Use of Helper-Dependent Adenoviral Vectors of Alternative Serotypes Permits Repeat Vector Administration

Home , Helper virus, Viral vector

Gene Therapy (1999) 6, 1565–1573  1999 Stockton Press All rights reserved 0969-7128/99 $15.00 http://www.stockton-press.co.uk/gt Use of helper-dependent adenoviral vectors of alternative serotypes permits repeat vector administration

RJ Parks1,2,3, CM Evelegh1 and FL Graham1,2 Departments of 1Biology and 2Pathology, McMaster University, 1280 Main Streeet West, Hamilton, Ontario, L8S 4K1, Canada

We have developed a new helper adenovirus (Ad) based expression in mouse liver after intravenous injection of the on serotype 2, Ad2LC8cCARP, for use in the Cre/loxP sys- Ad2-based hdAd showed that the vector could efficiently tem (Parks et al. Proc Natl Acad Sci USA, 1996; 93: transduce the liver, and produce high levels of a foreign 13565–13570) to generate Ad vectors deleted of all protein transgene, similar to those expressed by the hdAd gener- coding sequences (helper-dependent Ad vectors (hdAd)). ated with the Ad5 helper virus. Mice immunized with hdAd2 A comparison of Ad2LC8cCARP and our original helper produced Ad2-neutralizing antibodies, which did not cross- virus (based on serotype 5, Ad5LC8cluc) showed that the react with hdAd5. To determine if successful repeat Ad two helper viruses amplified hdAd with a similar efficiency, vector administration could be achieved by sequential use and resulted in a similar yield and purity after large-scale of alternative Ad serotypes, we injected mice with hdAd2 preparation of vector. In vitro, the resulting hdAd2 had a (hSEAP) followed 3 months later by a lacZ-expressing similar transduction efficiency and expression kinetics of hdAd of either the same or different serotype. Repeated transgene (␤-gal) as the hdAd5. An important feature of administration of hdAd2 resulted in a 30- to 100-fold the helper-dependent system is that all virion components, reduction in transgene expression compared with naive except the virion DNA, derive from the helper virus. Conse- animals. In contrast, no decrease in transgene expression quently, vectors produced with help from Ad2LC8cCARP was observed when the second vector was of a different were not neutralized by antibodies against Ad5, and vec- serotype. These results demonstrate that effective vector tors produced with Ad5 helper were resistant to neutraliz- readministration can be achieved by the sequential use of ing antibodies against Ad2. Analysis of transgene hdAds based on alternative serotypes.

Keywords: adenovirus; vectors; neutralizing antibody; serotype

Introduction some animal models,18–22 and is reported to decrease the formation of anti-Ad neutralizing antibodies.21 The use Recently, adenovirus (Ads) vectors have received con- of second-generation Ad vectors, which are further siderable attention for transgene delivery to mammalian deleted or attenuated in E2 or E4, can lead to decreased cells generally and for gene therapy in particular due to inflammatory responses and a longer duration of trans- several advantages over other vector systems, including gene expression,23–29 although not in all cases.10,11,30,31 high transduction efficiency of a variety of cell types com- However antibody responses against second-generation prised of both replicating and non-replicating cells, ease Ads were similar to those generated against first-gener- 1 of virus production, and relative safety. However, data ation vectors,11 thus compromising the ability to read- from preclinical and clinical studies have shown that Ads minister the vector. also have several disadvantages primarily due to the The development of systems for the generation of induction of both cellular and humoral immune helper-dependent Ad vectors (hdAd) which are deleted 2–13 responses to vector-derived antigens. Because of these for most if not all viral coding sequences,32–39 has allowed immune responses, administration of first-generation Ad production of hdAd which can provide long-term, high- vectors (ie deleted of early region 1 (E1) or E1/E3) has level transgene expression,40–42 and which result in sub- generally resulted in only transient transgene expression stantially reduced inflammatory and cellular immune and poor expression following repeat vector adminis- responses.41–43 However, as expected, deletion of all Ad 14–17 tration. Reintroduction of the E3 region, which enco- coding sequences does not overcome the humoral des functions involved in aiding virus escape from host immune response, and the resultant neutralizing anti- immune responses, can prolong transgene expression in bodies (JL Bramson, RJP and FLG, unpublished results), impair the effectiveness of hdAd vector readministration. In an attempt to prevent vector-directed immune Correspondence: FL Graham responses, many groups have explored the use of transi- 3Current address: Centre for Molecular Medicine, Ottawa General Hospi- tal Research Institute, 501 Smyth Road, Ottawa, Ontario, Canada, ent immune blockage at the time of Ad vector adminis- 44–58 K1H 8L6 tration, or the induction of tolerance to Ad. These Received 23 February 1999; accepted 11 May 1999 strategies have been somewhat successful, and allow Helper-dependent adenovirus type 2 vectors RJ Parks et al 1566 repeat vector administration; however, complications from Ad2, with the possible exception of pIX. The pIX and potential side-effects may make immune suppression gene is located immediately adjacent to E1, and encodes impractical for clinical use. An alternative strategy to a minor virion structural component that has not been allow for repeat vector administration is the sequential shown to be a major target for neutralizing antibody use of different Ad serotypes. Neutralizing antibodies activity,64,65 and is highly conserved between type 2 and formed against one serotype should have no effect on type 5 Ads (138 of 139 amino acids are identical between subsequent delivery of a different serotype, and this Ad2 and Ad5 pIX). Thus, the presence of an Ad5-pIX in approach has allowed repeat administration of first- our ‘Ad2’ hdAd vectors should not interfere with their generation Ad vectors.46,59–61 However, a reduced persist- ability to transduce cells in the presence of anti-Ad5 anti- ence of expression has been observed after the second bodies. We also included a fragment of lambda DNA administration,60 presumably due to cross-reacting cyto- within the E3 region of Ad2LC8cCARP as a stuffer to toxic T lymphocytes which can eliminate the transduced prevent RCA formation. If Ad2LC8cCARP were to cells.62 Over 40 different serotypes of human Ads have recombine with the Ad5 sequences contained in the 293 been isolated, suggesting that, in theory, Ad vectors of or 293Cre4 cells, it would generate a virus of approxi- different serotypes could be administered many times mately 39 kb, which greatly exceeds the upper limit for throughout the life of a patient. Ad DNA packaging.66 Consequently, as with our earlier The Cre/loxP system for producing helper-dependent helper Ad5LC8cluc, we have not observed RCA in our Ad vectors involves the use of a helper virus that con- helper virus preparations or in stocks of vector produced tains a packaging signal flanked by loxP sites.39 Upon using Ad2LC8cCARP (data not shown). infection of a 293-derived cell line that stably expresses the bacteriophage P1 Cre recombinase,63 the packaging Amplification of a hdAd using Ad2LC8cCARP signal is excised from almost all the helper virus DNA, To determine if Ad2LC8cCARP could amplify hdAd with rendering it unpackageable. The helper virus DNA an efficiency equal to that of Ad5LC8cluc, we transfected retains the ability to replicate and provides all of the duplicate 60-mm dishes of 293Cre cells with 5 ␮gof functions required in trans for the replication and packag- pRP1050 and, next day, infected the monolayers at a mul- ing of a hdAd. This system facilitates the generation of tiplicity of infection (MOI) of 5 with Ad2LC8cCARP. high-titer hdAd preparations with substantially reduced After complete CPE (approximately 72 h), the monolay- quantities of contaminating helper-virus. A key feature ers were harvested into the medium, freeze–thawed, and of the helper-dependent system is that the serotype of the an aliquot of the lysates analysed for lacZ-transducing hdAd is determined only by the helper virus. Therefore, particles. Ad2RP1050 was rescued at a frequency of in contrast to first-generation vectors that require the con- approximately 400 blue-forming units (b.f.u.) per pmol of struction of a new vector to switch serotypes, a series of transfected DNA, which is similar to the efficiency pre- genetically identical hdAds of different serotypes could viously observed for plasmids of comparable size using be generated simply by changing the serotype of the Ad5LC8cluc,67 suggesting that Ad2LC8cCARP could helper. indeed act as an effective helper virus in the 293Cre4 We have now constructed a helper virus based on the cells. We then subjected an aliquot of the crude lysates Ad2 serotype for use in the Cre/loxP system for the gen- (500 ␮l) to serial passage on Ad2LC8cCARP-infected eration of Ad vectors deleted of all Ad protein coding 293Cre cells, in order to determine the kinetics of vector sequences. Using this new virus and our previous helper amplification. As shown in Figure 2, Ad2RP1050 was virus that is based on Ad5, we produced genetically amplified using Ad2LC8cCARP at a rate similar to that identical hdAds that differ only in the virion protein previously observed for the Ad5LC8cluc helper virus components, which are derived from the helper virus. We (RJP and FLG, unpublished results). After four serial show that the vectors have identical expression character- passages on the helper virus-infected 293Cre cells, istics in vitro, regardless of the serotype, and that the Ad2RP1050 reached a titer of 3.4 × 107 b.f.u./ml. A large- sequential use of hdAd of different serotypes allows for scale preparation of Ad2RP1050 was performed, yielding successful repeat vector administration in vivo. 3 × 1011 b.f.u. from 20 150-mm dishes, with a helper virus contamination of 3.2 × 107 p.f.u./ml (approximately 0.01% of the Ad2RP1050 titer). We conclude that Results Ad2LC8cCARP can act as a helper virus in 293Cre4 cells with an efficiency similar to that of Ad5LC8cluc. Generation of an Ad2-based helper virus Ad2LC8cCARP contains an Ad5 left-end identical to that Effect of Ad5 neutralizing antibodies on hdAd2 of our previous helper virus Ad5LC8cluc including the In theory, an hdAd based on an Ad2 serotype should not ‘floxed’ packaging signal (Figure 1). As such, we be affected by Ad5 neutralizing antibodies, allowing for expected that the efficiency of Cre-mediated excision of vector readministration in animals previously treated the packaging signal (⌿) would be similar for the two with an Ad5-based hdAd. To determine if Ad2RP1050 viruses. Preliminary experiments to test for excision in was sensitive to antibodies generated against Ad5, 106 the 293Cre4 cell line showed that ⌿ was indeed excised b.f.u. of Ad2RP1050, Ad5RP1045 or Ad5CA35 were incu- from Ad2LC8cCARP with an efficiency similar to that bated with serial dilutions of Ad5-neutralizing serum, observed for Ad5LC8cluc (data not shown). Thus, and then used to infect A549 cells. Twenty-four hours Ad2LC8cCARP should act as an equally effective helper later, crude cell extracts were prepared from the infected virus in our Cre/loxP system, resulting in only low levels cells and assayed for ␤-gal activity. In this assay, infec- of helper virus contamination in the resulting vector tivity directly correlates with ␤-gal activity and thus neu- stocks. tralizaton of the vector by antibody leads to a corre- Ad2LC8cCARP encodes all structural proteins derived sponding reduction in ␤-gal activity. Both Ad5CA35 and Helper-dependent adenovirus type 2 vectors RJ Parks et al 1567

Figure 1 Construction of Ad2LC8cCARP. (a) Strategy for the generation of an Ad2 virus with a loxP-flanked packaging signal. (b) Strategy for the rescue of a stuffer segment into the E3 region of Ad2LC8c. Ad2LC8cCARP was constructed by a combination of molecular cloning and in vivo recombination techniques, as detailed in the Materials and methods. The final viral construct, Ad2LC8cCARP, is deleted of Ad sequences between 339 and 3533 bp (E1), and contains a packaging signal flanked by loxP sites. Ad2LC8cCARP also contains a 5.6 kb fragment of lambda DNA inserted within the E3 region. Regions of the Ad genome are delineated by whether they are of Ad2 or Ad5 origin, and the appropriate nucleotide number according to the conventional Ad2 or Ad5 map. The Ad2/5 crossover is left of the BspHI site located at nt 7882 on the conventional Ad5 map. Ad5 packaging signal (⌿), loxP sites (black triangles), Ad5 ITR (black arrow).

Ad5RP1045 were neutralized by incubation with the Ad5 antibodies, with an almost 100-fold decrease in ␤-gal activity at the highest antibody concentration (Figure 3). In contrast, no decrease in ␤-gal activity (or virus infectivity) was noted for Ad2RP1050. These results indicate that, as expected, the virion protein components derived from Ad2LC8cCARP and present in Ad2RP1050 are not sensitive to Ad5 neutralizing antibodies.

In vitro transgene expression Although Ad2RP1050 and Ad5RP1050 are genetically identical, and would be predicted to have virtually identical expression characteristics, the efficiency of cell transduction could be affected by the presence of different virion capsid proteins. It is also possible that subtle dif- ferences in the protein coat or core proteins contained within the virion might influence the efficiency of trans- port of the hdAd DNA to the nucleus or affect promoter activity. We therefore examined transgene expression of Figure 2 Amplification of Ad2RP1050 using Ad2LC8cCARP. After each Ad2RP1050 and Ad5RP1050 in transduced A549 cells. We serial passage, an aliquot of the resulting crude vector lysates was titered for the presence of lacZ-transducing particles (blue forming units – b.f.u.), chose A549 since E1-deleted first-generation vectors do as previously described.39 Amplifications were performed in duplicate, the not undergo productive infection in these cells, and our average b.f.u./ml is reported and error bars represent the maximum value. analysis of transgene expression over time would not be Helper-dependent adenovirus type 2 vectors RJ Parks et al 1568 complicated by the presence of small quantities of helper Generation of neutralizing antibodies in hdAd2- virus. Monolayers of A549 cells in 60-mm dishes were immunized animals transduced in duplicate with 106 b.f.u. of Ad2RP1050 or We would predict that animals immunized with a hdAd2 Ad5RP1050 and, at various times after transduction, would generate neutralizing antibodies to Ad2, but that crude protein extracts were prepared and analyzed for these antibodies would have no effect on a hdAd5. We ␤-gal activity. The hdAd2- and hdAd5-lacZ vectors had therefore analyzed serum collected at 28 days after injec- virtually identical expression characteristics in vitro over tion from mice injected i.v. with 5 × 1010 particles of the duration of the experiment (Figure 4). We conclude Ad2RP1046 for neutralizing antibodies to Ad2 or Ad5. that hdAd generated using an Ad2- or Ad5-based helper Serum samples were serially diluted and incubated with virus have identical transduction efficiencies and Ad2RP1050 or Ad5RP1050, and the resulting vector transgene expression characteristics in vitro. assayed for the ability to transduce A549 cells and express lacZ, as described above. All of the animals immunized with Ad2RP1046 produced neutralizing antibodies to Ad2, which resulted in a 30- to 100-fold decrease in Ad2RP1050 transduction at the highest antibody concentrations examined (Figure 5a). In contrast, there was no effect on Ad5RP1050, indicating that the hdAd2-immunized animals produced antibodies that were specific to Ad2 (Figure 5b). Therefore, it appears that helper-dependent Ad vectors based on alternative

Figure 3 Effect of Ad5 neutralizing antibodies on hdAd2 or hdAd5. Serial dilutions of Ad5 neutralizing antibodies were incubated for 1 h with 106 b.f.u. of Ad2RP1050 or Ad5RP1045, or 106 p.f.u. of Ad5CA35, a ﬁrst- generation Ad vector containing an identical expression cassette in place of the E1 region.74 The resulting vector was used to infect 22-mm dishes of A549 cells, and the quantity of ␤-gal assayed at 24 h after transduction. Using this assay, the quantity of ␤-gal expressed is proportional to the transduction efﬁciency (ie titer) of the vector.

Figure 4 In vitro expression from an Ad2- versus Ad5-based hdAd. Figure 5 Formation of Ad2-specific neutralizing antibodies in animals Ad2RP1050 and Ad5RP1045 are similar in structure, and contain an immunized with an hdAd2. Adult FVB/n mice (n = 3) were injected identical MCMV-lacZ expression cassette, but were generated using through the tail vein with 5 × 1010 particles of Ad2RP1046. At day 28 Ad2LC8cCARP and AdLC8cluc, respectively. Monolayers of A549 cells after injection, serum samples were collected and assayed for Ad neutraliz- in 60-mm dishes were transduced in duplicate with 106 b.f.u. of vector, ing antibodies, as described in the Materials and methods. Serial dilutions crude protein lysates prepared at various times after transduction, and of antibody were incubated with Ad2RP1050 (a) or Ad5RP1050 (b) for assayed for ␤-gal activity. The average of the duplicate samples is reported, 1 h, and assayed for transduction efficiency on A549 cells. The data from and the variation between the duplicates was less than 8%. all three mice are presented. Helper-dependent adenovirus type 2 vectors RJ Parks et al 1569 serotypes have the same general virion characteristics, against Ad2 resulted in an over 30-fold reduction in with respect to presentation of surface antigens, as first- transgene expression (4.0 × 105 versus 1.4 × 107 r.l.u. per generation Ad vectors. Based on these observations, we tissue) at day 3, and a 100-fold reduction (2.5 × 105 versus would predict that use of hdAd based on alternative Ad 2.6 × 107 r.l.u. per tissue) by day 6 after injection com- serotypes should permit vector readministration. pared with naive animals (Figure 6a). In contrast, no decrease in transgene expression compared with naive Use of hdAd of alternative serotype allows for repeat animals was observed when the re-administered vector vector administration was of a different serotype (Figure 6b). Taken together, Since mice immunized with the hdAd2 generated anti- these data indicate that the sequential use of hdAd of bodies to Ad2, and those antibodies did not cross-react alternative serotype is an effective strategy for vector with Ad5, subsequent delivery of hdAd5 to mice preim- readministration. munized with hdAd2 should lead to higher levels of transgene expression compared with readministration of Discussion hdAd2. Mice were immunized with 1010 particles of Ad2RP1046 and, 90 days later, injected with 108 b.f.u. of We have previously shown that hdAd can efficiently either Ad2RP1050 or Ad5RP1050. As a control, naive ani- transduce cells in vitro and in vivo, and can lead to long- mals were injected in parallel with the lacZ-expressing term transgene expression with dramatically reduced vectors. At 3 and 6 days after administration of the lacZ cellular and inflammatory responses, compared with vector, the animals were euthanized and the livers first-generation Ad vectors.41–43,68 Although the hdAd removed and assayed for ␤-gal activity. These experi- DNA does persist within nondividing or slowly cycling ments generated two notable results. First, the levels of cells for long periods of time, the episomal nature of Ads transgene expression in naive animals at days 3 and 6 (and hdAds) may mean that the vector DNA will eventu- for both hdAd2 (Figure 6a) and hdAd5 (Figure 6b) were ally be lost from the cell. Thus, although hdAds allow for similar (eg 1.4 × 107 ± 0.7 × 107 versus 7.5 × 107 ± 0.5 × 107 longer duration transgene expression than observed with r.l.u. per tissue at day 3 for Ad2RP1050 and Ad5RP1050, first-generation Ads, there may be a requirement for respectively). These data indicate that hdAds generated repeat vector administration in order to ‘boost’ transgene using the Ad2 helper virus are able to transduce cells in expression levels. The formation of neutralizing anti- vivo efficiently, and can lead to high levels of transgene bodies in animals exposed to Ads, which occurs due to expression, similar to Ad5-based hdAd. Second, adminis- the processing and presentation of virion proteins, would tration of a serotype 2 vector into animals immunized reduce the effectiveness of repeat administrations of hdAds, as found with first-generation Ad vectors. Use of alternative Ad serotypes, either of the same60,61 or differ- ent46,59 subgroup, allows effective repeat administration of first-generation vectors. We therefore undertook to: (1) construct and characterize a helper virus based on a different serotype than our Ad5-based helper virus; and (2) determine if sequential use of vectors derived from different serotypes could permit efficient hdAd vector readministration. Considering the overall genetic similarity of Ad2 and Ad5, it is not surprising that an Ad2-based helper virus with a floxed packaging signal was effective in amplify- ing helper-dependent Ads in the Cre/loxP system (Figure 2). We were able to produce genetically identical hdAds efficiently based on Ad2 or Ad5, and these vectors were equal in their ability to transduce cells and express transgene in vitro and in vivo (Figures 4 and 6). We would expect that other Ad serotypes with genomes similar to that of Ad5 could also be easily adapted to act as helper virus in the Cre/loxP system. In particular, Ads 1 and 6 belong to the same subgroup (C) as Ads 2 and 5, show high DNA sequence similarity to Ad5, and are relatively well characterized. Consequently, it should be straight- forward to develop additional helper viruses based on these two serotypes. However, less well characterized Ad serotypes from other subgroups, or other animal Ads (eg canine Ad) might prove more difficult to utilize as helper viruses. For example, genetic manipulation of other Ads might be hampered by inadequate molecular characteriz- Figure 6 Transgene expression from pre-immunized mice (hdAd2) using ation or lack of an E1 complementing cell line as useful either the same (hdAd2) or alternative (hdAd5) serotype. FVB/n mice were 10 as 293 cells. immunized with 10 particles of Ad2RP1046 and, 90 days later, injected Administration of a hdAd based on serotype 2 into ani- i.v. with 108 b.f.u. of Ad2RP1050 (a) or Ad5RP1050 (b). Three or 6 days after the second vector was administered, the mice were killed and the mals immunized against Ad2 resulted in a 30- to 100-fold livers assayed for ␤-gal activity. Each bar represents the average of two reduction in transgene expression compared with naive mice, and the error bars represent the maximum value. animals (Figure 6a). In contrast, no decrease in transgene Helper-dependent adenovirus type 2 vectors RJ Parks et al 1570 expression was observed when the re-administered vec- contains an approximately 22 kb fragment of eukaryotic tor was of a different serotype (Figure 6b). Taken DNA derived from the human hypoxanthine-guanine together, these data indicate that the sequential use of phosphoribosyltransferase (HPRT) gene, as described hdAd of alternative serotype is an effective strategy for elsewhere,68 pRP1050 is essentially identical to pRP1045, vector readministration. Interestingly, we did not observe but is deleted of a 1.2 kb StuI fragment from the HPRT a decrease in expression of lacZ between days 3 and 6 sequence, and has expression characteristics identical to in the hdAd2 (1st)–hdAd5 (2nd) treatment, as has been pRP1045 (RJP and FLG, unpublished results). pRP1046 is observed for similar treatments using first-generation Ad similar in structure to pRP1050, but contains the human vectors.60 The decrease in expression observed by Mack secreted alkaline phosphatase cDNA (hSEAP, Tropix) et al60 was attributed to cellular immune processes, replacing the lacZ gene. For clarity, hdAd will be desig- and this suggests that either hdAd do not elicit such nated with the appropriate serotype. For example, destructive processes or, alternatively, are poor targets. Ad5RP1050 and Ad2RP1050 are generated using Ad5LC- We have previously shown that use of hdAds can lead 8cluc and Ad2LC8cCARP, respectively. For Ad2RP1046, to prolonged transgene expression and reduced immune the titer of the vector was determined spectropho- = × 12 and inflammatory responses compared with first-gener- metrically, assuming 1A260 1.1 10 particles. ation Ad vectors.41–43 HdAds retain the other beneficial properties of Ad vectors, including virion stability during vector propagation and purification, and high transduc- Construction of a Cre/loxP helper virus based on Ad2 tion efficiency of replicating and quiescent cells, while eli- The Ad2-based helper virus, Ad2LC8cCARP was con- minating some of the obstacles and concerns that have structed using both molecular cloning and in vivo genetic been raised with respect to first- and second-generation recombination techniques (Figure 1). pLC8c was used to Ads. We have now shown that, should transgene provide the ‘left end’ of the helper virus (ie left ITR and expression levels decrease over time, the use of hdAds floxed packaging signal), and has been previously of alternative serotypes will permit readministration of a described.39 Plasmid, pLC8c, was cotransfected into 293 vector with the identical genotype. It is important to note cells with Ad2 genomic DNA digested with PshAI. The that, in our experiments, repeat administration was per- resulting viruses were screened by restriction analysis for formed with a different reporter gene than was carried those which contained the majority of protein coding by the vector used in the initial animal immunization. sequences derived from the Ad2 virus. One virus, desig- Since vector persistence (and hence transgene expression) nated Ad2LC8c, had resulted from a recombination event can be influenced by immune responses to both vector 7,11,14,43,69,70 to the left of a BspHI site located at 7882 bp on the Ad5 and transgene, the effectiveness of vector read- map. Thus, in Ad2LC8c all coding sequences for virion ministration using hdAds with different serotypes may capsid proteins, with the possible exception of pIX, are ultimately depend primarily on the immunogenicity of derived from Ad2. the therapeutic gene. The results of this study demon- Because of the requirement for multiple serial passages strate that, in the absence of transgene effects, the sequen- in order to increase the titer of hdAd, the formation of tial use of hdAd of alternative serotype is an effective replication competent adenovirus (RCA) is of concern. strategy for vector readministration. Once generated, RCA can rapidly outgrow the vector, leading to highly contaminated vector stocks. We have Materials and methods shown that the inclusion of a ‘stuffer’ segment within the E3 region, such that recombination between the helper Cell and virus culture virus and the Ad5 sequences contained in the 293 or All cell culture media and reagents were obtained from 293Cre cells results in a virus which is above the upper Gibco Laboratories (Grand Island, NY, USA). 29371 and packaging limit for Ad virions (approximately 105% of the wild-type genome),66 can eliminate the potential for A549 (human lung carcinoma, ATCC CCL 185) cells were 39 grown in monolayer in F-11 minimum essential medium RCA. We therefore designed an Ad2-based stuffer plas- supplemented with 100 U of penicillin per ml, 100 mg of mid which contained a fragment of lambda DNA located streptomycin per ml, 2.5 mg fungizone per ml, and 10% within the E3 region as follows (Figure 1b). pFG28, which fetal bovine serum for cell maintenance or 5% horse contains the right 40 map units of Ad2 (FLG, serum after virus infection. Recombinant Ad helper unpublished), was digested with PacI and ligated with viruses were grown and titered on 293 cells, as pre- PacI-digested pCARPABS1, which contains a 5.6 kb frag- viously described.72 The 293-derived cell line that stably ment of lambda DNA (22346–27972 bp of the conven- expresses the Cre recombinase, 293Cre4,63 was propa- tional lambda map) cloned into the unique BamHI site of gated in complete F-11 medium supplemented with 0.4 pABS1 (Neor).73 The resulting plasmid, pFG28CARP, was mg/ml G418. partially digested with HpaI, resulting in the loss of the Helper-dependent Ad vectors were propagated and neomycin resistance gene, part of the lambda DNA, and titered as previously described.39 pRP1045 is a hdAd other bacterial sequences derived from pABS1, and recir- deleted of all Ad protein coding sequences, but contain- cularized, generating pFG28CARPc. To transfer the ing an Ad5 head-to-tail inverted terminal repeat (ITR) stuffer segment to Ad2LC8c, DNA isolated from junction and packaging signal, as well as the E. coli ␤- Ad2LC8c virions was digested with BspHI and cotrans- galactosidase gene under the regulation of the murine fected with pFG28CARPc into 293 cells. The resulting cytomegalovirus immediate–early promoter (MCMV) viruses were screened by restriction analysis for those and Simian virus 40 polyadenylation (pA) sequence. To containing the lambda stuffer segment within the E3 maintain the size of the vector above the limit for efficient region, and one positive isolate, designated DNA packaging (approximately 28 kb),67 pRP1045 also Ad2LC8cCARP, was used for subsequent experiments. Helper-dependent adenovirus type 2 vectors RJ Parks et al 1571 Transgene expression studies gene product in the elimination of recombinant adenovirus- Methods for preparation of cell samples and assays for infected hepatocytes in vivo. Gene Therapy 1996; 3: 137–144. ␤-gal are described elsewhere.68 For in vivo expression 7 Dai Y et al. Cellular and humoral immune responses to adenovi- studies, adult female FVB/n mice (Harlan) were injected rual vectors containing factor IX gene: tolerization of factor IX and vector antigens allows for long-term expression. Proc Natl through the tail vein with the indicated vector dose in a ␮ Acad Sci USA 1995; 92: 1401–1405. volume of 200 l. For antibody titer determinations, 8 Gilgenkrantz H et al. Transient expression of genes transferred blood samples were removed by orbital bleed, incubated in vivo into heart using first-generation adenoviral vectors: role overnight at 4°C, and the serum cleared by two rounds of the immune response. Hum Gene Ther 1995; 6: 1265–1274. of centrifugation at 16 000 g for 5 min in a microcentri- 9 McCoy RD et al. Pulmonary inflammation induced by incom- fuge. The serum samples were then stored at −70°C. The plete or inactivated adenoviral particles. Hum Gene Ther 1995; 6: method for preparation and analysis of ␤-gal levels in the 1553–1560. liver of mice is described elsewhere.68 10 Morral N et al. Immune responses to reporter proteinsa nd high viral dose limit duration of expressiion with adenoviral vectors: comparison of E2a wild type and E2a deleted vectors. Hum Gene Virus neutralization assays Ther 1997; 8: 1275–1286. The Ad5 neutralizing antibodies used in these experi- 11 Christ M et al. Gene therapy with recombinant adenovirus vec- ments were generated in rabbits by injection of a first- tors: evaluation of the host immune response. Immunol Lett 1997; generation Ad5 vector (M Anton and FL Graham, 57: 19–25. unpublished) and Ad2 antibodies were made in mice by 12 Kafri T et al. Cellular immune response to adenoviral vector injecting a serotype 2 hdAd. Aliqouts (106 b.f.u. in 100 ␮l) infected cells does not require de novo viral gene expression: of a first-generation Ad vector (Ad5CA35),74 Ad5RP1045, implications for gene therapy. Proc Natl Acad Sci USA 1998; 95: AdRP1050 or Ad2RP1050, all containing an identical 11377–11382. ␤-gal expression cassette, were incubated with serial 13 van Ginkel FW et al. Adenoviral gene delivery elicits distinct pulmonary-associated T helper cell responses to the vector and dilutions (100 ␮l) of antibody-containing serum. After a ° to its transgene. J Immunol 1997; 159: 685–693. 1 h incubation at 37 C, the treated vectors were used to 14 Dong JY et al. Systematic analysis of repeated gene delivery into infect 22-mm dishes of A549 cells for 1 h, the monolayers animal lungs with a recombinant adenovirus vector. Hum Gene washed twice with PBS, maintenance medium replaced, Ther 1996; 7: 319–331. and the quantity of ␤-gal present within the cells assayed 15 Kaplan JM et al. Humoral and cellular immune responses of 24 h later. In this assay, the quantity of ␤-gal produced nonhuman primates to long-term repeated lung exposure to in the cells correlates directly with the efficiency of cell Ad2/CFTR-2. Gene Therapy 1996; 3: 117–127. transduction. 16 St George JA et al. Biological response of nonhuman primates to long-term repeated lung exposure to Ad2/CFTR-2. Gene Therapy 1996; 3: 103–116. Acknowledgements 17 Schulick AH et al. Established immunity precludes adenovirus- mediated gene transfer in rat carotid arteries. Potential for We thank Uma Sankar for excellent technical assistance, immunosuppression and vector engineering to overcome bar- and Jonathan Bramson for helpful discussions through- riers of immunity. J Clin Invest 1997; 99: 209–219. out the course of these studies. This work was supported 18 Lee MG et al. The constitutive expression of the immunomodul- − − by grants from the US National Institutes of Health, atory gp19k protein in E1 ,E3 adenoviral vectors strongly Natural Sciences and Engineering Research Council reduces the host cytotoxic T cell response against the vector. (NSERC), Medical Research Council (MRC), and the Gene Therapy 1995; 2: 256–262. 19 Poller W et al. Stabilization of transgene expression by incorpor- National Cancer Institute of Canada (NCIC), and by ation of E3 region genes into an adenoviral factor IX vector and Merck Research Laboratories. RJP was an NSERC Post- by transient anti-CD4 treatment of the host. Gene Therapy 1996; doctoral Fellow (PDF) and is currently an MRC PDF. FLG 3: 521–530. is a Terry Fox Research Scientist of the NCIC. All animal 20 Bruder JT et al. Expressionof gp19K increases the persistence of experiments were approved by and performed according transgene expression from an adenovirus vector in the mouse to the guidelines set by the Animal Research Ethics Board lung and liver. J Virol 1997; 71: 7623–7628. of McMaster University. 21 Ilan Y et al. Insertion of the adenoviral E3 region into a recombinant viral vector prevents antiviral humorla and cellular immune responses and permits long-term gene expression. Proc References Natl Acad Sci USA 1997; 94: 2587–2592. 22 Schowalter DB et al. Heterologous expression of adenovirus E3- 1 Hitt MM, Addison CL, Graham FL. Human adenovirus vectors gp19K in an E1a-deleted adenovirus vector inhibits MHC I for gene transfer into mammalian cells. Adv Pharmacol 1997; 40: expression in vitro, but does not prolong transgene expresssion 137–206. in vivo. Gene Therapy 1997; 4: 351–360. 2 Yang Y et al. Cellular immunity to viral antigens limits E1- 23 Engelhardt JF et al. Ablation of E2A in recombinant adeno- deleted adenoviruses for gene therapy. Proc Natl Acad Sci USA viruses improves transgene persistence and decreases inflam- 1994; 91: 4407–4411. matory response in mouse liver. Proc Natl Acad Sci USA 1994; 3 Yang Y, Wilson JM. Clearance of adenovirus-infected hepato- 91: 6196–6200. cytes by MHC class I-restricted CD4+ CTLs in vivo. J Immunol 24 Yang Y et al. Inactivation of E2a in recombinant adenoviruses 1995; 155: 2564–2570. improves the prospect for gene therapy in cystic fibrossis. Nat 4 Yang Y et al. Upregulation of class I major histocompatibility Genet 1994; 7: 362–369. complex antigens by interferon gamma is necessary for T-cell- 25 Goldman MJ et al. Transfer of the CFTR gene to the lung of mediated elimination of recombinant adenovirus-infected hepa- nonhuman primates with E1-deleted, E2a-defective recombinant tocytes in vivo. Proc Natl Acad Sci USA 1995; 92: 7257–7261. adenoviruses: a preclinical toxicology study. Hum Gene Ther 5 Yang Y, Su Q, Wilson JM. Role of viral antigens in destructive 1995; 6: 839–851. cellular immune responses to adenovirus vector-transduced 26 Gao GP, Yang Y, Wilson JM. Biology of adenovirus vectors with cells in mouse lungs. J Virol 1996; 70: 7209–7212. E1 and E4 deletions for liver-directed gene therapy. J Virol 1996; 6 Yang Y et al. Immune responses to viral antigens versus trans- 70: 8934–8943. Helper-dependent adenovirus type 2 vectors RJ Parks et al 1572 27 Dedieu JF et al. Long-term gene delivery into the livers of immu- vivo gene transfer to mouse synovium. Hum Gene Ther 1996; 7: nocompetent mice with E1/E4-defective adenoviruses. J Virol 499–506. 1997; 71: 4626–4637. 50 Smith TA et al. Transient immunosuppresion permits successful 28 Wang Q et al. Persistent transgene expression in mouse liver repetitive intravenous administration of an adenovirus vector. following in vivo gene transfer with a delta E1/delta E4 aden- Gene Therapy 1996; 3: 496–502. ovirus vector. Gene Therapy 1997; 4: 393–400. 51 Yang Y, Greenough K, Wilson JM. Transient immune blockade 29 Amalfitano A et al. Production and characterization of improved prevents formation of neutralizing antibody to recombinant adenovirus vectors with E1, E2b, and E3 genes deleted. J Virol adenovirus and allows repeated gene transfer to mouse liver. 1998; 72: 926–933. Gene Therapy 1996; 3: 412–420. 30 Fang B et al. Lack of persistence of E1− recombinant adenoviral 52 Zepeda M, Wilson JM. Neonatal cotton rats do not exhibit vectors containing a temperature-sensitive E2A mutation in dstructive immune responses to adenoviral vectors. Gene Ther- immunocompetent mice and hemophilia B dogs. Gene Therapy apy 1996; 3: 973–979. 1996; 3: 217–222. 53 Kaplan JM, Smith AE. Transient immunosuppression with 31 Lusky M et al. In vitro and in vivo biology of recombinant aden- deoxyspergualin improves longevity of transgene expression ovirus vectors with E1, E1/E2A, or E1/E4 deleted. J Virol 1998; and ability to readminister adenoviral vector to the mouse lung. 72: 2022–2032. Hum Gene Ther 1997; 8: 1095–1104. 32 Mitani K et al. Rescue, propagation, and partial purification of 54 Kuzmin AI, Finegold MJ, Eisensmith RC. Macrophage depletion a helper virus-dependent adenovirus vector. Proc Natl Acad Sci increasess the safety, efficacy and persistence of adenovirus- USA 1995; 92: 3854–3858. mediated gene transfer in vivo. Gene Therapy 1997; 4: 309–316. 33 Fisher KJ et al. Recombinant adenovirus deleted of all viral genes 55 Lieber A et al. The role of Kupffer cell activation and viral gene for gene therapy of cystic fibrosis. Virology 1996; 217: 11–22. expression in early liver toxicity after infusion of recombinant 34 Haecker SE et al. In vivo expression of full-length human dystro- adenovirus vectors. J Virol 1997; 71: 8798–8807. phin from adenoviral vectors deleted of all viral genes. Hum 56 Scaria A et al. Antibody to CD40 ligand inhibits both humoral Gene Ther 1996; 7: 1907–1914. and cellular immune responses to adenoviral vectors and facili- 35 Kochanek S et al. A new adenoviral vector: replacement of all tates repeated administration to mouse airway. Gene Therapy viral coding sequences with 28 kb of DNA independently 1997; 4: 611–617. expressing both full-length dystrophin and beta-galactosidase. 57 Wolff G et al. Enhancement of in vivo adenovirus-mediated gene Proc Natl Acad Sci USA 1996; 93: 5731–5736. transfer and expresssion by prior depletion of tissue macro- 36 Kumar-Singh R, Chamberlain JS. Encapsidated adenovirus min- phages in the target organ. J Virol 1997; 71: 624–629. ichromosomes allow delivery and expression of a 14 kb dystro- 58 Zsengeller ZK et al. Anti-T cell receptor antibody prolongs trans- phin cDNA to muscle cellss. Hum Mol Genet 1996; 5: 913–921. gene expression and reduces lung inflammation after aden- 37 Lieber A et al. Recombinant adenoviruses with large deleteions ovirus-mediated gene transfer. Hum Gene Ther 1997; 8: 935–941. generated by Cre-mediated excision exhibit different biological 59 Mastrangeli A et al. ‘Sero-switch’ adenovirus-mediated in vivo properties compared with first-generation vectors in vitro and gene transfer: circumvention of anti-adenovirus humoral in vivo. J Virol 1996; 70: 8944–8960. immune defenses against repeat adenovirus vector adminis- 38 Hardy S et al. Construction of adenovirus vectors through Cre- tration by changing the adenovirus serotype. Hum Gene Ther lox recombinations. J Virol 1997; 71: 1842–1849. 1996; 7: 79–87. 39 Parks RJ et al. A helper-dependent adenovirus vector system: 60 Mack CA et al. Circumvention of anti-adenovirus neutralizing removal of helper virus by Cre-mediated excision of the viral immunity by administration of an adenoviral vector of an alter- packaging signal. Proc Natl Acad Sci USA 1996; 93; 13565–13570. nate serotype. Hum Gene Ther 1997; 8: 99–109. 40 Chen HH et al. Persistence in muscle of an adenoviral vector that 61 Roy S et al. Circumvention of immunity to the adenovirus major lacks all viral genes. Proc Natl Acad Sci USA 1997; 94: 1645–1650. coat protein hexon. J Virol 1998; 72: 6875–6879. 41 Morsy MA et al. An adenoviral vector deleted for all viral coding 62 Smith CA et al. Extensive cross-reactivity of adenovirus-specific sequencces results in enhanced safety and extended expression cytotoxic T cells. Hum Gene Ther 1998; 9: 1419–1427. of a leptin transgene. Proc Natl Acad Sci USA 1998; 95: 7866–7871. 63 Chen L, Anton M, Graham FL. Production and characterization 42 Schiedner G et al. Genomic DNA transfer with a high-capacity of human 293 cell lines expressing the site-specific recombinase adenovirus vector results in improved in vivo gene expression Cre. Somat Cell Mol Genet 1996; 22: 477–488. and decreased toxicity. Nat Genet 1998; 18: 180–183. 64 Gahery-Segard H et al. Immune response to recombinant capsid 43 Morral N et al. High doses of a helper-dependent adenoviral proteins of adenovirus in humans: antifiber and anti-penton vector yield supraphysiological levels of alpha1-antitrypsin with base antibodies have a synergistic effect on neutralizing activity. negligible toxicity. Hum Gene Ther 1998; 9: 2709–2716. J Virol 1998; 72: 2388–2397. 44 Vilquin JT et al. FK506 immunosuppression to control the 65 Wohlfart C. Neutralization of adenoviruses: kinetics, stoichi- immune reactions triggered by first-generation adenovirus- ometry, and mechanisms. J Virol 1988; 62: 2321–2328. mediated gene transfer. Hum Gene Ther 1995; 6: 1391–1401. 66 Bett AJ, prevec L, Graham FL. Packaging capacity and stability 45 Jooss K, Yang Y, Wilson JM. Cyclophosphamide diminishes of human adenovirus type 5 vectors. J Virol 1993; 67: 5911–5921. inflammation and prolongs transgene expression following 67 Parks RJ, Graham FL. A helper-dependent system for aden- delivery of adenoviral vectors to mouse liver and lung. Hum ovirus vector production helps define a lower limit for efficient Gene Ther 1996; 7: 1555–1566. DNA packaging. J Virol 1997; 71: 3293–3298. 56 Kass-Eisler A et al. Circumventing the immune response to 68 Parks RJ et al. Effect of stuffer DNA and viral protein coding adenovirus-mediated gene therapy. Gene Therapy 1996; 3: 154– sequences on transgene expression from first generation and 162. helper-dependent adenoviral vectors 1999 (submitted). 47 Kolls JK et al. Use of transient CD4 lymphocyte depletion to 69 Michou AI et al. Adenovirus-mediated gene transfer: influence prolong transgene expression of E1-deleted adenoviral vectors. of transgene, mouse strain and type of immune response on per- Hum Gene Ther 1996; 7: 489–497. sistence of transgene expression. Gene Therapy 1997; 4: 473–482. 48 Lochmuller H et al. Transient immunosuppression by FK506 70 Tripathy SK et al. Immune responses to transgene-encoded pro- permits a sustained high-level dystrophin expression after aden- teins limit the stability of gene expression after injection of repli- ovirus-mediated dystrophin minigene transfer to skeletal cation-defective adenovirus vectors. Nature Med 1996; 2: 545– muscles of adult dystrophic (mdx) mice. Gene Therapy 1996; 3: 550. 706–716. 71 Graham FL et al. Characteristics of a human cell line transfor- 49 Sawchuk SJ et al. Anti-T cell receptor monoclonal antibody pro- med by DNA from human adenovirus type 5. J Gene Virol 1997; longs transgene expression following adenovirus-mediated in 36: 59–74. Helper-dependent adenovirus type 2 vectors RJ Parks et al 1573 72 Hitt M et al. Techniques for human adenovirus vector construction 74 Addison CL et al. Comparison of the human versus murine cyto- and characterization. Meth Mol Genet 1995; 7: 13–30. megalovirus immediate early gene promoters for transgene 73 Bett AJ. Construction and characterization of recombinant aden- expression by adenoviral vectors. J Gen Virol 1997; 78: 1653– ovirus type 5 vectors. Biology. McMaster: Hamilton, 1995. 1661.