Retroviral Stem Cell Gene Therapy

Concise Review Retroviral Stem Cell Gene Therapy MENZO HAVENGA, a PETER HOOGERBRUGGE, b,c DINKO VALERIO, a,b HELMUTH H.G. VAN ES a,b a Gene Therapy Section of the Department of Medical Biochemistry, Medical Faculty, LeidenUniversity, The Netherlands; b IntroGene BV, Leiden, The Netherlands; c Department of Paediatrics, Sophia Children s Hospital, Rotterdam, The Netherlands Key Words. Hemopoietic stem cell Retrovirus Gene therapy Murine studies Primate studies Vector Transduction ABSTRACT Long-term in vivo gene transfer studies in mice have shown that recombinant murine retroviruses are able to infect murine hemopoietic stem cells with high efficiency. Taken together the results indicated that the proviral structure was present at high frequency in circulating hemopoietic cells resulting in significant expression levels. Because of the success of these murine studies, it was believed that gene therapy would soon be applicable to treat a wide variety of congenital or acquired human diseases associated with the hemopoietic system. However, results from gene transfer studies in nonhuman primates and first human clinical trails have indicated that murine retrovirus infection of primate hemopoietic stem cells is inefficient. Although there are essential differences between the murine and primate gene therapy studies with respect to the recombinant viruses and transduction protocols used, these differences cannot solely account for the differences observed in infection efficiency. Therefore, in recent years effort has been spent on the identification of factors limiting retroviral transduction of primate hemopoietic stem cells. Increasing knowledge concerning hemopoiesis and retroviral infection has helped in identifying a number of limiting factors. Novel transduction strategies and tools have been generated which attempt to circumvent these limiting factors. These factors as well as the strategies that showed increased retroviral infection of primate hemopoietic stem cells will be discussed. Stem Cells 1997;15:162-179 INTRODUCTION Since 1980 when the feasibility of genetic modification of primitive hemopoietic mouse cells was demonstrated [1], investigators have been developing gene therapies for the treatment of a number of congenital and acquired human diseases. The pluripotent hemopoietic stem cell (PHSC) and its self-renewing capacity forms an ideal candidate for gene therapy because many of its progenitors are involved in human disorders. Although initial studies on transduction of hemopoietic cells focused on using the calcium phosphate precipitation technique [2], investigators soon switched to efficient gene delivery vehicles already present in nature retroviruses. In 1983 the first successful gene transfer with a murine leukemia retrovirus of mouse hemopoietic cells was reported [3]. Since then a number of important scientific hurdles have been overcome such as improvement of the transcription machinery of the murine retrovirus to increase expression of the transgene in human cells, the development of efficient retroviral packaging cell lines, the development of specific and sensitive safety tests, and increased knowledge on hemopoiesis. This progress has culminated in highly efficient retroviral transduction of murine PHSC showing that, upon repopulation, sustained long-term expression from introduced cdnas is detectable. By extrapolating the successes of these mouse studies it was believed that the treatment of patients would soon be a reality. However, once retroviral primate PHSC gene therapy experiments commenced, it became apparent that such an extrapolation was not justified. A better understanding of primate stem cell proliferation as well as the retroviral infection mechanism was clearly needed to enable the development of human stem cell gene therapy. Here we will present a survey of retroviral experimental stem cell gene therapy studies performed in mice and nonhuman primates and the data on clinical trails in human patients to date. Finally, putative factors limiting the retroviral transduction of the human stem cell and strategies aimed at increasing the retroviral transduction efficiency will be presented. TARGET DISEASES FOR STEM CELL GENE THERAPY Human congenital diseases which are manifested predominantly in one or more of the blood lineages are, in principle, target diseases for stem cell gene therapy, since all Correspondence: Dr. Helmuth H.G. van Es, IntroGene B.V., P.O. Box 2048, 2301 CA Leiden, The Netherlands. Accepted for publication January 6, 1997. AlphaMed Press 1066-5099/97/$5.00/0 STEM CELLS 1997;15:162-179

163 Retroviral Stem Cell Gene Therapy blood cells are derived from a common ancestor, the PHSC. There are, however, some limitations. First, the precise genetic defect causing the disease must be known. Second, the defect should not be dominant. In general, those diseases that can been treated by allogeneic bone marrow transplantation are candidates for stem cell gene therapy (Table 1). Ideally, the aberrant gene in the PHSC would be replaced by a correct copy, a process known as homologous recombination. Homologous recombination does occur in nature in mammalian cells but at a frequency of approximately one in one million which at present is too low for gene therapy purposes [4]. Therefore, stem cell gene therapy has focused on the addition of correct copies of a gene in the host genome. One possible disadvantage of this strategy is the potential transformation of proto-oncogenes due to insertional mutagenesis. Moreover, the vectors need to be able to overcome the silencing effect or several copies should be introduced to increase the chance of introducing the gene in a preferred site. RETROVIRUS-MEDIATED GENE DELIVERY Gene delivery vehicles for gene therapy purposes can be either viral or nonviral. A wide variety of viral delivery vectors is under investigation such as retrovirus, adenovirus, adeno-associated virus, herpes simplex virus, vaccinia virus, poliovirus, baculovirus and Sindbis virus. Nonviral delivery is pursued using either naked DNA or liposome-complexed DNA. There are two important criteria on which the choice of a gene delivery vehicle is based. First, the accessibility of the target tissue to be manipulated dictates whether host cell transduction can be performed in vivo or ex vivo and, secondly, the desired persistence of transgene expression, which can be either sustained or temporary. For stem cell gene therapy, sustained expression of the transgene is a prerequisite for its success and therefore only gene delivery vehicles that assure stable integration into the host genomic DNA are relevant. Consequently, retroviruses and possibly adeno-associated viral vectors are useful for stem cell gene therapy. Here, only retroviral vectors will be discussed. Murine leukemia retroviruses have been the vectors of choice since the start of the field of gene therapy. A general overview of the structure of murine leukemia retroviruses and their life cycle is depicted in Figure 1. Upon infection of the host cell, the viral RNA is released into the cytoplasm where it is converted into DNA by the viral enzyme reverse transcripts. Because the retroviral DNA cannot pass the nuclear envelope, it can only integrate in the host genome of cells that are going through mitosis [5]. Integration is facilitated by the viral enzyme integrase and the long terminal repeats present at both ends of the proviral structure. If retroviral integration occurs in a preferred site, the retroviral genes gag (core proteins), pol (reverse transcriptase and integrase), Table 1. Inherited human disorders which are candidates for stem cell gene therapy. The list includes only diseases associated with the hemopoietic system in which the single gene causing the disease has been cloned. The selection was based on the following sources: [128-130]. BMT: bone marrow transplantation; ND: not done. Disorders Deficiency Effect BMT Hurler s (MPS-1) α-l-iduronidase + Hurler-Scheie (MPS-1) α-l-iduronidase + Scheie (MPS-1) α-l-iduronidase + Hunter (MPS-II-XR severe) Iduronate sulfatase ± Hunter (MPS-II-XR mild) Iduronate sulfatase ± Hunter (MPS-II-AR) Iduronate sulfatase ± Sanfilippo s (MPS-IIIA) N-sulfatase - Sanfilippo s (MPS-IIIB) N-acetyl-α-D-glucosaminidase - Sanfilippo s (MPS-IIIC) α-glucosaminide-n-acetyltransferase - Sanfilippo s (MPS-IIID) N-acetyl-α-D-glucosaminide-6-sulphatase - Morquio s (MPS-IVA) Galactosamine-6-sulfate sulfatase ± Morquio s (MPS-IVB) β-galactosidase ± Maroteaux-Lamy (MPS-VI) N-acetylgalactosamine-4-sulphatase + Sly (MPS-VII) β-glucuronidase ND Gaucher β-glucocerebrosidase + Farber Acid ceramidase - Niemann-Pick Acid sphingomyelinase - Krabbe Galactocerebroside β-galactosidase ± Metachromatic leucodystrophy Arylsulfatase A ± Fabry α-galactosidase A ND Severe combined Adenosine deaminase + immunodeficiency (SCID) X-linked SCID Gamma-c receptor + XLA Bruton s tyrosine kinase + JAK-3 deficiency Jak-3 + Aspartylglycoaminuria Aspartylglucosaminidase ND Fucosidoses α-l-fucosidase ± Guibaud-Vainsel syndrome Carbonic anhydrase II ND Thallassemias α-, β-globin + G6PD deficiency Glucose 6-phosphate dehydrogenase ND PK deficiency Pyruvate kinase L-type ND Erythropoietic porphyria Uroporphyrinogen III synthase ND and env (envelope protein determining the host cell range or tropism of a retrovirus) are transcribed. In order to obtain infectious retroviral particles, full-length viral RNA, containing the packaging signal Ψ, is complexed with viral core proteins. These RNA-protein complexes are then released from the infected cells by budding, carrying the envelope molecules with it. For the production of infectious, replication-defective recombinant retrovirus particles carrying a gene of interest, packaging cells have been constructed (Fig. 2). Packaging cells are mammalian cell lines, genetically modified in such a

Havenga, Hoogerbrugge, Valerio et al. 164 A 5'-LTR U3 R U5 Promoter Enhancer SD Ψ P(-) SA gag pol env P(+) 3'-LTR (A)n U3 R U5 producer cell LTR Ψ + cdna LTR vector Ψ - Prom. gag pol pa Ψ - Prom. env pa Ψ + RNA transcript packaging Ψ - RNA transcripts packaging defective viral proteins infectious virus B reverse transcription integration Infection Ψ + LTR gag pol env LTR dsdna ssrna entrance Adhesion target cell reverse transcription receptor binding integration RNA transcript protein Ψ + LTR cdna LTR ssrna dsdna entrance Figure 1. A) Moloney murine leukemia proviral structure. Between the two long terminal repeats (LTRs) are the coding regions for the viral gag, pol, and env genes. Also shown are the splice donor (SD), splice acceptor (SA), packaging signal (Ψ), and primer binding sites (P[+] and P[-]). The LTRs are subdivided in three domains: U3, R, U5, which contain the transcriptional enhancer, promoter and poly-adenylation signal (AATAAA). Initiation of transcription and the beginning of the poly(a) tract is denoted by a horizontal arrow and (A)n, respectively. B) Wild type Moloney murine leukemia virus life-cycle. After binding to and entry into the target cell the viral (+) RNA genome is reverse transcribed into double-stranded DNA. The DNA is transported to the nucleus and integrates into the host genomic DNA, a process facilitated by the viral enzyme, integrase. The genes of the integrated retroviral DNA, or provirus, are transcribed and translated. The viral RNA is encapsulated with various viral proteins. The viral RNA-protein complexes subsequently bud from the cells. way that they express the viral proteins gag, pol, and env without producing infectious particles since the encoding mrnas do not harbor the packaging signal Ψ. Upon transfection with a retroviral construct carrying the packaging signal Ψ and a gene of interest, recombinant infectious particles are generated. Since the infectious particles lack the mrnas encoding reverse transcriptase and integrase, these virus preparations are, in principle, free of replication competent retrovirus (RCR). RCR is a major safety concern. Pathogenicity of replication competent amphotropic murine leukemia viruses in nonhuman primates was initially tested by injecting large amounts of amphotropic RCR intravenously or by implantation of virus-producing Figure 2. A packaging cell for recombinant infectious, replication defective retroviral particles. The viral genes necessary for packaging and integration are located on two plasmids which are introduced into mammalian cells: 1) a plasmid containing the viral gag and pol genes and 2) a plasmid containing the viral env gene. Introduction of a recombinant retroviral construct containing a gene of interest and a retroviral RNA packaging signal (Ψ) yields full length Ψ + RNA molecules which are packaged. Since these recombinant RNA molecules lack the gag, pol, and env genes, replication in infected target cells is impossible. autologous fibroblasts [6, 7]. Although these studies demonstrated rapid clearance of the murine retrovirus by the rhesus monkey sera, a later study demonstrated RCR induced T cell lymphomas in nonhuman primates [8]. By removing all overlapping sequences between the packaging constructs and the construct containing the gene of interest, the chance of generating RCR can be minimized [9-12]. STEM CELL GENE THERAPY STUDIES IN MICE To demonstrate the principle of gene transfer into PHSC, many researchers have used mice as an in vivo model. A schematic presentation of a bone marrow transplantation protocol used to assay retroviral transduction into murine PHSC is given in Figure 3. The protocol consists of pretreatment of

165 Retroviral Stem Cell Gene Therapy Bone marrow Donor Stimulate in vivo stem cell cycling by treatment of Ara-C, Velban, or 5-FU prior to bone marrow harvesting Retroviral transduction of murine bone marrow 1 st recipient Reconstitution from donor stem cells provides radioprotection Analysis of hemopoietic tissues to determine transgene presence and expression in total murine bone marrow 2 nd recipient Reconstitution from pluripotent stem cells (PHSC) provides radioprotection Analysis of hemopoietic tissues to determine transgene presence and expression in murine pluripotent stem cells (PHSC) Figure 3. Experimental design to study retroviral transduction into murine pluripotent hemopoietic stem cells. Myelosuppressive compounds such as 5- fluorouracil (5-FU) are administered to donor mice several days prior to bone marrow harvesting to trigger in vivo stem cell cycling. The murine bone marrow is then harvested and cocultured with a retrovirus producer cell line in the presence of growth factors. The transduced bone marrow is subsequently injected into lethally irradiated syngeneic recipient mice. Upon full reconstitution of the first recipients the bone marrow is retransplanted into secondary lethally irradiated syngeneic recipient mice. Southern analysis of genomic DNA extracted from several tissues and biochemical assays to determine expression indicate whether retroviral transduction of PHSCs was successful. mice with 5-fluorouracil (5-FU), cytosine arabinoside (Ara C) or Velban for two to four days because in vivo myelosuppression was shown to stimulate in vivo stem cell proliferation [13]. The harvested murine bone marrow cells are cocultured with retroviral producer cell lines, after which the nonadherent cells are isolated and infused in the tail vein of lethally irradiated recipient mice. Often, bone marrow from the primary recipients was harvested and injected into lethally irradiated secondary recipients to confirm PHSC transduction [13]. Most stem cell gene therapy studies in nonhuman primates and humans have been performed with supernatant of amphotropic retroviral producer cell lines, therefore it is relevant to categorize the data of the mouse studies according to: A) the retrovirus tropism used and B) the transduction protocol used. Table 2A is a compilation of long-term gene transfer studies in mice. These studies are identical with respect to virus tropism and transduction protocol. From the data it can be concluded that with ecotropic retroviral vectors and a cocultivation transduction protocol, murine PHSCs can be transduced efficiently (>1.0 provirus copies/cell on average) resulting in long-term persistence of the proviral structure and relatively high levels of expression of the transgene in primary and secondary recipients. Since the use of amphotropic retrovirus was considered a better model for human gene Table 2. Compilation of results from long-term in vivo studies in mice using either ecotropic (A) or amphotropic (B) virus in combination with a cocultivation protocol A. Ecotropic virus Primary recipient Transgene Period Provirus Protein Reference ADA 4 months <1.0 c/c >endogenous [131] ADA 5 months >1.0 c/c >endogenous [132] ADA 6 months <0.6 c/c <endogenous [133] ADA 14 months <0.7 c/c >endogenous [134] ADA 4 months <1.0 c/c hada+ [134] ADA 6 months >3.0 c/c <endogenous [135] β-globin 5 months <2.0 c/c <endogenous [136] β-globin 6 months <0.5 c/c <endogenous [137] β-globin 9 months <0.4 c/c <endogenous [138] GATA-1 10 months <0.5 c/c <endogenous [139] GC 7 months <2.0 c/c >endogenous [140] GC 3 months <1.0 c/c <endogenous [141] GC 8 months <2.0 c/c >endogenous [108] GC 6 months <3.0 c/c <endogenous [142] MDR 4 months <1.0 c/c not determined [143] MDR 8 months PCR + FACS + [144] B. Amphotropic virus Primary recipient Transgene Period Provirus Protein Reference GC 8 months <0.6 c/c Immunohistochemistry + [15] GC 6 months PCR + RNA + [14] ADA 6 months <0.05 c/c <endogenous [16] ADA 6 months <0.2 c/c ADA activity + [17] ADA 1.5 months Southern + ADA activity + [18] The number of studies included in these tables is limited to studies in which a retrovirus carrying a potentially therapeutic gene was used. ADA: adenosine deaminase; GATA-1: zinc-finger DNA-binding transcription factor; GC: glucocerebrosidase; MDR: multidrug resistance; c/c: the number of proviral copies per cell as established by Southern analysis. Expression of the introduced transgene is in most studies measured relative to the endogenous murine protein.

Havenga, Hoogerbrugge, Valerio et al. 166 therapy and because human cells including PHSCs are outside the host range of ecotropic retroviruses, several research groups have used cocultivation with amphotropic retroviral producer cells to infect murine PHSCs (Table 2B). These studies indicated that initial transduction efficiencies, as established by provirus positive colony forming units of the granulocyte-myeloid lineage (CFU-GM), long-term culture initiating cells (LTC-IC), and day 12 colonies isolated from the spleen of irradiated and transplanted recipients (MRA- CFUs) were similar using either ecotropic or amphotropic retroviruses [14-20]. However, in primary recipients several months after transplantation the frequency of PHSCs harboring the transgene was, in general, 10-fold lower as compared to ecotropic virus. In addition, pretreatment of mice with 5-FU did not increase the amphotropic retroviral transduction efficiency in contrast to the combination of 5-FU treatment with ecotropic virus [13, 17]. A small number of studies have been performed in which the transduction efficiency of ecotropic and amphotropic viruses was actually compared directly. One of these studies indicated that infection with amphotropic virus resulted in expression and thus transgene presence for less than eight weeks, whereas infection with ecotropic virus resulted in expression for more than 44 weeks after transplantation [21]. In a similar study, ecotropic virus was shown to be approximately 50-fold more efficient in transducing murine PHSCs as compared to amphotropic virus [22]. The effect of cocultivation versus supernatant transduction has also been compared with either ecotropic or amphotropic retroviral vectors. These studies showed that cocultivation, in general, resulted in an approximately fourfold higher transduction efficiency of murine PHSCs as compared to supernatant infection [13]. In conclusion these data suggest that infection of murine PHSC with amphotropic virus is not as efficient as with ecotropic virus. Ecotropic and amphotropic retroviruses differ in the receptor that is employed for virus entry, and the observed differences might simply be explained by significant differences in receptor expression levels in PHSCs. Indeed in a comparative study on mrna levels in mouse PHSCs (lin-c-kit bright ), it was demonstrated that ecotropic receptor (mcat1) mrna levels in these cells are high whereas amphotropic receptor (GLVR2) mrna levels were undetectable by reverse transcriptase polymerase chain reaction (PCR) [22]. In addition to the use of a different receptor, ecotropic and amphotropic viruses differ in their postadsorption pathways (see below). It was also shown that upon culturing the murine PHSC in interleukin 3 (IL-3), IL-6 and stem cell factor (SCF), mcat1 receptor RNA levels increased in murine lin-c-kit high cells but GLVR2 receptor RNA levels did not [22]. Although no increase in expression levels of amphotropic receptor mrna was observed in this study, increased infection with amphotropic vector of murine PHSC has been reported after addition of cytokines [14-20, 23]. This suggests that ex vivo stem cell proliferation plays an important role in the retroviral transduction process of PHSCs. Finally cocultivation infection is superior to supernatant infection of murine PHSC. This suggests that sufficient binding of retrovirus to target cells is another limiting factor since we believe that cocultivation decreases the distance between a retroviral particle and the target cell (see below). GENE THERAPY STUDIES IN NONHUMAN PRIMATES AND HUMANS Many researchers have applied variations of the transduction protocols found to be successful in mice, including in vivo pretreatment of 5-FU and cocultivation, to PHSCs from large animals. Since ecotropic viruses are unable to infect primate PHSCs, other tropisms including amphotropic and Gibbon ape leukemia (GaLV) based viruses were used [24]. The effect of in vivo bone marrow priming with 5-FU has been tested in rhesus monkeys. In initial studies using CFU-GM to test transduction efficiency, increased numbers of transgenepositive CFU-GM were found after injecting 5-FU less than seven days prior to bone marrow harvesting [25]. Long-term in vivo studies in nonhuman primates showed that prestimulation of bone marrow with 5-FU did not result in a significant increase in the retroviral transduction efficiency of primate PHSCs [26-28]. To investigate whether cocultivation increased retroviral transduction in primate progenitors, clonogenic assays such as CFUs were initially used. Cocultivation of amphotropic producer cells with canine bone marrow cells [29, 30] resulted in approximately 40% provirus positive CFU-GM as established by drug resistance. In contrast, only 5% drug resistant CFU-GM were scored using supernatant transduction [31]. These initial experiments indicated that, similar to the murine situation, cocultivation was superior to supernatant transduction. To assay for transduction of PHSCs, both protocols were tested in long-term in vivo studies in nonhuman primates. In the first studies, only the nonadherent cells were reinfused in lethally irradiated recipients after cocultivation, which resulted in impaired repopulation of the hemopoietic system. Gene transfer demonstrated greater than 0.1 provirus copies per cell in the bone marrow resulting in less than 0.01% human adenosine deaminase (ADA) activity in the peripheral blood as compared to endogenous monkey ADA levels. Parallel studies in which the rhesus bone marrow graft was transduced using a supernatant transduction protocol showed that reconstitution of the hemopoietic system was within normal limits. Gene transfer in these monkeys was 0.2%-0.5% of endogenous monkey ADA levels in the peripheral blood [32-34]. In another study, the repopulation

167 Retroviral Stem Cell Gene Therapy ability of the ex vivo manipulated graft was investigated using cocultivation transduction with producer cells genetically modified to produce gibbon IL-3 and human IL-6. Upon transplantation GM-CSF was administered to enhance hemopoiesis. Despite these precautions, reconstitution failed [35]. Of six rhesus monkeys transplanted in this study, three were found positive for the transgene for the duration of the study (<99 days). The impaired repopulation using the cocultivation protocol in rhesus monkeys may have been caused by the loss of stem cells either due to the adherence of stem cells to the retroviral producer cells or due to the loss of selfrenewal capacity during ex vivo manipulation. To overcome the loss of stem cells during cocultivation, two different approaches were reported. Bodine and coworkers harvested stem cells after cocultivation transduction by including mild trypsinization and showed that both transplanted rhesus monkeys fully repopulated. The ADA transgene, however, was never detected in either monkey. In parallel, three monkeys received bone marrow which was cocultured with a murine stromal cell line that produces membrane bound SCF. Transduction was performed by addition of retroviral supernatant. The two monkeys analyzed 11 months after transplantation were positive for the ADA provirus, and human ADA activity could be detected at 3% of endogenous monkey ADA activity [26]. Other researchers performed cocultivation transduction with irradiated producer cells to prevent further growth of these cells and harvested the adherent cells by trypsinization. As a consequence, the irradiated producer cells were coinjected with the stem cell graft. The presence of irradiated producer cells had no adverse effects, and the hemopoietic reconstitution in these animals was normal. Gene transfer, analyzed up to four and a half years after transplantation, indicated 0.1% provirus positive cells in lymph nodes as established by PCR [28, 36, 37]. Due to the limited number of primate PHSC transduction studies where cocultivation and supernatant protocols were compared directly, a clear conclusion on whether cocultivation is superior to supernatant infection cannot be drawn. In Figure 4, a schematic presentation of two essentially different supernatant transduction protocols for gene transfer into human PHSC is shown. These protocols are named short-term (Fig. 4A) and long-term (Fig. 4B) based on the length of the ex vivo period of the stem cell graft. The short-term protocol aims at limiting the ex vivo period of the stem cell graft as much as possible to ensure the maintenance of the self-renewal capacity of PHSCs. As a consequence, only those PHSCs that are in cycle at the moment of transduction will be transduced. The long-term protocol aims at ex vivo stem cell cycling without inducing differentiation during 21 days in culture. Increased cycling should result in increased numbers of vector-positive PHSCs. In the short-term supernatant infection protocol, CD34 + cells are isolated from total bone marrow. CD34 + cells are enriched for PHSCs and are capable of in vivo A B Reinfuse transducted cells into patient Isolate CD34 + cells from bone marrow or mobilized peripheral blood Reinfuse transducted cells into patient Isolate mononuclear cells from bone marrow or mobilized peripheral blood Day 0: Seed cells in medium containing retrovirus supplemented with human growth factors IL-3, IL-6, and SCF. Day 1: Refresh retroviral supernatant daily for 2, 3, or 4 days. Day x: Harvest nonadherent cells and adherent cells by trypsinization. Day 0: Seed mononuclear cells in medium containing virus. Day 7: Discard medium of cell culture and add non-virus containing medium. Day 8: Substitute half of the non-virus medium with medium containing virus. Day 14: Discard medium of cell culture and add non-virus containing medium. Day 15: Substitute half of the non-virus medium with medium containing virus. Day 21: Harvest adherent cell layer by trypsinization. Figure 4. Schematic presentation of the two essentially different retroviral supernatant transduction protocols used for the infection of primate and human PHSCs. A) Short-term supernatant infection protocol. CD34 + hemopoietic cells isolated from bone marrow, cord or mobilized peripheral blood are cultured for several hours or days in retrovirus-containing medium in the presence of growth factors prior to reinfusion into the recipient. Transduction of PHSC, albeit low, is usually only observed after myeloablative treatment. B) Longterm supernatant infection protocol. Bone marrow devoid of granulocytes and erythrocytes (postficoll) is seeded in retrovirus-containing medium in the presence of growth factors. The cells are cultured for 21 days during which the retrovirus supernatant is refreshed three times. After 21 days in culture, the adherent cells are reinfused into the recipient. Transduction of PHSC, albeit low, is observed with or without myeloablative treatment.

Havenga, Hoogerbrugge, Valerio et al. 168 repopulation of both myelosuppressed humans and monkeys [38, 39]. Using the short-term protocol, Xu et al. demonstrated efficient gene transfer into two rhesus monkeys. Transduction of rhesus CD34 + cells in the presence of IL-3, IL-6, and SCF for four days resulted in provirus positive granulocytes (0.1%) and B lymphocytes (14%) for more than one year after transplantation. Using this protocol an essential parameter was investigated, i.e., the requirement for myeloablation to obtain sufficient grafting levels of genetically modified PHSCs. Two rhesus monkeys receiving supernatant-transduced bone marrow without myeloablative treatment were found negative for the transgene within four months after transplantation [40]. These findings are reminiscent of a clinical study we conducted involving three nonmyeloablated ADA deficient patients in whom short-term transduced bone marrow was found positive for the ADA provirus until six to eight months after transplantation, but than became negative [41]. These findings suggested that myeloablative treatment is essential to prevent the incoming transduced stem cells from being outnumbered by endogenous stem cells. In contrast to the study described above, transgene persistence was reported for more than one year in three neonate ADA patients whose umbilical cord blood was reinfused without myeloablation after using a short-term transduction protocol [42]. The retroviral transduction protocol consisted of three exposures (24 h each) to the retroviral supernatant in the presence of IL-3, IL-6 and SCF. Semiquantitative PCR analysis of bone marrow CD34 + cells and clonogenic assays indicated a retroviral transduction efficiency of 1%-4%. This relatively high frequency contrasted with the frequency of vector-containing cells of the peripheral blood which ranged between 0.03%-0.001%. The authors suggested that although primitive progenitor cells may engraft without myeloablative therapy, they fail to undergo complete maturation in vivo. Upon decreasing the dose of polyethylene glycol-ada, the recombinant protein that all three patients received directly after birth, the number of vector-positive T lymphocytes increased. This suggests a selective survival advantage in vivo for transduced, corrected T cells as seen in allogeneic bone marrow transplantation studies [43]. A major difference between the two clinical ADA studies described above which might explain the observed difference in transgene persistence is the different source of PHSCs used bone marrow versus umbilical cord blood. Umbilical cord blood has a higher ex vivo proliferative capacity and engraftment potential as compared to bone marrow [44, 45]. Brenner and coworkers reported on the persistence of the neomycin transgene, as measured by drug resistant progenitors, in two patients 18 months after transplantation. A single, six h, exposure to the retroviral supernatant without the addition of growth factors was performed and the transduced cells were reinfused in myeloablated patients [46]. In yet another study both mobilized peripheral blood and bone marrow were used for gene marking using two distinguishable retroviral vectors. With a short-term supernatant transduction protocol, three out of nine patients showed persistence of the neomycin transgene for greater than 18 months after transplantation at a frequency of 0.1% to 0.01% provirus positive cells in the peripheral blood [47]. The same authors reported on a study in which both mobilized peripheral blood and growth factor primed bone marrow CD34 + cells from rhesus monkeys were transduced using a 96 h supernatant transduction protocol which resulted in 5% provirus positive circulating cells up to one year after transplantation [48]. The relatively high transduction efficiency was attributed to the in vivo treatment of GM-CSF and SCF which resulted in a threefold expansion of primate PHSC in bone marrow. In the long-term supernatant infection protocol postficoll purified bone marrow cells are seeded in culture flasks in retroviral supernatant. On day 7 and 14 of incubation the cultures are demidepopulated and the remaining cells are fed with nonvirus-containing medium. On day 8 and 15, one-half of the cell culture supernatant is replaced by fresh virus supernatant. Using this protocol for the transduction of canine PHSCs, approximately 10% transduction was reported as established by drug resistant CFU-GM three months after transplantation. This relatively high transduction efficiency declined to 1%, 21 months after transplantation. Since comparative results were obtained with myeloablated and nonmyeloablated dogs, these authors suggested that marrow conditioning is not required for the retention of genetically marked cells in combination with this specific protocol [49, 50]. This long-term transduction protocol is currently being tested in a gene marking study in myeloma patients that underwent therapeutic marrow ablation. Preliminary results using PCR indicate that the transgene was present at a frequency of 17% (bone marrow) and 1% (blood cells) in two patients 12 months after transplantation [51]. Compared to allogeneic bone marrow transplantation, efficient stem cell gene therapy would in principal allow one to omit severe myeloablative treatment. Clearly from the patient s point of view, a gene therapy protocol which has no myeloablation included is preferred and therefore should deserve full attention by those developing stem cell gene therapies. FACTORS LIMITING RETROVIRAL TRANSDUCTION OF PHSC The long-term gene marking studies in dogs, nonhuman primates and humans described above have demonstrated retroviral transduction of PHSC resulting in multilineage

169 Retroviral Stem Cell Gene Therapy transgene presence. However, a retroviral transduction efficiency leading to approximately 0.01%-5% provirus positive circulating cells is too low to expect clinical improvement for the majority of human diseases associated with the hemopoietic system. Therefore, a key question is how to increase the retroviral transduction efficiency into primate PHSC. This question is not easy to solve due to the nature of PHSCs long-term repopulating capacity of hemopoiesis. This implies that any study on retroviral transduction involves long-term follow-up after reinfusion of transduced cells for at least one year in the case of primates and humans. Obviously, this poses practical problems for the development of improved PHSC retroviral transduction methods. Therefore, assays which predict the characteristics of PHSCs are important tools in the development of stem cell gene therapies. Common in vitro assays such as CFUs and LTC-ICs do not assay for true PHSCs since there is a clear discrepancy between the retroviral transduction measured by these assays and the long-term in vivo studies. Dick and coworkers ambiguously showed that CFUs and LTC-ICs do not represent the cells capable of long-term reconstitution. With a retrovirus carrying a neomycin resistance gene, transduction of human cord blood-derived CFUs and LTC-ICs was shown to be as high as 80% to 70%, respectively. The transduced human cord blood cells were subsequently used to engraft severe combined immunodeficient (SCID) mice as a model for primitive cells (see below). The transplanted cells repopulated the mice, but were negative for the neomycin transgene, confirming poor PHSC transduction studies in man and monkey [52]. Alternative in vitro assays to monitor retroviral transduction are based on flow cytometric analysis using cell surface antigens that are normally not expressed on primitive hemopoietic cells. Examples of such genes are murine heatstable antigen [53], murine CD2 [54], the human nerve growth factor receptor (NGFR) [55], and the human homologue of heat-stable antigen, CD24, which is expressed on the surface of human B cells only [56]. Detection of expression of the introduced genes in conjunction with other cell surface antigens predicts in which population of hemopoietic cells retroviral infection is most efficient. However, these experiments should be evaluated with caution. In an experiment in which the NGFR marker gene was used to assess transduction efficiency, high NGFR expression levels were already found one hour after transduction as measured by flow cytometry. Moreover, The NGFR transgene could not be detected in unsorted cells, which at the time of genomic DNA extraction were 50% positive for the NGFR protein. The authors speculated that since the retroviral particles were budding from packaging cells which highly express NGFR protein, the protein might be copackaged, resulting in false positive flow cytometry signals [57]. Nevertheless, using these assays, conditions expected to enhance retroviral infection into human hemopoietic cells can be screened for. Putative factors limiting retroviral transduction of primate PHSCs are extracellular retroviral half-life, receptor expression on PHSC, intracellular retroviral half-life, possible antiviral host cell responses and ex vivo PHSC cycling. Table 3 gives an overview of a large number of studies which have identified factors that limit retroviral transduction of PHSCs. These studies will be discussed in detail below. EXTRACELLULAR HALF-LIFE, BROWNIAN MOTION AND VIRUS-CELL CONTACT At least two biological processes potentially limiting retroviral gene transfer are the short extracellular half-life of a retroviral particle and the distance a retroviral particle can travel in solution by Brownian motion. The half-life of a retroviral particle in cell culture medium is relatively short at 37 C, ranging between six and eight hours [58, 59]. Moreover the distance a retrovirus can travel by Brownian motion has been calculated to be less than 600 µm within one half-life [59, 60]. These two factors indicate that only those retroviral particles that are at close proximity to a target cell are able to bind to the cell. To increase the chance of binding, several techniques have been reported which aim at colocalizing the virus and the target cells. One such strategy is calcium phosphate precipitation of virus onto target cells [61]. This technique increased retroviral transduction 50-fold. However, since the increase was only reported on NIH/3T3 cells, the increased transduction and possible toxicity of calcium phosphate on human hemopoietic cells is unclear. Fragments of the extracellular matrix protein fibronectin have also been used to colocalize virus and target cells. For stem cell gene therapy purposes a number of fragments have been identified that bind both PHSCs and retrovirus, resulting in a 10- to 50-fold increase in transduction efficiency [62, 63]. Fibronectin-facilitated transduction of murine PHSCs with an ecotropic ADA retrovirus resulted in long-term transgene persistence in mice. Human ADA expression levels in the peripheral blood, six months after transplantation, were similar to endogenous murine ADA levels in the animals reconstituted with bone marrow transduced by either fibronectin-facilitated supernatant infection or cocultivation [64]. Supernatant infection with the ecotropic virus did not result in detectable levels of human ADA, in contrast to previous studies which demonstrated long-term persistence of the transgene using an ecotropic virus and supernatant infection [65]. Dick and coworkers showed preliminary data indicating that fibronectin-facilitated transduction of human cord blood CD34 + cells resulted in vector-positive SCID repopulating human cells [66]. Centrifugation or flow-through transduction aims at increasing the retroviral transduction efficiency by

Havenga, Hoogerbrugge, Valerio et al. 170 Table 3. Retroviral transduction of primate PHSCs: limiting factors. A compilation of results from studies that were performed in an attempt to overcome various putative factors thought to be involved in limiting retroviral transduction of primate PHSCs. See text for details Limiting factor Strategy Cell type Increase in transduction Reference Viral half-life Addition of dntps NIH/3T3 10-fold [99] (intracellular) (intervirion) Viral half-life Flow through transduction NIH/3T3 50-fold [59] (extracellular) Centrifugation CD34 + cells 5-fold [67] Fibronectin CD34 + cells 10-fold [62] CaPO4 precipitation NIH/3T3 50-fold [61] Polybrene CD34 + cells Protaminesulfate CD34 + cells Liposomes CD34 + cells 4-fold [71] Viral titer Sodium butyrate Packaging cells 10-fold [72] Temperature Packaging cells 10-fold [58] Receptor expression Phosphate depletion T cells 3-fold [87] Viral tropism CD34 + cells 2-fold [78] Liposome-retrovirus AS52 (hamster) 0.1%* [81] Adenovirus-retrovirus CD34 + cells 10%* [82] Adenovirus-facilitated CHO cells 60%* [83] retroviral infection PHSC cycling Anti-TGF-β CD34 + cells 4-fold [115] Stroma CD34 + cells 4-fold [40, 116, 117] *Titer of retrovirus on cells outside the host range as compared to titer of retrovirus on cells within the host range. enhancing the chance of virus binding to target cells. Centrifugation-mediated retroviral transduction has proved its use in increasing retroviral infection of human hemopoietic cells. Centrifugation-mediated transduction by three, twohour infections at 2,400 g was shown to enhance retroviral infection approximately sixfold on human CD34 + cord blood cells [67, 68]. Flow-through transduction utilizes porous membranes on which target cells are cultured and exposed to retrovirus by passing the supernatant through the membrane and thus pass the target cells [59, 69]. This technique demonstrated a 50-fold increase in transduction efficiency without the need for polybrene or protamine-sulfate. Also, transduction was no longer dependent on the virus titer since similar transduction efficiencies were obtained with virus titers ranging from 10 2 to 10 5 infectious particles. Moreover, since one volume of retroviral supernatant is being circulated over the target cells, only a small volume of the retroviral supernatant is required. Retrovirus binding to a target cell is a process facilitated by positively charged substances such as polybrene or protamine-sulfate. Addition of these substances greatly increases the retroviral infection efficiency by modulating the natural charge repulsion barrier. The importance to overcome this barrier was stressed by a recent study which showed that by combining cationic lipids with retrovirus there was a 10-fold increase in retroviral transduction in human fibrosarcoma cells as compared to the addition of polybrene [70]. The finding that addition of polybrene did not enhance the effect of the cationic lipids and that the effect was attainable by lipid treatment of either cells or retrovirus suggested that modulation of charge was responsible for the effect observed. Cationic lipid-mediated retroviral transduction has also been shown to increase the retroviral transduction efficiency in human CD34 + cells and CD34 + /CD38 cells [71]. Other approaches to increase retroviral transduction have focused on increasing the virus titer obtained with a particular producer by stimulating retroviral particle production and/or prolonging the extracellular half-life once the virus has been produced. Kotani et al. demonstrated a 10- to 100-fold increased virus titer when growing retroviral producer cells at 32 C instead of 37 C. This shift in temperature prolongs the extracellular half-life [58]. The use of sodium butyrate which increases RNA levels by generally enhancing promoter activity offers an alternative approach. Addition of sodium butyrate elevated the virus titer approximately 2- to 1,000-fold. The effect is not universal since the increase in RNA levels was shown to be dependent on the transgene present in the retroviral construct and on the retroviral producer cell used [72]. The significance of most of these strategies for increasing long-term in vivo transgene persistence remains to be investigated since CFUs and LTC-ICs were used to determine

171 Retroviral Stem Cell Gene Therapy the increase in transduction efficiency. For this purpose gene marking of human bone marrow or mobilized peripheral blood with retroviruses, in combination with flow cytometry and clinical gene marker studies, should be performed. EXPRESSION OF RETROVIRAL RECEPTORS The level of expression of retroviral receptors is another factor limiting transduction of primate PHSCs. By reverse transcriptase PCR it was demonstrated that mrna levels of the amphotropic receptor, GLVR2, were low in human CD34 + /CD38 + cells and very low in human CD34 + /CD38 cells [22]. This finding has been supported by GLVR2 protein studies on fresh CD34 + cells from bone marrow, peripheral blood and cord blood [73]. This implies that receptor expression on human primitive cells might limit transduction using amphotropic viruses. To overcome low levels of expression of the amphotropic receptor on human primitive cells two different approaches are under investigation. The first approach focuses on using alternative receptors or altogether abolishing the need for receptor binding for infection. The second approach is to increase the expression of retroviral receptor mrna levels in PHSCs. Concerning the use of other receptors, much effort has been spent on GaLV as an alternative to amphotropic retroviral vectors. This virus infects many mammalian species and a wide variety of cell types [24]. A retroviral packaging cell line has been constructed expressing the gag and pol proteins from Moloney leukemia virus and the envelope protein from GaLV [74]. A comparative study on mrna levels of the rat homologue of the GaLV receptor and the amphotropic virus receptor demonstrated that in rodent bone marrow the GaLV receptor is expressed at significantly higher levels [75]. Although increased numbers of vector-positive CFUs were scored using the GaLV virus, a long-term gene transfer study in baboons using both GaLV and amphotropic virus resulted in 0.1%-1% vector-positive bone marrow cells for both pseudotyped viruses [76]. These results correlate to another long-term gene transfer study in rhesus monkeys. In this study 0.1% of bone marrow cells were gene marked after transduction with a GaLV pseudotyped virus. Analysis of amphotropic and GaLV receptor mrna in rhesus bone marrow indicated that amphotropic receptor RNA levels were higher [77]. Increased transduction efficiency of human primitive progenitor cells has been reported using a GaLV pseudotyped retrovirus [78]. Within our group, retroviral transduction of human bone marrow cells with GaLV pseudotyped and amphotropic virus did not show a clear-cut difference between the two tropisms [79]. In addition, mrna levels of the GaLV receptor were found to be at lower levels compared to amphotropic receptor in different human CD34 + /CD38 subfractions [80]. Successful infection of cells which are normally outside the host range of a retrovirus has been reported. Innes and coworkers demonstrated infection with amphotropic retroviruses of Chinese hamster ovary cells in the presence of lipofectin. Lipofectin-mediated infection of cells lacking the receptor resulted in a titer of approximately 0.1% of the titer in cells which contained the homologous receptor [81]. Adams et al. demonstrated infection in HeLa cells when replication defective adenovirus and ecotropic retrovirus were simultaneously added to the cells [82]. Several observations made during this study suggested that the presence of the replication defective adenovirus enhances the entry of the retrovirus. The titer proved to be as high as 10% of the titer determined on cells within the host range of the retrovirus. The other approach, which aims at increasing expression of a retroviral receptor on target cells, has also demonstrated increased retroviral transduction. Adenovirus-mediated transient expression of the amphotropic receptor in HeLa cells was shown to increase retroviral transfer 10-fold [83]. Transient expression of the amphotropic receptor in Chinese hamster ovary cells increased retroviral transfer from 0% to 60%. Such a strategy might prove valuable since it has been reported that adenoviral vectors can infect human mononuclear cells and CD34 + and CD34 + /CD38 cells [84, 85]. Whether overexpression of members of this family of retroviral receptors is toxic to human primitive cells remains to be investigated. In a similar approach, adeno-associated virus was used to deliver the ecotropic receptor (mcat1) to a human cell (HeLa). Using flow cytometry it was shown that 80% of the HeLa cells expressed high levels of the mcat1 receptor. Infection with an ecotropic virus carrying a LacZ marker gene and subsequent βgal staining revealed only 30% infected cells [86]. Because both GLVR receptors function as sodium-dependent phosphate symporters, depletion of phosphate from the cell culture medium is expected to upregulate GLVR expression. Using a GaLV virus, supernatant infection in combination with centrifugation, phosphate depletion and low temperature incubation (32 C) demonstrated a 50-fold increase in transduction efficiency on peripheral blood lymphocytes. With an amphotropic virus, a 25-fold increase was observed [87]. The contribution of phosphate depletion on increased transduction was approximately threefold in this study. The effect of phosphate depletion on retroviral transduction of primate PHSC has not yet been reported. Although these studies described here demonstrate increased retroviral transduction by increasing the number of retroviral receptors on the cell surface, other studies have indicated that perhaps auxiliary receptors, in addition to the primary receptors, are involved in successful retroviral infection. The studies in which an ecotropic or amphotropic receptor was introduced in