VIROLOGY 180,849-H2 (1991)
Characterization of Recombinant Helper Retroviruses from Moloney-Based Vectors in Ecotropic and Amphotropic Packaging Cell Lines M . SCARPA,`u l D . COURNOYER,* D . M . MUZNY,t K . A . MOORE,* J . W . BELMONT, * 't AND C . T . Caskey*'t *Institute for Molecular Genetics, and tHoward Hughes Medical Institute . Bay/or College of Medicine, Houston, Texas 77030 Received July 6, 1990 ; accepted October 10, 1990
We have characterized the recombinant replication-competent retrovirus (Rev) arising from piN2-derived vectors in the packaging cell lines T2 (ecotropic) and PA317 (amphotropic) . Detailed restriction patterns and sequence of the envelope region of these RCVs has indicated that they arose from recombination events between the virus plasmids used to create the packaging cell line and the vectors . There was no evidence of recombination involving endogenous murine retroviral sequences in the packaging cell line or in transduced hematopoietic cells . In addition, we have confirmed that the mutation of the start codon of the pXM5(N2) derivatives gag+ sequence drastically decreased the occurrence of RCV production . These results offer encouragement that the risk of Rev production can be adequately decreased in gene therapy applications of defective retrovirus vectors, e tssf Academic Press, Inc . Replication-defective retrovirus vectors are commonly used for gene transfer into eukaryotic cells and have been widely investigated for their potential application to somatic gene therapy . The most frequently used vectors derive from the Moloney murine leukemia virus (Me MLV) (1) . The removal of trans-acting sequences (gag, pot, and env) and their substitution with a disease-related coding sequence, drug resistance, or reporter gene while encoding the viral proteins in so-called "packaging" cell lines has allowed creation of defective retrovlruses that can rnedlate highly efficient gene transfer (for a review see (2, 3)) . A major problem with the use of retroviral vectors, particularly for gene therapy applications, is the possible creation of RCVs or "helper-viruses ." These presumably arise from recombination events between homologous sequences of the retroviral vector and the structural genes of Mo-MI V in the packaging cell line . It is formally possible, however, that RCV could originate from recombination between the vector and endogenous retroviral sequences of the target animal or human cells . Ecotropic and amphotropic viruses are the most common types of RCV ; although the production of xenotropic RCVs has been observed in mice that were transplanted with retrovirus-transduced bone marrow cells (4), these RCVs have not been shown to contribute to neoplasia or to the propogation of the vector . Also, it is very unlikely that such phenomena would be observed in humans .
Various strategies have been coined to prevent the generation of RCV (5) . Most are based on the presumption that the helper-virus has its origin in the virus-producing cells . However, none of the existing packaging cell lines can be considered entirely safe from this point of view of potential RCV production . Conventional procedures for the detection of RCV are either time consuming [5-12 days forthe "S+L-" test (6) or 4 weeks for the gene rescue method (7)] or cannot be used to detect amphotropic viruses ["XC" test (6)] . Also these are not very convenient for the detection of RCVs from animal tissues, particularly if these have been stored . In order to characterize the RCVs that were generated during the maintenance of adenosine deaminase (ADA)-transducing vectors (7) in 'P2 (8) and PA3 17 (9), we have developed a rapid method of detection for RCV based on polymerase chain reaction (RCV-PCR) technology (10, 71) . We have used two sets of PCR primers to amplify Moloney sequences between the 3' end of poi and the 5' end of env . In the presence of Moloney-related RCV, these primers will amplify a fragment of 500 by (primers for ecotropic virus) or 700 by (primers for amphotropic virus) from the genomic DNA of NIH3T3 cells which have been exposed to known virus-producing cell line supernatant or directly from the DNA of animal tissues . This assay requires only 36-48 hr for completion and therefore allows the rapid identification of RCV-producing clones (Fig, 1) . We performed an evaluation for the assay's sensitivity using six samples which were positive for helpervirus by the gene rescue method (7) . These were a viral supernatant from the ecotropic packaging cell line 'P2
' To whom requests for reprints should be addressed, at present address : Department of Pediatrics, University of Padova, Via Giustiniani 3, 35125 Padova, Italy . 849
0092-6822/91 $3 .00 copyright A 1991! by Academi Press, Inc. All bghls of reproduction in any form reserved .
850
SHORT COMMUNICATIONS ccotropic amplification 1
2
3
amphotropic amplification 4
a. BstEll
5
6
7
8
9
?. BstEll
700 by 500 by
FIG . 1 . RCV-PCR : Ecotropic amplification : Lane 1, NIH3T3 cells infected with 4124N2ADA ; lane 2, lymph node of a mouse positive for RCV ; lane 2, peripheral blood of a mouse positive for RCV . Amphotropic amplification : Lanes 4 and 5, two clones of NIH3T3 infected with PA317AN25ADA ; lanes 6 and 7, uninfected NIH3T3 amplified with ecotropic and amphotropic oligos, respectively ; Lanes 8 and 9, lymph node and peripheral blood of animals negative for RCV . Ecotropic and amphotropic viral supernatants are collected, filtered, and used to infect 4 x 10^ NIH3T3 cells overnight in the presence of 4 pg/ml of Polybrene (Sigma) . Cells are grown for 36-48 hr. Up to 1 pg of DNA is used in the I00-µl PCR mix (11) with 5 U of AmpliTaq (Perkier Elmer Cetus, Norwalk, CT .) . The samples are denaturated at 94` for 7 min . 30 cycles : annealing, 30 seconds at 55° ; elongation, 2 minutes at 72' ; melting, 1 minute at 94° . Twenty percent of the reaction is analyzed on a 1 .3 0/0 agarose gel . Ecotropic oligonucleotides : (1) 5'CAAGAGTTACTAACAGCCCCTCTCTCCAAG, and (2) 5'TGGTTGCCAGAAGTTGCCCATACCGTCTCC, Amphotropic oligonucleotides : (3) 5'GAACCATCAAGGAGACTTTAACTAAATTAA, and (4) 5'AAGGCATCTTGTACAGTTCCCAGGAGGGAG .
transfected with the pXM5(N2) derivative retroviral vector pAN2ADA (7), two supernatants from the amphotropic packaging cell line PA31 7 infected with GP +E-86 derived RCV-free pdN2yADA virus (M . Scarps etal., in preparation), two lymph nodes, and one peripheral blood sample from transplanted C57BL/6J mice (Jackson Laboratories, Bar Harbor, ME) which had received syngeneic bone marrow cells that had been cocultivated with "P2 pAN2ADA-producing cells . RCV-PCR specifically amplified fragments of 500 or 700 by from ecotropic or amphotropic RCVt samples, respectively . No amplification was obtained from NIH3T3 cells that were exposed to marker rescuenegative virus supernatants . Both assays correlated . RCV-PCR could detect one helper-infected cell in 10 5 uninfected cells (data not shown) . RCV-PCR cannot be applied directly to RNA purified from virus-containing supernatant, however, as copackaging of RNA sequences lacking the packaging signal can occur in some viral particles, leading to false positive results . To investigate the origin of these RCVs, we amplified the entire ecotropic and amphotropic env genes of the above samples and generated a restriction map of these products with the following restriction enzymes : BamHl, Kpnl, Taql, Ncol, EcoRl, Hindlll, Sacl, Nhel, Xbal, Scal, C/al, Pstl, Pvul, Smal, Boll . The restriction
map of the ecotropic products is identical to the map of the Mo-MLV env. One product was sequenced which revealed >95% of homology with the env gene contained in the plasmid pMOV"P- used to generate the 'P2 cell line . Similarly, the restriction map of the amphotropic products was identical to the map of the amphotropic 4070A env gene (12) cloned in the plasmid pPAM3 used to generate the PA317 packaging cell line, and the sequence of one product showed >92% sequence homology with the 4070A env gene . To our knowledge, this is the first molecular characterization of RCV resulting from recombination events between defective retrovirus vectors and the proviral sequence used to generate packaging cell lines . Two major regions of homology are shared between the vectors and the proviral packaging constructions and could potentially be involved in the recombination events (Fig . 2) . The first region of homology occurs upstream from the ADA cDNA sequence and is from bases 2653 to 3126 of pAN2ADA ; the homology between pXMS derivatives and the viral structural genes of the packaging cell lines is 100% in that region . This fragment (derived from the Mo-MLV gag sequence) was included in the pXM5(N2) vector to improve the efficiency of vector packaging ; it results in an approximately 10-fold increase in titer (13) . This region of ho-
SHORT COMMUNICATIONS
85 1
RCV PRODUCTION gag pMOVV'-
pot
env
LTR
pXMS(N2)
LTR
+
`P
I
+
pPAM3
gag
pot
4070A env
a
pMOVw-
I F~\-~ 11 p
p4N2stADA 'P
pPAM3
I 1-
7 a
FIG . 2 . Mechanisms of generation of RCV . Recombination events exploiting homology regions between pXM5(N2) derivative vectors and the packaging plasmid pNOv4'-integrated in T2 cells or with the plasmid pPAM3 used to create the PA317 cells lead to the creation of RCV . the substitution of the gag+ start codon AUG with the stop codon UAA terminates the translation of the gag, poi, and env genes, therefore diminishing the risk of RCV production (see text for details) . (®) = Bacterial transposon Tn5 phosphotransferase gene conferring resistance to aminoglycoslde G418 ; (t) SV40 poly(A) signal ; (•) ADA cDNA .
mology is in fact sufficient to generate RCV when pXMS derivatives are produced in d/2 cells . The second region of homology occurs downstream from the ADA cDNA between bases 4805 and 4906 of pAN2ADA . It corresponds to bases 1904 through
to a UAA stop codon (pAN2stADA vector) . The early termination of translation generated by this alteration
2002 of the 4070A sequence with a degree of homol-
successful use of this concept to diminish the risk of RCV production from the pLNL6 vector maintained in the packaging cel' lines PA317 and PE501 (14) . We
ogy of 93% . Recombination events involving this region are also required to reconstitute an intact RCV genome from vectors produced in PA317 cells . This results in a much lower frequency of RCV production in PA31 7 than in 'P2 cells (9) . Our data show that ecotropic and amphotropic helper-virus generated in the 'V2 and PA317 cell lines derived from recombination with the plasmid used to generate these cell lines . This suggests that most recombination events take place between homologous sequences which have been introduced into the cells . Thus, by modifying these sequences, we should be able to decrease significantly the incidence of RCV generation . As a precaution against the generation of RCV, we mutated the normal start codon of the gag+ sequence
should reduce the chance of production of RCV unless there was an additional recombination or gene conversion event . Another group has previously reported the
wanted to assess the effect of a similar modification in our ADA-transducing vector (pAN2ADA) and to evaluate its efficiency in preventing RCV generation in the packaging cell line *2, which is more prone to produce RCV in presence of pXMS(N2) derivatives . This modification did in fact prevent the production of RCV when pAN2stADA was maintained for several months in *Y2 cells . We have, in fact, transfected *2 cells with pAN2ADA and pAN2stADA constructions and followed the RCV production with gene rescue assays and RCV-PCR for a period of over 28 weeks . As expected, pAN2ADA generated RCV after the third week after transfection ; in contrast, pAN2stADA did not pro-
852
SHORT COMMUNICATIONS
duce any RCV for the entire length of the experiment . The introduction of this mutation was also shown to have no effect on the expression of the transduced human ADA in human hematopoietic cells (D . Cournoyer et al., in preparation) . In conclusion, the generation of RCV from replication-defective retrovirus vector appears to be a preventable problem . In fact, safer packaging cell lines have now been created in which the structural genes for the production of retrovirus have been split between two plasmids, increasing to three the minimum number of recombination events required to form an intact viral genome (15, 16) . Using these cell lines to produce pAN2ADA and pzN2stADA, we did not observe the production of ecotropic oramphotropic RCV . These safer designs represent important progress toward the development of gene therapy approach applicable to human disorders .
ACKNOWLEDGMENTS The authors want to thank Dr . R . C. Mulligan, from the Massachusetts Institute of Technology, for the plasmid containing the µ-chain immunoglobulin enhancer and Dr . G . R . MacGregorforeritical reading of the manuscript . This work has been supported by the Howard Hughes Medical Institute (1 .W .B . and C .T.C .) . M .S . Was the recipient of a Cystic Fibrosis Foundation Postdoctoral Fellowship F054-9 . D .C . was the recipient of a Senior Fellowship from the National Cancer Institute of Canada . K .A .M . is the recipient of U .S .P .H .S . Individual National Research Service Award F32 RR05034 .
REFERENCES 7 . SCHINNICK, T . M ., LERNER, R. A ., and SUTCLIFFE. 1 . G ., Nature (London) 293, 543 548 (1981) . 2. FRIEDMAN, T ., Science 244, 1275-1281 (1989) . 3. ScARPA, M ., JONES, S . N ., and CASKEY, C .T ., Corr. Opin . Ped. 1, 453-464(1989). 4, KALEKO, M ., GARCIA, J . V ., OSBORNE, W . R . A ., and MILLER, A . D ., Blood 8, 1/32-1741 11990) . 5. MILLER, A . D ., ,Hum. Gene The . . 1, 5-14 (1990) . 6. JAKOBY, W. B . and PASTAN, I . H ., Eds ., "Methods in Enzymology," Vol . LVIII, pp . 421-424 . Academic Press, San Diego, 19/9 . 7. BELMONT, J . W ., MACGREGOR, G . R ., WAGER-SMITH, K ., FLETCHER, F . A ., MOORE, K . A ., HAWKINS, D ., VILLALON, D ., CHANG, S . M .-W ., and CASKEY, C . T ., Mol. Cell. Biol. 8, 5116-5125 (1988) . S. MANN, R ., MULLIGAN, R . C ., and BALTIMORE, D ., Ce1733, 153-159 (1983) 9. MILLER, A . D ., and BUTTIIMORE, C ., )Vol. Cell. 8iol. 6, 2895-2902 (1986) . 10 . SAIKI, R . K ., GELFAND, D . H ., STOFFEL, S ., SCHARF, S . J ., HIGUCHI, R . . HORN . G . T ., MULLIS K . B ., and ERLICH, H . A ., Science 239, 487-491(1988). 17 . KOGAN, S . C ., DOHERTY, M ., GITSOHIER, 1 ., N . EngL J. Med . 317, 985-990(1987). 12 . OTT, D ., FRIEDERICH, R ., and REIN, A ., J. Virol, 64, 757-766 (1990) . 13 . ARMENTANO, D . . Yu, S . F . . KANTOFF, P . W ., VON RUIDEN . T., ANDERSON, W . F ., and GILBOA, E ., 1. V7rel. 61, 1647-1650 (1987) . 14 . MILER, A . D ., and ROSMAN, G . J ., Biotechniques7, 980-990 (1989) . 15. MARKOWITZ, D ., GOFF, S., and BANK, A .,1. Virol. 62, 1120-1124 (1988) . 16 . MARKOWITZ . D ., GOFF, S . . and BANK, A ., Virology 167, 400-406 (1989) .
Characterization of Recombinant Helper Retroviruses from Moloney-Based Vectors in Ecotropic and Amphotropic Packaging Cell Lines M . SCARPA,`u l D . COURNOYER,* D . M . MUZNY,t K . A . MOORE,* J . W . BELMONT, * 't AND C . T . Caskey*'t *Institute for Molecular Genetics, and tHoward Hughes Medical Institute . Bay/or College of Medicine, Houston, Texas 77030 Received July 6, 1990 ; accepted October 10, 1990
We have characterized the recombinant replication-competent retrovirus (Rev) arising from piN2-derived vectors in the packaging cell lines T2 (ecotropic) and PA317 (amphotropic) . Detailed restriction patterns and sequence of the envelope region of these RCVs has indicated that they arose from recombination events between the virus plasmids used to create the packaging cell line and the vectors . There was no evidence of recombination involving endogenous murine retroviral sequences in the packaging cell line or in transduced hematopoietic cells . In addition, we have confirmed that the mutation of the start codon of the pXM5(N2) derivatives gag+ sequence drastically decreased the occurrence of RCV production . These results offer encouragement that the risk of Rev production can be adequately decreased in gene therapy applications of defective retrovirus vectors, e tssf Academic Press, Inc . Replication-defective retrovirus vectors are commonly used for gene transfer into eukaryotic cells and have been widely investigated for their potential application to somatic gene therapy . The most frequently used vectors derive from the Moloney murine leukemia virus (Me MLV) (1) . The removal of trans-acting sequences (gag, pot, and env) and their substitution with a disease-related coding sequence, drug resistance, or reporter gene while encoding the viral proteins in so-called "packaging" cell lines has allowed creation of defective retrovlruses that can rnedlate highly efficient gene transfer (for a review see (2, 3)) . A major problem with the use of retroviral vectors, particularly for gene therapy applications, is the possible creation of RCVs or "helper-viruses ." These presumably arise from recombination events between homologous sequences of the retroviral vector and the structural genes of Mo-MI V in the packaging cell line . It is formally possible, however, that RCV could originate from recombination between the vector and endogenous retroviral sequences of the target animal or human cells . Ecotropic and amphotropic viruses are the most common types of RCV ; although the production of xenotropic RCVs has been observed in mice that were transplanted with retrovirus-transduced bone marrow cells (4), these RCVs have not been shown to contribute to neoplasia or to the propogation of the vector . Also, it is very unlikely that such phenomena would be observed in humans .
Various strategies have been coined to prevent the generation of RCV (5) . Most are based on the presumption that the helper-virus has its origin in the virus-producing cells . However, none of the existing packaging cell lines can be considered entirely safe from this point of view of potential RCV production . Conventional procedures for the detection of RCV are either time consuming [5-12 days forthe "S+L-" test (6) or 4 weeks for the gene rescue method (7)] or cannot be used to detect amphotropic viruses ["XC" test (6)] . Also these are not very convenient for the detection of RCVs from animal tissues, particularly if these have been stored . In order to characterize the RCVs that were generated during the maintenance of adenosine deaminase (ADA)-transducing vectors (7) in 'P2 (8) and PA3 17 (9), we have developed a rapid method of detection for RCV based on polymerase chain reaction (RCV-PCR) technology (10, 71) . We have used two sets of PCR primers to amplify Moloney sequences between the 3' end of poi and the 5' end of env . In the presence of Moloney-related RCV, these primers will amplify a fragment of 500 by (primers for ecotropic virus) or 700 by (primers for amphotropic virus) from the genomic DNA of NIH3T3 cells which have been exposed to known virus-producing cell line supernatant or directly from the DNA of animal tissues . This assay requires only 36-48 hr for completion and therefore allows the rapid identification of RCV-producing clones (Fig, 1) . We performed an evaluation for the assay's sensitivity using six samples which were positive for helpervirus by the gene rescue method (7) . These were a viral supernatant from the ecotropic packaging cell line 'P2
' To whom requests for reprints should be addressed, at present address : Department of Pediatrics, University of Padova, Via Giustiniani 3, 35125 Padova, Italy . 849
0092-6822/91 $3 .00 copyright A 1991! by Academi Press, Inc. All bghls of reproduction in any form reserved .
850
SHORT COMMUNICATIONS ccotropic amplification 1
2
3
amphotropic amplification 4
a. BstEll
5
6
7
8
9
?. BstEll
700 by 500 by
FIG . 1 . RCV-PCR : Ecotropic amplification : Lane 1, NIH3T3 cells infected with 4124N2ADA ; lane 2, lymph node of a mouse positive for RCV ; lane 2, peripheral blood of a mouse positive for RCV . Amphotropic amplification : Lanes 4 and 5, two clones of NIH3T3 infected with PA317AN25ADA ; lanes 6 and 7, uninfected NIH3T3 amplified with ecotropic and amphotropic oligos, respectively ; Lanes 8 and 9, lymph node and peripheral blood of animals negative for RCV . Ecotropic and amphotropic viral supernatants are collected, filtered, and used to infect 4 x 10^ NIH3T3 cells overnight in the presence of 4 pg/ml of Polybrene (Sigma) . Cells are grown for 36-48 hr. Up to 1 pg of DNA is used in the I00-µl PCR mix (11) with 5 U of AmpliTaq (Perkier Elmer Cetus, Norwalk, CT .) . The samples are denaturated at 94` for 7 min . 30 cycles : annealing, 30 seconds at 55° ; elongation, 2 minutes at 72' ; melting, 1 minute at 94° . Twenty percent of the reaction is analyzed on a 1 .3 0/0 agarose gel . Ecotropic oligonucleotides : (1) 5'CAAGAGTTACTAACAGCCCCTCTCTCCAAG, and (2) 5'TGGTTGCCAGAAGTTGCCCATACCGTCTCC, Amphotropic oligonucleotides : (3) 5'GAACCATCAAGGAGACTTTAACTAAATTAA, and (4) 5'AAGGCATCTTGTACAGTTCCCAGGAGGGAG .
transfected with the pXM5(N2) derivative retroviral vector pAN2ADA (7), two supernatants from the amphotropic packaging cell line PA31 7 infected with GP +E-86 derived RCV-free pdN2yADA virus (M . Scarps etal., in preparation), two lymph nodes, and one peripheral blood sample from transplanted C57BL/6J mice (Jackson Laboratories, Bar Harbor, ME) which had received syngeneic bone marrow cells that had been cocultivated with "P2 pAN2ADA-producing cells . RCV-PCR specifically amplified fragments of 500 or 700 by from ecotropic or amphotropic RCVt samples, respectively . No amplification was obtained from NIH3T3 cells that were exposed to marker rescuenegative virus supernatants . Both assays correlated . RCV-PCR could detect one helper-infected cell in 10 5 uninfected cells (data not shown) . RCV-PCR cannot be applied directly to RNA purified from virus-containing supernatant, however, as copackaging of RNA sequences lacking the packaging signal can occur in some viral particles, leading to false positive results . To investigate the origin of these RCVs, we amplified the entire ecotropic and amphotropic env genes of the above samples and generated a restriction map of these products with the following restriction enzymes : BamHl, Kpnl, Taql, Ncol, EcoRl, Hindlll, Sacl, Nhel, Xbal, Scal, C/al, Pstl, Pvul, Smal, Boll . The restriction
map of the ecotropic products is identical to the map of the Mo-MLV env. One product was sequenced which revealed >95% of homology with the env gene contained in the plasmid pMOV"P- used to generate the 'P2 cell line . Similarly, the restriction map of the amphotropic products was identical to the map of the amphotropic 4070A env gene (12) cloned in the plasmid pPAM3 used to generate the PA317 packaging cell line, and the sequence of one product showed >92% sequence homology with the 4070A env gene . To our knowledge, this is the first molecular characterization of RCV resulting from recombination events between defective retrovirus vectors and the proviral sequence used to generate packaging cell lines . Two major regions of homology are shared between the vectors and the proviral packaging constructions and could potentially be involved in the recombination events (Fig . 2) . The first region of homology occurs upstream from the ADA cDNA sequence and is from bases 2653 to 3126 of pAN2ADA ; the homology between pXMS derivatives and the viral structural genes of the packaging cell lines is 100% in that region . This fragment (derived from the Mo-MLV gag sequence) was included in the pXM5(N2) vector to improve the efficiency of vector packaging ; it results in an approximately 10-fold increase in titer (13) . This region of ho-
SHORT COMMUNICATIONS
85 1
RCV PRODUCTION gag pMOVV'-
pot
env
LTR
pXMS(N2)
LTR
+
`P
I
+
pPAM3
gag
pot
4070A env
a
pMOVw-
I F~\-~ 11 p
p4N2stADA 'P
pPAM3
I 1-
7 a
FIG . 2 . Mechanisms of generation of RCV . Recombination events exploiting homology regions between pXM5(N2) derivative vectors and the packaging plasmid pNOv4'-integrated in T2 cells or with the plasmid pPAM3 used to create the PA317 cells lead to the creation of RCV . the substitution of the gag+ start codon AUG with the stop codon UAA terminates the translation of the gag, poi, and env genes, therefore diminishing the risk of RCV production (see text for details) . (®) = Bacterial transposon Tn5 phosphotransferase gene conferring resistance to aminoglycoslde G418 ; (t) SV40 poly(A) signal ; (•) ADA cDNA .
mology is in fact sufficient to generate RCV when pXMS derivatives are produced in d/2 cells . The second region of homology occurs downstream from the ADA cDNA between bases 4805 and 4906 of pAN2ADA . It corresponds to bases 1904 through
to a UAA stop codon (pAN2stADA vector) . The early termination of translation generated by this alteration
2002 of the 4070A sequence with a degree of homol-
successful use of this concept to diminish the risk of RCV production from the pLNL6 vector maintained in the packaging cel' lines PA317 and PE501 (14) . We
ogy of 93% . Recombination events involving this region are also required to reconstitute an intact RCV genome from vectors produced in PA317 cells . This results in a much lower frequency of RCV production in PA31 7 than in 'P2 cells (9) . Our data show that ecotropic and amphotropic helper-virus generated in the 'V2 and PA317 cell lines derived from recombination with the plasmid used to generate these cell lines . This suggests that most recombination events take place between homologous sequences which have been introduced into the cells . Thus, by modifying these sequences, we should be able to decrease significantly the incidence of RCV generation . As a precaution against the generation of RCV, we mutated the normal start codon of the gag+ sequence
should reduce the chance of production of RCV unless there was an additional recombination or gene conversion event . Another group has previously reported the
wanted to assess the effect of a similar modification in our ADA-transducing vector (pAN2ADA) and to evaluate its efficiency in preventing RCV generation in the packaging cell line *2, which is more prone to produce RCV in presence of pXMS(N2) derivatives . This modification did in fact prevent the production of RCV when pAN2stADA was maintained for several months in *Y2 cells . We have, in fact, transfected *2 cells with pAN2ADA and pAN2stADA constructions and followed the RCV production with gene rescue assays and RCV-PCR for a period of over 28 weeks . As expected, pAN2ADA generated RCV after the third week after transfection ; in contrast, pAN2stADA did not pro-
852
SHORT COMMUNICATIONS
duce any RCV for the entire length of the experiment . The introduction of this mutation was also shown to have no effect on the expression of the transduced human ADA in human hematopoietic cells (D . Cournoyer et al., in preparation) . In conclusion, the generation of RCV from replication-defective retrovirus vector appears to be a preventable problem . In fact, safer packaging cell lines have now been created in which the structural genes for the production of retrovirus have been split between two plasmids, increasing to three the minimum number of recombination events required to form an intact viral genome (15, 16) . Using these cell lines to produce pAN2ADA and pzN2stADA, we did not observe the production of ecotropic oramphotropic RCV . These safer designs represent important progress toward the development of gene therapy approach applicable to human disorders .
ACKNOWLEDGMENTS The authors want to thank Dr . R . C. Mulligan, from the Massachusetts Institute of Technology, for the plasmid containing the µ-chain immunoglobulin enhancer and Dr . G . R . MacGregorforeritical reading of the manuscript . This work has been supported by the Howard Hughes Medical Institute (1 .W .B . and C .T.C .) . M .S . Was the recipient of a Cystic Fibrosis Foundation Postdoctoral Fellowship F054-9 . D .C . was the recipient of a Senior Fellowship from the National Cancer Institute of Canada . K .A .M . is the recipient of U .S .P .H .S . Individual National Research Service Award F32 RR05034 .
REFERENCES 7 . SCHINNICK, T . M ., LERNER, R. A ., and SUTCLIFFE. 1 . G ., Nature (London) 293, 543 548 (1981) . 2. FRIEDMAN, T ., Science 244, 1275-1281 (1989) . 3. ScARPA, M ., JONES, S . N ., and CASKEY, C .T ., Corr. Opin . Ped. 1, 453-464(1989). 4, KALEKO, M ., GARCIA, J . V ., OSBORNE, W . R . A ., and MILLER, A . D ., Blood 8, 1/32-1741 11990) . 5. MILLER, A . D ., ,Hum. Gene The . . 1, 5-14 (1990) . 6. JAKOBY, W. B . and PASTAN, I . H ., Eds ., "Methods in Enzymology," Vol . LVIII, pp . 421-424 . Academic Press, San Diego, 19/9 . 7. BELMONT, J . W ., MACGREGOR, G . R ., WAGER-SMITH, K ., FLETCHER, F . A ., MOORE, K . A ., HAWKINS, D ., VILLALON, D ., CHANG, S . M .-W ., and CASKEY, C . T ., Mol. Cell. Biol. 8, 5116-5125 (1988) . S. MANN, R ., MULLIGAN, R . C ., and BALTIMORE, D ., Ce1733, 153-159 (1983) 9. MILLER, A . D ., and BUTTIIMORE, C ., )Vol. Cell. 8iol. 6, 2895-2902 (1986) . 10 . SAIKI, R . K ., GELFAND, D . H ., STOFFEL, S ., SCHARF, S . J ., HIGUCHI, R . . HORN . G . T ., MULLIS K . B ., and ERLICH, H . A ., Science 239, 487-491(1988). 17 . KOGAN, S . C ., DOHERTY, M ., GITSOHIER, 1 ., N . EngL J. Med . 317, 985-990(1987). 12 . OTT, D ., FRIEDERICH, R ., and REIN, A ., J. Virol, 64, 757-766 (1990) . 13 . ARMENTANO, D . . Yu, S . F . . KANTOFF, P . W ., VON RUIDEN . T., ANDERSON, W . F ., and GILBOA, E ., 1. V7rel. 61, 1647-1650 (1987) . 14 . MILER, A . D ., and ROSMAN, G . J ., Biotechniques7, 980-990 (1989) . 15. MARKOWITZ, D ., GOFF, S., and BANK, A .,1. Virol. 62, 1120-1124 (1988) . 16 . MARKOWITZ . D ., GOFF, S . . and BANK, A ., Virology 167, 400-406 (1989) .
