Proc. Natl. Acad. Sci. USA Vol. 86, pp. 1287-1291, February 1989 Genetics Highly attenuated vaccinia mutants for the generation of safe recombinant (human immunodeficiency virus type 1 gag and env genes) DOLORES RODRIGUEZ*, JUAN-RAMON RODRIGUEZ*, JOSE F. RODRIGUEZ*, DAVID TRAUBERt, AND MARIANO ESTEBAN*t Departments of *Biochemistry, , and Immunology, and tMedicine, State University of New York Health Science Center at Brooklyn, Brooklyn, NY 11203 Communicated by Chandler McCuskey Brooks, October 20, 1988

ABSTRACT An attenuated vaccinia virus mutant with gene and HIV gag and envelope genes, we have demon- specific genetic lesions has been used to develop a vehicle for strated the advantage ofbasing recombinant on this safer live recombinant virus vaccines. The mutant virus 48-7 attenuated variant. has an 8-MDa deletion starting 2.2 MDa from the left end ofthe viral genome and point mutations in the gene encoding the 14-kDa fusion protein that determines the plaque-size pheno- MATERIALS AND METHODS type of the virus. Using the highly sensitive reporter gene Construction of Vaccinia Recombinants Expressing Lu- luciferase, we have shown that this mutant can generate ciferase and HIV Envelope and gag Genes. A 1892-base-pair recombinant viruses that infect cultured cells and animals with (bp) BamHI fragment encoding the luciferase gene (16) was normal vaccinia virus tropism. Insertion of the envelope and inserted into the vaccinia virus expression vector pSC11 and gag genes of human immunodeficiency virus type 1 into the recombinant viruses were selected by marker rescue by attenuated vaccinia mutant resulted in their efficient expres- following standard protocols (17). Luciferase activity was sion and precursor processing in infected cultured cells. Infec- measured with a spectrophotometer and results were quan- tion of mice with human immunodeficiency virus-vaccinia titated as described (18). For the construction of HIV- recombinant viruses elicited human immunodeficiency virus- vaccinia recombinants, an Xba I-Xho I DNA fragment specific antibodies. Using mice pretreated with cyclophospha- containing the coding region of HIV env was created by mide as a model for immunosuppression, the reduced virulence standard recombinant DNA techniques from a full-length of the mutant recombinant virus was clearly evident. These HIV type 1 (HIV-1) DNA genome (19). This fragment findings demonstrate that the highly attenuated vaccinia virus extends to the Ssp I site 66 bp upstream from the start site of mutant 48-7 can be used to generate effective and safer translation of the env gene and 100 bp downstream from the vaccines. termination codon. The env DNA fragment was blunt-ended by treatment with the Klenow fragment of DNA polymerase Genetically engineered recombinants of vaccinia virus are I and cloned into the Sma I site of the vaccinia virus insertion now used as eukaryotic expression vectors and, because they vector pSC11. As a result ofthis cloning strategy, we isolated are potent immunogens, have a potential as vaccines against a plasmid containing HIV env coding sequence under the a broad spectrum of infectious pathogens of animals and control of the 7.5-kDa vaccinia virus promoter. A similar human significance (1-3). However, in humans, accidental strategy was used to construct vaccinia virus insertion vaccination with live vaccinia virus of an immunocompro- vectors containing the intact HIV gag gene. A BssHII-EcoRI mised host can DNA fragment containing the coding region of gag was lead to serious and lethal complications, such isolated from the full-length HIV-1 DNA. This fragment as systemic vaccinia infection with encephalitis (4, 5). This contains 79 bp upstream from the start site of translation of point was confirmed by the vaccination of a soldier who the gag gene and, from the termination site, it contains 2358 unknowingly was seropositive for human immunodeficiency bp of the pol gene. The luciferase, env, and gag genes were virus (HIV) (6). To diminish the chances of producing such inserted in the correct orientation relative to the vaccinia complications, an expert panel from the World Health Or- 7.5-kDa promoter and into the thymidine kinase region of ganization has recommended the development of attenuated wild-type and mutant viruses by DNA homologous recom- strains of vaccinia virus that carry identifiable genetic mark- bination. P -Galactosidase-producing plaques (17) were ers upon which to base the construction of future recombi- picked, cloned three times, and amplified in African green nant vaccines (7). Attenuated variants of vaccinia virus have monkey kidney cells (BSC-40) grown in culture. been obtained by several methods, including (i) inactivation of the viral thymidine kinase gene (8), (ii) generation of deletions at the left end of the viral genome (9, 10), (iii) RESULTS mutation of the gene encoding a 14-kDa fusion protein that Expression and Tissue Tropism of Attenuated Vaccinia influences viral plaque size (11), (iv) alterations in size of Recombinants. Since attenuated viruses might multiply less virion core proteins (12, 13), and (v) insertion of the inter- efficiently in infected animals than wild-type virus, our first leukin 2 gene that greatly decreases virus virulence in interest was to document the extent of viral gene expression athymic nude mice (14, 15). and tissue tropism of attenuated recombinants. This was In this report we describe the effectiveness of recombinant accomplished by introducing the highly sensitive reporter vaccines based on an attenuated vaccinia virus mutant that gene luciferase into vaccinia DNA. Two classes of recombi- has specific genetic lesions. Using luciferase as a reporter nants were obtained, one derived from wild-type virus and

The publication costs of this article were defrayed in part by page charge Abbreviations: HIV, human immunodeficiency virus; pfu, plaque- payment. This article must therefore be hereby marked "advertisement" forming unit(s). in accordance with 18 U.S.C. §1734 solely to indicate this fact. tTo whom reprint requests should be addressed. 1287 Downloaded by guest on September 28, 2021 1288 Genetics: Rodriguez et al. Proc. Natl. Acad. Sci. USA 86 (1989) A

3 10-

W.~2 -6 WRLUC ARA-C -- MUT7LUC ARA-C

W o L.I'x

10-

1 3 5 TIME (hr) 24 B

CD _ on

_ ._ -Wr LL E )'a a.jQ1 C

Un

0~~~~~~~~~

4) 00)Do% 1&'mr-~~~~~~~:oC la l.@ ~~~C la WR LUC MUT7LUC 72 hours after vaccination

FIG. 1. Expression of the firefly luciferase gene by wild-type and mutant recombinants of vaccinia virus. The construction of vaccinia luciferase recombinants and assays of luciferase activity have been described (18). (A) Expression of luciferase in cultured cells. BSC-40 cells were infected with wild-type vaccinia luciferase-recombinant viruses (WRLUC, open symbols) or with mutant 48-7 luciferase-recombinant viruses (MUT7LUC, solid symbols) at 5 plaque-forming units (pfu) per cell with (A, A) or without (o, *) cytosine arabinonucleoside (ARA-C) at 40,u g/ml. At various times after infection, cells were collected and cell extracts prepared in luciferase buffer (100 mM potassium phosphate/1 mM dithiothreitol) were tested for luciferase activity in the presence ofluciferin and ATP (18). (B) Expression of luciferase in organs of infected mice. BALB/c female mice, 6 weeks old (Charles River Breeding Laboratories), were inoculated i.p. with 5 x 105 pfu of recombinant vaccinia viruses and 3 days later tissues were washed in isotonic phosphate-buffered saline (PBS), weighed, homogenized in luciferase buffer containing 1 mM phenylmethylsulfonyl fluoride and leupeptin at 10ttg/ml (5t 1/mg of tissue), and supernatants were assayed for luciferase activity (18). WRLUC, wild-type vaccinia luciferase-recombinant; MUT7LUC, mutant 48-7 luciferase-recombinant. Downloaded by guest on September 28, 2021 Genetics: Rodriguez et al. Proc. Natl. Acad. Sci. USA 86 (1989) 1289 A 1 2 3 4 5 6 7 8 the other derived from the attenuated vaccinia mutant 48-7. Mutant 48-7 has an 8-MDa deletion starting 2.2 MDa from the left end of the viral genome and two point mutations on the gene encoding the 14-kDa fusion protein that result in high attenuation (refs. 10-12; S. C. Gong, C. F. Lai, and M. E., A*_- unpublished results). Vaccinia recombinants containing the ENV luciferase gene were then tested for luciferase activity both in cultured cells and in experimental animals. As shown in Fig. 1A, the levels of luciferase activity were equivalent in A. monkey cells infected with the luciferase-wild-type recom- binant or with the luciferase-mutant 48-7 recombinant. In infected mice luciferase activity was observed in spleen and liver, the known target organs for wild-type vaccinia virus B I 2 3 4 5 6 7 8 multiplication (20). The levels of luciferase activity found in spleens of mice inoculated with the luciferase-mutant recom- than in mice inoculated with 8; - p56 binant virus were =30% lower ii. mutant recombi- ..p32 the wild-type recombinant virus. Thus, the ~ ~ ~ ~~~~~ nant virus efficiently expresses a foreign gene and shows the GAG same tissue tropism as the wild-type virus. Expression of HIV Envelope and gag Genes by Attenuated . ,n / OFI.:':fLF _ Vaccinia Recombinants. To assess that the attenuated vac- 0* cinia virus mutant has a potential, we constructed recombinant viruses containing either the HIV-1 env or the gag gene and studied their expression and immunogenicity. Because mature retroviral gag and envelope proteins are derived from precursor molecules by processing (21-23), we first examined processing of env and gag by immunoprecip- C 1 2 3 4 5 6 7 itation analysis. The results are shown in Fig. 2 A and B. In

-N M ____ .- gp 160 B

:..- 4- I&-P ~ 1 g --9P120 678 9 domok -205

-116 A 1 2 3 4 5

205 - __. gpl60 -66 116 > __- p56 - _ 66 ft- 45- .45 --

29 - FIG. 2. Expression of HIV env and gag by vaccinia (wild-type 40 .29 and mutant) recombinant viruses. Immunoprecipitation of env (A) B. .-... A and gag (B) proteins from lysates of HIV-vaccinia virus-infected cells, as examined by NaDodSO4/PAGE. Hela S3 cells infected (5 pfu per cell) with vaccinia recombinants were labeled with [35S]_ FIG. 3. Humoral immune response elicited in mice by HIV- methionine (50,u Ci/ml; 1 Ci = 37 GBq) from 3 to 18 hr after infection vaccinia recombinants. (A) Four BALB/c mice per group were and cell extracts were immunoprecipitated with mouse monoclonal inoculated i.p. (5 x 107 pfu) with purified HIV-vaccinia env mutant antibodies against gpl20 (env) and p24 (gag) (DuPont). Lanes: 1, 3, recombinant virus, and 21 days later mice were revaccinated i.v. and 5, and 7, total cell extracts; 2, 4, 6, and 8, immunoprecipitates. i.p. with 108 pfu of recombinant virus. After 21 days, mice were given Uninfected cells (lanes 1 and 2), cells infected with wild-type virus a booster injection i.v. with 100tug of purified gpl60 isolated from (lanes 3 and 4), and cells infected with HIV-vaccinia recombinants, HIV-vaccinia-infected cells. Serum was collected from the tail vein from wild-type virus (lanes 5 and 6) and from mutant virus (lanes 7 6 days later and tested for specific antibodies on a Western blot by and 8), are shown. (C) Processing of env products in cells infected immunoperoxidase staining. Proteins were separated by a minigel with HIV-vaccinia env mutant virus (lanes 1-5). Six hours after (10% polyacrylamide). Lanes: 1, lysates of mutant virus-infected infection, Hela cells were pulse-labeled with [35S]methionine (l00 CV-1 cells 24 hr after infection; 2, lysates of HIV-vaccinia env .t Ci/ml) for 30 min and chased for various times with a 100-fold mutant-infected CV1 cells 24 hr after infection; 3, purified HIV- excess of unlabeled methionine, and cell extracts were immunopre- vaccinia env wild-type virions; 4, purified HIV-vaccinia env mutant cipitated with a monoclonal antibody to gpl20. Proteins were virions; 5, purified gpl60. (B) Mice were primed with purified HIV- resolved on a 10% NaDodSO4/polyacrylamide gel. Total cell lysate vaccinia gag mutant recombinant virus and 21 days later revacci- (lane 1) and immunoprecipitates after a 30-min pulse (lane 2) and after nated as above. Serum was collected 2 weeks after secondary chase periods of 30 min (lane 3), 60 min (lane 4), and 90 min (lane 5) vaccination and tested on a Western blot. Proteins were separated on are shown as well as purified HIV-1 (lane 6) and gpl20 released from a long gel (10% polyacrylamide). Lanes: 6, purified HIV-1 virus (a infected cells (lane 7). Released gpl20 was measured in culture reverse transcriptase mutant with unprocessed gag, isolated from the supernatants from mutant HIV-vaccinia env virus-infected Hela 8E5 human T-cell line); 7, purified gpl60; 8, lysates of HIV-vaccinia cells at 24 hr after infection after a 50-fold concentration by using an env wild-type virus-infected Hela cells 36 hr after infection; 9, lysates Amicon ultrafiltration cell. Lanes 6 and 7 show a blot reacted with of wild-type virus-infected Hela cells 36 hr after infection. Protein rabbit anti-HIV gpl60 serum and developed by immune peroxidase markers are myosin,, -galactosidase, bovine serum albumin, oval- staining. bumin, and carbonic anhydrase. Downloaded by guest on September 28, 2021 1290 Genetics: Rodriguez et al. Proc. Natl. Acad. Sci. USA 86 (1989) lysates of Hela S3 cells infected with wild-type HIV-vaccinia and gag proteins in serum samples was also confirmed by env virus, monoclonal antibodies against gp120 immunopre- ELISA with HIV-infected cell extracts (Pasteur Institute). cipitates both gp120 and its precursor gpl60 (Fig. 2A, lane 6). Reduced Virulence of Attenuated HIV-Vaccinia Recombi- These polypeptides were also detected in lysates of cells nants in Immunosuppressed Animals. Since it is known that infected with the mutant recombinant virus (lane 8). How- inoculation of immunocompromised individuals with vac- ever, these polypeptides were not detected in lysates of cinia virus may lead to generalized vaccinia and encephalitis uninfected cells (lane 2) or in cells infected with nonrecom- (4, 5) [and indeed, this has been documented in a vaccinated binant virus (lane 4). In lysates of cells infected with HIV- male seropositive for HIV (6)], it was important to assess the vaccinia gag recombinants (either wild-type or mutant virus), degree of virulence of attenuated HIV-vaccinia recombi- a monoclonal antibody against p24 has specific reactivity nants in an immunosuppressed host. Thus, BALB/c mice with polypeptides p56, p42, p32, and p24 (Fig. 2B, lanes 6 and were first treated with the immunosuppressor drug cyclo- 8). Next we establish a precursor-product relationship for phosphamide (24) and subsequently infected, and their sur- env by pulse-chase experiments (Fig. 2C). The polypeptide vival was evaluated daily (Fig. 4). As expected, in normal that was most labeled during the pulse period was gpl60, mice the mortality produced by the vaccinia mutant was which was chased into gpl20. The chased gpl20 was released about 100 times lower than that induced by wild-type virus extracellularly and was recovered in the cell culture super- (Fig. 4A). Significantly, the differences in mortality between natant (lane 7). Two other immunoreactive bands of 70 and the two viruses were greater with immunosuppressed mice 50 kDa (lane 7) were also found and appear to result from a (Fig. 4B). This was also the case with mutant recombinant single nick in gp120 (unpublished results). To determine if virus. As shown in Fig. 4 C and D, mutant HIV-vaccina env mutant HIV-vaccinia recombinants elicit specific antibodies virus caused markedly decreased mortality in immunosup- against env and gag proteins, mice were vaccinated with the pressed mice as compared with wild-type recombinant virus. recombinants and serum was tested by Western blot and ELISA. We found that mice vaccinated with mutant HIV- vaccinia env virus have weak antibody response to gpl60. DISCUSSION However, induction of antibodies to gp160 was markedly These studies demonstrate that a vaccinia virus mutant, 48-7, increased if vaccinated mice were given a booster injection containing two defined genetic lesions can be used to gener- with purified gpl60 (Fig. 3A, lane 5). Mice only vaccinated ate safe recombinant live vaccines. During infection mutant with mutant HIV-vaccinia gag virus had a strong immune recombinant viruses express significant levels of foreign response to gag (Fig. 3B, lane 6). Immune reactivity to env genes, are good immunogens, and have greatly reduced

A 100 j B

75

50-

25-

Viral inoculation, pfu C 10D 100

80-\<108 pfu 80 107 pfu

601 11 460;

40- ~~~~~~~~40-

20- ! 20 -4

5 10 15 5 10 15 Time after vaccination, days FIG. 4. Reduced virulence of HIV-vaccinia recombinants in immunosuppressed mice. Four BALB/c mice per group were inoculated i.p. (0.2 ml) with cyclophosphamide (Sigma) at 300 mg/kg in PBS, and 5 days later mice were treated with a second dose ofthe drug but at 150 mg/kg. Twenty-four hours later, untreated and drug-treated mice were inoculated i.p. with various doses of purified parental and recombinant HIV- vaccinia env viruses. Survival was followed daily. (A and B) Results obtained after 1 month with parental viruses in normal (A) and immunosuppressed (B) animals. o, MUT-7; *, WR. (C and D) Results obtained in imniunosuppressed animals with HIV-vaccinia env recom- binants. o, MUT-7 ENV;m, WR ENV. The surviving mouse (WR ENV) in D died on day 21. The survival pattern shown with mutant recombinant virus was maintained for more than 2 months. Downloaded by guest on September 28, 2021 Genetics: Rodriguez et al. Proc. Natl. Acad. Sci. USA 86 (1989) 1291 virulence in an immunosuppressed host. Expression of the 1. Panicali, D. & Paoletti, E. (1982) Proc. Natl. Acad. Sci. USA highly sensitive reporter gene luciferase in spleen and liver 79, 4927-4931. but not in other organs provided direct evidence that tissue 2. Mackett, M., Smith, G. L. & Moss, B. (1982) Proc. Natl. Acad. Sci. USA 79, 7415-7419. tropism of attenuated recombinant virus is the same as 3. Moss, B. & Flexner, C. (1987) Annu. Rev. Immunol. 5, 305- wild-type virus. Cells infected with attenuated HIV-vaccinia 324. viruses were efficient in their ability to process env and gag 4. Lane, J. M., Ruben, F. L., Neff, J. M. & Millar, J. D. (1969) precursors. Similar processing for HIV-1 env products has N. Engl. J. Med. 281, 1201-1208. been observed with other HIV-vaccinia recombinants de- 5. Arita, I. & Fenner, F. (1985) in Vaccinia Viruses as Vectorsfor rived from wild-type virus (25, 26). However, the processing Vaccine Antigens, ed. Quinnan, J. (Elsevier, New York), pp. 49-60. of HIV-1 gag precursors that we have observed with HIV- 6. Redfield, R. R., Wright, D. C., James, W. D., Jones, T. S., vaccinia recombinant has not been observed, and only the Brown, C. & Burke, D. S. (1987) N. Engl. J. Med. 316, 673- unprocessed form of gag p56 has been reported with a 676. wild-type HIV-vaccinia gag recombinant (27). Processing of 7. Brown, F., Schild, G. C. & Ada, G. L. (1986) Nature (London) gag precursors by our vaccinia recombinants is likely to be 319, 549-550. due to the function of the protease encoded by the 5' end of 8. Buller, R. M., Smith, G. L., Cremer, K., Notkins, A. L. & Moss, B. (1985) Nature (London) 317, 813-815. the pol gene contained in the BssHII-EcoRI fragment used to 9. Paez, E., Dallo, S. & Esteban, M. (1985) Proc. Natl. Acad. Sci. introduce the gag sequence as a result of ribosomal "frame- USA 82, 3365-3369. shifting" (28). Of significance in our study is that mice 10. Dallo, S. & Esteban, M. (1987) 159, 408-422. vaccinated with attenuated HIV-vaccinia recombinants de- 11. Dallo, S., Rodriguez, J. F. & Esteban, M. (1987) Virology 159, velop antibodies to env and gag proteins. The immune 423-432. response to the gag protein was stronger than to the env 12. Paez, E., Dallo, S. & Esteban, M. (1987) J. Virol. 61, 2642- 2647. protein. However, if after vaccination mice were given a 13. Maa, J.-S. & Esteban, M. (1987) J. Virol. 61, 3910-3919. booster injection i.v. with purified gpl60, we then observed 14. Ramshaw, I. A., Andrew, M. E., Phillips, S. M., Boyle, D. B. good antibody response to the env protein (Fig. 3A). & Coupar, B. E. H. (1987) Nature (London) 329, 545-546. In this study we have established an immunosuppressed 15. Flexner, C., Hugin, A. & Moss, B. (1987) Nature (London) 330, mouse model system to evaluate virulence of vaccinia virus 259-262. recombinants. In this animal-virus system, we observed 16. Ow, D. W., Wood, K. V., DeLuca, M., de Wet, J. R., Helin- ski, D. R. & Howell, S. H. (1986) Science 234, 856-859. clear differences in mortality between recombinant viruses 17. Chakrabarti, S., Brechling, K. & Moss, B. (1985) Mol. Cell. that correlated with the viral genetic markers (Fig. 4). In fact, Biol. 5, 3403-3409. inactivation ofwild-type vaccinia thymidine kinase locus that 18. Rodriguez, J. F., Rodriguez, D., Rodriguez, J.-R., McGowan, is known to reduce virulence in normal mice (8), has little E. & Esteban, M. (1988) Proc. Natl. Acad. Sci. USA 85, 1667- effect in decreasing virulence in immunosuppressed animals. 1671. However, a vaccinia recombinant with a left-end deletion and 19. Fisher, A. G., Collalti, E., Ratner, L., Gallo, R. C. & Wong- Staal, F. (1985) Nature (London) 316, 262-265. alterations in the 14-kDa fusion protein encoding gene has 20. Dales, S. & Pogo, B. G. T. (1981) Biology of Poxviruses, about 100 times less virulence in immunosuppressed animals Virology Monographs (Springer, New York), Vol. 18. than a wild-type recombinant with a thymidine kinase- 21. Muesing, M. A., Smith, D. H., Cabradilla, C. D., Benton, negative phenotype. Thus, a potential advantage in generat- C. V., Lasky, L. A. & Capon, D. J. (1985) Nature (London) ing highly attenuated vaccinia virus recombinants with de- 313, 450-458. letions and point mutations is that these types of recombi- 22. Sanchez-Pescador, R., Power, M. D., Barr, P. J., Steiner, K. S., Stempien, M. M., Brown-Shimer, S. L., Gee, W. W., nants will greatly decrease the risk of virus dissemination Renara, A., Randolph, A., Levy, J. A., Dima, D. & Luciw, after vaccination, particularly in an immunocompromised P. A. (1985) Science 227, 484-492. host. 23. Wain-Hobson, S., Sonigo, P., Damos, O., Cole, S. & Alizon, Considering the characteristics of the highly attenuated M. (1985) Cell 40, 9-17. vaccinia recombinants that we have generated, these recom- 24. Bach, J. F. (1975) Frontiers ofBiology (Elsevier Biomed., New binants and derivatives might be useful as immunogens in York), Vol. 41, pp. 173-225. studies of recognition specificity of immune T cells and in the 25. Chakrabarti, S., Rober-Guroff, M., Wong-Staal, F., Gallo, R. C. & Moss, B. (1986) Nature (London) 320, 535-537. design of safer vaccines against pathogens of veterinary and 26. Hu, S.-L., Kosowski, S. G. & Dalrymple, J. M. (1986) Nature human importance. (London) 320, 537-540. 27. Walker, B. D., Chakrabarti, S., Moss, B., Paradis, T. J., We thank J. A. Lewis and R. Bablanian for reading the manu- Flynn, T., Durmo, A. G., Blumberg, R. S., Kaplan, J. C., script, E. McGowan for luciferase assays, B. Moss for the vector Hirsch, M. S. & Schooley, R. T. (1987) Nature (London) 328, pSC 11, S. H. Howell for the luciferase plasmid pD0432, F. Ruscetti 345-348. for the plasmid containing the full-length HIV-1 DNA strain (HTLV- 28. Jacks, T., Power, M. D., Masiarz, F. R., Luciw, P. A., Barr, IIIB), and J. J. Skehel for the HIV-1 reagents. P. J. & Varmus, H. E. (1987) Nature (London) 331, 280-283. Downloaded by guest on September 28, 2021