Novel Testis-specific Protein-DNA Interactions Activate Transcription of the Mouse Protamine 2 Gene during Spermatogenesis*

The mouse protamines are expressed exclusively in postmeiotic male germ cells and are crucial for the compaction of chromatin during the late stages of spermatogenesis. The temporal expression of the two mouse protamines is transcriptionally regulated in the testis. Recent studies have demonstrated that ubiquitous and testis-specific proteins bind to the promoter of the mouse protamine 2 (mP2) gene. We have performed in vitro transcription and mobility shift assays to characterize the functional significance of the protein-DNA interactions within 180 base pairs upstream of the mP2transcription start site. Deletion and mutational analyses reveal two positive regulatory sequences for mP2 transcription at positions −59/−47 and −83/−72 of the mP2 promoter. The proximal element at −59/−47 binds to a novel testis-specific protein we name protamine-activating factor 1 (PAF-1). PAF-1 reaches high levels in round spermatids at the time of mP2transcription. Deletion of the −59/−47 sequence results in about a 3-fold reduction of mP2 transcription in vitro. Although the PAF-1 binding site (PAF-responsive element, PAF-RE), contains the sequence GTCA present in the cAMP-responsive element and is very similar to the estrogen-responsive element, mobility shift assays revealed that neither the cAMP-responsive element modulator nor the estrogen receptor is the protein(s) binding to PAF-RE. Competition mobility shift assays reveal that the second positive regulatory element at −83/−72 binds a Y-box-binding protein. Using in vitro transcription assays, a 5-fold decrease in mP2transcription is seen when both the PAF-RE and this Y-box are deleted. These data suggest that the testis-specific PAF-1 and a Y-box-binding protein are needed to activate mP2 transcription in postmeiotic male germ cells.

Spermatogenesis is a unique process of cellular differentiation in which diploid testicular stem cells differentiate into haploid spermatozoa (1,2). During meiosis and spermiogenesis, male germ cells undergo major morphological and biochem-ical changes. These changes are regulated by the stage-specific transcriptional and translational activation of a large number of genes under temporal and cell-specific regulation (3). In addition to numerous testis-specific genes, many testis-specific variants of somatic genes are also expressed during spermatogenesis (2,4,5).
Protamines are small, arginine-rich proteins expressed only during the postmeiotic stages of spermatogenesis (2,6). In the mouse, there are two protamines, mP1 and mP2, which differ in size and amino acid sequence and are transcribed from genetically linked single copy genes (6,7). The nuclear compaction of the sperm nucleus during spermiogenesis involves the sequential replacement of histones by transition proteins and then by protamines. These chromatin changes contribute to the cessation of transcription in the maturing spermatids. Since transcription of the two mouse protamine genes is temporally coordinated (8), it is likely that similar sequences in the 5Јflanking regions of the protamine genes participate in their regulation (9).
The promoters of the mP1 and mP2 genes have been used to direct transcription in transgenic mice. A 113-bp region of the mP1 promoter from Ϫ150 to Ϫ37 successfully directs spermatid-specific transcription (10,11). A number of ubiquitous and testis-specific proteins bind within this region (12), including a testis-specific nuclear protein (Tet-1), which binds to an 11-mer sequence at Ϫ64. Tet-1 has been proposed to activate the transcription of mP1 (13). Likewise, an 859-bp region of the mP2 gene immediately upstream of the transcription start site confers specific expression in round spermatids (14). Analyses of the mP2 promoter by in vitro transcription assays have identified a potential positive regulatory region from Ϫ170 to Ϫ82 (15). Mobility shift assays of the mP2 promoter region from Ϫ140 to Ϫ23 have detected the binding of ubiquitous and testis-specific proteins (16).
Recent studies analyzing protein-DNA interactions within the proximal mP2 promoter by DNase I footprinting and mobility shift assays have identified five protein-binding sites (17). Site 1 (Ϫ64/Ϫ48) contains the single core motif AGGTCA recognized by orphan nuclear receptors that are believed to be important for gonadal and brain development (18). Site 1 also contains a half-site of the cAMP-responsive element (CRE) that binds CRE-binding protein (CREB) or CRE modulator (CREM) (19) and a sequence similar to the estrogen receptor-responsive element (ERE) (20). Site 2 has a putative Y-box at Ϫ83/Ϫ72, which binds a family of sequence-specific DNA-binding proteins, one of them capable of activating germ cell-specific transcription (21). A testicular Y-box protein has also been shown to bind to a putative Y-box at Ϫ489/Ϫ478 of the mP2 promoter (22).
To characterize and determine the functional significance of the DNA-protein interactions at sites 1 and 2 of the mP2 promoter, in vitro transcription and mobility shift assays were performed. We find that both site 1 and site 2 are important to obtain maximal mP2 transcription in vitro and suggest that a novel orphan nuclear receptor and a Y-box-binding protein are essential to activate mP2 transcription during the late stages of spermatogenesis.

EXPERIMENTAL PROCEDURES
Construction of G-free Cassette Plasmids-The promoter-less pEcoRV(C 2 AT) plasmid, containing a G-free cassette, was derived from the plasmid p(C 2 AT)19 (23). DNA fragments representing various lengths of the mP2 promoter ( Fig. 2A) were amplified by the polymerase chain reaction (PCR). Each of the 5Ј-primers used in PCR has an EcoRI site at its 5Ј-end. Their sequences are 5Ј-CAG AAT TCA AAG CAA GAT GAG TAA C-3Ј (for Pr168); 5Ј-CCG AAT TCG CCC TCA CAG AGG GGA C-3Ј (for Pr115); and 5Ј-CAG AAT TCT TTA CCT TTA TAT ATG AGC C-3Ј (for Pr45). The 3Ј-primer (5Ј-TCC ACG TGG TGA TGA TGG TCT GG-3Ј) has an EcoRV site at its 5Ј-end. The PCR fragments were subcloned into the EcoRI and EcoRV sites upstream of the G-free cassette in pEcoRV(C 2 AT). The G-free cassettes together with their upstream mP2 promoter fragments were excised by digestion with EcoRI and SmaI and subcloned into the EcoRI and EcoRV sites of pBluescript SK(ϩ) plasmids to generate plasmids Pr168, Pr115, and Pr45. The Pr67 plasmid was prepared by amplifying Pr115 with a pBluescript primer encompassing the unique EcoRI site (5Ј-AAT TCC TGC AGC CCG GGG GA-3Ј) and an mP2 internal primer (5Ј-GGG CCG ACA GGT CAC AGT GG-3Ј) starting at Ϫ67 of the mP2 promoter. The open ends of the PCR fragments were then phosphorylated and ligated.
To construct Pr370, an mP2 promoter fragment from Ϫ870 to ϩ11 was amplified by PCR using the 5Ј-primer (5Ј-GGA ATT CTG GCC TGG CAT GTG C-3Ј) and the 3Ј-primer (5Ј-TCC CCC GGG GGT GGT GAT GAT GGT-3Ј). Subcloning of this DNA fragment into the EcoRI and EcoRV sites upstream of the G-free cassette in pEcoRV(C 2 AT) generated Pr870. The mP2 promoter from Ϫ370 to ϩ11 and the G-free cassette were then excised from Pr870 and subcloned into the HindIII and XbaI sites of pBluescript KS(ϩ). To remove the Y-boxes and the protamine-activating factor 1 (PAF-1)-responsive element (PAF-RE) in Pr370, two specific PCR primers were synthesized for each construct. Each pair of the primers has matched restriction enzyme sites at their 5Ј-ends. The primer sets used for the mutations were 5Ј-CGG GAT CCA AAG TTA CTC ATC TTG C-3Ј and 5Ј-CGG GAT CCG TTC CAG TCC TGC AAA CC-3Ј (for Pr370⌬Y1); 5Ј-GGA ATT CGG ATC CCA CCC TGC CC-3Ј and 5Ј-GGA ATT CAT GGG GTG GGC CGA CAG G-3Ј (for Pr370⌬Y2); and 5Ј-GAA GGC CTC ACG GTT TAC CTT TAT ATA TG-3Ј and 5Ј-GAA GGC CTT GCG TCG GCC CAC CCC-3Ј (for Pr370⌬RE). Using these primer sets, the Pr370 construct was amplified by PCR, and the open ends of the PCR fragments obtained were digested with the appropriate restriction enzymes and ligated. Double mutant constructs were similarly prepared by PCR using a Pr370 plasmid carrying the first mutation as DNA template. The sequences of all of the mP2 promoter constructs used for in vitro transcription assays were confirmed by DNA sequencing.
Purification of Germ Cells-Populations of germ cells were fractionated by sedimentation through a linear gradient of 2-4% bovine serum albumin at unit gravity in a Staput chamber as described previously (24).
Preparation of Nuclear Extracts-Mouse liver and testis nuclear extracts used for in vitro transcription and mobility shift assays were prepared as described previously (25,26). Nuclear protein extracts from purified germ cells were prepared as follows. Mouse spermatogenic cells were isolated (24) and washed twice in ice-cold phosphate-buffered saline. The cells were then resuspended in buffer A (10 mM HEPES, pH 7.9, 1.5 mM MgCl 2 , 10 mM KCl, 0.5 mM DTT, 0.5 mM phenylmethylsulfonyl fluoride) containing 0.05% Nonidet P-40 and homogenized briefly on ice with a Kontes pellet pestle. Protease inhibitors (1 mM benzamidine, 1% aprotinin, 1 g/ml pepstatin, and 1.1 g/ml leupeptin) were added to buffer A to minimize protein degradation. After incubation on ice for 15 min, the homogenate was centrifuged at 10,000 rpm in a microcentrifuge at 4°C for 10 min. The pellet was washed twice in buffer A and resuspended in buffer C (20 mM HEPES, pH 7.9, 0.42 M NaCl, 1.5 mM MgCl 2 , 0.2 mM EDTA, 0.5 mM DTT, 6.25% glycerol) containing protease inhibitors. After incubation on ice for 40 min, the nuclear suspension was centrifuged at 14,000 rpm in a microcentrifuge at 4°C for 15 min. The supernatant was collected and diluted with an equal volume of buffer D (20 mM HEPES, pH 7.9, 50 mM KCl, 0.2 mM EDTA, 0.5 mM DTT, 20% glycerol). The protein concentrations determined by Bio-Rad protein assays ranged from 1 to 5 mg/ml.
In Vitro Transcription Analyses of the mP2 Promoter Constructs-In vitro transcription was performed as described previously (23,27 Mobility Shift Assays-Double-stranded oligonucleotides were radiolabeled with Klenow polymerase, purified in 15% nondenaturing polyacrylamide gels, and recovered by overnight elution in 0.5 M ammonium acetate, 0.1% SDS, and 1 mM EDTA, pH 8.0. The radiolabeled oligonucleotides were then extracted once with phenol/chloroform, precipitated with ethanol, and dissolved in Tris-EDTA, pH 7.5. Labeled DNA probes (2-4 ϫ 10 4 cpm) were incubated with 5-10 g of protein extracts on ice for 20 min in a binding buffer containing 12 mM HEPES, pH 7.9, 60 mM KCl, 5 mM MgCl 2 , 0.2 mM EDTA, 0.5 mM DTT, 0.3 mM phenylmethylsulfonyl fluoride, 1-3 g of poly(dI-dC)⅐poly(dI-dC), and 12% glycerol. After incubation, the samples were resolved on 4% native polyacrylamide gels in 0.5 ϫ TBE (45 mM Tris borate, 1 mM EDTA, pH 8.5), and the protein-DNA complexes were detected by autoradiography. For competition mobility shift assays, protein extracts were preincubated with unlabeled specific or nonspecific oligonucleotides for 15 min before the addition of labeled DNA probes.
Isolation and Analysis of DNA-Protein Complexes-Mobility shift assays were performed as described above using radiolabeled DNA probes and liver or testis nuclear extracts. After the DNA-protein complexes were visualized in the wet gel by autoradiography, gel slices containing the complexes were excised from the polyacrylamide gels and the proteins were eluted at 37°C for 4 h in a buffer containing 20 mM HEPES, pH 7.6, 1 mM EDTA, 100 mM NaCl, 2 mM DTT, 20 g/ml bovine serum albumin, 0.1 mM phenylmethylsulfonyl fluoride, and 1% Triton X-100. The eluted proteins were precipitated with ice-cold 50% acetone, washed once with ice-cold 80% acetone, and resuspended in SDS-loading buffer. After being heated at 95°C for 5 min, the proteins were resolved on 10% SDS-polyacrylamide gels and electroblotted onto nitrocellulose membranes using the Bio-Rad Semi-Dry Trans-Blot SD system (Bio-Rad). The membranes were blocked at room temperature for 1 h with 5% nonfat dry milk in Tris-buffered saline (TBS) containing 20 mM Tris, pH 7.5, and 0.5 M NaCl. After several washes with TBST (TBS containing 0.05% Tween 20), the membranes were incubated for 1 h at room temperature in TBST, 1% bovine serum albumin with polyclonal antibodies to the Xenopus germ cell-specific Y-box-binding protein at a dilution of 1:20,000 (28). The filters were then washed with TBST and incubated for another hour at room temperature with horseradish peroxidase-conjugated protein A (Amersham) in TBST, 1% bovine serum albumin at a 1:8,000 dilution. After four final washes with TBST at room temperature, the Y-box proteins binding to the antibodies were detected by the ECL system (Amersham).

Deletion Analysis of the Mouse Protamine 2 Promoter-
DNase I footprinting analyses of the mP2 promoter from Ϫ370 to ϩ4 revealed five protein-binding sites (17). Nuclear proteins prepared from mouse liver and testis bind to sites 1, 2, and 4; whereas nuclear proteins from liver but not testis bind to site 3, and testis but not liver nuclear proteins bind to site 5. We denote the three protein-binding sites closest to the start of transcription (sites 1-3) as BS1 to BS3 (Fig. 1). BS1 contains the sequence AGGTCA recognized by orphan nuclear receptors (18), and the sequence GTCA inside this motif is also a half-site of the CRE, the binding site of CREB or CREM (19). BS1 also has an element at Ϫ59 to Ϫ47 with 11 of 13 nucleotides present in the consensus sequence of the ERE (20). By sequence homology, there are two putative Y-boxes present at Ϫ148 and Ϫ83 (BS2) in the mP2 proximal promoter. Both of them have 9 out of 12 nucleotides identical to the Y-box consensus sequence of Xenopus germ cell promoters (21).
To define in the proximal promoter of mP2 the importance of specific protein-DNA interactions on transcription, a series of 5Ј-deletion constructs containing various lengths of the mP2 promoter were subcloned upstream of a 400-bp G-free cassette ( Fig. 2A). Using these constructs with crude testis nuclear extracts, in vitro transcription assays generate transcripts of 411 nucleotides (Fig. 3A). A 331-nucleotide transcript from a construct containing 45 bp of the mP2 promoter subcloned upstream of a 320-bp G-free cassette was used as a control for each of the deletion constructs (Fig. 3A). Transcription of each altered promoter was expressed as a percentage of the activity of the Pr168 construct (Fig. 3B). Removal of the distal Y-box at Ϫ148 in the Pr115 construct did not significantly decrease transcription. However, when both of the Y-boxes were deleted in Pr67, transcription was decreased to about 65% of the control. Removal of the two Y-boxes and the PAF-RE sequence from the mP2 promoter dramatically reduced transcription to about 18% of the control Pr168. The greatly reduced transcription of Pr45 relative to Pr168 indicates that the proteins binding at BS1 and BS2 are crucial for the activation of mP2 transcription.
Analysis of the Sequence Elements in the mP2 Promoter-To better define the importance of the protein-binding sites BS1 and BS2 in activating mP2 transcription, a series of mP2 promoter fragments were generated in which the sequence elements at Ϫ59/Ϫ47 containing the PAF-RE or the Y-boxes at Ϫ148/Ϫ137 and Ϫ83/Ϫ72 were replaced by unrelated sequences (see Fig. 2B). No change in transcription was seen when the upstream Y-box at Ϫ148 (Pr370⌬Y1) was removed, whereas deletion of the downstream Y-box at Ϫ83 (Pr370⌬Y2) resulted in about a 40% reduction in transcription (Fig. 4). When both of the Y-box sequences were replaced with unrelated sequences (Pr370⌬Y), the relative transcription was reduced to about 50% of the control transcription. These data are consistent with our findings with the deletion constructs demonstrating that the proximal Y-box at Ϫ83 is more important than the more distal Y-box at Ϫ148 for mP2 transcription (Fig.  3). Moreover, when the PAF-RE sequence was deleted (Pr370⌬RE), transcription was decreased to about 38%, indicating the importance of this sequence element for mP2 transcription. Transcription was further reduced to 23% of the control Pr370 construct when the two Y-boxes and the PAF-RE sequence element were removed (Pr370⌬Y-RE). These results, in agreement with the 5Ј-deletion experiment (Fig. 3), confirm that the proximal putative Y-box at Ϫ83 and the PAF-RE at Ϫ59 are needed for maximal mP2 transcription.
Liver and Testis Nuclear Proteins Bind to BS1-To characterize the protein factors that bind to BS1, mobility shift assays were performed with liver and testis nuclear extracts using a radiolabeled BS1 oligonucleotide as DNA probe. One DNAprotein complex (complex I) was detected with liver and testis nuclear extracts (Fig. 5, lane 2), and two additional DNAprotein complexes (complexes II and III) were seen with testis nuclear extracts (Fig. 5, lane 6). To determine the specificity of protein binding to the BS1 DNA probe, a 100-fold molar excess The nucleotide sequence of the upstream region of the mouse protamine 2 gene from nucleotides Ϫ210 to ϩ5 is shown. The transcription start site is indicated by a right arrow and is assigned ϩ1. The putative TATA sequence is boxed, and the protein-binding sites detected previously by DNase I footprinting analyses (17) are underlined and denoted as BS1, BS2, and BS3. The two putative Y-boxes examined in this study are overlined and are designated Y-box 1 (at Ϫ148) and Y-box 2 (at Ϫ83), respectively. of unlabeled specific or nonspecific DNA oligonucleotides was added to the binding reactions before the addition of the radiolabeled probe. Unlabeled BS1 oligonucleotides greatly diminished both the DNA-protein complex I seen with liver extracts (Fig. 5, lane 3) and the DNA-protein complexes I, II, and III seen with testis nuclear extracts (Fig. 5, lane 7). No reduction in liver or testicular protein binding was detected with a nonspecific competitor oligonucleotide containing binding site BS2 (Fig. 5, lanes 4 and 8) or binding site BS3 (Fig. 5, lanes 5 and 9).
Since neither BS2 nor BS3 has homology to BS1, these data confirm the specificity of protein binding to BS1.
The Testicular Nuclear Proteins Binding to BS1 Are Developmentally Regulated-To further characterize the nuclear proteins that interact with BS1, mobility shift assays were performed using a radiolabeled BS1 DNA probe and nuclear extracts prepared from the testes of mice at different ages. Complex I was detected in the testis nuclear extracts of 12-dayold or 17-day-old mice (Fig. 6, lanes 1 and 2), and complexes II and III were first seen with nuclear extracts prepared from the testes of 22-day-old and adult mice (Fig. 6, lanes 3 and 4). These data indicate that the complexes II and III become detectable at the same time that postmeiotic germ cells begin to differentiate. Since mP2 is expressed only postmeiotically, the mobility shift assays and in vitro transcription data suggest that specific nuclear proteins present in postmeiotic germ cells bind to the BS1 site at Ϫ64/Ϫ48 of the mP2 promoter and activate its transcription.
To confirm which testicular cells express the proteins that form complexes II and III with the BS1 DNA probe, mobility shift assays were performed with nuclear extracts prepared from enriched germ cell populations (Fig. 6). Complex III was the major DNA-protein complex seen in the extracts from meiotic pachytene spermatocytes, although small amounts of complexes I and II were detected (Fig. 6, lane 5). The amounts of complexes II and III greatly increased in round spermatids (Fig. 6, lane 6), and elongated spermatids contain large amounts of complex III and a smaller amount of complex II (Fig. 6, lane 7). When equal amounts of proteins were assayed, we see that liver contains higher amounts of complex I than the differentiating germ cells (compare lane 8 to lanes 5-7 in Fig.  6). Mobility shift assays with nuclear extracts prepared from other somatic tissues and cells including kidney, spleen, Sertoli cells, and NIH 3T3 cells only detect complex I (data not shown). These data suggest that complex I is ubiquitous, whereas complexes II and III are germ cell-specific. Moreover, the amounts of complexes II and III increase greatly in round spermatids, the cell type in which protamine 2 is transcribed.
The Testicular Nuclear Proteins Binding to BS1 Are neither the Estrogen Receptor nor the CREM Protein-Since BS1 contains a putative ERE and a half-site of CRE, we performed mobility shift assays to determine whether the estrogen receptor or CREM binds to the BS1 site. Radiolabeled BS1 probe and whole cell extracts prepared from Chinese hamster ovary (CHO) cells stably transfected with human estrogen receptor were used in mobility shift assays. Although a prominent DNAprotein complex was detected with the CHO extract (Fig. 7A,  6 -9). The protein-DNA complexes formed were resolved on 4% native polyacrylamide gels and detected by autoradiography. To confirm binding specificity, 100-fold molar excesses of unlabeled double-stranded oligonucleotides as indicated at the tops of the lanes were used as competitors in binding reactions. Ϫ, no added competitor. The sequences of the sense strands of BS1, BS2, and BS3 oligonucleotides are depicted in Fig. 1. The arrow at the left indicates the ubiquitous DNA-protein complex (complex I), whereas the arrows at the right denote testis-specific protein-DNA interactions (complexes II and III).
FIG. 6. The testis-specific proteins binding to the BS1 in the mP2 proximal promoter are germ cell-specific and developmentally regulated. Nuclear protein extracts prepared from the testes of mice at different ages (lanes 1-4), enriched spermatogenic cells (lanes 5-7), or liver (lane 8) were incubated with 32 P-labeled BS1 DNA, and the DNA-protein complexes were analyzed on native 4% polyacrylamide gels. Complex I is ubiquitous, whereas complexes II and III are testisspecific. PS, pachytene spermatocytes; RS, round spermatids; ES, elongated spermatids. lane 1), its mobility differed significantly from DNA-protein complexes I, II, or III (Fig. 7A, lanes 2 and 3). When supershift experiments were performed using an antibody raised against the first 21 amino acids of the estrogen receptor, no change in electrophoretic mobilities or band intensities were observed for the DNA-protein complexes, suggesting that estrogen receptor is not the source of the protein in the DNA-protein complexes (data not shown).
The transcriptional activator CREM is highly expressed in postmeiotic cells (29,30). It is a member of a group of transcriptional regulators called CREMs that mediate cAMP-dependent transcription. Recent studies of CREM-deficient mice have shown that CREM is essential for mP2 transcription (31,32). Since BS1 contains a half-site of CRE, we performed mobility shift experiments with purified CREM-1 proteins using BS1 as probe. CREM-1 and CREM are two different CREM isoforms produced from the CREM gene by alternative splicing and are expected to have similar electrophoretic mobilities in native polyacrylamide gels (33). Two major DNA-protein complexes formed between BS1 and CREM-1 (Fig. 7B, lane 2). However, these DNA-protein complexes have different mobilities from those detected with liver (Fig. 7B, lane 3) or testis (Fig. 7B, lane 4) nuclear extracts. Thus, although purified CREM-1 binds to BS1, the DNA-protein complexes I, II, and III appear to be formed by other proteins.
Characterization of the Nuclear Proteins Binding to BS2 of the mP2 Promoter-The BS2 protein-binding site identified by DNase I footprinting (17) contains a putative Y-box sequence at Ϫ83 (Fig. 1). In gel shift assays using a radiolabeled BS2 oligonucleotide and nuclear extracts prepared from mouse liver, three major DNA-protein complexes are detected (Fig. 8,  lane 2). The three DNA-protein complexes are diminished by the addition of a 100-fold molar excess of unlabeled BS2 oligonucleotides (Fig. 8, lane 4) but not by adding equivalent amounts of the nonspecific BS1 oligonucleotide as competitor (Fig. 8, lane 3). The addition of unlabeled BS3 oligonucleotide (Fig. 1) specifically abolishes the formation of the slowest mi-grating complex but not of the other two complexes (Fig. 8, lane 5). When mouse testis nuclear extracts are analyzed by mobility shift assays using the radiolabeled BS2 DNA probe, two major protein-DNA complexes are seen (Fig. 8, lane 6). The slower migrating DNA-protein complex detected with testis has a mobility similar to that of the middle complex seen with liver nuclear extracts. Both of the complexes can be successfully competed by unlabeled BS2 (Fig. 8, lane 8) but not with equivalent amounts of the BS1 or BS3 oligonucleotides (Fig. 8,  lanes 7 and 9).
Since the BS2 protein-binding site contains a putative Y-box element, we performed competition mobility shift assays to determine whether the Y-box sequence is essential for the binding of nuclear proteins to BS2. Although the three DNAprotein complexes detected with liver nuclear extracts were successfully competed by the unlabeled BS2 oligonucleotide (Fig. 9, compare lanes 1 and 2), only the complex with intermediate mobility was diminished by the addition of the specific Y-box competitor (Fig. 9, lane 3). No competition was seen with the Y-box mutant oligonucleotide (Fig. 9, lane 4) (26). These data indicate that the Y-box is essential for protein binding in this complex. The addition of the CTF/NF-I consensus sequence that contains a CAAT box diminished the slowest migrating complex detected with liver nuclear extracts (Fig. 9, lane 5).
Two DNA-protein complexes are seen when testis nuclear extracts were used in mobility shift assays with the BS2 probe (Fig. 9, lane 6). Both were successfully competed by unlabeled BS2 oligonucleotide (Fig. 9, lane 7). As seen with liver nuclear extracts, the slower migrating complex was diminished by the addition of the Y-box oligonucleotide (Fig. 9, lane 8), but not by the addition of the Y-box mutant competitor (Fig. 9, lane 9) or by the CTF/NF-I consensus sequence (Fig. 9, lane 10). These data suggest that the binding of Y-box proteins to the BS2 probe generates the slower migrating complex detected with testis nuclear extracts.
The Testicular Y-box Protein Binding to BS2 Is the Mouse Homologue of the Xenopus Germ Cell-specific Y-box Binding Protein p54/p56 -Y-box proteins have been reported to specifically interact with the Y-box sequences in various germ cellspecific promoters and serve as transcriptional activators (21). The Xenopus protein p54/p56 is a germ cell-specific Y-boxbinding protein that has been shown to activate the transcription of Xenopus heat shock protein (hsp70) (34). Two mouse homologues of p54/p56 are present in the cytoplasmic extracts of mouse testis. These 48/52-kDa mouse proteins (p48/p52) are detected in pachytene spermatocytes with a maximal level of expression in round spermatids (35). To identify the mouse testicular nuclear protein that forms the slower migrating DNA-protein complex with the Y-box sequence in BS2, the protein in the complex was analyzed by Western blot assays using polyclonal anti-p54/p56 antibodies. As previously reported for testicular cytoplasmic extracts (35), two proteins of about 54 and 56 kDa are seen with control purified Xenopus proteins (Fig. 10, lane 1). One protein of about 52 kDa was detected in testis nuclear extracts (Fig. 10, lane 3) but not in liver nuclear extracts (Fig. 10, lane 2). When the protein in the testicular DNA-protein complex was recovered and analyzed by Western blotting, a protein of about 52 kDa was seen (Fig. 10,  lane 4).

Developmental Changes in Electrophoretic Mobility of the Y-box Proteins Binding to the Y-box within BS2-Previous
studies have shown that the mouse homologues of the Xenopus germ cell-specific protein p54/p56 are maximally expressed in round spermatids (35). To investigate the pattern of expression of the Y-box protein binding to the BS2 site of the mP2 promoter, mobility shift assays were performed with radiolabeled BS2 oligonucleotides and nuclear extracts prepared from prepuberal and adult mouse testes. Two DNA-protein complexes (U and L) were detected with the testis extracts obtained from adult mice (Fig. 11A, lane 4) and from prepuberal mice at 12, 17, or 22 days of age (Fig. 11A, lanes 1-3). Interestingly, the slower migrating complex detected with 12-day testis nuclear extract appears to migrate faster than that observed with testis nuclear extracts from 22-day-old and adult mice (Fig. 11A,  compare lane 1 with lanes 3 and 4). DNA-protein complexes of both mobilities are seen with BS2 and nuclear extracts from the testes of 17-day old mice (Fig. 11A, lane 2).
To determine whether the faster migrating variant of the slower migrating complex detected in gel shift assays with testis nuclear extracts from 12-day-old mice requires a Y-box for binding, competition mobility shift assays were performed using radiolabeled BS2 probes. As shown in Fig. 11B, the slower migrating complex formed with nuclear proteins from the testes of 12-day old mice migrates faster than that from adult testes (Fig. 11B, lanes 1 and 4). Both of the complexes are abolished by the addition of the Y-box oligonucleotide (Fig. 11B,  lanes 2 and 5), but equivalent amounts of the Y-box mutant show no competition (Fig. 11B, lanes 3 and 6).

DISCUSSION
In this study we have sought to define the importance of protein-DNA interactions in the proximal promoter (Ϫ168 to Ϫ45) of the mP2 gene by performing in vitro transcription analyses and mobility shift assays. Previous in vitro transcription studies of the mP2 promoter using nuclear extracts prepared from mouse testis have demonstrated that a positive regulatory sequence element is present between Ϫ170 and Ϫ82 FIG. 9. Competition mobility shift assays of the Y-box at ؊83 in the mP2 promoter. DNA-protein binding reactions were performed on ice with 32 P-labeled BS2 and 5 g of liver or testis nuclear extracts in the absence (Ϫ) (lanes 1 and 6) or presence of double-stranded oligonucleotides as competitors (lanes 2-5 and 7-10). The BS2 sequence is depicted in Fig. 1 9) are underlined. The Y-box, but not the Y mutant oligonucleotide, has been shown to bind specifically to Y-box-binding proteins (17). All competitors were added at a 100-fold molar excess before the addition of the radiolabeled DNA. The arrows at the left of the gel indicate the three specific DNA-protein complexes detected with the BS2 DNA probe.
FIG. 10. Western blot analysis of the protein in the slower migrating DNA-protein complex formed by testis nuclear extracts and the BS2 oligonucleotide. Gel mobility shift assays were performed using mouse testis or liver nuclear extract and radiolabeled BS2 probe. The gel slices containing the slower migrating DNA-protein complexes were excised, and the proteins were eluted and resolved on a 10% SDS-polyacrylamide gel. After being transferred to nitrocellulose membrane, the proteins were analyzed by Western blotting using polyclonal anti-p54/p56 antibodies.  1-4). The arrows at the left of the gel indicate the major protein-DNA complexes detected with BS2 and adult testis nuclear extracts. The slower migrating and faster migrating DNAprotein complexes formed by the BS2 probe and Y-box-binding proteins are denoted by the arrowheads labeled U and L, respectively. B, competition mobility shift analyses were performed with 32 P-labeled BS2 oligonucleotides and testis nuclear extracts from 12-day-old (lanes 1-3) or adult mice (lanes 4 -6) to demonstrate the specificity of binding to the Y-box at Ϫ83 of the mP2 promoter. The Y-box and Y mutant competition oligonucleotides are the same as those described in Fig. 9. U and L indicate the slower migrating and faster migrating DNA-protein complexes, respectively. (15), a region that binds a nuclear protein found in adult testis (16). DNase I footprinting analyses of the promoter region from Ϫ370 to ϩ65 revealed five protein-binding sites (17). Mouse liver and testis nuclear proteins bind to site 1 (Ϫ64/Ϫ48), site 2 (Ϫ87/Ϫ67), and site 4 (Ϫ239/Ϫ210); whereas liver but not testis proteins bind to site 3 (Ϫ202/Ϫ175), and testis but not liver nuclear proteins bind to site 5 (Ϫ328/Ϫ311).
We have focused on sites 1 and 2 because these two sites show considerable homology to the consensus sequences of known DNA-binding proteins. The sequence AGGTCA in BS1 is present in the responsive element recognized by orphan nuclear receptors (18). Moreover, site 1 (BS1) contains the 3Ј-half of the palindromic consensus sequence of the CRE. CRE is recognized by the postmeiotic transcription activator CREM (19,29). Sequence analysis also revealed that BS1 contains an element at Ϫ59/Ϫ47 similar to the consensus sequence of ERE (20). Based on homology, BS2 also has a sequence element at Ϫ83 in which 9 of 12 nucleotides are identical to the consensus Y-box sequence (21). Y-box-binding proteins have been shown to bind to Xenopus and mouse germ cell-specific promoters and serve as transcriptional regulators (22,26,34).
To define the importance of these sequence elements on mP2 transcription, 5Ј deletion constructs were generated, and their transcriptional activities were tested (Fig. 3). From in vitro transcription using mouse testis nuclear extracts, we conclude that BS1 and BS2 are essential for transcription, since their removal or alteration leads to significant reduction in mP2 transcription.
Mobility shift assays using BS1 as DNA probe detected one ubiquitous and two testis-specific DNA-protein complexes. The predominant testis-specific complex (complex III) was first detected in the testes of 22-day-old mice. Its abundance greatly increases from pachytene spermatocytes to round spermatids and appears to be present only in postmeiotic germ cells. We have named the protein forming this complex PAF-1 and named the sequence to which it binds PAF-RE. The AGGTCA motif in BS1 is similar to the binding sites of a group of transcription factors known as orphan receptors (18). These orphan receptors recognize a responsive element containing the sequence AGGTCA either as direct repeats separated by several nucleotides (36) or as a single motif (37). Mammalian orphan receptors have been found to play an important role in gonadal and brain development (18,38,39).
Germ cell nuclear factor GCNF/RTR, a germ cell-specific orphan nuclear receptor expressed during the development of oocytes and spermatocytes (38), is a possible candidate of PAF-1 due to their similar expression patterns. Expression of mouse RTR starts at day 17 after birth and reaches a high level at day 22 when spermatids appear (40). It binds as a homodimer to a response element containing the direct repeat of the sequence TCAAGGTCA, a half-site of steroidogenic factor-1 (41), and as a monomer to the single core motif TCAAGGTCA (40). GCNF/RTR is predominantly expressed in round spermatids with little or no expression in somatic cell types such as Sertoli cells (38,42), and it can bind to a sequence at Ϫ228/ Ϫ213 of the mP2 promoter containing the core motif (40). However, although the expression pattern of PAF-1 coincides with that of GCNF/RTR, their binding sites differ because BS1 has GAC instead of TCA upstream of the AGGTCA motif. Mobility shift assays showed that the T and C nucleotides at the 5Ј-end of the binding site are essential for GCNF binding (38). These crucial differences in the sequences recognized by PAF-1 and GCNF suggest that they are not the same protein.
Tr2-11 is another murine orphan receptor that has an expression profile similar to that of PAF-1 (43). High levels of testis-specific expression of Tr2-11 begin at postnatal day 18 and reach a maximum at day 22, the time at which postmeiotic germ cells become abundant. Tr2-11 is not detectable in other cell types including Sertoli cells (43), and it binds to a sequence containing AGGTCA. Considering the similarity in binding sites and germ cell-specific expression patterns, our data indicate that PAF-1 may be a novel member of this family of orphan nuclear receptors found in testes.
Despite the importance of CREM in regulating the transcription of testis-specific genes (31,32), our mobility shift and supershift assays using a BS1 probe and purified CREM-1 protein suggest that the testis-specific protein(s) that binds to the BS1 site in the mP2 promoter is not a CRE-binding protein.
The possibility of BS1 binding to other CREB/CREM isoforms is also unlikely, since the DNA-protein complexes formed between BS1 and liver or testis nuclear extracts could not be competed by the CRE consensus sequence (data not shown). Other studies have also reported testis-specific factors differing from CREM that interact with the CRE-like element in the proximal mP1 promoter and activate mP1 transcription (12,13). One testis-specific protein, Tet-1, has been proposed to differ from the family of CRE-binding factors based on differences in their protein binding sites (13). Since mP2 is not expressed in CREM mutant mice (31,32), our data suggest that CREM may affect mP2 transcription by binding to other CRE elements elsewhere in the mP2 gene or by controlling the expression of additional upstream transcription factor(s).
The protein-binding site BS1 also contains an element at Ϫ59/Ϫ47 that shares 11 of 13 nucleotides in the 13-bp ERE palindromic consensus sequence (5Ј-GGT CAC AGT GAC C-3Ј). Estrogen receptor is a ligand-activated transcription factor that belongs to the steroid receptor superfamily (44). Gene targeting experiments in male mice have demonstrated that the disruption of the estrogen receptor gene leads to a disorganized seminiferous epithelium and a decrease in the number of spermatogenic cells (45). The importance of 17-␤-estradiol and estrogen receptor in spermatogenesis is also suggested by the expression of aromatase in developing spermatids (46). However, we believe that the estrogen receptor is not PAF-1, since different DNA-protein complexes are formed between BS1 and protein extracts prepared from liver, testis, and CHO cells stably transfected with human estrogen receptor. Moreover, when liver, testis, and CHO cells are incubated with an oligonucleotide probe containing the ERE from the Xenopus vitellogenin A2 gene, we find DNA-protein complexes with different electrophoretic mobilities (data not shown). The DNA-protein complexes formed by the vitellogenin ERE and CHO whole cell extract, but not liver or testis nuclear extracts, can be supershifted by the addition of the anti-estrogen receptor antibody H222 (data not shown).
Eukaryotic Y-box-binding proteins represent a family of DNA-binding proteins that activate transcription of germ cellspecific genes by binding to the Y-box sequence (CTGATT-GG(C/T)(C/T)AA) (21,34). They also bind to RNA in a sequence-independent fashion (28,47). The mouse homologue of the Xenopus protein p48/p52 is only detected in germ cells with maximal expression in round spermatids (35) and has been shown to bind to a Y-box located at Ϫ489/Ϫ478 in the mP2 promoter (22). The BS2 site of the mP2 promoter contains a putative Y-box at Ϫ83. Mobility shift assays have detected proteins that specifically bind to this Y-box sequence. Western blot analysis of the testis DNA-protein complexes with polyclonal anti-p54/p56 antibodies has confirmed the binding of p48/p52 to BS2 (Fig. 10). It has been proposed that Y-box proteins bind to promoter elements and modify gene activity through the interaction with other tissue-specific trans-acting factors (48). The protein binding and the in vitro transcription data presented here provide the first direct evidence for this hypothesis, and we propose that p48/p52 facilitates the activation of mP2 transcription in the testis.
Interestingly, in mobility shift assays, the DNA-protein complexes formed between BS2 and the Y-box protein in the testes of 12-day-old mice migrate faster than those from the testes of 22-day-old and adult mice. The changes in mobility may be due to modification of the Y-box-binding proteins during development. Alternatively, since Y-box-binding proteins are capable of multimerization (49), they may form different multimers as the mouse approaches adulthood. Although low levels of p48/ p52 in nuclei cannot be detected by immunocytochemistry using anti-p54/p56 antibody, p48/p52 is first detected in the cytoplasm of early pachytene spermatocytes, its amount increases as the male germ cells develop, and it reaches a maximum in step 1-8 round spermatids (50). It appears that p48/p52 is modified when it is first detected in pachytene spermatocytes. As maximal p48/p52 production occurs with high levels of transcription during spermatogenesis, this modification may be crucial for its function as a transcriptional activator.
In summary, we have identified two sequence elements in the mP2 proximal promoter that are important for activating its transcription during spermatogenesis. The PAF-RE at Ϫ59/ Ϫ47 binds to a postmeiotically expressed germ cell-specific protein PAF-1 and induces a 3-fold increase in mP2 transcription. A second protein, p48/p52, binds to a Y-box sequence and further increases mP2 transcription. Together, these two DNAprotein interactions enhance mP2 transcription more than 5fold. We conclude that PAF-1 and p48/p52 act together to enhance transcription of the mP2 gene during the late stages of spermatogenesis.