WO2004070059A2 - Polymorphisms in the esr1 gene and diagnosis of an estrogen-sensitive disorder - Google Patents

Abstract

The method of the invention is useful for diagnosing an increased or decreased risk of an estrogen-sensitive disease in an individual. The evaluation of the risk of disease comprises determining whether the ERα gene or 3'UTR an individual comprises one or more of the polymorphisms defined herein as ISOE1, ISOE2, ISOE3, ERNE1, ERNE2, ISOC1, 2a1, 2a2, 2a3, 2a4, 2a5, 8B1, 8B2, 8C, 8E, 8H or 8I.

Description

DIAGNOSIS OF AN ESTROGEN - SENSITIVE DISORDER

Field of the Invention

This invention relates to the identification of numerous genetic variations in the estrogen receptor alpha (ERα) gene, which may be of importance in the diagnosis of disease.

Background of the Invention

Estrogen has an essential role in both male and female physiology. Responses to estrogen are observed in a diverse range of important systems, impacting the growth and function of breast and uterine tissue, the regulation of plasma cholesterol, vasomotor function, and bone homeostasis. The actions of estrogen are mediated via both classical steroid, nuclear receptor mechanisms, as well as by newly described non-nuclear pathways. Variations in any of the components of these signalling pathways might alter estrogen sensitivity and have important influences on an individual's risk of estrogen-responsive diseases such as osteoporosis, breast cancer, and atherosclerosis. The role of estrogens in regulating cellular metabolism in many tissues is well documented. Estrogens regulate cellular activity by interacting with specific intracellular receptor proteins. Two estrogen receptors have been identified to date: estrogen receptor alpha (ERα)¹ and estrogen receptor beta (ERβ)². The ERs belong to the steroid nuclear receptor family and act to regulate the transcription of estrogen responsive genes in a ligand-dependent manner.

The identification of receptors for estrogen advanced from an early stage largely through work performed in the laboratories of Jensen and Gorski ³'⁴. An open reading frame of 1785 nucleotides in the cDNA was found to correspond to a polypeptide of 595 amino acids and a molecular weight of 66200 Da ⁵'⁶' In addition, sequencing of the entire cDNA revealed a 4565 bp S'UTR ¹.

The ERα gene occupies a chromosomal location of 6q25.1 ⁷. The common 8 exons of ERα span more than 140 kb ⁷. In addition, at least 6 alternative exons extend further upstream ³⁴'³⁵.

Significant phenotypic correlations between bone parameters, and breast cancer risk ⁸'⁹, as well as atherosclerosis ^10"13, and insulin resistance ¹⁴ provides epidemiological evidence for the systemic nature of estrogen signalling within individuals. Differences in estrogen-dependent phenotypes may relate to levels of estrogen exposure, or to factors involved in the response to the hormone ("estrogen-sensitivity"). In turn, variation in response to estrogen may depend on a change in either the function of the receptor, its level of expression, or on variation in downstream pathways. Furthermore this variability may be either environmentally or genetically determined. Thus, the identification of genetic variants in estrogen signalling pathways may significantly influence both the beneficial and adverse effects of estrogen in the breast, bone and cardiovascular system. As with most human disease, breast cancer, osteoporosis, and cardiovascular disease are viewed as having a significant genetic component, and considerable effort is being made to find and study the susceptibility genes. Genetically-determined variable estrogen signalling may be a common factor in a number of these estrogen-sensitive disease processes, contributing towards both their heritability, and their correlation.

Screening of the ERα gene has identified numerous polymorphisms. However, to date, no common variants have been shown to change amino acids, or to have a significant effect on either the receptor expression, or function. Some of these variants, particularly two intron 1 restriction fragment length polymorphisms (RFLP: Xbal and PvuII) and an upstream variable dinucleotide repeat, have been examined for association with estrogen- dependent phenotypes. Phenotypes in which association with ERα polymorphisms have been demonstrated include risk of breast cancer ¹⁵'¹⁶ and fibroids ¹⁷, as well as the quantitative traits bone mineral density ^18"21, and body-mass index ²². In addition, a number of cardiovascular intermediate phenotypes have been associated with ERα polymorphisms. However, a lack of consistency between different studies, as well as a paucity of biologically plausible function for the examined variants has been problematic. Summary of the Invention

The present invention is based on the identification of several polymorphisms on the ERα gene, or its 3'UTR, which are implicated in estrogen-sensitive disease.

According to a first aspect of the invention, a method for diagnosing an increased or decreased risk of estrogen-sensitive disease in an individual, comprises the step of identifying whether the individual comprises one or more of the polymorphisms in the ERα gene or 3'UTR designated herein as: IsoEl, IsoE2, IsoE3, ERNEl, ERNE2, IsoCl, 2al, 2a2, 2a3, 2a4, 2a5, 8B1, 8B2, 8C, 8E, 8H and 81.

According to a second aspect of the invention, an isolated polynucleotide, useful in diagnosing whether a subject has or is predisposed to an estrogen-sensitive disease, comprises at least 15 contiguous nucleic acids and hybridises under highly stringent conditions to a region of the ERα gene or the 3 'UTR comprising a polymorphism specified above, or its complement.

According to a third aspect of the invention, a diagnostic kit comprises a polynucleotide as defined above contained in an individual container.

Brief Description of the Drawings

The invention is described with reference to the accompanying drawings, wherein:

Figure 1, is a chromatogram illustrating the formation of the three heteroduplexes in AmplEb, the x axis displays column retention time in minutes, the y axis represents absorbance;

Figure 2, illustrates the location and nature of ERα polymorphisms;

Figure 3, illustrates the results of a restriction enzyme digest of three polymorphisms in Amp-lEb; Figure 4, illustrates the results of a restriction enzyme digest of two polymorphisms in ERNEl adn ERNE2;

Figure 5, illustrates the results of WANE analysis of amplicon 8E;

Figure 6, shows electropherograms demonstrating the apparent length variation in amplicon 8Ea between different individuals; Figure 7, shows the allelic variation at ERNE 1 polymorphic locus, resulting in the introduction of a puntative NF-1 binding site;

Figure 8, illustrates the functional role of the ERNE2 (C>T) polymorphism, resulting in loss of negative transcriptional regulation by ERNE; and

Figure 9, illustrates the role of the 3'UTR of the ERα mRNA in the regulation of message stability; the 1000 bp UTR 2 segment is shown to be responsible for most of the regulation; sub-segmentation of the UTR 2 into three pieces (A, B and C) disrupted the instability of the region and recombination of A and B, but not A and C, or B and C resulted in restoration of instability. Description of the Invention The present invention utilises known methods of genetic analysis to determine whether a particular subject has, or is predisposed to, an estrogen-sensitive disorder. As used herein, "genetic predisposition" refers to an increased likelihood that a given subject has or is likely to develop an estrogen-sensitive disorder, given the presence of a particular genomic sequence (polymorphism).

The reference to "estrogen-sensitive disorder" is intended to refer to the related group of diseases, such as, cardiovascular disease, artherosclerosis, osteoporosis and insulin resistance including obesity.

The term "allele" is used herein to refer to variants of a nucleotide sequence. The term "polymorphism" as used herein refers to the occurrence of two or more alternative. genomic sequences or alleles between or among different genomes or subjects. A single nucleotide polymorphism (SNP) is a single base pair change. Typically, a SNP is the replacement of one nucleotide by another nucleotide at the polymorphic site. Deletion of a single nucleotide, or insertion of a single nucleotide, also gives rise to a single nucleotide polymorphism.

The term "highly stringent hybridisation conditions" refers to conditions which permit only complementary polynucleotides from hybridising. Suitable conditions include: Overnight incubation at 42°C in a solution comprising: 50% formamide, 5x SSC (150 mM NaCl 15mMtrisodium citrate), 50mM sodium phosphate (pH 7.6), 5xDenhardt's solution, 10% dextran sulfate, and 20 μg/ml denatured, sheared salmon sperm DNA, followed by washing the filters in 0. lx SSC at about 65°C. Having identified the SNPs, it will be apparent to the skilled person how to detect polymorphisms for a particular subject, and to make a diagnosis. Methods for the detection of a polymorphism are known in the art, and include: polymerase chain reaction- restriction fragment length polymorphism (PCR-RFLP) (e.g. Ju et al, Proceeding of the Eleventh International Histocompatibility Workshop and Conference, Vol. 2, 1992, pgs 317-319); PCR sequencing; ligase chain reaction (LCR) Abravaya et al, 1995, NAR, 23(4): 675-682); oligotyping using Sequence Specific Primers (SSP) (e.g. Olerup et al, Proceedings of the Eleventh International Histocompatibility Workshop and Conference" Vol. 2, 1992, pgs 315-317); oligotyping using Sequence Specific Oligonucleotide Probes (SSOP) (Tiercy et al, Immunobiology of HLA, Vol. LI, pp. 248-250, 1987, Springer Nerlag, New York); Single-stranded conformation polymorphism (SSCP) (Yap et al, Feb/1992, Trends in Genetics, 8(2):49; andOritaetα/., 1989, Genomics, 5: 874-879); and direct sequencing of the relevant genomic region. The present invention is not limited to any particular method for detecting a polymorphism.

A preferred method for the identification of the presence of a SNP is to use the LightCycler system (Lohmann et al, Biochemica, 2000; (4): 23-28) developed by Roche Molecular Biochemicals. The LightCycler system enables the amplification and real-time detection of a polynucleotide, allowing accurate quantification. The system permits the detection and genotyping of single nucleotide polymorphisms by utilising a function known as melting curve analysis. During the melting curve analysis the LightCycler instrument monitors the temperature-dependent hybridisation of sequence specific hybridisation probes to single stranded DNA.

An alternative detection method is to use what are referred to in the art as "Molecular beacons" . Molecular beacons are oligonucleotides designed for the detection and quantification of target nucleic acids. The oligonucleotides usually comprise self- hybridising portions that, in the absence of a target nucleic acid, form a stem-loop structure. A fluorescent moiety and a quencher moiety are attached at each end of the oligonucleotide, and are positioned adj acently when the oligonucleotide is in the stem-loop orientation. Fluorescence is effectively prevented by the quencher moiety in this orientation. The loop portion of the oligonucleotide is complementary to a specific target nucleic acid and, in the presence of the target, hybridisation to the target occurs disrupting the stem loop orientation, separating the fluor and quencher, resulting in an increase in detectable fluorescence. The use of the molecular beacons approach to the detection of SNPs is disclosed in US 6548254.

Using the known sequence information for the ERα gene region or the 3'UTR, , including the polymorphisms disclosed herein, it is possible to design hybridisation probes for use in the LightCycler system, or any hybridisation based system.

The design of suitable polynucleotide/hybridisation probes will be apparent to the skilled person. The probes will usually comprise the polymorphic site, e.g. the SNP.

The polynucleotides/hybridisation probes may be detectably labelled, e.g. fluorescently labelled, using methods and labels known in the art, e.g. as used in the detection methods referred to above.

The polynucleotides/hybridisation probes may be immobilised to a support surface, for use in a diagnostic assay. Suitable support materials are known in the art and include, ceramics, plastics, glass and silicon materials. Methods for immobilising polynucleotides to a support material are also known in the art. Polynucleotide array technology (DNA chips) are suitable for use in the invention, for screening of biological samples. Arrays that include the desired immobilised polynucleotides can be produced on a customised basis by various companies, including HySeq. In general, the arrays employ immobilised polynucleotide probes that are complementary to target sequences from a biological sample (e.g. from a subject). In the context of the present invention, the target sequence will include a polymorphism as disclosed herein.

The polynucleotides to be used as probes in a diagnostic method will usually comprise at least 8, 10, 15, 20 or 50 consecutive nucleic acids derived from the appropriate ERα gene or 3'UTR region, including one or more of the polymorphic sites disclosed herein.

Polynucleotides may also be designed to act as primers to amplify polynucleotides that may comprise a polymorphism.

One or more polynucleotides may be used to characterise/determine more than one different SNP.

Table 1 identifies the ERα polymorphisms of the invention, and indicates the location of each polymorphism in the ERα gene. The position of each polymophism is given relative to the location of the major transcriptional start site on the BAC contig (+49378 on BAC sequence AL356311). Reference is made to BAC sequence with NCBI accession numbers as follows: A = AL035695, B = AL356311 and C = AL078572. Screening for novel polymorphisms in regulatory regions of ERα

Denaturing high performance liquid chromatography (DHPLC) was used to scan the ERα gene for sequence variation, using the WAVE DNA Fragment Analysis System (Transgenomic). Genomic DNA was amplified from DNA obtained from a panel of 32 unrelated healthy Caucasian subjects. The screening of 64 chromosomes provides adequate power (>90%) to detect polymorphisms occurring at an allele frequency of greater than 5% ²⁴.

26 PCR amplicons were designed to analyse regions of potential regulatory importance across the ERα gene, in the promoter, and UTR. Amplicons were approximately 600 bp in length to allow maximum efficiency whilst remaining within the limits of the screening capabilities of the technology, and remaining suitable for sequencing. Overlap of adjacent amplicons was designed to ensure complete coverage during the screening process. Regions chosen for DHPLC analysis spanned across 200 kb upstream of the ERα gene. Specifically, the sequence immediately 5' to each of the alternate splicing exon 1 ' s ( 1 A- 1 F) was included in these amplimers, as was the sequence of the known negative regulatory element, ERNE ³⁶. The ERα amplimers and primers used to generate each are shown in Table 2. In addition, this table identifies the relative position of these amplimers to the coding sequence, and to relevant regulatory regions.

To address the extent of variation in the 4.2 kb 3'UTR, primers were designed to amplify 11 overlapping amplimers (Ex 8A-8K; Table 1). Although the entire exonic and splice junction sequences have been screened previously , exons 1, 5, and 8 of ERα were included in our analysis in order to ensure consistency between our protocol and other investigators of the ERα gene.

TABLE 2

Table 2 shows the primer pair combinations used to amplify regulatory regions of 10 the human ERα gene for subsequent analysis.

Amplimers forming heteroduplexes on DHPLC analysis were sequencedto identify the site and nature of the polymorphism. Where possible, the presence of genetic variants was independently confirmed by gain or loss of restriction endonuclease site(s) (either naturally occurring or engineered by altering a primer). Restriction enzymes used for specific polymorphisms are shown in Table 3. The addition or removal of a restriction enzyme site is indicated by + or - respectively, ^indicates that the restriction site was engineered by PCR.

Nature and extent of genetic variation in the human ERα gene:

Twenty two different amplicons, amplified from a panel of 32 unselected individuals, were screened by DHPLC, covering a total of 13,120 bp across promoter, 5'UTR, exonic and 3'UTR sequence. Figure 1 shows a chromatogram demonstrating the formation of 3 different heteroduplexes in amplicon AmplEb. The heteroduplexes identified in this manner were sequenced in order to confirm the presence of novel polymorphisms. In all, 20 polymorphisms were newly identified in the ERα regulatory regions by DHPLC and sequencing. Figure 2 summarizes the location of the biologically interesting variants detected. These include polymorphisms in: (i) a promoter region of 5' exon E, (ii) the negative regulatory region in 5' region- ERNE, and (iii) the 3'UTR. No new polymorphisms were identified in any of the coding regions of ERα examined; however 2 novel variants were identified in intron 5.

The presence of genetic variants was independently confirmed by gain or loss of restriction endonuclease site(s), as shown in Figures 3-4 . In addition to the SNPs identified above, it was observed that 2 amplicons, 8E and

8G, produced chromatograms with an unusual, wide peak. While sequencing of amplicon 8G revealed the deletion of 1 adenine in a run of 9 in a single individual, the broad peak observed on chromatograms of amplicon 8E was observed in all 32 individuals. Figure 5 shows the chromatograms (upper panel) of 4 individuals. A broad peak is visible that stretches from 2.6 to 3.5 min (retention time) in all individuals. In contrast, the more typical peaks presented in Figure 1 have a width of less than 0.4 min. Apparent insertion/deletion polymorphisms are present in both polyadenine and polythymidine tracts in all samples (middle panel). However, sequencing of TA cloned PCR product from a single individual revealed more than 2 apparent alleles, suggesting that the observed variation in length of polyadenine track in the amplified product was either confounded by, or entirely due to PCR slippage artefact.

Optimization of capillary electrophoresis for identification of single nucleotide length variations in polyadenine and polythymidine tracts of 3'UTR

Capillary electrophoresis was next employed to examine for length variation in the polyadenine and polythymidine tracts of the ERα 3'UTR, using a series of primers designed to flank the 2 putative variable regions (Table 4).

Table 4

Figure 6 shows the initial results of capillary electrophoresis analysis of amplicon 8Ea (using primers 8E-seqA F and A's R) confirming the presence of variation in allele length. Homozygotes for both short and long alleles, differing by one nucleotide were provisionally identified (upper panel). However, in addition to the major peaks corresponding to the true allelic length, and the minor peaks, consistent with shorter PCR slippage products, an additional peak was observed of one nucleotide greater length than the major peak. This is likely to be the result of the non-templated addition of an adenine at the 3' end of the amplicon by thermophilic polymerases- a well recognized potential source of error in genotyping studies employing Taq DNA polymerase to amplify microsatellite loci. As a result, the heterozygous electropherogram is difficult to genotype reliably, with the dominant amplicon of the longer allele containing a component from the non-templated ademne addition of the shorter allele, and with the dominant amplicon of the shorter allele having the additional component from slippage products of the longer allele.

In order to remove the artefactual products resulting from the nontemplated addition of a nucleotide by Taq polymerase, a method known as "pigtailing" was employed. The placement of the sequence 5' GTTTCTT-3' on the 5' end of reverse primers has been observed previously to result in nearly 100% adenylation of the 3 'end of the forward strand ²⁶. The addition of this "pigtail" to the reverse primer (A's R Pig; Table 4) greatly improved the ease of identification of major alleles. Figure 6 (lower panel) shows the capillary electrophoresis elution profiles of amplicon 8Ea after "pigtailing". Dominant peaks for homozygous individuals are seen at the maximum length (92 and 93 bp), with only slippage peaks at smaller lengths. In the case of the heterozygote, the shorter allele (i.e. 14 A's) has the greatest amplitude (at 92), as the result of addition of its own true peak with the slippage products from the longer allele. This results in the resemblance of the elution profile to a "Mexican hat".

Allelic variation was also confirmed across the polythymidine track of the 3'UTR. Allelic Frequency of ERα polymorphisms

Analysis of the frequency and distribution of genotypes at the newly identified ERα polymorphic loci was examined in a number of different study populations.

The twelve prioritized polymorphisms were genotyped in up to a total of 1821 individuals from 6 different study populations. Allele frequencies were deduced from genotypic frequencies by gene counting. The frequency of each variant in these 0 populations is summarized in Table 5.

Table 5

In Table 5, values represent the fraction of chromosomes with the rare allele. The number of chromosomes genotyped (n), and the nationality of the subjects are included in brackets. It is important to note that the TG and Sylvhia cohorts were selected based on the presence of coronary artery disease and hypertension respectively. Haplotype reconstruction of ERα polymorphisms

Haplotypes of the 12 newly identified biallelic polymorphisms (10 SNPs and 2 length variations) distributed throughout the ERα gene were assigned to the unrelated individuals of both the TG vascular study group and the AK bone study group by use of an inferential algorithm, PHASE ² . 0 The identity and frequency of these multi-locus haplotypes across the 10 SNP loci considered are presented for the two studies in Table 6. Haplotypes 1, 2 and 3, differing at only a single nucleotide, accounted for 68% of the total chromosomes assigned. Furthermore, haplotypes likely to be derived directly from these three main branches account for over 85% of those observed. Haplotype 6 is likely to be derived directly from the founder haplotype following an ancestral recombination event and accounts for a further 6% of haplotypes.

Table 6

Combined AK Bone Study TG Vascular Study

Haplotypes n % n % n %

1 ACGCTCTTTC 174 30.1 126 33.3 48 24

2 ATGCTCTTTC 123 21.3. 84 22.2 39 19.5

3 ACGTT.CTTTC 98 17.0 57 15.1 41 20.5

4 ATGTTCTTTC 47 8.1 31 8.2 16 8

5 ACACTCTTTC 40 6.9 25 . 6.6 15 7.5

6 ACGCTCTGCT , 33 5.7 22 5.8 11 5.5

7 CCGCTCTTTC 17 2.9 12 3.2 5 2.5

8 ATGCTCTGCT 12 2.1 4 1.1 8 4

9 ACGCCCTTTC 6 1.0 0 0.0 6 3

10 ACATTCTTTC 5 0.9 3 0.8 2 1

11 ACGCTCTGTC 4 0.7 2 0.5 2 1

12 ACGCTTGTTC 3 0.5 3 0.8 0 0

13 ACGCTTTTTC 2 0.3 2 0.5 0 0

14 ACACCCTTTC 2 0.3. 1 0.3 1 0.5

15 ACGCTCTTCC 2 0.3 0 0.0 2 1

16 ATGTTCTTTT 0.2 0.3 0 0

17 ATGCTTGGT.C 0.2 0.3 0 0

18 ATGCTCGGTC 0.2 0.3 0 0

19 ACGCTCGTTC 0.2 0.3 0 0

20 ACGCTTGGTC 0.2 0.3 0 0

21 CCGTTCTTTC 0.2 0.3 0 0

22 CTGCTCTTTC 0.2 0 0.0 .1 0.5

23 ACGCTCTTCT 0.2 0 0.0 1 0.5

24 CTGTTCTTTC. 0.2 0 0.0 1 0.5

25 ATGTTTGGTC 0.2 0 0.0 1 0.5

578 100.0 378 100.0 200 100

The individual polymorphic loci represented in Table 6 are (from left to right) IsoE3, IsoE2, IsoEl, ERNE 1, ERNE 2, 2al, 2a2, 2a3, 2a4, and 2a5. Potential functionally important polymorphisms in 5' region of ERα

The vast extent and complexity of the upstream sequence and promoter region of ERα means that efficient screening for novel mutations that may effect ERα transcriptional regulation requires some knowledge of important regulatory regions and organization. Thus, this study targeted regions which occur over 150 kb over the upstream region of ERα for the investigation of biologically interesting polymoφhisms. Consequently, the panel of variants that were identified provide substantial novel value for the genetic analysis of ERα. Polymorphisms in 5' regulatory regions with the greatest potential for functional importance are those in exon IE promoter, and those in the immediate vicinity of ERNE. The upstream exon IE is the alternatively spliced, untranslated exon of the predominant ERα mRNA isoform expressed in endothelial cells, hepatic cells, and bone tissue. Thus, the identification of allelic variation in the sequence immediately 5' to this exon may have implications for the transcriptional regulation of ERα in these tissues. Similarly, the negative regulatory element ERNE may influence cells mediating atherogenic actions of estrogen. Variations in these regions may therefore be novel risk factors for vascular disease via altered ERα expression.

The recent appreciation of the role of the 3 'UTR in regulating the stability of ERα mRNA is supportive of functional significance of nucleotide variations in this region. In this study a high incidence of polymorphism we identified in the 3 'UTR. Of most interest is the variation observed in polyadenine and polythymidine tracks. The sequence of this region is suggestive of a structural role for these tracks, with likely pairing of the adenines with uracils. The potential importance of length variation at these loci necessitated rigorous optimization and validation of genotyping methods which was achieved prior to further study.

The significant linkage disequilibrium observed between the (TA)_n variable repeat and PvuII loci with newly identified ERα polymorphisms is of particular interest given the large number of association studies published on these markers. The association of these variants with a number of estrogen-sensitive phenotypes, despite the lack of a biological plausible function has led investigators to consider the possible causal role of other variants in close linkage disequilibrium. The location of newly identified ERα variants in regions thought important in both transcriptional and post-transcriptional regulation suggests that some may be strong candidates for such biological action. The most significant linkage disequilibrium observed between the PvuII and (TA)_n loci and newly described ERα SNPs, was with the common ERNEl variant. In contrast, the lack of significant linkage disequilibrium between the polyadenine and polythymidine alleles and the remaining ERα loci is an important consideration for interpreting any future studies using these loci. More than for other markers, such an association would be suggestive of a direct causal role.

Functional Analysis of Novel ERα Polymorphisms

Through targeted screening of regulatory regions spread over 300 kb of the ERα, it has been possible to identify numerous polymorphisms which represent plausible candidates for future analysis of allelic association with disease. The following section investigates the possible influence of these polymorphisms on ERα expression. The impact of variants in the IE promoter and ERNE on transcriptional regulation is examined by luciferase reporter assays. Computer modelling was used to examine the possible biological importance of length variation in the polyadenine and polythymidine tracks of the 3'UTR. A panel of reporter gene constructs was used to identify potential functional variants in the the negative element ERNE. In addition, in silico studies were performed to predict potentially important biological influences of allelic variation at loci shown to be polymorphic.

Functional effects of allelic variation in the ERNE amplicon

The mechanism of negative transcriptional regulation by the 102 bp ERNE element is not known. The present study has identified 2 polymorphisms in the 465 bp ERNE amplicon screened by WAVE analysis: ERNEl located 110 bp prior to the element itself; and ERNE2, located 32 bp after the beginning of the element (ERNE 1 and ERNE2). Predicted transcriptional binding sites of this sequence containing ERNE, the two polymorphic loci, and the surrounding sequence were also examined. T>C allelic variation at ERNE 1 resulted in the introduction of a putative NF-1 binding site as shown in Figure 7. In contrast, ERNE2 does not lie in the immediate vicinity of a predicted binding site, and allelic variation does not result in the introduction of a site recognized by electronic database screening.

Promoter studies presented earlier confirmed a negative regulatory role of the ERNE element on transcription in human hepatic (Hep3B), and endothelial (hMEC) cell lines. Similar transient transfection experiments were used in the present study, in order to examine the effect of the novel ERNE2 polymorphism on the effect of the ERNE element. Again, the ERNE element containing the common ERNE2 T allele was associated with approximately 60% reduction in promoter activity in Hep3B cells. However, a functional role for the ERNE2 T>C polymorphism was suggested by the loss of significant negative transcriptional regulation by ERNE in the luciferase reporter constructs containing the C variant (Figure 8). Due to the noise of the assay, the differences between the reporter activities of constructs containing the ERNE2-C and ERNE2-T alleles did not reach formal statistical significance. The trend towards reversal of negative transcriptional effect of ERNE element in association with the ERNE2 C allele was also observed, but to a lesser degree, in endothelial cells, as well as after placement of the element after the luciferase gene. Influence of polyadenine and polythymidine length variation on mRNA folding Given the demonstrated role of the 3 'UTR in regulation of ERα mRNA stability, it was of interest to explore the structural effect of the polyadenine and polythymidine repeat regions on RNA folding, and to investigate the functional effect of variation in their length. Sequence spanning the second 1000 bp of the 3 'UTR (UTR2); ³⁷ was chosen for analysis as this region contains both the polyadenine and polythymidine tracks, and has been previously shown to exert almost all of the "instability" attributed to the 4.2 kb ERα 3'UTR.

The structural predictive programme "MFOLD" was used to investigate the contribution of the polyadenine and polythymidine tracks on the folding of the RNA. This programme predicts the folding of an RNA molecule by finding a conformation of minimum free energy, using published values of stacking and destabilizing energies ²⁷'²⁸ Over 50 optimal and suboptimal secondary RNA structures for the UTR2 sequence were predicted by MFOLD, presented in order of increasing free energy. A comparative analysis of these predictions revealed that the helical element produced by complementary base pairing between the A's and U's was a consistent feature with a predicted minimal energy decomposition of -16.1 kcal/mol (with 14 As and 17 Ts). It is of particular note that this helix provided the greatest contribution to the minimization of free energy of any structural element in the sequence analysed.

MFOLD was next used to examine the influence of length variation at the polyadenine and polythymidine repeats on the predicted RNA structure, with particular focus on the change of energy of important structural elements. The length of A's and T's in the submitted sequence was chosen to reflect those variations observed to occur naturally in the population (14 or 15 A's; and 16 or 17 T's). Further analysis of the sequence containing additional length variation was also performed: firstly, for interest, to examine the effect of deviating from those alleles naturally occurring; and secondly, as the prediction of true size of these alleles may have overestimated the size of the repeat regions due to uncertainty about the addition of an extra nucleotide dictated by the pigtailed reverse primer. Polymorphic variation in the length of the polyadenine and polythymidine repeats resulted in alterations in the thermodynamic properties of the helical element, illustrated in Table 7. The predictions for free energy of this helix were consistent between the many optimal and suboptimal foldings. The naturally occurring haplotype of polyadenine and polythymidine length variants with the minimal free energy of decomposition (putatively the most stable) was that containing 15 A's and 17 T's (-17.0 kcal/mol) . The greatest reduction in minimal free energy with naturally occurring variation in repeat length was observed in a change from 17 to 16 T's. However, this repeat haplotype was very rarely observed (6% of the total chromosomes examined in the AK population. The more commonly occurring reduction in the length of the polyadenine repeat also reduced the minimal free energy, but to a lesser degree (-16.1 kcal/mol). The reduction from 17 to 16 Ts did not alter minimal free energy when occurring, as most commonly observed, on a background of 14 A's.

Table 7

In Table 7, the free energy of the entire 1000 bp segment UTR2 is also provided in brackets. The predicted free energies are in kcal/mol (1 kcal= 4.18 J) at 37°C. Combinations of 14 or 15 A's and 16 or 17 T's are thought to represent naturally occurring haplotypes (bold). The order of frequency of the repeat haplotypes commonly observed in the population is as follows: 1) 14 As/17Ts (46%); 2) 15 As/17Ts (28%); 3) 15As/16Ts (20%); and 4) 14As/16Ts (6%).

In silico prediction thus suggests that both polyadenine and polythymidine length variation in the ERα 3'UTR may significantly alter the thermodynamics of secondary mRNA folding. Impact of polymorphisms in the ERNE amplicon Differences in transcriptional activity of the ERNE-reporter constructs representative of the different ERNE2 alleles are suggestive that this variant is functionally altering the site of binding of an important, but unidentified factor. There are no known transcriptional factor binding sites in this 102 bp region, and extensive in vitro experimentation has failed to identify the responsible factor. Thus, it is not possible to identify a precise mechanism for the inhibition of negative regulation by the single nucleotide polymorphism. Indeed, conversely, the identification of this altered response by this variation may be useful in narrowing down the site of action of the unidentified factor.

In contrast to the case of the ERNE2 polymorphism, analysis of putative transcriptional binding sites around ERNEl suggested that T>C variation at this locus results in the introduction of a potential binding site for nuclear factor- 1 (NF-1). NF-1 proteins constitute a large family of DNA binding proteins which are known to promote the initiation of adenovirus replication, as well as regulating the transcription of viral and cellular genes ^29"31. The binding sites for NF-1 have been reported in a wide variety of promoters, and they exhibit flexibility in their sequences, as well as their ability to act as either transcriptional activators, or repressors. Alteration of transcriptional regulation by the introduction of transcription factor binding by a SNP has previously been observed for a number of other genes, including the TNF (tumour necrosis factor) gene in which the variant allele results in the binding of OCT-1.

Impact of allelic variation in the polyadenine and polythymidine tracks of the ERα 3'UTR

A major determinant of RNA folding is the formation of stable base pairs, through the creation of hydrogen bonds between complementary bases (C-G and A-U). Analysis of the ERα 3'UTR demonstrated the central role of complementary base pairing of the polyadenine and polythymdine tracks in the determination of RNA folding according to principles of minimum free energy. Furthermore, length variations in the polyadenine and polythymidine tracks, as previously shown to occur naturally in the population, substantially altered the predicted thermodynamics of the resulting helix.

In vitro analysis of RNA half-life demonstrated that the segment of UTR containing both the A's and T's exerted almost all the mRNA instability attributed to the entire 4.2 UTR, but that sub-fragmentation, resulting in their separation was associated with disruption of the instability. These observations are summarized in Figure 9. Computer modelling data suggests that this may be a direct consequence of the disruption of the helix formed by the complementary base pairing of the repeat regions, indicating that stable helix formation is necessary for the reduction of message half-life. The mechanism by which the polyadenine/polythymidine helix may reduce mRNA stability is not known. An unlikely explanation is that the strength of hydrogen bonds involved in the helix itself influences the stability of the transcript. More likely, is that the secondary structure, influenced by the thermodynamics of the helix, is involved in interactions with a regulatory protein that binds to destabilize the ERα mRNA.

It is interesting to note that the length variation in 3'UTR observed in all populations examined involves only one nucleotide difference. This is in strong contrast with many microsatellite elements, including the (TA)_n repeat in the ERα upstream region which varies between 13 and 26 repeat units in length³⁸. This limited length variation of the polyadenine and polythymidine repeats in the ERα 3'UTR may have a biochemical explanation, such as enzyme slippage at a specific site due to tertiary structure of the DNA, or relate to selection pressure against further allelic variation. For example, reducing the polyadenine length from 15 to 12 A's reduces the free energy of decomposition from -16.1 to -6.0 kcal/mol, a change which may have significant effects on RNA destabilization. It is again of interest to observe the low frequency of the haplotype combining two common alleles, the 92 polyadenine allele and the 115 polythymidine allele, which, of the naturally occurring haplotypes, is the one associated with the greatest change in free energy of decomposition, from -16.1 to -11.0 kcal/mol.

To summarize, altered thermodynamics of the polyadenine/polythymidine helix, as determined by variable repeat lengths, may result in differential mRNA stability. Association of novel ERα polymorphisms with altered estrogen sensitivity

The association of novel ERα polymorphisms with estimated cardiovascular disease risk was investigated through intermediate risk factors including dyslipidemia and measurements of endothelial function. In addition, the biological plausibility of the role of ERα polymorphisms in the genetic determination of variable estrogen signalling was addressed by examining for evidence of association with bone mineral density (BMD).

Genotypes at newly identified ERα polymorphisms were examined for association with estrogen-sensitive phenotypes in a series of study cohorts. Traits for which association was examined were predominantly quantitative in character, and were considered, based on prior observations, to be significantly influenced by estrogen. In addition, the quantitative phenotype was generally an intermediate phenotype with strong association to the complex disease of interest (e.g. BMD and osteoporosis; lipid levels and atherosclerosis).

Measurements of quantitative traits were examined for normal distribution prior to statistical analyses. Skewed measurements were transformed before further analysis by taking the natural logarithm, which resulted in an approximately normal distribution. Each marker was examined independently for an association between the marker state and variation in phenotype using ANONA (type III sums of squares). Values are expressed as mean±standard error of the mean (SEM) unless otherwise stated. Standard interaction models were used to determine whether the influence of a risk factor varied between genotypes. Interaction of risk factors and genotype was assumed if an interaction term, being the product of the risk factor and a variable defining the genotype groups, was significant when added to the regression model containing both variables. Multiple stepwise regression analyses were also performed for some phenotypes.

Given the likelihood that estrogen signalling might have either quantitative or qualitatively different effects in the 2 sexes, the genotype/phenotype relationships were initially analysed separately according to gender. If the sexes behaved similarly, the group was analysed as a whole, with sex included as a potential covariant in the model. However, if the pattern of relationship was different in the two sexes, statistical examination was performed separately on each gender.

The influence of allelic variation at novel ERα polymorphic loci on endothelial- dependent vasorelaxation

The association between ERα polymorphisms and endothelial-dependent vasorelaxation (EDR) in the TG vascular cohort was investigated.

The relationships between vascular function, the clinical risk factors and ERα genotype variants (IsoEl-3, ERNE 1-2, 2a3, and polyadenine and polythymidine variable repeats) were investigated using analysis of variance (ANOVA).

Table 8 summarises the EDR measurements stratified by sex and ERNE2 genotype. Significant differences in the genotype distribution of the rare C allele were observed between sexes (Chi Square score 10; p<0.05), with four of the 20 females (20%) and 3 of 81 males (3.5%) identified as carriers. In males, the presence of the C allele at ERNE2 was associated with substantially increased vasorelaxation in response to all 3 endothelium-dependent agonists (acetylcholine, bradykinin, and calcium ionophore: A23187). In the case of relaxation to acetylcholine (ACh) the augmentation in males with the C allele was 100% (55.5±10%, n=3; versus 24.7±1%, n=78; p<0.001, accounting for 17% of the total variability in vasorelaxation (Eta-squared=0.17). This association was independent of the classic vascular risk factors hypercholesterolaemia, smoking and diabetes. No significant relationship was detected between the C allele and EDR in females. In a control experiment, sodium nitroprusside (SNP)-induced maximal relaxation, which is known to occur independently of endothelial NO production, did not vary between individuals according to genotype at ERNE2 (p>0.05). The gender-dependent difference between the association of EDR and the C allele was shown to be significant in an interaction model (Significance of interaction term: p=0.004).

Table 8

In Table 8, mean relaxations to acetylcholine (ACh), Bradykinin (BK), and calcium ionophore (Ca) are expressed as a percentage of maximal agonist-induced contraction. Consistent with the observed improvement of endothelial-dependent vasomotion, the likelihood of having a myocardial infarct was reduced by a factor of 2 in individuals possessing the C allele at ERNE2 (odds ratio 0.45; CI 0.33-0.61; p<0.05).

A significant association of EDR with the variation at the ERNEl polymorphic locus (upstream of ERNE) was also observed. In this case, the presence of the C allele did not influence ACh- or bradykinin-induced vasorelaxation, but was associated with a reduced vasorelaxant response to calcium ionophore. Individuals homozygous for the C allele at ERNEl had a 27% reduced vasorelaxant response to calcium ionophore (CC: 28.5±5%, n=5; CT/TT: 39.2±2%, n=64; p<0.05), and displayed a concomitant 77% increase in basal production of the NO scavenger, superoxide (CC: 32.3±11 RLU/sec/mg, n=8; CT/TT: 18.2±1 RLU/sec/mg, n=79; ρ<0.05), also suggestive of endothelial dysfunction. These associations were independent of sex, hypertension, diabetes mellitus, smoking and age. A non-significant trend for increased incidence of diabetes mellitus was observed in individuals with the C allele, and may warrant further study in a larger population more extensively characterized for measures of insulin resistance. Genotypes at the 3 'UTR polyadenine and polythymidine repeats were significantly associated with ED V. The results for the polyadenine repeat variable in the combined male and female group are summarised in Table 9. Individuals homozygous for the 92 polyadenine allele had 25% greater relaxation than individuals homozygous for the longer 93 allele (p=0.05). The polyadenine genotype was more predictive of EDR in females than males, predicting 40.6 % of variability in this model (adjusted R²= 0.406, p=0.015). In particular, homozygous 92/92 females displayed a 50% increase in endothelium-dependent vasorelaxation compared to the combined 92/93 and 93/93 genotype group (29.7±5 versus 14.1±2; %; n=8-l l; pθ.01). This association was independent of smoking, but not of cholesterol or diabetic status. The association of the polyadenine genotype with EDR in the male subgroup alone was not significant.

Table 9

A smaller, but significant association was also detected between polythymidine length variants and EDR. Individuals possessing at least one copy of the 115 allele had a significantly reduced EDR compared with 116/116 homozygotes independent of clinical risk factors, or sex (21±2%, n=39 versus 27±2%, n=57; p<0.05).

In summary, association of novel ERα polymorphisms was observed with EDR. Individuals possessing the variant C allele at ERNE2 and the short polyadenine allele demonstrated an improved endothelial function, and those with the variant C ERNEl allele, as well as the short polythymidine allele had a reduction in endothelial function. No further genotype/phenotype associations were observed in this cohort. The effect of ERα genotype on plasma cholesterol

Estrogen is known to exert a beneficial effect on serum lipid profiles. In all studies in which cholesterol measurements were taken higher levels of HDL (p<0.001), and lower LDL (of variable significance in the different studies) cholesterol were observed in females compared to males. The presence of the C allele at ERNE2 was associated with an improved cholesterol profile in females. Specifically, female carriers of the C allele demonstrated a 49% (CI: 9-91%; p<0.05) reduction in LDL cholesterol, and an improved HDL/LDL ratio when compared with TT homozygotes (p<0.05). No differences in LDL, HDL or HDL/LDL ratio was associated with the ERNE2 genotype in the males. The sex- dependent difference in the association between LDL levels and the C allele was significant in the interaction model (Significance of interaction term: p=0.023).

An increase in HDL levels was found in females possessing the 92 polyadenine alleles, occurring in an additive manner. Polyadenine genotype was estimated to contribute to 6% of the total variability in HDL in females, with homozygotes for the short allele demonstrating a 20% increase in HDL (CI: 5-35%). A highly significant interaction between sex and ERα genotype in association with plasma HDL levels was determined using the interaction model (p<0.001).

Genotyping of the IsoE2, IsoEl, ERNEl and 2a3, as well as the PvuII polymorphic loci was also performed. However, no other ERα genotype/lipid associations reached statistical significance in this cohort. Association of ERα SNPs in variable blood pressure and cardiac morphology

A significant effect of genotype at the polyadenine locus on blood pressure in females was identified (SBP: adjusted R sq. 0.073, p=0.008; and DBP: adjusted R sq. 0.072, p=0.022). Although blood pressure did not differ significantly between the different homozygote groups, the 92/93 genotype displayed a reduction of 8 mm Hg (CI: 2-14) in systolic BP and a 5 mmHg (CI 1-9) in diastolic BP compared with homozygote 93/93 females. No significant differences in blood pressure were observed with ERα genotype in males. In addition to the independent effects of the polyadenine and ERNE2

(nonsignificant alone) variant alleles, the two variables interacted significantly with sex in their effect on systolic blood pressure (p<0.001).

No significant associations were observed between blood pressure and genotype at the IsoEl, IsoE2, ERNEl or PvuII polymorphic loci. The association of ERα genotype and gender with parameters of left ventricular hypertrophy was also examined. Measurements of interventricular septal thickness (INS),

PWT (posterior wall thickness), and LNMI were significantly reduced in females compared with males (p=0.003; p=0.001 and pθ.001). This gender effect was independent of systolic blood pressure and body mass index.

INS was associated significantly with genotype at the IsoEl locus after correction for systolic blood pressure and sex (p=0.015), increasing with the presence of the variant allele. Measurements of INS before correction were GG: 0.95±0.02 (n=58); GA: 1.03±0.13 (n=7); and AA: 1.29 (n=l). 6% of the variance in INS measurements was accounted for by differences between the IsoEl genotype groups (Eta-squared=0.056). In addition, there was a significant interaction between systolic blood pressure and IsoEl genotype in association with INS (p for interaction term=0.001). The relationship between the IsoEl genotype and hypertrophic response to systolic blood pressure was examined by use of linear regression. Carriers of the A allele were observed to have a greater hypertrophic response to increased blood pressure, with systolic blood pressure estimated to account for 95% of the variability in INS in individuals possessing the variant A allele (adjusted R squared for the INS BP relation±standard error of the estimate: 0.95±0.08; p<0005), compared with only 9% in GG homozygotes (adjusted R²= 0.092±0.17; p=0.012).

Given previous correlations between LVH and insulin resistance/metabolic syndrome ³²'³³, it was of interest to observe a significant increase in blood glucose in carriers of the IsoEl A allele. Furthermore, although blood glucose levels did not significantly influence LNH alone, after addition of blood glucose measurements to the model, IsoEl genotype was no longer a significant predictor of LVH (p=0.079), suggesting a degree of dependency on the insulin resistant state.

The association between ERα genotype and LNME did not reach formal significance. However, significant interaction terms were observed between systolic blood pressure and IsoEl genotype in association with LNME (p=0.039), and also with IVS (p<0.001). No association was observed between the IsoEl genotype and PWT.

ERNEl was highly associated with IVS in females, with the presence of a C allele related to increased wall thickness, even after correction for systolic blood pressure (Estimated marginal means TT: 1.18± 0.04, n=19; TC: 1.22±0.04, n=22; CC: 1.54±0.13, n=2; p<0.05). The interaction term for systolic blood pressure and the ERNEl genotype in the association with ventricular hypertrophy (IVS) was highly significant in females (pθ.001). A similar, but non-significant trend was observed in males, and for other hypertrophy measurements.

Genotype at 2a3 was also examined for association with blood pressure and left ventricular hypertrophy in this cohort. However, no significant associations were observed.

ERα genotype, insulin resistance and obesity

Individuals possessing one copy of the ERNE2 C allele had substantially increased plasma glucose compared with those that did not. This difference remained after inclusion of sex in the model (means after correction: TT: 5.01±0.03 mM, n=219; TC: 5.48±0.18 mM, n=6; p=0.01). A similar, but non-significant trend was observed when females were analyzed in isolation. No significant differences in insulin levels were observed in relation to ERα genotype, and the association of ERNE2 genotype with blood glucose remained strong after inclusion of insulin as a covariant in the model (p=0.003), suggesting a possible peripheral resistance to insulin in this group. Bone mineral density is associated with novel ERα polymorphisms

ERα genotype at IsoEl, IsoE2, ERNE, 2a3 loci was found to significantly influence a variety of BMD parameters. Polyadenine and polythymidine genotypes were not observed to effect BMD parameters.

The relationship between hormonal levels and ERα genotype was examined using ANO VA. Allelic variation at the ERNE 1 locus was associated with significant changes in plasma estrogen and testosterone levels (p=0.011 and p=0.020 respectively; Table 10).

The Eta-squared value suggested that 5.7% and 5.1% of the variation in estrogen and testosterone levels respectively, are attributable to ERNEl allelic variation in this model.

Estrogen was 19% higher(CI: 4-34%; p<0.01), and testosterone 21% higher (CI: 7-35%) in CC homozygotes compared to TT homozygotes (p<0.01). No differences in measures of adipose tissue, the predominant site of aromatization, were observed between individuals of different ERα genotype. Table 10

Further significant associations were observed between ERNE 1 genotype and BMD. After controlling for age, BMI and plasma estradiol levels, the presence of C allele at ERNEl locus was shown to be associated with lower BMD at all sites of bone measurements except for the spine. The "biological" noise in the measurement of BMD at the lumbar spine, resulting from a high level of osteophyte formation, may explain the lack of significance of the similar trend for ERNEl genotype and BMD association at this site.

REFERENCE LIST

I. Green S, Walter P, Kumar V, et al. Human oestrogen receptor cDNA: sequence, expression and homology to v- erb-A. Nature. 1986;320:134-139. 2. Kuiper GG, Enmark E, Pelto-Huikko M, et al. Cloning of a novel receptor expressed in rat prostate and ovary. Proc Natl Acad Sci USA. 1996;93:5925-5930.

3. Barnea A, Gorski J. Estrogen-induced protein. Time course of synthesis. Biochemistry JID - 0370623. 1970;9:1899-1904.

4. Toft D, Gorski J. A receptor molecule for estrogens: isolation from the rat uterus and preliminary characterization. Proc Natl Acad Sci USA JID - 7505876.

1966;55:1574-1581.

5. Walter P, Green S, Greene G, et al. Cloning of the human estrogen receptor cD ^'A. Proc Natl Acad Sci USA. 1985;82:7889-7893.

6. Greene GL, Gilna P, Waterfield M, et al. Sequence and expression of human estrogen receptor complementary DNA. Science. 1986;231 : 1150-1154.

7. Ponglikitmongkol M, Green S, Chambon P. Genomic organization of the human oestrogen receptor gene. EMBO J. 1988;7:3385-3388.

8. Buist DSM, LaCroix AZ, Barlow WE, et al. Bone mineral density and breast cancer risk in postmenopausal women. Journal of Clinical Epidemiology . 2001;54:417-422.

9. Nguyen TV, Center JR, Eisman JA. Association between breast cancer and bone mineral density: the Dubbo Osteoporosis Epidemiology Study. Maturitas. 2000;36:27-34.

10. Barengolts El, Berman M, Kukreja SC, et al. Osteoporosis and coronary atherosclerosis in asymptomatic postmenopausal women. Calcif Tissue Int.

1998;62:209-13.

I I . Uyama O, Yoshimoto Y, Yamamoto Y, et al. Bone changes and carotid atherosclerosis in postmenopausal women. Stroke. 1997;28:1730-2.

12. Kado DM, Browner WS, Blackwell T, et al. Rate of bone loss is associated with mortality in older women: a prospective study. J Bone Miner Res. 2000;15:1974-

1980.

13. Van Der Klift M, Pols HA, Hak AE, et al. Bone mineral density and the risk of peripheral arterial disease: the rotterdam study. Calcif Tissue Int. 2002;70:443-9.

14. Haffher SM, Bauer RL. The association of obesity and glucose and insulin concentrations with bone density in premenopausal and postmenopausal women.

Metabolism. 1993;42:735-8. 15. Andersen TI, Heimdal KR, Skrede M, et al. Oestrogen receptor (ESR) polymorphisms and breast cancer susceptibility. Hum.Genet. 1994;94:665-670.

16. Parl FF, Cavener DR, Dupont WD. Genomic DNA analysis of the estrogen receptor gene in breast cancer. Breast Cancer Res. Treat. 1989;14:57-64. 17. Weel AE, Uitterlinden AG, Westendorp IC, et al. Estrogen receptor polymorphism predicts the onset of natural and surgical menopause. JClin EndocrinolMetab. 1999;84:3146-3150.

18. Kobayashi S, Inoue S, Hosoi T, et al. Association of bone mineral density with polymorphism of the estrogen receptor gene. J Bone Miner. Res. 1996;11:306-311. 19. Sano M, Inoue S, Hosoi T, et al. Association of estrogen receptor dinucleotide repeat polymorphism with osteoporosis. Biochem.Biophys.Res.Commun. 1995;217:378-383.

20. Deng HW, Li J, Li JL, et al. Change of bone mass in postmenopausal Caucasian women with and without hormone replacement therapy is associated with vitamin D receptor and estrogen receptor genotypes. Hum Genet JID - 7613873. 1998;103:576- 585.

21. Han K, Choi J, Moon I, et al. Non-association of estrogen receptor genotypes with bone mineral density and bone turnover in Korean pre-, peri-, and postmenopausal women. Osteoporos Int. 1999;9:290-5. 22. Deng HW, Li J, Li JL, et al. Association of estrogen receptor-alpha genotypes with body mass index in normal healthy postmenopausal Caucasian women. J Clin EndocrinolMetab JID - 0375362. 2000;85:2748-2751.

23. Schubert EL, Lee MK, Newman B, et al. Single nucleotide polymorphisms (SNPs) in the estrogen receptor gene and breast cancer susceptibility. J.Steroid Biochem.Mol.Biol. 1999;71:21-27.

24. Risch N, Merikangas K. The future of genetic studies of complex human diseases. Science. 1996;273:1516-1517.

25. Stephens M, Smith NJ, Donnelly P. A new statistical method for haplotype reconstruction from population data. Am JHum Genet. 2001;68:978-89. 26. Brownstein MJ, Carpten JD, Smith JR. Modulation of non-templated nucleotide addition by Taq DNA polymerase: primer modifications that facilitate genotyping. Biotechniques. 1996;20:1004-6, 1008-10.

27. Zuker M, Jaeger JA, Turner DH. A comparison of optimal and suboptimal RNA secondary structures predicted by free energy minimization with structures determined by phylogenetic comparison. Nucleic Acids Res. 1991;19:2707-14. 28. Zuker M. On finding all suboptimal foldings of an RNA molecule. Science. 1989;244:48-52.

29. Gounari F, De Francesco R, Schmitt J, et al. Amino-terminal domain of NF1 binds to DNA as a dimer and activates adenovirus DNA replication. Embo J. 1990;9:559-66.

30. Osada S, Daimon S, Nishihara T, et al. Identification of DNA binding-site preferences for nuclear factor I-A. FEBSLett. 1996;390:44-6.

31. Jones KA, Kadonaga JT, Rosenfeld PJ, et al. A cellular DNA-binding protein that activates eukaryotic transcription and DNA replication. Cell 1987;48:79-89. 32. Lind L, Andersson PE, Andren B, et al. Left ventricular hypertrophy in hypertension is associated with the insulin resistance metabolic syndrome. J Hypertens. 1995;13:433-8.

33. Sasson Z, Rasooly Y, Bhesania T, et al. Insulin resistance is an important determinant of left ventricular mass in the obese. Circulation. 1993;88:1431-6. 34. Grandien K, Backdahl M, Ljunggren O, et al. Estrogen target tissue determines alternative promoter utilization of the human estrogen receptor gene in osteoblasts and tumor cell lines. Endocrinology. 1995;136:2223-2229.

35. Flouriot G, Griffin C, Kenealy M, et al. Differentially expressed messenger RNA isoforms of the human estrogen receptor-alpha gene are generated by alternative splicing and promoter usage. Mol.Endocrinol. 1998;12: 1939-1954.

36. Penolazzi L, Lambertini E, Aguiari G, et al. Cis element 'decoy' against the upstream promoter of the human estrogen receptor gene. Biochim.Biophys.Acta. 2000;1492:560-567.

37. Keaveney M, Parker MG, Gannon F. Identification of a functional role for the 3' region of the human oestrogen receptor gene. J. Mol.Endocrinol. 1993;10: 143-152.

38. Becherini L, Gennari L, Masi L, et al. Evidence of a linkage disequilibrium between polymorphisms in the human estrogen receptor alpha gene and their relationship to bone mass variation in postmenopausal Italian women. Hum Mol Genet. 2000;9:2043-2050.

Claims

1. An in vitro method for diagnosing whether an individual has or is genetically predisposed to an estrogen-sensitive disease, comprising the step of identifying whether the genome of the individual comprises one or more of the following polymorphisms in the ERα gene or 3'UTR: ISOEl, ISOE2, ISOE3, ERNEl, ERNE2, ISOCl, 2al, 2a2, 2a3, 2a4, 2a5, 8B1, 8B2, 8C, 8E, 8H and 81.

2. A method according to claim 1, comprising sequencing at least part of the ERα gene or 3'UTR, and comparing the resulting sequence against a reference sequence.

3. A method according to claim 1, comprising

(a) contacting the ERα gene, the 3'UTR, or a fragment thereof, with a polynucleotide under highly stringent conditions, the polynucleotide having a sequence that permits selective hybridisation to a ERα/3'UTR sequence that contains a polymorphism; and

(b) detecting hybridisation.

4. A method according to any preceding claim wherein the step of identification is carried out on a biological sample isolated from the individual.

5. An isolated polynucleotide useful in diagnosing whether a subject has or is predisposed to an estrogen-sensitive disease, comprising at least 15 contiguous nucleic acids and hybridising under highly stringent conditions to a region of the ERα gene or the 3'UTR comprising a polymorphism identified in claim 1, or its complement.

6. A diagnostic kit comprising a polynucleotide as defined in claim 5.