INTRODUCTION

Eukaryotic ribosomal RNA (rRNA) constitutes 75% of total cellular RNA and is synthesized by RNA polymerase I (pol I)¹ as a precursor to the 18, 5.8, and 28 S RNAs of the ribosome (1). The typical eukaryotic cell contains from several hundred (in vertebrates) to several thousand (in plants) rRNA genes organized in head-to-tail tandem arrays located at one or a few chromosomal sites (2). Each rDNA repeating unit consists of the transcribed pre-rRNA region and an intergenic spacer (IGS), whose length varies considerably (e.g. from ~2.5 kb in Saccharomyces cerevisiae (3) to ~30 kb in humans (4)). Within a given species, and even within the individual rDNA repeats of a single organism, the length of the IGS may be polymorphic (e.g. ~3 kb to ~9 kb in Xenopus laevis (5)). In all cases examined, this length polymorphism is due to differences in the numbers of repetitive sequence elements (6, 7, 8, 9, 10). Indeed, all metazoan organisms for which sequence information is available have one or more types of reiterated sequence elements that constitute a substantial portion of their promoter-proximal IGSs.

The IGS of X. laevis rDNA is composed almost entirely of four types of repeated elements (Fig. 1) (11, 12). The promoter-proximal portion of the IGS consists of blocks of 60/81-bp repeats (6-12 copies of 60- or 81-bp elements) that alternate with a pol I spacer promoter (SP). This unit is generally repeated two to three times/spacer, but can be repeated up to eight times. The SP is ~90% identical to the gene promoter (140 to 1) and the 60/81-bp repeats have ~80% identity to 50 bp of the gene promoter (121 to 72) (11, 13). The promoter-distal portion of the X. laevis IGS consists of ``0'' repeats, 34-bp elements reiterated ~2-10 times/spacer, and ``1'' repeats, 100-bp elements reiterated ~six to nine times/spacer (11, 12, 14).

The promoter-proximal portion of the X. laevis IGS can substantially affect transcription from the rRNA gene promoter (15, 16). The 60/81-bp repeats are pol I transcriptional enhancers because they function in both orientations and over considerable distances to stimulate transcription from an rDNA promoter located in cis, relative to one on a separate DNA molecule (17, 18, 19). The 60/81-bp repeats can also stimulate a promoter in cis in the absence of a competitor template (20), and their cis-stimulatory and trans-competitive effects can each be ~10-fold (17, 20). Although the SP does not stimulate transcription from the gene promoter by itself, it potentiates the enhancement observed from the 60/81-bp repeats, in a process that is not yet understood (21, 22).

In contrast to the transcriptional effects of the promoter-proximal repetitive elements, previous studies have concluded that the promoter-distal region of the X. laevis rDNA IGS containing the 0 repeats and the 1 repeats had no appreciable effect on transcription (19, 22, 23). However, the assays used in these experiments were considerably less sensitive than those used to demonstrate the effects of the promoter proximal elements, and they might not have detected the effects of the 60/81-bp repeats either. This report re-examines the transcriptional role of the 0/1 repetitive elements.

Following the X. laevis paradigm, the promoter-proximal repetitive elements of the rDNA IGS in mouse (24, 25), Arabidopsis (26), and Acanthamoeba (27), the promoter-proximal spacer promoter repeats in Drosophila (28, 29) and mouse (30), and the promoter-proximal spacer promoter and/or repetitive elements in rat (31) have been found to stimulate transcription from the respective cis-located gene promoters, much as in frog. Thus, there are many examples of promoter-proximal rDNA repeats acting to stimulate pol I transcription.

Using oocyte microinjection assays under conditions that can separately detect cis stimulation and trans competition, we have directly examined the effects of the promoter-distal half of the X. laevis IGS. We show that the 0 repeats, the 1 repeats, and the combined 0/1 repeats can significantly influence transcription from the rRNA gene promoter. In cis, these repeats serve as potent enhancers of both spacer promoter and gene promoter transcription, acting independent of orientation and over distances, both upstream and downstream of the initiation site. When located in trans, these promoter-distal repetitive elements instead act as inhibitors of transcription. By footprint competition, the 0 and 1 repeats were found to specifically interact with a factor that binds to the rDNA promoter; gel shift analysis indicates that the pol I transcription factor UBF can bind to these sequences. Thus, all of the criteria that establish the 60/81-bp repeats as pol I transcriptional enhancers also apply to the promoter-distal 0/1 transcriptional enhancers, indicating that virtually the entire X. laevis IGS consists of repetitive elements that enhance ribosomal transcription.

EXPERIMENTAL PROCEDURES

All subclones of the 0/1 region derive from pXlrs5, a subclone of pXlr14B (13) in which a 1.7-kb fragment starting at the HindIII site at the end of the 28 S coding region and extending to 92 nucleotides before the first BamHI (an exonuclease III/S1 deletion to remove all residues of the spacer promoter) was cloned between the HindIII and SmaI sites of pUC18. This construct includes the eight copies of the 34-bp 0 elements (2 of which are imperfect) and the six copies of the 100-bp 1 elements of this IGS (Ref. 14, and data not shown).

The 0.3-kb AvaII fragment of pXlrs5, containing the 0 repeats was S1-blunted and subcloned into the SmaI site of a pUC18 derivative in which a BglII site replaced the EcoRI site (pUC_Eco-Bgl), yielding pXlrs10. Oriented polymers (32) of one to four copies of the 0 region from pXlrs10, excised by BamHI-BglII digestion, were inserted into the BamHI site of a pUC18 derivative in which an EcoRI site replaced the HindIII site (pUC_Hind-Eco) yielding subcloned 0, 0₂, 0₃, and 0₄ repeats (pXlrs12_1-4).

The 0.85-kb HphI (S1-blunted)-SstI (polylinker site) fragment of pXlrs5 containing the subcloned 1 repeats was cloned into SmaI-SstI-digested pUC_Eco-Bgl, yielding pXlrs13. Oriented monomer and dimer constructs of the BamHI-BglII-excised 1 repeat of pXrs13 were constructed as above, yielding 1 and 1₂ clones (pXlrs14_1-2).

A subcloned 0/1 repeat, pXlrs16, was made by first cloning the ~1.3-kb BssHII (S1-blunted)-SstI (polylinker site) fragment of pXlrs5 into SmaI-SstI digested pUC_Eco-Bgl (yielding pXlrs15) and then excising this insert at the flanking polylinker BamHI and BglII sites and inserting it into the BamHI site of pUC_Hind-Eco.

The A and B plasmids are pUC-based subclones of the X. laevis rRNA gene promoter (residues 245 to +13), transcribing two different reporter sequences (see Ref. 20, and references therein). The monomer and oligomers of the 0 repeats, 1 repeats, and 0/1 repeats were then inserted in the forward and reverse (r) direction, upstream of the gene promoter, by cloning the EcoRI fragment containing these repeats from pXlrs12_1-4, pXlrs14_1-2, and pXlrs16 into the EcoRI site upstream of the promoter in the A or B constructs. This created the plasmids 0_1-4:A, 0_r1-4:A, 0:B, 0_r:B, 1_1-2:A, 1_r1-2:A, 1:B, 1_r:B, 0/1:A, 0/1_r:A, 0/1:B, and 0/1_r:B. Transcription in all these templates proceeds away from the HindIII end of the polylinker. A:0, A:1, and A:0/1 were constructed by cloning the same 0-, 1-, or 0/1-containing fragments from pXlrs12, pXlrs14, or pXlrs16 (EcoRI-excised, S1-blunted, and joined to HindIII linkers) into the HindIII site of the A construct. 0:200:A, 1:200:A, and 0/1:200:A were constructed by replacing a 40-bp AccI polylinker fragment of 0:A, 1:A, or 0/1:A with a 238-bp HpaII fragment of pBR322 (residues 1020-1258).

The E plasmid contains a subcloned block of 10 X. laevis 60/81-bp enhancer repeats (20). The E:A and E:B plasmids have the 60/81-bp block from E inserted upstream of the gene promoter in the A and B plasmids (20). A:E and e₂:A (containing one 60- and one 81-bp element) were made by cloning into pUC18 the 1.1-kb SalI or the 0.47-kb HindIII-SalI fragment from A:E(pBR) or e₂:A(pBR), respectively (20).

Template SP was made by subcloning the leftmost spacer promoter from pXlr14 (14) (SmaI fragment, residues 252 to +50 relative to this promoter, joined to ClaI linkers) into the AccI site of pUC18 and then inserting, between the upstream XbaI and EcoRI polylinker sites, the X. laevis promoter-proximal terminator (residues 243 to 188 relative to the gene promoter, originally cloned into the SmaI site of pUC18 as pXlT3² and excised with XbaI and EcoRI). E:SP contains the 60/81-bp E block joined upstream of residue 252 of the spacer promoter at an introduced SalI site. 0/1:SP contains the 0/1 region from pXlr14B (1.5-kb BssHII (S1 blunted and joined to ClaI linkers)-BamHI fragment) cloned between the BamHI site of SP and the AccI site of the vector recreating a native spacer segment. SP:E:A and 0/1:SP:E:A contain the rDNA inserts from SP and 0/1:SP (excised using the HindIII (Klenow-blunted) and EcoRI (polylinker) sites) inserted into E:A (cleaved with XbaI (S1-blunted) and EcoRI). For 5 S maxi, the 0.35-kb HindIII fragment containing the 5 S maxi gene of pXbs115/77 (33) was cloned into pUC18. The rDNA constructs are diagrammed in Fig. 2.

About 30 stage V-VI X. borealis oocytes from mature frogs were each injected in the nucleus with ~30 nl of plasmid DNA, to deliver: 0.6 fmol of template plasmid plus 1 fmol of the 5 S gene control plasmid in the cis stimulation assay, 0.3 fmol each of the template and the competitor plasmid in the trans competition assay, and 0.3 fmol of each template plasmid plus 1 fmol of the 5 S gene control plasmid in the cis/trans assays. The total DNA injected was brought to 1.5-3 ng with pUC18. Injections were performed as described (20, 34); the oocytes were centrifuged for 5 min at 30 × g immediately preceding injection and 10 µg/ml -amanitin was used to allow transcription of the 5 S control template. At 6 h post-injection, the nucleic acid was isolated (35, 36) and RNA from two oocyte equivalents was analyzed by S1 nuclease mapping (34, 36). The probes for the A and B gene transcripts were AvaI-HindIII fragments (residues 60 to +97 or 60 to +60) from the A or B plasmids, respectively, 5-end-labeled, denatured in 95% formamide, and separated on a nondenaturing 6% polyacrylamide gel. The probe for the X. laevis spacer promoter transcript was a BglI-HindIII fragment (residues 25 to +71) of the SP plasmid, 5-end-labeled and strand separated on a 4% polyacrylamide, 9 M urea gel, recovering the longer antisense strand. The 5 S maxi gene probe was a BamHI fragment of the template (residues 77 to +115), prepared similarly to the A and B gene probes except using a 5% polyacrylamide gel. The transcription signals were quantitated by densitometry or phosphorimaging.

The 25-µl binding reaction was 20 mM HEPES, pH 7.9, 105 mM KCl, 5 mM MgCl₂, 10% glycerol, 2 mM DTT, 0.14 mM EDTA and contained 5 µl of mouse S-100 extract (~10 mg of protein/ml), 200 ng of single-stranded calf thymus DNA, 400 ng of HhaI-digested competitor construct, and 1 ng of a HindIII-EcoRI fragment of the rDNA promoter (plasmid pUC 230: a 230-bp HhaI fragment of rDNA, residues 210 to +20, blunted and cloned into the SmaI site of pUC9),² 5-end-labeled at +20 (EcoRI). After 30 min at 30 °C, 800 units of exonuclease III was added in 25 µl of binding buffer and incubation was continued for 10 min at 30 °C. The DNA products were resolved on a 6% sequencing gel and visualized by autoradiography. Factors of the mouse cell S-100 extract can bind to and transcribe frog and mouse rDNA promoters with approximately equal efficacy (37). Similar exonuclease III footprints are observed at the 3 border of the stable complex region on both frog and mouse rDNA promoters,² but the most distinctive footprint is at the upstream border of the frog rDNA promoter, as used in this analysis.

Recombinant xUBF protein was synthesized in vitro from the expression plasmid pxUBF-CITE and purified (38). The mRNA transcription reaction (1.2 ml) contained 960 units of RNasin, 30 µg of plasmid template (XbaI-linearized), and 1500 units of T7 RNA polymerase, as recommended by the supplier (Life Technologies, Inc.). After 1 h at room temperature, an additional 750 units of T7 RNA polymerase was added for another 1-h incubation, then 80 µg of DNase I was added for 15 min at 37 °C, and the RNA (750 µg) was phenol/CHCl₃-extracted and ethanol-precipitated. The 10-ml in vitro translation reaction was 20 mM HEPES, pH 7.4, 100 mM KCl, 0.5 mM spermidine-(HCl)₃, 2 mM DTT, 8 mM creatine phosphate, 25 µM amino acids and contained 400 µg of the xUBF transcript and 4 ml of micrococcal nuclease-treated (39) reticulocyte lysate (Green Hectares, Oregon, WI). ³⁵S-Labeled xUBF tracer was synthesized in a 1/40 size reaction using 300 µCi of [³⁵S]methionine. After 90 min at 30 °C, the reactions were combined, diluted to 40 ml with CB100 (CBn (where n equals KCl concentration) is 25 mM HEPES, pH 7.4, n mM KCl, 5 mM MgCl₂, 20% glycerol, 1 mM DTT, 0.1 mM EDTA, and 0.5 mM PMSF), and loaded onto a 5-ml Q-Sepharose column in CB100. The column was washed with 30 ml each of CB100 and CB250, and then eluted with 20 ml of CB600. The xUBF fractions, identified by ³⁵S tracer, were combined diluted to 100 mM KCl, applied to a 5-ml Bio-Rex 70 column in CB100, washed with 25 ml each of CB100 and CB300, and eluted with a 30-ml linear gradient from CB300 to CB1400. The peak xUBF-containing fractions were combined, diluted to 100 mM KCl, applied to a 1-ml Mono-Q FPLC column in CB100, washed with 6 ml each of CB100 and CB350, and eluted with a 10-ml linear gradient from CB350 to CB600. Peak xUBF-containing fractions were dialyzed against 50 mM HEPES, pH 7.9, 200 mM KCl, 40% glycerol, 2 mM DTT, 0.2 mM EDTA, and 0.5 mM PMSF, yielding ~500 ng of purified xUBF.

UBF gel shift analysis was conducted as described (40). The 10-µl shift reaction was 10 mM HEPES, pH 7.9, 30 mM KCl, 10% glycerol, 0.3 mM DTT, 0.03 mM EDTA, and contained 3 µg of bovine serum albumin, 6 fmol of probe (a 5-end-labeled 282-bp enhancer fragment containing a tetramer of the 60/81-bp element), ~2 ng of purified recombinant xUBF, and the indicated amounts of isolated competitor fragment. Reactions were incubated 15 min at room temperature.

RESULTS

To examine whether the 0 and 1 repeats that constitute the promoter-distal portion of the X. laevis rDNA IGS (Fig. 1) could affect rDNA transcription, we initially utilized a template (0/1:SP; Fig. 2A) that contains the promoter-distal 0/1 repeats and their natural adjacent pol I initiation signal, the 3 spacer promoter, as well as a terminator element (16, 41) added to prevent read-through inhibition. The level of transcription from this construct was compared to that of an otherwise identical construct lacking the 0/1 repeats (SP; Fig. 2A) by separately coinjecting X. borealis oocytes with either of the rDNA templates plus a control 5 S maxi gene and then quantitating the resultant transcripts by S1 nuclease analysis (see ``Experimental Procedures''). The 0/1 repeats are indeed capable of stimulating transcription from the adjacent pol I promoter, by up to 30-fold (Fig. 3A, upper panel, lanes 1 and 2). This is the largest cis stimulation reported for a pol I transcriptional enhancer, reports of which have ranged from ~3-fold (mouse rRNA gene promoter by mouse rDNA enhancer; Ref. 25) to ~10-fold (X. laevis rRNA gene promoter by the 60/81-bp rDNA enhancer; Ref. 20). A comparable extent of stimulation was also observed for this promoter by a similarly located block of 60/81-bp enhancer repeats (Fig. 3A, upper panel, lane 3).

The RNA preparations were also assayed for the transcript of the co-injected control 5 S maxi gene to verify that the injections were reproducible (Fig. 3A, lower panel). (This analysis was routinely performed but is omitted from the later figures.) Reproducibility was also demonstrated by routinely performing each injection in duplicate. The duplicate results were usually within ~20% of each other; only one set of injections is shown in the figures. Furthermore, each experiment was repeated by microinjecting into the oocytes from more than one frog. Although, as expected, the quantitative level of transcription varied between different Xenopus individuals, the qualitative results were highly reproducible between individuals.

To determine if the 0/1 repeats could stimulate the rRNA gene promoter as well as the spacer promoter, transcription from a plasmid containing the A gene (a marked X. laevis rRNA gene promoter segment; Ref. 20) to which the 0/1 region had been joined upstream in cis (0/1:A, Fig. 2B), was similarly assessed by oocyte microinjection (Fig. 3B). The 0/1 repeats markedly stimulate transcription from the rRNA gene promoter (lanes 1 and 2). This level of stimulation (~20-fold) is slightly higher than that exerted by the 60/81-bp enhancer block (E:A, Fig. 3B, lane 8) in almost all frogs examined.

To determine whether the separated 0 and 1 repeats stimulate transcription, the templates 0:A and 1:A (Fig. 2B) were examined in cis stimulation assays (Fig. 3B). Individually, both the 0 repeats (eight 34-bp repeats) and the 1 repeats (six 100-bp repeats) also stimulate transcription from the gene promoter (lanes 1, 3, and 4). Their levels of stimulation are somewhat lower than for the combined 0/1 repeats (lane 2), with the relative order of cis-stimulatory potential being 0/1 > 60/81  1 0.

The cis-stimulatory effect of the 0, 1, and 0/1 repeats is not dependent on the sequence of the transcribed reporter region. Templates analogous to those of Fig. 3B but transcribing the rDNA B gene (20) give analogous results (data not shown). Furthermore, the observed cis stimulation is not attributable to an inhibitory sequence having been inadvertently introduced upstream of the basal promoter whose action is shielded by the inserted stimulatory elements, since the A and B genes have been cloned into pUC and pBR322 vectors at numerous sites and in both orientations without any appreciable effect on the level of their transcription (Ref. 20 and data not shown).

One of the distinguishing characteristics of the 60/81-bp pol I enhancers, like pol II enhancers, is that they are capable of stimulating their respective promoters in both orientations. To assess the orientation dependence of the 0/1, 0, and 1 repeats, the templates 0/1_r:A, 0_r:A, and 1_r:A (Fig. 2B) were examined in cis stimulation assays (Fig. 3B). In the reverse orientation these repeats, both together and separately, also stimulate transcription from the gene promoter (lanes 5-7). Thus the 0/1, 0, and 1 repeats share with recognized enhancers the ability to function in both orientations.

The second characteristic feature of the 60/81-bp enhancer repeats is their ability to stimulate a promoter over several hundred bp distance (although quantitatively, their stimulation diminishes at increasing distance; Ref. 20). To assess whether the 0 and 1 repeats also stimulate over a distance, a 0.2-kb segment of prokaryotic DNA was inserted to generate the templates 0:200:A, 1:200:A, and 0/1:200:A (Fig. 2B) in which the stimulatory elements are separated from the rRNA gene initiation site by 0.45 kb. These displaced 0, 1, and 0/1 repeats also stimulate transcription from a downstream gene promoter, to approximately the same extent (± ~25%) as in the parental 0:A, 1:A, and 0/1:A constructs (Fig. 4A).

The position dependence of the promoter-distal stimulatory elements was further examined using the templates A:0, A:1, and A:0/1 (Fig. 2B) in which the repeats were positioned 0.15-0.2 kb downstream of the rRNA gene initiation site. The 0, 1, and 0/1 repeats also stimulate when 3 of the promoter, and with nearly the same efficacy as when upstream (Fig. 4B, lanes 1-7). An analogous result is observed with a similarly positioned 60/81-bp enhancer block (Fig. 4B, lanes 8 and 9). The ability of the 60/81-bp repeats to function 3 of the promoter was also suggested earlier using a different kind of assay (18); an apparently contradictory result (20) was likely due to that particular clone. Thus, the 0 and 1 repeats exhibit the same stimulatory properties as the 60/81-bp enhancer repeats; therefore, the 0 and 1 repeats are likewise pol I enhancers.

The extent of cis-enhancement by the 60/81-bp repeats increases with the number of copies (Fig. 2B) (22, 42); in some frogs, this additive effect does not plateau until more than two blocks of the 60/81-bp repeats are present, while in other frogs it plateaus at one of these 0.7-kb blocks (Ref. 20 and data not shown; see also Ref. 19, where the regions examined included variable numbers of inactive or active spacer promoters). To determine if multimers of the 0 and 1 repeats can further stimulate transcription over the level obtained from a single block, these elements were polymerized to >1 kb of repeat sequence and similarly inserted upstream of the A gene (Fig. 2B). Because we anticipated variation between individuals (19), oocytes from several frogs were separately microinjected and compared using the cis-enhancement assay (Fig. 5). Increasing numbers of 0 repeats afforded increasing levels of stimulation over the basal promoter. Under conditions of weak basal promoter activity, the 0 polymer series can give an almost linear increase in transcription up to four copies of the 0 block (Fig. 5A, lanes 1-5), while in oocytes from other frogs, saturation was reached at about two copies (Fig. 5A, lanes 8-12). On a per length basis, the 0 repeat block (~0.3 kb with eight copies of the 34-bp 0 element) is about as efficient at stimulating transcription as the 60/81-bp enhancer repeats (0.5-2 times as efficient in different individuals). A similar analysis with a polymer of the 1 repeats revealed that in oocytes from some individuals the cis-stimulatory effect is already saturated with one copy of the 1 block (Fig. 5B), while in others additional repeats can afford more stimulation (data not shown).

Like the subcloned 60/81-bp enhancer block that inhibits transcription from an rDNA promoter when coinjected in trans (17), the subcloned 0/1 repeats (Fig. 2C) strongly inhibit rDNA promoter activity in trans (Fig. 6A, lanes 1, 5, and 6). Additionally, both the separated 0 and 1 repeats inhibit transcription in trans, and this effect increases with increasing numbers of repeats (1 repeats in Fig. 6A, lanes 1, 3, and 4; 0 repeats in Fig. 6B). These elements exhibit a trans-competitive effect of comparable magnitude to that of the 60/81-bp enhancer repeats on a per length basis. As expected, the trans-competitive strength of these repeats varies somewhat between individuals (e.g. Fig. 6, A and B, lanes 2), but in all cases these promoter-distal repeats have a substantial inhibitory effect on transcription of a pol I promoter in trans.

Analysis of the 60/81-bp enhancers frequently has utilized a cis/trans assay in which two different marked promoter-containing plasmids are coinjected, at least one of which bears the enhancer repeats (e.g. Ref. 19), as in Fig. 7A (lanes 1-3). Using the cis/trans assay, the 0/1 repeats (lane 4) exert an effect similar to the 60/81-bp repeats (lane 3), enhancing transcription of the promoter carrying the repeats (B gene) and inhibiting transcription of the coinjected promoter lacking the repeats (A gene). Thus, the effect of the 0/1 repeats is also highly obvious in the cis/trans enhancer assay.

We next examined whether the 0/1 repeats could function in their natural, promoter-distal spacer arrangement (see Figs. 1 and 2A), through the intervening promoter-proximal 60/81-bp enhancer and spacer promoter, to further stimulate transcription from the downstream gene promoter. When examined in frogs whose cis-stimulatory response is not saturated by the 60/81-bp repeats and the spacer promoter (Fig. 7B, lanes 1-3; see also Ref. 21), the additional presence of the 0/1 repeats directs still greater enhancement of gene promoter transcription (lane 4). In the individual of Fig. 7B, the presence of each of these additional elements increases transcription from the downstream gene promoter by about 2-fold.

The trans competition results (Fig. 6) suggest that the 0/1 repeats bind a factor that is required for transcription at the ribosomal promoter. To more directly examine whether the 0/1 repeats interact with a rDNA promoter-binding factor, we used footprint competition analysis. When the X. laevis rDNA promoter, 5-end-labeled downstream from the initiation site, is preincubated in transcriptionally competent cell extract, a complex is formed that blocks the progression of exonuclease III at the borders of the region required for stable transcription complex formation (43, 44), resulting in a strong exonuclease stop at residue 141 (Fig. 8).² The band that results from the blockage of exonuclease progression indeed represents a specific complex, because its formation is not inhibited by the presence of unlabeled plasmid vector alone (lane 4) or by several other subcloned nonspecific DNA sequences (data not shown), but is inhibited by rDNA promoter-containing plasmid (lane 9) and by 60/81-bp enhancer-containing plasmid (lane 8).² When used as unlabeled competitors, the plasmids containing the subcloned 0, 1, and 0/1 repeats similarly inhibit formation of the exonuclease III footprint on the promoter (lanes 5-7). This suggests that the 0 and 1 repeats in addition to the 60/81-bp repeats may both bind a factor in common with the rDNA promoter.

 Extract

Because the 0 and 1 repeats and 60/81-bp repeats all inhibit formation of the promoter footprint similarly (Fig. 8), they might all bind the same rDNA transcription factor. UBF is a polymerase I transcription factor that interacts within the upstream rDNA promoter domain in a functionally relevant manner (38, 45). UBF is also the only protein thus far reported to bind the 60/81-bp enhancer, based on a characteristic DNase I footprint (25, 46) and mobility shift (40) created by isolated or recombinant UBF on those repeats. On the basis of these results, UBF has been thought to be important for enhancer function (40, 46), even though its binding exhibits limited sequence specificity (see ``Discussion''). To examine whether UBF binds to the 0 and the 1 enhancer repeats like it binds to the 60/81-bp enhancer repeats, X. laevis UBF was synthesized by in vitro transcription/translation and was purified to homogeneity (Fig. 9A) (38). Incubation of recombinant UBF protein with a radiolabeled fragment containing two of the 60-bp and two of the 81-bp sequences results in a characteristic (40) UBF-mediated electrophoretic mobility shift (Fig. 9B, lane 1) that is inhibited by co-incubation with unlabeled 60/81-bp repeats (lanes 2-4) to a greater extent than by co-incubation with a comparable amount of nonspecific dA-dT DNA (lanes 11-13). Notably, both the isolated 0 repeats and isolated 1 repeats also inhibit this mobility shift (lanes 5-10) to an extent comparable to the 60/81-bp repeats. In separate experiments, we found that incubation with UBF also causes a mobility shift and discrete DNase I and exonuclease III footprint patterns on the isolated 0 elements and on the isolated 1 elements (data not shown). Thus, by all the criteria used to infer that the pol I transcription factor UBF binds to the 60/81-bp enhancer repeats, it binds in a similar manner to the 0 and the 1 enhancer repeats.

DISCUSSION

The X. laevis 0 and 1 spacer repeats are enhancers of pol I transcription. The X. laevis rDNA intergenic spacer consists almost entirely of four kinds of repeated elements. The promoter-proximal portion contains the 60/81-bp enhancer repeats and spacer promoters, both of which serve to stimulate pol I transcription, while the promoter-distal portion contains two other types of repeats, the 0 and 1 repeats (11, 23, 47). We report here that the 0/1 repeats, as well as the individual 0 and 1 repeats, are enhancers of pol I transcription and exhibit all of the known properties of the 60/81-bp enhancers (17, 19, 20). The 0, 1, and 0/1 repeats stimulate transcription from a promoter when located in cis (Fig. 3, A and B), when in the reverse orientation relative to the promoter (Fig. 3B), when moved hundreds of bp upstream from the promoter (Fig. 4A), and when located within the transcribed region downstream from the promoter (Fig. 4B). They also stimulate transcription from a promoter in cis relative to that from a coinjected promoter in trans (Fig. 7A). Thus, virtually the entire multikilobase X. laevis IGS is made up of pol I enhancer repeats.

The extent of cis stimulation afforded a pol I promoter by the 0 and 1 repeats increases with an increasing number of repeats until the saturation level is approached (Fig. 5). This saturation point varies between individual frogs, a phenomenon also observed with the 60/81-bp enhancers (20, 42). Notably, the level of cis stimulation afforded by the 0 and 1 repeats is comparable to and in most frogs somewhat greater than that afforded by similarly positioned 60/81-bp enhancer repeats, on a per length basis.

In oocytes from individuals in which the maximal level of enhanced transcription is very high relative to the level from the promoter alone, we also observe that the 0/1 repeats in their natural promoter-distal location in the IGS stimulate transcription from the downstream gene promoter, above the level directed by the intervening promoter-proximal 60/81-bp enhancers and spacer promoter (Fig. 7B). In the mature oocytes of other frogs, the level of gene promoter transcription appears maximal with a single 60/81-bp block and no additional stimulation from the more distal spacer promoter and 0/1 repeats is observed (data not shown). This presumably explains why previous studies concluded that the 0 and 1 repetitive elements do not exert a significant transcriptional effect (19, 22, 23). These conclusions were based entirely on the type of experiment of Fig. 7B, where detection of stimulation by the distal 0/1 repeats is frog-dependent; in the oocytes examined, the maximal level of transcription may have been reached in the absence of the 0/1 region. These early studies did not specifically examine the promoter-distal repeats in experiments like those of Fig. 3-7.

In trans competition assays, the subcloned 0, 1, and 0/1 enhancer repeats inhibit transcription from a coinjected ribosomal promoter on a separate plasmid molecule (Fig. 6), as do the 60/81-bp enhancer repeats (17, 20), suggesting that both types of enhancers may bind a pol I transcription factor (see also Refs. 23 and 48). This hypothesis was strengthened by the finding that formation of a footprint on the rDNA promoter that is diagnostic for a stable transcription complex is specifically inhibited by the 0, 1, and 0/1 repeats (Fig. 8), as well as by the 60/81-bp repeats, but not by various nonspecific DNAs. Additionally, the 1 repeats exhibit a strong sequence identity with the upstream domain of the rRNA gene promoter (Fig. 10A; a 63-bp segment of >75% identity), suggesting that a common factor may bind to the promoter and to the enhancers. The 60/81-bp repeats show a comparable extent of sequence homology with the central region of the rRNA gene promoter (a 43-bp segment of ~80% identity (114 to 72); Refs. 11, 12, 13); and an artificially polymerized version of the central region of the rRNA gene promoter can function as a pol I enhancer (49).

The pol I transcription factor UBF that acts at the rDNA promoter (see Refs. 38, 45, and 50, and references therein) is the only protein reported to bind to the 60/81-bp repeats and has been proposed to be important in mediating enhancement by the 60/81-bp repeats (25, 40, 46). Since analogous assays show that UBF can bind similarly to the 0 and the 1 repeats (Fig. 9), UBF is also a candidate for mediating the action of the 0 and 1 enhancers. However, competition studies using footprinting, gel shift, and UV cross-linking do not reveal the extent of binding specificity for UBF to the promoter or to any of the enhancers, as is typical for pol II transcription factors (Ref. 51; Fig. 9)³ and mutations of the 60/81-bp sequence that inhibit enhancer function can have minimal effect on UBF binding (42). Furthermore, UBF appears to recognize structural features of DNA rather than a certain nucleotide sequence (51, 52, 53, 54, 55). From such considerations (see also Refs. 49 and 56), it appears that UBF alone may not mediate transcriptional enhancement, and it remains to be proven whether UBF is indeed directly relevant to the in vivo action of pol I enhancers, both the 60/81-bp repeats and the 0 and 1 repeats.

The promoter-distal intergenic spacer of Xenopus borealis rDNA contains extensive repetitive regions that have a high degree of sequence identity to the 0 and 1 repeats of X. laevis. There is >70% identity over a 70-bp region of a sequence that is repeated nine times in the X. borealis IGS region 2 with the X. laevis 1 repeat (Fig. 10B, bottom). There is also 100% identity between a 13-bp sequence that is repeated three times in the more promoter-distal X. borealis IGS and sequences of the X. laevis 0 repeat (Fig. 10B, top). Furthermore, there is 70% identity between a 28-bp sequence that is repeated four times in the promoter-distal X. clivii IGS (57) and sequences of the X. laevis 0 repeat (Fig. 10B, top). These repetitive elements in the spacers of other Xenopus species that have sequence identities to the 0 and 1 repeats may also be promoter-distal enhancers of pol I transcription. Furthermore, in mouse, Drosophila, Arabidopsis, and Acanthamoeba where promoter-proximal repetitive enhancers have been demonstrated, there are also more promoter-distal repetitive elements (26, 27, 58, 59, 60, 61); these might also function as promoter-distal enhancers of pol I transcription.