Pem Homeobox Gene Regulatory Sequences That Direct Androgen-dependent Developmentally Regulated Gene Expression in Different Subregions of the Epididymis*

The epididymis is a useful model system to understand the mechanisms that govern region-specific gene expression, as many gene products display spatially restricted expression within this organ. However, surprisingly little is known about how this regulation is achieved. Here, we report regulatory sequences from thePem homeobox gene that drive expression in different subregions of the mouse epididymis in vivo. We found that the 0.3-kb 5′-flanking sequence (region I) from the Pemproximal promoter (Pem Pp) was sufficient to confer androgen-dependent and developmentally regulated expression in the caput region of the epididymis. Expression was restricted to the normal regions of expression of Pem in the caput (segments 2–4), but there was also aberrant expression in the corpus region. This corpus misexpression was extinguished when 0.6 kb of Pem Pp5′-flanking sequence was included in the transgene, indicating that one or more negative regulatory elements exist between 0.6 and 0.3 kb upstream of the Pem Pp start site (region II). When heterologous sequences were introduced upstream of the Pem Pp, expression was further restricted, mainly to caput segment 3, implying that the Pem Pp has segment-specific regulatory elements. To our knowledge, the regulatory regions we have identified are the shortest so far defined that dictate regionally localized expression in the epididymis in vivo. They may be useful for identifying the factors that regulate region-specific expression in the epididymis, for expressing and conditionally knocking out genes in different subregions of the epididymis, for treating male infertility, and for generating novel methods of male contraception.

The adult epididymis is a highly convoluted tubule divided into three different regions: the caput, the corpus, and the cauda. These regions are further organized into infra-regional segments divided by connective tissue called septula. This regionalization creates different luminal fluid environments that probably play a crucial role regulating sperm motility and fertilizing capacity within the caput and corpus, and storing mature spermatozoa in the cauda. The function and identity of each region are thought to be dictated by the particular constellation of proteins expressed in each region. For example, cystatin-related epididymal spermatogenic (Cres) 1 is restricted to the proximal end of the caput region, where it probably acts as a protease inhibitor that regulates proprotein processing (1,2). Nerve growth factor is preferentially expressed in the corpus region where it may act as a paracrine signaling molecule (3). Superoxide dismutase (E-SOD) is expressed in the cauda epididymis, where it probably serves as an antioxidant (4,5).
Despite the intricate and highly complex patterns of regionspecific gene expression in the epididymis that have been described (6,7), very little is known about the specific mechanisms that dictate them. Only a few promoters have been examined for expression in the epididymis in vivo (8 -11), and of these, only the murine epididymal retinoic acid-binding protein (mE-Rabp) promoter has been shown to drive faithful expression in the epididymis and not in any other tissues (10).
To define the regulatory sequences required for gene expression in the epididymis, we chose to use the Pem gene, which we showed previously is transcribed in the mouse and rat epididymis (12)(13)(14)(15)(16). Pem is the founding member of a recently defined homeobox gene subfamily that we named PEPP after the members that have been identified so far: Pem, Esx-1 (Spx-1), Psx-1, and Psx-2 (Gpbox) (17). All members of the PEPP family have related homeodomains (DNA-binding domains) interrupted by two introns at signature positions, are found on the X chromosome, and are preferentially expressed in reproductive tissues (14, 15, 18 -21).
Pem transcripts are derived from two promoters that are independently regulated in a tissue-specific manner (14,16). The distal promoter (Pem Pd) is preferentially expressed in placenta and ovary. The subject of this report, the proximal promoter (Pem Pp), is expressed exclusively in the male reproductive tissues epididymis and testis. Expression of the Pem Pp is restricted to somatic cells in the caput region of the epididymis and Sertoli cells in the testis (14,16,17,19,22). In both the mouse and rat epididymis, Pem transcripts are dramatically up-regulated between day 20 and 30 postpartum, suggesting that Pem Pp transcription is developmentally regulated in this organ (12,13). Experiments in gonadotropin-deficient rats have demonstrated that Pem expression in the epididymis requires androgen (13), as does Pem expression in both mouse and rat Sertoli cells (12,16).
In this study, we report the identification of Pem Pp 5Јflanking sequences sufficient to direct the normal pattern of expression of Pem in vivo. Analysis of transgenic mice containing different lengths of Pem Pp 5Ј-flanking sequence revealed that a 0.3-kb region immediately upstream of the Pem Pp transcription start site is sufficient to confer high levels of androgen-dependent and developmentally regulated expression in the caput region of the epididymis. We also identified another 0.3-kb regulatory region just upstream of this that restricts expression to the caput region by preventing expression in the corpus region. Lastly, introduction of heterologous sequences upstream of the Pem Pp generated an expression pattern that suggested that the Pem Pp contains elements that control expression in individual segments within the caput region. Our study is the first to indicate that multiple regulatory elements collaborate to generate regionally localized patterns of expression in the epididymis.

EXPERIMENTAL PROCEDURES
Animals-All experiments were performed in accordance with National Institutes of Health guidelines for care and use of animals. Castration was performed through the abdominal route to mice anesthesized with light isoflurane (as recommended by the M. D. Anderson Cancer Center animal care committee). Testosterone replacement was accomplished by subcutaneously inserting 0.5-cm-long testosterone capsules made from SILASTIC brand tubing (number 602-305, Dow Corning, Midland, MD) as described previously (23,24). Male mice 6 -8 weeks old were divided into three groups for sham, castrated, and castrated plus testosterone treatment for a duration of 7 days. The animals were then sacrificed, and their epididymides were collected.
Generation of Transgenic Mice-The Pem-121 transgene was made in three steps. First, a fragment containing the mouse Pem Pp transcription start site and 0.6-kb 5Ј-flanking sequence was subcloned into the EcoRV site of pBluescript (KS ϩ ). This fragment was generated by PCR from mouse genomic DNA using the primers mPem-L and mPem-M (Table I), which correspond to mouse Pem exons 2 and 6, respectively. Second, tetracycline-regulated (tet) promoter sequences excised from pUHC13-3 (25) using SalI were subcloned at the SalI site just upstream of the 0.6-kb Pem 5Ј-flanking sequences. Third, the bovine growth hormone 3Ј-untranslated and polyadenylation region (BGH pA), obtained by partial digestion of pGKneo c bpA (kindly provided by Dr. Martin Matzuk, Baylor College of Medicine) with BclI and NotI, was ligated downstream of Pem at the BamHI and NotI sites. Pem-213 was made by subcloning a 4.8-kb SalI/NotI fragment from Pem-121 (containing Pem sequences and the BGH pA region but without tet promoter sequences) into pGem-11Zf (Promega Inc.). Pem-214 was made by subcloning a 4.5-kb EcoRV/NotI fragment from Pem-213 (containing 0.3-kb Pem Pp 5Ј-flanking sequences, Pem coding exons and introns, and the BGH pA region) into KSϩ. Pem-212 was made by subcloning 0.6-kb Pem Pp 5Ј-flanking sequences into EGFP-hMT2A-N1 (kindly provided by Dr. Gilbert Cote) at the AseI and KpnI sites. This fragment, which encompasses Ϫ655 to Ϫ1 with respect to the Pem start ATG in exon 3, was generated by PCR from Pem-121 using the primers MDA-654 and -758 (Table I). EGFP-hMT2A-N1 is identical with EGFP-N1 (Clontech, Palo Alto, CA) except that it has hMT2A intron 1 inserted in the EGFP coding region at nt 898.
To generate transgenic mice, the transcription units from each of the four constructs were excised with restriction enzymes that uniquely cut at the 5Ј and 3Ј ends of the transcription units, gel-purified, and injected into the male pronuclei of c57/Bl6 mouse embryos.  (Table I).
Histological Examination and Immunohistochemical Analysis-Mouse testes were fixed in Bouin's solution (Sigma) overnight at room temperature. Samples were destained with 1 M saturated Li 2 CO 3 in 80% ethanol until the dye was no longer visible. After dehydration in increasing concentrations of ethanol, the tissues were embedded in paraffin wax, and 4-m tissue sections were cut and mounted on siliconized slides. The paraffin sections were dewaxed with xylene followed by absolute ethanol, hydrated in decreasing concentrations of ethanol, and rinsed with phosphate-buffered saline (PBS). The samples were treated with 3% H 2 O 2 in methanol for 15 min at room temperature to block endogenous peroxidase and then rinsed with PBS. A proteinblocking solution containing 3% normal goat serum was applied to the sections for at least 30 min at room temperature followed by overnight incubation at room temperature in a humidified chamber with antibody in 1ϫ PBS containing 1% normal goat serum. The samples were incubated with either polyclonal antisera against Pem (kindly provided by Dr. Carol MacLeod), Cres (kindly provided by Dr. Gail Cornwall; Ref. 1), Sgp2 (kindly provided by Michael Griswold; Ref. 26), or GFP (Clontech) at 1:500, 1:25,000, 1:200, and 1:5000 dilutions, respectively. The slides were then rinsed with PBS, incubated with peroxidase-labeled goat anti-rabbit immunoglobulin (Vectastain ABC kit, Vector Laboratories, Inc.) for 45 min at room temperature, rinsed again with PBS, incubated with ABC solution (Vectastain ABC kit) for 30 min, rinsed again with PBS, incubated with diaminobenzidine for 2-10 min, rinsed with distilled water, and counterstained with Harris hematoxylin (Sigma). The sections were then mounted with Permount and examined by brightfield microscopy. Low magnification images of epididymides were acquired using Sprint Scan 35 (Polaroid Inc.) and Path Scan enabler (Meyer Instruments, Houston, TX).
RNA Isolation and Ribonuclease Protection Analysis-Total tissue RNA was isolated as described previously by guanidinium isothiocynate lysis and centrifugation over a 5.7 M CsCl cushion (27). RNA was analyzed by RNase protection analysis, performed as described previously (12). Probe A, which contains 61 nt of Pem exon 6 and 250 nt of the BGH 3Ј-untranslated region, was generated by digesting Pem-121 with NdeI. Probe B, which contains 92 nt of Pem 3Ј-untranslated region, was generated by digesting a plasmid containing mouse Pem cDNA (Pem-27) with MslI. Probe C, which contains nt 482-693 (with respect to the ATG) of the GFP coding region, was amplified by PCR using the template Pem-212 and the primers MDA-981 and MDA-982 (Table I). The ␤-actin probe contains nt 135-169 of human ␤-actin exon 3 (GenBank TM accession number X00351). A set of RNA size markers generated from the century ladder template (Ambion, Inc.) was included in all gels.

Developmentally Regulated Expression in the Caput
Region of the Epididymis Directed by 0.6-kb Pem Pp 5Ј-Flanking Sequence-We showed previously that the Pem Pp is expressed in rodent epididymides and testes in a developmentally regulated manner (12)(13)(14)(15)(16). Pem expression in the mouse epididymis is restricted to the caput region (17,22). In the study reported here we set out to identify Pem Pp regulatory sequences that provide its region-specific and developmentally regulated expression pattern in the epididymis. Toward this goal, we generated a transgene containing 0.6-kb Pem Pp 5Ј-flanking sequences upstream of Pem coding exons and introns (Fig. 1A). Two independent transgenic mice lines were generated that contained this Pem-213 transgene: Pem-213. 6 and Pem-213. 10. RNase protection analysis with a transgene-specific probe (probe A) demonstrated that both of these transgenic mouse lines expressed the transgene in epididymis and testis but not any other tissues that we tested, including brain, kidney, liver, lung, muscle, thymus, tongue, skin, and vas deferens ( Fig. 1B and data not shown). Analysis with a probe that distinguishes between transgene and endogenous Pem transcripts (probe B) demonstrated that the level of transgene mRNA in epididymides from Pem-213.10 mice was ϳ50-fold higher than that of endogenous Pem mRNA (data not shown). Expression of endogenous Pem transcripts was not significantly altered in the transgenic mice (data not shown), indicating that the Pem protein expressed from the transgene did not autoregulate its own promoter. There was a dramatic increase in mRNA expression from the transgene between days 20 and 30 postpartum in epididymides from Pem-213.10 mice (Fig. 1C), which is the same expression pattern as that from the endogenous Pem gene (12). We conclude from these results that the 0.6-kb Pem Pp 5Ј-flanking sequence is sufficient to direct high levels of expression in the epididymis in a normal developmentally regulated manner.
To examine whether 0.6-kb Pem Pp 5Ј-flanking sequence is also sufficient to drive caput-specific expression, we performed immunohistochemical analysis on serial cross-sections of the epididymis by using a Pem polyclonal antisera shown previously to specifically recognize Pem protein (16,22,28). In agreement with Pitman et al. (22), we found that nontransgenic mice expressed Pem only in the caput region and that Pem was restricted to less than 5% of the somatic cells within the tubules in this region ( Fig. 2A). Pem-213.10 transgenic mice recapitulated this caput-specific expression pattern (Fig. 2C) and expressed Pem in a much higher percentage of the cells in this region than in littermate controls (Fig. 2B). The greater percentage of cells expressing Pem is consistent with our finding that Pem-213.10 mice expressed ϳ50-fold higher levels of transgene mRNA than they do endogenous Pem mRNA. As with the nontransgenic littermate controls, Pem-213 transgenic mice expressed Pem protein only in principal cells (not in basal or clear cells) (Fig. 2B) within the caput region. No Pem protein was observed in the corpus (Fig. 2C) or cauda (data not shown).
Within the caput, we found that Pem expression was restricted to segments 2-4 in Pem-213.10 mice (Fig. 2C). This same segment-specific expression pattern was also observed in Pem-213.6 mice and control littermates (data not shown). We judged the position of the segments in our sections based on 1) the natural septula between the segments, 2) known differences in cell size in different segments, and 3) by using Cres and Sgp2 as markers. In agreement with its reported expression pattern (1), we found that Cres expression was highest in segment 1 (the initial segment), at intermediate levels in segment 2 (the most proximal segment that expresses Pem), and at trace levels in more distal segments where it is probably deposited by luminal fluid (Fig. 2C). Sgp2, which is also a secreted protein, was present on spermatozoa and the luminal walls of somatic epididymal cells in caput segments 3-5 and the corpus (Fig. 2C), similar to its localization in the rat (26,29,30). In contrast to its rather ubiquitous presence as a secreted protein, Sgp2 was only present in the cytoplasm of some somatic cells in segments 3 and 4 (thus, it overlapped with Pem in these segments) and in a few cells in segment 5 (not observable at the magnification shown in Fig. 2C). We interpret this as indicating that Sgp2 is primarily secreted by cells in segments 3 and 4 and that its presence in more distal regions is mainly the result of luminal deposition. Based on this analysis using Sgp2 and Cres as markers, we conclude that the 0.6-kb Pem Pp 5Ј-flanking sequence directs Pem expression to its normal site of expression in segments 2-4 of the caput epididymis.
Expression in the Caput and Corpus Epididymis Directed by 0.3-kb Pem Pp 5Ј-Flanking Sequence-To further map the regulatory regions essential for epididymis expression, we generated the Pem-214 transgene, which contains only 0.3-kb Pem Pp 5Ј-flanking sequence (Fig. 3A). Analysis of two independent Pem-214 transgenic lines (Pem-214.8 and Pem-214.10) showed that like Pem-213 mice, Pem-214 mice expressed the transgene specifically in epididymis and testis ( Fig. 3B; other tissues tested but not shown are brain, kidney, lung, muscle, thymus, tongue, skin, and vas deferens). Both transgenic lines expressed much higher levels of transgene mRNA than endogenous Pem mRNA in the epididymis (60 -130-fold higher levels, based on RNase protection analysis with probe B; data not shown). The transgene displayed a normal pattern of developmentally regulated expression (Fig. 3C).
To determine whether this 0.3-kb Pem Pp 5Ј-flanking sequence also directs region-specific expression, we performed immunohistochemical analysis on epididymal sections from Pem-214 mice. We found that, like Pem-213 mice, Pem-214 mice expressed Pem protein in segments 2-4 (Fig. 2, D and E). Although we consistently observed expression in segments 2-4 in all Pem-214 mice that we checked, we noted that individual transgenic mice displayed differences in the number of cells that expressed Pem, particularly in segment 4 (compare Fig. 2,  D and E). Surprisingly, unlike Pem-213 mice, Pem-214 mice also expressed Pem in the corpus epididymis (Fig. 2, D-F). This strongly suggests that there are at least two regulatory regions in the Pem Pp 5Ј-flanking sequence that control region-specific expression in the epididymis (Fig. 3D) site (region II) contains one or more negative regulatory elements that prevent inappropriate Pem expression in the corpus.
Segment-specific Expression Pattern of Pem Altered by Introduction of Heterologous Sequences Upstream-To assess whether the context of Pem Pp influences its expression pattern, we determined the effect of introducing heterologous reg-ulatory sequences upstream of the Pem Pp. To do this, we generated and tested the Pem-121 transgene, which is identical to the Pem-213 transgene except that it contains tet promoter sequences upstream of the Pem Pp 5Ј-flanking sequences (Fig.  4A). The tet promoter is not active in the transgenic mice that we used for our study, as the mice we used do not contain the tTA transactivator protein necessary for expression from the tet promoter (25). Thus, in the present study, our only interest was to test the effect of introducing heterologous sequences on Pem Pp transcription. That there was no expression from the tet promoter in Pem-121 mice was confirmed by RNase protection analysis using a probe that distinguishes between tet promoter-and Pem Pp-derived transcripts (data not shown).
We found that, like Pem-213 mice, all three Pem-121 mice (Pem-121.2, Pem-121.5, and Pem-121.7) expressed Pem specifically in epididymis and testis ( Fig. 4B; other tissues tested but not shown are brain, eye, lung, muscle, prostate, thymus, tongue, skin, and vas deferens). Immunohistochemical analysis of epididymides from Pem-121 transgenic mice showed they had caput-specific Pem expression, but unlike in Pem-213 mice, the expression was restricted mainly to segment 3 (Fig. 2, G  and H). Only a few cells in segment 4 and no cells in segment 2 expressed Pem (Fig. 2, G and H). The absence of Pem expression in segment 2 was clearly evident by comparison with the expression pattern of Cres, which was expressed primarily in segments 1 and 2 (Fig. 2, G and H). This Pem expression pattern was clearly different from that of Pem-213, which was expressed at similar levels in segments 2-4 ( Fig. 2C; similar levels of expression in these three segments were observed even when the anti-Pem antibody was diluted to 1:1000 or 1:2000 to decrease the Pem signal; data not shown). The restriction of Pem protein expression in Pem-121 mice was due to the tet promoter sequences, not integration site-specific transcriptional effects, as we found the same expression pattern for both Pem-121 transgenic mice lines that expressed high levels of Pem protein (Pem-121.7 and Pem-121.5; Fig. 2, G and H,  respectively) and also for the line that expressed lower levels of Pem protein (Pem-121.2; data not shown). Our data suggest that tet promoter sequences may interfere with Pem Pp regulatory elements that drive expression in caput segments 2 and 4, but they have no effect on Pem Pp regulatory elements that direct expression to segment 3.

Androgen-dependent Expression in the Epididymis Directed by 0.3-kb Pem Pp 5Ј-Flanking Sequences-
We previously reported that Pem expression from the Pem Pp is androgen-dependent (12)(13)(14)16). To determine whether region I is sufficient to drive this regulation, we performed bilateral castration on Pem-214 mice. This led to a dramatic decrease in the levels of transgene expression when compared with that of the uncastrated (sham) control mice (Fig. 5). The same was observed when Pem-121.5 and Pem-121.7 mice were castrated (data not shown). Transgene mRNA expression was restored when the castrated Pem-214 mice were given a testosterone supplement (Fig. 5). These observations demonstrate that region I is sufficient to provide androgen-dependent expression in the epididymis.

Expression of a Heterologous Gene (GFP) Driven by Pem Pp 5Ј-Flanking Sequence in the Caput Region of the Epididymis-
Because it is possible that Pem exon or intron sequences contribute to the expression pattern of Pem Pp transcripts (e.g. they may harbor a transcriptional enhancer or post-transcriptional regulatory elements), we examined whether Pem Pp 5Ј-flanking sequences were sufficient to drive tissue-specific and region-specific expression of a heterologous gene. For this purpose we generated the Pem-212 transgene, which contains 0.6-kb Pem Pp 5Ј-flanking sequence upstream of the GFP reporter gene (Fig. 6A). The Pem-212 transgenic line that expressed the highest levels of GFP in the testis, Pem-212.9, was chosen for further analysis. Mice from this transgenic line expressed GFP mRNA specifically in testis and epididymis (Fig. 6B). Immunohistochemical analyses performed with an anti-GFP antibody demonstrated that GFP was expressed by cells in the caput region of the epididymis (Fig. 2I). As in Pem-213 mice, expression of the transgene in Pem-212 mice was restricted to segments 2-4 (data not shown). Because there were fewer GFP-positive cells in Pem-212 mice than there were Pem-positive cells in Pem-121, Pem-213, and Pem-214 mice, it is possible that Pem intron and/or exon sequences augment the proportion of expressing cells. Regardless, our results clearly demonstrate that 0.6 kb of Pem Pp 5Ј-flanking sequence is sufficient to drive the expression of a heterologous gene specifically in the testis and the caput region of the epididymis.

DISCUSSION
One of the most intriguing features of the epididymis is the strikingly regionalized expression pattern of genes and gene products along its length. These regional differences are thought to be essential for establishing unique microenviron-ments for sperm maturation, protection, and storage (6,(31)(32)(33)(34). Here, we have used the Pem Pp as a model system to identify regulatory elements that govern region-specific expression in the epididymis. Our studies in transgenic mice identified two regulatory regions that control Pem Pp gene expression in the epididymis (Table II). A 0.3-kb region immediately upstream of the transcription start site (region I) directed androgen-dependent and normal developmentally regulated expression in segments 2-4 of the caput (Fig. 3D). Region II, the 0.3-kb region immediately upstream of the region I, appears to contain one or more negative regulatory elements that prevent inappropriate expression in the corpus. To our knowledge, Pem Pp regions I and II are the shortest sequences so far defined that regulate gene expression in the epididymis in vivo.
Other epididymal promoters that have been characterized in transgenic mice are those for proenkephalin, glutathione peroxidase-5 (Gpx-5), cysteine-rich secretory protein (Crisp-1), Pax-2, and murine epididymal retinoic acid-binding protein (mE-Rabp). The first of these to be characterized was the proenkephalin promoter. A construct containing ϳ4 kb of proenkephalin 5Ј-flanking sequence was shown to be expressed in the caput region of the epididymis (11). This promoter construct was also expressed in several other organs, as expected given that proenkephalin is normally expressed in many tissues, including testis, prostate, vas deferens, seminal vesicles, uterus, and ovary. Another transgene construct shown to direct expression in the caput was one containing ϳ5 kb of 5Ј-flanking sequence from the Gpx-5 gene (8). Expression from this Gpx-5 promoter construct was highest in segment 4 of the epididymis, but like the proenkephalin promoter, the Gpx-5 promoter was expressed in several other tissues in transgenic mice, including testis and brain (8). Another promoter that is normally caputspecific, Crisp-1, was found to be expressed in the testis rather than the epididymis when expressed from a transgene containing 3.8-kb 5Ј-flanking sequence (8). Thus, the sequences required for expression of Crisp-1 in the epididymis have not yet been identified. The Pax-2 promoter, which is normally expressed in the developing urogenital system, was shown to be expressed in fetal epididymides when expressed from a transgene containing 8.5 kb of 5Ј-flanking sequence (9). This Pax-2 promoter construct was also widely expressed throughout all the epithelium in the urogenital tract. To our knowledge, the only regulatory sequences that have been identified that drive epididymis-specific expression are those from the mE-Rabp gene. Lareyre et al. (10) found that ϳ5 kb of mE-Rabp 5Јflanking sequence directed androgen-dependent caput-specific expression (in segments 2-5) in transgenic mice. At least some of the regulatory sequences responsible for mE-Rapb caputspecific expression are between 0.6 and 5 kb upstream of the transcription start site, as a transgene with 0.6-kb proximal mE-Rabp 5Ј-flanking sequences was not expressed in the epididymis (10).
In the study reported here, we found that Pem Pp region I (Fig. 3D) directed Pem expression in the caput region of the epididymis, suggesting that this 0.3-kb region has all the elements required for transcription in the caput (Fig. 2, D-F). Expression was restricted to segments 2-4, indicating that region I also has segment-specific regulatory elements. However, we found that the context of the Pem Pp determines which caput segments expresses Pem, suggesting that region I contains multiple regulatory elements that control transcription in individual segments. In particular, we found that introduction of transcriptionally inactive tet promoter sequences upstream of the Pem Pp extinguished Pem expression in segment 2 and drastically reduced the number of Pem-expressing cells in segment 4 (Fig. 2, G and H). It is not clear how the tet promoter  (Fig. l) was used, and the protected band was ϳ200 nt. sequences alter Pem Pp regulation, as the tet promoter sequences only contain a minimal promoter (transcription start site) with several bacterial tet repressor-binding sites upstream (25). One possibility is that the transcription start site recruits some components of the transcription apparatus that interact and interfere with segment 2-and 4-specific regulatory proteins bound to region I. This can be experimentally tested when the factors interacting with region I that dictate the regionally localized expression of the Pem pattern in the caput are identified.
We found that the regional specification of Pem expression within the epididymis was dictated not only by region I but also by region II (Fig. 3D). Deletion of region II permitted expression in the corpus (Fig. 2, D-F), suggesting that this 0.3-kb region harbors at least one negative regulatory element that silences expression in this region of the epididymis. We propose that this negative regulatory element in region II collaborates with positive regulatory elements in region I to dictate precise region-specific expression in the epididymis. Very little is known about either the intracellular or extracellular factors that dictate segment-specific expression in the epididymis (6,35). Although testis-derived growth factors (36,37) and spermatozoa-associated factors (38) are known to regulate gene expression in the epididymis, it is not clear whether any of these factors are responsible for directing region-specific expression. In the case of Pem, a candidate intracellular regulatory factor is the Ets transcription factor Pea3, which is expressed in the epididymis and has been shown to be capable of acting as either a transcriptional repressor or activator (35,39). Inspection of region II revealed two consensus Ets-binding sites at positions Ϫ429 to Ϫ434 and Ϫ387 to Ϫ377 (with respect to the Pem start ATG) that contain core sequences conserved between mouse and rat Pem (Fig. 3D) (14,40). Mutagenesis of these Ets binding followed by in vivo expression analysis will allow assessment of whether Pea3 or other Ets family members are involved in region I-mediated transcriptional repression.
Our study revealed that the Pem Pp sequences conferring androgen-dependent expression are in region I (Fig. 5). The expression of many other genes in the epididymis have been shown to be dependent on androgen (41)(42)(43)(44)(45). Many of these genes have androgen response elements (AREs) that permit them to be directly activated by androgen receptor (AR) (46 -50). There are at least two ARE consensus sites in region I, both of which are conserved in mouse and rat Pem (Fig. 3D), which suggests that Pem Pp expression may be directly regulated by androgen receptor. This is supported by the finding that mutation of either site prevents androgen-dependent reporter gene expression in transfected cell lines (40). We believe that ARE-2 mediates AR-dependent expression, whereas ARE-I may instead function as a general transcriptional element, as it overlaps with the Pem transcription initiation site and initiator (Inr) element (16).
Pem Pp region I also has sufficient regulatory sequences to recapitulate the normal developmentally regulated expression pattern of Pem ( Figs. 1 and 3). Interestingly, the dramatic induction of Pem transcripts between days 20 and 30 postpartum occurs when epididymal somatic cells undergo dramatic alterations in morphology. The principal cells become much more columnar, exhibit extensions of the Golgi apparatus, and have increased numbers of vesicles (51,52). The time frame between days 20 and 30 postpartum is also when there are increases in the levels of AR (53,54), testosterone, dihydrotestosterone (47,55), and the enzyme responsible for dihydrotestosterone production, 5-␣-reductase (56). Thus, the induction of Pem Pp transcription after day 20 could be the result of increased androgen and/or androgen responsiveness, which would also be consistent with our finding that region I is necessary for androgen-dependent transcription. It is also possible that other factors are responsible for the dramatic developmental induction of Pem Pp transcription, including germ cell-derived proteins.
In conclusion, we have for the first time identified relatively short promoter sequences that direct gene expression to the epididymis. In addition, we have defined sequences that provide androgen and developmentally regulated expression in a segment-specific manner. Our results suggest that this segment-specific expression is controlled by a complex set of regulatory elements upstream of the Pem Pp that collaborate to  6. Epididymis and testis-specific GFP mRNA expression conferred by 0.6-kb Pem Pp 5-flanking sequence. A, schematic diagram of the Pem-212 transgene. B, ribonuclease protection analysis of total cellular RNA (20 g) from Pem-212.9 adult mice tissues using probe C (the protected band was ϳ190 nt). direct region-specific expression in the caput and prevent inappropriate expression elsewhere. Thus, we propose that the Pem Pp will be a good model system to identify cis-and transacting factors that govern gene expression in the epididymis. Given that Pem is a member of the homeobox transcription factor family, these regulatory elements will probably serve to govern not only the pattern of Pem expression but will also indirectly regulate the expression of Pem target genes. The caput-specific promoter sequences that we identified could be used to selectively knock out, ectopically express, or overexpress foreign genes in different regions of the caput in vivo. Such studies may identify the function of gene products in the caput region of the epididymis and could also lead to novel approaches for male contraception and methods to correct infertility.