p23 is a heat shock protein 90 (Hsp90) cochaperone located in both the cytoplasm and nucleus that stabilizes unliganded steroid receptors, controls the catalytic activity of certain kinases, regulates protein-DNA dynamics, and is upregulated in several cancers. We had previously shown that p23-overexpressing MCF-7 cells (MCF-7+p23) exhibit increased invasion without affecting the estrogen-dependent proliferative response, which suggests that p23 differentially regulates genes controlling processes linked to breast tumor metastasis. To gain a comprehensive view of the effects of p23 on estrogen receptor (ER)-dependent and -independent gene expression, we profiled mRNA expression from control versus MCF-7+p23 cells in the absence and presence of estrogen. A number of p23-sensitive target genes involved in metastasis and drug resistance were identified. Most striking is that many of these genes are also misregulated in invasive breast cancers, including PMP22, ABCC3, AGR2, Sox3, TM4SF1, and p8 (NUPR1). Upregulation of the ATP-dependent transporter ABCC3 by p23 conferred resistance to the chemotherapeutic agents etoposide and doxorubicin in MCF-7+p23 cells. MCF-7+p23 cells also displayed higher levels of activated Akt and an expanded phosphoproteome relative to control cells, suggesting that elevated p23 also enhances cytoplasmic signaling pathways. For breast cancer patients, tumor stage together with high cytoplasmic p23 expression more accurately predicted disease recurrence and mortality than did stage alone. High nuclear p23 was found to be associated with high cytoplasmic p23, therefore both may promote tumor progression and poor prognosis by increasing metastatic potential and drug resistance in breast cancer patients. Cancer Res; 70(21); 8446–56. ©2010 AACR.

p23 is a ubiquitous and evolutionarily conserved protein first characterized as a component of unliganded progesterone receptor complexes (1, 2). p23 acts as a cochaperone, stabilizing the ATP-bound form of heat shock protein 90 (Hsp90), thereby stabilizing Hsp90 interactions with steroid receptors (3) and assisting Hsp90 in increasing the hormone-binding affinities of many steroid receptors (46). p23 is required for survival in mammals because p23-null mice die perinatally and fail to undergo proper lung development due in part to decreased activity of the glucocorticoid receptor (7). p23 regulates steroid receptors and telomerase recruitment to DNA (811), and stabilizes some kinases, including PKR, FAK-2, Flt3, and Raf (1215). In addition to Hsp90-dependent events, p23 might also regulate client proteins autonomously via its innate chaperone activity (16). There is also evidence that mammalian p23 is a cytoplasmic prostaglandin E synthetase (17), although this function is controversial because p23 knockout mice do not exhibit reduced prostaglandin E synthetase activity (18).

p23 has many implicated roles in cancer. We have previously shown that p23 overexpression increases ligand binding to the estrogen receptor α (ER) and enhances ER-dependent transcriptional activation (19, 20). We have also shown that p23 overexpression in human ER-positive MCF-7 breast cancer cells promotes a transition from noninvasive into migratory and invasive cells, without affecting cell proliferation (10). Mechanistically, p23 selectively increases ER-target gene transcription and ER recruitment to the estrogen response element (ERE) of the metastasis-associated gene TFF1 (pS2), rather than to that of proliferative genes like c-myc (10).

p23 protein expression increases with breast tumor stage (10) and is upregulated in metastatic cancers (21, 22). p23 also regulates the binding and activity of the oncoprotein telomerase (11, 23, 24), known to impact both early and late stages of tumorigenesis, including epithelial cell transformation (25), human mammary epithelial cell immortalization (26), and metastasis in ER-negative breast cancer cells (27, 28). In addition, p23 desensitizes mammalian cells to treatment with the Hsp90 inhibitor and chemotherapeutic agent geldanamycin, thereby protecting Hsp90 from geldanamycin inhibition and stabilizing Hsp90 client oncoproteins (15).

Given the implicated role of p23 in cancer and the fact that a modest increase in p23 protein expression causes such strong estrogen-dependent and -independent transcriptional and functional effects on migration and invasion in MCF-7 cells (10), we further characterized the effect of p23 on global gene expression in MCF-7 cells, and examined the expression of p23 protein in primary tumor samples with appended outcome data from women with breast cancer.

Cell lines

MCF-7 control and p23-overexpressing MCF-7 cells (MCF-7+p23) were cultured and hormone-starved as previously described (10). A single MCF-7+p23 overexpressing clone (clone 7) was used in the experiments. Validation of selected p23-sensitive genes was performed using an independent, previously described MCF-7+p23 clone (clone 8) with similar results (Supplementary Fig. S1; ref. 10).

Microarray analysis

Total RNA was isolated using the RNeasy kit (Qiagen) from MCF-7 control and MCF-7+p23 cells treated 16 hours with either ethanol vehicle or 1 nmol/L 17-β-estradiol (E2). Total RNA was reverse transcribed into cDNA and in vitro transcribed to cRNA. The cRNA was fragmented, labeled, hybridized to the human U133A 2.0 Affymetrix GeneChip, washed, and scanned. Data are representative of four different conditions performed in duplicate (a total of eight) and normalized using Robust Multichip Average Express. The primary data were analyzed using MultiExperiment Viewer to determine hierarchical clustering, and the ANOVA test was performed on each of the conditions to find those probe sets that were considered significant (P < 0.05). Genes were considered significant if they passed the ANOVA test and displayed a fold-change of >2.0 when the logarithmic values in MCF-7+p23 and control cells were compared. The L2L online microarray data analysis tool (http://depts.washington.edu/l2l/) was used to analyze the gene sets for gene ontology.

RNA isolation and quantitative real-time PCR

Cells were incubated in the absence (ethanol vehicle) or presence of 1 nmol/L E2 or 100 nmol/L 4-hydroxy-tamoxifen (TAM) for 16 hours before harvesting. Total RNA from MCF-7 control and MCF-7+p23 cells was extracted with TRIzol (Invitrogen) as described by the manufacturer. cDNA was synthesized from 1 μg of RNA using the Superscript III reverse transcriptase (Invitrogen) and random primer hexamers (Amersham Biosciences) following the manufacturer's instructions. cDNAs were amplified with the SYBR Green Taq Ready Mix (USB) or Fast Start Universal SYBR Master Mix (Roche) using MyiQ Single-Color Real-Time PCR Detection System from Bio-Rad. The primers used for quantitative real-time PCR are listed in Supplementary Table S1.

Chromatin immunoprecipitation experiments

Cells were hormone-starved for 3 days and treated with the ethanol vehicle or 10 nmol/L E2 for 45 minutes and then formaldehyde cross-linked, lysed, washed, and pelleted prior to sonication according to methods described (10). Sonication of chromatin, preparation of protein A beads, immunoprecipitation of chromatin, and semiquantitative PCR were modified from methods described (29). Chromatin (125 μg) was subjected to immunoprecipitation with 6 μg ER antibody (HC20, Santa Cruz Biotechnology), 2 μg acetyl-histone H3-K9/14 antibody (Millipore, Inc.), or equivalent amount of rabbit IgG (I-8140, Sigma). Following cross-link reversal, DNA was purified using Qiagen PCR purification kit. Primers used for PCR are listed in Supplementary Table S2.

RNA interference

ON-TARGET plus Non-Targeting (D-001810-10-05), ER (L-003401-00-0005), and ABCC3 (L-007312-00-0005) SMART pool siRNA were purchased from Thermo Scientific. Prior to the siER and siABCC3 RNA transfections, cells were hormone-starved for 1 and 3 days, respectively. Cells were transfected using oligofectamine (Invitrogen) according to the manufacturer's instructions. For ER RNA interference (RNAi) experiments, RNA and protein were isolated 36 hours after the transfection. mRNA expression levels were determined by quantitative real-time PCR, and protein levels were analyzed following extraction from cells using radioimmunoprecipitation assay buffer (10) and immunoblotting using antibodies for ER (1:1,000; HC20, Santa Cruz Biotechnology), α-tubulin (1:2,000; MMS-489P, Covance), and p23 (1:2,000; JJ3; MA3-414, Thermo Fisher). For ABCC3 RNAi experiments, 24 hours posttransfection cells were dissociated using Versene-EDTA and plated at the same density. Two days after plating, cells were treated for 24 hours with 100 ng/mL doxorubicin. Percent survival was determined by cell counting wells in each condition. RNA was extracted from the untreated cells and knockdown of ABCC3 mRNA expression was confirmed.

Chemotherapeutic resistance assays

Cells were hormone-starved for three days. Media were supplemented with 0.01 to 1 μg/mL doxorubicin hydrochloride (Adriamycin; D1515, Sigma), 0.02 to 20 μmol/L etoposide (E1383, Sigma), or 0.04 to 40 μmol/L camptothecin (C9911, Sigma) for 48 hours and percent survival was determined in quadruplicate for each condition using the sulphorhodamine B assay (30).

Immunohistochemistry and tissue microarray analysis

Tissue arrays contained tissue specimens from individuals who had undergone surgical removal of breast tumors at Ludwig Cancer Center in São Paulo, Brazil, between 1983 and 1993 with a 12-year follow up. Representative tissue cores 0.6 mm in diameter were extracted from each specimen and mounted in paraffin blocks and stained for p23 (JJ3; MA3-414, Thermo Fisher). Two board-certified pathologists scored the specimens from the breast tissue microarray in a blinded fashion. Each was scored for intensity of staining (0–3) in the nucleus and in the cytoplasm. For intensity scores, no staining above background was scored as 0, weak but convincing staining as 1, moderate staining as 2, and strong staining as 3. Spearman correlation coefficients were used to determine associations between variables. Age and disease characteristics were compared between cytoplasmic p23 levels using χ2 tests, t-tests, and Fisher's exact test. Univariate and multivariable Cox proportional hazards models were developed to predict recurrence or mortality. Models were selected using stepwise procedures in SAS with differences in the likelihood ratio test statistics as the criteria for model selection.

Higher p23 protein levels cause changes in gene expression in MCF-7 cells that are also observed in invasive breast cancers

Our previous study showed that a small increase in the amount of p23 protein expression caused specific estrogen-dependent and -independent changes in MCF-7 cell gene expression and function favoring tumor progression (10). We profiled genes globally regulated by p23 using Affymetrix microarray technology (Fig. 1A) on RNA extracted from hormone-starved control or MCF-7 cells stably expressing 2-fold more p23 protein (MCF-7+p23; ref. 10). The analysis identified 359 known human genes sensitive to p23 levels in the absence of estrogen treatment. Of these genes, 179 were upregulated and 180 were downregulated by >2-fold (genes and corresponding probe sets are listed in Supplementary Tables S5 and S6).

Figure 1.

The effects of p23 overexpression on gene expression in MCF-7 cells. A, depiction of the procedure used to examine p23-responsive genes. WT, wild type. B, microarray heat map and fold-change of six genes misregulated in invasive breast cancers with >5-fold change in expression between MCF-7 control and MCF-7+p23 cells in the absence of E2 treatment and verified by quantitative real-time PCR (C). Data are means of at least three independent experiments, normalized to GAPDH, and presented as relative fold-change (RFC) from that of MCF-7 control cells, set to 1. Error bars, SE. D, clusters of genes affected by p23 overexpression in the absence or presence of E2 treatment. Heat map reveals clusters of genes rendered unresponsive (clusters A1, A2, yellow bar) or responsive (cluster B, red bar) to E2 by p23 overexpression. Cluster C (green bar), genes inversely regulated by p23 and E2. Also listed are genes misregulated in MCF-7+p23 cells involved in migration and invasion, including CADM1, LAMB1, and MTSS1.

Figure 1.

The effects of p23 overexpression on gene expression in MCF-7 cells. A, depiction of the procedure used to examine p23-responsive genes. WT, wild type. B, microarray heat map and fold-change of six genes misregulated in invasive breast cancers with >5-fold change in expression between MCF-7 control and MCF-7+p23 cells in the absence of E2 treatment and verified by quantitative real-time PCR (C). Data are means of at least three independent experiments, normalized to GAPDH, and presented as relative fold-change (RFC) from that of MCF-7 control cells, set to 1. Error bars, SE. D, clusters of genes affected by p23 overexpression in the absence or presence of E2 treatment. Heat map reveals clusters of genes rendered unresponsive (clusters A1, A2, yellow bar) or responsive (cluster B, red bar) to E2 by p23 overexpression. Cluster C (green bar), genes inversely regulated by p23 and E2. Also listed are genes misregulated in MCF-7+p23 cells involved in migration and invasion, including CADM1, LAMB1, and MTSS1.

Close modal

In fact, 50% of the genes most upregulated and 61% of the genes most downregulated by p23 overexpression (i.e., those genes exhibiting >5-fold changes in expression between MCF-7-control and MCF-7+p23 cells in the microarray) were also misregulated in invasive breast cancers (Supplementary Fig. S2). Many of these genes were validated, including PMP22, AGR2, and ABCC3 (upregulated in the microarray 23.7-fold, 5.2-fold, and 5.1-fold, respectively, in MCF-7+p23 cells) and p8/NUPR1, TM4SF1, and Sox3 (downregulated 14.2-fold, 9.2-fold, and 7.9-fold, respectively) from independent MCF-7+p23 lines (Fig. 1B and Supplementary Fig. S1). PMP22 is an integral membrane protein known to play a role in cell-cell and cell-matrix interactions in multiple cell types, and is implicated in hereditary human neuropathies such as Charcot-Marie-Tooth syndrome (31). ABCC3 (MRP3) is an ABC transporter with specificity for cellular export of etoposides and antifolates, both of which are used as chemotherapeutic agents (32). AGR2 is a secretory protein, p53 inhibitor, and late-stage breast cancer marker (33, 34). As in MCF-7+p23 cells, PMP22, ABCC3, and AGR2 have also been shown to be upregulated in advanced cancers (3235). Downregulation of p8, TM4SF1, and Sox3, similar to what is observed in MCF-7+p23 cells, is also associated with breast cancer progression (36, 37). p8 is a nuclear phosphoprotein, TM4SF1 is a tetraspanin family member implicated in cell motility, and studies in Xenopus allude to Sox3 as a tumor suppressor through its ability to bind to β-catenin and inhibit T-cell factor signaling (38).

The gene expression changes observed in the microarray were validated by quantitative real-time PCR. PMP22, ABCC3, and AGR2 were upregulated 33-fold, 3.5-fold, and 1.5-fold, respectively, in MCF-7+p23 cells compared with control cells. p8, TM4SF1, and Sox3 were reduced by 8.6-fold, 30.8-fold, and 6.1-fold, respectively (Fig. 1C). Interestingly, genes upregulated by p23 overexpression also displayed varying degrees of dependency on Hsp90. As expected for a gene dependent upon Hsp90-p23 complexes, disruption of Hsp90-p23 interactions in MCF-7+p23 cells with either geldanamycin or 17-AAG, the geldanamycin analogue with lower toxicity and used in chemotherapeutic regimens, decreased expression of ABCC3 (Supplementary Fig. S3). The same held true for pS2 and c-myc, both well characterized ER target genes. By contrast, expression of AGR2 was not reduced, but rather was enhanced upon inhibition of Hsp90. This suggests that an Hsp90-sensitive pathway may repress AGR2 expression. Surprisingly, geldanamycin or 17-AAG treatments had no effect on the expression of PMP22, despite disruption of Hsp90-p23 interaction by geldanamycin (Supplementary Fig. S3). Thus, the effect of p23 on PMP22 gene expression seems Hsp90 independent. Collectively, these data show that p23 protein levels can modulate the expression of genes misregulated in advanced breast cancers and that certain p23-sensitive changes in gene expression are Hsp90 independent.

We also found that 357 genes were sensitive to the ER agonist E2 in MCF-7 cells, many of which became more or less responsive to E2 by p23 overexpression (Fig. 1D, Supplementary Fig. S4, Supplementary Table S7). Heat maps revealed functional similarities between these genes. For example, cluster A (Fig. 1D, yellow bar) represents genes that changed from E2 responsive to E2 unresponsive as a result of p23 overexpression and can be divided into two subclusters based on function. In cluster A1, 23% of genes are involved in transcriptional regulation, including Sox11, and in cluster A2, 61% of genes are involved in migration and cell adhesion. Cluster B (red bar) represents genes involved in a range of biological processes that changed from E2 unresponsive to E2 responsive. Cluster C (green bar) represents a group of genes also involved in a variety of biological processes that are inversely regulated by p23 and E2, including ion transport (16%) and regulation of transcription (13%). Although they did not fall into specific clusters, other genes characteristically involved in migration and invasion are also sensitive to p23 overexpression, and include cell adhesion molecule 1 (CADM1), laminin B1 (LAMB1), and metastasis suppressor 1 (MTSS1; Fig. 1B).

Gene ontology of the genes induced by p23 revealed that p23 overexpression particularly affects the expression of genes involved in metabolism (Supplementary Table S3) with the highest significance attributed to kynureninase (KYNU), an enzyme involved in tryptophan metabolism. KYNU is also upregulated in ER-negative breast cancer cells that have metastasized to the lung (39), further supporting a role of p23 in promoting tumor progression through altering gene expression patterns.

Higher p23 protein levels result in increased histone H3 acetylation at PMP22 and ABCC3 gene promoters

To test whether the mechanism of the p23-mediated increase in PMP22 or ABCC3 gene expression is through activation of transcription at corresponding promoters, chromatin immunoprecipitation (ChIP) experiments were performed to detect acetylation of histone H3 lysines 9 and 14 (H3-AcK9/14), marks of transcriptional activation. The PMP22 locus encodes two transcripts that contain alternative exons 1A and 1B (Fig. 2A). Because the 1B transcript is predominantly expressed in nonneuronal cells (40), we predicted that increases in histone acetylation at the PMP22 locus in MCF-7+p23 cells relative to control cells would be most likely to occur at the promoter of 1B (P2). Histone H3-AcK9/14 levels relative to input were higher at the PMP22 P2 in MCF-7+p23 cells than in control cells (Fig. 2B and C, top). A similar, albeit less robust increase in histone H3-AcK9/14 was also observed at the ABCC3 promoter (Fig. 2B and C, bottom). The ABCC3 promoter had basal levels of histone acetylation, which could be detected above background (nonspecific IgG) levels in control cells (Fig. 2B). This suggests that a mechanism of p23-mediated transcriptional activation is through increased recruitment of chromatin-modifying enzymes that hyperacetylate histone H3-K9/14 and reinitiate gene transcription from active promoters.

Figure 2.

Increased histone H3 acetylation at genes upregulated by p23 overexpression. A, schematic of PMP22 and ABCC3 loci. Arrows, transcriptional start sites (TSS); white boxes, exons; black bars, DNA promoter regions amplified by PCR to determine the level of histone H3-AcK9/K14. ChIP assays were performed using specific antibodies against histone H3-AcK9/14 or negative control IgG. Relative levels of histone H3-AcK9/K14 were determined using both semiquantitative (B) and quantitative real-time PCR (C). Data are averages from three independent experiments, normalized to input, and presented as relative fold-enrichment of histone H3-AcK9/K14 at corresponding promoters of MCF+p23 over MCF-7 control cells, set to 1. Error bars, SD.

Figure 2.

Increased histone H3 acetylation at genes upregulated by p23 overexpression. A, schematic of PMP22 and ABCC3 loci. Arrows, transcriptional start sites (TSS); white boxes, exons; black bars, DNA promoter regions amplified by PCR to determine the level of histone H3-AcK9/K14. ChIP assays were performed using specific antibodies against histone H3-AcK9/14 or negative control IgG. Relative levels of histone H3-AcK9/K14 were determined using both semiquantitative (B) and quantitative real-time PCR (C). Data are averages from three independent experiments, normalized to input, and presented as relative fold-enrichment of histone H3-AcK9/K14 at corresponding promoters of MCF+p23 over MCF-7 control cells, set to 1. Error bars, SD.

Close modal

Downregulation of PMP22 and ABCC3 gene expression by E2 is associated with ER recruitment to distal EREs

Our microarray data revealed that many genes positively regulated by p23 overexpression (e.g., PMP22 and ABCC3) are negatively regulated by E2 treatment. To further explore the E2-mediated repression of genes upregulated by p23, we treated MCF-7 control and MCF-7+p23 cells with E2 or the selective estrogen receptor modulator TAM and measured mRNA expression of PMP22 and ABCC3. E2 treatment repressed PMP22 and ABCC3 gene expression roughly 2-fold in both control and MCF-7+p23 cells (Fig. 3A and B, left). Moreover, when comparing total expression levels of PMP22 and ABCC3 between untreated control cells and E2-treated MCF-7+p23 cells, it is evident that E2 repression could not fully reverse the activation of these genes elicited by p23 overexpression. Interestingly, treatment with TAM increased expression of PMP22 and ABCC3 in both control and MCF-7+p23 cells (Fig. 3A and B, right). Thus, p23 overexpression furthers induces the expression of these genes in the presence of TAM, whereas activation of PMP22 and ABCC3 by p23 overexpression is partially reversed by E2.

Figure 3.

ER recruitment to distal EREs is associated with repression of PMP22 and ABCC3 gene expression. MCF-7 control (white bars) and MCF-7+p23 cells (black bars) were hormone-starved for 3 days, treated for 16 hours with 1 nmol/L E2 (A and B, left) or 100 nmol/L TAM (A and B, right), and mRNA was analyzed by quantitative real-time PCR. Data are means from six independent experiments and presented as in Fig. 1C. C, schematic of PMP22 and ABCC3 loci containing transcription start sites (arrows), ER binding elements as determined by global ChIP-Chip analysis (ERE, striped boxes), and control upstream regions (UPS, black boxes). Black bars, regions amplified by PCR to determine the level of ER recruitment. D, the relative amount of ER binding, in the presence or absence of E2, was determined by ChIP, followed by quantitative real-time PCR. Data are shown as the averages, with SE between replicate experiments, normalized to input for each sample, and expressed relative to ER recruitment at the corresponding UPS in MCF-7 control cells, set to 1. E, depletion of ER protein by RNAi using ER (siER) and control siRNAs (siControl) verified by immunoblotting. WB, Western blot. F, ABCC3 and PMP22 gene expression levels upon depletion of ER were obtained by quantitative real-time PCR. Shown are the averages obtained from three independent experiments normalized to GAPDH and presented as relative fold-change (RFC) between siER (+) and siControl (-) transfected cells, with siControl cells set to 1.

Figure 3.

ER recruitment to distal EREs is associated with repression of PMP22 and ABCC3 gene expression. MCF-7 control (white bars) and MCF-7+p23 cells (black bars) were hormone-starved for 3 days, treated for 16 hours with 1 nmol/L E2 (A and B, left) or 100 nmol/L TAM (A and B, right), and mRNA was analyzed by quantitative real-time PCR. Data are means from six independent experiments and presented as in Fig. 1C. C, schematic of PMP22 and ABCC3 loci containing transcription start sites (arrows), ER binding elements as determined by global ChIP-Chip analysis (ERE, striped boxes), and control upstream regions (UPS, black boxes). Black bars, regions amplified by PCR to determine the level of ER recruitment. D, the relative amount of ER binding, in the presence or absence of E2, was determined by ChIP, followed by quantitative real-time PCR. Data are shown as the averages, with SE between replicate experiments, normalized to input for each sample, and expressed relative to ER recruitment at the corresponding UPS in MCF-7 control cells, set to 1. E, depletion of ER protein by RNAi using ER (siER) and control siRNAs (siControl) verified by immunoblotting. WB, Western blot. F, ABCC3 and PMP22 gene expression levels upon depletion of ER were obtained by quantitative real-time PCR. Shown are the averages obtained from three independent experiments normalized to GAPDH and presented as relative fold-change (RFC) between siER (+) and siControl (-) transfected cells, with siControl cells set to 1.

Close modal

Recent data suggest that repression by ER requires E2-dependent recruitment of the receptor to the EREs of the target gene (41). Therefore, we examined ER recruitment to PMP22 and ABCC3 EREs predicted from ChIP-on-Chip data (42), as well as to regions upstream of PMP22 and ABCC3 promoters (UPS) that were not predicted to bind ER (Fig. 3C). In MCF-7+p23 cells, E2-dependent ER recruitment to PMP22 and ABCC3 EREs is enriched relative to control cells (Fig. 3D, compare black with dark grey bars, and Supplementary Fig. S5, compare differences between columns 14 and 6 relative to respective inputs), but not to the PMP22 or ABCC3 UPS regions (Fig. 3D and Supplementary Fig. S5). In fact, no ER binding could be detected above nonspecific IgG levels at PMP22 and ABCC3 UPS regions (Supplementary Fig. S5), indicating both as suitable negative control regions for ER binding. Thus, p23 overexpression increases E2-dependent ER recruitment specifically to the EREs of the E2-repressed genes PMP22 and ABCC3. This trend of increased ER recruitment upon p23 overexpression to PMP22 and ABCC3 EREs is similar to what has been observed previously at the pS2 ERE (10). However, for pS2, ER recruitment is associated with gene activation, rather than repression.

Interestingly, ER recruitment was evident at the PMP22 ERE and more so at the ABCC3 ERE, even in the absence of exogenously added E2 in MCF-7+p23 cells compared with control cells (Fig. 3D, compare light grey with white bars, and Supplementary Fig. S5, compare differences between columns 10 and 2 relative to respective inputs). This suggests that ER may also function as a repressor of PMP22 and ABCC3 gene expression under basal conditions when E2 levels are very low. To assess whether ER represses “basal” PMP22 and ABCC3 gene expression, we silenced ER expression using siRNA in both MCF-7 control and MCF-7+p23 cells and measured ABCC3 and PMP22 mRNA levels in the absence of exogenous E2. Loss of ER resulted in a significant induction of ABCC3 in both MCF-7 control and MCF-7+p23 cells under such basal conditions (Fig. 3E and F). Only a modest effect was observed on PMP22 gene expression. Therefore, ER represses PMP22 and ABCC3 gene expression under basal and high E2 situations.

Upregulation of ABCC3 in MCF-7+p23 cells imparts selective chemotherapeutic resistance

Because MCF-7+p23 cells express more ABCC3 mRNA (Fig. 1B–D) compared with control cells, and ABCC3 overexpression confers multidrug resistance by exporting from cells chemotherapeutic agents including etoposide (43) and doxorubicin (Adriamycin; ref. 44), we suspected that MCF-7+p23 cells would be more resistant than control cells to these chemotherapeutic agents. Using a sulphorhodamine B dye to stain for surviving cells following treatment with chemotherapeutic agents, MCF-7+p23 cells survived better than control cells in the presence of etoposide and doxorubicin (Fig. 4A and C). This resistance was selective in that MCF-7+p23 cells were no more resistant than control cells to camptothecin, a chemotherapeutic agent not exported by ABCC3 (Fig. 4B; ref. 43). Resistance to etoposide and doxorubicin, but not camptothecin, was validated using clonogenic assays (data not shown). To assess the requirement for ABCC3 in doxorubicin resistance, because it is commonly used to treat metastatic and recurrent breast cancers, we silenced the expression of ABCC3 using siRNA and measured sensitivity of MCF-7+p23 cells to doxorubicin. Our analysis revealed that MCF-7+p23 cells are nearly 50% more sensitive to doxorubicin when ABCC3 expression is reduced by 80% (Fig. 4D), indicating that doxorubicin resistance is mediated in part by ABCC3. Lack of complete sensitization can be attributed to other multidrug resistance proteins capable of exporting doxorubicin (43).

Figure 4.

Increased ABCC3 gene expression in MCF-7+p23 cells mediates doxorubicin resistance. A, MCF-7 control (blue) and MCF-7+p23 cells (pink) cells were treated with the indicated concentrations of etoposide, camptothecin (Cpt; B); or doxorubicin (C). Percent survival upon increasing concentrations of drugs was determined using the sulphorhodamine B dye, which coats cells evenly so optical density (OD) values are proportional to cell number. Data are expressed as averages of at least three independent experiments with untreated cells set at 100% survival. Error bars, SE. D, MCF-7+p23 cells were transfected with siRNA against ABCC3 (siABCC3) or a control siRNA (siControl), and percent survival of doxorubicin treated relative to untreated cells is shown (left). Data from quadruplicate points were averaged from two independent experiments; error bars, SE. ABCC3 mRNA levels from siABCC3 and siControls were measured by quantitative real-time PCR (right).

Figure 4.

Increased ABCC3 gene expression in MCF-7+p23 cells mediates doxorubicin resistance. A, MCF-7 control (blue) and MCF-7+p23 cells (pink) cells were treated with the indicated concentrations of etoposide, camptothecin (Cpt; B); or doxorubicin (C). Percent survival upon increasing concentrations of drugs was determined using the sulphorhodamine B dye, which coats cells evenly so optical density (OD) values are proportional to cell number. Data are expressed as averages of at least three independent experiments with untreated cells set at 100% survival. Error bars, SE. D, MCF-7+p23 cells were transfected with siRNA against ABCC3 (siABCC3) or a control siRNA (siControl), and percent survival of doxorubicin treated relative to untreated cells is shown (left). Data from quadruplicate points were averaged from two independent experiments; error bars, SE. ABCC3 mRNA levels from siABCC3 and siControls were measured by quantitative real-time PCR (right).

Close modal

High p23 expression correlates with shorter disease-free survival times in breast cancer patients

Because breast cancer cells that express more p23 have a higher invasive potential and are more drug resistant, we suspected that patients expressing high levels of p23 protein would have shorter periods of disease-free survival. We analyzed 213 specimens from a human breast tissue microarray for intensity of p23 staining in the nucleus and cytoplasm on a scale of 0 to 3, with 3 as the highest intensity staining (Supplementary Fig. S6).

Using univariate analysis we determined associations between cytoplasmic p23, nuclear p23, stage, and tumor size (T), nodal involvement (N), and metastasis (M; TNM status). We found that cytoplasmic p23 is significantly associated with nuclear p23 expression (Spearman correlation ρ = 0.37), as well as with stage (ρ = 0.23) and nodal involvement (ρ = 0.26), but does not correlate significantly with tumor size (ρ = 0.16) or metastasis (ρ = 0.12; Supplementary Table S4).

High cytoplasmic p23 (3+) is also associated with later-stage tumors (68% are stage III/IV) compared with moderate p23 (2+) expression (36% are stage III/IV; χ2, P < 0.01; Table 1). Interestingly 73% of the high cytoplasmic p23–expressing tumors displayed nodal involvement, whereas only 41% of the moderate cytoplasmic p23–expressing cancers are node positive (χ2, P < 0.01; Table 1). This is consistent with our observation that MCF-7+p23 cells are more migratory and invasive than control MCF-7 cells (10).

Table 1.

p23 expression in relation to stage and TNM from the breast tumor microarray

Samples (n = 213)Cytoplasm p23P
1+ (n = 2)2+ (n = 22)3+ (n = 189)(Tests 1+ & 2+ vs. 3+)
Age, years 
    Mean ± SD 46.0 ± 42.24 50.91 ± 10.27 55.12 ± 13.21 0.10* 
Nuclear p23 
    0 2 (100%) 4 (18%) 2 (1%) <0.0001 
    1+ 10 (45%) 33 (17%) 
    2+ 7 (32%) 69 (37%) 
    3+ 1 (5%) 85 (45%) 
Stage 
    I 3 (14%) 5 (3%) <0.01 
    II 1 (50%) 11 (50%) 55 (29%) 
    III 1 (50%) 8 (36%) 108 (57%) 
    IV 21 (11%) 
    1 3 (14%) 7 (4%) 0.09 
    2 1 (50%) 9 (43%) 54 (31%) 
    3 1 (50%) 3 (14%) 30 (17%) 
    4 6 (29%) 81 (47%) 
    Missing  17 
    0 1 (50%) 13 (59%) 50 (27%) <0.01 
    1 1 (50%) 6 (27%) 46 (24%) 
    2 3 (14%) 52 (28%) 
    3 40 (21%) 
    Missing   
    0 2 (100%) 22 (100%) 168 (89%) 0.14 
    1 21 (11%) 
Samples (n = 213)Cytoplasm p23P
1+ (n = 2)2+ (n = 22)3+ (n = 189)(Tests 1+ & 2+ vs. 3+)
Age, years 
    Mean ± SD 46.0 ± 42.24 50.91 ± 10.27 55.12 ± 13.21 0.10* 
Nuclear p23 
    0 2 (100%) 4 (18%) 2 (1%) <0.0001 
    1+ 10 (45%) 33 (17%) 
    2+ 7 (32%) 69 (37%) 
    3+ 1 (5%) 85 (45%) 
Stage 
    I 3 (14%) 5 (3%) <0.01 
    II 1 (50%) 11 (50%) 55 (29%) 
    III 1 (50%) 8 (36%) 108 (57%) 
    IV 21 (11%) 
    1 3 (14%) 7 (4%) 0.09 
    2 1 (50%) 9 (43%) 54 (31%) 
    3 1 (50%) 3 (14%) 30 (17%) 
    4 6 (29%) 81 (47%) 
    Missing  17 
    0 1 (50%) 13 (59%) 50 (27%) <0.01 
    1 1 (50%) 6 (27%) 46 (24%) 
    2 3 (14%) 52 (28%) 
    3 40 (21%) 
    Missing   
    0 2 (100%) 22 (100%) 168 (89%) 0.14 
    1 21 (11%) 

*t-test.

Fisher's exact test.

χ2 test.

We used Cox proportional hazards models for each variable individually to examine the association of that variable with recurrence or death (Table 2). High levels of cytoplasmic p23, stage III/IV disease, and high T, N, and M were all individually associated with increased recurrence and death. When these variables were considered jointly (multivariable), only cytoplasmic p23 expression and stage were associated with disease recurrence or death (from Cox models). Our findings indicate that individuals with stage I/II disease and high cytoplasmic p23 (3+) expression (stage I/II-p23high) are 2.35 times more likely to have recurrent disease or death than the same stage patient with low p23 (1+ or 2+) expression (stage I/II-p23low; Table 2 and Fig. 5). Moreover, patients with stage III/IV disease and cytoplasmic p23 1+ or 2+ expression (stage III/IV-p23low), or stage III/IV disease and cytoplasmic p23 3+ expression (stage III/IV-p23high) are 3.34 and 7.85 times more likely to have disease recurrence than those stage I/II-p23low individuals (Table 2 and Fig. 5). At 150 months postsurgery, the cumulative disease-free survival for patients with stage I/II-p23low tumors is estimated at 80%; the corresponding estimates are 50% of individuals with stage I/II-p23high, 44% with stage III/IV-p23low, and 18% with stage III/IV-p23high tumors (Fig. 5). Although stage alone is highly associated with recurrence or death (likelihood ratio = 45.77), the addition of cytoplasmic p23 expression to the Cox proportional hazard models significantly improves the model fit (likelihood ratio = 52.15). In summary, individuals with high levels of cytoplasmic p23 and stage III/IV disease are more likely to exhibit lymph node metastases and experience disease recurrence and mortality.

Table 2.

Disease recurrence as function of p23 expression and stage using a Cox proportional hazard model

UnivariateMultivariable
Hazard ratio (95% CI)PHazard ratio (95% CI)P
Age 1.01 (0.99-1.02) 0.16   
Cytoplasm p23 
    1+, 2+ 1.00 <0.01 1.00 0.02 
    3+ 2.97 (1.46-6.06) <0.01 2.35 (1.15-4.82)  
Nuclear p23 
    0, 1+ 1.00 0.17   
    2+, 3+ 1.33 (0.88-2.01)    
Stage 
    I, II 1.00 <0.0001 1.00 <0.0001 
    III, IV 3.47 (2.34-5.15)  3.34 (2.22-5.03)  
    1 1.00    
    2 7.10 (0.97-51.75) 0.05   
    3 8.04 (1.08-59.81) 0.04   
    4 17.38 (2.41-125.20) <0.01   
    0 1.00    
    1 1.52 (0.91-2.53) 0.11   
    2 6.81 (4.25-10.90) <0.0001   
    3 4.04 (2.43-6.72) <0.0001   
    0 1.00 <0.0001   
    1 11.41 (6.76-19.27)    
UnivariateMultivariable
Hazard ratio (95% CI)PHazard ratio (95% CI)P
Age 1.01 (0.99-1.02) 0.16   
Cytoplasm p23 
    1+, 2+ 1.00 <0.01 1.00 0.02 
    3+ 2.97 (1.46-6.06) <0.01 2.35 (1.15-4.82)  
Nuclear p23 
    0, 1+ 1.00 0.17   
    2+, 3+ 1.33 (0.88-2.01)    
Stage 
    I, II 1.00 <0.0001 1.00 <0.0001 
    III, IV 3.47 (2.34-5.15)  3.34 (2.22-5.03)  
    1 1.00    
    2 7.10 (0.97-51.75) 0.05   
    3 8.04 (1.08-59.81) 0.04   
    4 17.38 (2.41-125.20) <0.01   
    0 1.00    
    1 1.52 (0.91-2.53) 0.11   
    2 6.81 (4.25-10.90) <0.0001   
    3 4.04 (2.43-6.72) <0.0001   
    0 1.00 <0.0001   
    1 11.41 (6.76-19.27)    

Abbreviation: 95% CI, 95% confidence interval.

Figure 5.

p23 expression and clinical outcome in human breast cancer. Kaplan-Meier curves plotting time to recurrence (in months) versus survival probability are shown for patients with stage I/II disease and low p23 (red solid line), stage I/II disease and high p23 (blue dotted line), stage III/IV disease and low p23 (green dotted line), or stage III/IV disease and high p23 (purple dotted line). Hazard ratios were calculated using multivariable Cox proportional hazards models and are shown for patients classified according to stage (I/II or III/IV) and p23 expression levels (low, 1+ or 2+; high, 3+).

Figure 5.

p23 expression and clinical outcome in human breast cancer. Kaplan-Meier curves plotting time to recurrence (in months) versus survival probability are shown for patients with stage I/II disease and low p23 (red solid line), stage I/II disease and high p23 (blue dotted line), stage III/IV disease and low p23 (green dotted line), or stage III/IV disease and high p23 (purple dotted line). Hazard ratios were calculated using multivariable Cox proportional hazards models and are shown for patients classified according to stage (I/II or III/IV) and p23 expression levels (low, 1+ or 2+; high, 3+).

Close modal

We have determined that many genes affected by p23 overexpression in MCF-7 cells are also misregulated in advanced breast cancers. These genes display varying sensitivities to treatments with Hsp90 inhibitors, TAM, or E2, which could play a role in the responsiveness of tumor cells to antiestrogen treatment or chemotherapy. For example, ABCC3 is upregulated by p23 overexpression alone, is Hsp90 dependent, is further upregulated by TAM treatment, and is downregulated by E2 treatment. Functionally, ABCC3 confers selective resistance to chemotherapeutic agents in MCF-7+p23 cells when compared with control cells, and we observed preliminarily that siRNA-mediated or pharmacologic depletion of ER, using ICI 182,720 treatment, decreases the sensitivity of MCF-7 control cells to doxorubicin (data not shown). Thus, both gene expression data and functional studies show that p23 via ABCC3 may regulate chemotherapeutic resistance acquired in metastatic and recurrent ER-negative breast tumors.

How might p23 selectively affect gene expression? p23 has been suggested to play a role in the binding and clearance of steroid receptors at response elements (810). Our results also indicate that p23 increases recruitment of ER to EREs of E2-repressed genes ABCC3 and PMP22 (Fig. 3) as well as to ER-activated genes pS2 and cathepsin D (10). Thus, certain transcription factors, like steroid receptors, may utilize p23 to efficiently cycle on and off DNA to modulate gene transcription.

We find that “active” histone modifications (H3-AcK9/14) are present at the promoters of p23-sensitive genes (Fig. 2B and C). The fact that these modifications are also present at the promoters of some p23-sensitive genes in MCF-7 control cells where p23 is not overexpressed (Fig. 2B) indicates that p23 may not need to overcome the transcriptional barriers associated with gene silencing to upregulate gene expression. Therefore, a mechanism p23 may use to increase gene expression could be to recruit histone acetyltransferases (HATs) implicated in transcription reinitiation (i.e., GCN5), rather than those involved in transcription initiation (i.e., p300; ref. 45) to raise the transcriptional output from already active promoters.

Indeed, we observed a functional interaction between p23 and the HAT GCN5 by conducting a synthetic lethal screen in yeast, suggesting that p23 affects the activity of HAT pathways in an overlapping fashion with GCN5.9

9Oxelmark and Garabedian, unpublished.

Histone H3-K9/14 is a target for acetylation by GCN5 (39, 46, 47), which further implicates GCN5 as a HAT-mediating hyperaceylation at the promoters of p23-upregulated genes. Interestingly, two metabolic genes regulated by p23, ACSL1 and KYNU, which are upregulated in metastatic and recurrent breast cancers, are also known to regulate the activity of HATs and poly(ADP-ribose) polymerase chromatin remodeling enzymes (39, 46, 47). Therefore, a plausible model for p23-sensitive gene regulation is that p23 may influence the DNA binding dynamics of specific transcription factors to upregulate or downregulate gene expression. p23 may also modulate the recruitment of coactivators necessary to reinitiate transcription from active loci to increase gene expression. Further, amplification of gene regulation in situations of p23 overexpression may occur through the enhancement of p23-sensitive metabolic pathways that affect the activity of chromatin-modifying enzymes. Whether or not p23 is recruiting these factors directly is still under investigation.

p23 localizes to both the nucleus and cytoplasm in MCF-7 cells (Supplementary Fig. S7), and high cytoplasmic p23 correlates with shorter disease-free survival times for breast cancer patients, indicating that p23 can promote tumorigenesis by functioning in multiple cellular compartments. In fact, we observe increased activated Akt1 in the cytoplasm and an expanded phosphoproteome in the cytoplasm and nucleus of MCF-7+p23 compared with control cells, suggesting that p23 overexpression facilitates, either directly or indirectly, enhanced kinase activity and protein phosphorylation (Supplementary Fig. S8). Given that activated Akt is a potent survival signal, our findings also suggest that tumors overexpressing p23 would have a survival advantage (48). Therefore, we propose that cancers with high levels of p23 activate signaling pathways in the cytoplasm, which can also amplify p23-sensitive gene transcription in the nucleus.

Our findings strongly suggest that individuals with breast cancers that display high p23 protein levels are more likely to exhibit lymph node involvement and experience disease recurrence and mortality. This is consistent with our in vitro model that p23 overexpression elicits gene expression and phenotypic changes in MCF-7 cells akin to those that occur as breast tumors become invasive. Given that nodal status is the most significant factor in predicting survival (49), it will be important to find ways to reduce p23 expression to decrease lymph node metastases. Lowering p23 expression should also mitigate p23-mediated chemotherapeutic resistance (etoposide, doxorubicin). Recently, celastrol, the active compound in Thunder of God Vine root extracts used in traditional Chinese medicine, has been shown to bind and inactivate p23 by triggering its oligomerization into fibers (50). Thus, targeting p23 should reduce invasiveness and drug resistance in breast cancer cells and prevent disease recurrence.

No potential conflicts of interest were disclosed.

We thank Naoko Tanese, Keren Imberg, and Chaowei Wu for critically reading the manuscript, and Diego Acosta-Alvear for guidance with the ChIP assays.

Grant Support: Vilcek Foundation (N.E. Simpson), NIH/NCI P30 CA16087 (J.D. Goldberg), DOD Center of Excellence (W81XWH-04-1-0905; R.J. Schneider/J.D. Goldberg), Breast Cancer Research Foundation (R.J. Schneider/J.D. Goldberg), Avon Foundation (R.J. Schneider), NIH/NCRR M01 RR000096 (R.J. Schneider), Philip Morris USA (M.J. Garabedian).

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 18 U.S.C. Section 1734 solely to indicate this fact.

1
Johnson
JL
,
Beito
TG
,
Krco
CJ
,
Toft
DO
. 
Characterization of a novel 23-kilodalton protein of unactive progesterone receptor complexes
.
Mol Cell Biol
1994
;
14
:
1956
63
.
2
Johnson
JL
,
Toft
DO
. 
A novel chaperone complex for steroid receptors involving heat shock proteins, immunophilins, and p23
.
J Biol Chem
1994
;
269
:
24989
93
.
3
McLaughlin
SH
,
Sobott
F
,
Yao
ZP
, et al
. 
The co-chaperone p23 arrests the Hsp90 ATPase cycle to trap client proteins
.
J Mol Biol
2006
;
356
:
746
58
.
4
Bohen
SP
. 
Genetic and biochemical analysis of p23 and ansamycin antibiotics in the function of Hsp90-dependent signaling proteins
.
Mol Cell Biol
1998
;
18
:
3330
9
.
5
Freeman
BC
,
Felts
SJ
,
Toft
DO
,
Yamamoto
KR
. 
The p23 molecular chaperones act at a late step in intracellular receptor action to differentially affect ligand efficacies
.
Genes Dev
2000
;
14
:
422
34
.
6
Holley
SJ
,
Yamamoto
KR
. 
A role for Hsp90 in retinoid receptor signal transduction
.
Mol Biol Cell
1995
;
6
:
1833
42
.
7
Grad
I
,
McKee
TA
,
Ludwig
SM
, et al
. 
The Hsp90 cochaperone p23 is essential for perinatal survival
.
Mol Cell Biol
2006
;
26
:
8976
83
.
8
Freeman
BC
,
Yamamoto
KR
. 
Disassembly of transcriptional regulatory complexes by molecular chaperones
.
Science
2002
;
296
:
2232
5
.
9
Stavreva
DA
,
Muller
WG
,
Hager
GL
,
Smith
CL
,
McNally
JG
. 
Rapid glucocorticoid receptor exchange at a promoter is coupled to transcription and regulated by chaperones and proteasomes
.
Mol Cell Biol
2004
;
24
:
2682
97
.
10
Oxelmark
E
,
Roth
JM
,
Brooks
PC
,
Braunstein
SE
,
Schneider
RJ
,
Garabedian
MJ
. 
The cochaperone p23 differentially regulates estrogen receptor target genes and promotes tumor cell adhesion and invasion
.
Mol Cell Biol
2006
;
26
:
5205
13
.
11
Toogun
OA
,
Zeiger
W
,
Freeman
BC
. 
The p23 molecular chaperone promotes functional telomerase complexes through DNA dissociation
.
Proc Natl Acad Sci U S A
2007
;
104
:
5765
70
.
12
Donze
O
,
Abbas-Terki
T
,
Picard
D
. 
The Hsp90 chaperone complex is both a facilitator and a repressor of the dsRNA-dependent kinase PKR
.
EMBO J
2001
;
20
:
3771
80
.
13
Schmidt
MH
,
Chen
B
,
Randazzo
LM
,
Bogler
O
. 
SETA/CIN85/Ruk and its binding partner AIP1 associate with diverse cytoskeletal elements, including FAKs, and modulate cell adhesion
.
J Cell Sci
2003
;
116
:
2845
55
.
14
Yao
Q
,
Nishiuchi
R
,
Li
Q
,
Kumar
AR
,
Hudson
WA
,
Kersey
JH
. 
FLT3 expressing leukemias are selectively sensitive to inhibitors of the molecular chaperone heat shock protein 90 through destabilization of signal transduction-associated kinases
.
Clin Cancer Res
2003
;
9
:
4483
93
.
15
Forafonov
F
,
Toogun
OA
,
Grad
I
,
Suslova
E
,
Freeman
BC
,
Picard
D
. 
p23/Sba1p protects against Hsp90 inhibitors independently of its intrinsic chaperone activity
.
Mol Cell Biol
2008
;
28
:
3446
56
.
16
Bose
S
,
Weikl
T
,
Bugl
H
,
Buchner
J
. 
Chaperone function of Hsp90-associated proteins
.
Science
1996
;
274
:
1715
7
.
17
Tanioka
T
,
Nakatani
Y
,
Semmyo
N
,
Murakami
M
,
Kudo
I
. 
Molecular identification of cytosolic prostaglandin E2 synthase that is functionally coupled with cyclooxygenase-1 in immediate prostaglandin E2 biosynthesis
.
J Biol Chem
2000
;
275
:
32775
82
.
18
Lovgren
AK
,
Kovarova
M
,
Koller
BH
. 
cPGES/p23 is required for glucocorticoid receptor function and embryonic growth but not prostaglandin E2 synthesis
.
Mol Cell Biol
2007
;
27
:
4416
30
.
19
Knoblauch
R
,
Garabedian
MJ
. 
Role for Hsp90-associated cochaperone p23 in estrogen receptor signal transduction
.
Mol Cell Biol
1999
;
19
:
3748
59
.
20
Oxelmark
E
,
Knoblauch
R
,
Arnal
S
,
Su
LF
,
Schapira
M
,
Garabedian
MJ
. 
Genetic dissection of p23, an Hsp90 cochaperone, reveals a distinct surface involved in estrogen receptor signaling
.
J Biol Chem
2003
;
278
:
36547
55
.
21
Krebs
J
,
Saremaslani
P
,
Caduff
R
. 
ALG-2: a Ca2+ -binding modulator protein involved in cell proliferation and in cell death
.
Biochim Biophys Acta
2002
;
1600
:
68
73
.
22
Mollerup
J
,
Krogh
TN
,
Nielsen
PF
,
Berchtold
MW
. 
Properties of the co-chaperone protein p23 erroneously attributed to ALG-2 (apoptosis-linked gene 2)
.
FEBS Lett
2003
;
555
:
478
82
.
23
Holt
SE
,
Aisner
DL
,
Baur
J
, et al
. 
Functional requirement of p23 and Hsp90 in telomerase complexes
.
Genes Dev
1999
;
13
:
817
26
.
24
Forsythe
HL
,
Jarvis
JL
,
Turner
JW
,
Elmore
LW
,
Holt
SE
. 
Stable association of hsp90 and p23, but not hsp70, with active human telomerase
.
J Biol Chem
2001
;
276
:
15571
4
.
25
Hahn
WC
,
Counter
CM
,
Lundberg
AS
,
Beijersbergen
RL
,
Brooks
MW
,
Weinberg
RA
. 
Creation of human tumour cells with defined genetic elements
.
Nature
1999
;
400
:
464
8
.
26
Stampfer
MR
,
Garbe
J
,
Levine
G
,
Lichtsteiner
S
,
Vasserot
AP
,
Yaswen
P
. 
Expression of the telomerase catalytic subunit, hTERT, induces resistance to transforming growth factor β growth inhibition in p16INK4A(-) human mammary epithelial cells
.
Proc Natl Acad Sci U S A
2001
;
98
:
4498
503
.
27
Hochreiter
AE
,
Xiao
H
,
Goldblatt
EM
, et al
. 
Telomerase template antagonist GRN163L disrupts telomere maintenance, tumor growth, and metastasis of breast cancer
.
Clin Cancer Res
2006
;
12
:
3184
92
.
28
Goldblatt
EM
,
Gentry
ER
,
Fox
MJ
,
Gryaznov
SM
,
Shen
C
,
Herbert
BS
. 
The telomerase template antagonist GRN163L alters MDA-MB-231 breast cancer cell morphology, inhibits growth, and augments the effects of paclitaxel
.
Mol Cancer Ther
2009
;
8
:
2027
35
.
29
Acosta-Alvear
D
,
Zhou
Y
,
Blais
A
, et al
. 
XBP1 controls diverse cell type- and condition-specific transcriptional regulatory networks
.
Mol Cell
2007
;
27
:
53
66
.
30
Houghton
P
,
Fang
R
,
Techatanawat
I
,
Steventon
G
,
Hylands
PJ
,
Lee
CC
. 
The sulphorhodamine (SRB) assay and other approaches to testing plant extracts and derived compounds for activities related to reputed anticancer activity
.
Methods
2007
;
42
:
377
87
.
31
Houlden
H
,
Reilly
MM
. 
Molecular genetics of autosomal-dominant demyelinating Charcot-Marie-Tooth disease
.
Neuromolecular Med
2006
;
8
:
43
62
.
32
Kool
M
,
de Haas
M
,
Scheffer
GL
, et al
. 
Analysis of expression of cMOAT (MRP2), MRP3, MRP4, and MRP5, homologues of the multidrug resistance-associated protein gene (MRP1), in human cancer cell lines
.
Cancer Res
1997
;
57
:
3537
47
.
33
Liu
D
,
Rudland
PS
,
Sibson
DR
,
Platt-Higgins
A
,
Barraclough
R
. 
Human homologue of cement gland protein, a novel metastasis inducer associated with breast carcinomas
.
Cancer Res
2005
;
65
:
3796
805
.
34
Smirnov
DA
,
Zweitzig
DR
,
Foulk
BW
, et al
. 
Global gene expression profiling of circulating tumor cells
.
Cancer Res
2005
;
65
:
4993
7
.
35
Nies
AT
,
Konig
J
,
Pfannschmidt
M
,
Klar
E
,
Hofmann
WJ
,
Keppler
D
. 
Expression of the multidrug resistance proteins MRP2 and MRP3 in human hepatocellular carcinoma
.
Int J Cancer
2001
;
94
:
492
9
.
36
Jiang
WG
,
Watkins
G
,
Douglas-Jones
A
,
Mokbel
K
,
Mansel
RE
,
Fodstad
O
. 
Expression of Com-1/P8 in human breast cancer and its relevance to clinical outcome and ER status
.
Int J Cancer
2005
;
117
:
730
7
.
37
Abba
MC
,
Drake
JA
,
Hawkins
KA
, et al
. 
Transcriptomic changes in human breast cancer progression as determined by serial analysis of gene expression
.
Breast Cancer Res
2004
;
6
:
R499
513
.
38
Zorn
AM
,
Barish
GD
,
Williams
BO
,
Lavender
P
,
Klymkowsky
MW
,
Varmus
HE
. 
Regulation of Wnt signaling by Sox proteins: XSox17 α/β and XSox3 physically interact with β-catenin
.
Mol Cell
1999
;
4
:
487
98
.
39
Minn
AJ
,
Gupta
GP
,
Siegel
PM
, et al
. 
Genes that mediate breast cancer metastasis to lung
.
Nature
2005
;
436
:
518
24
.
40
Maier
M
,
Castagner
F
,
Berger
P
,
Suter
U
. 
Distinct elements of the peripheral myelin protein 22 (PMP22) promoter regulate expression in Schwann cells and sensory neurons
.
Mol Cell Neurosci
2003
;
24
:
803
17
.
41
Ellison-Zelski
SJ
,
Solodin
NM
,
Alarid
ET
. 
Repression of ESR1 through actions of estrogen receptor α and Sin3A at the proximal promoter
.
Mol Cell Biol
2009
;
29
:
4949
58
.
42
Carroll
JS
,
Meyer
CA
,
Song
J
, et al
. 
Genome-wide analysis of estrogen receptor binding sites
.
Nat Genet
2006
;
38
:
1289
97
.
43
Kruh
GD
,
Belinsky
MG
,
Gallo
JM
,
Lee
K
. 
Physiological and pharmacological functions of Mrp2, Mrp3 and Mrp4 as determined from recent studies on gene-disrupted mice
.
Cancer Metastasis Rev
2007
;
26
:
5
14
.
44
Liu
Y
,
Peng
H
,
Zhang
JT
. 
Expression profiling of ABC transporters in a drug-resistant breast cancer cell line using AmpArray
.
Mol Pharmacol
2005
;
68
:
430
8
.
45
Metivier
R
,
Penot
G
,
Hubner
MR
, et al
. 
Estrogen receptor-α directs ordered, cyclical, and combinatorial recruitment of cofactors on a natural target promoter
.
Cell
2003
;
115
:
751
63
.
46
Pitroda
SP
,
Khodarev
NN
,
Beckett
MA
,
Kufe
DW
,
Weichselbaum
RR
. 
MUC1-induced alterations in a lipid metabolic gene network predict response of human breast cancers to tamoxifen treatment
.
Proc Natl Acad Sci U S A
2009
;
106
:
5837
41
.
47
Takahashi
H
,
McCaffery
JM
,
Irizarry
RA
,
Boeke
JD
. 
Nucleocytosolic acetyl-coenzyme a synthetase is required for histone acetylation and global transcription
.
Mol Cell
2006
;
23
:
207
17
.
48
Liu
W
,
Bagaitkar
J
,
Watabe
K
. 
Roles of AKT signal in breast cancer
.
Front Biosci
2007
;
12
:
4011
9
.
49
Shek
LL
,
Godolphin
W
. 
Model for breast cancer survival: relative prognostic roles of axillary nodal status, TNM stage, estrogen receptor concentration, and tumor necrosis
.
Cancer Res
1988
;
48
:
5565
9
.
50
Chadli
A
,
Felts
SJ
,
Wang
Q
, et al
. 
Celastrol inhibits Hsp90 chaperoning of steroid receptors by inducing fibrillization of the Co-chaperone p23
.
J Biol Chem
2010
;
285
:
4224
31
.

Supplementary data