INTRODUCTION

Purine/pyrimidine mirror repeat elements (PMRs)¹ in the 5 upstream region of the human CFTR and MUC1 genes are sensitive to S1 nuclease and chemical probes of single-stranded DNA character, consistent with the formation of H-DNA in vitro under conditions of acidic pH and plasmid supercoiling (1, 2). A nuclear protein of approximately 27 kDa binds to the purine-rich strand of these elements but not to the pyrimidine-rich strand or to double-stranded oligonucleotides of the same sequence (1). Whether H-DNA exists in vivo and the precise biological function of these elements are not well established. The studies reported here were designed to determine if PMR elements in the MUC1 promoter play a role in regulating transcription of the MUC1 gene. The results suggest that mirror symmetry in the sequence and potential for H-DNA in the element do not influence transcriptional activity of this promoter in transient transfection assays.

MATERIALS AND METHODS

The human pancreatic adenocarcinoma cell line Capan-2 (ATCC) and the human pancreatic epithelioid carcinoma cell line PANC1 (ATCC) were grown in minimal essential medium plus 10% fetal calf serum and 2 mM L-glutamine. The human colon adenocarcinoma cell line HT-29 (ATCC) was grown in McCoy's 5a medium with the same serum supplements described above.

DNA containing 2.87 kbp upstream or 790 bp of sequence upstream of the MUC1 transcription start site was inserted into the polylinker of the pCAT E vector (Promega, Madison, WI), which contains an SV40 enhancer element 3 of the CAT gene. These plasmids were named pCAT-2870 and pCAT-790, respectively. Plasmid DNA was isolated by Wizard Minipreps (Promega). DNA used in transfection experiments was purified on Qiagen columns (Qiagen, Chatsworth, CA).

As internal controls to assess transfection efficiency, cells were cotransfected with vectors containing one of three different promoters directing expression of luciferase. Plasmid pCMV-LUC uses the CMV promoter (3), pGlu-SV (4) uses a parvovirus promoter, and pGLU-LUC (4) uses the SV40 early promoter. These vectors were gifts from Dr. Solon Rhode. Plasmid pDol-CMV-CAT (a gift from Dr. Angie Rizzino) was used as a positive control for CAT assays (5).

A PCR strategy was used to delete M-PMR3 (133 to 102) from plasmid pMAH 5 (which contains base pairs 405 to +33 of the MUC1 promoter; Ref. 1). For the PCR reaction, we used oligonucleotide primers that flanked and diverged from this site and included a unique restriction enzyme site (XhoI) at their 5 termini. Following PCR amplification (1.5 mM MgCl₂, 0.2 mM dNTPs, 100 pmol of each primer, 10 ng of plasmid DNA, 2.5 units of Taq polymerase in 100 µl; 35 cycles of 1 min at 94 °C, 2 min at 55 °C, 4 min at 72 °C), amplified DNA fragments with lengths corresponding to linear plasmid with the element deleted were restricted with XhoI, ligated with T4 DNA ligase (Life Technologies, Inc.), and used to transform XL1 Blue Escherichia coli. A segment of DNA including the deletion was cloned into the corresponding position in the pCAT-790 construct using two unique restriction enzyme sites that flanked the M-PMR3 site. This plasmid was designated pCAT-790(M-PMR3DEL), and subsequent modifications to this plasmid were named using a similar convention (described below and in Fig. 1).

A DNA fragment encoding scrambled sequence with a length corresponding to the native sequence was inserted in the M-PMR3 element by cloning a double-stranded oligonucleotide into the XhoI site of the plasmid that contained native MUC1 promoter sequence (pCAT-790). This oligonucleotide was designed so that the XhoI site at one end of the element was destroyed and a SacII site was introduced in its place. This permitted directional cloning of additional double-stranded oligonucleotides into this site, which encoded the different substitutions described in Fig. 1. All constructs were verified by sequencing.

Cells were plated at 60-70% confluence in six-well culture plates (35-mm diameter for each well), 18 h prior to transfection. Transfection was performed using the Lipofectin reagent (Life Technologies, Inc.). Each well containing Capan-2 or PANC1 cells was cotransfected with 0.2 µg of CAT construct and 0.2 µg of pCMV-LUC luciferase vector as a control for transfection efficiency. HT-29 cells were transfected with 1.0 µg of the indicated CAT construct and 1.0 µg of pCMV-LUC. Cells were trypsinized and harvested 48 h post-transfection, washed twice in TENS buffer, and resuspended in 0.25 M Tris-HCl buffer, pH 7.8. Cell lysates were prepared by four cycles of freezing and thawing. Half of each lysate was evaluated for CAT and luciferase activity, respectively (6). Lysates assayed for CAT activity were heated at 65 °C for 10 min, followed by centrifugation at 12,000 × g for 10 min. The supernatant was incubated with [¹⁴C]chloramphenicol (specific activity: 54 mCi/mM; Amersham) for 4 h at 37 °C and acetylated chloramphenicol was separated from unmodified substrate by a phase extraction with 2:1 mixture of 2,6,10,14-tetramethylpentadecane:xylene (Sigma). Luciferase activity was measured in a luminometer with reaction conditions suggested by the manufacturer (ALL 2010; Analytical Luminescence Lab). Data were acquired for 20 s with each sample. Dunnett's t test was used to determine statistical significance.

RESULTS

The ability of constructs containing the human MUC1 promoter with 2.87 kbp or 790 bp of upstream sequence to direct expression of the CAT reporter gene (pCAT E plasmid, with an SV40 enhancer downstream) was evaluated in Capan-2 cells. Similar to previous reports (7, 8, 9), no significant differences in CAT activity were observed between the constructs containing 2.87 kbp or 790 bp of upstream sequence (data not shown). The CAT construct with 790 bp of upstream sequence (pCAT-790) was used in studies that evaluated the effect of deleting and modifying the M-PMR3 element.

For all experiments reported here, the control for transfection efficiency was a cotransfected construct in which unrelated promoters (either CMV, parvovirus, or SV40 early; see ``Materials and Methods'') controlled expression of a luciferase reporter gene. The optimal ratios and amounts of experimental and control plasmids were determined in pilot experiments. It is possible, even under optimal conditions of cotransfection, that competition for transcription factors will occur between the two transfected promoters, and that this will affect promoter activity (10). Therefore, parallel experiments were performed without cotransfection of the luciferase constructs. These results showed that absolute values of CAT activity were higher without the cotransfection controls; however, the relative levels of expression among the different constructs were similar to the results reported above (data not shown).

Promoter activity was evaluated in three cell lines. Capan-2 expresses moderate levels of MUC1 mRNA, the PANC1 cells used here express very low levels of MUC1 mRNA, and HT-29 cells do not express detectable MUC1 mRNA. The MUC1 promoter-reporter construct pCAT-790 (Fig. 2) accurately reflected qualitative expression of MUC1; the cell lines Capan-2 and PANC1 expressed CAT and HT-29 did not express CAT. However, the quantitative results of CAT expression did not reflect the quantitative expression of MUC1 in that the low MUC1-expressing PANC1 cells gave higher CAT/luciferase expression levels compared to the high MUC1-expressing Capan-2 cells.

One set of experiments (Fig. 2) evaluated the effects of deleting or substituting alternate sequence for the M-PMR3 element. The sequences of the native M-PMR3 and the constructs used in these studies are shown in Fig. 1.

Deletion of the M-PMR3 element (pCAT-790(M-PMR3DEL)) resulted in a ~2-fold increase in CAT expression in Capan-2 cells (statistically significant; p < 0.05, Dunnett's t test). The same trend was observed in PANC1 cells; however, the increase was not statistically significant. Substitution of M-PMR3 with an unrelated sequence of the same length (pCAT-790(M-PMR3SCR)) resulted in a 2-fold increase in CAT expression in Capan-2 and a 7-fold increase in CAT expression in PANC1 cells (statistically significant; p < 0.05) relative to the construct with native sequence. The sequence in the substituted construct (pCAT-790(M-PMR3SCR)) included two unique restriction enzyme sites (XhoI and SacII) to facilitate additional modifications of this vector. To confirm that deletion or substitution of M-PMR3 was responsible for the slight enhancement of transcription, the effect of ``restoring'' the M-PMR3 element in place of the substituted sequence at these restriction sites was evaluated by preparing the (pCAT-790(M-PMR3RES)) construct (Fig. 1) and testing its activity in transient transfection assays (Fig. 2). This construct contained insertions around the M-PMR3 element compared to the native sequence because of the addition of these restriction enzyme sites (Fig. 1). The restoration of M-PMR3 reduced CAT expression levels in Capan-2 cells to levels that were not statistically different from CAT expression levels obtained with the constructs that contained the native sequence. In PANC1 cells, expression was reduced by restoration of M-PMR3, but not to a level that was statistically indistinguishable from the construct containing the native sequence (Fig. 2). These data suggest that M-PMR3 plays a detectable but subtle role in regulating MUC1 transcription.

Two additional constructs were prepared to determine whether the mirror character of M-PMR3 contributed to regulation of transcriptional activity. Mirror character is required for the DNA to adopt an H structure in vivo (11). The native M-PMR3 sequence contains two pyrimidine residues in the purine strand that render it an imperfect mirror repeat (1). In one construct, we repaired the imperfection in the native M-PMR3 sequence to render it perfect (pCAT-790(M-PMR3PER)). In a second construct, we disrupted the mirror character of the element by inverting the orientation of half of the element (pCAT-790(M-PMR3DIS)), eliminating (in theory) the possibility of H-DNA. These constructs were similar in length and sequence to pCAT-790(M-PMR3RES), and hence were compared for activity with this construct (Fig. 3). The results of these experiments showed no significant difference in CAT activity among constructs containing the restored PMR, the perfect PMR, or the disrupted PMR elements in either Capan-2 cells or PANC1 cells. These data do not support the hypothesis that mirror character in sequence and potential for H-DNA in M-PMR3 influences transcriptional activity.

DISCUSSION

Kovarik et al. (7) and Abe and Kufe (8) independently evaluated the ability of 2.87 kbp and shorter regions of DNA upstream from the transcription start site of MUC1 to control expression of a CAT reporter gene in transient transfection assays. Both groups found that constructs with approximately 700-800 bp of promoter sequence gave expression of the CAT gene. Moreover, tissue specific expression of MUC1 was seen in mice transgenic for a 10.6-kbp SacII fragment of DNA, including the human MUC1 gene and approximately 1.6 kbp of sequence 5 to the transcription start site and 1.3 kbp of sequence 3 to the polyadenylation site (9). Together, these data suggest that all elements necessary for controlling temporal and spatial expression of MUC1 are contained within the 10.6-kbp SacII fragment.

The results of transient transfection assays described in Fig. 2 show that 790 bp of sequence 5 to the MUC1 transcription start site confers reporter gene expression on pancreatic tumor cell lines that normally express MUC1. A MUC1-nonexpressing colon tumor cell line transfected with the same construct did not express the CAT reporter gene. These data indicate that this region of the MUC1 promoter can produce qualitatively accurate expression in transient transfection assays. Nonetheless, the levels of expression obtained with this portion of DNA did not accurately reflect the quantitative levels of MUC1 expression within the pancreatic tumor cell lines, since the PANC1 cell line, which expresses much lower levels of MUC1 than Capan-2 cells, consistently gave higher levels of CAT gene expression than Capan-2. This trend was also observed in experiments that did not include cotransfections of the CMV promoter-luciferase construct to control for transfection efficiency (data not shown); hence, it was probably not due to differences among the cell lines in activity of the CMV promoter. We propose that this result is caused by a lack of regulatory elements (such as an enhancer required for high level expression) from regions of MUC1 outside of the 790 bases of sequence being evaluated in these studies. The fact that constructs with up to 2.87 kbp of upstream sequence did not show greater expression in Capan-2 cells suggests that such an element is not upstream of the transcription start site and instead may reside within or 3 of the MUC1 gene.

One goal of this investigation was to determine if M-PMR3 contained elements that contributed to the transcriptional activity of the MUC1 promoter. The data presented in Fig. 2 suggest that M-PMR3 has a very modest affect on transcriptional activity. The magnitude of the effect of deleting the element (2-fold increase in transcription) or substituting alternate sequence for the element (2-7-fold increase in transcription) is relatively small. The fact that the effect (increased transcription) was partially reversed upon reinsertion of the element supports the hypothesis that this region of the MUC1 promoter contributes its biological function; however, the significance of these relatively small variations in transcriptional activity is not well understood at this time. It is notable that none of the modifications to M-PMR3 reported here had an effect on reporter gene expression in the MUC1-nonexpressing cell line HT-29. This demonstrates that the M-PMR3 element is not responsible for preventing expression of MUC1 in nonexpressing cell lines.

A second goal of these studies was to determine if sequence and/or mirror symmetry in this element contributed to transcriptional activity of the MUC1 promoter. Mirror symmetry and purine/pyrimidine strand asymmetry bestow on this element the ability to adopt an H-DNA conformation in vitro (1, 2). Disruption of mirror symmetry destroys the potential to form H-DNA. M-PMR3 is not a perfect mirror repeat element in that one half of the element contains two pyrimidine interruptions in the purine strand that can be viewed as insertions into one half of the mirror repeat element (1). Recent studies have shown that M-PMR3 predominantly forms the less common Hy-5 conformation under in vitro conditions and that this unusual H-DNA conformation can be explained by the insertion of these two pyrimidines into the element. The pyrimidine insertions can be accommodated in the Hy-5 triplex conformation, but these insertions hinder formation of the Hy-3 conformation because it renders the Hoogstein strand of the triplex too short to form continuous triplets (12). Moreover, removing the pyrimidine interruptions in this sequence by creating the perfect mirror repeat element M-PMR3PER (Fig. 1B) caused the element to adopt a canonical Hy-3 conformation (12). Thus, subtle variations in M-PMR3 sequence alter its ability to form certain isomers of H-DNA. With these considerations in mind, we designed a series of constructs to test whether MUC1 promoter activity is affected by: 1) disrupting mirror symmetry by inverting one half of the element or 2) creating a perfect mirror repeat within M-PMR3. The results of these studies (Fig. 3) demonstrate that promoter-reporter constructs containing disrupted mirror symmetry or perfect mirror sequence gave expression levels of reporter genes that were not significantly different from a construct that contained a reinserted native M-PMR3 element. This finding does not support the hypothesis that the intrinsic ability of this element to form H-DNA contributes to its biological activity. This finding is similar to a previous report of a PMR in the c-Ki-ras promoter. Transcription activation by the Ki-ras element (evaluated in vitro) was shown to be mediated by protein binding to duplex DNA, rather than through formation of H-DNA (13).

The hypothesis that formation of H-DNA isomers contribute to the activity of the MUC1 promoter is not conclusively disproved by the results of our experiments because of two caveats. First, transient transfection assays may have limited usefulness in evaluating transcriptional functions of some DNA structural elements. It is hypothesized that part of the energy for forming H-DNA in vivo comes from negative supercoiling that occurs upstream during transcription. The general chromatin structure surrounding a gene that is being transcribed (14) is probably very different from the structure of plasmid DNA in transient transfection assays, and it is unlikely that the same degree of upstream negative supercoiling is achieved in transfected plasmids. As such, transient transfection assays may not be sufficiently sensitive to detect effects on transcription that are mediated by altered DNA structure. A second caveat is that some of the constructs evaluated in these assays contained small insertions flanking the M-PMR3 element due to the method of assembly. It is possible that these alterations in sequence had other affects that compromised the novel function of this element, especially given its proximity to a putative Sp1 site that may be utilized (7).

In summary, deletion or substitution of the M-PMR3 element in the MUC1 promoter results in a slight stimulation of transcription when evaluated in transient transfection assays. This activity is observed in MUC1-expressing cell lines. Alterations in the element had no effect on transcription in a MUC1-nonexpressing cell line. The M-PMR3 element can form H-DNA under certain in vitro conditions; however, its activity in transient transfection assays is not correlated with its inherent ability to form H-DNA isomers.