Functional analysis of the DNA-stimulated ATPase domain of yeast SWI2/SNF2
Functional analysis of the DNA-stimulated ATPase domain of yeast SWI2/SNF2
Emilie
Richmond
and
Craig L.
Peterson*
Program in Molecular Medicine and Department of Biochemistry and Molecular
Biology, University of Massachusetts Medical Center,
Worcester
, MA 01605,
USA
Received July 8, 1996;
Revised and Accepted August 18, 1996
ABSTRACT
The yeast SWI2/SNF2 polypeptide is a subunit of the SWI/SNF protein complex that
is required for many transcriptional activators to function in a chromatin
context. SWI2 is believed to be the founding member of a new subfamily of DNA-stimulated ATPases/DNA helicases that includes proteins that function in
DNA repair (RAD5, RAD16, ERCC6), recombination (RAD54), transcription (MOT1,
ISWI, brm, BRG1, hBRM) and cell cycle control (STH1). We have created a set of
16 mutations within the SWI2 ATPase domain and have analyzed the functional
consequences of these mutations
in vivo
. We have identified residues within each of the seven ATPase motifs that are
required for SWI2 function. We have also identified crucial residues that are
interspersed between the known ATPase motifs. In contrast, we identify other
highly conserved residues that appear to be dispensable for SWI2 function. We
also find that single amino acid changes in ATPase motifs IV and VI lead to a
dominant negative phenotype. None of the 12 SWI2 mutations that disrupt SWI2
activity
in vivo
alter the assembly of the SWI/SNF complex. These studies provide an invaluable
framework for biochemical analysis of the SWI2 ATPase and for functional
analysis of other SWI2 family members.
INTRODUCTION
The SWI/SNF protein complex, which is composed of 11 different polypeptides (
1
-
3
), is required for the induced expression of a large number of diversely
regulated yeast genes, including
HO
,
SUC2,
INO1
and
ADH2
(for review see
4
)
.
Furthermore, the efficient functioning of several sequence-specific transcriptional activator proteins, such as yeast GAL4,
Drosphila
ftz, and mammalian steroid receptors, requires an intact SWI/SNF complex (
5
,
6
). Mutations that alter chromatin components partially alleviate the defects in
transcription due to inactivation of the SWI/SNF complex, and thus it has been
proposed that the primary role of this complex is to facilitate the function of
gene regulatory proteins in a chromatin environment (
7
-
10
). Recent
in vitro
studies are consistent with this role, as the purified SWI/SNF complex can use
the energy of ATP hydrolysis to facilitate the binding of a transcriptional
activator, GAL4-AH, to a nucleosome (
2
).
The
SWI2
gene encodes a 200 kDa polypeptide that is the ATPase subunit of the SWI/SNF
complex. The sequence of
SWI2
contains seven sequence motifs that are characteristic of members of the DEAD/H
superfamily (superfamily 2) of nucleic acid-stimulated ATPases and DNA helicases (
11
,
12
). The SWI/SNF complex has DNA-stimulated ATPase activity
in vitro
, but lacks detectable DNA helicase (
2
,
13
) or DNA tracking (
14
) activities. A single amino acid change in the putative nucleotide binding loop
(motif I) of SWI2 eliminates SWI2 function
in vivo
(
5
,
13
) as well as the ATPase activity of the SWI2 subunit and SWI/SNF complex
in vitro
(
2
,
13
). Functionally important amino acids within motifs I and II have also been
identified in other members of this superfamily (
16
-
18
); however, the importance of residues within motifs Ia, III, IV, V and VI are
not known for any DNA-stimulated ATPase.
A large number of proteins have been identified that contain ATPase domains that
are more similar to that of SWI2 than other members of the nucleic acid-dependent ATPase and helicase superfamily. This observation led to the
proposal that SWI2 may define a new subfamily of DNA-stimulated ATPases (
19
,
20
; reviewed in
12
,
21
). This subfamily includes putative functional homologues of SWI2, such as
Drosophila
brm, mouse brg1, human BRG1, and human hbrm, as well as proteins that function
in DNA repair, recombination, cell cycle control and transcriptional
repression. Viral and bacterial family members have also been described.
Although several family members have been shown to exhibit DNA-stimulated ATPase activity, no DNA helicase activity has been described
for any member of the SWI2 subfamily. Consistent with these results, Henikoff (
22
) has suggested that the ATPase motifs of the SWI2 subfamily are most closely
related to poxvirus DNA-dependent ATPases.
We have carried out an extensive mutational analysis of the SWI2 DNA-stimulated ATPase domain. Mutations were constructed in each of the seven
ATPase/helicase motifs as well as in three additional sequence elements that
are highly conserved only in members of the SWI2 subfamily. Residues have been
identified in each of these sequence elements that are crucial for SWI2
function
in vivo
. Several additional, highly conserved residues, however, do not appear to play
an important role in the activity of SWI2. Furthermore, single amino acid
changes in either motif IV or VI result in dominant negative phenotypes that
are more potent than previously identified dominant negative
SWI2
alleles.
Strains (Table
1
) are isogenic derivatives of the YPH strain set described by Sikorski and
Hieter (
23
). The GAL4 reporter, pEG28 (
24
), and the yIP5-based
SWI2
plasmids were targeted for integration at the
ura3-52
locus by digestion of the plasmids with
Apa
I or
Stu
I. Strains were grown in either S medium [6.7 g/l yeast nitrogen base without
amino acids (Difco Laboratories)], SGS medium (6.7 g/l yeast nitrogen base, 2%
galactose, 0.5% sucrose) or YEPD medium (2% yeast extract, 1% bacto-peptone, 2% glucose). S media were supplemented with amino acids as
previously described (
25
). Appropriate amino acids were omitted for selection of plasmids.
Plasmids
Point mutations were generated by oligonucleotide-directed mutagenesis. The
Sac
I-
Kpn
I fragment which contains the ATPase domain of SWI2 was cloned into the RF form
of M13mp18 (Pharmacia). An aliquot (100 [mu]l) of phage supernatant was used to infect a dut
-
ung
-
Escherichia coli
strain (CJ236) and cells were grown in the presence of 20 [mu]g/ml uridine. Single-stranded DNA was purified from the culture supernatant and processed
for mutagenesis as described (
26
). For each mutant template the entire
Sac
I-
Kpn
I insert was sequenced (Sequenase, USB). Mutagenic oligonucleotides (DNA
International) were as follows:
The mutagenized
Sac
I-
Kpn
I fragments were isolated from RF DNA and used to replace the wild-type
SWI2
sequences in plasmid pER13 (
SWI2
in yEP24) to generate the following plasmids: pER14 (P824A), p5-1 (D894EE895Q), pER3 (D894AE895A), pER16 (P932A), pER22 (W935A), pER17
(R994A), pER18 (H1061A), pER19 (K1088A), pER23 (D1093A), pER36 (R1164A), pER21
(W1185A), pER20(G1166A), pER 56 ([Delta]STRAGGLG) and p4-1(R1196K). The mutagenized
Sac
I-
Kpn
I fragment was also used to replace the wild-type
SWI2
fragment in plasmid pCP331 [
SWI2-HA-6HIS
in yIP5; (
27
)] to generate the following plasmids: pER48 (P824A), pER1 (D854AE855A), pER49
(P932A), pER50 (W935A), pER51 (R994A), pER52 (H1061A), pER53 (K1088A), pER54
(R1164A), pER55 (W1185A), pER57 ([Delta]STRAGGLG) and pCP437 (R1196K). An
Nco
I-
Sal
I fragment from each of the above yEP24-based vectors was used to replace the wild-type
SWI2
sequences in plasmid pCP337 [
SWI2
in pRS315 (
23
)] to generate the following plasmids: pER24 (P824A), pER25 (D854AE855A), pER26
(W935A), pER27 (R994A), pER28 (H1061A), pER29 (K1088A), pER37 (D1093A), pER45
(R1164A), pER39 (G1166A), pER38 (W1185A), pER58 ([Delta]STRAGGLG) and p4-1 (R1196K). The
Nco
I-
Sal
I fragments were also used to replace wild-type sequences in plasmid pCP345 [
SWI2
in pRS424 (
23
)] to generate the following high copy number plasmids: pER31 (D854AE855A),
pER33 (R994A), pER40 (K1088A), pER59 ([Delta]STRAGGLG) and pER42 (R1196K). All
SWI2
plasmids contain at least 900 base pairs (bp) of sequence upstream of the
translation start site. In every case, the wild-type version fully complemented all growth and transcriptional defects of
a
swi2
null allele.
Invertase assays
Invertase assays were performed essentially by the method of Celenza and Carlson
(
28
) as adapted by M. Schmidt (University of Pittsburgh, PA) for smaller culture
volumes. Briefly, 4 ml duplicate cultures were grown to mid-log phase in SD media (S with 2% glucose) and one culture was harvested
(repressed). The second culture was washed twice with water and suspended in S
media plus 0.05% glucose, allowed to grow for 2.5 h at 30oC and harvested (derepressed). The standard colorimetric assay was
performed (
28
). Results are reported as nmol of glucose released/OD
600
/min.
Preparation of extracts and immunoblotting
RESULTS
Figure
1
shows the sequence of ATPase motifs I, Ia, II, III, IV, V and VI within yeast
SWI2, yeast STH1,
Drosophila
brm and human BRG1. Additional sequence elements that are highly conserved
among SWI2 family members are also shown and denoted IVa, IVb, and Vb. Boxed
residues indicate conserved residues that were targeted for site-directed mutagenesis. In most instances mutations were designed that
resulted in a single amino acid change to an alanine residue; in addition, two
sets of double amino acid changes were constructed at positions 894 and 895,
D894E,E895Q and D894A,E895A, a conserved lysine within motif VI was changed to
an arginine (R1196K), and a small deletion was constructed that results in the
removal of eight amino acids from motif V ([Delta]STRAGGLG).
The functioning of
swi2
mutants was assessed
in vivo
by complementation of the transcriptional defects that result from the complete
deletion of the
SWI2
gene. Three different SWI/SNF-dependent target genes were investigated: (i) an
HO-lacZ
fusion gene, (ii) a GAL4-dependent reporter gene and (iii) the
SUC2
gene.
Effect of SWI2 ATPase motif mutations on HO expression
Motif
SWI2 allele
[beta]-galactosidase expression
(% wild-type)
a
wild-type (CY396)
100
swi2
[Delta]
2
I
K798A
2
Ia
P824A
12
II
D894E,E895Q
2
D894A,E895A
2
III
P932A
97
W935A
12
IV
R994A
1.5
IVa
H1061A
53
IVb
K1088A
1
D1093A
ND
V
R1164A
41
G1166A
ND
[Delta]STRAGGLG
2
Vb
W1185A
37
VI
R1196K
2
a
[beta]-galactosidase activities of YEPD liquid cultures were performed (25)
on at least three independent transformants, and the Miller units (33) were
averaged. Results are expressed as % of wild-type (13.2 Miller units). Results varied by less than 20%.
ND, not determined
HO-lacZ
A deletion of the
SWI2
gene results in a 100-fold reduction in expression of a chromosomal
HO-lacZ
fusion gene (
25
). Mutated
SWI2
genes were initially introduced into a
swi2
deletion strain on high copy number plasmids and a qualitative [beta]-galactosidase filter assay was performed. Eleven mutants showed
measurable defects in activity in this assay, whereas four mutants, W831A (Ia),
P932A (III), D1093A (IVb), and G1166A (V), showed levels of
HO-lacZ
expression that were indistinguishable from wild-type (data not shown). Wild-type
SWI2
and
the remaining 11 mutants were integrated at the
URA3
locus for a quantitative analysis of their defects in
HO-lacZ
expression (Table
2
). Seven mutants, K798A (I), D894E,E895Q (II), D895A,E896A (II), R994A (IV),
K1088A (IVb), [Delta]STRAGGLG (V), and R1196K (VI), had a null phenotype, reducing
HO-lacZ
expression to 1-2% of the wild-type level. Two mutants, P824A (Ia) and W935A (III), reduced
HO-lacZ
expression to 12-15% of the wild-type level, and three mutants, H1061A (IVa), R1164A (V) and W1185A
(Vb), had a mild effect, lowering expression to ~40-50% of the wild-type level. These results indicate that at least one residue
within each of the 10 targeted domains is important for
SWI2
-dependent expression of an
HO-lacZ
fusion gene.
Effect of SWI2 ATPase motif mutations on Gal4 function
Motif
SWI2
allele
[beta]-galactosidase expression
(% wild-type)
a
wild-type
100
swi2
[Delta]
4
I
K798A
7
Ia
P824A
50
II
D894E,E895Q
15
D894A,E895A
5
III
P932A
ND
W935A
44
IV
R994A
7
IVa
H1061A
71
IVb
K1088A
6
D1093A
86
V
R1164A
61
G1166A
72
[Delta]STRAGGLG
6
Vb
W1185A
71
VI
R1196K
7
a
Three transformants were grown in SGS medium and analyzed as described in Table
2. Results are expressed as % of wild-type (110.4 Miller units). Values varied by less than 20%.
ND, not determined.
GAL4-dependent transcriptional activation
The ability of GAL4 to activate transcription from templates that contain two
low affinity GAL4 binding sites is dependent upon an intact SWI/SNF complex (
5
). We tested whether activation of transcription by GAL4 was sensitive to a functional SWI2 ATPase domain.
SWI2
mutants were introduced on low copy number plasmids into strain CY408 which
contains a deletion of
SWI2
and an integrated GAL4 reporter; expression of [beta]-galactosidase reflects the capacity of GAL4 to activate
transcription. The capacity of these
swi2
mutants to support GAL4-dependent activation is similar to our results for expression of the
HO-lacZ
fusion gene (Table
3
). The same seven mutants, K798A (I), D894A,E895A (II), D894E,E895Q (II), R994A
(IV), K1088A (IVb), [Delta]STRAGGLG (V) and R1196K (VI), exhibited a phenotype similar to that of a
deletion of
SWI2
(5-15% of the wild-type levels of GAL4-dependent expression). Only three other SWI2 mutants, P824A
(Ia), W935A (III), and R1164A (V) had a significant effect on GAL4-dependent expression (50, 44, and 61% of wild-type, respectively). Four mutants had little effect on the ability
of GAL4 to activate transcription. These mutants, H1061A (IVa), D1093A (IVb),
G1166A (V) and W1185A (Vb), also had small effects on
HO-lacZ
expression (Table
2
and data not shown).
SUC2
The
SUC2
gene encodes the enzyme invertase which is required for yeast to ferment
sucrose.
SUC2
expression is repressed on glucose medium and derepressed upon glucose
starvation; this transcriptional induction requires a functional SWI/SNF
complex (
29
). We assayed the production of invertase enzyme in strains harboring each of
the integrated SWI2 ATPase mutants (Table
4
).
Effect of SWI2 ATPase motif mutations on invertase expression
Motif
SWI2
allele
Invertase activity
a
Repressed
Derepressed
% Wild-type
wild-type
7
356
100
swi2
[Delta]
11
72
20
I
K798A
12
49
14
Ia
P824A
3
220
62
II
D894E,E895Q
6
77
22
D894A,E895A
5
42
11
III
P932A
ND
ND
ND
W935A
5
111
31
IV
R994A
7
40
11
IVa
H1061A
4
161
45
IVb
K1088A
6
26
7
D1093A
6
419
118
V
R1164A
6
123
35
G1166A
6
392
110
[Delta]STRAGGLG
4
38
11
Vb
W1185A
4
165
46
VI
R1196K
3
50
14
a
Two independent transformants were analyzed and the average is shown. Invertase
activity is defined as nmol of glucose released/OD
600
/min. Results varied by less than 20%.
ND, not determined.
The defects in invertase expression due to ATPase motif mutations parallel very
closely the results for
HO-lacZ
expression. The seven SWI2 mutants, K798A (I), D894A,E895A (II), D894E,E895Q
(II), R994A (IV), K1088A (IVb), [Delta]STRAGGLG (V) and R1196K (VI), showed induced levels of invertase
expression similar to that of the
swi2
deletion strain. One mutant, P824A (Ia), had only a mild effect on invertase
induction, while four mutants, W935A (III), H1061A (IVa), R1164A (V) and W1185A
(Vb), had a moderate effect.
ATPase motif mutations do not affect SWI/SNF complex assembly
The effects of
swi2
mutations on transcription could be due to functional inactivation of the
SWI/SNF complex or to a defect in SWI2 stability or SWI/SNF complex assembly.
To test the latter two possibilities we analyzed the steady state level of SWI2
protein by western blot and the assembly of the SWI/SNF complex by gel
filtration (Fig.
2
).
ATPase motif mutants exert dominant negative effects on transcriptional
activation
Introduction of an
swi2
mutant,
swi2K798A
, on a high copy plasmid into a wild-type strain is able to reduce transcriptional activation by the mammalian
glucocorticoid receptor by 50% (
15
). This result was used to design a similar mutation within a human
SWI2
homologue, BRG1. Overexpression of this BRG1 mutant in a cultured human cell
line resulted in a dominant, selective inhibition of transcription driven by
the eIF[alpha] promoter (
15
). We tested whether four additional ATPase motif mutants also had a dominant
negative effect on GR-dependent transcriptional activation in yeast. Strain CY26 (
SWI2
+
) was transformed with three different plasmids: (i) a GR expression vector,
(ii) a GR reporter that contains three GR binding sites upstream of a
CYC1-lacZ
fusion gene, and (iii) a high copy number plasmid that contained either wild-type
SWI2
,
swi2K798A
(I),
swi2D894A,E895A
(II),
swi2R994A
(IV),
swi2K1088A
(IVb) or
swi2R1196K
(VI). Expression of
swi2R994K
(IV) and
swi2R1196K
(VI) mutants were more effective than the
swi2K798A
(I) mutant at blocking activation of transcription by GR (Fig.
3
), lowering GR-dependent activation to ~20% of the wild-type control. Expression of the other two ATPase mutants also
caused a decrease in GR-dependent activation, but the decrease was similar to that found for the
swi2K798A
(I) allele. These results are consistent with the formation of non-functional SWI/SNF complexes that interfere with the ability of the wild-type SWI/SNF complex to facilitate GR function in yeast.
DISCUSSION
SWI2 is a member of a large family of proteins that shares extensive sequence
homology within seven sequence motifs characteristic of DNA-dependent ATPases and DNA helicases. We have constructed mutations in each
of these motifs and characterized the consequences on SWI/SNF function
in vivo
(summarized in Table
5
)
.
We identified residues within each of these motifs that are required for SWI2
to facilitate expression of an
HO-lacZ
fusion gene, activation of transcription by GAL4, and induction of
SUC2
expression. In contrast, some residues were not important for SWI2 function
even though they are highly conserved in SWI2 family members (Fig.
1
and Table
5
).
Although the SWI2 subfamily of ATPases is comprised of at least 20 proteins, and
the entire helicase superfamily 2 contains over 100 members, very little is
known about the biochemical functions of the individual ATPase motifs. Motifs I
and II are believed to be crucial for ATP binding and ATP hydrolysis,
respectively; the function of the other five ATPase/helicase motifs is unknown.
Mutations in these motifs may disrupt SWI2 function either by impairing its
ability to bind ATP, to hydrolyze ATP, or to interact with its nucleic acid co-factor. Only in the case of the RNA helicase, eIF4A, has an extensive
mutational analysis been carried out (
17
,
30
,
31
). In this case a sequence motif related to SWI2 motif VI is believed to be
involved in binding the RNA cofactor (
31
). In addition, mutations in some ATPase/helicase motifs do not alter RNA-stimulated ATPase activity, but do eliminate RNA helicase activity of
eIF4A (
30
). Most of the ATPase/helicase motifs of this RNA helicase, however, are not
closely related to those of the SWI2 subfamily. In addition, the SWI2 subfamily
also contains additional sequence blocks that are highly conserved between
subfamily members, but are not conserved in other members of helicase
superfamily 2 (Fig.
1
). These additional motifs may be involved in a common biochemical activity that is specific to the SWI2 subfamily of DNA-stimulated ATPases. One possibility is that these motifs are involved in
ATP-dependent nucleosome disruption, since many SWI2 subfamily members appear
to function in a chromatin context (
32
). Purification of SWI/SNF complexes that contain mutant SWI2 subunits will
allow direct tests of the biochemical function of each sequence element within
the ATPase/helicase domain.
Effect of SWI2 ATPase motif mutations on SWI/SNF dependent gene expression
Motif
SWI2
allele
HO expression
Gal4 activator function
Invertase expression
a
Dominant negative
b
wild-type
+++
a
+++
a
+++
-
swi2
[Delta]
-
-
+
-
I
K798A
-
-
+
*
Ia
P824A
+
++
++
ND
W831A
+++
ND
ND
ND
II
D894E,E895Q
-
+
+
ND
D894A,E895A
-
-
+
*
III
P932A
+++
ND
ND
ND
W935A
+
+
+
ND
IV
R994A
-
-
+
**
IVa
H1061A
++
++
+
ND
IVb
K1088A
-
-
-
*
D1093A
+++
+++
+++
ND
V
R1164A
+
+
+
ND
G1166A
+++
++
+++
ND
[Delta]STRAGGLG
-
-
+
*
Vb
W1185A
++
++
+
ND
VI
R1196K
-
-
+
**
a
+++, 80-100% of wild-type activity; ++, 50-79% of wild-type activity; +, 10-49% of wild-type activity; -, 1-9% wild-type activity.
b
**, 50-79% inhibition of wild-type activity; *, 10-49% inhibition of wild-type activity.
- , no inhibition.
ND, not determined.
Each of the
SWI2
ATPase domain mutants that we have tested exhibit a dominant negative phenotype
for GR-dependent transcriptional activation. This phenotype is not surprising
since these SWI2 mutants are competent for assembly of functionally inactive
SWI/SNF complexes. We were surprised to find that these same
swi2
mutants did not show a dominant negative phenotype for
HO
,
SUC2
or GAL4-dependent expression (unpublished results). One possibility is that
transcriptional activation of many SWI/SNF-dependent genes only requires a very low level of active SWI/SNF complex.
The inability to observe a dominant negative phenotype may simply reflect the
limited overexpression achieved by high copy plasmids. In any event, these
results indicate that GR-dependent activation is a sensitive assay system for identification of
dominant negative alleles of
SWI2
. The value of this assay system has already been proven, as the
swi2K798A
allele was first identified in this assay, and it has been a useful model to
design dominant negative alleles of a mammalian SWI2 homolog (
15
). In our current study, we have identified two new
swi2
alleles,
swi2R994A
and
swi2R1196K
, that are even more effective than
swi2K798A
as dominant negatives in the GR activation assay. The engineering of similar
mutations in other SWI2 family members may result in better dominant negative
alleles that will prove invaluable for functional studies.
ACKNOWLEDGEMENTS
We would like to thank Martin Schmidt for the modified invertase assay protocol,
and Loree Griffin Burns, Carolyn Church Landel, Colin Logie, and Mark Wechser
for comments on the manuscript. This work was supported by a March of Dimes
Basil O'Conner Scholar Award FY94-0754 and NIH grant GM49650 to C.L.P.
C.L.P. is a Scholar of the Leukemia Society of America.
REFERENCES
1 Cairns,B.R., Kim,Y-J., Sayre,M.H., Laurent,B.C., and Kornberg,R.D. (1994) Proc. Natl. Acad. Sci. USA, 91, 1950-1954.MEDLINE Abstract
