(2000) 7, 1613–1623  2000 Macmillan Publishers Ltd All rights reserved 0969-7128/00 $15.00 www.nature.com/gt VIRAL TRANSFER TECHNOLOGY RESEARCH ARTICLE Characterization of novel safe lentiviral vectors derived from simian immunodeficiency (SIVmac251) that efficiently transduce mature human dendritic cells

DNe`gre1, P-E Mangeot2, G Duisit3, S Blanchard1, P-O Vidalain4, P Leissner5, A-J Winter5, C Rabourdin-Combe4, M Mehtali5,6, P Moullier3, J-L Darlix2 and F-L Cosset1 1Vectorologie Re´trovirale and The´rapie Ge´nique, 2LaboRe´tro, Unite´ de Virologie Humaine, U412 INSERM, IFR 74 and ENS de Lyon; 4Immunobiologie Fondamentale et Clinique, U503 INSERM IFR 74 and ENS de Lyon; 3Laboratoire de The´rapie Ge´nique, CHU Hotel Dieu, Nantes; and 5Transge`ne SA, Strasbourg, France

We describe the generation and the characterization of new gene transfer, expression of the GFP in the target cells lentiviral vectors derived from SIVmac251, a simian immuno- remained constant after several weeks, indicating that the deficiency virus (SIV). A methodical approach was used to vectors had been stably integrated into the of the engineer both efficient and safe packaging constructs host cells. Preparations of SIV vectors were systematically allowing the production of SIV viral core . SIV-vec- checked for the absence of replication-competent and tors encoding GFP (green fluorescent ) were gener- recombinant but remained negative, suggesting ated as VSV-G-pseudotyped particles upon transient the innocuousness of these novel gene delivery vectors. expression of the vector construct and helper functions in Side-to-side comparisons with vectors derived from HIV-1 293 cells. The SIV vectors were able to transduce efficiently (human immunodeficiency virus) indicated that the SIV various target cell types at low multiplicity of infection, includ- vectors were equally potent in transducing proliferating ing monocyte-differentiated human dendritic cells (DCs) target cells. Finally, we have determined the infectivity of SIV which retained their capacity to differentiate into mature DCs vectors pseudotyped with surface glycoproteins of several after gene transfer. Transduction of the DCs by the SIV vec- membrane-enveloped . Gene Therapy (2000) 7, tors was prevented when infections were performed in the 1613–1623. presence of AZT, a reverse-transcriptase inhibitor. After

Keywords: lentiviral vectors; retroviruses; dendritic cells; pseudotypes; safety; packaging

Introduction where the target tissue is proliferating11 or under regener- ation,12 the most important limitation of MLV-derived The obtention of retroviral vectors that are suitable for vectors is their inability to allow transgene integration an in vivo gene delivery has been a recurrent theme in into nondividing or slowly proliferating target cells.13–15 gene therapy research over the last decade. Substantial The requirements by target cells to undergo mitosis to progress has been achieved towards this goal: (1) human be successfully transduced by MLV-based vectors are not complement-resistant retroviral vectors can now be gen- clear although it is currently assumed that the MLV pre- erated thanks to the development of human cell-derived integration complex (PIC), that is formed with the retro- 1,2 packaging cell lines and to the use of specific retroviral viral genome and some viral core proteins after virion 1,3 envelope glycoproteins; (2) methods that optimize the penetration and uncoating, cannot be transported production and quality of retroviral vectors have been through the nuclear pore and must await the nuclear 4,5 developed; (3) several strategies aimed at targeting membrane breakdown that accompanies mitosis to gain 6–8 gene delivery and/or expression have been described; access into the host chromatin.14,16 In contrast to MLVs, and (4) new vectors that allow gene delivery into non- are able to infect post-mitotic cells.17,18 The 9,10 proliferating cells have now emerged. reasons for this major difference are still unclear but seem Retroviral vectors derived from murine leukemia to rely on the ability of the PIC of lentiviruses to be trans- viruses (MLVs) and from other type C and D mammalian ported across the nuclear membrane.19 Several karyo- retroviruses have proved to be extremely useful and philic determinants in the lentiviral PIC may interact adaptable in transducing target cells cultivated in vitro or with the nuclear import machinery20–22 and may allow ex vivo. Although they may be used in vivo in situations PIC translocation into the nucleus.23 Possible strategies to obtain MLV-derived retroviral vectors capable of infecting nondividing cells may there- Correspondence: F-L Cosset, LVRTG, INSERM U412, Ecole Normale Supe´rieure de Lyon, 46 Alle´e d’Italie, 69364 Lyon Cedex 07, France fore consist in the substitution of MLV proteins by the 6Present address: IntroGene BV, Leiden, The Netherlands corresponding HIV proteins that permit interaction with Received 14 April 2000; accepted 5 July 2000 the nuclear import machinery24 or in the introduction of Safe lentiviral vectors derived from SIVmac251 DNe`gre et al 1614 NLS sequences into MLV PIC proteins. As an alternative To allow the rigorous comparison of these SIV core and more effective strategy, several groups have reported expression constructs and in order to evaluate their maxi- the generation of retroviral vectors derived from lenti- mal productivity of infectious viral particles, we first viruses such as HIV-1,25,26 HIV-2,27 SIV,28,29 feline sought to determine their capacity to package the GFP- immunodeficiency virus,30 or encoding SIV-T1 vector derived from SIVmac251 (Figure virus.31,32 The development of -based vectors, 1b), rather than a minimal SIVmac251-based vector such particularly those derived from HIV-1, is a rapidly pro- as the SIV-RMES4 vector (Figure 1b). Indeed, while the gressing field and several studies have demonstrated that SIV-T1 vector still contains residual sequences derived these vectors can transduce nondividing cells ex from SIV regulatory genes, of which only has retained vivo15,26,33,34 and in vivo.25,35–39 its coding capacity (Table 1, and unpublished results), the Despite all the safeguards that have progressively been more refined SIV-RMES4 vector is tat dependent and pro- introduced in lentiviral vectors, the clinical acceptance of vides significantly lower titers when generated with tat- vectors derived from pathogenic lentiviruses is still a sub- gene negative packaging constructs such as pSIV5 and ject of debate. It is therefore essential to design vectors subsequent plasmids (Table 1). derived from a wide range of lentivirus types and to All the different SIV gag– expression constructs examine comparatively their properties in terms of trans- were comparatively evaluated for their capacity to com- duction efficiency and biosafety. Here, we report the plement in trans the formation of SIV vectors in a transi- characterization of lentivirus vectors derived from ent transfection assay. Thus, highly transfectable 293 SIVmac251, a nonpathogenic molecular clone of simian human embryo kidney cells were transfected with DNAs immunodeficiency virus that can efficiently replicate in of: (1) the tested SIV packaging plasmids (Figure 1a), (2) primate cells.40 We have examined the performance of the phCMV-G expression vector encoding VSV-G envel- several SIV vector packaging constructs that are equival- ope glycoproteins,45 and (3) the pSIV-T1 packaging-com- ent to the first,25 second41 and third42,43 generations of petent GFP-encoding SIV vector (Figure 1b). A fourth HIV-1-based vectors. The lentivirus vectors were pro- plasmid encoding SIVmac251 (pK-Rev) was or was duced following transient expression of plasmid DNAs not cotransfected in producer cells to allow expression of encoding the transfer vector and the packaging functions SIV core proteins by the pSIV8 packaging construct, that in 293 human cells. Both HIV-1-based and SIV derived does not express the rev gene, and, additionally, to lentiviral vectors were similar in their capacity to trans- determine if expression of viral particles by the other – duce proliferating target cells in side-to-side experiments. rev-gene expressing – packaging plasmids could be We also report here the characterization of SIV vectors optimized by providing further rev protein expression in pseudotyped with glycoproteins from various mem- the producer cells. For each component of the SIV vector brane-enveloped viruses that include VSV (vesicular production system, the optimal dose of plasmid DNA, stomatitis virus), MLV (murine leukemia virus), GALV allowing maximal transduction of target cells, was estab- (gibbon ape leukemia virus) and influenza virus. Finally, lished (data not shown). Transfection efficiencies higher we show here that monocyte-derived, nonproliferating, than 90% GFP-positive producer cells were routinely mature human dendritic cells could be efficiently trans- obtained and were, in fact, required to achieve the high- duced at low multiplicity of infection with the novel est formation of infectious SIV vector particles. Two days lentivirus vectors. after transfection, the supernatants of the transfected cells were harvested and used to infect proliferating sMAGI target cells. The SIV vectors were evaluated by both their Results transduction efficiencies, determined as percentages of GFP-positive target cells, and their infectious titers, Functional evaluation of SIV-derived packaging determined by end-point dilutions and followed by constructs FACS analysis of infected cells (Figure 1c). A series of expression plasmids encoding the viral core With the exception of the pSIV5 and pSIV6 plasmids, packaging functions of SIVmac251, an infectious mole- all the packaging plasmids were found to allow the cular clone of simian immunodeficiency virus (SIV), was efficient formation of SIV vector particles, resulting in the generated (Figure 1a). Our strategy was progressively to capacity to transduce between 50 and 100% of target cells, eliminate all the SIV sequences unnecessary for formation as determined 3 days after infection, and in infectious tit- of vector particles. Starting from a raw packaging con- ers of up to 2 × 1O7 TU/ml (Figure 1c). Transduction was struct (pSIV2 construct), from which only the gene inhibited by more than 95% when the target cells were and the 3ЈLTR were removed, we first aimed to generate pre-incubated with AZT, a inhibitor, tat-independent constructs in which all of the 5ЈLTR and thus ruling out passive protein or plasmid DNA transfer most of the leader sequences were removed and replaced (data not shown). Additionally, no differences in the per- by the puissant CMV early promoter/enhancer sequence centages of GFP-positive cells were noticed 2 months ver- (in pSIV3+ and all subsequent constructs). We then sus 3 days after transduction with the SIV vectors (data attempted to remove the vif, and accessory genes not shown). Altogether, these results indicated that GFP (in the pSIV4+ construct), and then the tat regulatory expression in target cells was achieved by stable trans- gene (in pSIV5 and following constructs). Finally, in gene delivery, mediated by the lentivirus replication and order to eliminate further SIV sequences, we questioned integration machinery. whether the rev/RRE combination could be replaced (in Only weak formation of SIV vectors could be obtained the pSIV6 construct) by a sequence of equivalent func- with the pSIV6 gag–pol expression plasmid, indicating tions derived from an avian , REV-A,44 or alter- that the REV-A CTE could not efficiently promote a natively, whether the SIV rev gene could be expressed Rev/RRE-independent expression of SIV genomic ectopically (in pSIV5, pSIV8 and pSIV10 constructs). mRNAs. Not surprisingly, pSIV8 construct mediated a

Gene Therapy Safe lentiviral vectors derived from SIVmac251 DNe`gre et al 1615

Figure 1 Design and evaluation of SIV packaging constructs. (a) Schematic representation of the pSIV2, pSIV3+, pSIV4+, pSIV5, pSIV6, pSIV8 and pSIV10 packaging plasmids derived from SIVmac251. The black boxes represent the SIV env gene and regulatory genes vif, vpx, vpr, tat, and , which have been progressively eliminated in the constructs. The dark-stripped boxes show the rev gene, the light-stripped boxes represent the rev- responsive element (RRE) or the REV-A R-U5-leader sequence (REV-A CTE). The deletions (⌬) introduced in the genome of SIVmac251 are indicated on the left. LTR, long terminal repeat; CMV, human cytomegalovirus early promoter; MSD, major splice donor site; ⌿, packaging sequence, SV40, simian virus 40 early promoter. (b) Schematic representation of the SIV-T1 and SIV-RMES4 GFP-carrying vectors propagated by the SIV packaging constructs. (c) Functional characterization of the SIV packaging constructs. sMAGI proliferating cells were transduced by the GFP-encoding SIV-T1 SIV vector complemented in trans with the indicated SIV packaging plasmids and VSV-G envelope glycoproteins. Production of SIV vectors was performed in the presence (black bars) or in the absence (stripped bars) of the pK-Rev expression plasmid. Mock, producer cells were transfected without any plasmid. The infectious titers (TU/ml) and the transduction efficiencies (% GFP-positive cells obtained after infection of 2.5 × 105 target cells using 1 ml of vector supernatants) are shown. poor production of SIV vector particles in the absence of Absence of RCRs in the SIV vector preparations rev expression. However, as expected, it was efficiently To detect replication-competent retroviruses (RCRs), two complemented in trans by the pK-Rev expression plas- assays were designed by using sMagi target cells, per- mid. Additionally, while a significant expression of viral missive to SIVmac251 replication and harboring a stably particles was obtained with pSIV5, a strong enhancement integrated lacZ transcription unit driven by the SIV/HIV of production of SIV vectors was achieved upon its tat-inducible HIV-1 LTR.46 Simian MAGI cells were prim- cotransfection with the pK-Rev plasmid, thus indicating arily infected with VSV-G-pseudotyped SIV vector par- that the rev gene carried by the pSIV5 plasmid was not ticles generated by using either the pSIV4+ or the pSIV8 efficiently expressed from the IRES sequence. In contrast packaging constructs and carrying either the SIV-T1 or to the pSIV5 and pSIV8 constructs, no significant differ- the SIV-RMES4 vectors. Infected sMagi cells were ences in viral production could be noticed using the checked for both GFP and ␤-galactosidase expression pSIV3+, pSIV4+ and pSIV10 packaging units, whether the (Table 1), which respectively indicate SIV vector trans- pK-Rev expression plasmid was or was not provided in duction efficiency and suggest the possible presence of a the producer cells (Figure 1c). Altogether, these data indi- tat-expressing recombinant retrovirus, and were grown cated that the pSIV3+, the pSIV4+, the pSIV10 and, when for up to 10 days to allow spreading of an eventual RCR. combined with the pK-Rev plasmid, the pSIV5 and pSIV8 The capacity of recombinant retroviruses to undergo rep- packaging constructs could efficiently produce SIV viral lication and amplification within this time-frame was core particles. The pSIV3+, pSIV4+ and pSIV8 packaging controlled using the reverse transcriptase (RT) activity constructs were therefore retained for further analysis. (data not shown) and the tat gene expression (Table 1) in

Gene Therapy Safe lentiviral vectors derived from SIVmac251 DNe`gre et al 1616 Table 1 Detection of replication-competent retroviruses

Helper/Vectora Primary Secondary infectionb infectionc

GFP ␤-Gal GFP ␤-Gal (%) (CFU/ml) (%) (CFU/ml)

Mock 0.28 Ͻ1 0.07 2 SIV4+/SIV-T1 96.84 6500 0.09 Ͻ1 SIV8/SIV-T1 95.40 2800 0.09 Ͻ1 SIV4+/SIV-RMES4 86.97 1 0.09 1 SIV8/SIV-RMES4 8.88 1 0.11 Ͻ1 SIVmac251d NA 7800 NA 9200

aSIV vector particles were generated with the indicated plasmids and pseudotyped with VSV-G glycoproteins. In parallel, to provide the background values of GFP and ␤-galactosidase expression in primary and secondary target cells, 293 producer cells were transfected with no DNA plasmids (Mock). bsMAGI target cells were infected with 1 ml of the SIV vector preparations, resulting in transduction efficiencies higher than 95% GFP-positive cells and in titers of up to 107 TU/ml, except when the SIV-RMES4 vector was propagated using the tat-negative SIV8 packaging construct. In parallel, sMAGI cells were X-gal stained after infection with dilutions of SIV vector preparations. The number of resulting ␤-galactosidase-CFU are indicated for infection with 1 ml of SIV vector supernatants. cOne ml of the supernatants of the primary infected cells, passaged for 10 days to allow amplification of an eventual RCR, were used to infect intact sMAGI cells. GFP and ␤-galactosidase expression was checked 3 days after infection. As a positive control for the vector mobilization assay, GFP vector-positive cells (primary cells) were transfected with both pSIV4+ and phCMV-G to provide SIV vector packaging functions. Secondary target cells, infected with the supernatant of these transfected cells, were transduced at an efficiency higher than 75% GFP-positive cells. dWild-type SIVmac251, grown on macaque PBLs, was used as positive control both for tat gene transduction and expression, as shown by ␤-galactosidase expression in primary infected sMAGI cells, and for RCR amplification. Secondary target cells were infected with 1 ml of the supernatant of the primary infected sMAGI cells, prealably inoculated with 1 ␮l of SIVmac251 and passaged for 10 days. Figure 2 Comparisons between SIV-based and HIV-1-based vectors. (a) NA, not applicable. Titration of lentiviral vectors on 293 (stripped bars) and or sMAGI (black bars) proliferating cells, as TU/ml. SIV and HIV-1 vectors were generated with VSV-G envelope glycoproteins, with the indicated GFP-encoding sMAGI cells infected by wild-type SIVmac251. The super- lentiviral vectors, and with the viral packaging functions encoded by the indicated plasmids. (b) Titration of cross-packaged lentiviral vectors on natants of the infected primary target cells were then 293 proliferating cells, as TU/ml. SIV and HIV-1 vectors were generated used to infect intact sMAGI cells, as secondary target with the indicated GFP-encoding lentiviral vectors, with VSV-G envelope cells, in order to assess for the putative mobilization of glycoproteins and with plasmids encoding SIV and HIV-1 core proteins: the GFP-containing SIV vector and the potential presence pSIV3+ and pCMV⌬8.2 plasmids (black bars) or pSIV4+ and of replication-competent retroviruses (Table 1). Detection pCMV⌬8.91 plasmids (stripped bars), expressing, or not, the vif, vpx (for of both GFP and ␤-galactosidase expression in the sec- SIV), vpr, and vpu (for HIV-1) accessory proteins, respectively. ondary target cells remained negative in all experiments performed, thus demonstrating the absence of mobil- T1 vector (Figure 1b) itself rather than via a parasite ization of the SIV vector and suggesting that the stocks packaging/transduction of the pSIV4+ helper genome, of SIV vectors were devoid of both RCRs and tat-recom- or, more unlikely, via an unexpected recombinant retro- binant retroviruses. Additionally, RT activity remained virus. To ascertain this point, similar experiments were negative in the supernatants of both primary and secon- performed by using the SIV-RMES4 minimal vector dary infected cells after several passages (data not (Figure 1b), which does not contain any SIV gene. shown), thus establishing the absence of RCRs in the SIV Although SIV vectors generated using the SIV-RMES4 vector preparations. vector and the pSIV4+ gag–pol construct could efficiently Of note, primary target cells infected with SIV-T1 vec- transduce the GFP marker into primary target cells, tors generated by using either pSIV4+ or pSIV8 packag- neither lacZ gene expression in primary target cells nor ing constructs were found positive for ␤-galactosidase GFP or ␤-galactosidase in secondary infected cells could expression (Table 1). Since the tat gene is absent from the be detected (Table 1). Identical results were obtained pSIV8 packaging construct, this result indicated that the when the SIV-RMES4 vector was propagated using the tat gene had been transduced most probably via the SIV- pSIV8 packaging construct although at lower titers (Table

Gene Therapy Safe lentiviral vectors derived from SIVmac251 DNe`gre et al 1617 1), thus demonstrating that pSIV4+ and pSIV8 were safe hemagglutinin. To minimize variations in expression, all packaging constructs when used to propagate minimal four glycoproteins types were produced using an SIV vectors. expression backbone identical to that used in previous experiments for VSV-G-pseudotype formation. Signifi- Comparison between HIV-1- and SIV-derived vectors cant differences were found in the capacity of these vari- The pSIV3+ and pSIV4+ viral core expression backbones ous glycoproteins to pseudotype SIV-vector particles, are equivalent in structure to the pCMV⌬R8.2 and although all SIV pseudotypes were found to allow trans- pCMV⌬R8.91 packaging plasmids, which express HIV-1 duction of 293 target cells (Figure 3a) and of other human gag and pol proteins, with or without lentiviral accessory cell types (data not shown). In comparison with VSV-G- proteins, respectively.41 These four SIV- or HIV-1-derived coated particles, HA-pseudotyped SIV vectors were packaging constructs were comparatively evaluated in about 200-fold less infectious, with titers of up to 2.5 × 104 side-to-side experiments for their capacity to generate TU/ml on 293 target cells. SIV vectors pseudotyped with SIV-based- or HIV-1-based vectors. In these experiments, 10A1-MLV and GALV envelope glycoproteins had simi- either the SIV-T1 and SIV-RMES4 SIV-based vectors or lar titers (between 1.3 × 105 and 2 × 105 TU/ml), but were the HRЈCMV-GFP HIV-1-based vector,35 all carrying respectively 20- and 30-fold less infectious on 293 target CMV promoter-driven GFP-expression cassettes, were cells than SIV particles coated with VSV-G glycoproteins used to transduce 293 human target cells as well as (Figure 3a). In contrast, when the same glycoproteins sMAGI simian target cells (Figure 2a). The SIV-T1 vector were used to coat MLV retroviral vectors, much weaker was more efficient than the SIV-RMES4 minimal vector variations were found in the infectious titers of the differ- and gave rise to approximately 10-fold higher titers. Tit- ent MLV pseudotypes (Figure 3b). These data indicated ers of the HRЈCMV-GFP HIV-1-based vector, of approxi- that VSV-G was the best glycoprotein for preparing mately 6 × 106 TU/ml, were intermediate between those pseudotypes of SIV vectors, although the other types of of the SIV-T1 and SIV-RMES4 SIV vectors on 293 target glycoproteins might be valuable for transduction of spe- cells. However, a significant discrepancy between SIV- cific target cell types in vivo. Indeed, while the transduc- derived and HIV-1-based vectors was found when tion efficiency of the VSV-G pseudotypes was higher than sMAGI simian versus 293 human target cells were com- 95%, thanks to their high titers allowing the use of high pared in the titration assays. In comparison with the MOIs (more than 10 infectious particles per cell), up to HRЈCMV-GFP HIV-1-vectors, titers of the SIV-T1 and 67% and 43% of GFP-positive cells could be obtained SIV-RMES4 SIV-vectors were five- to 20-fold better on the with the 101A-MLV and GALV SIV-vector pseudotypes, simian target cells than on the human target cells, and respectively, using much lower MOIs, of less than one vice versa (Figure 2a), most probably reflecting the adap- infectious particle per cell (Figure 3). tation of either lentiviruses with cells derived from their natural hosts. Transduction of human dendritic cells We next attempted to cross-package SIV-based or HIV- The SIV vectors were then evaluated for their ability to 1-based vectors with core particles derived from HIV-1 or transduce nonproliferative human dendritic cells (DCs), SIV, respectively (Figure 2b). HIV-1 core particles could obtained by IL4/GM-CSF-mediated differentiation of efficiently cross-package the SIV-T1 and SIV-RMES4 SIV- blood monocytes. DCs, as assessed by immunostaining derived vectors and allowed transduction of target cells using monoclonal antibodies against CD1a, CD80, CD86 with an efficiency decreased by four-fold, at most, com- and MHC class II molecules (Figure 4a), were transduced paratively to that of the same SIV-based vectors packaged by using the SIV-T1 vector and either the pSIV3+ or with homologous SIV core particles. In contrast, the infec- tious titers of the HRЈCMV-GFP HIV-1-based vector were strongly reduced, by up to 1000-fold, when they were generated with SIV core particles (Figure 2b). These results were consistent with other studies addressing the capacity of the SIV genome to be packaged in HIV-1 viral particles47 and the nonreciprocal packaging of HIV-2 gen- ome, which is closely related to that of SIV, into HIV-1 core particles.48 Since previous studies have shown that the reduction of sequence homology between vector and packaging components is an important parameter in decreasing the risk of the emergence of recombinant retroviruses,1 that SIV-based vectors can be efficiently generated with HIV-1 core particles may therefore provide an additional safety feature to lentiviral vectors.

Pseudotyping of SIV vectors SIV vectors characterized here so far have been generated as VSV-G-coated retroviral particles. Since the transduc- tion efficiency of target cells is dependent on the type of Figure 3 Infectivity of SIV- and MLV-vector pseudotypes. (a) SIV vectors + glycoprotein used to coat retroviral vectors,49–50 we next were generated with the SIV-T1 GFP-encoding vector, the pSIV4 determined the infectivity of SIV vector particles packaging plasmid and the indicated glycoproteins. Titers determined on 293 target cells, as TU/ml. (b) MLV vectors were generated with the generated with envelope glycoproteins derived from indicated glycoproteins. Titers determined on 293 target cells, as TU/ml. 10A1-MLV (murine leukemia virus), GALV (gibbon ape On the top of the bars, the transduction efficiencies obtained after infection leukemia virus) and with an influenza virus H7-HA of 2.5 × 105 target cells using 1 ml of vector supernatant are shown.

Gene Therapy Safe lentiviral vectors derived from SIVmac251 DNe`gre et al 1618

Figure 4 Transduction of human dendritic cells. (a) Immunophenotyping of monocyte-derived DCs. The proportion of viable cells positive for expression of the indicated cell surface antigens is shown. The histograms of cells stained with control isotype-matched antibodies are shown as dotted lines. (b) Results of transduction of proliferating 293 cells or nonproliferating immature or mature human dendritic cells (DCs), using VSV-G pseudotypes of GFP-encoding MLV-vectors (empty bars) or SIV-T1 SIV-vectors generated with the pSIV3+ or the pSIV4+ packaging constructs, expressing (grey bars), or not (stripped bars), the vif, vpx and vpr accessory proteins, respectively. Target cells were transduced using an MOI of one infectious viral particle per cells, as determined by titration assays on proliferating 293 target cells, using end-point dilutions of the vector stocks. Mock, infection of target cells using supernatants of producer cells transfected without any plasmid. (c) Immunophenotyping of GFP-transduced mature DCs. SIV vector- transduced cells (lower dot-plot) or mock-transduced cells (upper dot-plot), grown in the presence of CD40 ligand after infection, were stained with PE- conjugated Mab against MHC class II molecules and analyzed for both MHC class II (indicative of mature DCs) and GFP expression. (d) Optimization of a gene transfer protocol into human DCs. Immature DCs were transduced with either the SIV-T1 vector or the SIV-RMES4 minimal vector, both generated using the pSIV3+ packaging plasmid, and, as control, an MLV vector. MOIs of 10 infectious particles per target cell were applied. SIV vector- infected cells were or were not maturated by co-cultivation with human CD40L-expressing mouse fibroblasts for 3 and 5 days before determining the transduction efficacy. GFP transduction in the mature DCs was inhibited by up to 80% when infection was carried out in the presence of 10 ␮m AZT (data not shown).

pSIV4+ packaging constructs which express, or not using the same MOIs, the preparations of SIV-based express, most of the SIV accessory genes (except nef), vectors could transduce the human DCs, although with respectively. Retroviral vectors derived from MLV variable efficiencies (Figure 4b). While immature DCs (murine leukemia virus) were used as control. Both lenti- could be transduced with an efficiency of up to 8.5% viral and retroviral vectors, generated as VSV-G-pseudo- GFP-positive cells, GFP expression was detected in up typed particles, had similar titers on proliferating cells to 18% of the mature DCs, most probably owing to the (around 5 × 106 TU/ml) and were used to infect human increased metabolism of these activated cells. Treatment DCs with a low MOI, of one infectious particle per target of the target cells with AZT during and after infection cell (Figure 4b). Infected immature DCs were then co-cul- was found to prevent GFP transduction by more than tured either with irradiated mouse fibroblasts expressing 90% when monitored 3 days after transduction (data not the human CD40 ligand, to allow their differentiation shown), thus ruling out the possibility that GFP into mature DCs,51 or with irradiated mouse fibroblasts expression detected in cells transduced in the absence of expressing the human CD32, as control. GFP expression AZT was due to pseudo-transduction or phagocytosis of was measured 3 days after transduction and the mature plasmid DNA (data not shown). Maturation of DCs, a DC phenotype of the GFP-positive cells was ascertained pre-requisite in fulfilling their role of antigen-presenting by immunostaining of MHC class II molecules and FACS cells and activation of the immune system, is charac- analysis (Figure 4b and c). No GFP-positive cells were terized by up-regulation of MHC class II molecules. Since found in DCs infected with MLV-based vectors (Figure the same proportion of cells with high MHC class II 4b), owing to the incapacity of the latter type of vectors expression – approximately 90% – was detected in both to integrate into non proliferating cells.13 In contrast, the nontransduced DCs and the GFP-transduced DCs

Gene Therapy Safe lentiviral vectors derived from SIVmac251 DNe`gre et al 1619 (Figure 4c), our results indicated that SIV vector- than the transient production method, they represent an mediated gene transfer did not impair the maturation important step toward the large-scale production of lenti- and properties of DCs. viral vectors for clinical application. Of note, we have In comparison with SIV vectors generated with pSIV4+ recently described hybrid vectors derived from powerful helper plasmid, SIV vectors generated with the pSIV3+ viral vectors, such as herpes virus amplicons or recombi- packaging construct could transduce the DCs better nant adenoviruses, which were engineered to induce the (Figure 4b). Since no differences were found between the safe formation of high titer MLV-based vectors.57,58 We two SIV vector preparations for infection of proliferating are currently investigating the properties of lentiviral cells, the reproducible enhancement of transduction of vectors generated by using such hybrid vectors. DCs by SIV vectors generated with the pSIV3+ packaging Among the important issues raised by the utilization of construct probably resulted from the expression of viral vectors in gene therapy is the absence of replication- accessory proteins, such as vpx/vpr, which have been competent viruses, especially when the former are found to facilitate nuclear import of the lentiviral pre- derived from pathogenic viruses such as primate lenti- integration complex in some terminally differentiated viruses. In practice for lentiviral vectors, both the cell types.22,52 removal of the env, tat, and accessory genes and the To improve gene transfer and expression into DCs, we repartition of the transfer and packaging functions on sought to increase both the MOI and the duration of DC separate units with minimal sequence overlaps seems to maturation with the CD40 ligand. Using the optimal con- satisfy biosafety criteria and to prevent the emergence of ditions of 10 infectious particles per cell and 5 days of replication-competent retroviruses, as demonstrated here co-cultivation of the transduced DCs with CD40L- for the SIV-based vectors (Table 1). Furthermore, the expressing mouse fibroblasts, transduction efficiencies of obtention of self-inactivating vectors either for HIV-1- up to 33% GFP-positive cells could be obtained (Figure derived vectors59,60 or for our SIV-based vectors65 should 4d). In addition, at the same MOIs, no significant differ- add an additional safety level, by preventing vector mob- ence in transduction efficiency was found when the SIV- ilization upon infection with an exogenous lentivirus,61 T1 vector was compared with the minimal SIV-RMES4 for example. Moreover, our results show that either HIV- vector for infection of DCs (Figure 4d). 1-based or SIV-derived vectors can be cross-packaged in Altogether, these results indicated that SIV vectors core particles of the converse lentivirus, although only could efficiently transduce nonproliferating primary the propagation of SIV vectors into HIV-1 core particles human dendritic cells. Efficient gene transfer could be gave rise to high infectious titers (Figure 2b). The low obtained by using a minimal SIV vector and transduced nucleotide sequence homology between HIV-1 and DCs were found to retain their biological properties, such SIVmac251, about 33% for their non-coding regions, as MHC class II molecule up-regulation in response to should therefore further reduce the likelihood of recom- CD40L-induced maturation. Moreover, in addition to bination for such hybrid SIV/HIV-1 vector systems. providing an optimized protocol for ex vivo transduction Despite all the biosafety safeguards that have progress- of human DCs, we also report here that our SIV vectors ively been added, the use of lentiviral vectors derived could achieve an efficient gene transfer at a low MOI, a from pathogenic primate lentiviruses raises serious con- critical requirement for approaching in vivo gene cerns on their clinical acceptance. At present, it is not delivery. clear which type of lentivirus vector, derived from either primate or nonprimate lentiviruses, should be better Discussion accepted. The biosafety of either vector type should be proven in vivo in animal models for HIV-induced disease We report here the generation of novel lentivirus vectors, before their use in humans. However, since HIV-1 causes derived from SIVmac251. Our vectors could efficiently AIDS only in humans, there is no suitable animal model transduce nonproliferating cells, as established using in which the safety of HIV vectors could be evaluated, terminally differentiated human dendritic cells and side- thus providing a serious limitation for the pre-clinical to-side comparisons with HIV-1-based vectors. In a testing of HIV-based vectors. Since SIVmac251 is a manner similar to vectors derived from other types of causative agent of simian AIDS, the biosafety of lentiviral lentiviruses, SIV-derived vectors were produced in 293 vectors may therefore be tested in a monkey model. Thus cells by transient expression after transfection of plasmid the SIV vectors may provide a valid alternative to HIV- DNAs encoding the different components required for 1-based vectors, at least in the early phases of the clinical assembly of SIV vector particles. Although widely used testing of lentivirus vectors. for producing lentivirus vectors, the use of such methods Lentiviral vectors are promising tools for in vivo gene might prove difficult to scale-up vector production and delivery since, unlike MLV-based vectors, they may to pass quality control and validation processes, due to stably transduce quiescent cells in various organs.25,35–39 the various types of contaminants that may be found in To be of real benefit for in vivo administration protocols, vector stocks obtained by transient expression. Stable such vectors should be infectious at low MOIs, since only packaging cell lines have been produced for HIV-1-based MOIs of far less than one would be applicable in vivo. vectors, yet they were found to release low titers of infec- Here, we have shown that SIV-based vectors could tious particles,53–55 owing to the properties of proteins efficiently transduce nonproliferating monocyte-derived encoded by some lentiviral genes which may exert a mature human DCs, at an MOI of one infectious particle negative effect on the viability of the producer cells. per target cell, thus establishing the value of these novel Alternative systems for producing lentivirus vectors have lentiviral vectors. Moreover, GFP-transduced DCs could been obtained by generating packaging cell lines whose be differentiated in vitro into mature DCs, suggesting that expression can be induced.37,42,56 Although these they had retained their biological properties. We are cur- inducible packaging cell lines seem to be less efficient rently investigating whether DCs transduced in vitro with

Gene Therapy Safe lentiviral vectors derived from SIVmac251 DNe`gre et al 1620 various antigen-encoding SIV vectors can activate cyto- entry site) and a SIV rev cDNA, in which the two rev gene toxic T cells and lead to effective CTL vaccines. We also exons (nts 6504 to 6573 and nts 8785 to 9041) were fused, describe here a protocol that allows efficient gene transfer was inserted after the SIV RRE sequence of pSIV8, thus into monocyte-derived human DCs, resulting in trans- resulting in the pSIV5 expression plasmid. In pSIV6, duction of up to 33% target cells. derived from pSIV8, the SIV RRE sequence was replaced Additionally, we show here that SIV vectors can be by the REV-A (reticuloendotheliosis virus type A) R-U5- efficiently pseudotyped with surface glycoproteins leader region which may contain a constitutive transport derived from a wide range of membrane-enveloped element (CTE), as suggested by a recent study on SNV viruses, VSV, MLV, GALV and influenza virus. This (spleen necrosis virus), an avian retrovirus that shares finding opens the way to transducing cell types that may extensive sequence homology with REV-A.44 To express be refractory to infection with VSV-G pseudotypes. the rev gene from a transcription unit different from that Besides, VSV-G-coated vector particles may not be of the gag–pol genes, the SIVmac251 rev cDNA was suitable for an in vivo gene delivery by systemic inserted into the pSI mammalian expression plasmid administration, since, due to the wide distribution of its (Promega, Charbonnie`res, France) and placed under con- receptor – a lipid component of the plasma membrane – trol of the SV40 early promoter, chimeric ␤-globin gene they may stick on the surface of all cells encountered after intron and SV40 sequences, resulting in inoculation before they have the chance of reaching the the pK-Rev plasmid. The SV40-driven expression unit of gene therapy target cells. Additionally, since MLV envel- the rev gene was excised from pK-Rev and inserted ope glycoproteins and influenza virus hemagglutinin downstream into the polyadenylation sequence of pSIV8, could efficiently pseudotype the SIV vectors (Figure 3a), thus resulting in the pSIV10 packaging plasmid. it will be possible to exploit the wide range of receptor- targeted glycoproteins that have been engineered using Other plasmids these two proteins types,8 in order to modify the host The pSIV-T1 and pSIV-RMES4 (Figure 1b) are packaging- range of lentivirus vectors. competent SIVmac251-based vectors.65 Both SIV vectors contain the enhanced green fluorescent protein (GFP) marker gene under control of the CMV promoter placed Materials and methods between the SIVmac251 LTRs and leader sequences, and carry the SIVmac251 rev-responsive element (RRE) and Construction of SIV gag/pol expression plasmids minimal sequences of the gag and pol genes (the first 218 The BK28 molecular clone of SIVmac25140 was used to or 57 nts of the gag gene and the last 1007 nts or a 141 bps derive the SIV packaging constructs. The nucleotide pol gene-derived fragment encompassing the cPPT/CTS sequence of SIVmac251 has the GenBank accession num- regions (nts 5048 to 5189), for the pSIV-T1 or the pSIV- ber M19499. Starting from the SIVmac251 provirus (a RMES4 vectors, respectively). While the pSIV-T1 vector kind gift of Jim Mullins, Department of Microbiology, still contains residual sequences derived from some of the University of Washington, Seattle, WA, USA), a methodi- SIVmac251 accessory genes (Figure 1b), of which only the cal approach was used to engineer both efficient and safe tat gene has retained its coding capacity, pSIV-RMES4 is expression plasmids allowing the production of SIV viral a minimal vector and retains less than 15% nucleotides core proteins (Figure 1a). In the pSIV2 packaging plas- derived from the parental SIVmac251 RNA genome.65 mid, the U3 region of the 5Ј LTR of SIVmac251 (nts 1 to The pCMV⌬8.225 or pCMV⌬8.9141 HIV-1 packaging 506) was replaced by the human cytomegalovirus early– plasmids and the pHRЈCMV-GFP HIV-1-based vector35 immediate promoter and enhancer region (CMV were described elsewhere and were kindly provided by promoter). In addition, the 5Ј half of the env gene (nts Didier Trono (Department of Genetics and Microbiology, 6582 to 7981) was removed, leaving intact the RRE (Rev- University of Geneva, Switzerland). responsive element) sequence as well as the 5Ј and 3Ј The pTG5349 murine leukemia virus (MLV) packaging exons of the tat and rev regulatory genes. Furthermore, plasmid and the pTG13077 plasmid, encoding an MLV- the 3Ј LTR (nts 9444 to 10249) was replaced by an SV40 based vector containing a CMV-GFP internal transcrip- polyadenylation sequence, thus resulting in deletion of tional unit, were kindly provided by Transgene the 3Ј end of the nef gene. Finally, the initiation codon of (Strasbourg, France). nef was inactivated to prevent translation. pSIV3+ was The following plasmids, phCMV-G,45 phCMV-HA,62 generated by removing from the pSIV2 construct a phCMV-10A163 and phCMV-GALV (S Chapel-Fernandes sequence (nts 507 to 959) encompassing R, U5 and the 5Ј and F-LC, unpublished material) are envelope-expression half of the leader region (up to the major splice donor constructs that respectively encode the vesicular site (MSD), nts 967 to 980). In pSIV4+, a construct similar stomatitis virus (VSV) G protein, an influenza virus H7- in structure to pSIV3+, the genes encoding the vif, vpx HA hemagglutinin, the 10A1-MLV and the gibbon ape and vpr SIV accessory proteins were inactivated by par- leukemia virus (GALV) envelope glycoproteins, all under tial deletion (from nts 5388 to 6249) and inactivation of control of the CMV promoter, rabbit ␤-globin intron II the vif gene initiation codon. Further CMV promoter- and polyadenylation sequences. driven SIV packaging constructs were generated by PCR to remove most of the residual sequences of the accessory Cells genes. First, the 5Ј part of the pSIV4+ construct (from the The 293 human embryo kidney cell line (ATCC CRL- CMV promoter to the end of the pol gene – nt 5396) was 1573) and the CD4-positive sMAGI macaque cell line,46 fused to the SIV RRE sequence (nts 8021 to 8527), thus which expresses a lacZ gene under control of the resulting in the pSIV8 expression construct. Second, a SIV/HIV-tat-responsive HIV-1 LTR, were grown in cassette encompassing the sequence of EMCV DMEM (Life Technologies, Cergy Pontoise, France) sup- (encephalo-myocarditis virus) IRES (internal ribosomal plemented with 10% fetal bovine serum (FCS).

Gene Therapy Safe lentiviral vectors derived from SIVmac251 DNe`gre et al 1621 Human dendritic cells (DCs) were prepared from of 2.5 × 105 per well in six-well plates 1 day before trans- blood monocytes from Ficoll/Percoll-purified human duction. Serial dilutions of vector preparations were peripheral blood mononuclear cells after immunodeple- added to the cells in the presence of 6 ␮gof tion of T and B lymphocytes and NK cells, as previously polybrene/ml, and the cultures were incubated for 4 h at described.64 After 6 days of culture in the presence of 50 37°C. The vector-containing medium was then replaced ng/ml hrGM-CSF and 500 U/ml hrIL-4 (kindly provided with normal culture medium and the cells were incu- by the Schering-Plough Laboratory for Immunological bated for 72 h at 37°C. The percentage of GFP-positive Research, Dardilly, France), more than 95% of the cells cells was determined by FACS analysis, following were DCs as assessed by CD1a labeling. Cultures of the individualization of the transduced cells in trypsine and DCs were performed in 24-well flat-bottomed microtiter fixing of the cells in 4% formaldehyde in PBS. The plates (Falcon/Becton Dickinson, Le Pont de Claix, transduction efficiency was arbitrarily determined as the France), in a total volume of 1 ml, in RPMI 1640 (Life percentage of GFP-positive cells after transduction of Technologies) supplemented with 10 mm Hepes (Life 2.5 × 105 target cells with 1 ml of viral supernatant. The Technologies), 2 mml-glutamine (Life Tech- nologies), infectious titers, provided as transducing units (TU)/ml, 40 ␮g/ml gentamicin (Life Technologies), and 10% FCS were calculated by using the formula: Titer = %inf × (Boehringer Mannheim, Meylan, France). For immuno- (2.5 × 105/100) × d, where d is the dilution factor of the phenotyping, DCs or GFP vector-transduced DCs were viral supernatant and %inf is the percentage of GFP-posi- washed in ice-cold phosphate buffer saline (PBS) sup- tive cells as determined by FACS analysis using dilutions plemented with 1% bovine serum albumin (BSA) and 3% of the viral supernatant that transduce between 5% and human serum in PBS. DCs were stained using 2 ␮g/ml 10% of GFP-positive cells. ␤-Galactosidase expression in phycoerythrin (PE)-conjugated monoclonal antibodies vector-transduced sMAGI cells was assayed 72 h after (mAb) specific for CD1a (BL6; DAKO, Glostrup, infection. Cells were fixed and stained with X-gal Denmark), CD80 (L307.4; Becton Dickinson Immunocyto- (5-bromo-4-chloro-3-indolyl-␤-d-galactopyranoside) as chemistry Systems, San Jose, CA, USA), CD86 (BU63; Ser- previously described.1 ␤-Galactosidase-positive colonies otec Ltd, Oxford, UK) and MHC class II molecules (L243; were counted and the infectious titers of the tat-express- Becton Dickinson Immunocytochemistry Systems). DCs ing retroviruses were expressed as CFU/ml. were also stained with corresponding isotype-matched To infect the human DCs, 1 ml of diluted viral super- control mAb (Immunotech, Marseilles, France). Surface natants containing 105 TU of infectious particles, as expression was analyzed using a FACSCalibur flow cyto- determined beforehand on 293 target cells, was added to meter (Becton Dickinson, Pont-de-Claix, France). Viable 105 DCs suspended in 0.1 ml of DC medium. DCs were DCs were gated according to negative staining with pro- infected in the presence of 6 ␮g of polybrene/ml for 4 h pidium iodide. Data were collected from 5000 to 10 000 at 37°C, then washed and incubated for 48 h in normal viable cells and analyzed using CellQuest software DC medium in the presence of 2 × 104 irradiated (7000 (Becton Dickinson Immunocytochemistry Systems). rads) fibroblastic CD40 ligand- or CD32-transfected L mouse cells (both kindly provided by Schering-Plough Production of lentiviral and retroviral vectors Laboratory for Immunological Research) to allow, or not The VSV-G-pseudotyped HIV-1-derived vectors were allow, differentiation into mature DCs, respectively.51 To generated as previously described25 by transient transfec- rule out pseudo-transduction by free GFP or by GFP-con- tion in 293 cells of the pHRЈCMV-GFP vector construct taining pseudo-particles or by DNA plasmids released by (8.1 ␮g) with the VSV-G-expressing construct phCMV-G the producer cells and ingested by the DCs, experiments (2.7 ␮g) and either the pCMV⌬8.2 or pCMV⌬8.91 were performed in duplicate in the absence or the pres- packaging constructs (8.1 ␮g). MLV-derived vectors ence of 10 ␮m AZT (3Ј-azido-3Ј-deoxythymidine; Sigma- pseudotyped with VSV-G or other viral surface Aldrich, Saint Quentin Fallavier, France), an inhibitor of glycoproteins were generated in a similar manner by reverse transcriptase, which was added 60 min before transient transfection of the pTG5349 MLV packaging infection until transduction analysis. Transduction construct (13.2 ␮g) with pTG13077 MLV vector construct efficiency was then assessed by FACS analysis 2 and 5 (13.2 ␮g) and the envelope-expressing construct (3.1 ␮g). days after infection. SIV-derived vectors were generated by transfecting the SIV packaging constructs (8.1 ␮g) with either the pSIV- T1 or the pSIV-RMES4 vector constructs (8.1 ␮g) and the Acknowledgements envelope-expressing construct (2.7 ␮g) in the presence of 10.8 ␮g of either a carrier DNA (pUC19) or the pK-Rev We thank J Mullins for the SIVmac251 molecular clone, expression plasmid. Didier Trono and Romain Zufferey for the HIV-1 vector Plasmid DNAs were transfected into 2.4 × 106 293 cells and helper plasmids, Transgene SA for the MLV vector seeded the day before in 10-cm plates using the calcium- and helper plasmids, and the Schering-Plough Labora- phosphate transfection system (Life Technologies, tory for Immunological Research for the CD40 ligand- France) according to the manufacturer’s recommen- expressing cells. We thank Bertrand Boson for helpful dations. The medium (12 ml per plate) was replaced 15 h discussions and Roger Legrand for providing the after transfection, and supernatant was harvested 36 h SIVmac251 viral stock. This work was supported by Insti- after the start of the transfection, filtered through 0.45- tut National de la Sante´ et de la Recherche Me´dicale ␮m pore-sized membranes, and stored at −80°C. (INSERM), Agence Nationale pour la Recherche contre le SIDA (ANRS), Association Franc¸aise contre les Myo- Transduction assays pathies (AFM), Association pour la Recherche contre le To determine the transduction efficiency and infectious (ARC), Mutuelle Ge´ne´rale de l’Enseignement titer, 293 or sMAGI target cells were seeded at a density National (MGEN), and the European Community (EC).

Gene Therapy Safe lentiviral vectors derived from SIVmac251 DNe`gre et al 1622 References 23 Bukrinsky MI, Haffar OK. HIV-1 nuclear import: in search of a leader. Front Biosci 1999; 4: D772–781. 24 Deminie CA, Emerman M. Functional exchange of an oncoretro- 1 Cosset F-L et al. High titer packaging cells producing recombi- virus and a lentivirus matrix protein. J Virol 1994; 68: 4442–4449. nant retroviruses resistant to human serum. J Virol 1995; 69: 25 Naldini L et al. In vivo gene delivery and stable transduction of 7430–7436. nondividing cells by a lentiviral vector. Science 1996; 272: 263– 2 DePolo NJ et al. The resistance of retroviral vectors produced from human cells to serum inactivation in vivo and in vitro is 267. primate species dependent. J Virol 1999; 73: 6708–6714. 26 Reiser J et al. Transduction of nondividing cells using pseudo- 3 Takeuchi Y et al. Type C retrovirus inactivation by human typed defective high-titer HIV type 1 particles. Proc Natl Acad complement is determined by both the viral genome and pro- Sci USA 1996; 93: 15266–15271. ducer cell. J Virol 1994; 68: 8001–8007. 27 Arya SK, Zamani M, Kundra P. Human immunodeficiency virus 4 Kotani H et al. Improved methods of retroviral vector transduc- type 2 lentivirus vectors for gene transfer: expression and poten- tion and production for gene therapy. Hum Gene Ther 1994; 5: tial for helper virus-free packaging. Hum Gene Ther 1998; 9: 19–28. 1371–1380. 5 Smith KT, Shepherd AJ, Boyd JE, Lees GM. Gene delivery sys- 28 Schnell T et al. Development of a self-inactivating, minimal lenti- tems for use in gene therapy: an overview of quality assurance virus vector based on simian immunodeficiency virus. Hum and safety issues. Gene Therapy 1996; 3: 190–200. Gene Ther 2000; 11: 439–447. 6 Diaz RM, Eisen T, Hart IR, Vile RG. Exchange of viral 29 White SM et al. Lentivirus vectors using human and simian promoter/enhancer elements with heterologous regulatory immunodeficiency virus elements. J Virol 1999; 73: 2832–2840. sequences generates targeted hybrid long terminal repeat vec- 30 Poeschla E, Wong-Staal F, Looney D. Efficient transduction of tors for gene therapy of melanoma. J Virol 1998; 72: 789–795. nondividing human cells by feline immunodeficiency virus lent- 7 Jager U, Zhao Y, Porter CD. Endothelial cell-specific transcrip- iviral vectors. Nature Med 1998; 4: 354–357. tional targeting from a hybrid long terminal repeat retrovirus 31 Mitrophanous K et al. Stable gene transfer to the nervous system vector containing human prepro-endothelin-1 promoter using a non-primate lentiviral vector. Gene Therapy 1999; 6: sequences. J Virol 1999; 73: 9702–9709. 1808–1818. 8 Russell SJ, Cosset F-L. Modifying the host range properties of 32 Olsen JC. Gene transfer vectors derived from equine infectious retroviral vectors. J Gene Med 1999; 1: 300–311. anemia virus. Gene Therapy 1998; 5: 1481–1487. 9 Naldini L. Lentiviruses as gene transfer agents for delivery to 33 Gallichan WS et al. Lentivirus-mediated transduction of islet non-dividing cells. Curr Opin Biotechnol 1998; 9: 457–463. grafts with interleukin 4 results in sustained gene expression 10 Trono D. Lentiviral vectors: turning a deadly foe into a thera- and protection from insulitis. Hum Gene Ther 1998; 9: 2717–2726. peutic agent. Gene Therapy 2000; 7: 20–23. 34 Miyoshi H et al. Transduction of human CD34+ cells that 11 Smiley WR et al. Establishment of parameters for optimal trans- mediate long-term engraftment of NOD/SCID mice by HIV vec- duction efficiency and antitumor effects with purified high-titer tors. Science 1999; 283: 682–686. HSV-TK retroviral vector in established solid tumors. Hum Gene 35 Blo¨mer U et al. Highly efficient and sustained gene transfer in Ther 1997; 8: 965–977. adult neurons with a lentiviral vector. J Virol 1997; 71: 6641– 12 Kitten O, Cosset F-L, Ferry N. Highly efficient retroviral- 6649. mediated gene transfer into hepatocytes in vivo. Hum Gene Ther 36 Johnson LG, Olsen JC, Naldini L, Boucher RC. Pseudotyped 1997; 8: 1491–1494. human lentiviral vector-mediated gene transfer to airway epi- 13 Miller DG, Adam MA, Miller AD. Gene transfer by retrovirus thelia in vivo. Gene Therapy 2000; 7: 568–574. vectors occurs only in cells that are actively replicating at the 37 Kafri T et al. Sustained expression of genes delivered directly time of infection. Mol Cell Biol 1990; 10: 4239–4242. into liver and muscle by lentiviral vectors. Nat Genet 1997; 17: 14 Roe T, Reynolds TC, Yu G, Brown PO. Integration of murine 314–317. leukemia virus DNA depends on mitosis. EMBO J 1993; 12: 38 Miyoshi H, Takahashi M, Gage FH, Verma IM. Stable and 2099–2108. efficient gene transfer into the retina using an HIV-based lenti- 15 Uchida N et al. HIV, but not murine leukemia virus, vectors . Proc Natl Acad Sci USA 1997; 94: 10319–10323. mediate high efficiency gene transfer into freshly isolated 39 Naldini L et al. Efficient transfer, integration, and sustained G0/G1 human hematopoietic stem cells. Proc Natl Acad Sci USA long-term expression of the transgene in adult rat brains injected 1998; 95: 11939–11944. with a lentiviral vector. Proc Natl Acad Sci USA 1996; 93: 16 Lewis PF, Emerman M. Passage through mitosis is required for 11382–11388. oncoretroviruses but not for the human immunodeficiency 40 Kornfeld H et al. Cloning of HTLV-4 and its relation to simian virus. J Virol 1994; 68: 510–516. and human immunodeficiency viruses. Nature 1987; 326: 610– 17 Ho DD, Rota TR, Hirsch MS. Infection of monocyte/ 613. macrophages by human T lymphotropic virus type III. J Clin 41 Zufferey R et al. Multiply attenuated lentiviral vector achieves Invest 1986; 77: 1712–1715. efficient gene delivery in vivo. Nat Biotechnol 1997; 15: 871–875. 18 Lewis P, Hensel M, Emerman M. Human immunodeficiency 42 Dull T et al. A third-generation lentivirus vector with a con- virus infection of cells arrested in the cell cycle. EMBO J 1992; ditional packaging system. J Virol 1998; 72: 8463–8471. 11: 3053–3058. 43 Kim VN, Mitrophanous K, Kingsman SM, Kingsman AJ. Mini- 19 Bukrinsky MI et al. Active nuclear import of human immuno- mal requirement for a lentivirus vector based on human deficiency virus type 1 preintegration complexes. Proc Natl Acad immunodeficiency virus type 1. J Virol 1998; 72: 811–816. Sci USA 1992; 89: 6580–6584. 44 Butsch M et al. The 5Ј RNA terminus of spleen necrosis virus 20 Bukrinsky MI et al. A nuclear localization signal within HIV-1 contains a novel posttranscriptional control element that facili- matrix protein that governs infection of non-dividing cells. Nat- tates human immunodeficiency virus Rev/RRE-independent ure 1993; 365: 666–669. Gag production. J Virol 1999; 73: 4847–4855. 21 Gallay P, Hope T, Chin D, Trono D. HIV-1 infection of nondivid- 45 Yee JK, Friedmann T, Burns JC. Generation of high-titer pseudo- ing cells through the recognition of by the typed retroviral vectors with very broad host range. Meth Cell importin/karyopherin pathway. Proc Natl Acad Sci USA 1997; Biol 1994; 43: 99–112. 94: 9825–9830. 46 Chackerian B, Haigwood NL, Overbaugh J. Characterization of 22 Heinzinger NK et al. The Vpr protein of human immunodefi- a CD4-expressing macaque cell line that can detect virus after ciency virus type 1 influences nuclear localization of viral a single replication cycle and can be infected by diverse simian nucleic acids in nondividing host cells. Proc Natl Acad Sci USA immunodeficiency virus isolates. Virology 1995; 213: 386–394. 1994; 91: 7311–7315. 47 Rizvi TA, Panganiban AT. Simian immunodeficiency virus RNA

Gene Therapy Safe lentiviral vectors derived from SIVmac251 DNe`gre et al 1623 is efficiently encapsidated by human immunodeficiency virus 56 Kaul M, Yu H, Ron Y, Dougherty JP. Regulated lentiviral pack- type 1 particles. J Virol 1993; 67: 2681–2688. aging cell line devoid of most viral cis-acting sequences. Virology 48 Kaye JF, Lever AM. Nonreciprocal packaging of human 1998; 249: 167–174. immunodeficiency virus type 1 and type 2 RNA: a possible role 57 Duisit G, Salvetti A, Moullier P, Cosset F-L. Functional charac- for the p2 domain of Gag in RNA encapsidation. J Virol 1998; terization of adenoviral/retroviral chimeric vectors and their 72: 5877–5885. use for efficient screening of retroviral producer cell lines. Hum 49 Marandin A et al. Retroviral-mediated gene transfer into human Gene Ther 1999; 10: 189–200. CD34+/38− primitive cells capable of reconstituting long-term 58 Savard N, Cosset F-L, Epstein AL. Use of defective HSV-1 vec- cultures in vitro and in nonobese diabetic-severe combined tors harbouring gag, pol, and env genes to rescue defective immunodeficiency mice in vivo. Hum Gene Ther 1998; 9: 1497– retrovirus vectors. J Virol 1997; 71: 4111–4117. 1511. 59 Miyoshi H et al. Development of a self-inactivating lentivirus 50 Movassagh M et al. High-level gene transfer to cord blood pro- vector. J Virol 1998; 72: 8150–8157. genitors using gibbon ape leukemia virus pseudotype retroviral 60 Zufferey R et al. Self-inactivating lentivirus vector for safe and vectors and an improved clinically applicable protocol. Hum efficient in vivo gene delivery. J Virol 1998; 72: 9873–9880. Gene Ther 1998; 9: 225–234. 61 Bukovsky AA, Song JP, Naldini L. Interaction of human 51 Servet-Delprat C et al. Measles virus induces abnormal differen- immunodeficiency virus-derived vectors with wild-type virus in tiation of CD40 ligand-activated human dendritic cells. J Immu- transduced cells. J Virol 1999; 73: 7087–7092. nol 2000; 164: 1753–1760. 62 Hatziioannou T, Valsesia-Wittmann S, Russell S, Cosset F-L. 52 Campbell BJ, Hirsch VM. Vpr of simian immunodeficiency virus Incorporation of fowl plague virus hemagglutinin into murine of African green monkeys is required for replication in macaque leukemia virus particles and analysis of the infectivity of the macrophages and lymphocytes. J Virol 1997; 71: 5593–5602. pseudotyped retroviruses. J Virol 1998; 72: 5313–5317. 53 Carroll R et al. A human immunodeficiency virus type 1 (HIV- 63 Collins MKL, Weiss RA, Takeuchi Y, Cosset F-L. Expression sys- 1)-based retroviral vector system utilizing stable HIV-1 pack- tems. PCT/GB96/02061. WO 97/08330. 1996. aging cell lines. J Virol 1994; 68: 6047–6051. 64 Fugier-Vivier I et al. Measles virus suppresses cell-mediated 54 Corbeau P, Kraus G, Wong-Staal F. Efficient gene transfer by a immunity by interfering with the survival and functions of den- human immunodeficiency virus type 1 (HIV-1)-derived vector dritic and T cells. J Exp Med 1997; 186: 813–823. utilizing a stable HIV packaging cell line. Proc Natl Acad Sci USA 1996; 93: 14070–14075. Reference added in proof 55 Srinivasakumar N et al. The effect of viral regulatory protein 65 Mangeot P-E et al. Development of minimal lentivirus vectors expression on gene delivery by human immunodeficiency virus derived from simian immunodeficiency virus (SIVmac251) and type 1 vectors produced in stable packaging cell lines. J Virol their use for gene transfer into human dendritic cells. J Virol 1997; 71: 5841–5848. 2000 (in press).

Gene Therapy