- Email: [email protected]
Lack of cytotoxic property in a variant of Epstein–Barr virus latent membrane protein-1 isolated from nasopharyngeal carcinoma
Cellular Signalling 16 (2004) 1071 – 1081 www.elsevier.com/locate/cellsig
Lack of cytotoxic property in a variant of Epstein–Barr virus latent membrane protein-1 isolated from nasopharyngeal carcinoma Takeshi Nitta1, Ayako Chiba, Naoki Yamamoto, Shoji Yamaoka * Department of Molecular Virology, Graduate School, Tokyo Medical and Dental University, 1-5-45, Yushima, Bunkyo, Tokyo 113-8519, Japan Received 29 February 2004; accepted 1 March 2004 Available online 21 April 2004
Abstract Latent membrane protein 1 (LMP1) encoded by Epstein – Barr virus (EBV) is a membrane protein that activates multiple signaling pathways and transcription factors, including NF-nB. Our recent report demonstrated that expression of LMP1 induced programmed cell death in an NFnB-dependent manner. In this study, we demonstrate that a variant CAO-LMP1 derived from EBV-infected nasopharyngeal carcinoma (NPC) does not induce cell death unlike the prototype B95.8-LMP1, although both types of LMP1 show NF-nB activation to a similar extent. Studies with chimeric or mutated proteins identified two amino acids in the transmembrane domain, which are commonly substituted in NPC-derived LMP1 variants, being critical for cell death induction by B95.8-LMP1. Furthermore, we show that the B95.8 transmembrane domain cooperates with NF-nB to trigger cell death program. Thus, our results reveal a particular feature of the transmembrane domain of tumor-derived CAO-LMP1 and suggest its possible contribution to the pathogenesis of NPC. D 2004 Elsevier Inc. All rights reserved. Keywords: Epstein – Barr virus; LMP1; Nasopharyngeal carcinoma; NF-nB
1. Introduction Latent membrane protein 1 (LMP1) is a membrane protein encoded by Epstein –Barr virus (EBV) [1]. LMP1 is a type of oncogenic protein that is essential for EBV-mediated transformation of B lymphocytes and can induce malignant transformation of rodent fibroblast cell lines. LMP1 consists of a short cytoplasmic N-terminal domain, transmembrane domain containing six membrane-spanning segments and five intra/extracellular loops, and long cytoplasmic C-terminal domain [2]. The transmembrane domain of LMP1 is necessary for its self-multimerization on the plasma membrane and enables the C-terminal cytoplasmic domain to interact with specific proteins that transduce cellular signals [1]. In this manner, LMP1 activates multiple cellular signalAbbreviations: LMP1, latent membrane protein 1; EBV, Epstein – Barr virus; JNK, c-Jun N-terminal kinase; STAT, signal transducer and activator of transcription; NPC, nasopharyngeal carcinoma; EGFP, enhanced green fluorescent protein; aa, amino acid; PARP, poly ADP-ribose polymerase; DEVD-AFC, Asp-Glu-Val-Asp-7-amino-4-trifluoromethyl coumarin. * Corresponding author. Tel.: +81-3-5803-5178; fax: +81-3-5803-0124. E-mail address: [email protected] (S. Yamaoka). 1 Present address: Division of Experimental Immunology, Institute for Genome Research, University of Tokushima, 3-18-15 Kuramoto, Tokushima 770-8503, Japan. 0898-6568/$ - see front matter D 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.cellsig.2004.03.001
ing cascades leading to activation of NF-nB [1], c-Jun Nterminal kinase (JNK) [3– 5], signal transducer and activator of transcription (STAT) [6] and p38 mitogen-activated protein kinase [7]. Among them, activation of NF-nB is known to be required for the cell-transforming activity of LMP1 [8]. LMP1 variants have been obtained from geographically distinct EBV isolates or from different types of EBV-associated tumors. However, correlation has not been well established among their sequences, biological functions and possible contribution to EBV-associated malignancies [9– 11]. B95.8-LMP1, an LMP1 clone originally derived from a B lymphocyte, has been extensively studied as a prototype, resulting in elucidation of its signaling mechanisms and biological functions. Regardless of its transforming activity, B95.8-LMP1 was reported to inhibit cell proliferation [12,13] and colony formation [14] or to induce apoptosis [15,16] in several types of cell lines. Most recently, we demonstrated that expression of B95.8-LMP1 induced programmed cell death, which was dependent on activation of NF-nB [17]. Cell death induction by B95.8-LMP1 was prevented by specific inhibition of NF-nB activity and restored by reactivation of NF-nB. B95.8-LMP1 also induced activation of caspase-3, a cysteine protease involved in the execution step of apoptosis [18], in an NF-nB-dependent manner. This
1072
T. Nitta et al. / Cellular Signalling 16 (2004) 1071–1081
LMP1-induced cell death was also expected to be a model to study molecular mechanisms of NF-nB-dependent cell death induction, although its pathological consequences in EBVrelated diseases remained elusive. A clue to this question was found in the CAO-LMP1 variant, which was originally isolated from EBV-associated nasopharyngeal carcinoma (NPC) [19] and had amino acid substitutions most frequently in the N-terminus and transmembrane domains, and amino acid insertions or deletions in the cytoplasmic C-terminus [20]. This LMP1 variant has been reported to show more tumorigenic [21 – 23] and less immunogenic properties [24], enhanced signaling activities [25], and impaired induction of certain phenotypic markers [25]. Here, we demonstrate that CAO-LMP1, unlike B95.8-LMP1, does not induce cell death and that two amino acids in the transmembrane domain that are commonly substituted in NPCderived LMP1s are critical for cell death induction. Our results indicate that the transmembrane domain of LMP1 governs its killing property, and suggest possible contribution of the two amino acid substitutions to the pathogenesis of NPC.
B/B-LMP1: the N-terminus of CAO-LMP1 linked to the transmembrane and C-terminal domains of B95.8-LMP1; B/ B/C-LMP1: the N-terminal and transmembrane domains of B95.8-LMP1 linked to the C-terminus of CAO-LMP1; B/C/ C-LMP1: the N-terminus of B95.8-LMP1 linked to the transmembrane and C-terminal domains of CAO-LMP1; C/ C/B-LMP1: the N-terminal and transmembrane domains of CAO-LMP1 linked to the C-terminus of B95.8-LMP1; B/C/ B-LMP1: the transmembrane domain of CAO-LMP1 placed between the N-terminal and C-terminal domains of B95.8LMP1; C/B/C-LMP1: the transmembrane domain of B95.8LMP1 placed between the N-terminal and C-terminal domains of CAO-LMP1. These chimeric constructs were cloned into the pSG5 vector. To generate point mutations, Ile85Leu and Phe106Tyr in B95.8-LMP1, we employed the overlap PCR technique [32], using oligonucleotide primers
5V-GGAGCCCTTTGTCTACTCCTACTGATG-3V 5V-GGACAGGCATTGTACCTTGGAATTGTG-3V
2. Materials and methods
and their complimentary oligonucleotides. Positions mutated are underlined. For the reciprocal mutations Leu85Ile and Tyr106Phe in CAO-LMP1, primers
2.1. Cell culture
293T, Rat-1 and Plat-E [26] cells were maintained in DMEM supplemented with 10% fetal calf serum, 100 units/ ml of penicillin G and 100 Ag/ml of streptomycin. 2.2. Plasmids LMP1 expression vectors pSG5-LMP1 or pSG5-CAOLMP1, in which cDNAs of the prototype LMP1 (B95.8LMP1) or CAO-LMP1 were located downstream of the SV40 promoter, were previously described (kindly provided by Dr. Rowe, University of Wales College of Medicine) [27,28]. Plasmids pRc-CMV-relA [29], Ign-ConAluc and EF1-lacZ (a kind gift of Dr. Memet, Institut Pasteur) [30] were previously described. Plasmid pEGFP-N1 encoding the enhanced green fluorescent protein (EGFP) was purchased from CLONTECH. For retroviral expression, we subcloned B95.8- or CAO-LMP1 cDNAs in a retroviral vector pMRX-IRESEGFP [31]. For construction of B95.8/CAO chimeric proteins, an artificial KpnI site was first introduced at the position corresponding to the amino acid (aa) 192 and 193 of both types of LMP1, to split them into fragments encoding the Nterminal and transmembrane domains (aa 1 – 192) or Cterminal cytoplasmic domain (aa 193 – 386 or 404). The introduction of the KpnI site led to two amino acid substitutions of Gly192 and Thr193 for Ser192 and Asp193 in B95.8LMP1 or for Thr192 and Asp193 in CAO-LMP1. The Nterminus (aa 1 – 26) and transmembrane (aa 27 – 192) domains were separated at the StuI site. Finally, these DNA fragments derived from B95.8- or CAO-LMP1 were appropriately ligated to generate chimeric constructs as follows: C/
5V-GGAGGCCTTGGTATACTCCTACTGATG-3V 5V-GGACAGGCATTGTTCCTTGGAATTGTG-3V
and their complimentary oligonucleotides were used. Mutated DNA fragments were cloned into the pSG5 vector, generating the following plasmids; pSG5-B95.8(L) encoding B95.8-LMP1 Ile85Leu; pSG5-B95.8(Y) encoding B95.8-LMP1 Phe106Tyr; pSG5-B95.8(LY) encoding B95.8-LMP1 Ile85Leu Phe106Tyr; pSG5-CAO(I) encoding CAO-LMP1 Leu85Ile; pSG5-CAO(F) encoding CAOLMP1 Tyr106Phe; pSG5-CAO(IF) encoding CAO-LMP1 Leu85Ile Tyr106Phe. To generate LMP1/EGFP chimeric constructs, an artificial EcoRI site was introduced at the position corresponding to the aa 192– 194. DNA fragments encoding the N-terminal and transmembrane domains of different types of LMP1 were inserted into the EcoRI site of the pEGFP-N1 vector, generating pEGFP-B95.8, pEGFP-B95.8(LY) and pEGFP-CAO. The sequences of all these constructs were verified by DNA sequencing. Full construction details are available upon request. 2.3. Transient transfection and luciferase assay Approximately 6 105 293T cells were transfected by the calcium phosphate coprecipitation method. Flow cytometry analysis of cells transfected with EGFP revealed that transfection efficiencies ranged from 60% to 80%. For measurement of the luciferase activity, 0.1 Ag of IgnConAluc and EF1-lacZ was included in each transfection. Twenty hours after transfection, luciferase activity was determined as previously described and normalized on the basis of h-galactosidase activity [30].
T. Nitta et al. / Cellular Signalling 16 (2004) 1071–1081
1073
2.4. Retroviral infection
2.6. Antibodies and Western blot analysis
For production of retroviruses, Plat-E cells were transfected with pMRX-IRES-EGFP, pMRX-B95.8-IRES-EGFP or pMRX-CAO-IRES-EGFP. Forty-eight hours after transfection, culture supernatants containing retroviruses were collected and stored at 80 jC. Rat-1 cells were infected with these viruses in the presence of polybrene (10 Ag/ml), and cells expressing EGFP were observed under a fluorescent microscope.
Anti-LMP1 antibodies (CS.1 –4) were described previously (kindly provided by Dr. Rowe, University of Wales College of Medicine) [33]. Anti-poly ADP-ribose polymerase (PARP) (H-250), anti-GFP (FL) and anti-RelA antibodies were purchased from Santa Cruz. Whole-cell lysate was prepared by lysing cells with the RIPA buffer (20 mM Tris –HCl [pH 8], 137 mM NaCl, 1 mM MgCl2, 1 mM CaCl2, 10% glycerol, 1% Nonidet P-40, 0.5% deoxycholate, 0.1% SDS and 0.1 mM PMSF) and analyzed as described previously [30].
2.5. Cell death assay 2.7. Enzyme assay for caspase activity Cell death of 293T cells was quantified by counting the number of detached and adherent cells as previously described [17]. After collecting cells detached in medium, adherent cells on the culture plate were collected. The number of cells from each fraction was counted, and the ratio of dead cells was determined as (number of detached cells)/(total cell count).
Activity of caspase-3 family proteases was assessed by using the CPP32/Caspase-3 Fluorometric Protease Assay Kit (MBL) with the fluorescent substrate Asp-Glu-Val-Asp7-amino-4-trifluoromethyl coumarin (DEVD-AFC). Caspase activity was determined as the fluorescence intensity normalized by protein amount.
Fig. 1. B95.8-LMP1, but not CAO-LMP1 induces cell death. (A) 293T cells (6 105) were transfected with 1 Ag of the pSG5 (vector), pSG5-LMP1 or pSG5CAO-LMP1, and photographed 36 h later under a phase-contrast microscope. (B) The extent of cell death determined 36 h after transfection is shown as the ratio of detached cells to total cells, because cells, once detached, did not adhere to the plastic again nor survive in suspension. Whole-cell lysates (15 Ag protein) collected 20 h after transfection, a few hours before the onset of cell death, were subjected to Western blot analysis with anti-LMP1 antibodies. (C) NFnB-dependent transcriptional activity was assessed 20 h after transfection and expressed in fold induction relative to that of vector-transfected cells. Each transfection contained 0.1 Ag of Ign-ConAluc and EF1-lacZ, respectively. The experiments were repeated three times with essentially similar results.
1074
T. Nitta et al. / Cellular Signalling 16 (2004) 1071–1081
Fig. 2. The extreme N-terminal and C-terminal cytoplasmic domains of LMP1 are not involved in cell death induction. (A) Schematic representation of the structure of B95.8-LMP1, CAO-LMP1 and their chimeric derivatives. Solid black lines represent B95.8-LMP1 sequence and open lines denote CAO-LMP1 sequence. (B) 293T cells (6 105) were transfected with 1 Ag of pSG5-C/B/B-LMP1, pSG5-B/B/C-LMP1, pSG5-B/C/C-LMP1 or pSG5-C/C/B-LMP1, and photographed 36 h later. B and C represent B95.8- and CAO-derived LMP1 sequences, respectively. (C) Cells were transfected with 1 Ag of pSG5 (lane 1), pSG5-LMP1 (lane 2), pSG5-C/B/B-LMP1 (lane 3), pSG5-B/B/C-LMP1 (lane 4), pSG5-B/C/C-LMP1 (lane 5) or pSG5-C/C/B-LMP1 (lane 6). The extent of cell death was determined 36 h after transfection as described in Fig. 1. Whole-cell lysates (15 Ag protein) collected 20 h after transfection were subjected to Western blot analysis with anti-LMP1 antibodies. (D) NF-nB-dependent luciferase activity was assessed as described in Fig. 1. The experiments were repeated three times with essentially similar results.
T. Nitta et al. / Cellular Signalling 16 (2004) 1071–1081
3. Results 3.1. CAO-LMP1 does not induce cell death As reported previously, transient expression of B95.8LMP1 in 293T cells induced a remarkable cell death characterized by degeneration and detachment of the cells in an NF-nB-dependent manner [17]. The extent of cell death was quantified by determining the ratio of detached
1075
and adherent cells 36 h after transfection. In contrast, transient expression of an NPC-derived LMP1 variant, CAO-LMP1 did not induce degeneration or detachment of the transfected cells, while its expression level was comparable to that of B95.8-LMP1 (Fig. 1A and B). To confirm the observations obtained in transient transfection of 293T cells, B95.8-LMP1 or CAO-LMP1 was expressed through retroviral infection in a rat fibroblast cell line Rat-1. Expression of B95.8-LMP1 caused severe degeneration and
Fig. 3. The transmembrane domain of B95.8-LMP1 is responsible for induction of cell death. (A) Schematic representation of the structure of B/C/B-LMP1 and C/B/C-LMP1. (B) 293T cells (6 105) were transfected with 1 Ag of pSG5-B/C/B-LMP1 or pSG5-C/B/C-LMP1 and photographed 36 h later. (C) Cells were transfected with 1 Ag of pSG5 (lane 1), pSG5-LMP1 (lane 2), pSG5-CAO-LMP1 (lane 3), pSG5-B/C/B-LMP1 (lane 4) or pSG5-C/B/C-LMP1 (lane 5). The extent of cell death was determined 36 h after transfection. Whole-cell lysates (15 Ag protein) collected 20 h after transfection were subjected to Western blot analysis with anti-LMP1 antibodies. (D) NF-nB-dependent luciferase activity was assessed as described in Fig. 1. The experiments were repeated three times with essentially similar results.
1076
T. Nitta et al. / Cellular Signalling 16 (2004) 1071–1081
detachment of infected Rat-1 cells, whereas cells expressing CAO-LMP1 remained viable (data not shown). Both types of LMP1 achieved similar levels of NF-nB activation
(Fig. 1C), although previous studies suggested that CAOLMP1 activated NF-nB more efficiently than did B95.8LMP1 [20,25]. These results indicate that CAO-LMP1 is
Fig. 4. Mutation of two amino acids in the transmembrane domain of B95.8-LMP1 diminishes its killing property. (A) The positions of amino acids 85 and 106 in the structure of LMP1 are schematically represented (left). Alignments of partial amino acid sequences (81 – 90 and 101 – 110) of B95.8- and CAO-LMP1 are shown (right). Dots represent identical amino acids. The closed triangles indicate amino acids 85 and 106. (B) 293T cells (6 105) were transfected with 1 Ag of pSG5LMP1, pSG5-B95.8(L), pSG5-B95.8(Y) or pSG5-B95.8(LY), and photographed 36 h later. (C) Cells were transfected with 1 Ag of pSG5 (lane 1), pSG5-LMP1 (lane 2), pSG5-CAO-LMP1 (lane 3), pSG5-B95.8(L) (lane 4), pSG5-B95.8(Y) (lane 5) or pSG5-B95.8(LY) (lane 6). The extent of cell death was determined 36 h after transfection. Whole-cell lysates (15 Ag protein) collected 20 h after transfection were subjected to Western blot analysis with anti-LMP1 antibodies. (D) NFnB-dependent luciferase activity was determined as described in Fig. 1. The experiments were repeated three times, and the results were essentially reproducible.
T. Nitta et al. / Cellular Signalling 16 (2004) 1071–1081
defective in cell death induction despite its potent NF-nB activation. 3.2. The transmembrane domain of LMP1 directs its killing property To map the domain(s) of CAO-LMP1 responsible for its lack of cell death induction, we divided the LMP1 polypeptide into three parts, the N-terminal cytoplasmic domain (aa 1 –26), transmembrane domain (aa 27– 192) and C-terminal cytoplasmic domain (aa 193 –386 or 404), and thereafter generated chimeric constructs between B95.8- and CAOLMP1 (Fig. 2A). We first analyzed the functional consequences of the amino (N-) and carboxy (C-) terminal cytoplasmic domains. C/B/B-LMP1, which has the N-terminal cytoplasmic domain of CAO-LMP1 followed by the transmembrane and C-terminal cytoplasmic domains of B95.8-LMP1, induced cell death in 293T cells as efficiently as B95.8-LMP1 (Fig. 2B and C, lane 3). B/B/C-LMP1, which has the Nterminal and transmembrane domains of B95.8-LMP1 linked to the C-terminus of CAO-LMP1, also induced cell death, although to a slightly lesser degree (lane 4). In contrast, reciprocal chimeras, B/C/C-LMP1 and C/C/B-LMP1 failed to induce cell death (lanes 5 and 6). Western blot analysis revealed that the expression levels of these chimeric proteins were similar (Fig. 2C, bottom panel). Each protein showed different migration probably due to the difference in their molecular weight and net charge of amino acid residues. These chimeras also showed NF-nB activation to similar extents (Fig. 2D). These results indicate that neither the N-terminal or C-terminal cytoplasmic domains of LMP1 are critically involved in cell death induction, and suggest that the transmembrane domain of LMP1 is responsible for its killing property. Indeed, B/C/B-LMP1, which has the transmembrane domain of CAO-LMP1 and the other parts from B95.8-LMP1, lacked the killing property (Fig. 3A, B and C, lane 4). Its reciprocal chimera, C/B/C-LMP1, induced cell death, although it was moderate compared with that by B95.8-LMP1 (lane 5). Their protein expression levels and NF-nB activation were found similar to those of B95.8-LMP1 (Fig. 3C and D). Taken together, these results clearly indicate that the transmembrane domain of LMP1 determines its killing property.
1077
B95.8-LMP1 has Ile85 and Phe106, whereas CAO-LMP1 has the NPC-type residues (Leu85 and Tyr106). This information prompted us to investigate the functional consequences of these residues in the cell death induction. We introduced amino acid substitutions of Ile85Leu, Phe106Tyr or both in B95.8-LMP1, generating B95.8(L)LMP1, B95.8(Y)-LMP1 or B95.8(LY)-LMP1, respectively, and examined their ability to induce cell death in 293T cells. B95.8(L)-LMP1 behaved in a manner similar to B95.8-LMP1 with regard to cell death induction (Fig. 4B and C, lane 4). A slight reduction of cell death induction was observed in cells transfected with B95.8(Y)-LMP1 (lane 5). B95.8(LY)-LMP1, which has two amino acid substitutions, showed a remarkable reduction of the killing property (lane 6). The protein expression levels and extents of NF-nB activation were not significantly altered by these
3.3. Mutation of two amino acids in the transmembrane domain of B95.8-LMP1 diminishes its killing property Fourteen amino acids in the transmembrane domain (aa 27 –192) of CAO-LMP1 are different from those of B95.8LMP1 [20]. A recent report on the sequence of LMP1 genes from NPC patients or from healthy carriers [34] revealed that the 85th and 106th amino acids were Leu and Tyr, respectively, in nearly all the tested LMP1 clones derived from NPC patients, whereas approximately a half of healthy people had different residues in these positions that were in most cases Ile85 and Phe106. Interestingly,
Fig. 5. Transmembrane domain of LMP1 directs activation of caspase-3. (A) 293T cells (6 105) were transfected with 1 Ag of pSG5 (lane 1), pSG5-LMP1 (lane 2), pSG5-CAO-LMP1 (lane 3), pSG5-B95.8(LY) (lane 4), pSG5-B/C/B-LMP1 (lane 5) or pSG5-C/B/C-LMP1 (lane 6). Thirty hours after transfection, the caspase-3 activity that cleaves DEVD-AFC was determined as described in the Materials and methods and expressed relative to that of the vector-transfected cells. The results are representative of three independent experiments. (B) Whole-cell lysates collected 30 h after transfection were subjected to Western blot analysis with anti-PARP antibodies. The arrow and arrowhead indicate intact 116-kDa and processed 85-kDa PARP proteins, respectively.
1078
T. Nitta et al. / Cellular Signalling 16 (2004) 1071–1081
Fig. 6. Transmembrane domain of B95.9-LMP1 induces cell death and caspase activation in the presence of high NF-nB activity. (A) Schematic representation of the structure of LMP1 chimeras fused with EGFP. (B) 293T cells (6 105) were transfected with the plasmids (1 Ag each) as follows; lane 1: pEGFP-N1 and pcDNA3; lane 2: pEGFP-N1 and pRc-CMV-relA; lane 3: pEGFP-B95.8 and pcDNA3; lane 4: pEGFP-B95.8 and pRc-CMV-relA; lane 5: pEGFP-B95.8(LY) and pcDNA3; lane 6: pEGFP-B95.8(LY) and pRc-CMV-relA; lane 7: pEGFP-CAO and pcDNA3; lane 8: pEGFP-CAO and pRc-CMV-relA. The extent of cell death was determined 36 h after transfection (upper panel). NF-nB-dependent luciferase activity was assessed as described in Fig. 1 (middle panel). Values of lanes 1, 3, 5 and 7 are indicated. Whole-cell lysates (15 Ag protein) collected 20 h after transfection were subjected to Western blot analysis with anti-GFP or anti-RelA antibodies (bottom panels). The experiments were repeated three times with essentially similar results. (C) Caspase activation was evaluated by the proteolytic cleavage of PARP. Whole-cell lysates collected 36 h after transfection were subjected to Western blot analysis with anti-PARP antibodies. The arrow and arrowhead indicate intact 116-kDa and processed 85-kDa PARP proteins, respectively.
T. Nitta et al. / Cellular Signalling 16 (2004) 1071–1081
mutations (Fig. 4C and D). Amino acid substitutions of Ile for Leu at aa 85, Phe for Tyr at aa 106, or both in CAOLMP1, however, did not restore the ability to induce cell death (data not shown). Thus, the two amino acid residues, Ile85 and Phe106 are critical, but not solely responsible for the different killing properties of B95.8- and CAO-LMP1. 3.4. Caspase activation depends on the transmembrane domain of LMP1 Our previous report demonstrated that expression of B95.8-LMP1 in 293T cells induced activation of caspase3 [17]. In contrast, expression of CAO-LMP1 failed to increase the caspase-3-like activity that cleaves DEVDAFC, a fluorescent substrate of caspase-3 subfamily proteases (Fig. 5A). Impaired cleavage of PARP, a wellcharacterized substrate for caspase-3, confirmed that CAO-LMP1 could not activate caspase-3 in vivo (Fig. 5B). B95.8(LY)-LMP1 showed a remarkable reduction in the caspase-3 activation and PARP cleavage compared with B95.8-LMP1. B/C/B-LMP1 lost the ability to activate caspase-3, indicating that the transmembrane domain of CAO-LMP1 is responsible for its inability of caspase activation. Conversely, the results that C/B/C-LMP1 restored activation of caspase-3 and cleavage of PARP indicate a critical role for the transmembrane domain of B95.8LMP1 in caspase-3 activation. 3.5. The transmembrane domain of LMP1 alone induces cell death under elevated NF-jB activity Our previous report demonstrated that NF-nB activation was required, but not sufficient for the LMP1-induced cell death [17]. While NF-nB is activated via the C-terminal cytoplasmic domain of LMP1 [1], it remained unclear which domain(s) controls such an additional death signal(s). The result that modifications of the transmembrane domain of LMP1 greatly influences its killing property, but not activation of NF-nB suggests that the transmembrane domain controls a yet undefined death signal(s) independently of its C-terminus. To verify this hypothesis, the transmembrane domain alone should be expressed and tested for cell death induction. However, the level of expression of the truncated B95.8-LMP1 lacking the C-terminal cytoplasmic domain (aa 193 – 386) was much lower compared to that of the fulllength LMP1 (data not shown), suggesting that the Cterminal domain is essential for stable expression of this protein. Besides, the N-terminal domain was reported to be indispensable for proper orientation of the transmembrane domain [35]. Thus, we replaced the C-terminal cytoplasmic domain of B95.8-LMP1 with an unrelated protein, EGFP, generating a chimeric protein B95.8/EGFP (Fig. 6A). This chimeric protein was expressed at a level comparable to B95.8-LMP1. B95.8(LY)/EGFP and CAO/EGFP were generated and detected in the same manner (Fig. 6A). Several studies have previously shown a characteristic localization
1079
of LMP1 in patches at the plasma membrane [25]. Examination by fluorescent microscopy of transfected cells revealed that these LMP1/EGFP chimeric proteins exhibited similar patterns of membrane localization characterized by patches at the plasma membrane (data not shown). Coexpression of B95.8/EGFP with RelA, an active subunit of NF-nB, extensively killed the cells, whereas B95.8/EGFP alone caused only a limited cell death (Fig. 6B, lanes 3 and 4). Transactivation by RelA was partially reduced by coexpression of B95.8/EGFP (Fig. 6B, lane 4), consistent with a previous report that the N-terminal and transmembrane domains had inhibitory effects on NF-nB-dependent transactivation [36]. Co-expression of B95.8/EGFP with RelA also induced PARP cleavage (Fig. 6C, lane 4). These killing activities were obviously dependent on the transmembrane domain, since the induction of cell death and caspase activation were markedly reduced in B95.8(LY)/EGFP (lanes 5 and 6), or lost in CAO/EGFP (lanes 7 and 8). These results indicate that the transmembrane domain of LMP1 triggers death program under high NF-nB activity.
4. Discussion In this paper, we investigated the structural requirements for cell death induction by LMP1, exploring the distinct killing properties of B95.8-LMP1 and CAOLMP1. The variety in the LMP1 structure may potentially contribute to the development of EBV-related malignancies, and CAO-LMP1 has been studied as a representative LMP1 variant derived from NPC [37 –39]. Previous studies demonstrated functional differences between CAOLMP1 and the B lymphocyte-derived prototype B95.8LMP1; Rat-1 fibroblasts transfected with CAO-LMP1 grew in soft agar more efficiently than those with B95.8-LMP1 [23]; CAO-LMP1 showed more potent activation of NF-nB, JNK and STAT, compared with B95.8LMP1 [20,25,40]; CAO-LMP1 failed to induce growth inhibition, expression of certain cell surface molecules and secretion of cytokines in human epithelial cells that had been reported for B95.8-LMP1 [25]. Transient expression of B95.8-LMP1 induced cell death and caspase-3 activation in an NF-nB-dependent manner [17]. In contrast, CAO-LMP1 did not induce cell death or caspase-3 activation, while it activated NF-nB to a level similar to B95.8-LMP1. Our result that CAO-LMP1 shows NF-nB activation similar to, but not significantly greater than B95.8-LMP1, could partly be explained by the differences in the internal control and time points of the assay, or by different cellular backgrounds. Since activation of caspase3 and induction of cell death by B95.8-LMP1 required not only NF-nB activation but also some additional event(s) [17], the difference between these two types of LMP1 in cell killing was expected to result from such unknown function(s).
1080
T. Nitta et al. / Cellular Signalling 16 (2004) 1071–1081
Prior studies indicated that the transmembrane domain, but not the C-terminal activation domain of LMP1 was responsible for the functional differences between B95.8 and CAO LMP1 [20,28,37]. Our studies with B95.8/CAO chimeras show compelling evidence that the transmembrane domain is responsible for the different killing property between B95.8-LMP1 and CAO-LMP1. The substitution of the CAO-LMP1 transmembrane domain for that of B95.8-LMP1 almost completely abolished cell death induction, and the reciprocal manipulation, placing the transmembrane domain of B95.8 in the context of CAO-LMP1, conferred the ability to induce cell death on CAO-LMP1, indicating that the transmembrane domain is a critical determinant for cell death induction by LMP1. Substitution of two amino acids in the transmembrane domain, Ile85Leu and Phe106Tyr, diminished cell death induction by B95.8LMP1. These two residues appear to function cooperatively, since Ile85Leu or Phe106Tyr alone had no or only partial effect on cell death induction. These two residues are located in the third or fourth membrane-spanning segments, respectively [2,40], suggesting that they contribute to the conformation or membrane topology of LMP1 and that the amino acid substitutions may affect the interaction of the membrane-spanning and/or loop segments with cellular factor(s). However, changing these two amino acids on CAO-LMP1 to those of B95.8-LMP1 was not sufficient to confer the killing property on CAO-LMP1. This could partly be explained by the fact that CAO-LMP1 has additionally 12 amino acid substitutions in the transmembrane domain compared with B95.8-LMP1. Our results with LMP1/EGFP chimeras indicate that the transmembrane domains of LMP1 can direct death signal(s) under high NF-nB activity without its C-terminal cytoplasmic domain. In fact, B95.8/EGFP by itself induced weak cell death without any detectable caspase activation, but it induced marked cell death and caspase activation when coexpressed with RelA, both of which were significantly diminished by the two amino acid substitutions in the transmembrane domain. The biological functions of the transmembrane domain of LMP1 have not been fully elucidated. Roles of this domain so far characterized include the induction of self-association on the plasma membrane and localization in lipid rafts that contribute to its efficient signaling [41]. It was reported that the N-terminus and transmembrane domains of B95.8-LMP1 could inhibit cell proliferation and cellular protein synthesis [13,36]. These effects are explained by a very recent finding that the transmembrane domain of LMP1 can induce phosphorylation of eIF2a [42]. The transmembrane domain of B95.8-LMP1 alone was also shown to induce reorganization of the actin cytoskeleton, depending on Cdc42, a Rho-like GTPase [43]. Blake et al. [20] reported that the transmembrane domain of CAO-LMP1 contributes to its half-life longer than that of B95.8-LMP1, while CAO-LMP1 exhibited membrane localization similar to B95.8-LMP1 [25] (data not shown). Our present report has demonstrated for the first time that the
transmembrane domain of LMP1 governs caspase activation and cell death induction under high NF-nB activity. The transmembrane domain of CAO-LMP1 may fail to evoke death signal(s) or may activate survival programs that cancel cell death induction by B95.8-LMP1. Further studies on the functional difference between B95.8-LMP1 and CAO-LMP1 are under way to fully elucidate the molecular mechanisms of this NF-nB-dependent cell death induction. Finally, our results suggest possible contribution of CAO-LMP1 to the pathogenesis of NPC. B95.8-LMP1 requires NF-nB activity not only for cell death induction but also for cell transformation [8]. Previous observations that CAO-LMP1 caused cell transformation more efficiently than B95.8-LMP1 [21 –23] may partly be explained by its defect in the cytotoxic property and unaltered ability to activate NF-nB. In fact, the CAO-type variants of LMP1 are frequently present not only in NPC, but also in other EBVassociated lymphoproliferative disorders [37,39,40]. Importantly, amino acid changes similar to those of CAO-LMP1, in particular the Ile86Leu and Phe106Tyr substitutions, have been widely observed in other LMP1 variants derived from NPC cell lines [37,38], NPC biopsies [38] and NPC patients [34,39]. Thus, it would be interesting to examine if LMP1 variants isolated from a wide range of EBV-associated tumor cells also carry the Ile86Leu and Phe106Tyr substitutions, and are defective in cell death induction. Understanding the molecular mechanisms of B95.8-LMP1induced cytotoxicity and contribution of the variant LMP1 to oncogenesis will help to establish novel diagnostic or therapeutic approaches to EBV-related diseases.
Acknowledgements We would thank Dr. Rowe (University of Wales College of Medicine) for LMP1 expression vectors and antibodies, Dr. Memet (Institut Pasteur Paris) for EF1-lacZ and Dr. Baltimore (California Institute of Technology) for the relA cDNA. We also thank members of the Department of Molecular Virology, Tokyo Medical and Dental University, for helpful discussion and comments. This work was partly supported by Japan Human Science Foundation (grant K1040 to S.Y.).
References [1] E.D. Cahir McFarland, K.M. Izumi, G. Mosialos, Oncogene 18 (1999) 6959 – 6964. [2] S. Fennewald, V. van Santen, E. Kieff, J. Virol. 51 (1984) 411 – 419. [3] A.G. Eliopoulos, S.M. Blake, J.E. Floettmann, M. Rowe, L.S. Young, J. Virol. 73 (1999) 1023 – 1035. [4] A.G. Eliopoulos, L.S. Young, Oncogene 16 (1998) 1731 – 1742. [5] A. Kieser, E. Kilger, O. Gires, M. Ueffing, W. Kolch, W. Hammerschmidt, EMBO J. 16 (1997) 6478 – 6485. [6] O. Gires, F. Kohlhuber, E. Kilger, M. Baumann, A. Kieser, C. Kaiser,
T. Nitta et al. / Cellular Signalling 16 (2004) 1071–1081
[7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23]
[24] [25]
A. Hall, B. Scheffer, M. Ueffing, W. Hammerschmidt, EMBO J. 18 (1999) 3064 – 3073. A.G. Eliopoulos, N.J. Gallagher, S.M. Blake, C.W. Dawson, L.S. Young, J. Biol. Chem. 274 (1999) 16085 – 16096. Z. He, B. Xin, X. Yang, C. Chan, L. Cao, Cancer Res. 60 (2000) 1845 – 1848. F. Khanim, Q.Y. Yao, G. Niedobitek, S. Sihota, A.B. Rickinson, L.S. Young, Blood 88 (1996) 3491 – 3501. R.H. Edwards, F. Seillier-Moiseiwitsch, N. Raab-Traub, Virology 261 (1999) 79 – 95. D.M. Walling, N. Shebib, S.C. Weaver, C.M. Nichols, C.M. Flaitz, J. Webster-Cyriaque, J. Infect. Dis. 179 (1999) 763 – 774. J.E. Floettmann, K. Ward, A.B. Rickinson, M. Rowe, Virology 223 (1996) 29 – 40. A. Kaykas, B. Sugden, Oncogene 19 (2000) 1400 – 1410. W. Hammerschmidt, B. Sugden, V.R. Baichwal, J. Virol. 63 (1989) 2469 – 2475. J.J. Lu, J.Y. Chen, T.Y. Hsu, W.C. Yu, I.J. Su, C.S. Yang, J. Gen. Virol. 77 (1996) 1883 – 1892. L.F. Sheu, A. Chen, Y.H. Wei, K.C. Ho, J.Y. Cheng, C.L. Meng, W.H. Lee, Am. J. Pathol. 152 (1998) 63 – 74. T. Nitta, A. Chiba, A. Yamashita, M. Rowe, A. Israe¨l, M. Reth, N. Yamamoto, S. Yamaoka, Cell Signal. 15 (2003) 423 – 433. C. Borner, L. Monney, Cell Death Differ. 6 (1999) 497 – 507. L.F. Hu, E.R. Zabarovsky, F. Chen, S.L. Cao, I. Ernberg, G. Klein, G. Winberg, J. Gen. Virol. 72 (1991) 2399 – 2409. S.M. Blake, A.G. Eliopoulos, C.W. Dawson, L.S. Young, Virology 282 (2001) 278 – 287. L.F. Hu, F. Chen, X. Zheng, I. Ernberg, S.L. Cao, B. Christensson, G. Klein, G. Winberg, Oncogene 8 (1993) 1575 – 1583. X. Zheng, F. Yuan, L. Hu, F. Chen, G. Klein, B. Christensson, Int. J. Cancer 57 (1994) 747 – 753. A.M. Mehl, N. Fischer, M. Rowe, F. Hartmann, H. Daus, L. Trumper, M. Pfreundschuh, N. Mu¨ller-Lantzsch, F.A. Gra¨sser, Int. J. Cancer 76 (1998) 194 – 200. P. Trivedi, L.F. Hu, F. Chen, B. Christensson, M.G. Masucci, G. Klein, G. Winberg, Eur. J. Cancer 30A (1994) 84 – 88. C.W. Dawson, A.G. Eliopoulos, S.M. Blake, R. Barker, L.S. Young, Virology 272 (2000) 204 – 217.
1081
[26] S. Morita, T. Kojima, T. Kitamura, Gene Ther. 7 (2000) 1063 – 1066. [27] D. Liebowitz, R. Kopan, E. Fuchs, J. Sample, E. Kieff, Mol. Cell. Biol. 7 (1987) 2299 – 2308. [28] R.J. Johnson, M. Stack, S.A. Hazlewood, M. Jones, C.G. Blackmore, L.F. Hu, M. Rowe, J. Virol. 72 (1998) 4038 – 4048. [29] G.P. Nolan, S. Ghosh, H.C. Liou, P. Tempst, D. Baltimore, Cell 64 (1991) 961 – 969. [30] S. Yamaoka, G. Courtois, C. Bessia, S.T. Whiteside, R. Weil, F. Agou, H.E. Kirk, R.J. Kay, A. Israe¨l, Cell 93 (1998) 1231 – 1240. [31] T. Saitoh, H. Nakano, N. Yamamoto, S. Yamaoka, FEBS Lett. 532 (2002) 45 – 51. [32] S.N. Ho, H.D. Hunt, R.M. Horton, J.K. Pullen, L.R. Pease, Gene 77 (1989) 51 – 59. [33] M. Rowe, H.S. Evans, L.S. Young, K. Hennessy, E. Kieff, A.B. Rickinson, J. Gen. Virol. 68 (1987) 1575 – 1586. [34] P. Hahn, E. Novikova, L. Scherback, C. Janik, O. Pavlish, V. Arkhipov, J. Nicholls, N. Mu¨ller-Lantzsch, V. Gurtsevitch, F.A. Gra¨sser, Int. J. Cancer 91 (2001) 815 – 821. [35] W.F. Coffin III, K.D. Erickson, M. Hoedt-Miller, J.M. Martin, Oncogene 20 (2001) 5313 – 5330. [36] M.L. Sandberg, A. Kaykas, B. Sugden, J. Virol. 74 (2000) 9755 – 9761. [37] W.E. Miller, J.L. Cheshire, A.S. Baldwin Jr., N. Raab-Traub, Oncogene 16 (1998) 1869 – 1877. [38] S.T. Cheung, S.F. Leung, K.W. Lo, K.W. Chiu, J.S. Tam, T.F. Fok, P.J. Johnson, J.C. Lee, D.P. Huang, Int. J. Cancer 76 (1998) 399 – 406. [39] X.S. Zhang, K.H. Song, H.Q. Mai, W.H. Jia, B.J. Feng, J.C. Xia, R.H. Zhang, L.X. Huang, X.J. Yu, Q.S. Feng, P. Huang, J.J. Chen, Y.X. Zeng, Cancer Lett. 176 (2002) 65 – 73. [40] C.A. Fielding, K. Sandvej, A. Mehl, P. Brennan, M. Jones, M. Rowe, J. Virol. 75 (2001) 9129 – 9141. [41] A. Kaykas, K. Worringer, B. Sugden, EMBO J. 20 (2001) 2641 – 2654. [42] N. Lam, M.L. Sandberg, B. Sugden, J. Virol. 78 (2004) 1657 – 1664. [43] A. Puls, A.G. Eliopoulos, C.D. Nobes, T. Bridges, L.S. Young, A. Hall, J. Cell. Sci. 112 (1999) 2983 – 2992.
Lack of cytotoxic property in a variant of Epstein–Barr virus latent membrane protein-1 isolated from nasopharyngeal carcinoma Takeshi Nitta1, Ayako Chiba, Naoki Yamamoto, Shoji Yamaoka * Department of Molecular Virology, Graduate School, Tokyo Medical and Dental University, 1-5-45, Yushima, Bunkyo, Tokyo 113-8519, Japan Received 29 February 2004; accepted 1 March 2004 Available online 21 April 2004
Abstract Latent membrane protein 1 (LMP1) encoded by Epstein – Barr virus (EBV) is a membrane protein that activates multiple signaling pathways and transcription factors, including NF-nB. Our recent report demonstrated that expression of LMP1 induced programmed cell death in an NFnB-dependent manner. In this study, we demonstrate that a variant CAO-LMP1 derived from EBV-infected nasopharyngeal carcinoma (NPC) does not induce cell death unlike the prototype B95.8-LMP1, although both types of LMP1 show NF-nB activation to a similar extent. Studies with chimeric or mutated proteins identified two amino acids in the transmembrane domain, which are commonly substituted in NPC-derived LMP1 variants, being critical for cell death induction by B95.8-LMP1. Furthermore, we show that the B95.8 transmembrane domain cooperates with NF-nB to trigger cell death program. Thus, our results reveal a particular feature of the transmembrane domain of tumor-derived CAO-LMP1 and suggest its possible contribution to the pathogenesis of NPC. D 2004 Elsevier Inc. All rights reserved. Keywords: Epstein – Barr virus; LMP1; Nasopharyngeal carcinoma; NF-nB
1. Introduction Latent membrane protein 1 (LMP1) is a membrane protein encoded by Epstein –Barr virus (EBV) [1]. LMP1 is a type of oncogenic protein that is essential for EBV-mediated transformation of B lymphocytes and can induce malignant transformation of rodent fibroblast cell lines. LMP1 consists of a short cytoplasmic N-terminal domain, transmembrane domain containing six membrane-spanning segments and five intra/extracellular loops, and long cytoplasmic C-terminal domain [2]. The transmembrane domain of LMP1 is necessary for its self-multimerization on the plasma membrane and enables the C-terminal cytoplasmic domain to interact with specific proteins that transduce cellular signals [1]. In this manner, LMP1 activates multiple cellular signalAbbreviations: LMP1, latent membrane protein 1; EBV, Epstein – Barr virus; JNK, c-Jun N-terminal kinase; STAT, signal transducer and activator of transcription; NPC, nasopharyngeal carcinoma; EGFP, enhanced green fluorescent protein; aa, amino acid; PARP, poly ADP-ribose polymerase; DEVD-AFC, Asp-Glu-Val-Asp-7-amino-4-trifluoromethyl coumarin. * Corresponding author. Tel.: +81-3-5803-5178; fax: +81-3-5803-0124. E-mail address: [email protected] (S. Yamaoka). 1 Present address: Division of Experimental Immunology, Institute for Genome Research, University of Tokushima, 3-18-15 Kuramoto, Tokushima 770-8503, Japan. 0898-6568/$ - see front matter D 2004 Elsevier Inc. All rights reserved. doi:10.1016/j.cellsig.2004.03.001
ing cascades leading to activation of NF-nB [1], c-Jun Nterminal kinase (JNK) [3– 5], signal transducer and activator of transcription (STAT) [6] and p38 mitogen-activated protein kinase [7]. Among them, activation of NF-nB is known to be required for the cell-transforming activity of LMP1 [8]. LMP1 variants have been obtained from geographically distinct EBV isolates or from different types of EBV-associated tumors. However, correlation has not been well established among their sequences, biological functions and possible contribution to EBV-associated malignancies [9– 11]. B95.8-LMP1, an LMP1 clone originally derived from a B lymphocyte, has been extensively studied as a prototype, resulting in elucidation of its signaling mechanisms and biological functions. Regardless of its transforming activity, B95.8-LMP1 was reported to inhibit cell proliferation [12,13] and colony formation [14] or to induce apoptosis [15,16] in several types of cell lines. Most recently, we demonstrated that expression of B95.8-LMP1 induced programmed cell death, which was dependent on activation of NF-nB [17]. Cell death induction by B95.8-LMP1 was prevented by specific inhibition of NF-nB activity and restored by reactivation of NF-nB. B95.8-LMP1 also induced activation of caspase-3, a cysteine protease involved in the execution step of apoptosis [18], in an NF-nB-dependent manner. This
1072
T. Nitta et al. / Cellular Signalling 16 (2004) 1071–1081
LMP1-induced cell death was also expected to be a model to study molecular mechanisms of NF-nB-dependent cell death induction, although its pathological consequences in EBVrelated diseases remained elusive. A clue to this question was found in the CAO-LMP1 variant, which was originally isolated from EBV-associated nasopharyngeal carcinoma (NPC) [19] and had amino acid substitutions most frequently in the N-terminus and transmembrane domains, and amino acid insertions or deletions in the cytoplasmic C-terminus [20]. This LMP1 variant has been reported to show more tumorigenic [21 – 23] and less immunogenic properties [24], enhanced signaling activities [25], and impaired induction of certain phenotypic markers [25]. Here, we demonstrate that CAO-LMP1, unlike B95.8-LMP1, does not induce cell death and that two amino acids in the transmembrane domain that are commonly substituted in NPCderived LMP1s are critical for cell death induction. Our results indicate that the transmembrane domain of LMP1 governs its killing property, and suggest possible contribution of the two amino acid substitutions to the pathogenesis of NPC.
B/B-LMP1: the N-terminus of CAO-LMP1 linked to the transmembrane and C-terminal domains of B95.8-LMP1; B/ B/C-LMP1: the N-terminal and transmembrane domains of B95.8-LMP1 linked to the C-terminus of CAO-LMP1; B/C/ C-LMP1: the N-terminus of B95.8-LMP1 linked to the transmembrane and C-terminal domains of CAO-LMP1; C/ C/B-LMP1: the N-terminal and transmembrane domains of CAO-LMP1 linked to the C-terminus of B95.8-LMP1; B/C/ B-LMP1: the transmembrane domain of CAO-LMP1 placed between the N-terminal and C-terminal domains of B95.8LMP1; C/B/C-LMP1: the transmembrane domain of B95.8LMP1 placed between the N-terminal and C-terminal domains of CAO-LMP1. These chimeric constructs were cloned into the pSG5 vector. To generate point mutations, Ile85Leu and Phe106Tyr in B95.8-LMP1, we employed the overlap PCR technique [32], using oligonucleotide primers
5V-GGAGCCCTTTGTCTACTCCTACTGATG-3V 5V-GGACAGGCATTGTACCTTGGAATTGTG-3V
2. Materials and methods
and their complimentary oligonucleotides. Positions mutated are underlined. For the reciprocal mutations Leu85Ile and Tyr106Phe in CAO-LMP1, primers
2.1. Cell culture
293T, Rat-1 and Plat-E [26] cells were maintained in DMEM supplemented with 10% fetal calf serum, 100 units/ ml of penicillin G and 100 Ag/ml of streptomycin. 2.2. Plasmids LMP1 expression vectors pSG5-LMP1 or pSG5-CAOLMP1, in which cDNAs of the prototype LMP1 (B95.8LMP1) or CAO-LMP1 were located downstream of the SV40 promoter, were previously described (kindly provided by Dr. Rowe, University of Wales College of Medicine) [27,28]. Plasmids pRc-CMV-relA [29], Ign-ConAluc and EF1-lacZ (a kind gift of Dr. Memet, Institut Pasteur) [30] were previously described. Plasmid pEGFP-N1 encoding the enhanced green fluorescent protein (EGFP) was purchased from CLONTECH. For retroviral expression, we subcloned B95.8- or CAO-LMP1 cDNAs in a retroviral vector pMRX-IRESEGFP [31]. For construction of B95.8/CAO chimeric proteins, an artificial KpnI site was first introduced at the position corresponding to the amino acid (aa) 192 and 193 of both types of LMP1, to split them into fragments encoding the Nterminal and transmembrane domains (aa 1 – 192) or Cterminal cytoplasmic domain (aa 193 – 386 or 404). The introduction of the KpnI site led to two amino acid substitutions of Gly192 and Thr193 for Ser192 and Asp193 in B95.8LMP1 or for Thr192 and Asp193 in CAO-LMP1. The Nterminus (aa 1 – 26) and transmembrane (aa 27 – 192) domains were separated at the StuI site. Finally, these DNA fragments derived from B95.8- or CAO-LMP1 were appropriately ligated to generate chimeric constructs as follows: C/
5V-GGAGGCCTTGGTATACTCCTACTGATG-3V 5V-GGACAGGCATTGTTCCTTGGAATTGTG-3V
and their complimentary oligonucleotides were used. Mutated DNA fragments were cloned into the pSG5 vector, generating the following plasmids; pSG5-B95.8(L) encoding B95.8-LMP1 Ile85Leu; pSG5-B95.8(Y) encoding B95.8-LMP1 Phe106Tyr; pSG5-B95.8(LY) encoding B95.8-LMP1 Ile85Leu Phe106Tyr; pSG5-CAO(I) encoding CAO-LMP1 Leu85Ile; pSG5-CAO(F) encoding CAOLMP1 Tyr106Phe; pSG5-CAO(IF) encoding CAO-LMP1 Leu85Ile Tyr106Phe. To generate LMP1/EGFP chimeric constructs, an artificial EcoRI site was introduced at the position corresponding to the aa 192– 194. DNA fragments encoding the N-terminal and transmembrane domains of different types of LMP1 were inserted into the EcoRI site of the pEGFP-N1 vector, generating pEGFP-B95.8, pEGFP-B95.8(LY) and pEGFP-CAO. The sequences of all these constructs were verified by DNA sequencing. Full construction details are available upon request. 2.3. Transient transfection and luciferase assay Approximately 6 105 293T cells were transfected by the calcium phosphate coprecipitation method. Flow cytometry analysis of cells transfected with EGFP revealed that transfection efficiencies ranged from 60% to 80%. For measurement of the luciferase activity, 0.1 Ag of IgnConAluc and EF1-lacZ was included in each transfection. Twenty hours after transfection, luciferase activity was determined as previously described and normalized on the basis of h-galactosidase activity [30].
T. Nitta et al. / Cellular Signalling 16 (2004) 1071–1081
1073
2.4. Retroviral infection
2.6. Antibodies and Western blot analysis
For production of retroviruses, Plat-E cells were transfected with pMRX-IRES-EGFP, pMRX-B95.8-IRES-EGFP or pMRX-CAO-IRES-EGFP. Forty-eight hours after transfection, culture supernatants containing retroviruses were collected and stored at 80 jC. Rat-1 cells were infected with these viruses in the presence of polybrene (10 Ag/ml), and cells expressing EGFP were observed under a fluorescent microscope.
Anti-LMP1 antibodies (CS.1 –4) were described previously (kindly provided by Dr. Rowe, University of Wales College of Medicine) [33]. Anti-poly ADP-ribose polymerase (PARP) (H-250), anti-GFP (FL) and anti-RelA antibodies were purchased from Santa Cruz. Whole-cell lysate was prepared by lysing cells with the RIPA buffer (20 mM Tris –HCl [pH 8], 137 mM NaCl, 1 mM MgCl2, 1 mM CaCl2, 10% glycerol, 1% Nonidet P-40, 0.5% deoxycholate, 0.1% SDS and 0.1 mM PMSF) and analyzed as described previously [30].
2.5. Cell death assay 2.7. Enzyme assay for caspase activity Cell death of 293T cells was quantified by counting the number of detached and adherent cells as previously described [17]. After collecting cells detached in medium, adherent cells on the culture plate were collected. The number of cells from each fraction was counted, and the ratio of dead cells was determined as (number of detached cells)/(total cell count).
Activity of caspase-3 family proteases was assessed by using the CPP32/Caspase-3 Fluorometric Protease Assay Kit (MBL) with the fluorescent substrate Asp-Glu-Val-Asp7-amino-4-trifluoromethyl coumarin (DEVD-AFC). Caspase activity was determined as the fluorescence intensity normalized by protein amount.
Fig. 1. B95.8-LMP1, but not CAO-LMP1 induces cell death. (A) 293T cells (6 105) were transfected with 1 Ag of the pSG5 (vector), pSG5-LMP1 or pSG5CAO-LMP1, and photographed 36 h later under a phase-contrast microscope. (B) The extent of cell death determined 36 h after transfection is shown as the ratio of detached cells to total cells, because cells, once detached, did not adhere to the plastic again nor survive in suspension. Whole-cell lysates (15 Ag protein) collected 20 h after transfection, a few hours before the onset of cell death, were subjected to Western blot analysis with anti-LMP1 antibodies. (C) NFnB-dependent transcriptional activity was assessed 20 h after transfection and expressed in fold induction relative to that of vector-transfected cells. Each transfection contained 0.1 Ag of Ign-ConAluc and EF1-lacZ, respectively. The experiments were repeated three times with essentially similar results.
1074
T. Nitta et al. / Cellular Signalling 16 (2004) 1071–1081
Fig. 2. The extreme N-terminal and C-terminal cytoplasmic domains of LMP1 are not involved in cell death induction. (A) Schematic representation of the structure of B95.8-LMP1, CAO-LMP1 and their chimeric derivatives. Solid black lines represent B95.8-LMP1 sequence and open lines denote CAO-LMP1 sequence. (B) 293T cells (6 105) were transfected with 1 Ag of pSG5-C/B/B-LMP1, pSG5-B/B/C-LMP1, pSG5-B/C/C-LMP1 or pSG5-C/C/B-LMP1, and photographed 36 h later. B and C represent B95.8- and CAO-derived LMP1 sequences, respectively. (C) Cells were transfected with 1 Ag of pSG5 (lane 1), pSG5-LMP1 (lane 2), pSG5-C/B/B-LMP1 (lane 3), pSG5-B/B/C-LMP1 (lane 4), pSG5-B/C/C-LMP1 (lane 5) or pSG5-C/C/B-LMP1 (lane 6). The extent of cell death was determined 36 h after transfection as described in Fig. 1. Whole-cell lysates (15 Ag protein) collected 20 h after transfection were subjected to Western blot analysis with anti-LMP1 antibodies. (D) NF-nB-dependent luciferase activity was assessed as described in Fig. 1. The experiments were repeated three times with essentially similar results.
T. Nitta et al. / Cellular Signalling 16 (2004) 1071–1081
3. Results 3.1. CAO-LMP1 does not induce cell death As reported previously, transient expression of B95.8LMP1 in 293T cells induced a remarkable cell death characterized by degeneration and detachment of the cells in an NF-nB-dependent manner [17]. The extent of cell death was quantified by determining the ratio of detached
1075
and adherent cells 36 h after transfection. In contrast, transient expression of an NPC-derived LMP1 variant, CAO-LMP1 did not induce degeneration or detachment of the transfected cells, while its expression level was comparable to that of B95.8-LMP1 (Fig. 1A and B). To confirm the observations obtained in transient transfection of 293T cells, B95.8-LMP1 or CAO-LMP1 was expressed through retroviral infection in a rat fibroblast cell line Rat-1. Expression of B95.8-LMP1 caused severe degeneration and
Fig. 3. The transmembrane domain of B95.8-LMP1 is responsible for induction of cell death. (A) Schematic representation of the structure of B/C/B-LMP1 and C/B/C-LMP1. (B) 293T cells (6 105) were transfected with 1 Ag of pSG5-B/C/B-LMP1 or pSG5-C/B/C-LMP1 and photographed 36 h later. (C) Cells were transfected with 1 Ag of pSG5 (lane 1), pSG5-LMP1 (lane 2), pSG5-CAO-LMP1 (lane 3), pSG5-B/C/B-LMP1 (lane 4) or pSG5-C/B/C-LMP1 (lane 5). The extent of cell death was determined 36 h after transfection. Whole-cell lysates (15 Ag protein) collected 20 h after transfection were subjected to Western blot analysis with anti-LMP1 antibodies. (D) NF-nB-dependent luciferase activity was assessed as described in Fig. 1. The experiments were repeated three times with essentially similar results.
1076
T. Nitta et al. / Cellular Signalling 16 (2004) 1071–1081
detachment of infected Rat-1 cells, whereas cells expressing CAO-LMP1 remained viable (data not shown). Both types of LMP1 achieved similar levels of NF-nB activation
(Fig. 1C), although previous studies suggested that CAOLMP1 activated NF-nB more efficiently than did B95.8LMP1 [20,25]. These results indicate that CAO-LMP1 is
Fig. 4. Mutation of two amino acids in the transmembrane domain of B95.8-LMP1 diminishes its killing property. (A) The positions of amino acids 85 and 106 in the structure of LMP1 are schematically represented (left). Alignments of partial amino acid sequences (81 – 90 and 101 – 110) of B95.8- and CAO-LMP1 are shown (right). Dots represent identical amino acids. The closed triangles indicate amino acids 85 and 106. (B) 293T cells (6 105) were transfected with 1 Ag of pSG5LMP1, pSG5-B95.8(L), pSG5-B95.8(Y) or pSG5-B95.8(LY), and photographed 36 h later. (C) Cells were transfected with 1 Ag of pSG5 (lane 1), pSG5-LMP1 (lane 2), pSG5-CAO-LMP1 (lane 3), pSG5-B95.8(L) (lane 4), pSG5-B95.8(Y) (lane 5) or pSG5-B95.8(LY) (lane 6). The extent of cell death was determined 36 h after transfection. Whole-cell lysates (15 Ag protein) collected 20 h after transfection were subjected to Western blot analysis with anti-LMP1 antibodies. (D) NFnB-dependent luciferase activity was determined as described in Fig. 1. The experiments were repeated three times, and the results were essentially reproducible.
T. Nitta et al. / Cellular Signalling 16 (2004) 1071–1081
defective in cell death induction despite its potent NF-nB activation. 3.2. The transmembrane domain of LMP1 directs its killing property To map the domain(s) of CAO-LMP1 responsible for its lack of cell death induction, we divided the LMP1 polypeptide into three parts, the N-terminal cytoplasmic domain (aa 1 –26), transmembrane domain (aa 27– 192) and C-terminal cytoplasmic domain (aa 193 –386 or 404), and thereafter generated chimeric constructs between B95.8- and CAOLMP1 (Fig. 2A). We first analyzed the functional consequences of the amino (N-) and carboxy (C-) terminal cytoplasmic domains. C/B/B-LMP1, which has the N-terminal cytoplasmic domain of CAO-LMP1 followed by the transmembrane and C-terminal cytoplasmic domains of B95.8-LMP1, induced cell death in 293T cells as efficiently as B95.8-LMP1 (Fig. 2B and C, lane 3). B/B/C-LMP1, which has the Nterminal and transmembrane domains of B95.8-LMP1 linked to the C-terminus of CAO-LMP1, also induced cell death, although to a slightly lesser degree (lane 4). In contrast, reciprocal chimeras, B/C/C-LMP1 and C/C/B-LMP1 failed to induce cell death (lanes 5 and 6). Western blot analysis revealed that the expression levels of these chimeric proteins were similar (Fig. 2C, bottom panel). Each protein showed different migration probably due to the difference in their molecular weight and net charge of amino acid residues. These chimeras also showed NF-nB activation to similar extents (Fig. 2D). These results indicate that neither the N-terminal or C-terminal cytoplasmic domains of LMP1 are critically involved in cell death induction, and suggest that the transmembrane domain of LMP1 is responsible for its killing property. Indeed, B/C/B-LMP1, which has the transmembrane domain of CAO-LMP1 and the other parts from B95.8-LMP1, lacked the killing property (Fig. 3A, B and C, lane 4). Its reciprocal chimera, C/B/C-LMP1, induced cell death, although it was moderate compared with that by B95.8-LMP1 (lane 5). Their protein expression levels and NF-nB activation were found similar to those of B95.8-LMP1 (Fig. 3C and D). Taken together, these results clearly indicate that the transmembrane domain of LMP1 determines its killing property.
1077
B95.8-LMP1 has Ile85 and Phe106, whereas CAO-LMP1 has the NPC-type residues (Leu85 and Tyr106). This information prompted us to investigate the functional consequences of these residues in the cell death induction. We introduced amino acid substitutions of Ile85Leu, Phe106Tyr or both in B95.8-LMP1, generating B95.8(L)LMP1, B95.8(Y)-LMP1 or B95.8(LY)-LMP1, respectively, and examined their ability to induce cell death in 293T cells. B95.8(L)-LMP1 behaved in a manner similar to B95.8-LMP1 with regard to cell death induction (Fig. 4B and C, lane 4). A slight reduction of cell death induction was observed in cells transfected with B95.8(Y)-LMP1 (lane 5). B95.8(LY)-LMP1, which has two amino acid substitutions, showed a remarkable reduction of the killing property (lane 6). The protein expression levels and extents of NF-nB activation were not significantly altered by these
3.3. Mutation of two amino acids in the transmembrane domain of B95.8-LMP1 diminishes its killing property Fourteen amino acids in the transmembrane domain (aa 27 –192) of CAO-LMP1 are different from those of B95.8LMP1 [20]. A recent report on the sequence of LMP1 genes from NPC patients or from healthy carriers [34] revealed that the 85th and 106th amino acids were Leu and Tyr, respectively, in nearly all the tested LMP1 clones derived from NPC patients, whereas approximately a half of healthy people had different residues in these positions that were in most cases Ile85 and Phe106. Interestingly,
Fig. 5. Transmembrane domain of LMP1 directs activation of caspase-3. (A) 293T cells (6 105) were transfected with 1 Ag of pSG5 (lane 1), pSG5-LMP1 (lane 2), pSG5-CAO-LMP1 (lane 3), pSG5-B95.8(LY) (lane 4), pSG5-B/C/B-LMP1 (lane 5) or pSG5-C/B/C-LMP1 (lane 6). Thirty hours after transfection, the caspase-3 activity that cleaves DEVD-AFC was determined as described in the Materials and methods and expressed relative to that of the vector-transfected cells. The results are representative of three independent experiments. (B) Whole-cell lysates collected 30 h after transfection were subjected to Western blot analysis with anti-PARP antibodies. The arrow and arrowhead indicate intact 116-kDa and processed 85-kDa PARP proteins, respectively.
1078
T. Nitta et al. / Cellular Signalling 16 (2004) 1071–1081
Fig. 6. Transmembrane domain of B95.9-LMP1 induces cell death and caspase activation in the presence of high NF-nB activity. (A) Schematic representation of the structure of LMP1 chimeras fused with EGFP. (B) 293T cells (6 105) were transfected with the plasmids (1 Ag each) as follows; lane 1: pEGFP-N1 and pcDNA3; lane 2: pEGFP-N1 and pRc-CMV-relA; lane 3: pEGFP-B95.8 and pcDNA3; lane 4: pEGFP-B95.8 and pRc-CMV-relA; lane 5: pEGFP-B95.8(LY) and pcDNA3; lane 6: pEGFP-B95.8(LY) and pRc-CMV-relA; lane 7: pEGFP-CAO and pcDNA3; lane 8: pEGFP-CAO and pRc-CMV-relA. The extent of cell death was determined 36 h after transfection (upper panel). NF-nB-dependent luciferase activity was assessed as described in Fig. 1 (middle panel). Values of lanes 1, 3, 5 and 7 are indicated. Whole-cell lysates (15 Ag protein) collected 20 h after transfection were subjected to Western blot analysis with anti-GFP or anti-RelA antibodies (bottom panels). The experiments were repeated three times with essentially similar results. (C) Caspase activation was evaluated by the proteolytic cleavage of PARP. Whole-cell lysates collected 36 h after transfection were subjected to Western blot analysis with anti-PARP antibodies. The arrow and arrowhead indicate intact 116-kDa and processed 85-kDa PARP proteins, respectively.
T. Nitta et al. / Cellular Signalling 16 (2004) 1071–1081
mutations (Fig. 4C and D). Amino acid substitutions of Ile for Leu at aa 85, Phe for Tyr at aa 106, or both in CAOLMP1, however, did not restore the ability to induce cell death (data not shown). Thus, the two amino acid residues, Ile85 and Phe106 are critical, but not solely responsible for the different killing properties of B95.8- and CAO-LMP1. 3.4. Caspase activation depends on the transmembrane domain of LMP1 Our previous report demonstrated that expression of B95.8-LMP1 in 293T cells induced activation of caspase3 [17]. In contrast, expression of CAO-LMP1 failed to increase the caspase-3-like activity that cleaves DEVDAFC, a fluorescent substrate of caspase-3 subfamily proteases (Fig. 5A). Impaired cleavage of PARP, a wellcharacterized substrate for caspase-3, confirmed that CAO-LMP1 could not activate caspase-3 in vivo (Fig. 5B). B95.8(LY)-LMP1 showed a remarkable reduction in the caspase-3 activation and PARP cleavage compared with B95.8-LMP1. B/C/B-LMP1 lost the ability to activate caspase-3, indicating that the transmembrane domain of CAO-LMP1 is responsible for its inability of caspase activation. Conversely, the results that C/B/C-LMP1 restored activation of caspase-3 and cleavage of PARP indicate a critical role for the transmembrane domain of B95.8LMP1 in caspase-3 activation. 3.5. The transmembrane domain of LMP1 alone induces cell death under elevated NF-jB activity Our previous report demonstrated that NF-nB activation was required, but not sufficient for the LMP1-induced cell death [17]. While NF-nB is activated via the C-terminal cytoplasmic domain of LMP1 [1], it remained unclear which domain(s) controls such an additional death signal(s). The result that modifications of the transmembrane domain of LMP1 greatly influences its killing property, but not activation of NF-nB suggests that the transmembrane domain controls a yet undefined death signal(s) independently of its C-terminus. To verify this hypothesis, the transmembrane domain alone should be expressed and tested for cell death induction. However, the level of expression of the truncated B95.8-LMP1 lacking the C-terminal cytoplasmic domain (aa 193 – 386) was much lower compared to that of the fulllength LMP1 (data not shown), suggesting that the Cterminal domain is essential for stable expression of this protein. Besides, the N-terminal domain was reported to be indispensable for proper orientation of the transmembrane domain [35]. Thus, we replaced the C-terminal cytoplasmic domain of B95.8-LMP1 with an unrelated protein, EGFP, generating a chimeric protein B95.8/EGFP (Fig. 6A). This chimeric protein was expressed at a level comparable to B95.8-LMP1. B95.8(LY)/EGFP and CAO/EGFP were generated and detected in the same manner (Fig. 6A). Several studies have previously shown a characteristic localization
1079
of LMP1 in patches at the plasma membrane [25]. Examination by fluorescent microscopy of transfected cells revealed that these LMP1/EGFP chimeric proteins exhibited similar patterns of membrane localization characterized by patches at the plasma membrane (data not shown). Coexpression of B95.8/EGFP with RelA, an active subunit of NF-nB, extensively killed the cells, whereas B95.8/EGFP alone caused only a limited cell death (Fig. 6B, lanes 3 and 4). Transactivation by RelA was partially reduced by coexpression of B95.8/EGFP (Fig. 6B, lane 4), consistent with a previous report that the N-terminal and transmembrane domains had inhibitory effects on NF-nB-dependent transactivation [36]. Co-expression of B95.8/EGFP with RelA also induced PARP cleavage (Fig. 6C, lane 4). These killing activities were obviously dependent on the transmembrane domain, since the induction of cell death and caspase activation were markedly reduced in B95.8(LY)/EGFP (lanes 5 and 6), or lost in CAO/EGFP (lanes 7 and 8). These results indicate that the transmembrane domain of LMP1 triggers death program under high NF-nB activity.
4. Discussion In this paper, we investigated the structural requirements for cell death induction by LMP1, exploring the distinct killing properties of B95.8-LMP1 and CAOLMP1. The variety in the LMP1 structure may potentially contribute to the development of EBV-related malignancies, and CAO-LMP1 has been studied as a representative LMP1 variant derived from NPC [37 –39]. Previous studies demonstrated functional differences between CAOLMP1 and the B lymphocyte-derived prototype B95.8LMP1; Rat-1 fibroblasts transfected with CAO-LMP1 grew in soft agar more efficiently than those with B95.8-LMP1 [23]; CAO-LMP1 showed more potent activation of NF-nB, JNK and STAT, compared with B95.8LMP1 [20,25,40]; CAO-LMP1 failed to induce growth inhibition, expression of certain cell surface molecules and secretion of cytokines in human epithelial cells that had been reported for B95.8-LMP1 [25]. Transient expression of B95.8-LMP1 induced cell death and caspase-3 activation in an NF-nB-dependent manner [17]. In contrast, CAO-LMP1 did not induce cell death or caspase-3 activation, while it activated NF-nB to a level similar to B95.8-LMP1. Our result that CAO-LMP1 shows NF-nB activation similar to, but not significantly greater than B95.8-LMP1, could partly be explained by the differences in the internal control and time points of the assay, or by different cellular backgrounds. Since activation of caspase3 and induction of cell death by B95.8-LMP1 required not only NF-nB activation but also some additional event(s) [17], the difference between these two types of LMP1 in cell killing was expected to result from such unknown function(s).
1080
T. Nitta et al. / Cellular Signalling 16 (2004) 1071–1081
Prior studies indicated that the transmembrane domain, but not the C-terminal activation domain of LMP1 was responsible for the functional differences between B95.8 and CAO LMP1 [20,28,37]. Our studies with B95.8/CAO chimeras show compelling evidence that the transmembrane domain is responsible for the different killing property between B95.8-LMP1 and CAO-LMP1. The substitution of the CAO-LMP1 transmembrane domain for that of B95.8-LMP1 almost completely abolished cell death induction, and the reciprocal manipulation, placing the transmembrane domain of B95.8 in the context of CAO-LMP1, conferred the ability to induce cell death on CAO-LMP1, indicating that the transmembrane domain is a critical determinant for cell death induction by LMP1. Substitution of two amino acids in the transmembrane domain, Ile85Leu and Phe106Tyr, diminished cell death induction by B95.8LMP1. These two residues appear to function cooperatively, since Ile85Leu or Phe106Tyr alone had no or only partial effect on cell death induction. These two residues are located in the third or fourth membrane-spanning segments, respectively [2,40], suggesting that they contribute to the conformation or membrane topology of LMP1 and that the amino acid substitutions may affect the interaction of the membrane-spanning and/or loop segments with cellular factor(s). However, changing these two amino acids on CAO-LMP1 to those of B95.8-LMP1 was not sufficient to confer the killing property on CAO-LMP1. This could partly be explained by the fact that CAO-LMP1 has additionally 12 amino acid substitutions in the transmembrane domain compared with B95.8-LMP1. Our results with LMP1/EGFP chimeras indicate that the transmembrane domains of LMP1 can direct death signal(s) under high NF-nB activity without its C-terminal cytoplasmic domain. In fact, B95.8/EGFP by itself induced weak cell death without any detectable caspase activation, but it induced marked cell death and caspase activation when coexpressed with RelA, both of which were significantly diminished by the two amino acid substitutions in the transmembrane domain. The biological functions of the transmembrane domain of LMP1 have not been fully elucidated. Roles of this domain so far characterized include the induction of self-association on the plasma membrane and localization in lipid rafts that contribute to its efficient signaling [41]. It was reported that the N-terminus and transmembrane domains of B95.8-LMP1 could inhibit cell proliferation and cellular protein synthesis [13,36]. These effects are explained by a very recent finding that the transmembrane domain of LMP1 can induce phosphorylation of eIF2a [42]. The transmembrane domain of B95.8-LMP1 alone was also shown to induce reorganization of the actin cytoskeleton, depending on Cdc42, a Rho-like GTPase [43]. Blake et al. [20] reported that the transmembrane domain of CAO-LMP1 contributes to its half-life longer than that of B95.8-LMP1, while CAO-LMP1 exhibited membrane localization similar to B95.8-LMP1 [25] (data not shown). Our present report has demonstrated for the first time that the
transmembrane domain of LMP1 governs caspase activation and cell death induction under high NF-nB activity. The transmembrane domain of CAO-LMP1 may fail to evoke death signal(s) or may activate survival programs that cancel cell death induction by B95.8-LMP1. Further studies on the functional difference between B95.8-LMP1 and CAO-LMP1 are under way to fully elucidate the molecular mechanisms of this NF-nB-dependent cell death induction. Finally, our results suggest possible contribution of CAO-LMP1 to the pathogenesis of NPC. B95.8-LMP1 requires NF-nB activity not only for cell death induction but also for cell transformation [8]. Previous observations that CAO-LMP1 caused cell transformation more efficiently than B95.8-LMP1 [21 –23] may partly be explained by its defect in the cytotoxic property and unaltered ability to activate NF-nB. In fact, the CAO-type variants of LMP1 are frequently present not only in NPC, but also in other EBVassociated lymphoproliferative disorders [37,39,40]. Importantly, amino acid changes similar to those of CAO-LMP1, in particular the Ile86Leu and Phe106Tyr substitutions, have been widely observed in other LMP1 variants derived from NPC cell lines [37,38], NPC biopsies [38] and NPC patients [34,39]. Thus, it would be interesting to examine if LMP1 variants isolated from a wide range of EBV-associated tumor cells also carry the Ile86Leu and Phe106Tyr substitutions, and are defective in cell death induction. Understanding the molecular mechanisms of B95.8-LMP1induced cytotoxicity and contribution of the variant LMP1 to oncogenesis will help to establish novel diagnostic or therapeutic approaches to EBV-related diseases.
Acknowledgements We would thank Dr. Rowe (University of Wales College of Medicine) for LMP1 expression vectors and antibodies, Dr. Memet (Institut Pasteur Paris) for EF1-lacZ and Dr. Baltimore (California Institute of Technology) for the relA cDNA. We also thank members of the Department of Molecular Virology, Tokyo Medical and Dental University, for helpful discussion and comments. This work was partly supported by Japan Human Science Foundation (grant K1040 to S.Y.).
References [1] E.D. Cahir McFarland, K.M. Izumi, G. Mosialos, Oncogene 18 (1999) 6959 – 6964. [2] S. Fennewald, V. van Santen, E. Kieff, J. Virol. 51 (1984) 411 – 419. [3] A.G. Eliopoulos, S.M. Blake, J.E. Floettmann, M. Rowe, L.S. Young, J. Virol. 73 (1999) 1023 – 1035. [4] A.G. Eliopoulos, L.S. Young, Oncogene 16 (1998) 1731 – 1742. [5] A. Kieser, E. Kilger, O. Gires, M. Ueffing, W. Kolch, W. Hammerschmidt, EMBO J. 16 (1997) 6478 – 6485. [6] O. Gires, F. Kohlhuber, E. Kilger, M. Baumann, A. Kieser, C. Kaiser,
T. Nitta et al. / Cellular Signalling 16 (2004) 1071–1081
[7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] [19] [20] [21] [22] [23]
[24] [25]
A. Hall, B. Scheffer, M. Ueffing, W. Hammerschmidt, EMBO J. 18 (1999) 3064 – 3073. A.G. Eliopoulos, N.J. Gallagher, S.M. Blake, C.W. Dawson, L.S. Young, J. Biol. Chem. 274 (1999) 16085 – 16096. Z. He, B. Xin, X. Yang, C. Chan, L. Cao, Cancer Res. 60 (2000) 1845 – 1848. F. Khanim, Q.Y. Yao, G. Niedobitek, S. Sihota, A.B. Rickinson, L.S. Young, Blood 88 (1996) 3491 – 3501. R.H. Edwards, F. Seillier-Moiseiwitsch, N. Raab-Traub, Virology 261 (1999) 79 – 95. D.M. Walling, N. Shebib, S.C. Weaver, C.M. Nichols, C.M. Flaitz, J. Webster-Cyriaque, J. Infect. Dis. 179 (1999) 763 – 774. J.E. Floettmann, K. Ward, A.B. Rickinson, M. Rowe, Virology 223 (1996) 29 – 40. A. Kaykas, B. Sugden, Oncogene 19 (2000) 1400 – 1410. W. Hammerschmidt, B. Sugden, V.R. Baichwal, J. Virol. 63 (1989) 2469 – 2475. J.J. Lu, J.Y. Chen, T.Y. Hsu, W.C. Yu, I.J. Su, C.S. Yang, J. Gen. Virol. 77 (1996) 1883 – 1892. L.F. Sheu, A. Chen, Y.H. Wei, K.C. Ho, J.Y. Cheng, C.L. Meng, W.H. Lee, Am. J. Pathol. 152 (1998) 63 – 74. T. Nitta, A. Chiba, A. Yamashita, M. Rowe, A. Israe¨l, M. Reth, N. Yamamoto, S. Yamaoka, Cell Signal. 15 (2003) 423 – 433. C. Borner, L. Monney, Cell Death Differ. 6 (1999) 497 – 507. L.F. Hu, E.R. Zabarovsky, F. Chen, S.L. Cao, I. Ernberg, G. Klein, G. Winberg, J. Gen. Virol. 72 (1991) 2399 – 2409. S.M. Blake, A.G. Eliopoulos, C.W. Dawson, L.S. Young, Virology 282 (2001) 278 – 287. L.F. Hu, F. Chen, X. Zheng, I. Ernberg, S.L. Cao, B. Christensson, G. Klein, G. Winberg, Oncogene 8 (1993) 1575 – 1583. X. Zheng, F. Yuan, L. Hu, F. Chen, G. Klein, B. Christensson, Int. J. Cancer 57 (1994) 747 – 753. A.M. Mehl, N. Fischer, M. Rowe, F. Hartmann, H. Daus, L. Trumper, M. Pfreundschuh, N. Mu¨ller-Lantzsch, F.A. Gra¨sser, Int. J. Cancer 76 (1998) 194 – 200. P. Trivedi, L.F. Hu, F. Chen, B. Christensson, M.G. Masucci, G. Klein, G. Winberg, Eur. J. Cancer 30A (1994) 84 – 88. C.W. Dawson, A.G. Eliopoulos, S.M. Blake, R. Barker, L.S. Young, Virology 272 (2000) 204 – 217.
1081
[26] S. Morita, T. Kojima, T. Kitamura, Gene Ther. 7 (2000) 1063 – 1066. [27] D. Liebowitz, R. Kopan, E. Fuchs, J. Sample, E. Kieff, Mol. Cell. Biol. 7 (1987) 2299 – 2308. [28] R.J. Johnson, M. Stack, S.A. Hazlewood, M. Jones, C.G. Blackmore, L.F. Hu, M. Rowe, J. Virol. 72 (1998) 4038 – 4048. [29] G.P. Nolan, S. Ghosh, H.C. Liou, P. Tempst, D. Baltimore, Cell 64 (1991) 961 – 969. [30] S. Yamaoka, G. Courtois, C. Bessia, S.T. Whiteside, R. Weil, F. Agou, H.E. Kirk, R.J. Kay, A. Israe¨l, Cell 93 (1998) 1231 – 1240. [31] T. Saitoh, H. Nakano, N. Yamamoto, S. Yamaoka, FEBS Lett. 532 (2002) 45 – 51. [32] S.N. Ho, H.D. Hunt, R.M. Horton, J.K. Pullen, L.R. Pease, Gene 77 (1989) 51 – 59. [33] M. Rowe, H.S. Evans, L.S. Young, K. Hennessy, E. Kieff, A.B. Rickinson, J. Gen. Virol. 68 (1987) 1575 – 1586. [34] P. Hahn, E. Novikova, L. Scherback, C. Janik, O. Pavlish, V. Arkhipov, J. Nicholls, N. Mu¨ller-Lantzsch, V. Gurtsevitch, F.A. Gra¨sser, Int. J. Cancer 91 (2001) 815 – 821. [35] W.F. Coffin III, K.D. Erickson, M. Hoedt-Miller, J.M. Martin, Oncogene 20 (2001) 5313 – 5330. [36] M.L. Sandberg, A. Kaykas, B. Sugden, J. Virol. 74 (2000) 9755 – 9761. [37] W.E. Miller, J.L. Cheshire, A.S. Baldwin Jr., N. Raab-Traub, Oncogene 16 (1998) 1869 – 1877. [38] S.T. Cheung, S.F. Leung, K.W. Lo, K.W. Chiu, J.S. Tam, T.F. Fok, P.J. Johnson, J.C. Lee, D.P. Huang, Int. J. Cancer 76 (1998) 399 – 406. [39] X.S. Zhang, K.H. Song, H.Q. Mai, W.H. Jia, B.J. Feng, J.C. Xia, R.H. Zhang, L.X. Huang, X.J. Yu, Q.S. Feng, P. Huang, J.J. Chen, Y.X. Zeng, Cancer Lett. 176 (2002) 65 – 73. [40] C.A. Fielding, K. Sandvej, A. Mehl, P. Brennan, M. Jones, M. Rowe, J. Virol. 75 (2001) 9129 – 9141. [41] A. Kaykas, K. Worringer, B. Sugden, EMBO J. 20 (2001) 2641 – 2654. [42] N. Lam, M.L. Sandberg, B. Sugden, J. Virol. 78 (2004) 1657 – 1664. [43] A. Puls, A.G. Eliopoulos, C.D. Nobes, T. Bridges, L.S. Young, A. Hall, J. Cell. Sci. 112 (1999) 2983 – 2992.