Cross-Linking of Estrogen Receptor
to Chromatin in Intact MCF-7 Human
Breast Cancer Cells: Optimization
and Effect of Ligand
Carol K. Wrenn and Benita S. Katzenellenbogen
Department of Physiology and Biophysics
University of Illinois and
University of Illinois College of Medicine
Urbana, Illinois 61801
To investigate the effect of ligand (be it hormone,
antihormone, or no hormone) on the interaction between estrogen receptor (ER) and chromatin, we
have used formaldehyde as a cross-linking agent in
intact MCF-7 human breast cancer cells. After a 1to 2-h hormone treatment, the cells are exposed for
8 min to formaldehyde, which is added directly to
their culture medium to minimize environmental perturbation. Nuclei are prepared from formaldehydetreated cells and their contents are fractionated on
CsCI density gradients to separate DNA-protein
complexes from free protein. Peak gradient fractions
are assayed for the presence of specific proteins by
immunoblot of sodium dodecyl sulfate-polyacrylamide gel patterns. Using this approach, we find that
0.15% formaldehyde is optimal for cross-linking ER
to chromatin. We detect ER and the large subunit of
RNA polymerase II with DNA from formaldehydetreated, but not from untreated cells. On the other
hand, actin (a cytoplasmic protein) and small nuclear
ribonucleoprotein particle proteins (nuclear RNA
binding proteins) are not cross-linked to DNA. Therefore, cross-linking appears to be selective and fractionation is efficient. Interestingly, we detect similar
levels of ER (as well as RNA polymerase II) with
DNA from formaldehyde-treated cells, regardless of
whether the cells are preexposed to estrogen (17/?estradiol at 10~8 M), antiestrogen (IC1164,384 at 10~7
or 10~6 M), or no hormone. These results, using
covalent cross-linking in intact cells, indicate that
both ligand-occupied and unoccupied ER are associated with chromatin. (Molecular Endocrinology 4:
1647-1654, 1990)
ous) are governed to a large extent by estrogen, it is
important to understand the mode of estrogen action.
According to our current understanding, steroid hormones modulate growth and development via their
interaction with specific intracellular receptor proteins.
Upon hormone binding, steroid hormone receptors
function as trans-acting transcriptional regulatory factors (1-3). DNA elements that modulate hormonal effects on transcription in vivo have been correlated with
receptor binding sites in vitro for several hormoneregulated genes. A comparison of DNA sequences
yields a consensus 13-nucleotide sequence for estrogen-responsive elements (EREs) and a 15-nucleotide
sequence for glucocorticoid and progesterone-responsive elements (3), which are palindromic and bind receptor homodimers (4, 5).
Although the mechanism of transcriptional enhancement is unknown, recent reports indicate that functional
interactions between receptors and other transcription
factors may be involved. Several laboratories have demonstrated a synergistic action of steroid responsive and
other promoter elements (CCAAT box, CACCC motif,
NF1 and SP1 binding sites) on transcription from reporter plasmids (6-9). In addition, in vivo footprinting
studies have shown steroid-dependent recruitment of
transcription factors onto the promoters of hormone
responsive genes (10, 11). The latter results indicate
that hormone induction of transcription involves receptor-mediated establishment of a transcription complex
at the promoter rather than activation of a preexisting
complex. Whether the interaction between steroid receptors and other transcription factors is direct via
protein-protein interaction, or indirect by modification of
DNA structure, is unknown.
We are particularly interested in the effect of ligand
on estrogen receptor (ER)-chromatin interaction. Demonstration that unoccupied ER is localized primarily in
the nucleus (12-16) raises the possibility that ER is
associated with responsive genes in the absence of
hormone, and is merely activated upon hormone binding. Indeed, Lees et al. (17) used gel mobility shift to
demonstrate that unoccupied ER can bind specifically
INTRODUCTION
Since the proliferation and phenotype of cells from
female reproductive tissues (both normal and cancer0888-8809/90/1647-1654$02.00/0
Molecular Endocrinology
Copyright © 1990 by The Endocrine Society
1647
Vol 4 No. 11
MOL ENDO-1990
1648
with an ERE in vitro under the conditions of their assay.
However, Kumar and Chambon (4) as well as Martinez
and Wahli (18) detected slight or no binding of unoccupied ER using somewhat different conditions. In addition, in vivo footprinting revealed no protection of
EREs flanking an estrogen inducible gene in hormonenaive liver from chicken (11). The latter indicates that
any association between unoccupied ER and DNA in
vivo must be relatively weak or nonspecific; however,
no other systems nor other genes have been tested in
intact cells. On the other hand, in all cases tested,
estrogen-ER and antiestrogen-ER complexes bind
strongly to the same DNA elements in vitro (4,18), and
probably in vivo as well (19). By definition, antiestrogens
antagonize estrogen action through competitive binding
of ER, without functional activation. However, several
antiestrogens function as weak agonists alone (20, 21).
Presumably, these various ER-ligand complexes interact differently with the DNA and/or with other transcription factors. Indeed, Meyer et al. (22) demonstrated that
progesterone receptor and ER compete, directly or
indirectly, for factors that mediate their enhancer functions. Treatment with antiestrogens eliminated the competition, such that progesterone receptor-controlled reporter genes were transcribed at an elevated level.
To study the effect of ligand on ER-chromatin association, we have used formaldehyde as a cross-linking
agent in intact MCF-7 human breast cancer cells. Formaldehyde rapidly penetrates into cells and forms covalent protein-protein and protein-nucleic acid bridges.
Cross-linking occurs within minutes, thereby preventing
large-scale redistribution of macromolecules. Solomon
and co-workers (23) used formaldehyde on intact Drosophila cells to identify specific histone H4 and RNA
polymerase II contact sites within hsp70. In addition,
Schouten (24) demonstrated the use of formaldehyde
cross-linking to study proteins associated with pBR322
DNA and mRNA in Escherichia coli cells.
Here, we report the optimization and detection of
ER-chromatin cross-linking via formaldehyde in intact
MCF-7 cells. We detect comparable levels of ER-chromatin cross-linking regardless of whether the cells are
exposed to estrogen, antiestrogen, or no hormone at
all, indicating that ER associates with chromatin even
in the absence of ligand.
Formaldehyde was added directly to the culture medium and exposure time was limited to 8 min, as
described by Solomon et al. (23), to minimize environmental perturbation and molecular rearrangement. Nuclei were prepared and their contents were fractionated
on CsCI gradients to separate DNA from free protein,
and fractions were assayed for the presence of ER by
immunoblot.
Figure 1 illustrates the effect of increasing formaldehyde concentration on the absorbance profiles of CsCI
gradients. Using untreated cells (0% HCHO), RNA pellets to the bottom, DNA forms an absorbance maximum
at 260 nm just above this, and protein remains at the
top of the gradient (280 nm maximum). Therefore,
fractions are collected from the bottoms of gradients to
avoid contamination of DNA with free protein. Formaldehyde (0.05%, 0.15%, and 0.25%) cross-links proteins
to the nucleic acids, reducing their densities. Thus, the
RNA and DNA peaks move up the gradient with increasing formaldehyde concentration. The lower peak of 260
nm absorbance is eliminated when nuclear lysates are
digested with RNase A before gradient fractionation,
and material from the middle, major peak of 260 nm
absorbance is degraded by DNase I (data not shown),
confirming the identities of these peaks as RNA and
DNA, respectively. Use of formaldehyde concentrations
greater than 0.75% causes the nuclear material to
become increasingly insoluble. Peak heights are reduced or, ultimately, eliminated (data not shown).
To determine the profile of ER on the CsCI gradients
illustrated in Fig. 1, fractions corresponding to regions
of the absorbance profile were pooled, dialyzed to
remove CsCI, and digested with nucleases. Residual
proteins were fractionated by sodium dodecyl sulfate
(SDS)-polyacrylamide gel electrophoresis, transferred
to nitrocellulose, and probed with anti-ER monoclonal
0%
1
0.15%
/
0.25%
n
\A
A260
A280
4
RESULTS
Optimization of Formaldehyde Concentration for
ER-Chromatin Cross-Linking
Formaldehyde was used as a cross-linking agent in
intact MCF-7 cells to study the interaction between ER
and chromatin. Initial experiments were designed to
determine the minimum amount of formaldehyde
needed to maximize ER-chromatin linkage, yet avoid
extensive cross-linking to remote structures via intermediate molecules. To this end, MCF-7 cells were
treated with increasing concentrations of formaldehyde.
0
i
i
I
5
10
15
20
traction
Fig. 1. Effect of Formaldehyde Concentration on CsCI Gradient
Absorbance Profiles
MCF-7 cells were treated with 10~8 M E2 for 1 h, followed
by 0, 0.05, 0.15, or 0.25% formaldehyde for 8 min. Nuclei
were prepared, broken by sonication in 0.5% Sarkosyl, and
their contents fractionated on CsCI gradients, as described in
Materials and Methods. Six tenths ml fractions were collected
from the bottoms of gradients, and their absorbance at 260
(solid lines) and 280 nm (broken lines) was read and plotted.
Estrogen Receptor-Chromatin Cross-Linking
1649
antibody, H222 (25). In Fig. 2A, all material from RNA
(R), DNA (D), and inter-DNA/protein (X) regions are
compared with 5% of the free protein (P). In the absence
of formaldehyde, ER is detected only in the free protein
fraction, as expected. However, when cells are treated
with 0.05% formaldehyde, a trace of ER is detected in
the DNA fraction; and at 0.15% formaldehyde, a significant portion of ER fractionates with the DNA. Approximately 5-10% of the total ER recovered is in the DNA
fraction at 0.15% formaldehyde. At 0.25% formaldehyde, less ER is detected in the DNA fraction than at
0.15%. As shown in Fig. 2C, this reduction in ER is a
consequence of poor recovery. Insoluble material (I),
that is, material not released by sonication of detergentlysed nuclei and removed by centrifugation before loading the soluble material (S) onto gradients, contains
.05
RDX
.25
.15
%HCHO
P R D X P R D P R D P
-66kDa
e
,-45
RDX
P R D X P R D P R D P
.05
.15
.25
%HCHO
Fig. 2. Distribution of ER and Actin in Cross-Linked Nuclei
Fractions from CsCI gradients containing RNA (R), DNA (D),
the region between DNA and protein (X), and protein (P),
based upon absorbance profiles in Fig. 1, were pooled and
assayed by Western blot for ER (A) and actin (B). Zero percent
HCHO:R = fractions 1-2, D = 3-7, X = 8-11, P = 12-19.
0.05% HCHO:R = fractions 1-3, D = 4-9, X = 10-11, P =
12-19. 0.15% HCHO:R = fractions 1-4, D = 5-11, P = 1219. 0.25% HCHO:R = fractions 1-6, D = 7-11, P = 12-19.
Pooled fractions were dialyzed against 10 mM Na-HEPES, 1
mM Na-EDTA, 0.5 mM Na-EGTA, pH7.5. R, D, and X were
precipitated with ethanol and digested with nucleases as described in Materials and Methods. All of R, D, and X, and 5%
of P prepared from two T150 cultures (approximately 4-6 x
107 cells) were concentrated by TCA precipitation, fractionated
on a 10% polyacrylamide-SDS gel, and transferred to nitrocellulose for immune blot. Molecular weight standards indicated are bovine albumin (66 kDa) and egg albumin (45 kDa).
C, Five percent of the soluble nuclear lysate/sonicate (S),
reserved before loading gradients, and 100% of the insoluble
pelleted material (I) were assayed for ER by immunoblot as in
A. Residual ER was extracted from the I material by boiling in
SDS loading solution (65 mM Tris-HCI, pH 6.8, 2% SDS, 10%
glycerol, 2 mM j8-mercaptoethanol, 0.03% bromophenol blue).
increasing amounts of ER as formaldehyde cross-linking is increased. Comparison of the ER levels in soluble
(S), or total material before gradient fractionation (using
a shorter exposure than that shown in Fig. 2C), indicates that roughly equal amounts of ER are present in
each preparation. Thus, 0.15% formaldehyde appears
to be optimal for ER-chromatin cross-linking, and was
used in subsequent experiments.
Selectivity of Cross-Linking
As mentioned, our aim is to link closely associated
molecules and avoid extensive cross-linking. To test
the extent of cross-linking within the cell, a portion of
the blot shown in Fig. 2A [around 45 kilodaltons (kDa)]
was probed with a monoclonal antibody against actin
(26). Since actin is a component of the cytoskeleton,
and is not present in nuclei, only trace contaminating
amounts should be detected in nuclear preparations.
Accordingly, a trace of actin is detected in nuclei not
exposed to formaldehyde (Fig. 2B; 0% HCHO). Increasing amounts are found with increasing concentrations
of formaldehyde treatment, presumably via cross-linking to remnants of the plasma membrane or other
structures that pellet with the nuclei. However, actin is
detected only in free protein fractions, and never with
DNA. Therefore, cross-linking does not extend from
chromatin into the cytoplasm to a large degree.
To test the extent of cross-linking within nuclei, a blot
similar to that in Fig. 2A and B was probed with Y12
anti-Sm monoclonal antibody for small nuclear ribonucleoprotein particle (SnRNP) proteins (27). Since the
SnRNP proteins are components of particles used for
processing RNA, it was anticipated that they would
fractionate with RNA (rather than DNA) after crosslinking. Indeed, a portion of SnRNPs are detected in
RNA fractions after cross-linking with 0.25% formaldehyde (data not shown), but not with 0.15% (the optimum concentration for ER-chromatin cross-linking). Unfortunately, however, SnRNPs from untreated cells are
detected in the upper to middle regions of CsCI gradients (Fig. 3B, -HCHO-RNase lanes). As a consequence, free SnRNPs overlap with cross-linked DNA
on gradients used to fractionate formaldehyde-treated
nuclei, and it is impossible to distinguish free SnRNPs
from any that might be cross-linked nonspecifically to
DNA (Fig. 3B, +HCHO-RNase lanes). To separate
SnRNP proteins in free particles from any cross-linked
to chromatin, nuclear lysates were digested with RNase
A before fractionation on CsCI gradients. As shown in
Fig. 3A, the distribution of ER is essentially unaffected
by RNase digestion. On the other hand, the 27 and 28
kDa SnRNP proteins are detected with cross-linked
DNA and free protein regions of the gradient when
RNase is omitted (Fig. 3B, +HCHO-RNase), but fractionate almost exclusively with free protein after RNase
digestion (Fig. 3B, -l-HCHO+RNase). Thus, the SnRNP
proteins do not cross-link with chromatin, indicating
that cross-linking conditions are selective and confirming that the fractionation methods are efficient.
Vol 4 No. 11
MOL ENDO-1990
1650
+ RNase
-RNase
±HCHO
X P
D
P
X
P
D P
k
-66kDa
-29
B
X P
D P
X
P
D
P
Fig. 3. Selectivity of Cross-Linking within Nuclei
MCF-7 cells were treated with 10~8 M E2 for 1 h. Half the
cultures were left untreated, while the other half were exposed
to 0.15% HCHO for 8 min. Nuclei were prepared and lysed in
0.5% Sarkosyl, 1 mM Na-EDTA, 0.5 ITIM Na-EGTA, 10 mM NaHEPES, pH 7.5. Chromatin was sheared, and one preparation
of each + and -HCHO was digested with 100 Mg/ml RNase
A at 37 C for 30 min. The nuclear lysates were fractionated
on CsCI density gradients and assayed by immunoblot, as in
Figs. 1 and 2, for ER (A) and the 27 and 28 kDa SnRNP
proteins (B). X is from the region of -HCHO gradients that
corresponds to cross-linked DNA (D) from +HCHO gradients
(see Fig. 1). X and D represent 50%, while P is 5% of the
material prepared from one T150 culture. Molecular weight
standards indicated are bovine albumin (66 kDa) and carbonic
anhydrase (29 kDa).
with chromatin, since ER is cross-linked to DNA by
formaldehyde (Fig. 4B; compare D fractions ± HCHO).
In other experiments, 10" 6 M ICI was used with similar
results (not shown). Berry and co-workers (34) have
shown that 10~6 M ICI completely antagonizes the
stimulatory effect of 10~8 M E2 on transcription from the
pS2 gene, so this same level of ICI should be more
than enough to antagonize any estrogenic substances
that may remain in the culture medium.
Interestingly, unoccupied receptor is cross-linked to
chromatin by formaldehyde as well (Fig. 4B, 0). Furthermore, no major or reproducible difference has been
detected in the proportion of total ER recovered with
cross-linked DNA regardless of whether it is exposed
to ligand or not, indicating that both ligand occupied
and unoccupied ER associate with chromatin. Presumably, there are differences between ER-chromatin interactions in the presence and absence of ligand, since
unoccupied ER has been shown to be retained more
weakly with nuclei during cell fractionation (15). However, these differences are not detected as differences
in cross-linking efficiency using formaldehyde.
It may be noted that a second band of slightly slower
electrophoretic mobility is detected for ligand-occupied
receptor, in addition to the band seen for unoccupied
ER. The latter is seen both in the presence and absence
of formaldehyde treatment (Fig. 4B, P lanes ± HCHO;
0
+
Effect of Ligand on Cross-Linking
To investigate the effect of ligand on ER-chromatin
interaction, MCF-7 cells were grown in the absence of
phenol red (28,29) and serum was treated with charcoal
dextran to eliminate estrogens from the culture medium.
These cells were treated with estrogen or antiestrogen,
or were untreated. Studies in our laboratory and others
(30,31) show complete occupancy of ER in intact MCF7 cells within 1 h using 10~9 or 10~8 M 17/3-estradiol.
Furthermore, Brown and co-workers (32) have demonstrated that transcriptional enhancement from pS2,
a primary response to estrogen treatment in MCF-7
cells, is nearly maximal after a 1-h treatment with 10~8
M 17/3-estradiol. Therefore, these conditions were used
for estrogen exposure. The antiestrogen ICI 164,384
was used because it behaves as a pure estrogen
antagonist (20, 2 1 , 33). Since ER has a lower affinity
for the antiestrogen than for estradiol and the cells are
less permeable to the antiestrogen, a longer exposure
(2 h) and a higher concentration (10~7 M) were used.
Estrogen-treated, antiestrogen-treated, and untreated
cells were exposed to formaldehyde or were left unexposed. These were fractionated essentially as above
and assayed for the presence of ER in DNA-containing
fractions.
As expected from reports of in vitro and in vivo
studies in other laboratories (4,18,19), ER exposed to
estrogen (E2) or antiestrogen (ICI) appears to interact
ICI
-
E2
+
+
+HCHO
D P D P D P D P D P
'
B
•"••I
**
" -205kDa
-66
D P D P D P D P D P
Fig. 4. Effect of Hormone and Antihormone on ER and RNA
Polymerase II Cross-Linking
MCF-7 cells grown in the absence of known estrogens were
untreated (0), treated with antihormone (ICI; 10"7 M IC1164,384
for 2 h), or treated with hormone (E2; 10~8 M E2 for 1 h) before
formaldehyde exposure (0.15% HCHO for 8 min). Nuclei were
prepared, lysed in 0.5% Sarkosyl, and the DNA sheared.
Nuclear lysates were fractionated on CsCI gradients, as in Fig.
1. Peak absorbing fractions were pooled and dialyzed against
10 mM Tris-HCI, 0.1 mM EDTA, pH 7.5. Material from 50% of
a T150 culture was adjusted to contain 1 mM CaCI2 and 1 mM
MgCI2, and digested with nucleases. Residual material from
these DNA fractions (D), and free protein (P) from the corresponding gradients (10% of a T150 culture), were assayed for
RNA polymerase II (A) and ER (B) by immunoblot. Molecular
weight standards indicated are myosin (205 kDa) and bovine
albumin (66 kDa).
Estrogen Receptor-Chromatin Cross-Linking
and data not shown). Furthermore, ER prepared in the
presence of protease inhibitors (0.2 ITIM phenylmethylsulfonyl fluoride, 0.5 Mg/ml leupeptin, 0.7 Mg/ml pepstatin, and 7.5 HIM EDTA, as recommended by Boehringer
Mannheim Biochemicals, Indianapolis, IN) appears the
same as that prepared in the absence of these inhibitors
(data not shown). Thus, the faster migrating species
(seen in the absence of hormone) is probably not a
proteolytic product of the slower migrating species,
suggesting that a portion of ER may be modified in
response to ligand. Changes in receptor mobility associated with ligand exposure have been noted previously
for ER (35) and progesterone receptor (36, 37) and, in
some cases, have been shown to reflect changes in
receptor phosphorylation.
As mentioned above, ER binds to specific elements
in estrogen responsive genes regardless of whether its
ligand is an estrogen or an antiestrogen (4, 18); however, the ICI-ER complex does not activate transcription
from these genes (20, 34). Presumably, RNA polymerase II, the enzyme responsible for this transcription, will
be less active in cells exposed to the antiestrogen. To
test this hypothesis, the top portion of the blot shown
in Fig. 4B was probed with antibodies against the large
subunit of RNA polymerase II (38). As shown in Fig.
4A, no major difference is detected in the proportion of
RNA polymerase II recovered with cross-linked DNA,
regardless
of
hormone/antihormone
treatment
(+HCHO, D lanes). Although these results seem to
refute our hypothesis, it is likely that hormone and
antihormone exposures were too short (1-2 h) to elicit
detectable changes in RNA polymerase ll-chromatin
association since the fraction of genes affected by ER
at this time may be too small to make a noticeable
difference.
Ultraviolet (UV) Cross-Linking
UV light was used as an alternative cross-linking agent
in an attempt to confirm the results obtained with
formaldehyde. Unlike formaldehyde, UV cross-linking
requires intimate protein-DNA contact and produces
few protein-protein linkages. Gilmour and Lis (39) used
UV cross-linking to determine the distribution of RNA
polymerase II on heat shock protein genes in Drosophila
tissue culture cells, so similar methods were used with
MCF-7 cells. However, whereas Drosophila cells were
maintained in suspension cultures, MCF-7 cells grow
more vigorously when attached to surfaces. Therefore,
MCF-7 cells were seeded directly onto the dishes used
for UV treatment. In addition, the light source used for
UV treatment had only 30% the output intensity of that
used by Gilmour and Lis. To compensate for this difference, the serum content of the MCF-7 culture medium
was reduced to 0.2% before UV exposure. It was
determined that medium containing 0.2% serum has an
absorbance at the lamp output wavelength (254 nm)
30% that of medium containing the usual 5% serum.
Since Gilmour and Lis used medium containing 12%
serum, it is likely that the MCF-7 cells in the present
1651
study received higher doses of UV light than the Drosophila cells in the reported work. Nonetheless,
whereas Gilmour and Lis (39) reported less than 10%
cross-linking of RNA polymerase II to Drosophila DNA,
we found that less than 1 % of RNA polymerase II was
cross-linked to DNA in MCF-7 cells {vs. 20-50% using
HCHO), and ER was rarely detectable (data not shown).
Furthermore, relatively long UV exposure was necessary (15-30 min vs. 8 min for HCHO), which could allow
time for molecular rearrangement. Longer UV treatment
times resulted in noticeable protein breakdown.
DISCUSSION
Several studies have shown that the bulk of ER, and
indeed most steroid and thyroid hormone receptors,
are localized in the cell nucleus even in the absence of
hormone. Therefore, it is important to determine
whether these unliganded receptors are present on
DNA or if hormone is needed to enhance receptor-DNA
association. Studies on ER association with DNA in
vitro have given conflicting results. While some have
shown that hormone is not required for DNA association
(17), other in vitro studies as well as in vivo footprinting
assays indicate that little, if any, receptor binds to DNA
in the absence of hormone (4,11,18).
We have taken a different approach to investigate
the issue of receptor-DNA association by using formaldehyde to cross-link ER and chromatin in intact cells.
Our method has the advantage that covalent crosslinking is achieved within minutes, under conditions that
should allow little redistribution of macromolecules.
Since formaldehyde treatment can lead to extensive
cross-linking, we determined the minimum formaldehyde concentration needed to yield readily detectable
ER-chromatin linkage and tested selectivity of crosslinking by assaying for the presence of proteins in DNA
fractions which are not known to bind DNA. 0.15%
formaldehyde was found to be optimal. At this concentration, 5 to 10% of ER recovered from cells is crosslinked to chromatin; whereas actin and SnRNP proteins
are not. These results demonstrate that cross-linking is
not extensive, and that our fractionation techniques are
efficient.
We have demonstrated that the presence of hormone
or antihormone has little effect on the magnitude of ERchromatin cross-linking. It is not surprising that estrogen-ER and antiestrogen-ER complexes cross-link with
chromatin, since several published reports have shown
that both complexes bind to the same DNA elements
(4, 18, 19). Indeed, some antiestrogens enhance transcription from EREs when administered alone (20, 21).
On the other hand, the exact intracellular localization of
unoccupied ER is unknown. Although it has been reported that the unoccupied ER is localized predominantly in the nucleus (12-16), its association with specific nuclear structures has not been investigated. Our
results with formaldehyde cross-linking indicate that
MOL ENDO-1990
1652
both ligand-occupied and unoccupied ER associate
with chromatin. To confirm and extend our findings, we
attempted to use UV light as a cross-linking agent.
Whereas formaldehyde produces protein-protein and
protein-DNA linkages via a methylene bridge, UV links
proteins directly to DNA and a more intimate proteinDNA association can be assumed. However, using
techniques similar to published reports (39), we found
that UV cross-linking of ER to chromatin in intact MCF7 cells was too inefficient for practical application (see
Results). Therefore, we can not say whether the association between ER and DNA is direct or indirect. We
note that Kumar and Chambon (4) report inefficient UV
cross-linking of ER to the consensus, perfectly palindromic ERE sequence in vitro, even though a bromodeoxyuridine substituted DNA sequence was used to
enhance cross-linking. Thus, UV cross-linking is inefficient under what seem to be optimal conditions, and
may not be suitable for the study of a very rare protein
in vivo.
Further characterization of covalently cross-linked
ER-chromatin complexes should allow discrimination
between specific and nonspecific DNA interactions. For
example, identification of specific regions of estrogenresponsive genes, such as that for pS2, that cross-link
with ER should be possible. Using this approach, Gilmour and Lis (39) identified regions of Drosophila heat
shock protein genes associated with RNA polymerase
II after UV cross-linking, and Solomon et al. (23) identified regions from one of the same genes which interact
with histone H4 and RNA polymerase II after formaldehyde cross-linking. In addition, formaldehyde crosslinking may provide the means to identify other proteins
associated with ER-DNA complexes. Since transcriptional activation seems to involve interaction between
ER and other transcription factors, it is of interest to
examine the effect of ligand on the pattern of proteins
that copurify with ER after cross-linking.
MATERIALS AND METHODS
Cultivation and Hormone Treatment of MCF-7 Cells
MCF-7 human breast cancer cells, originally obtained from the
Michigan Cancer Foundation, were maintained in Eagle's minimal essential medium (MEM) with Hank's salts (GIBCO, Grand
Island, NY) containing phenol red and supplemented with 10
ITIM HEPES (pH 7.6), 5% charcoal dextran treated calf serum
(CCS), 100 U penicillin/ml, 0.1 mg streptomycin/ml, 25 ^g
gentamycin/ml, 6 ng bovine insulin/ml, and 3.75 ng hydrocortisone/ml (MEM+5%CDCS) at 37 C. Cells were transferred to
the same medium, with (Figs. 1-3) or without phenol red (Fig.
4), in 150 cm2 Corning (T150) flasks approximately 1 week
before an experiment. The medium was replaced on two or
three occassions, one of which was 15-20 h before an experiment. Cells were grown to near confluency, or approximately
2-4 x 107cells/T150 flask.
For hormone or antihormone treatments, the medium was
removed and E2 (10"8 M) or ICI 164,384 (ICI, 10" 6 or 10"7 M)
was added. After mixing, the medium was returned to the
flask, and incubation at 37 C was continued for 1 (E2 treated)
or 2 h (ICI treated).
Vol 4 No. 11
Formaldehyde Cross-Linking and Fractionation of Cells
Procedures for formaldehyde treatment, preparation of nuclei,
and release of nuclear contents were adapted from methods
described by Solomon et al. (23). The culture medium was
removed, a freshly prepared solution containing 1 1 % formaldehyde (from a 37% formaldehyde, 10-15% methanol stock
solution; Fisher Scientific), 0.1 M NaCI, 1 mM Na-EDTA, 0.5
nriM Na-EGTA, 50 mM Na-HEPES, pH 7.5, was added to the
medium- to a final formaldehyde concentration of 0.05 to
0.25%, and the mixture was returned to the flask. After an 8min incubation at 37 C, the medium was removed and the
cells were washed twice with 10 ml PBS (10 mM sodium
phosphate, pH 7.4, 0.15 M NaCI). Cells were scraped from the
flasks into 10 ml PBS, and the flasks were rinsed with an
additional 10 ml PBS. Cells were collected at 225 x g for 5
min, resuspended in PBS (5 ml/T150), and repelleted. Washed
cells were lysed by resuspending in 0.25% Triton X-100, 10
mM Na-EDTA, 0.5 mM Na-EGTA, 10 mM Na-HEPES, pH 7.5
(5 ml/T150), and incubating 10 min on ice. Nuclei were collected at 2000 x g for 10 min, and were washed in 0.2 M
NaCI, 1 mM Na-EDTA, 0.5 mM Na-EGTA, 10 mM Na-HEPES,
pH 7.5 (5 ml/T150), with a 10-min incubation on ice before
repelleting. Washed nuclei were lysed in 0.5% Sarkosyl, 1 mM
Na-EDTA, 0.5 mM Na-EGTA, 10 mM Na-HEPES, pH 7.5 (0.5
ml/T150). The nuclear lysate was either sonicated, using a
Branson model 200 sonifier microtip at power setting 5 for
two 30-sec bursts separated by 30-sec cooling on ice; or
sheared, using two passages through a 20 gauge and eight
passages through a 26 gauge needle. Insoluble material was
removed from sonicated nuclei by centrifugation at 12,000 x
g for 10 min. The supernatant was layered over block gradients
(39) consisting of 3.8 ml 1.04 g CsCI/ml, 3.0 ml 0.67 g CsCI/
ml, and 2.5 ml 0.40 g CsCI/ml in 0.5% Sarkosyl, 1 mM NaEDTA, 0.5 mM Na-EGTA, 10 mM Na-HEPES, pH 7.5 (1 ml or
2xT150/gradient). Gradients were spun at 25,000 rpm, 20 C,
in a Beckman SW41 rotor for 45-48 h. Six tenth ml fractions
were collected from the bottoms of tubes, and their absorbance was read at 260 and 280 nm.
UV Cross-Linking
Procedures for UV cross-linking were adapted from Gilmour
and Lis (39). MCF-7 cells were seeded onto 85 or 135-mm
plastic petri plates (Falcon, Oxnard, CA) in MEM+5%CDCS
containing 0.2% NaHCO3, and were grown to near confluency
at 37 C in a CO2 atmosphere. Cells were rinsed twice with
MEM containing 5% or 0.2% CDCS, but lacking phenol red,
and were incubated with 10~8 M E2 in the same medium at 37
C for 1 h. These cultures were placed over ice on an orbital
platform shaker at 4 C, and shaken to keep the medium
uniformly chilled. The lids were removed from the plates, and
the cells were irradiated at 254 nm with two 15 W germicidal
lamps. Depth of the medium was approximately 3 mm, and
light intensity was 1.2 W/cm2 (1.2 x 10" ergs/cm2 x sec) at
the surface.
Preparation and Fractionation of Proteins
Fractions corresponding to regions of the CsCI gradient absorbance profile were pooled and dialzyed against either 1 mM
Na-EDTA, 0.5 mM Na-EGTA, 10 mM Na-HEPES, pH 7.5 (Figs.
1 and 2), or 10 mM Tris-HCI, 0.1 mM Na-EDTA, pH 7.5 (Figs.
3 and 4), with three changes of buffer. The protein content of
free protein fractions was determined using BCA protein assay
reagent according to the manufacturer's instructions (Pierce,
Rockford, IL) to ensure that similar amounts were recovered
for each preparation. Generally, twice as much free protein is
recovered with nuclei from formaldehyde-treated cells because
cytoplasmic proteins cross-link to nuclei or to structures that
pellet with nuclei (Fig. 2B and data not shown); however,
hormone or antihormone treatment did not affect the amount
of protein recovered.
1653
Estrogen Receptor-Chromatin Cross-Linking
Dilute protein-nucleic acid material (Fig. 2) was concentrated
by ethanol precipitation, using sodium dodecyl sulfate (SDS)
added to 1%, sodium acetate added to 0.3 M, and 4 vol
ethanol (all based upon the original volume), at - 2 0 C overnight. Precipitates were pelleted at 12,000 x g for 45 min and
rinsed twice with 70% ethanol at - 2 0 C. Nucleic acids were
degraded as described by Blanco et al. (40), using 50 ^g
DNase I (Worthington, Freehold, NJ, protease-free) and 750 U
micrococcal nuclease (Worthington)/ml in 10 mM Tris-HCI, pH
7.5, 1 mM CaCI2, 1 mM MgCI2, 0.1 mM phenylmethylsulfonyl
fluoride for 30 min at 37 C.
Residual material from nuclease treated fractions, as well
as protein from the tops of gradients, were precipitated by
adding trichloroacetic acid (TCA) to 10% and incubating 1-2
h on ice. TCA precipitates were pelleted at 12,000 x g, rinsed
twice with ice-cold ethanol-ether (1:1), and air dried. Pellets
were resuspended and boiled in SDS loading solution (see
legend to Fig. 2) for 15-20 min, and fractionated by SDSpolyacrylamide gel electrophoresis essentially as described by
Laemmli (41). Gel patterns were transferred to nitrocellulose
(Gelman BioTrace NT) in 25 mM Tris, 192 mM glycine, 20%
methanol, 0.1% SDS at 220 mA (40 V) overnight. Protein
patterns were visualized by staining the nitrocellulose with
0.1% amido black in 25% methanol, 7.5% acetic acid for a
few minutes, followed by rinsing with deionized distilled water.
Immunoblotting
Rat antihuman ER monoclonal antibody, H222 (25), was a gift
from G. Greene (University of Chicago, Chicago, IL). Rabbit
anW-Drosophila RNA polymerase II (38), affinity purified using
the carboxy-terminal repeat sequences from the yeast enzyme
large subunit, was a gift from A. Greenleaf (Duke University,
Durham, NC). Mouse antiactin monoclonal antibody was purchased from ICN (clone C4;26). Y12 anti-Sm monoclonal antibody against SnRNP proteins (27) was a gift from J. Steitz
(Yale University, New Haven, CT).
Immune blots were essentially as described by Greene et
al. (42). All treatments were at room temperature with agitation. To saturate the nitrocellulose with protein, blots were
treated with 3% Carnation instant nonfat dry milk in TBS (50
mM Tris-HCI, pH 7.5, 150 mM NaCI, 1 mM NaN3) for 45 min.
Tween 20 was added to 0.05% and the treatment was continued for an additional 45 min. Blocked membranes were incubated with primary antibody for 1-2 h (H222 at 4 Mg/ml, antiRNA polymerase II at 0.25 ng/m\, and antiactin or anti-Sm at
1:1000 dilutions). ER and actin blots were incubated with
secondary antibodies, rabbit antirat immunoglobulin G or rabbit
antimouse immunoglobulin G (Zymed), at 1 /xg/ml for 1-2 h.
Finally, all blots were incubated with 125l-protein A (New England Nuclear, Boston, MA; low specific activity) at 0.5 ^l/ml for
2 h. Antibody and protein A incubations were in 0.2% Tween
2 0 , 1 % milk in TBS, and blots were washed with 0.05% Tween
20 in TBS after each incubation. Kodak X-Omat AR5 film was
exposed to the radiolabeled blots using Dupont Cronex intensifying screens.
Acknowledgments
We thank Geoffrey Greene, Arno Greenleaf, and Joan Steitz
for providing ER, RNA polymerase II, and SnRNP antibodies,
and Alan Wakeling and ICI Pharmaceuticals for providing the
antiestrogen.
Received July 2,1990. Revision received August 22,1990.
Accepted August 22,1990.
Address requests for reprints to: Dr. Benita S. Katzenellenbogen, Department of Physiology and Biophysics, University
of Illinois, 524 Burrill Hall, 407 South Goodwin Avenue; Urbana,
Illinois 61801.
*This work was supported by NIH Grants CA-18119 and
5T32-HD-7028. Portions were presented at the 72nd Annual
Meeting of The Endocrine Society, Atlanta, GA, June 1990
(Abstract 856).
REFERENCES
1. Yamamoto KR 1985 Steroid receptor regulated transcription of specific genes and gene networks. Annu Rev Genet
19:209-252
2. Evans RM 1988 The steroid and thyroid hormone receptor
superfamily. Science 240:889-895
3. Beato M 1989 Gene regulation by steroid hormones. Cell
56:335-344
4. Kumar V, Chambon P 1988 The estrogen receptor binds
tightly to its responsive element as a ligand-induced homodimer. Cell 55:145-156
5. Tsai SY, Carstedt-Duke J, Weigel NL, Dahlman K, Gustafsson J-A, Tsai M-J, O'Malley BW 1988 Molecular interactions of steroid hormone receptor with its enhancer
element: evidence for receptor dimer formation. Cell
55:361-369
6. Strahle U, Schmid W, Schutz G 1988 Synergistic action
of the glucocorticoid receptor with transcription factors.
EMBO J 7:3389-3395
7. Schule R, Miller M, Kaltschmidt C, Renkawitz R 1988
Many transcription factors interact synergistically with
steroid receptors. Science 242:1418-1420
8. Ankenbauer W, Strahle U, Schutz G 1988 Synergistic
action of glucocorticoid and estradiol responsive elements. Proc Natl Acad Sci USA 85:7526-7530
9. Tsai SY, Tsai M-J, O'Malley BW 1989 Cooperative binding
of steroid hormone receptors contributes to transcriptional synergism at target enhancer elements. Cell
57:443-448
10. Cordingly MG, Riege AT, Hager GL1987 Steroid-dependent interaction of transcription factors with the inducible
promoter of mouse mammary tumor virus in vivo. Cell
48:261-270
11. Wijnholds J, Philipsen JNJ, AB G 1988 Tissue-specific
and steroid-dependent interaction of transcription factors
with the oestrogen-inducible apo VLDLII promoter in vivo.
EMBO J 7:2757-2763
12. King WJ, Greene GL 1984 Monoclonal antibodies localize
oestrogen receptor in the nuclei of target cells. Nature
307:745-749
13. Welshons WV, Lieberman ME, Gorski J 1984 Nuclear
localization of unoccupied oestrogen receptors. Nature
307:747-749
14. Welsons WV, Krummel BM, Gorski J 1985 Nuclear localization of unoccupied receptors for glucocorticoids, estrogens, and progesterone in GH3 cells. Endocrinology
117:2140-2147
15. Welshons WV, Cormier EM, Wolf MF, Williams PO, Jordan
VC 1988 Estrogen receptor distribution in enucleated
breast cancer cell lines. Endocrinology 122:2379-2386
16. Picard D, Kumar V, Chambon P, Yamamoto K1990 Signal
transduction by steroid hormones: nuclear localization is
differentially regulated in estrogen and glucocorticoid receptors. Cell Regul 1:291-299
17. Lees JA, Fawell SE, Parker MG 1989 Identification of two
transactivation domains in mouse oestrogen receptor.
Nucleic Acids Res 17:5477-5488
18. Martinez E, Wahli W 1989 Cooperative binding of estrogen receptor to imperfect estrogen-responsive DNA elements correlates with their synergistic hormone-dependent enhancer activity. EMBO J 8:3781-3791
19. Webster NJG, Green S, Jin JR, Chambon P 1988 The
hormone-binding domains of the estrogen and glucocorticoid receptors contain an inducible transcription activation function. Cell 54:199-207
20. Weaver CA, Springer PA, Katzenellenbogen BS 1988
Vol 4 No. 11
MOL ENDO-1990
1654
21.
22.
23.
24.
25.
26.
27.
28.
29.
30.
Regulation of pS2 gene expression by affinity labeling and
reversibly binding estrogens and antiestrogens: comparison of effects on the native gene and on pS2-chloramphenicol acetyltransferase fusion genes transfected into
MCF-7 human breast cancer cells. Mol Endocrinol 2:936945
Thompson EW, Katz D, Shima TB, Wakeling AE, Lippman
ME, Dickson RB 1989 ICI164,384, a pure antagonist of
estrogen-stimulated MCF-7 cell proliferation and invasiveness. Cancer Res 49:6929-6934
Meyer M-E, Gronemeyer H, Turcotte B, Bocquel M-T,
Tasset D, Chambon P 1989 Steroid hormone receptors
compete for factors that mediate their enhancer function.
Cell 57:433-442
Solomon MJ, Larsen PL, Varshavsky A 1988 Mapping
protein-DNA interactions in vivo with formaldehyde: evidence that histone H4 is retained on a highly transcribed
gene. Cell 53:937-947
Schouten JP 1985 Hybridization selection of covalent
nucleic acid-protein complexes. 2. Cross-linking of proteins to specific E. coli mRNAs and DNA sequences by
formaldehyde treatment of intact cells. J Biol Chem
260:9929-9935
Greene GL, Sobel NB, King WJ, Jensen EV 1984 Immunochemical studies of estrogen receptors. J Steroid
Biochem 20:51-56
Lessard JL1988 Two monoclonal antibodies to actin: one
muscle selective and one generally reactive. Cell Motil
Cytoskel 10:349-362
Lerner EA, Lerner MR, Janeway CA Jr, Steitz JA 1981
Monoclonal antibodies to nucleic acid-containing cellular
constituents: probes for molecular biology and autoimmune disease. Proc Natl Acad Sci USA 78:2737-2741
Berthois Y, Katzenellenbogen JA, Katzenellenbogen BS
1986 Phenol red in tissue culture media is a weak estrogen: implications concerning the study of estrogen-responsive cells in culture. Proc Natl Acad Sci USA
83:2496-2500
Bindal RD, Katzenellenbogen JA 1988 Bis(4-hydroxyphenyl) [2-(phenoxysulfonyl) phenyl] methane: isolation
and structure elucidation of a novel estrogen from commercial preparations of phenol red (phenolsulfonphthalin).
J Med Chem 31:1978-1983
Eckert RL, Katzenellenbogen BS 1982 Effects of estro-
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
gens and antiestrogens on estrogen receptor dynamics
and the induction of progesterone receptor in MCF-7
human breast cance rcells. Cancer Res 42:139-144
Horwitz KB, McGuire WL 1978 Nuclear mechanisms of
estrogen action: effects of estradiol and anti-estrogens
on estrogen receptors and nuclear receptor processing.
J Biol Chem 253:8185-8191
Brown AMC, Jeltsch J-M, Roberts M, Chambon P 1984
Activation of pS2 gene transcription is a primary response
to estrogen in the human breast cancer cell line MCF-7.
Proc Natl Acad Sci USA 81:6344-6348.
Wakeling AE, Bowler J 1987 Steroidal pure antiestrogens.
J Endocrinol 112:R7-R10
Berry M, Nunez A-M, Chambon P 1989 Estrogen-responsive element of the human pS2 gene is an imperfectly
palindromic sequence. Proc Natl Acad Sci USA 86:12181222
Golding TS, Korach KS 1988 Nuclear estrogen receptor
molecular heterogeneity in the mouse uterus. Proc Natl
Acad Sci USA 85:69-73
Wei LL, Sheridan PL, Krett NL, Francis MD, Toft DO,
Edwards DP, Horwitz KB 1987 Immunologic analysis of
human breast cancer progesterone receptors. 2. Structure, phosphorylation, and processing. Biochemistry
26:6262-6272
Read LD, Snider CE, Miller JS, Greene GL, Katzenellenbogen BS 1988 Ligand-modulated regulation of progesterone receptor messenger ribonucleic acid and protein in
human breast cancer cell lines. Mol Endocrinol 2:263-271
Greenleaf AL, Bautz EKF 1975 RNA polymerase B from
Drosophila melanogaster larvae. Purification and partial
characterization. Eur J Biochem 60:169-179
Gilmour DS, Lis JT 1985 In vivo interactions of RNA
polymerase II with genes of Drosophila melanogaster. Mol
Cell Biol 5:2009-2018
Blanco J, Kimura H, Mueller GC 1987 A method for
detecting protein-DNA interactions at sites of chromatin
replication. Anal Biochem 163:537-545
Laemmli UK 1970 Cleavage of structural proteins during
the assembly of the head of bacteriophage T4. Nature
227:680-685
Greene GL, Harris K, Bova R, Kinders T, Moore B, Nolan
C1988 Purification of T47D human progesterone receptor
and immunohcemical characterization with monoclonal
antibodies. Mol Endocrinol 2:714-726