A Novel Protein, RTN-Xs, Interacts with Both Bcl-Xl and Bcl-2 on Endoplasmic Reticulum and Reduces Their Anti-Apoptotic Activity

A novel protein, RTN-xS, interacts with both Bcl-xL and Bcl-2 on endoplasmic reticulum and reduces their anti-apoptotic activity

Shinji Tagami1,2,3, Yutaka Eguchi1,3, Manabu Kinoshita1, Masatoshi Takeda2 and Yoshihide Tsujimoto*,1,3

1Department of Medical Genetics, Biomedical Research Center, Osaka University Graduate School of Medicine, Osaka, Japan; 2Department of Neuropsychiatry, Osaka University Graduate School of Medicine, Osaka, Japan; 3CREST of Japan Science and Technology Corporation (JST), 2-2 Yamadaoka, Suita, Osaka 565-0871, Japan

Bcl-2 and Bcl-xL serve as critical inhibitors of apoptosis eects on the mitochondria, which play an essential triggered by a broad range of stimuli, mainly acting on role in apoptotic signal transduction (reviewed by the mitochondria. We identi®ed two members of the Green and Reed, 1998; Tsujimoto and Shimizu, 2000). reticulon (RTN) family as Bcl-xL binding proteins, i.e., Various apoptotic stimuli have been shown to induce NSP-C (RTN1-C) and a new family member, RTN-xS, mitochondrial changes that result in the release of both of which did not belong to the Bcl-2 family and apoptogenic factors like mitochondrial cytochrome c were predominantly localized on the endoplasmic (Liu et al., 1996) into the cytoplasm, which leads to reticulum (ER). RTN-xS interacted with both Bcl-xL activation of caspase-9 through Apaf-1 (Zou et al., and Bcl-2, increased the localization of Bcl-xL and Bcl-2 1997; Li et al., 1997). The release of cytochrome c is on the ER, and reduced the anti-apoptotic activity of prevented by anti-apoptotic Bcl-2 family members Bcl-xL and Bcl-2. On the other hand, NSP-C interacted (Yang et al., 1997; Kluck et al., 1997). Several eects only with Bcl-xL, aected the localization of Bcl-xL, and of them on the mitochondria have been proposed, such reduced Bcl-xL activity, but had no eect on Bcl-2. as closure of the voltage-dependent anion channel These results suggest that RTN family proteins can (VDAC) (Shimizu et al., 1999), the adenine nucleotide modulate the anti-apoptotic activity of Bcl-xL and Bcl-2 translocator (Marzo et al., 1998) or the permeability by binding with them and can change their localization to transition pore (Zamzami et al., 1996; Narita et al., the ER. Oncogene (2000) 19, 5736 ± 5746. 1998), as well as the binding and sequestration of Apaf-1 (Pan et al., 1998; Hu et al., 1998). Keywords: RTN-xS; Bcl-xL; Bcl-2; endoplasmic reticu- Although the eects of Bcl-2 on the mitochondria lum; apoptosis have been studied intensively, little is known about the eects of Bcl-2 on the endoplasmic reticulum (ER), where anti-apoptotic Bcl-2 family proteins are also Introduction localized. Bcl-xL and Bcl-2 regulate the release of calcium from intracellular stores, which are primarily Apoptosis plays an important role in a variety of located in the ER (Bay et al., 1993; Lam et al., 1994; biological events, including morphogenesis, mainte- He et al., 1997). It was also reported that Bcl-2 nance of tissue homeostasis, and removal of harmful targeted to the ER can inhibit Myc-induced apoptosis cells (reviewed by Kerr et al., 1972; Arends and Wyllie, (Zhu et al., 1996). These results suggest that Bcl-2 1991). Apoptotic signal transduction pathways acti- located on the ER might act to prevent apoptosis, vated by various stimuli converge into a phylogenically although the mechanisms involved are not clearly conserved common pathway, which is driven by understood. caspases and regulated by the Bcl-2 family (reviewed Bcl-xL and/or Bcl-2 are known to interact with by Salvesen and Dixit, 1997; Thornberry and Lazebnik, various proteins, such as pro-apoptotic Bcl-2 family 1998). Caspases are activated by proteolytic cleavage of members and other proteins from outside the Bcl-2 their proforms, and then cleave various cellular family. Among them, SMN (Iwahashi et al., 1997), Bis substrates (reviewed by Salvesen and Dixit, 1997; (Lee et al., 1999), and BAG-1 (Takayama et al., 1995) Thornberry and Lazebnik, 1998). are reported to enhance the anti-apoptotic eect of Bcl- The Bcl-2 family is characterized by the conservation 2. Bap31 (Ng et al., 1997), and SERCA (Kuo et al., of Bcl-2 homology (BH) domains (reviewed by Adams 1998) are localized on the ER, but their eects on the and Cory, 1998; Tsujimoto, 1998), and it consists of activity of Bcl-xL or Bcl-2 remain unknown. pro-apoptotic molecules (i.e. Bax, Bak, Bik, Bad, Bim The reticulon family gene 1 (RTN1) was identi®ed by and Bid) and anti-apoptotic molecules (i.e. Bcl-2, Bcl- antibodies that stained a subset of neuroendocrine xL, and Bcl-w) (reviewed by Adams and Cory, 1998; tissues and neoplasms (Roebroek et al., 1993), and was Tsujimoto, 1998). Recent studies have focused on their formerly called neuroendocrine-speci®c protein (NSP) gene. RTN1 produces three splice variants, which are designated NSP-A, -B, and -C. On the other hand, expression of its family genes, RTN2 and RTN3 is *Correspondence: Y Tsujimoto, Department of Medical Genetics, almost ubiquitous (Roebroek et al., 1998; Moreira et Biomedical Research center, Osaka University Graduate School of al., 1999). Medicine, 2-2 Yamadaoka, Suita, Osaka 565-0871, Japan Received 24 July 2000; revised 13 September 2000; accepted 22 Immunohistochemical studies have revealed that September 2000 RTN proteins are anchored to the ER (Senden et al., Translocation of Bcl-xL and Bcl-2 to ER by RTN-xS S Tagami et al 5737 1996; GrandPre et al., 2000) probably through two residues, and lacked residues 186 to 1004 of RTN-xL putative transmembrane domains in the homologous (Figure 1a). The C-terminal amino acid sequence of C-terminal region, so these proteins were renamed the RTN-xS (aa 188 ± 373) shared 69% identity with that of reticulon (RTN) family (Roebroek et al., 1998). Little NSP-C (Figure 1b), and was the region where RTN is known about the function of the RTN family in family proteins showed signi®cant homology with each apoptosis. other (Roebroek et al., 1993, 1998; Moreira et al., Here we describe two Bcl-xL-interacting proteins 1999). RTN-xL and RTN-xS contained two putative belonging to the RTN family, NSP-C and a novel transmembrane domains, as do other RTN family member that we have termed RTN-xS. Unlike NSP-C, proteins (Figure 1b and Roebroek et al., 1993, 1998; RTN-xS can also interact with Bcl-2. RTN-xS reduced Moreira et al., 1999). the anti-apoptotic activity of both Bcl-xL and Bcl-2, Northern blot analysis revealed that the longest while NSP-C only reduced that of Bcl-xL. Change in transcript (about 5.4 kb) was speci®cally expressed in subcellular localization of Bcl-2 family proteins from brain and testis, as well as at low levels in heart and the mitochondria to the ER was thought to be involved skeletal muscle (Figure 2), consistent with RTN-xL in both cases. being isolated from the brain library. A transcript of about 3 kb should correspond to RTN-xS, because RTN-xS is 2457 bp shorter than RTN-xL. RTN-xS was expressed in all tissues examined, except for the liver Results (Figure 2). At least two other transcripts were observed in some tissues, but we did not characterize them. Identification of RTN-x S Northern blotting using NSP-C as a probe yielded To search for proteins interacting with Bcl-xL,we signals with dierent sizes from those obtained using screened human fetal and adult brain libraries using the RTN-x probe (data not shown). the yeast two-hybrid system, with full-length human Bcl-x as the bait. Fifty-seven b-galactosidase-positive L RTN-x interacts with both Bcl-x and Bcl-2 in clones were obtained from 26107 transformants, and S L mammalian cells were found to contain 19 bad, seven bax, three Bnip3L, and one bid gene, as well as 27 non-bcl-2 family clones. To examine the interaction of RTN-xL, RTN-xS, and Among the 27 clones, one clone carried the entire NSP-C with Bcl-xL or Bcl-2 in mammalian cells, N- coding sequence of NSP-C (RTN1-C) (Roebroek et al., terminal HA-tagged RTN-xL, RTN-xS, and NSP-C 1993) and two clones (4-20s and 15-97) corresponded expression plasmids were constructed (pCAGGS-HA- to a new member of the RTN family, which was RTN-xL, pCAGGS-HA-RTN-xS, and pCAGGS-HA- designated RTN-x. Analysis using the yeast two-hybrid NSP-C, respectively). We transiently co-transfected system revealed that both 4-20s and NSP-C bound to COS-7 cells with these plasmids together with Bcl-xL, but only 4-20s bound to Bcl-2 (Table 1). pCAGGS-Bcl-xL, pCAGGS-Bcl-2, or a control empty We obtained the full-length cDNA sequence of vector. Western blot analysis of immune complexes RTN-x as described in Materials and methods. There recovered with anti-HA serum revealed that both Bcl- were two kinds of cDNA, which may have been xL and Bcl-2 were co-immunoprecipitated with HA- produced by alternative splicing. The longer cDNA, RTN-xS only when co-expressed with HA-RTN-xS which we termed RTN-xL, contained 3576 bp encoding (Figure 3a,b). Identical results were obtained when 1192 amino acid residues, and was identical to the D98/AH2 cells were used instead of COS-7 cells (data human cDNA clone KIAA0886 in the DNA data base not shown). In addition, an anti-Bcl-xL polyclonal (Nagase et al., 1998). The shorter form, termed RTN- antibody (pAb) and an anti-Bcl-2 pAb immunopreci- xS, consisted of 1119 bp encoding 373 amino acid pitated HA-RTN-xS when co-expressed with Bcl-xL and Bcl-2, respectively (data not shown). These results indicated that RTN-xS bound to both Bcl-xL and Bcl-2 in mammalian cells, consistent with the outcome of the yeast two-hybrid experiments (Table 1). No interaction Table 1 Results of yeast two-hybrid analysis of RTN-xL with Bcl-xL or Bcl-2 could be detected Plasmids b-galactosidase activity (data not shown), probably because of its lower level of bait prey expression or the eect of its large N-terminal region. Interaction of RTN-x with truncated Bcl-x and pBTM116-Bcl-xL pGAD10-4-20s +++ S L pBTM116-Bcl-xL pGAD10-NSP-C +++ Bcl-2, which lack the C-terminal membrane-anchoring pBTM116-Bcl-2 pGAD10-4-20s ++ domain (37 and 13 amino acid residues, respectively), pBTM116-Bcl-2 pGAD10-NSP-C 7 was also detected by the same methods, but was pBTM116-Bcl-xL pGAD100-Bad +++ pBTM116-Bcl-2 pGAD10-Bad +++ somewhat weaker (data not shown), suggesting that the pBTM-RAD51 pGAD10-4-20s 7 interaction of RTN-xS with Bcl-xL or Bcl-2 was not pBTM-RAD51 pGAD10-NSP-C 7 mediated through their C-terminal hydrophobic re- gions. The L40 reporter yeast strain was cotransfected with the indicated plasmids. Individual transfectants were plated in SD/-Trp/-Leu/-His We also tested whether endogenous RTN-XS could medium and grown for 4 days after which b-galactosidase activity associate with endogenous Bcl-xL and Bcl-2, using was assayed. The bad gene construct isolated in this study was used Jurkat cells treated with a protein cross linker; DTBP as a positive control of Bcl-xL- and Bcl-2-binding protein, while as described in Materials and methods. As shown in RAD51 was used as a negative control. Symbols represent the time required for development of a strong blue color: +++, less than Figure 3c,d, both the anti-Bcl-xL monoclonal antibody 1 h; ++, 1 ± 2 h; +, 2 ± 4 h; 7, more than 8 h or not grown on the (mAb) and anti-Bcl-2 mAb immunoprecipitated en- SD plates dogenous RTN-xS, indicating that endogenous RTN-xS

Oncogene Translocation of Bcl-xL and Bcl-2 to ER by RTN-xS S Tagami et al 5738

Figure 1 Primary structure of the RTN-x gene products. (a) Diagram of the structures of RTN-xL (upper) and RTN-xS (lower). The positions corresponding to 4-20s and 15-97 are underlined. Amino acid residues 186 to 1004 of RTN-xL are not present in RTN-xS.(b) Comparison of the amino acid sequences of RTN-xS and NSP-C. The identical amino acid residues are highlighted and two putative transmembrane domains are underlined

could associate with endogenous Bcl-xL and Bcl-2 in and Bcl-2, while NSP-C interacted with Bcl-xL but not mammalian cells. with Bcl-2. When HA-NSP-C was used instead of HA-RTN-xS, Bcl-x but not Bcl-2 was coimmunoprecipitated with L RTN-x is mainly localized on the ER HA-NSP-C (Figure 3a,b). Bcl-2 was not detected even S after longer exposure (data not shown). Consistently, Because reticulon family members including NSP-C are an anti-Bcl-xL pAb immunoprecipitated HA-NSP-C, known to be localized on the ER (Senden et al., 1996; while an anti-Bcl-2 pAb did not (data not shown). GrandPre et al., 2000), the subcellular localization of These results indicated that HA-NSP-C interacted RTN-xS was examined by the two dierent methods; with Bcl-xL, but not with Bcl-2, which was also biochemical fractionation and immunocytochemistry. consistent with the ®ndings obtained using the yeast When transiently transfected D98/AH2 cells were two-hybrid system. From these observations, we biochemically fractionated, both RTN-xS and NSP-C concluded that RTN-xS interacted with both Bcl-xL were mainly recovered in the microsomal fraction, with

Oncogene Translocation of Bcl-xL and Bcl-2 to ER by RTN-xS S Tagami et al 5739 heart brain placenta lung liver muscle skeletal kidney pancreas spleen thymus prostate testis ovary small intestine colon peripheral blood

Figure 2 Tissue distribution of human RTN-x mRNA. Northern blot analysis was performed using 32P-labeled cDNA probes for human RTN-x. The positions of mRNA for RTN-xL and RTN-xS are indicated by arrowheads. Asterisks indicate the transcripts that were not characterized

a lesser amount also being detected in the crude results were obtained for NSP-C, Bcl-xL and Bcl-2 nuclear and mitochondria-enriched fractions (Figure (Figure 4b ± d, N+). 4a,b). Immunocytochemical analysis of COS-7 cells transfected with RTN-x revealed a reticular pattern of S Subcellular localization of Bcl-2 and Bcl-x is altered by RTN-x and its colocalization with a typical ER L S RTN-x resident chaperone protein, Grp78 (Bip) (Figure 5a), S similarly to the previous immunohistochemical studies Since the predominant subcellular localization of RTN- on reticulons including NSP-C (Senden et al., 1996; xS diers from that of Bcl-xL or Bcl-2 (Figure 4), we GrandPre et al., 2000). These results indicate that examined whether the subcellular localization of Bcl-xL RTN-xS is mainly localized to the ER. The results and Bcl-2 was altered by RTN-xS or not. D98/AH2 together with the sequence homology between RTN-xS cells stably expressing Bcl-xL or Bcl-2 were transiently and NSP-C (Figure 1b), revealed that RTN-xS is a transfected with RTN-xS, NSP-C, or empty vector, member of reticulon family. followed by fractionation and Western blotting. The Bcl-2 was mainly localized in the mitochondria- quality of each fractionation procedure was veri®ed by enriched fraction and the nuclear envelope, with a the detection of known compartment-speci®c proteins, small amount in the microsomal fractions (Figure 4c), including VDAC as a mitochondrial protein and as reported previously (Krajewski et al., 1993; Akao presenilin-1 as an ER protein. Although VDAC was et al., 1994). On the other hand, Bcl-xL was detected in the crude nuclear fraction, this was due to predominantly recovered in the cytoplasmic fraction the presence of partially broken cells and/or mitochon- under our fractionation conditions (Figure 4d). dria. However, there was no contamination of the However, immunocytochemical study showed that a mitochondria-enriched fraction by the microsome- majority of Bcl-xL was localized on the mitochondria enriched fraction, and vice versa (Figure 6a,b,c, 3rd or peripheral to the mitochondria (Figure 5b), and 4th panels). Bcl-xL recovered in the microsomal consistent with the previous observation (Gonzalez fraction greatly increased from 15+5to50+4 and et al., 1994). The discrepancy in the results of 32+3% when cells were transfected with RTN-xS or biochemical fractionation and immunocytochemistry NSP-C, respectively (Figure 6a,b,c, top panels), of Bcl-xL subcellular localization was also observed in suggesting translocation of Bcl-xL to the microsomal previous studies (Hsu et al., 1997; Jia et al., 1999; fraction. Similar change in subcellular localization of Hausmann et al., 2000), and was probably due to Bcl-2 proteins from the mitochondria-enriched fraction loose association of Bcl-xL to the mitochondria and to the microsomal fraction was observed when RTN-xS other membranes. was transfected (Figure 6b, 2nd panel). When cells When treated with 0.5% NP40, which strips o the were transfected with NSP-C, which did not interact nuclear envelope, the nuclear fraction lost virtually all with Bcl-2 (Figure 3b), the translocation of Bcl-2 was + RTN-xS protein (Figure 4a, N ), suggesting that RTN- not observed (Figure 6c, 2nd panel). It is noteworthy xS was localized on the nuclear envelope. Identical that because the eciency of the transfection in each

Oncogene Translocation of Bcl-xL and Bcl-2 to ER by RTN-xS S Tagami et al 5740 L L X L X X /Bcl- /Bcl-2 S S X X L L L X X X NSP-C/Bcl-2 Vector/Bcl-2 RTN- NSP-C/Bcl- Vector/Bcl- RTN- /Bcl- /Bcl-2 S S X X HA HA HA HA HA HA Vector/Bcl- NSP-C/Bcl- NSP-C/Bcl-2 α NRS α NRS α NRS α NRS α NRS α NRS RTN- RTN- Vector/Bcl-2 L X Bcl- Bcl-2 α α NMI NHI

Figure 3 Interaction of Bcl-xL and Bcl-2 with RTN-xS or NSP-C. (a and b) pCAGGS-HA-RTN-xS, pCAGGS-HA-NSP-C, or the empty vector was cotransfected with pCAGGS-Bcl-xL (a) or pCAGGS-Bcl-2 (b) into COS-7 cells and cell lysates were immunoprecipitated (IP) using anti-HA antiserum (aHA) or normal rabbit serum (NRS) as a control. Each sample was subjected to SDS ± PAGE and Western blotting using an anti-Bcl-xL mAb (a: top) or an anti-Bcl-2 mAb (b: top). Then the same ®lters were reprobed with an anti-HA mAb to con®rm the presence of HA-RTN-xS and HA-NSP-C in the immunoprecipitates (bottom). Part of each precleared lysate was also loaded to show equivalence with Bcl-xL (a) or Bcl-2 (b) in the lysates. (c and d) Jurkat cells were treated with DTBP, and then cell lysates were immunoprecipitated using an anti-Bcl-xL mAb (aBcl-xL) and normal mouse IgG (NMI) as a control (c), or using an anti-Bcl-2 mAb (aBcl-2), and normal hamster IgG (NHI) as a control (d). RTN-xS in the immunoprecipitates was detected using anti-RTN-xS antibody

sample was approximately 50%, about half of the cells not shown). Transfection with Bcl-2 and RTN-xS used in the biochemical fractionation should show the showed the similar change in staining pattern of Bcl- normal localization for Bcl-xL and Bcl-2. 2, while transfection with NSP-C did not alter the To con®rm that subcellular localization of Bcl-xL is staining pattern of Bcl-2 (data not shown). Together changed from the mitochondria to the ER when RTN- with the results of subcellular fractionation, these xS is expressed, immunocytochemical studies were ®ndings suggested that Bcl-xL and Bcl-2 partly changed performed. When COS-7 cells were transfected with their subcellular localization to the ER, by binding Bcl-xL alone, an antibody speci®c to Bcl-xL predomi- with RTN-xS or NSP-C. nantly stained the mitochondria (Figure 5b), and only to a lesser extent the ER (Figure 5d). When cells were RTN-x reduces the anti-apoptotic activity of Bcl-x and cotransfected with Bcl-x and RTN-x , immunostain- S L L S Bcl-2 ing of Bcl-xL showed a lesser degree of overlap with Mito Tracker (Figure 5c) and a greater degree of We next examined the eect of RTN-xS or NSP-C on overlap with Bip (Figure 5e). When NSP-C was used the preventive activity of Bcl-xL and Bcl-2 against instead of RTN-xS, similar results were observed (data apoptosis induced by tunicamycin (TM) and stauros-

Oncogene Translocation of Bcl-xL and Bcl-2 to ER by RTN-xS S Tagami et al 5741 using apoptosis induced by STS. Like TM-induced apoptosis, the ability of Bcl-xL to prevent apoptosis by STS was reduced by RTN-xS and NSP-C, although the eect was weaker (Figure 7c). Similarly, the anti- apoptotic activity of Bcl-2 against STS was reduced by RTN-xS, but not by NSP-C (Figure 7d). Therefore, inhibition of the anti-apoptotic activity of Bcl-xL and Bcl-2 by RTN family members was not restricted to ER stress-induced apoptosis. From these results, we concluded that RTN-xS and NSP-C reduced the anti-apoptotic activity of Bcl-xL and Bcl-2 by speci®c binding, with one of the mechanisms involving the sequestration of Bcl-xL and Bcl-2 from the mitochondria by RTN family proteins.

Discussion

In the present study, we showed that a novel member of the RTN family, RTN-xS, interacted with both Bcl-xL and Bcl-2, while another family member, NSP- C, only interacted with Bcl-xL in mammalian cells as well as in the yeast two-hybrid system. Binding with RTN-xS or NSP-C partly changed the subcellular Figure 4 Subcellular localization of RTN-x and NSP-C S localization of Bcl-xL and Bcl-2 to the ER. protein. D98/AH2 cells transfected with pCAGGS-HA-RTN-xS Consistent with their binding speci®city, RTN-x (a), pCAGGS-HA-NSP-C (b), pCAGGS-Bcl-2 (c), or pCAGGS- S decreased the anti-apoptotic activity of both Bcl-x Bcl-xL (d) were subjected to biochemical fractionation as L described in Materials and methods. Each fraction corresponding and Bcl-2, while NSP-C only reduced that of Bcl-xL to equal numbers of cells was analysed by Western blotting using and not that of Bcl-2. Because Bcl-xL and Bcl-2 are an anti-HA mAb (a and b), an anti-Bcl-2 mAb (c), or an anti-Bcl- + thought to exert their anti-apoptotic activity mainly xL mAb (d). T, total cell lysate; N, crude nuclear fraction; N , nuclear fraction after treatment with 0.5% NP40; Mt, mitochon- through an eect on the mitochondria (reviewed by dria-enriched fraction; Ms, microsomal fraction; C, cytoplasmic Green and Reed, 1998; Tsujimoto and Shimizu, fraction 2000), one mechanism for impairment of the anti- apoptotic activity of Bcl-2 family proteins by RTN-xS and NSP-C is likely to involve their change in subcellular localization from the mitochondria to the porine (STS). TM is a speci®c inhibitor of N- ER. glycosylation in the ER, and thus causes ER stress, Sequence alignment analysis showed that RTN-xS while STS is a wide-range kinase inhibitor. and NSP-C were not members of the Bcl-2 family. When stable vector transfectants of D98/AH2 cells Among Bcl-2-interacting proteins that do not belong were transiently transfected with the empty vector, to the Bcl-2 family (reviewed by Tsujimoto, 1998), about 70% of the cells underwent apoptosis after 48 h only 14-3-3 is well documented to regulate the of exposure to ER stress by treatment with TM (Figure subcellular localization of Bad, a pro-apoptotic Bcl-2 7a,b). Apoptosis was reduced to about 11% in D98/ family member (Zha et al., 1996). Protein 14-3-3 binds AH2 cells overexpressing Bcl-xL or Bcl-2 (Figure 7a,b). to Bad that has been phosphorylated, and prevents its When these cells were transiently transfected with the translocation to the mitochondria, consequently pre- pCAGGS-RTN-xS, apoptosis increased to about 45% venting apoptosis mediated by Bad (Zha et al., 1996). (Figure 7a,b). Similar results were obtained when NSP- On the other hand, none of the Bcl-2-interacting C was introduced into D98/AH2 cells overexpressing proteins from outside the Bcl-2 family have been Bcl-xL but not D98/AH2 cells overexpressing Bcl-2 reported to be involved in the subcellular transloca- (Figure 7a,b). Because NSP-C sensitized the vector tion of Bcl-xL and Bcl-2. RTN-xS and NSP-C transfectants of D98/AH2 cells to TM to a similar speci®cally interact with anti-apoptotic Bcl-2 family extent as RTN-xS (Figure 7a,b, closed diamonds and proteins and not with pro-apoptotic family members, triangles), prevention of the anti-apoptotic activity of for example Bax (unpublished observations), indicat- Bcl-xL and Bcl-2 by RTN family proteins was not due ing that RTN-xS and NSP-C are the ®rst non-Bcl-2 to the additive eect of the treatment and/or their family proteins that have been found to regulate the transfection. Furthermore, RTN-xS and NSP-C did not subcellular localization of anti-apoptotic Bcl-2 family aect the levels of Bcl-2 and Bcl-xL (Figure 3a,b, and members. data not shown). These results indicated the RTN-xS It was reported that anti-apoptotic Bcl-2 family decreased the anti-apoptotic activity of both Bcl-xL and proteins changed their subcellular localization upon Bcl-2, while NSP-C only reduced that of Bcl-xL in apoptotic stimulation. For example, some of the Bcl-2 response to ER stress. on the ER moves to the mitochondria when CEM cells To examine whether the binding-dependent reduc- are treated with TNF-a (Jia et al., 1999). Bcl-xL, which tion of the anti-apoptotic activity of Bcl-xL and Bcl-2 is recovered in both the soluble fraction and the by RTN-xS and NSP-C also occurred with other membrane pellet under unstimulated conditions, be- apoptotic stimuli, similar experiments were performed comes predominantly membrane-associated within 2 h

Oncogene Translocation of Bcl-xL and Bcl-2 to ER by RTN-xS S Tagami et al 5742

Figure 5 Change in subcellular localization of Bcl-xL from mitochondria to endoplasmic reticulum by RTN-xS. COS-7 cells transfected with HA-RTN-xS alone (a), Bcl-xL alone (b, d) or Bcl-xL and RTN-xS (c, e) were immunostained by anti-HA mAb (a, left panel, green), anti-Bcl-xL mAb (b ± e, left panel, green), and by anti-Bip mAb (a, d, e middle panel, red) or stained with Mito Tracker (b, c middle panel, red) as described in Materials and methods. Right side panels in a ± e were overlaid images of left and middle panels, and yellow staining indicates the extent of colocalization. Shown are the representatives of three independent experiments

after the induction of apoptosis by dexamethasone or blocked their anti-apoptotic activity. Thus, localization g-irradiation (Hsu et al., 1997). Translocation of Bcl-xL of anti-apoptotic Bcl-2 proteins to the mitochondria and Bcl-2 to the mitochondria upon apoptotic might be regulated by their interaction with RTN-xS, stimulation could be involved in the rescue of NSP-C, and other RTN family proteins. mitochondrial dysfunction and the prevention of Although several studies have shown that Bcl-2 is apoptosis, because retention of these proteins on the localized to the ER and nuclear envelope in addition ER by binding to overexpressed RTN-xS and NSP-C to the mitochondrial membrane (Krajewski et al.,

Oncogene Translocation of Bcl-xL and Bcl-2 to ER by RTN-xS S Tagami et al 5743

Figure 6 Change in subcellular distribution of Bcl-xL and Bcl-2 by RTN-xS or NSP-C. D98/AH2 cells stably overexpressing Bcl-xL (top panels in a, b, and c) or Bcl-2 (second panels in a, b, and c) were transfected with the empty vector (a), HA-RTN-xS (b), or HA-NSP-C (c). Cells were fractionated as described in the legend to Figure 4 and analysed by Western blotting using anti-Bcl-xL and anti-Bcl-2 mAbs. A fraction (%; mean+s.d.) of Bcl-xL or Bcl-2 recovered in three dierent fractions derived from low speed supernatant (a mitochondria-enriched fraction, a microsomal fraction, and a cytosolic fraction) was calculated from three independent experiments, and shown under each panel. The translocation of Bcl-xL to the ER by RTN-xS or NSP-C and that of Bcl-2 by RTN-xS were signi®cant (**P50.02), while that of Bcl-2 by NSP-C was not signi®cant (#P40.2). Fractionated lysates from Bcl-xL-expressing cells were also subjected to Western blotting using anti-VDAC mAb (third panels in a, b, and c), or an anti- presenilin-1 pAb (bottom panels in a, b, and c). Fractionated lysates from Bcl-2-expressing cells showed virtually identical results on Western analysis with anti-VDAC and anti-presenilin-1 antibodies

1993; Akao et al., 1994), the role of Bcl-2 in the ER al., 1997). RTN-xS might interact with Bcl-xL and has remained unclear when compared with that in Bcl-2 to prevent the formation of a Bcl-2-containing the mitochondria. As shown in Figure 7, reduction anti-apoptotic complex on the ER and/or to of the anti-apoptotic activity of Bcl-xL and Bcl-2 by facilitate the release of apoptotic factors. RTN-xS or that of Bcl-xL by NSP-C was stronger in RTN family proteins, including RTN-xS and NSP- TM-induced apoptosis than STS-induced one, sug- C, seem to be a useful tool to investigate the eects gesting that the complex of Bcl-xL or Bcl-2 with of Bcl-xL and Bcl-2 on the ER. Recently, it was RTN-xS or NSP-C formed on ER, or the over- reported that the product of the human cDNA clone expression of RTN-xS and NSP-C itself might have KIAA0886 (designated Nogo-A, which is the same some role in inducing apoptosis under ER stress. as RTN-xL) and its rat homologue could be Recent studies have suggested that ER stress triggers localized on the surface of oligodendrocytes or the formation of unfolded proteins and thus leads to myelin where it inhibited neurite outgrowth (Grand- the induction of several heat shock-related proteins Pre et al., 2000; Chen et al., 2000), although the through phosphorylation of an ER protein, IRE1 majority of Nogo-A was localized on the ER. It is (reviewed by Kaufman, 1999). Several candidate of interest to note that neuronal overexpression of molecules have been suggested to be released from Bcl-2 promotes neurite outgrowth and axonal the ER by ER stress. Such factors might also regeneration both in vivo (Chen et al., 1997) and activate initiator caspases, such as caspase-12, which in vitro (Sato et al., 1994; Oh et al., 1996). Although was recently shown to be essential for apoptosis myelin-associated Nogo-A acts on neurons (Grand- induced by ER stress (Nakagawa et al., 2000). It is Pre et al., 2000; Chen et al., 2000), it is conceivable possible that Bcl-xL and Bcl-2 form a complex with that neurite outgrowth could be regulated by an these caspases and some unknown adopter mole- intrinsic neuronal Nogo-A/RTN-xL. Therefore, it is cules, like the apoptosome found on the mitochon- possible that Bcl-xL and Bcl-2 promote neurite drial membrane (Pan et al., 1998; Hu et al., 1998). outgrowth through inhibition of the cell surface Bap31, which interacts with Bcl-xL and Bcl-2 and is expression of Nogo-A/RTN-xL by sequestration to cleaved by caspase-8 (Ng et al., 1997), is one the ER. Taken together with the recent suggestion candidate for such an adopter. It was also reported that ER stress is involved in some neurodegenerative that Bcl-2 interacts with the sarcoplasmic Ca2+ diseases such as Alzheimer's disease, further studies ATPase pump (Kuo et al., 1998), and that Bcl-xL on RTN and Bcl-xL/Bcl-2 may provide some insights and Bcl-2 can regulate the release of calcium from not only into the eects of Bcl-xL/Bcl-2 on the ER intracellular stores, which are primarily located in but also into neurodegenerative diseases and axonal the ER (Bay et al., 1993; Lam et al., 1994; He et regeneration.

Oncogene Translocation of Bcl-xL and Bcl-2 to ER by RTN-xS S Tagami et al 5744

Figure 7 Eect of RTN-xS and NSP-C on the anti-apoptotic activity of Bcl-xL and Bcl-2 in cells exposed to TM and STS. D98/AH2 cells stably expressing Bcl-xL (a and c) or Bcl-2 (b and d) were transiently transfected with pCAGGS-HA-RTN-xS (^), pCAGGS-HA-NSP-C (~), or the empty vector (&) together with a reporter pEGFP-F plasmid. Vector transformants were similarly transfected (closed symbols). After culture for 36 h, cells were treated with 1.5 mg/ml of TM (a and b)or1mM of STS (c and d) and were examined for apoptosis at the indicated times as described in Materials and methods. Results were expressed as mean+s.d. calculated from three independent experiments (***P50.003, **P50.01, *P50.03, all of which were regarded as signi®cant, while #P40.07 as not signi®cant)

Materials and methods 18) and anti-Bcl-2 pAb (N-19) were obtained from Santa Cruz Biotechnology. Anti-presenilin-1 pAb was kindly Plasmid construction provided by Dr Takashi Kudo (Osaka University). For use in this study, anti-RTN-x pAb was raised by immunizing pBTM116-Bcl-x was constructed for yeast two-hybrid S L rabbits with the peptide 48DLEELEVLERKPAAGLSA65-C, screening to produce Bcl-x fused to the DNA-binding L and anity-puri®ed by the antigenic peptide immobilized to domain of LexA. pBTM116-Bcl-2 and pBTM116-Rad51 CNBr-activated SepharoseTM 4B (Amersham Pharmacia were described previously (Iwahashi et al., 1997). The genes Biotech). for HA-RTN-xS and HA-NSP-C were constructed so that the HA epitope were fused to the second codon of RTN-xS and NSP-C using PCR primers. HA-RTN-xS, HA-NSP-C, human Yeast two-hybrid screening Bcl-x , and human Bcl-2 cDNAs were cloned in the vector L The L40 yeast strain transfected with pBTM116-Bcl-x was pCAGGS bearing a chicken b-actin promoter and CMV L used to screen oligo (dT) and/or random-primed human fetal enhancer for expression in mammalian cells. and adult brain cDNA fusion libraries in a Gal4-activating domain vector (pGAD10, Clontech). Details of the methods employed were described previously (Iwahashi et al., 1997). Antibodies Anti-HA mAb (clone 12CA5), anti-HA antiserum, anti-Bcl- cDNA cloning of RTN-x xL mAb (#44), anti-Bcl-2 mAb (6C8), anti-VDAC mAb (31HL), and Anti-Grp78 (Bip) mAb (10C3) were purchased Searching the EST database, we identi®ed THC 167126 and from Boehringer Mannheim, Medical and Biochemical THC 161296 carrying information of 5'-neighboring region of Laboratory (MBL, Japan), Transduction Labs, Pharmingen, 4-20 s clone. To obtain the cDNA for the entire coding Calbiochem, and StressGen, respectively. Anti-Bcl-xL pAb (S- region of RTN-x, the human fetal brain 5' STRETCH cDNA

Oncogene Translocation of Bcl-xL and Bcl-2 to ER by RTN-xS S Tagami et al 5745 library (Clontech) was screened by the plaque hybridization Version 3.1 (Fuji Film). Student's t-test was used to evaluate procedure using the PCR-generated 713 bp fragment (from signi®cance in dierences between amounts of Bcl-xL and the ®rst base pair of THC 161296 to the 295th bp of the THC Bcl-2 recovered in each fraction and those in respective 167126) as a probe. Initially, eight independent clones were control fraction. Less than 5% of probability was regarded as obtained from 16106 clones. Since we could not ®nd an in- signi®cant dierence in the present experiments. frame stop codon in upstream region from candidate initiation codons in these clones, the screening was done Immunocytochemistry again using the clone with the longest cDNA as a probe, and we obtained four independent clones from 16106 clones with COS-7 cells were plated at 20% con¯uence onto 33 mm poly- an in-frame stop codon at 126 bp upstream of the putative L-lysine coated dishes. Cells were transiently transfected with initiation site in the cDNA. The sequence of the 5' region was a total of 0.5 mg of indicated plasmids. Thirty-six hours after con®rmed by sequencing the RT ± PCR fragment obtained transfection, cells were incubated in 50 ng/ml of Mito using human brain mRNA as a template. The resultant Tracker Orange, CMTM-Ros (Molecular Probes) to stain cDNA had a 3576 bp open reading frame, and was named the mitochondria for 10 min. Cells with or without Mito RTN-xL. We also obtained six independent clones that lacked Tracker treatment were washed with PBS, ®xed in 4% base pairs 556 to 3013 of the longer form, giving an 1119 bp paraformaldehyde, and permeabilized in 0.5% Triton X-100. open reading frame, which was named RTN-xS. Immunostaining was performed using anti-HA mAb or anti- Bcl-xL mAb, followed by Alexa 488-conjugated anti-mouse IgG antibody (Molecular Probes). In some cases, cells were Northern blotting subsequently stained with anti-Bip mAb and Alexa 543- Human poly(A)+ RNA tissue blots (Clontech) were conjugated anti-mouse IgG antibody. Cells were then hybridized according to the manufacturer's instructions using examined with a confocal ¯uorescence microscope (Zeiss). 32P-labeled DNA fragments corresponding to amino acid residues 45 to 204 of RTN-x as a probe. The blot was ®nally S Apoptosis assay washed with 16SSC containing 0.1% SDS for 30 min at 658C. Stable transformants of D98/AH2 cells expressing human Bcl-xL were generated by electroporation. Stable transformants of D98/AH2 cells expressing human Bcl-2 were Immunoprecipitation and Western blotting described previously (Iwahashi et al., 1997). Cells (2 ± COS-7 cells (2 ± 56106) were transfected with 3 mg each of 36105) were transiently transfected with indicated plasmids the indicated plasmids using LipofectAMINETM (Gibco (1 mg) together with the reporter plasmid pEGFP (green BRL). Immunoprecipitation was performed as described ¯uorescent protein)-F (0.1 mg; Clontech) using LipofectA- elsewhere (Iwahashi et al., 1997). In some experiments, MINE 2000TM. After culture for 36 h, cells were treated with immunoprecipitation was carried out using Jurkat cells 1.5 mg/ml of tunicamycin (Sigma) or 1 mM of staurosporine treated with a cleavable chemical cross-linker DTBP (Wako Biochemicals, Japan) for the indicated periods, (Dimethyl 3,3'-dithiobispropionimidate-2HCl) as described stained with 10 mM Hoechst 33342 (Calbiochem), and elsewhere (Narita et al., 1998). Immune complexes were examined under a ¯uorescent microscope. At least 200 subjected to Western blot analysis using speci®c antibodies GFP-positive cells were examined to determine whether they and an enhanced chemiluminescence system (Amersham). were viable or apoptotic from their nuclear morphology. The mean+s.d. of the ratio of apoptotic cells was calculated from three independent experiments. Student's t-test was used to Subcellular fractionation evaluate the eects of RTN-xS and NSP-C on Bcl-xL and Bcl- D98/AH2 cells (2 ± 56106), derived from HeLa cells, were 2 expressing cells. transiently transfected with 3 mg each of the indicated plasmids using LipofectAMINE 2000TM (Gibco BRL). After culture for 36 h, the cells were collected, washed with PBS, Abbreviations and suspended in a hypotonic solution (10 m HEPES ± M ER, endoplasmic reticulum; mAb, monoclonal antibody; NaOH, pH 7.4, 10 m MgCl ,42mM KCl) for 5 min on ice. M 2 pAb, polyclonal antibody; TM, tunicamycin; STS, staur- Then the cells were passed through a 30-gauge needle once osporine; VDAC, voltage-dependent anion channel and centrifuged at 600 g for 10 min to collect a crude nuclear pellet, which contained both nuclei and residual undisrupted cells. The supernatant was centrifuged at 7000 g for 10 min to yield a mitochondria-enriched pellet. This supernatant was Acknowledgments further centrifuged at 100 000 g for 90 min to yield a We thank Dr Kudo for providing anti-presenilin-1 anti- microsomal pellet and a cytosolic fraction. All of the pellets body. We also thank Mr Kawate and Dr Kashiwagi for were dissolved in 26PBS, 1% NP40, 0.5% sodium help with the immunocytochemical studies. This work was deoxycholate, and 0.1% sodium dodecyl sulfate. To obtain supported in part by Grants-in-Aid for Scienti®c Research a nuclear fraction without the outer membrane, cells were on Priority Areas and for COE Research from the suspended in the hypotonic solution with 0.5% NP40 and Japanese Ministry of Education, Science, Sport and were centrifuged at 12 000 g for 10 min. Three independent Culture. experiments were performed in order to examine the translocation of Bcl-xL and Bcl-2. A fraction (%, mean+s.d.) of Bcl-xL or Bcl-2 in three dierent fractions derived from Accession number low speed supernatant (a mitochondrial-enriched fraction, a The nucleotide sequences of the RTN-xL and RTN-xS have microsomal fraction, and a cytosolic fraction) was calculated been submitted to the DDBJ/EMBL/GenBank (Accession by densitometric analysis using Science Lab 98 Image Gauge No. AB040462 and AB040463, respectively).

References

Adams JM and Cory S. (1998). Science, 281, 1322 ± Akao Y, Otsuki Y, Kataoka S, Ito Y and Tsujimoto Y. 1326. (1994). Cancer Res., 54, 2468 ± 2471.

Oncogene Translocation of Bcl-xL and Bcl-2 to ER by RTN-xS S Tagami et al 5746 Arends MJ and Wyllie AH. (1991). Int. Rev. Exp. Pathol., 32, Moreira EF, Jaworski CJ and Rodriguez IR. (1999). 223 ± 254. Genomics, 58, 73 ± 81. Bay G, Miyashita T, Williamson JR and Reed JC. (1993). J. Nagase T, Ishikawa K, Suyama M, Kikuno R, Hirosawa M, Biol. Chem., 268, 6511 ± 6519. MiyajimaN,TanakaA,KotaniH,NomuraNandOhara Chen DF, Schneider GE, Martinou JC and Tonegawa S. O. (1998). DNA Res., 5, 355 ± 364. (1997). Nature, 385, 434 ± 439. Nakagawa T, Zhu H, Morishima N, Li E, Xu J, Yankner BA Chen MS, Huber AB, van der Haar ME, Frank M, Schnell L, and Yuan J. (2000). Nature, 403, 98 ± 103. Spillmann AA, Christ F and Schwab ME. (2000). Nature, Narita M, Shimizu S, Ito T, Chittenden T, Lutz RJ, Matsuda 403, 434 ± 439. H and Tsujimoto Y. (1998). Proc.Natl.Acad.Sci.USA, GonzalezGM,PerezBR,DingL,DuanL,BoiseLH, 95, 14681 ± 14686. Thompson CB and Nunez G. (1994). Development, 120, NgFW,NguyenM,KwanT,BrantonPE,NicholsonDW, 3033 ± 3042. Cromlish JA and Shore GC. (1997). J. Cell Biol., 139, GrandPre T, Nakamura F, Vartanian T and Strittmatter 327 ± 338. SM. (2000). Nature, 403, 439 ± 444. Oh YJ, Swarzenski BC and O'Malley KL. (1996). Neurosci. Green DR and Reed JC. (1998). Science, 281, 1309 ± 1312. Lett., 202, 161 ± 164. Hausmann G, O'Reilly LA, van Driel R, Beaumont JG, Pan G, O'Rourke K and Dixit VM. (1998). J. Biol. Chem., Strasser A, Adams JM and Huang DC. (2000). J. Cell 273, 5841 ± 5845. Biol,. 149, 623 ± 634. Roebroek AJ, van de Velde HJ, Van Bokhoven A, Broers JL, He H, Lam M, McCormick TS and Distelhorst CW. (1997). Ramaekers FC and Van de Ven WJ. (1993). J. Biol. Chem., J. Cell Biol., 138, 1219 ± 1228. 268, 13439 ± 13447. Hsu YT, Wolter KG and Youle RJ. (1997). Proc. Natl. Acad. Roebroek AJ, Contreras B, Pauli IG and Van de Ven WJ. Sci. USA, 94, 3668 ± 3672. (1998). Genomics, 51, 98 ± 106. Hu Y, Benedict MA, Wu D, Inohara N and Nunez G. (1998). Salvesen GS and Dixit VM. (1997). Cell, 91, 443 ± 446. Proc. Natl. Acad. Sci. USA, 95, 4386 ± 4391.1. Sato N, Hotta K, Waguri S, Nitatori T, Tohyama K, Kerr JF, Wyllie AH and Currie AR. (1972). Br.J.Cancer., Tsujimoto Y and Uchiyama YJ. (1994). J. Neurobiol., 25, 26, 239 ± 257. 1227 ± 1234. Iwahashi H, Eguchi Y, Yasuhara N, Hanafusa T, Matsuza- Senden NH, Timmer ED, Boers JE, van de Velde HJ, wa Y and Tsujimoto Y. (1997). Nature, 390, 413 ± 417. Roebroek AJ, Van de Ven WJ, Broers JL and Ramaekers Jia L, Macey MG, Yin Y, Newland AC and Kelsey SM. FC. (1996). Eur. J. Cell Biol., 69, 197 ± 213. (1999). Blood, 93, 2353 ± 2359. Shimizu S, Narita M and Tsujimoto Y. (1999). Nature, 399, Kaufman RJ. (1999). Genes Dev., 13, 1211 ± 1233. 483 ± 487. Kerr JF, Wyllie AH and Currie AR. (1972). Br. J. Cancer, 26, Takayama S, Sato T, Krajewski S, Kochel K, Irie S, Millan 239 ± 257. JA and Reed JC. (1995). Cell, 80, 279 ± 284. Kluck RM, Bossy-Wetzel E, Green DR and Newmeyer DD. Thornberry NA and Lazebnik Y. (1998). Science, 281, (1997). Science, 275, 1132 ± 1136. 1312 ± 1316. Krajewski S, Tanaka S, Takayama S, Schibler MJ, Fenton W Tsujimoto Y. (1998). Genes Cells., 3, 697 ± 707. and Reed JC. (1993). Cancer Res., 53, 4701 ± 4714. Tsujimoto Y and Shimizu S. (2000). FEBS Lett., 466, 6 ± 10. KuoTH,KimHR,ZhuL,YuY,LinHMandTsangW. Yang J, Liu J, Bhalla K, Kim CN, Ibrado AM, Cai J, Peng (1998). Oncogene, 17, 1903 ± 1910. TI, Jones DP and Wang X. (1997). Science, 275, 1129 ± Lam M, Dubyak G, Chen L, Nunez G, Miesfeld RL and 1132. Distelhorst CW. (1994). Proc. Natl. Acad. Sci. USA, 91, ZamzamiN,SusinSA,MarchettiP,HirschT,Gomez- 6569 ± 6573. Monterrey I, Castedo M and Kroemer G. (1996). J. Exp. LeeJH,TakahashiT,YasuharaN,InazawaJ,KamadaS Med., 183, 1533 ± 1544. and Tsujimoto Y. (1999). Oncogene, 18, 6183 ± 6190. Zha J, Harada H, Yang E, Jockel J and Korsmeyer SJ. Li P, Nijhawan D, Budihardjo I, Srinivasula SM, Ahmad M, (1996). Cell, 87, 619 ± 628. Alnemri ES and Wang X. (1997). Cell, 91, 479 ± 489. Zhu W, Cowie A, Wasfy GW, Penn LZ, Leber B and Liu X, Kim CN, Yang J, Jemmerson R and Wang X. (1996). Andrews DW. (1996). EMBO J., 15, 4130 ± 4141. Cell, 86, 147 ± 157. Zou H, Henzel WJ, Liu X, Lutschg A and Wang X. (1997). MarzoI,BrennerC,ZamzamiN,JurgensmeierJM,Susin Cell, 90, 405 ± 413. SA, Vieira HL, Prevost MC, Xie Z, Matsuyama S, Reed JC and Kroemer G. (1998). Science, 281, 2027 ± 2031.

Oncogene