THE JOURNAL OF BIOLOGICAL CHEMISTRY 0 1992 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 267, No. 9, Issue of March 25, pp. 5842-5846, 1992 Printed in U. S. A.

Effect of Replacement of “Zinc Finger” Zinc on Estrogen Receptor DNA Interactions*

(Received for publication, October 11,1991)

Paul F. Predki and Bibudhendra SarkarS From the Department of Biochemistry Research, Hospital for Sick Children, Toronto, Ontario M5G 1x8 and the Department of Biochemistv, University of Toronto, Toronto, Ontario M5S IA8, Canada

Exposure of bovine estrogen receptor to the metal chelators EDTA and 1,lO-phenanthroline results in a loss of nonspecific DNA binding, presumably because of the removal of “zinc finger” zinc. Nonspecific DNA binding, as measured by a DNA-cellulose binding as- say, can be restored by dialysis of the aporeceptor against buffer containing zinc, cadmium, and cobalt but not with buffer containing copper or nickel. More detailed studies were carried out using a bacterially expressed polypeptide encompassing the DNA binding domain of the human estrogen receptor. Apopolypep- tide fails to bind DNA specifically, as measured by mobility shift assay using a consensus estrogen re- sponse element hexamer containing oligonucleotide, but DNA binding was restored by dialysis of the apo- polypeptide against buffer containing zinc, cadmium, and cobalt but not with buffer containing copper or nickel. Dissociation constants of zinc- and cadmium- reconstituted polypeptide for the estrogen response element hexamer (66 and 48 nM, respectively) are vir- tually indistinguishable from native polypeptide (& = 48 nM) whereas cobalt-reconstituted polypeptide has a lower affinity (& = 720 nM). However, native, zinc-, cadmium-, and cobalt-reconstituted polypeptides gave identical results in a methylation interference assay. Competition experiments with zinc and copper or nickel suggest that copper and nickel are able to bind to zinc finger residues but do so nonproductively. The relative affinities copper > cadmium > zinc > cobalt > nickel for the polypeptide were determined by a zinc blot competition assay. The ability of cadmium and cobalt to substitute for zinc in the zinc fingers demon- strates a structural “flexibility” in the DNA binding domain as each of these metals has slightly different ionic radii. On the other hand, subtle differences in DNA binding affinity and/or specificity could exist, which may not be detectable here. Also, the ability of metals to substitute for zinc in the DNA binding domain suggests that metal substitution in these zinc fingers in vivo may be of relevance to the toxicity and/or carcin- ogenicity of some of these metals.

The steroid hormone receptor superfamily is a group of cytoplasmic receptors which act as transcriptional enhancer proteins, binding specifically to short DNA sequences (hor- mone response elements) and controlling the transcription of

* This research was supported by the Medical Research Council of Canada. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

3 To whom all correspondence should be addressed.

a number of genes (1). NMR solution structure determina- tions for both estrogen and glucocorticoid receptor DNA binding domain polypeptides (2,3) as well as a glucocorticoid receptor DNA binding domain crystal structure (4) have been done. These estrogen and glucocorticoid receptor domain structures are highly related, and both are similarly stabilized by two zinc atoms each coordinated to 4 cysteine residues. Mutagenesis experiments have demonstrated that specific DNA contacts appear to be mediated by residues at the base of the amino terminus of the first finger and the region immediately after this (5, 6). The second finger appears to have regions responsible for dimerization of the DNA binding domain and the orientation of the first finger (5, 7). In vitro studies with glucocorticoid receptor DNA binding domain polypeptide mutants also support and elaborate on this work

The effects of a number of metals on steroid hormone receptors have been reported. A number of metals, for in- stance, have been observed under certain conditions to inhibit steroid binding to the cognate receptors, probably because of the involvement of thiols in steroid binding. Low concentra- tions of zinc have been reported to inhibit lZ5I-T3 binding to rc-erbA p proteins although zinc had no significant effect on lZ5I-T3 binding to rc-erbA LY under identical conditions (10). Although high concentrations of zinc have been found to have no effect on steroid binding to the glucocorticoid receptor an inhibitory effect of cadmium exposure on dexamethasone binding was observed (11). In another study of receptor-metal interactions, the ability of in vitro copper exposure to unpu- rified estrogen receptor at 37 “C but not at 4 “C to stimulate estradiol binding to the receptor has led to the proposal that copper may be a significant mediator of estradiol action in vivo (12). Additionally, immobilized metal ions have been used to demonstrate metal ion-specific binding domains on the surface of the estrogen receptor (13). The specific sites of interaction of these metal ions, however, remains unknown.

Detailed studies of the interactions of different metals with the hormone receptor metal binding sites within the DNA binding domain have been lacking. We report here an in vitro study of the effects of zinc replacement with cadmium, cobalt, copper, and nickel on the DNA binding properties of calf estrogen receptor and a bacterially expressed human estrogen receptor DNA binding domain polypeptide. The ability of different metals to substitute for zinc gives information on the structural contribution of metal toward DNA binding and identifies metal interactions that may be of relevance to metal toxicity and carcinogenicity.

( 8 , ~ .

EXPERIMENTAL PROCEDURES

Materi~k-[2,4,6,7-~H]Estradiol (specific activity 110 Ci/mmol) was purchased from Amersham Corp.; unlabeled estradiol was ob- tained from Sigma. Metal salts were from Aldrich Chemical CO. The

5842

Effect of Zinc Replacement in Zinc Finger 5843 estrogen receptor DNA binding domain expression vector pET31 HE81 was a kind gift from P. Chambon (Institut de Chimie Biolo- gique, Strasburg, France). Molecular biological supplies were from Pharmacia LKB Biotechnology Inc. Nitrocellulose filters, 45 pm, were obtained from HBI (New York). Oligonucleotides were synthe- sized by the Pharmacia/Hospital for Sick Children Biotechnology Service Centre and purified by C-18 SepPak and polyacrylamide gel electrophoresis before use.

Receptor Transformation-Calf uteri obtained immediately after slaughter were stripped of fat, cut into small pieces, and rinsed in homogenization buffer (40 mM Tris, 10% glycerol, 1 mM DTT,' 1 mM phenylmethylsulfonyl fluoride, pH 7.4) a t 4 "C. Homogenization buffer was added at a ratio of 1:1 (w/w), and this was homogenized 10 times in 15-5 bursts in a Waring Blender. The homogenate was ultracentrifuged for 90 min at 150,000 X g. The supernatant was divided into 5-ml aliquots and frozen at -70 "C or used immediately. The protein concentration was determined to be 16 mg/ml as meas- ured by the method of Lowry et al. (14).

The cytosol prepared as described above was made 10 nM in [3H] estradiol and agitated gently at 0 "C for 2 h. KC1 was added to a concentration of 0.4 M and the solution incubated at 28 "C for 45 min. Free and loosely bound [3H]estradiol was removed by incubation with washed Norit A charcoal (0.05%) or dextran-coated charcoal (0.05%). Receptor concentration was determined by isotopic dilution.

DNA Binding Domain Polypeptide-The 113-amino acid estrogen receptor DNA binding domain polypeptide encoded in the plasmid pET31 HE81 is under the control of the T7 RNA polymerase (15). Similar to the purification of a glucocorticoid receptor DNA binding domain polypeptide (16) the plasmid was transformed (17) into the Escherichia coli strain BL21 (DE3) pLysS, which was grown to logarithmic phase in ZB media (Asm = 0.7-0.9), and expression of the polypeptide was induced with 1 mM isopropyl 1-thio-/3-D-galactopy- ranoside. Three h after induction the cells were collected and resus- pended in -10 volumes of lysis buffer (50 mM Tris, pH 7.5, 1 mM EDTA, 10% glycerol, 5 mM DTT, 500 mM NaC1) and lysed by freezing to -70 "C and thawing at 0 "C. Sodium deoxycholate was added to a concentration of 0.05%, and the mixture was centrifuged at 60,000 X g for 60 min at 4 "C. The supernatant was brought to 0.2% polyeth- yleneimine and centrifuged at 35,000 X g for 30 min at 4 "C. The supernatant was dialyzed against MSA buffer (12 mM HEPES, 4 mM Tris, 60 mM KCl, 12% glycerol, 1% DTT, pH 7.9) and then fraction- ated on a DNA-cellulose column (1 X 5 cm) preequilibrated in the same buffer. The partially purified polypeptide was eluted by increas- ing the concentration of KC1 to 400 mM. A 30% ammonium sulfate precipitate was prepared, resuspended in 20 mM sodium phosphate (1 mM DTT) buffer, pH 7.3, fractionated on a CM-Bio-Gel A column (1 X 10 cm), and eluted with a 1 M NaC1, 20 mM phosphate (1 mM DTT), pH 7.6, gradient.

Apoproteins-Aporeceptor was prepared by dialyzing transformed cytosol against homogenization buffer made 10 mM in 1,lO-phenan- throline overnight a t 4 "C. For metal replacement studies aporeceptor was dialyzed against binding buffer made 10 PM in cadmium, cobalt, copper, nickel, or zinc at 4 "C for 6 h. Receptor not exposed to 1,lO- phenanthroline was dialyzed against buffer with 10 PM zinc as a positive control (thus defining 100% binding to DNA-cellulose). Apo- receptor was dialyzed against binding buffer without added metal for a negative control. Fractionation of the cytosol containing [3H]estra- diol on a Sephacryl S-300 column and scintillation counting of the eluted fractions showed no difference between aporeceptor treated with 10 PM metals or unexposed to metals, demonstrating that under these conditions the metal exposure has no apparent effect on estra- diol binding.

Apopolypeptide was prepared by dialysis of polypeptide against 50 mM citric acid, 10 mM phenanthroline at 4 "C for 3 h. For replacement studies apopolypeptide was dialyzed at 4 "C for 3 h against MSA buffer with the indicated concentration of metal.

DNA Binding Assays-Nonspecific DNA binding was determined by binding the receptor to DNA-cellulose. Transformed receptor in binding buffer (80 mM KCl, 40 mM Tris, pH 7.4, 10% glycerol, 1 mM DTT) was incubated with 10-20 mg of DNA-cellulose at room tem- perature for 30 min with gentle shaking. After this the DNA-cellulose was washed three times with binding buffer until no detectable radioactivity eluted. Receptor was released from the DNA-cellulose by incubating the mixture in binding buffer made 400 mM in KCl.

The abbreviations used are: DTT, dithiothreitol; ERE, estrogen response element; HEPES, 4-(2-hydroxyethyl)-l-piperazineethane- sulfonic acid.

Radioactivity in the aqueous phase was determined by scintillation counting. This procedure was also followed using cellulose in place of DNA-cellulose as a control, and counts from this (which were negli- gible) were subtracted from the DNA-cellulose counts.

Specific binding of the polypeptide to an ERE was measured using a mobility shift assay (18) with a low ionic strength polyacrylamide gel using a double-stranded "P-labeled synthetic oligonucleotide con- taining an ERE consensus hexamer (CTGAGATGACCTGCAGCT). Gels were exposed to x-ray film for varying amounts of time to ensure linearity of response. To measure dissociation constants, bound and free DNA levels were determined by densitometer scanning (LKB Ultroscan XL) of the x-ray film exposed to the gel.

Methylation interference assays were performed essentially as described (19) using dimethyl sulfate-methylated ERE consensus hexamer in the mobility shift assay. DNA was run on a 20% sequenc- ing gel.

Dissociation Constants-Before determination of dissociation con- stants, polypeptide was dialyzed against MSA buffer with 0.1 mM EDTA for 3 h to remove loosely bound metal and then dialyzed against MSA buffer for 3 h. Concentration of active polypeptide was determined by saturation binding to the ERE hexamer oligonucleo- tide. Dissociation constants were determined by double-reciprocal

TABLE I Reconstitution of aporeceptor DNA-cellulose binding by dialysis

against 10 p~ metals The method was as described under "Experimental Procedures."

Receptor not exposed to 1,lO-phenanthroline was dialyzed against buffer with zinc as a positive control (thus defining 100% binding to DNA-cellulose). Aporeceptor was dialyzed against binding buffer without added metal for a negative control (apoprotein).

Metal % Bound

Control (zinc) 100.0 f 4.3 Apoprotein 2.9 f 0.7 Cadmium 94.3 * 5.5 Cobalt 96.0 f 7.1 Copper 0.7 f 0.3 Nickel 2.6 -t 1.2 Zinc 98.5 f 4.7

1 + bound

FIG. 1. Mobility shift assay of native (A) , apo ( B ) , and zinc- reconstituted (C) polypeptides. The methods are as described under "Experimental Procedures."

I Cd Co Cu Ni Zn

L 2

FIG. 2. Mobility shift assay of cadmium-, cobalt-, copper-, nickel-, and zinc-reconstituted polypeptides, as indicated. The methods are as described under "Experimental Procedures."

5844 Effect of Zinc Replacement in Zinc Finger

Ilr

FIG. 3. Dissociation constants for native and zinc-, cadmium-, and co- balt-reconstituted polypeptides with ERE hesamer-containing oligonu- cleotide as indicated. The methods are as described under “Experimental Pro- cedures.” 20

I /r

I O

0 0 I 2

I /A

TABLE I1 Dissociation conatants of polypeptide binding to ERE hexamer

sequence as determined by double-reciprocal plot analysis of mobility shift assay results

The method was as described under “ExDerimental Procedures.” Polypeptide form K d

nM

Native 48 Zinc 66 Cadmium 48 Cobalt 720

plots, in which r = [bound DNA]/[total DNA] and A = [free poly- peptide], where concentrations are expressed in molarity.

Zinc Blots-Zinc blots were performed essentially as described by Makowski et al. (20). Proteins were separated electrophoretically on a 15% sodium dodecyl sulfate-polyacrylamide gel and transferred electrophoretically to a nitrocellulose membrane. This was washed three times for 15 rnin in wash buffer (100 mM Tris, 50 mM NaCl, 5 mM CaCIZ, 1 mM DTT), once for 1 min in wash buffer without DTT, and for 30 min in the same buffer with 500 pCi/liter (- 4 X IO-’ M) 65ZnC1z (under an argon atmosphere) with or without other metals present in varying molar ratios. This was washed for 1 min in wash buffer without DTT and twice for 10 min in wash buffer with DTT and then dried. The dried membrane was exposed to x-ray film (Kodak XAR) at room temperature for 2-3 days and developed. Blots were exposed to x-ray film for varying amounts of time to ensure linearity of response.

RESULTS

Estrogen Receptor Reconstitution-It has been reported pre- viously (21) that exposure of bovine estrogen receptor to the metal chelators EDTA and 1,lO-phenanthroline results in a concentration-dependent loss of the nonspecific (DNA-cellu- lose) DNA binding property of the estrogen receptor, probably because of chelation of receptor zinc. We used a 12-h dialysis a t 4 “C against buffer with 5 mM 1,lO-phenanthroline to generate aporeceptor. This treatment reduced levels of non- specific DNA binding to a near zero level. Attempts to recon- stitute the DNA binding properties of the receptor by the

Ilr

20

llr

IO

0 3 4 0 I 2 3

I x 10-8) l / A ( x 10’)

direct addition of zinc or other metals were unsuccessful. It was possible, however, to reconstitute binding by dialysis of the aporeceptor against buffer containing 10 I . ~ M zinc (Table I). Dialysis against cadmium- and cobalt-containing buffers was also capable of reconstituting nonspecific DNA binding of the aporeceptor although similar effects were not observed with copper and nickel (Table I). The inability of copper or nickel to restore DNA binding could be because of a lack of or improper binding at the metal site of the DNA binding domain or could be mediated indirectly by effects of these metals a t other sites in the receptor. To differentiate these possibilities experiments were performed with an estrogen receptor DNA binding domain polypeptide.

DNA Binding Domain Polypeptide Recomtitutwn-Synthe- sis of the DNA binding domain polypeptide in the T7 expres- sion system was induced with isopropyl l-thio-fi-D-galacto- pyranoside, and the polypeptide was purified as described. Apopolypeptide was generated by dialysis of the polypeptide against citric acid/l,lO-phenanthroline. When apopolypep- tide was dialyzed against buffer without metal no specific binding, as measured by a mobility shift assay with a consen- sus ERE hexamer-containing oligonucleotide, could be de- tected (Fig. l). Specific DNA binding was, however, restored by dialysis against buffer containing 10 p~ zinc, cadmium, or cobalt but not with buffer containing copper or nickel as shown in Fig. 2. Dissociation constants for DNA binding domain polypeptide and zinc, cadmium, and cobalt reconsti- tuted apopolypeptide with the ERE consensus hexanucleo- tide-containing oligonucleotide were determined from mobil- ity shift results using a double-reciprocal binding plot (Fig. 3). Native DNA binding domain polypeptide and zinc- and cadmium-reconstituted polypeptide all had very similar affin- ities for the ERE hexamer whereas cobalt-reconstituted poly- peptide had a decreased affinity compared with native poly- peptide (Table 11). Methylation interference assays with the ERE hexameric sequence and native and zinc-, cadmium-, and cobalt-reconstituted polypeptides were virtually identical

Effect of Zinc Replacement in Zinc Finger

FIG. 4. Methylation interference assays of native, zinc-, cadmium-, and cobalt-reconstituted polypep- tides with ERE hexamer-containing oligonucleotide. The method is as de- scribed under “Experimental Proce- dures.” In each case bound (E) and free ( F ) DNA lanes are indicated, and the ERE hexameric sequence is denoted in bold. The specific guanine residue in the ERE hexamer which is “required” for specific binding is indicated by an arrow.

A B C D

-4

Native B F

Zn Cd co B F B F B F

t- bound

- free

FIG. 5. Mobility shift assay of zinc + nickel (C) and zinc + copper ( D ) competition experiments. Nickel ( A ) and copper (E) controls are also shown. The methods are as described under “Exper- imental Procedures.”

(Fig. 4). All demonstrated specific interactions only with the guanine of the ERE hexamer (TGACCT) as expected.

Competition Experiments-Todetermine whether or not copper or nickel interacts with the polypeptide, the ability of these metals to affect reconstitution of the apopolypeptide by zinc was determined by competition experiments. As a con- trol, native polypeptide or zinc-reconstituted polypeptide was dialyzed against buffer containing copper or nickel a t 4 “C for 3 h. Mobility shift assays revealed that this had no effect on specific DNA binding. For the competition experiment apo- polypeptide was dialyzed against buffer containing zinc with either copper or nickel. As a control, copper or nickel was added to zinc-reconstituted polypeptide at the same concen- tration. The ability of apopolypeptide reconstituted in the presence of zinc plus copper to bind to the ERE hexamer was severely diminished compared with reconstitution with zinc alone. A similar, although less dramatic, effect was observed with nickel and zinc (Fig. 5).

Zinc Blot Competition Experiments-To determine the rel- ative affinities of the metals for the polypeptide a zinc blot competition experiment was undertaken. Zinc blots were done in the presence or absence of competing cadmium, cobalt, copper, nickel, or zinc and exposed to x-ray film. After devel- oping the film, zinc binding was quantitated densitometri- cally. Copper competed most effectively with the zinc and

5845

G A

G A

T

G 4 - A

C

C

T

G

c

A

nickel the least. The relative order of affinities was deter- mined to be copper > cadmium > zinc > cobalt > nickel.

DISCUSSION

Primary sequence homology with the Xenopus zinc finger protein transcription factor IIIA (22) and earlier work on exposure of transformed estrogen receptor to metal chelators (21) suggested a zinc dependence for DNA binding. Subse- quent mutagenesis (23), NMR (2, 3), and crystallographic (4) investigations of hormone receptor DNA binding domains have confirmed this, demonstrating that the DNA binding domain of the receptor is a zinc-stabilized structure (described as a “zinc twist” (24) as it bears little structural resemblance to a transcription factor IIIA-type zinc finger). Consistent with these studies we observe that nonspecific DNA binding of the transformed estrogen receptor can be eliminated by exposure of the protein to the metal chelators EDTA and 1,lO-phenanthroline and recovered by dialysis of the protein against buffers containing 10 ~ L M zinc, cadmium, and cobalt but not copper or nickel. Similar results have been obtained with reconstitution of specific binding as measured by a nitrocellulose filter binding assay using a radiolabeled ERE- containing DNA fragment (data not shown). These results are consistent with the expected structural contribution of zinc because zinc, cadmium, and cobalt are known to coordi- nate with tetrahedral geometry, and are all capable of binding to cysteine sulfhydryls, which are the zinc ligands. The ina- bility of aporeceptor DNA binding to be restored by copper or nickel is also not unexpected. Square planar geometries are found to be more common for nickel. Thus, normally tetra- hedrally coordinated zinc fingers may form distorted fingers with nickel, incapable of binding DNA. Copper, however, in the form of copper (I), has a high affinity for sulfhydryl ligands but has less stringent geometric requirements and therefore would not be expected to permit proper folding of the DNA binding domain. On the other hand, copper or nickel may simply fail to bind to the cysteine residues at all. It must also be cautioned that these results do not eliminate the possibility that nickel and copper inhibits DNA binding by some other mechanism such as through indirect interactions with the DNA binding domain.

To characterize more thoroughly the interactions of these metals with the DNA binding domain of the receptor a DNA binding domain polypeptide was expressed in E. coli and

5846 Effect of Zinc Replacement in Zinc Finger

purified. Mobility shift assays with a hexameric ERE consen- sus-containing oligonucleotide demonstrated that the treat- ment of the aporeceptor with citric acid/phenanthroline did indeed destroy specific DNA recognition. In good agreement with the receptor experiments, dialysis of the aporeceptor against buffer containing zinc, cadmium, or cobalt but not copper or nickel restored specific DNA binding. Native poly- peptide dialyzed similarly against copper or nickel retained DNA binding properties, suggesting that these metals do not interact with properly folded metal-stabilized polypeptide in a manner detectable by a mobility shift assay.

The inability of copper or nickel to reconstitute specific DNA binding could result from improper or lack of binding of these metals to the metal binding cysteine residues. To determine whether or not these metals were interacting with the polypeptide two competition assays were employed. Apo- polypeptide was dialyzed against buffer with zinc (10 FM) with or without copper or nickel (10 or 100 p ~ , respectively) and specific DNA binding measured with a mobility shift assay. The ability of apopolypeptide reconstituted in the presence of zinc plus copper to bind to the ERE hexamer was severely diminished compared with reconstitution with zinc alone. A similar, although less dramatic, effect was observed with nickel and zinc. These results suggest that copper and to a lesser extent nickel can compete with zinc for metal-binding residues in the polypeptide.

These results are supported by the zinc blot competition assays. Polypeptide was electrophoresed on a sodium dodecyl sulfate-polyacrylamide gel and transferred electrophoretically to nitrocellulose. Blots were probed with 65Zn in the presence or absence of equimolar cadmium, cobalt, copper, nickel, or zinc and exposed to x-ray film. In the case of nickel and cobalt it was necessary to employ 100-fold molar excesses to observe an effect. 65Zn binding was quantitated by densitometry. Com- petition was observed with all metals although to varying degrees. A relative metal affinity of copper > cadmium > zinc > cobalt > nickel was obtained. The ability of each of these metals to compete with zinc demonstrates that they do inter- act with metal-binding residues in the polypeptide.

Our data suggest that some heavy metals could exert an effect on hormone receptors in vivo via their ability to affect the DNA binding properties of the estrogen receptor as dem- onstrated here in uitro. Low micromolar concentrations of both copper (and to some extent nickel) have, for example, been demonstrated here to have an inhibitory effect on DNA binding of DNA binding domain polypeptide. Also, cobalt can substitute for zinc in the estrogen receptor, but the slight difference in DNA binding affinity could have an effect on gene expression. Although cadmium can replace zinc without a noticeable difference, differences in DNA binding affinity not detectable in our assays may exist which could, perhaps, be manifested in physiological systems. The possibility of cobalt-substituted fingers generating DNA-damaging free radicals while bound to and thus damaging genetic regulatory/ response elements must also be considered. We have recently

observed such activity in vitro,2 and this has been proposed as a mechanism of metal induced carcinogenesis in vivo (25).

Acknowledgment-The DNA binding domain polypeptide expres- sion vector pET31 HE81 was a gift from P. Chambon.

1. 2.

3.

4.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

REFERENCES Evans, R. M. (1988) Science 240,889-895 Hard, T., Kellenbach, E., Boelens, R., Maler, B. A., Dahlman,

K., Freedman, L. P., Carlstedt, D. J., Yamamoto, K. R., Gus- tafsson, J. A., and Kaptein, R. (1990) Science 249 , 157-160

Schwabe, J. W., Neuhaus, D., and Rhodes, D. (1990) Nature 348 , 458-461

Luisi, B. F., Xu, W. X., Otwinowski, Z., Freedman, L. P., Yama- moto, K. R., and Sigler, P. B. (1991) Nature 362,497-505

Danielsen, M., Hinck, L., and Ringold, G. M. (1989) Cell 6 7 ,

Mader, S., Kumar, V., de Verneuil, H., and Chambon, P. (1989) Nature 338 , 271-274

Umesono, K., Giguere, V., Glass, C. K., Rosenfeld, M. G., and Evans, R. M. (1988) Nature 336 , 262-265

Dahlman, W. K., Wright, A., Gustafsson, J. A,, and Carlstedt, D. J. (1991) J. Biol. Chem. 266 , 3107-3112

Zilliacus, J., Dahlman-Wright, K., Wright, A., Gustafsson, J. A., and Carlstedt-Duke, J. (1991) J. Biol. Chem. 266 , 3101-3106

Lu, C., Chan, J . Y., and Walfish, P. G. (1990) Biochem. Int. 21 ,

Simons, S. S., Jr., Chakraborti, P. K., and Cavanaugh, A. H. (1990) J. Biol. Chem. 266, 1938-1945

Fishman, J. H., and Fishman, J. (1988) Bwchem. Biophys. Res. Commun. 162,783-788

Hutchens, T. W., Li, C. M., Sato, Y., and Yip, T. T . (1989) J. Biol. Chem. 264 , 17206-17212

Lowry, 0. H., Rosebrough, N. J., Farr, A. L., and Randall, R. J. (1951) J. Biol. Chem. 193,265-275

Studier, F. W., Rosenberg, A. H., Dunn, J. J., and Dubendorff, J. W. (1990) Methods Enzymol. 186,60-89

Freedman, L. P., Luisi, B. F., Korszun, Z. R., Basavappa, R., Sigler, P. B., and Yamamoto, K. R. (1988) Nature 334, 543- 546

1131-1138

191-198

17. Sambrook, J., Fritsch, E. F., and Maniatis, T. (1989) Molecular Cloning: A Laboratory Manual, pp. 182-184, Cold Spring Har- bor Laboratory, Cold Spring Harbor, NY

18. Fried, M., and Crothers, D. M. (1981) Nucleic Acids Res. 9,6505- 6525

19. Baldwin, A. S. (1989) in Current Protocols in Molecular Biology (Ausubel, F. M., Brent, R., Kingston, R. E., Moore, D. D., Seidman, J. G., Smith, J. A., and Struhl, K., eds) pp. 12.3.1- 12.3.6, John Wiley & Sons, New York

20. Makowski, G. S., Lin, S.-M., Brennan, S. M., Smilowitz, H. M., Hopfer, S. M., and Sunderman, F. W., Jr. (1991) Biol. Trace Element Res. 29,93-109

21. Sabbah, M., Redeuilh, G., Secco, C., and Baulieu, E. E. (1987) J. Biol. Chem. 262,8631-8635

22. Miller, J., McLachlan, A. D., and Klug, A. (1985) EMBO J. 4 ,

23. Severne, Y., Wieland, S., Schaffner, W., and Rusconi, S. (1988)

24. Vallee, B. L., Coleman, J. E., and Auld, D. S. (1991) Proc. Natl.

25. Sunderman, F. W., Jr., and Barber, A. M. (1988) Ann. Clin. Lab.

1609-1614

EMBO J. 7,2503-2508

Acad. Sei. U. S. A. 8 8 , 999-1003

Sci. 18, 267-288

* P. F. Predki and B. Sarkar, unpublished results.