Department of Biophysics, King's College London, UK. Received 12 ... at position 110 in chicken histone H3 to generate a disulphide bridge across the interface ...
Volume 8 Number 22 1980
Nucleic Acids Research
An examination of models for chromatin transcription
H.J.Gould, G.J.Cowling, N.R.Harborne and J.Allan Department of Biophysics, King's College London, UK
Received 12 September 1980 ABSTRACT
Structural studies have revealed that chromatin is composed of repeating units or nucleosomes having two distinct domains, the nucleosome core and the linker region. The nucleosome core comprises 146 base pairs of DNA wound in one and three quarter turns around an octamer of histones made up of two symmetrical tetramers (1). It may be inferred on topological grounds that this structure must be perturbed during chromatin transcription and replication since the histone core bridges the supercoil which blocks the passage of polymerase along the template and prevents the unwinding of DNA required for enzymatic copying. A number of mechanisms for freeing the DNA template may be envisaged, and one detailed model, based on symmetrical dissociation of the histone tetramers, has been proposed (2). Here we present evidence against such unpairing or indeed any detachment of histones from the octamer during chromatin transcription, and we give reasons for favouring a transcriptional mechanism based upon the separation of the octamer from at least one strand of the DNA.
INTRODUCTION The most explicit model for chromatin transcription is that of Weintraub
et al (2), in which the nucleosome is forced to dissociate along its dyadaxis
the unpaired histone tetramers remaining attached at the extremities of the We have submitted this model to a critical test by introducing a covalent cross-link to prevent the dissociation of the histone tetramers. This can be achieved with minimal disturbance of the native structure, for we have found it possible to oxidise the thiol groups located at position 110 in chicken histone H3 to generate a disulphide bridge across the interface between the two H3 molecules at the dyad axis of the octamer (3). Neither this modification, nor the more drastic cross-linking of histone octamers with Lomant's reagent (4), inhibit the transcription of chromatin
linear DNA segment.
in vitro by E. coli RNA polymerase.
©) IRL Press Limited, 1 Falconberg Court, London W1V 5FG, U.K.
5255
Nucleic Acids Research MATERIALS AND METHODS INTRODUCTION OF COVALENT CROSS-LINKS BETWEEN HISTONES IN SITU
Polynucleosomes were prepared by micrococcal nuclease digestion of erythrocyte nuclei, and a homogeneous fraction of 20 nucleosome-long chains was obtained by sucrose gradient centrifugation. Histones Hi and H5 were
removed by DNA cellulose chromatography (5).
The sulphydryl groups of the
histone H3 pairs in the nucleosomes were catalytically oxidised with copper
phenanthroline (6).
Prior to oxidation, polynucleosomes (A260nm=1-2) were
dialysed into 50 mM triethanolamine, 0.5 mM ethyleneglycolbis-(aminoethylethej tetraacetic acid (EGTA), 0.1 mM phenylmethylsulphonyl fluoride (PMSF), pH 8.0
(TEP) buffer, and then incubated with 0.1 volume of a solution containing 25 mM 1,10 o-phenanthroline (Sigma) and 12.5 mM copper sulphate for 24 hrs. at 40 with constant agitation. The treatment was terminated by making the solution 10 mM in ethylenediaminetetraacetic acid (EDTA) and dialysing out the reagent against 80 mM NaCt, 10 mM Tris-HCt, pH 7.5, 0.1 mM EDTA, 0.1 mM PMSF. Control samples were similarly treated in the absence of copper phenanthroline. For reduction of the disulphide bonds, the oxidised sample was made 1 M in 2-mercaptoethanol and incubated at 40 for 24 hours. Samples were prepared for histone analysis by lyophilisation from the TEP buffer, dissolution in sample buffer without 2-mercaptoethanol and heating at 1000 for 1 min. Electrophoresis was performed in 15% polyacrylamide gel slabs according to Laemmli (7). After electrophoresis, the protein was stained with Coomassie Brilliant Blue. Hi- and H5-depleted polynucleosomes (A260=1-2) were dialysed against 0.1 M borate buffer, 60 mM NaCt, 0.5 mM EDTA and 0.1 mM PMSF, pH 9.0. Dithiobis-
(succinimidylpropionate) (Lomant's reagent) (8), dissolved at 50 mg/ml in dimethylformamide was added to give 1 mg/ml, and the sample was incubated for 3 h. at 250. The reaction was terminated by adding Tris-HCZ buffer (10 mM),
followed by dialysis against 80 mM PMSF, pH 7.5.
NaCa, 10
mM Tris-HCt, 0.1 mM EDTA, 0.1 mM
Histones were analysed by electrophoresis in 5% polyacrylamide
gels according to Weber and Osborn (9). INTRODUCTION OF COVALENT CROSS-LINKS BETWEEN HISTONES BY RECONSTITUTION To extract DNA from polynucleosomes, solutions were made 0.1% in sodium
dodecylsulphate (SDS), and were then extracted twice with phenol and twice with chloroform.
DNA was precipitated from the aqueous phase made 0.3 M in
sodium chloride by the addition of two volumes of ethanol.
After washing the precipitates twice with 70% ethanol, the DNA was dissolved in 2 M NaCZ, 0.5 mM EDTA, 0.1 mM PMSF, 20 mM
5256
Tris-HCt, pH 8.0.
Nucleic Acids Research Core histones were prepared by chromatography on hydroxyapatite as previously described (5). H3 dimerization was essentially complete in samples used(Fig 1. Prior to reconstitution the histones were dialysed against 2 M NaCt, 0.5 mM EDTA, 0.1 mM PMSF, 20 mM Tris-HCi, pH 8.0. The reconstitution procedure was based on that described by Germond and co-workers (10). Equal volumes of DNA and core histones, each at 200 ig/ml were mixed and incubated at 370 according to the following schedule: (1) un-
diluted for 1 hr.; diluted with 0.5 mM EDTA, 0.1 mM PMSF, 20 mM Tris-HCU,pH 8.0, to the varying final NaCi concentrations indicated for (2) 1 hr. at 1.6 M NaCk; (3) 1 hr. at 1.2 M NaCt; (4) 2 hrs. at 0.85 M NaCk; (5) 2 hrs. at 0.75 M NaCt; (6) 2 hrs. at 0.65 M NaCk; (7) 1 hr. at 0.5 M NaCt; (8) 1 hr. at 0.2 M NaCt; and (9) overnight at 0.08 M NaCi.
The reconstitution was carried out
in the presence or absence of 2-mercaptoethanol (10 mM). TRANSCRIPTION OF POLYNUCLEOSOMES
After oxidation or chemical cross-linking of histone octamers in situ or reconstitution, polynucleosomes and controls (unmodified polynucleosomes
and DNA) were transcribed with E. coli RNA polymerase (Sigma Type III). Samples of the native and oxidised polynucleosomes and DNA were additionally
fixed after dialysis into 80 mM NaC2, 10 mM triethanolamine, 0.1 mM EDTA, 0.1 mM PMSF, pH 8.0, by addition of 0.1 volume of a neutralised solution of 10%
v/v formaldehyde and incubation for 24 hrs. at 40 (11).
Formaldehyde was
removed by dialysis against the original buffer, and then against 80 mM NaCk, 10 mM
Tris-fCk,
0.1 mM EDTA, 0.1 mM PMSF, pH 8.0.
To form the initiation complex with RNA polymerase, the template (7
DNA) in a solution containing 80 mM
KCi,
0.5 mM EGTA, 50 mM
Tris-HCi,
ig
6 mM
Mgc2t2' 0.4 mM ATP, 0.4 mM GTP, 0.05 mM UTP, and 50 pCi (5,6- H) UTP (43 Ci/ mmol, Amersham) was mixed with 1 unit of E. coli RNA polymerase in a total This pre-incubation was sufficient to volume of 0.5 ml for 10 min. at 37 obtain maximum synthesis in the subsequent elongation reaction. To initiate polymerisation, CTP (0.4 mM) was added together with rifampicin (10 ig/ml; CIBA) to inhibit re-initiation, and the incubation continued for a further 15 min. to complete the synthesis of RNA chains. the addition of 10
pl of 500
Transcription was terminated by
mM EDTA.
To extract RNA, the reaction mixture was made 0.1% in SDS, 0.1 M
NaCZ,
and 5 pg of E. coli transfer RNA was added as carrier. After phenol/chloroform extraction, the aqueous phase was applied directly to a Sephadex G50
column (0.7 x 45 cm) and the RNA was eluted with 10 mM NaCi, 1 mM EDTA, 0.1% SDS, 10 mM Tris-HCR, pH 7.5, precipitated by the addition of two volumes of
5257
Nucleic Acids Research
Co
rs
N
*
I
ci:.
..
a' = c_
5258
Nucleic Acids Research ethanol, washed with 70% ethanol and dried in vacuo. The RNA size distribution was determined by formamide gel electrophoresis (12). ( H) RNA transcripts and ( C) RNA size markers (2.5xlO dpm, 23S, 16S and 4S RNA isolated from E. coli grown in the presence of ( C)uracil) were lyophilised and dissolved in 10 pl of buffered deionised formamide containing 0.005% bromophenol blue, 20% sucrose. This mixture was heated to 500 and subjected to electrophoresis on 4% polyacrylamide-99% formamide cylindrical gels (0.5 x 6.5 cm). After electrophoresis at 2 mA/ tube, the gels were soaked in water for 1 hr., frozen at -70O and sliced. The slices (1.6 mm) were treated with 0.3 ml of Soluene 350 (Packard), heated 1-2 hrs. at 500, mixed with 3.5 ml Permablend/toluene scintillant (Packard) and counted in a liquid scintillation spectrometer. RESULTS The Hl- and H5-depleted polynucleosomes used as templates in this study Were catalytically oxidised in situ, using copper phenanthroline. Essentially quantitative conversion of histone H3 to the disulphide form results, and
this can be reversed by reduction with 2-mercaptoethanol (Fig. 1A). We established that the structures of the cross-linked and native materials are identical in respect of the characteristic 10 base-pair ladder on DNase I digestion (13) and the sedimentation coefficient (s 20 =11.3S) in the case of nucleosome monomers. Extensive characterisation of oxidised poly-
nucleosomes has also shown that the structure is unperturbed (Cowling et al, to be published). This accords with the previous observations on chromatin reconstituted with core histones containing the oxidised H3 dimer in place
Figure 1 A. Analysis of Histone H3 Monomers and Dimers from Polynucleosomes oxidised In Situ. (1) and (6) histone markers; (2) histones from control polynucleosomes, carried through the same manipulations as the oxidised sample without addition of the catalyst; (3) histones from polynucleosomes oxidised in situ; (4) The same as (3) after reduction by 2-mercaptoethanol; (5) the same as (3) after transcription in vitro. The positions of monomer (M) and dimer (D) in the electrophoresis pattern are indicated. B. Analysis of Histone H3 Monomers and Dimers from Reconstituted Polynucleosomes. Histones from polynucleosomes reconstituted in the absence (1) or presence(2) of 2-mercaptoethanol. Monomer (M) and dimer (D) shown as in A. C. Analysis of Histone Oligomers from Polynucleosomes cross-linked In Situ. (1) Histones cross-linked with Lomant's reagent; (2) Histones from polynucleosomes incubated in the borate buffer without Lomant's reagent. The positions of free (monomeric) core histones (1) and the cross-linked octamer (8) are indicated. 5259
Nucleic Acids Research of monomer (3). To assay for template activity, polynucleosomes (fractionated to yield a narrow size distribution with a weight average of 20 nucleosomes) were
mixed with E. coli RNA polymerase (at a ratio of one polynucleosome chain per enzyme molecule), together with three of the four nucleoside triphos-
phates required for transcription.
After 10 min. shown to be sufficient for
all the enzyme molecules to form initiation complexes with the polynucleosome
templates, rifampicin was added to prevent re-initiation, and RNA synthesis was started by addition of the fourth nucleoside triphosphate and continued
for 15 min. to ensure completion of RNA chains.
Preliminary experiments
showed that the transcription of both native and oxidised polynucleosomes was linear during the first 5 min. of the incubation, but that a significant
amount of RNA was synthesised in the remaining time.
The presence of
rifampicin prevents the preferential transcription of the protein-free linker regions resulting from the premature termination of synthesis when the polymerase encounters a nucleosome followed by re-initiation of RNA synthesis. The template activity of the polynucleosomes can now be assessed both from the yields of RNA and the size distribution.
As a further demonstration
that the nucleohistone complex (and not free DNA) was transcribed, we showed that synthesis was inhibited when the histone was cross-linked to the DNA
with formaldehyde (14). It seems clear that the treatment of formaldehyde does not cause a redistribution of the core histones on the Hl- and H5depleted polynucleosomes (15). The oxidation of the histone H3 pairs did not diminish the template
activity of the polynucleosomes or change the molecular weight distribution of the RNA product (Fig. 2A,C).
Ebr both the oxidised and native templates,
over 90% of the transcripts exhibited chain lengths greater than the average
nucleosome spacing of 200 base pairs.
The mean size of the RNA corresponded
to the length of the DNA in a pentanucleosome, but the distribution was broad
and extended up to the full length (4000 nucleotides) of the DNA. Formaldehyde treatment of both oxidised and native polynucleosome dramatically inhibited the synthesis of high molecular weight RNA (greater than the length of transcripts corresponding to the protein-free linker regions, which would co-migrate with the 4S marker), whereas formaldehyde treatment of the DNA had no such effect (not shown, cf. Ref. 14). Reduction of the disulphide bonds in the oxidised sample (Fig. 1A) did not affect the template activity, the molecular weight distribution of the transcript, or the results obtained after formaldehyde treatment (not shown).
5260
Nucleic Acids Research 2015 108 6 4
2
2015 1086 4
1
2
1
8
6 2 4
2
Il 0
Cl)
0
x
E
E
Q
-o
_l
Cl
3
10
20
30
10
20
30
SLICE NUMBER Fig. 2, Analysis of Molecular Weight Distribution in Polynucleosome Transcripts. Transcripts from (A) untreated polynucleosomes; (B) DNA extracted from polynucleosomes; (C) oxidised polynucleosome (See Fig. lA, track 3); (D) polynucleosomes containing cross-linked octamers (See Fig. lC, track 1). Samples were analysed directly, 0-0 or after treatment with formaldehyde S-0. The positions of the 23S, 16S and 4S RNA markers are indicated. The numbers (N=l,2...20) on the abcissa at the top of the diagram indicate the calculated positions of DNA transcripts corresponding to N nucleosome equivalents, ie. multiples of 200 nucleotides.
5261
Nucleic Acids Research From the yields of RNA (ca. 50 ng/pg DNA) we calculate that about 10%
of the DNA sequence in the polynucleosome templates was transcribed.
Over
90% of the histone H3 was oxidised in these experiments from densitometry of the gels (Fig. 1A), but we could not exclude the presence of a small propor-
tion of unoxidised monomer which, if localised in particular regions of the template, might account for the observed level of transcription. To eliminate this possibility, we reconstituted polynucleosomes using oxidised core histones obtained in the normal isolation procedure (see above), assuming that the minor proportion of unoxidised monomer would be randomly distributed
within the resulting chains.
Reconstitutions were performed in the presence
and absence of 2-mercaptoethanol, resulting in polynucleosomes containing H3 monomers and mainly dimers, respectively (Fig. 1B). The size distribution
of RNA transcribed from these templates (Fig. 3) and from native polynucleosomes or polynucleosomes oxidised in situ (Fig. 2A,C) were indistinguishable.
The high molecular weight transcripts from the latter cannot therefore be ascribed to stretches of un-oxidised nucleosomes.
We have also shown that
incubation in the transcription system (lacking 2-mercaptoethanol) does not
12
:0
o
x
E
10
8
-o I
6 4
2 10
20
30
10 SLICE NUMBER
20
30
Fig 3. Molecular Weight Distribution in Reconstituted Polynucleosome Transcripts. Template prepared in the presence of 2-mercaptoethanol (A); absence of 2-mercaptoethanol (B). The positions of the 23S, 16S and 4S markers are indicated.
5262
Nucleic Acids Research lead to the scission of the disulphide bridge (Fig. 1A).
These results
exclude a transcription mechanism involving the symmetric dissociation of histone tetramers. Next the histones were extensively and randomly cross-linked with the
bifunctional reagent dithiobis (succinimidylpropionate) (8) to yield an octamer which no longer dissociated in sodium dodecylsulphate (Fig. 1C and Ref. 4). This material also served as a template for the synthesis of high molecular weight RNA chains in the same amount as the native polynucleosome control (Fig. 2D). Given the extensive cross-linking in this system, it would seem that we may go further and assert that no histones need to dissociate from the octamer for transcription to proceed. The experiments on formaldehyde-treated chromatin lead to a more positive conclusion: reaction with formaldehyde in our conditions cross-links all the histones to the DNA, as a total absence of free histones in the gel electrophoretic patterns demonstrates (results not shown; Ref. 15), One observes instead protein-containing complexes too large to enter the polyacrylamide gel Formaldehyde is known also to form cross-links
at the origin of migration.
between histones H2A and H2B and of both histones H2A and H2B with histone H4 (16), but the loss of transcriptional activity on formaldehyde treatment may be taken to result from the protein-DNA cross-links, since the more extensive inter-histone bridges introduced with the bifunctional acylating reagent have no such effect (Fig. 1C and 2D).
DISCUSSION
Polynucleosomes containing only core histones provide a simple and
convenient model system for analysing the mechanism of transcription of nucleosome DNA, These templates lack extraneous components (histone Hl, nonhistone proteins and RNA) and are soluble, native in structure, highly active in transcription, and easily manipulated for experimental purposes, as our results show.
The choice of bacterial polymerase is also dictated by
convenience, since it is readily available and more active (by at least an order of magnitude) than eukaryotic polymerase in initiating synthesis of RNA chains on chromatin templates (17). Since we are not concerned with transcriptional specificity here, we have preferred to take advantage of the high polymerase activity. As noted, we observe an average of one RNA tran-
script per chain of template, and about 10 percent of the total DNA is transcribed. There is no reason a priori to suspect that the nucleosome should behave differently towards, say, chicken erythrocyte polymerase II,
5263
Nucleic Acids Research but it should be possible to compare the modes of transcription by the
heterologous and homologous enzymes.
Other workers have employed both
native and reconstituted polynucleosomes containing only core histones and either E. coli or eukaryotic polymerase for transcription studies (14-18-25). Our system affords the particular advantage that the distribution of the nucleosome cores is undisturbed by the method of Hl and H5 depletion (5), since the nucleosome spacing, and even the distribution, with respect to the DNA sequence (or 'phasing'), may be important in transcription.
There have been few previous attempts to arrive at a model for transcription by the approach of selective chemical modification of chromatin
in vitro.
In a recent study somewhat analogous to ours, Wasylyk and Chambon
constructed artificial 'minichromosomes' from SV40 DNA and histone octamers
cross-linked with dimethylsuberimidate (18), The reconstituted material served as a substrate for transcription by E. coli RNA polymerase, albeit only at grossly elevated salt concentrations, reaching an optimum at 0.8 M NaCi. The difficulty of transcribing the DNA in the modified nucleosomes was ascribed to the increased affinity of the octamer for DNA, due to an enhanced basicity of the histones with amidine side chains. In our experiments Lomant's reagent was used to give an unionised product (8), and transcription was
observed in the normal range of ionic strength (Fig. 2D).
The results of the
two types of substitution, taken together, support the view that electro-
static interactions between the histones and DNA stabilise chromatin in an inactive conformation.
The requirement for histones to dissociate from DNA
during transcription is also indicated by the inhibitory effects of formalde-
hyde treatment (Fig. 2 and Ref. 14). The histone cross-linking experiments suggest that the-octamer dissociates from the DNA in toto. If symmetrical histone contacts on both strands of the DNA are envisaged (26), the dissociation must be a cooperative process. It has been argued that transcriptionally active chromatin possesses
its full complement of histones and retains indeed the capacity to form nucleosomes (27-29) with the same periodicity (30) as the bulk of inactive
chromatin. Thus the octamer, partly disengaged from the DNA, evidently remains lodged at or near its original binding site. At least two types of mechanism consistent with both the topological requirements for transcription and the observed behaviour of nucleosomes may be proposed: since histone
octamers are capable of interacting with comparable affinities with single and double-stranded DNA (31), they may in the transcriptionally active state
dissociate completely from one strand and remain attached to the complementary
5264
Nucleic Acids Research They could, alternatively, transfer themselves to a neighbouring site; this could be either a nucleosome, since free octamers have been shown to associate with intact nucleosomes (32,33), or a segment of naked DNA, since (in Hl-depleted chromatin) the histone cores can be readily induced to slide
strand.
along the chain (34) or to exchange with denuded DNA (35).
Further experiments
to test and refine these models are in progress.
ACKNOWLEDGEMENTS
This work was supported by a Medical Research Council Grant G979/1094/SA and a Wellcome Senior Fellowship for J.A. We thank Dr. W.B. Gratzer for valuable criticism and advice.
REFERENCES
1.
2. 3. 4, 5.
6. 7. 8. 9. 10. 11.
12. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23.
24,
McGhee, J. and Felsenfeld, G. (1980), Ann. Rev. Biochem. 49,
1115-1156. Weintraub, H., Worcel, A., and Alberts, B. (1976), Cell 9, 409-417. Camerini-Otero, R.D, and Felsenfeld, G. (1977), Proc. Natl. Acad. Sci. USA 74, 5519-5523. Thomas, J.O. and Kornberg, R.D. (1975), Proc. Natl. Acad. Sci. USA 72, 2626-2630. Allan, J., Staynov, D.Z., and Gould, H. (1980), Proc. Natl. Acad. Sci. USA 77, 885-889. Kobashi, K. (1968), Biochim. Biophys. Acta 158, 239-245. Laemmli, U.K. (1970), Nature 227, 680-685. Lomant, A.J. and Fairbanks, G. (1976), J. Mol. Biol. 104, 243-261. Weber, K. and Osborn, M. (1969), J. Biol. Chem. 244, 4406-4412. Germond, J.-E., Ballard, M., Oudet, P., and Chambon, P. (1976), Nucl. Acids res. 3, 3173-3192. Brutlag, D., Schlehuber, C. and Bonner, J. (1969), Biochemistry 8, 3214-3218. Staynov, D.Z., Pinder, J.C., and Gratzer, W.B. (1972), Nature New Biol. 235, 108-110. Noll, M., (1974), Nucl. Acids Res. 1, 1573-1578. Williamson, P. and Felsenfeld, G., (1978), Biochemistry 17, 5695-5705. Varshavsky, A.J., Sundin, O., and Bohn M. (1979), Cell 16,453-466. Martinson, H.G., True, R., Lau, C.K., and Mehrabian, M. (1979), Biochemistry 18, 1075-1082. Cedar, H., (1975), J. Mol. Biol, 95, 257-269. Wasylyk, B., and Chambon, P., (1980), Eur. J, Biochem. 103, 219-226. Meneguzzi, G., Pignatti, P.F., and Milanesi, G., (1976), Biochem. Biophys. Res. Commun. 72, 626-633. Cremisi, C., Chestier, A., Dauguet, C., and Yaniv, M., (1977), Biochem, Biophys. Res. Commun. 78, 74-82. Mathis, D.J., Oudet, P., Wasylyk, B., and Chambon, P., (1978), Nucl. Acids Res. 5, 3523-3547. Wasylyk, B., and Chambon, P., (1979), Eur. J. Biochem. 98, 317-327. Wasylyk, B., Thevenin, G., Oudet, P., and Chambon, P., (1979), J. Mol. Biol. 128, 411-440. Gariglio, P., Llopis, R., Oudet, P., and Chambon, P., (1979), J. Mol Biol. 131, 75-105. 5265
Nucleic Acids Research 25. 26. 27.
Meneguzzi, G., Chenciner, N., and Milanesi, G., (1979), Nucl Acids Res. 6, 2947-2960. Finch, J.T., Lutter, L.C., Rhodes, D., Brown, R.S., Rushton, B., Levitt, M., and Klug, A., (1977), Nature 269, 29-36. Axel, R., Cedar, H., and Felsenfeld, G., (1975), Biochemistry 14,
2489-2495. 28. 29.
30. 31. 32. 33. 34. 35.
5266
Lacy, E., and Axel, R., (1975), Proc. Natl. Acad. Sci. USA 72, 3978-3982. Kuo. M.T., Sahasrabuddhe, G.G., and Saunders, G.F., (1976), Proc. Natl. Acad. Scio USA 73, 1572-1575. Gottesfeld, J.M., and Melton, D.A., (1978), Nature 273, 317-319. Palter, K.B., and Alberts, B.M., (1979), J. Biol. Chem. 254, 11160-11169. Voordouw, G., and Eisenberg, H., (1978), Nature 273, 446-448. Stein, A. (1979), J. Mol. Biol. 130, 103-134. Spadafora, C., Oudet, O., and Chambon, P., (1979), Eur. J. Biochem. 100, 225-235. Germond, J.E., Bellard, M., Oudet, P., and Chambon, P., (1977), Nucl. Acids Res. 3, 3173-3192.
