Lentiviruses in Gene Therapy Clinical Research

JB Connolly Gene Therapy Advisory Committee, Department of Health, London, UK

Gene therapy vectors derived from lentiviruses offer many generational effects and dissemination of new viruses from potentially unique advantages over more conventional retro- gene therapy patients. Investigators proposing to conduct viral gene delivery systems. Principal amongst these is their this type of research should take due account of the potential ability to provide long-term and stable gene expression and risks for interaction of lentiviral gene therapy vectors with to infect non-dividing cells, such as neurons. However, the other retroviral elements in human subjects, such as Human use of lentiviral-based vectors in the clinic also raises speciﬁc Immunodeﬁciency Virus. In addition, strict quality control for safety and ethical issues. Concerns include the possible replication competent lentiviruses and suitable measure- generation of replication competent lentiviruses during vector ments of lentiviral infectious particle number will be required production, mobilisation of the vector by endogenous retro- before these types of viral vector can proceed to the clinic. viruses in the genomes of patients, insertional mutagenesis Gene Therapy (2002) 9, 1730–1734. doi:10.1038/sj.gt.3301893 leading to cancer, germline alteration resulting in trans-

Keywords: lentivirus; retrovirus; insertional mutagenesis; oncogenesis; replication competent retrovirus (RCR); human immunodeﬁciency virus (HIV)

Introduction vectors, as part of its ‘horizon-scanning’ activity. Those discussions are presented here. Gene therapy vectors derived from lentiviruses (lentivectors) raise specific safety, ethical, and public health anxieties beyond those of more conventional gene Discussion delivery systems. They mirror concerns that were raised in the development of retroviral vectors, but which need What do lentiviral vectors newly offer to the field of to be reconsidered in light of the novel properties of gene therapy? lentivectors and the accumulation of scientific and Unlike other retroviruses, lentivectors do not necessarily clinical knowledge since that time. Possibilities include require cell division for proviral integration and produc- (i) the generation of replication competent lentiviruses tive infection1 (cf. Miller et al2) and so offer a gene (RCLs) during vector production due to recombination delivery method that does not need growth factor of vector plasimds or in vivo due to mobilization of administration to induce cell proliferation in cells vector proviral DNA by infectious retroviruses such as transduced ex vivo. Additionally, lentivectors are suffi- human immunodeficiency virus (HIV); (ii) insertional ciently robust for in vivo administration, potentially mutagenesis due to lentivector proviral DNA integra- facilitating the transduction of cells that would not be tions, potentially leading to oncogenesis; (iii) germline amenable by other gene therapy vectors. When appro- alteration and transmission of transgene to offspring; priately pseudotyped, lentiviral vectors can transduce and (iv) the dissemination of new lentiviruses beyond stem, dendritic, T-cells, and neurons.3 Lentiviruses also the patient. To date, in no country have authorities transduce cells with high efficiency and stability of formally approved the use of lentivectors for clinical transgene expression. Hence, one compelling reason for research. the use of lentivectors is that they may provide a vehicle The Gene Therapy Advisory Committee (GTAC) for the long-term in vivo clinical management of chronic considers the acceptability of all proposals to conduct diseases, such as Parkinson’s disease.4 gene therapy in the United Kingdom based on ethical concerns, practical issues, and scientific merit. As part of Previous experiences with retroviral vectors its responsibilities, GTAC also examines new and Many of the questions on lentivector safety can be emerging trends in gene therapy research and advises framed with reference to previous retroviral studies. ministers on developments in gene therapy and their ‘Gold Standard’ principles for murine oncoretroviral implications. GTAC held a forum in London in March vector manufacture have already been established, as 2002 to consider the benefits and risks of lentiviral have assays for the testing of Replication Competent Retroviruses (RCRs).5 At the time of GTAC’S forum in March 2002 no cancers had been linked to adminis- Correspondence: JB Connolly, Gene Therapy Advisory Committee, Department of Health, 652C Skipton House, 80 London Road, London tration of oncoretroviral vectors or oncoretroviral SE1 6LH, UK vector-transduced cells in gene therapy patients. Received 29 July 2002; accepted 31 July 2002 However, since then, in one patient a case of Lentiviruses in gene therapy JB Connolly 1731 uncontrolled lympho-proliferation has been associated Lentivector design and production with the administration of a gene therapy retroviral Lentivectors are generated using a ‘split-component’ vector [http://agmed.sante.gouv.fr/htm/10/filcoprs/ production system, the overall objective being to make indco.htm]. In addition, a study in mice has indicated each component less and less complete in function, to the that murine bone marrow cells transduced with a point where infectious viral particles can only be Moloney murine leukaemia virus-based retroviral vector produced in the packaging cell and not from the final containing a transgene encoding a truncated form of vector preparation. Typically, producer cell lines are low-affinity nerve growth factor receptor (dLNGFR) transfected with (i) the vector plasmid, containing the caused leukaemia.6 The leukaemic cells contained the therapeutic payload, lentiviral LTRs for host cell integra- vector integrated by the Evi1 gene, a potential mutagenic tion and perhaps the Rev-responsive element (RRE) for ‘hit’ that by itself, or combined with postulated disrup- most efficient vector production; (ii) a plasmid encoding tion of intracellular signalling pathways by dLNGFR, the gag and pol viral structural genes, in order to supply appeared necessary for oncogenic events. In primate reverse transcriptase and integration functions for the studies, lymphomas have occurred from administration therapeutic vector particles; and (iii) plasmids encoding of high titres of RCRs to immunosuppressed monkeys.7,8 envelope proteins for the therapeutic viral particles and In humans, HIV integration may be responsible for some perhaps Rev protein. Other strategies for split vector lymphomas.9–12 systems have been designed whereby genetic elements have been inserted into helper plasmids to transcription- ally partition structural and envelope helper compo- Insertional mutagenesis and oncogenesis nents. None of the starter plasmids are, by themselves, Lentiviruses typically insert into host DNA as a single capable of functioning as autonomous lentiviruses. In non-rearranged copy. It is unknown how many integra- addition, most accessory genes have been removed tion events yield the desired level of transgene expres- during the process. Only the therapeutic vector contains sion. However, up to 20 insertions can be detected in any the packaging signal and thus, in theory, infectious one cell.13 Integration is known to trigger DNA repair particles should contain only the envisaged therapeutic mechanisms. Retroviral infection can also result in the payload. Typically, the env gene encoding the G- generation of non-productive, ‘dead-end’ circular mole- glycoprotein from vesicular stomatitis virus (VSV-G) is cules. These may, in themselves, be toxic to the cell.14–16 used.21 Such pseudotyping of therapeutic vectors alters There appears to be no sequence specificity for integra- the native tropism of lentiviral vectors to allow infection tion by the retroviral agents used in current vector of a broad range of host cell types. The principal concern systems, although the structure of chromatin may play a rests on the chance of recombination between the split- role. Retroviral proviral DNA integration following component sequences at the DNA or RNA level.22 The infection with RCRs clearly has the potential to cause risk of such events can be minimized to apparently oncogenic events in animals, particularly in relation to negligible levels, but can never be eliminated. Apart haematological malignancies. As mentioned, one study from lack of overt sequence homology between plas- in mice has also shown that a replication-deficient mids, codon usage can be further optimized to provide oncoretroviral vector is associated with leukaemia an additional barrier to recombination.23 The net result of development.6 However, pre-clinical rodent studies such engineering is that multiple recombination events may not provide reliable predictions on the insertional would be required to produce infectious RCL. mutagenesis/oncogenesis issue for clinical administra- In theory, plasmids with gag, pol or VSV-G env could be tions of lentivectors or lentivector-transduced cells, as packaged into viral particles. Cross-packaging of the the number of events required for transformation could therapeutic vector by other retroviruses and of other be less than in humans, perhaps due to differences in retroviruses by the vector should occur at demonstrably telomerase regulation.17 In some haematological malig- low levels. Conditional packaging systems are worth nancies, a relatively minor number of mutagenic events, exploring.24 These could include dependence, for exam- perhaps as low as three, is required for complete ple, on synthetic transfer RNA; or packaging constructs transformation. For other cancers, up to eight may be using a heterologous promoter only capable of function necessary.18 Cell types may have differential potential for in producer cell lines. oncogenesis, for example, haematopoetic cells versus cells Self-inactivating (SIN) vectors are deleted within the 30 of solid tumours. Although numerous alterations of the LTR.25 This allows SIN lentivectors to productively infect coding and cis-acting sequences have been made in and integrate into target cell populations, but generation third-generation lentivectors, a central component in all of proviral transcripts is blocked. Hence, SIN vectors of these systems is the LTR. This may itself contribute to may be a sensible precaution, as they are less likely to be the oncogenic potential of some retroviruses.19 mobilized by other RCLs. However, SIN vectors are For the longer term, it may be useful to establish a difficult to manufacture and although phase I trials database of retroviral integration sites from the genomes would be unaffected because of the inherent need for less of gene therapy patients to determine definitively material, larger scale endeavours could be hampered by whether there are preferential integration sites and an insistence on their use. Imposing restrictions on further characterize the safety profile of these vectors. lentivector design flexibility at this stage may negatively These data could be incorporated into the outcome of impact this area of research. long-term monitoring of patients’ health.20 Ultimately, While it seems wise to use the same vector preparation such long-term monitoring of gene therapy patients may in both patients and pre-clinical animal studies, vectors represent the only reliable way to assess the oncogenic with human gene promoters may behave significantly potential of lentivectors in humans. differently in animal hosts. Because lentivectors are

Gene Therapy Lentiviruses in gene therapy JB Connolly 1732 manufactured in eukaryotic cells, the level of contamina- occurring to a significant extent in the long-term culture tion of the viral envelope with producer cell membrane of the producer cells, as can happen using the VSV-G molecules should be considered. Finally, viral particle envelope. Whilst stable cell lines are highly desirable, for number should be estimated by immunoassays based on phase I studies, reliable, validated transiently transfected Gag (for example, p24 Gag for HIV-1, where 1 pg of p24 production systems may prove acceptable. Gag corresponds to 12 000 physical particles), Pol (PERT Standard tests for RCL in vector preparations involve a assays) or quantitative immunostaining methods, if variety of in vitro infectivity-type assays, usually employ- available. These assessments do not distinguish between ing some tissue culture amplification procedures fol- full (viral RNA genome-containing) and empty particles. lowed by measurement of physical, biochemicals, or Full, potentially bio-active, particles may be estimated by biological parameters indicative of RCL propagation.29 quantitative PCR on lentiviral genomes. Both particle These could include p24 Gag measurements, enzyme number and in vitro potency (as measured by integration marker rescue or rescue of cell proliferation. In addition, capacity or by functional assay of transgene product) RCL genomes can be measured by quantitative RCR should be estimated in each lentivector lot, so that a techniques assuming that appropriate primers are avail- consistent product for ex vivo transductions can be able. There are opportunities for RCL generation in vivo, achieved. In the case of in vivo applications, this would but any RCL that arises will have an Env different from allow doses to be (i) accurately administered and (ii) the VSV-G of the lentivectors. For example, if the vector properly standardized across studies. genome is mobilized by HIV-1, it will have HIV-1 Env. Such RCLs only infect human cells and therefore there Replication competent lentiviruses can be difficulties in establishing appropriate animal The generation of RCLs in lentivector gene therapy trials model systems. Immunodeficient rodents, such as SCID- is the greatest safety preoccupation of the field. RCLs hu mice, engrated with human haematolymphoid cells, could occur during manufacture of lentivectors or in may be suitable for checking the transmissability of RCLs patients via mobilization by infectious retroviruses, such in vivo.30 as during parallel infection with HIV-1 or HIV-2. Most Finally, patients should be monitored for vector endogenous retroviral sequences in the human genome mobilization. It is worth remembering that the long-term have large deletions and so cannot produce infectious persistence of proviral DNA could result in RCL virus, but the possibility that retroviral function could be generation at any point in the patient’s future. supplied in trans or that the vector genome could recombine with human endogenous oncoretroviral se- Safety and ethics for HIV-based lentivectors quences merits consideration. Removal of sequence The biology and pathogenesis of HIV is well-studied. homology in split-component systems reduces the risk This offers both advantages and disadvantages in the use during lentivector production. Experience from other of HIV-based lentivectors. The use of HIV-based vectors retroviral vector studies suggests that any RCLs could be in HIV-negative individuals may result in their partial predicted from triple recombination product between seroconversion, although this could be distinguished vector, packaging plasmid, and env-containing plasmids. from an actual HIV-1 infection. In HIV-infected patients, Although other envelopes are being explored,26 VSV-G is no new HIV viral material would be introduced from currently the preferred envelope choice in most lenti- HIV-lentivectors. However, should an HIV-negative vectors. As VSV-G confers broad host cell range, carry- individual become infected with wild-type HIV during over of VSV-G env DNA or mRNA sequences into vector the course of a lentivector study, there could be preparations should be minimized by cogent manufac- ambiguity as to the perceived source of infection, turing and downstream purification strategies. It is although the origin of the infecting virus could be traced worth noting that the presence of a high concentration to an exogenous virus in such a case. This would need to of vector material can cause interference in the detection be given careful consideration prior to research propo- of RCLs. Because all RCLs will have Gag/Pol (most sals. Finally, in HIV patients, mobilization of the likely, system-derived), measurements of polymerase therapeutic vector is possible. However, since any gene activity, such as product- enhanced reverse transcriptase therapy patient could become HIV infected during a trial, (PERT) assays, are to be encouraged.27,28 Aliquots from arguments for and against the exclusion of HIV-positive the viral lot can be chemically disrupted and a standard patients would need to be considered carefully. template for reverse transcriptase supplied to the solution. In the presence of Pol activity, cDNA is Germline transmission generated and a PCR reaction amplifies this sequence. The consequences of potential germline alterations via These assays can be highly sensitive, where the detection retroviral vectors remain unknown. The risk of fertiliza- of 10–100 Pol-positive particles is possible. However, it is tion from a transduced germ cell appears to be vanish- undesirable to generate a lentivirus encoding VSV-G and ingly low, but remains unquantified. Apart from vector Gag/Pol simply to calibrate these assays or to monitor integration, therapeutic gene expression that may be RCL infectivity assays, since such a ‘positive control’ beneficial in one context could be deleterious another, virus could itself be pathogenic. However, alternative such as during embryogenesis. Thalidomide and other approaches should be taken to validate the sensitivity of current cancer chemotherapies raise similar issues for these assays. patient reproduction. Importance rests on the nature of The probability of recombination after transient the patient group. Many trial populations are restricted transfection with three plasmids is much higher than to those beyond the age of reproduction. However, the the probability of recombination in stable producer cell use of barrier contraception should remain a requirement lines containing single-copy cassettes with integrated for entry into gene therapy trails. It is feasible to monitor elements. This is so provided that superinfection is not both sperm and ova for the presence of vector material,

Gene Therapy Lentiviruses in gene therapy JB Connolly 1733 using relatively non-invasive and conventional tech- Due account will have to be taken of the risks of niques. However, a universal requirement for germline interaction with HIV or other retroviral elements in monitoring would seem over-zealous. Instead, pre- research subjects. A risk–benefit analysis for each clinical studies investigating vector biodistribution in lentivector study should be carried out beforehand. The suitable animal models should provide an indication of lack of patient safety data means that risks are likely to the potential for germline contamination. Should pre- weighted more heavily than benefits. Investigators clinical studies reveal vector presence in germline tissue, wishing to conduct lentiviral trials are encouraged to then monitoring for vector in sperm and/or ova may be contact the secretariat of their national gene therapy required. supervisory body for discussions prior to any formal protocol submission. In GTAC’s and the Department of Adverse events Health’s assessments of lentiviral vectors, the ‘Precau- Lentivectors do not raise any new specific issues with tionary Principle’ will be invoked.33 regard to ‘short-term’ adverse event reporting and patient monitoring. Although systems are primarily designed to detect effects from therapeutic payloads, Acknowledgements tumour progression or the appearance of new malignancies may be detected by pre-existing adverse event- I thank Mary Collins, Jonathon Rohll, Boro Dropulic, reporting mechanisms. Long-term monitoring of patients Doug Jolly, Olivier Danos, Anthony Meager, James Neil, is the most robust method for the detection of adverse Luigi Naldini, Sue Kingsman, Norman Nevin, David events caused by insertional mutagenesis. Harrison, Alex Markham, Martin Gore, Terry Hamblin, Andrew Lever, Ian Hart, Jayne Spink, Daniel Gooch and HIV, FIV, and EIAV: do we need more than one type of Margaret Straughan for their contributions to this forum lentivector? and article. There are reasons, other than commercial, to proceed with different types of lentivector.31,32 Each type may offer unique advantages, for example, in tropism and cell-infectivity range. The role of HIV as a human Disclaimer immune system pathogen is well-delineated. Non- This article is a personal account by the author of the primate lentiviruses have no known human pathogeni- discussions during GTAC’s forum on Lentiviruses and city but, importantly, are less well understood. The does not necessarily represent the policy of GTAC or the absence of data does not indicate the absence of effect Department of Health. and pathogenic features of non-primate lentiviruses would probably only be revealed by clinical study. Other factors that may influence vector choice are differential (i) toxicities of viral gene products from and (ii) host immune responses to the various lentiviral vectors. It is References also worth noting that there are non-dividing cells that are not susceptible to infection by certain lentiviruses. 1 Naldini L et al. In vivo delivery and stable transduction of For example, HIV-vectors do not infect a sub-population nondividing cells by a lentiviral vector. Science 1996; 272: 263– of haematopoietic stem cells. Thus, the pursuit of 267. multiple vector types should eventually allow the 2 Miller DG, Adam MA, Miller AD. Gene transfer by retrovirus targeting of a broader spectrum of cells. vectors occurs only in cells that are actively replicating at the time of infection. Mol Cell Biol 1990; 10: 4239–4242. 3 Kay MA, Glorioso JC, Naldini L. Viral vectors for gene therapy: Conclusions the art of turning infectious agents into vehicles of therapeutics. Nat Med 2001; 7: 33–40. The use of lentiviruses raises issues for patient safety, 4 Kordower JH et al. Parkinson’s disease: neurodegeneration public health, and the public perception of gene therapy. prevented by lentiviral vector delivery of GDNF in primate Although GTAC cannot currently offer prescriptive models of Parkinson’s disease. Science 2000; 290: 767–773. advice to researchers on the use of lentivectors, many 5 Chen J, Reeves L, Cornetta K. Safety testing for replication- of the issues discussed herein will form the basis of any competent retrovirus associated with gibbon ape leukemia lentiviral protocol assessment. Investigators will also virus-pseudotyped retroviral vectors. Hum Gene Ther 2001; 12: need to consider the implications of lentivector studies 61–70. for research staff, for example, in the case of needle-stick 6LiZet al. Murine leukemia induced by retroviral gene marking. 296 injury and seroconversion. Given the unknown risks Science 2002; : 497. 7 Donahue RE et al. Helper virus induced T cell lymphoma in associated with lentiviral administration, compelling nonhuman primates after retroviral mediated gene transfer. J reasons justifying the choice of vector, the patient group, Exp Med 1992; 176: 1125–1135. including immune system status, and the approach (ex 8 Vanin EF, Kaloss M, Broscius C, Nienhaus AW. Characterisation vivo versus in vivo) will be required. These will be over of replication competent retroviruses from nonhuman primates and above the requirements for protocols involving with virus-induced T-cell lymphomas and observations regard- vectors with established clinical safety track records. In ing the mechanisms of oncogenesis. J Virol 1994; 68: 4241–4250. addition, appropriate measurements of lentivector in- 9 Ng VL. McGrath MS. The immunology of AIDS-associated fectious particles, stringent quality control for RCLs, and lymphomas. Immunol Rev 1998; 162: 293–298. precautions against and measurements for the mobiliza- 10 Herndier B et al. Acquired immunodeficiency symdrome- tion of other retroviruses in patients will be necessary. associated T cell lymphoma: evidence for human immunodefi-

Gene Therapy Lentiviruses in gene therapy JB Connolly 1734 ciency virus type 1-associated T-cell transformation. Blood 1992; the packaging constructs and endogenous retroviral sequences. J 79: 1768–1774. Virol 1994; 72: 2663–2670. 11 Mack KD et al. HIV insertional characteristics of the proto- 23 Kotsopoulou E et al. A Rev-independent human immunodefi- oncogene c-fes in AIDS associated lymphomagenesis. J Acquir ciency virus type 1 (HIV-1)-based vector that exploits a codon- Immune Defic Syndr Hum Retrovirol 1994; 14: A44. optimised HIV-1 gag-pol Gene. J Virol 2000; 74: 4839–4852. 12 Shiramizu B, Herndier BG, McGrath MS. Identification of a 24 Dull T et al. A third generation lentivirus vector with a common clonal human immunodeficiency virus integration site conditional packaging system. J Virol 1998; 72: 8463–8471. in human immunodeficiency virus-associated lymphomas. 25 Miyoshi H et al. Development of a self-inactivating lentivirus Cancer Res 1994; 54: 2069–2072. vector. J Virol 1998; 72: 8150–8157. 13 O. Danos, personal communication. 26 Mazarakis ND et al. Rabies virus glycoprotein pseudotyping of 14 Pang S et al. High levels of unintegrated HIV-1 DNA in brain lentiviral vectors enables retrograde axonal transport and access tissue of AIDS dementia patients. Nature 1990; 343: 85–89. to the nervous system after peripheral delivery. Hum Mol Genet 15 Pauza CD, Galindo JE, Richman DD. Reinfection results in 2001; 10: 2109–2121. accumulation of unintegrated viral DNA in cytopathic and 27 Boni J, Pyra H, Schupbach J. Sensitive detection and quantifica- persistent human immunodeficiency virus type 1 infection in tion of particle-associated reverse transcriptase in plasma of CEM cells. J Exp Med 1990; 172: 1035–1042. HIV-infected individuals by the product-enhanced reverse 16 Kameoka M et al. High susceptiblity of U937-dervied subclones transcriptase (PERT) assay. J Med Virol 1996; 49: 23–28. to human immunodeficiency virus type 1 infection correlates 28 Martin-Rendon E et al. New methods to titrate eiav-based with accumulation of unintegrated circular viral DNA. Virus lentiviral vectors. Mol Ther 2002; 5: 566–570. Genes 1996; 12: 117–129. 29 Rohll JB et al. Design, production, safety, evaluation and clinical 17 Kim NW et al. Specific association of human telomerase activity applications of nonprimate lentiviral vectors. Methods Enzymol with immortal cells and cancer. Science 2002; 266: 2011–2015. 2002; 346: 466–500. 18 T. Hamblin, personal communication. 30 Bonyhadi ML, Kaneshima H. The SCID-hu mouse: an in vivo 19 Mukhopadhyay R, Medina D, Butel JS. Expression of the mouse model for HIV-1 infection in humans. Mol Med Today 1997; 3: mammary tumor virus long terminal repeat open reading frame 246–253. promotes tumorigenic potential of hyperplastic mouse mam- 31 Poeschia EM, Wong-Staal F, Looney DJ. Efficient transduction of mary epithelial cells. Virology 1995; 211: 74–93. non-dividing human cells by feline immunodeficiency virus 20 Nevin NC, Spink J. Gene Therapy Advisory Committee: long- lentiviral vectors. Nature Med 1998; 4: 354–357. term monitoring of patients participating in gene therapy. Hum 32 Olsen JC. Gene transfer vectors derived from equine infectious Gene Ther 2000; 11: 1253–1255. anaemia virus. Gene Ther 1998; 5: 1481–1487. 21 Burns JC et al. Vesicular stomatitis virus G glycoprotein 33 Lord Phillips of Worth Matravers. Return to an Order of the pseudotyped retroviral vectors: concentration to very high titre Honourable House of Commons dated October 2000 for the and efficient gene transfer into mammalian and non-mammalian Report, evidence and supporting papers of the Inquiry into the cells. Proc Natl Acad Sci USA 1993; 90: 8033–8037. emergence and identification of Bovine Spongiform Encephalo- 22 Chong H, Starkey W, Vile RG. A replication competent pathy (BSE) and variant Creutzfeldt-Jakob Disease (vCJD) and retrovirus arising from a split-function packaging cell line was the action taken in response to it up to 20 March 1996. HM generated by recombination events between the vector, one of Stationary Office, UK. [www.bse.org.uk].

Gene Therapy