Thyroxine-dependent Modulation of Actin Polymerization in Cultured Astrocytes A NOVEL, EXTRANUCLEAR ACTION OF THYROID HORMONE*

depolymerization specifically blocks the rapid hormone-dependent inactivation of type II iodothyronine 5’-deiodinase. internalization

without thyroxine also showed a disorganized actin cytoskeleton, and 10 nM thyroxine or 10 nM reverse triiodothyronine normalized the actin cytoskeleton appearance within 20 min; 10 nM 3,3',5-triiodothyronine had no effect. These data show that thyroxine modulates the organization of the actin cytoskeleton in astrocytes and suggest that regulation of actin polymerization may contribute to thyroid hormone's influence on arborization, axonal transport, and cell-cell contact in the developing brain.
Thyroid hormone plays an essential role in the growth and development of the mammalian brain influencing such diverse events as dendritic arborization, myelinization, expression of key enzymes in neurotransmitter synthesis, and the glial/ neuronal cell ratio (1, 2). However, specific effects of thyroid hormone excess or deficiency on cell metabolism in the adult brain have been elusive, even though the presence of substantial numbers of nuclear thyroid hormone receptors is well documented in brain (3)(4)(5). Since most, if not all, of the effects * This work was supported by National Institutes of Health Grant DK38 772 (to J. L. L.) and the Swiss National Research Foundation Grant 3.943.0.84.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 11 To whom reprint requests should be addressed.
of thyroid hormone are thought to be initiated by the selective binding of 3,5,3'-triiodothyronine (T3)l to these receptors, the inability of the adult brain to respond to thyroid hormone remains poorly understood.
In most animals the bioactive form of thyroid hormone, TB, is derived from the enzyme-catalyzed 5'-deiodination of Tq, the major secretory product of the thyroid gland (6). This enzyme-catalyzed bioactivation of thyroid hormone accounts for >75% of the T3 found in the circulation (7,8). The brain differs from other hormone-responsive tissues in that it relies almost exclusively on intracellular conversion of Tq to Ts rather than the circulation, as its source of bioactive TS (6-8). Interestingly, the activity of type II iodothyronine 5'deiodinase, the brain enzyme catalyzing this reaction, shows rapid 5-to lo-fold increases after thyroidectomy and equally rapid decrements with hormone replacement (9-11). This ability of the brain to modulate type II iodothyronine 5'deiodinase content and thus intracellular T3 production appears to be a homeostatic mechanism to preserve intracerebral TS levels within narrow limits and is likely to play a role in preventing pathophysiological changes in cerebrocortical metabolism in hyper-or hypothyroidism (10). The mechanism(s) by which thyroid hormone regulates levels of this plasma membrane-bound enzyme in the brain have been studied both in vivo and in cell culture and do not appear to involve the classical nuclear receptor. Comparison of the ability of individual iodothyronines to diminish cerebrocortical type II iodothyronine 5'-deiodinase activity in hypothyroid rats showed Tq = rTg >> TS (10, ll), in sharp contrast to the rank order of potency for binding to the nuclear T3 receptor (T3 > Tq > rT3). Thyroid hormone-induced increases in enzyme inactivation/degradation are unaffected by inhibitors of transcription or translation and can be accounted for by a selective increase in enzyme degradation (12). Comparable results have been obtained in cell culture models using dispersed fetal (13)  We have recently shown that the T4dependent increase in enzyme inactivation is energy-dependent and that cytochalasins specifically block enzyme loss (18). In this study, we determined the effects of thyroid hormone on actin content and microfilament organization in astrocytes and examined some of the interrelationship(s) between the T4-dependent regulation of this short-lived membrane-bound enzyme and the actin cytoskeleton.
The results demonstrate that the organization of the actin cytoskeleton in astrocytes is dynamically regulated by thyroid hormone. In addition, T4's influence on the polymerization state of the actin cytoskeleton is temporally related to the T,-induced changes in type II 5'deiodinase. MATERIALS AND

RESULTS
Thyroxine has been shown to regulate type II iodothyronine 5'-deiodinase levels by modulating the rate of enzyme degradation without altering enzyme synthesis. Cytochalasins se-lectively arrest this process (18). Since type II iodothyronine 5'-deiodinase is a short-lived integral membrane protein, associated with neurilemmal membranes (28), then a likely initial step in the degradation pathway is enzyme internalization. To determine whether the steady-state levels and/or the tsflt of type II iodothyronine 5'-deiodinase are influenced by the cell actin content, we varied the actin content of astrocytes by altering the plating density of the primary dispersions as described by Goldman and Chiu (29). Depicted in Fig. 1A is the effect of initial plating density on type II iodothyronine 5'-deiodinase inactivation.
The rate of enzyme disappearance in cycloheximide-blocked cultures was inversely related to the initial plating density, being fastest in 42-@yy*) w=--Actin (43k) 31-FIG. 1. Effects of initial plating density on the rate of type II iodothyronine 5'-deiodinase inactivation and the composition of the Triton-insoluble cytoskeleton in bt2cAMP-stimulated astrocytes. A, the rate of type II iodothyronine 5'-deiodinase (Fi'D-II) inactivation was determined in triplicate (25-cm' flasks) in cells at passage three as described under "Materials and Methods." Data are presented as the percent of the starting enzyme activity of closely agreeing (+lO%) triplicate cultures. SF, serum-free culture conditions; FCS, serum containing culture conditions. B, SDS-PAGE of the Triton-insoluble cytoskeleton in astrocytes. Aliquots (15 pg of protein) of the Triton-insoluble pellets obtained from cells at passage three were separated on 12.5% acrylamide slab gels as described under "Materials and Methods." Lane a, cells derived from primary cultures plated initially at 50,000 cells/cm*; lane b, cells plated at 400,000 cells/cm'; lane c, cells plated at 1,200,OOO cells/cm'. cells derived from the lowest initial density and slowest in cells grown from dispersions plated at the highest density.
Shown in Fig. 1B are the effects of increasing the initial plating density on composition of the Triton-insoluble cytoskeleton in astrocytes. As expected, F-actin was most abundant in cells derived from primary dispersions seeded at low density and decreased proportionally as the initial plating density increased. These changes in F-actin content were maintained through three passages indicating the density of the primary dispersion was a prime factor in determining actin levels in astrocytes. In contrast, the intermediate filament proteins, GFAP and vimentin, were much less dependent upon initial plating density with the apparent changes in quantity shown in lanes a-c (Fig. 1B) Table I. Steady-state enzyme levels increased proportionately from 43 to 123 units/mg protein and the tH increased >3-fold in cells grown from dispersions in which the plating density was increased 20-fold. These increases in type II iodothyronine 5'-deiodinase were due primarily to changes in the inactivation rate since enzyme production rates, as calculated from the steady-state enzyme levels and the disappearance rate constants (k), were relatively unchanged at 349 f 34 (S.E., n = 4) units/mg protein/h in the cells obtained from the four different initial plating densities. F-actin levels were also inversely related to the initial plating density and remained so for at least three passages, decreasing -4-fold as the plating density increased 20-fold (Table I). These changes in F-actin paralleled the increase in total cell actin, thus the ratio of F-to G-actin remained constant (data not shown). had a marked effect on the actin content of the Triton-insoluble cytoskeleton of astrocytes. Astrocytes grown in the thyroid hormone-deficient medium (serum-free medium) had only 40% of the F-actin of cells grown in serumsupplemented medium (8.1 uersus 12.8% of the total Tritoninsoluble protein), and addition of 10 nM T, restored the F- Quadruplicate flasks (25 cm") of confluent cells were grown from cultures seeded at increasing initial plating density and subcultured three times. Type II iodothyronine 5'-deiodinase was induced by treating the cells for 16 h with 0.5 mM b&AMP, 100 nM hydrocortisone. Enzyme half-times were determined from linear semilogarithmic plots of the disappearance of type II catalytic activity in cycloheximide-blocked cells versus time as described under "Materials and Methods." F-actin content was determined as described under "Materials and Methods" and is expressed as AU/mg cell protein (AU = arbitrary absorbance units). Data are from two representative experiments, and results are presented as means of closely agreeing (+ 10%) quadruplicate flasks. actin content to control levels (11.4 uersus 12.8% of total Triton-insoluble protein). Identical reductions in F-actin content were observed when 5% (v/v) hypothyroid rat serum was substituted for the calf serum in the culture medium (data not shown), indicating that other serum factors had little, if any, influence on F-actin content under these culture conditions. Total cellular actin, GFAP, and vimentin were unaffected by the presence or absence of serum (data not shown).
To examine whether T, and/or btncAMP treatment contributed to the changes in F-actin content by stimulating actin transcription, we determined the effects of Tq and btncAMP on @-actin mRNA levels. As shown in Fig. 2 Fig. 3 is the time course for the T4-dependent decrease of type II 5'-deiodinase in astrocytes. Addition of Tq (-100 pM "free" hormone) to bt*cAMP-treated cells grown in serum-free medium resulted in a rapid 50% loss of activity during the first 8 min followed by a progressively slower loss 1 2 18Sw FIG. 2. @-Actin mRNA levels in cultured glial cells.
Confluent primary cultures (100,000 cells/cm* initial plating density) were subcultured at 1 x lo4 cell/cm* and grown to confluence. Cells were then treated with 0.5 mM bt,cAMP, 100 nM hydrocortisone for 16 h to induce type II iodothyronine 5'-deiodinase activity as indicated. lo-fig aliquots of total RNA were separated on a 1.2% agarose/ formaldehyde gel, transferred to Duralose, and probed with p-actin "'P-cDNA as described under "Materials and Methods." Lane 1, 10 pg of RNA from untreated cells grown in 10% serum-containing medium; lane 2, 10 fig of RNA from cells grown in serum-free medium and treated with 1 XTIM bt*cAMP and 100 nM hydrocortisone for 16 h. Astrocytes were grown in serum-free medium and stimulated as described under "Materials and Methods." T, (final concentration 10 nM in Hank's solution containing 1 mg/ml BSA) was added at the start of the experiment and triplicate flasks of cells harvested at the times indicated. Type II 5'-deiodinating activity was determined in an aliquot of the cell sonicates as described under "Materials and Methods" and the results reported as the percent of starting enzyme activity. Data are reported as the means of closely agreeing (+lO%) triplicate determinations. ; seeded at 4 X lo5 cells/cm*) were treated for 16 h with 0.5 mM btncAMP and 100 nM hydrocortisone to induce type II iodothyronine 5'-deiodinase. Flasks were then pretreated for 15 min with PBS, 0.1 mM cycloheximide, or 10 FM actinomycin D. 10 nM T, was then added as indicated and triplicate flasks harvested at 0, 5, 10, and 20 min. F-actin was determined on the cells from individual flasks as described under "Materials and Methods." Data are expressed as arbitrary absorbance units (AU) per mg of total cell protein and are reported as the means of closely agreeing (+lO%) triplicates. of catalytic activity that plateaued at 35-40% of the serumfree control. The rapid effects of Tq on the F-actin content of astrocytes are shown in Table II. F-actin levels followed a similar time course to that of type II iodothyronine 5'-deiodinase, with a rapid -2-fold increase during the first 5-10 min and a plateau after -10 min of treatment. Proportional decreases in G-actin were observed with no change in total actin content during this short period of Tq treatment (data not shown). Neither cycloheximide nor actinomycin D pretreatment blocked the T1-induced increases in F-actin in astrocytes indicating that neither translation nor transcription were required for this thyroid hormone effect.
Effects of Thyroid Hormone on Microfilament Organization in B&AMP-stimulated Astroglial Cells-Because both cyclic nucleotides and cytochalasins promote structural reorganiza-cytoskeleton, a time course comparable with the hormonetion of actin (30), and cytochalasins antagonize the T4-de-induced changes in the type II iodothyronine 5'-deiodinase pendent regulation of type II iodothyronine 5'-deiodinase inactivation shown in Fig. 3. (18), we examined the effects of thyroid hormone on microfilament organization in astrocytes (Fig. 4). A complex net-DISCUSSION work of actin bundles was present in astrocytes grown in The ability of cytochalasins to block the T1-dependent serum-containing medium (Fig. 4a), and this network was inactivation of type II iodothyronine 5'-deiodinase (18) raised markedly reduced in both amount and fiber length in the the possibility that there was a functional relationship beabsence of thyroid hormone (Fig. 4b). Tq replacement alone tween the organization of the actin cytoskeleton and the was sufficient to restore the organization of the actin cyto-inactivation of this membrane-bound enzyme. In this study, skeleton (Fig. 4~) to that observed in control cultures.
we have shown that (i) increases in the cell actin content The rapid effects of iodothyronine replacement on the were accompanied by decreases in the tnh of type II iodothyappearance of the astrocyte actin cytoskeleton are illustrated ronine 5'-deiodinase with no change in the enzyme production in panels d-f of Fig. 4. Cells exposed to 10 nM TC1 for 20 min rate; (ii) Tq and rT., promoted F-actin formation and increased showed little or no reorganization of their actin cytoskeleton the rate of enzyme inactivation, whereas Ts was ineffective; (Fig. 4d), similar to T.l's lack of effect on enzyme inactivation (iii) the time courses for T,-dependent formation of F-actin at these concentrations (10,18). In contrast, cells treated for and the increase in type II iodothyronine 5'-deiodinase turnonly 20 min with either 10 nM Tq (Fig. 4e)  with 10 nM T:,, 10 nM T., or 10 nM rT:c, respectively. polymerization did not require continued protein synthesis or transcription.
These data suggest that the extranuclear Tqdependent modulation of type II 5'-deiodinase depends, in part, upon alterations in the polymerization state of the actin cytoskeleton.
Actin comprises a major fraction of the astrocyte cytoskeleton in uiuo (31) and in cultures of immature and mature astrocytes (32). The ability to alter the actin content of cultured astrocytes by changing the initial plating density (29), and the observation of Ciesielski-Treska et al. (32) that a 24-h exposure to bt,cAMP had no effect on total actin content in these cells, allowed us to manipulate actin levels in cells expressing type II iodothyronine 5'-deiodinase.
Steady-state enzyme levels decreased and the t,,* of the enzyme grew progressively shorter as the actin content increased in the cultured cells. Interestingly, changes in the total actin content of glial cells had no effect on the rate of type II iodothyronine 5'-deiodinase synthesis suggesting that alterations in the composition of the cytoskeleton were responsible for the changes in the inactivation/degradation of this enzyme.
A relationship between the T1-dependent modulation of the biological half-life of type II iodothyronine 5'-deiodinase and cell actin levels was further strengthened by examination of thyroid hormone's influence on F-actin content. Cells grown in the absence of thyroid hormone contained -40% less Factin than cells grown in presence of Tq. Addition of Tq alone normalized F-actin in cells grown in serum-free medium suggesting that thyroid hormone was essential for normal actin polymerization.
Importantly, the T4-dependent changes in Factin content were not mediated by hormone-induced changes in transcription or translation, since neither cycloheximide nor actinomycin D affected this cell response. In addition, pactin mRNA levels were unaffected by T.+ These data are consistent with the proposed extranuclear site of action for thyroid hormone (12,18).
Examination of the actin cytoskeleton in cultured astrocytes confirmed thyroid hormone's influence on the microfilament network. Astrocytes grown in the absence of thyroid hormone had a poorly developed actin cytoskeleton compared with cultures grown in medium supplemented with serum or with Tq alone. Iodothyronines induced a rapid reorganization of the poorly developed actin cytoskeleton in cells grown in serum-free medium, requiring as little as 20 min of exposure to thyroid hormone. Both Tq and the metabolically inactive metabolite, rT3, promoted repolymerization of the actin cytoskeleton, while an equimolar concentration of TB was ineffective. These results are identical to earlier reports of the effects of these iodothyronines on type II 5'-deiodinase turnover in vivo (10, 11) and in the cultured astrocyte (18).
The specific events that mediate thyroid hormone's regulation of actin polymerization remain to be established. The finding that this hormone action is not transcriptionally mediated (18,33) suggests that Tq may bind to and modulate the activity of one or more actin-binding proteins. Several actin-binding proteins capable of depolymerizing actin filaments or anchoring the actin filament to the membrane have been identified in the brain (34)(35)(36) and these may be targets for T, binding and subsequent dissociation from the arrested actin fibril.
T1-dependent changes in actin polymerization may be particularly important during brain development. Both neurite outgrowth and dendritic spine formation are developmentally programmed events that are highly correlated with an increase in actin filaments (37-39). These processing events depend, in part, on specific interactions between the actin cytoskele-ton, integrins in the neuronal cell membrane (40)(41)(42)(43), and the neurite-promoting domain of the laminin (44). Thus, the ability of thyroid hormone to modulate actin polymerization provides an attractive model by which this hormone could regulate neuronal process formation and ultimately cell-cell interactions.
The recent work of Faivre-Sarrailh and Rabie (33) is in agreement with this idea. They reported that the cerebellum of congenitally hypothyroid rats showed 60-70% decrements in F-actin content with no change in total actin pool and that Tq administration normalized the F-actin content of the cerebellum. These findings obtained in intact rats parallel our observations on the T4-dependent changes in actin polymerization in the cultured astrocyte and demonstrate that this culture model will be useful in unraveling the molecular events that mediate the TI-dependent regulation of actin polymerization in the central nervous system.