Stimulation by Thyroid Hormone Analogues of Red Blood Cell Ca2+-ATPase Activity in Vitro CORRELATIONS BETWEEN HORMONE STRUCTURE AND BIOLOGICAL ACTIVITY IN A HUMAN CELL SYSTEM*

Human red blood cell membrane Ca2’-ATPase activ- ity is stimulated in vitro by physiological concentrations ( 1 0 ” O M) of L-thyroxine (L-T,) and 3,5,3’-triiodo- L-thyronine (L-T3). This human cell system has been utilized to examine a series of iodothyronine and io- dotyrosine analogues for structure-activity relation- ships. Analogue purity was verified by high pressure liquid chromatography. Analogues were studied at a concentration of 1 0 ” O M and the stimulatory effect of each analogue was compared with that of L-TI in this system. Essential to Ca*’-ATPase stimulation were oc-cupation of the 3 and 5 phenyl positions by iodide, bromide, or methyl groups, the L-configuration of the alanine side chain, side chain length equal to that of alanine, and a perpendicular (skewed) conformation of the two rings. The 4’-hydroxyl group is not essential to Ca2’-ATPase stimulation in this model system. TS was 76% as active as T, in stimulating Ca2’-ATPase activity. The stimulatory effect of 3,5-dimethyl-3’- isopropyl-L-thyronine and 3,5,3’,5’-tetrabromo-~- thyronine approximated that of L-T4. Selected tyrosine absence of ATP were also included. Each experiment compared effects of up to ten analogues with concurrent effects of L-T, and L-T, on membranes from a single donor. Results were calculated as means -t- S.E. of three or more experiments conducted on membranes from two or three donors. The intra-assay coefficient of variation of measurements of Ca2*-ATPase activity was t3%, whereas interassay variation was 210% due to variability in basal enzyme activity introduced by freezing and thawing of membranes and differences in basal Ca2+-ATPase activity among cell donors. In each experiment, the absolute increase in Ca2+-ATPase activity (pmol of Pi/mg. 90 min) induced by L-T, was deter- mined and the effect of each analogue on enzyme activity was expressed as a per cent of activity stimulated by L-T,. Statistical Analysis-Significance of thyroid hormone and analogue stimulation of Ca”-ATPase activity was determined by paired t test. The enzyme activity of each sample incubated with hormone or analogue was compared with that of a simultaneous basal control sample that lacked hormone or analogue. The absolute increase in activity obtained with each analogue, if found to he significant ( p < 0.05, and at least 25% of the L-T, effect), was then compared with the L-T,-associated Ca2+-ATPase increase in the same experiments.

Uptake of a-amino-isobutyrate and cycloleucine by isolated rat thymocytes is stimulated by thyroid hormone and the effects of thyroid hormone analogues in this system have been studied (15). The uptake of 8-deoxyglucose in the same cell system is also responsive to thyroid hormone, at a concentration closer to physiologic than used in many prior studies (16,17). Recent studies of analogue binding to thymocyte membranes have shown direct correlation with the ability of the analogues to stimulate 2-deoxyglucose uptake (18). In the rat erythrocyte membrane, thyroid hormone can affect acetylcholinesterase (19) and Ca2+-stimulable, Mg2+-dependent adenosine triphosphatase (Ca"-ATPase) activity (20). The effect may be enzyme stimulation or inhibition, depending on the fatty acid composition of the membrane. Concentrations of hormone and analogues in these studies were between and lo-' M (20).
We have recently shown that the Ca2+-ATPase activity of human red blood cell membrane is maximally stimulated in vitro by physiologic concentrations of thyroid hormone M ) (21). Stimulation of Ca2+-ATPase by thyroid hormone is dependent upon hormone-binding to the red cell membrane (22) and upon the presence of calmodulin, a cytoplasmic activator protein for Ca2+-ATPase (23). This red cell Ca2+-ATPase assay system is a human cell model of extranuclear thyroid hormone action, the use of which in preliminary studies has indicated that D-amino and acetic acid analogues of iodothyronines do not stimulate Ca2+-ATPase activity (21). We have now used this human cell Ca2+-ATPase assay to investigate a wide variety of iodothyronine and iodotyrosine analogues and have characterized the importance of substituent groups to the stimulation of human red cell Ca2+-ATPase activity.
The majority of hormone analogues were dissolved in 4% propylene glycol, 0.4 N KOH prior to use. T, amine was initially placed in propylene glycol/KOH with sufficient 95% ethanol added to dissolve the analogue completely; n-acetyl DIMIT was dissolved initially in 95% ethanol. Subsequent dilutions of analogues were made in 10 mM Tris buffer, pH 7.45. Solutions of T, and T, were made with the same diluent as each analogue for use in concurrent assay samples.
Human Red Blood Cell Membranes-Red cell membranes were prepared hypotonically, as previously reported (21) from heparinized venous blood samples obtained from healthy euthyroid human volunteers. The research protocol was approved by the institutional Human Research Committee. The cells were pelleted by centrifugation and lysed in 10 volumes of hypotonic Tris buffer (10 mM) in the presence of 0.1 mM EDTA; the membranes were then washed twice in isotonic NaCl, twice in 10 mM Tris buffer, and suspended in 10 mM Tris buffer to a protein concentration of 1 mg/ml, as measured by the method of Lowry et al. (24), using BSA as the standard. The membranes are washed free of endogenous thyroid hormone during this preparation (21). Membranes were stored in 10 mM Tris, pH 7.4, a t -70 "C for 1-5 days until used.
Incubation of Membranes Prior to CaZ+-ATPase Assay-Membranes (1 mg of protein/ml of buffer) were incubated a t 37 "C for 60 min prior to ATPase assay, together with L-T,, L-T,, hormone analogue, or control diluent.
Ca'+-ATPase Actiuity-Membrane enzyme activity was assayed by our previously described method (21,23) in which hydrolysis of 1 mM Nan-ATP at 37 "C is quantified in the presence and absence of 0.15 mM Ca2+ by the liberation of P, (25). Results are reported as pmol of Pi/mg of membrane protein. 90-min assay period. Also present in the assay medium were 0.1 mM EGTA, 75 mM NaC1, 25 mM KCl, 1 mM MgCl,, and 25 mM Tris, pH 7.4. As determined by an ion-specific electrode (Orion 93-20, Orion Research, Inc., Cambridge, MA), the free Ca2+ in the assay system was 20 p~. The concentrations of Ca2+ and ATP have been found by us to optimize enzyme activity in both the presence and absence of thyroid hormone.* All enzyme measurements were made in duplicate in each experiment with simultaneous controls containing hormone diluent but no hormone or analogue. Controls for spontaneous hydrolysis of ATP and for liberation of endogenous phosphate from membranes in the absence of ATP were also included. Each experiment compared effects of up to ten analogues with concurrent effects of L-T, and L-T, on membranes from a single donor. Results were calculated as means -t-S.E. of three or more experiments conducted on membranes from two or three donors. The intra-assay coefficient of variation of measurements of Ca2*-ATPase activity was t 3 % , whereas interassay variation was 210% due to variability in basal enzyme activity introduced by freezing and thawing of membranes and differences in basal Ca2+-ATPase activity among cell donors. In each experiment, the absolute increase in Ca2+-ATPase activity (pmol of Pi/mg. 90 min) induced by L-T, was determined and the effect of each analogue on enzyme activity was expressed as a per cent of activity stimulated by L-T,.
Statistical Analysis-Significance of thyroid hormone and analogue stimulation of Ca"-ATPase activity was determined by paired t test. The enzyme activity of each sample incubated with hormone or analogue was compared with that of a simultaneous basal control sample that lacked hormone or analogue. The absolute increase in activity obtained with each analogue, if found to he significant ( p < 0.05, and at least 25% of the L-T, effect), was then compared with the L-T,-associated Ca2+-ATPase increase in the same experiments.
F. B. Davis and P. J. Davis, unpublished observations. this increase was 76% of that seen with T4, as indicated in Table I. Effect of Ring Iodination and Substitution with Bromide, Methyl, and Isopropyl Groups-The effects of various iodothyronines and ring-substituted analogues on Ca2+-ATPase activity are shown in Table I. L-T2, with iodination in the 3 and 5 positions on the inner ring, retained 77% of the effect of Tq in our assay system. In contrast, 3',5'-L-T2 had only 28% of the T, effect, and 3,3'-L-T2 and 3'-monoiodo-~thyronine were inactive. While L-T3 was 76% as active as T4, rT3 was inactive.
Modification of Alanyl Side Chain: Acetic and Propionic Acid Analogues, Decarboxylated Analogues, and Ethyl Esterification-The deaminated analogues, tetrac and triac, with one less carbon in the side chain than found in L-T, and L-T3, were ineffective in this Ca2+-ATPase system (Table 11).
The decarboxylated analogue of L-T3, T3amine, was 54% as effective as L-T4, and 71% as effective as L-T, in stimulating enzyme activity (Table 11). Ethyl esterification of L-T4 resulted in loss of enzyme stimulating action.
Modification of the Diphenyl Ether Linkage, D-configuration-Modification of T:? by substitution of an -Sbridge in place of the ether group resulted in complete loss of Ca'+-ATPase stimulating activity, as indicated in Table 111. D-T, and D-T, were ineffective in stimulating Ca2'-ATPase activity.
Effect of Tyrosine Analogues-The effects of several ringsubstituted tyrosine analogues on Ca2+-ATPase activity are

Effect of tyrosine analogues on red cell membrane &'+-ATPase activity
Determination of enzyme activity, analogue effect, and significance is outlined in the legend to Table I Table IV. Iodinated, brominated, nitro-, and methylsubstituted analogues all possessed enzyme stimulatory activity, while tyrosine itself, at 10"' M concentration, acted as an inhibitor of Ca*+-ATPase activity. While the decarboxylated analogue, 3,5-diiodotyramine, stimulated Ca2'-ATPase activity, the deaminated analogue, 3,5-diiodo-4-hydroxyhenylpropionic acid was nonstimulatory. The noniodinated parent compounds of these analogues were also inactive in this enzyme system.

DISCUSSION
The human erythrocyte membrane Ca2+-ATPase assay used in the present study is an in vitro model of extranuclear thyroid hormone action and affords the opportunity to define structure-activity relationships of thyroid hormone analogues in a human cell system. The enzyme assay is based on the method of Strittmatter et al. (26) which employs Tris buffer and generates enzyme activity levels which are lower than those reported by other methods (27)(28)(29). We have recently explored the effects of buffers on hormone-sensitive Ca2+-ATPase (23) and have concluded that the imidazole buffer shown by Farrance and Vincenzi (30) and others (27) to yield high levels of enzyme activity inhibits thyroid hormone action on red cell membranes in a concentration-dependent manner. We have established elsewhere that stimulation of Ca2+-ATPase by T4 is optimal at a hormone concentration of 10"O M in both human erythrocyte (21,23) and rabbit sarcolemmal (31) membranes. The observation that T4 is more effective than T, in stimulating Ca2+-ATPase activity in the human red cell is consistent with our previous studies and is a departure from the generalization that 3,5,3'-trisubstituted thyronines are more active than the related 3,5,3',5'-tetrasubstituted compounds (32). On the other hand, our red cell membrane preparation is incapable of converting T4 to T, (21) and the sensitivity of the preparation to T, may, therefore, not be surprising.

Thyroid Hormone and Ca2'-ATPase
Based on our analysis of Ca2+-ATPase stimulation in the human red blood cell, we show that, at a physiologic concentration, thyroid hormone action requires a number of structural conditions. First, the inner ring 3 and 5 positions must be occupied by functional groups of a size comparable to that of iodine, if not iodine itself. While 3,5-T2 was 77% as effective as T4, 3',5'-L-T2 was minimally effective, and 3,3'-L-T2 and 3'-monoiodo-~-thyronine not at all stimulatory. Reverse T3, lacking iodine in the 5-position, was inactive. The tetrabromo derivative of T4 was equal to T, in Ca2+-ATPase stimulation, and 3,5-dibromo-~-thyronine was 67% as effective as its iodinated counterpart, 3,5-T2. DIMIT, with 3,5-methyl substitutions and a 3'-isopropyl group, was fully as active as T3 in this red cell system. DIMIT has received much attention as an in vivo thyromimetic analogue (3,33,34), and its activity in our in vitro system correlates well with its known in vivo activities. Acetylation of the amino group of DIMIT results in loss of Ca2+-ATPase effect.
Second, it is necessary for the side chain to be of a critical length, a t least propionyl, for some enzyme stimulating activity to be retained in deaminated hormone analogues. Both tetrac and triac, which contain one less side chain carbon than T, and Tar are inactive in the red cell Ca"-ATPase system. Addition of another carbon to the side chain of tetrac, forming tetraiodothyropropionic acid, results in enzyme stimulating ability which is 50% of the effect of T4. The decarboxylation of TS, to form T3amine, leads to 50% retention of enzyme stimulating effect as compared with T, and 71% as compared with T3. Modification of the carboxyl group by esterification to T4-ethyl ester resulted in complete loss of Ca"-ATPase stimulation by the analogue.
Third, the perpendicular (skewed) conformation of two rings, conferred by the diorthosubstituted phenyl ether linkage, is necessary for maintenance of thyroid hormone stimulating effect. This is suggested by the observation that those analogues which are not sterically required to be skewed (e.g. rT3, 3,3'-T2, 3'-T1) have no stimulatory effect. Fourth, the Lconformation is also necessary for hormone action; both D-T4 and D-Ts were ineffective. Fifth, elimination of the 4'hydroxyl group still permits stimulation of Ca2+-ATPase activity.
Last, since tyrosine analogues with 3-or 3,5-substitutions by iodide, bromide, methyl, or nitro-groups retained the ability to stimulate red cell membrane Ca*+-ATPase, it is apparent that two rings are not necessary for this hormone effect. A tyramine derivative, lacking the side chain carboxyl group, retained stimulatory activity with iodide ions present in the 3,5-positions. With removal of the amino group, however, stimulatory activity was lost, even in the presence of 3-and 5-position iodination. Thus, the primary prerequisite for this stimulatory effect is recognition of a diortho-substituted ty-rosy1 moiety, followed by the stereospecifically oriented substituted diphenyl ether ring system. This is also the first observation of a hormonogenic response from selected single ring thyroid hormone analogues. Single ring analogues do not bind to the serum transport proteins, nor to the nuclear receptor.
Our findings with regard to the effect of thyroid hormone analogues are generally in agreement with in vitro and i n vivo comparisons of biological potency carried out in animal systems, as summarized extensively by Jorgensen (13,32) and Cody (14), except that in our assay system T4 is more active than Tn, and selected tyrosine analogues are stimulatory. Stimulation of rat thymocyte 2-deoxy-D-glucose uptake by Lisomers of rT3, 3,5-T2, and thyronine, as studied by Segal and Ingbar, has been demonstrated a t analogue concentrations of . ...~.~ ~~ ~~~~ . to M (16) and is calcium-dependent (17). These authors have recently compared the effect of analogues on [Iz5I]T3 binding to rat thymocytes and have shown that rT3, 3,5-T2, and thyronine had relative affinities for T3-binding sites on thymocyte membranes which correlated well with their relative stimulatory effect on 2-deoxy-~-glucose uptake (18). These findings contrast with our previous demonstration that tetrac, which binds to red cell membranes, displacing T4, also inhibits T4 stimulation of Ca2+-ATPase activity; tetrac has no intrinsic Ca"-ATPase stimulating effect (22). Our previous binding studies support the thesis that thyroid hormone stimulation of Ca2+-ATPase in red cell membranes requires hormone binding to the membranes.
Studies of binding of thyroid hormone analogues to nuclei in a variety of animal cells (6-10) have shown that D-analogues as well as tetrac and triac are bound avidly despite the trivial metabolic activity of these compounds in intact animals (13). Rapid i n vivo metabolism and excretion of D-analogues, as compared with L-isomers (13), and a relative decrease in plasma membrane transport of D-analOgueS (35) have been postulated to explain the occasional discrepancies between nuclear-binding and biological potency observed with the Disomers. The avidity of the acetic acid analogues for nuclear and cell membranes is probably due to their lipophilic nature. Nuclear binding studies have also shown that the 4'-hydroxyl group is critical, but the ether bridge not critical for binding (7). These findings are in contrast to ours, and probably reflect the existence of a variety of hormone receptor sites in different subcellular fractions, which recognize different portions of the analogue molecule.
Gal0 et al. (20) have examined the effect of thyroid hormone analogues on erythrocyte Ca2+-ATPase in the rat. In their study, T3 was 100-fold as effective as T4 in stimulating Ca2+-ATPase activity, but the hormones were stimulatory only in rats fed a diet containing lard as the principal fat source; with an increase in dietary saturated fatty acids leading to membrane lipid alterations, thyroid hormones became inhibitory to Ca2+-ATPase activity. In these rat erythrocyte studies, Disomers of T3 and T4 as well as tetrac, triac, TP, and monoiodothyronine were inactive in stimulating Ca2+-ATPase activity in animals fed the lard-containing diet.
The extrapolation to man of structure-activity relationships described in animal models has some hazard. We have already reported that there are some important interspecies differences in susceptibility of red cell Ca2+-ATPase activity to iodothyronine (36). It is also clear that thyroid hormone action on Na,K-dependent adenosine triphosphatase (Na,K-ATPase), a nucleus-dependent hormone effect, appears to be different in man and rat (37,38).
Whether the Ca2'-ATPase effect of thyroid hormone in the human red cell membrane applies to human cells of nonhematopoietic origin is not yet clear. We have found that the rabbit and human erythrocyte Ca2+-ATPases are similar in terms of hormone responsiveness (36). We have recently determined that the sarcolemmal Ca2+-ATPase of rabbit myocardium is readily stimulated i n vitro by 10"' M L-T4 and L-T3, but not stimulated by D-T4, tetrac, and triac (31). In the rabbit heart system, L-T, and L-T3 are equipotent. In animal systems, the Ca2+-ATPase in the red cell appears to be representative of plasma membrane Ca2+-ATPase in a variety of tissues (39).
We have shown that stimulation of human red cell Ca2+-ATPase activity by physiologic concentrations of T4 and TS is accompanied by enhanced calcium efflux in vitro in the intact erythrocyte (40,41), thus providing a functional correlate of the Ca2+-ATPase studies in membrane vesicles. We Thus, we believe that the effect of iodothyronines on Ca2+-ATPase activity in the human erythrocyte membrane is an index of biologic activity of thyroid hormone in the intact cell.