INTRODUCTION

Top Abstract Introduction Materials and Methods Results Discussion References

Mouse mammary tumor virus (MMTV) can be transmitted through milk or through the germ line (1). Although most mouse strains carry several MMTV proviruses, only some transmit an exogenous virus. The nonpathogenic MMTV proviruses are transcribed, but little or no envelope protein is produced. However, the same subtype synthesizes superantigen (SAg) from the viral transcript, which in turn renders the host resistant to infections. Each provirus is characterized by a distinct SAg that interacts with the T-cell receptor V element through its COOH-terminal portion (19). Different mouse strains are characterized by a distinct pattern of proviruses. For example, MTV-3 is found in 129 and GR mice (8, 19) but not in C57BL/6 and BALB/C mice (18).

Transcription of the proviral genome originates within the 5' long terminal repeat (LTR) and proceeds to a transcriptional stop and polyadenylation site within the 3' LTR. Both prolactin and glucocorticoids control transcription from the MMTV LTR through distinct regulatory sequences (6, 10, 15, 23, 26). As a result of hormonal stimulation during pregnancy, virus production dramatically increases during lactation (2). Three types of transcripts can be detected: a full-length RNA of approximately 8 kb encoding the gag and pol products, and two shorter transcripts encoding the envelope protein (env) and superantigen (sag), respectively (Fig. 4A). The latter two transcripts are splice products and share the 5' splice donor site. MMTV lacks an oncogene and probably induces tumors by acting as an insertional mutagen that activates the expression of cellular oncogenes juxtaposed to the insertion site in the host chromosome (16, 17). A mammary-specific enhancer within both LTRs is responsible for the activation of promoters located outside the integration site. In some cases, transcription originating from the 3' LTR has been linked to the activation of juxtaposed oncogenes.

The MMTV LTR has been a rich source for discoveries of transcription elements and mechanisms of transcriptional regulation. Transcription originating in the LTR is induced by steroid hormones and is controlled by numerous transcription elements, including binding sites for the glucocorticoid receptor, NF1, Oct proteins, and many others. Transcriptional activation in mammary tissue increases during pregnancy and peaks during lactation, similar to that seen for milk proteins, such as the whey acidic protein (WAP) (3). Transcriptional activation of WAP during pregnancy is controlled by prolactin through the Jak2/Stat5 pathway (13). Mutation of the Stat5a binding site (TTCNNNGAA) within the promoter fully abrogates transcription in transgenic mice (11), and deletion of the Stat5a gene from mice leads to downregulation of WAP (14). It has recently been shown that both Stat5a and Stat5b bind to the MMTV LTR and are required for its activation in mammary tissue in vivo (20).

We have generated mice with an inactive Stat5a gene and established its critical role in mammary gland development and function (12). In a quest to identify genes and signaling pathways that are controlled by the prolactin-Jak2/Stat5 axis (either directly or indirectly), we prepared subtractive cDNA libraries. This approach led us to the discovery of transcripts that originate in both LTRs of the MTV-3 locus, proceed into the juxtaposed host genome, and result in a family of splice and polyadenylation variants. Furthermore, our study provided insight into the capacity of Stat5a to negatively regulate transcription from the MMTV LTRs.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and Methods Results Discussion References

Mice. The ES cell line used for gene targeting was derived from the 129SvEv strain. Stat5a-hemizygous (+/) mice were subsequently bred with C57BL/6 mice, and the Stat5a-null allele was maintained in a mixed background. This is commonly done since 129 mice are poor breeders and have small litter sizes. The MTV-3 locus characterized in this report is present in the 129 but not C57BL/6 strain. It is located on chromosome 11 (70.5 centimorgams [cM]) approximately 10 cM distal to the Stat5a locus. We have observed a recombination frequency of less than 10% between the MTV-3 and Stat5a loci, which explains the cosegregation of the Stat5a-null and MTV-3 alleles.

cDNA subtraction library. To identify genes overexpressed in Stat5a-null mice at parturition compared to control littermates, we generated a subtractive library. PCR-based subtractive hybridization was performed with the Clontech PCR-Select DNA subtraction kit (Catalog no. K1804-1). Briefly, in this procedure, tester cDNA (containing the cDNAs to be cloned) was synthesized from RNA obtained from mammary tissue of mice homozygous null for Stat5a. Driver cDNA (used in excess as a reference cDNA) was synthesized from RNA from littermates containing two intact copies of the Stat5a gene. After restriction of tester and driver cDNAs with RsaI to obtain shorter, blunt-end molecules, two tester populations with different adapters were created. The driver cDNA had no adapters. Subsequent hybridization steps led to the enrichment of differentially expressed sequences from which templates for PCR amplification were generated. By using suppression PCR, only differentially expressed sequences were amplified exponentially. The background was reduced, and differentially expressed sequences were further enriched in the second amplification step.

RNA isolation and Northern blotting. To verify the differential expression between Stat5a-null and control mice, RNA blot analyses were performed (experimental details and extensive Northern analyses on 22 genes can be viewed online [14a]). RNA was prepared from mammary tissue of Stat5a-null mice, wild-type littermates, and C57BL/6 mice within 12 h after parturition and analyzed by Northern blotting. Liver RNA was used as a nonmammary control. Total RNA was extracted using the acid guanidinium thiocyanate-phenol-chloroform method (16). Poly(A)⁺ RNA was isolated by using a Poly(A) Quick mRNA isolation kit (Stratagene). The poly(A)⁺ RNA was used for both production of the cDNA library and Northern blot analyses. Twenty micrograms of total RNA of each sample was fractionated in a 1.3% formaldehyde gel. The RNA was transferred to GeneScreen Plus nylon membranes and UV cross-linked in a UV-Stratalinker 1800. PCR fragments amplified with specific internal primers for each clone were labeled with [³²P]dCTP, using a Prime-It II kit (Stratagene) according to the manufacturer's protocol, and used as probes for RNA expression analysis.

Isolation of cDNA clones containing sequences from probes 1 and 2. A cDNA library was generated (Edge Bio Systems vector pEAK8) from mRNA isolated from mammary tissue of Stat5a-null mice within 12 h after parturition. Approximately 2 million nonamplified colonies were plated on 40 140-mm-diameter LB-ampicillin plates. The colonies were transferred to colony/plaque screen hybridization transfer membranes, air dried, fixed by UV irradiation, autoclaved for 5 min, and dried for 10 min in a steam autoclave. Hybridizations were performed for 12 h with [-³²P]dCTP-labeled probes 1 and 2 at 65°C overnight. Positive clones were isolated and colony purified in an additional hybridization screen. Individual positive colonies were transferred into LB broth, grown overnight, and used to inoculate 200 ml of terrific broth with 100 µg of ampicillin per ml. Plasmids were isolated by using a Maxi (Qiagen) plasmid purification kit.

Isolation and subcloning of genomic BAC clones. Probe 2 was used to isolate corresponding bacterial artificial chromosome (BAC) clones from a 129SvJ library (Genome Systems). Hybridization of nylon filters containing individual BAC clones spotted at high density was performed at 65°C according to the protocol of the supplier (Genome Systems). Two BAC clones were isolated, and a restriction map was established by Southern blotting followed by hybridizations with probes 1 and 2. BamHI, HindIII, and EcoRI fragments of both isolated BAC clones were subcloned into the pBluescript II SK (Stratagene) and pZErO-1 (Invitrogen) cloning vectors.

Sequence analysis. Sequencing of plasmid DNA and PCR products was performed by using a standard protocol for cycle sequencing with the ABI PRISM 310 Genetic Analyzer (Perkin-Elmer). Sequence analysis was performed with the EditView 1.0 and Sequencher 3.0 software packages.

Chromosomal localization. Genome Systems determined the chromosomal localization of the BAC clone. The BAC clone, which hybridized to the hh-1 and hh-2 transcripts of the hitchhiker locus, was used as a probe in a fluorescence in situ hybridization (FISH) analysis. BAC clone DNA was labeled with digoxigenin-dUTP by nick translation. Labeled probe was combined with sheared mouse DNA and hybridized to normal metaphase chromosomes derived from mouse embryo fibroblast cells in a solution containing 50% formamide, 10% dextran sulfate, and 2× SSC (1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate). Specific hybridization signals were detected by incubating the hybridized slides in fluorescein-conjugated antidigoxigenin antibodies followed by counterstaining with 4',6-diamidino-2-phenylindole (DAPI). The initial experiment resulted in specific labeling of the terminus of a medium-sized chromosome, which was believed to be chromosome 11 on the basis of DAPI staining. We conducted a second experiment in which a probe specific for the centromeric region of chromosome 11 was cohybridized with the BAC clone. This experiment resulted in specific labeling of the centromere and the telomeric region of chromosome 11. This demonstrated that the DNA from the BAC clone is located in the telomeric of chromosome 11, in the area that corresponds to band 11E2. Of the total of 80 metaphase cells analyzed, 73 exhibited specific labeling.

RT-PCR. Reverse transcription (RT) was performed with the Superscript preamplification system for first-strand cDNA synthesis (GIBCO BRL). Five micrograms of total RNA was transcribed with 1 µl of 2 µM primer at 50°C in total volume of 12 µl. Gene-specific primers were used for the hitchhiker locus (RT primer 5'-TGCTGTCTCCGCTTCCACTGG-3') and for Stat5b (RT primer 5'-CTGGTCCATGTTGGCTGGC-3'). After cDNA synthesis, the RNA was removed with RNase T₁. PCR amplifications of 1 µl of cDNA product from the above reactions were performed with gene-specific primers. The hitchhiker fragment was amplified by using 5' primer 5'-GGCTGCTGCCTCATTAGG-3' and 3' primer 5'-GTGTTGTATCCCTGACTGC-3' in 25 cycles at an annealing temperature of 53°C. The Stat5b fragment was amplified by using 5' primer 5'-GACACTTGCTTCTGCTGG-3' and 3' primer 5'-CAGAGGCTGGTTCGCGAAGCC-3' in 25 cycles at an annealing temperature of 55'C. PCR products were analyzed by electrophoresis in a 2.5% agarose gel.

Nucleotide sequence accession numbers. The following accession numbers have been assigned to the sequences submitted to GenBank: AF120673 (genomic sequence of the hitchhiker locus without the MTV-3 insertion), AF118272 (hh-1 cDNA), AF118273 (hh-2 cDNA), AF118558 (cDNAs for hh-3, hh-4, and hh-5), AF118847 (hh-6 cDNA, full-length sequence of EST 1333114), AF119341 (5' LTR of MTV-3), and AF119342 (3' LTR of MTV-3).

	RESULTS

Top Abstract Introduction Materials and Methods Results Discussion References

Isolation of transcripts from Stat5a-null mammary tissue. In an attempt to isolate genes and identify prolactin controlled signaling pathways in mammary tissue, we prepared subtractive cDNA libraries from Stat5a-null and control mice (9a). A PCR-based subtraction approach was used to clone those mRNAs which are preferentially expressed in mammary tissue from Stat5a-null mice at parturition compared to control mice (see Material and Methods). For this purpose, RNA was prepared from littermates which were either homozygous null for Stat5a or contained two intact copies of the Stat5a gene. The ES cell line containing the targeted Stat5a allele had a 129 genotype, and Stat5a +/ mice were bred with C57BL/6 (see Materials and Methods for the rationale). Initially we identified 96 differentially expressed cDNA clones (see Material and Methods) and further analyzed clones (named probe 1 and probe 2 in subsequent experiments) whose expression appeared to be confined to mammary tissue of Stat5a-null mice (Fig. 1). Probe 1 (see Fig. 1E and 4A for locations of the probes) hybridized to three major transcripts (transcripts I, II, and III) with approximate sizes of 1.5, 2.5, and 4.5 kb (Fig. 1B). Transcripts IV and V were detectable but substantially weaker (Fig. 1B). Probe 2 preferentially hybridized to transcripts III and IV (Fig. 1A). Transcript V was weak but detectable. Neither probe 1 nor 2 hybridized to any transcripts from mammary tissue of C57BL/6 mice and wild-type littermates from the Stat5a-null mice (Fig. 1A and B). A sag probe was used to identify transcripts that contain endogenous MMTV sequences (Fig. 1C). sag, env, and gag/pol transcripts were abundant in mammary tissue from Stat5a-null mice and less abundant in wild-type littermates (Fig. 1C). Hybridization of the blot with probes specific for WAP and -casein demonstrated that -casein is expressed at similar levels in Stat5a-null mice and control littermates, whereas expression of WAP is reduced in the absence of Stat5a (14) (Fig. 1D). A tissue survey revealed that the presence of the transcripts was confined to mammary tissue (data not shown). No sequence similarity was detected between probes 1 and 2, and no matching sequence was found in GenBank.

View larger version (80K): [in this window] [in a new window]   FIG. 1.   Transcripts from the hitchhiker locus and MMTV proviruses. Total RNA was isolated from mammary tissue of Stat5a-null mice (lanes a and b) and wild-type littermates (lanes c and d) at day 13 of pregnancy (lanes a and c) and at parturition (lanes b and d). The membrane was hybridized with probe 2 located at the 3' end in exon F (A) and then hybridized with probe 1 located at the 5' end in exon F (B), with the sag probe located in the 3' LTR (C), and finally with oligonucleotides specific for WAP and -casein (-ca) RNA (D). (E) Positions of probes in the hitchhiker locus. Positions and coordinates of the probes are described in Materials and Methods. I, II, III, IV, and V are the major transcripts derived from the hitchhiker locus. Note that at parturition, WAP RNA is reduced in Stat5a-null mice as described elsewhere (14).

Developmental regulation. The activities of genes at different stages of mammary gland development provides insight into their transcriptional regulation and potentially into their function. Thus, we determined levels of the transcripts corresponding to probes 1 and 2 during pregnancy and lactation. The developmental patterns obtained with probes 1 and 2 were strikingly similar (Fig. 2). Expression in mammary tissue was low during the early stages of pregnancy but sharply increased around day 13 and further increased until day 18 of pregnancy (Fig. 2). Expression declined dramatically within 1 day after parturition.

View larger version (62K): [in this window] [in a new window]   FIG. 2.   Expression of RNAs from the hitchhiker locus at different stages of mammary development. Total RNA was isolated from mammary tissue of Stat5a-null (left) and wild-type littermates (right) from the virgin state (V), days 9, 13, 16 and 18 of pregnancy, and at parturition (L). (A) Hybridization with probe 1 located at the 5' end of exon F (Fig. 1 and 4). (C) Hybridization with probe 2 located at the 3' end in exon F (Fig. 1 and 4). (B and D) Hybridization with glyceraldehyde-3-phosphate dehydrogenase for normalization purposes. I, II, III, IV, and V are the major transcripts derived from the hitchhiker locus.

Chromosomal localization. Deletion of genes by homologous recombination results in major territorial changes, including the addition of a transcriptionally active neomycin gene cassette in the targeted locus. To exclude that the novel transcripts were derived from the targeted Stat5a locus, we determined the chromosomal localization of the genes represented by probes 1 and 2. We screened a BAC library with probe 2 and identified two clones. Southern blot analyses showed that the two BAC clones also hybridized to probe 1, suggesting for the first time that probes 1 and 2 were part of one genetic locus. We used one of the BAC clones in a FISH analysis and determined that it hybridized to chromosome 11E (Fig. 3). This region contains a number of genes, including Grb2 and the MTV-3 provirus. The MTV-3 provirus is present in several mouse strains, such as GR and 129, but not in C57/BL6. Since the Stat5 locus resides on chromosome 11E60, approximately 10 cM from the FISH signal, it is unlikely that the new transcripts are the result of the Stat5a deletion.

View larger version (111K): [in this window] [in a new window]   FIG. 3.   Chromosomal localization of the hitchhiker locus at the telomere of chromosome 11. (A) Hybridization of BAC clone DNA containing the hitchhiker locus to metaphase chromosomes derived from mouse embryo fibroblast cells resulted in specific labeling (arrows) of the distal end of a medium-sized chromosome which was believed to be chromosome 11 on the basis of DAPI staining. (B) The second hybridization of metaphase chromosomes with both the BAC clone DNA containing the hitchhiker locus (specific signals shown by arrows) and a probe which is specific for the centromeric region of chromosome 11 resulted in the specific labeling of the centromere and the telomeric region of chromosome 11.

Transcripts. In an attempt to define the nucleotide sequences of the novel transcripts and to deduce the amino acid sequence, cDNA libraries were generated from mammary tissue of Stat5a-null mice. Five cDNA clones were isolated with probe 2, and two cDNA clones were isolated with probe 1. Sequence analyses revealed that probes 1 and 2 were derived from transcripts with alternative splice sites, alternative exons, and different polyadenylation sites (Fig. 4A). Based on the sequence of the seven cDNA clones, we predicted the existence of at least seven different transcripts (Table 1).

View larger version (18K): [in this window] [in a new window]   FIG. 4.   Structure of the hitchhiker locus and its transcripts. (A) Open rectangles represent the 5' and 3' LTRs of MTV-3. Exons located 3' of MTV-3 are shown as solid boxes. B' and C' represent internal splice sites used in the transcript hh-1 (B). C' is also used by hh-2, and C is used by hh-6. EST probes 1 and 2 are indicated (the coordinates of the probes are shown in Materials and Methods). The asterisks in exon F point to the polyadenylation sites used in different transcripts. MMTV transcripts are presented above the genomic structure, and hitchhiker transcripts are shown below. Dashed lines represent sequences removed from the primary transcripts by splicing. (B) Structures of the cDNAs isolated. hh-6 corresponds to an EST from the soars mammary gland NbMMG library. (C) Transcripts that contain sequences from the env gene and the hitchhiker locus. RT-PCR analyses were performed with a primer in exon B followed by nested PCRs (see Materials and Methods). This confirmed the existence of RNAs, which probably start in the 5' LTR and continue through the MTV-3 genome and the hitchhiker locus, followed by various splicing patterns. These transcripts would account for the Northern signals that are larger than 4 kb and are specifically detected with hitchhiker probes.

Genomic structure. Two BAC clones from a 129SvJ library were isolated and sequenced, and the genetic map of the locus was determined (Fig. 4A; accession no. AF120673). Based on the sequence of 13,000 nucleotides (nt) from the genomic region and sequences of the seven cDNA clones and seven additional ESTs retrieved from GenBank, we propose the structure depicted in Fig. 4. The locus consists of the MTV-3 provirus and at least six additional exons, two of which (exons B and C) harbor additional functional internal splice sites, as well as four distinct polyadenylation sites within exon F and one in exon G (Fig. 4A). Exon A coincides with the first 250 nt of the MMTV transcript which originates within the 5' MMTV-LTR and proceeds into the MMTV splice site located 130 nt downstream of the 5' LTR. Transcript hh-2 contains exon A' which contains the transcribed sequences from the 5' or 3' LTR. Since exon A' is followed by exon B which is linked directly to the 3' LTR, it is likely that this transcript originates in the 3' LTR. Using 5'RACE, we determined that the start sites of hh-1 and hh-2 are located in the 5' LTR and 3' LTR, respectively (not shown). Exon B is used in its entirety in clone hh-2 (Fig. 4). An alternative splice site inside exon B is used in clone hh-1. Both clone hh-1 and clone hh-2 contain the 3' part of exon C and the entire exon D. While clone hh-1 splices directly into exon F, clone hh-2 picks up exon E before it splices into exon F (Fig. 4). The 3' ends of clones hh-1 and hh-2 coincide at the same polyadenylation site within exon F. However, based on the isolated cDNA clones, there are at least three additional polyadenylation sites within exon F, and use of the 3'-most one results in a transcript of approximately 4 kb (Fig. 4). Exon G has been identified as a match of the genomic sequence with an EST from mammary tissue (see below). All splice sites used adhered to the consensus sequence (Table 2).

ESTs from GenBank. Searching GenBank with the seven cDNAs isolated in our screen resulted in two kinds of matches. Several hundred ESTs matched a 150-bp sequence within exon F. This sequence represents a middle repetitive element. More informative was the match to six ESTs from a normalized cDNA library obtained from virgin mammary tissue of the C57BL/6 strain. In addition to confirming the genomic structure, it extended the number of splice sites and exons used. Of particular interest is clone hh-6 (originally accession no. AI159140; updated to AF118847), which has its 5' end within exon B. At the end of exon B it splices into the 5' end of exon C. In contrast, both hh-1 and hh-2 use a splice site within exon C (Fig. 4). Thus, hh-6 represents a splice variant not detected in our clones. Another EST with genomic sequences 3' of the known exon F was detected and is now defined as exon G (Fig. 4). Given the multiplicity of exons as well as intra or exon splice sites, we predict further transcripts.

Mouse strains 129SvJ and C57BL/6. The 129 but not the C57BL/6 strain contains the MTV-3 provirus (7). Given that the EST clones that match the hitchhiker transcripts were derived from a mammary cDNA library from C57BL/6 mice, it is possible that transcription at this locus is independent of the MMTV provirus. Alternatively, the source of the library may not have been C57BL/6 but a strain that contains MTV-3. We used RT-PCR assays to establish whether C57BL/6 mice express transcripts from the hitchhiker locus. The RT primers matched sequences in exon C' and the second primer was located in the transcribed part of the MMTV LTR. Using this primer set, we expected two fragments, one corresponding to the hh-1 transcript and the second to corresponding hh-2. RNA from mammary tissue of postpartum Stat5a-null, 129SvJ, and C57BL/6 mice was analyzed. The expected fragments were detected in tissue from the Stat5a-null and 129SvJ mice but not the C57BL/6 mice (Fig. 5). As a control we amplified Stat5b, which can be detected in mammary tissue from all three mice (Fig. 5). This finding strongly suggests that the mammary gland cDNA library (library 403; European IMAGE Consortium) was not derived from a C57BL/6 mouse. In the process of sequencing the hitchhiker locus from the two BAC clones from the 129SvJ genomic library, it became clear that they represented two different alleles, one containing and one lacking the MTV-3 provirus. We sequenced 20,000 bp from both alleles and were able to determine the point of proviral insertion (GenBank accession no. AF120673). Importantly, our results from the BAC clone demonstrate that the genomic library in question is not pure 129. 

View larger version (82K): [in this window] [in a new window]   FIG. 5.   Absence of MTV-3-hitchhiker fusion transcripts in C57BL/6 mice. RT-PCR assays were performed as described in Materials and Methods. Mammary tissue from postpartum Stat5a-null (a), Stat5a +/ (b), and C57BL/6 (c) mice was analyzed for the presence of hitchhiker (A) and Stat5b (B) transcripts. The two bands in panel A represent the hh-1 and hh-2 transcripts.

Influence of Stat5a on transcription from MMTV LTRs. Recently Qin and coworkers reported that transcription from the MMTV LTR in mammary tissue is dependent on a Stat5a binding site within the LTR and the presence of Stat5a (20). Since the hitchhiker cDNAs were isolated from a library enriched for transcripts preferentially expressed in the absence of Stat5a, we evaluated this aspect. Specifically, we determined the transcriptional activity of the MTV-3 hitchhiker locus and that of other endogenous MMTVs in mammary tissue from Stat5a-null mice, control 129 mice, and C57BL/6 mice. Analysis with probe 2 demonstrated the expected RNA fragments in the Stat5a-null mice (Fig. 6A). Fragments of the same size and similar intensity were detected in mammary RNA from lactating 129 mice. As expected, no signal was detected in C57BL/6 mice. Since the transcripts were detected in Stat5a-null mice, activation of the novel transcripts within the MTV-3 locus does not depend on the presence of Stat5a. We further analyzed expression of all MMTV proviruses by hybridizing the RNA with a sag probe, which detects all proviral transcripts. Levels of expression of sag, env, and gag/pol were similar in Stat5a-null and control 129 mice (Fig. 6B) but much lower in C57BL/6 mice.

View larger version (73K): [in this window] [in a new window]   FIG. 6.   Transcription from the MMTV 5' LTR in the absence of Stat5a. RNA from mammary tissue from postpartum Stat5a-null (lanes 1 to 3), Stat5a +/ (lane 4), C57BL/6 (lanes 5 to 7), and 129SvJ (lanes 8 to 10) mice was analyzed for hitchhiker transcripts by using probe 2 (A) and for MMTV transcripts by using a sag probe (B). Locations of the probes are shown in Fig. 1E. Probe 2, containing no MMTV sequences, detects the major hitchhiker transcripts III, IV, and V.

	DISCUSSION

Top Abstract Introduction Materials and Methods Results Discussion References

The MTV-3 provirus is located on chromosome 11 and contains a complete yet defective genome (9). Viral transcripts are detected, but the Env proteins are present at low amounts, suggesting a defect in translation or posttranslational processing (9). We have discovered that a large portion of transcripts originating in the 5' LTR of MTV-3 do not encode viral transcripts but splice into the juxtaposed mouse genome. In addition, transcripts originate in the 3' LTR of MTV-3 and proceed into 10 kb of flanking genomic sequence and pick up hitchhiking exons. These findings imply that the insertion of proviruses can have consequences in the evolution of the genome.

Hitchhiked exons in the MTV-3 locus. We detected two classes of transcripts, those that originate within the 5' LTR of MTV-3 and those that originate in the 3' LTR. Both classes proceed for at least 10 kb into the genomic flanking sequences. Transcripts originating in the 5' LTR used the native MMTV splice site, picked up at least six different exons, and utilized at least five different polyadenylation sites. Transcripts originating in the 3' LTR proceeded directly into exon B and essentially followed the same splice pattern as transcripts originating in the 5' LTR. All splice donor and acceptor sites shared appropriate consensus sequences. None of the transcripts exhibited any significant open reading frame, suggesting that they do not encode mature peptides. From Northern blot analyses, it appears that a large portion of the transcripts from the 5' LTR proceed into the hitchhiker locus and do not form mature MMTV transcripts. Since it is not possible to generate unique hybridization probes for MTV-3, we cannot determine accurately the percentage of transcripts that enter the hitchhiker locus. Similarly, a large number of transcripts originate from the 3' LTR. The majority of hitchhiker transcripts have sizes between approximately 1.5 and 4 kb, which can be accounted for by the MMTV and genomic exons identified over a stretch of 20 kb. However, additional RNA species were seen on Northern blots, and both large-scale genomic sequencing and EST data mining should provide more details on the full structure of this locus.

Evolutionary considerations. Acquired retroviruses can induce tumors through the deregulation of juxtaposed genes (17). In contrast to these selection events, no apparent selection occurs for proviral genomes that are passed on through the germ line. Transcription in the hitchhiker locus originates in both the 5' and 3' LTRs and proceeds past the 3' LTR for at least another 10 kb. The primary transcript contains at least six exons, and the combination of different splice donor and acceptor sites and several polyadenylation sites leads to a plethora of polyadenylated transcripts. This raises several questions about the influence of proviral genomes in the evolution of the genome. Several scenarios can be envisioned. First, transcription from the LTRs in conjunction with the generation of multiple polyadenylated RNAs is of no consequence, even over large time spans. Second, the observed transcription interferes with native genes in this region. Third, the transcripts serve as a pool for future genes. At this point there is no experimental evidence for the latter two hypotheses. We have to assume that the presence of hitchhiker transcripts is the result of (i) weak transcriptional stop sites in the 3' LTR and (ii) the presence of splice donor and acceptor sites and polyadenylation sites in the juxtaposed genomic territory.

Transcriptional regulation by Stat5. Stat5 was originally discovered by Wakao and coworkers and described as a mammary gland-specific factor that could activate transcription from the -casein gene promoter in HC-11 cells (25). It had been speculated that Stat5 could be the global activator of genes in mammary tissue. The two isoforms of Stat5, Stat5a and Stat5b, exhibit conservation of 96% and are activated by cytokines such as prolactin. Both Stat5a (14) and Stat5b (24) genes have been deleted from the mouse genome, and the observed physiological consequences are different. This is strong support that these molecules, despite their conservation, elicit different responses in vivo. In particular, Stat5a but not Stat5b is required for mammary development and function (14, 22). Although many genes that contain GAS sites can be activated by Stat5a or Stat5b in transfected tissue culture cells, no in vivo target gene other than the WAP gene (14) has been identified. Recently, Qin and coworkers reported the presence of a Stat5 binding site within the MMTV LTR and demonstrated that transcription from the LTRs of MTV-17 and MTV-3 is dependent on both Stat5a and Stat5b (20). They showed that transcription from the LTRs in mammary tissue, but not lymphoid tissue, of either Stat5a-null or Stat5b-null mice was reduced to less than 10% compared to control mice. These findings were important since they demonstrated for the first time that heterodimers formed between Stat5a and Stat5b were required for transcription of a specific promoter. However, we could not repeat these results in over 30 Stat5a-null mice. The amounts of transcripts derived from the MTV-3 locus and from all MMTV proviruses combined are the same in Stat5a-null and control 129 mice. Transcription from the MTV-3 locus may even be higher in the absence of Stat5a. We cloned and sequenced the MTV-3 locus from the 129 mice and could confirm the presence of GAS (Stat5 binding) sites in both LTRs (AF119341 and AF119342). At this point, we cannot explain these differences. Our finding that transcription from the MMTV LTRs is not affected, or is only marginally altered, by the presence of Stat5a fully agrees with our earlier studies of Stat5a-null mice (14). The presence of GAS site in a promoter, and its functionality in tissue culture cells, does not necessarily reflect its relevance in vivo.