Proc. Nati. Acad. Sci. USA Vol. 82, pp. 6927-6931, October 1985 Cell Biology Insertion of the bacterial gpt gene into the germ line of mice by retroviral infection (preimplantation mouse embryos/germ-line integration/DNA methylatlon/5-azacytidine/gene expression) DETLEV JAHNER*t, KIRSTEN HAASE*, RICHARD MULLIGANt, AND RUDOLF JAENISCH*t *Heinrich-Pette-Institut fur Experimentelle Virologie und Immunologie an der Universitat Hamburg, Martinistr. 52, 2000 Hamburg 20, Federal Republic of Germany; and tWhitehead Institute for Biomedical Research, and Department of Biology, Massachusetts Institute of Technology, 9 Cambridge Center, Cambridge, MA 02142 Communicated by Robert A. Weinberg, July 1, 1985

ABSTRACT Mouse substrains genetically transmitting the Retroviral vectors have been constructed that can exogenous Moloney murine leukemia virus (Mo-MuLV) at a transduce foreign genes into a variety ofcell types (12-22). In single locus have been derived previously by infection of previous experiments, a vector transducing a bacterial mark- preimplantation embryos. Here we explore the potential of er gene has been introduced into postimplantation mouse retroviral vectors for transferring nonviral genes into the germ embryos by injection of virus, and the transduced gene was line ofmice. Preimplantation mouse embryos were cocultivated shown to be active in many somatic tissues ofanimals derived with a cell line that produces a recombinant retrovirus whose from injected embryos (17). Viral insertion into germ line genome carries the Escherichia cofi gene gpt. We show that the cells was, however, not obtained in these experiments. vector sequence was inserted into the genome of the embryo To explore the potential of retroviral vectors to transfer and into the germ line at a frequency similar to that for the nonretroviral genes into the germ line of mice, we have Mo-MuLV-helper sequence. A new mouse strain, Mgpt-1, was exposed preimplantation embryos to MSVgpt virus, which is developed that is homozygous for a single MSVgpt proviral a Mo-MuLV-derived vector carrying the Escherichia coli genome. The proviral sequences were highly methylated and gene gpt. We show that the vector is inserted into germ-line not expressed in tissues of Mgpt-1 mice. When cells derived cells at a similar frequency as Mo-MuLV helper virus. The from transgeneic animals were treated with 5-azacytidine, the inserted gene is highly methylated and transcriptionally proviral sequences were not methylated and were transcrip- inactive but can be activated by treatment of cells with tionally activated. These results indicate that nonviral genes 5-azacytidine. that are under the control of the viral long terminal repeat are inactivated when transferred into the germ line of animals. MATERIALS AND METHODS The transfer of foreign genes into an animal's germ line represents a powerful tool for the study of gene regulation Mice. Embryos were derived from mice of strain 129. The during mammalian development. To date, the primary meth- origin of Mov-9 and Mov-13 mice has been described (8). od employed to generate transgeneic mice has involved the Cell Lines and Infection of Embryos. NIH 3T3, XC, and microinjection of recombinant DNA into the zygote Cl-1-lA cells were grown as described (23). The gpt-7-3T3 pronucleus (1-5). Although resulting mouse strains show cell line, derived by infection of NIH 3T3 cells with MSVgpt variable expression of the inserted gene, in a number of virus and Mo-MuLV, has been described (24). Four-to-eight- instances the expected pattern of tissue-specific expression cell mouse embryos were isolated and infected by cocultiva- was observed. A potential alternative to the method of tion with gpt-7-3T3 cells as described (7, 8). Primary Cell Culture. Lung tissue of Mgpt-1 mice was microinjection involves the exposure of early embryos to sliced, rinsed in Dulbecco's phosphate-buffered saline, and retroviruses (6-8). In contrast to the complex, tandemly incubated for 60 min at 370C in Dulbecco's modified Eagle's repeated structure that is frequently observed after introduc- medium supplemented with 10% fetal bovine serum and with tion of DNA into the germ line by microinjection, retroviral and cells infection of preimplantation embryos leads to the stable 0.25% collagenase (Sigma). Bone marrow spleen well-defined that can be were prepared as described (25). integration of single, genomes gpt Virus Assay. Primary lung fibroblasts (2 x 10W) were analyzed easily at the DNA and RNA levels as well as by infected with Mo-MuLV from Cl-i-lA cells. Two weeks assays for viral infectivity. Exposure of preimplantation later, 2 x 10i of these cells were cocultivated for 3 days with mouse embryos to Moloney leukemia virus (Mo-MuLV) 5 x results in efficient integration of proviral sequences into the 105 NIH 3T3 cells in the presence of Polybrene (Aldrich) derived and were subsequently tested for growth in selective medium embryo's genome and into the germ line of animals containing mycophenolic acid (25 ,ug/ml) as selective drug from the infected embryos (6-8). The viral sequences are de (26). novo methylated soon after infection and are not expressed Similarly, 107 bone marrow or spleen cells were coculti- (9). Upon further development of the embryo, these inserted with 2 in vated for 3 days in the presence of Polybrene x 10 proviral sequences remain methylated and suppressed NIH 3T3 cells in RPMI 1640 medium/10% fetal bovine serum most differentiated cells (9). In contrast, retroviruses intro- supplemented with either 10% supernatant from WEHI-3 duced into postimplantation embryos can replicate in an cells (for bone marrow cells) or 50 gg of bacterial lipopoly- unrestricted manner and remain unmethylated in all somatic These media were tissues ofthe developing embryo (for review, see refs. 10 and saccharide per ml (for spleen cells). kindly 11). Abbreviations: Mo-MuLV, Moloney murine leukemia virus; MSVgpt virus, a Mo-MuLV-derived vector containing the Esche- The publication costs of this article were defrayed in part by page charge richia coli gene gpt; SV40, simian virus 40; wt, wild-type. payment. This article must therefore be hereby marked "advertisement" tPresent address: Salk Institute, P.O. Box 85800, San Deigo, CA in accordance with 18 U.S.C. §1734 solely to indicate this fact. 92138. 6927 Downloaded by guest on September 28, 2021 'orf7%28Cell Biology: Rihner et al. Proc. Natl. Acad. Sci. USA 82 (1985) by T. Franz (Heinrich-Pette-Institut). Three days B provided A E P+ S X Pv con- later, the cells were tested for growth under selective a b c d ditions (26). -E } Analysis of DNA and RNA. All restriction enzymes were obtained from Boehringer Mannheim or from Bethesda Research Laboratories and used as recommended by the supplier. Preparation of high molecular weight DNA, elec- trophoresis in 0.8% agarose gels, blotting to nylon mem- branes, and hybridizations were as described (27). Plasmid pSV2 (24) was used to detect gpt and simian virus 40 (SV40) sequences. A BamHII-HindIII fragment from plasmid pL10 (24) served as a gpt-specific probe. al (I) collagen sequences were detected with plasmid HF677 (28). The synthesis of a "Dot cDNA specific for Mo-MuLV has been described (7). .~~~~~~SS i blot" hybridizations were performed with 5 jxg of DNA prepared from tail, and RNA was prepared and analyzed as described (28). gpt [Xanthine (Guanine) Phosphoribosyltransferase, EC 2.4.2.22] Assay. Extracts were prepared from 2-10 x 106 cells by repeated freezing and thawing in 50 mM Tris Cl (pH 7.5). Extract (10 jig of protein) was incubated for 60 min at 370C (29), using ['Cixanthine (Amersham) instead of [14C]gua- FIG. 1. (Upper) Restriction enzyme analyses of mouse 100-19 nine. The amount of [14C]xanthosine 5'-phosphate was de- and its offspring. pSV2 (24) was the hybridization probe (see termined by binding to DEAE plates and measurement of Materials and Methods). (A) EcoRI- and Sst I-digested DNAs from radioactivity in a liquid scintillation counter. liver ofmouse 100-19 (lanes a and b, respectively) and from tail ofone of his first-generation offspring (lanes c and d, respectively). EcoRI does not cleave the MSVgpt provirus (see band E); Sst I cuts in the RESULTS long terminal repeats (open bars in map) to yield bandS (see fragment S above map). Due to incomplete transfer of larger fragments to the nylon membrane, bands generated by EcoRI were less intense than Germ-Line Integration of an MSVgpt Proviral Genome and those derived by Sst I digestion of the same amount of DNA. (B) Derivation of Mice Homozygous at the Mgpt-1 Locus. Analysis of the MSVgpt provirus (see map) in DNA from 100-19 Preimplantation mouse embryos were cocultivated overnight offspring (lanes a) and from the gpt-7-3T3 cell line (lanes b) by the with gpt-7-3T3 cells (7, 8). This cell line produces both enzymes indicated above each pair of lanes: E, EcoRV; P, Pst I; 5, MSVgpt and Mo-MuLV helper virus, the latter at a 10-fold Sst I; X, Xba I; Nv, Pvu II. (Lower) Restriction map of MSVgpt excess (see Table 4). Seven mice were derived from infected provirus. Restriction sites are abbreviated as described above; and tail DNAs were analyzed for the presence of additional sites: A, Ava I. The gpt sequence within the provirus (24) embryos the hatched bar. proviral sequences. Three mice were shown to carry only is indicated by while one animal had Mo- Mo-MuLV helper sequences, named MuLV helper as well as MSVgpt proviral sequences inte- sion of the MSVgpt genome. This proviral locus was grated in its genome. This animal (no. 100-19) was bred with Mgpt-1. normal mice to test for genetic transmission of proviral To derive a mouse strain homozygous for the Mgpt-1 sequences. Restriction enzyme analyses of DNA from male locus, heterozygous animals were mated and the number of 100-19 and from his offspring are shown in Fig. 1. Digestion MSVgpt copies present in DNA from the offspring was of DNA from animal 100-19 with EcoRI, an enzyme which determined. Fig. 2 shows a Southern blot of Sst I digested does not cleave the Mo-MuLV genome, showed the presence DNA hybridized with the gpt probe and with an al(I) collagen of six copies of helper virus in different organs (data not probe, which served as an internal standard. A relatively shown). Most of these proviral copies were present at less higher intensity of the MSVgpt band was detected in 2 of 14 than one copy per nucleus, revealing mosaicism ofthe animal offspring, suggesting that these mice were homozygous at the as discussed previously (7, 8). Some of these bands were Mgpt-1 locus. This was confirmed genetically. Breeding of recovered in the offspring of male 100-19 (Table 1), which the homozygous animals resulted in offspring of only indicates genetic transmission of Mo-MuLV sequences. homozygous genotype, whereas backcrossing to the parental The presence and structure of MSVgpt sequences were strains resulted in 100% heterozygous offspring (Table 2). analyzed by hybridizing the gpt-specific probe to Southern blots. Liver DNA of male 100-19 revealed a single, very faint band after EcoRI digestion (Fig. 1A, lane a); this band was Table 1. Genetic transmission of Mo-MuLV and MSVgpt present in 2 of 21 tested offspring (Fig. lA, lane c; Table 1) sequences to 21 offspring of male 100-19 at a higher intensity. This indicates that male 100-19 was mosaic for a single MSVgpt provirus that was transmitted Genotype No. of offspring genetically to some of his offpsring. Digestion with Sst I, Mo-MuLV/wt* iit which cleaves in the viral long terminal repeat (LTR), MSVgpt/wt 2* confirmed the presence of a single proviral copy (Fig. LA, wt/wt 8 Digestion of DNA from offspring carrying lanes b and d). 100-19 was mated with mice of strain 129 (wt/wt). sequences and of DNA from gpt-7-3T3 cells with Mouse parental MSVgpt The genotypes of offspring were determined by Sst I digestion and different enzymes that cleave within the proviral genome Southern blotting oftail DNAs, using a Mo-MuLV-specific cDNA (7) gave identical internal fragments (Fig. 1B; compare restric- or the gpt probe pSV2 (24). tion map of MSVgpt provirus, Fig. 1 Lower), indicating that *wt, Wild-type "allele" (i.e., no proviral integration). the integrated MSVgpt provirus was not rearranged. When tEcoRI analysis revealed differences in genotypes among these animals carrying a single provirus were bred with animals of offspring due to the segregation of the multiple Mo-MuLV genomes the parental strain 129, 50% of the offspring inherited the present in the germ line of mouse 100-19. proviral sequence (Table 2), indicating Mendelian transmis- *MSVgpt/wt mice carried no Mo-MuLV sequences. Downloaded by guest on September 28, 2021 Cell Biology: Jahner et al. Proc. Natl. Acad. Sci. USA 82 (1985) 6929

Table 2. Derivation of mice homozygous at the Mgpt-J locus a b c d e f g h i -5.3 No. of offspring

Cross Mgpt-1/Mgpt-1 Mgpt-l/wt wt/wt mgpt - .21.:0 Mgpt-1/wt x wt/wt 0 28 29 Mgpt-l/wt x Mgpt-l/wt 2 9 3 SV40 Mgpt-l/Mgpt-1 x Mgpt-1/Mgpt-I 7 0 0 Mgpt-1/Mgpt-J x wt/wt 0 18 0 The genotypes of offspring from matings with wt/wt mice were analyzed by dot blot hybridization using plasmid pSV2; the geno- types of all other mice were determined by quantitative Southern blotting as described in the legend to Fig. 2.

These results show that animals homozygous at the Mgpt-J FIG. 3. RNA blot analysis of Mgpt-1 fibroblasts and of locus (Mgpt-1 mice) are fully viable. mycophenolic acid-resistant derivatives. RNAs were prepared and MSVgpt Provirus Is Not Expressed in Mgpt-1 Mice. To analyzed as described in Materials and Methods with probe pSV2 examine the expression of the genetically transmitted (lane a) or a gpt-specific fragment (lanes b-i). Lane a: hybridization MSVgpt provirus, RNA was prepared from various tissues of of the SV40 sequences present in the Mgpt-1 cell line. Lane b: the Mgpt-1 mice and analyzed by blot hybridization. Because no absence of gpt sequences in the same RNA. Lanes c, e, f, and g: hybridization to gpt sequences was detectable, a virus- hybridization of gpt (bands Mgpt) from Mgpt-1 mycophenolic acid- colony assay was applied that has been shown to detect the resistant clones 3, 4, 5, and 7, respectively (compare Tables 3 and 4). i: of expression of gpt-transducing vector in a small fraction of Lanes d, h, and RNA from 3T3/Mgpt3 cells (see legend Table animal cells For 4), gpt-7-3T3 cells, and NIH 3T3 cells, respectively. Markers of 5.3 (17). this, primary lung fibroblasts isolated and 2.0 kilobases were derived by digestion of pSV2 with Pvu II and were from Mgpt-1 mice infected with helper virus. When with Pvu II plus BamHI, respectively, and run in parallel. NIH 3T3 cells were exposed to culture medium from the Mo-MuLV-infected lung cells and grown under selective conditions to detect MSVgpt virus, no evidence of viral Induced by 5-Azacytidine. Fibroblasts from Mgpt-1 mice were rescue was obtained. Similarly, an attempt to infect NIH 3T3 transformed with SV40 to produce permanent tissue culture cells by cocultivation of lung fibroblasts, bone marrow cells, lines and then were analyzed for Mgpt-J expression by RNA was no or spleen cells from (Mgpt-1 x Mov-13)Fl and (Mgpt-1 x blot analysis. Although SV40 RNA present, MSVgpt Mov-9)F1 mice failed to indicate activation of MSVgpt virus. transcripts were detected (Fig. 3, lanes a and b). As expected, Because the endogenous Mo-MuLV in Mov-9 and Mov-13 the cells were sensitive to treatment with mycophenolic acid. mice is expressed in most somatic cells (8, 23) and therefore Treatment of these cells with 5-azacytidine, a drug that could have rescued the MSVgpt virus, these experiments reduces the level of DNA methylation in mammalian cells strongly support that the MSVgpt provirus is not transcribed (30), induced mycophenolic acid-resistant clones (Table 3). in cells of Mgpt-1 mice. Blot analysis ofRNA from these resistant clones showed the presence of a 3.5-kilobase MSVgpt-specific RNA (Fig. 3, The Provirus in Mgpt-1 Mice Is Methylated and Can Be lanes c, e, f, and g) that was indistinguishable from MSVgpt RNA present in gpt-7-3T3 cells (lane h). Furthermore, gpt gpt/gpt ii't/gpt Wt/itt p activity was detected at substantially higher levels in extracts from mycophenolic acid-selected clones than in extracts from the parental cell line (Table 4). Infection of mycophenolic acid-selected clones, but not of the unselected parental cells, with Mo-MuLV helper virus induced production ofinfectious ia -v MSVgpt virus (Table 4). NIH 3T3 cells infected with this virus produced titers of MSVgpt virus that were comparable

Table 3. 5-Azacytidine induces resistance to mycophenolic acid s --ifillm ililiv if.W.. s (MPA) in Mgpt-1 cells MPA-resistant colonies 5-Azacytidine, /iM Assay 1 Assay 2

alsj- -A V 0 0 0 1 0 0 3 1 0 10 18 15 25 36 5 50* 0 0 FIG. 2. Distinction of mice homozygous (gpt/gpt), heterozygous Mgpt-1 cells were derived by SV40 transformation ofprimary lung (gpt/wt), or negative (wt/wt) for integration ofthe MSVgpt provirus. fibroblasts from Mgpt-1 mice as described (31). Cells (2 x 105) were Nylon membranes carrying Sst I-digested DNAs from offspring of treated with various concentrations of 5-azacytidine in Dulbecco's heterozygous parents and from plasmid pMSVgpt (lane p), which modified Eagle's medium/10%o fetal bovine serum for 24 hr. After contains the MSVgpt genome (24), were hybridized with probes for incubation without 5-azacytidine for an additional 24 hr, the cells gpt and al(I) collagen sequences (see Materials and Methods). were incubated in medium containing MPA at 25 ,ug/ml to select for Relative intensities of bands representing the MSVgpt genomes the Gpt+ phenotype (26). The number of MPA-resistant cell clones (fragment S, Fig. 1 Lower) and al(I) collagen sequences clearly was determined 14 days later. Results of two independent assays are distinguished the individual genotypes. Bands V in lane p were given. derived by hybridization to plasmid sequences. *This concentration was highly toxic. Downloaded by guest on September 28, 2021 6930 Cell Biology: Jdhner et al. Proc. Natl. Acad. Sci. USA 82 (1985) Table 4. Gpt activity* and rescuable MSVgpt sequencest in generated in the DNA from unselected Mgpt-1 cells (lane a) mycophenolic acid-resistant Mgpt-1 (Mgpt/MPA) cell clones or from liver, kidney, or lung of Mgpt-1 mice (lanes g-i). [14C]XMP MSVgpt titer, Mo-MuLV titer, These results indicate that 5-azacytidine-induced nonmeth- formed, colony-forming XC plaque units ylation correlates with transcriptional activation of Cells cpm units x 10-2/ml x 10-2/ml the Mgpt-J provirus. Once virus is transcribed, superinfec- tion with helper virus results in the efficient rescue of the Mgpt-1 1,850 0 1000 MSVgpt viral genome. Mgpt/MPA3 20,600 1 500 Mgpt/MPA4 26,100 0.8 3 Mgpt/MPA5 17,400 0.5 3 DISCUSSION Mgpt/MPA7 6,500 2 200 This paper demonstrates the potential of retroviral vectors gpt-7-3T3 113,800 300 5000 for introducing foreign genes into the germ line of mice. The 3T3/Mgpt3 ND 600 1000 E. coli gpt gene carried by a replication-defective viral vector Cl-1-lA ND 0 1000 was chosen as a marker gene because in vitro selection for NIH 3T3 4,600 0 0 drug resistance is highly sensitive and allows detection of a ND, not determined. small percentage of cells that express the transduced gene. *Extracts of the fibroblast cell line from Mgpt-1 mice (Mgpt-1), Preimplantation mouse embryos were cocultivated with cells individual MPA selected clonal derivatives (Mgpt/MPA, see Table producing the retrovirus vector, because this method of 3), gpt-7-3T3 cells (see Materials and Methods), and an NIH 3T3 infection has been shown to be an efficient procedure to cell line were analyzed for Gpt activity by the formation of generate mice carrying Mo-MuLV proviral genomes in their xanthosine 5'-phosphate (XMP) from a-D-ribosyl diphosphate germ line (7, 8). The number ofvector and helper-virus copies 5-phosphate and xanthine (see Materials and Methods). in the line of the mosaic tAfter infection of Mgpt-1 cells and its derivatives with Mo-MuVL present in somatic tissues and germ from Cl-1-lA (31) cells, the titers of MSVgpt and Mo-MuLV were animal derived from infected embryos approximately reflects determined by infection of 3T3 cells followed by growth in selective the ratio ofvector and helper virus produced by the gpt-7-3T3 medium and assay on XC cells, respectively (see Materials and cells that were used for infection. Because helper virus Methods). 3T3/Mgpt3 cells were derived by infection of 3T3 cells cannot replicate in cells of the preimplantation embryo (10), with virus from Mgpt/MPA3 cells, followed by growth in selective this indicates that vector and helper-virus sequences are medium. inserted into the genome of early embryonic cells with equal efficiency. MSVgpt and helper sequences segregated inde- the pendently in the next generation and a mouse strain, Mgpt-1, to that of gpt-7-3T3 cells (Fig. 3, lane d), indicating that was derived carrying only a single MSVgpt provirus in its rescued virus had similar biological properties to the parental germ line. Thus, cocultivation of mouse embryos with cell virus used for infection of the embryos. lines producing high titers of retroviral vectors will allow the The methylation pattern of MSVgpt sequences before and efficient transfer of foreign genes into the mammalian germ after 5-azacytidine treatment was analyzed by double diges- line. Because replication of virus is not needed for germ-line tion with the methylation-insensitive restriction enzyme integration, the use of cell- lines that carry packaging- BstEII, which has no recognition sequence in the provirus, defective helper viruses for such experiments (24) would and the methylation-sensitive enzyme Ava I (see Fig. 1 avoid the infection of embryos with replication-competent Lower, map of MSVgpt). Fig. 4 shows that the viral Ava I retroviruses which can cause thymic leukemia upon activa- sites were unmethylated in the DNAs from 5-azacytidine- tion. treated and mycophenolic acid-selected cell clones (lanes Animals homozygous at the Mgpt-l locus were derived by b-e) as well as from gpt-7-3T3 cells (lane f), since their BstEII screening offspring of heterozygous parents. Homozygous fragments B were completely digested by Ava I, generating mice showed no phenotypic alterations in viability, fertility, the internal viral fragment A. No such fragments were or appearance, indicating that proviral insertion at the Mgpt-1

a b c d e f g h i 1 2 1 2 1 2 1 21 2 1 2 1 21 2 1 p _" -._ -B 4m .. m

I U

de __ -_ IEFA

FIG. 4. Nonmethylation of MSVgpt sequences in mycophenolic acid-resistant derivatives of Mgpt-1 fibroblasts. DNAs were prepared and analyzed with probe pSV2. B indicates the position of the BstEII fragment (lanes 1) carrying the MSVgpt provirus within flanking cellular sequences (see Fig. 1) in DNAs from Mgpt-1 cells (lanes a); Mgpt-1 mycophenolic acid-resistant clones 3, 4, 5, and 7 (lanes b-e, respectively); gpt-7-3T3 cells (lane f); and liver, kidney, and lung of Mgpt-1 mice (lanes g-i, respectively). The position of the internal viral Ava I fragment A (see Fig. 1), derived by secondary digestion with this enzyme in lanes 2, is shown in lane p, which contains a partial Ava I digest of plasmid pMSVgpt (30). Downloaded by guest on September 28, 2021 Cell Biology: Ahner et al. Proc. Natl. Acad. Sci. USA 82 (1985) 6931 locus did not result in insertional mutagenesis. So far, Warren, R. & Palmiter, R. D. (1981) Cell 27, 223-231. mutations by retroviral infection have been induced only 4. Stewart, C., Harbers, K., Jahner, D. & Jaenisch, R. (1983) after viral integration into primordial germ cells at the Science 221, 760-762. whereas none 14 5. Stewart, T. A., Pattengale, P. K. & Leder, P. (1984) Cell 38, postimplantation stage, of proviral genomes 627-637. in Mov substrains recovered from embryos exposed at the 6. Jaenisch, R. (1976) Proc. Natl. Acad. Sci. USA 73, 1260-1264. preimplantation stage had induced a mutation (27). It there- 7. Jahner, D. & Jaenisch, R. (1980) Nature (London) 287, fore seems that retroviral infection ofpreimplantation mouse 456-458. embryos is less likely to cause insertional mutagenesis than 8. Jaenisch, R., Jahner, D., Nobis, P., Simon, I., LUhler, J., microinjection of recombinant DNA into the zygote Harbers, K. & Grotkopp, D. (1981) Cell 24, 519-529. pronucleus, a procedure that is highly mutagenic (31, 32). 9. Jahner, D., Stuhlmann, H., Stewart, C. L., Harbers, K., Restriction enzyme analysis revealed a high level of LUhler, I., Simon, J. & Jaenisch, R. (1982) Nature (London) methylation ofthe genetically transmitted MSVgpt provirus, 298, 623-628. and viral expression was not detected by RNA blot-hybrid- 10. JAhner, D. & Jaenisch, R. (1984) in DNA Methylation, eds. Razin, A., Cedar, H. & Riggs, A. (Springer, New York), pp. ization analysis or by a sensitive colony assay. Both a high 189-219. level of methylation and a low or undetectable level of 11. Jaenisch, R. & Jahner, D. (1984) Biochim. Biophys. Acta 782, expression are characteristic of Mo-MuLV proviral copies 1-9. carried in Mov substrains (23). MSVgpt virus was, however, 12. Shimotohno, K. & Temin, H. M. (1981) Cell 26, 67-77. activated in Mgpt-1 cells upon treatment with 5-azacytidine, 13. Cepko, C., Roberts, B. & Mulligan, R. (1984) Cell 37, resulting in subclones that were drug-resistant. The provirus 1059-1062. in these clones was almost completely unmethylated at Ava 14. Miller, A. D., Ong, E. S., Rosenfeld, M. G., Verma, I. M. & I sites. The fraction ofcells selectable for drug resistance was Evans, R. M. (1984) Science 225, 993-998. small, however, and the level of gpt expression was 80-90%' 15. Joyner, A., Keller, G., Phillips, R. A. & Bernstein, A. (1983) lower than in 3T3 cells Nature (London) 305, 556-558. infected with the induced virus. This 16. Williams, D. A., Lemischka, I. R., Nathan, D. G. & Mulligan, suggests that the chromosomal position ofthe provirus at the R. C. (1984) Nature (London) 310, 476-480. Mgpt-1 locus may interfere with efficient viral expression (8) 17. Stuhlmann, H., Cone, R., Mulligan, R. C. & Jaenisch, R. or that the cell clones are selected for a low level ofproviral (1984) Proc. Natl. Acad. Sci. USA 81, 7151-7155. expression due to a toxic effect of the gpt protein on the cell 18. Rubenstein, J. L. R., Nicolas, J. F. & Jacob, F. (1984) Proc. (33). Natl. Acad. Sci. USA 81, 7137-7140. The use of5-azacytidine to activate suppressed genes is not 19. Cone, R. D. & Mulligan, R. C. (1984) Proc. NatI. Acad. Sci. restricted to cells growing in tissue culture. Silent proviral USA 81, 6349-6353. genes carried in the line of Mov substrains can 20. Sorge, J., Cuthing, A. E., Erdman, V. D. & Gautsch, J. W. germ also be (1984) Proc. NatI. Acad. Sci. USA 81, 6627-6631. efficiently activated in different tissues by drug injection into 21. Reik, W., Weiher, H. & Jaenisch, R. (1985) Proc. NatI. Acad. postnatal mice (34). This will allow manipulation of the Sci. USA 82, 1141-1145. expression ofgenes that are suppressed after transfer into the 22. Wagner, E., Vanek, M. & Vennstrom, B. (1985) EMBO J. 4, germ line, and it will permit exploration of the effect of 663-666. activation on the organismic level. 23. Stuhlmann, H., Jdhner, D. & Jaenisch, R. (1981) Cell 26, The results described here extend previous observations 221-232. on efficient inactivation ofMo-MuLV in transgeneic mice by 24. Mann, R., Mulligan, R. C. & Baltimore, D. (1983) Cell 33, DNA methylation (9, 10, 11, 35-37). In addition, the provirus 153-159. interacts with the host genome and de novo 25. Nobis, P. & Jaenisch, R. (1980) Proc. NatI. Acad. Sci. USA induces methyla- 77, 3677-3681. tion of flanking host sequences (38) and a change of an open 26. Mulligan, R. C. & Berg, P. (1981) Proc. NatI. Acad. Sci. USA chromatin conformation to a more compact one (39). Both 78, 2072-2076. these effects have been shown to be correlated with the 27. Jaenisch, R., Harbers, K., Schnieke, A., Lohler, J., virus-induced block ofal(I) collagen transcription in Mov-13 Chumakov, I., Jahner, D., Grotkopp, D. & Hoffmann, E. mice. It is likely that the viral long terminal repeat (LTR) is (1983) Cell 32, 209-216. the important determinant in position-dependent suppression 28. Schnieke, A., Harbers, K. & Jaenisch, R. (1983) Nature of gene activity. Retroviral vectors with selectable genes (London) 304, 315-319. under control ofan internal nonretroviral promoter have been 29. Nuesch, J. & Schumperli, D. (1984) Gene 32, 243-249. shown to be more in carci- 30, Jones, P. A. & Taylor, S. M. (1980) Cell 20, 85-93. expressed efficiently embryonal 31. Palmiter, R., Wilkie, T., Chen, H. & Brinster, R. (1984) Cell noma cells (18, 22) than are vectors with LTR-driven genes 36, 869-877. (35). It therefore will be of great interest to transfer such 32. Wagner, E., Covarrubias, L., Stewart, T. & Mintz, B. (1983) vectors into the germ line and compare their expression in Cell 35, 647-655. transgeneic animals to that of the vectors used in this study. 33. Chapman, A. B., Costello, M. A., Lee, R. & Ringold, G. (1983) Mol. Cell. Biol. 3, 1421-1429. We thank Doris Grotkopp for excellent technical assistance. This 34. Jaenisch, R., Schnieke, A. & Harbers, K. (1985) Proc. NatI. work was supported by grants from the Stiftung Volkswagenwerk Acad. Sci. USA 82, 1451-1455. and the Deutsche Forschungsgemeinschaft and by Grants HD19015 35. Stewart, C. L., Stuhlmann, H., Jahner, D. & Jaenisch, R. from the National Institutes of Health and PO1-CA38497 from the (1982) Proc. Natl. Acad. Sci. USA 79, 4098-4102. National Cancer Institute. 36. Gautsch, J. W. & Wilson, M. C. (1983) Nature (London) 301, 32-37. 1. Costantini, F. & Lacy, E. (1981) Nature (London) 294, 92-94. 37. Niwa, O., Yohota, Y., Ishide, H. & Sugahara, T. (1983) Cell 2. Wagner, E. F., Stewart, T. A. & Mintz, B. (1981) Proc. Nati. 32, 1105-1113. Acad. Sci. USA 78, 5016-5020. 38. Jdhner, D. & Jaenisch, R. (1985) Nature (London), in press. 3. Brinster, R. L., Chen, H. Y., Trumbauer, M., Senear, A. W., 39. Breindl, M., Harbers, K. & Jaenisch, R. (1984) Cell 38, 9-16. Downloaded by guest on September 28, 2021