Advances in Brief

[CANCER RESEARCH 62, 1289–1295, March 1, 2002] Advances in Brief

Stanniocalcin 2 Is an Estrogen-responsive Gene Coexpressed with the Estrogen Receptor in Human Breast Cancer1

Toula Bouras,2 Melissa C. Southey, Andy C. Chang, Roger R. Reddel, Dorian Willhite, Richard Glynne, Michael A. Henderson, Jane E. Armes, and Deon J. Venter Cancer Functional Genomics Unit, Murdoch Children’s Research Institute, 10th Floor Royal Children’s Hospital, Parkville, Victoria 3052, Australia [T. B., D. J. V.]; Department of Pathology, The University of Melbourne, Victoria 3010, Australia [M. C. S., J. E. A., D. J. V.]; Molecular Pathology Laboratory, Victorian Breast Cancer Research Consortium, Australia [J. E. A.]; Department of Surgery, St Vincent’s Hospital, Fitzroy, Victoria 3065, Australia [M. A. H.]; Children’s Medical Research Institute, Westmead, Sydney, NSW 2145, Australia [A. C. C., R. R. R.]; and Eos Biotechnology, South San Francisco, California 94080 [D. W., R. G.]

Abstract believed to include genetic or epigenetic aberrations occurring at the level of ER signaling. Differences in gene expression are likely to explain the phenotypic Recent cDNA microarray analyses have identified widespread dif- variation between hormone-responsive and hormone-unresponsive breast ferences in gene expression between ER-positive and ER-negative cancers. In this study, DNA microarray analysis of ϳ10,000 known genes and 25,000 expressed sequence tag clusters was performed to identify breast cancer clinical samples. Specifically, in two independent stud- genes induced by estrogen and repressed by the pure antiestrogen ICI 182 ies representing a total of 83 breast carcinomas, the clustering of 780 in vitro that correlated with estrogen receptor (ER) expression in global gene expression patterns divided tumors into two major groups primary breast carcinomas in vivo. Stanniocalcin (STC) 2 was identified as distinguished by ER status (3, 4). As yet, no systematic analysis has one of the genes that fulfilled these criteria. DNA microarray hybridiza- been described to unify the observed changes and determine which tion showed a 3-fold induction of STC2 mRNA expression in MCF-7 cells genes represent novel estrogen-responsive genes. We undertook a < in 3 h of estrogen exposure and a 3-fold repression in the presence of comprehensive analysis of genes responsive to estrogen and the pure antiestrogen (one-way ANOVA, P < 0.0005). In 13 ER-positive and 12 antiestrogen ICI 182 780 in the well-studied, ER-positive breast ER-negative breast carcinomas, the microarray-derived mRNA levels observed for STC2 correlated with tumor ER mRNA (Pearson’s correla- cancer cell line model, MCF-7 (5), using a high-density Affymetrix ϳ P < 0.0001) and ER protein status (Spearman’s rank array capable of measuring 35,000 genes and expressed sequence ;0.85 ؍ tion, r P < 0.0001). The expression profile of STC2 was tag clusters. To further pinpoint estrogen target genes with potential ;0.73 ؍ correlation, r further confirmed by in situ hybridization and immunohistochemistry on clinical relevance, these in vitro data were combined with the expres- a larger cohort of 236 unselected breast carcinomas using tissue microar- sion profiles of a cohort of 13 ER-positive and 12 ER-negative rays. STC2 mRNA and protein expression were found to be associated primary breast carcinomas, and the results were evaluated on an with tumor ER status (Fisher’s exact test, P < 0.005). The related gene, independent panel of 236 clinical breast cancer samples. Using this STC1, was also examined and shown to be associated with ER status in approach, STC2, a homologue of a glycoprotein hormone originally breast carcinomas (Fisher’s exact test, P < 0.05). This study demonstrates the feasibility of using global gene expression data derived from an in vitro found to regulate calcium/phosphate homeostasis in bony fish (6), was model to pinpoint novel estrogen-responsive genes of potential clinical identified as an estrogen-responsive gene that was also differentially relevance. expressed between ER-positive and ER-negative breast carcinomas.

Introduction Materials and Methods

3 The amount of ER protein in breast tumors is frequently used to Human Breast Cell Lines. MCF-7 cells obtained from the American Type group breast cancer patients in a clinical setting, both as a prognostic 4 Culture Collection were cultured at 37°C in 5% CO2 in RPMI 1640 supple- indicator and in predicting the likelihood of response to treatment with mented with 10% FCS (Sigma Chemical Co., Castle Hill, New South Wales, antiestrogens, such as tamoxifen (1). Additional patient information is Australia) to 25% confluence. The cells were then washed three times with gained from tumor levels of PR protein, as the gene for the PR is PBS to remove residual serum and grown for 24 h in phenol red-free RPMI up-regulated in response to estrogen and, thus, used as a surrogate 1640 supplemented with 10% charcoal-stripped FCS (Sigma Chemical Co.). marker of ER activity (2). A current difficulty with the hormonal Cells were then treated with 100 nM 17␤ estradiol or vehicle (ethanol) for management of breast cancer patients is the diversity observed in the periods of 30 min to 48 h before mRNA was harvested. In parallel, for the clinic beyond that predicted by the measurement of steroid receptor antiestrogen experiments, cells were grown to 25% confluence in complete protein levels in tumor tissue. The molecular basis of the differences RPMI 1640 supplemented with 10% FCS. Cells were then washed three times with PBS and serum starved for another 24 h in phenol red-free RPMI. Serum in clinical behavior observed between ER-positive and ER-negative was then readded for another 24 h in phenol red-free RPMI 1640, and cells breast carcinomas and the failure to respond to antiestrogen therapy is were treated with 50 nM of the antiestrogen ICI 182 780 (AstraZeneca Phar- maceutical LP, Wilmington, DE) or vehicle (ethanol) for periods between 30 Received 10/2/01; accepted 1/18/02. min and 48 h. Cells were lysed, and mRNA was isolated using TRIzol reagent The costs of publication of this article were defrayed in part by the payment of page (Life Technologies, Inc.) as specified by the manufacturer. All experiments charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. were replicated independently three times. In parallel to mRNA extractions, 1 Supported by the University of Melbourne, the Peter MacCallum Cancer Institute, the ability of estrogen to induce cell proliferation and ICI 182 780 to arrest cell and the Victorian Breast Cancer Research Consortium. growth was verified by fluorescence-activated cell sorting analysis of pro- 2 To whom requests for reprints should be addressed, at Cancer Functional Genomics Unit, Murdoch Children’s Research Institute, 10th Floor, Royal Children’s Hospital, pidium iodide-stained cells. mRNA was also extracted from five actively Parkville, Victoria 3052, Australia. Phone: 61-3-8341-6231; Fax: 61-3-9348-1391; E- growing (in the presence of serum) breast cancer cell lines, two ER positive mail: [email protected]. (MCF-7 and BT-474) and three ER negative (MDA-MB-231/453/435). 3 The abbreviations used are: ER, estrogen receptor; ERE, estrogen-response element; HRE, hormone response element; PR, progesterone receptor; DIG, digoxigenin; STC, stanniocalcin. 4 Internet address: http://www.atcc.org. 1289

Tumor Specimens. mRNA was extracted from a total of 25 archival fresh of SSC (ϫ2, ϫ1, and ϫ0.1) for 15 min. To digest, unbound probe sections frozen breast carcinomas obtained from patients undergoing surgery at the were treated with RNase A (final concentration: 20 ␮g/ml) for 1 h at 37°C, Peter MacCallum Cancer Institute. A pathologist grossly dissected histologi- followed by two washes in PBS. Slides were then incubated in blocking cally verified tumor tissue from normal adjacent tissue. All mRNA was solution [Roche Blocking Reagent 1% (w/v) in buffer 1, 100 mM maleic acid, extracted by the standard guanidinium thiocyanate methodology. The ER and 150 mM NaCl (pH 7.5)] for 30 min. Blocking solution (100 ␮l) containing protein status of the tumors was determined clinically by immunohistochem- 0.75 units/ml Anti-DIG[Fab]-AP (Roche Diagnostics Australia Pty. Ltd.) was Ն istry. Tumors were deemed ER positive if 10% of cells stained at an intensity added to each section, coverslipped, and incubated for1hatroom temperature. of weak or above. Of the 25 tumors, 13 were classified as ER positive and 12 Probe localization was visualized by adding nitroblue tetrazolium/5-bromo-4- ER negative. Of the 13 ER-positive cases, 6 were weakly staining, 5 moder- chloro-3-indolyl phosphate (Roche Diagnostics Australia Pty. Ltd.) color rea- ately, and 2 strongly. Appropriate institutional approval was obtained for all gent, according to the manufacturer’s instructions. Sections were then rinsed in tumor material used in this study. water to stop color development, counterstained with 0.1% methyl-green, and Microarray Hybridization. Oligonucleotide arrays (Eos Biotechnology- mounted using Kaiser’s glycerol gelatin (Merck, Whitehouse Station, NJ). specified Affymetrix 43K GeneChip Set) composed of 10,000 human genes Scoring of in Situ Hybridization. Each arrayed tissue sample was probed and 25,000 human expressed sequence tag clusters were used for hybridization. The protocols used for poly(A)ϩ mRNA purification, cDNA synthesis, in with a sense and antisense probe for STC1 and STC2. Staining of normal vitro transcription, chip hybridization, and statistical analysis are described by human endometrium with STC1/STC2 probes served as a positive and internal Glynne et al. (7). negative tissue control. For each tumor array, two independent hybridization Tumor Specimens and Tissue Arrays. Archived formalin-fixed, paraffin- experiments were performed. Tumors were deemed to express STC1/STC2 embedded specimens of primary breast carcinoma were retrieved from files at mRNA if there was positive staining in the two independent experiments that the Department of Pathology at the Peter MacCallum Cancer Institute and the was greater than staining with the cognate sense strand probe and negative Department of Surgery at St. Vincent’s Hospital. Cases used from the Peter control probe. Positive staining ranged from moderate to strong staining in a MacCallum Institute represented women undergoing surgery from 1996 to small percentage of scattered tumor cells (5–10%), to Ն10% of tumor cells 1998 and from St. Vincent’s Hospital from 1991 to 1994. All specimens were staining at an intensity of weak to strong compared with the sense control. obtained surgically from patients and in Յ1 h fixed in 10% buffered formalin Immunohistochemistry. Polyclonal antibodies for STC1 (rabbit) and for 24 h and then embedded in paraffin wax in routine manner. Storage of STC2 (sheep) were generated as described previously (8). Staining was per- paraffin blocks was at room temperature. A total of 236 primary breast tumors formed on 3-␮m sections of formalin-fixed, paraffin wax-embedded tissue was used to construct tissue arrays, representing primary invasive breast rehydrated through graded alcohols. Antigen retrieval was used for the ER, PR, carcinoma. The Institutional Ethics Committees of the Peter MacCallum Can- STC1, and STC2 antibodies, which consisted of 2 min of heating under cer Institute and St Vincent’s Hospital approved the use of archival tumor pressure in a pressure cooker in 10 mM sodium citrate (pH 6.0). Sections were tissue for this study. stained using a DAKO Autostainer (DAKO Corp., Carpinteria, CA) using Probe Labeling by in Vitro Transcription of DNA with DIG. Linearized established protocols as described by Armes et al. (9). The primary antibody cDNA probe (1 ␮g) was in vitro transcribed and labeled with DIG, according was applied at the following dilutions in 10% FCS made up in 50 mM Tris-HCl to the manufacturer’s instructions (Roche Diagnostics Australia Pty. Ltd., (pH 7.6) and 0.05% Tween 20 for 30 min at room temperature: ER (1:200; Melbourne, Victoria, Australia). Labeled probe (5 ␮l) was electrophoresed on DAKO), PR (1:800; DAKO), pS2 (1:100; Novocastra Laboratories), STC1 a 1.5% agarose gel to check for riboprobe integrity. Serial dilutions were used (1:3000), and STC2 (1:50; diluted in background reducing diluent, instead of to estimate the concentration of labeled probe against DIG-labeled control FCS; DAKO). Negative controls were incubated in 10% FCS minus the mRNA (Roche Diagnostics Australia Pty. Ltd.), according to the manufacturer’s instructions. primary antibody. Biotinylated secondary antibodies were detected with ϩ Slide Preparation and Probe Hybridization. STC1 and STC2 mRNA streptavidin peroxidase by using the LSAB 2 kit (pS2) or the LSAB kit (ER, was localized by in situ hybridization. For each gene, separate tissue array- PR, STC1, and STC2; DAKO). The final color reaction was carried out using ϩ derived slides were probed with an in vitro-transcribed, DIG-labeled antisense aminoethylcarbazole as a chromogen (AEC ; DAKO) and counterstained and sense probe (negative control). In addition, a cytokeratin-19 probe was with hematoxylin. used to verify the integrity of tissue mRNA. For all of the antibodies, the intensity of staining and proportion of positive In situ hybridization was performed in RNase-free conditions on 5-␮m cells were determined for each case, according to a method described by sections of tissue arrays mounted on APES (3-aminopropyltriethoxysilane)- Armes et al. (9). Briefly, a semiquantitative estimate of expression levels of the coated slides and dried for2hat60°C. Sections were dewaxed and then placed antigen was based on the combined score for the proportion of staining cells into 0.2 M HCl, shaken at room temperature for 20 min, and then washed twice and the intensity of staining. The proportion score represented the estimated in diethyl pyrocarbonate-treated sterile water for 5 min with shaking at room percentage of positive cells as a fraction of tumor cells (0, Ͻ10%; 1, 11–25%; temperature. Sections were then digested with 1 ␮g/ml proteinase K in pre- 2, 26–50%; 3, 51–75%; 4, 76–90%; and 5, Ͼ91%.). The intensity score warmed digest buffer [100 mM Tris (pH 8.0) and 50 mM EDTA (pH 8.0)] for represented the average staining intensity for positive cells (0, none; 1, weak; 30 min at 37°C. The digest was stopped by washing slides in 0.2% glycine/ 2, moderate; and 3, strong). Levels of staining were derived as follows: PBS for 10 min at 4°C. Sections were then washed briefly in 0.1 M tri- samples with an intensity score of 0 or having Ͻ10% of cells staining were ethanolamine before adding acetic anhydride for 5 min to a final concentration designated negative; samples with intensity score of 1 in Ն10% of cells of 0.25%. were designated weak. For intensity levels 2 and 3, combined scores of 2–3 ␮ Prehybridization was performed by adding 100 l of prehybridization were designated as weak, 4– 6 as moderate, and 7 or 8 as strong expression. solution [50% deionized formamide, 24 mM Tris (pH 7.4), 1 M EDTA, 375 mM For analysis purposes, staining of all antibodies was considered positive if NaCl, 10% Dextran Sulfate, 1 ϫ Denharts Solution, 250 ␮g/ml yeast total Ն10% of cells stained unequivocally at an intensity of weak, moderate, or mRNA, 100 ␮g/ml single-stranded DNA, and 250 ␮g/ml tRNA] to each strong compared with the no primary antibody control. For STC1, STC2, section. Sections were then covered with a coverslip and incubated for2hat and pS2, an additional, reproducible staining pattern was also observed in 50°C (STC1 probe) or at 60°C (STC2 and CK19 probes). For the hybridization reaction, DIG-labeled probe was denatured for 5 min at 65°C and added to a subset of the cases. This consisted of moderate-to-strong unequivocal hybridization solution (same as prehybridization solution with 0.1% SDS, staining in scattered tumor cells (5–10%). For analysis purposes, this 0.1% sodium thiosulfate, and 0.1 mM DTT) to a final probe concentration of pattern was also considered indicative of positive staining for these anti- 200 ng/ml. This probe was thoroughly mixed with the hybridization solution gens. Results were obtained for two independent immunohistochemistry by vortexing for 1 min and then denatured for 15 min at 65°C. Hybridization experiments. solution (150 ␮l) with probe was then added to each slide and incubated at Statistics. Statistical comparisons between groups were assessed by stand- 50°C (STC1 probe) or at 60°C (STC2 and CK19 probes). ard contingency table analysis using two-tailed Fisher’s exact test. Data were Slide Washes and Probe Visualization. Slides were then washed for 15 analyzed using the program StatXact 4 (Cytel Software Corp., Cambridge, min in 2 ϫ SSC at room temperature, followed by decreasing concentrations MA). Pearson’s and Spearman’s rank correlation were performed using Graph- 1290

Fig. 1. A, hierarchical cluster diagram of gene expression data of the top 10 differentially expressed genes across 25 breast carcinomas (ER positive n ϭ 13 versus ER negative n ϭ 12, unpaired t test) that were also regulated by estrogen and or ICI 182 780 treatment in MCF-7 cells (ANOVA, P Յ 0.0002). Data are presented in a matrix format: row, a gene; column, a sample. The dendogram at the top indicates the degree of relatedness between the tumor samples by the height of the nodes. B, summary of in vitro ER activation/inhibition experiments. Kinetics of estrogen and ICI 182 780 transcriptional regulation of the same top 10 genes in MCF-7 cells over a 48-h time course. Additionally shown in the bottom panel are the kinetics of two known estrogen target genes, PR and pS2, for comparison. STC2 was identified as an estrogen-regulated and ICI 182 780-repressed gene that was coexpressed with ER mRNA and protein across 25 breast carcinomas. Pseudocolor scale represents absolute raw data values in A; B, the percentage change in expression after treatment of MCF-7 cells with estrogen or ICI 182 780.

Pad InStat version 3.00 for Windows 95 (GraphPad Software, San Diego CA).5 780 regulated in MCF-7 cells (although not the subject of further Nominal Ps are given without adjustment for multiple comparisons. A cutoff discussion in this study) is shown in Fig. 1. STC2 was differentially P of 0.05 was taken to indicate significance. An agglomerative hierarchical expressed between ER-positive and ER-negative tumors (ER status by clustering method was used to investigate the relationship among the mRNA immunohistochemistry, unpaired t test, P ϭ 0.001) and had mRNA expression patterns of the tumor samples. Data were imported into the Spotfire levels that correlated with ER mRNA levels (Pearson’s correlation, Decision Site 6.2 for Functional Genomics program and reordered in similarity n ϭ 25, r ϭ 0.85; P Ͻ 0.0001) and ER protein status (Spearman’s in expression patterns over the 25 samples using the agglomerative clustering ϭ ϭ Ͻ algorithm for hierarchical clustering. For the in vitro MCF-7 experiments, rank correlation, n 25, r 0.73; P 0.0001). STC2 mRNA ϳ expression profiles were analyzed by one-way ANOVA for statistical signif- expression in estrogen-stimulated MCF-7 cells showed a 3-fold icance specifying time, treatment (estrogen or vector; ICI 182 780 or vector), increase in expression Յ3 h of treatment and remained elevated at 24 experiment repeat, and treatment by time interaction as factors in the model. and 48 h (ANOVA, P ϭ 0.0003). After treatment of actively growing The probability that treatment with estrogen had no effect on the expression MCF-7 cells with the pure antiestrogen ICI 182 780, STC2 mRNA Ϫ level of STC2 in MCF7 cells was 3 ϫ 10 4; the probability that treatment with levels decreased 3-fold in Յ6 h (ANOVA, P ϭ 0.00005). Taken ICI 182 780 had no effect on the expression level of STC2 in MCF7 cells was together, these results suggest that ER signaling is sufficient to induce ϫ Ϫ5 5 10 . high levels of STC2 mRNA in this in vitro experiment and that this Results pathway involving ER and STC2 is active in breast cancers in vivo. Interestingly, the closely related family member STC1 was found by Identification of STC-2 As an in Vitro Estrogen-induced/ICI 182 microarray analysis to be 2–6-fold higher in 3 of the 13 ER-positive 780-repressed Gene Coexpressed with the ER in Breast Carci- tumors and also in the ER-positive cell line MCF-7, relative to the noma Samples by Microarray Analysis. Microarray hybridization average expression in 12 ER-negative tumors and 3 ER-negative cell analysis of total mRNA extracted from MCF-7 cells after estrogen/ICI lines (MDA-MB- 231/453/435), respectively, warranting further in- 182 780 treatment and from 25 primary breast cancer specimens vestigation in breast carcinomas in vivo. generated expression data on Յ35,000 genes in each of the samples. To investigate the potential regulatory mechanisms of estrogen- Through this analysis, 299 genes were found to be regulated by mediated STC2 mRNA induction, the 5Ј upstream genomic sequence estrogen and/or ICI 182 780 in MCF-7 cells, with a significance value of the human STC2 gene was examined. The human STC2 gene, cutoff of P Յ 0.0005. To focus on genes that were also associated located on chromosome 5, is contained within the working draft with ER expression in breast carcinomas, the expression profile of the genomic DNA contig NT_023132.6 (National Center for Biotechnol- 299 estrogen and/or ICI 182 780-responsive genes found in MCF-7 ogy Information). From this contig, 4 kb upstream of the translation cells were compared between 13 ER-positive and 12 ER-negative start site of the human STC2 gene were scanned by computer-assisted breast tumors. STC2 was identified as one of the genes that fulfilled homology searches using the publicly available PatSearch 1.1 pro- these criteria. The expression profile of the top 10 differentially gram to identify known consensus sites for eukaryotic transcription expressed genes (including STC2) between ER-positive and ER- factors (10). Three regions of the 5Ј upstream sequence of the STC2 negative tumors that were also found to be estrogen and/or ICI 182 gene localized between nucleotide positions Ϫ3349 to Ϫ3113, Ϫ1763 to Ϫ1525, and Ϫ198 to Ϫ21 were identified as potential estrogen- 5 Internet address: http://www.graphpad.com. responsive enhancer elements. The 5Ј upstream region is numbered 1291

Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2002 American Association for Cancer Research. STANNIOCALCIN 2 IN BREAST CANCER relative to the translation start codon ATG (ϩ1). The most proximal Table 1 Relationship between STC2 mRNA/protein expression and steroid hormone element (Ϫ198 to Ϫ21) contained a half site ERE at position Ϫ198 markers in human breast carcinomas (5Ј-GGTCA-3Ј) 20 nt upstream of five closely spaced, potential A. STC2 mRNA statusa,b Ϫ Ј Ј Ϫ Ј Sp1-binding sites at positions 173 (5 -GGGAGG-3 ), 129 (3 - Pc GGAGGG-5Ј), Ϫ99 (5Ј-GGGAGG-3Ј), Ϫ92 (3Ј-GGAGGG-5Ј), and Antigena,b Positive (%) Negative (%) Pos vs. Neg at Ϫ21 (5Ј-GGGAGG-3Ј). Alignment of this proximal site (starting at ER (n ϭ 190) Ϫ Ј Positive 50 (75) 48 (39) 198 in the human gene) with the 5 flanking region of the mouse Ͻ Ϫ Negative 17 (25) 75 (61) 0.0001 STC2 gene (starting with an ERE half site at 187) up until the start PR (n ϭ 191) codon revealed significant homology over the whole of the aligned Positive 49 (71) 50 (41) sequence (87% identity). Specifically, the half site ERE and four of Negative 20 (29) 72 (59) 0.0001 pS2 (n ϭ 167) the five downstream Sp1 sites were 100% conserved between the Positive 46 (72) 59 (57) mouse and human STC1 gene within this proximal element, adding Negative 18 (28) 44 (43) 0.08 weight to the potential transcriptional importance of this site. In the B. STC2 protein statusa,b human STC2 gene, the two more distal sites (Ϫ1763 to Ϫ1525) and Pc (Ϫ3349 to Ϫ3113) were also found to contain similar ER/Sp1 ele- Antigena,b Positive (%) Negative (%) Pos vs. Neg Ϫ Ϫ Ϫ ments. The 1763 to 1525 site contains an ERE half site at 1763 ER (n ϭ 161) (5Ј-TGACCT-3Ј) followed by five closely spaced Sp1 sites at Ϫ1610 Positive 65 (64) 23 (39) (5Ј-GGGCGGGG-3Ј), Ϫ1599 (5Ј-GGGGCGGGG-3Ј), Ϫ1545 (3Ј- Negative 37 (36) 36 (61) 0.003 PR (n ϭ 165) GGAGGG-5Ј), Ϫ1529 (5Ј-GGGAGG-3Ј), and at Ϫ1525 (5Ј- Positive 63 (59) 26 (44) GGGGCGAGT-3Ј). The most distal site (Ϫ3349 to Ϫ3113) contained Negative 43 (41) 33 (56) 0.07 Ϫ Ј Ј pS2 (n ϭ 159) an ERE half site at 3349 (5 -TGACC-3 ), followed by an imperfect Positive 77 (72) 21 (40) palindromic ERE at Ϫ3262 (5Ј-AGGTCAnnnCTGCCT-3Ј). This se- Negative 30 (28) 31 (60) 0.0002 quence differs from the consensus palindromic ERE (5Ј-AGGT- a Not all cases stained for each antibody/probe were scored because of a loss of tissue Ј from slide during staining procedure. CAnnnTGACCT-3 ) in three bases of one of the half sites and was b Ϫ Ј Ј Ϫ Proportion showing Ն10% of cells positive at an intensity of weak and above, in followed by three Sp1 sites at 3216 (3 -GGAGGG-5 ), 3130 addition to moderate/strong staining in scattered tumor cells (Ͻ10%) for STC1, STC2, (5Ј-GGGAGG-3Ј), and at Ϫ3113 (3Ј-GGAGGG-5Ј). Comparable 5Ј and pS2. c upstream regulatory sequences with closely spaced Sp1 sites, with or Comparisons between groups (STC2 positive versus STC2 negative) were assessed by standard contingency table analysis using two-tailed Fisher’s exact test. without, upstream imperfect palindromic EREs or ERE half sites have been identified as estrogen-responsive enhancer elements in an increasing number of ER target genes, including PR, cyclin D1, and of the ER (Fisher’s exact test, P Ͻ 0.0001) and PR (Fisher’s exact test, bcl-2 (11–13). In this model, the ER is thought to bind directly to the P ϭ 0.0001; Table 1). A trend was also observed between STC2 ERE half site, indirectly by interacting with proteins bound to the Sp1 mRNA status and pS2 (Fisher’s exact test, P ϭ 0.08). sites, or a combination of these two mechanisms. The human STC1 Antibodies to STC2 were used to compare mRNA expression with genomic sequence (4 kb) contained within the working draft genomic protein expression and evaluate the relationship of STC2 protein contig NT_008130.7 (National Center for Biotechnology Informa- expression with ER, PR, and pS2 status (Table 1). Consistent with tion) was also examined for eukaryotic transcription factor-binding STC2 mRNA expression, STC2 immunoreactivity was also observed sites. No potential palindromic EREs or clustering of ERE/Sp1- in breast tumor cells. Infiltrating lymphocytes and stromal cells were binding sites were identified. An imperfect HRE for the androgen/ negative. Staining was observed in the cytoplasm of cells, as expected glucocorticoid/progesterone/mineralocorticoid receptor subfamily for a secreted glycoprotein hormone. Again, the patterns of immuno- was identified at position Ϫ2621 upstream of the translation start site. histochemical staining observed for STC2 were similar to that ob- This sequence 5Ј-AGGACAnnnTGTTCT-3Ј differed form the com- served for STC2 mRNA by in situ hybridization. STC2 immunore- mon consensus HRE for this steroid receptor subfamily by a single activity ranged from intensely staining isolated cells to positive nucleotide in one of the half sites, 5Ј-AGAACAnnnTGTTCT-3Ј. staining in a majority of tumor cells. For tumors showing STC2 In Situ Analysis of STC2 mRNA and Protein Levels in Primary mRNA positivity, 83% also displayed protein expression by STC2 Breast Carcinomas and Relationship to ER, PR, and pS2. For antibody. The minority of discordant cases may reflect the numerous STC2 to be a useful clinical marker of ER activity, we reasoned that post-transcriptional mechanisms that regulate the stability and turn- STC2 mRNA should be transcribed in transformed epithelial cells and over of RNA and protein. For the STC2 mRNA-negative cases, about that expression of the message in ER-positive tumors should be half was observed to be immunoreactive for STC2 protein, perhaps accompanied by an increased frequency of STC2 protein expression reflecting increased sensitivity for protein detection over mRNA in ER-positive tumors. To answer these questions with a high level of detection. A significant association was observed between immuno- statistical confidence, we used tissue microarrays of 236 invasive histochemical staining for STC2 protein and positivity for ER (Fish- breast cancer samples from different patients (Table 1) and examined er’s exact test, P ϭ 0.003) and pS2 immunoreactivity (Fisher’s exact the relationship between STC2 mRNA and protein levels with the test, P ϭ 0.0002; Table 1). A trend was observed between STC2 and protein status of ER and its target genes PR and pS2. Staining for PR protein status, although this did not quite reach statistical signif- STC2 transcripts after hybridization of breast tumor sections with icance (Fisher’s exact test, P ϭ 0.07). digioxenin-labeled riboprobes revealed specific expression of STC2 On the basis of the microarray results of high STC1 mRNA levels in the tumor cells. The expression pattern for STC2 mRNA in tumors in a subset of the ER-positive tumors, in situ hybridization and was variable, ranging from: (a) negative to (b) scattered cells staining immunohistochemistry on tissue arrays were also performed for moderate to strong and to (c) the majority of tumor cells staining STC1. mRNA positivity for STC1 was identified in 82 of 169 (49%) strongly. A positive STC2 mRNA staining pattern was observed in 75 tumors. For STC1 mRNA expression status, a positive association of 216 (35%) tumors. When tumors were divided into two subgroups was observed with ER (Fisher’s exact test, P ϭ 0.0002) and PR according to their mRNA expression pattern for STC2 as either immunoreactivity (Fisher’s exact test, P ϭ 0.01; Table 2). The rela- positive or negative, an association was observed for protein staining tionship between mRNA and protein expression of both STC1 and 1292

Table 2 Relationship between STC1 mRNA/protein expression and steroid hormone nohistochemical staining for STC1 protein and positivity for ER markers in human breast carcinomas (Fisher’s exact test, P ϭ 0.01) and PR (Fisher’s exact test, P ϭ 0.007; A. STC1 mRNA statusa,b Tables 2). No association was observed between STC1 and pS2 ϭ Pc immunoreactivity (Fisher’s exact test, P 0.2) Antigena,b Positive (%) Negative (%) Pos vs. Neg Discussion ER (n ϭ 155) Positive 54 (73) 35 (43) Negative 20 (27) 46 (57) 0.0002 The genome-wide analysis of MCF-7 cells exposed to estrogen/ PR (n ϭ 158) antiestrogen and 25 breast carcinoma samples of different ER status Positive 50 (66) 37 (45) identified STC2 as an estrogen-regulated gene that was coexpressed Negative 26 (34) 45 (55) 0.01 pS2 (n ϭ 150) with ER mRNA in breast carcinomas. By in situ hybridization and Positive 50 (65) 45 (62) immunohistochemistry on 236 unrelated primary breast carcinomas, Negative 27 (35) 28 (38) 0.7 we show for the first time that both STC2 mRNA and protein B. STC1 protein statusa,b expression are associated with ER protein status in clinical breast Pc carcinoma specimens. Interestingly, the mRNA and protein expres- Antigena,b Positive (%) Negative (%) Pos vs. Neg sion of the related family member STC1 was also found to be ER (n ϭ 130) correlated with ER protein status in the breast cancer specimens. Positive 51 (63) 19 (39) Although STC1/2-positive tumors were more likely to coexpress ER, Negative 30 (37) 30 (61) 0.01 PR (n ϭ 131) not all ER-positive cases expressed STC1/2, suggesting that the ex- Positive 51 (64) 20 (39) pression of STC1/2 may represent a particular subclass of ER-positive Negative 29 (36) 31 (61) 0.007 pS2 (n ϭ 129) cases, e.g., a recent cDNA microarray analysis has indicated the Positive 54 (68) 27 (55) presence of at least two distinctive groups of ER-positive tumors, each Negative 26 (32) 22 (45) 0.2 with a different prognosis and a characteristic gene expression profile a Not all cases stained for each antibody/probe were scored because of a loss of tissue encompassing variable mRNA expression of the ER and ER target from slide during staining procedure. b Proportion showing Ն10% of cells positive at an intensity of weak and above, in genes (14). Hence, STC1/2 expression may have added value in addition to moderate/strong staining in scattered tumor cells (Ͻ10%) for STC1, STC2, dichotomizing ER-positive patients regarding ER pathway activity or and pS2. prognosis. c Comparisons between groups (STC1 positive versus STC1 negative) were assessed by standard contingency table analysis using two-tailed Fisher’s exact test. Further insight was provided by the examination of two ER target genes, PR and pS2. STC1 and STC2 mRNA levels were both correlated with PR protein expression. STC2 protein levels were also STC2 with ER status in breast carcinomas is shown in Fig. 2. No correlated positively with the expression of another ER target gene, association was observed between STC1 mRNA and pS2 protein pS2. Although STC2 mRNA levels did show a trend for a positive expression (Fisher’s exact test, P ϭ 0.7). For tumors showing STC1 association with pS2 positivity, this relationship was not statistically mRNA expression, 75% also displayed protein expression by the significant. This may relate to an increased sensitivity for detection of STC1 antibody. A positive association was observed between immu- STC2 protein over mRNA. Alternatively, tumor cells that are immu-

Fig. 2. Examples of STC1/STC2 mRNA and protein expression in ER-positive (tumor 1 and 2) and ER-negative (tumor 3) breast carcinoma sections. Staining in A, D, and G represent immunohistochemical detection of ER protein. Tumor 1, STC1 mRNA (B) and STC1 protein (C) staining. Tumor 2, STC2 mRNA (E) and protein (F) staining. Tumor 3, STC2 mRNA (H) and protein staining (I) and is also representative of STC1 expression in this tumor.

1293

Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2002 American Association for Cancer Research. STANNIOCALCIN 2 IN BREAST CANCER noreactive to STC2 may not necessarily transcribe the gene but may genes that may be of importance to clinical disease. STC2 was represent the site of action because of the active sequestration of the identified as a novel estrogen target gene, which when tested by in situ hormone. Indeed, others have observed this expression pattern for hybridization and immunohistochemistry in a larger cohort of breast STC1 mRNA versus protein in mammalian kidney and ovary (15, 16). carcinomas, correlated positively with ER and PR status, as did the In addition, STC1/2 protein levels may be high, although mRNA related STC1 gene. The coexpression of STC1 and STC2 with ER expression is nondetectable because of low-level transcription cou- protein in a subset of ER-positive breast carcinomas suggests that the pled to a low rate of protein turnover. two genes may play a role in the biology of some estrogen-responsive The estrogen responsiveness of STC2 in vitro is consistent with the tumors and that STC1/2 expression data may provide additional study by Charpentier et al. (17), who identified STC2 as a novel information from standard practice regarding ER pathway activity. To estrogen target gene in MCF-7 cells using Serial Analysis of Gene potentially exploit this information in the management of breast Expression. In addition, we identified three potential ER/Sp1-binding cancer, additional investigation is needed to address the physiological Ј elements in the 5 upstream region of the human STC2 gene, compa- function of STC genes in the mammary gland and determine how rable with the estrogen-responsive enhancer elements found in nu- STC1/2 expression patterns in breast carcinomas relate to patient merous other ER target genes (11–13). Functional studies are now response with antiestrogen treatment. needed to determine whether any of these three ER/Sp1 sites identified in the 5Ј genomic sequence of the human STC2 gene bind ER, Acknowledgments Sp1, or ER/Sp1 complexes and confer estrogen responsiveness. The Ј absence of such ER/Sp1 motifs in the 5 upstream region of the STC1 We thank Melanie Trivett for her help in obtaining antibodies and optimiz- gene was consistent with the lack of transcriptional regulation of ing immunohistochemical and in situ hybridization conditions, Elena Prov- STC1 mRNA by estrogen in MCF-7 cells. However, the presence of enzano for her cell culture expertise, Gino Somers and Andrew Holloway for an imperfect HRE for the androgen/glucocorticoid/progesterone/ help with the design of cell culture experiments, Katrina Bell for her bioin- mineralocorticoid receptor subfamily suggests that STC1 may be formatics expertise, and members of the J. E. A. and D. J. V. laboratories under regulation by alternate steroid receptors. This may explain the involved in the construction of tissue arrays. We also thank John Hopper and observed correlation between STC1 with PR and, hence, also to ER in Gareth Price for reading the manuscript and their helpful comments. the breast carcinomas studied. STC was originally identified as an antihypercalcemic hormone in References bony fish produced by the corpuscles of Stannius, a specialized endocrine 1. Osborne, C. K. Steroid hormone receptors in breast cancer management. Breast gland adjacent to the kidney (18). In fish, rising plasma calcium levels Cancer Res. Treat., 51: 227–238, 1998. stimulate STC synthesis and secretion (19). STC counteracts hypercal- 2. Bloom, N. D., Tobin, E. H., Schreibman, B., and Degenshein, G. A. The role of progesterone receptors in the management of advanced breast cancer. Cancer (Phila.), cemia by slowing calcium uptake in the gills, increasing phosphate 45: 2992–2997, 1980. reabsorption in the renal proximal tubules, and inhibiting intestinal cal- 3. Martin, K. J., Kritzman, B. M., Price, L. M., Koh, B., Kwan, C. P., Zhang, X., cium transport (20, 21). To date, most studies to delineate the role of Mackay, A., O’Hare, M. J., Kaelin, C. M., Mutter, G. L., Pardee, A. B., and Sager, R. Linking gene expression patterns to therapeutic groups in breast cancer. Cancer mammalian STC have focused on STC1. Consistent with its role in fish, Res, 60: 2232–2238, 2000. STC1 has been observed to regulate mineral homeostasis in mammals. 4. Perou, C. M., Sorlie, T., Eisen, M. B., van de Rijn, M., Jeffrey, S. S., Rees, C. A., Infusion of recombinant human STC1 into rats increases renal phosphate Pollack, J. R., Ross, D. T., Johnsen, H., Akslen, L. A., Fluge, O., Pergamenschikov, reabsorption (22). Furthermore, application of STC1 to the serosal sur- A., Williams, C., Zhu, S. X., Lonning, P. E., Borresen-Dale, A. L., Brown, P. O., and Botstein, D. Molecular portraits of human breast tumours. Nature (Lond.), 406: face of rat or pig duodenal mucosa decreases calcium absorption and 747–752, 2000. increases phosphate uptake (23). 5. Levenson, A. S., and Jordan, V. C. MCF-7: the first hormone-responsive breast Additional information concerning the role of mammalian STC1/2 cancer cell line. Cancer Res, 57: 3071–3078, 1997. 6. Chang, A. C., and Reddel, R. R. Identification of a second stanniocalcin cDNA in has come from cellular localization studies. Of particular interest are mouse and human: stanniocalcin 2. Mol. Cell. Endocrinol., 141: 95–99, 1998. expression studies performed in the mouse ovary during different 7. Glynne, R., Akkaraju, S., Healy, J. I., Rayner, J., Goodnow, C. C., and Mack, D. H. reproductive states. STC1 is highly expressed in mouse and human How self-tolerance and the immunosuppressive drug FK506 prevent B-cell mitogen- esis. Nature (Lond.), 403: 672–676, 2000. ovary, where it is produced in the theca-interstitial cells and thought 8. Jellinek, D. A., Chang, A. C., Larsen, M. R., Wang, X., Robinson, P. J., and Reddel, to act in a paracrine manner to regulate corpus luteum function (15). R. R. Stanniocalcin 1 and 2 are secreted as phosphoproteins from human fibrosarcoma In this recent study by Deol et al. (15), the expression of mouse STC1 cells. Biochem. J., 350: 453–461, 2000. 9. Armes, J. E., Trute, L., White, D., Southey, M. C., Hammet, F., Tesoriero, A., was shown to be dramatically up-regulated in the ovary during preg- Hutchins, A. M., Dite, G. S., McCredie, M. R., Giles, G. G., Hopper, J. L., and nancy and lactation. This increase in ovarian STC1 was accompanied Venter, D. J. Distinct molecular pathogeneses of early-onset breast cancers in BRCA1 by the detection of STC1 immunoreactivity in the mouse serum, and BRCA2 mutation carriers: a population-based study. Cancer Res, 59: 2011–2017, 1999. suggesting a role for ovarian-derived STC1 as an endocrine regulator 10. Heinemeyer, T., Wingender, E., Reuter, I., Hermjakob, H., Kel, A. E., Kel, O. V., of mammary gland morphogenesis or milk production (15). Taken Ignatieva, E. V., Ananko, E. A., Podkolodnaya, O. A., Kolpakov, F. A., Podkolodny, together with our finding of the estrogenic regulation of STC2 ex- N. L., and Kolchanov, N. A. Databases on transcriptional regulation: TRANSFAC, TRRD and COMPEL. Nucleic Acids Res., 26: 362–367, 1998. pression in breast cancer cells, the published data suggest that STC 11. Petz, L. N., and Nardulli, A. M. Sp1 binding sites and an estrogen response element genes may also be subject to dynamic hormonal regulation in the half-site are involved in regulation of the human progesterone receptor A promoter. breast. Additionally of interest, in light of STC’s well-documented Mol. Endocrinol., 14: 972–985, 2000. role in Ca2ϩ regulation, are the findings that other calcium-mobilizing 12. Castro-Rivera, E., Samudio, I., and Safe, S. Estrogen regulation of cyclin D1 gene expression in ZR-75 breast cancer cells involves multiple enhancer elements. J. Biol. proteins, including parathyroid hormone-related protein and also os- Chem, 276: 30853–30861, 2001. teoprotogenin ligand (also known as RANKL), play a role in breast 13. Dong, L., Wang, W., Wang, F., Stoner, M., Reed, J. C., Harigai, M., Samudio, I., physiology, the organ required for transmission of maternal calcium to Kladde, M. P., Vyhlidal, C., and Safe, S. Mechanisms of transcriptional activation of bcl-2 gene expression by 17␤-estradiol in breast cancer cells. J. Biol. Chem., 274: neonates in mammals (24, 25). Knockout studies of STC genes are 32099–32107, 1999. now needed to more precisely delineate their role in mammary gland 14. Sorlie, T., Perou, C. M., Tibshirani, R., Aas, T., Geisler, S., Johnsen, H., Hastie, T., physiology. Eisen, M. B., van de Rijn, M., Jeffrey, S. S., Thorsen, T., Quist, H., Matese, J. C., Brown, P. O., Botstein, D., Eystein Lonning, P., and Borresen-Dale, A. L. Gene In conclusion, global transcript analysis of an experimental breast expression patterns of breast carcinomas distinguish tumor subclasses with clinical cancer cell line model of ER activation and inhibition can predict implications. Proc. Natl. Acad. Sci. USA, 98: 10869–10874, 2001. 1294

15. Deol, H. K., Varghese, R., Wagner, G. F., and Dimattia, G. E. Dynamic regulation of 21. Sundell, K., Bjornsson, B. T., Itoh, H., and Kawauchi, H. Chum salmon (Oncorhyn- mouse ovarian stanniocalcin expression during gestation and lactation. Endocrinol- chus keta) stanniocalcin inhibits in vitro intestinal calcium uptake in Atlantic cod ogy, 141: 3412–3421, 2000. (Gadus morhua). J. Comp. Physiol. [B], 162: 489–495, 1992. 16. Stasko, S. E., and Wagner, G. F. Stanniocalcin gene expression during mouse 22. Wagner, G. F., Vozzolo, B. L., Jaworski, E., Haddad, M., Kline, R. L., Olsen, H. S., urogenital development: a possible role in mesenchymal-epithelial signaling. Dev. Rosen, C. A., Davidson, M. B., and Renfro, J. L. Human stanniocalcin inhibits renal Dyn., 220: 49–59, 2001. phosphate excretion in the rat. J. Bone Miner. Res., 12: 165–171, 1997. 17. Charpentier, A. H., Bednarek, A. K., Daniel, R. L., Hawkins, K. A., Laflin, K. J., 23. Madsen, K. L., Tavernini, M. M., Yachimec, C., Mendrick, D. L., Alfonso, P. J., Gaddis, S., MacLeod, M. C., and Aldaz, C. M. Effects of estrogen on global gene Buergin, M., Olsen, H. S., Antonaccio, M. J., Thomson, A. B., and Fedorak, R. N. expression: identification of novel targets of estrogen action. Cancer Res., 60: Stanniocalcin: a novel protein regulating calcium and phosphate transport across 5977–5983, 2000. mammalian intestine. Am. J. Physiol., 274: G96–G102, 1998. 18. Wagner, G. F., Hampong, M., Park, C. M., and Copp, D. H. Purification, character- ization, and bioassay of teleocalcin, a glycoprotein from salmon corpuscles of 24. Wysolmerski, J. J., McCaughern-Carucci, J. F., Daifotis, A. G., Broadus, A. E., and Stannius. Gen. Comp. Endocrinol., 63: 481–491, 1986. Philbrick, W. M. Overexpression of parathyroid hormone-related protein or parathy- 19. Wagner, G. F., Gellersen, B., and Friesen, H. G. Primary culture of teleocalcin cells roid hormone in transgenic mice impairs branching morphogenesis during mammary from rainbow trout corpuscles of Stannius: regulation of teleocalcin secretion by gland development. Development, 121: 3539–3547, 1995. calcium. Mol. Cell. Endocrinol., 62: 31–39, 1989. 25. Fata, J. E., Kong, Y. Y., Li, J., Sasaki, T., Irie-Sasaki, J., Moorehead, R. A., Elliott, 20. Lu, M., Wagner, G. F., and Renfro, J. L. Stanniocalcin stimulates phosphate reab- R., Scully, S., Voura, E. B., Lacey, D. L., Boyle, W. J., Khokha, R., and Penninger, sorption by flounder renal proximal tubule in primary culture. Am. J. Physiol., 267: J. M. The osteoclast differentiation factor osteoprotegerin-ligand is essential for R1356–R1362, 1994. mammary gland development. Cell, 103: 41–50, 2000.

1295

Downloaded from cancerres.aacrjournals.org on September 26, 2021. © 2002 American Association for Cancer Research. Stanniocalcin 2 Is an Estrogen-responsive Gene Coexpressed with the Estrogen Receptor in Human Breast Cancer

Toula Bouras, Melissa C. Southey, Andy C. Chang, et al.

Cancer Res 2002;62:1289-1295.

Updated version Access the most recent version of this article at: http://cancerres.aacrjournals.org/content/62/5/1289

Cited articles This article cites 24 articles, 8 of which you can access for free at: http://cancerres.aacrjournals.org/content/62/5/1289.full#ref-list-1

Citing articles This article has been cited by 20 HighWire-hosted articles. Access the articles at: http://cancerres.aacrjournals.org/content/62/5/1289.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://cancerres.aacrjournals.org/content/62/5/1289. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.