Effective and safe gene therapy for colorectal carcinoma ... - Nature

Gene Therapy (1999) 6, 83–90  1999 Stockton Press All rights reserved 0969-7128/99 $12.00 http://www.stockton-press.co.uk/gt

Effective and safe gene therapy for colorectal carcinoma using the cytosine deaminase gene directed by the carcinoembryonic antigen promoter G Cao1, S Kuriyama2, J Gao1, M Kikukawa2, L Cui1, T Nakatani2, X Zhang1, H Tsujinoue2, X Pan1, H Fukui2 and Z Qi1 1

Department of Microbiology, Second Military Medical University, Shanghai, China; and 2Third Department of Internal Medicine, Nara Medical University, Kashihara, Japan

We have recently isolated carcinoembryonic antigen (CEA) promoter regions consisting of 419 bp and 204 bp from CEA-producing human colorectal carcinoma (CRC). We constructed CEA419/CD and CEA204/CD retroviruses carrying the bacterial cytosine deaminase (CD) gene directed by the CEA promoter regions. pCD2 retroviruses carrying the CD gene directed by the retrovirus long terminal repeat promoter were also used. CEA419/CD or CEA204/CD retrovirus-infected CRC cells were found to be susceptible to 5-fluorocytosine (5-FC), while non-CRC cells infected with the same retroviruses were not. CD-transduced CRC xenografts in nude mice were sensitive to 5FC treatment, resulting in arrest of tumor growth. When mice with intraperitoneally disseminated CRCs were given

intraperitoneal injections of CEA419/CD retrovirus-producing cells followed by 5-FC treatment, significantly prolonged survival rates were observed compared with animals injected with pCD2 retrovirus-producing cells followed by 5-FC treatment. Importantly, bone marrow suppression was not observed in animals injected with CEA419/CD retrovirus-producing cells and 5-FC, while profound bone marrow suppression was observed in those injected with pCD2 retrovirus-producing cells and 5-FC. These results indicate that effective and safe in vivo gene therapy for advanced CRC may be feasible by transferring the CD gene controlled by the CEA promoter followed by 5-FC treatment.

Keywords: colorectal carcinoma; carcinoembryonic antigen promoter; cytosine deaminase; in vivo gene therapy; retrovirus; bone marrow suppression

Introduction Colorectal carcinoma (CRC) is the second leading cause of malignancy in Western countries, accounting for 151 000 new cases and 61 000 deaths annually in the USA.1 In Japan its incidence has been increasing rapidly over the last 20 years and it is now the most common malignancy, after gastric cancer, accounting for 30 000 deaths per year. In spite of intensive efforts, there is still no satisfactory treatment that significantly improves the overall survival rate of patients with CRC.2 A significant number of patients die from recurrent disease, with the liver being the most common site for metastases. Peritoneal metastasis is the next most frequent type of recurrence and also a serious obstacle to the curative treatment of CRC.3,4 Gene therapy could provide an innovative therapeutic approach for the treatment of CRC. Among various strategies for gene therapy against cancer, the delivery of a suicide gene into tumor cells followed by the corresponding prodrug treatment, such as the combination of Escherichia coli cytosine deaminase (CD) gene transduction and Correspondence: S Kuriyama, Third Department of Internal Medicine, Nara Medical University, 840 Shijo-cho, Kashihara, Nara 634–8522, Japan Received 20 May 1998; accepted 12 August 1998

5-fluorocytosine (5-FC) treatment, is one of the most promising approaches. CD is an enzyme found in many bacteria and fungi which deaminates cytosine to uracil, while normal mammalian cells do not possess CD. It also deaminates the relatively nontoxic 5-FC to the highly toxic 5-fluorouracil (5-FU). 5-FU has been used most commonly for the treatment of CRC, alone or in combination with other chemotherapeutic agents. It causes cell death by inhibiting both DNA and RNA syntheses.5 However, its effectiveness is limited by systemic toxicity specifically associated with high doses. In marked contrast to 5-FU, which has a narrow therapeutic index, 5-FC is not toxic to humans at therapeutic doses.6 It is, therefore, desirable to produce a high local concentration of 5-FU at the tumor site. Localized in vivo generation of 5-FU as a novel strategy for antitumor therapy was reported by Nishiyama et al.7 They implanted capsules containing CD into subcutaneous tumors growing in rats. This was followed by systemic administration of 5-FC. Considerable antitumor activity without observable toxic effects was evident using this approach. With the advancement of recombinant DNA techniques, a similar strategy was recently implemented using a genetic approach. Mullen et al8 and Huber et al9 genetically modified tumor cells to express CD. They implanted the modified cells into mice and demonstrated local 5-FU generation and antitumor effects. Expression of the CD gene has been achieved

Gene therapy for CRC using CEA promoter G Cao et al

84

recently in various cell lines, such as fibrosarcoma, adenocarcinoma including CRC lines, gliosarcoma and hepatocellular carcinoma.10–20 Those genetically modified cells were selectively sensitive to 5-FC compared with unmodified cells. Therefore, transfer of the CD gene followed by 5-FC treatment could be a promising approach for various types of cancer, if tumor cell-selective transduction or expression of the CD gene could be feasible in vivo. CEA is a Mr 180 000 cell surface glycoprotein originally described by Gold and Freedman21 as an antigenic component in cancers derived from gastrointestinal tract epithelium. Although later studies have shown that CEA is also expressed on the apical surface of epithelial cells in the normal colon,22,23 levels of its expression are usually much higher in malignant colon and other cancers of epithelial cell origin compared with normal adult colon. Therefore, by coupling the CEA promoter with the CD gene, CD gene expression is expected to be restricted to tumor cells that produce high levels of CEA.16 Consequently, in the presence of 5-FC, a high local concentration of 5-FU can be produced in the CEA-producing tumor to kill tumor cells, while sparing the surrounding normal tissues. We have recently isolated CEA promoter regions from high CEA-producing human CRC.24 In the present study, we have constructed retrovirus vectors carrying the bacterial CD gene under the transcriptional control of the CEA promoter regions. This study examined whether infection of retroviruses carrying the CD gene under the control of the CEA promoter regions can cause killing of CEA-producing tumor cells without affecting CEA-nonproducing cells in the presence of 5-FC. Furthermore, to examine the antitumor effect against advanced CRC, athymic nude mice were inoculated intraperitoneally with CEA-producing human CRC cells and retrovirusproducing cells. Using this animal model, the efficacy and safety of using the Moloney murine leukemia virus long terminal repeat promoter and CEA promoter to direct CD gene expression were examined.

CEA-negative cell lines. Previous studies have shown a correlation between CEA protein levels and CEA transcription.25–27 It has been also shown that most of the CEA protein is secreted into the medium.16 CEA protein concentrations of media were, therefore, quantified by using immunoassay to evaluate transcriptional levels of the CEA gene in the cells used in experiments. As shown in Table 1, human CRC LoVo and SW1463 cells were shown to produce considerably high levels of the CEA protein. Conversely, human renal cell carcinoma RC9406 cells, human cervix carcinoma HeLa cells and human hepatocellular carcinoma SMMC7721 cells were shown not to produce detectable levels of the CEA protein.

Sensitivity to 5-FC of cells transduced with the CD gene CEA promoter regions isolated from high CEA-producing human colorectal carcinoma, located from −309 to +110 bp and from −135 to +69 bp upstream from the transcriptional start site of the CEA gene, were employed to direct the CD gene expression in a retrovirus vector (Figure 1). Before estimating the sensitivity to 5-FC, the sensitivity to 5-FU of the cell lines was assessed. As shown in Table 1, all parental cell lines were found to be sensitive to 5-FU with values of IC50, defined as the dose required for 50% cytotoxicity, being 1.1 to 3.5 ␮m. The sensitivity of CD gene-transduced and parental cells to 5-FC was then examined by culturing cells in the presence of various concentrations of 5-FC for 4 days. As shown in Table 1, all pCD2 retrovirus-infected cell lines were found to be much more sensitive to 5-FC compared with the corresponding parental cells and IC50 values to 5-FC were approximately 1/21 to 1/34. CEA419/CD or CEA204/CD retrovirus-infected CRC cells, LoVo and SW1463, were also shown to be much more susceptible to 5-FC compared with the corresponding parental cells and exhibited 20- to 27-fold higher sensitivity to 5-FC. Conversely, CEA419/CD or CEA204/CD retrovirusinfected non-CRC cells, RC9406, HeLa and SMMC7721, exhibited slightly higher sensitivity to 5-FC compared with the corresponding parental cells, and there was a small shift in IC50 to 5-FC. These results indicate that transduction of the CD gene directed by the CEA promoter can exclusively eliminate CEA-producing cells in the presence of 5-FC. Ability to cause killing of CEA-producing cells was not significantly different between the CEA promoter regions located from −309 to +110 bp and

Results CEA production ability of tumor cells To estimate the specificity of the CEA promoter regions, it is necessary to identify and rank CEA-positive and

Table 1 CEA protein expression and 5-FU sensitivity of parental cells and 5-FC sensitivity of parental and retrovirus-infected cells Cell lines

LoVo SW1463 RC9406 HeLa SMMC7721 a

CEA amounta

813 ± 171 691 ± 92 BTb BT BT

IC50 to 5-FCd (␮m)

IC50 to 5-FUc (␮m)

1.1 1.4 1.9 2.1 3.5

Parental

pCD2

7800 4642 11200 8634 13240

368 186 340 368 389

CEA419/CD CEA204/CD

Nanograms of CEA per mg protein per 7 days. BT, below threshold of 0.5 ng of CEA. Results are means of three separate experiments. d Results are means of four separate experiments. e Values are significantly different at P ⬍ 0.01 using Student’s t test. b c

Ratio of IC50

387 172 10640e 8024 11024

342 224 7420e 6804 9341

Parental:CEA419/CD Parental:CEA204/CD 20.2 27.0 1.1 1.1 1.2

22.8 20.7 1.5 1.3 1.4


85

Figure 1 Structure of retrovirus vectors. pCD2 retrovirus vector is composed of the Moloney murine sarcoma virus long terminal repeat (5′ LTR), the bacterial CD gene, simian virus 40 early region promoter (SV40) as an internal promoter, the neomycin phosphotransferase (neo) gene which confers G418 resistance on transfected cells and the Moloney murine leukemia virus long terminal repeat (3′ LTR). CEA419/CD and CEA204/CD retrovirus vectors contain the CEA promoter regions consisting of 419 bp and 204 bp, respectively, isolated from a high CEA-producing human CRC as internal promoters to direct the expression of the bacterial CD gene. ⌿+ indicates the extended retrovirus packaging signal. Arrows indicate transcriptional initiation sites and direction of transcription.

from −135 to +69 bp upstream from the transcriptional start site of the CEA gene. However, the IC50 value to 5FC for CEA419/CD retrovirus-infected RC9406 cells was significantly higher than that for CEA204/CD retrovirusinfected ones. Furthermore, 5-FC IC50 values for CEA419/CD retrovirus-infected HeLa and SMMC7721 cells were also higher than those for the corresponding CEA204/CD retrovirus-infected ones, although the differences were not statistically significant at P = 0.052 and 0.053, respectively, using Student’s t test. These results suggest that the CEA promoter region located from −309 to +110 bp upstream from the transcriptional start site can induce CEA-producing cell-selective expression of exogenous genes more stringently compared with the CEA promoter region located from −135 to +69 bp upstream from the transcriptional start site.

In vivo sensitivity of CD gene-transduced CRC to 5-FC To examine in vivo sensitivity of CD gene-transduced CRC to 5-FC, CRC xenograft experiments were performed. Parental, CEA204/CD retrovirus-infected or pCD2 retrovirus-infected LoVo cells were inoculated subcutaneously into the flank regions of athymic BALB/cnu/nu mice. Twenty-five days after inoculation, mice received intraperitoneal injections of 5-FC or PBS for 7 consecutive days. As shown in Figure 2, there were no significant differences in tumor volume among the groups at the time of initiation of 5-FC or PBS treatment, and mean tumor volume of each group was between 88 mm3 and 102 mm3. Mice inoculated with parental LoVo cells and treated with 5-FC developed rapidly growing tumors. Similarly, mice inoculated with CEA204/CD retrovirus-infected LoVo cells and treated with PBS also developed rapidly growing tumors. All animals in these groups died or were killed due to tumor overloading within 60 days after the inoculation. Conversely, tumors of the mice inoculated with CEA204/CD or pCD2 retrovirus-infected cells followed by 5-FC treatment grew slowly until day 40, and then did not grow significantly or were even reduced. All animals in these two groups stayed alive over a 60-day observation per-

Figure 2 Effect of 5-FC treatment on LoVo CRC xenografts. Parental, pCD2 retrovirus-infected and CEA204/CD retrovirus-infected LoVo cells (2 × 106 cells per mouse) were inoculated into the flank regions of athymic BALB/c-nu/nu mice. Twenty-five days later, animals inoculated with parental (open square, n = 5), or pCD2 retrovirus-infected (open circle, n = 5) LoVo cells were dosed intraperitoneally with 700 mg/kg 5-FC from day 25 to day 31, as indicated by the bar. Animals inoculated with CEA204/CD retrovirus-infected LoVo cells were dosed intraperitoneally with 5-FC (closed circle, n = 5), or PBS (closed square, n = 5) from day 25 to day 31. Each data point represents the mean ± s.d. of five animals. Data are not shown in the Figure after any of animals in the group died.

iod. After the end of the observation period, all animals inoculated with CEA204/CD or pCD2 retrovirus-infected LoVo cells followed by 5-FC treatment were killed and subcutaneous tumors were resected for routine histology. Consistent with the previous report,12 significant necrotic and fibrotic areas with infiltration of fibrocytes and lymphocytes were evident in all the resected tumors. There were no apparent differences between 5-FC-treated tumors consisting of CEA204/CD and pCD2 retrovirusinfected LoVo cells (data not shown).

Therapeutic effect on disseminated CRC by in vivo gene therapy To investigate the efficacy of in vivo gene therapy using the CEA promoter regions to direct the CD gene on established tumors, an animal model with intraperitoneally disseminated LoVo tumors that closely resemble human advanced CRC was used. Athymic nude mice received an intraperitoneal inoculation of LoVo cells. PA317 retrovirus-packaging cells, CEA419/CD retrovirus-producing cells or pCD2 retrovirus-producing cells were injected intraperitoneally at days 3, 5 and 7 after the inoculation of LoVo cells. 5-FC, 5-FU or PBS was administered intraperitoneally from day 10 to day 16. Figure 3 shows the survival rates of animals after the inoculation of LoVo cells. All the mice that received intraperitoneal injections of 5-FU without receiving injections of retrovirus-producing cells died within 18 days. All the mice injected with PA317 retrovirus-packaging cells followed by 5-FC treatment and all the mice injected with CEA419/CD retrovirus-producing cells followed by PBS treatment died within 20 and 25 days, respectively. Conversely, the survival rate of mice injected with pCD2 retrovirus-producing cells followed by 5-FC treatment was significantly higher compared with those mice in the above-men-


86

Figure 3 Effect of in vivo gene therapy on disseminated CRCs. Athymic nude mice received an intraperitoneal injection of parental LoVo cells (2 × 106 cells per mouse) and divided randomly 3 days later into the following five groups according to treatment schedules: (a) injection of PA317 retrovirus-packaging cells followed by 5-FC treatment (closed square, n = 10); (b) injection of pCD2 retrovirus-producing cells followed by 5-FC treatment (open square, n = 10); (c) injection of CEA419/CD retrovirus-producing cells followed by 5-FC treatment (open circle, n = 10); (d) injection of CEA419/CD retrovirus-producing cells followed by PBS treatment (closed circle, n = 10); (e) injection of the serum-free medium followed by 5-FU treatment (closed triangle, n = 10). The asterisk means that the survival rate of animals in the group (c) is significantly higher than that of animals in the group (b) at P ⬍ 0.01, using ␹2 test.

tioned groups, and the last mouse died 50 days after inoculation with LoVo cells. Furthermore, the survival rate of mice injected with CEA419/CD retrovirus-producing cells followed by 5-FC treatment was significantly higher than that of mice injected with pCD2 retrovirusproducing cells followed by 5-FC treatment, and 70% of the mice were still alive even 50 days after inoculation with LoVo cells.

Treatment-related bone marrow suppression To examine the treatment-related cytotoxicity on the bone marrow, athymic nude mice were inoculated intraperitoneally with LoVo cells and received the treatment as described above. Animals were killed 17 days after inoculation with LoVo cells and bone marrow tissues from femora and sterna were examined. To examine the treatment-related bone marrow suppression, ratios of karyocytes:mature erythrocytes (K:E) in the bone marrow were estimated. Naive BALB/c-nu/nu mice have relatively high K:E ratios compared with humans, and their K:E ratios generally range from 60% to 70%. As shown in Figure 4, K:E ratios of the mice injected with CEA419/CD retrovirus-producing cells followed by PBS treatment were not affected, while those of the mice treated with 5-FU were profoundly suppressed, with mean K:E ratios in the bone marrow of femora and sterna being 3% and 2%, respectively. Significant bone marrow suppression was not observed in the mice injected with CEA419/CD retrovirus-producing cells followed by 5-FC treatment, with mean K:E ratios in the bone marrow of femora and sterna being 62% and 57%, respectively. Conversely, the bone marrow of the mice injected with pCD2 retrovirusproducing cells followed by 5-FC treatment was mark-

Figure 4 Evaluation of treatment-related cytotoxicity on the bone marrow. The bone marrow was extracted from the femur (hatched bar) and sternum (open bar) of disseminated CRC model mice injected with CEA419/CD retrovirus-producing cells followed by PBS treatment (A, n = 10), CEA419/CD retrovirus-producing cells followed by 5-FC treatment (B, n = 10), pCD2 retrovirus-producing cells followed by 5-FC treatment (C, n = 10) and serum-free media followed by 5-FU treatment (D, n = 5). Values are expressed as means ± s.d.

edly suppressed, with K:E ratios in the bone marrow of femora and sterna being 5% and 7%, respectively.

Discussion To achieve effective and safe in vivo gene therapy for cancer using suicide genes, it is necessary to induce strong, tumor-selective expression of suicide genes. Because tumor-selective receptor-mediated vector delivery systems have not been established, the use of tumor-selective promoters, such as a CEA promoter and ␣-fetoprotein promoter, to direct expression of suicide genes is the most promising alternative. It has been shown already that the CEA promoter isolated from a human genomic library can induce tumor-selective expression of exogenous genes in human CRC,16 gastric cancer28,29 and lung cancer,30 in which CEA is highly produced. Recently, we have isolated CEA promoter regions consisting of 419 bp and 204 bp from a high CEA-producing human CRC, and demonstrated that they can direct more than 20-fold higher expression of a luciferase reporter gene in CEAproducing CRC cells compared with CEA-nonproducing cells.24 Furthermore, in CEA-producing CRC cells, the CEA promoter regions were shown to be as active as the SV40 enhancer/promoter, which has been used as a general, strong promoter to direct expression of therapeutic genes.24 Richards et al16 have demonstrated that the CEA promoter is located between −90 and +69 bp upstream from the transcriptional start site. Chen et al31 have also demonstrated that the CEA promoter region between −123 and −28 bp upstream from the transcriptional start site contains binding sites for several important trans-acting


factors such as Sp-1 and upstream stimulatory factor. We have isolated CEA promoter regions including the core sequences from a high CEA-producing human CRC and linked them with the bacterial CD gene. Upstream regions of the CEA gene located between −309 and +110 bp and between −135 and +69 bp upstream from the transcriptional start site were found to be able to direct CEAproducing cell-selective CD gene expression, resulting in 20- to 27-fold higher susceptibility to 5-FC compared with the corresponding parental cells. There was a tendency for the CEA promoter region between −309 and +110 bp to induce more stringent killing of CEA-producing cells than that between −135 and +69 bp, indicating that the 5′ flanking region of the CEA gene between −309 and −135 bp may contain some sequences that provide CEAproducing cell-selective gene expression. As shown previously,12,14,16,17,32 established subcutaneous CRCs composed of CD gene-transduced cells were susceptible to systemic 5-FC treatment. It should be stressed that the bulk of G418-selected cells were used as CD gene-transduced cells in the present study, although the best CD gene-transduced clone was used in much of the literature. Therefore, the results demonstrated here may be more realistic when gene therapy using the CD gene is employed in a clinical setting. CRCs consisting of bulky CD gene-transduced cells exhibited an approximately two-fold reduction by 5-FC treatment compared with controls and did not grow or were even reduced thereafter. Previous in vivo studies have shown that the host’s immune responses play an important role in the antitumor effects caused by the CD/5-FC system.15,32 Although athymic nude mice do not possess an intact T cell system, they have T cell-independent immune systems, such as natural killer cells. Therefore, it is possible that while CD gene-transduced CRC cells were continuously eliminated by 5-FC treatment, T cell-independent immune responses against CRC cells were elicited in nude mice, resulting in growth arrest or even reduction of CRC. For gene therapy to be a promising clinical modality for the treatment of CRC, it is essential to target tumor cells, while sparing other normal cells. We have shown here that transduction of the CD gene under the transcriptional control of the CEA promoter makes CEA-producing CRC cells susceptible to 5-FC without affecting the sensitivity to 5-FC of CEA-nonproducing cells. Hirschowitz et al14 have demonstrated the efficacy on CRC of adenovirus-mediated in vivo transduction of the CD gene directed by a CEA promoter followed by 5-FC treatment. In their studies, animals inoculated subcutaneously with CRC cells were used as an in vivo CRC model, which is apparently different from clinical manifestations. Therefore, we used an animal model bearing CRC cells intraperitoneally to imitate human advanced disseminated CRCs. Retrovirus-mediated in vivo transduction of the CD gene was shown to confer susceptibility to 5-FC on intraperitoneally disseminated CEA-producing CRCs. Significantly prolonged survival periods were observed in animals bearing disseminated CRCs by intraperitoneal injections of pCD2 or CEA419/CD retrovirus-producing cells followed by 5-FC treatment. Importantly, the survival rate of animals injected with CEA419/CD retrovirus-producing cells was significantly higher than that of animals injected with pCD2 retrovirus-producing cells. Another important concern as well as efficacy is safety

of the treatment, because safety is a major issue for the clinical application of gene therapy. Retroviruses possess the ability of selectively infecting and achieving integration into the genome of dividing cells.33 Retroviruses are, therefore, an attractive vector for tumor-selective gene transfer, because most normal cells surrounding cancers are in a quiescent, nonreceptive stage of cell growth. There are, however, noncancerous normal dividing cells, such as bone marrow cells. In the intraperitoneally disseminated CRC experiments, six of 10 animals injected with pCD2 retrovirus-producing cells followed by 5-FC treatment and eight of 10 animals treated with 5-FU died without massively disseminated CRCs. Subsequent experiments revealed that systemic administration of 5-FC caused severe bone marrow suppression in animals injected intraperitoneally with pCD2 retrovirus-producing cells. Conversely, significant bone marrow suppression was not observed in animals injected with CEA419/CD retrovirus-producing cells followed by 5-FC treatment. These results may indicate that the death of animals with pCD2 retrovirus-producing cells followed by 5-FC treatment was due to the tumor-burden and bone marrow suppression, while the death of animals injected with CEA419/CD retrovirus-producing cells followed by 5-FC treatment was due to the tumor burden. These results may imply the safety of gene therapy with CEA promoter-directed CD gene and 5-FC treatment. To prove the usefulness of the strategy using in vivo transduction of the CEA promoter-directed CD gene followed by 5-FC treatment against human CRC, more investigations have to be performed. Specifically, better vectors have to be developed, because current vectors, such as liposomes, retroviruses and adenoviruses, have their own limitations for clinical application. The results demonstrated here, however, indicate the potential efficacy and safety of transferring the CD gene directed by the CEA promoter in vivo followed by systemic administration of 5-FC for the treatment of patients with CRCs.

Materials and methods Cell lines and cell culture Human colorectal carcinoma cell line LoVo and SW1463, human cervix carcinoma cell line HeLa, murine embryo fibroblast cell line NIH3T3, ecotropic retrovirus-packaging cell line Psi2 and amphotropic retrovirus-packaging cell line PA317 were purchased from the American Type Culture Collection (Rockville, MD, USA). Human hepatocellular carcinoma cell line SMMU7721 and human renal cell carcinoma cell line RC9406 were established by the Departments of Pathology and Microbiology of the Second Military Medical University (Shanghai, China), respectively. The cells were grown in RPMI 1640 medium containing 10% heat-inactivated fetal calf serum, 100 units/ml ampicillin and 100 mg/ml streptomycin at 37°C in a humidified 5% CO2 atmosphere. CEA secretion of cells was measured following a 7-day accumulation using the CEA-EIA Monoclonal One-step kit (Abbott Laboratories, Abbott, IL, USA). CEA expression was standardized based on the protein content of cell pellets using the BioRad protein assay kit (Hercules, CA, USA).

87


88

Construction of retrovirus vectors Construction of CR4Ca-CEA419/LUC and CR4CaCEA204/LUC plasmids, which contain a luciferase reporter gene under the transcriptional control of the CEA promoter regions isolated from a high CEA-producing human colorectal carcinoma, has been described elsewhere.24 CR4Ca-CEA419/LUC and CR4Ca-CEA204/ LUC plasmids contain the core sequences of CEA promoter regions ranging from −309 to +110 bp and from −135 to +69 bp upstream from the transcriptional start site of the CEA gene, respectively. Both the promoter regions were released from the plasmids by digestion with SacI and HindIII and inserted into the SacI/HindIII site of the linearized pSP72 plasmid (Promega, Madison, WI, USA), resulting in pCEA419/SP72 and pCEA204/SP72 plasmids. Both the CEA promoter regions were again released from the pCEA419/SP72 and pCEA204/SP72 plasmids by digestion with ClaI and HindIII and inserted into the ClaI/HindIII site of the linearized LNCX retrovirus vector,34 kindly provided by AD Miller (Fred Hutchinson Cancer Research Center, Seattle, WA, USA), to create CEA419/LNCX and CEA204/LNCX retrovirus vectors. The bacterial CD gene was excised out from the pCD2 retrovirus vector,10 kindly provided by CA Mullen (MD Anderson Cancer Center, Houston, TX, USA), blunt-ended with the Klenow fragment, ligated with HindIII adapters, and then digested with HindIII and BamHI. The resulting 1.5-kb CD gene fragment was inserted into the HindIII/BamHI site of the linearized CEA419/LNCX and CEA204/LNCX retrovirus vectors, resulting in construction of CEA419/CD and CEA204/CD retrovirus vectors (Figure 1). Recombinant retrovirus production CEA419/CD and CEA204/CD retrovirus vector constructs were converted to the corresponding retroviruses by the transinfection protocol. The retrovirus vector constructs were transfected by using LipofectAMINE Reagent (Gibco BRL, Grand Island, NY, USA) according to the protocol provided by the manufacturers into the ecotropic retrovirus-packaging cell line Psi2, as described previously.35 Filtered supernatants from ecotropic retrovirus-producing cells were then added to the amphotropic retrovirus-packaging cell line PA317 in the presence of 8 ␮g/ml Polybrene (Sigma Chemical, St Louis, MO, USA). After 48 h, the cells were split and plated in the medium containing 400 ␮g/ml active G418 (Gibco BRL). Individual G418-resistant clones were isolated by cloning cylinders and tested for ability to produce recombinant retroviruses. The titers of recombinant viral particles were determined by infecting NIH3T3 cells with serial dilutions of the culture medium from each clone, as described previously.36 The high-titer retrovirus-producing clones selected for subsequent experiments had titers of approximately 1 × 106 colony-forming units/ml as assayed by G418 selection of infected NIH3T3 cells. The culture supernatant of the high-titer retrovirus-producing clones was collected, passed through a 0.45-␮m pore filter (Millipore, Bedford, MA, USA), and stored at −70°C. It served as a source for infectious recombinant retroviruses. pCD2 retrovirus-producing clones with titers of approximately 1 × 106 colony-forming units/ml were also established in a similar manner and used for subsequent experiments.

In vitro gene transfer Freshly prepared cells were infected with the recombinant retroviruses at 37°C, 5% CO2 for 4–6 h in the medium containing 8 ␮g/ml polybrene, as described previously.37 Retrovirus-infected cells were selected by addition of 400 ␮g/ml active G418, and the bulk of G418resistant cells were expanded and used as retrovirusinfected cells for subsequent experiments. Evaluation of in vitro sensitivity to 5-FC Cells were plated at a density of 1 × 103 cells per well in 96-well, flat-bottomed tissue culture plates in the media containing various concentrations (100 ␮m–50 000 ␮m) of 5-FC (Sigma). The cells were then cultured at 37°C for 4 days and the number of viable cells was measured using the MTT assay as described previously.36 The IC50 for each cell line was calculated using a curve-fitting parameter based on the Marquardt method.38 Sensitivity of cells to 5-FU was also estimated by a similar manner. Antitumor effect by CD/5-FC system To examine the in vivo sensitivity of CD gene-transduced cells to 5-FC, parental, pCD2 retrovirus-infected and CEA204/CD retrovirus-infected LoVo cells were suspended in the serum-free RPMI 1640 medium at a concentration of 2 × 107 cells/ml, and 200-␮l inoculum volumes were injected subcutaneously into the flank regions of female athymic BALB/c-nu/nu mice, purchased from the Experimental Animal Center (Shanghai, China). 5-FC (700 mg/kg body weight/day) was administered intraperitoneally, 25 days later, for 7 consecutive days, to animals inoculated with parental or pCD2 retrovirusinfected LoVo cells. Animals inoculated with CEA204/CD retrovirus-infected LoVo cells had 5-FC or PBS administered intraperitoneally for 7 days. Each group consisted of five animals. Tumor development was observed for 60 days and tumor volume was calculated every 2 or 3 days according to the formula: V (mm3) = A (mm) × B (mm)2/2 (A = largest diameter; B = smallest diameter). Animals were killed before the end of the observation period when they developed excessively large tumors or when there were other signs of animal distress. All animal experiments were performed using approved protocols and in accordance with recommendations for the proper care and use of laboratory animals. In vivo gene therapy for disseminated CRC To examine the therapeutic effects of in vivo CD gene transduction by means of retrovirus vectors followed by 5-FC treatment on disseminated CRCs, athymic BALB/cnu/nu mice received an intraperitoneal inoculation of 2 × 106 parental LoVo cells suspended in 200 ␮l of the serum-free medium and divided randomly 3 days later into the following five groups according to treatment schedules: (a) injection of PA317 retrovirus-packaging cells followed by 5-FC treatment (n = 10); (b) injection of pCD2 retrovirus-producing cells followed by 5-FC treatment (n = 10); (c) injection of CEA419/CD retrovirus-producing cells followed by 5-FC treatment (n = 10); (d) injection of CEA419/CD retrovirus-producing cells followed by PBS treatment (n = 10); (e) injection of the serum-free medium followed by 5-FU treatment (n = 10). In groups (a) to (d), 2 × 106 PA317 retrovirus-packaging cells, pCD2 retrovirus-producing cells and CEA419/CD retrovirus-producing cells were suspended in 200 ␮l of


the serum-free medium containing 16 ␮g/ml polybrene and injected intraperitoneally into mice at days 3, 5 and 7 after the intraperitoneal inoculation of LoVo cells. In group (e), 200 ␮l of the serum-free medium containing 16 ␮g/ml polybrene were injected intraperitoneally into mice at days 3, 5 and 7. 5-FC (700 mg/kg body weight/day), 5-FU (500 mg/kg body weight/day) or PBS was then administered intraperitoneally to animals from day 10 to day 16. Survival of animals was observed everyday.

Evaluation of bone marrow suppression by the CD/5-FC system An animal model with intraperitoneally disseminated CRCs was produced and treated as described above. Animals were killed on the day after termination of 5-FC, 5FU or PBS administration. Bone marrow was extracted from the femur of each mouse, smeared quickly on a glass slide, dried with a fan and fixed with methanol for 5 to 10 min. The slides were then stained with a mixture of Giemsa and Wright staining solutions containing methylene blue, eosin, azure, methanol and glycerol for 30 min. To examine the bone marrow suppression, karyocytes, referred to cells with nuclei, and mature erythrocytes were counted under a microscope and K:E ratios were calculated. Bone marrow was also extracted from the sternum of each mouse, fixed in 10% buffered formalin, embedded in paraffin and serially sectioned. K:E ratios were determined five times for each mouse. Statistics Results are expressed as means ± s.d. Standard descriptive statistics, Student’s t test and the ␹2 test, were used. A P value of ⬍0.05 was considered to indicate a significant difference between groups.

Acknowledgements We are grateful to Dr A Mitoro for statistical analysis of the results. This work was supported in part by the National Natural Science Foundation of China (39600142, 39500147, 39600172) and by Grant-in-Aid for Scientific Research (B) (07457141, 10470140) from the Ministry of Education, Science, Sports and Culture of Japan.

References 1 Finkelstein SD, Sayegh R, Christensen S, Swalsky PA. Genotypic classification of colorectal adenocarcinoma. Cancer 1993; 71: 3827–3838. 2 Cortesi E et al. Advanced colorectal cancer. J Surg Oncol 1991; 48 (Suppl. 2): 112–115. 3 Turk PS, Wanebo HJ. Results of surgical treatment of nonhepatic recurrence of colorectal carcinoma. Cancer 1993; 71: 4267–4277. 4 Vaughn DJ, Haller DG. Nonsurgical management of recurrent colorectal cancer. Cancer 1993; 71: 4278–4292. 5 Calabresi P, Chabner BA. Antineoplastic agents. In: Gilman AG, Rall T, Nies AS, Taylor P (eds). Goodman and Gilman’s The Pharmacological Basis of Therapeutics, 8th edn. Pergamon Press: New York, 1990, pp 1209–1263. 6 Bennett J. Antifungal agents. In: Gilman AG, Rall T, Nies AS, Taylor P (eds). Goodman and Gilman’s The Pharmacological Basis of Therapeutics, 8th edn. Pergamon Press: New York, 1990, pp 1165–1181. 7 Nishiyama T et al. Antineoplastic effects in rats of 5-fluorocytosine in combination with cytosine deaminase capsules. Cancer Res 1985; 45: 1753–1761.

8 Mullen CA, Coale MM, Lowe R, Blaese RM. Tumors expressing the cytosine deaminase suicide gene can be eliminated in vivo with 5-fluorocytosine and induce protective immunity to wildtype tumor. Cancer Res 1994; 54: 1503–1506. 9 Huber BE et al. Metabolism of 5-fluorocytosine to 5-fluorouracil in human colorectal tumor cells transduced with the cytosine deaminase gene: significant antitumor effects when only a small percentage of tumor cells express cytosine deaminase. Proc Natl Acad Sci USA 1994; 91: 8302–8306. 10 Mullen CA, Kilstrup M, Blaese M. Transfer of the bacterial gene for cytosine deaminase to mammalian cells confers lethal sensitivity to 5-fluorocytosine: a negative selection system. Proc Natl Acad Sci USA 1992; 89: 33–37. 11 Austin EA, Huber BE. A first step in the development of gene therapy for colorectal carcinoma: cloning, sequencing, and expression of Escherichia coli cytosine deaminase. Mol Pharmacol 1992; 43: 380–387. 12 Huber BE et al. In vivo antitumor activity of 5-fluorocytosine on human colorectal carcinoma cells genetically modified to express cytosine deaminase. Cancer Res 1993; 53: 4619–4626. 13 Kuriyama S et al. Bacterial cytosine deaminase suicide gene transduction renders hepatocellular carcinoma sensitive to the prodrug 5-fluorocytosine. Int Hepatol Commun 1995; 4: 72–79. 14 Hirschowitz EA et al. In vivo adenovirus-mediated gene transfer of the Escherichia coli cytosine deaminase gene to human colon carcinoma-derived tumors induces chemosensitivity to 5-fluorocytosine. Hum Gene Ther 1995; 6: 1055–1063. 15 Consalvo M et al. 5-Fluorocytosine-induced eradication of murine adenocarcinomas engineered to express the cytosine deaminase suicide gene requires host immune competence and leaves an efficient memory. J Immunol 1995; 154: 5302–5312. 16 Richards CA, Austin EA, Huber BE. Transcriptional regulatory sequences of carcinoembryonic antigen: identification and use with cytosine deaminase for tumor-specific gene therapy. Hum Gene Ther 1995; 6: 881–893. 17 Ohwada A, Hirschowitz EA, Crystal RG. Regional delivery of an adenovirus vector containing the Escherichia coli cytosine deaminase gene to provide local activation of 5-fluorocytosine to suppress the growth of colon carcinoma metastatic to liver. Hum Gene Ther 1996; 7: 1567–1576. 18 Dong Y et al. In vivo replication-deficient adenovirus vectormediated transduction of the cytosine deaminase gene sensitizes glioma cells to 5-fluorocytosine. Hum Gene Ther 1996; 7: 713–720. 19 Rowley S et al. Cytosine deaminase gene as a potential tool for the genetic therapy of colorectal cancer. J Surg Oncol 1996; 61: 42–48. 20 Kanai F et al. In vivo gene therapy for ␣-fetoprotein-producing hepatocellular carcinoma by adenovirus-mediated transfer of cytosine deaminase gene. Cancer Res 1997; 57: 461–465. 21 Gold P, Freedman SO. Demonstration of tumor-specific antigens in human colonic carcinomata by immunological tolerance and absorption techniques. J Exp Med 1965; 121: 439–462. 22 Egan ML, Pritchard DG, Todd CW, Go VLW. Isolation and immunological and chemical characterization of carcinoembryonic antigen-like substances in colon lavages of healthy individuals. Cancer Res 1977; 37: 2638–2643. 23 Baranov V, Yeung M-W, Hammarstrom S. Expression of carcinoembryonic antigen and nonspecific cross-reacting 50-kDa antigen in human normal and cancerous colon mucosa: comparative ultrastructural study with monoclonal antibodies. Cancer Res 1994; 54: 3305–3314. 24 Cao G et al. Comparison of carcinoembryonic antigen promoter regions isolated from human colorectal carcinoma and normal adjacent mucosa to induce strong tumor-specific gene expression. Int J Cancer 1998; 78: 242–247. 25 Abbasi AM et al. Localization of CEA messenger RNA by in situ hybridization in normal colonic mucosa and colorectal adenocarcinomas. J Pathol 1992; 168: 405–411. 26 Kim J et al. Expression of carcinoembryonic antigen and related genes in lung and gastrointestinal cancers. Int J Cancer 1992; 52: 718–725.

89


90

27 Jothy S, Yuan S-Y, Shirota K. Transcription of carcinoembryonic antigen in normal colon and colon carcinoma. In situ hybridization study and implication for a new in vivo functional model. Am J Pathol 1993; 143: 250–256. 28 Tanaka T et al. Adenovirus-mediated prodrug gene therapy for carcinoembryonic antigen-producing human gastric carcinoma cells in vitro. Cancer Res 1996; 56: 1341–1345. 29 Lan K-H et al. In vivo selective gene expression and therapy mediated by adenoviral vectors for human carcinoembryonic antigen-producing gastric carcinoma. Cancer Res 1997; 57: 4279–4284. 30 Ozaki T et al. Gene therapy for carcinoembryonic antigen-producing human lung cancer cells by cell type-specific expression of herpes simplex virus thymidine kinase gene. Cancer Res 1994; 54: 5258–5261. 31 Chen C-J, Li L-J, Maruya A, Shively JE. In vitro and in vivo footprint analysis of the promoter of carcinoembryonic antigen in colon carcinoma cells: effect of interferon-␥ treatment. Cancer Res 1995; 55: 3873–3882. 32 Mullen CA, Petropoulos D, Lowe RM. Treatment of microscopic

33

34 35

36

37

38

pulmonary metastases with recombinant autologous tumor vaccine expressing interleukin 6 and Escherichia coli cytosine deaminase suicide genes. Cancer Res 1996; 56: 1361–1366. Miller DG, Adam MA, Miller AD. Gene transfer by retrovirus vectors occurs only in cells that are actively replicating at the time of infection. Mol Cell Biol 1990; 10: 4239–4242. Miller AD, Rosman GJ. Improved retroviral vector for gene transfer and expression. BioTechniques 1989; 7: 980–990. Yoshimatsu T, Tamura M, Kuriyama S, Ikenaka K. Improvement of retroviral packaging cell lines by introducing the polyomavirus early region. Hum Gene Ther 1998; 9: 161–172. Kuriyama S et al. Bystander effect caused by suicide gene expression indicates the feasibility of gene therapy for hepatocellular carcinoma. Hepatology 1995; 22: 1838–1846. Kuriyama S et al. Gene therapy for the treatment of hepatoma by retroviral-mediated gene transfer of the herpes simplex virus thymidine kinase. Int Hepatol Commun 1993; 1: 253–259. Marquardt DW, Siam J. An algorithm for least square estimation of non-linear parameters. J Soc Indust Appl Math 1963; 11: 431– 441.