|Home | About | Journals | Submit | Contact Us | Français|
PTTG Binding Factor (PBF; PTTG1IP) is a relatively uncharacterized oncoprotein whose function remains obscure. Because of the presence of putative oestrogen response elements (ERE) in its promoter, we assessed PBF regulation by oestrogen. PBF mRNA and protein expression were induced by both diethylstilbestrol and 17ß-estradiol in oestrogen receptor alpha (ERα) positive MCF-7 cells. Detailed analysis of the PBF promoter showed that the region −399 to −291 relative to the translational start site contains variable repeats of an 18 bp sequence housing a putative ERE half-site (gcccctcGGTCAcgcctc). Sequencing the PBF promoter from 122 normal subjects revealed that subjects may be homozygous or heterozygous for between 1 and 6 repeats of the ERE. ChIP and oligonucleotide pull down assays revealed ERα binding to the PBF promoter. PBF expression was low or absent in normal breast tissue, but was highly expressed in breast cancers. Subjects with greater numbers of ERE repeats demonstrated higher PBF mRNA expression, and PBF protein expression positively correlated with ERα status. Cell invasion assays revealed that PBF induces invasion through Matrigel, an action that could be abrogated both by siRNA treatment and specific mutation. Further, PBF is a secreted protein, and loss of secretion prevents PBF inducing cell invasion. Given that PBF is a potent transforming gene, we propose that oestrogen treatment in post-menopausal women may up-regulate PBF expression, leading to PBF secretion and increased cell invasion. Further, the number of ERE half sites in the PBF promoter may significantly alter the response to oestrogen treatment in individual subjects.
Breast carcinogenesis requires multiple genetic changes, including the altered expression and function of tumour suppressor genes and oncogenes. Most human breast cancers evolve from normal epithelial cells in terminal duct lobular units through a series of increasingly abnormal stages over long periods of time. Key stages in this progression are hyperplasia, atypical hyperplasia, in situ carcinoma, invasive carcinoma and, finally, metastatic disease (1). Invasion into surrounding stroma defines the transition from in situ to invasive carcinoma. However, most defects responsible for the development and progression of malignant disease remain unknown (1).
Described in only 9 publications (2-10), 4 of which are from our own group (2;3;6;8), Pituitary Tumor Transforming Gene binding factor (PBF) was identified through its ability to interact with PTTG1, the human securin (4). First isolated in 1997 (11), PTTG1 is an oestrogen-regulated gene (12;13) previously implicated in breast cancer, with highest expression in invasive and metastatic breast cancers (14). Its binding partner PBF has not, however, been studied in the context of breast cancer before.
Initially identified in 1998 (10), PBF comprises 6 exons spanning 24 Kb within chromosomal region 21q22.3. The 180 amino acid peptide sequence of PBF shares no significant homology with other human proteins, but is highly conserved across a wide diversity of animal species (73% homology to mouse, 67% frog, 60% chicken, 52% zebra fish), suggesting both unique functionality and significant evolutionary importance.
PBF is ubiquitously expressed (15), but a decade after its cloning, very little has been reported concerning its function. We previously characterised PBF expression in thyroid cancers, and demonstrated it to be a transforming gene in vitro, and to be tumourigenic in vivo (8). Further, high PBF expression was independently associated with poor prognosis in human thyroid cancer. Most recently, we showed that PBF repressed iodide uptake in thyroid cells, both through transcriptional regulation (3) and through altered subcellular trafficking (6).
PBF is thus a relatively uncharacterised transforming gene that plays a part in multiple cellular processes, particularly in the setting of endocrine neoplasia. We now present extensive data, outlined below, that suggest PBF represents an entirely novel gene of direct relevance to breast cancer.
Breast tumour paraffin-embedded samples arranged on tissue microarrays (TMAs) were available from 146 patients (16). There were 6 cases of ductal carcinoma in situ, 4 cases of medullary carcinoma, 1 atypical medullary carcinoma, 7 cases of mucinious type, 2 cases of tubular carcinoma, 22 cases of lobular carcinoma, 1 lobular-papillary and 1 benign tumour. A further 101 cases were designated pathologically as being of no special type (breast tumours not fitting the histological categories above, but tending to be invasive ductal carcinomas that could not be further characterised on morphological grounds) (16). Clinical follow-up, encompassing tumour grade, vascular invasion, lymph node stage (LNS), Nottingham Prognostic Index (NPI) and ER status was available for the breast cancer series. Normal breast paraffin-embedded samples were available from US Biomax (Rockville, USA) (n = 6).
MCF-7 human Caucasian breast adenocarcinoma cells were obtained from the European Collection of Cell Cultures (ECACC) in April 2008. Low passage number cells obtained from the original stock were maintained in RPMI 1640 medium (Gibco) supplemented with 10 % fetal bovine serum (FBS). Cells were treated with diethylstilbestrol (DES) and 17β-estradiol (EST) at final concentrations of 10 nM and 20 nM, and with ICI 182780 (Faslodex/ Fulvestrant) at 100 nM and 1 μM, in RPMI 1640 phenol red free medium (Gibco) supplemented with 10 % charcoal stripped serum.
Total RNA was extracted from MCF-7 cells using the Sigma Trisol Kit, as previously (3). RNA was isolated from paraffin-embedded tissues on microscope slides using the Pinpoint™ Slide RNA isolation System II kit (Zymo Research, USA). RNA was reverse transcribed using the Reverse Transcription System (Promega), as previously (3). Expression of specific mRNAs was determined using the 7500 real time PCR system (Applied Biosystems) (8). Western blot analyses were performed as we have described previously (2;6;17;18).
Briefly, MCF-7 cells were cross-linked by addition of 1 % final concentration formaldehyde directly to 1.5 × 106 mid-exponential cells. After sonication samples were immunoprecipitated with Protein A Agarose beads (Upstate Biotech, Lake Placid, New York) with 5 μg ER antibody (Santa Cruz, CA), alongside an IgG sample as a background control (Abcam, Cambridge, UK). Immunocomplexes were then sequentially washed and crosslinks removed before DNA extraction. PBF promoter primers of sequence 5′GCA-GCC-CTT-TAG-GAT-GGA-G and 5′GAG-GAA-AGG-AGC-CTG-GTA-GC were then used with 5 μl of ChIP DNA material for analysis by semi-quantitative PCR.
Genomic DNA from normal and tumourous colorectal samples and normal thyroid specimens had previously been extracted (17;19). Normal DNA was also obtained from whole blood from patients with normal thyroid function (20). Genomic DNA was isolated from breast tumour and normal tissue which had been formalin fixed and paraffin embedded using the Pinpoint Slide DNA Isolation System™ kit (Zymo Research, USA).
Two primer sets were designed to amplify the region of the PBF promoter which contains a repeated 18bp sequence. Forward (F) and Reverse (R) sequences were: Primer Set 1(F): 5′-GCG-CTC-CCC-TAG-TCC-CCT-3′; Primer Set 1(R): 5′-GCG-AGG-AGA-GCG-GCT-GA-3′; Primer Set 2(F): 5′-GCA-GCC-CTT-TAG-GAT-GGA-G-3′; Primer Set 2(R): 5′-GAG-GAA-AGG-AGC-CTG-GTA-GC-3′. Product sizes were 220 bp and 286 bp, respectively. Sequencing was carried out to confirm ERE repeat numbers using the forward primer of Primer Set 1 or 2.
The pGL3_PBFpromoter construct was created by inserting bases −510 to −211 relative to the translational start site into the pGL3 basic vector (Promega). Cells were harvested in Passive Lysis Buffer (PLB: Promega). The Dual Luciferase Reporter Assay System (Promega) was used, and data were expressed as a ratio of Renilla luciferase activity.
Pull-down assays were essentially as described previously (21). Oligonucleotides of a consensus ERα binding sequence (22), 5′-GTC-CAA-AGT-CAG-GTC-ACA-GTG-ACC-TGA-TCA-AAG-TT- 3′, PBF ERE sequence (half sites in bold), 5′-CTC-GCC-CCT-CGG-TCA-CGC-CTC-GCC-CCT-CGG-TCA-CGC-CTC-GCC-CCT-CGG-TCA-CGC-CTC-GCC-CCT C- 3′ and mutant PBF ERE sequence (base changes underlined), 5′-CTC-GCC-CCT-CAA-TTT-CGC-CTC-GCC-CCT-CAA-TTT-CGC-CTC-GCC-CCT-CAA-TTT-CGC-CTC-GCC-CCT-C-3′ were 5′end-labelled with biotin and incubated with recombinant ERα protein (P2187; Invitrogen Ltd, Paisley, UK). In competition reactions, recombinant protein was additionally incubated with 1-2 nmoles of non-biotinylated oligonucleotide. DNA/protein complexes were captured with 0.1 mg of magnetic streptavidin beads (Promega, Madison, USA). Subsequently, bound proteins were probed with an anti-ERα antibody (sc-543; Santa Cruz, CA, USA).
Normal and tumourous breast specimens were immunostained using our specific rabbit polyclonal antibody to PBF (1:200). For negative controls the primary antibody was replaced by 2 % normal serum. Sections were counterstained with Mayers Haematoxylin, and blinded scoring carried out according to the intensity (0 = not present, +1= least intense, +3 = most intense) and percentage (<25, 25-50, 50-75, >75) of PBF staining.
MCF-7 cells transfected with vector only (VO), haemagglutinin (HA)-tagged PBF (PBF-HA), or the PBF mutant Δ29-93 were seeded onto BD Falcon™ cell culture inserts (8 μM pore size). Subsequently, 800 μl of RPMI 1640 medium supplemented with 20 % FBS, or charcoal stripped serum, was added to the well below the BD Falcon™ cell culture inserts. For knock-down, cells were treated with either 50 nM of Scrambled (negative control # 1, Ambion) or PBF specific siRNAs #14399 and #147350 (mixed in equal quantities; Ambion). After 24 or 48 hours cells were fixed and stained using Mayers Haematoxylin and Eosin (Sigma).
Cells were transfected or treated with diethylstilbestrol and 17β-estradiol at final concentrations of 10 nM and 20 nM. 24 and 48 hours post-PBF transfection, or 48 hours post-oestrogen treatment, 100 μg of MTT was added to each well, as previously (23).
Cell lysates were harvested in RIPA buffer. Cell medium was removed from MCF-7 cells and centrifuged to remove cellular debris. Supernatants were then added to 3 volumes of 100 % ethanol, centrifuged, and the pellet resuspended in RIPA buffer.
Cells were maintained in 1 ml of medium containing 2/3 standard medium and 1/3 leucine-free equivalent, along with 10 μCi L-leucine [3,4,5-3H] (MP Biomedicals). Media extracts were centrifuged to remove floating cells. To obtain cell lysate fractions, cells were lysed in RIPA buffer. 5 μl of rabbit polyclonal anti-PBF antibody (6) or 5 μl normal rabbit control serum were added to cell medium and lysate fractions. Immunocomplexes were pelleted by centrifugation at 13,000g, and disintegrations per minute measured. Western blot analysis of PBF immunoprecipitation confirmed that PBF was specifically pulled down.
MCF-7 cells were transfected with 1 μg PBF-HA and 1 μg chromogranin A-GFP (24) on coverslips. Cells were fixed and permeabilised before blocking. Rabbit polyclonal anti-HA (Y-11) antibody (Santa Cruz Biotechnology, Santa Cruz, CA, USA) was used as primary antibody, and Alexa Fluor 594-conjugated goat anti-rabbit IgG (Invitrogen) as secondary antibody. Hoechst stain was used to visualise nuclei (1:1000).
Data were analysed using Sigma Stat (SPSS Science Software UK Ltd). Normally distributed data were analysed using a two sample Student's t-test. The Mann Whitney Rank Sum test was used for comparison between two groups of non-parametric data. Data containing categorical information were analysed using the Chi-squared test and Fisher's Exact test.
The binding partner of PBF – PTTG1 – is an oestrogen regulated gene implicated in the aetiology of breast cancer (14). We therefore examined whether PBF might also be oestrogen regulated. PBF mRNA expression was induced 48 hours after treatment by 10 nM and 20 nM diethylstilbestrol (DES) and 10 and 20 nM 17ß-estradiol (EST) in ERα-positive MCF-7 cells (Figure 1A). PBF protein expression levels were also significantly up-regulated 2 to 3-fold compared with vehicle controls (Figure 1B). Further, treatment with the anti-oestrogen ICI 182780 at 100 nM and 1 μM concentrations repressed 17ß-estradiol stimulation of PBF expression in a dose-dependent manner (Figure 1C).
To investigate whether 17ß-estradiol stimulation of PBF expression occurred directly through the PBF promoter, crosslinking chromatin immunoprecipitation (ChIP) assays were carried out as described in the Materials and Methods. In each case (N=4 experiments), oestrogen receptor alpha bound to the human PBF promoter, with most pronounced binding occurring between 24 and 48 hours post oestrogen treatment in MCF-7 cells (Figure 1D).
In silico analysis of the human PBF promoter identified the region −399 to −292 relative to the translational start site to be replete with putative EREs (Figure 2A). Preliminary sequencing of this region revealed that it contained variable repeat numbers of an 18 bp sequence housing a putative consensus ERE half-site (gcccctcGGTCAcgcctc) (Figure 2A). PCR and sequencing of the region was additionally examined with a separate set of primers, confirming the existence of polymorphic numbers of 18 bp repeats (Figure 2A).
A panel of 92 genomic DNA samples was available to us, DNA being prepared from normal thyroid and colon tissue, and from tumourous colon. PCR and sequencing revealed that subjects may be homozygous or heterozygous for between 1 and 6 repeats of the 18 bp region housing the ERE (Figure 2B). We next examined ERE repeat number in formalin-fixed paraffin-embedded breast tumour microarray (TMA) samples. Of the ~60 tumours assessed, 27 tumours gave informative sequencing data for ERE repeat number, and 3 out of the 6 normals yielded unambiguous sequence data. These studies revealed that breast tumours could either be homozygous (3/3 repeats) or heterozygous (3/5 repeats), whereas the N=3 normal breast specimens were all homozygous for 3 repeats (Figure 2B). None of the other variants were detected in breast DNA. Overall, as each 18 bp repeat houses a putative consensus ERE half site, subjects therefore differ in the number of EREs present in their PBF promoter.
Having determined that PBF is regulated by oestrogen and that ERα binds to the PBF promoter in ChIP assays, and having identified a polymorphic region in the human PBF promoter containing variable numbers of putative ERE half sites, we next examined the oestrogen responsiveness of this fragment of the promoter. 20 nM DES and EST both induced significant luciferase activity compared to vehicle-treated cells after 24 hours (Figure 2 C). At 48 hours, the effect was more pronounced (DES: 1.8 ± 0.2-fold, p < 0.01 compared with vehicle, N = 3; EST: 1.7 ± 0.3-fold, p<0.05 compared with vehicle, N = 3) (Figure 2 C).
Next, we investigated whether the polymorphic ERE half sites identified within the promoter region −399 to −292 were capable of binding oestrogen receptor alpha (ERα) directly. We performed pull-down assays using oligonucleotides containing either a series of 3 wild type or 3 mutated PBF ERE half site repeats. Recombinant human ERα bound specifically to the biotinylated double-stranded PBF ERE oligonucleotide (Figure 2D), which could be competed out by incubation with increasing excesses of unlabelled PBF oligonucleotide. Further, recombinant ERα bound a biotinylated consensus double stranded ERE oligonucleotide, and binding was repressed by competition with unlabelled wild type PBF ERE oligonucleotide, but not by the mutated PBF oligonucleotide in which the EREs had been abolished. As PBF mRNA was induced 1.5 to 3-fold at identical timepoints and at identical doses of oestrogen (Figure 1), these data indicate that the short promoter region −510 to −211 is positively regulated by oestrogen and confers most, but not all, of PBF's responsiveness to DES and EST. In addition, the specific ERE half site region identified within the promoter region −399 to −292 is capable of binding ERα, suggesting that the main mechanism of oestrogen regulation of PBF is directly via the polymorphic EREs of the proximal promoter.
We next investigated whether PBF was expressed in human breast tumours. Initial mRNA investigations utilised 20 breast TMA specimens and 6 normal breast specimens. PBF mRNA expression was apparent in RNA extracted from 18/20 TMA tumour samples, with a mean TaqMan RT-PCR ΔCt of approximately 8 to 10, suggesting robustly detectable levels of expression (Figure 3A). Reverse transcriptase negative (RT-) controls confirmed that amplification in the tumour samples examined was not an artefact of genomic DNA contamination (data not shown). In contrast, PBF mRNA was not detected (ND) in any of the 6 normal breast specimens analysed (Figure 3A).
PBF protein expression was next quantified in a larger series of normal breast specimens (N=8) and TMA tumour samples (N=146) through immunohistochemistry, using our rabbit polyclonal antibody (6). Examples of scoring intensities are provided in Figure 3B. PBF expression was low or absent in normal breast tissue, whereas it was strongly expressed in epithelial cells of all tumour types and grades of breast tumour assessed (Figure 3C). Specificity of staining was confirmed in negative control experiments (data not shown). Importantly, ERα status positively and significantly correlated with the percentage of PBF protein expression (p <0.001). However, the remaining phenotypic end-points examined (tumour grade, vascular invasion, lymph node stage or the Nottingham Prognostic Index) were not associated with PBF staining intensity or percentage expression.
Next, we investigated the relationship between ERE number and PBF mRNA expression. 8 fixed tumour specimens and 3 normal breast samples yielded both mRNA and promoter sequencing data. PBF expression was apparent in 86 % of 3/5 heterozygotes, but only 25 % of 3/3 homozygotes (p = 0.044). Thus, PBF is oestrogen-regulated, its expression is higher in ER-positive than ER-negative tumours, and a greater number of ERE repeats is associated with higher PBF expression.
The exact function of PBF in cell transformation is not known (8). Given that PBF is induced in breast cancer, and that invasion and metastasis are critical processes in breast cancer progression, we next assessed whether PBF might play a role in cell invasion.
When 1×105 cells/well were seeded in invasion assays, PBF over-expression was associated with a 2.5 ± 0.6-fold increased cell invasion compared to VO treatment at 24 hours (N=6, p<0.01), and 6.1 ± 2.9-fold increase at 48 hours (N=6, p<0.01; Figure 4A). At a density of 2×105 cells/well, PBF induced a 2.5 ± 0.3-fold increase in cell invasion after 48 hours (N=6, p<0.01).
To determine whether the enhanced invasiveness of MCF-7 cells following PBF transfection reflected an increase in cell proliferation, MCF-7 cells were transfected with VO or PBF, and cell number estimated after 24 and 48 hours using MTT assays (Figure 4B). PBF did not significantly increase the proliferation of MCF-7 cells after 24 hours, but did marginally (~10 %) increase cell number after 48 hours. These data suggest that increased proliferation does not explain PBF's influence upon invasion.
As oestrogen induces PBF, and MCF-7 cells are well documented as an oestrogen-responsive cell line, we next examined the influence of treatment with 10 nM DES on cell invasion. DES treatment induced MCF-7 cell invasion by approximately 2.3-fold compared with vehicle treatment (N = 3, p < 0.001) (Figure 5A). 10 nM and 20 nM diethylstilbestrol and 17ß-estradiol treatment did not significantly alter MCF-7 cell proliferation (data not shown).
Subsequently, we validated transient knock-down of PBF in MCF-7 cells using siRNA. 50 nM of a PBF-specific siRNA elicited ~80 to 90 % knock-down compared with a scrambled control (Figure 5B). Cell invasion assays were then repeated after 24 hours in the presence of 10 nM DES or vehicle. Critically, the increased cell invasion observed after DES treatment in the presence of a scrambled control (103 ± 17 cells, p = 0.002, N = 2) was abolished when PBF was simultaneously knocked down using a PBF specific siRNA (27 ± 14 cells, P=NS, N=2) (Figure 5C). These data suggest that oestrogen induction of MCF-7 cell invasion through Matrigel is mediated via PBF.
As the induction of invasion by PBF could not be explained by increased cell number, and because cell invasion is a process frequently associated with secretion, we assessed whether PBF is a secreted protein. MCF-7 cells were grown in the presence of 3,4,5-3H-L-leucine, cell medium fractions harvested after 24 hours, and immunoprecipitations carried out with our PBF antibody (Figure 6). Immunoprecipitation of labelled PBF revealed that PBF is indeed a secreted protein, with vector-only transfected MCF-7 cells demonstrating that approximately 20 % of total cellular PBF is secreted over the experimental timeframe of 24 hours (Figure 6A). This was significantly enhanced by transient over-expression of PBF (49 ± 3 % secretion, N = 3, p < 0.05 compared with VO). Further, a mutant of PBF lacking amino acids 29 to 93 (and hence a functional signal peptide and two potential putative glycosylation sites (6)), showed reduced secretion into the cell medium compared with wild type (28 ± 7 %, N = 3), but did not differ statistically from VO (Figure 6A).
To confirm that PBF is secreted into the cell medium, we further carried out Western blotting for HA-tagged PBF. Cell lysates demonstrated successful transfection of HA-tagged PBF (Figure 6B). Wild type PBF is a putative glycoprotein (10), which runs as a doublet at around 25 to 30 kDa. Mutant Δ29-93 was, as anticipated, smaller, and ran at approximately 20 kDa (Figure 6B). Whereas wild type PBF was easily detectable in cell media, confirming that PBF is a secreted protein, Mutant Δ29-93 was not apparent.
To investigate the mechanism by which PBF is secreted, MCF-7 cells were co-transfected with HA-tagged PBF and chromogranin A-GFP. Chromogranin A-GFP is a chimeric photoprotein that is transported via the regulated pathway for exocytosis (24), and has recently been used as a marker for secretory granules in MCF-7 cells (25). As demonstrated previously in other cell lines, PBF-HA was found within intracellular vesicles, often detected towards the periphery of cells (6), in keeping with a secretory phenotype (Figure 6C). Although these vesicles were localised in a similar pattern to the chromogranin A-GFP-labelled secretory granules, no colocalisation with PBF was observed. This suggests that PBF is secreted via the constitutive pathway of secretion rather than via regulated secretion. Overall then, PBF is secreted by MCF-7 cells, higher expression results in increased secretion, and deletion of the amino acid region 29-93 results in significantly attenuated secretion into the cell medium.
Having shown that PBF is secreted by MCF-7 cells, we investigated the relationship between secretion and invasion. As before, PBF induced significant cell invasion when over-expressed (269 ± 52 invading cells, N = 10 experiments, p = 0.018 compared with VO control; Figure 6C). However, Mutant Δ29-93 failed to induce cell invasion compared with VO (161 ± 45 invading cells, N = 10 experiments, p = 0.518 compared with VO). Thus, secretion of PBF contributes significantly to its induction of cell invasion.
In summary, PBF is a relatively uncharacterised proto-oncogene which is induced by oestrogen in MCF-7 cells, shows increased expression in breast cancer, and which stimulates cell invasion, at least in part through secretion. On this basis, PBF warrants further and intensive study in the context of breast cancer initiation and progression.
Numerous genetic changes governing the initiation, progression and metastasis of breast cancer have already been described, but new genetic markers and therapeutic targets are vital for continued progress in addressing the approximately half-million global deaths annually from the disease. Based on our investigations, we propose four principle lines of evidence suggesting a critical role of PBF in breast cancer. First, PBF is highly expressed in breast tumours, where its expression correlates with oestrogen receptor positivity. Second, it is regulated by diethylstilbestrol and 17β-estradiol, both at the mRNA and protein level. Third, PBF upregulation results in significant MCF-7 cell invasion through Matrigel, a phenomenon highly dependent upon PBF secretion. Fourth, repressing PBF expression in the face of oestrogen stimulation prevents oestrogen mediated induction of cell invasion. Hence, oestrogen regulates PBF expression in MCF-7 cells, tumours show increased levels of the protein, and its functional property in promoting invasion can be ameliorated both through mutation and knock-down.
Whilst highly conserved, the 180 amino acid peptide sequence of PBF shares no significant homology with other human proteins. Its exact mechanisms of action have not been divined, but it has been shown to induce tumours in nude mice (8), to regulate expression and function of the sodium iodide symporter (3;6), and to interact with the human securin, PTTG (4). Our experiments are the first to report a role for PBF in the breast, and to elucidate a function in cell invasion. We now propose that oestrogen regulates PBF mRNA and protein expression, and suggest that this is mediated predominantly by a cluster of ERE half sites ~300 bp upstream of the translational start site.
One surprising finding of our preliminary sequence analysis was that the human PBF promoter is polymorphic for an 18 bp repeat housing a putative ERE half site. Importantly, a higher number of putative EREs was statistically associated with greater PBF mRNA expression in the breast, oestrogen receptor positive tumours having significantly increased PBF protein expression compared to receptor negative tumours. Although these findings would need to be confirmed in a larger series of samples, a higher number of EREs in a breast tumour would therefore predict a greater response to circulating oestrogen and hence increased expression of a known transforming gene.
Oestrogen has previously been shown to enhance expression of PTTG, the binding partner of PBF (12;13), and PTTG upregulation has been described in breast cancer (14). Whilst it was not the focus of the current investigation, it would be interesting to correlate PTTG and PBF expression in individual breast tumours. The functional implications of simultaneous over-expression are hard to gauge, given that both proteins are inherently multifunctional and likely to have dependent and independent modes of action. However, future studies may delineate the individual contributions of each gene to breast cancer initiation and progression.
Our in vitro experiments were predominantly carried out in MCF-7 cells, which remain the ‘gold standard’ ERα-positive and oestrogen-sensitive breast cell line (26). Further, MCF-7 cells are weakly invasive (27), allowing us to perform physiologically relevant invasion assays.
Our wider clinical associations did not reveal striking associations between PBF expression and tumour phenotype, which was unexpected. Clinical associations were hampered by a lack of a matched normal:tumour cohort, which would allow a more detailed interrogation of the association between promoter polymorphism and clinical outcome. Further, fresh tissue would have allowed us to perform Western blotting for PBF expression, which would have provided more quantitative expression data than through immunohistochemistry on FFPE slides.
The mechanism by which PBF induces cell transformation is not known. Our current study suggests that this might be via the induction of cell invasion. MCF-7 cells showed potent increases in invasion through Matrigel in response to PBF transfection. A mutant PBF which was not secreted into the cell medium lost this phenotype. Further, a specific siRNA entirely blocked oestrogen induction of cell invasion. Thus, we hypothesise that oestrogen induces PBF which in turn drives breast cell invasion, at least in part through secretion. As oestrogen stimulation of PBF expression was abrogated by co-treatment with the anti-oestrogen ICI 182780, oestrogen receptor antagonists might be useful in treating tumours with high PBF expression, to block potential cell invasion mediated by PBF.
In summary, we present evidence that the poorly characterised proto-oncogene PBF has particular relevance to breast tumourigenesis, particularly with respect to progression. Based on our in vitro and ex vivo findings we predict that oestrogen treatment in post-menopausal women may up-regulate PBF expression, leading to increased PBF secretion. Given that individuals have different numbers of ERE repeats, it is likely that oestrogen will stimulate PBF expression variably, with a potentially critical impact upon breast cell invasion.
We thank Professor Richard Cheney (University of North Carolina at Chapel Hill, NC, USA) for the provision of the chromogranin A-GFP construct.
Financial support: This work was supported by grants from the IDA Cooper Foundation and the Medical Research Council.