Solubilized Nuclear “Receptors” for Thyroid Hormones PHYSICAL CHARACTERISTICS AND BINDING PROPERTIES, EVIDENCE FOR MULTIPLE FORMS*

Tissues regulated by thyroid hormones contain chromatin-localized “receptors” that may be involved in the actions of these hormones. In this report, we describe some properties of these receptors after their solubilization from rat liver nuclei and their separation from nucleic acids and basic proteins. The nuclear extract and partially purified preparations contain a dominant class of binding which have a high affinity for triiodothyronine (3,5,3’-triiodo-L-thyronine, - 1 nM) and for the potent isopropyl diiodothyronine Kd -1 nM) and also thyroxine - 5 and reverse triiodothyronine (3,3’,5’- triiodo+thyronine, - 20 nM). This elutes on Sephadex G-100 in an included peak which has a Stokes radius of 35 A and sediments on glycerol gradients 3.5 S. these data a molecular weight ratio of 50,500 and a frictional ratio of I.4 were calculated, the is somewhat

Tissues regulated by thyroid hormones contain chromatin-localized "receptors" that may be involved in the actions of these hormones. In this report, we describe some properties of these receptors after their solubilization from rat liver nuclei and their separation from nucleic acids and basic proteins.
There was a sharp decline in triiodothyronine binding by this component above pH 8.7 (optimum around pH 7.6) where there is marked dissociation of the 4' phenolic hydroxyl of triiodothyronine (pK, -8.5). A similar decrease in thyroxine (pK, -6.7) binding with pH increases in this range was not observed. Thus, ionization of the phenolic hydroxyl may influence binding.
The solubilized preparations can also contain a minor specific-binding component that can be identified by binding analyses, and by G-100 or quaternary aminoethyl Sephadex chromatography. This component has a much lower affinity for triiodothyronine and isopropyl diiodothyronine than for thyroxine as compared to the major component. It probably has a pH optima around 6.0 and demonstrates an apparent tendency to aggregate. The minor component was not always identified by direct Scatchard analysis and may be generated in part from the major component as it was more commonly observed after storage or purification of the nuclear extract. Thus, at least two thyroid hormone-binding components can be present in extracts of purified rat liver nuclei; the minor component may be an altered form or subunit of the major component. The relative binding activities of triiodothyronine, isopropyl diiodothyronine, and thyroxine by the major component, similar to those in intact nuclei, parallel the biological potencies of these compounds, and suggest that the dominant binding is by biologically relevant receptors. Since ionization of the phenolic hydroxyl may influence binding, the lower activity of thyroxine relative to triiodothyronine may in part be due to the fact that at physiological pH, the phenolic hydroxyl of thyroxine is more dissociated than is that of triiodothyronine. The finding that this receptor is somewhat asymmetrical provides an indication of the shape of an intrinsic chromatin protein implicated in specific gene regulation.

Nuclear
Thyroid Hormone "Receptors" c&l were assayed for purity by radioimmunoassay using purified antibody (Calbiochem) specific for triiodothyronine and thyroxine." All triiodothyronine lots were found to contain less than 0.01% contaminating thyroxine. Thyroxine was used without further purification if it contained less than 0.5% cross-reactivity with the antibody to triiodothyronine.
However, the synthesis schemes for reverse triiodothyronine (45) and the 3'-isopropyl analog (46)  2 ml/g of tissue, to give a final temperature of about 4" after thawing.
The liver pieces were then drained using coarse nylon mesh, and added to ice cold Solution B (2.1 M sucrose, 6.5 mM MgCl,, 0.1 mM spermine), 3 ml/g of tissue. The mixture was then homogenized for three 1-min periods, separated by one minute of cooling using a Tekmar homogenizer (Cincinnati, Oh.) at maximum speed. This preparation, when stained with 1% crystal violet, was observed by light microscopy to contain cell debris, intact nuclei, and no whole cells. Nuclei were pelleted through 2.1 M sucrose (25,000 x g, 90 min) using an equal volume of homogenate and 2.1 M sucrose. The nuclear pellet was washed by gentle suspension in 25 volumes of Buffer C (20 mM Tricine, 2 mM CaCl,, 1 mN MgCl,, pH 7.6) containing 0.5% Triton X-100, and repelleted (500 x g, 10 min). The final nuclear pellet was resuspended in 10 volumes of cold Buffer D (20 mM Tris (pH 8.01, 0.25 M sucrose, 1 mM EDTA, 0.1 mM dithiothreitol, 5% glycerol), and saturated (4.1 M) ammonium sulfate was added to make 0.2 M. This suspension was then sonicated (model 185, Heat Systems-Ultrasonics, Inc., Plainview, N.J.) in 20ml aliquots using 85 watts average power in two 15-s periods separated by one minute of ice cooling. The sonicated mix was finally centrifuged (45,000 x g, 30 min) and the supernatant medium (designated nuclear extract) was stored in liquid nitrogen. This extract did not lose binding activity on freezing or after prolonged storage (up to 6 months) in liquid nitrogen. Endogenous triiodothyronine in the nuclear extract was less than 1 PM as measured by radioimmunoassay.6 Partial Purification -All steps were carried out at O-4". The nuclear extract was titrated to pH 9.2 by addition, with stirring, of 1.0 M Tris base and was then dialyzed against 100 volumes of Buffer E ' According to the manufacturer (Calbiochem), the antibody to triiodothyronine cross-reacts less than 0.001% with thyroxine. The thyroxine antibody preparations cross-react less than 0.05% with triiodothyronine.
Thus, the thyroxine antibody preparation can be used to detect less than 0.1% contamination and the triiodothyronine antibody can detect less than 0.01% contamination. ' School of Pharmacy, University of California, San Francisco, Calif. 94143.
fi Since binding assays routinely used 0.1 nM or higher concentrations of hormone, 1 PM endogenous cold triiodothyronine would be insignificant.
It was therefore considered unnecessary to thyroidectomize rats to lower endogenous hormone levels.
(0.1 M NaCl, 0.05 M Tris, 1.0 mhr dithiothreitol, pH 9.2) for 60 min. The dialyzed sample was centrifuged (25,000 x g for 30 min). The supernatant medium was loaded onto a QAE-Sephadex (strongly basic anion exchanger using diethyl-(2-hydroxypropyl)aminoethyl as the functional group) A-50 column (1.0 ml sample/l.0 ml bed volume) equilibrated in Buffer E, and eluted with one bed volume of Buffer E followed by two bed volumes of Buffer F (0.2 M (NH&SO,, 25 mM citric acid, 50 mM sodium phosphate, 1.0 mM dithiothreitol, pH 5.7). Protein concentration was measured by precipitating 0.1 ml of each fraction with an equal volume of 7% trichloroacetic acid in a standard assay (47). Aliquots (50 ~1) of each fraction were assayed for binding activity as described below, except that 1.0 nM hormone was used in the incubation mix. The fractions corresponding to peaks of hormone binding activity (see "Results") were separately pooled and precipitated by the addition of 0.3 g/ml of (NH&SO, (Schwarzl Mann, enzyme grade) and then centrifuged after 4 h (5000 x g for 10 min). The supernatant medium was discarded and the pellet, containing about 80% of the nuclear extract binding activity, was resuspended in 10 mM Tris (pH 8.51, 10 mM dithiothreitol, 0.6 M NaCl (1 ml for each 3 ml of starting nuclear extract) with negligible loss in binding activity.
The resuspended mixture was centrifuged (25,000 x g, 20 min), and the resulting supernatant medium containing all of the binding activity was stored in liquid nitrogen. Binding Assay -The binding assay mixture contained a limiting number of binding sites (10 to 200 PM in the added nuclear extract or QAE-purified preparation), 5 nM l'2"Iltriiodothyronine or l'2511thyroxine, and Buffer G (50 mM sodium phosphate (pH 7.61, 0.2 M ammonium sulfate, to make a total volume of 0.5 ml. To measure "nonspecific" binding (481, parallel tubes were prepared that were identical except that they also contained unlabeled triiodothyronine or thyroxine, respectively, at lOOO-fold excess. Incubations were for 60 min at 37" or 120 min at 25" (based on kinetic studies, see "Results" G-100 (Fig. 1) shows a minor peak which is excluded from the gel, and a major peak that is included and which elutes in a position corresponding to a Stokes radius of 35 A when compared to standard proteins (53). This finding is similar to those reported previously (21,341. The data in Fig. 1 (Fig. 3). As shown, the excluded peak is essentially abolished after the DNase treatment, whereas no specific DNase effect was observed on the included peak. Thus, even though heterogeneity of the excluded peak can be identified by the competitive experiment shown in Fig. 1, at least part of the peak from the crude extract may be due to DNA.protein complexes or aggregates. These data also suggest that the predominant triiodothyronine-binding activity is due to the component that is included on Sephadex G-100.
Partial Purification with Removal of Histones and DNA -Previous evidence suggested that the nuclear thyroid hormone receptor is a non-histone protein (17,22,34). As an initial approach to purification, we therefore, chose chromatography on QAE-Sephadex (a strongly basic anion exchanger, see "Materials and Methods") because of its potential for binding neutral and acidic proteins (and not histones) at high pH. In addition, the nuclear extract was titrated to pH 9.2 and dialyzed to low ionic strength prior to chromatography ("Materials and Methods"). The step results in a visible precipitation of basic proteins and nucleic acids, but there is no loss of binding activity from the supernatant medium. The binding activities eluted from QAE-Sephadex in two peaks (Fig. 4) 3. Agarose gel filtration and DNase treatment of nuclear extract.
The column dimensions and elution rate were as described in Fig. 1 (Fig. 4).
The thyroid hormone-binding activity in Pool A, labeled with either 1 nM ['251]triiodothyronine or ['""Ilthyroxine eluted only in the excluded volume from Sephadex G-100 (data not shown).
However, the triiodothyronine binding in Pool B also showed a major included peak on Sephadex G-100 which is identical in elution position to the included peak from nuclear extract (Fig. 5). Since Pool B does not contain detectable DNA (Table  I), it is likely that the minor excluded peak (Fig. 5) represents protein aggregates rather than nucleoprotein complexes.
Thus, even though DNA-protein associations may cause receptors to be excluded from the gel in the nuclear extract, other aggregations in the absence of DNA can also occur. Fig. 5 also shows the elution profile on Sephadex G-100 of Pool B equilibrated with ['*"Ilthyroxine. Of note is that the ratio of triiodothyronine/thyroxine bound by each peak at this concentration is quite different (excluded peak ratio: 1.6/l, included peak ratio: 11/l). These data indicate that more than one binding form may be present, even in Pool B. The binding activities in Pool B represent over 90% of the total triiodothyronine-binding activity eluted from the QAE-Sephadex column (Fig. 4). Since triiodothyronine has much greater hormonal activity in the rat than thyroxine (about lofold higher) and the major binding activity in Pool B shows a higher affinity for triiodothyronine than for thyroxine, the binders in this fraction are more likely candidates for further study as receptors.
Pool B also has several advantages for receptor characterization.
In contrast to nuclear extract, Pool B shows a linear hormone binding over a wide range of protein concentrations (Fig. 6). Also, this fraction is more stable during incubations at 37" (Fig. 7) than is the crude extract.
These results suggest  (60)) and thyroxine (pK, = 6.7 (60)) is also pH-dependent and could contribute to the pH profile if an intact hydroxyl is important for the hormone-receptor association (e.g. by hydrogen bonding (61)). In fact, there is a sharp decline in triiodothyronine binding around pH 8.5 to 9.0 which is in the region where increasing pH results in major dissociation of the 4'-hydroxyl. If an intact phenolic hydroxyl is important for the hormone's association with the receptor, then a dramatic effect on thyroxine binding would not be expected near pH 8.5 to 9.0 because the 4'-hydroxyl of thyroxine (pK,, = 6.7) becomes mostly dissociated at a lower pH. This is the case (Fig. 8). This hypothesis further predicts a sharp decline in thyroxine binding around the pK, for this compound. This is observed, however, data in this portion of the curve are more difficult to interpret since most of the binding is probably due to the minor component.
Hormone Binding -The binding of increasing concentrations of triiodothyronine and thyroxine by the nuclear extract, shown in Fig. 9 in the form of a Scatchard plot (62) O.&ml reaction mix containing 5 nM hormone. The sample was chilled to 4" and 0.5 ml was chromatographed as described in the legend to Fig. 1 and the amount of radioactivity was determined in each fraction. FIG. 6 (right). Linearity of the binding assay. Increasing amounts of nuclear extract (0)  Incubations were terminated by chilling (5 min in an ice bath) and were then assayed for specifically bound hormone as described under "Materials and Methods." by this criterion to a single class of sites with an approximately 5-fold higher affinity for triiodothyronine than for thyroxine.
However, in some preparations (data not shown) a minor component with a lower affinity for triiodothyronine was also observed.
The latter was more commonly observed in extracts which had been stored for a longer period of time. A Scatchard analysis of the Pool B-purified preparation also shows the presence of at least two binding components. The data (Fig. 10)   In the crude extract all of the binding of ['Yltriiodothyronine is more readily inhibited by nonradioactive triiodothyronine than by thyroxine (Fig.  11A). With [rz51]thyroxine most of the binding is more readily inhibited by triiodothyronine than by thyroxine; however, some of the ['Ylthyroxine binding is more readily inhibited by thyroxine than by triiodothyronine.
These data again suggest that there is a minor component which binds thyroxine more avidly than triiodothyronine.
As mentioned above, this lower affinity triiodothyronine binding component is not always apparent on a direct Scatchard plot of triiodothyronine binding (Fig. 9). This heterogeneity is also not apparent in the competition studies using ['Yltriiodothyronine ( Fig. 11 A, and C) since more than 90% of the binding in nuclear extract at the low concentration of ['Yltriiodothyronine used is by the major component (which has a higher affinity for triiodothyronine than for thyroxine). By contrast, a higher proportion of the binding of a low concentration of ['2511thyroxine by nuclear extract is due to the minor form of the receptor. The heterogeneity would therefore be more readily observed in the competition analysis using [Y]thyroxine.
A similar pattern of competition inhibition is observed using the Pool B-purified material. With ['Yltriiodothyronine, all of the binding is more readily inhibited by triiodothyronine than by thyroxine, whereas some heterogeneity is again suggested when YI]thyroxine is used. Of note is that with both the nuclear extract and the Pool Bpurified preparation, the biologically potent isopropyl diiodothyronine is as effective as triiodothyronine in the inhibition of the major binding component. This provides an additional criterion of specificity suggesting that the major binder is a biologically active receptor. By contrast, isopropyl diiodothyronine has a lower affinity for the minor component (Fig. 11, B and D) than either thyroxine or triiodothyronine.
A surprising finding is that reverse triiodothyronine does competitively inhibit all of the ['2"Iltriiodothyronine binding with an avidity that is about 25% (at 50% competition) of thyroxine and about 3% of triiodothyronine.
Since these studies indicated a competitive capacity of reverse triiodothyronine which is greater than previously reported (15, 19), we felt that it was important to consider the possibility that contaminants with other thyronine compounds that have a higher affinity for the receptor could result in artifactual competition data. The synthesis scheme used for the reverse triiodothyronine used here yields only the L-isomer and has no apparent chemical mechanism to form triiodothyronine or thyroxine (45). Thin layer chromatography studies provide a further indication of the reverse triiodothyronine purity since it migrated as a single spot and was resolved from 3,3'-diiodothyronine, triiodothyronine, and thyroxine (data not shown). Further, triiodothyronine radioimmunoassay demonstrated less than 0.01% triiodothyronine contamination. Therefore, it is unlikely that contaminants in the reverse triiodothyronine preparations are responsible for the competition. However, the possibility remained that a single outer ring deiodination during the binding reaction could form 3,3'-diiodothyronine and that this product could be responsible for the observed competition of triiodothyronine binding to the receptor. This possibility was tested by incubating nuclear extract under standard binding conditions with [1251]reverse triiodothyronine, extracting the reaction mixture with an equal volume of butanol:ethanol (3:1, v:v) and then examining the extracted products by thin layer chromatography ("Materials and Methods"). More than 96% of the extracted radioactivity migrated with authentic reverse triiodothyronine and less than 2% of the radioactivity migrated with authentic 3,3'-diiodothyronine. Therefore, there is little conversion under our reaction conditions. For the observed competition to be due to 3,3'diiodothyronine the affinity of this compound for the receptor would have to be 20-fold higher than that of reverse triiodothyronine. This is not the case since Jorgensen and BolgerR found (using an identical assay system) that the binding of 3,3'diiodothyronine is actually weaker than that of reverse triiodothyronine (61). This result has also been confirmed by us. It is also unlikely that other breakdown products of reverse triiodothyronine account for the observed competitive inhibition since further deiodination appears to reduce biologic ac-* These workers have also confirmed our findings on reverse triiodothyronine binding to nuclear extract.

Nuclear Thyroid
Hormone "Receptors" 7395 tivity and receptor binding (17,25). All of these considerations strongly suggest that reverse triiodothyronine is responsible for the observed competition and does bind to the receptors. Based on these data, we calculate an apparent equilibrium dissociation constant9 (K,) for reverse triiodothyronine under these conditions to be 30 nM.
Triiodothyronine Binding by Other Proteins -It was postulated that the nuclear receptors for thyroid hormones may also have intrinsic poly(A) polymerase activity (63). Because the thyroid hormone receptors represent only a very small proportion of the total protein in our preparations or those of other workers (17,22,24,34), any enzyme activity could be due to proteins other than the receptor. However, since several enzymes involved in nuclear function have been purified nearly to homogeneity from thyroid target tissues and were available to us, it seemed reasonable to ask if any of these have thyroid hormone-binding activity. We were unable to demonstrate triiodothyronine binding by purified DNA polymerase, terminal deoxynucleotidyltransferase, poly(A) polymerase, or RNA polymerase I. Therefore, it is unlikely that any of these enzymes, themselves, are or contain the thyroid hormone receptors. Of course, these data do not exclude the possibility that the receptors modulate the activity of these enzymes or have intrinsic activity which is similar to that of the enzymes tested.

DISCUSSION
In the present studies, characteristics of thyroid hormonebinding sites, solubilized from rat liver nuclei, are presented. We found that after some purification of the salt extracted receptor, a preparation could be obtained which is stable at 37", and which has other desirable characteristics such as lmearity of binding over a wide range of protein concentrations. The data with the partially purified preparations also indicate that the receptor binding of triiodothyronine and thyroxine does not require basic proteins (histones) or DNA since these components have been removed.
In the nuclear extract and partially purified preparations (Pool B) the dominant binding activity has characteristics suggestive that it is a thyroid hormone receptor. This component binds triiodothyronine with a higher affinity than thyroxine and also has a high affinity for the biologically potent isopropyl diiodothyronine.
The major binding component is included in Sephadex G-100 gels (Stokes radius 35 A) and migrates around 3.5 S in glycerol gradients. From these data a molecular weight ratio of 50,500 was estimated. The value is lower than that reported by Surks et al. (21,34) who used only the molecular sieve approach, a method that would tend to overestimate the size of the receptor if it is asymmetrical or contains a significant amount of bound water. In fact, the present studies suggest that the thyroid hormone receptor is somewhat asymmetrical since the estimated frictional ratio of 1.4 is greater than can be attributed to bound water alone.
It is of interest to compare the properties of the thyroid hormone receptors with those obtained for another class of regulating proteins that interact with chromatin, the steroid receptors. The latter differ from thyroid hormone receptors in that they are not ordinarily present in chromatin in the ab-9 The apparent equilibrium dissociation constant was calculated from the following equation derived from mass action considerations: K, = C,,K,/(K, + T), whereK, = apparent equilibrium dissociation constant of the competitor reverse triiodothyronine, K, = apparent dissociation constant for triiodothyronine, C,, = molarity of the competition that produws 50% competition, and T = concentration of [lZSIltriiodothyronine competed.
sence of the hormone. Further, even in the presence of the hormone, the steroid receptors are not as tightly associated with the chromatin as the thyroid hormone receptors (as measured by ease of extraction). The steroid receptors are also much more asymmetrical than are the thyroid hormone receptors; for instance, the progesterone receptors isolated from chick oviduct have frictional ratios of 1.7 and 1.9 (64, 65). It would appear that proteins which regulate gene expression at the level of chromatin may therefore vary considerably in their shapes; extreme asymmetry, as seen with the steroid receptors, is apparently not a general characteristic of proteins involved in genomic regulation. The present data also demonstrate that extracts of purified liver nuclei or partially purified receptor preparations contain thyroid hormone-binding components which differ from the major form of the receptor. One component binds the hormone nonspecifically and is not ordinarily reported because its contribution to the total binding is subtracted when specific binding is calculated. However, this nonspecific component can represent a significant part of the bound radioactivity which is excluded from Sephadex G-100 (Fig. 1). We also demonstrated a second component which binds triiodothyronine specifically but which differs from the principal form of the receptor in that it has a higher affinity for thyroxine than for triiodothyronine or isopropyl diiodothyronine.
This component was evident in Scatchard analyses (Fig. lo), competition studies (Fig.  ll), and was suggested by the pH studies (Fig. 8). An enrichment of this component can also be obtained by QAE-Sephadex chromatography (Fig. 4, Pool A) or by Sephadex G-100 (Fig. 5, excluded fraction). This component was not always observed by Scatchard analysis of the crude nuclear extract. Curiously, when this extract was stored on ice or was partially purified, the minor component could be clearly demonstrated by a subsequent Scatchard analysis. This apparent generation of the minor component following fractionation led us to a more extensive study in which strong evidence has been obtained that the major receptor component can undergo qualitative changes during storage or purification.iO Such "altered" receptors bind thyroxine more avidly than triiodothyronine and can account for the appearance of the minor component. The fact that the minor component can be generated also suggests that its presence is not due to contamination of the nuclear preparation by extranuclear binders.
An analysis of the effect of pH on binding indicates that careful attention must be given to this parameter when measuring binding activity; pH is especially critical when determining triiodothyronine-thyroxine cross-reactivity. For example, at pH values in the 7.4 to 6.2 range, this cross-reactivity reflected the biological potency of these two hormones, but at high pH, the total binding of thyroxine exceeded that of triiodothyronine. The finding of a dramatic decline in triiodothyronine binding around the pK, for its phenolic hydroxyl, which was not observed in this range for thyroxine (which has a lower pK,) also suggested that the integrity of the phenolic hydroxyl is important for binding. If this is true, then the higher potency of triiodothyronine (relative to thyroxine) may be attributed in part to the fact that its phenolic hydroxyl is less dissociated at physiologic pH; however, steric and other functions of the 5'-iodine may also contribute.
The competition data suggest that reverse triiodothyronine has a high affinity (K, -30 nM) for binding to the major form of the intranuclear receptor. This affinity is about 25% that of lo J. Ring, K. R. Latham, and J. D. Baxter, manuscript in preparation.