The expression of multiple isoforms of the MUC1 gene [2, 4, 7, 9] in a variety of cell types [2–6] and the active role of the gene in different physiological processes  suggest a fine-tuned transcriptional regulation. Some specific questions that we have addressed in this study include: 1) Does the MUC1 promoter regulate differential expression of isoform specific mRNAs? 2) Is the expression of human MUC1 gene transfected into heterologous (mouse) cells relevant to its endogenous expression in homologous (human) cells? 3) How does estrogen regulate MUC1 transcription? Some of these questions have been answered in this study, others are still issues for our ongoing investigations.
Numerous overlapping and densely distributed cis-elements that potentially could bind a multitude of transcription factors demonstrate the structural and functional complexity of the MUC1 promoter. Further evidence of the MUC1 promoter complexity is the presence of several elements that participate in the development of RNA-Pol II initiation complexes (TATA-boxes, GC-boxes and initiators) and multiple transcription start sites (CAP-sites).
Usually, eukaryotic promoters that regulate transcription by RNA polymerase II belong to one of four types: 1) TATA-box-containing promoter (facultative genes) ; 2) GC-box-containing promoter (housekeeping genes) ; 3) Initiator-containing promoter (first described in genes expressed in B-cells) ; 4) "Dual" promoter that have both TATA- and GC-elements and drive both TATA- and GC-boxes specific transcription (found in cathepsin D and methallotheonin genes) . The MUC1 promoter demonstrates properties characteristic for promoters of different types. It contains several TATA- and GC-boxes and a number of initiators (see "Additional files 1, 2, 3"). It contains multiple cis-elements that potentially transcribe the MUC1 gene in many different cell types and tissues and cis-elements specific for viral promoters [2, 3, 5, 28–30]. These peculiarities of the MUC1 promoter allow us to classify it as a "mixed polypotent promoter". We believe that studies of eukaryotic transcription will reveal increasing number of genes with this type of promoter.
Our bioinformatic study of the MUC1 promoter resulted in construction of the MUC1 cis-element map. This map has several advantages. Using the map, one may design a functional analysis of transcription factors potentially competing for overlapping cis-elements in the MUC1 promoter. It also allows prediction of cells in which the MUC1 gene might be expressed. Based on the content of the MUC1 cis-elements, we predicted expression of the MUC1 gene in lymphoid and muscle cells that had been thought to be MUC1 non-producing cells. Subsequently, the MUC1 expression in these cells was confirmed experimentally . Obviously, in a single study it is impossible to evaluate functional activities of all cis-elements detected in the MUC1 promoter. Nevertheless, several cis-elements that were identified by our bioinformatic approach (MZF-1, Sp1 and STAT) have been already found active in transcriptional regulation of the MUC1 gene in vivo [28–30].
Transient transfection of plasmids that have different deletions within the promoter DNA can identify promoter regions necessary for the expression of specific mRNA isoforms. We combined this approach with our bioinformatics information to experimentally characterize the MUC1 promoter activity. In our study, the Dpr plasmid that contained a full-length MUC1 promoter supported expression only of the MUC1/SEC isoform (Fig. 1B, lane – 1). In contrast, the plasmid DprΔ2154 lacking a ~2 kb fragment deleted from the 5'-end of the promoter demonstrated higher activity and supported expression of both MUC1/SEC and MUC1/TM mRNA (Fig. 1B, lanes – 4 and 5, respectively), indicating that within the deleted fragment are transcriptional repressor binding sites. Our cis-element map supports this hypothesis, since the deleted fragment of the MUC1 promoter (-2872/-718) contains multiple binding sites for transcriptional repressors: 9 sites for δEF1-repressor (-2180/-2170, -1546/-1536, -1522/-1512, -1486/-1476, -1462/-1452, -1418/-1408, -1175/-1165, -1029/-1019 and -1008/-998), 7 sites specific for the Gfi-1 repressor (-2627/-2515, -2589/-2574, -2530/-2515, -1738/-1723, -1435/-1420, -1306/-1291 and -1141/-1126), 5 sites specific for YY1-repressor (-2786/-2776, -2772/-2762, -1428/-1418, -1395/-1385 and 1203-1193) and a single ELP repressor binding site (-1172/-1165) (see "Additional files 1, 2").
Plasmid DprΔ2446, lacking an additional 292 bp, elevates MUC1/TM expression but leads to a decrease in the expression of the MUC1/SEC mRNA compared to DprΔ2154 (Fig 1B, lanes – 7 and 8). These observations suggest that the deleted 292 bp fragment might contain both additional sites for MUC1/TM mRNA repressors as well as sites important for MUC1/SEC expression. Indeed, as the map shows, the deleted fragment contains cis- elements specific for two repressors, YY1- (-670/-660) and δEF1- (-472/-462), in addition to several cis-elements specific for mammary epithelial cell activators: CTF/NF, ESE1, MAF, MGF, MP4 and RME [31–34] (see "Additional file 3").
According to results obtained with the DprΔ2839 plasmid, a minimal set of transcription elements (one TATA-box, one CAP-site and several cis-elements specific for transcriptional activators RCE, SRE and ETF-RE) (see "Additional file 3") is sufficient for expression of MUC1/SEC isoform (Fig. 1B, lane – 10). Transcriptional activity of the minimal MUC1 promoter has been observed in transfection studies [35, 36]. Notably, two very different plasmids, one of which contains the full MUC1 promoter (Dpr) and the other which contains only a minimal promoter sequence (DprΔ2839), drive similar expression of the MUC1/SEC isoform. It appears that in our experimental conditions, the positive and negative effects of the cis-elements in the full promoter seem to balance-out each other resulting in expression of MUC1/SEC similar to that of the minimal promoter.
The results demonstrating higher activity of the truncated promoter (DprΔ2154, DprΔ2446) compared to the full length MUC1 promoter (Dpr) are not surprising. Earlier, Kovaric et al  showed that expression of the CAT-test gene driven by the full length MUC1 promoter was almost half than that directed by 743 bp promoter fragment. Abe and Kufe  also observed higher CAT activity when the test-gene was driven by a smaller promoter fragment (-686/+33) than by a larger one (-1656/+33). Moreover, according to these authors, the 114 bp fragment of the MUC1 promoter located between -598 and -485 bp demonstrated the highest CAT gene expression. Superimposition of the promoter fragments analyzed in these studies and our MUC1 cis- element map leads to similar schematics of activating and repressing regions.
One of the issues addressed in our research was whether the expression of the human MUC1 gene transfected into heterologous (mouse) cells is relevant to its endogenous expression in homologous (human) cells. The results described in this study showed that in our experimental conditions, the patterns of the MUC1 gene expression in both cell systems were different. In mouse cells, human MUC1 gene (Dpr plasmid) directed expression of the MUC1/SEC isoform, whereas in human cells, the predominant mRNA was MUC1/TM. In transfected mouse DA3 cells, we could not detect human MUC1/Y isoform, whereas in human T47D and MCF7 cells this isoform was observed. The basis for such differential expression is presently being studied.
One of the important issues in regulation of MUC1 expression is the role of estrogen and estrogen receptors. Several studies have showed that the MUC1 gene positively responds to estrogen [16–19]. However, in our study, we observed that not all MUC1 isoforms responded to estrogen in the same manner. The MUC1/SEC mRNA was expressed in T47D cells (ER+ clone 10) only after treatment with estrogen. In contrast to T47D cells, in ER-positive MCF7 cells, expression of the MUC1/SEC mRNA was observed in the absence of estrogen, although, similarly to T47D cells, estrogen increased and 4-OHT decreased its expression. The MUC1/TM mRNA could be expressed both in T47D estrogen receptor positive cells (clone 10) and estrogen receptor negative cells (clone 8) as well as in MCF7 cells. Moreover, this expression was not affected by 4-OHT. The dependence of the MUC1/Y isoform expression on estrogen and ER was not clear cut. On one hand, we observed some increase in MUC1/Y expression in T47D and MCF7 cells after estrogen treatment but on the other hand, 4-OHT did not inhibit its expression. Further experiments will be needed to clarify this matter.
The above data suggest that the expression of MUC1 isoforms in MCF7 cells somehow differs from their expression in T47D cells. Perhaps relevant to our observations is that MCF7 and T47D cells express different levels of steroid receptors. In MCF7 cells, ER is expressed at much higher levels than progesterone receptors (PR), whereas in T47D cells, the expression of PR is higher than that of ER [37, 38]. Since estrogen regulates the transcription of the ER gene , it appears that T47D cells may require exogenous supplements of estrogen to activate expression of the ER gene. In contrast, the endogenous expression levels of estrogen receptors in MCF7 cells might be high enough to support expression of the MUC1/SEC isoform. In accordance with our results, Hurd et al  observed expression of hyperphosphorylated retinoblastoma protein (ppRB) in MCF7 but not in T47D cells when cells incubated without estrogen. A gradual increase of its expression after treatment with estrogen was observed in MCF7 cells in time-dependent manner. In T47D cells, longer estrogen treatment was necessary to detect ppRB than in MCF7 cells.
Our data demonstrating the responsiveness of the MUC1/SEC isoform expression in human epithelial cells to inhibitory effects of 4-OHT are in agreement with observations that the expression of MUC1 gene in human adenocarcinoma cells is also sensitive to antiestrogens [41, 42]. These data suggest that, in human cells, estrogen may regulate the MUC1 gene transcription by interaction with ER directing them to cis- ERE in MUC1 promoter. The results obtained by us with EMSA using T47D cell lysates support this hypothesis. However, they are in contradiction with the observations made in a mouse system. Studying the role of ER in transcriptional regulation of the mouse Muc1 gene, Zhou et al  concluded that ER did not directly bind to the cis- ERE of the murine Muc1 promoter. We found that all cis-ERE detected in the human MUC1 promoter could form complexes with human ERα in vitro. Several factors may explain this discrepancy. First, different cis-EREs were used in both studies. Although human and mouse MUC1 promoters have high degree of homology, their cis-EREs demonstrate pronounced diversity. We have analyzed the mouse Muc1 and human MUC1 promoter sequences and found that each promoter contains six cis-EREs. Comparison of these elements revealed both homology and differences in their sequences. Second, Zhou et al  analyzed binding of estrogen receptors that have been in vitro translated, whereas in our binding assays we used ERα endogenously synthesized in T47D cells that express human MUC1 gene.
It should be noted, that, although all MUC1 cis-EREs bound ERα, the properties of the complexes developed by different cis-EREs were different. Several cis-EREs, (ERE1, ERE3, ERE4 and ERE5) containing only half of the classical palindrome sequence produced a weak, fast migrating complex with ERα. Interestingly, ERE2, which also contains only half of the ERE palindrome sequence, developed two complexes that at least partially correspond to those observed with classical ERE from the vitellogenin gene and with "putative" ERE-6 of the MUC1 gene. The different electrophoretic mobility of the complexes might be explained by content of ER-cofactors in the complexes. It is not clear why oligonucleotides that have identical or very similar ERE core sequences recruit different cofactors to ER-containing complexes, however, the flanking sequences may play a crucial role in this process . Importantly, all complexes developed with tested cis-ERE contained ERα since the binding of ERα was specific and could be inhibited by antibodies developed against the DNA-binding domain of human ERα.
Whereas our in vitro studies clearly showed that physical binding of ERα with cis- EREs of the MUC1 promoter occurs, definitive in vivo binding still remains to be proven. Orientation of ERE within the synthetic oligonucleotides appears not to be important for in vitro binding, but orientation and distribution of EREs among cis-elements within promoter, are presumably crucial for proper in vivo binding [44–46]. A special approach will be needed to study this issue. In the MUC1 promoter, two estrogen responsive elements (ERE-1 and the "putative" ERE-6), have direct sequence orientation while others (ERE-2, ERE-3, ERE-4 and ERE-5) have opposite orientation. Three elements (ERE-1, ERE-2 and ERE-3) are located relatively distant from the active TATA-box and from each other. Three others (ERE-4, ERE-5 and the "putative" ERE-6) are located in close proximity to each other between nucleotides -389 and -337 (see "Additional files 1, 2, 3") and may function as an estrogen responsive unit [47, 48].
Very few mRNAs that are known to be directly regulated by estrogen in mammary epithelial cells are actually induced via canonical EREs . In fact, most estrogen-responsive genes identified to date contain one or more imperfect EREs or multiple copies of an ERE half-site rather than the classical ERE [50, 51]. The MUC1 cis-EREs also are not absolutely identical to the consensus sequence. Although the affinity of the estrogen receptor for the classical ERE is higher than for imperfect ERE-like sequences, most of the imperfect EREs bind ER . The MUC1 cis-EREs, although imperfect, also bound ER in vitro. It is becoming clear that EREs function as allosteric modulators of ER conformation [53, 54]. The conformational changes in ER induced by individual ERE sequences lead to specific association of the receptor with other transcription factors and assist in differential transcription of estrogen-responsive genes [54, 55]. In light of these data, the imperfect MUC1 cis-EREs might be significant, perhaps by differential usage of ERE in diverse cells recruiting distinct cofactors for the MUC1 expression.
We have discussed the possible involvement of ER and MUC1 EREs in regulation of the MUC1 gene transcription. However, it is known that ER may regulate gene transcription also by interaction with other transcription factors (STAT, AP1, EGFR or NFkB) without direct binding to ERE . Further studies are needed to understand the process of MUC1 transcription in vivo and to elucidate the mechanism by which estrogen activates transcription of the MUC1 gene.
Although our study revealed some new and important features of the MUC1 promoter cis-element content and structure, the precise mechanisms by which these cis-sites are involved in the regulation of MUC1 expression have not been fully elucidated. On one hand, elements within the promoter could determine usage of different transcription start sites specific for individual MUC1 isoforms. The presence of multiple CAP-sites in the MUC1 promoter together with previously documented multiple transcription start sites of the MUC1 gene in T47D cells  support this hypothesis. On the other hand, the promoter cis-elements might be involved in regulating of alternative splicing of a single pre-mRNA common to all MUC1 isoforms. A growing body of evidence suggests that transcription and splicing are highly coordinated processes both at the structural and functional levels [57–61]. For instance, it has been shown that mutations introduced into promoter cis-elements could change the alternative splicing patterns . Moreover, the RNA pol II large subunit physically associates with spliceosomes and the SR proteins that regulate alternative splicing act through specific promoter occupation .
In light of these data, we suggest that cis-elements of the MUC1 promoter may be involved in mechanisms that regulate both transcription and splicing of the MUC1 pre-mRNAs. In this study, we used total RNA extracted from transfected cells. However, for a better understanding of the role of the MUC1 promoter in transcription and splicing of MUC1 isoforms, an additional study of the 5'-ends of nuclear pre-mRNA is needed. Additionally, the effect of different mutations within cis-sites of the MUC1 promoter on isoform expression could be more thoroughly dissected. These experiments are currently in progress.