Repeated sequence elements found upstream of the ribosomal gene promoter in Xenopus function as RNA polymerase I‐specific transcriptional enhancers. Here we describe an in vitro system in which these enhancers function in many respects as in vivo. The principal requirement for enhancer function in vitro is the presence of a high concentration of upstream binding factor (UBF). This system is utilized to demonstrate that enhancers function by increasing the probability of a stable transcription complex forming on the adjacent promoter. Species differences in UBF are utilized to demonstrate that enhancers do not act by recruiting UBF to the promoter, rather UBF performs its own distinct role at the enhancers. UBF function in enhancement differs from that at the promoter, as it is flexible with respect to both the species of UBF and the enhancer element employed. Additionally, we identify a potential role for the mammalian UBF splice variant, UBF2, in enhancer function. We demonstrate that the TATA box binding protein (TBP)‐containing component of Xenopus RNA polymerase I transcription, Rib1, can interact with an enhancer–UBF complex. This suggests a model in which enhancers act by recruiting Rib1 to the promoter.
Transcription of the repeated genes that encode 18S and 28S rRNAs by RNA polymerase I (RNA pol I) requires a combination of cis‐acting regulatory sequences and trans‐acting protein factors that interact with these sequences (Reeder, 1992). Regulatory sequences include promoter and enhancer elements. Xenopus enhancer elements comprise blocks of interspersed 60 and 81 bp repeats that have sequence homology to the gene promoter (Busby and Reeder, 1983; Moss, 1983; Reeder et al., 1983, Labhart and Reeder, 1984, 1985). Xenopus oocyte injection experiments demonstrate that blocks of these elements confer a competitive advantage on a linked promoter. Repetitive sequence elements are also found in the intergenic spacer, immediately upstream of the gene promoter, in rodent species. In the mouse, these repeated elements are 140 bp in length and have been demonstrated to have enhancer activity in both mouse and Xenopus systems (Kuhn et al., 1990; Pikaard et al., 1990). Unlike Xenopus, mouse enhancer elements have no sequence similarity with the gene promoter. A sequence element with enhancer function has also been identified in the yeast intergenic spacer (Schultz et al., 1993).
RNA pol I transcription minimally requires the trans‐acting factors upstream binding factor (UBF) and a second factor termed SL1 in humans (Bell et al., 1988), TIF 1B (Schnapp et al., 1990) or Factor D (Mishima et al., 1982; Tower et al., 1986) in mouse and Rib1 in Xenopus (McStay et al., 1991a). For reasons of clarity, we will collectively refer to this second factor in mammals as SL1. Human SL1 is comprised of TATA box binding protein (TBP) and TBP‐associated factors (TAFIs) of 48, 63 and 110 kDa (Comai et al., 1992, 1994; Zomerdijk et al., 1994). Likewise, mouse SL1 is comprised of TBP and TAFIs of 48, 68 and 95 kDa (Eberhard et al., 1993). Recently, we have demonstrated that Rib1 is comprised of TBP and TAFIs (Bodeker et al., 1996). However, the size and number of these TAFIs has not yet been determined.
UBF, which binds to DNA sequences within the promoter (Bell et al., 1988; Pikaard et al., 1989), has been termed an ‘architectural’ transcription factor because of its remarkable propensity to bend and loop DNA (Reeder et al., 1995). Electron spectroscopic imaging has demonstrated that a single dimer of Xenopus UBF (xUBF) can organize ∼180 bp of DNA into a loop of almost 360° (Bazett‐Jones et al., 1994; Stefanovsky et al., 1996). A ligase‐mediated probe circularization assay was used to show the ability of xUBF to loop short DNA fragments (Putnam et al., 1994). This architectural ability is conferred by the presence of multiple HMG box DNA binding motifs in UBF (Jantzen et al., 1990; McStay et al., 1991b; Hu et al., 1994). HMG box proteins have also been characterized as key architectural elements in RNA polymerase II systems (for review, see Grosschedl et al., 1994)
SL1 on its own binds very poorly to DNA, but in the presence of human UBF (hUBF) it binds tightly and specifically to DNA sequences within the human promoter (Bell et al., 1988). Similarly, xUBF and Rib1 can combine to form a stable transcription complex on the Xenopus promoter (McStay et al., 1991a). The interaction of Rib1 and SL1 with their respective promoters requires precise architecture of UBF. This is demonstrated by the observation that a UBF molecule with an inappropriate number of HMG boxes (generated by the deletion or insertion of additional HMG boxes) is non‐functional in transcription (Bell et al., 1989; McStay et al., 1991b; Jantzen et al., 1992; Cairns and McStay, 1995).
In addition to binding to DNA sequences within the promoter, UBF can give rise to a DNase I footprint over Xenopus (Pikaard et al., 1989) and mouse (Pikaard et al., 1990) enhancer sequences. Addition of a large molar excess of enhancer elements to transcription reactions can inhibit transcription initiation (Pikaard et al., 1989; Kuhn et al., 1990). This competitive effect can be relieved partially by the addition of excess UBF (Pikaard et al., 1989). These observations have been interpreted as evidence for the involvement of UBF in enhancer function (Pikaard et al., 1989). To date, there has been no direct evidence for the involvement of UBF in enhancer action. Here we describe an in vitro system in which enhancers function in many respects as in vivo. This system has allowed us for the first time to demonstrate directly the involvement of UBF in Xenopus enhancement and to also show that enhancers act by increasing the probability of stable transcription complex formation on a linked promoter.
One can imagine two classes of model of enhancer action with respect to UBF. Enhancers may act by recruiting UBF to the promoter. Alternatively, UBF may perform its own distinct role at the enhancer that does not involve translocation to the promoter. Here we have exploited the observed species specificity of UBF in promoter function to distinguish between these two models and demonstrate that UBF does indeed have a role in enhancement distinct from that at the promoter. We demonstrate that Rib1 can interact with an enhancer–UBF complex. This suggests a model of enhancer action in which an enhancer–UBF complex recruits Rib1 to the promoter. Additionally, we show that mammalian UBF2, previously shown not to function at mammalian promoters (Jantzen et al., 1992; Kuhn et al., 1994), may have a role in enhancer function.
High concentrations of xUBF facilitate enhancer function in vitro
S100 extracts prepared from Xenopus culture cells support accurate and efficient transcription initiation by RNA pol I (McStay and Reeder, 1986). The amount of extract typically used in transcription reactions (20 μl) contains 5–10 ng of UBF. This UBF can be removed efficiently from transcription extracts by immunodepletion (Cairns and McStay, 1995). Full transcription activity can be restored by the addition of recombinant xUBF. Thus, immunodepletion of xUBF does not result in co‐precipitation of any other essential factor. We have demonstrated previously that UBF makes protein–protein contacts with Rib1 (Bodeker et al., 1996). These interactions do not occur in the presence of the α‐UBF polyclonal antisera used in this immunodepletion. Thus Rib1 does not co‐precipitate with UBF. We have used this depleted extract to test enhancer function over a wide range of xUBF concentrations, starting from amounts lower than that normally found in the transcription extract and increasing to ∼50 times the level observed in the non‐depleted extract (Figure 1). The xUBF used in these experiments has been produced in a baculovirus expression system and has the same specific activity in transcription as purified xUBF (see Materials and methods for details). Promoter‐only templates (pGem40 and pGem52, Figure 1A), in competition with each other, are transcribed with equal efficiency at every xUBF input tested (Figure 1B). Promoter plus enhancer (pGem40EX, Figure 1A) and promoter‐only (pGem52) templates in competition are also transcribed with equal efficiency at low xUBF inputs (0–10 ng). However, at higher xUBF inputs (50–500 ng), pGem40EX progressively out‐competes pGem52 (Figure 1B). Quantitation of these transcription signals shows that at an input of 500 ng of xUBF, an enhancer‐containing template is 26‐fold more transcriptionally active than a promoter‐only template (Figure 1C). This level of enhancer function is as high, if not higher, than levels observed in vivo (Busby and Reeder, 1983; Moss, 1983; Reeder et al., 1983, Labhart and Reeder, 1984, 1985). Thus it appears that the inability of S100 extracts to support enhancement is principally a function of UBF concentration. We have also shown that untreated S100 extract supplemented with baculovirus‐produced xUBF supports enhancement (data not shown). We believe that the role of UBF in enhancement is direct and involves the interaction of UBF with enhancer DNA. This is consistent with the observation that enhancers can compete for transcription initiation in trans and that this effect can be rescued by the addition of excess UBF (Pikaard et al., 1989). It is worth noting that at low UBF inputs in the experiment described above (0–10 ng), enhancers appear to titrate UBF from promoters both in cis and trans. (Figure 1B, compare reactions 3, 5 and 7 with 4, 6 and 8). We believe that the high levels of UBF required to observe enhancer action in vitro are a closer reflection of the in vivo situation (see Discussion).
Enhancers function in vitro in a position‐ and orientation‐independent manner
One of the defining characteristics of enhancer action is that it is position and orientation independent with respect to the promoter (Busby and Reeder, 1983; Reeder et al., 1983; Labhart and Reeder, 1984, 1985). We have compared enhancer function on plasmid templates with enhancers located immediately upstream of the promoter (pGem40EX) or located 1 kb upstream in the reverse orientation with respect to the promoter (pGem40EX*, Figure 1A). pGem40EX and pGem40EX* are transcribed 30‐ and 20‐fold respectively more efficiently than a promoter‐only template (pGem52) in competition (Figure 1D). It appears that in the presence of a high concentration of xUBF, Xenopus RNA pol I enhancers function similarly to in vivo.
Enhancers function by increasing the probability of stable complex formation on a linked promoter
Enhancers potentially could influence a linked promoter in two distinct ways. Enhancers could alter the nature of the transcription complex on the linked promoter resulting in higher levels of transcription initiation from that promoter. Alternatively, they could increase the probability of transcription complex formation on a linked promoter. In order to discriminate between these mechanisms, we have used the in vitro system to investigate the timing of enhancer action and the stability of the transcription complexes formed on enhancer‐linked and enhancer‐minus promoters.
When the enhancer‐containing template, pGem40EX, is competed against the promoter‐only template, pGem52, we observe a linear increase in transcription signal from each template over the course of the standard 2 h transcription reaction (Figure 2A). In addition, we have calculated that the enhancement varies by at most 2‐fold throughout the course of the reaction (Figure 2B). These observations suggest that the enhancer effect is established early in the reaction, possibly during stable transcription complex formation.
A template commitment experiment demonstrates that transcription complexes formed on enhancer‐linked and enhancer‐minus promoters are indistinguishable in terms of their long‐term stability (Figure 3A). Transcription complexes were formed on either promoter‐only or enhancer‐linked templates by pre‐incubation in extract supplemented with 200 ng of xUBF. An equal amount of a second template is added prior to the initiation of transcription by the addition of MgCl2 and nucleotide triphosphates. In each case, very low levels of transcription are observed from the second template relative to the first. Thus enhancer‐linked and enhancer‐minus promoters are equally capable of preempting transcription complex formation on a second template throughout the course of the standard reaction.
In a related experiment, enhancer‐containing and promoter‐only templates (pGem40EX and pGem52 respectively) were pre‐incubated separately with extract containing high levels of xUBF (200 ng). After combining equal amounts of each pre‐incubation mix, transcription was initiated as before. Using this protocol, we observe equal levels of transcription from both templates (Figure 3B, reaction 2). This is in contrast to the situation when templates are mixed prior to pre‐incubation and the standard enhancer effect is observed (Figure 3B, reaction 1). In addition to demonstrating that enhancers function during stable complex formation, this experiment further shows that competition is required to observe enhancer function in vitro. The only plausible explanation for these results is that in this system enhancers act by increasing the probability of stable complex formation on a linked promoter. Furthermore, transcription complexes formed on promoters with or without enhancers are transcribed with equal efficiency.
The role of UBF in enhancer function is distinct from that at the promoter
Having demonstrated that enhancers exert an influence on stable complex formation, one can imagine two models to explain the role of UBF in this regard. In the first model, one could envisage that enhancers act by recruiting UBF to the promoter. The second model proposes that UBF has its own distinct function at the enhancer. This function involves long‐range effects on stable complex formation at the linked promoter. Note that in this model UBF does not translocate from enhancer sequences to the promoter. The observation that enhancers only function in vitro at high UBF concentrations combined with the high concentration of UBF in the nucleolus (see Discussion) argues against the first model. UBF would not be expected to be limiting under these conditions. Here we exploit both the species specificity of UBF function at the promoter and the presence of two forms in mammalian cells to demonstrate conclusively that UBF does indeed have a function in enhancement distinct from that at the promoter.
Xenopus and human UBFs are strikingly similar at the amino acid sequence level (Bachvarov and Moss, 1991; McStay et al., 1991a), the only major difference being the presence of an additional HMG box DNA binding motif (HMG box 4) in hUBF. This similarity is reflected in the ability of UBF to bind with comparable affinities to both promoters and enhancers of a heterologous species (Bell et al., 1989; Pikaard et al., 1989, 1990). Despite this conservation in sequence, xUBF and hUBF are not interchangeable with respect to promoter function (Bell et al., 1989). We have shown recently that the absence or presence of HMG box 4 appears to be the sole determinant of this specificity (Cairns and McStay, 1995). Another feature of UBF is that it occurs in two forms (UBF1 and UBF2) in all mammals studied to date (O'Mahony and Rothblum, 1991). Typically, each form represents 50% of total UBF. These different forms arise as a result of alternative splicing. UBF1 is fully functional in transcription. In UBF2, 37 amino acids are deleted from HMG box 2 rendering it inactive in promoter function (Jantzen et al., 1992; Kuhn et al., 1994). The in vivo role of UBF2 has remained a puzzle.
Recombinant hUBF1 and hUBF2 were produced in a baculovirus expression system (Figure 4A) (see Materials and methods for details). These UBFs were compared with xUBF for function at the Xenopus promoter (Figure 4B). Curiously, hUBF1 appears to have some limited promoter activity in this experiment. This is a consequence of the high level of input. At lower input amounts (5–20 ng), hUBF1 has no measurable activity (our unpublished observation). hUBF2 is inactive in Xenopus promoter function at every input amount tested.
hUBF1 and 2 are both incapable of exerting a dominant‐negative effect on the function of xUBF at the promoter. Even in the presence of 200 ng of hUBF1 or hUBF2, 5 ng of xUBF is sufficient to confer full promoter activity (Figure 4C, compare reactions 1, 5 and 7). We presume this is a consequence of the high off rate of UBF from promoter sequences in the absence of Rib1 (Putnam and Pikaard, 1992). This observation allowed us to test specifically the ability of hUBF1 and 2 to function in the enhancement while maintaining promoter function with a low amount of xUBF. At low xUBF input (5 ng), enhancer‐plus and enhancer‐minus templates (pGem40EX and pGem52 respectively) are transcribed with equal efficiency (Figure 4D, reaction 2). At high xUBF input (200 ng), enhancer‐plus templates are transcribed 17‐fold more efficiently than promoter‐only templates (Figure 4D, reaction 4). When a low input of xUBF (5 ng, sufficient to support promoter function) is supplemented with a high input of either hUBF1 (200 ng) or hUBF2 (200 ng), enhancer function is maintained; 70‐ and 16‐fold enhancement respectively (Figure 4D, reactions 6 and 8). It should be pointed out that hUBF2, unlike hUBF1, does not stimulate total transcription over that observed with 5 ng of xUBF alone. However, the competitive advantage conferred by enhancers in cis is still observed. These experiments demonstrate that forms of UBF that cannot function at the promoter can nonetheless function in transcrptional enhancement, and provide conclusive evidence that enhancer sequences do not function by recruiting UBF to the promoter. Instead, we conclude that UBF has a distinct function at the enhancer that must involve long‐range effects on promoter activity. A final and important conclusion from this experiment is that we have identified a potential role for mammalian UBF2 in vivo, that is transcriptional enhancement.
Enhancer function is flexible with respect to the species of enhancer element and UBF
The observation that hUBF1 and 2 can support the enhancement of Xenopus RNA pol I transcription suggests that there are less constraints on UBF in enhancer than promoter function. This point is underscored further by the observation that mouse enhancers function in Xenopus oocytes and Xenopus enhancers in mouse cells (Kuhn et al., 1990; Pikaard et al., 1990). To address this point further, we have investigated mouse enhancer function in the Xenopus in vitro system. Promoters with linked Xenopus or mouse enhancers (pGem40EX and pGem40EM respectively) were tested in competition with a promoter‐only template (pGem52) in the immunodepleted extract supplemented with 200 ng of xUBF (Figure 5A). In this experiment, 21‐ and 15‐fold enhancement was observed with Xenopus and mouse enhancers respectively. hUBF1 and 2 were then compared with xUBF in their ability to enhance transcription from a Xenopus promoter linked to mouse enhancers (Figure 5B). The templates pGem40EM and pGem52 were transcribed in competition in immunodepleted extract supplemented with 200 ng of xUBF (Figure 5B, reactions 1 and 2), 5 ng of xUBF plus 200 ng of hUBF1 (Figure 5B, reactions 3 and 4) or 5 ng of xUBF plus 200 ng of hUBF2 (Figure 5B, reactions 5 and 6). This experiment demonstrates that hUBF1 and 2 can indeed function in the enhancement of transcription from a Xenopus RNA pol I promoter by linked mouse enhancers elements. Thus we can conclude that RNA pol I transcriptional enhancement is flexible with respect to both the species of enhancer element and the UBF employed. These observations suggest that there is a fundamental difference between the functions of UBF at the enhancer and at the promoter. This point will be addressed in the Discussion. UBF has been demonstrated to bind to both mouse and Xenopus enhancer sequences despite any obvious sequence homology (Pikaard et al., 1990). Therefore, the observation that mouse enhancer elements function in the Xenopus in vitro system strengthens the connection between UBF binding to enhancer sequences and enhancer function.
Rib1 can interact with a UBF–enhancer complex
In the preceding experiments, we have demonstrated that enhancers increase the probability of stable complex formation on a linked promoter. The most likely mechanism is that enhancers act by recruiting some component of the stable transcription complex to the promoter. We have demonstrated that enhancers do not act by recruiting UBF to the promoter. As Rib1 is the other component of the stable transcription complex (McStay et al., 1991a), we suggest that the enhancer–UBF complex recruits Rib1 to the promoter. A prediction of this model is that Rib1 can interact with an enhancer–UBF complex. Consistent with this, we have demonstrated recently that UBF can make multiple protein contacts with Rib1 in a DNA‐independent manner (Bodeker et al., 1996). It is entirely possible, however, that once bound to enhancer DNA these interacting surfaces of UBF are unavailable for interaction with Rib1. In order to test whether Rib1 can interact with enhancer‐bound UBF, we have prepared a DNA affinity column that contains Xenopus enhancer sequences covalently linked to Sepharose beads. This column was pre‐loaded with xUBF, then a fraction from a heparin–Sepharose column that contains Rib1 activity was applied. We have demonstrated previously that Rib1 is the only activity present in this heparin fraction that is required for RNA pol I transcription (Bodeker et al., 1996). Bound proteins were eluted subsequently from the enhancer–UBF column and tested for Rib1 activity by combining them with a heparin 0.4 M fraction that contains both UBF and RNA pol I. Using this protocol, we demonstrate that Rib1 can bind to an enhancer–UBF complex (Figure 6A, compare lanes 2 and 3). In a control experiment, we show that no Rib1 activity is recovered from the enhancer–Sepharose column in the absence of xUBF (lane 4). Thus we can conclude that Rib1 can bind to an enhancer–UBF complex but not to enhancer sequences alone. As predicted by the observation that hUBF1 and hUBF2 can function in the enhancement of a Xenopus promoter (Figure 4C), Rib1 can also bind to complexes of enhancers with hUBF1 and 2 (Figure 6B).
High concentrations of UBF facilitate enhancer action in vitro
Here we have described a Xenopus in vitro system in which enhancers function in many respects as they do in vivo. The key feature of this system is that a high concentration of UBF is required for enhancer function. This result is not unexpected, especially when one considers the abundance and subcellular localization of UBF in vivo. Quantitative Western blotting has shown that a Xenopus culture cell contains 5×105–106 molecules of UBF (C.Cairns and B.McStay, unpublished observation). Immunolocalization experiments show that the majority of this UBF is localized to the nucleolus (Jantzen et al., 1990; Chan et al., 1991). This results in a very high local concentration of UBF in vivo. It is estimated that in Xenopus there are 500 repeats of the ribosomal gene per haploid genome, with one gene promoter and on average two spacer promoters per repeat. We can calculate, therefore, that there is between a 150‐ and a 300‐fold molar excess of UBF to promoters in vivo. We observe enhancer function in vitro with only a 5‐ to 30‐fold molar excess of UBF over plasmid templates.
Enhancers function during stable complex formation
An examination of the time course of enhancer action reveals that the enhancer effect is established early in the transcription process. These in vitro results are in agreement with in vivo observations. Enhancers can function in trans in oocytes when catenated with a promoter‐only template (Dunaway and Droge, 1989). Since catenated plasmids are resolved almost immediately following microinjection, one can conclude that enhancers act early in vivo.
Template commitment experiments demonstrate no measurable difference in the stability of transcription complexes formed on promoters with or without linked enhancers. In addition, when stable complexes are pre‐formed independently on enhancer‐containing and promoter‐only templates, equivalent levels of transcription are observed from each template when they subsequently are co‐transcribed. The most reasonable interpretation of these results is that enhancers act by increasing the probability of stable transcription complex formation rather than by increasing the transcription rate from the linked promoter. An electron microscopy‐based analysis of transcription complexes on templates injected into Xenopus oocytes also indicates that enhancers act by increasing the probability of complex formation on a linked promoter in vivo (Oshein et al., 1996). Furthermore, it was demonstrated that enhancers do not alter the re‐initiation rate of transcription from a promoter in cis. It remains to be seen if enhancers in other RNA pol I systems behave in a similar manner. However, the observation that Xenopus and mouse enhancers function across species suggests that this may be generally applicable at least in vertebrates. Recent work on the mechanism of action of RNA pol II enhancers in vivo has come to a similar conclusion. Mammalian cell transfection experiments have been used to demonstrate that enhancers act by increasing the probability of transcription from a linked promoter rather than the rate of transcription from that promoter (Walters et al., 1995, 1996).
UBF has a role in enhancer action distinct from that at the promoter
Since UBF and Rib1 are required for stable transcription complex formation on the Xenopus promoter (McStay et al., 1991a), it seems likely that enhancers act by recruiting one or other of these factors. Previously, it has been suggested that enhancers act by recruiting limiting amounts of UBF to the promoter (Pikaard et al., 1989). This could arise by UBF transiently interacting with enhancer sequences then translocating to the promoter. The observation that enhancers only function in vitro when UBF is non‐limiting for promoter function argues against this model. Alternatively, one could imagine that multimerization of UBF over enhancer sequences may have a synergistic effect on UBF binding at the adjacent promoter. However, the observation that enhancers function when separated from the promoter by 1 kb of vector sequences argues against this model. Another argument against this model is the observation that an enhancer‐containing plasmid supplied in trans but catenated with a promoter‐only plasmid can support enhancer function in vivo (Dunaway and Droge, 1989). However, the most persuasive argument against models where enhancers recruit UBF to the promoter is that forms of UBF that do not function at the Xenopus promoter, hUBF1 and 2, can nonetheless function in enhancement. From this experiment, we conclude that the role of UBF in enhancement is distinct from that at the promoter. We can also infer from this that enhancer function requires simultaneous and independent loading of promoter and enhancer elements with UBF. Consequently, high levels of UBF are required to facilitate the occupancy of both enhancer and promoter elements on a significant fraction of template molecules.
The role of UBF in enhancer action
How does enhancer‐bound UBF increase the probability of stable transcription complex formation at a linked promoter? At this point, it is worth considering further recent work relating to the mechanism of action of RNA pol II transcriptional enhancers. RNA pol II transcriptional enhancers typically contain multiple binding sites for an array of sequence‐specific DNA binding proteins. These DNA binding proteins have activation domains that interact with components of the RNA pol II basic transcription machinery, the major target being transcription factor IID (TFIID) (Tjian and Maniatis, 1994). TFIID is comprised of TBP and eight or more TAFIIs. Different activation domains contact different components of TFIID (Goodrich and Tjian, 1994). Current evidence suggests that RNA pol II enhancers function by recruiting TFIID to the promoter and that this is mediated by the combined interactions of an array of activation domains with a subset of TAFIIs (Sauer et al., 1995a,b). An additional complication has been identified from studies of the T‐cell receptor α and the human β interferon gene enhancers, where HMG box proteins have been identified as essential components of enhancer function (Thanos and Maniatis, 1992; Du et al., 1993; Giese et al., 1995). These HMG box proteins perform an ‘architectural’ role in enhancer function (Grosschedl et al., 1994). RNA pol II enhancer function therefore requires both ‘architectural’ components and components with affinity for an element of the basic transcription machinery.
Like RNA pol II enhancer binding factors, UBF has been demonstrated to interact with another key component of the basic RNA pol I transcription machinery, SL1 (Beckman et al., 1995) or Rib1 (Bodeker et al., 1996). SL1 in mammals and Rib1 in Xenopus are the RNA pol I equivalent of TFIID. UBF makes protein–protein contacts with the TAFI48 and TBP components of SL1 (Kwon and Green, 1994; Beckman et al., 1995). A likely mechanism of enhancer action therefore is that an enhancer–UBF complex recruits Rib1 to the promoter. Consistent with this model, we have demonstrated that Rib1 can interact with an enhancer–UBF complex. We presume that this interaction is mediated through protein–protein contacts with UBF since no binding to enhancer sequences is observed in the absence of UBF. Indeed, we have already demonstrated that UBF makes multiple protein contacts with Rib1 in solution (Bodeker et al., 1996). This raises the question as to why high levels of UBF present in these transcription reactions do not repress transcription by titrating out Rib1. We presume that this protein–protein interaction in solution is weaker than that between Rib1 and the promoter–UBF complex or indeed the enhancer–UBF complex.
As discussed above, UBF has been described as an architectural transcription factor (Reeder et al., 1995). It is interesting to speculate that RNA pol I transcriptional enhancement has an ‘architectural’ component and that this is mediated by UBF. This proposed architectural role for UBF in enhancement must be fundamentally different from that at the promoter. Enhancer function appears to be flexible with respect to the species of enhancer sequence or UBF. Promoter function of UBF is exquisitely sensitive to the species of UBF employed (Bell et al., 1989; Cairns and McStay, 1995). One possible explanation for this difference is that Rib1 interacts with enhancers by protein–protein contacts but with the promoter by a combination of protein–protein and protein–DNA interactions, as has been demonstrated for SL1 (Bell et al., 1988; Beckman et al., 1995).
We realize that all of the experiments described here were performed in crude extracts and that there is the potential for other, as yet unidentified factors to be involved in enhancer function. However, this does not alter the principal conclusion of this work, namely that UBF has a role in enhancement distinct from that at the promoter. The ability of UBF to perform these multiple roles in RNA pol I transcription should not be surprising, especially when one considers that it is a transcription factor dedicated to the synthesis of a single essential gene product.
Materials and methods
The plasmids pGem40 and pGem52 were constructed by cloning the −245/+50 SalI–BamHI fragment of ψ40 and the −245/+62 SalI–BamHI fragment of ψ52 respectively (Labhart and Reeder, 1984) into the SalI and BamHI sites of the vector pGem3 (Promega). pGem40EX was constructed by cloning a −970/+50 SalI–BamHI fragment of pXlr401 (Labhart and Reeder, 1984) into the SalI and BamHI sites of the vector pGem3. In the plasmid pGem40EX*, a single enhancer block (−970/−245) was subcloned from pGem40EX as a SmaI–HincII fragment into the unique SspI site present in the vector sequences of the plasmid pGem40. In the resulting plasmid, the enhancer block is in a reverse orientation with respect to, and 1 kb upstream of, the promoter. The plasmid pMrE13 contains the mouse ribosomal gene intergenic spacer sequences −1930/−169 cloned as a SalI fragment into the vector pUC9 (Kuhn et al., 1990). This entire insert was subcloned as a HindIII–SalI partially digested fragment into the HindIII and SalI sites of the plasmid pGem40. The resulting plasmid is called pGem40EM. The plasmid pGemEX(8) contains eight blocks of enhancer elements (−970/−245) in a tandem array, cloned as a SalI–XhoI fragment into the SalI and XhoI sites of a modified pGem3 vector.
Transcription extracts and assays
S100 transcription extracts in CB100 [25 mM HEPES pH 7.9, 100 mM KCl, 0.2 mM EDTA, 1 mM dithiothreitol (DTT), 20% glycerol and protease inhibitors] were prepared from the Xenopus laevis cell line, XlK2, as described previously (McStay and Reeder, 1990). Immunodepletion of UBF from S100 extract has been described elsewhere (Cairns and McStay, 1995). Briefly, S100 extract (5.0 ml) was incubated with 50 μl of α‐xUBF antiserum on ice for 30 min then chromatographed three times over a 1.0 ml protein A–Sepharose Fast Flow column (Pharmacia). Immunodepletion was >95% efficient as determined by Western blotting.
In some experiments, fractionated extracts were employed. Heparin 0.4 M and Rib1 fractions have been described previously (McStay et al., 1991a). The heparin 0.4 M fraction elutes from heparin–Sepharose with 400 mM KCl and contains both RNA pol I activity and UBF. The Rib1 fraction elutes from heparin–Sepharose with 600 mM KCl and contains no detectable RNA pol I activity or UBF.
In transcription reactions, immunodepleted extract (20 μl) was combined with baculovirus‐expressed UBF (in 1 μl volume) and incubated with template DNA (400 ng total) for 10 min on ice. Reactions were initiated by the addition of 20 μl of transcription buffer (25 mM HEPES pH 7.9, 80 mM KCl, 12 mM MgCl2, 10 mM creatine phosphate, 1 mM DTT, 100 μg/ml α‐amanitin, 1 mM nucleotide triphosphates) and incubation at 25°C. The final reaction conditions were: 25 mM HEPES pH 7.5, 90 mM KCl, 6 mM MgCl2, 10% glycerol, 1 mM DTT, 5 mM creatine phosphate, 50 μg/ml α‐amanitin, 1 U/μl RNasin (Promega), 0.5 mM NTPs and 10 μg/ml template DNA. Unless otherwise stated, reactions were allowed to proceed for 2 h. Reactions were terminated and transcripts were detected using S1 nuclease protection with a probe prepared from pGem40 to detect transcripts from pGem40, pGem40EX, pGem40EX* and pGem40EM and a probe prepared from pGem52 to detect transcripts from that template (Labhart and Reeder, 1986). Note that in reactions containing templates in competition, the reaction was split and probed independently for transcripts from each promoter type. Transcription signals were quantitated by Phosphorimaging using a GS‐250 Molecular Imager (Bio‐Rad).
Baculoviral expression of UBF
Full‐length xUBF cDNA was cloned into the baculovirus transfer vector pBluebac (Invitrogen) as described previously (Hu et al., 1994). Full‐length hUBF1 and 2 cDNAs were cloned as BamHI–EcoRI fragments into the BamHI and EcoRI sites in the transfer vector pVL1393 (Invitrogen). Recombinant virus was produced for each cDNA clone using the Baculogold transfection system (Pharmingen). For large‐scale protein production, Sf9 cells (100–500 ml in spinner culture) were harvested 3 days post‐infection and nuclear extracts were prepared as previously described (Dignam et al., 1983). Nuclear extracts were chromatographed successively over BioRex 70 (Bio‐Rad) and Mono Q (Pharmacia) ion exchange columns, where UBF elutes with 500 and 450 mM KCl respectively. At this point, recombinant UBF was >95% pure as judged by Coomassie staining of SDS–polyacrylamide gels (see Figure 4A).
Enhancer affinity chromatography
An enhancer DNA affinity column was prepared by digesting pGemEX with HindIII and covalently binding it to CNBr‐activated Sepharose CL2B as previously described (Kadonaga et al., 1986). The final DNA concentration on the affinity resin was 80 μg/ml. A Rib1 fraction (250 μl) was loaded onto enhancer–Sepharose columns (100 μl) that had been pre‐loaded or not with baculovirus‐expressed UBF (10 μg). Columns were washed repeatedly with CB100. Bound proteins were eluted with CB800 (CB100 with 800 mM KCl) and dialysed against CB100.
We thank Ingrid Grummt for the plasmid pMrE13, and Steve Keyse and MacDara Bodeker for critical reading of the manuscript. Supported by The Wellcome Trust (B.M.)
- Copyright © 1997 European Molecular Biology Organization