Inhibition of human immunodeficiency virus type 1

Proc. Natl. Acad. Sci. USA Vol. 86, pp. 7191-7194, September 1989 Medical Sciences

Phosphorothioate and cordycepin analogues of 2',5'-oligoadenylate: Inhibition of human immunodeficiency virus type 1 reverse transcriptase and infection in vitro (antiviral agents/RNase L)

DAVID C. MONTEFIORI*, ROBERT W. SOBOL, JR.t, SHI Wu Lit, NANCY L. REICHENBACHt, ROBERT J. SUHADOLNIKt, RAMAMURTHY CHARUBALAt, WOLFGANG PFLEIDERERf, ANN MODLISZEWSKI*, W. EDWARD ROBINSON, JR.*, AND WILLIAM M. MITCHELL* *Department of Pathology, Vanderbilt University, School of Medicine, Nashville, TN 37232; tDepartment of Biochemistry, Temple University School of Medicine, Philadelphia, PA 19140; tFakultat fur Chemie, Universitat Konstanz, Konstanz D-7750, Federal Republic of Germany

Communicated by Frank Lilly, June 26, 1989

ATP. 2-5A, in turn, binds to and activates a latent endoribonuclease (RNase L) that cleaves viral and cellular RNA, thereby inhibiting protein synthesis (12). A series of reports (13-22) described the synthesis and ability of phosphorothioate and cordycepin analogues of 2-5A to replace authentic 2-5A in latent RNase L activation. In this report we demonstrate that some of these same analogues inhibit HIV-1 RT and protect target cells from HIV-1 infection in vitro.

Natural antiviral activity can be mediated by ABSTRACT the interferon-induced synthesis of 2',5'-oligoadenylates (2SAs) and subsequent RNase L activation by these molecules. Analogues of 2-5A that are biologically active and metabolically stable were synthesized and analyzed for antiviral activity against the human immunodeficiency virus type 1 (HIV-1). Replacement of the 3' hydroxyl group of the adenosine moieties of 2-5A with hydrogen atoms (i.e., cordycepin analogues of 2-5A) converted authentic 2-5A trimer into anti-HIV-1 agents in vitro. These cordycepin analogues of 2-5A also inhibited partially purified HIV-1 reverse transcriptase. Introduction of chirality into the 2',5'-phosphodiester internucleotide linkages or 5'-phosphate moieties of the 2-5A molecule (i.e., phosphorothioate analogues of 2-5A) converted authentic 2-5A into more potent inhibitors of HIIV-1 reverse transcriptase. However, these phosphorothioate 2-5As demonstrated little or no anti-HIV-1 activity in vitro. Thus, some analogues of 2-5A may form a class of anti-HIV-l drugs with possible' pleiotropic activities that include activation of latent RNase L and inhibition of reverse transcription.

MATERIALS AND METHODS Cells and Virus. The highly HIV-1 permissive CD4+ lym-

phoblastoid cell line MT-2 (23, 24) was used as target cells for infection with HIV-1 (strain HTLV-IIIB) produced in H9 cells (25). Stock cultures were grown and maintained in RPMI 1640 medium containing 12% (vol/vol) heat-inactivated fetal bovine serum and 50 ,ug of gentamicin per ml and incubated at 37°C. Viral titers in this study, which are given as a multiplicity of infection (i.e., infectious virus particles per cell), were determined by end-point microtitration on MT-2 cells as described (26). Enzymatic and Chemical Syntheses of the 2-5A Analogues. The enzymatic and chemical syntheses of the 2',5'cordycepin trimer and tetramer cores and their 5'-mono-, di-, and -triphosphates and their structural elucidations have been described (13-22). The enzymatic and chemical syntheses of the 2',5'-phosphorothioate tetramer monophosphates and their structural elucidation were essentially as described for the respective trimer syntheses (13, 14, 16). 2',5'-Adenylyl-3'-deoxyadenylyladenosine core (A-C-A) was synthesized as described (17). TLC analyses [Eastman Chromagram, 13254; solvent, isobutyric acid/ammonia/water, 66:1:33 (vol/vol)] revealed ultraviolet-absorbing regions for the 5'-mono-, -di-, and -triphosphates (Rf values were 0.66, 0.58, 0.50, respectively). Polyethylenimine-cellulose TLC (Brinkman; solvent, 0.25 M ammonium bicarbonate) revealed ultraviolet-absorbing regions with charges of 4-, 5-, and 6-, respectively, which agreed with the charges for authentic 2-5A trimer core, 5'-mono-, -di-, and -triphosphates (Rf values: 5'-mono-, 0.58; 5'-di-, 0.38; 5'-triphosphate, 0.19). Chemicals. The trimer of adenylic acid with 2',5'phosphodiester linkages and a 5'-triphosphate (p3A3) and the 5'-dephosphorylated p3A3 (A3) were obtained from Pharma-

Human immunodeficiency virus type 1 (HIV-1), the etiologic agent of acquired immunodeficiency syndrome (AIDS), is a lentivirus that primarily infects CD4+ lymphocytes (T4 cells) and cells of monocyte/macrophage lineage ultimately resulting in diverse immune perturbations and ensuing host susceptibility to severe and life-threatening opportunistic infections (1-3). As in all retroviruses, an essential feature of HIV-1 replication is reverse transcription of the plus-strand RNA genome into DNA, a process that requires an RNAdependent DNA polymerase, also known as reverse transcriptase (RT) (4, 5). This enzyme is viral-encoded and is found associated with genomic RNA in mature HIV virions (6). The relative restriction of RT to retroviruses and viruses requiring a short reverse-transcription step makes RT a major target for antiviral, and particularly for antiretroviral, therapeutic intervention. In fact, drugs such as 3'-azido3'-deoxythymidine (AZT) (7), 2',3'-dideoxynucleosides (8), and phosphonoformate (9) have been shown to effectively block HIV-1 reverse transcription in vitro at concentrations that had no cellular toxicity. Replication of many viruses is also blocked by the 2',5'oligoadenylate [2',5'-ppp(Ap)nA or 2-SA] system, which is part of the host's natural antiviral response induced by interferons (for reviews, see refs. 10 and 11). In the presence of interferons, increased 2',5'-oligoadenylate synthetase gene expression results in increased synthesis of 2-5A from

Abbreviations: HIV-1, human immunodeficiency virus type 1; RT, reverse transcriptase; 2-SA, 2',5'-oligoadenylates; p3A3, trimer of adenylic acid with 2',5'-phosphodiester linkages and a 5'-triphosphate; A3, 5'-dephosphorylated p3A3; C, p3C3, C3, etc., cor-

dycepin (3'-deoxyadenosine), 2',5'-cordycepin analogues of p3A3, 5'-dephosphorylated p3C3, etc.; p3A3aS, 5'-0-[(Sp)-1-P-thiotriphos-

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.

phoryl]-(Rp)-P-thioadenylyl-2',5'-(Rp)-P-thioadenylyl-2',5'-adenosine; A-C-A, 2',5'-adenylyl-3'-deoxyadenylyladenosine core.

7191

Medical Sciences: Montefiori et al.

7192

Proc. Natl. Acad. Sci. USA 86

cia. Cordycepin and adenosine 5'-[thio]monophosphate (AMP[S]) were purchased from Sigma. Mismatched doublestranded RNA [poly(I) poly(C12-U), Ampligen] was obtained as a lyophilized powder in salt from HEM Research, Rockville, MD. RT Assays. Virus was concentrated from cell-free (0.45-,um pore size filtered) H9/HTLV-IIIB-conditioned culture supernatants by centrifugation at 18,000 rpm for 4 hr at 200C in a Beckman JA-20 rotor. A viral pellet obtained from 50 ml of conditioned culture fluid was dissolved in 0.5 ml of a solution containing 17 mM Tris'HCl (pH 7.8), 3 mM dithiothreitol, 55 mM KCl, 0.32% Triton X-100, and 33% (vol/vol) glycerol. This viral lysate was stored at -20'C and was used as a source of HIV-1 RT. RT reactions were performed in 100-,ul reaction volumes containing 40 mM Tris HCl (pH 7.8), 4 mM dithiothreitol, 50 mM KCl, 10 mM MgCl2, 0.0325% Triton X-100, 3 tLM [3H]dTTP (80 Ci/mmol; 1 Ci = 37 GBq; NEN) and poly(A)-(dT)15 (2.5 ,g/ml) template-primer after the addition of 10 ,ul of enzyme. Inhibitors were added at various concentrations after adjusting water volumes so that reaction volumes remained constant. Reaction mixtures were incubated at 37°C for 1 hr in a humidified environment and terminated by adding 2 ml of 10% (wt/vol) ice-cold trichloroacetic acid. Precipitate was collected on 0.45-,Lm (pore size) cellulose-acetate Millipore filters that were then dissolved in 10 ml of 3a70B aqueous scintillant (Research Products International, Mt. Prospect, IL) and the cpm were quantitated using a Beckman LS 6800 liquid scintillation spectrometer. Infection Assays. Anti-HIV-1 activities of various compounds were detected and quantitated by an in vitro microtiter infection assay as described (26). Briefly, MT-2 cells were added to 96-well microdilution plates containing 1:2 serial dilutions of effector in triplicate. Virus was added at a multiplicity of infection of 1 and the plates were incubated at 37°C in a humidified 5% C02/95% air environment for 3 days. Viable cells were then quantitated by vital dye (neutral red) uptake by poly(L-lysine)-adherent cells as a measure of cytopathic effect. At this time, virus control wells (cells and virus in the absence of effectors) exhibited greater than 90% cytolysis. Percent protection was defined by the range of A5,0 readings occurring between cell control wells (cells in the absence of virus and effectors) and virus control wells. The multiplicity of infection was determined by end-point microtitration on MT-2 cells in this same assay.

(1989)

Cell Toxicity Assays. Cell toxicities were quantitated using MT-2 cells in microdilution plates as described above with the exceptions of omitting virus and replacing virus control wells with empty (blank) wells. The range of A540 readings occurring between cell control wells and blank wells after 3 days of incubation was used to calculate percent viable cells in test wells.

RESULTS AND DISCUSSION The short half-life of authentic 2-5A in biological systems (12-18) is an acknowledged disadvantage in the control of viral replication. In the studies described here, this disadvantage was surmounted by selectively altering 2-SA to provide analogues with increased metabolic stability. Modification of 2-SA at the 3'-hydroxyl groups provides various 2-SA analogues (i.e., 2'-5'-cordycepin 5'-mono-, -di-, and -triphosphates) with remarkably increased metabolic stability to 2'-phosphodiesterase and cellular nucleases, while maintaining the ability to activate RNase L (15, 18, 19, 21, 22). Modification at the 2',5'-internucleotide linkages by phosphorothioate substitution has provided stereochemically modified 2-SA molecules with profoundly differing stereochemical, physicochemical, and biological properties (13, 14, 16). The replacement of oxygen with sulfur on the phosphorus not only introduces chirality into the phosphodiester bond of the 2',5'-internucleotide linkage or 5'-phosphate moieties of 2-SA but also produces changes in the degree of hydration, ionic interactions, and hydrogen bonding (13, 14). 100

80-

,60M t-.

100 C

0

>

:E

80M

0..

0

I

40-

.0

15

2.

.

1.0

1.5

2.0

2.5

60

0

100

200

300

400

FIG. 1. Effect of cordycepin analogues of 2-5A (shown as jkM) on HIV-1 RT. Percent inhibition values are relative to a control set of reactions containing no effectors. Control values averaged 129,000 cpm and background values averaged 10,000 cpm. *, p2C3; a, pC4; *, p3C3; O, pC3; A, C3; A, A-C-A.

0

0.5

FIG. 2. Inhibition of HIV-1 RT by phosphorothioate stereochemical analogues of 2-5A (shown as ,1uM). Values for incorporated radioisotope were 201,000 cpm, average for control reactions (no effectors), and 13,500 cpm, average for background. 0, p3A3aS; o, pSpRpRp; *, pSpSpSp; o, pRpSpSp; A, pRpRpRp.


Proc. Natl. Acad. Sci. USA 86 (1989)

7193

1-

0

Co

2 .0 0

0

at

JM

_v

0 c

0 0

juM

1u *i 80

at

60 0

40

20

0o

E

0

2

6

lo

0

_-

12.5

JM

25

.uM

50 0

01

0.2

0.3 ,uM

OA

0.5

FIG. 3. Effect of cordycepin analogues of 2-5A (shown as AM) on HIV-1 infection in vitro. Each data point represents the average of three values. Standard deviations were less than 10%o of average values. (A) C3. (B) pC3. (C) A-C-A. (D) pC4. (E) pRpSpSp.

The biological activity and metabolic stability ofthese 2-5A analogues made them attractive candidates for testing antiviral activity against HIV-1. Authentic 2-5A and a series of structural and stereochemical 2-5A analogues were tested by measuring anti-HIV-1 activity in an in vitro MT-2 cell infection assay and by examining the ability to inhibit HIV-1 RT. RT was not inhibited by A3 or p3A3 at 0.25-256 ,uM, by cordycepin at 25-400 ,uM, or by cordycepin 5'-triphosphate (p3C) at 25-200 AM (data not shown). In contrast, concentration-dependent enzyme inhibition was observed for 2'5'-cordycepin trimer core (C3), cordycepin trimer 5'-mono-, -di-, and -triphosphates (pC3, p2C3, p3C3), 2',5'-cordycepin tetramer 5'-monophosphate (pC4), and A-C-A (Fig. 1). p2C3 was the most effective cordycepin 2-5A analogue inhibitor of RT (50%o inhibition at 75 ,uM) followed by pC4 (50%0 inhibition at 118 uM), p3C3 (50% inhibition at 132 ILM), A-C-A (50% inhibition at 160 uM), pC3 (50% at 290 ,uM), and C3 (40% at 400 MM). The 2',5'-phosphorothioate tetramer 5'-monophosphate analogues of 2-5A (i.e., pA4 stereoisomers) were much more effective inhibitors of HIV-1 RT activity than the 2',5'cordycepin analogues. The pSpRpRp, pSpSpSp, pRpSpSp, and pRpRpRp diastereomers§ exhibited concentration-dependent inhibition up to 58%, 44%, 43%, and 30%, respectively, at 2.5 AM (Fig. 2). The most effective inhibitor of HIV-1 RT was the 2' ,5'-phosphorothioate trimer 5'-triphosphate {i.e., 5'0-[(Sp)-1-P-thiotriphosphoryl]-(Rp)-P-thioadenylyl-2' ,5 '(Rp)-P-thioadenylyl-2',5'-adenosine (p3A3aS)}, with 50% inhibition observed at 0.5 AM. AMP[S] (i.e., adenosine 5'0-phosphorothioate) did not inhibit RT activity at concentrations up to 200 gM (data not shown), thus suggesting

§pSpRpRp, pSpSpSp, pRpSpSp, and pRpRpRp tetramer 5'-monophos-

phates are 5'-monophosphorylated tetramer phosphorothioate analogues of A4 with Rp and Sp stereoconfigurations in the three chiral centers with assignment of configuration from the 5' terminus to the 2' terminus.

that degradation products were not responsible for the inhibition observed with the 2',5'-phosphorothioates. The strong inhibition of RT observed for the phosphorothioates suggested that they may possess potent antiretroviral activity. When tested for activity against HIV-1 using an in vitro MT-2 cell microtiter infection assay, however, only the pRpSpSp stereoisomer had activity. The antiHIV-1 activity of pRpSpSp was optimum at 0.04 juM, a concentration that provided 32% protection in the infection assay (Fig. 3). Furthermore, authentic A3 had no anti-HIV-1 activity when tested at exogenous concentrations of 8-250 ,uM in vitro (data not shown). Possible reasons why the remaining stereoisomers (i.e., the pSpRpRp, pSpSpSp, and pRpRpRp diastereomers and p3A3aS) had no in vitro antiHIV-1 activity include (i) the molecules are impermeable to the cell or (ii) they are metabolically altered intracellularly. However, their potent inhibition of RT should encourage the search for analogous prodrugs that are readily internalized, resist metabolic inactivation, and may even become activated intracellularly. In contrast to the phosphorothioates, potent concentration-dependent anti-HIV-1 activity was observed for the 2-SA analogues C3 (Fig. 3A) and pC3 (Fig. 3B) at exogenous concentrations of 62.5-125 ,LM and for A-C-A (Fig. 3C) at 4-8 ,uM. These activities were 84%, 80%, and 81%, respectively, of the optimum anti-HIV-1 activity provided by mismatched double-stranded RNA (data not shown). This is significant anti-HIV-1 activity considering that mismatched double-stranded RNA is a potent anti-HIV-1 drug in vitro (27, 28). Less-potent concentration-dependent anti-HIV-1 activity was observed for pC4 (Fig. 3D) at 12.5-50 ,uM. No toxicity was observed for C3 or A-C-A at their most effective antiviral concentrations whereas mild toxicity was observed at some of these optimum antiviral concentrations of pC3 and pC4. Complete inhibition of RT activity was never achieved even at the highest concentrations of 2',5'-cordycepin analogues of 2-SA tested, while lower concentrations of three of these analogues (i.e., A-C-A, C3, and pC3) elicited potent

7194


Proc. Natl. Acad Sci. USA 86 (1989)

antiviral activity in vitro (Figs. 1 and 3). These results may be an indication that a mechanism other than RT inhibition, such as RNase L activation, was responsible for the antiviral activity observed. However, the possibility that the cordycepin analogues of 2-5A were either concentrated intracellularly (19) or metabolically altered to more active RT inhibitors cannot be ruled out. To determine ifthe antiviral activity ofthe 2',5'-cordycepin analogues of 2-5A was due to degradation, antiviral and anticellular effects of cordycepin were measured. Cordycepin demonstrated no anti-HIV-1 activity in vitro while it inhibited MT-2 cell viability at concentrations as low as 5 AM in toxicity assays. Cordycepin also did not inhibit RT activity up to 400 ,M. This is in contrast to C3 that had pronounced antiviral activity but no cell toxicity at 62.5 AuM (Fig. 3). These findings support an earlier report that 0.55% and 1.3%, respectively, of 3H- and 32P-labeled C3 were taken up intact by lymphocytes in culture (19). It would appear that under the experimental conditions described here, C3 does not serve as a prodrug of cordycepin. p3A3 and a synthetic oligomeric mixture (dimer to hexamer) have been reported to inhibit avian and mammalian RTs (29) where avian myeloblastosis virus RT was inhibited by 50%o at 75 AM p3A3. It is shown here that with HIV-1 RT there is no inhibition by p3A3 at 200 pM. However, introduction of chirality into the p3A3 molecule (i.e., p3A3aS) resulted in dramatic inhibition of HIV-1 RT (50% inhibition at 0.5 AM p3A3aS). Therefore, the formation of an inhibitory complex may be facilitated by (i) a more direct interaction between the 2',5'-phosphorothioate analogues and HIV-1 RT and/or (ii) increased stability of the 2',5'-phosphorothioates compared to authentic 2-5A (13-15). A similar observation has been reported by Broder and coworkers (30). They demonstrated that nuclease-resistant 3',5'-phosphorothioate oligodeoxynucleotides exhibit potent antiviral activity against HIV-1. However, no attempts were made to separate the diastereoisomers. Further, it is unlikely that 3',5'-linked oligodeoxynucleotides would be capable of activating RNase L. The precise mechanism by which the 2-5A analogues described here exert their in vitro anti-HIV-1 activity remains to be established. It is not clear whether the antiviral activity was a direct result of RNase L activation, RT inhibition, or an unknown mechanism. If RT inhibition is involved, the lack of cellular toxicity at effective concentrations, especially for A-C-A (Fig. 3), indicates a certain degree of specificity for the HIV-1 RT over cellular polymerases. Future studies will be aimed at comparisons using purified HIV-1 RT and cellular polymerases. However, the results do suggest that 2-SA analogues may have therapeutic utility. In any event, phosphorothioate and cordycepin analogues of 2-SA offer intriguing possibilities for further development. We thank Mr. Clyde W. (Pete) Parrish and Mrs. Jean Kolodziej for their aid in preparing this manuscript. This study was supported by grants from the National Institutes of Health (AI 25272) (W.M.M.), U.S. Public Health Service (POI CA-29545) (R.J.S.), Federal work study awards (R.W.S.), and a University Fellowship awarded by the graduate school, Temple University (S.W.L.).

2. 3. 4. 5. 6. 7.

8. 9. 10.

11. 12. 13. 14.

15. 16.

17.

18. 19.

20. 21. 22.

23. 24. 25. 26. 27.

28. 29. 30.

1.

Klatzmann,

D.,

Barre-Sinoussi,

F.,

Nugeyre,

M. T., Gauguet,

C., Vilmer, E., Griscelli, C., Brun-Vezinet, F., Rouzioux, C.,

Gluckman, J. C., Chermann, J.-C. & Montagnier, L. (1984) Science 225, 59-63. Wong-Staal, F. & Gallo, R. (1985) Nature (London) 317, 395-403. Ho, D. D., Pomerantz, R. J. & Kaplan, J. C. (1987) N. Engl. J. Med. 317, 278-286. Temin, H. J. & Mitzutani, Y. (1970) Nature (London) 226, 1211-1213. Baltimore, D. (1970) Nature (London) 226, 1209-1211. Veronese, F. M., Copeland, T., DeVico, A. L., Rahman, R., Oroszlan, S., Gallo, R. C. & Sarngadharan, M. G. (1986) Science 231, 1289-1291. Mitsuya, H., Weinhold, K. J., Furman, P. A., St. Clair, M. H., Lehrman, S. N., Gallo, R. C., Bolognesi, D., Barry, D. W. & Broder, S. (1985) Proc. Nati. Acad. Sci. USA 82, 7096-7100. Mitsuya, H. & Broder, S. (1986) Proc. Natl. Acad. Sci. USA 83, 1911-1915. Sandstrom, E. G., Kaplan, J., Byington, R. E. & Hirsch, M. (1985) Lancet i, 1480-1482. Williams, B. R. G. & Silverman, R. H., eds. (1985) The 2-5A System (Liss, New York). Lengyel, P. (1982) Annu. Rev. Biochem. 51, 251-282. Hovanessian, A. G. & Kerr, I. M. (1979) Eur. J. Biochem. 93, 515-526. Karik6, K., Sobol, R. W., Jr., Suhadolnik, L., Li, S. W., Reichenbach, N. L., Suhadolnik, R. J., Charubala, R. & Pfleiderer, W. (1987) Biochemistry 26, 7127-7135. Karik6, K., Li, S. W., Sobol, R. W., Jr., Suhadolnik, R. J., Charubala, R. & Pfleiderer, W. (1987) Biochemistry 26, 71367142. Karik6, K., Reichenbach, N. L., Suhadolnik, R. J., Charubala, R. & Pfleiderer, W. (1987) Nucleosides Nucleotides 6, 497-500. Suhadolnik, R. J., Lee, C., Karik6, K. & Li, S. W. (1987) Biochemistry 26, 7143-7149. Charubala, R. & Pfleiderer, W. (1980) Tetrahedron Lett. 21, 4077-4080. Doetsch, P., Wu, J. M., Sawada, Y. & Suhadolnik, R. J. (1981) Nature (London) 291, 355-358. Suhadolnik, R. J., Doetsch, P. W., Devash, Y., Henderson, E. E., Charubala, R. & Pfleiderer, W. (1983) Nucleosides Nucleotides 2, 7143-7149. Karik6, K., Li, S. W., Sobol, R. W., Jr., Reichenbach, N. L., Charubala, R., Pfleiderer, W. & Suhadolnik, R. J. (1987) Nucleosides Nucleotides 6, 173-184. Lee, C. & Suhadolnik, R. J. (1983) FEBS Lett. 157, 205-209. Nyilas, A., Vrang, L., Drake, A., Oberg, B. & Chattopadhyaya, J. (1986) Acta Chem. Scand. 840, 678-688. Miyoshi, I., Kubonishi, I., Yoshimoto, S., Akagi, T., Ohtsuky, Y., Shiraishi, Y., Nagata, K. & Hinuma, Y. (1981) Nature (London) 294, 770-771. Haertle, T., Carrera, C. J., McDougal, J. S., Sowers, L. C., Richman, D. D. & Carson, D. A. (1988) J. Biol. Chem. 263, 5870-5875. Popovic, M., Sarngadharan, M. G., Read, E. & Gallo, R. C. (1984) Science 224, 497-500. Montefiori, D. C., Robinson, W. E., Jr., Schuffman, S. S. & Mitchell, W. M. (1987) J. Clin. Microbiol. 26, 231-235. Montefiori, D. C. & Mitchell, W. M. (1987) Proc. Natl. Acad. Sci. USA 84, 2985-2989. Montefiori, D. C., Robinson, W. E., Jr., & Mitchell, W. M. (1988) Antiviral Res. 9, 47-56. Liu, D. K. & Owens, G. F. (1987) Biochem. Biophys. Res. Commun. 145, 291-297. Matsukura, M., Shinozuka, K., Zon, G., Mitsuya, H., Reitz, M., Cohen, J. S. & Broder, S. (1987) Proc. Natl. Acad. Sci. USA 84, 7706-7710.