Discussion
The expression of ERC4, ERD3 and ERD5 variant messenger RNAs relative 		 to WT-ER messenger RNA expression within adjacent normal and neoplastic human 		 breast tissues was investigated using previously described semi-quantitative 		 reverse transcription PCR assays [18,19,24]. These assays allow the 		 determination of the expression of ER-α variant messenger RNA relative to 		 WT-ER messenger RNA using a very small amount of starting material, and offer 		 the advantage of allowing investigators to work with histopathologically well 		 characterized human breast tissue regions. It should be noted, however, that 		 the sensitivities of the assays used in this study differed from each other. 		 The triple-primer PCR assay has previously been set up to allow the 		 determination of ERC4 relative expression in tumor samples with very low ER 		 levels, as measured by ligand-binding assay [24].
We showed that, in samples with a detectable level of ERC4 messenger 		 RNA using a standardized RNAse protection assay, the relative expression of 		 this variant to WT-ER messenger RNA expression was similar to the relative 		 expression of ERC4 PCR product obtained after triple-primer PCR [24]. Triple-primer PCR assay applied to the detection of ERC4 		 messenger RNA in 18 matched normal and tumor breast tissues gave a measurable 		 value of expression in 36 out of the 36 samples studied. This contrasts with 		 the detection of 30 out of 36 and 33 out of 36 obtained using ERD3-specific and 		 ERD5-specific primers, respectively. These differences in sensitivity probably 		 result from different primer set efficiencies under our experimental 		 conditions.
A higher ERC4 messenger RNA relative expression in tumor components 		 compared with the normal adjacent tissue component has been observed in the 		 ER-positive/PR-positive subgroup. This result is in agreement with our previous 		 data [19] obtained by comparing ERC4 messenger RNA 		 expression between independent normal reduction mammoplasty samples and a group 		 of ER-positive/PR-positive breast tumors. Even though a higher ERC4 messenger 		 RNA relative expression was observed in the tumor component of 12 out of 18 		 cases, this difference did not reach statistical significance. This absence of 		 statistically significant differences might result from the low number of 		 matched cases studied or from the different biology of ER-negative cases. 		 Further studies are needed to clarify this issue and to draw any conclusion 		 regarding the expression of ERC4 messenger RNA in ER-negative samples.
ERC4 variant messenger RNA has previously been shown [26] to be more highly expressed in ER-positive tumors that 		 show poor prognostic characteristics (presence of more than four axillary lymph 		 nodes, tumor size >2 cm, aneuploid, high percentage S-phase cells) than in 		 ER-positive tumor with good prognostic characteristics (absence of axillary 		 lymph node, tumor size <2 cm, diploid, low percentage S-phase cells). 		 Moreover, in that previous study, a higher ERC4 messenger RNA expression was 		 also observed in ER-positive/PR-negative tumors, as compared with 		 ER-positive/PR-positive tumors. interestingly, we have also recently reported 		 similar levels of expression of ERC4 messenger RNA in primary breast tumors and 		 their concurrent axillary lymph node metastases [24]. 		 Taken together, these data suggest that the putative role of the ERC4 variant 		 might be important at different phases of breast tumorigenesis and tumor 		 progression; alteration in ERC4 messenger RNA expression and resulting 		 modifications in ER signaling pathway probably occur before breast cancer cells 		 acquire the ability to metastasize. Transient expression assays revealed that 		 the protein encoded by ERC4 messenger RNA was unable to activate the 		 transcription of an estrogen responsive element-reporter gene or to modulate 		 WT-ER protein activity [17]. The biologic significance 		 of the changes observed in ERC4 messenger RNA expression during breast 		 tumorigenesis and tumor progression therefore remains unclear.
A trend toward a higher relative expression of ERD3 messenger RNA in 		 the normal breast tissue components compared with adjacent neoplastic tissue 		 was found (10 out of 13 cases), which reached statistical significance when the 		 ER-positive subgroup only was considered. These data are in agreement with the 		 recently published report of Erenburg et al [20] who showed a decreased relative expression of ERD3 		 messenger RNA in neoplastic breast tissues and breast cancer compared with 		 independent reduction mammoplasty and breast tumor. Transfection experiments 		 performed by those investigators showed that the activation of the 		 transcription of the pS2 gene by estrogen was drastically reduced in the 		 presence of increased ERD3 expression. Moreover, ERD3 transfected MCF-7 human 		 breast cancer cells had a reduced saturation density, exponential growth rate 		 and in vivo invasiveness, as compared with control cells. These data 		 led the authors to hypothesize that the reduction of ERD3 expression could be a 		 prerequisite for breast carcinogenesis to proceed. They suggested that if high 		 levels of ERD3 could attenuate estrogenic effects in normal breast tissue, low 		 levels might lead to an excessive and unregulated mitogenic action of 		 estrogen.
We observed a significantly higher relative expression of ERD5 		 messenger RNA in breast tumor components compared with matched adjacent normal 		 breast tissue. These data confirm our previous observations [18] performed on unmatched normal and neoplastic human breast 		 tissues. Upregulated expression of this variant has already been reported in 		 ER-negative/PR-positive tumors, as compared with ER-positive/PR-positive tumors 		 [15,27], suggesting a possible 		 correlation between ERD5 messenger RNA expression and breast tumor progression. 		 Interestingly, ERD5 messenger RNA can be detected in human pituitary adenomas, 		 but not in normal pituitary samples [28]. This 		 underscores the putative involvement of this ER variant in other tumor systems. 		 Even though it has been suggested that ERD5 could be related to the acquisition 		 of insensitivity to antiestrogen treatment (ie tamoxifen) [29,30], accumulating data refute a 		 general role for ERD5 in hormone-resistant tumors [14,25,31,32]. Only ER-positive pS2-positive tamoxifen resistant tumors 		 have been shown to express significantly higher levels of ERD5 messenger RNA, 		 as compared with control tumors [33]. Taken together, 		 these data suggest that the exact biologic significance of ERD5 variant 		 expression during breast tumorigenesis and breast cancer progression, if any, 		 remains unclear.
Among all the articles published so far on ER variants, only one has 		 investigated ER variant expression between normal and neoplastic matched 		 samples. Okada et al [33] recently reported a 		 study performed on 15 cases. They observed an apparent difference in ER variant 		 messenger RNA expression between adjacent normal and tumor samples. That study 		 was performed using a less sensitive PCR approach, however, because PCR 		 products were stained using ethidium bromide, and no attempt was made to 		 quantify ER variant messenger RNA expression relative to WT-ER messenger RNA 		 expression.
In conclusion, we have shown that the relative expression of ERC4 and 		 ERD5 variant mRNAs was increased in human breast tumor tissue, as compared with 		 normal adjacent tissue, whereas the expression of ERD3 variant messenger RNA 		 was decreased in breast tumor tissues. These results, which confirm previous 		 data obtained on independent human breast tissue samples [18,19], suggest that the expressions of 		 several ER-α variant messenger RNAs are deregulated during human breast 		 tumorigenesis. Further studies are needed to determine whether these changes 		 are transposed at the protein level. Only the use of specific antibodies that 		 are able to recognize specifically the different ER variant proteins putatively 		 encoded by these variant messenger RNAs will allow this issue the be addressed. 		 Furthermore, the putative role of ER-α variants in the mechanisms that 		 underlie breast tumorigenesis remain to be determined.