Abstract
Members of the recently recognized SRC-1 family of transcriptional coactivators interact with steroid hormone receptors to enhance ligand-dependent transcription. AIB1, a member of the SRC-1 family, was cloned during a search on the long arm of chromosome 20 for genes whose expression and copy number were elevated in human breast cancers. AIB1 amplification and overexpression were observed in four of five estrogen receptor–positive breast and ovarian cancer cell lines. Subsequent evaluation of 105 unselected specimens of primary breast cancer found AIB1 amplification in approximately 10 percent and high expression in 64 percent of the primary tumors analyzed. AIB1 protein interacted with estrogen receptors in a ligand-dependent fashion, and transfection of AIB1 resulted in enhancement of estrogen-dependent transcription. These observations identify AIB1 as a nuclear receptor coactivator whose altered expression may contribute to development of steroid-dependent cancers.
Gene amplification is a frequent mechanism of increased gene expression in human cancers. In breast cancer, commonly amplified chromosomal regions are derived from 17q12, 8q24, and 11q13 and encode erbB-2, c-myc, and cyclin D1, respectively (1). Molecular cytogenetic studies of breast cancers have revealed the occurrence of additional regions of increased DNA copy number whose target genes are unknown, including 20q (2). Recently, we used chromosome microdissection and hybrid selection to clone expressed sequences from 20q in an attempt to identify genes of biological significance (3). In this fashion, we isolated partial cDNAs for a candidate target gene termed AIB1 (amplified in breast cancer-1), which was ubiquitously expressed in normal human tissues (3). We now report that AIB1 is a member of the SRC-1 family of nuclear receptor (NR) coactivators, that it is amplified and overexpressed in breast and ovarian cancer cell lines as well as in breast cancer biopsies, that it interacts with estrogen receptor (ER), and that it functions to enhance ER-dependent transcription.
Sequence analysis of partial AIB1 cDNAs provided the first evidence of similarity between AIB1 and the SRC-1 family. SRC-1 and TIF2 are closely related transcriptional coactivators recently isolated on the basis of their affinity for NRs (4, 5). Although the mechanism of action of SRC-1 has not been completely elucidated, in addition to interacting with NRs, SRC-1 binds to the transcriptional integrators CREB binding protein (CBP) and the closely related p300, which interact directly with the basal transcription machinery (6).
To further characterize AIB1, the full-length cDNA was cloned and sequenced (7), revealing an open reading frame that encodes a protein of 1420 amino acids with a predicted molecular mass of 155 kD (Fig. 1). Database searches with BLASTP identified a highly significant similarity of AIB1 with TIF2 (45% amino acid identity) and SRC-1 (33% amino acid identity) (8). Like TIF2 and SRC-1, AIB1 contains a basic helix-loop-helix (bHLH) domain preceding a PAS (Per/Arnt/Sim) region, serine- and threonine-rich regions, and a charged cluster. There is also a glutamine-rich region that, unlike SRC-1 and TIF2, contains a polyglutamine tract. AIB1 also contains three copies of the conserved LXXLL motif (L = leucine, X = any amino acid), which was recently demonstrated to be critical to the coactivator receptor interaction (9, 10).
Deduced amino acid sequence and structural motifs of AIB1 (GenBank accession number AF012108) (22). Residues highlighted in black are identical in AIB1, TIF2, and SRC-1; those identical with TIF2 (GenBank accession numberX97674) or SRC-1 (GenBank accession number U59302) are highlighted in gray or boxed, respectively. Three copies of the LXXLL NR interaction motif are underlined. Structural features of AIB1 include bHLH and PAS domains (with the highly conserved PAS A and B regions in dark gray), serine- and threonine-rich regions, and a group of charged residues (+/−). The glutamine-rich region and the polyglutamine tract are also indicated. The alignment was generated with DNASTAR software.
Because of this strong sequence similarity, we evaluated the amplification and expression of AIB1 in a series of ER-positive and -negative breast and ovarian cancer cell lines (11). AIB1 gene copy number was determined by fluorescence in situ hybridization (FISH) (Fig. 2). High-level amplification of AIB1 (>20-fold) was observed in three ER-positive breast carcinoma cell lines (BT-474, MCF-7, and ZR75-1) and in one ovarian carcinoma cell line (BG-1) (Fig. 2, A and B). Overall, AIB1 was amplified in four of five ER-positive cell lines tested and in zero of six ER-negative cell lines (12). To determine whether AIB1 amplification also occurred in uncultured cells from tumor biopsies, we screened 105 unselected breast cancer specimens for AIB1 amplification by FISH. Ten specimens of primary tumors (9.5%) demonstrated amplification of AIB1, although the amplification levels were not as high as in the cell lines (13).
Bicolor FISH analysis demonstrates AIB1 amplification (red signals) in breast cancer cell line ZR75-1 (A), ovarian cancer cell line BG-1 (B), and two uncultured breast cancer samples (C). Intrachromosomal amplification of AIB1 (arrows) is apparent in metaphase chromosomes of ZR75-1 and BG1, and numerous copies of AIB1 are resolved in the adjacent interphase nuclei. The Spectrum Orange (Vysis)-labeled AIB1 P1 probe (3) was hybridized with a biotinylated reference probe for 20q11 (RMC20P037) (A and B) or a fluorescein-labeled probe for 20p (RMC20C039) (C), which appear green.
Previous interphase FISH studies have demonstrated that amplification of chromosome 20q in breast cancer is complex and involves several distinct variably coamplified chromosomal segments derived from 20q11, 20q12, and 20q13 (14). Importantly, in cancer cell lines BG-1 and ZR75-1, amplification of AIB1 (which maps to 20q12) has occurred independently of both the 20q11 and 20q13 regions (12). A similar pattern of amplification with higher copy number at AIB1 than elsewhere on 20q was also found in two tumor specimens (Fig. 2C). Although most instances of AIB1 amplification were observed in conjunction with increased 20q13 copy number, cases of 20q13 amplification with normal AIB1 copy number also were identified. These results indicate that AIB1 defines an independently selected region of amplification on 20q that takes part in a complex process of gene amplification involving multiple target regions on the same arm of the chromosome.
AIB1 expression was examined first in tumor cell lines with and without AIB1 amplification and compared with expression of ER, SRC-1, and TIF2 by Northern blotting. In direct concordance with its amplification status, AIB1 was highly overexpressed in the four ER-positive cell lines with increased AIB1 copy number (BT-474, MCF-7, ZR75-1, and BG-1) (Fig. 3A). In contrast, expression of TIF2 and SRC-1 remained relatively constant in all cell lines tested (Fig. 3A). AIB1 expression was then examined in primary breast malignancies by mRNA in situ hybridization to 75 of the tumor specimens previously used for FISH analysis (15). Normal mammary epithelium expressed moderate amounts of AIB1 mRNA (Fig. 3B). Among the tumors, high levels of AIB1 expression were observed in all 10 cases with AIB1 amplification (Fig. 3B). In addition, AIB1 expression was increased relative to normal mammary epithelium in 38 of 65 (58%) of the remaining tumors or 64% overall. This observation indicates that overexpression of AIB1 by mechanisms other than amplification also occurs frequently in human breast cancers. ER status, as determined by immunohistochemistry, was not strongly correlated with either AIB1 amplification or expression in the unselected tumors (12, 15). However, both clinical specimens that demonstrated independent high-level AIB1 amplification were from postmenopausal patients and were ER and progesterone receptor–positive. One of these specimens was a metastasis from a patient who subsequently responded favorably to treatment with tamoxifen.
(A) Increased expression of AIB1 in the AIB1 amplified cell lines BT-474, ZR-75-1, MCF7, and BG-1 is apparent on Northern analysis. The blot was hybridized sequentially with the indicated probe to compare AIB1 expression with that of ER, TIF2, and SRC-1. To avoid cross-hybridization between these related genes and to match signal intensities, we used similar-sized probes from the 3′ untranslated regions of AIB1, TIF2, and SRC-1 (23). Each of these probes detected a signal in normal mammary RNA (Clontech) on longer exposure. Electrophoresis, transfer, and hybridization of 15 μg of total RNA was performed by standard methods (24). We used a β-actin probe as a control for loading error. (B) Dark-field microscopic image of AIB1 mRNA in situ expression demonstrating moderate expression in normal breast epithelium (left) and high-level expression of AIB1 in malignant mammary epithelial cells carrying increased copies of the AIB1 gene (right) with marked cell to cell heterogeneity.
We then sought to determine whether expression of AIB1 increases ER ligand-dependent transactivation. To accomplish this, we performed transient transfection assays to examine the effect of increasing amounts of AIB1 on transcription of an ER-dependent reporter (16). These results conclusively demonstrated that cotransfection of AIB1 led to a dose-dependent increase in estrogen-dependent transcription (Fig.4A). This increase was not observed when the estrogen antagonist 4-hydroxytamoxifen (4-OHT) was substituted for 17β-estradiol or when the estrogen response element (ERE) was removed from the reporter plasmid (Fig. 4A). The degree of coactivation was not as pronounced as that reported for SRC-1 and TIF2, but it was comparable to that observed with the very recently described murine SRC-1 family member p/CIP (10). Further evidence of ER–AIB1 interaction was provided by a glutathione S-transferase (GST) pulldown assay (17), which demonstrated that a GST fusion protein that contains the region predicted to contain the NR interaction domain (residues 605 to 1294 of AIB1) associates with ER in a ligand-dependent fashion (Fig. 4B). These results demonstrate that AIB1 interacts directly with ER and increases estrogen-dependent transcriptional activity.
AIB1 increases estrogen-dependent transcription from an ER reporter plasmid in vivo and interacts with the ER in GST pulldown assays. (A) CV-1 cells were transiently transfected with 250 ng of ER expression vector (pHEGO-hyg), 1.0 μg of luciferase reporter plasmid (pGL3.luc.3ERE or pGL3 lacking ERE), and increasing amounts of pcDNA3.1-AIB1 and incubated in the absence (open bars) or presence of 10 nM 17β-estradiol (E2) (solid bars) or 100 nM 4OHT (hatched bars). Luciferase activity is expressed in RLU. Data are means of three determinations from one of four replicate experiments. Error bars indicate 1 SD. (B) AIB1 protein binds to ER in a ligand-dependent fashion. To examine the interaction of ER with AIB1 in vitro, we incubated unliganded or estradiol-treated ER with glutathione-Sepharose beads containing GST, GST-AIB.T1 (amino acids 605 to 1294), or GST-AIB.N1 (amino acids 1 to 194). Note that interaction is specific for the AIB1.T1 fragment that contains the LXXLL motifs. Bound ER was visualized by SDS-PAGE, immunoblotting, and chemiluminescence.
This study has identified AIB1 as a member of the SRC-1 family of transcriptional coactivators that is frequently amplified and overexpressed in breast cancer. Its amplification, effect on estrogen-dependent transcription, and interaction with ER implicate AIB1 as an important component of the estrogen-response pathway. On the basis of the observation that GRIP1, the mouse ortholog of TIF2, is a coactivator of multiple NRs (18) and the presence in AIB1 of the newly described LXXLL NR interaction motifs (9, 10), we do not expect that AIB1 coactivation is confined to ER alone. It is likely that ER is a major but not exclusive target of AIB1–NR interaction in mammary epithelial cells. Accumulation of excess quantities of the usually limiting AIB1 protein could have a profound effect on the expression of numerous genes that are normally regulated by NRs. Additionally, interaction with CBP/p300 is important in the function of SRC-1, TIF2, and p/CIP; it also is likely to be important for AIB1 function (6, 10). Thus, AIB1 overexpression could potentially perturb signal integration by CBP/p300 and affect multiple signal transduction pathways. Our observations of AIB1 amplification suggest that the resulting dysregulation of gene expression provides a selective advantage for tumor growth.
↵* To whom correspondence should be addressed. E-mail: pmeltzer{at}nhgri.nih.gov
