Characterization of DNA binding and retinoic acid binding properties ...

Proc. Natl. Acad. Sci. USA Vol. 88, pp. 3559-3563, May 1991

Physiology/Pharmacology

Characterization of DNA binding and retinoic acid binding properties of retinoic acid receptor NA YANG*t, ROLAND SCHLLEt, DAVID J. MANGELSDORFt,

AND

RONALD M. EVANSt

*Department of Chemistry, University of California, San Diego, La Jolla, CA 92093; and tHoward Hughes Medical Institute, The Salk Institute for Biological Studies, La Jolla, CA 92037

Contributed by Ronald M. Evans, December 28, 1990

ABSTRACT High-level expression of the full-length human retinoic acid receptor (RAR) a and the DNA binding domain of the RAR in Escherichia coli was achieved by using a T7 RNA polymerase-directed expression system. After induction, full-length RAR protein was produced at an estimated level of 20% of the total bacterial proteins. Both intact RAR molecules and the DNA binding domain bind to the cognate DNA response element with high specificity in the absence of retinoic acid. However, this binding is enhanced to a great extent upon the addition of eukaryotic cell extracts. The factor responsible for this enhancement is heat-sensitive and forms a complex with RAR that binds to DNA and exhibits a distinct migration pattern in the gel-mobility-shift assay. The interaction site of the factor with RAR is localized in the 70-amino acid DNA binding region of RAR. The hormone binding ability of the RARa protein was assayed by a charcoal absorption assay and the RAR protein was found to bind to retinoic acid with a Kdof 2.1 x 10'10M.

RAR and the factor is localized in the DNA binding domain of RAR. Activation of RAR is known to be triggered by the association with its ligand RA. However, determination of the accurate dissociation constant of RA with RAR from mammalian sources has been hampered because of the interference of the cellular RA binding protein in mammalian tissues (15). In contrast, the RAR produced in E. coli binds RA with high affinity, displaying a Kd of 2.1 x 10-1o M.

All-trans-retinoic acid (RA), a vitamin A derivative, is a biologically active vertebrate morphogen. It has profound effects on cellular differentiation, pattern formation, and embryonic development. RA also has a striking effect on regenerating limbs and has been implicated as a natural morphogen in chicken and frog embryogenesis (refs. 1-5 and references therein). Functions of RA are presumed to be mediated in part by its nuclear receptor proteins that are members of the steroid hormone receptor superfamily. Three species of RA receptor (RAR) cDNAs have been cloned and are referred to as a, ,, and y (refs. 6-10). However, despite the fact that certain genes regulated by RAR have been identified (refs. 11-13 and references therein), the molecular events that occur between RA signaling and the end biologic effects are largely unknown as yet. To understand how RAR interacts with its cognate DNA response sequence and its ligand RA as well as the mechanisms of its regulation, we have expressed the full-length human RARa protein and the DNA binding domain of RAR in Escherichia coli under the control of T7 polymerase (14). After isopropyl ,B-Dthiogalactopyranoside (IPTG) induction, up to 20% of the bacterial proteins is full-length RAR protein. In vitro DNA binding studies show that the bacterially expressed fulllength RAR protein and the 70-amino acid DNA binding domain bind specifically to a 27-mer oligonucleotide defined as the RAR response element in the promoter region of the RAR,B gene (refs. 11 and 13). However, the affinity of this binding is greatly enhanced by a heat-sensitive factor in eukaryotic cells. Upon addition of cellular extracts to the gel-mobility-shift assay, the RAR-DNA complex is further retarded. We conclude that in vivo, RAR protein requires another protein factor for high-affinity DNA binding. We show herein that at least one site of the interaction between

MATERIALS AND METHODS Expression of Human (h) RARa and RARDBD in E. coli. The cloning procedure of full-length hRARa cDNA was described in ref. 16. For expression of the DNA binding domain of RAR, the cDNA sequence of RARa corresponding to amino acids 87-156 (6) was PCR-amplified. An Nco I site and a BamHI site flanking the 5' and 3' ends, respectively, were put in for subsequent cloning into the pET-8c vector (14). The resulting plasmids pET-8cRARDBD or pET-8cRAR for fulllength RAR expression were then transformed into E. coli BL21(DE3)plysS. A single colony was picked and inoculated in LB medium with ampicillin (25 ,ug/ml) and chloramphenicol (25 /Lg/ml). The culture was grown at 370C until cell density corresponding to an OD6w of 0.7 was reached. Isopropyl 83-D-thiogalactopyranoside (IPTG; Boehringer Mannheim) was added to 0.4 mM to induce RAR protein. Induction continued for 3 hr and cells were collected by centrifugation. The cell pellet was then resuspended in 10o of the original volume of the lysis buffer (20 mM Hepes, pH 7.9/60 mM KCI/2 mM dithiothreitol). Cells were lysed by two cycles of freezing/thawing. Glycerol [20%o (vol/vol)] was then added. Centrifugation at 100,000 x g was performed to separate soluble proteins from cell debris. The supernatant was saved and used for subsequent DNA binding and hormone binding analyses. Gel-Mobility-Shift Assay. DNA binding activity was assayed by a gel-retardation assay. 32P-labeled DNA oligonucleotide probes (8 x 104 dpm; 1 ng) were incubated with 1 ,1 of pET-8cRAR- or pET-8cRARDBD-transformed BL21(DE3)plysS extracts in the binding buffer [6 mM KCI/22 mM Hepes, pH 7.9/0.2 mM dithiothreitol/5 mM spermidine/8% glycerol/2% (vol/vol) Ficoll/0. 1% Nonidet P-40]. Poly(dIdC) (1 ,ug) was present in each binding reaction mixture as the nonspecific DNA competitor. Eukaryotic cell extracts were prepared as described in ref. 17. Total protein of cell extracts (5 ,ug) was used in the binding reaction. For heat inactivation, cell extracts were treated at 65°C for 10 min before addition to the binding reaction. For the competition assay, a 25-fold molar excess of unlabeled oligonucleotide was added simultaneously with the radioactive probe. Incubation was carried out on ice for 10 min. Separation of bound

The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.

Abbreviations: RA, retinoic acid; RAR, RA receptor; IPTG, isopropyl ,B-D-thiogalactopyranoside; h, human; ,lRE, /3-responsive element; GR, glucocorticoid receptor.

3559

3560

Physiology/Pharmacology: Yang et al.

probes from free probes was by PAGE on a 5% gel in 0.5 x TBE (44 mM Tris/44 mM boric acid/0.5 mM EDTA, pH 8.0). Electrophoresis was performed at 40C. The gel was preelectrophoresed for 1.5 hr at 200 V. The dried gel was autoradiographed with an intensifying screen at -70'C with Kodak XAR film. RA Binding. The charsorb method (18) was used to determine the hormone binding activity of RAR. A desired amount of 3[H]RA was dried down to evaporate trace of ethanol and mixed thoroughly with 1 ul of RAR containing bacterial lysate and 79 /.l of mock BL21(DE3)plysS lysate in the RA binding buffer (0.12 M KCl/8 mM TrisHCl, pH 7.4/8% glycerol/4 mM dithiothreitol/0.24 mM phenylmethylsulfonyl fluoride). The final volume was 1 ml. To determine nonspecific binding, a 100-fold molar excess of unlabeled RA was mixed with radioactive RA before addition of the lysate. Incubation was carried out at 40C for 16 hr in the dark. Then 500 1.l of 3% (wt/vol) activated charcoal (Sigma) was added at the end of incubation, mixed well, and incubated at 40C for 15 min. Charcoal was activated according to ref. 18 and was then brought to 3% in 0.15 M NaCl/0.1 M Na2HPO4/0.039 M NaH2PO4/0.1% gelatin/0.015 M NaN3, pH 7.03. Charcoalabsorbed free RA was then separated from bound RA by centrifugation at 15,000 x g for 15 min. The supernatant was collected and radioactivity was measured in 20 ml of Ecolume (ICN Biochemicals). RESULTS Expression of hRARa and the DNA Binding Domain of RAR in E. coli. To produce the RAR protein in E. coli, a full-length hRAR coding sequence was inserted into the high-level expression plasmid pET-8c vector (Fig. 1A). As indicated in the diagram, the hRARa sequence starting from the initiation methionine was inserted into the Nco I-BamHI cloning site of the vector. The resulting plasmid allows the production of a full-length nonfusion RAR protein under the regulation of the T7 promotor. Cleavage of the first methionine residue during bacterial synthesis makes this protein product one amino acid shorter than wild-type RAR. This was confirmed by microsequencing. pET-8cRARa plasmid was transformed into E. coli BL21(DE3)plysS, which harbors in its genome the T7 polymerase gene driven by an IPIG-inducible UV5 promotor. After a 3-hr IPTG induction, high-level expression of RAR protein with the predicted molecular mass of 54 kDa was observed (Fig. 1B, lanes 1 and 2). The estimated level of induced RAR protein was about 20% of the total bacterial proteins (Fig. 1B, lane 2). Two freeze/thaw cycles were sufficient to lyse bacteria cells to completion because of the low-level constitutive expression of lysozyme in BL21(DE3)plysS (14). RAR proteins were recovered in the supernatant fraction as shown in Fig. 1B, lane 3, which indicates that full-length RAR protein is highly soluble. The DNA binding domain of RARa was induced at a slightly lower level compared to full-length RAR (data not shown). About 60%o of the RARDBD polypeptide can be recovered in the soluble fraction of the bacterial lysate. DNA Binding Property of the RAR. We have analyzed the in vitro DNA binding activity of the RAR by a gel-mobilityshift assay. The cognate DNA sequence used in the assay is a 27-mer oligonucleotide from the promotor region of the RARj3 gene. This sequence contains a perfect direct repeat of the nucleotide sequence motif GTTCAC and has been defined as a natural RAR response element, which is both necessary and sufficient for mediating RA responsiveness (11, 13). As shown in Fig. 2, RAR expressed in E. coli binds the 8-responsive element (I3RE) to give rise to a retarded band (lane 3). This binding is highly specific for the RAR response element but not for other DNA sequences as shown by the competition experiment. Whereas binding of RAR

Proc. Natl. Acad. Sci. USA 88 (1991) SD 1

A

hRARa

B

1

2

3

w w s I PTG

-

+

+

97,400--66,200--

--

_W 45,000--

40-

RARcx

__

31,000-

21 ,500---

_

FIG. 1. Construction of pET-8cRAR and expression of RAR protein in E. coli. (A) A schematic presentation of plasmid pET8cRAR. hRARa coding sequence starting from the first methionine is presented as an open box. First and last amino acids of RAR are indicated by numbers 1 and 462. Nco I, BamHI, and Asp718 cloning sites are indicated. The Shine-Dalgarno sequence (SD) is preceded by T7 promotor T7410-slO. (B) Expression of hRARa in E. coli. Before (lane 1) and after (lane 2) a 3-hr IPTG induction, cells in 20 ,ul of cell culture were lysed in loading buffer and proteins were resolved on a 10%o polyacrylamide gel containing SDS and stained with Coomassie brilliant blue R. Lane 3 is the soluble fraction (20 ul) from a culture after 3 hr of IPTG induction. Molecular weight markers are labeled on the left of the gel. RARa protein with a molecular mass of 54 kDa is indicated by the arrow on the right. Lanes: W, whole cell; S, soluble fraction. +, IPTG induction; -, no induction.

with /3RE is competed by an excess amount of unlabeled PRE oligonucleotide (lanes 4-6), neither the glucocorticoid receptor responsive element nor the inactive RAR response element derivative mtl (16) can compete the binding (lanes 8 and 7, respectively). In vitro DNA binding activity is independent of the presence or absence of RA. The DNA binding activity of RAR was assigned to a 66-amino acid peptide sequence residing in the middle of the RAR molecule based on sequence homology of RAR with other receptors of the family (6). To demonstrate that this region of the receptor is a functionally independent domain

Physiology/Pharmacology: Yang et al. 1 2 competitor - -

3 -

4 -

5

6

7

Proc. Natl. Acad. Sci. USA 88 (1991) mock

8

ORE - mtl GRE

cell extract

RAR RAR DBD -+! [7- +

Is

1

FIG. 2. DNA binding properties of hRARa assayed by gelmobility shift. A 32P-labeled oligonucleotide (1 ng) containing /RE was incubated with 2 ,u1 of a RARa-containing bacterial lysate in the absence of a responsive element (-, lane 3) or the presence of 5 ng, 10 ng, or 25 ng of unlabeled /BRE (lanes 4-6, respectively) or 25 ng of mtl or 25 ng of glucocorticoid receptor responsive element (GRE) competitor oligonucleotides (lanes 7 or 8, respectively) for 10 min at 0WC. Lanes 1 and 2 are control experiments. Lane 1 has no protein added. Lane 2 has 2 A.l of mock bacterial extract added. Mock bacterial extract is prepared from a BL21(DE3)plysS culture after a 3-hr IPTG induction. Poly(d-dC) (1 fug) was added to each reaction mixture as a nonspecific DNA competitor. The DNA-protein complex was then resolved from free probe on a 5% nondenaturing polyacrylamide gel.

structure that has specific DNA binding activity, we expressed the polypeptide sequence from amino acid 87 to amino acid 156 of RAR encompassing the DNA binding domain in E. coli. As shown in Fig. 3, this 70-amino acid polypeptide binds to the PRE (lane 5). (Due to the tendency to overexposure of other lanes in this gel, lane 5 was much underexposed.) High-Affinity RAR DNA Binding Requires a Protein Factor Present in Eukaryotic Cell Extracts. Both intact RAR and the DNA binding domain of RAR bind the DNA probe with high specificity. However, this binding can only be detected under low-stringency condition. The presence of KCI or prolonged incubation will dissociate the DNA-protein complex completely, indicating that the binding has a very high off rate and hence a very low association constant. We therefore speculated that a nuclear factor might exist to promote or improve RAR binding to DNA. To address this question, cell extracts from eukaryotic sources were added to the in vitro binding reaction mixture. Interestingly, in the presence of 5 ,ug of total proteins from monkey kidney COS cell extracts, RAR displayed a much stronger DNA binding activity (Fig. 3, compare lane 4 with lane 3). Other mammalian cell extracts tested including F9 cells and HeLa cells showed the same activity. Drosophila melanogaster Schneider cell extracts also conferred this DNA binding enhancing activity (Fig. 4, compare lane 5 with lane 4). In the presence and absence of cell extracts, the RAR-DNA complex showed different mobilities on the gel (Figs. 3 and 4). These further retarded bands suggest that RAR, DNA, and a third factor formed a new complex. This complex was very stable'under high salt and high-stringency conditions.

2

3561

3

_

4

5

6

FIG. 3. Eukaryotic cellular factor enhances RAR DNA binding and the factor interacts with the DNA binding domain of RAR. Gel retardation was done under the same conditions as in Fig. 2. Total proteins of COS cell extracts (5 Ag) were added to the binding reaction mixtures (+; lanes 2, 4, and 6). Lanes: 1 and 2, 2 Al of BL21(DE3)plysS lysate; 3 and 4, 2 p.1 of RAR-transformed bacterial lysate; 5 and 6,22A1 of RARDBD-transformed bacterial lysate. Arrows on the left indicate the RAR-DNA complex (lane 3) and the RARDNA-cellular factor complex (lane 4). The arrow on the right indicates the RARDBD-DNA-cellular factor complex. -, Cell extract not added.

DNA Binding Enhancing Factor Is Heat-Sensitive and Interacts with the RAR DNA Binding Domain. To test the stability of the factor, we pretreated the cell extracts at 650C for 10 min. The DNA binding enhancing activity is completely destroyed after this treatment (Fig. 4, lane 6). Most interestingly, when 5 pug of total proteins from a COS cell extract was added to RARDBD in this analysis, a much enhanced DNA binding activity was detected (Fig. 3, compare lane 6 with lane 5). The new complex of the RARDBD and the DNA binding enhancing factor is very stable under

high-stringency conditions. Based on these experiments, we conclude that a protein factor in eukaryotic cells is required for high-affinity DNA binding of the RAR protein. And at least one site of this interaction is located in the DNA binding domain of RAR. RAR Protein Binds to RA with High Affinity. The charcoal absorption hormone binding assay was used to analyze the ligand binding property of RAR (18). The linear plot of the saturation binding isotherm for RA-RAR binding is shown in Fig. 5A. Nonspecific binding determined by addition of a 100-fold molar excess of unlabeled RA showed a linear response. Specific binding of RAR protein was obtained by subtracting values for nonspecific binding from total binding. No specific binding was detected in a mock-transformed BL21 bacteria cell lysate. A dissociation constant of 2.1 x 10-10 M was determined by Scatchard analysis (Fig. 5B).

DISCUSSION The key event in the process of RAR activation and subsequent gene transcription is the association of RAR with RA. The efficient binding of bacterially expressed RAR to RA is significant for several reasons. First, the binding kinetics have been difficult to measure in mammalian cells because of the relatively low levels of RAR compared to the high concentrations of the endogenous cellular RA binding protein. This is further confounded by the high affinity of the cellular RA binding protein for RA (Kd = 7-42 nM) (19). Thus these properties conspired to preclude identification of the

3562

Physiology/Pharmacology: Yang et al. control

RAR

A

I x

G)

Proc. Natl. Acad. Sci. USA 88 (1991)

x~

a)

I -x

a)

10000 .

-x

CD

8000

-

6000. V c

4000.

0 co

I

2000 .

1.2

3[H]R.A. (nM)

B 0.30 -

It

1

2 3 4 5 6

co

0

FIG. 4. DNA binding enhancing activity is present in Schneider cells and the factor is heat-sensitive. Gel retardation was done under the same condition as in Fig. 2. An untransformed bacterial lysate (lanes 1-3) and a RAR-transformed bacterial lysate (lanes 4-6) were incubated with cell extracts. Lanes: 1 and 4, no protein (-); 2 and 5, 4 jug of proteins of Schneider cell extracts (extr.); 3 and 6, 4 ,.g of proteins of Schneider cell extracts pretreated at 650C for 10 min [extr. (H.I.)]. Arrows indicate the migration of the RAR-DNA complex in the absence (lower arrow) or presence (upper arrow) of Schneider cell extracts.

endogenous RAR protein prior to the cloning of its cDNA. The expression studies confirm that bacterially synthesized RARa possesses an intrinsic affinity for RA (Kd = 2.1 x 10-10 M) that is close to its half-maximal values of stimulating cellular differentiation and transcriptional activation. In addition, it has been suggested that glucocorticoid receptor (GR) hormone binding activity is modulated by another protein factor hsp90. When expressed in E. coli in the absence of hsp90 GR protein products bind dexamethasone with a Kd value about two magnitudes lower than the GR protein expressed in eukaryotic cells. In contrast, our data showed that no posttranslational modification or other eukaryotic cellular factors are required for high-affinity RA binding. This is consistent with the experimental observation that unlike GR and progesterone receptor (PR), RAR is free of association with hsp90 in the cell. The DNA binding domain of nuclear receptors has been defined as a 66-amino acid peptide sequence located in the middle of the primary sequence of the receptor molecules, based on the fact that this region is highly conserved in all members of the family. We now demonstrate that this central core of the receptor molecule when expressed in E. coli functions as an independent domain fully capable of binding DNA in a sequence-specific fashion. We have also shown that this peptide sequence forms a complex with a cellular protein factor that exhibits remarkably improved DNA binding kinetics, suggesting that in vivo this sequence interacts with another factor as a prerequisite for the RAR-DNA binding. In conclusion, we have developed a high-expression system for producing large quantities of nuclear receptor proteins in E. coli. Bacterially expressed receptors exhibit full

0.20-

0.10 -

K~d- 2.1 x10 'oM .I 0.00

0.01

0.02

.

0.03

0.04

.

.

.

I

0.05

.

I

.

0.06

0.07

3 [1IR.A. (nM)

FIG. 5. Saturation curve and Scatchard analysis. (A) RA binding assayed by the charcoal absorption assay. Total bacterial extracts (1 ,l) containing RAR were incubated with [3H]RA for 16 hr at 40C in the dark in 1 ml of binding buffer. Nonspecific binding was determined by addition of 100-fold molar excess of unlabeled RA. Free RA was separated from bound RA by charcoal absorption. No specific binding was seen in mock-transformed BL21(DE3)plysS extracts. o, Nonspecific binding activity; total binding activity; *, RAR-specific binding activity. (B) Scatchard analysis of [3HJRA binding in extract prepared from pET-8cRAR-transformed IPTGinduced BL21 cells. Each point was assayed in duplicate. The line was best-fitted by a Cricket graph program. The Kd value was calculated as 2.1 x 10-10 M. was

o,

DNA binding and hormone binding activities. We have also identified a eukaryotic cellular protein factor whose presence greatly facilitates RAR-DNA binding. Expression of the functional hRARa protein provides us with the possibility of studying the biochemical and physical chemical properties of hRAR molecules as well as the molecular mechanisms of transcriptional activation mediated by RAR. 1. Brockes, J. P. (1989) Neuron 2, 1285-1294. 2. Thaller, C. & Eichele, G. (1987) Nature (London) 327,625-628. 3. Roberts, A. B. & Sporn, M. B. (1984) in The Retinoids, eds. Sporn, M. B., Roberts, A. B. & Goodman, D. S. (Academic, Orlando, FL), Vol. 2, pp. 209-286. 4. Maden, M., Ong, D. E., Summerbell, D. & Chytil, F. (1988) Nature (London) 335, 733-735. 5. Durston, A. J., Timmermans, J. P. M., Hage, W. J., Hendriks, H. F. J., de Vries, N. J., Heideveld, M. & Nieuwkoop, P. D. (1989) Nature (London) 340, 140-144. 6. Giguere, V., Ong, E. S., Segui, P. & Evans, R. M. (1987) Nature (London) 330, 624-629. 7. Petkovich, M., Brand, N. J., Krust, A. & Chambon, P. (1987) Nature (London) 330, 444-450. 8. Benbrook, D., Lernhardt, E. & Pfahl, M. (1988) Nature (London) 333, 669-672. 9. Brand, N., Petkovich, M., Krust, A., Chambon, P., de The, H.,

Physiology/Pharmacology: Yang et al.

10. 11.

12. 13. 14.

Marchio, A., Tiollais, D. & Dejean, A. (1988) Nature (London) 332, 850-853. Krust, A., Kastner, P. H., Petkavich, M., Zelent, A. & Chambon, P. (1989) Proc. Nadl. Acad. Sci. USA 86, 5310-5314. de The, H., Vivanco-Ruiz, M., Tiollais, P., Stunnenberg, H. & Dejean, A. (1990) Nature (London) 343, 177-180. Vasios, G. W., Gold, J. D., Petkovich, M., Chambon, P. & Gudas, L. J. (1989) Proc. NatI. Acad. Sci. USA 86, 9099-9103. Sucov, H. M., Murikami, K. K. & Evans, R. M. (1990) Proc. Nadl. Acad. Sci. USA 87, 5392-53%. Studier, W. F., Rosenberg, A. H. & Dunn, J. J. (1990) Methods Enzymol. 185, 60-89.

Proc. Nati. Acad. Sci. USA 88 (1991)

3563

15. Chytil, F. & Ong, D. E. (1984) in The Retinoids, eds. Sporn, M. B., Roberts, A. B. & Goodman, D. S. (Academic, Orlando, FL), Vol. 2, pp. 89-123. 16. Schule, R., Mangeldorf, D. J., Umesono, K., Borado, J. & Evans, R. M. (1990) Cell 61, 497-504. 17. Damm, K., Thompson, C. C. & Evans, R. M. (1989) Nature (London) 339, 583-597. 18. Dokoh, S., Pike, J. W., Chandler, J. S., Mancini, J. M. & Haussler, M. R. (1981) Anal. Biochem. 116, 221-222. 19. Haussler, M. R., Donaldson, C. A., Kelly, M. A., Mangelsdorf, D. J., Bowden, G. T., Meinke, W. J., Meyskens, F. L. & Sidell, N. (1984) Biochim. Biophys. Acta 803, 54-62.