Identification, Molecular Cloning, and Characterization of the Sixth Subunit of Human Transcription Factor TFIIIC*

TFIIIC in yeast and humans is required for transcription of tRNA and 5 S RNA genes by RNA polymerase III. In the yeast Saccharomyces cerevisiae, TFIIIC is composed of six subunits, five of which are conserved in humans. We report the identification, molecular cloning, and characterization of the sixth subunit of human TFIIIC, TFIIIC35, which is related to the smallest subunit of yeast TFIIIC. Human TFIIIC35 does not contain the phosphoglycerate mutase domain of its yeast counterpart, and these two proteins display only limited homology within a 34-amino acid domain. Homologs of the sixth TFIIIC subunit are also identified in other eukaryotes, and their phylogenic evolution is analyzed. Affinity-purified human TFIIIC from an epitope-tagged TFIIIC35 cell line is active in binding to and in transcription of the VA1 gene in vitro. Furthermore, TFIIIC35 specifically interacts with the human TFIIIC subunits TFIIIC63 and, to a lesser extent, TFIIIC90 in vitro. Finally, we determined a limited region in the smallest subunit of yeast TFIIIC that is sufficient for interacting with the yeast TFIIIC subunit ScTfc1 (orthologous to TFIIIC63) and found it to be adjacent to and overlap the 34-amino acid domain that is conserved from yeast to humans.

TFIIIC in yeast and humans is required for transcription of tRNA and 5 S RNA genes by RNA polymerase III. In the yeast Saccharomyces cerevisiae, TFIIIC is composed of six subunits, five of which are conserved in humans. We report the identification, molecular cloning, and characterization of the sixth subunit of human TFIIIC, TFIIIC35, which is related to the smallest subunit of yeast TFIIIC. Human TFIIIC35 does not contain the phosphoglycerate mutase domain of its yeast counterpart, and these two proteins display only limited homology within a 34amino acid domain. Homologs of the sixth TFIIIC subunit are also identified in other eukaryotes, and their phylogenic evolution is analyzed. Affinity-purified human TFIIIC from an epitope-tagged TFIIIC35 cell line is active in binding to and in transcription of the VA1 gene in vitro. Furthermore, TFIIIC35 specifically interacts with the human TFIIIC subunits TFIIIC63 and, to a lesser extent, TFIIIC90 in vitro. Finally, we determined a limited region in the smallest subunit of yeast TFIIIC that is sufficient for interacting with the yeast TFIIIC subunit ScTfc1 (orthologous to TFIIIC63) and found it to be adjacent to and overlap the 34-amino acid domain that is conserved from yeast to humans.
Human RNA polymerase III transcribes genes encoding small non-translated RNAs. Transcription of these genes requires accessory factors that recognize the RNA polymerase III promoter elements. Genes with type 2 internal promoters (A and B boxes; tRNA-or VA genes) are recognized by TFIIIC, whereas the 5S RNA gene (type 1 internal promoter; A and C boxes as well as an intermediate element) is, in addition, dependent on TFIIIA. Type 3 promoter-containing genes (distal sequence element, proximal sequence element (PSE), and TATA-box; U6 RNA and 7SK RNA genes) require the PSE binding transcription factor (PTF; PBP; SNAPc) that binds upstream of the transcription initiation site (for review, see Refs. 1 and 2). TFIIIC alone or in conjunction with TFIIIA (type 2 or type 1 promoters, respectively) recruits TFIIIB-␤ (3), which allows the subsequent recruitment of RNA polymerase III. PSE binding transcription factor interacts with and incorporates TFIIIB-␣ (3) into the preinitiation complex at type 3 promoters, allowing the recruitment of RNA polymerase III (for review, see Refs. 1 and 2). Human TFIIIC was separated into two functional entities, TFIIIC1 and TFIIIC2 (4,5), of which the TFIIIC2 module was shown to be required for sequence-specific DNA binding to type 2 and, in conjunction with TFIIIA, to human type 1 RNA polymerase III promoters (5,6). Human TFIIIC2 was resolved into a five-subunit (220, 110, 102, 90, and 63 kDa) complex (7,8), and the cDNAs encoding the individual subunits were cloned (9 -13). Footprinting experiments revealed that TFIIIC2 interacts strongly with the B-box of type 2 promoters, and this interaction was reinforced as well as extended over the A-box to sequences close to the transcription initiation site by the addition of TFIIIC1. In addition, in vitro transcription with partially purified TFIIIC1 demonstrated that it was absolutely required for transcription of type 2 promoter-containing genes (5,14). Later, TFIIIC1 was also shown to be required for transcription of type 1 and type 3 promoters by RNA polymerase III (6). Although the exact composition of TFIIIC1 remains unclear, it was shown to at least comprise human Bdp1, a component of human TFIIIB-␣ and TFIIIB-␤ (15). These two forms of human TFIIIB contain, besides Bdp1, the TATA-binding protein and the TFIIB-related factors Brf1 (TFIIIB-␤) or Brf2 (TFIIIB-␣), respectively (16,17). TFIIIC2, TFIIIC1, and TFIIIB-␤, in conjunction with RNA polymerase III, are able to specifically initiate and also terminate transcription from type 2 promoters in vitro, but the efficiency of transcription is strongly enhanced by the addition of the RNA polymerase II co-activators PC3 (topoisomerase 1) or PC4 (18).
tRNA gene transcription in vitro and in vivo in the yeast Saccharomyces cerevisiae requires TFIIIC, TFIIIB, and RNA polymerase III. TFIIIB is composed of TBP, Bdp1, and Brf1 and is, thus, very similar in its composition to both forms of human TFIIIB. RNA polymerase III in yeast and humans was reported to contain 17 subunits that are conserved between the two species (for review, see Refs. 1, 2, and 19). However, S. cerevisiae TFIIIC was shown to be composed of six instead of five subunits for its human counterpart. The subunits of S. cerevisiae TFIIIC are encoded by the TFC1, -3, -4, -6, -7, and -8 genes (20 -27). Functional and/or sequence homologs have been identified in human cells for Tfc1 (HsTFIIIC63), Tfc3 (HsTFIIIC220), Tfc4 (HsTFIIIC102), Tfc6 (Hs-TFIIIC110), and for Tfc8 (HsTFIIIC90) (9 -13) but not for Tfc7. S. cerevisiae TFIIIC is composed of two subdomains, A and B (28). Tfc4 (131), Tfc1 (95), and Tfc7 (55) constitute A, whereas B is composed of Tfc3 (138), Tfc6 (91), and of Tfc8 (60) (27). The Tfc3 and Tfc6 subunits of B bind to B-box and close to transcription termination sequences, respectively (25,29,30). Tfc8 (60) is thought to connect the A and B subassemblies of TFIIIC (27). A represents the A-box and transcription initiation site binding module of TFIIIC. Nevertheless, it has recently been published that recombinant B is able to recruit TBP and to stimulate TFIIIB-directed transcription of a TATA box-containing tRNA gene, implying a combined contribution of A and B to TBP recruitment during preinitiation complex formation (31). Important roles for the A component Tfc4 (131) in recruitment of TFIIIB through interaction with its Brf1 and Bdp1 components (24,(32)(33)(34)(35)(36)(37)(38) and of RNA polymerase III via interactions with its C53 (39) and ABC10␣ subunits (40) have been published.
The Tfc1 (95) and Tfc7 (55) subunits of A could be specifically cross-linked to the vicinity of the A-box (29). In addition, Tfc7 (55) was shown to interact and to form a distinct subcomplex with Tfc1 (95), leading to the model that Tfc7 (55) is located at the interface of A and B, in close proximity to Tfc1 (95) and of Tfc8 (60) (26,27). Recently, all six subunits of S. cerevisiae TFIIIC were shown to be required and sufficient for transcription of tRNA and 5 S RNA genes in vitro in a reconstituted transcription system composed of all recombinant factors and highly purified RNA polymerase III (41).
Although the DNA binding properties and protein-protein interaction studies of ScTfc7 (55) suggested an important role for this subunit in TFIIIC function, all attempts to identify a related sixth subunit of human TFIIIC have hitherto failed. Here we report the in silico identification, molecular cloning, and characterization of this sixth subunit of human TFIIIC. We were able to identify this subunit by sequence homology searches, although it exhibits only limited sequence conservation with its yeast ortholog. The human protein, HsTFIIIC35, stably assembles into the TFIIIC2 complex, which is active in binding to and transcription of type 2 promoters by RNA polymerase III in vitro.

EXPERIMENTAL PROCEDURES
Buffers and Plasmids-BC buffer contained 20 mM Hepes/ KOH, pH 7.9, 10% glycerol, 1.5 mM MgCl 2 , 3 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, and KCl as appropriately indicated. An EST encompassing the cDNA encoding HsTFIIIC35 was obtained from the German Resource Center for Genome Research, RZPD (IMAGp998F0713364Q). Oligonucleotides were purchased from Sigma Genosys. All constructs were verified by sequencing using an ABI-310 genetic analyzer. The VA1 plasmid for in vitro transcription was previously described (3).
Purification of FLAG/HA 4 /Myc-HsTFIIIC-The cDNA encoding HsTFIIIC35 was cloned into the pFH-IRES1neo vector (Clontech) that was modified by introducing DNA sequences in between the EcoRV and NotI restriction sites encoding sequentially the FLAG, HA, and Myc epitopes to obtain the pFHM vector. At the 3Ј-end of the DNA sequence encoding the Myc epitope, an NheI restriction site was introduced that together with the BamHI site served for ligating the HsTFIIIC35 cDNA in-frame with the triple epitope. For that purpose the HsTFIIIC35 cDNA was amplified by employing sense (HsTFIIIC35-FNheI, 5Ј-CATGCTAGCA-TGGCGGCGGCGGCGGACG) and antisense (HsTFIIIC35-RBamHI, 5Ј-CATGGATCCTTTCTAAGGCAGCATTTG-AGTTTCC) primers followed by restriction with NheI and BamHI in appropriate buffers before ligation into the pFHM vector. The establishment of the HeLa cell line stably expressing epitope-tagged HsTFIIIC35, its maintenance in DMEM, and the preparation of nuclear extracts were as described (42). Before immunoaffinity purification procedures, 150 mg of nuclear extracts containing FLAG-HA-Myc (FHM)-HsTFIIIC35 or FLAG-HsTFIIIC110 were subjected to chromatography over phosphocellulose P11 in BC buffer as described (3). The fraction eluting between 350 and 600 mM KCl contained 600 g/ml protein and the majority of TFIIIC activity. 15 mg of this fraction from each cell line were subsequently further purified over M2-agarose as described (42) and eluted with 30 l of BC60 containing 250 ng FLAG peptide/l.
Preparation of Recombinant HsTFIIIC35 and Generation of Antibodies-The cDNA encoding HsTFIIIC35 was used as a template for PCR amplification with HsTFIIIC35-FNdeI sense (5Ј-GGAATTCCATATGGCGGCGGCGGCGGACG) and HsTFIIIC35-RBamHI antisense primers. The PCR product was restricted with NdeI and BamHI and ligated into NdeI-BamHI sites of the pET-His 6 -GST bacterial expression vector that was derived from the pET15 vector (Novagen; a gift from S. Fribourg) to yield the pET-His 6 -GST-HsTFIIIC35 vector. This construct was used for the expression of recombinant His 6 -GST-HsTFIIIC35 in BL21-Gold (DE3) as described (42), and the expressed protein was purified over nickel-nitrilotriacetic acid-agarose (Qiagen) under denaturing conditions in the presence of 8 M urea. The fusion protein was renatured by dialysis against BC60 buffer. Non-renaturated proteins were removed by centrifugation at 20,000 ϫ g for 5 min in a tabletop centrifuge at 4°C. The renatured protein was either directly used for functional assays or it was cleaved by tobacco etch virus (TEV) protease for removing the GST fusion protein. Cleaved recombinant full-length HsTFIIIC35 was resolved by SDS-12.5%PAGE, stained by Coomassie Brilliant Blue, excised from the gel, and used for the generation of anti-HsTFIIIC35 antibodies. SDS-PAGE and Western blot assays were essentially as described (3).
In Vitro Transcription and Gel Mobility Shift Assays (EMSA)-In vitro transcription was performed as described (14,17). Transcripts were resolved by 8 M urea-6% PAGE. EMSA binding reactions were performed at 30°C for 60 min in BC60 supplemented with 30 g of bovine serum albumin, 1 g of poly(dI-dC):(dI-dC) in a final volume of 20 l. The pVA1 plasmid was digested with EcoRI and HindIII restriction enzymes, and the resulting VA1 fragment was dephosphorylated and gelpurified. Thereafter, it was 5Ј-labeled by T4 polynucleotide kinase and [␥-32 P]ATP (PerkinElmer Life Sciences). All DNA binding reactions contained 10000 cpm DNA probe (ϳ1 ng of DNA) and the appropriately indicated protein fractions. Separation of DNA from DNA-protein complexes was achieved by native 5% PAGE in 0.5ϫ Tris-buffered EDTA using 1.5-mm gel spacers. After electrophoresis, polyacrylamide gels were dried and exposed to X-Omat MR film autoradiography (Eastman Kodak Co.).
Protein-Protein Interaction (GST pulldown) Assays-In five separate reactions recombinant His 6 -GST-HsTFIIIC35 or GST alone (1 g) bound to glutathione-agarose was incubated with 20 l of reticulocyte lysate containing 35 S-labeled recombinant human TFIIIC subunits in BC250 plus 0.1% Nonidet P-40. The beads were washed extensively with BC400 plus 0.1% Nonidet P-40. Retained proteins were eluted by boiling the beads in SDS loading buffer and subjected to SDS-5% PAGE (HsTFIIIC220) or SDS-9% PAGE (HsTFIIIC110, HsTFIIIC102, HsTFIIIC90, HsTFIIIC63). The gels were stained with Coomassie Brilliant Blue, dried, and subsequently exposed to X-Omat MR film autoradiography.
Protein Detection by Immunofluorescence-Cells grown on glass coverslips were fixed in PBS containing 3.7% formaldehyde (Sigma) for 20 min, treated with PBS containing 50 mM NH 4 Cl for 10 min, and permeabilized with PBS, 0.5% Triton X-100 for 5 min at room temperature. Between individual incubation steps, cells were washed 3 times for 2 min in PBS. After blocking with PBS, 0.2% gelatin for 1 h at room temperature, cells were incubated overnight at 4°C with specific primary antibodies diluted in blocking buffer, washed in PBS, and then incubated at room temperature for 1 h with anti-mouse Alexa-Fluor 488-and anti-rabbit AlexaFluor 594-conjugated secondary antibodies (Molecular Probes) diluted 1/400 in blocking buffer. Nuclei were counterstained by adding Hoechst 33528 (1 g/ml) to PBS in one of three final wash steps. The slides were mounted in Fluoromount medium (Molecular Probes) and dried at room temperature before being analyzed by confocal microscopy using a LSI 510 Meta Microscope (Zeiss). For Alexa-Fluor 488, AlexaFluor 594, and Hoechst 33528 visualization, we chose as excitation wavelength 488 nm (Ar laser), 543 nm (HeNe laser), and 405 nm (diode laser) and detection wavelength frames 505-540, 561-753, and 420 -480 nm, respectively. Primary antibodies were diluted in blocking buffer as follows: mouse anti-c-Myc (clone 9E10; Sigma) 1/100; rabbit anti-HsTFIIIC110 (11) 1/2000. Two-hybrid Assay-Two-hybrid analyses were essentially as described (39) employing the FRYL library for genome-wide screen (43).

RESULTS
We were prompted to search for genes encoding homologs of S. cerevisiae (Sc) Tfc7 (55) in the genomes of other eukaryotes, because all data obtained in S. cerevisiae indicated that this subunit is indispensable for transcription by RNA polymerase III. However, despite all attempts, no counterpart of ScTfc7 (55) could hitherto be identified in any other organism.
Identification of Potential Orthologs of S. cerevisiae Tfc7 in Several Eukaryotic Genomes-A TblastN search (Gish, W. (1996 -2002) Blast) with the ScTfc7 protein sequence as query was run on the Schizosaccharomyces pombe genome at The Sanger Institute. This search revealed a weak homology of amino acids 321-348 of ScTfc7 with a hypothetical protein encoded by a short intronless ORF of S. pombe (44). However, this ORF is yet unannotated as coding for a true protein because its first 26 nucleotides belong to the terminal exon of the preceding ORF (SPAC1250.03) that encodes the S. pombe ortholog of S. cerevisiae ubiquitin-conjugating enzyme Ubc4. The here identified unannotated S. pombe ORF (5097850 -5098212 in chromosome A, antiparallel) presumably encodes an S. pombe protein representing a potential ortholog of ScTfc7, and we tentatively refer to it as Sfc7(accession number SPAC1250.07), following the nomenclature already proposed by Huang and Maraia (1).
The Tfc7 Core Motif-A 34-amino acid sequence of Sfc7 (positions 35-68), exhibiting the highest homology to ScTfc7, was used as a seed for a Psi-Blast search (45) and yielded after four Psi-Blast iterations potential orthologs from various higher eukaryotes (shown in Fig. 1 and listed in supplemental Table 1). Among these proteins, a human protein (C6orf51; locus 6q21, GenBank TM accession NP_612417) encompassing 213 amino acids was identified as a potential ortholog of ScTfc7. Importantly, Psi-Blast searches identified only one ScTfc7 homologous protein in each eukaryotic genome. In S. pombe and higher eukaryotes, the proteins identified do not contain sequences homologous to the N-terminal part of ScTfc7 (see below). The 34-residue core motif used for the search represents the only protein sequence conserved across all species, and we, thus, refer to it as the Tfc7 core motif. Only a single proline residue at position 11 is strictly conserved (Fig.  1). Of the 34 positions, 13 are characterized by aliphatic or aromatic residues (Fig. 1, in blue with asterisks at the bottom). Plants (Arabidopsis thaliana and Oriza sativa japonica) display a single intervening residue (lysine) at position 19, whereas larger intervening sequences are found in Debaryomyces hansenii and Candida albicans at the same location and at position 23 of the core motif.
Conservation of Both Parts of the Chimeric Yeast Tfc7 Subunit across All Species-ScTfc7 (encoded by YOR110w) is composed of two specialized parts (26). The N-terminal part (residues 1-278) contains the signature of enzymes related to the glycolytic pathways such as phosphoglycerate mutase (PGM) and is not required for cell viability. In contrast, the C-terminal part of ScTfc7 (residues 279 -435) is required and sufficient for viability and for the formation of a functional TFIIIC complex. For clarity, we designate these two parts as "PGM" (N-terminal amino acids 1-278) and "transcription domain" (TD; C-termi-nal amino acids 279 -435). It should be noted that the Tfc7 core motif (amino acids 317-350 in ScTfc7) is an integral part of the TD. In the S. cerevisiae genome, YOR110w occupies the last position of the duplicated block 50 (8 duplicated genes (46)), and its twin ORF, YNL108c, solely encodes the PGM part. It was, therefore, hypothesized (26) that the original PGM part was that encoded by YNL108c and that the chimeric structure of ScTfc7 (encoded by YOR110w) resulted from a fusion between the last gene of the translocated block (YNL108c, PGM only) and a gene encoding an ancestral transcription factor (the TD is not included in the duplicated block; see Fig. 2A).
We searched and examined the homologs of YOR110w and YNL108c in eukaryotic genomes to sharpen the phylogeny of these two genes. As a matter of fact, all hemiascomycetes up to the distant yeast Yarrowia lipolytica display the same chimeric structure as YOR110w (Fig. 2B), showing that the fusion between the PGM and TD parts is by far more ancient than the whole genome duplication that took place in the Saccharomyces clade after the divergence from the Kluyveromyces clade (46) (noted as WGD in Fig. 2B). In S. pombe and all other eukaryotes in which we identified potential orthologs of Tfc7 (55), no fusion to a PGM-like part could be observed. Altogether, these data support that the PGM-TD fusion arose at the root of hemiascomycetes after the divergence from S. pombe and other archiascomycetes. Remarkably, in the three genomes, S. cerevisiae, Saccharomyces castellii, and Candida glabrata, which underwent the whole genome duplication and subsequent chromosomal rearrangements, potential orthologs of YNL108c, are also present, suggesting that in hemiascomycetes the original PGM part was that encoded by YOR110w (instead of YNL108c as previously hypothesized (26)) and that the sequence encoding the PGM part of YOR110w was translocated without that encoding the TD part to form YNL108c (Fig. 2C). Independently, in Y. lipolytica, a duplication of YOR110w took place, and one of the copies encodes a degenerate TD part that is probably not functional.
The Human Tfc7 Core Motif-containing Protein Colocalizes with HsTFIIIC110 in HeLa Cells-We wanted to know whether the human protein exhibiting homology to the ScTfc7 core motif is a constituent of human (Hs)TFIIIC or not. First we analyzed whether this protein colocalizes with a bona fide subunit of HsTFIIIC (HsTFIIIC110) in HeLa cells. To address this question, we generated a HeLa cell line stably expressing this homolog as an FHM epitope-tagged fusion protein ("Experimental Procedures"). By employing immunofluorescence and confocal microscopy ("Experimental Procedures"), we were able to detect the FHM epitope-tagged fusion protein with a monoclonal anti-Myc antibody (9E10) and HsTFIIIC110 with antibodies directed against the C-terminal 595 amino acids of TFIIIC110 (11). The majority of both proteins was found in nuclei of HeLa cells (Fig. 3), indicating that both proteins could be present in the HsTFIIIC complex.
HsTFIIIC35 Is an Integral Subunit of Human TFIIIC-We next analyzed whether we could purify HsTFIIIC by immunoaffinity chromatography from the cell line, stably expressing the FHM epitope-tagged human protein comprising the Tfc7 core motif. For that purpose, we prepared nuclear extracts from this cell line and subjected them to chromatography over phosphocellulose (P11). The fraction eluting between 350 and 600 mM KCl was used to affinity-purify the FHM epitope-tagged protein via M2-agarose ("Experimental Procedures"). The purified samples were subsequently analyzed by SDS-9%PAGE and silver stain (Fig. 4A) or by Western blot with anti-HsTFIIIC antibodies specifically directed against individual HsTFIIIC subunits (Fig. 4B). Analysis of the purified complex by SDS-9% PAGE and silver stain revealed the presence of five proteins of 220, 110, 102, 90, and 63 kDa that were likely to correspond to the known subunits of HsTFIIIC. In addition, protein bands of 35 and 40 kDa could be visualized (Fig. 4A, lane 3). As a control, we analyzed an affinity-purified HsTFIIIC sample from a FLAG epitope-tagged HsTFIIIC110 cell line (18) ("Experimental Procedures") that, as expected, contained HsTFIIIC with its five characterized subunits (Fig. 4A, lane 2). The purification of HsTFIIIC from the epitope-tagged HsTFIIIC110 cell line also resulted in the co-purification of several polypeptides smaller  Table 1. than 63 kDa, comprising 2 that migrated with 40 and 35 kDa, that could be related to the proteins purified from the cell line stably expressing the FHM epitope-tagged human protein with homology to the Tfc7 core motif. We wanted to know whether the five proteins described above indeed represented the known subunits of HsTFIIIC. To address this question, we performed Western blot analyses with antibodies directed against individual subunits of HsTFIIIC. As shown in Fig. 4B, all five known subunits of HsTFIIIC could be detected in samples that were affinity-purified from the cell line stably expressing the FHM-tagged human Tfc7 core motif protein (lane 2). The migration was identical with that of HsTFIIIC subunits, puri-fied from the epitope-tagged Hs-TFIIIC110 cell line with the exception of HsTFIIIC110 itself that migrated slightly slower due to its FLAG tag (compare in Fig. 4, A,  lanes 2 and 3, and B, lanes 1 and 2).
The human Tfc7 core motif protein itself migrated with 35 kDa after SDS-9% PAGE, and we designated it as HsTFIIIC35. HsTFIIIC35 purified from the FHM epitope-tagged HsTFIIIC35 cell line was detected by employing monoclonal antibodies that were directed against the Myc peptide (9E10), recognizing the Myc tag (Fig. 4B, lane 2, second  panel from the bottom). Hs-TFIIIC35 could likewise be detected in purified samples from a cell line stably expressing FLAG-tagged HsTFIIIC110 by antibodies that were directed against HsTFIIIC35 ("Experimental Procedures"), indicating that it indeed represents the sixth subunit of human TFIIIC (Fig.  4B, lane 1, lowest panel). These anti-HsTFIIIC35 antibodies detected a protein of about 40 kDa in affinitypurified samples from the epitopetagged HsTFIIIC35 cell line that most likely corresponds to the FHM-tagged HsTFIIIC35 fusion protein. Silver staining of purified human TFIIIC also revealed that HsTFIIIC35 is only faintly stained (Fig. 4A), probably explaining why it was not identified earlier by biochemical approaches.
Epitope-tagged HsTFIIIC35 Assembles into the HsTFIIIC Complex That Is Active in DNA Binding and Transcription-We next aimed at analyzing whether HsTFIIIC that was affinity-purified from FHM-HsTFIIIC35 cell line nuclear extracts was active in binding to and transcription of type 2 promoters. As shown in Fig. 5A, Hs-TFIIIC purified from the HsTFIIIC35 cell line bound to the VA1 gene. The shift produced in EMSA was indistinguishable from the one that was produced by affinity-purified HsTFIIIC from the FLAG-HsTFIIIC110 cell line (Fig. 5A, compare lanes 2  and 3). Incubation of TFIIIC⅐VA1⅐DNA complexes with anti-FLAG antibodies resulted in the appearance of a slightly slower migrating complex regardless of whether TFIIIC was prepared from the FLAG-HsTFIIIC110 or from the FHM-HsTFIIIC35 cell line (Fig. 5A, lanes 4 and 5). Incubation of TFIIIC⅐VA1⅐DNA complexes with anti-Myc antibodies only gave rise to a supershift if FHM-HsTFIIIC35 preparations were employed, reflect-  JUNE 8, 2007 • VOLUME 282 • NUMBER 23

JOURNAL OF BIOLOGICAL CHEMISTRY 17183
ing that these preparations, but not FLAG-HsTFIIIC110 preparations, contained a Myc tag (Fig. 5A, compare lanes 6 and 7). The band migrating in between the TFIIIC-VA1 shift and free VA1 DNA, which is indicated with an asterisk, was previously observed (6). These authors could not correlate this band to a specific footprint in the VA1 gene nor could they prevent its formation by competition with excess unlabeled B-box oligo, and this band was, therefore, considered nonspecific (6). Next, we depleted nuclear extracts from HsTFIIIC activity by mild heat treatment (Fig. 5B, compare lanes 1 and 2) (8). To the depleted extracts, we added roughly equimolar amounts of HsTFIIIC that was either purified from the FLAG-Hs-TFIIIC110 or the FHM-HsTFIIIC35 cell line (Fig. 5C, estimation of HsTFIIIC amounts in individual HsTFIIIC preparations was by Western blot with antibodies directed against Hs-TFIIIC102). Both preparations of HsTFIIIC were comparably able to restore transcription of the VA1 gene to the depleted extracts (Fig. 5B, compare lanes 3-5 with 6 -8), underscoring that HsTFIIIC35 is an integral component of HsTFIIIC.
HsTFIIIC35 Interacts with HsTFIIIC63 and HsTFIIIC90-In S. cerevisiae, ScTfc7 (55) was shown to interact with ScTfc1 (95), the ortholog of HsTFIIIC63 (26). We wanted to know whether HsTFIIIC35 directly interacts with HsTFIIIC63 or with other subunits of HsTFIIIC. For that purpose, we individually expressed the five known HsTFIIIC subunits in coupled in vitro transcription-translation reactions in the presence of [ 35 S]methionine (Fig. 6, lower panel). The labeled proteins were employed for in vitro GST pulldown experiments showing that the GST fusion protein of HsTFIIIC35, but not GST alone, was able to specifically interact with and retain recombinant Hs-TFIIIC63 and to a lesser extent recombinant HsTFIIIC90 (Fig.  6A, upper panel, lanes 8 and 6). The other subunits of HsTFIIIC (HsTFIIIC220, HsTFIIIC110, and HsTFIIIC102) were not significantly more strongly retained by the GST-HsTFIIIC35 fusion protein (Fig. 6A, upper panel, lanes 2, 4, and 10) than by GST alone (Fig. 6A, lanes 1, 3, and 9). These experiments indicated that the interaction of ScTfc7 (55) with ScTfc1 (95) may have been conserved during evolution. These experiments also showed that HsTFIIIC35 interacted with the human ortholog of ScTfc8 (60), HsTFIIIC90, which had not been reported from S. cerevisiae.
Amino Acids in ScTfc7 (55) Adjacent to and Overlapping the Conserved Domain Are Involved in the Interaction with ScTfc1 (95)-In light of the limited sequence conservation of ScTfc7 (55) and HsTFIIIC35 but the conserved ability of the two proteins to interact with ScTfc1 (95) and the orthologous Hs-TFIIIC63, respectively, we aimed at finding out which amino acids in ScTfc7 (55) are required for the interaction with ScTfc1 (95). For that purpose, we carried out a two-hybrid screen of a fragmented genomic DNA library (43) with ScTfc1 (95) as bait. This screen identified six DNA fragments of ScTfc7 (55), encoding partial ScTfc7 proteins that all contained amino acids 281-322, adjacent to the domain that we found to be conserved from yeast to humans (Fig. 7A). Five of the six identified ScTfc7 (55) fragments contained the entire domain (amino acids 317-351) that was conserved from S. cerevisiae to humans and other higher eukaryotes, indicating that this part of ScTfc7 (55) may also be involved in establishing an interaction with ScTfc1 (95). For confirming that this part of ScTfc7 (55) is by itself able of interacting with ScTfc1 (95), we cloned a DNA fragment, encoding amino acids Thr-278 to Phe-350 into vectors for directed two-hybrid analysis. Fusion of ScTfc7-(278 -350) (55-(278 -350)) to the activation domain of Gal4 was sufficient to establish an interaction with ScTfc1 (95) (Fig. 7B), underscoring that this part of ScTfc7 (55) is indeed sufficient for interacting with ScTfc1 (95). However, no interaction of the two proteins could be observed if ScTfc7-(278 -350) (55-(278 -350)) was fused to the DNA binding domain, and ScTfc1 (95) was fused to the activation domain of Gal4, FIGURE 3. The human Tfc7 core motif protein colocalizes with HsTFIIIC110 in HeLa cells. Analysis of the intracellular location of the human Tfc7 core motif protein and of HsTFIIIC110 by immunofluorescence and confocal microscopy (as described under "Experimental Procedures") is shown. FHM human Tfc7 core motif protein is detected by a mouse anti-Myc antibody (9E10), and HsTFIIIC110 is detected by rabbit anti-HsTFIIIC110 antibodies. The primary antibodies were recognized by using anti-mouse AlexaFluor 488 (green fluorescence)-or anti-rabbit AlexaFluor 594 (red fluorescence)-conjugated secondary antibodies. Nuclei have been counterstained with Hoechst dye in blue, and whole cells were visualized by phase contrast microscopy. An overlay of all four images is shown at the bottom. which could be explained by improper folding of the fusion proteins (Fig. 7B, upper panel). As a positive control, we examined the interaction of full-length ScTfc7 (55) and ScTfc1 (95) that was previously published (26) (Fig. 7B,  lower panel). Taken together, these data indicate that amino acids in ScTfc7 (55) that are adjacent to and overlapping the conserved region (Tfc7 core motif) are important for the interaction with ScTfc1 (95).

DISCUSSION
The identification and molecular cloning of human TFIIIC35 probably completes the discovery of human TFIIIC subunits. Hs-TFIIIC35 lacks the N-terminal part of ScTfc7 (55) that shows homology to PGM family proteins. It was previously shown for ScTfc7 (55) that deletion of the entire N-terminal domain (55-⌬N3) had no effect on the growth of yeast, although the 55-⌬N3 deletion mutant bound tDNA promoter elements less efficiently and was less active in transcription of these genes. These data together with the finding that partial deletion of the ScTfc7 N terminus (55-⌬N1) slowed down growth of yeast in ethanol-or glycerol-containing media led to the hypothesis that ScTfc7 N-terminal sequences may play an auxiliary role in transcription (26,47). This auxiliary role may be contributable to participation of the ScTfc7 N-terminal sequences in structuring ScTFIIIC and/or maintaining its stability. The fact that HsTFIIIC35 as well as its potential orthologs in S. pombe and other higher eukaryotes lack the PGM part indicates that this part of ScTfc7 (55) may not be involved in essential interactions with DNA, with TFIIIB, or the polymerase itself in S. cerevisiae. It may rather play a role in the formation of the threedimensional structure of ScTFIIIC. However, this contribution seems not to be essential, even in S. cerevisiae, since the deletion of the entire N terminus is not lethal (26).  As a result of the identification of HsTFIIIC35, it seems as if the subunit composition of TFIIIC has been conserved from yeast to humans (summarized in Fig. 8). Although we cannot rule out the possibility that human TFIIIC is composed of more than six distinct subunits, we do not believe it to be probable. Human TFIIIC, affinity-purified via its epitope-tagged Hs-TFIIIC35 subunit, contains six major polypeptides as evidenced by SDS-PAGE and silver stain, which could be attributed to the previously identified and to the here-reported subunits of Hs-TFIIIC (Fig. 4). A minor polypeptide of about 35 kDa is recognized by antibodies directed against HsTFIIIC35 and represents either a truncated form of epitope-tagged HsTFIIIC35 or endogenous HsTFIIIC35 that co-purifies with TFIIIC. If the latter was the case, HsTFIIIC would contain two copies of Hs-TFIIIC35 (in this purification one epitope-tagged and one untagged). With respect to this possibility, it was hypothesized that ScTFIIIC may contain two copies of ScTfc1 (95) (29), which interacts with ScTfc7 (55). However, it was later on shown with epitope-tagged ScTfc1 (95) that ScTFIIIC only contains one copy of ScTfc1 per complex (48). Although we cannot rule out the possibility that HsTFIIIC35 is present in two copies per HsTFIIIC complex, we believe it to be unlikely and favor the possibility that the 35-kDa band corresponds to a truncated form of epitope-tagged HsTFIIIC35.  35 S-labeled HsTFIIIC subunits that were retained in GST pulldown assays, performed as described under "Experimental Procedures" (SDS-9% PAGE for HsTFIIIC110, HsTFIIIC102, HsTFIIIC90, and HsTFIIIC63; SDS-5% PAGE for HsTFIIIC220). Molecular weight marker proteins (Bio-Rad broad range) are appropriately indicated on the left of each gel. Lanes 1, 3, 5, 7 and 9, proteins that were retained by GST. Lanes 2,4,6,8,and 10, proteins that were retained by GST-HsTFIIIC35. GST or GST-HsTFIIIC35, respectively, were incubated with 35 S-labeled HsTFIIIC110 (lanes 1 and 2), HsTFIIIC102 (lanes 3 and 4), HsTFIIIC90 ( lanes 5 and 6), HsTFIIIC63 (lanes 7 and 8), or HsTFIIIC220 (lanes 9 and 10). The autoradiograph of the interaction experiment is shown in the upper panel. As determined by scanning and ImageQuant analysis (GE Healthcare), the ratio of retention of 35 S-labeled HsTFIIIC subunits by GST-HsTFIIIC35 versus GST was 2.9 for HsTFIIIC110, 4.7 for HsTFIIIC102, 10 for HsTFIIIC90, and 11.3 for HsTFIIIC63. Because of the lack of retained protein and, thus, detectable specific signals in lanes 9 and 10, the ratio was not determined for HsTFIIIC220. We considered a ratio of retention by GST-HsTFIIIC35 versus GST of 5 as the cutoff for specific interaction. 20% of each in vitro transcribed and translated, [ 35 S]methionine-labeled TFIIIC subunit that was employed for the interaction experiments is shown in the lower panel. It furthermore seems as if certain, but not all protein-protein interactions within TFIIIC and of TFIIIC with TFIIIB have been conserved during evolution. Our finding that HsTFIIIC35 interacts with HsTFIIIC63 is complementary to the interaction of ScTfc7 (55) and ScTfc1 (95) that was previously shown (26). Also, ScTfc4 (131) and HsTFIIIC102 were both shown to interact with ScTfc1 (95)/HsTFIIIC63 and with the TFIIIB components TBP and Brf1, implying that mechanisms of TFIIIB recruitment have been conserved (12, 24, 32-34, 36 -38). Other interactions of ScTFIIIC subunits like those of ScTfc3 (138) with ScTfc6 (91) (25), ScTfc4 (131) (49), or ScTfc1 (95) (50) have hitherto not been analyzed with human proteins. On the other hand, an interaction of ScTfc7 (55) and of ScTfc8 (60) that would correspond to the here-shown interaction of HsTFIIIC35 and HsTFIIIC90 has not yet been reported. Although such an interaction may exist in S. cerevisiae, a thermosensitive mutation of ScTfc8 (60) could not be rescued by overexpression of ScTfc7 (55), but only by overexpression of TBP or Bdp1 (27). Thus, our finding that Hs-TFIIIC35 interacts with HsTFIIIC90 in vitro has no published correlate in the yeast S. cerevisiae. If we also take into account that interactions of HsTFIIIC90 with both Hs-TFIIIC63 and HsTFIIIC220 (13) have no published correlate in S. cerevisiae and that, furthermore, the interaction of ScTfc8 (60) and ScTBP (27) could not be reproduced with human proteins in vitro (13), we may speculate that the incorporation of HsTFIIIC90 into the TFIIIC complex depends at least in part on protein-protein interactions that are different from those of its S. cerevisiae ortholog ScTfc8 (60). However, the interaction of HsTFIIIC90 and of Hs-TFIIIC110 was recently reproduced with ScTFC8 (60) and a C-terminal fragment of ScTfc6 (91) (31), indicating that other protein-protein contacts of ScTfc8 (60) within the TFIIIC complex have been conserved.
Taking into account all published data allows the suggestion that ScTfc7, ScTfc8, and ScTfc1 in S. cerevisiae as well as Hs-TFIIIC35, HsTFIIIC90, and HsTFIIIC63 in human cells are located in close vicinity to each other. Furthermore, the interaction of HsTFIIIC35 with HsTFIIIC63, the latter of which was co-immunoprecipitated with HsTFIIIC102, permits tentatively positioning HsTFIIIC35 downstream the A-box. In addition, the interaction of HsTFIIIC35 and Hs-TFIIIC90 as well as the previously published interactions of HsTFIIIC90 with HsTFIIIC63, HsTFIIIC220, and Hs-TFIIIC110 (13) support a model that positions HsTFIIIC35 and HsTFIIIC90 in between the A-box binding module (HsTFIIIC63 and HsTFIIIC102) and the B-box binding module of HsTFIIIC220 and HsTFIIIC110 in the center of human TFIIIC. This model is in accordance with that of ScTFIIIC, which based on biochemical and genetic protein-protein interaction data as well as on photocross-linking data, suggested that ScTfc7 (55) and ScTfc8 (60) are located at the interface of A and B (Refs. 26, 27, and 29; for review, see Ref. 19).
In addition to the general conservation of protein-protein interactions within TFIIIC, the two-hybrid screen data presented in this manuscript suggest that the Tfc7 core motif, which has been conserved from yeast to humans, and amino acids adjacent to this domain are essential for the interaction of ScTfc7 (55) and ScTfc1 (95). Thus, the conservation of this region may be due to its importance in TFIIIC complex formation.
The two S. cerevisiae subunits ScTfc7 (55) and ScTfc1 (95) were furthermore shown to form a subcomplex that is distinct of ScTFIIIC (26). Although HsTFIIIC35 and HsTFIIIC63 interact in vitro, we did not obtain direct evidence that the human orthologs form a similar subcomplex in HeLa cells. However, our attempts to obtain hints for the existence of such a subcom- plex with recombinant HsTFIIIC subunits led to the identification of an interaction of HsTFIIIC35 with HsTFIIIC90 that was not reported for the orthologous S. cerevisiae TFIIIC subunits (ScTfc7 (55) and ScTfc8 (60); see above).
In summary, the conservation of the subunit composition from S. cerevisiae TFIIIC to human TFIIIC2 and the general conservation of protein-protein interactions within TFIIIC are opposed to a considerable divergence of the amino acid sequences of individual TFIIIC subunits from yeast to man (Fig.  8). This contrast may reflect that TFIIIC in yeast and humans must recognize and bind to the highly conserved promoter elements of tRNA genes and is, thus, forced to maintain its subunit composition. At the same time, yeast and human TFIIIC may need to respond to and integrate different stimuli for regulating RNA polymerase III transcription. These regulatory processes may involve the interaction with species-specific proteins and, thus, require the amino acid sequence divergence that can be observed from yeast to human TFIIIC subunits.