Regulation of Type II Iodothyronine 5’-Deiodinase by Thyroid Hormone lNHlBlTlON OF ACTIN POLYMERIZATION BLOCKS ENZYME INACTIVATION IN CAMP-STIMULATED GLIAL CELLS*

Jack L. Leonard#$jll, Catherine A. Siegrist-Kaiser811 , and Claudia J. ZuckermanS From the *Department of Physiology and Biophysics, Harvard Medical School, Boston, Massachusetts 02115, the SDepartments of Physiology and Nuclear Medicine, University of Massachusetts Medical School, Worcester, Massachusetts 01655, and the 1) Thyroid Research Unit, Endocrine Division, Department of Medicine, University of Geneva, Geneva, Switzerland

The cellular events mediating the rapid, thyroid hormone-dependent modulation of membrane-bound, type II iodothyronine 5'-deiodinase were studied in dibutyryl cyclic AMP(b&cAMP)-treated brain astrocytes. Unstimulated cells had undetectable type II B'-deiodinating activity.
Treating the cells with b&CAMP and hydrocortisone induced enzyme expression by increasing transcription of the enzyme or an essential enzyme related protein(s), with steady-state levels of type II 5'-deiodinase attained after 8 h of btscAMP treatment. Glial cells grown in the absence of thyroid hormone had lo-15-fold higher levels of 5'-deiodinating activity than cells grown in the presence of serum. The increased type II 5'-deiodinating activity observed in serum-free cultures was due to a prolonged enzyme half-life with no change in the rate of enzyme synthesis. Addition of thyroxine or 3,3',5'-triiodothyronine to the serum-free culture medium resulted in a concentration-dependent fall in steady-state enzyme levels, with EC& values of -0.4 nM. 3,5,3'-Triiodothyronine was at least loo-fold less effective. Chloroquine, NH&l, tunicamycin, colchicine, and monodansylcadavarine had no effect on the tH of the enzyme, while both carbonyl cynanide m-chlorophenylhydrazone and cytochalasins completely blocked the inactivation of the type II 5'-deiodinase.
These data indicate that in glial cells, an intact actin-cytoskeleton is required for thyroid hormone to modulate the energy-dependent regulation of the half-life of the short-lived, membrane-bound enzyme, type II 5'-deiodinase.
Thyroid hormone metabolism plays a fundamental role in determining the intracellular levels of bio-active 3,5,3'-triiodothyronine (T3)l in the brain. Recent studies have shown that >90% of the TB bound to brain cell nuclei is derived from intracellular thyroxine (T4) to T3 conversion, rather than from circulating TB (l-3). The key enzyme in this pathway is type 11 iodothyronine 5'-deiodinase (5'D-11) (for review see Ref. 1). Type 11 5'-deiodinase is a short-lived, integral membrane protein associated with neurolemmal membranes (4). lnterestingly, cerebrocortical enzyme activity fluctuates with changes in thyroid status, increasing 5-lo-fold 1 day after thyroidectomy and decreasing -90% within 30 min of hormone replacement (5,6). These dynamic, adaptive changes in 5'-deiodination serve to maintain near-normal intracellular Ts levels in the brains of neonatally hypothyroid rats given only 1/10 the daily replacement dose of T4 (7), indicating that the brain has regulatory pathway(s) that monitor and defend intracerebral T3 levels.
Recent studies in rats have shown that the thyroid hormone-dependent changes in type 11 5'-deiodinase are mediated by regulation of enzyme inactivation rather than enzyme synthesis (8,9). The demonstration that enzyme levels are more sensitive to T4 than to T3 (7, lo), together with the observation that this action of thyroid hormone is not blocked by inhibitors of transcription or translation (8), suggests a novel, extranuclear site of action for thyroid hormone in the brain.
The cellular mechanism for this thyroid hormone response has been difficult to study in intact animals, and cell culture models expressing thyroid hormone metabolizing enzymes such as dispersed fetal (11) or neonatal rat brain (12,13), and the neuroblastoma cell line NB41A3 (14), suffer from cell heterogeneity and/or low 5'-deiodinase levels. Our demonstration that CAMP induces abundant type 11 5'-deiodinase in cultured astrocytes (15) has been exploited to examine the T1-dependent regulation of this short-lived, membrane-bound enzyme. The data show that Tq and rT3 rapidly modulate the rate of enzyme inactivation in these cells, whereas TS is a much less effective hormone. In addition, inactivation and/or degradation of this membrane-bound enzyme was found to be energy dependent and to depend on the structural integrity of the actin-cytoskeleton. MATERIALS AND T4 Modulation of the tlh of Type II 5'-Deiodin~e (16) and purified as previously described (17) (22). Cells were then scraped from the dish and the Triton-insoluble cytoskeleton was collected by centrifugation at 12,000 x g for 10 min at 4 "C. Cytoskeleton proteins were dissolved in polyacrylamide gel electrophoresis sample buffer composed of 50 mM Tris buffer, pH 6.8, 1% (w/v) sodium dodecyl sulfate, 140 mM /3-mercaptoethanol,

RESULTS
Cell monolayers consisted of flat polygonal cells interspersed with occasional clusters of arborized cells. Greater than 95% of the cells contained glial fibrillary acidic protein (data not shown), an intermediate filament characteristic of astrocytes (26). Glial cells, under our culture conditions, showed a doubling time of 25 + 2 h that was unaffected by 1 mM bt,cAMP for up to 48 h.
In agreement with earlier work (15), bt,cAMP treatment resulted in the time-dependent appearance of type II 5'deiodinase activity that plateaued after 8 h of stimulation ( Fig. 1). Bt,cAMP was required to maintain steady-state levels of type II 5'-deiodinase, since removal of the cyclic nucleotide resulted in the progressive loss of the enzyme after a 60-min lag period. In subsequent studies, bt*cAMP was present  Fig. 2. Cells grown in hypothyroid medium showed IO-15-fold higher steady-state levels of 5'-deiodinating activity than cells grown in euthyroid medium. Addition of increasing amounts of either T4 or rTa resulted in the concentration-dependent fall in steady-state enzyme levels with an EDbO of 0.4 nM or -60 pM "free" hormone (determined by equilibrium dialysis, Fig. 2). T3 was at least IO&fold less effective in modulating steady-state enzyme levels.

Effects of Individual
Growth Factors on Bt,cAMP Induction of Type II 5'-Deiodinase-To examine whether one or more of the components in the chemically defined medium contributed to the increased expression of type II 5'-deiodinase, components were individually examined for their ability to influence 5'-deiodinating activity. In the absence of bt,cAMP, none of the compounds tested induced type II 5'-deiodinase (data not shown). Data in Table I show that fibroblast growth factor, putrescine, and prostaglandin FPa also had no effect on bt*cAMP-induced enzyme activity. In contrast, hydrocortisone increased 5'-deiodinating activity -2-fold, whereas insulin depressed enzyme levels -40%.
Hydrocortisone stimulation of type II 5'-deiodinase was concentration dependent (Fig. 3A) with an EDso of 4 nM. As shown in Fig. 3B, the fractional disappearance of 5'-deiodinating activity in cycloheximide-blocked, btzcAMP-stimulated cells was unchanged by hydrocortisone, suggesting that the glucocorticoid-dependent increase in type II 5'-deiodinase levels was most likely due to increased enzyme production.
The effect of actinomycin D on enzyme induction by bt,cAMP is shown in Fig. 4. At the start of the experiment, 1 mM bt,cAMP was added to the medium of all cells followed, at 30 min intervals, by addition of 10 pM actinomycin D. Actinomycin D-blocked cells were incubated for an additional 60 min and 5'-deiodinating activity determined.   Table II is the effect of actinomycin D on hydrocortisone amplification of type II 5'-deiodinase in stimulated glial cells. In bt2cAMP-stimulated cells expressing steadystate levels of type II 5'-deiodinase, 40 nM hydrocortisone induced further accumulation of type II 5'-deiodinase, and actinomycin D abolished the glucocorticoid-induced increase in enzyme.

Effects of Thyroid
Status on the Turnover of Type II 5'-Deiodinase-The influence of thyroid hormone on the tH of type II 5'-deiodinase is shown in Fig. 5. Cycloheximidedependent inhibition of protein synthesis in cells grown in euthyroid medium resulted in the exponential disappearance of type II 5'-deiodinase with a tlh of 20 min, while the enzyme in cells grown in hypothyroid medium showed a tlh of 300 min. Enzyme production rates, calculated from steady-state enzyme levels and the disappearance rate constant (k), were essentially the same in euthyroid and hypothyroid media indicating that thyroid hormone had little or no effect on enzyme synthesis (Table III). The data in Table III also show a modest reduction in b&CAMP-inducible type II 5'-deiodinase in cells after prolonged culture (i.e. passage 6). This decrement in enzyme activity was routinely observed and was progressive so that after passage 8, inducible type II 5'deiodinating activity was only 20% of that observed at passage 2 (data not shown). For this reason, experiments were done on cells between the 2nd and 6th passage.

Effects of Antimetabolites
and Cytoskeletal Inhibitors on the Turnover of Type II 5'-Deiodinase-The clearance of many integral membrane proteins is often initiated by internalization, followed by delivery to lysosomes. Since the degradation pathway for type II 5'-deiodinase was unknown, we examined the energy dependence, cell structural requirements, and role of lysosomes in the turnover of this short-lived enzyme. Bt*cAMP-stimulated cells, at steady state with respect to type II 5'-deiodinase, were exposed to selected inhibitors in the presence or absence of cycloheximide and incubated for 30 min at 37 "C. As shown in Table IV, cycloheximide reduced enzyme activity by 78% at 30 min and 91% at 60 min. ATP depletion by addition of 20 PM carbonyl cyanide m-chlorophenylhydrazone, resulted in a 45% decrease in type II 5'deiodinase, and cycloheximide did not further decrease 5'deiodinating activity in carbonyl cyanide m-chlorophenylhydrazone-treated cells. Lysosomotrophic agents (chloroquine and NH&l), inhibitors of endocytosis (monodansylcadaverine) or glycosidation (tunicamycin), and microtubule disruption (colchicine) had little or no effect on enzyme levels either in the absence or presence of cycloheximide. On the other hand, colchicine rapidly reversed the bt*cAMP-stimulated contraction of the glial cell borders, so that within 30 min the morphology of the bt*cAMP-treated cell was indistinguishable from that of unstimulated controls (data not shown). In contrast, cytochalasin B and the more selective microfilament inhibitor, dihydrocytochalasin B, preserved -75% of the type II 5'-deiodinase in the cycloheximide-blocked cell. Shown in Fig. 6 is the effect of 10 pM cytochalasin B on the tlh of type II 5'-deiodinase.
In the presence of cytochalasin B and cycloheximide, type II 5'-deiodinase activity progressively disappeared for the first 10 min. Thereafter, enzyme activity remained nearly constant in cytochalasin B-treated cultures, while 5'-deiodinating activity continued to decrease exponentially in cycloheximide-blocked cells. Pretreatment with cytochalasin B for 15 min prior to cycloheximide treatment abolished the rapid first phase of loss of type II 5'deiodinase suggesting that this initial phase occurred during a period when the actin polymerization was not completely inactivated (data not shown). Analysis of the effects of cyto- chalasin B on the actin-cytoskeleton showed that filamentous actin in cytochalasin B-treated cells decreased by 25% after 5 min and 43% after 15 min, as judged by densitometric analysis of the 43-kDa actin band present in the tritoninsoluble cytoskeleton after sodium dodecyl sulfate-polyacrylamide gel electrophoresis.
These data suggest that -15 min of cytochalasin treatment is required to inhibit actin polymerization sufficiently to decrease the cellular filamentous actin content to levels that impair the inactivation of type II iodothyronine 5'-deiodinase. In the absence of cycloheximide, cytochalasins promoted the further accumulation of type II 5'-deiodinase at a rate (1.7-2.4%/min) 60-80% of the estimated enzyme production rate (3%/min; (the product of type II iodothyronine 5'-deiodinase activity at steady-state (100%) x k)) ( Table IV) indicating that depolymerization of the actin-cytoskeleton selectively blocked the loss of type II 5'-deiodinase without altering the synthetic pathway for the integral membrane enzyme.  glial celki Cells were grown in euthyroid (DMEM/lO% fetal bovine serum (FBS) or hypothyroid (chemically defined, CD-serum) medium for 5 days and treated with 1 mM bt*cAMP for 16 h to induce steady-state levels of type II 5'-deiodinase. Enzyme activity was determined in cell sonicates as described under "Materials and Methods" and reported as units, where 1 unit equals 1 fmol I-released.min-'.mg protein-'. Fractional disappearance of the enzyme was determined as described in legend to Fig. 5. Production rate = steady-state enzyme activity (units) X fractional turnover (h-l). Passage

DISCUSSION
In this study, we have shown that the btpcAMP-stimulated glial cell is an excellent model system for the characterization of the molecular events regulating the turnover of type II 5'deiodinase in the central nervous system. Steady-state levels of type II 5'-deiodinase, 50-100 times those observed in the cerebral cortex of hypothyroid rats (5,7), are readily achieved within 8 h of stimulation with b&AMP (15,27) and are maintained for up to 24 h in the presence of the cyclic nucleotide. Cellular differentiation does not appear to mediate the induction of type II 5'-deiodinase, since bt,cAMP did not slow the growth rate of the cultured astrocytes and previous work has shown that butyrate, a differentiating agent (28), did not induce type II 5'-deiodinase (15,27).
The appearance of type II 5'-deiodinating activity stimulated by cyclic nucleotides was preceded by a transcriptional event(s) and synthesis of new mRNA encoding either the enzyme or an essential enzyme regulatory protein quickly reached non-rate limiting levels. Interestingly, enzyme induc- tion was amplified by hydrocortisone by a mechanism requiring increasing transcription. However, the glucocorticoid effect was only seen in the presence of bt,cAMP indicating an absolute requirement for cyclic nucleotide for the expression of type II 5'-deiodinase in the cultured astrocyte. Thyroid hormone had a marked influence on glial cell type II 5'-deiodinase with IO-20-fold increases in activity achieved in cells grown in the absence of thyroid hormone, similar to T4 Modulation of the tlA of Type II 5'-Deiodinase that found in rat brain (5)(6)(7)(8)29). Comparison of the enzyme production rates in cells grown in euthyroid and hypothyroid media revealed that thyroid hormone had no effect of synthesis; the increased 5'-deiodinating activity observed in hypothyroid cultures was entirely accounted for by diminished enzyme degradation/inactivation. These data are in agreement with earlier work that showed that the tl/, for cerebrocortical type II 5'-deiodinase was prolonged lo-15-fold in the hypothyroid rat (8) and further that the acute effects of thyroid hormone on the enzyme tlh did not require continued transcription or translation (8). Thus, CAMP-stimulated glial cell faithfully mimics the intact rat brain with respect to thyroid hormone-dependent regulation of this key membranebound enzyme.
The influence of individual iodothyronines on steady-state enzyme levels in the stimulated astrocyte was identical to those previously determined in uiuo. T, and rT3, a metabolically inactive iodothyronine, were -loo-fold more potent than T3 in modulating enzyme levels, as had been reported by Silva and Leonard (7) and Kaiser et al. (10) in viuo, and by St. Germain (29) in NB41A3 cells, and half-maximal hormonal effects were observed with biologically relevant concentrations. The rank order of potency of this limited series of iodothyronines differs markedly from that for nuclear TS receptor (for reviews see Refs. 30, 31) and is consistent with the proposed extra-nuclear site of action. Additional support for an extranuclear site of thyroid hormone action is provided by the observation that under our culture conditions few, if any, nuclear T3 receptors are present in glial cells (32). In contrast, others have reported modest levels of nuclear TX receptors in cultured glial cells obtained from primary dispersions of fetal mouse cerebral cortex grown without further subculture (33) and in C-6 astrocytomas cells (34, 35). Thus, it seems likely that continued subculture of astrocytes may result in loss of the nuclear TB receptor, without affecting the ability of the cell to respond to thyroid hormone by modulating type II 5'-deiodinase. Our preliminary survey of potential cellular mechanisms mediating the T4-dependent modulation of the degradation/ inactivation of the membrane-bound enzyme revealed that this was an energy-dependent process requiring an intact actin-cytoskeleton.
Lysosomotrophic agents and inhibitors of endocytosis and glycosidation had little or no effect on type II 5'-deiodinase suggesting that neither enzyme internalization by selective endocytosis nor glycosidation during synthesis contributed to the short half-life of the enzyme. The failure of colchicine to influence the dynamics of type II 5'-deiodinase induction and/or inactivation, despite a profound effect on the morphology of the btpcAMP-stimulated astrocyte indicates that turnover of this enzyme does not require intact microtubules.
Both adequate intracellular ATP and an intact actin-cytoskeleton were found to be essential for the rapid inactivation/ degradation of membrane-bound type II 5'-deiodinase.
Car-bony1 cyanide m-chlorophenylhydrazone-treated cells showed the expected modest decrease in protein synthesis and therefore type II 5'-deiodinase synthesis. However, enzyme activity was not further diminished by complete inhibition of protein synthesis with cycloheximide.
These data demonstrate that in addition to the well-known energy dependence of translation, inactivation/degradation of type II 5'-deiodinase requires ATP.
The most impressive antagonists of type II 5'-deiodinase turnover were cytochalasins.
Cytochalasins halted the loss of type II 5'-deiodinase in cells unable to synthesize protein and stimulated further accumulation of enzyme in cells capable of translation. Since cytochalasin B is an affinity ligand for the glucose transporter (36, 37), as well as an inhibitor of microfilament polymerization (38), we used the microfilamentspecific derivative, dihydrocytochalasin B (39), and both compounds yielded identical results. In preliminary studies, shortterm glucose depletion (4 h) had no effect on the disappearance kinetics of 5'-deiodinating activity in astrocytes grown in euthyroid medium, demonstrating that the cytochalasin effect was mediated through its ability to depolymerize the actin-cytoskeleton, rather than by altering glucose entry.' The specific interrelationships between the actin-cytoskeleton and the rapid inactivation/degradation of type II 5'deiodinase remain to be determined.
It is clear, however, that dynamic regulation of the structural integrity of the actincytoskeleton may play an important role in the inactivation pathway for this membrane-bound enzyme and thereby exert a direct effect on steady-state levels of the enzyme catalyzing intracellular Ts production in the brain. Recent studies by St. Germain (9) have attempted to address this issue by examining the nature of the thyroid hormone dependence of type II 5'-deiodinase turnover in neuroblastoma cells. A parallelism between the substrate specificity and enzyme cofactor requirements and the ability of these compounds to alter enzyme inactivation in NB41A3 cells led to the proposal that substrate-induced inactivation of type II 5'-deiodinase is a key factor in determining the biological half-life, and thereby cellular levels, of this enzyme. Our demonstration that disruption of the actin-cytoskeleton abolishes the inactivation of type II 5'-deiodinase in astrocytes grown in the presence of thyroid hormone suggests that substrate-enzyme interactions alone are insufficient to account for the TI-dependent modulation of type II 5'-deiodinase turnover. Thus, in a cell culture model system mimicking all of the known regulatory pathways of type II 5'-deiodinase in the brain, the T1-dependent regulation of biological half-life of the enzyme is mediated by an energy-dependent process requiring an intact actin-cytoskeleton.
Characterization of the interactions between thyroid hormone and the actin-cytoskeleton, without involvement of the nuclear T3 receptor, should provide new insights into the potential mechanisms available for this metabolically potent hormone to modulate neuronal arborization, cell-cell communication, and the structural abnormalities observed in the brain in congenital hypothyroidism.