Loss of FADD Protein Expression Results in a Biased Fas-Signaling Pathway and Correlates with the Development of Tumoral Status in Thyroid Follicular Cells

Loss of FADD protein expression results in a biased Fas-signaling pathway and correlates with the development of tumoral status in thyroid follicular cells

Le´ a Tourneur1, Sylvie Mistou1, Francine-Marie Michiels2, Vale´ rie Devauchelle1, Laurent Renia1, Jean Feunteun2 and Gilles Chiocchia*,1

1De´partement d’Immunologie, Institut Cochin, INSERM U567, CNRS UMR 8104, IFR 116, Universite´ Rene´ Descartes, 27 rue du fbg St-Jacques, 75014 Paris, France; 2Laboratoire d’Oncologie Mole´culaire, CNRS Unite´ de Recherche Associe´e 1158, Institut Gustave Roussy, 39 rue Camille Desmoulins, 94805 Villejuif, France

Downregulation of proapoptotic molecules like Fas or (Djerbi et al., 1999; Medema et al., 1999) has been caspase 8, or upregulation of antiapoptotic molecules like reported during tumor progression. FLICE inhibitory protein has been suggested to be a Activation of Fas by its ligand leads to recruitment of regulatory mechanism set up by tumor cells to block the the adapter molecule Fas-associated protein with death death signal received via death receptors. In an in-depth domain (FADD), which, in turn, activates FLICE/ study of the Fas/FasL-signaling pathway in thyroid tumor caspase 8, a process leading to apoptosis (Nagata, 1997). development, we have demonstrated that tumor cells FLIP, an antiapoptotic protein, can protect cells from specifically downregulate the multideath receptor adapter apoptosis by preventing FADD–caspase 8 association Fas-associated death domain (FADD). The regulation of (Irmler et al., 1997). Fas death domain-associated FADD expression occurred only at the protein level. protein (Daxx), a second adapter molecule, defines a Furthermore, in the absence of FADD, Fas-signaling different signaling pathway downstream of Fas (Yang resulted in accelerated growth of thyrocytes. Since et al., 1997). It can independently activate both the Jun thyrocytes also acquired FasL expression during tumor N-terminal kinases (JNK) and p38-mitogen-activated development, the absence of FADD protein could lead to protein kinases (p38-MAPK) pathways through the greater resistance to numerous death receptor-mediated apoptosis signal-regulating kinase 1 (ASK1) intermedi- apoptosis, stimulation of their own proliferation through ate (Ichijo et al., 1997; Chang et al., 1998). It seems that Fas/FasL interaction, and the capacity to counter-attack the JNK pathway generally leads to apoptosis, whereas the infiltrating lymphocytes. p38 kinase can have pro- or antiapoptotic effects, the Oncogene (2003) 22, 2795–2804. doi:10.1038/sj.onc.1206399 final result depending on multiple factors such as the type of cell and its state of differentiation. Recently, the Keywords: FADD; Fas ligand; thyroid; tumor kinase receptor-interacting protein (RIP) was reported to be a third adapter for Fas (Holler et al., 2000). It can transduce Fas-mediated caspase-dependent or -indepen- Introduction dent cell death. The thyroid is of particular interest regarding the role Dysregulation of the Fas pathway is responsible for of Fas/FasL since Fas is constitutively expressed by hypo- or hyperproliferative disorders that are as a result of excess or faulty apoptosis, respectively. FasL expres- normal thyroid follicular cells (TFC), and FasL expression can occur in numerous pathological thyroid sion on tumor cells allows the cells to escape the immune conditions (Baker, 1999), including cancer (Mitsiades system by killing infiltrating lymphocytes that express et al. et al. the receptor Fas, although this point is still controversial , 1999), Hashimoto’s thyroiditis (Giordano , 1997) and Grave’s disease (Hiromatsu et al., 1999). (Nagata, 1996; Strand et al., 1996; Favre-Felix et al., In the thyroid gland, guanine nucleotide stimulatory 2000; Restifo, 2000). Usually, when tumor cells coex- factor (Gs) activates adenylate cyclase in response to press Fas and FasL molecules, they turn out to be thyrotropin stimulation, leading to cyclic AMP level resistant to Fas-mediated cell death. Among the elevation and subsequent thyroid cell proliferation regulatory mechanisms that may take place to block (Dumont et al., 1992). Mutations of the a subunit of the death signal, downregulation of proapoptotic the Gs factor (gsp mutations), resulting in constitutive molecules such as Fas (Moller et al., 1994) or caspase activation of adenylate cyclase, are found in approxi- 8 (Teitz et al., 2000), or upregulation of antiapoptotic mately 30% of human thyroid toxic adenomas and 10% molecules such as FLIP (FLICE inhibitory protein) of human thyroid carcinomas (Suarez et al., 1991; Said et al., 1994). Transgenic mice expressing a mutant form *Correspondence: G Chiocchia; E-mail: [email protected] of the Gsa subunit directed to TFC (gsp mice) have Received 17 July 2002; revised 14 January 2003; accepted 16 January turned out to be a powerful model for studying human 2003 thyroid tumors. These mice have inherited with the gsp FADD protein in development of thyroid tumor L Tourneur et al 2796 transgene a highly penetrant predisposition to develop molecule in mouse thyroid AAC. We analysed thyroids thyroid hyperplasia that eventually transforms into at various stages of tumor development by semiquanti- hyperfunctioning thyroid adenomas at the age of 8 tative reverse transcriptase-polymerase chain reaction months (Michiels et al., 1994). Interestingly, adenocar- (RT–PCR) and Western blot. Neither FasL mRNA nor – cinomas of the thyroid have been observed in the FasL protein could be detected in normal thyroid tissue progeny of animal breed of a specific genetic back- from NTG mice (not shown). In nonpathological (NP) ground, suggesting that the gsp-linked phenotype is thyroid glands from gsp mice, FasL mRNA and protein sensitive to the genetic modifiers (F-M Michiels, were absent in all mice tested (Figure 1a, b, respectively), unpublished observations). whereas they could be detected in hyperplastic (HP) We investigated the presence of a wide array of thyroids, and at higher levels in thyroid AAC (Figure signaling molecules of the Fas cascade in these mice. 1a, b). Moreover, we detected both the membrane- Our results show that FasL expression was acquired bound (40 kDa) and soluble (26 kDa) forms of FasL in during tumor development, with a stronger expression AAC. By contrast, RT–PCR and Western blot analysis in adenoma/adenocarcinoma (AAC) than in hyperpla- showed that Fas mRNA and protein were expressed sia, but not in normal thyroid cells. Among the adapters both in NP and pathological thyroids (Figure 1a, b). for Fas, only FADD was regulated, in that FADD protein expression was weak or completely lost in AAC, although FADD mRNA was present. Furthermore, in FADD protein, but not Daxx and RIP, is lost during an in vitro model of thyrocytes culture, the absence of tumor development FADD resulted in a Fas signaling leading to faster Since Fas is known to act through the recruitment of growth of the thyrocytes. adapter molecules, we first investigated the modulation of FADD, Daxx and RIP (Figure 2). mRNAs of these Fas adapters were detectable in thyroids from NTG and Results transgenic mice, and their respective levels of expression of mRNAs were increased during the course of tumor Fas ligand expression is acquired during tumor development, with a maximal expression observed in development AAC (Figure 2a, b). Western blot analysis showed that high levels of FADD protein were detected in all NP Since FasL expression has been reported in human (3/3) and HP (4/4) glands from gsp mice. Interestingly, thyroid carcinoma, we studied the presence of this FADD protein expression was weak (2/5) or completely

l a a c a i m g o lo a n o i i h s rc t la a a a p m c p r o o n e n n o p e e y d d N H A A

425 bp FasL

366 bp Fas

NP HP b AAC L1 L2 Membrane-bound 40 kDa FasL Soluble 26 kDa

NP HP AAC L1 L2

45-48 kDa Fas

Figure 1 Fas ligand expression is acquired during tumor development. (a) Semiquantitative RT–PCR of Fas and FasL mRNAs and (b) Western blot analysis of Fas and FasL proteins expression by thyroid glands from gsp transgenic mice at various stages of tumor development. Clone H11 antibody was used for FasL protein detection. Ampliﬁed fragments length and proteins size are indicated. L1 and L2 are for differential protein expression by the two lobes of the same adenoma

Oncogene FADD protein in development of thyroid tumor L Tourneur et al 2797

Figure 2 Loss of the adapter protein FADD during tumor development. (a) Semiquantitative RT–PCR analysis of FADD, Daxx and RIP mRNAs expression by thyroid glands from gsp transgenic mice at various stages of tumor development. Amplified fragments size is indicated. b-Actin is used as control. FADD expression was evaluated by using two sets of primers covering the entire coding sequence (not shown). (b) FADD, Daxx, RIP and b-actin transcripts were analysed by semiquantitative RT–PCR as in (a), and bands were quantified by densitometry. Results are the mean of four NP, five HP and nine AAC for FADD; three NP, five HP and nine AAC for Daxx; three NP, two HP and six AAC for RIP. Data are expressed as the percentage of b-actin values. (c) Western blot analysis of the three adapter molecules expression by thyroid glands from gsp mice at various stages of tumor development. L1 and L2 are for differential protein expression by the two lobes of the same adenoma. Proteins size is indicated. We detected an approximately 70 kDa form of Daxx. Whatever the tumoral status of the gland, we could not detect the 120 kDa form of Daxx even in the nucleus (not shown). We used two different antibodies to confirm those results. Immunohistochemistry confirmed Daxx and RIP expression and showed that FADD protein expression was limited to restricted areas or completely lost in AAC (not shown). (d) Bands obtained in (c) were quantified by densitometry. Results are the mean of three NP, four HP and five AAC for FADD; one NP, two HP and five AAC for Daxx; one NP, three HP and five AAC for RIP, and are expressed as mean density in arbitrary units. *Po0.05, **Po0.001.

lost (3/5) in AAC (Figure 2c, d). By contrast, Western showed that procaspase 2 long (48 kDa), a pro- blot analysis showed that RIP and Daxx proteins were apoptotic form of caspase 2, was always expressed in both expressed at all stages of tumor development thyroids, regardless of the tumoral status of the gland. (Figure 2c). Daxx protein expression was unchanged or By contrast, both antiapoptotic forms of caspase 2 moderately decreased in AAC (Figure 2d). Although we (procaspase 2 short (35 kDa) (Droin et al., 2001) and/or observed an interindividual heterogeneity, the overall caspase 2 long prodomain (15 kDa) (Droin et al., 2000)) expression of RIP protein was only slightly increased in were only expressed in tumoral specimens. Although the AAC compared to NP thyroid gland (Figure 2d). 35 kDa form was barely expressed in AAC, high levels of the 15 kDa form were detected (Table 1). These Biased Fas-signaling pathway in the absence of FADD results suggest that the RIP-caspase 2 pathway was not the main Fas-signaling pathway during tumor develop- The above results show that both Daxx and RIP, but ment. Thus, in the absence of FADD, Fas signaling not or few FADD, proteins were expressed in thyroid seems to act mainly through a Daxx–ASK1 pathway. AAC. The next step was to study the downstream effector proteins of these signaling pathways. Western Inhibition of the kinase pathway during tumor blot analysis of ASK1 protein, the intermediate between development Daxx and kinase pathways, showed that ASK1 expression strongly increased during tumor growth (Figure 3a), We further investigated the two main kinase-signaling whereas the level of expression of HSP27, an inhibitor of pathways downstream of ASK1. the Daxx/ASK1 interaction, was unchanged (Figure 3b). We detected MKK3/MKK6-phosphorylated active On the other hand, RIP kinase can induce apoptosis proteins in gsp thyroids regardless of the tumoral status because of caspase 2 activation. Western blot analysis of the gland (Figure 3c). Since total MKK3 was strongly

Oncogene FADD protein in development of thyroid tumor L Tourneur et al 2798 a NP HP AAC L1 L2

154 kDa ASK1

b NP HP AAC L1 L2

27 kDa HSP 27

c NP HP AAC L1 L2

40 kDa Total MKK3

NP HP AAC L1 L2 Phospho-MKK3/MKK6 42 kDa (ser189/207)

d NP HP AAC L1 L2 45 kDa SEK1/MKK4

NP HP AAC L1 L2 Phospho-SEK1 46 kDa (thr261)

e NP HP AAC L1 L2

Phospho-p54 54 kDa Phospho-p46 Total SAPK/JNK 46 kDa

NP HP AAC L1 L2 Phospho-p54 Phospho-SAPK/JNK (thr183/tyr185) Phospho-p46

Figure 3 Molecules expression of kinases pathway under Fas-signaling dependency. Western blot analysis of total and/or active phosphorylated forms of the kinases pathway molecules expression by thyroid glands from gsp mice at various stages of tumor development: (a) ASK1 (b) HSP27, (c) MKK3 and MKK6, (d) SEK1/MKK4 and (e) SAPK/JNK. The ser189 phospho-MKK3 and ser207 phospho-MKK6 proteins, which are active forms of the kinases, are both recognized indifferently by the antibody we used. Total MKK3 and total SAPK/JNK antibodies detect both phosphorylated and unphosphorylated MKK3 and SAPK/JNK proteins, respectively. Proteins size is indicated. L1 and L2 are for differential protein expression by the two lobes of the same adenoma

decreased during tumor development (Figure 3c), these MAPK protein were detected in all NP thyroids, results suggest that in HP thyroids and AAC, only whereas its expression was low in HP glands, and very MKK6 protein is activated. weak or absent in AAC. A decrease or absence of p38- The average expression of total p38-MAPK, the MAPK activation correlated with a decrease or absence kinase downstream of MKK3/MKK6, was increased of total MKK3 expression (Figures 3c and 4b). in HP and AAC glands (Figure 4a). By contrast, The other kinases activated by ASK1 are SEK1/ expression of the p38-phosphorylated active form was MKK4 and SAPK/JNK molecules. We detected higher strongly decreased during tumor development levels of total SEK1 protein in AAC than in NP and HP (Figure 4b). High levels of thr180/tyr182 phospho-p38- glands. Nevertheless, the levels of thr261 phospho-SEK1

Oncogene FADD protein in development of thyroid tumor L Tourneur et al 2799 Table 1 Caspases expression of the Fas-signaling pathway Nonpathological thyroid Hyperplastic thyroid Adenoma/adenocarcinoma Caspase 8 +/Àa +/À +/À Caspase 10 + + + Procaspase 3 + + + Caspase 2 (48 kDa) ++ ++ or +++ ++ or +++ Caspase 2 (35 kDa) ÀÀÀor +/À Caspase 2 (15 kDa) ÀÀor + À or + or ++

Results were obtained by Western blot analysis and/or by immunohistochemistry and whenever possible were conﬁrmed by RT–PCR (not shown) aÀ: no; +/À: very weak; +: weak; ++: medium; and +++: high expression

Figure 4 Inhibition of the p38-MAPK pathway during tumor development. (a) Western blot analysis of total and (b) active phosphorylated forms of p38-MAPK expression by thyroids glands from gsp mice at various stages of tumor development. L1 and L2 are for differential protein expression by the two lobes of the same adenoma. Bands obtained were quantiﬁed by densitometry. Results are expressed as mean density in arbitrary units active form were higher in NP thyroids than in HP with collagenase H (Figure 5b), and FADD protein glands or AAC (Figure 3d). could not be detected up to 10 days of culture (not Although the total and phospho-SAPK/JNK protein shown). From these results, it was important to ﬁnd out levels were similar in all groups of mice, the apparent whether loss of FADD protein was an intrinsic property molecular weight of phosphorylated proteins was higher of cultured TFC or a generalized FADD protein in NP thyroids compared to diseased glands (Figure 3e). expression regulation mechanism. We chose hepatocytes These results suggest that SAPK/JNK proteins are to compare with thyrocytes because both are epithelial phosphorylated at both thr183 and tyr185 sites in NP cells, and liver tissue requires collagenase H treatment thyroids, but that HP glands and AAC are monopho- before culture. Both ex vivo tissues expressed FADD sphorylated because of the lack of phospho-SEK1- protein (Figure 5b). Western blotting of collagenase H- dependent thr183 phosphorylation. Since dual phos- treated and in vitro-cultured hepatocytes showed the phorylation of SAPK/JNK is necessary for kinase presence of FADD protein (Figure 5c). These results activity, the ASK1/SEK1/JNK pathway, as the ASK1/ suggest that loss of FADD protein expression was MKK3/p38-MAPK-signaling pathway, is inhibited dur- restricted to TFC, and that this in vitro culture system ing tumor development. could be used to elucidate the consequences of FADD protein expression regulation by thyrocytes. In vitro cultured thyrocytes do not express FADD protein Fas signaling in the absence of FADD increases TFC In the course of our study, we investigated the presence growth of molecules of the Fas-signaling pathway in in vitro cultures of NTG thyrocytes. Although FADD mRNA To evaluate the biological consequences of the Fas/FasL was expressed (Figure 5a), FADD protein was unde- interaction in the absence of FADD protein, we cultured tectable (Figure 5b). Loss of FADD protein expression TFC with or without anti-Fas antibody and determined was induced shortly after dissociation of thyroid tissue [3H]thymidine incorporation of thyrocytes every day

Oncogene FADD protein in development of thyroid tumor L Tourneur et al 2800 70 p<0.001 a Ex vivo In vitro 60 p<0.05 50 liver cultured thyrocytes p<0.02 40 30 695 bp 20 10 0 d % of proliferation increase d 1 2 3 4 5 b i id e H r o o r r e Day of culture r u s e y s lt y e v h a u t li t n th c y Figure 6 Fas signaling in the absence of FADD increases TFC o o e d c g o o iv iv e r a t t r growth. Normal thyrocytes were cultured with or without clone Jo2 v v l i y l a v h anti-Fas antibody for 24–120 h of exposure. DNA replication was x x o e t r n 3 E E C t I determined every day until confluence by adding [ H]thymidine for the last 12 h of culture. Data represent percentage of increase of anti-Fas versus control cultures. Each point is mean c.p.m. of six 28 kDa wells. Measure of [3H]thymidine incorporation of control TFC cultures varied from 343740, 632751, 7187104, 891747 and 1382745 c.p.m. from days 1 to 5, respectively. Measure of [3H]thymidine incorporation of anti-Fas-treated TFC varied from c 392735, 864762, 10697116, 1437774 and 1507788 c.p.m. from H ed r days 1 to 5, respectively r e r u s e t e e l t v as v li i u y n l c c o o d o t iv ge e r t t a In vitro cultured thyrocytes v la i l a v p id x o e e r o in r n h r t E C t I e y s v y 2 i h n c l t i 3 in t a t 1 t 2 o o s s 3 y c G l 2 y 1 v v c a i i M c ro 3 L G v v t a a t 1 + + t n -I c c M x G -I -I x o a I a + 28 kDa FADD E E C PI L M P PI L 28 kDa Figure 5 In vitro cultured thyrocytes did not express FADD protein. (a) Semiquantitative RT–PCR analysis of FADD mRNA expression by ex vivo liver and in vitro culture of thyroid cells from In vitro cultured thyrocytes id n normal NTG mice. In vitro cultured thyrocytes were analysed 24 h is o i t r t y s s y after confluence (day 7 of culture). (b) Western blot analysis of h 2 e t n c 3 n t i a i t t 1 t 2 FADD protein expression by ex vivo, collagenase H treated, and in o o s c s 3 y G v l 2 a y 1 v i c c i 3 L M vitro cultured thyrocytes (24 h after confluence) or (c) hepatocytes v ro a a G v t t 1 + + t I I n c I - c M x x - G - (24 h of culture) from normal NTG mice. Results of two o a a + Daxx E E C PI L M PI PI L independent in vitro cultures of hepatocytes are shown. Amplified fragment and protein sizes are indicated 70 kDa

Figure 7 Proteasome inhibitors did not induce FADD protein expression on cultured TFC. Western blot analysis of FADD and until confluence (Figure 6). After 24 h of treatment, Daxx adapter proteins expression by subconfluent normal thyr- there was no significant difference between DNA ocytes cultures treated overnight with proteasome inhibitors. synthesis of control and anti-Fas-treated TFC. There- Proteasome inhibitor I (PI-I) (1 mm) , lactacystin (1 mm) and after, anti-Fas antibody treatment induced a significant MG132 (0.1 mm) were used alone or in combination. The same results were obtained using 10 mm of PI-I, 10 mm of lactacystin and proliferation increase (37% (Po0.02), 49% (Po0.05) 1 mm of MG132, alone or in combination (not shown). Ex vivo liver and 61% (Po0.001) after 48, 72 and 96 h of treatment, and testis were used as positive controls for FADD and Daxx respectively). No further difference between control and (70 kDa form) proteins expression, respectively. Proteins size is anti-Fas-treated TFC growth could be detected when indicated cultures reached confluence (after 120 h of treatment). No difference was observed between control and control Ig-treated TFC (data not shown). associated, was able to induce FADD protein expression in TFC cultures (Figure 7), whereas Daxx protein Proteasome inhibitors did not induce FADD protein expression was highly increased after treatment of TFC expression on cultured TFC by combining PI-I and lactacystin or lactacystin and MG132 (Figure 7). As mRNA for FADD but not protein was expressed by cultured thyrocytes, an explanation could be that FADD protein was degraded after synthesis. Subcon- fluent thyroid epithelial cells cultures were treated with Discussion proteasome inhibitors, including proteasome inhibitor I (PI-I), lactacystin and MG132, alone or combined, and In mice, transgenic expression of a mutant form of the FADD expression was tested by Western blot analysis. Gsa subunit directed to TFC induces adenomas or None of these inhibitors, even at high dose or adenocarcinomas. We focused on Fas/FasL- and Fas-

Oncogene FADD protein in development of thyroid tumor L Tourneur et al 2801 signaling molecule expression, because FasL expression ment of TNFR-associated death domain (TRADD), on tumor cells could confer a double advantage to these which can, in turn, bind FADD and TNFR-associated cells by inducing apoptosis of invading cytotoxic factor 2 (TRAF2). As FADD protein is not expressed in lymphocytes (O’Connell et al., 1996), and by stimulating thyroid AAC, TNF stimulation would result only in a their own proliferation (Desbarats and Newell, 2000). TRAF2-dependent transducing signal. Furthermore, However, the regulatory mechanisms that occur in such FADD-deficient embryonic fibroblasts were shown to a context are poorly understood. be resistant to apoptosis induced by TNFR1 (Yeh et al., We focused on the expression of a large array of 1998), suggesting that FADD is essential to TNF- molecules involved in the Fas/FasL pathway during the mediated cell death. We found that FADD-deficient in vivo transformation process of murine TFC. We TFC were resistant to TNFa-mediated cell death (not showed that during this process: (1) FasL expression is shown), supporting the idea that tumoral TFC also gained; (2) FADD protein, but not mRNA, is lost; (3) acquire resistance to TNFR1-induced apoptosis. FADD Fas signaling in the absence of FADD results in a Daxx- is also necessary for death receptor 3 (DR3)-induced mediated signal; (4) the inactivated status of the p38- apoptosis (Yeh et al., 1998). The requirement of FADD MAPK is correlated with an MKK3 decrease and for TNF-related apoptosis-inducing ligand (TRAIL)- absence of FADD protein expression; and (5) Fas mediated cell death of tumor cells through two death signaling, in the absence of FADD, accelerated TFC receptors, DR4 and DR5, was also demonstrated growth. This is the first instance in which spontaneous (Kuang et al., 2000). All these data show that apoptosis FADD protein disappearance has been associated with induced by Fas, TNFR1, DR3, DR4 and DR5 needs the any pathological consequence. FADD molecule, implying that all these death pathways Inhibition of FADD function by gene targeting or by might be blocked in the absence of FADD. Interest- expression of dominant-negative interfering mutant has ingly, human primary thyrocytes, as well as the KAT5 been described on T lymphocytes, and was shown to human papillary thyroid carcinoma-derived cell line, render T-cells resistant to apoptosis induced by Fas were shown to express DR4 and DR5, but to resist (Newton et al., 1998; Zhang et al., 1998). In these TRAIL-induced apoptosis (Bretz et al., 1999). Thus, models, FADD defects also resulted in an inhibition of FADD deficiency would confer on tumor cells multiple mitogen-induced T-cell proliferation, demonstrating resistance against death receptor-induced apoptosis. that FADD can transmit both apoptotic and growth But what are the molecular mechanisms that regulate and/or survival signals. Conversely, it was found that expression of FADD protein? FADD deficiency in rag-1À/À mice induced the develop- One classical mechanism for the cells to degrade a ment of thymic lymphoma, indicating that FADD can protein is the targeting to the proteasome. However, we also act as a tumor suppressor (Newton et al., 2000). found that none of the proteasome inhibitors we tested These results demonstrate that FADD protein could was able to restore FADD protein expression. have various effects, depending on the time of its In eucaryotic cells, translation of mRNAs is mainly expression, the tissue and the developmental status of because of a cap-dependent mechanism through the tissue. In our model, we showed that FADD protein m7GpppN cap structure recognition (Gingras et al., might have a tumor suppressive function. Indeed, Fas 1999). On the other hand, some mRNAs are translated signaling in TFC, in the absence of FADD, promotes by a cap-independent mechanism. In this case, ribosome cell growth. These results were confirmed by the higher entry occurs at a specific region in the 50 end of mRNA proliferative rate of FADD-negative TFC expressing called the internal ribosome entry segment (IRES). Few FasL, isolated from thyroid-specific FasL transgenic mRNAs possessing IRES have been described in mice, compared to NTG thyrocytes (not shown). mammalians and, to our knowledge, all are implicated In T lymphocytes, Fas ligation can result either in in the control of cell growth (Vagner et al., 1995; apoptosis or in proliferation depending on previous Bernstein et al., 1997; Miller et al., 1998). For instance, antigenic stimulation (Desbarats et al., 1999). In human c-myc mRNA can be translated by both cap-dependent dermal fibroblasts, although anti-Fas antibody stimu- and -independent mechanisms. It has been demon- lates proliferation of low-Fas-expressing cells, anti-Fas strated that the p38-MAPK-signaling pathway is treatment triggers apoptosis of high-Fas-expressing implicated in this cap-independent translation (Stoneley fibroblasts (Freiberg et al., 1997). More surprisingly, in et al., 2000). Interestingly, no cap site was found in liver, Fas engagement can accelerate organ regeneration FADD mRNA (Kim et al., 1996), and the phospho-p38- after partial hepatectomy, instead of inducing fulminant MAPK pattern of expression was very similar to that of hepatitis (Desbarats and Newell, 2000). These data FADD (Figures 2c and 4b). Thus, FADD mRNA demonstrate that signaling through Fas can result either regulation of cap-independent translation could occur in apoptosis or in proliferation, depending on the type via the p38-MAPK pathway. of cell and the environmental conditions. In tumor cells, Our data suggest a new role for the Fas/FasL anti-Fas-promoted cell growth has been reported, and pathway in the etiology of thyroid tumor. In the absence involved hematological as well as nonhematological of FADD protein, but in the presence of Fas and FasL tumors (Owen-Schaub et al., 1993). molecules, thyrocytes would have numerous advantages: FADD is a signaling molecule common to numerous high resistance to Fas and other death receptor- death receptors. Stimulation of tumor necrosis factor mediated apoptosis; stimulation of their own prolifera- (TNF) receptor 1 (TNFR1) with TNF leads to recruit- tion through Fas/FasL interaction (Figure 8); and the

Oncogene FADD protein in development of thyroid tumor L Tourneur et al 2802

a Fas b Fas Cytosol Ligand Cytosol Ligand

Extracellulary Extracellulary Fas Fas

Cytosol Cytosol RIP RIP RAIDD X? RAIDD Apoptosis Pro-Caspase 2 Short FADD Caspase 2 Long-Pro

Daxx Pro-Caspase 2 Long Daxx Pro-Caspase 2 Long Pro-Caspase 8 HSP27 HSP27 ? ASK 1 Caspase 10 ? ASK 1 HSP27

MKK3 SEK1 MKK6 MKK3 MKK66 JNK Caspases p38 Effective caspases Mitochondrial pathway

PROLIFERATION

APOPTOSIS Apoptosis Induction of pro/antiapoptotic genes Figure 8 Biased Fas-signaling pathway in the absence of FADD. (a) Signal transmission through Fas can operate through three different adapters. (b) In the absence of FADD, only Daxx and RIP proteins can transduce a Fas-mediated signal. However, RIP signaling was not elicited in FADD-deﬁcient thyroids, and ASK1 intermediate expression strongly increased in AAC without consequent p38-MAPK or JNK pathways activation. We can make three hypotheses concerning the mechanism of Fas-signaling transduction in cancer TFC in the absence of FADD: (1) a particular Daxx-ASK1-MKK6 signaling is selected, (2) Daxx and/or RIP could lead to a still unknown transducing pathway, and (3) a still unknown adapter (X) for the Fas receptor is used

capacity to counter-attack the infiltrating lymphocytes, 500 ml to 1 ml Tri-Reagent per thyroid (Molecular Research allowing immune escape (Batteux et al., 2000; Tourneur Center, Cincinnati, Ohio, USA). For the in vitro culture et al., 2001). Understanding how FADD protein model, TFC were cultured in 24-well, flat-bottomed plates expression is lost will enable us to envisage new (model 3524, Costar, Cambridge, MA, USA). At subcon- therapeutic approaches. Restoration of FADD function fluence, TFC cultures were washed, and RNA was isolated in FasL+ thyrocytes would induce greater Fas-mediated using 250 ml/well Tri-Reagent. Contaminating DNA was eliminated by DNAse (Ozyme, Saint Quentin Yvelines, cell death susceptibility, and inhibit Fas-induced pro- France) treatment. Reverse transcription was performed on liferation of these cells, leading to tumor regression. 1–2 mg total RNA using Superscript II reverse transcriptase Moreover, our results could probably be extended to (Life Technologies, Inc., Cergy Pontoise, France). cDNA many types of cancer. (50 ng) was amplified by PCR using 0.5 U Taq polymerase (Life Technologies) and specific primers: FasL (425 bp): sense 50-CAGCAGTGCCACTTCATCTTGG-30, antisense 50-TTCACTCCAGAGATCAGAGCGG-30; Fas (366 bp): Materials and methods sense 50-CTTCTA-CTGCGATTCTCCTG-30, antisense 50- CTTGCCCTCCTTGATGTTAT-30; FADD (695 bp): sense Mice 50-TGGATGGGATTGAGGAGAAG-30, antisense 50- Gsp transgenic mice have been described elsewhere (Michiels AAACCACAGTCCTCACAGGG-30; Daxx (478 bp): sense et al., 1994). CBA/J female mice (Iffa Credo, L’Arbresle, 50-CCAACCTACCCCCCAATC-30, antisense 50-GCCTCA- France) were used at 8–10 weeks of age. All mice were GAGGCAGTGTTTTC-30; RIP (399 bp): sense 50-CAGGAG- maintained in standard environmental conditions, and allowed CAAGAGGTCATTCTG-30, antisense 50-TGTTATCAAA to adapt to their environment for 1 week before the GATGGCTTGGTG-30; FLICE (488 bp): sense 50- experiments. Thyroid tumor development was monitored by GCTGACCTCAGATGTGAGAGG-30, antisense 50-TCCCA 0 0 measurement of T4 and T3 serum levels. TTGAAGATGTACACAGC-3 . b-Actin ((348 bp): sense 5 - TGGAATCCTGTGGCATCCATGAAAC-30, antisense 50- TAAAACGCAGCTCAGTAACAGTCCG-30) was used as Reagent control. For quantification, serial, twofold dilutions of PI-I, lactacystin and MG132 were purchased from Calbiochem cDNA were amplified for 25 cycles using b-Actin primers. (Calbiochem-Novabiochem Corp., La Jolla, CA, USA). cDNA giving equivalent amount of b-Actin were then amplified for 24–30 PCR cycles. The RT–PCR products were subjected to electrophoresis on a 2% agarose gel and stained Semiquantitative RT–PCR with ethidium bromide. Amplified fragments were quantified Thyroids of gsp mice were collected at various stages of by densitometry using biocapts and bio-profil bio1ds soft- tumor development and RNA was isolated from cells using wares.

Oncogene FADD protein in development of thyroid tumor L Tourneur et al 2803 Western blot Nonadherent cells were removed by washing to obtain TFC pure cultures. Cells reached subconfluence after 5–6 days of Thyroids of gsp mice were collected at various stages of tumor culture. development, and total proteins were extracted with lysis buffer (10 mm Tris-HCl, 150 mm NaCl pH 7.8, 1% NP40, 1 mm PMSF, 1 mg/ml aprotinin and 1 mg/ml leupeptin), containing a Culture of primary hepatocytes cocktail of protein phosphatase inhibitors (Calbiochem- BALB/c hepatocytes were prepared according to Mazier et al. Novabiochem Corp.). TFC were cultured in vitro in six-well, (1985) with minor modifications. Briefly, cells were isolated by flat-bottomed plates (model 3516, Costar). At subconfluence, perfusion of the liver fragments with collagenase (Boehringer, TFC were washed and total proteins were extracted. Sample Mannheim, Germany), following purification in a 60% Percoll concentrations were determined using a micro-BCA protein gradient (Pharmacia Biotech, Uppsala, Sweden). Purified assay reagent kit (Pierce, Rockford, IL, USA), and 25 mgof hepatocytes showed at least 95% viability and purity as total proteins was subjected to SDS–PAGE and transferred to assessed by Trypan blue dye exclusion. They were cultured in nitrocellulose (NEN Life Sciences, Boston, MA, USA). After William’s E medium (Gibco, Irvine, UK) supplemented with blotting, membranes were probed with specific primary 5% fetal calf serum and 1% penicillin–streptomycin solution antibodies: anti-FasL (H11) was manufactured by Alexis following incubation at 371C for 24 h in a 3.5% CO2 Biochemicals (San Diego, CA, USA); anti-Fas (FL-335), atmosphere. FADD (MC19), Daxx (M112), RIP (H-207), caspase 10 (H- 131), ASK1 (N-19), SEK1/MKK4 (C-20), HSP27 (M-20) were obtained from Santa Cruz Biotechnology (Santa Cruz, CA, Proliferation assay USA); anticaspase 8 was from Chemicon International TFC were cultured in 96-well, flat-bottomed plates (model (Temecula, CA, USA); anti-caspase 3 proform (46) was from 3599, Costar) at approximately 18 000 cells/well in the absence Transduction Laboratories (Lexington, KY, USA); anti- or presence of clone Jo2 anti-Fas antibody (0.5 mg/ml, caspase 2 long and short (G310-1248) were from BD PharMingen) and in the absence or presence of polyclonal PharMingen (San Diego, CA, USA); antiphospho-SEK1/ control Ig (0.5 mg/ml, PharMingen). Proliferation of thyrocytes MKK4 (thr261), total SAPK/JNK, phospho-SAPK/JNK was determined from 24 to 120 h. Thyrocytes were pulsed with (thr183/tyr185), total MKK3, phospho-MKK3/MKK6 0.5 mCi of [3H]thymidine during the last 12 h of culture. Results (ser189/207), total p38-MAPK, phospho-p38-MAPK are expressed as mean c.p.m. of six wells. (thr180/tyr182) were obtained from New England Biolabs (Beverly, MA, USA). The proteins were visualized with the enhanced chemiluminescence technique (Amersham Pharma- cia Biotech, Orsay, France). Bands obtained were quantified Abbreviations by densitometry using biocapts and bio-profil bio1ds soft- AAC, adenoma/adenocarcinoma; Daxx, Fas death domain- wares. associated protein; FADD, Fas-associated protein with death domain; HP, hyperplastic; NP, nonpathological; NTG, non- Culture of primary thyrocytes transgenic; TFC, thyroid follicular cells. Thyroid tissue was obtained from control CBA/J female mice. Animals were killed and the two thyroid lobes removed after intracardiac puncture. The collected lobes were digested in Acknowledgements 1.5 mg/ml collagenase H (Boehringer Mannheim, Mannheim, The assistance of Gise` le Baudry in the animal care is gratefully Germany) solution for 45 min at 371C. TFC were then cultured acknowledged. This work was supported by the ‘Institut at 371C in a 5% CO2 atmosphere, in RPMI 1640 medium National de la Sante´ et de la Recherche Me´ dicale’ (INSERM). supplemented with 5% fetal bovine serum, 1% penicillin– F-M Michiels and J Feunteun are supported by the ‘Associa- streptomycin solution and 5 Â 10À5 m 2-mercaptoethanol. tion pour la Recherche sur le Cancer’ (ARC).

References

Baker Jr JR, (1999). J. Clin. Endocrinol. Metab., 84, 2593– Djerbi M, Screpanti V, Catrina AI, Bogen B, Biberfeld P and 2595. Grandien A. (1999). J. Exp. Med., 190, 1025–1032. Batteux F, Lores P, Bucchini D and Chiocchia G. (2000). J. Droin N, Beauchemin M, Solary E and Bertrand R. (2000). Immunol., 164, 1681–1688. Cancer Res., 60, 7039–7047. Bernstein J, Sella O, Le SY and Elroy-Stein O. (1997). J. Biol. Droin N, Rebe C, Bichat F, Hammann A, Bertrand R and Chem., 272, 9356–9362. Solary E. (2001). Oncogene, 20, 260–269. Bretz JD, Rymaszewski M, Arscott PL, Myc A, Ain KB, Dumont JE, Lamy F, Roger P and Maenhaut C. (1992). Thompson NW and Baker Jr JR. (1999). J. Biol. Chem., 274, Physiol. Rev., 72, 667–697. 23627–23632. Favre-Felix N, Fromentin A, Hammann A, Solary E, Chang HY, Nishitoh H, Yang X, Ichijo H and Baltimore D. Martin F and Bonnotte B. (2000). J. Immunol., 164, (1998). Science, 281, 1860–1863. 5023–5027. Clemens MJ, Bushell M, Jeffrey IW, Pain VM and Morley SJ. Freiberg RA, Spencer DM, Choate KA, Duh HJ, Schreiber (2000). Cell Death Differ, 7, 603–615. SL, Crabtree GR and Khavari PA. (1997). J. Invest. Desbarats J and Newell MK. (2000). Nat. Med., 6, 920–923. Dermatol., 108, 215–219. Desbarats J, Wade T, Wade WF and Newell MK. (1999). Gingras AC, Raught B and Sonenberg N. (1999). Annu. Rev. Proc. Natl. Acad. Sci. USA, 96, 8104–8109. Biochem., 68, 913–963.

Oncogene FADD protein in development of thyroid tumor L Tourneur et al 2804 Giordano C, Stassi G, De Maria R, Todaro M, Richiusa P, Nagata S. (1997). Cell, 88, 355–365. Papoff G, Ruberti G, Bagnasco M, Testi R and Galluzzo A. Newton K, Harris AW, Bath ML, Smith KG and Strasser A. (1997). Science, 275, 960–963. (1998). EMBO J., 17, 706–718. Hiromatsu Y, Hoshino T, Yagita H, Koga M, Sakisaka S, Newton K, Harris AW and Strasser A. (2000). EMBO J., 19, Honda J, Yang D, Kayagaki N, Okumura K and Nonaka 931–941. K. (1999). J. Clin. Endocrinol. Metab., 84, 2896–2902. O’Connell J, O’Sullivan GC, Collins JK and Shanahan F. Holler N, Zaru R, Micheau O, Thome M, Attinger A, Valitutti (1996). J. Exp. Med., 184, 1075–1082. S, Bodmer JL, Schneider P, Seed B and Tschopp J. (2000). Owen-Schaub LB, Meterissian S and Ford RJ. (1993). J. Nat. Immunol., 1, 489–495. Immunother., 14, 234–241. Ichijo H, Nishida E, Irie K, ten Dijke P, Saitoh M, Moriguchi Pomerance M, Abdullah HB, Kamerji S, Correze C and T, Takagi M, Matsumoto K, Miyazono K and Gotoh Y. Blondeau JP. (2000). J. Biol. Chem., 275, 40539–40546. (1997). Science, 275, 90–94. Restifo NP. (2000). Nat. Med., 6, 493–495. Irmler M, Thome M, Hahne M, Schneider P, Hofmann K, Said S, Schlumberger M and Suarez HG. (1994). J. Endocrinol. Steiner V, Bodmer JL, Schroter M, Burns K, Mattmann C, Invest., 17, 371–379. Rimoldi D, French LE and Tschopp J. (1997). Nature, 388, Stoneley M, Chappell SA, Jopling CL, Dickens M, MacFar- 190–195. lane M and Willis AE. (2000). Mol. Cell Biol., 20, 1162– Kim PK, Dutra AS, Chandrasekharappa SC and Puck JM. 1169. (1996). J. Immunol., 157, 5461–5466. Strand S, Hofmann WJ, Hug H, Muller M, Otto G, Strand D, Kuang AA, Diehl GE, Zhang J and Winoto A. (2000). J. Biol. Mariani SM, Stremmel W, Krammer PH and Galle PR. Chem., 275, 25065–25068. (1996). Nat. Med., 2, 1361–1366. Mazier D, Beaudoin RL, Mellouk S, Druilhe P, Texier B, Suarez HG, du Villard JA, Caillou B, Schlumberger M, Trosper J, Miltgen F, Landau I, Paul C and Brandicourt O, Parmentier C and Monier R. (1991). Oncogene, 6, 677–679. Guguen-guillouzo C, Langlois P, Gentilini M. (1985). Teitz T, Wei T, Valentine MB, Vanin EF, Grenet J, Valentine Science, 227, 440–442. VA, Behm FG, Look AT, Lahti JM and Kidd VJ. (2000). Medema JP, de Jong J, van Hall T, Melief CJ and Offringa R. Nat. Med., 6, 529–535. (1999). J. Exp. Med., 190, 1033–1038. Tourneur L, Malassagne B, Batteux F, Fabre M, Mistou S, Michiels FM, Caillou B, Talbot M, Dessarps-Freichey F, Lallemand E, Lores P and Chiocchia G. (2001). J. Immunol., Maunoury MT, Schlumberger M, Mercken L, Monier R 167, 1338–1346. and Feunteun J. (1994). Proc. Natl. Acad. Sci. USA, 91, Vagner S, Gensac MC, Maret A, Bayard F, Amalric F, Prats 10488–10492. H and Prats AC. (1995). Mol. Cell Biol., 15, 35–44. Miller DL, Dibbens JA, Damert A, Risau W, Vadas MA and Yang X, Khosravi-Far R, Chang HY and Baltimore D. (1997). Goodall GJ. (1998). FEBS Lett., 434, 417–420. Cell, 89, 1067–1076. Mitsiades N, Poulaki V, Mastorakos G, Tseleni-Balafouta ST, Yeh WC, Pompa JL, McCurrach ME, Shu HB, Elia AJ, Kotoula V, Koutras DA and Tsokos M. (1999). J. Clin. Shahinian A, Ng M, Wakeham A, Khoo W, Mitchell K, El- Endocrinol. Metab., 84, 2924–2932. Deiry WS, Lowe SW, Goeddel DV and Mak TW. (1998). Moller P, Koretz K, Leithauser F, Bruderlein S, Henne C, Science, 279, 1954–1958. Quentmeier A and Krammer PH. (1994). Int. J. Cancer, 57, Zhang J, Cado D, Chen A, Kabra NH and Winoto A. (1998). 371–377. Nature, 392, 296–300. Nagata S. (1996). Nat. Med., 2, 1306–1307.

Oncogene