tissues were obtained from the repository at the in homozygous females. Pathology Department of Harper University Hospital of Wayne State University.We excluded those cases, which Determination of Inactivation Status of the X-chromo- received any major treatment before surgery and biopsy somes sample collection.We used 20 frozen bulk tumor and adjacent Similar to genomic DNA amplification, cDNA from human normal tissues in this study.For laser-assisted microdissection, breast cancer cell lines and human breast cancer and adjacent histologic sections were prepared from 25 formalin-fixed, normal/benign tissues was amplified, and the resulting paraffin-embedded tissues and were stained with H&E for products were digested and analyzed on nondenaturing microscopic evaluation.In these 25 samples, tumor tissues acrylamide gel to detect the activation/inactivation status of were identified and microdissected from the 5-Am histologic the X-chromosomes.Samples that showed two bands (158 sections.Adjacent normal tissues from each case were and 133 for IDS and 79 and 58 for G6PD) were identified to microdissected as a control. possess two active X-chromosomes through the reactivation of an inactive X-chromosome, whereas a single band may DNA and RNA Extraction represent either two active X-chromosomes through duplica- We used the 25 microdissected tumor and adjacent normal tion or only one active X-chromosome depending on the band samples and 20 bulk frozen tumor and adjacent normal tissues density relative to undigested PCR product.Primer sequence as well as human and murine breast cancer or mammary and conditions are described previously (29) and shown in epithelial cell lines for RNA and genomic DNA extraction. Table 4B.

Table 4. Primer Sequences for Genotyping and for the Determination of Activation Status

Gene Primers Sequence 5¶-3¶ Exon PCR product size Enzymes

A.Primer sequences for genotyping and PCR product size IDS IDS 3b (sense) CCAAAGAAGGGAGGGTCCAC Intron 3 151 HpaII IDS 4 (antisense) AGACCAGCTATACGGAGAATCAC*C 4 G6PD G6PD-E10 (sense) GCTGGACCTGACCTACGGCAACA 10 183 PvuI G6PVPDPVU1(antisense) GAAGACGTCCAGGATGAGGCGA*TC 11

B.Primer sequence for reverse transcription-PCR analysis and PCR product size IDS IDS 3a (sense) TGTACGACTTCAACTCCTACTGGA 3 158 HpaII IDS (antisense) AGACCAGCTATACGGAGAATCAC*C 4 G6PD G6PD-E10 (sense) GCTGGACCTGACCTACGGCAACA 10 79 PvuI G6PVPDPVU1 (antisense) GAAGACGTCCAGGATGAGGCGA*TC 11

Mol Cancer Res 2007;5(2). February 2007 Downloaded from mcr.aacrjournals.org on October 2, 2021. © 2007 American Association for Cancer Research. 180 Thakur et al.

Reverse Transcription-PCR for the Expression of Putative rehydrated in graded ethanol followed by antigen retrieval Tumor Suppressor Genes and Oncogenes and peroxidase blocking.Primary antibody to RbAp46, RNA extracted from human breast cancer cell lines, Rsk4, or Cldn2 (1:50 dilution) was applied for 90 min then human breast cancer and adjacent normal tissues, and mouse sections were incubated with biotinylated goat anti-rabbit NMuMG sublines was reverse transcribed to prepare cDNA. secondary antibody (1:1,000) for 60 min at room temper- For subsequent PCR amplification, a maximum of 2 ALof ature.Antigen was detected by using a complex of avidin each cDNA sample was used per 25-AL PCR reactions. and horseradish peroxidase with diaminobenzidine staining The specific primers and PCR product size are listed in system.Sections were then lightly counterstained with Table 5A and B.The amplified products were electrophoret- hematoxylin, dehydrated, and mounted.Omitting the prima- ically separated on ethidium bromide containing 1.2% ry antibody from the immunohistochemical procedure and agarose gels. replacing it with non-immune rabbit serum acted as negative controls.Besides, different primary antibodies also served as Immunostaining for Putative Tumor Suppressor Gene mutual controls.Immunoreactivity for each antigen was and Oncogene Proteins evaluated by examination of the sections under bright-field Formalin-fixed, paraffin-embedded human breast cancer light microscopy independently by two experienced observ- sections (5 Am thick) were deparaffinized in xylene and ers (H.N. and A.T.).

Table 5. Human and Mouse Primer Sequences

Gene name Accession no.Primer sequence Size of PCR product (bp) T m (jC)

A.Human primer sequences and conditions for reverse transcription-PCR APXL X83543 Upstream, 5¶-CGGATGTTCATTGGAGACC-3¶ 275 58 Downstream, 5¶-TTCACGAAGTGCTCATAGTC-3¶ 58 IL13RA1 NM_001560 Upstream, 5¶-ACACAGTCAGAGTAAGAGTC-3¶ 223 58 Downstream, 5¶-CAGGATCAGGAATTGGAGG-3¶ 58 GPC4 NM_001448 Upstream, 5¶-TCAAGTCGAAAAGTTGCTCG-3¶ 249 58 Downstream, 5¶-ATTCATCAAACTTCTTGTAACG-3¶ 58 Tm4sf6 AY888941 Upstream, 5¶-AGACATGAGATTAAGAACAGC-3¶ 293 58 Downstream, 5¶-ACGACTCCCATTTCTGACTC-3¶ 60 Pdk3 NM_005391 Upstream, 5¶-CAAGCAGATCGAGCGCTAC-3¶ 420 60 Downstream, 5¶-GGTTGGTATAAAACCGATCC-3¶ 58 DKC1 AJ224481 Upstream, 5¶-TTCTGGACAAGCATGGGAAG-3¶ 374 61 Downstream, 5¶-AGCTTCAAGTGGCCTTCACT-3¶ 59 Rsk4 NM_014496 Upstream, 5¶-GCCCCAATGATACTCCTGAA-3¶ 396 60 Downstream, 5¶-CATCCAGTTTGGCCTAGGAA-3¶ 60 Smc1l1 BC080185 Upstream, 5¶-GGGCATTCTCATCAAAGCTC-3¶ 447 60 Downstream, 5¶-ATTTTGCCCAGCTCCTTCTT-3¶ 60 RbAp46 NM_002893 Upstream, 5¶-CATGCAGATATGGCAAATGG-3¶ 401 60 Downstream, 5¶-TGCATATTTGGGAACAGCAA-3¶ 60 Fgf13 BC018238 Upstream, 5¶-AATTGGCTTCTCAGGCTCAA-3¶ 595 60 Downstream, 5¶-TGCACAGGACAGAAAACTGC-3¶ 60 Cldn2 BC014424 Upstream, 5¶-ACTGTCCATCGGAAGATGCT-3¶ 441 60 Downstream, 5¶-ACGCTGAGGAAGTTCTCCAA-3¶ 60 Rsk2 BC096303 Upstream, 5¶-TTTGCAAATGGTCCTGATGA-3¶ 367 60 Downstream, 5¶-GGGCTGTTGAGGTGATTTTT-3¶ 60

B.Mouse primer sequences and conditions for reverse transcription-PCR APXL XM_486797 Upstream, 5¶-CGGATGTTCATTGGAGACC-3¶ 275 58 Downstream, 5¶-TTCACGAAGTGCTCATAGTC-3¶ 58 IL13RA1 NM_133990 Upstream, 5¶-ACACAGTCAGAGTAAGAGTC-3¶ 223 58 Downstream, 5¶-CAGGATCAGGAATTGGAGG-3¶ 58 GPC4 BC006622 Upstream, 5¶-TCAAGTCGAAAAGTTGCTCG-3¶ 249 58 Downstream, 5¶-ATTCATCAAACTTCTTGTAACG-3¶ 58 Tm4sf6 NM_019656 Upstream, 5¶-AGACATGAGATTAAGAACAGC-3¶ 293 58 Downstream, 5¶-ACGACTCCCATTTCTGACTC-3¶ 60 Pdk3 NM_145630 Upstream, 5¶-CAAGCAGATCGAGCGCTAC-3¶ 420 60 Downstream, 5¶-GGTTGGTATAAAACCGATCC-3¶ 58 Fgf13 AF020737 Upstream, 5¶-CATTTTCTGCCCAAACCACT-3¶ 378 60 Downstream, 5¶-AATGCTTGGCACTCTTTTGC-3¶ 60 Cldn2 NM_016675 Upstream, 5¶ AATCTTCATGGCATCCTTCG-3¶ 437 60 Downstream, 5¶-CCAATGGCCAAAGTCATTCT-3¶ 60 DKC1 AJ224481 Downstream, 5¶-GACCGGAAGCCATTACAAGA-3¶ 374 61 Downstream, 5¶-GCTACCACCTCATGGGAAGA-3¶ 59 Rsk4 NM_014496 Upstream, 5¶-GTGGGTGCCAAAGTTTTGAT-3¶ 396 60 Downstream, 5¶-CAAACCACATGGAAATCAGG-3¶ 60 Smc1l1 BC080185 Upstream, 5¶-GGGCATTCTCATCAAAGCTC-3¶ 447 60 Downstream, 5¶-ATTTTGCCCAGCTCCTTCTT-3¶ 60 RbAp46 NM_002893 Upstream, 5¶-CATGCAGATATGGCAAATGG-3¶ 401 60 Downstream, 5¶-TGCATATTTGGGAACAGCAA-3¶ 60 Rsk2 NM_148945 Upstream, 5¶-ATGGAGTTTGCCGTGAAGAT-3¶ 360 60 Downstream, 5¶-TGCAAAGCCAAAATCACAAA-3¶ 60

Mol Cancer Res 2007;5(2). February 2007 Downloaded from mcr.aacrjournals.org on October 2, 2021. © 2007 American Association for Cancer Research. X-Linked Genes and Breast Cancer 181

the fibroblastic phenotype of Ras-transformed epithelial cells.Mol Biol Cell References 1997;8:2329 – 44. 1.Heard E, Clerc P, Avner P.X-chromosome inactivation in mammals.Annu Rev Genet 1997;31:571 – 610. 17.Gregg XT, Kralovics R, Prchal JT.A polymorphism of the X-linked gene IDS increases the number of females informative for transcriptional clonality 2.Lyon MF.Geneaction in the X-chromosome of the mouse ( Mus musculus L.). assays.Am J Hematol 2000;63:184 – 91. Nature 1961;190:372 – 3. 18.Thakur A, XH, Wang Y, Bollig A, Biliran H, Liao DJ.Role of X-linked 3.Buller RE, Sood AK, Lallas T, Buekers T, Skilling JS.Association between genes in breast cancer.Breast Cancer Res Treat 2005;93:135 – 43. nonrandom X-chromosome inactivation and BRCA1 mutation in germline DNA of patients with ovarian cancer.J Natl Cancer Inst 1999;91:339 – 46. 19.Berns K, Hijmans EM, Mullenders J, et al.A large-scale RNAi screen in human cells identifies new components of the p53 pathway.Nature 2004;428: 4.Liao DJ, Du QQ, Yu BW, Grignon D, Sarkar FH.Novelperspective: focusing 431 – 7. on the in reproductive cancers.Cancer Invest 2003;21:641 – 58. 20.Dummler BA, Hauge C, Silber J, et al.Functional characterization of human 5.Mitelman F.Chromosomes, genes, and cancer.CA Cancer J Clin 1994;44: RSK4, a new 90-kDa ribosomal S6 kinase, reveals constitutive activation in most 133 – 5. cell types.J Biol Chem 2005;280:13304 – 14. 6.Brown CJ.Role of the X chromosome in cancer.JNatl Cancer Inst 1996;88: 21.Myers AP, Corson LB, Rossant J, Baker JC.Characterization of mouse Rsk4 480 – 2. as an inhibitor of fibroblast growth factor-RAS-extracellular signal-regulated 7.Camargo M, Wang N.Cytogenetic evidence for the absence of an inactivated kinase signaling.Mol Cell Biol 2004;24:4255 – 66. X chromosome in a human female (XX) breast carcinoma cell line.Hum Genet 22.Zhang TF, Yu SQ, Deuel TF, Wang ZY.Constitutive expression of Rb 1980;55:81 – 5. associated protein 46 (RbAp46) reverts transformed phenotypes of breast cancer 8.Ghosh SN, Shah PN.Probable mechanism for the loss of Barr body in human cells.Anticancer Res 2003;23:3735 – 40. female tumor with special reference to breast cancer.Med Hypotheses 1981;7: 23.Zhang TF, Yu SQ, Loggie BW, Wang ZY.Inducible expression of RbAp46 1099 – 104. activates c-Jun NH2-terminal kinase-dependent apoptosis and suppresses 9.Ghosh SN, Shah PN, Banerji AP.Loss of X-chromatin and X-chromosome progressive growth of tumor xenografts in nude mice.Anticancer Res 2003;23: activation in HeLa cells.Acta Cytol 1979;23:169 – 72. 4621 – 7. 10.Rajeswari S, Ghosh SN, Shah PN, Borah VJ.Barr body frequency in the 24.Sirchia SM, Ramoscelli L, Grati FR, et al.Loss of the inactive X human breast cancer tissue.A prospective study on its prognostic value and its chromosome and replication of the active X in BRCA1-defective and wild-type correlation with specific oestradiol receptors.Eur J Cancer 1977;13:99 – 102. breast cancer cells.Cancer Res 2005;65:2139 – 46. 11.Wang N, Cedrone E, Skuse GR, Insel R, Dry J.Two identical active X 25.McDonald HL, Gascoyne RD, Horsman D, Brown CJ.Involvementof the X chromosomes in human mammary carcinoma cells.Cancer Genet Cytogenet chromosome in non-Hodgkin lymphoma.Genes Chromosomes Cancer 2000;28: 1990;46:271 – 80. 246 – 57. 12.Buekers TE, Lallas TA, Buller RE.Xp22.2– 3 loss of heterozygosity is 26.Terracciano LM, Bernasconi B, Ruck P, et al.Comparative genomic hybridization analysis of hepatoblastoma reveals high frequency of X- associated with germline BRCA1 mutation in ovarian cancer.Gynecol Oncol 2000;76:418 – 22. chromosome gains and similarities between epithelial and stromal components. Hum Pathol 2003;34:864 – 71. 13.Kristiansen M, Langerod A, Knudsen GP, Weber BL, Borresen-Dale AL, 27.Van Dyke DL, Abraham JP, Maeda K, Weiss L, Poel M.Multiple active X Orstavik KH.High frequency of skewed X inactivation in young breast cancer chromosomes in myelofibrosis with myeloid metaplasia.Cancer Genet Cytogenet patients.J Med Genet 2002;39:30 – 3. 1981;3:137 – 44. 14.Giordano SH, Cohen DS, Buzdar AU, Perkins G, Hortobagyi GN.Breast 28.Gilbert SL, Sharp PA.Promoter-specific hypoacetylation of X-inactivated carcinoma in men: a population-based study.Cancer 2004;101:51 – 7. genes.Proc Natl Acad Sci U S A 1999;96:13825 – 30. 15.Rudas M, Schmidinger M, Wenzel C, et al.Karyotypicfindings in two cases 29.el-Kassar N, Hetet G, Briere J, Grandchamp B.Clonality analysis of of male breast cancer.Cancer Genet Cytogenet 2000;121:190 – 3. hematopoiesis in essential thrombocythemia: advantages of studying T lympho- 16.Zhong C, Kinch MS, Burridge K.Rho-stimulated contractility contributes to cytes and platelets.Blood 1997;89:128 – 34.

Mol Cancer Res 2007;5(2). February 2007 Downloaded from mcr.aacrjournals.org on October 2, 2021. © 2007 American Association for Cancer Research. Aberrant Expression of X-Linked Genes RbAp46, Rsk4, and Cldn2 in Breast Cancer

Archana Thakur, KM Wahidur Rahman, Jack Wu, et al.

Mol Cancer Res 2007;5:171-181.

Updated version Access the most recent version of this article at: http://mcr.aacrjournals.org/content/5/2/171

Cited articles This article cites 28 articles, 7 of which you can access for free at: http://mcr.aacrjournals.org/content/5/2/171.full#ref-list-1

Citing articles This article has been cited by 10 HighWire-hosted articles. Access the articles at: http://mcr.aacrjournals.org/content/5/2/171.full#related-urls

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://mcr.aacrjournals.org/content/5/2/171. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.

Downloaded from mcr.aacrjournals.org on October 2, 2021. © 2007 American Association for Cancer Research.