Regulation of ALF Promoter Activity in Xenopus Oocytes

Background In this report we evaluate the use of Xenopus laevis oocytes as a matched germ cell system for characterizing the organization and transcriptional activity of a germ cell-specific X. laevis promoter. Principal Findings The promoter from the ALF transcription factor gene was cloned from X. laevis genomic DNA using a PCR-based genomic walking approach. The endogenous ALF gene was characterized by RACE and RT-PCR for transcription start site usage, and by sodium bisulfite sequencing to determine its methylation status in somatic and oocyte tissues. Homology between the X. laevis ALF promoter sequence and those from human, chimpanzee, macaque, mouse, rat, cow, pig, horse, dog, chicken and X. tropicalis was relatively low, making it difficult to use such comparisons to identify putative regulatory elements. However, microinjected promoter constructs were very active in oocytes and the minimal promoter could be narrowed by PCR-mediated deletion to a region as short as 63 base pairs. Additional experiments using a series of site-specific promoter mutants identified two cis-elements within the 63 base pair minimal promoter that were critical for activity. Both elements (A and B) were specifically recognized by proteins present in crude oocyte extracts based on oligonucleotide competition assays. The activity of promoter constructs in oocytes and in transfected somatic Xenopus XLK-WG kidney epithelial cells was quite different, indicating that the two cell types are not functionally equivalent and are not interchangeable as assay systems. Conclusions Overall the results provide the first detailed characterization of the organization of a germ cell-specific Xenopus promoter and demonstrate the feasibility of using immature frog oocytes as an assay system for dissecting the biochemistry of germ cell gene regulation.


Introduction
The factors and mechanisms that control transcriptional regulation in spermatocytes and oocytes of higher organisms have not been as well characterized as those in somatic cells [1][2][3]. This is due in part to the fact that germ cells cannot be propagated in cell culture and because cell-free extracts from complex tissues such as the testis are composed of contaminating subpopulations of germ cells and somatic cells. As a result, studies on the mechanisms of mammalian germ cell gene expression have tended to rely on somatic cell culture systems-where germ cell genes should normally be off-or on cell-free extracts derived from mixed somatic and germ cell populations from whole tissue sources.
Despite these issues, many regulatory factors have been proposed as regulators of germ cell gene expression. For a few of these, including CREMt and alternative general transcription factors such as TRF2 and TAF105, gene knockouts have demonstrated specific effects on fertility [4][5][6][7][8][9]. In contrast, it has been more difficult to show that regulatory factors identified in transfected somatic cells or those identified by in vitro protein-DNA interactions have genuine physiological roles in regulating germ cell genes.
In this report we ask whether frog oocytes, used in the early studies of gene regulation to define core promoter elements such as the TATA box, would be useful for characterizing germ cell promoter architecture and regulation [10][11][12]. There are several advantages to this approach. First, immature oocytes (stages I-VI) from frogs are very large and promoter constructs can be tested for activity by direct injection into the oocytes themselves. Second, cell-free extracts can be made in sufficient quantities to allow biochemical studies. Third, the set of basal transcription factors responsible for promoter recognition in germ cells, including oocytes, consists of a physiologically appropriate set that includes TRF3, ALF, and several TAF variants [1,[13][14][15][16][17]. These factors are different from TFIID components used for promoter recognition and activation in somatic cells, making oocytes a natural environment for dissecting the mechanisms of germ cell gene regulation. Thus, the approach allows a germ cell-specific promoter to be matched with cells, in this case oocytes, where the endogenous gene would normally be on.
To evaluate the use of Xenopus oocytes as a germ cell transcription system we have used the ALF gene as the model. ALF is a paralog of the large (a/b) subunit of TFIIA [18,19] and it plays a TFIIA-like role in stabilizing TBP (TFIID) to TATA elements within promoter DNA [20]. Characterization of the ALF promoter in mouse resulted in the identification of a number of candidate transcription factors, including possible somatic repressors. These factors included the zinc-finger proteins SP1, SP3, CTCF and the winged helix transcription factor RFX1 [21]. Studies on the ALF gene in Xenopus have shown that its expression is similar to that observed in mice and is restricted to spermatocytes and oocytes [22][23][24][25]. The germ cell-specific expression pattern suggests that the gene has retained the same functional role across these species and that the regulatory mechanisms that control expression are conserved.
In this report we have isolated the promoter from the Xenopus ALF gene using a PCR-based genome walking approach and characterized it with respect to initiation site by both RT-PCR and 59-RACE analysis. Interestingly, the low level of sequence homology between ALF promoters identified in X. laevis, X. tropicalis, mouse, rat, human, chimpanzee, chicken, and others made it difficult to predict regulatory elements on the basis of sequence homology alone. Nevertheless, functional assays showed that the promoter is able to drive high level expression of a downstream reporter in microinjected oocytes. Fine-scale deletion and mutational analysis resulted in the definition of a very small (63 bp) promoter, making it one of the smallest active germ cell promoter sequences identified thus far. Additional promoter constructs were used to show that the promoter consists of two discrete elements, A and B, both of which were necessary for activity. Finally, mobility shift assays using oocyte-derived extracts revealed multiple complexes which interacted with the A and B elements. Overall, the results showed that frog oocytes provide an effective system to study core promoter architecture and regulatory factor interactions for germ cell specific genes.

Materials and Methods
Isolation of the X. laevis ALF promoter The X. laevis ALF promoter was isolated with the Promoter-Finder system (Clontech). In brief, genomic DNA was prepared from liver and digested with blunt-end six-cutters HincII, PvuII, EcoRV, or StuI. After ligation to adaptor primers the resulting fragments were used as templates for PCR reactions with an adaptor-specific primer and a downstream gene specific primer (GSP1; 59-GGGTTAGCCGAATGGGCCATGA-39) derived from the Xenopus ALF cDNA [15]. After nested reamplification reactions with a second gene-specific primer (GSP2; 59-GCCTAACCGGAAGTTGGAACCA-39), PCR products were cloned into the pGEM T-easy vector (Promega) and sequenced. All of the products had perfect sequence identity to an overlapping section from the 59-end of the ALF cDNA.
For the M1A series of constructs, mutations were introduced into the parent 85 bp wild type promoter (ALF85). For M1A, the first 6 bp from the 59 end were converted to 59-CTGCGC-39; for M2A, 5 bp from 7 to 11 were mutated to 59-GTTTT-39; for M3A, 11 bp from 12 to 22 were mutated to 59-CCGCGGCAGTG-CAGTCG-39(the first 6 bp of this sequence is a SacII restriction site); for M4A, 10 bp from 23 to 32 were mutated to 59-CCGCGGCTTGCGGGTT-39 and for M5A, the 8 bp from 33 to 40 were mutated to 59-CCGCGGGCGCAATG-39. Deletion constructs D1A, D2A and D3A were generated by deleting 11 bp, 21 bp and 31 bp from the 39 end of WTA respectively.
The M5B series of constructs were based on the ALF85 construct to which an additional 11 base pairs of endogenous sequence had been added at the 39 end (ALF85+). In addition, these constructs all contained an EcoR1 site within the mutated region. For M5B, 8 bp from 33 to 40 were mutated to (59-GAATTCTG-39); for M6B, 12 bp from 41 to 52 were mutated to ''TGGTTGAATTCT''; for M7B, 12 bp from 53 to 64 were mutated to 59-TGTGAATTCGTG-39; for M8B, 12 bp from 65 to 76 were mutated to (59-TTGGAATTCGGG-39); for M9B, 9 bp from 77 to 85 were mutated to (59-GTGAATTCG-39); for M10B, 11 bp from 86 to 96 were mutated to (59-GAATTCTGTTG-39). The AM1-AM10 and BM1-BM6 series of mutant constructs were prepared using specific PCR primers, and the exact sequences are shown in the relevant figure.

Bisulfite methylation analysis
Genomic DNA was isolated from Xenopus laevis oocytes and liver using the Wizard SV Genomic DNA Purification System (Promega) and processed using the EpiTect Bisulfite Kit (Qiagen). In brief, DNAs were subject to repeated denaturing (99uC for 5 min) with several incubation steps (60uC for 25 min, 85 min, and 175 min). Reactions were cleaned up with EpiTect spin columns (Qiagen) and purified DNA was used as a template for PCR reactions with modified ALF promoter specific primer F (59-ATGTGTTTTTTGAATATTTATTAGTAAT-39) and primer R (59-ATCTCCCATAACTACTTTAATTCCTTAAAC-39). Final products of 275 bp were cloned into the pGEM T-easy vector (Promega) and sequenced.

Preparation of oocytes and kidney cells
Oocyte-positive Xenopus laevis females (NASCO) were anesthetized with ethyl 3-aminobenzoate methanesulfonate and oocytes were removed surgically. Isolated oocytes were treated with 0.2% collagenase (Invitrogen) in OR2 medium for 3 hours at RT. Oocytes were then washed with OR2 media and incubated with penicillin-streptomycin solution (Sigma) at 20uC overnight. Stage V and VI oocytes were selected under an Olympus SZ-40 stereomicroscope. Xenopus laevis kidney epithelial cells XLK-WG (ATCC) were cultured at 32uC in complete growth medium composed of 60% RPMI 1640 and 20% FBS.
Oocyte microinjections, cell transfections, and luciferase assays For microinjections, ,2 ng DNA was injected into 30 stage V/ VI oocytes using a Nanoject II injector (Drummond) and TIP10XV119 needles (World Precision). After 24 hour incubation in OR2 media, healthy oocytes were selected and centrifuged for 10 min at 14,000 rpm at 4uC and the aqueous supernatant was collected. Extracts were diluted 100-fold and luciferase activity was measured in a Turner TD-20 luminometer with the Luciferase Assay System (Promega). Reporter activity was normalized to measured protein concentrations and each experiment was performed a minimum of three times and in most cases four to six times using oocyte preparations from different frogs. This repetition was important to control for interindividual differences in oocyte quality and microinjection efficiency.
XLK-WG kidney cells were grown to 50-80% confluence on 6well plates and transfected with 1 mg of DNA using 3 ml of the FuGENE 6 reagent (Roche Applied Science). The pGL3-basic vector served as the negative control. Whole cell extracts were prepared after 24 h and luciferase activity was assayed.

Analysis of RNA by Northern blot
Northern blotting of RNA from microinjected promoter constructs was performed as follows. RNA prepared from injected and uninjected oocytes was isolated using the Trizol reagent (Invitrogen). RNAs (12 mg) were loaded onto a formaldehydecontaining gel and transferred to a Zeta-Probe Blotting Membrane (Bio-Rad). The PCR-generated hybridization probe spanned nucleotides 128-771 of the luciferase reporter and was labeled with [a-32 P] dCTP using Ready-To-Go DNA labeling beads (Amersham). A 260 bp PCR fragment from the X. laevis 5S RNA gene was used as an RNA normalization control.

Electrophoretic mobility shift analysis
Oligonucleotide probes from the X. laevis ALF promoter used in bandshift assays were prepared by T4-kinase labeling with [c-32 P] ATP. Oocyte extracts were prepared from collagenase treated oocytes centrifuged twice at 14000 rpm at 4uC for 15 min in HB buffer (50 mM Tris-Cl, pH 7.5, 10% glycerol, 5 mM Mg-Acetate, 0.2 mM EDTA, 0.5 mM DTT, and 0.1 mM PMSF) in the presence of 1% protease inhibitor (Sigma).

Identification of the Xenopus laevis ALF promoter
Since the genome sequence of X. laevis is not available, the first step in this project was to isolate the promoter of the ALF gene from purified genomic DNA. This was accomplished by a genome walking approach that involved an upstream-directed gene specific ALF primer located within the 59-UTR ( Figure 1A). The largest product (1745 bp) was generated with HincII digested DNA. This product, as well those derived from DNA digested with other enzymes, showed an identical match to 70 base pairs in the 59-UTR of the ALF cDNA, confirming that the genomic sequences are directly upstream of the ALF mRNA.
Computer analysis of the 1745 bp HincII sequence revealed several unique sequence features, including a 587 bp long DNA transposon (position 21739 to 21203) ( Figure 1B). In addition, two short repetitive motifs were identified by visual inspection. One of these had an approximately 15 bp long ''TA'' rich fragment repeated at least four times, while the other was composed of six repetitions of a ''TA-GC'' rich element with length of about 15-20 bp ( Figure 1B).

Mapping the ALF initiation site
We determined the transcription start sites of the endogenous ALF gene using two different approaches. One approach involved a series of upstream primers located at various positions within the putative promoter region. These primers were used in combination with a common downstream primer to generate RT-PCR products from oocyte RNA ( Figure 1C). The downstream primer was positioned so that amplification would occur only from spliced mRNAs and not from contaminating genomic DNA. Control reactions included a primer (S0) that was located a known distance (256 bp) upstream of the common primer. Results with primers S3, S4, S5, and S6 all generated PCR products corresponding in size to primer location, indicating that ALF mRNAs spanning these regions did exist ( Figure 1C lanes 4, 5, 6, 7). Among the products, those derived from primers S5 and S6 (lanes 6, 7) were of greater intensity than those derived from primers S3 and S4 (lanes 4, 5). The results place the most upstream initiation site somewhere between the S2 and S3 primers, and place a stronger downstream site between the S4 and S5 primers. The failure of S1 and S2 to generate a signal (lanes 2 and 3) defines an upper limit beyond which transcription does not initiate. Overall, the results show that transcription can begin as far as 160 bp upstream of the ATG codon, but that the strongest signals occur approximately 50 bp upstream.
A second approach to map the ALF gene start site involved 59-RACE analysis. In these experiments a primer located approxi-mately 200 bp downstream of the ATG start codon was used for first strand cDNA synthesis reactions followed by reamplification with an adaptor primer and a nested gene-specific primer. The resulting PCR products were cloned and sequenced to determine their endpoints. The results, summarized in Figure 1D, reveal multiple start sites with frequencies ranging from one and seven. Most sites (,80%) mapped between 16 to 66 nucleotides upstream of the ATG codon, consistent with the strong RT-PCR signals observed with primers S5 and S6 in Figure 1C. The most distal RACE product (observed once at position 2146) was at a position consistent with signals observed with primer S3. Sequencing showed that all the RACE clones matched genomic DNA and there was no evidence for an upstream exon. Collectively, the promoter isolation and start site mapping experiments led us to conclude that we had correctly isolated the X. laevis ALF promoter region, and showed that the gene possessed multiple transcription start sites.
Many germ cell promoters are hypomethylated when active in germ cells and hypermethylated when silenced in somatic cells [24,26,27]. To test whether this was also true for the Xenopus ALF gene we performed bisulfite methylation analysis of genomic DNA isolated from liver and oocyte tissue. The results with 11 liverderived clones showed an average methylation in the ALF promoter region of 55%, whereas oocyte-derived clones showed an average methylation of 8% ( Figure 1E). Except for clones numbered 25 and 26, the remainder of the samples were either unmethylated or were methylated at only two positions. Since the DNA used in this experiment was derived only from oocytes, together with any contaminating follicle cells, the results suggest an association between a demethylated promoter state and activity.

Alignment of ALF promoters from different species
We next asked whether homology between the X. laevis ALF promoter and corresponding ALF promoters from other organisms would help identify functional sequence elements. To address this question, the ALF gene from X. laevis was compared to those from human, chimpanzee, macaque, mouse, rat, horse, pig, cow, dog, chicken, and the frog Xenopus tropicalis. The sequences were identified by BLAST search and included ,140 bp upstream of the first exon.
Comparisons among these sequences showed that they can be grouped into two main categories (Figure 2). The first category include ten sequences including nine from mammals and one from chicken. A separate second category included the two frog sequences, X. laevis and X. tropicalis. Despite the fact that the sequences in both categories are germ cell specific, the sequences were quite divergent and did not enable the identification of putative functional elements based on homology alone. The highest pairwise alignment scores occurred between highly related organisms such as human and chimpanzee (100%), human and macaque (95.6%), and mouse and rat (95.1%). Interestingly, alignment of the two frog sequences, X. laevis and X. tropicalis, showed a comparatively weak similarity of 59.6%, presumably reflecting evolutionary distance and pseudotetraploidization in X. laevis [28,29]. However, the two frog sequences did display two homologous subregions. One of these (region 1 in Figure 2) was 71.9% similar over 32 nt, while another (region 2), just upstream of the ATG codon, was 73.7% similar over 19 nt.

Identification of a minimal ALF promoter of 63 bp
Microinjection experiments initially involved three constructs, a full-length version (ALF1.7) and two 59 deletions (ALF1.0 and ALF0.25), all of which were designed to drive a luciferase reporter in the pGL3-Basic vector ( Figure 3A). Construct ALF1.0 removed the DNA transposon while construct ALF0.25 removed both the transposon and the simple sequence repeats. A positive control construct placed the luciferase reporter under the control of the Figure 2. Alignment of ALF promoter sequences from different species. ALF promoters from twelve different species (mouse, rat, chimp, human, macaque, horse, pig, cow, dog, chicken, X. tropicalis, and X. laevis) were aligned. Conserved regions (shaded) could be identified among the first ten of these organisms. The two frog-derived sequences were only weakly similar and were shaded separately. The table shows pairwise identity scores. doi:10.1371/journal.pone.0006664.g002 herpes simplex virus thymidine kinase (HSV-TK) promoter, while the empty vector served as a promoterless negative control (pGL3 Basic). Constructs were injected into stage V/VI Xenopus oocytes and assayed for luciferase after 24 hours of incubation. The results showed that the ALF constructs were all active, and that even the shortest ALF0.25 construct retained as much activity as ALF1.7 ( Figure 3A). The low activity of the positive control showed that it was not as active here as would typically be expected in somatic cells.
An additional series of promoter deletions, ALF250, ALF205, ALF165, ALF125, ALF85, and ALF45, were also constructed. All of these were active except for ALF45 in which sequences between 85 and 45 had been removed ( Figure 3B). This observation showed that the region required for full activity was 85 bp or less.
Although ALF and other germ cell genes are normally expressed only in germ cells, they typically will show some activity when introduced into somatic cells. Here we compared the relative activity of the ALF0.25 construct in both oocytes and in XLK-WG kidney epithelial cells ( Figure 3C). The results showed that the relative activity of the ALF construct was about 4-fold higher in oocytes compared to a pGL3-TK control, whereas in somatic cells the activity was about 2-fold lower compared to that same control.
We also tested the relative activity of various ALF promoter constructs after transfection into XLK-WG cells and after microinjection into oocytes. The results, normalized to the activity of ALF250, showed that the deletion constructs retained full activity in oocytes until sequences between 63 and 45 were removed ( Figure 3D). In contrast, the activity of these same constructs in kidney cells showed a progressive length-dependent decline in activity. Although the basis for this decline is not known, the differences highlight the fact that somatic cells, although capable of supporting some level of transcription, do not exhibit the same regulatory profile as seen in oocytes. The use of the ALF63 construct in these experiments further refines the minimal ALF promoter from 85 base pairs to a region of not more than 63 nucleotides. The very small size of the promoter is consistent with earlier observations on mammalian germ cell promoters [22].

Identification of the core promoter elements
To determine the location of elements within the ALF promoter that are critical for activity in oocytes, several additional series of mutation/deletion constructs were made. One such series involved base substitutions between 285 and 245 in the ALF85 parent, along with three deletion mutants D1A, D2A, and D3A ( Figure 4A). Mutant M5A and all three deletion mutants showed a dramatic loss of luciferase activity ( Figure 4B). Surprisingly, alteration of the CGCAAA sequence, which appeared to be similar to a conserved TTCAAA motif in the mammalian sequences (Figure 2), had no effect on transcription.
Because the D1A deletion mutant was close to the junction between the promoter and reporter, and because that junction included a potential CA initiation site, we wanted to exclude the possibility that construct design had resulted in new sites of transcription initiation. We therefore created a second series of mutants based on a parent construct ALF85+ that contained 11 additional nucleotides extending downstream to the ATG translation initiation codon ( Figure 4A). Microinjection assays of these mutants revealed two regions that were critical for full activity ( Figure 4C). One of these was defined by mutant constructs M5B and M6B and which were overlapping with M5A. This was termed the A element. The second region was defined by mutant constructs M8B, M9B, and deletion construct D1A. This was termed the B element. The similarity in the results obtained using two different series of mutant constructs strongly suggested that A and B were bona fide regulatory elements.
Luciferase assays measure the amount of reporter protein available at the time of assay and reflect the combined effects of regulation at the transcriptional and translational levels. Because the A and B elements were close to the ATG codon and potentially within the 59-UTR of the transcribed RNA itself, we asked if their effects might be translational rather than transcriptional. To address this issue Northern blots were performed with RNA prepared from microinjected oocytes and a luciferase-specific hybridization probe ( Figure 4D). The results showed that different constructs generated different steady-state levels of reporter RNA, and that the amount of RNA correlated well with the amount of luciferase activity. The results suggest that the mutations affect transcription and therefore define the location of promoter regulatory elements.

Fine structure mapping of core promoter elements A and B
To further narrow the location and sequence of the A and B elements we prepared a series of triplet nucleotide substitutions that covered the A (AM1 to AM10) and B (BM1 to BM6) elements ( Figure 5A). The parent for these constructs was ALF63, the shortest region which retained full activity. Microinjection results for the A-series show diminished activity for the AM5, AM6 AM7, and AM8 constructs and normal activity in the others ( Figure 5B). Results with the B-series of mutants showed a loss of activity for BM2, BM3 and BM4, but normal activity for BM1 and BM5 ( Figure 5C). Overall, the data show that the promoter contains two distinct functional elements, A (59-GCGTTACGTCAGA-39) and B (59-AACTTCCGG-39).

Protein complex formation on the ALF promoter
We next asked if DNA binding proteins in oocyte extracts could recognize the ALF promoter and whether the sites of factor interactions coincided with the locations of the A and B elements. To address this point, mobility shift assays were performed using an ALF85 promoter fragment as a probe, together with whole cell extracts from stage V/VI oocytes ( Figure 6A, 6B). In the absence of any competitor three main complexes were observed ( Figure 6C, lane 1). Competition for complex 1 occurred when oligo P2-05 was added (lane 2), and partial to complete competition occurred for complexes 2 and 3 when oligo P2-89 as added (lane 6). The use of an additional set of mutant competitors based on P2-05 and P2-89 were used to narrow down the likely region of binding ( Figure 6D, 6E). As summarized in Figure 6A, complex 1 bound near a palindromic sequence AACGCGTT that partially overlapped with functional promoter element A. Partial competition for complex 1 was also observed with competitor P2-78 ( Figure 6C, lane 5), perhaps because it contained an AACGTGGTT element that was similar to AACGCGTT. Complexes 2 and 3 recognized a sequence in the left half of P2-89 ending in AACTTCC. This element was located entirely within functional element B. The overlap between the sites of promoter mutations and the approximate sites of complex formation suggests that the factors responsible for complex formation may also be important for promoter activity. Computer predictions suggest that element A might harbor sites for leucine zipper-type factors such as ATF1, CREB, and c-jun and others, while element B might harbor sites for ETS domain-containing factors such as c-ETS-1, Elk-1, Pu.1 and others. Although we have not yet verified these predictions, the identification of specific factors will be an important next step in understanding how this promoter is regulated.

Rearrangement of the A and B elements and oocyte maturation
We also examined the activity of constructs in which the relative locations of the A and B elements had been altered. Mutations in which the orientation was maintained but the spacing was altered showed a modest increase in activity for the 25 deletion, a slight decrease for the 210 deletion, and unaltered levels for the 215, +5, +10, and +30 constructs ( Figure 7A). Separation of the A and B elements by insertion of a 256 spacer showed activity similar to wildtype, and individual mutations of the A element in the MA construct and in the MAB construct showed diminished activity ( Figure 7B). The results show that the context of the A and B elements is important and that the rearrangements have complex effects on activity.
The activity of the ALF promoter was also examined in oocytes induced to mature by the addition of progesterone ( Figure 7C). In particular, the farther cells were into the maturation program at the time of injection, the greater the reduction in promoter activity. This result is consistent with the idea that maturation initiates a period of transcriptional quiescence during which events are primarily driven by translational activation of a pool of stored maternal RNAs.

Discussion
Studies on the mechanisms of transcription require a source of cells or cell extracts in which the promoter of interest displays the correct pattern of expression and inducibility. In the case of germ cell promoters this would ideally involve spermatocytes and oocytes. However, these cells undergo meiosis and cannot be propagated in cell culture. This has led to the use of cell-free extracts from whole testis, a tissue which contains germ cells in many stages of differentiation together with associated somatic cells, as well as the use of somatic cells which, although capable of transcribing transfected germ cell promoters, normally exhibit silencing of endogenous germ cell genes. Although candidate regulators can be been identified by such approaches, it has generally been difficult to prove that those factors are important for expression in the subpopulation of germ cells where the target gene is actually on. To begin to address these issues, we describe efforts to characterize a germ cell promoter by matching it to a cell type, in this case oocytes from X. laevis, where it is normally on. A summary of how somatic and germ cell transcription factors activate or silence endogenous and exogenous germ cell promoters is illustrated in Figure 8.
The use of Xenopus oocytes has a number of unique advantages. First, early studies of gene expression in these cells were based on the recognition that they are transcriptionally very active. Such studies helped define core promoter elements such as the TATA box and were important for developing methods of oocyte preparation and microinjection [10,12]. Second, recent studies have shown that the basal transcription machinery present in oocytes is distinct from that in somatic cells [1]. This machinery includes a set of oocyte-specific germ cell substitutes for core promoter recognition factors like TBP, TFIIA, and TAFs [15,17,25]. These variants replace their somatic counterparts and are therefore critical for the recognition and regulation of germ cell genes. The presence of these variants in oocytes means that these cells provide an environment where it is possible to match a germ cell promoter to the set of factors (e.g. TRF3, ALF, etc.) that normally control its expression (Figure 8). Third, oocytes can be isolated in quantities sufficient for the preparation of cellfree extracts needed for in vitro protein-DNA interaction assays. This would not be as easily possible with mammalian oocytes because such cells must be individually dissected from ovarian tissue. Disadvantages to the system include the need for sufficient experimental replication of transcription assays to compensate for microinjection damage or frog-to-frog variation in oocyte quality. In addition, the X. laevis genome sequence is not available, and orthologous germ cell promoters from frogs and mammals, although conserved in terms of expression patterns, may diverge at the sequence level. On balance however, the ease of oocyte purification and the fact that they contain regulatory factors that are physiologically relevant make these cells an attractive system to address questions about the biochemistry of germ cell gene expression.
Interestingly, alignment of ALF promoter sequences from twelve diverse species showed poor similarity when all were included, and the similarity between two related frog species X. laevis and X. troplicalis was itself only about only 60%. The two regions of homology identified in the two frog species, as well as the CGCAAAA motif that appeared to line up with a conserved TATA-like TTCAAAA sequence in mammalian promoters, were not critical for transcription activity in oocytes. These observations raise questions about the role of conserved and nonconserved elements in mediating germ cell-specific expression. One possibility is that divergent regions of the promoter might be responsible for germ cell specific gene expression while conserved motifs might be important for somatic silencing. Germ cell-expressed genes and somatic cell-expressed genes evolve at different rates, and we speculate that the greater conservation of somatic factors would require a correspondingly greater conservation of target sites whereas more rapidly evolving germ cell factors might coevolve with rapidly changing regulatory sites. Since the current study was limited to oocytes, however, we do not yet know if the sequences active for expression are also responsible for somatic silencing, or if silencing might be due to a separate, more conserved region of the promoter.
Deletion analysis demonstrates that the minimal active region is only about 63 base pairs, about half that described for the mouse ALF gene using transgenic experiments [22]. The conclusions reinforce results of previous studies which have shown germ cell promoters to be quite small. Interestingly, transcription start site mapping experiments for the endogenous ALF gene suggest that initiation occurs at sites adjacent to or upstream of the active promoter domain defined in construct microinjection experiments. Further, bisulfite sequencing experiments to assay methylation status show that the endogenous promoter is relatively demethylated in DNA isolated from purified oocytes, consistent with the general correlation between demethylation and germ cell expression [24,26,27]. Overall, these similarities suggest that germ cell gene regulation in Xenopus oocytes is similar to other germ cell expression systems, including those of mammals.
Fine scale mutational analysis revealed two regions of the promoter (A and B) that, when altered, resulted in changes in both RNA levels and luciferase reporter activity. The results suggest that these two domains, despite their proximity to the 59-UTR and ATG codon, are likely to be promoter regulatory elements rather than translational regulatory elements. Moreover, oligonucleotide competition assays showed that oocyte extracts contain factors that interact specifically with sequences within or adjacent to the A and B elements. Although computer analysis provides predictions about what these factors might be, these predictions will need to be verified in future studies using factor-specific antibodies and purification procedures. The promoter must also be recognized by oocyte-specific basal factor variants such as TRF3 and ALF in order to form a complete, transcriptionally active, preinitiation complex. Indeed, chromatin immunoprecipitation assays have demonstrated interactions between one such factor, TRF3, and a microinjected H2B promoter [17]. Whether the A and B elements are involved in direct interactions with the core transcription machinery or whether the complexes identified in bandshift assays mediate later steps in preinitiation complex assembly are interesting questions that can be addressed in future studies.
Previous efforts to characterize germ cell promoters have emphasized their relatively small size and sequence diversity and have identified many site-specific transcription factors possibly involved in their regulation. The current paper extends this work using an approach in which a germ cell promoter is introduced into a cell type, Xenopus oocytes, where the endogenous gene itself is naturally active. The results demonstrate the feasibility of the approach and define a very short Xenopus-specific germ cell promoter that can be used as a model to study regulatory factors and other mechanisms that are important for germ cell gene regulation.