Helper virus-free transfer of human immunodeficiency virus type 1 vectors

Journal of General Virology (1995), 76, 691 696. Printed in Great Britabz 691 Helper virus-free transfer of human immunodeficiency virus type 1 vectors Jennifer H. Riehardson,~ Jane F. Kaye, Lisa A. Child and Andrew M. L. Lever* University of Cambridge Department of Medicine, Addenbrooke's Hospital, Hills Road, Cambridge CB2 2QQ, UK Recombinant vectors based on the type 1 human immunodeficiency virus (HIV-1) can be used to deliver genes into cells expressing the HIV receptor, CD4. We have used a transient RNA packaging system to compare the safety and efficacy of HIV-1 vector transduction by wild-type and replication-deficient helper viruses. Helper virus-free vector transfer was consistently achieved when the helper virus gag-pol and env genes were expressed from separate plasmids such that two recombination events were required to form an infectious genome. Other forms of attenuation, such as deletion of the 5' W region, were inadequate to prevent helper virus transmission. Vector transduction by the wild-type and non-replicating helper viruses occurred with comparable efficiency except in instances where efficient vector RNA expression was dependent upon transactivating factors supplied by the helper virus. These data demonstrate the feasibility of safe gene transfer using HIV-1 vectors. Retrovirus vectors currently offer the most efficient means of stable gene transfer into vertebrate cells. Recombinant vectors based on human immunodeficiency virus type 1 (HIV-1) provide in addition a means of targeting foreign genes to cells expressing CD4, the HIV receptor, and could be used to deliver genes with therapeutic or antiviral potential into primary human T cells. Vectors based on HIV-1 may have other advantages over murine retroviral vectors, such as the ability to infect non-dividing cells (Weinberg et al., 1991 ; Lewis et al., 1992) and the potential for using the HIV-1 regulatory genes tat and rev to achieve inducible expression of the vector RNA. One factor currently limiting the utility of HIV-1 vectors is the absence of an efficient system for vector packaging. In this paper we examine aspects of helper virus design that will contribute to the design of a safe and effective packaging system. Retroviral vectors, in which some or all of the viral genes are replaced by foreign sequences, require a separate helper virus in order to be packaged and transferred to the target cell. To avoid transmission of the helper virus itself, vector systems based on murine and avian retroviruses have used helper viruses which have a cis-acting replication defect such as deletion of the RNA packaging signal (termed q~ or E). This cis-acting element allows selective encapsidation of the viral genomic RNA. In all retroviruses studied, an essential * Author for correspondence. Fax +44 1223 336846. t Present address: Division of Human Retrovirology, Dana-Farber Cancer Institute, 44 Binney Street, Boston, MA 02115, USA. component of the RNA packaging signal is located at the 5' end of the genome, often encompassing or downstream of the major splice donor (Linial & Miller, 1990; Richardson et al., 1993 and references therein). Cell lines stably transfected with the mutant proviruses have been termed packaging cell lines and produce virion particles devoid of the ~- helper virus genome but capable of encapsidating a ud+ vector RNA. In practice, removal of the packaging signal has proved insufficient to prevent helper virus transmission because of low-level encapsidation of the W- genome and recombination between the helper virus and vector RNAs. In the more recent packaging lines based on Moloney murine leukaemia virus (M-MuLV), reticuloendotheliosis virus (REV) and avian leukosis virus (ALV), the scope for helper virus transfer has been substantially reduced by replacement of the helper virus long terminal repeat (LTR) sequences with alternative promoter or polyadenylation signals and/or by the physical separation of gag-pol and env genes (Miller, 1990; Dougherty et al., 1989; Cosset et al., 1990). These modifications increase the number of recombination events required to generate an infectious genome. We recently described a series of HIV-1 vectors and identified several that could be packaged and transduced with high efficiency by a replication-competent (wild-type) helper virus. In this paper we show that helper virus-free transfer of such vectors can be achieved using strategies similar to those employed in murine and avian retrovirus packaging cell lines. A transient packaging system was used to compare the transducing capability and infectivity of wild-type HIV- 1 and a series of replication-deficient helper viruses. 5 to 0001-2771 1995 SGM

692 Short communication (a) HVP ]J fl ~ 6.4 kb gag [pr0 }..v I pu.o A A LGRPL IN gag [ env ]PURO 4-5 kb (b) g.g I.o, env m (i) HXBc2 TGGTGAGTACGCCAAAAATTTFGACTAGCGGAGGCTAGAAGGAGAGAGATG A (ii) HXB,~P1 (iii) phxbap2 TGGTGAGTACGCC A TGGTGAGTACT~ A -GGCTAGAAGGAGAGAGATG GAGATG (iv) phxbap1altr ~ ga~ [ pol env (v) phxbapi~env ~ gag [ pol ~ env + + plenv [-~'--F'" env k'1 Fig. 1. (a) HIV-1 vectors. Hatched and filled boxes represent exons of tat and rev respectively. The underlined area within env represents the Rev-responsive element. Relevant splice donor (~) and splice acceptor (A) sites are indicated. (b) Helper virus constructs used for packaging. 10 lag each of vector and packaging plasmids (Fig. 1) were cotransfected into COS-1 cells using the DEAEdextran technique (Selden, 1987). All of the plasmids contain a simian virus 40 (SV40) origin of replication. Virus-containing supernatants were harvested 72 h later and used immediately to infect CD4-positive or -negative cells. A colony-forming assay was used to estimate the efficiency of vector transduction by the various helper virus plasmids (Richardson et al., 1993). Briefly, 0.45 lamfiltered vector stocks were inoculated onto HeLa cells expressing human CD4 or CD8 and the number of puromycin-resistant colonies was scored 10 to 12 days later. Vector stocks were also used to infect the CD4- positive cell line Jurkat tat (Rosen et al., 1986). Puromycin selection was applied 24 h later at 0.5 lag/ml and cells containing the transduced vector were cultured for molecular and virological analysis. The helper virus constructs used were (i) phxbc2, a fully infectious molecular clone of HIV-1 ; (ii) phxbap 1, a previously described mutant of phxbc2 which has a 19 bp deletion between the major splice donor and gag initiation codon, resulting in impaired viral replication (Lever et al., 1989); (iii) phxbap2, which has a 36 bp deletion between the splice donor and gag initiation codon (this deletion results in a more severe replication defect); (iv) phxbapialtr, a non-infectious derivative of phxbap1 which was obtained by deleting sequences downstream of the XhoI site (8897) (this removes the entire 3" LTR without affecting viral protein synthesis; the polyadenylation site used in this construct

Short communication 693 Table 1. Vector transduction efficiency by different helper virus constructs Puro R (c.f.u./ml)* HeLaT4 Helper Secondary Packaging HeLaT8 virus vector Vector plasmid/s) 1 1 2 3 4 TCIDs0 ~" transfer HVP HXBc2 0 188 115 46 2300 2590 + HVP HXBAP1 0 140 140 68 ND 690 + HVP HXBAP2 0 195 ND ND 3770 127 + HVP HXBAP1AYLTR 0 190 61 85 5770 < 10 + HVP Dual plasmid 0 101 16 20 2424 0 - LGRPL HXBc2 0 ND 16 22 804 ND + LGRPL Dual plasmid 0 ND 3 4 32 ND -- * Normalized for a typical reverse transcriptase activity of 106 c.p.m./ml. f Infectious virus titre expressed as syncytium-forming units/ml of COS supernatant; normalized for reverse transcriptase activity of 106 c.p.m./ml: mean of four experiments. ND, Not done. has not been determined); (v) phxbap1aenv, a noninfectious mutant ofphxbap1 lacking a BglII fragment (7041-7621) in the env gene. The env-deleted provirus required a complementary envelope-expressing plasmid to produce intact virion particles and we refer to this below as a 'dual plasmid' helper virus system. Table 1 summarizes the vector transfer results. The HVP vector, a tat- and rev-positive vector that also contains gag protease and env sequences (Richardson et al., 1993) was transduced with moderate efficiency by all of the helper virus constructs. There was no significant difference between the transduction titres obtained using the wild-type, ul-, and LTR viruses. A small but reproducible decrease in titre was seen when the dual plasmid helper system was used. No colonies were obtained when HeLa cells expressing the CD8 molecule were used for infection, clearly demonstrating the CD4- specificity of the transducing virus. The actual titres were somewhat variable between experiments, suggesting that the COS cell system is rather sensitive to minor variations in experimental conditions. LGRPL is a Tat and Rev-inducible vector, containing gag and env sequences but lacking Tat and Rev functions. Using the wild-type HXBc2 helper virus, the LGRPL vector gave ~ fourfold fewer colonies than HVP. We have shown elsewhere that both vectors are efficiently encapsidated (Richardson et al., 1993) and the lower titre may reflect differences in transfection efficiency or expression of the two vector plasmids. Transduction of the LGRPL vector was consistently reduced by a factor of about ten when the dual plasmid helper system was used. The selectable marker (puro gene in LGRPL) is expressed using the 5' LTR promoter and is sensitive to transactivation by the HIV-1 Tat protein. Infection of the target cells by helper virus may therefore improve the colony-forming ability of this vector by Tat-mediated transactivation of the vector LTR. To compare the infectivity of the wild-type and replication-deficient helper viruses, titres of infectious virus in the COS cell supernatants were measured using a TCIDs0 assay (Johnson & Byington, 1990). Briefly, Jurkat tat cells were infected with serial fourfold dilutions of virus (six replicates at each dilution) and wells were scored for the presence of syncytia 7 to 10 days later. The results in Table 1 clearly demonstrate the reduced infectivity of the mutant proviruses HXBAP1 and HXBAP2 but also indicate that this level of attenuation is not sufficient to prevent helper virus transmission. Infectious virus was also recovered on one of three occasions when the HVP vector was cotransfected with the non-infectious provirus HXBAP 1ALTR, which lacks a 3' LTR. This was not unexpected as a single recombination event between the vector and helper virus would generate an infectious genome. Infectious virus was reduced to undetectable levels by expressing Gag-Pol and Env proteins from separate plasmids so that two recombination events were required to generate an infectious genome. Further data on helper virus transmission was obtained by monitoring the target cell cultures for evidence of infection. These were assayed for HIV-1 cytopathic effects (syncytium formation) and reverse transcriptase activity over a 4 to 6 week period (Ports, 1990). RNA isolated from the cells was also screened for the presence of helper virus RNA by Northern blot analysis and cellfree culture supernatant from the transduced cells was inoculated onto fresh Jurkat tat cells to test for horizontal (or secondary) transfer of the vector. The wild-type helper virus was invariably transmitted to target cell cultures, as determined by all of the above criteria (Table 1 and Fig. 2). In contrast, ceils transduced using the dual plasmid packaging system showed no evidence of reverse transcriptase activity or cytopathic effect during a 6 week monitoring period. Moreover, supernatants from these

694 Short communication (a) 2~ (c) "d a:,.~,...a,..a,..a,-...1,-.1,..a 1 28S -- 23 -- 9.4 --6-6 --4-4 118s --2-3 --2.0 28S 18S 1 (b) 2 3 ~ gag ~a~ gag l pol 533 I 6 7 1~ env I PURO 3-50 2-20 HVP 5 3.88 LGRPL 1~ 4 ]pro -~ ~ env ~ XI~1 N env I PURO ~ HXBC2 357 Fig. 2. (a) Northern blot analysis of cells transduced with the HVP vector. Filters were hybridized with puro gene (upper panel) or pol gene (lower panel) probes. Each lane represents a separate experiment and the helper virus used for transduction is indicated above each lane. (b) Diagram showing the expected SacI restriction fragments of the HXBc2 provirus, HVP and LGRPL. Recombination in the region shown would explain the R band. (c) Southern blot analysis of Jurkat tat cells following stable transfection or transduction of the HVP and LGRPL vectors. Genomic DNA digested with SacI was probed with a SacI fragment of LGRPL. Lane 1, cells transfected with HVP; lanes 2 and 3, cells transduced with HVP using wild-type (HXBc2) and dual plasmid helper virus systems; lane 4, HTLVIII~-infected cells; lane 5, cells transfected with LGRPL; lanes 6 and 7: cells transduced with LGRPL using wild-type (HXBc2) and dual plasmid helper systems. R indicates a putative recombination product described in the text.

Short communication 695 cultures failed to transmit puromycin resistance to fresh Jurkat tat cells, consistent with the absence of packaging functions in the cells. A cumulative total of ~ 30 ml of dual plasmid-derived vector stock was screened in this way for replication-competent virus and none was detected. Northern blot analysis confirmed the presence of the intact vector RNA in the target cells and showed no evidence of helper virus infection in 4/4 independent experiments when the dual plasmid packaging system was used (Fig. 2a). Total RNA was extracted from the cells and hybridized with puro and pol gene probes as previously described (Richardson et al., 1993). The puro probe revealed the expected HVP vector transcripts of approximately 6.1, 4.2, and 2.1 kb (and a minor 1"7 kb species, probably arising from a cryptic 3' splice site). Using a helper virus-specific probe (pol fragment 4648-5743), full-length (9 kb) helper virus RNA could be demonstrated in target cells when the wild-type (HXBc2) virus was used for transduction but was not detectable when vector transduction was achieved using the dual plasmid helper system. Southern blot analysis was used to confirm the presence of intact vector in target cell cultures. Using a SacI fragment of LGRPL to probe SacI-digested genomic DNA, fragments of the expected sizes for HVP (2.20 kb and 3"50 kb) and LGRPL (3.88 kb) were seen. Bands of 3'57 kb and 5'33 kb corresponding to the helper provirus can be seen when infectious helper virus was used but not after transfer using the dual plasmid helper system (Fig. 2). Where both infectious virus and vector are present, an additional fragment labelled R in Fig. 2(c) can be seen. From its length (~ 4 kb) we postulate that this fragment is the result of recombination between the vector and helper virus which would produce an extended env +puro fragment of 4.08 kb. Cellular RNA prepared at the same time as the DNA showed little evidence of such a recombination (Fig. 2), but this may relate to the kinetics of formation or expression of the recombinant provirus. The development of HIV- 1-based vectors as a safe and effective means of gene delivery carries the dual advantages that genes can be targeted to CD4-positive cells and that, subsequently, vector gene expression can be induced by viral trans-activators such as the Tat and Rev proteins. We have shown here that helper virus-free transmission of HIV-1-based vectors is both feasible and CD4-specific. Helper virus-free vector transfer was routinely achieved when the helper virus gag-pol and env genes were expressed from separate plasmids such that two recombination events were required to form an infectious genome. Other forms of attenuation such as deletion of the 5' W region were inadequate to prevent helper virus transmission as the W-deleted genomes were encapsi- dated to a low extent. Infectious virus was also recovered in circumstances where a single recombination event between a defective helper virus and the vector RNA could generate an infectious genome. The vector titres obtained using this transient packaging system were considerably lower than those obtained when the same vectors are packaged by a wildtype helper virus in T cell lines (Richardson et al., 1993). This partly reflects the lower concentration of virus particles in the COS cell supernatant, but even after normalizing for reverse transcriptase activity, transduction titres were ~ 10-fold lower using the COS cell system. Other groups have reported similar low titres when using COS-1 cells for HIV-1 vector packaging, suggesting that this is not an optimal cell type for the assembly of HIV-1 vectors (Hayashi et al., 1992; Shimada et al., 1991; Buchschacher & Panganiban, 1992). With the development of stable packaging cell lines capable of producing high titres of helper virus-free vector, HIV-1 vectors should ultimately prove a safe and effective tool for the genetic manipulation of CD4- expressing cells. The system may find wide application in molecular immunology and has clear potential for a genetic approach to the treatment of HIV infection. We thank Dr J. Sodroski for a number of plasmids used in this study and Charles Wright for technical assistance. This work was supported by the Medical Research Council (UK) AIDS Directed Programme, NATO and the Sykes Trust. References BUCHSCHACHER, G. L. & PANGANIBAN, A. T. (1992). Human immunodeficiency virus vectors for inducible expression of foreign genes. Journal of Virology 66, 2731-2739. COSSET, F. L., LEGRAS, C., CHEBLOUNE, Y., SAVATIER, P., THORAVAL, P., THOMAS, J. L., SAMARUT, J.~ NIGON, V. M. VERDIER, G. (1990). A new avian leukosis virus-based packaging cell line that uses two separate transcomplementing helper genomes. Journal of Virology 64, 1070-1078. DOUGHERTY, J. P., WISNIEWSKI, R., YANG, S., RHODE, B. W. & TEMIN, H. W. (1989). New retrovirus helper cells with almost no nucleotide sequence homology to retrovirus vectors. Journal of Virology 63, 3209-3212. HAYASHIL T., SHIODA, T., IWAKURA, Y. & SEIIBUTA, H. (1992). RNA packaging signal of human immunodeficiency virus type 1. Virology 188, 590-599. JOHNSON, V.A. & BYINGTON, R.E. (1990). Quantitative assays for virus infectivity. In Techniques in HIV Research, pp. 71-76. Edited by A. Aldovini & B. D. Walker. New York: Stockton Press. LEVER, A. M. L., GOTTLINGER, H., HASELTINE, W. & SODROSKI, J. (1989). Identification of a sequence required for efficient packaging of human immunodeficiency virus type 1 RNA into virions. Journal of Virology 63, 4085-4087. LEWIS, P., HENSEL, M. & EMERMAN, M. (1992). Human immunodeficiency virus infection of cells arrested in the cell cycle. EMBO Journal 11, 2053-3058. LINIAL, M. & MILLER, A. D. (1990). RNA Packaging. In Retroviruses: Strategies of Replication, pp. 125-152. Edited by R. Swanstrom & P. K. Vogt. Berlin: Springer-Verlag. MILLER, A. D. (1990). Retrovirus packaging cells. Human Gene Therapy 1, 5-14.

696 Short communication POTTS, B.J. (1990). ~Mini' reverse transcriptase (RT) assay. In Techniques in H1V Research, pp. 102-106. Edited by A. Aldovini & B. D. Walker. New York: Stockton Press. RICHARDSON, J. H, CHILD, L. A. & LEVER, A. M. L. (1993). Packaging of human immunodeficiency virus type 1 RNA requires c/s-acting sequences outside the 5' leader region. Journal of Virology 67, 39974005. ROSEN, C.A., SODROSKI, J.G., CAMPBELL, K, & HASELTtNE, W.A. (1986). Construction of recombinant routine retroviruses that express the human T-cell leukemia virus type III trans-activator genes. Journal of Virology 57, 379-384. SELDEN, R. F. (1987). Transfection using DEAE-dextran. In Current Protocols in Molecular Biology, vol. t, unit 9.2. Edited by R. Ausubel, R. E. Brent, D. D. Kingston, J. G. Moore, J. A. Seidman, J. A. Smith & K. Struhl. New York: Wiley Interscience. SHIMADA, T., FUJtl, H., MITSUYA, H. & NIENHUIS, A.W. (1991). Targeted and highly efficient gene transfer into CD4 + cells by a recombinant human immunodeficiency virus retroviral vector. Journal of Clinical Investigation 88, 1043-1047. WEINBERG, J. B., MATTHWEWS, T. J., CULLEN, B. R. & MALIM, M. H. (1991). Productive human immunodeficiency virus type 1 (HIV-1) infection of non proliferating human monocytes. Journal of Experimental Medicine 174, 1477-1482. (Received 11 July 1994; Accepted 24 October 1994)