(2006) 25, 1143–1152 & 2006 Nature Publishing Group All rights reserved 0950-9232/06 $30.00 www.nature.com/onc ORIGINAL ARTICLE Merlin facilitates ubiquitination and degradation of transactivation-responsive RNA-binding

JY Lee1,8, HJ Moon1, WK Lee2, HJ Chun3, CW Han4, Y-W Jeon5, Y Lim6, YH Kim1, T-P Yao7, K-H Lee1,2, T-Y Jun1,4, HK Rha8 and J-K Kang8

1Catholic Neuroscience Center, The Catholic University of Korea, Seoul, Korea; 2Department of Pharmacology, The Catholic University of Korea, Seoul, Korea; 3Department of Biomedical Science, The Catholic University of Korea, Seoul, Korea; 4Department of Orthopedics Surgery, The Catholic University of Korea, Seoul, Korea; 5Department of Psychiatry, The Catholic University of Korea, Seoul, Korea; 6Department of Occupational and Environmental Medicine, The Catholic University of Korea, Seoul, Korea; 7Duke University, Durham, NC, USA and 8Department of Neurosurgery, The Catholic University of Korea, Seoul, Korea

The Nf2 tumor suppressor codes for merlin, a protein Introduction whose function is largely unknown. We have previously demonstrated a novel interaction between merlin and Neurofibromatosis type 2 (NF2) is a predominantly TRBP, which inhibits the oncogenic activity of TRBP. In inherited disorder characterized by the development spite of the significance of their functional interaction, its of schwann cell tumors and other brain tumors (Evans molecular mechanism still remains to be elucidated. In this et al., 1992a, b). or the loss of heterozygosity report, we investigated how merlin inhibits the oncogenic of the Nf2 locus has been detected in various tumors of activity of TRBP in association with cell growth conditions. the nervous system, such as schwannomas, meningiomas In the embryonic kidney 293 cell line, the level of and ependymomas (Gusella et al., 1999). In addition, endogenous merlin increased, whereas that of endogenous inactivation of Nf2 specifically in the schwann cells leads TRBP significantly decreased along with the increase in to the development of schwannomas and schwann cell cell confluence. We demonstrated that the carboxyl- hyperplasia in mice (Giovannini et al., 2000). terminal region of TRBP was responsible for this The Nf2 gene encodes merlin, which was named for its phenomenon using stable cell lines expressing deletion structural similarity to the ERM family mutants of TRBP. The overexpression of merlin decreased (, and ) (Gusella et al., 1996; Tsukita the protein level of TRBP, and the ubiquitin-like subdomain and Yonemura, 1997; Bretscher et al., 2002; Sun et al., of merlin’s FERM domain was important for this activity. 2002). Merlin and ERM protein share the FERM We also demonstrated that TRBP is ubiquitinylated and domain, which is composed of three structural modules the ubiquitinylated forms of TRBP are accumulated by (F1, F2 and F3) that together form a compact clover- ectopically expressed merlin or cell confluence in the shaped structure (Pearson et al., 2000). Especially, F1 is presence of MG132, a proteasome inhibitor. Furthermore, very similar to the structure of ubiquitin (Vijay-Kumar we showed that the regulation of TRBP in response to cell et al., 1987). confluence was abolished upon knockdown of merlin Recent work has demonstrated that the protein level expression by specific small interfering RNA. Finally, we and states of merlin are affected by the showed that ectopically expressed merlin restored cell–cell growth conditions of cells, such as confluence, loss of contact inhibition in cells stably expressing TRBP but not adhesion or serum deprivation (Gutmann et al., 1999). in TRBPDc. These results suggest that merlin is involved in The phosphorylation of merlin at 518 can be the regulation of TRBP protein level by facilitating its induced by an active form of Rac or cdc42, but not of ubiquitination in response to such cues as cell–cell contacts. Rho (Shaw et al., 2001). The Rac/cdc42-induced phos- Oncogene (2006) 25, 1143–1152. doi:10.1038/sj.onc.1209150; phorylation of merlin at serine 518 is mediated by a - published online 10 October 2005 activated , Pak (Kissil et al., 2002; Xiao et al., 2002). Such phosphorylation can disrupt the intramole- Keywords: Nf2; merlin; tarbp2; TRBP; ubiquitination; cular interaction of merlin and its association with the degradation , and this induces a change in the subcellular localization of merlin in LLC-PK1 cells (Shaw et al., 2001; Kissil et al., 2002). On the other hand, merlin Correspondence: Professor YH Kim, Neuroscience Genome Research inhibits the activation of Pak1 through direct binding to Center, 4011, Catholic Research Institute of Medical Science, The the Pak1 PBD, and it thereby interferes with the Pak1 Catholic University of Korea, 505 Banpo-dong, Socho-ku, Seoul 137- recruitment to focal adhesions (Kissil et al., 2003). 701, Korea. E-mail: [email protected] TRBP belongs to the family of double-stranded RNA Received 13 December 2004; revised 25 August 2005; accepted 26 August (dsRNA)-binding proteins, and it was originally identi- 2005; published online 10 October 2005 fied as an HIV-1 TAR RNA-binding protein with two Merlin facilitates degradation of TRBP JY Lee et al 1144 clearly defined dsRNA-binding domains (dsRBDs) and full-length human TRBP DNA or the derivatives of human a C-terminal basic region (St Johnston et al., 1992; TRBP DNA. At 24 h after transfection, the cells were replated Gatignol et al., 1993; Bass et al., 1994; Kharrat et al., at the ratios of 3:1, 9:1, 27:1 and 50:1, and the G418-resistant 1995; Daher et al., 2001). TRBP dsRBD2 binds TAR colonies were selected with 400 mg/ml G418 for 2 weeks. Before with a higher affinity than dsRBD1 does, because the each experiment using the constructed stable cell lines, Western blot analyses were performed to certify the expression of the former contains a KR-helix motif (St Johnston et al., integrated TRBP. 1992; Bannwarth et al., 2001). 0 The murine homolog of TRBP binds the 3 untrans- Antibodies, immunoprecipitation and Western blotting lated region of Prm1 protamine RNA, represses its The antibodies for merlin (sc-331, sc-28247 and a specific batch translation and plays a physiological role in spermato- of sc-332, Santa Cruz Biotechnology, Santa Cruz, CA, USA), genesis (Lee et al., 1996; Zhong et al., 1999). TRBP ubiquitin (Santa Cruz Biotechnology, Santa Cruz, CA, USA), directly binds the dsRNA-activated protein kinase V5-epitope (Invitrogen, San Diego, CA, USA) and b-actin (PKR), which has antiviral and antiproliferative effects. (Sigma Inc.) were obtained from commercial sources. Poly- TRBP inhibits the ability of PKR to phosphorylate clonal anti-TRBP antibody was produced in a rabbit using a eucaryotic translation initiation factor 2 (eIF-2), leading KLH-conjugated peptide (RSPPMELQPPVSPQQSECNPV- to its inactivation (Park et al., 1994, Benkirane et al., GALQ) as an epitope (Peptron Inc., Daejeon, Korea), and the antibody product was tested by ELISA and Western analysis 1997). In relation to the inhibition of PKR activity, (data not shown). TRBP was demonstrated to play a growth-promoting For immunoprecipitation or Western blotting, the cells were role and it has oncogenic potential (BenKirane et al., lysed in RIPA-B buffer (0.5% Nonidet P-40, 20 mM Tris, pH 1997). However, it has not been elucidated how TRBP’s 8.0, 50 mM NaCl, 50 mM NaF, 100 mM Na3VO4,1mM DTT function is regulated in cells. and 50 mg/ml PMSF) for 1 h on ice. The insoluble materials In our previous study, we identified TRBP as a were removed by centrifugation at 12 000 r.p.m. for 20 min at merlin-binding protein in a yeast two-hybrid screen and 41C. The supernatant was then subjected to SDS–PAGE and carboxyl-terminal regions of each protein as a domain Western blotting. The blots were blocked in phosphate- responsible for their interaction (Lee et al., 2004). The buffered saline containing 5% skim milk and 0.05% Tween consequences of merlin–TRBP interaction include in- 20, and the blots were then incubated sequentially with the primary antibody and HRP-conjugated secondary antibody. hibitory effects on TRBP-mediated cell proliferation, Detection was performed according to the enhanced chemilu- anchorage-independent cell growth, oncogenic transfor- minescence protocol (Amersham, Arlington Heights, IL, USA). mation and tumor development in nude mice (Lee et al., 2004). Therefore, it may be possible to suggest that Enrichment of ubiquitinated protein merlin has a role as a cellular regulator of TRBP Polyubiquitin affinity beads, in which the ubiquitin-associated function through certain mechanisms. domain of Rad23 was immobilized, and the control beads were In this report, we aimed to understand the mechanism obtained from Calbiochem-Novabiochem. The whole-cell by which merlin inhibits the oncogenic activity of extracts from HEK293 cells were obtained using a lysis buffer TRBP. We found that TRBP is degraded by ubiquitina- consisting of 50 mM HEPES (pH 7.5), 5 mM EDTA, 150 mM tion under the high-confluence cell growth conditions NaCl, 1% Triton X-100 and complete protease inhibitor (Roche Applied Science). A 1 mg portion of each whole-cell and this is facilitated by merlin. extract was incubated with 40 ml of beads in the lysis buffer for 4 h at 41C, with constant mixing. The beads were then washed three times with the same buffer, and the bound proteins were Materials and methods extracted in SDS buffer and analysed by Western blotting using anti-ubiquitin antibody (Santa Cruz Biotechnology Inc.) Plasmids and adenovirus or anti-TRBP antibody. The plasmids for wild-type merlin and full-length human TRBP and its derivatives were described in previous reports (Lee et al., RNA interference to Nf2 2004). The adenovirus expressing merlin was kindly gifted by Dr Stealtht small interfering RNA (siRNA) duplex oligonucleo- Jeun. The virus was propagated in the human embryonic kidney tides were designed based on the sequences specific to human 293 (HEK293) cell line, and the viral titers were determined by a Nf2 cDNA (50-GGACAAGAAGGUACUGGAUCAUGAU- limiting-dilution bioassay in HEK293 cells. 30 and 50-AUCAUGAUCCAGUACCUUCUUGUCC-30). Stealtht RNAi Negative Control Med GC was used as an Cell culture and transient transfection experimental control. The antisense and sense siRNA oligo- NIH3T3, A172 and HEK293 cells were obtained from the nucleotides with a dTdT 30-overhang were synthesized and American Type Culture Collection (Manassas, VA, USA). All annealed by Invitrogen Inc. (San Diego, CA, USA). Transfec- the cell lines were maintained in DMEM that was supplemen- tion of HEK293 cells with the siRNA duplexes was performed ted with 10% FBS and antibiotics. The Gene Porter2 reagent according to the manufacturer’s instructions using Lipofecta- (Gene Therapy System Inc.) and Lipofectamine 2000 (Invitro- minet 2000. gen, San Diego, CA, USA) were used for transfection according to the manufacturer’s recommendations. For the same amount of total transfected DNA, the pcDNA3.1 plasmid was used in all the transfections. Results

Construction of stable cell lines Downregulation of endogenous TRBP by confluency NIH3T3 cells were plated at a density of 3 Â 105 cells per We have previously reported that merlin interacts with 35 mm dish and they were transfected with plasmids containing TRBP and inhibits its oncogenic activity such as

Oncogene Merlin facilitates degradation of TRBP JY Lee et al 1145 proliferation, anchorage-independent growth and tu- is involved in the downregulation of TRBP in response morigenesis in a nude mouse model (Lee et al., 2004). to cell confluence. However, the mechanism used for doing so has not been elucidated. Based on the function of TRBP and merlin The carboxyl-terminal region of TRBP is responsible for as a putative oncogene and a tumor suppressor, TRBP decrease according to cell confluency respectively, we were interested in whether their func- To further support downregulation of TRBP level by tional interaction could induce changes in the level of cell confluency, we investigated which region of TRBP each protein owing to the growth condition of cells such was responsible for this. Several stable NIH3T3 cell lines as the state of cell confluence and the cell-to-cell contact were constructed, with each expressing wild-type TRBP (Fisher and Yeh, 1967). To test this hypothesis, we first or its two deletion mutants (TRBPDR and TRBPDC; examined the cellular levels of endogenous TRBP and Figure 2a) or EGFP. Each stable cell line was plated at merlin by performing immunoblotting on HEK293 cells increasing degrees of confluency; the protein levels of that were plated at increasing cell densities (Figure 1a). ectopically expressed TRBP and EGFP, and the As shown in Figure 1b, the levels of merlin increased endogenous merlin, were examined by immunoblotting. along with the increase in cell confluency, and this was In all cases, the levels of endogenous merlin increased particularly prominent when the majority of cells had along with increasing cell confluency (Figure 2b–e). The established cell-to-cell contacts (upper panel). This level of the N-terminal deletion form of TRBP is a well-known phenomenon (Sherman et al., 1997; (TRBPDR) decreased along with increasing confluency, Shaw et al., 1998a, b). Strikingly, the levels of TRBP similarly to that of the wild-type TRBP (Figures 2b and decreased along with increasing confluency, which d). However, the levels of the C-terminal deletion form showed an opposite result to that of merlin (Figure 1b, of TRBP and the negative control of EGFP did not middle panel). Considering the interaction between show any change with respect to increasing cell merlin and TRBP, these data may suggest that merlin confluency (Figures 2c and e). These results indicated

Figure 2 The carboxyl-terminal region of TRBP is responsible for its decrease depending upon cell confluence. Immunoblot analysis of merlin and TRBP proteins in NIH3T3 cells stably expressing TRBP-V5 (b), TRBPDC-V5 (c), TRBPDR-V5 (d) and EGFP (e). The anti-V5 epitope antibody was used to detect TRBP proteins, Figure 1 High cell density reduces TRBP protein level in HEK293 and b-actin was detected as an internal control. The confluences of cells. (a) The phase-contrast images of HEK293 cells plated in cells in each lane are as follows: lane 1, 2.5 Â 105; lane 2, 5 Â 105; 60 mm dishes at increasing cell densities as described. The next day, and lane 3, 1 Â 106 in a 60 mm dish. The TRBP constructs are the cells were used for immunoblotting as shown in lanes 1–4 of (b). depicted in (a). TRBP contains three domains: two dsRNA-binding (b) Immunoblot analysis of merlin, TRBP and b-actin using the domains (dsRBD1 and dsRBD2) and a C-terminal region. Deletion total cell lysates of HEK293 cells from (a). The representative notation: TRBPR lacks dsRBD1; TRBPDC lacks the C-terminal results shown are from three independent experiments. region.

Oncogene Merlin facilitates degradation of TRBP JY Lee et al 1146 that the confluency-induced TRBP decrease is a specific process, and it depends on the carboxyl-terminal region of TRBP.

Merlin plays a role in TRBP decrease The above results showed that the levels of merlin and TRBP inversely correlated with cell confluency. Pre- viously, we have also shown the specific interaction and functional counteraction between merlin and TRBP (Lee et al., 2004). Therefore, we assumed that the decrease in the level of TRBP according to cell confluency might be caused by the increase in the level of merlin. To prove this hypothesis, an increasing amount of merlin expression plasmid was cotransfected with a given amount of TRBP expression plasmid and 0.5 mg of pcDNA3.1-LacZ into NIH3T3 cells, and their cellular levels were examined. As shown in Figure 3a, the level of overexpressed TRBP decreased along with the increase in merlin levels. To see whether the merlin effect also occurs with endogenous TRBP, the adeno- viruses expressing merlin (Ad-NF2) were infected to A172, a human glioblastoma cell line, and the level of endogenous TRBP was examined. As shown in Figure 3b, the level of endogenous TRBP was decreased by Ad-NF2 in a dose-dependent manner, but it was not decreased by the GFP-expressing adenovirus (Ad-GFP) in A172 cells. These results strongly suggest that the overexpression of merlin decreases the TRBP protein level. To further support the direct role of merlin, endo- genous merlin was knocked down by a siRNA against merlin expression (si-NF2), and the TRBP protein level Figure 3 Merlin is responsible for the decrease of TRBP. (a) was examined. As shown in Figure 3c, the endogenous Immunoblot analysis for merlin, TRBP and LacZ in NIH3T3 cells TRBP protein level specifically increased when si-NF2 transfected with the indicated amounts of TRBP and merlin expression plasmids with 0.5 mg pcDNA-lacZ-his plasmid. (b) was transfected. Taken together, these results suggest Immunoblot analysis of merlin and TRBP in A172 cells infected that merlin plays a role as a regulator of the TRBP with the merlin-expressing adenovirus at the indicated MOIs. (c) protein level. Immunoblot analysis of merlin and TRBP in HEK293 cells transiently transfected with the siRNA to merlin (si-NF2) or a control siRNA (si-control). Mock represents the nontransfected Ubiquitin-like subdomain of merlin’s FERMdomain is control. In order to ensure confluent status, all cell lysates were responsible for TRBP decrease, and phosphorylation of prepared at 2 days after transfection or infection. merlin at S518 does not affect it To further support the specific role of merlin, we performed a domain analysis of merlin to determine which region of merlin has the ability to induce the revealed that the wild-type and deletion mutants of decrease of TRBP. Various deletion mutants of merlin merlin were expressed at comparable levels (Figure 4b, were constructed (Figure 4a). Wild-type merlin and its middle panels). When comparing the domains in the deletion mutants were each expressed in HEK293 cells deletion mutants, this result suggested that the ubiqui- together with V5 epitope-tagged TRBP and pcDNA3.1- tin-like subdomain of merlin’s FERM domain is lacZ. Subsequent Western blotting with anti-V5 anti- important for the decrease in TRBP level by merlin body revealed that the wild-type and M3 deletion overexpression. We have quantified TRBP-V5 levels mutants of merlin clearly decreased the level of from three independent experiments (Figure 4c) and ectopically expressed TRBP and the M2 and M4 arrived at the same conclusion. mutants of merlin showed relatively weak effects. It was previously shown that phosphorylation of Notably, the level of ectopically expressed TRBP was merlin at serine 518 leads to relocalization and not affected by the expression of M1 mutant, which inactivation of merlin, whereas a merlin mutant of lacks the ubiquitin-like subdomain (F1 subdomain) of S518A functions as a constitutive active form (Shaw merlin’s FERM domain. Additionally, the level of et al., 2001). To examine the effect of merlin phosphor- ectopically expressed TRBP was moderately decreased ylation at serine 518, we observed the amount of by M1 and M3 mutant coexpression (Figure 4b, upper exogenous TRBP in NIH3T3 cells coexpressing each panel). Western blotting with anti-merlin antibody of wild-type, nonphosphorylatable mutant type (S518A)

Oncogene Merlin facilitates degradation of TRBP JY Lee et al 1147

Figure 4 Ubiquitin-like subdomain of merlin’s FERM domain is responsible for the decrease of TRBP. (a) Schematic presentation of wild-type and deletion mutants of merlin. (b) Immunoblot analysis of merlin and TRBP in NIH3T3 cells transiently transfected with the expression plasmids for merlin proteins described in (a) and/or TRBP-V5. (c) Three independent immunoblotting images of TRBP were scanned with GS-800 Calibrated Densitometer (Bio-Rad Laboratories, Richmond, CA, USA) and analysed by Quantity One 4.2.2 program. Statistical analysis was performed with Student’s t-test (*Po0.05 and **Po0.01 versus TRBP only). (d) Immunoblot analysis for TRBP and merlin proteins in NIH3T3 cells transfected with the plasmids expressing TRBP-V5, and the wild-type or mutant (S518A and S518D) merlin proteins. In order to ensure confluent status, all cell lysates were prepared at 2 days after transfection. The asterisks indicate putative depredation products of each protein.

and phosphorylation mimicking mutant type (S518D) possible mechanisms could be the ubiquitin–proteasome merlin proteins. As shown in Figure 4d, the exogenous pathway. In our first attempt to elucidate this, TRBP protein level was decreased by all three forms we examined whether TRBP by itself is ubiquitinylated of merlin, indicating that merlin phosphorylation at in HEK293 cells. After treatment of cells for 15 h with serine 518 does not affect its ability to induce TRBP MG132, a proteasome inhibitor, the endogenous TRBP decrease. level increased according to MG132 concentration (Figure 5a, middle panel). Moreover, a ladder of high molecular weight endogenous TRBP bands, a Ubiquitinylation of endogenous TRBP characteristic of polyubiquitinylated proteins, was de- As described above, TRBP is dramatically decreased by tected by TRBP immunoblotting and this was also merlin expression and it interacts with merlin, as was increased in amount by MG132 treatment (Figure 5a, previously demonstrated by us (Lee et al., 2004). upper panel). Additionally, the ubiquitin-like domain of merlin’s To demonstrate that the high molecular weight ladder FERM domain is responsible for merlin-induced TRBP of TRBP was due to ubiquitinylation, we used decrease. Taken together, if merlin immediately reg- polyubiquitin affinity beads to pull the proteins down ulates TRBP level at the protein level, one of the from the whole-cell extracts of HEK293 cells that were

Oncogene Merlin facilitates degradation of TRBP JY Lee et al 1148 ubiquitinylated and MG132, a specific proteasome inhibitor, can stabilize TRBP and also its ubiquitiny- lated forms.

Merlin facilitates TRBP degradation via the ubiquitin– proteasome pathway We demonstrated in the above experiments that merlin decreases the amount of TRBP protein and that TRBP can be ubiquitinylated. To address merlin’s function in TRBP ubiquitinylation, HEK293 cells were transfected with an increasing amount of merlin expression plas- mids, and the cells were treated with 5 mM MG132 for 15 h to inhibit the proteasomal degradation of TRBP. Subsequently, cell lysates were immunoprecipitated with anti-TRBP antibody and then immunoblotted with anti- ubiquitin antibody to show the ubiquitinylated TRBP. In cells expressing exogenous merlin, ubiquitinylated high molecular mass proteins were detected in the anti- TRBP immunoprecipitates and their amounts were enhanced by the increase in merlin expression (Figure 6a). To further confirm merlin-induced TRBP ubiquitiny- lation, the same experiment was performed with M3 and M4 deletion mutants of merlin. As shown in Figure 6b, endogenous ubiquitinylated TRBP was detected in wild-type merlin and in the M3 deletion mutant. However, the M4 deletion mutant did not increase ubiquitinylated TRBP. Given the above ob- servations, this result indicated that merlin may facil- itate TRBP ubiquitinylation via the ubiquitin-like F1 Figure 5 Ubiquitination of TRBP in HEK293 cells. (a) HEK293 subdomain of merlin. cells were left untreated or treated with MG132, a specific proteasome inhibitor, as indicated. The total cell extracts were immunoblotted with anti-V5 epitope antibody to detect the tagged TRBP. Both the long- and short-exposure images are presented to Merlin mediates confluency-dependent TRBP regulation show the ubiquitination and stabilization of TRBP by MG132, To elucidate if TRBP ubiquitinylation is regulated by respectively. (b) Equivalent amounts of whole-cell extracts of cell confluence via merlin, we investigated TRBP HEK293 cells treated as above were applied to the polyubiquitin ubiquitinylation in HEK293 cells with various cell affinity beads or to the control beads. The pulled down proteins confluences. As shown in Figure 7a, endogenous were analysed by Western blotting using anti-TRBP antibody. (c) HEK293 cell lysates were immunoprecipitated (IP) with anti-TRBP ubiquitinylated TRBP was increased along with the antibody or preimmune serum (PIS), and then immunoblotted (IB) increase in cell confluence. This result suggested that with anti-ubiquitin antibody. All cell lysates were prepared at TRBP was regulated by cell confluence. confluent status. Furthermore, we investigated confluency-dependent TRBP regulation in merlin knockdown cells. As shown in Figure 7b, si-control-transfected HEK293 cells showed confluency-dependent TRBP regulation. How- treated with MG132. Subsequent Western blotting with ever, merlin knockdown cells did not show confluency- anti-TRBP antibody revealed a ladder of polyubiquiti- dependent TRBP regulation and also basal TRBP level nylated TRBP bands from the pulled down proteins was significantly higher than si-control-transfected (Figure 5b). The ubiquitinylated TRBP bands were NIH3T3 cells. Taken together, these data suggested shifted up with similar size intervals, and the lowest that TRBP protein level was controlled by confluency band was about 6 kDa apart from the original position. via merlin. These bands might have corresponded to the conjuga- tion of one, two, three and four ubiquitins (Figure 5b, asterisks). Merlin overexpression restores contact inhibition in To further support TRBP ubiquitination, the cell TRBP-overexpressing NIH3T3 cells lysates of HEK293 cells treated with MG132 were first Previously, we demonstrated that NIH3T3 cells immunoprecipitated with anti-TRBP antibody. The overexpressing TRBP exhibit transformed phenotypes, subsequent immunoblotting with anti-ubiquitin anti- such as the loss of contact inhibition, the increase of body revealed a smear band of polyubiquitinylated cell proliferation, anchorage-independent growth and TRBP bands with high molecular weights (Figure 5c). tumorigenesis in nude mouse (Benkirane et al., 1997; Taken together, these results suggest that TRBP is Lee et al., 2004). To elucidate the biological function of

Oncogene Merlin facilitates degradation of TRBP JY Lee et al 1149

Figure 6 Overexpressed merlin increases the ubiquitinylated TRBP in HEK293 cells. (a) Immunoblot analysis of exogenously expressed merlin (top panel) and ubiquitinylated TRBP (middle panel) in HEK293 cells transfected with increasing amounts of merlin- expressing plasmid (pNF2). (b) Immunoblot analysis of endogenous ubiquitinylated TRBP in HEK293 cells transfected with wild-type, M3 and M4 merlin-expressing plasmid (right panel). Immunoblot analysis of exogenously expressed wild-type, M3 and M4 deletion mutant of merlin with anti-merlin C-term (left panel) and anti-merlin N-term antibody (center panel). The cells were treated with 5 mM MG132 for 15 h at 1 day after transfection. All the cell lysates were immunoprecipitated (IP) with anti-TRBP antibody and immunoblotted (IB) with anti-ubiquitin antibody. In order to ensure confluent status, all cell lysates were prepared at 2 days after transfection or infection. The asterisks indicate the positions of the merlin mutants expected from their calculated molecular weights.

merlin, we investigated contact inhibition in the TRBPDc-expressing cell line showed less TRBP-, TRBPDc- and EGFP-overexpressing NIH3T3 contact inhibition than other cell lines. Merlin-expres- cells that were infected with merlin-expressing adeno- sing adenovirus infection restored cell–cell contact virus or not. inhibition in the TRBP stable cell line. In contrast, As shown in Figure 8, the TRBP- and TRBPDc- merlin-expressing adenovirus infection did not show expressing stable cell line (TRBP, TRBPDc) showed a significant effect on TRBPDc-expressing and significantly higher growth rates than the control EGFP-expressing cell lines. TRBPDc, which cannot cells (EGFP). Additionally, cell–cell contact inhibition bind to merlin, is a stable protein, regardless of was observed in EGFP- and TRBP-expressing cells, cell confluency (Figure 2c). These data suggest that but the TRBP-expressing cell line had greater cell merlin can rescue contact inhibition in TRBP-over- number than the EGFP-expressing cell line. In contrast, expressing cells.

Oncogene Merlin facilitates degradation of TRBP JY Lee et al 1150

Figure 8 Merlin overexpression rescued loss of contact inhibition by TRBP overexpression in NIH3T3 cells. NIH3T3 stable cell lines expressing the indicated proteins were plated on six-well plates with 2 Â 105 cells/well. Merlin-expressing adenovirus was infected with 100 MOI at 1 day after seeding. The cell number was counted on indicated days. The results shown are the average of three independent experiments. Statistical analysis was performed with Student’s t-test (*Po0.05 versus each mock infection).

Figure 7 Confluency-dependent TRBP regulation is mediated by growth conditions as cell confluency (Shaw et al., merlin-dependent ubiquitinylation of TRBP. (a) Immunoprecipita- 1998a, b). Furthermore, cell confluency influences mer- tion-coupled immunoblot analysis of endogenous ubiquitinylated TRBP in HEK293 cells that show deferent confluency. The lin’s function such as its interaction with Pak1, resulting confluences of cells in each lane are as follows: lane 1, 7.5 Â 105; in the inhibitory effect on Pak1 activation (Kissil et al., lane 2, 1.5 Â 106; lane 3; and lane 4, 3 Â 106 in a 100 mm dish. (b) 2003). Given the correlation of merlin’s function and cell Immunoblot analysis of endogenous merlin (upper panel), TRBP confluency, we were interested to know whether growth (middle panel) and b-actin (bottom panel) in HEK293 cells that were transfected with si-control and si-NF2 RNA. The confluences inhibition by high cell confluency might cause a change of cells in each lane are as follows: lane 1, 2.5 Â 105; lane 2, 5 Â 105; in the cellular level of TRBP. Our data showed that the and lane 3, 1 Â 106 in a 60 mm dish. level of endogenous TRBP is decreased by increasing confluency, and it is inversely correlated with the level of merlin in HEK293, a human kidney cell line (Figure 1). Discussion Furthermore, we also showed that the carboxyl-terminal region of TRBP is responsible for this phenomenon After positional cloning of the Nf2 gene, its encoded using stable NIH3T3 cell lines expressing deletion protein, merlin, was revealed as a member of a family of mutants of TRBP (Figure 2). Notably, we described in cytoskeleton:membrane linkers, and it functionally a previous report (Lee et al., 2004) that the merlin– plays a role both as a tumor suppressor and metastasis TRBP interaction depends on the same carboxyl- suppressor. However, the molecular basis for the terminal region of TRBP. Therefore, it could be growth-suppressing function of merlin still remains speculated that the merlin–TRBP interaction may be rather elusive. We recently reported that merlin interacts necessary for the confluency-induced downregulation of specifically with TRBP via its carboxyl-terminal region, TRBP. Although we did not directly see the importance and it consequently inhibits the oncogenic activities of of their interaction, we proved the role of merlin in TRBP that induce cell proliferation, anchorage-inde- decreasing TRBP using transient expression systems and pendent cell growth, oncogenic transformation and siRNA against merlin (Figure 3). Taken together, it tumor development in nude mice (Lee et al., 2004). would be reasonable to suggest that increased merlin However, the regulation mechanism and downstream expression either from exogenous or endogenous events of their interaction remain to be elucidated. We sources induced the decrease in TRBP level. now provide several lines of evidence suggesting a novel On the domain analysis of merlin (Figure 4b), function of merlin to induce ubiquitination and degra- ubiquitin-like F1 subdomain of merlin was responsible dation of TRBP in response to the signals by cell for the downregulation of TRBP level. It is interesting confluency. that this domain was also shown in our previous report Previous study has demonstrated that the cellular to be responsible for the merlin-dependent downregula- level of merlin was increased at high cell confluency, and tion of level (Kim et al., 2004). Therefore, the F1 that merlin was differentially phosphorylated by such subdomain of merlin appears to play role(s) in the

Oncogene Merlin facilitates degradation of TRBP JY Lee et al 1151 downregulation of some of its target proteins in cells. to exclude the possibility that the decrease in TRBP is The TRBP level was apparently decreased by the M3 simply a secondary result of merlin’s growth inhibitory mutant of merlin lacking the TRBP-binding region (Lee effect. First, we showed that TRBPDc, which cannot et al., 2004). It may be possible that endogenous merlin bind to merlin, is a stable protein regardless of cell and other ERM protein are involved in this phenom- confluency (Figure 2). Furthermore, ectopically ex- enon. Merlin and ERM protein form homo- or hetero- pressed merlin cannot restore contact inhibition in cells oligomer in normal cells (Bretscher et al., 2000). stably expressing this merlin-resistant TRBPDc mutant, Therefore, if the M3 mutant formed an oligomer with although it can do so in cell expressing wild-type TRBP endogenous merlin, TRBP may bind with the M3– cells (Figure 8). Finally, we showed that downregulation merlin oligomer. of TRBP in response to cell confluence was abolished Phosphorylation of merlin can represent one mechan- upon knockdown of merlin expression by a specific ism of regulating the function of merlin by modulating siRNA (Figure 7b). These results show that TRBP its folding states, being inactive to inhibit cell growth degradation is not simply a secondary effect of cell when phosphorylated (Shaw et al., 1998a, b; Rong et al., confluency. It requires both a functional merlin and 2004; Surace et al., 2004). The phosphorylation state of TRBP–merlin interaction. merlin at serine 518 was recently suggested to be very Considering the growth-promoting role of TRBP, it is important for its function in suppressing cell growth and not so strange that the cellular TRBP level is tightly motility. The of serine 518 to alanine (S518A) regulated by the ubiquitin–proteasome degradation resulted in the constitutive active form of merlin, while pathway. In terms of merlin, it has been recently its mutation to glutamate (S518D), mimicking a reported that a deletion mutant of merlin frequently ‘phosphorylated’ form of merlin, abolished its activities observed in NF2 patients is efficiently degraded by the (Surace et al., 2004). In this report, we also examined the ubiquitin–proteasome pathway (Gautreau et al., 2002). function of these two mutations of merlin (S518A and However, this is the first report that associates merlin S518D), but they showed no effect on TRBP down- with the function of inducing ubiquitination and regulation (Figure 4d). However, our previous results degradation of target proteins. demonstrated that hypophosphorylated merlin prefer- Although the molecular machinery for merlin indu- entially binds to TRBP (Lee et al., 2004). In addition, cing proteasomal degradation of TRBP is still largely phospho-amino-acid analyses have demonstrated that unknown, the data presented here represent a first step merlin is phosphorylated on multiple serine and/or toward understanding how merlin, a tumor suppressor, threonine residues (Shaw et al., 1998a, b). Therefore, we regulates the growth-promoting activity of TRBP. cannot exclude the possibility that other merlin phos- Taken together, our results suggest that merlin may phorylation sites can affect its function to downregulate serve its tumor suppressor role in response to the signals TRBP. from increased cell-to-cell contact at least in part by We demonstrated that TRBP was ubiquitinylated at inducing the proteasomal degradation of TRBP. the physiological level (Figure 5), and exogenous merlin and M3 mutant expression can induce the ubiquitinyla- tion of endogenous TRBP in HEK293 cells (Figure 6). Abbreviations Furthermore, we showed TRBP ubiquitinylation was increased by cell confluency (Figure 7a) and merlin NF2, neurofibromatosis type 2; TRBP, trans-activation- expression was responsible for confluency-dependent responsive RNA-binding protein; dsRBD, double-stranded TRBP regulation (Figure 7b). Finally, we demonstrated RNA-binding domain. that ectopically expressed merlin can restore the contact inhibition in cells stably expressing TRBP (Figure 8). This line of evidence suggested that merlin facilitates Acknowledgements ubiquitinylation and proteasomal degradation in re- We thank Dr BH Choi for the invaluable advice and Dr CY sponse to cell confluency. Choi for the helpful comments and all the other members of We cannot directly show merlin-dependent TRBP our laboratory for their valuable assistance. This study was ubiquitinylation using an in vitro ubiquitinylated assay supported by a grant of the Korean Health 21 R&D Project, because E3 for TRBP has not yet been Ministry of Health Welfare, Republic of Korea (00-PJ3-PG6- identified. However, we provided the following evidence GN02-0002, 01-PJ3-PG6-GN07-0004).

References

Bannwarth S, Talakoub L, Letourneur F, Duarte M, Purcell Bretscher A, Edwards K, Fehon RG. (2002). Nat Rev Mol Cell DF, Hiscott J et al. (2001). J Biol Chem 276(52): 48803– Biol 3(8): 586–599. 48813. Daher A, Longuet M, Dorin D, Bois F, Segeral E, Bannwarth Bass BL, Hurst SR, Singer JD. (1994). Curr Biol 4(4): 301–314. S et al. (2001). J Biol Chem 276(36): 33899–33905. Benkirane M, Neuveut C, Chun RF, Smith SM, Samuel CE, Evans DG, Huson SM, Donnai D, Neary W, Blair V, Newton Gatignol A et al. (1997). EMBO J 16(3): 611–624. V et al. (1992a). J Med Genet 29(12): 847–852. Bretscher A, Chambers D, Nguyen R, Reczek D. (2000). Annu Evans DG, Huson SM, Donnai D, Neary W, Blair V, Teare D Rev Cell Dev Biol 16: 113–143. et al. (1992b). J Med Genet 29(12): 841–846.

Oncogene Merlin facilitates degradation of TRBP JY Lee et al 1152 Fisher HW, Yeh J. (1967). Science 155(762): 581–582. Park H, Davies MV, Langland JO, Chang HW, Nam YS, Gatignol A, Buckler C, Jeang KT. (1993). Mol Cell Biol 13(4): Tartaglia J et al. (1994). Proc Natl Acad Sci USA 91(11): 2193–2202. 4713–4717. Gautreau A, Manent J, Fievet B, Louvard D, Pearson MA, Reczek D, Bretscher A, Karplus PA. (2000). Cell Giovannini M, Arpin M. (2002). J Biol Chem 277(35): 101(3): 259–270. 31279–31282. Rong R, Surace EI, Haipek CA, Gutmann DH, Ye K. (2004). Giovannini M, Robanus-Maandag E, van der Valk M, Niwa- Oncogene 23: 8447–8454. Kawakita M, Abramowski V, Goutebroze L et al. (2000). Shaw RJ, Henry M, Solomon F, Jacks T. (1998a). Mol Biol Genes Dev 14(13): 1617–1630. Cell 9(2): 403–419. Gusella JF, Ramesh V, MacCollin M, Jacoby LB. (1996). Curr Shaw RJ, McClatchey AI, Jacks T. (1998b). J Biol Chem Opin Genet Dev 6(1): 87–92. 273(13): 7757–7764. Gusella JF, Ramesh V, MacCollin M, Jacoby LB. (1999). Shaw RJ, Paez JG, Curto M, Yaktine A, Pruitt WM, Saotome Biochim Biophys Acta 1423(2): M29–36. I et al. (2001). Dev Cell 1(1): 63–72. Gutmann DH, Sherman L, Seftor L, Haipek C, Sherman L, Xu HM, Geist RT, Saporito-Irwin S, Howells N, Hoang Lu K, Hendrix M. (1999). Hum Mol Genet 8(2): Ponta H et al. (1997). Oncogene 15(20): 2505–2509. 267–275. St Johnston D, Brown NH, Gall JG, Jantsch M. (1992). Proc Kharrat A, Macias MJ, Gibson TJ, Nilges M, Pastore A. Natl Acad Sci USA 89(22): 10979–10983. (1995). EMBO J 14(14): 3572–3584. Sun CX, Robb VA, Gutmann DH. (2002). J Cell Sci 115(Part Kim H, Kwak NJ, Lee JY, Choi BH, Lim Y, Ko YJ et al. 21): 3991–4000. (2004). J Biol Chem 279(9): 7812–7818. Surace EI, Haipek CA, Gutmann DH. (2004). Oncogene 23(2): Kissil JL, Johnson KC, Eckman MS, Jacks T. (2002). J Biol 580–587. Chem 277(12): 10394–10399. Tsukita S, Yonemura S. (1997). Curr Opin Cell Biol 9(1): 70–75. Kissil JL, Wilker EW, Johnson KC, Eckman MS, Yaffe MB, Vijay-Kumar S, Bugg CE, Cook WJ. (1987). J Mol Biol 194(3): Jacks T. (2003). Mol Cell 12(4): 841–849. 531–544. Lee JY, Kim H, Ryu CH, Kim JY, Choi BH, Lim Y et al. Xiao GH, Beeser A, Chernoff J, Testa JR. (2002). J Biol Chem (2004). J Biol Chem 279(29): 30265–30273. 277(2): 883–886. Lee K, Fajardo MA, Braun RE. (1996). Mol Cell Biol 16(6): Zhong J, Peters AH, Lee K, Braun RE. (1999). Nat Genet 3023–3034. 22(2): 171–174.

Oncogene