Inflammatory Cytokine Regulation of Fas-mediated Apoptosis in Thyroid Follicular Cells*

The occurrence of apoptosis in thyroid follicular cells induced by Fas activation has been a subject of much debate. This is due, in part, to the fact that no physiologically relevant treatment conditions have been reported to cause rapid and extensive Fas-mediated apoptosis in thyroid cells, whereas treatment with the protein synthesis inhibitor cycloheximide prior to Fas activation allows for massive cell death. This indicates that the Fas signaling pathway is present but that its function is blocked in the overwhelming majority of cultured thyroid cells. To reconcile the conflicting reports, we set out to identify physiologically relevant conditions in which rapid, massive thyroid cell apoptosis in response to Fas activation could be demonstrated. We determined that susceptibility to Fas-activated apoptosis could be influenced by certain combinations of inflammatory cytokines. Although no single cytokine was effective, pretreatment of thyroid cells with the combination of γ-interferon and either tumor necrosis factor-α or interleukin 1β allowed for massive Fas-mediated apoptosis. Susceptibility to Fas-induced death correlated with an increase in expression of a tunicamycin-inhibitable high molecular weight form of Fas but not with aggregate expression of Fas.

Apoptosis is an important mechanism in mammalian development, homeostasis, and immune response (1,2). The processes involved in apoptosis are tightly regulated, and alterations in their function may result in disorders, including autoimmune disease and cancer (3)(4)(5). Apoptosis may play an important role in thyroid homeostasis and disease. Morphological changes indicative of apoptosis have been observed in normal thyroid glands and appear at increased frequency in thyroid tissue affected with chronic autoimmune thyroiditis (6 -8). Apoptosis is one mechanism by which cytotoxic T lymphocytes can destroy thyrocytes in thyroiditis, which in turn may lead to hypothyroidism (6,8). In contrast, the suppression of apoptosis may contribute to thyroid proliferative diseases including goiter, cancer, and Graves' disease (7,8). However, little is known about the mechanisms and regulation of apoptotic signaling in thyroid cells. It is important to define the signaling components of apoptosis present in thyroid follicular cells because they may provide insights into potential pathogenic mechanisms and lead to development of pharmacological interventions for treatment of thyroid disease.
A potentially important pathway in signaling apoptosis in the thyroid involves the Fas death-inducing receptor. Fas is a type I membrane protein and a member of the tumor necrosis factor receptor family (9,10). It is found constitutively expressed on lymphocytes and has been detected on many cells of nonhematopoeitic origin (9,11). Activation of Fas by Fas ligand (FasL) 1 initiates intracellular signals that result in death of the cell (12). Regulation or modulation of this pathway can occur at multiple levels throughout the pathway. This may include changes in the level of the expression of Fas or its ligand (9,11,13,14), the regulation of components of intracellular signaling (2,(15)(16)(17), and expression of proteins that promote survival, such as members of the Bcl-2 gene family (2,5,16,18,19). The Fas pathway is a major mechanism of T lymphocyte-mediated cytotoxicity (9,20). It is therefore possible that this system plays a role in the pathogenesis of thyroiditis, because cytotoxic T lymphocytes are abundantly present in the thyroid in areas where apoptosis is observed (6).
Three reports have presented different conclusions as to the expression and regulation of Fas and the induction of apoptosis in thyroid cells. Kawakami et al. (21) reported that Fas is constitutively expressed on thyroid cells but suggested that it does not function to induce apoptosis unless the cells are exposed to IFN␥ or IL-1␤. In addition, they reported that TSH regulates the expression of Fas and Fas-mediated apoptosis. In contrast, Giordano et al. (22) suggested that Fas was not constitutively expressed on thyroid cells but that its expression was induced by IL-1␤ in thyroiditis-derived cell cultures leading to autologous apoptosis by thyrocyte expressed FasL. However, these investigators did not use a cell death assay specific for apoptosis (rather than necrosis). In addition, the acute and massive apoptosis that would be expected from autologous Fas activation, as suggested by the Giordano hypothesis, is not observed in thyroiditis. The reliability of reagents used in this report also have been questioned (23)(24)(25). We reported that Fas was constitutively expressed and only minimally regulated by either IFN␥ or IL-1␤ (26) and showed, in direct contrast to the results of Giordano et al. (22), that thyroid cells are resistant to Fas-mediated cell death regardless of treatment with either IFN␥ or IL-1␤, except in the presence of the protein synthesis inhibitor cycloheximide (26). This confirmed the functionality of the Fas cell death pathway in thyroid cells but suggested the presence of a labile inhibitor. In addition, we showed no TSH regulation of Fas-mediated apoptosis.
To resolve the contradictions between these studies, we examined Fas-mediated cell death in thyroid cells by several different methods and under various treatment conditions. We found that the combination of IFN␥ with either TNF␣ or IL-1␤ provides signals that facilitate rapid and extensive Fas-mediated thyroid cell apoptosis. The inflammatory cytokine-induced susceptibility to Fas-mediated apoptosis correlated with expression of a tunicamycin-inhibitable high molecular weight form of Fas. We will discuss possible mechanisms of Fas-induced cell death signaling in thyroid cells and the potential role of Fas in thyroid disease.

EXPERIMENTAL PROCEDURES
Cell Culture-Normal thyroid tissue was obtained from patients at thyroidectomy from the uninvolved, contralateral lobes of thyroids with tumors. All excised tissues were prepared for cell culture as described previously (26). The primary cultures were passaged in CellGro Complete medium (Mediatech, Herndon, VA) supplemented with 20% NuSerum IV (Collaborative Biomedical Products, Bedford, MA), 100 units/ml penicillin, 100 g/ml streptomycin, and TSH (10 milliunits/ ml). All cell culture experiments were performed in the presence of TSH. Thyroglobulin staining was performed as described (26) to determine thyrocyte purity, and only cultures that were Ͼ95% thyroglobulin positive were used for experiments. Cells were treated with cytokines and Abs as described in the figure legends at the following concentrations: 100 units/ml IFN␥ (Roche Molecular Biochemicals), 50 ng/ml TNF␣ (Collaborative Biomedical Products), 50 units/ml IL-1␤ (Sigma), 1 g/ml mouse anti-Fas clone CH11 agonist (Upstate Biotechnology, Lake Placid, NY), 1 g/ml purified IgM (Sigma), 10 g/ml purified goat anti-recombinant human TNF-RI agonist Ab, and 5.0 g/ml purified mouse anti-TNF-RI antagonist monoclonal Ab (both R & D Systems, Minneapolis, MN). Tunicamycin (Sigma) was dissolved to make a stock solution of 0.5 mg/ml in 25 mM Tris-HCl, pH 9.5, and stored at Ϫ70°C.
Determination of Cell Viability, Death, and Apoptosis-Cell viability was determined by MTT assay (27) as described in Ref. 26. Cell death was determined by staining cells with fluorescein diacetate (green fluorescence) and propidium iodide (red fluorescence). 1 ϫ 10 4 cells were acquired for each sample and quantitated on a FACScan flow cytometer, and data were analyzed by CellQuest software (both Becton Dickinson, Franklin Lakes, NJ).
Apoptosis was determined by analysis of DNA fragmentation (TUNEL staining). Flow cytometry analysis of DNA fragmentation was performed as described (28). Green and red fluorescence of 1 ϫ 10 4 cells/sample was determined on a FACScan flow cytometer and analyzed by CellQuest software. Apoptotic cells were defined as having both euploid DNA content as measured by propidium iodide (red fluorescence) staining and DNA fragmentation greater than control cells (TUNEL positive) as measured by anti-bromodeoxyuridine-fluorescein isothiocyanate (green fluorescence) staining. Dead cells were defined as subdiploid, whereas other cells (euploid, TUNEL negative) were considered live.
Determination of Fas Expression-Fas expression was determined by flow cytometry and Western analysis. For flow cytometry analysis, thyroid cells were lifted from 10-cm culture dishes using trypsin/EDTA. After washing with phosphate-buffered saline, 1-2 ϫ 10 6 cells/ml were fixed with 1% formaldehyde in phosphate-buffered saline for 30 min on ice. Cells were spun at 200 ϫ g for 10 min at 4°C and then permeabilized with 0.1% Triton X-100 in phosphate-buffered saline with 0.1% bovine serum albumin and 0.02% sodium azide (PBA). After washing with phosphate-buffered saline, cells were stained with a monoclonal mouse anti-Fas monoclonal antibody clone UB2 (Medical & Biological Laboratories, Co., Watertown, MA). Controls included cells stained using isotype control Ab. After 1 h of incubation, cells were washed with PBA and then incubated with fluorescein isothiocyanate-conjugated anti-mouse IgG (Jackson ImmunoResearch, West Grove, PA) for 1 h. Finally, cells were washed with PBA, and fluorescence was analyzed and quantitated as described above. Western analysis of Fas was performed as described previously (26) using anti-human Fas Ab C-20 (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) at a 1:500 dilution. Results were visualized by ECL (Amersham Pharmacia Biotech) followed by autoradiography and quantitated by densitometry.
RNase Protection Assays-RNA was isolated from cells using Trizol Reagent according to the manufacturer's protocol (Life Technologies, Inc.). RiboQuant MultiProbe RNase Protection Assay System (Pharmingen, San Diego, CA) was used for the detection and quantitation of multiple, specific mRNA species. 32 P-Labeled antisense RNA probes were prepared using the Human Apoptosis hAPO-3c Template Set (Pharmingen), which included probes for human Fas and glyceraldehyde-phosphate dehydrogenase. This was performed as described previously (26) according to the manufacturer's protocol. Transcripts were quantified by autoradiography followed by densitometry. Relative amounts of message were corrected for RNA loading by comparison with the glyceraldehyde-phosphate dehydrogenase band intensity for each sample.
Computer Software-Graphing and calculations of standard deviation were performed using DeltaGraph 4.0 (DeltaPoint Inc., CA). Quantitation of autoradiograms was performed using Quantity One (Bio-Rad).

RESULTS
Inflammatory Cytokines Induce Susceptibility to Fas-mediated Cell Death in Thyroid Cells-To reconcile conflicting reports of inflammatory cytokine regulated Fas-mediated apoptosis in thyroid cells, we used three different methods to analyze cell death. MTT assays were used as an initial screening technique to measure relative cell viability without measuring dead or dying cells. Flow cytometry assays were used to confirm MTT results by simultaneously staining live and dead cells with fluorescein diacetate and propidium iodide. Flow cytometry assessment was also used to differentiate between healthy cells and live cells undergoing apoptosis by staining with bromodeoxyuridine and propidium iodide. Fig. 1A displays the results of a representative MTT viability assay of thyroid cell cultures in response to pretreatment with combinations of inflammatory cytokines with and without subsequent Fas activation. IFN␥, TNF␣ or, IL-1␤ alone had minimal effects on cell viability. Subsequent Fas activation, accomplished by addition of Fas cross-linking clone CH11 monoclonal Ab, did not additionally alter viability compared with control IgM Ab treatment. Significant decreases in MTT values were seen in cultures treated with the combination of IFN␥ with either TNF␣ or IL-1␤ or the addition of all three cytokines but not with TNF␣ combined with IL-1␤. Visual observation of these cultures showed very few rounded up, nonadherent cells, which suggests that the lower MTT values were primarily due to slowed cell growth rather than cell death. However, treatment of cultures with IFN␥, along with either TNF␣ or IL-1␤, followed by Fas activation showed massive loss of viability (Ͼ60%) within 18 h. Visual observation of these cultures showed that many cells had rounded up and become nonadherent, suggesting the loss of viability is due to apoptosis rather than slowed growth.
To quantitate the cell death observed visually, we performed live/dead staining quantitated by flow cytometry. Fig. 1B displays the results as a percentage of live cells for each cytokine pretreatment with and without subsequent Fas activation. Maximal death (Ͼ45% above controls) occurred in cultures treated with the combination of IFN␥ with either TNF␣ or IL-1␤ or with all three cytokines after subsequent Fas activation. Significant death could also be detected in cultures treated with IL-1␤ (24%) or TNF␣ combined with IL-1␤ (26%) after subsequent Fas activation. The only condition where significant death occurred without Fas activation was with the combination of IFN␥, TNF␣, and IL-1␤ (25% above untreated control). These experiments confirmed the earlier observation that IFN␥ combined with either TNF␣ or IL-1␤ does not kill thyroid cells but merely slows growth. The slowed growth also confirms that the inflammatory cytokines are biologically active.
Fas-induced Cell Death Regulated by Inflammatory Cytokines Is Apoptotic-Most recent reports of apoptosis in thyroid cell cultures did not demonstrate a clear distinction between apoptosis and necrosis (21,22,29,30). This distinction is important because the difference in the inflammatory response to necrotic cells compared with apoptotic cells is significant. Apoptosis eliminates cells without an inflammatory response, whereas necrosis elicits an immune response (5). Fig. 1C displays the results of flow cytometry to document that the cell death shown in Fig. 1B was due to apoptosis. Fas activation in these studies was performed for only 7 h before analysis as compared with 18 h in the previous assay (Fig. 1B). This modification was performed to identify cells in the process of apoptosis before actual cell death had occurred. Rapid and massive apoptosis induced by Fas activation occurred only after pretreatment with the combination of IFN␥ and TNF␣ (59% compared with Ͻ1% without Fas activation). The number of dead cells were also greatly increased (19% compared with 2% without Fas activation) in these cultures. A total of 78% of the cells were dead or dying within 7 h of Fas activation under these conditions. Pretreatment with the combination of IFN␥ and IL-1␤ prior to Fas activation also showed significant apoptosis (12% compared with Ͻ1%) and dead cells (5% compared with 1%). The apoptotic response after IFN␥/IL-1␤ treatment is diminished when compared with the same treatments quantitated by live/dead staining in Fig. 1B. This is probably due to the shorter period of clone CH11 Ab exposure (7 h versus 18 h) prior to measurement. No other combination showed more than 7% apoptosis.

Increased Susceptibility to Apoptosis Does Not Correlate with Alterations of Fas Expression by Inflammatory Cytokines-Previous reports have suggested that increased susceptibility to
Fas-mediated cell death in thyrocytes treated with cytokines was due to increased expression of Fas (21,22). To determine whether expression of Fas correlates with susceptibility to apoptosis in our system, we used RNase protection assays, flow cytometry, and Western analysis to determine relative levels of Fas expression after cytokine treatments. The RNase protection assay shown in Fig. 2A displays relative expression of Fas mRNA after cytokine treatment. Quantitation by densitometry shows that Fas mRNA is increased moderately and to the same degree by IFN␥ and TNF␣ (1.2-fold), but it is not significantly further enhanced by the conditions of enhanced cell death susceptibility: the combination of IFN␥ with either TNF␣ (1.4fold) or IL-1␤ (1.5-fold). IL-1␤ alone did not increase Fas mRNA (0.95-fold). Fig. 2B displays the results of flow cytometry determination of Fas expression. We used geometric mean channel fluorescence as a measure of Fas expression. IFN␥ with TNF␣ increased Fas expression 2.2-fold over the untreated control. IFN␥ combined with IL-1␤ increased Fas expression 1.6-fold. TNF␣ combined with IL-1␤ increased Fas expression to a similar extent (1.5-fold) as IFN␥ and IL-1␤ despite significantly less susceptibility to Fas-mediated apoptosis. Western analysis was used to confirm the flow cytometry results. Fig.  2C shows an autoradiogram of the analysis with densitometric quantitation of the bands listed below each lane. Under conditions that increase Fas susceptibility, IFN␥ with TNF␣ increased total Fas expression only 1.51-fold, and IFN␥ with IL-1␤ increased expression 1.34-fold. Conditions that did not increase Fas susceptibility had an increase in total Fas protein expression of between 1.03-and 1.20-fold. A more significant correlation with apoptosis is seen with increased expression of high molecular weight forms of Fas protein. The high molecular weight forms (upper bands in Fig. 2C) increase 7.7-fold and 4.7-fold when cultures are treated with IFN␥ combined with TNF␣ or IL-1␤, respectively. Other cytokine treatments increase high molecular weight Fas expression by only 1.7-2.3-fold. The data in Fig. 2 suggest that an increase in total Fas expression is not responsible for increased cell death susceptibility.

Tunicamycin Inhibits High Molecular Weight Fas Expression and Fas-induced Cell Death-
The correlation between Fasmediated cell death susceptibility and the high molecular weight forms of Fas suggested that glycosylation may be important for its activity. To demonstrate whether the high molecular weight form of Fas is due to glycosylation, we performed Western analysis on lysates from thyroid cells that had been treated with the glycosylation inhibitor tunicamycin simultaneously with addition of IFN␥ and TNF␣. Fig. 3A shows that tunicamycin inhibits the increased expression of the high molecular weight form of Fas that is induced by treatment with IFN␥ and TNF␣. This demonstrates that IFN␥ and TNF␣ promote glycosylation of Fas in thyroid cells. We then investigated whether tunicamycin could prevent Fas-mediated cell death in inflammatory cytokine-treated thyroid cells. Fig. 3B shows the results of viability staining of thyroid cells pretreated with tunicamycin prior to clone CH11 Ab treatment to induce Fas-mediated cell death. Interestingly, tunicamycin alone had considerable cytotoxicity, which could be inhibited by inflammatory cytokine treatment. The addition of clone CH11 Ab to inflammatory cytokine-treated cells increased cell death from 10 to 31% as expected. Tunicamycin was able to reduce this increase by one-half down to 21%.

FIG. 1. Inflammatory cytokines induce susceptibility to Fasmediated apoptosis in thyroid cells.
A, thyroid cells plated at a density of 5000 cells/well in a 96-well microtitre plate were treated for 72 h with inflammatory cytokines. Clone CH11 or control Ab was added for an additional 18 h prior to performing the MTT viability assay. Each treatment was performed in triplicate, and standard deviation was calculated at 95% confidence. This experiment is representative of results from five independent experiments each using thyrocyte cultures derived from different patient samples. B, thyroid cells were plated at a density of 2 ϫ 10 5 /6-cm plate and treated with cytokines for 72 h prior to addition of clone CH11 or control Ab. Adherent cells were harvested by trypsinization and combined with the previously saved nonadherent cells of the same culture 18 h after addition of Ab. Cells were stained for flow cytometry, and 1 ϫ 10 4 cells/sample were assayed for determination of live and dead cells. An asterisk denotes p Ͻ 0.005 (chi-squared test) for clone CH11 Ab treatment when compared with control Ab for each cytokine treatment. This experiment is representative of results from four independent experiments each using thyrocyte cultures derived from different patient samples. C, thyroid cells were plated at a density of 2 ϫ 10 5 cells/6-cm plate and treated for 72 h with inflammatory cytokines prior to treatment with clone CH11 or control Ab for an additional 7 h. Cells were harvested as described in B and prepared for flow cytometry to measure subdiploid (dead) and euploid/TUNEL positive (apoptotic) cells. Only intact cells were used in the analysis as identified by forward and side scatter. An asterisk denotes p Ͻ 0.001 (chi-squared test) for the difference in number of euploid/TUNEL positive cells after clone CH11 Ab treatment when compared with control Ab for each cytokine treatment. This experiment is representative of results from two independent experiments, each using thyrocyte cultures derived from different patient samples.

TNF Receptor 1 Is Involved in Mediating a Fas-mediated Cell
Death Susceptibility Signal-To further clarify the specific pathways involved in thyroid cell susceptibility to Fas-mediated apoptosis, we used specific receptor neutralizing and activating Abs to determine which of the two TNF receptors was responsible for transmitting the signal observed when we added TNF␣. We used an agonist Ab specific for TNF-R1 activation and antagonist Abs for both TNF-R1 and TNF-R2. As shown in Fig. 4A, the agonist Ab for TNF-R1 can replace TNF␣ when combined with IFN␥ to provide susceptibility to Fasmediated cell death. Agonist Ab combined with IFN␥ allows for a 53% reduction in viability by Fas activation, which is comparable with TNF␣ combined with IFN␥ (73%) as measured by MTT viability assay. IFN␥, without the addition of TNF␣ or cytokines for 72 h prior to harvesting. Results of quantitation, in units of intensity, of upper, middle, and lower bands and totals are listed below each lane. This experiment is representative of results from three independent experiments each using thyrocyte cultures derived from different patient samples.

. Tunicamycin inhibits high molecular weight Fas expression and Fas-induced cell death.
A, Western analysis of thyroid cells plated at 2 ϫ 10 5 cells/6-cm plate and treated with inflammatory cytokines and tunicamycin (1 g/ml) for 72 h prior to harvesting cell lysates. An overexposure of the tunicamycin and IFN␥/TNF␣ treatment lane is included at the far right to emphasize the lack of the high molecular weight form under these conditions. This experiment is representative of results from two independent experiments, each using thyrocyte cultures derived from different patient samples. B, cell viability of thyroid cultures plated at 2 ϫ 10 5 cells/6-cm plate and treated with inflammatory cytokines and tunicamycin (1 g/ml) for 48 h prior to incubation with clone CH11 or control Ab for an additional 18 h. Nonadherent cells were washed off, and medium was replaced before addition of Ab. Viability was measured as in Fig. 1B. Only intact cells (Ͼ3500/sample) were used in the analysis as identified by forward and side scatter. An asterisk denotes p Ͻ 0.001 (chi-squared test) for the difference in viability of tunicamycin treated cells compared with cells not treated with tunicamycin. This experiment is representative of results from three independent experiments each using thyrocyte cultures derived from different patient samples.
agonist Ab, provides only a 30% loss of viability, and agonist Ab alone has no effect on Fas-mediated cell death. Also, the addition of an antagonist Ab for TNF-R1 was able to block Fasmediated apoptosis of cells pretreated with IFN␥/TNF␣ (Fig.  4B). This clearly shows the involvement of TNF-R1 signaling in providing Fas susceptibility. The antagonist Ab for the other known receptor for TNF␣, TNF-R2, did not show any ability to inhibit apoptosis under the same conditions. 2

DISCUSSION
Apoptosis clearly plays an important role in autoimmune chronic (Hashimoto's) thyroiditis, a disorder that often results in hypothyroidism. The inability of researchers to more precisely define the pathogenesis of autoimmune thyroiditis suggests a great complexity to this disease. The contradictory reports of Fas expression and Fas-mediated apoptosis in thyroid cells re-enforces this complexity. In this paper, we report data that may suggest an explanation for some of these inconsistencies. We set out to determine physiologically relevant conditions in which rapid and extensive Fas-mediated apoptosis could occur. The combination of IFN␥ with either TNF␣ or IL-1␤ provided susceptibility to nearly complete and total apoptotic cell death through Fas activation. This massive apoptosis occurs within 7 h of activation. Greater than 78% of thyroid cells are either apoptotic or dead as measured by flow cytometry. This is consistent with Fas-mediated cell death signaling in other systems (10,31), but this is in stark contrast to data provided by Giordano et al. (22), who required 72 h to reach comparative levels of thyroid cell death after pretreatment with IL-1␤, and Kawakami et al. (21,29), who found Յ40% dead cells after 48 h pretreatment with IFN␥ and Ͻ20% after TSH and IFN␥, followed by 18 h of Fas activation. In addition, all of our experiments were performed in the presence of TSH, suggesting that this growth factor may not protect against Fas-mediated cell death. In our system, combinations of inflammatory cytokines other than IFN␥ with either TNF␣ or IL-1␤ provide much less significant susceptibility. The need for this synergistic effect to induce apoptosis suggests an explanation for the contrasting reports of IL-1␤ induction of Fas susceptibility. It is possible that cultures of thyroiditis-derived cells contained contaminating, activated, IFN␥-producing T lymphocytes known to be present in the intact thyroid (32). Therefore, when IL-1␤ was added to the cultures, the cells would become susceptible to Fas-mediated apoptosis because of the provision of the two signals required for this to happen. Other inflammatory cytokines may also be present to varying degrees depending on the presence of contaminating lymphocytes (IFN␥ and TNF␣) or macrophages (TNF␣ and IL-1␤) that produce these cytokines. Therefore we took great care to remove all nonadherent cells from our cultures by repeated washings. Reverse transcription-polymerase chain reaction analysis to detect rearranged T-cell receptor genes confirmed the absence of those cells when this method is used. Despite this care, we encountered considerable variability in IFN␥-induced susceptibility to Fas-mediated apoptosis. This could be due to prior in vivo exposure to or the variable presence of contaminating cells producing either TNF␣ and/or IL-1␤.
It is apparent from our data that both TNF␣ and IL-1␤ provide signals that synergize with IFN␥ for inducing Fasmediated death susceptibility, and it is possible that they may provide the same signal. Activation of their respective receptors generate overlapping signals including the secondary messengers ceramide and nitrous oxide and the transcription activator NFB (33)(34)(35)(36). Ceramide alone kills thyroid cells, whereas fumonisin, an inhibitor of an enzyme in the ceramide pathway, does not affect Fas-mediated death. 2 This suggests that this pathway is not involved with providing susceptibility to Fas-mediated cell death. The specific intracellular pathways involved in this process remain to be determined. One clue, as we have discerned, is that TNF-R1 activation can generate one of the two required signals.
One important question that remains to be satisfactorily answered regarding Fas-mediated death in the thyroid is whether FasL is expressed and whether it can induce thyroid cell suicide in cells expressing Fas. Our data suggest that it is unlikely that significant amounts of FasL are expressed by thyroid cells because this would cause massive death by suicide/fratricide in our experiments, without the addition of Fasactivating Ab. The change in cell viability by treatment with IFN␥ combined with TNF␣ or IL-1␤, without Fas activation (as measured by MTT assay) is clearly not apoptosis in the flow cytometry assays. If FasL was constitutively expressed or induced by inflammatory cytokines, as suggested by some researchers (37), then we would expect massive thyroid cell death after IFN␥/TNF␣ treatment. Treatment of cells with a Fas 2 J. Bretz, unpublished observation.
FIG. 4. TNF-R1 is involved in the Fas cell death susceptibility signal pathway. Thyroid cells were plated for MTT viability assays as described in Fig. 1A. Cells were then treated with combinations of inflammatory cytokines and TNF-R1 agonist Ab (A) or antagonist (B) for 48 h prior to clone CH11 Ab treatment for an additional 24 h. This experiment is representative of results from two independent experiments each using thyrocyte cultures derived from different patient samples.
antagonist Ab (clone ZB4) did not inhibit the loss of viability induced by IFN␥ combined with TNF␣ or IL-1␤ treatment, 2 confirming a lack of Fas activation. Inconsistent reports of Fas-induced apoptosis in thyroid cells may be due to two other factors. Some of the apoptosis activating Fas antibodies, clone DX2 an IgG isotype for example, do not induce apoptosis in thyroid cells. 2 In our hands only clone CH11, an IgM antibody, sufficiently cross-links to activate Fas signaling. Also, differences in interpretation of results because of detection of cell death by different methods may overestimate or underestimate the effect of different agents.
The question remains as to why an inflamed thyroid is not destroyed upon being infiltrated by activated FasL-expressing immune cells. There are situations in which the thyroid is infiltrated by immune cells but is not destroyed (Graves' disease and post-partum thyroiditis). One answer might be that the cellular immune response differs in that IFN␥and TNF␣producing helper 1 subset of T lymphocytes are predominant in Hashimoto's thyroiditis, whereas helper 2 subset T lymphocytes (IL-4-producing) are predominant in Graves' disease. Another possibility is that the thyroid cells are capable of fighting off infiltrating cells via induced expression of death ligands, such as TRAIL, which we have shown to be functionally expressed by thyroid cells (38). TRAIL is capable of killing activated T lymphocytes (39,40) and therefore could protect the thyroid cells.
We have shown by three methods that Fas expression in thyroid cells is regulated by inflammatory cytokines. The conditions of greatest susceptibility to Fas-mediated apoptosis show the greatest increase in Fas. However, the expression levels are not substantially greater than other conditions that show little or no Fas-mediated death. It is unlikely that either the 1.5-fold increase measured by Western analysis, the 1.5fold increase measured by RNase protection assay, or the 2.2fold increase measured by flow cytometry can account for the vast differences in Fas-mediated apoptosis when compared with the respective 1.2-, 1.3-, and 1.5-fold increases in Fas observed with conditions that were not associated with increased susceptibility. In contrast, susceptibility does correlate with induction of high molecular weight forms of Fas, which is due to increased glycosylation. This suggests that glycosylation may play an important role in signaling. This is of interest because, although Fas is known to be glycosylated (41), there are no reports of Fas glycosylation being an important regulatory process. It is possible that glycosylation affects transport and cell surface expression of Fas. Our data indicate that inhibition of Fas glycosylation can reduce Fas-mediated cell death. Another possibility is that the critical event in Fasmediated apoptosis is not Fas expression but the regulation of a Fas pathway inhibitor. Our previous report (26) showing cycloheximide-induced susceptibility suggests the presence of a labile inhibitor that may be eliminated or inactivated by the combination of IFN␥ with TNF␣ or IL-1␤. A known inhibitor of the Fas pathway is I-FLICE (42,43). We found no evidence of regulation of this inhibitor by cycloheximide or any combination of cytokines in thyroid cells. 2 These two hypotheses are not mutually exclusive. Both points of regulation (glycosylation and labile inhibitor) may influence Fas signaling. Supporting this idea are two points: 1) tunicamycin only partially inhibits Fas-mediated cell death despite the almost total ablation of the glycosylated form of Fas and 2) cycloheximide-inducible Fasmediated cell death occurs in the absence of Fas glycosylation. 2 Our data will provide important guidance in the search for the inhibitor(s) of this pathway.