175

Molecular and Cellular Endocrinology, 70 (1990) 175-184 Elsevier Scientific Publishers Ireland, Ltd.

MOLCEL

02269

Human c-e& A protein expressed in Escherichia coli: changes in hydrophobicity upon thyroid hormone binding Kazuo Ichikawa ‘, Kiyoshi Hashizume ‘, Shuichi Furuta 2, Takashi Osumi 2, Takahide Miyamoto ‘, Keishi Yamauchi ‘, Teiji Takeda 1 and Takashi Yamada ’ ’ Department

of Gerontology, Endocrinology, and Metabolism, ’ Department of BiochemistT, Asahi 3-l-1, Matsumoto

Shinshu University School of Medicine,

390, Japan

(Received 12 September 1989; accepted 5 January 1990)

Key words: c-erb A; Escherichia coli expression vector; Thyroid hormone receptor; Aqueous two-phase partitioning study; Affinity labeling

The human c-erb AB gene sequence was inserted in an Escherichia coli expression vector plasmid. The E. coli cells transformed with this plasmid produced proteins with molecular masses of 52 and 50 kDa. These products bound 3,5,3’-triiodo-r_-thyronine (T3) with an affinity constant of 4.3 X lo9 liter/mol. The order of affinity for iodothyronine analogs was triiodothyroacetic acid > T3 > 3,5,3’-triiodo-D-thyronine > L-thyroxine. Affinity labeling experiments showed that the 50 kDa protein was covalently labeled with [‘*‘I]T3, and this was competed by triiodothyroacetic acid, T,, and L-thyroxine (from potent to weaker competitor). The c-erb A protein bound to calf thymus DNA-cellulose and the binding was inhibited by 0.3 M KC1 or 10 mM pyridoxal 5’-phosphate. Aqueous two-phase partitioning studies showed that the c-erb A product became less hydrophobic upon Ts or triiodothyroacetic acid binding. The same finding was obtained when Tj bound to partially purified rat liver nuclear thyroid hormone receptor. However, thyroxine binding globulin became more hydrophobic upon T3 binding. Since the T3 molecule partitioned preferentially into the upper polyethylene glycol-rich phase, the alteration of partitioning behavior of thyroxine binding globulin was explained by a simple additive effect of Ts. In contrast, the alteration of partitioning behavior of the c-erb A product or receptor reflected a conformational transition upon T3 binding. The c-erb A protein expressed in E. coli showed various characteristics similar to classical thyroid hormone receptor and may be useful in studying the structure and function of the thyroid hormone receptor.

Introduction The proto-oncogene c-erb A is a cellular homolog of v-erb A, which is a component of the avian erythroblastosis virus gene. It was first identified

Address for correspondence: Kazuo Ichikawa, M.D., Department of Gerontology, Endocrinology, and Metabolism, Shinshu University School of Medicine, Asahi 3-1-1, Matsumoto 390, Japan. 0303-7207/90/$03.50

in chicken DNA (Vennstrom and Bishop, 1982). Recognition of sequence homology between the human glucocorticoid receptor and the v-erb A oncogene prompted the finding that the human and chicken c-erb A products may be high affinity receptors for thyroid hormone (Sap et al., 1986; Weinberger et al., 1986). Similarities in molecular weight, binding affinity for 3,5,3’-triiodo-Lthyronine, relative affinity for iodothyronine analogs and in abundance in tissue between the c-erb

0 1990 Elsevier Scientific Publishers Ireland, Ltd.

A product and thyroid hormone receptor suggest that they are identical. In the present study, a human c-e& A gene was inserted into an E. c&i expression vector plasmid and the c-erb A products were expressed. c-erb A products expressed in E. coli showed hormone binding and calf thymus DNA binding similar to those of classical thyroid hormone receptor. Binding of 3,5,3’-triiodo-Lthyronine caused changes in hydrostatic properties of the c-e& A products which represent a conformational alteration of the c-erb A products. Materials

and methods

DNA restriction, ligation, and electrophoresis were performed as described by Maniatis et al. (1982). 3,5,3’-Triiodo-L-thyronine, 3,5,3’-triiodoD-thyronine, pyridoxal 5’-phosphate and L-thyroxine were from Sigma Chemical Co. (St. Louis, MO, U.S.A.). Isopropyl-&D-thiogalactopyranoside was from Wako Chemical Co. (Osaka, Japan). Phenylmethylsulfonyl fluoride, polyethylene glycol (molecular weight 7800-9000), and dithiothreitol were from Nakarai Chemical Co. (Kyoto, Japan). Expression plasmid pKK 233-2, double-str~ded calf thymus DNA-cellulose and Dextran T-500 were from Pharmacia Fine Chemicals (Piskataway, NJ, U.S.A.). Restriction enzymes were purchased from Takara Shuzo Co. (Tokyo, Japan). [1251]3,5,3’-Triiodo-L-thyronine (3000 pCi/pg) was from New England Nuclear (Boston, MA, U.S.A.). Dowex l-X8, Cl-, 200-400 mesh, anion-exchange resin was from Bio-Rad (Richmond, CA, U.S.A.). Disuccinimidyl suberate was from Pierce Chemical Co (Rockford, IL, U.S.A.).

pheA 12, cloned from a human placenta cDNA library, contained the entire coding region of c-erb A/3 (Weinberger et al., 1986). PEA 101 (Weinberger et al., 1986), containing the pheA 12 sequence inserted into the EcoRI site of pGEM 3 (Promega Biotec), was cut with EcoRI. The pheA 12 insert was isolated by agarose gel electrophoresis followed by electroelution, and blunt-ended using mung bean nuclease. This DNA fragment was inserted into pKK 233-2 which had been cut with NcoI and blunt-ends using mung bean nuclease. A plasmid containing pheA 12 cloned in

the correct orientation was selected. The resulting plasmid itself directed by production of c-erb A/3 protein when introduced into E. colt, though at a very low level. To improve the yield of the protein, the DNA sequence containing the c-erb A/3 cDNA and the promoter region was transferred to a pUC plasmid, which has an extremely high copy number in E. co/i cells. For this purpose, the plasmid was digested with EcoRI and ScaI, and the fragment containing pheA 12 was ligated to the larger fragment of pUC 8 after digesting with EcoRI and ScaI. This plasmid (pNTR) was used for the expression of the human placental c-erb A in E. co/i. This construction destroyed the initiation ATG codon located in the NcoI site of pKK233-2. Instead, the first methionine codon (ATG) of the translated region of the c-erb A cDNA is located 13 bases downstream of the lacZ ribosome binding sequence (AGGA) of pKK233-2. The result of dideoxy sequencing (Sanger et al., 1977) of the 5’ junction was AGGAAACAGACCGCGGTATG (underlined are the 1acZ ribosome binding sequence of pKK233-2 and the initial methionine codon of c-erb A cDNA). Therefore this plasmid was expected to produce non-fusion c-erb A protein. In fact, proteins produced were 52 kDa and 50 kDa. These values agree to molecular masses of polypeptides initiating translation at methionine 1 and 27 of the translated region of c-erb A cDNA that were calculated to be 52 kDa and 49 kDa from the predicted amino acid sequences.

Preparation of c-erb A proiein E. coli (JM 103) transformed with pNTR was propagated in L-broth with 0.5% glucose in the presence of 50 pg/rnl of ampicillin at 37” C for 12-14 h. Thereafter, culture was continued in the same media, without glucose, in the presence of 1 mM isopropyl-/3-D-thiogalactopyranoside at 37 o C for 14 h. E. cofi were collected by centrifugation and suspended in 15 mM potassium phosphate buffer pH 7.4. The cells were sonicated for 3 min in the presence of 5 mM EDTA and 0.5 mM phenylmethylsulfonyl fluoride, frozen at - 70 o C for 15 min, and sonicated for 3 min again. Insoluble materials were removed by centrifugation at 12,000 x g at 2O C for 20 min. Most of the c-erb A product was recovered in the supernatant (Fig. 1).

Preparation of rat liuer nuclear receptor Preparation of rat liver nuclear extract was performed as described previously by Torresani and DeGroot (1975). The receptor was further purified using a hydroxylapatite column, ammonium sulfate precipitation, and Sephadex G-150 column chromatography, as previously reported (Ichikawa and DeGroot, 1987), except that the receptor was eluted from the hydroxylapatite column using 200 mM phosphate buffer instead of a linear phosphate gradient and that 50 mM KCl, 1 mM EDTA, 5 mM Tris-HCl pH 7.4 and 1 mM dit~ot~eitol were used for the elution buffer of the Sephadex G-150 column.

3,5,3’-Triiodo-L-thyronine binding assay Assay of binding of [‘251]3,5,3’-triiodo-Lthyronine to the c-erb A protein or receptor was performed as previously described (Ichikawa and DeGroot, 1987). Dowex anion-exchange resin was used for the separation of bound [‘251]3,5,3’-triiodo-r.,-thyronine from free [‘251]3,5,3’-triiodo-~thyronine (Torresani and DeGroot, 1975).

Cross-linking of ~“‘I]3,5,3’-triiodo-~-thyr~~i~e and the c-erb A products E. coli extract was incubated with 0.3 nM [1251]3,5,3’-triiodo-L-thyronine in the absence or presence of unlabeled iodothyronine in 0.3 M NaCl and 15 mM potassium phosphate pH 7.4 at 22 o C for 2 h. Samples were cooled in an ice bath and free { ‘251]3,5,3’-triiodo-L-thyronine was removed by two treatments with Dowex anion-exchange resin. Fleshly prepared ~su~~~dyl suberate (50 mM in dimethyl sulfoxide) was added to a final concentration of 0.5 mM. After 30 min of incubation in an ice bath, a one-third volume of 15% sodium dodecylsulfate/l5% 2-mercaptoethano1/188 mM Tris-HCl pH 6.8/30% glycerol/ 0.006% bromophenol blue in H,O was added to quench the reaction. Samples were kept in boiling water for 3 rain and analyzed by polyacrylamide gel electrophoresis followed by autoradiography. In some experiments, labeling was quantitated by densitometry using Image Analyzer, type TIB 100 (I~unomedica Co., Tabata City, Shizuoka, Japan).

Preparation of thyroid hormone-free or [‘251f3,5,3’triiodo-L-thyronine-o~e~pied ~arnple~ Rat liver receptors occupied and unoccupied by 3,5,3’-triiodo-L-thyronine were prepared as follows. Receptors were first incubated with Dowex resin at 22°C for 120 min during which time the samples were vortexed at 15-min intervals. The samples were then centrifuged to remove the resin. By this procedure, total and specific protein-bound 3,5,3’-triiodo-L-thyronine decreased to 25.7 rf 1.5 and 22.9 f 0.88, respectively, of initial values, when examined using [‘251]3,5,3’-triiodo-Lthyro~e-labeled receptor. However, maximal 3,5,3’-total-thyro~ne-binding capacity of the receptor was not significantly altered by this treatment. Since the amount of endogenous 3,5,3’-triiodo+thyronine in the nuclear extract determined by radioimmunoassay was 57.1% of the maximal 3,5,3’-triiodo-r..-thyronine-binding capacity of the receptor in the extract, at least 85% of the receptor became free of 3,5,3’-triiodo-L-thyronine by this procedure. These procedures were omitted for c-erb A protein expressed in E. coli since no endogenous 3,5,3’-t~iodo-L-thyronine was detected. 3,5,3~-T~~~L-thyro~ne-free receptors or c-erb A protein were divided into three parts: 10 nM [1251]3,5,3’-triiodo-L-thyronine (total binding), 10 nM [‘251]3,5,3’-triiodo-L-thyronine plus 5 PM unlabeled 3,5,3’-triiodo-L-thyronine (non-specific binding), and solutions without 3,5,3’-triiodo-r.thyronine were added to lo-15 ~1 of E. coli extract of c-erb A protein or 0.1 ml of rat receptors at a final volume of 0.6 ml (50 mM Tris-Cl, pH 7.4,1 mM EDTA, and various salt concentrations). The samples were incubated for 2 h at 22* C to produce 3,5,3’-t~~~L-th~o~ne-~cupied and -unoccupied receptors or c-erb A protein, cooled in an ice bath for 15 mm and used for the studies. Thyroxine binding globulin occupied or unoccupied by 3,5,3’-triiodo-L-thyronine was prepared as follows. 10 ~1 of human serum in 0.6 ml of 50 mM Tris-Cl, pH 7.4, 1 mM EDTA and various salts was incubated with 40 mg Dowex resin at 22” C for 14 h with continuous mixing. This treatment removed 97.3% of endogenous thyroxine and 99.7% of endogenous 3,5,3’-triiodo-L-thyronine. The 3,5,3’-triiodo+thyronine binding capacity of the serum, however, did not change sig~ficantly with this procedure. Resin was removed by

178

centrifugation and 0.12 ml each was diluted to 0.6 ml of the same buffer. They were incubated without or with 6.3 nM [‘251]3,5,3’-triiodo-Lthyronine in the presence or absence of 3 PM unlabeled 3,5,3’-triiodo+thyronine at 22OC for 2 h. These were used as hormone-free and 3,5,3’-triiodo-L-thyronine-occupied thyroxine binding globulin for the study. Thyroxine binding prealbumin did not interfere with the results, since it does not bind 3,5,3’-triiodo-L-thyronine in any buffer system (Oppenheimer, 1968). This was further confirmed by the similar 3,5,3’-triiodo-L-thyronine binding activity obtained in the presence of barbital buffer (pH 8.6) which interferes with thyroid hormone binding to the thyroxine binding prealbumin. Aqueous two-phase partitioning study The aqueous two-phase partitioning study was performed as described previously (Hansen and Gorski, 1985). Dextran and polyethylene glycol stock solutions were prepared at 20.2% (by mass) in 1 mM EDTA and 50 mM Tris-Cl, pH 7.4. The equilibrated undiluted phase system, composed of 9% (by mass) each of dextran and polyethylene glycol, various concentrations of salts, and 1 mM dithiothreitol, was formed by combining the respective stock solutions and vortexing rapidly. Aliquots (0.9 ml) of this turbid polymer and salt mixtures were added to 0.6 ml of samples either occupied or unoccupied with [‘251]3,5,3’-triiodoL-thyronine. Samples contained lo-15 ~1 of E. coli extract for c-erb A protein, 0.1 ml of partially purified rat liver receptor, or 2 ~1 of human serum. The final mixture was composed of 5.4% (by mass) each of dextran T-500 and polyethylene glycol, 50 mM Tris-HCl pH 7.4, 1 mM EDTA, 1 mM dithiothreitol, various concentrations of salts and protein sample. The mixture was vigorously vortexed for 20 s during which time partitioning equilibrium was reached, and then centrifuged at 600 x g for 5 min for the phase separation. The partition between the two phases and the final recovery of the receptor were not affected by the length of vortexing from 10 to 40 s. After the phase separation, 0.2 ml of each phase was placed in a separate tube. If the 3,5,3’-triiodo-Lthyronine-bound protein was partitioned, 0.8 ml of 0.38 M KCl, 1 mM MgCl,, 20 mM potassium

phosphate, pH 8.0, and 1 mM dithiothreitol (buffer A) containing 40 mg Dowex anion-exchange resin was added into each phase and protein-bound [‘251]3,5,3’-triiodo-L-thyronine was determined. If the 3,5,3’-triiodo-L-thyronine-unoccupied protein was partitioned, 0.4 ml buffer A containing 10 nM [‘251]3,5,3’-triiodo-L-thyronine with or without 5 PM unlabeled 3,5,3’-triiodo-L-thyronine was added into each phase. After 2 h of incubation at 22” C, samples were cooled in an ice bath and 0.4 ml of buffer A containing 40 mg Dowex resin was added, and specifically bound [‘251]3,5,3’-triiodoL-thyronine was determined in each phase. Specifically bound [‘251]3,5,3’-triiodo-L-thyronine was calculated by subtracting non-specific [‘251]3,5,3’-triiodo-L-thyronine binding from total [‘251]3,5,3’-triiodo-L-thyronine binding. 3,5,3’-Triiodo-L-thyronine binding activity and stability of the 3,5,3’-triiodo-L-thyronine-receptor complex in the presence of the phase solutions were checked in each assay. The partition coefficient was calculated as the ratio of receptor concentration in the upper phase H,O to that in the lower phase H,O. H,O content was 89.9% in the upper phase and 79.0% in the lower phase. H,O content in each phase and the volume ratio of the phases were not affected by the difference in the buffer and salt. Binding of 3,5,3’-triiodo-L-thyronine to the receptor or c-erb A protein reached a plateau after 90 min of incubation at 22 o C and did not change for up to 180 min of incubation at neutral pH in the presence or absence of these phase solutions, when incubated with either a small amount of 3,5,3’-triiodo-L-thyronine or a saturating amount for 3,5,3’-triiodo-L-thyronine. The receptor or c-erb A protein was incubated with [“‘1]3,5,3’-triiodo+ thyronine for 2 h in the present study. Although the amount of 3,5,3’-triiodo-L-thyronine bound to the receptor or c-erb A protein was the same in the presence of these solutions when a saturating amount of 3,5,3’-triiodo-L-thyronine was used, the amount of bound 3,5,3’-triiodo-L-thyronine was the highest in buffer A, followed by the lower phase, and upper phase when a small amount of 3,5,3’-triiodo-L-thyronine was used. Scatchard analysis (Scatchard, 1949) of 3,5,3’-triiodo-Lthyronine binding to the receptor or c-erb A protein showed approx. half affinities both in the presence of the upper and lower phases when

compared to that in buffer A. However, maximal 3,5,3’-triiodo-L-thyronine binding capacities were almost the same in these solutions. Based on these results, we used a receptor saturating dose of 3,5,3’-triiodo-L-thyronine (10 nM of 3,5,3’-triiodoL-thyronine for less than 7 nM of the receptor) to determine the amount of the receptor or c-erb A protein in each phase. When [‘251]3,5,3’-triiodo-Lthyronine-bound c-erb A or receptor, deprived of free [ 1251]3 5 3’-triiodo-L-thyronine, was incubated with 1 & unlabeled 3,5,3’-triiodo-L-thyronine at 18” C, half-dissociation times of [‘251]3,5,3’-triiodo-L-thyronine from c-erb A or receptor were 1.6 to 1.8 times longer in these phase solutions than in buffer A, indicating that 3,5,3’-triiodo-Lthyronine-bound receptor is rather stabilized in these phase solutions. Dissociation of [‘*‘1]3,5,3’triiodo-L-thyronine from the c-erb A protein or receptor was negligible at 2” C in these phase solutions. The partition coefficients of the 3,5,3’triiodo-L-thyronine-occupied and -unoccupied protein and recoveries of the 3,5,3’-triiodo-rthyronine binding activity did not change when the samples were kept at 2” C for 60 min after the phase separation, again suggesting that the [‘*‘1]3,5,3’-triiodo-L-thyronine-receptor complex and 3,5,3’-triiodo-r_-thyronine binding activity of the receptor or c-erb A protein are stable in this system. For thyroxine binding globulin, the amount of unoccupied thyroxine binding globulin determined after phase separation was corrected for a slight decrease in 3,5,3’-triiodo-L-thyronine binding activity of the thyroxine binding globulin in each phase. However, such corrections did not significantly affect the results, since the decrease in binding was very small. Dissociation of 3,5,3’triiodo-L-thyronine from the thyroxine binding globulin was negligible. Bound 3,5,3’-triiodo-r.thyronine was proportional to the amount of serum and was not affected by the different amounts of 3,5,3’-triiodo-L-thyronine (3-20 nM) added in the assay, suggesting that all the thyroxine binding globulin was saturated by 3,5,3’-triiodo-Lthyronine and the amount of the thyroxine binding globulin was correctly measured. Separation of bound and free 3,5,3’-triiodo-L-thyronine by Dowex resin was not affected in the presence of these phase solutions. In some experiments, partitioning of protein

between two phases was analyzed using sodium dodecylsulfate-polyacrylamide gel electrophoresis. In these experiments, unlabeled iodothyronine, instead of [‘251]3,5,3’-triiodo-r_-thyronine, was used. Results Expression of human c-erb A protein Proteins with molecular masses of 52 and 50 kDa were produced (Fig. 1) when E. coli transformed with pNTR were induced by 1 mM isopropyl-j%D-thiogalactopyranoside in the absence of glucose. 52 kDa and 50 kDa products correspond to polypeptides initiating translocation at methionine 1 and 27, respectively, and these molecular masses agree to those calculated from predicted amino acid sequences (52 kDa and 49 kDa, respectively). [‘251]3,5,3’-Triiodo-~-thyronine binding activity was also induced by the same treatment. Scatchard analysis (Scatchard, 1949) showed that these proteins bound 3,5,3’-triiodo-Lthyronine with an affinity constant of 4.3 x lo9 liter/m01 (Fig. 2) and maximal binding capacity of 80 pmol 3,5,3’-triiodo-r_-thyronine/mg protein.

kDo

-92.5 -

68.0

- 45.0

-31.0

I*r 123

45

-21.5

6

Fig. 1. Expression of c-erb A. After 14 h of propagation, E. coli harboring pNTR were incubated for 12 h either in L-broth with 0.5% glucose (lanes 1-3) or in L-broth with 1 mM isopropyl-p-D-thiogalactopyranoside without glucose (lanes 4-6). E. co/i extract prepared by sonication in the absence (lanes 1 and 4) or presence (lanes 2 and 5) of 0.4 M NaCl and whole E. coli cells (lanes 3 and 6) were run on a 10% polyacrylamide gel and stained with Coomassie brilliant blue R-250. Positions of molecular mass markers are indicated.

180 5

0.4

0

Ko

q

4.3 x 10’

liter / mol

h

IODOTHYRONtNE 0

50 Bound

100

fpMf

Fig. 2. Scatchard piot analysis of 3,5,3’-triiodo-r.-thyronine (Ta) binding to the c-erb A product expressed in E. coli. Cell extract of E. coli containing the c-erb A product was incubated with 36 pM [“‘I]T, and varying amounts (O-3.0 nM) of unlabeled Ts. Duplicate tubes containing in addition 0.3 pM unlabeled T3 were used for the determination of non-specific binding. Cell extract of E. coli without the c-erb A product did not show any specific Ts binding

ANALOGUE CONCENTRATION(M)

Fig. 3. Potency of iodothyronine analogs to displace [‘251]3,5,3’-triiOdo-L-thyTOnine (Ts) binding to the c-erb A product expressed in E. coli. Cell extract of E. coli containing the c-erb A product was incubated with 0.1 nM (‘2sI~s in the presence of various amounts of 3,5,3’-triiodothyroacetic acid (A), T3 (O), 3,5,3’-triiodo-D-thyronine (o), or t-thyroxine (m). Bound [‘2sI]T, was determined and expressed as percent of [lzsI]T, binding in the absence of unlabeled iodothyronine.

kDo

kDo

When the amount of c-e& A proteins in the extract was estimated from the protein staining of electrophoresis gel, 50-608 of the c-erb A protein bound 3,5,3~-t~od~L-thyro~ne, assuming that the c-erb A protein has a single 3,5,3’-triiodo+ thyronine binding site. The order of affinity of iodothyronine analogs was triiodothyroacetic acid > 3,5,3’-triiodo-L-thyronine z 3,5,3’-triiodo-Dthyronine S- L-thyroxine (Fig. 3). E. coli extract which did not express c-erb A products showed no specific 3,5,3’-triiodo-L-thyronine binding. Crosslinking of [‘251j3,5,3’-triiodo-x_-thyronine to E. co/i extract which expressed the oerb A protein, followed by sodium d~ecylsulfate-~lyac~la~de gel electrophoresis and autoradio~aphy, showed that 50 kDa protein was covalently labeled with [1251]3,5,3’-triiodo-L-thyronine (Fig. 4). 63% of the labeling was inhibited by 1 nM of 3,5,3’-triiodot-thyronine, when analyzed using densitometry. Inhibition was more potent using triiodothyroacetic acid (66% inhibition) and was weaker using t-thyroxine (35% inhibition) than 3,5,3’-triiodo-Lthyronine. 1 ,uM of 3,5,3’-triiodo-L-thyronine completely inhibited the labeling. These data show that cross-linking reflects limited capacity high affinity binding of 3,5,3’-t~i~o-L-th~o~e. E. co& extract which did not express the c-erb A

92.5

,_,’ __: __,. _,_-‘-1. ,. ,( ,. -; _.. :~:,:’ ,;;,

-

@=&t-J-

‘:

45.0-

-50

‘,

31.021.51234567 L

Time

after induction fh)

Unlabeled iodothyronine

- 0.5

-

-

I

*

14

-

.g 7 $;-I; 1 InM

i+yr

I IpM

Fig. 4. Affinity labeling of the c-erb A products with [‘251]3,5,3’-triiodo-L-thyronine (Ts). After 14 h of propagation, E. coli cells were incubated for 12 h in L-broth with 0.5% glucose (lane l), or in L-broth with 1 mM isopropyl-&nthiogalactopyranoside without glucose (lanes 3-7). Lane 2 is the cell extract of E. coli harvested 0.5 h after the induction (the second incubation). Equivalent protein contents of E. co& extracts were cross-linked as described in Materials and Methods. 0.3 nM f’2sI]T~ in the absence (lanes 1-3) or presence of 1 nM 3,5,3’-t~~othyr~cetic acid (lane 4), 1 nM t-thyroxine (lane 5),1 nM Ta (lane 6) or 1 FM Ts (lane 7) was used.

181

3

2 III!iI

-2

,9 PYr

f \ 1

5’P

I ’ I 4 008

0 0 FRACTION

protein showed no cross-linking of [tz51]3,5,3’triiodo-L-thyronine (Fig. 4, lane 1). The c-erb A protein expressed in E. coli bound to calf thymus DNA-cellulose column at 0.05 M KC1 and was eluted at 0.2-0.3 M KC1 from the column (Fig. 5, left panel). 10 mM of pyridoxal 5’-phosphate demonstrated the inhibition of the binding of the c-erb A protein to calf thymus DNA-cellulose (Fig. 5, right panel). Previous studies indicated the involvement of Schiff base in the binding of thyroid (Ichikawa and DeGroot, 1987) or steroid (Cake et al., 1978) hormone receptor to DNA which is potently inhibited by pyridoxal 5’-phosphate. No c-erb A product bound to plain cellulose. These results demonstrated that human c-erb A products have thyroid hormone binding and calf thymus DNA binding activities quite similar to classical thyroid hormone receptors.

I w x-1

5

;

G t iz

-1

L :

‘a

0

5

10

NUMBER

Fig. 5. DNA-cellulose column profile of c-e& A protein. E. coli extract containing c-e& A protein (total protein of 30 mg with 2.4 run01 3,5,3’-ttiiodo+thyronine (Ts) binding capacity) was adjusted to 0.4 M KC1 by adding solid KCl. Polyethylenimine was added to a final concentration of 0.05%. This was centrifuged at 10,000 x 8 for 20 min at 4O C in order to remove insoluble materials including DNA. This was diluted 8 times by adding 10 mM potassium phosphate pH 7.5,1 mM EDTA, 1 mM dithiothreitol and 20% glycerol and applied to a DNAcellulose column (25 ml column volume with 30 mg doublestranded calf thymus DNA attached). The column was washed with 2 column volumes of 0.05 M KCl, 1 mM EDTA, 1 mM dithiothreitol, and 10 mM potassium phosphate pH 7.5. Linear KC1 gradient (left panel) or 10 mM pyridoxal 5’-phosphate (right panel) in 10 mM potassium phosphate pH 7.5, 1 mM EDTA, and 1 mM dithiothreitol was used for the elution of the protein. Fractions (10 ml each during loading and washing, and 3 ml each during elution) were collected. 2 pl from each fraction was assayed for protein content (0). Specific ]12SI]T, binding (0) was also determined using 32,000 cpm of [‘251]Ts. 1, loading of the sample; w, washing of the column.

TABLE

Alteration of the partition coefficients of the c-erb A protein upon 3,.5,3’-triiodo-L-thyronine binding Partitioning of charged molecules between two phases is a function of an electrical potential difference across the interface, which is caused by the different partitioning behavior of the anion and cation. Therefore the types and concentrations of salts affect partitioning behavior of charged molecules (Hansen and Gorski, 1985).

1

CONFORMATIONAL TRANSITION THYRONINE (Ts) OCCUPANCY

OF

c-erb

A

PROTEIN

AND

NUCLEAR

RECEPTOR

UPON

3,5,3’-TRIIODO-L-

Aqueous two-phase partitioning study of c-erb A protein, partially purified rat liver receptor, thyroxine binding globulin, and Ts was performed using 50 mM Tris-HCl pH 7.4,l mM EDTA with the salt composition indicated. T-,-binding capacity of 6.2 pmol (c-erb A protein). 1.4 pmol (rat liver receptor), and 0.33 pmol (thyroxine binding globulin) was used. Numbers in parentheses are total Ts binding activity (pmol) recovered from both phases. Means* SD of three determinations are shown. Similar results were obtained from another experiment. Variations in partition coefficients among different experiments are about 7% of mean values. Specific radioactivity of [i2’I]Ts was adjusted to 1000 cpm per fmol Ts by adding unlabeled Ts.

salt

Partition

coefficient

(Ts binding

c-erb A protein T3

0.3 M KC1 0.1 M KC1 0.3+M LiCl 0.05 M NaCl

(-)

activity

recovered;

pmol) Thyroxine

Rat liver receptor T3

(+I

T3

(-)

T3

(+I

T3

binding

t-1

globulin T3

T3

(+)

-

0.250 f 0.001 (4.11 f 0.03)

0.182 f 0.002 (4.14 f 0.04)

0.254 f 0.011 (1.16 *0.08)

0.193 f 0.012 (1.21 *0.05)

0.248 f 0.006 (0.30 *0.02)

0.276 * 0.004 (0.29 kO.01)

1.20*0.00

0.581 (5.83 0.072 (5.15

0.542 (5.78 0.024 (5.17

0.668 f 0.008 (1.09 f 0.05) -

0.493 f 0.014 (1.11 *0.07) _

0.612 *0.009 (0.27 f 0.03) _

0.649 f 0.011 (0.27 f 0.02)

1.54 * 0.00

f 0.002 kO.04) f 0.001 +cO.O3)

f f f f

0.003 0.03) 0.000 0.03)

182

The pH value of the solvent also affects partitioning of proteins by altering the electrical charge of the proteins (Hansen and Gorski, 1985). Under the conditions employed in the present study, the c-erb A protein was more enriched in the lower dextran-rich phase than in the upper polyethylene glycol-rich phase, giving a partition coefficient lower than 1.0. As shown in Table 1, the partition coefficient of the c-erb A protein without 3,5,3’-triiodo-t-thyronine was higher than when occupied by 3,5,3’-triiodo-t-thyronine, indicating that 3,5,3’-t~odo-L-thyro~ne binding caused changes in electrostatic properties of the c-erb A protein (Table 1). A similar alteration of the partition coefficient was observed using rat liver nuclear receptor. In contrast, thyroxine binding globulin showed an increase in the partition coefficient upon 3,5,3’-triiodo-L-thyronine binding. Since 3,5,3’-triiodo+thyronine partitioned preferentially into the upper phase, giving a partition coefficient higher than 1.2, changes in the partition coefficient of the c-erb A products or receptors after 3,5,3’-triiodo+thyronine binding cannot be explained by a simple additive effect of the 3,5,3’triiodo+thyronine molecule but may reflect a confo~ational change of these proteins. This suggests that the c-erb A proteins, as well as rat liver nuclear receptors, undergo a conformationdl transition upon 3,5,3’-triiodo+thyronine binding. Similar findings were obtained when the 3,5,3’typo-L-th~o~ne-occupied and 3,5,3’-triiodo-Lthyronine-free c-erb A products were analyzed by an aqueous two-phase partitioning study followed by polyacrylamide gel electrophoresis. The 3,5,3’t~odo-L-Sloane-o~upied c-erb A product (50 kDa protein) was less abundant in the upper phase in comparison to the 3,5,3’-triiodo-L-thyroninefree c-erb A product (Fig. 6). The change in the partition coefficient was not caused artificially by the presence of free 3,5,3’-t~iodo-L-thyro~ne (Fig. 6, lanes 11 and 12) but required binding of 3,5,3’triiodo-L-thyronine to the c-e& A product. The degree of the change in partition behavior correlated to the occupancy of the c-erb A products by 3,5,3’-t~~~L-thyro~ne (Fig. 6, lanes 3-8). A similar decrease in partition coefficient was observed using triiodothyroacetic acid (Fig. 6, lanes 9 and 10). In this experiment, 0.05 M NaCl was used since the difference in the partition coeffi-

1

2

3 4

5

6 7

8 9101112

Fig. 6. Aqueous two-phase partitioning study of the c-erb A product. Cell extract of E. coil containing the c-erb A products (95 nM of 3,5,3’-triiodo-L-thyronine (TX) binding capacity} was incubated at 22*C for 2 h in the absence (lanes 3, 4, 11 and 12) or presence of 30 nM Ts (lanes 5 and 6), 1 pM Ts (lanes 7 and V), or 1 pM 3,5,3’-triiodo-t_-thyroacetic acid (lanes 9 and 10). Aqueous two-phase partitioning study using 0.05 M NaCl was performed as described in Materials and Methods. In lanes 11 and 12, 1 I_IMTs was added just before vortexing for partitioning. After phase separation, 10% trichloroacetic acid (final concentration) was added to an ahquot of each phase. After incubation in ice for 30 rnin, protein was precipitated by centrifugation, and washed twice with 10% trichioroacetic acid. The pellet was washed 3 times with ether/ethanol, 1: 1 (v/v). Samples were dried, dissolved in 5% sodium dodecylsulfate/5% 2-mercaptoethanol/62.5 mM Tris-HCl pH 6.8/10% glycerol/ 0.004% bromophenol blue, kept in boiling water for 3 mitt and electrophoresed in a sodium dodecylsulfate-10% polyacrylarnide gel. The gel was stained with Coomassie brilliant btue R-250. Lanes 3, 5, 7, 9 and 11 are from 0.2 ml of the upper phases and lanes 4, 6, 8, 10 and 12 are from 0.02 ml of the lower phases. E. coli extract without c-erb A protein was used for the aqueous two-phase partitioning study and samples from the upper phase (lane 1) and Iower phase (lane 2) were analyzed. The arrow indicates the position of the 50 kDa c-erb A protein. Similar results were obtained from four other experiments.

cients of 50 kDa c-erb A products between 3,5,3’triiodo-L-thyronine-occupied and 3,5,3’-triiodo-Lthyronine-unoccupied states was almost 3-fold at this salt content, and the difference was clear when analyzed by electrophoresis. Discussion Proteins with molecular masses of 50 and 52 kDa were produced when E. co& transfo~ed

183

with pNTR were induced by 1 mM isopropyl-/?D-thiogalactopyranoside. Affinity labeling experiments revealed that the 50 kDa protein bound [‘251]3,5,3’-triiodo-r_-thyronine. Affinity labeling was inhibited partially by 1 nM 3,5,3’-triiodo-Lthyronine and completely by 1 PM 3,5,3’-triiodoL-thyronine, demonstrating that labeling reflected high affinity and limited capacity binding. The order of affinity of iodothyronine analogs in displacing the [‘251]3,5,3’-triiodo-L-thyronine label was characteristic for the thyroid hormone receptor (Samuels et al., 1979; Ichikawa and DeGroot, 1987) and was identical to what was obtained with reversible equilibrium binding. This indicates that the c-erb A proteins expressed in E, coli display hormone binding indistinguishable from classical thyroid hormone receptor. The reason the 52 kDa proteins were not covalently labeled with [‘251]3,5,3’-triiodo-L-thyronine was not clear. The 52 kDa protein may not be a hormone binding form of c-erb A products. Alternatively, the coupling efficiency of [‘251]3,5,3’-triiodo-L-thyronine and 52 kDa protein was lower than that of [‘251]3,5,3’-triiodo-L-thyronine and 50 kDa protein, and we could not detect it. Another possibility is that the 52 kDa protein did not undergo proper protein folding in E. coli, so that the tertiary structure of the product did not allow the hormone binding. Binding of the c-erb A proteins to calf thymus DNA-cellulose column was demonstrated (Fig. 5) and the c-erb A protein was eluted at 0.2-0.3 M KC1 or 10 mM pyridoxal 5’-phosphate from the column. These results are almost identical to those obtained using rat liver nuclear thyroid hormone receptors (Ichikawa and DeGroot, 1987). Although the DNA binding demonstrated here is not sequence specific, non-sequence-specific DNA binding of the receptor is controlled by histones or other nuclear proteins and is important in determining the localization of the receptors in chromatin (Ichikawa et al., 1987). Changes in hydrophobicity of the c-erb A protein upon thyroid hormone binding were shown in the present study. Since the changes in partition coefficient of the c-erb A protein were opposite to that expected from a simple additive effect of the 3,5,3’-triiodo-L-thyronine molecule, and to that observed for thyroxine binding globulin, our findings indicate that the c-erb A protein, like thyroid

hormone receptor, undergoes conformational transition upon thyroid hormone binding. An alternative possibility would be that the presumably hydrophobic pocket of the receptor is obscured by the hormone and that the appearance of charged residues from the amino and carboxy group of the triiodothyronine molecule resulted in decreased hydrophobicity. However, this is unlikely for the following reasons. (1) The carboxy group of 3,5,3’-triiodo+thyronine is important for the interaction with the receptor and is not considered to be exposed to the surface of the receptor (Apriletti et al., 1983). (2) Although the amino group of 3,5,3’-triiodo+thyronine may be exposed to the surface of the receptor (Apriletti et al., 1983) it also exposes the amino group on the surface (Penski and Marshall, 1969) and caused opposite changes in partition coefficient upon binding to thyroxin binding globulin. (3) Triiodothyroacetic acid, which lacks the amino group of triiodothyronine and shows potent thyromimetic properties (Samuels et al., 1979) also decreased the hydrophobicity of the c-erb A protein. A similar conformational alteration upon hormone binding was demonstrated for steroid hormone receptors using an aqueous two-phase partitioning study (Hansen and Gorski, 1985, 1986) and was considered to be related to the biological activation of the receptors by the hormone. Conformational changes in the receptor molecule may result in increased affinity of the receptors to the hormone-responsive element of the target gene. Alternatively changes in hydrostatic properties of the receptors may result in changes in interaction of the receptor with other proteins necessary for transcriptional activities. Involvement of proteinprotein interaction in the transcriptional activation has recently been suggested for the nuclear receptor family or other truns-acting factors (Theveny et al., 1987; Horikoshi et al., 1988). Together with the recent findings that the c-erb A products recognize specific DNA sequences in the 5’ flanking region of the rat growth hormone gene, which carry thyroid hormone responsiveness (Glass et al., 1987) our data provide evidence that the c-erb A products meet the requirements for being thyroid hormone receptor. Our plasmid, pNTR, when expressed in E. coli, produced a sufficient amount of the c-erb A protein, possibly the thyroid

184

hormone receptor. Recently, several proteins which regulate eukaryotic gene expression were produced using E. coli expression vectors (Kadonaga et al., 1987; Angel et al., 1988; Bohmann et al., 1988; Nagai et al., 1988). They were shown to be biologically active and useful in studying the mechanism by which eukaryotic gene expression is regulated. pNTR, an E. coli expression vector plasmid containing the human placenta c-erb A cDNA, will be useful in studying the structure and function of the thyroid hormone receptor. Acknowledgement We are grateful to Dr. T. Hashimoto (Professor of Biochemistry, Shinshu University) for support and encouragement. References Angel, P., Allegretto, E.A., Okino, ST., Hattori, K., Boyle, W.J., Hunter, T. and Karin, M. (1988) Nature 332,166-171. Apriletti, J.W., David-Inouye, Y., Baxter, J.D. and Eberhardt, N.L. (1983) in Molecular Basis of Thyroid Hormone Action (Oppenheimer, J.H. and Samuels, H.H., eds.), pp. 67-96, Academic Press, New York. Bohmann, D., Bos, T.J., Admon, A., Nishimura, T., Vogt, P.K. and Tjian, R. (1988) Science 238, 1386-1392. Cake, M.H., DiSorbo, D.M. and Litwack, G. (1978) J. Biol. Chem. 253, 4886-4891. Glass, C.K., France, R., Weinberger, C., Albert, V.R., Evans, R.M. and Rosenfeld, M.G. (1987) Nature 329, 738-741.

Hansen, J.C. and Gorski, J. (1985) Biochemistry 24,6078-6085. Hansen, J.C. and Gorski, J. (1986) J. Biol. Chem. 261, 1399013996. Horikoshi, M., Hai, T., Lin, Y.-S., Green, M.R. and Roeder, R.G. (1988) Cell 54, 1033-1042. Ichikawa, K. and DeGroot, L.J. (1987) Proc. Natl. Acad. Sci. U.S.A. 84, 3420-3424. Ichikawa, K., Bentley, S., Fee, M. and DeGroot, L.J. (1987) Endocrinology 121, 893-899. Kadonaga, J.T., C-utter, K.R., Masiarz, F.R. and Tjian, R. (1987) Cell 51,1079-1090. Maniatis, T., Fritsch, E. and Sambrook, J. (1982) Molecular Cloning, A Laboratory Manual. Cold Spring Harbor Laboratories, Cold Spring Harbor, NY. Nagai, K., Nakaseko, Y., Nasmyth, K. and Rhodes, D. (1988) Nature 332, 284-286. Oppenheimer, J.H. (1968) New Engl. J. Med. 278, 1153-1162. Penski, J. and Marshall, J.S. (1969) Arch. Biochem. Biophys. 135, 304-310. Samuels, H.H., Stanley, F. and Casanova, J. (1979) J. Clin. Invest. 63, 1229-1240. Sanger, F., Nicklen, S. and Coulson, A.R. (1977) Proc. Natl. Acad. Sci. U.S.A. 74, 5463-5467. Sap, J., Munoz, A., Damm, K., Goldberg, Y., Ghysdeal, J., Leutz, A., Beug, H. and Vennstrom, B. (1986) Nature 324, 635-640. Scatchard, G. (1949) Ann. N.Y. Acad. Sci. 51, 660-672. Theveny, B., Bailly, A., Rauch, C., Rauch, M., Delain, E. and Milgrom, E. (1987) Nature 329, 79-81. Torresani, J. and DeGroot, L.J. (1975) Endocrinology 96, 1201-1209. Vennstrom, B. and Bishop, J.M. (1982) Cell 28, 135-143. Weinberger, C., Thompson, C.C., Ong, E.S., Lebo, R., Gruel, D.J. and Evans, R.M. (1986) Nature 324, 641-646.