GB2361769A - Estrogen receptors - Google Patents

Abstract

A method of detection of breast cancer in cells comprising correlating the levels and/or combination of ER b isoforms present in the cells with the absence or presence of breast cancer, wherein the presence of an ER b isoform indicates the presence of breast cancer. In a preferred embodiment, the ER b isoform is Er b 2 ( b cx), and the presence of this isoform indicates the presence of a tamoxifen-resistant breast cancer type.

Description

1 2361769 ESTROGEN RECEPTORS

The present invention relates to the field of estrogen receptors.

Estrogen is a modulator of cellular growth and differentiation (Warner, M, , et al (1999) Current Opinion in Obstetrics in Gynecology 11: 249-54). In females, the major targets of estrogen are the mammary gland and uterus, however, in both males and females, estrogen is essential for the maintenance of the bone, cardiovascular system, brain and urogenital tract (Simpson, E, R., et al (1998) Mol. & Cell Endocrinol. 145: 55-9; Wenger, N. K., et al (1999) Current Opinion in Cardiology 14: 292-7). Estrogen mediates most of its functions through two specific intracellular receptors, ERa and ERP, which act as hon-none-dependent transcriptional regulators (Kuiper, G. G., et al (1997) Endocrinology, 138: 863-870) Because it is essential for the growth and development of the mammary gland, estrogen has been associated with promotion and growth of breast cancer. Numerous animal studies show that estrogen can induce and promote breast cancer, and it has been demonstrated that removal of the ovaries or administration of antiestrogens can oppose this (Liao, D. Z., et al (1998) Carcinogenesis 19: 2173-80; Koibuchl, Y., et al (1999) Int. J Mol. Med. 4: 145-8). The mechanisms through which estrogen induces epithelia] growth in the breast are not clear, as ERa-containing epithelial cells in normal breast tissue do not proliferate (Zeps, N., et al (1998) Differentiation, 62: 221-226; Clarke, R. B., et al (1997) Breast Cancer Res. Treat. 45: 121-133). The prevailing concept is that estrogen stimulates secretion of growth factors from breast stroma and these factors stimulate epithelial cells to proliferate (Wiesen, J. F., et al (1999) Development 126: 335-344).

One possible explanation of estrogen action on epithelial cells is that it is the ERP-containing cells which proliferate. However, this does not appear to be the case in rodent mammary glands (Saji, S., et al (2000) Proc. Natl. Acad. Sci USA, 97: 337-342). In rodent breast tissue, ERP is constitutively expressed in approximately 70% of epithelial cells regardless of the endocrine state of the breast. ERU is highly expressed during lactation when it is co-expressed with ERP in over 70% of epithelial cells. In the breast tissue of virgin, pregnant and post lactation rodents, few cells express ERP, and there are very few cells where the two receptors are colocalized. During the highest proliferative phase of the breast i.e. pregnancy, there is very little expression of ERCJ and high expression of ERP. However, most of the cells which express the proliferating cell antigen contain neither receptor. These results seem to indicate that the presence of estrogen receptors in epithelial cells prevents these cells from proliferating. Such suppression could be explained in several ways, one of the simplest being that the estrogen receptors down regulate growth factor receptors.

Growth factors and their receptors have been extensively studied in the mammary gland (Lippman, M. E. (1996) Cancer Treatnient & Research 87: 26383; Marsh, S. K., et al (1999) Oncogene 18: 1053-60). It would therefore follow that malignancy would be a state where growth factor receptors have escaped from negative regulation by estradiol. In support of this theory, studies have shown that in sublines of breast cancer cells in culture, loss of ERcx is accompanied by an increase in growth factor levels and growth factor receptor levels, coupled with higher levels of phosphotyrosine residues, indicating an increased tyrosine kinase activity (EI-Ashry, D., et a/ (1997) Oncogene 15: 423-35).

Suggestions that expression of ERP may contribute to the initiation and progression of chemical carcinogen -induced neoplastic transformation in breast have come from one study where expression of ERP was induced in chemical carci no gen -transformed human breast epithelial cells (Hu, Y. F., et al (1998) Int. J Oncol. 12: 1225-1228). The more transformed cells had higher levels of ERP expression. In addition, the expression of mRNAs for both ERs in normal and malignant human breast tissue was determined by RT-PCR. This study showed that in normal breast tissue, expression of ERP predominated, with 22% of samples exclusively expressing ERP. This was not observed in any of the breast tumor samples investigated. Most breast tumors expressed ER(x, either alone or in combination with ERP. Interestingly, those tumors that coexpressed ERa and ER were node positive and tended to be of higher grade (Hu, Y. F., (1998) supra). In other RT-PCR studies, it was Z:' suggested that expression of ERP in human breast tumors could be a marker of endocrine responsiveness, i.e. loss of responsiveness, since the level of ERP mRNA was found to be lower in PR(progesterone receptor) tumors compared with PR- tumors (Dotzlaw, H., (1997) J Clin. Endocrinol. Wetab. 82: 2371-4; Dotzlaw, H., (1999) Cancer Res. 59:

3 529-32). ERP protein has also been detected in three human breast tissues of unspecified histopathology (Fuqua, S. A. W., et al (1999) Cancer Res. 59: 5425-5428).

The presence of significant amounts of ER(x in breast cancer tissue at the time of diagnosis is generally taken as an indication of hormone dependence (DeSombre, E. R., and Jensen, E. V., (1997) Cancer Medicine, 5th Edition pp 421-429) and on this basis, treatment with antiestrogen (tamoxifen) therapy is now the first-line therapy for metastatic disease (Encamacion, C. A., et al (1993) Breast Cancer Res. & Treatment 26: 23746). With optimal definition or receptor positivity, about two-thirds of patients with ER(x positive breast tumors will respond favourably to tamoxifen treatment or other endocrine manipulation, whereas a few patients classified as receptor-negative also respond (DeSombre, E. R., and Jensen, E. V., (1997) supra). Despite initial benefits, most patients on tarnoxifen therapy eventually relapse with tumors which not only have become tamoxifen resistant but which actually are stimulated by the medication. Loss of ERa does not account for this "acquired tamoxifen resistance" (Encamacion, C. A. et al (1993) supra). An animal model for this phenomenon suggests involvement of an ERcc mutation (Jordan, V. C., et al (1995) Endocrine-Related Cancer 2: 45-51). However, there may be other explanations, one being the possibility that the receptor becomes activated by a mechanism independent of a binding ligand (Encarnacion, C. A., et al (1993) supra). Recent studies have shown that ER(x can be activated in the absence of ligands (Gangolli, E. A., et al (1997) J Steroid Biochem. Mol. Biol., 61: 1-9; Tremblay, A., et al (1999) Molecular Cell 3: 513-519). Activation by phosphorylation via kinases in growth factor- activated pathways is a distinct possibility that requires further consideration.

In order to determine whether conclusions drawn from these rodent and human studies are applicable to all types of human breast malignancies, ERa and ERP proteins were measured using several techniques in various human samples obtained from both benign and malignant breast tissue.

Accordingly, a first aspect of the present invention provides a method of detection of breast cancer in cells comprising correlating the levels and/or combination or ERP isoforms 4 present in the cells with the absence or presence of breast cancer, wherein the presence of a particular ERP isofon-ns indicates the presence of a breast cancer, A further aspect of the invention provides a method of detection of breast cancer in a patient comprising: a) isolating a sample of breast tissue from the patient; and b) testing the sample for the presence of an ERP isoform; wherein the presence of an ERP isoform indicates the presence of a tamoxifen-resistant breast cancer.

The ER isoform to be detected may be ER2 (pex) and the presence of ERP (Pcx) may indicate the presence of a tamoxifen - resistant breast cancer type.

A further aspect of the invention provides a method of detection of breast cancer in a patient, comprising: a) isolating a sample of breast tissue from the patient; and b) testing the sample for the presence of cytosolic ERa; and c) correlating the level of cylosolle HRct detected with the presence of cancer cells in a the sample; wherein the absence of cytosolic ER(x or the presence of cytosolic ER(x at low levels indicates the presence of fibrocystic disease and/or medullary cancer, and the presence of cytosolic ERu. at higher levels indicates the presence of invasive ductal cancer.

A further aspect of the invention provides a diagnostic kit for the detection of breast cancer, comprising a means for the detection of the presence of ERP isoform(s) in a sample of cells. Preferably, the list includes anti-ERa antibody and an antibody for the detection of ERP isoforins.

Most preferably, the kit includes antibody for the detection of the ER 2(pcx) isoform.

Embodiments of the invention will now be described, by way of example only, and with reference to the accompanying Figures I to 5 in which.

Fig. I is a composite picture showing sucrose density gradient profiles and Western blots; Fig. 2 shows a section of breast stromal cells stained for the presence of ERP; Fig. 3 shows sections of breast tissue samples stained for the presence of ER(X and ERP; Fig. 4 shows human prostate cells stained for the presence of ERP (panel A) and stained to indicate the presence of tissue (panel B); and Fig. 5 shows: A) a schematic diagram of the reported exon-intron junction of human ERP sequences; B) a representative gel picture of RT-PCR analysis; and Q a table summarising the results of RNA expression studies.

The following abbreviations are used in the description: ER - estrogen receptor PgR - progesterone receptor RT-PCR - reverse transcriptasepolymerase chain reaction LBD - ligand binding domain HRT - hormone replacement therapy OCP - oral contraceptive pill DHC - depot hon-none contraceptive DCIS - ductal carcinoma in situ APAAP - alkaline phosphatase, antialkaline phosphatase PBS-T - phosphate buffered saline with 0.055% Tween 20 TBS - Tris-buffered saline

1) Antibodies The following antibodies were used:

The primary antibodies used were H222 monoclonal ERa antibody (Abbott) diluted 11500 and 1633, a polyclonal antibody, raised against the complete ERP ligand binding domain, diluted 115000. The secondary antibodies used in Western blots were raised in goats and were peroxidase conjugated anti-mouse and anti-rabbit immunoglobulins and anti-rabbit inmunoglobulins. Both were obtained from DAKO and used at 1/10,000 dilution. For 6 immunohistochemical studies the antibodies used were: a chicken anti-ER IgY which was I raised against the whole ERP protein and characterized previously in SaJi et al (Saji, S., et al (2000) Proc. Natl. Acad. Sci. US.4, 97: 337-342); mouse monoclonal ERoc 19G, rabbit anti-chicken peroxidase conjugated IgG (anti-chicken-HRP); and mouse alkaline phosphatase, anti-alkaline phosphatase (APAAP). All of these reagents were obtained from DAKO.

Table I shows the clinical histopathic details of the 24 breast samples which were analysed.

TABLE 1

Age Menopausal Pathology Histology L-V Nodal ERcc PgR & Hormone Size (Cm) type invasion status status 1 28 Pre OCP Fibrocystic + 2 31 Pre OCP Fibrocystic + 3 28 Pre OCP Fibroadenos - is 4 27 Pre Fibrosis - + 21 Pre Depo HC Fibroevstic - 6 54 Post never 0.-' DCIS No _'Ve Microinv.

7 50 unknown: 1.0 DCIS No 1. Not done + + i 8 47 Pre Inv. ductal 2 No Not done......

9 65 Post no HRT 4 Inv. ductal 3 Yes i-ve +++ +++ 56 Post HRT 3.3 Inv. ductal 2 No -ve +++ +++ 11 49 Post HRT 4.0 Inv. ductal 2 No -ve +++ 12 65 Post OCP 2.2 Inv. ductal 3 No Not done +± 13 58 Post 4.0 Inv. ductal 3 Yes 2/9 +ve +++ +±-] 14 59 Post 4 Inv. ductal 2 Yes -ve 43 unknown 2.6 Inv. ductal 2 Yes 3/8 +ve ++...

16 52 Post 5.5 ln,,. ductal 3 Yes 2/26 +ve 17 45 Pre 1.3&- 1.2 Inv. ductal 2 Yes 4/12 ±,,,c +++ +++ 18 42 Pre 5.5 fnv. ductal 2 Yes 35/35+ve......

19 43 Pre 1.9 Inv. ductal 2 Yes 3/18 +ve +++ 333 Pre 4.0 Medullary Yes 2/24 +ve - 21 47 Pre 3.3 Medullarv No -ve - 22 38 Pre OCP 2 Medullary No -ve - 23 54 unknown 4 Medullary No unknown - 24 51 Post never 4 Medullarv No -ve - DCIS=Ductal in-situ OCP= oral contraceptive pill DHC=depot hormone contraceptive HRT= hormone replacement therapy 7 Never = never used OCP or HRT 2) Sucrose densiql gradients Breast tissue samples were frozen in liquid nitrogen and pulverized in a dismembranator (Braun Melsungen, Germany) for 45 sec at 1800 RPM, Pulverized tissue was added to a buffer composed of 10 mM Tns chloride, pH 7.5, 1.5 mM EDTA and 5 mM sodium molybdate, using I ml per 100 tg tissue. Cytosol was obtained by centrifugation of the homogenate at 204,000 x g for I h in a 70Ti rotor at 4'C. MCF-7 cell cytosol was a generous gift from Abbott Laboratories.

Gradients were carried out as follows and as described in Greene et al (Greene, G. L., et al (1977) Proc. NatI. Acad. Sci. USA., 74: 3681-3685). Breast tissue and tumor extracts were incubated for 3 h at OOC with 10 nM ['H] estradiol- 1 7P either in the presence or absence of a 50-fold excess of radio inert 17P estradiol, and the bound and unbound steroids were separated with dextran-coated charcoal. Sucrose density gradients [10-30% (w/v) sucrose] were prepared in a buffer containing 10 mM Tris- HCI, 1.5 mM EDTA, I mM (x-monothioglycerol, and 10 mM KCL Samples of 200 pI were layered on 3.5 ml of gradients and centrifuged at 4'C for 16 h at 300,000 x g in a Beckman L79K ultracentrifuge with SW-60Ti rotor. Successive 100 d fractions were collected from the bottom by paraffin oil displacement and assayed for radioactivity by liquid scintillation counting. For Western blotting, &actions were first precipitated with TCA, and the precipitate resuspended in methanol. Samples were placed on dry lee for 30 min and the protein recovered by centrifugation. Protein pellets were dissolved in SDS sample buffer and resolved by sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) in gradient gels 4-20%.

The results can be seen in Table 2.

TABLE2

Patient Estrogen binding Western blotting Western blotting ERP isoform # frnol/mg ERcc ERP 8S 4S cytosol nucleus cytosol nucleus I neg 329 ++ Pos... POS 2 neg 23 neg neg ++ pos PCX>P 1 3 neg 407 neg pos... POS 8 On sucrose uradients with lOnM ['H]estradiol as the ligand, as depicted in Flo. 1, there are two distinct peaks of estradiol binding detectable in low salt extracts of human breast samples. In Fig. 1, sucrose density gradients were performed with low salt extracts of breast samples in the presence of 10riM [3H] estradiol. Every fraction from the gradients was precipitated and analyzed by Western blotting. Filters were probed with ER(x monoclonal H222 (lower blot) and ERP LBD IgG (upper blot). Limitation in sample size prevented reliable analysis of samples 11, 12 and IS. There are clear differences between samples in the quantity of estradiol bound and the ratio of the two peaks. ERU is known to sediment as an SS peak in low salt gradients (Greene, G. L., et al (1977) supra) and ERP as a 4S peak (Saji, S., et al (20001) supra). ERu in MCF-7 cytosol was used as a control in all gradient runs and it sediments as an 8S peak with no evidence of a 4S peak.

The specific activity of the estradiol in the binding assay was 84 cpm/fmoI and the protein content in each cytosol aliquot was about 100 pg. With a background count of around 30 cpm/tube, and since the peaks were spread into as few as two fractions and as many as 15 fractions, a lower limit of sensitivity of about 20 fmol/mg cytosolic protein was achieved. Cytosolic ERa was not detectable in this binding assay in medullary cancer or in fibrocystic disease. With invasive ductal cancer samples, 8 out of 12 samples had levels of

4 neg - -I-± - + pos Pos neg neg + 4- POS ++ Pos Pex>>P1>P5 6 neg 59... POS _+++ POS 7 120 54 neg POS Pos 8 12.3 104 nec POS Pos 9 3,700 neg i POS Pos only Pex neg 95 Pos... Pos Pl>P5=Pex 11 181 neg ne. POS POS 12 482 neg neg neg pos 13 13 47.5 neg Pos + + pos 14 neg 114 neg Pos pos Pcx=P5 no P I 22.6 190 + POS ++ Pos 16 neg neg POS _++ pos Pex=pl=P5 17 neg neg neg pos _++ Pos 18 472 330 ±+ POS Pos 19 213 neg ++ neg ++ POS neg neg ++ Pos pos 21 neg neg neg Pos pos Pex>>P5>>P1 22 neg 250 neg POS Pos 23 neg 919 neg neg Pos Pcx>p]=P5 24 neg 68 + Pos... pos 9 cytosolic ERu_ ranging from 13-3,700 fmol/mg protein. Unactivated ERCX was present in 15 out of 24 samples and did not appear to segregate with breast disease type.

3) Western blotting of low salt extracts Every fraction from each sucrose gradient was precipitated, resuspended in SDS sample buffer, divided into two equal parts, and analyzed by Western blotting for the presence of ERa and ERP.

All tissue handling was done at 40C. Nuclear ER was extracted with decavanadate as described by Fritsch et al. (Fritsch, M., et al (1999) Biochemistly, 38: 6987-6996). Pellets from the low salt extracts were homogenized for a few seconds using Polytron PT3100 apparatus in 4 volumes of PBS containing 2 mM decavanadate, I mM EDTA, pH 7.4 and two protease inhibitor cocktail tablets (Boehringer-Manheim, Germany) per 50 ml. The homogenates were then centrifuged for I h at 105,000 x g to obtain the supernatants as tissue extracts, and the protein contents of the tissue extracts were measured using a Bio-Rad Protein Assay with BSA as the standard.

Proteins were resolved on SDS polyacrylamide gels, either 9% polyacrylamide or prefon-ned gradient gels of 4-20% acrylamide (NOVEX, San Diego, CA), with a Tris-glycine buffer system. Transfer to ProBlot membranes (Applied Biosystems) by electroblotting was either by semi-dry blotting or in a tns-glycine buffer. The following molecular weight markers were used: Kaleidoscope prestained standards (Bio-Rad Laboratories), Low molecular weight electrophoresis calibration kit (Amersham Pharmacia, Sweden) and High range protein molecular weight markers (Promega, Madison, WI). Signals were developed using "SuperSignal West Fernto Maximun Sensitivity Substrate" from PIERCE.

ERa was present only in the 8S peak, never in the 4S peak. There was one sample [4] which did not bind estradiol but which had distinct and specific bands of the correct size on Western blots with the H222 antibody. The 4S peak always contained ERP with a molecular weight of 6063 kD. There were five samples [9, 16, 17, 20 and 21] which had no estrogen binding in the 4S peak but had distinct and specific signals of the correct size on Western blots. In some of these samples ERP was highly expressed and was found throughout the gradient (see Fig. I). Vanadate extracts of breast nuclear fractions all I'D In contained ERP by Western blotting.

As indicated in Table 2, ERct was present in all but four of the samples. If the presence of ER(x in the nucleus after low salt extraction is an indication of activated receptor, then in some breasts ERP can be activated while ER(x remains unactivated. Surprisingly, four women identified clinically as postmenopausal [13, 14, 16 and 24] and who were not ingesting any known ER ligands, had activated ERck.

4) Correlation between ER(x, ERP and Node Status From Table 2, if the breast samples are sorted into two orroups, node negative and node L_ positive, it is clear that patients whose cvtosols show only ER are node negative, whereas node-positive patients all have ERa present in the cytosol.

There was a good agreement between ERct immunohistochemistry and the 8S form of ER (p 0.007; k 0.5 1).

Only a single Western blot of breast tissue was negative for ER. There was no correlation between ER(x and ERP by any method of measurement. There -,vas a significant correlation between PgR status as measured by immunohistochemistry and ERa measured by immunohistochemistry (p<0.001) or by sucrose density gradients (p=0.007), but these correlations did not reach statistical significance when PgR was determined by a ligand binding assay. There was no association between PgR and ERP. There was evidence that Z.7 1 presence of ER(x correlated with invasive ductal carcinoma as opposed to medullary carcinoma, which were all ER(x negative by immunohistochemistry (p0.003). Similarly, there was a significant association between ERa as measured by 8S estrogen binding and invasive ductal carcinoma when compared to fibrocystic disease (p=0.003). There was no evidence of association between ERP (as measured by binding of estradiol in the 4S peak or by Western blotting) with histopathology.

11 In the following experiments, all immunohistochemistry and immunostaining were carried out as described below:

6) Immunohistochemistry All immunohistochemistry was performed on paraffin sections of breast tissue obtained from the histopathology archive at Channg Cross Hospital, London. In all experiments, paraffin sections (2gm) were dewaxed in xylene and rehydrated through graduated alcohol to water. Endogenous peroxidase was blocked by incubation for 15 min with a solution of 0.6% hydrogen peroxide and antigen retrieval was performed by microwaving sections in 0.0 1 M citrate buffer, pH 6.0, for 20 min at 800 watts.

a) Single immunostaining with chicken anti-ERP Tissue sections were incubated for 10 min at room temperature with normal rabbit serum diluted at 1/20 in PBS containing 0. 1 % bovine serum albumin and 0. 1 % sodium azlde (antibody diluent). Chicken anti-ERP (1/ 1000) in antibody diluent was then applied and the sections incubated overnight at 4'C. Negative controls consisted of substitution of the primary antibody with PBS and, as a further specificity control, primary antibody was used after being absorbed with its antigen. Prior to addition of the secondary antibody, sections were rinsed in PBS containing 0.05% Tween 20 (PBS-T). Rabbit anti-chicken IgG-HRP (I/1000 dilution) in PBS was applied to the sections. After 60 minutes, the sections were washed with PBS-T. Peroxidase was developed with hydrogen peroxide and diaminobenzidine tetrahydrochloride and a hematoxylin counterstain was applied. The sections were then dehydrated through graduated alcohol to xylene and mounted with Pertex (Histolab Products AB, Gothemburg, Sweden).

b) Single Immunostaining with anti ERct monoclonal antibody Tissue sections were incubated for 10 min in normal rabbit serum diluted at 1/20 in antibody diluent. This was followed by overnight incubation at 4'C with ERoc monoclonal (Dako) at 1/100 dilution in antibody diluent. For the negative control, PBS was substituted for the primary antibody. Prior to addition of secondary antibodies, sections were rinsed with PBS- T. Rabbit anti-mouse IgG at 1/100 dilution was applied 12 to the sections and after 20 minutes, the slides were washed with PBS-T followed by a TBS wash. Mouse APAAP,I.j"50 dilution, was applied for 30 minutes, and the alkaline phosphatase was developed with naphthol AS-MX and fast Blue BB.

C) Double immunostaininig, for ER(x and ERP Tissue sections were incubated for 10 min at room temperature with normal rabbit serum diluted 1 /20 in antibody diluent. This was followed by an overnight incubation at 4'C with a mixture composed of ERcj. monoclonal (1/100 dilution) and chicken ER IgY (1/1000 dilution). PBS was used in place of this mixture in the negative controls. Prior to addition of secondary antibodies, the sections were washed with PBS-T. A mixture of rabbit antimouse I-G (1/100 dilution) and rabbit anti-chicken IgG (1/1000) was added and after 30 min of incubation, the slides were washed with TBS. A second mixture composed of APAAP (1,150) and rabbit anti- chicken IgG (1/1000) was added and incubation continued for 30 min. After washing, alkaline phosphatase was developed with naphthol ASMX to produce a blue color and peroxidase with hydrogen peroxide and 3-amino-9-ethyl carbazole to give a red color.

Immunohistochemistry for ER and double staining for ER(x and ERP was performed on 6 of the 24 breast samples. These consisted of a norinal breast [sample 4], a ductal carcinoma in situ (i.e where the carcinoma has not spread) [sample 7], three invasive ductal carcinomas [samples 9, 11, & IS], and a niedullary carcinoma [sample 23]. Examples of this staining are shown in Fig.2. Five out of the six samples were clearly ERP- positive. Samples 4 and 23 were predominantly ER-positive with, very few ER(X-containing cells. Fig 2A shows nuclear staining of ERP in stromal cells from a norinal breast [sample 4]. A further example of stromal ER staining is shown in Fig. 2B [sample 18]. Double staining for ERa and ERP revealed that there was variability in the degree of co-localization, with some patients having a majority of cells co-expressing the receptors [samples 7, 11 and 18] while in samples 4, 9 and 223, the majority of cells were either ERU or ERP positlve with relatively few areas of colocalization. Fig 3 shows areas where there is co-localization. After absorption of the ERP antibody with its antigen, no immunostaining was seen (Fig 4). Other negative controls Nvere similarly unstained.

7) Preparation of RNA and RT-PCR Total RNA was extracted from frozen breast tissue sections using RNAwiz (Ambion, Austin, TX) according to the manufacturer's instructions, and quantified spectrophotometncally. 5 tg of total RNA was treated with 5tg of RNase-free DNase (Promega, Madison, Wl) for 60 min at 37C to remove genomic DNA from the RNA samples. After inactivation of DNase for 10 min at 65T, samples were reverse-transcribed using SuperScript preamplification system (Life Technologies, Inc., Gaithersburg, MD) in a final volume of 20Ll according to the manufacturer's instructions.

I The primer sets for ERP variants were as follows:

ERP LBD U, 5'GAGCTCAGCCTGTTCGACC; ERP LBD L, 5'-GGCCTTGACACAGAGATATTC; ERP delta 5 U, 5'-ATGATGATGTCCCTGACCAAG ERP IU, 5'CGATGCTTTGGTTTGGGTGAT (Moore, J. T., et at (1998) Biochem. Biophys. Res. Comm., 247: 75-78); ERP I L, 5'GCCCTCTTTGCTTTTACTGTC (Moore, J. T., et at (1998) supra); and ERP 2L, 5'CTTTAGGCCACCGAGTTGATT (Moore, J. T., et at (1998) supra).

The location of each of the primers and the size of their respective products are shown in Fig.5 (P-Actin primers were 5CTGGCACCACACCTTCTAC for sense and 5'GGGCACAGTGTGGGTGAC for antisense).

mRNAs from 8 representative breast samples were evaluated by RT-PCR. Six primer pairs were designed to amplify specific PCR products from each variant mRNA.

Identification of ERP isoforms by RT-PCR RT-PCR with various specific primers was carried out to examine whether there was a correlation between the presence of ERP variants and lack of estradiol binding. As shown in Fig.5A, six primer pairs were designed to detect the following ERP isoforms:

14 1) the exon 5 deletion isofon-n (65) (Vladusic, E. A., et al (1998) Cancer Res. 58: 210-214) 2) 18 amino acid insertion in the LBD (Hanstein, B., et al (1999) Afol. Endocrinol. 13:

129-137; Maruyama, K., et al (1998) Biochenz. Biophys. Res. Commun. 246: 142-147) 3 3) truncated variants of C-terminal regiori such as (2 (pex) and P5 (Moore, J. T., et al (1998) Biochem. Biophys. Res. Coinni. 247: 75-78; Ogawa, S., et al (1998) Nucleic Acids Res. 26: 3505-35 12).

The results are shown in Fig. 5 and are summarized in Fig. 5B and 5C.

In panel A of Fig. 5 the reported exon-intron junction of human ERP is indicated. Models of each variant protein and the location of primers are shown. The estimated sizes of PCR products which arise from each of the primer sets are indicated on the right side. Panel B of Fig. 5 shows a representative picture of RT-PCR analysis. The wild type and variant cDNAs retrotranscribed from each sample's mRNA were amplified as described earlier and were visualized on 2% agarose gel with ethidiurn bromide. The primer sets are I indicated on the left side and arrows on the right side indicate the cDNA from which the PCR product is derived. In panel C of Fig. 5 is a summary of results from triplicate experiments. The presence of absence of variant mRNAs is indicated as - or -, respectively. For evaluation of the relative expression of each variant (wt, P2(Pcx), P5) using primers I U, I L and 2 L, PCR product band intensity was evaluated and intensities are indicated as -++, ++, -+. None of the tested sampleshad LBD insertion isoforms, (2nd panel of Fig. 5B) (Tremblay, A., et al (1999) Molecular Cell. 3: 5 11-519). The 8S variant of ERP was detected in samples 5, 16 and 21 which had no estradiol binding on sucrose gradients as well as in samples 10, 14, and 23, which had no estradiol binding in the 8S peak but did have a 4S peak (see first panel of Fig. 5B). Triple primer PCR, used for evaluating the ratio of the various isoforms (see third panel of Fig. 513), revealed that the expression of C-terminally truncated isoforms is common. Sample 9 (invasive ductal carcinoma, grade 3) had CXCILISIvelv EPP2 (ERP cx) and had very high binding of estradiol in the 8S peak. Sample 14, which had estradiol binding in the 4S peak on sucrose gradients had no wild type (wt) C-terminal sequence.

Therefore, it appears that ERP is present in breast tissue whether it is benign or malignant.

In summary:

(1) The ERP C-tenninal variant, ERpcx, reported to be repressor of ER(x action (Ogawa, S., et al (1998) Nucleic Acids Res. 26: 3505-3512), is quantitatively important in the breast.

(2) The major ER in human breast stroma is ERP.

(3) Cytosolic ERoc is usually absent or present at very low levels in fibrocystic disease and in medullary cancer, whereas high levels of cytosolic ERa can be found in invasive ductal cancer.

This study is the first to carry out successful immunohistochemistry for ERP on paraffin sections and to perform dual staining for ERa and ERP. Previously, frozen breast sections have been used for ERP immunohistochernistry and for staining of serial sections for ERCC and ERP (37). With frozen sections, as in the present study, ERP and ERP are found to be coexpressed in the majority of ER-expressing cells in some patients. In the other patients, there are few areas of colocalization with the majority of cells positive for either ERU or ERP. Coexpression has biological implications given the possibility of heterodimerization between the two receptors, but the occurrence and significance of heterodimerization of ERu. and ERP in normal and malignant breast tissue has yet to be established. In the stroma, ERct was the only ER detected. The lack of detectable ER(x in human breast stroma has been observed previously (McCarty, K. S. Jr., et al (1986) Cancer Res. (suppl.) 46: 4244s-4248s). In rodents, ER(x is present in the breast stroma and is responsible for secretion of epithelial cell mitogens (growth factors) in response to estradiol. Tissue recombinants with stroma and epithelium from ERcc knock out mice demonstrated that ERcc in the stroma is responsible for epithelial growth in response to estradiol (Cunha, G. R., et al (1997) J Mam. Gland Biol. Neoplasia, 2: 393-402). Whether the distribution of ERs in different cellular compartments of the breast is different in rodents and humans or whether there is variation between women needs to be further examined.

16 Another unexpected finding of the present study is the presence of ERP isofonris which do not bind estradiol but are clearly present as proteins as indicated by Western blotting. The presence of ERP isoforms in these samples has been examined by RT-PCR and the data clearlv indicate that low estrogen binding in the 4S part of the gradient is associated with high levels of ERP2 (pex) variant and low levels of EPPI. ERP isoforms have been previously descnbed in breast cancer at the mRNA level (Leygue, E., et al (1999) Cancer Res., 59: 1175-1179, Vladusic, E. A. et al (1998) supi-a; Hanstem, B., et al (1999) supra). However, the corresponding proteins have not been studied. In addition to extensions of the N-tenninus, three groups have reported the cloning of a 503 amino acid long ERP isoform with an IS amino acid residues in-frame insertion in the LBD (Chu S. and Fuller PJ: (1997) Mol Cell Endocrinol 132: 195-199; Petersen DN. et at. (1998) Endocrinology 139: 1082-1092; Hanstein B et al. (1999) Mol Endocrinol 13: 129-137), in the splice junction inbetween exon 5 and 6 (Enmark E ei al. (1997) J Clin Endocrinol Met 82: 4258-4265). In contrast to the N-terminal extended ER isoforms the affinity of ER-503 for 17p-estradiol (E2) and other known estrogen receptor ligands was shown to be lower than for ER(x and for ERP-485. As a consequence of that ERP-503 was shown to require 100-1000-fold higher concentration of E2 for activation of transcription of a reporter gene (Petersen DN et al. (1998) supi-a, Hanstein B et al. (1999) supra). Others have, however, reported that the ERP-503 isoforin does not bind ligand and that it acts as a dominant negative regulator of estrogen action by suppressing the estrogen-dependent ERU- and ERP- mediated gene transcription activation, respectively (Maruyama K et al. (1998) Biochem Blophys Res Comm 246: 142147). Both ERP-503 and ERP-485/530 bind to a consensus estrogen response element (ERE) and heterodimerize with each other and with ER(x (Petersen DN et al. (1998) supi-a, Hanstein B et al. (1999) supm). The coactivator SRC-1 has been shown to interact with both ERa and ERP-485 in an estrogendependent manner but not with ERP-503) (Hanstein B et al. (1999) supra), which, in part, may explain the need for very high concentrations of E2 to activate target gene expression by ERP-503 (Petersen DN et al. (1998) supra, Hanstein B et al. (1999) supi-a). Transenipts encoding additional ERP isoforms with variations at the extreme C-terminus have been found in testis cDNA libraries (0gawa S et al. (1998 b) Biochem Biophys Res Comm 243: 122-126; Moore JT et al. (1998) Biochern Biophys Res Comm 247: 7578). ERpcx (Ogawa S et al.

17 (1998 b) supra) is identical to ERP-530 except for the last 61 C-tenninal amino acids (exon 8 encoding part of helix I I (H 11) and H 12), which are replaced by 26 unique amino acid residues. Due to the exchange of the last ERP-530 exon the ERPcx isoform lacks amino acid residues important for ligand binding as well as amino acids that constitute the core of the activation function 2 (AF2), previously shown to be important for 11 gand -dependent receptor: coactivator interactions (Danielian P S. et al. (1992) EMBO J. 11: 1025-1033; Henttu PMA et al. (1997) Mol Cell Biol 17: 1832-1839; Shibata H et al. (1997) Recent Prog Horm Res 52: 141165; Brzozowski AM et al. (1997) Nature 389: 753-758; Shlau AK et al. (1998) Cell 95: 927-937; Danmont BD et al. (1998) Genes Dev 12: 3343-3356; Feng W et al. (1998) Science 280: 1747-1749). It was therefore not an unexpected finding that the ERpcx isoform showed no ligand binding activity or had no capacity to activate transcription of an estrogensensitive reporter gene (Ogawa S et al. (1998 b) supra). Surprisingly, however, ERPcx was reported to have no affinity for a consensus estrogen response element (ERE) in a gel-shift assay. Furthermore, ERPcx showed preferential heterodimerization with ER(x rather than with ERP, inhibiting ERa DNA binding. Functionally the heterodimerization of ERPcx with ERa was reported to have a dominant negative effect on ligand- dependent ERa reporter-gene transactivation (Ogawa S et al. (1998 b) supra). Of the five ERP isoforms (ERP 1-5) described by Moore et al. (Moore JT et al. (1998) Blochem Biophys Res Comm 247: 75-78), ERl corresponds to the previously described ERP-530 (Ogawa S et al. (1998 a) supra) and the ERP2 variant is most likely identical to ERPcx (Ogawa S et al. (1998 b) supra). However, ERP 3-5 are novel splice variants with an exchange of the last exon of ERP-530 for previously unknown 3'exons (Moore JT et al. (1998) supra). Similar to ERPcx, neither of the novel C- terminal splice variants, ERP 3-5, can be expected to bind ligand or activate transcription from an ERE driven reporter vector as they all, like ERPcx, lack amino acids important for ligand binding and the core of AF2. However, in contrast to what was reported for ERPcx (Ogawa S et al. (1998 b) supra), two of the C-terminal splice variants, ERP 2 and 3, were shown to bind to a consensus ERE (Moore JT et al. (1998) supra). Sample 9, in the present study, has ERPcx as the dominant ERP isoform but clearly has ERP protein on Western blots, indicating that EPP2 (Pcx) is, indeed, expressed as a protein. Furthermore, this sample had very high levels of binding of estradiol in the 8S is region of the sucrose gradient where ERct was abundant. These data indicate that EPP2 (Pcx) does not prevent binding of estradiol to ERcf,. Post menopausal women have low levels of circulating estrogen, yet the present experimental results show that such patients have ERa tightly bound in the nucleus of their tumor cells. In order to develop a fuller picture of the role of antiestrogens in breast cancer treatment, both ER(X and ERP must be measured, not only in the cytosol but also in the nucleus. Furthermore, it is not sufficient to speak of ERP alone in breast tissue since ERP isoforrns, which have defective estrogen binding, and which can inhibit ERct and ERP transcriptional activation, are present in the tissue.

ERP2 (Pcx) has a dominant negative effect on ERcf. and wild type ERP and does not bind I to estradiol. Therefore, if ERP-2 (Pex) is present in a tumour cell, the protective effect of ERP (as an inhibitor of hori-none-induced cell proliferation) is lost, resulting in uninhibited hormone-induced cell proliferation. The resulting tumour will also exhibit resistance to treatment using Tamoxifen due to the dominant negative effect of ERP2 (cx).

19

Claims (9)

Claims

1. A method of detection of breast cancer in cells comprising correlating the levels and/or combination of ERP isoforrns present in the cells with the absence or presence of breast cancer, wherein the presence of a particular ERP isofon-n indicates the presence of breast cancer.

2. A method of detection of breast cancer in a patient comprising: a) isolating a sample of breast tissue from the patient; and b) testing the sample for the presence of an ERP isoform; wherein the presence of a ERP isoform indicates the presence of a tamoxifen-resistant breast cancer.

3. A method according to claim I or claim 2 wherein the ERP isoform is ERP2(pex).

4. A method according to claim 3 wherein the presence of ERP2(pcx) indicates the presence of a tamoxifen-resistant breast cancer type.

5. A method of detection of breast cancer in a patient, comprising: a) isolating a sample of breast tissue from the patient; and b) testing the sample for the presence of cytosolic ERa; and c) correlating the level of cytosolic ERu. detected with the presence of cancer cells in a the sample; wherein the absence of cytosolic ERcc or the presence of cytosolic ERO, at low levels indicates the presence of fibrocystic disease and/or medullary cancer, and the presence of cytosolic ERa at higher levels indicates the presence of invasive ductal cancer.

6. A diagnostic kit for the detection of breast cancer, comprising a means for the detection of the presence of ERP isoform(s) in a sample of cells.

7. A diagnostic kit according to claim 6 in which the kit comprises an ERa selective antibody and at least one antibody for the detection of ari ERP isoform.

8. A diagnostic kit according to claim 7 in which the kit comprises antibodies for the detection of at least two different ER isofomis.

9. A diagnostic kit according to claim 7 or 8 in which the kit includes an antibody for the detection of the ERP2 (pex) lsoform.