Human Transcription Factor TFIIIC2 Specifically Interacts with a ... - NCBI

3 downloads 14995 Views 3MB Size Report
In addition, the 5S rRNA gene binding site and the tRNA-type gene B-box sequence did not ... t This paper is dedicated to Arthur Kornberg on the anniversary.
Vol. 9, No. 11

MOLECULAR AND CELLULAR BIOLOGY, Nov. 1989, P. 4941-4950

0270-7306/89/114941-10$02.00/0 Copyright © 1989, American Society for Microbiology

Human Transcription Factor TFIIIC2 Specifically Interacts with Unique Sequence in the Xenopus laevis 5S rRNA Genet

a

LEE G. FRADKIN,"12t STEVEN K. YOSHINAGA,2'3§ ARNOLD J. BERK,2'3 AND ASIM DASGUPTA' 2* Department of Microbiology and Immunology and Jonsson Comprehensive Cancer Center, School of Medicine,' and Molecular Biology Institute2 and Department of Microbiology,3 University of California, Los Angeles, California 90024-1747 Received 12 June 1989/Accepted 14 August 1989

Transcription factor TFLIC2 derived from human cells is required for tRNA-type gene transcription and binds with high affinity to the essential B-box promoter element of tRNA-type genes. Although 5S rRNA genes contain no homology with the tRNA-type gene B box, we show that TFIIIC2 is also required for Xenopus laevis 5S rRNA gene transcription. TFIIIC2 protected an approximately 30-base-pair (-10 to + 18) region of a Xenopus 5S rRNA gene from DNase I digestion. This region, which spanned the transcription start site, included sequences that are highly conserved among eucaryotic 5S rRNA genes and have no homology with the B-box sequence of tRNA genes. Mutation of the TFIIHC2-binding site reduced transcription of the 5S rRNA gene by a factor of 10 in HeLa cell extracts. Methylation of C residues within the TFIIIC2-binding site interfered with binding of TFIIIC2. These results suggest a role of the TFIIIC2-binding sequence in 5S rRNA gene transcription. In addition, the 5S rRNA gene binding site and the tRNA-type gene B-box sequence did not compete with each other for binding to TFIIIC2 any better than did an unrelated DNA sequence, indicating that TFIIIC2 interacts with 5S rRNA genes and tRNA-type genes through separate DNA-binding domains or polypeptides. stably associated with a SS rRNA gene when challenged with another 5S rRNA gene in a template commitment assay (25). However, TFIIIA plus TFIIIC together form a stable complex on 5S genes (2, 22, 25, 52). Another transcription factor, TFIIIB, is also required for transcription of both tRNA-type and 5S rRNA genes (reviewed in reference 20). TFIIIB has not been observed to bind to DNA but is stably bound to transcription complexes (2, 12, 44), perhaps through proteinprotein contacts with TFIIIC and, in the case of the 5S rRNA genes, possibly TFIIIA. The majority of studies of transcription factor interactions with the 5S rRNA gene have been performed by using the Xenopus gene and Xenopus or human extracts and factors. Human cell extracts and partially purified TFIIIB and TFI IIC (in concert with Xenopus TFIIIA) efficiently transcribe the Xenopus gene. The study of human factor interactions with the human gene has been precluded by the lack of a cloned transcriptionally active human 5S rRNA gene. Recently, the human 5S rRNA gene was chemically synthesized on the basis of the sequence of the major 5S rRNA species present in human cells (43). This gene does not contain the 5'- or 3'-flanking sequences that would be present in a gene cloned from a chromosomal locus and would therefore prevent examination of the effect of potential extragenic promoter elements. When purified from Saccharomyces cerevisiae, TFIIIC appears to have two domains. One domain is responsible for interaction with the A-box promoter element of a tRNA gene, and the other is responsible for interaction with the B box (18, 27). Purification of TFIIIC activity from human cells in culture resulted in the separation of two activities required to reconstitute TFIIIC activity, designated TFIIIC1 and TFIIIC2 (57). TFIIIC1 and TFIIIC2 activities can be separated by either anion-exchange chromatography (57) or sequence-specific DNA affinity chromatography (11). A potentially analogous separation has been effected for the silkworm PolIII in vitro transcription system (31). Both

Transcription factor TFIIIC is required for transcription of both tRNA-type and 5S rRNA genes in eucaryotes (42; reviewed in reference 20). This requirement of the same transcription factor for both major types of genes transcribed by RNA polymerase III has been rationalized in terms of the A-box promoter element found in both types of genes. The basic promoter elements of the tRNA-type genes (which include the adenovirus type 2 VA I RNA genes [15]) and of 5S rRNA genes (reviewed in reference 20) are diagrammed in Fig. 1. The tRNA-type genes contain two essential promoter elements, termed the A and B boxes. These are approximately 11 base pairs each and are separated by about 35 base pairs in the most actively transcribed tRNA genes, although the intervening distance can vary considerably. 5S rRNA genes also contain an A box that is functionally equivalent to the A box of tRNA genes (9), but there is no recognizable B-box homology in the 5S rRNA gene. Rather, another sequence, termed the C box, is essential for 5S rRNA transcription in Xenopus laevis oocyte extracts (3, 34-36, 38). These promoter elements have been shown to be binding sites for polymerase III (PolIII) transcription factors. TFI IIA interacts closely with the 5S rRNA gene C box (39) and protects both the C-box and A-box regions from DNase I digestion (14). TFIIIC was shown to bind with high affinity to the B box of tRNA genes (1, 25, 27, 48) and with lower affinity to the A box (17, 27, 47, 48) in nuclease protection assays. TFIIIC was also observed to bind specifically to 5S rRNA genes (8, 44), but its affinity was too weak to remain Corresponding author. t This paper is dedicated to Arthur Kornberg on the anniversary of his 70th birthday. *

t Present address: Department of Biochemistry, Stanford University Medical Center, Stanford, CA 94305. § Present address: Department of Biochemistry and Biophysics, University of California, San Francisco, CA 94143. 4941

4942

MOL. CELL. BIOL.

FRADKIN ET AL.

A

-+8

-+19

A Box

I

(31-93)

--X.,

B Box GG1rCGANTCC

TRGCN

YNGG

MGCCMGCAGG

B

+49

+60

7I[Box Z

T1GGATGGGAG +80

+90

C Box

+120

FIG. 1. Structures of the intragenic promoters of the tRNA-type (A) and the 5S rRNA (B) genes. The numbers shown correspond to nucleotide positions relative to the start site of transcription in the indicated gene. The homology between the A-box regions is indicated. The consensus sequence for the tRNA-type gene promoter elements is taken from Pieler et al. (35). Arrows indicate transcription initiation nucleotides. N, Any nucleotide; R, purine; Y, pyrimidine.

TFIIIC1 and TFIIIC2 activities are required, together with TFIIIB and Pollll, for in vitro transcription of tRNA-type genes (12, 57). TFIIIC2 has a molecular weight of approximately 500,000 and binds with high affinity (Ks = 2 x 101l M-1) to the B box of the VA I gene through a polypeptide of approximately 230,000 daltons (4). TFIIIC1 has a molecular weight of approximately 200,000 and appears to interact with the A-box region of the VA I gene (57). The separation of human TFIIIC into two activities required for the transcription of tRNA-type genes raised the question of which of these fractions was required for transcription of 5S rRNA genes. TFIIIC1 seemed likely to be required for the transcription of both types of genes, since it appears to interact with the A-box promoter element common to both gene types. But the requirement of TFIIIC2 was unclear, since 5S rRNA genes do not contain the B box to which the factor bind in tRNA genes. Here we report that TFIIIC1 and TFIIIC2 are both required for 5S rRNA gene transcription. DNase I footprinting showed that affinity-purified TFIIIC2 bound to a highly conserved region of 5S rRNA gene from -10 to + 18. Methylation of this binding site by HaeIII methylase interfered with TFIIIC2 binding. A linker-scanning mutation from +8 to 15 (40) resulted in a 10-fold reduction of in vitro transcription of the 5S rRNA gene in HeLa nuclear extracts. This region of the 5S rRNA gene bears no homology to the B box of tRNA-type genes. Furthermore, the two sequences do not compete with each other's binding to highly purified TFIIIC2, suggesting that the interaction of TFIIIC2 with the 5S rRNA gene sequence is mediated by a different polypeptide or domain of the protein than the one required for its interaction with the B box of tRNA-type genes. MATERIALS AND METHODS Recombinant plasmids. pVA I contains the adenovirus type 2 VA I gene upon a SalI-BalI fragment cloned into pUC18 (M. Schmidt, unpublished data). pXbsF201 contains the Xenopus borealis somatic 5S rRNA gene cloned in pUC9 (37; the kind gift of F. Razvi and A. Worcel). pMaxi is the +20 X. borealis maxigene cloned in pBR322 (38). The LS X/Y-MT mutant series (graciously provided by M. Sands and D. Bogenhagen) consists of linker-scanning mutants of the wild-type X. borealis gene cloned in tandem with a 5S rRNA maxigene in a pBR325 derivative (40). Cell culture and extract preparation. Spinner HeLa and 293 cells were grown in S-MEM (GIBCO Laboratics, Grand

Island, N.Y.) supplemented with 5% newborn calf serum, amino acids, and glutamine. Cells were harvested at 4 x i05 to 10 x 105 cells per ml, and nuclear extracts (13) or high-salt S100 extracts (56) were prepared. DNase I footprinting assays. Binding reactions and DNase I footprinting assays were done as previously described (57). DNase I footprinting reaction mixtures contained approximately 134 ng and 1 ,ug of TFIIIA and TFIIIC2, respectively. The radioactive probes for the DNAse protection assay were prepared as follows. For the VA I gene probe, pVA I was linearized with Sall, dephosphorylated, and 5' end labeled with [8-32P] ATP, using T4 polynucleotide kinase. A secondary digestion with EcoRI was performed, and the resultant 247-base-pair probe was purified by polyacrylamide gel electrophoresis, followed by electroelution, phenol extraction, and ethanol precipitation. For the 5S rRNA gene probe, the noncoding stand was labeled by kinase treatment at the Hindlll site, secondary digestion with BamHI, and purification as described above. The coding strand was labeled by 3' end filling at the HindIlI site with the Klenow fragment of polymerase I and one a-32P-labeled deoxynucleoside triphosphate. The secondary digestion and purification were as for the noncoding strand.

Oligonucleotide competition footprinting experiments were performed by using ligated oligonucleotides. The VA I gene B-box sequence used was 5'-GGATTCCGGGGTTCG AACCCC-3'. The 5S rRNA gene TFIIIC2-binding site sequence used was 5'-TGCTCGCCTACGGCCATACCACCC T-3'. The human T-cell leukemia virus type II long terminal repeat sequence used (kindly provided by R. Gaynor) was 5'-ACAAGGCTCTGACGATTACCCCCTGCC-3'. HaeIII methylase was used as recommended by the manufacturer (New England BioLabs, Inc., Beverly, Mass.). The completeness of methylation was checked by the resistance of the radioactive methylated probe to digestion by HaeIII restriction endonuclease. In vitro transcription. Transcription reactions and analysis of transcripts were done as previously described (16). Transcriptions were performed for 90 min at 30°C, and reaction mixtures contained 250 ng of template DNA in a 40-,ul volume. Unless indicated otherwise, all reconstituted transcription reaction mixtures contained 20 ,ul of TFIIIC1 (0.8 mg of protein per ml), 5 ,ul of TFIIIC2 (1.2 mg of protein per ml), and 5 ,ug of TFIIIB-PolIII. In addition, 140 ng of purified TFIIIA was included in 5S rRNA transcription reaction

5S rRNA GENE-TFIIIC INTERACTION

VOL. 9, 1989

mixtures. Transcription in crude extracts was performed with 20 ,ug of total protein. Preparation of transcription factors. TFIIIA was prepared from stage 1 Xenopus oocytes (Xenopus 1, Ann Arbor, Mich.) as previously described (46). Silver staining revealed only trace contaminants; the purity of the preparation was estimated to be greater than 95%. TFIIIB was prepared from 293 cell high-salt S100 extracts (57) by step elution from phosphocellulose, followed by gradient elution from phosphocellulose. It was previously shown that this second gradient elution is required to free TFIIIB preparations from contaminating TFIIIC1 (57). PoIIII copurifies with TFIIIB such that it is not a limiting component for in vitro transcription. TFIIIC1 was prepared from 293 cell high-salt S100 extracts by fast protein liquid chromatography (FPLC) Mono Q ion-exchange chromatography of the phosphocellulose C fraction as previously described (57). TFIIIC2 was purified from HeLa cell nuclear extracts (13) by a method described in detail elsewhere (58). Briefly, ammonium sulfate-precipitated nuclear extract was dissolved and chromatographed over Sephacryl S300 (Pharmacia, Inc., Piscataway, N.J.) gel filtration medium. The protein peak in the void volume was bound to and step eluted from a VA I gene B-box oligonucleotide affinity column. Active fractions were then applied to and step eluted from an FPLC Phenyl-Superose hydrophobic interaction column. Active fractions were then applied to and gradient eluted by a salt gradient from an FPLC Mono Q ion-exchange column. This preparation was used in the in vitro transcription reactions shown in Fig. 3 and was approximately 25,000-fold purified from the crude extract. For all other experiments, HeLa cell nuclear extract was prepared and chromatographed on phosphocellulose to prepare the C fraction as described elsewhere (13). Phosphocellulose C fraction protein was precipitated by addition of ammonium sulfate to 45% saturation (at 4°C) over 30 min. After being stirred for 30 min on ice, the precipitate was collected by centrifugation at 11,000 rpm for 30 min in an SS34 rotor (Ivan Sorvall, Inc., Norwalk, Conn.) and dissolved in buffer Z (0.1 M KCI, 20 mM N-2-hydroxyethylpiperazine-N'-2-ethanesulfonic acid [HEPES; pH 8.0], 0.1 ,uM dithiothreitol, 20%o glycerol). A 500-mg amount of protein in 50 ml of buffer Z was loaded on a 1.0-liter Sephacryl S300 column (5 by 45 cm) preequilibrated with buffer Z. The protein peak eluted in the void volume was pooled (80 ml; 2 mg of protein per ml), diluted to 90 mM KCl with water, and made 5 mM MgCl2, and poly (dI-dC)-poly (dI-dC) was added to 9 ,ug/ml. The protein was incubated in batch for 1 h with a B-box DNA affinity resin prepared according to a modification of the procedure of Kadonaga and Tjian (24) as described in detail elsewhere (58) with 5 to 8 mg of protein per ml of resin. The slurry was poured into a silanized Econo column (2.5 by 10 cm; Bio-Rad Laboratories, Richmond, Calif.), the flowthrough was collected, and the column was washed with three column volumes of buffer Z plus 0.1% Nonidet P-40. After elution with 3 column volumes of buffer Z made 0.2 M KCl-0.1% Nonidet P-40, TFIIIC2 activity assayed by DNase I footprinting on the VA I gene was eluted with buffer Z made 0.4 M KCl-0.1% Nonidet P-40-50 ,ug of acetylated bovine serum albumin per ml (58). Phenylmethysulfonyl fluoride and aprotinin (Trysylol; FBA Pharmaceuticals, New York, N.Y.) were added to 0.5 mM and 100 Kallikrein units per ml, respectively. This preparation is called affinitypurified TFIIIC2 and had approximately 400-fold-higher

Fraction

4943

B

A

TFmIA B

CI ± + C2 ++

VAI-

t

++ + + +

+0

0 +0 + c) U)

I

*

lane 1 2 3

-5

456 789

FIG. 2. Requirement for TFIIIC2 and TFIIIC1 for in vitro transcription of the Xenopus 5S rRNA gene. Transcription reactions were performed by using fractionated PollIl transcription factors or HeLa cell S100 extracts as described in Materials and Methods. TFIIIC2 purified through the oligonucleotide affinity step was used. A 250-ng amount of pVA I (containing the adenovirus type 2 VA I gene; lanes 1 to 3) or 250 ng of pXbsF201 (containing the X. borealis wild-type 5S rRNA gene; lanes 4 to 9) was included in the transcription reactions. An autoradiograph of labeled RNA products from the reactions is shown. The factors included in the incubation mixture are indicated above the lanes. a-Amanitin (cA) was present at 300 ,ug/ml where indicated.

specific activity in a DNase I footprinting assay on the VA I gene than did the nuclear extract.

RESULTS TFIIIC2 is required for 5S rRNA transcription. To test whether TFIIIC2 is required for 5S rRNA transcription, in vitro transcription reactions were performed with partially purified transcription factors. As shown previously (11, 12, 57), both the TFIIIC1 and TFIIIC2 fractions were required for VA I RNA transcription (Fig. 2, lanes 1 to 3). In addition to TFIIIA and fraction B (which provides both PolIII and TFIIIB activities [12, 13]), both TFIIIC1 and TFIIIC2 were also required for 5S rRNA transcription (Fig. 2, lanes 4 to 9). To determine whether the same TFIIIC2 protein was required for both VA I RNA and 5S rRNA transcription, reactions were performed with highly purified TFIIIC2. TFIIIC2 was purified approximately 25,000-fold from a HeLa cell nuclear extract as described previously (58). Fractions from the final FPLC Mono Q column eluted with a KCl gradient were assayed for in vitro transcription of VA I RNA (Fig. 3A) and 5S rRNA (Fig. 3B) in reaction mixtures containing PolIll and the other required PollIl transcription factors. Using the VA I gene as a template, TFIIIC2 transcriptional activity was observed in fractions 9 to 12, with

4944

MOL. CELL. BIOL.

FRADKIN ET AL.

A. 0

* 4

tb

* ,

0_

_4

.

.p OmP

4x

-..-

c

W. W-

-~0-

-VAl

FRACTION

FT 6 7 8 910 111213141516182022

B.

s + -t. ,w _ _

a

0 a

.A

a

*

,

-5S

C. 205-

_- 230 -

116- _

97-X

*-110

.

-

j

as

100

66-1 45 _L

M

OFT

2021 22t M--FRACTION 1618 27 8910111214 19 1 7 l O

FIG. 3. Copurification of TFIII2 activities required for VA I and 5S rRNA transcription. Fractions (5 ,ul) from the final step in TFIIIC2 purification, gradient elution from an FPLC Mono Q column (58), were assayed for VA I RNA transcription (A) and 5S rRNA transcription (B). (C) Silver-stained sodium dodecyl sulfate-polyacrylamide gel of the column fractions.

maximal activity in fraction 10. The peak of TFIIIC2 activity required for 5S rRNA transcription precisely coeluted with the activity required for VA I transcription (Fig. 3B). A silver-stained sodium dodecyl sulfate-polyacrylamide gel of these fractions is shown in Fig. 3C. The darkeststaining bands near the bottom of the gel were due to bovine serum albumin carrier, which must be added to purified TFIIIC2 to maintain activity. Bovine serum albumin, which was added to the protein fraction applied to the Mono Q column, eluted in a peak that overlapped with the peak of TFIIIC2 activity. Fractions 9 to 12 contained a polypeptide of =-230 kilodaltons (kDa) equivalent in size to the TFIIIC2 polypeptide that had been UV cross-linked to VA I DNA (4). Two other polypeptides of -110 and =100 kDa were also

prominent in these fractions and coeluted with the =230-kDa polypeptide. Since these polypeptides coeluted with the 230-kDa polypeptide during earlier steps in the purification and native TFIIIC2 has a molecular size of -=500 kDa, the TFIIIC2 protein may be composed of these three polypeptides (58). The =110- and -100-kDa polypeptides might also be partially proteolyzed forms of the -230-kDa polypeptide. On the basis of the results of Fig. 3, we conclude that the same or very tightly associated polypeptides in TFIIIC2 are required for VA I RNA and 5S rRNA transcription. Sequence-specific binding of TFIIIC2 to the 5S rRNA gene. Since TFIIIC2 binds with high affinity to an essential promoter region of tRNA-type genes, we asked whether this protein might bind to specific promoter elements of the 5S

*_ ~ C

VOL. rRNA GENE-TFIIIC INTERACTION VS 9, 1989 U

VI C3