ABSTRACT
The
ROX1
gene of
Saccharomyces cerevisiae
encodes a protein required for the repression of genes expressed under anaerobic
conditions. ROX1 belongs to a family of DNA binding proteins which contain the
high mobility group motif (HMG domain). To ascertain whether the HMG domain of
ROX1 is required for specific DNA binding we synthesized a series of ROX1
protein derivatives, either
in vitro
or in
Escherichia coli
as fusions to glutathione S-transferase (GST) protein, and tested them for their ability to bind to
DNA. Both ROX1 proteins that were synthesized
in vitro
and GST-ROX1 fusion proteins containing the intact HMG domain were able to bind
to specific target DNA sequences. In contrast, ROX1 proteins which contained deletions within the HMG domain were no longer capable of binding to DNA. The
oligomerization of ROX1
in vitro
was demonstrated using affinity-purified GST-ROX1 protein and ROX1 labelled with [
35
S]methionine. Using various ROX1 protein derivatives we were able to demonstrate
that the domain required for ROX1-ROX1 interaction resides within the N-terminal 100 amino acids which constitute the HMG domain.
Therefore, the HMG domain is required for both DNA binding activity and
oligomerization of ROX1.
In the yeast
Saccharomyces cerevisiae
a family of genes has been identified whose expression is coordinately
regulated by oxygen. Transcription of genes in this family is repressed in the
presence of oxygen and heme (
1
-
3
). These genes are dispersed throughout the genome and include
HEM13
, encoding coproporphyrinogen oxidase, an enzyme which catalyses the sixth step in the heme biosynthetic
pathway. Other members of this family include
ANB1
and
COX5b
, respectively encoding an isoform of translational factor eIF-5A and an isoform of cytochrome oxidase subunit V.
Repression of these genes in the presence of oxygen and heme is mediated by the
product of the
ROX1
gene. Mutations in
ROX1
were first identified based on their ability to allow constitutive expression
of the anaerobic gene
ANB1
and
ROX1
was thus shown to encode a repressor (
4
). ROX1 is also required for repression of
COX5b
and
HEM13
by oxygen and heme (
3
,
5
). Transcription of
ROX1
is
induced in the presence of oxygen or heme (
6
). Thus levels of the ROX1 repressor are increased under aerobic conditions and
serves to repress expression of the oxygen- and heme-repressed genes. Analysis of the upstream regulatory sequences of
ANB1
revealed the target for repression mediated by ROX1 protein to be two operator
sites each containing the consensus sequence YYYATTGTTCTC; deletion of these operator sites resulted in high level expression of
ANB1
in the presence of oxygen (
7
). Inspection of the upstream non-coding regions of other ROX1-regulated genes revealed the presence of two copies of the operator
consensus in
COX5b
(
8
) and five copies of this sequence in
HEM13
(T.Keng, unpublished observations). Deletion of these sites in
COX5b
and in
HEM13
also resulted in constitutive expression of these genes (
8
; T.Keng, unpublished observations).
DNA sequence analysis of
ROX1
revealed that it encodes a protein of 368 amino acids and that the N-terminal portion of ROX1 protein shows homology with the high mobility
group (HMG) class of proteins (
9
). The HMG1 and HMG2 proteins were first described as acid-soluble, non-histone components of mammalian chromatin (
10
).
Although these proteins are quite abundant, their function is not yet fully
understood. These proteins contain two repeats, A and B, which share
significant homology with each other, and an acidic domain which can interact
with histone H1. The function of repeats A and B was revealed when Jantzen
et al.
(
11
) identified a DNA binding motif in hUBF, a transcription factor for human RNA
polymerase I, by sequence alignment with HMG1. On the basis of its homology
with the A and B repeats of HMG1 this novel DNA binding motif was named the HMG
domain. This domain is composed of an 85 amino acid stretch containing many
basic amino acid residues. Since the initial discovery of this novel motif a
number of HMG domain proteins have been identified (
12
). These include the
Schizosaccharomyces pombe
regulatory proteins Ste11, Mat-a1 and Mat-Mc (
13
-
15
), the mammalian sex determining factor SRY (
16
,
17
), the lymphoid-specific transcription factors LEF-1 (TCF-1[alpha]) and TCF-1 (
18
-
21
), the mitochondrial transcription factor mtTF1 (
22
) and the yeast ARS binding protein ABF-2 (
23
), as well as the ROX1 protein (
9
). Comparison of the sequences of the various members of this family of proteins
has revealed the existence of subfamilies of HMG proteins (
24
). One subfamily is made up of the HMG1 and HMG2 proteins, the ARS binding
protein ABF-2, UBF and the mtTF-1 protein. These proteins are present in all cell types, contain
multiple HMG domains and recognize DNA with no sequence specificity. The second subfamily includes proteins that show tissue-specific expression, interact specifically with restricted DNA sequences and contain only a single HMG domain. Members of this
subfamily include the lymphoid factors LEF-1 and TCF-1, the sex-determining protein SRY and the fungal transcription factors
Ste11, Mat-Mc and Mat-a1.
In order to delineate the functional domains of the ROX1 protein we expressed
the proposed HMG domain of ROX1 both
in vitro
in wheat germ extracts and
in vivo
as a fusion to glutathione S-transferase (GST) in
E.coli.
We asked if the HMG domain portion of ROX1 was capable of binding specifically
to an oligonucleotide from the
HEM13
promoter which we have shown to be necessary for repression of
HEM13
. Our analysis indicates that ROX1 protein derivatives containing the entire HMG
domain, be they
in vitro
translated, GST-ROX1 fusion proteins or ROX1 derivatives separated from the GST moiety by
thrombin cleavage, were able to bind to DNA with sequence specificity. We also
demonstrate that the HMG domain of ROX1 is required for the ability of ROX1 to
form oligomers
in vitro
.
Escherichia coli
strains MC1061 and DH5[alpha] were used for plasmid propagation.
Escherichia coli
strain SG935, containing a mutation in the Lon protease, was used for expression of GST-ROX1 fusion proteins. Bacterial cells were grown in Luria Broth
supplemented with 100 [mu]g/ml ampicillin.
Restriction digests and other manipulations were performed essentially as
described (
25
). Oligonucleotide primers were synthesized by either the Regional DNA Synthesis
Laboratory (University of Calgary, Canada) or the Sheldon Biotechnology Centre
(McGill University, Canada). Plasmid pSPROX1, used for
in vitro
transcription of
ROX1
, was generated by polymerase chain reaction using the following primers: 5'-G
Plasmid pGEX-ROX1 full-length (FL) was constructed by cloning an
Eco
RI-
Hin
dIII fragment containing the entire
ROX1
coding region from the plasmid pSPROX1 into the
Eco
RI and
Aat
II sites of a pGEX-2T plasmid (Pharmacia) modified to contain an extra 5 nt before the
Eco
RI restriction site. Plasmid pGEX-ROX1 176-368 was constructed by digesting pGEX-ROX1 FL with
Bam
HI and
Bgl
II and religating the vector, thereby generating an in-frame fusion of GST to the C-terminal half of the
ROX1
ORF. Plasmid pGEX-ROX1 1-176 was constructed by digesting pGEX-ROX1 FL at the unique
Bgl
II site and blunt ending with the Klenow fragment of DNA polymerase. The blunt-ended DNA was then circularized by ligation. This plasmid produces a GST
fusion protein containing the first 176 amino acids of ROX1 followed by nine
amino acids coded by an alternative reading frame before a stop codon is
reached. Plasmid pGEX-ROX1 1-101 was generated by polymerase chain reaction using primer 1 and 5'-GCGGATCCTCACTCGATTTCCTTCAA-3' (primer 3), which introduces an in-frame stop codon just after codon
101 in
ROX1
. The amplified product was cloned into the PCRII vector using the TA Cloning
Kit (In Vitrogen). An
Eco
RI-
Bam
HI fragment from this plasmid was subsequently ligated into
Eco
RI/
Bgl
II-digested vector pGEX-ROX1 FL, generating an in-frame fusion of GST to the N-terminal 101 amino acids of ROX1. Plasmid pGEX-ROX1 1-50 was generated as follows. Plasmid pGEX-ROX1 FL was digested with
Bam
HI and
Bgl
II. The
Bam
HI-
Bgl
II fragment was purified and digested with
Rsa
I, which cleaves the DNA 150 bp downstream from the
ROX1
start codon, generating a blunt-ended fragment. The purified
Bam
HI-
Rsa
I fragment was then ligated into the plasmid pGEX-2T cleaved with
Bam
HI and
Sma
I. This construction resulted in an in-frame fusion of GST to the N-terminal 50 amino acids of ROX1.
The pSPROX1 plasmid, in which the
ROX1
coding region is placed downstream of the SP6 promoter, was used to generate
transcripts which were subsequently translated
in vitro
. The ROX1 FL template was generated by digesting pSPROX1 with
Hin
dIII, while templates for the derivatives ROX1 1-286, ROX1 1-175, ROX1 1-100 and ROX1 1-58 were generated by digesting pSPROX1 with
Ssp
I,
Bgl
II,
Taq
I and
Hpa
II, respectively. The DNA templates were then extracted with phenol and precipitated with
ethanol. RNA was synthesized using SP6 RNA polymerase and 5 [mu]g of each linear template in a 50 [mu]l volume according to the manufacturer's instructions (Promega). Reactions were incubated for 60 min at 40oC and terminated by addition of RQ1 DNase, followed by incubation
at 37oC for 15 min. The reactions were extracted with phenol and the RNA
precipitated with ethanol and resuspended in 10 [mu]l distilled water that had been treated with diethylpyrocarbonate. Aliquots of each RNA (2 [mu]l) were then added to wheat germ extracts (Promega) together with 37.5 [mu]Ci [
35
S]methionine (Amersham) for
in vitro
translation. Aliquots of 5 [mu]l of each sample were then analysed on denaturing 15% SDS-polyacrylamide gels. Gels were fixed in isopropanol and acetic acid
and treated with Amplify fluorographic reagent (Amersham) for 30 min before
they were dried and exposed to X-ray film.
Expression of GST fusion proteins was performed essentially as described (
26
), with some modifications. Saturated overnight cultures of
E.coli
strain SG935 containing GST-ROX1 fusion expression plasmids were diluted 10-fold in LB ampicillin medium and the cultures were grown for 90 min at 32oC. Expression of the fusion proteins was induced with 0.15 mM
isopropyl thiogalactoside and the cultures were grown for an additional 3 h at
32oC. Cells were harvested by centrifugation at 5000 r.p.m. for 5 min and
lysed by two pulse sonications of 30 s. Triton-X-100 was added to a final concentration of 1% (v/v) and the
suspension was centrifuged for 5 min at 10 000 r.p.m. A 50% slurry of glutathione-Sepharose beads (Pharmacia) was added to the supernatant and protein
binding was allowed to proceed for 15 min at 4oC. Following binding, the beads were washed three times with phosphate-buffered saline (PBS) and the suspension was centrifuged at 500
r.p.m. for 1 min to collect the beads. GST fusion proteins were left bound to
the glutathione-Sepharose beads or, alternatively, were eluted from the beads with a
solution of 50 mM Tris, pH 8.0 and 20 mM glutathione. The GST portion of the
fusion protein bound to glutathione-Sepharose beads was removed by cleavage with thrombin. Proteins were then
analysed on a 10 or 15% SDS-polyacrylamide gel.
GST or GST-ROX1 derivatives immobilized on glutathione-Sepharose beads were incubated for 2 h at 4oC on a rotating platform with
in vitro
translated
35
S-labelled ROX1 protein derivatives in binding buffer (120 mM NaCl, 50 mM
Tris, pH 8.0, 0.5% Nonidet P-40) in a total volume of 200 [mu]l. The beads were then washed four times with 1 ml binding buffer
before bound proteins were eluted by boiling in 30 [mu]l SDS-polyacrylamide gel electrophoresis sample loading buffer (
25
). The eluted proteins were analysed on 10 or 15% SDS-polyacrylamide gels.
DNA binding assays were performed in a total volume of 15 [mu]l containing DNA binding buffer (4 mM Tris, pH 8.0, 4 mM MgCl
2
, 100 mM KCl, 12% glycerol), 100 ng poly(dI[middot]dC) and 0.5-1.5 ng
32
P-labelled oligonucleotides containing the ROX1 binding site. Two double-stranded oligonucleotides with different sequences were used in
these studies. The first, RS33, consisted of complementary oligonucleotides 5'-AATTCTTTG
In order to define the region of the ROX1 protein necessary for DNA binding we
generated derivatives of ROX1 with C-terminal deletions using
in vitro
transcription/translation systems. The
ROX1
gene was inserted into a plasmid downstream of an SP6 promoter and deletions
were generated by digesting the DNA with different restriction endonucleases
(Fig.
1
A). The linearized DNA templates were subsequently transcribed with SP6 RNA
polymerase and the transcripts were translated in a wheat germ extract.
Aliquots of the [
35
S]methionine-labelled proteins were analysed by SDS-PAGE to verify that synthesis had occurred and that the proteins
were produced in similar amounts (Fig.
1
B). From the autoradiogram each labelled protein was found to be of a size
consistent with its predicted molecular weight. Proteins of lower molecular weight that were also observed in some cases were probably due
to premature termination of translation. No labelled protein was observed when
no exogenous RNA was added to wheat germ extract (Fig.
1
B).
The full-length ROX1 protein synthesized
in vitro
was tested for its ability to interact with DNA in a sequence-specific manner by an electrophoretic mobility shift assay. A labelled 32
bp double-stranded DNA, RS32, containing sequences from the
HEM13
promoter defined to be involved in repression, was used as a probe. The left
panel in Figure
2
depicts an autoradiogram of binding reactions performed in the presence of
increasing amounts of unlabelled competitor DNA with the same sequence as the
labelled probe. In the absence of specific competitor DNA protein-DNA complexes were detected (Fig.
2
, lane 2). The complexes were barely detectable when a 10-fold excess of the unlabelled specific DNA was added to the reaction (Fig.
2
, lane 3) and disappeared when a 50- or 100-fold excess of the unlabelled DNA was included in the binding
mixture (Fig.
2
, lanes 4 and 5). In contrast, no detectable decrease in complex formation was
observed when an excess of an unrelated 31 bp fragment, RNS31, was added to the
reactions as a non-specific competitor (Fig.
2
, lanes 8-10). No complex formation was detected when wheat germ extract incubated
in the absence of
ROX1
RNA was added to labelled RS32 DNA (Fig.
2
, lane 1). These experiments demonstrate that ROX1 protein synthesized
in vitro
is able to form a protein-DNA complex with a specific target DNA.
In order to obtain a more homogeneous source of ROX1 protein we expressed ROX1
as a fusion to GST in
E.coli.
The entire
ROX1
ORF was cloned in-frame into the pGEX-2T plasmid to generate the plasmid GST-ROX1 FL. This plasmid was subsequently used to construct fusion proteins between GST and different
regions of the ROX1 protein (Fig.
4
A). Plasmids containing the different fusions were introduced into
E.coli
strain SG935, which harbors a mutation in the Lon protease. Fusion proteins were
purified using a single step affinity purification with glutathione-Sepharose beads.
Purified proteins were separated by SDS-PAGE to verify the expression and size of each protein. Figure
4
B shows expression of the 26 kDa GST protein alone, as well as that of various
GST-ROX1 fusion proteins. The full-length ROX1 protein fused to GST migrated with an apparent
molecular weight of 68 kDa, a size consistent with the predicted molecular
weight of 66 kDa. The GST-ROX1 N-terminal fusion proteins, GST-ROX1 1-176, GST-ROX1 1-101 and GST-ROX1 1-50, also gave the
expected molecular sizes (Fig.
4
B). The C-terminal fusion protein GST-ROX1 176-368 migrated with an apparent molecular weight of 43 kDa,
closely approximating the size predicted by its ORF. An advantage of this
system is that these proteins are made in greater quanties than the equivalent
proteins synthesized by
in vitro
translation. This was verified by Coomassie blue staining of proteins
synthesized by both methods.
Gel mobility shift assays performed with the various fusion proteins
demonstrated that GST-ROX1 fusion proteins containing a minimum of 101 N-terminal amino acids were capable of binding to the labelled RS33
DNA fragment containing the same ROX1 binding site from the
HEM13
promoter as used with
in vitro
synthesized ROX1 proteins (data not shown). In addition, the inability of the
GST-ROX1 176-368 fusion protein to bind to the RS33 DNA fragment ruled out the
possibility that a second independent DNA binding domain existed in ROX1 at the
C-terminus.
In this report we show that the full-length ROX1 repressor is capable of sequence-specific binding to an operator region in the regulatory region of
HEM13
. This DNA binding activity of ROX1 is mediated by the HMG domain at its N-terminus. We also demonstrate that the HMG domain is required for
oligomerization of ROX1
in vitro
.
ROX1 is a sequence-specific DNA binding protein. Full-length ROX1 synthesized
in vitro
in wheat germ extracts was able to specifically bind to a fragment of DNA
containing sequences required for repression of
HEM13.
Neither a 100-fold excess of a non-homologous DNA fragment nor a vast excess of poly(dI[middot]dC) could compete for binding to ROX1. DNA binding assays
utilizing truncated versions of ROX1 synthesized
in vitro
indicated that the N-terminal 100 amino acids of the ROX1 protein are essential for DNA
binding. This region includes the HMG domain, which lies between residues 9 and
93 (
9
).
In vitro
translated ROX1 1-58, which is missing 35 amino acids of the HMG domain, and GST-ROX1 1-50, which is missing 43 amino acids of the HMG domain, are
both unable to bind to DNA fragments containing the operator site.
Interestingly, all ROX1 derivatives containing a minimum of 175 N-terminal amino acids gave rise to multiple protein-DNA complexes which migrated with the same mobilities as complexes
formed with full-length ROX1 protein. Formation of multiple complexes with these
derivatives could be due to the ability of ROX1 to form oligomers
in vitro
and the multiple protein-DNA complexes detected may represent complexes of oligomers of ROX1 with
DNA. Alternatively, this phenomenon could be due to the presence of protease-sensitive sites within the ROX1 protein. The different ROX1 protein
derivatives would be degraded to form distinct species, which could then form
multiple specific protein-DNA complexes. However, we believe the latter possibility to be
improbable, given the fact that the multiple protein-DNA complexes appear whether we utilize
in vitro
synthesized ROX1 or
E.coli
-derived ROX1 proteins.
The oligomerization of ROX1 was investigated using both GST-ROX1 fusion proteins and
in vitro
labelled ROX1. Our analyses indicate that ROX1 protein is capable of
oligomerization and that the HMG domain at the N-terminus of ROX1 is required for formation of oligomers. The ROX1 protein
contains a stretch of 22 amino acids from residues 102 to 123 which contains 16
glutamine residues. Such stretches of glutamine residues are a feature common
to eukaryotic transcriptional activator proteins and are thought to be required
for protein-protein interactions (
27
). In the protein interaction assay the ROX1 1-100
in vitro
translated protein was capable of interacting with GST-ROX1 FL protein, indicating that the polyglutamine tract in ROX1 is not
required for ROX1 oligomerization.
It is interesting to note that while the
in vitro
translated ROX1 1-100 protein gave rise to a single protein-DNA complex in the DNA binding assay, the
E.coli
-produced, thrombin-cleaved ROX1 1-101 protein gave rise to a protein-DNA complex of lower mobility in addition to one of
the same mobility as that of the
in vitro
synthesized protein. Formation of the additional complex may be explained by the different concentrations of the two proteins used in the DNA
binding assays. The concentration of
E.coli
-produced ROX1 1-101 in the DNA binding assays was higher than that of
in vitro
translated ROX1 1-100 and the presence of a higher concentration of protein in the binding
assay would favor the formation of oligomeric species. When the DNA binding
assay was carried out with lower concentrations of ROX1 1-101 protein only the complex with a higher mobility was detected (data
not shown).
Although the precise mechanism of repression by ROX1 protein remains unclear, it
must involve the TUP1 and SSN6 (CYC8) proteins. In strains deleted for
TUP1
or
SSN6,
expression of ROX1-regulated genes is observed under repressing conditions (
9
,
28
). In particular, we have constructed
ssn6
::
LEU2
and
tup1
::
LEU2
disrupted strains each containing an integrated copy of a
HEM13
-
lacZ
fusion at the
TRP1
locus. When expression of this fusion was tested under repressing conditions
derepressed levels of
HEM13
-
lacZ
activity were detected (data not shown). This suggests that the TUP1/SSN6
complex is an integral component of the machinery required for repression of
HEM13
.
The TUP1 and SSN6 proteins are also required for activity of a number of other DNA binding repressor proteins that function to regulate a wide
range of activities, including cell type and catabolite repression (
29
-
31
). While they do not themselves contact DNA, TUP1 and SSN6 are believed to form
a complex that is recruited by specific DNA binding repressors to the promoter
(
30
,
32
-
37
). Affinity chromatography experiments have detected only a tenuous SSN6-ROX1 interaction (
37
). We have also failed to detect any interaction of SSN6 with ROX1 (data not
shown). This may be due to the complexity of the interaction. For example,
interaction of ROX1 with SSN6 may only occur in the presence of TUP1. Alternatively, ROX1 binding to its cognate DNA may be required for
interaction with SSN6. Interaction of SSN6 with ROX1 may occur through what we
believe is the repression domain of ROX1, which is located at the C-terminus. A strain containing a
rox1
::
LEU2
disruption, as well as an integrated copy of a
HEM13
-
lacZ
fusion, was transformed with plasmids containing different deletion derivatives
of
ROX1.
Ability of these derivatives to restore repression to the strain was examined. A
ROX1
construct containing a deletion of sequences coding for the 82 C-terminal amino acids, which is fully capable of
in vitro
DNA binding, was unable to repress expression of
HEM13
-
lacZ
in vivo
(data not shown).
Thus we believe that ROX1 can function to recruit the TUP1/SSN6 complex to
repress transcription through an as yet unidentified domain of ROX1 at the C-terminus. In addition, the ability of ROX1 to repress transcription of
genes must be intimately associated with its ability to bind DNA. Deletion of
binding sites for ROX1 upstream of the
HEM13
transcription start site results in a large increase in expression of
HEM13
under repressing conditions (T.Keng, unpublished observations).
In this study we have demonstrated that ROX1 is a specific DNA binding protein
and that the HMG domain at the N-terminus of ROX1 functions in DNA binding. In addition, we postulate that
repression by ROX1 is mediated by an uncharacterized motif found in the C-terminal portion of ROX1.
Similar results have been obtained in a report published while our paper was
under revision (
38
). These authors show that the first 100 amino acids of ROX1 are sufficient for
DNA binding. In addition, they also show that the C-terminus of ROX1 is required for repression. They also demonstrate that
mutants in the HMG domain of ROX1, which bind with reduced affinity, show a reduced ability to
repress target genes.
We are grateful to V.Bilanchone and M.Cumsky for advice concerning the DNA
binding assays. R.Zitomer kindly provided the YCpROX1 plasmid and R.Trumbly the
SSN6
and
TUP1
disruption plasmids. In addition, we would like to thank S.Chronopoulos and
S.L.Chan for help with antibody production, D.J.Briedis for graciously
providing space and equipment and A.Staffa, P.Ulycznyj and I.Siboo for valuable
discussions. CDF was supported by grants from the Natural Sciences and
Engineering Research Council of Canada (NSERC) and by the Fonds Pour la
Formation de Chercheurs et L'Aide a la Recherche of Quebec. This work was
supported by research grant MA10010 from the Medical Research Council of Canada
and OGP0138319 from NSERC, as well as by grants from the Special General Fund
of NSERC administered by McGill University.
+
Present address: Apex Bioscience Inc., 2810 Meridian Parkway, Suite 120, Durham,
NC 27713-2277, USA
To demonstrate that the proteins bound to glutathione-Sepharose were in fact fusion proteins between GST and ROX1 and that
these proteins were devoid of
E.coli
-derived proteins we performed immunoblot analysis on the expressed
proteins using a rabbit polyclonal antibody raised against the GST-ROX1 1-50 protein (data not shown).
REFERENCES
Return