May 23, 2015 - binding (Bogenhagen et al., 1982; Gottesfeld and Bloomer,. 1982; Emerson .... in cells treated for 18 h with 50 Ag/ml aphidicolin, an inhibitor of ...
The EMBO Journal vol.10 no.9 pp.2569-2576, 1991
Glucocorticoids are required for establishment and maintenance of an alteration in chromatin structure: induction leads to a reversible disruption of nucleosomes over an enhancer Andreas Reik, Gunther Schutz2 and A.Francis Stewart1 Institut fOir Zell- und Tumorbiologie, Deutsches Krebsforschungszentrum, Im Neuenheimer Feld 280, D-6900 Heidelberg, FRG
'Present address: European Molecular Biology Laboratory. Postfach 10.2209, D-6900 Heidelberg, FRG 2Corresponding author Communicated by M.Beato
Induction of the rat tyrosine aminotransferase gene (TAT) with glucocorticoid hormones leads to formation of a nuclease hypersensitive site at the hormonedependent enhancer located 2.5 kb upstream of the start site of transcription. This enhancer comprises binding sites for the glucocorticoid receptor and additional transcription factors which are only recognized after hormone administration, as demonstrated by genomic footprinting. We show here that the alteration in chromatin structure occurs very rapidly and is also rapidly reversible after withdrawal of the hormone. At all stages nuclease hypersensitivity at this enhancer parallels the transcriptional activity of the TAT gene indicating that transcriptional stimulation requires ongoing enhancer activity. The enhancer region is nucleosomal before induction and after removal of the hormone but the nucleosomal pattern is disturbed in the presence of the hormone. The rapidity of chromatin changes implies that neither disruption nor reassembly of the nucleosomes is dependent on DNA replication. Addition of the glucocorticoid antagonist RU486 to induced cells has effects similar to washing out the glucocorticoid hormone, showing that maintenance of the hypersensitive site requires the ongoing binding of agonist by glucocorticoid receptor. These results describe an inducible enhancer which is regulated by a dynamic balance between two types of protein -DNA complexes, one of which is nucleosomal. Key words: DNA replication/DNase I hypersensitive site/ in vivo footprinting/RU 486/steroid hormone action
Introduction The DNA of eukaryotic cells is packaged into nucleosomes and higher-order chromatin structures. This property has important implications for essential processes such as transcription and replication. Potentially it confers to the DNA a level of regulation beyond the information encoded by the DNA sequence itself. Investigations to examine this proposition have largely focused on the role of nucleosomes, which are now regarded as universal repressors of transcriptional initiation. Nucleosome reconstitution in vitro has shown that initiation of transcription by RNA polymerases can be inhibited by nucleosomes assembled on (C Oxford University Press
the promoter region (Sergeant et al., 1984; Knezetic and Luse, 1986; Lorch et al., 1987). Some transcription factors can, however, relieve nucleosome-mediated repression when they are added before or at the start of chromatin assembly because they are able to compete with histones for DNA binding (Bogenhagen et al., 1982; Gottesfeld and Bloomer, 1982; Emerson and Felsenfeld, 1984; Workman and Roeder, 1987; Almouzni et al., 1990). Further support for the repressive role of nucleosomes comes from studies with yeast (for review, see Grunstein, 1990). For example, the replication efficiency of a plasmid can be influenced by the positioning of a nucleosome over its replication origin (Simpson, 1990). Transcription of certain yeast genes, located next to transposable elements, is affected by changes in histone gene dosage (Clark-Adams et al., 1988). Reduction of nucleosome density by depleting histone H4 activates several inducible genes including phoS (Kim et al., 1988; Han et al., 1988; Han and Grunstein, 1988). Activation of the phoS gene normally requires phosphate starvation and is accompanied by trans-acting factor-mediated removal of positioned nucleosomes from the promoter region (Almer et al., 1986; Fascher et al., 1990). In this case, it appears that reduction of nucleosome density by lowering H4 levels short-circuits the normal activation pathway by depleting nucleosomes from the phoS promoter and thereby allowing transcription factors access to their binding sites. The glucocorticoid receptor (GR) is a transcription factor that exerts its function, at least in part, by changing chromatin structure. Glucocorticoid induction of a number of genes is accompanied by a perturbation of chromatin structure around the GR binding sites, termed glucocorticoid regulatory elements (GREs), resulting in hypersensitivity of this region to DNase I. The mechanisms, however, leading to hypersensitive site formation and maintenance are not well understood (for a review, see Gross and Garrard 1988). In the case of the mouse mammary tumor virus (MMTV), it has been shown that glucocorticoid treatment of cells containing episomal copies of the MMTV promoter leads to the formation of a DNase I hypersensitive site (HS site) by the removal of a positioned nucleosome located on the GREs (Richard-Foy and Hager, 1987). To gain insights into the mechanisms involved in the establishment and maintenance of HS sites, we study here a case where glucocorticoid treatment of rat hepatoma cells leads to the induction of a HS site in endogenous chromatin (Becker et al., 1984; Nitsch et al., 1990). This site appears 2.5 kb upstream of the tyrosine aminotransferase (TAT) gene and accompanies the activation of a glucocorticoid-dependent enhancer, which increases the transcription rate of the TAT gene. The cis elements responsible for glucocorticoid induction have been located within the HS site by genomic footprinting, in vitro binding studies with purified GR and transfection analyses (Becker et al., 1986; Jantzen et al., 1987). These cis elements include the GRE and also CCAAT
2569
A.Reik, G.Schutz and A.F.Stewart
and CACCC box motifs. Further transfection experiments indicate that factors interacting with these motifs exert synergistic effects with GR on transcription rates (Strahle et al., 1988). We exploit the precision offered by ligand administration to living cells to examine the following questions: (i) Is the hormone required for induction and maintenance of the alterations in the chromatin structure? (ii) What is the nature of the structural alterations? (iii) How do these changes affect transcription of the gene? We find that hypersensitive site formation and transcriptional stimulation are rapid and also rapidly reversible after removal of the hormone. Induction of the gene is accompanied by GR-mediated alterations within the hypersensitive region, while deinduction is characterized by the reassembly of a nucleosomal structure. In this case, nucleosome disruption and reassembly occur too rapidly to require DNA replication.
termination of corticosterone action (Figure 2a, lanes 23-25). Furthermore, the -2.5 HS site remained induced when the washout procedure was performed with media containing corticosterone during washout and subsequent incubation or when dexamethasone was used as inducer (results not shown). No effect of the washout on the other HS sites is observed. The transcription rate of the TAT gene (Figure 2b and c) closely parallels the diminishing hypersensitive reaction and rapidly returns to basal level.
These data argue that the maintenance of the -2.5 HS site and enhancer function require the continual binding of hormone by the glucocorticoid receptor. Since the doubling time of FTO-2B cells in our experiments is 20 h and the chromatin response occurs on more than half of the TAT gene templates (see Figure 4), the rapidity of HS site -
7.
Results
.1
..
Glucocorticoid-induced changes in chromatin structure are rapid and accompanied by transcriptional activation To analyse the changes in TAT chromatin structure in response to the presence or removal of the hormonal inducer, the physiological hormone corticosterone, rather than the synthetic inducer dexamethasone, was used. We reasoned that the tighter binding of dexamethasone by GR (Bell and Jones, 1982) might slow GR release from DNA upon withdrawal of the hormone from the medium. The concentration of corticosterone used for induction was shown by titration to be sufficient for a full hypersensitive reaction (data not shown). The probing strategy of the HS analyses and a diagram of the TAT upstream region are shown in Figure la. A time-course of the formation of the HS site (Figure lb) and of the stimulation of transcription using the same nuclei preparations (Figure lc,d) was performed. As Figure 1 shows, both processes occur with a similar timecourse. The changes upon induction are first seen at 10 min, and are complete after 40 min. The run-on data confirm the kinetics of TAT and PEPCK stimulation by glucocorticoid hormones found in primary hepatocytes by Schmid et al., (1987). Hybridization to pUC18 and the transcription rate of the GAPDH gene, which is not inducible by glucocorticoids, are used as controls. -
The altered chromatin structure is rapidly reversed after hormone withdrawal Besides the need to respond rapidly to a stimulus, it is also important for the cell to possess mechanisms to terminate the response when the stimulus ceases. In comparison with glucocorticoid induction, the mechanisms of deinduction have not been well examined. In particular it is unclear whether the changes in chromatin structure seen upon induction persist after withdrawal of the hormone. Therefore we performed a time course of deinduction by washing the hormone out of the cells and analysed TAT chromatin features and transcription rate. The HS site analysis shows that, following incubation for 30-60 min following washout of the inducer, the -2.5 HS site is no longer seen (Figure 2a). The loss of hypersensitivity results from hormone withdrawal and is not due to programmed
2570
H,.
i
li'
II
.,l
201 llii
111111
4)
IIir
60
11111160 uninducell -i--- -; r i-h n
-
5-
.-
-
:
mF 1.1
*
:'
4.'
{I
i:
O e
'. ,.i-.
46
t7l;
..".
'..
%I V
Fig.
-f&
-=-_-
_.C
., ,,
1. Corticosterone leads to
rapid changes in TAT chromatin representation of the indirect end-labelling strategy used and the hypersensitive sites in the TAT upstream region. P indicates the promoter region. -1.0 HS refers to a HS located 1 kb upstream of the start site of transcription. -2.5 HS refers to the glucocorticoid inducible HS site which is located 2.5 kb upstream. Transcription factor binding sites that lie within the -2.5 HS site implicated by previous studies to be important for enhancer function are schematically represented. SH800 denotes the probe used to indirectly end-label the HindIlI fragment depicted. structure and
transcription
rate. a. Schematic
b. FT02B cells were treated with 10-7 M corticosterone for the times indicated. Nuclei were isolated and subjected to digestion with different concentrations of DNase I (lane 1: 0 units; lanes 2, 5, 8, 11, 14, 17: 30 units; lanes 3, 6, 9, 12, 15, 18: 60 units; lanes 4, 7, 10, 13, 16, 19: 100 units). Subsequently the DNA was purified, cleaved with HindIll, separated on a 1.25% agarose gel and subjected to indirect end-labelling analysis. Hypersensitive sites are marked on the right. c. Run-on analysis of the TAT gene transcription rate made with the same nuclei preparations used in (b). Aliquots of the nuclei were used in run-on transcription assays and the transcribed RNA was analysed by hybridization to DNA fragments immobilized on nitrocellulose membranes. The times of induction are indicated above and the DNA fragments used are TAT: tyrosine aminotransferase, PEPCK: phosphenolpyruvate-carboxykinase, GAPDH: glyceraldehyde
phosphate dehydrogenase, pUC18: vector control for background hybridization. d. The run-on analysis was quantified by scintillation counting and the values for TAT transcription (filled squares) and for GAPDH transcription (open circles) are plotted as c.p.m. versus induction time.
Steroid hormone action and chromatin structure
formation and disappearance demonstrates that replication is not required for either process. Both processes also occur in cells treated for 18 h with 50 Ag/ml aphidicolin, an inhibitor of DNA polymerase a (results not shown). attr r
*
|Idr Li
r,rls|d. mret f
-
w
_n
:q
in, ]
ir
_
!r
r~~~~~~~~~~~~~4
_-
0.
VW
A
-
-
.a
a
a
- -4
_
Fig. 2. The induced change in chromatin structure is rapidly reversed after hormone withdrawal. FTO-2-B cells were induced with 10-7 M corticosterone for 60 min, followed by washout of the hormone and continued incubation of the cells for the times indicated. Thereupon nuclei were harvested and processed for analysis as described in Figure 1. Lanes 23-25 show cells induced in parallel with 10-7 M corticosterone for the duration of the time course (120 min). a. Hypersensitive site analysis as described in Fig. 1. b. Run-on transcription analysis of the same nuclei preparations. c. Graphical representation of (b). Filled squares = TAT transcription rates obtained from corticosterone induced samples; open squares = TAT transcription rates obtained from samples after hormone washout; closed and open circles = GAPDH transcription rates from corticosterone induced and washout samples respectively.
-
-
_-
-
_
mm1-
u
55 J
a
--
ow
.
-
-
-
ft 4m
-
a 9
-
. -
-
-
-
-4
Fig. 3. The antagonist RU486 abrogates the effect of corticosterone on chromatin. Cells were either treated with 10-7 M corticosterone for the times indicated or with 10-6 M RU486 alone or first with 10-7 M corticosterone for 1 h followed by the addition of 10-6 M RU486 for the times indicated. a. Hypersensitive site formation was analysed as described in Figure 1. b. Run-on transcription analysis of the same nuclei. c. Graphical representation of (b). Filled squares = TAT transcription rates obtained after induction with corticosterone; open squares = TAT transcription rates obtained after addition of RU486 to corticosterone induced cells; open triangles = TAT transcription rates obtained after addition of RU486 alone.
The glucocorticoid antagonist RU486 abrogates the hypersensitive reaction induced by corticosterone RU486 is a synthetic antagonist that strongly binds to the glucocorticoid and progesterone receptors (Groyer et al., 1987) and prevents induction of target genes by these receptors, including TAT induction by glucocorticoid hormones (Schmid et al., 1987). We investigated the influence of RU486 on formation and maintenance of the -2.5 HS site, and found that RU486 does not induce a strong hypersensitive reaction under the conditions tested (Figure 3a). Furthermore, when added in a 10-fold excess to cells previously induced with corticosterone, RU486 causes the decline of the -2.5 hypersensitive site. The stimulated level of TAT transcription also declines under these conditions of incubation with RU486 and rapidly returns to basal level (Figure 3b,c). As expected, RU486 alone does not significantly stimulate transcription. These results support the conclusion drawn from the washout experiment (Figure 2) that the maintenance of the -2.5 HS site and enhancer activity require ongoing binding of agonist by the glucocorticoid receptor. Furthermore they indicate that RU486 is not able to sustain these events even when they have been established by corticosterone. The transcription rate of the PEPCK gene also diminishes upon hormone washout (Figure 2b) and RU486 competition (Figure 3b) indicating that our observations are not unique to the TAT gene.
Susceptibility to restriction enzyme cleavage delineates the boundaries of the hypersensitive region In order to examine the chromatin structure of the TAT GRE region before and after induction in more detail, we probed this region with restriction enzymes. The frequency with which most restriction enzymes cut in nuclei is a measure of how accessible their respective recognition sites are and therefore provides an accurate way to map the borders of the hypersensitive region. The effect of glucocorticoid induction on accessibility of a variety of restriction enzyme cleavage sites in the region of the GRE is shown in Figure 4b. Figure 4a shows an example of these results. The cleavage data fall into three classes. First, all restriction sites that show greatly enhanced cleavage after induction are located between -2640 and -2310. Second, restriction sites around -2680, -2500 and -2310 show high levels of cleavage before induction, and some of these sites show little or no increase after induction. The occurrence of local regions of relative sensitivity to restriction enzymes in uninduced chromatin separated by 180 and 190 bp suggest that this chromatin region is comprised of positioned nucleosomes whose linkers display a high frequency of restriction cleavage. This places one nucleosome between -2680 and -2500 and the next between -2500 and -2310. These co-ordinates are compatible with a recent mapping of micrococcal nuclease cleavages in the same region of chromatin by indirect end-labelling at low resolution (Carr and Richard-Foy, 1990). Third, flanking sites outside the hypersensitive region at -2951, -2735 and -2252 show little or no significant increases in cleavage after induction. Hence the chromatin response observed here may involve the disruption of two nucleosomes, and the dominant element of this enhancer, the GRE located at -2509 to -2495 (Jantzen et al., 1987), would lie on the 3' end of the nucleosomal core or in the central linker region.
2571
A.Reik, G.Schutz and A.F.Stewart
*:.{ ..l. l. 4h.
.!ii i;
,IF
I
-
4h
wo v
::iX..i
RuAh
(9h. .'
____
K.
"h
Ni.
Lri
-R.rh
vt
:' e .
h
Rui
RU u h; _3 h _E [
I
t;
:'4,
!; i.-1"I
:.flt
I'm
Fig. 4. Cleavage susceptibility of chromatin to restriction enzymes delineates the boundaries of the hypersensitive site. Isolated nuclei from uninduced cells and from cells induced with 10-6 M dexamethasone were treated with various restriction enzymes. The DNA was purified, digested with HindIll and BamHI and the extent of restriction cleavage in chromatin assessed by indirect end-labelling using either EB670 (panel a and not shown) or BX400 (not shown). a. Example showing the result of an XbaI digest of isolated nuclei from uninduced and induced cells. The bands arise from cleavage at XbaI sites at -3091 and -2562. b. Summary of the results obtained with the various restriction enzymes employed. Autoradiographs were scanned and the frequency of cutting in nuclei from induced cells (filled squares) and in nuclei from uninduced cells (open circles) was plotted against the position of the restriction site. Each symbol denotes one recognition site and the differences between the frequencies of cutting before and after induction are visualized by shading of the area between the two values. The restriction enzymes were: Hinfl (- 1937, -2683), RsaI (-2172, -2508), MboI (-2252, -2349), AspI (-2312), DdeI (-2320, -2616, -2687), Alul (-2331, -2494, -2592), PflMI (-2376), FokI (-2503), StyI (-2530, -2731), XbaI (-2562), ScrFI (-2581,-2667), MboII (-2620), Sau96I (-2634), Bspl286 (-2671), BanlI (-2735), BglII (-2951), EcoRI (-3050). (See Oddos et al., 1989, for the nucleotide sequence.)
The degree of restriction enzyme cleavage observed (Figure 4) demonstrates that more than half of the TAT gene templates participate in the glucocorticoid-induced chromatin response.
The -2.5 HS site is characterized by a reversible disruption of nucleosomes To establish whether the hypersensitive reaction is caused by a change in nucleosomal structure, we used micrococcal nuclease, an enzyme which preferentially cuts the linker regions between nucleosomes. Isolated nuclei were digested with micrococcal nuclease thereby generating ladders of DNA fragments starting with the 150 bp long mononucleosomal band followed by bands corresponding to nucleosome multimers. Based on the results described in the preceding section, we selected a hybridization probe that is located entirely within the hypersensitive region. With this probe (Figure 5a) it can be seen that the DNA around the GRE is indeed nucleosomal as assessed by micrococcal nuclease cleavage. The nucleosomal organization is disturbed after induction and is rapidly restored after the washout of the hormone or the addition of RU486. Rehybridization of the same filter to a TAT gene fragment located further upstream (Figure 5b) showed no change in the nucleosomal pattern after hormone treatment, indicating that the same amount of DNA was loaded in all lanes and that the change in nucleosomal organization is specific to the -2.5 HS site. Genomic footprinting detects hormone-dependent binding of proteins to the -2.5 HS site We next wished to ascertain whether the factors which bind within the -2.5 HS site in chromatin in the presence of -
2572
.
l....
,
i.
A:. -.- " ....
__
w.
W
W
:1:.,..
:g.
:?::,
r
x-
-40 At
ir.
-_ I,
Fig. 5. Reversible loss of nucleosomes from the hypersensitive region. FTO 2B cells were subjected to different induction and deinduction protocols as indicated, and then nuclei were isolated and treated with 375 U/ml of micrococcal nuclease for 5 min (lanes 2, 4, 6, 8, 10, 12) 50 min (lanes 3, 5, 7, 9, 11, 13) or incubated for 50 min without micrococcal nuclease in the digestion buffer (lane 1). DNA was isolated, electrophoresed through a 1.25% agarose gel and blotted for Southern analysis. Lane 14 contains a hybridization control (total, undegraded, genomic DNA digested with AluI or Hinfl) included to verify that the hybridization signal obtained was specific to the probe employed. The blot was hybridized with probe TF187 (-2495 to -2308; panel a) and then stripped and hybridized with probe XA163 (-9207 to 9044; panel b). The positions of these probes are depicted below each panel along with the location of the relevant restriction sites giving rise to the specific signals obtained in lanes 14. (Alul sites at positions -9043, -9112, -9182 evidently produced fragments too small to give a hybridization signal with probe XA163). Abbreviations used are: un, uninduced; Cort., induction with 10-7 M corticosterone; wo, induction with 10-7 M corticosterone for 1 h, followed by washout and 1 h incubation; RU, treatment with 10-6 M RU486; i.M., internal marker. or
hormone can still be observed upon hormone withdrawal. To pursue this objective, we employed a genomic footprinting procedure making use of Taq polymerase (Figure 6, panel a). This procedure differs somewhat from previously described genomic footprinting procedures employing Taq polymerase (Saluz and Jost, 1989, Mueller and Wold, 1989, Pfeifer et al., 1989) in that we use unlabelled primers to amplify linearly local regions of genomic DNA. After electrophoresis, the DNA is electroblotted and the resulting filter hybridized to a uniformly labelled RNA probe that abuts, but does not overlap, the primer. Thus the specificity of the result relies on filter hybridization and not on the fidelity of primer hybridization during the rounds of Taq polymerase
amplification.
Linear amplification of the DNA around the -2.5 kb hypersensitive site was achieved using this strategy. Previously, hormone-induced changes in DMS reactivity in the region around the GRE have been studied on unamplified genomic DNA by Becker et al., (1986). We observe identical patterns of DMS reactivity and also the same small, but reproducible changes of reactivity that occur upon glucocorticoid induction (Figure 6, lanes 1 and 2). After hormone washout, DMS reactivities return to the uninduced pattern (Figure 6, lane 3). To extend our analysis we performed genomic footprinting using an improved method
Steroid hormone action and chromatin structure a IS
456 an CO W-O
I
'Lll CO %0
-314-6-
-2410-
I reat cells wiLh DvS or DNase I
1Lvse with SDS clean up DNA
-2436-7 C-
_llo
":..... I*
-_
-w.
_s
.....
E
-.2442'01"
3(9 rounds of linear anplificaLion
with chosen 27mer Run products on sequencing gel.
iI
#X_:-.
.
-2456-9 -2469-2480-
-2489/92 -
* -
U NO
**-
qv...f
.
q
blot
Hybnrdize blot w-ith single-stranded probe to indirectly end-label amplified DNA
.,
-
.2519-20 -
El*. U
-T-
'^ 1-
-2529-31-
-2538-
El
_
-._
i'
-2505 -
a
_-1
x,ase iA
-2547-9-
DM5S Fig. 6. Genomic footprinting in vivo indicates hormone-dependent reversible binding of proteins to the TAT -2.5 HS. Cells were treated with DMS DNase I and a linear amplification of the signal from the TAT -2.5 kb region was obtained by the use of Taq polymerase as outlined in a. The primer used corresponds in sequence to the upper strand of the TAT gene from -2736 to -2709. The RNA probe was transcribed from the FokI site (-2489) to the Hinfl site (-2682). b. Genomic footprint. The lanes are from uninduced cells (un), cells induced for 1 h with 10-7 M corticosterone (co) and from cells induced with corticosterone for 1 h followed by a washout and a 1 h incubation (wo). Lanes 1 -3 show DMS reactivity patterns and lanes 4-6 show DNase I cleavage patterns and hormone dependent changes that have been reproducibly observed in at least five experiments, are indicated (open squares = protections; filed squares = enhancements). The location of homologies to known transcription factor binding sites are also indicated on the right. On the left, sequence coordinates are shown.
or
to cleave chromatin in permeabilized cells with DNase I (Stewart et al., 1991). Upon hormone induction, a series
of cleavage alterations is observable throughout the -2.5 HS site. These cleavage changes are compatible with factors interacting with previously identified motifs; however, the precise identity of these factors remains to be established (Grange et al., 1991). Of more interest are the changes observable over the major GRE. This region is flanked by DNase I cleavages at approximately -2510 and -2492 which show no, or little, change upon hormone administration (Figure 6, lanes 4 and 5; in the centre of panel b, cleavage protections and enhancements reproducibly observed in five experiments are noted). Between these two sites, cleavages at approximately -2504, -2500 and -2494 show protections when hormone is present. This pattern of cleavage protection is the same as that previously obtained in vitro in DNase I footprinting experiments using partially purified GR (Jantzen et al., 1987) and supports the previous conclusion (Becker et al., 1986), that GR is bound to this region in vivo upon hormone induction. More importantly, we find that all DMS reactivities and DNase I cleavages return to the uninduced pattern after hormone washout (Figure 6, lanes 3 and 6), indicating that protein binding in this region has rapidly reverted to the preinduced state.
Discussion Mechanistic aspects of glucocorticoid-induced chromatin responses The glucocorticoid enhancer upstream of the TAT gene offers a model to study the relationship between DNA binding by transcription factors and DNA packaging by chromatin proteins. All available evidence argues that the
inducible hypersensitive site studied here is caused by the disruption of nucleosomes in the region. The changes we observe in chromatin occur too rapidly to be dependent on replication and are apparently related simply to the presence of hormone bound by the glucocorticoid receptor. Cleavage of the disrupted region with nucleases indicates that a stretch of -370 bp becomes accessible after induction. This is compatible with the removal of two nucleosomes, which may be defined by restriction enzyme cleavage to lie between -2680 and -2500, and -2500 and -2310. Genomic footprinting (Becker et al., 1986; and Figure 6) and transfection experiments (Jantzen et al., 1987; Strahle et al., 1988) indicate that the activated GR, whose major binding site lies between -2509 and -2495, promotes other transcription factors to bind within the disrupted region. Other cases of inducible hypersensitive site formation have been linked to the removal of positioned nucleosomes. In the MMTV LTR, one nucleosome is removed upon binding of the glucocorticoid receptor to GREs located on the nucleosome, and the transcription factors NF- 1 and TFIID are recruited to establish an active promoter (Cordingley et al., 1987; Richard-Foy and Hager, 1987). In the phoS gene of yeast, stimulation of transcription is linked to the removal of four positioned nucleosomes, mainly mediated by a single transcription factor, whose binding sites are located on the central nucleosome linker, and also on the flanking nucleosome cores (Almer et al., 1986; Fascher et al., 1990). In the case of the glucocorticoid receptor, three mechanisms of nucleosome disruption can be envisaged. Either the receptor alone disrupts the nucleosomes first and thereby facilitates the access of other factors to their binding sites, or nucleosome disruption occurs by concerted action of the receptor and the other factors, or nucleosome disruption is a consequence of processes such as transcription 2573
A.Reik, G.Schutz and A.F.Stewart
(Lee and Garrard, 1991) or acetylation/deacetylation (Bresnik et al., 1990) and activated GR exploits the transiently disrupted state. While we cannot distinguish between these possibilities for the TAT gene, some results obtained with the MMTV LTR support the first model. Cordingley et al. (1987) studied chromatin in isolated nuclei and detected factors binding the NFl and TATA box motifs after, but not before, glucocorticoid induction. However, they could not detect factor binding to the GREs. Subsequently, in vitro footprinting experiments (Briiggemeier et al., 1990; Perlmann et al., 1990) indicated competition for DNA binding between the glucocorticoid receptor and NFI. This is partly attributable to overlap of their respective binding sites. In vitro chromatin reconstitution has shown that the GR can bind the GREs when they are located on a nucleosome (Perlmann and Wrange, 1988) and that binding of GR may make the NFl binding site more accessible (Pina et al., 1990). These observations suggest that GR initiates events by nucleosome destabilization but need not remain bound to DNA for NFI and other factors to bind. This model does not explain, however, certain observations made on the mechanism of hormone responsiveness of the MMTV promoter including one case where NFI is apparently not involved (Cato et al., 1988) and the recent observation of co-operative binding between GR and OTFI on the MMTV promoter (Briiggemeier et al., 1991). In the case of the TAT gene, we find that maintenance of the -2.5 HS site requires the continuous presence of glucocorticoids. This can be most easily explained by postulating that GR has to remain continuously bound to the GRE for companion factors to be bound. Our observations are also compatible with a more elaborate model. This invokes a cycle of nucleosomal destabilization by the activated GR, followed by the establishment of a GR-independent complex that stimulates transcription. For TAT, this complex must have a rapid rate of dissociation and, in the continuous presence of activated GR, the cycle must be rapidly reinitiated. The inducible TAT chromatin response involves the establishment of a transcriptional enhancer 2.5 kb upstream of the transcriptional start site. The MMTV chromatin response involves the establishment of a promoter. It will be interesting to inquire, by hormone washout and RU486 competition experiments, whether the MMTV response is rapidly turned over. Such experiments, coupled to genomic footprinting, may distinguish between the alternative mechanisms proposed for the MMTV response to hormones. The chromatin response studied here is an intermediate step in glucocorticoid-mediated signal transduction. Three mechanisms by which signals are transduced through chromatin to stimulate transcription rapidly have been described (Weih et al., 1990). First, heat shock in yeast and epidermal growth factor in A431 cells appear to stimulate transcription by activation of factors already stably bound to DNA (Sorger and Pelham, 1988; Herrera et al., 1989). Second, heat shock in Drosophila and cyclic AMP stimulation of the rat tyrosine aminotransferase gene are mediated by increasing the DNA binding activity of factors which bind within pre-existing HS sites (Thomas and Elgin, 1988; Weih et al., 1990). Third, GR action as described here involves the activation of a DNA binding activity and the establishment of a HS site. This HS site represents a dynamic balance between a nucleosomal state and an enhancer complex. We reason that this third class of
2574
chromatin response may be a mechanism which ensures that transcriptional stimulation is rapidly down-regulated upon withdrawal of the stimulus.
Replication-independent nucleosome disruption and reassembly DNA in the eukaryotic cell is condensed by its association with histones and other chromatin proteins. It has been reasoned that this packaging results in global transcriptional repression mediated by exclusion of sequence-specific DNA binding proteins and the transcription apparatus from regulatory regions. Support for this proposition has come from nucleosome reconstitution and in vitro transcription experiments which demonstrate that prior deposition of nucleosomes over promoter regions prevents transcriptional initiation and reduces basal levels of transcription (Wasylyk and Chambon, 1979; Sergeant et al., 1984; Knezetic and Luse, 1986; Lorch et al., 1987; Workman and Roeder, 1987; Matsui, 1987; Almouzni et al., 1990; Workman et al., 1991). If regions of the genome are repressed by packaging into nucleosomes, how do sequence-specific DNA binding proteins access their binding sites? This question led to the proposition that nucleosome disruption by replication forks provides an opportunity for sequence-specific DNA binding proteins to compete with nucleosome reassembly for regulatory regions (Brown, 1984). Supporting evidence for chromatin remodelling by replication has come from studies in vitro (Wolffe and Brown, 1986; Gruss et al., 1990), in yeast (Miller and Nasmyth, 1984) and after introduction of DNA into mammalian cells (Ceregini and Yaniv, 1984; Enver et al., 1988). Although a number of cases of
replication-independent transcriptional activation have been reported (e.g. Chiu and Blau, 1984), these studies have not been accompanied by detailed examinations of changes in the relevant chromatin regions. In this study we have examined replication-independent changes in chromatin and our data extend previous observations by indicating that some transacting factors can disrupt nucleosomes in the absence of DNA replication (Zaret and Yamamoto, 1984; RichardFoy and Hager, 1987) and that the disrupted nucleosome can also reassemble in the absence of DNA replication. Replication-independent nucleosome disruption and reassembly also accompany the passage of RNA polymerases (Wu et al., 1979; De Bernardin et al., 1986; Lorch et al., 1987; see Stewart et al., 1990; Ericsson et al., 1990; Pfaffle et al., 1990 for further discussion). In this case recent data imply that the histones do not dissociate from the DNA but remain loosely attached through their N-terminal regions (Nacheva et al., 1989). Similar mechanisms may be involved in the chromatin response observed here. The proximity of
bound, disrupted nucleosomes or, for example, the repetitive destabilization of protein -DNA complexes by the passage of RNA polymerase, may explain the difference between the rapid reversion of GR DNA binding after loss of the agonist observed here in chromatin and the stability of ligandindependent binding observed in vitro (Willmann and Beato, 1986; Schmid et al., 1989; Wrange et al., 1990). It is worth considering that we have only observed the
-2.5 HS response in cells and tissues of hepatic origin. The HS response in endogenous chromatin fails to occur in TAT non-expressing cells (e.g. fibroblasts) that are able to activate, upon glucocorticoid administration, transiently transfected constructs containing the glucocorticoid
Steroid hormone action and chromatin structure
responsive sequences (Nitsch et al., 1990; E.Schmid and A.F.Stewart, unpublished observations). This suggests that the -2.5 HS response requires a priming event, for example, locus activation. Although such a priming event may be dependent on replication-mediated chromatin remodelling, our observations describe a condition where nucleosomes do not constitute a barrier to the access of a transcription factor but participate in a reversible, replication-independent chromatin response.
Materials and methods Plasmids Single copy TAT fragments were cloned into pBluescribe M 13 (Stratagene). The following subclones were used: pSH800 (Saul -HindM, + 700/ + 1500), pTX 250 (TthlllI-XbaI, -2308/-2562), pXA650 (XbaI-ApaI, -9207/-8586), pHF200 (Hinfl-FokI, -2682/-2489), pEB670 (EcoOI09-BamHI, -1973/-1301), pBX400 (BglII-XbaI, -2951/ -2562). The plasmids were linearized at appropriate restriction sites to give the templates for RNA probe synthesis. Cell culture FT02B cells were grown to confluency in DMEM/Ham's F12 (1:1) with 10% fetal calf serum, 2 mM glutamine, 100 u/mi penicillin, 100 pg/ml streptomycin and 10 mM HEPES (pH 7.4). Prior to harvesting cells were incubated overnight in serum-free medium, and 1 h before induction fresh serum-free medium was added. Hormone washout was performed by rinsing the cells five times in prewarmed serum-free medium. Preparation of nuclei, nuclease digestion and run-on analysis Approximately 108 FT02B cells were harvested in the culture media using a Teflon policeman and pelleted. After resuspension in 20 mi of cold NI buffer (15 mM Tris-HCI, pH 7.5, 60 mM KCI, 15 mM NaCI, 300 mM sucrose, 5 mM MgCl2 0.5 mM DTT) an equal volume of the same buffer containing 1% NP40 was added and left on ice for 5 min. Nuclei were pelleted (3000 g, 5 min.) and resuspended in NI buffer. One fifth was removed and added to tubes containing 4 ml of run-on buffer (50 mM HEPES, 5 mM MgCl2, 0.5 mM DTT, 1 pg/ml BSA, 25% glycerol). The nuclei were pelleted again, resuspended in NI or run-on buffer as relevant and repelleted. Nuclei for run-on transcription were resuspended in run-on buffer to give a final volume of 100 1l, frozen in liquid N2 and stored at -70'C. Nuclei for DNase I digestion were resuspended in nuclei buffer to give a final volume of 400 Al, and 100 IL aliquots added to tubes on ice containing 0, 30, 60 or 100 units of DNase I (Worthington). After 10 min, EDTA was added to 20 mM. Digestions of nuclei with micrococcal nuclease (Boehringer) were performed in the same manner except that the final nuclei pellet was resuspended in NI buffer containing 1 mM CaCI2 instead of 5 mM MgCI2. Digestion was performed at room temperature for the times indicated in the legends. Nuclei preparations for restriction enzyme digestion were performed similarly except that the restriction buffer contained: 10 mM Tris-HCI pH 7.4, 50 mM NaCI, 10 mM MgCI2, 0.2 mM EDTA, 0.2 mM EGTA, 0.15 mM spermine, 0.5 mM spermidine and 1 mM 3mercaptoethanol. Typically 200 U of each enzyme were used for S x 106 nuclei and the digestion was performed for 1 h at 370C. Subsequently nuclei were lysed and DNA purified as described (Stewart et al., 1991). The DNA was then digested with HindIII (Figures 1-3) or BamHI and HindlIl (Figure 4). The DNA was electrophoresed through a 1.25% agarose gel, blotted to Pall Biodyne B membrane according to the manufacturer's recommendations and probed as indicated in the figure legends. Run-on transcription was performed as described (Stewart and
Schutz, 1987). Genomic footprinting and probe synthesis Methylation of whole cells was performed by addition of 0.2% DMS to PBS/10 mM HEPES, pH 7.4 and incubation at room temperature for 5 min. The plates were rinsed twice with PBS/10 mM HEPES pH 7.4 and the cells were lysed by the addition of 1% SDS, 50 mM Tris-HCI pH 8.0, 20 mM EDTA. After two phenol extractions, DNA was purified by two precipitations from ammonium acetate. Piperidine cleavage of the DNA was performed as described (Becker and Schuxtz, 1988). DNase I samples for genomic footprinting were obtained by permeabilizing whole cells with 0.2% NP40 or 0.05% lysolecithin as described (Stewart et al., 1991) using 300 U DNase I.
DMS and DNase I treated genomic DNAs were subjected to a linear amplification reaction. In the experiment shown in Figure 6, DNase I cleaved genomic DNA was treated with terminal transferase as described (Tanguay et al., 1990) before being subjected to linear amplification, however similar results were obtained when this
was
omitted. The
amplification reaction
contained, in 100 sl: 30 ytg DNA, 40 ng of a 27mer primer corresponding to the TAT sequence -2736 to -2709, 500 1,M of each dNTP, 40 mM NaCl, 10 mM Tris-HCI (pH 8.9), 5 mM MgCI2, 0.17 mg/ml BSA and 10 U Taq polymerase (Perkin-Elmer). Thirty cycles at 64'C for 1 min, 72°C for 1.7 min and 94°C for 1 min were performed. The reaction was stopped and DNA was purified by phenol -chloroform extraction followed by precipitation from 0.3 M Na-acetate and rinsing of the pellet with 75% ethanol. The DNA was dried, resuspended in loading buffer and denatured by heating to 95'C. Afterwards the DNA was analysed on sequencing gels as described (Becker and Schutz, 1988) except that RNA probes synthesized
with UTP 3000 Ci/mmol were used. The probe synthesis was performed in 10 gl. UTP (Amersham, 250 jtCi) was dried down and resuspended in 40 mM Tris-HCI pH 8.0, 80 mM MgCl2, 2 mM spermidine, 50 mM NaCl, 10 mM DTT, 330 itM each of ATP, GTP, CTP, 15 units of RNasin (Promega) and 1 Ag of linearized template. Then 10 units of T7 or T3 polymerase (Stratagene) were added. Incubation was for 15 min at 37'C followed by DNase digestion of the template by addition of 15 11 containing 50 mM Tris-HCI pH 7.5, 5 mM MgCl2, 1 mM DTT, 5 isg tRNA and 1 U RNase-free DNase (Promega) and further incubation at 37'C for 15 min. The reactions were stopped by addition of hybridization mix and hybridizations were performed according to Church and Gilbert (1984). Probe syntheses for Southern blots were performed similarly except that reaction volumes were twice those cited above and 100 ACi of UTP (800 Ci/mmol) (not dried down) were used.
Acknowledgements We are grateful to E.Schmid for help with the run-on transcription assay and G.Pfeifer for technical suggestions. We thank W.Schmid, C.DeVack, G.Kelsey and M.Nichols for critically reading the manuscript, C.Schneider for secreterial assistance and W.Fleischer for photography. This work was supported by the Deutsche Forschungsgemeinschaft through SFB 229 and the Leibniz Programm, and the Fonds der Chemischen Industrie.
References Almer,A., Rudolph,H., Hinnen,A. and Horz,W. (1986) EMBO J., 5, 2689-2696.
Almouzni,G., Mechali,M. and Wolffe,A.P. (1990) EMBO J., 9, 573-582. Becker,P., Renkawitz,R. and Schutz,G. (1984) EMBO J., 3, 2015-2020. Becker,P.B., Gloss,B., Schmid,W., Strahle,U. and Schutz,G. (1986) Nature, 324, 686-688. Becker,P.B. and Schutz,G. (1988) In Setlow,J.K. (ed.) Genetic Engineering, Principles And Methods. Plenum Press, New York. Vol. 10, pp 1-19. Bell,P.A. and Jones,T.R. (1982) In Agarwal,M.K. (ed.) Hormone Antagonists. Walter de Gruyter & Co., Berlin, New York, 391-405. Bogenhagen,D.F., Wormington,W.M. and Brown,D.D. (1982) Cell, 28, 413-421.
Bresnick,E.H., John,S., Berard,D.S., LeFebvre,P. and Hager,G.L. (1990) Proc. Natl. Acad. Sci. USA, 87, 3977 3981. Brown,D.D. (1984) Cell, 37, 359-365. Bruiggemeier,U., Rogge,L., Winnacker,E.-L. and Beato,M. (1990) EMBO J., 9, 2233-2239. Bruggemeier,U., Kalff,M., Franke,S., Scheidereit,C. and Beato,M. (1991) Cell, 64, 565 572. Carr,K.D. and Richard-Foy,H. (1990) Proc. Natl. Acad. Sci. USA, 87, -
-
9300-9304.
Cato,A.C.B., Skroch,P., Weinmann,J., Butkeraitis,P. and Ponta,H. (1988) EMBOJ., 7, 1403 - 1410. Cereghini,S. and Yaniv,M. (1984) EMBO J., 3, 1243-1253. Chiu,C.-P. and Blau,H.M. (1984) Cell, 37, 879-887. Church,G.M. and Gilbert,W. (1984) Proc. Natl. Acad. Sci. USA, 81, 1991 -1995.
Clark-Adams,C.D., Norris,D., Osley,M.A., Fassler,J.S. and Winston,F. (1988) Genes Dev., 2, 150-159. Cordingley,M.G., Riegel,A.T. and Hager,G.L. (1987) Cell, 48, 261-270. De Bernardin,W., Koller,T. and Sogo,J.M. (1986) J. Mol. Biol., 191, 469 -482.
Emerson,B.M. and Felsenfeld,G. (1984) Proc. Natl. Acad. Sci. USA, 81, 95-99.
2575
A.Reik, G.Schutz and A.F.Stewart Enver,T., Brewer,A.C. and Patient,R.K. (1988) Mol. Cell. Biol., 8, 1301- 1308.
Ericsson,C., Grossbach,U., Bjorkroth,B. and Daneholt,B. (1990) Cell, 60, 73-83.
Fascher,K.-D., Schmitz,J. and Horz,W. (1990) EMBO J., 9, 2523 -2528. Gottesfeld,J. and Bloomer,L.S. (1982) Cell, 28, 781-791. Grange,T., Roux,J., Rigaud,G. and Pictet,R. (1991) Nucleic Acids Res., 19, 131-139. Gross,D.S. and Garrard,W.T. (1988) Annu. Rev. Biochem., 57, 159- 197. Groyer,A., Schweizer-Groyer,G., Cadepond,F., Mariller,M. and Baulieu,E.-E. (1987) Nature, 328, 624-626. Grunstein,M. (1990) Trends Genet., 6, 395-400. Gruss,C., Gutierrez,C., Burhans,W.C., DePamphilis,M.L., Koller,T. and Sogo,J.M. (1990) EMBO J., 9, 2911-2922. Han,M. and Grunstein,M. (1988) Cell, 55, 1137-1145. Han,M., Kim,.U.-J., Kayne,P. and Grunstein,M. (1988) EMBO J., 7, 2221 -2228. Herrera,R.E., Shaw,P.E. and Nordheim,A. (1989) Nature, 340, 68-70. Jantzen,H.-M., Strihle,U., Gloss,B., Stewart,F., Schmid,W., Boshart,M., Miksicek,R. and Schutz,G. (1987) Cell, 49, 29-38. Kim,U.J., Han,M., Kayne,P. and Grunstein,M. (1988) EMBO J., 7, 2211-2219. Knezetic,J.A. and Luse,D.A. (1986) Cell, 45, 95-104. Lee,M.-S. and Garrard,W.T. (1991) EMBO J., 10, 607-615. Lorch,Y., LaPointe,J.W. and Kornberg,R.D. (1987) Cell, 49, 203-2 10. Matsui,T. (1987) Mol. Cell. Biol., 7, 1401-1408. Miller,A.M. and Nasmyth,K.A. (1984) Nature, 312, 247-251. Mueller,P.R. and Wold,B. (1989) Science, 246, 780-786. Nacheva,G.A., Guschin,D.Y., Preobrazhenskaya,O.V., Karpov,V.L., Ebralidse,K.K. and Mirzabekov,A.D. (1989) Cell, 58, 27-36. Nitsch,D., Stewart,A.F., Boshart,M., Mestril,R., Weih,F. and Schutz,G. (1990) Mol. Cell. Biol., 2, 1044-1051. Oddos,J., Grange,T., Carr,K.D., Matthews,B., Roux, J., Richard-Foy,H. and Pictet,R. (1989) Nucleic Acids Res., 17, 8877-8878. Perlmann,T. and Wrange,O. (1988) EMBO J., 7, 3073-3079. Perlmann,T., Eriksson,P. and Wrange,O. (1990) J. Biol. Chem., 265, 17222-17229. Pfaffle,P., Gerlach,V., Bunzel,L. and Jackson,V. (1990) J. Biol. Chem., 265, 16830-16840. Pfeifer,G.P., Steigerwald,S.D., Mueller,P.R., Wold,B. and Riggs,A.D. (1989) Science, 246, 810-813. Pina,B., Bruggmeier,U. and Beato,M. (1990) Cell, 60, 719-731. Richard-Foy,H. and Hager,G.L. (1987) EMBO J., 6, 2321-2328. Saluz,H. and Jost,J.-P. (1989) Proc. Natl. Acad. Sci. USA, 86, 2602-2606. Schmid,E., Schmid,W., Jantzen,M., Mayer,D., Jastorff,B. and Schdtz,G. (1987) Eur. J. Biochem., 165, 499-506. Schmid,W., Striihle,U., Schutz,G., Schmitt,J. and Stunnenberg,H. (1989) EMBO J., 8, 2257-2263. Sergeant,A., Bohmann,D., Zentgraf,H., Weiher,H. and Keller.W. (1984) J. Mol. Biol., 180, 577-600. Simpson,R.T. (1990) Nature, 343, 387-389. Sorger,P.K. and Pelham,H.R.B. (1988) Cell, 54, 855-864. Stewart,A.F. and Schdtz,G. (1987) Cell, 50, 1109-1117. Stewart,A.F., Herrera,R.E. and Nordheim,A. (1990) Cell, 60, 141-149. Stewart,A.F., Reik,A. and Schdtz,G. (1991) Nucleic Acids Res., 19, 3157. Straihle,U., Schmid,W. and Schdtz,G. (1988) EMBO J., 7, 3389-3395. Tanguay,R.L.,Pfeifer,G.P. and Riggs,A.D. (1990) Nucleic Acids Res., 18, 5902. Thomas,G.H. and Elgin,S.C.R. (1988) EMBO J., 7, 2191-2201. Wasylyk,B. and Chambon,P. (1979) Eur. J. Biochem., 98, 317-327. Weih,F., Stewart,A.F., Boshart, M., Nitsch,D. and SchUtz,G. (1990) Genes Dev., 4, 1437-1449. Willmann,T. and Beato,M. (1986) Nature, 324, 688-691. Wolffe,A.P. and Brown,D.D. (1986) Cell, 47, 217-227. Workman,J.L. and Roeder,R.G. (1987) Cell, 51, 613-622. Workman,J.L., Taylor,I.C.A. and Kingston,R.E. (1991) Cell, 64, 533-544. Wrange,O., Eriksson,P. and Perlmann,T. (1989) J. Biol. Chem., 264, 5253-5259. Wu,C., Wong,Y.-C. and Elgin,S.C.R. (1979) Cell, 16, 807-814. Zaret,K.S. and Yamamoto,K.R. (1984) Cell, 38, 29-38. Received on 12 April 1991; revised on May 23. 1991
2576
