Inhibition of Heat Shock Protein 90, a Novel RET/PTC1-associated Protein, Increases Radioiodide Accumulation in Thyroid Cells*

RET/PTC1 is a rearranged form of the RET tyrosine kinase commonly seen in papillary thyroid carcinomas. It has been shown that RET/PTC1 decreases expression of the sodium/iodide symporter (NIS), the molecule that mediates radioiodide therapy for thyroid cancer. Using proteomic analysis, we identify hsp90 and its co-chaperone p50cdc37 as novel proteins associated with RET/PTC1. Inhibition of hsp90 function with 17-allylamino-17-demothoxygeldanamycin (17-AAG) reduces RET/PTC1 protein levels. Furthermore, 17-AAG increases radioiodide accumulation in thyroid cells, mediated in part through a protein kinase A-independent mechanism. We show that 17-AAG does not increase the total amount of NIS protein or cell surface NIS localization. Instead, 17-AAG increases radioiodide accumulation by decreasing iodide efflux. Finally, the ability of 17-AAG to increase radioiodide accumulation is not restricted to thyroid cells expressing RET/PTC1. These findings suggest that 17-AAG may be useful as a chemotherapeutic agent, not only to inhibit proliferation but also to increase the efficacy of radioiodide therapy in patients with thyroid cancer.

Members of the heat shock protein family function as molecular chaperones, ensuring the proper folding of newly translated proteins (1). Heat shock protein 90 (hsp90) 1 is an abundant cytosolic protein, which is highly conserved across a number of species. hsp90 mediates the folding of a limited number of client proteins, including steroid hormone receptors and signaling kinases. The ability of hsp90 to facilitate client protein folding requires the presence of co-chaperone proteins, which vary depending on the client protein (2). In the case of client kinases, the interaction with hsp90 is mediated by the co-chaperone p50 cdc37 .
A number of signaling kinases, such as phosphatidylinositol 3-kinase (3), Raf (4), AKT (5), IKK (6), c-Src (7), and ErbB2 (8), have been demonstrated to interact with hsp90. Structural studies of hsp90 have demonstrated the presence of a molecular clamp, whose open or closed state is regulated by the presence of ATP (9). The co-chaperone p50 cdc37 has been shown to interact with the molecular clamp, maintaining hsp90 in the open state and facilitating client protein loading (10). In contrast, hsp90 function can be inhibited by several members of the ansamycin family of antibiotics, including herbimycin A, geldanamycin, and 17-allylamino-17-demethoxygeldanamycin .
Ansamycin antibiotics were thought to function as tyrosine kinase inhibitors (11), however, it was subsequently demonstrated that their molecular target is hsp90 (12,13). Geldanamycin and 17-AAG have been shown to reduce cellular proliferation and induce apoptosis, and interest has risen in their use as anti-neoplastic agents (14). The utility of geldanamycin, however, is limited by significant hepatotoxicity (15). In contrast, 17-AAG is well tolerated in animal studies and has recently entered phase I clinical trials (14).
Geldanamycin and 17-AAG represent novel chemotherapeutic agents for thyroid cancer. Park et al. (16) have demonstrated that geldanamycin decreased expression of c-Raf, inhibited thyroid cell proliferation, reduced Matrigel invasion, and induced apoptosis. Recently, the ability of 17-AAG to decrease thyroid cell proliferation has also been demonstrated, although no increase in apoptosis was observed (17). Thyroid cancer is associated with mutations in several different kinase pathways, including ras and B-raf, and with activation of the AKT pathway (18). In addition, several chromosomal rearrangements involving the RET gene have also been identified in papillary thyroid cancer and named RET/PTC (19).
RET/PTC chimeric oncoproteins have been identified in as many as 40% of papillary thyroid carcinomas in the general population (20) and 60 -70% of cases following the 1986 Chernobyl nuclear disaster (21). All RET/PTC isoforms contain the C-terminal tyrosine kinase domain of the RET linked to the N-terminal portion of an activating gene. These N-terminal proteins are expressed in thyroid follicular cells and are capable of dimerization, resulting in constitutive activation of the RET kinase domain (19). PTC1 is the most common rearrangement seen in the general population, whereas PTC3 was most highly prevalent in short-latency tumors after Chernobyl (22). To develop novel molecular therapies for PTC-induced thyroid cancer, we have conducted a proteomic study to identify novel PTC1-associated factors.
We have identified hsp90 and p50 cdc37 as novel PTC1-associated proteins. Treatment with 17-AAG reduced PTC1 expression levels and increased radioiodide accumulation in thyroid cells, most likely through reduced iodide efflux. The effect of 17-AAG on radioiodide accumulation required preservation of the PKA signaling pathway, which maintained NIS expression at the cell surface. These results suggest that 17-AAG may be useful as a chemotherapeutic agent, not only to inhibit proliferation but also to increase the efficacy of radioiodide therapy in patients with thyroid cancer.
Generation and Immunoprecipitation of FLAGPTC Constructs-To construct the FLAG-C1 vector, two complementary primers were designed based on the FLAG sequence, flanked by PinA1 and BspE1 sites (5Ј-ACC GGT CGC CAC CAT GGA CTA CAA GGA CGA CGA CGA CAA GT and 5Ј-TCC GGA CTT GTC GTC GTC GTC CTT GTA GTC CAT GGT GGC GA). EGFP-C1 vector (Clontech) was digested with PinA1/ BspE1 to remove the green fluorescent protein sequence, and the annealed FLAG primers were ligated into the vector backbone.
FLAGPTC1 was generated by subcloning PTC1 from pRC-CMV-PTC1 using EcoRV and DraI sites into FLAG-C1 vector, which had been cut with XhoI and filled in. FLAGPTC3 was created by subcloning PTC3 from pcDNA3-PTC3 using HindIII and EcoRV into FLAG-C1 cut with HindIII and SmaI. For FLAGPTC1 Y/F mutants, PTC1Y/F DNA fragments were released with EcoRI digestion from pBS-PTC1⌬E, pBS-PTC1⌬P, and pRC-CMV-PTC1⌬G, and then subcloned into FLAGPTC1 vector cut with EcoRI.
Transfection of FLAGPTC constructs was performed using Lipo-fectAMINE 2000 (Invitrogen). After 15 h, immunoprecipitation was performed with FLAG-M2-agarose (Sigma) using 500 -1500 g of whole cell extract (0.5% Nonidet P-40, 10 mM Tris (pH 7.5), and 100 mM NaCl). Equal amounts of extracts were incubated with 40 l of FLAG-M2agarose (Sigma) for 2 h at 4°C. The pellets were washed twice and eluted using 150 ng/l 3ϫ FLAG peptide (Sigma) in TBS at 4°C for 30 min. The eluted peptides were separated on polyacrylamide gels. Coomassie Blue staining was performed using GelCode Blue Stain Reagent (Pierce).
MALDI-TOF Analysis of FLAGPTC-associated Proteins-SDS-gel sample bands were cut from Coomassie-stained gels and submitted the Proteomics Core Facility. The samples were transferred to the MassPrep station (PerkinElmer Life Sciences) for automated in-gel protein digestion, using WinPREP Multiprobe II software. Briefly, gel pieces were first de-stained with ammonium bicarbonate/acetonitrile and reduced with dithiothreitol. The samples were then alkylated with iodoacetamide, washed with ammonium bicarbonate, and dehydrated with acetonitrile. In-gel digestion was carried out with 6-ng/l trypsin (Promega) in 50 mM ammonium bicarbonate for 5 h at 37°C. The digested peptides were washed and concentrated with ZipTip C18 tips. The samples were eluted directly onto the Micromass MALDI-TOF target plate with 2 l of matrix solution (2 mg/ml recrystallized ␣-cyano-4-hydroxycinnamic acid in 50% acetonitrile/0.2% trifluoroacetic acid) and air-dried.
Mass spectra of the resulting peptides were recorded on the MALDI-TOF spectrometer in reflectron mode. Prior to data collection, calibration was performed using the following standards: angiotensin I (average molecular mass 1296.5Da), renin (average molecular mass, 1759.0 Da), and adrenocorticotropic hormone 18 -39 clip (average molecular mass, 2465.199 Da). Resulting peptides were matched with their corresponding proteins with the Protein Prospector (available at prospector.ucsf.edu/) by searching the non-redundant data base maintained at the NCBI (available at www3.ncbi.nlm.nih.gov/). The following parameters are used for the search: mass tolerance, 0.1 Da; allowed incomplete cleavages, 1; acetylation of the N terminus, alkylation of cysteine, and oxidation of methionine were considered as possible modifications.
Generation of PC-Cl3 Cell Lines with Constitutive and Doxycyclineinducible Expression of PTC1-To generate a constitutively expressing cell line, PTC1 was cloned into the pLNCX vector (Clontech) and transfected into Phoenix retroviral packaging cells. After 48 h, the media was harvested, centrifuged at 1000 rpm to pellet producer cells, and then filtered through a 0.22-m filter. PC-Cl3 cells transduced with PTC1 retrovirus were selected with 400 g/ml G418 (Invitrogen) and screened by Western blot analysis.
To generate a PC-Cl3 cell line with inducible expression, PTC1 was cloned into the multiple cloning region of pTRE (Clontech). Phoenix cells were transfected with either pCMV-TetOn (Clontech) or pTRE-PTC1, and retrovirus was collected. PC-Cl3/pCMV-TetOn cells were selected with 400 g/ml G418 and screened with transient transfection of TRE-luciferase. A highly inducible clone was subsequently infected with pTRE-PTC1 retrovirus to generate PC-Cl3/TetOn-PTC1 cell lines. Clones were selected with 200 g/ml hygromycin and expanded in media containing 200 g/ml G418 and 100 g/ml hygromycin.
Generation of HeLa Cell Lines with Doxycycline-inducible Expression of NIS-Human NIS was cloned into the multiple cloning site of pTRE (Clontech). Phoenix cells were transfected with pTRE-hNIS and retroviruses were harvested. HeLa TetOn cells (Clontech) were transduced with TRE-hNIS viral supernatant, and stable clones selected in media containing 250 mg/ml hygromycin (Clontech). The clones were expanded in media containing 100 mg/ml hygromycin.
Radioactive Iodide Uptake and Efflux Studies-When cells were seeded, higher numbers of cells were included in wells treated with 17-AAG, due to its anti-proliferative effect. After treatment with 17-AAG, one well was counted for each treatment group, and the remaining wells were incubated with media containing 2.0 Ci of Na 125 I in 5 M non-radioactive NaI for 30 min at 37°C in 5% CO 2 . Cells were washed twice with ice-cold HBSS and lysed with cold 95% ethanol. The cell lysate was collected and counted for 125 I activity.
For efflux studies, the cells were incubated with media containing Na 125 I as described above. They were then washed twice and incubated in 1 ml of HBSS plus 5 M non-radioactive NaI. After 2 min, the buffer was collected, and fresh HBSS plus 5 M non-radioactive NaI was added. This was repeated every 2 min for a total of 10 min, after which the cells were lysed. The total uptake for each well was calculated as the sum of the efflux washes and the lysate.
Preparation of Cell Lysate and Western Blot Analysis-Cells were washed twice in phosphate-buffered saline and lysed in 0.5% Nonidet P-40, 100 mM NaCl, and 10 mM Tris (pH 7.5) containing 1:500 Protease Inhibitor Mixture Set III (Calbiochem), 1 mM NaF, and 1 mM sodium orthovanadate. The extract was clarified by centrifugation at 14,000 rpm for 20 min at 4°C. Protein concentration was determined by the Bradford method, and the lysate was separated on a 4 -15% polyacrylamide gel and transferred to a nitrocellulose membrane. Membranes were blocked in 5% dry milk plus TBST (10 mM Tris (pH 8.0), 150 mM NaCl, 0.1% Tween 20), and Western blot to detect PTC1 was performed as previously described (27). Antibodies against hsp90, p50 cdc37 , PLC␥ were obtained from Santa Cruz, and FLAG M2 monoclonal antibody was purchased from Sigma.
Crude, post-nuclear membrane fractions were prepared as described (28), and 2 g was resolved on a 4 -15% polyacrylamide gel. Western blot analysis was performed using PA716 polyclonal rabbit ␣-rat NIS antibody diluted 1:2500 (a kind gift from Dr. Bernard Rousset, INSERM, Lyon, France). Antibody against the E subunit of V-ATPase (kindly provided by Dr. Beth Lee, The Ohio State University) was diluted 1:1000.
Immunofluorescence-TetOn-PTC1 or parental PC-Cl3 cells (2 ϫ 10 4 ) were seeded in 4-well chamber slides (Fisher Scientific). For wells treated with 17-AAG, 3 ϫ 10 4 cells were seeded. After 24 h, 2 g/ml doxycycline was added. 12 h later, 17-AAG was added to a final concentration of 3 M for 24 h. The cells were washed with phosphatebuffered saline, fixed in 4% paraformaldehyde for 15 min, and blocked with 4% donkey serum for 1 h. The slides were incubated for 1 h with a PA716 ␣-NIS rabbit polyclonal antibody diluted 1:2500, followed by Cy3-conjugated donkey ␣-rabbit IgG antibody (Jackson ImmunoResearch) diluted 1:250 for 30 min. Cells were visualized using a fluorescent microscope equipped with a 63ϫ oil immersion objective lens.
Statistical Analysis-Statistical comparison of radioactive iodide uptake and efflux was performed using the paired t test test. A p value Ͻ0.05 was considered to be statistically significant.

RESULTS
Characterization of FLAGPTC Constructs-To facilitate coimmunoprecipitation of PTC-associated proteins, we generated FLAGPTC1 and FLAGPTC3 constructs. The FLAG tag was placed at the N-terminal region of the PTC proteins to avoid interference with the C-terminal kinase domain. Following transfection into 293 cells, FLAGPTC1 and FLAGPTC3 expression was confirmed by Western blot (Fig. 1A), using ␣-FLAG M2 antibody or ␣-RET/PTC C1 antibody. In addition, co-immunoprecipitation using ␣-FLAG M2-agarose followed by Western blot demonstrated that PLC␥, a known PTC1-interacting protein (29), is associated with FLAGPTC1 (Fig. 1B).
To determine whether PTC1 was associated with hsp90 and p50 cdc37 in thyroid cells, we transiently transfected ARO human thyroid cancer cells with FLAGPTC1 or FLAG-C1 plasmids. Western blot analysis of ␣-FLAG M2 immunoprecipitates demonstrated that both hsp90 and p50 cdc37 were associated with FLAGPTC1 but not with FLAG-C1 vector (Fig. 2B). In addition, we explored the requirement for specific tyrosines in the kinase domain of PTC1 to mediate interaction with hsp90 and p50 cdc37 . Mutations in tyrosines 294, 404, and 451 were studied, due to their well characterized roles in differentially mediating downstream signaling (19). Single phenylalanine mutants of these tyrosines in PTC1 showed no decrease in their ability to interact with hsp90 and p50 cdc37 , compared with wild-type PTC1 (Fig. 2B).
Inhibition of hsp90 Function with 17-AAG Reduces PTC1 Protein Levels-To determine the effect of inhibiting hsp90 function on PTC1, we employed PC-Cl3-TetOn/PTC1 cells (TetOn/PTC1), a rat immortalized thyroid cell line in which PTC1 expression is induced by doxycycline. Treatment with 17-AAG for 24 h resulted in a dose-dependent decrease in the expression level of PTC1 (Fig. 3A). The decrease in PTC1 expression was not the result of reduced levels of hsp90, as hsp90 levels were unchanged by 17-AAG.
The temporal profile of PTC1 expression in response to 17-AAG was assessed by inducing PTC1 expression for 12 h, followed by the addition of 17-AAG to a final concentration of 3 M for varying lengths of time. A rapid decrease in PTC1 expression levels occurred within 6 h of 17-AAG treatment (Fig. 3B, lane 2 versus 3). Due to continued induction of PTC1 by doxycycline, a slight increase in PTC1 levels occurred at 24 h, relative to the 6-h exposure (Fig. 3B, lane 3 versus 5). Taken together, inhibition of hsp90 function with 17-AAG results in a rapid decrease in PTC1 expression levels.
17-AAG Increases Radioiodide Accumulation in Thyroid Cells-Expression of PTC1 has been previously reported to decrease NIS expression (31,32) and radioiodine accumulation (RAIU) (33). Indeed, we found that NIS protein levels (Fig. 4B,  lane 1 versus 3) and RAIU (Fig. 4A) were reduced in TetOn/ PTC1 compared with parental PC-Cl3 cells. Treatment with 17-AAG significantly increased radioiodide accumulation in TetOn/PTC1 cells (p Ͻ 0.05), and perchlorate inhibition (33) demonstrated that the radioiodide uptake following 17-AAG treatment was mediated by NIS. However, following 17-AAG treatment, NIS protein levels were decreased in PC-Cl3 cells and unchanged in TetOn/PTC1 cells, despite an increase in RAIU. Thus, the increase in RAIU by 17-AAG was not the result of increased amount of NIS protein. Interestingly, 17-FIG. 1. Expression of FLAGPTC constructs. A, FLAGPTC proteins can be detected by Western blot using M2 antibody against FLAG or C1 antibody against PTC. 293FT cells were transfected with the indicated DNA constructs, and Western blot was performed using 50 g of whole cell extract (WCE). B, an N-terminal FLAG tag does not interfere with the ability of PTC1 to co-immunoprecipitate PLC␥ in 293FT cells. Western blot for total PLC␥ was performed using 50 g of WCE and FLAG immunoprecipitation was performed using 1.5 mg of WCE.
FIG. 2. hsp90 and p50 cdc37 are novel PTC1-associated factors. A, Coomassie Blue staining shows four sets of PTC-associated proteins in 293FT cells: p90A-C, p70A-B, p50, and p30 (arrows). Asterisks indicate the location of FLAGPTC proteins. Co-immunoprecipitation was performed using 2.5 mg of WCE extract from transfected 293FT cells. B, hsp90 and p50 cdc37 are associated with FLAGPTC1 in ARO human thyroid cancer cells. Mutations in tyrosines 294, 404, and 451 do not affect the association of FLAGPTC1 with hsp90 and p50 cdc37 . Co-immunoprecipitation was performed using 500 g of whole cell extract (WCE), and Western blot analysis was performed using 20 g WCE. AAG increased RAIU in TetOn/PTC1 cells to a level greater than that of parental PC-Cl3 cells.

17-AAG Increases RAIU in Part through a PKA-independent Mechanism-NIS-mediated RAIU is determined by NIS expression levels at the cell surface. Withdrawal of TSH has
been demonstrated to reduce cell surface NIS expression (34), and inhibition of PKA results in decreased NIS cell surface trafficking. 2 Indeed, treatment of parental and TetOn/PTC1 cells with myristoylated PKI, a PKA inhibitor, resulted in reduced RAIU (Fig. 5A) and decreased expression of cell surface NIS (Fig. 5B). Following 17-AAG treatment, however, NIS surface expression was unchanged in both TetOn/PTC1 and parental PC-Cl3 cells, compared with untreated cells. Treatment with PKI prior to the addition of 17-AAG reduced both RAIU (p Ͻ 0.05) and NIS cell surface expression. Regardless, RAIU was higher in cells treated with both PKI and 17-AAG (Fig. 5A) compared with those treated with PKI alone (p Ͻ 0.05), suggesting that 17-AAG increases RAIU in part through a PKA-independent mechanism.
17-AAG Decreases Radioiodide Efflux-RAIU represents the steady-state accumulation of radioiodide, the net outcome of uptake and efflux. Although perchlorate inhibition demonstrates that the uptake of iodide following 17-AAG treatment is mediated by NIS, 17-AAG did not increase either NIS protein levels or NIS cell surface expression. We thus examined the possibility that 17-AAG may increase RAIU by decreasing iodide efflux (Fig. 6). In the absence of 17-AAG, the rate of iodide efflux in parental cells was slightly faster than in TetOn/PTC1 cells. Treatment with 17-AAG, however, significantly reduced iodide efflux in both TetOn/PTC1 and parental PC-Cl3 cells. Consistent with the results seen in RAIU (Fig. 4A), iodide efflux was markedly slower in 17-AAGtreated TetOn/PTC1 cells. These results suggest that the increase in RAIU induced by 17-AAG is contributed by decreased iodide efflux.

17-AAG Reduces PTC1 Protein Levels and Increases RAIU in PC-Cl3 Cells Constitutively
Expressing PTC1-We wanted to verify that the decrease in PTC1 protein levels and increase in RAIU by 17-AAG was not the result of a clonal effect in the TetOn/PTC1 cell line. Thus, we examined the effect of 17-AAG on the PC-Cl3/PTC1 cell line, which constitutively expresses PTC1. The addition of 17-AAG dramatically reduced PTC1 protein levels in PC-Cl3/PTC1 cells (Fig. 7A). In addition, 17-AAG significantly increased RAIU in PC-Cl3/PTC1 cells (p Ͻ 0.05), similar to that in TetOn/PTC1 cells (Fig. 7B). These results demonstrate that the ability of 17-AAG to decrease PTC1 and increase RAIU was neither a clonal effect nor an artifact of the TetOn system.  RAIU by 17-AAG was not restricted to PC-Cl3 cells. As shown in Fig. 8A, 17-AAG also increased RAIU in the immortalized FRTL-5 rat thyroid line (p ϭ 0.06). In contrast, 17-AAG did not increase RAIU in two human thyroid cancer cell lines, ARO and TPC1, which had lost NIS-mediated radioiodine accumulation (Fig. 8A). We also compared the effect of 17-AAG in COS-7 cells transiently transfected with NIS and in HeLa-TetOn/NIS cells. In these two non-thyroid cell lines expressing exogenous NIS, 17-AAG did not increase RAIU (Fig. 8B). In addition, treatment with 17-AAG did not alter the rate of iodide efflux in HeLa-TetOn/NIS cells (Fig. 8C).

DISCUSSION
In this study, we have identified hsp90 and p50 cdc37 as novel proteins associated with PTC1 and demonstrated that inhibition of hsp90 function with 17-AAG reduces PTC1 protein levels. Furthermore, treatment with 17-AAG increased radioiodide accumulation in thyroid cells, and this increase was not restricted to PTC1-expressing cells. The ability of 17-AAG to increase RAIU was unrelated to NIS expression level and appeared to be contributed by decreased iodide efflux. These findings suggest that 17-AAG may be useful as a chemotherapeutic agent, not only to inhibit proliferation but also to increase the efficacy of radioiodide therapy in patients with thyroid cancer.
hsp90 functions as a chaperone for a limited group of proteins, including kinases, such as c-Src and AKT, and transcription factors, such as p53. It has been demonstrated that the affinity of hsp90 is higher for v-src, the oncogenic mutant form of the protein, than for the cellular homolog, c-src (7). Similarly, hsp90 does not interact well with native p53 protein but is critical for stabilizing the mutant form of the protein (35). In this context, it is not surprising that PTC oncoproteins interact with hsp90. It is likely that a chimeric PTC oncoprotein would be inherently unstable, with a short half-life. This is supported by the rapid decrease in PTC1 protein levels observed following inhibition of hsp90 function with 17-AAG. The dependence of oncogenic proteins on the chaperone function of hsp90 may partially explain why inhibitors such as 17-AAG are more cytotoxic in tumor cells than in normal cells.
The interaction between hsp90 and PTC1 does not appear to require activation of the tyrosine kinase of PTC1. In our study, we found that Y/F mutants of PTC1 in tyrosines 294, 404, and 451 interacted equally well as wild-type PTC1 with hsp90. Tyrosine 294 is located in the kinase domain of PTC1, in a position conserved within tyrosine kinases (36). Mutation of this tyrosine to phenylalanine results in loss of transforming ability in NIH3T3 cells and significantly decreases autophosphorylation and the phosphorylation of other cellular proteins (37). These findings are consistent with the current model for hsp90 function, in which hsp90 mediates the folding of a limited set of nascent, inactive proteins (38).
As PTC1 has been shown to decrease NIS expression and function (32,33), we anticipated that 17-AAG would increase RAIU by increasing NIS expression and function. Surprisingly, NIS levels were decreased in PC-Cl3 and unchanged in TetOn/ PTC1 cells by 17-AAG. In addition, 17-AAG increased RAIU not only in TetOn/PTC1 cells but also in parental PC-Cl3 cells. This finding indicates that the increase in RAIU by 17-AAG is not solely due to decreased PTC1 expression. Furthermore, the increase in radioiodine accumulation in TetOn/PTC1 cells by 17-AAG was consistently greater than or equal to that of parental PC-Cl3 cells. It is interesting to note that 17-AAG decreased NIS expression in parental PC-Cl3 cells, suggesting that hsp90 may function as a chaperone for NIS. Previously, it has been demonstrated that hsp90 regulates the protein levels of other integral membrane proteins. Ficker et al. (39) have reported that the cardiac potassium channel hERG is associated with hsp90, and treatment with geldanamycin blocked formation of the mature, glycosylated protein. In the case of the cystic fibrosis transmembrane receptor, hsp90 is also essential for glycosylation and maturation (40). Following treatment with geldanamycin, steady-state levels of cystic fibrosis transmembrane receptor protein declined, whereas cell surface expression was unchanged. Likewise, we found that 17-AAG reduced NIS steadystate levels in PC-Cl3 cells while maintaining cell surface localization. Thus, whereas hsp90 may be essential for NIS folding, the increase in RAIU by 17-AAG is not due to an increase in NIS expression levels.
The increase in radioiodide accumulation by 17-AAG in thyroid cells appears to involve a PKA-independent mechanism and a decreased rate of iodide efflux. Gerard et al. (41) have reported that forskolin, a PKA activator, had no effect on iodide efflux and TSH treatment weakly increased the rate of iodide efflux in cultured porcine thyroid follicles. These results suggest that iodide efflux is regulated primarily through PKAindependent pathways. Instead, iodide efflux was stimulated by TPA, a protein kinase C (PKC) agonist. In addition, activation of phospholipase C (PLC) and PKC are required for ATPinduced increase of iodide efflux in FRTL-5 cells (42). The decreased iodide efflux by 17-AAG was more pronounced in TetOn/PTC1 than parental PC-Cl3 cells. Interestingly, PTC1 activates both PLC␥ and PKC⑀ (29,43). Thus, it is possible that 17-AAG decreases iodide efflux in thyroid cells through downregulation of the PLC/PKC pathway.
The ability of 17-AAG to down-regulate kinase activity, however, is selective. Basso et al. (5) have reported that 17-AAG reduces protein levels of AKT and PDK1 in the MCF-7 human breast cancer cells without affecting the expression of S6K, SGK-1, several isoforms of PKC, or the catalytic subunit of PKA. It is noteworthy that PKA activity also appears to be intact following 17-AAG treatment in rat immortalized thyroid cells. Inhibition of PKA function with PKI resulted in reduced NIS protein at the cell surface, either in the presence or in the absence of 17-AAG (Fig. 5B), whereas treatment with 17-AAG alone had no effect on NIS cell surface localization. Thus, PKA signaling is preserved in thyroid cells following treatment with 17-AAG.
The increase of radioiodine accumulation by 17-AAG occurred in immortalized thyroid cells with endogenous NIS function. We found that 17-AAG increases RAIU by decreasing iodide efflux without increasing the amount of NIS protein. Consistent with this mechanism, 17-AAG did not increase RAIU in two human thyroid cancer cell lines that have lost NIS expression. However, the inability to increase RAIU in non-thyroid cells suggests that the effect of 17-AAG on iodide efflux may be unique to certain cell types, including thyroid cells.
Thyroid cancers of follicular origin typically respond poorly to chemotherapy (44). Treatment of thyroid cell lines with geldanamycin derivatives, however, has been shown to have a number of therapeutic effects, including inhibition cell growth, reduced invasiveness, and increased apoptosis (16,17). Because the majority of thyroid tumors maintain TSH-inducible RAIU, radioiodide is used as the primary tool for the detection and treatment of recurrent and metastatic thyroid cancer. However, some thyroid tumors fail to accumulate radioiodine to a level sufficient for effective therapy, and much effort is ongoing to increase RAIU in thyroid tumors (45,46). Thus, it is of clinical significance that 17-AAG increases RAIU in thyroid cells expressing endogenous NIS. In addition, our results suggest that 17-AAG may increase RAIU in thyroid tumors to a greater extent than normal thyroid tissue. Taken together, 17-AAG represents a novel class of chemotherapeutic agents for thyroid cancer, not only for its antineoplastic effects but also for its ability to enhance the efficacy of radioiodide therapy.