[Frontiers in Bioscience 3, d376-398, March 26, 98]

Table of Conents
 Previous Section   Next Section


Katherine M. Hannan, Ross D. Hannan and Lawrence I. Rothblum

Henry Hood Research Program, Department of Molecular and Cellular Physiology, Penn State College of Medicine, Weis Centre for Research, 100 N. Academy Avenue, Danville, PA 17822-2618

Received 3/18/98 Accepted 3/23/98


Essential components required for efficient rDNA transcription include the rRNA genes, RNA polymerase I, RNA polymerase I associated factors and a number of rDNA specific trans-acting factors such as SL-1, the homologue of TFIID, and UBF (17). In addition, other proteins have been reported to be components of the transcription initiation complex and may participate in the regulation of rDNA transcription (17). The specific contributions of these factors to the regulation of rDNA is poorly understood. However, they present interesting links between transcription by RNA polymerase II and RNA polymerase I.

4.1. The rRNA Genes

There are approximately 150-200 copies of mammalian rRNA genes (rDNA) present per haploid genome. In general, the genes are distributed among several chromosomes and arranged in tandem, head to tail arrays with the coding regions of the primary transcript being separated by nontranscribed or intergenic spacer regions (figure 1). The length of the transcript generated from the rDNA varies from ~8 kb (yeast, Drosphila and Xenopus) to ~13 kb (mammals), and this appears to be dependent on the length of the external and internal transcribed spacer regions (2,5,6). Although examination of the sequences of the rDNA promoters of different genera fails to demonstrate significant sequence identity, there is a high degree of conservation between the functional elements (3). In fact, the human, mouse, frog and rat rDNA promoters all share a similar molecular anatomy (4). In addition to the promoters, the vertebrate rDNA repeats contain terminator elements as well as additional transcription elements within the intergenic spacer. Surprisingly, the nontranscribed spacer of the yeast rDNA repeat also functions in termination and the anatomy of the yeast promoter is more similar to those of the vertebrate rRNA genes than the Acanthamoeba rDNA promoter is.

4.1.1. rDNA Promoter

Functionally, the vertebrate rDNA promoter consists of 2 domains, the core promoter element (CPE:~+6 to -31, with respect to the transcription initiation site), and the upstream promoter element (UPE) which extends from the CPE (-30) to ~ -167 (figure 2) (3,4,5,6,17). The CPE is necessary and sufficient for in vitro transcription, and is required but not sufficient for in vivo transcription. The UPE is not absolutely required for transcription initiation in vitro. However, it can stimulate transcription from the CPE under stringent conditions in vitro and is required for transcription in vivo (38). Transcription from the CPE occurs without the formation of a stable preinitiation complex, and the experimental evidence suggests that the UPE is essential for the formation of the stable preinitation complex in vitro (39,40). Studies using deletion, point, and linker scanning mutants have demonstrated that the UPE is important for transcription but have, with one or two exceptions, failed to identify critical nucleotides within the CPE. Interestingly, the CPE has been demonstrated to consist of at least two functional domains, and individual nucleotides, such as the Gs at -7 and -16, have been demonstrated to fulfill critical roles both in vitro and in vivo (2,41).

Figure 2. Schematic depiction of the factors interacting with the 45S rRNA promoter. The factors include SL-1, UBF, TFIC (IC), PAF53 and RNA polymerase I (RNA Pol I). A. Illustration of the DNA binding sites with which the factors interact. These include the upstream promoter element (UPE) and the core promoter element (CPE). B. Illustration of a model where UBF bends the rRNA promoter bringing two molecules of SL-1 in contact. Note that the stoichiometry of SL-1 to the promoter is not known.

Experiments using distant-altering mutations demonstrated an interesting relationship between the UPE and CPE. For example, Pape et al. (42) demonstrated that altering the spacing between the UPE and the CPE of the Xenopus rDNA promoter allowed that promoter to be transcribed efficiently by mouse extracts (42). In addition, distant altering mutants of the rat rDNA promoter suggested that the distances between the UPE and CPE were critical for initiation. However, this response was not uniform across the entire UPE, suggesting that different segments of the UPE must have different functional roles, or are "neutral" with respect to their role in the structure of the preinitiation complex (43). To date, the results from published studies are consistent with a model in which the protein complexes that form on the UPE interact with and possibly stabilize those complexes bound to the CPE. This, might then enhance the rate of passage through the rate-limiting steps involved in the formation of an open initiation complex. At least two transcription factors have been shown to interact with the UPE and CPE, these are UBF and SL-1 (41,44,45,46). Interestingly, although the model by which this is accomplished on vertebrate promoters is physically different than that proposed for Saccharomyces promoter, it is biochemically similar (discussed below).

4.1.2. Intergenic Spacer

The intergenic spacer lies between the transcribed regions and is bound at both ends by transcription termination signals (2,4,5,17). In Xenopus laevis the intergenic spacer is punctuated by 2-7 spacer promoters and in turn these are separated by six to twelve 60 and 81 bp directly repeating elements (47). The spacer promoter is almost a perfect duplication of the rRNA promoter with as high as ~ 90% homology in the regions -145 to +4, and in the imperfect copy of a 42 bp sequence (active core), that localizes to the -72 to -114 region of the gene promoter (48). However, in rat and mouse the spacer and 45S promoters contain only one conserved block of 12-13 bp which includes the Gs at -7 and -16 (3,48). The spacer promoter is transcribed by RNA polymerase I producing a transcript which terminates just upstream of the rRNA promoter, ~ -167 bp. In Xenopus, the spacer promoter may also enhance transcription from the gene promoter, possibly by delivering RNA polymerase I to the gene promoter (49). However, other studies contradict this observation (48).

The intergenic spacer of Xenopus contains additional repetitive elements. The most notable of these are the 60 and 81 bp repetitive elements which are homologous to a portion of the 45S promoter. The 81 bp elements are identical to the 60 bp elements except they contain an additional 21 bp of unique sequence (47). The 3 end of the intergenic spacer (-2300 to -3950), i.e. the region near the 3 end of the 45S rRNA transcript, is a region that shows little homology with either the spacer promoter or the 60/81 bp elements. It consists of at least two repetitive elements (repeat 0 and 1) and some non-repetitive elements (figure 1) (48).

The intergenic spacers of Xenopus, yeast, Drosophila, mouse and rat, contain elements which enhance transcription from their "major" promoters (3,4,5). In Xenopus, the cis-acting 60 or 81 bp repeat elements, enhance transcription from both the 40S preribosomal RNA and the spacer promoters. In this case the rate of transcription has been shown to be directly proportional to the number of repeat elements and independent of their orientation or distance from the promoter (4,5). Such characteristics are typical of enhancers described in RNA polymerase II transcription. Other enhancer elements have been reported in the rat and mouse intergenic spacers, including the 130 bp element which comprises the variable region of the rDNA repeat (41) and the 37 bp enhancer motif localized in the rat 174 bp non-repetitive region which is able to enhance both RNA polymerase I and RNA polymerase II transcription (3). Additional 140 bp and 200 bp element have also been identified in rat and yeast, respectively (3,50). To date the mechanism by which these elements enhance rDNA transcription is not clear. The repeated elements in the mammalian intergenic spacers and the 60/81 bp repeats of the Xenopus spacers have been shown to bind UBF and to act across species. However, it is not clear if UBF is the only factor that binds to the repeated elements and if UBF is solely responsible for enhancer activity.

4.1.3. Terminators

At the 3 end of the primary transcript of mammalian 45S rRNA genes lie several copies of a 17 bp motif, referred to as the Sal box (figure 1). The Sal box functions as orientation dependent terminators of transcription (2,4,17,18,51,52). The 13 bp promoter proximal terminator (T0) located ~ -167 bp +1 is a Sal box (figure 1). In both cases, the terminator elements act as binding sites for the 105 kDa RNA polymerase I transcription termination factor, TTF-1 (51,53). TTF-1 binds to DNA in a polymerase specific but not a species specific manner. This suggests that TTF-1 once bound to the terminator site acts by interacting with one of the unique subunits of RNA polymerase I (53). Interestingly, recent studies indicate that TTF-1 can associate with RNA polymerase I in the absence of DNA (R. Hannan and L. Rothblum, unpublished observation).

In vertebrates, the process of transcription termination requires two steps: i) RNA polymerase I pausing and its subsequent release; and ii) release and processing of the 3 end of the pre-rRNA (4,18). Mammalian transcription termination requires TTF-1 for the pausing of RNA polymerase I ~11 bp upstream of the Sal box. Interestingly, the second step in termination requires a T-rich element upstream of the TTF-1 binding site and a releasing factor (51,54). In Xenopus a terminator factor (54) that binds to the T3 box in the intergenic spacer has been identified, Rib2 (54,55).

Yeast rDNA repeats contain unique termination elements, which are comprised of two domains. The first domain consists of an 11 bp element, sometimes referred to as a REB1 element, which serves as the binding site for Reb1p (2,56,57). Transcription termination for yeast RNA polymerase I also requires about 46 bp of T-rich 5' flanking sequence. It has been suggested that the Reb1p-DNA complex comprises a pause element, while the 5' flanking sequence contains a release element. In contrast to the TTF-1-DNA complex, the Reb1p-DNA complex is not specific for polymerase I (57,58,59).

The terminators may also serve secondary functions. The promoter proximal terminator element (T0) not only serves to terminate transcripts originating from the spacer promoter, but it may also activate transcription. It has been suggested that T0 may function to prevent promoter occlusion. Occlusion is a phenomemon by which transcription through a promoter disrupts the semistable preinitiation complex (3). Thus, in the absence of TTF-1, the transcription factors SL-1 and/or UBF would be displaced from the rDNA by a polymerase that was transcribing the promoter. Recent studies suggest that TTF-1 dependent transcription activation is dependent on chromatin and involves repositioning nucleosomes in an ATP dependent fashion (29,60). However, the exact mechanism by which TTF-1 catalyzes these functions is unknown. It has also been proposed that TTF-1 may function in DNA replication, as it results in the arrest of replication fork movement and thus directs DNA replication in the same direction as transcription (61).

4.2. Proteins Involved in rDNA Transcription

4.2.1. RNA polymerase I

The core mammalian RNA polymerase I is a large, complex enzyme with a total approximate Mr of 500-600,000 and the subunit composition is yet to be confirmed. Varying reports suggest a subunit composition ranging from 11 subunits and 2-3 associated factors to only 2 large and 3-4 smaller subunits depending on the purification method implemented (62,63,64,65,66). Two recent studies, employing different purification schemes, report that mammalian RNA polymerase I is composed of at least 12 subunits with 3 associated factors (PAFs) (67, 68). In contrast, yeast RNA polymerase I has been subject to a detailed series of studies, and fourteen subunits have been identified and cloned (62,64,69).

To date only four of the mammalian RNA polymerase I subunits have been cloned, including the two largest subunits of 190 kDa and 127 kDa, which are analogous to the archaebacterial b and bsubunits, respectively (62,67). The other cloned subunits, AC40 and AC19 (64,69), are common to both RNA polymerase I and III. The yeast subunits are classified into three groups: i) four core subunits: b-like (A190), b-like (A135) and two which are similar to the bacterial asubunits (AC40 and AC19); ii) five subunits common to all three RNA polymerases: ABC27, 23, 14.5, 10a, 10b; and iii) five specific subunits: A49, 43, 34.5, 14, 12.2 (62). There is a large degree of sequence conservation between the homologous mammalian and yeast homologous RNA polymerase I subunits and also between the RNA polymerases I, II and III subunits themselves (62). Interestingly, the b and b subunit of yeast RNA polymerase I are more identical to the b and b subunit of rat RNA polymerase I than they are to the band b subunit of yeast RNA polymerase II.

The majority of the yeast RNA polymerase I subunits are essential for growth especially the five ABC and two AC subunits (62). However, the A34.5 and A49 subunits are not strictly essential for cell growth. For example, mutations of A49 generate slow growing colonies with only reduced RNA polymerase I activity illustrating that it is important, but not essential for cell viability (62). Identification of the specific functions of the RNA polymerase I subunits has been limited and restricted mainly to the yeast system. Experimental evidence to date demonstrated that the A190 and A135 subunits cross-link to nascent chain RNA and contain putative Zn2+ fingers. Other subunits containing putative zinc binding domains include A12.2, ABC10aand ABC10b. Independent studies have suggested that Zn2+ binding may be essential for activity and/or the structural integrity of RNA polymerase I (62).

In order for functional RNA polymerase I to initiate transcription it must recognize and bind the transcription initiation site. The A190, A135, A125 and ABC23 subunits have been implicated in this process. However, the domain(s) involved in this process are yet to be defined (62). Furthermore, the A135 subunit contains a putative nucleotide binding domain suggestive of a role in elongation (62,70). In addition, studies suggest that A190 may also play a role in elongation since resistance to a-amanitin, a drug which interferes with chain elongation, maps to the b subunit of RNA polymerase II (62,71).

In order for RNA polymerase I to mediate transcription initiation and elongation, it needs to interact with other proteins. To date a number of proteins termed RNA polymerase I associated factors, such as TFIC, Factor C*, TIF-IA, TIF-IC and PAF53 (68,72,73,74) have been shown to closely interact with RNA polymerase I. In addition, there is evidence suggesting that RNA polymerase I itself may interact with the transcription factor UBF. One study has demonstrated that UBF interacts with a 62kDa subunit of murine RNA polymerase I in vitro (75). However, RNA polymerase I purified by another group did not contain a 62 kDa subunit, and that laboratory reported an interaction between UBF and the 180, 114 and 44 kDa subunits of mouse RNA polymerase I, as well as with PAF53 in vitro (64,68). The association of RNA polymerase I with PAF53 has been confirmed (67,76). However, the interaction between RNA polymerase I and UBF has proven more problematic (67,77). The reasons for these disparities are unclear and further investigation is required.

4.2.2. RNA Polymerase I Associated Factors TFIC, TIF-IA and Factor C*

TFIC (15), TIF-IA (78) and Factor C* (72,79) are factors closely associated with RNA polymerase I and thought to be intimately involved in regulating its activity under certain growth conditions. The majority of studies examining these factors have been carried out in systems where rDNA transcription is virtually shut off. For example, depriving tumor cells of essential nutrients and growth factors, or treating hormone sensitive lymphosarcoma cells with glucocorticosteroids reduces both rDNA transcription and RNA polymerase I ability to initiate specific transcription (70,80,81). In each case the identified RNA polymerase I associated factor has been shown to restore the ability of RNA polymerase I to initiate specific transcription (15,72,78). Moreover, studies indicate that while not critical for the formation of a stable pre-initiation complex, all three factors are required for the formation of the first phosphodiester bond of nascent pre-rRNA (78,80). Interestingly, a factor, with properties similar to the mammalian polymerase associated factors, has been identified in yeast. This protein, Rrnp3, has been shown to interact directly with RNA polymerase I, independent of the DNA. It has been suggested that Rrnp3 stimulates the recruitment of the polymerase to the stable complex containing the rDNA promoter, and the Rrn6/7/11 and the Rrn5/9/10 complexes (82).

It has been suggested that TFIC, TIF-IA and Factor C may well represent the same biological component. However, there are no antibodies to these factors and none have been cloned. Thus, this conclusion remains to be verified. Accordingly, our understanding of the specific role these factors play in rDNA transcription is limited.

Arguments against the possibility that these activities represent the same factor include differences in the subunit composition of the purified factors. For example, mouse TFIC activity co-purifies with three polypetides present in a stoichiometric ratio of 1:1:1, with approximate molecular mass of 55, 50 and 42 kDa (81). In contrast, TIF-IA activity purifies with one 75 kDa polypeptide (78). In addition, TIF-IA has been found to be far less abundant in cells compared to TFIC. Moreover, TIF-IA can be liberated from the initiating complex and recycled to facilitate transcription from other templates (78). In contrast, Factor C* or TFIC functions stoichiometrically in vitro (79,83), i.e. it can activate one round of transcription and is then "used up." Interestingly, if elongation is halted within a critical distance (54 bp) Factor C* remains active (79).

It is not possible to compare all of the properties reported for each factor, as different laboratories have carried out different characterizations. For example, mouse TFIC activity was reported to be heat stable (15), and mouse TIF-IA has not been found to be species specific. TIF-IC

TIF-IC has been identified as a 65kDa factor associated with RNA polymerase and is required for the assembly of the initiation complex, formation of first internucleotide bond and chain elongation (84). TIF-IC contributes to the chain elongation by stimulating elongation and suppressing RNA polymerase I pausing. TIF-IC also inhibits nonspecific initiation and supports the synthesis of full-length, run-off transcripts (84). However, as with the above RNA polymerase I associated activities, there are no antibodies to TIF-IC and this factor has not been cloned. This precludes a more complete analysis of its contribution to rDNA transcription. PAFs

Recently, three polymerase associated factors have been isolated and cloned from mouse cells, PAF53, PAF51 and PAF49 (68). All three factors are tightly associated with RNA polymerase I but dissociable under certain purification conditions, indicating that they are probably not core subunits of this enzyme. PAF53 and PAF51 are structurally related proteins since they are both recognized by anti-PAF53 antibodies. It is as yet unknown if PAF51 is a degradative product, an alternatively spliced isoform, or a post translationally modified form of PAF53. PAF49 however, is not detected by anti-PAF53 antibodies thus it appears to be a distinctively different protein (68).

PAF53 is associated with RNA polymerase I purified from exponentially growing 3T3 cells but not with RNA polymerase from quiescent NIH3T3 cells. In addition, antibodies to PAF53 block specific, but not random, transcription from the rDNA promoter. These observations suggest that PAF53 is not involved in template binding, nucleotide incorporation, polymerization activity or elongation. Instead they suggest that PAF53 is required for initiation of specific transcription from the rDNA promoter (68). In vitro studies indicate that PAF53 has the potential to interact with the transcription factor, UBF (68), suggesting a role for this factor in the recruitment of RNA polymerase I to the initiation complex. However, the mechanism by which PAF53 contributes to the regulation of rDNA transcription remains to be elucidated.

4.2.3. rDNA Trans-acting Factors

There are at least two trans-acting factors required for efficient transcription of rDNA by RNA polymerase I. In mammals, they are referred to as SL-1 (selectivity factor 1) and UBF (upstream binding factor). Studies in Acanthamoeba, described below, have unambiguously identified a multimeric complex, TIF-IB, with properties and a functional role similar to SL-1. Interestingly, studies on transcription by yeast RNA polymerase I, have identified two complexes, described below, with properties similar to what might be considered to be a combination of SL-1 and UBF. Briefly, SL-1 is absolutely required for rDNA transcription in vitro (85,86). In contrast, UBF is not absolutely required for specific initiation on the rDNA promoter in vitro, although its addition to UBF-depleted extracts increases the efficiency of in vitro transcription in a dose dependent manner (87,88,89). In addition, overexpression of UBF1 in cell lines or primary cultures of cardiomyocytes is sufficient to directly increase transcription of a reporter for rDNA transcription (90), as well as the endogenous rRNA genes (R. Hannan and L. Rothblum, unpublished observation). Factors which bind to the core promoter element

There is evidence that several of the rDNA transcription factors may interact with the core promoter element. However, the experimental evidence accumulated from studies on the mechanism of rDNA transcription in Saccharomyces and Acanthamoeba suggests that SL-1, or rather its paralogues in those systems, must be considered the primary factor that interacts with the CPE.

Saccharomyces cerevisiae core factor: Genetic studies in yeast demonstrated that TATA-binding protein (TBP), the highly studied component of TFIID, was involved in transcription by RNA polymerase I and III, before biochemical studies had demonstrated that it was a component of any of the factors which interacted with the core promoter elements of the rDNA (91,92,93,94,95). Subsequent biochemical and genetic experiments confirmed that TBP was a component of the human, mouse, yeast, and Acanthamoeba rDNA transcription systems. Biochemical and genetic studies in yeast provided evidence for two transcription factors, referred to as UAF (upstream activation factor) and CF (core factor). UAF (discussed below) is a multiprotein transcription factor which consists of at least five proteins. Both biochemical and genetic analyses confirm that CF is also a multiprotein transcription factor, and it consists of at least three proteins, Rrn6p, Rrn7p and Rrn11p (96,97). CF interacts with the core promoter element, but does not by itself form a stable DNA-protein complex. However, in the presence of UAF, which forms a stable complex with the upstream element (98). CF becomes committed to the template and directs the initiation of transcription. It is not clear, at this time, whether TBP is a component of CF as suggested by Lin et al. (99) or if TBP is a "bridge" between CF and UAF (100).

Acanthamoeba TIF-IB: Studies on rDNA transcription have demonstrated that one protein, TIF-IB, is the TBP-containing transcription factor that binds the rDNA promoter to form the committed complex (86,101,102). While, TIF-IB has not been cloned, it has been purified to homogeneity and its interactions with the rDNA promoter have been studied extensively. TIF-IB consists of TAFIs of 145, 99, 96, and 91 kDa as well as TATA-binding protein. Site-specific cross-linking experiments demonstrated that the TIF-IB contacts mapped from -19 to -66 (86,102,103). Interestingly, TBP, as part of TIF-IB, only made contact with promoters derivatized between -38 and -43. This site is 22 bp upstream of the bend in the promoter induced by contact with TIF-IB, and consistent with the hypothesis that the DNA binding region of TBP may not be as involved in DNA-binding by TIF-IB as it is in TIF-ID (86).

Subsequent studies on the the interaction of TIF-IB and RNA polymerase I with the A. castellanii promoter demonstrated that TIF-IB could direct transcription from a core promoter terminated at -6. Additional cross-linking experiments demonstrated that, when assayed in combination, both TAFI96 and the 133 kDa subunit of RNA polymerase I interacted with the region between -1 and -7. This region contains a conserved sequence which is present in a large number of rRNA promoters: n(g/r)(g/r)Gt(T/A)aTnTAgGG(a/g)gAn (A=+1). This leads to the hypothesis that the CPE of RNA polymerase I promoters contains both an upstream site that interacts with TIF-IB and an Inr-like element that strengthens the interaction between TIF-IB and the promoter (104). Interestingly, these observations, using state-of- the-art techniques and highly purified reagents are consistent with conclusions drawn in earlier studies using various promoter mutants (105 and references therein).

SL-1: The mammalian homologue of TIF-IB is referred to as SL-1. Like TIF-IB, SL-1 is a "basal" rDNA transcription initiation factor capable of directing multiple rounds of RNA polymerase I recruitment to the rDNA promoter. SL-1 was first identified and its subunits cloned in humans (106). Subsequently, homologous proteins have been identified in rat (SL-1) (107), mouse (TIF-IB, factor D) (101,108,109,110), and frog (Rib1) (111). SL-1 exists as a complex containing the TATA-binding protein (TBP) and at least three RNA polymerase I specific TBP associated factors (TAFIs) (106, 109).

As mentioned, TBP is a the subunit common to the fundamental transcription factors for all three nuclear transcription systems. In every case, the functional regions of TBP are localized to the highly conserved C-terminal domain, which consists of two copies of an imperfect repeat of 61-62 amino acids. This region is sufficient for the correct assembly of SL-1 and is necessary for transcriptional activity (112).

In contrast to TBP, the RNA polymerase I TAFs exhibit no homology to the TAFs involved in transcription by RNA polymerase II or III (86,106). In addition, the molecular masses of the RNA polymerase I TAFs differ between species for example, the human TAFIs are 110, 63, and 48 kDa (106), and the mouse TAFs 95, 68 and 48 kDa (108,110). TAFI48 exhibits the highest degree of conservation among species and contains two stretches near the N-terminus which are imperfectly repeated at the C-terminal (106,110). The largest TAFIs, mouse TAFI95 and human TAF1110, are the least conserved, demonstrating only 66% identity at the amino acid level. The second largest TAFs also differ. Human TAFI63 contains an unique 40 amino acid N-terminal extension and mouse TAFI68 has 66 unique amino acids in its C-terminal region. Both proteins contain two putative Zn2+ fingers, although mTAF168 may have a third Zn finger (106,110). To date, the 5 end of the cDNA for human TAFI68 has not been cloned (106).

The mechanism determining the association of TBP with the TAFIs rather than other TAFs to form TFIID, TFIIIB and SNAPc is not known. In vitro experiments demonstrated that when TBP is bound to any of the TAFIIs, it will no longer bind the TAFIs, and vice versa (106). These studies suggest a mutually exclusive binding and that this binding specificity will direct the formation of the promoter- or polymerase-selective TBP-TAF complexes (106). SL-1 activity could be reconstituted from the three human TAFIs (106). However, functional mouse SL-1 could not be reconstituted from recombinant mouse TBP and TAFIs, although they did form a high molecular weight complex (109), and could complex with the human TAFIs. This observations suggests that SL-1 may contain additional components, or that additional factors may be required to mediate the interaction between SL-1 and RNA polymerase I.

Formation of a stable SL-1 complex involves multivalent contacts between TBP and the TAFs as well as between the individual TAFs (106,109). These contacts appear to be conserved, as the interactions between the mouse TAFs and human TBP appears to be the same as that observed with the human TAFs (106). How the SL-1 complex interacts with the rDNA promoter and thus mediates rDNA transcription is currently under investigation.

The original studies on human SL-1 suggested that, by itself, SL-1 was not a DNA-binding protein (113). Human SL-1 did not footprint the human rDNA promoter. However, the addition of SL-1 to a UBF footprinting assay resulted in a 5 extension of the UBF footprint (113,114,115). Rat SL-1 was found to be sufficient to drive transcription from a promoter that extended from -37 to +164 (44), but footprinted over the UPE of the rat rDNA promoter (44). It was noted that the position of that footprint was similar to the "extension" of the human UBF footprint by human SL-1 (44). Interestingly, mouse SL-1 yielded a "disperse" footprint, but that footprint included the CPE (113,116). This binding was abolished by a mutation at -16 with respect to +1. The same mutation results in a decrease in rDNA transcription, confirming that mouse SL-1 binding is required for promoter recognition and transcription initiation (116). Studies on the purified, recombinant hTAFIs suggest that hSL-1 is a DNA-binding protein. One study reported that both hTAFI110 and hTAFI63 bound to the rDNA promoter (117), while a second paper demonstrated that human TAFI48 and TAFI63 (or mouse TAFI68) can bind to DNA (106,118).

Although it has not been tested, experiments examining the interactions between the core promoter binding factors suggest an ordered strength of DNA-binding, mSL-1>rSL-1>hSL-1. This may explain the relative importance of UBF in these various transcription systems. In this regard it should be noted that A. castellanii TIF-IB has an very strong affinity for its promoter (kDa of 30x10-9), and it is a matter of discussion if there is a UBF-like activity in that organism (119).

The interaction between SL-1 and UBF appears to be critical for UBF-dependent activation of transcription (115). It has been suggested that basal rDNA transcription requires SL-1 and the CPE, while elevated levels of transcription also require UBF and the distal promoter elements (44). Coimmunoprecipitation studies demonstrated that SL-1 can bind to UBF in the absence of DNA (77,120). UBF antibody depleted extracts of SL-1 activity but not TFIIIB activity, demonstrating that this is a specific interaction (77, 120). In vitro studies suggest that this interaction may be mediated by the SL-1 components, TBP and TAFI48 (117). However, the domains of the proteins involved are as yet undetermined. These studies suggest that SL-1 serves to communicate between UBF and RNA polymerase I.

The interaction between SL-1 and the rDNA promoter is species specific, e.g. human SL-1 is required for transcription from the human rDNA promoter (3,4,121,122). In contrast, UBF and RNA polymerase I are, at least to some degree, interchangeable between species (116,123). For example, extracts prepared from primate cells that actively transcribe the human rRNA promoter fail to initiate transcription from a rodent rRNA promoter, but will do so when supplemented with either mouse or rat SL-1 (4,44,116,124). However, this only extends so far, a similar study showed that frog and human extracts could not be "reprogrammed" to accurately transcribe one anothers genes (125). [Interestingly, mouse extracts would initiate transcription on the Xenopus promoter but at +4 (3,126). Another publication reported that rat SL-1 can utilize primate RNA polymerase I transcription machinery (127)] The subunit of SL-1 responsible for reprogramming has not been identified. The two largest TAFs are the least conserved thus, are the most likely candidates for conferring species specificity. However, this needs to be established. UBF

UBF has been cloned from humans (84), mice (128), rats (123,129) and Xenopus (130,131). In Acanthamoeba a 125kDa protein has also been identified which has some functional characteristics similar to UBF (3). UBF is a highly conserved protein. Human and rat UBF1 are 97% identical, and there is only one, nonconservative amino acid change between the two (128). Even between mammals and Xenopus there is a 73% conservation of the amino acids overall. This conservation becomes 90% when the N-terminal domains are compared (2).

Purified UBF consists of two polypeptides, UBF1 and UBF2, the sizes of which vary depending upon the species (128,129). The human and rodent UBF isoforms are 97 kDa (UBF1) and 94 kDa (UBF2), whereas in Xenopus laevies they are 85 and 83 kDa (18,129). The mouse UBF gene consists of 21 exons extending over 13 kb (128). Transcription of this gene generates a single transcript which results in the mRNA for UBF1 (764 amino acids), or, due to alternative splicing at exon 8, UBF2. The result of the splicing event is that UBF2 mRNA contains an in frame deletion of 37 amino acids in HMG box 2 (128,129). In contrast, Xenopus UBF1 and UBF2 are generated by transcription from two different genes (130), and there is evidence for additional UBF genes or pseudogenes in the Xenopus genome (131). The xUBF1 gene encodes a protein which has 93% identity to xUBF2 and contains an insert of 22 unique amino acids between HMG box 3 and 4 (3,130,131).

Both isoforms of UBF can bind to the rDNA promoter, form homo or hetero dimers in solution and bind to synthetic DNA cruciforms with a similar affinity (46,85,132). However, UBF1 has been shown to be a more potent activator of transcription in vitro. UBF2 is 1/3 to 1/10 as active as UBF1 (87,132). This suggests that the activity difference is due to the alteration of HMG box 2 as found in UBF2. One study suggests that UBF2 may have a unique function in the formation of loops between the enhancers of the gene promoter (133). However, when COS cells overexpress p21h-ras they only express UBF1 and these cells are viable (132). This suggests that, at least in COS cells, UBF2 is not essential for cell viability (132). One report found that the ratio of UBF1/UBF2 in a cell reflected the growth state of the cell, i.e. the ratio of UBF2/UBF1 was approximately two in stationary 3T3 or MH134 cells (128) and the ratio approached 1 upon nutritional upshift. It has also been reported that the ratio of UBF1 to UBF2 changes with development. For example, during differentiation of F9 cells or mouse embryogenesis the ratio of UBF1 to UBF2 decreases both at the mRNA and protein level (132). Cumulatively, these data raise the possibility that UBF2 may have an as yet unrecognized function(s).

The mechanism(s) by which UBF expression is regulated is unknown. It is clear that the gene is subject to both positive and negative regulation (9,11,134). However, further analysis is required in order to determine the direct mechanisms by which UBF expression is regulated.

UBF Structure: The dominant structural elements of UBF are the HMG boxes which are similar to the DNA binding domain of the chromosomal high mobility group proteins 1 and 2 (HMG-1, HMG-2) (89). Other proteins belonging to this family include T-lymphocyte receptor a-enhancer factor, sex determining region Y protein, mitochondrial transcription factor and the yeast mitochondrial nonhistone protein NHP6 (3).

HMG boxes are usually 80 amino acids long (4). The number "found" in UBF depends upon the stringency of homology to the consensus sequence used for classification. Thus, there are reports of four to six HMG boxes in UBF (75,89,128). For example, when the definition of a HMG box is applied stringently, mammalian UBF and Xenopus UBF have four and three HMG boxes, respectively (4,75,135,136). However, many papers cite six and five boxes respectively. Interestingly, each HMG box appears to play a specific role (131,135). An HMG box cannot be replaced with another box from the same protein. However, they can be replaced with the same HMG box from a distantly related species. UBF requires the correct number and order of HMG boxes for it to function in transcript (135). The finding that Xenopus UBF failed to activate transcription in an extract from human cells may be explained by the observation that Xenopus UBF lacks HMG box3 as found in human UBF (131).

Functions other than DNA-binding have been assigned to the HMG boxes. Part of the UBF nucleolar localization signal is found in the NH2-terminal region, which includes HMG box 1, (137,138). In addition, nuclear transport, requires a short 24 amino acid sequence near HMG box 5 as well as the CO2H-terminus (137). Similar to other HMG-like proteins UBF contains a highly acidic CO2H-terminal domain and an NH2-terminal dimerization domain. The acidic CO2H-terminal domain of UBF consists of a stretch of 89 amino acids of which 68% are Glu or Asp, 25% are serines, and 7% glycine (89). The acidic regions are interrupted by conserved serine-rich blocks (3). The NH2-terminal dimerization domain contains two short regions that are hypothesized to form amphipathic helices similar to a helix-loop-helix motif. The dimerization domain is also required for optimal DNA binding along with at least one HMG box (43,136), and additional HMG boxes appear to stabilize DNA-binding (43,139).

Mechanism of UBF Action: The action of UBF depends on the formation of homo- and/or hetero-dimers (3,136,148) and its binding to DNA, via the minor groove (140). Various manuscripts have reported that UBF binds to the CPE, UPE, spacer promoters and the enhancer repeats in the intergenic spacer (44,46,47,89,123). In addition, Xenopus UBF can bind on each side of the promoter proximal terminator (3). Interestingly, footprinting analysis has demonstrated that the DNase footprint obtained with UBF depends on the rDNA promoter being footprinted, and is independent of the origin of the UBF used in the assay (18,46,89,123). As discussed above, this would suggest that rDNA promoters share underlying structural similarities despite their sequence differences. In general, UBF footprints the rDNA promoter in the UPE, from ~ -50 to ~ -130 (89). However, UBF can also protect the CPE, from ~ -45 to ~ +20 (43, 89,123). As discussed previously, a mutation at either the guanine at -16 or -7 eliminates promoter activity (41,44,89), but did not affect UBF binding to the DNA (41). As mentioned, the interaction of UBF with SL-1 results in an extension of the UBF footprint (44,113,114,115,141). This is believed to be part of the mechanism by which UBF facilitates the generation of the pre-initiation complex on the promoter.

Initially it was reported that UBF bound predominantly to a GC rich consensus sequence (4). However, recent studies suggest that UBF may recognize a specific DNA structure, such as synthetic DNA cruciforms, four way junctions and tRNA, rather than a sequence (2,132,137,139). The domains of UBF required for DNA binding and the DNA binding sequence recognized are controversial. UBF with every HMG box deleted, except HMG box 1, is able to bind DNA (as long as the dimerization domain is present). The addition of HMG boxes increases the strength of DNA binding (4,89,136). These results may be explained if UBF binding to DNA is the result of a summation of multiple HMG box-DNA contacts. This would also lessen the requirement that any single sequence (DNA recognition site) be stringently maintained.

UBF binds to DNA and by inducing folding and bending shortens the DNA contour by ~190 bp (141,142). This generates a disk-like UBF-DNA complex which has been referred to as an enhancersome (2,43,141). The enhancersome contains a low-density protein core around which the DNA loops, probably by in-phase bending. UBF can force the DNA to generate a 3600 loop with a diameter of 19 nm (141). In this structure the HMG boxes would interact with the promoter in a colinear manner. This model is consistent with those reported by Xie et al. (41). In that study spacing changes of half a helical turn significantly decreased rDNA promoter activity, while a full turn only mildly affected promoter activity (41). Thus it would appear that UBF binds to the DNA and bends it. It is not clear if UBF binds cruciform DNA and then bends the remaining DNA to form an enhancersome or if UBF binds to the DNA, bends it facilitating the formation of a cruciform. In either case, bending the rDNA promoter would make it possible for the two bound SL-1s to interact generating the pre-initiation complex model illustrated in figure 2.

Regulation of UBF Specific Activity: Cells can modify the specific activity of UBF by at least two different mechanisms: i) phosphorylation; and ii) sequestration. UBF is a phosphoprotein (13), especially the CO2H-terminal tail which can be extensively phosphorylated (88). The ability of UBF to trans-activate the rDNA promoter is reduced when UBF is treated with phosphatase (13) or when the phosphorylated CO2H-terminal tail is deleted (88). This suggests that phosphorylated UBF is the more active form of the transcription factor. UBF contains numerous consensus motifs for characterized kinases including CKII and MAPK. Interestingly, treatment of UBF with CKII in vitro increases UBF phosphorylation (13,88,139) and enhances transcriptional activity (88). However, it is not known whether or not UBF is an endogenous substrate of CKII in vivo. In many models of growth the phosphorylation status of UBF has been demonstrated to correlate positively with the rate of rDNA transcription (13,88,143). These experiments are discussed in more detail in section 5.4.2.

The activity of UBF can also be down-regulated by the interaction of UBF with the product of the retinoblastoma susceptibility gene, Rb110 (10,144). Studies indicate that the ability of UBF to transactivate the rDNA promoter is severely compromised when UBF is sequestered, either directly or indirectly, by Rb110. Moreover, the physiological relevance of this mechanism in the regulation of rDNA transcription has been demonstrated in vivo (10). These experiments are discussed in greater detail in section 5.4.2.

Cellular Distribution: In some cell types the localization of UBF can change during the cell cycle. For example, during early S phase there is an increased association of UBF, RNA polymerase I and SL-1 within the nucleolus (21). The increase in UBF association with nucleoli may be due to an increase in the ability of UBF to compete with the histones for binding to the rDNA. Typically, UBF localizes to the FC and DFC of the nucleolus where it forms small bead like structures in a folded filament pattern (22). This distribution is sustained during the G2 phase when the cells are actively transcribing the rDNA (23). At the end of G2, when rDNA transcription is "shut off", UBF, RNA polymerase I and SL-1 accumulate in the mitotic NORs forming a few intensive spots on the chromosomes (21). Ku/E1BF

Ku/E1BF was originally detected as an human autoantigen reacting with antibodies from patients with rheumatic disorders and has now been widely identified in a number of species (3). Ku/E1BF exists as a heterodimer of two polypeptides, 70 and 86 kDa polypeptide (145,146,147). Ku/E1BF tends to bind DNA in a non-specific manner, while recent studies have shown that it binds the rDNA promoter with high specificity (145,148,149). Interestingly, when Ku/E1BF is added to cell-free transcription assays it can affect the rate of rDNA transcription (to be discussed below).

Ku/E1BF has been shown to be the DNA-binding component of the DNA-dependent protein kinase (DNA-dependent PK) (150,151). DNA-dependent PK is a nuclear, serine/threonine protein kinase consisting of a 350 kDa catalytic subunit and Ku/E1BF. The enzyme is most active when bound to DNA, a process dependent on Ku/E1BF. To date DNA-dependent PK has been shown to be important in various cellular processes such as, cell signaling, DNA replication, RNA polymerase II transcription activation and DNA repair. CPBF

Interestingly Ku/E1BF has been shown to interact with another potential rDNA transcription factor, CPBF (core promoter-binding factor) (3). CPBF is a rDNA binding protein, which has been isolated from both rat mammary adenocarcinoma ascites and HeLa cells. CPBF purifies as two polypeptides of 44 and 39 kDa (152). The 44 kDa peptide binds to Ku/E1BF (146). Both CPBF peptides specifically interact with the rDNA core promoter sequence resulting in trans-activation of the rDNA promoter in vitro (152). Moreover, CPBF and Ku/E1BF function synergistically to enhance RNA polymerase I transcription (146).

CPBF has been found to be the rat homologue of human USF which also consists of two peptides, 44 and 43 kDa (153). USF is a basic helix-loop-helix zipper, DNA binding protein which specifically binds E-boxes in genes transcribed by RNA polymerase II. Interestingly, USF and CPBF bind to the same E-box in the rat rRNA promoter suggesting a possible mechanism for their action on rDNA transcription. Oligonucleotides to the E-box sequence inhibit rDNA transcription possibly by preventing USF/CPBF binding to the DNA (154). Topoisomerases

Topoisomerases are enzymes which modulate DNA topology by catalyzing cleavage-rejoining reactions of the phosphodiester bonds. There are two classes of topoisomerases, I and II. Topoisomerase II includes both a(170 kDa) and b (180 kDa) isoforms (27,155). Topoisomerase may act as a swivel, relieving torsonal stress generated during transcription. This would allow for a rotation of the transcribed DNA segments without having to turn any other part of the DNA or the transcription ensemble (155,156).

Both topoisomerase I and II are nuclear enzymes. Topoisomerase I and II aare found in both the nucleoplasm and nucleolus. Topoisomerase II bis exclusively localized in the nucleolus (20,157,158,159,160). Topoisomerase I preferentially associates with actively transcribed regions of chromatin, and has also been implicated in the regulation of rDNA transcription (159,160). Topoisomerase I has been demonstrated to be required for rDNA transcription and replication in yeast (157,161). p16

p16 is an HMG-like, DNA-binding protein isolated from Novikoff hepatoma ascites and Hela cells (162). p16 binds to the oligo d(A).d(T) tracts found both within the UPE (-620 to -417) and external transcribed spacer (+352 to +525) of the rat rRNA gene. p16 was demonstrated to stimulate rDNA transcription in a dose dependent and saturable fashion when either of those sites was in cis with the target promoter. To date, this factor has not been cloned, thus, further details on the nature and mechanism of the interaction between p16 and the rDNA transcription apparatus are not available.

4.3. Formation of Preinitiation Complexes

The transcription cycle involves four distinct steps: i) initiation; ii) promoter clearance; iii) elongation; and iv) termination (4). Initiation involves the assembly of the pre-initiation complex on the rDNA promoter, isomerization of the closed pre-initiation complex to form an initiation competent open complex and, finally, generation of the first phosphodiester bond (84). Once the first bond is formed, and the promoter is cleared, an alteration occurs in RNA polymerase I conformation which commits the enzyme to undergo RNA chain elongation (163). Elongation involves catalyzing the processive addition of ribonucleotides to the 3 end of the growing RNA chain until specific attenuation or termination signals are encountered. Lastly, transcription is terminated and the product of the polymerase released from the template (84).

The preponderance of evidence is consistent with the model that regulation of rDNA transcription occurs at the level of formation of the initiation complex (163,164). Once formed, the initiation complex is quite stable and may remain in place on the promoter, even as initiation rates vary widely, suggesting that the number of complexes is determined by limiting the amounts of a transcription factor (164). Once these complexes form, the actual rate of transcription will depend upon the ability of RNA polymerase I to recognize the complex and initiate transcription. The steps which may be involved in this regulation are dependent on the system being investigated.

A working model for initiation can be based on studies from Acanthamoeba castellani. In this system, TIF (SL-1) binds to the promoter, in the absence of either UBF (or a UBF-like factor) or RNA polymerase I, and causes a distinctive DNAse I footprint (163,165). TIF then recruits RNA polymerase I by protein-protein interactions. RNA polymerase I binding results in an extension of the TIF footprint to include the +1 site (164). Upon the addition of nucleotide triphosphates, elongation occurs and the RNA polymerase I footprint moves down the template leaving the original TIF footprint behind (163).

In mammalian and Xenopus transcription systems, initiation is also believed to be a multistaged process. In mammals, initiation involves SL-1 binding to the core promoter a process which is facilitated by UBF, and possibly TIF-IC, to form a stable pre-initiation complex (166). This complex is stable for a number of rounds of transcription and able to recruit RNA polymerase I and the RNA polymerase I associated factors, such as TIF-IC, TIF-IA and PAF53, to form the second pre-initiation complex. The result of these steps is a complex which, with the addition of ATP/CTP (mouse; GTP/CTP in human) and further NTPs, becomes an initiation competent complex (166,167). The complex is now ready for elongation during which time RNA polymerase I moves past the initiation complex and leaves the pre-initiation complex intact.