Proc. Nati. Acad. Sci. USAVol. 86, pp. 5963-5967, August 1989Medical Sciences

Studies of the mechanisms of action of the antiretroviral agentshypericin and pseudohypericin

(AIDS/therapeutic/human immunodeficiency virus/polycyclic aromatic diones/virus assembly)

GAD LAVIE*, FRED VALENTINEt, BRANDI LEVIN*, YEHUDA MAZURt, GLORIA GALLO*, DAVID LAVIEt,DAVID WEINER§, AND DANIEL MERUELO*¶*Department of Pathology, Kaplan Cancer Center, and tDepartment of Medicine, New York University Medical Center, 550 First Avenue, New York, NY10016; tDepartment of Organic Chemistry, The Weizmann Institute of Science, Rehovot 76 100, Israel; and §Department of Pathology and Medicine,University of Pennsylvania, Philadelphia, PA 19104

Communicated by Michael Heidelberger, May 8, 1989

ABSTRACT Administration of the aromatic polycyclicdione compounds hypericin or pseudohypericin to experimen-tal animals provides protection from disease induced by ret-roviruses that give rise to acute, as well as slowly progressive,diseases. For example, survival from Friend virus-inducedleukemia is significantly prolonged by both compounds, withhypericin showing the greater potency. Viremia induced byLP-BM5 murine immunodeficiency virus is markedly sup-pressed after infrequent dosage of either substance. Thesecompounds affect the retroviral infection and replication cycleat least at two different points: (i) Assembly or processing ofintact virions from infected cells was shown to be affected byhypericin. Electron microscopy of hypericin-treated, virus-producing cells revealed the production of particles containingimmature or abnormally assembled cores, suggesting the com-pounds may interfere with processing of gag-encoded precur-sor polyproteins. The released virions contain no detectableactivity of reverse transcriptase. (ii) Hypericin and pseudohy-pericin also directly inactivate mature and properly assembledretroviruses as determined by assays for reverse transcriptaseand infectivity. Accumulating data from our laboratories sug-gest that these compounds inhibit retroviruses by unconven-tional mechanisms and that the potential therapeutic value ofhypericin and pseudohypericin should be explored in diseasessuch as AIDS.

We recently reported (1) that two naturally occurring poly-cyclic aromatic diones, hypericin and pseudohypericin, pos-sess antiretroviral activity. The two compounds, which arederived from the plants of the Hypericum genus (St.Johnswort) (2-5), markedly suppress the spread of murineretrovirus infections both in vivo and in vitro (1). We havenow compared the mechanisms of action and the therapeuticpotentials of different doses of these agents in two murineretroviral systems. In the Friend virus system (6-8) thecompounds can preclude the onset of acute Friend virus-induced erythroleukemia. In the other system a more slowlyprogressing, fatal form of murine immunodeficiency is in-duced by the LP-BM5 virus (9, 10); hypericin and pseudo-hypericin prevent development of significant viremia andminimize disease in mice infected with the LP-BM5 virus.The modes of action of the two agents are unusual.

Hypericin and pseudohypericin interfere with assembly and/or processing of viral components, significantly decreasingthe number of morphologically mature viral particles pro-duced by infected cells without effects on intracellular levelsof viral mRNA and viral protein synthesis (1). In addition, asdetermined by assays for infectivity and reverse transcrip-tase, the compounds directly inactivate mouse and human

retroviruses, including human immunodeficiency virus(HIV). To our knowledge no other antiretroviral drug pos-sesses the capacity to inactivate directly intact retrovirions aswell as interfere with their assembly and/or processing.

MATERIALS AND METHODSMice. BALB/cJ, BALB/cBy, and C57BL/6J mice used in

these studies were bred at our animal facility at New YorkUniversity Medical Center.

Cell Cultures. AQR and B10.T(6R) murine leukemia celllines, infected with and secreting radiation leukemia virus(RadLV), were derived and adapted to growth in culture asdescribed (11). SC1 cells secreting LP-BM5 ecotropic andMink focus xenotropic virus were a gift from H. C. Morse III(National Institutes of Health). The human CEM T cell linewas used to propagate HIV isolated by coculturing bloodmononuclear cells from infected individuals with cells fromnormal donors.

Collection of Blood Samples. Mice were bled from theretroorbital plexus of the eye. Serum was then separated andfrozen at -70'C until used.

Viruses, Infectious Procedures, and Reverse TranscriptaseAssays. Details and protocols have been published (1, 9-13).An important modification of protocols was introduced thatmarkedly diminishes in vitro toxicities associated with hy-pericin and pseudohypericin. These agents are always pre-mixed with medium containing 10-20% fetal calf serumbefore incubation with cells or virions. In addition, theconcentrations of cells used for initial exposure to theseagents is as high as possible (e.g., 5 x 106 cells per ml). Afterexposure to the agents, cells are diluted to their normalconcentrations for proper in vitro maintenance and growth(2-5 x 105 cells per ml) (F.V., D.M., V. Itri, and G.L.,unpublished work).

Extraction of Hypericin and Pseudohypericin. Hypericin(C30H1608, Mr 504.43) and pseudohypericin (C30H1609, Mr520.43) were isolated from the herbs of Hypericum triquet-rifolium Turra during the flowering time of the plants asdescribed (1).

Preparation of Cells for EM. After incubation in culturecells were pelleted and fixed in 1% glutaraldehyde/0.1 Mcacodylate hydrochloride, pH 7.3, stained with OS04, andembedded in spur resin for examination of thin sections.

RESULTSIn Vivo Effects of Hypericin and Pseudohypericin. Fig. 1

illustrates the effects of a single i.v. administration of hyper-

Abbreviations: HIV, human immunodeficiency virus; RadLV, radi-ation-induced leukemia virus.9To whom reprint requests should be addressed.

5963

The publication costs of this article were defrayed in part by page chargepayment. This article must therefore be hereby marked "advertisement"in accordance with 18 U.S.C. §1734 solely to indicate this fact.

5964 Medical Sciences: Lavie et al.

A

>

c 0 80 160 240

_W

01

0 B

50

0 80 160 240

Days after FV injection

FIG. 1. BALB/c mice (10 per group) were inoculated i.v. with106 focus-forming units of Friend virus (FV) (o); the control groupreceived phosphate-buffered saline. (A) Hypericin was given at dosesof 10 (a), 50 (o), and 150 (o) ,ug per mouse at the time Friend viruswas inoculated. (B) Pseudohypericin was given at doses of 10 (n), 50(o), and 150 (o) kg per mouse. Survival was followed for 240 days.

icin and pseudohypericin on the survival of BALB/c miceinfected with Friend virus. All animals receiving a 150-pgdose survived for the entire follow-up period of 240 dayswhen the compounds were coadministered with Friend viruson the day of virus inoculation. All untreated, virus-inoculated control animals were dead within 23 days. Lowerdoses (50 and 10 ,ug per mouse) are protective but do not leadto 100% survival. It is also evident at the lower doses thathypericin is a more potent antiretroviral agent than pseudo-hypericin.Although the Friend virus system permits testing the

activity of the compounds in an acute infection system, thismodel has several drawbacks. Transformation of erythroidprecursor cells occurs rapidly after virus inoculation. Thusthe system permits little delay in the administration ofhypericin or pseudohypericin to animals.To overcome these disadvantages, further studies of the in

vivo effects of hypericin and pseudohypericin were con-ducted on other retroviruses that give rise to slower andgradually progressing disease such as the LP-BM5 murineleukemia virus [which causes the murine acquired immuno-deficiency syndrome (MAIDS)]. Mice infected with LP-BM5virus develop lymphoadenopathy, splenomegaly, hypergam-maglobulinemia, T- and B-cell polyclonal proliferation, andprofound immunosuppression associated with enhanced sus-ceptibility to infection (9). Analysis of spleen and lymph nodecells from LP-BM5-infected mice has shown many similari-ties between the early course of this disease and that ofHIV-associated AIDS (9).

Fig. 2 shows that injection of LP-BM5 virus into C57BL/6mice results in onset of viremia after day 21. This viremia isprevented, at least during the 90-day follow-up period, by i.p.injections of hypericin or pseudohypericin once every 2weeks beginning on day 14 after virus inoculation. The

E

20

-5

0 7 14 21 28 35 42 49 63 77 84 91Days after infection with LP-BM5 virus

FIG. 2. Inhibition of LP-BM5 murine leukemia virus-inducedviremia by hypericin and pseudohypericin. LP-BM5 virus (associ-ated with murine acquired immunodeficiency disease) was inocu-lated i.p. into C57BL/6 mice (o, LP-BM5 alone). Hypericin (A) andpseudohypericin (o) (150 ,ug per mouse) were given i.p. once every2 weeks beginning on day 14 after infection. The mice were bled fromthe periorbital plexus ofthe eye at weekly intervals. Sera were frozenin liquid N2 and assayed for reverse transcriptase (RT). Enzymaticactivities in sera from hypericin- and pseudohypericin-treated as wellas from untreated mice are shown for the first 91 days postinocula-tion. Each experimental group consisted of four mice.

biweekly treatment with hypericin or pseudohypericin sig-nificantly ameliorates the lymphoproliferative disease in-duced by LP-BM5 (Table 1), a remarkable result consideringthe long intervals between treatments. LP-BM5 rapidly in-duces lymphoproliferation, causing marked changes in CD5+and CD8+ cells in spleen and lymph nodes within a few days(data not shown). Injections of hypericin and pseudohyper-icin beginning on days 21, 28, or 33 post-virus inoculation,when viremia is already present, lead to prompt reduction ofviremia (data not shown).

Hypericin Interferes with Assembly of Viral Molecules orProcessing of Viral Polyproteins Essential for Maturation. Theeffect of hypericin on assembly of virus is demonstrated bythe following experiments. Transformed AQR cells produc-ing murine RadLV (11) were treated with hypericin at variousdoses at 37°C for 30 min. Cell-free compound was removedby extensive washing, and subsequent virus production wasmonitored for 24 and 48 hr by analysis of reverse transcrip-tase activity in the culture supernatants. Fig. 3A shows amarked decline in the release ofreverse transcriptase activityinto the growth medium. As measured after 24 or 48 hr, 50%inhibition of virus release occurred in the presence of hyper-icin at -0.1 ,g/ml; complete inhibition for at least 24 hr ofvirus release was obtained with 1 ,ug/ml. Cell viability wasnot affected by this treatment.These findings indicate that release of mature virus detect-

able by reverse transcriptase assays into the growth mediumwas inhibited by intracellular or membrane-bound hypericin.Such a result could be explained by inhibition of virusproduction at any of several steps involved in the virus

Table 1. Effects of hypericin or pseudohypericin treatment onspleen size of LP-BM5-infected mice

Inhibition ofSpleen weight, g splenomegaly,

Treatment ± SE %Phosphate-buffered saline (i.p.) 0.104 ± 0.01LP-BM5 virus (i.p.) 1.886 ± 0.44LP-BM5 virus (i.p.)+ hypericin (150 ,g) 0.706 ± 0.03 66

LP-BM5 virus (i.p.)+ pseudohypericin (150 ,ug) 0.999 ± 0.34 50Antiretroviral agents were administered i.p. every 2 weeks begin-

ning on day 14. Spleen weight data are presented as mean (n = 4).

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

Proc. Natl. Acad. Sci. USA 86 (1989) 5965

replication cycle, starting from transcription of proviralDNA. However, we previously showed that hypericin andpseudohypericin do not affect viral mRNA steady-state lev-els, translation of viral mRNA, or transport of viral proteinsto the cell surface (1).Whether assembly or budding of viral particles is inhibited

by hypericin has been examined by EM. In a manner similarto that indicated in Fig. 3A, infected AQR cells that produceRadLV were washed four times with phosphate-bufferedsaline, incubated for 1 hr with hypericin, fixed, embedded,and sectioned forEM 24 hr later. Electron micrographs showthat treatment with hypericin at 10 ,.g/ml does not eliminateparticle production (Fig. 3B) compared with untreated cells(Fig. 3C). However, the majority of budded particles haveeither immature cores, lacking the electron-translucent icosa-hedral structures characteristic of C type retroviruses, orcontain improperly formed cores. These observations have

0.1 1 10 50 14Concentration of hypericin, .tg

been quantitated in Table 2. Hypericin, therefore, eitherinterferes with the proper assembly of virus cores, interfereswith the processing of viral proteins needed for core matu-ration, or disrupts formed mature cores.

Direct Inactivation of Retroviruses by Hypericin andPseudohypericin. Fig. 4 A (with RadLV) and B (with HIV)shows that hypericin and pseudohypericin can inactivatedirectly murine and human retroviruses as measured byreverse transcriptase assays. The inhibitory capacity of hy-pericin is again shown to exceed that of pseudohypericin.This inhibition is not a direct effect on the active enzyme itselfas shown previously (1). One explanation for these findingsis that hypericin interferes with release of the enzyme inactive form from its proenzymes or packaged organizationwithin the viral core. Another possibility is that loss ofreverse transcriptase activity results from the lysis of viralparticles by hypericin or pseudohypericin, preventing parti-

B

..t.jEA.*4 $-J orofitll4s;

; #spt

r_I_

:*V:_

*f .

.,.t'"'A.w*Qs*l'i

....*" ti

C

FIG. 3. (A) Effect of hypericinon production and budding ofRad-LV from infected AQR cells asmeasured by release of viral re-verse transcriptase (RT) into thegrowth medium after a 30-minpulse treatment ofAQR cells with0.1, 1, 10, 25, and 50 ,g/ml ofhypericin. Treated and untreatedcells were cultured, and growthmedium samples were collectedafter 24 (e) and 48 (o) hr of incu-bation. Virus was pelleted at100,000 x g, and the pellet wasanalyzed for reverse transcrip-tase. (B) Effect of hypericin asdetermined by EM; infected AQRcells were treated for 1 hr withhypericin at 10 ,g/ml. (x 200,000.)(C) Electron micrograph of un-treated cells. (x200,000.)

.):3

CD

V

'D

._

E

_o

cc0

I0-

-.

4*.i

Medical Sciences: Lavie et al.

I.- ,:

.1A.

5966 Medical Sciences: Lavie et al.

Table 2. Analysis of morphologically mature virus particlessecreted by AQR cells during 24-hr incubation with andwithout hypericin

Particles with

Fields Virus icosahedral cores, %Hypericin, scored, particles, Excluding Including

,ug/ml no. no. ghosts ghosts0 21 559 66 ± 21 46 ± 201 16 230 17 ± 16 11 ± 11

10 37 659 22 ± 16 17 ± 12Numbers given are ±SE.

cle pelleting during ultracentrifugation (100,000 x g), whichoccurs before the assay. To investigate this hypothesis, weexamined reverse transcriptase activity in both pellets andsupernatants from hypericin-treated virus preparations. Re-verse transcriptase activity could not be found in eitherfraction (Fig. 4A and data not shown), making it unlikely thatlysis of virus is the mechanism for elimination of viral reversetranscriptase activity by hypericin or pseudohypericin.

DISCUSSIONThis paper describes efforts to elucidate the mode(s) of actionby which hypericin and pseudohypericin exert their antiret-roviral activity. The data are suggestive of mechanisms attwo different levels. The first mechanism is at the level of adirect action of the compounds against virus and virus-infected cells. The other mechanism involves the connectionbetween function of these agents and their chemical struc-tures. The study was complicated by data suggesting that thecompounds affect multiple phases of the virus replication andinfection cycles. Hypericin and pseudohypericin seem tointerfere with proper assembly and maturation of virus coresof particles budding from cells. However, these compoundsalso inactivate both the infectivity and reverse transcriptaseactivity of intact viruses that have undergone proper assem-

20k A

I0

1

Ea~

Hypericin, /tg/ml

0

x

E

0.5

.t

20 40 60 80Concentration of compound, pg/ml

100

FIG. 4. Reverse transcriptase (RT) analyses after direct expo-sure of virus to hypericin (e) and pseudohypericin (o). Virus wascollected from AQR cells producing RadLV and exposed to hyper-icin for 1 hr at 40C (A) and CEM cells producing HIV exposed tohypericin and pseudohypericin for 1 hr at 370C (B). Virus was thenpelleted at 50,000 x g, lysed with 0.5% Triton X-100, and analyzedfor reverse transcriptase.

bly and budding. Elsewhere (F.V., D.M., V. Itri, and G.L.,unpublished work; and D.W., G.L., W. V. Williams, D.L.,M. I. Greene, and D.M., unpublished work) we show thathypericin interferes with de novo infection of cells with HIVand viral-mediated syncytia formation. Accounting for allthese effects with a simple, unifying explanation is clearlydifficult.

Several explanations might account for the interferencewith proper assembly and/or maturation of retroviruses byhypericin. The occurrence of immature cores in virus prep-arations has been attributed in the Rauscher and Moloneyleukemia systems to the absence of processing of gag-precursor polyproteins (Pr65gag) (14, 15). Studies in both theavian and murine retrovirus systems have shown that viralparticles can be formed in the absence of envelope glyco-proteins (15-20), reverse transcriptase (15, 21-23), or ge-nomic RNA (15, 24, 25). By contrast, physical particles arenot generated when avian virus gag-related polyproteins areeither not cleaved or synthesized (15, 26-29).

It has been shown with murine retroviruses that thematuration of the core structure after virus budding, as seenin EM, may relate to cleavage of the precursor polyproteins(15, 30, 31). Deletion mutations in the protease regions of thevirus genome have been shown to lead to the production ofimmature forms of virus core (15, 32). The potential effectsof hypericin and pseudohypericin on protease activity needevaluation.

Alternatively, hypericin could interfere with importantsteps occurring in virus assembly at the cell membrane. Forexample, a hypothesis has been proposed suggesting that gagand gag-pol polyproteins become associated at the cell-surface membrane to interact at their amino termini with atransmembrane segment of the envelope-glycoprotein com-plex and interact at their carboxyl termini with genomic RNA(15, 33). Hypericin and pseudohypericin are known to belypophylic molecules that can intercalate in the cell mem-brane. Thus, possibly these compounds disrupt the gag andgag-pol polyprotein interactions within the plasma mem-brane essential for assembly and/or the encapsulation ofRNA into viral particles. For example, these compounds mayselectively bind to certain viral polyproteins or proteins andinterfere with their packaging role.As stated above, hypericin and pseudohypericin treatment

also results in complete inactivation of reverse transcriptasein murine and human viruses. It is important to clarify thatthis effect is not due to interference by hypericin or pseudo-hypericin with the reverse transcriptase enzymatic reactionbecause hypericin or pseudohypericin do not affect reversetranscriptase activity when commercially purified enzyme isused. An explanation for this direct inactivation of virusconsistent with the interference with polyprotein processingby hypericin that leads to improper assembly and/or matu-ration of retrovirus particles would be as follows (for whichno data are yet available). Reverse transcriptase within thecore of intact virus would be assumed to be an inactiveenzyme or proenzyme kept nonfunctional either by tightassociation with another viral core component [such as RNA(34)] or by its tertiary structure within the virion core. Suchan inactive form might be rendered active by viral-encodedproteases or by other mechanism (e.g., kinases) duringinfection of cells or whenever virions are disrupted. Hyper-icin could then interfere with this activation or release ofreverse transcriptase. Purified reverse transcriptase is notaffected by hypericin, because enzyme activation or releasehad already occurred during purification of the enzyme fromvirus particles.The difference in activities between hypericin and pseudo-

hypericin is an important additional observation. The avail-ability of two structurally related compounds that differ in asingle hydroxyl group side-chain substitution, yet have sig-

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

Proc. Natl. Acad. Sci. USA 86 (1989) 5967

AOH 0 OH

B

OH 0 OH

HOQ i)CH2-OH

OH 0 OH

FIG. 5. Chemical structure of hypericin (A) and pseudohypericin(B).

nificantly different biological activities, enables a preliminaryinsight into structure-function relationships. With a maincommon structural feature of a flat core of eight fusedaromatic rings encircled by six phenolic hydroxyls (see Fig.5), the substitution for the seventh hydroxyl group thatoccurs in pseudohypericin clearly diminishes antiretroviralactivity. Further study of this structure-function relationshipmay help guide development of even more powerful antiret-roviral agents.An important aspect of the biological activity of hypericin

and pseudohypericin, only briefly demonstrated in this manu-script, is their anti-HIV activity. The anti-HIV activity ofthese compounds will be further documented elsewhere.Data presented here, in ref. 1, and in the work to be publishedindicate the need to perform the requisite studies of preclin-ical toxicology and clinical trials to determine the potentialtherapeutic value of hypericin and pseudohypericin in dis-eases such as AIDS.

1. Meruelo, D., Lavie, G. & Lavie, D. (1988) Proc. Natl. Acad.Sci. USA 85, 5230-5234.

2. Brockmann, H. V., Falkenhausen, E. H., Neeff, R., Dorlars,A. & Budde, G. (1950) Naturwissenschaften 37, 540-545.

3. Brockmann, H. V., Falkenhausen, E. H., Neeff, R., Dorlars,A. & Budde, G. (1951) Chem. Ber. 84, 865-887.

4. Brockmann, H. V., Franssen, U., Spitzner, D. & Augustiniak,H. (1974) Tetrahedr. Lett. 23, 1991-1994.

5. Brockmann, H. V. & Spitzner, D. (1975) Tetrahedr. Lett. 1,37-40.

6. Steeves, R. A., Eckner, R. J., Bennett, M., Mirand, E. A. &Trudel, P. J. (1971) J. Natl. Cancer Inst. 46, 1209-1217.

7. Friend, C. (1957) J. Exp. Med. 105, 307-326.8. Troxler, D. H. & Scolnick, E. M. (1978) Virology 85, 17-27.

9. Klinken, S. P., Frederickson, T. N., Hartley, J. W., Yetter,R. A. & Morse, H. C., III (1988) J. Immunol. 140, 1123-1131.

10. Mosier, D. E., Yetter, R. A. & Morse, H. C., III (1985) J. Exp.Med. 161, 766-784.

11. Bach, R. G. & Meruelo, D. (1984) J. Exp. Med. 160, 270-285.12. Meruelo, D., Lieberman, M., Deak, B. & McDevitt, H. 0.

(1977) J. Exp. Med. 146, 1088-1095.13. Levy, J. A. & Shimabukuro, J. (1985) J. Infect. Dis. 152,

734-738.14. Yoshinaka, Y., Ishigame, K., Ohmo, T., Kageyama, S., Shi-

bata, K. & Luftig, R. (1980) Virology 100, 130-140.15. Weiss, R., Teich, N., Varmus, H. & Coffin, J., eds. (1984) RNA

Tumor Viruses (Cold Spring Harbor Lab., Cold Spring Harbor,NY), Vol. 1, pp. 513-648.

16. Weiss, R. A. (1969) J. Gen. Virol. 5, 529-539.17. Scheele, C. M. & Hanafusa, H. (1971) Virology 45, 401-410.18. Kawais, S. & Hanafusa, H. (1973) Proc. Nati. Acad. Sci. USA

70, 3492-3497.19. Linial, M., Fenno, J., Burnette, V. N. & Rohrschneider,

L. M. J. (1980) J. Virol. 36, 280-290.20. Ramsay, G. & Hagman, M. J. (1980) Virology 106, 71-81.21. Hanafusa, H., Baltimore, D., Smoler, D., Watsm, K. F.,

Yaniv, A. & Spiegelman, S. (1972) Science 177, 1188-1191.22. Eisenman, R. N., Mason, W. S. & Linial, M. (1980) J. Virol.

36, 62-78.23. Ramsay, G. & Hayman, M. J. (1980) Virology 106, 71-81.24. Levin, J. G., Grimley, P. M., Ramseur, J. M. & Berzesky,

J. K. (1974) J. Virol. 14, 152-161.25. Linial, M., Mederiros, E. & Hayward, W. S. (1978) Cell 15,

1371-1381.26. Eisenman, R., Vogt, V. M. & Diggelmann, H. (1975) Cold

Spring Harbor Symp. Quant. Biol. 39, 1067-1075.27. Eisenman, R., Burnette, W. N., Zucco, F., Diggelman, H.,

Heater, P., Tsichilis, P. & Coffin, J. (1980) in Biosynthesis,Modification, and Processing of Cellular and Viral Polypro-teins, eds. Koch, G. & Richter, D. (Academic, New York).

28. Bister, K., Hayman, M. J. & Vogt, P. K. (1977) Virology 83,431-448.

29. Hayman, M. J., Royer-Pokora, B. & Graf, T. (1979) Virology92, 31-45.

30. Yoshinaka, Y. & Luftig, E. (1977) Proc. Natl. Acad. Sci. USA74, 3446-3450.

31. Pinter, A. & der Harven, E. (1980) Virology 99, 103-110.32. Katoh, I., Yoshinaka, Y., Rein, A., Shibuya, M., Odaka, T. &

Oroszlan, S. (1985) Virology 145, 280-292.33. Bolognesi, D. P., Montelaro, R. C., Frank, H. & Schafer, W.

(1978) Science 199, 183-186.34. Panet, A., Haseltine, W. A., Baltimore, D., Peters, G.,

Harada, F. & Dahlberg, J. E. (1975) Proc. Natl. Acad. Sci.USA 72, 2535-2539.

Medical Sciences: Lavie et al.