A Novel Transcription Initiation Factor (TIF), TIF-IE, Is Required for Homogeneous Acanthamoeba castellanii TIF-IB (SL1) to Form a Committed Complex*

The fundamental transcription initiation factor (TIF) for ribosomal RNA expression by eukaryotic RNA polymerase I, TIF-IB, has been purified to near homogeneity fromAcanthamoeba castellanii using standard techniques. The purified factor consists of the TATA-binding protein and four TATA-binding protein-associated factors with relative molecular weights of 145,000, 99,000, 96,000, and 91,000. This yields a calculated native molecular weight of 460,000, which compares well with its mass determined by scanning transmission electron microscopy (493,000) and its sedimentation rate, which is close to RNA polymerase I (515,000). Both impure and nearly homogeneous TIF-IB exhibit an apparent equilibrium dissociation constant of 56 ± 3 pm. However, although impure TIF-IB can form a promoter-DNA complex resistant to challenge by other promoter-containing DNAs, near homogeneous TIF-IB cannot do so. An additional transcription factor, dubbed TIF-IE, restores the ability of near homogeneous TIF-IB to sequester DNA into a committed complex.

In eukaryotic transcription, a fundamental transcription initiation factor marks the promoter for subsequent events leading to the recruitment of RNA polymerase (1). TFIIIB serves this role for RNA polymerase III transcription (2), TFIID for RNA polymerase II (3), and for ribosomal RNA genes, this fundamental factor is dubbed TIF-IB, 1 SL1, factor D, Rib1, core factor, or TFID (reviewed in Refs. 1 and 4 -8). These factors all contain a common subunit, the TATA-binding protein (TBP) (9,10), which is associated with a variable number of polymerasespecific subunits, the TATA-binding protein associated factors (TAFs). In human SL1, TBP is associated with three TAFs with M r ϭ 110,000, 63,000, and 48,000 (9); in mouse, the TAF I s have M r ϭ 95,000, 68,000, and 48,000 (11).
Lower eukaryotes contain a slightly modified factor. In Saccharomyces cerevisiae, a genetically identified complex called core factor or CF, made up of Rrn6p, Rrn7p, and Rrn11p, appears to function on the core promoter in a manner similar to vertebrate TIF-IB/SL1 (12)(13)(14). Surprisingly, none of the cloned Rrn6/7 or -11 genes from the S. cerevisiae complex show any sequence similarity to the vertebrate TAF I s, and TBP is not tightly associated with the TAF I s. In Acanthamoeba castellanii, a single transcription factor, TIF-IB, is necessary and sufficient to form a complex on the promoter and recruit RNA polymerase I for multiple rounds of transcription (15)(16)(17)(18)(19).
The ability of TIF-IB to form a stable complex on the core promoter varies considerably from species to species. At one end of the spectrum, human SL1 can bind only very weakly or not at all to the rRNA promoter by itself, based on its ability to mediate transcription in vitro in the presence of only RNA polymerase I (20). Instead, an accessory factor, upstream binding factor (UBF), binds first to an upstream promoter element and apparently aids the binding of SL1 either by interacting with it (21) or by altering the structure of the DNA in the region bound by SL1 (22). The UBF footprint persists following committed complex formation, showing it remains in the complex (21). In Xenopus laevis, Rib1 similarly cannot form a stable complex without UBF, but for a different reason; in addition to its ability to alter the DNA structure of the promoter (22), in a DNA-independent mechanism UBF prevents the dissociation of TBP from the rather unstable Rib1 (23). In rat and mouse, UBF can also stimulate TIF-IB binding. However, in these species UBF is not required for stable association of the TBPcontaining factor with the promoter, and thus TIF-IB is sufficient for specific initiation in these systems (6,24). A. castellanii TIF-IB is at the farthest extreme of the spectrum of TBP⅐TAF I complexes in its ability to interact very strongly with the promoter in the absence of any other assembly or architectural proteins.
Recently, a distinct transcription factor dubbed upstream activation factor (UAF) has been genetically identified in Saccharomyces cerevisiae. Like UBF, this multiprotein complex functions when bound to the upstream promoter element, stabilizing binding of CF to the DNA (25). However, UAF can commit the template, but CF cannot. Curiously, UAF appears to act on CF via a bridging TBP molecule. The latter is reminiscent of one of the roles of Xenopus UBF, stabilizing the interaction of TBP with the TAF I s (23). Whether UBF or UAF homologs are obligatory parts of the committed complex in A. castellanii is not clear. We have shown that the DNA in the A. castellanii committed complex is not wrapped or looped (26) as in a UBF complex (22,27), and we cannot identify an upstream promoter element in vitro or a UBF or UAF footprint consistent with those found in vertebrates or yeast. TBP is a stable subunit of A. castellanii TIF-IB (28). Thus, A. castellanii TIF-IB appears to have some functions not found in the other factors.
In this paper, we describe a procedure to obtain A. castellanii TIF-IB in a nearly homogeneous form. We show that polypeptides with apparent molecular weights of 145,000, 99,000, 96,000, and 91,000 copurify with TBP and TIF-IB activity. The homogeneous TIF-IB was functionally tested. It is capable of driving specific transcription initiation in an in vitro system consisting of only TIF-IB and RNA polymerase I purified to near homogeneity. However, unlike TIF-IB from earlier stages of purification, homogeneous TIF-IB is incapable of forming as persistent a complex with the core promoter in a template commitment assay. Despite this finding, the apparent dissociation constant between TIF-IB and the promoter-DNA is identical between fractions capable and incapable of template commitment. The ability to commit the template can be restored by adding a partially purified factor, which we dub TIF-IE. 2

Specific Transcription Run-off Assay
The standard assay for TIF-IB was carried out in a final volume of 50 l containing a 500 M concentration each of ATP, CTP, and UTP; 25 M GTP; 5 Ci of [␣-32 P]GTP (NEN Life Science Products; 3000 Ci/ mmol); 100 mM KCl; 10 mM MgCl 2 ; 20 mM HEPES-KOH (pH 7.9); 10% (v/v) glycerol; 0.1 mM EDTA; 0.5 mM dithiothreitol; and 50 ng of linearized plasmid DNA, pEBH10/NdeI (15) or pAr6/HindIII. pAr6, which contains the rRNA promoter from Ϫ683 to ϩ219, was derived from pAr4 (29) by excision of the EcoRI-DdeI fragment encompassing the transcription initiation site, filling in its ends with the Klenow fragment of DNA polymerase I, and ligating it into the SmaI site of pUC8. The reactions were started by the addition of individual fractions (1-5 l) containing TIF-IB and 30 milliunits of heparin-Sepharose-purified RNA polymerase I (30). Incubation was at 25°C for 30 min. The reactions were terminated by the addition of 50 l of stop buffer containing 1 mg/ml proteinase K and 1% SDS, followed by incubation at 50°C for 60 min. Nucleic acids were precipitated by the addition of 200 l of 3 M ammonium acetate, 0.125 mg/ml linear polyacrylamide, and 750 l of 95% ethanol. Nucleic acids were pelleted in an Eppendorf centrifuge for 30 min at maximum speed. The pellets were washed with 1 ml of 75% ethanol; dried under vacuum; suspended in 5 l of 95% deionized formamide, 10 mM EDTA, 0.1% bromphenol blue, 0.1% xylene cyanol; and analyzed on a denaturing (7 M urea) 6% polyacrylamide sequencing gel, 0.5ϫ Tris borate-EDTA (31).
TIF-IB activity was determined in a run-off transcription assay by carrying out a titration of each pool containing TIF-IB. The specific radioactivity of GTP in the assay and the number of nmol of incorporated GMP were determined by simultaneous exposure of phosphor storage screens to a known volume of the reaction mixture containing [␣-32 P]GTP and the runoff products in the dried polyacrylamide gel, followed by quantification using ImageQuant version 4.1 software. One unit of TIF-IB activity is defined as the amount mediating incorporation of 1 nmol of [ 32 P]GMP into a 309-nucleotide run-off in the standard 30 min assay. The 309-nucleotide run-off product contains 76 G residues.

Template Commitment Assay
The minimum amount of template required to bind all the available TIF-IB in the reaction was determined for each template and used in the following protocol. Template (pAr6 and/or pEBH10) was preincubated with TIF-IB or TIF-IB plus TIF-IE under the standard assay conditions for 10 min, except RNA polymerase I was omitted. Since some of the experiments described here were done with very pure and dilute components, bovine serum albumin (0.5 mg/ml) was included in the preincubation mixtures to stabilize the protein components and help prevent them from binding nonspecifically to the walls of the reaction vessel. The second template or buffer was added, and preincubation continued for another 10 min. RNA polymerase I was added to start the RNA synthesis phase, which proceeded for another 30 min. Run-off RNAs were analyzed as described above.

Electrophoretic Mobility Shift Assay (EMSA)
Electrophoretic mobility shift assays were carried out as described (28), except the electrophoretic buffer was the Tris-glycine buffer used for protein gels minus the SDS (32), and electrophoresis was carried out at room temperature.

Estimation of Equilibrium Dissociation Constants and Kinetic Dissociation Rate Constants
An accurately known amount of promoter DNA (EcoRI/HindIII-cut pEBH10; Ϫ120 to ϩ80) was labeled with 32 P by fill-in with the Klenow fragment of DNA polymerase I, and its specific activity was estimated by spotting on DEAE filter paper discs (DE81, Whatman, Fairfield, NJ) and washing as for an RNA polymerase nonspecific assay (30). A fixed amount of TIF-IB was titrated with known amounts of this labeled DNA as described, except no competitor DNA was added (28), and the resulting mixture analyzed in an EMSA as described above. ImageQuant version 4.1 software was used to analyze the amount of free DNA and TIF-IB⅐DNA complex present at each DNA concentration, which was then plotted according to the method of Scatchard. The kinetic dissociation rate constant for the TIF-IB⅐DNA complex was estimated by forming the complex for 10 min, as above, and then adding a 100-fold molar excess of unlabeled promoter DNA for various time periods. In the experiment shown in Fig. 7, start times for each time point were staggered so that all the incubations ended simultaneously. At the end of the incubation, the samples were rapidly chilled on ice, immediately loaded onto the EMSA gel, and electrophoresed. Data were analyzed by Phosphor Imager analysis.

Polyacrylamide Gel Electrophoresis and Staining of Proteins
Proteins were precipitated with chloroform-methanol (33), resuspended in 1ϫ SDS loading buffer and electrophoresed through an SDS-10% or 7.5% polyacrylamide gel by standard methods (34). Gels were stained using Coomassie Brilliant Blue R-250 as described (31) or with silver (35).

Purification of RNA Polymerase I and TIF-IE
RNA polymerase I was purified from a whole cell extract as described by Spindler et al. (30). The TIF-IE used in the study described herein was separated from RNA polymerase I at the last step, glycerol gradient rate zonal sedimentation, but the yield is variable from successive preparations of RNA polymerase I. We have recently found a larger pool of TIF-IE in the BioRex70 fraction of TIF-IB, which can be further purified by heparine-Ultrogel A4R (IBF Biotechnics, Paris) and ratezonal sedimentation (data not shown). In both cases, TIF-IE is not yet homogeneous.

Preparation of Promoter-DNA-Sepharose 4B
Both strands of the A. castellanii rRNA promoter from Ϫ70 to Ϫ15 were chemically synthesized incorporating five point mutations (GϪ53A, CϪ51T, GϪ37T, CϪ22T, and GϪ20C) which increase promoter strength, presumably by increasing the binding strength to TIF-IB (36). Four base 5Ј-extensions (G residues at Ϫ70 and C residues at Ϫ15) were included so that the annealed oligonucleotide could be oligomerized. The ends of the oligomerized DNA were filled using the Klenow fragment of Escherichia coli DNA polymerase I and cloned into the HincII site of Bluescribe(Ϫ) (Stratagene, La Jolla, CA). A fragment containing four head-to-tail copies of the Ϫ70 to Ϫ15 promoter sequence were excised using EcoRI and HindIII, purified over Sephacryl S-500, and coupled to cyanogen bromide-activated Sepharose CL-4B as described (37), except unreacted cyanogen bromide-derivatized Sepharose was inactivated with 1 M Tris (pH 8).

Protein Concentration Assay
Protein concentration in early fractions was estimated using a modified Bradford microassay procedure (Bio-Rad) with bovine ␥-globulin as the standard protein according to the manufacturer's directions. For the final glycerol gradient-purified fraction, the protein concentration was estimated by silver staining.

Purification of TIF-IB
All steps were carried out at 0 -4°C unless otherwise noted. A crude nuclear extract was fractionated by ammonium sulfate precipitation; the TIF-IB-containing fraction (0.5-1.82 M (NH 4 ) 2 SO 4 ) also contained the components needed for transcription of RNA polymerase III-transcribed genes (28,38), while RNA polymerase I was found in the 1.82-3.56 M (NH 4 ) 2 SO 4 fraction (28). This fraction is called the "nuclear extract/pol III cut." DEAE-Sepharose Fast Flow Chromatography-The TIF-IB-containing fraction from 225 g (wet weight) of A. castellanii cells containing 1700 mg of protein and 22.5 units of TIF-IB activity was dialyzed against 100 mM KCl in HEG 20 (50 mM HEPES, pH 7.9, 0.2 mM EDTA, 20% glycerol, 1 mM dithiothreitol, 0.1 mM phenylmethanesulfonyl fluoride) to remove ammonium sulfate and convert the extract to 100 mM KCl. This was loaded onto a 50-ml DEAE-Sepharose fast flow column (2.5 ϫ 10.2-cm) equilibrated to 100 mM KCl in HEG 20 at a linear flow rate of 10 cm/h (0.5 column volume/h). This low flow rate is necessary to achieve optimal binding of TIF-IB to the exchange medium. The column was washed with 5 column volumes of HEG 20 containing 100 mM KCl at a linear flow rate of 25 cm/h. The column was developed with a 5-column volume linear gradient from 100 to 500 mM KCl in HEG 20 at 25 cm/h. Eighty fractions were collected and assayed for TIF-IB by a run-off transcription assay. TIF-IB reproducibly elutes with a peak at 280 mM KCl. RNA polymerase III is primarily in the unbound flowthrough, but a small amount elutes ahead of and overlapping with the TIF-IB. The active fractions were pooled and diluted to 150 mM KCl in HEG 20 .
BioRex Promoter-DNA Affinity Column Chromatography-The active fractions from the BioRex 70 column were pooled and dialyzed against H 20 EG 10 (20 mM HEPES, pH 7.9, 0.2 mM EDTA, 10% glycerol, 1 mM dithiothreitol, 0.1 mM phenylmethanesulfonyl fluoride) containing 150 mM KCl. The pool was prepared for loading onto the DNA affinity column by adding MgCl 2 to 10 mM, salmon sperm DNA to 0.09 mg/mg of protein, and sufficient 2-fold concentrated H 20 EG 10 to maintain a constant buffer concentration. This final set of conditions was found to be optimal for binding to promoter DNA by EMSA. It is the same as the optimal set of conditions for transcription, except salmon sperm DNA was added to compete nonspecific DNA-binding proteins, and the KCl concentration was increased from 100 to 150 mM. The latter is allowable because of the point mutations, which specifically increase TIF-IB binding to the promoter (see "Preparation of Promoter-DNA-Sepharose 4B"). Five ml of promoter-DNA-Sepharose 4B was mixed with the prepared BioRex 70 pool and incubated with gentle agitation at room temperature (22°C) for 1 h. The affinity medium was then loaded at 4°C into a 1.5 ϫ 2.8-cm column and the supernatant recirculated through the column at a linear flow rate of 5 cm/h until it had passed over the column three times. The column was washed with 5 column volumes of H 20 EG 10 containing 150 mM KCl, 10 mM MgCl 2 and 0.1% Nonidet P-40 (v/v) at 5 cm/h (1 column volume/h). The column was developed with a 5-column volume gradient from 150 to 1000 mM KCl in the same buffer at 5 cm/h. 40 fractions were collected and analyzed as above. TIF-IB eluted as a peak centered at 460 mM KCl. Promoter-DNA affinitypurified TIF-IB could be concentrated using a Microcon-10 microconcentrator according to the manufacturer's instructions (Amicon, Beverly, MA) or by diluting to 150 mM KCl with H 20 EG 10, binding to a 1-ml (0.9 ϫ 1.6 cm) BioRex-70 column and eluting with a step gradient of 600 mM KCl in H 20 EG 10 .
Rate Zonal Sedimentation in Glycerol Gradients-17.5-35% (v/v) glycerol gradients in 50 mM HEPES, pH 7.9, 0.2 mM EDTA, 1 mM dithiothreitol, 0.1 mM phenylmethanesulfonyl fluoride, 0.1% Nonidet P-40 (v/v), 100 mM KCl were prepared in 13 ϫ 51-mm polyallomer centrifuge tubes. 200 l of the concentrated promoter-DNA affinity column pool was layered on the top of each gradient and centrifuged for 15 h at 47,500 rpm in a Beckman SW50.1 rotor at 4°C. The gradients were fractionated by pumping Fluorinert FC-40 (ISCO, Lincoln, NE) into the bottom of the tubes using an ISCO density gradient fractionator. 23 fractions were collected and assayed as above. The peak of TIF-IB is reproducibly in fractions 13 and 14, which is just above the position to which A. castellanii RNA polymerase I sediments in the same gradients (centered at fraction 15). The yield of TIF-IB is based upon the sum of the activities in the most active fractions. The amount of protein in these fractions is too low to estimate accurately using a modified Bradford assay. Instead, protein concentrations were estimated from silver staining and densitometric scanning.

RESULTS
TIF-IB was purified from a nuclear extract using DEAE-fast flow, BioRex 70, and promoter-DNA affinity chromatography followed by rate zonal sedimentation in a glycerol gradient ( Fig. 1 and Table I). Because there are inhibitors in the nuclear extract (28) leading to an apparent increase in activity after the first purification step, the DEAE-Sepharose fast flow fraction was assigned as 100% in Table I. The promoter-DNA affinity column alone purified TIF-IB 226-fold. The factor at this point is nearly homogeneous (Fig. 2). Even after the promoter-DNA affinity column step, the polypeptides that make up TIF-IB can be readily discerned (Fig. 2B), along with a number of contaminating polypeptides that are present in variable amounts from preparation to preparation (cf. Fig. 2B and Fig. 3B, lane L). In the 7.5% gel shown in Fig. 2, the TBP is run off the bottom of the gel; however, the presence of TBP in the purified TIF-IB as well as in the promoter-DNA⅐TIF-IB complex has been demonstrated in previous studies (26,28). We have also found that multiple rounds of promoter-DNA affinity chromatography did not significantly improve the purity of TIF-IB. The glycerol gradient-purified TIF-IB had a specific activity of 211 units/mg of protein and was purified approximately 16,000-fold from the TIF-IB-containing nuclear extract, or 55,000-fold from the whole cells (39).
Subunit Composition of TIF-IB-We scaled up the amount of TIF-IB loaded on each glycerol gradient and stained the gel with Coomassie Blue (Fig. 3B), which yields a better estimate of the relative stoichiometries of the polypeptides present in the purified protein and avoids artifacts of silver staining (see below).
Six major polypeptides are present in the peak of TIF-IB (Fig. 3B) whose amounts correspond with the activity estimated from the transcription run-off assay (Fig. 3A). The relative molecular weights of these polypeptides are 145,000, 99,000, 96,000, 91,000, 32,500, and 31,000. The latter two polypeptides have been shown by Western blotting to be TBP (M r ϭ 32,500) and the degradation product of TBP (M r ϭ 31,000) (28). The stoichiometry of the four high molecular weight polypeptides appears to be 1:1:1:1 in the Coomassiestained gel; the 145-kDa polypeptide stains less efficiently with silver (cf. Fig. 2B). This stoichiometry is constant from preparation to preparation, suggesting that all four polypeptides are independent components of TIF-IB, not degradation products of each other. Furthermore, site-specific photo-cross-linking shows that the 145-and 96-kDa TAF I s cannot derive from any of the higher molecular weight subunits by proteolysis (26). Polypeptides of 140,000, 114,000, 112,000, and 58,000 are also present but are in very low stoichiometry, and these are not consistently observed in successive preparations. In addition, the 58-kDa polypeptide does not exactly correlate with the TIF-IB activity. We sometimes observe a faintly stained polypeptide of about 185 kDa. This has been observed in multiple preparations but is always in low amounts. However, we cannot rule out the possibility that one of the lower molecular weight polypeptides is actually a degradation product of this 185-kDa polypeptide. We conclude that TIF-IB has a subunit architecture with TAF I s of 145, 99, 96, and 91 kDa and TBP, most likely in equal stoichiometry (see "Discussion").
Purified TIF-IB Cannot Commit a DNA Template-A. castellanii TIF-IB and some metazoan counterparts are capable of forming complexes on rRNA promoters (16,40) that are stable through multiple rounds of transcription. Once TIF-IB is bound to a template, the factor is committed to it and will not be released to form a preinitiation complex on a second template added subsequently. Thus, if template is added to an in vitro transcription system in an amount sufficient to bind all of the input TIF-IB, subsequently added templates are not transcribed, yielding an assay for stable complex formation (template commitment assay). In contrast, Xenopus (41) and Saccharomyces cerevisiae (25) require additional transcription factor(s) for template commitment. While impure preparations of A. castellanii TIF-IB can commit the template (16), the ability of nearly homogeneous TIF-IB to do so had not been tested.
Promoter-DNA affinity-purified TIF-IB, like less pure fractions (16), is capable of template commitment (Fig. 4, lanes  1-4). The pAr6 template (DNA A) produces a 240-nucleotide run-off RNA (Fig. 4, lane 1), and the pEBH10 template (DNA B) produces a 309-nucleotide run-off plus a pause product at about 270 nucleotides (Fig. 4, lane 2). When equal amounts of both templates are present in the preincubation, each template should be transcribed in equal amounts. As expected, after correction for the number of potentially labeled nucleotides in the two products, pEBH10 is transcribed at its expected level (Fig. 4, lane 3). However, when only pAr6 was present with a Protein concentrations were determined as described under "Experimental Procedures." b Chosen as the starting point for yield calculation because of inhibitors in the nuclear extract.

FIG. 2. Elution profile of TIF-IB from the promoter-DNA affinity column.
A, the specific run-off transcription product phosphor image and corresponding fraction number, load (L), and flow-through (FT) are indicated. B, a 7.5% silver-stained SDS-polyacrylamide gel of the fractions whose activities are shown in A. The relative molecular weights, in thousands, of the polypeptides identified to be components of TIF-IB are marked on the right. TIF-IB in the first preincubation and pEBH10 was added during the second preincubation, transcription of pEBH10 was drastically reduced (Fig. 4, lane 4). The first template, pAr6, remained fully active (Fig. 4, lane 4), demonstrating the formation of a strong committed complex on pAr6.
In contrast, glycerol gradient-purified TIF-IB does not produce a committed complex on the template. When pAr6 alone is in the primary preincubation and pEBH10 is added only during the second preincubation, the activity of pEBH10 is at its expected level (Fig. 4, lane 8), essentially identical to when both templates are present simultaneously in the first incubation (Fig. 4, lane 7). Therefore, it appears that during the last step of purification some additional component has been separated from TIF-IB that helps stabilize its complex with the promoter (see "Discussion").
A Novel Transcription Factor Can Confer Template Commitment on Glycerol Gradient-purified TIF-IB-Initial attempts at identifying this component from glycerol gradient fractions of the TIF-IB met with no success, presumably because it is too dilute. However, site-specific photo-cross-linking near the transcription initiation site revealed that the addition of partially purified RNA polymerase I resulted in significantly increased proximity of the TAF I 96 to the DNA (42). Based upon the hypothesis that this might be related to commitment, subsequent tests found that some preparations of partially purified RNA polymerase I could confer commitment (data not shown). Further purification separated from the polymerase a component with this activity, which we dubbed TIF-IE. More recently, we have found that larger amounts of this factor are present in early stages of the TIF-IB purification (data not shown).
TIF-IE separated from RNA polymerase I or TIF-IB by ratezonal sedimentation on a glycerol gradient. It sediments with an apparent native molecular weight similar to bovine serum albumin, 66,000 (data not shown), but at this stage we cannot identify the polypeptide composition of TIF-IE. When TIF-IE is added to TIF-IB during the first incubation of a template com-mitment assay, it confers upon TIF-IB the ability to commit the template (Fig. 5, lane 7). This commitment is not dependent upon which template is added during the first incubation (cf. lanes 7 and 8). TIF-IE alone in the preincubation does not result in commitment (data not shown). Therefore, this novel factor appears to act by altering the binding affinity of TIF-IB for the promoter. The molecular mechanism by which this is accomplished is currently under investigation.
The Equilibrium Binding Constants of TIF-IB Purified through the Promoter Affinity Column or the Glycerol Gradient Are Identical-TIF-IB purified through the glycerol gradient is unable to form a committed complex on the promoter but is still able to mediate specific transcription in the presence of partially purified RNA polymerase I (Fig. 3A). Further, based upon specific transcription, the yield of TIF-IB is high (Table I). Therefore, we were puzzled as to why TIF-IB with a presumed lower binding affinity for the promoter did not result in a decrease in transcription activity. A direct estimate of the equilibrium binding constant was made by EMSA (Fig. 6). Surprisingly, this analysis showed TIF-IB purified to these two stages exhibited nearly identical apparent equilibrium dissociation constants, 59 pM for the promoter-DNA affinity-purified TIF-IB and 53 pM for the glycerol gradient-purified factor; these are identical within the error of the experiment.
TIF-IB Purified through the Glycerol Gradient Forms a Rapid Equilibrium Complex with the Core Promoter-From the analysis of the equilibrium binding constant and its inability to commit template, we concluded that glycerol gradient-purified TIF-IB must be in rapid equilibrium between its bound and free forms, while at earlier stages of purification, TIF-IB must have very low "on" and "off" rates from the promoter. We tested this notion by adding a 100-fold molar excess of cold promoter DNA to a complex of TIF-IB prebound for 10 min to 32 P-labeled promoter-DNA and incubating for various periods of time.  These reactions were staggered in real time so that all of the reactions were stopped simultaneously by placing them on ice and immediately loading them onto the EMSA gel. Once in the polyacrylamide gel, the caging effect of the gel presumably holds the remaining complexes together (43,44). As seen in Fig. 7, two-thirds of the glycerol gradient-purified TIF-IB⅐DNA complex dissociated within the first minute of incubation, while the promoter-DNA affinity-purified TIF-IB was stable for the entire 20-min incubation period. Thus, the glycerol gradientpurified TIF-IB is in rapid equilibrium between its free and bound form, while TIF-IB from the earlier stage of purification forms a persistent complex with the promoter DNA. Despite the kinetic dissimilarities of these two complexes, they retain the identical equilibrium binding constant, so the association and dissociation rate constants for glycerol gradient-purified TIF-IB must be proportionately increased (see "Discussion").

DISCUSSION
The fundamental transcription initiation factor, TIF-IB, was purified to apparent homogeneity from an A. castellanii nuclear extract using ammonium sulfate fractionation, DEAE-Sepharose fast flow, BioRex-70, and promoter-DNA affinity chromatographies and rate zonal sedimentation. This is only the fourth fundamental RNA polymerase I transcription factor to be purified to homogeneity (9,(11)(12)(13)(14) and the first to be purified completely by classical methods; the others utilized an immunoprecipitation step. The purified factor contains TBP and four TAF I s with apparent molecular weights of 145,000, 99,000, 96,000, and 91,000.
The TBP subunit was shown previously to be present in the glycerol gradient-purified factor, by Western blotting with monospecific antibody against A. castellanii TBP. TBP is also in the committed complex, as shown by supershifting it with the same antibody in an electrophoretic mobility shift assay (28). Evidence in support of the other four polypeptides being part of the factor includes the following. 1) TBP and all four of these subunits sediment together and in parallel with the transcription-mediating activity ascribed to TIF-IB. 2) These four subunits were found in the same relative amounts in multiple preparations in approximately an equimolar ratio, suggesting that they make up a single protein oligomer; 3) At both early and late stages of the purification, TBP and all four of the putative TAF I s, and only these polypeptides, photo-crosslink to multiple but subunit-specific sites across the entire rRNA core promoter, which has been derivatized with photoactive deoxyribonucleotides. This photo-cross-linking is eliminated when competed with underivatized promoter-containing DNA, but not when competed by nonspecific DNA (26,42). 4) In rate zonal sedimentation, TIF-IB sediments to approximately the same position as A. castellanii RNA polymerase I, which has a native molecular weight of 515,000 (30). This compares well with the calculated molecular weight for TIF-IB of 460,000, assuming it is made up of TBP and the four TAF I s. 5) Scanning transmission electron microscopy of committed complexes is consistent with the calculated molecular weight of TIF-IB (493,000 from STEM and 460,000 from gel electrophoresis). In the aggregate, this is extremely strong evidence that the subunit composition of A. castellanii TIF-IB is TBP-TAF I 145-TAF I 99 -TAF I 96 -TAF I 91, with each subunit present once in the transcription factor.
It is interesting that A. castellanii TIF-IB has four TAF I s, while in contrast human TIF-IB (9), murine TIF-IB (11), and S. cerevisiae core factor (12)(13)(14) have only three TAF I s. Perhaps this accounts for the somewhat unique properties of A. castellanii TIF-IB when compared with the other factors; it binds unusually tightly to the core promoter, allowing the production of footprints and electrophoretic mobility shifts and driving large amounts of template into complex. In addition, there is no apparent need in A. castellanii for additional transcription factors to mediate recruitment of RNA polymerase I and transcription initiation, even when nearly homogeneous RNA polymerase I and TIF-IB are used in the assay. In contrast, yeast rRNA transcription requires another factor for the formation of sarcosyl-resistant complexes, Rrn3p (45). RRN3 is a required gene. Rrn3p does not become associated with the stable preinitiation complex, but preincubation of Rrn3p with RNA polymerase I stimulates the latter's recruitment to the promoter (45). In murine systems, two additional transcription factors have been implicated in initiation, TIF-IC and TIF-IA (or TFIC or C*) (reviewed in Refs. 6 and 46). These are also required for formation of the first phosphodiester bond, are not part of the persistent complex on the promoter, and thus bear some similarities to yeast Rrn3p. Finally, another additional factor, PAF53, has been found associated with RNA polymerase I in mouse (47,48). It is argued that PAF53 bridges UBF and RNA polymerase I during the recruitment of the polymerase to the promoter. In the case of A. castellanii, it is formally possible that the additional factor(s) are associated with TIF-IB, as an additional TAF I , rather than with RNA polymerase I. This would account for the additional TAF I in this species. There is a precedent for differential association of RNA polymerase I transcription factor subunits; while TBP is clearly a component of human SL1 (9) and mouse TIF-IB (11), TBP is more transiently associated with Xenopus Rib1 (23). In S. cerevisiae, UAF more tightly associates with TBP than core factor (7,49); thus, this subunit might be considered to be part of the former rather than the latter factor.
Finally, although the sizes of the TAF I subunits from yeast when compared with human and mouse are very similar, these polypeptides exhibit very significant differences at the primary sequence level. Indeed, the S. cerevisiae TAF I s bear no discernable sequence similarity to their size counterparts in humans or mice (12)(13)(14). Thus, it is perhaps not too surprising that A. castellanii TIF-IB has a different subunit architecture than the TIF-IB functional homologs from other species.
In less pure preparations, TIF-IB is capable of forming a persistent complex on the core promoter, one that is resistant to challenge by promoter-containing DNAs added later. This complex, called the committed complex, remains through multiple rounds of transcription. This is also a property of partially purified TIF-IB homologs from mouse and rat. In contrast, yeast CF cannot commit a template unless both UAF and TBP are also present (49). We were surprised to find that A. castellanii TIF-IB in its near homogeneous state loses the ability to commit an rRNA gene template (Fig. 4). Nevertheless, homogeneous TIF-IB can mediate specific transcription from the rRNA promoter (Figs. 3A and 4) and has the same strong equilibrium binding as its less pure preparations (Fig. 6). The latter reflects a difference in the half-life of the committed complex between nearly homogeneous and impure preparations. For the impure TIF-IB, the rate constant for the off reaction is very small (t1 ⁄2 measured in hours), but for the homogeneous TIF-IB, the half-life of the complex is approximately 45 s (Fig. 7). This translates into an apparent rate constant for the "on" reaction of about 3 ϫ 10 8 M -1 s -1 , close to that predicted for a diffusion-limited reaction (50).
An additional transcription factor, dubbed TIF-IE, can confer upon TIF-IB the ability to commit the ribosomal DNA template. This additional component is present in impure preparations of RNA polymerase I and TIF-IB but can be separated by rate zonal sedimentation, where it sediments with a native molecular weight of about 66,000. TIF-IE by itself cannot commit the template but must be present with TIF-IB and the template to do so. In this respect, it is different from yeast UAF, which can commit template on its own. The mechanism and subunit composition of TIF-IE are currently under investigation.
One of the curiosities of rRNA transcription is its strong species specificity (51,52). As the study of rRNA promoters and transcription factors initially proceeded, the similarities between the various species were significantly more striking than the differences, creating a mystery as to what determined species specificity. However, as more details about the RNA polymerase I transcription factors are discovered, significant differences between species are coming to light. In particular, we show here that A. castellanii TIF-IB has one more TAF I than other species, and these have molecular weights that are quite different from other species. A. castellanii TIF-IB contains tightly associated TBP, similar to vertebrate TIF-IB/SL1. In contrast, yeast CF is deficient in TBP, and while yeast TAF I s are similar in size to those identified in vertebrates, they have FIG. 7. Homogeneous TIF-IB forms a rapid equilibrium complex with the promoter. The off rate for the complex between TIF-IB and the promoter was estimated by incubating, for various periods of time, excess unlabeled promoter DNA with a preformed complex. The remaining complex was estimated by electrophoretic mobility shift and quantified by phosphor imaging. A, the EMSA phosphor image. B, plot of the complex remaining at the times shown for the promoter-DNA affinity-purified (solid triangles) or the glycerol gradient-purified (solid squares) TIF-IB. no sequence similarity. The novel transcription factor, TIF-IE, identified for the first time in this publication presents yet another possible component of the RNA polymerase I transcription system. Perhaps polypeptides with similar functions are aggregated into "factors" differently in diverse species, as does TBP in yeast when compared with other species, resulting in factors that cannot be exchanged between species. However, the final step in the rRNA transcription initiation process is identical for all organisms. It will be interesting to see how such a diverse set of transcription factors function in the recruitment of the relatively conserved class I RNA polymerase.