Degradation and recycling of the substrate-binding subunit of type II iodothyronine 5'-deiodinase in astrocytes.

Thyroxine dynamically regulates levels of type II iodothyronine 5′-deiodinase (5′D-II) by modulating enzyme inactivation and targeting the enzyme to different pathways of internalization. 5′D-II is an ∼200-kDa multimeric protein containing a 29-kDa substrate-binding subunit (p29) and an unknown number of other subunits. In the absence of thyroxine (T4), p29 is slowly endocytosed and transported to the lysosomes. T4 treatment rapidly activates an actin-mediated endocytotic pathway and targets the enzyme to the endosomes. In this study, we have characterized the influence of T4 on the intracellular trafficking of 5′D-II. We show that T4 accelerates the rate of 5′D-II inactivation by translocating the enzyme to the interior of the cell and by sequestering p29 in the endosomal pool without accelerating the rate of degradation of p29. This dichotomy between the rapid inactivation of catalytic activity and the much slower degradation of p29 is consistent with the reuse of p29 in the production of 5′D-II activity. Immunocytochemical analysis with a specific anti-p29 IgG shows that pulse affinity-labeled p29 reappears on the plasma membrane ∼2 h after enzyme internalization in the presence of T4, indicating that p29 is recycled. Despite the ability of p29 to be recycled in the T4-treated cell, 5′D-II catalytic activity requires ongoing protein synthesis, presumably of another enzyme component(s) or an accessory enzyme-related protein. In the absence of T4, enzyme inactivation and p29 degradation are temporally linked, and pulse affinity-labeled p29 is internalized and sequestered in discrete intracellular pools. These data suggest that T4 regulates fundamental processes involved with the turnover of integral membrane proteins and participates in regulating the inter-relationships between the degradation, recycling, and synthetic pathways.

Type II iodothyronine 5Ј-deiodinase (5ЈD-II) 1 is a multimeric integral membrane protein (ϳ200 kDa) that catalyzes T 4 to T 3 conversion in the brain (1)(2)(3)(4). Thyroid hormone, specifically T 4 , dynamically regulates 5ЈD-II levels by modulating the biologi-cal half-life of this short-lived protein without altering enzyme synthesis (5). In the absence of T 4 , enzyme levels are high, and 5ЈD-II inactivation is slow. T 4 promotes interactions between the enzyme and the F-actin microfilaments that lead to 5ЈD-II inactivation and enzyme internalization, resulting in a rapid fall in enzyme levels (6 -8). This action of T 4 is independent of transcription or translation, indicating an extranuclear mode of action, and has been extensively characterized in cultured astrocytes that lack functional thyroid hormone receptors (9). Indeed, T 4 is Ͼ100-fold more potent than the transcriptionally active T 3 in regulating 5ЈD-II inactivation (1,(5)(6)(7)(8).
The 29-kDa substrate-binding subunit (p29) of 5ЈD-II is covalently modified by the alkylating affinity label N-bromoacetyl-L-T 4 (BrAcT 4 ), allowing the enzyme to be identified without measuring catalytic activity (10). Utilizing cultured astrocytes that retain all of the 5ЈD-II regulatory aspects seen in the brain in vivo and that express high levels of 5ЈD-II activity in the presence of cyclic AMP (11), we have previously shown that T 4 dynamically regulates 5ЈD-II levels by directing the enzyme to use different pathways of internalization (8). Under T 4 -deficient conditions, p29 is internalized via the traditional endocytotic pathway and is slowly transported through the endosomes to the lysosomes. In contrast, T 4 treatment activates specific protein-F-actin interactions involved in actinmediated endocytosis and targets the enzyme to the endosomes, where the internalized 5ЈD-II-containing vesicle remains without subsequent transit to the lysosomes. Recycling of the catalytically active enzyme back to the plasma membrane has not been observed in the short time frames examined previously; thus, the fate of the endosomal pool of internalized 5ЈD-II remains uncertain in the presence of T 4 .
The endosomes are a collection of vesicles located in the perinuclear space (12,13). Vesicles internalized via endocytosis initially exist as early endosomes and contain a mixture of polypeptides with differing fates. Vesicles containing proteins to be recycled, such as the transferrin and insulin receptors, are directed back to the plasma membrane (14), while vesicles containing proteins destined for degradation evolve into late endosomes and are targeted to the lysosomes. In the secretory pathway, vesicles containing newly synthesized membrane proteins pass from the endoplasmic reticulum to the Golgi stack before exiting through the trans-Golgi network, where they are targeted to the plasma membrane directly or routed through the endosomes (15,16). From the endosomes, the secretory vesicles may be directed to the plasma membrane, to the lysosomes, or to vesicle storage pools. The sorting mechanism that determines the destination of the differing vesicles that compose the endosomes is still unclear (17).
The T 4 -dependent regulation of the pathways of 5ЈD-II inactivation/internalization suggests a role for hormonal regulation in vesicle-mediated protein transport (8). In this study, we show that 1) the rate of degradation of the 29-kDa substrate-binding subunit of 5ЈD-II (p29) is unaffected by T 4 ; 2) in the presence of T 4 , p29 is recycled back to its site of action on the plasma membrane; and 3) despite recycling of p29, 5ЈD-II catalytic activity requires de novo synthesis of additional enzyme components. Since T 4 determines whether p29 is recycled back to the plasma membrane or is routed to the lysosomes, these data suggest that, in addition to regulating the internalization pathway of 5ЈD-II, T 4 participates in regulating the inter-relationships between the degradation, recycling, and synthetic pathways.

EXPERIMENTAL PROCEDURES
Materials-Pregnant (16 -17-day gestation) rats were obtained from Charles River Laboratories (Kingston, NY). T 4 and BSA was purchased from Sigma. T 3 was obtained from Henning GmbH, and dihydrocytochalasin B was obtained from Calbiochem. Dulbecco's modified Eagle's medium (DMEM), antibiotics, Hanks' solution, and 0.25% trypsin were obtained from Life Technologies, Inc., and defined bovine calf serum (heat-inactivated) was from Hyclone Laboratories. Culture flasks were obtained from Nunc, and 24-well tissue culture plates were obtained from Falcon. All other reagents used were of the highest purity commercially available.
Cell Culture-Rat type I astrocyte cultures were obtained by enzymatic dispersion of neonatal rat brains as described previously (18). Cells were grown in a humidified atmosphere of 5% CO 2 and 95% air at 37°C in DMEM supplemented with 15 mM sodium bicarbonate, 33 mM glucose, 1 mM sodium pyruvate, and 15 mM HEPES, pH 7.4, with 10% (v/v) calf serum, 50 units/ml penicillin, and 90 g/ml streptomycin. The culture medium was changed three times weekly, and cells were subcultured (2-3 ϫ 10 4 cells/cm 2 ) when they reached confluence (7-10 days). Confluent cells from passages 2-4, containing Ͼ95% astrocytes as determined by staining for the astrocyte-specific protein, glial fibrillary acidic protein (19), were utilized for experiments.
Induction and Measurement of 5ЈD-II Activity-Steady-state levels of 5ЈD-II were induced by incubating the cells for 8 h in a defined medium consisting of DMEM and 0.1% BSA Ϯ thyroid hormone, followed by a 16-h stimulation with 1 mM Bt 2 cAMP and 100 nM hydrocortisone. Cells were harvested by scraping in ice-cold 8 mM sodium phosphate buffer, pH 7.4, containing 2.7 mM KCl and 137 mM NaCl and collected by centrifugation. 5ЈD-II activity was determined in cell sonicates by the iodide release method at 2 nM rT 3 and 20 mM dithiothreitol in the presence of 1 mM propylthiouracil. Units are expressed as femtomoles of I Ϫ released per hour.
The turnover of 5ЈD-II was determined by assaying enzyme activity 5-60 min after inhibition of protein synthesis with 100 M cycloheximide in 50 mM HEPES and 1 mg/ml BSA in buffered Hanks' solution, pH 7.0, in the presence or absence of 10 nM T 4 . The turnover of 5ЈD-II in the presence of BFA was determined by preincubating cells with 5 g/ml BFA for 15 min before the addition of 100 M cycloheximide. Cells were then collected as described above and assayed for 5ЈD-II activity.
Degradation of p29 -The p29 polypeptide was affinity-labeled in confluent astrocytes expressing steady-state levels of 5ЈD-II in the absence of thyroid hormone by incubation for 20 min with either 0.4 or 10 nM BrAc[ 125 I]T 4 , 1 mM dithiothreitol, and 50 mM HEPES in buffered Hanks' solution, pH 7.0. The labeling medium was removed, and the cells were washed free of unbound affinity label and then chased for 30 -120 min with either 10 nM T 4 or T 3 or with no hormone in 50 mM HEPES and 1 mg/ml BSA in buffered Hanks' solution, pH 7.0. Cells were collected and sonicated. Affinity-labeled proteins were identified after SDS-polyacrylamide gel electrophoresis and autoradiography. The quantity of p29 was determined by scanning densitometry.
Immunocytochemistry-Cells were seeded onto glass coverslips (22 ϫ 22 mm) coated with poly-D-lysine (10 g/ml) and grown for 24 h. 5ЈD-II activity was induced as described above. Cells were affinity-labeled for 20 min with 10 nM BrAcT 4 in buffered Hanks' solution containing 10 mM dithiothreitol and 50 mM HEPES, pH 7.0, and washed and chased for 0 -120 min in DMEM with 1 mg/ml BSA, 1 mM Bt 2 cAMP, and 100 nM hydrocortisone in the presence and absence of 10 nM T 4 . All cells were treated with 10 M colchicine to relax the cell borders for 60 min prior to fixation. Cells were fixed in iced 4% paraformaldehyde and permeabilized with iced methanol. 5ЈD-II was identified by incubation with an antiserum raised against p29 (see accompanying paper (31)) or with an anti-T 4 IgG that recognizes the BrAcT 4 moiety, and immune complexes were identified by incubation with Texas Red-conjugated anti-rabbit IgG. Coverslips were then mounted and examined by laser scanning confocal microscopy with a depth of resolution of 0.5 m.
Statistical Analysis-Where indicated, results were analyzed by unpaired Student's t test.

RESULTS
Degradation of p29 -The 29-kDa substrate-binding subunit of 5ЈD-II (p29) is selectively affinity-labeled by BrAcT 4 (4, 10). In the absence of T 4 , p29 is internalized and targeted to the lysosomes, whereas in the presence of T 4 , p29 remains indefinitely in the endosomal pool localized in the perinuclear space of the cell (7,8). In light of the different metabolic fates of proteins residing in these two intracellular compartments, we examined the effect of thyroid hormone on the degradation of p29. Cells were grown in thyroid hormone-free medium and stimulated with Bt 2 cAMP and hydrocortisone to express high levels of 5ЈD-II (11). P29 was affinity-labeled, and cells were then chased with either 10 nM T 4 or T 3 or with no hormone for up to 90 min. The cellular content of p29 and the levels of catalytically active 5ЈD-II were analyzed at 0 -90 min after the addition of hormone. As shown in Fig. 1A, the disappearance of p29 over time was unaffected by thyroid hormone, while Ͼ80% of the catalytic activity of 5ЈD-II was lost due to the presence of T 4 (Fig. 1B). Little or no other smaller affinity-labeled degradation products from the affinity-labeled p29 protein were observed over this time frame (data not shown).
Shown in Table I are the steady-state levels of p29 in cells grown in the presence and absence of T 4 . As reported previ- ously, the quantity of BrAcT 4 -labeled p29 (labeled at a concentration of 0.4 nM BrAcT 4 ) (see accompanying paper (31)) in T 4 -deficient cells was ϳ2-fold greater than that in T 4 -treated astrocytes. In contrast, 5ЈD-II activity in T 4 -deficient cells was ϳ10-fold higher than that in T 4 -treated astrocytes (Table I) (6,10). This apparent discrepancy between the p29 content and 5ЈD-II activity levels is likely to be due to different quantification techniques for catalytic activity and p29 (10) and the ability of BrAcT 4 to label the p29 subunit of catalytically inactive 5ЈD-II (see accompanying paper (31)). Since T 4 has no effect on the fractional disappearance rate of p29 (Fig. 1A), the calculated appearance rate of this polypeptide in the presence of T 4 is approximately half that seen in the absence of hormone (Table I). As shown previously (6,10), the production rate of catalytic activity is unaffected by thyroid hormone (Table I). Thus, the production and disposal of the p29 substrate-binding subunit of 5ЈD-II differ from the production and disposal of 5ЈD-II catalytic activity.
Effect of Cycloheximide on 5ЈD-II Catalytic Activity-The differences between the effects of T 4 on the respective t1 ⁄2 values of p29 and 5ЈD-II catalytic activity, our observation that p29 accumulates in the endosomal pool in the presence of T 4 (8), and the observation that p29 is found in the endosomal pool in catalytically inactive astrocytes (see accompanying paper (31)) raise the possibility that this multimeric enzyme is recycled. To examine this possibility, recycling of 5ЈD-II activity was determined in protein synthesis-blocked cells kept for extended periods of time. Cells expressing 5ЈD-II activity in T 4 -replete medium were incubated in the presence and absence of 100 M cycloheximide for up to 4 h, and 5ЈD-II activity was assayed. As shown in Fig. 2, 5ЈD-II activity fell to undetectable levels in cycloheximide-blocked cells, with little or no recovery of catalytic activity observed at any time point. Cell viability, as determined by trypan blue exclusion, was unaffected over the 4-h treatment period with cycloheximide (data not shown). Enzyme levels remained constant in the presence of continued protein synthesis. Thus, the maintenance of 5ЈD-II catalytic activity in astrocytes requires ongoing protein synthesis.
Effect of Brefeldin A on 5ЈD-II Activity-BFA is a fungal toxin that dramatically alters vesicle trafficking within the cell, causing fusion of the Golgi/endoplasmic reticulum/endosomal organelles and blocking transport of vesicle proteins to the lysosomes (20 -22). This leads to a functional separation of the secretory, recycling, and degradative pathways and also promotes membrane protein recycling (20,21). For example, BFA increases the accumulation of the receptors for insulin-like growth factor II and insulin on the plasma membrane, two receptors that are recycled through the endosomes in fibroblasts (22). To determine if 5ЈD-II activity could be recycled from the endosomes in astrocytes, we used BFA to maximize transit through the membrane protein recycling pathways.
As shown in Fig. 2, the addition of BFA to cycloheximideblocked, T 4 -replete cells had no effect on the fall of 5ЈD-II activity. In the absence of BFA, there was little or no recovery of catalytic activity observed at any time point. Next, we examined the effects of BFA on steady-state levels of 5ЈD-II in the presence of continued protein synthesis (Fig. 3). BFA had no effect on steady-state levels in T 4 -replete cells during treatment periods up to 60 min. In contrast, BFA treatment led to a steady increase in 5ЈD-II activity in T 4 -deficient cells, resulting in a doubling of enzyme activity by 60 min. Analysis of the disappearance kinetics of 5ЈD-II revealed that BFA prolonged the biological half-life of the catalytically active enzyme in both T 4 -deficient (t1 ⁄2 ϭ 2.1 versus 3.0 h Ϫ1 ) (Fig. 4A) and T 4 -replete (t1 ⁄2 ϭ 0.2 versus 0.3 h Ϫ1 ) (Fig. 4B) astrocytes. Since the biological half-life of 5ЈD-II activity in T 4 -replete cells was prolonged in the presence of BFA and the levels of enzyme remained unchanged, BFA treatment must cause a proportional fall in enzyme production. 5ЈD-II generation in T 4 -deficient cells was minimally affected by BFA, as there was a concordant increase in both t1 ⁄2 and enzyme levels. These data imply that the sources of the catalytically active holoenzyme are different in cells grown in the presence and absence of T 4 .
One explanation for these differential effects of BFA on 5ЈD-II activity is that one or more of the enzyme-associated polypeptides are recycled in cells grown in the presence of T 4 . Thus, in T 4 -treated cells where 5ЈD-II is rapidly internalized to the endosomes (7,8), BFA treatment should prevent the recycling enzyme polypeptide(s) from joining the de novo enzyme production pathway, leading to an accumulation of inactive enzyme precursors in the fused vesicle pool and thereby decreasing the observed enzyme production rate. To test this hypothesis, we determined the effects of BFA on the accumulation of catalytically active 5ЈD-II after depolymerization of the F-actin cytoskeleton in T 4 -treated astrocytes. Depolymerization of F-actin by dihydrocytochalasin B blocks T 4 -mediated endocytosis and results in the accumulation of 5ЈD-II activity at a rate equal to the production rate of enzyme catalytic activity (6). An accelerated rate of accumulation of catalytic activity is expected if a pool of enzyme precursors is formed in the presence of BFA.
As shown in Fig. 5, the addition of dihydrocytochalasin B to T 4 -replete astrocytes caused the steady accumulation of 5ЈD-II activity as reported previously (6). Pretreating the T 4 -replete cells with BFA increased the initial rate of enzyme accumula-  tion by ϳ2-fold in dihydrocytochalasin B-blocked cells. This effect was transient in cells pretreated for only 30 min with BFA, while a 60-min pretreatment period led to a sustained increase in the rate of 5ЈD-II activity accumulation. These data suggest that BFA treatment results in an increase in the stor-age of enzyme-related polypeptides in an intracellular pool.
Immunocytochemical Analysis of the Intracellular Transit of the Substrate-binding Subunit of 5ЈD-II-Direct examination of the intracellular transit of the enzyme polypeptide(s) was performed by using a specific IgG directed against the affinitylabeled 29-kDa substrate-binding subunit of 5ЈD-II (anti-p29 IgG) (see accompanying paper (31)). T 4 -deficient astrocytes were grown on glass coverslips, pulse affinity-labeled with BrAcT 4 , and then chased with either 10 nM T 4 or no hormone. Cells were fixed after increasing periods of time, stained with anti-p29 IgG, and examined by laser scanning confocal microscopy.
As shown in Fig. 6A (arrows), punctate staining is present in a "rim" pattern at the periphery of the cell in the thyroid hormonedeficient, affinity-labeled astrocyte, while the interior of the cell is largely devoid of immunoreactive staining. This rim pattern is consistent with a plasma membrane location for the staining. After a 20-min treatment with T 4 , the majority of the staining is located in the perinuclear space within the cell (Fig. 6B, P), with little immunoreactive staining remaining on the plasma membrane (arrows). These patterns are consistent with previous studies done with an anti-T 4 IgG that show that affinity-labeled polypeptide(s), predominantly p29, are located on the cytoplasmic leaflet of the plasma membrane in the thyroid hormonedeficient cell and are translocated to the perinuclear space within 20 min of exposure to T 4 (7). Similarly, previous studies have shown that the majority of affinity-labeled p29 after short-term exposure (20 min) to T 4 resides in the endosomal pool (8). After a 2-h exposure to T 4 , prominent punctate staining is again found at the periphery of the cell in a rim pattern (Fig. 6C, white arrows), along with clusters of staining present diffusely throughout the cell (black arrow). Since anti-p29 IgG selectively recognizes affinity-labeled p29 (see accompanying paper (31)), these data show that pulse-labeled p29 was internalized and recycled back to the plasma membrane in T 4 -treated astrocytes. In contrast to the T 4 -treated cells, immunoreactive staining in the T 4 -deficient cell is clustered into discrete regions within the cell (Fig. 6D, black arrow), with little staining remaining on the plasma membrane (white arrows). Previous studies have shown that affinitylabeled p29 is transported from the plasma membrane to discrete intracellular pools identified as lysosomes in T 4 -deficient cells (8).
The effects of BFA on the recycling of pulse-labeled p29 are shown in Fig. 7. Pulse-labeled p29 was identified with an anti-T 4 IgG that recognizes the BrAcT 4 -labeled enzyme as described previously (7). Punctate staining in a rim pattern is present in the cells treated with T 4 for 2 h, indicating a plasma membrane location (Fig. 7, upper left panel, arrows). This pattern is identical to the pattern observed in Fig. 6C using anti-p29 IgG and is again consistent with the recycling of p29 back to the plasma membrane. Treatment with BFA blocked the reappearance of p29 on the plasma membrane in the presence of T 4 since the immunofluorescence remains in clusters within the cell and rim staining is not present (Fig. 7, upper right  panel). In the absence of T 4 , the immunoreactive staining is predominantly clustered within the cell at 2 h, with little effect of BFA observed on the distribution of immunofluorescence (Fig. 7, lower panels). DISCUSSION In this study, we have characterized the intracellular trafficking of 5ЈD-II in cAMP-stimulated astrocytes and have shown that T 4 regulates fundamental processes involved in the turnover of integral membrane proteins. Specifically, T 4 shuttles plasma membrane proteins to the endosomes by switching from an actin-independent to an actin-mediated internalization pathway. Once in the endosomes, T 4 modulates the sorting mechanism to select recycling pathways over degradation pathways. Thus, T 4 participates in regulating the inter-relationships between the degradation, recycling, and synthetic pathways.
The evidence the T 4 shifts 5ЈD-II to a recycling pathway is 3-fold. First, there is a dichotomy between the rapid inactivation of catalytic activity and the much slower degradation of the 29-kDa substrate-binding subunit (p29) in euthyroid cells, consistent with the reuse of this protein in the production of catalytically active 5ЈD-II. Second, there is an increased rate of accumulation of 5ЈD-II activity in brefeldin A-treated euthyroid cells after depolymerization of the F-actin microfilaments, consistent with the storage of enzyme-related polypeptides in an intracellular pool that is available to the synthetic pathway. Finally, using anti-p29 antisera, pulse-labeled p29 reappears on the plasma membrane ϳ2 h after internalization in the presence of T 4 .
In contrast, the 5ЈD-II-containing vesicles in T 4 -deficient cells are sorted through the endosomes to the degradation pathway, ending up in the lysosomes (8). Enzyme inactivation parallels p29 degradation in the absence of T 4 . Interestingly, the degradation rate of p29 is similar whether it is sorted directly to the lysosomes in the absence of T 4 or is routed through one or more recycling sequences in the presence of T 4 .
Since 5ЈD-II is a multimeric enzyme, there are two possibilities for recycling of this subunit. First, the holoenzyme may remain intact waiting for the synthesis of an accessory protein that either targets the enzyme to the plasma membrane or activates catalytic activity once it arrives at the plasma membrane. Consistent with this hypothesis, we showed that p29 in catalytically inactive astrocytes is part of a 180 -200-kDa complex found in the endosomes (see accompanying paper (31)). Cyclic AMP induces the transcription of a 5ЈD-II "activating factor," leading to the translocation of 5ЈD-II to the plasma membrane, activation of catalytic activity, and an increase in the apparent molecular mass of 5ЈD-II to the 200 -220-kDa range. Alternatively, the holoenzyme may be disassembled, with one or more subunits shuttled to the lysosomes and p29 combining with other newly synthesized subunits. At least one component of the catalytically active enzyme requires de novo synthesis since ongoing protein synthesis is required to maintain catalytic activity. It is likely that this latter component is the cAMP-inducible activating factor.
Previous work on the T 4 -dependent regulation of 5ЈD-II activity identified the inactivation/internalization pathway as the primary site of action of T 4 (6 -8, 23, 24). Indeed, when measuring catalytic activity, T 4 markedly increases the rate of inactivation of enzyme activity without affecting the production rate of 5ЈD-II activity (6). However, analysis of the effects of T 4 on the turnover of the p29 subunit revealed a more complex production pathway for catalytically active 5ЈD-II. Brefeldin A, a fungal toxin that disconnects the secretory, recycling, and degradation pathways for membrane proteins, decreased both the production rate and inactivation rate of the catalytically active enzyme in euthyroid cells, but had little effect on the production rate in hypothyroid cells. These data FIG. 6. Immunocytochemistry of the intracellular transit of 5D-II. Astrocytes were grown on coverslips, and steady-state levels of 5ЈD-II activity were induced as described under "Experimental Procedures." Cells were affinity-labeled with 10 nM BrAcT 4 for 20 min and then chased with either 10 nM T 4 or no hormone for up to 2 h. Cells were fixed and stained with an anti-p29 IgG, and immune complexes were visualized by indirect immunofluorescence. Images were obtained by confocal microscopy at the level of the nucleus. Shown are representative images obtained from 30 -40 images per condition. A, thyroid hormone-deficient, affinity-labeled cells, no chase period; B, thyroid hormone-deficient, affinity-labeled cells treated with T 4 for 20 min; C, thyroid hormone-deficient, affinity-labeled cells treated with T 4 for 2 h; D, thyroid hormone-deficient, affinity-labeled cells incubated for 2 h without hormone. N, nucleus; P, perinuclear space. Marker bars ϭ 10 m.

FIG. 7. Effect of brefeldin A on the intracellular transit of 5D-II.
Astrocytes were grown on coverslips, and steady-state levels of 5ЈD-II activity were induced as described under "Experimental Procedures." Cells were affinity-labeled with 10 nM BrAcT 4 for 20 min and then chased for 2 h in the presence (right panels) and absence (left panels) of BFA (5 g/ml). Cells were fixed and stained with an anti-T 4 IgG, and immune complexes were visualized by indirect immunofluorescence. Images were obtained by confocal microscopy at the level of the nucleus. Marker bar ϭ 10 m. Upper panels, cells chased in serumfree medium with 10 nM T 4 ; lower panels, cells chased in serum-free (SF) medium without hormone.
indicate that the production of catalytically active 5ЈD-II differs in the presence and absence of T 4 .
One possible explanation for this dichotomy is that the production of 5ЈD-II in the presence of T 4 uses both de novo synthesis and preformed subunits recycled from the endosomes. In euthyroid astrocytes, we found that BFA prevented the return of 5ЈD-II to the plasma membrane and led to an accumulation of 5ЈD-II polypeptide(s) in intracellular membrane pools (Fig. 7). In contrast, in hypothyroid cells, all components of catalytically active 5ЈD-II are produced de novo, and BFA did not affect the production rate. Since at least one component of catalytically active 5ЈD-II needs to be newly synthesized in both cases, it appears that fusion of vesicles from the Golgi and endosomal pool as well as de novo synthesis of all 5ЈD-II subunits contribute to the production of active 5ЈD-II in euthyroid cells, while only the latter process contributes to production of active 5ЈD-II in hypothyroid cells. Thus, our initial finding that steady-state production rates of enzyme activity are unaffected by T 4 (6) suggests that the rate-limiting synthetic step is the same in the presence and absence of T 4 .
Our current model of the regulatory events that modulate 5ЈD-II activity is shown in Fig. 8. The synthetic pathway includes the production of 5ЈD-II polypeptides and a putative activating factor that leads to translocation of the enzyme to the plasma membrane and activation of catalytic activity. This activating factor may be a component of 5ЈD-II itself or simply an accessory/ chaperon protein. As shown in Fig. 8A, T 4 initiates inactivation of 5ЈD-II, located on the cytoplasmic surface of the plasma membrane (7), by promoting the binding of the enzyme to the F-actin cytoskeleton (6, 7). The enzyme is then transported to the endosomes by an actin-mediated mechanism (7, 8), presumably in-volving a molecular motor protein (25,26). From the endosomes, 5ЈD-II is shuttled back into the synthetic pathway, where it is recycled back to its active state on the plasma membrane. In hypothyroid cells or in cells treated with transcriptionally active T 3 , the 5ЈD-II polypeptide(s) are internalized by an actin-independent pathway (Fig. 8B) (6, 7). The 5ЈD-II-containing vesicles are then sorted through the endosomes to the degradation pathway, ending up in the lysosomes (8).
The routing of vesicles through the endosomes to degradative, recycling, and synthetic pathways is regulated by targeting signals provided by vesicle-associated proteins located on the vesicle membrane. These vesicle proteins include regulatory coat proteins (17,20), small Ras-like GTP-binding proteins (17), and proteins isolated primarily from synaptic vesicles, including the synapsins and the calcium-binding synaptotagmin (27)(28)(29). The binding of these proteins to, or removal from, vesicles results in the targeting of the vesicles to their ultimate destination. One potential mechanism by which T 4 may regulate vesicle sorting is by binding to one of these regulatory proteins and thus modulating the targeting signals for the p29-containing vesicle. A potential candidate protein to be regulated by T 4 is the synaptic vesicle protein synaptotagmin, which participates in vesicle recycling in the nerve terminal (27,28). In addition to promoting vesicle recycling, synaptotagmin also influences the morphology of the actin cytoskeleton (29,30), two cellular events regulated by T 4 in astrocytes. In summary, the regulation of 5ЈD-II activity in cultured astrocytes suggests that thyroid hormone regulates the basic mechanisms by which cells synthesize, degrade, and recycle proteins.