and human immunodeficiency virus type I replication in

Proc. Nati. Acad. Sci. USA Vol. 91, pp. 9715-9719, October 1994

Medical Sciences

Ribozyme-mediated suppression of Moloney murine leukemia virus and human immunodeficiency virus type I replication in permissive cell lines (retrovirus/packaging sequence/gene therapy)

LUN-QUAN SUN, DAVID WARRILOW*, Li WANG, CRAIG WITHERINGTON, JANET MACPHERSON,

AND GEOFF SYMONDSt

The R. W. Johnson Pharmaceutical Research Institute-Sydney, 74 McLachlan Avenue, Rushcutters Bay, New South Wales 2011, Australia

Communicated by W. J. Peacock, May 24, 1994 (received for review November 3, 1993)

ABSTRACT Several hammerhead ribozymes targeted to different sites within the retroviral packaging () sequences of the Moloney murine leukemia virus (Mo-MLV) and the human immunodeficiency virus type 1 (HIV-1) were designed and shown to cleave target RNA in vitro at the chosen sites. The engineered ribozymes, as well as antisense sequence complementary to the Mo-MLV 4epackaging region, were cloned into the 3' untranslated region of the neomycin-resistance gene (neo). This was coupled to the simian virus 40 early promoter within the pSV2neo vector. For the ribozymes against the Mo-MLV Or site, the constructs were transfected into MoMLV-infected and virus-producing mouse NIH 3T3 cells. With the exception of one of the single ribozymes (the one least effective in cutting target RNA in vitro), all of the constructs effectively (70-80%) suppressed retrovirus production. These results demonstrate a direct correlation between in vitro deavage and in vivo ribozyme-mediated virus suppression. In addition, a ribozyme targeted to the HIV-1 packang site was engineered into the same vector and transfected into the human T-cell line SupT1. The transfectants were cloned and then challenged with HIV-1. When compared to vector-transfected control cells, a snficant reduction in HIV-1 production was observed as measured by p24 and syncytia formation assays. This study demonstrates a feasible approach to the suppression of retrovirus replication by targeting the * packaging site with hammerhead ribozymes.

MLV is useful for assessing the efficacy of antiviral ribozymes because (i) it is an RNA virus, an ideal target for ribozyme action, (ii) levels can be assayed in tissue culture systems, and (iii) it induces two assessable phenotypes in mice-viremia and leukemia. Anti-Mo-MLV ribozymes can thus be tested for their ability to reduce virus in cells in culture and viremia and subsequent leukemia in animals. Among several important elements for retroviral, including HIV-1 and Mo-MLV, replication are cis-acting viral genomic sequences necessary for the specific packaging of viral RNA into virus particles (11, 12). Recent studies have indicated that antisense RNA complementary to the Mo-MLV packaging sequence could inhibit virus replication-dependent leukemia induction in mice (13). In the present report, we have used Mo-MLV as a first step for assessing proof of principle for efficacy of anti-retroviral ribozymes and to show a correlation between in vitro cleavage of the Mo-MLV packaging sequence and in vivo suppression of Mo-MLV replication mediated by hammerhead ribozymes. In addition, we demonstrate that a ribozyme construct targeted to the HIV-1 qi packaging region can significantly inhibit HIV-1 infectivity and viral production in a target cell population. The present results show a feasible approach to the suppression of retroviral replication by targeting hammerhead ribozymes to the retroviral 4i packaging site.

Ribozymes are RNA molecules that possess enzymatic, self-cleavage activities. Several features of ribozymes make them attractive as potential antiviral agents, in particular their specificity of cleavage and ability to cleave multiple substrate molecules (1). Haseloff and Gerlach (2) demonstrated that hammerhead ribozymes can be efficiently and specifically targeted to various sites along RNA transcripts in vitro. Because of both its inherent simplicity and the fact that it can be incorporated into a variety of flanking sequence motifs without changing its site-specific cleavage capabilities, the hammerhead ribozyme has been used in studies of gene suppression both in vitro and in vivo. Several reports have indicated that the hammerhead ribozyme may function in human T-ceil lines as a potential anti-human immunodeficiency virus type 1 (HIV-1) agent (3-8). In addition, a hairpin ribozyme has also been shown to be able to inhibit HIV-1 expression in a transient assay system (9, 10). Both HIV-1 and Moloney murine leukemia virus (MoMLV) belong to the class retrovirus. HIV-1 has become the main target in the potential use of gene therapy for AIDS, and the approaches that have been used in tissue culture systems include antisense- and ribozyme-based technology. Mo-

MATERIALS AND METHODS Cells. 3T3 Mo-MLV was derived from mouse NIH 3T3 cells infected with Mo-MLV (M. Miller and G.S., unpublished data). Both 3T3 Mo-MLV and XC cells (ATCC CCL 165) were maintained in Dulbecco's modified Eagle's medium containing 10%o fetal calf serum (FCS) (CytoSystems, Sydney, Australia). SupTl cells (ATCC CRL-1942.; a human T-lymphoma cell line) were cultured in RPMI 1640 medium plus 10% FCS. Design and Construction of Ribozymes. For anti-Mo-MLV ribozymes (Fig. 1), four sites were chosen in the Mo-MLV packaging region according to the presence of a GUC sequence within a potentially accessible site of the RNA secondary structure derived from Zucker's FOLDRNA program (14). The sites were designated 243, 274, 366, and 553, based on their distance from the 5' end of the viral transcript (15). Two types of ribozyme were designed: three single ribozymes targeted individually to sites 243, 274, and 366 with arms 12 nt long and one multiple ribozyme targeted to Abbreviations: Mo-MLV, Moloney murine leukemia virus; HIV, human immunodeficiency virus; Rz, ribozyme. *Present address: Waite Agricultural Research Institute, Urrbrae, South Australia. tTo whom reprint requests should be addressed.

The publication costs of this article were defrayed in part by page charge

payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact.

9715

9716

Medical Sciences: Sun et al.

Proc. Natl. Acad Sci. USA 91 Bal I

Bal I

MoMLV RNA 212 243 274 Antisense Multiple Rz Rz243 Rz274 Rz366

366

553

747

FIG. 1. Ribozyme targeting sites within the Mo-MLV packaging region. Four GUC cleavage sites are at nt 243, 274, 366, and 553 of pMLV-1. An antisense sequence was made to the region (nt 212-747) and single (Rz243, Rz274, Rz366) and multiple ribozymes with intervening antisense sequences were also engineered.

all four sites with intervening antisense arms of the length of sequences between each of the target sites. The single ribozymes were constructed by cloning an artificial doublestranded insert with overhanging Pst I and EcoRl ends into pGEM3Zf(+). The resulting plasmids were pGEM243, pGEM274, and pGEM366. The multiple ribozyme was constructed by a variation of standard in vitro mutagenesis protocols (16). This plasmid was called pGEM-M7. For the anti-HIV ribozyme, one GUA site was chosen in the HIV-1 (HIVSF2) packaging region. The ribozyme flanking sequence is from nt 736-765 numbered from the 5' end of the HIV genome (CGGCGACUGGUGAiIJACGCCAAUUUUUGAC). As for the previous constructs, the synthetic ribozyme insert was cloned into either the pGEM transcription vector for in vitro cleavage assay or a Sma I site in the 3' untranslated region of the neo gene of pSV2neo vector by blunt-ended ligation. The latter construct was termed pSVRzoFHIV. Successful cloning and sequence integrity were confirmed by DNA sequencing. In Vitro Cleavage. In vitro cleavage reactions were conducted on in vitro generated transcripts from pGEMO, a plasmid containing the Bal I/Bal I fragment of the Mo-MLV 4i packaging sequence, and pGEM550, which encompasses the 5' untranslated region of HIV-1 including the packaging region. Runoff transcriptions were performed with a RNA transcription kit (Promega). For cleavage reactions, the ribozyme and substrate (molar ratio, 1:1) were preincubated at 80°C for 2 min, followed by 30 min of incubation at 37°C in the presence of 50 mM Tris HCl (pH 7.5) and 10 mM MgCl2. Transfection of 3T3 Mo-MLV Cells with Vectors Containing Ribozymes or Antisense Sequences. As noted, the ribozyme inserts and an antisense control were cloned into a Sma I site in the 3' untranslated region of the neo gene of the pSV2neo vector by blunt-ended ligation. The resultant vectors, termed pSV243, pSV274, pSV366, pSVM7, and pSVasi (the antisense construct), respectively, were transfected into 3T3 Mo-MLV cells by a calcium phosphate transfection protocol. Positive colonies were scored after 9-12 days in culture with G418 (500 ,g/ml). For each construct, four to seven colonies were isolated with cloning cylinders. These colonies were grown, stored in liquid N2, and then used in further assays. XC Assay. Mo-MLV from virus-producing cells was titrated as described (17) except that Polybrene (8 ug/ml) (Sigma) was present during infection. This assay was based on the ability of Mo-MLV to induce syncytia (fused cells) on plates of XC cells. Viral RNA Dot-Blot Assay. Supernatant (1 ml) from a 16-hr culture of NIH 3T3 virus-producing cells was clarified by centrifuging (9000 x g; 10 min; 4°C) in a microcentrifuge. Viral RNA was precipitated in 8% PEG 8000 and 0.5 M NaCl. After phenol/chloroform extraction, RNA was blotted onto a positively charged nylon membrane (Zeta-Probe; Bio-Rad) in an alkali transfer solution and hybridized with an RNA probe from pGEMi as described (18). i,

i

(1994)

Molecular Analyses. The integration of transfected DNA in 3T3 Mo-MLV cells was confirmed by the method of Southern. An RNA protection assay was used to detect ribozyme or antisense construct expression in the transfected 3T3 cells. To confirm expression of the neo-ribozyme chimeric RNA in human SupTi cells, Northern analysis was performed by using a neo-specific probe. Transfection of pSV-Rqi-HIV Construct into T Lymphocytes. The anti-HIV packaging site construct, pSV-R*-HIV, was electroporated into SupTi cells. Exponentially growing cells were harvested and the number of viable cells was determined by dye exclusion. The cells were washed with PBS and resuspended at a density of 1 x 107 viable cells per ml in RPMI 1640 medium without FCS but containing 10 mM dextrose and 0.1 mM dithiothreitol; 0.4 ml of the cell suspension and 10 pg of pSV-RO-HIV plasmid DNA were used per electroporation in 0.4-cm cuvettes (Bio-Rad). The cell and DNA mixture were subjected to a single pulse of 960 puF, 200 V from a Gene Pulser (Bio-Rad). Cells in the cuvette were then incubated for 10 min at room temperature and seeded at 1 x 106 cells per ml in a Petri dish (5%6 C02/95% air; 370C). At 48 hr after electroporation, the cells were selected in medium supplemented with G418 (800 ug/ml). Nine to 12 days later, positive colonies were isolated and grown as clonal isolates to be then used in a HIV-1 challenge assay. ByV-1 Infection. The clonal ribozyme construct-expressing cells, plus controls, were infected with HIV-1 SF2 or SF33 isolates at a multiplicity of infection of 0.1. Using SF33, 2 hr postinfection the cells were washed and 10 ml of fresh medium was added. Every 3-4 days, the number of single viable cells and syncytia was counted. Using SF2, a similar protocol was employed with aliquots of the supernatant taken for p24 assay by using the Coulter HIV-1 p24 ELISA kit every 3-4 days.

RESULTS In Vitro Ribozyme-Catalyzed Cleavage of Mo-MLV and HIV-l *Packaging Sequences. Using the strategy of Haseloff and Gerlach (2), we designed three single ribozymes and one multiple nbozyme against Mo-MLV as shown in Fig. 1. ABal I/Bal I fragment of the q, packaging sequence of Mo-MLV (derived from pMLV-1; ref. 15) was cloned into pGEM3Zf(+) and transcribed as a substrate for in vitro ribozyme cleavage analysis. For three of the four ribozyme constructs, incubation of a 32P-labeled qi transcript with 32P-labeled ribozyme RNA (in approximately equimolar amounts) led to efficient cleavage of the substrate under mild physical conditions (37°C; 10 mM MgCl2/50 mM Tris HCl, pH 7.5). As shown in Fig. 2, the size of the cleaved Mo-MLV 4i/fragments produced by ribozyme (Rz) 274 and Rz366 was consistent with the location of predicted sites for cleavage, resulting in bands of 62 nt plus 473 nt (Rz274; lane 6) and 154 nt plus 381 nt (Rz366; lane 8), respectively. The multiple ribozyme (Rz-M7) produced four fragments of 50, 92, 187, and 240 nt (lane 10) as predicted, as well as several partially cleaved fragments (lane 4). For Rz243, there was no visible cleavage at 37°C (Fig. 2) and relatively weak cleavage, yielding appropriate sized fragments, at 50°C (data not shown). Under the same conditions, the anti-HIV-1 q, ribozyme was also assayed for its ability to cleave the substrate. Two cleavage products (P1 and P2) of expected sizes 292 and 336 bp were observed, showing a correlation between the catalytic activity of ribozyme and time (Fig. 3). Integration and Expression of Ribozyme Constructs in Transfected 3T3 Mo-MLV Cells. To study in vivo efficacy of the ribozymes, each of the single rbozymes and the multiple ribozyme were cloned into pSV2neo. These constructs, plus antisense and vector alone controls, were transfected into a clonal 3T3 Mo-MLV cell line (M. Miller and G.S., unpub-

-.t,r10 syncytia. Mock, uninfected SupT1 cells treated in the standard infection procedure but without HIV-1.

brane of the transfected SupT1 cells were similar to the levels found for the untransfected cells (data not shown). In addition, Northern analysis of the transfected SupT1 cells demonstrated stable expression of the neo-ribozyme chimeric mRNA up to 3 months in culture (Fig. 6).

DISCUSSION To date, ribozymes have been designed targeting certain regions of the HIV-1 genome including the 5' leader sequence (6, 8-10) and gag (3) and tat genes (ref. 7; L.Q.S., L.W., and G.S., unpublished data). In the present study, a specific site within retroviral genomes has been chosen for ribozyme targeting. As for other retroviruses, the efficient packaging of HIV-1 genomic RNA into virus particles requires a specific sequence, termed the 4i packaging region (11, 12). This region is found in a highly structured conserved stem-loop. Comparison among 18 published HIV-1 strains has shown almost absolute nucleotide conservation in the base-paired regions required to maintain this structure (19). It appeared to us that the functional and structural features of the qi packaging region made it a desirable ribozyme target site for the suppression of HIV-1 replication. We first explored this potential strategy in a Mo-MLV model system. In vitro, three of the four ribozymes were able to effectively cleave the substrate RNA molecule. After transfection of these ribozyme constructs into MoMLV-infected cells, the same ribozymes that showed efficient in vitro cleavage exhibited an ability to suppress Mo-MLV replication in vivo. This was not due to differen1

pSV2neo

(1994)

2

3

4

3 SupT1 -T

XI

E

100 C

10>

|

1

6

9

12

Days post-infection

FIG. 5. Effect of ribozyme on HIV-1 replication in stably transfected SupT1 cells. Clones containing anti-HIVqi ribozyme, pSV2neo vector alone, and parental SupTl cells were challenged with HIV-1 SF2 (multiplicity of infection, 0.1) on day 0. Samples from days 6, 9, and 12 were collected and subjected to p24 antigen analysis. The p24 values (mean ± SE) from three independent clones are plotted in the graph. Ordinate is a logarithmic scale.

FIG. 6. Long-term expression of ribozyme constructs in transfected SupT1 cells. Total RNA (15 ug) from SupT1 cells (lane 1), SupT1 cells with pSV2neo vector (lane 2), and ribozyme constructexpressing SupT1 cells in culture for 1 week after cloning (lane 3) and for 3 months (lane 4) were blotted onto nylon membrane and hybridized with a neo-specific probe. Arrowheads indicate sizes of the neo transcript from SupTi cells with pSV2neo and the neoribozyme chimeric transcript from the ribozyme constructexpressing cells (upper band).

Proc. Natl. Acad. Sci. USA 91 (1994)

Medical Sciences: Sun et al. tial expression of the constructs and argues for the suppression of viral replication being due to ribozyme activity. According to Alford et al. (20), who described the RNA / packaging region of Mo-MLV, and our own data utilizing the Zuker folding program (14) (data not shown), sites 274 and 366 are within domains 4 and 6 of the RNA secondary structure of the Mo-MLV fpackaging region (20). These two domains are present in a segment of the packaging signal that appears to be absolutely required for qi function (21). After initiation of these experiments, it was shown that site 243, despite being unpaired, is close to domain 1 and was shown to be protected from all chemical modification by the compounds dimethyl sulfate, kethoxal, and 1-cyclohexyl-3-(2morpholinoethyl)carbodiimide metho-p-toluene sulfonate (20). These results suggested that domain 1 pairs with another sequence that lies outside the sequence analyzed (20). Therefore, the most likely explanation for the inability of Rz243 to function both in vitro and in vivo lies in the inaccessibility of the target sequence within domain 1. On the basis of the results in the Mo-MLV system, we constructed an anti-HIV-1 4i packaging site ribozyme (RzHIVO) and demonstrated the efficacy of this ribozyme construct in inhibiting HIV-1 infectivity and replication in the human T-cell line SupT1. The inhibitory effect of the ribozyme is not due to a non-specific inhibitory effect of the vector sequence as the introduction of vector alone did not suppress HIV-1 infectivity or replication, nor is it due to a reduction in the number of CD4 receptors in the constructtransfected cells. As to the possible mechanism of ribozyme action, several points can be made. The level at which inhibition occurs could be multiple. In the Mo-MLV system, the ribozyme constructs were introduced into already virusproducing cells (3T3 Mo-MLV) in which Mo-MLV provirus was integrated. In this case, the only steps at which the ribozymes could act are postintegration, and inhibition at these steps was clearly sufficient to inhibit viral replication. By contrast, the constructs in the HIV-1 system were transfected into uninfected cells (SupTl), which were subsequently challenged with HIV-1 and shown to be protected. In this case, inhibition could be at all the levels noted above at both pre- and postintegration steps including virus packaging. In addition to analysis of the correlation between in vitro cutting and in vivo activities, different designs were used in this study: a multiple ribozyme (long antisense sequence with four ribozyme domains inserted at different sites), single ribozymes, and antisense sequence alone. From our in vitro cleavage data, the multiple ribozyme appeared to be more efficient in cleaving target substrate than the individual single ribozymes. Another report has also described greater apparent efficacy of cutting by a multiple ribozyme (22). It can be envisaged that a multiple ribozyme may be more efficient due to the ability to cut at different sites and the fact that these sites may have different accessibilities in different conformations of the RNA molecule. By contrast, in in vivo viral suppression assays there was no apparent difference between the single and multiple ribozyme. This may indicate that in vitro cleavage is more sensitive than the in vivo assay, and/or that in vivo the substrate and ribozyme conformations can be modified by the cellular environment to modulate efficiency of action. In both systems, our results indicate that ribozyme cleavage of targets within the i packaging site has caused a suppression of virus production. As shown in the result from the Mo-MLV system, antisense-expressing cells also appeared to be protected from Mo-MLV replication at a level

9719

similar to ribozyme-expressing cells. This is consistent with the published observation in transgenic mice in which a 540-base antisense sequence to the Mo-MLV i region was used (13). This does not, however, detract from the ribozyme result, which shows a varying level of inhibition dependent on the in vitro cleavage capability. Our data suggest that the ribozyme construct effects seen are not due to an antisense effect since Rz243 does not exhibit inhibition of syncytia or reduction in RNA levels, yet like the other ribozymes it has the same size antisense arms. However, a longer , site antisense (535 bases) sequence can also work. We conclude that ribozymes targeted to the i packaging region of retroviruses, both murine and human, can be highly effective in suppressing both infectivity and replication in vivo. The use of anti-+i site ribozymes singly or in combination may be a useful gene-therapy approach to the treatment of retroviral diseases such as AIDS. We thank Phil Jennings, Wayne Gerlach, Yoichi Takayama, Lynn Bonham, and Peter Rowe for helpful discussions and Jagdeesh Pyati for providing a subclone of HIV-1 genome pRASH-5. The work was initiated at the Children's Medical Research Institute, Westmead, Australia, and was supported by a Gene Shears Research and Development Contract. 1. Rossi, J. J., Cantin, E. M., Sarver, N. & Chang, P. F. (1991) Pharmacol. Ther. 50, 245-254. 2. Haseloff, J. & Gerlach, W. J. (1988) Nature (London) 334,

585-591. 3. Sarver, N., Cantin, E. M., Chang, P. S., Zaia, J. A., Ladne, P. A., Stephens, D. A. & Rossi, J. J. (1990) Science 247, 1222-1225. 4. Sioud, M. & Drlica, K. (1991) Proc. Nati. Acad. Sci. USA 88, 7303-7309. 5. Dropulic, B., Lin, N. H., Martin, M. A. & Jeang, K.-T. (1992) J. Virol. 66, 1432-1441. 6. Weerasinghe, M., Liem, S. E., Asad, S., Read, S. E. & Joshi,

S. (1991) J. Virol. 65, 5531-5534. 7. Lo, K. M. S., Biasolo, M. A., Dehni, G. P. & Haseltine, W. A. (1992) Virology 190, 176-183. 8. Homann, M., Tzortzakaki, S., Rittner, K., Sczakiel, G. & Tabler, M. (1993) Nucleic Acids Res. 21, 2809-2814. 9. Qiwang, J. 0., Hampel, A., Looney, D. J., Wong-Staal, F. & Rappaport, J. (1992) Proc. Natl. Acad. Sci. USA 89, 1080210806. 10. Yu, M., Qjwang, J. O., Yamada, O., Hampel, A., Rappaport, J., Looney, D. & Wong-Staal, F. (1993) Proc. Natl. Acad. Sci. USA 90, 6340-6344. 11. Man, R. & Baltimore, D. (1985) J. Virol. 54, 401-407. 12. Lever, A. M. L., Gottlinger, H., Haseltine, W. & Sodroski, J. (1989) J. Virol. 63, 4085-4087. 13. Han, L., Yun, J. S. & Wagner, T. E. (1991) Proc. Natl. Acad. Sci. USA 88, 4313-4317. 14. Zuker, M. & Stiegler, P. (1991) Nucleic Acids Res. 9, 133-148. 15. Coffin, J. (1985) in RNA Tumor Viruses, eds. Weis, R., Teich, N., Varmus, H. & Coffin, J. (Cold Spring Harbor Lab. Press, Plainview, NY), Vol. 2, pp. 766-782. 16. Warrilow, D., Takayama, Y. & Symonds, G. (1992) BioTechniques 13, 42-43. 17. Gautsch, J. W. & Meier, H. (1976) Virology 72, 509-513. 18. Reed, K. C. & Mann, D. A. (1985) Nucleic Acids Res. 13, 7207-7221. 19. Harrison, G. P. & Lever, A. M. L. (1992) J. Virol. 66, 41444153. 20. Alford, R. L., Honda, S., Lawrence, C. B. & Belmont, J. W.

(1991) Virology 183, 611-619.

21. Adam, M. A. & Miller, A. D. (1988) J. Virol. 62, 3802-3806. 22. Chen, C.-J., Banerjea, A. C., Harmison, G. G., Haglund, K. & Schubert, M. (1992) Nucleic Acids Res. 20, 4581-4589.