In Vitro Packaging of an Infectious Recombinant Adeno-Associated Virus 2

Home , In vitro, Recombinant virus

L Ding, S Lu and NC Munshi Division of Hematology/Oncology, Department of Medicine, University of Arkansas for Medical Sciences and John L McClellan VA Medical Center, Little Rock, AR, USA

Adeno-associated virus 2 (AAV), a human parvovirus, has not documented until now. In this study, we produced in properties such as stable chromosomal integration, high vitro infectious recombinant AAV virions containing the infectivity and lack of known human pathogenicity, making neomycin resistance (NeoR) and the human CD-16 it a potentially useful vector for human gene therapy. AAV (FC␥RIIIa) gene. Our in vitro packaging of AAV has the requires a helper virus, such as an adenovirus, for optimal same characteristics as AAV produced in vivo and demon- replication and packaging in mammalian cells. Although strates the feasibility of developing a safe packaging replication of the wild-type AAV genome has been demon- method for AAV to be used in gene therapy. strated in vitro, packaging of infectious viral particles was

Keywords: adeno-associated virus; in vitro packaging; gene therapy

Introduction a crude extract of HeLa cells previously infected with adenovirus-2 (Ad-2) and transfected with AAV Adeno-associated virus 2 (AAV), a nonpathogenic sequences. human parvovirus, causes latent infection of mammalian cells by integration into the host genome.1 However, co- infection with adenovirus or herpes virus (helper viruses) Results is necessary to cause AAV to replicate, excise and pro- With the background of successful replication of wild- 2 duce infectious viral particles. The 145-bp inverted ter- type AAV in vitro,15,16 we attempted to study the in vitro minal repeats (ITRs) are the sole cis-acting elements replication and eventual packaging of recombinant AAV responsible for rescue, replication and packaging of the sequences. We studied in vitro replication of DNA 3–5 virus. Transcripts from the p40 promoter produce viral sequences between AAV ITRs in the presence of required capsid protein while the p5 and p19 promoters code for essential elements that were provided through crude cell 6 rep genes important in viral replication. The AAV rep78 extracts. The same procedure was used to determine protein mediates AAV replication by its ATP-dependent whether or not the replicated sequences are packaged 7 site-specific endonuclease and DNA helicase activity. into AAV virions to determine its potential as a vector Peak production of both rep and capsid proteins in for gene transfer. human cells occurs between 12 and 24 h after infection.8 The recombinant virus is produced by cotransfection of In vitro AAV DNA replication a human cell line with (1) a recombinant plasmid contain- To determine if recombinant AAV could be packaged in ing the gene of interest flanked by the 145-bp ITRs; (2) a vitro, we transfected adenovirus-2-infected HeLa cells plasmid containing all of the AAV sequences with modi- with a plasmid containing AAV sequences (p␭For 9,10 fications to prevent its packaging; and (3) adenovirus. pAAV/AD) encoding the viral structural and nonstruc- 11,12 In vitro packaging of lambda DNA and in vitro rep- tural genes. Both p␭F and pAAV/AD have sequence 13,14 lication and supercoiling of simian virus 40 DNA modifications to prevent them from being packaged. A have been well characterized. Recently, the in vitro repli- crude cell extract was used to study replication and pack- cation of wild-type adeno-associated virus DNA, as aging as described in the Materials and methods. assayed by production of full-length excised AAV DNA Replication of the recombinant plasmid pLM10 con- 15,16 resistant to DpnI digestion was reported. However, in taining the LacZ gene between AAV ITRs was studied by vitro replication of recombinant virions and packaging of incubating the recombinant plasmid with the crude cell wild-type or recombinant AAV has not been demon- lysate, nucleotide and deoxynucleotide triphosphates and strated previously. In the present study, we demonstrate other ions for 24 h. Subsequently, low molecular weight in vitro production of functional recombinant AAV from DNA was extracted by the described method and DNA was analyzed by Southern blot analysis. To confirm DNA replication, a DNA sample was digested with DpnI, Correspondence: NC Munshi, University of Arkansas for Medical which digests only methylated DNA; newly replicated Sciences, 4301 W Markham, Slot 508, Little Rock, Arkansas 72205, USA DNA will be unmethylated in at least one strand and not Received 8 May 1997; accepted 20 June 1997 digested. As seen in Figure 1, a DpnI-resistant species is In vitro packaging of recombinant AAV L Ding et al 1168 CD16-specific primers. Various negative controls were utilized to confirm the specificity of packaging. The cell extract used for in vitro packaging must contain all AAV rep and cap proteins and for efficient production of each protein, cells must be transduced with plasmid containing AAV sequences (p␭F or pAAV/Ad), as well as with adenovirus. In the control samples where the cell lysate was prepared without p␭F or pAAV/Ad transfection or without adenovirus infection, no PCR bands are seen (Figure 2, lanes 3 and 4). Similarly, inactivation of the cell lysate before in vitro packaging results in no PCR amplification (Figure 2, lane 5). Production of recombinant AAV also requires a gene of interest between the AAV ITRs, so absence of AAV ITRs or linearization of the plasmid between AAV ITRs will not allow production of the recombinant AAV. As seen in Figure 2, lanes 6 and 8, no PCR bands are observed; this suggests a failure to package. However, when all components of the system are present and intact, a 462-bp PCR fragment is amplified (Figure 2, lane 7), indicating successful packaging of recombinant virus. All of these requirements are similar to those needed for in vivo packaging of AAV.

Infectivity of in vitro packaged AAV Biological activity of the in vitro packaged virus was investigated by examining its infectivity of human cells. When HeLa cells are infected with different in vitro packaged extracts, the expression of NeoR gene can be studied by looking at development of resistance to the neomycin analogue G-418. In our study, only cells infected with extracts containing all the components, developed resistance to G418 at an active concentration of 800 ␮g/ml. All

Figure 1 Replication of recombinant AAV plasmid in a cell-free system was tested. Ad-2-infected cells were transfected with pAAV/Ad and lysates were incubated with pLM10. After a 24-h incubation at 37°C low molecular weight DNA was isolated by Hirt extraction and half of the sample was further digested with DpnI for 2 h at 37°C. The products were separated on 1% agarose gel and analyzed by Southern blot with 32P-labeled LacZ sequences. Lanes 1 and 2 represent positive control where replication was studied in vivo in intact cells, while lanes 3–6 have studies performed in vitro. Lanes 2 and 4 show the presence of bands suggestive of replicative form while the absence of the band in lane 6 suggests no replication. Lane 1, Ad-2 + pAAV/Ad; lane 2, Ad-2 + pAAV/Ad+DpnI; lane 3, Ad-2 + pAAV/Ad; lane 4, Ad-2 + pAAV/Ad+ Dpn I; lane 5, Ad- 2; lane 6, Ad-2 + Dpn I.

detected in the in vivo (Figure 1, lane 2) and in vitro (Figure 1, lane 4) replicated samples. Similar bands are not seen in Figure 1, lane 6, where the cells were not transfected with pAAV/Ad, which provides the rep gene needed for replication.

In vitro packaging of the recombinant AAV Packaging of the recombinant AAV in vitro was then examined. Plasmid pMM-11 was incubated in the cell Figure 2 Detection of packaging of a recombinant AAV by PCR amplifi- extracts under conditions described in the Materials and cation using a human CD-16-specific primer-pair. Lane 1, positive control, plasmid pMM11; lane 2, negative control; lane 3, cell extracts containing methods. All non-packaged, and hence non-protected, + −␭ ␭ + DNA was destroyed by treating the mixture with a pro- Ad2 pMM11 ( F); lane 4, F pMM11 (-Ad-2); lane 5, heat-inactivated cell extract with AD-2 + ␭F + pMM11; lane 6, AD-2 + ␭F + pMM- longed exposure (2 h) to a high concentration (1000 4 (no termini); lane 7, Ad-2 + ␭F + pMM11 (all components); lane 8, ␮g/ml) of DNase I. Subsequently packaged and pro- Ad-2 + ␭F + pMM11 digested with BglII. The expected amplified fragment tected DNA was examined by PCR amplification using using CD-16 primer-pair is 462 bp. In vitro packaging of recombinant AAV L Ding et al 1169 other cells infected with extracts lacking even one of the necessary components of the viral packaging system failed to show a G-418-resistant colony. The viral titer as measured by the number of neomycin-resistant colonies was Ͻ103/ml for in vitro packaged virus and 104–105/ml for in vivo packaged viruses. The presence of NeoR in the infected cell colonies was demonstrated by PCR analysis of the clones of the resistant cells after 4 weeks’ growth (Figure 3). A recombinant AAV containing LacZ gene was also packaged in vitro. HeLa cells were infected with in vitro packaged virus and control cells were infected with extracts with the LacZ gene inserted at a site other than between the AAV ITRs, making it unlikely to be packaged. As seen in Figure 4, the presence of blue cells con- firms transduction of HeLa cells while control extracts fail to transduce cells. Prior transfection of the cells with wild-type AAV as a source of rep and cap products, or transfection of plasmid- containing AAV sequences were equally effective in gen- erating virus. In either case, a 24-h incubation period was required for virus generation (data not shown). Attempts to package AAV by first preparing the cell extract and then adding Ad-2, p␭F or pAAV/AD, and pMM-11 sim-

Figure 4 The biological activity of in vitro packaged AAV was tested by infecting the HeLa cells with recombinant virus, packaged in vitro using a plasmid containing LacZ gene between AAV ITRs, while control cells were infected with in vitro packaged AAV where the LacZ plasmid was without AAV termini. The cells were stained for LacZ expression 48 h later. Cells infected with in vitro packaged material containing LacZ gene between AAV ITRs stain blue suggesting transduction and biological activity, while LacZ plasmid without AAV ITR does not show staining, suggesting inability of the plasmid to be packaged and transduced.

ultaneously did not generate the virus; this suggests that the intact cell is needed for the initial expression of the viral structural and nonstructural genes.

Discussion The results presented here describe successful in vitro production of a recombinant DNA virus using a lysed cell extract. In vitro packaging of lambda DNA has been described previously, in which an extract from a phage- infected E. coli supplies the mixture of proteins and pre- cursors required for encapsidation of lambda DNA.11,12 Similarly, in vitro replication and supercoiling of simian virus 40 DNA have been well characterized.13,14 Soluble cell extracts from the nucleus and cytoplasm of human 293 cells have been shown to efficiently replicate a DNA template containing the origin of simian virus 40 replication.17 The replication of wild-type AAV sequences have been described previously.16 Using a plasmid-containing duplex form of AAV DNA in pBR322, replication of AAV sequences was demonstrated by production of full-length excised AAV DNA resistant to DpnI digestion Figure 3 PCR analysis of neomycin-resistant clones of KB cells using in an extract of cells co-infected with AAV and aden- NeoR gene-specific primer pair. Lane 1, neomycin-resistant clones of KB cells infected with in vivo packaged virus; lane 2, neomycin-resistant ovirus. This report also indicated additional requirements clones of KB cells infected with in vitro packaged virus; lane 3, negative for in vitro replication: AAV sequences, AAV termini and control; lane 4, positive control, plasmid pMM-11. excision of the AAV insert.15,16 Similarly, complex eukar- In vitro packaging of recombinant AAV L Ding et al 1170 yotic DNA replication in a cell-free system has also been vitro produced rep and cap genes may obviate the need described.18 However, ours is the first report of in vitro for plasmid transfection of the cells and adenoviral infec- replication and packaging of the recombinant virions. tion. This also raises the possibility that we may be able The in vitro replication and packaging of recombinant to produce recombinant AAV without the need for aden- AAV described here conforms to all the characteristics of oviral infection, eliminating some safety and regulatory in vivo packaging such as presence of functional AAV rep concerns. and cap genes and packaging of genes only when they The data presented here demonstrate the feasibility of have been inserted between ITRs.10 AAV packaging in a cell-free system. This system should The in vitro packaged virus is detected by a PCR ampli- allow further studies to evaluate the role of different fication of a fragment of the gene, as the packaged DNA cellular components in AAV packaging and the role of is protected from DNase I action. One unit of DNase I AAV-encoded proteins in the packaging of the virus. completely digests 1 ␮g of plasmid DNA in 10 min at Because AAV is becoming an attractive vector for gene 37°C. We used DNase I at a very high concentration and transfer,19–21 the availability of an in vitro packaging sys- for a prolonged period of time to digest the in vitro pack- tem promises to offer a safer packaging method, which aged extracts. Initially, they are digested with DNase I at in turn will help develop AAV as a vector for human 100 ␮g/ml for 1 h and then 1000 ␮g/ml for one gene therapy. additional hour to guarantee complete digestion of the unpackaged plasmid. A variety of negative controls was used for two Materials and methods reasons. First, the negative results so obtained confirm the efficacy of the DNase I in destroying all unpackaged Cells and viruses plasmid, thus proving that the positive PCR reaction A nasopharyngeal carcinoma cell line, KB and HeLa cells shows amplification of true DNA protected from DNase were cultured in RPMI supplemented with 10% fetal calf ° I digestion. Second, these controls also demonstrate the serum and 1% penicillin-streptomycin at 37 Cin5%CO2. various requirements for packaging of the AAV. Clearly Human adenovirus type 2 and wild-type AAV were the cell lysate prepared without adenovirus infection or kindly given by Dr A Srivastava, Indiana University without AAV sequences that produce rep and cap pro- School of Medicine, Indianapolis, IN, USA. teins is unable to package AAV. For in vivo packaging to occur, the gene of interest must be between the ITRs. In Plasmids our negative controls, the gene to be packaged was not Plasmid pAAV/Ad (a gift from Dr J Samulski, University between the ITRs or BglII endonuclease was used to of North Carolina, Chapel Hill, NC, USA) and p␭F (a gift digest pMM11 giving a linear DNA with the gene cleaved from Dr P Hermonat, University of Arkansas for Medical between the AAV ITRs. We show that both these Sciences, Little Rock, AR, USA) have been previously manipulations lead to failure of packaging in our system, described.9,10 Plasmid pMM11 contains a SV-40 as would be expected in in vivo packaging. An additional promoter-driven human CD16 gene and thymidine kin- confirmation of the requirement for functional AAV pro- ase (TK) promoter driven neomycin resistance gene teins in the lysate was shown by the inability to package between AAV ITRs. pMM4 contains the same construct when previously heat-inactivated lysate was used. without the AAV ITRs. The plasmid pLM10 contains the Although PCR is a very sensitive technique, able to detect LacZ gene driven by the phosphoglycerokinase (PGK) even a single copy of DNA, all of the negative controls promoter between AAV ITRs and plasmid pLM2 was the with similar unpackaged plasmid contents provide same construct without AAV ITRs. further evidence that sequence amplification reliably interprets the packaging. Preparation of cell extracts The titer obtained through this method has been Cell extracts for in vitro packaging were produced from extremely low suggesting that the in vitro packaging HeLa cells infected with Ad-2 and co-transfected with described here may not be as efficient as in vivo packag- AAV/Ad or ␭F using the calcium phosphate transfection ing. Various possibilities can explain this finding. This method as described.22 Twenty-four hours later, cells may be due to inefficient in vitro replication of the plas- were harvested by scraping. Following two washes with mid, leading to decreased availability of replicative DNA PBS (pH 7.4), the cells are resuspended in PBS at a con- for packaging. Inadequate amounts and inadequate func- centration of 2 × 107 cells/ml and subjected to three tioning of AAV rep and cap proteins may contribute to freeze–thaw cycles (−80°C and 37°C). The cell lysates the inefficiency of the system. Additionally, the reaction were cryopreserved at −80°C in smaller aliquots. The con- volume in the in vitro system compared to the cell volume trol lysate was prepared in the absence of specific compo- in vivo points to a dilution effect that may decrease the nents: without Ad-2 infection, without AAV/AD or ␭F probability of packaging. Correction of the mechanisms transfection. possibly responsible for inefficient packaging will be important when broadly applying this method to pro- In vitro replication and packaging duce recombinant AAV. The in vitro recombinant AAV replication was carried out In vitro packaging provides a simple method for pro- in a 200-␮l reaction mixture containing 30 mm HEPES; 7 m m ␮ duction of recombinant AAV without the need for m MgCl2;4m ATP; 4 g BSA (bovine serum expertise or extensive understanding of the biology of albumin); 200 ␮m each of CTP, GTP and UTP; 100 ␮m AAV. It also decreases the packaging time from between each of dATP, dGTP, dTTP and dCTP; 400 mm creatine 48 and 72 h to only 24 h. With the suggested modifi- phosphate; 1 ␮g creatine phosphokinase; and 150 ␮l cell cations, better titers may be obtained for clinical appli- lysate. Then 10 ␮g recombinant AAV plasmid pLM10 cation. It remains to be seen whether the addition of in was added. The reaction mixture was incubated for 24 h In vitro packaging of recombinant AAV L Ding et al 1171 at 37°C and analyzed for DNA replication and virus cycles of incubation at 94°C (30 s), 61°C (30 s), and 72°C packaging. (45 s), followed by a 10-min extension period at 72°C. The amplified fragments were separated by electrophoresis Detection of DNA replication on 2% agarose gel and stained by ethidium bromide. Low molecular weight DNA was extracted from the reac- The second plasmid packaged in vitro was pLM10, con- tion mixture by Hirt extraction.23 Briefly, a 100-␮l reaction taining the LacZ gene between AAV ITR or LM2 contain- mixture was mixed with 400 ␮l Hirt solution at room ing the LacZ gene without AAV ITRs. The packaged pro- temperature for 10 min; NaCl was added to a final con- ducts were digested by DNase I as described and HeLa centration of 1.0 m, and kept at 4°C overnight. The pro- cells were infected by the packaged products, and after ducts were then extracted with phenol:chloroform, pre- 48 h, stained by X-gal to show virus-infected cells. cipitated by ethanol, and dissolved in 20 ␮l water. Half of the sample was digested with DpnI for 2 h at 37°C. Acknowledgements The products were separated on 0.8% agarose gel by electrophoresis and transferred to a nylon membrane This work was supported in part by grants from the by the standard method.24 The nylon membrane was American Cancer Society (DHP 153), and the USPHS’s probed with a LacZ-specific probe and visualized by National Heart Blood and Lung Institute (HL-55695) and autoradiography. National Cancer Institute (CA71092), and the Phi Beta Psi sorority. We thank the Office of Grants and Scientific Detection of packaged viruses Publications at the Arkansas Cancer Research Center The recombinant AAV plasmid used in this study, (Paula Card-Higginson and Mary Dornhoffer) for edi- pMM11, containing the SV40 promoter-driven CD16 torial assistance with the manuscript. gene and the TK promoter-driven neomycin-resistant gene, was packaged in vitro. The packaged products were References digested by DNase I at 100 U/ml for 1 h and then at 54 ° 1 Berns KI. Parvovirus replication. Microbiol Rev 1990; : 316–329. 1000 U/ml for 1 h at 37 C to destroy all the unpackaged 2 Berns KI, Bohenzky RA. Adeno-associated viruses: an update. plasmid and final product was analyzed by PCR by Adv Vir Res 1987; 32: 243–306. amplifying a segment of the CD16 gene. 3 Samulski RJ, Berns KI, Tan M, Muzyczka N. Cloning of adeno- associated virus into pBR322: rescue of intact virus from the Polymerase chain reaction (PCR) recombinant plasmid in human cells. Proc Natl Acad Sci USA The primers used spanned a region of 462 bp in CD16 1982; 79: 2077–2081. gene, from bases 224–244 (5′-AGAGCCTCATCT 4 Samulski RJ, Srivastava A, Berns KI, Muzyczka N. Rescue of CAAGCCAG) to bases 686–665 (5′-GGAGTACCATCA adeno-associated virus from recombinant plasmids: gene correction within the terminal repeats of AAV. Cell 1983; 33: 135–143. CCAAGCAG). Each 100-␮l reaction consisted of 5 ␮lof m m 5 Srivastava A. Replication of the adeno-associated virus DNA in vitro packaged product, 2.5 m MgCl2, 0.2 m dNTPs, termini in vitro. Intervirology 1987; 27: 138–147. 50 mm KCl, 10 mm Tris-HCl, pH 8.3, 1 unit AmpliTaq 6 Hermonat PL et al. Genetics of adeno-associated virus: isolation DNA polymerase (Perkin Elmer Cetus, Emeryville, CA, and preliminary characterization of adeno-associated virus type USA), and 50 pmol synthetic primers. After overlay with 2 mutants. J Virol 1984; 51: 329–339. a drop of mineral oil, DNA was amplified using 35 cycles 7 Im DS, Muzyczka N. The AAV origin binding protein Rep68 is of incubation at 94°C (60 s), 64°C (2 min), and 72°C(3 an ATP-dependent site-specific endonuclease with DNA hel- min) in an automated cycler (Perkin Elmer Cetus DNA icase activity. Cell 1990; 61: 447–457. thermal cycler). PCR products were electrophoresed on 8 Redemann BE, Mendelson E, Carter BJ. Adeno-associated virus rep protein synthesis during productive infection. J Virol 1989; 2% agarose gels and visualized by ethidium bromide 63: 873–882. staining or analyzed by Southern blots using a CD16-spe- 9 Hermonat PL, Muzyczka N. Use of adeno-associated virus as cific DNA probe. a mammalian DNA cloning vector: transduction of neomycin resistance into mammalian tissue culture cells. Proc Natl Acad Infectivity of in vitro packaged viruses Sci USA 1984; 81: 6466–6470. HeLa cells were infected by the in vitro packaged pMM11 10 Samulski RJ, Chang LS, Shenk T. Helper-free stocks of recombi- product and subjected to G418 selection at the concen- nant adeno-associated viruses: normal integration does not tration of 400 ␮g/ml. After 2 weeks of selection, G418- require viral gene expression. J Virol 1989; 63: 3822–3828. resistant clones were counted, with some clones picked 11 Rosenberg SM, Stahl MM, Kobayashi I, Stahl FW. Improved in vitro packaging of coliphage lambda DNA: a one-strain system and grown further for PCR analysis. The HeLa cells were free from endogenous phage. Gene 1985; 38: 165–175. trypsinized and 1 × 105 cells were pelleted, washed once ␮ 12 Rosenberg SM. Improved in vitro packaging of lambda DNA. with PBS, and resuspended in 5 l PBS. Cells were lysed Meth Enzymol 1987; 153: 95–103. by adding 20 ␮l water and heating the samples at 95°C 13 Wobbe CR, Dean F, Weissbach L, Hurwitz J. In vitro replication for 10 min. After digestion with proteinase K (100 ␮g/ml) of duplex circular DNA containing the simian virus 40 DNA at 55°C for 1 h, the cell lysate was heated at 95°C for 10 origin site. Proc Natl Acad Sci USA 1985; 82: 5710–5714. min to inactivate proteinase K and then stored at −20°C 14 Stillman BW, Gluzman Y. Replication and supercoiling of sim- before PCR. The presence of NeoR gene was detected by ian virus 40 DNA in cell extracts from human cells. Molec Cell PCR analysis. Each 100-␮l reaction consisted of 5 ␮l cell Biol 1985; 5: 2051–2060. lysate, 1.5 mm MgCl , 0.2 mm dNTPs, 50 mm KCl, 10 mm 15 Ni TH et al. In vitro replication of adeno-associated virus DNA. 2 J Virol 1994; 68: 1128–1138. Tris-HCl, pH 8.3, 1 unit AmpliTaq DNA polymerase 16 Hong G, Ward P, Berns KI. In vitro replication of adeno-associa- (Perkin Elmer Cetus) and 50 pmol synthetic primers. The ted virus DNA. Proc Natl Acad Sci USA 1992; 89: 4673–4677. R ′ Neo specific primer pair used was 5 GGT GGA GAG 17 Ward P, Gerns KI. In vitro rescue of an integrated hybrid adeno- GCT ATT CGG CTA TGA 3′ and 5′ ATC CTG ATC GAC associated virus/simian virus 40 genome. J Molec Biol 1991; 218: AAG ACC GGC TTC 3′. DNA was amplified using 35 791–804. In vitro packaging of recombinant AAV L Ding et al 1172 18 Munshi NC, Gabig TG. Growth factor-dependent initiation of 22 Srivastava CH et al. Construction of a recombinant human par- DNA replication in nuclei isolated from an interleukin 3-depen- vovirus B19: adneo-associated virus 2 (AAV) DNA inverted ter- dent murine myeloid cell line. J Clin Invest 1990; 85: 300–304. minal repeats are functional in an AAV-B19 hybrid virus. Proc 19 Flotte TR, Carter BJ. Adeno-associated virus vectors for gene Natl Acad Sci USA 1989; 86: 8078–8082. therapy. Gene Therapy 1995; 2: 357–362. 23 Hirt B. Selective extraction of polyoma DNA from infected 20 Einerhand MP et al. Regulated high-level human beta-globin mouse cell cultures. J Molec Biol 1967; 26: 365–369. gene expression in erythroid cells following recombinant adeno- 24 Southern EM. Detection of specific sequences among DNA frag- associated virus-mediated gene transfer. Gene Therapy 1995; 2: ments separated by gel electrophoresis. J Molec Biol 1975; 98: 336–343. 503–517. 21 Walsh CE et al. Phenotypic correction of Fanconi anemia in human hematopoietic cells with a recombinant adeno-associated virus vector (see comments). J Clin Invest 1994; 94: 1440–1448.