In Vivo Gene Delivery and Stable Transduction of
`Nondividing Cells by a Lentiviral Vector
`Luigi Naldini, Ulrike Blomer, Philippe Gallay, Daniel Ory,
`Richard Mulligan, Fred H. Gage, Inder M. Verma,* Didier Trono
`
`A retroviral vector system based on the human immunodeficiency virus (HIV) was de-
`veloped that, in contrast to a murine leukemia virus-based counterpart, transduced
`heterologous sequences into HeLa cells and rat fibroblasts blocked in the cell cycle, as
`well as into human primary macrophages. Additionally, the HIV vector could mediate
`stable in vivo gene transfer into terminally differentiated neurons. The ability of HIV-based
`viral vectors to deliver genes in vivo into nondividing cells could increase the applicability
`of retroviral vectors in human gene therapy.
`
`polyadenylation [poly(A)] site from the in-
`sulin gene was substituted for the 3' long
`terminal repeat (LTR) at the end of the nef
`reading frame (I 1). This design eliminated
`cis-acting sequences crucial for packaging,
`reverse transcription, and integration of
`transcripts derived from the packaging plas-
`mid (12). To broaden the tropism of the
`vector, we used a second plasmid that en-
`codes a heterologous envelope protein for
`pseudotyping the particles generated by
`pCMVAR9 (13). Two variants of this con-
`struct were used: One variant encodes the
`amphotropic envelope of MLV (Ampho),
`and the other encodes the G glycoprotein
`of vesicular stomatitis virus (VSV G) (14).
`The latter envelope offers the additional
`advantage of high stability, which allows for
`
`HIV provirus
`
`FT?UlI
`
`Gaq
`
`I Pro I
`
`Pol
`
`Packaging
`construct
`
`SD T
`
`SD AT
`
`Until now, gene therapy protocols have
`often relied on vectors derived from retro-
`viruses such as murine leukemia virus
`(MLV) (1,
`2). These vectors are useful
`because the genes they transduce are inte-
`grated into the genome of the target cells, a
`desirable feature for long-term expression.
`However, these retroviral vectors can only
`transduce dividing cells, which limits their
`use for in vivo gene transfer in nonprolifer-
`ating cells such as hepatocytes, myofibers,
`hematopoietic stem cells, and neurons (3,
`4). The optimal gene transfer system would
`include a retroviral vector based on a virus,
`such as HIV and other lentiviruses, that can
`integrate into the genome of nonproliferat-
`ing cells. In vitro, HIV can infect primary
`cultures of monocyte-derived macrophages
`(5) as well as cell cycle-arrested CD4+
`HeLa or T lymphoid cells (6). Central to
`this ability are karyophilic determinants
`contained in two virion proteins, matrix
`(MA) and Vpr. These proteins interact
`with the nuclear import machinery and me-
`diate the active transport of the HIV pre-
`integration complex through the nucleo-
`pore (7-9).
`A three-plasmid expression system was
`used to generate HIV-derived retroviral
`vector particles by transient transfection, as
`described for other vectors (10) (Fig. 1).
`Plasmid pCMVAR9, the packaging con-
`struct, contains the human cytomegalovirus
`(hCMV) immediate early promoter, which
`drives the expression of all viral proteins
`required in trans. This plasmid is defective
`for the production of the viral envelope and
`the accessory protein Vpu. The packaging
`signal (T) and adjacent sequences were
`deleted from the 5' untranslated region, but
`the 5' splice donor site was preserved. A
`
`L. Naldini, U. Bi6mer, P. Gallay, F. H. Gage, I. M. Verma,
`D. Trono, Salk Institute, 10010 North Torrey Pines Road,
`La Jolla, CA 92037, USA.
`D. Ory and R. Mulligan, Whitehead Institute for Biomedi-
`cal Research, 9 Cambridge Center, Cambridge, MA
`02142, USA.
`*To whom correspondence should be addressed.
`
`particle concentration by ultracentrifugation
`(15). The third plasmid, the transducing
`vector (pHR'), contains cis-acting sequences
`of HIV required for packaging, reverse tran-
`scription, and integration, as well as unique
`restriction sites for the cloning of heterolo-
`complementary DNAs (cDNAs).
`gous
`Nearly 350 base pairs of gag as well as env
`sequences encompassing the Rev response
`element (RRE) flanked by splice signals
`were included in the pHR' vector (16).
`This design had a dual purpose: first, to
`increase packaging efficiency, as both gag
`and env RNA determinants have been dem-
`onstrated to enhance this process (17), and
`second, to allow the efficient transcription
`and cytoplasmic export of full-length vector
`transcripts only in the presence of the HIV
`Tat and Rev regulatory proteins, both of
`which are encoded by the packaging plas-
`mid, pCMVAR9. In the absence of these
`transacting factors, the only detectable ex-
`pression originated from the internal pro-
`moter in the vector (18). The Escherichia
`coli ,-galactosidase (P-gal) or the firefly
`luciferase coding sequences were inserted
`into pHR' downstream of the hCMV im-
`mediate early promoter to serve as reporter
`genes.
`Replication-defective retroviral parti-
`cles were generated by transient cotrans-
`fection of 293T human kidney cells with
`the
`three-plasmid
`combination
`(19).
`MLV-derived packaging and transducing
`vectors served as controls (20). Media
`from the various transfectants were first
`
`Enf
`I
`
`IV-i-f
`
`Env
`
`"*---Tat
`'*--- Rev
`
`RRE
`
`I
`
`I._|
`
`b-I.lm
`
` on September 17, 2008
`
`www.sciencemag.org
`
`Downloaded from
`
`Pol
`
`I
`
`LEL &.11,
`R-e-Tat
`*- ~Rev
`
`I~Poif(A)
`
`t
`
`Ne
`Ne
`
`Env-coding
`plasmids
`
`MLV (Ampho) rP0iy("A
`or
`
`RO
`
`VSV G
`
`Transfer
`vector
`
`Fig. 1. Schematic representation of the HIV provirus and the three-plasmid expression system used for
`generating a pseudotyped HIV-based vector by transient transfection. Only the relevant portion of each
`plasmid is shown. For the HIV provirus, the coding region of viral proteins, including the accessory
`proteins, is shown. The splice donor site (SD) and the packaging signal () are indicated. In the packaging
`construct pCMVAR9, the reading frames of Env and Vpu are blocked (X. In the Env-coding plasmid, the
`coding region of 4070a amphotropic MLV envelope is flanked by a MLV LTR and a SV40 poly(A) site. The
`VSV G coding region is flanked by the CMV promoter and a poly(A) site. In the transfer vector pHR', the
`gag gene is truncated and out of frame (X), and the internal promoter CMV is used to drive expression of
`either P-galactosidase (lacZ) or luciferase cDNA. The Rev responsive element (RRE) and splice acceptor
`site (SA) are shown.
`
`SCIENCE * VOL. 272
`
`*
`
`12 APRIL 1996
`
`263
`
`Page 1 of 5
`
`KELONIA EXHIBIT 1009
`
`

`

`for transduction frequency on
`assayed
`growing 208F rat fibroblasts (21). HIV-
`based 3-gal vectors yielded titers of 0.8
`(±1.7) X 105 (n = 3) transducing units
`(TU) per milliliter with the MLV(Am-
`pho) envelope and 4 (±1.5) x 105 (n =
`6) TU/ml with the VSV envelope. These
`titers are comparable with those obtained
`with MLV-based vectors produced by the
`same method-105 TU/ml with its own
`envelope, and 5 x 105 Tu/ml when
`pseudotyped with the VSV envelope-
`and significantly higher than those previ-
`ously reported for other HIV-based vectors
`(17, 22). Potentially contributing to this
`increased efficiency is the incorporation of
`accessory HIV-1 genes into the packaging
`construct, including nef that markedly en-
`
`Table 1. Relative transduction of cells at different
`stages of the cycle by HIV- and MLV-based vec-
`tors. Results are expressed relative to the trans-
`duction obtained by the vector in growing cells.
`Multiplicity of infection was matched for both vec-
`tors. Abbreviations: arr., arrested; d, days; repl.,
`replated.
`
`Infected
`culture
`
`Tranduction efficiency
`
`HIV-based
`vector
`
`MLV-based
`vector
`
`1
`0.97 t 0.02
`0.71 + 0.22
`
`1
`0.05 ± 0.01
`0.08 ± 0.01
`
`1
`0.45 ± 0.02
`0.29 + 0.02
`0.23 + 0.01
`0.17 + 0.01
`
`1
`0.11 ± 0.01
`0.03 ± 0.02
`0.02 ± 0.01
`0.01 ± 0.01
`
`HeLa cells*
`Growing
`G1-S-arr.
`G2-arr.
`208F cellst
`Growing
`G0-arr. 4 d
`G0-arr. 7 d
`G0-arr. 11 d
`G0-arr. 15 d
`208F cellst
`Growing
`1
`1
`0.05 ± 0.02
`0.08 + 0.02
`Go
`Go repl. 2 d
`0.50 + 0.03
`0.08 ± 0.02
`Go repl. 4 d
`0.43 + 0.04
`0.08 + 0.02
`Go repl. 8 d
`0.08 ± 0.02
`0.46 + 0.07
`*Human HeLa cells were arrested in G1-S by aphidicolin
`treatment or in G2 by exposure to 40 grays (1 gray = 100
`rads) of gamma radiation (25) and infected with ,8-gal
`vector pseudotyped with MLV (Ampho) envelope. Trans-
`duction was scored by X-Gal staining of the cultures 48
`hours after infection. Results are the mean + SEM deter-
`mination from four experiments.
`tRat 208F fibro-
`blasts were plated at low density and either infected the
`following day (growing) or grown to confluence, switched
`to medium containing 5% calf serum and 2 ,uM dexa-
`methasone (3), and further incubated for the indicated
`number of days (d) before infection with luciferase vectors
`pseudotyped with VSV G protein. Transduction was
`scored by measuring luminescence in cell extracts 48
`hours after infection. Results are the mean ± SD of rep-
`licated determinations from a representative experiment
`tRat 208F fibroblasts
`of a total of five performed.
`either growing or arrested in Go for 3 weeks were infected
`with ,B-gal vectors pseudotyped with the MLV (Ampho)
`envelope. Transduction was scored by X-Gal staining
`either 48 hours after infection (growing and Go) or 48
`hours after replating (repl.) at low density Go cultures
`trypsinized at the indicated days after infection (Go re-
`plated X d). Results are expressed relative to the number
`of blue cell foci obtained by infecting growing cells and
`are the mean ± SD of replicated determinations from a
`representative experiment of a total of four performed.
`
`Fig. 2. Reverse transcription and
`nuclear import of the HIV-based
`vector genome in fibroblasts grow-
`ing or arrested in Go, Cultures of
`208F fibroblasts were plated at low
`density and either infected the fol-
`lowing day (growing) or grown to
`confluence and further incubated
`for the indicated number of days
`(Go X days) before infection with
`HIV-based luciferase vector pseu-
`dotyped or not (AEnv) with VSV en-
`velope. At the indicated time in
`hours after infection,
`cells were
`lysed and assayed by PCR with
`primers specific for various prod-
`ucts of reverse transcription, as
`previously described (9, 39). A sam-
`ple of the PCR reaction was ana-
`lyzed by Southern (DNA) blot with a 32P-labeled HIV proviral DNA probe. EL, early products (strong stop
`DNA); LL, late linear products (generated after the second template switch); Ci, two-LTR circles (formed
`in the nucleus).
`
`Ci: fl
`
`& .,
`
`.
`
`1
`
`G
`Growing
`
`1
`
`1
`
`GG
`7 days
`
`1
`
`1
`
`1
`
`21 days
`
`264
`
`SCIENCE * VOL. 272
`
`*
`
`12 APRIL 1996
`
`that inactivate integrase. HIV-1 mutants in
`which the expression of integrase is abro-
`gated by the introduction of a stop codon at
`its 5' end do not reverse transcribe their
`genome efficiently (26). When this muta-
`tion was introduced into the packaging
`construct, it completely prevented trans-
`duction by the resulting vector particles.
`Furthermore, whereas a
`3-gal vector made
`with the wild-type packaging construct had
`a transduction efficiency of 940 TU per
`nanogram of p24 in growing or G1-S--arrest-
`ed cells, a single amino acid change [from
`acid
`valine
`position 64
`aspartic
`to
`at
`(D64V)] in the HIV-1 integrase sequence,
`previously demonstrated to severely decrease
`the activity of this enzyme but not to affect
`any other step of infection (27), reduced the
`efficiency to 54 and 130 TU per nanogram
`of p24 in growing and GC-S-arrested cells,
`respectively (28). Efficient gene transfer in
`both settings was thus dependent on reverse
`transcription as well as integration. Taken
`together, these results indicate that the
`unique features of HIV can be transferred to
`a replication-defective retroviral vector, al-
`lowing
`transduction
`of
`nonproliferating
`cells.
`To test the transduction of cells arrested
`in Go, we grew cultures of rat 208F fibro-
`blasts to confluence and then maintained
`them in Go by density-dependent inhibi-
`tion of growth in the presence of dexameth-
`asone (3). The HIV-based vector was sig-
`nificantly more efficient than its MLV
`equivalent. However, its transduction rate
`decreased as a function of time between
`growth arrest and infection (Table 1). Cells
`growth-arrested for 4 days were transduced
`at levels that were 45% of those observed in
`dividing cells. However, in cells that had
`been maintained in Go for 15 days, the
`relative transduction decreased to 17%.
`The MLV-based vector was significantly
`
` on September 17, 2008
`
`www.sciencemag.org
`
`Downloaded from
`
`hances virion infectivity (23).
`The HIV-derived vector system used
`here is devoid of helper virus per se. Fur-
`thermore, the use of a three-plasmid com-
`bination and of a heterologous envelope,
`as well as the removal of multiple cis-
`acting sequences from the packaging vec-
`tor, makes it unlikely that a replication-
`competent recombinant would be gener-
`ated. The potential transfer of packaging
`functions from producer to target cells was
`assayed by testing for the production of
`the tat and gag gene products in vector-
`transduced cells. Neither protein was de-
`tected, which, considering the sensitivity
`of the assays we used (24), implied that
`the transfer of packaging functions was at
`least three orders of magnitude less effi-
`cient than that of vector sequences. Fur-
`thermore, conditioned medium from seri-
`ally passaged. transduced cells did not
`transfer the reporter gene to naive cells
`(24).
`HIV- and MLV-derived vectors were
`compared for their ability to transduce cells
`blocked at various stages of the cell cycle.
`HeLa cells were growth-arrested at the G1-S
`boundary or at the G2 phase of the cycle by
`aphidicolin treatment or gamma irradia-
`tion, respectively (25). The arrested state of
`the cells at the time of infection was veri-
`fied by propidium iodide staining of the
`DNA and by flow cytometry (18). An HIV-
`based retroviral vector expressing 3-gal was
`as efficient at transducing G1-S- and G2-
`arrested as proliferating HeLa cells, whereas
`its MLV counterpart was only 5 to 8% as
`effective (Table 1). The wider variability
`observed in the transduction of HeLa cells
`arrested by gamma irradiation was perhaps
`due to the cytotoxicity of the treatment.
`To test whether the HIV-based vector
`integrates in the host cell genome, we used
`packaging constructs carrying mutations
`
`Time
`(hours):
`EL: ,
`
`LL:
`
`2 2
`
`2 8 24
`
`2 8 24
`
`2 8 24
`
`-
`
`Page 2 of 5
`
`

`

`ar import of the HIV preintegration com-
`plex by this peptide (9). Neither MA-Vpr
`double mutations nor NLS peptide treat-
`ment affected the ability of the vectors to
`transduce dividing cells (18). The require-
`ment for interaction with the cellular nu-
`clear import machinery, together with the
`lack of significant transduction by the
`that gene
`demonstrates
`MLV vector,
`transfer by the HIV vector did occur in
`nonproliferating macrophages and not
`simply in a small proportion of dividing
`cells in the culture.
`To test if HIV-based vectors can deliv-
`er genes in vivo, we injected highly con-
`centrated stocks of HIV- or MLV-based
`3-gal vectors pseudotyped with VSV G
`protein bilaterally into the corpus striatum
`and hippocampus of adult female rat
`brains (36). Seven or 30 days later the
`brains were removed, sectioned, and pro-
`cessed for immunocytochemistry. Analysis
`with the light microscope showed no
`pathological change in the injected areas
`of the brains, except for a limited deposit
`of debris and lining-up of scavenger cells
`along the needle tract in brains examined
`1 week after injection. These findings
`were even less apparent 1 month after the
`injection. Areas of p-gal-positive cells
`were detected surrounding all injected
`sites for both HIV-based and MLV-based
`vectors. In brains injected with the HIV-
`
` on September 17, 2008
`
`www.sciencemag.org
`
`Downloaded from
`
`n
`
`I--
`
`more affected by the growth arrest. In its
`case, the residual transducing activity re-
`flected the fraction of cells still undergoing
`division, as assessed by propidium iodide
`staining of the cell DNA followed by flow
`cytometry (29). Whereas vector particles
`entered GO-arrested and dividing cells with
`comparable efficiencies (30), they were sig-
`nificantly defective for reverse transcription
`in Go cells
`(Fig.
`2), which resembles a
`phenomenon observed
`in HIV-infected
`quiescent T lymphocytes (31). Neverthe-
`less,
`stable
`transduction
`a
`intermediate
`must have been established, because replat-
`ing and proliferation of Go cells up to 8 days
`after infection revealed titers as high as
`50% of those obtained in dividing cells
`(Table 1). In contrast, inducing cell divi-
`sion even 1 day after inoculation did not
`rescue the MLV-derived vector. The gener-
`ation of a stable infection intermediate by
`the HIV-based vector offers an advantage
`for delivering genes into targets such as
`hematopoietic stem cells. Indeed, it may
`alleviate the need for inducing the prolifer-
`ation of these cells ex vivo, a manipulation
`that can affect their pluripotentiality.
`The decreased transduction efficiency
`of the HIV vector in GO-arrested fibro-
`blasts may partly reflect suboptimal con-
`centrations of intracellular deoxynucleo-
`tides (32). Whether a similar limitation
`would preclude gene transfer into termi-
`
`nally differentiated primary cells could not
`be inferred from these observations and
`was therefore assessed directly. The HIV-
`based luciferase vector, pseudotyped with
`the VSV G protein, was tested for its
`ability to transduce human monocyte-de-
`rived primary macrophages (33). Signifi-
`cant levels of luciferase activity were de-
`tected in an envelope-dependent manner
`(Table 2). In contrast, only background
`levels of luciferase activity were measured
`in macrophages inoculated with a compa-
`rable VSV G-pseudotyped MLV-based
`vector (34). To rule out that the HIV
`vector was infecting a small proportion of
`macrophages that were proliferating, we
`generated mutant packaging constructs
`where Vpr and the nuclear localization
`signal (NLS) present in the MA protein
`were inactivated (35). At least one of
`these two elements is essential for viral
`infection in macrophages, because they
`mediate nuclear import of the HIV prein-
`tegration complex (7-9). A vector assem-
`bled from a mutant packaging construct in
`which both Vpr and the MA NLS are
`inactivated was severely reduced in its
`ability to transduce macrophages (Table
`2). Similarly, NLS peptide treatment pre-
`vented transduction by a vector produced
`from a Vpr-defective packaging construct,
`thus corroborating the previously demon-
`strated inhibition of MA-mediated nucle-
`
`Stnatum
`1 week
`
`Striatum
`4 weeks
`
`Hippocampus
`4 weeks
`
`Fig. 3. The in vivo transduction of adult rat neurons. Confocal microscope
`images of sections from brains injected with HIV-based 1-gal vectors stained
`by immunofluorescence for 1-gal, NeuN, and glial fibrillary acidic protein
`(GFAP). The images obtained from each individual staining and from their
`overlap are shown, as indicated on the top. Representative fields of the area
`
`surrounding the injection site are shown for a section from striatum 1 week
`and 4 weeks and from hippocampus 4 weeks after injection of the vector.
`Several cells doubly labeled for 1-gal and NeuN (arrows) are evident in the
`sections. The overall pattern was reproduced in all five animals (three exam-
`ined after 7 days, and two after 30 days) injected with the HIV-based vector.
`
`SCIENCE * VOL. 272
`
`*
`
`12 APRIL 1996
`
`265
`
`Page 3 of 5
`
`

`

` on September 17, 2008
`
`www.sciencemag.org
`
`Downloaded from
`
`Virology 198, 336 (1994); E. Vicenzi etal., J. Virol. 68,
`7879 (1994).
`13. P. Lusso etal., Science 247, 848 (1990); D. H. Spec-
`tor et al., J. Virol. 64, 2298 (1990); N. R. Landau, K.
`A. Page, D. R. Littman, ibid. 65,162 (1991).
`14. Plasmid pSV-A-MLV-env [K. A. Page, N. R. Landau,
`D. R. Littman, J. Virol. 64, 5270 (1990)] encodes the
`amphotropic envelope of the 4070 Moloney leuke-
`mia virus under the transcriptional control of the MLV
`LTR. Plasmid pMD.G encodes the envelope protein
`G of the vesicular stomatitis virus under the tran-
`scriptional control of the CMV promoter.
`15. J. C. Burns, T. Friedman, W. Driever, M. Burrascano,
`J.-K. Yee, Proc. Natl. Acad. Sci. U.S.A. 90, 8033
`(1993); J.-K. Yee et al, ibid. 91, 9564 (1994).
`16. Plasmid pHR' was constructed by cloning a frag-
`ment of the env gene encompassing the RRE and a
`splice acceptor site between the two LTRs of the
`HIV-1 proviral DNA. The gag gene was truncated
`and its reading frame blocked by a frameshift muta-
`tion. pHR'-CMVLacZ was generated by cloning a
`3.6-kbp Sal l-Xho
`fragment containing the CMV
`promoter and the E coli lacZ gene (encoding P-gal)
`from plasmid pSLX-CMVLacZ (20). pHR'-CMVLucif
`was made by replacing a Bam HI-Xho fragment in
`pHR'-CMVLacZ, containing the lacZ gene, with a
`fragment from pGEM-luc
`1.7-kbp Bam HI-Xho
`(Promega) containing the firefly luciferase gene.
`17. G. L. J. Buchschacher and A. T. Panganiban, J.
`Virol. 66, 2731 (1992); J. H. Richardson, L. A. Child,
`A. M. L. Lever, ibid. 67, 3997 (1993); C. Parolin, T.
`Dorfman, G. Palu, H. G6ttlinger, J. Sodroski, ibid. 68,
`3888 (1994); J. H. Richardson, J.
`F. Kaye, L. A.
`Child, A. M. L. Lever, J. Gen. Virol. 76,691 (1995); R.
`D. Berkowitz, M.-L. Hammarskj6ld, C. Helga-Maria,
`D. Rekosh, S. P. Goff, Virology 212, 718 (1995); J. F.
`Kaye, J. H. Richardson, A. M. L. Lever, J. Virol. 69,
`6588 (1995).
`18. L. Naldini et al., unpublished data.
`19. A total of 40 ,ug of plasmid DNA was used for the
`transfection of a 10-cm-diameter plate of 293T
`in the following
`cells
`proportions:
`10 p.g
`of
`pCMVAR9, 20 p.g of pHR', and 10 p.g of env plas-
`mid, as described [C. Chen and H. Okayama, Mol.
`Cell. Bio. 7, 2745 (1987)]. Conditioned medium
`was harvested 48 to 60 hours after transfection,
`subjected to low-speed centrifugation,
`filtered
`through 0.45-pm filters, and assayed for p24 Gag
`antigen by enzyme-linked immunosorbent assay
`(ELISA) (DuPont). The average vector yield was 50
`to 80 ng of p24 per milliliter.
`20. MLV-based vectors were produced from transient
`transfection in 293T cells of the following plasmids.
`pSLX-CMVLacZ [R. Scharfmann, J. H. Axelrod, I.
`M. Verma, Proc. Natl. Acad. Sci. U.S.A. 88, 4626
`(1991)] is a MLV-derived vector carrying a hCMV-
`driven E. coli lacZ gene. The pCL plasmid series
`carries a hybrid CMV-LTR promoter that allows for
`CMV-driven transcription in the packaging cell and
`reconstitution of a functional LTR in the target cell
`(R. Naviaux, E. Costanzi, M. Haas, I. Verma, in
`preparation). The luciferase gene was cloned in
`vector pCLNCX, creating pCLNCLuc. MLV-based
`vectors with the cognate MLV (Ampho) envelope
`were produced by the cotransfection of either of
`the vector plasmids with the amphotropic packag-
`ing plasmid pCL-Ampho. VSV G-pseudotyped
`vectors were produced by the cotransfection of
`either of the vector plasmids with the MLV gag-pol
`packaging plasmid pCMV-GAGPOL and the VSV
`G plasmid.
`21. Rat 208F cells were infected overnight in six-well
`plates with serial dilutions of conditioned medium
`from 293T transient transfectants or with concen-
`trated viral stocks in culture medium supplemented
`with polybrene (8 p.g/ml). The medium was re-
`placed, the cells further incubated for 36 hours,
`and expression of ,-gal scored by X-Gal staining.
`Titers were calculated by counting the number of
`foci of blue cells per well and dividing that number
`by the dilution factor. Transduction of the reporter
`gene was only observed when the packaging vec-
`tor and Env-coding plasmid had been cotrans-
`fected in 293T cells; no transduction was observed
`when either plasmid was omitted or when the HIV-
`based vector was cotransfected with an MLV-
`
`Our results lend strong credence to the
`that HIV-based vectors
`idea
`transduce
`genes efficiently and can be used for in vivo
`gene delivery. Because retroviruses inte-
`grate in the genome of the target cells,
`transduction
`repeated
`is
`unnecessary.
`Therefore, in contrast to an adenoviral vec-
`tor capable of in vivo gene delivery, prob-
`lems linked to the humoral response to
`injected viral antigens can be avoided (38).
`Furthermore, the vectors described here are
`consequently,
`replication
`defective;
`the
`transduced cells lack viral protein that
`could trigger a cellular immune response. A
`major goal of our work was to establish a
`proof of principle that lentiviral vectors can
`be used for stable in vivo gene delivery in
`nondividing cells. For human experimenta-
`tion, it may be more prudent to develop
`vectors derived from nonhuman lentivi-
`ruses such as simian immunodeficiency vi-
`rus, bovine immunodeficiency virus,
`or
`equine infectious anemia virus. We believe
`that the generation of safe and efficacious
`lentiviral vectors will significantly advance
`the prospects of human gene therapy.
`
`REFERENCES AND NOTES
`
`1.
`
`I. M. Verma, Mol. Med. 1, 2 (1994); R. C. Mulligan,
`Science 260, 926 (1993); R. G. Crystal, ibid. 270,
`404 (1995); J. M. Leiden, N. Engl. J. Med. 333, 871
`(1995); R. Sanders Williams, Nature Med. 1, 1137
`(1995).
`2. A. D. Miller, D. G. Miller, J. V. Garcia, C. M. Lynch,
`Methods Enzymol. 217, 581 (1993).
`3. D. G. Miller, M. A. Adam, A. D. Miller, MoI. Cell. Biol.
`10, 4239 (1990).
`4. T. Roe, T. C. Reynolds, G. Yu, P. O. Brown, EMBO J.
`12, 2099 (1993); P. F. Lewis and M. Emerman, J.
`Virol. 68, 510 (1994).
`5. J. B. Weinberg, T. J. Matthews, B. R. Cullen, M. H.
`Malim, J. Exp. Med. 174,1477 (1991).
`6. P. Lewis, M. Hensel, M. Emerman, EMBO J. 11,
`3053(1992).
`7. M. I. Bukrinsky et al., Nature 365, 666 (1993); N. K.
`Heinzinger et al., Proc. Nati. Acad. Sci. U.S.A. 91,
`7311 (1994).
`8. U. von Schwedler, R. S. Kombluth, D. Trono, Proc.
`Natl. Acad. Sci. U.S.A. 91, 6992 (1994).
`9. P. Gallay, S. Swingler, C. Aiken, D. Trono, Cell 80,
`379 (1995); P. Gallay, S. Swingler, J. Song, F. Bush-
`man, D. Trono, ibid. 83, 569 (1995); P. Gallay, V.
`Stitt, C. Mundy, M. Oettinger, D. Trono, J. Virol. 70,
`1027 (1996).
`10. N. R. Landau and D. R. Littman, J. Virol. 66, 5110
`(1992); W. S. Pear, G. P. Nolan, M. L. Scott, D.
`Baltimore, Proc. Natl. Acad. Sci. U.S.A. 90, 8392
`(1993); M. H. Finer, T. J. Dull, L. Qin, D. Farson, M. R.
`Roberts, Blood 83, 43 (1994); Y. Soneoka et al.,
`Nucleic Acids Res. 23, 628 (1995).
`11. Briefly, plasmid pCMVAR9 was constructed from an
`env-defective version of pR9, an infectious molecular
`clone of proviral HIV-1 DNA made by cloning the Bss
`HII-Bam HI fragment of NL4.3 in pR7, with insertion
`of a Mlu linker at the Stu site that frameshifts the
`env reading frame [D. Trono, M. B. Feinberg, D.
`Baltimore, Cell 59,113 (1989)1. A 39-base pair (bp)
`intemal deletion in the qi sequence (12) was intro-
`duced, and the 3' HIV long terminal repeat (LTR) was
`replaced with the poly(A) site of insulin genomic
`DNA. The 5' LTR and leader sequences of HIV were
`substituted with a 0.8-kbp fragment containing the
`CMV promoter.
`12. A. Lever, H. Gottlinger, W. Haseltine, J. Sodroski, J.
`Virol. 63, 4085 (1989); A. Aldovini and R. A. Young,
`ibid. 64,1920 (1990); J. Luban and S. P. Goff, ibid.
`68, 3784 (1994); H.-J. Kim, K. Lee, J. J. O'Rear,
`
`SCIENCE * VOL. 272
`
`*
`
`12 APRIL 1996
`
`m
`
`Table 2. Transduction of human monocyte-de-
`rived macrophages. Primary cultures of human
`macrophages prepared from different donors were
`incubated
`with
`HIV-based
`luciferase
`vectors
`pseudotyped with VSV G protein and generated
`either from wild-type (pCMVAR9) or mutant pack-
`aging plasmids carrying inactivating mutations in
`the vpr gene (AVpr) or both in the vpr gene and the
`MA NLS [AVpr ANLS MA (33, 34)]. Macrophage
`cultures were incubated with 100 puM of peptide
`whose sequence corresponded either to the SV40
`T antigen NLS to block NLS-dependent nuclear
`import, or to its reverse, inactive orientation (Rev.
`NLS), starting 1 hour before and throughout infec-
`tion, as previously described (8). Luminescence
`was measured in cell extracts 48 hours after infec-
`tion. Transduction was dependent on active nucle-
`ar import of the vector in target cells, as it was
`inhibited by mutations inactivating Vpr and the MA
`NLS in the packaging plasmids and when infected
`cells were incubated with NLS peptide.
`
`Packaging
`plasmid
`
`Cell
`treatment
`
`Luminescence
`(RLU)*
`
`Donor 1
`
`Donor 2
`
`-
`Rev. NLS
`NLS
`Rev. NLS
`NLS
`-
`
`0
`17,873
`16,225
`8,447
`1,141
`3,501
`
`Wild typet
`Wildtype
`Wild type
`AVpr
`AVpr
`AVprANLS MA
`*Luminescence in relative units above background of 50
`p.d of infected macrophages extract.
`tAs a control,
`this plasmid was not pseudotyped with VSV G protein.
`
`0
`17,785
`15,322
`9,687
`1,348
`2,787
`
`based vector, a variety of ,3-gal-positive
`cells with a morphology resembling neu-
`oligodendrocytes,
`rons,
`and astrocytes
`could be detected (18). To further identify
`the cell types transduced by both vectors,
`we used confocal microscopy after immu-
`nofluorescence staining with antibodies
`specific for ,B-gal,
`glial fibrillary acidic
`protein (GFAP, a marker for astrocytes),
`and NeuN (a marker for terminally differ-
`(37). Sections from
`entiated neurons)
`brains injected with the MLV-based vec-
`tor contained cells either labeled only for
`3-gal and GFAP (18).
`l3-gal or for both
`The MLV vector was unable to transduce
`neurons because no cells labeled for both
`3-gal and NeuN were detected. In con-
`trast, the striatum of animals injected with
`the HIV-based vector showed multiple
`3-gal and NeuN
`cells double-labeled for
`(Fig.
`3, top panel), demonstrating the
`ability of the HIV-based vector to infect
`and transduce genes in terminally differ-
`entiated neurons. NeuN and p-gal double-
`positive cells were also detected in the
`hippocampus of brains injected with the
`HIV vector. As expected, the HIV-based
`vector was also able to transduce astroglial
`cells (18). The expression of ,B-gal in neu-
`rons in the striatum and the hippocampus
`could be detected after a 30-day period,
`the longest time tested (Fig. 3, bottom two
`panels).
`
`266
`
`Page 4 of 5
`
`

`

`institution-approved protocols and in a biosafety
`level 3 environment. Adult female Fischer 344 rats
`were anesthetized [ketamine (44 mg per kilogram
`of body weight), acepromazine (0.75 mg/kg), and
`xylazine (4 mg/kg) in 0.9% NaCI, intraperitoneally],
`positioned in a stereotactic head frame, and slowly
`injected with 2 1d of vector stock into the striatum
`[anteroposterior (AP), +0.2; mediolateral (ML),
`±3.5; dorsoventral (DV), -4.5] and hippocampus
`(AP, -3.5, ML, 3.0; DV, -4.0) bilaterally. Seven or-
`30 days after injection the rats were deeply anes-
`thetized and perfused with 4% cold paraformalde-
`hyde and 0.2% glutaraldehyde intracardially. The
`brains were removed, postfixed 24 hours, saturat-
`ed in 30% sucrose, and sectioned on a freezing
`microtome (40-p.m sections). Light microscopy
`sections were stained with avidin-biotin peroxidase
`(Vectastain Elit, Vector Labs) and diaminobenzi-
`dine. Immunofluorescence triple labeling was con-
`ducted with rabbit antibody to p-gal (anti-p-gal )
`(1:1000, Cortex), mouse monoclonal anti-NeuN
`(1: 4), and guinea pig anti-GFAP (1:250, Advanced
`Immunochemical). Secondary antibodies coupled to
`fluorescent markers CY5, dichlorotriazinyl amino fluo-
`rescein, and Texas Red were used at 1: 250 dilution.
`Slices were mounted with diazobicyclooctane/polyvi-
`nyl alcohol mounting medium and analyzed by confo-
`cal scanning laser microscopy (Bio-Rad MRTC600).
`Fluorescent signals were collected, digitally color-en-
`hanced, and superimposed. False-color images were
`generated electronically with Adobe Photoshop (Ado-
`be System).
`37. R. J. Mullen, C. R. Buck, A. M. Smith, Development
`116, 201 (1992).
`38. R. M. Knowles et al., N. Engl. J. Med. 333, 823
`(1995).
`39. For the polymerase chain reaction (PCR) assay, cul-
`tures were incubated with vector concentrated by
`ultracentrifugation and pretreated with deoxyribonu-
`(DNase I) (20 p.g/ml for 2 hours at 370C),
`clease
`washed, trypsinized, and extracted for PCR as pre-
`viously described (9). To adjust for the different con-
`tent of cellular DNA in growing and confluent cul-
`tures, each PCR reaction contained an equal volume
`(2 p.d for EL and LL, 7.5 p.l for Ci) of both growing and
`GO-arrested cell extract, either one of which had
`been infected. The sequence of HIV-specific primers
`are as follows [positions of nucleotides in the
`HIV-1 HXB2D sequence, according to L. Ratner et al.,
`Nature 313, 277 (1985), are indicated in parenthe-
`ses]. LTR5: GGCTAACTAGGGAACCCACTGCTT
`(496 to 516); LTR6: CTGCTAGAGATTTTCCA-
`CACTGAC (635 to 612); 5NC2: CCGAGTCCT-
`GCGTCGAGAGAGC (698 to 677); LTR8: TCCCAG-
`GCTCAGATCTGGTCTMC (488 to 465 and 9572 to
`9549); and LTR9: GCCTCAATAAAGCTTGCCTTG
`(522 to 542 and 9606 to 9626). LTR5 plus LTR6
`amplifies minus-strand strong stop DNA, LTR5 plus
`5NC2 amplifies double-stranded molecules generat-
`ed after the second template switch, and LTR8 plus
`LTR9 amplifies two LTR circles. A series of logarith-
`mic dilutions of pHR' plasmid used as a template
`showed linearity of the PCR reaction over the early
`time points.
`40. We are grateful to G. Nolan, A. Leavitt, and R. J.
`Mullen for providing reagents; members of the
`Verma, Gage, and Trono laboratories for helpful sug-
`gestions; and J. Stevenson for critical reading of the
`manuscript.