Molecular Cancer Therapeutics CTRC-AACR San Antonio Breast Cancer Symposium
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
Cancer Research Clinical Cancer Research
Cancer Epidemiology Biomarkers & Prevention Molecular Cancer Therapeutics
Molecular Cancer Research Cancer Prevention Research
Cancer Prevention Journals Portal Cancer Reviews Online
Annual Meeting Education Book Meeting Abstracts Online
This Article
Right arrow Abstract Freely available
Right arrow Full Text (PDF)
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Becker, M.
Right arrow Articles by Fichtner, I.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Becker, M.
Right arrow Articles by Fichtner, I.
Mol Cancer Ther. 2005;4:151-170
© 2005 American Association for Cancer Research

Distinct gene expression patterns in a tamoxifen-sensitive human mammary carcinoma xenograft and its tamoxifen-resistant subline MaCa 3366/TAM

Michael Becker1, Anette Sommer2, Jörn R. Krätzschmar2, Henrik Seidel2, Hans-Dieter Pohlenz2 and Iduna Fichtner1

1 Max-Delbrueck-Center for Molecular Medicine, Experimental Pharmacology, and 2 Research Laboratories, Schering AG, Berlin, Germany

Requests for reprints: Michael Becker, Max-Delbrueck-Center for Molecular Medicine, Robert-Rössle-Strasse 10 D-13092 Berlin, Germany. Phone: 49-30-9406-2702; Fax: 49-30-9406-3823. E-mail: mibecker{at}mdc-berlin.de


   Abstract
Top
 Abstract
Introduction
 Materials and Methods
 Results
 Discussion
 References
 
The reasons why human mammary tumors become resistant to tamoxifen therapy are mainly unknown. Changes in gene expression may occur as cells acquire resistance to antiestrogens. We therefore undertook a comparative gene expression analysis of tamoxifen-sensitive and tamoxifen-resistant human breast cancer in vivo models using Affymetrix oligonucleotide arrays to analyze differential gene expression. Total RNAs from the tamoxifen-sensitive patient-derived mammary carcinoma xenograft MaCa 3366 and the tamoxifen-resistant model MaCa 3366/TAM were hybridized to Affymetrix HuGeneFL and to Hu95Av2 arrays. Pairwise comparisons and clustering algorithms were applied to identify differentially expressed genes and patterns of gene expression. As revealed by cluster analysis, the tamoxifen-sensitive and the tamoxifen-resistant breast carcinomas differed regarding their gene expression pattern. More than 100 transcripts are changed in abundance in MaCa 3366/TAM as compared with MaCa 3366. Among the genes that are differentially expressed in the tamoxifen-resistant tumors, there are several IFN-inducible and estrogen-responsive genes, and genes known to be involved in breast carcinogenesis. The genes neuronatin (NNAT) and bone marrow stem cell antigen 2 (BST2) were sharply up-regulated in MaCa 3366/TAM. The differential expression of four genes (NNAT, BST2, IGFBP5, and BCAS1) was confirmed by Taqman PCR. Our results provide the starting point for deriving markers for tamoxifen resistance by differential gene expression profiling in a human breast cancer model of acquired tamoxifen resistance. Finally, genes whose expression profiles are distinctly changed between the two xenograft lines will be further evaluated as potential targets for diagnostic or therapeutic approaches of tamoxifen-resistant breast cancer.

Key Words: tamoxifen • breast cancer • endocrine resistance • gene expression profiling • interferon-inducible gene


   Introduction
Top
 Abstract
 Introduction
Materials and Methods
 Results
 Discussion
 References
 
Breast cancer is the most common type of cancer in women of the Western world. Due to advances in early detection and treatment, breast cancer survival rates have increased markedly over the past decades. After surgery, estrogen receptor {alpha} (ER{alpha})–positive breast cancer is usually treated with endocrine therapy. Tamoxifen, a nonsteroidal antiestrogen, also termed selective estrogen receptor modulator, is the first-line therapy for premenopausal and, until recently, also for postmenopausal hormone receptor–positive women (1). For postmenopausal women, three different aromatase inhibitors are now available that might replace tamoxifen as first-line therapy in the future. Tamoxifen is effective both as adjuvant therapy and for advanced disease of hormone-responsive breast cancer and can prevent breast cancer in high-risk patients (2). In many cases, however, therapies fail and women die from recurrent, endocrine-resistant breast cancer. Prognosis of hormone-dependent breast cancer as well as treatment strategies are mainly determined by the presence of the ER{alpha} and the progesterone receptor (PR). Two thirds of the patients who present with breast cancer are ER{alpha} positive (3). Treatment of estrogen-dependent breast cancer with an antiestrogen like tamoxifen inhibits tumor growth. ER{alpha}- and PR-positive breast cancer have a better response rate than ER{alpha}- and PR-negative breast cancers (3). However, a large number of originally tamoxifen-sensitive tumors develop resistance after several months of treatment while still expressing the ER{alpha} (4).

Antagonism of tamoxifen has been attributed to the antiestrogenic activity of tamoxifen in which the active metabolite of tamoxifen, 4-hydroxytamoxifen (4-OHT), competes with E2 for binding to the ER{alpha}. Activation of the transcriptional activation domain AF-2 of the ER{alpha}, but not AF-1, is prevented by 4-OHT because coactivators can no longer bind to the antagonist-occupied ER{alpha} (5). Clinical manifestation of tamoxifen resistance is now often interpreted as a manifestation of increased tamoxifen agonism and as a switch from tamoxifen-dependent growth inhibition to growth stimulation (6, 7). The phenomenon of tamoxifen resistance is poorly understood and genetic mechanisms have been proposed, but mutations in the ER{alpha} are rare events in both patients with tamoxifen-resistant breast cancer and in various cell culture models of tamoxifen resistance (8, 9). In the MaCa 3366/TAM xenograft model, mutations in the ER{alpha} ligand binding domain were not identified (10, 11). Tamoxifen resistance most likely is a multicausal phenomenon. A disturbance of growth and survival pathways namely of growth factors, their receptors, extracellular proteins, proteases like kallikrein 10, immediate-early genes, transcription factors, cell cycle regulators, signal transduction molecules like BCAR1/p130Cas, phosphorylation of the ER{alpha} by protein kinase A, and alterations in the uptake, retention, and metabolism of tamoxifen might all contribute to tamoxifen resistance (12–18). Clues to the mechanisms of tamoxifen resistance could therefore be gained from an understanding of the numerous effects that tamoxifen produces at the gene expression level.

We decided to approach the question of tamoxifen resistance in a xenograft system. Xenografts of human tumors resemble the clinical situation much more closely than cell lines do (11). The xenograft tumor line MaCa 3366/TAM is one of the very few in vivo preclinical models in which antiestrogen resistance was induced in a clinically adapted manner. By direct transplantation of a ductal invasive carcinoma with moderate differentiation from a postmenopausal woman onto nude mice, the xenograft tumor line MaCa 3366 was established (19). To study tamoxifen resistance in an in vivo model, the tamoxifen-resistant xenograft tumor line MaCa 3366/TAM was developed by treatment of the tamoxifen-sensitive parental human xenograft tumor MaCa 3366 with the antiestrogen tamoxifen during successive passaging over 2 years (10)(11). MaCa 3366 and MaCa 3366/TAM are both ER{alpha} and PR positive. In both xenograft tumor lines, the PR is inducible by E2 indicating that the ER{alpha}-dependent transcriptional regulation is still intact.

In recent years, several gene expression profiling studies were performed to identify genes that are differentially expressed in human breast cancer which then allow to classify tumors and to predict outcome (20–27). In in vitro breast cancer cell culture models, the influence of estrogen or antiestrogen treatment was also analyzed (28–37).

It was therefore our aim to compare the gene expression profile of the parental tamoxifen-sensitive mammary carcinoma xenograft model MaCa 3366 and its tamoxifen-resistant derivative MaCa 3366/TAM on a genome-wide scale to analyze the molecular events that lead to the tamoxifen-resistant phenotype in an in vivo system. By comparing the gene expression pattern of MaCa 3366, untreated or after short-term tamoxifen treatment, and MaCa 3366/TAM, either untreated, after short-term or permanent tamoxifen treatment, we expected to identify genes which respond to short-term tamoxifen treatment and those which distinguish the tamoxifen-sensitive from the tamoxifen-resistant phenotype. Different types of gene expression profiles can be envisioned which might either indicate an involvement of a particular gene in the biological process and the mechanism of tamoxifen resistance or which can potentially be exploited as marker genes of tamoxifen resistance in a diagnostic or prognostic approach. By using clustering algorithms as well as pairwise comparisons of sample groups, we focussed on those genes, which distinguish the tamoxifen-sensitive from the tamoxifen-resistant phenotype.


   Materials and Methods
Top
 Abstract
 Introduction
 Materials and Methods
Results
 Discussion
 References
 
Animal Experiment
Animals. For the animal experiments, 50 female nude mice (Bom: NMRI-nu/nu) per xenograft experiment, ages 4 to 6 weeks and weighing 20 to 24 g, were used. Breeding and keeping conditions have been described (38). All animal experiments were done according to the United Kingdom Coordinating Committee on Cancer Research Guidelines for the Welfare of Animals in Experimental Neoplasia and with the permission of the responsible local authorities (G V247/98).

Tumor Transplantation. The s.c. transplantation of the tumor fragments (size, 4 x 4 x 4 mm3) was done under Radenarkon anesthesia (40 mg/kg i.p. Etomidat, Asta Medica, Frankfurt, Germany). The diameter of the tumors was measured once weekly using a caliper-like mechanical instrument and the tumor volume (V) was calculated according to the empirical equation V = (length x width2)/2. The median volumes of each group were normalized to the initial tumor volume resulting in the relative tumor volume. In all the experiments, tumor-bearing mice received estradiol supplementation [estradiol valeriate (E2D), 0.5 mg/kg once/wk i.m.]. This supplementation leads to physiologic levels of serum E2 (25-984 pg/mL) that are comparable to the human situation (25–600 pg/mL depending on the follicular phase).

Substances. The following substances were used: E2D (Jenapharm, Jena, Germany) and tamoxifen (Sigma, Chemie GmbH, Taufkirchen, Germany).

Treatment Modalities. Two independent experiments were done: All MaCa 3366 and MaCa 3366/TAM transplanted animals received E2D (0.5 mg/kg) injections once a week. The last passage of MaCa 3366/TAM before the start of an experiment was cultivated without tamoxifen treatment. MaCa 3366 and MaCa 3366/TAM were either left untreated, or treated with 50 mg/kg tamoxifen 24 hours before sacrifice (short-term tamoxifen treatment). An additional group of MaCa 3366/TAM xenograft animals was treated once a week with 50 mg/kg tamoxifen during the course of the experiment starting on day 1 (permanent tamoxifen treatment; Fig. 1). As permanent tamoxifen treatment of MaCa 3366 led to complete remission of tumors, this group could not be included in the analysis. From histologic and fluorescence-activated cell sorting analyses of MaCa 3366 and MaCa 3366/TAM, the amount of murine tissue in a human breast cancer xenograft tumor was estimated to be as low as 5% to 10%.3



View larger version (25K):
[in this window]
[in a new window]
  		Figure 1. Experimental workflow. Five nude mice each were transplanted in the flank with the human breast cancer xenograft model MaCa 3366 and MaCa 3366/TAM and all animals were supplemented with E2D (0.5 mg/kg) injections once a week. MaCa 3366 and MaCa 3366/TAM were either left untreated, or treated with 50 mg/kg tamoxifen (TAM) 24 hours before sacrifice (short-term TAM treatment), and MaCa 3366/TAM were treated once per week with 50 mg/kg TAM in the course of the experiment (permanent TAM treatment). Two independent xenograft experiments were conducted. RNA was isolated from the tumors and RNA from the five animals per treatment group was pooled. The cRNA from the first xenograft experiment was hybridized to Affymetrix HuGeneFL arrays, the cRNA from the second experiment to Hu95Av2 arrays. Both data sets were subjected to supervised and unsupervised analyses that were done with the Expressionist software (Genedata).

 
Extraction of RNA
Tumor tissue samples (size, 2 x 2 x 2 mm3) were taken from the sacrificed animals. Samples were snap frozen and stored in liquid nitrogen until use. Total RNA of tumor samples was prepared using the RNeasy Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer's recommendations. A DNase I (Qiagen) digestion step was included to eliminate DNA contamination.

Affymetrix DNA Chip Hybridization
Isolated total RNA was checked for integrity and concentration using the RNA 6000 Assay and RNA LabChips on the Agilent Bioanalyzer 2100 (Agilent Technologies, Inc., Palo Alto, CA). The cRNA was prepared as described above (39). The HuGeneFL and Hu95Av2 arrays (Affymetrix, Santa Clara, CA) were hybridized, washed, and stained with streptavidin-phycoerythrin (Moleculer Probes, Eugene, OR), biotinylated anti-streptavidin (Vector Laboratories, Burlingame, CA) and streptavidin-phycoerythrin according to the standard antibody amplification protocol for eukaryotic targets. The arrays were scanned with a confocal scanner from Affymetrix at 570 nm.

Data Analysis
The Affymetrix arrays were normalized by employing the Maximum Order Preserving Set method. This method identifies a subset of genes which does not change its expression between two experiments. The intensity scatter plot for all genes from this restricted set is approximated by a nonlinear regression function, which is then used for normalization as described above (39). The normalized expression data sets were loaded into the CoBi database (Genedata, Basel, Switzerland) and analyzed with the Genedata Expressionist software version 5.0 (40, 41). The expression data were filtered with the coefficient of variation and on the variant data set three different fold change–based two-group tests (intergroup gap, valid value proportions, and ratio of means) were done. A ratio of means is the ratio of the mean expression values for two sample groups under investigation. A group mean is calculated as the exponentialized arithmetic mean of all logarithmized expression values within a group. This is identical to the geometric mean of all expression values. For the valid value proportions analysis, all possible pairwise combinations of a sample from the first sample group and another sample from the second sample group are created. If the first group contained m samples and the second group contained n samples, then there were m x n such sample pairs. Subsequently, for each sample pair, the ratio of their corresponding expression values is calculated and compared with a threshold which is an analysis variable. The percentage of sample pairs which pass this threshold is determined. Finally, this type of analysis returns all probe sets for which the proportion of sample pairs with above-threshold expression ratios exceeds a given percentage which is another analysis variable. The intergroup gap method identifies probe sets which have consistently higher expression values in one of the two sample groups. If expression values in the first sample group are smaller than in the second sample group, then the intergroup gap is the ratio between the smallest value of the second group and the largest value of the first group. The intergroup gap method then returns all probe sets for which this ratio is above a given threshold. Probe sets with overlapping ranges of expression values are never returned by this type of analysis. Statistical tests could not be applied because only two or three samples were present in each sample group.

Probe Sets. Profiling human genes expressed in a human breast cancer–derived xenograft model transplanted onto nude mice using the human-specific Affymetrix arrays HuGeneFL and Hu95Av2, will mostly uncover genes specifically expressed in human cancer cells in xenograft tumors that might comprise a mixture of human- and mouse-derived cells. Using the human-specific Affymetrix arrays in contrast to cDNA arrays has the advantage that 20independent probes as perfect match oligonucleotides interrogate a particular gene. Unless a murine gene is 100% homologous to a human gene, this provides very high specificity compared with a continuous stretch of DNA on a cDNA array. The probe sets are usually situated either at the extreme 3' end of the cDNA or in the 3' untranslated region of the human genes. In the latter case, there is a very low probability that a human-specific probe set detects a murine gene because the homology between human and murine sequence is usually very low in the 3' untranslated region. In addition, the presence of a set of 20 mismatch oligonucleotides provides a further degree of specificity as only those probe sets are incorporated in the determination of a condensed signal intensity for which the signal intensity on the perfect match is larger than on the mismatch oligonucleotide. The probe sets on the two array generations HuGeneFL and Hu95Av2 were changed by Affymetrix such that different probe sets are present on both arrays. If genes are interrogated on both types of arrays by independent probe sets or if more than one probe set interrogates a particular gene on one array then the differential expression of human genes is further corroborated.

Unsupervised Analysis. For the unsupervised analysis, genes with invariant expression were filtered out by the coefficient of variation analysis. HuGeneFL data were filtered with a coefficient of variation (CV) ≥ 0.03175 resulting in a selection of 503 genes. For the analysis of Hu95Av2 data, a CV ≥ 0.04050 selected 538 genes. The CV was chosen so that a manageable number of probe sets was obtained. k-Means clustering was applied and genes were sorted into 10 different clusters.

Supervised Analysis. To select for variant genes a CV ≥ 0.001091 (CV ≥ 0.001097 for HuGeneFL array) was applied to the data derived from the Hu95Av2 array over all hybridization samples and 4,887 (2,305 for HuGeneFL array) probe sets were selected from a total of 12386 probe sets (6,574 for HuGeneFL array). Next, three different fold change–based two-group tests [intergroup gap >1.5-fold (1.3-fold), valid value proportions 40% (40%), ratio of means 1.8-fold (1.6-fold)] were used to identify differentially expressed probe sets between MaCa 3366/TAM (untreated, short-term, and permanent tamoxifen treatment) and MaCa 3366 (untreated, short-term tamoxifen treatment). [Values in brackets for HuGeneFL array.] Those 154 [50] probe sets overlapping (intersection) and those 362 [140] probe sets identified with any of the three independently conducted, fold change–based tests were further subjected to a k-means clustering resulting in two groups consisting of 57 [37] and 144 [125] probe sets up-regulated and 97 [13] and 218 [67] down-regulated, respectively, in all MaCa 3366/TAM compared with all MaCa 3366 tumors. To compare differentially expressed genes on the two array formats, the 144 and 125 up-regulated and 218 and 67 down-regulated probe sets from the Hu95Av2 and HuGeneFL array, respectively, were subjected to manual inspection combined with a BLAST analysis of the genes corresponding to the probe set identifiers. Differential expression of several genes was detected with more than one probe set. Nonredundant genes that are regulated and present on both array types and genes present on only one array, but strongly regulated, are displayed.

Quantitative Real-time PCR
Six genes (BST2, NNAT, IGFBP5, ESRRG, PTK6, and BCAS1) were selected and analyzed by quantitative real-time reverse transcription-PCR based on TaqMan-chemistry (Table 1). Hydroxymethylbilane synthase (HMBS) was used as housekeeping gene (Table 1). Predeveloped reagents containing a specific primer pair/probe mix, premixed TaqMan Reverse Transcription Reagents and TaqMan Universal MasterMix were obtained from Applied Biosystems GmbH (Weiterstadt, Germany). For BCAS1 and HMBS, new primer pair/probe combinations were designed by the Assay-by-Design-service at Applied Biosystem using the template sequences (Table 1). Pooled RNA from n = 5 animals from the five treatment groups obtained from the two independent xenograft experiments was reverse transcribed with the Reverse Transcription Reagents (Applied Biosystems) and contained 200 ng RNA in a final volume of 60 µL. The RT reaction was carried out at 25°C for 10 minutes, 48°C for 30 minutes, and 95°C for 5 minutes. The PCR was done in 96-well microtiter plates in an ABI Prism 7700 Sequence Detection System (Applied Biosystems). The reaction mixture consisted of 10 µL Taqman Universal MasterMix, 1 µL primer/probe-mix, and 2 µL of the RT reaction mix in a final volume of 20 µL. Following an initial step at 95°C for 10 minutes for denaturation of the DNA and activation of the Taq enzyme, 45 cycles consisting of denaturation at 95°C for 15 seconds and annealing/extension at 60°C for 1 minute were done. Each sample was tested in triplicate.


View this table:
[in this window]
[in a new window]
  		Table 1. TaqMan assays

 
The relative expression of each gene was determined on the basis of the threshold cycle (CT value). With cDNA from testis, placenta, and prostate (Clontech, Palo Alto, CA) standard curves were generated. For each CT value, a corresponding ng amount of cDNA was calculated. With the housekeeping gene HMBS the quantity of cDNA was normalized. Then, the normalized target gene expression was divided by the normalized gene expression value of the calibrator (i.e. untreated MaCa 3366 xenograft) to obtain the fold change of the gene in the treatment groups relative to the untreated MaCa 3366 xenograft. The SD of the normalized target gene expression relative to HMBS is calculated from the initial SDs of the target gene and HMBS.


   Results
Top
 Abstract
 Introduction
 Materials and Methods
 Results
Discussion
 References
 
Originating from the tamoxifen-sensitive human mammary carcinoma xenograft MaCa 3366, the tamoxifen-resistant model MaCa 3366/TAM was successfully established by the treatment of tumor-bearing nude mice with tamoxifen for 2 years during routine passaging. Tumor fragments were transplanted s.c. to female nude mice, which were supplemented with E2D to support tumor growth. For the analysis of differential gene expression, the five treatment groups were MaCa 3366 and MaCa 3366/TAM untreated and with short-term tamoxifen treatment, and MaCa 3366/TAM under permanent tamoxifen treatment (Fig. 1). In order to distinguish the effects of short-term tamoxifen treatment on gene expression from the effects of permanent tamoxifen treatment on resistance, the MaCa 3366 and the MaCa 3366/TAM xenograft mice were treated once with tamoxifen 24 hours before sacrifice (short-term tamoxifen treatment). The tumor growth curve of the five treatment groups from the second xenograft experiment are displayed (Fig. 2). Although MaCa 3366/TAM tumors grow more slowly than MaCa 3366 tumors growth of MaCa 3366/TAM tumors is independent of tamoxifen treatment. MaCa 3366 tumors treated with tamoxifen showed complete remission (data not shown).



View larger version (23K):
[in this window]
[in a new window]
  		Figure 2. Tumor growth curves of MaCa 3366 and MaCa 3366/TAM xenografts on nude mice (n = 5) supplemented with estradiol valeriate (E2D, 0.5mg/kg per week i.m.). MaCa 3366 and MaCa 3366/TAM were either left untreated, or treated with 50 mg/kg TAM 24 hours before sacrifice (short-term TAM treatment), and MaCa 3366/TAM were treated once per week with 50 mg/kg TAM treatment in the course of the experiment (permanent TAM treatment). Treatment was initiated 1 week after tumor transplantation and continued until the end of the experiment.R elative tumor volume (rtv) ± over time.

 
Tumor samples from both xenograft lines were used for differential gene expression analysis. Total RNA was isolated from the tumor tissues, RNA from five animals per group was pooled, and the cRNA from the first and second xenograft experiments was hybridized to Affymetrix HuGeneFL and Hu95Av2 arrays, respectively (Fig. 1). By comparing the gene expression pattern of MaCa 3366 untreated or after short-term tamoxifen treatment and MaCa 3366/TAM either untreated, after short-term or permanent tamoxifen treatment, we expect to identify genes that respond to short-term tamoxifen treatment and genes that distinguish the tamoxifen-sensitive from the tamoxifen-resistant phenotype.

We adopted two different approaches in an attempt to define gene expression patterns segregating with a tamoxifen-resistant compared with a tamoxifen-sensitive xenograft phenotype (Fig. 1). For unsupervised analysis, k-means clustering with positive correlation as distance measure was applied to the variant gene set of 503 genes on the HuGeneFL array and 538 genes on the Hu95Av2 array and genes were sorted into 10 different clusters. Profiles for the 10 clusters on the HuGeneFL and on the Hu95Av2 array are depicted in Fig. 3A and B, respectively. When the two independent xenograft experiments were hybridized to two different array types, the corresponding clusters on the two array types were identified first and afterwards genes differentially expressed in these clusters were analyzed. Cluster 5 on HugeneFL corresponds to cluster 9 on Hu95Av2 and contains genes, which are solely sharply up-regulated in MaCa 3366/TAM treated permanently with tamoxifen. Similarly, cluster 8 on HuGeneFL corresponds to cluster 3 on Hu95Av2 and contains genes that are up-regulated in all three MaCa 3366/TAM xenografts independent of tamoxifen or length of tamoxifen treatment. Moreover, the profile of cluster 4 in the HuGeneFL data set corresponds to the profile of cluster 5 in the Hu95Av2 data set and contains genes that are up-regulated with increasing exposure of the MaCa3366 xenografts to tamoxifen. However, as probe sets and expression values vary between the data sets on the HuGeneFL and Hu95Av2 array, a complete concordance of clusters cannot be expected. Applying k-means clustering on the Hu95Av2 array variant data set consisting of 538 genes, the following prominent gene clusters were identified: (1) genes that are remarkably up-regulated in all MaCa 3366/TAM xenografts (cluster 3: 88 genes), (2) genes that are noticeably down-regulated in all MaCa 3366/TAM xenografts (cluster 1: 104 genes), (3) genes that are up-regulated by short-term tamoxifen treatment in MaCa 3366 and MaCa 3366/TAM (cluster 6: 114 genes, cluster 7: 29 genes, cluster 10: 44 genes), (4) genes that are down-regulated by short-term tamoxifen treatment in MaCa 3366 and MaCa 3366/TAM (cluster 2: 16 genes), and (5) genes that are increasingly up-regulated in the MaCa 3366 and MaCa 3366/TAM xenograft tumors with extended exposure to tamoxifen (cluster 5: 51 genes). Surprisingly, we also found genes that are only up-regulated after permanent tamoxifen treatment in MaCa 3366/TAM (cluster 9: 33 genes).



View larger version (22K):
[in this window]
[in a new window]
  		Figure 3. On the variant 503 genes from the HuGeneFL array and the 538 variant genes on the Hu95Av2 array, k-means clustering was applied to obtain 10 clusters. Screenshots from the Expressionist software (Genedata). MaCa 3366 (A) untreated, (B) short-term TAM treated; MaCa 3366/TAM (C) untreated, (D) short-term TAM treated, and (E) permanently TAM treated.

 
To identify differentially expressed probe sets between MaCa 3366/TAM (untreated, short-term, and permanent tamoxifen treatment) and MaCa 3366 (untreated and short-term tamoxifen treated), a supervised analysis was done and three different fold change-based tests were applied (see Material and Methods) resulting in four groups with 57 (37) probe sets up-regulated and 97 (13) down-regulated on stringent criteria and 144 (125) probe sets up-regulated and 218 (67) down-regulated on less stringent criteria, respectively, in all tamoxifen-resistant compared with all tamoxifen-sensitive samples on the Hu95Av2 array (values for HuGeneFL in brackets). The redundancy (i.e. more than one probe set mapping to one particular gene) was removed by a BLAST analysis of the corresponding gene identifiers as supplied by Affymetrix, and the Online Mendelian Inheritance in Man (OMIM) description was attributed to each probe set. To compare differentially expressed genes on the two different array formats, the 144 up-regulated probe sets from the Hu95Av2 and 125 up-regulated probe sets from the HuGeneFL array and the 218 down-regulated probe sets from the Hu95Av2 and 67 down-regulated probe sets from the HuGeneFL array, respectively, were subjected to a BLAST analysis and annotation by Gene Ontology (GO) and OMIM. Nonredundant genes (n = 109) were regulated >2-fold in MaCa 3366/TAM as compared with MaCa 3366. The expression values of differentially expressed nonredundant genes present on both array types or present on only one array, but regulated with a fold change >2, were ranked and the genes showing the strongest differential expression in the tamoxifen-resistant compared with the tamoxifen-sensitive human breast cancer xenograft model are displayed (Table 2A and B). If more than one probe set was given for a particular gene, then the probe sets with the highest fold change determined the rank on the list. Differences in the expression values can be explained on the one hand by different tumor growth characteristics in the xenograft mice from the two independent animal experiments which should, however, be leveled out by pooling five animals per treatment group and, on the other hand, by the fact that from array generation HuGeneFL to Hu95Av2 the number of probe sets on the array was increased from 6,574 to 12,386 explaining why some genes are only present on the Hu95Av2 array.


View this table:
[in this window]
[in a new window]
  		Table 2 Summary of the differentially regulated genes in the human breast cancer xenograft model MaCa 3366/TAM compared with MaCa 3366

 
Functional Categories of Differentially Expressed Genes
For the 109 nonredundant genes differentially expressed and listed in Table 2A and B, either a known function or an inferred function was identified from OMIM and GO nomenclature (42, 43). GO stores controlled vocabulary organized under the three independent ontologies "biological process," "molecular function," and "cellular component." The differentially expressed genes were categorized according to these ontologies (Fig. 4). In addition, we introduced three categories reflecting the biological system (i.e., "known association with cancer", "IFN-inducible genes" (Table 2A), and "estrogen-responsive genes" (Table 2A and B). As all three GOs can be attributes of a particular gene, they can be assigned independently to it. In the GO biological process, molecular function, and cellular component, the distribution of genes to the categories was very similar for genes that are up- or down-regulated in MaCa 3366/TAM (Fig 4A and B) with three exceptions: the immune/defense response group, the IFN-inducible genes, and the lipid metabolism group. Numerous IFN-inducible genes were found that were coordinately up-regulated in MaCa 3366/TAM (Table 2A; Fig. 4A). Similarly, numerous estrogen-responsive genes were down-regulated in MaCa 3366/TAM (Table 2B; Fig. 4B). Examples for up- and down-regulated genes are displayed in Fig. 5. The graphs show the high degree of reproducibility of the two independent xenograft experiments and the reproducibility of results even if different array generations (i.e., HuGeneFL and Hu95Av2 with different probe sets are used).



View larger version (31K):
[in this window]
[in a new window]
  		Figure 4. Gene Ontology categories. The 54 genes that are up-regulated (A) and (B) the 55 genes that are down-regulated in MaCa 3366/TAM were categorized according to involvement in "biological processes", "molecular function", and according to "cellular component". The GO "biological process" is divided into immune/defense response, development/morphogenesis, cell growth, cell signaling, metabolism, and apoptosis. The GO "molecular function" is divided into signal transduction, transcription, protein synthesis/modification/transport, and lipid metabolism. In addition, we introduced three categories reflecting the biological system (i.e., "IFN-inducible genes," "known association with cancer," and "estrogen-responsive genes"). The GO category "cellular component" is divided into the four components: membrane, nucleus, intracellular, and extracellular. One gene can be a member of more than one GO category.

 


View larger version (35K):
[in this window]
[in a new window]
  		Figure 5. Affymetrix expression profile of selected genes. The values are derived from probe sets on the Hu95Av2 array (black and dark gray) and/or HuGeneFL array (light gray; see Table 2A and B for reference). The gene symbol here is the OMIM abbreviation, if available. A, examples of genes up-regulated in MaCa 3366/TAM compared with MaCa 3366 are neuronatin (NNAT), bone marrow stromal cell antigen 2 (BST2), UDP-glucuronosyltransferase 2 family, member B15 (UGT2B15), breast carcinoma amplified sequence 1 (BCAS1), myxovirus resistance 1 (MX1), IFN-induced gene with tetratricopeptide repeats 1 (IFIT1), IFN-induced transmembrane protein 2 (IFITM2), and IFN {alpha}–inducible protein 27 (IFI27). B, examples of genes down-regulated in MaCa 3366/TAM compared with MaCa 3366 are insulin-like growth factor binding protein 5 (IGFBP5), PDZ domain containing 1 (PDZK1), estrogen-related receptor {gamma} (ESRRG), and insulin-like growth factor 1 receptor (IGF1R).

 
IFN-Inducible Genes. Very prominently, among the genes that are up-regulated in the tamoxifen-resistant compared with the tamoxifen-sensitive breast cancer xenograft model MaCa 3366, a set of genes was discovered that was annotated as IFN-inducible (Table 2A and B; Figs. 4A and 5A, and data not shown) such as IFITM2 (34-fold), G1P2 (12-fold), G1P3 (14-fold), IFI44 (3.3-fold), IFITM1 (2.9-fold), IFIT1 (3.5-fold), MX1 (5.2-fold), NMI (2.9-fold), ISGF3G (2.4-fold), and PKRK (1.9-fold) that were all increased. The gene IFI27 was found to be up-regulated 11-fold in MaCa 3366/TAM treated permanently with tamoxifen compared with the MaCa 3366 model, but it was not up-regulated in MaCa 3366/TAM untreated or treated with tamoxifen only short-term. Numerous genes were identified which are down-regulated by short-term tamoxifen treatment of MaCa 3366 and are also down-regulated in MaCa 3366/TAM (Table 2B; Figs. 4B and 5B, and data not shown). Members of this cluster of genes with decreased expression are cyclin D1 (2.1- to 2.3-fold depending on the probe set), GATA3 (1.9-fold), PDZK1 (4.7-fold), IGFBP5 (8.4- to 10.3-fold depending on the probe set), gap junction protein 43 [GJA1, gap junction protein {alpha}1, connexin 43 (3.8-fold)], and the homeobox transcription factors MSX1 and MSX2 (2.4- and 3.2-fold, respectively). Among the genes decreased in tamoxifen-resistant xenografts, either untreated or short-term tamoxifen treated, are also the estrogen-related receptor {gamma} (ESRRG) and insulin-like growth factor 1 receptor (IGF1R; Fig. 5B).

Several genes that are associated with breast cancer are differentially expressed, [e.g., breast carcinoma amplified sequence 1 (BCAS1) was up-regulated 4.1-fold in MaCa 3366/TAM under permanent tamoxifen treatment compared with untreated MaCa 3366]. The UDP-glycosyltransferase 2 family, member B15 (UGT2B15) is increased 3.2- to 7.9-fold depending on the probe set in MaCa 3366/TAM under permanent tamoxifen treatment compared with MaCa 3366. In addition, many genes that had not previously been implicated in tamoxifen resistance were also identified: among the genes sharply up-regulated on both array types in the tamoxifen-resistant xenograft MaCa 3366/TAM, either untreated or short-term tamoxifen treated, are bone marrow stem cell antigen-2 (BST2) and neuronatin (NNAT).

Validation of Array Data with Quantitative Reverse Transcription-PCR. To confirm the differential gene expression from the array experiments employing an independent method, we did real-time quantitative reverse transcription-PCR in xenograft RNA samples (untreated and short-term tamoxifen treated MaCa 3366, untreated, short-term, and permanent tamoxifen-treated MaCa 33666/TAM) pooled from five animals each and from two independent animal experiments. We selected four genes with strong differential expression (NNAT, BST2, BCAS1, and IGFBP5) and two genes with a weaker differential expression (ESRRG and PTK6). The normalized gene expression levels relative to the expression in untreated MaCa 3366 as calibrator are shown in Fig. 6. TaqMan analysis clearly confirmed the increase of NNAT, BST2, and BCAS1 and the decrease of IGFBP5 in MaCa 3366/TAM (Fig. 6). NNAT and BST2 were sharply up-regulated in MaCa 3366/TAM under permanent tamoxifen treatment compared with untreated MaCa3366 as measured by TaqMan analysis in RNA derived from the first and second xenograft experiment. The down-regulation of ESRRG in MaCa 3366/TAM compared with MaCa 3366 on the Hu95Av2 array (2.2-fold) was confirmed with the Taqman assay (2-fold). However, ESRRG was not decreased in RNA from the first xenograft experiment as tested by Taqman PCR (data not shown). The increase (2.4 on Hu95Av2 array) of the breast cancer-related tyrosine kinase PTK6/Brk in MaCa 3366/TAM compared with MaCa 3366 was not confirmed by Taqman PCR (data not shown).



View larger version (25K):
[in this window]
[in a new window]
  		Figure 6. Representation of Taqman-PCR results. Columns, mean values for each cDNA derived from pooled RNAs from five animals; bars, ±SD. Each real-time Taqman PCR was done in triplicate. The amount of a given mRNA was first normalized to the level of the housekeeping gene HMBS. Fold differences were then computed using normalized values (standard curve method). The fold change is expressed relative to the amount of mRNA present in the calibrator MaCa 3366. Neuronatin (NNAT), bone marrow stromal cell antigen 2 (BST2), and BCAS1 are up-regulated in MaCa 3366/TAM. The same expression value of BCAS1 was measured for untreated MaCa 3366/TAM as in the calibrator in the RNA sample from the first xenograft experiment. Insulin-like growth factor binding protein 5 (IGFBP5) is down-regulated in MaCa 3366/TAM. Experiments with RNA from the first (gray columns) and second (black columns) xenograft experiments.

 

   Discussion
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
References
 
In the past, numerous estrogen-responsive genes were identified. Many of them are down-regulated after short-term tamoxifen treatment and after permanent tamoxifen treatment in the tamoxifen-resistant breast cancer xenograft MaCa 3366/TAM. Members of this group of genes which previously were shown to be estrogen-responsive and which are down-regulated in MaCa 3366/TAM are cyclin D1 (65), GATA3 (66), PDZK1 (59), gap junction protein 1 GJA1 (64), and the homeobox transcription factors MSX1 and MSX2 (60). MSX2 was shown to be induced by E2 in the human breast cancer cell line MCF-7 (61). The growth factor amphiregulin, which modulates growth by binding to the epidermal growth factor receptor, was sharply down-regulated in MaCa 3366/TAM. It is well known that estrogen induces the expression of amphiregulin in human breast carcinoma cells and that 4-OHT interferes with the induction of amphiregulin by estrogen in cell lines (67). Down-regulation of IGFBP5 (62, 63) was confirmed by real-time PCR (Fig. 6). The IGF1R was down-regulated in MaCa 3366/TAM. In a comparative study ofdifferent antiestrogen-resistant cell lines, reduced levelsofIGF1R were detected and it was proposed that this protein might be a general marker for antiestrogen resistance (14).

NNAT was found to be sharply up-regulated in MaCa 3366/TAM compared with MaCa 3366 (Table 2A; Fig. 5). By real-time PCR, the strong differential expression was confirmed (Fig. 6). NNAT is highly expressed in the brain of the murine fetus and neonate and at much lower levels in the brain of the adult (69). The function of NNAT is unknown, but its functioning as a regulator of ion channels during brain development has been proposed (70). Thus far, no association of NNAT expression with breast cancer and tamoxifen-resistant breast cancer in particular has been shown.

Breast carcinoma amplified sequence 1 (BCAS1) was up-regulated (4.1-fold on the Hu95Av2 array) in MaCa 3366/TAM treated permanently with tamoxifen compared with MaCa 3366 (Table 2A). By real-time PCR, the up-regulation was confirmed (Fig. 6). BCAS1 also known as novel amplified in breast cancer-1 (NABC1) lies in 20q13.3, a region often amplified in breast carcinoma, and is associated with aggressive tumor behavior. BCAS1 was found to be overexpressed in breast carcinomas and certain breast cancer cell lines on the RNA and protein level (71), (72). Surprisingly, BCAS1 mRNA is expressed at high levels only in normal brain and prostate (72). Currently, the function of BCAS1 is unknown; however, overexpression of human BCAS1 in NIH-3T3 cells did not affect the growth rate or anchorage-independent growth suggesting that BCAS1 is not a typical proto-oncogene (72).

Metabolism of steroid hormones are subject to glucuronidation which is a major pathway for the elimination of steroids in liver and several other steroid target tissues. Glucuronidation is catalyzed by UDP-glycosyltransferases, which transfer glucuronic acid to a wide variety of endogenous and exogenous compounds. The UDP-glycosyltransferase 2 family, member B15 (UGT2B15) is up-regulated 3.2- to 7.9-fold depending on the probe set in MaCa 3366/TAM under permanent tamoxifen treatment compared with MaCa 3366 (Table 2A; Fig. 5). It was shown that recombinant UGT2B15 selectively glucuronidates cis-4-OHT, the active metabolite of tamoxifen (73). It is of interest that cis-4-OHT existed at a higher concentration than trans-4-OHT in a MCF-7 breast carcinoma xenograft model of tamoxifen-administered mice who had become resistant to tamoxifen in the course of tamoxifen treatment (74). In addition, it was shown that acquired resistance to tamoxifen is associated with markedly reduced cellular levels of tamoxifen together with an accumulation of theless antiestrogenic cis-4-OHT in human breast cancer patients who had become resistant to tamoxifen (75). Possibly, tamoxifen administration induces the expression of UGT2B15, which metabolizes and inactivates cis-4-OHT. Interestingly, the phenol sulfotransferase SULT1A that was shown to sulfate 4-OHT (73) was induced by tamoxifen in the human breast cancer cell line ZR-75-1, as shown by SAGE analysis (32), suggesting a role for tamoxifen metabolizing enzymes in the development of tamoxifen resistance.

Many IFN-inducible genes were up-regulated in MaCa 3366/TAM as compared with the MaCa 3366 xenograft (Table 2A). Via activation of the JAK signal transducer and activator of transcription (STAT) pathway, IFNs but also other cytokines like IL-6 elicit multiple biological responses, which are mediated by the proteins encoded by IFN-inducible genes. Phosphorylated STAT proteins bind DNA sequences in the promoters of IFN-inducible genes that are called {gamma}-IFN activated sequences and IFN-stimulated response elements (76). Two IFN-inducible genes, the double-stranded RNA-activated protein kinase PRKR and MX1, which belongs to the dynamin family of large GTPases (77), had an increased expression in MaCa 3366/TAM compared with MaCa 3366 (Tables 2A and 3, Fig. 5). Interestingly, MX1 was previously identified in a comparative genomic and proteomic analysis of the human breast cancer cell line T47D and its antiestrogen (Faslodex , ZM 182780)-resistant derivative T47D-r (39)(78). In this study, MX1 was sharply up-regulated in the antiestrogen-resistant cell line T47D-r on the mRNA and protein level.

The biological functions of many other IFN-inducible genes that were coordinately up-regulated in the tamoxifen-resistant as compared with the parental xenograft tumor such as IFITM1, IFITM2, GIP2 and GIP3 (45), IFIT1(51), and IFI44 (49) is still completely unknown (79). In contrast to the other IFN-inducible genes, IFI27, which was initially identified from E2-treated human breast cancer cells (58), was up-regulated 11-fold in MaCa 3366/TAM treated permanently with tamoxifen compared with MaCa 3366 but it was not up-regulated in untreated or short-term tamoxifen treated MaCa 3366/TAM (Fig. 5).

In the tamoxifen-resistant breast cancer xenograft model the bone marrow stromal cell antigen 2 (BST2) was sharply up-regulated and differential expression was confirmed by real-time PCR (Fig. 6). BST2 is also known as HM1.24 antigen and is a type II membrane glycoprotein that is preferentially expressed on terminally differentiated B cells(80). The promoter contains three STAT3 DNA-binding sites (81). Expression of BST2 might therefore directly be induced by STAT3. A human anti-BST2 antibody was generated which is currently being tested for the immunotherapy of multiple myeloma (82). Recently, it has been shown that the rat homologue of BST2/HM1.24 has both a cell surface and an intracellular localization and it is efficiently internalized from the cell surface (83).

The fact that IFN-inducible genes are up-regulated in xenograft tumors is not unprecedented. When human breast tumors were analyzed by gene expression profiling, a set of coexpressed genes was identified which was annotated as being IFN-stimulated (20, 21). The strength of expression of these genes correlated positively with the protein expression of STAT1 and the authors suggested that high levels of STAT1 and STAT3 could increase the expression of a set of IFN-inducible genes (20). Moreover, in a gene expression profiling analysis of ductal carcinoma in situ as compared with invasive ductal carcinoma (84), IFN-inducible genes were most strongly up-regulated in the ductal carcinoma in situ phase of the breast carcinomas. Although we did not identify STAT1 and STAT3 as transcriptionally up-regulated genes in MaCa 3366/TAM by Affymetrix array analysis, it is conceivable that they are post-transcriptionally activated (e.g., by growth factor receptor or receptor tyrosine kinase mediated phosphorylation). Additionally, it is well known that STAT3 and to some extent also STAT1 can be activated in breast carcinomas (85) and that constitutively active mutants of STAT3 are oncogenic (86–89). Activated STAT3 promotes the growth and survival of tumor cells thereby contributing to malignancy (86, 89). STAT1 and STAT3 can be activated by various receptor and non-receptor tyrosinekinases (e.g., the epidermal growth factor receptor and c-Src; refs. 86, 88). Moreover, it was shown in vitro that E2-mediated inhibition of IL-6–induced STAT3 activation was reversed by tamoxifen indicating that under permanent tamoxifen treatment STAT3 could indeed be hyperactivated (90, 91) One may assume that heterodimers of the transcription factors STAT1/STAT2 and STAT11/STAT3, which bind to IFN-stimulated response elements (76) are involved in the up-regulation of MX1 and BST2. Thus, the up-regulation seen for MX1 and BST2 in MaCa 3366/TAM could be explained by an activated STAT pathway in these cells. Activation of STAT3 in MaCa 3366/TAM might be responsible for the up-regulation of the numerous IFN-inducible genes that contain an IFN-stimulated response element in their promoter.

In conclusion, this study provides a detailed analysis of the genes that are up- and down-regulated in an in vivo model of tamoxifen resistance. More than 100 transcripts are changed in abundance in MaCa 3366/TAM as compared with MaCa 3366. Our future studies will aim at identifying which of the genes in the tamoxifen-resistant xenograft model contribute most to the phenotype. Among the genes that are differentially expressed in the tamoxifen-resistant tumors, there are several IFN-inducible genes, estradiol-regulated genes, and genes known to be involved in breast carcinogenesis. The genes NNAT and BST2 were strongly increased in MaCa 3366/TAM. The differential expression of four genes (NNAT, BST2, IGFBP5, and BCAS1) was confirmed by TaqMan PCR. To examine a larger number of differentially expressed genes, ideally in tissue samples obtained from tamoxifen-treated patients and patients who became resistant in the course of tamoxifen therapy, we are currently designing TaqMan Low Density Arrays with which we will determine whether overexpression of these genes is a hallmark of tamoxifen-resistant breast cancer. Very recently, based on gene expression profiling, a two-gene expression ratio involving the expression ratio of HOXB13 versus IL17BR was shown to predict clinical outcome in breast cancer patients treated with tamoxifen (92) strongly supporting this concept. An extension of this approach should help to pinpoint which genes might be predictive of tamoxifen-resistant growth.


   Acknowledgments
 
We thank Daniela Sauvageot and Daniela Schilling for expert technical assistance for the Affymetrix experiments; Britta Büttner for animal experiments; Claudia Neumann for Taqman analysis; and Jens Hoffmann from Experimental Oncology, Schering AG, Berlin, Germany and Andreas Hohn from Genedata, Basel, Switzerland for stimulating discussions.


   Footnotes
 
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 accordance with U.S.C. Section 1734 solely to indicate this fact.

Note: M. Becker and A. Sommer contributed equally to this work.

2 Fichtner, I. and Becker M., unpublished observation. Back

Received 8/18/04; revised 11/ 2/04; accepted 11/ 8/04.


   References
Top
 Abstract
 Introduction
 Materials and Methods
 Results
 Discussion
 References
 

  1. Wong ZW, Ellis MJ. First-line endocrine treatment of breast cancer: aromatase inhibitor or antioestrogen? Br J Cancer 2004;90:20–5.[CrossRef][Medline]
  2. Wickerham L. Tamoxifen: an update on current data and where it can now be used. Breast Cancer Res Treat 2002;75:7–12.[CrossRef]
  3. Robertson JF. Oestrogen receptor: a stable phenotype in breast cancer. Br J Cancer 1996;73:5–12.[Medline]
  4. Robertson JF, Cannon PM, Nicholson RI, Blamey RW. Oestrogen and progesterone receptors as prognostic variables in hormonally treated breast cancer. Int J Biol Markers 1996;11:29–35.[Medline]
  5. Shiau AK, Barstad D, Loria PM, et al. The structural basis of estrogen receptor/coactivator recognition and the antagonism of this interaction by tamoxifen. Cell 1998;95:927–37.[CrossRef][Medline]
  6. Katzenellenbogen BS, Katzenellenbogen JA. Defining the "S" in SERMs. Science 2002;295:2380–1.[Abstract/Free Full Text]
  7. Schafer JM, Bentrem DJ, Takei H, Gajdos C, Badve S, Jordan VC. A mechanism of drug resistance to tamoxifen in breast cancer. J Steroid Biochem Mol Biol 2002;83:75–83.[CrossRef][Medline]
  8. Tonetti DA, Jordan VC. The role of estrogen receptor mutations in tamoxifen-stimulated breast cancer. J Steroid Biochem Mol Biol 1997;62:119–28.[CrossRef][Medline]
  9. Madsen MW, Reiter BE, Larsen SS, Briand P, Lykkesfeldt AE. Estrogen receptor messenger RNA splice variants are not involved in antiestrogen resistance in sublines of MCF-7 human breast cancer cells. Cancer Res 1997;57:585–9.[Abstract/Free Full Text]
  10. Naundorf H, Becker M, Lykkesfeldt AE, et al. Development and characterization of a tamoxifen-resistant breast carcinoma xenograft. Br J Cancer 2000;82:1844–50.[CrossRef][Medline]
  11. Fichtner I, Becker M, Zeisig R, Sommer A. In vivo models for endocrine-dependent breast carcinomas: special considerations of clinical relevance. Eur J Cancer 2004;40:845–51.
  12. Clarke R, Skaar TC, Bouker KB, et al. Molecular and pharmacological aspects of antiestrogen resistance. J Steroid Biochem Mol Biol 2001;76:71–84.[CrossRef][Medline]
  13. Katzenellenbogen BS, Montano MM, Ekena K, Herman ME, McInerney EM. William L. McGuire Memorial Lecture. Antiestrogens: mechanisms of action and resistance in breast cancer. Breast Cancer Res Treat 1997;44:23–38.[CrossRef][Medline]
  14. Brockdorff BL, Heiberg I, Lykkesfeldt AE. Resistance to different antiestrogens is caused by different multi-factorial changes and is associated with reduced expression of IGF receptor I {alpha}. Endocr Relat Cancer 2003;10:579–90.[Abstract]
  15. van der Flier S, Brinkman A, Look MP, et al. Bcar1/p130Cas protein and primary breast cancer: prognosis and response to tamoxifen treatment. J Natl Cancer Inst 2000;92:120–7.[Abstract/Free Full Text]
  16. Dorssers LC, van der Flier S, Brinkman A, et al. Tamoxifen resistance in breast cancer: elucidating mechanisms. Drugs 2001;61:1721–33.[CrossRef][Medline]
  17. Luo LY, Diamandis EP, Look MP, Soosaipillai AP, Foekens JA. Higher expression of human kallikrein 10 in breast cancer tissue predicts tamoxifen resistance. Br J Cancer 2002;86:1790–6.[CrossRef][Medline]
  18. Michalides R, Griekspoor A, Balkenende A, et al. Tamoxifen resistance by a conformational arrest of the estrogen receptor {alpha} after PKA activation in breast cancer. Cancer Cell 2004;5:597–605.[CrossRef][Medline]
  19. Naundorf H, Fichtner I, Büttner B, Frege J. Establishment and characterization of a new human oestradiol- and progesterone-receptor-positive mammary carcinoma serially transplantable in nude mice. J Cancer Res Clin Oncol 1992;119:35–40.[CrossRef][Medline]
  20. Perou CM, Jeffrey SS, van de Rijn M, et al. Distinctive gene expression patterns in human mammary epithelial cells and breast cancers. Proc Natl Acad Sci U S A 1999;96:9212–7.[Abstract/Free Full Text]
  21. Perou CM, Sørlie T, Eisen MB, et al. Molecular portraits of human breast tumours. Nature 2000;406:747–52.[CrossRef][Medline]
  22. Sørlie T, Perou CM, Tibshirani R, et al. Gene expression patterns of breast carcinomas distinguish tumor subclasses with clinical implications. Proc Natl Acad Sci U S A 2001;98:10869–74.[Abstract/Free Full Text]
  23. Gruvberger S, Ringner M, Chen Y, et al. Estrogen receptor status in breast cancer is associated with remarkably distinct gene expression patterns. Cancer Res 2001;61:5979–84.[Abstract/Free Full Text]
  24. van't Veer LJ, Dai H, van de Vijver MJ, et al. Gene expression profiling predicts clinical outcome of breast cancer. Nature 2002;415:530–6.[CrossRef][Medline]
  25. van de Vijver MJ, He YD, van't Veer LJ, et al. A gene-expression signature as a predictor of survival in breast cancer. N Engl J Med 2002;347:1999–2009.[Abstract/Free Full Text]
  26. Ma XJ, Salunga R, Tuggle JT, et al. Gene expression profiles of human breast cancer progression. Proc Natl Acad Sci U S A 2003;100:5974–9.[Abstract/Free Full Text]
  27. Sotiriou C, Neo SY, McShane LM, et al. Breast cancer classification and prognosis based on gene expression profiles from a population-based study. Proc Natl Acad Sci U S A 2003;100:10393–8.[Abstract/Free Full Text]
  28. Charpentier AH, Bednarek AK, Daniel RL, et al. Effects of estrogen on global gene expression: identification of novel targets of estrogen action. Cancer Res 2000;60:5977–83.[Abstract/Free Full Text]
  29. Inoue A, Yoshida N, Omoto Y, et al. Development of cDNA microarray for expression profiling of estrogen-responsive genes. J Mol Endocrinol 2002;29:175–92.[Abstract]
  30. Levenson AI, Svoboda KM, Pease KM, et al. Gene expression profiles with activation of the estrogen receptor {alpha}-selective estrogen receptor modulator complex in breast cancer cells expressing wild-type estrogen receptor. Cancer Res 2002;62:4419–26.[Abstract/Free Full Text]
  31. Omoto Y, Hayashi SA. Study of estrogen signaling using DNA microarray in human breast cancer. Breast Cancer 2002;9:308–11.[Medline]
  32. Seth P, Krop I, Porter D, Polyak K. Novel estrogen and tamoxifen induced genes identified by SAGE (Serial Analysis of Gene Expression). Oncogene 2002;21:836–43.[CrossRef][Medline]
  33. Soulez M, Parker MG. Identification of novel oestrogen receptor target genes in human ZR75-1 breast cancer cells by expression profiling. J Mol Endocrinol 2001;27:259–74.[Abstract]
  34. Cunliffe HE, Ringner M, Bilke S, et al. The Gene expression response of breast cancer to growth regulators. Cancer Res 2003;63:7158–66.[Abstract/Free Full Text]
  35. Frasor J, Danes JM, Komm B, Chang KC, Lyttle CR, Katzenellenbogen BS. Profiling of estrogen up- and down-regulated gene expression in human breast cancer cells: insights into gene networks and pathways underlying estrogenic control of proliferation and cell phenotype. Endocrinology 2003;144:4562–74.[Abstract/Free Full Text]
  36. Hodges LC, Cook JD, Lobenhofer EK, et al. Tamoxifen functions as a molecular agonist inducing cell cycle-associated genes in breast cancer cells. Mol Cancer Res 2003;1:300–11.[Abstract/Free Full Text]
  37. Frasor J, Stossi F, Danes JM, Komm B, Lyttle CR, Katzenellenbogen BS. Selective estrogen receptor modulators: discrimination of agonistic versus antagonistic activities by gene expression profiling in breast cancer cells. Cancer Res 2004;646:1522–33.
  38. Naundorf H, Arnold W. Breeding of nude mice under improved conventional conditions. Z Versuchstierkd 1981;23:77–9.[Medline]
  39. Huber M, Bahr I, Krätzschmar JR, et al. Comparison of proteomic and genomic analyses of the human breast cancer cell line T47D and the antiestrogen-resistant derivative T47D-r. Mol Cell Proteomics 2004;3:43–55.[Abstract/Free Full Text]
  40. Gmuender H. Perspectives and challenges for DNA microarrays in drug discovery and development. BioTechniques 2002;32:152–8.[Medline]
  41. Kun Y, How LC, Hoon TP, et al. Classifying the estrogen receptor status of breast cancers by expression profiles reveals a poor prognosis subpopulation exhibiting high expression of the ERBB2 receptor. Hum Mol Genet 2003;12:3254–8.
  42. Online Mendelian Inheritance in Man, OMIM (TM) database on the Internet. Baltimore (MD): McKusick-Nathans Institute for Genetic Medicine, Johns Hopkins University and Bethesda (MD): National Center for Biotechnology Information, National Library of Medicine; 2000 cited 2004 Jun. Available from http://www.ncbi.nlm.nih.gov/omim.
  43. Ashburner M, Ball CA, Blake JA, et al. Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat Genet 2002;25:25–9.
  44. Lewin AR, Reid LE, McMahon M, Stark GR, Kerr IM. Molecular analysis of a human interferon-inducible gene family. Europ J Biochem 1991;199:417–23.[Medline]
  45. Porter AC, Chernajovsky Y, Dale TC, Gilbert CS, Stark GR, Kerr IM. Interferon response element of the human gene 6-16. EMBO J 1988;7:85–92.[Medline]
  46. Lebrun SJ, Shpall RL, Naumovski L. Interferon-induced upregulation and cytoplasmic localization of Myc-interacting protein Nmi. J Interferon Cytokine Res 1998;18:767–71.[Medline]
  47. Veals SA, Schindler C, Leonard D, et al. Subunit of an {alpha}-interferon-responsive transcription factor is related to interferon regulatory factor and myb families of DNA-binding proteins. Mol Cell Biol 1992;12:3315–24.[Abstract/Free Full Text]
  48. Deblandre GA, Marinx OP, Evans SS, et al. Expression cloning of an interferon-inducible 17-kDa membrane protein implicated in the control of cell growth. J Biol Chem 1995;270:23860–6.[Abstract/Free Full Text]
  49. Kitamura A, Takahashi K, Okajima A, Kitamura N. Induction of the human gene for p44, a hepatitis-C-associated microtubular aggregate protein, by interferon-{alpha}/ß. Eur J Biochem 1994;224:877–83.[Medline]
  50. Watanabe T, Inoue S, Hiroi H, Orimo A, Kawashima H, Muramatsu M. Isolation of estrogen-responsive genes with a CpG island library. Mol Cell Biol 1998;18:442–9.[Abstract/Free Full Text]
  51. Wathelet M, Moutschen S, Defilippi P, et al. Molecular cloning, full-length sequence and preliminary characterization of a 56-kDa protein induced by human interferons. Eur J Biochem 1986;155:11–7.[Medline]
  52. Cavailles V, Dauvois S, L'Horset F, et al. Nuclear factor RIP140 modulates transcriptional activation by the estrogen receptor. EMBO J 1995;14:3741–51.[Medline]
  53. Roberts M, Wallace J, Jeltsch JM, Berry M. The 5-prime flanking region of the human pS2 gene mediates its transcriptional activation by estrogen in MCF-7 cells. Biochem Biophys Res Commun 1988;151:306–13.[CrossRef][Medline]
  54. Westley BR, May FE. Oestrogen regulates cathepsin D mRNA levels in oestrogen responsive human breast cancer cells. Nucleic Acids Res 1987;15:3773–86.[Abstract/Free Full Text]
  55. Augereau P, Miralles F, Cavailles V, Gaudelet C, Parker M, RochefortH. Characterization of the proximal estrogen-responsive element of human cathepsin D gene. Mol Endocrinol 1994;8:693–703.[Abstract]
  56. Meurs E, Chong K, Galabru J, et al. Molecular cloning and characterization of the human double-stranded RNA-activated protein kinase induced by interferon. Cell 1990;62:379–90.[CrossRef][Medline]
  57. Horisberger MA, Wathelet M, Szpirer J, et al. cDNA cloning and assignment to chromosome 21 of IFI-78K gene, the human equivalent of murine Mx gene. Somat Cell Mol Genet 1988;14:123–31.[CrossRef][Medline]
  58. Rasmussen UB, Wolf C, Mattei MG, et al. Identification of a new interferon-{alpha}-inducible gene (p27) on human chromosome 14q32 and its expression in breast carcinoma. Cancer Res 1993;53:4096–101.[Abstract/Free Full Text]
  59. Ghosh MG, Thompson DA, Weigel RJ. PDZK1 and GREB1 are estrogen-regulated genes expressed in hormone-responsive breast cancer. Cancer Res 1999;60:6367–75.
  60. Friedmann Y, Daniel CW. Regulated expression of homeobox genes Msx-1 and Msx-2 in mouse mammary gland development suggests a role in hormone action and epithelial-stromal interactions. Dev Biol 1996;177:347–55.[CrossRef][Medline]
  61. Phippard DJ, Weber-Hall SJ, Sharpe PT, et al. Regulation of Msx-1, Msx-2, Bmp-2 and Bmp-4 during foetal and postnatal mammary gland development. Development 1996;122:2729–37.[Abstract]
  62. Parisot JP, Leeding KS, Hu XF, DeLuise M, Zalcberg JR, Bach LA. Induction of insulin-like growth factor binding protein expression by ICI 182,780 in a tamoxifen-resistant human breast cancer cell line. Breast Cancer Res Treat 1999;55:231–42.[CrossRef][Medline]
  63. Huynh H, Yang XF, Pollak M. A role for insulin-like growth factor binding protein 5 in the antiproliferative action of the antiestrogen ICI 182780. Cell Growth Differ 1996;7:1501–6.[Abstract]
  64. Yu W, Dahl G, Werner R. The connexin43 gene is responsive to estrogen. Proc R Soc London B Biol Sci 1994;255:125–32.[Medline]
  65. Sabbah M, Courilleau D, Mester J, Redeuilh G. Estrogen induction of the cyclin D1 promoter: involvement of a cAMP response-like element. Proc Natl Acad Sci U S A 1999;96:11217–22.[Abstract/Free Full Text]
  66. Hoch RV, Thompson DA, Baker RJ, Weigel RJ. GATA-3 is expressed in association with estrogen receptor in breast cancer. Int J Cancer 1999;84:122–8.[CrossRef][Medline]
  67. Martinez-Lacaci I, Saceda M, Plowman GD, et al. Estrogen and phorbol esters regulate amphiregulin expression by two separate mechanisms in human breast cancer cell lines. Endocrinology 1995;136:136–92.
  68. Liu HL, Golder-Novoselsky E, Seto MH, Webster L, McClary J, Zajchowski DA. The novel estrogen-responsive B-box protein (EBBP) gene is tamoxifen-regulated in cells expressing an estrogen receptor DNA-binding domain mutant. Mol Endocrinol 1998;12:1733–48.[Abstract/Free Full Text]
  69. Kikyo N, Williamson CM, John RM, et al. Genetic and functional analysis of neuronatin in mice with maternal or paternal duplication of distal Chr 2. Dev Biol 1997;190:66–77.[CrossRef][Medline]
  70. Dou D, Joseph R. Cloning of human neuronatin gene and its localization to chromosome-20q 11.2-12: the deduced protein is a novel "proteolipid". Brain Res 1996;723:8–22.[CrossRef][Medline]
  71. Collins C, Rommens JM, Kowbel D, et al. Positional cloning of ZNF217 and NABC1: genes amplified at 20q13.2 and overexpressed in breast carcinoma. Proc Natl Acad Sci U S A 1998;95:8703–8.[Abstract/Free Full Text]
  72. Beardsley DI, Kowbel D, Lataxes TA, et al. Characterization of the novel amplified in breast cancer-1 (NABC1) gene product. Exp Cell Res 2003;290:402–13.[CrossRef][Medline]
  73. Nishiyama T, Ogura K, Nakano H, et al. Reverse geometrical selectivity in glucuronidation and sulfation of cis- and trans-4-hydroxytamoxifens by human liver UDP-glucuronosyltransferases and sulfotransferases. Biochem Pharmacol 2002;63:1817–30.[CrossRef][Medline]
  74. Osborne CK, Coronado E, Allred DC, Wiebe V, DeGregorio M. Acquired tamoxifen resistance: correlation with reduced breast tumor levels of tamoxifen and isomerization of trans-4-hydroxytamoxifen. J Natl Cancer Inst 1991;83:1477–82.[Abstract/Free Full Text]
  75. Osborne CK, Wiebe VJ, McGuire WL, Ciocca DR, DeGregorio MW. Tamoxifen and the isomers of 4-hydroxytamoxifen in tamoxifen-resistant tumors from breast cancer patients. J Clin Oncol 1992;10:304–10.[Abstract]
  76. Levy DE, Darnell JE Jr. Stats: transcriptional control and biological impact. Nat Rev Mol Cell Biol 2002;3:651–62.[CrossRef]