© 2000 Nature America, Inc. 0929-1903/00/$15.00/ϩ0 www.nature.com/cgt

Targeted killing of carcinoembryonic antigen (CEA)- producing cholangiocarcinoma cells by polyamidoamine dendrimer-mediated transfer of an Epstein-Barr virus (EBV)-based plasmid vector carrying the CEA promoter

Saiyu Tanaka,1 Masaki Iwai,1 Yoshinori Harada,1 Teruhisa Morikawa,1 Akira Muramatsu,1 Takahiro Mori,1 Takeshi Okanoue,1 Kei Kashima,1 Hiroko Maruyama-Tabata,2,3 Hideyo Hirai,3 Etsuko Satoh,3 Jiro Imanishi,3 and Osam Mazda3 1Third Department of Internal Medicine, and Departments of 2Pediatrics and 3Microbiology, Kyoto Prefectural University of Medicine, Kamikyo, Kyoto, Japan.

The present study reports a novel nonviral method to efficiently and specifically target carcinoembryonic antigen (CEA)-producing cholangiocarcinoma (CC) cells in vitro. Epstein-Barr virus (EBV)-based and conventional plasmid vectors were constructed that possess the ␤-galactosidase (␤-gal) or herpes simplex virus-1 (HSV-1) thymidine kinase (Tk) genes as well as tandem repeats of the human genomic sequence Ϫ82 to Ϫ42 bp from the transcriptional start site of the CEA gene. The plasmids were transfected by means of polyamidoamine dendrimer into CEA-positive (HuCC-T1) or -negative cell lines. Transfection of the conventional plasmid vector with the CEA promoter and ␤-gal gene resulted in a very low or undetectable level of marker gene expression even in the CEA-positive cell line. Transferring the HSV-1 Tk gene by conventional plasmid did not affect the susceptibility of HuCC-T1 cells to ganciclovir. In marked contrast, strong ␤-gal expression was specifically obtained in HuCC-T1 cells by transfecting the EBV-based plasmid in which the CEA promoter and a ubiquitous promoter (SR␣) are employed to drive the EBV-encoded nuclear antigen 1 (EBNA1) and ␤-gal genes, respectively (pTES.␤). Furthermore, CEA-positive but not -negative tumor cells were rendered highly susceptible to ganciclovir when transfected with the EBV-based vector that carries the CEA promoter-EBNA1 and SR␣-HSV-1 Tk genes (pTES.Tk). These results strongly suggest that the EBV-based plasmid vector/cationic polymer system (EBV/polyplex) equipped with the CEA promoter provides an efficient nonviral method for the targeted gene therapy of CEA-producing malignancies. Cancer Gene Therapy (2000) 7, 1241–1249

Key words: Epstein-Barr virus-based vector; carcinoembryonic antigen promoter; polyamidoamine dendrimer; EBV-encoded nuclear antigen 1; cholangiocarcinoma; gene therapy.

he prognosis of cholangiocarcinoma (CC) is poor. ers, such as ␣-fetoprotein and carcinoembryonic antigen TSurgery is the only effective therapy, but in most (CEA) promoters, shows promise. The ␣-fetoprotein cases, CC is diagnosed too late to be completely resected. promoter has been used successfully to target hepato- 1,2 Only 15–20% of patients have resectable tumors. Pa- cellular carcinoma (HCC) in combination with adenovi- tients with unresectable tumor do not survive 5 years after rus or retrovirus vectors.4–7 3 surgery, and their median survival is 8 months. There is no CEA is a widely used tumor marker, especially for effective chemotherapy or radiation therapy, and little cancers in the gastrointestinal tract. It is a 180-kDa progress has been made in the treatment of this cancer. A glycoprotein with an oncofetal expression pattern.8 In new strategy is needed to treat advanced CC, and gene some earlier studies, the CEA promoter was employed therapy is an attractive choice. to target CEA-producing tumors.9–13 Most of these For effective and safe suicide gene therapy, it is studies employed a DNA fragment corresponding to the necessary to induce strong, tumor-specific gene expres- CEA gene upstream regulatory regions originally re- sion. To this end, the use of the tumor-specific promot- ported by Schrewe et al14 (Ϫ316 to ϩ107 from the transcriptional start site). Richards et al9 reported that the sequence between Ϫ41 and Ϫ18 was essential for Received December 30, 1999; accepted April 15, 2000. transcription from the CEA promoter. Multimerization Address correspondence and reprint requests to Dr. Osam Mazda, Ϫ Ϫ Department of Microbiology, Kyoto Prefectural University of Medicine, of the sequence 89 to 40 resulted in an increase in Kamikyo, Kyoto 602-8566, Japan. E-mail address: [email protected] both the expression level and specificity for CEA-posi- m.ac.jp tive cells in a copy number-dependent manner. Tumor-

Cancer Gene Therapy, Vol 7, No 9, 2000: pp 1241–1249 1241 1242 TANAKA, IWAI, HARADA, ET AL: NONVIRAL GENE TRANSFER TO CEA-POSITIVE TUMOR CELLS specific gene expression was achieved with retrovirus or adenovirus vectors.9,15,16 In the case of nonviral vectors, however, transfection with plasmid vectors carrying the CEA promoter did not result in sufficient therapeutic outcome.17,18 Nonviral vector systems may provide safe gene deliv- ery due to their low immunogenicity and cell toxicity. However, the nonviral systems devised to date have generally been insufficient in terms of intensity and longevity of gene expression. This may be the reason why nonviral systems engaging the CEA promoter were not successful in eliciting a therapeutic outcome specific for CEA-producing tumor cells. Figure 1. a: 5Ј regulatory region of human CEA gene. b: Structure of We have demonstrated previously that extremely ef- the CEA promoter used in this study. Quadrupled sequences of the ficient transfection can be achieved by using Epstein- enhancer region (Ϫ82 to Ϫ42 from the transcriptional start site) 19–23 Barr virus (EBV)-based plasmid vector. The EBV- (filled box) and a single minimum promoter region (Ϫ42 to ϩ69) are based plasmid vector carries the EBV-encoded nuclear included. antigen 1 (EBNA1) gene and origin of replication (oriP) region from the EBV genome.24,25 Through binding to oriP, EBNA1 facilitates nuclear localization, binding to Cell Bank (Tsukuba, Japan), respectively. These cells were the nuclear matrix, and replication of the plasmid DNA. maintained in RPMI 1640 medium (Life Technologies, Grand Island, NY) supplemented with 10% fetal bovine sera, 2 mM Another function of EBNA1 is transcriptional up-regu- L-glutamine, 100 U/mL penicillin, and 100 ␮g/mL streptomy- lation, and these functions make this vector highly 26–30 cin. A4573 cells were maintained in RPMI 1640 medium efficient. No infectious virus particle can be pro- supplemented with 10% fetal bovine sera, 2 mM L-glutamine, duced using this vector. Recently, we have reported that 0.1 mM nonessential amino acids, 1 mM sodium pyruvate, 100 use of the EBV-based plasmid vector combined with U/mL penicillin, and 100 ␮g/mL streptomycin. polyamidoamine dendrimer (PAAD) (EBV/polyplex) resulted in a markedly high expression in human HCC CEA production assay and Ewing’s sarcoma cell lines, suggesting the potential Cells were plated at a density of 2 ϫ 105 cells/well in 24-well usefulness of the EBV/polyplex system for gene thera- cell culture plates (Corning, Corning, NY); after 48 hours of py.31,32 cultivation at 37°C in 5% CO2/95% humidified air, the culture However, no study has been reported in which a medium was renewed. Seven days later, the supernatant was tumor-specific promoter was employed in the EBV- collected and the CEA concentration was measured by solid based plasmid vector to target tumor cells. In this study, phase radioimmunoassay using a CEA RIABEAD kit (Dain- we combined the multimerized CEA promoter with the abot, Tokyo, Japan). CEA expression was standardized based on the protein content of cells measured by a protein assay EBV/polyplex system to establish an efficient nonviral 33 strategy to target CEA-producing cancers. kit. Transfection by PAAD MATERIALS AND METHODS Cells were seeded into wells of 6-well plates (Becton Dickinson Plasmid vectors Labware, Franklin Lakes, NJ) at a density of 1 ϫ 105 (HuCC- T1) or 3 ϫ 105 (SSP-25) cells/well and incubated at 37°C in 5% The CEA promoter was constructed according to the pub- ␮ 14 Ϫ ϩ CO2/95% humidified air. On the following day, 2 gofDNA lished sequence. The upstream region 89 to 77 bp from ␮ the transcriptional start site of the CEA gene was amplified by was mixed with 30 g of PAAD (generation 5) (Qiagen, Hilden, Germany) in RPMI 1640 and incubated for 10 minutes polymerase chain reaction with a pair of primers (sense primer, ␮ 5Ј-TATGATATCAGGGGAGGGGACAGAGGACA; anti- at room temperature. The complex was added to 600 Lof sense primer, 5Ј-GGGATCGATCGAGAAGAGCTTGAGT- complete medium and immediately transferred to cells that TCCA). The resultant polymerase chain reaction product was had been washed with phosphate-buffered saline (PBS). The inserted into the pBluescript II phargemid (Stratagene, La cells were incubated at 37°C in 5% CO2/95% humidified air for Jolla, Calif). Tetraplicated CEA promoter was constructed by 2 hours, and the medium was renewed. a standard molecular biology technique (Fig 1). Four plasmid ␤ vectors, pSES.␤, pS.␤, pSES.Tk, and pS.Tk (Figs 2 and 3), have 5-bromo-4-chloro-3-indolyl -D-galactoside (X-Gal) been described previously.21,31 The other plasmid vectors, staining ␤ ␤ ␤ ␤ pT. , pTES. , pSET. , pTET. , pT.Tk, and pTES.Tk (Figs 2 Cells were fixed with 1% glutaraldehyde/PBS for 10 minutes, and 3), were constructed by substituting the tetraplicated CEA ␣ ␤ ␤ washed three times with PBS, and subsequently incubated for promoter for the SR promoter in pS. , pSES. , pS.Tk, or 3 hours at 37°C in X-Gal staining solution (0.05% (vol/vol) pSES.Tk. X-Gal (Nacalai Tesque, Kyoto, Japan), 1 mM MgCl2, 150 mM Cell lines and cell culture NaCl, 3 mM K4 [Fe(CN)6],3mMK3[Fe(CN)6], 60 mM Na2HPO4,40mMNaH2PO4, and 0.1% Triton X-100). The HuCC-T1 and SSP-25 cells were purchased from the Health reaction was terminated by replacing the solution with 1 mM Science Research Resources Bank (Osaka, Japan) and Riken ethylenediaminetetraacetic acid/PBS.

Cancer Gene Therapy, Vol 7, No 9, 2000 TANAKA, IWAI, HARADA, ET AL: NONVIRAL GENE TRANSFER TO CEA-POSITIVE TUMOR CELLS 1243

Figure 2. Schematic diagram of ␤-gal expression vectors. Maps of pSES.␤, pS.␤, pT.␤, pTES.␤, pSET.␤, and pTET.␤ are shown. Amp, ampicillin-resistant gene; ␤-gal, Escherichia coli ␤-gal gene; EBNA1, EBNA1 gene; polyA, poly(A) additional signal; SR␣,SR␣ promoter; CEA, tetraplicated CEA promoter.

Figure 3. Schematic diagram of HSV-1 Tk expression vectors. Maps of pSES.Tk, pS.Tk, pT.Tk, and pTES.Tk are shown. Amp, ampicillin-resistant gene; EBNA1, EBNA1 gene; polyA, poly(A) additional signal; SR␣,SR␣ promoter; CEA, tetraplicated CEA promoter; Tk, HSV-1 Tk gene.

Cancer Gene Therapy, Vol 7, No 9, 2000 1244 TANAKA, IWAI, HARADA, ET AL: NONVIRAL GENE TRANSFER TO CEA-POSITIVE TUMOR CELLS

Table 1. CEA production ␮ 10,000 M. After further incubation in 5% CO2/95% humid- Cell lines CEA secretion* (mean Ϯ SD) ified air at 37°C for 120 hours, Alamar blue (Alamar Bio- sciences, Sacramento, Calif) was added according to the HuCC-T1 7.0 Ϯ 0.7 manufacturer’s protocol. The cells were further cultured for 4 SSP-25 BT† hours, and the OD of each well was measured with a micro- A4573 BT plate reader at test and reference wavelengths of 570 and 600 nm, respectively. *Nanograms of CEA/mg protein/7 days. †BT, below the threshold of 0.5 ng of CEA. RESULTS ␤ ␤ -galactosidase ( -gal) assay Production of CEA protein by tumor cell lines Cells were scraped off dishes, washed twice with PBS, and In the present study, we employed two CC cell lines, resuspended in 50 ␮L of tris(hydroxymethyl)aminomethane- HCl (pH 7.8). After two cycles of freezing and thawing, the HuCC-T1 and SSP-25, as well as a Ewing’s sarcoma cell lysate was centrifuged at 15,000 rpm for 5 minutes. The ␤-gal line, A4573. Earlier reports have shown that the tran- activity in the supernatant fraction was assayed with a ␤-gal scription level of the CEA gene correlates with the CEA 36–38 assay kit (Invitrogen, San Diego, Calif) according to the protein level and that most of the CEA protein is manufacturer’s protocol. The optical density (OD) was mea- secreted into medium.9 Therefore, we measured the sured at 420 nm. Activity was calculated using the following CEA content in culture media. CEA was detected in the ␤ ϭ ϫ formula: -gal units ([OD420 380]/30)/mg protein, where media of HuCC-T1 cells but not in the media of SSP-25 380 is a conversion factor and 30 is the time of incubation in ␤ and A4573 cells (Table 1). The results are consistent minutes. The -gal activity was standardized based on the with those published earlier.39 amount of protein in the supernatant, as determined with a 33 protein assay kit. Comparison of transfection efficiencies with pSES.␤ and pS.␤ Alamar blue assay The susceptibility of cells to ganciclovir (GCV) was examined To test the effectiveness of the plasmid vectors carrying 34,35 EBNA1 and oriP, we transfected the three cell lines with by Alamar blue assay as described previously. Briefly, ␤ ␤ quadruplicate aliquots of cells were seeded in 96-well, flat- pSES. ,a -gal expression vector containing EBNA1 bottom microtiter plates (Falcon, Lincoln Park, NJ) (5 ϫ 103 and oriP, or pS.␤, a conventional plasmid vector (Fig 2). cells in 200 ␮L of complete medium per well). After 24 hours, A total of 2 ␮g of DNA was combined with 30 ␮gof GCV was added at various concentrations ranging from 0 to PAAD to transfect the cells in each well of a 6-well plate.

Figure 4. ␤-gal expression in tumor cell lines transfected with pSES.␤ or pS.␤ /PAAD. HuCC-T1 (1 ϫ 105 cells), SSP-25 (3 ϫ 105 cells), and A4573 (3 ϫ 105 cells) cells were plated on 6-well plates. After 24 hours, the cells were incubated for 2 hours with 2 ␮g of pSES.␤ (f)orpS.␤ (Ⅺ) combined with 30 ␮g of PAAD. After 3 days, ␤-gal activities were measured. The data represent the mean Ϯ SE of triplicate determinations.

Cancer Gene Therapy, Vol 7, No 9, 2000 TANAKA, IWAI, HARADA, ET AL: NONVIRAL GENE TRANSFER TO CEA-POSITIVE TUMOR CELLS 1245

Figure 5. X-Gal staining of transfected cells. HuCC-T1 (a,d), SSP-25 (b,e), or A4573 (c,f) cells were transfected with pSES.␤/PAAD (a–c) or pTES.␤/PAAD (d–f) and stained with X-Gal on day 3 posttransfection (original magnification is ϫ25).

On day 3 posttransfection, the ␤-gal activities were possesses the ␤-gal gene under the control of the CEA measured. As shown in Figure 4, the ␤-gal activities promoter (pT.␤) was also constructed. provided by pSES.␤ /PAAD were 3.4 to 13.4 times We tested the efficiency and specificity of these plas- higher than those provided by pS.␤ /PAAD in all of the mids by transfecting them via PAAD into HuCC-T1, cell lines. X-Gal staining also revealed that HuCC-T1, SSP-25, and A4573 cells. pT.␤ did not provide high ␤-gal SSP-25, and A4573 cells transfected with pSES.␤ activity in the cell lines tested (Fig 6d). This indicates /PAAD showed high-level ␤-gal expression. (Represen- that a conventional plasmid vector with CEA promoter tative data are shown in Fig 5, a–c.) is insufficient to induce high-level marker gene expres- The data are compatible with our previous studies sion even in CEA-positive cells. When pSET.␤ was demonstrating the superiority of the EBV plasmid to transfected, the ␤-gal activity in HuCC-T1 was about conventional plasmid vector in various tumor as well as ␤ 23,31,32 one-seventh of that given by pSES. (Fig 6a). A consid- primary cells. erable level of ␤-gal activity was also detected in A4573, ␤ ␤ indicating that pSET. was lacking not only in activity Cell type-specific expression of the -gal gene but also in specificity. The ␤-gal activities provided by ␤ Because pSES.␤ has a strong universal promoter (SR␣), pTET. were very weak in both CEA-positive and ␤-gal was strongly expressed in both CEA-positive and -negative cell lines (Fig 6c). In contrast, when pTES.␤ -negative cell lines transfected with pSES.␤. To target was transfected, both SSP-25 and A4573 cells exhibited CEA-producing cells, we constructed three EBV-based very low levels of ␤-gal activity, whereas HuCC-T1 had plasmids, pTES.␤, pSET.␤, and pTET.␤, that contain strong enzyme activity (Fig 6b). The X-Gal staining also the CEA promoter upstream of the EBNA1 gene, ␤-gal confirmed the strong expression of ␤-gal exclusively in gene, or both (Fig 2). A conventional plasmid vector that the CEA-producing tumor cells (Fig 5, d-f). These

Cancer Gene Therapy, Vol 7, No 9, 2000 1246 TANAKA, IWAI, HARADA, ET AL: NONVIRAL GENE TRANSFER TO CEA-POSITIVE TUMOR CELLS

Figure 7. Kinetics of ␤-gal expression given by pTES.␤. HuCC-T1 cells (1 ϫ 105 cells) were transfected with pTES.␤ as described in the legend to Figure 4. ␤-gal activities were measured on the indicated days after the transfection. Means Ϯ SE of triplicate determinations are shown.

PAAD. At 72 hours posttransfection, cells were cultured in the presence of various concentrations of GCV, and 5 days later, their viabilities were measured by Alamar blue assay. Both cell lines transfected with pSES.Tk were rendered highly susceptible to GCV compared with mock-transfected cells (Fig 8, a and e). The transfection with pS.Tk moderately affected the susceptibilities of these cell lines to GCV (Fig 8b). These observations with pSES.Tk and pS.Tk are consistent with our previ- ous results using HCC and Ewing’s sarcoma cell lines.31,32 The pT.Tk/PAAD did not significantly affect the susceptibility of either cell line to GCV; the susceptibil- ity curves were similar to those of mock-transfected cells (Fig 8, c and e). This observation is compatible with the results for pT.␤ (Fig 6d). Compared with pS.Tk, the Figure 6. ␤-gal is specifically expressed in CEA-positive cells upon suicide effect provided by pT.Tk was weaker, indicating ␤ ϫ 5 f ϫ 5 pTES. transfection. HuCC-T1 (1 10 cells) ( ), SSP-25 (3 10 that the promoter activity of the CEA promoter is cells) (u), or A4573 (3 ϫ 105 cells) (Ⅺ) cells were transfected with ␣ pSET.␤ (a), pTES.␤ (b), pTET.␤ (c),orpT.␤ (d) as described in the weaker than that of SR (Figs 4 and 6). Therefore, a legend to Figure 4. After 3 days, ␤-gal activities were measured. therapeutic effect cannot be expected with this plasmid Means Ϯ SE of triplicate determinations are shown. vector. In marked contrast, the GCV sensitivity of HuCC-T1 was sufficiently elevated by pTES.Tk, whereas that of SSP-25 was not altered by the same plasmid (Fig 8d). The percent cell survival curve of pTES.Tk-trans- results indicate the usefulness of pTES.␤ as a plasmid fected SSP-25 cells was similar to that of the mock- for tumor-specific transfection. transfected cells (Fig 8, d and e), whereas HuCC-T1 cells We examined the kinetics of the ␤-gal activity in transfected with pTES.Tk exhibited nearly the same HuCC-T1 cells provided by pTES.␤ transfection. Strong percent cell survival curve as pSES.Tk-transfected ␤-gal activity was observed from 2 to 7 days after HuCC-T1 cells (Fig 8, a and d). This finding indicates transfection (Fig 7). After day 7, the expression gradu- that pTES.Tk specifically affects CEA-positive cells, ally diminished, but even on day 21 posttransfection, leaving the CEA-negative cells unaffected. ␤-gal activity was detectable. In the aspects of clinical usage, the killing effects of these vectors were tested by culturing cells in the presence of 10 Targeted killing of CEA-producing cells by pTES.Tk and 30 ␮M GCV after the transfection. It has been To investigate whether CEA-positive tumor cells can be reported that patients systemically administered this pro- specifically killed by a nonviral vector system, four drug exhibit a serum concentration of 2.7 to 22.3 ␮M.40 In plasmid vectors carrying the herpes simplex virus-1 the presence of 10 ␮M GCV, HuCC-T1 cells transfected (HSV-1) thymidine kinase (Tk) gene were constructed with pTES.Tk showed a marked reduction in viability, (pSES.Tk, pS.Tk, pT.Tk, pTES.Tk) (Fig 3). They were whereas SSP-25 cells transfected with the same plasmid transfected into HuCC-T1 or SSP-25 cells by means of remained viable (data not shown). Even in the presence of

Cancer Gene Therapy, Vol 7, No 9, 2000 TANAKA, IWAI, HARADA, ET AL: NONVIRAL GENE TRANSFER TO CEA-POSITIVE TUMOR CELLS 1247

Figure 8. Targeted killing of CEA-positive cells by pTES.Tk. HuCC-T1 (f) or SSP-25 (Ⅺ) cells were transfected with pSES.Tk (a), pS.Tk (b), pT.Tk (c), or pTES.Tk (d) as described in the legend to Figure 4. Mock-transfected cells (e) were treated with PAAD alone. After 2 days, 5 ϫ 103 cells were seeded into the wells of 96-well plates. On the following day, various concentrations of GCV were added to the cells. Five days later, the viabilities of the cells were assessed by Alamar blue assay. Each point represents the mean Ϯ SE of quadruplicate determinations.

30 ␮M GCV, which is beyond the peak serum level, the CEA-positive or -negative cells. The results strongly sug- viability of the SSP-25 cells transfected with pTES.Tk was gest that pTES.Tk but not pT.Tk can elicit significant preserved (Fig 9), indicating CEA-negative cells were safe. therapeutic effects. Transfection with pT.Tk did not affect the viability of either DISCUSSION

In the present study, we have established an in vitro model system for targeted suicide gene therapy of CEA-producing cancer by using EBV/polyplex. Trans- fection with a non-EBV, conventional plasmid vector carrying the same promoter did not result in significant marker gene expression or effective killing of the CEA- positive cells. Therefore, the present results demonstrate the usefulness of EBV-based but not conventional plas- mid vectors in targeted suicide gene therapy for cancer. We and others have reported that plasmid vectors carrying oriP but not EBNA1 were effective in vitro in targeting EBV-associated lymphoma cells.20,41–43 These particular tumor cells express endogenous EBNA1, Figure 9. Cell viability on day 5 of culture with a clinical dose of which allows very strong expression when an oriP- GCV. HuCC-T1 (left) or SSP-25 (right) cells were transfected with the indicated plasmids as described in the legend to Figure 8. Two bearing plasmid is introduced. Because most normal days later, 5 ϫ 103 cells were seeded into the wells of a 96-well cells do not express EBNA1, this strategy elicits specific plate. On the following day, GCV was added to a final concentration killing of EBNA1-positive tumor cells. However, cancers of 30 ␮M, and 5 days later, the viability of the cells was evaluated by not latently infected with EBV cannot be targeted by this Alamar blue assay. Each point represents the mean Ϯ SE of method. To transfer the suicide gene into EBNA1- quadruplicate determinations. negative cancer cells, we used an EBV-vector carrying

Cancer Gene Therapy, Vol 7, No 9, 2000 1248 TANAKA, IWAI, HARADA, ET AL: NONVIRAL GENE TRANSFER TO CEA-POSITIVE TUMOR CELLS both oriP and EBNA1 and succeeded in effectively studies are now underway to improve the vector system killing the cancer cells both in vitro31 and in vivo.32 In in terms of specificity and effectiveness and to evaluate previous studies, we employed a ubiquitous promoter the in vivo efficacy of the system. and did not aim at targeting tumor cells. Therefore, the present study is the first in which an EBV-negative tumor was targeted by an EBV-based plasmid vector ACKNOWLEDGMENTS equipped with a tumor-specific promoter. To our knowledge, there have been two earlier studies in We thank Dr. M. Enjoji for cell lines. We also thank F. which targeted gene transfer into CEA-positive tumors was Hoffman-La Roche (Basel, Switzerland) for kindly providing attempted by transfection with conventional plasmid vec- GCV. This research was supported by a grant-in-aid for tors via nonviral methods. Fichera et al17 used liposome to scientific research (No. 11670523) from the Ministry of Edu- cation, Japan. deliver a plasmid vector in which the 490-bp sequence corresponding to the human CEA promoter was included 18 to regulate the luciferase gene as a marker. Kurane et al REFERENCES used a plasmid vector containing a chimeric promoter consisting of the CEA promoter (a 424-bp fragment up- 1. Chen MF, Jan YY, Wang CS, et al. Clinical experience in stream of the translational start site originally reported by 14 20 hepatic resections for peripheral cholangiocarcinoma. Schrewe et al ) and an enhancer from the immediate early Cancer. 1989;64:2226–2232. gene of cytomegalovirus. Antibody against the Lewis Y 2. Schlinkert RT, Nagorney DM, Van Heerden JA, et al. antigen was also employed to improve the specificity of the Intrahepatic cholangiocarcinoma: clinical aspects, pathol- vector.18 In both studies, however, only marker genes were ogy, and treatment. HPB Surg. 1992;5:95–101. transferred. Although specific expression was achieved in 3. Nakeeb A, Pitt HA, Sohn TA, et al. Cholangiocarcinoma: CEA-positive cells, the expression level was very low, with a spectrum of intrahepatic, perihilar, and distal tumors. the duration of gene expression being short. No therapeu- Ann Surg. 1996;224:463–475. tic effect was shown, probably due to the low level of 4. Kaneko S, Hallenbeck P, Kotani T, et al. Adenovirus- mediated gene therapy of hepatocellular carcinoma using expression obtained with the non-EBV-type plasmid vec- cancer-specific gene expression. Cancer Res. 1995;55:5283– tors. 5287. PAAD is a positively charged synthetic polymer cre- 5. Ido A, Nakata K, Kato Y, et al. Gene therapy for 44–47 ated recently. It is a macromolecule that is spherical hepatoma cells using a retrovirus vector carrying herpes in shape and composed of repeating polyamidoamino simplex virus thymidine kinase gene under the control of subunits. It has been shown that PAAD is an efficient human ␣-fetoprotein gene promoter. Cancer Res. 1995;55: vehicle to transduce genetic materials into mammalian 3105–3109. 6. Kanai F, Shiratori Y, Yoshida Y, et al. Gene therapy for cells. We have reported recently that by using PAAD, ␣ extremely efficient gene transfer can be achieved in -fetoprotein-producing human hepatoma cells by adeno- tumor cells both in vitro31,32 and in vivo.32 Therefore, the virus-mediated transfer of the herpes simplex virus thymi- dine kinase gene. Hepatology. 1996;23:1359–1368. EBV-plasmid vector/dendrimer complexes constructed 7. Bui LA, Butterfield LH, Kim JY, et al. In vivo therapy of in the present study may also be applied to experimental hepatocellular carcinoma with a tumor-specific adenoviral animal models, taking advantage of the in vivo effective- vector expressing interleukin-2. Hum Gene Ther. 1997;8: ness of PAAD. 2173–2182. Transfection with pTET.␤ resulted in very poor marker 8. Gold P, Freedman SO. Specific carcinoembryonic antigens of gene expression in both CEA-positive and -negative cell the human digestive system. J Exp Med. 1965;122:467–481. lines. This may be ascribed to the relatively weak activity of 9. Richards CA, Austin EA, Huber BE. Transcriptional the CEA promoter even in CEA-positive cells.13,48 pTES.␤ regulatory sequences of carcinoembryonic antigen: identi- could be transduced effectively and exclusively into CEA- fication and use with cytosine deaminase for tumor-specific positive cells, whereas pSET.␤ showed less selectivity. The gene therapy. Hum Gene Ther. 1995;6:881–893. 10. Osaki T, Tanio Y, Tachibana I, et al. Gene therapy for results suggest that the expression level of the EBNA1 carcinoembryonic antigen-producing human lung cancer gene is critically important to the expression of the trans- cells by cell type-specific expression of herpes simplex virus gene. If a small amount of EBNA1 is first expressed in the thymidine kinase gene. Cancer Res. 1994;54:5258–5261. transfected cells, the molecule may induce the expression 11. DiMaio JM, Clary BM, Via DF, et al. Directed enzyme of not only marker but also secondary EBNA1 molecules, pro-drug gene therapy for pancreatic cancer in vivo. which in turn induce further stronger expression of both Surgery. 1994;116:205–213. genes. In this way, the EBNA1 gene driven by a weak 12. Lan KH, Kanai F, Shiratori Y, et al. In vivo selective gene promoter may elicit an intensive expression of the trans- expression and therapy mediated by adenoviral vectors for gene, particularly if the marker gene is driven by a strong human carcinoembryonic antigen-producing gastric carci- promoter. noma. Cancer Res. 1997;57:4279–4284. 13. Tanaka T, Kanai F, Lan KH, et al. Adenovirus-mediated In conclusion, targeted gene transfer was achieved by gene therapy of gastric carcinoma using cancer-specific using an EBV-based plasmid vector/PAAD equipped gene expression in vivo. Biochem Biophys Res Commun. with a tumor-specific promoter. The system may be 1997;231:775–779. applicable not only to CC but to other CEA-producing 14. Schrewe H, Thompson J, Bona M, et al. Cloning of the malignancies, such as gastric and colon cancers. Further complete gene for carcinoembryonic antigen: analysis of

Cancer Gene Therapy, Vol 7, No 9, 2000 TANAKA, IWAI, HARADA, ET AL: NONVIRAL GENE TRANSFER TO CEA-POSITIVE TUMOR CELLS 1249

its promoter indicates a region conveying cell type-specific gene expression in hepatocellular carcinoma cells by Epstein- expression. Mol Cell Biol. 1990;10:2738–2748. Barr virus-based plasmid vectors combined with poly- 15. Cao G, Kuriyama S, Gao J, et al. In vivo gene transfer of amidoamine dendrimer. Cancer Gene Ther. 2000;7:27–36. a suicide gene under the transcriptional control of the 32. Maruyama-Tabata H, Harada Y, Matsumura T, et al. carcinoembryonic antigen promoter results in bone mar- Effective suicide gene therapy in vivo by EBV-based row transduction but can avoid bone marrow suppression. plasmid vector coupled with polyamidoamine dendrimer. Int J Oncol. 1999;15:107–112. Gene Ther. 2000;7:53–60. 16. Tanaka T, Kanai F, Okabe S, et al. Adenovirus-mediated 33. Bradford MM. A rapid and sensitive method for the prodrug gene therapy for carcinoembryonic antigen-pro- quantitation of microgram quantities of protein utilizing ducing human gastric carcinoma cells in vitro. Cancer Res. the principle of protein-dye binding. Anal Biochem. 1976; 1996;56:1341–1345. 72:248–254. 17. Fichera A, Michelassi F, Arenas RB. Selective expression 34. Ahmed SA, Gogal RM Jr, Walsh JE. A new rapid and of carcinoembryonic antigen promoter in cancer cell lines: simple non-radioactive assay to monitor and determine the targeting strategy for gene therapy in colorectal cancer. Dis proliferation of lymphocytes: an alternative to [3H]thymi- Colon Rectum. 1998;41:747–774. dine incorporation assay. J Immunol Methods. 1994;170: 18. Kurane S, Krauss JC, Watari E, et al. Targeted gene 211–224. transfer for adenocarcinoma using a combination of tu- 35. Hirai H, Shimazaki C, Yamagata N, et al. Effects of throm- mor-specific antibody and tissue-specific promoter. Jpn J bopoietin (c-mpl ligand) on growth of blast cells from pa- Cancer Res. 1998;89:1212–1219. tients with transient abnormal myelopoiesis and acute myelo- 19. Mazda O, Satoh E, Yasutomi K, et al. Extremely efficient blastic leukemia. Eur J Haematol. 1997;59:38–46. gene transfection into lympho-hematopoietic cell lines by 36. Abbasi AM, Chester KA, MacPherson AJ, et al. Localiza- Epstein-Barr virus-based vectors. J Immunol Methods. tion of CEA messenger RNA by in situ hybridization in 1997;204:143–151. normal colonic mucosa and colorectal adenocarcinomas. 20. Hirai H, Satoh E, Osawa M, et al. Use of EBV-based J Pathol. 1992;168:405–411. vector/HVJ-liposome complex vector for targeted gene 37. Kim J, Kaye FJ, Henslee JG, et al. Expression of carcino- therapy of EBV-associated neoplasms. Biochem Biophys embryonic antigen and related genes in lung and gastro- Res Commun. 1997;241:112–118. intestinal cancers. Int J Cancer. 1992;52:718–725. 21. Satoh E, Osawa M, Tomiyasu K, et al. Efficient gene 38. Jothy S, Yuan SY, Shirota K. Transcription of carcinoem- transduction by Epstein-Barr-virus-based vectors coupled bryonic antigen in normal colon and colon carcinoma; in with cationic liposome and HVJ-liposome. Biochem Bio- situ hybridization study and implication for a new in vivo phys Res Commun. 1997;238:795–799. functional model. Am J Pathol. 1993;143:250–257. 22. Satoh E, Hirai H, Inaba T, et al. Successful transfer of 39. Miyagiwa M, Ichida T, Tokiwa T, et al. A new human ϩ ADA gene in vitro into human peripheral blood CD34 cholangiocellular carcinoma cell line (HuCC-T1) produc- cells by transfecting EBV-based episomal vectors. FEBS ing carbohydrate antigen 19/9 in serum-free medium. In Lett. 1998;441:39–42. Vitro Cell Dev Biol. 1989;25:503–510. 23. Tomiyasu K, Satoh E, Oda Y, et al. Gene transfer in vitro 40. Sommadossi J-P, Bevan R, Ling T, et al. Clinical pharmaco- and in vivo with Epstein-Barr virus-based episomal vector kinetics of ganciclovir in patients with normal and impaired results in markedly high transient expression in rodent renal function. Rev Infect Dis. 1988;10:S507–S514. cells. Biochem Biophys Res Commun. 1998;253:733–738. 41. Judde JG, Spangler G, Magrath I, et al. Use of Epstein- 24. Sugden B, Marsh K, Yates J. A vector that replicates as a Barr virus nuclear antigen-1 in targeted therapy of EBV- plasmid and can be efficiently selected in B-lymphoblasts associated neoplasia. Hum Gene Ther. 1996;7:647–653. transformed by Epstein-Barr virus. Mol Cell Biol. 1985;5: 42. Evans TJ, Brooks L, Farrell PJ. A strategy for specific 410–413. targeting of therapeutic agents to tumor cells of virus- 25. Yates JL, Warren N, Sugden B. Stable replication of associated cancers. Gene Ther. 1997;4:264–267. plasmids derived from Epstein-Barr virus in various mam- 43. Kenney S, Ge JQ, Westphal EM, et al. Gene therapy malian cells. Nature. 1985;313:812–815. strategies for treating Epstein-Barr virus-associated lym- 26. Rawlins DR, Milman G, Hayward SD, et al. Sequence- phomas: comparison of two different Epstein-Barr virus- specific DNA binding of the Epstein-Barr virus nuclear based vectors. Hum Gene Ther. 1998;9:1131–1141. antigen (EBNA-1) to clustered sites in the plasmid main- 44. Haensler J, Szoka FC Jr. Polyamidoamine cascade poly- tenance region. Cell. 1985;42:859–868. mers mediate efficient transfection of cells in culture. 27. Middleton T, Sugden B. Retention of plasmid DNA in Bioconjug Chem. 1993;4:372–379. mammalian cells is enhanced by binding of the Epstein- 45. Kukowska-Latallo JF, Bielinska AU, Johnson J, et al. Barr virus replication protein EBNA1. J Virol. 1994;68: Efficient transfer of genetic material into mammalian cells 4067–4071. using Starburst polyamidoamine dendrimers. Proc Natl 28. Gahn TA, Sugden B. An EBNA-1-dependent enhancer Acad Sci USA. 1996;93:4897–4902. acts from a distance of 10 kilobase pairs to increase 46. Tang MX, Redemann CT, Szoka FC Jr. In vitro gene expression of the Epstein-Barr virus LMP gene. J Virol. delivery by degraded polyamidoamine dendrimers. Bio- 1995;69:2633–2636. conjug Chem. 1996;7:703–714. 29. Kirchmaier AL, Sugden B. Plasmid maintenance of deriva- 47. Tang MX, Szoka FC Jr. The influence of polymer structure tives of oriP of Epstein-Barr virus. J Virol. 1995;69:1280– on the interactions of cationic polymers with DNA and 1283. morphology of the resulting complexes. Gene Ther. 1997; 30. Jankelevich S, Kolman JL, Bodnar JW, et al. A nuclear 4:823–832. matrix attachment region organizes the Epstein-Barr viral 48. Lan KH, Kanai F, Shiratori Y, et al. Tumor-specific gene plasmid in Raji cells into a single DNA domain. EMBO J. expression in carcinoembryonic antigen-producing gastric 1992;11:1165–1176. cancer cells using adenovirus vectors. Gastroenterology. 31. Harada Y, Iwai M, Tanaka S, et al. Highly efficient suicide 1996;111:1241–1251.

Cancer Gene Therapy, Vol 7, No 9, 2000