| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Institut de Biologie animale (M.T.-P., W.W.)
Université de Lausanne Bâtiment de Biologie CH-1015
Lausanne, Switzerland
Swiss Institute for Experimental Cancer
Research (M.T.-P., S.M.G.) CH-1066 Epalinges, Switzerland
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
Regulatory regions of genes are often maintained free of nucleosomes, as indicated by a high degree of nuclease sensitivity. Although it is generally accepted that sequence-specific factors are involved in the generation of these nuclease-hypersensitive sites, the mechanism by which this is achieved is still unclear. Possibly, the regulatory regions become accessible during DNA replication to sequence-specific factors produced at specific stages of development and tissue differentiation, such that these factors can compete with histones for DNA binding. Alternatively, transcription factors that bind to DNA between or within a nucleosome may themselves promote nucleosome remodeling or attract the appropriate factors to allow the assembly of a functional transcription machinery (Ref. 3 and references therein).
Although nucleosomes are generally thought to play an inhibitory role, DNA sequence-dependent constraints could position a nucleosome in such a way that it facilitates transcription. For instance, a static loop in the DNA may be generated by a positioned nucleosome, which juxtaposes otherwise distant cis-acting elements and favors the interaction and clustering of transcription factors to achieve stimulation of gene activity (1, 6).
The induction of gene expression by nuclear hormone receptors associated with chromatin has served as a model to analyze the interplay between inducible transcription factors and nucleosomal arrangement. A classic example is the opening of the mouse mammary tumor virus (MMTV) promoter by the glucocorticoid receptor (GR) (7). A well positioned nucleosome restricts the access of CTF/NFI to DNA. However, the activated GR can bind to glucocorticoid response elements at the surface of a positioned nucleosome leading to its perturbation in vivo, which enables CTF/NFI to bind and stimulate transcription (Refs. 8, 9 and references therein). A similar mechanism of nucleosome invasion by an activator that is not a nuclear receptor occurs during PHO5 induction in yeast (10). Others have extended such studies with different experimental systems showing disruption of chromatin structures after hormonal stimulation (11, 12, 13), confirming the prevailing view that nucleosomes are inhibitory and need to be structurally altered during the transcription activation process.
Another classic model system for investigating hormone-induced gene expression is provided by the vitellogenin genes (14). A role for several transcription factors, including the estrogen receptor (ER), liver-enriched and ubiquitous factors, has been suggested in the control for the liver- specific expression of these genes (Ref. 15 and references therein; Refs. 16, 17, 18). In vivo experiments have revealed alterations in chromatin structure in the promoter region of these genes after estrogen stimulation (19, 20, 21). Unexpectedly, in vitro transcription experiments showed that the combination of chromatin assembly and transcription factor binding involving a NFI-like factor strongly enhanced transcription from the Xenopus vitellogenin gene B1 promoter (22). Moreover, ER-dependent stimulation of transcription in nuclear extracts suggested that a positioned nucleosome, which wrapped DNA around the histone octamer core to create a loop, bringing distal ER and other binding sites close to proximal promoter elements, potentiates transcriptional activation (6). Although it was suggested that this nucleosome-mediated clustering of transcription factors close to the initiation site was responsible for transcriptional potentiation, it remained to be determined whether a nucleosome is positioned similarly in vivo.
The demonstration that the ER functions in yeast (23) makes it possible to study changes in chromatin structure that are dependent on the receptor and its activation by hormone. In this work, we have reconstituted an estrogen-responsive model system in Saccharomyces cerevisiae, to test whether the nucleosome positioning detected by in vitro chromatin assembly of the vitellogenin gene B1 promoter also occurs in vivo. We have screened for a functional interaction between the human estrogen receptor (hER) and CTF/NF1 (hereafter referred to as CTF1) similar to that observed in vitro, and we have examined how the nucleosomal organization of the vitellogenin gene B1 promoter in yeast affects its estrogen-dependent regulation.
The results from the yeast system demonstrate functional cooperation between hER and CTF1 fusion proteins on a synthetic promoter and indicate that the proline-rich transactivation domain of CTF1 is responsible for this interaction. Chromatin structure analysis in vivo of the natural vitellogenin B1 promoter in yeast cells reveals a nucleosome with multiple positions in the same region as that identified in in vitro assembly experiments. However, in yeast this nucleosome assumes a favored position more promoter proximal, such that the CTF1-binding site is included within the nucleosomal DNA. The location of this nucleosome is not altered by hormone-induced transcription nor by the presence of CTF1 during chromatin assembly. Thus, although hER does bind and activate the B1 promoter when organized into chromatin in yeast, CTF1 is unable to bind its consensus within the B1 promoter chromatin, and consequently fails to activate either alone or in concert with hER. Differences between yeast and mammalian nucleosomes, promoter flanking sequences, and/or the absence of accessory factors may be responsible for this discrepancy.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
). The
correct expression of hER in yeast was analyzed by Western blot (Fig. 2A), and the in vitro
DNA-binding activity in the presence or absence of hormone and
antagonists was analyzed by a gel retardation assay (Fig. 2B).
In agreement with previous finding (23), no DNA-protein complex is
formed when estradiol is omitted (Fig. 2B, lane 2), even though the hER
is highly expressed (see Western blot, Fig. 2A). In contrast,
increasing the amounts of estradiol added to galactose-induced cultures
enhances the levels of specific DNA-protein complex formed in
vitro. In extracts of cells grown in presence of 5
µM of the antiestrogens, tamoxifen, nafoxidine, and ICI
164,384 (hereafter called ICI), alone, no complexes were detected
although hER levels were similar to those found in extracts of cells
grown in absence or presence of estrogen (data not shown). Furthermore,
the levels of complexes were reduced in protein extracts from cells
grown with both 10 nM estradiol and 5 µM
antagonist compared with extracts of cells grown in the presence of
estradiol alone (Fig. 2B), even though the hER was expressed in similar
amounts (Fig. 2A). Together, these results demonstrate that the hER
produced in yeast cells is strictly dependent on estradiol for binding
to DNA in vitro. Moreover, anti-estrogens do not enhance
binding, but rather reduce it, when they are added to the culture
medium together with estrogen.
|
|
). ß-Galactosidase activity was determined
in protein extracts from transformed yeast cells grown in the absence
or presence of increasing amounts of estradiol (Table 1). In the absence of hormone, a basal
ß-galactosidase activity was detected, which corresponds to the
amount found in yeast cells transformed with the control p2HG
expression plasmid that does not carry hER sequences. Depending on the
copy number of the hER expression plasmid, maximal stimulation is
reached with 1 µM estradiol for the CEN-based
plasmid (single-copy number centromere-containing plasmid), and with 50
nM estradiol for the 2µ-based hER expression
plasmid (Table 1). To analyze the effects of antiestrogens on the
transcriptional activity of the hER in yeast on a CEN-based
(single-copy) expression plasmid, ß-galactosidase activity was
determined in protein extracts isolated from galactose-induced
cells grown in the presence of 5 µM antagonist, with or
without the addition of 60 nM estradiol. None of the tested
antagonists, tamoxifen, nafoxidine, or ICI, showed any agonist activity
in the absence of estrogen (Table 2). In
the presence of 60 nM estradiol and 5 µM of
either tamoxifen, nafoxidine, or ICI, the ß-galactosidase levels drop
by 40–55% compared with those with 60 nM estrogen alone
(Table 2). The effect of antiestrogens on the transcriptional activity
of hER is less pronounced than that observed on in vitro DNA
binding (Fig. 2B). This discrepancy, however, reflects the nature of
the two different methods used: ß-galactosidase accumulates in the
yeast cell during the growth in the presence of overexpressed hER,
whereas the gel retardation assay represents the DNA binding activity
of hER at a given timepoint. Our results allow us to conclude that the
expressed hER mediates a strictly estrogen-dependent and an
antagonist-sensitive activation of transcription in yeast.
|
|
). First, whole-cell extracts were
tested for CTF1 DNA-binding activity by gel retardation assay (Fig. 3C). Only extracts isolated from galactose-induced cells show a
specific DNA-protein complex (lane 2), in contrast to extracts isolated
from cells grown in the presence of glucose, in which no complexes are
formed (lane 1).
|
) containing one copy of a
consensus CTF/NFI DNA-binding site (29) upstream of the minimal yeast
CYC1 promoter that drives expression of lacZ
coding sequences. ß-Galactosidase activity was determined in protein
extracts from yeast cells transformed with both the CTF1 expression
plasmid and the CTF/NFI-dependent reporter plasmid. CTF1 induces a weak
7-fold stimulation of ß-galactosidase activity as compared with the
basal activity of the reporter gene cotransformed with the control p2GH
expression plasmid lacking CTF1-coding sequences (Fig. 3D). To test
whether this low transcription activation mediated by CTF1 reflects the
weak activity in yeast of proline-rich transactivation domains, the
latter was replaced by the C-terminal 129 amino acids (aa) of yeast
Gal4, which is an acidic transcription activation domain (see
Materials and Methods and Fig. 3A). The expression and
DNA-binding activity of this fusion protein, CTF1-Gal4, was tested by
gel retardation assay (Fig. 3C). An abundant and specific DNA-protein
complex is only visible in extracts isolated from galactose-induced
cells (lane 4). Furthermore, the hybrid CTF1-Gal4 factor is 50 times
more active in the transcription induction assay than wild-type CTF1
(Fig. 3D). Similarly, a fusion protein comprising the full-length CTF1
and the C-terminal 80 aa of the herpes simplex virus protein VP16 is 95
times stronger than CTF1 (Fig. 3, A and D). These results indicate that
the proline-rich transactivation domain of human CTF1 functions poorly
in yeast, although it does not negatively affect the function of an
acidic domain when associated with it in a fusion protein (VP16-CTF1).
Finally, full substitution of the proline-rich domain by the GAL4
acidic domain activates strongly in vivo.
Functional Interaction between hER and CTF1 in Yeast
Having shown that hER as well as CTF1 and CTF1 derivatives can be
expressed in yeast and that they transactivate their corresponding
reporter genes, we tested the ability of these transcription factors to
function additively or synergistically on a promoter containing both
binding sites. To this end, an ER- and CTF1-dependent reporter plasmid
was constructed by inserting two consensus EREs [(ERE)2]
upstream of the CTF/NFI element in pLGCTF/NFI to give pLGERE-CTF/NFI
(Fig. 4B). In addition, expression
plasmids were constructed that coexpress the hER and CTF1
derivatives under the control of the yeast GAL1
promoter (Fig. 4A). To overcome saturation of induction, which would
mask synergism due to a high overexpression of hER and CTF1-derivatives
with 2µ-based plasmids, a CEN-based plasmid was used for hER
and CTF1 derivative coexpression. This reduces the level of expression
of hER and CTF1 derivatives by about 50-fold, although their
transcriptional activity remains similar (data not shown).
ß-Galactosidase activity was determined in protein extracts from
yeast cells transformed with plasmids expressing either hER, CTF1-GAL4
and hER, VP16-CTF1 and hER, or VP16-CTF
and hER (VP16-CTF is a
VP16-CTF1 derivative, in which the C-terminal 100 aa proline-rich
domain of CTF1 was deleted). Since the activity of hER is strictly
dependent on estrogen (see above), the stimulation mediated by CTF1 or
the CTF1 derivatives can be determined in coexpression
experiments in the absence of hormone. When CTF1 is coexpressed with
hER, the hormone-dependent stimulation is similar to that obtained with
hER alone, indicating that there is no synergism between the two
factors (data not shown). The highest transcriptional stimulation of
the reporter gene is obtained when the hER and VP16-CTF1 fusion protein
are coexpressed in the presence of estrogen. The induced level is
16,000-fold over that of basal promoter activity (Fig. 4C). This strong
estrogen-dependent activation of the reporter promoter represents an
almost 3-fold higher activity than the sum of the transcription
activities obtained with each activator tested separately (Fig. 4C). In
contrast, the stimulation obtained with CTF1-Gal4 and hER is additive;
stimulation obtained with VP16-CTF and hER is slightly more
than additive, suggesting the absence of functional interaction and a
weak synergism, respectively.
|
) of the ß-galactosidase activities from each
individual transformants grown in the presence (E+,
combined effects of CTF1 derivatives and hER) or absence
(E-, effect of CTF1 derivatives only) of estrogen,
i.e. [E+ - E-] (Fig. 4D). The results confirm that the CTF1-Gal4 construct that lacks the
proline-rich activation domain does not synergize with the hER.
Similarly, VP16-CTF, which also lacks the proline-rich domain of
CTF1, shows only a 2-fold higher (E+ -
E-) activity than seen with hER alone. This suggests that
the acidic activation domain of VP16 alone can only weakly synergize
with the hER in yeast. However, since the VP16-CTF1 fusion potentiates
the action of hER by a factor of 4.7, we conclude that the proline-rich
domain of CTF1, in association with the acidic domain of VP16, creates
a strong synergistic activation.
Analysis of the X. laevis Vitellogenin Promoter B1 in
Yeast
We have demonstrated above that CTF1 derivatives and hER are
capable of synergistic activation in yeast of a synthetic promoter
composed of two estrogen-responsive elements and one copy of a
consensus CTF/NFI site, which thus allow functional cooperativity
between factors binding to them. However, it is not clear whether this
holds true for a natural estrogen-dependent promoter with its native
arrangement of ER and CTF/NFI-binding sites and its propensity for
forming nucleosomes. To answer this, we analyzed the promoter of the
vitellogenin B1 gene of X. laevis (30), which contains three
imperfect copies of the ERE at -314, -334 [forming the
estrogen-responsive unit (ERU)], and -555 as well as a strong
CTF/NF1 site at -101 (15, 26, 28). Either the full-length vitellogenin
B1 promoter (-596 until -42) or a 3'-deleted version of it (-596
until -235) was subcloned upstream of the yeast CYC1
minimal promoter driving lacZ, to create yeast reporter
plasmids dependent on the vitellogenin B1 promoter region (Fig. 5). In the latter, the EREs are brought
close to the TATA box, and the strong CTF/NFI binding site is deleted.
In these constructs, the vitellogenin B1 TATA box region (-41 until
+1) was replaced by the yeast CYC1 TATA box sequences (-245
until +1), because the former shows a very high basal activity in
yeast.
|
.
In the absence of estrogen, CTF1-Gal4 showed a weak 3-fold stimulation,
and VP16-CTF1 was unable to stimulate the full-length vitellogenin B1
promoter (pVitB1-CYC1), although both CTF1 and a CTF1 derivative were
able to bind the CTF/NF1 site in gel retardation assays in
vitro (data not shown). Control experiments showed that both
CTF1-derivatives are fully competent for activation from a synthetic
promoter containing a single CTF/NF1-binding site. Furthermore, the
integrity of this binding site in the full-length B1 promoter construct
was confirmed by sequencing. On the other hand, hER mediates a more
than 40-fold estrogen-dependent activation of the full-length promoter
and an even stronger activation (
200-fold) of its 3'-truncated
version (pLGVitB1). Finally, the VP16-CTF1 fusion fails to synergize
with hER on the vitellogenin B1 promoter, although the two proteins
activate synergistically on a synthetic promoter, which bears both
binding sites immediately upstream of the TATA box (see Fig. 4).
Compared with the activation of this latter synthetic promoter, the
stimulation by the hER of the full-length vitellogenin B1 promoter and
its 3'-deleted version is at least 20 times and 5 times lower,
respectively. From these results, we conclude that the EREs in the
native vitellogenin B1 promoter are functional in yeast in a
distance-dependent manner. In contrast, the CTF/NFI-binding site very
weakly promotes transcription by binding CTF1-Gal4 (100–150 times less
than on the synthetic promoter) and is unable to do so by binding CTF1
or VP16-CTF1 either on their own or in combination with hER.
|
). In comparison to the
cleavage of naked DNA, three preferred MNase cleavage sites are
centered around positions -430, -270, and -100 bp (±20 bp) relative
to the start site of transcription on the vitellogenin B1 promoter,
with the latter position being slightly diffuse. The pattern is the
same in the absence or presence of hER and estradiol, and it is
consistent with nucleosomes being positioned between the cleavage sites
as the enzyme cuts preferentially in the linker regions between
nucleosomes.
|
), all DED1
promoter upstream sequences were removed, except for 13 bp upstream
from the unique DED1 TATA box. Analysis of ß-galactosidase
activities with the VitB1-DED1 fusion promoter showed a 80- to 100-fold
higher basal activity as compared with the VitB1-CYC1 fusion promoter
(Table 3). The presence of CTF1 derivatives induces ß-galactosidase
activity 2- to 4-fold over basal level, although the hER-mediated
activation was a mere 4-fold in the presence of estrogen (Table 3).
Analysis of the VitB1-DED1 fusion promoter on a CEN-based reporter
plasmid reduced the basal activity 35-fold as compared with the
2µ-based reporter plasmid, and in turn a 10-fold rather than a 4-fold
enhanced transactivation was observed with the estrogen-stimulated hER
(Table 3).
Indirect end-labeling analysis of yeast nuclei containing the
VitB1-DED1 fusion promoter reveals three preferred MNase cleavages
sites around positions -430, -270, and -100 bp (±20 bp, Fig. 6B),
as seen with the VitB1-CYC1 promoter, although the most
promoter-proximal cleavage at position -100 bp is now precisely
positioned. This shows that the surrounding DNA sequences influence the
positioning of the promoter proximal nucleosome (34). Consistent with
the masking of the CTF/NF1-binding site by the positioned nucleosome,
in none of the conditions tested (CEN- vs. 2µ-based
vitellogenin B1 reporter- or expression plasmids, with variable
estrogen concentrations) did CTF1 and CTF1 derivatives show synergism
with hER (Table 3 and data not shown).
We can draw four conclusions from these results. First, the
vitellogenin promoter sequences are able to position nucleosomes in
yeast as they do in vitro. We note, however, that the
nucleosome positioned between -300 and -140 bp after in
vitro nucleosome reconstitution (6) shifts slightly more promoter
proximal (between -270 and -100 ± 20) in yeast. Interestingly,
the nucleosomal DNA encompasses the CTF/NFI-binding site and raises the
possibility that the site is inaccessible due to the nucleosome.
Second, the unique transcription initiation site in the VitB1-DED1
construct reduces the variation in nucleosome positioning observed on
the VitB1-CYC1 promoter, but conserves the nucleosomal phasing over the
vitellogenin B1 promoter. Third, the region containing the EREs also
falls within a nucleosome, yet these EREs remain accessible to hER.
Fourth, in addition to the positioned nucleosomes over the vitellogenin
promoter, a 150- to 200-bp periodic MNase cleavage pattern is visible
over the lacZ reporter gene, while the CYC1
promoter region and the proximal lacZ gene are either
nucleosome free or do not bear positioned nucleosomes (Fig. 6A).
High-Resolution Footprinting Analysis on the Vitellogenin B1
Promoter in Yeast
To analyze the structure of the vitellogenin B1
promoter with high resolution, we used MNase and deoxyribonuclease I
(DNase I) digestions on gently lysed spheroplasts, followed by primer
extension, which allows base pair resolution of nucleosome locations as
well as the detection of other proteins bound to DNA (35, 36).
Digestions were performed on estrogen-stimulated nuclei isolated from
yeast cells transformed with p2HG-hER or p2HG insert-free control and
pVitB1-CYC1. The genomic footprinting analyses showed identical
nuclease cleavage patterns both in the presence and absence of hER
(activated or silent promoter, data not shown), consistent with the
indirect end-labeling analyses above (see Fig. 6A).
The vitellogenin B1 promoter reveals an unexpectedly complex MNase
cleavage pattern. Analysis of the promoter proximal MNase cleavage
sites in nuclei shows a protection from around -60 (data not shown)
until -171 (±2), with slightly higher accessibility at -98 and -135
(±2, Fig. 7; compare lanes 1+2 with 3+4;
see asterisks). Between -175 and -300 (±2) we also
observe an altered pattern compared with naked DNA with both enhanced
cleavage sites (open arrows) and protected sites (not
indicated) and with a pronounced MNase-specific hypersensitive region
between -225 and -250 (±2; Fig. 7, lane 1). There is also a
pronounced site of cleavage near -180. Similar complex patterns have
been observed by Tanaka et al. (36) and interpreted as
identifying a nucleosome with multiple positions.
|
, solid
arrows). Taken with the end-labeling results (Fig. 6A), we
interpret the complex digestion pattern over the analyzed promoter
region (MNase protection over 100 bp and MNase and DNase I
hypersensitive sites up- and downstream of it) to indicate a nucleosome
whose position is not strictly fixed, but has a restricted mobility
resulting in alternative positions within the region of interest, as
suggested by the scheme of Fig. 7. This proposed nucleosome covers the
CTF/NFI-binding site at -114 to -101, and there is no indication of a
CTF1-specific footprint (data not shown). Thus, the lack of
CTF1-dependent transcriptional activation from this promoter is
consistent with the idea that this nucleosome masks the CTF/NFI-binding
site.
The analysis of the more distal vitellogenin B1 promoter sequence shows
that the region between -340 and -300, including two imperfect EREs
at -302 to -334, is partially protected from MNase digestion (Fig. 7). These observations hold true whether or not hER and estrogen are
present. The fact that hER can stimulate transcription 50 fold from
this promoter indicates that the hER can bind the EREs even though they
might be associated with a nucleosome (see Fig. 6).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
hER Stimulates the Native Vitellogenin B1 Promoter in Yeast
We show here that the hER functions in yeast as it does in
vertebrate cells, confirming observations first made by Metzger
et al. (23). Under our conditions hER requires estrogen to
stimulate transcription of a simple responsive promoter that consists
of an ERE placed directly upstream of the minimal yeast CYC1
promoter. More importantly, hER is also able to activate the natural
vertebrate vitellogenin B1 promoter in yeast cells. In this system, the
antiestrogens, tamoxifen, nafoxidin, and ICI, have no agonistic effect,
but rather they antagonize estrogen action. Previous analyses of the
role of estrogen and antiestrogens in ER DNA binding and
transactivation led to conflicting results, most likely due to the
experimental conditions, as well as the different sources and forms of
ER (see Ref. 37 and references therein). Since the natural vitellogenin
gene B1 promoter is readily stimulated in yeast by the liganded ER, we
were able to analyze potential interactions of the hER with other
transcription factors, notably with CTF1, in a controlled chromatin
environment.
The Proline-Rich Transactivation Domain of CTF1 Is Required for a
Synergistic Interaction with the hER in Yeast
Our previous demonstration of a functional interaction between
members of the CTF/NFI family and hER in vitro suggested
that their synergistic interaction results from the stronger tethering
of a common factor that participates in the transcriptional machinery,
rather than from cooperative DNA binding (28). Furthermore, the
activation domains of both hER and CTF1 were required for the observed
synergism (27). In yeast we find that the proline-rich activation
domain of CTF1 is roughly 50–100 times weaker than the acidic
transciptional activation domains of Gal4 or VP16 (see also Refs. 38, 39). However, addition of the strong acidic transcription
activation domain of VP16 to the N terminus of CTF1 resulted in a
5-fold synergism between the hER and VP16-CTF1 on a synthetic promoter.
Synergism was observed only when VP16-CTF1 and hER were expressed from
CEN-based plasmids that avoid high levels of overexpression, suggesting
either that the promoter is quickly saturated or that a yeast factor
involved in transactivation is limiting and is titrated out by the
overexpressed polypeptides. Similar results were found previously in
HeLa cells expressing GAL4-CTF1 fusion proteins and hER on a synthetic
promoter (27). Interestingly, no synergism could be detected when the
proline-rich transcription activation domain of CTF1 was replaced by
the acidic transcription activation domain of Gal4, and only a 2-fold
synergism was observed when the proline-rich transcription activation
domain of CTF1 was deleted in the VP16-CTF chimera. Thus, we
conclude that in yeast, as in vertebrate cells, the proline-rich domain
of CTF1 participates in the strong synergism between VP16-CTF1 and hER.
The proline-rich domain might either make the acidic domain more
accessible or with the aid of this latter domain it may itself promote
interaction between hER, CTF1, and yeast coactivators. Since the
proline domain does not activate on its own, it is unlikely that it
interacts with the basal transcription machinery.
A Positioned Nucleosome Masks the CTF/NFI Site on the Vitellogenin
B1 Promoter in Yeast
We were unable to detect stimulation of the X. laevis
vitellogenin B1 promoter by CTF1 and observed only a weakly
transactivating activity of CTF1 derivatives, although all can bind the
CTF/NFI site of the vitellogenin B1 promoter in vitro. This
result suggests that the CTF/NFI-binding site is masked in yeast cells.
The vitellogenin B1 promoter contains a strong nucleosome positioning
element that places a nucleosome between -300 and -140,
i.e. close to the CTF/NF1 site, during in vitro
chromatin assembly (6). Analyses of both MNase- and DNase I-digested
yeast nuclei are consistent with the presence of a nucleosome with
multiple positions between -250 and -80 on the vitellogenin B1
promoter. The nucleosome appears to be shifted slightly toward the
transcription initiation site when compared with that mapped from
in vitro chromatin assembly (6). The local heterogeneity
revealed by the high-resolution MNase mapping has also been observed in
high-resolution chromatin structure analyses of the yeast
URA3 gene (36). These authors also detect unexpectedly
complex digestion patterns for MNase and DNase I, despite the clear
chromatin features observed by low-resolution mapping. From their
high-resolution mapping of six nucleosomes on the URA3 gene
they could conclude that despite the local heterogeneity revealed by
the high-resolution mapping, the low-resolution map is a reasonably
accurate representation of the chromatin structure. In view of these
results, our low- and high-resolution nuclease-sensitivity patterns are
most consistent with a population of overlapping nucleosome positions
over the proximal vitellogenin B1 promoter, which cover the CTF/NFI
site (-101 to -114). This appears to render the site inaccessible to
CTF1 or to derivatives of CTF1 and may explain the lack of
transcriptional activation by these factors on the vitellogenin
promoter in yeast.
It was recently shown that the proline-rich domain of CTF1 interacts specifically with histone human H3, and full-length CTF1 was shown to alter the interaction of reconstituted nucleosomal cores with DNA (40). Were this true in yeast, we might expect that CTF1 would displace the positioned nucleosome on the B1 promoter. Recent studies suggest, however, that CFT1 interacts with yeast histone H3 with significantly reduced affinity (N. Mermod, personal communication), possibly rendering it unable to remodel yeast chromatin even during assembly after DNA replication. Consistently, using a specific DNA-bending sequence to direct CTF/NFI-binding sites to different positions around an in vitro reconstituted nucleosome, DNA-binding studies using CTF1 showed a 100- to 300-fold reduced binding affinity for all nucleosomal targets as compared with free DNA (41). It was concluded that nucleosomal inhibition of CTF/NFI binding was an inherent characteristic of the factor, which is in excellent agreement with our observations on the vitellogenin promoter. However, there is also some support from in vivo experiments for the idea that CTF might bind to a preexisting nondisplaced nucleosome on the MMTV promoter (42, 43). In vivo, transcriptional activation of this promoter by the GR recruits CTF1 and the octamer factor to their binding sites within the chromatin modified by the receptor (44). After transient GR activation, these binding sites are reincorporated into the positioned nucleosome, and transcription is repressed (45). A CTF2-dependent hormone activation of this promoter has also been shown in yeast (42). A porcine CTF2 that lacks 69 aa at its C terminus, which is produced through differential splicing of the NFI gene transcript (38), was used and shown to bind its site within a nucleosome. The strongly synergistic interaction between GR and CTF2 was dependent on low-level expression of GR, low ligand concentration, and high-level expression of CTF2 (42).
The variety of results from in vitro and in vivo
analyses of the chromatin structure of the MMTV promoter (42, 43, 44, 45, 46, 47, 48, 49) may
reflect the effects of flanking sequences on nucleosome positioning
(34). Consistently, in our analyses we observe that the position of the
most promoter-proximal nucleosome is influenced by the minimal yeast
promoter fused to the vitellogenin B1 sequences (compare Fig. 6, A and
B). We were obliged, however, to use yeast sequences for the minimal
promoter elements (either CYC1 or DED1) because
the proximal region of the vitellogenin B1 promoter (-41 to +1)
confers a very high basal activity in yeast (data not shown). The
exchange of yeast for Xenopus sequences in this region may
also account for the shift of the nucleosome from its position mapped
in vitro (-300 to -140) to the position from
high-resolution mapping in yeast (-250 to -80).
Surprisingly, despite the strong activating potential of hER during estrogen stimulation, no qualitative differences were observed by end-labeling analyses on the positioned nucleosome (-270 to -100 ± 20) in the presence or absence of hER and estradiol. Our analysis suggested the presence of a second nucleosome from -430 to -270 (±20), a region of the vitellogenin B1 promoter that contains the ERU, suggesting that hER can bind to the ERU even when wrapped around a nucleosome. Alternatively, the hER-inducible expression might derive from only a small subpopulation of plasmids.
From our previous in vitro studies, we postulated that the folding of DNA around a positioned nucleosome on the vitellogenin B1 promoter creates a static loop, which facilitates the interaction of the ER with transcription factors bound to the proximal promoter elements (6). We show here that the deletion of most of the nucleosome positioning sequences enhances transcription about 5-fold, as compared with the full-length promoter, indicating that the nucleosome-positioning sequences are not essential for activation in yeast. However, the deletion of these nucleosome-positioning sequences does not significantly increase estrogen-dependent transcription of the same promoter in transient transfection studies using primary hepatocytes (50). The nucleosome-induced loop should not only favor the interaction of ER with CTF1, but also with proximal promoter-bound factors such as the liver-enriched factors CCAAT/enhancer-binding protein (C/EBP) and hepatic nuclear factor 3 (HNF3) (see Ref. 15), that bind between the -100 boundary of the positioned nucleosome and the TATA box. Although these factors are lacking in yeast, it should now be possible to test whether a nucleosome loop aids activation by hER when HNF3 and/or C/EBP are coexpressed with hER in yeast carrying the vitellogenin gene B1 promoter. It must, of course, be kept in mind that several different aspects of chromatin structure influence the access of transcription factors.
In conclusion, we have investigated the importance of a positioned nucleosome on the hormonal activation of the vertebrate vitellogenin gene B1 promoter in yeast, in particular with respect to the synergism between the hER and CTF1. The hER appears to bind and activate transcription without disrupting a positioned nucleosome downstream of the EREs. This same nucleosome appears to mask the CTF/NFI site and prevent functional interaction between the hER and CTF1 in yeast. In view of these results, it is now of interest to define the position of nucleosomes in hepatocytes with the vitellogenin B1 promoter embedded in its natural context, to determine whether the mechanism of vitellogenin gene B1 induction relies on the interaction of the ER with members of the CTF/NFI family, or with those of the C/EBP and HNF3 families. This should contribute to our further understanding of the tissue-specific expression of this gene.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
-
l)-N-methylundecanamide)
was a gift from Dr. A. Wakeling, (ICI Pharmaceuticals, Alderley Park,
Macclesfield, UK). 17ß-Estradiol, tamoxifen, nafoxidine, and general
reagents were purchased from Sigma Chemical Co. (St. Louis, MO). Yeast
medium components were obtained from Difco Laboratories (Detroit,
MI).
Yeast Strains
The protease-deficient strain GA24 (MATa
GAL+ pep4–3 ura3 his3 bar1 suc29)
was used for all experiments. All transformations into this strain were
accomplished by using the lithium acetate transformation protocol
(51).
Construction of Plasmids
The expression vector p2HG-hER was obtained as follows. The hER
vector, ER (hERval400-mutant), and p2HG (52) were gifts of
D. Picard. The vector ER was digested with BamHI, and the
hER cDNA was subcloned into the BamHI-digested plasmid
pGAL1, resulting in pGAL1-hER. The plasmid pGAL1 contains a 907-bp
SalI-BamHI fragment of the yeast
GAL1-GAL10 promoter (53) subcloned into the SalI
and BamHI sites of YEp24 (50). pGAL1-hER was digested with
SalI and ApaI, and the resulting 3.6-kb fragment
containing the promoter GAL1-GAL10, hER, and part of YEp24
was subcloned into the SalI- and ApaI-digested
p2HG, yielding plasmid p2HG-hER. The human CTF1 expression plasmid was
constructed as follows. The human CTF1 cDNA was isolated as a 1.7-kb
NcoI-XhoI fragment from pCTF1 (54) and subcloned
into the NcoI and XhoI sites of p2HG-GAL1. To
obtain p2HG-GAL1, a MluI linker was inserted into the filled
BamHI site of pGAL1, and NcoI and XhoI
sites were added before the 907-bp yeast GAL1-GAL10 promoter
was inserted as a SalI-XhoI fragment into p2HG.
p2HG-CTF1/hER resulted from subcloning the 3.6-kb
SalI-ApaI fragment, containing the hER under the
control of the yeast GAL1 promoter, from pGAL1-hER into the
XhoI-ApaI sites of p2HG-CTF1. The hybrid protein
(CTF1-GAL4) comprises the first 398 aa of human CTF1 fused to the
C-terminal 129 aa of yeast GAL4. The expression vector for the fusion
protein (CTF1-GAL4) was derived as follows. A 444-bp
PvuII-EcoRI fragment containing the acidic
activation domain of GAL4 was isolated from pSG4 (55) and ligated into
the Klenow-filled BglII and EcoRI sites of pCTF1,
replacing the proline-rich transcription activation domain of CTF1. The
resulting plasmid, pBS-CTFGAL4, was digested with NcoI and
XhoI, and the 1.7-kb fragment was subcloned into p2HG-CTF1
to create p2HG-CTFGAL4. p2HG-CTFGAL4/hER arose then from p2HG-CTFGAL4
as p2HG-CTF1/hER above. The fusion protein (VP16-CTF1) comprises the
C-terminal 80 aa of the herpes simplex virus protein VP16 fused
N-terminal to the human CTF1. The expression vector for the hybrid
protein (VP16-CTF1) was constructed as follows. The CTF1 cDNA was
isolated from pCTF1 as a Klenow-filled SpeI and
XhoI fragment and inserted downstream from the acidic
transcription activation domain of VP16 into the Klenow-filled
EcoRI site of pSD10 (56), resulting in pSD-VP16CTF1. Then
the KpnI and SacII sites in pSD-VP16CTF1 were
flush-ended with T4 DNA polymerase, and the GAL1 promoter
and VP16CTF1-containing fragment were isolated and ligated into the
Klenow-filled and dephosphorylated SalI site of
p2HG-hER, resulting in p2HG-VP16CTF1/hER. The fusion protein
(VP16-CTF) has the C-terminal 100 aa long proline-rich domain
deleted by digesting pSD-VP16CTF1 with BglII and
XhoI and religation of the Klenow blunt-ended sites
resulting in pSD-VP16CTF. p2HG-VP16CTF/hER arose then from
pSD-VP16CTF as p2HG-VP16CTF1/hER above.
CEN-based expression vectors were obtained by replacing the 2µ containing 2.6-kb SpeI-NotI fragments in p2HG-hER, p2HG-CTFGAL4/hER, and p2HG-VP16CTF1/hER with a 2.5-kb SpeI-NotI fragment containing the ARS1 and CEN4.
Reporter plasmids were constructed as follows. The parent vector used
for all reporter plasmids was pLG669Z (51). To obtain
pLGSmaI-XhoI, pLG669Z was digested by
SmaI and XhoI, the XhoI site was
filled by Klenow DNA polymerase, and the SmaI and filled
XhoI sites were religated, restoring the filled
XhoI site. pLGERE was obtained from D. Picard and is the
same ERE reporter plasmid as used in Ref. 52 . The plasmid pLGCTF/NF1
was obtained by introducing a synthetic oligonucleotide corresponding
to the consensus CTF/NFI site into the Klenow-filled XhoI
site of pLGSmaI-XhoI. pLGERE-CTF/NFI was
constructed by cloning the consensus CTF/NF1 site oligomer into the
HincII and SmaI sites of pBSKS+ (51),
resulting in pBS-CTF/NF1. Then a 75-bp BamHI fragment,
containing two consensus EREs, was ligated into the BamHI
site of pBS-CTF/NFI. The resulting pBSERE-CTF/NFI was then digested
with SpeI and the site was filled with Klenow polymerase and
further digested with XhoI. The XhoI and filled
SpeI fragment, containing the (ERE)2-CTF/NFI DNA elements,
was then subcloned into the XhoI-SmaI-digested
pLG669Z, yielding pLGERE-CTF/NFI. The VitB1-based reporter plasmids
were constructed by adding a XhoI linker into the
Klenow-filled BglII site of pB1(-596/+8)CAT8+ (57) and then
subcloning the VitB1 promoter as a HincII-XhoI
fragment into the SmaI-XhoI sites of pLG669Z,
resulting in pVitB1-CYC1. pLGVitB1 was obtained by subcloning the
0.8-kb ApaI-SspI fragment from pVitB1-CYC1 into
the ApaI and Klenow-filled XhoI sites of
pLGSmaI-XhoI, and pVitB1-DED1 was obtained by
subcloning the 1.1-kb ApaI-XhoI fragment fom
pVitB1-CYC1 into the ApaI-XhoI sites of pLS168
(58). CEN-based reporter plasmids were obtained by replacing the 2µ
containing the 1.9-kb SpeI fragment in pVitB1-CYC1 and
pVitB1-DED1 with a 2.5-kb SpeI fragment containing the
ARS1 and CEN4.
Growth of Yeast Cells and Preparation of Protein Extracts
Cells were grown and protein extracts prepared as follows.
Transformed yeast cells were grown in synthetic drop-out medium lacking
uracil and histidine (51) and supplemented with 2% (wt/vol) glucose,
3% (vol/vol) glycerol, and 2% (wt/vol) lactate. At late-log phase the
cultures were diluted (1:100) into the same medium without glucose and
the growth continued for 24 h, after which galactose was added to
2% (wt/vol) to induce the yeast GAL1 promoter and estrogen
or antagonists as needed. After 8 h, cells were harvested, washed
first with water and then with buffer A [50 mM Tris-HCl
(pH 7.5), 1 mM EDTA, 1 mM EGTA, 10% (vol/vol)
glycerol, 15 mM MgCl2, 5 µM
ZnSO4, 50 mM NaCl, 20 mM
Na-molybdate, 25 mM sodium fluoride, 1 mM
sodium bisulfite, 1 mM dithiothreitol, 1 mM
phenylmethylsulfonyl fluoride (PMSF), and 1 µg/ml each of the
following protease inhibitors: antipain, benzamidine, chymostatin,
leupeptin, and pepstatin A]. The washed cells from a 15-ml culture
were resuspended in 0.5 ml buffer A in 15-ml round bottom plastic tubes
and frozen rapidly in liquid nitrogen. Cells were thawed in ice and
0.5 g glass beads (250 µm) was added, after which the cells were
vortexed at maximum speed in the cold, 6 x 1 min bursts with
cooling in ice between. Glass beads and cell debris were removed by
centrifugation (10 min at 15,000 x g, 4 C). The
supernatants were transferred into 1.5-ml microcentrifuge tubes, and
the centrifugation was continued for 30 min in a microcentrifuge,
13,000 x g, 4 C. The protein concentration in the extracts
was determined by the Bradford method, using the Bio-Rad protein assay
(Bio-Rad, Richmond, CA).
Antibody Preparation
An Escherichia coli ß-galactosidase
(Met1-Asp374) and X. laevis ER
(Asp276-His508) fusion protein was cloned into
pEX2 (59) and expressed in E. coli. The
overexpressed fusion protein precipitated inside the bacterial cell and
could be easily purified as insoluble protein. The fusion protein was
separated by SDS-PAGE (51), and the corresponding 67-kDa band was cut
out and lyophilized. About 10 µg of fusion protein were rehydrated,
mixed with complete Freund’s adjuvant, and injected ip into a rabbit.
The animal was boosted at 6-week intervals with a mixture of fusion
protein and incomplete Freund’s adjuvant and bled 1–2 weeks after
each boost.
Immunoblot Analysis
Protein extracts were separated on 8% SDS-polyacrylamide gels
and transferred to nitrocellulose as described (51). The (1:1000)
diluted polyclonal antibody against the X. laevis ER was
used for the indirect detection of the yeast expressed hER. Specific
antigen-antibody complexes were detected with the enhanced
chemiluminescence (ECL) reagent (Amersham, Arlington Heights, IL).
DNA Mobility Shift Assay
Protein extracts in buffer A were preincubated with poly
(dI·dC)·poly (dI·dC), (0.1 mg/ml) at 0 C for 15 min, after which
the 32P-end-labeled DNA probe was added to a final volume
of 20 µl and the incubation was continued at 20 C for 15 min. The
protein-DNA complexes were analyzed on nondenaturing 5% polyacrylamide
gels (15, 51).
Assay of ß-Galactosidase
The analysis of the ß-galactosidase activity in the protein
extracts was performed as described in Ref. 51 using the given formula
to calculate the specific activity.
Preparation of Yeast Nuclei and Nuclease Digestions
Transformed yeast cells were grown mainly as described above,
except the growth in galactose and estradiol was reduced to 5–6 h.
Nuclei were isolated from 2 liters of culture, about 16 g wet
cells, following the method described (35). Cells were harvest by
centrifugation and spheroplasted as described in Ref. 60 at 10 ml/g wet
cells in synthetic drop-out medium lacking uracil and histidine and
supplemented with 1 µM estradiol, 2% (wt/vol) galactose,
3% (vol/vol) glycerol, 2% (wt/vol) lactate, 1.1 M
sorbitol, 0.15 mg/ml Zymolyase (Seikagaku, Chuo-ku, Tokyo), and
800 U/ml lyticase (60). Spheroplast formation continued at 30 C for 30
min with gentle agitation. After centrifugation at 4 C, spheroplasts
were washed twice by gentle but thorough resuspension in 20
mM K-PO4 buffer (pH 6.5) supplemented with 1.1
M sorbitol and 0.5 mM PMSF at 4 C. Washed
spheroplasts were resuspended in 80 ml buffer F [20 mM
PIPES (pH 6.5), 18% (wt/vol) Ficoll 400, 0.125 mM
spermidine, 0.05 mM spermine, 0.5 mM EDTA, and
1 mM PMSF] and divided into four 40-ml polycarbonate
tubes. Each tube was vortexed at maximum speed for two 5-min bursts at
4 C. The sheared spheroplasts (25 ml each) were carefully layered onto
18 ml buffer GF [20 mM PIPES (pH 6.5), 20% (vol/vol)
glycerol, 7% (wt/vol) Ficoll 400, 0.125 mM spermidine,
0.05 mM spermine, 0.5 mM EDTA, and 1
mM PMSF] and centrifuged at 22,500 x g
for 30 min at 4 C. Supernatants were discarded and each pellet
resuspended in 20 ml buffer F (without EDTA). Vortexing was repeated as
above, and unlysed cells and cell debris were removed by centrifugation
at 3,400 x g for 15 min at 4 C. Nuclei containing
supernatants were transferred into fresh tubes and were divided
appropriately so that some could be resuspended in digestion buffer and
some in control buffer. Nuclei were recovered by centrifugation:
22,500 x g for 30 min at 4 C. Supernatants were
discarded, and the crude nuclei-containing pellets were resuspended as
follows. For undigested controls the nuclei were resuspended in 0.5
ml/g wet cells control buffer [10 mM HEPES (pH
7.5), 0.125 mM spermidine, 0.05 mM spermine, 10
mM EDTA, and 1 mM PMSF], divided into 220-µl
aliquots, and DNA was purified as described below. Nuclei to be
digested were resuspended in 0.5 ml/g wet cells digestion buffer
[10 mM HEPES (pH 7.5), 0.5 mM
MgCl2, 3 mM CaCl2, and 1
mM PMSF], divided into 200-µl aliquots, and preincubated
at room temperature or at 37 C for 5 min. Nuclease digestions were
started by adding 10 µl MNase or DNase I diluted in digestion buffer.
The digestions were stopped after 10 min by the addition of 10 µl 0.2
M EDTA. DNA from nondigested and digested nuclei was
purified by incubation for 30 min at 37 C with 10 µl (10 mg/ml)
ribonuclease A, after which 30 µl 22% (vol/vol) sarkosyl, 30 µl
1.7 M NaCl, and 30 µl (10 mg/ml) proteinase K were added
in this order with mixing between. The proteinase K digestion was
performed at 37 C for 3–6 h, after which 30 µl 5 M
NaClO4 were added, followed by extensive phenol and
phenol/chloroform extractions. DNA was precipitated with 990 µl
ethanol at room temperature, washed with 75% (vol/vol) ethanol, and
resuspended in 100 µl 10 mM HEPES (pH 7.5).
Indirect End-Label Analysis
The cleavage sites of MNase-digested yeast nuclei were
located by the indirect end-label method (61, 62). Purified DNA from
digested nuclei and digested DNA controls were cleaved by
XbaI, and the DNA fragments were separated by agarose gel
electrophoresis and blotted onto nylon filter according to Southern
(51). Vitellogenin B1-specific DNA sequences were probed with a
32P-end-labeled 150-bp XbaI-AccI
fragment (-596 to -446 of the vitellogenin B1 promoter).
Primer Extension Assay
MNase and DNase I cleavage sites were located by primer
extension assay mainly as described (35). The oligonucleotide used as
primer includes the X. laevis vitellogenin B1 promoter up to
bp -53 and overlaps the CYC1 gene (SG102:
5'-CCTGGCGGATCTGCTCGAGGGATCTGGCGCATGTGC-3'). The DNA/primer mix was
denatured at 93 C for 90 sec, annealed at 62 C for 4 min, and then
extended at 72 C for 3 min. This cycle was repeated 35 times with 10
min at 72 C for final extension. Reactions were terminated by
chloroform extraction and ethanol precipitation. Dideoxy sequencing
reactions using Taq polymerase were performed on plasmid DNA
as described (35), using the same primer extension reactions as above.
The DNA products were analyzed on 8% polyacrylamide/7 M
urea sequencing gels with wedge spacers (51).
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
This work was supported by grants from the Swiss National Science Foundation to S.M.G. and to W.W. and by the Etat de Vaud. M.T.-P. was supported by the Swiss Institute for Experimental Cancer Research.
Received for publication January 27, 1998. Revision received May 6, 1998. Revision received July 7, 1998. Accepted for publication July 8, 1998.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONRESULTSDISCUSSIONMATERIALS AND METHODSREFERENCES |
|---|
