Integrase-Defective Lentiviral Vectors: Progress and Applications

MB Banasik1,2 and PB McCray Jr1,2 1Department of Pediatrics, Program in Gene Therapy, Carver College of Medicine, The University of Iowa, Iowa City, IA, USA and 2Interdisciplinary Graduate Program in Genetics, Carver College of Medicine, The University of Iowa, Iowa City, IA, USA

Lentiviral vectors (LVs) offer the advantages of a large in which transient gene expression is desired. Several recent packaging capacity, broad cell tropism or speciﬁc cell-type publications outline the development and initial biological targeting through pseudotyping, and long-term expression characterization of such vectors. Here, we discuss the poten- from integrated gene cassettes. However, transgene integ- tial applications and new directions for the development of ration carries a risk of disrupting gene expression through integration-defective LVs. insertional mutagenesis and may not be required for all Gene Therapy (2010) 17, 150–157; doi:10.1038/gt.2009.135; applications. A non-integrating LV may be beneﬁcial in cases published online 22 October 2009

Keywords: HIV; FIV; episome; zinc-ﬁnger nuclease; vaccine

Introduction necessary for integration.9,10,14 IN then facilitates strand transfer through a one-step mechanism that both induces Lentiviral vectors (LV) are an important class of gene a staggered break in the phosphodiester bonds of the transfer vectors. LV have a large packaging capacity target DNA and joins a single vector DNA end to the (410 kb transgene cassette1,2) and envelope glycoprotein target DNA through a transesterification reaction.11,14,15 pseudotyping achieves broad or specific tropism as Once both vector DNA ends are attached, host enzymes necessary. These vectors are also associated with low repair the resulting gapped intermediate, generating an immunogenicity. Importantly, they transduce mitotically integrated provirus (Figure 2).10,14,16 quiescent cell types, including many relevant targets for Although integration is generally described as the gene transfer. One of the greatest benefits of LVs is their specific end point of lentiviral gene transfer, DNA ability to integrate a transgene copy into target cell episomes are also generated by lentiviruses and their chromosomes, allowing for long-term expression. How- derived vectors.15,17,18 Among these, the linear DNA ever, this process carries finite risks of detrimental episome is the precursor to the integrated provirus insertional mutagenesis, prompting examination of (Figure 2).11,15,19 Two types of circular episomes with intact alternatives to vector-mediated integration. viral coding regions are also produced. Homologous recombination (HR) within the LTRs generates a circular episome with a single LTR (1-LTR circle).15,17 In addition, LV transduction biology ligation of nicks in the circular DNA intermediates during the final stage of reverse transcription also results in 1-LTR After vector internalization and reverse transcription in circle episome formation.15,20 Non-homologous end joining the cytoplasm, the vector double-stranded DNA is of the linear episome results in a circular episome with two incorporated into the pre-integration complex. A critical adjacent LTRs (2-LTR circle) (Figure 2).15,17,21 Generally, component of the integration complex is viral integrase 1-LTR circles are more prevalent than 2-LTR circles, but (IN), a 32–45 kDa protein,3–7 which catalyzes viral DNA ratios vary.17,18,22–24 Both 1- and 2-LTR episomes can express integration into the host genome (Figure 1).8–10 Once the proteins because the genome or transgene remains intact.25,26 integration complex reaches the nucleus, IN mediates IN can also auto-integrate within the viral DNA itself, but integration between vector and host DNA. the disruption of the viral coding region generally renders Before nuclear translocation,4,11,12 IN catalyzes the the resulting products inactive.17,27 removal of two nucleotides from the 30 end of the Episomes do not persist in dividing cells because they reverse-transcribed genome at an invariant CA dinucleo- are not integrated and lack an origin of replication (ORI). tide.9,10,13 This 30-processing forms hydroxyl groups It was originally hypothesized that 1- and 2-LTR episomes were rapidly degraded as shown by PCR Correspondence: Dr PB McCray Jr, Department of Pediatrics, 240G analysis of 2-LTR circles present in peripheral blood EMRB, Carver College of Medicine, The University of Iowa, Iowa mononuclear cells from HIV infected patients.18 How- City, IA 52242, USA. ever, subsequent studies determined that the apparent E-mail: [email protected] half-life of these episomes correlated with the rate of cell Received 13 June 2009; revised 16 September 2009; accepted 18 23,28 September 2009; published online 22 October 2009 division. Therefore, viral episomes are stable and the Integrase-defective LVs: progress and applications MB Banasik and PB McCray Jr 151

Figure 1 Diagram of HIV IN. A schematic representation of the three domains of IN. Sites of IDLV mutations are indicated by arrows. Mutations affecting the catalytic triad are in bold. Mutations affecting genomic DNA binding are underlined. Other IDLV mutations are italicized.

cancer. Retroviral integration site selection was initially hypothesized to be random, based largely on in vitro studies.6,32 However, the risks of insertional mutagenesis were highlighted during the X-SCID retroviral gene therapy trials.33–37 Unexpectedly, five patients developed T-cell such as leukemia as a result of vector treatment.33,34,36,37 According to the model of site selection at the time, multiple oncogenic events should have been exceedingly rare. Extensive integration site mapping studies subsequently showed that retroviral integration site selection is not random and varies between gammaretroviruses, lentiviruses and other classes of integrating viral and non-viral vectors. Studies show that gammaretroviruses preferentially integrate in 50 flanking regions of genes near transcription start sites, increasing the possibility of affecting nearby gene expression.32 Unlike gammaretroviruses, LV do not exhibit an increased integration site preference for 50 flanking regions and avoid CpG islands, a character- istic feature found near transcription start sites.32 Instead, this vector class integrates across the entire transcrip- tional unit.32,38–41 A recent study showed that LV with chimeric gammaretroviral LTRs were still significantly less genotoxic than gammaretroviral vectors.42 Therefore, LV site selection preferences are distinct from gammaretroviral vectors and may be less genotoxic for therapeutic uses. Though the probability of aberrant gene activation or disruption are less likely after LV integration, insertional mutagenesis risks must be considered for clinical applications.2,30,33,34,43 One approach to circumvent detrimental insertional Figure 2 Representation of integration process. Integration begins 0 mutagenesis risks is the generation of a vector with with IN multimerization and vector DNA binding. LTR 3 cleavage properties of specific ‘safe’ site integration.44,45 However, generates hydroxyl groups necessary for the catalytic process. IN binds the host genomic DNA and integration occurs. A variety of mutations permanent transgene expression is not necessary or block different steps of this process, indicated below each stage, desirable in all applications. An alternative to a perma- rendering the vector IN defective. This prevents provirus formation nently integrated transgene is transient expression from and vector DNA remains as one of three primary episome types. an integration-defective LV vector (termed IDLV, also INdef, IN- or NILV). Non-integrating gammaretroviral vectors have been described,46 but because they cannot observed loss occurred from dilution of the episomes transduce non-dividing cells, several clinically relevant during cell replication.23,28 cell types are not effectively targeted. However, LV transduce slowly dividing cell types; therefore, episomal vector DNA is diluted slowly and persists.47–49 Several LV integration reports have detailed the production of HIV IDLVs.13,22,44,47–60 In addition, feline immunodeficiency As LVs introduce a permanent transgene copy into the virus (FIV) IDLVs are also described.48,61 host genome, they achieve persistent gene expression.15,19,29–31 This may correct a monogenic recessive disorder. However, LV integration is inherently muta- IDLV mutations genic and could induce detrimental side effects depending on the specific host gene disrupted. For instance, the IN catalyzes LV integration, with HIV IN the most activation of a protooncogene or disruption of a tumor extensively studied. IN has three separate protein suppressor gene could lead to the development of domains as determined by proteolysis and functional

Gene Therapy Integrase-defective LVs: progress and applications MB Banasik and PB McCray Jr 152 studies (Figure 1).12,62 The N-terminal domain contains from non-integrating lentiviral infections26,72,76 and other an HH-CC zinc-finger binding domain that primarily non-integrating viruses. binds viral DNA5,63,64 and facilitates required IN After confirmation of lentiviral episomal persis- multimerization.65,66 The C-terminal domain is the least tance,23,28 the potential of IDLVs was further examined. conserved retroviral IN domain5,67,68 and binds non- The first successful IDLVs exhibited roughly equivalent specific target DNA.63,69 The IN core domain contains the gene expression levels in growth arrested cells from both D-D35E catalytic amino-acid triad motif, comprised of HIV- and FIV-derived vectors.48 Some later reports also amino-acids D64, D116 and E152 in HIV.6,70 These indicated equivalent levels of gene expression from residues are absolutely conserved among retroviral INs IDLVs.13,49 In general, LV and IDLV expression equiv- and are similar to other transposase active sites.3,4,12,63,70–72 alency was derived from the percentage of cells Mutations within IN often form IDLVs. Class I transduced as opposed to more sensitive means, such mutations refer to specific integration deficiencies that as measuring total protein production.13,48,49 leave other viral processes intact. Class II mutations Other publications suggested that IDLVs do not impair multiple stages of the viral life cycle.48,72 Class I support levels of gene expression equivalent to integrat- IN mutations are more suitable than class II mutations ing LVs. An early IDLV report comparing both the for generating IDLVs, because the pleotropic effects percentage of cells transduced and GFP mean fluores- associated with class II mutations make them unsuitable cence intensity observed a slight reduction of IDLV for vector development. A variety of point mutations efficiency.58 Later work using similar methodologies produce IDLVs and affect different stages of the confirmed that while IDLVs had similar transduction integration process (Figures 1 and 2). percentages, GFP fluorescence values were sup- The most commonly used IDLV mutations introduce a pressed.22,47,50–52,58,59 The level of reduction varies missense amino acid in the catalytic triad (Figure 1). An between reports, from approximately one half50,52 to alteration in any of these residues inactivates the catalytic one tenth that of integrating LV.22,51 Observations that an properties of IN. Of these potential mutation target sites, IDLV generated reduced humoral immune responses D64 is most commonly altered,13,22,44,48–53,55–57,60 but may also indicate reduced protein expression.54 The mutations to D116 have also been reported.51,54,58,59 results from our laboratory indicate that FIV IDLV Additional mutations affect IN DNA binding, linear luciferase reporter transgene expression is B10-fold less episome processing or IN multimerization (Figure 2). IN than a matched integrating FIV LV control (MB Banasik must successfully bind both vector DNA and genomic and PB McCray, unpublished observation). DNA to support integration. Vector DNA binding is A recent study by Bayer et al.22 focused extensively on impaired by altering Q148.50 By mutating the RRK(262– the differences in gene expression achieved from HIV 264) motif in the C-terminal domain, IN fails to bind IDLVs. Initially, the authors wished to clarify discrepan- target genomic DNA.47,50,73 Both N120 and W235 are also cies from early, non-functional HIV IDLV reports.61,74,75 implicated in genomic DNA binding and generate IDLV These initial IDLVs contained full-length LTRs, as mutations when altered.50 In addition, mutating the opposed to U3 deleted, self-inactivating (SIN) LTRs. K264/K266/K273 triad impairs both target DNA binding The authors hypothesized that the HIV full-length LTRs and strand transfer.47,50 As IN multimerization is contain cis acting regulatory elements that prevented required for proper function, blocking mutations, such episomal transgene expression. To address this, LVs and as those to H12, also generate IDLVs.51 For successful IDLVs with differing lengths of LTR SIN deletions were IN-mediated integration to occur, the vector LTRs compared for transduction and transgene expression are cleaved at an invariable CA dinucleotide. Mutations efficiency. The LTR length did not affect the roughly in these LTRs (att mutants) also render the vector equivalent percent transduction by LVs or IDLVs. integration deficient.13,50 However, the level of transgene expression was altered in IDLVs. A longer SIN LTR deletion improved both GFP and luciferase transgene expression. Even so, HIV IDLVs IDLV biology yielded significantly reduced gene expression. The inclusion of a long SIN deletion improved the IDLV Although IDLVs could be an important alternative to mean fluorescence and luciferase expression to B10–15% either integrating LVs or other non-integrating vectors, of a integrating LV.22 Therefore, despite early accounts of most envisioned IDLV applications require effective equivalent levels of gene expression from IDLVs, levels of transgene expression. The biological differences mounting evidence suggests that lentiviral episomes and similarities of IDLVs and integrating LVs were are a weaker template for effective gene expression. compared in several independent studies. Primary areas Most reports of IDLVs note similar vector titers of focus include transgene expression and the level of compared with integrating LVs. The most common integration deficiency. titering method is a p24 ELISA assay,13,50,52,53,57 with Early HIV IDLVs were ineffective at transgene additional comparisons of RT activity,48,54 vector expression.25,61,74,75 In vivo brain injections comparing genomes53 and transducing units as determined by a expression from HIV LV and D64V IDLV found no visual marker.13,22,50 Three groups report lower transduc- transgene expression from the IDLV.74 A later report tion efficiency in IDLVs as compared with LVs when examined the potential efficacy of HIV D64V IDLV in normalized to p24 content.47,50,58,59 Vargas et al.58,59 liver transduction experiments and again no expression reported B25% reduction in transduction from IDLVs was observed.75 Retinal injections of an FIV IDLV also as compared with integrating LVs. Philippe et al.47 yielded no transgene expression.61 From these early reported a 10-fold decrease in TU/ng p24 in IDLVs, experiments, integration was hypothesized to be and Apolonia et al.50 reported a fivefold decrease using required for LV expression, despite conflicting evidence similar methods. Although these findings may reflect the

Gene Therapy Integrase-defective LVs: progress and applications MB Banasik and PB McCray Jr 153 earlier noted lower levels of transgene expression, they transduction of hematopoietic stem cells, a rapidly may also represent a deficiency in initial vector particle dividing population, in which expression was noted for formation. Titer differences may also vary depending on 1 month.13 Both muscle and dendritic cells supported the lentiviral backbone. We have reproducibly observed IDLV expression for up to 3 months.50,52 Several studies a 10-fold decrease in FIV IDLV titers when quantified by show efficient IDLV expression in the brain, with the transducing units, viral genomes or RT activity (MB longest duration examined for 4 months.22,47,49,56 The Banasik and PB McCray, unpublished observation). longest IDLV brain transduction report was particularly Although LVs and IDLVs should transduce similar cell notable because the vector was delivered in utero.56 The types, endogenously expressed proteins may further same report also followed transgene expression in the influence transgene expression from unintegrated epi- spinal cord for 14 days, but did not examine gene somes. IDLVs had different expression levels in various expression duration further.56 Liver transduction with T-cell-derived cell lines and primary cells.77 IDLVs also IDLV vectors was observed for as long as 6 months.22,50 failed to express in the corpus callosum and expression To date, the longest IDLV study was in murine retina. was reduced in astrocytes as compared with integrating After transduction, efficient GFP expression was LVs.22 Although these are limited examples, they raise observed in the retinal pigment epithelium for up to 9 the possibility that endogenous factors may influence months.49 expression from non-integrated episomes. IDLVs lack viral protein-mediated integra- 4,13,22,44,47–51,53–55,57,72,77,78 tion. However, IN-independent INdef vector therapeutic applications insertion still functions in transduced cells.13,77,78 The level of background integration varies by report, even As IDLV-mediated transgene expression is long lasting in when examining the same IDLV mutation. Residual slowly dividing cell types,22,47,49,50,56 non-integrated, integration from the most commonly used D64 catalytic corrective gene expression may be achieved in these mutation ranges from 104 times less4,13,77 to 10 times tissue types. Both mouse and rat models of retinitis less55 than integrating LVs, with additional reports pigmentosa had visual improvement, as shown by indicating that D64 IDLVs are 103 to 102 times less electroretinography, after transgene appropriate IDLV integrative.50,51,53,57 D116 IDLV mutants are associated delivery.49 Phenotypic correction was stable for 8 weeks with 103 to 102 times less integration.4,51 Similar varia- after injection with no loss of improvement in IDLV- tions in residual integration are also reported for treated animals.49 RRK(262–264) and att IDLVs, ranging from 104 to 102- Although IDLVs lack viral protein-mediated proviral fold less and 104 to 10-fold less, respectively.13,47,50,77 insertion, several reports have used IDLVs as a platform Although the integration frequency of IDLV is greatly for transgene integration through alternate pathways. reduced, background integration that occurs at a similar One option is to use episomes as templates for HR for frequency as naked DNA may be unavoidable. either gene-specific correction or ‘safe site’ integration. Okada et al. used a neomycin resistance transgene cassette flanked by ‘arms’ complementary to the calme- Persistence of expression from IDLV gin gene to generate mouse ES cells with a site-specific transgene insertion. Several neomycin resistant ES cell Although IDLV episomal DNA expresses a transgene, lines were established after IDLV treatment, and a small this vector form dissipates with cell division.23,28 The proportion had gene-specific integration (0.83%).55 Gene duration of transgene expression varies depending on targeting from IDLVs was less efficient than more how rapidly transduced cells turn over. The length of traditional plasmid electroporation. The authors this expression has been characterized in a number of hypothesized that vector DNA incorporation into the in vitro and in vivo settings. integration complex may hinder access to endogenous Expression duration in cell culture systems is highly HR components.55 Two other groups have also studied transient because of rapid cell turnover.22,47,51 If cell targeted integration through HR, but induced double- division is decreased, expression can be maintained for a stranded breaks, which improve HR frequency. By greater time. IDLV-mediated expression was short lived introducing two IDLVs carrying either an ISce-I restric- in dividing CRFK and HT1080 cells. However, cell tion enzyme expression cassette or a non-expressed lacZ- growth arrest before IDLV transduction enhanced ex- eGFP HR template, Cornu and Cathomen51 reported site- pression duration.48 A similar duration increase was specific targeting and correction of a disrupted lacZ- observed between dividing and non-dividing C2C12 eGFP gene. More significantly, Lombardo et al.44 used cultures.50 Alternatively, if the transgene cassette carries IDLVs expressing zinc-finger nucleases to induce site- a viral ORI, IDLV transgene expression persists if specific double-stranded breaks. The authors primarily necessary trans proteins are supplied. By including the used a three IDLV transduction protocol, in which the SV40 ORI in IDLVs, transgene expression lasted 56 days, vectors expressed one of the two necessary zinc-finger the final time point examined, in the SV40 T-antigen nucleases or carried the donor template. Using this containing 293T cell line. This contrasted to parallel method, successful gene correction was observed at the transductions in 293 cells, with no SV40 T-antigen, in IL2RG locus. In addition, site-specific integration into the which transgene expression was short lived.58 CCR5 locus targeted 50% of transduced cultured cells In vivo studies have primarily focused on the and 5% of human ES cells.44 The CCR5 locus is transduction of slowly dividing tissue populations to hypothesized to be ‘safe’ for targeted integration because maximize IDLV expression duration. In all cases, individuals homozygous for null mutations in the gene transgene expression remained constant throughout the have no apparent health detriments.44,79 More recently, observed experiment. An exception was the ex vivo Lombardo et al. reported evidence that targeted insertion

Gene Therapy Integrase-defective LVs: progress and applications MB Banasik and PB McCray Jr 154 of gene cassettes at the CCR5 locus results in aberrant expression.81–83 Minicircle plasmid technology can re- expression of nearby endogenous genes and lower levels move these undesirable bacterial sequences, but proto- of integrated transgene expression, as compared with the cols for clinical grade preparations are still in their AAVS1 locus. These preliminary results suggest that infancy.83 Negri et al.54 transduced BALB/c mice with an targeted integration at the CCR5 locus may not be as IDLV expressing HIV-1 gp120. After a single dose, robust desirable as earlier envisioned.80 Additional studies cellular and humoral immune responses were detected should further clarify this situation. in IDLV-treated animals for up to 90 days, the limit of the IDLVs may also serve as a platform for DNA experiment. These results were a promising proof-of- transposon system delivery and expression. By including principle for using IDLVs as vaccines, but efficacy testing necessary cis elements in the transgene cassette, vector was limited to quantification of immune response and episomes can serve as appropriate templates for trans- not expanded to viral challenge. Recently, Coutant et al.52 poson-mediated gene addition. An IDLV expressing the further showed IDLV vector immunization. Mice were yeast Flp recombinase conferred site-specific integration treated with a single vector dose expressing secreted in an engineered human cell line.53 By including Flp West Nile virus envelope protein. The IDLV elicited a recombinase targets in a transgene cassette with a similar magnitude of response as compared with hygromycin-B construct lacking the ATG start codon, integrating LV at higher doses. After immune response the authors screened targeted integrations in 293 cell confirmation, animals were challenged with a lethal dose lines that contained matching Flp recombinase targets of West Nile virus. The immune response elicited by the sites downstream of a functional promoter and ATG IDLV conferred protective immunity and survival.52 codon. After dual IDLV transduction with Flp and These results illustrate the potential for IDLV vaccine the faulty hygromycin-B cassette, hygromycin-B resistant applications. colonies were recovered (58±6 colonies per 7 Â 105 cells transduced), though at a much lesser efficiency than when integrating LV or plasmid expression of Flp Future directions were used.53 Two recent reports described Sleeping Beauty (SB) IDLVs show promise for achieving gene expression transposase-mediated transposition when delivered without integration, preserving some benefits of LVs, using an IDLV.57,60 Unlike retroviral integration, SB whereas reducing the potentially detrimental risk of transposition shows little preference for specific genomic insertional mutagenesis. Numerous reports support the regions.57,60 Both studies constructed two IDLVs, one ability of these vectors to confer long-term gene expres- expressing a hyperactive SB transposase and a second sion in slowly dividing cell types for potentially containing a transposable element with a drug resistance corrective gene expression. These reports also highlight transgene cassette and necessary inverted repeats. additional applications of these vectors as delivery Integration site analysis in drug resistant colonies platforms for alternate integrative pathways, suicide revealed that cells transduced with the dual IDLVs gene therapy and vaccines. As discussed below, several showed an integration pattern similar to SB, and areas for further research remain. dissimilar from LV integration preferences (87 sites57 Earlier reports have noted lower gene expression and 119 sites60 analyzed). Furthermore, no viral-derived levels from IDLVs,22,47,50–52,58,59 presumably because elements (that is LTRs) were integrated with the episomes are not as effective as chromosomal vector transposon cassette.60 Therefore, using IDLVs as platform proviruses. The mechanism for this difference between for transposon delivery could reduce the risk of inser- episomal and vector provirus transgene expression is tional mutagenesis by disfavoring transgene insertion in currently unknown. As LVs integrate into transcription- gene coding regions. ally active regions, local chromatin architecture and IDLVs also show promise for suicide gene therapy transcription factor-binding sites may aid expression. As applications in virally transformed cancers. As discussed episomes are unintegrated, these regional features are earlier, effective episomal replication occurred in the lacking. Similarly, various trans activating factors are SV40 T-antigen expressing 293T cell line if the SV40 ORI hypothesized to affect episome expression and the was included in the transgene cassette.58 Extending these prevalence of these factors may vary among target cell results, a later study used the thymidine kinase suicide type.22,77 Although identification of these trans acting gene to render cells sensitive to ganciclovir treatment.59 factors may prove difficult, knowledge of such elements In IDLV transduced 293T cells, ganciclovir treatment was could improve IDLV design by several means. These cytotoxic even 4 weeks post transduction because of non- include ensuring that desirable trans acting factor- integrated lentiviral episome replication.59 Although binding sites are retained, incorporating additional these studies used SV40 transformed cells, other viruses, binding sites to increase expression efficiency, and such as the human papilloma virus and Epstein–Barr focusing on IDLV development of LV backbones with virus, could be targeted if their specific ORIs were favorable trans acting factor binding. Identification of included in the transgene cassette.59 such trans acting factors may also suggest cell and tissue As IDLV expression is inherently transient, it may be targets with favorable levels of expression. ideal for vaccines. Selected viral or bacterial genes could In addition, increased in vivo testing of IDLVs would be expressed, eliciting a protective immune response. be beneficial. Though LVs and IDLVs should transduce Purified, plasmid-free IDLV vaccine vectors offer alter- similar cell profiles when pseudotyped equivalently, natives to plasmid DNA vaccines. The bacterial plasmid differences in cell-type transduction in the brain have backbone and drug resistance genes contain many CpG been noted.22 Although this may be a unique result, it sequences, which are prone to silencing and have been would be interesting to learn if any other cell transduc- implicated in transient and low levels of transgene tion differences occur after IDLV transduction. The

Gene Therapy Integrase-defective LVs: progress and applications MB Banasik and PB McCray Jr 155 majority of in vivo testing has focused on longitudinal 7 Talbott RL, Sparger EE, Lovelace KM, Fitch WM, Pedersen NC, studies of IDLV expression of a marker trans- Luciw PA et al. Nucleotide sequence and genomic organization gene.22,47,49,50,52,56 As reviewed above, limited studies of feline immunodeficiency virus. Proc Natl Acad Sci USA 1989; show potential applications for corrective gene expres- 86: 5743–5747. sion or vaccines. These reports indicate a favorable 8 Nisole S, Saib A. Early steps of retrovirus replicative cycle. profile for IDLV because of the sustained visual Retrovirology 2004; 1:9. correction or protective immunity achieved after admin- 9 Craigie R, Fujiwara T, Bushman F. The IN protein of Moloney istration of a single vector dose.49,52,54 Therapeutic gene murine leukemia virus processes the viral DNA ends and expression studies could be expanded to other tissue accomplishes their integration in vitro. Cell 1990; 62: 829–837. types with robust INdef vector expression. IDLV delivery 10 Katz RA, Merkel G, Kulkosky J, Leis J, Skalka AM. The avian of non-viral integration platforms, such as zinc-finger retroviral IN protein is both necessary and sufficient for integrative recombination in vitro. Cell 1990; 63: 87–95. nucleases-mediated HR or transposase-mediated 44,57,60 11 Katz RA, Skalka AM. The retroviral enzymes. Annu Rev Biochem integration, show promise in vitro. It would be 1994; 63: 133–173. interesting to translate these procedures in vivo, 12 Engelman A, Craigie R. Identification of conserved amino acid especially in tissue types inaccessible to plasmid deliv- residues critical for human immunodeficiency virus type 1 84–86 ery. IDLVs that are pseudotyped with a cell-type integrase function in vitro. J Virol 1992; 66: 6361–6369. specific envelope glycoprotein could be systemically or 13 Nightingale SJ, Hollis RP, Pepper KA, Petersen D, Yu XJ, Yang C topically delivered, bypassing the need for direct tissue et al. Transient gene expression by nonintegrating lentiviral injection. Not all cell types are easily transfectable, vectors. Mol Ther 2006; 13: 1121–1132. limiting the usefulness of plasmid-based transgene 14 Mizuuchi K. Polynucleotidyl transfer reactions in transpositional expression in some settings.87 By harnessing the entry DNA recombination. J Biol Chem 1992; 267: 21273–21276. properties of viral vectors, the inefficiencies of plasmid 15 Goff SP. Retroviridae: the retroviruses and their replication. In: delivery can be overcome. The potential for suicide gene David M Knipe PMH, Diane E Griffin, Malcolm A Martin, therapy of virus transformed cells was shown in vitro Robert A Lamb, Bernard Roizman, Stephen E Strauss (ed). Fields with SV40 altered cells.59 Not only would it be Virology, 4th edn, vol. 2. Lippincott Williams & Wilkins: informative to test this system in vivo, it would also be Philadelphia, 2001. pp 1871–1940. of interest to confirm that other transformative viruses 16 Yoder KE, Bushman FD. Repair of gaps in retroviral DNA could be targeted. integration intermediates. J Virol 2000; 74: 11191–11200. In conclusion, IDLVs represent a class of novel viral 17 Farnet CM, Haseltine WA. Circularization of human immuno- vectors with several established and emerging applica- deficiency virus type 1 DNA in vitro. J Virol 1991; 65: 6942–6952. tions. These vectors support lower levels of gene 18 Pauza CD, Trivedi P, McKechnie TS, Richman DD, Graziano FM. 2-LTR circular viral DNA as a marker for human immunode- expression, and efforts should be made to identify ficiency virus type 1 infection in vivo. Virology 1994; 205: 470–478. potential expression limiting trans acting factors and to 19 Pages JC, Bru T. Toolbox for retrovectorologists. J Gene Med 2004; boost episomal gene expression. A number of potential 6 (Suppl 1): S67–S82. applications have been shown for IDLVs, but further 20 Miller MD, Wang B, Bushman FD. Human immunodeficiency in vivo testing is required to show the utility of these virus type 1 preintegration complexes containing discontinuous techniques. plus strands are competent to integrate in vitro. J Virol 1995; 69: 3938–3944. 21 Li L, Olvera JM, Yoder KE, Mitchell RS, Butler SL, Lieber M et al. Conflict of interest Role of the non-homologous DNA end joining pathway in the early steps of retroviral infection. EMBO J 2001; 20: 3272–3281. The authors declare no conflict of interest. 22 Bayer M, Kantor B, Cockrell A, Ma H, Zeithaml B, Li X et al. A large U3 deletion causes increased in vivo expression from a nonintegrating lentiviral vector. Mol Ther 2008; 16: 1968–1976. References 23 Butler SL, Johnson EP, Bushman FD. Human immunodeficiency virus cDNA metabolism: notable stability of two-long terminal 1 Kumar M, Keller B, Makalou N, Sutton RE. Systematic repeat circles. J Virol 2002; 76: 3739–3747. determination of the packaging limit of lentiviral vectors. Hum 24 Nandi JS. Unintegrated viral DNA as a marker for human Gene Ther 2001; 12: 1893–1905. immunodeficiency virus 1 infection in vivo and in vitro. Acta Virol 2 Sinn PL, Sauter SL, McCray Jr PB. Gene therapy progress and 1999; 43: 367–372. prospects: development of improved lentiviral and retroviral 25 Brussel A, Sonigo P. Evidence for gene expression by vectors—design, biosafety, and production. Gene Therapy 2005; unintegrated human immunodeficiency virus type 1 DNA 12: 1089–1098. species. J Virol 2004; 78: 11263–11271. 3 Leavitt AD, Shiue L, Varmus HE. Site-directed mutagenesis of 26 Stevenson M, Haggerty S, Lamonica CA, Meier CM, Welch SK, HIV-1 integrase demonstrates differential effects on integrase Wasiak AJ. Integration is not necessary for expression of human functions in vitro. J Biol Chem 1993; 268: 2113–2119. immunodeficiency virus type 1 protein products. J Virol 1990; 64: 4 Leavitt AD, Robles G, Alesandro N, Varmus HE. Human 2421–2425. immunodeficiency virus type 1 integrase mutants retain in vitro 27 Shoemaker C, Hoffman J, Goff SP, Baltimore D. Intramolecular integrase activity yet fail to integrate viral DNA efficiently integration within Moloney murine leukemia virus DNA. J Virol during infection. J Virol 1996; 70: 721–728. 1981; 40: 164–172. 5 Shibagaki Y, Holmes ML, Appa RS, Chow SA. Characterization 28 Pierson TC, Kieffer TL, Ruff CT, Buck C, Gange SJ, Siliciano RF. of feline immunodeficiency virus integrase and analysis of Intrinsic stability of episomal circles formed during functional domains. Virology 1997; 230: 1–10. human immunodeficiency virus type 1 replication. J Virol 2002; 6 Shibagaki Y, Chow SA. Central core domain of retroviral 76: 4138–4144. integrase is responsible for target site selection. J Biol Chem 29 Barquinero J, Eixarch H, Perez-Melgosa M. Retroviral vectors: new 1997; 272: 8361–8369. applications for an old tool. Gene Therapy 2004; 11(Suppl 1): S3–S9.

Gene Therapy Integrase-defective LVs: progress and applications MB Banasik and PB McCray Jr 156 30 Baum C, Schambach A, Bohne J, Galla M. Retrovirus vectors: bility to expression in nondividing cells: analysis with class I toward the plentivirus? Mol Ther 2006; 13: 1050–1063. integrase mutants. J Virol 2004; 78: 2906–2920. 31 Rodrigues T, Carrondo MJ, Alves PM, Cruz PE. Purification of 49 Yanez-Munoz RJ, Balaggan KS, MacNeil A, Howe SJ, Schmidt M, retroviral vectors for clinical application: biological implications Smith AJ et al. Effective gene therapy with nonintegrating and technological challenges. J Biotechnol 2007; 127: 520–541. lentiviral vectors. Nat Med 2006; 12: 348–353. 32 Wu X, Li Y, Crise B, Burgess SM. Transcription start regions in 50 Apolonia L, Waddington SN, Fernandes C, Ward NJ, Bouma G, the human genome are favored targets for MLV integration. Blundell MP et al. Stable gene transfer to muscle using non- Science 2003; 300: 1749–1751. integrating lentiviral vectors. Mol Ther 2007; 15: 1947–1954. 33 Hacein-Bey-Abina S, von Kalle C, Schmidt M, Le Deist F, 51 Cornu TI, Cathomen T. Targeted genome modifications Wulffraat N, McIntyre E et al. A serious adverse event after using integrase-deficient lentiviral vectors. Mol Ther 2007; 15: successful gene therapy for X-linked severe combined immuno- 2107–2113. deficiency. N Engl J Med 2003; 348: 255–256. 52 Coutant F, Frenkiel MP, Despres P, Charneau P. Protective 34 Hacein-Bey-Abina S, Von Kalle C, Schmidt M, McCormack MP, antiviral immunity conferred by a nonintegrative lentiviral Wulffraat N, Leboulch P et al. LMO2-associated clonal T cell vector-based vaccine. PLoS One 2008; 3: e3973. proliferation in two patients after gene therapy for SCID-X1. 53 Moldt B, Staunstrup NH, Jakobsen M, Yanez-Munoz RJ, Science 2003; 302: 415–419. Mikkelsen JG. Genomic insertion of lentiviral DNA circles 35 Hacein-Bey-Abina S, Le Deist F, Carlier F, Bouneaud C, Hue C, directed by the yeast Flp recombinase. BMC Biotechnol 2008; De Villartay JP et al. Sustained correction of X-linked severe 8: 60. combined immunodeficiency by ex vivo gene therapy. N Engl J 54 Negri DR, Michelini Z, Baroncelli S, Spada M, Vendetti S, Med 2002; 346: 1185–1193. Buffa V et al. Successful immunization with a single injection 36 Howe SJ, Mansour MR, Schwarzwaelder K, Bartholomae C, of non-integrating lentiviral vector. Mol Ther 2007; 15: Hubank M, Kempski H et al. Insertional mutagenesis combined 1716–1723. with acquired somatic mutations causes leukemogenesis follow- 55 Okada Y, Ueshin Y, Hasuwa H, Takumi K, Okabe M, Ikawa M. ing gene therapy of SCID-X1 patients. J Clin Invest 2008; 118: Targeted gene modification in mouse ES cells using integrase- 3143–3150. defective lentiviral vectors. Genesis 2009; 47: 217–223. 37 Hacein-Bey-Abina S, Garrigue A, Wang GP, Soulier J, Lim A, 56 Rahim AA, Wong AM, Howe SJ, Buckley SM, Acosta-Saltos AD, Morillon E et al. Insertional oncogenesis in 4 patients after Elston KE et al. Efficient gene delivery to the adult and fetal CNS retrovirus-mediated gene therapy of SCID-X1. J Clin Invest 2008; using pseudotyped non-integrating lentiviral vectors. Gene 118: 3132–3142. Therapy 2009; 16: 509–520. 38 Mitchell RS, Beitzel BF, Schroder AR, Shinn P, Chen H, Berry CC 57 Staunstrup NH, Moldt B, Mates L, Villesen P, Jakobsen M, Ivics et al. Retroviral DNA integration: ASLV, HIV, and MLV show Z et al. Hybrid lentivirus-transposon vectors with a random distinct target site preferences. PLoS Biol 2004; 2: E234. integration profile in human cells. Mol Ther 2009; 17: 1505–1514. 39 Schroder AR, Shinn P, Chen H, Berry C, Ecker JR, Bushman F. 58 Vargas Jr J, Gusella GL, Najfeld V, Klotman ME, Cara A. Novel HIV-1 integration in the human genome favors active genes and integrase-defective lentiviral episomal vectors for gene transfer. local hotspots. Cell 2002; 110: 521–529. Hum Gene Ther 2004; 15: 361–372. 40 Wang GP, Ciuffi A, Leipzig J, Berry CC, Bushman FD. HIV 59 Vargas Jr J, Klotman ME, Cara A. Conditionally replicating integration site selection: analysis by massively parallel pyrose- lentiviral-hybrid episomal vectors for suicide gene therapy. quencing reveals association with epigenetic modifications. Antiviral Res 2008; 80: 288–294. Genome Res 2007; 17: 1186–1194. 60 Vink CA, Gaspar HB, Gabriel R, Schmidt M, McIvor RS, 41 Kang Y, Moressi CJ, Scheetz TE, Xie L, Tran DT, Casavant TL Thrasher AJ et al. Sleeping beauty transposition from noninte- et al. Integration site choice of a feline immunodeficiency virus grating lentivirus. Mol Ther 2009; 17: 1197–1204. vector. J Virol 2006; 80: 8820–8823. 61 Loewen N, Leske DA, Chen Y, Teo WL, Saenz DT, Peretz M et al. 42 Montini E, Cesana D, Schmidt M, Sanvito F, Bartholomae CC, Comparison of wild-type and class I integrase mutant-FIV Ranzani M et al. The genotoxic potential of retroviral vectors is vectors in retina demonstrates sustained expression of inte- strongly modulated by vector design and integration site grated transgenes in retinal pigment epithelium. J Gene Med selection in a mouse model of HSC gene therapy. J Clin Invest 2003; 5: 1009–1017. 2009; 119: 964–975. 62 van Gent DC, Oude Groeneger AA, Plasterk RH. Identification 43 Metais JY, Dunbar CE. The MDS1-EVI1 gene complex as a of amino acids in HIV-2 integrase involved in site-specific retrovirus integration site: impact on behavior of hematopoietic hydrolysis and alcoholysis of viral DNA termini. Nucleic Acids cells and implications for gene therapy. Mol Ther 2008; 16: Res 1993; 21: 3373–3377. 439–449. 63 Khan E, Mack JP, Katz RA, Kulkosky J, Skalka AM. Retroviral 44 Lombardo A, Genovese P, Beausejour CM, Colleoni S, Lee YL, integrase domains: DNA binding and the recognition of LTR Kim KA et al. Gene editing in human stem cells using zinc finger sequences. Nucleic Acids Res 1991; 19: 851–860. nucleases and integrase-defective lentiviral vector delivery. Nat 64 Ellison V, Brown PO. A stable complex between integrase and Biotechnol 2007; 25: 1298–1306. viral DNA ends mediates human immunodeficiency virus 45 Maeder ML, Thibodeau-Beganny S, Osiak A, Wright DA, integration in vitro. Proc Natl Acad Sci USA 1994; 91: 7316–7320. Anthony RM, Eichtinger M et al. Rapid ‘open-source’ engineer- 65 Engelman A, Bushman FD, Craigie R. Identification of discrete ing of customized zinc-finger nucleases for highly efficient gene functional domains of HIV-1 integrase and their organization modification. Mol Cell 2008; 31: 294–301. within an active multimeric complex. EMBO J 1993; 12: 46 Yu SS, Dan K, Chono H, Chatani E, Mineno J, Kato I. Transient 3269–3275. gene expression mediated by integrase-defective retroviral 66 Ellison V, Gerton J, Vincent KA, Brown PO. An essential vectors. Biochem Biophys Res Commun 2008; 368: 942–947. interaction between distinct domains of HIV-1 integrase med- 47 Philippe S, Sarkis C, Barkats M, Mammeri H, Ladroue C, Petit C iates assembly of the active multimer. J Biol Chem 1995; 270: et al. Lentiviral vectors with a defective integrase allow efficient 3320–3326. and sustained transgene expression in vitro and in vivo. Proc Natl 67 Craigie R. HIV integrase, a brief overview from chemistry to Acad Sci USA 2006; 103: 17684–17689. therapeutics. J Biol Chem 2001; 276: 23213–23216. 48 Saenz DT, Loewen N, Peretz M, Whitwam T, Barraza R, Howell 68 Lu R, Limon A, Ghory HZ, Engelman A. Genetic analyses of KG et al. Unintegrated lentivirus DNA persistence and accessi- DNA-binding mutants in the catalytic core domain of human

Gene Therapy Integrase-defective LVs: progress and applications MB Banasik and PB McCray Jr 157 immunodeficiency virus type 1 integrase. J Virol 2005; 79: 77 Nakajima N, Lu R, Engelman A. Human immunodeficiency 2493–2505. virus type 1 replication in the absence of integrase-mediated dna 69 Engelman A, Hickman AB, Craigie R. The core and carboxyl- recombination: definition of permissive and nonpermissive terminal domains of the integrase protein of human immuno- T-cell lines. J Virol 2001; 75: 7944–7955. deficiency virus type 1 each contribute to nonspecific DNA 78 Gaur M, Leavitt AD. Mutations in the human immunodeficiency binding. J Virol 1994; 68: 5911–5917. virus type 1 integrase D,D(35)E motif do not eliminate provirus 70 Kulkosky J, Jones KS, Katz RA, Mack JP, Skalka AM. Residues formation. J Virol 1998; 72: 4678–4685. critical for retroviral integrative recombination in a region that is 79 Lim JK, Glass WG, McDermott DH, Murphy PM. CCR5: no highly conserved among retroviral/retrotransposon integrases longer a ‘good for nothing’ gene—chemokine control of West and bacterial insertion sequence transposases. Mol Cell Biol 1992; Nile virus infection. Trends Immunol 2006; 27: 308–312. 12: 2331–2338. 80 Lombardo A, Cesana D, Genovese P, Provasi E, Maruggi G, 71 Drelich M, Wilhelm R, Mous J. Identification of amino acid Binini C et al. Characterization of potential genomic ‘safe harbor’ residues critical for endonuclease and integration activities of for efficient targeted gene addition with zinc finger nucleases. HIV-1 IN protein in vitro. Virology 1992; 188: 459–468. Mol Ther 2009; 17: S168. 72 Wiskerchen M, Muesing MA. Human immunodeficiency virus 81 Mairhofer J, Grabherr R. Rational vector design for efficient non- type 1 integrase: effects of mutations on viral ability to integrate, viral gene delivery: challenges facing the use of plasmid DNA. direct viral gene expression from unintegrated viral DNA Mol Biotechnol 2008; 39: 97–104. templates, and sustain viral propagation in primary cells. J Virol 82 Chen ZY, He CY, Meuse L, Kay MA. Silencing of episomal 1995; 69: 376–386. transgene expression by plasmid bacterial DNA elements in vivo. 73 Lutzke RA, Plasterk RH. Structure-based mutational analysis Gene Therapy 2004; 11: 856–864. of the C-terminal DNA-binding domain of human immuno- 83 Mayrhofer P, Blaesen M, Schleef M, Jechlinger W. Minicircle-DNA deficiency virus type 1 integrase: critical residues for production by site specific recombination and protein-DNA protein oligomerization and DNA binding. J Virol 1998; 72: interaction chromatography. J Gene Med 2008; 10: 1253–1269. 4841–4848. 84 Davidson BL, Boudreau RL. RNA interference: a tool for 74 Naldini L, Blomer U, Gage FH, Trono D, Verma IM. Efficient querying nervous system function and an emerging therapy. transfer, integration, and sustained long-term expression of the Neuron 2007; 53: 781–788. transgene in adult rat brains injected with a lentiviral vector. 85 Ge Q, Filip L, Bai A, Nguyen T, Eisen HN, Chen J. Inhibition of Proc Natl Acad Sci USA 1996; 93: 11382–11388. influenza virus production in virus-infected mice by RNA 75 Park F, Ohashi K, Kay MA. Therapeutic levels of human factor interference. Proc Natl Acad Sci USA 2004; 101: 8676–8681. VIII and IX using HIV-1-based lentiviral vectors in mouse liver. 86 Snove Jr O, Rossi JJ. Expressing short hairpin RNAs in vivo. Nat Blood 2000; 96: 1173–1176. Methods 2006; 3: 689–695. 76 Wu Y. HIV-1 gene expression: lessons from provirus and 87 Scherr M, Eder M. Gene silencing by small regulatory RNAs in non-integrated DNA. Retrovirology 2004; 1: 13. mammalian cells. Cell Cycle 2007; 6: 444–449.

Gene Therapy