Mouse Mammary Tumor Virus Sequences Responsible for Activating Cellular Oncogenes

Abstract

Integration of mouse mammary tumor virus (MMTV) near the int genes results in the inappropriate expression of these proto-oncogenes and initiates events that lead to the formation of mammary adenocarcinomas. In most cases, the MMTV provirus integrates in a transcriptional orientation opposite that of the int genes. We have used a novel, vector-based system designed to recapitulate the integration of MMTV upstream of the int-2 promoter. Compared to a cellular promoter or another retroviral promoter, the MMTV long terminal repeat (LTR) in this configuration is particularly efficacious at activating the int-2 promoter. The sequences responsible for enhancing the activity of the int-2 promoter map to two domains in the 5′ end of the MMTV LTR. One domain is a previously defined element; the second is an element delineated by these studies that acts synergistically with the first. Both of these elements display mammary cell-specific activity. Thus, even though the MMTV promoter itself is weak without hormonal stimulation, viral integration can position the 5′ LTR elements to efficiently activate transcription from cellular proto-oncogenes. Other functional elements in the LTR have little effect on the activation of the int-2 promoter. Even stimulation of the MMTV promoter with steroid hormones only modestly activates transcription from the int-2 promoter, suggesting that the 5′ elements of the LTR are the predominant determinants of the tissue- and orientation-specific activation of cellular promoters by MMTV.

Mouse mammary tumor virus (MMTV) is a type B retrovirus that is transmitted either endogenously through the germ line or exogenously as infectious viral particles in the mother’s milk (32). Although the virus is latently transforming, MMTV does not encode an oncogene. Like other retroviruses, proviral copies of MMTV integrate throughout euchromatin. Analysis of integration sites in MMTV-induced tumors demonstrated preferential integration near a limited set of genes, termed the int genes. Viral integration is associated with increased and inappropriate expression of the int genes, which in turn plays a key role in the disruption of cellular homeostasis leading to mammary tumorigenesis (reviewed in reference 37). The provirus is integrated frequently outside of the coding region of the int genes and only occasionally within (6). Transcription of MMTV at both upstream and downstream integration sites is almost always directed away from the int genes. Thus, increased oncogene expression occurs via an enhancer insertion mechanism rather than a promoter insertion event (28).

A common target for MMTV integration is the fibroblast growth factor (FGF) family of genes. Three of the nine family members, int-2 (Fgf-3), hst (Fgf-4), and Fgf-8, are activated by MMTV proviral insertion (29, 30, 36). The int-2 locus was the second MMTV integration site identified and is the most frequent target of MMTV integration in BALB/CfC3H and RIII mice (4). The int-2 (Fgf-3) gene encodes a 27-kDa protein that is expressed during embryogenesis but not in normal adult tissues (21, 39). Transgenic mice bearing int-2 coding sequences linked to the MMTV long terminal repeat (LTR) develop mammary gland and prostatic hyperplasia, suggesting that int-2 can act as a potent epithelial growth factor in these organs (23).

MMTV is almost exclusively expressed in the mammary gland, but very low levels of message can be detected in salivary glands, lungs, kidney, and lymphoid tissues (11). Enhancer elements that influence the mammary cell-restricted expression of MMTV have been mapped to the 5′ end of the LTR. When progressive deletions from the 5′ end of the LTR were made, MMTV LTR-growth hormone transgenes were less efficiently transcribed in the mammary glands of transgenic mice (35). A 5′ fragment of the LTR (nucleotides −1185 to −863) upstream of a minimal MMTV promoter (−109 to +103) was as effective at targeting expression of a transgene to the mammary gland as a full-length LTR (35). Enhancer elements have been further characterized by assaying the promoter activity of deletion constructs by transient transfection, comparing mammary and nonmammary cell lines. Gel mobility shift and footprinting assays have delineated a complex array of protein binding sites in the 5′ end of the MMTV LTR (13, 14, 18–20, 41). The activity of the MMTV promoter is greatly enhanced by steroid hormones, most notably glucocorticoids and progestins (40). Without hormones the MMTV promoter is weak, even in the mammary gland, yet activates int expression independently of hormone (34). int-2 expression itself is not significantly increased by hormone (34).

In this study, we investigated the hypothesis that the MMTV LTR is particularly effective as an enhancer, activating an oppositely oriented promoter. Furthermore, activation of the int genes by mammary cell-specific enhancer elements in the 5′ end of the MMTV LTR was assessed. We have constructed a series of vectors that recapitulate MMTV integration at the int-2 genomic locus and enable simultaneous monitoring of expression from both the int-2 and MMTV promoters. The arrangement of these vectors mimics the arrangement of transcriptional control sequences found in MMTV-induced tumors. Using these vectors in transient transfection experiments, we have defined the MMTV sequences that activate cellular proto-oncogenes in a mammary cell-specific manner.

MATERIALS AND METHODS

Plasmids.

Murine int-2 promoter sequences were derived from the int-2c plasmid, obtained from the American Type Culture Collection (Rockville, Md.). A 1,089-bp SacI/NgoMI fragment containing the three defined int-2 promoters (8) was inserted upstream of the luciferase reporter gene in pXP2 (24) at the SacI and BglII sites of the multiple cloning site to make int-2/luciferase. The chloramphenicol acetyltransferase (CAT) reporter gene was inserted upstream of the int-2/luciferase transcription unit in the opposite transcriptional orientation to create the vector int-2/luciferase + CAT (ILC).

MMTV or other promoter sequences were introduced into the ILC vector to direct transcription of the CAT gene and enhance expression directed by the int-2 promoter. Except as noted, the MMTV LTR sequences that were introduced into the ILC vector were derived from an MMTV-CAT vector that has a chimeric LTR derived from the C3H and GR strains of MMTV (GenBank accession no. J02274 and V01175, respectively). The LTR is predominantly derived from the C3H strain, with GR sequences spanning from −291 to +83 (AlwNI to PpuMI). C3H sequences comprise −1194 to −292 and +84 to +99. A chimeric MMTV LTR was used because the promoter activity of the C3H strain is barely measurable and the GR strain, while behaving like the chimera, lacks restriction sites used to generate many of the deletions including those that define the mammary cell-specific elements.

Approximately 500 bp of the Rous sarcoma virus (RSV) promoter were inserted upstream of the CAT gene in ILC to create ILC+RSV. ILC+FAS was made by excising the full-length fatty acid synthase (FAS) promoter (−2195 to +65) from pFAS-CAT FL (1) and inserting this fragment into the ILC vector at the HindIII and NcoI sites.

Cell culture.

The T47D(A1-2) (27) human breast cancer, Ltk− mouse fibroblast, and HeLa human cervical carcinoma cell lines were cultured in minimum essential medium supplemented with 5% fetal bovine serum (FBS; HyClone, Logan, Utah), 10 mM HEPES, 1% nonessential amino acids, 2 mM L-glutamine, penicillin (50 U/ml), and streptomycin (50 mg/ml). G418 (CalBiochem, San Diego, Calif.) was added to the T47D(A1-2) culture medium at 200 μg/ml. The COMMA1D mouse mammary (5, 17) and the CHO-K1 Chinese hamster ovary cell lines were cultured in Dulbecco’s modified Eagle medium–nutrient mixture F-12 containing 5% FBS, glutamine, penicillin, and streptomycin. Additionally, the COMMA1D cells were supplemented with insulin (10 μg/ml; Sigma, St. Louis, Mo.) and epidermal growth factor (7.5 ng/ml; Upstate Biotechnology Inc., Lake Placid, N.Y.). The DU145 human prostate adenocarcinoma cell line was cultured in RPMI medium containing 10% FBS, penicillin, and streptomycin. All cell culture reagents were purchased from Gibco BRL (Grand Island, N.Y.) unless otherwise noted.

Transient transfections.

Twenty-four hours before transfection, 1.4 × 10⁶ cells were plated per 60-mm-diameter culture dish or 5 × 10⁵ cells were plated per well of a six-well plate. All cells were plated in duplicate dishes for each plasmid tested. T47D(A1-2) and Ltk− cells were transiently transfected by the DEAE-dextran–dimethyl sulfoxide (DMSO) shock method (16). Cells were exposed for 2 h to 1 ml of growth medium containing DEAE-dextran at 200 μg/ml, 100 μM chloroquine, test plasmid at 2 μg/ml, and pCH110 (an internal control plasmid expressing β-galactosidase from a simian virus 40 promoter [10]) at 0.2 μg/ml. The transfection medium was removed, and the cells were shocked with DMSO (15% DMSO in 1 ml of shock buffer [137 mM NaCl, 5 mM KCl, 0.7 mM Na₂HPO₄, 6 mM glucose, 21 mM HEPES]) for 6 min. COMMA1D cells were transiently transfected by a calcium phosphate method (33) using 20 μg of test plasmid and 2 μg of pCH110 per ml. After a 4-h incubation with DNA, the cells were shocked with 15% glycerol for 2 min. HeLa and CHO-K1 cells were transiently transfected by using LipofectAmine (Gibco BRL). Each well of a six-well plate received 7 μl of lipid, 1 μg of test plasmid, and 0.25 μg of pCH110. After a 5-h incubation, the lipid was washed away and replaced with growth medium. DU145 cells were transiently transfected by using 10 μl of SuperFect (Qiagen, Valencia, Calif.), 2 μg of test plasmid, and 0.25 μg of pCH110 with a 5-h incubation, after which the lipid was washed away and replaced with growth medium. Cells were allowed to grow for 48 to 72 h before harvest.

Assays for reporter gene activity.

To harvest the cells, dishes were washed and lysed, and cell debris was removed by centrifugation (25). Protein concentrations, CAT assays, and luciferase assays were performed as described previously (2, 25, 26). T47D(A1-2) cells have 10 copies of an MMTV-luciferase vector (pHHLuc) stably integrated into the genome (27). In these cells, basal levels of luciferase activity are virtually undetectable and hormonal induction of the integrated MMTV promoter by progestins is very low. We subtracted these background levels from our experimental results. β-Galactosidase activity was measured from 2 to 5 μl of cell extract by using a Galactolight/Galacton-Plus kit (Tropix, Bedford, Mass.). Duplicate aliquots were assayed from each reaction; results are presented as means ± standard errors of the means (SEM).

RESULTS

MMTV sequences enhance the activity of the int-2 promoter in a mammary cell-specific fashion.

To test the hypothesis that MMTV activates the int-2 promoter, we constructed a series of vectors that recapitulate MMTV integration at the int-2 genomic locus and enable simultaneous monitoring of expression from both the int-2 and MMTV promoters. Schematics of these constructs are shown in Fig. 1, and details of the cloning are given in Materials and Methods. The arrangement of these vectors mimics the arrangement of transcriptional control sequences found in MMTV-induced tumors. Transcription from the MMTV promoter is directed in the opposite orientation with respect to int-2 promoter driven transcription. Using this series of constructs, we transiently transfected both mammary and nonmammary cell lines. The CAT activity is a measure of MMTV promoter activity, and luciferase activity represents int-2 promoter activity.

FIG. 1.

Vectors to assess the MMTV LTR sequences involved in enhancing the activity of the int-2 promoter. At the top is a schematic outlining the construction of reporter vectors that mimic MMTV integration at the int-2 locus. Note the divergent orientation of the two transcription units. The mouse int-2 promoter was cloned into the promoterless luciferase expression vector pXP2. Next, the CAT reporter gene was inserted into int-2/luciferase in the opposite transcriptional orientation, to create the ILC construct. The ILC construct served as the parent vector for the remaining constructs in this series, with various promoters inserted in a position to drive expression of the CAT gene. Below is shown the linear arrangement of each construct and any alterations made to the MMTV or int-2 promoters. The black box and white circle within the MMTV promoter represent enhancer elements located in the 5′ end of the LTR. Note that the elements are not drawn to scale. The int-2 promoter fragment used is about the size of the MMTV LTR and is reported to have three start sites (8).

In T47D(A1-2) human breast cancer cells, the int-2 promoter alone was able to direct very little luciferase activity (Fig. 2A). The int-2 gene is not normally expressed in adult mammary epithelium (39). Introduction of the CAT reporter gene into the vector had no effect on int-2 promoter activity. However, the presence of a full-length MMTV promoter in an opposite transcriptional orientation greatly enhanced the activity of the int-2 promoter, evidenced by the increased luciferase expression. The activity of the luciferase reporter gene was dependent on the presence of the int-2 promoter. When the int-2 promoter was deleted from the vector (ILC+MMTV Δint-2), luciferase activity was reduced by 85%. Cryptic promoter sequences in the LTR or vector sequences may account for the small residual activity. These results indicate that our vector-based system recapitulates, in a breast cancer cell line, the effect of MMTV integration at the int-2 genomic locus. Results for the COMMA1D mouse mammary epithelial cell line were similar (Fig. 2B). The int-2 promoter was silent, or nearly so, until the MMTV LTR was inserted upstream. Results for transiently transfected nonmammary cell lines, such as Ltk− mouse fibroblast cells, HeLa human cervical carcinoma cells, CHO-K1 Chinese hamster ovary cells, or DU145 human prostate carcinoma cells, were markedly different (Fig. 2C to F). The int-2 promoter alone had high basal activity, and the presence of the MMTV promoter had little effect on int-2 promoter activity. If anything, the MMTV LTR inhibited the activity of the int-2 promoter. Thus, the ability of MMTV to activate the int-2 promoter appears to be restricted to mammary cells.

FIG. 2.

Effect of MMTV sequences on int-2 promoter activity in various cell lines. T47D(A1-2) human breast cancer (A), COMMA1D mouse mammary (B), Ltk− mouse fibroblast (C), HeLa human cervical carcinoma (D), CHO-K1 Chinese hamster ovary (E), and DU145 human prostate adenocarcinoma (F) cell lines were transiently transfected with the indicated vectors as described in Materials and Methods. Whole-cell extracts were assayed for luciferase activity, which was normalized to β-galactosidase activity. The bar graphs represent the averages of three to five experiments where each condition was performed in duplicate. The error bars represent the SEM. Results are expressed as a percentage of the activity measured for the ILC+MMTV construct, which was set at 100%.

We then compared the activating potential of the MMTV promoter relative to two other promoters, the cellular promoter from the FAS gene and the strong viral promoter from RSV. Again, CAT activity is a measure of MMTV promoter activity and luciferase activity represents int-2 promoter activity. While the FAS promoter was approximately twice as active as the MMTV promoter (Fig. 3A), it was not able to enhance the activity of the int-2 promoter in T47D(A1-2) cells. The RSV promoter was almost 30 times more active than the MMTV promoter in this same cell line yet activated int-2 to about 60% of the level induced by the MMTV promoter. Thus, in mammary cells the MMTV LTR effectively activates a second promoter despite exhibiting relatively weak promoter activity itself. In contrast, in Ltk− mouse fibroblast cells (Fig. 3B), neither the MMTV promoter nor other promoters could further activate the constitutive int-2 promoter.

FIG. 3.

Preferential activation of int-2 by MMTV. T47D(A1-2) cells (A) or Ltk− cells (B) were transfected with vectors containing the MMTV, FAS, and RSV promoters, which direct an oppositely oriented transcription unit upstream of the int-2 promoter. Both luciferase and CAT activities were measured and normalized to β-galactosidase activity. Luciferase activity is a measure of int-2 promoter activity (white bars), and CAT activity a measure of the activity of the indicated promoter (black bars). The bar graphs represent the averages of three to four experiments performed in duplicate, with the error bars representing the SEM. Results are expressed as a percentage of the activity measured for the ILC+MMTV construct, which was set at 100%. The average raw CAT values for ILC+MMTV were 25.8 pmol/min with an assay background of 0.21 pmol/min in T47D(A1-2) cells and 57.9 pmol/min with an assay background of 0.37 pmol/min in Ltk− cells. Note that the graph containing the ILC+RSV data is on a different scale in panel A.

Contribution of MMTV sequences in activating the int-2 promoter.

The surprising ability of the MMTV LTR to enhance the activity of the oppositely oriented int-2 promoter relative to its own promoter strength suggested that the enhancing activity may be separable from MMTV promoter activity. Results of the next series of experiments support this conclusion.

The MMTV LTR used in these studies is a chimera of LTR sequences from the C3H and GR strains of MMTV (see Materials and Methods). In T47D(A1-2) mammary carcinoma cells, in the absence of steroid induction, the C3H promoter is very weak whereas the GR promoter is stronger. The chimeric LTR has somewhat greater basal transcriptional activity than even the GR LTR (unpublished observation). The ILC+C3H MMTV construct contains MMTV LTR sequences from the C3H LTR. As shown in Fig. 4A, the C3H promoter had little basal transcriptional activity. CAT expression directed by the C3H promoter was only 3% of that of the chimeric LTR. Nonetheless, the C3H LTR still commanded a significant enhancement of luciferase activity directed by the int-2 promoter.

FIG. 4.

Contribution of MMTV sequences in enhancing the activity of the int-2 promoter. (A) Constructs with a weak MMTV LTR (C3H strain) or an MMTV LTR with a deletion that removes the basal promoter were assessed for the ability to activate the int-2 promoter in T47D(A1-2) cells. (B) Constructs that contain deletions throughout the middle of the MMTV LTR were assessed for the ability to activate the int-2 promoter. Constructs with deletions in the 5′ MMTV LTR sequences were assessed by transient transfection in T47D(A1-2) cells (C) and Ltk− cells (D). The data represent means for three to six experiments ± 1 SEM.

Deletion of the proximal MMTV promoter gave a similar result (Fig. 4A). The ILC+Δprom fwd MMTV construct contains the chimeric MMTV LTR, with a deletion removing the proximal MMTV promoter from −108 to +99. The deleted region contains the TATA box, two hormone response element half-sites, and binding sites for nuclear factor I (NF-I) and Oct-1 (9). While the transcriptional activity of the ILC+Δprom fwd MMTV construct was severely impaired in T47D(A1-2) cells, it retained a substantial ability to enhance the activity of the int-2 promoter. The promoterless LTR was equally effective in the opposite orientation (ILC+Δprom MMTV rev), demonstrating that the enhancer elements within the MMTV LTR activate the int-2 promoter in either orientation. Thus, weak or disabled MMTV promoters that retain the 5′ LTR sequences are still able to activate the int-2 promoter and do so in an orientation-independent fashion.

We made a series of deletions along the MMTV LTR to test which sequences are responsible for enhancing the activity of the int-2 promoter. The constructs tested in the analysis shown in Fig. 4B remove sequences from the middle of the MMTV LTR, where regulatory elements that inhibit expression in certain nonmammary cells have been mapped (3, 7, 12, 15, 20, 22, 31). The ILC+ΔClaI/StuI construct removes 224 bp from the MMTV promoter, from −862 to −638. The ILC+ΔStuI/AlwNI construct deletes 347 bp from −638 to −291. The removal of these regions of the MMTV promoter had only modest effects on MMTV promoter activity and on the activation of the int-2 promoter in T47D(A1-2) cells. Therefore, neither the middle portion nor the proximal promoter region of MMTV can account for the enhancing activity that MMTV exerts over the int-2 promoter.

Role of the 5′ enhancer elements in activating int-2.

It has been suggested that enhancer elements in the 5′ end of the MMTV LTR play a role in activating cellular proto-oncogenes because the 5′ end of the LTR is almost always positioned closest to the int locus upon integration of the MMTV provirus (reviewed in reference 28). We tested this hypothesis by deleting three regions at the 5′ end of the MMTV LTR: a 100-bp deletion to the FspI site (ILC+ΔFspI), a 240-bp deletion to the EarI site (ILC+ΔEarI), and an internal deletion of 76 bp between the BsaBI and ClaI sites (ILC+Δ76). Deleting sequences upstream of the FspI site had no effect on the MMTV promoter itself or on the activation of the int-2 promoter in T47D(A1-2) cells (Fig. 4C). A 5′ deletion to the EarI site removed the previously identified Ban2 enhancer element (13, 14, 18), designated by the black box in Fig. 1. When this enhancer was removed, MMTV was no longer able to support transcription from its own promoter or to enhance the activity of the int-2 promoter.

A 76-bp internal deletion from BsaBI to ClaI (−938 to −862) identifies a new enhancer element (represented by the white circle in Fig. 1). When this region was deleted (ILC+Δ76), the MMTV promoter was no longer functional and the ability to activate the int-2 promoter was completely lost. This effect was also seen when the ILC+Δ76 construct was tested in COMMA1D mouse mammary cells (data not shown). Thus, the two enhancer elements defined by the EarI deletion and the BsaBI/ClaI (Δ76) deletion appear to function synergistically in mammary cells, since neither element alone was sufficient to activate int-2. Both deletions appear to be mammary cell-specific enhancer elements, as they had no effect on MMTV promoter activity in fibroblast cells (Fig. 4D). Taken together, our results demonstrate that the 5′ enhancer elements are necessary and sufficient to enhance the activity of the int-2 promoter and appear to function in a mammary cell-specific fashion.

To further characterize the enhancing activity defined by the BsaBI/ClaI (Δ76) deletion, two smaller deletions within this region were made. ILC+ΔBB MMTV contains a deletion from BsaBI to Bsu36I, removing 35 bp, and ILC+ΔBC MMTV has a deletion of 38 bp, from Bsu36I to ClaI. These constructs were transiently transfected in both T47D(A1-2) and Ltk− cells to test their activity. When either half of this Δ76 region was removed, MMTV promoter activity was severely compromised in mammary cells and the ability to enhance int-2 promoter activity was also lost (Fig. 5A). It appears sequences from both halves of the Δ76 region are required for enhancing the activity of both the MMTV and int-2 promoters. Again, this region of the MMTV LTR appears to play a role in mammary cell specificity, as there was no change in promoter activity in Ltk− fibroblast cells transfected with these two deletion constructs (Fig. 5B).

FIG. 5.

Effect of smaller deletions or point mutations in the 5′ MEM element on enhancing the activity of the int-2 promoter. Two smaller deletions that remove the 5′ and the 3′ half of the MEM element (ILC+ΔBB and ILC+ΔBC, respectively) were made in the MMTV LTR and tested in T47D(A1-2) (A) or Ltk− cells (B). (C) Three or five base changes were made in protein binding sites located within the MEM element in the context of the ILC+MMTV construct. (D) T47D(A1-2) cells were transiently transfected with these vectors, and both MMTV and int-2 promoter activities were assessed. The data represent means for three to seven experiments ± 1 SEM.

We made point mutations in two previously characterized binding sites within the ΔBC region (Fig. 5C). A consensus binding site for NF-I had been identified at −896 to −882 (14), and a binding site for an unidentified factor called mammary cell-activating factor (MAF) (19, 38) is located just upstream of the NF-I site. As shown in Fig. 5D, the point mutations in the NF-I site decreased the activity of MMTV to 15% of wild-type MMTV activity, but the construct with the mutated MAF site retained approximately 50% of wild-type MMTV activity. The ability to enhance the int-2 promoter in mammary cells was proportional to the amount of MMTV promoter activity displayed by these constructs. Because the ΔBB region and the NF-I site have never been implicated in the mammary cell-specific expression of MMTV, and the MAF mutation had only modest effects, we have designated this novel enhancer element the MEM (mammary-specific enhancer of MMTV) element.

Independent activity of the MEM element.

To test the function of the MEM element independent of other 5′ MMTV LTR sequences, we made a construct containing three copies of the MEM element upstream of a minimal MMTV promoter (Fig. 6A). As shown in Fig. 6B, the minimal MMTV promoter (−108 to +99) could not support transcription of the CAT reporter gene in mammary cells and showed no enhancement of int-2 promoter activity. However, three copies of the MEM element placed upstream of the minimal MMTV promoter were able to enhance the activity of the int-2 promoter to levels equal to that of the full-length MMTV LTR and enhanced MMTV-mediated transcription to levels four times that of the full-length LTR in a mammary cell line. In fibroblast cells (Fig. 6C), the minimal MMTV promoter was four times more active than the full-length MMTV LTR, due to the removal of negative regulatory elements. The addition of the MEM element was somewhat inhibitory, reducing both CAT and luciferase activities. These data support the idea that the MEM element can function as a mammary cell-specific enhancer element on its own.

FIG. 6.

The MEM element can function as an independent enhancer unit. (A) Three copies of the MEM element (from the HinfI site at −956 to the ClaI site at −862) were placed upstream of a minimal MMTV promoter (−108 to +99) in the context of the ILC vector to assess the enhancing activity of the MEM element alone. T47D(A1-2) cells (B) or Ltk− cells (C) were transiently transfected with these constructs to test their transcriptional activity.

Effect of steroid hormones on MMTV and int-2 promoter activity.

The MMTV LTR has been used as the prototypical hormone-responsive promoter for studies of steroid hormone action. Because there is a large increase in transcription from the MMTV promoter in response to steroid hormones, we wanted to see what influence a hormone-activated MMTV LTR had over the activity of the int-2 promoter. Figure 7 shows the results of transient transfections of T47D(A1-2) cells with the ILC+MMTV construct. As expected, there was a large increase in MMTV-driven transcription in response to R5020, a synthetic progestin. The induction in CAT activity over basal levels was approximately 15-fold. There was also an increase in int-2-mediated transcription (3.7-fold), but it was substantially less than the level of induction of the MMTV promoter activity. We conclude that the ability of MMTV to enhance the activity of the int-2 promoter is primarily dependent on the upstream LTR elements and is less dependent on steroid hormones.

FIG. 7.

Hormone induction of the MMTV promoter and its effect on int-2 promoter activity in T47D(A1-2) cells. The ILC+MMTV construct was tested by transient transfection in T47D(A1-2) cells. Experiments were performed as described in the text except that cells were treated with 10 nM R5020 20 h before harvesting. The bar graphs represent the means of three to six experiments performed in duplicate, with the error bars indicating the SEM. Values are expressed as a ratio of hormone-induced transcription over basal transcription, to give the fold induction. Locations of the hormone response elements (HRE) are shown relative to the two 5′ MMTV LTR elements that activate int-2-directed transcription (black box and white circle).

DISCUSSION

The 5′ end of the MMTV LTR is thought to play a role in activating the promoters of cellular proto-oncogenes. This hypothesis has been based on the observation that the 5′ end of the integrated MMTV provirus, where tissue-specific enhancer elements are thought to reside, is positioned closest to the promoter of the activated cellular proto-oncogene (28). Our experimental results provide a direct link between the 5′ enhancer elements and their role in enhancing the activity of a cellular proto-oncogene.

The int-2 gene is silent in mammary epithelium until a copy of the MMTV LTR is inserted upstream of the int-2 promoter in an opposite transcriptional orientation (39). We have used a vector-based system that recapitulates in vivo circumstances to delineate sequences in the MMTV LTR responsible for enhancing the activity of a second promoter (int-2). As it is in vivo, the int-2 promoter is inactive when transfected into a nontumor mouse mammary epithelial cell line, COMMA1D, or into a human mammary tumor cell line, T47D(A1-2). The introduction of MMTV sequences in the vector upstream of the int-2 promoter strongly activates int-2 promoter activity. In contrast, the int-2 promoter is constitutively active in nonmammary cell lines, and MMTV does not enhance this activity further.

By testing constructs with deletions across the MMTV LTR, we have shown that two domains in the 5′ end of the LTR enhance the activity of a cellular proto-oncogene in a synergistic manner. A cluster of four binding sites, previously designated as the Ban2 enhancer (13, 14, 18), corresponds to the 5′-most int-2-activating domain. This region includes binding sites for AP-2, partially characterized proteins, such as an Ets-related protein, a member of the NF-I/CTF family, and an uncharacterized factor (mp4). Deletion of this region from the MMTV LTR results in a large reduction of MMTV promoter activity itself, as well as the ability to enhance the activity of int-2. The Ban2 enhancer functions preferentially in mammary cells. Removal of this region has no effect on MMTV promoter activity in fibroblast cells (compare results for the ILC+ΔEar construct in Fig. 4C and D).

We have characterized a second enhancer element 3′ of the Ban2 enhancer that we call the MEM element. As with the ΔEar construct, deletion of the MEM element results in a loss of both MMTV promoter activity and, coordinately, the ability to enhance int-2 promoter activity. The Ban2 and MEM elements appear to function synergistically, since neither element can compensate for the loss of the other, although multimerization of the MEM element can strongly activate transcription in mammary cells. MEM is a complex enhancer composed of multiple binding sites. Previous DNase I footprinting studies identified three protected regions in the MEM element, designated F4, F5, and F2 (19). The F2 footprint is in the 3′ region of the functionally defined MEM element that contains a consensus NF-I binding site and a poorly characterized site for MAF. MAF was proposed to be a mammary cell-specific factor, even though its binding activity was present in both mammary and nonmammary cell lines (19). Based on homology of the footprinted sequences, the MMTV LTR was thought to contain two MAF sites, the second in the Ban2 enhancer. However, subsequent studies have determined that the latter MAF site binds an Ets-related factor whereas the downstream site in the MEM element does not have the GGAA core sequence required for binding Ets proteins (38). In our hands, the point mutations made in the F2/MAF binding site had only modest effects on the transcriptional activity of MMTV and the ability to enhance int-2 expression in a mammary cell line. In contrast, the base changes made in the F2/NF-I site reduced MMTV promoter activity to the same extent as deleting the whole region of the LTR (compare ILC+ΔBC in Fig. 5A with ILC+mNF-I in Fig. 5D). Previous studies had not implicated this NF-I site as being important for the mammary cell specificity of MMTV (14, 19), yet it appears to play a role in this phenomenon.

The F4 and F5 binding sites in the MEM element are poorly characterized. We have not observed F5 footprinting in nuclear extracts from T47D(A1-2) cells. The F4 region was shown to bind a factor present in all cell lines, as well as a factor found only in nonmammary cells (19). On this basis, it was suggested that the F4 region did not contribute to the mammary cell-specific expression of MMTV. Our functional data indicate that the F4 region is vitally important for both the mammary cell-specific expression of MMTV and the activation of cellular oncogenes. When this region is deleted from the MMTV LTR in the ILC+ΔBB construct, there is a loss of both MMTV and int-2 promoter activities in a mammary cell line (Fig. 5A) but no change in a fibroblast cell line (Fig. 5B). Furthermore, we found in 6 of 9 mammary cell lines a binding activity to sequences in this region that is not present in any of 11 nonmammary cell lines of both epithelial and nonepithelial origin (unpublished data). More work is needed to characterize the activities that define the MEM element.

For mouse strains GR, BR, and RIII, the growth of mammary tumors induced by MMTV begins as a very pregnancy-dependent phenomenon, with the size of the tumor increasing with each subsequent pregnancy but regressing at the withdrawal of lactogenic hormones (4). After three or four pregnancies, the tumors become pregnancy independent and continue to grow in the absence of hormones. It was of interest, therefore, to see what effect a hormone-activated MMTV LTR had on the activity of the int-2 promoter. As shown in Fig. 7, there is a modest (less than fourfold) increase in int-2 promoter activity in the presence of the synthetic progestin R5020. However, the MMTV promoter itself is stimulated more than 15-fold over basal activity. Thus, the amount of enhanced int-2 activity does not correlate with the increase in strength of the MMTV promoter, and much of the int-2 activation occurs independent of hormonal stimulation. These data are consistent with the findings of Sonnenberg et al. (34). Making use of a mammary cell line (RAC-10P) that has a copy of MMTV integrated near the int-2 locus, they demonstrated by Northern analysis that the levels of int-2 transcript did not change after addition of dexamethasone, although there was a large stimulation of MMTV-driven transcription.

By using a novel vector-based system that recapitulates the MMTV integration at the int-2 proto-oncogene locus, we have demonstrated the remarkable ability of MMTV to enhance expression of an oppositely oriented promoter despite the fact that the MMTV promoter itself is relatively weak. This is most evident with the weak C3H strain promoter. We have defined two discrete elements in the MMTV LTR responsible for this phenomenon: (i) the previously characterized Ban2 enhancer and (ii) the MEM element described in this report. Thus, a 213-bp region in the 5′ end of the MMTV LTR, containing multiple binding sites for cis-acting factors, is responsible for the mammary cell-specific expression of MMTV as well as for directing the inappropriate expression of cellular genes in a tissue-specific fashion. In future studies, we will seek to identify these cis-acting factors and determine how the combination of multiple transcription factors is able to confer tissue-specific gene expression.