Gene Therapy (2004) 11, 79–84 & 2004 Nature Publishing Group All rights reserved 0969-7128/04 $25.00 www.nature.com/gt BRIEF COMMUNICATION Restoration of tumor-suppressor activity in human tumor cells in vitro and in their xenografts in vivo by recombinant avian adenovirus CELO-p53

DY Logunov1,5, GV Ilyinskaya2, LV Cherenova1, LV Verhovskaya1, MM Shmarov3, PM Chumakov4, BP Kopnin2 and BS Naroditsky3 1Institute of Agricultural Biotechnology, Moscow, Russia; 2Institute of Carcinogenesis, Blokhin Research Center, Moscow, Russia; 3Gamaleya Research Institute of Epidemiology and Microbiology, Moscow, Russia; and 4Engelhardt Institute of Molecular Biology, Moscow, Russia

Human adenovirus (Ad) vectors are extensively used as various organs on the background of pre-existing immunity gene transfer vehicles. However, a serious obstacle for the to human Ad5. The created CELO-p53 vector restored the use of these vectors in clinical applications is due to pre- function of the p53 tumor suppressor both in cultured human existing immunity to human Ads affecting the efficacy of gene tumor cells in vitro and in their xenografts in nude mice in transfer. One of the approaches to circumvent host immune vivo. The latter effect was accompanied by inhibition of response could be the development of vectors based on non- tumor growth. Noteworthily, the delivery of CELO-p53 led to human Ads that are able to transduce genes into human activation of p53 target genes in cells showing inactivation of cells. In this study, we explored the possibility of using avian endogenous p53 by three different mechanisms, that is, in Ad CELO vectors as gene-transfer vehicles. For this purpose, the human epidermoid carcinoma A431, lung adenocarcino- we constructed a set of recombinant CELO viruses and ma H1299, and cervical carcinoma HeLa. demonstrated that they are able to deliver transgenes into Gene Therapy (2004) 11, 79–84. doi:10.1038/sj.gt.3302146

Keywords: cancer; avian adenovirus CELO; p53 tumor suppressor

Vectors based on the human adenoviruses (Ads) are Under these circumstances, a sufficient level of widely used for gene delivery.1,2 However, despite the transgene expression may be achieved either by transient obvious advantages of Ads as highly efficient gene immunosuppression, or by increased vector dose. vehicles, their clinical application is severely restricted Unfortunately, this can lead to undesired side effect or by natural immunity.3–5 Indeed, as the majority of adults toxicity.15 A more attractive approach avoiding the (485%) demonstrate systemic anti-Ad antibodies against problems associated with natural immunity in humans widespread Ad serotypes,6–8 injections of Ad vectors into might involve the use of non-human Ad vectors.16–19 recipients with pre-existing immunity (except aerosol Avian Ad CELO (FAV-1) seems to be a promising administration) were shown to trigger robust production candidate for the development of alternative Ad vectors of anti-Ad neutralizing antibodies.5 Production of high that are able to transfer genes into different human cells, titers of neutralizing antibodies significantly reduced the with the efficiencies comparable to those of Ad5 efficacy of gene transfer upon administration of similar vectors.17–20 There are three features of the CELO virus serotype-based Ad vectors.9–11 Several other mechanisms that distinguish it from human Ads and provide restricting the expression of Ad vectors have been additional advantages for gene therapy applications. reported. They include an increase of Ad-specific anti- First, there is no pre-existing immune response to this body action through the deposition of C on the virus in human population. Second, some peculiarities of viral surface;12 the development of crossreactivity of its capsid architecture allow one to change the tropism of response of the CD4 þ and CD8 þ T cells to a wide range CELO vectors. Indeed, unlike other Ads, the CELO virus of human Ad serotypes through conservative antigens is characterized by two different-length fibers at each associated with class I or II MHC,3,13,14 etc. vertex of a virion.20 The CELO long fiber is responsible for the recognition of the coxsackie-Ad receptor (CAR),20–24 and, unlike the short CELO fiber, it is Correspondence: Dr BS Naroditsky, Gamaleya Research Institute of dispensable for both virus assembly and its penetration Epidemiology and Microbiology, Gamaleya Street, 18, Moscow 123098, into permissive cells.20 Thus, it seems that the directed Russia 5Current address: Gamaleya Research Institute of Epidemiology and modifications within the long fiber can alter virus Microbiology, Gamaleya Street, 18, Moscow 123098, Russia specificity to definite cell types without significant Received 07 February 2003; accepted 08 July 2003 change of its natural tropism. Third, the conventional Restoration of p53 tumor-suppressor activity DY Logunov et al 80 propagation system for CELO virus is low-cost chicken It was previously established that the development of embryos.18,25 anti-Ad5 neutralizing antibodies results in significant However, despite the above potential advantages of restriction of further gene transfer with Ad5.9–11 To CELO vectors to date, there were no detailed studies characterize the potential usefulness of CELO vectors as devoted to testing their application in vivo. As the first an additional delivery system for gene therapy, we step in this direction, we characterized the abilities of performed a crossadministration experiment. Mice were CELO vectors (i) to deliver transgenes into different immunized with Ad5 or CELO viruses intraperitoneally. tissues, (ii) to transfer transgenes into hosts bearing The anti-Ad5 and anti-CELO sera showed no cross- preformed anti-hAd5 antibodies, and (iii) to achieve the inhibition of tumor growth following delivery of the p53 tumor-suppressor gene into mice bearing human tumors. The recombinant CELO viruses were constructed as described earlier17 by homologous recombination between the CELO virus genome digested with SwaI restriction endonuclease and a plasmid containing CMV promoter-driven cassettes expressing green fluorescent (GFP), secreted alkaline phosphatase (SEAP) or p53 tumor suppressor, flanked by a fragment of CELO virus DNA. The details of the design and the scheme of the resulting recombinant CELO viruses are shown in Figure 1. We tested the ability of resultant vectors for gene transfer in vivo. Intravenous injections of the recombinant CELO-SEAP virus into mice resulted in a high levels of SEAP in the blood sera. The CELO virus DNA was found in the liver, spleen, heart, and kidney (Figure 2). Local vector delivery to lungs and muscles was also successful (Figure 2). Interestingly, upon intramuscular delivery, the initial levels of SEAP in blood were similar to those observed after intratracheal injection. However, the intramuscular administration resulted in a shorter time of SEAP expression, which was probably due to more rapid clearance of CELO DNA from muscles (Figure 2).

Figure 1 Scheme of the recombinant CELO-p53, CELO-GFP, CELO- SEAP genome (mu. map units; dl deletion). Recombinant variants of avian Ad CELO were constructed as described previously.17 Briefly, the plasmid containing CMV promoter and cDNAs of either p53, GFP, or SEAP and flanked with the CELO genome fragment (88.8–100 map units) was cotransfected into LMH cells along with the CELO DNA digested with SwaI. Recombinant Ads were plaque-purified twice, propagated, and isolated by two CsCl density centrifugation. Before injections, the viruses were dialyzed as described previously.17

Figure 2 Ability of the CELO vector to deliver transgenes into different organs. The upper panel shows the SEAP activity in the blood sera of immunocompetent D2&I mice (n¼24) after intratracheal, intramuscular, and intravenous injection of recombinant virus CELO-SEAP (5 Â 1011 particles/injection). SEAP was measured as described by Berger et al.26 At the indicated time points, blood samples were collected and the sera were prepared. To inhibit the endogenous phosphatase activity, the sera were preheated at 651C for 10 min. Next, the SEAP activity was tested. The bottom panel shows Southern-blot analysis of total DNAs isolated from different organs of mice injected with CELO-SEAP (5 Â 1011 particles/ injection) intramuscularly, intratracheally, or via the ear vein. Southern- blot analysis was performed as described by Lusky et al.27 At day 7, postintravenous injection was carried out and, at the indicated time points, the total genomic DNAs were extracted from tested organs, digested with restriction endonuclease PvuI, and analyzed using a 32P-labeled PCR product from 30 261 to 30 503 bp of CELO genome.

Gene Therapy Restoration of p53 tumor-suppressor activity DY Logunov et al 81 reactivity both in cross-virus neutralization assay and in ELISA. In agreement with this observation, we found that the CELO vectors are capable of gene transfer in the context of pre-existing anti-Ad5 antibodies, with an efficiency comparable to that in the control animals (Figure 3). Indeed, there was no statistically significant difference in the level of SEAP expression between the group of mice immunized with Ad5 and nonimmunized animals. In contrast, animals with preformed anti-CELO antibodies demonstrated a low-level transient expression of SEAP after injection with CELO-SEAP vector. Our results show that the CELO vectors efficiently escape humoral anti-Ad5 response, thus allowing to carry out gene transfer in hosts that were subjected to Ad5 vectors in the past. Then, we explored the possibility of application of the CELO-p53 vector to cancer gene therapy. The efficiency of the p53 gene delivery and the effects of its expression were tested in various human tumor cell lines in vitro, and in the same cells grown as xenografts in nude mice. First, the CELO-p53 virus was tested in vitro. Its ability to express p53 was demonstrated by Western-blot analysis (Figure 4a). Next, we studied the ability of the CELO-p53 vector to restore p53 function in human tumor cell lines (HeLa, A431, and H1299) in which the activity of endogenous p53 was disrupted by different mechanisms. The HPV18-positive HeLa cells (cervical carcinoma) show greatly reduced levels of p53 due to its association with the viral E6 oncoprotein.29 The A431 cells (epider- moid carcinoma) express the mutated form of the p53 gene with substitution at codon 273 (His273).30 The H1299 cells (lung adenocarcinoma) have partial homo- zygous deletion of the p53 gene and no expression of the p53 protein.31 In order to monitor the activity of introduced p53, we used derivatives of the above cell lines bearing b-galactosidase reporter gene under the control of a p53-inducible promoter. After infection with the CELO-p53 virus, we observed p53-mediated b-galactosidase expression in all of the cell lines (Figure 4b). The A431/waf1ConALacZ cells displayed a somewhat lower activation of LacZ expres- sion in response to transduction by CELO-p53. This may have two explanations. First, we found that, after infection of the A431 cells with control CELO-SEAP virus, they show four-fold lower production of the secreted alkaline phosphatase as compared with the Figure 3 SEAP gene transfer by CELO vector in the context of preformed HeLa and H1299 cells (Figure 4c). Therefore, transduction anti-Ad5 and anti-CELO antibodies. Immunocompetent D2&I mice were of the A431 cells by CELO virus is less efficient. Second, immunized intraperitoneally with CELO (n¼6) or Ad5 (n¼6) purified the A431 cells express high levels of His273 p53 mutant, viruses (50 mkg per injection) using the routine protocol.28 The titers were which might interfere with the activity of the exogenous defined as the reciprocal of the highest dilution of a sample for which the 32,33 optical density (OD) exceeded at least two times the OD of the negative p53 through dominant-negative inhibition. controls. They were estimated as geometric mean titer (lg)7s.d. The titers For further in vivo experiments, we have chosen the of anti-CELO and anti-Ad5 sera were 4.9570.16 and 4.7570.23, A431 cells because they represent the most frequent type respectively. The crossreactive antibodies to CELO or Ad5 were absent of p53 alterations (missense mutation) in various human for both anti-Ad5 and anti-CELO sera. In addition, the crossreactivities of 30 anti-CELO and anti-Ad5 sera were tested using the virus neutralization tumors. Moreover, the testing efficiency of the vector in 5 such a complicated context (the decreased efficiency of assay, by their ability to prevent the infection of 293 cell line with Ad5 or LMH cells with CELO. The neutralizing antibody titer (per 5 ml of serum) transduction, the dominant-negative inhibition of exo- was calculated as the reciprocal of the highest dilution showing o50% genous p53 by the endogenous His273 mutant) could cytopathic effect, and was estimated as the geometric mean of neutralizing allow assessment of real potential of this vector for gene antibodies titer (lg)7s.d. The anti-CELO and anti-Ad5 antibodies were therapy. As in the in vitro experiment, the re-establish- virus-specific and titers were equal to 3.770.15 for anti-CELO and 7 ment of the p53 function in vivo was monitored by the 3.4 0.20 for anti-Ad5 sera. There was no crossreactivity of anti-CELO expression of b-galactosidase driven by the p53-depen- and anti-Ad5 sera. The immunized mice were injected intravenously with 1011 particles of CELO-SEAP or CELO-GFP recombinant viruses. dent promoter. Six A431/waf1ConALacZ tumor xeno- Measurement of the SEAP activity at different time points after virus grafts grown in nude mice were injected once with injection showed the ability of CELO vectors to circumvent the preformed 5 Â 1011 particles of the CELO-p53 virus. Injection anti-Ad5 humoral response.

Gene Therapy Restoration of p53 tumor-suppressor activity DY Logunov et al 82

Figure 5 Transcriptional activity of p53 in A431/waf1ConALacZ tumors after injection with the CELO-p53 virus. Tumors were injected with CELO-p53 (b, d), CELO-GFP (a, c), and PBS (PBS not shown), and removed 48 h after treatment. Frozen sections of tumors (a, b) were stained with X-gal as described above (magnification  50). Nonfixed tumor sections (c, d) showing the expression of GFP protein (magnification  250).

Figure 4 Restoration of p53 function in tumor cells by the recombinant CELO-p53 virus. (a) Western-blot analysis of p53 expression. Leghorn male hepatoma (LMH) cells were infected with CELO-p53 virus at MOIs of 10, harvested at 18 h and lysed. Proteins were separated using SDS- PAGE and transferred to a polyvinylidene difluoride (PVDF) membrane. P53 protein was detected by incubation with monoclonal mouse anti- human p53 antibody (p53 DO-1; Santa Cruz Biotechnology, Santa Cruz, USA). After washing, the membrane was incubated with the secondary horseradish peroxidase-conjugated goat anti-mouse IgG antibody. HRP was detected using DAB. (b) p53-responsive expression of LacZ. Cells were infected with CELO-p53 and CELO-GFP viruses at MOIs 10, 100, and 500 (mock not shown). b-Galactosidase staining was performed as described.39 After 48 h post-transduction, the cells were washed with PBS, fixed for 15–30 min in ice-cold fixing solution with 1 mM MgCl2, Figure 6 Effect of recombinant CELO-p53 virus on A431/waf1ConALacZ and stained in 0.2% X-gal, 3.3 mM K4Fe(CN)6, 3.3 mM K3Fe(CN)6, 0.02% tumors in a nude mice model. Female D2&I nude mice (n¼18) were NP-40, in PBS in the dark at 371C for 4–5 h. (c) SEAP activity in culture inoculated s.c. in the dorsal flank with 106 A431/waf1ConA cells medium from A431/waf1ConALacZ, HeLa/waf1ConALacZ, H1299/ suspended in 200 ml of PBS. When palpable nodules were established, waf1ConALacZ cells infected with CELO-SEAP and Ad5-SEAP viruses. tumors were injected with PBS (100 ml), CELO-GFP (5 Â 1011 particles/ The cells were infected with CELO-SEAP and Ad5-SEAP viruses at MOIs 100 ml), or CELO-p53 (5 Â 1011 particles/100 ml). A total of four injections 10, 100, and 500. SEAP activity was determined after 48 h, as described were given every 48 h. Tumor volume was calculated as (width2) Â (length) Â 0.5. Differences in the tumor volumes for the treatment groups were compared using the Student’s t-test. Statistical significance was defined as Po0.05. with the CELO-GFP virus, or with PBS, was used as controls. X-gal staining was performed on frozen sections of tumors harvested 48 h after injection. Only demonstrates that repeated injections of the CELO-p53 tumors injected with the CELO-p53 virus showed strong virus inhibit tumor growth. b-galactosidase expression. In contrast, all control tumors The results of our study suggest that transduction by demonstrated low background expression of b-galacto- CELO-p53 vectors can restore the functional activity of sidase (Figure 5). p53 in tumor cell lines with different modes of p53 The next question was whether injection with the inactivation. Importantly, the transduction by the CELO- CELO-p53 virus could inhibit the growth of A431/ p53 vector can overcome the dominant-negative effect of waf1ConALacZ tumors. The established A431/waf1 endogenous mutant p53. Interestingly, we noticed that ConALacZ tumors grown on a separate set of mice the control CELO-GFP virus showed some tumoricidal were injected with 5 Â 1011 particles of CELO-p53 (n¼6) effect as well. This independence of the therapeutic or CELO-GFP (n¼6) viruses, or PBS (n¼6). Each tumor transgene effect of the Ads can be explained by the was injected four times at 48 h intervals to maximize the activation of the immune response, at least partially transduction of p53. The mean final volume of tumors mediated through IFN-g and the action of NK cells.34 injected with the CELO-p53 virus was significantly Additionally, the CELO vectors could act through smaller than that of control tumors injected with the inhibition of histone deacetylase HDAC1 activity, as the GELO-GFP virus or PBS (Figure 6). This result clearly CELO genome contains the GAM1 gene encoding the

Gene Therapy Restoration of p53 tumor-suppressor activity DY Logunov et al 83 inhibitor of HDAC1.35 It was observed that the inhibitors 12 Loser P et al. Ovine adenovirus vectors promote efficient gene of HDAC can suppress tumor progression by growth delivery in vivo. Gene Ther Mol Biol 1999; 4: 33–43. arrest, or induced differentiation.36–38 Other 13 Flomenberg P, Piaskowski V, Truitt RL, Casper JT. possibilities also cannot be excluded. Characterization of human proliferative T cells response to In summary, our results indicate that the CELO-based adenoviruses. J Infect Dis 1995; 171: 1090–1096. vectors might represent a promising tool for gene 14 Smith CA, Woodruff LS, Rooney C, Kitchigan GR. Extensive therapy applications. Clearly, additional studies are cross-reactivity of adenovirus-specific cytotoxic T cells. Hum required to assess their biosafety. According to our Gene Ther 1998; 9: 1419–1427. preliminary data, the CELO-p53 virus shows no onco- 15 Marshall E. Gene therapy death prompt review of adenovirus genicity during at least 6 months; however, more vector. Science 1999; 286: 2244–2245. thorough testing is required. Understanding the impor- 16 Loser P et al. Advances in the development of non-human viral DNA-vectors for gene delivery. Curr Gene Ther 2002; 2: 161–171. tance of increased efficiency of cell-specific gene transfer, 17 Shmarov MM et al. Eukaryotic vectors of Celo avian adenovirus we are currently developing new-generation vectors genome, carrying GFP and human IL-2 genes. Mol Gen Microbiol with modified elements of the capsid, in particular those Virol 2002; 2: 25–30. of the long fiber. 18 Michou AI, Lehrmann H, Saltik M, Cotten M. Mutational analysis of the avian adenovirus CELO, which provides a basis for gene delivery vectors. J Virol 1999; 73: 1399–1410. Acknowledgements 19 Francois A et al. Construction of avian adenovirus CELO recombinants in cosmids. J Virol 2001; 11: 5288–5301. We are grateful to Dr M Cotten (Institute for Molecular 20 Tan PK, Michou A, Bergelson JM, Cotten M. Defining CAR as a Pathology, Vienna, Austria) for providing the LMH cells, cellular receptor for the avian adenovirus CELO using a genetic and to Dr K Doronin for stimulating discussions. The analysis of the two viral fibre proteins. J Gen Virol 2001; 82: work was supported by the Russian Foundation for Basic 1465–1472. Research (BSN, BPK, and PMC) and the International 21 Bergelson JM et al. Isolation of a common receptor for coxsackie Research Scholars Program of the Howard Hughes B viruses and adenoviruses 2 and 5. Science 1997; 275: 1320–1323. Medical Institute (BPK). 22 Roelvink PW et al. The coxsackievirus-adenovirus receptor protein can function as a cellular attachment protein for adenovirus serotypes from subgroup A, C, D, E and F. J Virol 1998; 72: 7909–7915. References 23 Stevenson SC et al. Human adenovirus serotype 3 and 5 bind two different cellular receptors via the fiber head domain. J Virol 1 Corti O et al. A single adenovirus vector mediates doxycycline- 1995; 69: 2850–2857. controlled expression of human tyrosine hydrolase in brain 24 Tomko RP et al. Expression of the adenovirus receptor and its grafts of human neural progenitors. Nat Biotechnol 1999; 17: interaction with the fiber knob. Exp Cell Res 2000; 255: 47–55. 349–354. 25 Laver WG, Younghusband HB, Wrigley NG. Purification and 2 Dewey RA et al. Chronic brain inflammation and persistent properties of chick embryo lethal orphan virus (an avian herpes simplex virus 1 thymidine kinase expression in survivors adenovirus). Virology 1971; 45: 598–614. of syngenic glioma treated by adenovirus-mediated gene 26 Berger J, Hauber J, Hauber R, Geiger R, Cullen BR. Secreted therapy: implication for clinical trials. Nat Med 1999; 5: placental alkaline phosphatase: a powerful new quantitative 1256–1263. indicator of in eukaryotic cells. Gene 1988; 66: 3 Yang Y, Li Q, Ertl HC, Wilson JM. Cellular and humoral immune 1–10. responses to viral antigens create barriers to lung-directed gene 27 Lusky M et al. In vitro and in vivo biology of recombinant therapy with recombinant adenoviruses. J Virol 1995; 69: adenovirus vectors with E1, E1/E2A, or E1/E4 deleted. J Virol 2004–2015. 1998; 72: 2022–2032. 4 Kass-Eisler A et al. The impact of developmental stage, route of 28 Verhovskaya LV et al. Enzyme-linked immunosorbent assay for administration and the immune system on adenovirus-mediated determination of specific antibodies to canine adenovirus hexon. gene transfer. Gene Therapy 1994; 1: 395–402. Agric Biotechnol 1994; 4: 36–42. 5 Harvey BG et al. Variability of human systemic humoral immune 29 Liang XH et al. Co-localization of the tumor suppressor protein responses to adenovirus gene transfer vectors administered to p53 and human papilloma virus E6 protein in human cervical different organs. J Virol 1999; 73: 6729–6742. carcinoma cell lines. Oncogene 1993; 8: 2645–2652. 6 Huebner RG et al. Adenoidal–pharingeal–conjuctivital agents. A 30 Park DJ et al. Transcriptional and DNA-binding abilities of newly recognized group of common viruses of the respiratory endogenous p53 in p53 mutant cell lines. Oncogene 1994; 9: system. N Engl J Med 1954; 251: 1077–1086. 1899–1906. 7 Jung D, Wigand R. Epidemiology of group II adenoviruses. Am J 31 Chen JY et al. Heterogenety of transcriptional activity of mutant Epidemiol 1967; 85: 311–319. p53 protein and p53 DNA target sequences. Oncogene 1993; 8: 8 Schmitz H, Wigand R, Hienrich W. Worldwide epidemiology of 2159–2166. human adenovirus infections. Am J Epidemiol 1983; 117: 455–466. 32 Sigal A, Rotter V. Oncogenic mutations of the p53 tumor 9 Kass-Eisler A et al. Circumventing the immune response suppressor: the demons of the guardian of the genomes. to adenovirus-mediated gene therapy. Gene Therapy 1996; 3: Cancer Res 2000; 60: 6788–6793. 154–162. 33 Chene P. In vitro analysis of the dominant negative effect of p53 10 Mack CA. Circumvention of anti-adenovirus neutralizing mutants. J Mol Biol 1998; 281: 205–209. immunity by administration of an adenoviral vector of an 34 Korst RJ et al. Adenovirus gene transfer vectors inhibit growth of alternate serotype. Hum Gene Ther 1997; 8: 99–109. lymphatic tumor metastases independent of a therapeutic 11 Mastrangeli A et al. ‘Sero-switch’ adenovirus-mediated in vivo transgene. Hum Gene Ther 2001; 12: 1639–1649. gene transfer: circumvention of anti-adenovirus humoral 35 Colombo R et al. The adenovirus protein GAM1 interferes defences against repeat adenovirus vector administration by with sumoylation of histone deacetylase 1. EMBO Rep 2002; 3: changing the adenovirus serotype. Hum Gene Ther 1996; 7: 79–87. 1062–1068.

Gene Therapy Restoration of p53 tumor-suppressor activity DY Logunov et al 84 36 Butler LM et al. The histone deacetylase inhibitor SAHA arrest 38 Imanishi R et al. A histone deacetylase inhibitor enhances killing cancer cell growth, up-regulates thioredoxin-binding protein-2, of undifferentiated thyroid carcinoma cells by p53 gene therapy. and down-regulates thioredoxin. Proc Natl Acad Sci USA 2002; J Clin Endocrinol Metab 2002; 87: 4821–4824. 99: 11700–11705. 39 Komarova EA et al. Transgenic mice with p53-responsive LacZ: 37 Vigushin DM, Coombes RC. Histone deacetylase p53 activity varies dramatically during normal development and inhibitors in cancer treatment. Anticancer Drugs 2002; 13: determines radiation and drug sensitivity in vivo. EMBO J 1997; 1–13. 16: 1391–1400.

Gene Therapy