Replication-competent Lentivirus Analysis of Clinical Grade Vector Products

Transcription

1 original article Replication-competent Lentivirus Analysis of Clinical Grade Vector Products Kenneth Cornetta 1 3, Jing Yao 1, Aparna Jasti 1, Sue Koop 1, Makhaila Douglas 1, David Hsu 4, Larry A Couture 4, Troy Hawkins 1 and Lisa Duffy 1 1 Department of Medical and Molecular Genetics, Indiana University School of Medicine, Indianapolis, Indiana, USA; 2 Department of Medicine, Indiana University School of Medicine, Indianapolis, Indiana, USA; 3 Department of Microbiology and Immunology, Indiana University School of Medicine, Indianapolis, Indiana, USA; 4 Center for Applied Technology Development, Beckman Research Institute of City of Hope, Duarte, California, USA Lentiviral vectors are now in clinical trials for a variety of inherited and acquired disorders. A challenge for moving any viral vector into the clinic is the ability to screen the vector product for the presence of replication- competent virus. Assay development for replication-competent lentivirus (RCL) is particularly challenging because recombination of vector packaging plasmids and cellular DNA leading to RCL has not been reported with the current viral vector systems. Therefore, the genomic structure of a RCL remains theoretical. In this report, we describe a highly sensitive RCL assay suitable for screening vector product and have screened large-scale vector supernatant, cells used in vector production, and cells transduced with clinical grade vector. We discuss the limitations and challenges of the current assay, and suggest modifications that may improve the suitability of this assay for screening US Food and Drug Administration (US FDA)- licensed products. Received 14 August 2010; accepted 17 November 2010; published online 21 December doi: /mt Introduction The clinical application of lentiviral vectors requires a sensitive and reliable test for detecting replication-competent lentivirus (RCL) to ensure patients are not inadvertently exposed to replicating virus. 1 6 Many investigators are seeking to generate clinical lentiviral vectors through the transient transfection method using plasmids expressing the transgene vector and the viral genes required for virion formation. To expand the range of cells susceptible to vector transduction, the HIV-1 envelope glycoprotein (which restricts infection to CD4 + cells) is replaced with an alternative envelope, most commonly the vesicular stomatitis virus G glycoprotein (VSV-G). 7 Many clinical investigators are developing third-generation lentiviral vectors that have been modified to remove accessory proteins from the vector and packaging plasmids. 8,9 As an added safety precaution, Rev responsive elements may be retained in the vector, thus requiring Rev expression during vector production. The most likely source of RCL is recombination between transfer vector and packaging construct sequences used in vector production. In addition, the possibility of recombination between packaging plasmids and human endogenous retroviral sequences must also be considered. 10 If vector is generated using transient transfection methods, detecting RCL can be further complicated by contamination of vector supernatants with packaging plasmid DNA that contain the same viral sequences likely to be present in a RCL. 11,12 A limited number of RCL assays have been described in the literature. A PCR assay to detect tat sequences from a packaging construct in cultures infected with lentiviral vectors has been described for bovine Jembrana disease viral vectors. 13 As tat has been removed from most third-generation HIV-1-based lentiviral packaging constructs, this assay has limited applicability for many clinical applications. Syncytia formation assays for testing vector-transfected cells, producer cells, or transduced cells have been developed using cell lines permissive for HIV-1 infection. 14 However, these assays detect a fully competent, env-containing lentivirus and may not detect the type of RCL generated with current HIV-1 vectors. Marker rescue assays involving mobilization of an integrated marker provirus following infection of an indicator cell line with RCL have also been used for RCL but whether an unusual recombinant virus will rescue the marker vector is unknown. 1,4 Assays based on sensitive measures of reverse transcriptase activity do provide broad base screening for retroviruses but are associated with a significant rate of false positive as background activity varies with the cell type and media used for cell growth. 15,16 In developing detection methods for RCL, the C8166 human T cell line has been shown to be highly infectable with HIV-1 and lentiviral vectors pseudotyped with the VSV-G envelope As important, this cell line is able to amplify HIV-1 at high titer. Previously, we compared detection of virus using a commercially available p24 enzyme-linked immunosorbent assay (ELISA) method (sensitivity of 3 pg/ml 36,000 viral particles), a real-time PCR assay for VSV-G env DNA (sensitivity of ~5 50 copies/0.1 µg of genomic DNA), and a PCR assay (psi-gag PCR) that detects early recombination between the vector plasmid and the gag/pol plasmid (sensitivity 10 copies/0.1 µg of genomic DNA). These detection assays are not sensitive enough to detect small numbers of virus particles but are equally sensitive when limiting number Correspondence: Kenneth Cornetta, Indiana University School of Medicine, Department of Medical and Molecular Genetics, IB 130, 975 West Walnut Street, Indianapolis, Indiana 46202, USA. kcornett@iupui.edu Molecular Therapy vol. 19 no. 3, mar

2 RCL Testing of Clinical Products of viruses are first amplified to high titer on cell lines such as the C8166 T cell line. 18 As the true nature of a RCL remains theoretical, we chose two methods of virus detection to improve the chance of detecting an unusual recombinant rather than rely on a single read-out assay. Our assay currently uses the p24 ELISA to detect HIV-1 capsid protein along with PCR for psi-gag recombination. The latter was chosen based on our prior observations of rare recombinations occurring in vector preparation. 18 We have utilized this assay to screen large-scale vector preparations for RCL and the experience and refinements in the assay are presented in this report. Results Assay design We developed a RCL assay that is modeled after those currently accepted by the US Food and Drug Administration (US FDA) for detecting replication-competent murine retroviruses (RCR). 20 For RCL detection, the test article is first amplified in the C8166 T cell line that is permissive for viral infection. 18 The virus is allowed to replicate during the 3-week culture period (amplification ) so virus below the limits of detection at the start of the assay will be amplified and readily detected after 3 weeks (Figure 1). Our preliminary experience with the assay led us to add an indicator rather than test material at the end of the amplification for two reasons. First, when testing concentrated vector material (which has a high p24 content), we occasionally observed low levels of p24 in the amplification even after 3 weeks of culture. Therefore, the possibility of a slow-growing RCL cannot be excluded. Secondly, when evaluating cultures for viral DNA, we previously reported that there can be rare recombinations between vector and packaging plasmids. 18 Although the recombinations do (3 weeks) Vector RCL 7 10 days Assay by PCR and ELISA Figure 1 Schematic representation of the replication-competent lentivirus (RCL) assay. Vector product is used to transduce C8166 cells. Small amount of virus which may be present in vector preparations will propagate in the C8166 culture over the 3 weeks of the amplification typically yielding virus concentration in excess of X ng/ml of p24. Cell-free media is harvested at the end of the amplification and used to transduce naive C8166 cells, which are propagated in the indicator. At the end of the indicator, the media is evaluated for virus using the p24 enzyme-linked immunosorbent assay (ELISA) assay. Cells are evaluated for evidence of virus using PCR amplification with primers within the viral packaging sequence and the gag gene region (psi-gag PCR). not reconstitute a RCL, they may be present (at very low levels) in the amplification cells leading to a false positive assay. The indicator ensures that the assay detects a true RCL by requiring passage from amplification cells to naive C8166 cells. For a positive, we have selected the attenuated HIV-1 virus R This lacks the HIV-1 vif, vpr, vpu, and nef accessory genes that are also deleted in most third-generation lentiviral packaging systems. It should be noted that this virus expresses the native HIV-1 envelope whereas all of the vector testing in this report express the VSV-G glycoprotein envelope. While a RCL that arises from vector products is likely to also express the VSV-G envelope, we believed that generating a VSV-G pseudotyped replication-competent HIV-1 virus with the wide host range conferred by the VSV-G envelope represented an unacceptable risk to laboratory personnel. Optimizing the concentration of vector product Although our initial test articles were unconcentrated vector material, most final products intended for clinical use will be concentrated to high titer. Upon testing concentrated vector, we noted significant inhibition of C8166 growth using material produced at our facility and in the laboratory of our collaborators and with vectors that expressed a variety of different transgenes (data not shown). To help define the relationship between inhibition and vector dose, CSCGW vector [an HIV-1-based vector expressing green fluorescent protein (GFP) and pseudotyped with the VSV-G envelope] was generated by transient transfection then concentrated 100-fold by ultracentrifugation. C8166 and MRC-5 cells were exposed to undiluted, 1:10, and 1:100 dilutions of vector and the effect on cell expansion was measured after 48 hours (Figure 2a). The addition of concentrated vector inhibited cell growth in a dose-dependent manner, with the greatest effect noted in MRC-5 cells. To help determine the optimal dose for use in the RCL assay, C8166 cells were exposed to CSCGW vector at defined amounts of p24 for 4 hours (the length of vector exposure in the RCL assay). As shown in Figure 2b, growth inhibition was noted with ~60% of the cells present after exposure to vector at the 10,000 ng/ml concentration. We have also observed growth inhibition using clinical grade vector produced at Indiana University (Indianapolis, IN) and at the City of Hope (data not shown). This data suggest that care must be taken when screening concentrated vectors. For example, MRC-5 cells are commonly used in the in vitro virus assay for adventitious agents, a certification assay that has been required by the US FDA before clinical use of gene therapy products. The data also suggests that at a vector concentration of 1,000 ng/ml the growth inhibition of C8166 cells is modest with ~90% of cells surviving. The concentration of vector is also important when optimizing the sensitivity of the RCL assay. For example, vector may be present in great excess relative to any RCL; the vector particles could block receptors and decrease the sensitivity of RCL detection. This was assessed by introducing a small amount of the GFP-expressing vector (CGCGW, sufficient to transduce 10 20% of C8166 cells) with varying concentrations of the CSO vector (a vector that lacks an expressed transgene). Infectivity was monitored by assessing the number of GFP-expressing cells after transfection. As shown in Figure 2c, there was a dose-dependent decrease in the number vol. 19 no. 3 mar. 2011

3 ,000 5,000 10,000 RCL Testing of Clinical Products of GFP-expressing cells indicating that excess of CSO vector particles limited the ability of CSCGW vector to transduce the target cells. At 1,000 ng/ml, this inhibition is quite modest (~80%) and is not predicted to significantly decrease the sensitivity of the RCL assay. Therefore, for the RCL assay, we determine the p24 concentration on all vector supernatants and, if needed, dilute the material to 1,000 ng/ml and do not exceed a ratio of 1,000 ng of p24/ C8166 cells. Performance of assay s Critical to the usefulness of a clinical RCL detection assay is its ability to minimize false negatives and false positives. As shown in Figure 3a, our initial assay incorporated three positive and three negative cultures at the start of the amplification and three negative and five positive cultures at the start of the indicator. All positive cultures were inoculated with the R8.71 attenuated HIV-1 virus at the TCID 50 (~0.5 IU/culture). We chose five cultures for the indicator positive since virus would only have 7 days to expand. For the assay to be acceptable, all negative cultures must be negative for RCL and at least one of three amplification positive cultures, and at least one of the five-indicator positive cultures must be positive. For a culture to be positive, the amount of p24 must be above background for the ELISA, and a band of the expected size for the psi-gag PCR must be present on Southern blot, or both. Using this series of s, we have performed 13 RCL assays under Good Manufacturing Practice conditions and the results are presented in Table 1 (assays 1 13). a C8166 MRC-5 Total cell number Control Undiluted 1:10 1:100 Camp b Total cell number C8166 Exp 1 C8166 Exp 2 p24 ng/ml c 120 Percent GFP-expressing cells ,000 ng/ml 5,000 ng/ml 10,000 ng/ml Figure 2 Effect of concentrated lentiviral vector on the growth and infectivity of C8166 cells. (a) The effect on concentrated vector was assessed in the nonadherent C8166 cell line and the adherent MRC-5 cell line. Cells were exposed to concentrated CSCGW vector for 4 hours and cell number counted after 48 hours. Concentrated vector (~5,000 ng/ml) was used undiluted or diluted 1:10 and 1:100. Camptothecin was used as a positive toxin. The data represents the average of three cultures ± SD. (b) The effect of concentrated vector on the growth of C8166 cells was measured at three concentration, 1,000, 5,000, and 10,000 ng/ml of p24 using the CSCGW vector. Duplicate cultures were exposed to vector preparations on day 0 and the total number of cells was determined in each culture after 5 or 6 days of growth (experiment 1 or 2, respectively). Data represents the average of two cultures ± SD for two independent experiments. (c) The inhibitory effect of concentrated vector on C8166 cell infectivity was measured by introducing increasing amounts of a null vector (CSO) into a lentiviral vector preparation of the green fluorescent protein (GFP)- expressing CSCGW vector. The x-axis represents the amount of CSO vector as measured by p24 ELISA. The y-axis represents the GFP-expressing cells as a percentage of (no CSO vector). The percentage of cells expressing GFP ranged between 13.5 and 20.1%. The data represents the average of three independent experiment ± SD. Molecular Therapy vol. 19 no. 3 mar

4 RCL Testing of Clinical Products a Initial assay b Modified assay Negative Positive IP negative IP positive Negative Positive Inhibition Media 0.5 IU Media 5 IU 50 IU + vector Media 0.5 IU Acceptance criteria In terms of performance of the p24 ELISA assay there were no false positive assays detected in the negative cultures. By ELISA, amplification positive cultures were positive in 33 of 39 cultures (Table 1, column 5) and indicator positive were positive in 46 of 60 cultures (Table 1, column 9). For psi-gag PCR testing, all negative s were negative except for one indicator where a faint band was seen in one of two duplicate lanes that was interpreted upon further testing as a false positive. For amplification positive s 30 of 30 cultures were positive (Table 1, column 6) whereas 39 of 50 were positive when inoculated at the start of the indicator (Table 1, column 10). These findings suggest that the p24 ELISA may be more sensitive than the psi-gag Southern when detecting low levels of virus (i.e., after only 7 days of expansion from the TCID 50 ). Both ELISA and PCR performed well when higher concentrations of virus were being analyzed (after expansion in the amplification and indicator ). Our data also suggests that our estimate of 0.5 IU/culture was close to the actual amount, with the actual inoculum being ~0.5 1 IU/culture. p24 0/3 + 0/3 + 1/3 + 1/3 + 0/3 + 0/3 + 1/5 + 0/3 + 0/3 + 1/3 + 2/3 + PCR 1/5 + 1/3 + 2/3 + Figure 3 Controls for the replication-competent lentivirus (RCL) assay: inoculation and acceptance criteria. (a) The initial version of the RCL assay utilized three negative s setup at the start of the amplification and three at the start of the indicator. All cultures are assayed for RCL at the end of the indicator. Two sets of positive s were utilized, both inoculating cultures with R8.71 attenuated HIV-1 virus at the TCID 50. Three cultures are set up at the start of the amplification and five cultures inoculated at the start of the indicator. Five cultures were chosen for the indicator positive since virus expansion will only be occurring for 7 days and the level of virus after this time is expected to be low. However, both the amplification positive s and the indicator positive s must have at least one culture positive for RCL for the assay to be valid. The assay is considered positive is the p24, psi-gag PCR, or both are positive. The established acceptance required all negative cultures to be negative for RCL. (b) The RCL assay was modified to remove the indicator negative and positive s. The inoculum at the start of the amplification was increased to 5 IU per positive culture. An inhibitor was added to the assay in which the test articles (at a concentration 1,000 ng/ml of p24) are spiked with 50 IU. At least one of the amplification positive s and two of the inhibitor s must be positive for the assay to be valid. IP, intraperitoneal. As a quality, we also measured p24 on the material collected at the end of the amplification in all 30 assays (data not shown). There was no instance where a positive p24 detected at the end of the amplification was negative after the indicator, providing assurance that the passage of cell-free media from the amplification cultures to the indicator does not decrease virus detection. There were three assays that failed to meet the acceptability criteria. As discussed above, one negative culture had a false positive psi-gag PCR assay. This analysis did not require retesting of vector product. Two assays failed because the positive s failed to meet specifications, with one assay failing at the amplification and the other at the indicator. Since these failures called into question the sensitivity of the assay, the entire assays were repeated. On rerunning the assay, the positive s met acceptance criteria and the test articles were found to be free of RCL. Revising assay s The failure of 2 of our first 13 assays to meet acceptance criteria because the positive cultures were negative led us to reevaluate the selection of 0.5 IU as the inoculum for positive cultures. At this level, we expect that only 50% of cultures would be positive but even using a careful methodology for preparing the positive s, consistently inoculating this low level of virus into a culture is challenging. Coincident with this finding, we began an evaluation of assay sensitivity when the test article is a lentiviral vector designed to inhibit HIV-1 replication. We conducted a qualification protocol where the rhiv7-shi-tar-ccr5rz viral vector 21,22 (generated at the City of Hope and diluted to 1,000 ng/ ml) was spiked with the R8.71-attenuated HIV-1 virus at concentrations ranging from 5 to 5,000 IU. The rhiv7-shi-tar-ccr5rz vector contains three sequences targeted against HIV-1 with a U6 Pol III promoter-driven short hairpin RNA targeting the rev and tat mrnas of HIV-1, a U6 transcribed nucleolar-localizing TAR RNA decoy, and a VA1-derived Pol III cassette that expresses an anti-ccr5 ribozyme. As shown in Table 2, the RCL assay was capable of detecting replicating virus, even at inoculums as low as 5 IU. Based on our experience with the first 13 GMP RCL assays, and our finding with the above qualification protocol, we revised the s in our RCL assay (see Figure 3b). Since the indicator negative and positive s correlated with amplification s, we no longer perform them and rely solely on the s initiated at the start of the amplification. To decrease the likelihood that all positive s would be negative, we inoculate the positive cultures with 5 IU of virus. Furthermore, we have added an inhibition where test articles (i.e., vector product or cells) are spiked with 50 IU of R8.71 virus. Three inhibition cultures are inoculated and carried through the amplification and indicator. At least two of the three must be positive for the assay to be acceptable. We continue to utilize vector product at a maximum concentration of 1,000 ng/ml and do not exceed C8166/1,000 ng of p24. To date, the modified RCL assay has proven reproducible and sensitive. We have not had to repeat an assay due to failure of positive s nor have we seen significant inhibition by vector products (Table 3) vol. 19 no. 3 mar. 2011

5 RCL Testing of Clinical Products Table 1 Performance of the RCL assay positive and negative s Initiation negative p24 ELISA negative psi-gag PCR positive p24 ELISA positive psi-gag PCR negative p24 ELISA negative psi-gag PCR positive p24 ELISA positive psi-gag PCR 1 10/3/2005 0/3 0/3 3/3 3/3 0/3 0/3 4/5 3/5 2 9/20/2005 0/3 0/3 3/3 3/3 0/3 0/3 4/5 4/5 3 3/30/2005 0/3 0/3 2/3 (3/3) ND (3/3) 0/3 0/3 0/5 (5/5) ND (5/5) 4 5/16/2006 0/3 0/3 3/3 3/3 0/3 0/3 5/5 3/5 5 8/15/2006 0/3 0/3 3/3 3/3 0/3 0/3 5/5 5/5 6 9/5/2006 0/3 0/3 3/3 3/3 0/3 0/3 3/5 3/5 7 9/11/2006 0/3 0/3 3/3 3/3 0/3 1/3 a 5/5 3/5 8 9/21/2006 0/3 0/3 3/3 3/3 0/3 0/3 5/5 5/5 9 10/16/2006 0/3 0/3 3/3 3/3 0/3 0/3 5/5 5/5 10 2/27/2007 0/3 0/3 0/3 (3/3) ND (3/3) 11 2/27/2007 0/3 0/3 3/3 3/3 0/3 0/3 5/5 5/5 12 4/3/1997 0/3 0/3 3/3 3/3 0/3 0/3 5/5 3/5 13 4/23/2007 0/3 (0/3) ND (0/3) 1/3 (1/3) ND (2/3) 0/3 (0/3) ND (0/3) 0/5 (3/5) ND (3/5) 14 9/17/2007 0/6 0/6 6/6 6/ /22/2007 0/3 0/3 3/3 3/3 16 3/12/2008 0/3 0/3 3/3 3/3 17 6/5/2008 0/3 0/3 3/3 3/3 18 6/11/2008 0/3 0/3 3/3 3/3 19 7/9/2008 0/3 0/3 3/3 3/ /27/2008 0/3 0/3 3/3 3/3 21 4/8/2009 0/3 0/3 3/3 3/3 22 6/1/2009 0/3 0/3 3/3 3/3 23 6/8/2009 0/3 0/3 3/3 3/3 24 6/17/2009 0/3 0/3 3/3 3/ /19/2009 0/3 0/3 3/3 3/ /22/2009 0/3 0/3 3/3 3/3 27 1/5/2010 0/3 0/3 3/3 3/3 28 1/25/2010 0/3 0/3 3/3 3/3 29 2/10/2010 0/3 ND 3/3 ND 30 3/1/2010 0/3 0/3 3/3 3/3 Initial assay b 0/39 0/36 33/39 30/30 0/36 1/33 46/60 39/50 Modified assay b 0/54 0/51 54/54 51/51 Total b 0/93 0/90 87/93 86/87 0/36 1/33 46/60 39/50 Abbreviations: ELISA, enzyme-linked immunosorbent assay; ND, not done; RCL, replication-competent lentivirus. The data is presented as the number of positive cultures over the total number of cultures performed. Values in parenthesis represent data from repeat assays performed when the initial assay did not meet acceptance criteria. a Assay results did not meet acceptance criteria and was determined on subsequent testing to represent a false positive. b Excludes repeat testing results. Performance of the RCL assay in screening vector products Table 3 summarizes the vector products tested and the results of the inhibition s for materials tested using the modified RCL assay. All of the vector products have been produced using transient transfection methods, the majority using third- generation vector systems. 8 With a total of 30 assays performed under GMP conditions, there has been no evidence of RCL detected in vector supernatants, end-of-production cells, engineered cell lines, and clinical samples transduced with lentiviral vectors. For materials intended for clinical use, we have analyzed ng of p24, which is estimated to represent ~ virions. We have performed screening of end-of-production cells for 17 products, and screened an additional seven cell lines to be used in generating clinical trial material. This represents screening of ~ cells generated in six different GMP facilities (~20% of the tested material was generated at Indiana University). Molecular Therapy vol. 19 no. 3 mar

6 RCL Testing of Clinical Products Table 2 Detection of attenuated HIV-1 contaminating the rhiv7-shi- TAR-CCR5RZ vector expressing sequences aimed at suppressing HIV-1 replication Split 1 Media alone 0/3 0/3 0/3 Vector alone 3/3 0/3 0/3 5 IU 0/3 3/3 3/3 Vector + 5 IU 3/3 3/3 3/3 Vector + 50 IU 3/3 2/3 3/3 Vector IU 3/3 3/3 3/3 Vector + 5,000 IU 3/3 3/3 3/3 Samples were assessed for p24 after the first media change (split 1), the amplification, and the indicator. The number of positive cultures over the total number of cultures analyzed is presented. In this assay, all positive cultures had levels of virus virions/ml. Discussion When evaluating I trials using novel agents, safety not efficacy is the major factor used by the US FDA in determining whether these early trials may move forward. Inadvertent exposure of subjects to pathogens that contaminate gene therapy products is a top safety concern. In this article, we demonstrate consistent performance of a RCL detection assay. The testing of clinical grade material provides further evidence that the likelihood of RCL development using current lentiviral vector systems is low. The initial retroviruses used for gene therapy were based on the murine leukemia viruses. RCR that arise during production of murine leukemia virus-based vectors are known to cause malignancy in mice and immunosuppressed nonhuman primates. 23,24 Murine leukemia virus-based vectors are generally produced by packaging cell lines and RCR was frequently detected in early versions of these cell lines RCR development has been decreased by minimizing homology between vector and packaging cell sequences and segregating packaging genes Still, recombinations have been detected between packaging gene sequences and endogenous sequences within the packaging cells. 35,36 The detailed evaluation of RCR that arise during the course of retroviral vector productions has facilitated the development of RCR assays and shaped US FDA recommendations for vector testing. 37 Lentiviral vectors are replacing murine leukemia virus-based retroviral vector due to their improved ability to transduce nondividing cells, and their lower risk of insertional mutagenesis. 11,38 46 Another advantage of the lentiviral system is an apparent lack of RCL generation during vector production. This may be due, in part, to the use of transient transfection methods for lentiviral production. The ability to segregate the vector components onto four plasmids, the use of self-inactivating long terminal repeat s, and retention of Rev dependence also contributes to the safety profile of lentiviral vectors. To date, no RCL has been reported using this method although vector material has not previously been screened as extensively as we described in this article. A key component of any RCL assay is the positive. While a VSV-G pseudotyped HIV-1 virus may better represent a RCL arising through recombination, that assumption remains theoretical and a VSV-G pseudotyped HIV-1 presents significant risk to laboratory personnel. Most importantly, the positive is used to establish that the sensitivity of the assay is consistent from assay to assay and the R8.71 HIV-1 is sufficient to provide this assurance. In terms of sensitivity, our assay appears to be able to detect R8.71 at 1 IU but with ~10% of cultures failing acceptance criteria in our initial RCL assay, we increased the positive inoculum from the TCID 50 (0.5 IU/culture) to 5 IU/culture. While it is possible that the original failure of virus detection in positive cultures represents variability in the assay, the finding of consistent positives at the 5 IU suggest that the technical challenges of diluting virus to the TCID 50 and the statistical probabilities of relying on three cultures at the TCID 50 are the cause. Preparation of our positive is done through a defined procedure under GMP protocol with extensive testing to establish the IU, a factor critical in setting the limits of detection of our assay (1 5 IU). We demonstrate that the vector decreases the sensitivity of RCL detection, but by diluting vector to 1,000 ng/ ml and maintaining a high ratio of cells to vector (1,000 ng for C8166 cells) this effect can be minimized. This was shown when comparing two VSV-G pseudotyped virions (Figure 2c) and when the R8.71 virus is spiked into vector known to inhibit HIV-1 replication (Table 2). The assay described here was overdesigned by intent. Since lentiviral vectors are just entering the clinic, an assay, which could provide a high level of assurance that RCL could be detected, was required. To this aim, we relied on two methods of virus detection (p24 ELISA and psi-gag PCR) to provide redundant virus detection. This redundancy minimizes the possibility that a technical problem with a single assay (ELISA or PCR) would lead to a false positive or negative result. Two assays may also increase the chance of detecting a RCL that arose from vector sequences recombining with endogenous retroviral sequences. As the psi-gag PCR method involves detection of the PCR product by Southern blotting using a radiolabeled probe, we are now looking to replace this cumbersome assay with quantitative PCR for the VSV-G envelope DNA. We have previously shown the VSV-G quantitative PCR has similar sensitivity to psi-gag PCR in detecting RCL and are currently validating this alternative approach. 18 We would continue to test for p24 that is expected to be positive, even if the RCL contained human endogenous retroviral or other viral envelope sequences. The amount of vector used per cell (1,000 ng/ml p24 exposed to C8166 cells or cocultured with C8166 cells) is approximately five times the ratio used in RCR assays. This excess of C8166 cells does suggest that the number of infections per cell is low, and may explain the ability of our assay to detect RCL in the presence of an anti-hiv-1 vector. Although our current assay has shown excellent sensitivity, it will be technically challenging to meet the current US FDA guidance requiring testing of 5% of the vector final product as manufacturing methodology allows further scale-up. Currently, the average clinical production scale is in the order of l unconcentrated, which requires in excess of C8166 cells for a single RCL assay. To facilitate RCL testing of larger batches, we are now evaluating the possibility of decreasing the number of cells per 1,000 ng of p24. Purification of vector should also decrease toxicity and allow higher concentrations of vector to be used in the RCL assay. This will need to be assessed experimentally to validate that the vol. 19 no. 3 mar. 2011

7 RCL Testing of Clinical Products Table 3 Test articles evaluated for RCL Assay number Test article GMP supernatant (pg/ml p24) End-of-production cells (total cell number) 1 Lentiviral vector Lentiviral vector Lentiviral vector Lentiviral vector Miscellaneous cells (total cell number) 3 Primary cells Transduced cell line Primary cells Lentiviral vector Lentiviral vector Lentiviral vector Lentiviral vector Lentiviral vector Lentiviral vector Lentiviral vector Primary cells Transduced cell line Miscellaneous supernatant (pg/ml p24) Inhibition p24 ELISA (# positive/total) Inhibition psi-gag PCR (# positive/total) 14 Lentiviral vector /3 3/3 14 Lentiviral vector /3 3/3 15 Transduced cell line /3 3/3 16 Lentiviral vector 7,302 3/3 3/3 16 Lentiviral vector /3 3/3 17 Lentiviral vector /3 3/3 17 Lentiviral vector 6,248 3/3 3/3 18 Transduced cell line /3 3/3 18 Lentiviral vector /3 3/3 19 Lentiviral vector /3 3/3 19 Lentiviral vector /3 3/3 20 Primary cells 3 Insuff Insuff 21 Lentiviral vector /3 3/3 21 Lentiviral vector /3 3/3 22 Transduced cell line 240 3/3 3/3 23 Transduced cell line /3 3/3 24 Transduced cell line /3 3/3 24 Transduced cell line 255 3/3 3/3 25 Lentiviral vector /3 3/3 26 Lentiviral vector /3 3/3 27 Lentiviral vector /3 3/3 28 Lentiviral vector /3 3/3 28 Lentiviral vector /3 3/3 28 Lentiviral vector /3 3/3 29 Packaging cell line 100 3/3 Alt 29 Packaging cell line /3 Alt 30 Transduced cell line Indet Indet 30 Transduced cell line 320 3/3 3/3 Total material tested Products tested Abbreviations: ELISA, enzyme-linked immunosorbent assay; RCL, replication-competent lentivirus. In the inhibition assay test, articles were incubated with 50 IU of attenuated HIV-1 virus. Molecular Therapy vol. 19 no. 3 mar

8 RCL Testing of Clinical Products increased concentration of test material does not decrease the sensitivity of the RCL assay. The ratio of C8166 to test material may also need to be adjusted based on the vector properties (i.e., a higher ratio when the test article is a vector that inhibits HIV- 1). In addition, there was good correlation between the indicator and the amplification p24 data, and the possibility of eliminating the indicator or performing it only when the results of the amplification are equivocal, could be considered. The data presented here serves as a starting point for future RCL assay modifications as vector technology and manufacturing mature. The current US FDA requirements to screen vector supernatant for RCL has proven to be a challenging endeavor. Concentrated vector pseudotyped with VSV-G can be toxic to a variety of cells, and the assay is difficult to perform even with modest scale clinical productions (10 50 l range). As production methodologies move to larger scale batches, what percentage of the final product is feasible and appropriate for RCL testing will require additional experience and assay development. The data presented in this paper does provide evidence that the incidence of RCL in lentiviral vector preparations is low and should be considered when assessing regulatory guidelines. This data may also be useful when assessing the need to test transduced cells for RCL, as currently recommended by the US FDA for transduced cells products incubated ex vivo for >4 days. While modifications are being tested to better adapt to the larger vector productions, additional work will be required as alternative pseudotypes to the VSV-G envelope and non-hiv-1-based vectors are developed for clinical trials. Materials and Methods Cell lines, and vector preparations. The C (C8166; human T-lymphocyte) was obtained from the AIDS Research and Reference Reagent Program (Rockville, MD) and maintained in RPMI The HEK293T, HEK293, and MRC-5 cell lines were obtained from the American Type Culture Collection and maintained in Dulbecco s modified Eagle s medium. RPMI and Dulbecco s modified Eagle s medium (Invitrogen, Carlsbad, CA) are supplemented with 10% fetal calf serum (Hyclone, Logan, UT), 2 mmol/l l-glutamine (Invitrogen), and 100 units/ml penicillin and 100 µg/ ml streptomycin (Invitrogen). C8166 cells are maintained in cell banks and only those cells at 15 passages are utilized at the start of an assay. Lentiviral vectors were produced by transient transfection of HEK293T cells (at a ratio of cells/75 cm 2 ) using calcium phosphate transfection (Profection kit; Promega, Madison, WI). Packaging plasmids were obtained from Cell Genesys (Foster City, CA) and include pmdlg (6.6 μg/75 cm 2 ), prsv/rev (3.3 μg/75 cm 2 ), and the VSV-G glycoprotein expression plasmid pmd.g (4.62 μg/75 cm 2 ). Vector plasmids (18 μg/75 cm 2 ) include the GFP-expressing vector pcdna-hiv-cs-cgw (provided by Philip Zoltick, Children s Hospital, Philadelphia, PA) and the null vector CSO (provided by Donald Kohn, UCLA, Los Angeles, CA). After refeeding cells with fresh medium hours post-transfection, vector-containing supernatants were harvested 48 hours after transfection, and filtered through a 0.45-µm filter. Nonclinical vector was concentrated 100-fold by ultracentrifugation (50,000g for 1 hour at 4 C in a Beckman ultracentrifuge Optima XL-100k using a 45-Ti fixed-angle rotor). Clinical vector preparation were treated with Benzonase to remove residual plasmids as previously described 53 and concentrated fold by ultrafiltration using a MiniKros disposable hollow fiber tangential flow modules (Spectrum Laboratories, Rancho Dominguez, CA). Growth inhibition assessment. To assess potential inhibition of vector on cell growth, serial dilutions of concentrated CSCGW vector were used to transduce C8166 or MRC-5 cells. Vector was pelleted by ultracentrifugation (described above), and the pellet was resuspended in media with 10% fetal calf serum at 1/100th the original volume. Aliquots of the concentrated material were diluted 1:10 and 1:100 in serum-containing media and the vector material was placed directly onto pelleted C8166 or monolayers of MRC-5 where the media had been removed. After a 4-hour incubation in the presence of 8 μg/ml polybrene (Sigma, St Louis, MO), the vector was removed and fresh media was added to the culture (RMPI or Dulbecco s modified Eagle s medium based media for C8166 or MRC-5, respectively). Alternatively, a defined concentration of CSCGW vector (0, 1,000, 5,000, and 10,000 ng) was also used to evaluate C8166 cell growth, using the same procedure. Viable cell counts were assessed 2 6 days after vector exposure. Inhibition of transduction by vector particles. The impact of competing vector particles on transduction efficiency was assessed by combining the GFP-expressing CSCGW with the null CSO vector. A limiting amount of CSCGW vector stock (previously determined to provide 15 25% GFPexpressing C8166 cells) was mixed with concentrated CSO vector (1,000, 5,000, and 10,000 ng/ml). The vector combination was used to transduce C8166 cells for 4 hours (8 μg/ml in polybrene) and the percentage of GFPexpressing cells was assessed after 72 hours using flow cytometry as previously described. 12 Positive preparation. The attenuated HIV-1 positive is generated by transfecting HEK293T cells with the pr8.71 plasmid (provided from Cell Genesys, 112 µg plasmid/ cells/) using the Profection kit (Promega). After a media change with RPMI at 24 hours, cell-free supernatant (0.45-µm filter) is collected 48 hours after transfection. Material is frozen in 0.5 ml aliquots and frozen at less than 70 C. Potency is assessed by preparing ml flat bottom culture tube with cells/tube and incubating cells overnight. On day 0, six tubes are centrifuged and cells are resuspended in 1 ml of virus with polybrene at 8 µg/ml. The six dilutions tested range from 10 3 to 10 8 of the frozen viral stock. Three negative tubes are also prepared. After 4-hour incubation, cells are pelleted, resuspended in RMPI, and transferred to 6-well plates. Cells are maintained in log- growth for at least 12 days, after which time cells are pelleted, and the media is filtered (0.45 µm) and assessed for p24 by ELISA and cultures read as positive or negative. RCL assays. Before initiation of the assay, the p24 content of each test article is assessed to determine the number of C8166 cells required. For aqueous test article, C8166 cells/ml of test material is required, and the maximal concentration of p24 used is 1,000 ng/ml. In this assay, all culture tubes/flasks are incubated at 37 C, 5% CO 2. On day 1, triplicate negative, inhibition, and positive cultures are prepared by adding 12 ml of RPMI and C8166 cells into nine 50-ml culture tubes (flat-bottom 10 cm 2 ). On day 0, three negative cultures are prepared by centrifuging cells by replacing the media with 1 ml of RPMI and polybrene (8 µg/ml). Three inhibition s are prepared in a similar manner except the cultures are inoculated with 1 ml of the test article, 50 IU of R8.71 virus, and 8 µg/ml polybrene. Three positive s are prepared by inoculating cultures with 1 ml of RPMI, 5 IU of R8.71 virus, and 8 µg/ml polybrene. Test article flasks are prepared on day 1 based on the volume of material to be tested. For clinical productions, multiple flasks are usually required. The amount of cell suspension ( /ml) added to each flask is equal to the volume of the test article (after dilution to 1,000 ng/ml). To accommodate the large volumes required for clinical production testing the following options are used: 12 ml/75-cm 2 flask or 50 ml culture tube, 30 ml/175-cm 2 flask, 60 ml/300 cm 2 flask. On day 0, cells are transferred into a 50-ml conical tube, the media is removed after centrifugation, and the test article added to resuspend the cells and the mixture is returned to the flask with polybrene (8 µg/ml) vol. 19 no. 3 mar. 2011

9 RCL Testing of Clinical Products After 4-hour incubation, the negative, inhibition, and positive culture cells are centrifuged and the media is replaced with 12 ml of fresh RPMI medium and transferred to 75-cm 2 flasks. For test articles, cultures from each tubes/flasks are centrifuged after 4-hour incubation and media is replaced with fresh RPMI medium (12 30 ml/75-cm 2 flask, ml/ 175-cm 2 flask, ml/300 cm 2 flask) and returned to the original flask. : Cells are maintained in log growth through the 3-week period by passing cells a minimum of five times during the 2-week period. For test articles in larger flasks, the initial splits are performed so that cells are placed in smaller size flasks with each subsequent split until the material is in a 75-cm 2 flask. Split ratio should not exceed 1:10 for larger flasks (300 and 175-cm 2 flasks) and 1:5 for 75- cm 2 flasks. To meet acceptability criteria, cultures must contain sufficient cells for passage on days 3 4, 6 8, 9 11, 12 14, and of the assay. After at least five splits (3 weeks of culture), cells are removed from the flask, centrifuged, and resuspended in 12 ml of fresh RPMI. At least 24 hours later, the media from each flask is collected and filter through a 0.45-µm filter. Filtered supernatant from each of the flasks is used to inoculate the corresponding indicator culture. For a test article, the material from all flasks are pooled and used to inoculate two test article indicator flasks. : In the last week of the amplification, naive C8166 cells are expanded for the indicator. The day before harvesting media from the indicator, nine flasks are prepared plus an additional two flasks for each test article analyzed in the assay. Each flask is to contain C8166 cells in 4 ml of RPMI. After incubating overnight, cells are centrifuged and resuspended in 1 ml of filtered media from the corresponding amplification culture (with polybrene 8 µg/ml). After 4 hours, cells are centrifuged and resuspended in 4 ml of RPMI. After 24 hours, all contents in a culture tube are transferred to a 75-cm 2 flask with 12 ml of fresh RPMI medium. For the test article flasks, 12 ml of RPMI is added to duplicate 75-cm 2 flasks along with C8166 cells for each of the test article flasks from the amplification (e.g., if there are 25 amplification flasks for a test article C8166 cells are added to each of 2 flasks). After incubating overnight, cells are centrifuged and resuspended in amplification media with polybrene (8 µg/ml). One milliliter of the pooled media for each of the amplification test article flasks is used for each duplicate flask. After 4 hours, cells are centrifuged and resuspended in 12 ml of RPMI. After 48 hours, all of the contents are transferred to a 175-cm 2 flask with 30 ml of fresh RPMI medium. Six days after the cells are inoculated, media is removed and fresh RPMI (12 ml/75 cm 2, 30 ml/175 cm 2 ) is added. At least 24 hours after media change, contents are transferred from flasks to a 15-ml conical tubes and centrifuged. Media is removed and individually filtered (0.45 µm) and analyzed for p24 content. Cells are then collected for DNA isolation. Coculture assay: When setting up the inhibitor s for the amplification, test articles cells are incubated with C8166 cells and 50 IU of R8.71 virus. For the test article cultures, the ratio is 1 test articles cell for every 5 C8166 cells in the amplification. The cultures are processed in a similar manner to aqueous test article except special attention is given to prevent overgrowth of the test article. Specifically, cultures are assessed 2 hours after passage and if there are significant adherent cells noted, the nonadherent cells are removed and replated into new flasks. Interpretation of results: At the end of the indicator, the assay is acceptable if (i) all three negative s are negative for p24 antigen and psi-gag sequences; (ii) two of three inhibition flasks are positive for p24 antigen and psi-gag; and (iii) at least one of the positive s flasks are positive for p24 antigen and psi-gag at the indicator s. If the s are acceptable and the test article flasks are negative for p24 antigen and psi-gag then the test article is reported as negative for RCL. If the test article is positive for both p24 and psi-gag, the sample is interpreted as RCL positive. If there is disparity in the p24 and psi-gag results of a test article, the indicator is repeated and extended for >14 days before samples are harvested. Our assay also defines criteria when samples do not meet the acceptability criteria. If one of the negative s tested positive by p24 or psi-gag PCR (but not both), the samples are reassayed and if negative the negative is considered acceptable. If the inhibition is negative for p24 and psi-gag in 2 of the three cultures it is likely that an inhibitor of RCL infection or amplification is present. If the other s are otherwise acceptable, the assay is considered acceptable but further analysis is performed to define the level of inhibition by spiking test article (1,000 ng/ml) with 100, 500, or 5,000 IU and assessing RCL detection by p24 ELISA after a 3-week amplification. The results are included in the assay findings and interpretation. If s do not otherwise meet the acceptability criteria, then the assay is considered inconclusive and repeated. Assessing p24 concentration. The p24 ELISA assay was performed using a commercially available kit (Alliance HIV-1 p24 ELISA kit; Perkin Elmer, Waltham, MA). The kit range is pg/ml. All measurements were done in duplicate. Psi-gag PCR assessment. Primers used in the psi-gag PCR are: GrecF1 (5-CAGGACTCGGCTTGCTGAA-3), and GrecR2 (5-TGTCTTATGTC CAGAATGCT-3). After PCR amplification, the product is transferred to a nitrocellulose membrane and probed with the P 32 oligo probe GrecP (5-AAGATTTAAACACCATGCTA-3). In addition to test articles, two negative s are used in the PCR assay; water as a no-template and genomic DNA from untransduced C8166 cells. A PCR with 100 copies of pr8.71 plasmid in a background of C8166 genomic DNA serves as the positive. Negative and positive must demonstrate the expected result for the assay to be acceptable. The psi-gag methodology is described in detail in Sastry et al. 18 and the assay has been modified to include a second PCR for human β-globin to validate that the test article DNA is of sufficient quantity and quality. ACKNOWLEDGMENTS The authors would like to thank for assistance from Lakshmi Sastry, Scott Cross, Philip Zoltick, and Donald Kohn. This work was supported by a grant from the National Gene Vector Laboratory (U42RR11148) and the National Gene Vector Biorepository (P40 RR024928). T.H. received training grant support (T32 HL Basic Science Studies on Gene Therapy of Blood Disease). K.C. was supported, in part, by the Indiana Genomic Initiative (INGEN) created through a grant from the Lilly Endowment, Inc. K.C. is a founder of Rimedion Inc. but there is no financial conflict of interest with the work described in this manuscript. REFERENCES 1. Segall, HI, Yoo, E and Sutton, RE (2003). Characterization and detection of artificial replication-competent lentivirus of altered host range. Mol Ther 8: Delenda, C, Audit, M and Danos, O (2002). Biosafety issues in lentivector production. Curr Top Microbiol Immunol 261: Mautino, MR, Ramsey, WJ, Reiser, J and Morgan, RA (2000). Modified human immunodeficiency virus-based lentiviral vectors display decreased sensitivity to transdominant Rev. Hum Gene Ther 11: Srinivasakumar, N and Schuening, FG (1999). A lentivirus packaging system based on alternative RNA transport mechanisms to express helper and gene transfer vector RNAs and its use to study the requirement of accessory proteins for particle formation and gene delivery. J Virol 73: Evans, JT and Garcia, JV (2000). Lentivirus vector mobilization and spread by human immunodeficiency virus. Hum Gene Ther 11: Wu, X, Wakefield, JK, Liu, H, Xiao, H, Kralovics, R, Prchal, JT et al. (2000). Development of a novel trans-lentiviral vector that affords predictable safety. Mol Ther 2: Cronin, J, Zhang, XY and Reiser, J (2005). Altering the tropism of lentiviral vectors through pseudotyping. Curr Gene Ther 5: Dull, T, Zufferey, R, Kelly, M, Mandel, RJ, Nguyen, M, Trono, D et al. (1998). A third-generation lentivirus vector with a conditional packaging system. J Virol 72: Molecular Therapy vol. 19 no. 3 mar