Therapy (2000) 7, 1707–1714  2000 Macmillan Publishers Ltd All rights reserved 0969-7128/00 $15.00 www.nature.com/gt VIRAL TRANSFER TECHNOLOGY REVIEW ARTICLE Production of first generation adenovirus vectors: a review

X Danthinne1 and MJ Imperiale2 1O.D. 260 Inc, Mountain States Medical Research Institute, and VA Medical Center, Boise, ID; and 2Department of and Immunology and Center for , University of Michigan Medical School, Ann Arbor, MI, USA

In the past decade, adenovirus vectors have generated other settings, such as cancer and vaccination, or simply to tremendous interest, especially in gene therapy applications. transfect cell lines that are refractory to other transfection In the so-called ‘first generation’ adenovirus vectors, the methods. Until a few years ago, the construction of first gen- transgenes are inserted in place of the E1 region, or less eration adenovirus vectors was a labor-intensive and time- often the E3 region. Although second-generation and helper- consuming process. More than 20 methods have appeared dependent adenovirus vectors will probably prevail in the that facilitate their construction and are reviewed below. future in applications that require long-term gene expression, Gene Therapy (2000) 7, 1707–1714. first generation adenovirus vectors will remain very useful in

Keywords: adenovirus; E1; viral vectors; gene therapy; methods

Introduction Adenovirus vectors have become a very popular tool for gene transfer into mammalian cells.1–7 They have numer- ous advantages: they can infect a wide variety of dividing or nondividing cells; they are easily purified to high tit- ers; the strains commonly used to construct recombinant (Ad2, Ad5) are well characterized; they can accommodate up to 37 kb foreign genetic material;8 their genome rarely integrates into the host chromosome, which is suitable for applications requiring transient gene expression;9 and hybrid adenovirus/retrovirus or adenovirus/adeno-associated vectors can be used to achieve integration of transgenes into the genome and long-term gene expression.10,11 Over the years, many methods for manipulating the viral genome have been developed. We will review the approaches that are used to produce standard, so-called first generation adeno- Figure 1 Genome structures of first generation, second generation and virus vectors. gutted adenovirus vectors. The locations of the early and late Three classes of adenovirus vectors are currently being units (arrows) in the adenovirus genome (black bar) are shown on top of developed for gene therapy purposes (Figure 1). First the figure. The ITRs are symbolized by triangles. The packaging signal ⌿ generation adenovirus vectors are made by substituting is highlighted by an open box. Nonadenoviral sequences in gutless vectors are represented by a thin line. The sizes of the largest deletions are indi- an expression cassette for the E1 and/or E3 regions. The cated for each region, and the maximal insert sizes are indicated for each E1 region, located at the left end of the 36 kb adenovirus type of vector. genome, encodes proteins necessary for the expression of the other early and late . Various deletions of this region have been made, with a maximum of 3150 bp. are not essential for in vitro, and there- Since the E1 products are necessary for viral growth, they fore no complementing cell line is necessary.15 However, must be provided in trans in specific cell lines, such as in some applications, it might be desirable to retain or 12 13 14 293, 911, or PER.C6. The E3 region encodes products even increase the expression of some E3 products: for that counteract host defense mechanisms; these products instance, the adenovirus ‘death protein’ E3–11.6K, which facilitates the release of viral particles from infected cells,16 or gp19K, whose constitutive expression reduces Correspondence: MJ Imperiale, Department of Microbiology and Immu- 17 nology, University of Michigan Medical School, 1500 E Medical Center the host cytotoxic T cell response against the vector and Drive, 6310 Cancer Center, Ann Arbor, MI 48109-0942, USA increases the persistence of transgene expression on its Received 12 March 2000; accepted 11 July 2000 own,18 but possibly not in the context of constitutive Production of first generation adenovirus vectors X Danthinne and MJ Imperiale 1708 expression of the entire E3 region.19 Up to 3.1 kb of the E3 restriction fragments that correspond to the left end of region have been deleted.20 Since adenovirus can package the genome and are carried over during the transfection approximately 38 kb without affecting growth rate and can religate inside the cell and generate replication-com- viral titer,8 E1-deleted adenoviruses can accept insertions petent adenovirus.43,44 Therefore two or three rounds of of up to 5.1 kb, while E1/E3-deleted viruses allow the purification by plaque assay must be performed, and cloning of about 8.2 kb. multiple plaques must be analyzed to identify the recom- First-generation adenovirus vectors elicit a significant binant virus. A second disadvantage of this method is immune response in vivo, mainly due to the de novo syn- that the recombination event, which is required to link thesis of viral proteins.21,22 Therefore, additional genes the left adenovirus sequences carrying the gene of inter- necessary for viral DNA replication have been inacti- est to the right end sequences, is inefficient in mam- vated, giving rise to the second generation of adenovirus malian cells. Therefore it usually takes up to 2 weeks for vectors. Various cell lines have been constructed that the recombinant virus to appear. express the E2a DNA-binding protein,23–25 the E2b-enco- Three methods have been developed to facilitate the ded terminal protein and viral DNA polymerase,26–28 or detection and purification of the recombinant virus. By all or most of the E4 products.29–35 The corresponding using a viral DNA in which the E1 region is replaced by a deletions on the viral genome should allow the insertion cassette expressing the HSV thymidine kinase (tk) gene,45 of expression cassettes up to 14 kb. A third class of counter-selection against the parental background virus adenovirus vectors is represented by the so-called ‘gut- can be performed by adding ganciclovir to the agar over- ted’ or ‘helper-dependent’ adenoviruses.36–41 These vec- lay. This counter-selection is reliable only on a secondary tors are deleted for all viral genes, but retain the cis-acting plaque assay, however, as the primary plaques obtained sequences necessary for viral DNA replication and pack- directly after transfection may originate from cells that aging. Such viruses can in theory accommodate multiple have taken up multiple viral genomes and contain both transgenes, totaling about 37 kb. However their the tk-expressing parental virus and the recombinant production is tedious and needs to be optimized. virus. Similarly, viral DNA in which the E1 region is First generation adenovirus vectors remain an invalu- replaced by a cassette expressing E. coli ␤-galactosidase43 able tool for gene transfer. Their transgene capacity is or a green fluorescent protein46 can be used as parental compatible with most applications and they are easy to DNA. Plaques containing background virus are detected grow to high titers. They can be used not only in gene either by staining with X-gal, or by fluorescence therapy, particularly in settings in which the immune microscopy, respectively. These two methods reportedly response is not an issue such as cancer, but also to allow the isolation of the recombinant virus during the express recombinant proteins for vaccination purposes, first round of purification. It would be more cautious or simply to transduce cell lines that are refractory to however, and proper virological procedure, to perform a other transfection methods. Until a few years ago, the second round of plaque purification, as suggested for the construction of adenovirus vectors was a time-consuming tk-expressing virus. process, requiring 2 to 3 months of intensive labor. The The adenovirus DNA–terminal protein complex can past few years have seen the proliferation of techniques generate viral plaques with an efficiency two to three which facilitate the construction of first generation vec- orders of magnitude higher than naked DNA prepared tors, and which are reviewed below. with proteinase. For decades, such complexes have been used to construct recombinant adenoviruses.47–50 More Methods requiring DNA recombination in recently, a technique has been adapted in which the DNA–terminal protein complex is digested with a restric- mammalian cells tion enzyme at several sites in the Ad5 genome, better The classical method to generate E1-substituted aden- preventing the regeneration of the parental virus.51 The ovirus vectors requires recombination between two DNA expression cassette is cloned into a cosmid containing a molecules, one carrying sequences mapping to the very full-length copy of the adenovirus genome. Both the cir- left end of the adenovirus genome and the gene of inter- cular cosmid and the digested viral DNA are transfected est, and the other carrying sequences that slightly overlap into helper cells. A few hundred plaques, which arise by the 3Ј-most viral sequences on the first molecule and con- recombination of the full-length cosmid-derived genome tinue to the right end of the adenovirus genome. The lat- with the two terminal protein-linked genome ends, are ter, which we will refer to as right end sequences for sim- usually obtained using this procedure, among which plicity, can be either a linear DNA purified from virions, about 70% are positive for the presence of the or a plasmid. This recombination takes place inside E1- expression cassette. expressing cells such as 293 cells, generating the desired To overcome the inefficiency of homologous recombi- recombinant viral DNA. nation in mammalian cells, a technique has been Originally, the right end of the adenovirus genome was developed that uses the P1 Cre-lox recom- obtained by digesting viral DNA with a restriction bination system.52 This system is composed of three enzyme(s) that cuts in the E1 region (Figure 2a).42 Because elements: (1) a recombinant adenovirus that contains two of the small difference in size between the undigested loxP sites flanking the packaging signal; (2) a shuttle vec- and digested viral , it is difficult to verify the com- tor which contains the left ITR, the packaging signal, the pleteness of the digest by agarose gel electrophoresis or expression cassette and a loxP site; (3) a 293-derived cell to separate the two DNAs for purification. This can be a line which expresses the phage P1 Cre recombinase. To problem because undigested viral DNA will generate generate E1-substituted adenovirus vectors, the shuttle virus more efficiently than the recombination product, vector carrying the gene of interest and the viral DNA leading to the contamination of the recombinant virus are co-transfected into the helper cells. First, an intramol- preparation with the parental virus. Moreover, the small ecular recombination occurs between the two loxP sites

Gene Therapy Production of first generation adenovirus vectors X Danthinne and MJ Imperiale 1709

Figure 2 (a) Construction of E1-substituted adenovirus vectors by classical recombination in helper cells. See text for explanations and Figure 1 legend for symbol definitions. The double arrow represents the expression cassette. Hollow and black viruses at the bottom of the panel indicate parental and recombinant viruses, respectively. (b) Construction of E1-substituted adenovirus vectors from infectious circular adenovirus DNAs. Ori, Amp: colE1 origin of replication and ampicillin-resistance gene, respectively. The tables summarize the main characteristics of both techniques and their derivatives. In each table, features that are mentioned more than once are referred to by numbers in parenthesis.

present in the viral DNA. This generates an adenovirus virus is not possible, and therefore a limited number of genome which is able to replicate and express viral genes, plaques needs to be analyzed. but is unable to be packaged. A second recombination Another approach to construct recombinant adeno- occurs between the loxP site of this product and that on virus vectors relies on the ability of circular adenovirus the shuttle vector, generating the desired recombinant DNAs to generate virus upon transfection into 293 cells virus. The main disadvantage of this method is the per- (Figure 2b).20,54 These circular viral DNAs contain an sistence of the parental virus in the preparation, which ampicillin resistance gene and a bacterial origin of repli- can represent up to 30% of the viral population after the cation in place of the E1 region, and are therefore able to first passage, and still constitutes 0.2% after three replicate in E. coli as plasmids. These vectors have been passages. Therefore, this method requires a careful veri- subsequently modified to increase their size to about 40 fication of the identity of the recombinant virus. kb, or have been deleted for their packaging signal. These A very recent improvement of this technique was features render the parental molecules unable to be pack- obtained by using adenovirus DNA isolated from plas- aged into virions, and therefore a homogeneous recombi- mids instead of viral particles.53 A large plasmid was con- nant virus preparation should be obtainable. In this sys- structed that contains the Ad5 backbone devoid of the tem, the gene of interest is cloned into a small plasmid left ITR, the packaging signal, and the E1 sequences. This containing the left end of the adenovirus genome. The plasmid, linearized at a site flanking the right ITR, is co- resulting plasmid, linearized, is co-transfected into helper transfected into helper cells with a shuttle plasmid that cells with the circular adenovirus plasmid. Recombi- contains the left adenovirus sequences and the expression nation must occur between both DNAs to create a mol- cassette. Virus is generated upon homologous DNA ecule that can be replicated and packaged to generate the recombination between both DNA molecules. Unlike the recombinant virus. The same technique can be used to above-described methods, contamination with parental insert expression cassettes in the E3 region, or in both the

Gene Therapy Production of first generation adenovirus vectors X Danthinne and MJ Imperiale 1710 E1 and E3 regions. However, manipulation of these large recombinant viral DNA in vitro using the Cre-lox recom- plasmids in E. coli is rendered difficult because of the bination system (Figure 3a).60 The expression cassette is presence of a 200-bp long palindromic sequence that cloned into a small intermediate vector that contains the results from the juxtaposition of the left and right adeno- left ITR, packaging signal, a multiple cloning site, and a virus ITRs in the plasmid construct. Such structures are loxP site. The resulting plasmid is linearized at a site known to be unstable in E. coli and deleterious for bac- flanking the left ITR, and incubated with Cre recombi- terial growth.55 Moreover, the formation of plaques upon nase and a linearized cosmid DNA which contains transfection of circular adenovirus DNA has been another loxP site in place of the E1 region. Since the reported to be inefficient,41 probably because imbedded cosmid DNA lacks the left ITR and packaging signal, adenovirus ITRs initiate DNA synthesis less efficiently only the recombination products are able to generate than those at the ends of linear DNA.56 A 293 cell line virus after transfection in helper cells. Recombinant virus constitutively expressing the adenovirus preterminal pro- can be effectively obtained 7–9 days after transfection. tein and DNA polymerase can be used to overcome this Three methods have been published that describe the replication problem.41 generation of recombinant adenovirus by recombination As for the techniques using linear viral DNAs, this in E. coli.30,61,62 All three methods involve two elements, method was improved using a large plasmid containing the entire adenovirus genome the Cre-lox recombination system. In a first embodiment, (or just the right end), and a small shuttle plasmid con- loxP sites were inserted in both the large circular aden- taining the expression cassette flanked by sequences ovirus DNA and the small shuttle plasmid.57 Recombi- homologous to the region to be targeted. In the technique nant adenovirus is obtained as a result of Cre-mediated developed by Chartier et al,61 the large adenovirus plas- recombination between both plasmids after their cotrans- mid is linearized at a unique site located in the E1 region fection into a 293 cell line expressing Cre recombinase. or near the E3 region. This DNA is introduced into a The frequency of virus rescue using this specific recombi- RecA+ E. coli strain together with a DNA fragment con- nation system is significantly higher (approximately 30- taining the gene of interest flanked by E1 or E3 fold) than by in vivo homologous recombination. In a sequences. Recombination between both DNA molecules second embodiment, the E1 region of the circular form generates a plasmid containing the expression cassette in of adenovirus DNA is replaced with a cosmid vector the desired region, with an efficiency ranging from 20 flanked by loxP sites.58 As above, this replacement pro- to 100%. Background colonies arise when the adenovirus duces a DNA molecule too large to be packaged into plasmid is not completely linearized or when the purified virions. Expression cassettes are inserted between the fragment is contaminated with the vector backbone car- loxP-flanked cosmid backbone and the adenovirus gen- rying an ampicillin-resistance gene. Transformants are ome via packaging into phage ␭. The resulting cosmid is screened for the presence of the gene of interest and the transfected into helper cells that express the Cre recombi- recombinant plasmid is transferred to a standard RecA− nase. An intramolecular Cre-lox mediated recombination E. coli strain in order to isolate larger quantities of plas- excises the cosmid vector backbone and produces a mol- mid DNA. The plasmid is finally digested to release a ecule that can be packaged into virions. Using this linear adenovirus DNA, and transfected into helper cells. method, viral plaques are observed between 8 and 10 This method is very powerful for introducing expression days after transfection. cassettes in the E1 or E3 region, or mutations in any region of the genome as long as a unique restriction site is present in that region.63 Methods reconstructing the recombinant A very similar method was designed to introduce adenovirus genome before transfection into expression cassettes specifically into the E1 region, and mammalian cells facilitate the recovery of the recombinant E. coli clones.30 The large adenovirus plasmid is introduced into E. coli A second category of methods is based on the construc- as a supercoiled DNA in order to increase transformation tion of the recombinant adenovirus genome before trans- efficiency, and a kanamycin-resistance gene, provided by fection into helper cells. All these methods but one59 have the shuttle vector containing the expression cassette, is the advantage that the transfections are performed using used to select the recombinant clones. With this tech- DNAs isolated from bacterial clones, and generate virus nique, about 60% of the kanamycin-resistant colonies preparations devoid of parental virus. The only potential contain the desired recombinant plasmid. As above, concern is the absence of selective pressure for the aden- recombinant plasmid DNA is purified, and the viral ovirus sequences in E. coli, which could lead to mutations chromosome is released by restriction digestion and that impair the viability of the virus. transfected into the appropriate cell line. In vitro DNA ligation was already in use two decades The third recombination method in E. coli involves a ago to construct recombinant adenoviruses.59 A large two-step gene replacement protocol.62 A co-integrate DNA fragment obtained by digesting virion DNA in the plasmid is first obtained by recombination between a E1 region is ligated with a plasmid DNA fragment con- large adenovirus plasmid and a small shuttle vector that taining the adenovirus left ITR, the packaging signal, and contains the gene of interest, and selected for in the pres- the expression cassette. Transfection of the ligation pro- ence of two antibiotics. The final recombinant plasmid is ducts into helper cells results in the production of the generated via a second homologous recombination event recombinant virus. However, this technique is prone to that excises from the co-integrate B. subtilis sacB contamination of the virus progeny by the parental virus sequences, which encodes a protein that kills the and therefore careful virus purification by plaque assay bacterium in the presence of sucrose. The technique is essential. generates a 50–50 mixture of parental and recombinant More recently, a method was designed to construct the plasmids. It therefore requires a thorough verification of

Gene Therapy Production of first generation adenovirus vectors X Danthinne and MJ Imperiale 1711

Figure 3 Construction of E1-substituted adenovirus vectors by Cre-lox recombination in vitro (a) or cosmid construction (b). See text for explanations and Figure 1 and 2 legends for symbol definitions. The phage P1 loxP site is represented by a black thick arrow. cos, kan: phage ␭ cohesive site and kanamycin-resistance gene, respectively.

the recombinant adenovirus plasmid by restriction or is achieved with an efficiency approaching 50%. In the Southern analysis before isolation of the viral genome second method,65,66 a counter-selection against the par- and transfection into cells. Compared to the two methods ental adenovirus plasmid is performed by digesting the described above, this technique does not require the lin- ligation reaction with a third enzyme. This raises the earization of either the adenovirus or the shuttle plasmid cloning efficiency to 90%. The recombinant adenovirus before transformation. Moreover, this method has the plasmid is amplified, digested to expose the inverted ter- advantage of allowing insertion of mutations or hetero- minal repeats and then transfected into helper cells to res- logous sequences at any location in the adenovirus cue the virus. Viral plaques are visible after 7 to 10 days. genome. Since transformation of E. coli with large plasmids gen- A few techniques have been designed to insert erates a background of smaller plasmids due to expression cassettes in adenovirus plasmids by ligation rearrangements, other techniques have been designed rather than recombination. These procedures are more that reconstitute the sequence of the recombinant virus straightforward since only one E. coli strain is used for in a cosmid. This technology is particularly well suited both plasmid construction and amplification, and there for the cloning of the 36-kb long adenovirus genome, is less danger of unpredicted recombination events that since phage lambda preferentially packages DNAs rang- might go undetected and generate noninfectious viral ing from 38 to 51 kb.67 In a first approach, the cosmid DNA. Two similar methods use large plasmids which containing the sequence of the recombinant virus is contain the entire sequence of the adenovirus genome, directly transfected into helper cells, apparently without with one or two unique recognition sites for intron-enco- linearization of the DNA before transfection.68 This ded endonucleases in place of the E1 region.64–66 The method is inefficient in generating plaques, probably expression cassettes are cloned into small shuttle vectors because of the inefficiency of replicating viral genomes that contain these specific sites, and are subsequently whose ends are imbedded.56 To overcome this problem, transferred into the large adenovirus plasmids. In the another cosmid vector was developed, from which the first method,64 construction of the recombinant plasmid recombinant adenovirus genome can be released by

Gene Therapy Production of first generation adenovirus vectors X Danthinne and MJ Imperiale 1712 restriction digestion before transfection into helper cells.69 source of contamination, methods that use exclusively This cosmid vector contains unique cloning sites in the plasmid DNA for the transfections will maximize the E1 and E3 regions, allowing the insertion of two trans- chance of obtaining the right clone and minimize labor genes into the adenovirus genome. since fewer plaques will have to be analyzed. Methods In another approach, an adenovirus cosmid is con- that reconstitute the sequence of the recombinant aden- structed using three vectors containing the left end, the ovirus before transfection into helper cells also avoid the middle region and the right end of the adenovirus gen- homologous recombination step in helper cells, which is ome.70 One or two expression cassettes are inserted into clearly another rate-limiting step for virus generation. All the E1 or the E3 region, using the left or the right end these techniques but two53,60 require one or two vector, respectively. By choosing a set of three or four additional cloning steps in E. coli, but in compensation, restriction enzymes that are present naturally in the allow one to transfect helper cells with larger amounts of adenoviral genome and allow for directional cloning, DNA and to recover viral plaques more rapidly. A poten- large DNA fragments are generated, and ligated to recon- tial problem of these methods is the instability of the struct a cosmid containing the sequence of the recombi- large adenovirus plasmids in E. coli. This is particularly nant virus. After purification, the cosmid DNA is the case for plasmids that contain large palindromes com- digested to release the viral genome and transfected into prising the ITRs.20,58 Reconstituting the sequence of the a packaging cell line. Virus is recovered on average 7 virus by recombination in E. coli is slightly less straight- days later. This method has the advantage of allowing forward than using conventional ligation approaches, the introduction of two independent expression cassettes because at least one more step is needed to amplify the at different places in the genome in two cloning steps. adenovirus plasmid. These recombination-based However, the choice of restriction enzymes suitable for methods also present the danger of unpredicted recombi- directional cloning is limited given the size of the gen- nation events that might go undetected and generate ome, and the purification of large DNA fragments on noninfectious adenovirus plasmids. It is generally a good agarose gels is difficult. To circumvent these problems, idea, therefore, to use more than one independent clone two improved methods have been designed that facilitate for transfection. The advantage of some of these tech- the construction of a cosmid containing one or two niques,61,62 however, is that they are helpful for introduc- expression cassettes in the E1 and/or the E3 region ing mutations at any place in the adenovirus genome. (Figure 3b).71,72 The expression units are cloned into small Finally, construction of adenovirus plasmids by ligation bacterial vectors, and thereby linked to positive selection in E. coli is greatly facilitated by using cosmid technology bacterial markers. Using versatile sets of restriction since almost 100% of the bacterial clones are positive.70–72 enzymes that flank these sequences and allow for direc- Replication-competent adenoviruses are generated by tional cloning, the expression units and the positive selec- recombination between the recombinant viral DNA and tion markers are transferred to large plasmids containing the adenovirus sequences integrated into the genome of the adenovirus sequences, with an efficiency approaching the helper cells. The occurrence of RCA is very low in 100%. The cosmid DNA is purified, linearized and trans- the first passages of the virus74 and can be avoided if cell fected into helper cells. As for the other plasmid-based lines such as PER.C6,14 which do not contain sequences methods, viral plaques appear on average 7 days, and overlapping the viral chromosome, and appropriate vec- sometimes as early as 4 days, after transfection. tors are used. Contamination by variants derived from Finally, a method to reconstruct the sequence of the the recombinant virus is a problem that can affect all recombinant adenovirus DNA in yeast before transfec- methods. Since the viral DNA polymerase does not have tion into helper cells has also been reported.73 These the same degree of proofreading activity as its cellular investigators used a yeast artificial chromosome which counterpart,75 the mutation rate of viral DNA is much contains the full-length sequence of adenovirus DNA, higher than that of cellular DNA. A mutant virus could and inserted the gene of interest using a double-step gene have a growth advantage over the desired recombinant replacement procedure. Unfortunately, this technique has virus if, for instance, the latter expresses a gene that inter- proven difficult to reproduce. feres with the viral replication cycle, and mutations in viral genes on the recombinant chromosome might inter- Which method to use? fere with growth of the recombinant itself. A priori it is difficult to predict whether an expression cassette, or a Time and labor are the major factors that come into con- specific DNA sequence inserted into the genome, will be sideration when choosing a method to construct adeno- detrimental to virus growth. One should therefore be virus vectors, and ideally they should be minimized. The cautious in drawing conclusions from experiments using individual investigator’s comfort with the various experi- crude adenovirus extracts obtained directly from trans- mental procedures is also of obvious importance. All the fected cells, even if those were transfected with a plasmid techniques described above include one or several clon- isolated from a single E. coli clone. Ideally, two rounds ing steps in E. coli, transfection into a packaging cell line, of plaque assays should be performed and all stocks and the purification of one or several candidate recombi- should be checked for the presence of RCA, particularly nant viruses by plaque assay. The most time-consuming when in vivo studies are being performed. step is certainly the last one. Is this step necessary for all methods, and if yes, how many plaque assays are The future of first generation adenovirus needed? Contamination of a recombinant viral preparation can vectors originate from three sources: the parental virus, repli- Several straightforward methods are now available to cation competent adenoviruses (RCA), and mutant construct first-generation recombinant adenoviruses. viruses. Since the parental virus is generally the major Each investigator should choose a method, taking into

Gene Therapy Production of first generation adenovirus vectors X Danthinne and MJ Imperiale 1713 account the advantages and disadvantages of each, along 18 Bruder J, Jie T, McVey D, Kovesdi I. Expression of gp19K with his or her own experimental strengths. However, it increases the persistence of transgene expression from an aden- should be kept in mind that a recombinant virus might ovirus vector in the mouse lung and liver. J Virol 1997; 71: prove unstable and therefore plaque assays should 7623–7628. always be performed, regardless of the method used. 19 Schowalter D et al. Heterologous expression of adenovirus E3- gp19K in an E1a-deleted adenovirus vector inhibits MHC I Although first generation adenovirus vectors might be expression in vitro, but does not prolong transgene expression supplanted in the future by gutted vectors, especially in in vivo. Gene Therapy 1997; 4: 351–360. replacement gene therapy settings that require long-term 20 Bett AJ, Haddara W, Prevec L, Graham FL. An efficient and gene expression, they will remain an invaluable alterna- flexible system for construction of adenovirus vectors with tive for applications in which the immune response can insertions or deletions in early regions 1 and 3. Proc Natl Acad be tolerated or might even be desirable, such as cancer Sci USA 1994; 91: 8802–8806. or vaccination, for recombinant protein expression, or to 21 Gilgenkrantz H et al. Transient expression of genes transferred transfer genes into cell lines that are refractory to other in vivo into heart using first-generation adenoviral vectors: role transfection methods. of the immune response. Hum Gene Ther 1995; 6: 1265–1274. 22 Yang Y, Su Q, Wilson J. Role of viral antigens in destructive cellular immune responses to adenovirus vector-transduced Acknowledgements cells in mouse lungs. J Virol 1996; 70: 7209–7212. MJI acknowledges the support of NIH grants GM34902 23 Zhou H, O’Neal W, Morral N, Beaudet A. Development of a and HL64762. XD is the recipient of a fellowship from complementing cell line and a system for construction of aden- ovirus vectors with E1 and E2a deleted. J Virol 1996; 70: 7030– the Mountain States Medical Research Institute. 7038. 24 Gorziglia M et al. Elimination of both E1 and E2 from aden- References ovirus vectors further improves prospects for in vivo human gene therapy. J Virol 1996; 70: 4173–4178. 1 Berkner K. Development of adenovirus vectors for the 25 Lusky M et al. In vitro and in vivo biology of recombinant aden- expression of heterologous genes. Biotechniques 1988; 6: 616–629. ovirus vectors with E1, E1/E2A, or E1/E4 deleted. J Virol 1998; 2 Benihoud K, Yeh P, Perricaudet M. Adenovirus vectors for gene 72: 2022–2032. delivery. Curr Opin Biotechnol 1999; 10: 440–447. 26 Amalfitano A, Begy C, Chamberlain J. Improved adenovirus 3 Grunhaus A, Horwitz M. Adenoviruses as cloning vectors. packaging cell lines to support the growth of replication-defec- Semin Virol 1992; 3: 237–252. tive gene-delivery vectors. Proc Natl Acad Sci USA 1996; 93: 4 Haddada H, Cordier L, Perricaudet M. Gene therapy using 3352–3356. adenoviral vectors. In: Doerfler W, Bohm P (eds). The Molecular 27 Amalfitano A, Chamberlain J. Isolation and characterization of Repertoire of Adenoviruses. Springer-Verlag: Berlin, 1995, pp packaging cell lines that coexpress the adenovirus E1, DNA 297–306. polymerase, and preterminal proteins: implications for gene 5 Horwitz M. and their replication. In: Fields BN therapy. Gene Therapy 1997; 4: 258–263. et al (eds). . Raven Press: New York, 1990, pp 1679–1721. 28 Langer S, Schaack J. 293 cell lines that inducibly express high 6 Kovesdi I, Brough D, Bruder J, Wickham T. Adenoviral vectors levels of adenovirus type 5 precursor terminal protein. Virology for gene transfer. Curr Opin Biotechnol 1997; 8: 583–589. 1996; 221: 172–179. 7 Trapnell B, Gorziglia M. Gene therapy using adenoviral vectors. 29 Gorziglia M et al. Generation of an adenovirus vector lacking Curr Opin Biotechnol 1994; 5: 617–625. E1, E2a, E3, and all of E4 except open reading frame 3. J Virol 8 Bett A, Prevec L, Graham F. Packaging capacity and stability of 73 human adenovirus type 5 vectors. J Virol 1993; 67: 5911–5921. 1999; : 6048–6055. 9 Harui A, Suzuki S, Kochanek S, Mitani K. Frequency and stab- 30 He T-C et al. A simplified system for generating recombinant ility of chromosomal integration of adenovirus vectors. J Virol adenoviruses. Proc Natl Acad Sci USA 1998; 95: 2509–2514. 1999; 73: 6141–6146. 31 Brough D et al. A gene transfer vector-cell line system for com- 10 Recchia A et al. Site-specific integration mediated by a hybrid plete functional complementation of adenovirus early regions adenovirus/adeno-associated virus vector. Proc Natl Acad Sci E1 and E4. J Virol 1996; 70: 6497–6501. USA 1999; 96: 2615–2620. 32 Gao G, Yang Y, Wilson J. Biology of adenovirus vectors with 11 Lieber A, Steinwaerder D, Carlson C, Kay M. Integrating aden- E1 and E4 deletions for liver-directed gene therapy. J Virol 1996; ovirus-adeno-associated virus hybrid vectors devoid of all viral 70: 8934–8943. genes. J Virol 1999; 73: 9314–9324. 33 Yeh P et al. Efficient dual transcomplementation of adenovirus 12 Graham FL, Smiley J, Russel WC, Nairn R. Characteristics of a E1 and E4 regions from a 293-derived cell line expressing a mini- human cell line transformed by DNA from human adenovirus mal E4 functional unit. J Virol 1996; 70: 559–565. type 5. J Gen Virol 1977; 36: 59–74. 34 Krougliak V, Graham F. Development of cell lines capable of 13 Fallaux F et al. Characterization of 911, a new helper cell line complementing E1, E4, and protein IX defective adenovirus type for the titration and propagation of early region 1-deleted aden- 5 mutants. Hum Gene Ther 1995; 6: 1575–1586. oviral vectors. Hum Gene Ther 1996; 7: 215–222. 35 Wang Q, Jia X, Finer M. A packaging cell line for propagation 14 Fallaux F et al. New helper cells and matched early region 1- of recombinant adenovirus vectors containing two lethal gene- deleted adenovirus vectors prevent generation of replication- region deletions. Gene Therapy 1995; 2: 775–783. competent adenoviruses. Hum Gene Ther 1998; 9: 1909–1917. 36 Kochanek S et al. A new adenoviral vector: replacement of all 15 Wold W, Tollefson A, Hermiston T. E3 transcription unit of viral coding sequences with 28 kb of DNA independently ␤ adenovirus. In: Doerfler W, Bohm P (eds). The Molecular Reper- expressing both full-length dystrophin and -galactosidase. Proc toire of Adenoviruses. Springer-Verlag: Berlin, 1995, pp 237–274. Natl Acad Sci USA 1996; 93: 5731–5736. 16 Tollefson A et al. The adenovirus death protein (E3–11.6K) is 37 Parks R et al. A helper-dependent adenovirus vector system: required at very late stages of infection for efficient cell lysis removal of by Cre-mediated excision of the viral and release of adenovirus from infected cells. J Virol 1996; 70: packaging signal. Proc Natl Acad Sci USA 1996; 93: 13565–13570. 2296–2306. 38 Fisher K et al. Recombinant adenovirus deleted of all viral genes 17 Lee M, Abina M, Haddada H, Perricaudet M. The constitutive for gene therapy of cystic fibrosis. Virology 1996; 217: 11–22. expression of the immunomodulatory gp19k protein in E1−,E3− 39 Kumar-Singh R, Chamberlain J. Encapsidated adenovirus min- adenoviral vectors strongly reduces the host cytotoxic T cell ichromosomes allow delivery and expression of a 14 kb dystro- response against the vector. Gene Therapy 1995; 2: 256–262. phin cDNA to muscle cells. Hum Mol Genet 1996; 5: 913–921.

Gene Therapy Production of first generation adenovirus vectors X Danthinne and MJ Imperiale 1714 40 Steinwaerder D, Carlson C, Lieber A. Generation of adenovirus in a cosmid and the Cre-loxP recombination system. Hum Gene vectors devoid of all viral genes by recombination between Ther 1999; 10: 1845–1852. inverted repeats. J Virol 1999; 73: 9303–9313. 59 Stow N. Cloning of a DNA fragment from the left-hand ter- 41 Hartigan-O’Connor D, Amalfitano A, Chamberlain J. Improved minus of the adenovirus type 2 genome and its use in site- production of gutted adenovirus in cells expressing adenovirus directed mutagenesis. J Virol 1981; 37: 171–180. preterminal protein and DNA polymerase. J Virol 1999; 73: 60 Aoki K et al. Efficient generation of recombinant adenoviral vec- 7835–7841. tors by cre-lox recombination in vitro. Mol Med 1999; 5: 224–231. 42 Chinnadurai G, Chinnadurai S, Brusca J. Physical mapping of 61 Chartier C et al. Efficient generation of recombinant adenovirus a large-plaque mutation of adenovirus type 2. J Virol 1979; 32: vectors by homologous recombination in Escherichia coli. J Virol 623–628. 1996; 70: 4805–4810. 43 Schaack J, Langer S, Guo X. Efficient selection of recombinant 62 Crouzet J et al. Recombinational construction in Escherichia coli adenoviruses by vectors that express ␤-galactosidase. J Virol of infectious adenoviral genomes. Proc Natl Acad Sci USA 1997; 1995; 69: 3920–3923. 94: 1414–1419. 44 Munz P, Young C. End-joining of DNA fragments in adenovirus 63 Gustin K, Imperiale M. Encapsidation of viral DNA requires the transfection of human cells. Virology 1991; 183: 160–169. adenovirus L1 52/55-kilodalton protein. J Virol 1998; 72: 7860– 45 Imler J-L et al. An efficient procedure to select and recover 7870. recombinant adenovirus vectors. Gene Therapy 1995; 2: 263–268. 64 Souza D, Armentano D. Novel cloning method for recombinant 46 Davis AR, Meyers K, Wilson JM. High throughput method for adenovirus construction in Escherichia coli. Biotechniques 1999; 26: creating and screening recombinant adenoviruses. Gene Therapy 502–508. 1998; 5: 1148–1152. 65 Mizuguchi H, Kay MA. Efficient construction of a recombinant 47 Jones N, Shenk T. Isolation of deletion and substitution mutants adenovirus vector by an improved in vitro ligation method. Hum of adenovirus type 5. Cell 1978; 13: 181–188. Gene Ther 1998; 9: 2577–2583. 48 Berk A et al. Pre-early adenovirus 5 gene product regulates syn- 66 Mizuguchi H, Kay M. A simple method for constructing E1- and thesis of early viral messenger RNAs. Cell 1979; 17: 935–944. E1/E4-deleted recombinant adenoviral vectors. Hum Gene Ther 49 Galos R, Williams J, Shenk T, Jones N. Physical location of host- 1999; 10: 2013–2017. range mutations of adenovirus type 5; deletion and marker- 67 Feiss M, Siegle D. Packaging of the bacteriophage lambda chro- rescue mapping. Virology 1980; 104: 510–513. mosome: dependence of cos cleavage on chromosome length. 50 Volkert F, Young C. The genetic analysis of recombination using Virology 1979; 92: 190–200. adenovirus overlapping terminal DNA fragments. Virology 1983; 68 Fu S, Deisseroth A. Use of cosmid adenoviral vector cloning 125: 175–193. system for the in vitro construction of recombinant adenoviral 51 Miyake S et al. Efficient generation of recombinant adenoviruses vectors. Hum Gene Ther 1997; 8: 1321–1330. using adenovirus DNA–terminal protein complex and a cosmid 69 Kojima H, Ohishi N, Yagi K. Generation of recombinant aden- bearing the full-length virus genome. Proc Natl Acad Sci USA ovirus vector with infectious adenoviral genome released from 1996; 93: 1320–1324. cosmid-based vector by simple procedure allowing complex 52 Hardy S et al. Construction of adenovirus vectors through cre- manipulation. Biochem Biophys Res Commun 1998; 246: 868–872. lox recombination. J Virol 1997; 71: 1842–1849. 70 Danthinne X. New vectors for the construction of double recom- 53 Anderson R et al. A simple method for the rapid generation of binant adenoviruses. J Virol Meth 1999; 81: 11–20. recombinant adenovirus vectors. Gene Therapy 2000; 7: 1034– 71 Danthinne X, Werth E. New tools for the generation of E1- 1038. and/or E3-substituted adenoviral vectors. Gene Therapy 1999; 7: 54 McGrory WJ, Bautista DS, Graham FL. A simple technique for 80–87. the rescue of early region I mutations into infectious human 72 Danthinne X. Simultaneous insertion of two expression cassettes adenovirus type 5. Virology 1988; 163: 614–617. into adenovirus vectors (submitted for publication). 55 Leach D, Stahl F. Viability of l phages carrying a perfect palin- 73 Ketner G et al. Efficient manipulation of the human adenovirus drome in the absence of recombination nucleases. Nature 1983; genome as an infectious yeast artificial chromosome clone. Proc 305: 448–451. Natl Acad Sci USA 1994; 91: 6186–6190. 56 Hay R, Stow N, McDougall I. Replication of adenovirus mini- 74 Lochmuller H et al. Emergence of early region 1-containing rep- chromosomes. J Mol Biol 1984; 175: 493–510. lication-competent adenovirus in stocks of replication-defective 57 Ng P et al. A high-efficiency Cre/loxP-based system for con- adenovirus recombinants (delta E1 + delta E3) during multiple struction of adenoviral vectors. Hum Gene Ther 1999; 10: 2667– passages in 293 cells. Hum Gene Ther 1994; 5: 1485–1491. 2672. 75 King A et al. Processive proofreading by the adenovirus DNA 58 Tashiro F, Niwa H, Miyazaki J-I. Constructing adenoviral vec- polymerase. Association with the priming protein reduces exon- tors by using the circular form of the adenoviral genome cloned ucleolytic degradation. Nucleic Acids Res 1997; 25: 1745–1752.

Gene Therapy