ABSTRACT
We have determined the location of
cis
-acting elements that are important for the expression of
RPO21
and
RPO22,
genes that encode the two largest subunits of RNA polymerase II (RNAPII) in
Saccharomyces cerevisiae
. A series of 5
'
-end deletions and nucleotide substitutions in the upstream regions of
RPO21
and
RPO22
were tested for their effect on the expression of
lacZ
fusions of these genes. Deletion of sequences from -723 to -693 in
RPO21
, which disrupted two Reb1p-binding sites and an Abf1p-binding site, resulted in a 10-fold decrease in expression. A T-rich region downstream of these sites was also important
for expression. Deletion of sequences from -437 to -392 in the
RPO22
-upstream, which resulted in a 30-fold decrease in expression, indicated that the Reb1p- and Abf1p-binding sites in this region were important for
RPO22
expression, as was a T-rich sequence immediately downstream of these sites. The
RPO21
and
RPO22
upstream regions were capable of interacting
in vitro
(gel-mobility-shift assays) with Reb1p and Abf1p. The similarities in the type and organization of elements
in the upstream regions of
RPO21
and
RPO22
suggest that expression of these genes may be regulated coordinately.
RNA polymerase II (RNAPII) is responsible for the synthesis of mRNAs and some
small nuclear RNAs in eukaryotic cells. In the yeast
Saccharomyces cerevisiae
, the enzyme is composed of subunits encoded by 12 different genes that are scattered throughout the yeast genome (
1
,
2
). Five of the subunits are also found in RNAPI, which synthesizes rRNA, and in
RNAPIII, which synthesizes tRNA, 5S RNA and U6 snRNA.
Little is known about the biosynthesis of RNAPII subunits in yeast, despite the
fact that this enzyme plays a central role in the expression of thousands of
genes whose transcript levels are likely to be influenced by the cellular
amount of active RNAPII (
3
). For this reason it is important to understand the mechanisms involved in
maintaining the appropriate amount of enzyme. A consideration of the
biosynthesis of multi-subunit enzymes such as RNAPII leads to the question of whether subunit
synthesis is coordinated so as to produce the required molar ratio of subunits.
In
Escherichia coli
, coordinate synthesis of the three proteins that make up the core RNAP results
in subunit levels that reflect closely the stoichiometry of each subunit in the
enzyme (
4
). Although the mechanisms involved in this regulation are not understood fully,
it is clear that coordinate synthesis of the two largest subunits of bacterial
RNAP, [beta] and [beta]', is due in part to the fact that the genes encoding these
subunits lie in the same operon and thus are transcribed from the same
promoters (
4
).
One mechanism for generating similar amounts of expression from genes that act
in a common biochemical pathway is the use of similar transcriptional control
elements. For example, in
S.cerevisiae
the Gal4p transcription factor acts through a defined upstream regulatory
sequence (UAS) to control the expression of a number of genes required for
galactose metabolism (
5
). Transcription of ribosomal-protein genes is controlled in a similar manner. In this case, the
expression of many of the yeast genes that encode protein subunits of the
ribosome is controlled by similar
cis
-acting elements, which are located upstream of the start site of
transcription. These elements often contain one or two binding sites for Rap1p
(
6
) or Abf1p (
7
), transcription factors that have important roles in coordinating the
expression of ribosomal-protein genes (
8
).
A similar mechanism may affect the expression of genes encoding subunits of
RNAPIII. An analysis of the upstream regions of
RPC160
and
RPC40
, two of the genes encoding subunits of RNAPIII in
S.cerevisiae
, has shown that binding sites for the transcription factor Abf1p are present (
9
). A deletion analysis of the upstream region of
RPC40
suggested that the binding site for Abf1p plays a significant role in the
expression of this gene (
9
). It is possible that Abf1p has a role in ensuring a similar level of
expression from genes encoding subunits of RNAPIII.
We have undertaken an investigation of the biosynthesis of RNAPII in
S.cerevisiae
in order to gain a better understanding of the mechanisms that maintain a
normal level of this enzyme. Our approach has been to determine which elements
in the upstream regions of the genes encoding the two largest subunits of
RNAPII, namely
RPO21
(also called
RPB1
) and
RPO22
(also called
RPB2
), are important for their expression. We describe a similarity in the type, organization and function of
cis
-acting elements in the two promoters.
The
S.cerevisiae
strain used in this study was W303-1a (
MAT
a
can1-100
his3-11
,
15
leu2-3
,
112
trp1-1
ura3-1
ade2-1
). Growth media were prepared and yeast transformations were performed as
described (
10
,
11
). All plasmids were propagated in
E.coli
strain XL-Blue (
12
).
DNA manipulations were performed essentially as described (
13
). pYF1495 is a yeast shuttle-vector (
LEU2
; 2[mu]m origin of replication) carrying an in-frame fusion of
RPO21
and
lacZ
. Details of the construction of this and other plasmids described here may be
obtained from the corresponding author. A series of 5"-end deletions was generated in the
RPO21
-upstream region by digestion with
Hin
dIII at -1585 (the A of the
RPO21
initiation codon is +1) and
Bal
31 exonuclease. The ends were filled with Klenow fragment of DNA polymerase I and
Hin
dIII linkers were ligated to the ends.
Hin
dIII-
Avr
II fragments (AvrII is at +171) from this set of 5"-end deletions were cloned into the same sites of pYF1495. The
rpo21-
[Delta]
UAS
allele was constructed by: (i) inserting a
Hin
dIII linker at an end-filled
Bst
EII site after -724 in the
RPO21
-upstream region; (ii) cutting the resulting plasmid with
Hin
dIII to isolate a 860 bp
Hin
dIII-
Hin
dIII fragment containing
RPO21
sequences from -1585 to -724; (iii) cloning this fragment in the
Hin
dIII site of a derivative of pYF1495 that has a 5' -deletion in the upstream region of
RPO21
up to -692. The resulting internal deletion/insertion is shown in Figure
2
A (bottom).
pYF1476 is a yeast shuttle-vector (
LEU2
; 2[mu]m origin of replication) carrying an in-frame fusion of
RPO22
and
lacZ
. 5'-end deletions of the
RPO22
-upstream region were generated by cleaving at unique sites in the upsteam region of
RPO22
, filling the ends, adding
Hin
dIII linkers, digesting with
Hin
dIII and religating the DNA.
Hin
dIII-
Bam
HI fragments (encompassing sequences from the 5"-end up to +81) were cloned into the same sites of pYF1476.
Plasmids with mutations in the
RPO21
and
RPO22
UASs were generated either by using oligonucleotides with UAS element mutations
and a second downstream oligonucleotide to amplify the UAS sequence by PCR or
by using a PCR-based site-directed mutagenesis technique that was described previously (
15
). The amplified regions were cloned and sequenced to confirm that the mutated
plasmids contained only the desired mutations. One of the site-directed mutations introduced three nucleotide substitutions in the putative Abf1p-binding site in the
RPO22
-upstream region. At position -383, a G is changed to A and at positions -393 and -392, CG is replaced with AA (see Fig.
2
B, top). The latter two substitutions created a new
Eco
RI site at -394 to -389 in the
RPO22
-upstream region. The new
Eco
RI site was used to create a 5'-deletion (to -391) in the upstream region of
RPO
22 shown in the fourth line of Figure
2
A. The asterisk in Figure
2
2
A refers to the additional G to A substitution at -383 in the upstream region, which, as
mentioned above, introduced another alteration in the Abf1p-binding site. The
rpo22
-n
UAS
allele was created by subcloning a
Hin
dIII-
Hin
cII fragment containing sequences of the
RPO
22-upstream region from -1200 to -438 and an
Eco
RI-
Bam
HI fragment containing
RPO
22 sequences from -391 to +82 into pYF1476, creating a
RPO22
-
lacZ
fusion gene with the deletion/insertion mutation shown in Figure
2
B (bottom). Note that this allele also contains the G to A substitution at -383.
Plasmids were generated with the UAS of
RPO21
cloned upstream of a UAS-less
CYC1-lacZ
gene. The vector, pDJ22, was derived from pLG670-Z (
15
), a yeast shuttle-vector (
URA3
, 2[mu]m) containing a fusion of
CYC1
-upstream sequences to the
lacZ
gene. pDJ22 was created by deleting sequences in pLG670-Z from the
Sma
I site at the 3'-end of
URA3
to the
Sph
I site 12 bp upstream of the 5'-most TATA box of
CYC1
and replacing them with a
Bgl
II linker. DNA fragments from wild-type and mutated
RPO21
-upstream sequences were amplified by PCR using two oligonucleotides, DA23 (5'-GGGGATCCGACTATCATACGGTAACC-3') and DA24 (5'-GGGGATCCACCGACAATCGTCTTTAG-3'). Amplified products
contained sequences from -740 to -674 that encompass the UAS of
RPO21
. The PCR products were digested with
Bam
HI and were cloned into the
Bgl
II site of pDJ22. Resulting plasmids were sequenced to confirm the number of
inserts, insert orientation and the absence of unwanted mutations.
[beta]-galactosidase activity was measured as described by Miller (
16
). Cells were grown in selective medium at 30oC. For each measurement, [beta]-galactosidase activity was determined on at least three
independent cultures.
Assays were performed as described (
17
) except that DNA-protein mixtures were separated on 5% acrylamide gels. Probes and non-labeled-competitor DNA from
RPO21
were generated by PCR using the primers DA23 and DA24, described above. The non-labeled-competitor DNA from the
RPO22
-upstream region was a
Hin
dIII-
Rsa
I fragment containing -437 to -209 of
RPO22
-upstream sequence. The probe with the Reb1p site from the
GAL1-GAL10
intergenic region (provided by C. Brandl) was a 104 bp
Bam
HI-
Eco
RI fragment from plasmid
his3
-GG227 (
18
). The probe containing the Abf1p site from
MAT
a
was the 136 bp
Bam
HI fragment of pB2ABF1 (
19
) (provided by L. McBroom). Probes were end-labeled using Klenow fragment of DNA polymerase I and [[alpha]-
32
P]dATP.
Proteins used for gel-shift assays were either: (i) purified yeast Abf1p, provided by A. Buchman
(
17
) or (ii) an extract from an
E.coli
strain expressing
REB1
(
20
; strain provided by J. Warner) or (iii) yeast whole-cell extract from a wild-type yeast strain (provided by S. Nouraini) that was prepared as
described (
21
). Estimations of the relative intensities of complexes were obtained by PhosphorImager analysis, using ImageQuant software (Fig.
4
A), or by scanning autoradiographs and quantifying band intensities using NIH
image software (Fig.
5
).
The
RPO21
-upstream region and various constructs were sequenced by the chain-termination method (
22
), using a combination of promoter deletions and primers that had been synthesized on the basis of the
RPO21
or
RPO22
sequences. Single- or double-stranded DNA templates were used.
Site-directed mutagenesis was used to introduce substitutions in the
RPO21
UAS (Fig.
2
). Plasmids were constructed that contain 1585 bp of
RPO21
-upstream region with mutations in the three elements of the UAS (Fig.
2
B, line h), mutations in any single site (Fig.
2
B, lines b-d) or mutations in any two of the sites (Fig.
2
B, lines e-g). The chosen mutations (Fig.
2
A, top) have been shown in other Reb1p- and Abf1p-binding sites (
9
,
25
,
27
) to reduce protein binding
in vitro
or transcriptional activation activity
in vivo
.
When mutations were introduced into all three sites (Fig.
2
B, line h), a 10-fold decrease in expression was observed, equal to the decrease observed
with a deletion that removed all or part of the three elements that make up the
UAS (Fig.
2
B, line i). When any one of the three elements was wild-type, and the other two were mutated, 50-70% of wild-type activity was observed (Fig.
2
B, lines e, f and g). Finally, activation of transcription by UAS sequences with
mutations in any single element (the other two being wild-type) was at levels that ranged from 70 to 85% of wild-type. These data suggest that the
RPO21
UAS is composed of three distinct functional elements, all of which are
necessary for full activity and all of which are partially redundant.
A DNA fragment carrying the
RPO21
UAS was inserted in a position upstream of the
CYC1
basal promoter and start sites of a
CYC1-lacZ
fusion gene devoid of other upstream-activating sequences. The DNA fragment used in these experiments spans nucleotides -740 to -674 and contains the two putative Reb1p-sites and the putative Abf1p-binding site.
A plasmid (pYF1476) was constructed with 1200 bp of
RPO22
- upstream sequences as well as 27 codons of the ORF fused in-frame to
lacZ
. It has been shown by S1-nuclease analysis of
RPO22
transcripts that the start site of transcription is at -260 +- 25 bp (the A of the initiation codon of
RPO22
is defined as +1) (
29
).
[beta]-galactosidase assays of a series of 5'-end deletions revealed that sequences upstream of -440 were not required for full expression (Fig.
3
A). The removal of 46 bp to position -391 resulted in a >30-fold decrease in expression. The DNA sequences of the
RPO22
-upstream region (
29
) that lie between -410 and -380 resemble the consensus-binding sites for Reb1p and Abf1p (Fig.
3
B); sites with the same non-conformity with the Reb1p consensus have been shown previously to bind
Reb1p (
25
,
30
). The sequences that match the Abf1p-consensus-binding site are in the opposite orientation to those in the UAS of
RPO21
. Since a deletion that removed all or part of the putative Reb1p- and Abf1p-binding sites resulted in decreased expression of
RPO22-lacZ
, this region is important for the expression of
RPO22
. We refer to this region as the
RPO22
UAS.
DNA fragments derived from the promoters of
RPO21
and
RPO22
were used as probes or competing DNA in electrophoretic mobility shift assays
(EMSAs) (
32
). Incubation of a DNA fragment containing the wild-type
RPO21
UAS with a crude yeast whole-cell extract (WCE), revealed two DNA-protein complexes (complexes 1 and 2, Fig.
4
A, lane 2). Mutation of the Abf1p-binding site (lane 3) or of the two Reb1p-binding sites (lane 4) caused a modest (<2-fold) reduction in the amount of complex 1. In contrast,
mutations in all three elements abolished complex 2 and reduced complex 1 by
>10-fold (compare lanes 2 and 5). We conclude that the Reb1p and Abf1p
elements of the UAS of
RPO21
are able to bind to proteins in a sequence-specific manner. Furthermore, mutations that reduce the activity of the
UAS
in vivo
(Fig.
2
B) also reduce the strength of DNA-protein interactions
in vitro
.
An extract from an
E.coli
strain over-expressing
REB1
(provided by J. Warner) (
20
) was used to test whether the
RPO21
UAS can bind Reb1p. With the wild-type
RPO21
UAS, a complex migrated at the same position as complex 1 (Fig.
4
A, compare lanes 2 and 7). Two additional faster-migrating complexes were also observed; these probably result from interactions between the probe and proteolytic fragments of Reb1p. Formation of complex 1 with a probe that carried mutations in both potential Reb1p-binding sites was reduced by 10-fold (lane 9). Mutations in the putative Abf1p-binding site had no significant effect on the
interaction of Reb1p with the probe (lane 8) and a probe with mutations in all
three elements showed a pattern similar to that with mutations in the two Reb1p-binding sites (lane 10). These data suggest that Reb1p can bind
specifically
in vitro
to the UAS of
RPO21
. Although the mutations in the Reb1p sites did not completely abolish complex
formation
in vitro
, the reduction in binding is sufficient to impair their function in the
RPO21
promoter
in vivo
(see Fig.
2
B). Residual binding of Reb1p to the mutated sites
in vivo
may account for the unexpectedly large activation of the
CYC1
promoter by the
RPO21
UAS carrying mutations in both Reb1p sites (leaving a single Abf1p site, Fig.
2
B; row g), perhaps through a synergistic interaction between Abf1p and Reb1p (see
Discussion).
When wild-type
RPO21
UAS DNA was combined with purified Abf1p (
17
) a single major complex was observed that migrated in a position corresponding
to complex 1 (Fig.
4
A, compare lanes 2 and 12) and this was reduced 15-fold by mutations in the putative Abf1p-binding site (lane 13). Mutations in the two Reb1p-binding sites showed no effect (lane 14), while mutations in
all three sites showed a similar pattern to that seen with mutations in only
the Abf1p site (lane 15). These data suggest that Abf1p interacts specifically
with the UAS of
RPO21
. Mutations that disrupt the function of the element
in vivo
(Fig.
2
B), also disrupt the interaction of Abf1p with the UAS
in vitro
.
The suggestion that the UAS of
RPO21
contains two Reb1p-binding sites was tested by adding to a wild-type
RPO21
UAS probe increasing amounts of extract from
E.coli
expressing
REB1
. At low levels of Reb1p, a single major complex (complex 1) was seen, which was chased into the slower-migrating complex 2 with the addition of increasing Reb1p concentration (Fig.
4
B, lanes 2-4). This complex migrated to the same position as complex 2 (Fig.
4
A, lane 2 and Fig.
4
B, lane 7). We conclude that the
RPO21
UAS has two binding sites for Reb1p.
The observation that elements of the
RPO21
UAS are partially redundant
in vivo
(Fig.
2
B) may reflect the fact that only a subset of the three sites is occupied at one
time. Radioactively-labeled wild-type
RPO21
UAS was incubated with purified Abf1p and an extract from
E.coli
expressing
REB1
in order to test this possibility. The presence of a third complex (complex 3,
Fig.
4
B, lane 6) at a higher gel position suggested that all three sites in the UAS
can be occupied simultaneously
in vitro
. Complex 3 was also detected with a yeast whole-cell extract (lane 7).
These results may indicate that complex 1, which is formed with a yeast whole-cell extract, is a mixture of two complexes, each of which include one
molecule of Abf1p or one molecule of Reb1p. This hypothesis is supported by the
observation that complex 1 is reduced by the introduction of mutations in all
three binding sites (Fig.
4
A, lane 5). Similarly, complex 2 probably contains a mixture of probe bound
either to two molecules of Reb1p, or to one molecule of Abf1p and one of Reb1p.
This hypothesis is supported by the observation of a reduction in complex 2
when either the Abf1p-binding site (Fig.
4
A, lane 3) or both of the Reb1p-binding sites (Fig.
4
A, lane 4) are mutated. Finally, complex 3 (Fig.
4
B, lane 7) may represent the
RPO21
UAS probe bound to two molecules of Reb1p and one molecule of Abf1p. Regardless
of the exact composition of each complex, the formation of three complexes is
clear and these complexes are abolished by mutations that have been shown to
reduce the binding of Abf1p and Reb1p. Thus, we conclude that the
RPO21
UAS contains two sites that bind Reb1p and one site that binds Abf1p.
Reb1p was combined with a radioactively-labeled DNA fragment containing a known Reb1p-binding site from the
GAL1-GAL10
intergenic region (
18
); DNA fragments containing the UASs of
RPO21
and
RPO22
were used as non-labeled competitors. As expected, a single major complex was formed, which
was reduced 15-fold by the addition of a 10-fold molar excess of non-labeled wild-type
RPO21
UAS (Fig.
5
, compare lanes 2 and 3). Non-labeled competitor carrying mutations in both Reb1p-binding sites reduced the amount of complex by <2-fold (lane 6). Competitors with a mutation in only one Reb1p-binding site (either one) were almost as effective as
the wild-type UAS, suggesting that both sites are able to bind Reb1p (lanes 4 and
5). A 10-fold molar excess of a DNA fragment with the wild-type UAS of
RPO22
was able to compete for binding of Reb1p to the Reb1p-binding site of
GAL1-GAL10
(Fig.
5
, lane 7); however, DNA with mutations in the putative Reb1p-binding site of
RPO22
failed to compete (lane 8).
These data suggest that the UAS of
RPO22
can interact with Reb1p
in vitro
in a sequence-specific manner, since mutations that impair the activity of the Reb1p
site
in vivo
(Fig.
3
C) also reduce the ability of the site to interact with Reb1p
in vitro
.
Competition experiments with a complex formed with purified Abf1p and a
radioactively-labeled DNA fragment from
MAT
a
(
19
) yielded similar results (Fig.
5
, lanes 10-14), suggesting that the
RPO22
UAS can interact specifically with Abf1p.
Our results indicate that the expression of two genes, which encode subunits unique to RNAPII in
S.cerevisiae
, is controlled by similar
cis
-acting upstream elements. Mutation of two Reb1p- binding sites and an Abf1p-binding site results in a 10-fold decrease in
RPO21
expression, while mutation of a Reb1p-binding site and an Abf1p-binding site results in a >30-fold decrease in the expression of
RPO22
. In both genes, the UAS is immediately upstream of a T-rich sequence which, when removed, decreases expression by an additional
10-fold.
We suggest that the similar elements in the UASs of
RPO21
and
RPO22
may serve to control coordinate synthesis of stoichiometric amounts of each subunit, which would minimize the wasteful expenditure of
cellular energy associated with the production of one subunit in excess of
others.
Regulation of
RPO21
and
RPO22
expression is similar in some ways to the regulation of genes that encode
components of the ribosome, another multisubunit complex. Ribosomal protein
(rprotein) genes are controlled by Abf1p or the related protein, Rap1p (
8
,
33
,
34
) and are regulated through the
RAS
/cAMP signal-transduction pathway. The increase in mRNA transcribed from many rprotein
genes due to a nutritional upshift depends on two kinds of upstream elements: a
T-rich stretch (
35
) and a functional Rap1p- (
35
-
37
) or Abf1p- (
35
) binding site. The increase in gene expression also requires cAMP-dependent-protein kinases (
36
,
37
) and
de novo
protein synthesis (
37
), indicating that phosphorylation of Rap1p (
36
) may not be sufficient for the response. Abf1p also is phosphorylated when
cells are shifted to a rich carbon source (
38
), suggesting that this modification may have a role in the regulation of
rprotein gene expression.
Abf1p (TAF, BAF) and Reb1p (Grf2p), members of a family of highly-abundant DNA-binding proteins [which also includes Rap1p (
6
)], are involved in many functions. Abf1p binds the promoters of many genes,
including those which encode rproteins (
33
,
34
) and proteins involved in glycolytic functions (
39
-
41
) and is important for silencing at the
HMR
locus (
7
,
42
). Abf1p also has a role in DNA replication, since its binding is required for
the function of some autonomously-replicating sequences (ARSs) and mutations in
ABF1
that confer a temperature-sensitive phenotype on yeast, result in mitotically unstable
ARS-CEN
plasmids at the semi-permissive temperature (
43
). Reb1p (Grf2p)-binding sites are important for the expression and termination of rRNA transcripts (
44
), have roles in the activation and repression of
ENO1
(
30
,
39
) and were shown to be important for creating a nucleosome-free
GAL1-10
intergenic region (
45
). Reb1p-binding sites stimulate transcription
in vivo
(
25
), but reduce the expression of
CYC1
when interposed between the UAS and TATA box (
46
). Reb1p also binds to telomeres and centromeres (
25
).
How could these multi-purpose proteins function in coordinate expression of genes encoding
subunits of RNAPII? As shown in Figure
6
, subunits unique to RNAPII encode seven genes which [with the exception of
RPO29
(
RPB9
), in which only one potential Reb1p-binding site was found in the upstream sequences] have two or more potential-binding sites for Reb1p and/or Abf1p and one or two T-rich stretches. In all cases these elements cluster within ~100 bp of each other. Although the orientation and
relative positions of the sites vary, the similarity of the sequences and
relative proximity to each other suggest that they might serve as functional
units for the maintenance of similar (coordinate) levels of gene expression
under various growth conditions.
Our results suggest that a T-rich sequence downstream of the
RPO21
UAS, and perhaps the one downstream of the
RPO22
UAS, are important for the expression of these genes. Homopolymeric dA:dT
tracts show weak enhancement of the expression of a minimal
CYC1
promoter but can act synergistically with Abf1p, Reb1p and other (usually
upstream) DNA-binding factor sites to enhance gene expression (
17
,
25
). The synergistic activity of the elements depends on the distance between
them, since activation of gene expression falls rapidly when sequences are
interposed between the elements (
25
). Since they are spaced closely, it is possible that the UASs of
RPO21
and
RPO22
also act in synergy with the downstream T-rich tracts.
Our results indicate that an Abf1p-binding site located downstream of two mutated Reb1p-binding sites is able to enhance the expression of a
CYC1
-minimal promoter by 200-fold (Fig.
2
B). This level of enhancement is higher than reported previously for a single
Abf1p site. Buchman and Kornberg (
17
) used a
CYC1-lacZ
reporter plasmid (pCZ[Delta]) similar to the one used in this study, which also contained a cloning
site 12 bp upstream of the 5'-most TATA site of the
CYC1
promoter. In a study of the effect of seven different single Abf1p-binding sites, Buchman and Kornberg observed a maximum stimulation of 9-fold (table 2 in ref.
17
). They did, however, report a synergistic enhancement of up to 56-fold with two adjacent Abf1p sites. We suggest that the large enhancement of transcription which we observe with a single Abf1p site is due to a synergistic interaction between this site and the two mutated Reb1p sites, which may bind Reb1p at a significant level, albeit reduced compared to normal. In support of this hypothesis,
we note the observation of residual binding of Reb1p to mutated sites
in vitro
(Fig.
4
A), even though the activity of these sites apparently is destroyed in the
context of the promoter of
RPO21
.
We found that the elements of the
RPO21
UAS are partially redundant in their roles as activators of
RPO21
expression. The same elements, however, are not redundant when placed in the
context of a minimal promoter. We also note that the
RPO21
UAS has a greater effect on the expression of the
CYC1
minimal promoter (400-fold) than it has in its normal context in the promoter of
RPO21
(10-fold). These apparent discrepancies may be due to the presence of other
elements in the
RPO21
promoter. It has been suggested that homopolymeric dA:dT tracts may enhance the transcriptional-activating effect of other nearby elements that bind transcription factors by freeing the region of nucleosomes that
may interfere with the binding of the factors (
47
). If chromatin interference is a major impediment to the function of DNA-binding factors and a sequence in the promoter is present to counteract
chromatin interference, then perhaps the effect of a single DNA-binding factor is not enhanced significantly by the binding of an additional factor. However, if a chromatin-modulating sequence is not present (for example, in a minimal
promoter), then the binding of a second or third factor may stimulate the
activity of the first factor in counteracting the negative effects of
chromatin. This stimulation may be the result of enhanced binding or perhaps
through increased interactions with general transcription factors that may also
be adversely affected by chromatin (
48
).
We thank Brenda Andrews, Jack Greenblatt, Shahrzad Nouraini and Ian Donaldson
for critical comments on the manuscript and other members of our group for
helpful discussions. We also thank Christopher Brandl, Leonard Guarente, Linda
McBroom and Alessio Vassarotti for providing plasmids, Jonathan Warner for
providing strains and Andrew Buchman for the gift of purified Abf1p protein.
This work was supported by a grant from the Medical Research Council of Canada.
*To whom correspondence should be addressed at present address: Banting and Best
Department of Medical Research, University of Toronto, 112 College Street, Toronto, Ontario M5G 1L6, Canada. Tel: +1 416 946 3017; Fax: +1 416 978 8528;
Email: james.friesen@utoronto.ca
Present addresses:
+
Banting and Best Department of Medical Research, University of Toronto, 112
College Street, Toronto, Ontario M5G 1L6, Canada and
[sect]
Bio-Mega/Boehringer Ingelheim Recherche Inc., 2100 Cunard, Laval, Quebec H7S
2G5, Canada
REFERENCES
Return