Oncogene (2000) 19, 2877 ± 2886 ã 2000 Macmillan Publishers Ltd All rights reserved 0950 ± 9232/00 $15.00 www.nature.com/onc Hepatitis B -related insertional implicates SERCA1 gene in the control of apoptosis

Mounia Chami1,7, Devrim Gozuacik1,7, Kenichi Saigo1,2,7, Thierry Capiod3, Pierre Falson4, Herve Lecoeur5, Tetsuro Urashima2, Jack Beckmann6, Marie-Lyse Gougeon5, Michel Claret3, Marc le Maire4, Christian Bre chot1 and Patrizia Paterlini-Bre chot*,1

1U-370 INSERM, Necker Institute, 75015 Paris, France; 2Second Department of Surgery, Chiba University, Chiba 263-8852, Japan; 3U-442 INSERM, Universite Paris Sud, 91405 Orsay, France; 4URA CNRS 2096, CEA Saclay, 91191 Gif sur Yvette, France; 5Pasteur Institute, 75015 Paris, France; 6Centre National GeÂnotypage, 91057 Evry, France

We have used the Hepatitis B Virus DNA as a and Yun, 1998) and NF-kB signaling (Chirillo et al., probe to identify genes clonally mutated in vivo,in 1996). It has also been suggested that it directly human liver cancers. In a tumor, HBV-DNA was found interacts with cellular proteins controlling cell growth to be integrated into the gene encoding Sarco/Endoplas- (p53 (Truant et al., 1995; Wang et al., 1994a)), DNA mic Reticulum Calcium ATPase (SERCA), which pumps repair (ERCC (Wang et al., 1994b) and UVDDB (Lee calcium, an important intracellular messenger for cell et al., 1995; Sitterlin et al., 1997)), senescence (Sun et viability and growth, from the cytosol to the endoplasmic al., 1998), NFkB activity (Weil et al., 1999), basal reticulum. The HBV X gene cis-activates transcription (RBP5 (Cheong et al., 1995) and RMP chimeric HBV X/SERCA1 transcripts, with splicing of (Dorjsuren et al., 1998)), signal transduction (ATF/ SERCA1 exon 11, encoding C-terminally truncated CREB (Williams and Andrisani, 1995)) and with SERCA1 proteins. Two chimeric HBV X/SERCA1 proteasome (Sirma et al., 1998). Depending on the proteins accumulate in the tumor and form dimers. In experimental conditions, HBV X has also been vitro analyses have demonstrated that these proteins reported to accelerate (Koike et al., 1994) or inhibit localize to the ER, determine its calcium depletion and progression (Benn and Schneider, 1995), and induce cell death. We have also shown that these to induce (Chirillo et al., 1997; Sirma et al., 1999; biological e€ects are related to expression of the Terradillos et al., 1998) or inhibit apoptosis (Wang et SERCA, rather than of the viral moiety. This report al., 1995). Furthermore, HBV X induces HCC in involves for the ®rst time the expression of mutated certain transgenic mice (Kim et al., 1991). However, in SERCA proteins in vivo in a tumor cell proliferation and a di€erent genetic context, HBV X transgenic mice in vitro in the control of cell viability. Oncogene (2000) exhibit only increased susceptibility to chemical 19, 2877 ± 2886. carcinogenes (Slagle et al., 1996), or an accelerated development of c-myc induced HCC (Terradillos et al., Keywords: SERCA1; Hepatitis B Virus; calcium; 1997). apoptosis; cancer In addition to these mechanisms, the clonal integra- tion of HBV-DNA into the host cell genome has been found in more than 90% of HBV-related HCCs Introduction (Paterlini and Bre chot, 1994). HBV-DNA integration may directly promote genetic instability or lead to cis- Chronic HBV infection is a major etiologic factor of activation of growth-related genes (Paterlini and hepatocellular carcinoma (HCC) (Chang et al., 1997; Bre chot, 1994). In certain isolated cases, the viral Hildt et al., 1996). Although no transforming viral genome was found to be integrated into the Retinoic oncogene has so far been identi®ed, HBV-induced acid receptor b gene (Dejean et al., 1986), and Cyclin chronic liver in¯ammation is thought to play an A2 gene (Wang et al., 1990), leading to the expression important role in (Buendia, 1992; of chimeric transcripts and proteins, the transforming Paterlini and Bre chot, 1994). In addition, the expres- activity of which was subsequently proved (Berasain et sion of HBV X, PreS2/S (Kekule , 1994), and PreS1/ al., 1998; Garcia et al., 1993). Reports have also been PreS2/S (Chisari et al., 1989) viral proteins has been made of HBV-DNA integration in the gene encoding associated with the development of HCC. In particular, mevalonate kinase in the PLC/PRF/5 cell line (Graef et the 17 kDa HBV X encoded protein has been shown to al., 1994), and in the Carboxypeptidase N locus in a activate in trans a wide variety of cellular and viral human HCC (Pineau et al., 1996). In both instances, genes (reviewed in Weil et al., 1999). HBV X activates chimeric transcripts were identi®ed in the tumor cells, signal transduction pathways such as Ras/Raf/MAP but the e€ect of their in vitro expression on cell kinase (Benn and Schneider, 1994), Jak1-STAT (Lee phenotype has not been investigated so far. On the other hand, a number of studies have failed to identify HBV-DNA integration into cellular DNA coding sequences, so that insertional mutagenesis has been *Correspondence: P Paterlini-Bre chot, Unite INSERM 370, 156 rue considered as a rare event in HBV-related carcinogen- de Vaugirard, 75015 Paris, France 7The ®rst three authors contributed equally to this work esis (Koike, 1998). Received 12 November 1999; revised 3 April 2000; accepted 4 April In contrast, in the woodchuck animal model, 2000 Woodchuck hepadnavirus (WHV) integration into the HBV-related insertional SERCA1 mutagenesis in a HCC M Chami et al 2878 N-myc2 and c-myc oncogenes is a frequent event, (446 in Figure 1A), fused in frame to SERCA1 cDNA occurring in more than 70% of tumors, and its nucleotide 153 in exon 3. Interestingly, SERCA1 transforming e€ect has been well established (Buendia, cDNA was characterized by the splicing of exon 11, 1992). Interestingly, in some tumors, WHV-DNA leading to a 22 codon frameshift and a premature stop integration occurs in a locus called win, which is codon in exon 12. Three of eight clones exhibited an situated at more than 150 Kb from the N-myc2 gene additional splicing of SERCA1 exon 4. These two (Fourel et al., 1994). chimeric cDNAs and encoded proteins will be referred In HBV-related, human HCC, however, the actual to as Xt/St+4 and Xt/St74: Xt for HBV X truncated impact of insertional mutagenesis has not yet been sequence fused to truncated SERCA1 (St), with or explored exhaustively. This is due to technical limita- without exon 4. tions of the genomic libraries previously used for identifying HBV-DNA integration sites, which pre- Chimeric HBV X/SERCA1 proteins predicted structure cluded large scale screening. prevents calcium pumping Using an original, Alu-PCR-based approach devel- oped in our laboratory (Minami et al., 1995), we have According to the cDNA sequence, the chimeric initiated a screening program of HBV-related HCC, proteins have their SERCA1 N-terminal 51 aminoacids searching for HBV/cellular DNA junctions. In the replaced by 148 aminoacids of the HBV X protein and present study, we report on a liver tumor cell an additional residue encoded by the viral/cellular in proliferation carrying the HBV-DNA integration into frame junction (Figure 2A). Xt/St+4 protein lacks six the SERCA1 gene. putative SERCA1 transmembrane segments (M5-M10), SERCA proteins play a pivotal role in regulating which include ®ve of the six Ca2+ binding residues cellular calcium (Pozzan et al., 1994), which, in turn, (Glu-771, Asn-796, Thr-799, Asp-800, and Glu-908) acts as an intracellular messenger involved in a broad (MacLennan et al., 1997) as well as the cytoplasmic range of specialized and basic cellular activities, loop between transmembrane segments 6 and 7 that ranging from muscle contraction, transmission of also controls Ca2+ binding (Falson et al., 1997). A neuronal signal, cell proliferation and death (Berridge mutated peptide of 22 aminoacids is encoded at the C- et al., 1998). Three distinct SERCA genes encode a terminal end. In addition to that, Xt/St-4 protein lacks number of di€erentially expressed isoforms (Wu et al., a peptide (aa. 74 ± 108) encoding the second putative 1995). SERCA1a and 1b are mainly found in fast- transmembrane segment (M2) and the last ®ve C- twitch skeletal muscle, SERCA2a in slow-twitch terminal residues of M1. The expected size of the Xt/ skeletal and cardiac muscle, SERCA2b is ubiquitous St+4 and Xt/St74 chimeric proteins is 56 and and SERCA3 which also displays isoforms is expressed 53 kDa, respectively. in several tissues, including hemopoietic, endothelial Thus, although these proteins retain the putative and secretory epithelial cells. We show here that HBV calcium binding residue Glu 309 and the phosphory- integration into the SERCA1 gene in a human liver latable Asp-351, consistent with previously reported tumor cis-activates SERCA1 chimeric transcripts mutants (MacLennan et al., 1998), they cannot encoding chimeric proteins. We also provide evidence function as calcium pumps. that the in vitro expression of these transcripts and of their SERCA moiety induces ER calcium depletion and Chimeric proteins are expressed and form dimers in the apoptosis. Our study therefore highlights the implica- tumor tion of SERCA1 gene in clonal cell expansion and the control of cell viability. We then looked for these chimeric and truncated proteins in the tumorous tissue (T86) by Western blot. A chimeric protein (53/56 kDa), reacting with both anti-HBV X (anti-X) and anti-SERCA1 antibodies Results (Figure 2B), was clearly detectable in the tumorous tissue. High resolution analysis demonstrated two HBV-DNA integration into the SERCA1 gene drives the proteins, of 53 and 56 kDa (Figure 2C), which size expression of chimeric transcripts was consistent with the cDNA sequences. We con- By using an HBV X speci®c and an Alu-speci®c primer sistently observed a weaker upper band (56 kDa), as (Minami et al., 1995), we have isolated a PCR compared to the strongest lower band (53 kDa). This fragment of 498 bp from a liver tumorous DNA contrasted with the higher number of clones (5/8) (T86). In this fragment, the HBV X gene was fused encoding the 56 kDa protein. This discrepancy may be in frame with SERCA1 gene exon 3 (67 bp) (Zhang et due to the non representative number of clones or to al., 1995) and intron 3 (247 bp) (MacLennan, personal di€erent conformation and reactivity to the antibody communication) (Figure 1A). of the two proteins. A Northern blot analysis performed on tumorous Oligomerization of SERCA proteins has been (T86) and non tumorous mRNAs showed a 3.2 kb already reported (Martonosi, 1996). Therefore, in this transcript, hybridizing with both HBV and SERCA1 setting, we looked for heterodimers of endogenous liver probes, and expressed in T86 but not in adjacent, non- SERCA isoform (SERCA2b) and chimeric HBV X/ tumorous liver tissue (Figure 1B). A cDNA library, SERCA1 proteins in the tumorous tissue (Figure 2C). constructed with poly(A)+ mRNAs from T86, yielded Anti-SERCA1 antibody showed absence of hetero- eight recombinant clones (Figure 1C). Their 5' ends logous aggregates of SERCA2b with chimeric proteins showed the same HBV X sequence, starting at nt 1309 (predicted size: 163 ± 166 kDa) and, interestingly re- HBV (adr4 subtype) (Raney and McLachlan, 1991), vealed prominent aggregates of chimeric proteins 64 bp before the HBV X ATG, up to nucleotide 1819 (106 kDa), possibly corresponding to chimeric dimers,

Oncogene HBV-related insertional SERCA1 mutagenesis in a HCC M Chami et al 2879

Figure 1 Integration of HBV into SERCA1 gene in a tumorous tissue (T86) and expression of chimeric transcripts. (A) Structure of the PCR product showing the HBV integration in frame into the third exon of SERCA1 gene. HBV speci®c primer (MD26C), and Alu speci®c primer (AluTag 5) were used to amplify the HBV/cellular DNA junction: 8 indicates the 1819th nucleotide of the HBV genome (adr4 genotype), * indicates the 153 nucleotide of SERCA1 cDNA sequence. The amino acid mutated by the HBV integration is underlined. The SERCA1 gene structure is depicted below. Exonic sequences are represented as black boxes. (B) Expression of chimeric HBV X/SERCA1 mRNA in T86. The same Northern blot of poly(A)+ RNA (10 mg) from various tissues (M, muscle (positive control for SERCA1 probe); T86, tumorous tissue T86; NT, adjacent non tumorous tissue NT86; NL, normal + liver and HepG2 , HepG2 cells transfected with HBV (positive control for HBV probe)) was hybridized with SERCA1 and HBV probe, and then normalized with the GAPDH probe. A 3.2 kb transcript hybridizing with both HBV and SERCA1 probe is present in T86 (arrow), but not in NT. In T86 the HBV probe also shows the expression of the 2.1 kb HBV surface transcript. (C) Hybrid cDNAs cloned from T86. Xt/St+4, clones with splicing of SERCA1 exon 11 (-ex11) leading to a frameshift (black box) and a stop codon in exon 12. Xt/St74, clones with exon 11 and exon 4 (-ex4) splicing. X, HBV X sequence in non-heated tumor sample, whereas SERCA1 in which was not signi®cantly lower than that found in normal fast skeletal muscle (SR) underwent only slight three, di€erent, normal livers (Figure 2D) and cell lines aggregation (220 kDa) under the same conditions. The transfected with the Xt/St constructs (data not shown). expression of SERCA2b was sustained in T86 at a level SERCA3 mRNA was expressed at very low and similar

Oncogene HBV-related insertional SERCA1 mutagenesis in a HCC M Chami et al 2880 levels in both tumorous and corresponding non- vectors with or without fusion to GFP or Myc tag tumorous tissues (data not shown). sequences. The subcellular distribution of the encoded proteins was studied in transiently transfected CCL13 cells, Chimeric SERCA proteins are localized to the ER using immuno¯uorescence and scanning confocal We then investigated whether chimeric proteins microscopy (Figure 4). Xt/St+4 and Xt/St74 pro- localize to the ER. According to their predicted teins, as well as their SERCA moieties (St+4 and structure, they lack 51 (Xt/St+4) or 51+34 (Xt/ St74), fused to GFP (Figure 4a,c) or to the Myc St74) N-terminal SERCA1 residues, which include peptide (Figure 4b), colocalized with ER (Figure 4a ± the 28 N-terminal amino acids sequence previously c), but not with mitochondria and golgi (Figure 4d,e). reported as the minimal domain for SERCA proteins The untargeted Xt moiety was mostly nuclear (Figure retention in the ER (Guerini et al., 1998). Constructs 4f), without any ER localization, indicating that the expressing chimeric proteins, their HBV X and Xt part of the chimeric proteins is targeted to the ER SERCA1 moieties, as well as wild type HBV X and by the SERCA moiety. This result led us to clone the SERCA1 proteins (Figure 3) were cloned in expression Xt sequence downstream to a prolactin cassette in order to obtain its targeting to the ER (Figure 3: Xtpro).

Chimeric SERCA proteins induce depletion of ER Ca2+ pools We ®rstly tried to analyse the rate of calcium uptake on isolated microsomal fractions. We tested whether we might use transiently transfected cells, sorted by FACS to obtain equivalent amounts of microsomes from cells transfected with di€erent constructs. However, we noticed a massive cell death in cells transfected with chimeric cDNAs after sorting, avoiding to perform the test. We then investigated whether chimeric proteins induce, upon in vitro expression, modi®cations of the ER and cytosolic Ca2+ pools (Figure 5). Indeed, if the SERCA pump activity is reduced, the ER free calcium content will reach a new equilibrium at lower Ca2+ levels. The 2+ cytosolic free [Ca ]i concentration was studied in cells transiently expressing Xt/St+4-GFP or SERCA1-GFP constructs and in non-transfected cells (NTC). No signi®cant di€erence was found in resting cytosolic 2+ [Ca ]i in Xt/St+4 transfected cells (64+4nM, n=16) when compared to controls (NTC: 61+4nM, n=18 and SERCA1: 66+5nM, n=22) (Figure 5b). In contrast, Xt/St+4 expression decreased the ER

Figure 2 Chimeric proteins are expressed and form dimers in the tumorous tissue (T86). (A) A schematic representation of chimeric proteins: numbers below drawings indicate transmembrane domains. D351: phosphorylatable aspartic acid 351. (.) trans- membrane and (o) cytoplasmic calcium binding residues. The HBV X is depicted as broken line, the peptide created by frameshift is shown as a dotted line and the peptide encoded by exon 4 and deleted in Xt/St74 is in gray. (B) Chimeric protein (X/SERCA) is expressed in T86. The same Western blot (14% SDS ± PAGE) was successively hybridized with anti-HBV X and anti-SERCA1 antibodies. X Ag (21 kDa), puri®ed recombinant HBV X protein (a: 100 ng, b: 50 ng, c: 50 ng mixed with 10 mgof normal liver microsomes); T86, T86 microsomes (50 mg); SR, SERCA1 from rabbit muscle microsomes (50 ng). (C) Western blot (7% SDS ± PAGE) of heated (h) and non-heated (nh) microsomes under mildly denaturing conditions (NL and T86: 50 mg; SR: 20 ng). A band of 106 kDa corresponding to chimeric proteins dimers is present in the T86 non heated sample. (D) SERCA2b protein expression in T86 compared to three normal Figure 3 Schematic representation of the di€erent cloned livers (NL1-3). Western blot analysis was performed on the total constructs. Xt/St+4 and Xt/St74: chimeric clones corresponding protein and normalized with anti-tubulin antibody. The to chimeric cDNAs; St+4 and St74, SERCA moieties of SERCA2b accumulation was not signi®cantly di€erent in T86 chimeric clones; Xt, HBV X moiety; Xtpro, HBV X moiety as compared to normal livers directed to the ER membrane by means of the prolactin cassette

Oncogene HBV-related insertional SERCA1 mutagenesis in a HCC M Chami et al 2881

Figure 4 Chimeric proteins are localized to ER whereas truncated HBV X moiety (Xt) is mainly nuclear. CCL13 cells transiently transfected with Xt/St+47GFP (green) (a), St+47GFP (green) (c) and with Xt/St747Myc (stained with anti-Myc antibody, green) (b) were immunolabeled with anti-ER antibody (red). Cyto¯uorograms with red in abscissae and green in ordinate, demonstrate colocalization of chimeric proteins and SERCA moiety to the ER. The same cells transfected with Xt/St747GFP (green) (d), Xt/St+47GFP (green) (e) and immunolabeled with anti-mitochondria (red) (d), or anti-Golgi antibodies (red) (e) clearly showed that chimeric proteins do not colocalize to these organelles. Cells transfected with Xt moiety (Xt-GFP, green) and immunolabeled with anti-lamin antibodies (red), which speci®cally label the nuclear membrane, showed that the Xt untargeted moiety is mostly localized into the nucleus (f) calcium contents to 75 and 80% of the levels found in Chimeric SERCA proteins induce apoptosis non-transfected and SERCA1 transfected cells, respec- tively (Figure 5c). However, as the whole cell Morphological studies were conducted in CCL13 population included 40% of transfected cells (eval- (Figure 6A,B) and HuH7 cells (data not shown), by uated by GFP expression), the actual level of ER Ca2+ counting the number of apoptotic bodies in GFP depletion related to Xt/St+4 expression is expected to expressing cells (see Materials and methods). They be higher. By comparing the results obtained with demonstrated that both Xt/St+4 and Xt/St74 SERCA1 and SERCA1-GFP, we assessed that fusion proteins, as well as their SERCA moieties, St+4 and to GFP did not per se interfere with ER Ca2+ St74, induce apoptosis to a highly signi®cant degree contents (data not shown). We therefore concluded (Figure 6B). FACScan analysis of mitochondrial that the expression of Xt/St+4 protein leads to a structure by Nonyl Acridine Orange (NAO) incorpora- 2+ signi®cant reduction in the level of exchangeable Ca tion (Petit et al., 1995) in HepG2 cells con®rmed these stored in the ER. The high rate of cell death induced data (Figure 6C). The analysis is based on mitochon- by the Xt/St74 construct precluded such analysis and drial membrane cardiolipins modi®cation, which occurs encouraged us to look for apoptosis. early during apoptosis and leads to a lower incorpora-

Oncogene HBV-related insertional SERCA1 mutagenesis in a HCC M Chami et al 2882

Figure 5 Chimeric proteins signi®cantly reduce the ER Ca2+ pool. (a) Estimation of ER and non-ER Ca2+ contents (see Materials and methods). (b) The size of the TBuBHQ-sensitive ER Ca2+ pool (ER [Ca2+]) was signi®cantly reduced in Xt/St+4- transfected cells compared to non-transfected cells (NTC) and SERCA1 transfected cells. No signi®cant change in the non-ER Ca2+ pool was detected under any of the three experimental conditions. (c) The ER Ca2+ pool/total mobilizable Ca2+ ratio was 51+3% in NTC cells, 38+2% in Xt/St+4 transfected cells and 47+3% in SERCA1 transfected cells. Results are expressed as mean+s.e.m. *, P50.001; **, P50.02 (Student t-test)

tion of NAO. The percentage of apoptotic cells was estimated according to lower NAO incorporation associated to smaller cell size. These results led us to conclude that both chimeric constructs induce apopto- sis in three di€erent cell lines. In an independent experiment we also tested, by FACS analysis, whether transient transfection of HepG2 cells with a construct encoding a truncated SERCA protein having a normal Figure 6 Chimeric proteins induce apoptosis in transfected cells. N-terminal end and a truncated C-terminal end (A) Apoptotic bodies (arrows) were visualized by confocal microscopy (scale bars: 10 mm) 48 h after transfection of CCL13 (SERCA1+4711) was associated to apoptosis. As cells. (B) Percentage of apoptotic cells recorded in at least 300 compared with the empty vector, the SERCA1+4711 transfected CCL13 cells; values are the mean of two independent construct induced apoptosis at a statistically signi®cant experiments. *, P50.001; **, P50.0001; NS, not signi®cant, level (22% vs 6%). calculated vs GFP transfected cells value. GFP, cells transfected with the empty pEGFP-N1 vector. (C) FACScan analysis of apoptosis in HepG2 transfected cells by mitochondrial incorpora- Targeting of Xt to the ER abrogates its proapoptotic and tion of NAO (Nonyl-Acridine Orange) (ordinate) and cell size transactivating property (abscissae). The percentage of apoptotic cells (black dots), with lower NAO incorporation associated to a smaller cell size, is By using both morphological counting of apoptotic shown in each case. NTC, non transfected cells, pcDNA-3, cells bodies (Figure 6B) and FACScan analysis of mito- transfected with empty pcDNA3.1 vector chondrial membrane by NAO incorporation (Figure 6C), we observed that both the wild type (Xwt) and the Discussion truncated untargeted HBV X moiety (Xt) induce apoptosis in three di€erent cell lines. In contrast with Our work reports on a new case of cis-activation of a these ®ndings, the ER targeted HBV X moiety (Xtpro) cellular gene upon HBV-DNA integration in a human barely induced apoptosis, reaching non signi®cant hepatocellular carcinoma. It also identi®es for the ®rst (Figure 6B) or border-line (Figure 6C) levels. Thus time SERCA1 as a potential target for mutation in a the proapoptotic e€ect of chimeric constructs is mainly human cancer. related to their SERCA moiety. Since the proapoptotic SERCA2b is the endogenous SERCA in liver e€ect of HBV X has been associated to its transactivat- tissues, the SERCA1 gene expression being detectable ing property, we analysed the transactivating e€ect of only by RT ± PCR. In the liver tumor we studied, the Xtpro as compared to Xwt and to the Xt moiety. In viral HBV X gene promoter cis-activates two chimeric our system (see Materials and methods), Xwt displayed HBV X/SERCA1 transcripts, characterized by exon 11 transactivating activity on NFkB reporter gene (Sirma splicing and a premature stop codon in exon 12. et al., 1999). In contrast with these ®ndings, we were These transcripts encode chimeric proteins with their not able to evidence a signi®cant transactivating C-terminal end truncated and their N-terminal end activity of Xtpro and Xt/St+4 in triplicate experiments replaced by the C-terminally truncated HBV X (Figure 7). protein.

Oncogene HBV-related insertional SERCA1 mutagenesis in a HCC M Chami et al 2883 heterodimers or aggregates with endogenous SERCA2b proteins and thus alter their biological activity. However, heterodimers were not detectable by Western blot, even under weak denaturing conditions. Instead, dimers of chimeric proteins were found in the tumorous tissue. HBV X may have favored this dimerization in our model, since in vitro expressed HBV X has indeed been shown to dimerize (Lin and Lo, 1989). These results suggest that chimeric proteins may determine calcium leakage from the ER by embedding the ER membrane (ER overload) (Pahl and Baeuerle, 1997) or by forming a cation pore upon dimerization of the two M4 transmembrane segments, which contain a calcium binding residue (Rice and MacLennan, 1996). Finally, calcium release may also follow the accumulation of unfolded proteins in the Figure 7 Targeting of HBV X moiety (Xt) to the ER decreases its transactivating activity. CCL13 cells were co-transfected with ER, leading to an unfolded protein response (Kauf- pcDNA3.1 empty vector (pcDNA-3), wild type HBV X (Xwt), man, 1999). The chimeric SERCA proteins are Xt/St+4 chimeric construct, and Xtpro (Xt targeted to the ER) expected, however, to be inserted into the ER expression plasmids and with reporter vector p-NF-kB-LUC and membrane. Overexpression of the SERCA moieties p-b-actin-b-galactosidase (see Materials and methods). A de- induce apoptosis, while overexpression of the Xtpro creased transactivating activity was observed with Xtpro and the chimeric construct Xt/St+4, both targeting the Xt to the ER, as construct, which is also inserted into the ER compared to the wild type HBV X protein (Xwt) membrane, does not. In agreement with this result, we have also observed that overexpression of the Pro- GFP construct does not induce a decreased ER When compared with the functional structure of wild calcium content (data not shown). Finally, overexpres- type SERCA1, HBV X/SERCA1 proteins have lost sion of truncated SERCA proteins (having the normal calcium binding residues and the ATP binding domain. SERCA1 N-terminal end, and therefore presumably They are therefore unable to pump calcium. Further, correctly folded) induce apoptosis in the same propor- they have also lost 51 N-terminal residues, including tions as compared with the chimeric SERCA proteins. the 28 residues previously identi®ed as the minimal Therefore, even if we cannot formally rule out a minor SERCA1 ER targeting signal (Guerini et al., 1998). contribution of ER overload and/or unfolding protein Using immuno¯uorescence and confocal analysis, we response to the ER calcium depletion and apoptosis, have shown that these chimeric proteins still localize to our data demonstrate that mutated SERCAs are the ER, and do not colocalize with mitochondria or involved in the control of apoptosis. Golgi. Since previous reports (Bayle et al., 1997) ER calcium depletion has been shown to cause showed that the ®rst four predicted SERCA transmem- repression of protein synthesis and folding (Brostrom brane domains, which are retained in the chimeric and Brostrom, 1990) which can induce apoptosis protein, act as signal anchor and stop transfer (Nicotera and Orrenius, 1998). SERCA genes, includ- sequences, chimeric proteins are presumed to be ing SERCA1, have consistently been shown to control inserted into the ER membrane. We cannot, however, Inositol 1,4,5-triphosphate (IP3)-mediated calcium rule out misfolding of these proteins into the ER waves (Camacho and Lechleiter, 1993; Morgan and lumen. Jacob, 1998) and it has been recently shown that With respect to HBV X localization, our results raise calcium wave amplitude and duration di€erentially two interesting points. Wild type HBV X shows a activate certain transcription factors (Dolmetsch et al., predominant cytoplasmic localization (Sirma et al., 1997). In this setting, modi®cation of the SERCA 1998) and it is unclear whether a small fraction function, in the tumor we studied, could have translocates or not to the nucleus (Weil et al., 1999). deregulated cell viability and growth. In fact, the It has been recently shown that large C-terminal transient expression of chimeric proteins induced cell deletions promote HBV X nuclear localization (Sirma death in three HCC-derived cell lines. In line with et al., 1999). Although localization signals have not previous results from our and other laboratories, we been mapped to date, our study shows that deletion of also found that the wild type HBV X protein induces only ®ve C-terminal residues is sucient to drive HBV apoptosis. The HBV X moiety, which has lost only ®ve X to the nucleus. Moreover, we have shown that, in C-terminal residues as compared to the wild type form, the chimeric proteins, the HBV X moiety is driven to also demonstrated a proapoptotic e€ect, but only in its the ER by the SERCA moiety. free form. Namely, apoptotic activity was lost by Cells transiently transfected with the chimeric targeting the HBV X moiety to the ER through a construct display a signi®cant depletion of ER calcium. prolactin cassette. As with previously reported data, Several di€erent, non-exclusive mechanisms may showing that the proapoptotic e€ect of HBV X account for this depletion. Chimeric proteins may correlates with its transactivation activity, we found downregulate endogenous SERCA2b expression. How- that targeting the HBV X moiety to the ER ever, we failed to demonstrate a signi®cantly decreased signi®cantly reduced its transactivating property. expression of SERCA2b in the tumorous tissue (T86) Finally, transient transfection of the SERCA moi- as compared to normal livers and in cells transfected eties massively induced apoptosis. Taken together, with the chimeric constructs. SERCA proteins have these data demonstrate that the SERCA moieties been shown to dimerize. Chimeric proteins may form account for most of the apoptotic activity due to

Oncogene HBV-related insertional SERCA1 mutagenesis in a HCC M Chami et al 2884 chimeric proteins. In agreement with our results, Ma et PCR and RT ± PCR protocols and primer sequences al. (1999) have recently provided evidence for a direct implication of SERCA2a in apoptotic cell death. DNA was extracted from tumorous and non tumorous This is the ®rst report demonstrating that direct tissues as previously described (Paterlini et al., 1990). In order mutation of the SERCA1 gene a€ects the control of to amplify HBV/cellular DNA junctions in the tumorous tissue, Alu-PCR was performed as previously reported cell death. SERCA1 gene have been (Minami et al., 1995). reported in some patients with Brody disease (Mac- Lennan et al., 1998) characterized by autosomal recessive inheritance of SERCA1 (ATP2A1) gene RNA isolation and Northern blot analysis mutations on chromosome 16p12.1 ± 12.2. In these Total cellular RNA was extracted using Trizol reagent (Life patients, muscle relaxation is impaired because of the Technologies). mRNA isolation was done using Dynabeads expression of mutated SERCA1 proteins, which are oligo (dT) (Dynal). Northern blot was performed as unable to pump calcium. Recently, SERCA2 gene described (Church and Gilbert, 1984). Full-length SERCA1 mutations have been reported in Darier-White disease cDNA and full-length HBV cDNA were used as speci®c probes. Blots were normalized with a full-length GAPDH (Sakuntabhai et al., 1999), a rare dermatosis with cDNA probe. dominant inheritance characterized by focal areas of separation between suprabasal epidermal cells, and abnormal keratinocytes di€erentiation. Darier-White cDNA cloning disease SERCA2 mutations therefore highlight the fact Double strand cDNA was synthesized using the Marathon that aberrations of intracellular calcium levels may lead TM cDNA Ampli®cation kit (Clontech) starting from 1 mgof to a distinct program of altered cell growth and poly(A)+ RNA from T86. The cDNA 5' and 3' ends were di€erentiation (Peacocke and Christiano, 1999). ampli®ed and sequenced with adaptor primer AP1 (Mara- Our work constitutes the ®rst report on an in vivo thon kit) and HBV-junction primers Hsh-s (5'-accagcaccatg- caagctggtgatagagca-3' and Hsh-r (5'-accagcaccatgcaagctgg- SERCA1 gene mutation occurring selectively in tgatagagca-3'). Full-length hybrid cDNAs were ampli®ed tumorous cells and not in adjacent, non tumorous with HBV X-s (5'-tttccatggctgctaggctgtactgccaa-3') and liver tissue. Chromosomal aberrations involving the SERCA1 ex23-r (5'-cacagctctgcctgaagatgtgtcact-3') primers 16p12 region, which contains the SERCA1 gene, have and cloned into pMos vector (Gibco ± BRL). been detected in several tumors, including pancreatic endocrine tumors, colorectal carcinomas and semino- Isolation of microsomal fraction and immunoblotting mas (Chung et al., 1998; Ottesen et al., 1997; Pirc- Danoewinata et al., 1996). Thus, depending on the Liver microsomes were isolated as described (Heilmann et al., cellular environment, mutations of SERCA1 might 1985). Proteins were treated with concentrated urea (20 mg/ 50 mg protein), SDS 1.8% and b-mercaptoethanol 9%. After deregulate cell viability and growth. SDS ± PAGE (14%), proteins were electrotransferred onto Our data provide new information concerning the PVDF membranes (Immobilon). For immunodetection, we impact of HBV-related insertional mutagenesis. This used rabbit polyclonal antibody raised against HBV X mechanism has been considered as a rare event in protein (gift from P Arbuthnot), guinea-pig polyclonal human liver cancer. The number of tumors in which antibody 79 B raised against rabbit fast twitch muscle HBV integrants were studied in detail has remained, SERCA1 (gift from AM Lompre ), mouse monoclonal however, low (around 30 in a 20-year period). We have antibody raised against human SERCA2 (anti-SERCA2, therefore developed an HBV-Alu PCR based technique clone IID8, Tebu) and mouse monoclonal antibody raised to rapidly isolate viral/cellular junctions (Minami et al., against human b-tubulin (Boehringer Mannheim). Following 1995). This approach has allowed us to isolate 18 reaction with a horseradish peroxydase-coupled secondary antibody (goat anti rabbit or anti-guinea-pig or anti-mouse), HBV/cellular DNA junction fragments; among these, immunodetection was visualized by the super signal kit new cellular genes have been recently identi®ed (work (Pierce). Dimers or aggregates were sought with 7% SDS ± in progress). Thus, our results suggest that the actual PAGE in non-heated samples under mildly denaturing prevalence of HBV-related cis-activation of cellular conditions (without urea) (Soulie et al., 1996). genes should be reassessed. In conclusion, our results stimulate further analysis Plasmid constructs and transient transfection of SERCA gene mutations in cancer, and reinforce the concept that the pharmacological modulation of The constructs shown in Figure 3 were subcloned into intracellular calcium may have an impact in cancer pcDNA3.1, pcDNA3.1/Myc-His (Invitrogen) and pEGFP-N1 (Clontech) and were used to transfect hepatoma derived cell treatment (Kohn and Liotta, 1995). lines (CCL13, HuH7 and HepG2) by the calcium phosphate coprecipitation method (10 mg/dish). The Xtpro construct was obtained by subcloning Xt downstream the ER membrane targeting cassette (prolactin peptide signal, b Materials and methods lactamase sequence and IgM stop transfer signal, gift of V Lingappa). In addition to that, we also cloned the prolactin Tissues cassette upstream to the GFP sequence (ProGFP) and the SERCA1+4711 constructs. Tumorous (T86) and non tumorous (NT86) frozen liver tissues were obtained at surgery from a 65-year-old Japanese, Immunofluorescence HBsAg negative, HBV-DNA positive, anti-HCV positive patient with primary liver cancer. The tumorous tissue was Cells were ®xed in 4% paraformaldehyde and permeabilized classi®ed as HCC Edmonson's grade 2. Chronic active for 30 s in acetone/methanol (1 : 1). Non-speci®c binding sites hepatitis without cirrhosis was found in the non tumorous were blocked with 3% bovine serum albumin. Immunostain- tissue. Other tumors used for this study were diagnosed as ing was performed with anti-ER polyclonal antibody (gift primary liver cancer after pathological examination. from D Louvard), anti-myc monoclonal antibody (9E10,

Oncogene HBV-related insertional SERCA1 mutagenesis in a HCC M Chami et al 2885 Santa Cruz), anti-mitochondria polyclonal antibody (gift (NAO) incorporation (Petit et al., 1995). Adherent and from C Marsac), anti-Golgi monoclonal antibody (gift from ¯oating cells were incubated for 30 min at 378C in PBS M Bornens), and anti-lamin B antibody. Immunodetection containing 4 mM NAO and harvested cells were then analysed was carried out using a ¯uorescein isothiocyanate (FITC) in the FACScan ¯ow cytometer. Data from 20 000 events (green), texas red or Cy5 (red)-conjugated second antibody. were recorded on a logarithmic scale and analysed using cell Coverslips were analysed using confocal laser scanning Quest software. microscopy (Zeiss, LSM 510). Transactivation assay Cell calcium contents CCL13 cells were cotransfected with 2 mg of each of reporter CCL13 cells were loaded with the ¯uorescent Ca2+ indicator vectors: p-NF-kB-LUC, (containing the luciferase gene under fura2-AM (1.2 mM) for 40 min at 378C in loading bu€er that the control of the NF-kB promoter) and p-b-actin-b- does not contain Ca2+. Successive applications of EGTA galactosidase (b-galactosidase gene under control of b-actin (0.1 mM), 2,5-Di-tert-butylhydroquinone (TBUBHQ, 50 mM), promoter) and 5 mg of the di€erent activators (wild type X, speci®cally inhibiting SERCA and thus releasing the ER Xtpro, Xt/St+4 and the empty expression plasmid calcium, and of ionomycin (IONO, 2 mM), which is a non- pcDNA3). Luciferase and b galactosidase analyses were speci®c calcium ionophore, enabled us to determine the performed according to standard protocols (Sirma et al., 2+ 2+ increase in [Ca ]i due to the ER and non-ER Ca pools, 1999). Luciferase activity was normalized against b galacto- respectively (Figure 5a). The cytosolic-free Ca2+ concentra- sidase activity. Data presented are the mean values of 2+ tion ([Ca ]i) was determined by the ratio of 340/380 nm transfections done in triplicate experiments. 2+ ¯uorescence intensities using a Kd for Ca -dye of 225 nM, and the maximum and minimum ¯uorescence responses were determined as described (Grynkiewicz et al., 1985).

Acknowledgments Apoptosis We thank P Champeil for critical reading of the paper and Transfected cells were analysed 48 h after transfection with most helpful discussion. This work was supported by GFP-fusion constructs. The percentage of apoptotic nuclei grants from INSERM, LNC no 8981 and ARC no 9327 was evaluated in GFP-positive cells using the DNA-dye 7- (France). M Chami, D Gozuacik and K Saigo are recipient amino actinomycin D (7-AAD). Alterations to mitochondrial of fellowships from LNC (France), TUBITAK (Turkey) structure were assessed following Nonyl Acridine Orange and FRM (France), respectively.

References

Bayle D, Weeks D, Hallen S, Melchers K, Bamberg K and Dorjsuren D, Lin Y, Wei W, Yamashita T, Nomura T, Sachs G. (1997). J. Recept. Signal. Transduct. Res., 17, Hayashi N and Murakami S. (1998). Mol. Cell. Biol., 18, 29 ± 56. 7546 ± 7555. Benn J and Schneider R. (1994). Proc. Natl. Acad. Sci. USA, Falson P, Menguy T, Corre F, Bouneau L, de Gracia AG, 91, 10350 ± 10354. Soulie S, Centeno F, Moller JV, Champeil P and le Maire, Benn J and Schneider R. (1995). Proc. Natl. Acad. Sci. USA, M. (1997). J. Biol. Chem., 272, 17258 ± 17262. 92, 11215 ± 11219. Fourel G, Couturier J, Wei Y, Apiou F, Tiollais P and BerasainC,PatilD,PereraE,HuangS,MoulyHand Buendia MA. (1994). EMBO J., 13, 2526 ± 2534. Bre chot C. (1998). Oncogene, 16, 1277 ± 1288. Garcia M, De The H, Tiollais P, Samarut J and Dejean A. Berridge MJ, Bootman MD and Lipp P. (1998). Nature, 395, (1993). Cell Biol., 90, 89 ± 93. 645 ± 648. Graef E, Caselmann W, Wells J and Koshy R. (1994). Brostrom CO and Brostrom MA. (1990). Annu. Rev. Oncogene, 9, 81 ± 87. Physiol., 52, 577 ± 590. Grynkiewicz G, Poenie M and Tsien RY. (1985). J. Biol. Buendia MA. (1992). Adv. Cancer Res., 59, 167 ± 226. Chem., 260, 3440 ± 3450. Camacho P and Lechleiter JD. (1993). Science, 260, 226 ± GueriniD,Garcia-MartinE,ZeccaA,GuidiFandCarafoli 229. E. (1998). Acta. Physiol. Scand. Suppl., 643, 265 ± 273. Chang MH, Chen CJ, Lai MS, Hsu HM, Wu TC, Kong MS Heilmann C, Spamer C and Gerok W. (1985). J. Biol. Chem., et al. (1997). New Engl. J. Med., 336, 1855 ± 1859. 260, 788 ± 794. CheongJH,YiM,LinYandMurakamiS.(1995).EMBO J., Hildt E, Hofschneider P and Urban S. (1996). Virology, 7, 14, 143 ± 150. 333 ± 347. Chirillo P, Falco M, Puri PL, Artini M, Balsano C, Levrero Kaufman RJ. (1999). Genes Dev., 13, 1211 ± 1233. M and Natoli G. (1996). J. Virol, 70, 641 ± 644. Kekule AS. (1994). Primary liver cancer: Etiological and Chirillo P, Pagano S, Natoli G, Puri P, Burgio V, Balsano C progression factors, Vol. 13. Bre chot C. (ed). CRC Press: and Levrero M. (1997). Proc. Natl. Acad. Sci. USA, 94, London. 8162 ± 8167. Kim C, Koike K, Saito I, Miyamura T and Jay G. (1991). Chisari FV, Klopchin K, Moriyama T, Pasquinelli C, Nature, 351, 317 ± 320. DunsfordHA,SellS,PinkertCA,BrinsterRLand Kohn E and Liotta L. (1995). Cancer Res., 55, 1856 ± 1862. Palmiter RD. (1989). Cell, 59, 1145 ± 1156. Koike K. (1998). Hepatitis B virus: molecular mechanisms in Chung DC, Brown SB, Graeme-Cook F, Tillotson LG, disease and novel strategies for therapy, Vol. 3. Koshy R Warshaw AL, Jensen RT and Arnold A. (1998). Cancer and Caselmann WH (eds). Imperial College Press: Res., 58, 3706 ± 3711. London, pp. 133 ± 160. Church GM and Gilbert W. (1984). Proc. Natl. Acad. Sci. Koike K, Moriya H, Yotsuyanagi H, Iino S and Kurokawa USA, 81, 1991 ± 1995. K. (1994). J. Clin. Invest., 94, 44 ± 49. Dejean A, Bougueleret L, Grzeschik KH and Tiollais P. Lee TH, Elledge SJ and Butel JS. (1995). J. Virol., 69, 1107 ± (1986). Nature, 322, 70 ± 72. 1114. Dolmetsch RE, Lewis RS, Goodnow CC and Healy JI. Lee YH and Yun Y. (1998). J. Biol. Chem., 273, 25510 ± (1997). Nature, 386, 855 ± 858. 25515.

Oncogene HBV-related insertional SERCA1 mutagenesis in a HCC M Chami et al 2886 Lin MH and Lo SC. (1989). Biochem. Biophys. Res. SakuntabhaiA,Ruiz-PerezV,CarterS,JacobsenN,Burge Commun., 164, 14 ± 21. S, Monk S, Smith M, Munro CS, O'Donovan M, Ma TS, Mann DL, Lee JH and Gallinghouse GJ. (1999). Cell Craddock N, Kucherlapati R, Rees JL, Owen M, Lathrop Calcium, 26, 25 ± 36. GM, Monaco AP, Strachan T and Hovnanian A. (1999). MacLennan DH, Rice WJ and Green NM. (1997). J. Biol. Nat. Genet., 21, 271 ± 277. Chem., 272, 28815 ± 28818. SirmaH,GianniniC,PoussinK,PaterliniP,KremsdorfD MacLennan DH, Rice WJ, Odermatt A and Green NM. and Bre chot C. (1999). Oncogene, 18, 4848 ± 4859. (1998). Acta. Physiol. Scand. Suppl., 643, 55 ± 67. Sirma H, Weil R, Rosmorduc O, Urban S, Israel A, Martonosi AN. (1996). Biochim. Biophys. Acta., 1275, 111 ± Kremsdorf D and Bre chot C. (1998). Oncogene, 16, 117. 2051 ± 2063. Minami M, Poussin K, Bre chot C and Paterlini P. (1995). Sitterlin D, Lee TH, Prigent S, Tiollais P, Butel JS and Genomics, 29, 403 ± 408. Transy C. (1997). J. Virol., 71, 6194 ± 6199. Morgan AJ and Jacob R. (1998). J. Physiol. (Lond)., 513, Slagle B, Lee T, Medina D, Finegold M and Butel J. (1996). 83 ± 101. Mol. Carcinog., 15, 261 ± 269. Nicotera P and Orrenius S. (1998). Cell Calcium, 23, 173 ± Soulie S, Moller JV, Falson P and le Maire M. (1996). Anal. 180. Biochem., 236, 363 ± 364. Ottesen AM, Kirchho€ M, De-Meyts ER, Maahr J, Gerdes Sun BS, Zhu X, Clayton MM, Pan J and Feitelson MA. T, Rose H, Lundsteen C, Petersen PM, Philip J and (1998). Hepatology, 27, 228 ± 239. Skakkebaek NE. (1997). Genes Chromosomes Cancer, 20, Terradillos O, Billet O, Renard C, Levy R, Molina T, Briand 412 ± 418. P and Buendia M. (1997). Oncogene, 14, 395 ± 404. Pahl HL and Baeuerle PA. (1997). Trends Biochem. Sci., 22, Terradillos O, Pollicino T, Lecoeur H, Tripodi M, Gougeon 63 ± 67. ML, Tiollais P and Buendia MA. (1998). Oncogene, 17, Paterlini P and Bre chot C. (1994). Primary liver cancer: 2115 ± 2123. Etiological and progression factors, Vol. 12. Bre chot C. Truant R, Antunovic J, Greenblatt J, Prives C and Cromlish (ed). CRC Press: London, pp. 167 ± 190. J. (1995). J. Virol., 69, 1851 ± 1859. PaterliniP,GerkenG,KhemenyD,FrancoD,D'ErricoA, Wang J, Chenivesse X, Henglein B and Bre chot C. (1990). Grigioni W, Wands J, Kew M, Pisi E, Tiollais P and Nature, 343, 555 ± 557. Bre chot C. (1990). Viral hepatitis and liver disease. Wang X, Forrester K, Yeh H, Feitelson M, Gu J and Harris Hollinger FB, Lemon SM and Margolis HS (eds). C. (1994a). Proc. Natl. Acad. Sci. USA, 91, 2230 ± 2234. Williams and Wilkins: Baltimore, pp. 556 ± 559. Wang X, Gibson M, Vermeulen W, Yeh H, Forester K, Peacocke M and Christiano AM. (1999). Nat. Genet., 21, Stuà rzbecher H, Hoeijmekers J and Harris C. (1995). 252 ± 253. Cancer Res., 55, 6012 ± 6016. Petit PX, Lecoeur H, Zorn E, Dauguet C, Mignotte B and WangXW,ForresterK,YehH,FeitelsonMA,GuJRand Gougeon ML. (1995). J. Cell. Biol., 130, 157 ± 167. Harris CC. (1994b). Proc. Natl. Acad. Sci. USA, 91, Pineau P, Marchio A, Terris B, Mattei MG, Tu ZX, Tiollais 2230 ± 2234. P and Dejean A. (1996). J. Virol., 70, 7280 ± 7284. WeilR,SirmaH,GianniniC,KremsdorfD,BessiaC, Pirc-Danoewinata H, Bull JP, Okamoto I, Karner J, Dargemont C, Bre chot C and Israel A. (1999). Mol. Cell. Breiteneder S, Liebhardt A, Budinsky A and Marosi C. Biol., 19, 6345 ± 6354. (1996). Wien Klin. Wochenschr., 108, 752 ± 758. Williams JS and Andrisani OM. (1995). Proc. Natl. Acad. Pozzan T, Rizzuto R, Volpe P and Meldolesi J. (1994). Sci. USA, 92, 3819 ± 3823. Physiol. Rev., 74, 595 ± 635. WuKD,LeeWS,WeyJ,BungardDandLyttonJ.(1995). Raney AK and McLachlan A. (1991). of Am. J. Physiol., 269, C775 ± C784. Hepatitis B Virus. McLachlan A. (ed). CRC Press. pp. 1 ± Zhang Y, Fujii J, Phillips M, Chen H-S, Karpati G, Yee W- 26. C, Schrank B, Cornblath D, Bobylan K and MacLennan Rice WJ and MacLennan DH. (1996). J. Biol. Chem., 271, D. (1995). Genomics, 30, 415 ± 424. 31412 ± 31419.

Oncogene