Proc. Nati. Acad. Sci. USAVol. 90, pp. 158-162, January 1993Medical Sciences

Photodynamic inactivation of infectivity of human immunodeficiencyvirus and other enveloped viruses using hypericin and rose bengal:Inhibition of fusion and syncytia formation

(vesicular stomatitis virus/influenza virus/Sendai virus/hemolysis)

JOHN LENARD*, ARNOLD RABSONt, AND ROGER VANDEROEF**Department of Physiology and Biophysics, Robert Wood Johnson Medical School, University of Medicine and Dentistry of New Jersey, 675 Hoes Lane,Piscataway, NJ 08854-5635; and tDepartment of Molecular Genetics and Microbiology, Center for Advanced Biotechnology and Medicine, Piscataway,NJ 08854

Communicated by Aaron J. Shatkin, September 29, 1992 (received for review May 15, 1992)

ABSTRACT The mechanism of the antiviral activity ofhypericin was characterized and compared with that of rosebengal. Both compounds inactivate enveloped (but not unen-veloped) viruses upon illumination by visible light. Humanimmunodeficiency and vesicular stomatitis viruses were pho-todynamically inactivated by both dyes at nanomolar concen-trations. Photodynamic inactivation of fusion (hemolysis) byvesicular stomatitis, influenza, and Sendai viruses was inducedby both dyes under similar conditions (e.g., I50 = 20-50nM forvesicular stomatitis virus), suggesting that loss of infectivityresulted from inactivation of fusion. Syncytium formation,between cells activated to express human immunodeficiencyvirus gpl20 on their surfaces and CD4+ cells, was inhibited byillumination in the presence of 1 ,uM hypericin. Hypericin androse bengal thus exert similar virucidal effects. Both presum-ably act by the same mechanism-namely, the inactivation ofthe viral fusion function by singlet oxygen produced uponillumination. The implications of this photodynamic antiviralaction for the potential therapeutic usefulness of both hypericinand rose bengal are discussed.

Microscopic observations of hypericin's intrinsic red flu-orescence in cultured cells showed it to be localized in plasmamembranes (2). This suggests that hypericin acts in a mannerquite different from the well-known nucleoside analogs.Observations of hypericin's activities in vitro have led tosuggestions that it may act on viral assembly (1, 2) or byinhibiting RT (4). Since these experiments did not controllight as a variable, their interpretation is difficult.

Hypericin has a long and well-documented history as aphotodynamic compound. It is the active principle containedin plants of the Hypericum genus; ingestion by grazinganimals causes photopoisoning (hypericism) leading to skinirritation, fever, and death (8, 9). Hypericin's photodynamicactivity against membranes in vitro was also noted very early,when it was found to be a promoter of light-induced hemol-ysis (H. F. Blum, quoted in ref. 10). In this regard it resem-bled rose bengal, another potent inducer of photodynamichemolysis (11-13).

As the worldwide epidemic of human immunodeficiencyvirus (HIV) infection and AIDS grows, the rapid develop-ment of effective, affordable agents with anti-HIV activity isan urgent need. To date, the most effective anti-HIV agentshave been nucleoside analogs that inhibit the HIV reversetranscriptase (RT). Recently, much attention has been givento classes of potential antiviral agents that target otheraspects of the viral life cycle. In this regard, hypericin hasbeen regarded as a promising potential anti-AIDS drug since1988, when it was reported that a single injection of thiscompound prevented splenomegaly and death in mice in-fected with Friend leukemia virus. Hypericin was "coadmin-istered" with the virus; 50% survival was conferred by 10 jigof hypericin, whereas 150 Ag was needed for completeprotection (1, 2). In addition, biweekly injections of hypericininto mice greatly reduced the viremia resulting from infectionwith the slower acting LP-BM5 murine leukemia virus (2).Others, however, could not replicate some of these results(ref. 3, note added in proof; N. R. Stevenson and J.L.,unpublished results).

Subsequent reports rapidly established two importantfacts: (i) hypericin inactivated a wide variety of lipid-containing (i.e., enveloped) viruses, but was ineffectiveagainst viruses lacking membranes (3-7); and (ii) light in-creased the virucidal potency of hypericin against severaldifferent enveloped viruses by at least 100-fold (6, 7).

ICHCHb

Cl HO 0 OH

Rose Bengal Hypeiln

The similarities between hypericin and rose bengal extendto their mechanism of action. Both compounds catalyze thelight-induced formation of singlet oxygen, a highly reactiveoxidizing species that is largely or completely responsible fortheir membrane-disruptive effects (14-16). In a further par-allel, rose bengal has been shown to inactivate envelopedviruses upon illumination. The photodynamic inactivation ofvaccinia virus by rose bengal was attributed to alterations inviral proteins, as opposed to nucleic acids (17). Rose bengal'svirucidal activity against herpes simplex virus was noted asa complication attending its use in a common diagnosticprocedure for ocular disease (18). However, in contrast tohypericin, rose bengal has not previously been evaluated asa potentially useful antiviral agent.

In view of these very similar properties of hypericin androse bengal, a comparison of their effects on enveloped virus

Abbreviations: HIV, human immunodeficiency virus; VSV, vesic-ular stomatitis virus; RT, reverse transcriptase; PMA, phorbol12-myristate 13-acetate.

158

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.

Proc. Natl. Acad. Sci. USA 90 (1993) 159

infectivity and fusion was carried out. Fusion was chosen forstudy because it is the membrane-specific process that allenveloped viruses must perform. As reported below, bothcompounds appear to act by the same mechanism. Both arepotent inactivators of fusion by all the enveloped virusesstudied. Loss of fusion activity appears to account for theinactivation of infectivity of both vesicular stomatitis virus(VSV) and HIV, induced in the presence of light, by either ofthese photodynamic agents.

MATERIALS AND METHODSViruses. Sendai (Z strain) and influenza (APR/8/34 strain)

viruses were grown in 10- to 11-day-old embryonated eggs andpurified on 5-40% potassium tartrate gradients. VSV Indiana(Birmingham strain) was grown in BHK-21F cells and purifiedas described (19). HIV-1 (LAV strain) was prepared bypassage in phytohemagglutinin-stimulated human peripheralblood lymphocytes and was a kind gift of Kathleen Clouse andthe Viral Biology Unit (Georgetown University).

Reagents. Rose bengal was obtained from Sigma, andhypericin was from Sigma or Atomergic (Farmingdale, NY).Stock solutions of rose bengal (100 AM) were made up inwater. Stock solutions of hypericin (400 AuM) were made upby adding 1 vol of methanol or dimethyl sulfoxide to the solidcompound, heating gently as necessary (especially withmethanol) to dissolve, and then adding 9 vol of phosphate-buffered saline at pH 7.2 (PBS) containing 0.1% Tween 80 (5).Samples were prepared under conditions of reduced light;that is, lights immediately overhead were turned off, butincident light remained sufficient to carry out the requiredoperations (measured illumination, <1 footcandle; 1 footcan-dle = 10.76 lx).

Illumination. The purified or stock viruses were suspendedin translucent plastic tubes or plate wells in 100 ul of PBScontaining the desired concentrations of hypericin or rosebengal. Illumination was carried out for 1 h in ice by using astandard fluorescent desk lamp (containing two Philips F15T8/CW 15-W bulbs) placed 5 cm above the sample. Illumi-nation was 8-900 footcandles, as measured with a SpectraCandela light meter.

Hemolysis. After illumination of the virus sample, 1.4 ml ofbuffer and 0.1 ml of a 10% suspension offresh washed humanerythrocytes in PBS were added, and the sample was mixed.PBS was used for Sendai virus-induced hemolysis. Citricacid/phosphate buffer at pH 5.0 (20) was used for influenzavirus- and VSV-induced hemolysis; the buffer containedDEAE-dextran (3 ,ug/ml) for the VSV experiments (21). Thesuspension was incubated in the dark for 1 h on ice, followedby 1 h at 37°C. The samples were then centrifuged at 1200 xg for 10 min at 4°C, and absorbance of the supernatant wasmeasured at 590 nm. Values corresponding to complete(100%) hemolysis were determined in each experiment fromsamples lysed in distilled water. The amount of virus used ineach experiment (1-5 ,ug of virus protein) was controlled soas to produce 30-80% hemolysis under these conditions.

Infectivity. Stock VSV was illuminated in the presence ofhypericin or rose bengal as described above and then used toinfect BHK-21F cells at a multiplicity of infection of 5. Theinfection was allowed to proceed overnight. The cell mediumwas then collected, serially diluted, and titrated for plaque-forming units on BHK-21F cell monolayers. Similarly, HIV-1was treated with hypericin or rose bengal, illuminated, andthen added to 4 x 105 MT-4 cells (22) at a multiplicity ofinfection of 0.01 in 1.5 ml of RPMI 1640 with 10% (vol/vol)fetal calf serum (GIBCO/BRL) in 24-well plates. Culturemedium was changed on days 4 and 6, and aliquots of culturesupernatant were harvested daily for RT assay. Cell viabilitywas determined by trypan blue dye exclusion.HIV RT Assay. HIV production in infected cultures was

determined by a 32P-based RT assay previously described

(23) in which samples were incubated with RT cocktail for 1.5h and then spotted on DEAE paper and washed as described.RT activity was determined by quantitation of 32P bound tothe filter by using a Molecular Dynamics phosphorimagerfollowing a 3-h exposure to a phosphorimaging screen.HIV gpl20-Induced Syncytium Formation. A culture of

ACH-2 cells was induced to active production of HIV byincubation with phorbol 12-myristate 13-acetate (PMA) asdescribed (24, 25). After 24 h of induction, 4 x 103 ACH-2cells were suspended in 100 gl of buffer in a microtiter dish,treated with hypericin or rose bengal with or without illumi-nation, then mixed with 4 x 105 Sup-T1 cells (a CD41 T-cellline that forms large syncytia upon HIV infection; ref. 26),and incubated overnight at 370C; "dark samples" werewrapped in foil. Syncytia were identified by microscopicexamination, and those exceeding a minimum size offour celldiameters were counted.SDS/PAGE. SDS/PAGE was carried out by the method of

Laemmli (27) using 10% polyacrylamide gels with 6% stack-ing gels.

RESULTSInactivation of Infectivity. Stock preparations of HIV or

VSV were mixed with various concentrations of hypericin orrose bengal and illuminated prior to infecting their respectivehost cells as described. Dark samples were prepared andincubated identically, but were shielded from the light bywrapping in aluminum foil. Results with VSV are shown inFig. 1, and those with HIV are shown in Fig. 2. Both viruseslost infectivity upon illumination with either photosensitizer,in a concentration-dependent manner.The concentration of hypericin or rose bengal required to

cause 50% inactivation (150) under standard illuminationconditions is shown in Table 1. Values for VSV inactivationwere 20 nM hypericin and 50 nM rose bengal. The '50 valuefor loss of HIV infectivity depended upon the time of assay.Thus, on day 3 postinfection 150 for rose bengal was =8 nM,while on day 5 postinfection it was 50 nM; a similar increaseoccurred for hypericin between day 5 and day 7 postinfec-tion. The higher I50 on later days resulted because the lowestconcentrations delayed, but did not completely prevent, newvirus production (Fig. 2). This delay was due to a 3 logreduction in infective titer in the viral stock, induced by thedrug treatment and illumination (data not shown). Completeinhibition of HIV infectivity was achieved by using 30 nM

C.0

0 30 100 300 1000 300010,000 10 30 100 300 1000

Hypericin, nM Rose bengal, nM

FIG. 1. Photodynamic inactivation of infectivity of VSV byhypericin (A) or rose bengal (B). pfu, Plaque-forming units.

Medical Sciences: Lenard et al.

160 Medical Sciences: Lenard et al.

a 300-

cc

4 5 6 -7 8 1

Dav at.

300 -

200- 400~

100A 2001

01 01

mock 0 10 30 100 300 1000 1090 mock 0 10 30 100 300 10001000dark arK

Hypeicin. nM Rose bergalI,

FIG. 2. Photodynamic inactivation of HIV-1 infectivity by hypericin (A and C) or rose bengal (B and D). (A and B) RT activity present inthe culture supernatant at the indicated times was measured as described in Materials and Methods. Numbers refer to concentrations (nM) ofthe appropriate drug. (C and D) RT activity values determined 5 days postinfection as a function ofdrug concentration. RT activity was recordedas phosphorimager pixel units x 104. The values shown are the average of three independent wells at each point. Error bars indicate standarddeviations. HIV, untreated control.

hypericin or 100 nM rose bengal. No inactivation was seen inthe presence of 1 AM hypericin or rose bengal in the absenceof light. Further, growth of MT-4 cells was not inhibited bythe illuminated HIV stocks containing 1 ,AM hypericin or rosebengal, showing that inhibition ofHIV replication was due toeffects on the virus, not on the cells.

Inactivation of VSV, Influenza Virus, and Sendai Virus-Induced Hemolysis. In view of the broad antiviral activitydemonstrated by hypericin and by rose bengal against envel-oped but not nonenveloped viruses, it was of interest todetermine the effects of these agents on viral fusion. This isconveniently measured for many viruses by virus-inducedhemolysis (21, 28, 29), since fusion of the viral envelope withthe erythrocyte membrane is an essential prerequisite toleakage of hemoglobin. To test for photodynamic inactivationof viral fusion, therefore, hypericin or rose bengal was addedto purified virus preparations and illuminated as described;hemolytic activity was then assayed in the dark. As shown inFig. 3, the fusion activity of VSV and influenza and Sendaiviruses was inactivated by nanomolar concentrations of bothdrugs. Inactivation was absolutely dependent upon light (Fig.3) and increased with increasing time of illumination (data notshown). I5o values differed for each virus/photosensitizercombination, but most fell in the 20-80 nM range at the virusconcentration used (Table 1).

It was possible that the inactivation offusion shown in Fig.3 might have resulted from an impaired ability of the virus tobind to the erythrocyte cell surface. To test for this, hemag-glutination titers were determined for aliquots of influenzaand Sendai viruses that had been photodynamically treated asin Fig. 3. Hemagglutination was unaffected by any concen-

tration of either dye (data not shown). This indicates thatfusion rather than binding was inactivated by the photody-namic action of the dyes.The susceptibility of erythrocytes to direct photodynamic

hemolysis (i.e., in the absence of virions) by hypericin androse bengal were also measured. Both reagents induceddirect photodynamic hemolysis (data not shown), confirmingprevious reports (10-13). 150 values were much higher thanthose for viral inactivation of hemolysis, however (Table 1).It should be noted that the drug concentrations shown in Fig.3 were those that were present during illumination of the

Table 1. 15o values for photodynamic effects of rose bengal orhypericin on hemolysis, infectivity, and protein cross-linking

I5o (±50%o), nMVirus Function Hypericin Rose bengalNone Hemolysis at pH 7.2 14,000 350

Hemolysis at pH 5 6700VSV Inactivation of hemolysis 20 50

Inactivation of infectivity 25 20Cross-linking of G 1,000 500Cross-linking of M 5,000 1000

HIV-1 Inactivation of infectivity <10 (5)* 8 (3)*20 (7)* 50 (5)*

Inf Inactivation of hemolysis 80 50Sendai Inactivation of hemolysis 220 35

Inf, influenza.*The numbers in parentheses are the days at which the assay wasperformed.

Proc. Natl. Acad Sci. USA 90 (1993)

Proc. Natl. Acad. Sci. USA 90 (1993) 161

E 0.8eniSendai Sendai0.8 -

0

0.6T

~0.4-

d: 0.2-

O~ F.mI0.6

0.4-

0.2-

C30 10030010003000 3 10 30 1003001000

Hypericin, nM Rose bengal, nM

FIG. 3. Photodynamic inactivation of hemolysis of VSV (A andB), Sendai virus (C and D), or influenza virus (E and F) by hypericin(A, C, and E) or rose bengal (B, D, and F). A relative hemolysis valueof 1.0 equals 30-80o hemolysis.

virus; as described in Materials and Methods, they were

diluted 15-fold prior to the hemolysis assay.Inactivation of 1IV gp120-Induced Syncytium Formation.

Viral fusion activity may also be assayed by the formation ofcell syncytia mediated by cells expressing the viral fusionprotein on their plasma membranes. For HIV, syncytiaformation involves the fusion of uninfected CD4+ cells withcells expressing cell surface HIV gpl20. ACH-2 cells are a

chronically infected T-cell line in which high levels of HIVprotein expression can be induced by treatment with phorbol

ALig h t ---- '7

Hypericin, uM 0 0.41 1 1 0 10...- --- W--- -

Table 2. Effect of hypericin or rose bengal and light onsyncytium formation by HIV-producing ACH-2 cellswith SUP-T1 cells

Treatment of ACH-2 cells* Syncytia per platetNone 155 ± 34None (no PMA) 3 ± 31 AM hypericin + light 4 ± 41 AM hypericin (dark) 227 ± 641AM rose bengal + light 38 ± 31AM rose bengal (dark 205 ± 17

*Treatment 24 h after induction of HIV with PMA, except asindicated.tAverage ± SD of three plates.

esters (24, 25). Induced ACH-2 cells efficiently form syncytiawhen incubated with Sup-T1 cells (26). ACH-2 cells weretreated with hypericin or rose bengal, with or without light,24 h after induction of HIV protein expression by PMA, inorder to study the effects ofthese treatments on HIV-inducedcell fusion. Syncytium formation between these cells andSup-T1 cells was completely abolished by 1 1LM hypericin,but illumination was required; syncytium formation was

unaffected by hypericin in the dark (Table 2). Treatment with1 gM rose bengal decreased syncytium production by >80P/oand was similarly light dependent (Table 2).

Cross-L of Viral Proteins. Treatment of membraneswith rose bengal has been reported to result in the cross-

linking of membrane proteins, evidenced by their disappear-ance from their usual positions in SDS/PAGE and theiraccumulation at the top of the gel (13, 15). We thereforeexamined the effects of hypericin and rose bengal on cross-

linking ofVSV proteins. As shown in Fig. 4, both compoundscaused a.similar pattern of protein cross-linking. Of the threemajor VSV proteins, the integral membrane protein G wasmost readily cross-linked, followed by M, a peripheral mem-brane protein that partially penetrates the viral bilayer (30).The nucleocapsid protein, N, which binds the genomic RNAbut is not membrane-associated, was not significantly cross-linked by either com ound. These findings provide an addi-tional indication that the predominant action of both hyper-icin and rose bengal is at the viral membrane and that bothdyes are exerting similar effects.

DISCUSSIONIn this report we have provided evidence that the photody-namic inactivation of enveloped viruses by hypericin may

Medical Sciences: Lenard et al.

BLight

Rose bengal, nM 0 3 " CD 301 00 300' 1 000 1--. - - M.W

" -w -..

."mm - - -

M -_AOb m_ -Ws

FIG. 4. Photodynamically induced cross-linking of VSV proteins by hypericin (A) or rose bengal (B). L, G, M, and N are the major VSVproteins.

162 Medical Sciences: Lenard et al.

arise chiefly from the destruction of its fusion function. Thefusion activity of three different enveloped viruses wasphotoinactivated by concentrations of hypericin ranging from20 to 220 nM. HIV gpl20-mediated cell fusion, as measuredby syncytium formation, was photoinactivated completely by1 gM hypericin and >80% by 1 ,uM rose bengal. The higherconcentrations used for inhibition of syncytium formationwere likely necessitated by the larger amounts of membranepresent in the cell samples as compared to viral stocks.

Similar concentrations of hypericin have previously beenreported to inactivate Sindbis virus and murine cytomegalo-virus in a light-dependent fashion (7). Higher concentrationswere required for photodynamic inactivation of equine in-fectious anemia virus (6), and of other lipid-containing vi-ruses in experiments in which light was not controlled (1-5).Comparison of the effects of hypericin and rose bengal

permitted the following observations. (i) Both dyes actsimilarly, inactivating infectivity and/or fusion in all fourenveloped viruses tested and promoting similar cross-linkingof viral membrane proteins (Table 1). Both dyes are efficientproducers of singlet oxygen upon illumination (14-16), andboth are known to associate with biological membranes.Reactions of singlet oxygen induced by either compound thuspresumably cause most or all of the observed effects. (ii)Loss offusion function may be the general reason enveloped,but not nonenveloped, viruses are inactivated by hypericinand light. In support of this suggestion, the 150 for VSVinactivation of infectivity and that for inactivation of hemno-lysis were very similar, using either hypericin or rose bengalas photosensitizer (Table 1).As clinical trials of hypericin are currently under develop-

ment,t it is critical to understand the mechanism by whichthis drug exerts its antiretroviral effects. Most importantly,we have demonstrated that light is essential for all theantiviral activities of hypericin and rose bengal that werestudied. In this connection, it should be noted that the actionof hypericin and rose bengal resembles quite closely that ofseveral other photodynamic antiviral compounds: merocya-nine-540, certain porphyrin derivatives, and certain phthalo-cyanine derivatives (31-33). All of these associate withmembranes and are thought to work at least in part bygeneration of singlet oxygen (34). In common with theseagents, hypericin and rose bengal might prove to be suitableagents for photodynamic inactivation of enveloped virusespresent in blood or blood products, although lack of speci-ficity might be a problem with any of these agents. Thepossibility of illuminating the blood of HIV patients afteradministration of a photodynamic dye might also be inves-tigated, as has been reported for psoralen (35).The close similarity of effects seen with hypericin and rose

bengal and their similarity to effects seen with other photo-dynamic compounds suggests that several of these should beconsidered together and compared as potential anti-HIVagents, both in vivo and in vitro.

tGulick, R., Lui, H., Anderson, R., Kollias, N., Hussey, S. &Crumpacker, C., 8th International Conference on AIDS, July19-24, 1992, Amsterdam, abstr, PoB3D18.

This study was supported in part by National Institutes of HealthGrant AI-13002 to J.L.

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

2. Lavie, G., Valentine, F., Levin, B., Mazur, Y., Gallo, G.,Lavie, D., Weiner, D. & Meruelo, D. (1989) Proc. Natl. Acad.Sci. USA 86, 5963-5967.

3. Tang, J., Colacino, J. M., Larsen, S. H. & Spitzer, W. (1990)Antiviral Res. 13, 313-326.

4. Schinazi, R. F., Chu, C. K., Babu, J. R., Oswald, B. J., Saal-mann, V., Cannon, D. L., Eriksson, B. F. H. & Nasr, M.(1990) Antiviral Res. 13, 265-272.

5. Andersen, D. O., Weber, N. D., Wood, S. G., Hughes, B. G.,Murray, B. K. & North, J. A. (1991) Antiviral Res. 16, 185-1%.

6. Carpenter, S. & Kraus, G. A. (1991) Photochem. Photobiol. 53,169-174.

7. Hudson, J. B., Lopez-Bazzocci, I. & Towers, G. H. N. (1991)Antiviral Res. 15, 101-112.

8. Blum, H. F. (1941) Photodynamic Action and Diseases Causedby Light (Reinhold, New York).

9. Knox, J. P. & Dodge, A. D. (1985) Plant Cell Environ. 8,19-25.

10. Pace, N. & Mackinney, G. (1941) J. Am. Chem. Soc. 63,2570-2574.

11. Blum, H. F., Pace, N. & Garrett, R. L. (1937) J. Cell. Comp.Physiol. 9, 217-239.

12. Valenzeno, D. P. & Pooler, J. P. (1982) Photochem. Photobiol.35, 343-350.

13. Barratt, M. D., Evans, J. C., Lewis, C. A. & Rowlands, C. C.(1982) Chem.-Biol. Interact. 38, 215-230.

14. Duran, N. & Song, P.-S. (1986) Photochem. Photobiol. 43,677-680.

15. Valenzeno, D. P. (1987) Photochem. Photobiol. 46, 147-160.16. Giulivi, C., Sarcansky, M., Rosenfeld, E. & Boveris, A. (1990)

Photochem. Photobiol. 52, 745-751.17. Turner, G. S. & Kaplan, C. (1968) J. Gen. Virol. 3, 433-443.18. Roat, M. I., Romanowski, E., Araullo-Cruz, T. & Gordon,

Y. J. (1987) Arch. Ophthalmol. 105, 1415-1417.19. Miller, D. K., Feuer, B. I., Vanderoef, R. & Lenard, J. (1980)

J. Cell Biol. 84, 421-429.20. Gomori, G. (1955) Methods Enzymol. 1, 138-146.21. Bailey, C. A., Miller, D. K. & Lenard, J. (1984) Virology 133,

111-118.22. Harada, S., Koyanagi, Y. & Yamamoto, N. (1985) Science 229,

563-566.23. Willey, R., Smith, D. H., Laskey, L. A., Theodore, T. S.,

Earl, P. L., Moss, B., Capon, J. & Martin, M. A. (1988) J.Virol. 62, 139-147.

24. Clouse, K. A., Powell, D., Washington, I., Poli, G., Strebel,K., Farrar, W., Barstad, P., Kovacs, J., Fauci, A. S. & Folks,T. M. (1989) J. Immunol. 142, 431-438.

25. Folks, T. M., Clouse, K. A., Justement, J., Rabson, A., Duh,E., Kehrl, J. H. & Fauci, A. S. (1989) Proc. Natl. Acad. Sci.USA 86, 2365-2368.

26. Hoxie, J. A., Alpers, J. D., Ratkowski, J. L., Huebner, K.,Haggerty, B. S., Cedarbaum, A. J. & Reed, J. C. (1986) Sci-ence 234, 1123-1127.

27. Laemmli, U. K. (1970) Nature (London) 227, 680-685.28. Lenard, J. & Miller, D. K. (1981) Virology 110, 479-482.29. Lenard, J., Bailey, C. A. & Miller, D. K. (1982) J. Gen. Virol.

62, 353-355.30. Lenard, J. & Vanderoef, R. (1990) J. Virol. 64, 3486-3491.31. Sieber, F., O'Brien, J. M., Krueger, G. J., Schober, S. L.,

Burns, W. H., Sharkis, S. J. & Sensenbrenner, L. L. (1987)Photochem. Photobiol. 46, 707-711.

32. North, J., Freeman, S., Overbaugh, J., Levy, J. & Lansman,R. (1992) Transfusion 32, 121-128.

33. Horowitz, B., Williams, B., Rywkin, S., Prince, A. M., Pas-cual, D., Geacintov, N. & Valinsky, J. (1991) Transfusion 31,58-64.

34. Kalyanaraman, B., Felix, J. B., Sieber, F., Thomas, J. P. &Girotti, W. (1987) Proc. Natl. Acad. Sci. USA 84, 2999-3003.

35. Bisaccia, E., Berger, C. & Klainer, A. S. (1990) Ann. Int. Med.113, 270-275.

Proc. Natl. Acad. Sci. USA 90 (1993)