Expression of oestrogen receptor beta (ERβ1) protein in human breast cancer biopsies

Oestrogen action is mediated via specific receptors that act as ligand-activated transcription factors. A monoclonal antibody specific to the C-terminus of human oestrogen receptor beta has been characterized and the prevalence of expression of oestrogen receptor beta protein investigated in a well defined set of breast cancers. Reverse transcription-polymerase chain reaction analysis of RNA from tissue biopsies detected oestrogen receptor beta in all samples examined. The anti-oestrogen receptor beta antibody cross reacted specifically with both long (∼59 Kd) and short (∼53 Kd) forms of recombinant oestrogen receptor beta. Western blot analysis of breast tumours contained both forms of oestrogen receptor beta protein although in some samples lower molecular weight species (32–45 Kd) were identified. Fifty-one breast cancer biopsies were examined using immunohistochemistry; 41 (80%) were immunopositive for oestrogen receptor alpha, 48 (94%) were immunopositive for oestrogen receptor beta and 38 (74.5%) co-expressed both receptors. Expression of oestrogen receptor beta was exclusively nuclear and occurred in multiple cell types. There was no quantitative relationship between staining for the two ERs although in tumours in which both receptors were present immunoexpression of oestrogen receptor alpha was invariably more intense. The significance of oestrogen receptor beta protein expression in breast cancers to therapy remains to be determined but the availability of a well characterized antibody capable of detecting oestrogen receptor beta in archive material will facilitate the process. British Journal of Cancer (2002) 86, 250–256. DOI: 10.1038/sj/bjc/6600035 www.bjcancer.com © 2002 The Cancer Research Campaign

Until recently it was accepted that the major effects of oestrogen on the growth and development of the breast and its tumours was mediated through a single oestrogen receptor (ERa, Green et al, 1986). Ligand binding assays and immunohistochemical studies indicated that most breast tumours possessed such receptors and their presence was associated with the likelihood of response to endocrine therapy (McGuire et al, 1982;Jordan et al, 1988;Miller, 1996). However in 1996 an additional ER isotype, usually known as ERb, was identified in rat (Kuiper et al, 1996) and human (Mosselman et al, 1996). Both receptors share significant sequence homology within their DNA and ligand binding domains but are encoded on different chromosomes . Studies in vitro have demonstrated that although both ERa and ERb bind oestradiol with equal affinity (Kuiper et al, 1997) these receptors may have differential responses to some oestrogen agonists and antagonists (Watanabe et al, 1997;Barkhem et al, 1998;Jones et al, 1999;Sun et al, 1999). Notably ERb appears to have a higher affinity for phytoestrogens, including genestein, than does ERa (Kuiper et al, 1997). When present within in the same cell, ERa and ERb have the capacity to form either homo-or heterodimers (Pace et al, 1997) and the proportions of the different isotypes may be critical to modulation of gene expression (Hall and McDonnell, 1999). Studies in mammary tissues of the rat have suggested that one role of ERb may be to antagonize ERa-mediated actions in epithelial cells (Saji et al, 2000), a function supported by data from in vitro cell transfections (Hall and McDonnell, 1999).
To date studies demonstrating the expression of ERb in breast cancer tissues have largely been confined to the demonstration of expression of ERb mRNA (Dotzlaw et al, 1997;Leygue et al, 1998;Speirs et al, 1999;Vladusic et al, 2000). Messenger RNAs encoding variant forms of both ERa (Bollig and Miksicek, 2000) and ERb (Lu et al, 1998) have been identified in breast cancers and in breast cancer cell lines and there has been considerable debate over the role of such variants in cancer progression (Balleine et al, 1999;Huang et al, 1999).
The present investigation was designed to characterize the expression of ERb and ERa proteins in a series of 51 breast cancers; some samples were also subjected to analysis for mRNAs by RT -PCR. We have made use of specific monoclonal antibodies and used both immunohistochemistry on well-fixed tissues in which the cellular architecture has been preserved as well as Western analysis of tissue extracts. These investigations have demonstrated wide spread expression of ERb protein and provide new information important for further exploration of the relationship between the co-expression of ERb and ERa and the in response of breast cancers to endocrine therapies.

Patients and tissue samples
Samples of breast were obtained from 51 consecutive patients presenting to the Edinburgh Breast Unit with diagnosis of breast cancer who had given informed consent for tissue to be used for research purposes. Samples were snap frozen to provide material for extraction of RNA or protein, or fixed in 10% neutral buffered formaldehyde for 16 to 24 h then stored in 70% (w v 71 ) ethanol prior to processing into paraffin wax at the Department of Pathology using standard procedures.

Antibodies
The anti-hERa mouse monoclonal antibody (code 1D5) was obtained from DAKO (Cambridge, UK). A peptide located at the C-terminus of hERb (Mosselman et al, 1996) (CSPAEDSKS-KEGSQNPQSQ) was used to prepare a monoclonal antibody in mice according to standard methods and positive clones were identified by ELISA using recombinant human ERb (P2466, PanVera, Madison, WI, USA) (Saunders et al, 2000). This antibody has been used previously to demonstrate expression of ERb using human ovarian tissue sections (Saunders et al, 2000).

Western analysis
Two forms of recombinant human ERb1 were obtained from Pan Vera (Madison, WI, USA). These were hERb1 'short', a *53 Kd form of the receptor (bs) synthesized from a cDNA (Mosselman et al, 1996) lacking the first potential start site for translation (Ogawa et al, 1998a), and hERb1 'long' (bL) the larger protein (*59 Kd) synthesized from the full length cDNA (Ogawa et al, 1998a). Recombinant hERa (*66 Kd) was also obtained from Pan Vera. Gel analysis and blotting were carried out as described previously (Saunders et al, 2000). Briefly, proteins were extracted from frozen biopsy specimens by rapid homogenization of tissue in denaturing/loading buffer (50 mM Tris-HCl pH 6.8, 100 mM DTT, 2% SDS, 0.1% bromophenol blue, 10% glycerol, all from Sigma). Recombinant proteins (0.5 mg lane 71 ), tissue extracts (30 -50 mg total protein) and prestained protein molecular weight markers (BioRad) were separated on denaturing minigels containing an acrylamide gradient from 4 to 20% (w v 71 ) polyacrylamide (Novex, San Diego, CA, USA). Membranes were incubated overnight with the mouse monoclonal anti hERb1 (code M9) at 1 in 500 or mouse monoclonal anti-hERa (code1D5) at 1 in 100; both the antibodies were diluted in TBST containing 5% normal donkey serum. Bound antibodies were detected using rabbit anti-mouse IgG and the ECL visualization system (Amersham, Bucks, UK) according to the manufacturer's instructions.

Immunohistochemistry
Sections (4 mm) were mounted on Superfrost coated slides (BDH, Poole, Dorset, UK) dewaxed and rehydrated in gradient alcohols and distilled water. Endogenous peroxidases were blocked with 3% hydrogen peroxide for 10 min and sections were subjected to heat-induced antigen retrieval in 0.01 m citrate buffer, pH 6.0 (Norton et al, 1994) before staining with specific antibodies as outlined below.

Anti-ERa All staining for ERa was carried out in the Pathology
Department of the Western General Hospital. An endogenous biotin block was carried out by applying 100 ml egg white blocking solution for 30 min. Anti-ERa, (Dako) was diluted 1 in 50 in biotin diluent for primary antibodies (PBS, goat serum and dbiotin), and incubated in the sections for 60 min at room temperature. The secondary antibody, biotinylated anti-mouse Ig(Vector Laboratories) was diluted 1 : 2000, in 'background reducing diluent' (Dako) and applied to sections for 30 min at room temperature. The tertiary system (ABC-HRP, Dako) was applied as per manufacturer's instructions for 30 min at room temperature. The tissue was visualized by immersing sections in 3,3'-diaminobenzidine tetrahydrochloride (DAB) for 5 min. Sections were counterstained using Mayers haematoxylin (Sigma-Aldrich, Poole, Dorset), dehydrated through gradient alcohols and mounted.
Anti-ERb Immunolocalization was undertaken as described in detail in Saunders et al (2000), Sections were blocked for 30 min in normal rabbit serum (NRS, Diagnostics Scotland, Carluke) diluted 1 : 4 in TBS containing 5% BSA (NRS/TBS/BSA), rinsed briefly in TBS and an avidin biotin block performed using reagents from Vector (Peterborough, UK). Anti-ERb antibody was diluted 1 : 40 in NRS/TBS and incubated on sections overnight at 48C. Sections were washed twice for 5 min each time in TBS and incubated with rabbit anti mouse, (Dako, Cambridge, UK) diluted 1 : 500 in NRS/TBS/BSA. Thereafter, bound antibodies were visualized by incubation with 3,3'-diaminobenzidine tetra-hydrochloride (liquid DAB cat K3468, DAKO). Sections were counterstained with haematoxylin.
Images were captured using an Olympus Provis microscope (Olympus Optical Co, London, UK) equipped with a Kodak DCS330 camera (Eastman Kodak Co., Rochester, NY, USA), stored on a Macintosh PowerPC computer and assembled using Photoshop 5.5 (Adobe, Mountain View, CA, USA).

Quantitation of immunohistochemical staining
Quantitation was based on a scoring system reported in detail previously (Allred et al, 1998;Leake et al, 2000). This method is based on a composite additive score of intensity 0 -3 and proportion of malignant epithelial cells staining 0 -5. This gives a range from 0 -8 for each tissue. Samples were analyzed using the SPSS package (version 10 for Macintosh; SPSS Inc, Chicago, IL, USA) and plotted as a box and whisker plot. No correlation between ERa and ERb scores was detected.

Detection of mRNAs for ERa and ERb in breast cancer samples
All samples tested (n=9) were positive for ERb following RT -PCR ( Figure 1). This signal always appeared greater than those for ERa and was present in both ERa positive and negative samples. Actin was amplified from all samples although the efficiency of the reaction was variable.

Specificity of antisera and extraction of ER proteins from breast cancer biopsies
On Western blots ( Figure 2) antibodies directed against ERa and ERb bound to either recombinant ERa or recombinant ERb protein depending upon the isotype to which they were directed. These results were consistent with previously published data (Saunders et al, 2000); no binding of the ERb specific monoclonal to ERa was observed ( Figure 2, lower panel, lane a). The anti-hERb monoclonal that was directed against a peptide at the C-terminus of hERb bound to both short (Mosselman et al, 1996) and long (Ogawa et al, 1998a) forms of ERb. This result is consistent with data that has demonstrated that the difference in size of the long and short forms of ERb is due to use of alternative start sites for translation within the full length mRNA and that the C-termini of both proteins are identical.
Tissue biopsied from eight tumours, that were histologically shown to be cancers, were also examined. The predominant form of the ERa protein ( Figure 2, upper panel) extracted from all biopsies migrated with an apparent molecular size (*66 Kd) identical to recombinant ERa run in a parallel lane (a). In only two samples (lanes 6 and 7) did we see evidence of expression of shorter/variant ERa proteins.
The amount of ERb protein detected in extracts from cancer biopsies was highly variable (Figure 2 lower panel). It was notable that in six of the eight samples proteins migrating with apparent molecular sizes corresponding to both long (*59 Kd) and short (53 Kd) ERb were present. We have found that this antibody recognizes ERb protein extracted from human ovary, prostate (Saunders et al, 2000) endometrium and testis and human cell lines (MCF-7, Ishikawa, unpublished observations). In breast tumour samples that appeared to contain high levels of expression of full length ERb (numbers 1, 3, 4, 7, 8) several lower molecular weight protein species with apparent molecular weights from 32 to 45 Kd were detected.

Immunolocalization of oestrogen receptors
Typical examples of immunostaining for ERa and ERb are shown in Figures 3 and 4 respectively. Staining for ERa (Figure 3) was  (7) contained a sample prepared without reverse transcriptase. Note that although a cDNA specific for ERb was amplified from all samples, the amount of ERa cDNA amplified from the same sample set was highly variable. The anti-ERb1 antibody bound to both long (bL) and short (bs) forms of recombinant hERb but not to recombinant hERa (a). Proteins migrating with the same apparent molecular size as recombinant ERa (a, upper panel, arrowhead) were detected in all breast samples (lanes 1 to 8, note identical samples were used for both gels and are loaded in the same order). In sample numbers 6 and 7 additional lower molecular weight forms of ERa were present. Variable amounts of ERb proteins were detected in the same samples. Proteins migrating with the same apparent molecular size as both long and short forms of ERb proteins (arrowheads) were detected in breast sample numbers 1, 3, 4, 6, 7, 8; additional lower molecular weight variants were present in these same extracts but samples 2 and 5 lacked significant levels of ERb.

Molecular and Cellular Pathology
Expression of ERb1 in human breast cancer biopsies PTK Saunders et al predominantly nuclear and almost exclusively restricted to malignant epithelium (insets A' and B') in this tissue series. Note that the malignant tissues illustrated in Figure 4A,B are the same as those in Figure 3A,B (codes 5580 and 5667 respectively) and clearly illustrate that ERa expression (Figure 3) can occur in the presence ( Figure 4A) or absence ( Figure 4B) of ERb. Expression of ERb was almost exclusively nuclear and often appeared granular and heterogeneous ( Figure 4A'). Expression of ERb was noted in a wider range of cells than was ERa and was found in non-malignant components of the tumour including normal glandular elements ( Figure 4D arrows), blood vessels, adipose tissue and stromal cells (asterisks) as well as in non-invasive intraduct cancers ( Figure 4C).

Quantitation of immunohistochemical staining
Most of the tumours (48 out of 51) displayed staining for ERb in malignant epithelium with a range of scoring between 2 and 7 (median score 4.5). ERa staining was found in 41 out of 50 tumours with a range of scoring between 6 and 8 (median score 7.5). Quantitatively it was possible to identify ERa-positive, ERbpositive tumours (38 out of 51, Figures 3A and 4A) as well as ERa-positive, ERb-negative tumours (3 out of 51, Figure 3B compared with Figure 4B; 2 out of 51). ERa-negative, ERb-positive tumours were detected (10 out of 51) but we observed no double negatives. There was no quantitative relationship between immunohistochemical scores for ERa and ERb ( Figure 5).

DISCUSSION
Many breast cancers, like the normal tissue from which they are derived, appear sensitive to oestrogens. The major action of oestrogen appears to be mediated by specific receptor proteins that act as nuclear transcription factors. Until recently, studies have concentrated on the ERa member of the family and these have clearly demonstrated the involvement of the protein in maintaining the growth of hormone sensitive tumours. As a consequence ERa measurements have been used to select patients for endocrine therapy and the protein has become a therapeutic target by which to treat patients with breast cancer. Nevertheless there have been paradoxical observations such as tumours regressing following endocrine deprivation therapy in apparently ERa negative disease. Oestrogen responses in ERa knockout mice and the differential effects of anti-oestrogens in tissues and tumours were also unexplained. Our ability to correlate ER status with outcome of therapy has been complicated by the finding of a second oestrogen receptor (ERb) which can bind oestrogens including oestradiol and tamoxifen with high affinity (Kuiper et al, 1996(Kuiper et al, , 1997Mosselman et al, 1996). As a result there has been a major effort to delineate the role of ERb in the natural history of breast cancer. Many papers have reported that the mRNAs for both ERa and ERb are expressed in breast cancer cell lines (Watanabe et al, 1997;Moore et al, 1998;Vladusic et al, 2000), in breast cancer tissue (Dotzlaw et al, 1997) and in the normal human and rodent mammary gland (Moore et al, 1998;Saji et al, 2000). Studies that have compared levels of expression of the mRNAs encoding the two receptors have reported that the amount of ERb mRNA does not appear to be correlated with that of ERa (Dotzlaw et al, 1997;Iwao et al, 2000;Vladusic et al, 2000) consistent with expression of the receptors by different genes . Some studies have reported that up-regulation/over expression of ERb mRNA may be correlated with development of oestrogen-independent tumour growth and a poor prognosis (Speirs et al, 1999;Iwao et al, 2000).
Modelling studies using ERa have defined the amino acids within the protein which interact with natural as well as synthetic oestrogens and anti-oestrogens (Ekena et al, 1997). The major determinants of ligand binding are conserved between ERa and ERb consistent with their ability of both to bind oestradiol (Kuiper et al, 1997). Barkhem et al (1998) have used cell lines stabily transfected with either ERa or ERb to test the affinity and potency of widely used anti-oestrogens including tamoxifen, raloxifine and ICI 164,384 and concluded that the ligand binding cavity of ERb is more different to that of ERa than can be anticipated from the primary sequence. Recently novel non-steroidal ligands that show subtype specific binding affinity and transcriptional potency have been identified (Sun et al, 1999) and ligand-dependent differences in the ability of ERa and ERb to recruit co-activators following exposure to xenoestrogens described (Routledge et al, 2000). ERdriven gene activation can be determined by the formation of homo-or hetero-dimers, the cell type, and whether the ligand-activated receptors bind to a promotor containing ERE or an AP-1 site (Watanabe et al, 1997;Jones et al, 1999). Furthermore the experience with studies on ERa has been that mRNA is not necessarily Expression of ERb1 in human breast cancer biopsies PTK Saunders et al translated into protein make it essential that assays for ERb are performed at the level of protein.

Molecular and Cellular Pathology
The monoclonal antibody used to detect ERb in the present study was raised against a peptide at the C-terminus of human ERb1 (Mosselman et al, 1996;Moore et al, 1998). This peptide is not conserved in any of the ERb variants formed by alternative splicing of the F domain of the protein (Moore et al, 1998;Ogawa et al, 1998b) and does not recognize recombinant ERb2/bcx on Western blots (unpublished observations). Similarly Western blotting indicated that the monoclonal antibody identified ERb but not ERa in breast cancers. Most of the ERb1 protein detected in the extracts from the breast cancers migrated with the same apparent size as the 'long' and 'short' forms of recombinant ERb1, which are formed by translation from different ATGs in the mRNA (Mosselman et al, 1996;Ogawa et al, 1998a). We did not detect proteins corresponding in size to those that could be translated from mRNAs deleted in exons 5 or 6 (Lu et al, 1998;Brandenberger et al, 1999) predicted to be 16.8 and 13 Kd respectively. The most prominent proteins other than full length ERb1 migrated between 30 and 36 Kd these could represent use of alternative start sites, translation from an exon 2 deleted mRNA (*35 Kd) or translation of protein from mRNA deleted for both exons 5 and 6 (AF074599) which is predicted to be *43 Kd (short) or *49 Kd (long) from the mRNA sequence. It is notable that mRNAs corresponding to alternatively spliced forms of ERb have been detected in breast cancer tissues and cell lines (Lu et al, 1998;Moore et al, 1998;Vladusic et al, 1998;Iwao et al, 2000) as well as in normal human tissues (Ogawa et al, 1998b;Scobie et al, 2001). Furthermore, monoclonal antibodies directed against the N terminus of ERb have detected expression of proteins other than full length ERb in breast cancer cell lines (Fuqua et al, 1999) which might have been formed by translation of alternatively spliced mRNAs. During the course of the present study we found that recombinant ERb proteins (both from commercial sources and prepared in house) degrade if subjected to a single freeze-thaw cycle or following prolonged storage even at low temperatures (7708C). Therefore although considerable attention was paid to extraction of the breast tumour samples and to the storage of extracts we believe that the most likely explanation for the lower molecular weight bands identified in samples containing the highest levels of ERb1 is that these are breakdown products of the full length protein which have formed during handling of the protein extracts.
We have used our ERb1 specific monoclonal antibody to immunolocalize ERb1 in a series of breast cancers as well as in other human and primate tissues (Saunders et al, 2000;Scobie et al, 2001). The present study has demonstrated the presence of ERb1 in cell nuclei not only the malignant epithelium but also non-malignant elements of most breast cancers. The qualitative and quantitative expression of ERb was independent of that of ERa. We have observed that ERb1 was also expressed in multiple types of non-cancer cells within the breast tissue and this will therefore further complicate the assessment of ERb status. For example, methods such as RT -PCR or Western blotting which use tissue extracts may contain a contribution from cells other than those derived from the malignant component of the tumour. It will therefore be important to quantify expression in different compartments of the breast separately. This precludes the simple use of Western and Northern blotting together with other technologies in which tissue is homogenized and extracted.
Whilst our studies were being written up three reports describing immunolocalization of ERb to breast cancer samples were published. Mann et al (2001) used a rabbit polyclonal antibody directed against the N-terminus of human ERb on formalin fixed samples; on the Western blot shown in their article multiple bands are shown, the most prominent of which appeared shorter than the recombinant standard and this may reflect degradation of protein in their extracts or non-specific reactivity of the antibody used. In their paper immunopositive staining of human breast cancer for ERb was present in 66 and 70% of the two sets of samples reported but no mention was made of immunopositive staining of cells other than those of the malignancy. The authors mentioned the potential cross-reactivity of their antibody with isoforms of ERb including ERbcx (Ogawa et al, 1998b) which will not occur with the antibody used in the current study. It is notable that the polyclonal rabbit antibody used by Omoto et al (2001) is raised to an identical part of the ERb1 protein to our monoclonal and we would therefore expect similar results to our own. In their study they used frozen sections of tissue and found that only 59% (52 out of 88) were positive for ERb, with only 38% of the ERa negative samples expressing the ERb subtype. This proportion is much lower than in the current study or in the tissue set studied by Jarvinen et al (2000) who used frozen sections fixed briefly with Zamboni's, and found 60% of cancers contained ERb1 positive cells using a commercial polyclonal antibody raised to the same region of the protein. The need to use frozen sections clearly limits the utility of these antibodies and highlights an important difference with the reagent used in the present study which appears capable of identifying ERb1 in material fixed by formalin, methacarn (unpublished observations) or Bouins (Saunders et al, 2000). In studies using fixed samples from human tissues including ovary, placenta, vas deferens, testis and endometrium we have used monoclonal and polyclonal antibodies to localize ERb proteins (Saunders et al, 2000;Critchley et al, 2001;Scobie et al, 2001). In all cases we find the protein to be nuclear in location in agreement with the findings using fixed tissues of human breast (present study) the only exceptions being dividing cells, and some myoid cell types where background staining of the cytoplasm associated with the secondary antibodies was a problem. We have detected cytoplasmic staining using some commercial anti ERb antibodies especially those that have not been affinity purified and with some secondary antibodies especially those raised in goats (unpublished observations). These findings may explain some of the cytoplasmic staining seen in the figures published by others (Jarvinen et al, 2000;Mann et al, 2001;Omoto et al, 2001).
In conclusion, we believe that to assess the responsiveness of breast cancers to oestrogenic and anti-oestrogenic stimuli it will be necessary to measure both ERa and ERb at the level of protein.
The presence of ERb in both malignant and non-malignant components of breast tumours means that assessments in individual compartments may also be required. This approach is being utilized in our ongoing studies. Quantification of immunoexpression for ER isotypes. Box and whisker plot summarising the relationship between score for ERa (x-axis) and ERb (y-axis) for each sample. Solid horizontal line shows the median for the data, the top of the box the 25th percentile, the bottom the 75th percentile and the additional lines the range of the data. Note that there were no samples with an ERa score of 1 to 5.