MOLECULAR AND CELLULAR BIOLOGY, July 1996, p. 3308–3316 Vol. 16, No.

7
0270-7306/96/$04.0010
Copyright q 1996, American Society for Microbiology

TFG3/TAF30/ANC1, a Component of the Yeast SWI/SNF Complex

That Is Similar to the Leukemogenic Proteins ENL and AF-9
BRADLEY R. CAIRNS, N. LYNN HENRY, AND ROGER D. KORNBERG*
Department of Structural Biology, Stanford University School of Medicine, Stanford, California 94305
Received 16 November 1995/Returned for modification 9 January 1996/Accepted 11 April 1996

The SWI1/ADR6, SWI2/SNF2, SWI3, SNF5, and SNF6 gene products are all required for proper transcrip-
tional control of many genes in the yeast Saccharomyces cerevisiae. Genetic studies indicated that these gene
products might form a multiprotein SWI/SNF complex important for chromatin transitions preceding tran-
scription from RNA polymerase II promoters. Biochemical studies identified a SWI/SNF complex containing
these and at least six additional polypeptides. Here we show that the 29-kDa component of the SWI/SNF com-
plex is identical to TFG3/TAF30/ANC1. Thus, a component of the SWI/SNF complex is also a member of the
TFIIF and TFIID transcription complexes. TFG3 interacted with the SNF5 component of the SWI/SNF com-
plex in protein interaction blots. TFG3 is significantly similar to ENL and AF-9, two proteins implicated in hu-
man acute leukemia. These results suggest that ENL and AF-9 proteins interact with the SNF5 component of
the human SWI/SNF complex and raise the possibility that the SWI/SNF complex is involved in acute leukemia.

Many genetic and biochemical experiments have established ATPase activity associated with the SWI2/SNF2 component
an important role for chromatin in the repression of transcrip- copurifies with the complex (3, 7, 42). Recent studies indicate
tion (15). For example, a DNA molecule containing a binding that the yeast SWI/SNF complex utilizes ATP to perturb nu-
site for a transcriptional activator may display a much lower cleosome structure and to assist GAL4 derivatives in their
level of activator occupancy when the binding site is packaged binding to nucleosomal DNA (7). Similar results have been
in a nucleosome (59). In addition, there is considerable evi- obtained with highly purified fractions derived from human
dence that many nonhistone chromatin components assist in cells containing a complex with a homolog of SWI2/SNF2 (23,
the establishment or maintenance of chromatin structure and 30).
contribute to repression. Before transcription can occur, this We and others have previously demonstrated that SWI/SNF
repression must be overcome to make the DNA accessible to complex contains several polypeptides in addition to those
both activators and the basal transcription machinery. Such described above (3, 7). Here, we identify one of these polypep-
chromatin structural changes apparently occur at the yeast tides as TFG3, a protein which is also an integral member of
PHO5 and SUC2 promoters, where positioned nucleosomes two other yeast transcription factor complexes, TFIID and
are disrupted under derepressing growth conditions (20, 48). TFIIF.
Importantly, this disruption requires neither DNA replication
nor transcription from the promoter to occur (48). The ob- MATERIALS AND METHODS
served chromatin transition thus precedes transcription and is
not coupled to it. Yeast strains. A Saccharomyces cerevisiae strain lacking the TFG3 gene was
prepared by one-step gene replacement with the knockout plasmid pPL2 and the
Genetic studies from several laboratories identified five of parent strain YPH499 (MATa his3-D200 trp1-D3 ura3-52 leu2 ade2-101 lys2-806).
the components of a multiprotein complex which may help The disruption plasmid pPL2 was prepared and cut as described previously (18).
mediate chromatin transitions. They include the products of The tfg3D::LEU2/TFG3 heterozygous diploid strain is a derivative of W303 and
the SWI1/ADR6, SWI2/SNF2, SWI3, SNF5, SNF6, and SNF11 was prepared as described previously (18). The swi1D strain CY58 was a gift of
Craig Peterson, University of Massachusetts Medical Center, Worcester, Mass.,
genes, all of which (except SNF11) are required for proper and is congenic to YPH499.
control of the same set of yeast promoters (13, 17, 32, 33, 40, Purification of the SWI/SNF complex. The SWI/SNF complex was purified
43, 54). Their action in a complex was inferred from experi- from two strains, BJ926 and Fleischmann’s active yeast (gift of Fleischmann’s
ments with LexA-SNF2 and LexA-SNF5 fusion proteins; both Yeast Inc., Oakland, Calif.). Extractions from 12 kg of cells and the first three
chromatographic steps (Bio-Rex 70, DEAE-Sephacel, and hydroxylapatite) were
chimeric proteins required all five SWI and SNF proteins to performed as described previously (47). Peak fractions from hydroxylapatite
activate transcription from a promoter containing a LexA up- were further resolved on Mono Q. Adsorbed proteins were eluted with a linear
stream activating sequence (UAS). Additional genetic studies gradient of 200 to 1,000 mM potassium acetate in buffer A (20 mM Tris-acetate
suggest that these proteins antagonize the repressive effects of [pH 7.6], 20% glycerol, 1 mM dithiothreitol, 1 mM EDTA, 2 mg of chymostatin
per ml, 2 mM pepstatin A, 0.6 mM leupeptin, 2 mM benzamidine, 1 mM phen-
chromatin (4, 58). For example, extragenic suppressors of both ylmethylsulfonyl fluoride, 0.01% Nonidet P-40). Peak fractions of the SWI/SNF
swi and snf mutants reveal mutations in genes encoding his- complex eluting at 750 mM potassium acetate were diluted twofold in buffer A
tones and nonhistone chromatin components (29, 51). In ad- lacking potassium acetate, bound batchwise for 4 h at 48C to an SNF6 antibody
dition, the chromatin transitions that occur at the SUC2 pro- immunoaffinity resin (2.5-ml bed volume) (described below), washed in a column
with 50 ml of buffer A containing 700 mM potassium acetate and 0.2% Nonidet
moter in the absence of glucose repression in wild-type yeast P-40, and eluted with 5 M urea. The purification was monitored by immunoblot
cells do not occur in snf2 or snf5 mutants (20). Biochemical analysis with antisera to SNF6 and SWI3.
purification verified that these SWI and SNF proteins form The complexes derived from the two cell sources had identical polypeptide
a stable complex and demonstrated that a DNA-dependent compositions and were therefore combined to obtain peptide sequence informa-
tion. To the combined 5 M urea eluates (8 ml) were added 120 mg of insulin,
deoxycholate to a final concentration of 0.2%, and trichloroacetic acid to a final
concentration of 10%. After incubation for 1 h at 48C and centrifugation at
13,000 3 g for 20 min at 48C, the resulting pellets were washed with 200 ml of
* Corresponding author. Phone: (415) 723-6988. Fax: (415) 723- acetone at 48C, resuspended in 23 sodium dodecyl sulfate (SDS) loading buffer,
8464. and heated at 658C for 2 min. Approximately 200 mg of SWI/SNF complex was

3308
VOL. 16, 1996 TFG3 IS A COMPONENT OF YEAST SWI/SNF 3309

separated in a single lane of an SDS–9% acrylamide gel and transferred for 12 h and a BamHI site at the 39 end of the TFG3 gene by PCR with Taq polymerase,
at 10 V/cm to a polyvinylidene difluoride membrane (Bio-Rad Trans Blot) in a plasmid pPL2 (a genomic subclone of TFG3), and the oligonucleotide primers
transfer buffer containing 25 mM Tris (pH 8.3), 192 mM glycine, 10% methanol, 59-ATATGAATTCGTAGCGACAGTAAAAAGAACCATC-39 and 59-ATAT
and 0.005% SDS. Proteins were revealed by staining with Ponceau S and excised. GGATCCTTACTCGGTATTTTTCTT-39. The 59 oligonucleotide primer in-
Peptide sequencing was performed by the Harvard Microchemistry Facility, cludes the three amino acids encoded by the first exon of TFG3 and directs
Cambridge, Mass. amplification from the second exon. Amplified DNA products were digested with
An alternative procedure was used to purify the complex to near homogeneity EcoRI and BamHI and cloned into plasmid pGBT9 (2mm origin, TRP1), which
without the use of an immunoaffinity column. Peak fractions from the Mono Q directs expression of GAL4 1–147 from the yeast ADH1 promoter (a gift of P.
column were pooled, dialyzed against buffer B (20 mM N-2-hydroxyethylpipera- Bartel and S. Fields, State University of New York, Stony Brook, N.Y.). The
zine-N9-2-ethanesulfonic acid [HEPES; pH 7.6], 20% glycerol, 1 mM dithiothre- DNA sequence flanking the EcoRI junction was confirmed by DNA sequencing.
itol, 1 mM EDTA, 2 mg of chymostatin per ml, 2 mM pepstatin A, 0.6 mM Plasmid pLEXA-TFG3 was prepared by cloning the EcoRI-BamHI fragment
leupeptin, 2 mM benzamidine, 1 mM phenylmethylsulfonyl fluoride, 0.01% Non- from pGAL4-TFG3 into the corresponding sites in pSH2-1.
idet P-40), containing 100 mM potassium acetate, applied to a TSK-heparin b-Galactosidase assays. Transcriptional activation in vivo was quantified by
column, and eluted in buffer B with a linear gradient of 200 to 800 mM potassium b-galactosidase assays with the reporter plasmid pRS1840 (2mm origin, URA3),
acetate. Peak fractions (from 450 to 490 mM potassium acetate) were pooled, which contains a single high-affinity binding site for the Escherichia coli repressor
dialyzed against buffer B, applied to a Mono S column, and eluted in buffer B LexA (a gift of S. Hanes). Strains containing pRS1840 (alone or in combination
with a linear gradient of 200 to 800 mM potassium acetate. Peak fractions (from with plasmids expressing LexA fusion proteins) were grown to an optical density
350 to 380 mM potassium acetate) were pooled, dialyzed against buffer A at 600 nm of 0.5 to 0.8 in 100 ml of synthetic medium containing the appropriate
containing 100 mM potassium acetate, applied to a DEAE-Sephacel column, and amino acids and 2% glucose. The cell suspension was centrifuged at 5,000 3 g for
eluted in buffer A with a linear gradient of 200 to 800 mM potassium acetate. The 10 min, washed with sterile water, harvested, and suspended in 0.5 ml of buffer
peak of the SWI/SNF complex eluted at 470 mM potassium acetate. Because of A containing 400 mM sodium chloride. The cells were disrupted by beating with
the presence of contaminants eluting at 400 to 450 mM potassium acetate, glass beads for 9 min at 48C. The extract was centrifuged at 13,000 3 g for 10 min,
fractions preceding the peak were about 25 to 40% pure and fractions after the and the supernatant was centrifuged again at 13,000 3 g for 10 min. b-Galac-
peak were about 40 to 70% pure, as determined by electrophoretic separation tosidase activities, determined as described previously (3), are given in units per
and staining with Coomassie blue dye. milligram of protein in the whole-cell extract.
Immunoprecipitation and immunodepletion experiments. Crude SNF6 and Far Western (protein-blot) analyses. A DEAE-Sephacel fraction (4 mg; ap-
affinity-purified TFG3 antiserum were coupled to protein A-Sepharose as de- proximately 70% SWI/SNF complex) and 1 mg of homogeneous TFIIF were
scribed previously (3). To prepare anti-SNF6 and anti-TFG3 immune complexes, separated on an SDS–8% acrylamide gel, transferred to nitrocellulose, and
the peak SWI/SNF fraction from the Mono S column (25 ml, 15 mg) was first renatured as described by Vinson et al. (65). The filter was incubated overnight
diluted twofold in buffer A lacking potassium acetate and precleared with 20 ml at 48C in hybridization buffer (20 mM HEPES [pH 7.6], 75 mM KCl, 0.1 mM
of 50% protein A-Sepharose beads (in buffer A containing 0.1 M potassium EDTA, 2.5 mM MgCl2, 1 mM dithiothreitol, 0.05% Nonidet P-40, 1% dry milk),
acetate). Samples were sedimented in a microcentrifuge, and the supernatants probed with 5 ml (10% of the in vitro transcription-translation mixture [Promega
were incubated for 3 h at 48C with either 20 ml of 50% protein A-Sepharose Inc.]) of 35S-labelled full-length TFG3 protein, washed with hybridization buffer,
beads coupled to anti-SNF6 or 20 ml of 50% protein A-Sepharose beads coupled and exposed to XAR film for 5 days.
to anti-TFG3. The beads were sedimented in a microcentrifuge, and the resulting
“IP” supernatant was conserved, while immune complexes on the beads were
washed three times with 500 ml of buffer A containing 600 mM potassium acetate
RESULTS
and 0.2% Nonidet P-40 and eluted twice with 20 ml of 5 M urea. Large-scale purification of the yeast SWI/SNF complex. The
To prepare anti-SNF6 immune complexes with SWI/SNF lacking TFG3, the
peak fraction from Mono S (60 ml, 40 mg; ,5% SWI/SNF complex) was pre-
yeast SWI/SNF complex was purified from whole-cell extracts
cleared with 50 ml of 50% protein A-Sepharose beads, incubated for 4 h at 48C by three ion-exchange steps followed by anti-SNF6 immuno-
with 50 ml of protein A-Sepharose beads conjugated to anti-SNF6 antibodies, affinity chromatography. The anti-SNF6 immunoaffinity col-
centrifuged, and the supernatant (120 ml) was saved. Immune complexes were umn was washed extensively with a buffer containing 700 mM
washed three times with 1 ml of buffer A containing 500 mM potassium acetate
and 0.2% Nonidet P-40 and were boiled for 5 min in 40 ml of SDS loading buffer.
potassium acetate and nonionic detergent. The eluate of this
To load half of the IP supernatant in a single lane of an SDS-containing gel, the column, in 5 M urea, contained at least 11 polypeptides; under
material was precipitated with deoxycholate (0.2%) and trichloroacetic acid these elution conditions, SNF6 remained bound to the column
(10%). (Fig. 1). Immunoblot analysis confirmed that four of these
Coimmunoprecipitation assays with SWI/SNF complex and TFIIF lacking
TFG3. TFIIF lacking TFG3 (TFIIF*) was purified as described previously (18).
polypeptides were SWI1/ADR6, SWI2/SNF2, SWI3, and SNF5
Highly purified SWI/SNF complex from DEAE-Sephacel was dialyzed against (results not shown). One additional polypeptide with an ap-
buffer A containing 100 mM potassium acetate, and 20 ml of this fraction (6 mg; parent molecular mass of 20 kDa, encoded by the SNF11 gene,
40% SWI/SNF complex) was precleared with 25 ml of 50% protein A-Sepharose is observed only when the SDS-polyacrylamide gel is stained
(in buffer A containing 100 mM potassium acetate) by rocking for 1 h at 48C. The
sample was centrifuged, and the supernatant (40 ml) was mixed with 20 ml (50 ng)
with Coomassie blue dye (64). Previous work has identified
of near-homogeneous TFIIF* (in buffer A [400], where [400] indicates the four additional polypeptides (SWP59, SWP61, SWP73, and
millimolar salt concentration). The sample was diluted to a final volume of 100 SWP82) as members of the SWI/SNF complex and three other
ml with buffer A [100], creating a potassium acetate concentration of 180 mM. polypeptides (p29, p57, and p77) as possible members (3, 7).
The sample was rocked at 48C for 90 min, and then 40 ml of 50% protein
A-Sepharose beads conjugated to SNF6 antibody (in buffer A [100]) was added,
Protein p29 was present in all preparations.
creating a final potassium acetate concentration of 160 mM. The sample was p29 is identical to TFG3/ANC1. The polypeptides from ap-
rocked for 4 h at 48C and centrifuged briefly in a microcentrifuge at 4,000 3 g, proximately 200 mg of SWI/SNF complex were separated by
and the supernatant was removed and stored in liquid nitrogen. The pellets were SDS-polyacrylamide gel electrophoresis (PAGE), transferred
washed twice with 1 ml of buffer A [200] and once with 1 ml of buffer A [500] and
then eluted with two 25-ml aliquots of 5 M urea.
to a synthetic membrane, excised, and treated with a protease.
Antibodies. Polyclonal antisera to SNF6 and SNF5 were prepared as previ- The following high-confidence sequence from p29 was ob-
ously described (3). Polyclonal antiserum to SWI3 was a gift of Craig Peterson. tained: IEEQGWGGFPLDISVF. An identical sequence of
Affinity-purified TFG3 antiserum was prepared as previously described (18). residues is present in the TFG3 protein at positions 76 to 91.
Immunoblot analyses. All immunoblots were incubated for 8 h with a 1:500
dilution of primary antibody in TBS–2% milk for 8 h and then for 1 h with a
We have previously reported that TFG3 is an integral com-
1:2,000 dilution of goat anti-rabbit secondary antibody (Bio-Rad) conjugated to ponent of both yeast TFIID and yeast TFIIF, two factors
alkaline phosphatase. Immunoblots were developed as described previously (3). required for basal transcription by RNA polymerase II (18,
Plasmids. Plasmids directing the expression of LexA fusion proteins are all 19). Yeast TFIIF is a complex of three proteins: TFG1, TFG2,
derivatives of pSH2-1 (HIS3, 2mm origin), which directs the expression of LexA
1–87 from the ADH1 promoter (a gift of S. Hanes). Plasmids pLEXA-SNF2,
and TFG3. TFG1 and TFG2 are significantly similar to the
pLEXA-SNF5, pLEXA-SNF6 (gifts of B. Laurent), and pLEXA-GAL4 (pSH17- mammalian Rap74 and Rap30 proteins, respectively, which
4; a gift of S. Hanes) have been described previously (16, 32, 33). Plasmid form mammalian TFIIF (14, 18, 22). Both yeast TFIIF and
pLEXA-GRact (pGNLX, TRP1, 2mm origin; a gift of K. Yamamoto) directs mammalian TFIIF form a complex with RNA polymerase II,
expression of the first 452 amino acids of the rat glucocorticoid receptor fused to
amino acids 1 to 87 of LexA from the yeast GPD promoter (60). Plasmid
and each is required for polymerase II transcription in vitro
pGAL4-TFG3 contains the first 147 amino acids of GAL4 fused to the entire (18, 36). TFG1 and TFG2 are essential genes (18). In contrast,
TFG3 coding sequence. It was prepared by creating an EcoRI site at the 59 end TFG3 protein is at most stimulatory to transcription, and de-
3310 CAIRNS ET AL. MOL. CELL. BIOL.

acetate and 0.2% Nonidet P-40. SDS-PAGE analysis and stain-

ing with silver revealed only the other members of the SWI/
SNF complex (Fig. 2A). Although the immunoaffinity column
was eluted with 5 M urea, the antigen (TFG3) remained bound
to the antibody. In a converse experiment, SDS-PAGE and im-
munoblot analysis of immune complexes prepared with the
peak Mono S fraction and anti-SNF6 antibodies revealed a
protein with an apparent molecular mass of 29 kDa that comi-
grated with the TFG3 component of yeast TFIIF (Fig. 2B and
C, respectively). Other members of the SWI/SNF complex
such as SNF6 and SWI3 were fully depleted from the Mono S
fraction by anti-TFG3 antibodies, showing that all SWI/SNF
complexes in the fraction contained TFG3 (Fig. 2D). In addi-
tion, we have determined that TFG3 is not a component of
other yeast transcription factor complexes such as TFIIE or
TFIIH. We conclude that TFG3 is not simply an abundant
protein with a high nonspecific affinity for proteins but, rather,
an integral component of multiple complexes.
Deletion of TFG3 does not confer many of the phenotypes
associated with SWI/SNF complex. Mutations in or deletions
of either SWI1/ADR6, SWI2/SNF2, SWI3, SNF5, or SNF6 con-
fer similar phenotypes, including slow growth and defects in
transcriptional activation of several genes including SUC2 (en-
coding invertase) and HO (encoding an endonuclease required
for mating-type switching) (5, 39, 50, 58). Under anaerobic
conditions, swi and snf mutants grow very poorly on solid me-
dium containing either sucrose, galactose, or raffinose (Snf2
phenotype). Anaerobic conditions can be mimicked by the ad-
FIG. 1. Purified SWI/SNF complex contains a 29-kDa protein. The SWI/SNF dition of antimycin A (1 mg/ml; an inhibitor of electron trans-
complex was purified on a large scale by the immunoaffinity procedure as de- port in the respiratory chain) to the medium. This treatment
scribed in Materials and Methods. The eluate from the anti-SNF6 immunoaf-
finity column (5 mg) was separated in a SDS–10% acrylamide gel and stained dramatically reduces the growth rate of swi and snf mutants on
with Coomassie blue dye. sucrose, galactose, or raffinose but has only a modest effect on
their growth on glucose (39).
Welch and Drubin (56) have shown that tfg3/anc1 mutants
are temperature sensitive (Ts2), but a possible Snf2 phenotype
letion of the TFG3 gene confers a temperature-sensitive phe- was not reported. We prepared a strain lacking the TFG3 gene
notype (18, 56). Yeast TFIIF is known to form a stable com- and tested it for certain phenotypes and defects characteristic
plex with RNA polymerase II holoenzyme, which is a of swi and snf mutants. As observed with such mutants, the
combination of RNA polymerase II, SRB proteins, GAL11, tfg3D strain grew slowly at 308C on solid media lacking anti-
and several other polypeptides; this complex enables a re- mycin A and containing 2% glucose, compared with its wild-
sponse to activators (27). ANC1, which is identical to TFG3, type parent. Several of the phenotypes displayed by the tfg3D
was isolated by Welch et al. in a screen for mutants which strain, however, contrasted with phenotypes reported for mu-
fail to complement a temperature-sensitive allele of actin tants in other SWI/SNF complex components. For example,
(57). However, TFG3 protein resides exclusively in the nu- the presence of 1 mg of antimycin A per ml had little effect on
cleoplasm, which is devoid of actin, further suggesting a role the growth of tfg3D cells on solid rich medium containing
for TFG3 in transcriptional regulation (see Discussion) glucose, sucrose, galactose, or raffinose at either the permissive
(56). (308C) or semipermissive (348C) temperature. In addition, the
Immune complexes formed with anti-TFG3 antisera contain tfg3D strain grew more poorly in galactose than in sucrose, a
all other members of the SWI/SNF complex. Further evidence relative difference not observed with either the wild-type par-
that TFG3 is a member of the SWI/SNF complex was obtained ent or other swi and snf mutants. To ensure that the Snf1
by immunoprecipitation and immunoblot analyses. SWI/SNF phenotype was not due to a reversion of our haploid tfg3D
complex devoid of TFIIF or TFIID was prepared for this pur- strain, we observed the segregation pattern of the Snf pheno-
pose. Following three ion-exchange steps, chromatography type in eight tetrads derived from our tfg3D::LEU2/TFG3 het-
on Mono Q yielded SWI/SNF complex free of mediator and erozygous diploid strain. Each of the eight four-spore tetrads
TFIID (determined by immunoblot analyses with anti-TBP, tested produced two Leu1 Ts2 Snf1 spores and two Leu2 Ts1
anti-SRB4, and anti-SRB5 antibodies) but containing detect- Snf1 spores. These results show that the lack of TFG3 does not
able amounts of TFIIF. Mono S chromatography then sepa- cause a Snf2 phenotype. Finally, in contrast to a congenic
rated the SWI/SNF complex from TFIIF. After removal of swi3D strain, the tfg3D strain will not form colonies at 378C on
TFIIF, TFG3 cofractionated precisely with the other members rich media containing glucose.
of the SWI/SNF complex. Likewise, during further fraction- Additional differences between the tfg3D strain and strains
ation of TFIIF, TFG3 cofractionated precisely with the other lacking other SWI and SNF components were observed in
components of TFIIF (data not shown). transcriptional activation (Table 1). Others have shown that
The peak Mono S fraction (approximately 20% SWI/SNF fusion of the DNA-binding domain of the bacterial repressor
complex) was immunoprecipitated with affinity-purified anti- LexA (amino acids 1 to 87) to SNF2, SNF5, SNF6, GAL4, or
TFG3 antibodies coupled to protein A-Sepharose and then the activation domain of the rat glucocorticoid receptor (GRact)
extensively washed with a buffer containing 700 mM potassium creates fusion proteins that are potent activators when assayed
VOL. 16, 1996 TFG3 IS A COMPONENT OF YEAST SWI/SNF 3311

FIG. 2. Analysis of anti-SNF6 and anti-TFG3 immune complexes with SDS-PAGE and immunoblots. Immune complexes were formed with either anti-SNF6 or
anti-TFG3 antibody conjugated to protein A-Sepharose and the peak Mono S fraction (20 mg; ,5% SWI/SNF complex) as described in Materials and Methods. (A)
Half of the immune eluate from anti-SNF6 (lane 1) or anti-TFG3 (lane 2) immune complexes was separated on an SDS–10% acrylamide gel and stained with silver.
Since the immunoprecipitations are performed without reducing agents (to lower the background of antibody released), the SWI2/SNF2 protein gradually becomes
linked to the immunoaffinity column and can be detected in the immune pellet after other SWI/SNF components have been eluted with 5 M urea (data not shown).
(B) Analysis of the anti-SNF6 immune eluates (lane 1) and homogeneous TFIIF (lane 2) by SDS-PAGE and silver staining. (C) Immunoblot analysis of TFIIF and
the SWI/SNF complex with affinity-purified anti-TFG3 antisera. Homogeneous TFIIF (100 ng; lane 2) and one-fifth of the immune eluate from the anti-SNF6 im-
munoprecipitation (lane 1) were separated on an SDS–10% acrylamide gel, immunoblotted, and probed with affinity-purified anti-TFG3 antisera. (D) Anti-TFG3
antibodies can immunodeplete members of the yeast SWI/SNF complex. The untreated Mono S fraction (3.0 mg; lane 1), one-fifth of the IP supernatant (3.0 mg; lane
2), and two-fifths of the anti-TFG3 immune eluate (lane 3) were separated on an SDS–10% acrylamide gel, immunoblotted, and probed with either anti-SNF6 or anti-SWI3
antiserum.

with a b-galactosidase reporter plasmid containing a single was a potent activator in all four strains, further demonstrating
LexA-binding site (16, 31–33, 60). These studies demonstrated that activation by LexA-SNF2 does not require TFG3. Unlike
that the previously identified components of the SWI/SNF LexA-SNF2 or LexA-SNF5, LexA-SNF6 is reported not to
complex are required for these hybrid proteins to activate at require other components of the complex to activate transcrip-
full capacity. tion (33). We observed that LexA-SNF6 activated transcription
We found that LexA-GRact fusion protein was a potent at reduced capacity in the tfg3D strain compared with the
activator in both wild-type and tfg3D strains but did not activate wild-type strain. Immunoblot analysis with anti-SNF6 antibod-
transcription in the congenic swi1D strain. In addition, both ies revealed, however, that at least 10-fold less LexA-SNF6
LexA-SNF2 and LexA-SNF5 fusion proteins activated tran- protein was present in the tfg3D strain than in the wild-type
scription at full capacity in the tfg3D strain but no activation cells, which may account for our observation (data not shown).
was observed in a swi1D strain. To ensure that this observation Finally, LexA-GAL4 is a potent activator in the wild-type
was not due to a reversion in our tfg3D strain, we transformed strain but did not activate significantly in either tfg3D or swi1D
LexA-SNF2 into four strains derived from one tetrad of our cells. Again, we assessed LexA-GAL4 expression by immuno-
heterozygous tfg3D::LEU2/TFG3 diploid strain. LexA-SNF2 blot analysis and determined that at least 20-fold less LexA-
GAL4 protein is present in tfg3D transformants than in wild-
type transformants (data not shown). We have previously
TABLE 1. Potency of LexA-activator fusion proteins reported that the activation observed with a reporter plasmid
assessed in b-galactosidase assaysa
containing a single GAL4-binding UAS element is reduced
b-Galactosidase activity (U/mg) of: approximately sixfold in tfg3D cells (18). Thus, the dramatic
Activator expressed reduction in activation observed with LexA-GAL4 may be ex-
Wild type tfg3D swi1D
plained in part by a combination of these effects. These results
LexA-1-87 (control) 12 8 ND demonstrate that (in contrast to strains lacking SWI1, SWI2/
LexA-SNF2 2,039 1,651 5 SNF2, SWI3, SNF5, or SNF6 proteins) TFG3 protein is not
LexA-SNF5 2,353 1,980 4 required for certain activators to operate at full capacity. In
LexA-SNF6 8,300 1,124 127 addition, the activators tested that displayed a reduced capac-
LexA-GRact(1–452) 1,863 3,733 2
LexA-GAL4 6,200 32 65 ity to activate transcription are correspondingly present at re-
duced levels in tfg3D cells.
a
The indicated LexA fusion proteins were expressed and b-galactosidase All swi mutants are defective in the transcription of the HO
assays were performed as described in Materials and Methods. Transformants
also contained the reporter pRS1840, which contains one 22-bp LexA operator
gene, which encodes an endonuclease required for mating-type
upstream of the minimal GAL1 promoter fused to the b-galactosidase gene. The switching. To assess the extent to which TFG3 is required for
values reported are the average of at least four transformants. HO activation, we assessed the UAS activity of the SWI5/
3312 CAIRNS ET AL. MOL. CELL. BIOL.

FIG. 3. TFG3 interacts with SNF5 in a far Western analysis. (A) Far Western analysis with 35S-labelled TFG3 protein and renatured TFIIF or SWI/SNF complex.
Homogeneous TFIIF (1 mg, lane 1) and a DEAE-Sephacel fraction (4 mg, 70% SWI/SNF complex; lane 2) were separated on an SDS–8% acrylamide gel, transferred
to nitrocellulose, and renatured. The filter was probed with 35S-labelled full-length TFG3 protein, washed with hybridization buffer, and exposed to XAR film for 5
days. (B) The nitrocellulose filter was probed and developed sequentially with affinity-purified antiserum to TFG1, polyclonal antiserum to SNF5, and affinity-purified
antiserum to SWI3. (C) SDS-PAGE analysis of the DEAE-Sephacel fraction. The DEAE-Sephacel fraction (2 mg; 70% SWI/SNF complex) was separated on an
SDS–6% acrylamide gel and stained with silver.

PHO2(GRF10) element from the HO promoter (carried on TFG3 interacts with the SNF5 component of the SWI/SNF
plasmid M632) in wild-type and tfg3D cells (2). We have ob- complex in a far Western analysis. To determine which com-
served a 10- to 20-fold reduction in b-galactosidase levels in ponent(s) of the SWI/SNF complex interacts with TFG3, pro-
certain swi and snf mutant cells containing this reporter rela- tein blots of the complex were probed with 35S-labelled TFG3
tive to wild-type cells (2a). Extracts from wild-type cells con- protein. A fraction derived from DEAE-Sephacel (approxi-
tained 14 U of activity compared with 10 U observed from mately 70% pure SWI/SNF complex), along with a fraction of
tfg3D cells, indicating that deletion of TFG3 has little effect on homogeneous TFIIF, was separated by SDS-PAGE. The pro-
activation with this element. teins were transferred to nitrocellulose, renatured, and probed
A LexA-TFG3 fusion protein is only a weak activator. To with 35S-labelled TFG3 protein. Autoradiography revealed an
further investigate the role of TFG3 in transcriptional activa- interaction of TFG3 with a protein of approximately 110 kDa
tion, we fused the DNA-binding domain of GAL4 (amino acids in TFIIF and with a protein of approximately 120 kDa in the
1 to 147) or of LexA (amino acids 1 to 87) to the amino SWI/SNF complex (Fig. 3A). We have shown previously that
terminus of TFG3. The resulting vectors, pGAL4-TFG3 and the 110-kDa component of TFIIF is TFG1 and that the SWI/
pLEXA-TFG3, contained a high-copy 2mm origin and the SNF complex contains two proteins, SWI3 and SNF5, that
strong constitutive ADH1 promoter. Expression of LexA- migrate on SDS-PAGE at approximately 120 kDa (3, 18). To
TFG3 in tfg3D cells fully complemented their growth de- clarify the identity of the protein interacting with TFG3, the
fect at 308C and their temperature sensitivity at 378C, whereas blot was probed and developed sequentially with affinity-puri-
expression of the GAL4-TFG3 fusion protein in tfg3D cells fied antiserum to TFG1, polyclonal antiserum to SNF5, and
only partially complemented these defects. Immunoblot anal- affinity-purified antiserum to SWI3 (Fig. 3B). Superimposition
ysis with anti-TFG3 antibodies showed that GAL4-TFG3 pro- of the autoradiograph with the immunoblot clearly revealed
tein was expressed at slightly higher levels in wild-type than SNF5 as the interacting protein. By performing the immuno-
tfg3D transformants and that this level was approximately five- blot analysis in this order, we also verified that the DEAE-
fold lower than the amount of endogenous TFG3 protein in Sephacel fraction utilized did not contain any TFG1 protein.
wild-type cells (data not shown). Electrophoresis of the fraction in a 6% acrylamide–SDS gel
The ability of these TFG3 fusion proteins to activate tran- followed by staining with silver revealed no additional polypep-
scription was assessed with b-galactosidase reporter plasmids tide migrating near SNF5 (Fig. 3C). To help clarify the region
containing a single binding site for either GAL4 or LexA as the responsible for these interactions, a truncation of TFG3 that
UAS. GAL4-TFG3 protein activated transcription only weakly removed 88 amino acids from the carboxy terminus (TFG3DC)
in both wild-type and tfg3D cells. Extracts prepared from wild- was prepared. 35S-labelled TFG3DC failed to interact with
type or tfg3D cells expressing GAL4-TFG3 contained only 112 either SNF5 or TFG1, indicating that this region is required for
or 86 U of activity, respectively, compared with 5 U obtained these interactions or for proper protein folding.
with the reporter plasmid alone. Extracts prepared from wild- If TFG3 utilizes the same domain(s) for binding to SNF5 as
type or tfg3D cells expressing LexA-TFG3 contained only 35 or for interaction with TFG1, SNF5 and TFG1 might contain
29 U of activity, respectively, compared with 6 U obtained with regions of homology. To test this possibility, we performed a
a control reporter plasmid. The lack of strong activation by rigorous comparison of the protein sequences of SNF5 and
either LexA-TFG3 or GAL4-TFG3 may be a result of more TFG1 with the program PEP-Align (IntelliGenetics). No sig-
than one factor. First, either alone or as a member of a com- nificant regions of homology were detected, indicating that
plex, they may repress or interfere with activation. Second, TFG3 does not physically interact with identically positioned
these proteins may not interact well with the SWI/SNF com- residues in both SNF5 and TFG1.
plex or may be sequestered in the more abundant TFIID or Activation of the DNA-dependent ATPase does not cause the
TFIIF complexes. Finally, since LexA-SWI3 is only a weak SWI/SNF complex to disassemble, or to assemble with TFIIF
activator, although it will fully complement a SWI3 mutant or RNA polymerase II holoenzyme in vitro. Our observation
(54), not all fusions to SWI/SNF complex members necessarily that TFG3 is a member of three distinct complexes involved in
create potent activators. transcriptional activation raises the possibility that TFG3 pro-
VOL. 16, 1996 TFG3 IS A COMPONENT OF YEAST SWI/SNF 3313

vides a physical link between these complexes. To investigate

this possibility, we tested whether TFG3 displays homotypic
interactions. 35S-labelled TFG3 failed to interact with TFG3
protein derived from the SWI/SNF complex in far Western
analyses (data not shown). Furthermore, no significant binding
of either homogeneous TFIIF or nearly homogeneous RNA
polymerase II holoenzyme (a complex of core RNA polymer-
ase II and mediator complex, containing TFIIF and other
proteins) to immunoprecipitates of the SWI/SNF complex was
observed under either moderate- or high-ionic-strength condi-
tions (160 to 180 mM or 500 mM potassium acetate, respec-
tively). These potential interactions are not enabled by activa-
tion of the SNF2/SWI2 DNA-dependent ATPase, since prior
incubation of these factors and SWI/SNF immunoprecipitates
with Mg21, ATP, and DNA under conditions of low ionic FIG. 4. The composition and interactions of the SWI/SNF complex and
strength (75 mM potassium acetate) did not effect an associa- TFIIF purified from cells lacking TFG3. (A) Immunoblot analysis of immuno-
tion stable to moderate- or high-ionic-strength conditions. precipitations with anti-SNF6 antibodies and SWI/SNF complex partially puri-
fied from tfg3D extracts. Immune complexes were formed with anti-SNF6 anti-
Another possible reason for the presence of TFG3 protein in bodies conjugated to protein A-Sepharose and the peak Mono S fraction as
multiple complexes is that it masks or protects a region used described in Materials and Methods. Immune complexes were washed three
for interactions between these complexes. According to this times with 1 ml of buffer A containing 500 mM potassium acetate and 0.2%
idea, a conformational change by or the dissociation of TFG3 Nonidet P-40 and then boiled for 5 min in 40 ml of SDS loading buffer. To
load half of the IP supernatant, the material was precipitated with deoxycholate
from the SWI/SNF complex may uncover a site of interaction. (0.2%) and trichloroacetic acid (10%). The untreated Mono S fraction (20 mg;
To test this idea, we investigated whether activation of the approximately 5% SWI/SNF complex; lane 1), half of the precipitated IP super-
DNA-dependent ATPase would cause the dissociation of natant (20 mg, lane 2), and half of the anti-TFG3 immune precipitate (lane 3)
TFG3 (or other components) from the remainder of the com- were each separated on two SDS–10% acrylamide gels, immunoblotted, and
probed with either anti-SNF2, anti-SWI3, or anti-SNF5 antiserum. (B) TFIIF
plex. Immune complexes were formed with the peak Mono S lacking TFG3 does not interact with native SWI/SNF complex. TFIIF lacking
fraction and either anti-SNF6 or anti-TFG3 coupled protein TFG3 (TFIIF*) was purified as previously described (18). The peak DEAE-
A-Sepharose beads. The immune precipitates were washed Sephacel fraction of the native SWI/SNF complex (6 mg; 40% SWI/SNF com-
extensively and supplemented with a buffer and other compo- plex) was mixed with 20 ml (100 ng) of near-homogeneous TFIIF*. The sample
was rocked at 48C for 90 min, and 40 ml of 50% protein A-Sepharose beads
nents previously shown to be optimal for the DNA-dependent conjugated to SNF6 antibody was then added. The final potassium acetate
ATPase activity (3). None of the components of the SWI/SNF concentration in the sample was 160 mM. The sample was rocked for 4 h at 48C
complex were released from either anti-SNF6 or anti-TFG3 and centrifuged briefly in a centrifuged briefly in a microcentrifuge at 4,000 3 g,
immune complexes, indicating that either the activation of the and the supernatant was removed and stored in liquid nitrogen. The pellets were
washed extensively and then eluted with two 25-ml aliquots of 5 M urea.
DNA-dependent ATPase is not coupled to complex disassem-
bly or only a small percentage of the complexes in this assay
were involved in producing the observed DNA-dependent
ATPase activity (data not shown). tected in the immune precipitate. The SWI/SNF complex was,
A stable complex containing many SWI and SNF compo- however, efficiently precipitated, as evidenced by the lack of
nents is present in extracts from tfg3D cells. To determine SWI3 protein in the IP supernatant and its presence in the
whether a SWI/SNF complex could assemble in tfg3D cells, we precipitate (Fig. 4B). Identical results were obtained when the
monitored purification of the complex form tfg3D cells by im- reactions were performed either with a fourfold excess of SWI/
munoblot analyses with antisera against SWI2/SNF2, SWI3, SNF complex over TFIIF* or with stoichiometric amounts of
SNF5, and SNF6 proteins. All four of these proteins copurified TFIIF* or wild-type TFIIF.
in the four chromatographic steps used (data not shown). Im- TFG3 protein is not required for the DNA-dependent ATPase
munoprecipitation of the peak Mono Q fraction with anti- activity of SWI/SNF. Highly purified preparations of the SWI/
SNF6 antibodies coupled to protein A-Sepharose immuno- SNF complex contain a potent DNA-dependent ATPase that is
depleted SWI2/SNF2, SWI3, SNF5, and SNF6 (Fig. 4A). associated with the SWI2/SNF2 component (3, 7). To deter-
Although the immunoprecipitate was extensively washed with mine whether SWI/SNF complex lacking TFG3 protein retains
a buffer containing 500 mM potassium acetate and 0.2% Non- DNA-dependent ATPase activity, we assayed the supernatant
idet P-40, all four proteins remained associated. These results of the anti-SNF6 immunoprecipitation of the SWI/SNF com-
demonstrate that a stable complex consisting of at least SWI2/ plex lacking TFG3 for this activity. As a control, this peak
SNF2, SWI3, SNF5, and SNF6 proteins can be isolated from Mono S fraction was subjected to a mock immunoprecipitation
tfg3D cells. These results also indicate that TFG3 is not required with protein A-Sepharose beads conjugated to preimmune se-
for SNF5 to stably interact with other components of the complex. rum. Anti-SNF6 coupled beads were able to immunodeplete
TFIIF from a tfg3D strain does not interact stably with the the majority of the DNA-dependent ATPase activity; only 25%
SWI/SNF complex from wild-type cells. If separate domains of of the activity remained in the supernatant. In contrast, the
TFG3 are responsible for interaction with the SWI/SNF com- mock-treated supernatant exhibited 85% of its original activity.
plex and TFIIF, a single TFG3 molecule might tether the two These results suggest that the SWI/SNF complex isolated from
complexes together. To test this possibility, we purified TFIIF tfg3D cells is associated with a DNA-dependent ATPase activity.
to near homogeneity from tfg3D cells, isolating a complex of two
proteins, TFG1 and TFG2 (TFIIF*). Interaction of TFIIF* DISCUSSION
with the SWI/SNF complex was assessed by coimmunoprecipi-
tation with anti-SNF6 antibodies. No evidence of stable inter- Recent studies have demonstrated that many transcriptional
action was obtained; under conditions of moderate ionic activators do not exert their positive effects solely through
strength (160 mM potassium acetate), TFIIF* was not immu- factors required for basal transcription. Genetic and biochem-
nodepleted with anti-SNF6 antibodies, nor was TFIIF* de- ical studies identified TFIID and SWI/SNF as two large com-
3314 CAIRNS ET AL. MOL. CELL. BIOL.

plexes involved in transcriptional activation (11, 33, 43, 60). lated in both humans and D. melanogaster, and both homologs
TFIID, a complex of seven or eight polypeptides in addition to are believed to be associated in a large complex that contains
the TATA-binding protein, also participates in both basal and a SWI2/SNF2 homolog (9, 24). A human SWI/SNF complex
activated transcription (11, 44). Genetic studies with S. cerevi- (hSWI/SNF) has been partially purified, and contains a
siae identified several SWI and SNF polypeptides and sug- polypeptide which is a SWI2/SNF2 homolog (24, 30). Sequenc-
gested that these proteins formed a complex (13, 32, 33, 40, 43, ing of the other components of hSWI/SNF complex may still
54). Biochemical studies verified this prediction, confirmed the reveal a homolog of TFG3.
DNA-dependent ATPase activity of yeast SWI2/SNF2 protein, The presence of a SNF5 homolog in hSWI/SNF complex,
and revealed the presence of several additional tightly associ- coupled with the discovery of a SNF5 homolog in flies, suggests
ated polypeptides (3, 7). that the composition of the complex may be conserved in
In contrast to TFIID and TFIIF, both of which contain at evolution. Thus, homologs of other yeast SWI/SNF complex
least one component that is required for basal level transcrip- members are likely to be found in analogous complexes in
tion in vitro, no member of the SWI/SNF complex appears to other organisms. Here, we report that TFG3 interacts strongly
be required for the basal reaction (3, 18, 25). In addition, many with SNF5 in a far Western analysis, suggesting that TFG3
components of TFIID are encoded by essential genes whereas homologs may interact with SNF5 homologs in higher cells.
deletions of SWI/SNF complex members confer only slow- Studies on the mammalian homologs of SWI2/SNF2 and
growth or conditional phenotypes (18, 32, 43, 45). Genetic and SNF5 suggest the involvement of these proteins in diverse and
biochemical studies with S. cerevisiae and biochemical experi- important cellular processes. For example, the human ho-
ments with mammalian cells all suggest that the primary role of molog of SNF5, INI1, binds to human immunodeficiency virus
SWI/SNF complex may be to help transcriptional activators
type 1 integrase and stimulates its DNA-joining activity 20-fold
bind to sites occluded by chromatin components (6, 7, 23, 30).
in vitro (24, 49). In addition, studies on hBrm show a correla-
This work identifies the 29-kDa member of the yeast SWI/
tion between the expression of hBrm and the ability of the
SNF complex as TFG3/TAF30/ANC1. This protein is remark-
glucocorticoid receptor to activate transcription (37). Consis-
able in many respects. First, it is tightly associated with TFIIF,
TFIID, and the SWI/SNF complex and is thus a member of tent with these observations, SWI2/SNF2 is required for glu-
three complexes known to contribute to transcriptional activa- cocorticoid receptor activation in S. cerevisiae (60). Further-
tion in yeast cells. TFG3 is not a component of TFIIH, TFIIE, more, the ATPase domain of another human SWI2/SNF2 homo-
or any other characterized yeast basal transcription factor log, BRG1, can be swapped for the corresponding domain in
(17a). The phenotypes of tfg3D strains, however, demonstrate SWI2/SNF2 and will enable activation by the glucocorticoid
that TFG3 is not essential for many of the tasks associated with receptor in S. cerevisiae (26). Recent studies also suggest that
these complexes, such as basal-level transcription (TFIID and BRG1 binds to the retinoblastoma suppressor protein and may
TFIIF) and nonfermentative growth on various carbon sources assist this protein in inducing tumor suppressor activity (10).
(SWI/SNF complex). As further shown here, TFG3 is not re- The SWI/SNF complex may also be involved in leukemo-
quired for transcriptional activation by certain activators genesis. Reciprocal chromosomal translocations are often as-
shown previously to depend strongly on other components of sociated with acute human leukemias, and many of these trans-
the SWI/SNF complex. locations result in the fusion of the HRX (human trithorax)
TFIIF and TFIID complexes have been purified to homo- gene (also called ALL-1) to one of several other genes located
geneity from mammalian cells, and almost all of the tightly on other chromosomes (38, 53). HRX is significantly similar in
associated components of these complexes have been cloned two regions to the Drosophila transcriptional activator trithorax
and sequenced (14, 18, 21, 22, 28). Neither of these mamma- (35). The first region of similarity encodes a zinc finger-con-
lian complexes has been shown to contain a mammalian ho- taining domain. The other region, which exhibits exceptional
molog of yeast TFG3. Despite the lack of a TFG3 homolog as similarity (60.5% identical and 82% similar over a 215-amino-
a member of mammalian TFIIF or TFIID, the possibility re- acid region), is located at the carboxy terminus of HRX (53).
mains that a TFG3 homolog interacts with these complexes or The human protein contains a putative DNA-binding region
is present at substoichiometric levels (18). near the amino terminus composed of “AT hooks”, which are
Our observation that TFG3 occurs in three distinct com- believed to be involved in binding to the minor groove of DNA,
plexes raises the possibility that it acts as a bridge between whereas the D. melanogaster protein lacks such a region. Ge-
these complexes. Since TFG3 is not an essential gene, these netic experiments in D. melanogaster strongly suggest that the
interactions, or the role of TFG3 in these interactions, must D. melanogaster SWI2/SNF2 homolog brahma assists certain
not be essential. Furthermore, TFG3 is unlikely to play a homeotic activators such as trithorax in relieving the repressive
crucial structural role in these complexes, because deletions of effects of chromatin, perhaps through a D. melanogaster coun-
any of the other components of TFIIF or TFIID confer lethal- terpart of the yeast SWI/SNF complex (51). These results raise
ity (18, 45). TFG3 may, however, perform a similar nonessen- the possibility that human HRX protein interacts with a human
tial function in all three complexes, such as assisting in the SWI/SNF complex.
binding of these complexes to RNA polymerase II or chroma- All chromosomal translocations involving HRX result in the
tin components. Although we have not been able to demon- production of chimeric proteins, and it has been suggested that
strate stable interactions between complexes containing TFG3, these fusion proteins may improperly alter the transcriptional
many conditions still remain to be tested, including possible program (38, 53). The HRX fusion proteins retain the amino-
interactions with TFIID. Another possibility is that TFG3 terminal DNA-binding region of HRX but do not always retain
masks or protects essential interaction sites on these com- the carboxy-terminal region. Two of the most common fusions
plexes, possibly regulating the fidelity of interactions between are to the ENL or AF-9 genes, which are strikingly similar to
complexes. In strains that lack TFG3, these essential interac- one another. This similarity has led others to suggest that the
tions may still take place but may not be properly regulated. gene products perform analogous functions in the cell and
Homologs of SWI2/SNF2 have been identified in S. cere- therefore may perform similar roles in promoting leukemia
visiae, Drosophila melanogaster, mice, and humans (8, 12, 26, when fused to HRX protein (38). In addition, both ENL and
34, 37, 41, 52). In addition, homologs of SNF5 have been iso- AF-9 contain nuclear targeting signals, and immunofluores-
VOL. 16, 1996 TFG3 IS A COMPONENT OF YEAST SWI/SNF 3315

FIG. 5. TFG3 is significantly similar to the human ENL and AF-9 proteins. An alignment of a homologous 50-amino-acid region of TFG3, SC33KB_3, ENL, and
AF-9 proteins is shown. The amino acid position of the residue that begins each region is indicated. Residues that are identical among at least three of the proteins
are displayed as white letters on a black background, and conserved residues (determined with a PAM-150 matrix) are displayed on a shaded background.

cence experiments demonstrate that ENL resides in the nu- 6. Clark-Adams, C. D., D. Norris, M. A. Osley, J. S. Fassler, and F. Winston.
cleus (38, 46). 1988. Changes in histone gene dosage alter transcription in yeast. Genes
Dev. 2:150–159.
Welch and Drubin previously reported that TFG3 is signif- 7. Côté, J., J. Quinn, J. L. Workman, and C. L. Peterson. 1994. Stimulation of
icantly similar to ENL and AF-9 (56). We find that TFG3 is GAL4 derivative binding to nucleosomal DNA by the yeast SWI/SNF com-
also similar to the predicted polypeptide encoded by the un- plex. Science 265:53–60.
characterized open reading frame SC33KB_3. As calculated by 8. Davis, J. L., R. Kunisawa, and J. Thorner. 1992. A presumptive helicase
(MOT1 gene product) affects gene expression and is required for viability in
the BLAST program, the probability that the maximum score the yeast Saccharomyces cerevisiae. Mol. Cell. Biol. 12:1879–1892.
between the 50 most highly conserved residues of TFG3 and 9. Dingwall, A. K., S. J. Beek, C. M. McCallum, J. W. Tamkun, G. V. Kalpana,
AF-9 would occur at random is 3.9 3 10210 (Fig. 5). The cor- S. P. Goff, and M. P. Scott. 1995. The Drosophila snr1 and brm proteins are
responding probability between TFG3 and ENL is 6.2 3 1027 related to yeast SWI/SNF proteins and are components of a large protein
complex. Mol. Biol. Cell 6:777–791.
(1, 56). The similarity between TFG3 and SC33KB_3 is ob- 10. Dunaief, J. L., B. E. Strober, S. Guha, P. A. Khavari, K. Alin, J. Luban, M.
served through almost the entire length of these proteins, and Begemann, G. R. Crabtree, and S. P. Goff. 1994. The retinoblastoma protein
the corresponding probability (as calculated by BLAST) is and BRG1 form a complex and cooperate to induce cell cycle arrest. Cell 79:
7.2 3 10243. Importantly, the regions of greatest similarity 119–130.
11. Dynlacht, B. D., T. Hoey, and R. Tjian. 1991. Isolation of coactivators
between TFG3 and ENL/AF-9 are the same as those between associated with the TATA-binding protein that mediate transcriptional ac-
ENL and AF-9. The homology between TFG3, AF-9, and ENL, tivation. Cell 61:563–576.
together with our observation of TFG3 as a member of the 12. Elfring, L. K., R. Deuring, C. M. McCallum, C. L. Peterson, and J. W.
yeast SWI/SNF complex, led us to the proposal that ENL and/ Tamkun. 1994. Identification and characterization of Drosophila relatives of
the yeast transcriptional activator SNF2/SWI2. Mol. Cell. Biol. 14:2225–2234.
or AF-9 protein associates with a human SWI/SNF complex. 13. Estruch, F., and M. Carlson. 1990. SNF6 encodes a nuclear protein that is
Whereas activators may ordinarily interact with SWI/SNF com- required for expression of many genes in Saccharomyces cerevisiae. Mol. Cell.
ponents in a regulated fashion, fusion of the HRX activator to Biol. 10:2544–2553.
ENL or AF-9 may result in persistent interaction with the 14. Finkelstein, A., C. F. Kostrub, J. Li, D. P. Chavez, B. Q. Wang, S. M. Fang,
J. Greenblatt, and Z. F. Burton. 1992. A cDNA encoding RAP74, a general
SWI/SNF complex and thus to uncontrolled access to sites initiation factor for transcription by RNA polymerase II. Nature (London)
normally occluded by chromatin. The resulting changes in gene 355:464–467.
expression may play a role in initiating acute leukemias. Ac- 15. Grunstein, M. 1990. Histone function in transcription. Annu. Rev. Cell Biol.
cording to this idea, it is not necessary for the TFG3 homolog 6:643–678.
16. Hanes, S. D., and R. Brent. 1989. DNA specificity of the bicoid activator
to be a positive-acting factor or even a factor required for the protein is determined by homeodomain recognition helix residue 9. Cell 57:
activity of SWI/SNF complex; it need only interact with the 1275–1283.
complex. In addition, the presence of a putative TFG3 ho- 17. Happel, A. M., M. S. Swanson, and F. Winston. 1991. The SNF2, SNF5, and
molog in S. cerevisiae raises the possibility that this factor is a SNF6 genes are required for Ty transcription in Saccharomyces cerevisiae.
Genetics 128:69–77.
member of a related complex. Future studies of TFG3 and 17a.Henry, L., J. Svejstrup, and J. Feaver. Unpublished observations.
SC33KB_3 function in S. cerevisiae may help us understand 18. Henry, N. L., A. M. Campbell, W. J. Feaver, D. Poon, P. A. Weil, and R. D.
how ENL and AF-9 participate in leukemogenesis in humans. Kornberg. 1994. TFIIF-TAF-RNA polymerase II connection. Genes Dev.
8:2868–2878.
19. Henry, N. L., M. H. Sayre, and R. D. Kornberg. 1992. Purification and
ACKNOWLEDGMENTS characterization of yeast RNA polymerase II general initiation factor g.
J. Biol. Chem. 267:23388–23392.
We thank Yang Li for protein fractions and informative discussions. 20. Hirschhorn, J. N., S. A. Brown, C. D. Clark, and F. Winston. 1992. Evidence
We also thank Stefan Bjorklund, Jesper Svejstrup, Y.-J. Kim, and Jan that SNF2/SWI2 and SNF5 activate transcription in yeast by altering chro-
LaPointe for protein fractions; Roger Brent, Steve Hanes, Keith matin structure. Genes Dev. 6:2288–2298.
Yamamoto, David Stillman, and Brehon Laurent for plasmids; and 21. Hisatake, K., S. Hasegawa, R. Takada, Y. Nakatani, and R. G. Roeder. 1993.
Craig Peterson for the swi3D strain. The p250 subunit of native TATA-box binding factor is the cell cycle regu-
N.L.H. is supported by a Howard Hughes Medical Institute Predoc- latory protein CCG1. Nature (London) 362:179–181.
toral Fellowship. B.R.C. is supported by a postdoctoral fellowship from 22. Horikoshi, M., H. Fujita, J. Wang, R. Takada, and R. G. Roeder. 1991. Nu-
cleotide and amino acid sequences of RAP30. Nucleic Acids Res. 19:5436.
the American Cancer Society. This work is also supported by National
23. Imbalzano, A. N., H. Kwon, M. R. Green, and R. E. Kingston. 1994. Facil-
Institutes of Health grant GM36659 to R.D.K. itated binding of TATA-binding protein to nucleosomal DNA. Nature (Lon-
don) 370:481–485.
REFERENCES 24. Kalpana, G. V., S. Marmon, W. Wang, G. R. Crabtree, and S. P. Goff. 1994.
1. Altschul, S. F., W. Gish, W. Miller, E. W. Myers, and D. J. Lipman. 1990. Binding and stimulation of HIV-1 integrase by a human homolog of yeast
Basic local alignment search tool. J. Mol. Biol. 215:403–410. transcription factor SNF5. Science 266:2002–2007.
2. Brazas, R. M., and D. J. Stillman. 1993. The Swi5 zinc-finger and Grf10 25. Kelleher, R. J., III, P. M. Flanagan, D. I. Chasman, A. S. Ponticelli, K.
homeodomain proteins bind DNA cooperatively at the yeast HO promoter. Struhl, and R. D. Kornberg. 1992. Yeast and human TFIIDs are interchange-
Proc. Natl. Acad. Sci. USA 23:11237–11241. able for the response to acidic transcriptional activators in vitro. Genes Dev.
2a.Cairns, B. Unpublished data. 6:296–303.
3. Cairns, B. R., Y. J. Kim, M. H. Sayre, B. C. Laurent, and R. D. Kornberg. 26. Khavari, P. A., C. L. Peterson, J. W. Tamkun, D. B. Mendel, and G. R.
1994. A multisubunit complex containing the SWI1/ADR6, SWI2/SNF2, Crabtree. 1993. BRG1 contains a conserved domain of the SWI2/SNF2
SWI3, SNF5, and SNF6 gene products isolated from yeast. Proc. Natl. Acad. family necessary for normal mitotic growth and transcription. Nature (Lon-
Sci. USA 91:1950–1954. don) 366:170–174.
4. Carlson, M., and B. C. Laurent. 1994. The SNF/SWI family of global tran- 27. Kim, Y.-J., S. Bjorklund, Y. Li, M. H. Sayre, and R. D. Kornberg. 1994. A
scriptional activators. Curr. Opin. Cell Biol. 6:396–402. multiprotein mediator of transcriptional activation and its interaction with
5. Carlson, M., B. C. Osmond, and D. Botstein. 1981. Mutants of yeast defec- the C-terminal repeat domain of RNA polymerase II. Cell 77:599–608.
tive in sucrose utilization. Genetics 98:25–40. 28. Kokubo, T., D. G. Gong, J. C. Wootton, M. Horokoshi, R. G. Roeder, and Y.
3316 CAIRNS ET AL. MOL. CELL. BIOL.

Nakatani. 1994. Molecular cloning of Drosophila TFIID subunits. Nature 44. Poon, D., and P. A. Weil. 1993. Immunopurification of yeast TATA-binding
(London) 367:484–487. protein and associated factors. J. Biol. Chem. 268:15325–15328.
29. Kruger, W., and I. Herskowitz. 1991. A negative regulator of HO transcrip- 45. Reese, J. C., L. Apone, S. S. Walker, L. Griffin, and M. R. Green. 1994. Yeast
tion, SIN1 (SPT2), is a nonspecific DNA-binding protein related to HMG1. TAFIIs in a multisubunit complex required for activated transcription. Na-
Mol. Cell. Biol. 11:4135–4146. ture (London) 371:523–527.
30. Kwon, H., A. N. Imbalzano, P. A. Khavari, R. E. Kingston, and M. R. Green. 46. Rubnitz, J. E., J. Morrissey, P. A. Savage, and M. L. Cleary. 1994. ENL, the
1994. Nucleosome disruption and enhancement of activator binding by a gene fused with HRX in t(11; 19) leukemias, encodes a nuclear protein with
human SWI/SNF complex. Nature (London) 370:477–481. transcriptional activation potential in lymphoid and myeloid cells. Blood
31. Laurent, B. C., and M. Carlson. 1992. Yeast SNF2/SWI2, SNF5, and SNF6 84:1747–1752.
proteins function coordinately with the gene-specific transcriptional activa- 47. Sayre, M. H., H. Tschochner, and R. D. Kornberg. 1992. Reconstitution of
tors GAL4 and Bicoid. Genes Dev. 6:1707–1715. transcription with five purified initiation factors and RNA polymerase II
32. Laurent, B. C., M. A. Treitel, and M. Carlson. 1990. The SNF5 protein of from Saccharomyces cerevisiae. J. Biol. Chem. 267:23376–23382.
Saccharomyces cerevisiae is a glutamine- and proline-rich transcriptional 48. Schmid, A., K. D. Fascher, and W. Horz. 1992. Nucleosome disruption at the
activator that affects expression of a broad spectrum of genes. Mol. Cell. yeast PHO5 promoter upon PHO5 induction occurs in the absence of DNA
Biol. 10:5616–5625. replication. Cell 71:853–864.
33. Laurent, B. C., M. A. Treitel, and M. Carlson. 1991. Functional interdepen- 49. Shih, C. C., J. P. Stoye, and J. M. Coffin. 1988. Highly preferred targets for
dence of the yeast SNF2, SNF5, and SNF6 proteins in transcriptional acti- retroviral integration. Cell 53:531–537.
vation. Proc. Natl. Acad. Sci. USA 88:2687–2691. 50. Stern, M. J., R. E. Jensen, and I. Herskowitz. 1984. Five SWI genes are
34. Laurent, B. C., X. Yang, and M. Carlson. 1992. An essential Saccharomyces required for expression of the HO gene in yeast. J. Mol. Biol. 178:853–868.
cerevisiae gene homologous to SNF2 encodes a helicase-related protein in a 51. Swanson, M. S., and F. Winston. 1992. SPT4, SPT5, and SPT6 interactions:
new family. Mol. Cell. Biol. 12:1893–1902. effects on transcription and viability in Saccharomyces cerevisiae. Genetics
35. Mazo, A. M., D. H. Huang, B. A. Mozer, and I. B. David. 1990. The trithorax 132:325–336.
gene, a trans-acting regulator of the bithorax complex in Drosophila, encodes 52. Tamkun, J. W., R. Deuring, M. P. Scott, M. Kissinger, A. M. Pattatucci, T. C.
a protein with zinc-binding domains. Proc. Natl. Acad. Sci. USA 87:2112– Kaufman, and J. A. Kennison. 1992. brahma: a regulator of Drosophila
2116. homeotic genes structurally related to the yeast transcriptional activator
36. McCracken, S., and J. Greenblatt. 1991. Related RNA polymerase-binding SNF2/SWI2. Cell 68:561–572.
regions in human RAP30/74 and Escherichia coli s70. Science 253:900–902. 53. Tkachuk, D. C., S. Kohler, and M. L. Cleary. 1992. Involvement of a ho-
37. Muchardt, C., and M. Yaniv. 1993. A human homologue of Saccharomyces molog of Drosophila trithorax by 11q23 chromosomal translocations in acute
cerevisiae SNF2/SWI2 and Drosophila brm genes potentiates transcriptional leukemias. Cell 71:691–700.
activation by the glucocorticoid receptor. EMBO J. 12:4279–4290. 54. Treich, I., B. R. Cairns, T. Santos, E. Brewster, and M. Carlson. 1995.
38. Nakamura, T., H. Alder, Y. Gu, R. Prasad, O. Canaani, N. Kamada, R. P. SNF11, a new component of the yeast SNF-SWI complex that interacts with
Gale, B. Lange, W. M. Crist, P. C. Nowell, C. M. Croce, and E. Canaani. a conserved region of SNF2. Mol. Cell. Biol. 15:4240–4248.
1993. Genes on chromosomes 4, 9, and 19 involved in 11q23 abnormalities in 55. Vinson, C. R., K. L. LaMarco, P. F. Johnson, W. H. Landschulz, and S. L.
acute leukemia share sequence homology and/or common motifs. Proc. Natl. McKnight. 1988. In situ detection of sequence-specific DNA binding activity
Acad. Sci. USA 90:4631–4635. specified by a recombinant bacteriophage. Genes Dev. 2:801–806.
39. Neigeborn, L., and M. Carlson. 1984. Genes affecting the regulation of SUC2 56. Welch, M. D., and D. G. Drubin. 1994. A nuclear protein with sequence
gene expression by glucose repression in Saccharomyces cerevisiae. Genetics similarity to proteins implicated in human acute leukemias is important for
108:845–858. cellular morphogenesis and actin cytoskeletal function in Saccharomyces
40. O’Hara, P. J., H. Horowitz, G. Eichinger, and E. T. Young. 1988. The yeast cerevisiae. Mol. Biol. Cell 5:617–632.
ADR6 gene encodes homopolymeric amino acid sequences and a potential 57. Welch, M. D., D. B. N. Vinh, H. H. Okamura, and D. G. Drubin. 1993.
metal-binding domain. Nucleic Acids Res. 16:10153–10169. Screens for extragenic mutations that fail to complement act1 alleles identify
41. Okabe, I., L. C. Bailey, O. Attree, S. Srinivasan, J. M. Perkel, B. C. Laurent, genes that are important for actin function in Saccharomyces cerevisiae.
M. Carlson, D. L. Nelson, and R. L. Nussbaum. 1992. Cloning of human and Genetics 135:265–274.
bovine homologs of SNF2/SWI2: a global activator of transcription in yeast 58. Winston, F., and M. Carlson. 1992. Yeast SNF/SWI transcriptional activa-
S. cerevisiae. Nucleic Acids Res. 20:4649–4655. tors and the SPT/SIN chromatin connection. Trends Genet. 8:387–391.
42. Peterson, C. L., A. Dingwall, and M. P. Scott. 1994. Five SWI/SNF gene 59. Workman, J. L., I. C. A. Taylor, and R. E. Kingston. 1991. Activation
products are components of a large multi-subunit complex required for domains of stably bound GAL4 derivatives alleviate repression of promoters
transcriptional enhancement. Proc. Natl. Acad. Sci. USA 91:2905–2908. by nucleosomes. Cell 64:533–544.
43. Peterson, C. L., and I. Herskowitz. 1992. Characterization of the yeast SWI1, 60. Yoshinaga, S. K., C. L. Peterson, I. Herskowitz, and K. R. Yamamoto. 1992.
SWI2, and SWI3 genes, which encode a global activator of transcription. Cell Roles of SWI1, SWI2, and SWI3 proteins for transcriptional enhancement
68:573–583. by steroid receptors. Science 258:1598–1604.